196 24 35MB
English Pages 433 Year 2016
NEUROIMAGING GENE TICS
NEUROIMAGING GENE TICS Principles and Practices
EDITED BY
Kristin L. Bigos, PhD Division of Clinical Pharmacology and Pharmacogenetics Lieber Institute for Brain Development D e p a r t m e n t o f M e d i c i n e , D i v i s i o n o f C l i n i c a l P h a r m a c o l o g y, a n d Depar tment of Psychiatr y and Behavioral Sciences Johns Hopkins School of Medicine Baltimore, MD
Ahmad R. Hariri, PhD Department of Psychology and Neuroscience Duke University D u r h a m , N C
Daniel R. Weinberger, MD Lieber Institute for Brain Development D e p a r t m e n t s o f P s y c h i a t r y a n d B e h a v i o r a l S c i e n c e s , N e u r o l o g y, a n d N e u r o s c i e n c e ; a n d t h e M c Ku s i c k - N a t h a n s I n s t i t u t e o f G e n e t i c M e d i c i n e Johns Hopkins School of Medicine Baltimore, MD
1
1 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trademark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016
© Oxford University Press 2016 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-in-Publication Data Names: Bigos, Kristin L., editor. | Hariri, Ahmad R., 1972–, editor. | Weinberger, Daniel R. (Daniel Roy) Title: Neuroimaging genetics : principles and practices / edited by Kristin L. Bigos, Ahmad R. Hariri, Daniel R. Weinberger. Description: Oxford ; New York : Oxford University Press, [2016] | Includes bibliographical references and index. Identifiers: LCCN 2015012030 | ISBN 9780199920211 Subjects: | MESH: Neuroimaging. | Brain—physiology. | Brain Chemistry—genetics. | Mental Disorders—genetics. Classification: LCC QP376.6 | NLM WL 141.5.N47 | DDC 612.8/2—dc23 LC record available at http://lccn.loc.gov/2015012030 This material is not intended to be, and should not be considered, a substitute for medical or other professional advice. Treatment for the conditions described in this material is highly dependent on the individual circumstances. And, while this material is designed to offer accurate information with respect to the subject matter covered and to be current as of the time it was written, research and knowledge about medical and health issues is constantly evolving and dose schedules for medications are being revised continually, with new side effects recognized and accounted for regularly. Readers must therefore always check the product information and clinical procedures with the most up-to-date published product information and data sheets provided by the manufacturers and the most recent codes of conduct and safety regulation. The publisher and the authors make no representations or warranties to readers, express or implied, as to the accuracy or completeness of this material. Without limiting the foregoing, the publisher and the authors make no representations or warranties as to the accuracy or efficacy of the drug dosages mentioned in the material. The authors and the publisher do not accept, and expressly disclaim, any responsibility for any liability, loss or risk that may be claimed or incurred as a consequence of the use and/or application of any of the contents of this material.
9 8 7 6 5 4 3 2 1 Printed in the United States of America on acid-free paper
CONTENTS
Contributors 1. NEUROIMAGING GENETICS
Kristin L. Bigos, Ahmad R. Hariri, and Daniel R. Weinberger
vii 1
10. GENETICS OF BRAIN STRUCTURE
Jason L. Stein, Derrek P. Hibar, and Paul M. Thompson 11. IMAGING GENETICS OF REWARD PROCESSING IN THE HUMAN BRAIN
Caroline F. Zink
157
167
PA R T I . I M AG I N G G E N E T I C S A N D N E U R O C H E M I S T R Y 2. MOLECULAR NEUROIMAGING GENETICS
Patrick M. Fisher and Gitte Moos Knudsen
15
PA R T I V. I M AG I N G G E N E T I C S A N D D I S E A S E 1 2. IMAGING GENETICS OF SCHIZOPHRENIA
3. NEURORECEPTOR IMAGING OF GENETIC VARIATION IMPACTING THE SEROTONIN TRANSPORTER
Greg Perlman, Christine DeLorenzo, J. John Mann, and Ramin Parsey 4. IMAGING OF GENETIC VARIATION IMPACTING DOPAMINE TRANSMISSION PARAMETERS
Elsmarieke van de Giessen, Rassil Ghazzaoui, Raj Narendran, and Anissa Abi-Dargham 5. IMAGING GENETICS OF DOPAMINE SYNAPTIC TERMINAL ACTIVITY
Giuseppe Blasi and Alessandro Bertolino
31
13. IMAGING GENETICS OF DEPRESSION
Matthew D. Sacchet, Lara C. Foland-Ross, and Ian H. Gotlib 49 14. IMAGING GENETICS OF ANXIETY DISORDERS
Mats Fredrikson
67
PA R T I I . I M AG I N G G E N E T I C S A N D DRU G D I S C OV E R Y 6. VARIABILITY OF ANTIDEPRESSANT DRUG RESPONSE: CONTRIBUTION OF IMAGING GENETICS STUDIES
Ulrich Rabl, Bernhard M. Meyer, and Lukas Pezawas 7. IMAGING GENETICS OF PHARMACOLOGICAL RESPONSE IN PSYCHIATRIC DISORDERS
Philip R. Szeszko and Anil K. Malhotra
81
Daniel Paul Eisenberg, Ranjani Prabhakaran, and Karen F. Berman 9. IDENTIFYING UNANTICIPATED GENES AND MECHANISMS IN SERIOUS MENTAL ILLNESS: GWAS-BASED IMAGING GENETICS STRATEGIES
Steven G. Potkin, Theo G. M. van Erp, Shichun Ling, Fabio Macciardi, and Xiaohui Xie
15. IMAGING GENETICS OF BIPOLAR DISORDER
Ole A. Andreassen and Martin Tesli 16. GENETIC NEUROIMAGING STUDIES OF BASAL GANGLIA DISORDERS
Trevor W. Robbins, James B. Rowe, and Roger A. Barker 17. IMAGING GENETICS OF ANTISOCIAL BEHAVIOR AND PSYCHOPATHY
Hayley M. Dorfman and Joshua W. Buckholtz 101
PA R T I I I . I M AG I N G G E N E T I C S A N D G E N E T I C D I S C OV E R Y 8. IMAGING GENETICS OF WILLIAMS SYNDROME
Thomas M. Lancaster, Joanne L. Doherty, David E. Linden, and Jeremy Hall
113
187
209
223
233
241
251
PA R T V. I M AG I N G G E N E T I C S A N D T H E E N V I R O N M E N T 18. INCORPORATING THE ENVIRONMENT INTO NEUROGENETICS RESEARCH: AN IMAGING GENE BY ENVIRONMENT INTERACTIONS (IG X E) APPROACH
Luke W. Hyde, Ryan Bogdan, and Ahmad R. Hariri
267
PA R T V I . I M AG I N G G E N E T I C S A N D C O G N I T I O N 141
19. IMAGING GENETICS OF EPISODIC MEMORY
Björn Rasch, Susanne Erk, Andreas Papassotiropoulos, and Dominique J.-F. de Quervain
289
v
20. IMAGING GENETICS OF WORKING MEMORY
Tristram A. Lett, Eva J. Brandl, Daniel J. Müller, Andreas Heinz, and Henrik Walter
309
PA R T V I I . I M AG I N G G E N E T I C S O F C O G N I T I V E AG I N G 21. NEUROBIOLOGY AND GENETICS OF COGNITIVE AGING: INSIGHTS FROM NEUROIMAGING STUDIES
John C. Muse, Milap A. Nowrangi, Daniel R. Weinberger, and Venkata S. Mattay
PA R T V I I I . I M AG I N G G E N E T I C S A N D M U LT I - L O C U S M O DE L S 23. IMAGING GENETICS OF THE HYPOTHALAMICPITUITARY-ADRENAL AXIS: IMPLICATIONS FOR PSYCHOPATHOLOGY
Nadia S. Corral-Frias, Lindsay J. Michalski, Christina R. Di Iorio, and Ryan Bogdan
397
327 Index
411
22. NEUROIMAGING GENETICS OF ALZHEIMER’S DISEASE 357
Theresa M. Harrison, Alison C. Burggren, and Susan Y. Bookheimer
v i C ontents
CONTRIBUTORS
Anissa Abi-Dargham, MD Department of Psychiatry Columbia University New York State Psychiatric Institute New York, NY Ole A. Andreassen MD, PhD NORMENT–KG Jebsen Centre University of Oslo and Oslo University Hospital Oslo, Norway Roger A. Barker, MD, PhD Department of Clinical Neurosciences Behavioural and Clinical Neuroscience Institute John van Geest Centre for Brain Repair University of Cambridge Cambridge, UK Karen F. Berman, MD Section on Integrative Neuroimaging Clinical and Translational Neuroscience Branch Intramural Research Program, NIMH, NIH, DHHS Bethesda, MD Alessandro Bertolino, MD, PhD NORD DTA Neuroscience Hoffman–La Roche Ltd. Basel, Switzerland Giuseppe Blasi, MD Department of Basic Medical Sciences Neuroscience and Sense Organs University of Bari Aldo Moro Bari, Italy Ryan Bogdan, PhD Department of Psychology Washington University in St. Louis St. Louis, MO
Susan Y. Bookheimer, PhD Department of Psychiatry and Biobehavioral Sciences University of California, Los Angeles Los Angeles, CA Eva J. Brandl, MD Department of Psychiatry and Psychotherapy Charité Universitätsmedizin Berlin, Germany Joshua W. Buckholtz, PhD Department of Psychology Harvard University Cambridge, MA Alison C. Burggren, PhD Department of Psychiatry and Biobehavioral Sciences University of California, Los Angeles Los Angeles, CA Nadia S. Corral-Frias, PhD Psychiatry Department Washington University in St. Louis St. Louis, MO Christine DeLorenzo, PhD Department of Psychiatry Stony Brook University Stony Brook, NY Dominique J.-F. de Quervain, MD Division of Cognitive Neuroscience University of Basel Basel, Switzerland Christina R. Di Iorio, MA Department of Psychology Washington University in St. Louis St. Louis, MO
vii
Joanne L. Doherty, BA BM BCh MRCPsych Institute of Psychological Medicine and Clinical Neurosciences Cardiff University Cardiff, UK Hayley M. Dorfman, BA Department of Psychology Harvard University Cambridge, MA Daniel Paul Eisenberg, MD Section on Integrative Neuroimaging Clinical and Translational Neuroscience Branch Intramural Research Program, NIMH, NIH, DHHS Bethesda, MD Susanne Erk, PhD Division of Mind and Brain Research Department of Psychiatry and Psychotherapy Charité–Universitätsmedizin Berlin, Campus Mitte Berlin, Germany
Jeremy Hall, MA MB BChir MPhil PhD MRCPsych Neuroscience and Mental Health Research Institute Cardiff University School of Medicine Cardiff, UK Theresa M. Harrison, BS Department of Psychiatry and Biobehavioral Sciences University of California, Los Angeles Los Angeles, CA Andreas Heinz, MD, PhD Department of Psychiatry and Psychotherapy Charité Universitätsmedizin Berlin, Germany Derrek P. Hibar, PhD Imaging Genetics Center, Institute for Neuroimaging & Informatics Keck School of Medicine of the University of Southern California Los Angeles, CA
Patrick M. Fisher, PhD Neurobiology Research Unit Rigshospitalet Copenhagen, Denmark
Luke W. Hyde, PhD Department of Psychology Center for Human Growth and Development University of Michigan Ann Arbor, MI
Lara C. Foland-Ross, PhD Department of Psychology Stanford University Stanford, CA
Gitte Moos Knudsen, DMSc Neurobiology Research Unit Rigshospitalet Copenhagen, Denmark
Mats Fredrikson, PhD, DMSc Department of Psychology Uppsala University Uppsala, Sweden
Thomas M. Lancaster, PhD Neuroscience and Mental Health Research Institute Cardiff University School of Medicine Cardiff, UK
Rassil Ghazzaoui MA Department of Psychiatry Columbia University New York State Psychiatric Institute New York, NY
Tristram A. Lett, PhD Department of Psychiatry and Psychotherapy Charité Universitätsmedizin Berlin, Germany
Ian H. Gotlib, PhD Department of Psychology Stanford University Stanford, CA
David E. Linden, MD, DPhil Institute of Psychological Medicine and Clinical Neurosciences Cardiff University Cardiff, UK
v i i i C ontributors
Shichun Ling, BA Department of Psychiatry and Human Behavior University of California, Irvine Irvine, CA
Milap A. Nowrangi, MD Department of Psychiatry and Behavioral Sciences Johns Hopkins University School of Medicine Baltimore, MD
Fabio Macciardi, MD, PhD Department of Psychiatry and Human Behavior University of California, Irvine Irvine, CA
Andreas Papassotiropoulos, MD Division of Molecular Neuroscience Faculty of Psychology University of Basel Basel, Switzerland
Anil K. Malhotra, MD Center for Psychiatric Neuroscience Feinstein Institute for Medical Research Manhasset, NY J. John Mann, MD Department of Psychiatry Columbia University New York, NY Venkata S. Mattay, MD Division of Cognitive Neuroscience and Imaging Genetics Lieber Institute for Brain Development Department of Neurology Johns Hopkins School of Medicine
Ramin Parsey, MD, PhD Department of Psychiatry Stony Brook University Stony Brook, NY Greg Perlman, PhD Department of Psychiatry Stony Brook University Stony Brook, NY Lukas Pezawas, MD Department of Psychiatry and Psychotherapy Medical University of Vienna Vienna, Austria
Bernhard M. Meyer, MSc Department of Psychiatry and Psychotherapy Medical University of Vienna Vienna, Austria
Steven G. Potkin, MD Department of Psychiatry and Human Behavior University of California-Irvine Irvine, CA
Lindsay J. Michalski, BA Department of Psychology Washington University in St. Louis St. Louis, MO
Ranjani Prabhakaran, PhD Section on Integrative Neuroimaging Clinical and Translational Neuroscience Branch Intramural Research Program, NIMH, NIH, DHHS Bethesda, MD
Daniel J. Müller, MD, PhD Department of Psychiatry University of Toronto Toronto, Canada John C. Muse, BA Division of Cognitive Neuroscience and Imaging Genetics Lieber Institute for Brain Development Baltimore, MD Raj Narendran, MD Department of Radiology University of Pittsburgh Pittsburgh, PA
C ontributors
Ulrich Rabl, MD Department of Psychiatry and Psychotherapy Medical University of Vienna Vienna, Austria Björn Rasch, PhD Division of Cognitive Biopsychology and Methods Department of Psychology University of Fribourg Fribourg, Switzerland
ix
Trevor W. Robbins, PhD Department of Psychology Behavioural and Clinical Neuroscience Institute University of Cambridge Cambridge, UK James B. Rowe, FRCP, PhD Department of Clinical Neurosciences Behavioural and Clinical Neuroscience Institute MRC Cognition and Brain Sciences Unit University of Cambridge Cambridge, UK
Paul M. Thompson, PhD Imaging Genetics Center, Institute for Neuroimaging & Informatics Keck School of Medicine of the University of Southern California Los Angeles, CA Elsmarieke van de Giessen, MD, PhD Department of Psychiatry Columbia University New York State Psychiatric Institute New York, NY
Matthew D. Sacchet, ScB Neurosciences Program Stanford University Stanford, CA
Theo G. M. van Erp, PhD Department of Psychiatry and Human Behavior University of California, Irvine Irvine, CA
Jason L. Stein, PhD Neurogenetics Program, Department of Neurology UCLA School of Medicine, Los Angeles Los Angeles, CA
Henrik Walter, MD, PhD Department of Psychiatry and Psychotherapy Charité Universitätsmedizin Berlin, Germany
Philip R. Szeszko, PhD Center for Psychiatric Neuroscience Feinstein Institute for Medical Research Manhasset, NY
Xiaohui Xie, PhD Department of Computer Science University of California, Irvine Irvine, CA
Martin Tesli, MD, PhD NORMENT–KG Jebsen Centre University of Oslo and Oslo University Hospital Oslo, Norway
Caroline F. Zink, PhD Division of Cognitive Neuroscience and Imaging Genetics Lieber Institute for Brain Development Department of Psychiatry and Behavioral Sciences Johns Hopkins School of Medicine Baltimore, MD
x C ontributors
1. NEUROIMAGING GENE TICS Kristin L. Bigos, Ahmad R. Hariri, and Daniel R. Weinberger
I
maging genetics offers an extraordinary tool with which to explore and evaluate the functional impact of brain-relevant genetic variation. The field of neuroimaging genetics has grown exponentially; in less than 20 years, nearly 40,000 articles have been published based on imaging genetics (www.pubmed.gov). Imaging genetics has been used as a tool by researchers in diverse areas of psychiatry, neurology, neuropsychology, and cognitive and affective neuroscience to study the genetic components of brain structure, neurochemistry, and functional differences in the living human brain. Imaging genetics studies have identified mechanisms of risk for clinical syndromes, have established biologic validation by elucidating mechanisms and pathways that confer genetic risk, and conversely have identified genes that contribute to structural and functional variation in brain circuitries. This book reviews the basic principles of neuroimaging techniques and their application to imaging genetics. The work presented in this volume elaborates on examples of the application of imaging genetics to understand neurochemical systems and pathways, to explore relationships between genetics and the structural and functional connectivity in the human brain, and to provide insight into mechanisms of risk for psychiatric and neurologic illness. Early techniques for identifying genotype-phenotype associations in the living brain began with the electroencephalogram (EEG) (Vogel et al. 1979). The emergence of functional magnetic resonance imaging (fMRI) as a high-resolution radioactivity-free imaging technique and the increasing interest in studying intermediate biologic phenotypes in complex medical disorders led to an explosion in neuroimaging research. Nearly 20 years ago, it was suggested that fMRI, because of its unique potential to map individual physiologic responses, could become a phenotyping strategy for genetic discovery (Weinberger et al. 1996). The first study to report an association of genetic
variation and a modern neuroimaging measure was a radionuclide imaging study with single-photon emission computed tomography (SPECT) that reported an association between a functional variation in the dopamine transporter gene (SLC6A3) and binding of [I-123]β-CIT to dopamine transporters in the striatum (Heinz et al. 2000; see Chapter 2 by Fisher and Knudsen in this volume). The first MRI-based neuroimaging genetics study was reported by Bookheimer and colleagues, investigating the effect of variation in the apolipoprotein E (APOE) gene, associated with risk for Alzheimer’s disease, on brain activity during memory tasks in healthy older adults with intact cognition (Bookheimer et al. 2000; see Chapter 22 by Harrison, Burggren, and Bookheimer in this volume). They showed that carriers of the risk-associated ε4 allele had greater activity in regions affected by Alzheimer’s disease, including hippocampal, parietal, and prefrontal regions during a memory task, which was thought to reflect inefficient mnemonic processing in the hippocampus (Bookheimer et al. 2000). At virtually the same time, Egan and colleagues reported the use of a similar strategy to test the effects of the catechol-O-methyltransferase (COMT) Val108/ 158Met genotype on prefrontal cortical function in patients with schizophrenia, their healthy siblings, and healthy unrelated subjects (Egan et al. 2001). This COMT variation is a functional single nucleotide polymorphism (SNP) that affects activity of the enzyme and thus, analogous to APOE4, offers a specific biological tool for probing the activity of a specific system related to the biology of the gene. COMT Val allele carriers, who have relatively increased cortical dopamine catabolism and therefore less cortical dopamine, showed greater activity in the prefrontal cortex during a working memory task in each of the clinical samples (Egan et al. 2001). This finding of prefrontal inefficiency during working memory is characteristic
1
of patients with schizophrenia and unaffected siblings, and this imaging marker is now a well-established intermediate phenotype of schizophrenia (see Chapter 12 by Lancaster, Doherty, Linden, and Hall in this volume). The association of this phenotype with COMT genotype was consistent with data from rodents and non-human primates that cortical dopamine tunes the efficiency of cortical microcircuits during working memory. The remarkable finding from this study is that healthy subjects carrying the COMT Val allele that had been associated with risk for schizophrenia show a similar pattern of inefficiency in the prefrontal cortex, even when their performance in the working memory task is not impaired, suggesting that the neural mechanism of association of the Val allele with risk for schizophrenia is based on this effect (Egan et al. 2001). This observation encouraged adopting the strategy of testing the effects of genetic variation on intermediate brain phenotypes in healthy subjects, rather than in patient populations, which have inherent confounders, including medication effects (Weinberger et al. 1996; Hariri and Weinberger 2003; Hariri, Drabant, and Weinberger 2006; Bigos and Hariri 2007; Bigos and Weinberger 2010). Though the COMT Val/Met variation has not been sustained as a consistent association with risk for schizophrenia, the COMT Val association with frontal cortical function during executive cognition has been confirmed in a number of subsequent studies and a meta-analysis (Mier, Kirsch, and Meyer-Lindenberg 2010). Susceptibility for major psychiatric disorders is significantly accounted for by genetics (Moldin and Gottesman 1997), and recent large-scale genome-wide association (GWA) studies have identified many genetic variants associated with risk for psychiatric illness. For example, the Psychiatric Genomics Consortium identified 108 independent genetic loci associated with risk for schizophrenia, in a sample of approximately 150,000 subjects (Schizophrenia Working Group 2014). The effects of these genetic variants include those involved in the development and function of brain regions responsible for specific cognitive and emotional processes, as well as some that may provide targets for future therapies. Because statistical association with clinical diagnosis does not establish biological significance or identify a mechanism of risk, imaging genetics is a uniquely valuable strategy for extending statistical evidence with biological data based on the living brain and will become an important strategy for extending and validating these new genetic associations. Another early seminal paper using this strategy measured the effect of genetic variation in the serotonin
transporter gene (SLC6A4) on amygdala reactivity (Hariri et al. 2002). Individual differences in serotonin function have been associated with affect and temperament in both human and animal models (Manuck et al. 1998). The serotonin transporter (5-HTT), which regulates the amount of serotonin in the synapse, is the target of most drugs used to treat depression and anxiety (reviewed in Chapter 3 by Perlman, DeLorenzo, Mann, and Parsey in this volume). There is a common polymorphism in the promoter region of the serotonin transporter (5-HTTLPR), and subjects who carry the short allele of this variant report increased temperamental anxiety and depression, though this association with temperament had been inconsistent and controversial (Bigos and Hariri 2007). This fMRI study found that even healthy subjects who have at least one copy of the short allele had increased amygdala reactivity to threatrelated facial expressions, suggesting that the biological basis of the temperament associations was responsivity of the amygdala to the perception of threat (Hariri et al. 2002). The data from this early study helped to establish imaging genetics as having the biological resolution to uncover neural mechanisms of clinical associations that are generally inconclusive. This research and others have shown that weak and inconsistent effects of genetic variation at the level of human cognition, emotion, and behavior are much more strongly associated with imaging phenotypes. The 5-HTTLPR study of Hariri et al. (2002) led to a number of replications and a meta-analysis (Hariri et al. 2002; Manuck et al. 1998; Munafò et al. 2009; Munafò, Freimer, et al. 2009); however, some studies showed varying effects, in general, because of variation of the specifics of the task paradigm. The clinical associations with the 5-HTTLPR variation have been weak, inconsistent, and controversial (Munafò, Freimer, et al. 2009; Risch et al. 2009). The effect of the gene on a biologic measure in the brain linked to its molecular function has so far been more consistent and stronger, confirming the assumption that the penetrance of gene effects will be greater at the level of brain biology than at the level of behavior. Principles of imaging genetics can also be applied to structural imaging methodologies, including voxel-based morphometry (VBM) and diffusion tensor imaging (DTI). However, unlike fMRI, in which every patient serves as his or her own control by measuring the level of activation in response to stimuli as compared to baseline or a neutral condition, there is still surprisingly limited information about the tissue components of structural measures, and the non-genetic determinants of variance in morphometric
2 cha p ter 1 : N euroimaging G enetics
techniques and DTI are difficult to identify and control (Scholz et al. 2009). Indeed, recent large-scale structural imaging studies of genetic variation have been surprisingly disappointing in identifying common variants associated with variation in brain structural measures (Hibar et al. 2015). One possible explanation for this result is that many non-genetic and non-structural variables contribute to signal with structural imaging measures, including fluid balances, exercise, medications, and perfusion of the tissue, to name a small list of many. In functional imaging studies involving so-called activation paradigms, many of these variables are controlled to some extent because of the within-subject design of imaging activation paradigms, whereas in structural imaging, there is no baseline against which the individual measures can be referenced. Psychiatric illnesses such as schizophrenia and bipolar disorder are complex genetic disorders likely resulting from alterations in multiple genes and environmental factors as well as interactions between risk factors. Imaging genetics can also be used to detect epistasis, biologic interactions between genes, using a similar strategy to the single candidate gene approaches. Imaging genetics also has been adopted to validate higher order genetic interactions (e.g., interaction among three or more genes), which are likely to be very important in complex behavioral disorders but are exceedingly difficult to validate in biologic systems relevant to human brain function. I M AG I N G G E N E T I C S A N D N E U RO C H E M I S T R Y This book was designed to survey the scope of imaging genetics in its current incarnation. It begins with Chapter 2 of this volume, a review by Fisher and Knudsen, which describes molecular neuroimaging as a technique for assaying specific neuromolecular processes. Molecular neuroimaging tools such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT) represent the most sensitive techniques available for non-invasively assaying molecular processes in the human brain in vivo, through the introduction of trace doses of a radioactive molecule targeting a specific protein of interest (e.g., receptor, transporter, enzyme, etc.). Molecular neuroimaging tools like PET and SPECT assay the density, affinity, and spatial distribution of specific neuromolecular processes in vivo, providing high-resolution data about underlying brain chemistry, including glucose metabolism, neurotransmission, enzyme activity, or protein availability. PET or SPECT can be used, in theory, cha p ter 1 : N euroimaging G enetics
to measure any aspect of molecular signaling for which a suitable radio-labeled molecule can be developed. In turn, molecular neuroimaging can be a powerful tool for linking molecular processes with relevant behavioral phenotypes and neuropsychiatric illness and, like fMRI, can help reveal the role of specific neurobiological mechanisms affecting risk for these illnesses, which may aid in the development of new treatment strategies. Advances in neuroimaging technologies have encouraged the use of multimodal imaging to elucidate the mechanisms behind the brain circuitry involved in mental illness. A study by Fisher and colleagues combined PET with fMRI to determine the contribution of 5-HT1A auto-receptors to amygdala reactivity (Fisher et al. 2006). This study found that 5-HT1A auto-receptor density (measured by PET) predicted 30%–44% of the variability in amygdala reactivity (measured by fMRI), which suggests that reduced capacity for negative feedback regulation of 5-HT release is associated with increased amygdala reactivity. Perlman, DeLorenzo, Mann, and Parsey in Chapter 3 of this volume review the molecular neuroimaging of genetic variation impacting the serotonin transporter. As previously mentioned, the serotonin transporter (SLC6A4, 5-HTT, or SERT) plays a critical role in regulating synaptic serotonin flux in the human nervous system. Chapter 3 by Perlman and colleagues introduces genetic neuroimaging of SERT using PET. They review research on the role of SERT in the nervous system and catalog the complex genetic variation in the SERT gene (SLC6A4). They outline strategies and considerations for measuring SERT using PET, and include general factors that may influence PET assessment of SERT (e.g., demographics, genetic variation, medical conditions). Finally, they summarize findings from studies using SERT PET to understand psychiatric phenotypes, and future directions for this research. Chapter 4 of this volume, by van de Giessen, Ghazzaoui, Narendran, and Abi-Dargham, reviews the impact of genetic variation on dopamine transmission parameters. The neurotransmitter dopamine is involved in several neuropsychiatric disorders such as schizophrenia, substance dependence, and Parkinson’s disease. This chapter reviews studies that examined the association between genetic polymorphisms and imaging measures of the dopamine system in the brain. These imaging measures assess dopamine synthesis capacity, dopamine receptors, dopamine transporters, and dopamine degradation in vivo in humans using PET and SPECT imaging. They describe the most frequently studied associations–for example, between a polymorphism (3' VNTR) in the dopamine transporter gene and expression of the striatal dopamine transporter protein. They also 3
describe the future of genetics and imaging of dopaminergic transmission, which is rapidly advancing by increasing sample sizes, by building multi-center collaborations, and by widening the focus to epigenetics and genes that are indirectly related to the dopamine system. Blasi and Bertolino in Chapter 5 review imaging genetics of synaptic terminal activity, primarily based on functional MRI, as opposed to the previous chapter using molecular neuroimaging, primarily PET. This review describes the details of using fMRI to study the functional impact of genetic variation of a variety of neurotransmitter systems, particularly dopamine. Genetic variation related to D2 receptors modulates cognitive and emotional phenotypes that can be measured using molecular assays, imaging tools, or behavioral tests, which will inform the study of brain disorders with strong links to dopaminergic function, such as schizophrenia. Variation of gene sequence is likely not the only determinant of the physiology and pathophysiology of dopamine-related brain functions, and therefore focusing on the interaction between environmental factors and dopamine-related genetic variation through epigenetic mechanisms may further shed light on the biological determinants of the physiology of the brain and on aspects of its dysregulation in brain disorders. I M AG I N G G E N E T I C S A N D DRU G DI S C OV E R Y The landscape of imaging genetics is vast and varied. In addition to the identification and validation of genetic associations in the living brain, studies of pharmacological interventions on brain structural MRI and fMRI measures and their modulation by genetic variation is an emerging area of research interest. Chapter 6 of this volume, by Rabl, Meyer, and Pezawas, provides an in-depth review of imaging approaches to understanding drug effects related to antidepressant drugs and the application of imaging genetics to elucidate the brain mechanisms of these pharmacological interventions. These authors also highlight the complexity of interpretations of genetic associations in brain states associated with illness, in this case depression, as genetic associations are very likely modified by illness state, by developmental environments, and by stress and cognitive states as well. While an ambitious goal of imaging genetics is to enable an understanding of individual factors that underlie illness and its treatment, the application of imaging genetics in clinical settings is a distant possibility. Rabl and colleagues eloquently remind us of the challenges in this regard and the improbability of clinical application anytime soon.
Szeszko and Malhotra, in Chapter 7, review studies using imaging genetics to study pharmacological responses in psychiatric disorders. There is considerable heterogeneity in treatment response across major psychiatric disorders, and the results of large-scale clinical trials cannot provide information for clinicians at the individual patient level. The identification of psychiatric patients who are non-responsive to psychotropic medications is a critical priority for clinicians, as these patients are at increased risk for long-term treatment resistance, severe functional disability, common medical morbidity, and in some cases, increased mortality given the risk of suicide. Although traditional pharmacogenetic studies assessing treatment response have yielded several novel findings, these studies generally have low effect sizes due to heterogeneity in diagnosis and behavioral outcomes. While studies directed at discerning predictors of treatment response using neuroimaging in psychiatric disorders are few so far, this chapter reviews these studies and outlines how imaging genetics should be used in the future to understand current treatments and drugs in development. The ability to identify neural circuitries that are improved by drugs and their relationship to genetic variants may explain underlying neurobiological processes and may provide useful biomarkers to predict response. They also point out that imaging genetics could be a valuable approach to evaluate brain penetration of current and novel psychiatric and neurologic drugs, and genes that encode central nervous system (CNS) transporters for drugs, such as ABCB1 (P-gp transporter), could be used to predict brain penetration in individual patients. Knowing how a particular polymorphism affects brain structure and/or function could further enhance treatment strategies such as dosing, and/or the use of medications with different pharmacologic properties. The pharmacological imaging genetics studies reviewed in Chapter 7 suggest that the identification of specific genetics-imaging relationships could increase the variance explained by individual candidate genes and thus could be tailored to patients at the individual level. I M AG I N G G E N E T I C S A N D G E N E T I C DI S C OV E R Y Many of the principles and practices of imaging genetics have been most clearly instantiated in the study of highly penetrant genetic syndromes. As Eisenberg, Prabhakaran, and Berman describe in Chapter 8, a crowning example of this is the study of Williams syndrome (WS). As they point out, the study of simpler genetic syndromes
4 cha p ter 1 : N euroimaging G enetics
represents a proving ground for imaging genetics without the inherent insensitivity and signal to noise problems of understanding the role of common variants in complex brain-related phenotypes. The WS arena of imaging genetics also illustrates how easy it is to mistakenly believe that association of genotype with phenotype is a direct manifestation of their linkage. Even in the context of the highly penetrant WS deletion and of well-established neuroimaging phenotypes associated with discrete cognitive characteristics of the disorder, the potential confounder of general intellectual deficiency is almost impossible to control. The studies of WS ask a more general question about brain development and neuroimaging phenotypes: Is there a consistent, robust change in brain structure and function that can be attributed to the WS genetic lesion? Because most patients with this syndrome have the same hemi-deletion, a common phenotype is reliably linked to the locus of the genetic lesion. Despite general cognitive developmental deficits, a cognitive profile of visual spatial deficits and relatively preserved language abilities suggests a syndromal cognitive signature. Despite a number of imaging findings in WS patients that are not consistent from one study to another, in general, structural and functional abnormalities of the parietal occipital region are consistent and replicable characteristics across many studies. These authors also tackle the more difficult task of attributing the unique sociability and lack of stranger anxiety of WS subjects to structural and functional analysis of amygdala and related circuitries, particularly involving orbital and insular cortices. The studies they review illustrate, in their inconsistencies, the difficulties inherent in understanding complex social behavior in the context of both the inadequacies of imaging methodology as an anatomical tool and the limitations of our current neural circuitry models of social behavior. While most of the existing imaging genetics literature is focused on relationships between genes with some prior probability, based on existing neurobiological information of being associated with specific brain phenotypes, there has been increasing interest in imaging genetics of the GWA approach to gene discovery. GWA is an agnostic search of common variants across the genome for possible association with imaging phenotypes. Potkin and colleagues (Chapter 9 of this volume) performed a structured review of the imaging genetics literature to examine studies of schizophrenia and bipolar disorder that used a GWA approach to gene discovery and then developed predictors of neuroimaging data from polygenic constructs based on the GWA statistics. They also introduce multivariate approaches to imaging-based phenotyping, as early cha p ter 1 : N euroimaging G enetics
studies of single-locus models have changed in both the genomic and the neuroimaging fields. The association of genotype with phenotype in any genetic association study assumes that what is called a phenotype is indeed an observable characteristic of a genetic trait. In imaging genetics, the apparent heritability of imaging measures has been explored in only a relatively few instances. In Chapter 10 of this volume, Stein, Hibar, and Thompson highlight this with respect to studies of structural imaging, noting that while heritability estimates of brain and cortical volume, based primarily on comparisons of monozygotic (MZ) to dizygotic (DZ) twins are relatively high, estimates of other structural measures, like DTI and measures of cortical thickness, can be much lower. They appropriately caution that methodological error may contribute to these differences in heritability measurements. Small structures are, in general, more difficult to measure accurately, and brain structures that deviate further from stereotactic landmarks used to group scans for statistical analyses will require more adjustment in DZ than in MZ twins because of shape and other deformations, potentially corrupting measures of heritability. Imaging genetics studies of the role of common genetic variation in complex traits, like analogous strategies for gene discovery in complex illnesses, must recognize the challenges of likely small effect associations across diverse and heterogeneous samples. One approach to this problem, which has become de rigueur in disease genetics, is to amass very large samples in collaborative consortia and to perform meta-analyses of the various datasets. The ENIGMA consortium initiated and described by Stein and colleagues in Chapter10 is the current state-of-the-art consortium for genetic studies of large-sample structural imaging data sets. Imaging genetics begins with the assumption that imaging measures are heritable. There is still relatively little demonstration of this for most of the commonly performed imaging studies, including studies of various cortical and subcortical structures, and of functional phenotypes. It is well established that brain volume is highly heritable, but more subtle regional measures are less clearly so. The standard practice of comparing quantitative measures of brain structure in healthy MZ and DZ twins is potentially confounded by the greater shape similarities of MZ twins of a pair, making the imaging technique of spatial averaging biased in these comparisons. Such methodological uncertainties are likely to be especially problematic in studies comparing patient samples to healthy control samples, and also in comparing relatives of patient samples to healthy samples. 5
Candidate gene studies have been criticized for not achieving genome-wide statistical significance in virtually all studies, but the question of whether they are hypothesis driven to the extent that such statistical correction is unjustified is an area of active debate. Chapter 11 by Zink on the imaging genetics of DA system genes and reward processing illustrates both the appeal of hypothesis-driven studies and the conundrums associated with them. Many of the genetic variants implicated in DA imaging studies have been validated in biological studies as influencing the activity of brain DA signaling, making hypothesis-driven study both possible and lawful. However, the results of most of the studies in this realm, as she notes, are at times consistent with what is known about DA and reward circuitry in the brain, but at other times are inconsistent. Genetic imaging studies of the experience of reward also highlight that context matters in terms of clinical samples. So, as Zink emphasizes, genes implicated in reward processing can be associated with reward responses specifically related to illness or behavioral disorders, such as obesity and drug and alcohol abuse, and the directionality can vary depending on the value of the rewarding stimulus in these various contexts. This again may illustrate how population-based studies differ in their associations with genes that influence reward circuitry, or even cognitive and emotional circuitries, as population samples may vary in terms of these components. Finally, Zink points out the obvious, that lumping all reward imaging paradigms into a reward phenotype for genetic association is both simplistic and misguided. This is undoubtedly true of phenotypes related to other imaging paradigms, as reward is a complex neural function that engages a complex brain circuitry, as the responses are critically dependent on a number of characteristics of an individual, the stimulation and demands of the paradigm, and the genetic background. I M AG I N G G E N E T I C S A N D D I S E A S E As Lancaster and colleagues illustrate in Chapter 12, studies of relatives of patients with schizophrenia show varying imaging abnormalities depending on the specific study, but there seems to be virtually no finding seen with imaging in association with schizophrenia that has not been reported in at least one study of relatives of patients with schizophrenia. It is difficult to not have concerns about the potential for artifacts to confound these studies, especially because, unless relatives are very carefully screened and matched with controls on various behaviors and characteristics that also affect imaging measures, the findings are difficult to interpret.
There is advocacy for the translation of basic imaging genetics research into systematic programs for improving quality of life by leveraging the target pathways identified into opportunities for early intervention and prevention. Indeed, identifying predictive biomarkers of risk and resilience is a lofty goal of much imaging genetics research. This clinical goal is clear in Chapter 13 by Sacchet et al. on major depression, Chapter 14 by Fredrikson on anxiety disorders, and Chapter 15 by Andreassen and Tesli on bipolar disorder. In identifying specific neurobiological pathways through which genetic risk may be expressed in these disorders, imaging genetics is slowly but surely shedding light on not only disease etiology and pathophysiology, but also on the next generation of targets for developing more effective treatment and, ultimately, prevention. Imaging genetics is positioned to similarly advance first our understanding and subsequently treatment of neurological disorders, including Parkinson’s disease and Huntington’s disease, as described in Chapter 16 by Robbins, Rowe, and Barker. In Chapter 17, Dorfman and Buckholtz take on one of the more challenging and controversial issues in behavioral science, the degree to which biological factors can be associated with antisocial behavior. They argue that the structural imaging literature is relatively consistent in showing volume reductions in fronto-limbic regions in subjects with these characteristics. These results echo archival claims from phrenological studies and from anatomical studies of the early twentieth century of frontal lobe biological handicaps in such individuals. The question of whether these so-called structural imaging results are really of a structural nature is still, in our view, unresolved. The problem of physiological confounders in structural imaging, mentioned earlier, is difficult to exclude in populations with the social histories of antisocial study samples. An analogous challenge occurs in trying to interpret the functional imaging studies that fairly consistently show that antisocial individuals have a less active amygdala response and amygdala coupling during emotional stimuli, altered reward-related activity, and attentional system engagement. Studies from decades ago showed that such subjects do not have as robust changes in blood pressure, heart rate, or skin conductance during similar exposures. The nagging conundrum here, as with many functional imaging studies, is “which is the cart and which is the horse?” The genetic associations to antisocial behavior are remarkably consistent in some respects, despite inconsistencies in others. This apparent paradox may illustrate an important lesson for complex genetics studies. The convergence of data suggesting that altered serotonin signaling is associated with antisocial behavior, particularly impulsive
6 cha p ter 1 : N euroimaging G enetics
antisociality, is impressive. While individual samples may vary in the specific gene implicated and sometimes in the directionality of the association, the role of genetic variation in the serotonin-signaling molecular machinery is difficult to dismiss. I M AG I N G G E N E T I C S A N D T H E E N V I RO N M E N T The world of imaging genetics, like the world of clinical genetics, is devoting considerable attention to genes that emerge from large population-based GWA studies because the statistics are robust. In general, these studies take a single-locus, additive approach to genetic association, and with studies involving many tens of thousands of subjects, power is sufficient to find association that survives statistical correction for testing all common variants in the genome. As statistically compelling as these findings are, the effect sizes across diverse and heterogeneous populations are very small, with odds ratios usually less than 1.2. This raises the question of whether in certain contexts (i.e., specific genomes or environmental events that may also influence risk) the genetic effects can be exaggerated or diminished. This is the principle of gene-by-environment (GxE) interactions, which is extensively discussed in Chapter 18 by Hyde, Bogdan, and Hariri. While current data about GxE interactions in higher order brain function is very limited, as Hyde and colleagues point out, the evidence thus far is surprisingly tantalizing in suggesting that environmental variation can have large interactive effects with specific genes, potentially explaining some of the weak effects of individual loci and also some of the inconsistencies in the literature restricted to single-locus genetic association. Hyde et al. also emphasize in Chapter 18 the complexity of measurement of the environment and the likelihood that inconsistencies in the literature can reflect, at least in part, this complexity, as well as the complex nature of the interactions themselves. The complexity of behavioral diagnosis and the inconsistencies that arise from this complexity are also critical issues that plague the current behavioral genetics literature and especially the literature about potential GxE interactions. Approaches to modeling GxE interactions also may be critical in observing these effects, as indirect and mediating influences of individual factors may complicate any attempt to show direct influences. A gene may affect brain function only in a specific context, such as during a specific time of life or only under select experiential events. The potential levels of complexity of modeling gene and environment interactions are challenging but are cha p ter 1 : N euroimaging G enetics
likely to be necessary if we are to understand how genetic risk at the population level is translated into pathogenesis at the individual subject level. I M AG I N G G E N E T I C S A N D C O G N I T I O N Among its many and varied contributions to our understanding of the biological basis of individual differences, imaging genetics has helped illuminate specific mechanisms shaping the emergence of variability in core cognitive and behavioral processes, including episodic memory, working memory, stress responsiveness, and anxiety. In many cases, imaging genetics studies of cognition and behavior have overlapped considerably with those targeting the modulatory effects of specific molecular signaling pathways on brain function (see Chapters 3, 4, and 5) In this way, imaging genetics has not only revealed important neuromodulatory effects of numerous molecules, but also the relevance of these same signaling mechanisms for variability in cognition and behavior supported by the common functional brain architecture targeted across these studies. Consequently, such imaging genetics research has provided a foundation from which to investigate etiologic and pathophysiologic mechanisms of diseases associated with dysfunction in these core processes. Several chapters in the volume provide an illustration of the synergy between understanding normal and abnormal processes made possible through imaging genetics. In Chapter 19, Rasch, Erk, Papassotiropoulos, and de Quervain provide an overview of imaging genetics studies of episodic memory. Broadly, they highlight how imaging genetics has provided important mechanistic insights by demonstrating that polymorphisms in genes involved in multiple signaling pathways (e.g., BDNF, COMT, HTR2A) associated with individual differences in episodic memory are also associated with variability in the functional response of the hippocampal formation and the prefrontal cortex, which play critical roles in encoding and retrieval of declarative memories. More pointedly, they detail a series of studies with the gene WWC1 (KIBRA), demonstrating how imaging genetics can provide a “missing link” between initial studies establishing an association between a candidate gene and a behavior, and specific biological mechanisms through which the gene-behavior associations may emerge. In the imaging genetics studies of KIBRA and two other genes, CAMTA1 and PRNP, increased memory-related hippocampal activity was associated with alleles predicting poorer overall episodic memory performance. 7
In Chapter 20, Lett, Brandl, Müller, Heinz, and Walter note similar patterns in their review of imaging genetics studies of working memory. Before surveying the existing literature reporting associations between functional variants in specific candidate genes and individual differences in working memory-related brain function, the authors provide a useful discussion of the general heritability of working memory and related neural phenotypes, including gray matter volume and white matter connectivity of the dorsolateral prefrontal cortex (DLPFC). Such estimates of the overall genetic contribution to variability in a phenotype can usefully inform the search for specific genetic variants by providing an upper boundary for the expected variance. Unfortunately, the heritability of many neural phenotypes, especially the diverse functional measures that have been explored with fMRI, is not fully articulated, leaving uncertainty in the utility of identifying variants accounting for a small amount of overall variance. Certainly, extending research on the broad heritability of target neural phenotypes will be of great value in imaging genetics. Of the many fMRI measures of cortical function, several have been shown to be familial and presumably heritable. These include prefrontal engagement during working memory and hippocampal engagement during episodic memory (Rasetti and Weinberger 2011). The authors’ review of candidate genetic variants within dopaminergic (e.g., COMT), glutamatergic (e.g., DTNBP1), and GABAergic (e.g., GAD1) pathways, as well as novel genes (e.g., CACNA1C) identified from GWA studies of related phenotypes, including the clinical syndromes of bipolar disorder and schizophrenia, reveals that alleles associated with poorer working memory or risk for disorder also generally map onto relatively exaggerated activity in underlying neural regions, most notably the DLPFC during working memory tasks. Interestingly, such relatively increased activity in the presence of normal or poorer episodic or working memory, which is generally interpreted as less efficient information processing as circuits are compromised, has been observed in normal aging as well as in premorbid stages of diseases characterized by memory loss, including Alzheimer’s disease and mild cognitive impairment.
IMAGING GENETICS OF ADVANCING AGE Muse, Nowrangi, Weinberger, and Mattay in Chapter 21 provide a review of imaging genetics research into trajectories of cognitive aging. As part of their synthesis of a vast literature, they describe how variation in episodic memory-related genes including KIBRA and BDNF, as
well as working memory-related genes including COMT, map onto individual differences in age-related memory and cognitive decline. Variation in these genes also predicts relative preservation of brain structure in the form of gray matter volumes and white matter fractional anisotropy with age. Muse et al. also discuss the moderation of age-related changes in cognition and underlying neural circuit structure and function by a common functional polymorphism in APOE, which has been of central importance in the imaging genetics of Alzheimer’s disease (AD). In Chapter 22, reviewing the imaging genetics of AD, Harrison, Burggren, and Bookheimer provide a detailed review of the replicated association between the ε4 allele of APOE, which confers significant risk for AD, with inefficient (i.e., relatively exaggerated) activity of the hippocampal formation and prefrontal cortex during declarative memory tasks. In fact, the seminal study of Bookheimer et al. (2000) demonstrating exaggerated hippocampal formation activity during episodic memory encoding in asymptomatic carriers of the ε4 allele has stood the test of time as a valid observation of a gene effect in the living brain, which launched the imaging genetics era. Similar to the imaging genetics research in Williams syndrome (Chapter 8), Harrison et al. also discuss the growing interest in using Down syndrome (DS), where there is a significantly higher occurrence of AD than in the general population, as a model through which novel age-related genetic variation can be identified. As Down syndrome already shares an abnormality of the APP (amyloid precursor protein) gene linked with familial, autosomal dominant AD, additional investigation of genes located on chromosome 21, which is duplicated in DS, may reveal novel variants affecting aging and cognition. Finally, Harrison et al. highlight the growing importance of neural phenotypes associated with AD in driving the discovery of new risk genes. Through coordinated efforts such as those represented by the Alzheimer’s Disease Neuroimaging Initiative (ADNI), generating large neuroimaging data sets in AD cases and controls, neural phenotypes such as decreased gray matter in the hippocampal formation are proving to be useful in discovering novel gene variants contributing to both normal and abnormal aging. I M AG I N G G E N E T I C S A N D M U LT I - L O C U S M O DE L S An important extension of imaging genetics research identifying associations between individual loci and neural phenotypes, as highlighted in the chapters reviewed earlier, is the construction of multi-locus profile scores, which
8 cha p ter 1 : N euroimaging G enetics
attempt to capture a greater proportion of genetically driven variability in a candidate signaling pathway or across the genome. For example, recent studies have reported larger effects for a multi-locus profile score of variability in dopamine signaling on reward-related brain function than any of the individual loci used to construct the profile (Nikolova et al. 2011; Stice et al. 2012). Here and in similar studies (Pearson-Fuhrhop et al. 2013; Pearson-Fuhrhop et al. 2014; Davis and Loxton 2013; Foshee et al. 2014) the multi-locus profile is constructed based on prior biological evidence, including findings from imaging genetics, for a directional effect (i.e., greater or lesser) of each locus on the magnitude of signaling. A similar approach is described in Chapter 19 by Rasch et al. for a multi-locus profile of excitatory neural signaling on episodic memory and hippocampal activity. Exemplary of this multi-locus approach, Chapter 23 by Corral-Frias Michalski, Di Iorio, and Bogdan reviews ongoing efforts to model individual differences in the responsiveness of the hypothalamic-pituitary-adrenal (HPA) axis by constructing biologically informed multi-locus genetic profiles. As detailed in their chapter, this approach has already produced interesting results, with one study linking a biologically informed 10-SNP HPA axis profile with elevated cortisol reactivity to an acute stressor and decreased limbic brain gray matter as a function of early life stress. Here, the review of Corral-Frias et al. dovetails nicely with Chapter 18 by Hyde et al., encouraging the examination of gene-environment interactions on the emergence of individual differences in imaging phenotypes. Corral-Frias et al. also review two examples of how multi-locus genetic profiles can be constructed, based not on prior candidate gene studies of directional biological effects but on results of discovery-based association studies of clinical or intermediate biological phenotypes leading to the identification of “risk” alleles, which then can be differentially weighted to create a broader “risk” profile. As the authors diligently remark, much of this work, like that in other imaging genetics research, awaits replication before related findings can be considered in the context of treatment or prevention of psychopathology. Nevertheless, the use of multi-locus profiles holds considerable promise in imaging genetics research seeking to understand the impact of variability in specific signaling pathways on the brain or to identify risk for illness. I M AG I N G G E N E T I C S A N D F U T U R E A PP L I C AT I O N S As reviewed in this book, the future of neuroimaging genetics research will include more complex genetic cha p ter 1 : N euroimaging G enetics
models, including the multi-locus approach described in the preceding section, also known as a polygene risk score, as well as more in-depth studies on epigenetics and gene-byenvironment interactions. As discussed in Chapter 2, the development of better molecular neuroimaging probes and probes that aim to measure novel molecular processes will be critical in future PET and SPECT studies. As described in Chapter 11, fMRI task design will be critical for understanding complex brain circuitries and would provide more congruent interpretations of imaging genetics results. Multi-model imaging tools should be increasingly exploited to understand the impact of specific genetic variants at an individual level, but some imaging techniques should be used with caution. Thomas and colleagues examine diffusion tensor imaging (DTI) or diffusion-weighted imaging (DWI) MRI as a method of tractography (Thomas et al. 2014). They show that the accuracy of DWI is limited by technical factors, including noise, artifacts, and data undersampling. Their data on the macaque brain show that even with high-quality DWI data, current methodologies do not result in high anatomical accuracy. Because DTI/ DWI is a highly derived measure based on voxel-averaged estimates of local fiber orientation, it is difficult to determine long-range anatomical projections. This makes DTI data even more difficult to interpret than structural MRI in a non-lesioned brain. Buckner and colleagues review intrinsic functional connectivity MRI (fcMRI) of the resting brain (Buckner, Krienen, and Yeo 2013). They show that slow (< 1 Hz) intrinsic fluctuations in hemodynamics can be measured by fMRI and then simply correlated with fluctuations in other brain regions. However, they agree that resting-state functional connectivity measures cannot be used as a direct proxy for anatomical connectivity because of technical artifacts that confound interpretation. Functional correlations exist between regions that are not directly connected, and coupling is modulated by tasks and recent experience, and is dynamic within an individual over time. Additionally, these measures are confounded by head movement, but also physiological artifacts such as cardiac and respiratory rhythms. Smith and colleagues believe that there is no default mode network, as others believe can be measured during a resting-state fMRI (Smith et al. 2012). Instead, they suggest that there are multiple “temporal functional modes,” which are functionally distinct and spatially overlapping (Smith et al. 2012). They are likely a concatenation of multiple transient mental states that vary across individuals, based on the experience of the MRI environment and the mental state of the subject, which makes it very difficult to interpret the mechanisms associated with resting-state 9
findings. Many of the published resting-state associations are undoubtedly related to the systematic variations in the experience of resting state in individuals. More sophisticated statistical analyses will be necessary as genetic and brain circuitries models become dated with the publication of more complex models. Dynamic causal modeling (DCM) is a method to model functional and directional connectivity in fMRI data (Friston, Harisson, and Penny 2003). DCM uses Bayesian modeling to provide quantifiable estimates of the effective strength of synaptic connections among neuronal populations and their context-dependent modulation (i.e., directionality between connections), as reviewed by Stephan et al. (2010). DCM and other statistical modeling techniques, such as structural equation modeling, will contribute to a better understanding of the structural and functional connectivity in human brain. In a recent review of the state of imaging genetics, Carter et al. give recommendations to enhance the informativeness and replicability of imaging genetics studies (Carter et al. in press), including (1) data-driven selection of candidate genes and imaging measures, (2) informed design and analysis including correcting for multiple comparisons, and (3) independent replications and data-sharing tools. However, Meyer-Lindenberg et al. show that after using a corrected p-value threshold (FWE or FDR p TC > CC) (Hirvonen et al. 2004). The DRD2 C957T SNP is a synonymous polymorphism, meaning that although this variant is located within the coding region of the DRD2 gene, it does not affect the amino acid sequence of the D2 receptor; but a study of its effects in vitro indicate that the T-allele is associated with decreased mRNA stability, translation, and activity-induced up-regulation of DRD2 expression (Duan et al. 2003). A later study from the same group reported that this polymorphism was associated with significant differences in [11C]raclopride K D such that the C-allele was associated with increased K D (i.e., CC > TC > TT) while no significant differences in Bavail were reported (Hirvonen et al. 2009). That is, group differences in [11C]raclopride affinity for D2 appeared to mediate the observed group differences in D2-binding potential. In terms of binding potential, an increase in K D in the C-allele would increase the value of the denominator (remember that binding potential is proportional to Bavail/K D), leading to decreased dopamine D2-binding potential. The authors interpreted this K D-specific effect as the result of this genetic polymorphism possibly affecting dopamine tone.
Acquiring two PET scans within a single cohort with the aim of estimating Bavail and K D has its advantages, such as the ability to determine meaningful information that cannot be acquired with a single PET scan (e.g., estimates of Bavail and K D), but disadvantages include financial and practical constraints related to the acquisition of two PET scans per participant. In the case of the DRD2 C957T polymorphism, in vitro studies suggest that the T allele should be associated with decreased translation, resulting in reduced D2 Bavail. However, the authors reported that differences in binding potential were primarily due to differences in K D. Animal models or in vitro studies evaluating direct effects of genetic variation on transcription, translation, post-transcriptional modification, or other mechanisms are invaluable guides, but the function of these neurobiological systems in humans in vivo is complexly regulated by additional mechanisms that may modulate the effect of genetic variants observed elsewhere. Molecular neuroimaging genetics is the best available methodology for evaluating genetic effects on molecular mechanisms in humans in vivo, and the acquisition of two PET scans within a single cohort for the purpose of estimating Bavail and K D can link genetic effects with these two specific outcome measures. However, an obvious limitation of this approach is that it becomes difficult to acquire sufficiently large sample sizes. One of the largest molecular neuroimaging genetic studies to date included a sample size of 133 individuals (Klein et al. 2010), but most studies are typically smaller than 60–70 individual data sets. The collection of two scans in half as many participants may significantly limit the power of studies to detect potentially subtle genetic effects. Thus, while the acquisition of two PET scans is appealing in the context of developing our understanding of how genetic variants shape variability in molecular mechanisms in vivo, its benefits and limitations need to be carefully considered when determining a study design. GENE TIC SOURCES OF VARIABILIT Y IN OPIOD NEUROTRANSMISSION
The dopamine system is unique in that it is a neurotransmitter system for which there is a well-validated and commonly used radiotracer for measuring endogenous neurotransmitter release (Laruelle 2000). In the case of measuring endogenous serotonin neurotransmission, recent studies provide preliminary evidence for some PET radiotracers that may be sensitive to endogenous release, as reviewed in Paterson et al. (2010) and others (Sibon et al. 2008; Finnema et al. 2010; Pinborg et al. 2012; Selvaraj et al. 2012), which would provide an interesting opportunity to evaluate genetic sources
2 2 Part I : I maging G enetics and N eurochemistry
of variability in endogenous serotonin signaling. Opioid neurotransmission is another system for which there is some evidence for the capacity to measure endogenous release using molecular neuroimaging. Activation of opioid receptors, through a family of endogenous neurotransmitters (including endorphins) and exogenous activators (including morphine), can have profound effects on behavior and the perception of physical or emotional pain (Ribeiro et al. 2005). A molecular neuroimaging genetics study evaluated the association between μ-opioid receptor binding, assessed with [11C] carfentanil PET, and a genetic variant within the gene coding for catechol-O-methyl-tranferase (COMT) (Zubieta et al. 2003). COMT is an enzyme that degrades catecholamines, including dopamine and norepinephrine, and variation in catecholamine signaling has been associated with variation in endogenous opioid signaling (George and Kertesz 1987). An SNP within the COMT gene (rs4680) results in an amino acid switch from valine → methionine at position 158 (Val158Met). The Met allele of this variant is associated with decreased enzymatic activity, negatively affecting the capacity to regulate dopamine signaling (Lotta et al. 1995). In a cohort of healthy individuals, Zubieta and colleagues reported that a sustained pain protocol (infusion of hypertonic saline into a jaw muscle) affected [11C] carfentanil binding in the nucleus accumbens in a manner dependent on COMT Val158Met genotype status (Zubieta et al. 2003). Specifically, [11C]carfentanil binding increased approximately 15%–20% in Val/Val carriers, did not change in Val/Met carriers, and decreased approximately 15% in Met/Met carriers. The authors interpreted their finding as evidence that genetic variation affecting the capacity to regulate catecholamine signaling is associated with variability in pain-induced opioid neurotransmission. The authors also noted a significant effect of COMT Val158Met on [11C]carfentanil binding in the nucleus accumbens at baseline, prior to pain stimulation, such that the Val allele was associated with reduced binding (i.e., Val/Val < Val/Met < Met/Met). As mentioned previously, without a measure of receptor occupancy, it is impossible to dissociate the effects of differences in receptor availability and differences in endogenous neurotransmitter occupancy on the specific binding of radiotracers that are sensitive to endogenous neurotransmitter release. Thus, it is not clear whether COMT Val158Met confers differences in endogenous opioid levels in a baseline state or in response to pain stimulation. However, these results provide an intriguing example of how molecular neuroimaging genetics can reveal effects of genetic variation that putatively affect one neurotransmitter system on molecular signaling of another neurotransmitter system. cha p ter 2 : M olecular N euroimaging G enetics
A more recent study reported that an SNP (A118G) within the gene coding for the μ-opioid receptor (OPRM1) was significantly predictive of [11C]carfentanil binding (Ray et al. 2011). The authors reported that carriers of the G-allele, which in mice is associated with decreased μ-opioid mRNA expression and receptor protein (Mague et al. 2009), showed evidence for reduced [11C]carfentanil binding within anterior cingulate cortex, amygdala, caudate, and thalamus, but not nucleus accumbens. The authors reported significant effects when considering only smokers, but the effect of OPRM1 genotype status appeared to be similar for non-smoking healthy controls, too. As mentioned earlier, if one assumes that [11C]carfentanil binding is sensitive to endogenous opioid neurotransmission, then it is not clear whether these observed differences reflect differences in receptor availability or differences in μ-opioid receptor occupancy. This is an example where the acquisition of two PET scans at baseline, one with high and low specific radioactivity, would offer the opportunity to estimate Bavail and K D independently and to evaluate the source of differences in [11C]carfentanil binding in greater detail. Regardless, these two studies provide intriguing examples of using molecular neuroimaging genetics to link inter-individual variability in opioid neurotransmission with genetic variants that putatively affect opioid signaling both directly (OPRM1 A118G) and indirectly (COMT Val158Met). GENE TIC EFFECTS ON TSPO BINDING AFFINIT Y AND BINDING POTENTIAL
As we described earlier, the affinity of a radiotracer for its target protein is described by the inverse of its dissociation constant (1/K D) and is a proportional factor in computing binding potential. Translocator protein (TSPO) 18 kDa (kDa is an abbreviation for the weight unit, kilo-Daltons), previously known as the peripheral benzodiazepine receptor, is a putative marker of neuroinflammation and microglial activation, which are hypothesized to be relevant in the pathophysiology of neurological illnesses including schizophrenia, dementia, and Alzheimer’s disease (Cagnin et al. 2004; Venneti, Lopresti, and Wiley 2006; Edison et al. 2008). [11C]PBR928 and other PET radiotracers were developed for the purpose of measuring TSPO levels in vivo but substantial variability in binding across participants was identified, with some participants showing almost negligible specific binding (Kreisl et al. 2010). It was recently identified that a common SNP (rs6971) involving a C → T mutation within the gene coding for TSPO (TSPO) results in an amino acid switch from alanine (Ala) 23
to threonine (Thr) at position 147 and significantly reduces binding affinity (measured in blood platelets) such that Ala/Ala homozygotes showed 80% higher binding than Ala/Thr individuals, and Thr/Thr individuals show negligible specific [11C]PBR928 binding (Owen et al. 2012; Kreisl et al. 2013). Similar group differences in TSPO binding have been observed for other PET radiotracers and have contributed to efforts aimed at modeling such effects (Guo et al. 2012; Mizrahi et al. 2012). This example highlights the dramatic effects that a genetic variant within the coding region of a gene can have on binding potential by directly affecting radiotracer binding affinity. The minor allele frequency (0.3 in Caucasians) is high enough that this polymorphism must be considered when performing PET studies measuring TSPO binding. Additionally, there is emerging evidence that this polymorphism affects de novo production of neurosteroids and anxiety (Costa et al. 2009; Costa et al. 2012). For this specific example, it appears that differences in binding due to this genetic polymorphism can be effectively accounted for in statistical models and that doing so improves the ability to identify differences in [11C]PBR928 binding between healthy and schizophrenic individuals (Kreisl et al. 2013). E XPAND THE MODEL OF A GENE TIC POLY MORPHISM: 5 -H T T LPR AND 5 - H T 4 B I N D I N G
A canonical model for how genetic variation affects signaling is through direct regulation of expression or function of the gene with which the polymorphism is associated. However, these direct effects can have modulatory effects on related aspects of the same neurotransmitter system and possibly additional neurotransmitter systems. The 5-HT4 receptor is expressed in brain regions including striatum, neocortex, and hippocampus and has been associated with memory function as well as regulation of serotonin release (Lucas et al. 2005; Haahr et al. 2012). Our lab recently reported that neocortical 5-HT4 binding, assessed with [11C]SB207145 PET, was 9% higher in LA/LA individuals compared to LG and S carriers of the commonly studied tri-allelic 5-HTTLPR polymorphism (including rs25531) within the gene coding for the serotonin transporter (SLC6A4) (Fisher et al. 2012). This finding is consistent with a model based on the in vitro effects of the 5-HTTLPR (i.e., the LG- and S-alleles are associated with reduced transcription of the serotonin transporter and elevated serotonin levels) and studies in animals indicating that 5-HT4 levels are inversely related to serotonin levels (Vidal et al. 2009; Licht, Knudsen, and Sharp 2010; Jennings et al. 2011). In the context of imaging genetics studies, genetic
polymorphisms are useful proxies for modeling differences in neuromolecular signaling mechanisms, which in turn may bias related aspects of brain function or behavior. Molecular neuroimaging genetics provides a unique capacity to develop our interpretation of how genetic polymorphisms affect related molecular mechanisms by providing in vivo evidence for its effects on many molecular mechanisms within a neurotransmitter system or perhaps across multiple neurotransmitter systems. In the case of the 5-HTTLPR, it is likely that any direct functional effects on 5-HTT expression, regulation, or function will contribute to variability in serotonin receptor expression profiles. Thoroughly evaluating receptor systems affected by the 5-HTTLPR benefits its usefulness to model differences in many aspects of serotonin signaling. EPIGENETICS
The role of epigenetic mechanisms in regulating gene function and modulating the effects of genetic polymorphisms is a new frontier in understanding how variation in genes contributes to variability in complex behaviors and related risk for illness. A recent study highlights its application in the field of molecular neuroimaging. Wang and colleagues evaluated the effects of the 5-HTTLPR polymorphism, within the SLC6A4 gene that codes for the 5-HTT, on serotonin synthesis, assessed with [11C]AMT PET (Wang et al. 2012). [11C]AMT is a radiotracer that mimics the uptake of tryptophan (precursor to serotonin) into neurons where it is believed to become irreversibly trapped (Chugani and Muzik 2000). Due to characteristics of this specific radiotracer, it does not reflect binding potential but trapping rate, termed K*, which can be estimated in PET scans with radiotracers that have irreversible kinetics (Chugani and Muzik 2000). Wang and colleagues reported no effect of the 5-HTTLPR on serotonin synthesis (Wang et al. 2012). However, they found that methylation at so-called “CpG” sites within the promoter region of the SLC6A4 gene was inversely associated with serotonin synthesis in vivo ([11C]AMT K*) and suppressed SLC6A4 transcription in vitro. That is, putatively decreased 5-HTT gene function due to hyper-methylation of the SLC6A4 promoter region is associated with decreased serotonin synthesis, assessed with [11C]AMT PET. The authors also reported that methylation was positively associated with high childhood-limited aggression, suggesting that this epigenetic regulation may be linked to aggressive behavioral phenotypes (Wang et al. 2012). Methylation was assessed in peripheral white blood cells (i.e., monocytes and T-cells), which were assumed to reflect genetic methylation patterns in neurons.
2 4 Part I : I maging G enetics and N eurochemistry
The application of epigenetics in the context of molecular neuroimaging is certainly in its infancy. Although we are beginning to appreciate the critical role of epigenetic modification in shaping gene expression, an outstanding limitation of this and other similar studies is that the association between methylation in peripheral cells and serotonergic or other neurons is not well understood, and epigenetic regulation including DNA methylation may be a more dynamic process than previously thought (Baker-Andresen, Ratnu, and Bredy 2013). A recent study detailing how some of the complex interactions between genetic variation and epigenetic modification can in turn affect transcription of the gene coding for the 5-HT2A receptor (HTR2A) underscores the potentially important role played by epigenetic factors in shaping gene function (Falkenberg et al. 2010). As our understanding of these interactions develops, molecular neuroimaging stands as a potent tool for evaluating the extent to which those effects are observable in vivo and can be linked to subsequent effects on brain function and behavior.
O PP O R T U N I T I E S F O R F U T U R E R E S E A RC H DEVELOPMENT OF NOVEL MOLECUL AR NEUROIMAGING TOOLS
Within this and related chapters we have highlighted the strengths and limitations of molecular neuroimaging genetics in revealing genetic sources of variability in molecular mechanisms that can be observed in vivo. However, the extent to which molecular neuroimaging genetics can be applied is limited by the availability of radiotracers that can be used in humans and that effectively index specific target proteins. For example, the norepinephrine system is known to play a critical role in arousal and attention and has been linked to neuropsychiatric disorders including depression and post-traumatic stress disorder (Callado et al. 1998; Aston-Jones, Rajkowski, and Cohen 1999; Southwick et al. 1999). Nevertheless, there has been limited development of radiotracers suitable for measuring aspects of the norepinephrine system (e.g., α2 receptor binding, norepinephrine transporter binding) in humans, in vivo (Jakobsen et al. 2006; Takano et al. 2008). Obviously, this limits the ability to evaluate noradrenergic signaling in humans, in vivo, and to further evaluate how genetic polymorphisms shape inter-individual variability in this behaviorally relevant neurotransmitter system. Recent studies, including radiotracer validation studies noted earlier, offer promise for the cha p ter 2 : M olecular N euroimaging G enetics
development of suitable radiotracers, which is necessary before genetic sources of variation can be effectively evaluated. Additionally, there is preliminary evidence that [11C] yohimbine, a radiotracer for the noradrenergic α2 receptor, may be susceptible to displacement by endogenous norepinephrine release (Landau, Doudet, and Jakobsen 2012). Cannabanoid (CB) receptors within the brain are activated by endogenous neuromodulators known as “endocannabinoids” (eCB), which can have pronounced effects on behavior, including anxiety (Viveros, Marco, and File 2005). A recent imaging genetics study with fMRI reported that a C385A polymorphism (rs32440, sometimes referred to as Pro129Thr) within the gene coding for fatty-acid amide hydrolase (FAAH), an enzyme that regulates CB receptor signaling by degrading eCBs, was associated with emotion- and reward-related brain function (Hariri et al. 2009). Along with this study, converging evidence suggests that cannabinoid signaling may play a critical role in fear extinction and the pathophysiology of post-traumatic stress disorder (PTSD) (Neumeister 2013). Interestingly, the recent development of novel radiotracers has made it possible to assay CB1 receptor binding and FAAH enzyme availability in humans in vivo with [11C]OMAR and [11C] CURB PET, respectively (Wong et al. 2010; Rusjan et al. 2013). A notable limitation of CB1 radiotracers such as [11C]OMAR is that an effective reference region devoid of specific CB1 binding has not been identified, so the primary outcome measure is regional distribution volume, V T, which reflects the sum of specific and non-specific binding. However, a recent study reported elevated CB1 V T in subcortical and cortical regions in detoxified alcohol patients compared to healthy controls, indicating that this metric may effectively track pathophysiologically relevant variation in CB1 signaling (Neumeister et al. 2012). Based on in vitro studies, the A allele of FAAH C385A is associated with reduced cellular expression of FAAH (Chiang et al. 2004). It will be interesting to see whether PET studies can identify a similar association in vivo with [11C]CURB PET and furthermore whether this polymorphism can be linked to alterations in CB1 binding with [11C]OMAR PET, which would implicate this receptor mechanism in mediating the effects of this polymorphism on threat- and reward-related brain function. Monoamine oxidase A (MAO-A) is a critical regulator of central monoamine levels (Shih, Chen, and Ridd 1999). A common variable number tandem repeat polymorphism (VNTR) of a 30 base-pair sequence within the gene coding for the MAO-A enzyme affects MAOA gene expression such that the 3.5- and 4-repeats show higher expression in cell lines whereas the 3- and 25
5-repeats show lower expression (Sabol, Hu, and Hamer 1998). Based on imaging genetics studies, this polymorphism has been linked to threat- and reward-related brain function in a gene-by-environment manner such that the low-expressing enzyme (i.e., higher monoamine signaling due to diminished degradation), coupled with negative early life experiences, can increase the likelihood of developing aggressive behavioral phenotypes (Buckholtz and Meyer-Lindenberg 2008). Despite this model, one study reported that MAOA VNTR status in a cohort of healthy males was not significantly predictive of MAO-A enzyme activity, assessed with [11C] clorgyline PET (Fowler et al. 2007). In the case of [11C] clorgyline, the primary outcome measure is λk 3, which is a model term that is proportional to the number of catalytically active MAO-A molecules (Fowler et al. 2001). A recent study using another MAO-A PET radiotracer, [11C]harmine, reported that acute tryptophan depletion (i.e., decrease in central serotonin levels) and acute sinemet challenge (i.e., increase in central dopamine levels) significantly decreased and increased [11C]harmine binding, respectively, consistent with a model wherein MAO-A levels reflect monoamine levels in specific brain regions (Sacher et al. 2012). No studies have reported the effects of genotype on [11C]harmine binding, and like CB1 radiotracers, a limitation of this radiotracer is the lack of a valid reference region, so V T represents the primary outcome measure. Thus, there are opportunities to develop our understanding of genetic sources of variability in MAO-A levels, as measured in vivo with PET. Additionally, the MAOA VNTR model suggests that this polymorphism should affect monoaminergic levels (low-expressing alleles associated with higher monoaminergic levels), which could be observed with relevant PET radiotracers. Thus, another avenue for future research could involve evaluating the effects of this polymorphism on PET radiotracers indexing receptor, transporter binding, or other features of monoamine systems. Novel radiotracer development represents a key avenue through which molecular neuroimaging can evolve and benefit future research by allowing an opportunity to index behaviorally relevant molecular processes in vivo with molecular neuroimaging. Beyond the previous examples, there are additional behaviorally relevant molecular systems for which suitable molecular neuroimaging radiotracers are not available in humans. In the context of molecular neuroimaging genetics, novel radiotracer development is essential to evaluate specific molecular mechanisms that mediate the effects of genetic variation on cognition or behavior and may affect risk for neurological illness.
MULTIMODAL NEUROIMAGING GENE TICS
Molecular neuroimaging is the most effective method available for estimating molecular signaling pathways in humans in vivo. Within the imaging genetics framework, outcome measures such as binding potential represent a useful intermediate endophenotype for modeling genetic variation effects on molecular mechanisms that in turn modulate distributed neural pathways that affect complex behaviors. We strongly encourage the consideration of molecular neuroimaging within a multimodal neuroimaging framework, including estimates of brain and distributed circuit function, assessed with fMRI, electroencephalography (EEG), or other tools for assessing behaviorally relevant brain function. Molecular neuroimaging and such neuroimaging tools are complementary methodologies that can more directly link in vivo brain chemistry and brain function, benefiting our understanding of how specific molecular mechanisms shape brain function (Fisher and Hariri 2012). Molecular neuroimaging studies also require high-resolution MRI in order to define specific regions of interest, so there is a compelling opportunity available to acquire fMRI data while already in the MR scanner. With the recent introduction of combined MR/PET scanners, it has become possible to simultaneously image MR-based physiological outcomes and molecular neuroimaging measures of interest. Integrating such data sets with genetic information offers the exciting prospect of more directly linking molecular pathways that mediate the effects of genetic variation on variability in brain function and behavior. Such models would likely require larger sample sizes to remain effectively powered, which is difficult considering practical constraints, but focused multimodal neuroimaging studies based on strong a priori associations (and perhaps targeting recruitment based on genotype information) will benefit the payoff in terms of effectively mapping biological pathways mediating individual differences in behavior and related risks for psychopathology, and possibly informing the development of novel therapeutic targets. MULTITR ACER PROTOCOLS
A strength of molecular neuroimaging is that it provides high-resolution imaging of specific molecular processes in vivo. However, a limitation is that all positron emitting radionuclides result in the generation of a pair of photons at the same specific energy level (511 keV). This means that it is impossible to distinguish signal resulting from one radiotracer compared to another and that multiple
2 6 Part I : I maging G enetics and N eurochemistry
molecular neuroimaging scans within a single cohort must be collected in serial. This creates practical constraints on studies interested in collecting multiple molecular neuroimaging scans. This is unfortunate because most molecular neuroimaging studies collect detailed information about a single receptor or transporter without information about the rest of a neurotransmitter system. For example, in the case of the serotonin system there are many receptor families (e.g., 5-HT1-7) and many subtypes within those families (e.g., 5-HT1A, 5-HT1B, 5-HT1D, etc.) (Barnes and Sharp 1999; Sharp et al. 2007). A high-resolution [11C]DASB PET scan can provide detailed information about 5-HTT binding but complementary information about specific serotonin receptor systems would remain elusive. Although multi-tracer studies suggest that related PET measures are correlated, there is likely unique and possibly complementary information about molecular signaling from individual PET radiotracers (Erritzoe et al. 2010; Tuominen et al. 2013). Multi-tracer protocols, in which multiple molecular neuroimaging scans with various radiotracers are acquired, can provide a more comprehensive picture of a distributed neurotransmitter system. This would allow for the opportunity to evaluate to what degree specific genetic variants shape not only individual features of a neurotransmitter system but distributed aspects of a neurotransmitter system. Statistical models that go beyond relatively common general-linear models—including multivariate statistical modeling techniques such as partial-least squares, latent variable analysis, and machine learning—may reveal features related to multiple components of a neurotransmitter system more efficiently and link these features with genetic variability, which can effectively expand our model of how specific genetic variants shape underlying neurobiology. C O N C L U DI N G R E M A R K S Evaluating the link between genetic variants and related measures of brain function, behavior, or illness is a powerful approach for identifying sources of variability in neurobiological mechanisms related to neuropsychiatric and other illnesses. Effectively linking genetic variants with various outcomes is critically dependent upon a sound model for how a specific genetic variant affects gene expression, protein/enzymatic levels, or other molecular mechanisms. The capacity for molecular neuroimaging PET and SPECT to index these aspects of brain chemistry in humans in vivo grants it a critical position in facilitating our understanding of the molecular mechanisms that mediate the effects of genetic variation on brain function, behavior, and related cha p ter 2 : M olecular N euroimaging G enetics
risk for illness. Molecular neuroimaging genetics offers the opportunity to evaluate putative genetics effects in vivo, revealing novel associations between genetic variants and related molecular mechanisms that may not have been initially evaluated within an in vitro or animal model. This second feature is important because developing our understanding of how a specific genetic variant affects various aspects of neurotransmitter signaling (e.g., receptor function, enzymatic activity, etc.) benefits the capacity to apply that genetic variant as a broader model for differences in neurotransmitter systems and their subsequent effects on brain function and behavior. Additionally, the development of novel radiotracers, including radiotracers sensitive to changes in endogenous neurotransmitter release, offers novel opportunities to expand the application of a genetic polymorphism as a model for differences in neurotransmitter signaling. Similar to other neuroimaging fields, the use of imaging genetics with molecular neuroimaging has developed greatly in recent years. Related findings have provided novel insights into how genetic variability shape inter-individual variability in aspects of brain chemistry, which may in turn affect brain function, behavior, and related risk for illness. A lesson learned from molecular neuroimaging genetics studies is that multiple regulatory pathways control these complex molecular systems. It is the exception rather than the norm for a single “smoking gun” genetic variant to explain a disproportionate degree of the variability in the molecular neuroimaging signal for a specific molecular target. The extent to which genetic variation predicts variability in these measures is likely to be distributed across multiple polymorphisms, possibly covering multiple genes. Additionally, our developing understanding of the mechanisms that shape gene function, including genetic variation, epigenetic phenomena, and post-transcriptional modifications (e.g., alternative splicing), provides a wealth of opportunities to evaluate sources of variability in neurobiological mechanisms. This distributed effect model precipitates the need for sophisticated statistical models that move beyond simply evaluating whether a single genetic variant predicts variability in the regional binding potential for a specific radiotracer. The application of some of these types of models is described elsewhere throughout this book and will undoubtedly guide future key findings from molecular neuroimaging genetics studies. Another requirement of these more sensitive models is larger sample sizes than those typically collected for a single PET or SPECT study. Logistical and financial constraints associated with molecular neuroimaging often limit the sample sizes, but collaboration between research groups—pooling related 27
data sets, or using other means to generate sufficiently large data sets—is critically important for advancing our understanding of how genes shape variability in our underlying neurobiology. We can see clear examples that regulation of gene expression is complex, and as we attempt to more thoroughly map the genetic and environmental factors that shape inter-individual variability in neurobiology it is imperative that we effectively appreciate the complexity of these associations by developing research studies well suited to test these hypotheses. Otherwise, we will be limited in accomplishing the various goals that motivate such research, including understanding how variation emerges, identifying risk for illness, facilitating more effective treatment, and promoting behavioral strategies that help prevent illness onset.
REFERENCES Albert PR, Benkelfat C. (2013). The neurobiology of depression— revisiting the serotonin hypothesis. II. Genetic, epigenetic and clinical studies. Philos Trans R Soc Lond B Biol Sci. 368 (1615): 20120535. Aston-Jones G, Rajkowski J, Cohen J. (1999). Role of locus coeruleus in attention and behavioral flexibility. Biol Psychiatry. 46(9): 1309–1320. Baker-Andresen D, Ratnu VS, Bredy TW. (2013). Dynamic DNA methylation: a prime candidate for genomic metaplasticity and behavioral adaptation. Trends Neurosci. 36(1): 3–13. Barnes NM, Sharp T. (1999). A review of central 5-HT receptors and their function. Neuropharmacology. 38: 1083–1152. Blier P, de Montigny C. (1998). Possible serotonergic mechanisms underlying the antidepressant and anti-obsessive-compulsive disorder responses. Biol Psychiatry. 44: 313–323. Brody AL, Mandelkern MA, Olmstead RE, Scheibal D, Hahn E, Shiraga S, Zamora-Paja E, Farahi J, Saxena S, London ED, McCracken JT. (2006). Gene variants of brain dopamine pathways and smoking-induced dopamine release in the ventral caudate/nucleus accumbens. Arch Gen Psychiatry. 63(7): 808–816. Buckholtz JW, Meyer-Lindenberg A. (2008). MAOA and the neurogenetic architecture of human aggression. Trends Neurosci. 31(3): 120–129. Burghardt PR, Love TM, Stohler CS, Hodgkinson C, Shen PH, Enoch MA, Goldman D, Zubieta JK. (2012). Leptin regulates dopamine responses to sustained stress in humans. J Neurosci. 32(44): 15369– 15376. Cagnin A, Rossor M, Sampson EL, Mackinnon T, Banati RB. (2004). In vivo detection of microglial activation in frontotemporal dementia. Ann Neurol. 56(6): 894–897. Callado LF, Meana JJ, Grijalba B, Pazos A, Sastre M, Garcia-Sevilla JA. (1998). Selective increase of alpha2A-adrenoceptor agonist binding sites in brains of depressed suicide victims. J Neurochem. 70(3): 1114–1123. Carson RE, Daube-Witherspoon ME, Herscovitch P. (1998). Quantitative functional brain imaging with positron emission tomography. San Diego, CA: Academic Press. Chiang KP, Gerber AL, Sipe JC, Cravatt BF. (2004). Reduced cellular expression and activity of the P129T mutant of human fatty acid amide hydrolase: evidence for a link between defects in the endocannabinoid system and problem drug use. Hum Mol Genet. 13(18): 2113–2119.
Chugani DC, Muzik O. (2000). Alpha[C-11]methyl-L-tryptophan PET maps brain serotonin synthesis and kynurenine pathway metabolism. J Cereb Blood Flow Metab. 20(1): 2–9. Cornea-Hebert V, Riad M, Wu C, Singh SK, Descarries L. (1999). Cellular and subcellular distribution of the serotonin 5-HT2A receptor in the central nervous system of adult rat. J Comp Neurol. 409:187–209. Costa B, Pini S, Abelli M, Gabelloni P, Da PE, Chelli B, Calugi S, Lari L, Cardini A, Lucacchini A, Cassano GB, Martini C. (2012). Role of translocator protein (18 kDa) in adult separation anxiety and attachment style in patients with depression. Curr Mol Med. 12(4): 483–487. Costa B, Pini S, Gabelloni P, Da PE, Abelli M, Lari L, Preve M, Lucacchini A, Cassano GB, Martini C. (2009). The spontaneous Ala147Thr amino acid substitution within the translocator protein influences pregnenolone production in lymphomonocytes of healthy individuals. Endocrinology. 150(12): 5438–5445. Czesak M, Lemonde S, Peterson EA, Rogaeva A, Albert PR. (2006). Cell-specific repressor or enhancer activities of Deaf-1 at a serotonin 1A receptor gene polymorphism. J Neurosci. 26: 1864–1871. David SP, Murthy NV, Rabiner EA, Munafo MR, Johnstone EC, Jacob R, Walton RT, Grasby PM. (2005). A functional genetic variation of the serotonin (5-HT) transporter affects 5-HT1A receptor binding in humans. J Neurosci. 25: 2586–2590. Domino EF, Evans CL, Ni L, Guthrie SK, Koeppe RA, Zubieta JK. (2012). Tobacco smoking produces greater striatal dopamine release in G-allele carriers with mu opioid receptor A118G polymorphism. Prog Neuropsychopharmacol Biol Psychiatry. 38(2): 236–240. Duan J, Wainwright MS, Comeron JM, Saitou N, Sanders AR, Gelernter J, Gejman PV. (2003). Synonymous mutations in the human dopamine receptor D2 (DRD2) affect mRNA stability and synthesis of the receptor. Hum Mol Genet. 12(3): 205–216. Edison P, Archer HA, Gerhard A, Hinz R, Pavese N, Turkheimer FE, Hammers A, Tai YF, Fox N, Kennedy A, Rossor M, Brooks DJ. (2008). Microglia, amyloid, and cognition in Alzheimer’s disease: an [11C](R)PK11195-PET and [11C]PIB-PET study. Neurobiol Dis. 32(3): 412–419. Erritzoe D, Holst K, Frokjaer VG, Licht CL, Kalbitzer J, Nielsen FA, Svarer C, Madsen J, Knudsen G. (2010). A nonlinear relationship between cerebral serotonin transporter and 5-HT(2A) receptor binding: an in vivo molecular imaging study in humans. J Neurosci. 30(9): 3391–3397. Falkenberg VR, Gurbaxani BM, Unger ER, Rajeevan MS. (2010). Functional genomics of serotonin receptor 2A (HTR2A): interaction of polymorphism, methylation, expression and disease association. NeuroMolecular Medicine. 13: 66–76. Farde L, Nordstrom AL, Wiesel FA, Pauli S, Halldin C, Sedvall G. (1992). Positron emission tomographic analysis of central D1 and D2 dopamine receptor occupancy in patients treated with classical neuroleptics and clozapine: relation to extrapyramidal side effects. Arch Gen Psychiatry. 49(7): 538–544. Finnema SJ, Varrone A, Hwang TJ, Gulyas B, Pierson ME, Halldin C, Farde L. (2010). Fenfluramine-induced serotonin release decreases [11C]AZ10419369 binding to 5-HT1B-receptors in the primate brain. Synapse. 64(7): 573–577. Fisher PM, Hariri AR. (2012). Linking variability in brain chemistry and circuit function through multimodal human neuroimaging. Genes Brain Behav. 11(6): 633–642. Fisher PM, Holst KK, Mc Mahon B, Haahr MH, Madsen K, Gillings N, Baare WF, Jensen PS, Knudsen GM. (2012). 5-HTTLPR status predictive of neocortical 5-HT(4) binding assessed with [(11)C] SB207145 PET in humans. Neuroimage. 62(1): 130–136. Fisher PM, Muñoz KE, Hariri AR. (2008). Identification of neurogenetic pathways of risk for psychopathology. Am J Med Genet C. 148C: 147–153. Fowler JS, Alia-Klein N, Kriplani A, Logan J, Williams B, Zhu W, Craig IW, Telang F, Goldstein R, Volkow ND, Vaska P, Wang GJ.
2 8 Part I : I maging G enetics and N eurochemistry
(2007). Evidence that brain MAO A activity does not correspond to MAO A genotype in healthy male subjects. Biol Psychiatry. 62(4): 355–358. Fowler JS, Logan J, Ding YS, Franceschi D, Wang GJ, Volkow ND, Pappas N, Schlyer D, Gatley SJ, Alexoff D, Felder C, Biegon A, Zhu W. (2001). Non-MAO A binding of clorgyline in white matter in human brain. J Neurochem. 79(5): 1039–1046. George SR, Kertesz M. (1987). Met-enkephalin concentrations in striatum respond reciprocally to alterations in dopamine neurotransmission. Peptides. 8(3): 487–492. Guo Q, Owen DR, Rabiner EA, Turkheimer FE, Gunn RN. (2012). Identifying improved TSPO PET imaging probes through biomathematics: The impact of multiple TSPO binding sites in vivo. Neuroimage. 60(2): 902–910. Haahr ME, Fisher P, Holst K, Madsen K, Jensen CG, Marner L, Lehel S, Baare W, Knudsen G, Hasselbalch S. (2012). The 5-HT(4) receptor levels in hippocampus correlates inversely with memory test performance in humans. Hum Brain Mapp. 34(11): 3066–3074. Hariri AR, Gorka A, Hyde LW, Kimak M, Halder I, Ducci F, Ferrell RE, Goldman D, Manuck SB. (2009). Divergent effects of genetic variation in endocannabinoid signaling on human threat- and reward-related brain function. Biol Psychiatry. 66(1): 9–16. Hariri AR. (2009). The neurobiology of individual differences in complex behavioral traits. Ann Rev Neurosci. 32: 225–247. Hesselgrave N, Parsey RV. (2013). Imaging the serotonin 1A receptor using [11C]WAY100635 in healthy controls and major depression. Philos T Ro Soc B. 368(1615): 20120004. Hirvonen M, Laakso A, Nagren K, Rinne JO, Pohjalainen T, Hietala J. (2004). C957T polymorphism of the dopamine D2 receptor (DRD2) gene affects striatal DRD2 availability in vivo. Mol Psychiatry. 9(12): 1060–1061. Hirvonen MM, Laakso A, Nagren K, Rinne JO, Pohjalainen T, Hietala J. (2009). C957T polymorphism of dopamine D2 receptor gene affects striatal DRD2 in vivo availability by changing the receptor affinity. Synapse. 63(10): 907–912. Hurlemann R, Matusch A, Kuhn KU, Berning J, Elmenhorst D, Winz O, Kolsch H, Zilles K, Wagner M, Maier W, Bauer A. (2008). 5-HT2A receptor density is decreased in the at-risk mental state. Psychopharmacology (Berl). 195: 579–590. Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, Gunn RN, Holden J, Houle S, Huang SC, Ichise M, Iida H, Ito H, Kimura Y, Koeppe RA, Knudsen GM, Knuuti J, Lammertsma AA, Laruelle M, Logan J, Maguire RP, Mintun MA, Morris ED, Parsey R, Price JC, Slifstein M, Sossi V, Suhara T, Votaw JR, Wong DF, Carson RE. (2007). Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab. 27: 1533–1539. Jakobsen S, Pedersen K, Smith DF, Jensen SB, Munk OL, Cumming P. (2006). Detection of alpha2-adrenergic receptors in brain of living pig with 11C-yohimbine. J Nucl Med. 47(12): 2008–2015. Jennings KA, Licht CL, Bruce A, Lesch KP, Knudsen GM, Sharp T. (2011). Genetic variation in 5-hydroxytryptamine transporter expression causes adaptive changes in 5-HT4 receptor levels. Int J Neuropsychopharmacol. 15(8): 1099–1107. Klein AB, Trajkovska V, Erritzoe D, Haugbol S, Madsen J, Baare W, Aznar S, Knudsen GM. (2010). Cerebral 5-HT2A receptor and serotonin transporter binding in humans are not affected by the val66met BDNF polymorphism status or blood BDNF levels. J Cereb Blood Flow Metab. 30(11): e1–e7. Kraus C, Hahn A, Savli M, Kranz GS, Baldinger P, Hoflich A, Spindelegger C, Ungersboeck J, Haeusler D, Mitterhauser M, Windischberger C, Wadsak W, Kasper S, Lanzenberger R. (2012). Serotonin-1A receptor binding is positively associated with gray matter volume: a multimodal neuroimaging study combining PET and structural MRI. Neuroimage. 63(3): 1091–1098. Kreisl WC, Fujita M, Fujimura Y, Kimura N, Jenko KJ, Kannan P, Hong J, Morse CL, Zoghbi SS, Gladding RL, Jacobson S, Oh U,
cha p ter 2 : M olecular N euroimaging G enetics
Pike VW, Innis RB. (2010). Comparison of [(11)C]-(R)-PK 11195 and [(11)C]PBR28, two radioligands for translocator protein (18 kDa) in human and monkey: Implications for positron emission tomographic imaging of this inflammation biomarker. Neuroimage. 49(4): 2924–2932. Kreisl WC, Jenko KJ, Hines CS, Hyoung LC, Corona W, Morse CL, Zoghbi SS, Hyde T, Kleinman JE, Pike VW, McMahon FJ, Innis RB. (2013). A genetic polymorphism for translocator protein 18 kDa affects both in vitro and in vivo radioligand binding in human brain to this putative biomarker of neuroinflammation. J Cereb Blood Flow Metab. 33(1): 53–58. Landau AM, Doudet DJ, Jakobsen S. (2012). Amphetamine challenge decreases yohimbine binding to alpha2 adrenoceptors in Landrace pig brain. Psychopharmacology (Berl). 222(1): 155–163. Laruelle M. (2000). Imaging synaptic neurotransmission with in vivo binding competition techniques: a critical review. J Cereb Blood Flow Metab. 20: 423–451. Lemonde S, Du L, Bakish D, Hrdina P, Albert PR. (2004). Association of the C(-1019)G 5-HT1A functional promoter polymorphism with antidepressant response. Int J Neuropsychopharmacol. 7: 501–506. Lemonde S, Turecki G, Bakish D, Du L, Hrdina PD, Bown CD, Sequeira A, Kushwaha N, Morris SJ, Basak A, Ou XM, Albert PR. (2003). Impaired repression at a 5-hydroxytryptamine 1A receptor gene polymorphism associated with major depression and suicide. J Neurosci. 23: 8788–8799. Licht CL, Knudsen GM, Sharp T. (2010). Effects of the 5-HT4 receptor agonist RS67333 and paroxetine on hippocampal extracellular 5-HT levels. Neurosci Lett. 476: 58–61. Lothe A, Boni C, Costes N, Bouvard S, Gorwood P, Lavenne F, Alvarez M, Ryvlin P. (2010). 5-HT1A gene promoter polymorphism and [18F]MPPF binding potential in healthy subjects: a PET study. Behav Brain Funct. 6: 37. Lotta T, Vidgren J, Tilgmann C, Ulmanen I, Melen K, Julkunen I, Taskinen J. (1995). Kinetics of human soluble and membranebound catechol O-methyltransferase: a revised mechanism and description of the thermolabile variant of the enzyme. Biochemistry. 34(13): 4202–4210. Lucas G, Compan V, Charnay Y, Neve RL, Nestler EJ, Bockaert J, Barrot M, Debonnel G. (2005). Frontocortical 5-HT4 receptors exert positive feedback on serotonergic activity: viral transfections, subacute and chronic treatments with 5-HT4 agonists. Biol Psychiatry. 57: 918–925. Mague SD, Isiegas C, Huang P, Liu-Chen LY, Lerman C, Blendy JA. (2009). Mouse model of OPRM1 (A118G) polymorphism has sex-specific effects on drug-mediated behavior. Proc Natl Acad Sci U S A. 106(26): 10847–10852. Miner LAH, Backstrom JR, Sanders-Bush E, Sesack SR. (2003). Ultrastructural localization of serotonin2A receptors in the middle layers of the rat prelimbic prefrontal cortex. Neuroscience. 116: 107–117. Mizrahi R, Rusjan PM, Kennedy J, Pollock B, Mulsant B, Suridjan I, De Luca V, Wilson AA, Houle S. (2012). Translocator protein (18[thinsp]kDa) polymorphism (rs6971) explains in-vivo brain binding affinity of the PET radioligand [lsqb]18F[rsqb]-FEPPA. J Cereb Blood Flow Metab. 32(6): 968–972. Neumeister A. (2013). The endocannabinoid system provides an avenue for evidence-based treatment development for PTSD. Depress Anxiety. 30(2): 93–96. Neumeister A, Normandin MD, Murrough JW, Henry S, Bailey CR, Luckenbaugh DA, Tuit K, Zheng MQ, Galatzer-Levy IR, Sinha R, Carson RE, Potenza MN, Huang Y. (2012). Positron emission tomography shows elevated cannabinoid CB (1) receptor binding in men with alcohol dependence. Alcohol Clin Exp Res. 36(12): 2104–2109. Owen DR, Yeo AJ, Gunn RN, Song K, Wadsworth G, Lewis A, Rhodes C, Pulford DJ, Bennacef I, Parker CA, StJean PL, Cardon LR, Mooser VE, Matthews PM, Rabiner EA, Rubio JP. (2012). An 18-kDa translocator protein (TSPO) polymorphism explains
29
differences in binding affinity of the PET radioligand PBR28. J Cereb Blood Flow Metab. 32(1): 1–5. Parsey RV, Ogden RT, Miller JM, Tin A, Hesselgrave N, Goldstein E, Mikhno A, Milak M, Zanderigo F, Sullivan GM, Oquendo MA, Mann JJ. (2010). Higher serotonin 1A binding in a second major depression cohort: modeling and reference region considerations. Biol Psychiatry. 68(2): 170–178. Parsey RV, Olvet DM, Oquendo MA, Huang YY, Ogden RT, Mann JJ. (2006). Higher 5-HT1A receptor binding potential during a major depressive episode predicts poor treatment response: preliminary data from a naturalistic study. Neuropsychopharmacology. 31: 1745–1749. Paterson LM, Tyacke RJ, Nutt DJ, Knudsen GM. (2010). Measuring endogenous 5-HT release by emission tomography: promises and pitfalls. J Cerebr Blood F Met. 30: 1682–1706. Pinborg LH, Feng L, Haahr ME, et al. (2012). No change in [(1)(1)C] CUMI-101 binding to 5-HT(1A) receptors after intravenous citalopram in human. Synapse. 66(10): 880–884. Pinborg LH, Arfan H, Haugbol S, Kyvik KO, Hjelmborg J, Svarer C, Frokjaer VG, Paulson OB, Holm S, Knudsen GM. (2008). The 5-HT2A receptor binding pattern in the human brain is strongly genetically determined. Neuroimage. 40: 1175. Price JC. (2003). Principles of tracer kinetic analysis. Neuroimag Clin N Am. 13: 689–704. Ray R, Ruparel K, Newberg A, Wileyto EP, Loughead JW, Divgi C, Blendy JA, Logan J, Zubieta JK, Lerman C. (2011). Human mu opioid receptor (OPRM1 A118G) polymorphism is associated with brain mu-opioid receptor binding potential in smokers. Proc Natl Acad Sci U S A. 108(22): 9268–9273. Ribeiro SC, Kennedy SE, Smith YR, Stohler CS, Zubieta JK. (2005). Interface of physical and emotional stress regulation through the endogenous opioid system and mu-opioid receptors. Prog Neuropsychopharmacol Biol Psychiatry. 29(8): 1264–1280. Richardson-Jones JW, Craige CP, Guiard BP, Stephen A, Metzger KL, Kung HF, Gardier AM, Dranovsky A, David DJ, Beck SG, Hen R, Leonardo ED. (2010). 5-HT1A autoreceptor levels determine vulnerability to stress and response to antidepressants. Neuron. 65(1): 40–52. Richardson-Jones JW, Craige CP, Nguyen TH, Kung HF, Gardier AM, Dranovsky A, David DJ, Guiard BP, Beck SG, Hen R, Leonardo ED. (2011). Serotonin-1A autoreceptors are necessary and sufficient for the normal formation of circuits underlying innate anxiety. J Neurosci. 31(16): 6008–6018. Rusjan PM, Wilson AA, Mizrahi R, Boileau I, Chavez SE, Lobaugh NJ, Kish SJ, Houle S, Tong J. (2013). Mapping human brain fatty acid amide hydrolase activity with PET. J Cereb Blood Flow Metab. 33(3): 407–414. Sabol SZ, Hu S, Hamer D. (1998). A functional polymorphism in the monoamine oxidase A gene promoter. Hum Genet. 103(3): 273–279. Sacher J, Rabiner EA, Clark M, Rusjan P, Soliman A, Boskovic R, Kish SJ, Wilson AA, Houle S, Meyer JH. (2012). Dynamic, adaptive changes in MAO-A binding after alterations in substrate availability: an in vivo [(11)C]-harmine positron emission tomography study. J Cereb Blood Flow Metab. 32(3): 443–446. Selvaraj S, Turkheimer F, Rosso L, Faulkner P, Mouchlianitis E, Roiser JP, McGuire P, Cowen PJ, Howes O. (2012). Measuring endogenous changes in serotonergic neurotransmission in humans: a [(11)C] CUMI-101 PET challenge study. Mol Psychiatry. 17(12): 1254–1260. Sharp T, Boothman L, Raley J, Queree P. (2007). Important messages in the ‘post’: recent discoveries in 5-HT neurone feedback control. Trends Pharmacol Sci. 28: 629–636. Shih JC, Chen K, Ridd MJ. (1999). Monoamine oxidase: from genes to behavior. Annu Rev Neurosci. 22: 197–217.
Sibon I, Benkelfat C, Gravel P, Aznavour N, Costes N, Mzengeza S, Booij L, Baker G, Soucy JP, Zimmer L, Descarries L. (2008). Decreased [18F]MPPF binding potential in the dorsal raphe nucleus after a single oral dose of fluoxetine: a positron-emission tomography study in healthy volunteers. Biol Psychiatry. 62(12): 1135–1140. Southwick SM, Bremner JD, Rasmusson A, Morgan CA, III, Arnsten A, Charney DS. (1999). Role of norepinephrine in the pathophysiology and treatment of posttraumatic stress disorder. Biol Psychiatry. 46(9): 1192–1204. Takano A, Varrone A, Gulyas B, Karlsson P, Tauscher J, Halldin C. (2008). Mapping of the norepinephrine transporter in the human brain using PET with (S,S)-[18F]FMeNER-D2. Neuroimage. 42(2): 474–482. Thompson P, Cannon TD, Toga AW. (2002). Mapping genetic influences on human brain structure. Ann Med. 34(7–8): 523–536. Tuominen L, Nummenmaa L, Keltikangas-Jarvinen L, Raitakari O, Hietala J. (2013). Mapping neurotransmitter networks with PET: an example on serotonin and opioid systems. Hum Brain Mapp. 35(5): 1875–1884. Urban NB, Slifstein M, Meda S, Xu X, Ayoub R, Medina O, Pearlson GD, Krystal JH, Abi-Dargham A. (2011). Imaging human reward processing with positron emission tomography and functional magnetic resonance imaging. Psychopharmacology (Berl). 221(1): 67–77. Venneti S, Lopresti BJ, Wiley CA. (2006). The peripheral benzodiazepine receptor (Translocator protein 18kDa) in microglia: from pathology to imaging. Prog Neurobiol. 80(6): 308–322. Vidal R, Valdizan EM, Mostany R, Pazos A, Castro E. (2009). Longterm treatment with fluoxetine induces desensitization of 5-HT4 receptor-dependent signalling and functionality in rat brain. J Neurochem. 110(3): 1120–1127. Viveros MP, Marco EM, File SE. (2005). Endocannabinoid system and stress and anxiety responses. Pharmacol Biochem Behav. 81(2): 331–342. Volkow ND, Wang GJ, Fowler JS, Logan J, Schlyer D, Hitzemann R, Lieberman J, Angrist B, Pappas N, MacGregor R. (1994). Imaging endogenous dopamine competition with [11C]raclopride in the human brain. Synapse. 16(4): 255–262. Wang D, Szyf M, Benkelfat C, Provençal N, Turecki G, Caramaschi D, Côté SM, Vitaro F, Tremblay RE, Booij L. (2012). Peripheral SLC6A4 DNA methylation is associated with in vivo measures of human brain serotonin synthesis and childhood physical aggression. PLoS ONE. 7: e39501. Willeit M, Praschak-Rieder N. (2010). Imaging the effects of genetic polymorphisms on radioligand binding in the living human brain: a review on genetic neuroreceptor imaging of monoaminergic systems in psychiatry. Neuroimage. 53: 878–892. Wong DF, Kuwabara H, Horti AG, Raymont V, Brasic J, Guevara M, Ye W, Dannals RF, Ravert HT, Nandi A, Rahmim A, Ming JE, Grachev I, Roy C, Cascella N. (2010). Quantification of cerebral cannabinoid receptors subtype 1 (CB1) in healthy subjects and schizophrenia by the novel PET radioligand [11C]OMAR. Neuroimage. 52(4): 1505–1513. Wright IC, Sham P, Murray RM, Weinberger DR, Bullmore ET. (2002). Genetic contributions to regional variability in human brain structure: methods and preliminary results. Neuroimage. 17(1): 256–271. Zubieta JK, Heitzeg MM, Smith YR, Bueller JA, Xu K, Xu Y, Koeppe RA, Stohler CS, Goldman D. (2003). COMT val158met genotype affects mu-opioid neurotransmitter responses to a pain stressor. Science. 299(5610): 1240–1243.
3 0 Part I : I maging G enetics and N eurochemistry
3. NEURORECEPTOR IMAGING OF GENE TIC VARIATION IMPACTING THE SEROTONIN TRANSPORTER Greg Perlman, Christine DeLorenzo, J. John Mann, and Ramin Parsey
W H Y S T U DY S E RO T O N I N? In humans, the neurotransmitter serotonin (5-hydroxytryptamine, or 5-HT) is linked to a variety of basic biological functions (e.g., sleep, appetite, etc.), complex traits (e.g., mood, cognition, personality, etc.), and psychiatric diseases (e.g., mood disorders, substance abuse disorders, anxiety disorders, etc.). Driven in part by the diversity of behaviors influenced by serotonin, basic neuroscience and applied clinical science have endeavored to describe the structure and the function of the serotonin system in the central and peripheral nervous system. This chapter will describe genetic neuroimaging of the serotonin transporter (5-HTT, SERT, SLC6A4 gene), a key mechanism in the regulation of intra-synaptic serotonin in the human nervous system. Due to its higher resolution, we primarily discuss SERT as imaged by positron emission tomography (PET), which has proven an exceptionally useful tool for in vivo measurement of SERT binding in humans. In the first section, we review the serotonin system with a focus on SERT and the genetics of SERT. In the second section, we review the measurement of SERT using PET methodology. Next, we review general factors that may influence measurement of SERT using PET, such as demographics, genetic variation in the SERT gene, and association with serotonin receptors. In the following section, we briefly discuss findings that link SERT to phenotypes of interest, such as psychiatric disorders and personality. In the final section, we discuss future directions for SERT PET imaging research. Serotonin is a phylogenetically old neurotransmitter evolving 700 million years ago, well before the bifurcation of vertebrae and non-vertebra organisms (Peroutka and Howell 1994). Reflecting its diversified functional role, serotonin can be found in (1) several tissues in the
human body, such as the nervous system, the gastrointestinal tract, and blood platelets; (2) all mammalian species, including humans, mouse, rat, and canine; and (3) most non-vertebra species, such as worms and insects. That means that the many functions of serotonin in humans can be studied in animal models, such as rodents or baboons. In the human brain, all neurons with serotonin receptors originate in the dorsal raphe nuclei (DRN) or median raphe nuclei (MRN) (Visser, van Waarde, et al. 2011). These cell bodies innervate virtually all regions of the human brain, which means that serotonin can be found in virtually all areas of the human brain. Perhaps due to its evolutionary age, there are several distinct classes of serotonin receptors, as well as subclasses (Peroutka 1994; Barnes and Sharp 1999; Hoyer, Hannon et al. 2002). These serotonin receptor classes and subclasses differentially populate brain regions, show distinct patterns of regional density in the brain, can have an inhibitory or excitatory role in regulating neural firing and serotonin release, activate or inhibit several second messenger systems via G- protein coupled receptors, and can be found pre-synaptically or post-synaptically, which presumably allows one neurotransmitter to have so many different behavioral/cognitive effects. While serotonin is generally studied for its role as a neurotransmitter, serotonin also plays a key role in neurodevelopment and in both neurogenesis and enhancing neuronal survival (Whitaker-Azmitia 2001). While the serotonin system is diverse and complex, the role of the SERT in the system is relatively simple. Namely, SERT’s role is the re-uptake of synaptic serotonin. SERT density in the brain varies by brain region, and is located on pre-synaptic nerve terminals, cell bodies, dendrites, and axons in all brain regions innervated by 31
serotonin neurons. Brain regions with the greatest SERT density are the raphe nuclei, the midbrain structures innervated by the raphe nuclei, thalamus, hypothalamus, and striatum. Other subcortical regions, such as hippocampus, amygdala, and cingulate cortex, have an intermediate density of SERT. Regions of the cortex and cerebellum have the lowest density of SERT. Although the cerebellum appears to have the lowest density of SERT in the human brain, it is not devoid of SERT, which can be found in midline nuclei and even in the cerebellar cortex (Cortes, Soriano, et al. 1988; Laruelle, Vanisberg, et al. 1988; Kish, Furukawa, et al. 2005). GENE TICS OF SERT
The last few decades have seen identification of the genes that encode all of the serotonin specific receptors, as well as genes that regulate serotonin synthesis (Murphy, Fox, et al. 2008). The gene encoding the structure of the SERT protein—solute carrier family 6 (neurotransmitter transporter, serotonin), member 4 (SLC6A4)—has been located on chromosome 17 (i.e., 17q11.1-q12; Ramamoorthy, Bauman, et al. 1993). One well-studied polymorphism in SLC6A4 is an insertion/deletion of a 44-base pair repeat found in the 5'-flanking upstream regulatory promoter region. This polymorphism is designated “5-HTTLPR” in the literature. The short 5-HTTLPR allele is generally referred to as the “S” allele, while the longer 5-HTTLPR allele is generally referred to as the “L” allele. The long allele has a polymorphism designated as LA and LG (rs25531), and carriers of the LA variant have greater expression than the carriers of LG and S alleles, which have a similar level of expression (Heils, Teufel, et al. 1996; Lesch, Bengel, et al. 1996; Hu, Lipsky, et al. 2006). Most PET binding potential studies that compared L allele carriers to S allele carriers were confounded by the presence of undetected LG alleles in the “L” allele carrier group (Parsey, Hastings, et al. 2006a). Further exploration has revealed several additional polymorphisms and allele differences in populations of Asians and Europeans (Kunugi, Hattori, et al. 1997; Nakamura, Ueno, et al. 2000; Sakai, Nakamura, et al. 2002), and between Asians, Africans, and Europeans (Gelernter, Kranzler, et al. 1997). Another study has shown that there are different variants of the S allele (Wendland, Martin, et al. 2006). In addition, several variable number tandem repeats (VNTR) have been identified outside the promoter region (Lesch, Balling, et al. 1994; Wendland, Martin, et al. 2006). At least a few VNTR appear to have functional significance (Hranilovic, Stefulj, et al. 2004; Ali, Vasiliou, et al. 2010). Thus, while the greater focus has
been on studying S and LA, and occasionally LG, there are several other identified functional variations in SLC6A4. Several of these variants have yet to be studied using neuroimaging methods, such as PET. Thus, it is unclear how these variants impact binding potential estimates in genetic neuroimaging studies. Several studies have attempted to identify the impact of SLC6A4 gene variants on individual differences in behavior. In humans, the S allele is associated with heightened emotional reactivity, greater risk of mood and anxiety psychopathology (e.g., social phobia, major depressive disorder, seasonal affective disorder, etc.), and elevated personality traits related to neuroticism and fearfulness (Lesch, Bengel, et al. 1996; Sen, Burmeister, et al. 2004; Lasky-Su, Faraone, et al. 2005; Praschak-Rieder, Willeit, et al. 2008). Similarly, cognitive psychology studies have reported evidence that S allele carriers preferentially attend toward negative valance stimuli (Pergamin-Hight, Bakermans-Kranenburg, et al. 2012). These findings from human studies are consistent with findings about SERT and anxiety/depression models in studies of mice. For example, SERT knockout mice exhibit greater anxiety and depression-related behaviors relative to mice with the SERT gene (Lira, Zhou, et al. 2003). Moreover, adult mice treated with paroxetine during early periods of neurodevelopment exhibit greater anxiety/depression-related behaviors on par with that seen in SERT knockout mice (Ansorge, Zhou, et al. 2004). In a seminal study, Caspi et al. (2003) found that S allele carriers demonstrated higher rates of adulthood depression in response to stressful life events than L allele carriers with a comparable history of childhood adversity. Several studies replicated this genotype by childhood adversity interaction (Caspi, Hariri, et al. 2010; Uher, Caspi, et al. 2011), while several studies could not replicate the finding (Munafo, Durrant, et al. 2009; Risch, Herrell, et al. 2009). Paradoxically, there are meta-analyses suggesting that there is (Karg, Burmeister, et al. 2011) and is not (Munafo, Durrant, et al. 2009; Risch, Herrell, et al. 2009) an interaction between SERT genotype and childhood adversity on outcomes such as depression. Studies with monkeys, which afford relatively more experimental control than human studies, have found that genetic variants in the serotonin transporter gene may interact with childhood stress to shape adult social behavior (Champoux, Bennett, et al. 2002), as well as serotonin metabolism (Bennett, Lesch, et al. 2002). Mouse studies also provide suggestive evidence for an interaction between SERT genotype and childhood stressful experiences in shaping adult social behavior (Jones, Smith, et al. 2010).
3 2 Part I : I maging G enetics and N eurochemistry
Several functional and structural magnetic resonance imaging (MRI) studies have examined the neural response to salient stimuli as a function of SLC6A4 (Hariri, Drabant, et al. 2006). Carriers of the S allele, relative to the L allele, show greater amygdala blood- oxygen level dependent (BOLD) reactivity in humans (Hariri et al. 2002; Munafo, Brown, et al. 2008). The S allele is also associated with greater BOLD signal functional connectivity between the amygdala and ventromedial prefrontal cortex (PFC) during the presentation of aversive images (Heinz, Braus, et al. 2005), and less functional connectivity between the amygdala and posterior anterior cingulate cortex (ACC) (Hariri, Drabant, et al. 2006). S allele carriers have also been found to exhibit reduced gray matter volume in the perigenual anterior cingulate cortex and amygdala, as well as less structural connectivity between those regions (Pezawas, MeyerLindenberg, et al. 2005). Thus, there is growing evidence that SLC6A4 shapes individual differences in neurobiological parameters such as structural and functional connectivity and in vivo function, which in turn may modulate vulnerability to emotional problems. In summary, the gene encoding SERT has been identified, along with several functional variants. A growing literature has identified the low-expressing S allele as increasing the risk for adult emotional problems, perhaps via an interaction with exposure to childhood adversity. Functional and structural neuroimaging studies suggest that the L/S allele may shape brain structure, connectivity, and function, particularly in the limbic region and in cortical-limbic connectivity. Many of the genetic variants in SLC6A4 await further study using genetic neuroimaging methodology. Given that several of these polymorphisms may have functional consequences, it is fair to conclude that there is much to learn about the role that SLC6A4 variations play in human behavior and brain function. P E T I M AG I N G A N D S E R T Positron emission tomography (PET) is the optimal imaging method to assess 5-HT binding in humans. Alternative methods have provided several insights about serotonin and have provided the groundwork for pursuing genetic neuroimaging studies of SERT in humans. For example, postmortem studies provide a powerful strategy for determining regional densities of serotonin receptors, including SERT (Cortes, Soriano, et al. 1988; Laruelle, Vanisberg, et al. 1988; Rosel, Arranz, et al. 1997). When assessed in vivo, the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA) in cerebrospinal fluid or blood has been used as an index for
general serotonergic function (Visser, van Waarde, et al. 2011). However, 5-HIAA levels represent the output of the entire serotonin system working in concert. Thus, another approach is needed to identify which deviant aspect of the serotonin system accounts for or explains abnormal metabolite levels. For example, it would be difficult to ascertain that role of SERT per se in producing altered 5-HIAA levels. A third method, neuroendocrine challenge tests (e.g., tryptophan depletion) has allowed researchers to assess serotonin release and receptor activation in the neural circuitry associated with neuroendocrine function. While this is a useful strategy for identifying function in the serotonin system, it is not well suited for detecting regional brain differences. In comparison, PET measures of SERT binding allows mapping of regional brain effects associated with function. INTERPRETING SERT BINDING ESTIMATES
In this section, we review methodological considerations for SERT PET research. We begin with a review of the interpretations of increased/decreased SERT binding. We then briefly address the outcome measures that assess this binding and the modeling used to obtain these outcome measures. Finally, we review the most common PET ligands used in human studies for measuring SERT binding. It is worthwhile to provide a brief overview of interpretations of PET SERT binding. When an individual exhibits increased SERT binding in a PET study, the most straightforward interpretation is that the brain region of interest contains a greater density of SERT and not a greater affinity of the tracer for SERT. In other words, SERT binding is elevated because there are more available SERT proteins to which the ligand may bind. An increase in SERT density could be due to reduced SERT internalization secondary to elevated serotonin release, a specific SLC6A4 genotype that leads to greater SERT density, or a transcription factor that increases gene expression. Increased total ligand binding could also result from non-specific binding, rather than an effect of SERT per se (e.g., the binding estimate is inflated because the ligand bound to receptors other than SERT). Finally, some PET tracers are sensitive to endogenous competition, which can affect binding. This does not appear to be the case for SERT. As noted later in this chapter, some PET binding measures are able to disentangle specific and non-specific binding. Those methods that do account for non-specific binding will be able to rule out differences in non-specific binding as a cause of altered total ligand binding.
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
33
Low SERT binding values in a PET study have similar but diametrically opposite interpretations from those offered for elevated SERT binding values. The most straightforward interpretation is that low SERT binding reflects a lower density of SERT receptors in that brain region. A decrease in SERT density could be due to greater SERT internalization secondary to decreased serotonin release. Alternatively, low binding estimates might result from decreased SLC6A4 gene expression, or a polymorphism that lowered affinity for the ligand by altering its structure but did not change density. Finally, the change in binding could be due to a change in affinity of the tracer for SERT. That high or low binding estimates can result from several plausible mechanisms must be considered in the interpretation of a PET study linking SERT binding estimates to mood, behavior, cognition, or psychiatric diagnosis. Supplemental lines of evidence, such as results from postmortem studies or animal studies, can be informative in distinguishing effects due to affinity or density differences. PET does not usually allow for multiple scans to partition separate effects (i.e., receptor concentration [Bmax] vs. dissociation equilibrium constant [Kd]). Despite these limitations, PET is an immensely powerful tool by which to study SERT in vivo.
SERT Outcome Measures PET SERT studies report several different outcome measures, depending on the quantification technique used. A thorough review of definitions of these measures can be found elsewhere (Innis, Cunningham, et al. 2007). To summarize, there are five commonly outcome measures used in SERT studies: (1) the volume of distribution, V T. For this measure, the radioactivity in plasma from an arterial blood sample must be measured during the PET scan. V T refers to the volume of plasma that contains the same concentration of tracer as contained in the region of interest. V T measures the total binding within a region of interest: the sum of specific (SERT) and non-specific binding (anything other than SERT). (2) BPF is the closest measure to in vitro binding potential, defined as Bmax/K D. In practice, BPF is estimated using V T and one additional measure—the free fraction (f P). Free fraction is a measure of the percent of the tracer not bound to any proteins in blood (i.e., the fraction of tracer free to cross the blood-brain barrier and enter the
brain, where it can then bind to SERT). BPF represents the tracer specifically bound to SERT within a region of interest. It can be calculated by subtracting the non-specific binding (estimated from a reference region lacking specific binding) from the total binding (V T) in the region of interest and dividing by fp (BPF = [V T – V ND]/f P (Innis, Cunningham, et al. 2007). Because no region of the brain is completely devoid of SERT (Laruelle and Maloteaux 1989; Parsey, Kent, et al. 2006), this assumption is violated to a modest degree, which can lead to an under-estimation of specific binding estimates. Non-specific binding is referred to as V ND, where ND stands for non-displaceable, since SERT-specific drugs will not displace non-specifically bound tracer. (3) BPP is used when f P is unknown or measurements of f P are unreliable. It is calculated as BPP = V T – V ND. (4) BPND is calculated as [V T – V ND]/V ND. The advantage of BPND over other outcome measures is that it can be calculated without an arterial input function, which is much less burdensome for the research subject and the research team. This calculation is possible by using “reference region approaches” that allow the calculation of BPND directly (without estimation of V T). Although these approaches save time, money, and research subject burden, they are heavily dependent on the choice of reference region. This can be a challenge for SERT PET studies since, as mentioned earlier, an ideal reference region does not exist for SERT tracers. (5) Another popular measure that can be calculated using reference region approaches is distribution volume ratio (DVR), or the ratio of V T to V ND. From the equation for BPND, it is readily apparent that DVR is related to BPND (i.e., DVR = BPND + 1). See Chapter 2 for more details on binding potential equations. Several practical factors, such as availability of plasma data, may limit the choice of outcome measure. Except for BPND and DVR (an extension of BPND), all of the outcome measures listed earlier require arterial blood plasma sampling to obtain an input function, which is expensive and invasive. For this reason, plasma analysis is not always available. Another factor is the reliability of the reference region. All the measures listed earlier, with the exception of V T, require a valid estimate of non-specific binding (determined by binding in a reference region that is devoid of specific binding such that its total binding is only non-specific binding). When non-specific binding
3 4 Part I : I maging G enetics and N eurochemistry
estimates are not reliable (as is the case with SERT), V T/ f P may be preferred to V T because pharmacological interventions (or disease state) may affect the percent of free tracer (f P) (Miller, Hesselgrave, et al. 2013b). Therefore, comparing V T before/after interventions (or between groups) may include the confounding factor of free fraction differences in addition to SERT changes. A third factor is that tracer properties may restrict outcome measure possibilities. For example, [11C]McNeil 5652 (radiolabeled trans-1,2,3,5,6,10-β-hexahydro-6-[4-(methylthio) phenyl]pyrrolo-[2,1-a]-isoquinoline) was the first human widely used SERT PET tracer (Suehiro, Scheffel, et al. 1993a; Suehiro, Scheffel, et al. 1993b). However, the quantification of [11C]McNeil 5652 free fraction using a common technique (ultrafiltration) proved to be impossible because > 80% of the tracer will bind to the filter used in the measurement (Oquendo, Hastings, et al. 2007). Therefore, for this tracer, BPF estimates were not employed, and the best alternative outcome measure closest to in vitro estimate was BPP. Note that when plasma analysis is not performed, neither V T in the reference region nor f P can be measured, and only outcome measures that can be estimated by reference region approaches (BPND or DVR) are available. These estimates make the assumptions that non-specific binding and f P do not vary from scan to scan in the same subject and/ or between groups (e.g., across diagnoses). Because of these assumptions, differences between groups may be obscured (or exaggerated). For example, if V ND is actually a measure of non-specific binding plus some SERT binding (because the reference region contains SERT), then higher levels of SERT (in either depressed or control individuals) will also result in higher V ND. A group difference in binding may not be observed using reference region techniques, however, because BPND values ([V T – V ND]/V ND) will be artificially low by high V ND estimates since the numerator will be smaller and the denominator will be greater. Moreover, sometimes there is no real difference in specific binding, but since it is not estimated independently of non-specific binding, a difference in non-specific binding may produce an apparent difference in BPND. Therefore, in SERT studies, BPND and DVR outcome measures must be interpreted with caution.
SERT Modeling Considerations Aside from the choice of outcome measure, many other factors can affect SERT binding estimates, for example the resolution of the PET scanner (especially in relation to the size of the region of interest), total scanning time being sufficient
to capture equilibrium binding estimates for all brain regions of interest, choice of image reconstruction method, image-processing techniques including motion correction and co-registration of PET frames to an MRI or template for delineation of regions of interest, and assumptions about the component of the PET signal that are due to the tracer in blood. It is beyond the scope of this chapter to review each of these factors. However, one significant component of PET quantification is the choice of modeling technique. For SERT quantification, there are five models that are most commonly used. Brief, non-technical descriptions of these models are provided here: (1) The one tissue compartment model (1TC). In this model, the kinetics of the tracer are defined by two rate parameters: the rate constants in and out of the brain from the plasma (Innis, Cunningham, et al. 2007). This method requires plasma analysis and is useful when tracer kinetics make it difficult to detect differences between two tracer states—specifically bound tracer or non-specifically bound/free tracer. (2) The simplified reference tissue model (SRTM). This is an extension of the reference tissue model (RTM), and as such, does not require plasma analysis. In addition to the assumptions of RTM, SRTM can be applied when tracer kinetics are well-defined by a single tissue compartment (as in the 1TC). This additional assumption allows the use of three parameters to describe the tracer kinetics (instead of the four used by RTM) and solves issues in RTM related to convergence and standard errors. When SRTM is used, only BPND can be estimated (Lammertsma and Hume 1996). (3) Multilinear reference tissue models (MRTM). A combination of two MRTM models was developed specifically for analysis of [11C]-3-amino-4-(2-dimethylamino methyl-phenylsulfanyl)-benzonitrile ([11C]DASB, developed by Wilson, Ginovart, et al. 2002) without plasma input (reference region approach). In order to reduce outcome bias, the developers of this technique (Ichise, Liow, et al. 2003) rearranged the terms of the original MRTM model (called MRTM0) in order to remove a noisy term related to regional radioactivity. The resulting equation, referred to as MRTM, led to unstable estimates of BPND, but stable, unbiased estimates of the clearance rate (rate at which the tracer is cleared from the body) when using the reference tissue (cerebellum). Further modification of MRTM (called MRTM2) allowed the reduction of parameters from three to two by fixing this clearance rate across all
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
35
regions. This method was favorably compared to both SRTM and a variant of SRTM (SRTM2). (4) Logan graphical analysis is a graphical technique for estimating tracer binding (Logan, Fowler, et al. 1990). Using this technique, the radioactivity in the region of interest and plasma are transformed such that, beyond a certain scan time, the relationship between transformed variables is linear. The slope of this function is related to V T. In a further extension, Logan et al. (1996) showed that BPND could be estimated from this graphical technique by replacing the plasma radioactivity with the radioactivity in the reference region. This reference region approach is often referred to as the non-invasive Logan approach. (5) L ikelihood estimation in graphical analysis (LEGA). Estimation of the slope in the Logan approach is performed using ordinary least squares (OLS). LEGA solves a problem in the Logan method, namely that the transformed variables in the Logan technique contain noise (both on the x and y axis), which caused an under-estimation of binding that increases with increasing noise levels. In the LEGA approach, the noise is modeled directly, and likelihood theory is used to generate a reliable estimate of the slope parameter independent of noise. The consideration of which model to choose depends on the characteristics of the tracer and the study design or research question, and therefore a single model is not appropriate for all SERT studies. One way to make a model choice is with test-retest PET studies or the reliability and validity of alternative outcome measures. In these studies, the same subject is scanned twice, most often in the same day and under the same conditions. Multiple modeling techniques are applied to the test and retest data, and various metrics (including test-retest reliability, variance of the measure, and time needed to attain a stable measurement) are estimated for each model. Such test-retest studies were performed using [11C]DASB to determine the optimal modeling technique (Frankle, Slifstein, et al. 2006; Kim, Ichise, et al. 2006; Ogden, Ojha, et al. 2007). Reliable estimates have been found when reference region approaches (SRTM and MRTM2) were applied (Frankle, Slifstein, et al. 2006; Kim, Ichise, et al. 2006). However, the estimates differed from those obtained using plasma analysis. The 1TC and LEGA approaches were top performers in all of the metrics tested in Ogden, Ojha, et al. (2007). That work did not consider reference region approaches
because of the detectable levels of SERT binding in the cerebellum. Although [11C]McN5652 predated [11C]DASB, test-retest studies of [11C]McN5652 have not been performed. However, several studies examined the kinetics of this tracer (Szabo, Scheffel, et al. 1999; Buck, Gucker, et al. 2000; Parsey, Kegeles, et al. 2000). Among these studies, there is general consensus that the 1TC model provided the most reliable binding estimates.
SERT Ligand Comparison A number of radioactive ligands that bind to serotonin receptors, including SERT, have been reviewed elsewhere (Fletcher, Pike, et al. 1995; Saulin, Savli, et al. 2012). Here, we focus on ligands designed for the assessment of SERT binding. For several reasons, [11C]McN5652 has proven less than ideal (Parsey, Kegeles, et al. 2000; McCann, Szabo, et al. 2005). For example, [11C]McN5652 has relatively weak signal to noise (the proportion of total binding that is specific binding) in many important brain regions such as the prefrontal cortex (PFC) and anterior cingulate cortex, which renders it problematic for assessing SERT in these regions (Parsey, Kegeles, et al. 2000). In addition, [11C]McN5652 demonstrates high enough affinity to other receptors to hinder accurate measurement in regions with only moderate SERT density and high levels of these other receptors (Buck, Gucker, et al. 2000). Moreover, the time course for [11C]McN5652 to achieve equilibrium binding is somewhat long. It typically requires about 90–120 minutes of imaging using [11C]McN5652 to yield reliable measurements of SERT density. However, after 100 minutes of scanning, the passage of five half-lives of the isotope (11C) means there are fewer counts and less reliable estimation of the counts. Finally, as described earlier, it may be difficult to measure the plasma free fraction because the free fraction is too small to measure reliably (Huang, Hwang, et al. 2002). Several PET-SERT ligands were developed to improve assessment of SERT. A popular second-generation ligand, [11C]DASB (Wilson and Houle 1999; Houle, Ginovart, et al. 2000; Wilson, Ginovart, et al. 2002), offers an improvement over [11C]McN5652. Other tracers with more favorable properties include [11C] DAPP (Houle, Ginovart, et al. 2000), [11C]MADAM (Lundberg, Odano, et al. 2005; Lundberg, Halldin, et al. 2006), [11C] AFM (Huang, Narendran, et al. 2004), [11C] EADAM (Jarkas, McConathy, et al. 2005), [11C] HOMADAM (Jarkas, Votaw, et al. 2005), and [11C] AFE (Zhu, Guo, et al. 2004). Recently, several 18F labeled ligands have been developed
3 6 Part I : I maging G enetics and N eurochemistry
for SERT (Garg, Thopate, et al. 2007; Wang, Parhi, et al. 2008; Stehouwer and Goodman 2009; Huang, Huang, et al. 2012; Huang, Huang, et al. 2013). PET ligands may vary in several parameters, such as the kinetic rate, selectivity, and signal to noise ratio, and these may also vary by region of interest (Ametamey, Honer, et al. 2008). Studies that compare the ligands in the same species provide valuable information to guide researchers to select an optimal ligand. One such study conducted an examination of five SERT ligands: [11C]McN5652, [11C]ADAM, [11C]DASB, [11C]DAPA, and [11C]AFM. These five ligands were compared via an in vitro assessment of affinity and lipophilicity and in vivo assessment of SERT binding in baboons (Huang, Hwang, et al. 2002). Based on several contrasts, the authors concluded that [11C]DASB was particularly advantageous due to fast uptake into the brain (e.g., reduced imaging time). [11C]AFM was also identified as advantageous due to superior reliability of measurement in some regions with low density of SERT. In a followup study, [11C]McN5652 was compared to [11C]DASB in humans, using a 1TC model for both tracers (Frankle, Huang, et al. 2004). Both ligands’ regional binding estimates converged in identifying brain regions with the greatest V T (i.e., midbrain) and least amount of V T (i.e., cerebellum), and were comparable to relative regional SERT density findings from postmortem studies. Consistent with the findings in baboons, [11C]DASB outperformed [11C] McN5652 by estimating about half as much V T in the cerebellum. This finding was interpreted to mean that [11C] DASB has lower non-specific binding. However, unlike the findings from baboons, the kinetic properties of [11C]DASB in human subjects were not superior to [11C]McN5652; the optimal scan time for both ligands was about 90 minutes. Other studies in humans using [11C]DASB recommended a 90–100 minute scan duration (Frankle, Slifstein, et al. 2006). Thus, it is not clear if [11C]DASB is advantageous in terms of scan duration. In another study, [11C]McN5652 and [11C]DASB were compared in three baboons at three time points under three experimental manipulations: a control baboon, a baboon with reduced SERT density due to MDMA exposure, and a baboon with reduced SERT availability due to pre-treatment with paroxetine (Szabo, McCann, et al. 2002). The latter two experimental conditions were employed to establish whether the ligands could detect lower SERT density or fewer available SERT binding sites. In this study, statistical tests were used to determine that a 1TC fit the data better than other models. Both ligands (1) exhibited strong agreement as to the rank order of SERT density by brain region, (2) detected significantly reduced
V T in the baboon exposed to MDMA, and (3) detected reduced V T in the midbrain and limbic regions following paroxetine pretreatment. However, the ligands differed in some key contrasts that generally support the preferred use of [11C]DASB. For example, relative to [11C]McN5652, the kinetics for [11C]DASB were much more slowed by paroxetine pretreatment. This implies that [11C]DASB was more affected by competition with paroxetine for SERT, and that [11C]McN5652 was less affected, presumably because it bound to other receptors once SERT was blocked by paroxetine. Furthermore, the [11C]DASB V T values, but not the [11C]McN5652 values, were significantly reduced in cortical regions in the paroxetine pre-treated baboon relative to the control baboon, although these effects were not significant after controlling for multiple comparisons. However, that the paroxetine pre-treatment effects on cortical V T were relatively weaker for [11C]McN5652 implies that [11C]McN5652 will bind to other targets in the cortex when SERT is not available. Thus, [11C]DASB, relative to [11C]McN5652, offers a superior estimate of SERT binding because it has less non-specific binding/more selective binding for SERT. Importantly, this study also utilized in vitro methods to quantify loss of SERT following MDMA exposure (Szabo, McCann, et al. 2002). Both ligands were found to have grossly underestimated the loss of SERT following MDMA exposure based on in vitro analysis. One advantage of [11C]DASB is its popularity. The ligand has become well characterized through its use in many studies. For example, a recent large-scale study (n = 95) assessed individuals with [11C]DASB, as well as several other ligands (targeting 5-HT1A, 5-HT1B, and 5-HT2A) to create a normative database for PET studies (Savli, Bauer, et al. 2012). Access to such data may help researchers employing [11C] DASB in their own studies by providing a basis of comparison of results. S E R T I M AG I N G R E S E A RC H F I N D I N G S In this section, we briefly review five factors that may influence measurement of SERT PET: (1) demographic factors (i.e., age, gender, and ethnicity), (2) seasonal effects, (3) health-related variables, (4) SLC6A4 effects on SERT binding and serotonin receptor binding, and (5) serotonin receptor influences on SERT binding. For several reasons, these five factors can inform SERT PET research. First, these factors may be linked to the same psychopathologies that are studied using SERT. For example, women are much more likely to be diagnosed with major
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
37
depression than men (Kessler, McGonagle, et al. 1993), suicidal behavior increases in men after the age of 70 years (Szanto, Gildengers, et al. 2002), vulnerability to depression decreases in women after 55 years of age (Bebbington, Dunn, et al. 1998), and winter months are associated with increased risk for depression. SERT PET is ideally suited to test whether age-, gender-, or seasonal-related changes in psychopathology can be accounted for by corresponding changes in SERT. Second, these factors may need to be controlled for in SERT PET studies either by study design (e.g., matched sampling; conducting all studies within the same season) or post facto statistical correction. Finally, these factors represent substantive research questions in their own right (e.g., developmental neuroscience, biological rhythms, genetic neuroimaging, etc.). DEMOGR APHIC VARIABLES
The SERT PET literature contains several studies of mixed-gender adult samples spanning a wide age range (e.g., 18–60 years old). In general, results are mixed as to whether demographics are associated with SERT PET binding.
Gender In one [11C]DASB PET study using SRTM for quantification, women relative to men exhibited a 55% reduction in SERT BPND, (Jovanovic, Lundberg, et al. 2008). The gender difference was not uniform, however; midbrain regions showed the greatest reduction. Interestingly, this study also reported that women exhibited 39% more 5-HT1A receptors, a gender difference similar to what has been reported elsewhere for the 5-HT1A receptor (Parsey, Oquendo, et al. 2002). Another study reported reduced [11C]DASB BPND (using MRTM2) in women in the putamen, but not in the thalamus, caudate, or cerebellum (Kalbitzer, Erritzoe, et al. 2010). In yet other studies (including one performed using plasma analysis), gender differences were reported to be non-significant using [11C]DASB and BPND (MRTM2) or V T/fp (LEGA) (Cannon, Ichise, et al. 2006; Miller, Hesselgrave, et al. 2013b).
applies constraints from previous plasma analysis. Another group reported reductions in [11C]DASB BPND (using MRTM2) by age in the putamen and thalamus but not caudate or cerebellum (Kalbitzer, Erritzoe, et al. 2010). A third study reported a decrease in [11C]DASB binding (using MRTM2) in the thalamus with age in a sample of healthy controls, but no correlation in a sample with bipolar disorders (Cannon, Ichise, 2006). A fourth study reported a decline in [11C]DASB BPND (using MRTM2) with age, where age was entered as a covariate term in a study of depression (Meyer, Wilson, et al. 2001). Conversely, Bose, Mehta, et al. (2011) reported a positive correlation between age and [11C]DASB BPND (assessed using the Logan graphical approach with plasma analysis) in 2 out of 16 brain regions in a sample of 42 adult male participants aged 25–60 years-old. However, because the correlations were not significant after correcting for multiple tests, they did not mention in which brain regions the correlations were initially detected as significant. In yet more studies, non-significant correlations between SERT and age were reported (Parsey, Hastings, et al. 2006b; Miller, Hesselgrave, et al. 2013b). Evidence for age-related change in regional density of serotonin receptors (Wong, Wagner, et al. 1984; Meltzer, Smith, et al. 1998; Moses-Kolko, Price, et al. 2011) is consistent with the idea that there may be changes in SERT as well, possibly due to up-regulation or down-regulation, respectively. It is difficult to draw firm conclusions about age and gender effects on SERT, given that many studies have reported no effect, and the studies reporting effects were inconsistent in terms of region and direction. Although evidence is largely mixed as to whether age and gender affect SERT PET measurements, it is important to stress that few studies specifically sampled a large cohort of male and female adults in each decade of life. In addition, some inconsistencies may have arisen from differences in modeling strategies. Given that basic demographic data are likely to be collected routinely in SERT PET studies, it may be possible to pool data across studies for a meta-analysis of age and gender effects.
Ethnicity Age To date, a few PET studies reported evidence that SERT declines with age. Yamamoto, Suhara, et al. (2002) examined 28 healthy males between the ages of 20 and 79 using [11C]McN5652 and found evidence of age-related decline in BPND in the thalamus and midbrain of about 10% per decade using a modified Logan graphical approach that
Very few studies have reported how ethnic differences might influence SERT PET studies. One study by Praschak-Rieder, Kennedy, et al. (2007) reported that Caucasians with two copies of the L A allele (homozygotes) exhibited higher [11C]DASB BPND, as assessed from non-invasive Logan graphical analysis in the putamen, than LA allele homozygote non-Caucasians. However, it is
3 8 Part I : I maging G enetics and N eurochemistry
unclear if the LA allele functions differently by ethnicity or was moderated by other genes that were ethnicity specific. SEASONALIT Y
In a study of 88 healthy adults assessed over a 4-year period, Praschak-Rieder, Willeit, et al. (2008) found greater [11C] DASB BPND using non-invasive Logan graphical analysis in several brain regions in participants assessed in the winter/fall relative to spring/summer. Moreover, less sunshine (a proxy for season) was also associated with greater levels of BPND across several brain regions. A second group of researchers replicated this finding (except in the midbrain), and examined SERT genotype as a moderator of the effect of seasonality (Kalbitzer, Erritzoe, et al. 2010). They reported that S allele carriers, relative to L allele homozygotes, showed a greater increase in [11C]DASB BPND (using MRTM2) during periods with less sunlight. Thus, when using BPND, it may be necessary to control for seasonality effects on SERT in studies that recruit participants across the year (Praschak-Rieder and Willeit 2012). HE ALTH-REL ATED VARIABLES
One recent study reported evidence that cortical and subcortical [11C]DASB BPND, as assessed using MRTM2 , correlates inversely with body mass index (Erritzoe, Frokjaer, et al. 2010). Such a finding has implication for obesity research, but also, if replicated, would suggest a necessary covariate in PET SERT studies generally.
heterozygotes using [11C]McN5652 BPND as estimated from the non-invasive Logan approach (Shioe, Ichimiya, et al. 2003). A third study found no biallelic effect or triallelic effect on [11C]DASB BPND in a sample of 63 adults using the Logan graphical approach with plasma analysis (Murthy, Selvaraj, et al. 2010). In addition, a fourth study of 83 adults also failed to detect an effect of L A alleles (i.e. 0, 1, or 2) on [11C]DASB binding derived using LEGA (Miller, Hesselgrave, et al. 2013b). The latter two studies are notable because of the relatively large sample sizes, but also because they utilized [11C]DASB, while the former two studies with null findings utilized [11C]McN5652. Along these lines, a study of 29 rhesus monkeys reported no difference between S allele homozygotes and L allele carriers on a measure of SERT DVR using [11C]DASB and MRTM2 (Christian, Fox, et al. 2009). These findings suggest that SLC6A4 variation may not impact SERT density (or, at the very least, not impact SERT binding). Three studies employing [11C]DASB and reference region approaches have provided evidence that S carriers exhibited less SERT BPND, but only in one specific region of interest per study: the putamen (Praschak-Rieder, Kennedy, et al. 2007), midbrain (Reimold, Smolka, et al. 2007a), and caudate (Kalbitzer, Erritzoe, et al. 2010). One study found that relative to the S allele carriers, L allele carriers exhibited more [11C]DASB BPND across all brain regions using the Logan graphical approach with plasma analysis (Bose, Mehta, et al. 2011). Thus, evidence is mixed about whether LA, S, and LG impact PET SERT assessment. OTHER SEROTONIN INFLUENCES
SLC6A4 VARIATION ON SERT BINDING
A number of studies have examined the effect of the SLC6A4 polymorphisms on SERT binding. Given that SLC6A4 codes the structure of SERT and that several SLC6A4 polymorphisms are functional, it is reasonable to expect a correlation between SLC6A4 genotype and SERT binding values. Paradoxically, several studies using several methodologies have failed to provide such evidence. For example, no difference by SLC6A4 genotype has been reported in postmortem studies of SERT density (Sugden, Tichopad, et al. 2009) and in analysis of serotonin in blood platelet (Patkar, Berrettini, et al. 2004). Several PET studies have reported no genotype effect on SERT. For example, Parsey, Hastings, et al. (2006a) found no differences in [11C]McN5652 BPP (using LEGA) between LG, S, and LA carriers (n = 67) across several brain regions. A similar null finding was reported in a study of 27 adults comparing S homozygotes, L homozygotes, and
The link between SLC6A4 polymorphisms and 5-HT1A receptor density has been the focus of a few studies. [11C] WAY-100635, or N-[2-[4(2-methoxyphenyl)-1-piperazinyl]ethyl]-N-(2-pyridyl) cyclohexanecarboxamide, is a popular ligand for assessing 5-HT1A, and has been utilized concurrently to study the impact of SLC6A4 variation on 5-HT1A binding. One study found reduced 5-HT1A BPND in S allele carriers relative to L allele homozygotes in a mixed gender sample (David, Murthy, et al. 2005). In a sample of women, S allele carriers, relative to L allele homozygotes, exhibited greater serotonin1A BPP in the anterior cingulate, but not in other brain regions (Lee, Bailer, et al. 2005) and this was basically replicated in a mixed gender sample using BPND and a different tracer (Lothe, Boni, et al. 2009), except for some regional differences. A fourth study found no difference in serotonin1A BPND between S allele carriers and L allele homozygotes (Borg, Henningsson, et al. 2009).
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
39
Although less studied relative to serotonin1A, SLC6A4 may be associated with serotonin2A receptor binding (Erritzoe, Frokjaer, et al. 2010), and genetic variation in the serotonin2A gene may influence SERT PET density (Laje, Cannon, et al. 2010). In addition, genes outside the serotonin system may impact SERT binding. For example, the Val66Met polymorphism in the brain-derived neurotrophic factor (BDNF) gene may modulate SERT expression, and vice versa (Henningsson, Borg, et al. 2009), but findings have been inconsistent here as well (Klein, Trajkovska, et al. 2010).
ASSOCIATION BE T WEEN SERT BINDING AND 5-HT RECEPTOR BINDING
Several studies have examined the correlation between SERT binding and binding of a ligand designed to target a 5-HT receptor. These studies are valuable because they suggest that SERT’s part in the serotonin system is dynamic and interactive with individual receptors, rather an independent part of the serotonin system determined solely by SLC6A4 genotype. In other words, they provide evidence that genetic variation in the serotonin system can influence SERT density, and therefore likely SERT binding. In several studies, a positive correlation has been observed between SERT and 5-HT1A binding. For example, one group employed the ligand [11C] MADAM to assess SERT BPND and [11C]WAY-100635 to assess 5HT1A BPND and found a positive correlation in the brain stem raphe and hippocampus, but no correlation in the frontal cortex (Lundberg, Borg, et al. 2007). In a second study, [11C] WAY-100635 BPND and [11C]DASB BPND (using Logan graphical approach and plasma analysis) were positively correlated in the raphe nucleus, insula, and superior temporal gyrus, and trended toward significance in the inferior frontal region (Bose, Mehta, et al. 2011). There was also suggestive evidence that 5-HT1A receptor density in the raphe may be correlated with SERT density throughout the brain. Conversely, the correlation between SERT BPND in the raphe and a measure of total brain 5-HT1A BPND was not as robust, suggesting that SERT density in the raphe may not have far-reaching consequences for 5-HT1A density. It should be noted that Bose et al. (2011) conducted several statistical tests, making it likely that the statistical correction hindered the detection of subtle true effects. In a third study, Jovanovic, Lundberg, et al. (2008) presented correlations between 5-HT1A BPND, and SERT BPND (using
[11C]MADAM and SRTM) in the cortex, hippocampus, and dorsal raphe by gender, although with the caveat that sample size was small for such analyses. One positive correlation reached statistical significance in the hippocampus for women (r = 0.93) but not men (r = 0.73), although both correlations are consistent with the idea that 5-HT1A and SERT are positively correlated in the hippocampus. However, there was suggestive evidence of a region by gender interaction in the correlation between 5-HT1A BPND and SERT BPND. For example, the correlation in the dorsal raphe was more substantial in men (r = 0.74) than women (r = −0.06), although neither correlation reached statistical significance due to sample size. In a fourth study, a negative correlation between [11C] WAY-100635 BPND and [11C] DASB BPND (using MRTM 2) was observed in the cingulate, insula, and several cortical regions (Takano, Ito, et al. 2011), although the use of BPND with [11C] WAY-1006535 can produce misleading results due to the presence of 5-HT1A receptors in the cerebellum (Parsey, Ogden, et al. 2010). Altogether, there are mixed findings from studies linking SERT binding to SLC6A4, and mixed findings linking SERT binding to binding from ligands designed to assess other serotonin receptors. These findings may be considered suggestive of two broad conclusions. First, SLC6A4 may not be the only factor, genetic or environmental, that influences SERT binding. There may be up- or down-regulation in SERT based on serotonin release, which implicates any and all factors affecting serotonin receptor function in the etiology of SERT density. Second, SLC6A4 appears to impact other aspects of the serotonin system besides SERT, including the serotonin receptors that may in turn up-regulate and down-regulate SERT density. Perhaps, the SLC6A4 genotype has an upstream or downstream effect on the SERT receptor density that manifests or disappears developmentally through an interaction with other serotonin receptors. Those effects may also vary by gender and or region of interest, a further complexity that requires study. L I N K BE T W E E N S E R T BI N DI N G A N D PHENOT YPES OF INTEREST SERT PET studies have examined the role of SERT in several phenotypes of interest (e.g., psychiatric, medical, etc.), and several reviews have been published (Parsey and Mann 2003; Veltman, Ruhe, et al. 2010; Jones and Rabiner 2012). While this is not an exhaustive literature review, it helps illuminate the diversity of potential applications for SERT PET.
4 0 Part I : I maging G enetics and N eurochemistry
DEPRESSION
Several lines of evidence support the hypothesis that serotonin plays a role in clinical depression. This has often been referred to as the “serotonin hypothesis” of depression. First, psychiatric medications that bind to SERT, such as the selective serotonin reuptake inhibitors (SSRIs), are often effective in the treatment of depression. Second, postmortem studies have found a lifetime history of depression to be associated with decreased serotonin receptor density (Stockmeier 2003), abnormalities in SERT mRNA in the dorsal raphe (Anisman, Du, et al. 2008), and less binding in dorsal and ventral PFC (Underwood, Kassir, et al. 2012) and dorsal raphe (Arango, Underwood, et al. 1995; Arango, Underwood, et al. 2001; Baumann, Bielau, et al. 2002). Third, studies have shown a link between serotonin genetics and risk for major depressive disorder (MDD) (Caspi, Sugden, et al. 2003; Karg, Burmeister, et al. 2011). Fourth, the rodent model of clinical depression (e.g., surgical ablation of the olfactory bulb) is associated with greater SERT density (Grecksch, Zhou, et al. 1997). Fifth, studies of serotonin metabolites have reported reductions in individuals with a history of MDD, relative to individuals without a history of MDD (Asberg 1997; Rosa-Neto, Diksic, et al. 2004; Mann and Currier 2007). Several PET studies of depression have examined the 5-HT1A receptor using [11C]WAY-100635, but results have been mixed, in part because of choice of outcome measure (Smith and Jakobsen 2009). Some studies have reported reduced binding of the 5-HT1A receptor in MDD (Drevets, Frank, et al. 1999; Sargent, Kjaer, et al. 2000; Savitz and Drevets 2012), while other studies have reported the opposite effect (Parsey, Olvet, et al. 2006). Recently, the latter group has replicated their [11C]WAY-100635 result (higher 5-HT1A in depressed subjects) as well as reconciled the source of these discrepant findings (Parsey, Ogden, et al. 2010; Miller, Hesselgrave, et al. 2013a). In these Parsey et al. studies, BPF (the closest measure to in vitro estimates) was estimated using plasma analysis and the cerebellar white matter as a reference region. This group was able to show that using suboptimal quantification techniques, including a reference region approach, BPND as an outcome measure, and cerebellar gray matter as a reference region (which has measureable specific binding), yields results in the opposite direction. This is a prime example of the inaccuracies that can occur when applying reference region approaches. The SERT PET literature on depression is similarly composed of mixed findings, although no study has demonstrated whether choice of outcome measure explains the inconsistencies. Two studies reported an increase in SERT
binding in MDD. In one study that compared unmedicated depressed patients and controls, there was increased [11C]DASB BPND (quantified using MRTM2) in the thalamus, periaqueductal gray matter (PAG), insula, and striatum, which also correlated with severity of anxiety and depression (Cannon, Ichise, et al. 2007). In a second study, people with MDD, relative to controls, exhibited higher [11C]McN5652 SERT DVR (quantified using a 1TC model) in left PFC and right cingulate, but not midbrain regions or the occipital cortex (Reivich, Amsterdam, et al. 2004). In other studies using various outcome measures and modeling techniques (including full quantification with an arterial input function), there were no differences in SERT binding between MDD and controls (Meyer, Houle, et al. 2004; Bhagwagar, Murthy, et al. 2007). Still more studies reported decreased SERT binding in MDD. In one study, the amygdala and midbrain showed reductions in [11C]McN5652 BPP in MDD relative to controls, and the effect was greater for unmedicated MDD cases (Parsey, Hastings, et al. 2006b). This suggests that antidepressant usage normalized SERT density in MDD patients. In addition, SERT BPP was not correlated with MDD severity, days of MDD without medication, or suicide completion (Parsey, Hastings, et al. 2006b). In a second study using [11C]DASB and MRTM2 , MDD patients relative to controls showed reduced BPND in the several regions, including the thalamus, and BPND in the thalamus and amygdala correlated with ratings of anxiety (Reimold, Batra, et al. 2008). Correlations with depression ratings, on the other hand, were not significant. In a third study, individuals with a positive family history for depression exhibited reduced [11C]DASB BPND quantified using MRTM2 in the DLPFC and ACC relative to low-risk individuals (Frokjaer, Vinberg, et al. 2009). In a fourth study, reduced [11C]McN5652 BPP in the amygdala, midbrain, and anterior cingulate at baseline predicted non-response to treatment among MDD patients relative to controls (Miller, Oquendo, et al. 2008). One explanation for the mixed effects is that medicated MDD patients may experience a normalization of SERT, while their natural state is reduced SERT (Parsey, Hastings, et al. 2006b). That unmedicated high-risk individuals also exhibit reduced binding is consistent with the idea that MDD is associated with reduced SERT binding, and that medication usage may obscure such effects (Frokjaer, Vinberg, et al. 2009). In addition, etiologic heterogeneity may obscure the association between depression and SERT. For example, a recent study found that lower SERT binding characterized depressed individuals who also attempt suicide, but not depressed
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
41
individuals without a history of suicide attempts (Miller, Hesselgrave, et al. 2013). Methodological decisions, such as choice of outcome measure, may also have contributed to mixed findings. BIPOL AR DISORDER
In studies that compare individuals with BP to controls, one study observed increased [11C]DASB BPND (using MRTM2) in thalamus, dorsal ACC, medial PFC, and insula, and decreased BPND in raphe (Cannon, Ichise, et al. 2006). In another study, [11C]MCN5652 BPP (using LEGA) has been found to be reduced across regions (Oquendo, Hastings, et al. 2007). ANXIE T Y DISORDERS
There are relatively few studies of anxiety disorders using SERT PET methodologies. Findings about SERT binding in obsessive-compulsive disorder (OCD) have been mixed. Initial findings suggested that there was no difference between OCD patients and controls on [11C]MCN5652 BPP using a 1TC model (Simpson, Lombardo, et al. 2003). Another study found reduced [11C]DASB BPND quantified using MRTM 2 in the thalamus and midbrain in a group with OCD relative to controls (Reimold, Smolka, et al. 2007b). A third study found reduced [11C]DASB BPND quantified using MRTM 2 in the insula and frontal cortex (Matsumoto, Ichise, et al. 2010). Relative to controls, panic disorder was associated with greater SERT BPND across several brain regions using [11C]MADAM and SRTM, but only for males (Maron, Toru, et al. 2011). NEUROCOGNITIVE DISORDERS
Parkinson’s disease is associated with the death of dopamine-generating neurons, but this does not preclude the serotonin system from being affected. Increased [11C]DASB BPND (using the non-invasive Logan graphical approach) has been observed in the dorsolateral PFC in depressed Parkinson’s patients who were not using antidepressants (Boileau, Warsh, et al. 2008). In addition, increased [11C]DASB BPND, similarly quantified, has been observed across regions in Parkinson’s patients with depression (Politis, Wu, et al. 2010a; Politis, Wu, et al. 2010b). Studies have also examined SERT in Alzheimer’s disease (Meltzer, Smith, et al. 1998).
SUBSTANCE USE
Several drugs of abuse have direct neurotoxic properties that may affect the functioning of the serotonin system and SERT in particular. Ecstasy (MDMA; 3,4-methylenedioxy-methamphetamine) is known to be neurotoxic for neurons containing serotonin receptors and SERT. Using SERT PET to study humans, MDMA users exhibit lower binding in several brain regions (McCann, Szabo, et al. 1998; Buchert, Thomasius, et al. 2003; McCann, Szabo, et al. 2005). However, partial recovery of SERT is believed to occur because remitted MDMA users showed more normative SERT binding (Buchert, Thomasius, et al. 2003; McCann, Szabo, et al. 2005). A second drug of abuse linked to SERT and studied using PET methodology is alcohol. One study found that recovering alcoholics exhibit reduced SERT using [11C]McN5652 and several outcome measures (VT, BPP, DVR, quantified by the 1TC model) relative to healthy controls (Szabo, Owonikoko, et al. 2004), while another study found no difference using [11C]DASB BPND and MRTM2 (Brown, George, et al. 2007). PERSONALIT Y RESEARCH
There is growing interest in understanding the biological basis of normal range personality, due in part to the overlap between normal personality and psychiatric illness (Krueger and Markon 2006). Genetic neuroimaging of personality is a relatively understudied area, and very little is known about SERT PET correlates of personality. Attempts to establish a link between SLCA4 genetic variation and personality have yielded mixed evidence, due in part to inconsistent findings between cohorts (Terracciano, Balaci, et al. 2009) and because different instruments purporting to assess the same personality trait seem to differentially relate to SLCA4 genetic variation (Sen, Burmeister, et al. 2004; Munafo, Freimer, et al. 2009). Although there are several models of personality, the Big Five model is perhaps the most well-studied and validated model, with very strong links to psychopathology (Kotov, Gamez, et al. 2010). One study found a positive correlation between [11C]DASB PET BPND quantified using MRTM2 in the thalamus and neuroticism, but no other effects were detected (Takano, Arakawa, et al. 2007). In a study with monkeys, SERT availability (BPND, assessed using [11C]DASB and MRTM2) was correlated with anxious temperament in the medial temporal lobe (e.g., amygdala, hippocampal region; Oler, Fox, et al. 2009).
4 2 Part I : I maging G enetics and N eurochemistry
MEDICAL/HE ALTH PHENOT Y PES
There is growing interest in uncovering the association between medical phenotypes and SERT using PET. For example, HIV patients with comorbid depression have been found to exhibit increased [11C]DASB BPND, quantified with SRTM, in some subcortical regions compared to HIV patients without depression (Hammoud, Endres, et al. 2010). One study found that participants with chronic fatigue syndrome relative to healthy participants exhibited reduced [11C]McN5652 BPND (using SRTM) in the anterior cingulate cortex (Yamamoto, Ouchi, et al. 2004). CHALLENGES
There are several challenges that impede progress in the field of clinical genetic neuroimaging. One challenge is that participants in psychiatric studies are often on medication or have a history of psychotropic medication use, which may act directly, or indirectly, to normalize SERT density. That these medication effects may alter SERT density may lead to spurious conclusions about the role of SERT in a clinical phenotype. Drug-free participants, drug naïve participants, and high-risk participants prior to the onset of disease should be studied when possible to minimize confounding medication effects. However, these cohorts may not be representative of the population under study. A second issue is that clinical phenotypes often show etiologic heterogeneity. In other words, SERT may explain depression in some cases but not all, and the extent to which a sample is heterogeneous will attenuate effect sizes or mask effects completely.
F U T U R E DI R EC T I O N S SERT PET has several clinical applications, particularly in terms of advancing our understanding of existing treatments, such as psychotropic medication. First, several studies have utilized SERT PET is to elucidate the mechanism of action for SSRIs (Meyer, Wilson, et al. 2001; Parsey, Kent, et al. 2006; Voineskos, Wilson, et al. 2007). These studies typically involve a SERT PET assessment prior to medication onset in order to establish a baseline level of SERT in each participant. Then, after the medication trial, participants undergo a second SERT PET assessment. The change in SERT binding between assessments by region is then used to infer where the treatment is binding in the brain, and to quantify the percent of SERT occupied by the drug. This kind of study design can be used in conjunction
with groups assigned to different medication dosages in order to examine dose response curves and identify optimal dosage. Second, studies have begun to explore how genetic variation in SERT correlates with or predicts treatment response (Smeraldi, Zanardi, et al. 1998; Pollock, Ferrell, et al. 2000; Serretti, Artioli, et al. 2005; Serretti, Benedetti, et al. 2005; Serretti, Mandelli, et al. 2007), and how PET findings might be used to predict treatment response (Lanzenberger, Kranz, et al. 2012). Along these lines, results from a meta-analysis of several studies suggest that depressed SERT L allele homozygotes, relative to depressed S allele carriers, show better response to SSRI treatment (Serretti, Kato, et al. 2007). These lines of research could be merged to examine genotype and treatment response using genetic neuroimaging methodologies (Ruhe, Ooteman, et al. 2009). Third, SERT PET may be used in the discovery and development of novel drugs (Talbot and Laruelle 2002; Guilloteau and Chalon 2005; Nemeroff and Vale 2005; DeLorenzo, Lichenstein, et al. 2011), but few studies have utilized this strategy to find drugs that bind to SERT specifically (DeLorenzo, Milak, et al. 2011). An important direction for future research is to understand why SERT binding is so inconsistently related to SLC6A4 variation and phenotypes of interest. One method that might be helpful is to widen the neurobiological phenotype of interest, such as exploring multivariate serotonin phenotypes as outcomes (e.g., SERT PET + fMRI + platelet analysis) rather than just SERT. Integrating SERT PET with other neuroimaging methods, such as functional magnetic resonance imaging, may also help elucidate the impact of SLC6A4 variation on neural physiology (Fisher and Hariri 2012). Alternatively, gains may be made by considering multiple functional polymorphisms within SLC6A4 or across serotonin genes. Perhaps, rather than group participants based on one polymorphism (e.g., L allele vs. S allele), it may be advantageous to score each participant based on several SLC6A4 variants or serotonin variants. Indeed, due in part to the recognition that common genetic variations are unlikely to have more than small effects on complex traits (Plomin, Haworth, et al. 2009), researchers have begun to explore strategies to sum effects across several polymorphisms within a neurotransmitter system (Derringer, Krueger, et al. 2012). Alternatively, increasing sample sizes and standardizing methods and analytic strategies may also help produce more consistent findings (Button, Ioannidis, et al. 2013; Falcone, Smith, et al. 2013). As is evident from the analysis presented in the preceding sections, research groups are using different outcome measures and various modeling techniques in their SERT studies. It is essential to
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
43
understand the assumptions of each of these techniques in order to know if these assumptions are violated for a given study. The lack of an ideal reference region is a particularly significant issue in SERT PET research. Given that differences in methodology has been proven responsible for disparate results in 5HT1A research, a lack of uniformity in SERT studies may have caused disparate results when reference region techniques are used. For this reason, future studies may consider using V T/f P for SERT tracers, in order to yield more consistent, reliable PET estimates. REFERENCES Ali FR, SVasiliou SA, et al. (2010). Combinatorial interaction between two human serotonin transporter gene variable number tandem repeats and their regulation by CTCF. J Neurochemistry. 112(1): 296–306. Ametamey SM, Honer M, et al. (2008). Molecular imaging with PET. Chem Rev. 108(5): 1501–1516. Anisman H, Du L, et al. (2008). Serotonin receptor subtype and p11 mRNA expression in stress-relevant brain regions of suicide and control subjects. J psychiat Neurosci. 33(2): 131–141. Ansorge MS, Zhou M, et al. (2004). Early-life blockade of the serotonin transporter alters emotional behavior in adult mice. Science. 306(5697): 879–881. Arango V, Underwood MD, et al. (2001). Serotonin 1A receptors, serotonin transporter binding and serotonin transporter mRNA expression in the brainstem of depressed suicide victims. Neuropsychopharmacology. 25(6): 892–903. Arango V, Underwood MD, et al. (1995). Localized alterations in preand postsynaptic serotonin binding sites in the ventrolateral prefrontal cortex of suicide victims. Brain Res. 688(1–2): 121–133. Asberg M. (1997). Neurotransmitters and suicidal behavior. The evidence from cerebrospinal fluid studies.” Ann N Y Acad Sci. 836: 158–181. Barnes NM, Sharp T (1999). A review of central serotonin receptors and their function. Neuropharmacology. 38(8): 1083–1152. Baumann B, Bielau H, et al. (2002). Circumscribed numerical deficit of dorsal raphe neurons in mood disorders. Psychol Med. 32(1): 93–103. Bebbington PE, Dunn G, et al. (1998). The influence of age and sex on the prevalence of depression conditions: Report from the National Survey of Psychiatric Morbidity. Psychol Med. 28(1): 9–19. Bennett AJ, Lesch KP, et al. (2002). Early experience and serotonin transporter gene variation interact to influence primate CNS function. Mol Psychiatry. 7(1): 118–122. Bhagwagar Z, Murthy N, et al. (2007). SerotoninT binding in recovered depressed patients and healthy volunteers: a positron emission tomography study with [11C]DASB. American J Psychiatry 164(12): 1858–1865. Boileau I, JWarsh JJ, et al. (2008). Elevated serotonin transporter binding in depressed patients with Parkinson’s disease: a preliminary PET study with [11C]DASB. Movement Disord. 23(12): 1776–1780. Borg J, Henningsson S, et al. (2009). Serotonin transporter genotype is associated with cognitive performance but not regional serotonin1A receptor binding in humans. Int J Neuropsychopharmacol. 12(6): 783–792. Bose SK, Mehta MA, et al. (2011). Presynaptic serotonin1A is related to serotoninT receptor density in the human brain. Neuropsychopharmacology. 36(11): 2258–2265.
Brown AK, George DT, et al. (2007). PET [11C]DASB imaging of serotonin transporters in patients with alcoholism. Alcohol Clin Exp Res. 31(1): 28–32. Buchert R, Thomasius R, et al. (2003). Long-term effects of “ecstasy” use on serotonin transporters of the brain investigated by PET. J Nucl Med. 44(3): 375–384. Buck A, Gucker PM, et al. (2000). Evaluation of serotonergic transporters using PET and [11C](+)McN-5652: assessment of methods. J Cereb Blood Flow Metab. 20(2): 253–262. Button KS, Ioannidis JP, et al. (2013). Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci. 14(5): 365–376. Cannon DM, Ichise MM, et al. (2006). Serotonin transporter binding in bipolar disorder assessed using [11C]DASB and positron emission tomography. Biol Psychiatry. 60(3): 207–217. Cannon DM, Ichise M, et al. (2007). Elevated serotonin transporter binding in major depressive disorder assessed using positron emission tomography and [11C]DASB; comparison with bipolar disorder. Biol Psychiatry 62(8): 870–877. Caspi A, Hariri AR, et al. (2010). Genetic sensitivity to the environment: the case of the serotonin transporter gene and its implications for studying complex diseases and traits. Am J Psychiatry. 167(5): 509–527. Caspi A, Sugden K, et al. (2003). Influence of life stress on depression: moderation by a polymorphism in the serotoninT gene. Science 301(5631): 386–389. Champoux M, Bennett A, et al. (2002). Serotonin transporter gene polymorphism, differential early rearing, and behavior in rhesus monkey neonates. Mol Psychiatry. 7(10): 1058–1063. Christian BT, Fox AS, et al. (2009). Serotonin transporter binding and genotype in the nonhuman primate brain using [C-11]DASB PET. NeuroImage. 47(4): 1230–1236. Cortes R, Soriano E, et al. (1988). Autoradiography of antidepressant binding sites in the human brain: localization using [3H]imipramine and [3H]paroxetine. Neuroscience. 27(2): 473–496. David SP, Murthy NV, et al. (2005). A functional genetic variation of the serotonin (serotonin) transporter affects serotonin1A receptor binding in humans. J Neuroscience. 25(10): 2586–2590. DeLorenzo C, Lichenstein S, et al. (2011). SEP-225289 serotonin and dopamine transporter occupancy: a PET study. J Nucl Med. 52(7): 1150–1155. DeLorenzo C, Milak MS, et al. (2011). In vivo positron emission tomography imaging with [(1)(1)C]ABP688: binding variability and specificity for the metabotropic glutamate receptor subtype 5 in baboons. Eur J Nucl Med Mol I. 38(6): 1083–1094. Derringer J, Krueger RF, et al. (2012). The aggregate effect of dopamine genes on dependence symptoms among cocaine users: cross-validation of a candidate system scoring approach. Behav Genet. 42(4): 626–635. Drevets WC, Frank E, et al. (1999). PET imaging of serotonin 1A receptor binding in depression. Biol Psychiatry 46(10): 1375–1387. Erritzoe D, Frokjaer VG, et al. (2010). Cerebral serotonin transporter binding is inversely related to body mass index. NeuroImage 52(1): 284–289. Falcone M, Smith RM, et al. (2013). Neuroimaging in psychiatric pharmacogenetics research: the promise and pitfalls. Neuropsychopharmacology. 38(12): 2327–2337. Fisher PM, Hariri AR. (2012). Linking variability in brain chemistry and circuit function through multimodal human neuroimaging. Genes Brain Behav. 11(6): 633–642. Fletcher A, Pike VW, et al. (1995). Visualization and characterization of serotonin receptors and transporters in viv and in man. Sem Neurosci. 421–431. Frankle WG, Huang Y, et al. (2004). Comparative evaluation of serotonin transporter radioligands 11C-DASB and 11C-McN 5652 in healthy humans. J Nucl Med. 45(4): 682–694.
4 4 Part I : I maging G enetics and N eurochemistry
Frankle WG, Slifstein M, et al. (2006). Estimation of serotonin transporter parameters with 11C-DASB in healthy humans: reproducibility and comparison of methods. J Nucl Med. 47(5): 815–826. Frankle WG, Slifstein M, et al. (2006). Estimation of serotonin transporter parameters with 11C-DASB in healthy humans: reproducibility and comparison of methods. J Nucl Med. 47(5): 815–826. Frokjaer VG, Vinberg M, et al. (2009). High familial risk for mood disorder is associated with low dorsolateral prefrontal cortex serotonin transporter binding. NeuroImage. 46(2): 360–366. Garg S, Thopate SR, et al. (2007). 3-Amino-4-(2-((4-[18F]fluorobenzyl)methylamino)methylphenylsulfanyl)benzonitrile, an F-18 fluorobenzyl analogue of DASB: synthesis, in vitro binding, and in vivo biodistribution studies. Bioconjugate Chem. 18(5): 1612–1618. Gelernter J, Kranzler H, et al. (1997). Serotonin transporter protein (SLC6A4) allele and haplotype frequencies and linkage disequilibria in African- and European-American and Japanese populations and in alcohol-dependent subjects. Hum Genet. 101(2): 243–246. Grecksch G, Zhou D, et al. (1997). Influence of olfactory bulbectomy and subsequent imipramine treatment on 5-hydroxytryptaminergic presynapses in the rat frontal cortex: behavioural correlates. BR J Pharmacol. 122(8): 1725–1731. Guilloteau D, Chalon S. (2005). PET and SPECT exploration of central monoaminergic transporters for the development of new drugs and treatments in brain disorders. Curr Pharm Design. 11(25): 3237–3245. Hariri AR, Mattay VS, et al. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science, 297(5580), 400–403. Hammoud DA, Endres CJ, et al. (2010). Imaging serotonergic transmission with [11C]DASB-PET in depressed and non-depressed patients infected with HIV. NeuroImage. 49(3): 2588–2595. Hariri AR, Drabant EM, et al. (2006). Imaging genetics: perspectives from studies of genetically driven variation in serotonin function and corticolimbic affective processing. Biol Psychiatry. 59(10): 888–897. Heils A, Teufel A, et al. (1996). Allelic variation of human serotonin transporter gene expression. Jo Neurochem. 66(6): 2621–2624. Heinz A, Braus DF, et al. (2005). Amygdala-prefrontal coupling depends on a genetic variation of the serotonin transporter. Nat Neurosci. 8(1): 20–21. Henningsson S, Borg J, et al. (2009). Genetic variation in brain-derived neurotrophic factor is associated with serotonin transporter but not serotonin-1A receptor availability in men. Biol Psychiatry. 66(5): 477–485. Houle S, Ginovart N, et al. (2000). Imaging the serotonin transporter with positron emission tomography: initial human studies with [11C]DAPP and [11C]DASB. Eur J Nucl Med. 27(11): 1719–1722. Hoyer D, Hannon JP, et al. (2002). Molecular, pharmacological and functional diversity of serotonin receptors.” Pharmacol Biochem Behav. 71(4): 533–554. Hranilovic D, Stefulj J, et al. (2004). Serotonin transporter promoter and intron 2 polymorphisms: relationship between allelic variants and gene expression. Biol Psychiatry. 55(11): 1090–1094. Hu XZ, Lipsky RH, et al. (2006). Serotonin transporter promoter gain-of-function genotypes are linked to obsessive-compulsive disorder. Am J Hum Genet. 78(5): 815–826. Huang WS, Huang SY, et al. (2013). PET imaging of the brain serotonin transporters (SERT) with N,N-dimethyl-2-(2-amino-4-[18F] fluorophenylthio)benzylamine (4-[18F]-ADAM) in humans: a preliminary study. Eur J Nucl Med Mol I. 40(1): 115–124. Huang YY, Huang WS, et al. (2012). Synthesis and comparison of 4-[18F]F-ADAM, 2-[18F]F-ADAM, N-Desmethyl-4-[18F] F-ADAM and [18F]F-AFM as serotonin transporter imaging agents. Appl Radiat Isotopes. 70(10): 2298–2307. Huang Y, Hwang DR, et al. (2002). Comparative evaluation in nonhuman primates of five PET radiotracers for imaging the serotonin transporters: [11C]McN 5652, [11C]ADAM, [11C]
DASB, [11C]DAPA, and [11C]AFM. J Cereb Blood Flow Metab. 22(11): 1377–1398. Huang Y, Narendran R, et al. (2004). A PET imaging agent with fast kinetics: synthesis and in vivo evaluation of the serotonin transporter ligand [11C]2-[2-dimethylaminomethylphenylthio)]5-fluorophenylamine ([11C]AFA). Nucl Med Biol. 31(6): 727–738. Ichise M, Liow JS, et al. (2003). Linearized reference tissue parametric imaging methods: application to [11C]DASB positron emission tomography studies of the serotonin transporter in human brain. J Cereb Blood Flow Metab. 23(9): 1096–1112. Innis RB, Cunningham VJ, et al. (2007). Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab. 27(9): 1533–1539. Jarkas N, McConathy J, et al. (2005). Synthesis and characterization of EADAM: a selective radioligand for mapping the brain serotonin transporters by positron emission tomography. Nucl Med Biol. 32(1): 75–86. Jarkas N, Votaw JR, et al. (2005). Carbon-11 HOMADAM: a novel PET radiotracer for imaging serotonin transporters. Nucl Med Biol. 32(3): 211–224. Jones KL, Smith RM, et al. (2010). Combined effect of maternal serotonin transporter genotype and prenatal stress in modulating offspring social interaction in mice. Int J Devel Neurosci. 28(6): 529–536. Jones T, Rabiner EA. (2012). The development, past achievements, and future directions of brain PET. J Cereb Blood Flow Metab. 32(7): 1426–1454. Jovanovic H, Lundberg J, et al. (2008). Sex differences in the serotonin 1A receptor and serotonin transporter binding in the human brain measured by PET. NeuroImage. 39(3): 1408–1419. Kalbitzer J, Erritzoe D, et al. (2010). Seasonal changes in brain serotonin transporter binding in short serotonin transporter linked polymorphic region-allele carriers but not in long-allele homozygotes. Biol Psychiatry. 67(11): 1033–1039. Karg K, Burmeister M, et al. (2011). The serotonin transporter promoter variant (serotoninTLPR), stress, and depression meta-analysis revisited: evidence of genetic moderation. Arch Gen Psychiatry. 68(5): 444–454. Kessler RC, McGonagle KA, et al. (1993). Sex and depression in the National Comorbidity Survey: I. Lifetime prevalence, chronicity and recurrence. J Affect Disord. 29(2–3): 85–96. Kim JS, Ichise M, et al. (2006). PET imaging of serotonin transporters with [11C]DASB: test-retest reproducibility using a multilinear reference tissue parametric imaging method. J Nucl Med. 47(2): 208–214. Kish SJ, Furukawa Y, et al. (2005). Regional distribution of serotonin transporter protein in postmortem human brain: is the cerebellum a SERT-free brain region? Nucl Med Biol. 32(2): 123–128. Klein AB, Trajkovska V, et al. (2010). Cerebral serotonin2A receptor and serotonin transporter binding in humans are not affected by the val66met BDNF polymorphism status or blood BDNF levels. J Cereb Blood Flow Metab. 30(11): e1–7. Kotov R, Gamez W, et al. (2010). Linking “big” personality traits to anxiety, depressive, and substance use disorders: a meta-analysis. Psychol Bull. 136(5): 768–821. Krueger RF, Markon KE. (2006). Understanding psychopathology: melding behavior genetics, personality, and quantitative psychology to develop an empirically based model. Curr Dir Psychol Sci. 15(3): 113–117. Kunugi H, Hattori M, et al. (1997). Serotonin transporter gene polymorphisms: ethnic difference and possible association with bipolar affective disorder. Mol Psychiatry. 2(6): 457–462. Laje G, Cannon DM, et al. (2010). Genetic variation in HTR2A influences serotonin transporter binding potential as measured using PET and [11C]DASB. Int J Neuropsychopharmacol. 13(6): 715–724. Lammertsma AA, Hume SP. (1996). Simplified reference tissue model for PET receptor studies. Neuroimage. 4(3 Pt 1): 153–158.
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
45
Lanzenberger R, Kranz GS, et al. (2012). Prediction of SSRI treatment response in major depression based on serotonin transporter interplay between median raphe nucleus and projection areas. NeuroImage. 63(2): 874–881. Laruelle M, Maloteaux JM. (1989). Regional distribution of serotonergic pre- and postsynaptic markers in human brain. Acta Psychiatr Scand Suppl. 350: 56–59. Laruelle M, Vanisberg MA, et al. (1988). Regional and subcellular localization in human brain of [3H]paroxetine binding, a marker of serotonin uptake sites. Biol Psychiatry. 24(3): 299–309. Lasky-Su JA, Faraone SV, et al. (2005). Meta-analysis of the association between two polymorphisms in the serotonin transporter gene and affective disorders. Am J Med Genet. 133B(1): 110–115. Lee M, Bailer UF, et al. (2005). Relationship of a serotonin transporter functional polymorphism to serotonin1A receptor binding in healthy women. Mol Psychiatry. 10(8): 715–716. Lesch KP, Balling U, et al. (1994). Organization of the human serotonin transporter gene. J Neural Trans Gen. 95(2): 157–162. Lesch KP, Bengel D, et al. (1996). Association of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory region. Science. 274(5292): 1527–1531. Lira A, Zhou M, et al. (2003). Altered depression-related behaviors and functional changes in the dorsal raphe nucleus of serotonin transporter-deficient mice. Biol Psychiatry. 54(10): 960–971. Logan J, Fowler JS, et al. (1990). Graphical analysis of reversible radioligand binding from time-activity measurements applied to [N-11C-methyl]-(-)-cocaine PET studies in human subjects. J Cereb Blood Flow Metab. 10(5): 740–747. Logan J, Fowler JS, et al. (1996). Distribution volume ratios without blood sampling from graphical analysis of PET data. J Cereb Blood Flow Metab. 16(5): 834–840. Lothe A, Boni C, et al. (2009). Association between triallelic polymorphism of the serotonin transporter and [18F]MPPF binding potential at serotonin1A receptors in healthy subjects. NeuroImage. 47(2): 482–492. Lundberg J, Borg J, et al. (2007). A PET study on regional coexpression of serotonin1A receptors and serotoninT in the human brain. Psychopharmacology. 195(3): 425–433. Lundberg J, Halldin C, et al. (2006). Measurement of serotonin transporter binding with PET and [11C]MADAM: a test-retest reproducibility study. Synapse. 60(3): 256–263. Lundberg J, Odano I, et al. (2005). Quantification of 11C-MADAM binding to the serotonin transporter in the human brain. J Nucl Med. 46(9): 1505–1515. Mann JJ, Currier D. (2007). A review of prospective studies of biologic predictors of suicidal behavior in mood disorders. Arch Suicide Res. 11(1): 3–16. Maron E, Toru I, et al. (2011). Gender differences in brain serotonin transporter availability in panic disorder. J Psychopharmacol. 25(7): 952–959. Matsumoto R, Ichise M, et al. (2010). Reduced serotonin transporter binding in the insular cortex in patients with obsessive-compulsive disorder: a [11C]DASB PET study. NeuroImage. 49(1): 121–126. McCann UD, Szabo Z, et al. (1998). Positron emission tomographic evidence of toxic effect of MDMA (“Ecstasy”) on brain serotonin neurons in human beings. Lancet. 352(9138): 1433–1437. McCann UD, Szabo Z, et al. (2005). Quantitative PET studies of the serotonin transporter in MDMA users and controls using [11C]McN5652 and [11C]DASB. Neuropsychopharmacology. 30(9): 1741–1750. Meltzer CC, Smith G, et al. (1998). Serotonin in aging, late-life depression, and Alzheimer’s disease: the emerging role of functional imagi ng.”Neuropsychopharmacology. 18(6): 407–430. Meyer JH, Houle S, et al. (2004). Brain serotonin transporter binding potential measured with carbon 11-labeled DASB positron emission tomography: effects of major depressive episodes and severity of dysfunctional attitudes. Arch Gen Psychiatry. 61(12): 1271–1279.
Meyer JH, Wilson AA, et al. (2001). Occupancy of serotonin transporters by paroxetine and citalopram during treatment of depression: a [(11) C]DASB PET imaging study. Am J Psychiatry. 158(11): 1843–1849. Miller JM, Hesselgrave N, et al. (2013a). Brain serotonin 1A receptor binding as a predictor of treatment outcome in major depressive disorder. Biol Psychiatry. 74(10): 760–767. Miller JM, Hesselgrave N, et al. (2013b). Positron emission tomography quantification of serotonin transporter in suicide attempters with major depressive disorder. Biological psychiatry 74(4): 287–295. Miller JM, Oquendo MA, et al. (2008). Serotonin transporter binding as a possible predictor of one-year remission in major depressive disorder. J Psychiat Res. 42(14): 1137–1144. Moses-Kolko EL, Price JC, et al. (2011). Age, sex, and reproductive hormone effects on brain serotonin-1A and serotonin-2A receptor binding in a healthy population. Neuropsychopharmacology. 36(13): 2729–2740. Munafo MR, Brown SM, et al. (2008). Serotonin transporter (serotoninTLPR) genotype and amygdala activation: a meta-analysis. Biol Psychiatry. 63(9): 852–857. Munafo MR, Durrant C, et al. (2009). Gene X environment interactions at the serotonin transporter locus. Biol Psychiatry. 65(3): 211–219. Munafo MR, Freimer NB, et al. (2009). serotoninTLPR genotype and anxiety-related personality traits: a meta-analysis and new data. Am J Med Genet Part B. 150B(2): 271–281. Murphy DL, Fox MA, et al. (2008). How the serotonin story is being rewritten by new gene-based discoveries principally related to SLC6A4, the serotonin transporter gene, which functions to influence all cellular serotonin systems. Neuropharmacology. 55(6): 932–960. Murthy NV, Selvaraj S, et al. (2010). Serotonin transporter polymorphisms (SLC6A4 insertion/deletion and rs25531) do not affect the availability of serotoninT to [11C] DASB binding in the living human brain. NeuroImage. 52(1): 50–54. Nakamura M, Ueno S, et al. (2000). The human serotonin transporter gene linked polymorphism (serotoninTLPR) shows ten novel allelic variants. Mol Psychiatry. 5(1): 32–38. Nemeroff CB, Vale WW. (2005). The neurobiology of depression: inroads to treatment and new drug discovery. J Clin Psychiatry. 66(Suppl 7): 5–13. Ogden RT, Ojha A, et al. (2007). In vivo quantification of serotonin transporters using [(11)C]DASB and positron emission tomography in humans: modeling considerations. J Cereb Blood Flow Metab. 27(1): 205–217. Oler JA, Fox AS, et al. (2009). Serotonin transporter availability in the amygdala and bed nucleus of the stria terminalis predicts anxious temperament and brain glucose metabolic activity. J Neuroscience. 29(32): 9961–9966. Oquendo MA, Hastings RS, et al. (2007). Brain serotonin transporter binding in depressed patients with bipolar disorder using positron emission tomography. Arch Gen Psychiatry. 64(2): 201–208. Parsey RV, Hastings RS, et al. (2006a). Effect of a triallelic functional polymorphism of the serotonin-transporter-linked promoter region on expression of serotonin transporter in the human brain. Am J Psychiatry. 163(1): 48–51. Parsey RV, Hastings RS, et al. (2006b). Lower serotonin transporter binding potential in the human brain during major depressive episodes. Am J Psychiatry. 163(1): 52–58. Parsey RV, Kegeles LS, et al. (2000). In vivo quantification of brain serotonin transporters in humans using [11C]McN 5652. J Nucl Med. 41(9): 1465–1477. Parsey RV, Kent JM, et al. (2006). Acute occupancy of brain serotonin transporter by sertraline as measured by [11C]DASB and positron emission tomography.” Biol Psychiatry. 59(9): 821–828. Parsey RV, Mann JJ. (2003). Applications of positron emission tomography in psychiatry. Semin Nucl Med. 33(2): 129–135.
4 6 Part I : I maging G enetics and N eurochemistry
Parsey RV, Ogden RT, et al. (2010). Higher serotonin 1A binding in a second major depression cohort: modeling and reference region considerations. Biol Psychiatry. 68(2): 170–178. Parsey RV, Olvet DM, et al. (2006). Higher serotonin1A receptor binding potential during a major depressive episode predicts poor treatment response: preliminary data from a naturalistic study. Neuropsychopharmacology. 31(8): 1745–1749. Parsey RV, Oquendo MA, et al. (2002). Effects of sex, age, and aggressive traits in man on brain serotonin serotonin1A receptor binding potential measured by PET using [C-11]WAY-100635. Brain Res. 954(2): 173–182. Patkar AA, Berrettini WH, et al. (2004). Relationship between serotonin transporter gene polymorphisms and platelet serotonin transporter sites among African-American cocaine-dependent individuals and healthy volunteers. Psychiat Genet. 14(1): 25–32. Pergamin-Hight L, Bakermans-Kranenburg MJ, et al. (2012). Variations in the promoter region of the serotonin transporter gene and biased attention for emotional information: a meta-analysis. Biol Psychiatry. 71(4): 373–379. Peroutka SJ. (1994). Molecular biology of serotonin (serotonin) receptors. Synapse. 18(3): 241–260. Peroutka SJ, Howell TA. (1994). The molecular evolution of G protein-coupled receptors: focus on 5-hydroxytryptamine receptors. Neuropharmacology. 33(3–4): 319–324. Pezawas L, Meyer-Lindenberg A, et al. (2005). serotoninTLPR polymorphism impacts human cingulate-amygdala interactions: a genetic susceptibility mechanism for depression. Nat Neurosci. 8(6): 828–834. Plomin R, Haworth CM, et al. (2009). Common disorders are quantitative traits. Nat Rev Genet. 10(12): 872–878. Politis M, Wu K, et al. (2010a). Depressive symptoms in PD correlate with higher serotoninT binding in raphe and limbic structures. Neurology. 75(21): 1920–1927. Politis M, Wu K, et al. (2010b). Staging of serotonergic dysfunction in Parkinson’s disease: an in vivo 11C-DASB PET study. Neurobiol Disease. 40(1): 216–221. Pollock BG, Ferrell RE, et al. (2000). Allelic variation in the serotonin transporter promoter affects onset of paroxetine treatment response in late-life depression. Neuropsychopharmacology. 23(5): 587–590. Praschak-Rieder N, Kennedy J, et al. (2007). Novel serotoninTLPR allele associates with higher serotonin transporter binding in putamen: a [(11)C] DASB positron emission tomography study. Biol Psychiatry. 62(4): 327–331. Praschak-Rieder N, Willeit M. (2012). Imaging of seasonal affective disorder and seasonality effects on serotonin and dopamine function in the human brain. Curr Topics Behav Neurosci. 11: 149–167. Praschak-Rieder N, Willeit M, et al. (2008). Seasonal variation in human brain serotonin transporter binding. Arch Gen Psychiatry. 65(9): 1072–1078. Ramamoorthy S, Bauman AL, et al. (1993). Antidepressant- and cocaine-sensitive human serotonin transporter: molecular cloning, expression, and chromosomal localization. Proc Natl Acad Sci U S A. 90(6): 2542–2546. Reimold M, Batra A, et al. (2008). Anxiety is associated with reduced central serotonin transporter availability in unmedicated patients with unipolar major depression: a [11C]DASB PET study. Mol Psychiatry. 13(6): 606–613, 557. Reimold M, Smolka MN, et al. (2007a). Midbrain serotonin transporter binding potential measured with [11C]DASB is affected by serotonin transporter genotype. J Neural Trans. 114(5): 635–639. Reimold M, Smolka MN, et al. (2007b). Reduced availability of serotonin transporters in obsessive-compulsive disorder correlates with symptom severity—a [11C]DASB PET study. J Neural Trans. 114(12): 1603–1609. Reivich M, Amsterdam JD, et al. (2004). PET brain imaging with [11C](+)McN5652 shows increased serotonin transporter availability in major depression. J Affect Disord. 82(2): 321–327.
Risch N, Herrell R, et al. (2009). Interaction between the serotonin transporter gene (serotoninTLPR), stressful life events, and risk of depression: a meta-analysis. JAMA. 301(23): 2462–2471. Rosa-Neto P, Diksic M, et al. (2004). Measurement of brain regional alpha-[11C]methyl-L-tryptophan trapping as a measure of serotonin synthesis in medication-free patients with major depression. Arch Gen Psychiatry 61(6): 556–563. Rosel P, Arranz B, et al. (1997). High affinity [3H]imipramine and [3H]paroxetine binding sites in suicide brains. J Neural Trans. 104(8–9): 921–929. Ruhe HG, Ooteman W, et al. (2009). Serotonin transporter gene promoter polymorphisms modify the association between paroxetine serotonin transporter occupancy and clinical response in major depressive disorder. Pharmacogenet Genomics. 19(1): 67–76. Sakai K, Nakamura M, et al. (2002). The silencer activity of the novel human serotonin transporter linked polymorphic regions. Neurosci Lett. 327(1): 13–16. Sargent PA, Kjaer KH, et al. (2000). Brain serotonin1A receptor binding measured by positron emission tomography with [11C] WAY-100635: effects of depression and antidepressant treatment. Arch Gen Psychiatry. 57(2): 174–180. Saulin A, Savli M, et al. (2012). Serotonin and molecular neuroimaging in humans using PET. Amino Acids. 42(6): 2039–2057. Savitz JB, Drevets WC. (2012). Neuroreceptor imaging in depression. Neurobiol Disease. 52, 49–65. Savli M, Bauer A, et al. (2012). Normative database of the serotonergic system in healthy subjects using multi-tracer PET. NeuroImage. 63(1): 447–459. Sen S, Burmeister M, et al. (2004). Meta-analysis of the association between a serotonin transporter promoter polymorphism (serotoninTLPR) and anxiety-related personality traits. Am J Med Genet Part B. 127B(1): 85–89. Serretti A, Artioli P, et al. (2005). Pharmacogenetics in the treatment of depression: pharmacodynamic studies. Pharmacogenet Genomics. 15(2): 61–67. Serretti A, Benedetti F, et al. (2005). The influence of serotonin transporter promoter polymorphism (SERTPR) and other polymorphisms of the serotonin pathway on the efficacy of antidepressant treatments. Prog Neuropsychopharm Biol Psychiatry.29(6): 1074–1084. Serretti A, Kato M, et al. (2007). Meta-analysis of serotonin transporter gene promoter polymorphism (serotoninTLPR) association with selective serotonin reuptake inhibitor efficacy in depressed patients. Mol Psychiatry. 12(3): 247–257. Serretti A, Mandelli L, et al. (2007). Serotonin transporter gene influences the time course of improvement of “core” depressive and somatic anxiety symptoms during treatment with SSRIs for recurrent mood disorders. Psychiatry Res. 149(1–3): 185–193. Shioe K, Ichimiya T, et al. (2003). No association between genotype of the promoter region of serotonin transporter gene and serotonin transporter binding in human brain measured by PET. Synapse. 48(4): 184–188. Simpson HB, Lombardo I, et al. (2003). Serotonin transporters in obsessive-compulsive disorder: a positron emission tomography study with [(11)C]McN 5652. Biol Psychiatry. 54(12): 1414–1421. Smeraldi E, Zanardi R, et al. (1998). Polymorphism within the promoter of the serotonin transporter gene and antidepressant efficacy of fluvoxamine. Mol Psychiatry. 3(6): 508–511. Smith DF, Jakobsen S. (2009). Molecular tools for assessing human depression by positron emission tomography. Eur Neuropsychopharmacol. 19(9): 611–628. Stehouwer JS, Goodman MM. (2009). Fluorine-18 radiolabeled PET tracers for imaging monoamine transporters: dopamine, serotonin, and norepinephrine.” PET Clinics. 4(1): 101–128. Stockmeier CA. (2003). Involvement of serotonin in depression: evidence from postmortem and imaging studies of serotonin receptors and the serotonin transporter. J Psychiatric Res. 37(5): 357–373.
cha p ter 3 : G enetic V ariation I m p acting the S erotonin T rans p orter
47
Suehiro M, Scheffel U, et al. (1993a). A PET radiotracer for studying serotonin uptake sites: carbon-11-McN-5652Z. J Nucl Med. 34(1): 120–127. Suehiro M, Scheffel U, et al. (1993b). [11C]McN5652 as a radiotracer for imaging serotonin uptake sites with PET. Life Sciences. 53(11): 883–892. Sugden K, Tichopad A, et al. (2009). Genes within the serotonergic system are differentially expressed in human brain. BMC Neuroscience. 10: 50. Szabo Z, McCann UD, et al. (2002). Comparison of (+)11C-McN5652 and 11C-DASB as serotonin transporter radioligands under various experimental conditions. J Nucl Med. 43(5): 678–692. Szabo Z, Owonikoko T, et al. (2004). Positron emission tomography imaging of the serotonin transporter in subjects with a history of alcoholism. Biol Psychiatry. 55(7): 766–771. Szabo Z, Scheffel U, et al. (1999). Kinetic analysis of [11C]McN5652: a serotonin transporter radioligand. J Cereb Blood Flow Metab. 19(9): 967–981. Szanto K, Gildengers A, et al. (2002). Identification of suicidal ideation and prevention of suicidal behaviour in the elderly. Drugs Aging 19(1): 11–24. Takano A, Arakawa R, et al. (2007). Relationship between neuroticism personality trait and serotonin transporter binding. Biol Psychiatry. 62(6): 588–592. Takano H, Ito H, et al. (2011). Serotonergic neurotransmission in the living human brain: a positron emission tomography study using [(1)(1)C]dasb and [(1)(1)C]WAY100635 in young healthy men. Synapse. 65(7): 624–633. Talbot PS, Laruelle M. (2002). The role of in vivo molecular imaging with PET and SPECT in the elucidation of psychiatric drug action and new drug development. Eur Neuropsychopharmacol. 12(6): 503–511. Terracciano A, Balaci L, et al. (2009). Variants of the serotonin transporter gene and NEO-PI-R neuroticism: no association in the BLSA and SardiNIA samples. Am J Med Genet Part B. 150B(8): 1070–1077. Uher R, Caspi A, et al. (2011). Serotonin transporter gene moderates childhood maltreatment’s effects on persistent but not single-episode depression: replications and implications for resolving inconsistent results. J Affect Disord. 135(1–3): 56–65.
Underwood MD, Kassir SA, et al. (2012). Neuron density and serotonin receptor binding in prefrontal cortex in suicide. Int J Neuropsychopharmacol. 15(4): 435–447. Veltman DJ, Ruhe HG, et al. (2010). Investigating serotonergic function using positron emission tomography: overview and recent findings. Curr Pharmaceut Design. 16(18): 1979–1989. Visser AK, van Waarde A, et al. (2011). Measuring serotonin synthesis: from conventional methods to PET tracers and their (pre)clinical implications. Eur J Nucl Med Mol I. 38(3): 576–591. Voineskos AN, Wilson AA, et al. (2007). Serotonin transporter occupancy of high-dose selective serotonin reuptake inhibitors during major depressive disorder measured with [11C]DASB positron emission tomography. Psychopharmacology. 193(4): 539–545. Wang JL, Parhi AK, et al. (2008). 2-(2’-((Dimethylamino) methyl)-4’-(3-[(18)F]fluoropropoxy)-phenylthio)benzenamine for positron emission tomography imaging of serotonin transporters. Nucl Med Biol. 35(4): 447–458. Wendland JR, Martin BJ, et al. (2006). Simultaneous genotyping of four functional loci of human SLC6A4, with a reappraisal of serotoninTLPR and rs25531. Mol Psychiatry. 11(3): 224–226. Whitaker-Azmitia PM. (2001). Serotonin and brain development: role in human developmental diseases. Brain Res Bull. 56(5): 479–485. Wilson AA, Ginovart N, et al. (2002). In vitro and in vivo characterisation of [11C]-DASB: a probe for in vivo measurements of the serotonin transporter by positron emission tomography. Nucl Med Biol. 29(5): 509–515. Wilson AA, Houle S. (1999). Radiosynthesis of carbon-11 labelled N-methyl-2-(arylthio) benzylamines: potential radiotracers for the serotonin reuptake receptor. J Labelled Compd Rad. 42(13): 1277–1288. Wong DF, Wagner HN, Jr., et al. (1984). Effects of age on dopamine and serotonin receptors measured by positron tomography in the living human brain. Science. 226(4681): 1393–1396. Yamamoto M, Suhara T, et al. (2002). Age-related decline of serotonin transporters in living human brain of healthy males. Life Sciences. 71(7): 751–757. Yamamoto S, Ouchi Y, et al. (2004). Reduction of serotonin transporters of patients with chronic fatigue syndrome. Neuroreport. 15(17): 2571–2574. Zhu Z, Guo N, et al. (2004). The new PET imaging agent [11C]AFE is a selective serotonin transporter ligand with fast brain uptake kinetics. Nucl Med Biol. 31(8): 983–994.
4 8 Part I : I maging G enetics and N eurochemistry
4. IMAGING OF GENE TIC VARIATION IMPACTING DOPAMINE TRANSMISSION PARAME TERS Elsmarieke van de Giessen, Rassil Ghazzaoui, Raj Narendran, and Anissa Abi-Dargham
I N T RO DU C T I O N Dopamine is a neurotransmitter that is important in physiological processes (e.g., reward and motor functions) and plays a role in many psychiatric and neurologic disorders (e.g. schizophrenia, attention deficit hyperactivity disorder [ADHD], substance abuse disorders, and Parkinson’s disease). The majority of dopaminergic neurons in the brain are located in the ventral tegmental area and substantia nigra, projecting to the striatum, the prefrontal cortex, and other regions. Dopamine is released from the axons and the signal is subsequently transmitted by post-synaptic dopamine receptors. Dopamine levels are regulated by proteins involved in dopamine synthesis and degradation, dopamine transporters and dopamine autoreceptors. Dysregulation of any of these components could result in abnormal function and disease. Polymorphisms in the genes encoding for proteins in the dopamine pathway can affect the expression and function of these proteins and increase vulnerability for dopamine-related disorders. Twin studies have indeed shown that heritability of schizophrenia and ADHD is up to 80% (Sullivan et al. 2003; Faraone et al. 2005). Although these disorders are multigenic, part of the heritability could be related to genetic variability in the dopamine pathway or other genes that indirectly influence the dopamine system. In order to better understand the influence of genetic polymorphisms on the dopamine system, it is possible to investigate the relation between these two with association studies. A way to measure the functioning of the dopamine system in vivo in humans is the use of positron emission tomography (PET) or single-photon emission computed tomography (SPECT) imaging. These techniques apply radioactively labeled ligands to measure
receptors and transporters. If a genetic polymorphism directly or indirectly influences expression of a receptor, such as D2, it could yield an association between this genetic polymorphism and the dopamine D2 receptor expression levels. This polymorphism could subsequently play a role in disorders with aberrant dopamine D2 receptor function. The aim of this chapter is to give an overview of the currently available association studies that have investigated the influence of genetic polymorphisms on PET/SPECT measures of the dopamine system. Different parameters have been studied, including dopamine receptors and transporters, dopamine release, synthesis, and degradation capacity. To our knowledge, two previous reviews have been published on this topic (Martinez et al. 2001a; Willeit and Praschak-Rieder 2010). Here we provide an update, and discuss the implications of the findings for dopamine-related disorders and for future investigations in this area. D O PA M I N E D 1 R EC E P T O R AVA I L A BI L I T Y Dopamine signals through G-protein coupled receptors consisting of two subfamilies: the dopamine D1- (D1, D5) and dopamine D2-like receptor subfamily (D2 , D3, D4) (Beaulieu and Gainetdinov 2011). In this paragraph we will review studies of the dopamine D1 receptors (D1R) and associations with genetic polymorphisms. D1R are widely expressed in both the striatum and the cortex and are located post-synaptically (Cortes et al. 1989). They have been implicated in cognitive functioning (Takahashi et al. 2008; Karlsson et al. 2009; Karlsson et al. 2011). 49
In schizophrenia, D1R availability is likely increased in the prefrontal cortex, especially in drug-naïve patients (Abi-Dargham et al. 2002; Abi-Dargham et al. 2012), although not all studies consistently report this observation (Okubo et al. 1997; Karlsson et al. 2002). D1R availability can be examined in humans using PET and the selective D1-dopamine receptor antagonist radioligands [11C]NNC112 or [11C]-SCH23390. There are three studies that have investigated the effect of genetic polymorphisms on D1R availability. The first study examined the relationship between polymorphisms in the gene encoding for catechol-O-methyltransferase (COMT) and D1R availability (Slifstein et al. 2008). The COMT gene is located on chromosome 22q11.2 and a well-known polymorphism in the gene is the functional Val158Met polymorphism. The Met variant catabolizes dopamine at a slower rate than the Val variant (Chen et al. 2004b), which can lead to higher synaptic dopamine levels. COMT plays a major role in the degradation of dopamine in the prefrontal cortex and has been shown to influence dopamine tone in the cortex and cortical functioning (Lachman et al. 1996). In this study, [11C]NNC112 binding in 11 healthy volunteers, who were homozygous for the Val allele, was compared with 17 Met carriers. Results showed higher dopamine D1 receptor binding (range 15%– 60%) in limbic and cortical regions with the COMT genotype Val/Val, when compared with either the Met/Val or the Met/Met COMT genotype. There were no significant differences found for the striatal regions. These results indicate that the Val158Met polymorphism is a genetic factor accounting for some of the variability across individuals in dopamine D1 receptor availability and provide further support for the role of COMT in regulating dopamine levels in cortical and limbic regions of the brain. Another study focused on autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE) and D1R binding (Fedi et al. 2008). This condition is characterized by clusters of brief motor seizures of frontal origin that take place during non-REM sleep. Mutations of genes for the neuronal nicotine acetylcholine (nACH) receptor subunits α4 (CHRNA4), β2 (CHRNB2), and α2 (CHRNA2) have been found in several families suffering from ADNFLE. Activation of pre-synaptic nicotinic receptors augments the release of dopamine in the striatum and the prefrontal regions and could therefore affect markers of the dopaminergic system. Twelve affected subjects with the α4-Ser248Phe mutation and 19 healthy volunteers were scanned using PET and [11C]SCH23390 (Fedi et al. 2008). The subjects with the α4-Ser248Phe mutation had reduced D1 receptor binding in the striatum (−10%), whereas there
were no significant group differences for cortical regions. The authors hypothesized that the reduced striatal D1 receptor binding would be a result of elevated extracellular dopamine levels, which would play a role in the pathophysiologic mechanism of ADNFLE. The third study focused on the autosomal dominant Huntington’s disease (HD). HD is a disorder characterized by severe neuronal loss in the caudate nucleus and putamen, while cortical abnormalities are less evident. The gene for the disorder has been found to be in the short arm of chromosome 4p16. Using [11C]SCH23390 and PET, D1R availability in five male controls was compared to five patients with a clinical diagnosis of HD and one asymptomatic gene carrier (Sedvall et al. 1994). The five patients showed a 50% reduction in the putamen volume. D1R density in the putamen was also reduced by 50% in the patients, leading to a total reduction of 75% in D1R. Findings were comparable for the caudate. Patients also tended to have reduced D1R availability in the frontal cortex. The asymptomatic carrier had volume and binding values for the putamen and caudate in the lower range of the controls. Therefore, the authors suggest that [11C]SCH23390 PET could be used to detect early brain degeneration in HD. In summary, these studies show that different genetic polymorphisms can influence D1R availability. The COMT Val158Met polymorphism is a common variant; thus, the effect of this polymorphism on D1R availability is clinically relevant and could play a role in disease risk. Indeed, the COMT polymorphism has been associated with alcohol dependence (Tiihonen et al. 1999). The Val allele has been associated with increased risk for schizophrenia (Egan et al. 2001) and, as described earlier, with increased D1R availability (Slifstein et al. 2008). This is in line with the findings that schizophrenia is likely associated with altered D1R availability (Okubo et al. 1997; Abi-Dargham et al. 2002; Karlsson et al. 2002; Abi-Dargham et al. 2012). The studies on ADNFLE and HD are based on single autosomal dominant genetic polymorphisms that cause the disease, and the results are therefore primarily important for these specific diseases. D O PA M I N E D 2 A N D D 3 R EC E P T O R AVA I L A BI L I T Y Dopamine D2 receptors (D2R) are abundantly expressed in the whole striatum, and at lower concentrations in the neocortex. Mesolimbic D2R in the striatum play an important role in reward-related behavior and have therefore been implicated in addiction and other psychiatric disorders
5 0 Part I : I maging G enetics and N eurochemistry
(Koob 1996). Imaging studies have shown that D2R availability is significantly decreased in several substance abuse disorders (e.g., alcohol [Volkow et al. 1996], methamphetamine [Volkow et al. 2001], and cocaine abusers [Volkow et al. 1997]) and obesity (Wang et al. 2001; de Weijer et al. 2011). D2R are also a major target of antipsychotics, and a meta-analysis shows that there is a modest increase in striatal D2R availability in schizophrenia (Weinberger and Laruelle 2002). A number of factors, including age (Rinne et al. 1993; Volkow et al. 1998) and gender (Pohjalainen et al. 1998b), have been associated with variations in D2 receptor density. However, the extent to which these factors may modify receptor density cannot explain the twoto threefold variability observed among healthy volunteers (Farde et al. 1995). Genetic factors might be behind the individual differences in D2R density and could affect the predisposition to substance abuse and other psychiatric disorders, in which D2R are involved. The presently available PET and SPECT imaging ligands bind to both D2 and D3 receptors. The ligands that are described in this section ([11C]raclopride and [11C] FLB457 for PET and [123I]IBZM for SPECT) are antagonists with similar affinities for the D2 and D3 receptors (Stark et al. 2007). [11C]raclopride and [123I]IBZM only provide information for the striatum and [11C]FLB457 only for extrastriatal regions due to its high affinity. Because there are significantly more D2 than D3 receptors in the cortex and in the striatum (in particular in the dorsal striatum, where the number of D3 receptors is negligible) (Seeman et al. 2006), we will describe the imaging findings as if they are reflecting D2 receptor binding, although it should be kept in mind that in the ventral striatum they partly reflect D3 receptor binding. DOPA MINE D2 RECEP T OR GENE (DRD2)
An obvious candidate gene that could influence D2R expression is the dopamine D2 receptor gene itself (DRD2). There are five different polymorphisms related to the DRD2 gene that have been assessed in relation to D2R binding: Taq1A, Taq1B, −141C Ins/Del, C957T, and rs1076560 (see Table 4.1). The Taq1A variant is a restriction fragment length polymorphism (RFLP) that is located in the 3' flanking region, 10kb downstream from the DRD2 gene (Grandy et al. 1989). It was later discovered that this variant is located in the coding region of the ANKK1 gene and leads to an amino acid substitution (Glu713Lys) (Neville et al. 2004). The A1 allele of the Taq1A variant has been associated with psychiatric disorders that involve impaired impulse
control, including substance abuse disorders, gambling, and binge eating disorder (Blum et al. 1990; Blum et al. 1991; Noble et al. 1993; Blum et al. 1995; Blum et al. 1996; Comings et al. 1996b; Persico et al. 1996; Lawford et al. 2000; Chen et al. 2004a; Messas et al. 2005; Davis et al. 2012). The association between alcohol dependence and the A1 allele has also been confirmed in a meta-analysis (Smith et al. 2008). There are three studies investigating a relation between the Taq1A variant and striatal D2R availability (Laruelle et al. 1998; Pohjalainen et al. 1998a; Jonsson et al. 1999a) and one focusing on extrastriatal D2R availability (Hirvonen et al. 2009b). Two of those (Pohjalainen et al. 1998a; Jonsson et al. 1999a) included only healthy subjects and used [11C]raclopride and PET. Pohjalainen et al. recruited 54 subjects and investigated the total number of D2R (Bmax), the dissociation constant (K D), which is inversely related to affinity, and the D2R binding potential (Bmax/K D) (Pohjalainen et al. 1998a). They did two [11C] raclopride PET scans per subject, one with high and one with low specific activity in order to determine these three different variables. The Taq1A A1 allele carriers had 12% lower binding potential than the non-carriers, but they did not have a difference in K D. These findings suggest that the A1 allele carriers have less D2R in the striatum, without structural changes of the receptor protein. The lower striatal D2R binding in A1 allele carriers was replicated in a sample of 56 healthy controls with [11C]raclopride PET (Jonsson et al. 1999a). The third study on striatal D2R binding did not find an association with the A1 allele (Laruelle et al. 1998). Differences in the samples could provide explanations for this discrepancy. The latter sample (Laruelle et al. 1998) consisted of 70 subjects; however, 23 of these subjects were patients with schizophrenia who showed an opposite effect of the A1 allele compared to the healthy controls (see Table 4.1). Furthermore, the applied imaging technique was [123I]IBZM SPECT, whereas the other studies used [11C]raclopride PET. Comings et al. (1999) suggested that molecular heterosis of the Taq1A variant (i.e., heterozygotes expressing a specific trait more or less than either homozygotes) could be an explanation for the discrepancies. Ethnic differences between samples could also play a role. Nevertheless, postmortem brain studies have consistently shown that striatal D2R expression is lower in Taq1A A1 carriers (Noble et al. 1991; Thompson et al. 1997; Ritchie and Noble 2003). Therefore, this seems to be a robust finding. When combining the information that alcohol dependence is associated with lower striatal D2R binding (Volkow et al. 1996) and with the Taq1A A1 allele (Smith et al. 2008) and that the Taq1A A1 is associated with lower striatal D2R binding, it seems plausible that the
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
51
TABLE 4.1
STUDIES ON DOPAMINE D 2 AND D 3 RECEPTOR AVAILABILIT Y AND POLYMORPHISMS RELATED TO THE DRD2 GENE Radioligand
N
Region
Result (percentage difference in BP)
Pohjalainen et al. (1998)
[11C]raclopride
54 HC
striatum
A1 carriers < A2/A2 −12%
Jonsson et al. (1999)
[ C]raclopride
56 HC
striatum
A1 carriers < A2/A2 −11%
Laruelle et al. (1998)
[
47 HC +
striatum
A1 carriers = A2/A2 −0.05%
Taq1A polymorphism
11
123
I]IBZM
23 SCZ
HC −9% SCZ +18%
[11C]FLB457
38 HC
extrastriatal
trend for A1 carriers > A2/A2
Jonsson et al. (1999)
[11C]raclopride
56 HC
striatum
B1 carriers < B2/B2 −11%
Laruelle et al. (1998)
[
47 HC +
striatum
B1 carriers = B2/B2 0%
Hirvonen et al. (2009) Taq1B polymorphism
123
I]IBZM
23 SCZ
HC −12% SCZ +40%
-141C Ins/Del polymorphism Jonsson et al. (1999)
[11C]raclopride
56 HC
striatum
Del carrier > Ins/Ins +12%
Pohjalainen et al. (1999)
[ C]raclopride
52 HC
striatum
Del carriers = Ins/Ins
Hirvonen et al. (2004)
[11C]raclopride
45 HC
striatum
CC < CT < TT CT + 3%, TT + 22%
Hirvonen et al. (2009a)
[11C]raclopride
45 HC
striatum
KD: CC > CT > TT
11
C957T polymorphism
Bmax: no difference Hirvonen et al. (2009b)
[11C]FLB457
38 HC
extrastriatal
CC > CT > TT
[123I]IBZM
37 HC
right putamen
T carriers < G/G
rs1076560 polymorphism Bertolino et al. (2010)
BP = binding potential; HC = healthy control; SCZ = schizophrenics
Taq1A allele is a risk factor for alcohol abuse, possibly due to its effect on D2R expression. The study on extrastriatal D2R binding and the Taq1A variant included 38 healthy male subjects who got an [11C]FLB457 PET scan (Hirvonen et al. 2009a). The A1 allele carriers tended to have higher extrastriatal D2R binding, although this was statistically not significant. There was no regional specificity for this trend. The trend is opposite to the findings for the striatum and may be related to the distinct functional roles of dopamine and D2R in the cortex versus the striatum. The Taq1B variant is located before the sequence encoding the transmembrane 1 part of the D2R. Taq1B is found to be in linkage disequilibrium with the Taq1A variant (Hauge et al. 1991; Ritchie and Noble 2003). Therefore, it is not surprising that there are associations between Taq1B
and substance abuse disorders (psychostimulant substance abuse [Persico et al. 1996], alcohol dependence [Blum et al. 1993], and polysubstance abuse [Smith et al. 1992]), which were also reported for Taq1A. Two of the studies that assessed the Taq1A variant and D2R binding also determined Taq1B genotype (Laruelle et al. 1998; Jonsson et al. 1999a). The study using [11C]raclopride PET in healthy subjects reported that carriers of the B1 allele had lower striatal D2R binding than B2 homozygotes (Jonsson et al. 1999a). However, the study using [123I]IBZM SPECT in healthy subjects and patients with schizophrenia did not find an association between the Taq1B genotype and D2R binding (Laruelle et al. 1998). As previously discussed, these contradicting results could be due to differences in imaging methods, inclusion of patients with schizophrenia versus only healthy subjects, ethnicity of the subject
5 2 Part I : I maging G enetics and N eurochemistry
sample, and lack of power. In particular, the discrepancy in results between the patients and healthy subjects might be an important factor (see Table 4.1). Replication will be necessary to determine whether there is a valid association between the Taq1B allele and D2R binding. A polymorphism in the DRD2 gene that is not in linkage disequilibrium with the Taq1A and Taq1B variants is −141C Ins/Del (Arinami et al. 1997). This is a functional polymorphism located in the 5' promoter region of the gene. The deletion allele leads to lower D2R expression in vitro (Arinami et al. 1997). The insertion allele has repeatedly been associated with schizophrenia (Arinami et al. 1997; Ohara et al. 1998; Jonsson et al. 1999b; Lafuente et al. 2008; Cordeiro et al. 2009), although not consistently (Stober et al. 1998; Breen et al. 1999; Tallerico et al. 1999; Parsons et al. 2007). The previously described study by Jonsson et al. (1999a), with healthy volunteers and [11C] raclopride PET, found that Del allele carriers had higher striatal D2R binding than Ins homozygotes, which contrasts with the in vitro finding. Pohjalainen et al. (1999) genotyped the −141C Ins/Del polymorphism in the same sample of healthy volunteers and [11C]raclopride PET that was described previously for the Taq1A allele (Pohjalainen et al. 1998a). They found no association between this polymorphism and striatal D2R binding or with K D. Because the results are inconsistent across studies, it is unfortunately not possible yet to draw reliable conclusions about the association between the −141C Ins/Del polymorphism and striatal D2R binding based on the two studies thus far. One previously mentioned study on extrastriatal D2R (Hirvonen et al. 2009b) reported no association with the −141C Ins/Del polymorphism. The fourth DRD2 polymorphism is C957T. This polymorphism is located in exon 7, and the T allele leads to altered mRNA folding and lower D2R expression in vitro (Duan et al. 2003). C957T is in strong linkage disequilibrium with both the Taq1A and the −141C Ins/ Del polymorphism (Duan et al. 2003). These polymorphisms probably interact with each other and might have additive effects. C957T, like −141C Ins/Del, has been associated with schizophrenia (Hoenicka et al. 2006; Monakhov et al. 2008). In the same samples that were previously studied for the Taq1A and −141C Ins/Del polymorphisms (Pohjalainen et al. 1998a; Hirvonen et al. 2009b), Hirvonen et al. (Hirvonen et al. 2004; Hirvonen et al. 2009a; Hirvonen et al. 2009b) assessed whether there would be an association between D2R binding and the C957T polymorphism. They found that the striatal D2R binding potential was highest in the T/T, followed by the C/T group, and lowest in the C/C group (Hirvonen et al.
2004), whereas in the thalamus and cortex this was the opposite (CC > C/T > TT) (Hirvonen et al. 2009b). The opposite findings for striatal and extrastriatal regions are in line with what was previously reported in this sample for the Taq1A variant (Pohjalainen et al. 1998a; Hirvonen et al. 2009b), probably because Taq1A and C957T are in linkage disequilibrium with each other. The striatal differences in binding potential were explainable by group differences in striatal K D, which was highest in C/C, intermediate in C/T and lowest in T/T, whereas Bmax was not different between groups (Hirvonen et al. 2009a). This suggests that the C957T polymorphism affects D2R affinity but not receptor density. At last, one study investigated the intronic SNP rs1076560, for which the T allele has been associated with decreased expression of the DRD2 short splice variant (expressed mainly pre-synaptically) relative to DRD2 long variant (post-synaptic) in the striatum and prefrontal cortex (Zhang et al. 2007). In a sample of 37 healthy subjects that were scanned with [123I]IBZM SPECT, it was found that T allele carriers had lower striatal D2R binding than the G/G homozygotes (Bertolino et al. 2010), suggesting that the T allele leads to overall lower expression of the D2R. The SNP also predicted a correlation between D2R signaling and prefrontal cortex activity during a working memory task, which could indicate that it might indirectly affect cognitive function. CAT ECHOL- O - ME T H Y L T R A N S F E R A S E G E N E (C O M T )
The COMT Val158Met polymorphism that has been described in the dopamine D1 receptor section, has also been studied in relation to D2R availability, both in healthy subjects (Hirvonen et al. 2010) and in subjects with 22q11 deletion syndrome (Boot et al. 2011). The healthy subject sample consisted of 38 and 45 Finnish healthy subjects, who were scanned with PET and the D2/D3R radioligands [11C]FLB457 and [11C]raclopride, respectively, to assess both cortical and striatal D2R availability (Hirvonen et al. 2010). The COMT Val158Met genotype has been involved in prefrontal cortex function and cognitive processing (Egan et al. 2001; Goldberg et al. 2003). Furthermore, COMT is the major component for synaptic dopamine clearance in the prefrontal cortex, whereas the dopamine transporter is primary for synaptic re-uptake of dopamine in the striatum. Therefore, the hypothesis was that only an effect of the COMT Val158Met polymorphism on cortical but not striatal D2R availability would be shown. However, there were no significant associations between the Val158Met
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
53
genotype and D2R binding (BPND), either in the cortical regions or in the striatum (striatal binding 2% lower in Met carriers than Val/Val homozygotes) (Hirvonen et al. 2010). Thus, the original hypothesis of a link between the COMT Val158Met genotype and baseline D2 receptor availability in the cortex was not supported. Patients with 22q11 deletion syndrome lack part of the 22q11 region in one copy of chromosome 22. The COMT gene is located in this region and therefore these patients have only one copy of the gene. Subsequently, these patients have decreased COMT expression, which is thought to result in abnormal dopamine function. Boot et al. (2011) investigated whether the COMT Val158Met polymorphism would affect striatal D2R availability in a sample of 15 patients with 22q11 deletion syndrome, who were scanned with [123I]IBZM and SPECT. Because all patients carry only one allele, Met hemizygotes were compared to Val hemizygotes. The Met hemizygotes had 17% lower D2R binding than the Val hemizygotes. This could possibly be explained by the fact that the Met allele leads to lower COMT activity, resulting in higher synaptic dopamine levels, which may subsequently lead to a down-regulation of D2R. Because these patients carry only one copy of the COMT gene, the effect of the polymorphism may be relatively large compared to controls, which may explain the association between the COMT Val158Met polymorphism in this patient group, which was not previously observed in healthy subjects (Hirvonen et al. 2010). OTHER GENES
Polymorphisms in genes that are not directly related to the dopamine system have been investigated in relation to D2R availability, including the leptin gene (LEP) and the PER2 gene. Leptin is a hormone secreted from adipose tissue and is involved in energy balance. It signals satiety, encourages lower food intake, and increases energy expenditure (Halaas et al. 1995). It has been shown that leptin can decrease activation of the ventral striatum, a region rich in dopaminergic innervation, in response to pictures of food in individuals suffering from leptin-deficiency (Farooqi et al. 2007). Preclinical research indicates that leptin influences the central dopamine system (Fulton et al. 2006). Therefore, Burghardt et al. (2012) assessed the association between leptin gene (LEP) polymorphisms and D2R binding. They genotyped four haplotype tagging SNPs (rs12706832, rs3828942, rs7791621, and rs7795794) located in intronic non-coding regions upstream/downstream to LEP in a sample of 50 healthy subjects. These subjects received a PET scan with [11C]raclopride. No association between any of
the SNPs and striatal D2R binding was observed, nor was there an association between circulating plasma leptin and D2R binding. The PER2 (Period 2) gene is a clock gene involved in regulation of the circadian rhythm. Shumay et al. (2012a) hypothesized that the PER2 gene might affect the regulation of D2R expression in the brain and subsequently might be linked to cocaine addiction (Volkow et al. 1990). Reasons for this hypothesis were that dopamine tone in the striatum is subject to circadian oscillation (Castaneda et al. 2004); that the PER2 gene regulates the reinforcing effects of cocaine in lab animals (Abarca et al. 2002); and that subjects suffering from substance-use disorders have unusual patterns of sleep and circadian rhythmicity (Kowatch et al. 1992; Morgan et al. 2008). Shumay et al. (2012a) genotyped 509 unrelated individuals for a newly identified VNTR in intron 3 of the PER2 gene. They found that the 4R/4R genotype was over-represented in the non-African American population and that genotype distributions were different between healthy controls and cocaine abusers depending on race. A subsample of 52 African American subjects received [11C]raclopride PET scans. The 4R and 3R homozygotes in this subsample had higher D2R binding than the 4R/3R heterozygotes in the putamen, caudate, and ventral stiratum. It is unclear, though, whether this subsample consists of healthy subjects, cocaine abusers, or both. As the authors mention themselves, they provide preliminary results that suggest that the PER2 genotype might affect D2R binding and could be a risk factor for cocaine addiction, but these results need replication. D O PA M I N E R E L E A S E Some PET and SPECT radioligands that bind to the dopamine D2 receptor can be displaced from the receptor by endogenous dopamine. This provides the opportunity to study changes in dopamine levels with PET and SPECT imaging. For example, an increase in synaptic dopamine levels after amphetamine administration will lead to a decrease in radioligand binding, which is proportional to the amount of dopamine released (Laruelle 2000). The decrease in radioligand binding most likely not only shows the displacement of the radioligand by dopamine, but probably also reflects other processes such as internalization of the D2R (Laruelle 2000; Ginovart 2005; Guo et al. 2010; Skinbjerg et al. 2010). However, the decrease in radioligand binding is often primarily interpreted as a measure of dopamine release, and this method has provided important insights in the alterations of dopamine
5 4 Part I : I maging G enetics and N eurochemistry
transmission in several disorders. Striatal dopamine release after an amphetamine challenge is increased in schizophrenia (Laruelle et al. 1996; Breier et al. 1997; Abi-Dargham et al. 1998), whereas it is lower than normal in substance abuse (Volkow et al. 1997; Martinez et al. 2005). This section will describe whether genetic polymorphisms affect dopamine release, as measured by radioligand displacement with SPECT and PET imaging. A substance that causes a very robust radioligand displacement is amphetamine. An important mechanism by which amphetamine induces dopamine release is by blocking and most likely reversing dopamine transport at the transporter, combined with vesicular release and increased phasic dopamine release (Sulzer 2011; Daberkow et al. 2013). The dopamine transporter (DAT) VNTR polymorphism has therefore been studied in relation to amphetamine-induced dopamine release (Martinez et al. 2001b). Thirty-one healthy controls and 29 patients with schizophrenia were genotyped for the DAT VNTR and scanned with [123I]IBZM and SPECT to determine [123I] IBZM displacement after amphetamine administration. However, there was no significant association between genotype and amphetamine-induced dopamine release, either in the overall sample, or in the patient or control group (patients: 10R/10R −14.5% displacement compared to −21% in 9R carriers; controls: 10R/10R −8.3% displacement compared to −6.6% in 9R carriers). Brody et al. (2006) studied dopamine release induced by nicotine in 45 tobacco-dependent smokers. The subjects were scanned with PET and [11C]raclopride, and 35 of them smoked a cigarette during a break in the scan. Furthermore, they were genotyped for the Taq1A polymorphism in the DRD2 gene, a VNTR in the DRD4 gene, a VNTR in the 3' untranslated region of the dopamine transporter gene, and the Val158Met polymorphism in the COMT gene. Some of these polymorphisms have previously been associated with smoking behavior (DRD2 Taq1A polymorphism: [Comings et al. 1996a]; DRD4 VNTR: [Shields et al., 1998]). For the subjects who smoked during scanning, those who were 9R allele carriers for the dopamine transporter VNTR (10R/10R −3.6% displacement compared to −14.8% in 9R carriers), who had fewer than 7 repeats for the DRD4 VNTR (≥ 7R carriers −2% displacement compared to −11.3% in < 7R carriers), and who were Val homozygotes for the COMT gene (Met carriers −4.5% displacement compared to −21.7% in Val/Val) had higher dopamine release after smoking than those with the alternate genotypes. There were no associations found with the DRD2 Taq1A polymorphism (A1 carriers −6.6% displacement compared to −9.6% in A2/A2). These results suggest
that dopamine system genotype polymorphisms explain part of the inter-individual variability in smoking-induced dopamine release, and they indicate that smoking-induced dopamine release has a genetic component. Pecina et al. (2013) studied SNPs in the DRD2 gene and the effect on stress-induced dopamine release to gain insight in the inter-individual variation in healthy subjects. They had shown that a haplotype-block composed of two SNPs (rs4274224 and rs4581480) influenced the hemodynamic responses of the dorsolateral prefrontal cortex (DLPFC) during reward expectation and the subgenual anterior cingulate cortices (sgACC) during implicit emotional processing. Furthermore, they found that rs4274224 affected the functional connectivity between the DLPFC and sgACC. Therefore, they chose to study the effect of rs4274224 on stress (pain)–induced dopamine release in a sample of 52 healthy subjects who were scanned with [11C] raclopride and PET. The subjects with genotype AG had lower dopamine release throughout the striatum after stress than subjects with AA or GG genotype. Thus, they conclude that DRD2 genotype has significant neurobiological effects in several functional domains, such as emotional, stress, and reward processing. In the same subject sample, the effect of the leptin gene (LEP) on stress-induced dopamine release was studied (Burghardt et al. 2012). As previously described in the section on dopamine D2 receptor availability, leptin can influence the central dopamine system. Leptin is also known for influencing the endocrine response to stress stimuli (Wilson et al. 2005). Burghardt et al. (2012) found that subjects with the GG genotype for the SNP rs12706832 had greater stress-induced dopamine release in the left nucleus accumbens, left putamen, and right ventral striatum (including part of the ventral putamen and caudate head). Dopamine release also positively correlated with circulating plasma leptin. These results suggest that leptin can influence stress-induced dopaminergic function, which could be important in pathological states such as obesity. Finally, the oxytocin gene (OXT) has been studied in relation to stress-induced dopamine release. Oxytocin is important in an array of reproductive and social activities expanding from pair bonding to maternal behavior to sexual behavior (Donaldson and Young 2008; Lee et al. 2009). It has also been shown to play a role in the integration and regulation of salient information, including painful stimuli (Scott et al. 2006; Gu and Yu 2007). Oxytocin also modulates central dopaminergic responses related to non-social behaviors and has anti-nociceptive properties (Kovacs et al. 1985; Pfister and Muir 1989). Love et al. (2012) focused on the oxytocin gene (OXT) and the dopaminergic system in
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
55
a subject sample of healthy controls (n = 55). They genotyped four haplotype tagging SNPs (rs4813625, rs877172, rs3761248, and rs2740210) located in regions close to the OXT gene. Stress-induced dopamine release was measured with [11C]raclopride PET. Assessments of pain, anxiety, well-being, interpersonal attachment, and affect were acquired. They found that female rs4813625 C allele carriers demonstrated higher stress-induced dopamine release in the rostral caudate (> 10-fold higher), whereas male subjects showed no significant differences between genotype groups. In addition, female rs4813625 C allele carriers had higher anxiety and lower emotional well-being scores, whereas higher stress-induced dopamine release was also associated with lower emotional well-being in these carriers. Thus, the variability within the oxytocin gene appears to explain inter-individual differences in dopaminergic responses to stress and to influence anxiety and emotional well-being, at least in females. D O PA M I N E T R A N S P O R T E R AVA I L A BI L I T Y The pre-synaptic dopamine transporter (DAT) is important for the regulation of the synaptic dopamine levels, in particular tonic dopamine. The DAT is abundantly expressed in the striatum and at lower density also in the midbrain (ventral tegmental area and substantia nigra) (Staley et al. 1995). DAT imaging with SPECT and PET is used diagnostically to detect in vivo degeneration of nigro-striatal cells in disorders like Parkinson’s disease and dementia with Lewy bodies (Booij et al. 1997; McKeith et al. 2007; Oh et al. 2012). DAT availability in ADHD has also been extensively studied and seems to relate to previous treatment with psychostimulants, which act on the DAT (Fusar-Poli et al. 2012). In the general population, DAT availability decreases with aging and tends to be higher in women compared to men (van Dyck et al. 1995; Lavalaye et al. 2000). There are several PET and SPECT ligands available to image DAT availability. The ligands used in this section are [123I]β-CIT, [123I]FP-CIT, [99mTc]TRODAT-1, and [123I]IPT for SPECT and [11C]cocaine and [11C]altropane for PET. The SPECT radioligands are all derived from cocaine and they are non-selective for DAT, although binding in the striatum predominantly reflects binding to DAT (Laruelle et al. 1993; Booij et al. 2007; Ziebell et al. 2010). For SPECT, [123I]β-CIT and [123I]FP-CIT are most used
because they have good specific to non-specific ratios and validated quantification methods. DOPAMINE TR ANSPORTER G E N E (SLC 6 A 3)
The DAT gene (SLC6A3) is located on chromosome 5p15.3 and includes a VNTR in the 3' untranslated region of the gene (Vandenbergh et al. 1992), which has been extensively studied. The VNTR consists of a 40-base pair repeat, and the most common alleles are the 9 repeat (9R) and 10 repeat (10R). This VNTR has repeatedly been implicated in bipolar disorder (Serretti and Mandelli 2008) and ADHD (Gizer et al. 2009; Sharp et al. 2009), although not consistently (Purper-Ouakil et al. 2005). The VNTR appears to affect DAT gene expression in vitro (Fuke et al. 2001; Michelhaugh et al. 2001; Mill et al. 2005). There are 11 studies that have investigated the in vivo association between the DAT VNTR and striatal DAT availability (for details, see Table 4.2). Five of them found that 9R allele carriers had higher DAT availability in striatal subregions compared to 10R homozygotes (Jacobsen et al. 2000; van Dyck et al. 2005; van de Giessen et al. 2009; Shumay et al. 2011; Spencer et al. 2013). However, four studies found no significant effect (Martinez et al. 2001b; Lynch et al. 2003; Krause et al. 2006; Lafuente et al. 2007) and two found the opposite effect (Heinz et al. 2000; Cheon et al. 2005). Differences in population, radioligand, and sample size could partially explain the inconsistent results. In this respect it is worth noting that the five studies that found higher DAT availability in 9R allele carriers had relatively large sample sizes and included predominantly healthy adult subjects. Recently, two meta-analyses have been performed (Costa et al. 2011; Faraone et al. 2014). Costa et al. (2011) included seven of the studies described in Table 4.2 (not included were Lynch et al. 2003; Cheon et al. 2005; Shumay et al. 2011; Spencer et al. 2013). They reported no significant effect of the DAT VNTR on striatal DAT availability, either in a meta-analysis of healthy subjects, or in an analysis including patient groups, although DAT availability was nominally higher in the 9R carriers (effect size g = 0.66). However, Faraone et al. (2014) included all PET and SPECT studies and found that the 9R carriers had significantly higher DAT availability than the 10R homozygotes. For the [123I]β-CIT SPECT studies this effect was only significant for the healthy subjects and not for patient populations. Including all SPECT studies in the meta-analysis introduced a lot of heterogeneity. The recent study of Shumay et al. (2011) showed that the effect (9R
5 6 Part I : I maging G enetics and N eurochemistry
TABLE 4.2
STUDIES ON STRIATAL DOPAMINE TRANSPORTER AVAILABILIT Y AND THE 3' VNTR IN THE DOPAMINE TRANSPORTER GENE Radioligand
N
Region
Note (percentage difference in BP)
Jacobsen et al. (2000)
[123I]β-CIT
44 (30 HC, 14 COC)
striatum
+13%
van Dyck et al. (2005)
[
I]β-CIT
96 HC
striatum
+9%
van de Giessen et al. (2009)
[123I]β-CIT
79 HC
striatum
+18%; Haplotype effect with 5' polymorphisms
Shumay et al. (2011)
[11C]cocaine
95 HC
caudate, ventral striatum
Effect significant in Caucasians, not in AA
Spencer et al. (2012)
[11C]altropane
68 (34 HC, 34 ADHD)
striatum
HC: +2.4%–7.5%
9R carriers > 10R/10R
123
ADHD: +2.2%–13.8% 9R carriers < 10R/10R Heinz et al. (2000)
[123I]β-CIT
25 (11 HC, 14 ALC)
putamen
−22%; Alcoholism not associated with DAT genotype and availability
Cheon et al. (2005)
[123I]IPT
11 ADHD children
basal ganglia
right: −72%, left: −68% Methylphenidate response better in 9R carriers
No effect Martinez et al. (2001)
[123I]β-CIT
43 (21 HC, 22 SCZ)
striatum
0%
Lynch et al. (2003)
[
Tc]TRODAT-1
166 (66 HC, 100 PD)
striatum
HC: 0%, PD: +1%
Krause et al. (2006)
[
Tc]TRODAT-1
22 ADHD
striatum
+2%
Lafuente et al. (2007)
[123I]FP-CIT
15 SCZ
striatum
+3.5%
99m
99m
BP = binding potential; 9R = 9-repeat allele; 10R = 10-repeat allele; COC = abstinent cocaine abusers; ADHD = attention deficit hyperactivity disorder; ALC = abstinent alcoholics; SCZ = schizophrenics; PD = Parkinson’s disease; AA = African Americans
carriers with DAT availability higher than 10R homozygotes) was only significant in Caucasians and not in African Americans, which suggests that ethnic background should be taken into account in future studies. In summary, the results of the available data and meta-analyses so far suggest that the 9R allele carriers of the DAT VNTR have higher DAT availability than 10R homozygotes, in particular in healthy subjects. Other polymorphisms in the SLC6A3 have also been studied in relation to striatal DAT availability. Drgon et al. (2006) had identified two SNPs (rs2652511 and rs2937639) in the 5' promotor region of the gene that accounts for most of the genetic variability in this region. Using [11C]cocaine and PET, they found in 15 subjects (of whom 6 were diagnosed with ADHD) that the allele combination C-G for these two SNPs was significantly more prevalent in subjects with high DAT availability in the ventral striatum than in subjects with low DAT availability, whereas the T-A combination was significantly more prevalent in subjects with low DAT availability. They also confirmed this finding in
51 postmortem brains. Van de Giessen et al. (2009) genotyped these two SNPs in a sample of 79 healthy subjects in addition to the 3' DAT VNTR (see Table 4.2). They found no direct effect of the SNPs on striatal DAT availability, but a haplotype analysis consisting of the SNPs and VNTR showed that the haplotype combination T-A-9R was significantly associated with higher striatal DAT expression (+31%) compared to the other haplotypes. The results of Drgon et al. (2006) and van de Giessen et al. (2009) are not consistent with each other, but suggest that polymorphisms in the promoter region of the gene need more attention in future studies to determine their exact role. Van de Giessen et al (2012) also report the results for four other SNPs that are linked to a newly identified splice variant of SLC6A3, exon 3B (rs420422 and rs462523 upstream of exon 3b; rs458609 and rs457702 downstream of exon 3b) in the same subject sample that they described previously (van de Giessen et al., 2009). These four SNPs did not affect DAT availability, though, nor did they modify the previously reported effect of the 3' VNTR.
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
57
Two of the previously described studies (Shumay et al. 2011; Spencer et al. 2013) genotyped both the 3' VNTR and a VNTR in intron 8 of SLC6A3. Shumay et al. (2011) reported that the 5R allele carriers for the intron 8 VNTR had higher DAT availability throughout the striatum relative to the 6R homozygotes. The intron 8 VNTR also showed an interaction with age. A haplotype analysis with the 3' VNTR (alleles 9R and 10R) showed that the allele combination 5R-9R and 5R-10R was associated with higher DAT availability in the caudate than the combinations 6R-9R and 6R-10R. However, Spencer et al. found no association between the intron 8 VNTR and striatal DAT availability, nor did they find an association with the 3' VNTR-intro 8 VNTR haplotype. Thus, the effect of the intron 8 VNTR in the DAT gene is unclear. OTHER GENES AFFEC TING DAT AVAIL ABILIT Y
Apart from polymorphisms in the DAT gene SLC6A3, polymorphisms in other genes might also affect DAT availability. There are three studies that investigated polymorphisms in the DRD2 gene. Laine et al. (2001) investigated the previously described Taq1A polymorphism and acquired [123I]β-CIT SPECT scans in 29 abstinent patients with alcohol dependence. The A1 allele carriers had +16% higher DAT availability than the A2 homozygotes. Given this finding, it is interesting that the A1 allele has also been associated with more depressive symptoms during alcohol withdrawal and that DAT availability is higher in patients with depression (Laasonen-Balk et al. 1999; Meyer et al. 2001). Laine et al. (2001) suggest that the underlying mechanism might be related to a gene variant that modifies DAT availability and that is in linkage disequilibrium with the Taq1A polymorphism, although this gene variant is still unknown. The two other studies focused on a functional SNP rs1076560 in the DRD2 gene, which we described in the section on dopamine D2 receptor availability. The same study that reported lower striatal D2R availability in T allele carriers (Bertolino et al. 2010) also found that these carriers had lower striatal DAT availability, measured with [123I]FP-CIT. This finding could be the result of regulation of DAT by pre-synaptic D2R auto-receptors, or it might be a compensatory mechanism to reduce transport of dopamine from the synapse, because of the lower D2R capacity, although this is speculative. In a subanalysis of this sample, it was found that the GG homozygotes had reduced connectivity in the medial prefrontal cortex in a component of the default mode network (measured with fMRI) and
this connectivity correlated positively with DAT availability (Sambataro et al. 2013). In subjects with GT genotype, the connectivity in the ventral striatum (with a striatal network including pallidum, dorsolateral prefrontal cortex, and superior parietal cortex) was negatively correlated with striatal DAT availability. The underlying mechanisms are still unclear, but these data suggest that the DRD2 SNP not only influences striatal DAT availability, but also affects brain connectivity in ventral striatal and prefrontal regions in healthy subjects. Because this SNP has only been studied in one sample, these data are still in need of replication. Finally, there is one study that investigated the GNβ3 gene (GNB3), which encodes for a subunit of the G-protein that is important for G-protein coupled receptors. Chen et al. (2011) hypothesized that pre-synaptic auto-receptors, which are G-protein coupled receptors, might regulate DAT expression and that the GNβ3 gene could therefore affect striatal DAT availability. Therefore, they genotyped the C825T polymorphism in this gene in 87 healthy subjects, who were scanned with [99mTc]TRODAT and SPECT. The subjects with CC genotype had 15% higher striatal DAT availability than the T allele carriers. This is an interesting finding, because it is the first study to show that genetic variation in a downstream signaling molecule affects the regulation of the striatal dopamine system. D O PA M I N E S Y N T H E S I S C A PAC I T Y Dopamine is synthesized in dopaminergic neurons, of which the majority is located in the midbrain projecting to the striatum. It is synthesized from tyrosine, which is converted to DOPA by tyrosine hydroxylase (TH), and DOPA is subsequently converted to dopamine by DOPA decarboxylase (DDC). The enzymes TH and DDC are therefore determinants of dopamine synthesis capacity. PET imaging with the radioligands 6-18F-fluoro-L-DOPA ([18F]FDOPA) and 6-18F-fluoro-meta-tyrosine ([18F]FMT), which are substrates for DDC and TH, respectively, can measure the synthesis capacity in vivo in the human brain. Therewith, it can provide a measure of the integrity of pre-synaptic dopamine neurons. [18F]FDOPA is a commonly used radiolabeled analog of the dopamine precursor L-DOPA, and it appears to correlate with dopamine synthesis in dopaminergic neurons (Tsukada et al. 1994; Cumming and Gjedde 1998). Research using these radioligands and PET has consistently shown that dopamine synthesis capacity is increased in schizophrenia (Howes et al. 2007; Kumakura et al. 2007) and is reduced in Parkinson’s disease (Brooks et al. 1990; Pavese et al. 2012).
5 8 Part I : I maging G enetics and N eurochemistry
Two studies examined whether genetic polymorphisms are associated with altered dopamine synthesis capacity. The first one (Laakso et al. 2005) focused on polymorphisms in or associated with the dopamine D2 receptor gene (DRD2): the previously described Taq1A, C957T, and –141C Ins/Del polymorphisms. Dopamine D2 auto-receptors regulate dopamine synthesis (Wolf and Roth 1990). By affecting the expression and effectiveness of the DRD2 autoreceptors in the brain, it is possible that these three polymorphisms influence dopamine synthesis regulation. Laakso et al. (2005) examined this in a sample that consisted of 33 healthy, non-smoking volunteers, who were scanned with PET and [18F]fluorodopa ([18F]FDOPA). The authors found that the A1 allele for Taq1A was associated with 18% higher [18F]FDOPA uptake in the putamen of healthy humans, indicating higher pre-synaptic dopamine levels in this brain region. The other two polymorphisms did not significantly affect [18F]FDOPA uptake. It is interesting to note that higher levels of [18F]FDOPA uptake have been observed in alcohol dependence (Tiihonen et al. 1998) (although striatal dopamine release is blunted [Martinez et al. 2005]) and that alcohol dependence has been associated with the A1 allele (Blum et al. 1991). Parkinson’s disease is a neurodegenerative illness, which includes symptoms such as rigidity, limb bradykinesia, and postural instability, which are linked to dopaminergic neuronal loss in the nigro-striatal tract. However, non-motor features, such as cognitive deficits, are also important disease characteristics (Martinez-Martin et al. 2007). Val158Met, a common functional polymorphism in the catechol-O-methyltransferase (COMT) gene, is linked with alterations in executive performance in fronto-striatal tasks (Tunbridge et al. 2004) thought to relate to changes in cortical dopamine levels, as COMT is the principal mode of dopamine clearance in the frontal cortex. Wu et al. (2012) investigated the in vivo variability in pre-synaptic dopamine synthesis capacity in patients with idiopathic Parkinson’s disease as a function of the COMT Val158Met polymorphism. The sample consisted of 20 patients with Parkinson’s disease, half of whom were homozygous for Val/Val and the other half for Met/Met. They used [18F]-DOPA and PET and applied a prolonged imaging protocol. There were no group differences in Ki values for the striatum. The authors report that Met/Met homozygotes had higher late Ki values in several cortical areas (superior frontal [+104%], anterior cingulate [+85%], orbitofrontal [+70%], and mid-frontal cortex [+132%]). However, it should be noted that Ki values for [18F]-DOPA cannot be reliably determined in cortical areas, which questions the significance of these findings.
M O N OA M I N E OX I DA S E A AC T I V I T Y Monoamine oxidase A (MAO-A) is an enzyme involved in the degradation pathway of dopamine. Apart from deaminating dopamine, it also deaminates serotonin, norepinephrine, and epinephrine and the dopamine catabolism product 3-methoxytyramine. It is related to monoamine oxidase B, which also degrades dopamine. The enzyme is located on the outer mitochondrial membrane and is abundantly expressed in the thalamus, but also present in cortical and subcortical regions (Fowler et al. 2005). MAO-A has been implicated in depression; for example, selective MAO-A inhibitors are used as anti-depressants (Culpepper 2012), and in aggression and antisocial behavior (Sjoberg et al. 2008; McDermott et al. 2009; Fergusson et al. 2012). It is possible to measure MAO-A activity in humans using PET and the selective MAO-A radioligands [11C]clorgyline (irreversible inhibitor) or [11C]harmine (reversible inhibitor) (Fowler et al. 2005). PET studies with these ligands have shown that MAO-A levels are elevated in the brains of patients with major depressive disorder (Meyer et al. 2006), but reduced in the brains of cigarette smokers (Fowler et al. 1996). MAO-A is encoded by the MAOA gene, which is located on the X chromosome (Xp11.23–11.4). There is a VNTR in the promoter region of the gene with two to five copies of a 30bp repeated sequence (Sabol et al. 1998). This polymorphism affects transcriptional levels, and its variants determine high or low levels of enzyme activity (Hotamisligil and Breakefield 1991). The frequency distribution of the variants differs between ethnic groups (Sabol et al. 1998). There are two studies that examined whether the MAOA high/low polymorphisms would affect MAO-A activity in the brain. Both studies are in men, because men are hemizygous for the X chromosome and the polymorphisms would thus be expected to show clear effects. Fowler et al. (2007) scanned 38 healthy men using PET and [11C] clorgyline, but found no differences in MAO-A brain activity between men with high and low MAOA variants (except for a trend of 12% lower MAO-A activity in the visual cortex in men with the low MAOA variant). Analysis of a subgroup of this sample (27 healthy men) also did not show an association between brain MAO-A activity and genotype, but in this subgroup MAO-A activity inversely correlated with the trait of aggression (Alia-Klein et al. 2008). These findings suggest that the MAOA VNTR does not affect brain MAO-A activity, although the studies may have been underpowered. Given the findings that both the MAOA VNTR (Manuck et al. 2000; Caspi et al. 2002) and brain
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
59
MAO-A activity (Alia-Klein et al. 2008) are associated with aggression, a relationship between these two may be expected. Recently, part of this relation might have been identified: methylation status of the MAO-A promoter region (which includes the VNTR) negatively correlated with brain MAO-A activity in 34 healthy men of the previously published sample (Shumay et al. 2012b). Also, methylation patterns were associated with MAO-A activity in this sample. It still needs to be demonstrated that MAOA promoter methylation is also associated with behavior. However, this finding shows the importance of epigenetic regulation on MAO-A expression in the brain. It is worth noting that the findings for the MAO-A VNTR were derived from the same subjects sample, so independent replication in another sample is needed.
DI S C U S S I O N The literature on dopaminergic imaging and genetics has grown rapidly over the last 15 years. In particular, DAT availability and the DAT 3' VNTR have been extensively studied. A recent meta-analysis showed that 9R allele carriers for this polymorphism have higher DAT availability than the 10R homozygotes, in particular in Caucasian healthy subjects. Several studies have also focused on D2R availability and DRD2 polymorphisms. The A1 allele of the Taq1A polymorphism seems to lead to increased D2R binding; however, this finding is not yet definitive. A general issue with the currently available studies is that they provide interesting leads for gene effects on brain function, but need replication. Many of them are underpowered, and studies regularly lead to conflicting results. Because of the lack of conclusive results, it is still too early to recommend standard genotyping of specific polymorphisms when imaging the DAT, D2R, or any other component of the dopaminergic system, either for research or in the clinical setting. A tool to overcome the low power of many studies could be the institution of imaging databases with genetic data, and the sharing of these databases among institutions. Such developments have started in the field of MR imaging (e.g., BRAINnet, http://www.brainnet.net/about/ brain-resource-international-database/) and might be an example for the field of dopamine imaging. Also, more meta-analyses will be valuable to determine the specific gene effects on the dopaminergic system. A recent development in the field is the interest in genes that are not directly related to the dopaminergic system but
may have indirect effects, for example genes for leptin and oxytocin (Burghardt et al. 2012; Love et al. 2012). If replicated and confirmed, these studies will allow new insights in how genes, and the proteins they encode for, might affect the dopaminergic brain system. Furthermore, genome-wide association studies (GWAS) in different dopamine-related psychiatric disorders have often shown that relevant genetic variants for these disorders are unrelated to the dopamine system, or are only indirectly related through signal transduction pathways (Psychiatric et al. 2009). These GWAS results could provide new candidate genes for future studies with dopaminergic imaging measures in these psychiatric patient groups. Another new development that the field might focus on in the future is epigenetics. A first study now showed that methylation status of the promoter region of the MAOA gene affects in vivo MAO-A activity in the brain (Shumay et al. 2012b), whereas a VNTR in the MAOA promoter region showed no significant effect (Fowler et al. 2007; Alia-Klein et al. 2008). So, the epigenetic modification of this gene seems more important than the genetic polymorphisms in this region. Recently, it was also found that dopamine D2-like antagonists can change chromatin structure in striatal neurons (Li et al. 2004), which could be part of the mechanisms of action of antipsychotics and may lead to differences in expression of receptors and other proteins. These types of modifications might explain differences in dopaminergic imaging measures between drug-naïve and drug-free patients with schizophrenia. The combination of genetic and epigenetic data could thus help to disentangle gene effects on dopaminergic imaging measures by taking the epigenetic modification into account. Finally, for a complete picture of the dopaminergic system, it is also important to study components of the system that have not been examined in relation to genetic polymorphisms. More specifically, there are several radioligands available to study the vesicular monoamine transporter type-2 (VMAT-2), which transports dopamine from the cytosol into synaptic vesicles (Kilbourn 1997; Lin et al. 2010). Furthermore, the availability of the radioligand [11C]-(+)-PHNO, which has high affinity for the dopamine D3 receptor (D3R), provides the opportunity to study D3R-related effects in D3R-rich brain regions such as the ventral striatum, globus pallidum, thalamus, and midbrain. The D3R has a recognized role in reward and motivation, and there is currently much interest in this receptor subtype for many neuropsychiatric disorders. For imaging of the dopamine D4 and D5 receptors, there are no radioligands available yet, which explains why associations with genetic polymorphisms have not been investigated yet.
6 0 Part I : I maging G enetics and N eurochemistry
In summary, the field of genetics and dopaminergic imaging is rapidly advancing. However, due to the low power of many studies and the lack of replications, it is still too early to draw definite conclusions about specific gene effects. The only evidence for a significant gene effects is the replicated association between the 3' VNTR in the DAT gene and striatal DAT availability. Increasing sample sizes by building (multi-center) databases, and widening the focus to epigenetics and genes that are indirectly related to the dopamine system, will contribute to successful progress in the field. REFERENCES Abarca C, Albrecht U, Spanagel R. (2002). Cocaine sensitization and reward are under the influence of circadian genes and rhythm. Proc Natl Acad Sci U S A. 99: 9026–9030. Abi-Dargham A, Gil R, Krystal J, et al. (1998). Increased striatal dopamine transmission in schizophrenia: confirmation in a second cohort. Am J Psychiatry. 155: 761–767. Abi-Dargham A, Mawlawi O, Lombardo I, Gil R, Martinez D, Huang Y, Hwang DR, Keilp J, Kochan L, Van HR, Gorman JM, Laruelle M. (2002). Prefrontal dopamine D1 receptors and working memory in schizophrenia. J Neurosci. 22: 3708–3719. Abi-Dargham A, Xu X, Thompson JL, Gil R, Kegeles LS, Urban N, Narendran R, Hwang DR, Laruelle M, Slifstein M. (2012). Increased prefrontal cortical D(1) receptors in drug naive patients with schizophrenia: a PET study with [(1)(1)C]NNC112. J Psychopharmacol. 26: 794–805. Alia-Klein N, Goldstein RZ, Kriplani A, Logan J, Tomasi D, Williams B, Telang F, Shumay E, Biegon A, Craig IW, Henn F, Wang GJ, Volkow ND, Fowler JS. (2008). Brain monoamine oxidase A activity predicts trait aggression. J Neurosci. 28: 5099–5104. Arinami T, Gao M, Hamaguchi H, Toru M. (1997). A functional polymorphism in the promoter region of the dopamine D2 receptor gene is associated with schizophrenia. Hum Mol Genet. 6: 577–582. Beaulieu JM, Gainetdinov RR. (2011). The physiology, signaling, and pharmacology of dopamine receptors. Pharmacol Rev. 63: 182–217. Bertolino A, Taurisano P, Pisciotta NM, Blasi G, Fazio L, Romano R, Gelao B, Lo BL, Lozupone M, Di GA, Caforio G, Sambataro F, Niccoli-Asabella A, Papp A, Ursini G, Sinibaldi L, Popolizio T, Sadee W, Rubini G. (2010). Genetically determined measures of striatal D2 signaling predict prefrontal activity during working memory performance. PLoS One. 5: e9348. Blum K, Braverman ER, Wood RC, Gill J, Li C, Chen TJ, Taub M, Montgomery AR, Sheridan PJ, Cull JG. (1996). Increased prevalence of the Taq I A1 allele of the dopamine receptor gene (DRD2) in obesity with comorbid substance use disorder: a preliminary report. Pharmacogenetics. 6: 297–305. Blum K, Noble EP, Sheridan PJ, Finley O, Montgomery A, Ritchie T, Ozkaragoz T, Fitch RJ, Sadlack F, Sheffield D. (1991). Association of the A1 allele of the D2 dopamine receptor gene with severe alcoholism. Alcohol. 8: 409–416. Blum K, Noble EP, Sheridan PJ, Montgomery A, Ritchie T, Jagadeeswaran P, Nogami H, Briggs AH, Cohn JB. (1990). Allelic association of human dopamine D2 receptor gene in alcoholism. JAMA. 263: 2055–2060. Blum K, Noble EP, Sheridan PJ, Montgomery A, Ritchie T, Ozkaragoz T, Fitch RJ, Wood R, Finley O, Sadlack F. (1993). Genetic
predisposition in alcoholism: association of the D2 dopamine receptor TaqI B1 RFLP with severe alcoholics. Alcohol. 10: 59–67. Blum K, Sheridan PJ, Wood RC, Braverman ER, Chen TJ, Comings DE. (1995). Dopamine D2 receptor gene variants: association and linkage studies in impulsive-addictive-compulsive behaviour. Pharmacogenetics. 5: 121–141. Booij J, de Jong J, de Bruin K, Knol R, de Win MM, van Eck-Smit BL. (2007). Quantification of striatal dopamine transporters with 123I-FP-CIT SPECT is influenced by the selective serotonin reuptake inhibitor paroxetine: a double-blind, placebo-controlled, crossover study in healthy control subjects. J Nuclear Med. 48: 359–366. Booij J, Tissingh G, Boer GJ, Speelman JD, Stoof JC, Janssen AG, Wolters EC, van Royen EA. (1997). [123I]FP-CIT SPECT shows a pronounced decline of striatal dopamine transporter labelling in early and advanced Parkinson’s disease. J Neurol Neurosurg Psychiat. 62: 133–140. Boot E, Booij J, Zinkstok JR, Baas F, Swillen A, Owen MJ, Murphy DG, Murphy KC, Linszen DH, Van Amelsvoort TA. (2011). COMT Val(158) met genotype and striatal D(2/3) receptor binding in adults with 22q11 deletion syndrome. Synapse. 65: 967–970. Breen G, Brown J, Maude S, Fox H, Collier D, Li T, Arranz M, Shaw D, StClair D. (1999). -141 C del/ins polymorphism of the dopamine receptor 2 gene is associated with schizophrenia in a British population. Am J Med Genet. 88: 407–410. Breier A, Su TP, Saunders R, Carson RE, Kolachana BS, de Bartolomeis A, Weinberger DR, Weisenfeld N, Malhotra AK, Eckelman WC, Pickar D. (1997). Schizophrenia is associated with elevated amphetamine-induced synaptic dopamine concentrations: evidence from a novel positron emission tomography method. Proc Natl Acad Sci U S A. 94: 2569–2574. Brody AL, Mandelkern MA, Olmstead RE, Scheibal D, Hahn E, Shiraga S, Zamora-Paja E, Farahi J, Saxena S, London ED, McCracken JT. (2006). Gene variants of brain dopamine pathways and smoking-induced dopamine release in the ventral caudate/nucleus accumbens. Arch Gen Psychiatry. 63: 808–816. Brooks DJ, Ibanez V, Sawle GV, Quinn N, Lees AJ, Mathias CJ, Bannister R, Marsden CD, Frackowiak RS. (1990). Differing patterns of striatal 18F-dopa uptake in Parkinson’s disease, multiple system atrophy, and progressive supranuclear palsy. Ann Neurology. 28: 547–555. Burghardt PR, Love TM, Stohler CS, Hodgkinson C, Shen PH, Enoch MA, Goldman D, Zubieta JK. (2012). Leptin regulates dopamine responses to sustained stress in humans. J Neurosci. 32: 15369– 15376. Caspi A, McClay J, Moffitt TE, Mill J, Martin J, Craig IW, Taylor A, Poulton R. (2002). Role of genotype in the cycle of violence in maltreated children. Science. 297: 851–854. Castaneda TR, de Prado BM, Prieto D, Mora F. (2004). Circadian rhythms of dopamine, glutamate and GABA in the striatum and nucleus accumbens of the awake rat: modulation by light. J Pineal Res. 36: 177–185. Chen CK, Hu X, Lin SK, Sham PC, Loh e, Li T, Murray RM, Ball DM. (2004a). Association analysis of dopamine D2-like receptor genes and methamphetamine abuse. Psychiatr Genet. 14: 223–226. Chen J, Lipska BK, Halim N, Ma QD, Matsumoto M, Melhem S, Kolachana BS, Hyde TM, Herman MM, Apud J, Egan MF, Kleinman JE, Weinberger DR. (2004b). Functional analysis of genetic variation in catechol-O-methyltransferase (COMT): effects on mRNA, protein, and enzyme activity in postmortem human brain. Am J Hum Genet. 75: 807–821. Chen PS, Yeh TL, Lee IH, Lin CB, Tsai HC, Chen KC, Chiu NT, Yao WJ, Yang YK, Chou YH. (2011). Effects of C825T polymorphism of the GNB3 gene on availability of dopamine transporter in healthy volunteers—a SPECT study. NeuroImage. 56: 1526–1530. Cheon KA, Ryu YH, Kim JW, Cho DY. (2005). The homozygosity for 10-repeat allele at dopamine transporter gene and dopamine transporter density in Korean children with attention deficit
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
61
hyperactivity disorder: relating to treatment response to methylphenidate. Eur Neuropsychopharmacol. 15: 95–101. Comings DE. (1999). Molecular heterosis as the explanation for the controversy about the effect of the DRD2 gene on dopamine D2 receptor density. Mol Psychiatry. 4: 213–215. Comings DE, Ferry L, Bradshaw-Robinson S, Burchette R, Chiu C, Muhleman D. (1996a). The dopamine D2 receptor (DRD2) gene: a genetic risk factor in smoking. Pharmacogenetics. 6: 73–79. Comings DE, Rosenthal RJ, Lesieur HR, Rugle LJ, Muhleman D, Chiu C, Dietz G, Gade R. (1996b). A study of the dopamine D2 receptor gene in pathological gambling. Pharmacogenetics. 6: 223–234. Cordeiro Q, Siqueira-Roberto J, Zung S, Vallada H. (2009). Association between the DRD2-141C Insertion/Deletion polymorphism and schizophrenia. Arq Neuropsiquiatr. 67: 191–194. Cortes R, Gueye B, Pazos A, Probst A, Palacios JM. (1989). Dopamine receptors in human brain: autoradiographic distribution of D1 sites. Neuroscience. 28: 263–273. Costa A, Riedel M, Muller U, Moller HJ, Ettinger U. (2011). Relationship between SLC6A3 genotype and striatal dopamine transporter availability: a meta-analysis of human single photon emission computed tomography studies. Synapse. 65: 998–1005. Culpepper L. (2012). The use of monoamine oxidase inhibitors in primary care. J Clin Psychiatry. 73(Suppl 1):37–41. Cumming P, Gjedde A. (1998). Compartmental analysis of dopa decarboxylation in living brain from dynamic positron emission tomograms. Synapse. 29: 37–61. Daberkow DP, Brown HD, Bunner KD, Kraniotis SA, Doellman MA, Ragozzino ME, Garris PA, Roitman MF. (2013). Amphetamine paradoxically augments exocytotic dopamine release and phasic dopamine signals. J Neurosci. 33: 452–463. Davis C, Levitan RD, Yilmaz Z, Kaplan AS, Carter JC, Kennedy JL. (2012). Binge eating disorder and the dopamine D2 receptor: genotypes and sub-phenotypes. Prog Neuropsychopharmacol Biol Psychiatry. 38: 328–335. de Weijer BA, van de Giessen E, van Amelsvoort TA, Boot E, Braak B, Janssen IM, van de Laar A, Fliers E, Serlie MJ, Booij J. (2011). Lower striatal dopamine D2/3 receptor availability in obese compared with non-obese subjects. EJNMMI Res. 1: 37. Donaldson ZR, Young LJ. (2008). Oxytocin, vasopressin, and the neurogenetics of sociality. Science. 322: 900–904. Drgon T, Lin Z, Wang GJ, Fowler J, Pablo J, Mash DC, Volkow N, Uhl GR. (2006). Common human 5' dopamine transporter (SLC6A3) haplotypes yield varying expression levels in vivo. Cell Mol Neurobiol. 26: 875–889. Duan J, Wainwright MS, Comeron JM, Saitou N, Sanders AR, Gelernter J, Gejman PV. (2003). Synonymous mutations in the human dopamine receptor D2 (DRD2) affect mRNA stability and synthesis of the receptor. Hum Mol Genet. 12: 205–216. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, Goldman D, Weinberger DR. (2001). Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci U S A. 98: 6917–6922. Faraone SV, Perlis RH, Doyle AE, Smoller JW, Goralnick JJ, Holmgren MA, Sklar P. (2005). Molecular genetics of attention-deficit/hyperactivity disorder. Biol Psychiatry. 57: 1313–1323. Faraone SV, Spencer TJ, Madras BK, Zhang-James Y, Biederman J. (2014). Functional effects of dopamine transporter gene genotypes on in vivo dopamine transporter functioning: a meta-analysis. Mol Psychiatry. 19: 880–889. Farde L, Hall H, Pauli S, Halldin C. (1995). Variability in D2-dopamine receptor density and affinity: a PET study with [11C]raclopride in man. Synapse. 20: 200–208. Farooqi IS, Bullmore E, Keogh J, Gillard J, O’Rahilly S, Fletcher PC. (2007). Leptin regulates striatal regions and human eating behavior. Science. 317: 1355. Fedi M, Berkovic SF, Scheffer IE, O’Keefe G, Marini C, Mulligan R, Gong S, Tochon-Danguy H, Reutens DC. (2008). Reduced striatal
D1 receptor binding in autosomal dominant nocturnal frontal lobe epilepsy. Neurology. 71: 795–798. Fergusson DM, Boden JM, Horwood LJ, Miller A, Kennedy MA. (2012). Moderating role of the MAOA genotype in antisocial behaviour. Br J Psychiatry. 200: 116–123. Fowler JS, Alia-Klein N, Kriplani A, Logan J, Williams B, Zhu W, Craig IW, Telang F, Goldstein R, Volkow ND, Vaska P, Wang GJ. (2007). Evidence that brain MAO A activity does not correspond to MAO A genotype in healthy male subjects. Biol Psychiatry. 62: 355–358. Fowler JS, Logan J, Volkow ND, Wang GJ. (2005). Translational neuroimaging: positron emission tomography studies of monoamine oxidase. Mol Imaging Biol. 7: 377–387. Fowler JS, Volkow ND, Wang GJ, Pappas N, Logan J, Shea C, Alexoff D, MacGregor RR, Schlyer DJ, Zezulkova I, Wolf AP. (1996). Brain monoamine oxidase A inhibition in cigarette smokers. Proc Natl Acad Sci U S A. 93: 14065–14069. Fuke S, Suo S, Takahashi N, Koike H, Sasagawa N, Ishiura S. (2001). The VNTR polymorphism of the human dopamine transporter (DAT1) gene affects gene expression. Pharmacogenomics J. 1: 152–156. Fulton S, Pissios P, Manchon RP, Stiles L, Frank L, Pothos EN, Maratos-Flier E, Flier JS. (2006). Leptin regulation of the mesoaccumbens dopamine pathway. Neuron. 51: 811–822. Fusar-Poli P, Rubia K, Rossi G, Sartori G, Balottin U. (2012). Striatal dopamine transporter alterations in ADHD: pathophysiology or adaptation to psychostimulants? A meta-analysis. Am J Psychiatry. 169: 264–272. Ginovart N. (2005). Imaging the dopamine system with in vivo [11C] raclopride displacement studies: understanding the true mechanism. Mol Imaging Biol.7: 45–52. Gizer IR, Ficks C, Waldman ID. (2009). Candidate gene studies of ADHD: a meta-analytic review. Human Genetics. 126: 51–90. Goldberg TE, Egan MF, Gscheidle T, Coppola R, Weickert T, Kolachana BS, Goldman D, Weinberger DR. (2003). Executive subprocesses in working memory: relationship to catechol-O-methyltransferase Val158Met genotype and schizophrenia. Arch Gen Psychiatry. 60: 889–896. Grandy DK, Litt M, Allen L, Bunzow JR, Marchionni M, Makam H, Reed L, Magenis RE, Civelli O. (1989). The human dopamine D2 receptor gene is located on chromosome 11 at q22-q23 and identifies a TaqI RFLP. Am J Hum Genet. 45: 778–785. Gu XL, Yu LC. (2007). Involvement of opioid receptors in oxytocin-induced antinociception in the nucleus accumbens of rats. J Pain. 8: 85–90. Guo N, Guo W, Kralikova M, Jiang M, Schieren I, Narendran R, Slifstein M, Abi-Dargham A, Laruelle M, Javitch JA, Rayport S. (2010). Impact of D2 receptor internalization on binding affinity of neuroimaging radiotracers. Neuropsychopharmacology. 35: 806–817. Halaas JL, Gajiwala KS, Maffei M, Cohen SL, Chait BT, Rabinowitz D, Lallone RL, Burley SK, Friedman JM. (1995). Weight-reducing effects of the plasma protein encoded by the obese gene. Science. 269: 543–546. Hauge XY, Grandy DK, Eubanks JH, Evans GA, Civelli O, Litt M. (1991). Detection and characterization of additional DNA polymorphisms in the dopamine D2 receptor gene. Genomics. 10: 527– 530. Heinz A, Goldman D, Jones DW, Palmour R, Hommer D, Gorey JG, Lee KS, Linnoila M, Weinberger DR. (2000). Genotype influences in vivo dopamine transporter availability in human striatum. Neuropsychopharmacology. 22: 133–139. Hirvonen M, Laakso A, Nagren K, Rinne JO, Pohjalainen T, Hietala J. (2004). C957T polymorphism of the dopamine D2 receptor (DRD2) gene affects striatal DRD2 availability in vivo. Mol Psychiatry. 9: 1060–1061. Hirvonen MM, Laakso A, Nagren K, Rinne JO, Pohjalainen T, Hietala J. (2009a). C957T polymorphism of dopamine D2 receptor gene affects striatal DRD2 in vivo availability by changing the receptor affinity. Synapse. 63: 907–912.
6 2 Part I : I maging G enetics and N eurochemistry
Hirvonen MM, Lumme V, Hirvonen J, Pesonen U, Nagren K, Vahlberg T, Scheinin H, Hietala J. (2009b). C957T polymorphism of the human dopamine D2 receptor gene predicts extrastriatal dopamine receptor availability in vivo. Prog Neuropsychopharmacol Biol Psychiatry. 33: 630–636. Hirvonen MM, Nagren K, Rinne JO, Pesonen U, Vahlberg T, Hagelberg N, Hietala J. (2010). COMT Val158Met genotype does not alter cortical or striatal dopamine D2 receptor availability in vivo. Mol Imaging Biol. 12: 192–197. Hoenicka J, Aragues M, Rodriguez-Jimenez R, Ponce G, Martinez I, Rubio G, Jimenez-Arriero MA, Palomo T. (2006). C957T DRD2 polymorphism is associated with schizophrenia in Spanish patients. Acta Psychiatr Scand. 114: 435–438. Hotamisligil GS, Breakefield XO. (1991). Human monoamine oxidase A gene determines levels of enzyme activity. Am J Hum Genet. 49: 383–392. Howes OD, Montgomery AJ, Asselin MC, Murray RM, Grasby PM, McGuire PK. (2007). Molecular imaging studies of the striatal dopaminergic system in psychosis and predictions for the prodromal phase of psychosis. Br J Psychiatry. Suppl 51: s13–18. Jacobsen LK, Staley JK, Zoghbi SS, Seibyl JP, Kosten TR, Innis RB, Gelernter J. (2000). Prediction of dopamine transporter binding availability by genotype: a preliminary report. Am J Psychiatry. 157: 1700–1703. Jonsson EG, Nothen MM, Grunhage F, Farde L, Nakashima Y, Propping P, Sedvall GC. (1999a). Polymorphisms in the dopamine D2 receptor gene and their relationships to striatal dopamine receptor density of healthy volunteers. Mol Psychiatry. 4: 290–296. Jonsson EG, Nothen MM, Neidt H, Forslund K, Rylander G, MattilaEvenden M, Asberg M, Propping P, Sedvall GC. (1999b). Association between a promoter polymorphism in the dopamine D2 receptor gene and schizophrenia. Schizophr Res. 40: 31–36. Karlsson P, Farde L, Halldin C, Sedvall G. (2002). PET study of D(1) dopamine receptor binding in neuroleptic-naive patients with schizophrenia. Am J Psychiatry. 159: 761–767. Karlsson S, Nyberg L, Karlsson P, Fischer H, Thilers P, Macdonald S, Brehmer Y, Rieckmann A, Halldin C, Farde L, Backman L. (2009). Modulation of striatal dopamine D1 binding by cognitive processing. Neuroimage. 48: 398–404. Karlsson S, Rieckmann A, Karlsson P, Farde L, Nyberg L, Backman L. (2011). Relationship of dopamine D1 receptor binding in striatal and extrastriatal regions to cognitive functioning in healthy humans. Neuroimage. 57: 346–351. Kilbourn MR. (1997). In vivo radiotracers for vesicular neurotransmitter transporters. Nuclear Med Biol. 24: 615–619. Koob GF. (1996). Hedonic valence, dopamine and motivation. Mol Psychiatry. 1: 186–189. Kovacs GL, Horvath Z, Sarnyai Z, Faludi M, Telegdy G. (1985). Oxytocin and a C-terminal derivative (Z-prolyl-D-leucine) attenuate tolerance to and dependence on morphine and interact with dopaminergic neurotransmission in the mouse brain. Neuropharmacology. 24: 413–419. Kowatch RA, Schnoll SS, Knisely JS, Green D, Elswick RK. (1992). Electroencephalographic sleep and mood during cocaine withdrawal. J Addict Dis. 11: 21–45. Krause J, Dresel SH, Krause KH, La Fougere C, Zill P, Ackenheil M. (2006). Striatal dopamine transporter availability and DAT-1 gene in adults with ADHD: no higher DAT availability in patients with homozygosity for the 10-repeat allele. World J Biol Psychiatry. 7: 152–157. Kumakura Y, Cumming P, Vernaleken I, Buchholz HG, Siessmeier T, Heinz A, Kienast T, Bartenstein P, Grunder G. (2007). Elevated [18F]fluorodopamine turnover in brain of patients with schizophrenia: an [18F]fluorodopa/positron emission tomography study. J Neurosci. 27: 8080–8087. Laakso A, Pohjalainen T, Bergman J, Kajander J, Haaparanta M, Solin O, Syvalahti E, Hietala J. (2005). The A1 allele of the human D2
dopamine receptor gene is associated with increased activity of striatal L-amino acid decarboxylase in healthy subjects. Pharmacogenet Genomics. 15: 387–391. Laasonen-Balk T, Kuikka J, Viinamaki H, Husso-Saastamoinen M, Lehtonen J, Tiihonen J. (1999). Striatal dopamine transporter density in major depression. Psychopharmacology (Berl). 144: 282–285. Lachman HM, Papolos DF, Saito T, Yu YM, Szumlanski CL, Weinshilboum RM. (1996). Human catechol-O-methyltransferase pharmacogenetics: description of a functional polymorphism and its potential application to neuropsychiatric disorders. Pharmacogenetics. 6: 243–250. Lafuente A, Bernardo M, Mas S, Crescenti A, Aparici M, Gasso P, Catalan R, Mateos JJ, Lomena F, Parellada E. (2007). Dopamine transporter (DAT) genotype (VNTR) and phenotype in extrapyramidal symptoms induced by antipsychotics. Schizophrenia Res. 90: 115–122. Lafuente A, Bernardo M, Mas S, Crescenti A, Aparici M, Gasso P, Goti J, Sanchez V, Catalan R, Carne X. (2008). -141C Ins/Del polymorphism of the dopamine D2 receptor gene is associated with schizophrenia in a Spanish population. Psychiatr Genet. 18:122–127. Laine TP, Ahonen A, Rasanen P, Pohjalainen T, Tiihonen J, Hietala J. (2001). The A1 allele of the D2 dopamine receptor gene is associated with high dopamine transporter density in detoxified alcoholics. Alcohol Alcoholism. 36: 262–265. Laruelle M. (2000). Imaging synaptic neurotransmission with in vivo binding competition techniques: a critical review. J Cereb Blood Flow Metab. 20: 423–451. Laruelle M, Abi-Dargham A, van Dyck CH, Gil R, D’Souza CD, Erdos J, McCance E, Rosenblatt W, Fingado C, Zoghbi SS, Baldwin RM, Seibyl JP, Krystal JH, Charney DS, Innis RB. (1996). Single photon emission computerized tomography imaging of amphetamineinduced dopamine release in drug-free schizophrenic subjects. Proc Natl Acad Sci U S A. 93: 9235–9240. Laruelle M, Baldwin RM, Malison RT, Zea-Ponce Y, Zoghbi SS, alTikriti MS, Sybirska EH, Zimmermann RC, Wisniewski G, Neumeyer JL, et al. (1993). SPECT imaging of dopamine and serotonin transporters with [123I]beta-CIT: pharmacological characterization of brain uptake in nonhuman primates. Synapse. 13: 295–309. Laruelle M, Gelernter J, Innis RB. (1998). D2 receptors binding potential is not affected by Taq1 polymorphism at the D2 receptor gene. Mol Psychiatry. 3: 261–265. Lavalaye J, Booij J, Reneman L, Habraken JB, van Royen EA. (2000). Effect of age and gender on dopamine transporter imaging with [123I]FP-CIT SPET in healthy volunteers. Eur J Nuclear Med. 27: 867–869. Lawford BR, Young RM, Noble EP, Sargent J, Rowell J, Shadforth S, Zhang X, Ritchie T. (2000). The D(2) dopamine receptor A(1) allele and opioid dependence: association with heroin use and response to methadone treatment. Am J Med Genet. 96: 592–598. Lee HJ, Macbeth AH, Pagani JH, Young WS, 3rd. (2009). Oxytocin: the great facilitator of life. Progress Neurobiol. 88: 127–151. Li J, Guo Y, Schroeder FA, Youngs RM, Schmidt TW, Ferris C, Konradi C, Akbarian S. (2004). Dopamine D2-like antagonists induce chromatin remodeling in striatal neurons through cyclic AMPprotein kinase A and NMDA receptor signaling. J Neurochem. 90: 1117–1131. Lin KJ, Weng YH, Wey SP, Hsiao IT, Lu CS, Skovronsky D, Chang HP, Kung MP, Yen TC. (2010). Whole-body biodistribution and radiation dosimetry of 18F-FP-(+)-DTBZ (18F-AV-133): a novel vesicular monoamine transporter 2 imaging agent. J Nuclear Med. 51: 1480–1485. Love TM, Enoch MA, Hodgkinson CA, Pecina M, Mickey B, Koeppe RA, Stohler CS, Goldman D, Zubieta JK. (2012). Oxytocin gene polymorphisms influence human dopaminergic function in a sexdependent manner. Biol Psychiatry. 72: 198–206. Lynch DR, Mozley PD, Sokol S, Maas NM, Balcer LJ, Siderowf AD. (2003). Lack of effect of polymorphisms in dopamine metabolism
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
63
related genes on imaging of TRODAT-1 in striatum of asymptomatic volunteers and patients with Parkinson’s disease. Movement Disord. 18: 804–812. Manuck SB, Flory JD, Ferrell RE, Mann JJ, Muldoon MF. (2000). A regulatory polymorphism of the monoamine oxidase-A gene may be associated with variability in aggression, impulsivity, and central nervous system serotonergic responsivity. Psychiatry Res. 95: 9–23. Martinez D, Broft A, Laruelle M. (2001a). Imaging neurochemical endophenotypes: promises and pitfalls. Pharmacogenomics. 2: 223–237. Martinez D, Gelernter J, Abi-Dargham A, van Dyck CH, Kegeles L, Innis RB, Laruelle M. (2001b). The variable number of tandem repeats polymorphism of the dopamine transporter gene is not associated with significant change in dopamine transporter phenotype in humans. Neuropsychopharmacology. 24: 553–560. Martinez D, Gil R, Slifstein M, Hwang DR, Huang Y, Perez A, Kegeles L, Talbot P, Evans S, Krystal J, Laruelle M, Abi-Dargham A. (2005). Alcohol dependence is associated with blunted dopamine transmission in the ventral striatum. Biol Psychiatry. 58:779–786. Martinez-Martin P, Schapira AH, Stocchi F, Sethi K, Odin P, MacPhee G, Brown RG, Naidu Y, Clayton L, Abe K, Tsuboi Y, MacMahon D, Barone P, Rabey M, Bonuccelli U, Forbes A, Breen K, Tluk S, Olanow CW, Thomas S, Rye D, Hand A, Williams AJ, Ondo W, Chaudhuri KR. (2007). Prevalence of nonmotor symptoms in Parkinson’s disease in an international setting: study using nonmotor symptoms questionnaire in 545 patients. Movement Disord. 22: 1623–1629. McDermott R, Tingley D, Cowden J, Frazzetto G, Johnson DD. (2009). Monoamine oxidase A gene (MAOA) predicts behavioral aggression following provocation. Proc Natl Acad Sci U S A. 106: 2118–2123. McKeith I, O’Brien J, Walker Z, Tatsch K, Booij J, Darcourt J, Padovani A, Giubbini R, Bonuccelli U, Volterrani D, Holmes C, Kemp P, Tabet N, Meyer I, Reininger C, Group DLBS. (2007). Sensitivity and specificity of dopamine transporter imaging with 123I-FPCIT SPECT in dementia with Lewy bodies: a phase III, multicentre study. Lancet Neurology. 6: 305–313. Messas G, Meira-Lima I, Turchi M, Franco O, Guindalini C, Castelo A, Laranjeira R, Vallada H. (2005). Association study of dopamine D2 and D3 receptor gene polymorphisms with cocaine dependence. Psychiatr Genet. 15: 171–174. Meyer JH, Ginovart N, Boovariwala A, Sagrati S, Hussey D, Garcia A, Young T, Praschak-Rieder N, Wilson AA, Houle S. (2006). Elevated monoamine oxidase a levels in the brain: an explanation for the monoamine imbalance of major depression. Arch Gen Psychiatry. 63: 1209–1216. Meyer JH, Kruger S, Wilson AA, Christensen BK, Goulding VS, Schaffer A, Minifie C, Houle S, Hussey D, Kennedy SH. (2001). Lower dopamine transporter binding potential in striatum during depression. Neuroreport. 12: 4121–4125. Michelhaugh SK, Fiskerstrand C, Lovejoy E, Bannon MJ, Quinn JP. (2001). The dopamine transporter gene (SLC6A3) variable number of tandem repeats domain enhances transcription in dopamine neurons. J Neurochem. 79: 1033–1038. Mill J, Asherson P, Craig I, D’Souza UM. (2005). Transient expression analysis of allelic variants of a VNTR in the dopamine transporter gene (DAT1). BMC Genetics. 6: 3. Monakhov M, Golimbet V, Abramova L, Kaleda V, Karpov V. (2008). Association study of three polymorphisms in the dopamine D2 receptor gene and schizophrenia in the Russian population. Schizophr Res. 100: 302–307. Morgan PT, Pace-Schott EF, Sahul ZH, Coric V, Stickgold R, Malison RT. (2008). Sleep architecture, cocaine and visual learning. Addiction. 103: 1344–1352. Neville MJ, Johnstone EC, Walton RT. (2004). Identification and characterization of ANKK1: a novel kinase gene closely linked to DRD2 on chromosome band 11q23.1. Hum Mutat. 23: 540–545.
Noble EP, Blum K, Khalsa ME, Ritchie T, Montgomery A, Wood RC, Fitch RJ, Ozkaragoz T, Sheridan PJ, Anglin MD. (1993). Allelic association of the D2 dopamine receptor gene with cocaine dependence. Drug Alcohol Depend. 33: 271–285. Noble EP, Blum K, Ritchie T, Montgomery A, Sheridan PJ. (1991). Allelic association of the D2 dopamine receptor gene with receptor-binding characteristics in alcoholism. Arch Gen Psychiatry. 48: 648–654. Oh M, Kim JS, Kim JY, Shin KH, Park SH, Kim HO, Moon DH, Oh SJ, Chung SJ, Lee CS. (2012). Subregional patterns of preferential striatal dopamine transporter loss differ in Parkinson disease, progressive supranuclear palsy, and multiple-system atrophy. J Nuclear Med. 53: 399–406. Ohara K, Nagai M, Tani K, Nakamura Y, Ino A, Ohara K. (1998). Functional polymorphism of -141C Ins/Del in the dopamine D2 receptor gene promoter and schizophrenia. Psychiatry Res. 81: 117–123. Okubo Y, Suhara T, Suzuki K, Kobayashi K, Inoue O, Terasaki O, Someya Y, Sassa T, Sudo Y, Matsushima E, Iyo M, Tateno Y, Toru M. (1997). Decreased prefrontal dopamine D1 receptors in schizophrenia revealed by PET. Nature. 385: 634–636. Parsons MJ, Mata I, Beperet M, Iribarren-Iriso F, Arroyo B, Sainz R, Arranz MJ, Kerwin R. (2007). A dopamine D2 receptor generelated polymorphism is associated with schizophrenia in a Spanish population isolate. Psychiatr Genet. 17: 159–163. Pavese N, Simpson BS, Metta V, Ramlackhansingh A, Chaudhuri KR, Brooks DJ. (2012). [(1)(8)F]FDOPA uptake in the raphe nuclei complex reflects serotonin transporter availability: a combined [(1) (8)F]FDOPA and [(1)(1)C]DASB PET study in Parkinson’s disease. NeuroImage. 59: 1080–1084. Pecina M, Mickey BJ, Love T, Wang H, Langenecker SA, Hodgkinson C, Shen PH, Villafuerte S, Hsu D, Weisenbach SL, Stohler CS, Goldman D, Zubieta JK. (2013). DRD2 polymorphisms modulate reward and emotion processing, dopamine neurotransmission and openness to experience. Cortex. 49: 877–890. Persico AM, Bird G, Gabbay FH, Uhl GR. (1996). D2 dopamine receptor gene TaqI A1 and B1 restriction fragment length polymorphisms: enhanced frequencies in psychostimulant-preferring polysubstance abusers. Biol Psychiatry. 40: 776–784. Pfister HP, Muir JL. (1989). Influence of exogenously administered oxytocin on central noradrenaline, dopamine and serotonin levels following psychological stress in nulliparous female rats (Rattus norvegicus). Int J Neurosci. 45: 221–229. Pohjalainen T, Nagren K, Syvalahti EK, Hietala J. (1999). The dopamine D2 receptor 5’-flanking variant, -141C Ins/Del, is not associated with reduced dopamine D2 receptor density in vivo. Pharmacogenetics. 9: 505–509. Pohjalainen T, Rinne JO, Nagren K, Lehikoinen P, Anttila K, Syvalahti EK, Hietala J. (1998a). The A1 allele of the human D2 dopamine receptor gene predicts low D2 receptor availability in healthy volunteers. Mol Psychiatry. 3: 256–260. Pohjalainen T, Rinne JO, Nagren K, Syvalahti E, Hietala J. (1998b). Sex differences in the striatal dopamine D2 receptor binding characteristics in vivo. Am J Psychiatry. 155: 768–773. Psychiatric GCCC, Cichon S, Craddock N, Daly M, Faraone SV, Gejman PV, Kelsoe J, Lehner T, Levinson DF, Moran A, Sklar P, Sullivan PF. (2009). Genomewide association studies: history, rationale, and prospects for psychiatric disorders. Am J Psychiatry. 166: 540–556. Purper-Ouakil D, Wohl M, Mouren MC, Verpillat P, Ades J, Gorwood P. (2005). Meta-analysis of family-based association studies between the dopamine transporter gene and attention deficit hyperactivity disorder. Psychiat Genet. 15: 53–59. Rinne JO, Hietala J, Ruotsalainen U, Sako E, Laihinen A, Nagren K, Lehikoinen P, Oikonen V, Syvalahti E. (1993). Decrease in human striatal dopamine D2 receptor density with age: a PET study with [11C]raclopride. J Cereb Blood Flow Metab. 13: 310–314.
6 4 Part I : I maging G enetics and N eurochemistry
Ritchie T, Noble EP. (2003). Association of seven polymorphisms of the D2 dopamine receptor gene with brain receptor-binding characteristics. Neurochem Res. 28: 73–82. Sabol SZ, Hu S, Hamer D. (1998). A functional polymorphism in the monoamine oxidase A gene promoter. Hum Genet. 103: 273–279. Sambataro F, Fazio L, Taurisano P, Gelao B, Porcelli A, Mancini M, Sinibaldi L, Ursini G, Masellis R, Caforio G, Di Giorgio A, Niccoli-Asabella A, Popolizio T, Blasi G, Bertolino A. (2013). DRD2 genotype-based variation of default mode network activity and of its relationship with striatal DAT binding. Schizophrenia Bull. 39: 206–216. Scott DJ, Heitzeg MM, Koeppe RA, Stohler CS, Zubieta JK. (2006). Variations in the human pain stress experience mediated by ventral and dorsal basal ganglia dopamine activity. J Neurosci. 26: 10789– 10795. Sedvall G, Karlsson P, Lundin A, Anvret M, Suhara T, Halldin C, Farde L. (1994). Dopamine D1 receptor number—a sensitive PET marker for early brain degeneration in Huntington’s disease. Eur Arch Psychiatry Clin Neurosci. 243: 249–255. Seeman P, Wilson A, Gmeiner P, Kapur S. (2006). Dopamine D2 and D3 receptors in human putamen, caudate nucleus, and globus pallidus. Synapse. 60: 205–211. Serretti A, Mandelli L. (2008). The genetics of bipolar disorder: genome ‘hot regions,’ genes, new potential candidates and future directions. Mol Psychiatry. 13: 742–771. Sharp SI, McQuillin A, Gurling HM. (2009). Genetics of attention-deficit hyperactivity disorder (ADHD). Neuropharmacology. 57: 590–600. Shields PG, Lerman C, Audrain J, Bowman ED, Main D, Boyd NR, Caporaso NE. (1998). Dopamine D4 receptors and the risk of cigarette smoking in African-Americans and Caucasians. Cancer Epidem Biomar Prev. 7: 453–458. Shumay E, Chen J, Fowler JS, Volkow ND. (2011). Genotype and ancestry modulate brain’s DAT availability in healthy humans. PloS One. 6: e22754. Shumay E, Fowler JS, Wang GJ, Logan J, Alia-Klein N, Goldstein RZ, Maloney T, Wong C, Volkow ND. (2012a). Repeat variation in the human PER2 gene as a new genetic marker associated with cocaine addiction and brain dopamine D2 receptor availability. Transl Psychiatry. 2: e86. Shumay E, Logan J, Volkow ND, Fowler JS. (2012b). Evidence that the methylation state of the monoamine oxidase A (MAOA) gene predicts brain activity of MAO A enzyme in healthy men. Epigenetics. 7: 1151–1160. Sjoberg RL, Ducci F, Barr CS, Newman TK, Dell’osso L, Virkkunen M, Goldman D. (2008). A non-additive interaction of a functional MAO-A VNTR and testosterone predicts antisocial behavior. Neuropsychopharmacology. 33: 425–430. Skinbjerg M, Liow JS, Seneca N, Hong J, Lu S, Thorsell A, Heilig M, Pike VW, Halldin C, Sibley DR, Innis RB. (2010). D2 dopamine receptor internalization prolongs the decrease of radioligand binding after amphetamine: a PET study in a receptor internalizationdeficient mouse model. NeuroImage. 50: 1402–1407. Slifstein M, Kolachana B, Simpson EH, Tabares P, Cheng B, Duvall M, Frankle WG, Weinberger DR, Laruelle M, Abi-Dargham A. (2008). COMT genotype predicts cortical-limbic D1 receptor availability measured with [11C]NNC112 and PET. Mol Psychiatry. 13:.821–827. Smith L, Watson M, Gates S, Ball D, Foxcroft D. (2008). Meta-analysis of the association of the Taq1A polymorphism with the risk of alcohol dependency: a HuGE gene-disease association review. Am J Epidemiol. 167: 125–138. Smith SS, O’Hara BF, Persico AM, Gorelick DA, Newlin DB, Vlahov D, Solomon L, Pickens R, Uhl GR. (1992). Genetic vulnerability to drug abuse. The D2 dopamine receptor Taq I B1 restriction fragment length polymorphism appears more frequently in polysubstance abusers. Arch Gen Psychiatry. 49: 723–727.
Spencer TJ, Biederman J, Faraone SV, Madras BK, Bonab AA, Dougherty DD, Batchelder H, Clarke A, Fischman AJ. (2013). Functional genomics of attention-deficit/hyperactivity disorder (ADHD) risk alleles on dopamine transporter binding in ADHD and healthy control subjects. Biol Psychiatry. 74: 84–89. Staley JK, Boja JW, Carroll FI, Seltzman HH, Wyrick CD, Lewin AH, Abraham P, Mash DC. (1995). Mapping dopamine transporters in the human brain with novel selective cocaine analog [125I]RTI-121. Synapse. 21: 364–372. Stark D, Piel M, Hubner H, Gmeiner P, Grunder G, Rosch F. (2007). In vitro affinities of various halogenated benzamide derivatives as potential radioligands for non-invasive quantification of D(2)-like dopamine receptors. Bioorg Med Chem. 15: 6819–6829. Stober G, Jatzke S, Heils A, Jungkunz G, Knapp M, Mossner R, Riederer P, Lesch KP. (1998). Insertion/deletion variant (-141C Ins/ Del) in the 5' regulatory region of the dopamine D2 receptor gene: lack of association with schizophrenia and bipolar affective disorder. Short communication. J Neural Transm. 105: 101–109. Sullivan PF, Kendler KS, Neale MC. (2003). Schizophrenia as a complex trait: evidence from a meta-analysis of twin studies. Arch Gen Psychiatry. 60: 1187–1192. Sulzer D. (2011). How addictive drugs disrupt presynaptic dopamine neurotransmission. Neuron. 69: 628–649. Takahashi H, Kato M, Takano H, Arakawa R, Okumura M, Otsuka T, Kodaka F, Hayashi M, Okubo Y, Ito H, Suhara T. (2008). Differential contributions of prefrontal and hippocampal dopamine D(1) and D(2) receptors in human cognitive functions. J Neurosci. 28: 12032–12038. Tallerico T, Ulpian C, Liu IS. (1999). Dopamine D2 receptor promoter polymorphism: no association with schizophrenia. Psychiatry Res. 85: 215–219. Thompson J, Thomas N, Singleton A, Piggott M, Lloyd S, Perry EK, Morris CM, Perry RH, Ferrier IN, Court JA. (1997). D2 dopamine receptor gene (DRD2) Taq1 A polymorphism: reduced dopamine D2 receptor binding in the human striatum associated with the A1 allele. Pharmacogenetics. 7: 479–484. Tiihonen J, Hallikainen T, Lachman H, Saito T, Volavka J, Kauhanen J, Salonen JT, Ryynanen OP, Koulu M, Karvonen MK, Pohjalainen T, Syvalahti E, Hietala J. (1999). Association between the functional variant of the catechol-O-methyltransferase (COMT) gene and type 1 alcoholism. Mol Psychiatry. 4: 286–289. Tiihonen J, Vilkman H, Rasanen P, Ryynanen OP, Hakko H, Bergman J, Hamalainen T, Laakso A, Haaparanta-Solin M, Solin O, Kuoppamaki M, Syvalahti E, Hietala J. (1998). Striatal presynaptic dopamine function in type 1 alcoholics measured with positron emission tomography. Mol Psychiatry. 3: 156–161. Tsukada H, Lindner KJ, Hartvig P, Langstrom B. (1994). Effect of 6R-L-erythro-5,6,7,8-tetrahydrobiopterin on the extracellular levels of dopamine and serotonin in the rat striatum: a microdialysis study with tyrosine or tryptophan infusion. Brain Res. 635: 59–67. Tunbridge EM, Bannerman DM, Sharp T, Harrison PJ. (2004). Catechol-o-methyltransferase inhibition improves set-shifting performance and elevates stimulated dopamine release in the rat prefrontal cortex. J Neurosci. 24: 5331–5335. van de Giessen E, de Win MM, Tanck MW, van den Brink W, Baas F, Booij J. (2009). Striatal dopamine transporter availability associated with polymorphisms in the dopamine transporter gene SLC6A3. J Nuclear Med. 50: 45–52. van de Giessen E, Nimgaonkar VL, Watson AM, Kodavali C, Baas F, Tanck MW, de Win MM, van den Brink W, Booij J. (2012). Association tests of striatal DAT availability and SNPs that impact a novel splice variant in the DAT gene. J Nuclear Med. 53: 839. van Dyck CH, Malison RT, Jacobsen LK, Seibyl JP, Staley JK, Laruelle M, Baldwin RM, Innis RB, Gelernter J. (2005). Increased dopamine transporter availability associated with the 9-repeat allele of the SLC6A3 gene. J Nuclear Med. 46: 745–751.
cha p ter 4 : G enetic V ariation I m p acting D o p amine T ransmission Parameters
65
van Dyck CH, Seibyl JP, Malison RT, Laruelle M, Wallace E, Zoghbi SS, Zea-Ponce Y, Baldwin RM, Charney DS, Hoffer PB. (1995). Age-related decline in striatal dopamine transporter binding with iodine-123-beta-CITSPECT. J Nuclear Med. 36: 1175–1181. Vandenbergh DJ, Persico AM, Hawkins AL, Griffin CA, Li X, Jabs EW, Uhl GR. (1992). Human dopamine transporter gene (DAT1) maps to chromosome 5p15.3 and displays a VNTR. Genomics. 14: 1104–1106. Volkow ND, Chang L, Wang GJ, Fowler JS, Ding YS, Sedler M, Logan J, Franceschi D, Gatley J, Hitzemann R, Gifford A, Wong C, Pappas N. (2001). Low level of brain dopamine D2 receptors in methamphetamine abusers: association with metabolism in the orbitofrontal cortex. Am J Psychiatry. 158: 2015–2021. Volkow ND, Fowler JS, Wolf AP, Schlyer D, Shiue CY, Alpert R, Dewey SL, Logan J, Bendriem B, Christman D, et al. (1990). Effects of chronic cocaine abuse on postsynaptic dopamine receptors. Am J Psychiatry. 147: 719–724. Volkow ND, Wang GJ, Fowler JS, Ding YS, Gur RC, Gatley J, Logan J, Moberg PJ, Hitzemann R, Smith G, Pappas N. (1998). Parallel loss of presynaptic and postsynaptic dopamine markers in normal aging. Ann Neurol. 44: 143–147. Volkow ND, Wang GJ, Fowler JS, Logan J, Gatley SJ, Hitzemann R, Chen AD, Dewey SL, Pappas N. (1997). Decreased striatal dopaminergic responsiveness in detoxified cocaine-dependent subjects. Nature. 386: 830–833. Volkow ND, Wang GJ, Fowler JS, Logan J, Hitzemann R, Ding YS, Pappas N, Shea C, Piscani K. (1996). Decreases in dopamine receptors but not in dopamine transporters in alcoholics. Alcohol Clin Exp Res. 20: 1594–1598.
Wang GJ, Volkow ND, Logan J, Pappas NR, Wong CT, Zhu W, Netusil N, Fowler JS. (2001). Brain dopamine and obesity. Lancet. 357: 354–357. Weinberger DR, Laruelle M. (2002). In Davis KL, et al., eds., Neuropsychopharmacology: the fifth generation of progress (pp. 883–885). Philadelphia: Lippincott, Williams, and Wilkins. Willeit M, Praschak-Rieder N. (2010). Imaging the effects of genetic polymorphisms on radioligand binding in the living human brain: a review on genetic neuroreceptor imaging of monoaminergic systems in psychiatry. Neuroimage. 53: 878–892. Wilson ME, Fisher J, Brown J. (2005). Chronic subcutaneous leptin infusion diminishes the responsiveness of the hypothalamicpituitary-adrenal (HPA) axis in female rhesus monkeys. Physiol Behav. 84: 449–458. Wolf ME, Roth RH. (1990). Autoreceptor regulation of dopamine synthesis. Ann N Y Acad Sci. 604: 323–343. Wu K, O’Keeffe D, Politis M, O’Keeffe GC, Robbins TW, Bose SK, Brooks DJ, Piccini P, Barker RA. (2012). The catechol-O-methyltransferase Val(158)Met polymorphism modulates fronto-cortical dopamine turnover in early Parkinson’s disease: a PET study. Brain. 135: 2449–2457. Zhang Y, Bertolino A, Fazio L, Blasi G, Rampino A, Romano R, Lee ML, Xiao T, Papp A, Wang D, Sadee W. (2007). Polymorphisms in human dopamine D2 receptor gene affect gene expression, splicing, and neuronal activity during working memory. Proc Natl Acad Sci U S A. 104: 20552–20557. Ziebell M, Holm-Hansen S, Thomsen G, Wagner A, Jensen P, Pinborg LH, Knudsen GM. (2010). Serotonin transporters in dopamine transporter imaging: a head-to-head comparison of dopamine transporter SPECT radioligands 123I-FP-CIT and 123I-PE2I. J Nuclear Med. 51: 1885–1891.
6 6 Part I : I maging G enetics and N eurochemistry
5. IMAGING GENE TICS OF DOPAMINE SYNAPTIC TERMINAL ACTIVIT Y Giuseppe Blasi and Alessandro Bertolino
I N T RO DU C T I O N Psychiatric disorders are currently defined based upon a collection of symptoms and signs. Discussions about the phenomenology of these disorders date back to the nineteenth century, when eminent psychiatrists attempted to dissect different behaviors into differential diagnoses. Since then, a series of interesting phenomenological paradigms have been put forth with the objective of better characterizing patients, obtaining greater knowledge about the pathophysiology and the etiology of these disorders, and finding new treatments. A new effort in the categorical definition and characterization of these disorders has recently been published in the form of the fifth edition of the Diagnostic and Statistic Manual for Mental Disorders (DSM-5). This manual has maintained its agnostic approach toward etiology by further describing the phenomenology of psychiatric disorders. After all these years, though, it is clear that the understanding of the etiology and the treatment of these disorders have not been significantly improved because of fundamentally different categorical definitions. Psychiatric disorders are by nature heterogeneous since they describe complex behaviors that are likely to emerge as a downstream effect of a series of factors, including the environment, complex interactions, and emergent phenomena. Germane to this heterogeneity is the notion that psychiatric disorders still have no biological validity or validation and that they cannot be easily reduced to unique molecular alterations or deficits. In this regard, the NIMH has launched a recent effort as a framework to guide the classification of patients for research studies based upon dimensions of behavior (Research Domain Criteria, RDoC). These dimensions are directly interfaced with genomics, neuroscience, and
behavioral science, in the hope of making progress in explicating etiology and suggesting new treatments. Even though a precise understanding and definition of nosology are still lacking, epidemiological studies have clearly demonstrated that major psychiatric disorders recognize genetic variation as the major cause of risk, including schizophrenia, bipolar disorder, and autism spectrum disorders (ASD) (McGuffin, Riley, and Plomin 2001). Consistently, major efforts are being made to characterize common and rare genetic variants that predispose to these disorders. Several genome-wide association (GWA) studies have demonstrated different variants crossing the genome-wide statistical threshold for schizophrenia, bipolar disorder, and ASD (Persico and Napolioni 2012; Craddock and Sklar 2013; Mowry and Gratten 2013). Other studies have demonstrated how rare variants across the genome increase risk for these disorders (Malhotra and Sebat 2012). Moreover, consistent with the assertion that these diagnostic categories do not have biological validity, several other reports have indicated how these variants have pleiotropic effects and map onto different nosological categories (Smoller 2013). However, in our opinion, these pleiotropic effects do not imply that these brain disorders are the same, or that diagnostic categories should be redefined based on this genetic information. Based on the experience gained with environmental risk factors, it would have been rather surprising not to find pleiotropic effects. For example, cigarette smoking is a risk factor for lung cancer, heart disease, and prostate cancer, but nobody questions that these are different disorders. It is indeed very evident that schizophrenia, bipolar disorder, and ASD fundamentally have a different phenomenology and different behavioral alterations, despite some similarities and unclear diagnostic boundaries in some cases.
67
Another point that has clearly been made by many recent genetic studies is that risk for disorders is explained by a large number of genetic variants which may also differ from one individual to another (Ripke et al. 2013). These studies make it clear that different combinations of genetic variants may predispose to these disorders. However, it is also clear that there has to be some common final pathway onto which genetic risk converges, otherwise there would be no identifiable clinical and diagnostic entity because each patient would present with a unique set of symptoms and signs. In other words, it is likely that genetic risk has convergence onto molecular pathways and brain circuitry that allows identification of common phenomenology (Harrison and Weinberger 2005). Thus, once genetic variation conferring risk for these disorders has been identified, we will face the challenge of putting this information together with the final phenotype. In fact, what is still in many ways lacking in the studies about genetic risk for these psychiatric disorders is knowledge of what genetic variation means in terms of brain function, which is the endgame of what these studies should provide. Imaging genetics is the effort to map genetic variation onto brain circuitry and activity. It has permitted a tremendous leap toward a new dimension of brain imaging because it allows to link two fields of neuroscience often proceeding separately. Using a “genetic” perspective, the use of brain imaging provides the means to investigate the association of variation in DNA architecture with complex phenotypes in the living human brain. Using a “brain imaging” perspective, the use of genetic information allows a thorough understanding of the biological bases of individual variation of function and structure of brain networks. This approach is particularly useful when investigating genetic variation associated with risk for complex disorders. One way to reduce the complexity of searching genes for such complex diseases is to use biological measures relevant to the pathophysiology, rather than to the diagnosis in and of itself. These biological measures have also been defined as “endophenotypes” or “intermediate phenotypes” (Gottesman and Gould 2003). These are phenotypic variables that should be heritable, that co-segregate with a psychiatric illness, yet that should be present even when the disease is not (i.e., state independent), and that can be found in non-affected family members at a higher rate than in the general population (Gottesman and Gould 2003). They should also be more easily quantifiable and involved in a biologically plausible mechanism of pathogenesis. In other words, endophenotypes should provide a genetically tractable target lying in the gap between gene and clinical diagnosis and useful to understand something
about the biology of a psychiatric disorder. Such phenotypes may improve the power of genetic studies by reducing genetic heterogeneity or providing phenotypes with a simpler genetic architecture (Moore, Le, and Fan 2013). All of these characteristics make imaging phenotypes an ideal tool for studying association with genetic variation in order to determine the etiology and pathophysiology of complex brain disorders. Importantly, the study of brain imaging phenotypes in healthy subjects is preparatory to investigation in patients. In fact, such investigation allows us to clarify the genotype-phenotype relationship in physiological conditions that are not biased by state variables as pharmacological treatment. Genetic variation in dopamine signaling and related brain imaging phenotypes are well suited to converge onto imaging genetics and the intermediate phenotype approach. This is particularly true for schizophrenia because (1) anomalies in prefrontal activity are considered a well-established intermediate phenotype for this brain disorder; (2) schizophrenia is a complex brain disorder with a heavy genetic load; and (3) dopamine is a strongly hypothesized constituent of the pathophysiology of schizophrenia and is a crucial modulator of prefrontal function. In this context, the D2 receptor, the main target of antipsychotic treatment to date, is a particularly relevant component of dopamine signaling. D2-related genetic variation will be the focus of this chapter, the aim of which is to exemplify how investigation of the relationship between variation in dopamine-related genes and brain imaging measures may help to progess toward a better understanding of risk for schizophrenia phenotypes. However, it should be kept in mind that such understanding cannot be achieved without considering brain imaging phenotypes in concert with other biological and behavioral measures, which together may allow us to draw a more general picture of the relevance of dopamine genes for the physiology and pathophysiology of schizophrenia. Even more relevant from a clinical point of view, this multi-phenotypic approach, including imaging genetics, may also add new knowledge on the link between genetic variation and individual variability in response to antipsychotic treatment. However, the complexity of the biology of the human brain does not only lie on the effects of stable and invariant characteristics of DNA. Regulation and dysregulation of brain physiology may also be associated with environmental inputs, inducing molecular cascades involving DNA adjustments. Relevant implications for gene-environment interactions are arising through the study of epigenetics, that is, the investigation of heritable changes in gene activity or function that are not associated with any change
6 8 Part I : I maging G enetics and N eurochemistry
of the DNA sequence itself (Moore, Le, and Fan 2013). Epigenetic mechanisms include DNA methylation, which is influenced by environmental stressors (Jirtle and Skinner 2007). The study of the relationship between genetic and epigenetic influxes is starting to shed light on how genetic variation modulates modes of interaction between environmental factors and gene products, and how these mechanisms reflect on regulation of brain physiology. Here we will also describe examples on this topic related to dopaminergic signaling. G E N E T I C VA R I AT I O N A N D D 2 D O PA M I N E R EC E P T O R S I G N A L I N G D2 receptors are strongly involved in disparate brain processes. For example, several studies support D2 modulation of emotion processing (Pezze and Feldon 2004), as indicated by results revealing that administration of D2 agonists or antagonists in animal models modulates emotional responses, including emotional reactivity (Gendreau et al. 1998) and learning (Greba, Gifkins, and Kokkinidis 2001). Furthermore, the well-known relationship between D2 and motor abilities has been elucidated by studies in animal models showing that knockout of D2 receptors and administration of D2 antagonists in mice impair locomotion and motor coordination (Kelly et al. 1998). Moreover, several reports have deeply characterized the relevance of D2 receptors for high-order cognitive processing subtended by the cortico-thalamo-striatal-circuit (Alexander, DeLong, and Strick 1986). For example, a series of results suggest that D2 receptors regulate prefrontal activity during working memory, acting locally and directly (Wang, Vijayraghavan, and Goldman-Rakic 2004) as well as indirectly via modulation of striatal signaling (Cropley et al. 2006; Kellendock et al. 2006). This crucial link is also supported by evidence indicating that D2 receptors are found in the prefrontal cortex on the body of pyramidal neurons, where they presumably regulate general excitability (Goldman-Rakic 1995, 1999), as well as in the striatum on pre- and post-synaptic terminals (Tisch et al. 2004). Consistent with the relevance of D2 receptors for multiple aspects of brain processing, D2 signaling abnormalities have been strongly implicated in the pathophysiology of brain disorders such as schizophrenia. This possibility is supported by the fact that D2 receptors are privileged targets of antipsychotic drugs, which antagonize their activity (Kapur, Zipursky, and Remington 1999). Consistently, previous reports have suggested an association between psychosis and relatively greater D2 density in striatum (Laruelle
1998). Moreover, clinical symptoms and cognitive deficits in patients have been associated with abnormal D2 signaling (Wang, Vijayraghavan, and Goldman-Rakic 2004; Kellendock et al. 2006; Durstewitz and Seamans 2008). On this basis, investigation of the D2 gene is crucial to discover how functional variation modulates D2 signaling in order to relate D2 genetic variability to neuronal physiology. Once functional genetic variation is established, its association with brain imaging phenotypes using an imaging genetics approach may increase the understanding of the effects of D2 genetic variability on physiology at the brain network level. The consistency of genotype-phenotype associations at molecular and brain network levels may in turn orient further investigation into the behavioral level. The relevance of findings for schizophrenia may be then inferred based on a general picture including molecular, brain imaging and behavioral correlates, which may shed light on aspects of the pathophysiology of this brain disorder. Using, at least in part, this line of reasoning, a series of studies have investigated the association of different phenotypes with variation in the D2 gene (DRD2), which is located on chromosome 11 (11q23). The single nucleotide polymorphism (SNP) rs1799732, referred to as −141C Ins/Del, has been one of the first DRD2 variants investigated. This is an insertion/deletion of a cytosine in the 5' flanking site of the gene, and it has been associated in vitro with DRD2 expression (Ins > Del) (Arinami et al. 1997). Furthermore, another study has reported an effect of this polymorphism on reward-related ventral striatal response (Ins/Del > Ins/Ins) (Forbes et al. 2009). Other studies have focused on another DRD2 polymorphism, rs6277 (also referred to as T957C). The T allele of this synonymous SNP has been associated with altered mRNA folding, stability, and translation (Duan et al. 2003). Not always consistently, other reports have linked the T allele with greater striatal (Hirvonen et al. 2004; erratum in Hirvonen et al. 2005) and lower extrastriatal (Hirvonen et al. 2009) D2 binding potential. Furthermore, this SNP has also been associated with behavioral phenotypes, as those related to reinforcement learning (Frank et al. 2007). Further investigation has been devoted to another genetic variant, rs1800497, also referred to as Taq1A. This polymorphism, previously believed to be located within DRD2, is actually positioned 9489 base pairs downstream from the 3′ end of this gene. This location pertains to the last exon of another gene, ANKK1, and implies an amino acid substitution in the related protein ankyrin (Glu713Lys). A series of binding studies have associated Taq1A with decreased striatal D2 density (Noble et al. 1991; Thompson et al. 1997;
cha p ter 5 : I maging G enetics of D o p amine S yna p tic T erminal A ctivity
69
Pohjalainen et al. 1998; Jonsson et al. 1999), even if this result has not always been replicated (Laruelle, Gelernter, and Innis 1998). Furthermore, other reports have implicated Taq1A in other brain phenotypes, including glucose metabolism (Noble et al. 1997) and midbrain volume (Cerasa et al. 2009), as well as anterior cingulate gray matter volume in interaction with another functional polymorphism, BDNF Val66Met (Montag et al. 2010). Moreover, Taq1A has also been linked with activity in the striatum and in medial frontal cortex, with the functional connection between the prefrontal cortex and the hippocampus during learning tasks, as well as with related behavior (Klein et al. 2007; Jocham et al. 2009). The association between Taq1A and anterior cingulate, putamen, and amygdala activity during the processing of negative facial expression has been also reported (Lee et al. 2011). Finally, the interaction between this polymorphism and administration of the dopamine agonist bromocriptine has also predicted activity of the reward network (Kirsch et al. 2006). Even if this body of results consistently indicates the association of Taq1A with brain physiology, it is not clear if these findings are because of a D2-dependent mechanism. As reported above, Taq1A is located in ANKK1, and it implies an amino acid substitution in the related protein. Thus, it is possible that the effects of Taq1A on imaging and behavioral findings involve different biological pathways. Another possibility is that DRD2 variants in linkage disequilibium with Taq1A may be more directly involved in modulating dopamine-related phenotypes previously linked to Taq1A. Overall, none of the reported studies has investigated the association of genetic variation with D2-related phenotypes ranging from the molecular to the behavioral level. This is the methodological approach that we have applied, starting from the notion that two isoforms of the D2 receptor are known. The D2 long isoform (D2L) has the canonical sequence and is mainly found at the post-synaptic level, where it transmits D2 signaling to intra-neuronal molecular cascades; the D2 short (D2S) isoform is mainly located pre-synaptically, where it serves as an autoreceptor and regulates dopamine synthesis and release (Usiello et al. 2000). These two isoforms are both coded by DRD2 by mechanisms of alternative splicing. In an extensive investigation of this gene, we have focused on the potential functional effects of regulatory polymorphisms affecting gene transcription and mRNA splicing analyzing 23 SNPs. Using allelic expression imbalance (AEI), we found that two intronic SNPs, rs2283265 (G/T) and rs1076560 (G/T), which are in high linkage disequilibrium (LD, 0.932 in Caucasians), affect the D2S/D2L ratio of splice variants
both in human prefrontal and striatal tissue, as well as in vitro using mini-gene constructs (G/T). In particular, the T allele shifts splicing from D2S to D2L, decreasing the D2S/ D2L ratio relative to the G allele. Importantly, these two SNPs regulate physiology in the prefrontal cortex and striatum during working memory, with GG subjects showing attenuated prefrontal and striatal fMRI BOLD response compared with T carrier individuals. Furthermore, rs1076560 and rs2283265 also modulate a series of behavioral phenotypes related to working memory and attentional performance, with GG subjects having greater cognitive accuracy compared to heterozygous individuals (Zhang et al. 2007). Together, the imaging and behavioral results suggest greater efficiency of GG subjects during cognitive processing, that is, greater behavioral performance in spite of lower prefrontal activity relative to individuals with the T allele (Zhang et al. 2007). Importantly, the −141C Ins/Del polymorphism cited above was not associated with detection of different mRNA expression in this report (Zhang et al. 2007). Furthermore, the Taq1A variation was in linkage disequilibrium with rs1076560 and rs2283265 in the study sample. In a following case-control study including healthy subjects and patients with schizophrenia (Bertolino et al. 2009), we replicated the association of the T allele of the DRD2 SNP rs1076560 (Zhang et al. 2007) in healthy subjects, with lower D2 short/long ratio prefrontal postmortem expression and with working memory behavioral performance. Furthermore, we also found similar molecular and behavioral results in patients with schizophrenia. On the other hand, prefrontal activity during working memory measured with fMRI revealed an SNP (rs1076560) by diagnosis interaction. In particular, healthy subjects with the T allele had reduced performance and greater brain activity (i.e., they are less efficient), while a different association was present in patients. Specifically, these individuals perform worse than controls and have reduced brain activity. In this physiological and behavioral context, T allele patients had reduced performance and reduced activity, which may indicate that rs1076560 T allele individuals with schizophrenia do not fully engage prefrontal–striatal resources. Together, these results suggest that the GG genotype is advantageous in both controls and patients. In another study (Bertolino et al. 2010), healthy subjects carrying the T allele of rs1076560 were also associated with a reduction of measures of striatal dopamine signaling, that is, binding of the radioligands [123I] iodobenzamide (IBZM) for D2 receptors and [123I] FP-CIT for the dopamine transporter. In the same
7 0 Part I : I maging G enetics and N eurochemistry
study, rs1076560 also differentially predicted the correlation between striatal D2 signaling and activity of the prefrontal cortex during working memory as measured with BOLD fMRI. Together, these findings support the notion that functional SNPs in the D2 receptor gene modulate D2 expression in the brain, conferring variability on neurobiological and behavioral responses associated with dopaminergic signaling and cognition. Furthermore, they also indicate that the relationship between specific brain imaging phenotypes and DRD2 genetic variation may be modified by the diagnosis of schizophrenia. D2 signaling is also relevant to physiology related to emotion processing. Studies in animal models have indicated that dopamine D2 receptors are involved in emotional behavior (Gendreau et al. 1998; Guarraci et al. 2000; Greba, Gifkins, and Kokkinidis 2001) and in activity of the amygdala and the prefrontal cortex, brain regions centrally implicated and functionally connected during emotion processing (Rosenkranz and Grace 2001; Pezze and Feldon 2004; Seamans and Yang 2004; Stein et al. 2007). Thus, the study of the association between DRD2 variation and emotional correlates is warranted. In another study (Blasi et al. 2009), the DRD2 functional SNP rs1076560 described earlier has been investigated with regard to its possible modulatory effect on brain responses during emotion processing and on emotional behavior in healthy subjects. The fMRI analysis indicated that GG subjects had greater left amygdala activity during implicit processing and greater left dorsolateral prefrontal cortex response during explicit processing of facial expressions. Consistently, activity in the dorsolateral prefrontal cortex of GG subjects was positively correlated with ratings of the emotional impact of facial stimuli presented during the task. At the behavioral level, GG subjects were slower when explicitly processing facial stimuli. Furthermore, GG subjects had reduced emotion control scores obtained with the Big Five Questionnaire, which measures a series of personality traits according to the Big Five Factors Model (McCrae and Costa 1987). Together, these findings suggest that DRD2 rs1076560 is associated with physiology and behavior related to emotion processing and that the G allele of this SNP confers greater reactivity to emotionally salient stimuli. Other findings have extended knowledge about the association of DRD2 rs1076560 with brain physiology. More in detail, a recent study (Fazio et al. 2011) has linked DRD2 rs1076560 with response in the motor network. In particular, healthy subjects homozygous for the G allele had lower fMRI responses in the left basal
ganglia, thalamus, primary motor cortex, and supplementary motor area, the activity of which was inversely correlated with left striatal DAT binding, as measured with [123I] FP-CIT SPECT. Furthermore, another investigation (Sambataro et al. 2013) has focused on the effects of genetic variation on the Default Mode Network (DMN), a set of brain regions with a greater activation during functional rest than during active task performance (Raichle et al. 2001). In this study (Sambataro et al. 2013), a complex association between DRD2 rs1076560 and DMN has been found, such that this SNP affected specific components of DMN as well as their relationship with striatal DAT binding. Consistent with these results, the association of rs1076560 with prefronto-striatal connectivity during the manipulation of working memory stimuli has been described in another study using dynamic causal models, in which the “inhibitory” effective connectivity from the dorsolateral prefrontal cortex to the striatum was increased in GG compared to T carrier individuals (Tan et al. 2012). By putting all these results together, a series of points may be inferred about their general meaning and the methodological approach used. As a first and more general point, a molecular-to-system approach is useful to clarify the possible biological significance at the brain network level of the specific molecular effects of genetic variation. In other words, more specific knowledge on functional molecular effects of DNA variability potentially allows more solid hypotheses on association of such variation with imaging phenotypes, as well as better interpretation of the results at the network level. Importantly, coherent molecular and brain network findings may open further scenarios implicating the investigation of genotype effects at the behavioral level. Given the biological distance between genes and behavior, such behavioral effects are not always easily conceivable. On the other hand, strong and replicated imaging genetic findings may push toward the investigation of genetic variation-behavior association in larger samples, which increase the power of detection. As a second and more particular point, these findings suggest that functional genetic variability in the DRD2 gene is associated with multiple brain phenotypes at the network level. This suggestion is also supported by relevance of D2 signaling for a broad range of brain processes including cognitive, emotional, and motor processing. Finally, it is worth noting that all these phenotypes have been related to schizophrenia, for the pathophysiology of which D2 signaling and genetic risk are considered key determinants (Moore, Le, and Fan 2013).
cha p ter 5 : I maging G enetics of D o p amine S yna p tic T erminal A ctivity
71
DRD2 VARIATION AND GENOT Y PEGENOT YPE INTERACTIONS
The previous enumerated results identify a link between physiological as well as behavioral phenotypes and DRD2 variation. However, D2 signaling is the product of complex interactions between the D2 receptor and several molecules acting at extra- and intra-neuronal sites. Thus, it is likely that the interaction between functional genetic variations within genes coding for such molecules modulates D2 signaling and related phenotypes. In other words, the physiological effects of a single genetic variation may be added to, or may vary as a function of, the genetic background determined by other polymorphisms in dopamine-related genes, configuring genotype-genotype additive or epistatic interactions. A specific case of this general picture is one related to the relationship between D2 receptors and the dopamine transporter (DAT), one of the main contributors of dopamine inactivation via reuptake in the synapsis. Earlier in vitro studies have indicated that DAT and D2 interact (Lee et al. 2007), especially on pre-synaptic terminals (Bolan et al. 2007). Functional variation in the DAT gene has also been characterized. In particular, a variable number of tandem repeat (VNTR) polymorphism in the DAT gene (SLC6A3) has been described (DAT 3′ VNTR) (62). Alleles of this polymorphism range from 3 to 11 repeats, with the 9- and 10-repeat alleles by far the most common (Vandenbergh et al. 1992). As compared with the 9-repeat allele, the 10-repeat allele has been associated with an increased gene expression both in vitro (Mill et al. 2002; VanNess, Owens, and Kilts 2005) and in vivo (Heinz et al. 2000). Even though other variants in this gene may also be significantly affecting the transcript (Pinsonneault et al. 2001), several studies have reported that the 10-repeat allele is associated with more focused cortical activity during memory and attention in healthy subjects (e.g., (Fossella et al. 2002; Bertolino et al. 2006). On this basis, another study (Bertolino et al. 2009) in healthy subjects tested the hypothesis that DRD2 rs1076560 interacts with DAT 3′ VNTR in shaping functional activity during memory performance as well as morphology of the brain. Results indicated a strong significant interaction between DRD2 and DAT genetic variants on prefrontal and striatal brain activity during memory processes. In particular, the differential effect of DAT 3′ VNTR on BOLD activity was greater in the context of the DRD2 GT genotype relative to GG. A similar modulation was present on striatal gray matter content. Moreover, all these prefrontal and striatal interactions were non-linear, fitting well with earlier models describing
the relationship between dopamine signaling and prefrontal neuronal activity. Specifically, this link may be shaped following an inverted-U function in which too little or too much dopamine is associated with non-optimal cognitive processing (Seamans and Yang 2004). These results were also corroborated by data in mice showing in vivo D2-DAT striatal interaction in wild-type and D2 knockout animals (Bertolino et al. 2009). Together, these findings provide support for the hypothesis that the pooled effects of genetic variation associated with D2 signaling and with mechanisms of dopamine catabolism in the synapsis shape the physiological inverted U relationship between dopamine signaling and prefronto-striatal network morphology and activity during cognition. Consistent with the relevance of the interaction between genetic variation in D2 and DAT for modulation brain phenotypes, another study (Meyer et al. 2012) has indicated an additive effect of Taq1A and DAT 3′ VNTR on error-related negativity (ERN) in children performing an inhibition task, such that Taq1 A1/DAT 9 carriers had more negative ERN. In another study (Stice et al. 2012) exploring reward-related networks, the variants DRD2 −141 Ins/Del and DAT 3′ VNTR have been studied together with others: a 48-bp exon 3 VNTR polymorphism in the gene coding for D4 receptors (DRD4), and an SNP in the cathecol-o-methyltransferase gene (COMT Val158Met). Results indicate a negative association between multilocus composite scores and striatal activity during monetary reward. This finding is interpreted by the authors in terms of a negative relationship between the dopamine signaling capacity in the reward network and striatal activity related to reward processing. Another piece of the puzzle relative to D2-related genetic interactions pertains to the intra-neuronal propagation of D2 signaling. Downstream from D2 receptors, different molecular pathways have been identified: the classic cAMP-PKA pathway and another cAMP-independent pathway that includes the serine/threonine protein kinase AKT1, which phosphorylates to inhibit another protein kinase, glycogen-synthase-kinase 3β (GSK-3β). The specific relationship between D2 receptor signaling and AKT1 has been elucidated by data indicating that D2 stimulation by dopamine inhibits AKT1 signaling through dephosphorylation via the β-arrestin 2/phosphatase PP2A complex (Beaulieu, Gainetdinov, and Caron 2007; Freyberg et al. 2010). Importantly, AKT1 levels in lymphoblasts and in the prefrontal cortex of patients with schizophrenia are reduced (Emamian et al. 2004; Thiselton et al. 2008). Furthermore, the D2 antagonist antipsychotic drug clozapine increases AKT1 and GSK-3β phosphorylation, as well
7 2 Part I : I maging G enetics and N eurochemistry
as total cellular and intranuclear levels of β-catenin (Kang et al. 2004), a crucial factor for gene expression that is inhibited by GSK-3β activity (Freyberg et al. 2010). The gene coding for AKT1 (14q32.32) has been associated with schizophrenia (Emamian et al. 2004; Ikeda et al. 2004; Schwab et al. 2005; Norton et al. 2007; Thiselton et al. 2008). Furthermore, the A allele in a synonymous SNP in AKT1 (rs1130233 −G > A) is associated with reduced AKT1 protein levels in lymphoblasts (Harris et al. 2005; Tan et al. 2008), reduced cognitive performance, and inefficient prefronto-striatal activity during cognition, as well as with diagnosis of schizophrenia (Tan et al. 2008). A recent report (Blasi et al. 2011) presents convergent results implicating epistatic interaction of DRD2 rs1076560 and AKT1 rs1130233 on a series of progressively more distal phenotypes from gene effects and implicated in schizophrenia. As phenotypes of interest, this study included molecular, brain imaging, and behavioral correlates as well as response to antipsychotic treatment. Results indicated that the interaction between the “risk” alleles of these polymorphisms—the rs1076560 T allele and the rs1130233 A allele—are associated in healthy subjects with reduced AKT1 protein levels and reduced phosphorylation of GSK-3β in peripheral blood mononuclear cells. These results are consistent in indicating that genotypic interaction between functional polymorphisms in DRD2 and AKT1 reflects on D2 signaling transduction along the D2 cAMP-independent pathway, with possible impact on mechanisms of genes involved in neurodevelopmental processes mediated by β-catenin and possibly relevant to schizophrenia. Brain imaging results in healthy subjects were consistent with molecular findings in indicating a DRD2 rs1076560 by AKT1 rs1130233 interaction on brain phenotypes. Here, the physiological increase of fMRI cingulate cortex response associated with a parametric increase in attentional control load was not present in DRD2 GT/AKT1 A carrier individuals, whose activity dropped off from the intermediate to the higher level of attentional control. Indeed, this pattern of activity was reminiscent of that found in patients with schizophrenia in an earlier study with the same task (Blasi et al. 2010). Notably, the interaction between rs1076560 and rs1130233 was also associated with correlates at the behavioral level during attentional processing. In fact, DRD2 GT/AKT1 A carrying subjects had reduced accuracy at the Continuous Performance Test relative to all other genotype configurations. Together, these results on the association of DRD2 and AKT1 genetic variation with attentional processes once again suggest their possible relationship with schizophrenia, for which attentional deficits are the key features (Weickert et al. 2000). Such a link was
further substantiated in the same study by the interaction of DRD2 rs1076560 and AKT1 rs1130233 on response to treatment with antipsychotics in patients with schizophrenia. In this case, DRD2 GT/AKT1 A carrier patients with schizophrenia had better response after 8 weeks of olanzapine monotherapy, as measured with the positive and negative syndrome scale (PANSS). These results are in line with the notion that the similar antipsychotic olanzapine blocks D2 signaling and with data showing that second-generation antipsychotics activate AKT1 (Kang et al. 2004) or mimic AKT1 activity, increasing GSK-3β phosphorylation (Li et al. 2000). Interestingly, patients with schizophrenia carrying the two “risk” alleles (DRD2 T and AKT1 A) had better response to treatment with olanzapine. This beneficial effect would seem at odds with the effects of these two alleles in healthy subjects and can be interpreted in different ways. One interpretation is that DRD2 T and AKT1 A patients may respond better because their dopamine cAMP-independent pathway is more profoundly altered in terms of dopamine D2 signaling, and thus there is more “room” for improvement by treatment with a drug that specifically acts on it. Another interpretation is that genetic variants interact with dysregulated levels of dopamine in patients, determining an effect that is not possible to elicit by studying only healthy subjects. A series of considerations may be derived looking at the entirety of these results. A first general point pertains to the relevance of the investigation of genetic interactions for disentangling the relationship between gene effects and schizophrenia-related phenotypes. This approach is also consistent with the obvious concept that complex phenotypes at brain imaging and behavioral levels are modulated by multiple mutually and epistatically interacting genetic variations. Furthermore, it is congruent with the strong possibility that genetic risk for schizophrenia is produced by multiple interacting variations, each adding a small effect to the pathophysiology of this brain disorder. As a more specific point, these results suggest that genetic variation within the cAMP-independent D2 signaling pathway is relevant for molecular, brain imaging, and behavioral phenotypes associated with schizophrenia. Consistent with this contention, in a recent study (Blasi et al. 2013) we have investigated if genetic variation in GSK-3β, which also participates in this pathway, is associated with prefrontal phenotypes and the diagnosis of schizophrenia. GSK-3β is implicated in a broad range of neurodevelopmental processes and crucially contributes to intraneuronal transduction of signaling from the extraneuronal space to the nucleus (Hur and Zhou 2010), including those related to serotonin, neuregulin, Wnt, and DISC1.
cha p ter 5 : I maging G enetics of D o p amine S yna p tic T erminal A ctivity
73
Consistent evidence also suggests that D2 signaling modifies the state of phoshorylation and thus the activity of GSK-3β via AKT1 (Beaulieu, Gainetdinov, and Caron 2007; Freyberg et al. 2010). Using the translational genetic approach described earlier, we have first performed in silico prediction on the functional consequences of GSK-3β genetic variation on gene function. Based on these predictions, we have selected an SNP, rs12630592, with possible effects on mechanisms of splicing and thus on GSK-3β mRNA expression. Thus, we have tested this prediction in the postmortem prefrontal cortex of non-psychiatric individuals, as well as in lymphocytes of healthy subjects. The results of these analyses were consistent in indicating that the T allele of this SNP, compared with the G allele, was associated with lower GSK-3β mRNA and protein expression, with lower expression of phosphorylated GSK-3β, as well as with lower expression of β-catenin, a GSK-3β substrate involved in nuclear processes including gene expression. At the brain imaging level, the T allele was associated with lower prefrontal cortical thickness and reduced prefrontal activity during working memory and attentional control processing. Finally, the T allele was also associated with diagnosis of schizophrenia in a case-control investigation. Again, considered together, these results add elements that highlight the relevance of the cAMP independent D2 signaling pathway for the pathophysiology of schizophrenia. Furthermore, they call for a further investigation of the genetic interactions between molecules within this signaling pathway to verify if epistatic genetic mechanisms increase the risk for molecular, brain imaging, and behavioral phenotypes of relevance for this brain disorder. BE YO N D I M AG I N G G E N E T I C S: I M AG I N G E P I G E N E T I C S O F D O PA M I N E S I G N A L I N G Imaging genetics is providing neuroscientists with new and important findings about the implication of genetic variation on in vivo brain phenotypes. Together with this increasing knowledge, new scenarios are taking the stage in neuroscientific research, based on the notions that genes by themselves are not the only modulators of the physiology of the brain. Environmental inputs are also central in conferring specific characteristics to individual response to specific stimuli, and to explain part of the variance associated with brain disorders (McGuffin, Riley, and Plomin 2001). Thus, the study of the relationship between genes and environment is of particular interest to understand how the brain works. This field has been particularly enriched by
epigenetics, the investigation of non-structural, heritable changes in gene expression arising without changes in the DNA sequence, but occurring via different mechanisms, including modification in DNA methylation (Pidsley and Mill 2011). Of note, studies in rodents have indicated that DNA methylation is sensitive to environmental stressful exposures (Jirtle and Skinner 2007; Szyf, McGowan, and Meaney 2008). Furthermore, other studies in animal models have revealed a relationship between memory and neuronal DNA methylation associated with early life stress (Miller and Sweatt 2007; Murgatroyd et al. 2009; Miller et al. 2010). Moreover, experiments in human monozygotic twins have suggested a relationship between dynamic epigenetic mechanisms and the environment (Fraga et al. 2005; Wong et al. 2010). The main part of DNA methylation occurs at a cytosine (C) preceding a guanine nucleotide (G), constituting a so-called CpG site. On the other hand, SNPs may alter CpG sites, thus hypothetically modulating DNA methylation. This is the case of a very well investigated SNP in the gene coding for COMT, a key enzyme for inactivation of prefrontal dopamine (Gogos et al. 1998). This non-synonymous SNP, rs4680 (Val158Met), implies a substitution of a guanine with an adenine, resulting in a valine to methionine change in the mature protein with an effect on its function. In particular, the guanine form is associated with greater activity and thus greater dopamine inactivation relative to the adenine form. Furthermore, COMT rs4680 guanine has also been associated with more complex phenotypes, including a blunted response to stress (Enoch et al. 2003; Smolka et al. 2005) and inefficient prefrontal and cingulate activity during cognitive processing (Blasi et al. 2005; Egan et al. 2001). More important for epigenetic mechanisms, while rs4680 guanine has a CpG site, rs4680 adenine does not, thus possibly implicating the modulation of trajectories of methylation of COMT. A recent study (Ursini et al. 2011) has investigated the possible relationship between COMT rs4680, COMT methylation, COMT expression, stress, prefrontal activity, and behavior associated with working memory processing. Results indicate that methylation of the guanine allele measured from peripheral blood mononuclear cells of homozygous healthy individuals was correlated negatively with lifetime stress and positively with working memory performance. Furthermore, it interacts with stress in modulating prefrontal activity during working memory. Specifically, lower methylation of the guanine allele and greater stress predict lower cortical efficiency. Moreover, COMT rs4680 guanine allele methylation was also inversely correlated with mRNA expression and protein levels, suggesting that
7 4 Part I : I maging G enetics and N eurochemistry
in vivo findings may be related to these molecular effects. These results were demonstrated by using COMT methylation in peripheral blood cells. However, there was a positive correlation between methylation of COMT in the prefrontal cortex and in PBMCs of rats, suggesting that peripheral COMT methylation is a good proxy for the study of brain phenotypes (Ursini et al. 2011). In summary, this study suggests that environmental stressors interact with structural DNA variation via methylation in modulating dopamine-related brain imaging and behavioral phenotypes associated with cognitive processing. Thus, it gives a data-based conceptual framework in which it is possible to delineate further lines of research investigating the relationship between genes and environment. Moreover, it calls for further investigation into the relevance of such mechanisms for brain disorders in terms of pathophysiology and therapeutic approaches. A F T E R WO RD After less than 15 years of imaging genetics research, it is becoming increasingly clear that the physiology of the brain in terms of imaging phenotypes cannot be completely understood if the effects of functional genetic variation are not taken into account. This is also true for the specific case of dopamine and related complex correlates, for which well-known models that link dopaminergic signaling and brain phenotypes have been posited. These models, which are largely centered on an inverted U relationship between dopamine stimulation and prefronto-striatal function, cannot entirely explain the results of several studies if the functional effects of dopamine-related genetic variation are not considered and modeled (Bertolino et al. 2009). As highlighted in this chapter, the complexity and the abundance of factors implicated in dopamine signaling make it difficult to put the pieces of the puzzle together in order to generate a general picture of the relationship between genetic variation and brain imaging phenotypes. However, knowledge about the disparate dopamine-related signaling pathways may help to focus the efforts on discovering genetically mediated epistatic interactions with an increasing level of complexity and acting on the physiology of the brain. Obviously, this complex research needs very large samples and the investigation of different phenotypes at different biological distances from gene effects. Without these options, it would be hard or impossible to begin to understand these complex mechanisms, as well as to be confident with the interpretation of results of imaging genetic studies. Importantly, the relationship between structural and
non-structural variation of DNA also driven by environmental factors increases the level of this complexity. A more accurate knowledge of the link between genetic variation in dopamine signaling and brain physiology and pathophysiology should take these aspects into account.
REFERENCES Alexander GE, DeLong MR, Strick PL. (1986). Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu Rev Neurosci. 9: 357–381. Arinami T, Gao M, Hamaguchi H, Toru M. (1997). A functional polymorphism in the promoter region of the dopamine D2 receptor gene is associated with schizophrenia. Hum Mol Genet. 6(4): 577–582. Beaulieu JM, Gainetdinov RR, Caron MG. (2007). The Akt-GSK-3 signaling cascade in the actions of dopamine. Trends Pharmacol Sci. 28(4): 166–172. Bertolino A, Blasi G. (2009). The genetics of schizophrenia. Neuroscience. 2009. Bertolino A, Blasi G, Latorre V, Rubino V, Rampino A, Sinibaldi L, et al. (2006). Additive effects of genetic variation in dopamine regulating genes on working memory cortical activity in human brain. J Neurosci. 26(15): 3918–3922. Bertolino A, Fazio L, Caforio G, Blasi G, Rampino A, Romano R, et al. (2009). Functional variants of the dopamine receptor D2 gene modulate prefronto-striatal phenotypes in schizophrenia. Brain. 132(Pt 2): 417–425. Bertolino A, Fazio L, Di Giorgio A, Blasi G, Romano R, Taurisano P, et al. (2009). Genetically determined interaction between the dopamine transporter and the D2 receptor on prefronto-striatal activity and volume in humans. J Neurosci. 29(4): 1224–1234. Bertolino A, Taurisano P, Pisciotta NM, Blasi G, Fazio L, Romano R, et al. (2010). Genetically determined measures of striatal D2 signaling predict prefrontal activity during working memory performance. PLoS One. 5(2): e9348. Blasi G, Lo Bianco L, Taurisano P, Gelao B, Romano R, Fazio L, et al. (2009). Functional variation of the dopamine D2 receptor gene is associated with emotional control as well as brain activity and connectivity during emotion processing in humans. J Neurosci. 29(47): 14812–14819. Blasi G, Mattay VS, Bertolino A, Elvevag B, Callicott JH, Das S, et al. (2005). Effect of catechol-O-methyltransferase val158met genotype on attentional control. J Neurosci. 25(20): 5038–5045. Blasi G, Napolitano F, Ursini G, Di Giorgio A, Caforio G, Taurisano P, et al. (2013). Association of GSK-3beta genetic variation with GSK3beta expression, prefrontal cortical thickness, prefrontal physiology, and schizophrenia. Am J Psychiatry. 170(8): 868–876. Blasi G, Napolitano F, Ursini G, Taurisano P, Romano R, Caforio G, et al. (2011). DRD2/AKT1 interaction on D2 c-AMP independent signaling, attentional processing, and response to olanzapine treatment in schizophrenia. Proc Natl Acad Sci U S A. 108(3): 1158– 1163. Blasi G, Taurisano P, Papazacharias A, Caforio G, Romano R, Lobianco L, et al. (2010). Nonlinear response of the anterior cingulate and prefrontal cortex in schizophrenia as a function of variable attentional control. Cereb Cortex. 20(4): 837–845. Bolan EA, Kivell B, Jaligam V, Oz M, Jayanthi LD, Han Y, et al. (2007). D2 receptors regulate dopamine transporter function via an extracellular signal-regulated kinases 1 and 2-dependent and phosphoinositide 3 kinase-independent mechanism. Mol Pharmacol. 71(5): 1222–1232. Cerasa A, Gioia MC, Tarantino P, Labate A, Arabia G, Annesi G, et al. (2009). The DRD2 TaqIA polymorphism associated with changed
cha p ter 5 : I maging G enetics of D o p amine S yna p tic T erminal A ctivity
75
midbrain volumes in healthy individuals. Genes Brain Behav. 8(4): 459–463. Craddock N, Sklar P. (2013). Genetics of bipolar disorder. Lancet. 381(9878): 1654–1662. Cropley VL, Fujita M, Innis RB, Nathan PJ. (2006). Molecular imaging of the dopaminergic system and its association with human cognitive function. Biol Psychiatry. 59(10): 898–907. Duan J, Wainwright MS, Comeron JM, Saitou N, Sanders AR, Gelernter J, et al. (2003). Synonymous mutations in the human dopamine receptor D2 (DRD2) affect mRNA stability and synthesis of the receptor. Hum Mol Genet. 12(3): 205–216. Durstewitz D, Seamans JK. (2008). The dual-state theory of prefrontal cortex dopamine function with relevance to catechol-omethyltransferase genotypes and schizophrenia. Biol Psychiatry. 64(9): 739–749 Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, et al. (2001). Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci U S A. 98(12): 6917–6922. Emamian ES, Hall D, Birnbaum MJ, Karayiorgou M, Gogos JA. (2004). Convergent evidence for impaired AKT1-GSK3beta signaling in schizophrenia. Nat Genet. 36(2): 131–137. Enoch MA, Xu K, Ferro E, Harris CR, Goldman D. (2003). Genetic origins of anxiety in women: a role for a functional catechol-Omethyltransferase polymorphism. Psychiatr Genet. 13(1): 33–41 Fazio L, Blasi G, Taurisano P, Papazacharias A, Romano R, Gelao B, et al. (2011). D2 receptor genotype and striatal dopamine signaling predict motor cortical activity and behavior in humans. Neuroimage. 54(4): 2915–2921. Forbes EE, Brown SM, Kimak M, Ferrell RE, Manuck SB, Hariri AR. (2009). Genetic variation in components of dopamine neurotransmission impacts ventral striatal reactivity associated with impulsivity. Mol Psychiatry. 14(1): 60–70. Fossella J, Sommer T, Fan J, Wu Y, Swanson JM, Pfaff DW, et al. (2002). Assessing the molecular genetics of attention networks. BMC Neuroscience. 3: 14. Fraga MF, Ballestar E, Paz MF, Ropero S, Setien F, Ballestar ML, et al. (2005). Epigenetic differences arise during the lifetime of monozygotic twins. Proc Natl Acad Sci U S A. 102(30): 10604–10609. Frank MJ, Moustafa AA, Haughey HM, Curran T, Hutchison KE. (2007). Genetic triple dissociation reveals multiple roles for dopamine in reinforcement learning. Proc Natl Acad Sci U S A. 104(41): 16311–16316. Freyberg Z, Ferrando SJ, Javitch JA. (2010). Roles of the Akt/GSK-3 and Wnt signaling pathways in schizophrenia and antipsychotic drug action. Am J Psychiatry. 167(4): 388–396. Gendreau PL, Petitto JM, Gariepy JL, Lewis MH. (1998). D2-like dopamine receptor mediation of social-emotional reactivity in a mouse model of anxiety: strain and experience effects. Neuropsychopharmacology. 18(3): 210–221. Gogos JA, Morgan M, Luine V, Santha M, Ogawa S, Pfaff D, et al. (1998). Catechol-O-methyltransferase-deficient mice exhibit sexually dimorphic changes in catecholamine levels and behavior. Proc Natl Acad Sci U S A. 95(17): 9991–9996. Goldman-Rakic PS. (1995). Cellular basis of working memory. Neuron. 14(3): 477–485. Goldman-Rakic PS. (1999). The “psychic” neuron of the cerebral cortex. Ann N Y Acad Sci. 868: 13–26. Gottesman, II, Gould TD. (2003). The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry. 160(4): 636–645. Greba Q, Gifkins A, Kokkinidis L. (2001). Inhibition of amygdaloid dopamine D2 receptors impairs emotional learning measured with fear-potentiated startle. Brain Res. 899(1–2): 218–226. Guarraci FA, Frohardt RJ, Falls WA, Kapp BS. (2000). The effects of intra-amygdaloid infusions of a D2 dopamine receptor antagonist on Pavlovian fear conditioning. Behav Neurosci. 114(3): 647–651.
Harris SL, Gil G, Robins H, Hu W, Hirshfield K, Bond E, et al. (2005). Detection of functional single-nucleotide polymorphisms that affect apoptosis. Proc Natl Acad Sci U S A. 102(45): 16297–16302. Harrison PJ, Weinberger DR. (2005). Schizophrenia genes, gene expression, and neuropathology: on the matter of their convergence. Mol Psychiatry. 10(1): 40–68; image 5. Heinz A, Goldman D, Jones DW, Palmour R, Hommer D, Gorey JG, et al. (2000). Genotype influences in vivo dopamine transporter availability in human striatum. Neuropsychopharmacology. 22(2): 133–139. Hirvonen M, Laakso A, Nagren K, Rinne JO, Pohjalainen T, Hietala J. (2004). C957T polymorphism of the dopamine D2 receptor (DRD2) gene affects striatal DRD2 availability in vivo. Mol Psychiatry. 9(12): 1060–1061. Hirvonen M, Laakso A, Nagren K, Rinne JO, Pohjalainen T, Hietala J. (2005). C957T polymorphism of the dopamine D2 receptor (DRD2) gene affects striatal DRD2 availability in vivo. Mol Psychiatry. 10(12): 889. Hirvonen MM, Lumme V, Hirvonen J, Pesonen U, Nagren K, Vahlberg T, et al. (2009). C957T polymorphism of the human dopamine D2 receptor gene predicts extrastriatal dopamine receptor availability in vivo. Prog Neuropsychopharmacol Biol Psychiatry. 33(4): 630–636. Hur EM, Zhou FQ. (2010). GSK3 signalling in neural development. Nat Rev Neurosci. 11(8): 539–551. Ikeda M, Iwata N, Suzuki T, Kitajima T, Yamanouchi Y, Kinoshita Y, et al. (2004). Association of AKT1 with schizophrenia confirmed in a Japanese population. Biol Psychiatry. 56(9): 698–700. Jirtle RL, Skinner MK. (2007). Environmental epigenomics and disease susceptibility. Nat Rev Genet. 8(4): 253–262. Jocham G, Klein TA, Neumann J, von Cramon DY, Reuter M, Ullsperger M. (2009). Dopamine DRD2 polymorphism alters reversal learning and associated neural activity. J Neurosci. 29(12): 3695– 3704. Jonsson EG, Nothen MM, Grunhage F, Farde L, Nakashima Y, Propping P, et al. (1999). Polymorphisms in the dopamine D2 receptor gene and their relationships to striatal dopamine receptor density of healthy volunteers. Mol Psychiatry. 4(3): 290–296. Kang UG, Seo MS, Roh MS, Kim Y, Yoon SC, Kim YS. (2004). The effects of clozapine on the GSK-3-mediated signaling pathway. FEBS Lett. 560(1–3): 115–119. Kapur S, Zipursky RB, Remington G. (1999). Clinical and theoretical implications of 5-HT2 and D2 receptor occupancy of clozapine, risperidone, and olanzapine in schizophrenia. Am J Psychiatry. 156(2): 286–293. Kellendonk C, Simpson EH, Polan HJ, Malleret G, Vronskaya S, Winiger V, et al. (2006). Transient and selective overexpression of dopamine D2 receptors in the striatum causes persistent abnormalities in prefrontal cortex functioning. Neuron. 49(4): 603–615. Kelly MA, Rubinstein M, Phillips TJ, Lessov CN, Burkhart-Kasch S, Zhang G, et al. (1998). Locomotor activity in D2 dopamine receptor-deficient mice is determined by gene dosage, genetic background, and developmental adaptations. J Neurosci. 18(9): 3470–3479. Klein TA, Neumann J, Reuter M, Hennig J, von Cramon DY, Ullsperger M. (2007). Genetically determined differences in learning from errors. Science. 318(5856): 1642–1645. Kirsch P, Reuter M, Mier D, Lonsdorf T, Stark R, Gallhofer B, et al. (2006). Imaging gene-substance interactions: the effect of the DRD2 TaqIA polymorphism and the dopamine agonist bromocriptine on the brain activation during the anticipation of reward. Neurosci Lett. 405(3): 196–201. Laruelle M. (1998). Imaging dopamine transmission in schizophrenia. A review and meta-analysis. Q J Nucl Med. 42(3): 211–221. Laruelle M, Gelernter J, Innis RB. (1998). D2 receptors binding potential is not affected by Taq1 polymorphism at the D2 receptor gene. Mol Psychiatry. 3(3): 261–265.
7 6 Part I : I maging G enetics and N eurochemistry
Lee BT, Lee HY, Han C, et al. (2011). DRD2/ANKK1 TaqI A polymorphism affects corticostriatal activity in response to negative affective facial stimuli. Behav Brain Res. 223(1): 36–41. Lee FJ, Pei L, Moszczynska A, et al. (2007). Dopamine transporter cell surface localization facilitated by a direct interaction with the dopamine D2 receptor. EMBO J. 26(8): 2127–2136. Li X, Rosborough KM, Friedman AB, et al. (2007). Regulation of mouse brain glycogen synthase kinase-3 by atypical antipsychotics. Int J Neuropsychopharmacol. 10(1): 7–19. Malhotra D, Sebat J. (2012). CNVs: harbingers of a rare variant revolution in psychiatric genetics. Cell. 148(6): 1223–1241. McGuffin P, Riley B, Plomin R. (2001). Genomics and behavior: toward behavioral genomics. Science. 291(5507): 1232–1249. McCrae RR, Costa PT, Jr. (1987). Validation of the five-factor model of personality across instruments and observers. J Pers Soc Psychol. 52(1): 81–90. Meyer A, Klein DN, Torpey DC, Kujawa AJ, Hayden EP, Sheikh HI, et al. (2012). Additive effects of the dopamine D2 receptor and dopamine transporter genes on the error-related negativity in young children. Genes Brain Behav. 11(6): 695–703. Mill J, Asherson P, Browes C, D’Souza U, Craig I. (2002). Expression of the dopamine transporter gene is regulated by the 3’ UTR VNTR: Evidence from brain and lymphocytes using quantitative RT-PCR. Am J Med Genet. 114(8): 975–979. Miller CA, Gavin CF, White JA, Parrish RR, Honasoge A, Yancey CR, et al. (2010). Cortical DNA methylation maintains remote memory. Nat Neurosci. 13(6): 664–666. Miller CA, Sweatt JD. (2007). Covalent modification of DNA regulates memory formation. Neuron. 53(6): 857–869. Montag C, Weber B, Jentgens E, Elger C, Reuter M. (2010). An epistasis effect of functional variants on the BDNF and DRD2 genes modulates gray matter volume of the anterior cingulate cortex in healthy humans. Neuropsychologia. 48(4): 1016–1021. Moore LD, Le T, Fan G. (2013). DNA methylation and its basic function. Neuropsychopharmacology. 38(1): 23–38. Mowry BJ, Gratten J. (2013). The emerging spectrum of allelic variation in schizophrenia: current evidence and strategies for the identification and functional characterization of common and rare variants. Mol Psychiatry. 18(1): 38–52. Murgatroyd C, Patchev AV, Wu Y, Micale V, Bockmuhl Y, Fischer D, et al. (2009). Dynamic DNA methylation programs persistent adverse effects of early-life stress. Nat Neurosci. 12(12): 1559– 1566. Noble EP, Blum K, Ritchie T, Montgomery A, Sheridan PJ. (1991). Allelic association of the D2 dopamine receptor gene with receptorbinding characteristics in alcoholism. Arch Gen Psychiatry. 48(7): 648–654. Noble EP, Gottschalk LA, Fallon JH, Ritchie TL, Wu JC. (1997). D2 dopamine receptor polymorphism and brain regional glucose metabolism. Am J Med Genet. 74(2): 162–166. Norton N, Williams HJ, Dwyer S, Carroll L, Peirce T, Moskvina V, et al. (2007). Association analysis of AKT1 and schizophrenia in a UK case control sample. Schizophr Res. 93(1–3): 58–65. Persico AM, Napolioni V. (2013). Autism genetics. Behav Brain Res. 251: 95–112. Pezze MA, Feldon J. (2004). Mesolimbic dopaminergic pathways in fear conditioning. Prog Neurobiol. 74(5): 301–320. Pidsley R, Mill J. (2011). Epigenetic studies of psychosis: current findings, methodological approaches, and implications for postmortem research. Biol Psychiatry. 69(2): 146–156. Pinsonneault JK, Han DD, Burdick KE, Kataki M, Bertolino A, Malhotra AK, et al. (2011). Dopamine transporter gene variant affecting expression in human brain is associated with bipolar disorder. Neuropsychopharmacology. 36(8): 1644–1655. Pohjalainen T, Rinne JO, Nagren K, Lehikoinen P, Anttila K, Syvalahti EK, et al. (1998). The A1 allele of the human D2 dopamine receptor
gene predicts low D2 receptor availability in healthy volunteers. Mol Psychiatry. 3(3): 256–260. Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. (2001). A default mode of brain function. Proc Natl Acad Sci U S A. 98(2): 676–682. Ripke S, O’Dushlaine C, Chambert K, Moran JL, Kahler AK, Akterin S, et al. (2013). Genome-wide association analysis identifies 13 new risk loci for schizophrenia. Nat Genet. 45(10): 1150–1159. Rosenkranz JA, Grace AA. (2001). Dopamine attenuates prefrontal cortical suppression of sensory inputs to the basolateral amygdala of rats. J Neurosci. 21(11): 4090–4103.Seamans JK, Yang CR. (2004). The principal features and mechanisms of dopamine modulation in the prefrontal cortex. Prog Neurobiol. 74(1): 1–58. Sambataro F, Fazio L, Taurisano P, Gelao B, Porcelli A, Mancini M, et al. (2013). DRD2 genotype-based variation of default mode network activity and of its relationship with striatal DAT binding. Schizophr Bull. 39(1): 206–216. Schwab SG, Hoefgen B, Hanses C, Hassenbach MB, Albus M, Lerer B, et al. (2005). Further evidence for association of variants in the AKT1 gene with schizophrenia in a sample of European sib-pair families. Biol Psychiatry. 58(6): 446–450. Smolka MN, Schumann G, Wrase J, Grusser SM, Flor H, Mann K, et al. (2005). Catechol-O-methyltransferase val158met genotype affects processing of emotional stimuli in the amygdala and prefrontal cortex. J Neurosci. 25(4): 836–842. Smoller JW. (2013). Disorders and borders: psychiatric genetics and nosology. Am J Med Genet B Neuropsychiatr Genet. 162(7): 559–578. Stein JL, Wiedholz LM, Bassett DS, Weinberger DR, Zink CF, Mattay VS, et al. (2007). A validated network of effective amygdala connectivity. Neuroimage. 36(3): 736–745. Stice E, Yokum S, Burger K, Epstein L, Smolen A. (2012). Multilocus genetic composite reflecting dopamine signaling capacity predicts reward circuitry responsivity. J Neurosci. 32(29): 10093–10100. Szyf M, McGowan P, Meaney MJ. (2008). The social environment and the epigenome. Environ Mol Mutagen. 49(1): 46–60. Tan HY, Chen AG, Kolachana B, Apud JA, Mattay VS, Callicott JH, et al. (2012). Effective connectivity of AKT1-mediated dopaminergic working memory networks and pharmacogenetics of anti-dopaminergic treatment. Brain. 135(Pt 5): 1436–1445. Tan HY, Nicodemus KK, Chen Q, Li Z, Brooke JK, Honea R, et al. (2008). Genetic variation in AKT1 is linked to dopamine-associated prefrontal cortical structure and function in humans. J Clin Invest. 118(6): 2200–2208. Thiselton DL, Vladimirov VI, Kuo PH, McClay J, Wormley B, Fanous A, et al. (2008). AKT1 is associated with schizophrenia across multiple symptom dimensions in the Irish study of high density schizophrenia families. Biol Psychiatry. 63(5): 449–457. Thompson J, Thomas N, Singleton A, Piggott M, Lloyd S, Perry EK, et al. (1997). D2 dopamine receptor gene (DRD2) Taq1 A polymorphism: reduced dopamine D2 receptor binding in the human striatum associated with the A1 allele. Pharmacogenetics. 7(6): 479–484. Tisch S, Silberstein P, Limousin-Dowsey P, Jahanshahi M. (2004). The basal ganglia: anatomy, physiology, and pharmacology. Psychiatr Clin North Am. 27(4): 757–799. Ursini G, Bollati V, Fazio L, Porcelli A, Iacovelli L, Catalani A, et al. (2011). Stress-related methylation of the catechol-O-methyltransferase Val 158 allele predicts human prefrontal cognition and activity. J Neurosci. 31(18): 6692–6698. Usiello A, Baik JH, Rouge-Pont F, Picetti R, Dierich A, LeMeur M, et al. (2000). Distinct functions of the two isoforms of dopamine D2 receptors. Nature. 408(6809): 199–203. Vandenbergh DJ, Persico AM, Hawkins AL, Griffin CA, Li X, Jabs EW, et al. (1992). Human dopamine transporter gene (DAT1)
cha p ter 5 : I maging G enetics of D o p amine S yna p tic T erminal A ctivity
77
maps to chromosome 5p15.3 and displays a VNTR. Genomics. 14(4): 1104–1106. VanNess SH, Owens MJ, Kilts CD. (2005). The variable number of tandem repeats element in DAT1 regulates in vitro dopamine transporter density. BMC Genetics. 6: 55. Wang M, Vijayraghavan S, Goldman-Rakic PS. (2004). Selective D2 receptor actions on the functional circuitry of working memory. Science. 303(5659): 853–856. Weickert TW, Goldberg TE, Gold JM, Bigelow LB, Egan MF, Weinberger DR. (2000). Cognitive impairments in patients with
schizophrenia displaying preserved and compromised intellect. Arch Gen Psychiatry. 57(9): 907–913. Wong CC, Caspi A, Williams B, Craig IW, Houts R, Ambler A, et al. (2010). A longitudinal study of epigenetic variation in twins. Epigenetics. 5(6): 516–526. Zhang Y, Bertolino A, Fazio L, Blasi G, Rampino A, Romano R, et al. (2007). Polymorphisms in human dopamine D2 receptor gene affect gene expression, splicing, and neuronal activity during working memory. Proc Natl Acad Sci U S A. 104(51): 20552–20557.
7 8 Part I : I maging G enetics and N eurochemistry
PA R T II. IMAGING GENE TICS AND DRUG DISCOVERY
6. VARIABILIT Y OF ANTIDEPRESSANT DRUG RESPONSE C ON T RIBU T ION OF IM AGING GE NE T IC S S T UDIE S
Ulrich Rabl, Bernhard M. Meyer, and Lukas Pezawas
I N T RO DU C T I O N The considerable disability and socioeconomic damage caused by major depressive disorder (MDD) calls for a better understanding of pathological mechanisms and treatment (Whiteford, Degenhardt, et al. 2013). MDD not only remains under-treated as well as under-recognized (Kupfer, Frank, et al. 2012), but available antidepressants are insufficient and are characterized by modest response rates on average, fueling vigorous debate about their potential benefits (Fountoulakis, Hoschl, et al. 2013). In fact, less than a third of MDD patients achieve complete remission with a single antidepressant, and a number of patients may even get worse under treatment compared to placebo (Trivedi, Rush, et al. 2006; Gueorguieva, Mallinckrodt, et al. 2011). Disregarding those patients who exhibit no or only partial drug response, a high relapse rate persists even in initially responding patients (McGrath, Stewart, et al. 2006). Another important concern is the delayed time of onset until remission is achieved, as well as frequent adverse drug reactions (ADRs) that accompany the use of available antidepressants (Taylor, Freemantle, et al. 2006). Despite the growing public attention toward MDD, drug discovery in psychiatry significantly lags behind other medical fields (Martinowich, Jimenez, et al. 2013). Beyond the search for more effective compounds, the development of biology-based diagnostic stratification of patients to available treatments is thought to have the potential to improve current treatment outcomes (Kapur, Phillips, et al. 2012). However, this endeavor is intrinsically linked to the biological understanding of the highly variable antidepressant drug response. Besides well-known variables affecting drug response, such as age, weight, comorbidity, gender, nutrition, and co-administered drugs, increasing
attention is being paid to the effects of genetic variation (Roden and George 2002). Considerable research has been devoted to this topic, jointly assessing genetic variants in candidate genes and response to current antidepressant treatments (Kato and Serretti 2010). However, while initial expectations in psychopharmacogenetics have been high, underpowered studies, small effects, and failed replications have been the rule rather than the exception (Tansey, Guipponi, et al. 2012). Despite known small effects of candidate genes putting their clinical use into question, some of the reported associations, such as for the candidate genes SLC6A4 and BDNF, have at least shown sufficient evidence in meta-analyses (Kato and Serretti 2010; Porcelli, Fabbri, et al. 2012; Niitsu, Fabbri, et al. 2013). Imaging genetics studies have demonstrated that candidate genes linked to variable antidepressant drug response also impact on brain function and structure (Scharinger, Rabl, et al. 2010). Another piece of evidence is provided by pharmacological magnetic resonance imaging (phMRI), which delineates the unique and common brain responses of antidepressant drugs, with distinct pharmacological profiles ranging from selective serotonin reuptake inhibitors (SSRIs) to the rapid antidepressant ketamine (Anderson, McKie, et al. 2008; De Simoni, Schwarz, et al. 2013). Together with studies assessing the structural alterations induced by depression and antidepressant drugs (Arnone, McKie, et al. 2012), these studies shed light on the orchestrated patterns of neural circuit responses during the course of antidepressant treatment. Taken together, this research suggests that genetic variants are implicated in antidepressant treatment response due to their effects on neural circuits that are involved in the brain’s response to antidepressant drug treatment. This chapter attempts to provide a comprehensive and
81
timely review of imaging genetics, pharmacogenetics, and phMRI studies as they relate to a mechanistic understanding of variable antidepressant drug response. N E U RO I M AG I N G M E T H O D S I N M DD R E S E A RC H Since we will cover studies assessing the effects of genetic variants as well as the impact of antidepressant therapy on brain structure and function measured with neuroimaging, this section gives an overview of the various imaging methods used in phMRI and imaging genetics in the context of depression. MRI has become a standard technique, both in the study of genetic effects and with regard to the investigation of antidepressant effects. Whereas positron emission tomography (PET) and single-photon emission computed tomography (SPECT) are mapping distinct neurotransmitter targets with tracer compounds, functional MRI (fMRI) provides less specific measures of drug or genetic effects beyond receptor occupancy or enzyme activity at a high spatial and temporal resolution without the need of injection of a contrast-agent or radioactive tracer (Wong, Tauscher, et al. 2009). The blood-oxygen-level dependent (BOLD) signal measured by fMRI is driven by local changes in the ratio of oxygenated to deoxygenated hemoglobin, which result from increased blood demand in response to neural activity. It is noteworthy, however, that phMRI might not be appropriate for some compounds with significant vasoactive properties. While this may complicate the interpretation of some drug effects, it is probably not a relevant issue for serotonergic drugs and subanaesthetic doses of ketamine that are used as antidepressants, which are generally not thought to have significant central vasoactive effects (Anderson, McKie, et al. 2008; De Simoni, Schwarz, et al. 2013). If it is unknown whether a compound is vasoactive, a breath-hold paradigm could be applied, because hypercapnia is known to significantly increase BOLD signaling via CO2-induced vasodilation (Carhart-Harris, Erritzoe, et al. 2012). While the majority of imaging genetic studies utilize cognitive or emotional tasks to drive activation in specific neural circuits, two investigative strategies can be distinguished with regard to phMRI, namely modulation phMRI and challenge phMRI (Anderson, McKie, et al. 2008). In modulation phMRI, the participants are asked to perform a task while in the scanner that is contrasted with a control condition (e.g., a less demanding or less emotionally challenging task), which allows identifying those brain structures that are specifically recruited by the task effect of interest.
To assess the impact of pharmacological treatment on the neural processes elicited by the task, participants or scan sessions can be randomized to treatment or a placebo to differentiate between effects of drug treatment and confounders such as learning or habituation effects in longitudinal assessments. Alternatively, the effects of the drug treatment can be compared to a baseline scan before treatment initiation or with scans of untreated subjects. However, due to ethical considerations, healthy subjects are frequently used as controls for treated patients, because otherwise an effective treatment would be withheld from symptomatic patients (Fu, Williams, et al. 2004; Lopez-Sola, Pujol, et al. 2010). In contrast to modulation phMRI, challenge phMRI does not necessarily require a task, but instead assesses BOLD signal changes as a consequence of pharmacological interventions (Anderson, McKie, et al. 2008). With respect to challenge phMRI, intravenous drug administration should be preferred to provide a rapid and consistent concentration increase of the studied drug during scanning. This is a limitation for drugs such as escitalopram, where only an oral formulation is licensed for medical use. There are several modeling approaches to identify voxels responsive to the drug treatment, and design considerations may affect the test-retest reliability (De Simoni, Schwarz, et al. 2013). Regressors can be based on known pharmacokinetic data or on the psychological response to the drug (Anderson, McKie, et al. 2008). More data-driven approaches may be advantageous, since they require fewer assumptions about the course of activation over time. Hereby, changes in the BOLD signal can be binned over time and compared to the baseline, or are analyzed via independent component analysis to unravel underlying patterns of activation (Beckmann and Smith 2004; Anderson, McKie, et al. 2008). Challenge phMRI is specifically useful as a translational tool from animals to humans since it models the acute effects of drugs without the requirement of a task (Martin and Sibson 2008). On the downside, challenge phMRI (1) requires intravenous drug administration, and (2) might be specifically susceptible to motion or drift artifacts due to long scan durations, and (3) is further complicated by high variability of the temporal profile of the response in between subjects or brain regions (De Simoni, Schwarz, et al. 2013). Finally, challenge phMRI is not applicable to study the long-term effects of drugs, which are specifically important in the case of classical antidepressants due to their delayed onset of action (Taylor, Freemantle, et al. 2006). In addition to activation studies, functional connectivity measures allow assessment of the communication between brain regions on a brain-systems level and the
8 2 Part I I : I maging G enetics and D rug D iscovery
effects of genetic variation or pharmacologic treatment on these networks. Functional connectivity provides a measure to assess temporal correlations in the fMRI signal across functionally related areas. Specifically, seed-based functional connectivity analysis explores which parts of the brain communicate with a given “seed” region by cross-correlating the time series of the seed with all other voxels. This can be either done with resting-state data or with task-based fMRI data. In the latter, task effects can be removed by regression or by restricting the correlation to the baseline condition (Fair, Schlaggar, et al. 2007). In contrast, psychophysiological interaction analysis (PPI), specifically uses task-based fMRI data by assessing which voxels in the brain change their correlative relationship with a seed region of interest in the context of the task or treatment (O’Reilly, Woolrich, et al. 2012). An alternative to seed-based approaches is independent component analysis, which partitions the BOLD signal fluctuations into a set of spatially independent components that can be tested for group effects (Beckmann and Smith 2004). Functional connectivity measures are increasingly adopted as intermediate phenotypes and have been found to be both modulated by antidepressant treatment as well as genetic variation (Anand, Li, et al. 2005; Pezawas, Meyer-Lindenberg, et al. 2005; Anand, Li, et al. 2007; Chen, Suckling, et al. 2008). Structural imaging techniques provide the possibility of investigating regional morphometric alterations driven by genetic effects or pharmacologic treatment. Manual tracing of regions of interest by researchers trained in neuroanatomy has long been the accepted standard, but full- or semi-automated techniques have gained popularity (Morey, Petty, et al. 2009). These techniques allow processing of large data sets with minimal user intervention and with increased reproducibility, as they are less susceptible to rater bias. Comparable to manual tracing, automated subcortical segmentation techniques assign neuroanatomical labels to each voxel based on voxel intensity, the intensity of neighboring voxels, and atlas-based prior probabilities (Fischl, Salat, et al. 2002). While manual tracing and automated segmentation techniques are limited to a priori regions of interest, voxel-based morphometry (VBM) is a straightforward method for voxel-wise comparisons of morphometric alterations across the whole brain. In VBM, high-resolution anatomical images of all subjects are spatially normalized into the same stereotactic space, followed by segmentation of the gray matter and smoothing to provide voxel-wise comparisons of the local concentration of gray matter (Ashburner and Friston 2000). In contrast to VBM, which does not account for the complex folding patterns of the
cortex, surface reconstruction algorithms provide geometrically accurate models of the gray/white and pial surface. These methods may be preferable over VBM for the study of genetic effects, since cortical thickness and surface area seem to be genetically independent (Winkler, Kochunov, et al. 2010). Beyond differences in regional anatomy, several methods also reflect aspects of neuronal “wiring” between regions. Structural or anatomical covariance is a group-level (across subjects) measure that assesses the correlation among regional structural measures. The anatomical measure that is correlated can either be derived from prior volume-based (gray matter density or deformation) or vertex-wise (thickness, area, or volume) analysis (Evans 2013). With diffusion tensor imaging (DTI), the integrity of white matter tracts within subjects can be investigated by exploiting differences in the directional mobility of water molecules (fractional anisotropy). There is evidence that structural covariance networks more closely resemble results obtained from functional connectivity than DTI tractography measures, reflecting that significant functional connectivity and structural covariance can be present in the absence of direct white matter connections (Evans 2013). N E U RO BI O L O G Y O F M A J O R DE P R E S S I V E DI S O R DE R In this section, we will review the current knowledge on the neurobiological alterations in the course of acute MDD, which are, at least partly, modulated by antidepressant treatment. MDD is defined in DSM-5 (APA 2013) as a distinct mood disorder requiring at least five symptoms of the following categories: (1) depressive mood, (2) lack of interest or pleasure, (3) weight or appetite alterations, (4) insomnia or hypersomnia, (5) psychomotor agitation or retardation, (6) fatigue or loss of energy, (7) feelings of worthlessness or guilt, (8) diminished ability to think or concentrate, or indecisiveness, and (9) recurrent thoughts of death, suicidal ideation, suicide plans or attempts. Additionally, it is required that at least one of the first two categories applies and lasts for a minimum duration of two weeks. The fact that MDD is solely based on behavioral criteria is a considerable hurdle in the development of valid neurobiological theories or animal models. For example, two subjects may fulfill the criteria for a major depressive episode according to the DSM without having even a single symptom in common (Nestler and Hyman 2010). Nevertheless, several remarkable observations could be made through the past years that have fostered our understanding of the processes
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
83
underlying depression. Stressful life events are known to trigger the onset of MDD, which may be modulated by genetic makeup (Kendler, Karkowski, et al. 1999; Caspi, Sugden, et al. 2003). Notably, even depressive episodes that are characterized as endogenous, which are traditionally thought to exhibit more pronounced neurobiological alterations, are at least partly driven by earlier environmental adversity, which may lead to changes in vulnerability pathways via the routes of epigenetics and neuroplasticity (Harkness and Monroe 2002; Dannlowski, Stuhrmann, et al. 2012). The crucial role of environmental adversity in depression is reflected in animal models, such as learned helplessness or chronic social defeat stress, which induce depressive behavior by confronting the animal with inescapable stressors (Nestler and Hyman 2010). While at least certain aspects of depression result from maladaptive responses to stress, the precise molecular machinery involved in these processes is still incompletely understood (Krishnan and Nestler 2008). The traditional “monoamine hypothesis of depression,” which posits that depression is caused by a decrease of monoamines in the brain, is suggested by the reoccurrence of depressive symptoms following acute tryptophan depletion and the efficiency of serotonergic drugs (Hirschfeld 2000). Although this hypothesis falls short of explaining several aspects such as the slow onset of action of classical antidepressants, several animal as well as human studies indeed suggest that alterations of the serotonergic system can be driven by uncontrollable stress. If a stressor is under the organism’s control, the medial prefrontal cortex inhibits the stress-induced activation of the dorsal raphe nucleus (DRN) (Amat, Baratta, et al. 2005). In turn, prolonged uncontrollable stress putatively results in exhaustion of the serotonergic system, leading to desensitization of 5-HT1A receptors in the DRN and serotonergic terminal regions (Jovanovic, Perski, et al. 2011; Rozeske, Evans et al. 2011). Likewise, reductions of 5-HT1A binding in acute MDD have been suggested by PET and postmortem studies (Savitz and Drevets 2013). While these findings suggest a principal role for the serotonergic system, the brain level alterations in MDD are widespread, including the anterior cingulate cortex (ACC), specifically in its subgenual portion (sACC), as well as the hippocampus and amygdala. Alterations of sACC metabolism, as well as reductions of sACC volume, possibly related to glial reductions, have been reported in MDD (Drevets, Price, et al. 1997; Ongur, Drevets, et al. 1998). The importance of the sACC has further been demonstrated by the efficacy of deep brain stimulation (DBS) of this area as a therapeutic option in refractory patients (Lozano, Mayberg, et al. 2008). The sACC is also
implicated in inhibitory top-down control of the amygdala, which exhibits increased responses to aversive stimuli in MDD (Hamilton, Etkin, et al. 2012). Reductions of ACC control over the amygdala have been found in acute MDD, reflected in decreased fractional anisotropy of the uncinate fasciculus (Johnstone, van Reekum, et al. 2007; Keedwell, Chapman, et al. 2012; de Kwaasteniet, Ruhe, et al. 2013). The ACC-amygdala circuit has further been found to be modulated by acute tryptophan depletion, 5-HT2A density, and genetic variation of the serotonin transporter, suggesting a strong regulatory influence of serotonergic signaling (Pezawas, Meyer-Lindenberg, et al. 2005; Fisher, Meltzer, et al. 2009; Passamonti, Crockett, et al. 2012). The ACC is also part of the default mode network (DMN), which is a network primarily active at rest, which reduces its activity during goal-directed behaviors (Gusnard, Akbudak, et al. 2001). Beyond the ACC, the DMN consists of the posterior cingulate cortex (PCC), the lateral inferior parietal lobes, and medial temporal structures including the hippocampi (Leech and Sharp 2013). DMN activity has been attributed to self-referential processing, internal thought processes, autobiographical memory retrieval, and mind wandering (Mason, Norton, et al. 2007; Berman, Peltier, et al. 2011; Hamilton, Furman, et al. 2011). DMN activity seems to increase during prolonged periods of stress (Soares, Sampaio, et al. 2013) and has been linked to ruminative thoughts in depressed subjects (Hamilton, Furman, et al. 2011; Bartova, Meyer, et al. 2015). Studies in depressed patients revealed a lack of suppression of DMN activity during goal-oriented tasks (Grimm, Boesiger, et al. 2009; Anticevic, Cole, et al. 2012). These alterations seem to be mostly driven by the ACC (Greicius, Flores, et al. 2007; Sheline, Barch, et al. 2009; Berman, Peltier, et al. 2011), while PCC connectivity may be reduced (Zhu, Wang, et al. 2012). Lacking suppression of the DMN, as is present in MDD, is associated with less efficient stimulus processing during goal-directed behaviors (Weissman, Roberts, et al. 2006). Therefore, these alterations may be related to observations of reduced working memory performance and rumination in depression, resulting in decreased efficiency of the dorsolateral prefrontal cortex (Matsuo, Glahn, et al. 2007; Bartova, Meyer, et al. 2015). Interestingly, DMN activity is explained to a considerable amount by 5-HT1A binding; therefore, altered DMN activity in depression may at least partly result from stress-induced desensitization of 5-HT1A autoreceptors (Rozeske, Evans, et al. 2011; Hahn, Wadsak, et al. 2012; Scharinger, Rabl, et al. 2014). The ACC is also connected to the hippocampus, another major target in depression research. Increased
8 4 Part I I : I maging G enetics and D rug D iscovery
functional connectivity, but decreased white matter integrity between the hippocampus and the ACC, has been found in MDD patients (de Kwaasteniet, Ruhe, et al. 2013). Granger causality analysis suggests that hippocampal activity predicts subsequent increases in the ACC in depression (Hamilton, Chen, et al. 2011). The hippocampus is specifically sensitive to stress, which results in an inhibition of hippocampal neurogenesis and plasticity due to downregulation of brain-derived neurotrophic factor (BDNF) (Sapolsky 2000; Tsankova, Berton, et al. 2006). Moreover, intact BDNF signaling in the hippocampus is obligatory for antidepressant behavioral effects (Wang, Dranovsky, et al. 2008). Hippocampal volume reduction has consistently been found in depression and may also constitute a vulnerability factor that mediates between environmental adversity and genetic risk (Rao, Chen, et al. 2010; Kempton, Salvador, et al. 2011). Notably, these alterations of hippocampal volume may not be detectable in younger patients with less severe symptoms, though hints of hippocampal memory dysfunction may still be present in this subgroup (Vythilingam, Vermetten, et al. 2004). Since the hippocampus is a major inhibitory regulator of the hypothalamus-pituitary-adrenal (HPA) axis, these alterations may also be linked to findings of increased cortisol awakening response, which constitutes a trait marker in depression (Dedovic, Duchesne, et al. 2009; Vreeburg, Hoogendijk, et al. 2009). Beyond these core regions of depression, meta-analytic evidence also suggests volume reductions in the basal ganglia, thalamus, hippocampus, frontal lobe, orbitofrontal cortex, striatum, and gyrus rectus (Kempton, Salvador, et al. 2011; Arnone, McIntosh, et al. 2012). Regarding alterations of brain function, increased activations in the thalamus, insula, and the dorsal anterior cingulate cortex, but reduced responses in the striatum and the dorsolateral prefrontal cortex, have further been reported (Hamilton, Etkin, et al. 2012).
R E S P O N S E T O A N T I DE P R E S S A N T T H E R A P Y I N T H E BR A I N CLASSICAL ANTIDEPRESSANTS
Modulation or challenge phMRI designs have been applied to delineate the brain sites of antidepressant action across the time course of antidepressant treatment, in healthy controls as well as depressed patients. Most of these studies have been focusing on SSRIs, which constitute the first-line treatment due to their beneficial side-effects profile (Bauer, Pfennig, et al. 2013). Interestingly, these studies highlight
regions that are also implicated in acute depression, with most evidence available for the amygdala and the ACC. Although the response to classical antidepressants is often described as having a delayed onset of effect that takes 2 to 3 weeks to become clinically evident, meta-analyses show that symptomatic improvement starts already by the end of the first week of use (Taylor, Freemantle, et al. 2006). On the other hand, acute SSRI therapy has been linked to elevated fear and suicide risk, suggesting an initial worsening of symptoms (Burghardt, Sullivan, et al. 2004; Browning, Reid, et al. 2007; Henry, Kisicki, et al. 2012). This potential increase in fear symptoms has been linked to findings of increased amygdala response in healthy subjects following acute treatment with citalopram or the norepinephrine reuptake inhibitor reboxetine (McKie, Del-Ben, et al. 2005; Bigos, Pollock, et al. 2008; Onur, Walter, et al. 2009). However, these observations are at odds with several studies that reported decreased amygdala responses following acute antidepressant therapy (Del-Ben, Deakin, et al. 2005; Takahashi, Yahata, et al. 2005; Anderson, Del-Ben, et al. 2007; Murphy, Norbury, et al. 2009; Rawlings, Norbury, et al. 2010; Anderson, Juhasz, et al. 2011). There is also preliminary evidence that acute treatment increases amygdala responses to pleasant stimuli, suggesting a neural basis for early beneficial effects (Taylor, Freemantle, et al. 2006; Rawlings, Norbury, et al. 2010; Tendolkar, van Wingen, et al. 2011). These partly contradictory findings for the amygdala are understandable given the low test-retest reliability of acute citalopram effects on the region that arises likely due to susceptibility artifacts (Klomp, van Wingen, et al. 2013). In contrast, the same study reported reproducible effects of acute citalopram treatment in the ACC, and the PCC in the direction of DMN deactivation decreases during a sensorimotor paradigm (Klomp, van Wingen, et al. 2013). While this observation is promising given the potential role of the DMN in depression, only a minority of studies reported effects of acute antidepressant therapy on ACC activation (McKie, Del-Ben, et al. 2005; Anderson, Juhasz, et al. 2011). Further, only limited evidence is available for acute effects in the insula (Anderson, Del-Ben, et al. 2007; Anderson, Juhasz, et al. 2011) and the orbitofrontal cortex (Del-Ben, Deakin, et al. 2005; McCabe, Mishor, et al. 2010). Studies reporting on long-term effects of antidepressant therapy are more conclusive and further show high congruence between healthy and patient samples. Sub-chronic or chronic treatment with citalopram, escitalopram, sertraline, fluoxetine, clomipramine, duloxetine, or bupropion has been associated with reduced amygdala activation, both in healthy subjects as well as in depressed
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
85
patients (Sheline, Barch, et al. 2001; Fu, Williams, et al. 2004; Harmer, Mackay, et al. 2006; Anand, Li, et al. 2007; Robertson, Wang, et al. 2007; Arce, Simmons, et al. 2008; de Almeida, Phillips, et al. 2010; Victor, Furey, et al. 2010; Windischberger, Lanzenberger, et al. 2010; van Marle, Tendolkar, et al. 2011; Arnone, McKie, et al. 2012; Tao, Calley, et al. 2012). In the same direction, activation reductions in the pregenual, subgenual, or dorsal ACC have been reported for citalopram, escitalopram, duloxetine, fluoxetine, and reboxetine treatment (Harmer, Mackay, et al. 2006; Norbury, Mackay, et al. 2008; Simmons, Arce, et al. 2009; de Almeida, Phillips, et al. 2010; Lopez-Sola, Pujol, et al. 2010; Tendolkar, van Wingen, et al. 2011; van Marle, Tendolkar, et al. 2011; Tao, Calley, et al. 2012), though conflicting results have been observed for the serotonin-norepinephrine reuptake inhibitor venlafaxine (Davidson, Irwin, et al. 2003). Likewise, chronic administration of fluoxetine, escitalopram, and bupropion resulted in diminished activity of the orbitofrontal cortex (Robertson, Wang, et al. 2007; Tao, Calley, et al. 2012; Macoveanu, Knorr, et al. 2013), while escitalopram or clomipramine treatment attenuated insula response (Arce, Simmons, et al. 2008; Simmons, Arce, et al. 2009; de Almeida, Phillips, et al. 2010). Further, effects in the hippocampus, parahippocampus, and the reward system have been reported (Fu, Williams, et al. 2004; McKie, Del-Ben, et al. 2005; Schaefer, Putnam, et al. 2006; Robertson, Wang, et al. 2007; Windischberger, Lanzenberger, et al. 2010; Anderson, Juhasz, et al. 2011; Stoy, Schlagenhauf, et al. 2012). Beyond these effects on regional activation patterns, several studies employing functional connectivity techniques suggest that antidepressants also impact neural network organization. These studies reveal an impact of antidepressants on the amygdala-ACC circuitry, resulting in increased coupling following chronic fluoxetine or sertraline treatment in depressed patients (Anand, Li, et al. 2005; Anand, Li, et al. 2007; Chen, Suckling, et al. 2008). Further, enhanced connectivity between the insula and the amygdala after 2 weeks of duloxetine treatment in healthy subjects has been reported (van Marle, Tendolkar, et al. 2011). Antidepressant treatment has also been associated with alterations of DMN connectivity. Acute as well as chronic antidepressant treatment has been found to decrease functional connectivity between key nodes of the DMN (Li, Liu, et al. 2013; Posner, Hellerstein, et al. 2013; van de Ven, Wingen, et al. 2013). While these results suggest that various classical antidepressants share common effects, with most evidence available for the amygdala-ACC circuitry, there are first
studies that directly compare compounds with different modes of action. To obtain a brain level correlate of the putative superior clinical efficiency of escitalopram, two studies compared citalopram and escitalopram given in a sub-chronic scheme in healthy volunteers (Windischberger, Lanzenberger, et al. 2010; Henry, Lauriat, et al. 2013). Despite comparable study designs, there was no overlap in the brain effects of escitalopram versus citalopram between both studies (Windischberger, Lanzenberger, et al. 2010; Henry, Lauriat, et al. 2013). Further, the noradrenergic and specific serotonergic antidepressant mirtazapine and the serotonin-norepinephrine reuptake inhibitor venlafaxine, as well as the noradrenaline reuptake inhibitor reboxetine and the SSRI citalopram, have been compared in single studies, but replications are needed to draw conclusions (McCabe, Mishor, et al. 2010; Frodl, Scheuerecker, et al. 2011). Given that the response to first-class antidepressants is delayed, several researchers attempted to use imaging of intermediate phenotypes as early predictors of antidepressant drug response (Fu, Steiner, et al. 2013). Though neuroimaging may be too expensive and sophisticated to be clinically applicable, the characterization of brain imaging of intermediate phenotypes may pave the way for more affordable biomarkers, such as blood tests (Holsboer 2008). A recent meta-analysis of PET and fMRI studies found that increased baseline activity in the pregenual ACC is predictive of a higher likelihood of improvement, while increased baseline activation in the insula and striatum has been associated with poor clinical outcome. Interestingly, there may be regionally distinct effects within the ACC, with a notable number of studies reporting greater baseline activation in the sACC to predict poorer response (Fu, Steiner, et al. 2013). Structural imaging studies demonstrate that antidepressants not only impact on brain activation and connectivity, but also modulate brain anatomy, at least to the extent that structural MRI can be interpreted to reflect structural neuroanatomy. With this regard, antidepressant therapy seems to counteract changes in brain structure related to acute MDD (Sheline, Gado, et al. 2003; Arnone, McKie, et al. 2012). While reductions of hippocampal volume have been found in acutely depressed subjects as well as in subjects at risk for depression (Rao, Chen, et al. 2010; Kempton, Salvador, et al. 2011), hippocampal volume seems to increase during antidepressant therapy (Kempton, Salvador, et al. 2011; Arnone, McKie, et al. 2012). This is in line with evidence from animal studies and postmortem studies in humans that suggests a stimulation of hippocampal neurogenesis by antidepressants that is required for the
8 6 Part I I : I maging G enetics and D rug D iscovery
behavioral effects of antidepressant treatment (Santarelli, Saxe, et al. 2003; Boldrini, Underwood, et al. 2009; Anacker, Zunszain, et al. 2011; Boldrini, Hen, et al. 2012). Interestingly, this stimulating effect on hippocampal volume may not be limited to antidepressants, but has also been reported for electroconvulsive therapy in depressed patients and cognitive behavioral therapy in patients with posttraumatic stress disorder (Nordanskog, Dahlstrand, et al. 2010; Levy-Gigi, Szabo, et al. 2013). Finally, reduced pretreatment hippocampal volume has been found to be a predictor of a lower likelihood of clinical response (Fu, Steiner, et al. 2013). Similar to volumetric alterations of the hippocampus, antidepressant medication may also lead to increases of amygdala volume, while decreases of amygdala volume have been found in unmedicated patients (Hamilton, Siemer, et al. 2008). KE TAMINE
Ketamine, an N-methyl-d-aspartate (NMDA) receptor antagonist, has been proven to be a fast acting antidepressant with a peak success rate after 24 hours between 25% and 85%, even in treatment-refractory patients. While these numbers are impressive, the long-term success of ketamine treatment is limited, reducing its usability to research and the treatment of acute refractory patients in the hospital setting (Aan Het Rot, Zarate, et al. 2012). Challenge phMRI suggests that ketamine results in a rather unspecific increase in various brain regions, including the cingulate cortex, temporal lobes, and hippocampus, but a focal decrease in the sACC (Deakin, Lees, et al. 2008; De Simoni, Schwarz, et al. 2013). Although not highlighted by these ketamine challenge phMRI studies, the pregenual ACC may be specifically important for the antidepressant response to ketamine. Measured with magnetoencephalography, reduced pretreatment activation in the pregenual ACC during working memory processing, but decreased response to fearful faces, predicted beneficial treatment responses to ketamine (Salvadore, Cornwell, et al. 2009; Salvadore, Cornwell, et al. 2010). According to magnetic resonance spectroscopy (MRS) studies, ketamine further leads to an immediate increase of glutamine in the pregenual ACC, suggesting enhanced glutamate neurotransmitter release, probably counteracting deficits of glutamatergic metabolism in depression (Rowland, Bustillo, et al. 2005; Walter, Henning, et al. 2009; Stone, Dietrich, et al. 2012). Beyond these regional alterations that are putatively beneficial in depression, acute ketamine treatment has also been linked to neural patterns related to schizophrenia-like positive and negative symptoms, such as increased prefrontal
activation and reduced DMN deactivation in cognitive tasks (Honey, Honey, et al. 2004; Honey, Honey, et al. 2005; Honey, Corlett, et al. 2008; Anticevic, Gancsos, et al. 2012). Ketamine has also been shown to modulate brain connectivity. Ketamine reduced the coupling between anterior and posterior midline compartments of the DMN, which may be elevated in depression (Sheline, Price, et al. 2010; Scheidegger, Walter, et al. 2012; Li, Liu, et al. 2013). Interestingly, a similar reduction of connectivity between ACC and PCC has been reported for the classic psychedelic psilocybin, which also has antidepressant effects (Carhart-Harris, Erritzoe, et al. 2012). These recent observations contribute to the regained interest in the potential implications of psychedelics for the understanding and treatment of mood disorders (Vollenweider and Kometer 2010). I M AG I N G G E N E T I C S I N T H E C O N T E X T O F VA R I A BL E A N T I DE P R E S S A N T DRU G R E S P O N S E Genetic variants implicated in the response to antidepressants can be either involved in pharmacokinetics, which refers to genes involved in processes such as absorption, distribution, metabolism, and elimination that influence the delivery of a drug to its target, or in pharmacodynamics, which relates to genes directly involved in antidepressant drug effects (Schosser and Kasper 2009). Although evidence is accumulating that some genetic variants implicated in the pharmacokinetics of antidepressant drugs may also be involved in the neurobiology of depression, most research with regard to brain function and structure has been focused on pharmacodynamic variants (Kirchheiner, Seeringer, et al. 2011; Persson, Sim, et al. 2013). Two of those candidate genes implicated in antidepressant pharmacodynamics, SLC6A4 and BDNF, have been consistently linked to variable drug response according to recent meta-analyses and intensively investigated with imaging genetics (Kato and Serretti 2010; Scharinger, Rabl, et al. 2010; Porcelli, Fabbri, et al. 2012; Niitsu, Fabbri, et al. 2013). SEROTONIN TRANSPORTER GENE
The monoamine hypothesis, which proposes that depression is caused by a deficiency in noradrenergic, dopaminergic, and predominantly serotonergic neurotransmission, led to the development of SSRIs, which constitute the first-line treatment for MDD (Krishnan and Nestler 2008; Bauer, Pfennig, et al. 2013). Serotonin transporters, which are
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
87
inhibited by SSRIs, reduce 5-HT levels in the synaptic cleft by re-uptake and are expressed by the serotonin transporter gene (SLC6A4). Since its identification, a functional variable number of tandem repeats (VNTR) polymorphism (5-HTTLPR) in the promoter region of SLC6A4 has been used by a steadily increasing number of researchers to investigate the downstream effects of variable transcriptional efficiency in SLC6A4 (Heils, Teufel, et al. 1996). Additionally, a single nucleotide polymorphism (rs25531) within the promoter region is often assessed together with the 5-HTTLPR, although its effect on transcription levels has not been consistently demonstrated (Martin, Cleak, et al. 2007). The short (S) allele of 5-HTTLPR results in decreased transporter expression, but seems to have only a small effect on 5-HTT binding in the adult brain, since PET studies are rather equivocal (Shioe, Ichimiya, et al. 2003; Van Dyck, Malison, et al. 2004; Parsey, Hastings, et al. 2006; Praschak-Rieder, Kennedy, et al. 2007; Reimold, Smolka, et al. 2007; Kalbitzer, Frokjaer, et al. 2009; Murthy, Selvaraj, et al. 2010; Ho, Ho, et al. 2013). Several variables such as smoking status, gender, or seasonal changes in daylight minutes are likely contributing to these inconsistent results (Kalbitzer, Erritzoe, et al. 2010; Kobiella, Reimold, et al. 2011). However, while this argues for a rather small effect of this variant on serotonergic neurotransmission, there may still be substantial impact of 5-HTTLPR on brain function and anatomy during critical periods of early neurodevelopment, where serotonin is involved in processes such as neural cell proliferation, migration, and differentiation (Gaspar, Cases, et al. 2003). Indeed, animal studies suggest a paradoxical effect, where genetically (5-HTTLPR) or pharmaceutically (SSRIs) evoked reductions in 5-HT reuptake during development lead to an increase of depressive symptoms in adult life, whereas treatment with SSRIs in the developed brain leads to an anxiolytic and antidepressant effect (Ansorge, Zhou, et al. 2004; Sibille and Lewis 2006; Homberg, Schubert, et al. 2010). On a behavioral level, 5-HTTLPR has been associated with anxiety-related personality traits and risk for depression, especially in interaction with environmental adversity (Lesch, Bengel, et al. 1996; Caspi, Sugden, et al. 2003). However, recent meta-analyses of these efforts are controversial and depend on the study inclusion criteria, suggesting profound heterogeneity between studies (Risch, Herrell, et al. 2009; Karg, Burmeister, et al. 2011). When stress is less broadly defined, 5-HTTLPR seems to be a significant moderator of the relationship between stress and depression risk, especially in the presence of childhood maltreatment and chronic stress, while this association is
more ambiguous with regard to stressful life events, especially when assessed with self-report questionnaires (Karg, Burmeister, et al. 2011). Given that serotonin transporters constitute the first-line drug target in antidepressant treatment, it is not surprising that 5-HTTLPR emerged as a major candidate variant for pharmacogenetic research. Although a remarkable series of studies and several meta-analyses have been devoted to the potential impact of 5-HTTLPR on antidepressant drug response, final conclusions are difficult to draw. While an earlier meta-analysis concluded that the 5-HTTLPR does not predict antidepressant response to a clinically useful degree, a more recent meta-analysis including a larger number of studies found the S allele to be linked to a less favorable antidepressant response in Europeans, while in East Asians it does not appear to play a major role (Taylor, Sen, et al. 2010; Porcelli, Fabbri et al. 2012). Further, the L allele of 5-HTTLPR has been associated with a reduced risk of experiencing side effects (Kato and Serretti 2010), though an effect of 5-HTTLPR on discontinuation from antidepressant treatment could not consistently be demonstrated (Crawford, Lewis, et al. 2013). While these inconsistent results suggest that the effects of 5-HTTLPR on behavioral phenotypes are small, imaging genetics studies suggest profound impact of this variant on the neural level. Amygdala hyper-reactivity to emotional stimuli in healthy S allele carriers is the most prominently investigated intermediate phenotype for 5-HTTLPR with numerous replications (Hariri, Mattay, et al. 2002; Bertolino, Arciero, et al. 2005; Hariri, Drabant, et al. 2005; Heinz, Braus, et al. 2005; Pezawas, Meyer-Lindenberg, et al. 2005; Canli, Qiu, et al. 2006; Smolka, Bühler, et al. 2007; Dannlowski, Ohrmann, et al. 2008; Williams, Gatt, et al. 2009; Dannlowski, Konrad, et al. 2010; Kobiella, Reimold, et al. 2011; Lonsdorf, Golkar, et al. 2011; Costafreda, McCann, et al. 2013). A recent meta-analysis found this effect to be statistically significant, though the true effect size may be much smaller than initially anticipated (Hariri, Mattay, et al. 2002; Murphy, Norbury, et al. 2013). The same direction of increased amygdala activity in S carriers was further observed in MDD patients (Dannlowski, Ohrmann, et al. 2007; Dannlowski, Ohrmann, et al. 2008; Friedel, Schlagenhauf, et al. 2009; Brockmann, Zobel, et al. 2011; Costafreda, McCann, et al. 2013). It has been suggested that these differences in amygdala function are driven by decreased activation to neutral stimuli, rather than increased activation to negative stimuli in carriers of the S allele (Canli, Omura, et al. 2005). Others demonstrated increased amygdala responses in the S allele not
8 8 Part I I : I maging G enetics and D rug D iscovery
only to clearly aversive stimuli but also to undefined and ambiguous cues (Heinz, Smolka, et al. 2007). The S allele has also been found to increase resting cerebral blood flow in the amygdala, though this was not replicated in a larger cohort (Rao, Gillihan, et al. 2007; Viviani, Sim, et al. 2010). In contrast to neutral and negative stimuli, 5-HTTLPR seems to have no effect on the amygdala response to happy faces (Dannlowski, Konrad, et al. 2010). Further, an interaction effect between life stress or childhood adversity and 5-HTTLPR on amygdala response was found (Canli, Qiu, et al. 2006; Lemogne, Gorwood, et al. 2011; Walsh, Dalgleish, et al. 2012). The 5-HTTLPR has also been investigated with regard to neural network organization. Using a variety of imaging modalities such as functional connectivity, structural covariance, and diffusion tensor imaging, a reduction in amygdala-cingulate interactions could be related to the S allele (Pezawas, Meyer-Lindenberg, et al. 2005; Pacheco, Beevers, et al. 2009). Since these results suggest a decrease in cingulate inhibitory efferents to the amygdala in S allele carriers, they provide a mechanistic explanation for increased amygdala reactivity in this group. PPI analysis has further been used to probe this network, indicating that the effect of 5-HTTLPR on amygdala-prefrontal interplay mediates inter-individual differences in decision-making (Roiser, De Martino, et al. 2009). Further, it has been found that this network can be modulated by volitional cognitive regulation, diminishing the effect of 5-HTTLPR (Schardt, Erk, et al. 2010; Lemogne, Gorwood, et al. 2011). The importance of the prefrontal-amygdala network is further highlighted by reports that the integrity of this circuit predicts a large proportion of variance in harm avoidance or trait anxiety (Pezawas, Meyer-Lindenberg, et al. 2005; Kim and Whalen 2009; Holmes, Lee, et al. 2012). Several studies also suggest an effect of 5-HTTLPR on brain anatomy, supporting the assumption that the effect of 5-HTTLPR takes place in early neurodevelopment. At least partly, these effects on brain anatomy may be the driving force behind the functional effects of 5-HTTLPR. In this line, a recent study demonstrated that the stronger amygdala reactivity in S carriers is mediated by amygdala structure, but not by acute 5-HTT binding levels as measured with [11C]DASB PET (Kobiella, Reimold, et al. 2011). Several studies confirm that amygdala volumes are decreased in S carriers (Pezawas, Meyer-Lindenberg, et al. 2005; Frodl, Koutsouleris, et al. 2008; Kobiella, Reimold, et al. 2011), though two studies reported the opposite (Scherk, Gruber, et al. 2009) or no effect at all (Stjepanovic, Lorenzetti, et al. 2013). Reflecting the intricate relationship between the amygdala and the ACC, decreased gray
matter volume of the ACC in S carriers has also been found (Pezawas, Meyer-Lindenberg, et al. 2005; Frodl, Koutsouleris, et al. 2008). Further support for the impact of this variant on ACC volume stems from a study in rhesus macaques that found similar alterations driven by the rhesus macaque orthologue of 5-HTTLPR (Jedema, Gianaros, et al. 2010). Results linking the S allele of 5-HTTLPR to smaller hippocampal volumes are more equivocal and may be further complicated by interactions with additional variables such as childhood adversity or gender (Frodl, Koutsouleris, et al. 2008; Cole, Weinberger, et al. 2011; Everaerd, Gerritsen, et al. 2012; Price, Strong, et al. 2013; Rabl, Meyer, et al. 2014). Finally, epistasis, which involves interaction effects between genetic variants in two or more genes, has been investigated with regard to 5-HTTLPR. Epistasis between 5-HTTLPR and the Val66Met polymorphism in the BDNF gene has been observed in relation to sACC volume and structural covariance between amygdala and sACC (Pezawas, Meyer-Lindenberg, et al. 2008). With regard to brain function, epistasis between a VNTR in MAOA and 5-HTTLPR affecting ACC activity has been reported (Passamonti, Cerasa, et al. 2008). Further, additive effects of 5-HTTLPR and the Val158Met polymorphism in COMT on amygdala reactivity have been found (Smolka, Bühler, et al. 2007). BRAIN-DERIVED NEUROTROPHIC FACTOR GENE
In the late 1990s, the neurotrophin hypothesis of depression emerged, which posited that depression is linked to reduced synaptic plasticity and neurogenesis, processes that are controlled by the most prevalent growth factor in the brain, BDNF (Duman, Heninger, et al. 1997; Autry and Monteggia 2012). In agreement with this hypothesis, there is strong evidence that serum BDNF levels reflect the current state of disease, exhibiting lower levels during depression that normalize after successful antidepressant treatment (Molendijk, Spinhoven, et al. 2013). BDNF expression in the hippocampus seems to follow the same course of reduced levels in the acute state and normalization during remission (Duman and Aghajanian 2012). Although reduced BDNF expression in knockout mice does not lead to depressive symptoms, it induces the inability to respond to antidepressant treatment. In turn, infusion of BDNF produces antidepressant effects. Opposite effects have been observed in the mesolimbic dopaminergic reward system, where increased BDNF levels result in depressive symptoms and increased stress susceptibility (Autry and Monteggia 2012; Duman and Aghajanian 2012).
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
89
Among the multiple polymorphic sites in the BDNF gene, a non-synonymous single nucleotide polymorphism (Val66Met, rs6265) has been intensively investigated, since it could be related to impaired intracellular trafficking and activity-dependent secretion of BDNF in Met carriers (Egan, Kojima, et al. 2003; Licinio, Dong, et al. 2009). Meta-analyses reveal no consistent association of Val66Met with MDD, though stratified analyses suggest an increased risk for male and geriatric carriers of the Met allele (Verhagen, Van Der Meij, et al. 2010; Pei, Smith, et al. 2012; Gyekis, Yu, et al. 2013). Similar to 5-HTTLPR, the effect of Val66Met may be modulated by environmental adversity (Elzinga, Molendijk, et al. 2011). In line with the primary role of BDNF in hippocampal function, a small but consistent effect of Val66Met on declarative memory performance has been found, with favorable responses in Val homzygotes (Kambeitz, Bhattacharyya, et al. 2012). BDNF seems to be involved in various clinical and non-clinical interventions to promote well-being, ranging from antidepressants to behavioral interventions such as regular physical activity or psychotherapy (Pollak, Monje, et al. 2008; Autry and Monteggia 2012; Davidson and McEwen 2012). Chronic administration of SSRIs increase BDNF-dependent neuroplasticity, while increased glutamate transmission following ketamine treatment leads to a fast induction of BDNF-mediated synaptogenesis and reversal of stress-related atrophy (Autry and Monteggia 2012; Duman and Aghajanian 2012). Consistent with this, Val66Met has been associated with antidepressant response, being the best candidate of 15 investigated polymorphisms in a recent meta-analysis (Niitsu, Fabbri, et al. 2013). Preliminary translational evidence in mice and humans suggests that Val66Met also determines the efficacy of ketamine treatment (Laje, Lally, et al. 2012; Liu, Lee, et al. 2012). Mirroring the important role of BDNF for hippocampal integrity, structural imaging genetics studies found an association of Val66Met with the hippocampus in terms of decreased volume in carriers of the Met allele (Pezawas, Verchinski, et al. 2004). This finding could be replicated by several groups and even in patient groups, though also negative reports exist. Nevertheless, meta-analyses support this effect, though effect sizes are smaller than originally anticipated (Hajek, Kopecek, et al. 2012; Kambeitz, Bhattacharyya, et al. 2012; Molendijk, Bus, et al. 2012). Meta-analyses also found significant heterogeneity in between studies, especially for patients (Kambeitz, Bhattacharyya, et al. 2012). This inhomogeneity could be caused by past or current medication, the influence of environmental adversity, age effects, and methodological
differences (Gatt, Nemeroff, et al. 2009; Cole, Weinberger, et al. 2011; Molendijk, Bus, et al. 2011; Rabl, Meyer, et al. 2014; Meyer, Huemer, et al. 2014). Volume reductions in Met carriers of Val66Met have also been shown in other brain regions that are relevant for MDD, such as the dorsolateral PFC (Pezawas, Verchinski, et al. 2004; Matsuo, Walss-Bass, et al. 2009), the parahippocampal gyrus (Montag, Weber, et al. 2009), and the ACC-amygdala circuit (Pezawas, Meyer-Lindenberg, et al. 2008; Sublette, Baca-Garcia, et al. 2008; Matsuo, Walss-Bass, et al. 2009; Montag, Weber, et al. 2009; Gerritsen, Tendolkar, et al. 2012). While a main effect of Val66Met on the ACC-amygdala circuit has not been consistently replicated, more complex models have been proposed, including epistatic effects of 5-HTTLPR and interactions with age and environmental adversity (Pezawas, Meyer-Lindenberg, et al. 2008; Sublette, Baca-Garcia, et al. 2008; Gatt, Nemeroff, et al. 2009; Gerritsen, Tendolkar, et al. 2012; Rabl, Meyer, et al. 2014). Complementing the effect on brain anatomy, functional imaging studies further demonstrate an effect of Val66Met on normal hippocampal functioning (Kambeitz, Bhattacharyya, et al. 2012). Meta-analytic evidence suggests increased hippocampal activation in Val homozygotes across a wide range of cognitive paradigms, including episodic memory, working memory, and decision-making, with no indication of publication bias (Egan, Kojima, et al. 2003; Hariri, Goldberg, et al. 2003; Kambeitz, Bhattacharyya, et al. 2012). This effect of Val66Met has been considerably larger than the effect on memory performance and hippocampal structure, though it may be inflated due to voxel selection bias (Kambeitz, Bhattacharyya, et al. 2012; Dodds, Henson, et al. 2013). Supporting the crucial role of BDNF function in memory consolidation, a recent multimodal longitudinal study linked sleep EEG slow-wave oscillations in the consolidation night and fMRI activation in medial PFC and left parahippocampal gyrus during retrieval at the next day to better memory performance in Val homozygotes compared to Met carriers (Mascetti, Foret, et al. 2013). Fear-processing paradigms are not consistent across studies, reporting increased responses of the amygdala, hippocampus, ACC, and insula in Met carriers (Montag, Reuter, et al. 2008; Lau, Goldman, et al. 2010; Mukherjee, Whalley, et al. 2011). Further, an impact of Val66Met on fear extinction learning has been found in mice and humans, which was in humans reflected in decreased activity of the ventromedial PFC, but increased activity of the amygdala in Met carriers (Soliman, Glatt, et al. 2010). These Val66Met-driven alterations of the fear extinction circuit have further been
9 0 Part I I : I maging G enetics and D rug D iscovery
linked to reduced fractional anisotropy in the uncinate fasciculus of Met carriers, which has also been found in MDD patients (Soliman, Glatt, et al. 2010; Carballedo, Amico, et al. 2012), but could not be replicated in an independent study (Tost, Alam, et al. 2013). The impact of Val66Met on white matter integrity may be even more widespread, though no consistent pattern emerged across studies (Chiang, Barysheva, et al. 2011; Tost, Alam, et al. 2013; Ziegler, Foret, et al. 2013). DI S C U S S I O N Despite enormous efforts by clinicians and researchers over the past years to track down the biological basis of variable antidepressant drug response, a transition from bench to bedside seems unlikely in the near future (Kapur, Phillips, et al. 2012). While the biological understanding of MDD and its treatment has significantly advanced, drug choice in the clinic is still based on a trial and error principle. Partly fueled by the completion of the Human Genome Project, much emphasis has been put on the stratification of patients based on genetic variation, but as of 2015 no genetic test is available for widespread clinical use (Tansey, Guipponi, et al. 2012). Even the most promising candidate genes such as SLC6A4 and BDNF only explain a negligible portion of the variance in treatment response (Kato and Serretti 2010; Porcelli, Fabbri, et al. 2012; Niitsu, Fabbri, et al. 2013). Nevertheless, it has been known since early twin and family studies that antidepressant drug response can be attributed to a considerable extent to genetic factors (Fabbri, Di Girolamo, et al. 2013). But why did the quest for the specific genetic markers turn out to be so burdensome? One reason may lie in the broad genetic architecture of antidepressant drug response that is only recently beginning to unfold due to the increasing resolution of genetic mapping techniques. Based on genome-wide association data, it has been estimated that common genetic polymorphisms, scattered across the genome, contribute approximately 42.0% of individual variation in antidepressant drug response (Tansey, Guipponi, et al. 2013). Although this figure seems remarkable, the practical predictive potential of these variants is much lower, since this heritability estimate reflects the upper bound of what can maximally be explained by these variants (Purcell 2013). In fact, a polygenic model in the largest genome-wide association study (GWAS) dataset available today had a remarkably low prediction value of 1.2% explained variance in treatment outcome (GENDEP, MARS, STAR*D Investigators, 2013). Notably, this lack of reliable findings is mirrored by studies investigating the
genetic contribution to MDD risk. While consistent results from GWAS are beginning to emerge for other common traits—including psychiatric disorders such as schizophrenia or bipolar disorder—MDD has been specifically difficult to track down. Recently, the most comprehensive GWAS for MDD was unable to identify any robust or replicable risk loci (Major Depressive Disorder Working Group of the Psychiatric, Ripke, et al. 2013). Partly, this may be explained by hidden genetic variation that is insufficiently covered by current genotyping arrays, such as rare variants of large effect or copy number variations. While these sources of genetic variation may ultimately be covered with advanced sequencing techniques, another potential area for improvement may not be related to genotyping, but phenotyping (Major Depressive Disorder Working Group of the Psychiatric, Ripke, et al. 2013). Common disorders such as depression are not distinct qualitative entities, but the extremes of quantitative traits (Plomin, Haworth, et al. 2009). Moreover, many specific patterns of symptoms gather under the umbrella of depression, although they most likely differ in their neurobiological signature (Nestler and Hyman 2010). The same problem applies to the phenotype of antidepressant drug response, which is measured using behavioral sum scores such as the Hamilton Depression Rating Scale. This scale has been criticized for a variety of psychometric flaws, including that the total score is of multidimensional and unclear meaning, unlikely related to a specific biological process (Bagby, Ryder, et al. 2004; Della Pasqua, Santen et al. 2010). Intermediate phenotypes that are reliably measurable and biologically meaningful may provide valid substitutes for these subjective behavioral phenotypes (Meyer-Lindenberg and Weinberger 2006). Clinical neuroimaging, phMRI, and imaging genetics studies show that the same neural systems are implicated in the neurobiology of MDD, affected by genetic variation and modulated by antidepressant treatment (Rabl, Scharinger, et al. 2010). The best studied intermediate phenotypes in MDD thus far, amygdala hyperactivity (Hamilton, Etkin, et al. 2012) and hippocampal volume reduction (Kempton, Salvador, et al. 2011), have both been found to be affected by various antidepressant treatments (Sheline, Barch, et al. 2001; Fu, Williams, et al. 2004; Anand, Li, et al. 2007; Robertson, Wang, et al. 2007; Victor, Furey, et al. 2010; Kempton, Salvador, et al. 2011; Arnone, McKie, et al. 2012; Tao, Calley, et al. 2012). Other potential intermediate phenotypes are beginning to emerge, such as connectivity measures of the DMN or the ACC-amygdala circuit (Pezawas, Meyer-Lindenberg, et al. 2005; Chen, Suckling, et al. 2008; Li, Liu, et al. 2013). Imaging genetics studies demonstrate that these
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
91
intermediate phenotypes are under genetic control, and, moreover, that distinct genetic variants act on distinct brain systems. While genetic variation in BDNF was successfully linked to hippocampal function and volume, SLC6A4 seems to predominantly affect the amygdala-ACC pathway (Pezawas, Meyer-Lindenberg, et al. 2005; Pacheco, Beevers, et al. 2009; Kambeitz, Bhattacharyya, et al. 2012; Murphy, Norbury, et al. 2013). Taken together, these candidate gene studies provide a consistent picture that links genetic variation to alterations in brain circuits that in turn are related to antidepressant treatment response, even though the impact of these variants on treatment response may be too small to be clinically exploited (Porcelli, Fabbri, et al. 2012; Niitsu, Fabbri, et al. 2013). It is therefore not unreasonable to assume that these well-defined intermediate phenotypes could provide quantitative measures to substitute invalid behavioral assessments. This strategy may not only have the potential to increase the power of both reverse (candidate gene studies) and forward (genome-wide association studies) genetics research on antidepressant drug response, but may also contribute to a biologically meaningful psychiatric classification framework, as proposed in the Research Domain Criteria (RDoC) project (Meyer-Lindenberg and Weinberger 2006; Insel, Cuthbert, et al. 2010). While the adoption of intermediate phenotypes for pharmacogenetics research on antidepressant drug response is still in its infancy, the strength of this approach has been successfully demonstrated in other domains (Falcone, Smith, et al. 2013). In a seminal study, an amphetamine treatment x COMT Val158Met genotype interaction effect on working memory–related prefrontal activation was found, revealing the inverted-U functional response curve to increasing dopamine signaling in the prefrontal cortex (Mattay, Goldberg, et al. 2003). This study clearly demonstrates that the joint investigation of genetic and drug effects can offer a direct window to brain physiology. Nevertheless, while the integration of neuroimaging into psychiatric pharmacogenetic research will likely show new directions toward the understanding of antidepressant drug response, there are several hurdles that must be overcome. Recent meta-analyses suggest that the effect sizes of many imaging genetics associations are smaller than initially anticipated and are affected by peak voxel selection bias (Kambeitz, Bhattacharyya, et al. 2012; Vul and Pashler 2012; Dodds, Henson, et al. 2013; Murphy, Norbury, et al. 2013). Hence, larger sample sizes are required than originally proposed for imaging genetics studies (Molendijk, Bus, et al. 2012; Murphy, Norbury, et al. 2013). This is specifically true for investigations of gene-environment interactions or epistasis (Pezawas, Meyer-Lindenberg, et al.
2008; Klengel and Binder 2013). While this may only be feasible in large-scale collaborative efforts, first experiences with multicenter imaging-genomics investigations, such as the IMAGEN study, suggest that the neuroimaging community is ready to meet the challenges posed by such an endeavor (Schumann, Loth, et al. 2010). Another concern stems from the so-called “researcher degrees of freedom,” which refers to flexibility in data collection, analysis, and report (Linden 2012). Neuroimaging research is especially prone to this error, with nearly every study being based on a unique data analysis pipeline (Carp 2012). When analysis conditions are constant, the false-positive rate has been demonstrated to be sufficiently controlled using standard correction procedures; however, given the current flexibility of experimental designs, the true type I error rate in imaging genetics studies is likely considerably larger (Meyer-Lindenberg, Nicodemus, et al. 2008; Carp 2012). Although this flexibility may have initially been an advantage for discovery and basic science, the development of clinical diagnostics and translation to bedside requires highly standardized and robust procedures (Valenzuela, Bartres-Faz, et al. 2011). Fortunately, an increasing number of researchers are today addressing the test-retest reliability of neuroimaging measures, as well as the impact of methodological decisions of their analyses. Therefore it is reasonable to assume that these results will soon be put into practice (Weissenbacher, Kasess, et al. 2009; Plichta, Schwarz, et al. 2012; De Simoni, Schwarz, et al. 2013; Klomp, van Wingen, et al. 2013). The consideration of the above-mentioned issues will further strengthen the clinical relevance of intermediate phenotypes, which offers an unprecedented opportunity to go below the surface of behavioral measures of antidepressant drug response by revealing their biological mechanisms (Meyer-Lindenberg and Weinberger 2006). The increasing sophistication of genetic mapping techniques (finally exome or whole genome sequencing) and reliability of intermediate phenotypes, as well as studies in larger samples, are indicating that a thorough understanding of variable antidepressant drug response may be well under way.
REFERENCES Aan Het Rot M, Zarate CA, Jr., Charney DS, Mathew SJ. (2012). Ketamine for depression: where do we go from here? Biol Psychiatry 72(7): 537–547. Amat J, Baratta MV, Paul E, Bland ST, Watkins LR, Maier SF. (2005). Medial prefrontal cortex determines how stressor controllability affects behavior and dorsal raphe nucleus. Nat Neurosci. 8(3): 365–371.
9 2 Part I I : I maging G enetics and D rug D iscovery
American Psychiatric Association (APA). (2013). Diagnostic and statistical manual of mental disorders (5th ed., DSM-5). Arlington, VA: American Psychiatric Publishing. Anacker C, Zunszain PA, Cattaneo A, Carvalho L, Garabedian MJ, Thuret S, Price J, Pariante CM. (2011). Antidepressants increase human hippocampal neurogenesis by activating the glucocorticoid receptor. Mol Psychiatry. 16(7): 738–750. Anand A, Li Y, Wang Y, Gardner K, Lowe MJ. (2007). Reciprocal effects of antidepressant treatment on activity and connectivity of the mood regulating circuit: An fMRI study. J Neuropsychiatry Clin Neurosci. 19(3): 274–282. Anand A, Li Y, Wang Y, Wu J, Gao S, Bukhari L, Mathews VP, Kalnin A, Lowe MJ. (2005). Antidepressant effect on connectivity of the mood-regulating circuit: an FMRI study. Neuropsychopharmacology. 30(7): 1334–1344. Anderson IM, Del-Ben CM, McKie S, Richardson P, Williams SR, Elliott R, Deakin JF. (2007). Citalopram modulation of neuronal responses to aversive face emotions: a functional MRI study. Neuroreport. 18(13): 1351–1355. Anderson IM, Juhasz G, Thomas E, Downey D, McKie S, Deakin JF. Elliott R. (2011). The effect of acute citalopram on face emotion processing in remitted depression: a pharmacoMRI study. Eur Neuropsychopharmacol. 21(1): 140–148. Anderson IM, McKie S, Elliott R, Williams SR, Deakin JFW. (2008). Assessing human 5-HT function in vivo with pharmacoMRI. Neuropharmacology. 55(6): 1029–1037. Ansorge MS, Zhou M, Lira A, Hen R, Gingrich JA. (2004). Early-life blockade of the 5-HT transporter alters emotional behavior in adult mice. Science. 306(5697): 879–881. Anticevic A, Cole MW, Murray JD, Corlett PR, Wang XJ, Krystal JH. (2012). The role of default network deactivation in cognition and disease. Trends Cogn Sci. 16(12): 584–592. Anticevic A, Gancsos M, Murray JD, Repovs G, Driesen NR, Ennis DJ, Niciu MJ, Morgan PT, Surti TS, Bloch MH, Ramani R, Smith MA, Wang XJ, Krystal JH, Corlett PR. (2012). NMDA receptor function in large-scale anticorrelated neural systems with implications for cognition and schizophrenia. Proc Natl Acad Sci U S A. 109(41): 16720–16725. Arce E, Simmons AN, Lovero KL, Stein MB, Paulus MP. (2008). Escitalopram effects on insula and amygdala BOLD activation during emotional processing. Psychopharmacology. 196(4): 661–672. Arnone D, McIntosh AM, Ebmeier KP, Munafo MR, Anderson IM. (2012). Magnetic resonance imaging studies in unipolar depression: systematic review and meta-regression analyses. Eur Neuropsychopharmacol. 22(1): 1–16. Arnone D, McKie S, Elliott R, Juhasz G, Thomas EJ, Downey D, Williams S, Deakin JF, Anderson IM. (2012). State-dependent changes in hippocampal grey matter in depression. Mol Psychiatry. 18(12): 1265–1272. Arnone D, McKie S, Elliott R, Thomas EJ, Downey D, Juhasz G, Williams SR, Deakin JF, Anderson IM. (2012). Increased amygdala responses to sad but not fearful faces in major depression: relation to mood state and pharmacological treatment. Am J Psychiatry 169(8): 841–850. Ashburner J, Friston KJ. (2000). Voxel-based morphometry: the methods. Neuroimage. 11(6 Pt 1): 805–821. Autry AE, Monteggia LM. (2012). Brain-derived neurotrophic factor and neuropsychiatric disorders. Pharmacol Rev. 64(2): 238–258. Bagby RM, Ryder AG, Schuller DR, Marshall MB. (2004). The Hamilton Depression Rating Scale: has the gold standard become a lead weight? Am J Psychiatry. 161(12): 2163–2177. Bauer M, Pfennig A, Severus E, Whybrow PC, Angst J, Moller HJ, Task Force on Unipolar Depressive. (2013). World Federation of Societies of Biological Psychiatry (WFSBP) guidelines for biological treatment of unipolar depressive disorders, part 1: update 2013 on the acute and continuation treatment of unipolar depressive disorders. World J Biol Psychiatry. 14(5): 334–385.
Beckmann CF, Smith SM. (2004). Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans Med Imaging. 23(2): 137–152. Berman MG, Peltier S, Nee DE, Kross E, Deldin PJ, Jonides J. (2011). Depression, rumination and the default network. Soc Cogn Affect Neurosci. 6(5): 548–555. Bertolino A, Arciero G, Rubino V, Latorre V, De Candia M, Mazzola V, Blasi G, Caforio G, Hariri A, Kolachana B, Nardini M, Weinberger DR, Scarabino T. (2005). Variation of human amygdala response during threatening stimuli as a function of 5′HTTLPR genotype and personality style. Biol Psychiatry. 57(12): 1517–1525. Bartova L, Meyer BM, Diers K, Rabl U, Scharinger Ch, Popovic A, Pail G, Kalcher K, Boubela NR, Huemer J, Mandorfer D, Windischberger Ch, Sitte HH, Kasper S, Praschak-Rieder N, Moser E, Brocker B, Pezawas L. (2015). Reduced default mode network suppression during a working memory task in remitted major depression. Journal of Psychiatric Research. 64: 9–18. Bigos KL, Pollock BG, Aizenstein HJ, Fisher PM, Bies RR, Hariri AR. (2008). Acute 5-HT reuptake blockade potentiates human amygdala reactivity. Neuropsychopharmacology. 33(13): 3221–3225. Boldrini M, Hen R, Underwood MD, Rosoklija GB, Dwork AJ, Mann JJ, Arango V. (2012). Hippocampal angiogenesis and progenitor cell proliferation are increased with antidepressant use in major depression. Biol Psychiatry. 72: 562–571. Boldrini M, Underwood MD, Hen R, Rosoklija GB, Dwork AJ, Mann JJ, Arango V. (2009). Antidepressants increase neural progenitor cells in the human hippocampus. Neuropsychopharmacology. 34(11): 2376–2389. Brockmann H, Zobel A, Schuhmacher A, Daamen M, Joe A, Biermann K, Schwab SG, Biersack HJ, Maier W, Boecker H. (2011). Influence of 5-HTTLPR polymorphism on resting state perfusion in patients with major depression. J Psychiatr Res. 45(4): 442–451. Browning M, Reid C, Cowen PJ, Goodwin GM, Harmer CJ. (2007). A single dose of citalopram increases fear recognition in healthy subjects. J Psychopharmacol. 21(7): 684–690. Burghardt NS, Sullivan GM, McEwen BS, Gorman JM, Ledoux JE. (2004). The selective serotonin reuptake inhibitor citalopram increases fear after acute treatment but reduces fear with chronic treatment: A comparison with tianeptine. Biol Psychiatry. 55(12): 1171–1178. Canli T, Omura B, Haas BW, Fallgatter A, Constable RT, Lesch KP. (2005). Beyond affect: A role for genetic variation of the serotonin transporter in neural activation during a cognitive attention task. Proc Natl Acad Sci U S A. 102(34): 12224–12229. Canli T, Qiu M, Omura K, Congdon E, Haas BW, Amin Z, Herrmann MJ, Constable RT, Lesch KP. (2006). Neural correlates of epigenesis. Proc Natl Acad Sci U S A. 103(43): 16033–16038. Carballedo A, Amico F, Ugwu I, Fagan AJ, Fahey C, Morris D, Meaney JF, Leemans A, Frodl T. (2012). Reduced fractional anisotropy in the uncinate fasciculus in patients with major depression carrying the met-allele of the Val66Met brain-derived neurotrophic factor genotype. Am J Med Genet B Neuropsychiatr Genet 159B(5): 537–548. Carhart-Harris RL, Erritzoe D, Williams T, Stone JM, Reed LJ, Colasanti A, Tyacke RJ, Leech R, Malizia AL, Murphy K, Hobden P, Evans J, Feilding A, Wise RG, Nutt DJ. (2012). Neural correlates of the psychedelic state as determined by fMRI studies with psilocybin. Proc Natl Acad Sci U S A. 109(6): 2138–2143. Carp J. (2012). The secret lives of experiments: methods reporting in the fMRI literature. Neuroimage, 63(1): 289–300. Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H, McClay J, Mill J, Martin J, Braithwaite A, Poulton R. (2003). Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene. Science. 301(5631): 386–389. Chen CH, Suckling J, Ooi C, Fu CH, Williams SC, Walsh ND, Mitterschiffthaler MT, Pich EM, Bullmore E. (2008). Functional coupling of the amygdala in depressed patients treated with antidepressant medication. Neuropsychopharmacology. 33(8): 1909–1918.
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
93
Chiang MC, Barysheva M, Toga AW, Medland SE, Hansell NK, James MR, McMahon KL, de Zubicaray GI, Martin NG, Wright MJ, Thompson PM. (2011). BDNF gene effects on brain circuitry replicated in 455 twins. Neuroimage. 55(2): 448–454. Cole J, Weinberger DR, Mattay VS, Cheng X, Toga AW, Thompson PM, Powell-Smith G, Cohen-Woods S, Simmons A, McGuffin P, Fu CH. (2011). No effect of 5HTTLPR or BDNF Val66Met polymorphism on hippocampal morphology in major depression. Genes Brain Behav. 10(7): 756–764. Costafreda SG, McCann P, Saker P, Cole JH, Cohen-Woods S, Farmer AE, Aitchison KJ, McGuffin P, Fu CH. (2013). Modulation of amygdala response and connectivity in depression by serotonin transporter polymorphism and diagnosis. J Affect Disord. 150(1): 96–103. Crawford AA, Lewis G, Lewis SJ, Munafo MR. (2013). Systematic review and meta-analysis of serotonin transporter genotype and discontinuation from antidepressant treatment. Eur Neuropsychopharmacol. 23(10): 1143–1150. Dannlowski U, Konrad C, Kugel H, Zwitserlood P, Domschke K, Schöning S, Ohrmann P, Bauer J, Pyka M, Hohoff C, Zhang W, Baune BT, Heindel W, Arolt V, aSuslow T. (2010). Emotion specific modulation of automatic amygdala responses by 5-HTTLPR genotype. NeuroImage. 53(3): 893–898. Dannlowski U, Ohrmann P, Bauer J, Deckert J, Hohoff C, Kugel H, Arolt V, Heindel W, Kersting A, Baune BT, Suslow T. (2008). 5-HTTLPR biases amygdala activity in response to masked facial expressions in major depression. Neuropsychopharmacology 33(2): 418–424. Dannlowski U, Ohrmann P, Bauer J, Kugel H, Baune BT, Hohoff C, Kersting A, Arolt, V Heindel W, Deckert J, Suslow T. (2007). Serotonergic genes modulate amygdala activity in major depression. Genes, Brain and Behavior 6(7): 672–676. Dannlowski U, Stuhrmann A, Beutelmann V, Zwanzger P, Lenzen T, Grotegerd D, Domschke K, Hohoff C, Ohrmann P, Bauer J, Lindner C, Postert, Konrad C, Arolt V, Heindel W, Suslow T, Kugel H. (2012). Limbic scars: long-term consequences of childhood maltreatment revealed by functional and structural magnetic resonance imaging. Biol Psychiatry. 71(4): 286–293. Davidson RJ, Irwin W, Anderle MJ, Kalin NH. (2003). The neural substrates of affective processing in depressed patients treated with venlafaxine. Am J Psychiatry. 160(1): 64–75. Davidson RJ, McEwen BS. (2012). Social influences on neuroplasticity: stress and interventions to promote well-being. Nat Neurosci. 15(5): 689–695. de Almeida, JR, Phillips ML, Cerqueira CT, Zilberman M, Lobo D, Henna E, Tavares H, Amaro E, Gorenstein C, Gentil V, Busatto GF. (2010). Neural activity changes to emotional stimuli in healthy individuals under chronic use of clomipramine. J Psychopharmacol. 24(8): 1165–1174. de Kwaasteniet B, Ruhe E, Caan M, Rive M, Olabarriaga S, Groefsema M, Heesink L, van Wingen G, Denys D. (2013). Relation between structural and functional connectivity in major depressive disorder. Biol Psychiatry. 74(1): 40–47. De Simoni S, Schwarz AJ, O’Daly OG, Marquand AF, Brittain C, Gonzales C, Stephenson S, Williams SC, Mehta MA. (2013). Testretest reliability of the BOLD pharmacological MRI response to ketamine in healthy volunteers. Neuroimage. 64: 75–90. Deakin JF, Lees J, McKie S, Hallak JE, Williams SR, Dursun SM. (2008). Glutamate and the neural basis of the subjective effects of ketamine: a pharmaco-magnetic resonance imaging study. Arch Gen Psychiatry. 65(2): 154–164. Dedovic K, Duchesne A, Andrews J, Engert V, Pruessner JC. (2009). The brain and the stress axis: the neural correlates of cortisol regulation in response to stress. Neuroimage. 47(3): 864–871. Del-Ben, CM, Deakin JF, McKie S, Delvai NA, Williams SR, Elliott R, Dolan M, Anderson IM. (2005). The effect of citalopram pretreatment on neuronal responses to neuropsychological tasks in
normal volunteers: an FMRI study. Neuropsychopharmacology. 30(9): 1724–1734. Della Pasqua O, Santen GW, Danhof M. (2010). The missing link between clinical endpoints and drug targets in depression. Trends Pharmacol Sci. 31(4): 144–152. Dodds CM, Henson RN, Miller SR, Nathan PJ. (2013). Overestimation of the effects of the BDNF val66met polymorphism on episodic memory-related hippocampal function: a critique of a recent meta-analysis. Neurosci Biobehav Rev. 37(4): 739–741. Drevets WC, Price JL, Simpson JR, Jr, Todd RD, Reich T, Vannier M, Raichle ME. (1997). Subgenual prefrontal cortex abnormalities in mood disorders. Nature 386(6627): 824–827. Duman RS, Aghajanian GK. (2012). Synaptic dysfunction in depression: potential therapeutic targets. Science 338(6103): 68–72. Duman RS, Heninger GR, Nestler EJ. (1997). A molecular and cellular theory of depression. Arch Gen Psychiatry. 54(7): 597–606. Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, Zaitsev E, Gold B, Goldman D, Dean M, Lu B, Weinberger DR. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell 112(2): 257–269. Elzinga BM, Molendijk ML, Oude Voshaar RC, Bus BA, Prickaerts J, Spinhoven P, Penninx BJ. (2011). The impact of childhood abuse and recent stress on serum brain-derived neurotrophic factor and the moderating role of BDNF Val66Met. Psychopharmacology (Berl). 214(1): 319–328. Evans AC. (2013). Networks of anatomical covariance. Neuroimage. 80: 489–504. Everaerd D, Gerritsen L, Rijpkema M, Frodl T, I. van Oostrom I, Franke B, Fernandez G, Tendolkar I. (2012). Sex modulates the interactive effect of the serotonin transporter gene polymorphism and childhood adversity on hippocampal volume. Neuropsychopharmacology 37(8): 1848–1855. Fabbri C, Di Girolamo G, Serretti A. (2013). Pharmacogenetics of antidepressant drugs: an update after almost 20 years of research. Am J Med Genet B Neuropsychiatr Genet. 162(6): 487–520. Fair DA, Schlaggar BL, Cohen AL, Miezin FM, Dosenbach NU, Wenger KK, Fox MD, Snyder AZ, Raichle ME, Petersen SE. (2007). A method for using blocked and event-related fMRI data to study “resting state” functional connectivity. Neuroimage. 35(1): 396–405. Falcone M, Smith RM, Chenoweth MJ, Kumar Bhattacharjee A, Kelsoe JR, Tyndale RF, Lerman C. (2013). Neuroimaging in psychiatric pharmacogenetics research: the promise and pitfalls. Neuropsychopharmacology. 38(12): 2327–2337. Fischl B, Salat DH, Busa E, Albert M, Dieterich M, Haselgrove C, van der Kouwe A, Killiany R, Kennedy D, Klaveness S, Montillo A, Makris N, Rosen B, Dale AM. (2002). Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron. 33(3): 341–355. Fisher PM, Meltzer CC, Price JC, Coleman RL, Ziolko SK, Becker C, Moses-Kolko EL, Berga SL, Hariri AR. (2009). Medial prefrontal cortex 5-HT(2A) density is correlated with amygdala reactivity, response habituation, and functional coupling. Cereb Cortex. 19(11): 2499–2507. Fountoulakis KN, Hoschl C, Kasper S, Lopez-Ibor J, Moller HJ. (2013). The media and intellectuals’ response to medical publications: the antidepressants’ case. Ann Gen Psychiatry. 12(1): 11. Friedel E, Schlagenhauf F, Sterzer P, Park SQ, Bermpohl F, Ströhle A, Stoy M, Puls I, Hägele C, Wrase J, Büchel C, Heinz A. (2009). 5-HTT genotype effect on prefrontal-amygdala coupling differs between major depression and controls. Psychopharmacology. 205(2): 261–271. Frodl T, Koutsouleris N, Bottlender R, Born C, Jäger M, Mörgenthaler M, Scheuerecker J, Zill P, Baghai T, Schüle C, Rupprecht R, Bondy B, Reiser M, Möller HJ, Meisenzahl EM. (2008). Reduced gray matter brain volumes are associated with variants of the
9 4 Part I I : I maging G enetics and D rug D iscovery
serotonin transporter gene in major depression. Mol Psychiatry. 13(12): 1093–1101. Frodl T, Scheuerecker J, Schoepf V, Linn J, Koutsouleris N, Bokde AL, Hampel H, Moller HJ, Bruckmann H, Wiesmann M, Meisenzahl E. (2011). Different effects of mirtazapine and venlafaxine on brain activation: an open randomized controlled fMRI study. J Clin Psychiatry. 72(4): 448–457. Fu CH, Steiner H, Costafreda SG. (2013). Predictive neural biomarkers of clinical response in depression: a meta-analysis of functional and structural neuroimaging studies of pharmacological and psychological therapies. Neurobiol Dis. 52: 75–83. Fu CHY, Williams SCR, Cleare AJ, Brammer MJ, Walsh ND, Kim J, Andrew CM, Pich EM, Williams PM, Reed LJ, Mitterschiffthaler MT, Suckling J, Bullmore ET. (2004). Attenuation of the neural response to sad faces in major depression by antidepressant treatment: a prospective, event-related functional magnetic resonance imaging study. Arch Gen Psychiatry. 61(9): 877–889. Gaspar P, Cases O, Maroteaux L. (2003). The developmental role of serotonin: News from mouse molecular genetics. Nat Rev Neurosci. 4(12): 1002–1012. Gatt JM, Nemeroff CB, Dobson-Stone C, Paul RH, Bryant RA, Schofield PR, Gordon E, Kemp AH, Williams LM. (2009). Interactions between BDNF Val66Met polymorphism and early life stress predict brain and arousal pathways to syndromal depression and anxiety. Mol Psychiatry. 14(7): 681–695. Gerritsen L, Tendolkar I, Franke B, Vasquez AA, Kooijman S, Buitelaar J, Fernandez G, Rijpkema M. (2012). BDNF Val66Met genotype modulates the effect of childhood adversity on subgenual anterior cingulate cortex volume in healthy subjects. Mol Psychiatry. 17(6): 597–603. Greicius MD, Flores BH, Menon V, Glover GH, Solvason HB, Kenna H, Reiss AL, Schatzberg AF. (2007). Resting-state functional connectivity in major depression: abnormally increased contributions from subgenual cingulate cortex and thalamus. Biol Psychiatry. 62(5): 429–437. Grimm S, Boesiger P, Beck J, Schuepbach D, Bermpohl F, Walter M, Ernst J, Hell D, Boeker H, Northoff G. (2009). Altered negative BOLD responses in the default-mode network during emotion processing in depressed subjects. Neuropsychopharmacology 34(4): 932–843. Gueorguieva R, Mallinckrodt C, Krystal JH. (2011). Trajectories of depression severity in clinical trials of duloxetine: insights into antidepressant and placebo responses. Arch Gen Psychiatry. 68(12): 1227–1237. Gusnard DA, Akbudak E, Shulman GL, Raichle ME. (2001). Medial prefrontal cortex and self-referential mental activity: relation to a default mode of brain function. Proc Natl Acad Sci U S A. 98(7): 4259–4264. Gyekis JP, Yu W, Dong S, Wang H, Qian J, Kota P, Yang J. (2013). No association of genetic variants in BDNF with major depression: a meta- and gene-based analysis. Am J Med Genet B Neuropsychiatr Genet 162(1): 61–70. Hahn A, Wadsak W, Windischberger C, Baldinger P, Hoflich AS, Losak J, Nics L, Philippe C, Kranz GS, Kraus C, Mitterhauser M, Karanikas G, Kasper S, Lanzenberger R. (2012). Differential modulation of the default mode network via serotonin-1A receptors. Proc Natl Acad Sci U S A. 109(7): 2619–2624. Hajek T, Kopecek M, Hoschl C. (2012). Reduced hippocampal volumes in healthy carriers of brain-derived neurotrophic factor Val66Met polymorphism: meta-analysis. World J Biol Psychiatry. 13(3): 178–187. Hamilton JP, Chen G, Thomason ME, Schwartz ME, Gotlib IH. (2011). Investigating neural primacy in Major Depressive Disorder: multivariate Granger causality analysis of resting-state fMRI time-series data. Mol Psychiatry. 16(7): 763–772. Hamilton JP, Etkin A, Furman DJ, Lemus MG, Johnson RF, Gotlib IH. (2012). Functional neuroimaging of major depressive disorder: a
meta-analysis and new integration of base line activation and neural response data. Am J Psychiatry. 169(7): 693–703. Hamilton JP, Furman DJ, Chang C, Thomason ME, Dennis E, Gotlib IH. (2011). Default-mode and task-positive network activity in major depressive disorder: implications for adaptive and maladaptive rumination. Biol Psychiatry. 70(4): 327–333. Hamilton JP, Siemer M, Gotlib IH. (2008). Amygdala volume in major depressive disorder: a meta-analysis of magnetic resonance imaging studies.” Mol Psychiatry. 13(11): 993–1000. Hariri AR, Drabant EM, Munoz KE, Kolachana BS, Mattay VS, Egan MF, Weinberger DR. (2005). A susceptibility gene for affective disorders and the response of the human amygdala. Arch Gen Psychiatry. 62(2): 146–152. Hariri AR, Goldberg TE, Mattay VS, Kolachana BS, Callicott JH, Egan MF, Weinberger DR. (2003). Brain-derived neurotrophic factor val66met polymorphism affects human memory-related hippocampal activity and predicts memory performance. J Neurosci. 23(17): 6690–6694. Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D, Egan MF, Weinberger DR. (2002). Serotonin transporter genetic variation and the response of the human amygdala.” Science. 297(5580): 400–403. Harkness KL, Monroe SM. (2002). Childhood adversity and the endogenous versus nonendogenous distinction in women with major depression. Am J Psychiatry. 159(3): 387–393. Harmer CJ, Mackay CE, Reid CB, Cowen PJ, Goodwin GM. (2006). Antidepressant drug treatment modifies the neural processing of nonconscious threat cues. Biol Psychiatry. 59(9): 816–820. Heils A, Teufel A, Petri S, Stöber G, Riederer P, Bengel D, Lesch KP. (1996). Allelic variation of human serotonin transporter gene expression. J Neurochem. 66(6): 2621–2624. Heinz A, Braus DF, Smolka MN, Wrase J, Puls I, Hermann D, Klein S, Grüsser SM, Flor H, Schumann G, Mann K, Büchel C. (2005). Amygdala-prefrontal coupling depends on a genetic variation of the serotonin transporter. Nat Neurosci. 8(1): 20–21. Heinz A, Smolka MN, Braus DF, Wrase J, Beck A, Flor H, Mann K, Schumann G, Buchel C, Hariri AR, Weinberger DR. (2007). Serotonin transporter genotype (5-HTTLPR): effects of neutral and undefined conditions on amygdala activation. Biol Psychiatry. 61(8): 1011–1014. Henry A, Kisicki MD, Varley C. (2012). Efficacy and safety of antidepressant drug treatment in children and adolescents. Mol Psychiatry. 17(12): 1186–1193. Henry ME, Lauriat TL, Lowen SB, Churchill JH, Hodgkinson CA, Goldman D, Renshaw PF. (2013). Effects of citalopram and escitalopram on fMRI response to affective stimuli in healthy volunteers selected by serotonin transporter genotype. Psychiatry Res. 213(3): 217–224. Hirschfeld RM. (2000). History and evolution of the monoamine hypothesis of depression. J Clin Psychiatry. 61(Suppl 6): 4–6. Ho PS, Ho KK, Huang WS, Yen CH, Shih MC, Shen LH, Ma KH, Huang SY. (2013). Association study of serotonin transporter availability and SLC6A4 gene polymorphisms in patients with major depression. Psychiatry Res. 212(3): 216–222. Holmes AJ, Lee PH, Hollinshead MO, Bakst L, Roffman JL, Smoller JW, Buckner RL. (2012). Individual differences in amygdala-medial prefrontal anatomy link negative affect, impaired social functioning, and polygenic depression risk. J Neurosci. 32(50): 18087–18100. Holsboer F. (2008). How can we realize the promise of personalized antidepressant medicines? Nat Rev Neurosci. 9(8): 638–646. Homberg JR, Schubert D, Gaspar P. (2010). New perspectives on the neurodevelopmental effects of SSRIs. Trends Pharmacol Sci. 31(2): 60–65. Honey GD, Corlett PR, Absalom AR, Lee M, Pomarol-Clotet E, Murray GK, McKenna PJ, Bullmore ET, Menon DK, Fletcher PC. (2008). Individual differences in psychotic effects of ketamine are
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
95
predicted by brain function measured under placebo. J Neurosci. 28(25): 6295–6303. Honey GD, Honey RA, O’Loughlin C, Sharar SR, Kumaran D, Suckling J, Menon DK, Sleator C, Bullmore ET, Fletcher PC. (2005). Ketamine disrupts frontal and hippocampal contribution to encoding and retrieval of episodic memory: an fMRI study. Cereb Cortex. 15(6): 749–759. Honey RA, Honey GD, O’Loughlin C, Sharar SR, Kumaran D, Bullmore ET, DMenon DK, Donovan T, Lupson VC, Bisbrown-Chippendale R, Fletcher PC. (2004). Acute ketamine administration alters the brain responses to executive demands in a verbal working memory task: an FMRI study. Neuropsychopharmacology. 29(6): 1203–1214. Insel T, Cuthbert B, Garvey M, Heinssen R, Pine DS, Quinn K, Sanislow C, Wang P. (2010). Research domain criteria (RDoC): toward a new classification framework for research on mental disorders. Am J Psychiatry. 167(7): 748–751. GENDEP Investigators, MARS Investigators, STAR*D Investigators (2013). Common genetic variation and antidepressant efficacy in major depressive disorder: a meta-analysis of three genome-wide pharmacogenetic studies. Am J Psychiatry. 170(2): 207–217. Jedema HP, Gianaros PJ, Greer PJ, Kerr DD, Liu S, Higley JD, Suomi SJ, Olsen AS, Porter JN, Lopresti BJ, Hariri AR,Bradberry CW. (2010). Cognitive impact of genetic variation of the serotonin transporter in primates is associated with differences in brain morphology rather than serotonin neurotransmission. Mol Psychiatry. 15(5): 512–522. Johnstone T, van Reekum CM, Urry HL, Kalin NH, Davidson RJ. (2007). Failure to regulate: counterproductive recruitment of top-down prefrontal-subcortical circuitry in major depression. J Neurosci. 27(33): 8877–8884. Jovanovic H, Perski A, Berglund H, Savic I. (2011). Chronic stress is linked to 5-HT(1A) receptor changes and functional disintegration of the limbic networks. Neuroimage. 55(3): 1178–1188. Kalbitzer J, Erritzoe D, Holst KK, Nielsen FÅ, Marner L, Lehel S, Arentzen T, Jernigan TL, Knudsen GM. (2010). Seasonal changes in brain serotonin transporter binding in short serotonin transporter linked polymorphic region-allele carriers but not in long-allele homozygotes. Biol Psychiatry. 67(11): 1033–1039. Kalbitzer J, Frokjaer VG, Erritzoe D, Svarer C, Cumming P, Nielsen FÅ, Hashemi SH, Baaré WFC, Madsen J, Hasselbalch SG, Kringelbach ML, Mortensen EL, Knudsen GM. (2009). The personality trait openness is related to cerebral 5-HTT levels. NeuroImage. 45(2): 280–285. Kambeitz JP, Bhattacharyya S, Kambeitz-Ilankovic LM, Valli I, Collier DA, McGuire P. (2012). Effect of BDNF val(66)met polymorphism on declarative memory and its neural substrate: a meta-analysis. Neurosci Biobehav Rev. 36(9): 2165–2177. Kapur S, Phillips AG, Insel TR. (2012). Why has it taken so long for biological psychiatry to develop clinical tests and what to do about it? Mol Psychiatry. 17(12): 1174–1179. Karg K, Burmeister M, Shedden K, Sen S. (2011). The serotonin transporter promoter variant (5-HTTLPR), stress, and depression meta-analysis revisited: evidence of genetic moderation. Arch Gen Psychiatry. 68(5): 444–454. Kato M, Serretti A. (2010). Review and meta-analysis of antidepressant pharmacogenetic findings in major depressive disorder. Mol Psychiatry. 15(5): 473–500. Keedwell PA, Chapman R, Christiansen K, Richardson H, Evans J, Jones DK. (2012). Cingulum white matter in young women at risk of depression: the effect of family history and anhedonia. Biol Psychiatry. 72(4): 296–302. Kempton MJ, Salvador Z, Munafo MR, Geddes JR, Simmons A, Frangou S, Williams SC. (2011). Structural neuroimaging studies in major depressive disorder: meta-analysis and comparison with bipolar disorder. Arch Gen Psychiatry. 68(7): 675–690.
Kendler KS, Karkowski LM, Prescott CA. (1999). Causal relationship between stressful life events and the onset of major depression. Am J Psychiatry. 156(6): 837–841. Kim MJ, Whalen PJ. (2009). The structural integrity of an amygdala-prefrontal pathway predicts trait anxiety. J Neurosci. 29(37): 11614–11618. Kirchheiner J, Seeringer A, Godoy AL, Ohmle B, Maier C, Beschoner P, Sim EJ, Viviani R. (2011). CYP2D6 in the brain: genotype effects on resting brain perfusion. Mol Psychiatry. 16(3): 237, 333–241. Klengel T, Binder EB. (2013). Gene x environment interactions in the prediction of response to antidepressant treatment. Int J Neuropsychopharmacol. 16(3): 701–711. Klomp A, van Wingen GA, de Ruiter MB, Caan MW, Denys D, Reneman L. (2013). Test-retest reliability of task-related pharmacological MRI with a single-dose oral citalopram challenge. Neuroimage. 75: 108–116. Kobiella A, Reimold M, Ulshofer DE, Ikonomidou VN, Vollmert C, Vollstadt-Klein S, Rietschel M, Reischl G, Heinz A, Smolka MN. (2011). How the serotonin transporter 5-HTTLPR polymorphism influences amygdala function: the roles of in vivo serotonin transporter expression and amygdala structure. Transl Psychiatry. 1: e37. Krishnan V, Nestler EJ. (2008). The molecular neurobiology of depression. Nature. 455(7215): 894–902. Kupfer DJ, Frank E, Phillips ML. (2012). Major depressive disorder: new clinical, neurobiological, and treatment perspectives. Lancet 379(9820): 1045–1055. Laje G, Lally N, Mathews D, Brutsche N, Chemerinski A, Akula N, Kelmendi B, Simen A, McMahon FJ, Sanacora G, Zarate C, Jr. (2012). Brain-derived neurotrophic factor Val66Met polymorphism and antidepressant efficacy of ketamine in depressed patients. Biol Psychiatry. 72(11): e27–28. Lau JY, Goldman D, Buzas B, Hodgkinson C, Leibenluft E, Nelson E, Sankin L, Pine DS, Ernst M. (2010). BDNF gene polymorphism (Val66Met) predicts amygdala and anterior hippocampus responses to emotional faces in anxious and depressed adolescents. Neuroimage. 53(3): 952–961. Leech R, Sharp DJ. (2013). The role of the posterior cingulate cortex in cognition and disease. Brain. Lemogne C, Gorwood P, Boni C, Pessiglione M, Lehericy S, Fossati P. (2011). Cognitive appraisal and life stress moderate the effects of the 5-HTTLPR polymorphism on amygdala reactivity. Hum Brain Mapp. 32(11): 1856–1867. Lesch KP, Bengel D, Heils A, Sabol SZ, Greenberg BD, Petri S, Benjamin J, Müller CR, Hamer DH, Murphy DL. (1996). Association of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory region. Science. 274(5292): 1527–1531. Levy-Gigi E, Szabo C, Kelemen O, Keri S. (2013). Association among clinical response, hippocampal volume, and FKBP5 gene expression in individuals with posttraumatic stress disorder receiving cognitive behavioral therapy.” Biol Psychiatry. 74(11): 793–800. Li B, Liu L, Friston KJ, Shen H, Wang L, Zeng LL Hu D. (2013). A treatment-resistant default mode subnetwork in major depression. Biol Psychiatry. 74(1): 48–54. Licinio J, Dong C, Wong ML. (2009). Novel sequence variations in the brain-derived neurotrophic factor gene and association with major depression and antidepressant treatment response. Arch Gen Psychiatry. 66(5): 488–497. Linden DE. (2012). The challenges and promise of neuroimaging in psychiatry. Neuron. 73(1): 8–22. Liu RJ, Lee FS, Li XY, Bambico F, Duman RS, Aghajanian GK. (2012). Brain-derived neurotrophic factor Val66Met allele impairs basal and ketamine-stimulated synaptogenesis in prefrontal cortex. Biol Psychiatry. 71(11): 996–1005. Lonsdorf TB, Golkar A, Lindstom KM, Fransson P, Schalling M, Ohman A, Ingvar M. (2011). 5-HTTLPR and COMTval158met genotype gate amygdala reactivity and habituation. Biol Psychol. 87(1): 106–112.
9 6 Part I I : I maging G enetics and D rug D iscovery
Lopez-Sola M, Pujol J, Hernandez-Ribas R, Harrison BJ, Contreras-Rodriguez O, Soriano-Mas C, Deus J, Ortiz H, Menchon JM, Vallejo J, Cardoner N. (2010). Effects of duloxetine treatment on brain response to painful stimulation in major depressive disorder. Neuropsychopharmacology 35(11): 2305–2317. Lozano AM, Mayberg HS, Giacobbe P, Hamani C, Craddock RC, Kennedy SH. (2008). Subcallosal cingulate gyrus deep brain stimulation for treatment-resistant depression. Biol Psychiatry. 64(6): 461–467. Macoveanu J, Knorr U, Skimminge A, Sondergaard MG, Jorgensen A, Fauerholdt-Jepsen M, Paulson OB, Knudsen GM, Siebner HR, Kessing LV. (2014). Altered reward processing in the orbitofrontal cortex and hippocampus in healthy first-degree relatives of patients with depression. Psychol Med. 44(6): 1183–1195 1–13. Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium, Ripke S, Wray NR, Lewis CM, Hamilton SP, Weissman MM, Breen G, Byrne EM, Blackwood DH, Boomsma DI, Cichon S, … Sullivan PF. (2013). A mega-analysis of genome-wide association studies for major depressive disorder. Mol Psychiatry. 18(4): 497–511. Martin C, Sibson NR. (2008). Pharmacological MRI in animal models: a useful tool for 5-HT research? Neuropharmacology. 55(6): 1038–1047. Martin J, Cleak J, Willis-Owen SA, Flint J, Shifman S. (2007). Mapping regulatory variants for the serotonin transporter gene based on allelic expression imbalance. Mol Psychiatry. 12(5): 421–422. Martinowich K, Jimenez DV, Zarate CA, Jr. Manji HK. (2013). Rapid antidepressant effects: moving right along. Mol Psychiatry. 18(8): 856–863. Mascetti L, Foret A, Schrouff J, Muto V, Dideberg V, Balteau E, Degueldre C, Phillips C, Luxen A, Collette F, Bours V, Maquet P. (2013). Concurrent synaptic and systems memory consolidation during sleep. J Neurosci. 33(24): 10182–10190. Mason MF, Norton MI, Van Horn JD, Wegner DM, Grafton ST, Macrae CN. (2007). Wandering minds: the default network and stimulus-independent thought. Science. 315(5810): 393–395. Matsuo K, Glahn DC, Peluso MA, Hatch JP, Monkul ES, Najt P, Sanches M, Zamarripa F, Li J, Lancaster JL, Fox PT, Gao JH, Soares JC. (2007). Prefrontal hyperactivation during working memory task in untreated individuals with major depressive disorder. Mol Psychiatry. 12(2): 158–166. Matsuo K, Walss-Bass C, Nery FG, Nicoletti MA, Hatch JP, Frey BN, Monkul ES, Zunta-Soares GB, Bowden CL, Escamilla MA, Soares JC. (2009). Neuronal correlates of brain-derived neurotrophic factor Val66Met polymorphism and morphometric abnormalities in bipolar disorder. Neuropsychopharmacology. 34(8): 1904–1913. Mattay VS, Goldberg TE, Fera F, Hariri AR, Tessitore A, Egan MF, Kolachana B, Callicott JH, Weinberger DR. (2003). Catechol O-methyltransferase val158-met genotype and individual variation in the brain response to amphetamine. Proc Natl Acad Sci U S A. 100(10): 6186–6191. McCabe C, Mishor Z, Cowen PJ, Harmer CJ. (2010). Diminished neural processing of aversive and rewarding stimuli during selective serotonin reuptake inhibitor treatment. Biol Psychiatry. 67(5): 439–445. McGrath PJ, Stewart JW, Quitkin FM, Chen Y, Alpert JE, Nierenberg AA, Fava M, Cheng J, Petkova E. (2006). Predictors of relapse in a prospective study of fluoxetine treatment of major depression. Am J Psychiatry. 163(9): 1542–1548. McKie S, Del-Ben C, Elliott R, Williams S, del Vai N, Anderson I, Deakin JF. (2005). Neuronal effects of acute citalopram detected by pharmacoMRI. Psychopharmacology (Berl). 180(4): 680–686. Meyer BM, Huemer J, Rabl U, Boubele NR, Kalcher K, Berger A, Banschweski T, Barker G, Bokde A, Büchel Ch, Conrod P, Desrivières S, Flor H, Frouin V, Gallinat G, Garvan H, Heinz A, Ittermann B,
Jia T, Lathrop M, Martinot J-L, Nees F, Rietschel M, Smolka MN, Bartova L, Popovic A, Scharinger Ch, Sitte HH, Steiner H, Friedrich MH, Kasper S, Perkmann T, Praschak-Rieder N, Haslacher H, Esterbauer H, Moser E, Schumann G, Pezawas L. (2014). Oppositional COMT Val158Met effects on resting state functional connectivity in adolescents and adults. Brain Structure and Function. Epub ahead of print. Meyer-Lindenberg A, Nicodemus KK, Egan MF, Callicott JH, Mattay V, Weinberger DR. (2008). False positives in imaging genetics. NeuroImage. 40(2): 655–661. Meyer-Lindenberg A, Weinberger DR. (2006). Intermediate phenotypes and genetic mechanisms of psychiatric disorders. Nat Rev Neurosci. 7(10): 818–827. Molendijk ML, Bus BA, Spinhoven P, Kaimatzoglou A, Oude Voshaar RC, Penninx BW, I. M. H. van Elzinga BM. (2012). A systematic review and meta-analysis on the association between BDNF val(66) met and hippocampal volume—a genuine effect or a winners curse? Am J Med Genet B Neuropsychiatr Genet 159B(6): 731–740. Molendijk, ML, Bus BA, Spinhoven P, Penninx BW, Kenis G, rickaerts J, Voshaar RC, Elzinga BM. (2011). Serum levels of brain-derived neurotrophic factor in major depressive disorder: state-trait issues, clinical features and pharmacological treatment. Mol Psychiatry. 16(11): 1088–1095. Molendijk ML, Spinhoven P, Polak M, Bus BA, Penninx BW, Elzinga BM. (2013). Serum BDNF concentrations as peripheral manifestations of depression: evidence from a systematic review and meta-analyses on 179 associations (N=9484). Mol Psychiatry. 19(7): 791–800. Montag C, Reuter M, Newport B, Elger C, Weber B. (2008). The BDNF Val66Met polymorphism affects amygdala activity in response to emotional stimuli: Evidence from a genetic imaging study. NeuroImage. 42(4): 1554–1559. Montag C, Weber B, Fliessbach K, Elger C, Reuter M. (2009). The BDNF Val66Met polymorphism impacts parahippocampal and amygdala volume in healthy humans: Incremental support for a genetic risk factor for depression. Psychol Med. 39(11): 1831–1839. Morey RA, Petty CM, Xu Y, Hayes JP, Wagner HR, 2nd, Lewis DV, LaBar KS, Styner M, McCarthy G. (2009). A comparison of automated segmentation and manual tracing for quantifying hippocampal and amygdala volumes. Neuroimage 45(3): 855–866. Mukherjee P, Whalley HC, McKirdy JW, McIntosh AM, Johnstone EC, Lawrie SM, Hall J. (2011). Effects of the BDNF Val66Met polymorphism on neural responses to facial emotion. Psychiatry Res. 191(3): 182–188. Murphy SE, Norbury R, Godlewska BR, Cowen PJ, Mannie ZM, Harmer CJ, Munafo MR. (2013). The effect of the serotonin transporter polymorphism (5-HTTLPR) on amygdala function: a meta-analysis. Mol Psychiatry. 18(4): 512–520. Murphy SE, Norbury R, O’Sullivan U, PCowen PJ, Harmer CJ. (2009). Effect of a single dose of citalopram on amygdala response to emotional faces. Br J Psychiatry. 194(6): 535–540. Murthy NV, Selvaraj S, Cowen PJ, Bhagwagar Z, Riedel WJ, Peers P, Kennedy JL, Sahakian BJ, Laruelle MA, Rabiner EA, Grasby PM. (2010). Serotonin transporter polymorphisms (SLC6A4 insertion/deletion and rs25531) do not affect the availability of 5-HTT to [11C] DASB binding in the living human brain. NeuroImage. 52(1): 50–54. Nestler EJ, Hyman SE. (2010). Animal models of neuropsychiatric disorders. Nat Neurosci. 13(10): 1161–1169. Niitsu T, Fabbri C, Bentini F, Serretti A. (2013). Pharmacogenetics in major depression: a comprehensive meta-analysis. Prog Neuropsychopharmacol Biol Psychiatry. 45: 183–194. Norbury R, Mackay CE, Cowen PJ, Goodwin GM, Harmer CJ. (2008). The effects of reboxetine on emotional processing in healthy volunteers: an fMRI study. Mol Psychiatry. 13(11): 1011–1020. Nordanskog P, Dahlstrand U, Larsson MR, Larsson EM, Knutsson L, Johanson A. (2010). Increase in hippocampal volume after
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
97
electroconvulsive therapy in patients with depression: a volumetric magnetic resonance imaging study. J ECT. 26(1): 62–67. O’Reilly JX, Woolrich MW, Behrens TE, Smith SM, Johansen-Berg H. (2012). Tools of the trade: psychophysiological interactions and functional connectivity. Soc Cogn Affect Neurosci. 7(5): 604–609. Ongur D, Drevets WC, Price JL. (1998). Glial reduction in the subgenual prefrontal cortex in mood disorders. Proc Natl Acad Sci U S A. 95(22): 13290–13295. Onur OA, Walter H, Schlaepfer TE, Rehme AK, Schmidt C, Keysers C, Maier W, Hurlemann R. (2009). Noradrenergic enhancement of amygdala responses to fear. Soc Cogn Affect Neurosci. 4(2): 119–126. Pacheco J, Beevers CG, Benavides C, McGeary J, Stice E, Schnyer DM. (2009). Frontal-limbic white matter pathway associations with the serotonin transporter gene promoter region (5-HTTLPR) polymorphism. J Neurosci. 29(19): 6229–6233. Parsey RV, Hastings RS, Oquendo MA, Hu X, Goldman D, Huang YY, Simpson N, Arcement J, Huang Y, Ogden RT, Van Heertum RL, Arango V, Mann JJ. (2006). Effect of a triallelic functional polymorphism of the serotonin- transporter-linked promoter region on expression of serotonin transporter in the human brain. Am J Psychiatry. 163(1): 48–51. Passamonti L, Cerasa A, Gioia MC, Magariello A, Muglia M, Quattrone A, Fera F. (2008). Genetically dependent modulation of serotonergic inactivation in the human prefrontal cortex. NeuroImage 40(3): 1264–1273. Passamonti L, Crockett MJ, Apergis-Schoute AM, Clark L, Rowe JB, Calder AJ, Robbins TW. (2012). Effects of acute tryptophan depletion on prefrontal-amygdala connectivity while viewing facial signals of aggression. Biol Psychiatry. 71(1): 36–43. Pei Y, Smith AK, Wang Y, Pan Y, Yang J, Chen Q, Pan W, Bao F, Zhao L, Tie C, Wang Y, Wang J, Zhen W, Zhou J, Ma X. (2012). The brain-derived neurotrophic-factor (BDNF) val66met polymorphism is associated with geriatric depression: a meta-analysis. Am J Med Genet B Neuropsychiatr Genet. 159B(5): 560–566. Persson A, Sim SC, Virding S, Onishchenko N, Schulte G, Ingelman-Sundberg M. (2013). Decreased hippocampal volume and increased anxiety in a transgenic mouse model expressing the human CYP2C19 gene. Mol Psychiatry. 19(6): 733–741. Pezawas L, Meyer-Lindenberg A, Drabant EM, Verchinski BA, Munoz KE, Kolachana BS, Egan MF, Mattay VS, Hariri AR, Weinberger DR. (2005). 5-HTTLPR polymorphism impacts human cingulate-amygdala interactions: a genetic susceptibility mechanism for depression. Nat Neurosci. 8(6): 828–834. Pezawas L, Meyer-Lindenberg A, Goldman AL, Verchinski BA, Chen G, Kolachana BS, Egan MF, Mattay VS, Hariri AR, Weinberger DR. (2008). Evidence of biologic epistasis between BDNF and SLC6A4 and implications for depression. Mol Psychiatry. 13(7): 709–716. Pezawas L, Verchinski BA, Mattay VS, Callicott JH, Kolachana BS, Straub RE, Egan MF, Meyer-Lindenberg A, Weinberger DR. (2004). The brain-derived neurotrophic factor val66met polymorphism and variation in human cortical morphology. J Neurosci. 24(45): 10099–10102. Plichta MM, Schwarz AJ, Grimm O, Morgen K, Mier D, Haddad L, Gerdes AB, Sauer C, Tost H, Esslinger C, Colman P, Wilson F, Kirsch P, Meyer-Lindenberg A. (2012). Test-retest reliability of evoked BOLD signals from a cognitive-emotive fMRI test battery. Neuroimage. 60(3): 1746–1758. Plomin R, Haworth CM, Davis OS. (2009). Common disorders are quantitative traits. Nat Rev Genet. 10(12): 872–878. Pollak DD, Monje FJ, Zuckerman L, Denny CA, Drew MR, Kandel ER. (2008). An animal model of a behavioral intervention for depression. Neuron. 60(1): 149–161. Porcelli S, Fabbri C, Serretti A. (2012). Meta-analysis of serotonin transporter gene promoter polymorphism (5-HTTLPR) association with antidepressant efficacy. Eur Neuropsychopharmacol. 22(4): 239–258.
Posner J, Hellerstein DJ, Gat I, Mechling A, Klahr K, Wang Z, McGrath PJ, Stewart JW, Peterson BS. (2013). Antidepressants normalize the default mode network in patients with dysthymia. JAMA Psychiatry. 70(4): 373–382. Praschak-Rieder N, Kennedy J, Wilson AA, Hussey D, Boovariwala A, Willeit M, Ginovart N, Tharmalingam S, Masellis M, Houle S, Meyer JH. (2007). Novel 5-HTTLPR allele associates with higher serotonin transporter binding in putamen: a [11C] DASB positron emission tomography study. Biol Psychiatry. 62(4): 327–331. Price JS, Strong J, Eliassen J, McQueeny T, Miller M, Padula CB, Shear P, Lisdahl K. (2013). Serotonin transporter gene moderates associations between mood, memory and hippocampal volume. Behav Brain Res. 242: 158–165. Purcell SM. (2013). Antidepressant response and polygenes. Biol Psychiatry. 73(7): 600–601. Rabl U, Meyer BM, Diers K, Bartova L, Berger A, Mandorfer D, Popovic A, Scharinger C, Huemer J, Kalcher K, Pail G, Haslacher H, Perkmann T, Windischberger C, Brocke B, Sitte HH, Pollak DD, Dreher JC, Kasper S, Praschak-Rieder N, Moser E, Esterbauer H, Pezawas L. (2014). Additive gene-environment effects on hippocampal structure in healthy humans. Journal of Neuroscience. 34(30): 9917–9926. Rabl U, Scharinger C, Muller M, Pezawas L. (2010). Imaging genetics: implications for research on variable antidepressant drug response. Expert Rev Clin Pharmacol. 3(4): 471–489. Rao H, Gillihan SJ, Wang J, Korczykowski M, Sankoorikal GMV, Kaercher KA, Brodkin ES, Detre JA, Farah MJ. (2007). Genetic variation in serotonin transporter alters resting brain function in healthy individuals. Biol Psychiatry. 62(6): 600–606. Rao U, Chen LA, Bidesi AS, Shad MU, Thomas MA, Hammen CL. (2010). Hippocampal changes associated with early-life adversity and vulnerability to depression. Biol Psychiatry. 67(4): 357–364. Rawlings NB, Norbury R, Cowen PJ, Harmer CJ. (2010). A single dose of mirtazapine modulates neural responses to emotional faces in healthy people. Psychopharmacology (Berl). 212(4): 625–634. Reimold M, Smolka MN, Schumann G, Zimmer A, Wrase J, Mann K, Hu XZ, Goldman D, Reischl G, Solbach C, Machulla HJ, Bares R, Heinz A. (2007). Midbrain serotonin transporter binding potential measured with [11C]DASB is affected by serotonin transporter genotype. J Neural Trans. 114(5): 635–639. Risch N, Herrell R, Lehner T, Liang KY, Eaves L, Hoh J, Griem A, Kovacs M, Ott J, Merikangas KR. (2009). Interaction between the serotonin transporter gene (5-HTTLPR), stressful life events, and risk of depression: a meta-analysis. JAMA. 301(23): 2462–2471. Robertson B, Wang L, Diaz MT, Aiello M, Gersing K, Beyer J, Mukundan S, Jr, McCarthy G, Doraiswamy PM. (2007). Effect of bupropion extended release on negative emotion processing in major depressive disorder: A pilot functional magnetic resonance imaging study. J Clin Psychiatry. 68(2): 261–267. Roden DM, George, AL, Jr. (2002). The genetic basis of variability in drug responses. Nat Rev Drug Discov. 1(1): 37–44. Rowland LM, Bustillo JR, Mullins PG, Jung RE, Lenroot R, Landgraf E, Barrow R, Yeo R, Lauriello J, Brooks WM. (2005). Effects of ketamine on anterior cingulate glutamate metabolism in healthy humans: a 4-T proton MRS study. Am J Psychiatry. 162(2): 394–396. Rozeske RR, Evans AK, Frank MG, Watkins LR, Lowry CA, Maier SF. (2011). Uncontrollable, but not controllable, stress desensitizes 5-HT1A receptors in the dorsal raphe nucleus. J Neurosci. 31(40): 14107–14115. Salvadore G, Cornwell BR, Colon-Rosario V, Coppola R, Grillon C, Zarate CA, Jr, Manji HK. (2009). Increased anterior cingulate cortical activity in response to fearful faces: a neurophysiological biomarker that predicts rapid antidepressant response to ketamine. Biol Psychiatry. 65(4): 289–295. Salvadore G, Cornwell BR, Sambataro F, Latov D, Colon-Rosario V, Carver F, Holroyd T, DiazGranados N, Machado-Vieira R, Grillon C, Drevets WC, Zarate CA, Jr. (2010). Anterior cingulate
9 8 Part I I : I maging G enetics and D rug D iscovery
desynchronization and functional connectivity with the amygdala during a working memory task predict rapid antidepressant response to ketamine. Neuropsychopharmacology. 35(7): 1415–1422. Santarelli L, Saxe M, Gross C, Surget A, Battaglia F, Dulawa S, Weisstaub N, Lee J, Duman R, Arancio O, Belzung C, Hen R. (2003). Requirement of hippocampal neurogenesis for the behavioral effects of antidepressants. Science. 301(5634): 805–809. Sapolsky RM. (2000). Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Arch Gen Psychiatry. 57(10): 925–935. Savitz JB, Drevets WC. (2013). Neuroreceptor imaging in depression. Neurobiol Dis. 52: 49–65. Schaefer HS, Putnam KM, Benca RM, Davidson RJ. (2006). Event-related functional magnetic resonance imaging measures of neural activity to positive social stimuli in pre- and post-treatment depression. Biol Psychiatry. 60(9): 974–986. Schardt DM, Erk S, Nusser C, Nothen MM, Cichon S, Rietschel M, Treutlein J, Goschke T, Walter H. (2010). Volition diminishes genetically mediated amygdala hyperreactivity. Neuroimage 53(3): 943–951. Scharinger C, Rabl U, Kasess CH, Meyer BM, Hofmaier T, Diers K, Bartova L, Pail G, Huf W, Uzelac Z, Hartinger B, Kalcher K, Perkmann T, Haslacher H, Meyer-Lindenberg A, Kasper S, Freissmuth M, Windischberger C, Willeit M, Lanzenberger R, Esterbauer H, Brocke B, Moser E, Sitte HH, Pezawas L. Platelet serotonin transporter function predicts default-mode network activity. PLoS One. 9(3): e92543. Scharinger C, Rabl U, Sitte HH, Pezawas L. (2010). Imaging genetics of mood disorders. Neuroimage. 53(3): 810–821. Scheidegger M, Walter M, Lehmann M, Metzger C, Grimm S, Boeker H, Boesiger P, Henning A, Seifritz E. (2012). Ketamine decreases resting state functional network connectivity in healthy subjects: implications for antidepressant drug action. PLoS One. 7(9): e44799. Scherk H, Gruber O, Menzel P, Schneider-Axmann T, Kemmer C, Usher J, Reith W, Meyer J, Falkai P. (2009). 5-HTTLPR genotype influences amygdala volume. Eur Arch Psychiatry Clin Neurosci. 259(4): 212–217. Schosser A, Kasper S. (2009). The role of pharmacogenetics in the treatment of depression and anxiety disorders. Int Clin Psychopharmacol. 24(6): 277–288. Schumann G, E. Loth E, T. Banaschewski T, A. Barbot A, G. Barker G, C. Buchel C, P. J. Conrod PJ, J. W. Dalley JW, H. Flor H, J. Gallinat H, H. Garavan H, A. Heinz A, B. Itterman B, … Consortium. (2010). The IMAGEN study: reinforcement-related behaviour in normal brain function and psychopathology. Mol Psychiatry 15(12): 1128–1139. Sheline YI, Barch DM, Donnelly JM, Ollinger JM, Snyder AZ, Mintun MA. (2001). Increased amygdala response to masked emotional faces in depressed subjects resolves with antidepressant treatment: An fMRI study. Biol Psychiatry. 50(9): 651–658. Sheline YI, Barch DM, Price JL, Rundle MM, Vaishnavi SN, Snyder AZ, Mintun MA, Wang S, Coalson RS, Raichle ME. (2009). The default mode network and self-referential processes in depression. Proc Natl Acad Sci U S A. 106(6): 1942–1947. Sheline YI, Gado MH, Kraemer HC. (2003). Untreated depression and hippocampal volume loss. Am J Psychiatry. 160(8): 1516–1518. Sheline YI, Price JL, Yan Z, Mintun MA. (2010). Resting-state functional MRI in depression unmasks increased connectivity between networks via the dorsal nexus. Proc Natl Acad Sci U S A. 107(24): 11020–11025. Shioe K, Ichimiya T, Suhara T, Takano A, Sudo Y, Yasuno F, Hirano M, Shinohara M, Kagami M, Okubo Y, Nankai M, Kanba S. (2003). No association between genotype of the promoter region of serotonin transporter gene and serotonin transporter binding in human brain measured by pet. Synapse. 48(4): 184–188. Sibille E, Lewis DA. (2006). SERT-ainly involved in depression, but when? Am J Psychiatry. 163(1): 8–11.
Simmons AN, Arce E, Lovero KL, Stein MB, Paulus MP. (2009). Subchronic SSRI administration reduces insula response during affective anticipation in healthy volunteers. Int J Neuropsychopharmacol. 12(8): 1009–1020. Smolka MN, Bühler M, Schumann G, Klein S, Hu XZ, Moayer M, Zimmer A, Wrase J, Flor H, Mann K, Braus DF, Goldman D, Heinz A. (2007). Gene-gene effects on central processing of aversive stimuli. Mol Psychiatry. 12(3): 307–317. Soares JM, Sampaio A, Ferreira LM, Santos NC, Marques P, Marques F, Palha JA, Cerqueira JJ, Sousa N. (2013). Stress impact on resting state brain networks. PLoS One. 8(6): e66500. Soliman F, Glatt CE, Bath KG, Levita L, Jones RM, Pattwel SS, Jing D, Tottenham N, Amso D, Somerville LH, Voss HU, Glover G, Ballon DJ, Liston C, Teslovich T, Van Kempen T, Lee FS, Casey BJ. (2010). A genetic variant BDNF polymorphism alters extinction learning in both mouse and human. Science. 327(5967): 863–866. Stjepanovic D, Lorenzetti V, Yucel M, Hawi Z, Bellgrove MA. (2013). Human amygdala volume is predicted by common DNA variation in the stathmin and serotonin transporter genes. Transl Psychiatry. 3: e283. Stone JM, Dietrich C, Edden R, Mehta MA, De Simoni S, Reed LJ, Krystal JH, Nutt D, Barker GJ. (2012). Ketamine effects on brain GABA and glutamate levels with 1H-MRS: relationship to ketamine-induced psychopathology. Mol Psychiatry. 17(7): 664–665. Stoy M, Schlagenhauf F, Sterzer P, Bermpohl F, Hagele C, Suchotzki K, Schmack K, Wrase J, Ricken R, Knutson B, Adli M, Bauer M, Heinz A, Strohle A. (2012). Hyporeactivity of ventral striatum towards incentive stimuli in unmedicated depressed patients normalizes after treatment with escitalopram. J Psychopharmacol. 26(5): 677–688. Sublette ME, Baca-Garcia E, Parsey RV, Oquendo MA, Rodrigues SM, Galfalvy H, Huang YY, Arango V, Mann JJ. (2008). Effect of BDNF val66met polymorphism on age-related amygdala volume changes in healthy subjects. Prog Neuro-Psychopharmacol Biol Psychiatry. 32(7): 1652–1655. Takahashi H, Yahata N, Koeda M, Takano A, Asai K, Suhara T, Okubo Y. (2005). Effects of dopaminergic and serotonergic manipulation on emotional processing: a pharmacological fMRI study. Neuroimage. 27(4): 991–1001. Tansey KE, Guipponi M, Hu X, Domenici E, Lewis G,Malafosse A, Wendland JR, Lewis CM, McGuffin P, Uher R. (2013). Contribution of common genetic variants to antidepressant response. Biol Psychiatry. 73(7): 679–682. Tansey KE, Guipponi M, Perroud N, Bondolfi G, Domenici E, Evans D, Hall SK, Hauser J, Henigsberg N, Hu X, Jerman B, Maier W, Mors O, O’Donovan M, … Uher R. (2012). Genetic predictors of response to serotonergic and noradrenergic antidepressants in major depressive disorder: a genome-wide analysis of individual-level data and a meta-analysis. PLoS Med. 9(10): e1001326. Tao R, Calley CS, Hart J, Mayes TL, Nakonezny PA, Lu H, Kennard BD, Tamminga CA, Emslie GJ. (2012). Brain activity in adolescent major depressive disorder before and after fluoxetine treatment. Am J Psychiatry. 169(4): 381–388. Taylor MJ, Freemantle N, Geddes JR, Bhagwagar Z. (2006). Early onset of selective serotonin reuptake inhibitor antidepressant action: systematic review and meta-analysis. Arch Gen Psychiatry. 63(11): 1217–1223. Taylor MJ, Sen S, Bhagwagar Z. (2010). Antidepressant response and the serotonin transporter gene-linked polymorphic region. Biol Psychiatry. 68(6): 536–543. Tendolkar I, van Wingen G, Urner M, Verkes RJ, Fernandez G. (2011). Short-term duloxetine administration affects neural correlates of mood-congruent memory. Neuropsychopharmacology. 36(11): 2266–2275. Tost H, Alam T, Geramita M, Rebsch C, Kolachana B, Dickinson D, Verchinski BA, Lemaitre H, Barnett AS, Trampush JW, Weinberger
cha p ter 6 : V ariability of A ntide p ressant D rug R es p onse
99
DR, Marenco S. (2013). Effects of the BDNF Val66Met polymorphism on white matter microstructure in healthy adults. Neuropsychopharmacology. 38(3): 525–532. Trivedi MH, Rush AJ, Wisniewski SR, Nierenberg AA, Warden D, Ritz L, Norquist G, Howland RH, Lebowitz B, McGrath PJ, Shores-Wilson K, Biggs MM, Balasubramani GK, aFava M. (2006). Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: Implications for clinical practice. Am J Psychiatry. 163(1): 28–40. Tsankova NM, Berton O, Renthal W, Kumar A, Neve RL, Nestler EJ. (2006). Sustained hippocampal chromatin regulation in a mouse model of depression and antidepressant action. Nat Neurosci. 9(4): 519–525. Valenzuela M, Bartres-Faz D, Beg F, Fornito A, Merlo-Pich E, Muller U, Ongur D, Toga AW, anYucel M. (2011). Neuroimaging as endpoints in clinical trials: are we there yet? Perspective from the first Provence workshop. Mol Psychiatry. 16(11): 1064–1066. van de Ven V, Wingen M, Kuypers KP, Ramaekers JG, Formisano E. (2013). Escitalopram decreases cross-regional functional connectivity within the default-mode network. PLoS One. 8(6): e68355. Van Dyck CH, Malison RT, Staley JK, Jacobsen LK, Seibyl JP, Laruelle M, Baldwin RM, Innis RB, Gelernter J. (2004). Central serotonin transporter availability measured with [123I]β-CIT SPECT in relation to serotonin transporter genotype. Am J Psychiatry. 161(3): 525–531. van Marle HJ, Tendolkar I, Urner M, Verkes RJ, Fernandez G, van Wingen G. (2011). Subchronic duloxetine administration alters the extended amygdala circuitry in healthy individuals. Neuroimage. 55(2): 825–831. Verhagen M, Van Der Meij A, Van Deurzen PM, Janzing J, Arias-Vásquez A, Buitelaar J, Franke B. (2010). Meta-analysis of the BDNF Val66Met polymorphism in major depressive disorder: effects of gender and ethnicity. Mol Psychiatry. 15(3): 260–271. Victor TA, Furey ML, Fromm SJ, Ohman A, Drevets WC. (2010). Relationship between amygdala responses to masked faces and mood state and treatment in major depressive disorder. Arch Gen Psychiatry. 67(11): 1128–1138. Viviani R, Sim EJ, Lo H, Beschoner P, Osterfeld N, Maier C, Seeringer A, Godoy AL, Rosa A, Comas D, Kirchheiner J. (2010). Baseline brain perfusion and the serotonin transporter promoter polymorphism. Biol Psychiatry. 67(4): 317–322. Vollenweider FX, Kometer M. (2010). The neurobiology of psychedelic drugs: implications for the treatment of mood disorders. Nat Rev Neurosci. 11(9): 642–651. Vreeburg SA, Hoogendijk WJ, van Pelt J, Derijk RH, Verhagen JC, van Dyck R, Smit JH, Zitman FG, Penninx BW. (2009). Major depressive disorder and hypothalamic-pituitary-adrenal axis activity: results from a large cohort study. Arch Gen Psychiatry. 66(6): 617–626. Vul E, Pashler H. (2012). Voodoo and circularity errors. Neuroimage. 62(2): 945–948. Vythilingam M, Vermetten E, Anderson GM, Luckenbaugh D, Anderson ER, Snow J, Staib LH, Charney DS, Bremner JD.
(2004). Hippocampal volume, memory, and cortisol status in major depressive disorder: effects of treatment. Biol Psychiatry 56(2): 101–112. Walsh ND, Dalgleish T, Dunn VJ, Abbott R, St Clair MC, Owens M, Fairchild G, Kerslake WS, Hiscox LV, Passamonti L, Ewbank M, Ban M, Calder AJ, Goodyer IM. (2012). 5-HTTLPR-environment interplay and its effects on neural reactivity in adolescents. Neuroimage. 63(3): 1670–1680. Walter M, Henning A, Grimm S, Schulte RF, Beck J, Dydak U, Schnepf B, Boeker H, Boesiger P, Northoff G. (2009). The relationship between aberrant neuronal activation in the pregenual anterior cingulate, altered glutamatergic metabolism, and anhedonia in major depression. Arch Gen Psychiatry. 66(5): 478–486. Wang JW, Dranovsky A, Hen R. (2008). The when and where of BDNF and the antidepressant response. Biol Psychiatry. 63(7): 640–641. Weissenbacher A, Kasess C, Gerstl F, Lanzenberger R, Moser E, Windischberger C. (2009). Correlations and anticorrelations in resting-state functional connectivity MRI: a quantitative comparison of preprocessing strategies. Neuroimage. 47(4): 1408–1416. Weissman DH, Roberts KC, Visscher KM, Woldorff MG. (2006). The neural bases of momentary lapses in attention. Nat Neurosci. 9(7): 971–978. Whiteford HA, Degenhardt L, Rehm J, Baxter AJ, Ferrari AJ, Erskine HE, Charlson FJ, Norman RE, Flaxman AD, Johns N, Burstein R, Murray CJ, Vos T. (2013). Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010. Lancet. 382(9904): 1775–1786. Williams LM, Gatt JM, Schofield PR, Olivieri G, Peduto A, Gordon E. (2009). ‘Negativity bias’ in risk for depression and anxiety: brain-body fear circuitry correlates, 5-HTT-LPR and early life stress. NeuroImage. 47(3): 804–814. Windischberger C, Lanzenberger R, Holik A, Spindelegger C, Stein P, Moser U, Gerstl F, Fink M, Moser E, Kasper S. (2010). Area-specific modulation of neural activation comparing escitalopram and citalopram revealed by pharmaco-fMRI: a randomized cross-over study. NeuroImage. 49(2): 1161–1170. Winkler AM, Kochunov P, Blangero J, Almasy L, Zilles K, Fox PT, Duggirala R, Glahn DC. (2010). Cortical thickness or grey matter volume? The importance of selecting the phenotype for imaging genetics studies. Neuroimage. 53(3): 1135–1146. Wong DF, Tauscher J. Grunder G. (2009). The role of imaging in proof of concept for CNS drug discovery and development. Neuropsychopharmacology. 34(1): 187–203. Zhu X, Wang X, Xiao J, Liao J, Zhong M, Wang W, Yao S. (2012). Evidence of a dissociation pattern in resting-state default mode network connectivity in first-episode, treatment-naive major depression patients. Biol Psychiatry. 71(7): 611–617. Ziegler E, Foret A, Mascetti L, Muto V, Le Bourdiec-Shaffii A, Stender J, Balteau E, Dideberg V, Bours V, Maquet P, Phillips C. (2013). Altered white matter architecture in BDNF met carriers. PLoS One. 8(7): e69290.
1 0 0 Part I I : I maging G enetics and D rug D iscovery
7. IMAGING GENE TICS OF PHARMACOLOGICAL RESPONSE IN PSYCHIATRIC DISORDERS Philip R. Szeszko and Anil K. Malhotra
I N T RO DU C T I O N Psychiatric disorders represent the leading cause of years lost because of disability in young people aged 10–24 years (Gore et al. 2011), and epidemiological data indicate that there is a significant rising burden associated with mental and behavioral disorders (Murray and Lopez 2013). Thus, there is an increasing need for information that can inform treatment, especially that which can be tailored to patients at the individual level. The current approach for assessing the efficacy of psychotropic medications is the clinical trial. Unfortunately, however, samples used in clinical trials are notoriously heterogeneous, and the endpoints used to define treatment response are admittedly arbitrary, with lack of consistency across studies. Moreover, the use of large-scale randomized clinical trials and meta-analyses to summarize treatment results are often used to inform treatment decisions, but it is widely acknowledged that there is a significant loss of information at the individual (or even subgroup) level that could potentially be valuable for informing treatment decisions (Gerretsen et al. 2009). The identification of psychiatric patients who respond to treatment versus those who do not is a high priority for clinicians given that treatment-resistant patients are at increased risk for severe functional disability, medical morbidity, mortality, and suicide. For example, approximately 30% of patients with schizophrenia fail to respond to at least two adequate trials of antipsychotic medications and are thus considered treatment-resistant (Meltzer 1997; Kane et al. 1988). Given that some of the risks associated with pharmacotherapy can be significant, it would be extremely beneficial for clinicians to quickly and accurately differentiate responders from non-responders for resource allocation and to limit adverse events (Rosenheck
et al. 1999a, 1999b). The identification of treatment outcome biomarkers is thus a critical priority for translational psychiatry so that clinicians can begin to benefit from neuroimaging studies to directly influence patient care. However, pharmacogenetics studies within psychiatry have not yielded compelling results to date and thus the utilization of pharmacogenetics within clinical psychiatry has been extremely limited (Malhotra et al. 2012; Zhang and Malhotra 2013). Moreover, despite decades of neuroimaging studies demonstrating anatomic pathology in patients with psychiatric disorders, there continue to be tremendous gaps between this work and its translational significance and relevance for the treating clinician. The use of intermediate imaging phenotypes, which permit in vivo assessment of neuronal function, may help clarify the relationship between genetic variation and treatment response to psychotropic drugs (Blasi and Bertolino 2006). The investigation and use of structural and functional brain imaging and the concomitant identification of how these networks relate to response are quickly becoming critical considerations in drug discovery. The identification of specific gene-endophenotype linkages can improve our ability to explain individual variability at the candidate gene level, and some functional imaging modalities can be used to demonstrate “proof of concept” principles in drug design trials (Wise and Tracey 2006) and biomarkers in drug discovery (Wiedemann 2011). Few studies, however, have been able to assess the interaction of genotype and drug effects within the context of pharmacogenetic designs (Falcone et al. 2013). The combination of imaging and genetics for a better understanding of treatment response, targeting a specific network in the brain, could be a valuable research direction, but requires extensive knowledge regarding how specific psychotropic medications interact with a 101
given brain network to clarify the underlying mechanisms that contribute to these effects. In this chapter we review studies that have assessed the relationship between brain imaging and genetic variation related to pharmacological response in psychiatric disorders including schizophrenia, depression, bipolar disorder, and attention deficit hyperactivity disorder. To date the majority of studies have utilized magnetic resonance imaging (MRI), which may be ideally suited for integration with pharmacologic studies given the lack of ionizing radiation and the ability to scan patients repeatedly in the context of a clinical trial. Moreover, we argue that MRI offers the greatest potential for integration into clinical practice given the widespread availability of magnets in medical centers throughout the country. SCHIZOPHRENIA For several decades, the dysfunction of dopaminergic systems has been widely regarded as one of the leading theories contributing to schizophrenia neurobiology (Seeman 2010, 2011). In this context the use of imaging genetics could be utilized to clarify the relationship between variation in genes that play a role in dopamine processing and antipsychotic treatment response (Di Giorgio et al. 2009). Moreover, a better understanding of imaging genetics relationships could also serve to clarify the underlying mechanisms of cognitive dysfunction in schizophrenia that could serve as treatment targets for novel pharmacologic agents (Roffman et al. 2006). As an example of this approach, we describe a common functional polymorphism (Val[108/158]Met) within the catechol-O-methyltransferase (COMT) gene that has been linked to variability in enzyme activity and dopamine catabolism in humans that affects dopamine functioning (Lotta et al 1995; Weinshilboum et al. 1999). In seminal work, Egan and colleagues (2001) investigated the COMT genotype in relationship to performance on the Wisconsin Card Sorting Test (WCST) in 175 patients with schizophrenia, 219 unaffected siblings, and 55 controls. They reported that the COMT genotype was associated with WCST performance in an allele dosage manner and explained approximately 4% of the variance in perseverative errors. Moreover, using functional magnetic resonance imaging (fMRI), these authors demonstrated an effect of the COMT genotype on prefrontal functioning in three separate cohorts (ranging in size from 11 to 16 subjects) such that methionine (Met) carriers had a more “efficient” prefrontal cortical response. Other work suggests an additive linear effect
of the COMT genotype when investigating performance on the N-back task such that Val/Val homozygotes had the lowest performance and Met/Met homozygotes had the best performance (Goldberg et al. 2003). Moreover, the COMT Val(108/158)Met polymorphism predicted working memory performance in relationship to antipsychotic medications. Notably, patients homozygous for the COMT Met allele had significant performance improvements in working memory following treatment, in contrast to Val homozygotes, who did not demonstrate this effect. Taken together, these results indicate that the Val/Val genotype encodes for the high-activity enzyme and may be associated with lower dopamine concentration and concomitant worse performance compared to Met carriers. A logical extension of this work can test whether a given pharmacologic intervention for treating prefrontal dysfunction based on knowledge of an individual’s COMT genotype could be useful in treating schizophrenia (Apud and Weinberger 2007). Unfortunately, few studies in the literature have addressed this question in particular. Bertolino et al. (2004) investigated the relationship between COMT genotype and symptoms, working memory, and prefrontal cortical physiology in patients with schizophrenia treated with olanzapine. In their study, 30 patients with schizophrenia were treated with olanzapine for 8 weeks, and 20 patients completed fMRI studies while performing an N-back working memory test at 4 and 8 weeks. Their results indicated that Met allele load within the COMT gene predicted changes in working memory functioning and prefrontal cortical physiology following 8 weeks of olanzapine treatment. The authors observed a comparable effect for negative symptoms using the Positive and Negative Syndrome Scale (PANSS). Additional studies investigating the interaction of genetic variation in COMT and other antipsychotic agents that have different pharmacologic properties could enhance our understanding of prefrontal cortical dysfunction in schizophrenia and could provide novel targets for cognitive intervention. Another gene that has been implicated in dopamine functioning, which may be relevant to treatment response in schizophrenia and could yield important information in imaging genetics studies, is DRD2, which codes for the D2 dopamine receptor. In a meta-analysis, Zhang et al. (2010) investigated six studies that reported significant findings for the -141C Ins/Del polymorphism in a total sample size of 687 patients. They report that individuals who carried the deletion (Del) had poorer antipsychotic drug response compared to the Ins/Ins genotype. Eight studies assessed the Taq1A polymorphism and the antipsychotic response, yielding a total sample size of 748 patients. There were,
1 0 2 Part I I : I maging G enetics and D rug D iscovery
however, no significant differences in the response rate among A1 allele carriers compared to individuals with the A2/A2 genotype, or A2 allele carriers compared to individuals with the A1/A1 genotype. These findings thus suggest that variation within the DRD2 gene may play a role in antipsychotic treatment response, and imaging studies designed to investigate the relationship between genetic variation in this polymorphism and brain functioning could be a fruitful area for further investigation. In a double-blind crossover study, Kirsch et al. (2006) investigated the relationship between the DRD2 Taq1A polymorphism and the selective D2 receptor agonist bromocriptine on reward circuitry using fMRI in 24 healthy subjects. It is known that the DRD2 Taq1A polymorphism is linked to variability in the amount of dopamine D2 receptor density and is associated with reward regulation such that individuals carrying the A1 allele have a lower D2 receptor density (Ritchie and Noble 2003) and may be at increased risk of substance use (Conner et al. 2005; Esposito-Smythers et al. 2009). Kirsch et al. (2006) report an increase in reward system activation from placebo to bromocriptine only among individuals who were carrying the A1 allele and a concomitant performance increase when receiving bromocriptine. In another fMRI study investigating reversal learning, Cohen and colleagues (2007) scanned 22 subjects during placebo and once on the D2 receptor agonist cabergoline. Administration of cabergoline was associated with greater reward responsivity in the cingulate, striatum, and orbital frontal regions among A1+ individuals, but reductions in reward responsivity in these regions were observed among A1– subjects. Moreover, these authors observed reductions in task performance and frontostriatal connectivity in A1+ subjects (with the opposite effect in A1– subjects) that were consistent with an inverted-U pattern. Moreover, these authors reported that the effect of cabergoline on functional connectivity significantly predicted feedback-guided learning. Taken together, the studies by Kirsch et al. (2006) and Cohen et al. (2007) suggest that genetic variation in DRD2 Taq1A may be associated with reward responsiveness, which is highly relevant to the negative symptoms associated with schizophrenia. In a novel study, Blasi et al. (2011) investigated the epistatic relationship between the D2 and AKT1 genes in the D2/AKT1/GSK-3β signaling pathway, which have been implicated in dopamine processing and associated brain activity in schizophrenia, along with response to antipsychotics. Specifically, these authors measured the proteins AKT1 and GSK-3β and investigated fMRI response within the cingulate cortex during attentional control. In this study, 73 healthy individuals underwent fMRI scanning
while performing the visual attention control task, which requires an increasing amount of attention under different levels of control. Findings indicated that there were significant interactions between DRD2 genotype and load, and between AKT1 genotype and load for cingulate activity. Interestingly, these authors also reported an interaction between DRD2 rs1076560, AKT1 rs1130233, and load for the cingulate. Post hoc investigation of these differences indicated that cingulate responses at the higher load were greater than those at the intermediate attentional level across groups, but that DRD2 GT/AKT1 A carriers displayed lower activity at the highest load. Moreover, they reported that between-group differences in cingulate activity were also evident at the higher attentional load when comparing DRD2 GT/AKT1 A carriers versus DRD2 GT/ AKT1 GG and DRD2 GG/AKT1 GG subjects. Also, they report that the interaction of the two alleles was associated with greater improvement of PANSS scores in patients following treatment with olanzapine. These findings thus highlight the importance of epistasis in studying imaging genetics of pharmacologic response to better understand dysfunctional pathways in schizophrenia. DE P R E S S I O N Leading theories of depression have focused on abnormalities within the serotonin system that are strongly linked to behavioral and cognitive dysfunction (Willner, 1985). Approximately 60% of individuals with depression do not respond completely to antidepressant pharmacotherapy, and genetic factors may account for up to 50% of an individual’s ability to respond (Crisafulli et al. 2011). The serotonin transporter–linked polymorphic region (5-HTTLPR) and variants within the 5-hydroxytryptamine receptors (5-HT1A, 5-HT2A, 5-HT3A, 5-HT3B, and 5-HT6) have been highlighted as critical targets for understanding treatment response in depression. Clarifying how these genes are associated with treatment response could have significant translational impact such that imaging-genetics studies may provide novel information for better understanding the variability in antidepressant drug response in the context of the neurobiology of depression (Rabl et al. 2010). The majority of imaging genetics work in depression has focused on polymorphisms within serotonin genes linked to treatment response. Prior work by Fakra et al. (2009) indicates a significant effect of a common functional variation (C[-1019]G) in the human 5-HT(1A) gene (HTR1A) related to significantly lower threat-related amygdala reactivity. This polymorphism thus influences the neural
cha p ter 7 : I maging G enetics of Pharmacological R es p onse in Psychiatric D isorders
103
circuitry responsible for emotional behavioral processes and could be a future treatment target for anxiety and/ or depression. Moreover, these investigators reported that this effect was independent of another common functional polymorphism that has been linked to serotonin signaling, 5-HTTLPR. These authors reported that overall, HTR1A C(-1019)G predicted approximately 9% of interindividual variability in trait anxiety through its mediation of amygdala reactivity, thus providing a potential target for therapeutic agents. Abnormalities within the immune system have been hypothesized to play an important role in the pathogenesis of depression, with converging evidence for genetic mechanisms that contribute to these effects (Bufalino et al. 2013). In one study, Baune et al. (2010) investigated the relationship between genetic variation in the interleukin 1 beta gene (IL1B) and amygdala and anterior cingulate responsiveness during the presentation of emotional stimuli in the context of clarifying antidepressant treatment response. These investigators genotyped 256 Caucasian patients with depression, focusing on the rs16944, rs1143643, and rs1143634 polymorphisms within the IL1B gene. A subgroup of 32 depressed patients received fMRI while undergoing an emotional face paradigm. Overall, results indicated significant relationships between the GG genotypes of single nucleotide polymorphism (SNP) rs16944 and rs1143643 with non-remission after approximately 6 weeks of treatment. Moreover, it is also noteworthy that these authors reported that the number of G-alleles in both SNPs was linked to lower responsiveness in both the anterior cingulate and amygdala during emotional stimulation. Two studies by Domschke and colleagues highlight genetic variation in depression in relationship to treatment response and resistance that may be associated with abnormalities in the normal brain response to socially rewarding stimuli using a face paradigm. In a study exploring individuals with “anxious depression,” Domschke et al (2010) reported that the less active neuropeptide Y (NPY) rs16147 -399C allele was associated with slower response following 2 weeks of antidepressant treatment and a failure to achieve remission following 4 weeks of treatment. In complementary functional neuroimaging studies, these authors reported that genetic variation within the rs16147C allele was associated with stronger bilateral amygdala activation when visualizing threatening faces in an allele-dose manner. This study thus highlights the effects of genetic variation within NPY on antidepressant treatment response among individuals with “anxious depression,” and points to a neurobiological circuit that may mediate this effect. Similarly, Domschke et al. (2008) examined whether
variability in rs1049353 and rs12720071 within the cannabinoid receptor 1 gene (CNR1) was also linked to antidepressant treatment response. Their findings indicate that variation within CNR1 rs1049353 was associated with greater risk of treatment resistance to antidepressants and that variation within this same SNP was linked to lower amygdalae and striatum (putamen and globus pallidus) blood-oxygen-level dependent (BOLD) activation during the presentation of masked happy faces. Another study examined the relationship between treatment response in depression and variation within genes that have been implicated in serotonin functioning. Ruhé et al. (2009) examined SERT occupancy with iodine-123-labeled 2beta-carbomethoxy-3beta-(4-iodophenyl)-tropane ([123I] beta-CIT) single-photon emission computed tomography (SPECT) at baseline and following 6 weeks of paroxetine treatment in drug-naïve or drug-free depressed outpatients genotyped for the 5-HTTLPR SNP. The authors reported a significant positive relationship between SERT occupancy and clinical response, which was defined as percent reduction on the 17-item Hamilton Depression Rating Scale, but only among individuals who had the L(A)/L(A) genotype. These findings were interpreted to suggest that individuals who are L(A)/L(A) carriers may manifest a more “dynamic” serotonin system, which could portend a better response to serotonin reuptake inhibitors. Prior work has strongly implicated genetic variation within the Val66Met polymorphism of the brain-derived neurotrophic factor (BDNF) gene in brain plasticity and neurogenesis, especially within the hippocampus (Szeszko et al. 2005). In a study assessing depression, Gonul et al. (2011) reported that the Val/Val genotype may represent a vulnerability marker in major depression for hippocampal volume loss given its association to illness duration. Although a recent meta-analysis did not find any evidence of an association between BDNF and risk of major depression (Gyekis et al. 2013), another meta-analysis did indeed report that the BDNF Met allele was associated with greater risk of depression among older individuals (Pei et al. 2012). Moreover, this polymorphism has also been linked to treatment response and resistance. For example, in a comprehensive meta-analysis of pharmacogenetics studies in major depression, Niitsu and colleagues (2013) reported that the best SNP associated with antidepressant response was BDNF Val66Met and that it had a selective effect, in particular, on serotonin reuptake inhibitor treatment. Two studies investigated the relationship between the BDNF Val66Met polymorphism and treatment response/resistance in major depression using
1 0 4 Part I I : I maging G enetics and D rug D iscovery
neuroimaging. Cardoner and colleagues (2013) investigated the relationship between this polymorphism and abnormalities within brain regions (including the hippocampus) hypothesized to play a role in the neurobiology of depression and their relationship to treatment outcome in 37 patients. In their study, they observed a significant inverse relationship between left hippocampal volume and days to remission among Val66 homozygotes. They further report that right orbitofrontal volume was inversely associated with remission status among Met66 carriers. In another study, Alexopoulos and colleagues (2010) examined genetic variation within this polymorphism and remission rate in geriatric depression and whether this relationship could be associated with white matter microstructural abnormalities as inferred from diffusion tensor imaging. Geriatric patients with major depression were assigned to a 2-week placebo period and individuals with a Hamilton Depression Rating Scale (HDRS) of 18 or greater received escitalopram at a dosage of 10 mg daily for 12 weeks. Findings indicated that BDNF (Met) carriers were more likely to remit compared to BDNF (Val/Val) homozygotes following treatment and that white matter abnormalities in the inferior longitudinal fasciculum and corpus callosum were associated with lower remission. The interaction between BDNF (Val66Met) status and white matter disruptions predicting remission was not significant, however. BI P O L A R D I S O RDE R Comparatively little work has investigated the relationship between brain imaging and pharmacological response in bipolar disorder. Of all the treatments currently available, lithium remains the gold standard for bipolar disorder and may be one of the most logical choices to investigate using imaging genetics paradigms. In a comprehensive literature review, Ikeda and Kato (2003) identified several biological predictors of lithium response that could be incorporated into MRI spectroscopy studies, including brain lithium concentration, N-acetyl-aspartate (NAA), and myo-inositol. Moreover, Steen et al. (1998) have proposed that the enzyme inositol polyphosphate 1-phosphatase (INPP1) may be a putative target associated with successful lithium treatment. In their study, they reported preliminary evidence that the C973A transversion was evident in 6/9 lithium responders compared to only 1/9 non-responders, although findings were not evident within an independent sample.
The protein kinase glycogen synthase kinase-3 (GSK-3) is a serine/threonine kinase and its phosphorylation can inhibit activity in downstream regions. Prior pharmacological and behavioral animal studies suggest that lithium selectively inhibits GSK-3 and thus is associated with mood stabilization (Gould and Manji 2005), although the precise mechanisms regarding this effect are unknown at this time. GSK-3 may play a role as a therapeutic agent in the reduction of autoimmune neuroinflammation (Beurel et al. 2010; Rowse et al. 2012). In one study, Benedetti et al. (2013) investigated the relationship between GSK3-Beta promoter gene variants and white matter integrity as assessed using diffusion tensor imaging. In that study, the authors investigated the effects of lithium treatment on white matter integrity, the GSK3-B promoter rs334558 polymorphism, and white matter microstructure. Their findings indicated that the less active GSK3-B rs334558*C gene promoter variant, coupled with long-term administration of the GSK3-B inhibitor lithium, were associated with increased axial diffusivity in several white matter tracts of patients with bipolar disorder. Given that axial diffusivity may reflect properties of myelineation, these findings point to future studies that could address whether patients with the less active gene variant are more or less likely to respond to lithium pharmacotherapy. Abnormalities in the regulation of intracellular calcium (Ca(2+)) play a critical role in neuronal plasticity through gene expression and energy metabolism (Kawamoto et al. 2012) and have been reported to play a role in the neurobiology of bipolar disorder (Bhat et al. 2012; Solis-Chagoyán et al. 2013). Moreover, key risk SNPs in the CACNA1C gene identified in whole genome association studies in bipolar disorder have been linked to brain morphology (Kempton et al. 2009) and function (Bigos et al. 2010) among healthy individuals. These calcium signaling mechanisms may also be influenced by a key brain protein, B-cell lymphoma 2 (Bcl-2) (Bonneau et al. 2013). Machado-Vieira et al. (2011) used live cell fluorescence imaging to examine effects on intracellular Ca(2+) dynamics in lymphoblasts of 18 patients with bipolar disorder carrying the AA, AG, or GG variants of the Bcl-2 gene SNP rs956572. Their findings indicate that, compared to GG homozygotes, the variant AA expresses less BCL-2 messenger RNA (mRNA) and protein-exhibited elevated basal cytosolic Ca(2+). Notably, the abnormal behavior observed in the AA cells was reversed by lithium administration. These findings thus highlight an important role for abnormal Bcl-2 gene expression among individuals with the AA variant that is associated with abnormalities in Ca(2+) homeostasis.
cha p ter 7 : I maging G enetics of Pharmacological R es p onse in Psychiatric D isorders
105
AT T E N T I O N DE F I C I T H Y P E R AC T I V I T Y DI S O RDE R Considerable evidence supports the role of dopamine dysfunction in the neurobiology of attention deficit hyperactivity disorder (ADHD) (del Campo et al. 2011). Primary treatments for the disorder include methylphenidate, which is a potent blocker of the dopamine transporter, which when administered at low doses serves to increase attention and improve executive functioning among both healthy individuals and those with ADHD (Arnsten 2006). A meta-analysis by Yang et al. (2007) indicated that the 10-repeat allele of a variable number tandem repeat (VNTR) polymorphism in the dopamine transporter gene (SLC6A3) reportedly had a small, but statistically significant, role in genetic susceptibility to ADHD. Moreover, some data suggest that homozygosity of the 10-repeat allele within the dopamine transporter gene is linked to worse methylphenidate response in children with ADHD (see Wallis 2010 for a review). Using a pharmacogenetic approach, Kooij et al. (2008) investigated methylphenidate response in a double-blind study and dopamine transporter genetic polymorphisms. Their findings indicated that the VNTR in the 3' untranslated region significantly predicted a greater likelihood of methylphenidate response among individuals with a single 10-repeat allele compared to patients with the 10/10 genotype. In contrast, however, it should be acknowledged that a recent review by Contini et al. (2013) reported that a majority of findings from studies focusing on response to methylphenidate have been negative. Cheon and colleagues (2005) investigated the association between a 10-repeat allele within the dopamine transporter and methylphenidate response among children with ADHD using I-123-N-(3-iodopropen-2-yl)2beta-carbomethoxy-3beta-(4-chlorophenyl)tropane [123I]IPT single photon emission computed tomography (SPECT). They recruited 11 drug-naïve children with ADHD who were treated for approximately 8 weeks with methylphenidate. The goal of the study was to compare dopamine transporter (DAT) density between ADHD children with and without homozygosity for the 10-repeat allele at the dopamine transporter as well as its association with methylphenidate response. They reported that there was significantly greater DAT density in the basal ganglia among ADHD children who had the 10/10 genotype (n = 7) compared to children without the 10/10 genotype (n = 4). Moreover, they report that all ADHD subjects without the 10/10 genotype demonstrated a good response to methylphenidate treatment, in contrast to ADHD
subjects with the 10/10 genotype, where response was only 29%. Despite the caveat of a small sample size, the study highlights an association between DAT density in brain regions implicated in the pathogenesis of ADHD and genetic variation within the dopamine transporter gene. More recently, Szobot and colleagues (2011) investigated whether previously reported risk alleles within the SLC6A3 and DRD4 genes could predict striatal dopamine changes following treatment with methylphenidate in adolescents with ADHD and comorbid substance abuse. The adolescents received a SPECT scan using [Tc(99m)] TRODAT-1 at baseline and then again 3 weeks later, after receiving methylphenidate, to assess changes within striatal (caudate and putamen) dopamine transporter binding potential that were investigated specifically in relationship to genotype. These investigators found that homozygosity for the dopamine transporter 10-repeat allele in combination with the DRD4 7-repeat allele (7R) significantly predicated reductions in DAT following treatment with methylphenidate in the striatum, even after consideration of possible confounds. This study thus highlights how genetic variation regarding the treatment of ADHD can be informed by clarifying brain mechanisms in how response might occur. F U T U R E DI R EC T I O N S Future studies investigating imaging genetics of treatment response in psychiatric disorders could focus on several potentially important issues. One consideration relates to how genetic factors play a role in the ability of psychotropic medications to cross the blood-brain barrier (Bai 2010). For example, some data suggest that genetic variation within P-glycoprotein (P-gp) could play a role in the penetration of psychotropic medications to the brain. In this regard, Wang and colleagues (2009) investigated mice deficient in the P-gp gene (Abcb1a/b-/-) that were dosed intraperitoneally with aripiprazole. These investigators reported that although deficiency of P-gp was not significantly associated with plasma drug concentrations, there were dramatic effects on drug concentrations within brain tissue. It is therefore conceivable that factors influencing P-gp activity within the blood-brain barrier in humans may have implications for the therapeutic efficacy and tolerability of psychotropic medications, including side effect profiles. There are limited data, however, that focus on the relationship between imaging and genotype to monitor drug uptake within the brain, with no clear, consistent evidence reported to date. Thus, research is critically needed to examine how
1 0 6 Part I I : I maging G enetics and D rug D iscovery
pharmacogenetics predict individual outcomes by better understanding these transporter mechanisms. An example of this approach was described by Bankstahl et al. (2011), who developed a novel positron emission tomography (PET) imaging protocol to identify overactivity of P-glycoprotein in the blood-brain barrier that was associated with seizure activity, with the goal of detecting seizure-induced regional changes in P-glycoprotein activity to directly inform pharmacoresistance in epilepsy. An additional goal of future research relates to better understanding how genetic variation in drug metabolism is associated with efficacy and the side effect profile of psychotropic medications. In particular, genetic variation within CYP2D6 has been associated with psychotropic drug metabolism and thus could be an important consideration in understanding imaging genetics relationships. Unfortunately, few studies have prospectively examined the relationship between CYP2D6 genetic variation and its relationship to clinical efficacy in the management of psychiatric patients such as those diagnosed with schizophrenia (Ravyn et al. 2013). Moreover, of the studies conducted to date, few have reported a significant relationship between genotype and clinical efficacy, although the majority of prior work has been limited by small samples, retrospective designs, and low statistical power. A potentially fruitful area, however, is the investigation of the CYP2D6 genetic variation and side effects such as tardive dyskinesia (Lee and Kang 2011). In this regard it is noteworthy that the FDA has already revised labeling for pimozide to provide clinicians with genotype testing recommendations to enable greater therapeutic efficacy at the individual level (Rogers et al. 2012). Brain imaging has also been used to understand how the CYP2D6 genotype is associated with neural activity during functional brain imaging of cognitive tasks in a large cohort of healthy drug-free humans (Stingl et al. 2012). Similarly, Kirchheiner et al. (2011) reported that compared to extensive metabolizers, poor metabolizers had higher perfusion levels in the thalamus and that in exploratory whole brain analyses there was evidence for involvement of CYP2D6 in regions associated with alertness and/or serotonergic functioning. SUMMARY AND CONCLUSIONS In this chapter we have reviewed studies that have used imaging genetics to investigate pharmacological response in patients with psychiatric disorders including schizophrenia, depression, bipolar disorder and attention deficit hyperactivity disorder. Notably, there is a paucity of
research utilizing this strategy in some psychiatric disorders such as obsessive-compulsive disorder. Impediments to the field of imaging genetics of treatment response include the lack of well-defined outcomes in many studies. For example, there are significant problems associated with the definition of treatment response and thus the identification of patients who are treatment resistant may ultimately be more informative and may reduce heterogeneity in identifying outcomes in response to psychotropic medications. Although the identification of genotype can provide important information regarding outcome, it is widely acknowledged that optimal treatment will ultimately not be associated with the identification of a single gene, but rather with abnormalities within complex brain networks that can only be identified using brain imaging across multiple neuroimaging modalities. The identification of these networks may ultimately inform novel compound development that target dysfunction within circuits across disorders identified in the fifth edition of the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (DSM-5). The integration of pharmacogenetics with brain imaging thus offers the potential to inform treatment strategies at the individual level and will provide new opportunities for the future of personalized medicine. REFERENCES Alexopoulos GS, Glatt CE, Hoptman MJ, Kanellopoulos D, Murphy CF, Kelly RE Jr, Morimoto SS, Lim KO, Gunning FM. (2010). BDNF Val66met polymorphism, white matter abnormalities and remission of geriatric depression. Journal of Affective Disorders, 125(1), 262–268. Apud JA, Weinberger DR. (2007). Treatment of cognitive deficits associated with schizophrenia. CNS Drugs, 21(7), 535–557. Arnsten AF. (2006). Stimulants: therapeutic actions in ADHD. Neuropsychopharmacology, 31(11), 2376–2383. Bai JP. (2010). Drug delivery to the brain: considerations of genetic polymorphisms of blood-brain barrier transporters and imaging technologies. Therapeutic Delivery, 1(4), 563–570. Bankstahl JP, Bankstahl M, Kuntner C, Stanek J, Wanek T, Meier M, Ding XQ, Muller M, Langer O, Löscher W. (2011). A novel positron emission tomography imaging protocol identifies seizure-induced regional overactivity of P-glycoprotein at the blood–brain barrier. Journal of Neuroscience, 31(24), 8803–8811. Baune BT, Dannlowski U, Domschke K, Janssen DG, Jordan MA, Ohrmann P, Bauer J, Biros E, Arolt V, Kugel H, Baxter AG, Suslow T. (2010). The interleukin 1 beta (IL1B) gene is associated with failure to achieve remission and impaired emotion processing in major depression. Biological Psychiatry, 67(6), 543–549. Bertolino A, Caforio G, Blasi G, De Candia M, Latorre V, Petruzzella V, Altamura M, Nappi G, Papa S, Callicott JH, Mattay VS, Bellomo A, Scarabino T, Weinberger DR, Nardini M. (2004). Interaction of COMT Val108/158 Met genotype and olanzapine treatment on prefrontal cortical function in patients with schizophrenia. American Journal of Psychiatry, 161(10), 1798–1805.
cha p ter 7 : I maging G enetics of Pharmacological R es p onse in Psychiatric D isorders
107
Bhat S, Dao DT, Terrillion CE, Arad M, Smith RJ, Soldatov NM, Gould TD. (2012). CACNA1C (Cav1.2) in the pathophysiology of psychiatric disease. Progress in Neurobiology, 99(1), 1–14. Blasi G, Bertolino A. (2006). Imaging genomics and response to treatment with antipsychotics in schizophrenia. NeuroRx, 3(1), 117–130. Blasi G, Napolitano F, Ursini G, Taurisano P, Romano R, Caforio G, Fazio L, Gelao B, Di Giorgio A, Lacovelli L, Sinibaldi L, Poplizio T, Usiello A, Bertolino A. (2011). DRD2/AKT1 interaction on D2 c-AMP independent signaling, attentional processing, and response to olanzapine treatment in schizophrenia. Proceedings of the National Academy of Sciences, 108(3), 1158–1163. Benedetti F, Bollettini I, Barberi I, Radaelli D, Poletti S, Locatelli C, Pirovano A, Lorenzi C, Falini A, Colombo C, Smeraldi E. (2013). Lithium and GSK3-β Promoter gene variants influence white matter microstructure in bipolar disorder. Neuropsychopharmacology, 38(2), 313–327. Beurel E, Michalek SM, Jope RS. (2010). Innate and adaptive immune responses regulated by glycogen synthase kinase-3 (GSK3). Trends in Immunology, 31(1), 24–31. Bigos et al. (2010). Genetic variation in CACNA1C affects brain circuitries related to mental illness. Archives of General Psychiatry, 67(9), 939–945. Bonneau B, Prudent J, Popgeorgiev N, Gillet G. (2013). Non-apoptotic roles of Bcl-2 family: The calcium connection. Biochimica et Biophysica Acta (BBA)-Molecular Cell Research, 1833(7), 1755–1765. Bufalino C, Hepgul N, Aguglia E, Pariante CM. (2013). The role of immune genes in the association between depression and inflammation: a review of recent clinical studies. Brain, Behavior, and Immunity, 31, 31–47. Cardoner N, Soria V, Gratacòs M, Hernández-Ribas R, Pujol J, López-Solà M, Deus J, Urretavizcaya M, Estivill X, Menchón JM, Soriano-Mas C. (2013). Val66met bdnf genotypes in melancholic depression: effects on brain structure and treatment outcome. Depression and Anxiety, 30(3), 225–233. Cheon KA, Ryu YH, Kim JW, Cho DY. (2005). The homozygosity for 10-repeat allele at dopamine transporter gene and dopamine transporter density in Korean children with attention deficit hyperactivity disorder: relating to treatment response to methylphenidate. European Neuropsychopharmacology, 15(1), 95–101. Cohen MX, Krohn-Grimberghe A, Elger CE, Weber B. (2007). Dopamine gene predicts the brain’s response to dopaminergic drug. European Journal of Neuroscience, 26(12), 3652–3660. Conner BT, Noble EP, Berman SM, Ozkaragoz T, Ritchie T, Antolin T, Sheen C. (2005). DRD2 genotypes and substance use in adolescent children of alcoholics. Drug and Alcohol Dependence, 79(3), 379–387. Contini V, Rovaris DL, Victor MM, Grevet EH, Rohde LA, Bau CH. (2013). Pharmacogenetics of response to methylphenidate in adult patients with Attention-Deficit/Hyperactivity Disorder (ADHD): a systematic review. European Neuropsychopharmacology, 23(6), 555–560. Crisafulli C, Fabbri C, Porcelli S, Drago A, Spina E, De Ronchi D, Serretti A. (2011). Pharmacogenetics of antidepressants. Frontiers in Pharmacology, 2, 6. del Campo N, Chamberlain SR, Sahakian BJ, Robbins TW. (2011). The roles of dopamine and noradrenaline in the pathophysiology and treatment of attention-deficit/hyperactivity disorder. Biological Psychiatry, 69(12), e145–e157. Di Giorgio A, Sambataro F, Bertolino A. (2009). Functional imaging as a tool to investigate the relationship between genetic variation and response to treatment with antipsychotics. Current Pharmaceutical Design, 15(22), 2560–2572. Domschke K, Dannlowski U, Ohrmann P, Lawford B, Bauer J, Kugel H, Heindel W, Young R, Morris P, Arolt V, Deckert J, Suslow T, Baune BT. (2008). Cannabinoid receptor 1 (CNR1) gene: impact on antidepressant treatment response and emotion processing in
major depression. European Neuropsychopharmacology, 18(10), 751–759. Domschke K, Dannlowski U, Hohoff C, Ohrmann P, Bauer J, Kugel H, Zwanzger P, Heindel W, Deckert J, Arolt V, Suslow T, Baune BT. (2010). Neuropeptide Y (NPY) gene: Impact on emotional processing and treatment response in anxious depression. European Neuropsychopharmacology, 20(5), 301–309. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, Goldman D, Weinberger, DR. (2001). Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proceedings of the National Academy of Sciences, 98(12), 6917–6922. Esposito-Smythers C, Spirito A, Rizzo C, McGeary JE, Knopik VS. (2009). Associations of the DRD2 TaqIA polymorphism with impulsivity and substance use: preliminary results from a clinical sample of adolescents.Pharmacology Biochemistry and Behavior, 93(3), 306–312. Fakra E, Hyde LW, Gorka A, Fisher PM, Munoz KE, Kimak M, Halder I, Ferrell RE, Manuck SB, Hariri AR. (2009). Effects of HTR1A C (-1019) G on amygdala reactivity and trait anxiety. Archives of General Psychiatry, 66(1), 33–40. Falcone M, Smith RM, Chenoweth MJ, Bhattacharjee AK, Kelsoe JR, Tyndale RF, Lerman C. (2013). Neuroimaging in psychiatric pharmacogenetics research: the promise and pitfalls. Neuropsychopharmacology, 38(12), 2327–2337. Gerretsen P, Müller DJ, Tiwari A, Mamo D, Pollock BG. (2009). The intersection of pharmacology, imaging, and genetics in the development of personalized medicine. Dialogues in Clinical Neuroscience, 11(4), 363–376. Goldberg TE, Egan MF, Gscheidle T, Coppola R, Weickert T, Kolachana BS, Goldman D, Weinberger DR. (2003). Executive subprocesses in working memory: relationship to catechol-Omethyltransferase Val158Met genotype and schizophrenia. Archives of General Psychiatry, 60(9), 889–896. Gonul AS, Kitis O, Eker MC, Eker OD, Ozan E, Coburn K. (2011). Association of the brain-derived neurotrophic factor Val66Met polymorphism with hippocampus volumes in drug-free depressed patients. World Journal of Biological Psychiatry, 12(2), 110–118. Gore FM, Bloem PJ, Patton GC, Ferguson J, Joseph V, Coffey C, Sawyer SM, Mathers CD. (2011). Global burden of disease in young people aged 10–24 years: a systematic analysis. The Lancet, 377(9783), 2093–2102. Gould TD, Manji HK. (2005). Glycogen synthase kinase-3: a putative molecular target for lithium mimetic drugs. Neuropsychopharmacology, 30(7), 1223–1237. Gyekis JP, Yu W, Dong S, Wang H, Qian J, Kota P, Yang J. (2013). No association of genetic variants in BDNF with major depression: A meta- and gene- based analysis. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 162(1), 61–70. Ikeda A, Kato T. (2003). Biological predictors of lithium response in bipolar disorder. Psychiatry and Clinical Neurosciences, 57(3), 243–250. Kane J, Honigfeld G, Singer J, Meltzer H. (1988). Clozapine for the treatment-resistant schizophrenic: a double-blind comparison with chlorpromazine. Archives of General Psychiatry, 45(9), 789–796. Kawamoto EM, Vivar C, Camandola S. (2012). Physiology and pathology of calcium signaling in the brain. Frontiers in Pharmacology, 3, 61. Kempton MJ, Ruberto G, Vassos E, Tatarelli R, Girardi P, Collier D, Frangou S. (2009). Effects of the CACNA1C risk allele for bipolar disorder on cerebral gray matter volume in healthy individuals. The American Journal of Psychiatry, 166(12), 1413–1414. Kirchheiner J, Seeringer A, Godoy AL, Ohmle B, Maier C, Beschoner P, Sim EJ, Viviani R. (2011). CYP2D6 in the brain: genotype effects on resting brain perfusion. Molecular Psychiatry, 16(3), 333–341. Kirsch P, Reuter M, Mier D, Lonsdorf T, Stark R, Gallhofer B, Vaitl D, Hennig J. (2006). Imaging gene–substance interactions: the
1 0 8 Part I I : I maging G enetics and D rug D iscovery
effect of the DRD2 TaqIA polymorphism and the dopamine agonist bromocriptine on the brain activation during the anticipation of reward. Neuroscience Letters, 405(3), 196–201. Kooij JS, Boonstra AM, Vermeulen SH, Heister AG, Burger H, Buitelaar JK, Franke B. (2008). Response to methylphenidate in adults with ADHD is associated with a polymorphism in SLC6A3 (DAT1). American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 147(2), 201–208. Lee HJ, Kang SG. (2011). Genetics of tardive dyskinesia. International Review of Neurobiology, 98, 231–264. Lotta T, Vidgren J, Tilgmann C, Ulmanen I, Melen K, Julkunen I, Taskinen J. (1995). Kinetics of human soluble and membrane-bound catechol O-methyltransferase: a revised mechanism and description of the thermolabile variant of the enzyme. Biochemistry, 34(13), 4202–4210. Machado-Vieira R, Pivovarova NB, Stanika RI, Yuan P, Wang Y, Zhou R, Zarate CA Jr, Drevets WC, Brantner CA, Baum A, Laje G, McMahon FJ, Chen G, Du J, Manji HK, Andrews SB. (2011). The Bcl-2 gene polymorphism rs956572AA increases inositol 1, 4, 5-trisphosphate receptor–mediated endoplasmic reticulum calcium release in subjects with bipolar disorder. Biological Psychiatry, 69(4), 344–352. Malhotra AK, Zhang JP, Lencz T. (2012). Pharmacogenetics in psychiatry: translating research into clinical practice. Molecular Psychiatry, 17(8), 760–769. Meltzer HY. (1997). Treatment-resistant schizophrenia-the role of clozapine. Current Medical Research and Opinion, 14(1), 1–20. Murray CJ, Lopez AD. (2013). Measuring the global burden of disease. The New England Journal of Medicine, 369(5), 448–457. Niitsu T, Fabbri, C, Bentini F, Serretti A. (2013). Pharmacogenetics in major depression: a comprehensive meta-analysis. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 45, 183–194. Pei Y, Smith AK, Wang Y, Pan Y, Yang J, Chen Q, Pan W, Bao F, Zhao L, Tie C, Wang Y, Wang J, Zhen W, Zhou J, Ma X. (2012). The brain-derived neurotrophic-factor (BDNF) val66met polymorphism is associated with geriatric depression: a meta-analysis. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 159(5), 560–566. Rabl U, Scharinger C, Müller M, Pezawas L. (2010). Imaging genetics: implications for research on variable antidepressant drug response. Expert Review of Clinical Pharmacology, 3(4), 471–489. Ravyn D, Ravyn V, Lowney R, Nasrallah HA. (2013). CYP450 pharmacogenetic treatment strategies for antipsychotics: a review of the evidence. Schizophrenia Research, 149(1-3), 1–14. Ritchie T, Noble EP. (2003). Association of seven polymorphisms of the D2 dopamine receptor gene with brain receptor-binding characteristics. Neurochemical Research, 28(1), 73–82. Roffman JL, Weiss AP, Goff DC, Rauch SL, Weinberger DR. (2006). Neuroimaging-genetic paradigms: a new approach to investigate the pathophysiology and treatment of cognitive deficits in schizophrenia. Harvard Review of Psychiatry, 14(2), 78–91. Rogers HL, Bhattaram A, Zineh I, Gobburu J, Mathis M, Laughren TP, Pacanowski M. (2012). CYP2D6 genotype information to guide pimozide treatment in adult and pediatric patients: basis for the US Food and Drug Administration’s new dosing recommendations. The Journal of Clinical Psychiatry, 73(9), 1187–1190. Rosenheck R, Evans D, Herz L, Cramer J, Xu W, Thomas J, Henderson W, Charney D. (1999a). How long to wait for a response to clozapine: a comparison of time course of response to clozapine and conventional antipsychotic medication in refractory schizophrenia. Schizophrenia Bulletin, 25(4), 709–719. Rosenheck R, Dunn L, Peszke M, Cramer J, Xu W, Thomas J, Charney D. (1999b). Impact of clozapine on negative symptoms and on the deficit syndrome in refractory schizophrenia. American Journal of Psychiatry, 156(1), 88–93.
Rowse AL, Naves R, Cashman KS, McGuire DJ, Mbana T, Raman C, De Sarno P. (2012). Lithium controls central nervous system autoimmunity through modulation of IFN-γ signaling. PloS One, 7(12), e52658. Ruhé HG, Booij J, Reitsma JB, Schene AH. (2009). Serotonin transporter binding with [123I] β-CIT SPECT in major depressive disorder versus controls: effect of season and gender. European Journal of Nuclear Medicine and Molecular Imaging, 36(5), 841–849. Seeman P. (2010). Dopamine D2 receptors as treatment targets in schizophrenia. Clinical Schizophrenia & Related Psychoses, 4(1), 56–73. Seeman P. (2011). All roads to schizophrenia lead to dopamine supersensitivity and elevated dopamine D2High receptors. CNS Neuroscience & Therapeutics, 17(2), 118–132. Solís-Chagoyán H, Calixto E, Figueroa A, Montaño LM, Berlanga C, Rodríguez-Verdugo MS, Romo F, Jiménez M, Gurrola CZ, Riguelme A, Benítez-King G. (2013). Microtubule organization and L-type voltage-activated calcium current in olfactory neuronal cells obtained from patients with schizophrenia and bipolar disorder. Schizophrenia research, 143(2-3), 384–389. Steen VM, Lovlie R, Osher Y, Belmaker RH, Berle JO, Gulbrandsen AK. (1998). The polymorphic inositol polyphosphate 1-phosphatase gene as a candidate for pharmacogenetic prediction of lithium-responsive manic-depressive illness. Pharmacogenetics, 8, 259–268. Stingl JC, Brockmöller J, Viviani R. (2012). Genetic variability of drug-metabolizing enzymes: the dual impact on psychiatric therapy and regulation of brain function. Molecular Psychiatry, 18(3), 273–287. Szeszko PR, Lipsky R, Mentschel C, Robinson D, Gunduz-Bruce H, Sevy S, Ashtari M, Napolitano B, Bilder RM, Kane JM, Goldman D, Malhotra AK. (2005). Brain-derived neurotrophic factor val66met polymorphism and volume of the hippocampal formation. Molecular Psychiatry, 10(7), 631–636. Szobot CM, Roman T, Hutz MH, Genro JP, Shih MC, Hoexter MQ, Junior N, Pechansky F, Bressan RA, Rohde LA. (2011). Molecular imaging genetics of methylphenidate response in ADHD and substance use comorbidity. Synapse, 65(2), 154–159. Wallis D. (2010). The search for biomarkers for attention deficit/hyperactivity disorder. Drug News & Perspectives, 23(7), 438–439. Wang JS, Zhu HJ, Donovan JL, Yuan HJ, Markowitz JS, Geesey ME, DeVane CL. (2009). Aripiprazole brain concentration is altered in P-glycoprotein deficient mice. Schizophrenia Research, 110(1), 90–94. Weinshilboum RM, Otterness DM, Szumlanski CL. (1999). Methylation pharmacogenetics: catechol O-methyltransferase, thiopurine methyltransferase, and histamine N-methyltransferase. Annual Review of Pharmacology and Toxicology, 39(1), 19–52. Wiedemann K. (2011). Biomarkers in development of psychotropic drugs. Dialogues in Clinical Neuroscience, 13(2), 225–234. Willner P. (1985). Antidepressants and serotonergic neurotransmission: an integrative review. Psychopharmacology, 85(4), 387–404. Wise RG, Tracey I. (2006). The role of fMRI in drug discovery. Journal of Magnetic Resonance Imaging, 23(6), 862–876. Yang B, Chan RC, Jing J, Li T, Sham P, Chen RY. (2007). A meta-analysis of association studies between the 10-repeat allele of a VNTR polymorphism in the 3′-UTR of dopamine transporter gene and attention deficit hyperactivity disorder. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 144(4), 541–550. Zhang JP, Lencz T, Malhotra AK. (2010). Dopamine D2 receptor genetic variation and clinical response to antipsychotic drug treatment: a meta-analysis. The American Journal of Psychiatry, 167(7), 763–772. Zhang JP, Malhotra AK. (2013). Pharmacogenetics of antipsychotics: recent progress and methodological issues. Expert Opinion on Drug Metabolism & Toxicology, 9(2), 183–191.
cha p ter 7 : I maging G enetics of Pharmacological R es p onse in Psychiatric D isorders
109
PA R T III. IMAGING GENE TICS AND GENE TIC DISCOVERY
8. IMAGING GENE TICS OF WILLIAMS SYNDROME Daniel Paul Eisenberg, Ranjani Prabhakaran, and Karen F. Berman
I N T RO DU C T I O N Among the seeds of mental diversity and disorder, genetic variation represents an invaluable subject of study not only because it may harbor important diagnostic and prognostic information, but also because when integrated with meaningful neurophysiological measurements, it can be used to unearth biological mechanisms supporting brain function, which can direct novel, principled therapeutic target discovery. Yet, delineating genetic mechanisms underlying behavior through neuroimaging is a tall order. Given that many genes impact many functions of many brain cell types in many neural circuits important for many behaviors, any single genetic variant must compete with armies of other biological actors at each of the molecular, cellular, and circuit levels to ultimately impact a neuroimaging or behavioral readout. Additionally, at each stratum of organization, the result of a particular variant is likely to depend critically on the context provided by other variants (Nicodemus et al. 2010a; Nicodemus et al. 2010b; Eisenberg, Kleinman, and Berman 2013). When combined with expected limitations on the effect sizes of individual common variants, this “penetrance” problem can leave ambiguous their clinical relevancy unless ancillary investigation is undertaken. A further problem is establishing that observed genetic associations are due to the implicated gene and not a result of stratification and linkage disequilibrium. Finally, when the syndrome or behavioral condition under study is very heterogeneous and likely polygenic in origin, as is the case with most major psychiatric disorders, establishing precision in linking the clinical entity to associations between particular genetic variants and particular neuroimaging measures is not a trivial matter. Thus, one of the prime challenges in human neuroimaging genetics is to overcome inherent limitations on sensitivity and specificity.
The accompanying sections in this work attest to the fact that, while careful experimentation can ameliorate and even overcome many of these issues, studies in Williams syndrome (WS), a 7q11.23 microdeletion syndrome, hold a special place in the imaging genetics research canon as a model body of experimentation that confronts these challenges head-on. Work in WS holds particular promise because (1) the genetic abnormality responsible for the syndrome results in substantial gene dosage effects that translate to a large clinical effect size; (2) the WS deletion on chromosome 7 is tightly circumscribed, leaving little ambiguity as to the locale of the etiological variant; and (3) the hallmark cognitive and behavioral phenotypes in WS are stereotyped and reliably profiled, providing a clear translational path from genes to brain to behavior. Thus, Williams syndrome offers the imaging genetics literature a near ideal model system with which to test the basic question: How might genes impact brain systems to result in clinically significant changes in cognition and comportment? Here we first touch on what is known of both the genetic deficit causing Williams syndrome and its characteristic neurocognitive and personality profile, and we then delve into the rich recent history of imaging innovations aimed at marrying these two domains (genetic and clinical) under the canopy of specific systems-level neurobiological mechanisms. GENE TICS Williams syndrome arises from a rare (1:7500–1:20,000 live births) error during meiosis, typically resulting in hemizygous deletion of approximately 1.6 megabases (Mb) composing some 25 genes on 7q11.23. This chromosomal region is bounded by highly similar low-copy repeat (LCR) sequences, bookends that lend the WS
113
locus its predisposition to disruption largely via unequal DNA crossing-over during homologous recombination (Dutly and Schinzel 1996) and that most often result in a uniformly sized deletion across affected individuals. Only quite infrequently (< 5% of cases) is the consequence of this genetic segment’s meiotic instability a non-classical deletion size (Botta et al. 1999). In addition, the parents of WS individuals show enrichment for a paracentric inversion of the WS region, which results in no known sequelae aside from a potentially increased risk of having offspring with WS (Tam et al. 2008), and there is not yet consistent evidence for imprinting effects (Jurado et al. 1996; Wang et al. 1999). By the same token, the flanking low-copy repeat sequence break points may also lead to a duplication of the WS region, resulting in a clinical phenotype distinct from WS. Recently, a triplication has even been reported (Beunders et al. 2010).
CLINICAL PHENOT YPE MEDICAL PRESENTATION
The modern history of Williams syndrome began when the diagnosis of a relatively uncommon cardiovascular condition—supravalvular aortic stenosis—was predicted in a series of three unrelated pediatric cardiac patients based on shared facial characteristics (Williams, Barratt-Boyes, and Lowe 1961). To this day, structural cardiovascular abnormalities, including supravalvular aortic stenosis and pulmonary artery stenosis, are still frequently the issues that initially bring WS individuals to clinical attention and subsequent genetic diagnosis. Though penetrance is variable, and for many with WS, genetic predilection toward these abnormalities does not translate to clinically meaningful illness, early surgical intervention is often required to prevent progressive, potentially fatal cardiac disease (Collins 2013). These cardiovascular complications of WS are attributable to hemideletion of the elastin (ELN) gene within the WS critical region, which, among other functions, is particularly important in determining the mechanical properties of large blood vessels (Merla et al. 2012). For the purposes of neuroimaging genetics, this poses interesting questions for understanding neuroimaging observations of affected individuals. Overt neurovascular occurrences such as ischemic events and gross cerebral arteriopathy are infrequently observed in WS (Soper et al. 1995; Wollack et al. 1996), but could even relatively subtle, systematic alterations in cerebral vascular supply during early brain development contribute 114
to WS deletion-associated neural phenotypes? Recent angiographic work on a selective research cohort indicating normal patency in the proximal intracranial arteries in WS, along with the absence of the neuropsychological phenotype (discussed later in this chapter) in isolated ELN hemizygous deletion (Eisenberg et al. 1964; Ewart et al. 1994; Fryssira et al. 1997), provides important reassurance that this may not be a critical concern (Wint et al. 2013); however, the field awaits definitive developmental data ruling out this possibility. Beyond cardiovascular and morphological alterations, a host of other somatic conditions are also associated with the syndrome. These include early growth restriction and feeding problems, infantile hypercalcemia/hypercalciuria, hyperacusis, early puberty, dental malocclusion and microdontia, renal problems, and endocrinological and musculoskeletal (e.g., joint hypermobility, contractures) abnormalities (Pober and Morris 2007). Neurologically, there can be early hypotonia followed by hypertonia and hyperactive deep tendon reflexes, nystagmus, and mild coordination limitations. However, most intriguing are the hallmark neurocognitive and personality profiles that are clinically overt. The unambiguous expression of these behavioral profiles in WS individuals stands in striking contrast to the much subtler, subclinical differences in behavioral phenotypes predicted by other types of functional genetic variation currently under study in the imaging genetics field. COGNITIVE PROFILE
Intellectual disability, occurring in most, but not all, affected individuals, was noticed early in the syndrome’s description (Williams, Barratt-Boyes, and Lowe 1961) and, as discussed in greater detail later in this chapter, continues to pose both an important consideration for the interpretation of WS neuroimaging data and a promising focus for future research. Whether by differential engagement in cognitive functional neuroimaging paradigms or by virtue of inherent, neurodevelopmental differences underlying full-scale (domain-general) IQ variation, study cohorts with intellectual disability from WS (or any etiology) may introduce formidable obstacles to unbiased neural measurements. When pursuing characterization of other, domain-specific neurocognitive features of WS, some investigators have therefore adopted a variety of strategies to minimize the potential impact of IQ on outcome measurements, including using younger but “mental age”–matched, typically developing controls; using IQ-matched controls with non-WS developmental P art I I I : I maging G enetics and G enetic D iscovery
disorders; or studying only the rare individuals with both WS and normal-range IQs. Some of these strategies may arguably trade one type of confound for another. For example, matching IQ but not age may generate findings resulting from normative regional neurodevelopmental trajectories, yet matching with a non-WS developmental disorders cohort may lead to results due to abnormalities in the non-WS conditions. Thus, in WS neuroimaging research, accounting for the nature of general intellectual capacities in both study and control groups becomes a crucial challenge in order to best delineate deletion-related neural phenomena. With the accrual of greater clinical experience with WS, appreciation grew for areas of very specific cognitive deficits. In particular, visuospatial construction—the ability to use spatial relationships to assemble a whole from its individual parts—is markedly impaired. This has been demonstrated formally with the block-design task that requires building a cued black-and-white design with painted blocks (Bellugi et al. 1990). Even compared with children with other developmental disabilities, children with Williams syndrome perform poorly. Neuropsychological testing in large groups of children with Williams syndrome as well as other neurodevelopmental syndromes has revealed that the Williams neurocognitive profile—specifically, visuospatial construction deficits in conjunction with intact verbal IQ scores—is both a sensitive and specific marker for WS (Mervis et al. 2000). Because of this profile of non-verbal cognitive deficits, in stark contrast with relatively preserved gross language aptitude, and because of a long-standing, fundamental psycholinguistic debate over whether language is a distributed, inseparable correlate of cognitive function or a modular cognitive component with distinct neural underpinnings, WS became a condition showcased as evidence for language “modularity,” the argument being that the roots of language could not be embedded in the development of all other cognitive functions if a pronounced cognitive deficit could be unaccompanied by concomitant linguistic impairment (Pinker 1999). However, in fact, many language milestones are delayed in WS, and discreet aspects of language function are indeed affected by the WS microdeletion, including the acquisition of conditional and relational syntax (Mervis and Becerra 2007). The latter is in accord with the syndrome’s hallmark, striking difficulty in the processing of visuospatial relationships, suggesting that at least certain facets of language function and analogous non-verbal cognitive processes likely share common neurodevelopmental resources. chapter 8 : I maging G enetics of W illiams S yndrome
PERSONALIT Y PROFILE
In addition to a distinctive neurocognitive profile, WS is accompanied by a remarkable personality profile, notable for pronounced sociability (Tomc, Williamson, and Pauli 1990; Mervis and Klein-Tasman 2000; Mervis et al. 2003). This strong social drive to interact with others is evident at the earliest germs of social development, as even infants with WS spend increased amounts of time looking at both familiar and unfamiliar faces (Mervis et al. 2003), and continues to flower throughout childhood, accompanied by elevated empathy measurements (Mervis and Klein-Tasman 2000) and greater tendency to approach strangers (Mervis et al. 2003; Doyle et al 2004). In fact, the paucity of stranger anxiety in children with WS can be sufficient to pose a safety concern, and assertive instruction to prevent WS children from wandering astray with someone unfamiliar to them is often considered necessary (Riby et al. 2013). In contradistinction, this fearlessness is isolated to social situations, and WS individuals are vulnerable to non-social anxieties and specific phobias (Udwin and Yule 1991; Einfeld, Tonge, and Florio 1997; Dykens 2003; Woodruff-Borden et al. 2010). Notably, the WS social-affective phenotype is distinct from other developmental disorders, suggesting that it is not merely reflective of cognitive impairment, but rather constitutes a unique property of WS (Doyle et al. 2004). Like the WS neurocognitive profile, when experimentally operationalized and quantified, the personality profile is characterized by high sensitivity and specificity in distinguishing individuals with WS from those with other neurodevelopmental syndromes and represents an invaluable model for the social cognitive field (Klein-Tasman and Mervis 2003); unlike many neuropsychiatric conditions (e.g., schizophrenia, depression, autism) showing various deficits in affiliative behavior and complicated by symptoms that may interfere with social cognitive measurements, WS is unique in its brand of isolated social exuberance and promises to offer privileged and pointed insight into the complexities of social neurobiology. BR A I N P H E N O T Y P E The advent of neuroimaging technology has made possible the current understanding of the neural foundations of these unique neurocognitive and personality phenotypes. In the following text, research efforts focused on characterizing structural and functional imaging variation putatively linked to the WS visuospatial deficiency and hypersociability are emphasized. 115
VISUAL NEURAL SYSTEMS IN WILLIAMS SYNDROME
Because visual stimuli can be carefully controlled and easily adapted for use in a number of different animal or human experimental models, because early receptive cortical regions show regularized, hierarchically organized topographic features, and because of the precious nature of vision itself to any animal with eyes, foundational work in the neurosciences has often been pioneered in the realm of visual systems biology. Groundbreaking achievements in understanding visual information processing from investigations in non-human primates and later positron emission tomography (PET) studies in humans have delineated two key cortical processing streams in the brain—commonly described as a “where” pathway and a “what” pathway, which anatomically diverge after the primary visual cortex (Mishkin, Ungerleider, and Macko 1983; Haxby et al. 1991). The former pathway extends dorsally through the parietal cortex and distills information about spatial relationships among visual stimuli, whereas the latter pathway follows a ventral course through temporal cortices and carries information relating to the classification and identity of visualized objects. This dual-stream model of visual processing provides one road map upon which the specific circuit-level pathology in WS can be charted. As discussed earlier, WS is a human molecular genetic experiment of nature, in which one aspect of visual processing—spatial construction—is particularly perturbed, whereas other visual functions, such as facial recognition and discrimination, remain intact (Bellugi et al. 2000). The dual-stream model of visual processing would predict that it is specifically along the dorsal, “where,” pathway—and less so in ventral stream locations—that dysfunction relating to this visuospatial impairment occurs as a result of the WS microdeletion.
Primary Visual Cortex The cortical fountainhead of both dorsal and ventral streams is the primary visual cortex, where divergence of the two streams is hypothesized to be already demarcated by distinct microcircuitry and topography, with the magnocellular, peripheral visual cortex largely associated with the dorsal stream and the parvocellular, foveal visual cortex largely associated with the ventral stream. Evidence for primary visual cortex (V1) involvement in the WS visuospatial cognitive phenotype is scarce. An initial evaluation of rare postmortem primary visual cortex tissue, specifically extracted from presumptive 116
peripheral field regions of Brodmann’s Area 17 from three WS brains and three age- and sex-matched control brains, suggested differences in cell-packing density (though in a parvocellular projection layer in the left hemisphere only) and neuronal size distributions in WS (Galaburda et al. 2002). This is in contrast to findings in selected cortical layers of primary auditory cortex, where these same samples showed blunted asymmetry but not overall differences in cell-packing density or neuronal size (Holinger et al. 2005). However, as noted by Galaburda and colleagues, differences in BA17 cell-packing densities may be related to the nature of the substantially smaller total brain volumes in the WS specimens—a well-replicated observation in in vivo structural studies. Better understanding of the specificity and generalizability of these preliminary data must await additional histomorphometric studies in larger samples with greater regional coverage. Nonetheless, both structural and functional neuroimaging have not yet provided consistent evidence of primary visual cortex abnormalities in WS, suggesting that any in vivo sequelae of these findings at the V1 tier are unlikely observable with standard neuroimaging techniques. In particular, though collaborators performing in vivo structural studies analyzed at the lobar level have reported diminished occipital cortical volumes (Reiss et al. 2000; Reiss et al. 2004; Chiang et al. 2007) and increased occipital gyrification (Schmitt et al. 2002; Gaser et al. 2006), subsequent voxel-wise analyses have not localized specific structural abnormalities to primary visual cortex, but rather, to dorsal parieto-occipital locales, as will be described later in this chapter. The majority of functional imaging investigations employing visual stimuli have similarly failed to provide convincing evidence for primary visual cortex involvement in the WS visuospatial deficit (Meyer-Lindenberg 2004; Eckert et al. 2006). However, several exceptions exist. One study compared neural responses to face stimuli, to which WS individuals may differentially attend, as measured by eye-tracking (Porter, Shaw, and Marsh 2010; Riby et al. 2011) and discussed in greater detail later in this chapter. This study found diminished occipital activation in 11 individuals with WS (both children and adults) relative to age- and sex- (but not IQ- or performance-)matched controls (Mobbs et al. 2004). A second experiment by the same research group used a global visual-processing task, in which the overall shape of a figure composed of smaller congruent or non-congruent shapes was to be identified, and found a cluster that extended into inferior occipital cortex and fusiform gyrus, where 10 WS adults showed less activation than eight age- (but not IQ- or performance-)matched comparison participants (Mobbs P art I I I : I maging G enetics and G enetic D iscovery
et al. 2007a). However, most of these studies, both those with and without WS-associated differences observed in early visual cortex, were not designed to directly measure early visual-processing responses, but rather included visual control stimuli to eliminate early visual processing from the dependent measurement. In fact, because visual stimuli are first cortically processed in primary visual cortex, a finding of differential neural responses in this region to basic features of stimuli in WS would present a formidable challenge to the interpretation of abnormalities detected in higher-order visual areas. For this reason, one report on dedicated, surface-based retinotopic mapping is particularly significant in interpreting previous inferences about V1 function. This report’s experiments, in which fMRI using standard dynamic visual stimuli was employed to functionally define the boundaries of V1, demonstrated not only that WS individuals show the same spatial extent of primary visual cortex, but also the same location of V1 based on center of mass analysis (Olsen et al. 2009). Thus, the weight of the neurostructural and neurofunctional data to date supports the notion of primary visual cortex largely unscathed by the WS microdeletion, although fine-grained, subtle alterations in this region may become apparent with imaging at higher field strengths than have been available to date.
Temporo-Occipital Cortex After V1, the ventral stream flows through prestriate cortex and along the inferior longitudinal fasciculus (ILF) to reach inferior temporal gyrus ports. In concordance with clinical and neuropsychological data indicating relatively spared visual-object processing, neuroimaging support for specific, WS-associated abnormalities in visual temporo-occipital cortex is limited. Though many structural MRI studies have been conducted in WS (Table 8.1), in general, they have not found statistically significant inferior temporo-occipital gray matter volume or concentration abnormalities (MeyerLindenberg et al. 2004; Kippenhan et al. 2005; Boddaert et al. 2006; Faria et al. 2012), or gyrification disturbances (Schmitt et al. 2002; Kippenhan et al. 2005). One exception is a study of 15 WS children compared with age- and sex(but not IQ-)matched controls, which found diminished left inferior occipital and right fusiform gyri and greater left temporal (including fusiform gyrus) gray matter volumes in WS; however, these findings were absent after controlling for IQ (Campbell et al. 2009). A more recent study focusing on white matter identified nominally increased fusiform fiber tract volume, fractional anisotropy, and fiber chapter 8 : I maging G enetics of W illiams S yndrome
density in 20 WS individuals compared with 10 age- (but not IQ-)matched typically developed controls (Haas et al. 2012). In the same study, when a different control group (10 age- and IQ-matched developmentally delayed individuals) was used, consistent but nominal significance was found for fusiform fiber tract volume and fiber density differences in WS. These findings were absent in two control regions— the genu of the corpus callosum and the caudate—though an even smaller study has reported normal DTI measurements in the fusiform but reduced white matter volumes and increased fractional anisotropy in the caudate (Faria et al. 2012). Like structural studies, functional neuroimaging research has not consistently demonstrated neurophysiological abnormalities along ventral stream pathways in temporo-occipital cortex. One unreplicated study found that WS adolescents and adults show different inferior temporal activation during house (but not face) viewing, when compared to either an age- (but not IQ-)matched typically developed group or an approximate “mental age”–matched group of children (O’Hearn et al. 2011). However, in experiments designed to assay ventral stream responses during attention to the content (rather than the spatial characteristics) of visual stimuli, WS individuals show robust fusiform activation that is statistically indistinguishable from age-, gender-, and IQ-matched controls (Meyer-Lindenberg et al. 2004). Normal fusiform activation has been replicated in an overlapping cohort using a different paradigm (passive face and house viewing) (Sarpal et al. 2008), as well as in independent work (Mobbs et al. 2004; Mobbs et al. 2007a). Despite these largely negative findings in the fusiform gyrus, one proposed island of WS-associated differential function in the ventral stream is the fusiform face area, where specialized processing of viewed facial information occurs. Taking account of the hypersociability and potential differences in attention to, and engagement with, facial stimuli—despite generally intact facial recognition—in WS (Porter, Shaw, and Marsh 2010; Riby et al. 2011), it is reasonable to hypothesize some measureable neurophysiological differences in this region. In a study of passive viewing (faces, objects, places, and textures), native space analyses of activation volumes found, on average, larger functionally defined fusiform face areas in 13 adults with WS alongside normal response amplitudes and normal facial recognition measurements when compared to 13 age- (but not IQ-)matched comparison subjects (Golarai et al. 2010). Fusiform gyrus cortex responsive to other (non-face) objects was also investigated and showed no differences. Thus, subtle and specific differences in the functional topography of face-responsive fusiform cortex may 117
TABLE 8.1
Year
SUMMARY OF NEUROIMAGING LITERATURE ON VISUAL NEURAL SYSTEMS IN WILLIAMS SYNDROME Authors
Sample Characteristics
Demographic Matching
Type of Imaging
Task/Neuropsychological Measure/Rating Scale
Main Findings
Age, sex, handedness
fMRI
Gaze detection task
WS > TD: activation in PFC, MTL, Thal, and ACC.
Brain-Behavior Correlation
PRIMARY VISUAL CORTEX Functional and Structural Findings 2004
Mobbs et al. (^)
n = 20 adults (11 WS, 9 TD)
TD > WS: activation in primary and secondary visual cortex. WS = TD: activation in Amyg, FuG, and STS. 2009
Olsen et al. (†)
n = 20 adults (10 WS, 10 TD)
Age, sex, IQ, handedness
fMRI and MRI
Retinotopic mapping procedure, WASI, and visuospatial construction task performance (from Meyer-Lindenberg et al. 2004)
WS > TD: variability of V1 anatomical boundaries.
None
WS > TD: cortical thickness in perisylvian language-related cortex and gray matter volume in FuG.
WS = TD: spatial extent of functionally defined V1 surface area (after adjusting for total brain volume).
No correlation between IQ and hemispheric or V1 surface area in WS or TD groups. No correlation between visuospatial task performance and V1 surface area in WS.
TEMPORO-OCCIPITAL CORTEX Structural Findings 2005
Thompson et al. (^)
n = 82 children and adults (42 WS, 40 TD)
Age, sex
MRI
TD > WS: total cerebral volume. 2009
Campbell et al.
n = 30 children (15 WS, 15 TD)
Age, sex
MRI
Conner’s rating scale (inattention and hyperactivity symptom ratings), Strength and Difficulties Questionnaire (assessment of peer problems), WISC-III
WS > TD: gray matter in Temp and Fron lobes and ACC. TD > WS: volume of Par-Occ cortex and BG
2012
Haas et al.
n = 40 adults (20 WS, 20 TD, 10 DD)
Chronological age, mental age
DTI
None
WS > TD and DD: volume, fiber density, and FA of reconstructed fibers projecting through bilateral FuG.
2012
Faria et al.
n = 16 children and adults (8 WS, 8 TD)
Age, sex
DTI
None
WS > TD: FA in SLF, FOF, caudate, and cingulum. TD > WS: volume of caudate, globus pallidum, and anterior commissure (after adjusting for total brain volume); FA in corticospinal tract. WS = TD: gray matter volumes of FuG, PHG, MTG, cerebellum, and Amyg.
WS: Significant correlations between inattention, hyperactivity, and peer problem ratings with gray matter volume in cerebellar, Temp, and Pcu regions. No correlations between IQ and total gray or white matter volume for WS or TD.
Fuctional Findings 2008
Sarpal et al. ()
n = 19 adults (9 WS, 10 TD)
Age, sex, IQ, handedness
fMRI
Passive viewing (faces, houses, scrambled images)
WS: Altered functional connectivity of FFA and PPA with several regions (compared to TD). TD > WS: activation in IPS and BA 18 for houses. WS = TD: ventral stream activation for faces.
2010
2011
Golarai et al.
O’Hearn et al.
n = 30 adults (16 WS, 15 TD)
Age, sex
n = 16 adults (8 WS, 8 TD) and 8 children (TD: mental age controls)
Age (chronological and mental), sex
n = 28 adults (14 WS, 14 TD)
Age, sex
fMRI
fMRI
Passive viewing task (faces, objects, textures, places), Benton recognition test on upright faces, and WAIS-R or WAIS-III
WS > TD: FFA size FuG activation for faces.
Target detection (human faces, houses, shoes, cat faces, scrambled stimuli)
WS and TD adults > TD children: ventral stream activation for faces.
TD > WS: Amyg activation for faces.
WS: Significant (+) correlation between FFA size and Benton test performance. No correlation between IQ and FFA size in WS or TD.
TD adults and children> WS: ventral stream activation for houses.
PARIETO-OCCIPITAL CORTEX Structural Findings 2000
Reiss et al. (^)
MRI
None
WS > TD: ratio of frontal to posterior tissue. TD > WS: brain volume, cerebral white matter volume, Occ lobe volume.
2002
Schmitt et al. (^)
n = 34 adults (17 WS, 17 TD)
Age, sex
MRI
None
WS > TD: cortical gyrification in Par, Occ, and Fron regions.
2004
Reiss et al. (^)
n = 83 children and adults (43 WS, 40 TD)
Age, sex
MRI
WISC-R and WAIS-R
WS > TD: gray matter volume and density in FuG, Amyg, insula, STG, and OFC. TD > WS: gray matter density in superior Par gyri and gray matter volume in Thal and Occ cortex.
2005
Eckert et al.
n = 34 adults (17 WS, 17 TD)
Age, sex
MRI
None
TD>WS: superior Par gray matter volume (adjusted for total brain volume).
2005
Kippenhan et al. (†)
n = 27 adults (14 WS, 13 TD)
Age, sex, IQ, handedness
MRI
None
TD > WS: sulcal depth and gray matter volume of IPS/OccPar sulcus, CS, and OFC. (continued)
TABLE 8.1
CONTINUED
Year
Authors
Sample Characteristics
Demographic Matching
2006
Boddaert et al.
n = 20 children (9 WS, 11 TD)
Age, sex
2006
Eckert et al. (^)
n = 17 adults (8 WS, 9 TD)
2006
Gaser et al. (^)
2007
Hoeft et al. (^)
Type of Imaging
Task/Neuropsychological Measure/Rating Scale
Main Findings
MRI
None
TD > WS: gray matter volume in Par-Occ cortex.
Age, sex
MRI
None
TD > WS (Jacobian modulated and unmodulated): gray matter density in IPS.
n = 82 children and adults (42 WS, 40 TD)
Age, sex
MRI
None
WS > TD: gyrification of Par-Occ regions, Pcu, posterior and anterior cingulate, OFC, MPFC, and paracentral lobule.
n = 30 adults (10 WS, 10 TD, 10 DD)
Age, sex, handedness
DTI
WAIS-III
WS > TD: FA in SLF. WS = TD: FA in SLF and ILF.
2007
Marenco et al. (†)
n = 10 adults (5 WS, 5 TD)
Age, sex, IQ
DTI
None
WS: alteration in directionality of white matter fibers, deviation in course of posterior fiber tract, and reduction in lateralization of fiber coherence; abnormalities in course of ILF and SLF tracts that connect to IPS.
2011
Menghini et al.
n = 25 children and adults (12 WS, 13 TD)
Age, sex
MRI
Stanford-Binet Intelligence Scale, Boston Naming Test, Peabody Picture Vocabulary Test, Phrase Repetition Test, Grammar Comprehension Test, Categorical Fluency Test, WISC-R (Block Design), Developmental Test of Visual-motor Integration, Digit Span
WS > TD: regional gray matter density in Thal, superior Occ lobule, cuneus, insula, and cerebellum. TD > WS: regional gray matter density in posterior superior Par lobule, PHG, MFG and posterior central gyrus
Brain-Behavior Correlation
WS: Significant (-) correlation between FA in right SLF and visuospatial construction scores on WAIS-III Object Assembly subtest.
WS: (1) Significant (+) correlation between visuospatial and visuo-motor abilities and gray matter density in cerebellum, Par lobule, superior and orbital frontal gyri; (2) Significant (+) correlation between morpho-syntactic ability and regional gray matter density in cerebellum, SMA and FuG.
Functional and Structural Findings 2004
Meyer-Lindenberg et al. (†)
n = 24 adults (13 WS, 11 TD)
Age, sex, IQ, handedness
fMRI and MRI
Square completion task
TD > WS: (1) activation in Par regions adjacent to IPS during visuospatial construction and attention to location; (2) volume in ParOcc/IPS and OFC.
2007
Mobbs et al.
n = 18 adults (10 WS, 8 TD)
Age, sex
fMRI
Global processing task, WISC or WASI
TD > WS: activation in superior Par cortex.
No significant correlations between activation and IQ.
HIPPOCAMPAL CORTEX Structural Findings 2006
Van Essen et al.
n = 53 children and adults (16 WS, 37 TD)
Sex
MRI
None
WS > TD: Abnormal cortical folding in many areas (>30) spanning dorsoposterior to ventroanterior cortex in both hemispheres.
2010
Sampaio et al.
n = 28 adults (15 WS, 13 TD)
Age, sex
MRI
Digit Span from WISC-III and WAIS-III, Corsi Blocks Span, California Verbal Learning Test, Rey-Osterrieth Complex Figure
WS > TD: Hipp volume (adjusted for total brain volume).
fMRI: Passive viewing (faces, houses)
WS > TD: shape changes in mid-Hipp (locally larger structure in WS).
WS: lack of Hipp volume asymmetry.
Functional and Structural Findings 2005
Meyer-Lindenberg et al. (†)
n = 20 adults (10 WS, 10 TD)
Age, sex, IQ, handedness
fMRI, PET, MRS, and MRI
TD > WS: activation of faces and houses and regional cerebral blood flow in Hipp; NAA/Creation ration in anterior Hipp. Note: For all tables ^, †, &, and # symbols indicate overlap in participant pool across different studies.
No significant correlations between Hipp volume and memory measures in WS.
occur in WS, though the functional implications of this finding remain unclear, and replication is needed. Given the prolonged structural (Golarai et al. 2007) and functional (Golarai, Liberman, et al. 2010) development of the fusiform face area, whether such differences arise secondary to lifelong atypical social experience and drive will be an important research question for future confirmatory studies. In light of possible fusiform structural connectivity differences in WS (Haas et al. 2012), whether evolution of cortical tuning in this region is guided by input from other neural nodes implicated in WS provides another avenue of experimentation. Interestingly, in the Sarpal et al. (2008) face-house viewing study, while fusiform activation was no different in participants with WS relative to controls, functional connectivity analyses suggested that the dynamics of activity in ventral stream nodes (i.e., functionally defined locales in the fusiform face area and parahippocampal place area) correlated with the dynamics of activity in dorsal stream and social network regions differently in WS (Sarpal et al. 2008). Though dorsal-ventral stream cross-talk has been previously suggested (Zhong and Rockland 2003), these data likely point to large-scale network behaviors that will require further investigation.
Parieto-Occipital Cortex After V1, the dorsal stream climbs through prestriate cortex and along the superior longitudinal fasciculus, extending to parietal cortical way stations. Just as clinical and neuropsychological data consistently describe unambiguous visuospatial deficits in WS, neuroimaging studies have repeatedly evidenced WS-associated dorsal stream abnormalities, both in structure and in function. Though gross examination of small numbers of postmortem WS brains suggesting diminished parieto-occipital volumes offered preliminary anatomic clues as to disturbances in dorsal stream development (Galaburda and Bellugi 2000), in vivo neuroimaging techniques have proven invaluable in elucidating aberrant loci with increasingly greater precision. For instance, early studies of lobar volume suggested greater anterior-posterior ratios in WS (Reiss et al. 2000). Subsequent voxel-based morphometry comparisons of normal-range IQ WS adults with age-, sex-, and IQ-matched controls revealed diminished gray matter volume in the dorsal parieto-occipital sulcus/vertical part of the intraparietal sulcus in the group with WS (Meyer-Lindenberg et al. 2004). When this latter morphometric experiment was conducted either in WS children and age-matched controls or in independent adult 122
cohorts, diminished gray matter concentration in WS was again observed in a similar parieto-occipital junction locale (Reiss et al. 2004; Eckert et al. 2006; Boddaert et al. 2006; Campbell et al. 2009). Potentially consistent with these findings, manual tracings in 17 women with WS and 17 age-matched comparison women found diminished superior parietal cortex (but not inferior parietal cortex) gray matter volumes in WS (Eckert et al. 2005). Greater gyrification of parieto-occipital cortex in WS has been reported in both postmortem and in vivo lobar measurements with MRI (Galaburda and Bellugi 2000; Schmitt et al. 2002). Subsequent surface-based analyses of sulcal geometry in a normal-range IQ adult WS cohort revealed markedly shallow intraparietal sulci, adjacent to the site of the replicated gray matter deficit (Kippenhan et al. 2005). Notably, gray matter thickness per unit surface area was not significantly reduced in this area in an adolescent-adult group of low-IQ WS individuals compared to age- and sex-matched controls (Thompson et al. 2005), suggesting the possibility that any underlying gray matter deficits may be proportional to this sulcal anomaly (Kippenhan et al. 2005). Furthermore, follow-up white matter studies employing diffusion tensor imaging have identified associated differences in the structure of intraparietal sulcus white matter subserving the regions of gray matter abnormality in WS adults (Marenco et al. 2007). One study also found increased fractional anisotropy (FA) of the right superior longitudinal fasciculus, a fiber tract important for dorsal parietal interregional communication, which corresponded with worse visuospatial test scores in WS (Hoeft et al. 2007). Functional neuroimaging studies have corroborated these abnormalities in WS. During a square completion task that requires visuospatial construction judgments, adults with WS, relative to age-, sex-, and IQ-matched controls, demonstrated selective parietal cortex hypoactivation in a region adjacent to and downstream of the structural aberrancies just described (Meyer-Lindenberg et al. 2004). This finding is also consistent with a study that found superior parietal hypoactivation in WS during a global processing visual-discrimination task (Mobbs et al. 2007a). Thus, on balance, the literature to date has left little doubt that dorsal stream abnormalities are a key and prominent element of the WS neural phenotype. How these findings may impact the ability of downstream structures to process visual information and whether there exist additional bottlenecks to the flow of visuospatial information in higher-order visual processing locales are critical questions for understanding visual system networks in WS. P art I I I : I maging G enetics and G enetic D iscovery
Regardless, the consistency of these dorsal stream abnormalities, particularly in the structural domain, offers a valuable brain phenotype for exploration of neurogentic mechanisms.
Hippocampal Cortex The hippocampus receives important input from both ventral and dorsal visual streams, integrating spatial and non-spatial information to facilitate navigation and codify contextual information and episodic memory (Knierim, Lee, and Hargreaves 2006; Whitlock et al. 2008; Wang, Gao, and Burkhalter 2011), and, thus, represents an important area of investigation in WS. Evidence from a series of neuroimaging experiments in a cohort of normal-range IQ adults with and without WS has suggested selective deviations in hippocampal structure and function in WS (Meyer-Lindenberg et al. 2005a). In these experiments, manually traced, whole hippocampus volumes were not different between 7q11.23 deletion and non-deletion individuals; however, deformation-based morphometry revealed significantly greater mid- and dorsal head- subregional volumes in WS, along with a non-significant trend for diminished hippocampal volumes in dorsal/posterior subregions more closely associated with parietal/dorsal stream innervation and with visuospatial processing. This finding was accompanied by reduced anterior hippocampal basal blood flow, reduced BOLD response to visual stimuli (faces and houses), and diminished left hippocampal N-acetyl aspartate, a measure of neuronal integrity (Meyer-Lindenberg
VISUAL NEURAL SYSTEMS (a)
R
L
(b)
et al. 2005a). In a separate study of adolescents and adults with WS, deeper hippocampal/parahippocampal sulcal depths co-localizing to these findings have been observed (Van Essen et al. 2006). An independent investigation not controlling for IQ also reported a left hippocampal volumetric increase in WS individuals with moderate intellectual disability (Sampaio et al. 2010). Though formal tests for relationships between this array of hippocampal abnormalities and visuospatial functioning have not yet been completed, and though these changes may also have relevance for other domains of divergent function (e.g., mnemonic or other cognitive processes) in the 7q11.23 deletion syndrome, these findings suggest the possibility that neural aberrancies in WS may occur even in regions that are integrative and relatively downstream in visual processing. However, additional longitudinal work to understand the precise role of the hippocampus in visual neural systems in WS will be important to begin to delineate between secondary developmental/experiential abnormalities of this structure and direct genetic effects. In sum, the neuroimaging findings reviewed here (see summary Figure 8.1 and Table 8.1) have thus helped to support long-standing hypotheses about the contribution of dorsal stream dysfunction to Williams syndrome’s well-established visuospatial construction deficit. In forming a critical marker on the trail from genes to behavior, this body of work provides a foundation for further demarcation of the origins and development of the WS neurocognitive profile. For instance, which gene or genes in the Williams critical region might be responsible for this aspect of the
Structural Findings: Meyer-Lindenberg et al. (2004) Reiss et al. (2004) Meyer-Lindenberg et al. (2005b) Boddaert et al. (2006) Eckert et al. (2006) Campbell et al. (2009) Menghini et al. (2011) Jabbi et al. (2012) Functional Findings: Meyer-Lindenberg et al. (2004) Mobbs et al. (2004) Meyer-Lindenberg et al. (2005b) Mobbs et al. (2007)
O’Hearn et al. (2011) R
L
Jabbi et al. (2012)
Figure 8.1 VISUAL NEURAL SYSTEMS. Localization of selected findings in visual neural systems (primary visual cortex, temporo-occipital cortex,
parieto-occipital cortex, and hippocampal cortex) reflecting differences between WS and TD groups in (A) structure (gray matter volume or gray matter density) and (B) function (neural activation or regional cerebral blood flow). Loci indicate approximate locations of peak coordinates of findings and are color-coded by study. Size of loci does not represent extent of structural or functional findings.
chapter 8 : I maging G enetics of W illiams S yndrome
123
syndrome, and how does this genetic change translate to the observed neural abnormalities? SOCIO-EMOTIONAL NEURAL SYSTEMS IN WILLIAMS SYNDROME
Just as distinctive as the visuospatial aspect of the WS neurocognitive profile, the remarkable WS social-affective personality phenotype has motivated vigorous research aimed at investigating its neural mechanisms. In the following sections, we review this literature, focusing on structural and functional findings in the amygdala, orbitofrontal cortex, and the insula—key regions in an emotional salience network (Seeley et al. 2007) suggested to play a critical role in social and affective processing (Adolphs 2003; Lieberman 2007).
Amygdala In addition to its central role in emotional salience attribution and the processing of affectively charged stimuli (Adolphs et al. 1994; Morris et al. 1998; Whalen et al. 1998; Anderson and Phelps 2001; Fitzgerald et al. 2006; Paton et al. 2006), the amygdala engages in social calculations, including judgments of approachability and trustworthiness (Adolphs, Tranel, and Damasio 1998; Engell, Haxby, and Todorov 2007; Koscik and Tranel 2011; Winston et al. 2002). In consideration of the social-affective phenotype of WS (Bellugi et al. 1999), several studies have focused on the amygdala as a potential locus of structural and functional alterations in WS. The literature on volumetric differences in the amygdala between WS and typically developing control participants has been mixed. Although several studies have found disproportionately greater amygdala volumes in WS (after correction for total brain volume) (Reiss et al. 2004; Martens et al. 2009; Capitao et al. 2011; Haas et al. 2012; Jabbi et al. 2012), just as many have found no differences (Meyer-Lindenberg et al. 2004; Chiang et al. 2007; Faria et al 2011; Meda, Pryweller, and Thornton-Wells 2012; Menghini et al. 2011). Nonetheless, high-resolution, surface-based structural MRI evaluation of the amygdala has suggested that if amygdala volumes, on the whole, are greater in WS, such changes may be driven by regionally specific increases in the volumes of posterior cortical, lateral, and central amygdalar nuclei (Haas et al. 2012). Though ostensibly a variably expressed brain phenotype, amygdala enlargement might be consistent with this brain region’s hypothesized role in the expansive social-affective traits in WS. Evidence for this assertion comes from studies 124
that have demonstrated associations between alterations of amygdala structure and rating scale measurements of social-affective behavior. For example, in an experiment by Martens and colleagues, WS adults generated approachability ratings to unfamiliar faces that were both greater than age- (but not IQ-)matched typically developed controls and were positively correlated with their right amygdalar volumes (Martens et al. 2009). These findings suggest that larger amygdalar volumes in Williams syndrome could be associated with a greater tendency to view faces as approachable, and thus may represent amygdalar participation in WS hypersociability. Jabbi et al. (2012) also reported a positive correlation between gray matter volume of the amygdala in WS individuals and WS personality profile scores. In sum, these results indicate that at least for some subset of WS individuals, structural alterations of the amygdala may hold functional relevance for the unusual social and affective processing hypothesized to occur in WS, and may be an important clue as to the neural scaffolding of this condition’s unique personality profile. In accord with this notion, functional neuroimaging studies have demonstrated differential engagement of the amygdala in individuals with WS, but in a manner that is sensitive to methodology. Studies employing neutral facial stimuli have reported either no differences in amygdalar response between WS and typically developed controls (Mobbs et al. 2004), or reduced amygdalar response for WS individuals during a facial identity-matching task (Paul et al. 2009). In contrast, studies employing either implicit (e.g., gender-discrimination task) (Haas et al. 2010; Thornton-Wells, Avery, and Blackford 2011) or explicit (e.g., emotion-matching task) emotion-processing tasks that feature negative-valence facial stimuli (Kippenhan et al. 2005; Mimura et al. 2010) have more consistently demonstrated amygdala hypoactivation. Consistent with the possibility that this abnormal amygdala activation may be an important correlate of the lack of social fear in WS, recent work has found that blunted amygdala activation to fearful faces shows positive covariation with social fearlessness during implicit emotion processing (Haas et al. 2010). The fact that measures of hypersociability did not show this relationship indicates the likelihood that amygdala hypoactivation is not necessarily relevant to all aspects of WS social behavioral traits. Indeed, diffuse, primary amygdala compromise seems clearly inadequate to explain the entirety of the WS socio-emotional phenotype, which is not simply an attenuated Kluver-Bucy syndrome, but rather includes a striking dichotomy between social disinhibition/drive and non-social anxiety. Using an emotion-matching task, P art I I I : I maging G enetics and G enetic D iscovery
Meyer-Lindenberg et al. (2005; see also Munoz et al., 2008) executed a pair of fMRI experiments to examine the neural substrate of this behavioral domain specificity, discovering that whereas individuals with WS show diminished amygdala responses to fearful faces, the same individuals show increased amygdala responses to threatening scenes (Meyer-Lindenberg et al. 2005b; Munoz et al. 2010). Furthermore, Haas and colleagues have also reported increased amygdala responses to happy faces in WS using fMRI (Haas et al. 2009). These findings emphasize a broad, dynamic range of context-dependent amygdala function in WS, which makes intrinsic, isolated amygdala lesion models of WS, if not inaccurate, at the least incomplete. The need for further explication has increasingly become evident in work examining connections between the amygdala and other important nodes in social-affective neural networks. Recent structural studies have suggested differences in the white matter tracts that connect the amygdala to other neural structures involved in social and affective processing. In particular, adults with WS have shown reductions in white matter integrity, as measured by fractional anisotropy, of the uncinate fasciculus (UF) (Arlinghaus et al. 2011; Avery et al. 2012; Jabbi et al. 2012), as well as ventral amygdalofugal pathways (Avery et al. 2012), the inferior fronto-occipital fasciculus (IFOF) (Arlinghaus et al. 2011; Avery et al. 2012), and the ILF (Arlinghaus et al. 2011; Avery et al. 2012) (though increased lattice index, another measure of fiber coherence, has been found in the IFO as well [Marenco et al. 2007], and one study has found increased FA in regions of the UF and IFO [Hoeft et al. 2007]). Haas et al. (2013) recently investigated differences in the integrity of white matter pathways between children with WS and typically developing children. In contrast with the adult studies, this work identified increases in fractional anisotropy of the right IFOF and bilateral UF, emphasizing the need for longitudinal studies to better understand the developmental trajectories of white matter in WS. The results of this analysis demonstrated altered white matter integrity in WS children in bilateral fusiform gyrus, bilateral amygdala, and the left hippocampus. These results suggest that the genetic abnormality in WS is associated with alterations in the white matter integrity of neural regions and tracts subserving social and affective processing, and indicate that these alterations are present early in development. Functional imaging experiments have further supported the hypothesis of abnormal connections with the amygdala in WS. In addition to reductions in activity of the amygdala, Sarpal et al. (2008) reported impaired functional connectivity between the amygdala and other neural chapter 8 : I maging G enetics of W illiams S yndrome
regions involved in the processing of social stimuli in adults with WS (Sarpal et al. 2008). In particular, the authors found diminished correspondence of BOLD time courses between bilateral amygdala and functionally defined fusiform face area (FFA) in WS; they posited that this aberrant coupling may reflect impairment of face-specific bottom-up data flow from the fusiform to the amygdala and/or reciprocal affective modulatory signals failing to proceed. Either or both of these mechanisms could contribute to impaired social threat evaluation in WS individuals. This formulation is in line with the Haas et al. (2013) study of white matter composition in children with WS, which, in addition to the findings just cited, observed that both amygdala and fusiform regions show reduced white matter integrity (fractional anisotropy) in WS. Furthermore, in the MeyerLindenberg et al. (2005b) study, path analyses of regional BOLD responses to facial stimuli suggested an absence of normal covariance between amygdala and regulatory prefrontal cortical regions—particularly the orbitofrontal cortex—in WS, again in accord with structural white matter findings (Arlinghaus et al. 2011; Avery et al. 2012; Jabbi et al. 2012). In sum, these findings implicate not merely a “broken” amygdala, but rather an aberrantly regulated network of amygdala-linked sensory and regulatory neural regions involved in processing social and affective information, indicating that abnormalities in the functional interactions of these regions may contribute to the socialaffective phenotype of WS.
Orbitofrontal Cortex The evidence supporting a critical role for the orbitofrontal cortex (OFC) in this network is particularly strong. The OFC is an important mediator of social and affective processing and evaluation of the reinforcement value of stimuli (Adolphs 2003; Gottfried, O’Doherty, and Dolan 2003; Kringelbach and Rolls, 2004; O’Doherty 2004; Rolls 2004), and, as alluded to earlier, its regulation of the amygdala, to which it is structurally connected (Stefanacci and Amaral 2002; Ghashghaei, Hilgetag, and Barbas 2007), is important for mediating responses to emotionally and socially salient information (Ochsner et al. 2004; Phan et al. 2005; Murray and Izquierdo 2007; Indovina et al. 2011). OFC structural abnormalities have been observed in WS by multiple investigators; however, the nature of these structural alterations has been inconsistent across studies. Among voxel-based morphometry investigations, some have reported decreased gray matter volume in the OFC in WS individuals compared to typically developed controls (Meyer-Lindenberg 2004; Kippenhan et al. 2005); 125
however, other studies have reported increased gray matter volume (Reiss et al. 2004; Jabbi et al. 2012) or no differences (Chaing et al. 2007; Faria et al. 2011). To some degree, this discrepancy can be accounted for by the VBM approach used: when using Jacobian modulation and accounting for changes in brain shape (local expansion/contraction to the study template), data from Reiss et al. (2004), which, when originally analyzed without Jacobian modulation, suggested increased OFC gray matter volume, could be found to replicate the decreased OFC gray matter volume finding in WS (Eckert et al. 2006). Notably, the intraparietal gray matter result discussed earlier was retained regardless of methodology, implying that in the OFC, where substantial variability of shape and size in MRI-measured anatomy occurs, VBM-based measurements are particularly sensitive to methodology. Furthermore, the OFC finding in Jabbi et al. (2012), showing increased volume in WS, represented an extension of a cluster originating in the ventral anterior insula, discussed later in this chapter and distinct from other cited OFC finding locales. Studies employing surface-based analyses and measures of surface complexity have also reported OFC structural alterations in WS in the form of reduced sulcal depth (Kippenham et al. 2005; Van Essen et al. 2006), increased gyrification (Gaser et al. 2006), and both reduced surface area and increased cortical thickness in the OFC (Meda, Pryweller, and Thornton-Wells 2012). Furthermore, in addition to structural OFC abnormalities, several studies have reported abnormalities in white matter tracts connecting the OFC to the amygdala. Although several studies have reported structural alterations of orbitofrontal cortex in WS, as detailed above, only a couple have employed neuropsychological or other behavioral measures to explicitly link OFC structural alterations to the social-affective phenotype of WS. For instance, Gothelf et al. (2008) reported a positive correlation between the size of ventral anterior prefrontal cortex, which overlaps with the OFC, and the social use of language in persons with WS. In particular, those WS individuals with greater whole brain volume-adjusted ventral anterior prefrontal cortical volume—a measurement that was increased in WS, relative to control individuals—engaged in greater use of social evaluative devices, which are used to convey emotion and maintain the audience’s interest in the narrative. Likewise, Jabbi et al. (2012) demonstrated a positive correlation between gray matter volume of left lateral OFC and medial OFC and the WS personality profile scores described earlier in this text. These findings suggest an association between structural alterations of OFC and the extent of expression of social-affective behaviors in WS, but moreover, highlight the need for future experiments 126
to employ comprehensive phenotyping that includes neuropsychological and other behavioral measures to better understand hypothesized structure-behavior associations. Several experiments using PET and fMRI have suggested that the structural alterations noted earlier are accompanied by functional impairments in the OFC. An increase in resting regional cerebral flood flow (Jabbi et al. 2102) and altered OFC responses to affective stimuli (Meyer-Lindenberg et al. 2005b; Mimura et al. 2010) have been reported in WS at the sites of OFC gray matter volume alterations. In addition to the WS-associated differences in amygdalar responses to fearful/frightening (relative to neutral) social and non-social visual stimuli discussed earlier, Meyer-Lindenberg et al. (2005b) reported an absence of OFC activation for both of these stimulus conditions. Similarly, in a study using a face-matching task in a small sample of participants (9 typically developed controls and 7 individuals with WS), Mimura et al. (2010) found that the WS individuals showed less lateral OFC activation for angry (versus happy) facial stimuli, and they additionally reported greater medial OFC activation for happy (versus angry) facial stimuli in WS. Whether or not this observation suggests a disruption of usual medio-lateral organization of the OFC in WS, as the authors hypothesize, remains unclear (Kringelbach and Rolls 2004; Mimura et al. 2010), but on the whole, these neuroimaging studies implicating the OFC suggest a candidate neural correlate of abnormal social stimuli valuation in WS and invite further investigation into the functional organization of OFC circuitry and its impact on other, closely related socio-emotional network nodes. Indeed, as might be expected from the reported white matter tract abnormalities reported in bilateral ventral amygdalofugal pathways (Avery et al. 2012), the UF (Jabbi et al. 2010; Arlinghaus et al. 2011; Avery et al. 2012; Haas et al. 2013), the ILF (Marenco et al. 2007; Arlinghaus et al. 2011; Avery et al. 2012), and the IFOF (Marenco et al. 2007; Arlinghaus et al. 2011; Avery et al. 2012; Haas et al. 2013), functional coupling of the OFC with other neural regions involved in processing of emotionally and socially salient information has also been demonstrated to be impaired in WS. Using an anatomically informed model of prefrontal-amygdala circuitry and path analysis of the above-cited fMRI data, Meyer-Lindenberg et al. (2005b) additionally reported an absence of normal functional interactions between the OFC and both the amygdala and dorsolateral prefrontal cortex in WS individuals during processing of threatening faces. Furthermore, reduced functional coupling of the OFC with the FFA (Sarpal et al. 2008) and with the ventral anterior insula (Jabbi et al. P art I I I : I maging G enetics and G enetic D iscovery
2012) has been reported, augmenting the likelihood that the OFC’s regulatory function within the socio-emotional network may be aberrant in WS and a potential driver of atypical social and affective responses in WS.
Insular Cortex The insula plays an important role in social and emotional processing, particularly related to anxiety, empathy, and the integration of social, cognitive, and affective information (Damasio et al. 2000; Wright et al. 2003; Paulus and Stein 2006; Jabbi, Swart, and Keysers 2007; Seeley et al. 2007; Craig 2009; Singer, Critchley, and Preuschoff 2009). As these aspects of social and emotional processing are affected in WS, several studies have investigated whether the WS brain phenotype is characterized by structural and functional alterations of the insula. Both increases (Reiss et al. 2004; Eckert et al. 2006; Menghini et al. 2011; Jabbi et al. 2012) and decreases (Reiss et al. 2004; Cohen et al. 2010; Jabbi et al. 2012) in gray matter volume and density of insular cortex have been reported, and some studies (Meyer-Lindenberg et al. 2004) have also reported no differences in insular cortical volume between WS and typically developed control individuals (Chiang et al. 2007; Faria et al. 2011). Some of the inconsistency in these findings may be explained by local regional volumetric variation within the insula. Jabbi et al. (2012) demonstrated decreased gray matter volume in WS individuals in bilateral dorsal anterior and middle insula, as well as a region of increased gray matter volume in the right ventral anterior insula, compared to an age-, sex-, and IQ-matched control sample. The authors also showed that structural alterations of the insula have functional consequences for social and affective processing in WS, as the gray matter volume increase in the right ventral anterior insula of WS participants was found to be positively correlated with the degree to which each participants evidenced the WS personality profile. In addition to volumetric alterations of the insula, surface-based analyses have revealed deeper sulcal depths (Van Essen et al. 2006) and decreased surface area (Meda, Pryweller, and Thornton-Wells 2012) in the insula of individuals with WS. Further support for the insula as a neural substrate of social and affective impairments in WS comes from findings of altered insula function in WS. In addition to the structural alterations of insular cortex noted earlier, Jabbi et al. (2012) found reduced regional cerebral blood flow in the left dorsal anterior insula, right dorsal middle/posterior insula, and middle insula in persons with WS compared to typically developed control participants. Additionally, chapter 8 : I maging G enetics of W illiams S yndrome
increased regional cerebral blood flow was found in the right ventral anterior insula of WS individuals, corresponding to the same region in which increased gray matter volume was also found. Similar to the correlation between altered insular structure and WS personality profile scores, increased regional cerebral blood flow in the right ventral anterior insula of WS individuals was found to be positively correlated with their WS personality profile scores. Furthermore, the authors demonstrated alterations in the functional coupling of the anterior insula with other neural regions, including the OFC, amygdala, medial prefrontal cortex, and anterior cingulate cortex, compared to typically developed control individuals. This finding of altered functional connectivity corresponds with the structural finding of altered white matter integrity of the UF, which is an important conduit of information flow between the insula and both amygdala and OFC (Arlinghaus et al. 2011; Avery et al. 2012; Jabbi et al. 2012). Together, these findings suggest that the WS 7q11.23 deletion and/or associated developmental events may impact the structural and functional properties of the insula, as well as this structure’s connections to other neural regions involved in social and affective processing. In light of the literature supporting an important role for the insula in emotional awareness and the integration of emotion and cognition, to the extent that these abnormalities have shown association with the WS personality profile, they may, along with those observed in the closely linked amygdala and OFC (see summary Figure 8.2 and Table 8.2), ultimately help identify a potential neural substrate for the phenotype of social and affective processing impairments characteristic of WS, thereby providing a means to explore neurgenetic mechanisms.
ADDITIONAL SYSTEMS IN WS
Domain-General Cognitive Systems Though investigations of the visual and social systems reflect the two main corpora of WS imaging experimentation, it is important to recognize that there are other aspects of the clinical syndrome that remain without as well-defined neural correlates but that nonetheless deserve investigation. This is especially true of phenotypes considered less unique to WS, such as a leftward shift in the distribution of full-scale IQ scores not solely attributable to poor visuospatial functioning (Mervis et al. 2012). The widespread adoption by neuroimaging researchers of various strategies to control for IQ, as noted earlier, belies the general recognition that uncontrolled IQ variability may 127
SOCIO-EMOTIONAL NEURAL SYSTEMS
(a)
Structural Findings: Meyer-Lindenberg et al. (2004) Reiss et al. (2004) Menghini et al. (2011) Jabbi et al. (2012)
R
L
(b)
Functional Findings: Meyer-Lindenberg et al. (2005a) Haas et al. (2009) Paul et al. (2009)
Munoz et al. (2010) Mimura et al. (2011) Thornton-Wells et al. (2011) R
L
Jabbi et al. (2012)
Figure 8.2 SOCIO-EMOTIONAL NEURAL SYSTEMS. Localization of selected findings in socio-emotional neural systems (amygdala, orbitofrontal cortex, and
insular cortex) reflecting differences between WS and TD groups in (A) structure (gray matter volume or gray matter density) and (B) function (neural activation or regional cerebral blood flow). Loci indicate approximate locations of peak coordinates of findings and are color-coded by study. Size of loci does not represent extent of structural or functional findings.
impact imaging measurements of neural systems function in WS and may obscure some genetic effects of interest. This impact may be due to biases in how individuals with lower IQ might engage with cognitive tasks during functional neuroimaging assays or due to intrinsic neural differences from which the IQ impairments arise. Despite the latter’s importance for a fuller mechanistic understanding of genotype-phenotype relationships in WS—and potentially in other neurodevelopmental syndromes with intellectual disability, the neuroimaging signature of IQ impairment in WS remains unclear. That is in part because there is a dearth of well-powered, incisive neuroimaging studies specifically dedicated to understanding the range of general intellectual disability in WS. In selected works, secondary whole-brain exploratory analyses have examined IQ correlations with various structural and functional measures, albeit with mixed results and limited power. For instance, a tensor-based morphometric experiment found that in 18 WS individuals older than 30 years (but not in 19 younger WS adults), volumes of numerous diverse brain regions were positively correlated with performance but not verbal IQ scores (Chiang et al. 2007). Additionally, in a small cohort of WS children (n = 10), scores on the Snijders-Oomen Nonverbal Intelligence test, which includes a visuospatial component, have shown positive relationships with frontoparietal cortical volumes and surface area, and negative correlations with occipital gyrification indices (Fahim et al. 2012). Functional MRI studies have not offered any broader perspective on this issue, often limiting the search space and 128
number of observations with which to detect relationships with IQ, resulting in positive findings with suboptimal scope and negative findings with poor power. One recent study failed to find multiple comparison-corrected statistically significant relationships between IQ scores and region of interest BOLD response to hearing action sounds versus non-action sounds, hearing sounds versus silence, viewing angry faces versus fixation cross, or viewing negative versus neutral images, but study cohorts contained no more than 16 WS individuals (Pryweller et al. 2012). Likewise, three even smaller studies, unique in their execution of whole-brain, voxel-wise queries for full-scale IQ effects, reported the absence of significant within-group IQ effects on BOLD signal during response inhibition, face viewing, and global versus local shape discrimination, though notably, behavioral performance during imaging was predicted by IQ in the response inhibition study (Mobbs et al. 2007a; Mobbs et al. 2007b; Mimura et al. 2010). The remaining past reports examining IQ effects in WS did so largely in post hoc, limited region-of-interest analyses to bolster the claimed specificity of reported findings to the deletion itself, rather than general intellectual functioning deficits (Jackowski and Schultz 2005; Haas et al. 2009; Haas et al. 2012). An exception is one intriguing study using magnetic resonance spectroscopy that identified diminished N-acetyl aspartate—a measure of neural cell integrity—relative to choline and creatine in the cerebellum of WS individuals and found that this abnormality corresponded to broad cognitive deficits including standard IQ scores (Rae et al. 1998). Though not yet followed up, this result P art I I I : I maging G enetics and G enetic D iscovery
TABLE 8.2
Year
SUMMARY OF NEUROIMAGING LITERATURE ON SOCIO-EMOTIONAL NEURAL SYSTEMS IN WILLIAMS SYNDROME Authors
Sample Characteristics
Demographic Matching
Type of Imaging
Task/Neuropsychological Measure/Rating Scale
Main Findings
Brain-Behavior Correlation
WS: Significant (+) correlation between approachability ratings for negative faces and Amyg volume.
AMYGDALA Structural Findings 2009
Martens et al.
n = 44 children and adults (22 WS, 22 TD)
Age, sex, handedness
MRI
Adolph’s approachability rating scale
WS > TD: Amyg volume (adjusted for total brain volume) and higher approachability ratings.
2011
Capitao et al.
n = 33 children and adults (17 WS, 16 TD)
Age, sex
MRI
None
WS > TD: Amyg volume (adjusted for total brain volume).
2011
Arlinghaus et al.
n = 32 adults (16 WS, 16 TD)
Age, sex
DTI
None
WS > TD: FA in IFO/ILF, SLF, and UF. TD > WS: FA in splenium of corpus collosum, corona radiata, external capsules, cortico-spinal/ cortico-cerebellar tracts from pons through posterior limb of external capsules, and UF/ IFO, and SLF.
2012
Haas et al. (^)
n = 79 children and adults (39 WS, 40 TD)
Age, sex, handedness
MRI
WISC-IV or WASI
WS > TD: Amyg volume (adjusted for total brain volume).
No significant correlations between IQ and adjusted Amyg volume for TD or WS.
WS: regional specificity of structural alterations within Amyg. 2012
2012
2013
Meda, Pryweller, and Thornton Wells
n = 81 adults (31 WS, 50 TD)
Avery et al. (&)
n = 17 adults (8 WS, 9 TD)
Age, sex, handedness, race, non-social fear
DTI
n = 36 children (18 WS, 18 TD)
Age, sex
DTI
Hass et al.
Age, sex
MRI
Beck Anxiety Inventory (WS)
WS > TD: gray matter volume of FuG, Amyg, and OFC. TD > WS: surface area of inferior and superior Par cortices and Hipp.
KBIT-II
WS: No significant correlations between Beck Anxiety Inventory Scores and neural activation after correcting for multiple comparisons.
WS > TD: FA in region of IFO. TD > WS: FA in ventral Amygdalofugal pathways, UF, ILF, IFO.
Social Cognition subscale, Affect Recognition subscale, WISC-IV
WS > TD: FA values in social brain regions (e.g., FuG, Amyg) and related white-matter tracts (e.g., IFO and UF).
WS: Significant (+) correlation between FA values in right FuG and Social Cognition scores. No significant correlations between FA values in any regions and Affect Recognition scores or IQ. (continued)
TABLE 8.2
Year
CONTINUED Authors
Sample Characteristics
Demographic Matching
Type of Imaging
Task/Neuropsychological Measure/Rating Scale
Main Findings
N = 26 adults (13 WS, 13 TD)
Age, sex, IQ
fMRI
Emotion matching task
WS > TD: Amyg activation for threatening scenes.
Brain-Behavior Correlation
Functional Findings 2005
Meyer-Lindenberg et al. (†)
TD > WS: Amyg activation for threatening faces. WS: Impaired functional interactions between Amyg, PFC, and OFC. 2009
2009
Paul et al.
Haas et al. (^)
n = 34 adults (17 WS, 17 TD) and 17 c hildren (TD: mental age controls)
Age (chronological and mental), sex
fMRI: n = 27 adults (14 WS, 13 TD)
Age, sex
fMRI
Facial identity-matching task
TD (adults and children) > WS: Amyg activation. TD adults > WS and TD children: ventral Occ-Temp cortex activation.
fMRI and ERP
Implicit emotion-processing task
ERP: n = 70 (30 WS, 25 TD, 15 developmentally delayed)
WS > TD: Amyg activation for happy faces. TD > WS: Amyg activation for fearful faces.
2010
Haas et al. (^)
n = 12 WS adults
NA
fMRI
Implicit emotion-processing task and Salk Institute Sociability Questionnaire
Significant (-) correlation between social approach to strangers scores and Amyg activation for fearful faces.
2010
Munoz et al. (†)
n = 26 adults (13 WS, 13 TD)
Age, sex, IQ
fMRI
Emotion-matching task and emotion-labeling task
WS > TD: Amyg activation for fearful scenes (matching and labeling tasks). TD > WS: DLPFC activation for labeling task.
2011
Thornton-Wells et al. (&)
n = 30 adults (10 WS, 10 TD with extreme inhibited temperament, 10 TD Age, sex, handedness, race with extreme unhibited temperament)
Age, sex, handedness, race
fMRI
Implicit emotion-processing task; Novelty Seeking and Harm Avoidance subscales of Tridimensional Personality Questionnaire, State scale of State-Trait Anxiety Inventory, Social Phobia and A goraphobia subscales of Social Phobia and Anxiety Inventory, KBIT-II
WS > TD (inhibited temperament group): Amyg activation for fearful nonsocial images. WS > TD (uninhibited temperament group): Amyg activation for fearful and neutral social images.
No significant correlation between empathic sociability social expressions. No significant correlation between social approach scores and Amyg activation for all facial expressions.
ORBITOFRONTAL CORTEX Structural Findings 2008
Gothelf et al.
n = 79 adults (39 WS, 40 TD)
Age, sex
MRI
Social language use narrative task
WS > TD: volume of ventral anterior PFC and bending angle of corpus calosum.
WS: Significant (+) correlation between social language use and gray matter volume of ventral anterior PFC.
n = 14 adults (7 WS, 7 TD)
Age, sex
fMRI
Emotion matching task and WAIS-III
WS > TD: medial OFC activation for negative faces.
No significant correlations between IQ and neutral activation (whole brain analyses) for TD or WS.
Functional Findings 2010
Mimura et al.
TD > WS: Amyg and lateral OFC activation for negative faces. INSULAR CORTEX Structural Findings 2010
Cohen et al.
n = 22 adults (11 WS, 11 TD)
Age
MRI
None
TD > WS: total, anterior, and posterior volume of insula (adjusted for total brain volume).
2012
Jabbi et al. (†)
Structural MRI: n = 37 adults (14 WS, 23 TD)
Age, sex, IQ
PET, DTI, and MRI
Multidimensional Personality Questionnaire, WASI (WS), and WAIS-R (TD)
WS > TD: ventral anterior insula volume and regional cerebral blood flow.
DTI: n = 10 adults (5 WS, 5 TD) PET: n = 28 adults (14 WS, 14 TD) NOTE : For all tables ^, †, &, and # symbols indicate overlap in participant pool across different studies.
TD > WS: dorsal anterior insula volume and regional cerebral blood flow; Hipp formation/Amyg regional cerebral blood flow; FA of UF.
WS: Significant (+) correlation between WS personality profile scores and increase in ventral anterior insula volume and blood flow. No significant correlations between Verbal IQ WASI scores and insula PET or VBM measures.
is an interesting footnote to structural studies suggesting enlarged or preserved cerebellar volumes in WS (Schmitt et al. 2001; Chaing et al. 2007). Thus, overall, a coherent picture of neural systems implicated in the general intellectual disability occurring in many persons with WS has not yet emerged. One explanation is simply that there is not a parsimonious, syndrome-specific, measurable neural correlate of the domain-general cognitive deficiency that variably accompanies the WS critical region hemideletion. The experiments reported to date clearly do not provide sufficient evidence to support this hypothesis, but it is possible that there may be too many heterogeneous, independent neural factors responsible for determining the often mild to moderate extent of composite intellectual disability in the WS population to permit reliable signal detection with the necessarily limited sample sizes available for study. Alternatively, it may be that there are empirically discoverable neural correlates of domain-general cognitive impairment in WS. Because the genotype in WS is unique, but this particular behavioral phenotype (low IQ) is shared by other conditions, it is unclear whether such correlates will ultimately prove to be syndrome-specific. Either way, their identification would represent a step closer to understanding at least one neurogenetic route to intellectual disability, which may have valuable implications both for and beyond WS. In order to solve for these unknowns, the investigative approach typified by the current literature must be refined. In addition to improved sample sizes and more dedicated studies, removal of factors that may be adding noise to the measurement of domain-general cognition may prove important, such as any potentially independent effects of visuospatial construction impairment, which is consistently observed even in normal-range IQ WS individuals (Morris et al. 2003; Meyer-Lindenberg 2004). Statistically controlling for visuospatial aptitude or using IQ tests that do not rely on visuospatial construction (e.g., KBIT2; Pryweller et al. 2012) could be useful in this regard, as could functional neuroimaging paradigms designed for greater sensitivity to general IQ effects. Furthermore, many of the findings observed in WS but not clearly linked to domain-specific (e.g., visuospatial construction) deficits could be considered candidate phenotypes for testing (see summary Table 8.3).
Other Neural Systems Findings in Williams Syndrome Such candidate phenotypes are not rare in the WS literature and may lend insight into the broader ramifications of the WS deletion. Examples from structural neuroimaging 132
experiments range from broad, grossly noted observations such as overall cerebral volume reductions (Jernigan and Bellugi 1990; Galaburda et al. 2001; Schmitt et al. 2001; Wengenroth et al. 2010), to specific, diversely quantified midline white matter abnormalities. Regarding the latter, diminished volume and length of the corpus callosum in WS relative to age- (but not IQ-)matched, typically developed controls have been reported in adults (Schmitt et al. 2001a; Schmitt et al. 2001b)as well as mixed-age cohorts (Tomaiuolo et al. 2002; Luders et al. 2007; Martens et al. 2013; Sampaio et al. 2013). Several studies have also investigated subtler measurements of callosal morphology, finding a larger bending angle (Schmitt et al. 2001a; Tomaiuolo et al. 2002; Sampaio et al. 2013), greater posterior curvature (Sampaio et al. 2013), and increased relative thickness (Sampaio et al. 2013) in WS. Additional observations in other neural systems in WS merit mention and remain open areas for research. One experiment found diminished frontostriatal activation to response inhibition during a simple go-no-go task (Mobbs et al. 2007a). This initial result, originating from 11 WS individuals and age- (but not IQ-)matched controls, has yet to be replicated, leaving the field with the question of whether such findings (i.e., those in neural systems without strong a priori implication) represent an underappreciated facet of WS per se or less specific correlates of epiphenomena. In contrast to this solitary finding, several studies have examined auditory anatomy and processing in WS, generally inspired by anecdotal reports of greater affinity for music, though developmental adaptation to chronic hyperacusis may be just as relevant to their findings. Greater planum temporale volumes in WS have been reported and relate to both standardized academic test scores and musical pitch challenge performance (Martens, Reutens, and Wilson 2010). More recent work has replicated the increased auditory cortex volume finding, additionally identifying augmented gyrification in this region, resulting in a posterior shift of auditory stimulus-evoked activation by fMRI and magnetoencephalography (MEG) measurements. This same study further noted an enrichment for holistic (versus spectral) sound perception in WS and a corresponding leftward lateralization in auditory evoked P50 responses during MEG. Additionally, left-sided evoked responses were greater in WS than typically developed individuals (Wengenroth et al. 2010). These findings suggest that aberrancies in sensory systems in WS may not be limited to the visual domain, but do not directly speak to whether there is anything different about the manner in which individuals with WS perceive or process more P art I I I : I maging G enetics and G enetic D iscovery
TABLE 8.3
Year
SUMMARY OF NEUROIMAGING LITERATURE ON ADDITIONAL NEURAL SYSTEMS IN WILLIAMS SYNDROME Authors
Sample Characteristics
Demographic Matching
Type of Imaging
Task/Neuropsychological Main Findings Measure/Rating scale
Brain-Behavior Correlation
DOMAIN-GENERAL COGNITIVE SYSTEMS Structural Findings 2005
Jackowski and Schultz
n = 70 adults and children (28 WS, 22 TD, 20 TD IQ matches)
Age, sex, IQ
MRI
WISC-III
WS > TD: distance between dorsal extension of central sulcus and interhemispheric fissure.
No significant correlation with WISC scores found.
2007
Chiang et al. (^)
n = 80 children and adults (41 WS, 39 TD)
Age, sex
MRI
WISC-R and WAIS-R
WS > TD: volume in PFC, OFC, ACC, inferior Par cortex at ParOcc junction, STG, Amyg, Hipp, FuG, cerebellum.
WS: Significant (+) correlation between performance IQ scores and volume of several regions (including ACC, corpus callosum, Par, Occ, and PFC).
TD > WS: volume in Par and Occ lobes, Thal, BG, and midbrain (adjusted for total brain volume). 2012
Fahim et al.
n = 22 children (10 WS, 12 TD)
Age, sex
MRI
Snijders-Oomen Nonverbal Intelligence (SON) Tests and Vineland Social Maturity Scale (VSMS)
WS > TD: gyrification of Par lobe TD > WS: cortical volume and surface area in WS; cortical complexity of Fron and Par lobes; cortical thickness of Par lobe
WS & TD: Significant (+) correlations between SON scores, cortical volume. Significant (-) correlations between SON scores and gyrification index in Occ and Par cortices. Significant (+) correlations between VSMS scores and surface area of Fron and Temp cortices.
Functional Findings 2012
Pryweller et al. (&)
n = 29 WS children and adults
NA
fMRI
KBIT-II
No significant correlations between IQ and fMRI activation in extant datasets after correction for multiple comparisons.
n = 23 children and adults (6 WS, 3 Down Syndrome, 14 TD)
None
MRI
None
WS > TD: area of neocerebellar vermal lobules.
n = 30 WS children and adults
NA
OTHER NEURAL SYSTEMS Structural Findings 1990
1997
Jernigan and Bellugi
Brinkmann et al.
TD > WS: total cerebral volume MRI
None
No characteristic neural abnormalities noted in WS. (continued)
TABLE 8.3
CONTINUED
Year
Authors
Sample Characteristics
Demographic Matching
Type of Imaging
Task/Neuropsychological Main Findings Measure/Rating scale
2001
Schmitt et al.
n = 40 adults (20 WS, 20 TD)
Age, sex
MRI
None
WS > TD: posterior cerebellar vermis size (after controlling for intracranial area)
2001
Galaburda et al. (^)
n = 42 adults (21 WS, 21 TD)
Age, sex
MRI
None
WS = TD: ventral extent of central sulcus
Brain-Behavior Correlation
TD > WS: likelihood of dorsal central sulcus reaching medial brain surface. 2002
Tomaiuolo et al. (#)
n = 24 children and adults (12 WS, 12 TD)
Age, sex, handedness
MRI
None
WS > TD: convexity of corpus callosum, voxel intensity in mid- and caudal regions of body of corpus callosum. TD > WS: volume of corpus callosum.
2007
Luders et al. (#)
n = 24 children and adults (12 WS, 12 TD)
Age, sex, handedness
MRI
None
TD > WS: length, curvature, and thickness of corpus callosum.
2010
Martens et al.
n = 54 children and adults (25 WS, 29 TD)
Age, sex, handedness
MRI
Specimen Aural Tests (SAT) and Bentley Measures of Musical Abilities
WS > TD: planum temporale volume. WS = TD: planum temporale asymmetry and primary auditory cortical volumes.
Functional and Structural Findings 2003
2006
Levitin et al.
Mobbs et al. (^)
n = 10 adults (5 WS, 5 TD)
n = 20 adults (11 WS, 9 TD)
age, sex, handedness, musical experience
fMRI
Age, sex
fMRI
Passive listening task (music, noise, and baseline rest conditions)
WS > TD: Amyg activation for music.
Go/No-Go Response Inhibition Task
TD > WS: striatum, DLPFC, and dorsal ACC activation.
TD > WS: Temp lobe activation for music.
WS: Significant (+) correlations between planum temporal volume, total SAT, and Pitch #2 subset scores. No significant correlations for TD group temporale volumes or between temporale volumes and total Bentley scores for WS or TD.
2010
Thornton-Wells et al. (&)
Study 1: n = 26 children and adults (13 WS, 13 TD).
Age, Sex
fMRI
Study 2: 6 WS children and adults from Study 1.
Passive listening task, retinotopic mapping procedures, and color localizer
WS > TD: Occ (e.g., lingual gyrus, cuneus), Temp (e.g., STG), and insula activation for music.
No correlations between Occ lobe activation and musicality measures for WS or TD.
TD > WS: posterior cingulate, Thal, and ITG activation for music.
Study 3: 4 WS children and adults from Study 2.
WS: activation in early visual cortex for musical and non-musical auditory stimuli. 2010
Wengenroth et al.
MRI: n = 31 children and adults (11 WS, 20 TD)
Age, sex, musical training
MEG, fMRI, and MRI
MEG: n = 30 children and adults (10 WS, 20 TD)
Passive listening task (instruments and complex tones)
WS (fMRI): activation in Heschl’s gyrus extending to posterior duplications.
fMRI: n = 3 WS children and adults with duplications of Heschl’s gyrus 2013
Lense et al.
n = 26 adults (16 WS, 13 TD)
WS > TD: volume of Heschl’s gyrus (auditory cortex), incidence of complete posterior Heschl’s gyrus duplications.
Age, sex, handedness, MIS subscale or total scores
EEG
Cross-modal affective priming task, Music Interest Scale (MIS)
WS > TD: alpha power at frontal-central electrodes for happy musical primes, frontal-central gamma band activity for emotionally congruent music face pairs.
WS: Significant (+) correlations between MIS Emotional Reaction to Music subscale score (parent-reported) and alpha power difference between happy and sad musical primes.
NOTE : For all tables ^, †, &, and # symbols indicate overlap in participant pool across different studies. ACC = anterior cingulate cortex; Amyg = amygdala; BG = basal ganglia; Cre = creatine; CS = collateral sulcus; DD = developmentally delayed; DLPFC = dorsolateral prefrontal cortex; FA = fractional anisotropy; FFA = fusiform face area; FOF = fronto-occipital fasciculus; Fron = frontal; FuG = fusiform gyrus; Hipp = hippocampus; IFO = inferior fronto-occipital fasciculus; ILF = inferior longitudinal fasciculus; IPS = intraparietal sulcus; ITG = inferior temporal gyrus; KBIT = Kaufman Brief Intelligence Test; MPFC = medial prefrontal cortex; MTL = medial temporal lobe; NAA = N-acetylaspartate; Occ = occipital; OFC = orbitofrontal gyrus; Par-Occ = parieto-occipital; Par = parietal; Pcu = precuneus; PFC = prefrontal cortex; PHG = parahippocampal gyrus; PPA = parahippocampal place area; SLF = superior longitudinal fasciculus; SMA = supplementary motor area; STG = superior temporal gyrus; STS = superior temporal sulcus; TD = typically developed; Temp = temporal; Thal = thalamus; UF = uncinate fasciculus; WAIS = Wechsler Adult Intelligence Scale; WASI = Wechsler Abbreviated Scale of Intelligence; WISC = Wechsler Intelligence Scale for Children; WS = Williams syndrome
complex auditory stimuli. In a small, initial experiment measuring BOLD responses to music (versus noise), five individuals with WS did show enhanced amygdala and reduced superior/middle temporal gyri activation (Levitin et al. 2003). In a provocative follow-up study, however, these results were not corroborated. Instead, in addition to increased activation in insula, temporal cortex, and cerebellum, an unexpected, novel finding of abnormally increased extrastriate visual cortical responses to music was identified in many of the WS participants. A subset of these WS participants was studied again with a visual (color stimuli) localizer, and music-induced activations were confirmed (Thornton-Wells et al. 2010). In this publication, the authors are careful not to label WS individuals as obligatory synesthetes, and in fact, post-scan interviewing suggests the possibility that a propensity to relate to an unstructured, musical experience with diverse, dynamic imagery may play a role. Nonetheless, taken together, this interesting collection of work invites further investigation into the neural processing of both emotional and acoustical elements of complex auditory stimuli. CONCLUSION Williams syndrome is a rare condition in which hemizygosity of a small group of genes generates a distinctive neurocognitive and neurobehavioral profile that is accompanied by correlative dorsal stream and socioaffective network aberrancies, respectively. Because the causitive candidate gene set is small and unambiguous and the cognitive and behavioral phenotype is relatively stereotyped and easily measured, this fascinating syndrome provides an invaluable opportunity to study and define the relationships between genes, the neural networks they are programmed to act upon, and the complex behaviors that these neural systems mediate. Despite the inherent difficulty in recruiting large cohorts of affected individuals, substantial effect sizes and concerted neuroimaging experimentation have successfully provided in vivo evidence for specific anomalies in both these systems, supporting biologically plausible models linking a genetic defect, brain sequelae, and meaningful real-world functioning, and providing compelling brain phenotypes for exploration of neurogenetic mechanisms. Though the majority of this progress has occurred in defining the Williams syndrome neural phenotype, future advances must reach back to the molecular level to better understand from which genes, and through what
136
mechanisms, these system-level read-outs arise. In parallel to the multifaceted neural aspects of the 7q11.23 syndrome, the answer, too, may be manifold. For instance, mice lacking one copy of the GTF2I gene, which resides in the WS deleted region, appear to demonstrate a pro-social, but not visuospatial, phenotype consistent with Williams syndrome (Sakurai et al. 2011). Interestingly, duplication of this gene in animal models results in social anxiety, in accord with the phenotype observed in humans with 7q11.23 duplication (Mervis et al. 2012). However, disruption of GTF2IRD1, also in the telomeric portion of the deletion region (Li et al. 2009), may result in a diminished fear and aggression phenotype as well (Young et al. 2008). Complexity at the level of the neural phenotype can also be anticipated: diminished corpus callosal volume has been observed in CLIP2 (aka CYLN2; another WS deletion gene) knock-out mice (Hoogenraad et al. 2002; van Hagen et al. 2007), whereas increased corpus callosal volume accompanies GTF2IRD1 knock-out (van Hagen et al. 2007). CLIP2 knock-out mice additionally show diminished contextual fear conditioning and hippocampal synaptic plasticity (van Hagen et al. 2007). The complexity of these animal data emphasizes the need for translational work in humans, which should take advantage of matured neuroimaging technologies and individuals with partial WS-region deletions. Clinical reports have already identified several such individuals and have demonstrated variable cognitive and personality phenotypes (Heller et al. 2003; Morris et al. 2003; van Hagen et al. 2007; Smith et al. 2009; Antonell et al. 2010; Karmiloff-Smith et al. 2012), though debate as to causative molecules continues (Gray et al. 2006). Concerted neuroimaging of these individuals may be able to more sensitively identify relevant neurogenetic signatures and reliable genotype-phenotype associations. Future work must also be directed toward assessing the potential contribution of gene-gene interactions and gene-environment interactions to this fascinating syndrome, and must include particular attention to understanding the developmental trajectory of the above-mentioned phenotypes. The insights resulting from such investigations, coupled with those already provided by the WS neuroimaging genetics studies reviewed in this chapter, will undoubtedly yield a crucial guide for understanding how a singular meiotic error might yield a clinically compelling, multi-systemic neurodevelopmental condition, and perhaps, more generally, help trace at least one instructive path from molecules to mind.
P art I I I : I maging G enetics and G enetic D iscovery
AC K N OW L E D G M E N T The authors are supported by the Intramural Research Program, National Institute of Mental Health, National Institutes of Health, Bethesda, Maryland. REFERENCES Adolphs, R. (2003). Cognitive neuroscience of human social behaviour. Nat Rev Neurosci. 4: 165–178. Adolphs R, Tranel D, Damasio AR. (1998). The human amygdala in social judgment. Nature. 393: 470–474. Adolphs R, Tranel D, Damasio H, Damasio A. (1994). Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature. 372: 669–672. Anderson AK, Phelps EA. (2001). Lesions of the human amygdala impair enhanced perception of emotionally salient events. Nature 411, 305–309. Antonell A, et al. (2010). Partial 7q11.23 deletions further implicate GTF2I and GTF2IRD1 as the main genes responsible for the Williams-Beuren syndrome neurocognitive profile. J Med Genet. 47: 312–320. Arlinghaus LR, Thornton-Wells TA, Dykens EM, Anderson AW. (2011). Alterations in diffusion properties of white matter in Williams syndrome. Magn Reson Imaging. 29: 1165–1174. Avery SN, Thornton-Wells TA, Anderson AW, Blackford JU. (2012). White matter integrity deficits in prefrontal-amygdala pathways in Williams syndrome. Neuroimage. 59: 887–894. Bellugi U, Adolphs R, Cassady C, Chiles M. (1999). Towards the neural basis for hypersociability in a genetic syndrome. Neuroreport. 10: 1653–1657. Bellugi U, Bihrle A, Jernigan T, Trauner D, Doherty S. (1990). Neuropsychological, neurological, and neuroanatomical profile of Williams syndrome. Am J Med Genet. Suppl 6: 115–125. Bellugi U, Lichtenberger L, Jones W, Lai Z, St George MI. (2000). The neurocognitive profile of Williams Syndrome: a complex pattern of strengths and weaknesses. J Cogn Neurosci. 12(Suppl 1): 7–29. Beunders G, et al. (2010). A triplication of the Williams-Beuren syndrome region in a patient with mental retardation, a severe expressive language delay, behavioural problems and dysmorphisms. J Med Genet. 47: 271–275. Boddaert N, et al. (2006). Parieto-occipital grey matter abnormalities in children with Williams syndrome. Neuroimage. 30: 721–725. Botta A, et al. (1999). Detection of an atypical 7q11.23 deletion in Williams syndrome patients which does not include the STX1A and FZD3 genes. J Med Genet. 36: 478–480. Campbell LE, et al. (2009). Brain structural differences associated with the behavioural phenotype in children with Williams syndrome. Brain Res. 1258: 96–107. Capitao L, et al. (2011). MRI amygdala volume in Williams Syndrome. Res Dev Disabil. 32: 2767–2772. Chiang MC, et al. (2007). 3D pattern of brain abnormalities in Williams syndrome visualized using tensor-based morphometry. Neuroimage. 36: 1096–1109. Cohen JD, et al. (2010). Morphometry of human insular cortex and insular volume reduction in Williams syndrome. J Psychiatr Res. 44: 81–89. Collins RT, 2nd. (2013). Cardiovascular disease in Williams syndrome. Circulation. 127: 2125–2134. Craig AD. (2009). How do you feel—now? The anterior insula and human awareness. Nat Rev Neurosci. 10: 59–70.
chapter 8 : I maging G enetics of W illiams S yndrome
Damasio AR, et al. (2000). Subcortical and cortical brain activity during the feeling of self-generated emotions. Nat Neurosci. 3 1049–1056. Doyle TF, Bellugi U, Korenberg JR, Graham J. (2004). “Everybody in the world is my friend”: hypersociability in young children with Williams syndrome. Am J Med Genet A. 124A: 263–273. Dutly F, Schinzel A. (1996). Unequal interchromosomal rearrangements may result in elastin gene deletions causing the Williams-Beuren syndrome. Hum Mol Genet. 5: 1893–1898. Dykens EM. (2003). Anxiety, fears, and phobias in persons with Williams syndrome. Dev Neuropsychol. 23: 291–316 (2003). Eckert MA, et al. (2005). Evidence for superior parietal impairment in Williams syndrome. Neurology. 64: 152–153. Eckert MA, et al. (2006). To modulate or not to modulate: differing results in uniquely shaped Williams syndrome brains. Neuroimage. 32: 1001–1007. Einfeld SL, Tonge BJ, Florio T. (1997). Behavioral and emotional disturbance in individuals with Williams syndrome. AJMR. 102: 45–53. Eisenberg D, Kleinman J, Berman K. (2013). Epistasis and complex neuropsychiatric phenotypes. Psychiatry. 76: 293–295. Eisenberg R, Young D, Jacobson B, Boito A. (1964). Familial supravalvular aortic stenosis. Am J Dis Child 108: 341–347. Engell AD, Haxby JV, Todorov A. (2007). Implicit trustworthiness decisions: automatic coding of face properties in the human amygdala. J Cogn Neurosci. 19: 1508–1519. Ewart AK, Jin W, Atkinson D, Morris CA, Keating MT. (1994). Supravalvular aortic stenosis associated with a deletion disrupting the elastin gene. J Clin Invest. 93: 1071–1077. Fahim C, et al. (2012). Williams syndrome: a relationship between genetics, brain morphology and behaviour. J Intellect Disabil Res. 56: 879–894. Faria AV, et al. (2011). Quantitative analysis of brain pathology based on MRI and brain atlases—applications for cerebral palsy. Neuroimage. 54: 1854–1861. Faria AV, et al. (2012). Quantitative analysis of gray and white matter in Williams syndrome. Neuroreport. 23: 283–289. Fitzgerald DA, Angstadt M, Jelsone LM, Nathan PJ, Phan KL. (2006). Beyond threat: amygdala reactivity across multiple expressions of facial affect. Neuroimage. 30: 1441–1448. Fryssira H, et al. (1997). Fluorescent in situ hybridisation (FISH) for hemizygous deletion at the elastin locus in patients with isolated supravalvular aortic stenosis. J Med Genet. 34: 306–308. Galaburda AM, Bellugi UV. (2000). Multi-level analysis of cortical neuroanatomy in Williams syndrome. J Cogn Neurosci. 12(Suppl 1): 74–88. Galaburda AM, et al. (2001). Dorsal forebrain anomaly in Williams syndrome. Arch Neurol. 58: 1865–1869. Galaburda AM, Holinger DP, Bellugi U, Sherman GF. (2002). Williams syndrome: neuronal size and neuronal-packing density in primary visual cortex. Arch Neurol 59: 1461–1467. Gaser C, et al. (2006). Increased local gyrification mapped in Williams syndrome. Neuroimage 33: 46–54. Ghashghaei HT, Hilgetag CC, Barbas H. (2007). Sequence of information processing for emotions based on the anatomic dialogue between prefrontal cortex and amygdala. Neuroimage. 34: 905–923. Golarai G, et al. (2007). Differential development of high-level visual cortex correlates with category-specific recognition memory. Nat Neurosci. 10: 512–522. Golarai G, et al. (2010). The fusiform face area is enlarged in Williams syndrome. J Neurosci. 30: 6700–6712. Golarai G, Liberman A, Yoon JM, Grill-Spector K. (2010). Differential development of the ventral visual cortex extends through adolescence. Front Hum Neurosci. 3: 80. Gothelf D, et al. (2008). Association between cerebral shape and social use of language in Williams syndrome. Am J Med Genet A. 146A: 2753–2761.
137
Gottfried JA, O’Doherty J, Dolan RJ. (2003). Encoding predictive reward value in human amygdala and orbitofrontal cortex. Science 301: 1104–1107. Gray V, Karmiloff-Smith A, Funnell E, Tassabehji M. (2006). In-depth analysis of spatial cognition in Williams syndrome: a critical assessment of the role of the LIMK1 gene. Neuropsychologia. 44: 679–685. Haas BW, Sheau K, Kelley RG, Thompson PM, Reiss AL. (2012). Regionally specific increased volume of the amygdala in Williams syndrome: Evidence from surface-based modeling. Hum Brain Mapp. 35(3): 866–874. Haas BW, et al. (2009). Genetic influences on sociability: heightened amygdala reactivity and event-related responses to positive social stimuli in Williams syndrome. J Neurosci. 29: 1132–1139. Haas BW, et al. (2010). Individual differences in social behavior predict amygdala response to fearful facial expressions in Williams syndrome. Neuropsychologia. 48: 1283–1288. Haas BW, et al. (2012). Preliminary evidence of abnormal white matter related to the fusiform gyrus in Williams syndrome: a diffusion tensor imaging tractography study. Genes Brain Behav. 11: 62–68. Haas BW, et al. (2013). Altered microstructure within social-cognitive brain networks during childhood in Williams syndrome. Cereb Cortex. 24(10): 2796–2806. Haxby JV, et al. (1991). Dissociation of object and spatial visual processing pathways in human extrastriate cortex. Proc Natl Acad Sci U S A. 88: 1621–1625. Heller R, Rauch A, Luttgen S, Schroder B, Winterpacht A. (2003). Partial deletion of the critical 1.5 Mb interval in Williams-Beuren syndrome. J Med Genet. 40: e99. Hoeft F, et al. (2007). More is not always better: increased fractional anisotropy of superior longitudinal fasciculus associated with poor visuospatial abilities in Williams syndrome. J Neurosci. 27: 11960–11965. Holinger DP, et al. (2005). Relative sparing of primary auditory cortex in Williams syndrome. Brain Res. 1037: 35–42. Hoogenraad CC, et al. (2002). Targeted mutation of Cyln2 in the Williams syndrome critical region links CLIP-115 haploinsufficiency to neurodevelopmental abnormalities in mice. Nat Genet. 32: 116–127. Indovina I, Robbins TW, Nunez-Elizalde AO, Dunn BD, Bishop SJ. (2011). Fear-conditioning mechanisms associated with trait vulnerability to anxiety in humans. Neuron. 69: 563–571. Jabbi M, Swart M, Keysers C. (2007). Empathy for positive and negative emotions in the gustatory cortex. Neuroimage. 34: 1744–1753. Jabbi M, et al. (2012). The Williams syndrome chromosome 7q11.23 hemideletion confers hypersocial, anxious personality coupled with altered insula structure and function. Proc Natl Acad Sci U S A. 109: E860–866. Jackowski AP, Schultz RT. (2005). Foreshortened dorsal extension of the central sulcus in Williams syndrome. Cortex. 41: 282–290. Jernigan TL, Bellugi U. (1990). Anomalous brain morphology on magnetic resonance images in Williams syndrome and Down syndrome. Arch Neurol. 47: 529–533. Karmiloff-Smith A, et al. (2012). Social cognition in williams syndrome: genotype/phenotype insights from partial deletion patients. Front Psychol. 3: 168. Kippenhan JS, et al. (2005). Genetic contributions to human gyrification: sulcal morphometry in Williams syndrome. J Neurosci 25: 7840–7846. Klein-Tasman BP, Mervis CB. (2003). Distinctive personality characteristics of 8-, 9-, and 10-year-olds with Williams syndrome. Dev Neuropsychol. 23: 269–290. Knierim JJ, Lee I, Hargreaves EL. (2006). Hippocampal place cells: parallel input streams, subregional processing, and implications for episodic memory. Hippocampus. 16: 755–764. Koscik TR, Tranel D. (2011). The human amygdala is necessary for developing and expressing normal interpersonal trust. Neuropsychologia. 49: 602–611. 138
Kringelbach ML, Rolls ET. (2004). The functional neuroanatomy of the human orbitofrontal cortex: evidence from neuroimaging and neuropsychology. Prog Neurobiol. 72: 341–372. Levitin DJ, et al. (2003). Neural correlates of auditory perception in Williams syndrome: an fMRI study. Neuroimage. 18: 74–82. Li HH, et al. (2009). Induced chromosome deletions cause hypersociability and other features of Williams-Beuren syndrome in mice. EMBO Mol Med. 1: 50–65. Lieberman MD. (2007). Social cognitive neuroscience: a review of core processes. Annu Rev Psychol 58: 259–289. Luders E, et al. (2007). Callosal morphology in Williams syndrome: a new evaluation of shape and thickness. Neuroreport. 18: 203–207. Marenco S, et al. (2007). Genetic contributions to white matter architecture revealed by diffusion tensor imaging in Williams syndrome. Proc Natl Acad Sci U S A. 104: 15117–15122. Martens MA, Reutens DC, Wilson SJ. (2010). Auditory cortical volumes and musical ability in Williams syndrome. Neuropsychologia. 48: 2602–2609. Martens MA, Wilson SJ, Chen J, Wood AG, Reutens DC. (2013). Handedness and corpus callosal morphology in Williams syndrome. Dev Psychopathol. 25: 253–260. Martens MA, Wilson SJ, Dudgeon P, Reutens DC. (2009). Approachability and the amygdala: insights from Williams syndrome. Neuropsychologia. 47: 2446–2453. Meda SA, Pryweller JR, Thornton-Wells TA. (2012). Regional brain differences in cortical thickness, surface area and subcortical volume in individuals with Williams syndrome. PLoS One. 7: e31913. Menghini D, et al. (2011). Relationship between brain abnormalities and cognitive profile in Williams syndrome. Behav Genet. 41: 394–402. Merla G, Brunetti-Pierri N, Piccolo P, Micale L, Loviglio MN. (2012). Supravalvular aortic stenosis: elastin arteriopathy. Circulation Cardiovasc Genet 5: 692–696. Mervis CB, Becerra AM. (2007). Language and communicative development in Williams syndrome. Ment Retard Dev Disabil Res Rev. 13: 3–15. Mervis CB, Kistler DJ, John AE, Morris CA. (2012). Longitudinal assessment of intellectual abilities of children with Williams syndrome: multilevel modeling of performance on the Kaufman Brief Intelligence Test-Second Edition. Am J Intellect Dev Disabil. 117: 134–155. Mervis CB, Klein-Tasman BP. (2000). Williams syndrome: cognition, personality, and adaptive behavior. Ment Retard Dev Disabil Res Rev. 6: 148–158. Mervis CB, et al. (2000). The Williams syndrome cognitive profile. Brain Cogn. 44: 604–628. Mervis CB, et al. (2003). Attentional characteristics of infants and toddlers with Williams syndrome during triadic interactions. Dev Neuropsychol. 23: 243–268. Mervis CB, et al. (2012). Duplication of GTF2I results in separation anxiety in mice and humans. Am J Hum Genet. 90: 1064–1070. Meyer-Lindenberg A, et al. (2004). Neural basis of genetically determined visuospatial construction deficit in Williams syndrome. Neuron. 43: 623–631. Meyer-Lindenberg A, et al. (2005a). Functional, structural, and metabolic abnormalities of the hippocampal formation in Williams syndrome. J Clin Invest. 115: 1888–1895. Meyer-Lindenberg A, et al. (2005b). Neural correlates of genetically abnormal social cognition in Williams syndrome. Nat Neurosci. 8: 991–993. Mimura M, et al. (2010). A preliminary study of orbitofrontal activation and hypersociability in Williams Syndrome. J Neurodev Disord 2, 93–98 (2010). Mishkin M, Ungerleider LG, Macko KA. (1983). Object vision and spatial vision—2 cortical pathways. Trends Neurosci. 6: 414–417. P art I I I : I maging G enetics and G enetic D iscovery
Mobbs D, et al. (2004). Anomalous brain activation during face and gaze processing in Williams syndrome. Neurology. 62: 2070–2076. Mobbs D, et al. (2007a). Frontostriatal dysfunction during response inhibition in Williams syndrome. Biol Psychiatry. 62: 256–261. Mobbs D, et al. (2007b). Reduced parietal and visual cortical activation during global processing in Williams syndrome. Dev Med Child Neurol. 49: 433–438. Morris CA, et al. (2003). GTF2I hemizygosity implicated in mental retardation in Williams syndrome: genotype-phenotype analysis of five families with deletions in the Williams syndrome region. Am J Med Genet A. 123A: 45–59. Morris JS, et al. (1998). A neuromodulatory role for the human amygdala in processing emotional facial expressions. Brain. 121(Pt 1): 47–57. Munoz KE, et al. (2010). Abnormalities in neural processing of emotional stimuli in Williams syndrome vary according to social vs. non-social content. Neuroimage. 50: 340–346. Murray EA, Izquierdo A. (2007). Orbitofrontal cortex and amygdala contributions to affect and action in primates. Ann N Y Acad Sci. 1121: 273–296. Nicodemus, K. K. et al. (2010a). Biological validation of increased schizophrenia risk with NRG1, ERBB4, and AKT1 epistasis via functional neuroimaging in healthy controls. Arch Gen Psychiatry. 67: 991–1001. Nicodemus KK, et al. (2010b). Evidence of statistical epistasis between DISC1, CIT and NDEL1 impacting risk for schizophrenia: biological validation with functional neuroimaging. Hum Genet. 127: 441–452. Ochsner KN, et al. (2004). For better or for worse: neural systems supporting the cognitive down- and up-regulation of negative emotion. Neuroimage. 23: 483–499. O’Doherty JP. (2004). Reward representations and reward-related learning in the human brain: insights from neuroimaging. Curr Opin Neurobiol. 14: 769–776. O’Hearn K, et al. (2011). Object recognition in Williams syndrome: uneven ventral stream activation. Dev Sci. 14: 549–565. Olsen RK, et al. (2009). Retinotopically defined primary visual cortex in Williams syndrome. Brain 132: 635–644. Paton JJ, Belova MA, Morrison SE, Salzman CD. (2006). The primate amygdala represents the positive and negative value of visual stimuli during learning. Nature. 439: 865–870. Paul BM, et al. (2009). Amygdala response to faces parallels social behavior in Williams syndrome. Soc Cogn Affect Neurosci. 4: 278–285. Paulus MP, Stein MB. (2006). An insular view of anxiety. Biol Psychiatry. 60: 383–387. Perez Jurado LA, Peoples R, Kaplan P, Hamel BC, Francke U. (1996). Molecular definition of the chromosome 7 deletion in Williams syndrome and parent-of-origin effects on growth. Am J Hum Genet. 59: 781–792. Phan KL, et al. (2005). Neural substrates for voluntary suppression of negative affect: a functional magnetic resonance imaging study. Biol Psychiatry. 57: 210–219. Pinker, S. (1999). Words and rules: the ingredients of language, 1st ed. New York: Basic Books. Pober BR, Morris CA. (2007). Diagnosis and management of medical problems in adults with Williams-Beuren syndrome. Am J Med Genet C Semin Med Genet. 145C: 280–290. Porter MA, Shaw TA, Marsh PJ. (2010). An unusual attraction to the eyes in Williams-Beuren syndrome: a manipulation of facial affect while measuring face scanpaths. Cogn Neuropsychiatry. 15: 505–530. Pryweller JR, Avery SN, Blackford JU, Dykens EM, Thornton-Wells TA. (2012). The effect of intellectual ability on functional activation in a neurodevelopmental disorder: preliminary evidence from multiple fMRI studies in Williams syndrome. J Neurodev Disord. 4: 24. chapter 8 : I maging G enetics of W illiams S yndrome
Rae C, et al. (1998). Brain biochemistry in Williams syndrome: evidence for a role of the cerebellum in cognition? Neurology. 51: 33–40. Reiss AL, et al. (2000). IV. Neuroanatomy of Williams syndrome: a high-resolution MRI study. J Cogn Neurosci. 12(Suppl 1): 65–73. Reiss AL, et al. (2004). An experiment of nature: brain anatomy parallels cognition and behavior in Williams syndrome. J Neurosci. 24: 5009–5015. Riby DM, et al. (2011). Attention to faces in Williams syndrome. J Autism Dev Disord. 41: 1228–1239. Riby DM, Kirk H, Hanley M, Riby LM. (2013). Stranger danger awareness in Williams syndrome. J Intellect Disabil Res. 58(6): 572–582. Rolls ET. (2004). The functions of the orbitofrontal cortex. Brain Cogn 55: 11–29. Sakurai T, et al. (2011). Haploinsufficiency of Gtf2i, a gene deleted in Williams Syndrome, leads to increases in social interactions. Autism Res. 4: 28–39. Sampaio A, et al. (2010). Williams syndrome and memory: a neuroanatomic and cognitive approach. J Autism Dev Disord. 40: 870–877. Sampaio A, et al. (2013). Morphometry of corpus callosum in Williams syndrome: shape as an index of neural development. Brain Struct Funct. 218: 711–720. Sarpal D, et al. (2008). A genetic model for understanding higher order visual processing: functional interactions of the ventral visual stream in Williams syndrome. Cereb Cortex. 18: 2402–2409. Schmitt JE, Eliez S, Warsofsky IS, Bellugi U, Reiss AL. (2001). Enlarged cerebellar vermis in Williams syndrome. J Psychiatr Res. 35: 225–229. Schmitt JE, Eliez S, Bellugi U, Reiss AL. (2001a). Analysis of cerebral shape in Williams syndrome. Arch Neurol. 58: 283–287. Schmitt JE, Eliez S, Warsofsky IS, Bellugi U, Reiss AL. (2001b). Corpus callosum morphology of Williams syndrome: relation to genetics and behavior. Dev Med Child Neurol. 43: 155–159. Schmitt JE, et al. (2002). Increased gyrification in Williams syndrome: evidence using 3D MRI methods. Dev Med Child Neurol. 44: 292–295. Seeley WW, et al. (2007). Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci. 27: 2349–2356. Singer T, Critchley HD, Preuschoff K. (2009). A common role of insula in feelings, empathy and uncertainty. Trends Cogn Sci. 13: 334–340. Smith AD, Gilchrist ID, Hood B, Tassabehji M, Karmiloff-Smith A. (2009). Inefficient search of large-scale space in Williams syndrome: further insights on the role of LIMK1 deletion in deficits of spatial cognition. Perception 38: 694–701. Soper R, et al. (1995). Ischemic stroke and intracranial multifocal cerebral arteriopathy in Williams syndrome. J Pediatr. 126: 945–948. Stefanacci L, Amaral DG. (2002). Some observations on cortical inputs to the macaque monkey amygdala: an anterograde tracing study. J Comp Neurol. 451: 301–323. Tam, E. et al. (2008). The common inversion of the Williams-Beuren syndrome region at 7q11.23 does not cause clinical symptoms. Am J Med Genet A. 146A: 1797–1806. Thompson PM, et al. (2005). Abnormal cortical complexity and thickness profiles mapped in Williams syndrome. J Neurosci. 25: 4146–4158. Thornton-Wells TA, Avery SN, Blackford JU. (2011). Using novel control groups to dissect the amygdala’s role in Williams syndrome. Dev Cogn Neurosci. 1: 295–304. Thornton-Wells TA, et al. (2010). Auditory attraction: activation of visual cortex by music and sound in Williams syndrome. Am J Intellect Dev Disabil. 115: 172–189. Tomaiuolo F, et al. (2002). Morphology and morphometry of the corpus callosum in Williams syndrome: a T1-weighted MRI study. Neuroreport. 13: 2281–2284. 139
Tomc SA, Williamson NK, Pauli RM. (1990). Temperament in Williams syndrome. Am J Med Genet. 36: 345–352. Udwin O, Yule W. (1991). A cognitive and behavioural phenotype in Williams syndrome. J Clin Exp Neuropsychol. 13: 232–244. Van Essen DC, et al. (2006). Symmetry of cortical folding abnormalities in Williams syndrome revealed by surface-based analyses. J Neurosci. 26: 5470–5483. van Hagen JM, et al. (2007). Contribution of CYLN2 and GTF2IRD1 to neurological and cognitive symptoms in Williams syndrome. Neurobiol Dis. 26: 112–124. Wang MS, et al. (1999). Molecular and clinical correlation study of Williams-Beuren syndrome: no evidence of molecular factors in the deletion region or imprinting affecting clinical outcome. Am J Med Genet, 86: 34–43. Wang Q, Gao E, Burkhalter A. (2011). Gateways of ventral and dorsal streams in mouse visual cortex. J Neurosci. 31: 1905–1918. Wengenroth M, Blatow M, Bendszus M, Schneider P. (2010). Leftward lateralization of auditory cortex underlies holistic sound perception in Williams syndrome. PLoS One 5. 5(8): e12326. Whalen PJ, et al. (1998). Masked presentations of emotional facial expressions modulate amygdala activity without explicit knowledge. J Neurosci. 18: 411–418. Whitlock JR, Sutherland RJ, Witter MP, Moser MB, Moser EI. (2008). Navigating from hippocampus to parietal cortex. Proc Natl Acad Sci U S A. 105: 14755–14762.
140
Williams JC, Barratt-Boyes BG, Lowe JB. (1961). Supravalvular aortic stenosis. Circulation 24: 1311–1318. Winston JS, Strange BA, O’Doherty J, Dolan RJ. (2002). Automatic and intentional brain responses during evaluation of trustworthiness of faces. Nat Neurosci. 5: 277–283. Wint, D. P. et al. (2013). Intracranial arteries in individuals with the elastin gene hemideletion of Williams syndrome. AJNR Am J Neuroradiol. 35(1): 90–94. Wollack JB, Kaifer M, LaMonte MP, Rothman M. (1996). Stroke in Williams syndrome. Stroke 27: 143–146 (1996). Woodruff-Borden J, Kistler DJ, Henderson DR, Crawford NA, Mervis CB. (2010). Longitudinal course of anxiety in children and adolescents with Williams syndrome. Am J Med Genet C Semin Med Genet. 154C: 277–290. Wright CI, Martis B, McMullin K, Shin LM, Rauch SL. (2003). Amygdala and insular responses to emotionally valenced human faces in small animal specific phobia. Biol Psychiatry. 54: 1067–1076. Young EJ, et al. (2008). Reduced fear and aggression and altered serotonin metabolism in Gtf2ird1-targeted mice. Genes Brain Behav 7: 224–234. Zhong YM, Rockland KS. (2003). Inferior parietal lobule projections to anterior inferotemporal cortex (area TE) in macaque monkey. Cereb Cortex. 13: 527–540.
P art I I I : I maging G enetics and G enetic D iscovery
9. IDENTIF YING UNANTICIPATED GENES AND MECHANISMS IN SERIOUS MENTAL ILLNESS GWA S - BASED IM AGING GE NE T IC S S T R AT EGIE S
Steven G. Potkin, Theo G. M. van Erp, Shichun Ling, Fabio Macciardi, and Xiaohui Xie
I N T RO DU C T I O N Genome-wide association studies (GWAS) investigate the relationship or association between genetic markers, usually single-nucleotide polymorphisms (SNPs), along the entire genome with categorical (e.g., diagnoses) or quantitative traits (e.g., imaging phenotypes). In contrast, candidate gene association studies investigate the relationship between genetic markers at a particular locus (e.g., a gene hypothesized to have biological relevance for a particular disorder) and either categorical or quantitative traits. The GWAS approach has an advantage compared to the candidate gene approach because it does not require a priori knowledge of a trait’s etiological connection, or its function singly or as a part of a network (Potkin et al. 2009a; Stranger, Stahl, and Raj 2011); thus this approach can lead to unanticipated discoveries of genes associated with mental illness. While the advantages of this approach apply to mental illness in general, we will confine our examples to schizophrenia and bipolar subjects, as several SNPs are risk factors for the development of both disorders and for some of their shared clinical characteristics. We would prefer to include only relevant studies with both bipolar and schizophrenia subjects; however, given that there are very few such studies, we will include studies with either disorder. Why combine genetic and imaging data? Mental disorders are brain disorders that can be studied with functional brain imaging measures, which are known to be highly heritable. Studying brain imaging measures in neuropsychiatry without considering genetics neglects part of the risk for developing mental disorders such as schizophrenia and bipolar disorder. Similarly, studying genetics
in psychiatric disorders without determining their brain effects fails to understand (at least part of) the consequences of these genetic influences. The rationale for considering both schizophrenia and bipolar disorder is due to their overlap in clinical symptoms, medication response, and risk genes (Tesli et al. 2014), although we are aware that the two disorders also have significant differences in these domains. The approach to jointly study multiple psychiatric syndromes is in accord with the National Institute of Mental Health’s Research Domain Criteria (RDoC) approach of examining possible common mechanisms across disorders. The Psychiatric Genomics Consortium (PGC) cross-diagnosis working group has identified six candidate regions that are shared risk factors for both schizophrenia and bipolar disorders (Ruderfer et al. 2014). However, these data to date lack a functional phenotype that is likely to be critical in understanding the candidate regions’ role in these disorders. In this chapter we selectively review studies that combine both genome-wide genetic and functional imaging data in schizophrenia and/or bipolar disorder. An advantage of the GWAS approach is that it surveys the entire genome for associations between genetic markers and phenotypes (e.g., diagnosis, imaging, and/or clinical characteristics). Adequate SNP coverage is important, and in current studies at least 750,000 SNPs proportionately distributed across the genome are needed for detailed association studies. With imputation of SNPs absent in early SNP chips used for genotyping, this has become a realistic goal. Rich multimodal quantitative imaging phenotype data offer the potential of linking genetic risk with functional brain consequences and have several additional advantages over case-control studies that will be discussed
141
later in this chapter. We did not include linkage studies, which survey the whole genome with less than 10,000 SNP markers, because of the large number of genes contained in a typical linkage peak. We searched the PubMed database with the following criteria: the “All Fields” field contained “imaging,” “fMRI,” “functional MRI,” “resting state,” “resting state fMRI,” “GWAS,” “genome-wide association study,” “genetics,” “schizophrenia” and/or “bipolar”; or “MeSH Major Topic” contained “MRI,” “fMRI,” “functional MRI,” “imaging,” “GWAS,” “genome-wide association study,” “schizophrenia” and/or “bipolar”; and various combinations of these terms. We integrated the search results and limited them to papers that contained both the “GWAS” and “imaging” stems. We manually reviewed all the abstracts prior to September 1, 2014, and identified 12 relevant functional imaging genetics publications related to schizophrenia and/or bipolar through this extensive search. We have summarized these publications in Table 9.1 (in the following section). B AC KG RO U N D The integration of imaging and genetic data has the potential to not only avoid the limitations of studying genetics or brain imaging in isolation but can leverage their synergism. Brain imaging has productively been used to clarify the brain effects of candidate genes. These candidate genes can be a priori chosen or identified in GWAS. One example is the study by Walton et al. (2013a), who examined the structural and functional imaging associations with neurogranin SNPs that were identified in a GWAS of 12,945 schizophrenia and 34,591 healthy controls (Stefansson et al. 2009). Walton et al. (2013a) observed increased cortical inefficiency and increased gray matter thinning in schizophrenia patients with the neurogranin risk alleles. Prefrontal cortical inefficiency and gray matter thinning are both well-documented findings in schizophrenia. Walton et al. (2013a) linked these brain findings with specific neurogranin genetic risk, thereby moving toward potentially clarifying an underlying etiological mechanism. Brain imaging can also be used as a quantitative trait to discover unexpected risk genes for schizophrenia and bipolar disorder (see Table 9.1 and Figure 9.1). For example, in a sample of less than 200 subjects, the quantitative trait (QT) dorsolateral prefrontal cortex (DLPFC) BOLD activation measured during a working memory task
142
identified several unanticipated risk SNPs for schizophrenia. These statistically significant SNPs included TNIK (TRAF2 and NCK-interacting kinase) (Potkin et al. 2009b). Interestingly, the previously identified candidate gene DISC1 modulates TNIK kinase activity. DISC1 and TNIK are co-localized in dendritic spines of the hippocampus. Together they co-regulate synaptic GLUR1 and AMPA activity (Wang et al. 2011). Dysregulation in both receptors has been hypothesized in schizophrenia (Harrison and Weinberger 2005) and bipolar disorder (Wang et al. 2011). In addition to these biological mechanistic implications, a QT approach has distinct advantages in power over categorical diagnoses (i.e., healthy control vs. schizophrenia and/or bipolar disorder). QT approaches have approximately 4–10 times more statistical power (Purcell, Cherny, and Sham 2003; Potkin et al. 2009a; Shen et al. 2014) by making use of the entire distribution of trait values. This approach also avoids the often arbitrary or error-prone cutoff distinctions, for example, thresholds or criteria for the diagnosis of schizophrenia are different in ICD-10, DSM-IV, and DSM-5. In addition, a GWAS-QT analysis, compared to a GWAS-CT (categorical trait) analysis, offers an alternative strategy to discover unanticipated genes associated with schizophrenia and/or bipolar risk as described for TNIK. In identifying TNIK, DLPFC activation during a working memory task was the starting place, as aberrant activation is well-described in schizophrenia. The GWAS SNPs that were statistically significantly associated with the degree of DLPFC activation were then identified. This approach began with a brain imaging characteristic of schizophrenia and/or bipolar disorder and then identified genes (or SNPs, or other types of genetic variation) associated with that imaging phenotype. Such intermediate phenotypes may have greater sensitivity in clarifying the functional links related to the schizophrenia and/or bipolar risk genes than diagnoses, in part because they are quantitative and possibly also because they are closer to the biological mechanisms underlying the illness than symptom patterns. Symptoms or cognitive performance may be affected by many factors in addition to genetic risk, such as medication and social factors, as well as modifying genes. One option to avoid the potential confounds of medication and disease is to study subjects at genetic or clinical high risk (CHR) for developing these disorders, usually non-ill relatives or at-risk adolescents (see Table 9.1). Studies of genetic risk influences in high-risk subjects have the advantages of avoiding medication and illness effects that can affect brain activation
P art I I I : I maging G enetics and G enetic D iscovery
and interpretation of the results. On the other hand, clarifying mechanisms of disease are the primary motivation for imaging genetics studies. Non-ill high-risk subjects who are past the age of illness onset have a genetic vulnerability, but the vulnerability has been insufficient to produce the disease, possibly because of the lack of additional genetic or environmental influences or the presence of compensatory mechanisms. Studies of adolescents at CHR or offspring from parents with the illness have an advantage because they can be followed longitudinally such that both risk- and disease-related effects (in those who become ill) can be studied. However, these studies take many years to complete and therefore require considerable time and financial investment. Others have chosen to investigate the function of GWAS-identified candidates by conducting brain imaging studies in healthy controls. These studies also avoid confounds of medication and disease, but may ignore the perturbations of genetic networks that are responsible for the risk liability for disease. Previous GWAS have provided candidates that have subsequently been studied with brain imaging, producing useful data regarding the brain effects of these candidate genes. We do not discuss these studies, as their designs are straightforward. We focus on publications that combine GWAS and brain imaging in a single study. These studies offer, in our view, the most robust strategy for identifying unanticipated genetic influences in serious mental illness (SMI) and for understanding the underlying mechanisms, while presenting the most challenging methodological and statistical issues. This chapter focuses on addressing those challenges and highlighting the promise of this method. S T RU C T U R A L A N D F U N C T I O N A L N E U RO I M AG I N G Structural neuroimaging has been the most studied of brain phenotypes in psychiatry. Brain-genome associations can be identified at several levels. From a genomic perspective, we can examine candidate genes (identified by surrogate SNPs), part of the genome involved in biological pathways or networks, the entire genome (GWAS), or the DNA sequence (e.g., exome, or whole genome seqeuncing [WGS]). Current multi-center collaborative research efforts such as ENIGMA (Enhancing Neuro Imaging Genetics through Meta Analysis) can leverage imaging and genetic data on thousands of subjects and are beginning to reveal the genomic underpinning of brain structural differences across disorders (Stein et al. 2012).
From an imaging perspective, individual region of interest (ROI), brain circuits including connectivity between multiple ROIs, or the whole brain (e.g., voxel-wise or cortical surface) activations can represent the entry point (phenotype/dependent measure[s]) of an investigation. Figure 9.2 shows examples of studies in each category. In this chapter we review fMRI (functional MRI) and rsfMRI (resting state fMRI) but not EEG/EVP/ERP, MEG, PET, SPECT, or fNIR because we were unable to find GWAS combined with these imaging methods in schizophrenia and/or bipolar subjects, although the discussion and points raised also apply to these methods. Structural imaging, which is discussed in detail in other chapters of this book, is a more indirect measure of brain function than physiological activations, though the relationship between structural and functional brain measures remains insufficiently defined. In this chapter, we focus on functional brain imaging, as it likely reflects a more direct measure of the functional consequences of genetic risk than those reflected by structural imaging, though some genetic variation may contribute to both or be associated with functional and not structural brain changes, and vice versa. Moreover, abnormal brain structure may result in abnormal brain function, and abnormal brain function may result in changes in brain structure, shape, and/or volume. However, we believe that to understand underlying mechanisms, functional changes may be more sensitive and have greater specificity than structural alterations. We review an unbiased whole genome approach using SNP associations with functional phenotypes. As genes do not function in isolation but rather function within pathways or networks, we also consider genetic association of functional networks/pathways with targeted functional phenotypes. All of these analyses require corrections for multiple testing to minimize the risk of false positives (type 1 errors) when searching millions of variants in our genetic code; therefore statistical threshold corrections are discussed first. I M AG I N G G E N E T I C S S TAT I S T I C S Genetic studies have begun to implement a high statistical threshold (e.g., p-value < 5×10 -8 in case-control studies) to implicate a genetic variant with a trait or a disorder. Currently there is considerable debate of what an appropriate statistical threshold is for imaging genetic findings to avoid type 1 errors without over-corrections. The widely accepted threshold (Risch and Merikangas 1996; Barsh et al. 2012) for a conventional GWAS is nominal
chapter 9 : I dentifying U nanticipated G enes and M echanisms
143
TABLE 9.1
IMAGING GENETICS PUBLICATIONS IN SCHIZOPHRENIA AND BIPOLAR DISORDER
Imaging
Author
Year
Cohort (n)
Imaging Cohort
fMRI Phenotype
Positive SNPs/Total
Top SNPs
Genes
Analysis
FA
Sprooten et al.
2013
70 high risk for BD; 80 HC
70 high risk for BD; 80 HC
Fractional anisotropy
730K
10 SNPs
LPP, HEPACAM, HEPN1, ROBO4, CAMK2D
Association
FA
Whalley et al.
2013
70 BD/ MDD family members; 62 HC
70 BD/ MDD family members; 62 HC
Fractional anisotropy
Polygenic risk score using 730K GWAS
PRS with FA in R superior longitudinal fasciculus in MDD but not BD
N/A
Association w/multidimensional scaling
Resting state
Martin et al.
2014
78 SZ
33 SZ
Resting state cognitive control and default network
900K
Total deletion burden
Resting state
Meda et al.
2014
296 SZ; 300 BD; 324 HC; 385 relatives
190 SZ; 189 BD; 170 HC
Resting state default mode network (DMN)
Motor
Chen et al.
2012
92 SZ; 116 HC
92 SZ; 116 HC
Motor task in pre/post central gyri activation
253 of 5157 selected from 778K
253 SNPs comprising an ICA component
Parallel imaging and genetic ICA
Auditory
Liu et al.
2012
48 SZ; 40 HC
48 SZ; 40 HC
Auditory oddball task in thalamus anterior and posterior cingulate gyri
300K
26
Parallel imaging and genetic ICA; association, permutation
Cognitive Control
Rietschel et al.
2012
1169 SZ; 3714 HC for GWAS
122 HC
Flanker cognitive control task in anterior cingulate
475K
top GWAS SNP rs11154491; rs7112229, rs11819869, rs7130141 and rs12575668
ARGHAP18; AMBRA1, DGKZ, and CHRM4
ANOVA
Faces
Ousdal et al.
2012
51 SZ; 64 BD; 94 HC
Face processing in amygdala
546K
rs10014254, rs11722038, rs17529323
PHOX2B
Association
Association
Parallel imaging and genetic ICA
Main Effect p-Value
p-Value of Interaction
10^-5 - 10^-6
Percent of Variance Explained
Results
PubMed URL
10^-5 - 10^-7; top 10 SNPs associated with FA within each group, but in opposite directions of effect.
KEGG: axon guidance, ErbB-signaling neurotrophin signaling, phosphatidylinositol signaling, and cell adhesion
http://www.ncbi. nlm.nih.gov/ pubmed/23218918
N/A
Sig negative association between MDD PRS and FA values in the right SLF, ILF, and IFOF in the parietal region (pFWE < 0.05) but not BD
Negative correlation between PRS* and FA in MDD but not BD
http://www.ncbi. nlm.nih.gov/ pubmed/23453289
p < .001
N/A
DLPFC and putamen cognitive control connectivity decreaed as deletion burden increased. Dysregulated DMN positively associated with deletion burden.
http://www.ncbi. nlm.nih.gov/ pubmed/25036426
3 significant DMNs (p = 4.8E-4; 8.7E-5;5.3E-7)
NS
Pathways: NMDA-related long-term potentiation, PKA, immune response signaling, axon guidance, and synaptogenesis
http://www.ncbi. nlm.nih.gov/ pubmed/24778245
r = .29, pLL
Domschke, 2006
5HT1A -1019/C/G
PD (20)
BOLD/fMRI
Faces
↓ACC, ↓OFC,↓vmPFC, GG < CG + CC ↑ amygdala to happy faces, GG > CG + CC
Faces
↑ amygdala to happy faces, SS + SL > LL
5-HTTLPR Domschke, 2008
COMT Val158Met
PD (20)
BOLD/fMRI
Faces
↑ amygdala, ↑ vmPFC, ↑ OFC, Val + Val/Val + Met > Met/Met
Furmark, 2008
5-HTTLPR TPH2-703 G/T
SAD (25)
rCBF/PET
Placebo responders during public speaking
↓ amygdala, LL > SS + SL ↓ amygdala, GG > TG + TT Additive effects
Furmark, 2009
5-HTTLPR TPH2-703 G/T
SAD (34) HC (18)
rCBF/PET
Faces
↑ amygdala, SS + SL > LL ↑ amygdala, TT + TG > GG Interactive effects
Lau, 2009
5-HTTLPR/rs25531
Anx (31) HC (33)
BOLD/fMRI
Faces
↑ amygdala, LALA > SS + SL > LL ↑ PFC LALA >SS + SL > LL ↑ Brainstem
Lau, 2010
BDNF Val66Met
Anx (31) HC (31)
BOLD/fMRI
Faces
↑ amygdala,↑ anterior hippocampus Met/Met + Met/Val > Val/Val
Domschke, 2011
Neuropeptide S receptor rs324981 A/T
PD (20)
BOLD/fMRI
Faces
↓dl PFC, ↓OFC, ↓ACC, AA > AT + TT
Morey, 2011
5-HTTLPR
PTSD (22)
BOLD/fMRI
Trauma-related images
↑ v/PFC (rs1695628-transporter) GG > CG + CC ↑ amygdala (SCL6A4) SS > SL + LL
Reif, 2013
MAO-VNTR
PD/AG (39)
BOLD/fMRI
Fear conditioning
↑ ACC to CS+ (low activity > high activity)
SAD = social anxiety disorder; PD = panic disorder; PD/AG = panic disorder with agoraphobia; HC = healthy controls; Anx = anxious/depressive adolescents
chapter 1 4 : I maging G enetics of A nxiety D isorders
225
TABLE 14.2
IMAGING GENETICS STUDIES OF BRAIN STRUCTURE IN PATIENTS WITH ANXIETY DISORDERS
First Author, Year
Polymorphism
Patients/Controls
Imaging Modality
Task
Main Results
Schulz-Heik, 2011
COMT Val158Met
PTSD (51) HC (48)
MR/ROI-based volumetric analyses
Rest
↓ ACC, Val/Val < Val/Met + Met/Met [in PTSD not HC]
Mueller, 2013
BDNF Val66Met
Anx (39) HC (63)
MR/Voxel-based morphometry
Rest
↑ Insula, ↑ dACC Val/Val > Val/Met + Met/Met [in patients not HC]
Kim, 2013
COMT Val158Met
PD (26) HC (26)
MR/Fractional anisotropy
Rest
↑ Connectivity, post thalamic radiation, post and sup corona radiation, sup long fasciculus, sagittal striatum Met/Met + Met/Val>Val/Val [in PTSD not HC]
Anx = adolescents with anxiety (mainly SAD related) + comorbid depression; PTSD = post-traumatic stress disorder; PD = panic disorder; HC = healthy controls
Three research groups have reported the majority of the imaging genetics studies in the anxiety disorders. Our group (Mats Fredrikson and Tomas Furmark) has studied patients with SAD using rCBF PET (Furmark et al. 2004; Furmark et al. 2008; Furmark et al. 2009) while Katarina Domschke and her coworkers have investigated panic disorder using BOLD/fMRI (Domschke et al. 2006; Domschke et al. 2008; Domschke et al. 2011). Daniel Pines laboratory has studied adolescents with anxiety comorbid with depression, mainly reflecting social anxiety (Lau et al. 2009; Lau et al. 2010; Mueller et al. 2013). There is also a PTSD study (Morey 2011) and a fear-conditioning experiment in PD patients using BOLD/fMRI (Straube et al. 2014). The majority of studies involved measures of perfusion determined by rCBF and PET or BOLD and fMRI. Only a few evaluated genetic effects on brain structure. Because a recent review of all neurochemical imaging data in the anxiety disorders (Fredrikson et al. 2013) consistently suggests anxiety-related alterations in serotonergic (reflected in 5HT1A receptors) and GABAergic (reflected in benzodiazepine receptors) signaling, it is noteworthy that no study evaluated allelic variations affecting neurotransmission in patients with anxiety disorders, but only in addiction, ADHD, depression, and schizophrenia (Willeit and Praschak-Reider 2010). This is an area where progress easily can be made, and it would be interesting to evaluate if the two genetic factors defined by qualitative genetic methodology (Hettema et al. 2005) reflect allelic differences corresponding to differences in neurotransmission or relate more to functional measures of neural activity in emotion-initiating and modulating areas. FUNCTIONAL IMAGING GENETICS OF SOCIAL ANXIETY DISORDER
The first imaging genetics study in anxious patients compared the effect of the low- and high-expressing short and
long serotonin transporter alleles (5-HTTLPR) on brain function during symptom provocation in social anxiety disorder. Patients and healthy controls performed a speech in the PET scanner with or without an observing audience (Tillfors et al. 2001), and the rCBF reaction was related to allelic variation in the 5HTTLPR in the patients (Furmark et al. 2004). We found increased amygdala reactivity among s-carriers as compared to ll homozygotes, replicating previous findings in healthy volunteers (Hariri et al. 2002; see also the review by Munafò et al. 2008). To determine if the previously reported amygdala hyper-reactivity to harsh emotional displays in social anxiety disorder (SAD; Stein et al. 2000) is modulated by serotonergic genes, we evaluated the effect of the 703 G/T polymorphism in the TPH2 gene and its interaction with the serotonin transporter in patients with SAD as compared to healthy controls (Furmark et al. 2009). The TPH2 gene controls the rate limiting step in brain serotonin synthesis and has previously been related to increased amygdala reactivity in healthy T allele carriers (Brown et al. 2005). We found that amygdala reactivity determined with PET and rCBF was increased among s-carriers of the 5-HTTLPR allele and also in carriers of the T allele of the TPH2 polymorphism (Furmark et al. 2009). Amygdala reactivity to emotional faces tended to be somewhat higher in individuals with SAD than in healthy controls (HC), but the genotypic influences accounted for more of the variability than did the phenotypic diagnosis (Furmark et al. 2009). While both the S allele of the 5-HTTLPR and the T allele of TPH2 were associated with increased amygdala reactivity, we found no interaction between health status and genotype, suggesting that, at least for SAD, the effect of the serotonergic genotypes on brain reactivity is similar in health and disorder. Our sample was too small to evaluate gene/gene interactions with any great power, but we found an interaction such that carriers of both the serotonin transporter S allele
2 2 6 part iv : I maging G enetics and D isease
and the T allele of the TPH2 had higher reactivity than all other gene combinations, suggesting epistatic modulation of amygdala reactivity to perceived threat. A subsequent study further supported that the S allele of the serotonin transporter and of the T allele of the TPH2 genes act like anxiety risk factors, as we observed that responders to placebo treatment in a double blind randomized control trial were more likely to be carriers of the ll or G alleles of the 5HTTLPR and TPH2 genes, respectively (Furmark et al. 2008). This was recently conceptually replicated in a study where SAD patients were treated with a selective serotonin reuptake inhibitor (SSRI) and S and T allele carriers showed a poor response, while ll and GG homozygotes had the best treatment response (Rotberg et al. 2013). In our study, genotype predicted decreased amygdala reactivity, and we also found a significant gene-gene interaction with carriers of both the 5-HTTLPR l and the TPH2-703 G allele displaying the most pronounced amygdala reductions. Using mediation analyses, we could demonstrate (Furmark et al. 2008) that the most likely path was from genes to brain to behavior, providing support for the hypothesis that the brain endophenotype is driven genetically, and that genetically modulated brain activity in turn influences behavior. In summary, all SAD imaging genetics studies support the notion that genes regulating serotonin signaling influence amygdala reactivity. Both the initially enhanced and symptomatically driven reactivity, as well as amygdala reductions following remission, are modulated by the 5HTTLPR and the TPH2 genes. This supports part one of the tripartite brain-imaging hypothesis, that areas initiating emotion should be up-regulated in anxiety, and that this is genetically modulated. As we used a region of interest approach and did not evaluate activity in inhibitory brain areas such as the vmPFC or the ACC, our data (Furmark et al. 2004; Furmark et al. 2008; Furmark et al. 2009) are silent with respect to the second part of the tripartite hypothesis on genetic effects on areas modulating fear-induced amygdala reactivity. This is not the case for panic disorder because both Domschke and her coworkers (Domschke et al. 2006; Domschke et al. 2008; Domschke et al. 2011) as well as Reif et al. (2013) reported data allowing evaluation of the hypothesis that genes not only modulate amygdala reactivity but also activity in regulatory prefrontal areas. FUNCTIONAL IMAGING GENETICS O F PA N I C D I S O R D E R
Domschke et al. (2006) evaluated the effect of the 5HT1A -1019/C/G polymorphism in the gene determining receptor chapter 1 4 : I maging G enetics of A nxiety D isorders
sensitivity in a group of panic disorder patients and found that amygdala reactivity was similar in exposure to fearful and aversive faces but was increased in exposure to happy faces in GG carriers (see Table 14.1). GG homozygotes also displayed reduced activity in the PFC and OFC as compared to C allele carriers. Thus, only limited support was obtained for increased amygdala reactivity, but support was given to the notion that serotonergic receptor genes shape activity in areas involved in emotional regulation (cf. Etkin 2010). Because amygdala reactivity was enhanced in exposure to happy faces in s-carriers, data suggest that serotonergic genes may modulate reactions to biologically relevant stimuli in general, rather than aversive events in particular. This is consistent with the view that amygdala performance is driven by biological salience (Davis and Whalen 2001). Domschke et al. (2008) evaluated the effect of the Val158Met allele in the COMT gene, and reported that Val carriers as compared to Met homozygotes displayed increased fear-related amygdala reactivity, but also that they were characterized by an increased reactivity in the vmPFC and the OFC. Because Val is a risk allele for panic disorder in Caucasian populations (Hamilton et al. 2002; Domschke et al. 2004; Rothe et al. 2006), the results support that the allelic risk is associated with increased reactivity in emotion-initiating areas like the amygdala, but are at odds with the idea that reactivity in modulatory areas is attenuated. Both animal and human studies suggest that the neuropeptide S and its receptor are involved in the pathogenesis of anxiety. Using a multilevel approach, Domschke et al. (2011) recently demonstrated that the risk allele (rs324981) is epidemiologically associated with panic disorder in women. In addition, allele presence was related to reports of anxiety sensitivity, increased heart rate, and symptomatology supporting its phenotypical influence. Endophenotypic brain activity in the dorsolateral prefrontal, lateral orbitofrontal, and anterior cingulate cortex during the processing of fearful faces was decreased in AA homozygotes as compared to T carriers. In essence, for women carrying the risk receptor allele, data did not support an increased reactivity in the amygdala but were consistent with compromised activity in modulatory brain areas. Judging from their figures (Domschke et al. 2011, Fig. 4, p. 944), amygdala reactivity was slightly but not significantly higher among risk allele carriers. Thus, on balance, data support that compromised emotional regulation in the fear network among women with panic disorder in part is determined by allelic differences in the neuropeptide S receptor gene. Generally, in panic disorder the imaging genetics studies support a major role for the 5HT1A polymorphism rather 227
than the serotonin transporter, and also that genetic variation in dopaminergic and noradrenergic pathways reflected in COMT alleles bias corticolimbic processing in panic disorder patients. Using fear conditioning, Reif et al. (2013) reported that patients with panic disorder with the long VNTR allele for the MAO gene, being the high active allele, displayed reduced reactivity in the vmPFC but not increased amygdala reactivity during fear conditioning, supporting that functional areas modulating emotional learning but not areas initiating emotional reactions are compromised. As enhanced vmPFC activity previously has been related to superior consolidation of extinction (Phelps et al. 2004; Milad et al. 2005), data support that serotonin-signaling genes modulate learning processes involved in anxiety acquisition and maintenance in panic disorder patients. FUNCTIONAL IMAGING GENETICS O F P O S T- T R A U M AT I C S T R E S S D I S O R D E R
Morey et al. (2011) reported that the presentation of trauma related images in patients with PTSD elicited higher amygdala reactivity in S carriers than LL homygotes of the 5HTTLPR and also that the rs1695628 transporter polymorphism was associated with enhanced activity in the ventral PFC. FUNCTIONAL IMAGING GENETICS IN ANXIETY DISORDERS
Collectively, there is strong support that carriers of the S allele of the 5HTTLPR display enhanced amygdala reactivity to emotional challenge, and this seems to characterize SAD rather than PD. In SAD, allelic differences in both the 5HTTLPR and the TPH2 genes modulate amygdala reactivity separately and also in an epistatic manner. In panic disorder, both dopaminergic and serotonergic genes, as well as genes associated with the neuropeptide S receptor, influence brain function. Averaged over PTSD, SAD, and PD, 8 of 11 contrasts support that amygdala reactivity is enhanced as an effect of allelic differences in putative susceptibility genes for anxiety (cf. Domshcke and Dannlowski 2012 for a review) and 6 comparisons suggest that reduced activity in areas modulating emotion also is linked to the polymorphism driving amygdala reactivity. In SAD there are no indications that individual differences in genetic makeup modulate brain activity differently than in healthy controls, whereas in patients with PD the effects of the COMT allele seem to differ according to diagnostic status. Domschke et al. (2008) reported that Val carriers with
PD were characterized by increased reactivity, while Smolka et al. (2005) found the presence of the Met allele associated with enhanced reactivity in healthy controls. Even though Smolka et al. (2005) used a different design with aversive pictorial pictures, while Domschke et al. (2008) used emotional faces, and even though studies differ in baseline measures, the major difference is whether patients or healthy controls were studied (Domschke and Dannlowski 2010). This implies that the modulatory effect of dopaminergic but not serotonergic polymorphisms differ according to diagnostic status or that the effect of the 5-HTTLPR and TPH2 genes, on the one hand, and COMT, on the other, interact with the diagnosis. We cannot determine the difference, as no direct comparison has been performed except for the serotonergic polymorphisms in SAD (Furmark et al. 2009). The tripartite hypothesis of brain mechanisms in anxiety claims that reactivity is up-regulated in emotioninitiating areas like the amygdala and insula cortex, down-regulated in emotional-controlling areas such as the prefrontal and orbitofrontal corticies, and finally that the coupling between initiating and controlling areas is compromised. Imaging genetics studies strongly support that serotonergic and dopaminergic signaling pathways contribute to enhanced amygdala reactivity, which is coupled to compromised activity in emotion-modulating areas, although with somewhat less support for the latter. There are no imaging genetics studies that have tested if the effective connectivity between emotion-initiating and controlling areas differs as a function of polymorphisms. IMAGING GE NE T IC S IN S T R U C T U R AL S T U DIE S Mueller and coworkers reported that gray matter volumes (GMVs) in the amygdala and the anterior hippocampal area were smaller in adolescents with anxiety symptomatology, predominantly reflecting SAD. The BDNF genotype modulated volume alterations in the insula and the dorsal ACC, being larger in Val/Val homozygotes than Met carriers. This modulation was specific to anxious individuals, as the effect was not found in healthy young persons. It is pointed out that several previously published functional brain studies indicate a role for not only the dorsal ACC but also the anterior insula in anxious reactions (Paulus and Stein, 2006; Shin and Liberzon 2010). Schulz-Heik and coworkers (2011) based their study on the structural correlates of the Val158Met COMT polymorphism on the fact that in PTSD, studies suggest both
2 2 8 part iv : I maging G enetics and D isease
a compromised ACC structure and function (Woodward et al. 2006). They (Schulz-Heik et al. 2011) reported that right ACC volume was reduced in symptomatic PTSD patients carrying Val/Val alleles, when compared both to non-symptomatic Val/Val homozygotes and to patients and controls being Met carriers. This indicates that the Val158Met polymorphism modulates ACC volume differently in PTSD and normal healthy volunteers. As twin studies suggest that ACC reductions follow traumatic events rather than being a pre-existing PTSD risk factor (Kasai et al. 2008), the authors suggested that a decreased amount of intrasynapatic dopamine increases the vulnerability for dystrophic effects of trauma. Kim and coworkers (2013) studied white matter connectivity using measures of fractional anisotropy related to the Val158Met COMT polymorphism in PD. They reported increased white matter connectivity in Met carriers as compared to Val homozygotes in the posterior thalamic radiation, posterior and superior corona radiata, superior longitudinal fasciculus, and sagittal striatum in the right hemisphere. This may reflect increased fiber density, fiber myelination, and increased directional coherence, all of which may contribute to rapid neural transmission (Shimony et al. 1999; Kim et al. 2013). One imaging genetics study (Reetz et al. 2008) demonstrated atrophy in the hippocampus and frontal structures, as an effect of a mutation in PINK1 gene associated with the onset of Parkinson’s disease, in a mixed group of psychiatric patients including those diagnosed with panic disorder, depression, and schizophrenia. In summary, there is suggestive, but not conclusive, evidence that volumetric alterations and structural connectivity relate to polymorphisms in BDNF and COMT genes, thereby linking structure to function in imaging genetics studies of anxiety disorders. FEA R C ONDI TI O N I N G AN D IM AGING GEN ETI CS There might be an association between fear conditioning and both structural and functional effects of serotonergic and dopaminergic genes, supporting the theory that polymorphic modulation of fear-conditioning processes might be mechanistically involved in determining the clinical genetic association. First, carriers of the S allele of the 5HTTLPR gene show superior fear conditioning (Garpenstrand et al 2001; Lonsdorf et al. 2009) and tend toward retarded extinction. S-carriers are also more amygdala reactive, and amygdala reactivity supports chapter 1 4 : I maging G enetics of A nxiety D isorders
fear conditioning (Sehlmeyer et al. 2009). Hence, it has recently been demonstrated that the increased fear conditioning observed in S-carriers is mediated by increased amygdala responses to the conditioned cue (Hermann et al. 2012; Klucken et al. 2013). Furthermore, poor extinction of fear conditioning, a likely anxiety phenotype, is found among Val-homozygotes of the Val158Met polymorphism of the COMT gene (Loonsdorf et al. 2009), which also is associated with increased amygdala reactivity in anxious patients (Domschke 2008). Also, polymorphisms in genes in the HPA-axis related to stress reactivity affect acquisition and extinction of fear conditioning through amygdala prefrontal interactions (Ridder et al. 2012). In addition, poor extinction recall, another likely anxiety phenotype, is linked to reduced ACC volume (Milad et al. 2005), which also characterizes PTSD patients being homozygous for the Val allele of the Val158Met COMT polymorphism. This pattern of results is not likely to occur by chance, but may instead point toward enhanced fear conditioning as being etiologically involved in the anxiety disorders. Other learning-determined processes of relevance for anxiety etiology, such as disruption of reconsolidation (Schiller et al. 2010), are also amygdala linked (Ågren et al. 2012a) and under serotonergic genetic modulation (Ågren et al. 2012b). FEAR CONDITIONING, IMAGING GENETICS, AND ANXIETY DISORDERS
Some studies support the theory that patients with anxiety disorders acquire fear conditioning more easily and retain it better than non-anxious controls. In PTSD, for example, initial studies reported increased fear conditioning (Orr et al. 2000), and more recent studies have indicated compromised extinction retention (Milad et al. 2009). Patients with panic disorder over-generalize conditioned fear (Lissek et al. 2009). In social anxiety, delayed extinction (Hermann et al. 2002) and stronger fear conditioning using emotional social stimuli that tap into symptomatology (Lissek et al. 2008) have been observed, and this seems amygdala mediated (Pejic et al. 2013). This is interesting because all imaging genetics studies in the anxiety disorders have used stimuli drawn from an emotional social sphere. In summary, even though there are additional intermediate phenotypes like personality and anxiety sensitivity that also may have translational power, fear conditioning is a putative mechanism translating the functional and structural genetically modulated variations observed in the imaging genetics studies to anxiety symptomatology. 229
Simply, increased fear conditioning in situationally elicited anxiety disorders results from molecular genetic effects on structure and function in the nodes of the fear network of the brain, biasing the brain toward fear.
R EF ERENCES Ågren T, Engman J, Frick A, Björkstrand J, Larsson E-M, Furmark T, Fredrikson M. (2012a). Disruption of reconsolidation erases a memory trace in the human amygdala. Science. 337: 1550–1552. Ågren T, Furmark T, Eriksson E, Fredrikson M. (2012b). Human fear reconsolidation and allelic differences in serotonergic and dopaminergic genes. Transl Psychiatry. 2: e76. doi: 10.1038/tp.2012.5. Ahs F, Pissiota A, MichelgArd A, Frans O, Furmark T, Appel L, Fredrikson M. (2009). Disentangling the web of fear: amygdala reactivity and functional connectivity in spider and snake phobia. Psychiatry Res. 172: 103–108. American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders (DSM IV), 4th ed. Washington, DC: American Psychiatric Press. Barbas H, Zikopoulos B. (2006). Sequential and parallel circuits for emotional processing in primate orbitofrontal cortex. In Zald DH and Rauch SL, eds., The orbitofrontal cortex (pp. 57–91). New York: Oxford University Press. Blokland GA, de Zubicaray GI, McMahon KL, Wright MJ. (2012). Genetic and environmental influences on neuroimaging phenotypes: a meta-analytical perspective on twin imaging studies. Twin Res Hum Genet. 15: 351–371. Britton JC, Gold AL, Deckersbach T, Rauch SL (2009). Functional MRI study of specific animal phobia using an event-related emotional counting stroop paradigm. Depress Anxiety. 26: 796–805. Brown SM, Peet E, Manuck SB, Williamson DE, Dahl RE, Ferrell RE, Hariri AR. (2005). A regulatory variant of the human tryptophan hydroxylase-2 gene biases amygdala reactivity. Mol Psychiatry. 9: 884–888. Bryant RA, Felmingham K, Kemp A, Das P, Hughes G, Peduto A, Williams L. (2008). Amygdala and ventral anterior cingulate activation predicts treatment response to cognitive behaviour therapy for post-traumatic stress disorder. Psychol Med. 38: 555–561. Caseras X, Giampietro V, Lamas A, Brammer M, Vilarroya O, Carmona S, Rovira M, Torrubia R, Mataix-Cols D. (2010). The functional neuroanatomy of blood-injection-injury phobia: a comparison with spider phobics and healthy controls. Psychol Med. 40: 125–134. Critchley HD, Elliott R, Mathias CJ, Dolan RJ. (2000). Neural activity relating to generation and representation of galvanic skin conductance responses: a functional magnetic resonance imaging study. J Neurosci. 20: 3033–3040. Davis M, Whalen PJ. (2001). The amygdala: vigilance and emotion. Mol Psychiatry. 1: 13–34. Domschke K, Braun M, Ohrmann P, Suslow T, Kugel H, Bauer J, Hohoff C, Kersting A, Engelien A, Arolt V, Heindel W, Deckert J. (2006). Association of the functional-1019C/G 5-HT1A polymorphism with prefrontal cortex and amygdala activation measured with 3 T fMRI in panic disorder. Int J Neuropsychopharmacol. 3: 349–355. Domschke K, Dannlowski U. (2010). Imaging genetics of anxiety disorders. Neuroimage 53:822–831. Domschke K, Freitag CM, Kuhlenbaumer G, Schirmacher A, Sand P, Nyhuis P, Jacob C, Fritze J, Franke P, Rietschel M, Garritsen HS, Fimmers R, Nöthen MM, Lesch KP, Stogbauer F, Deckert J. (2004). Association of the functional V158M catechol-O-methyl-transferase polymorphism with panic disorder in women. Int J Neuropsychopharmacol. 7: 183–188. Domschke K, Ohrmann P, Braun M, Suslow T, Bauer J, Hohoff C, Kersting A, Engelien A, Arolt V, Heindel W, Deckert J, Kugel H. (2008).
Influence of the catechol-O-methyltransferase val158met genotype on amygdala and prefrontal cortex emotional processing in panic disorder. Psychiatry Res. 163: 13–20. Domschke K, Reif A, Weber H, Richter J, Hohoff C, Ohrmann P, Pedersen A, Bauer J, Suslow T, Kugel H, Heindel W, Baumann C, Klauke B, Jacob C, Maier W, Fritze J, Bandelow B, Krakowitzky P, Rothermundt M, Erhardt A, Binder EB, Holsboer F, Gerlach AL, Kircher T, Lang T, Alpers GW, Ströhle A, Fehm L, Gloster AT, Wittchen HU, Arolt V, Pauli P, Hamm A, Deckert J. (2011). Neuropeptide S receptor gene: converging evidence for a role in panic disorder. Mol Psychiatry. 9: 938–948. Etkin, A. (2010). Functional neuroanatomy of anxiety: a neural circuit perspective. Curr Top Behav Neurosci. 2:251–277. Etkin A, Wager TD. (2007). Functional neuroimaging of anxiety: a meta-analysis of emotional processing in PTSD, social anxiety disorder, and specific phobia. Am J Psychiatry. 164:1476–1488. Fredrikson M, Faria V. (20130). Neuroimaging in anxiety disorders. In Baldwin DS, ed., Anxiety disorders (vol. 29, pp. 47–66). Mod Trends Pharmacopsychiatry. Basel: Karger. Fredrikson M, Faria V, Furmark T. (2013). Neuroimaging of neurotransmission in anxiety disorders. Chapter 14 in Dierckx RAJO, Otte A, Boer JA den, De Vries EFJ, Van Waarde A, eds., PET and SPECT in psychiatry. Berlin: Springer Verlag. Fredrikson M, Furmark T, Olsson MT, Fischer H, Andersson J, Långström B. (1998). Functional neuroanatomical correlates of electrodermal activity: a positron emission tomographic study. Psychophysiology. 35:179–185. Fredrikson M, Wik G, Annas P, Ericson K, Stone-Elander S. (1995). Functional neuroanatomy of visually elicited simple phobic fear: additional data and theoretical analysis. Psychophysiology. 32: 43–48. Fredrikson M, Wik G, Greitz T, Eriksson L, Stone-Elander S, Ericson K, SedVall G. (1993). Regional cerebral blood flow during experimental phobic fear. Psychophysiology. 30: 126–130. Furmark T, Appel L, Henningsson S, Ahs F, Faria V, Linnman C, Pissiota A, Frans O, Bani M, Bettica P, Pich EM, Jacobsson E, Wahlstedt K, Oreland L, Långström B, Eriksson E, Fredrikson M. (2008). A link between serotonin-related gene polymorphisms, amygdala activity, and placebo-induced relief from social anxiety. J Neurosci. 28: 13066–13074. Furmark T, Henningsson S, Appel L, Ahs F, Linnman C, Pissiota A, Faria V, Oreland L, Bani M, Pich EM, Eriksson E, Fredrikson M. (2009). Genotype over-diagnosis in amygdala responsiveness: affective processing in social anxiety disorder. J Psychiatry Neurosci. 1: 30–40. Furmark T, Tillfors M, Garpenstrand H, Marteinsdottir I, Långström B, Oreland L, Fredrikson M. (2004). Serotonin transporter polymorphism related to amygdala excitability and symptom severity in patients with social phobia. Neurosci Lett. 362: 189–192. Garpenstrand H, Annas P, Ekblom J, Oreland L, Fredrikson M. (2001). Human fear-conditioning is related to dopaminergic and serotonergic biological markers. Behav Neurosci. 115: 358–364. Goossens L, Sunaert S, Peeters R, Griez EJ, Schruers KR. (2007). Amygdala hyperfunction in phobic fear normalizes after exposure. Biol Psychiatry. 62: 1119–1125. Hamilton SP, Slager SL, Heiman GA, Deng Z, Haghighi F, Klein DF, Hodge SE, Weissman MM, Fyer AJ, Knowles JA. (2002). Evidence for a susceptibility locus for panic disorder near the catechol-O-methyltransferase gene on chromosome 22. Biol Psychiatry. 51: 591–601. Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D, Egan MF, Weinberger DR. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science. 297: 400–403. Hermann A, Küpper Y, Schmitz A, Walter B, Vaitl D, Hennig J, Stark R, Tabbert K. (2012). Functional gene polymorphisms in the serotonin system and traumatic life events modulate the neural basis of fear acquisition and extinction. PLoS One. 7(9): e44352. Hermann A, Schäfer A, Walter B, Stark R, Vaitl D, Schienle A. (2009). Emotion regulation in spider phobia: role of the medial prefrontal cortex. Soc Cogn Affect Neurosci. 4: 257–267.
2 3 0 part iv : I maging G enetics and D isease
Hermann C, Ziegler S, Birbaumer N, Flor H (2002). Psychophysiological and subjective indicators of aversive pavlovian conditioning in generalized social phobia. Biol Psychiatry. 52: 328–337. Hettema JM, Annas P, Neale MC, Kendler KS, Fredrikson M. (2003). A twin study of the genetics of fear-conditioning. Arch Gen Psychiatry. 60: 702–708. Hettema JM, Neale MC, Kendler KS. (2001) A review and meta-analysis of the genetic epidemiology of anxiety disorders. Am J Psychiatry. 158: 156–164. Hettema JM, Prescott CA, Myers JM, Neale MC, Kendler KS. (2005). The structure of genetic and environmental risk factors for anxiety disorders in men and women. Arch Gen Psychiatry. 62: 182–189. Hopper JW, Frewen PA, van der Kolk BA, Lanius RA. (2007). Neural correlates of reexperiencing, avoidance, and dissociation in PTSD: symptom dimensions and emotion dysregulation in responses to script-driven trauma imagery. J Trauma Stress. 20:713–725. Kasai K, Yamasue H, Gilbertson MW, Shenton ME, Rauch SL, Pitman RK. (2008). Evidence for acquired pregenual anterior cingulate gray matter loss from a twin study of combat-related posttraumatic stress disorder. Biol Psychiatry. 63: 550–556. Kendler KS, Karkowski LM, Prescott CA. (1999). Fears and phobias: reliability and heritability. Psychol Med. 29: 539–553. Kessler RC, Berglund P, Demler O, Jin R, Walters EE. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry. 62: 593–602. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, Wittchen HU, Kendler KS. (1994). Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Survey. Arch Gen Psychiatry. 51: 8–19. Kim J, Dager S, Lyoo K (2012). The role of the amygdala in the pathophysiology of panic disorder: evidence from neuroimaging studies. Biol Mood Anxiety Disord. 2: 20. Kim B, Yoo E, Lee JY, Lee KS, Choe AY, Lee JE, Kwack K, Yook KH, Choi TK, Lee SH. (2013). The effects of the catechol-O-methyltransferase val158met polymorphism on white matter connectivity in patients with panic disorder. J Affect Disord. 147: 64–71. Klengel T, Mehta D, Anacker C, Rex-Haffner M, Pruessner JC, Pariante CM, Pace TW, Mercer KB, Mayberg HS, Bradley B, Nemeroff CB, Holsboer F, Heim CM, Ressler KJ, Rein T, Binder EB. (2013). Allele-specific FKBP5 DNA demethylation mediates gene-childhood trauma interactions. Nat Neurosci. 16: 33–41. Klucken T, Alexander N, Schweckendiek J, Merz CJ, Kagerer S, Osinsky R, Walter B, Vaitl D, Hennig J, Stark R. (2013). Individual differences in neural correlates of fear conditioning as a function of 5-HTTLPR and stressful life events. Soc Cogn Affect Neurosci. 3: 318–325. Lau JY, Goldman D, Buzas B, Fromm SJ, Guyer AE, Hodgkinson C, Monk CS, Nelson EE, Shen PH, Pine DS, Ernst M. (2009). Amygdala function and 5-HTT gene variants in adolescent anxiety and major depressive disorder. Biol Psychiatry. 65: 349–355. Lau JY, Goldman D, Buzas B, Hodgkinson C, Leibenluft E, Nelson E, Sankin L, Pine DS, Ernst M. (2010). BDNF gene polymorphism (Val66Met) predicts amygdala and anterior hippocampus responses to emotional faces in anxious and depressed adolescents. Neuroimage. 53: 952–961. Lissek S, Levenson J, Biggs AL, Johnson LL, Ameli R, Pine DS, Grillon C. (2008). Elevated fear conditioning to socially relevant unconditioned stimuli in social anxiety disorder. Am J Psychiatry. 165: 124–132. Lissek S, Rabin SJ, McDowell DJ, Dvir S, Bradford DE, Geraci M, Pine DS, Grillon C. (2009). Impaired discriminative fear-conditioning resulting from elevated fear responding to learned safety cues among individuals with panic disorder. Behav Res Ther. 47: 111–118. Lonsdorf TB, Weike AI, Nikamo P, Schalling M, Hamm AO, Ohman A (2009). Genetic gating of human fear learning and extinction: possible implications for gene-environment interaction in anxiety disorder. Psychol Sci. 2: 198–206. chapter 1 4 : I maging G enetics of A nxiety D isorders
Maddox SA, Schafe GE, Ressler KJ. (2013). Exploring epigenetic regulation of fear memory and biomarkers associated with post-traumatic stress disorder. Front Psychiatry. 4: 62–68. Maier S, Szalkowski A, Kamphausen S, Perlov E, Feige B, Blechert J, Philipsen A, van Elst LT, Kalisch R, Tüscher O. (2012). Clarifying the role of the rostral dmPFC/dACC in fear/anxiety: learning, appraisal or expression? PLoS One. 7(11): e50120. doi: 10.1371/ Milad MR, Pitman RK, Ellis CB, Gold AL, Shin LM, Lasko NB, Zeidan MA, Handwerger K, Orr SP, Rauch SL. (2009). Neurobiological basis of failure to recall extinction memory in posttraumatic stress disorder. Biol Psychiatry. 66: 1075–1082. Milad MR, Quinn BT, Pitman RK, Orr SP, Fischl B, Rauch SL. (2005). Thickness of ventromedial prefrontal cortex in humans is correlated with extinction memory. Proc Natl Acad Sci U S A. 102: 10706–10711. Morey RA, Hariri AR, Gold AL, Hauser MA, Munger HJ, Dolcos F, McCarthy G. (2011). Serotonin transporter gene polymorphisms and brain function during emotional distraction from cognitive processing in posttraumatic stress disorder. BMC Psychiatry. 11: 76. Mueller SC, Aouidad A, Gorodetsky E, Goldman D, Pine DS, Ernst M. (2013). Gray matter volume in adolescent anxiety: an impact of the brain-derived neurotrophic factor Val(66)Met polymorphism? J Am Acad Child Adolesc Psychiatry. 52(2): 184–195. Erratum in: J Am Acad Child Adolesc Psychiatry. 52: 329. Munafò MR, Brown SM, Hariri AR. (2008). Serotonin transporter (5-HTTLPR) genotype and amygdala activation: a meta-analysis. Biol Psychiatry. 63: 852–857. Orr SP, Metzger LJ, Lasko NB, Macklin ML, Peri T, Pitman RK. (2000). De novo conditioning in trauma-exposed individuals with and without posttraumatic stress disorder. J Abnorm Psychol. 109: 290–298. Paulus MP, Stein MB. (2006). An insular view of anxiety. Biol Psychiatry. 60: 383–387. Pejic T, Hermann A, Vaitl D, Stark R. (2013). Social anxiety modulates amygdala activation during social conditioning. Soc Cogn Affect Neurosci. 3: 267–276. Phelps EA, Delgado MR, Nearing KI, LeDoux JE. (2004). Extinction learning in humans: role of the amygdala and vmPFC. Neuron. 43: 897–905. Pissiota A, Frans O, Fernandez M, von Knorring L, Fischer H, Fredrikson M. (2002) Neurofunctional correlates of posttraumatic stress disorder: a PET symptom provocation study. Eur Arch Psychiatry Clin Neurosci. 252: 68–75. Protopopescu X, Pan H, Tuescher O, Cloitre M, Goldstein M, Engelien W, Epstein J, Yang Y, Gorman J, LeDoux J, Silbersweig D, Stern E. (2005). Differential time courses and specificity of amygdala activity in posttraumatic stress disorder subjects and normal control subjects. Biol Psychiatry. 57: 464–473. Reetz K, Lencer R, Steinlechner S, Gaser C, Hagenah J, Büchel C, Petersen D, Kock N, Djarmati A, Siebner HR, Klein C, Binkofski F. (2008). Limbic and frontal cortical degeneration is associated with psychiatric symptoms in PINK1 mutation carriers. Biol Psychiatry. 64: 241–247. Ridder S, Treutlein J, Nees F, Lang S, Diener S, Wessa M, Kroll A, Pohlack S, Cacciaglia R, Gass P, Schütz G, Schumann G, Flor H. (2012). Brain activation during fear conditioning in humans depends on genetic variations related to functioning of the hypothalamic-pituitary-adrenal axis: first evidence from two independent subsamples. Psychol Med. 1–11. Rotberg B, Kronenberg S, Carmel M, Frisch A, Brent D, Zalsman G, Apter A, Weizman A. (2013). Additive effects of 5-HTTLPR (serotonin transporter) and tryptophan hydroxylase 2 G-703T gene polymorphisms on the clinical response to citalopram among children and adolescents with depression and anxiety disorders. J Child Adolesc Psychopharmacol. 2: 117–122. Rothe C, Koszycki D, Bradweijn J, King N, De Luca V, Tharmalingam V, Macciardi F, Deckert J, Kennedy J. (2006). Association of the Val158Met catechol-O-methyltransferase genetic polymorphism with panic disorder. Neuropsychopharmacology. 31, 2237–2242. 231
Schiller D, Monfils MH, Raio CM, Johnson DC, Ledoux JE, Phelps EA. (2010). Preventing the return of fear in humans using reconsolidation update mechanisms. Nature. 463: 49–53. Schulz-Heik RJ, Schaer M, Eliez S, Hallmayer JF, Lin X, Kaloupek DG, Woodward SH. (2011). Catechol-O-methyltransferase Val158Met polymorphism moderates anterior cingulate volume in posttraumatic stress disorder. Biol Psychiatry. 70: 1091–1096. Erratum in: Biol Psychiatry. 71(4): 388. Sehlmeyer C, Schöning S, Zwitserlood P, Pfleiderer B, Kircher T, Arolt V, Konrad C. (2009). Human fear conditioning and extinction in neuroimaging: a systematic review. PLoS One. 4(6): e5865. doi: 10.1371/journal.pone.0005865. Review. Shimony JS, McKinstry RC, Akbudak E, Aronovitz JA, Snyder AZ, Lori NF, Cull TS, Conturo TE. (1999). Quantitative diffusion-tensor anisotropy brain MR imaging: normative human data and anatomic analysis. Radiology. 212: 770–784. Shin LM, Orr SP, Carson MA, Rauch SL, Macklin ML, Lasko NB, Peters PM, Metzger LJ, Dougherty DD, Cannistraro PA, Alpert NM, Fischman AJ, Pitman RK. (2004). Regional cerebral blood flow in the amygdala and medial prefrontal cortex during traumatic imagery in male and female Vietnam veterans with PTSD. Arch Gen Psychiatry. 61: 168–176. Shin LM, Wright CI, Cannistraro PA, Wedig MM, McMullin K, Martis B, Macklin ML, Lasko NB, Cavanagh SR, Krangel TS, Orr SP, Pitman RK, Whalen PJ, Rauch SL. (2005). A functional magnetic resonance imaging study of amygdala and medial prefrontal cortex responses to overtly presented fearful faces in posttraumatic stress disorder. Arch Gen Psychiatry. 62: 273–281. Shin LM, Liberzon I. (2010). The neurocircuitry of fear, stress, and anxiety disorders. Neuropsychopharmacology. 35: 169–191. Smolka MN, Schumann G, Wrase J, Grüsser SM, Flor H, Mann K, Braus DF, Goldman D, Büchel C, Heinz A. (2005). Catechol-Omethyltransferase val158met genotype affects processing of emotional stimuli in the amygdala and prefrontal cortex. J Neurosci. 25: 836–842.
Stein MB, Goldin PR, Sareen J, Zorrilla LT, Brown GG. (2000). Increased amygdala activation to angry and contemptuous faces in generalized social phobia. Biol Psychiatry. 47: 769–776. Straube T, Mentzel HJ, Miltner WH. (2007). Waiting for spiders: brain activation during anticipatory anxiety in spider phobics. Neuroimage. 37: 1427–1436. Straube B, Reif A, Richter J, Lueken U, Weber H, Arolt V, Jansen A, Zwanzger P, Domschke K, Pauli P, Konrad C, Gerlach AL, Lang T, Fydrich T, Alpers GW, Ströhle A, Wittmann A, Pfleiderer B, Wittchen HU, Hamm A, Deckert J, Kircher T. (2014). The functional-1019C/G HTR1A polymorphism and mechanisms of fear. Transl Psychiatry. 4: e490. doi: 10.1038/tp.2014.13 Tillfors M, Furmark T, Ekselius L, Fredrikson M. (2001). Social phobia and avoidant personality disorder as related to parental history of social anxiety: a general population study. Behav Res Ther. 39: 289–298. Whiteford HA, Degenhardt L, Rehm J, Baxter AJ, Ferrari AJ, Erskine HE, Charlson FJ, Norman RE, Flaxman AD, Johns N, Burstein R, Murray CJ, Vos T. (2013). Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010. Lancet. 382(9904): 1575–1586. doi: 10.1016/S0140-6736(13)61611-6. Epub 2013 Aug 29. Willeit M, Praschak-Rieder N. (2010). Imaging the effects of genetic polymorphisms on radioligand binding in the living human brain: a review on genetic neuroreceptor imaging of monoaminergic systems in psychiatry. Neuroimage. 3: 878–892. Williams LM, Kemp AH, Felmingham K, Barton M, Olivieri G, Peduto A, Gordon E, Bryant RA. (2006). Trauma modulates amygdala and medial prefrontal responses to consciously attended fear. Neuroimage. 29: 347–357. Woodward SH, Kaloupek DG, Streeter CC, Martinez C, Schaer M, Eliez S. (2006). Decreased anterior cingulate volume in combat-related PTSD. Biol Psychiatry. 59: 582–587.
2 3 2 part iv : I maging G enetics and D isease
15. IMAGING GENETICS OF BIPOLAR DISORDER Ole A. Andreassen and Martin Tesli
INTRODUCTI O N Bipolar disorder (BD) is a severe psychiatric disorder with a lifetime prevalence of approximately 1% worldwide (Merikangas et al. 2011). The core symptoms are mood swings, which manifest as outbursts of mania and depression, with psychotic symptoms such as hallucinations and delusions occasionally co-occurring with the mood episodes. Epidemiological studies have demonstrated that BD runs in families, and based on twin studies, heritability has been estimated to be 60%–85% (Smoller and Finn 2003). Additionally, cross-heritability with schizophrenia (SZ) has been reported in epidemiological studies (Lichtenstein et al. 2009). Despite these high heritability estimates, few risk genetic variants have been unambiguously identified, and the pathophysiological mechanisms remain elusive. The diagnosis is based solely on clinical assessment, with no guiding laboratory tests to corroborate the diagnostic evaluation. However, during the last couple of years, major advances in the field of psychiatric genetics as well as neuroimaging have identified new biological markers related to BD. Based on the results from these two disciplines, imaging genetics has provided us with new insights in the neurobiological correlates of the disorder. Imaging genetics studies seem particularly meaningful in BD, considering its high heritability, and this approach might add important information to our knowledge of disease mechanisms. While genetics gives researchers in the imaging field a tool for differentiating between genetically and environmentally conditioned brain anomalies, imaging enables researchers in the genetics field to gain insight into intermediate phenotypes between genotype and broad diagnostic category. This has been the traditional outcome measure in case-control studies. Imaging genetics studies also have the potential for developing a more holistic understanding of causal mechanisms in this common and
severe disorder. However, as BD has a paroxystic course, with state rather than trait characteristics as defining diagnostic markers, imaging genetics might impose some particular challenges, in addition to opportunities, for BD. Hence, the application and interpretation of imaging genetics studies in BD might differ slightly from other, more trait-defined psychiatric disorders, like SZ. In this chapter, we will give a short introduction to the latest findings from molecular genetic and neuroimaging studies, before presenting an overview of imaging genetics studies in BD, including different approaches, major results, and potential implications for future research and clinical practice. B IPOLAR DIS OR DE R AND GE NE T IC S In the past decade, three major approaches have been applied to identify genetic risk variants for BD: the functional candidate approach, the positional candidate approach, and genome-wide association (GWA) studies (Serretti and Mandelli 2008). The functional candidate approach builds on prior knowledge of physiological alterations in the disease, pharmacological mechanisms, and animal models. These studies have traditionally investigated genes related to the serotonin, dopamine, and noradrenaline neurotransmitter systems: catechol-O-methyltransferase (COMT), monoamine oxidase A (MAOA), and the serotonin transporter (5HTT). Due to evidence of overlapping phenotypes, there has also been a particular interest in genes associated with SZ, including disrupted in schizophrenia (DISC1), brain-derived neurotrophic factor (BDNF), the G72/G30 locus (D-amino acid oxidase activator [DAOA]), and Neuregulin 1 (NRG1). However, both positive and negative findings have been reported for these genetic variants. The same holds for the positional candidate approach, 233
in which risk variants are sought by investigating genomic regions associated with BD through linkage studies of affected and unaffected family members (Craddock and Sklar 2009). This lack of consistency from both candidate gene studies and linkage studies indicates that no single genetic marker has a major effect on increasing the risk for BD. Thus, taking into consideration the high heritability estimates, pre-GWA molecular genetic studies have provided evidence that BD is a polygenic rather than a monogenic disorder, in which many variants must interact with each other as well as with environmental factors to give rise to the disorder. Since the first large GWA study of BD in 2007 (WTCCC 2007), several genome-wide scans have been performed in case-control samples. In contrast to candidate gene studies, which are hypothesis-driven, GWA studies investigate millions of markers across the genome, without a prior hypothesis. An advantage of this “agnostic” approach is the possibility of identifying novel variants and pathways. However, with such a high number of genetic variants, GWA studies impose statistical challenges related to multiple testing correction. The major published BD GWA studies include the Wellcome Trust Case-Control Consortium (WTCCC 2007) study (n = 1868 cases and 2938 controls of British descent), in which a strong association was found for a single nucleotide polymorphism (SNP) in PALB2 (partner and localizer of BRCA2) (WTCCC 2007); a German/ American study (n = 1233 cases and 1439 controls) reporting a signal in DGKH (diacylglycerol kinase eta) (Baum et al. 2008); an American/British study (n = 1461 BDI cases and 2008 controls in discovery sample) with MYO5B (myosin5B) and TSPAN8 (tetraspanin-8) as the top hits (Sklar et al. 2008); and a collaborative study (n = 4387 cases and 6209 controls of European descent) identifying ANK3 (ankyrin3) and CACNA1C (alpha 1C subunit of the Ltype voltage-gated calcium channel) (Ferreira et al. 2008). The associations between ANK3 and CACNA1C and BD were later replicated by several research groups (Scott et al. 2009; Smith et al. 2009; Lee et al. 2011; PGC 2011; Takata et al. 2011; Tesli et al. 2011). The largest GWA study of BD thus far is a multi-center international mega-analysis (n = 11974 BD cases and 51,792 controls) undertaken by the Psychiatric GWAS Consortium (PGC) (Sullivan 2010). Findings from this study were published in the autumn of 2011 (PGC 2011), and confirmed the association between CACNA1C and BD, and gave further support to ANK3, in addition to identifying new genetic risk variants in the genes ODZ4 and SYNE1. Furthermore, the authors reported
enrichment in several genes encoding calcium channel subunits, suggesting a common pathway. Interestingly, when combining and comparing the BD results with the results from the SZ sample (n = 17839 SZ cases and 33,859 controls), three of these genes/regions were common for both disorders. Moreover, the PGC cross-disorder study, including BD, SZ, major depressive disorder (MDD), autism spectrum disorder (ASD), and attention-deficit hyperactivity disorder (ADHD) (n = 33,332 cases and 27,888 controls), reported enrichment of calcium channel genes in all five disorders, and a polygene risk score showed cross-disorder associations, most pronounced between BD and SZ (Smoller et al. 2013). These findings are in line with epidemiological data and support genetic overlap between BD and SZ, which seems to involve a large proportion of the polygenic architecture (Andreassen et al. 2013). But there is also evidence of molecular genetic differences, as immune-related genes and structural variants (CNVs) seem to be more involved in SZ than BD pathology (Bergen et al. 2012). Thus, the hypothesis of a partial overlap and a spectrum model of psychotic disorders (Craddock and Owen 2010) is strengthened. However, each of these genetic risk variants has a minor effect size, typically an odds ratio between1.1 and 1.2, and explains a tiny fraction of the clinical phenotype. Moreover, when added together, the most significant SNPs from the major GWA studies still do not explain more than approximately 3% of the variance (PGC 2011). With heritability estimates of 60%–85%, this leaves a gap of unexplained variance, often coined “missing heritability.” In order to explain more of the molecular genetic and neurobiological underpinnings of BD, it seems meaningful to assess the influence of genetic variants on brain structures and functions related to the clinical phenotype. B IPOLAR DIS OR DE R AND NE U ROIMAG I N G In this section we will focus on neuroimaging studies investigating structural biomarkers, functional biomarkers, and connectivity in BD. The first computed tomography (CT) and structural magnetic resonance imaging (sMRI) studies reported no or few changes in brain volume in BD subjects compared to healthy controls (Houenou et al. 2012). Subsequent studies have yielded inconsistent results, with enlarged ventricles and white matter hyperintensities as some of the most robust findings (Houenou et al. 2012; Kempton et al. 2008; McDonald et al. 2004). A recent study of a large sample (n = 173 patients with SZ spectrum disorder,
2 3 4 part iv : I maging G enetics and D isease
139 patients with BD, and 207 healthy control subjects) found cortical thinning, subcortical volume reductions, and ventricular enlargements in patients with BD type 1 compared with healthy controls, although not as pronounced as in SZ patients (Rimol et al. 2010). As for the functional biomarkers in BD, the most commonly used method has been functional magnetic resonance imaging (fMRI). Emotional paradigms have been widely applied, as neural regions and networks involved in emotional processing have been hypothesized to play an important role in the underlying mechanisms in BD. Although diverging results have been reported, meta-analyses have identified some consistent patterns of abnormal activation. There are several lines of evidence for hyperactivity of the ventral-limbic brain network and hypoactivity of dorsal brain structures (Houenou et al. 2012). One recent meta-analysis (n = 1040 BD cases and 1074 healthy controls) found underactivation in the inferior frontal gyrus (IFG) and putamen and overactivation in limbic areas in BD patients (Chen et al. 2011). Whereas the IFG dysregulations were found both in cognitive and emotional paradigms, the enhanced limbic activation was reported in emotional paradigms. As for the clinical states, this metaanalysis reported IFG hypoactivation during manic episodes but not in euthymic and depressed episodes (Chen et al. 2011). Another single study (n = 9 BD cases and 9 healthy controls) found manic patients to have hyperactivation in the left amygdala and bilateral hypoactivation in the lateral orbitofrontal cortex (Altshuler et al. 2005). A more recent study compared activation between BD cases in euthymic state (n = 15) with BD cases in depressed state (n = 15) and manic state (n = 15), as well as with healthy controls (n = 30) during a negative facial emotion matching task. The investigators found that euthymic and depressed patients have increased amygdala activation compared to healthy controls, but this increase was not observed in the manic BD cases (Hulvershorn et al. 2012). But these early interesting findings need to be replicated, in order to establish a model of state- and trait-specific markers in BD. On the basis of reports of white matter hyperintensities, abnormal anatomical connectivity has been proposed to play a role in BD. Diffusion tensor imaging (DTI) is a recently developed method to explore white matter and brain connectivity. In a meta-analysis on DTI findings in BD, two clusters of decreased fractional anisotropy (FA) were identified. The first in the right white matter close to the parahippocampal gyrus and the other close to the right anterior and subgenual cingulate cortex. The authors concluded that changes in FA might represent an underlying mechanism for emotional dysreguation and chapter 1 5 : I maging G enetics of B ipolar D isorder
changes in functional limbic connectivity in BD (Vederine et al. 2011). Functional connectivity in BD has been investigated with newly developed fMRI methods, enabling researchers to identify brain networks coupled together during a paradigm or at rest (Rogers et al. 2007). From the few studies of functional connectivity in BD, there is some evidence that the connection between amygdala/hippocampus and the prefrontal/perigenual cortex is altered (Womer et al. 2009). These findings are in line with other fMRI studies reporting ventral-limbic hyperactivation and hypoactivation in dorsal brain structures (Houenou et al. 2012), however, there are not any robust resting state fMRI patterns in BD yet. Despite some converging reports of structural and functional alterations in BD, it remains uncertain whether these abnormalities are genetically conditioned, or if they result from environmental factors, the course of the illness, or psychopharmacologic treatment. Imaging genetics studies might have the potential to address some of these questions. B IPOLAR DIS OR DE R AND IMAGING GE NE T IC S Imaging genetics is a newly developed discipline in the BD research field with a sharp increase in publications during the last couple of years. We will go through the imaging genetics studies in the same order as for the genetics and imaging studies; from genes identified in pre-GWA studies to those identified in GWA studies, and from structural studies through functional studies to those measuring anatomical and functional connectivity. Imaging genetics studies of structural phenotypes have focused mainly on single variants identified through large GWA studies. In particular, the BD risk gene CACNA1C has been extensively studied; sMRI studies of healthy controls have found the risk SNP rs1006737 to be associated with increased gray matter volume (n = 77 healthy adults) (Kempton et al. 2009), as well as a total cortical volume increase (n = 55 healthy subjects) (Wang et al. 2011). However, an effect of this SNP on gray matter volume could not be replicated in a study of 585 healthy individuals, but an association between SNPs in intron 3 of this gene and brainstem volume alterations were reported from the same study (Franke et al. 2010). In a study of 41 euthymic BD subjects and 40 controls, CACNA1C SNP rs1006737 risk allele carriers were shown to have increased gray matter density in right amygdala and right hypothalamus irrespective of diagnosis. Additionally, the authors reported an 235
interaction effect between genotype and diagnosis in the left putamen, which was smaller in BD cases with the risk allele than in the healthy controls (Perrier et al. 2011). A recent report of 517 individuals (121 BD cases, 116 SZ cases, 61 other psychosis cases, and 219 healthy controls) found no significant associations between nine BD risk SNPs in the genes CACNA1C, ANK3, ODZ4, and SYNE1 and eight brain structural measures found to be altered in BD, neither for single variants nor for the polygene risk score across all SNPs (Tesli et al. 2013). Taken together, there are few studies and little consistency for sMRI studies and BD risk variants. As for the imaging genetics functional measures in BD, more studies have been published, with some recent consistent results. Early reports investigated candidate genes involved in neurotransmission and BD risk. In 2002, Hariri and coworkers showed that individuals with the short allele of the serotonin transporter (5-HTT) promoter polymorphism exhibited enhanced amygdala activity in an fMRI emotional paradigm (Hariri et al. 2002). This finding might have relevance for BD, as the short allele has been related to reduced 5-HTT expression and function as well as increased risk of depression (Caspi et al. 2003). As dopamine dysregulation has been proposed to play a role in psychotic disorders, functional variants of the COMT gene have also been investigated in fMRI studies. The most robust findings include increased activity in carriers of the Val158 allele in the dorsal prefrontal cortex (PFC) during executive function (Blasi et al. 2005; Bertolino et al. 2006; Schott et al. 2006; Winterer et al. 2006; Mier et al. 2010), and increased activity in the ventral PFC and limbic regions in Met158 allele carriers during emotionally negative stimuli (Smolka et al. 2005; Drabant et al. 2006; Mier et al. 2010). However, variants in the COMT gene and the serotonin transporter gene have not been specifically related to BD in large molecular genetic studies (PGC 2011). In contrast, SNPs in CACNA1C and ANK3 have been consistently related to BD in recent major studies (Ferreira et al. 2008; Sklar et al. 2008; Schulze et al. 2009; PGC 2011). As for the sMRI genetics studies, the CACNA1C SNP rs1006737 has been a focus of attention in the fMRI field. In a study from 2010, risk allele carriers were found to exhibit increased hippocampus activity during emotional processing and increased prefrontal activity during executive cognition (Bigos et al. 2010). There were also trend-significant associations for increased amygdala activity. In accordance with this latter finding, three recent reports have found evidence for increased amygdala activity in CACNA1C SNP rs1006737 risk allele carriers during emotional paradigms. One of the studies consisted of
64 healthy volunteers (Wessa et al. 2010); the second study included 41 euthymic BD cases, 25 first-degree relatives without any Axis I disorder, and 50 healthy controls ( Jogia et al. 2011); whereas the third report comprised 250 individuals (n = 66 BD, 61 SZ, and 123 healthy controls) (Tesli et al. 2013). The second report also demonstrated an interaction between genotype and diagnosis in the right ventrolateral PFC, with decreased activity in BD patients with the risk allele compared to the relatives and healthy controls ( Jogia et al. 2011). CACNA1C risk allele carriers have also been shown to exhibit increased activation in the left inferior frontal gyrus and the left precuneus during a semantic verbal fluency task (Krug et al. 2010). Further, there is evidence of reduced bilateral hippocampal activity during episodic memory recall, diminished functional coupling between left and right hippocampal regions, as well as activation deficits of the subgenual anterior cingulate cortex in healthy risk allele carriers (Erk et al. 2010). Finally, healthy risk allele carriers have been associated with reduced activity in the right inferior parietal lobule during orienting and reduced activity in the medial frontal gyrus (MFG) during executive control of attention (Thimm et al. 2011). Other BD risk genetic variants investigated with fMRI include SNPs in the gene DGKH, implied in BD pathology in an early GWA study (Baum et al. 2008), although this finding was not confirmed in later studies (PGC 2011). One study (n = 81 individuals at high familial risk of BD and 75 healthy) reported an interaction effect between genotype and group in the left medial frontal gyrus, left precuneus, and right parahippocampal gyrus during a verbal fluency task (Whalley et al. 2012). Another study found carriers of the risk variant in the gene ODZ4 (rs12576775) to be associated with increased amygdala activity during reward sensitivity and reward expectation (n = 485 healthy individuals) (Heinrich et al. 2013). The GWA approach has also been applied to fMRI studies of BD patients. One study (n = 39 BD cases and 29 healthy controls) found right amygdala activation to be significantly associated with an SNP in the gene DOK5, a gene that is involved in neurotrophin signaling (Liu et al. 2010). A more recent study reported implied variants near a monoaminergic pathway gene (PHOX2B) in amygdala activity during a negative faces matching paradigm (n = 221) (Ousdal et al. 2012). As each susceptibility variant has been shown to confer a small increase in the risk of BD, using cumulative risk load instead of single variants might increase the statistical power of imaging genetics studies. This approach comprises polygenic risk score analyses, in which the total risk load of the genome is assessed, as well as pathway analyses, where
2 3 6 part iv : I maging G enetics and D isease
genes are clustered into functionally related categories. In one recent study, the polygenic risk score for BD is used as an independent variable, and fMRI activity during an executive processing/language task is used as outcome variable (n = 87 BD cases and 71 healthy controls). The authors found that an increasing polygenic risk was associated with enhanced activity in the anterior cingulate cortex and amygdala across both groups. However, there was no evidence of interaction effect between diagnostic group and polygenic risk score (Whalley et al. 2012). With regard to structural connectivity imaging genetics studies, one group reported decreased white matter integrity in healthy carriers of the ANK3 SNP rs10994336 risk allele in the anterior limb of the internal capsule, but no white matter alterations in risk allele carriers of ANK3 SNP rs9804190 (Linke et al. 2012). Another group found no differences in fractional anisotropy (FA) between risk allele carriers and carriers of the protective allele in ZNF804A SNP rs1344706 (Sprooten et al. 2012). There are more published studies investigating functional connectivity than structural connectivity with an imaging genetics approach in BD. Again, CACNA1C SNP rs1006737 has been a major focus of interest. One study found that risk allele carriers of this SNP had reduced functional connectivity within a corticolimbic frontotemporal neural system (Wang et al. 2011), whereas another study reported risk allele carriers to have reduced effective connectivity from MFG, an effect that was more pronounced in BD patients than in their unaffected relatives and healthy control subjects (Radua et al. 2013). Additionally, the risk allele in ZNF804A SNP rs1344706 has been related to altered connectivity for the dorsolateral PFC, hippocampus, and amygdala (Esslinger et al. 2009). The altered connectivity between the dorsolateral PFC and hippocampus was replicated in a later study, although with weaker effect (Paulus et al. 2013). Taken together, there is a growing number of imaging genetics studies in BD, investigating the effect of risk genetic variants on structural, functional, and connectivity phenotypes. The most consistent finding, and also the most studied relation, seems to be amygdala over-activation in risk allele carriers of a CACNA1C SNP. DIS C US S IO N On the basis of large multicenter GWA studies and progress in neuroimaging, there has been a steep increase in imaging genetic studies in BD in recent years. Except from some promising replications in the fMRI field, there seems to be chapter 1 5 : I maging G enetics of B ipolar D isorder
little consistency between studies and imaging modalities. One explanation for this inconsistency might be the little interest in replicating previous results in this new research discipline. Another possibility is low statistical power, which has been shown to be the case for BD case-control genetic studies, where even the largest current analyses have been proposed to be underpowered (PGC 2011). Single genetic variants have been found to have minor effect sizes at the diagnosis level (odds ratio of 1.1–1.2). This could also hold true at brain phenotype level, despite a possible increase in effect size due to closeness in the translational chain of causality in psychiatric disorders (Hariri and Weinberger 2003). BD has been demonstrated to be highly polygenic (PGC 2011; Andreassen et al. 2013), and brain structural and functional variation is probably polygenic as well (Stein et al. 2012). In this case, the single variant strategy in imaging genetics might inhibit researchers from releasing the underlying potential of this field, which could be evoked by using polygenic approaches (Andreassen et al. 2013.) and pathway analyses (Mattingsdal et al. 2013). These would increase the amount of genomic information and collapse it into fewer and more robust variables than in the GWA study or single variant approach. Also, some brain phenotype differences between BD cases and controls might result from environmental factors, course of the disease, or psychopharmacological treatment. Although imaging genetics studies could address some of these questions, it might be difficult to discern whether brain alterations are the result of genetic risk, disease course, or psychopharmacological treatment, as the patients with highest genetic risk might also have the most severe course and highest dose of medication. However, longitudinal imaging studies could address some of these challenges. Moreover, gene-environment interaction studies might add valuable information on the interplay between environmental risk factors in BD, like childhood trauma (Etain et al. 2008; Aas et al. 2012), and risk genetic variation. As for the different imaging genetics methods, there seems to be an overweight of fMRI studies compared to sMRI studies, in addition to more converging findings from the former method than from the latter. One could speculate that this phenomenon results from BD being a disorder caused by dysfunctional neuronal circuitry rather than impaired neurodevelopment. Although BD has been shown to share genetic risk with SZ, these common risk genes might be related to aberrant affective neuronal networks, whereas the MHC genes and CNVs implied in SZ but not in BD could be related to impaired neurodevelopment (Demjaha et al. 2012). In line with this argument, the most significant findings from BD GWA studies are 237
CACNA1C and ANK3, both genes involved in ion channel functioning (Ferreira et al. 2008), with little evidence of playing a substantial part in neurodevelopment. Future imaging genetic studies could answer these questions by applying several new approaches. First, replication studies should be undertaken, in order to confirm previous findings. Multicenter mega-analyses, like we have seen in the molecular genetics field, could be performed to increase the statistical power and in turn decrease the rate of type 2 errors as well as spurious findings. Second, polygenic approaches and pathway analyses should be conducted to explain larger proportions of the brain phenotypic variation, and to reveal underlying mechanisms. Third, functional imaging studies could further explore the relations between state and trait phenotypes in BD, with traditional methods in addition to novel approaches, like resting state activity/default mode networks. Finally, gene-gene, geneenvironment, and functional connectivity studies could help explain the different levels of interaction in this highly heritable and neurobiologically complex disorder. With regard to clinical implications, there is currently too much inconsistency and too little specificity to implement findings from imaging genetics studies as diagnostic biomarkers in BD. However, with continued cross-disciplinary effort the field of imaging genetics has the potential to identify relations between gene pathways and brain networks. Thus increasing the knowledge on disease mechanisms, and, hopefully, providing the clinician with a supplementary tool in diagnostic evaluation and treatment stratification. Further, enhanced comprehension of pathological processes might open new possibilities for the development of pharmacological treatment, or, at least, may increase our understanding of the neurobiological mechanisms of currently available medication. R EF ERENCES Aas M, Djurovic S, Athanasiu L, Steen NE, Agartz I, Lorentzen S, Sundet K, Andreassen OA, Melle I. (2012). Serotonin transporter gene polymorphism, childhood trauma, and cognition in patients with psychotic disorders. Schizophr Bull. 38(1): 15–22. Altshuler L, Bookheimer S, Proenza MA, Townsend J, Sabb F, Firestine A, Bartzokis G, Mintz J, Mazziotta J, Cohen MS. (2005). Increased amygdala activation during mania: a functional magnetic resonance imaging study. Am J Psychiatry. 162(6): 1211–1213. Andreassen OA, Thompson WK, Schork AJ, Ripke S, Mattingsdal M, Kelsoe JR, Kendler KS, O’Donovan MC, Rujescu D, Werge T, Sklar P; Psychiatric Genomics Consortium (PGC); Bipolar Disorder and Schizophrenia Working Groups, Roddey JC, Chen CH, McEvoy L, Desikan RS, Djurovic S, Dale AM. (2013). Improved detection of common variants associated with schizophrenia and bipolar disorder using pleiotropy-informed conditional false discovery rate. PLoS Genet. 9(4): e1003455.
Baum AE, Akula N, Cabanero M, Cardona I, Corona W, Klemens B, Schulze TG, Cichon S, Rietschel M, Nothen MM, Georgi A, Schumacher J, Schwarz M, Abou JR, Hofels S, Propping P, Satagopan J, Detera-Wadleigh SD, Hardy J, and McMahon FJ. (2008). A genome-wide association study implicates diacylglycerol kinase eta (DGKH) and several other genes in the etiology of bipolar disorder. Mol Psychiatry. 13(2): 197–207. Bergen SE, O’Dushlaine CT, Ripke S, Lee PH, Ruderfer DM, Akterin S, Moran JL, Chambert KD, Handsaker RE, Backlund L, Osby U, McCarroll S, Landen M, Scolnick EM, Magnusson PK, Lichtenstein P, Hultman CM, Purcell SM, Sklar P, and Sullivan PF. (2012). Genome-wide association study in a Swedish population yields support for greater CNV and MHC involvement in schizophrenia compared with bipolar disorder. Mol Psychiatry. 17(9): 880–886. Bertolino A, Rubino V, Sambataro F, Blasi G, Latorre V, Fazio L, Caforio G, Petruzzella V, Kolachana B, Hariri A, Meyer-Lindenberg A, Nardini M, Weinberger DR, and Scarabino T. (2006). Prefrontal-hippocampal coupling during memory processing is modulated by COMT val158met genotype. Biol Psychiatry. 60(11): 1250–1258. Bigos KL, Mattay VS, Callicott JH, Straub RE, Vakkalanka R, Kolachana B, Hyde TM, Lipska BK, Kleinman JE, and Weinberger DR. (2010). Genetic variation in CACNA1C affects brain circuitries related to mental illness. Arch Gen Psychiatry. 67(9): 939–945. Blasi G, Mattay VS, Bertolino A, Elvevag B, Callicott JH, Das S, Kolachana BS, Egan MF, Goldberg TE, Weinberger DR. (2005). Effect of catechol-O-methyltransferase val158met genotype on attentional control. J Neurosci. 25(20): 5038–5045. Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H, McClay J, Mill J, Martin J, Braithwaite A, Poulton R. (2003). Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science. 301(5631): 386–389. Chen CH, Suckling J, Lennox BR, Ooi C, Bullmore ET. (2011). A quantitative meta-analysis of fMRI studies in bipolar disorder. Bipolar Disord. 13(1): 1–15. Craddock N, Owen MJ. (2010). The Kraepelinian dichotomy—going, going … but still not gone. Br J Psychiatry. 196(2): 92–95. Craddock N, Sklar P. (2009). Genetics of bipolar disorder: successful start to a long journey. Trends Genet. 25(2): 99–105. Demjaha A, MacCabe JH, Murray RM. (2012). How genes and environmental factors determine the different neurodevelopmental trajectories of schizophrenia and bipolar disorder. Schizophr Bull. 38(2): 209–214. Drabant EM, Hariri AR, Meyer-Lindenberg A, Munoz KE, Mattay VS, Kolachana BS, Egan MF, Weinberger DR. (2006). Catechol O-methyltransferase val158met genotype and neural mechanisms related to affective arousal and regulation. Arch Gen Psychiatry. 63(12): 1396–1406. Erk S, Meyer-Lindenberg A, Schnell K, Opitz von BC, Esslinger C, Kirsch P, Grimm O, Arnold C, Haddad L, Witt SH, Cichon S, Nothen MM, Rietschel M, Walter H. (2010). Brain function in carriers of a genome-wide supported bipolar disorder variant. Arch Gen Psychiatry. 67(8): 803–811. Esslinger C, Walter H, Kirsch P, Erk S, Schnell K, Arnold C, Haddad L, Mier D, Opitz von BC, Raab K, Witt SH, Rietschel M, Cichon S, Meyer-Lindenberg A. (2009). Neural mechanisms of a genome-wide supported psychosis variant. Science. 324(5927): 605. Etain B, Henry C, Bellivier F, Mathieu F, Leboyer M. (2008). Beyond genetics: childhood affective trauma in bipolar disorder. Bipolar Disord. 10(8): 867–876. Ferreira MA, et. al. (2008). Collaborative genome-wide association analysis supports a role for ANK3 and CACNA1C in bipolar disorder. Nat Genet. 40(9): 1056–1058. Franke B, Vasquez AA, Veltman JA, Brunner HG, Rijpkema M, Fernandez G. (2010). Genetic variation in CACNA1C, a gene associated with bipolar disorder, influences brainstem rather
2 3 8 part iv : I maging G enetics and D isease
than gray matter volume in healthy individuals. Biol Psychiatry. 68(6): 586–588. Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D, Egan MF, Weinberger DR. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science. 297(5580): 400–403. Hariri AR, Weinberger DR. (2003). Imaging genomics. Br Med Bull. 65: 259–270. Heinrich A, Lourdusamy A, Tzschoppe J, Vollstadt-Klein S, Buhler M, Steiner S, Bach C, Poustka L, Banaschewski T, Barker G, Buchel C, Conrod P, Garavan H, Gallinat J, Heinz A, Ittermann B, Loth E, Mann K, Martinot JL, Paus T, Pausova Z, Smolka M, Strohle A, Struve M, Witt S, Flor H, Schumann G, Rietschel M, and Nees F. (2013). The risk variant in ODZ4 for bipolar disorder impacts on amygdala activation during reward processing. Bipolar Disord. 15(4): 440–445. Houenou J, d’Albis MA, Vederine FE, Henry C, Leboyer M, Wessa M. (2012). Neuroimaging biomarkers in bipolar disorder. Front Biosci. (Elite Ed) 4: 593–606. Hulvershorn LA, Karne H, Gunn AD, Hartwick SL, Wang Y, Hummer TA, Anand A. (2012). Neural activation during facial emotion processing in unmedicated bipolar depression, euthymia, and mania. Biol Psychiatry. 71(7): 603–610. Jogia J, Ruberto G, Lelli-Chiesa G, Vassos E, Maieru M, Tatarelli R, Girardi P, Collier D, and Frangou S. (2011). The impact of the CACNA1C gene polymorphism on frontolimbic function in bipolar disorder. Mol Psychiatry. 16(11): 1070–1071. Kempton MJ, Geddes JR, Ettinger U, Williams SC, Grasby PM. (2008). Meta-analysis, database, and meta-regression of 98 structural imaging studies in bipolar disorder. Arch Gen Psychiatry. 65(9): 1017–1032. Kempton MJ, Ruberto G, Vassos E, Tatarelli R, Girardi P, Collier D, Frangou S. (2009). Effects of the CACNA1C risk allele for bipolar disorder on cerebral gray matter volume in healthy individuals. Am J Psychiatry 166(12): 1413–1414. Krug A, Nieratschker V, Markov V, Krach S, Jansen A, Zerres K, Eggermann T, Stocker T, Shah NJ, Treutlein J, Muhleisen TW, Kircher T. (2010). Effect of CACNA1C rs1006737 on neural correlates of verbal fluency in healthy individuals. Neuroimage. 49(2): 1831–1836. Lee MT, Chen CH, Lee CS, Chen CC, Chong MY, Ouyang WC, Chiu NY, Chuo LJ, Chen CY, Tan HK, Lane HY, Chang TJ, Lin CH, Jou SH, Hou YM, Feng J, Lai TJ, Tung CL, Chen TJ, Chang CJ, Lung FW, Chen CK, Shiah IS, Liu CY, Teng PR, Chen KH, Shen LJ, Cheng CS, Chang TP, Li CF, Chou CH, Chen CY, Wang KH, Fann CS, Wu JY, Chen YT, Cheng AT. (2011). Genome-wide association study of bipolar I disorder in the Han Chinese population. Mol Psychiatry. 16(5): 548–556. Lichtenstein P, Yip BH, Bjork C, Pawitan Y, Cannon TD, Sullivan PF, Hultman CM. (2009). Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet. 373(9659): 234–239. Linke J, Witt SH, King AV, Nieratschker V, Poupon C, Gass A, Hennerici MG, Rietschel M, Wessa M. (2012). Genome-wide supported risk variant for bipolar disorder alters anatomical connectivity in the human brain. Neuroimage. 59(4): 3288–3296. Liu X, Akula N, Skup M, Brotman MA, Leibenluft E, McMahon FJ. (2010). A genome-wide association study of amygdala activation in youths with and without bipolar disorder. J Am Acad Child Adolesc Psychiatry. 49(1): 33–41. Mattingsdal M, Brown AA, Djurovic S, Sønderby IE, Server A, Melle I, Agartz I, Hovig E, Jensen J, Andreassen OA. (2013). Pathway analysis of genetic markers associated with a functional MRI faces paradigm implicates polymorphisms in calcium responsive pathways. Neuroimage. 70: 143–149. McDonald C, Zanelli J, Rabe-Hesketh S, Ellison-Wright I, Sham P, Kalidindi S, Murray RM, Kennedy N. (2004). Meta-analysis of magnetic resonance imaging brain morphometry studies in bipolar disorder. Biol Psychiatry. 56(6): 411–417. chapter 1 5 : I maging G enetics of B ipolar D isorder
Merikangas KR, Jin R, He JP, Kessler RC, Lee S, Sampson NA, Viana MC, Andrade LH, Hu C, Karam EG, Ladea M, Medina-Mora ME, Ono Y, Posada-Villa J, Sagar R, Wells JE, Zarkov Z. (2011). Prevalence and correlates of bipolar spectrum disorder in the world mental health survey initiative. Arch Gen Psychiatry. 68(3): 241–251. Mier D, Kirsch P, Meyer-Lindenberg A. (2010). Neural substrates of pleiotropic action of genetic variation in COMT: a meta-analysis. Mol Psychiatry. 15(9): 918–927. Ousdal OT, Anand BA, Jensen J, Nakstad PH, Melle I, Agartz I, Djurovic S, Bogdan R, Hariri AR, Andreassen OA. (2012). Associations between variants near a monoaminergic pathways gene (PHOX2B) and amygdala reactivity: a genome-wide functional imaging study. Twin Res Hum Genet. 15(3): 273–285. Paulus FM, Krach S, Bedenbender J, Pyka M, Sommer J, Krug A, Knake S, Nothen MM, Witt SH, Rietschel M, Kircher T, Jansen A. (2013). Partial support for ZNF804A genotype-dependent alterations in prefrontal connectivity. Hum Brain Mapp. 34(2): 304–313. Perrier E, Pompei F, Ruberto G, Vassos E, Collier D, Frangou S. (2011). Initial evidence for the role of CACNA1C on subcortical brain morphology in patients with bipolar disorder. Eur Psychiatry. 26(3): 135–137. PGC. (2011). Large-scale genome-wide association analysis of bipolar disorder identifies a new susceptibility locus near ODZ4. Nat Genet. 43(10): 977–983. Radua J, Surguladze SA, Marshall N, Walshe M, Bramon E, Collier DA, Prata DP, Murray RM, McDonald C. (2013). The impact of CACNA1C allelic variation on effective connectivity during emotional processing in bipolar disorder. Mol Psychiatry 18(5): 526–527. Rimol LM, Hartberg CB, Nesvag R, Fennema-Notestine C, Hagler DJ, Jr., Pung CJ, Jennings RG, Haukvik UK, Lange E, Nakstad PH, Melle I, Andreassen OA, Dale AM, Agartz I. (2010). Cortical thickness and subcortical volumes in schizophrenia and bipolar disorder. Biol Psychiatry. 68(1): 41–50. Rogers BP, Morgan VL, Newton AT, and Gore JC. (2007). Assessing functional connectivity in the human brain by fMRI. Magn Reson Imaging. 25(10): 1347–1357. Schott BH, Seidenbecher CI, Fenker DB, Lauer CJ, Bunzeck N, Bernstein HG, Tischmeyer W, Gundelfinger ED, Heinze HJ, Duzel E. (2006). The dopaminergic midbrain participates in human episodic memory formation: evidence from genetic imaging. J Neurosci. 26(5): 1407–1417. Schulze TG, Detera-Wadleigh SD, Akula N, Gupta A, Kassem L, Steele J, Pearl J, Strohmaier J, Breuer R, Schwarz M, Propping P, Nothen MM, Cichon S, Schumacher J, Rietschel M, and McMahon FJ. (2009). Two variants in Ankyrin 3 (ANK3) are independent genetic risk factors for bipolar disorder. Mol Psychiatry. 14(5): 487–491. Scott LJ, Muglia P, Kong XQ, Guan W, Flickinger M, Upmanyu R, Tozzi F, Li JZ, Burmeister M, Absher D, Thompson RC, Francks C, Meng F, Antoniades A, Southwick AM, Schatzberg AF, Bunney WE, Barchas JD, Jones EG, Day R, Matthews K, McGuffin P, Strauss JS, Kennedy JL, Middleton L, Roses AD, Watson SJ, Vincent JB, Myers RM, Farmer AE, Akil H, Burns DK, Boehnke M. (2009). Genome-wide association and meta-analysis of bipolar disorder in individuals of European ancestry. Proc Natl Acad Sci U S A. 106(18): 7501–7506. Serretti A, Mandelli L. (2008). The genetics of bipolar disorder: genome ‘hot regions,’ genes, new potential candidates and future directions. Mol Psychiatry. 13(8): 742–771. Sklar P, Smoller JW, Fan J, Ferreira MA, Perlis RH, Chambert K, Nimgaonkar VL, McQueen MB, Faraone SV, Kirby A, de Bakker PI, Ogdie MN, Thase ME, Sachs GS, Todd-Brown K, Gabriel SB, Sougnez C, Gates C, Blumenstiel B, Defelice M, Ardlie KG, Franklin J, Muir WJ, McGhee KA, MacIntyre DJ, McLean A, VanBeck M, McQuillin A, Bass NJ, Robinson M, Lawrence J, Anjorin A, Curtis D, Scolnick EM, Daly MJ, Blackwood DH, Gurling HM, Purcell SM. (2008). Whole-genome association study of bipolar disorder. Mol Psychiatry. 13(6): 558–569. Smith EN, Bloss CS, Badner JA, Barrett T, Belmonte PL, Berrettini W, Byerley W, Coryell W, Craig D, Edenberg HJ, Eskin E,
239
Foroud T, Gershon E, Greenwood TA, Hipolito M, Koller DL, Lawson WB, Liu C, Lohoff F, McInnis MG, McMahon FJ, Mirel DB, Murray SS, Nievergelt C, Nurnberger J, Nwulia EA, Paschall J, Potash JB, Rice J, Schulze TG, Scheftner W, Panganiban C, Zaitlen N, Zandi PP, Zollner S, Schork NJ, Kelsoe JR. (2009). Genome-wide association study of bipolar disorder in European American and African American individuals. Mol Psychiatry. 14(8): 755–763. Smolka MN, Schumann G, Wrase J, Grusser SM, Flor H, Mann K, Braus DF, Goldman D, Buchel C, Heinz A. (2005). Catechol-Omethyltransferase val158met genotype affects processing of emotional stimuli in the amygdala and prefrontal cortex. J Neurosci. 25(4): 836–842. Smoller JW, Craddock N, Kendler K, Lee PH, Neale BM, Nurnberger JI, Ripke S, Santangelo S, and Sullivan PF. (2013). Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis. Lancet. 381(9875): 1371–1379. Smoller JW, Finn CT. (2003). Family, twin, and adoption studies of bipolar disorder. Am J Med Genet C Semin Med Genet. 123C(1): 48–58. Sprooten E, McIntosh AM, Lawrie SM, Hall J, Sussmann JE, Dahmen N, Konrad A, Bastin ME, Winterer G. (2012). An investigation of a genomewide supported psychosis variant in ZNF804A and white matter integrity in the human brain. Magn Reson Imaging. 30(10): 1373–1380. Stein JL, et al. (2012). Identification of common variants associated with human hippocampal and intracranial volumes. Nat Genet. 44(5): 552–561. Sullivan PF. (2010). The psychiatric GWAS consortium: big science comes to psychiatry. Neuron 68 (2): 182–186. Takata A, Kim SH, Ozaki N, Iwata N, Kunugi H, Inada T, Ujike H, Nakamura K, Mori N, Ahn YM, Joo EJ, Song JY, Kanba S, Yoshikawa T, Kim YS, Kato T. (2011). Association of ANK3 with bipolar disorder confirmed in East Asia. Am J Med Genet B Neuropsychiatr Genet. 156B(3): 312–315. Tesli M, Koefoed P, Athanasiu L, Mattingsdal M, Gustafsson O, Agartz I, Rimol LM, Brown A, Wirgenes KV, Smorr LL, Kahler AK, Werge T, Mors O, Mellerup E, Jonsson EG, Melle I, Morken G, Djurovic S, Andreassen OA. (2011). Association analysis of ANK3 gene variants in nordic bipolar disorder and schizophrenia case-control samples. Am J Med Genet B Neuropsychiatr Genet. 156B(8): 969–974.
Tesli M, Skatun KC, Ousdal OT, Brown AA, Thoresen C, Agartz I, Melle I, Djurovic S, Jensen J, Andreassen OA. (2013). CACNA1C risk variant and amygdala activity in bipolar disorder, schizophrenia and healthy controls. PLoS One. 8(2): e56970. Tesli M, Egeland R, Sønderby IE et al. (2013). No evidence for association between bipolar disorder risk gene variants and brain structural phenotypes. J Affect Disord. 151(1): 291–297. Thimm M, Kircher T, Kellermann T, Markov V, Krach S, Jansen A, Zerres K, Eggermann T, Stocker T, Shah NJ, Nothen MM, Rietschel M, Witt SH, Mathiak K, Krug A. (2011). Effects of a CACNA1C genotype on attention networks in healthy individuals. Psychol Med. 41(7): 1551–1561. Vederine FE, Wessa M, Leboyer M, Houenou J. (2011). A meta-analysis of whole-brain diffusion tensor imaging studies in bipolar disorder. Prog Neuropsychopharmacol Biol Psychiatry. 35(8): 1820–1826. Wang F, McIntosh AM, He Y, Gelernter J, Blumberg HP. (2011). The association of genetic variation in CACNA1C with structure and function of a frontotemporal system. Bipolar Disord. 13(7–8): 696–700. Wessa M, Linke J, Witt SH, Nieratschker V, Esslinger C, Kirsch P, Grimm O, Hennerici MG, Gass A, King AV, Rietschel M. (2010). The CACNA1C risk variant for bipolar disorder influences limbic activity. Mol Psychiatry 15(12): 1126–1127. Whalley HC, Papmeyer M, Romaniuk L, Johnstone EC, Hall J, Lawrie SM, Sussmann JE, McIntosh AM. (2012). Effect of variation in diacylglycerol kinase eta (DGKH) gene on brain function in a cohort at familial risk of bipolar disorder. Neuropsychopharmacology. 37(4): 919–928. Whalley HC, Papmeyer M, Sprooten E, Romaniuk L, Blackwood DH, Glahn DC, Hall J, Lawrie SM, Sussmann J, McIntosh AM. (2012). The influence of polygenic risk for bipolar disorder on neural activation assessed using fMRI. Transl Psychiatry. 2: e130. Winterer G, Musso F, Vucurevic G, Stoeter P, Konrad A, Seker B, Gallinat J, Dahmen N, Weinberger DR. (2006). COMT genotype predicts BOLD signal and noise characteristics in prefrontal circuits. Neuroimage. 32(4): 1722–1732. Womer FY, Kalmar JH, Wang F, Blumberg HP. (2009). A ventral prefrontal-amygdala neural system in bipolar disorder: a view from neuroimaging research. Acta Neuropsychiatr. 21(6): 228–238. WTCCC. (2007). Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 447(7145): 661–678.
2 4 0 part iv : I maging G enetics and D isease
16. GENETIC NEUROIMAGING STUDIES OF BASAL GANGLIA DISORDERS Trevor W. Robbins, James B. Rowe, and Roger A. Barker
INTRODUCTI O N The combination of genetic and neuroimaging studies has proved to be a bold, but innovative, step in neuropsychiatry and neurology, and one that is already yielding new insights into many disorders. This chapter focuses on Parkinson’s disease, the most studied disorder with basal ganglia pathology. We also survey the utility of combined neuroimaging genetics for other basal ganglia disorders, such as Huntington’s disease, in which there is a well-known single genetic contribution via an autosomal dominant mode of inheritance. Neuroimaging genetics has yet to be developed in “sporadic” basal ganglia disorders such as multiple system atrophy, progressive supranuclear palsy, and dystonia. Although genome-wide association (GWA) studies have identified several significant genetic foci for these disorders, they have not yet led to stratified neuroimaging studies. THE GENETI C BACKG RO UN D TO PA R KI N SO N ’S D I SEASE Despite long-held views that the etiology of Parkinson’s disease (PD) results primarily from environmental causes leading to nigro-striatal dopamine (DA) cell degeneration, which is held to be mainly responsible for its characteristic motor symptoms, this simple pathophysiological picture has changed considerably in the past two decades. Thus, PD is now recognized to involve non-DA pathology, with ancillary and heterogeneous symptoms of cognitive deficit and depression, and may even originate outside the CNS. PD is subject to considerable genetic influence; at least 3%–5% of PD cases are now thought to result from monogenic forms of the disease, not including the
markedly increased risk of PD conveyed by mutations in the glucocerebrosidase (GBA) gene (Van der Vegt et al. 2009). Advances in neuroimaging methodology, including ligand-based positron emission tomography (PET), structural magnetic resonance imaging (MRI), and functional MRI (fMRI), have contributed enormously to our understanding of the underlying neuropathology. However, the powerful technique of combining neuroimaging methods with genetically stratified cases of PD patients (and indeed those of other basal ganglia disorders) is in its infancy. The basic premise underlying this strategy is a prominent theme of this volume, but basically depends on the identification of powerful new phenotypic descriptions of the disease and distinct functional brain networks associated with different genetic polymorphisms. The approach may (1) clarify the pathophysiological mechanisms underlying the expression of different variants of PD, including the prominent sporadic or idiopathic forms, (2) enhance the search for biomarkers for prodromal PD as well as other basal ganglia disorders, (3) identify possible compensatory changes in brain networks occurring as a consequence of the early stages of neurodegeneration, and (4) reveal fundamental aspects of normal CNS function. There are several sources of relevant genetic variation for PD; particularly important are those genes associated with the etiology, pathogenesis, and medication by DA agents. Briefly, the genetic and chromosomal loci linked to familial forms of PD are designated as PARK1–13, including six autosomal dominant forms (1, 3, 5, 8, 11, and 13) and four recessive forms (2, 6, 7, and 9). PARK1 and PARK4 are associated with alpha synuclein and early onset PD. PARK2, 6, and 7 are all associated with different variants of early PD. Of these, PARK2 is associated with 241
Parkin, and PARK6 with PINK1 (PTEN-induced putative kinase 1). PARK8 is associated with LRRK2 (leucine-rich repeat kinase 2) and has variable presentations, including classical “sporadic” PD (see Van der Vegt et al. 2009). Of late, heterozygote mutations of the GBA gene, linked to the development of Gaucher’s disease (GD), has been shown to be the commonest single gene underlying sporadic and familial PD (Sidransky and Lopez 2012) being present in about 10% of incident cases in England (Winder-Rhodes et al. 2013). Such mutations have been shown to increase not only the risk of developing PD but also the rate of cognitive decline (Sidransky and Lopez 2012). These findings have been confirmed in a prospective study of incident cases of newly diagnosed PD where the increased risk of progression to dementia was five times greater in those with a GBA mutation compared to those without (Winder-Rhodes et al. 2013). Furthermore, a cross-sectional study found cognitive impairment to be more frequent and more severe in patients with mutations of GBA than matched controls without GBA mutations (Sidransky and Lopez 2012). Despite this promising advance, there have been to date no functional imaging studies of GBA mutations in PD dementia. For other common polymorphisms such as the MAPT (Microtubule Associated Protein Tau) haplotype, the H1/ H1 haplotypic variant has been associated not only with an increased risk of developing Parkinson’s disease but also with accelerated progression to Parkinson’s disease dementia (Williams-Gray et al. 2013). The H1 haplotype is associated with an increase in 4-repeat tau isoforms in the brain, and this is thought to play a role in the pathogenic development of this condition (as well as progressive supranuclear palsy), although the exact mechanism for this is currently unclear (Williams-Gray et al. 2009). A recent in vitro study has suggested one possible mechanism for an interaction with tau cross-seeding to promote alpha synuclein aggregation (Guo et al 2013). The H1/H1 haplotype has also been associated with changes in cognition and behavior in PD, and associated regional brain activity (see further discussion later in this chapter). Apolipoprotein E (APOE) genotype is an established factor in the susceptibility to Alzheimer’s disease (AD). The establishment of pathological overlap between AD and PD dementia has stimulated research to determine whether there may be a role for APOE genotype in risk of PD dementia. However, a large meta-analysis has failed to show any clear support for this connection (Williams-Gray et al. 2009). Finally, several common genetic polymorphisms modulate DA transmission, including the catecholamine-
O-methyl transferease (or COMT) polymorphism and the DRD2 receptor Taq polymorphisms (DRD2 TaqIA). These polymorphisms are not thought to be pathogenic, but they affect behavioral and cognitive functions and the response to DA medications. T H E C OM T POLYMOR PH IS M AND PAR K INS ON’ S DIS E AS E A seminal finding driving this field has been the discovery that the COMT Val158Met polymorphism that putatively affects prefrontal cortical (PFC) dopamine (DA) metabolism exerts an influence on healthy human volunteers, which involves: (1) prominent behavioral and cognitive effects depending on the presence of homozygosity for valine/valine (Val/Val) or methionine/methionine (Met/ Met) alleles at residue 158 of the COMT gene; (2) differential cognitive responses to indirect DA agonists such as d-amphetamine; and (3) changes in PFC activation in standard working memory paradigms (see Egan et al. 2001; Mattay et al. 2003). In general, cognitive functions such as working memory are deficient in individuals with Val/Val alleles compared with Met/Met individuals and heterozygotes; such individuals also exhibit reduced “efficiency” of activation of the PFC in “n-back” tasks, measured using blood-oxygenlevel dependent signal (BOLD), functional MRI (Mattay et al. 2003). In contrast, performance may be enhanced by d-amphetamine in Val/Val individuals, but not necessarily improved in those with Met/Met alleles (Mattay et al 2003). This constellation of findings can best be accommodated within a Yerkes-Dodson formulation: that cognitive performance is an inverted-U shaped function of PFC dopamine activity. These observations are highly relevant to those disorders affecting basal ganglia function in view of (1) the close functional relationships between frontal cortex and basal ganglia circuitry, as mediated for example, by the well-known parallel cortico-striatal loops with partially segregated functions in cognition and behavior, and (2) the fact that PFC DA is ultimately compromised in Parkinson’s disease, but may be normal or even enhanced early in disease (Rakshi et al. 1999; Kaasinen et al. 2001; Rakshi et al. 2012). Armed with this background information, in collaboration with Daniel Weinberger and colleagues, we embarked on a study of cognitive function in PD (Foltynie et al. 2004). We used a set of tasks that had previously been shown to exhibit both improvements and deficits following L-dopa medication. Tasks included a sensitive test
2 4 2 part iv : I maging G enetics and D isease
of PFC function, the CANTAB Tower of London (or “Stockings of Cambridge”) test of planning, together with simple CANTAB tests of visual pattern recognition and short-term spatial memory capacity. The effects of L-dopa on these tasks differs according to their engagement of DA-dependent functions of the dorsal and ventral striatum (Swainson et al. 2000; Cools et al. 2001). The large sample of 288 patients diagnosed with Parkinson’s disease exhibited a normal range of Val/Val (n = 60), Met/Met (n = 73) and heterozygote COMT (n = 155) genotypes. To our considerable surprise, and after almost excessive checking of samples, we found the opposite of what had been hypothesized (the simple prediction that cognitive deficits would be greater in those PD patients with the Val/ Val polymorphism). In fact, it was the Met/Met patients who did worse, especially if they were (1) male and (2) dosed with L-dopa. This finding was initially difficult to understand, as it implied that elevated levels of PFC DA (as a consequence of the Met/Met alleles) were detrimental to cognition in Parkinson’s disease. However, that result could have been predicted on the basis of an established inverted U-shaped function, especially as (1) we had previously found that L-dopa medication could be detrimental to certain forms of cognitive function in Parkinson’s disease (Swainson et al. 2000; Cools et al. 2001; Cools et al. 2003; Rowe et al. 2008); (2) it has been repeatedly shown in early Parkinson’s disease that there is an up-regulation of PFC DA (Kaasinen et al. 2001; Rakshi et al. 2012), presumably reflecting the repeatedly demonstrated reciprocal balance that appears to hold for striatal and PFC DA function (Pycock et al. 1980; Roberts et al. 1994); and (3) Meyer-Lindenberg et al. (2005) found that COMT genotype remarkably affected midbrain DA synthesis, such p = 0.017
1.5
(b)
1.0
–0.5 –1.0
tro ls con
D
0.0
yP
Working memory
0.5
rl ea
Change in TOL per year
(a)
that the Met/Met genotype in healthy volunteers actually reduced midbrain DA function, presumably in keeping with this reciprocal relationship between cortical and striatal DA. What is nevertheless surprising is that this apparent inverted U-shaped relationship for PFC DA and cognitive performance is present even in the context of the massive loss of striatal DA that characterizes Parkinson’s disease. This account makes several assumptions relating to the effects of COMT on DA function in Parkinson’s disease that turn out to be correct. Thus, it is in fact the case that Parkinson’s disease patients with the Met/Met alleles exhibit greater levels of DA synthesis in the PFC (Wu et al. 2012). Moreover, medial PFC DA is initially elevated in PD (Rakshi et al. 1999; Rakshi et al. 2012), before it eventually does show signs of depletion. One could make the dramatic and seemingly counterintuitive prediction that the progression of Parkinson’s disease could conceivably lead to a reversal of the advantage that Met/Met patients have over Val/Val patients, with the latter exhibiting deficits in the long term. Remarkably, this result has been shown; when Parkinson patients were retested 5 years later, it was the Met/Met patients who performed better than the Val/ Val cases (see Figure 16.1; Williams-Gray et al. 2009). This finding reinforces the model of an “inverted-U” shaped curve that links PFC DA to cognitive performance (Williams and Goldman-Rakic 1995; Robbins 2010). The common finding in functional imaging studies is that the COMT genetic polymorphism has much more clear-cut effects on brain function than behavior because the function of brain circuits represents a more sensitive and selective intermediate phenotype, with less variation than cognitive measures. Correspondingly, neuroimaging reveals significant genotype-related effects with much
–1.5 –2.0
Val/Val
Val/Met
Met/Met
COMT genotype
Dopamine level in PFC Val
COMT
Met
Figure 16.1 (A) Progression of cognitive deficits in early Parkinson’s disease patients 5 years on from original testing; performance in the Met/Met
patients improves, suggesting that some forms of cognition are determined (B) by an inverted U-shaped function of dopamine function in the prefrontal cortex, modulated by alleles of the COMT gene. Reproduced from Williams-Gray et al. (2009) with permission from the guarantors of Brain.
chapter 1 6 : G enetic N euroimaging S tudies of B asal G anglia D isorders
243
smaller numbers than the several hundred that are normally required to show reliable effects of COMT using cognitive or behavioral measures alone (Nombela et al. 2014). This principle was upheld in the first genotype-stratified functional imaging study performed with Parkinson’s disease patients (Williams-Gray et al. 2007). The main objective of this study was to investigate the underlying neural basis of this genotype–phenotype effect in Parkinson’s disease using fMRI. We scanned 31 patients with early Parkinson’s disease who were homozygous for either valine (Val) (n = 16) or methionine (Met) (n = 15) at the COMT Val158Met polymorphism during performance of an executive task comprising both Tower of London (planning) and simple subtracting (“control”) problems (Figure 16.2). A cross-group comparison between genetic subgroups for imaging data revealed a significant reduction in BOLD signal across the frontoparietal network involved in planning in Met/Met compared with Val/Val patients. Hence, it was shown that COMT genotype impacts on executive function in Parkinson’s disease through
direct influences on frontoparietal activation. Importantly, COMT genotype did not interact with striatal activation on this task, consistent with the view that the enzyme product of COMT primarily regulates cortical dopamine metabolism. Furthermore, the directionality of the genotype–phenotype effect observed in this study, when interpreted in the context of the existing literature, added weight to the hypothesis that the relationship between PFC function and DA levels follows an inverted U-shaped curve. Response times for planning problems were significantly longer in Met compared with Val homozygotes, whereas response times for control problems did not differ. These behavioral differences were more subtle than in the original Foltynie et al. (2004) study, with a clear disparity in terms of response times (p = 0.039) but only a trend toward reduced accuracy in Met compared with Val homozygotes, which did not reach significance. This was probably because of relatively small subgroup sizes and subgroup matching in terms of global cognitive ability and IQ to minimize confounding influences in this functional neuroimaging
(a) R PPC
L PPC
R DLPFC
Underactivation in Met-Met PD Patients compared with Val-Val on the Tower of London fronto-executive planning task
R frontopolar region
(b)
(c)
6
p = 0.05*
Val/Val Met/Met
Arbitrary units
5
(n = 31)
4 3 2 1
CORRECT
0 –1
R frontopolar
PPC
R DLPFC
Control
ROI Figure 16.2 Underactivation of frontoparietal network in Parkinson’s disease patients with Met/Met homozygotes compared to Val/Val homozygotes on
the Tower of London test of planning. (A) Regional BOLD response for critical contrast. (B) Screen-shot of Tower of London test (a slightly different form of the test that was used in the actual study). (C) Quantification of reduced BOLD response in fronto-polar cortex, dorsolateral prefrontal cortex, and posterior parietal cortex. Reproduced from Williams-Gray et al. (2009), The Journal of Neuroscience, with permission, Copyright of the Society of Neuroscience; Tower of London screenshot courtesy of Cambridge Cognition.
2 4 4 part iv : I maging G enetics and D isease
study. However, the finding is consistent with the view that genetic influences are more readily discernible on structural aspects of brain function than on cognitive or behavioral output. A second study showed less clear-cut genotype and disease-dependent impairments for a different form of executive functioning that can conveniently be labeled as “cognitive flexibility” (Williams-Gray et al. 2008). Twenty-nine medicated patients with early PD, homozygous for Val (n = 16) or Met (n = 13) were studied. Genotype had a significant effect on task strategy; Val/Val patients shifted attention within (intra-dimensional shifting), rather than between (extra-dimensional shifting), perceptual dimensions, similar to the normal tendency in controls. By contrast, those PD patients with Met/ Met alleles failed to form an attentional “set”; that is, their attention was more “labile.” Analysis of the fMRI data suggests that the impaired ability to form an attentional set in Met homozygotes again reflected under-recruitment of frontoparietal areas. As in the earlier study, there was no interaction of genotype with caudate nucleus BOLD activation. Another form of cognitive flexibility, reversal learning, was not affected by COMT genotype in that study, consistent with data in non-human primates suggesting that it is 5-HT modulation of orbitofrontal cortex that is a more important factor in this behavior than involvement of catecholamines (Robbins and Roberts 2007). The authors again accommodated these findings in an inverted-U shaped model with Met/Met individuals being further to the right, along the descending limb of the curve. A further study examined the genotype-imaging interaction in healthy older volunteers, and showed that whereas Val/Val PD patients had shown set-like behavior associated with heightened dorsolateral prefrontal cortical activation, the elderly adults showed the opposite pattern of results, with Val/Val homozygotes exhibiting poor set formation (Fallon et al. 2013). The latter finding is in agreement with data from non-human primates indicating that PFC DA depletion impairs attentional-set formation (Crofts et al. 2001). Williams-Gray et al. (2009) again concluded that early PD patients have a relative excess of PFC DA, and this supports the inverted-U shaped relationship between PFC DA levels and executive function (in this case, attentional set formation). However, it appears more difficult to account for the original deficits exhibited by both medicated and unmedicated PD patients in extra-dimensional set-shifting (e.g., Downes et al. 1989) in these terms. A third study (Rowe et al. 2010) examined whether the COMT polymorphisms influenced age-related changes in gray matter density and volume in healthy controls and
PD patients. Intriguingly, although there were greater gray matter volumes in the insula and ventral PFC in younger Val/Val individuals, genotypic differences were confined to the premotor cortex later in life. These data were consistent with the developmental role of dopamine, and the changes in patients with only one allele of COMT arising from velo-cardio-facial syndrome. However, the COMT polymorphism did not interact with PD in determining age-related changes in gray matter. Overall, these studies of COMT and Parkinson’s disease have changed the way we think about the role of DA in cognitive impairments in PD. It is becoming increasingly clear that some aspects of cognition (such as reversal learning, gambling-like decision-making, and impulsivity) are potentially impaired by L-dopa medication, in contrast to functions such as working memory, which generally exhibit improvements (Cools et al. 2001; Swainson et al. 2000; Cools et al. 2003; Rowe et al. 2008; Nombela et al. 2014). The findings with COMT suggest that the same cognitive functions could be improved or impaired as a joint function of COMT genotype and disease duration, as well as dopaminergic treatments. Moreover, some cognitive functions may not depend at all on striatal or PFC DA loss, and may instead have a completely different neuropathological basis, necessitating non-DA-ergic medication approaches. This has led therefore to a new “dual syndrome” hypothesis of cognitive deficits in PD: a DA-dependent fronto-striatal component; and a non-DA, probably cortical component depending on cholinergic influences and a PD-associated posterior cortical Lewy body pathology (Kehagia et al. 2010). These findings place into context the possible utility of COMT inhibitors such as entacapone or tolcapone in the treatment of PD. OT H E R GE NE T IC INF LU E NC E S ON NE U ROC OGNIT IV E F U NC T IONING IN PAR K INS ON’ S DIS E AS E Among other genetic polymorphisms of relevance to Parkinson’s disease are those associated with brain-derived neurotrophic factor (BDNF), which has been prominently implicated in the survival and functioning of midbrain DA neurons and in the regulation of DA D3 receptor. Moreover, the BDNF Val66Met polymorphism modulates the hippocampal mediation of episodic memory in healthy volunteers (and in patients with schizophrenia) in a classic “genetic neuroimaging” study (Egan et al. 2003). For Parkinson’s disease, most studies have focused on such factors as age of onset and the development of L-dopa dyskinesias. However,
chapter 1 6 : G enetic N euroimaging S tudies of B asal G anglia D isorders
245
Foltynie et al. (2005) reported a gender-specific influence on planning in 291 patients (overlapping heavily with the cohort used in the COMT studies) with Parkinson’s disease, using the CANTAB Tower of London task. Those patients with low rates of BDNF secretion, as a consequence of Met alleles, performed significantly better than those with high rates. The effect was most apparent in women and among those patients with prior DA-ergic medication, consistent with previous evidence of interactive modulations of DA function by estrogen and BDNF that lead to relative DA hyperactivity in the PFC and consequent cognitive impairment. Interestingly, the effects of the COMT and BDNF polymorphisms within the same population of PD patients were entirely additive and independent. It would be of evident interest and importance to subject a selection of these PD patients with BDNF polymorphisms to a functional imaging study similar to that described above in order to determine whether modulations of cognitive function by COMT and BDNF in PD produce common effects in terms of regional BOLD responses during fMRI. However, the BDNF Val66Met polymorphism has not been established to be important in longitudinal studies of cognitive decline (Williams-Gray et al. 2013). The other major genetic polymorphism known to affect normal cognition that has also been studied in the context of PD is of the DA D2 receptor, TaqIA A1 and A2 alleles. Their contribution could be expected to be especially relevant to PD because of evidence that the DA D2 receptor mediates many of the therapeutic effects of L-dopa. The human DRD2 gene is located on chromosome 11q22-q23 (Grandy et al. 1989), and individuals with the A1 allele (genotypes A1/A1 and A1/A2) have reduced brain DA-ergic function compared with non-carriers (genotype A2/A2). This is consistent with studies of European PD populations reporting an increased risk of developing PD in the presence of the DRD2 TaqIA restriction fragment length polymorphism variant allele A1 (Grevle et al 2000; Oliveri et al. 2000). Bartres-Faz et al. (2006) directly compared 10 A1/A2 heterozygote PD patients with 14 A2/A2 homozygotes in a functional imaging setting using a complex sequential motor task. They found that the A1 allele PD carriers activated a larger network of bilateral cerebral areas, including cerebellar and premotor regions, than a group of PD non-carriers matched for clinical and demographic variables. The findings were interpreted as showing recruitment of compensatory cerebral resources during motor processing in PD patients carrying the A1 allele. A similar approach has been used to identify functional effects or compensation in PD patients with genetic mutations such as PARK2 (Parkin). Indeed, pre-symptomatic carriers
provide a powerful platform for studying the pathophysiology of disease, unconfounded by the major cognitive and motor changes associated with symptomatic disease. Several PET studies have shown that there is a latent pre-synaptic deficit of midbrain (nigro-striatal) DA neurotransmission in PD mutation carriers who have yet to show clinical symptoms. In order to test for possible changes in network function, Buhmann et al. (2005) contrasted fMRI activations for interoceptively selected as opposed to exteroceptively cued finger movements in Parkin mutation carriers. In sporadic PD there is attenuated activation of the rostral supplementary motor area and right dorsolateral PFC during internally but not externally elicited finger movements. In comparison, Parkin mutation carriers did not exhibit any task-related deficits but, relative to healthy controls, exhibited increased activation for internally selected movements in the left dorsal premotor cortex and in the right rostral cingulate motor area, which was also shown to be more strongly coupled in Parkin mutation carriers. This study thus indicates that a pre-synaptic DA deficit in the striatum modifies cortical processing of a motor task sensitive to basal ganglia dysfunction and also provides a “biomarker” for PD, assuming that the deficits found in Parkin-positive PD cases parallel those seen in the more common sporadic PD. Van Nuenen et al. (2009a, 2009b) went on to show that Parkin and PINK1 mutation carriers (pre-symptomatic) showed similar abnormal activation of the rostral SMA and premotor cortex. Once symptomatic, the activation of motor systems in Parkin-positive cases resembled sporadic PD, confirming the suitability of Parkin mutations for pre-symptomatic studies of PD. The paradigm could be used effectively in longitudinal studies to determine how such functional compensation eventually breaks down. A parallel study has focused on a voxel-based morphometry (VBM) study of the non-manifest heterozygote carriers of a Parkin or PINK1 mutation (Binkofski et al. 2007). These investigators showed that regional increases in gray matter volume of the posterior putamen and internal globus pallidus were inversely correlated with decreases in striatal 18F-Fluoro-Dopa uptake. Whereas the perspective of these studies has been primarily from a motor viewpoint, it is likely that a similar approach can be taken to cognitive symptoms, including parkinsonian dementia. Examples of this are our recent studies of the neurocognitive changes associated with the MAPT gene (Winder-Rhodes et al. 2013; Nombela et al. 2014). Using fMRI to map cortical changes in activity during the encoding of visual memories, we found that H1/HI homozygosity was associated with poorer memory recall in both PD patients and age-matched controls. These memory impairments were associated with hypoactivation of the
2 4 6 part iv : I maging G enetics and D isease
medial temporal lobe during memory encoding. There were some additional effects of MAPT on brain activation, however, these were found exclusively in PD patients; thus overall memory performance in H1 homozygotes with PD showed a pronounced hypoactivation of the medial temporal lobes relative to H2 carriers with PD; that is, this difference was only evident for PD patients and not for the control group, regardless of their genotype. We have therefore shown a link between MAPT haplotype, memory encoding, and the cognitive phenotype of PD. In addition, we have generated novel evidence that MAPT also modifies healthy individuals’ brain function during a memory task in the older population. GENETIC IMAG I N G I N HUNTINGTO N ’S D I SEASE Huntington’s disease (HD) is an autosomal, dominantly inherited progressive neurodegenerative disorder, predominantly of the basal ganglia and cortex, characterized by motor, cognitive, and psychiatric symptoms that are caused by an expanded cytosine-adenosine-guanine (CAG) trinucleotide sequence in exon 1 of the huntingtin gene (HTT) (Weir et al. 2011). That its genetic etiology is largely already understood means that one of the major issues in HD is not whether neuroimaging can help establish its diagnosis or particularly resolve its clinical heterogeneity. The issue is instead whether it can be used in tandem with genetic information more accurately to predict the course of its clinical features, potentially to enable therapeutic intervention before clinical features emerge (generally by the age of 45). For this purpose, the number of CAG repeats in the relevant HD gene can be very informative, younger onset being associated with greater CAG repeat number, but is not yet sufficient for very accurate predictions concerning the age of onset, rate of progression, and severity of clinical features. The logic of combining such genetic burden information with neuroimaging measures is clear inasmuch as the latter, through measures of structural and functional response, as well as functional connectivity, can presumably be relied upon to better predict cognitive, affective, and motor features. However, to date, it is unclear to what extent this strategy has been implemented successfully. To some extent, this approach has to be combined with evidence from other biomarkers, including biochemical ones. The genetic neuroimaging strategy would involve predictive testing of carriers of the CAG expansion within the HTT gene also being exposed to structural or functional neuroimaging methodologies, most obviously to relate
the degree of expansion quantitatively to imaging parameters or “biomarkers” (Kloppel et al. 2009: Weir et al. 2011). The prodromal cognitive picture may also inform which neural circuits are most likely at risk and include fronto-executive tasks such as the n-back working memory task (Georgiou-Karistianis et al. 2013), the Tower of London (Unschuld et al. 2013), and set-shifting (Lawrence et al. 1998; Gray et al. 2013), as well as obvious striatal targets such as motor sequence learning and self-timed finger tapping (reviews: Papp et al. 2011; Weir et al. 2011), with limbic involvement perhaps best assayed in tests of emotional recognition (Dogan et al. 2013). In structural terms, striatal atrophy correlates with age, time of disease onset and motor dysfunction, as well as CAG repeat length, being often present 15–20 years from disease onset with linear progression until that point. (Aylward et al. 1997, 2004; Harris et al. 1999; Paulsen et al. 2010). Quantitative analyses of neuroimaging findings in terms of CAG repeat number are quite rare. However, van Oostrom et al. (2005) did find that the rate of loss of striatal dopamine D2 receptor binding was highly significantly correlated with numbers of CAG repeats, as well as age, in 27 pre-manifest carriers of the HD mutation. Changes in cortical volume with MRI have been more variable in pre-manifest HD but may potentially explain phenotypic heterogeneity. Data for fMRI studies of hemodynamic response to neural activation by specific tasks have intriguingly revealed both increased and reduced activity (either BOLD or PET), the increases presumably reflecting compensatory functioning, where behavioral performance is intact (reviews: Kloppel et al. 2009; Weir et al. 2011). A logical inference might be that the former pattern is related to early neuronal dysfunction, whereas the latter is more likely related to imminent clinical abnormalities. Combined with information on age and genetic load in terms of numbers of CAG repeats, such findings might provide sensitive indicators of early and intermediate HD phenotypes. Possibly more sophisticated approaches in future will focus on functional connectivity. For example, Unschuld et al. (2013) performed BOLD fMRI and structural-MRI in 53 subjects with the HD mutation (41 pre-manifest and 12 early-onthe-course). Disease stage was estimated for each subject based on age, degree of putaminal atrophy, and the length of the CAG-repeat expansion. The Tower of London test was administered at three levels of difficulty as a test of fronto-executive function. While prodromal HD subjects performed similarly to controls, they exhibited reduced connectivity between the frontal cortex and the left premotor area.
chapter 1 6 : G enetic N euroimaging S tudies of B asal G anglia D isorders
247
There is an obvious additional need for longitudinal as distinct from cross-sectional studies of pre-manifest HD. One recent study performed a longitudinal follow-up of pre-manifest HD. Georgiou-Karistianis et al. (2013) measured 18-month longitudinal changes in both functional activation and functional connectivity during working memory as measured by performance on an N-back working memory task in pre-manifest HD (n = 27) and manifest HD (n = 17) and control participants (n = 23) in two separate scans. There were several major findings of interest. In the pre-manifest HD subjects, activation was increased longitudinally in the lateral and medial prefrontal, anterior cingulate, primary motor, and temporal areas cortically, and in the caudate and putamen subcortically. Those pre-manifest HD patients relatively far from predicted clinical onset showed further longitudinal increases in the right and left dorsolateral prefrontal cortex (DLPFC) compared with controls. During 1-BACK performance, functional connectivity between the right DLPFC and posterior parietal cortex, anterior cingulate, and caudate was also significantly reduced over 18 months. If genetic load in terms of CAG repeats can be linked convincingly to such changes in activation and connectivity, then the combined genetic-imaging approach may be of great utility in predicting future trajectories of confirmed HD patients. C ONCL US IONS Combined genetic and neuroimaging approaches are in their infancy in the analysis of neurodegenerative disorders of the basal ganglia, but promise much in terms of resolving the burgeoning heterogeneity of Parkinson’s disease, especially with respect to cognitive features, and for Huntington’s disease to its variable course. The studies we have reviewed here provide a platform for exploring further discoveries in the genetics of these disorders and their treatment, relating them to symptoms via the pathophysiology of defined neural circuitry. AC K NOW L EDG MEN TS We gratefully acknowledge support from the U.K. Parkinson’s Disease Society and NIHR funding of a Biomedical Research Centre at the University of Cambridge/Addenbrooke’s Hospital. JBR is a Wellcome Trust Senior Research Fellow in Clinical Science (103838).
The BCNI is jointly funded by the Medical Research Council and Wellcome Trust. Thanks to Sarah Mason for help with the literature review. R E F E R E NC E S Aylward EH, Li Q, Stine OC, Ranen N, Sherr M, Barta PE, Bylsma FW, Pearson GD, Ross CA. (1997). Longitudinal change in basal ganglia volume in patients with Huntington’s disease. Neurology. 48: 394–399. Aylward EH, Sparks BF, Field KM, Yallapragada, V, Shpritz BD, Rosenblatt A, Brandt J, Gourley LM, Liang K, Zhou H, Margolis RL, Ross CA. (2004). Onset and rate of striatal atrophy in preclinical Huntington’s disease Neurology. 63: 66–72. Batres-Faz D, Mart MJ, Junque C, Sole-Padulles C, Exquerra M, Braiten LB, Gaig C, Campdelacreu J, Mercader JM, Tolossa E. (2006). Increased cerebral activity in Parkinson’s disease patients carrying the DRD2 TaqIA A1 allel during a demanding motor task: a compensatory mechanism? Genes Brain Behav. 6: 588–592. Binkofski F, Reetz K, Gaser C, Hilker R, Hagenah J, Hedrich K, van Eimeren T, Thiel A, Buchel C, Pramstaller PP, Siebner HR, Klein C. (2007). Morphometric fingerprint of asymptomatic Parkin and PINK1 mutation carriers in the basal ganglia. Neurology. 69: 842–850. Buhmann C, Binkovski F, Klein C, Buchel C, van Eimeren van T, Erdmann C, Hedrich K, Kasten M, Hagenah J, Deuschl G, Pramstaller PP, Siebner HR. (2005). Motor reorganization in asymptomatic carriers of a single mutant Parkin allele: a human model for presymptomatic Parkinsonism. Brain. 128: 2281–2290. Cools R, Barker R, Sahakian BJ, Robbins TW. (2001). Enhanced or impaired cognitive function in Parkinson’s disease as a function of dopaminergic medication and task demands. Cerebral Cortex, 11: 1136–1143. Cools R, Barker RA, Sahakian BJ, Robbins, TW. (2003). L-Dopa medication remediates cognitive inflexibility, but increases impulsivity in patients with Parkinson’s disease. Neuropsychologia. 41: 1431–1441. Crofts HS, Dalley JW, Collins P, Van Denderen JCM, Everitt BJ, Robbins TW, Roberts AC. (2001). Differential effects of 6-OHDA lesions of the prefrontal cortex and caudate nucleus on the ability to acquire an attentional set. Cerebral Cortex 11: 1015–1026. Dogan I, Sass C, Mirzazade S, Kleiman A, Werner CJ, Pohl A, Schiefer J, Binkovski F, Schultz JB, Shah NJ, Reetz K. (2013). Neural correlates of impaired emotion processing in manifest Huntington’s disease. Soc Cogn Affect Neurosci. 9, 671–680. Downes JJ, Roberts AC, Sahakian BJ, Evenden JL, Morris RG, Robbins TW. (1989). Impaired extra-dimensional shift performance in medicated and unmedicated Parkinson’s disease: evidence for a specific attentional dysfunction. Neuropsychologia. 27: 1329–1343. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, Goldman D, Weinberger DR. (2001). Effect of COMT Val108/158 genotype on frontal lobe function and the risk for schizophrneia. Proc Natl Acad Sci U S A. 98: 6917–6922. Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, Zaitsev E, Gold B, Goldman D, Dean M, Lu B, Weinberger DR. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 112: 257–269. Fallon SJ, Williams-Gray CH, Barker RA, Owen AM, Hampshire A. (2013). Prefrontal dopamine levels determine the balance between cognitive stability and flexibility. Cerebral Cortex. 23: 361–369. Foltynie T, Goldberg TE, Lewis SJG, Blackwell AD, Kolachana BS, Weinberger DR, Robbins TW, Barker RA. (2004). Planning ability in Parkinson’s disease is influenced by the COMT val158met polymorphism. Movement Disord. 19: 885–891.
2 4 8 part iv : I maging G enetics and D isease
Foltynie T, Lewis SJG, Goldberg TE, Blackwell AD, Kolachana BS, Weinberger DR, Robbins TW, Barker RA. (2005). The BDNF Val66Met polymorphism has a gender specific influence on planning ability in Parkinson’s disease. J Neurology. 252: 833–838. Grandy DK, Marchionni MA, Makam H, Stofko RE, Alfano M, Frothingham L, Fischer JB, Burkehowie KJ, Bunzow JR, Server AC, Civelli O. (1989). Cloning of the CDNA and gene for a human D2-dopamine receptor. Proc Natl Acad Sci U S A. 86: 9762–9766. Grevle L, Guzey C, Hadidi H, Brennersted R, Idle JR, Aasly J. (2000). Allelic association between the DRD2 TaqI polymorphism and Parkinson’s disease. Movement Disord. 15: 1070–1074. Georgiou-Karistianis, N, Poudel GR., Dominguez DJF, Langmaid, R, Gray MA, Churchyard A, Chua P, Borowsky B, Egan GF, Stout, JC. (2013). Functional and connectivity changes during working memory in Huntington’s disease: 18 month longitudinal data from the IMAGE-HD study. Brain Cognition. 83(1): 80–91. Gray MA, Egan GF, Churchyard A, Chua P, Stout JC, Georgihou-Karistianis N. (2013). Prefrontal activity in Huntington’s disease reflects cognitive and neuropsychiatric disturbances: the IMAGE-HD study. Exp Neurol. 239: 218–228. Guo JL, Covell DJ, Daniels JP, Iba M, Stieber A, Zhang B, et al. (2013). Distinct alpha-synuclein strains differentially promote tau inclusions in neurons. Cell. 154(1):103–117. Harris GJ, Codori AM, Lewis RF, Schmidt E, Bedi A, Brandt J. (1999). Reduced basal ganglia blood flow and volume in pre-symptomatic gene-tested persons at risk for Huntington’s disease. Brain. 122: 1667–1678. Kaasinen V, Nurmi E, Bruck A, Bergman J, Solin O, Rinne JO. (2001). Increased frontal [(18)F] fluorodopa uptake in early Parkinson’s disease: sex differences in the prefrontal cortex. Brain. 124: 1125–1130. Kehagia A, Barker RA, Robbins TW. (2010). Neuropsychological and clinical heterogeneity of cognitive impairment and dementia in patients with Parkinson’s disease. Lancet Neurol. 9: 1200–1213. Kloppel S, Henley SM, Hobbs NZ, Wolf RC, Kassubek J, Tabrizi SJ, Frackowiak RSJ. (2009). Magnetic resonance imaging of Huntington’s disease: preparing for clinical trials. Neuroscience. 164: 205–219. Lawrence AD, Hodges JR, Rosser AE, Kershaw A, French-Constant C, Rubinsztein DC, Robbins TW, Sahakian BJ. (1998). Evidence for specific cognitive deficits in preclinical Huntington’s disease. Brain. 121: 1329–1341. Mattay VS, Goldberg TE, Fera F, Hariri AR, Tessitore A, Egan MF, Kolachana B, Callicott JH, Weinberger DR. (2003). Catechol-omethyltransferase val (158)-met genotype and individual variation in the response to amphetamine. Proc Natl Acad Sci U S A. 100: 6186–6191. Meyer-Lindenberg A, Kohn PD, Kolachana BS, Kipperhan S, McInerney-Leo A, Nussbaum R, Weinberger DR, Berman KF. (2005). Midbrain dopamine and prefrontal function in humans: interaction and modulation. Nat Neurosci. 8: 594–596. Nombela C, Rowe JB, Winder-Rhodes SE, Hampshire A, Owen AM, Breen DP, Duncan, GW, Khoo TK, Yarnall AJ, Firbank MJ, Chinnery PF, Robbins TW, O’Brien JT, Brooks, DJ, Burn DJ; ICICLE-PD study group, Barker RA. (2014). Genetic impact on cognition and brain function in newly diagnosed Parkinson's disease: ICICLE-PD study. Brain. 137: 2743–2758. Oliveri RL, Annesi G, Zappia M, et al. (2000). The dopamine D2 receptor gene is a susceptibility locus for Parkinson’s disease. Movement Disord. 15: 127–131. Papp KV, Kaplan RF, Snyder PJ. (2011). Biological markers of cognition in prodromal Huntington’s disease: a review. Brain Cognition. 77: 280–291. Paulsen JS, Nopoulos PC, Aylward E, Ross CA, Johnson H, Magnottta VA, Juhl A, Pierson RK, Mills J, Langbehn D, Nance M. (2010). Striatal and white matter predictors of estimated diagnosis for Huntington disease. Brain Res Bull. 82: 201–207.
Pycock C, Kerwin RW, Carter CJ. (1980). Effect of lesion of cortical dopamine terminals on subcortical dopamine receptors in rats. Nature. 286: 174–177. Rakshi JS, Uema T, Ito K, Bailey DL, Morrish PK, Ashburner J, Dagher A, Jenkins AH, Friston KJ, Brooks DJ. (1999). Frontal, midbrain and striatal dopaminergic function in early and advanced Parkinson’s disease: a 3D [(18)F]dopa-Pet study. Brain. 122: 1637–1650. Rakshi JS, Pavese N, Uema T, Ito K, Morrish PK, Bailey DL, Brooks DN. (2012). A comparison of the progression of early Parkinson’s disease in patients started on ropinirole or L-dopa: an F-18-dopa PET study. J Neural Transmission. 109: 1433–1443. Robbins TW. (2010). From behavior to cognition: functions of mesostriatal, mesolimbic and mesocortical dopamine systems. In Iversen LL, Iversen SD, Dunnett SB, Bjorklund A, ed., Dopamine handbook (pp. 203–214). New York: Oxford University Press. Robbins TW, Roberts AC. (2007). Differential regulation of fronto-executive function by the monoamines and acetylcholine. Cerebral Cortex. 17: i151–i160. Roberts AC, De Salvia MA, Wilkinson LS, Collins P, Muir JL, Everitt BJ, Robbins TW. (1994). 6-hydroxydopamine lesions of the prefrontal cortex in monkeys enhance performance on an analog of the Wisconsin Card Sort Test: possible interactions with subcortical dopamine. J Neurosci. 14: 2531–2544. Rowe JB, Hughes L, Ghosh BCP, Eckstein D, Williams-Gray CH, Fallon S, Barker RA, Owen AM. (2008). Parkinson’s disease and dopaminergic therapy—differential effects on movement, reward and cognition. Brain. 131: 2094–2105. Rowe JB, Hughes L, Williams-Gray CH, Bishop S, Fallon S, Barker RA, Owen AM. (2010). The val158met COMT polymorphism’s effect on atrophy in healthy aging and Parkinson’s disease. Neurobiol Aging. 31: 1064–1068. Sidransky E, Lopez G. (2012). The link between the GBA gene and parkinsonism. Lancet Neurol. 11(11): 986–998. Swainson R, Rogers RD, Sahakian BJ, Summers BA, Polkey CE, Robbins TW. (2000). Probabilistic learning and reversal deficits in patients with Parkinson’s disease or frontal or temporal lobe lesions: possible adverse effects of dopaminergic medication. Neuropsychologia. 38: 596–612. Unschuld PG, Liu X, Shanahan M, Margolis RL, Bassett SS, Brandt J, Schretlen DJ, Redgrave GW, Hua J, Hock C, Reading GA, van Zijl PC, Pekar JJ, Ross CA. (2013). Prefrontal executive function associated coupling relates to Huntington’s disease stage. Cortex. 49: 2661–2673. Van der Vegt JPM, Van Nuenen BFL, Bloem BR, Klien C, Siebner HR. (2009). Imaging the impact of genes on Parkinson’s disease. Neuroscience. 164: 191–204. Van Nuenen BFL, van Eimeren T, van der Vegt JPM, Buhmann C, Klein C, Bloem BR, Siebner HR. (2009a). Mapping preclinical compensation in Parkinson’s disease and imaging genomics approach. Movement Disord. 24: S703–S710. Van Nuenen BFL, Weis MM, Bloem BR, Reetz K, Van Eimeren T, Lohmann K, Hagenah J, Pramstaller PP, Binkovski F, Klein C, Siebner HR. (2009b). Heterozygous carriers of a Parkin or PINK1 mutation share a common functional endophenotype. Neurology. 72: 1041–1047. Van Oostrom JCH, Maguire RP, Verschuuren-Bemelmans CC, van der Duin LV, Pruim J, Roos RAC, Leenders KL. (2005). Striatal D2 dopamine receptors, metabolism, and volume in preclinical Huntington disease. Neurology. 65: 941–943. Weir D, Sturrock A, Leavitt BR. (2011). Development of biomarkers for Huntington’s disease. Lancet Neurol. 10: 573–590. Williams GV, Goldman-Rakic PS. (1995). Modulation of memory fields by dopamine D1 receptors in prefrontal cortex. Nature. 376: 572–575. Williams-Gray CH, Hampshire A, Robbins TW. Owen AM, Barker RA. (2007). Catechol O-methyltransferase Val158Met genotype
chapter 1 6 : G enetic N euroimaging S tudies of B asal G anglia D isorders
249
influences frontoparietal activity during planning in patients with Parkinson’s disease. J Neurosc. 27: 4832–4848. Williams-Gray CH, Hampshire A, Barker RA, Owen AM. (2008). Attentional control in Parkinson’s disease is dependent on COMT val (158) met genotype. Brain. 131: 397–408. Williams-Gray CH, Evans JR, Goris A, Foltynie T, Ban M, Robbins TW, et al. (2009). The distinct cognitive syndromes of Parkinson’s disease: 5 year follow-up of the CamPaIGN cohort. Brain. 132: 2958–2969. Williams-Gray CH, Goris A, Saiki M, Foltynie T, Compston DA, Sawcer SJ, et al. (2009). Apolipoprotein E genotype as a risk factor for susceptibility to and dementia in Parkinson’s disease. J Neurol. 256: 493–498. Williams-Gray CH, Mason SL, Evans JR, Foltynie T, Brayne C, Robbins TW, et al. (2013). The CamPaIGN study of Parkinson’s
disease: 10-year outlook in an incident population-based cohort. J Neurol Neurosurg Psychiatry, 84: 1258–1264. Winder-Rhodes SE, Evans JR, Ban M, Mason SL, Williams-Gray CH, Foltynie T, et al. (2013). Glucocerebrosidase mutations influence the natural history of Parkinson’s disease in a community-based incident cohort. Brain. 136: 392–399. Winder-Rhodes SE, Hampshire A, Rowe JB, Peelle JE, Robbins TW, Owen AM, Barker RA. (2013). Association between MAPT haplotype and memory function in patients with Parkinson’s disease and healthy aging individuals. Neurobiol Aging. 36: 1519–1528. Wu K, O’Keeffe D, Politis M, O’Keefe GC, Robbins TW, Bose SK, Brooks DJ, Piccini P, Barker RA. (2012). The catechol-Omethyltransferase Val(158)Met polymorphism modulates fronto-cortical dopamine turnover in early Parkinson’s disease: a PET study. Brain. 135: 2449–2457.
2 5 0 part iv : I maging G enetics and D isease
17. IMAGING GENETICS OF ANTISOCIAL BEHAVIOR AND PSYCHOPATHY Hayley M. Dorfman and Joshua W. Buckholtz
T
he problem of antisocial behavior has been with us since the dawn of our species. Wherever and whenever a group of humans seek to facilitate economic prosperity and community peace by establishing social norms, some within that group will persistently violate them. From the biblical Cain to the kunlangeta of the Yupi Eskimos and the arankan of Nigeria, nearly every culture on earth has had its peace and prosperity threatened by a small number of enduringly antisocial individuals. Today, antisocial behavior contributes to dysfunction and impairment across a range of psychiatric diagnoses, including borderline personality disorder (BPD), bipolar disorder, substance abuse, and some subtypes of major depression (e.g., angry/irritable). The two disorders in which antisocial behavior is most prominent, however, are antisocial personality disorder (APD) and psychopathy. These two disorders have been estimated to cost the United States over a trillion dollars per year in costs related to incarceration, treatment, and lost productivity. Yet in terms of our understanding of their causal neurobiology, they are among the most poorly characterized syndromes across the entire spectrum of psychopathology. This is due in part to social factors—by their very nature, antisocial offenders do not receive support from advocacy groups promoting their welfare and study—but also to the phenotypic heterogeneity of antisocial behavior and consequent challenges for clinical assessment. Here, we will review the current state of knowledge about the biological roots of antisocial behavior. To that end, we will (1) describe the clinical features of antisocial behavior, distinguishing between impulsive antisociality and psychopathy, (2) highlight insights from brain imaging work in antisocial populations, (3) discuss the genetic architecture of antisocial behavior generally, and (4) detail imaging genetic findings that shed light on the neurobiological
mechanisms through which heritable factors act to predispose antisociality. C LINIC AL F E AT U R E S AND AS S E S S M E N T OF ANT IS OC IAL B E H AV IOR One of the earliest clinical examinations of antisocial behavior dates to 1801, when French physician Philippe Pinel used the term “manie sans délire” (mania without delirium) to describe patients in his clinic who were impulsive and destructive, but showed no signs of psychosis (Pinel and Davis 1962). Many mark 1941 as the start of the “modern” era of research, as this was the year that Hervey Cleckley published his landmark book The Mask of Sanity. Cleckley noticed that a particular kind of troubling patient seemed to darken his clinic’s doorstep with disturbing regularity. These patients did not fit well in any of familiar category of psychopathology. They were persistently antisocial, but appeared to lack any recognizable motivation for their outrageous behavior. Most of them were superficially quite charming, while at the same time lacking any substantial emotional depth. They exhibited an absolute inability to accept responsibility for their behavior; this was coupled with an unnerving callousness toward others generally, and complete lack of remorse for their victims specifically. He termed them “psychopaths,” and it was then that “psychopathy” was first considered as a disorder in its own right. Clinical assessments, such as the Psychopathy Checklist—Revised (PCL-R) and Psychopathic Personality Inventory (PPI), parse the syndrome into two distinct symptom domains, or “factors” (Hare 1980; Lilienfeld and Andrews 1996). The first factor (F1) encompasses characteristic socioemotional aberrations, 251
such as glibness and superficiality; egocentricity and grandiosity; deficits in remorse, guilt, and empathy; and deceitful and manipulative behaviors. The second factor (F2) addresses the antisocial behaviors and lifestyle patterns evident in psychopaths. They show an insistent need for novelty and stimulation; are impulsive and show poor behavioral control; lack realistic long-term goals; and their lifestyle is peripatetic as well as parasitic, punctuated by frequent arrests for crimes ranging from theft and fraud to rape and murder (Kiehl and Hoffman 2011). Contrast these factors with the diagnostic criteria for APD, which highlight persistent criminal behavior, impulsivity, aggression, and irresponsibility (APA, 2000). Psychopathy F2 items overlap considerably with the DSM IV-TR criteria for APD, explaining in part the apparently high comorbidity between psychopathy and APD. However, the relationship between the two is asymmetric. Considering the entire population of criminal offenders, it is thought that ~20%– 30% meet criteria for psychopathy and ~85% meet criteria for APD. However, while the vast majority of those diagnosed with psychopathy would also merit an APD diagnosis, most offenders who meet criteria for APD would not be diagnosed as psychopaths (Kiehl and Hoffman 2011). This asymmetry highlights the fact that shared variance between APD and psychopathy reflects a common pattern of disinhibitory control, while unique variance reflects the social-interpersonal deficits that are relatively selective for psychopathy. We can thus separate psychopathy from impulsive antisociality on the basis of these social-interpersonal features. This distinction has a close parallel in the developmental psychopathology literature, with divisions between conduct disorder with and without “callous-unemotional” traits (CU+ or CU−) mirroring the division between APD and psychopathy. NEUROBIOL OG Y O F AN TI SO CI AL BEHA VIOR: B RAI N I MAG I N G Neuroimaging research in antisocial populations is challenging. There are significant practical hurdles associated with imaging incarcerated offenders or recruiting community volunteers with clinically relevant antisocial behaviors and traits. Assessment issues are also less than straightforward. As one example, the nosology of psychopathy is still a matter of heated debate. While the PCL-R was historically considered to be a gold-standard measure, some now question the degree to which it completely indexes the construct (Skeem and Cooke 2010). There is also much heated debate about how best to disentangle “psychopathy” from
the supervening construct of “antisociality.” Nevertheless, studies of brain structure, function, and connectivity are beginning to suggest specific neurobiological substrates for several characteristic features of antisocial behavior. STRUCTURE
In general, work to date suggests that antisocial behavior is associated with structural aberrations within frontolimbic and corticostriatal circuits. In particular, reductions in gray matter volume and cortical thickness have been observed in dorsolateral prefrontal cortex (DLPFC), anterior cingulate cortex (ACC), orbitofrontal cortex (OFC), and premotor cortex (Yang et al. 2005; Yang and Raine 2009; Ly et al. 2012). Similar reductions in gray matter volume have been observed in amygdala, hippocampus, and insula (Yang 2009; Yang et al. 2010; Ermer et al. 2012; Ermer et al. 2013). Moreover, while structural deficits in frontolimbic regions are evident in antisocial offenders, the opposite pattern is observed for corticostriatal circuitry. Several studies have reported increased striatal gray matter volume in psychopaths, which are observed even after controlling for potential confounds such as substance abuse history (Glenn et al. 2010; Shiffer et al. 2011; Ermer et al. 2012; Ermer et al. 2013). Notably, there is some evidence for factor selectivity with respect to these associations. Across studies, negative relationships between psychopathy and amygdala volume are strongest for F1 (Yang 2009; Ermer et al. 2013), while increased striatal (and, to some extent, ventromedial prefrontal) gray matter is relatively selective for F2 (Ermer et al. 2012). Taken together, available data indicate that emotional and interpersonal facets of antisocial behavior are driven by structural deficits within frontolimbic circuitry. This notion is supported, in part, by diffusion tensor imaging (DTI) studies showing lower white matter integrity within tracts that connect amygdala to prefrontal cortex. By contrast, relatively increased gray matter volume within corticostriatal circuits may give rise to impulsivity, aggression, and substance abuse in antisocial offenders. In the following section, we explore potential information-processing consequences of these observed structural associations. FUNCTION
Aversive Emotion Processing Given the profound social and interpersonal dysfunction in antisocial offenders, there is a wealth of data examining emotion processing in these individuals. In particular,
2 5 2 part iv : I maging G enetics and D isease
deficits in the processing of negative facial emotion expression and of aversive non-social stimuli are reasonably consistent findings. Compared to healthy volunteers, antisocial adults and CU+ adolescents showed relatively impaired recognition of fearful, sad, and surprised facial expressions (Blair 1999; Blair and Cipolotti 2000; Marsh and Blair 2008) and fearful and sad vocalizations (Blair et al. 2002; Blair et al. 2005). Brain imaging studies reporting amygdala hypoactivity in response to aversive social stimuli suggest that amygdala dysfunction may underlie this deficit (Lozier et al. 2014; Jones et al. 2009; Contreras-Rodríguez 2014; Marsh and Blair 2008). Psychopaths show reduced functional connectivity between vmPFC and amygdala at rest (Motzkin et al. 2011) and during emotion-processing tasks (Contreras-Rodríguez 2014), suggesting that aberrant prefrontal control of amygdala function may drive amygdala hypoactivity in psychopathy. Of note, some have reported that amygdala hyporeactivity to distress cues is selectively associated with the emotion-interpersonal facet of antisocial behavior, while the antisocial lifestyle facet is linked to enhanced amygdala engagement (Lozier et al. 2014; Coccaro et al. 2011; Hyde et al. 2014). This suggests that distinct clinical manifestations of antisocial behavior, hinging on either the presence or absence of emotional-interpersonal deficits, are the result of distinct—and indeed opposite—forms of amygdala dysregulation.
Reward, Motivation, and Learning Several groups have found evidence for dysfunctional reward processing in antisocial behavior. Buckholtz and colleagues (2010) found that community volunteers with impulsive-antisocial traits showed exaggerated amphetamine-induced striatal dopamine release and enhanced nucleus accumbens activity during the anticipation of monetary rewards (Buckholtz et al. 2010). The relationship between striatal activity and trait variation in antisociality has been replicated in both community and forensic samples (Bjork, Chen, and Hommer 2012; Pujara et al. 2013). Of note, the link between antisocial behavior and reward processing is not limited to the striatum. In a comparison of ADHD youths with and without conduct disorder, Rubia and colleagues (2009) found a selective association between conduct disorder and OFC dysfunction during a reward-pursuit task (Rubia et al. 2009). These alterations in reward response are particularly noteworthy in light of evidence that antisocial youth and adults show evidence of deficient feedback learning. For example, psychopaths show deficits in response reversal in the context of preserved attentional set-shifting (Mitchell et al. 2002),
suggesting dysfunctional stimulus-reinforcement learning (Newman, Patterson, and Kosson 1987; Mitchell et al. 2002; Blair 2004; Budhani and Blair 2005; Blair et al. 2006; Budhani, Richell, and Blair 2006). Youths with antisocial traits exhibit abnormal activity in the ventral medial prefrontal cortex (vmPFC) during punished reversal errors (Finger et al. 2008), and lower OFC and striatal activity in response to rewarded correct responses (Finger et al. 2011).
Selective Attention Studies of attentional dysfunction in antisocial offenders suggest distinct forms of attentional dysfunction in psychopathic versus impulsive-antisocial individuals. Psychopaths and impulsive-antisocial individuals show opposite performance patterns during the attentional blink paradigm, with the former showing reduced distracter interference and the latter showing an exaggerated attentional blink (Wolf et al. 2012). This and other work suggest that while impulsive-antisocial individuals have compromised attentional control, psychopaths are actually better able to focus on goal-directed tasks, but at the expense of goal-peripheral information that would otherwise be used to adaptively guide behavior. Some have argued that this attentional hyperfocus in psychopaths may underlie some of their deficits in affective processing (Newman et al. 2010; Sommers et al. 2012). Consistent with this, psychopaths show less behavioral interference (Mitchell et al. 2006), decreased amygdala activity, and increased prefrontal function (Larson et al. 2013) when exposed to threat-relevant information, but only when threat stimuli are presented after an alternative goal-directed task set has already been engaged.
Theory of Mind and Prosocial Concern Given their profound callousness and lack of empathy, it is perhaps unsurprising that psychopaths show deficits in theory of mind and prosocial concern. In fact, psychopaths and OFC lesion patients exhibit very similar impairments in theory of mind (ToM) tasks. Such deficits are selective for so-called “affective” ToM, which involves representing the emotional experience of others (Shamay-Tsoory et al. 2010). Similarly, psychopaths have reduced vmPFC activity while imagining or viewing others in pain (Decety, Chen, and Harenski 2013). However, while the conventional wisdom is that ToM deficits are selective for psychopathy over impulsive antisociality, recent work belies this neat distinction (Blair 2007). For example, Sebastian and colleagues (2012) found that CU traits in antisocial youths were negatively related to amygdala response during an affective ToM
chapter 1 7 : I maging G enetics of A ntisocial B ehavior and P sychopathy
253
task, but conduct disorder without CU traits was linked to exaggerated amygdala reactivity (Sebastian et al. 2012). Thus, as with aversive emotion processing, psychopaths and impulsive-antisocial individuals appear to have opposite patterns of amygdala dysfunction while representing the emotional states of others: namely, hyporeactivity in the former and hypereactivity in the latter. S EROTONIN AN D THE G EN ETI C A RC HITEC TU RE O F AN TI SO CI AL BEHAV IOR Twin and family studies suggest that genetic factors account for approximately 50%–60% of the variation in “broadband” antisocial behavior in adults (Rhee and Waldman 2002; Moffitt 2005; Ferguson 2010; Gunter, Vaughn, and Philibert 2010; Kendler et al. 2013). However, an ingenious series of twin studies in youth suggests distinct genetic architectures for psychopathic versus impulsive-antisocial facets of antisocial behavior. Viding and colleagues found only modest heritability (0.3) for antisocial behavior in the absence of CU traits. By contrast, heritability is moderate to high for such traits alone (.67), and very high for the combination of antisocial behavior + CU (.81) (Viding et al. 2005). Together, this suggests that psychopathy is under very strong genetic control, with environmental factors making relatively weak contributions to risk. Conversely, genetic contributions to impulsive-antisocial behavior are relatively modest, with environmental influences expected to play a key role in the expression of genetic risk. The high heritability for antisociality in general has spurred a search for associated genotypes. Behavioral genetics work has identified risk-linked variants in a range of neurobiological pathways. However, the most consistent associations, by far, have been to the serotonin (5-HT) system. 5HT CATABOLISM: MAOA
Brunner and colleagues (1993; Brunner, Nelen, et al. 1993) reported the first clear demonstration that a specific genetic variant could account for the intergenerational transmission of antisocial behavior. The gene in question—MAOA—remains the most well-characterized candidate gene for antisociality. The subjects in these studies were a large Dutch family in which a characteristic phenotype—including mild mental retardation, a propensity toward aggressive outbursts, and impulsively violent behavior, including rape, arson, assault, and attempted murder—had been observed in some of the men for many generations. As the syndrome appeared to be transmitted
via an X-linked mode of inheritance, Brunner’s linkage analyses of this family focused on the X chromosome. This work isolated the monoamine oxidase A (MAOA) locus on chromosome Xp11.23–11.4, and subsequent sequencing of the MAOA gene in probands revealed a point mutation (C936T) that leads to a premature stop codon. Consistent with this alteration in sequence, MAOA activity was essentially undetectable in fibroblasts cultured from affected males. MAOA encodes the mitochondrial catabolic enzyme monoamine oxidase A (MAO-A), which catalyzes the oxidative deamination of biogenic amines. Affected males showed marked alterations in monoamine metabolism, including a prominent reduction in urine 5-HIAA levels compared to unaffected males, Thus this X-linked mutation, present in all probands, appeared to produces a functional human MAOA knockout (Brunner et al. 1993; Brunner, Nelen, et al. 1993). MAO-A and MAO-B are the two known isoforms of MAOA; the genes encoding both map to adjacent sites on chromosome Xp11.23 (Grimsby et al. 1991). The monoamine oxidases are localized to the outer mitochondrial membrane in the presynaptic terminal of monoamine projection neurons (MAO-A) (Westlund et al. 1993; Arai et al. 2002) and in astrocytes (MAO-A and MAO-B) (Levitt, Pintar, and Breakefield 1982; Westlund et al. 1988), where they are able to regulate both the amount of intracellular substrate available for release and the degree of extra-synaptic monoamine inactivation. Therefore, functional genetic variation in MAOA is likely to disturb signaling at monoaminergic synapses throughout the brain (Shih and Thompson 1999). Both MAO-A and MAO-B are composed of 15 exons and demonstrate matching exon-intron organization, suggesting their derivation from the same ancestral gene by a duplication event (Grimsby et al. 1991). However, the two isoforms show significant divergence in regional expression and activity patterns (Westlund et al. 1985; Willoughby et al. 1988; Saura et al. 1996a; Saura et al. 1996b; Fowler et al. 1997; Jahng et al. 1997). Other differences between the two isoforms are potentially instructive for understanding the MAOA-aggression linkage. First, while dopamine (DA) is a good substrate for both isoforms of MAO, MAOA has higher affinity for serotonin (5-HT) and norepinephrine (NE). The relative importance of MAOA for regulating 5-HT and NE (i.e., vs. DA) is underscored by the finding that MAOA knockout mice show selective increases in brain 5HT and NE compared to wild-type, while DA is unaffected (Cases et al. 1995). Moreover, the neurochemical, neuromorphological, and behavioral phenotypes produced by the genetic deletion of MAO-B
2 5 4 part iv : I maging G enetics and D isease
are quite distinct from those of MAOA knockout mice. One potential explanation for the relative importance of MAO-A over MAO-B may pertain to the divergent developmental trajectories of MAO-A and MAO-B expression. MAO-A expression precedes that of MAO-B, and is the more critical for regulating monoamine catabolism ante partum. While MAO-A is present at adult levels at birth, MAO-B appears only postnatally and subsequently exhibits a striking increase (Tsang, Ho, and Wen 1986; Strolin Benedetti et al. 1992; Nicotra et al. 2004). While the C936T nonsense mutation characterized by Brunner is of course rare, being a private mutation held only by members of that specific Dutch family, many common polymorphic variants have been identified in the MAOA gene region. The best studied of these, by far, is a variable number of tandem repeats (VNTR) polymorphism located in the promoter region of the gene, 1.2 kilobase pairs upstream of its start site. Originally characterized by Sabol and colleagues, the MAOA upstream VNTR (hereafter, MAOA u-VNTR) is a 30bp repeat that has been shown to impact transcription in an in vitro heterologous expression system and MAOA activity in fibroblast cultures (Sabol and Hamer 1998). The presence of 3.5 or 4 repeats is associated with relatively higher MAOA expression and activity (therefore termed MAOA-H or “high” alleles), while the presence of 3 repeats results in relatively lower expression and activity (likewise termed an MAOA-L or “low” alleles) (Daruna and Kent 1976; Garpenstrand et al. 2001; Syagailo et al. 2001; Mertins et al. 2011). However, contradictory findings do exist, and the functional significance of the 2 and 5 repeats is still somewhat controversial (Balciuniene et al. 2002; Cirulli and Goldstein 2007). MAOA: ANIMAL WORK
Murine genetic deletion studies generally recapitulate the findings of Brunner and colleagues, with MAOA knockouts showing dramatically heightened reactive aggression, poor fear learning, and increased forebrain 5-HT and NE levels (Cases et al. 1995; Popova et al. 2001; Scott et al. 2008; Godar et al. 2010). Interestingly, a naturally occurring nonsense mutation (MAOAA863T) has also been identified in mice, providing a closer parallel to the genetic anomaly described by Brunner and colleagues. Mutant allele carriers exhibit high levels of reactive aggression, including increased fighting behaviors, amplified tail rattling, and reduced attack latency (Scott et al. 2008). Intriguingly, these animals also display aberrant exploratory behaviors: while they are slower to investigate objects and familiar environments,
they show much quicker unconditioned escape and fear responses to discrete threat stimuli and threatening contexts (Godar et al. 2010). Significant visual, locomotor, or olfactory abnormalities do not accompany these deficits, implying a selective impairment in emotional reactivity and risk assessment (Godar et al. 2010). Echoing the work outlined earlier, molecular and behavioral pharmacology studies suggest that the relationship between MAOA and aggression is specifically mediated by genetically determined dysregulation in serotonin signaling pathways. For example, murine MAOA knockouts show a significant increase in serotonin that is particularly dramatic early in life, with mutant pups showing a 9-fold increase compared to wild-type. Changes in DA and NE levels, while apparent, are much smaller in magnitude (Cases et al. 1995). Moreover, blunting post-synaptic serotonin signaling via the 5HT2A antagonist ketanserin reduces aggressive behavior in these animals (Shih and Chen 1999). Neuromorphological findings in MAOA knockouts also implicate excess serotonin signaling during development as a potential pathomechanism. MAOA-deficient mice show cytoarchitectonic changes in sensory cortex, which are replicated by pharmacological attenuation of MAO-A by clorgyline. This pharmacologically induced neuromorphological phenocopy is developmentally specific; clorgyline administration replicates the knockout phenotype only when administered during a critical developmental window. Changes in neuronal morphology (e.g., dendritic size, length, branching) have also been observed in the orbitofrontal cortex and basolateral amygdala of MAOA knockout mice (Bortolato et al. 2011). Finally, pharmacological depletion of serotonin, but not of catecholamines, rescues the phenotype in MAOA-deficient animals. Of note, this rescue is only observed during a specific developmental window (Cases et al. 1996), which is consistent with other work showing developmentally specific effects of aberrant serotonin signaling in producing adult affective illness (Gingrich et al. 2003; Ansorge et al. 2004; Gross and Hen 2004; Ansorge, Hen, and Gingrich 2007). Further, the observation of sensitive periods during development suggests that there may be critical windows during youth and early adolescence for mitigating the impact of genetic risk (de Boer et al 2009; Tost et al. 2010). MAOA: HUMAN WORK
Human behavioral genetic studies provide qualified support for the idea that lowered MAOA expression (e.g., in MAOA-L carriers) predisposes antisocial behavior. Impulsive-antisocial symptoms have been linked to reduced
chapter 1 7 : I maging G enetics of A ntisocial B ehavior and P sychopathy
255
MAOA activity in platelets and brain (Davis et al. 1983; Perris et al. 1984; cf Klinteberg et al. 1987; Mukasa et al. 1990; Belfrage, Lidberg, and Oreland 1992; Prochazka et al. 2003; Skondras et al. 2004; Eklund, Alm, and af Klinteberg 2005; Alia-Klein et al. 2008), and numerous associations with MAOA-L have been reported for substance abuse (Contini et al. 2006; Saito et al. 2002; Parsian et al. 2003), antisocial personality disorder, and other clinical diagnoses with antisocial features (Williams et al. 2009; Mertins et al. 2011; Reti et al. 2011). Associations with quantitative behavioral phenotypes linked to antisocial behavior have also been reported. For example, Gallardo-Pujol and colleagues (2012) found that MAOA-L allele carriers were substantially more aggressive in the point subtraction aggression paradigm than their MAOA-H counterparts, an effect that was magnified by provocations such as social exclusion and aversive acoustical stimuli (Gallardo-Pujol, Andrés-Pueyo, and Maydeu-Olivares 2012; Kuepper et al. 2013). Similarly, MAOA-L participants were significantly more likely than high-expressing allele carriers to administer physical punishment (hot sauce) to an opponent when they believed that 80% of their earnings had been taken in a power-to-take task (McDermott et al. 2009). Low MAO-A expression has also been linked to decreased cooperation in economic tasks, with MAOA-L males making lower contributions to community funds in public goods games (Mertins et al. 2011). Despite the positive findings reported earlier, support for MAOA-L as a causal pathogenic variant is, on the whole, mixed. For example, some groups have reported enhanced impulsive-antisocial behavior (Beitchman et al. 2004), elevated risk for Cluster B personality disorders ( Jacob et al. 2005), lower CSF 5-HIAA levels ( Jönsson et al. 2000), and blunted prolactin-fenfluramine (PRL-fen) response in MAOA-H, rather than MAOA-L allele carriers (Beitchman et al. 2004; Manuck et al. 2000; Manuck et al. 2002; Doudet et al. 1995). Associations with the MAOA-H allele are particularly common in female samples, and in the presence of specific environmental factors (e.g., prenatal nicotine exposure; see discussion later in this chapter) (Sjöberg et al. 2008; Wakschlag et al. 2010). 5HT CLEARANCE: SERT
The human serotonin transporter gene (SLC6A4; SERT) also contains a functional promoter VNTR (often termed the 5HTTLPR) that impacts transporter expression and has been linked to adverse psychiatric outcomes. Possessing 14 copies of a 22bp repeat cassette (the 5HTTLPR “short” allele) is associated with lower transporter availability and increased risk for negative emotionality phenotypes
compared to the “long” allele (16 repeats), particularly following exposure to childhood adversity (Holmes and Hariri 2003; Caspi et al 2005; Canli and Lesch 2007; Karg et al. 2011). While the study of this variant has most commonly occurred in the context of anxiety and depression, a parallel literature has arisen showing associations with aggressive behavior and related traits. Consistent with prior data linking impulsive aggression to reduced SERT availability in platelets (Coccaro et al. 1996; Coccaro, Lee, and Kavoussi 2010; Marazziti et al. 2010) and in brain (Lindström et al. 2004; Frankle et al. 2005; Sekine et al. 2006), individuals carrying the low-expressing short allele are at significantly increased risk for antisocial behavior. The short allele has been tied to impulsive violence in alcoholics and suicide attempters (Hallikainen et al. 1999; Baca-García et al. 2005); to recurrent physical violence in forensic inpatients (Retz et al. 2004) and incarcerated criminals (Liao et al. 2004); to antisocial personality disorder diagnosis (Lyons-Ruth et al. 2007); to increased impulsivity on performance-based measures (Paaver et al. 2007; Walderhaug et al. 2010); and to enhanced antisocial, novelty seeking, and hostility traits (Gerra et al. 2005; Gonda et al. 2009) in community-based samples of adults. In addition, short allele genotype predicts higher levels of aggressive behavior in children (Beitchman et al. 2006; Haberstick, Smolen, and Hewitt 2006) and increased aggressive behavior and impulsive-antisocial traits in adolescents (Sakai et al. 2006; Fowler et al. 2009; Sadeh, Javdani, and Jackson 2010). 5HT RELEASE: HTR1B
HTR1b, encoding the 5HT terminal autoreceptor 5HT1b, was initially proposed as an aggression susceptibility gene by an informative series of rodent knockout and behavioral pharmacology studies. Genetic deletion of the 5HT1b receptor in mice produces a dramatically aggressive phenotype (Saudou et al. 1994), and these mice also show profound deficits in impulse control (Bouwknecht et al. 2001). Consistent with the apparent pro-aggression effects of reducing 5HT1b function, 5HT1b agonists decrease offensive (Bell, Donaldson, and Gracey 1995) and frustration-induced (de Almeida 2002) aggression in rats, and attenuate anabolic steroid-induced aggression in hamsters (Grimes and Melloni 2005). Further, there appears to be some degree of regional specificity with respect to this action: de Almeida and colleagues (2006) found that micro-injections of a 5HT1b agonist in ventral orbitofrontal cortex—but not more lateral aspects of PFC—dose-dependently decreased aggressive responding in mice. Similar anti-aggressive effects were found
2 5 6 part iv : I maging G enetics and D isease
following micro-injections of a 5HT1b agonist into 5HTergic neurons of the dorsal raphe (Bannai et al. 2007), and—intriguingly, given the linkage between 5HT and type II alcoholism—these injections reduced escalated aggression seen after alcohol administration (Faccidomo, Bannai, and Miczek 2008). Of note, this receptor appears to play a role in impulsive choice behavior, as the 5HT1a/b agonist eltoprazine decreases impulsive responding in a delay aversion paradigm in rats (van den Bergh, Bloemarts, and Groenink 2006). Human genetic association data also support a role for HTR1b variation in predisposing antisociality. An early study reported that a synonymous coding variant in HTR1b (G861C) was associated with increased risk for antisocial behavior in alcoholics (Lappalainen et al. 1998), and the same allele linked to antisocial alcoholism in that study (861C) was subsequently found to be enriched in individuals with substance abuse and major depression (Huang et al. 2003). More recently, work by Jensen and colleagues (2009) identified a common polymorphic variant (rs13212041) in the 3' UTR of HTR1b that impacts microRNA-mediated repression of HTR1b expression. In that study, the A allele at that SNP was associated with relatively greater repression of HTR1b by miR-96. Critically, the authors found that this same allele predicted higher levels of antisocial behavior in young adults, and this association was selective for aggressive antisocial behaviors compared to non-aggressive antisocial behaviors (e.g., rule-breaking, cheating). Similar associations were subsequently reported to anger and hostility traits in rs13212041-A–carrying male community volunteers (Conner et al. 2010). Together, the extant human data suggest that genetic risk for aggression attributable to HTR1b variation is linked to genetically mediated decreases in HTR1b expression. Since lower HTR1b expression would lead to diminished autoreceptor control over 5HT release, the human genetic findings described here support the notion that dysregulated (enhanced) 5HT transmission predisposes antisocial behavior. 5HT SYNTHESIS: TPH2
5HT is synthesized from the amino acid L-tryptophan in a two-step process that commences with the hydroxylation of L-tryptophan to 5-hydroxytryptophan (the rate-limiting step in this process) by the enzyme tryptophan hydroxylase (TPH). Given the importance of TPH in regulating 5HT availability, a number of studies have examined the impact of genetic variation in TPH activity on antisocial phenotypes. Early rodent work found that TPH activity was higher in aggressive strains of mice, suggesting
that genetic variability in 5HT synthesis contributes to antisocial behavior (Popova 2006). Early human genetic association studies appeared to support this hypothesis. For example, an intron 7 SNP (A218C) in the TPH gene was linked, in a directionally consistent fashion, to lower 5-HIAA ( Jönsson et al. 1997), a blunted PRL-fen response (Manuck et al. 1999), increased impulsivity and criminality (New et al. 1998), and heightened levels of trait anger and aggression (Manuck et al. 1999; Rujescu et al. 2002; Hennig et al. 2005). Subsequently, however, a brain-specific TPH isoform (TPH2) was discovered (Walther and Bader 2003), encoded by a different gene (TPH2) from the one on which the A218C SNP resides. While this complicates interpretation of the earlier findings, Zhang and colleagues (2004) identified an SNP (C1473G, coding for a proline → arginine substitution) in the murine tph2 gene that selectively affects 5HT synthesis in brain. The arginine variant, which leads to lower 5HT synthesis and reduced midbrain TPH2 activity (Osipova, Kulikov, and Popova 2009), is associated with reduced aggression in mice (Kulikov et al. 2005; Osipova, Kulikov, and Popova 2009). In addition, a common human TPH2 haplotype is enriched in patients with borderline personality disorder and predicts antisocial behaviors and traits in these individuals (Perez-Rodriguez et al. 2010). These findings are suggestive of a role for TPH2 variation in human antisociality. However, unlike the MAOA and SERT VNTRs described earlier, little is known regarding the functional molecular effects of this putative risk haplotype. Moreover, few studies to date have directly examined associations between TPH2 variation and antisociality-related phenotypes in humans. GE NE × E NV IRONME NT INT E R AC T ION S Antisociality is genetically complex, with many small-effect variants contributing to risk in interaction with each other and with environmental factors. Childhood maltreatment in particular appears to be an important environmental moderator of genetic risk. In their landmark 2002 study, Caspi and colleagues found that early life maltreatment significantly strengthened the relationship between MAOA genotype and adult antisocial behavior. Specifically, MAOA-L men who were previously exposed to early life abuse exhibited significantly higher levels of violent behavior compared to MAOA-H men (Caspi et al. 2002). These data have been consistently replicated in both men and women (Foley et al. 2004; Huang et al. 2004; Ducci et al. 2006; Kim-Cohen et al. 2006; Nilsson et al. 2006; Widom and Brzustowicz 2006; Frazzetto et al. 2007; Reif et al.
chapter 1 7 : I maging G enetics of A ntisocial B ehavior and P sychopathy
257
2007; Ducci et al. 2008; Kinnally et al. 2009; Derringer et al. 2010; Edwards et al. 2010; Enoch et al. 2010; Fergusson et al. 2011; Haberstick et al. 2013; Weder et al. 2013). Similarly, childhood maltreatment has been found to potentiate the effect of 5HTTLPR short allele status on risk for adult antisocial behavior in both humans (Li and Lee 2010) and in non-human primates (Barr et al. 2003; Schwandt et al. 2010). NEURA L M ECHAN I SMS O F G EN ETI C R IS K : IM AGIN G G EN ETI CS Brain imaging phenotypes have proven especially useful for uncovering the systems-level mechanisms through which genetic factors exert their effects on risk for psychopathology (Meyer-Lindenberg and Weinberger 2006; Buckholtz and Meyer-Lindenberg 2008). Work to date suggests that known genetic susceptibility factors predispose impulsive-antisocial behavior by dysregulating the structure and function of corticolimbic circuitry for emotional arousal and regulation, and of frontoparietal circuitry for executive cognition. Early work in this area found decreased activity in anterior cingulate, ventrolateral prefrontal cortex, and posterior parietal cortex during a combined conflict-resolution and response inhibition in MAOA-L subjects (Fan et al. 2003; Passamonti et al. 2008). This is consistent with most recent work demonstrating reduced task-based and resting-state functional connectivity in MAOA-L individuals within fronto-cingulate networks for executive control (Ziermans et al. 2012; Clemens et al. 2014). In one of the most comprehensive assays of the systems-level consequences of MAOA genetic variation, Meyer-Lindenberg et al. (2006) used a battery of tasks to measure the impact of MAOA genetic variation on affective responsiveness, emotional memory, and impulse control. Across these paradigms, MAOA-L carriers showed a pattern of enhanced subcortical (amygdala, hippocampus, insula) activation in the context of reduced cortical (ACC, orbitofrontal cortex) recruitment (Meyer-Lindenberg et al. 2006), as well as decreased gray matter volume throughout the cingulate cortex and medial temporal lobe. These findings have since been independently replicated (for cingulate and orbitofrontal cortex) (Kalin et al. 2008; Roiser et al. 2009; Cerasa et al. 2010; Alia-Klein et al. 2011). A wealth of preclinical data suggests that prefrontal cortical regions exert a negative regulatory influence over amygdala (Salichon et al. 2001; Rebsam, Seif, and Gaspar 2002), and clinical work suggests that this negative regulation is impaired in illness (Coccaro et al. 2011). Buckholtz
and colleagues tested the hypothesis that risk-linked MAOA genetic variation predisposes antisocial behavior by disrupting corticolimbic circuit function. Using functional connectivity analyses in a sample of healthy volunteers, they reported aberrant functional coupling between amygdala and vmPFC in MAOA-L hemizygotes. The degree of functional coupling was linked to variation in trait negative affect (Buckholtz et al. 2007). Of note, connections between amygdala and vmPFC are sparse (Ongür and Price 2000; Ghashghaei and Barbas 2002; Ghashghaei, Hilgetag, and Barbas 2007), suggesting the involvement of another region with strong anatomical connections to both; subsequent analyses identified perigenual anterior cingulate as playing this mediating role. These findings, which have been replicated by others (Dannlowski et al. 2009), support the hypothesis that MAOA variation disrupts the functional coupling of amygdala with vmPFC to induce emotion regulation deficits similar to that seen in impulsive-antisocial behavior. The impact of 5HTTLPR genetic variation appears to have a similar impact on this circuitry. Structurally, short allele carriers show reduced gray matter volume in amygdala and cingulate (Pezawas et al. 2005; Frodl et al. 2008a; Pezawas et al. 2008) and additionally in hippocampus (Frodl et al. 2008a; Frodl et al. 2008b). Analogous structural findings have also been observed in rhesus monkeys, with carriers of an orthologous rh5-HTTLPR short allele demonstrating morphological deficits in amygdala, cingulate, and hippocampus ( Jedema et al. 2010). There is a wealth of data showing that the short allele carriers have enhanced amygdala function. This effect has been demonstrated during aversive face emotion processing (Hariri et al. 2002; Hariri et al. 2005; Pezawas et al. 2005; Canli et al. 2006; Munafò, Brown, and Hariri 2008; Hagen et al. 2011), at rest (Canli et al. 2005; Rao et al. 2007), during negative mood induction (Fortier et al. 2010), in response to threat in non-human primates (Kalin et al. 2008), and during economic decision-making (Roiser et al. 2009). In addition, short allele carriers show alterations in cingulate function during performance monitoring (Holmes, Bogdan, and Pizzagalli 2010) and response inhibition, mirroring the effect of the MAOA-L allele on cingulate activity during cognitive control. Short allele carriers also show enhanced amygdalavmPFC coupling (Heinz et al. 2005; Friedel et al. 2009) in the context of reduced amygdala-perigenual cingulate connectivity (Pezawas et al. 2005; Lemogne et al. 2011), which is in turn negatively correlated with a trait measure of negative emotionality. Alterations in cortical regulation of subcortical reactivity in short allele carriers may be due to changes in structural connectivity: short allele carriers
2 5 8 part iv : I maging G enetics and D isease
show relatively degraded white matter architecture in the uncinate fasciculus, a fiber tract that connects medial prefrontal cortex and amygdala (Pacheco et al. 2009). C ONCL US IO N Antisocial behavior is heterogeneous at the level of phenomenology, with the presence or absence of emotional/interpersonal deficits marking distinct subtypes. Importantly, this distinction is reflected at the level of systems neurobiology, as each subtype appears to have characteristic (and indeed, opposite) brain imaging phenotypes. Differential heritability for each suggests that their genetic architectures are dissociable as well. Imaging genomics work to date has yielded important mechanistic insights into the nature of this dissociation. On the whole, extant data on two of the best-studied candidate genes for antisocial behavior suggest that risk variants disrupt cortical regulation of amygdala reactivity. Moreover, imaging phenotypes help to clarify the nature of these associations. For both MAOA and 5HTTLPR, variants with prior associations to antisociality are linked to enhanced amygdala engagement during aversive emotion processing. This raises the possibility that variation in 5HT signaling pathway genes selectively contributes to the heritability of impulsive antisociality, but not psychopathy. It may well be the case that variation in other signaling systems with suggestive links to antisocial behavior—such as the neuropeptides oxytocin and vasopressin—independently drive risk for antisocial behavior by predisposing the emotional and interpersonal deficits that are specific to psychopathy. R EF ERENCES af Klinteberg B, Schalling D, Edman G, Oreland L, Asberg M. (1987). Personality correlates of platelet monoamine oxidase (MAO) activity in female and male subjects. Neuropsychobiology. 18: 89–96. Alia-Klein N, et al. (2008). Brain monoamine oxidase A activity predicts trait aggression. J Neurosci. 28: 5099–5104. Alia-Klein N, et al. (2011). Gene x disease interaction on orbitofrontal gray matter in cocaine addiction. Arch Gen Psychiatry. 68: 283–294. American Psychiatric Association (APA). (2000). Diagnostic and statistical manual of mental disorders (DSM-IV-TR, 4th ed., text revision). Washington, DC: APA. Ansorge MS, Hen R, Gingrich JA. (2007). Neurodevelopmental origins of depressive disorders. Curr Opin Pharmacol. 7: 8–17. Ansorge MS, Zhou M, Lira A, Hen R, Gingrich JA. (2004). Early-life blockade of the 5-HT transporter alters emotional behavior in adult mice. Science. 306: 879–881. Arai R, et al. (2002). Differential subcellular location of mitochondria in rat serotonergic neurons depends on the presence and the absence of monoamine oxidase type B. Neuroscience. 114: 825–835. Baca-García E, et al. (2005). A pilot genetic study of the continuum between compulsivity and impulsivity in females: the serotonin
transporter promoter polymorphism. Prog. Neuropsychopharmacol. Biol. Psychiatry 29: 713–717. Balciuniene J, Emilsson L, Oreland L, Pettersson U, Jazin E. (2002). Investigation of the functional effect of monoamine oxidase polymorphisms in human brain. Hum. Genet. 110: 1–7. Bannai M, Fish EW, Faccidomo S, Miczek KA. (2007). Anti-aggressive effects of agonists at 5-HT1B receptors in the dorsal raphe nucleus of mice. Psychopharmacology. 193: 295–304. Barr CS, et al. (2003). The utility of the non-human primate model for studying gene by environment interactions in behavioral research. Genes Brain Behav. 2: 336–340. Baskin Sommers A, Curtin JJ, Li W, Newman JP. (2012). Psychopathyrelated differences in selective attention are captured by an early event-related potential. Pers Disord. 3: 370–378. Beitchman JH, Mik HM, Ehtesham S, Douglas L, Kennedy JL. (2004). MAOA and persistent, pervasive childhood aggression. Mol Psychiatry. 9: 546–547. Beitchman JH, et al. (2006). Serotonin transporter polymorphisms and persistent, pervasive childhood aggression. Am J Psychiatry. 163: 1103–1105. Belfrage H, Lidberg L, Oreland L. (1992). Platelet monoamine oxidase activity in mentally disordered violent offenders. Acta Psychiatr Scand. 85: 218–221. Bell R, Donaldson C, Gracey D. (1995). Differential effects of CGS 12066B and CP-94,253 on murine social and agonistic behaviour. Pharmacol. Biochem. Behav. 52: 7–16. Blair KS, Morton J, Leonard A, Blair RJR. (2006). Impaired decision-making on the basis of both reward and punishment information in individuals with psychopathy. Pers Indiv Differ. 41: 155–165. Blair R. (1999). Responsiveness to distress cues in the child with psychopathic tendencies. Pers Indiv Differ. 27: 135–145. Blair R. (2007). Empathic dysfunction in psychopathic individuals. Empathy in mental illness. 3–16. Blair RJR. (2004). The roles of orbital frontal cortex in the modulation of antisocial behavior. Brain Cogn. 55: 198–208. Blair RJR, Budhani S, Colledge E, Scott S. (2005). Deafness to fear in boys with psychopathic tendencies. J Child Psychol & Psychiat. 46: 327–336. Blair RJR, Cipolotti L. (2000). Impaired social response reversal: a case of ‘acquired sociopathy.’ Brain. 123(Pt 6): 1122–1141. Blair RJR, et al. (2002). Turning a deaf ear to fear: impaired recognition of vocal affect in psychopathic individuals. J Abnorm Psychol. 111: 682–686. Bjork JM, Chen G, Hommer DW. (2012). Psychopathic tendencies and mesolimbic recruitment by cues for instrumental and passively obtained rewards. Biol Psychol. 89: 408–415. Bortolato M, et al. (2011). Social deficits and perseverative behaviors, but not overt aggression, in MAO-A hypomorphic mice. Neuropsychopharmacology. 36: 2674–2688. Bouwknecht JA, et al. (2001). Absence of 5-HT(1B) receptors is associated with impaired impulse control in male 5-HT(1B) knockout mice. BPS. 49: 557–568. Brunner HG, Nelen M, Breakefield XO, Ropers HH, van Oost BA. (1993). Abnormal behavior associated with a point mutation in the structural gene for monoamine oxidase A. Science. 262: 578–580. Brunner HG, et al. (1993). X-linked borderline mental retardation with prominent behavioral disturbance: phenotype, genetic localization, and evidence for disturbed monoamine metabolism. Am J Hum Genet. 52: 1032–1039. Buckholtz JW, Meyer-Lindenberg A. (2008). MAOA and the neurogenetic architecture of human aggression. Trends in Neurosciences 31: 120–129. Buckholtz JW, et al. (2007). Genetic variation in MAOA modulates ventromedial prefrontal circuitry mediating individual differences in human personality. Mol Psychiatry, 13: 313–324. Dannlowski U, et al. (2009). Reduced amygdala-prefrontal coupling in major depression: association with MAOA genotype and illness severity. Int. J. Neuropsychopharmacol. 12: 11–22.
chapter 1 7 : I maging G enetics of A ntisocial B ehavior and P sychopathy
259
Buckholtz JW, et al. (2010). Mesolimbic dopamine reward system hypersensitivity in individuals with psychopathic traits. Nat Neurosci. 13: 419–421. Budhani S, Blair R. (2005). Response reversal and children with psychopathic tendencies: success is a function of salience of contingency change. J Child Psychol. 46: 972–981. Budhani S, Richell RA, Blair RJR. (2006). Impaired reversal but intact acquisition: Probabilistic response reversal deficits in adult individuals with psychopathy. J Abnorm Psychol 115: 552–558. Canli T, Lesch KP. (2007). Long story short: the serotonin transporter in emotion regulation and social cognition. Nat Neurosci. 10: 1103–1109. Canli T, et al. (2005). Beyond affect: a role for genetic variation of the serotonin transporter in neural activation during a cognitive attention task. Proc Natl Acad Sci U S A. 102: 12224–12229. Canli T, et al. (2006). Neural correlates of epigenesis. Proc Natl Acad Sci U S A. 103: 16033–16038. Cases O, et al. (1995). Aggressive behavior and altered amounts of brain serotonin and norepinephrine in mice lacking MAOA. Science. 268: 1763–1766. Cases O, et al. (1996). Lack of barrels in the somatosensory cortex of monoamine oxidase A-deficient mice: role of a serotonin excess during the critical period. Neuron. 16: 297–307. Caspi A, et al. (2005). Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Focus. 3: 156–160. Caspi A, et al. (2002). Role of genotype in the cycle of violence in maltreated children. Science 297: 851–854. Cerasa A, et al. (2010). Morphological correlates of MAO A VNTR polymorphism: new evidence from cortical thickness measurement. Behav Brain Res. 211: 118–124. Cirulli ET, Goldstein DB. (2007). In vitro assays fail to predict in vivo effects of regulatory polymorphisms. Hum Mol Genet. 16: 1931–1939. Clemens B, et al. (2014). Effect of MAOA genotype on resting-state networks in healthy participants. Cereb Cortex. 25: 1771–1781. Coccaro EF, Kavoussi RJ, Sheline YI, Lish JD, Csernansky JG. (1996). Impulsive aggression in personality disorder correlates with tritiated paroxetine binding in the platelet. Arch Gen Psychiatry. 53: 531–536. Coccaro EF, Lee R, Kavoussi RJ. (2010). Inverse relationship between numbers of 5-HT transporter binding sites and life history of aggression and intermittent explosive disorder. J Psychiatr Res. 44: 137–142. Coccaro EF, Sripada CS, Yanowitch RN, Phan KL. (2011). Corticolimbic function in impulsive aggressive behavior. Biol. Psychiatry 69: 1153–1159. Conner TS, et al. (2010). Functional polymorphisms in the serotonin 1B receptor gene (HTR1B) predict self-reported anger and hostility among young men. Am J Med Genet B Neuropsychiatr Genet. 153B: 67–78. Contini V, Marques FZC, Garcia CED, Hutz MH, Bau CHD. (2006). MAOA-uVNTR polymorphism in a Brazilian sample: further support for the association with impulsive behaviors and alcohol dependence. Am J Med Genet B Neuropsychiatr Genet. 141B: 305–308. Contreras-Rodríguez, O. et al. (2014). Disrupted neural processing of emotional faces in psychopathy. Social Cogn. 9: 505–512. Daruna JH, Kent EW. (1976). Comparison of regional serotonin levels and turnover in the brain of naturally high and low aggressive rats. Brain Res. 101: 489–501. Davis BA, Yu PH, Boulton AA, Wormith JS, Addington D. (1983). Correlative relationship between biochemical activity and aggressive behaviour. Prog Neuropsychopharmacol Biol Psychiatry. 7: 529–535. de Almeida R. (2002). Aggression escalated by social instigation or by discontinuation of reinforcement (‘frustration’) in mice inhibition by anpirtoline: a 5-HT1B Receptor agonist. Neuropsychopharmacology 27: 171–181. de Almeida RMM, et al. (2006). 5-HT(1B) receptors, ventral orbitofrontal cortex, and aggressive behavior in mice. Psychopharmacology. 185: 441–450.
de Boer SF, Caramaschi D, Natarajan D, Koolhaas JM. (2009). The vicious cycle towards violence: focus on the negative feedback mechanisms of brain serotonin neurotransmission. Front Behav Neurosci. 3: 52. Decety J, Chen C, Harenski C. (2013). An fMRI study of affective perspective taking in individuals with psychopathy: imagining another in pain does not evoke empathy. Front. Hum. Neurosci. 7: 489. Derringer J, Krueger RF, Irons DE, Iacono WG. (2010). Harsh discipline, childhood sexual assault, and MAOA genotype: an investigation of main and interactive effects on diverse clinical externalizing outcomes. Behav Genet. 40: 639–648. Doudet D, et al. (1995). Cerebral glucose metabolism, CSF 5-HIAA levels, and aggressive behavior in rhesus monkeys. Am J Psychiatry 152: 1782–1787. Ducci F, et al. (2008). Interaction between a functional MAOA locus and childhood sexual abuse predicts alcoholism and antisocial personality disorder in adult women. Mol Psychiatry. 13: 334–347. Ducci F, et al. (2009). A functional polymorphism in the MAOA gene promoter (MAOA-LPR) predicts central dopamine function and body mass index. Mol Psychiatry. 11: 858–866. Edwards AC, et al. (2010). MAOA-uVNTR and early physical discipline interact to influence delinquent behavior. J Child Psychol Psychiatry. 51: 679–687. Eklund J, Alm PO, af Klinteberg B. (2005). Monoamine oxidase activity and tri-iodothyronine level in violent offenders with early behavioural problems. Neuropsychobiology. 52: 122–129. Enoch MA, Steer CD, Newman TK, Gibson N, Goldman D. (2010). Early life stress, MAOA, and gene-environment interactions predict behavioral disinhibition in children. Genes Brain Behav. 9: 65–74. Ermer E, Cope LM, Nyalakanti PK, Calhoun VD, Kiehl KA. (2013). Aberrant paralimbic gray matter in incarcerated male adolescents with psychopathic traits. J Am Acad Child Adolesc Psychiatry. 52: 94–103.e3. Ermer E, Cope LM, Nyalakanti PK, Calhoun VD, Kiehl KA. (2012). Aberrant paralimbic gray matter in criminal psychopathy. J Abnorm Psychol. 121: 649–658. Faccidomo S, Bannai M, Miczek KA. (2008). Escalated aggression after alcohol drinking in male mice: dorsal raphe and prefrontal cortex serotonin and 5-HT1B receptors. Neuropsychopharmacology. 33: 2888–2889. Fan J, Fossella J, Sommer T, Wu Y, Posner MI. (2003). Mapping the genetic variation of executive attention onto brain activity. Proc Natl Acad Sci U S A. 100: 7406–7411. Ferguson CJ. (2010). Genetic contributions to antisocial personality and behavior: a meta-analytic review from an evolutionary perspective. J Soc Psychol 150: 160–180. Fergusson DM, Boden JM, Horwood LJ, Miller AL, Kennedy MA. (2011). MAOA, abuse exposure and antisocial behaviour: 30-year longitudinal study. Br J Psychiatry. 198: 457–463. Finger EC, et al. (2008). Abnormal ventromedial prefrontal cortex function in children with psychopathic traits during reversal learning. Arch Gen Psychiatry. 65: 586–594. Finger EC, et al. (2011). Disrupted reinforcement signaling in the orbitofrontal cortex and caudate in youths with conduct disorder or oppositional defiant disorder and a high level of psychopathic traits. Am J Psychiatry. 168: 152–162. Foley DL, et al. (2004). Childhood adversity, monoamine oxidase a genotype, and risk for conduct disorder. Arch Gen Psychiatry. 61: 738–744. Fortier E, et al. (2010). Early impact of 5-HTTLPR polymorphism on the neural correlates of sadness. Neurosci Lett. 485: 261–265. Fowler JS, et al. (1997). Age-related increases in brain monoamine oxidase B in living healthy human subjects. Neurobiol Aging. 18: 431–435. Fowler T, et al. (2009). Psychopathy trait scores in adolescents with childhood ADHD: the contribution of genotypes affecting MAOA, 5HTT and COMT activity. Psychiatr Genet. 19: 312–319.
2 6 0 part iv : I maging G enetics and D isease
Frankle WG, et al. (2005). Brain serotonin transporter distribution in subjects with impulsive aggressivity: a positron emission study with [11C]McN 5652. Am J Psychiatry. 162: 915–923. Frazzetto G, et al. (2007). Early trauma and increased risk for physical aggression during adulthood: the moderating role of MAOA genotype. PLoS One. 2: e486. Friedel E, et al. (2009). 5-HTT genotype effect on prefrontal-amygdala coupling differs between major depression and controls. Psychopharmacology. 205: 261–271. Frodl T, et al. (2008a). Reduced gray matter brain volumes are associated with variants of the serotonin transporter gene in major depression. Mol. Psychiatry 13: 1093–1101. Frodl T, et al. (2008b). Reduced hippocampal volumes associated with the long variant of the tri- and diallelic serotonin transporter polymorphism in major depression. Am J Med Genet B Neuropsychiatr Genet. 147B: 1003–1007. Gallardo-Pujol D, Andrés-Pueyo A, Maydeu-Olivares A. (2012). MAOA genotype, social exclusion and aggression: an experimental test of a gene-environment interaction. Genes Brain Behav. 12: 140–145. Garpenstrand H, Annas P, Ekblom J, Oreland L, Fredrikson M. (2001). Human fear conditioning is related to dopaminergic and serotonergic biological markers. Behav Neurosci. 115: 358–364. Gerra G, et al. (2005). Serotonin transporter promoter polymorphism genotype is associated with temperament, personality traits and illegal drugs use among adolescents. J Neural Transm 112: 1397–1410. Ghashghaei HT, Barbas H. (2002). Pathways for emotion: interactions of prefrontal and anterior temporal pathways in the amygdala of the rhesus monkey. Neuroscience. 115: 1261–1279. Ghashghaei HT, Hilgetag CC, Barbas H. (2007). Sequence of information processing for emotions based on the anatomic dialogue between prefrontal cortex and amygdala. NeuroImage. 34: 905–923. Gingrich JA, Ansorge MS, Merker R, Weisstaub N, Zhou M. (2003). New lessons from knockout mice: the role of serotonin during development and its possible contribution to the origins of neuropsychiatric disorders. CNS Spectr. 8: 572–577. Glenn AL, Raine A, Yaralian PS, Yang Y. (2010). Increased volume of the striatum in psychopathic individuals. Biol. Psychiatry. 67: 52–58. Godar SC, et al. (2010). Maladaptive defensive behaviours in monoamine oxidase A-deficient mice. Int J Neuropsychopharm. 14: 1195–1207. Gonda X, et al. (2009). Association of the s allele of the 5-HTTLPR with neuroticism-related traits and temperaments in a psychiatrically healthy population. Eur Arch Psychiatry Clin Neurosci. 259: 106–113. Grimes JM, Melloni RH. (2005). Serotonin-1B receptor activity and expression modulate the aggression-stimulating effects of adolescent anabolic steroid exposure in hamsters. Behav Neurosci. 119: 1184–1194. Grimsby J, Chen K, Wang LJ, Lan NC, Shih JC. Human monoamine oxidase A and B genes exhibit identical exon-intron organization. Proc Natl Acad Sci USA. 88: 3637–3641. Gross C, Hen R. (2004). Genetic and environmental factors interact to influence anxiety. Neurotox Res. 6: 493–501. Gunter TD, Vaughn MG, Philibert RA. (2010). Behavioral genetics in antisocial spectrum disorders and psychopathy: a review of the recent literature. Behav Sci Law. 28: 148–173. Haberstick BC, Smolen A, Hewitt JK. (2006). Family-based association test of the 5HTTLPR and aggressive behavior in a general population sample of children. BPS 59: 836–843. Haberstick BC, et al. (2013). MAOA genotype, childhood maltreatment, and their interaction in the etiology of adult antisocial behaviors. Biol Psychiatry. 75: 25–30. Hagen von dem EAH, Passamonti L, Nutland S, Sambrook J, Calder AJ. (2011). The serotonin transporter gene polymorphism and the effect of baseline on amygdala response to emotional faces. Neuropsychologia. 49: 674–680. Hariri AR, et al. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science. 297: 400–403. Hariri AR, et al. (2005). A susceptibility gene for affective disorders and the response of the human amygdala. Arch Gen Psychiatry. 62: 146–152.
Hallikainen T, et al. (1999). Association between low activity serotonin transporter promoter genotype and early onset alcoholism with habitual impulsive violent behavior. Mol Psychiatry. 4: 385–388. Hare RD. (1980). A research scale for the assessment of psychopathy in criminal populations. Pers Indiv Differ. 1: 111–119. Heinz A, et al. (2005). Amygdala-prefrontal coupling depends on a genetic variation of the serotonin transporter. Nat Neurosci. 8: 20–21. Hennig J, Reuter M, Netter P, Burk C, Landt O. (2005). Two types of aggression are differentially related to serotonergic activity and the A779C TPH polymorphism. Behav Neurosci. 119: 16–25. Holmes AJ, Bogdan R, Pizzagalli DA. (2010). Serotonin transporter genotype and action monitoring dysfunction: a possible substrate underlying increased vulnerability to depression. Neuropsychopharmacology. 35: 1186–1197. Holmes A, Hariri AR. (2003). The serotonin transporter gene-linked polymorphism and negative emotionality: placing single gene effects in the context of genetic background and environment. Genes Brain Behav. 2: 332–335. Huang Y-Y, et al. (2003). Substance abuse disorder and major depression are associated with the human 5-HT1B receptor gene (HTR1B) G861C polymorphism. Neuropsychopharmacology 28: 163–169. Huang Y-Y, et al. (2004). An association between a functional polymorphism in the monoamine oxidase a gene promoter, impulsive traits and early abuse experiences. Neuropsychopharmacology. 29: 1498–1505. Hyde LW, Byrd AL, Votruba-Drzal E, Hariri AR, Manuck SB. (2014). Amygdala reactivity and negative emotionality: divergent correlates of antisocial personality and psychopathy traits in a community sample. J Abnorm Psychol. 123: 214–224. Jacob CP, et al. (2005). Cluster B personality disorders are associated with allelic variation of monoamine oxidase A activity. Neuropsychopharmacology. 30: 1711–1718. Jahng JW, et al. (1997). Localization of monoamine oxidase A and B mRNA in the rat brain by in situ hybridization. Synapse. 25: 30–36. Jedema HP, et al. (2010). Cognitive impact of genetic variation of the serotonin transporter in primates is associated with differences in brain morphology rather than serotonin neurotransmission. Mol Psychiatry. 15: 512–22, 446. Jensen KP, et al. (2009). A common polymorphism in serotonin receptor 1B mRNA moderates regulation by miR-96 and associates with aggressive human behaviors. Mol Psychiatry. 14: 381–389. Jones AP, Laurens KR, Herba CM, Barker GJ, Viding E. (2009). Amygdala hypoactivity to fearful faces in boys with conduct problems and callous-unemotional traits. Am J Psychiatry. 166: 95–102. Jönsson EG, et al. (1997). Tryptophan hydroxylase and catechol-Omethyltransferase gene polymorphisms: relationships to monoamine metabolite concentrations in CSF of healthy volunteers. Eur Arch Psychiatry Clin Neurosci. 247: 297–302. Jönsson EG, et al. (2000). A promoter polymorphism in the monoamine oxidase A gene and its relationships to monoamine metabolite concentrations in CSF of healthy volunteers. J Psychiatr Res. 34: 239–244. Kalin NH, et al. (2008). The serotonin transporter genotype is associated with intermediate brain phenotypes that depend on the context of eliciting stressor. Mol Psychiatry. 13: 1021–1027. Karg K, Burmeister M, Shedden K, Sen S. (2011). The serotonin transporter promoter variant (5-HTTLPR), stress, and depression meta-analysis revisited: evidence of genetic moderation. Arch Gen Psychiatry. 68: 444–454. Kendler KS, Patrick CJ, Larsson H, Gardner CO, Lichtenstein P. (2013). Genetic and environmental risk factors in males for self-report externalizing traits in mid-adolescence and criminal behavior through young adulthood. Psychol Med. 43: 2161–2168. Kiehl KA, Hoffman MB. (2011). The criminal psychopath: history, neuroscience, treatment, and economics. Jurimetrics. 51: 355–397. Kim-Cohen J, et al. (2006). MAOA, maltreatment, and gene-environment interaction predicting children’s mental health: new evidence and a meta-analysis. Mol Psychiatry. 11: 903–913.
chapter 1 7 : I maging G enetics of A ntisocial B ehavior and P sychopathy
261
Kinnally EL, et al. (2009). Parental care moderates the influence of MAOA-uVNTR genotype and childhood stressors on trait impulsivity and aggression in adult women. Psychiatr Genet 19: 126–133. Kuepper Y, Grant P, Wielpuetz C, Hennig J. (2013). MAOA-uVNTR genotype predicts interindividual differences in experimental aggressiveness as a function of the degree of provocation. Behav Brain Res. 247: 73–78. Kulikov AV, Osipova DV, Naumenko VS, Popova NK. (2005). Association between Tph2 gene polymorphism, brain tryptophan hydroxylase activity and aggressiveness in mouse strains. Genes Brain Behav. 4: 482–485. Lappalainen J, et al. (1998). Linkage of antisocial alcoholism to the serotonin 5-HT1B receptor gene in 2 populations. Arch Gen Psychiatry. 55: 989–994. Larson CL, et al. (2013). The interplay of attention and emotion: top-down attention modulates amygdala activation in psychopathy. Cogn Affect Behav Neurosci. 13: 757–770. Lemogne C, et al. (2011). Cognitive appraisal and life stress moderate the effects of the 5-HTTLPR polymorphism on amygdala reactivity. Hum Brain Mapp. 32: 1856–1867. Levitt P, Pintar JE, Breakefield XO. (1982). Immunocytochemical demonstration of monoamine oxidase B in brain astrocytes and serotonergic neurons. Proc Natl Acad Sci. U S A. 79: 6385–6389. Li JJ, Lee SS. (2010). Latent class analysis of antisocial behavior: interaction of serotonin transporter genotype and maltreatment. J Abnorm Child Psychol. 38: 789–801. Liao D-L, Hong C-J, Shih H-L, Tsai S-J (2004). Possible association between serotonin transporter promoter region polymorphism and extremely violent crime in Chinese males. Neuropsychobiology. 50: 284–287. Lilienfeld SO, Andrews BP. (1996). Development and preliminary validation of a self-report measure of psychopathic personality traits in noncriminal population. J Pers Assess. 66: 488–524. Lindström MB, et al. (2004). Impulsivity related to brain serotonin transporter binding capacity in suicide attempters. Eur Neuropsychopharmacol. 14: 295–300. Lozier, L. M., Cardinale, E. M., VanMeter, J. W., & Marsh A. A. (2014). Amygdala response to fear mediates the relationship between callous-unemotional traits and proactive aggression among children with conduct problems. JAMA Psychiatry.71: 627–636. Ly M, et al. (2012). Cortical thinning in psychopathy. Am J Psychiatry. 169: 743–749. Lyons-Ruth K, et al. (2007). Serotonin transporter polymorphism and borderline or antisocial traits among low-income young adults. Psychiatr Genet. 17: 339–343. Manuck SB, Flory JD, Ferrell RE, Mann JJ, Muldoon MF. (2000). A regulatory polymorphism of the monoamine oxidase-A gene may be associated with variability in aggression, impulsivity, and central nervous system serotonergic responsivity. Psychiatry Res. 95: 9–23. Manuck SB, Flory JD, Muldoon MF, Ferrell RE. (2002). Central nervous system serotonergic responsivity and aggressive disposition in men. Physiol Behav. 77: 705–709. Manuck SB, et al. (1999). Aggression and anger-related traits associated with a polymorphism of the tryptophan hydroxylase gene. BPS. 45: 603–614. Marazziti D, et al. (2010). Impulsivity, gender, and the platelet serotonin transporter in healthy subjects. NDT. 6: 9–15. Marsh AA, Blair RJR. (2008). Deficits in facial affect recognition among antisocial populations: a meta-analysis. Neurosci Biobehav Rev. 32: 454–465. McDermott R, Tingley D, Cowden J, Frazzetto G, Johnson DDP. (2009). Monoamine oxidase A gene (MAOA) predicts behavioral aggression following provocation. Proc Natl Acad Sci. 106: 2118–2123. Mertins V, Schote AB, Hoffeld W, Griessmair M, Meyer J. (2011). Genetic Susceptibility for individual cooperation preferences: the role of monoamine oxidase A gene (MAOA) in the voluntary provision of public goods. PLoS One. 6: e20959.
Meyer-Lindenberg A, Weinberger DR. (2006). Intermediate phenotypes and genetic mechanisms of psychiatric disorders. Nat Rev Neurosci. 7: 818–827. Meyer-Lindenberg A, et al. (2006). Neural mechanisms of genetic risk for impulsivity and violence in humans. Proc Natl Acad Sci U S A. 103: 6269–6274. Mitchell DGV, Colledge E, Leonard A, Blair RJR. (2002). Risky decisions and response reversal: is there evidence of orbitofrontal cortex dysfunction in psychopathic individuals? Neuropsychologia. 40: 2013–2022. Mitchell DGV, et al. (2006). Instrumental learning and relearning in individuals with psychopathy and in patients with lesions involving the amygdala or orbitofrontal cortex. Neuropsychology 20: 280–289. Moffitt, T. (2005). Genetic and environmental influences on antisocial behaviors: evidence from behavioral–genetic research. Advances in Genetics. 55: 41–104. Motzkin JC, Newman JP, Kiehl KA, Koenigs M. (2011). Reduced Prefrontal Connectivity in Psychopathy. J Neurosci. 31: 17348–17357. Mukasa H, Nakamura J, Yamada S, Inoue M, Nakazawa Y. (1990). Platelet monoamine oxidase activity and personality traits in alcoholics and methamphetamine dependents. Drug Alcohol Depend 26: 251–254. Munafò MR, Brown SM, Hariri AR. (2008). Serotonin transporter (5-HTTLPR) genotype and amygdala activation: a meta-analysis. Biol. Psychiatry. 63: 852–857. New AS, et al. (1998). Tryptophan hydroxylase genotype is associated with impulsive-aggression measures: a preliminary study. Am J Med Genet. 81: 13–17. Newman JP, Curtin JJ, Bertsch JD, Baskin-Sommers AR. (2010). Attention moderates the fearlessness of psychopathic offenders. Biol Psychiatry 67: 66–70. Newman JP, Patterson CM, Kosson DS. (1987). Response perseveration in psychopaths. J Abnorm Psychol. 96: 145–148. Nicotra A, Pierucci F, Parvez H, Senatori O. (2004). Monoamine oxidase expression during development and aging. Neurotoxicology. 25: 155–165. Nilsson KW, et al. (2006). Role of monoamine oxidase A genotype and psychosocial factors in male adolescent criminal activity. Biol Psychiatry. 59: 121–127. Ongür D, Price JL. (2000). The organization of networks within the orbital and medial prefrontal cortex of rats, monkeys and humans. Cereb Cortex. 10: 206–219. Osipova DV, Kulikov AV, Popova NK. (2009). C1473G polymorphism in mouse tph2 gene is linked to tryptophan hydroxylase-2 activity in the brain, intermale aggression, and depressive-like behavior in the forced swim test. J Neurosci Res. 87: 1168–1174. Paaver M, et al. (2007). Platelet MAO activity and the 5-HTT gene promoter polymorphism are associated with impulsivity and cognitive style in visual information processing. Psychopharmacology. 194: 545–554. Pacheco J, et al. (2009). Frontal-limbic white matter pathway associations with the serotonin transporter gene promoter region (5-HTTLPR) polymorphism. J Neurosci. 29: 6229–6233. Parsian A, Cloninger CR, Sinha R, Zhang, ZH. (2003). Functional variation in promoter region of monoamine oxidase A and subtypes of alcoholism: haplotype analysis. Am J Med Genet B Neuropsychiatr Genet. 117B: 46–50. Passamonti L, et al. (2008). Genetically dependent modulation of serotonergic inactivation in the human prefrontal cortex. NeuroImage 40: 1264–1273. Perez-Rodriguez MM, et al. (2010). Tryptophan-hydroxylase 2 haplotype association with borderline personality disorder and aggression in a sample of patients with personality disorders and healthy controls. J Psychiatr Res 44: 1075–1081. Perris C, Eisemann M, von Knorring L, Oreland L, Perris H. (1984). Personality traits and monoamine oxidase activity in platelets in depressed patients. Neuropsychobiology. 12: 201–205.
2 6 2 part iv : I maging G enetics and D isease
Pezawas L, et al. (2005). 5-HTTLPR polymorphism impacts human cingulate-amygdala interactions: a genetic susceptibility mechanism for depression. Nat Neurosci. 8: 828–834. Pezawas L, et al. (2008). Evidence of biologic epistasis between BDNF and SLC6A4 and implications for depression. Mol Psychiatry. 13: 709–716. Pinel, P. (1962). A treatise on insanity (D. Davis, Trans.). New York: Hafner. (Original work published 1801). Popova NK, et al. (2001). Behavioral characteristics of mice with genetic knockout of monoamine oxidase type A. Neurosci Behav Physiol. 31: 597–602. Popova NK. (2006). From genes to aggressive behavior: the role of serotonergic system. Bioessays 28: 495–503. Prochazka H, Anderberg UM, Oreland L, Knorring LV, Agren H. (2003). Self-rated aggression related to serum testosterone and platelet MAO activity in female patients with the fibromyalgia syndrome. World J Biol Psychiatry. 4: 35–41. Pujara M, Motzkin JC, Newman JP, Kiehl KA, Koenigs M. (2013). Neural correlates of reward and loss sensitivity in psychopathy. Soc Cogn Affect Neurosci. 9: 794–801. Rao, H. et al. (2007). Genetic variation in serotonin transporter alters resting brain function in healthy individuals. BPS. 62: 600–606. Rebsam A, Seif I, Gaspar P. (2002). Refinement of thalamocortical arbors and emergence of barrel domains in the primary somatosensory cortex: a study of normal and monoamine oxidase a knock-out mice. J Neurosci. 22: 8541–8552. Reif A, et al. (2007). Nature and nurture predispose to violent behavior: serotonergic genes and adverse childhood environment. Neuropsychopharmacology. 32: 2375–2383. Reti IM, et al. (2011). Monoamine oxidase A regulates antisocial personality in whites with no history of physical abuse. Compr Psychiatry. 52: 188–194. Retz W, Retz-Junginger P, Supprian T, Thome J, Rösler M. (2004). Association of serotonin transporter promoter gene polymorphism with violence: relation with personality disorders, impulsivity, and childhood ADHD psychopathology. Behav Sci Law 22: 415–425. Rhee SH, Waldman ID. (2002). Genetic and environmental influences on antisocial behavior: a meta-analysis of twin and adoption studies. Psychol Bull. 128: 490–529. Roiser JP, et al. (2009). A genetically mediated bias in decision making driven by failure of amygdala control. J Neurosci 29: 5985–5991. Rubia K, et al. (2009). Disorder-specific dissociation of orbitofrontal dysfunction in boys with pure conduct disorder during reward and ventrolateral prefrontal dysfunction in boys with pure ADHD during sustained attention. Am J Psychiatry. 166: 83–94. Rujescu D, et al. (2002). Association of anger-related traits with SNPs in the TPH gene. Mol Psychiatry. 7: 1023–1029. Sabol S, Hamer D. (1998). A functional polymorphism in the monoamine oxidase A gene promoter. Hum Genet. 103: 273–279. Sadeh N, Javdani S, Jackson JJ. (2010). Serotonin transporter gene associations with psychopathic traits in youth vary as a function of socioeconomic resources. J Abnormal Psychol. 119: 604–609. Saito T, et al. (2002). Analysis of monoamine oxidase A (MAOA) promoter polymorphism in Finnish male alcoholics. Psychiatry Res. 109: 113–119. Sakai JT, et al. (2006). Case-control and within-family tests for an association between conduct disorder and 5HTTLPR. Am J Med Genet B Neuropsychiatr Genet. 141: 825–832. Salichon N, et al. (2001). Excessive activation of serotonin (5-HT) 1B receptors disrupts the formation of sensory maps in monoamine oxidase a and 5-ht transporter knock-out mice. J Neurosci 21: 884–896. Saudou F, et al. (1994). Enhanced aggressive behavior in mice lacking 5-HT1B receptor. Science 265: 1875–1878. Saura J, et al. (1996a). Localization of monoamine oxidases in human peripheral tissues. Life Sci. 59: 1341–1349.
Saura J, et al. (1996b). Molecular neuroanatomy of human monoamine oxidases A and B revealed by quantitative enzyme radioautography and in situ hybridization histochemistry. Neuroscience. 70: 755–774. Schiffer B, et al. (2011). Disentangling structural brain alterations associated with violent behavior from those associated with substance use disorders. Arch Gen Psychiatry. 68: 1039–1049. Schwandt ML, et al. (2010). Gene-environment interactions and response to social intrusion in male and female rhesus macaques. Biol Psychiatry. 67: 323–330. Scott AL, Bortolato M, Chen K, Shih JC. (2008). Novel monoamine oxidase A knock out mice with human-like spontaneous mutation. Neuroreport 19: 739–743. Sebastian CL, et al. (2012). Neural responses to affective and cognitive theory of mind in children with conduct problems and varying levels of callous-unemotional traits. Arch Gen Psychiatry 69: 814–822. Sekine Y, et al. (2006). Brain serotonin transporter density and aggression in abstinent methamphetamine abusers. Arch Gen Psychiatry. 63: 90–100. Shamay-Tsoory SG, Harari H, Aharon-Peretz J, Levkovitz Y. (2010). The role of the orbitofrontal cortex in affective theory of mind deficits in criminal offenders with psychopathic tendencies. Cortex 46: 668–677. Shih JC, Chen K. (1999). MAO-A and -B gene knock-out mice exhibit distinctly different behavior. Neurobiology (Bp). 7: 235–246. Shih J, Thompson R. (1999). Monoamine oxidase in neuropsychiatry and behavior. The Am J Human Genet. 65: 593–598. Sjöberg RL, et al. (2008). A non-additive interaction of a functional MAO-A VNTR and testosterone predicts antisocial behavior. Neuropsychopharmacology. 33: 425–430. Skeem JL, Cooke DJ. (2010). Is criminal behavior a central component of psychopathy? Conceptual directions for resolving the debate. Psychol Assess. 22: 433–445. Skondras M, Markianos M, Botsis A, Bistolaki E, Christodoulou G. (2004). Platelet monoamine oxidase activity and psychometric correlates in male violent offenders imprisoned for homicide or other violent acts. Eur Arch Psychiatry Clin Neurosci. 254: 380–386. Strolin Benedetti M, Dostert P, Tipton KF. (1992). Developmental aspects of the monoamine-degrading enzyme monoamine oxidase. Dev Pharmacol Ther. 18: 191–200. Syagailo YV, et al. (2001). Association analysis of the functional monoamine oxidase A gene promoter polymorphism in psychiatric disorders. Am J Med Genet. 105: 168–171. Tost H, et al. (2010). A common allele in the oxytocin receptor gene (OXTR) impacts prosocial temperament and human hypothalamic-limbic structure and function. Proc Natl Acad Sci. 107: 13936–13941. Tsang D, Ho KP, Wen HL. (1986). Ontogenesis of multiple forms of monoamine oxidase in rat brain regions and liver. Dev Neurosci. 8: 243–250. Viding, E., Blair, R. J. R., Moffitt, T. E., & Plomin, R. (2005). Evidence for substantial genetic risk for psychopathy in 7-year-olds. J Child Psychol Psychiatry. 46: 592–597. van den Bergh FS, Bloemarts E, Groenink L. (2006). Delay aversion: effects of 7-OH-DPAT, 5-HT1A/1B-receptor stimulation and d-cycloserine. Pharmacol Biochem Behav. 85: 736–743. Wakschlag LS, et al. (2010). Interaction of prenatal exposure to cigarettes and MAOA genotype in pathways to youth antisocial behavior. Mol Psychiatry. 15: 928–937. Walderhaug E, Herman AI, Magnusson A, Morgan MJ, Landrø NI. (2010). The short (S) allele of the serotonin transporter polymorphism and acute tryptophan depletion both increase impulsivity in men. Neurosci Lett. 473: 208–211. Walther DJ, Bader M. (2003). A unique central tryptophan hydroxylase isoform. Biochem Pharmacol. 66: 1673–1680. Weder N, et al. (2013). MAOA genotype, maltreatment, and aggressive behavior: the changing impact of genotype at varying levels of trauma. Biol Psychiatry. 65: 417–424.
chapter 1 7 : I maging G enetics of A ntisocial B ehavior and P sychopathy
263
Westlund KN, Denney RM, Rose RM, Abell CW. (1988). Localization of distinct monoamine oxidase A and monoamine oxidase B cell populations in human brainstem. Neuroscience. 25: 439–456. Westlund KN, Denney RM, Kochersperger LM, Rose RM, Abell CW. (1985). Distinct monoamine oxidase A and B populations in primate brain. Science. 230: 181–183. Westlund KN, Krakower TJ, Kwan SW, Abell CW. (1993). Intracellular distribution of monoamine oxidase A in selected regions of rat and monkey brain and spinal cord. Brain Res. 612: 221–230. Widom CS, Brzustowicz LM. (2006). MAOA and the ‘cycle of violence:’ childhood abuse and neglect, MAOA genotype, and risk for violent and antisocial behavior. BPS. 60: 684–689. Williams LM, et al. (2009). A polymorphism of the MAOA gene is associated with emotional brain markers and personality traits on an antisocial index. Neuropsychopharmacology. 34: 1797–1809. Willoughby J, et al. (1988). Monoamine oxidase activity and distribution in marmoset brain: implications for MPTP toxicity. Neurosci Lett. 90: 100–106.
Wolf RC, et al. (2012). Reduced susceptibility to the attentional blink in psychopathic offenders: Implications for the attention bottleneck hypothesis. Neuropsychology. 26: 102–109. Yang Y. (2009). Localization of deformations within the amygdala in individuals with psychopathy. Arch Gen Psychiatry. 66: 986. Yang Y, Raine A. (2009). Prefrontal structural and functional brain imaging findings in antisocial, violent, and psychopathic individuals: a meta-analysis. Psychiat Res Neuroimag. 174: 81–88. Yang Y, Raine A, Colletti P, Toga AW, Narr KL. (2010). Morphological alterations in the prefrontal cortex and the amygdala in unsuccessful psychopaths. J Abnorm Psychol. 119: 546–554. Yang Y, et al. (2005). Volume reduction in prefrontal gray matter in unsuccessful criminal psychopaths. Biol Psychiatry. 57: 1103–1108. Zhang X, Beaulieu, J-M, Sotnikova TD, Gainetdinov RR, Caron MG. (2004). Tryptophan hydroxylase-2 controls brain serotonin synthesis. Science 305: 217. Ziermans T, et al. (2012). Working memory brain activity and capacity link MAOA polymorphism to aggressive behavior during development. Trans Psychiatry. 2: e85.
2 6 4 part iv : I maging G enetics and D isease
PA RT V. IMAGING GENETICS AND THE ENVIRONMENT
18. INCORPORATING THE ENVIRONMENT INTO NEUROGENETICS RESEARCH A N I MAGI NG- GENE- BY- ENVI RONME NT INTE RACTIONS (IG X E ) APPROACH
Luke W. Hyde, Ryan Bogdan, and Ahmad R. Hariri
GENES , EX PERI EN CE, AN D THE BR AI N A burgeoning synergy of disciplines and technologies are providing unique insights into how the dynamic interplay between genes, brain, and experience shapes complex behavior, especially risk for psychopathology. This interplay is being articulated at multiple levels of analysis from molecules to cells to neural circuits; from emotional responses to cognitive functions to personality; and from populations to families to individuals (Caspi and Moffitt 2006; Hariri 2009; Caspi, Hariri, Holmes, Uher, and Moffitt 2010; Meaney 2010). In this chapter we briefly review recent endeavors that highlight the potential value of such interdisciplinary research. We then provide perspectives on how existing approaches and methods could be leveraged further to advance understanding of the etiology, the pathophysiology, and ultimately, the treatment and prevention of psychopathology. Gene-environment interaction (G x E) (Moffitt, Caspi, and Rutter 2005) and imaging genetics (Hariri, Drabant, and Weinberger 2006) studies have both been very useful approaches to studying complex behavior, especially personality and psychopathology. G x E studies have emphasized the transactional and interacting nature of experience and the genome in the development of behavior, and imaging genetics studies have provided more proximal phenotypes and plausible mechanisms through which genes affect behavior. However, these approaches are not yet well integrated, though they have great potential to inform each other. In designing and carrying out studies that combine these methods, it is critically important
for researchers to understand and address challenges to progress that are inherent in each approach and to consider methods that address these challenges. Moreover, in order to fruitfully combine these approaches, it is also important to consider approaches to analyzing these types of data and to have an appreciation for biological mechanisms (e.g., epigenetics) through which genes and experience affect subsequent brain function and behavior. With careful consideration of all of these points, future research that combines G x E and imaging genetics approaches has the potential to greatly inform our understanding of psychopathology and to delineate more personalized and successful prevention and interventions. GE NE - E NV IRONME NT INT E R AC T IONS PR E DIC T ING B E H AV IOR A G x E interaction occurs when the relationship between an environmental experience (e.g., exposure to toxins, trauma, stress) and the emergence of altered physiological or behavioral responses (e.g., compromised immune function, psychopathology) is contingent on individual differences in genetic makeup (i.e., genetic polymorphisms, which are variations in DNA with a frequency of at least 1% in the population; functional genetic polymorphisms may reflect changes in a single [or multiple] base pair that can affect subsequent transcription of a gene or the structure of the resulting translated protein) (Moffitt et al. 2005). In other words, within G x E interaction studies, the effect of an environmental experience on outcome is conditional on genetic background (i.e., genotype) or, conversely, the effect
267
of individual genotype on behavior or health is conditional on an environmental experience (see Figure 18.1). For example, in key early work, Caspi and colleagues demonstrated, using a longitudinal study design, that well-established links between life stress and subsequent depressive symptoms were contingent on serotonin transporter linked polymorphic region (5-HTTLPR) genotype (Caspi et al. 2003). The 5-HTTLPR is a variable number of tandem repeats (VNTR) polymorphism in the promoter region of the serotonin transporter gene (SLC6A4). The serotonin transporter mediates active reuptake of synaptic serotonin and is thus critical to regulating the duration and magnitude of serotonin signaling. Based on the role of 5-HTTLPR on the physiology of the serotonin system, in this early G x E interaction study, individuals with the transcriptionally less efficient “short” allele (i.e., fewer transporter proteins available to clear serotonin from the synapse) had a strong and positive relationship between life stress and depressive phenotypes, whereas those with the “long” allele had little or no relationship between life stress and depression. These relationships are supported by meta-analysis (Karg, Burmeister, Shedden, and Sen 2011) (though not without a fair amount of controversy; see Munafo, Durrant, Lewis, and Flint 2009; Risch et al. 2009; Duncan and Keller 2011) and animal models (Caspi et al. 2010), and a wealth of other G x E studies have demonstrated similar relationships across other genes, environments, and phenotypes. For example, monoamine oxidase A (MAOA) genotype has been shown to moderate the relationship between maltreatment and antisocial behavior (Byrd and Manuck 2014; Caspi et al. 2002) and catechol O-methyltransferase (COMT) genotype has been shown to moderate the relationship between cannabis use and psychosis (Caspi et al. 2005). Theoretical reviews have revealed several key principles for conducting G x E research and evaluating resulting patterns (Moffitt et al. 2005; Caspi and Moffitt 2006): researchers should consider the heritability (the extent to which individual genetic differences contribute to phenotypic individual differences) of the target behavior, and then leverage knowledge generated in physiology and neuroscience to focus on polymorphisms in candidate genes that are of functional relevance to biological mechanisms that are sensitive to the environmental experience. A candidate gene is a gene whose protein product suggests that it may be involved in a phenotype of interest or a construct relevant to the phenotype or a gene that has been linked to a phenotype through
GWAS, an examination of genetic variation across the entire genome. Moreover, there should be evidence of variability in the response to the selected environmental experience, for which accurate measurement and, ideally, quantification should be available. Finally, there should be causal evidence linking the environmental experience with psychopathology. Because this approach does not presuppose a large main effect of single genetic variants (or experiences) on behavior but rather emphasizes an interaction with experience, carefully conducted studies of G x E interactions are instrumental in addressing several major issues that have arisen in behavioral genetics research. For example, G x E interaction studies may help to tackle the problem of “hidden heritability” (i.e., variance accounted for in twin studies of phenotypes that is unaccounted for by molecular genetic studies) raised by the general failure of genome-wide association studies (GWAS) (and specific genetic variants) to account for much of the variance attributed to heritable factors in quantitative studies (Maher 2008). By incorporating differences in environmental exposures, G x E interaction studies may help identify gene-behavior links that are weak across the entire population but strong in certain environments (e.g., after exposure to harsh parenting; Choe, Shaw, Hyde, and Forbes 2014; see also Jaffee et al. 2005; Tuvblad, Grann, and Lichtenstein 2006 for studies demonstrating that heritability varies by environmental experiences). Similarly, G x E interaction studies help to address the generally weak penetrance (i.e., the likelihood that a genotype will result in a phenotype) of polymorphisms in candidate genes (Maher 2008), and the lack of consistent replication in genetic association studies of complex behavior and psychopathology (Plomin 2005; Caspi and Moffitt 2006) by identifying environmental exposures that amplify genetic effects. G x E interaction research often represents a more plausible model of disease in which individual experiences and genetic makeup interact across development to influence relative risk, rather than more simplistic models hypothesizing independent effects of particular genetic variants or experiences. Moreover, G x E research is consistent with a growing literature supporting the existence of factors that make some individuals more or less susceptible to certain experiences. In these models, certain genes are markers for individuals who may be more susceptible or “plastic” to positive or negative experiences (Belsky et al. 2009; Belsky and Pluess 2009; Ellis and Boyce 2011; Pluess and Belsky 2013), and may help identify why only some individuals with the same experience (e.g., abuse) go on to experience psychopathology (e.g., depression, antisocial behavior).
2 6 8 part v : I maging G enetics and the E nvironment
CHALLENGES TO PROGRESS AND POSSIBLE ADVANCES IN G X E RESEARCH
Although G x E research has already advanced our understanding of the etiology of psychopathology, there are outstanding issues, limitations, and challenges to the field that deserve further consideration.
The Range of Effects First, it is unclear if G x E pertains only to harsh environments and undesirable outcomes (Heiming and Sachser 2010). Some authors have argued and provided evidence that “crossover” effects (in which a specific polymorphism is disadvantageous in some environments but advantageous in others) suggest that some polymorphisms cannot be cast as simply conferring relative “risk” but rather as shaping the range or “plasticity” or “differential susceptibility” to environmental triggers or contexts (Belsky and Pluess 2009; Ellis and Boyce 2011). Some have even argued that some genetic variation may even result in “vantage sensitivity” in which these individuals benefit more from positive experiences (Pluess and Belsky 2013). These models are both intuitively appealing and help to emphasize one point often lost in behavioral and psychiatric genetics research: a specific genetic polymorphism is unlikely to be uncovered that is a “depression gene,” responsible for a large majority of cases of disease. Rather, genes code for proteins, which in turn affect neurotransmission that biases a neural circuit’s response toward one outcome or another (e.g., more or less attention to threat), which may, in certain environments, lead to a higher likelihood of a negative (or positive) outcome. Thus, it is important to cast G x E interaction studies as highlighting that the long-term effect of genes are dependent on experience (and on the brain). Though these differential susceptibility models are important, others have argued that the limited empirical data thus far suggest that the hypothesized “plasticity” effects might not fall within a meaningful range of the data (i.e., actually observed in the real world; Manuck 2009). Fortunately, this question can be addressed through continued research, especially research that addresses enriching environments and positive outcomes (Pluess and Belsky 2013). Moreover, the application of model-fitting procedures can now be used to ascertain whether data fit better within the context of a plasticity model or a diathesis-stress model, thereby providing a means to assess this theoretical work empirically (Widaman et al. 2012; Belsky, Pluess, and Widaman 2013; Bogdan, Agrawal, Gaffrey, Tillman, and Luby 2014). chapter 1 8 : I ncorporating the E nvironment
What Is an Environmental Experience? A second outstanding issue reflects controversy in the use and definition of “environment” in G x E interaction research (Manuck and McCaffery 2010). Typically, environment refers to both experiential phenomena, including childhood abuse or adult stressors like divorce or unemployment (Caspi et al. 2003), and exposure to physical forces such as toxins, natural disasters (e.g., hurricanes, tornadoes), and acts of violence (e.g., war, terrorism; Franklin and Mansuy 2010). However, experiential phenomena and physical forces differ crucially in the degree to which the affected individual can contribute to the environmental trigger: little to none with physical forces but possibly a significant amount with experiential factors. There is a wealth of research showing that individuals vary in the degree to which they influence their own experience of adversities, and many “experiences” may thus be correlated with genotype (Scarr and McCartney 1983; O’Connor, Deater-Deckard, Fulker, Rutter, and Plomin 1998; Burt 2008; Jaffee 2011). “Experiences” that may be influenced by an individual (e.g., the extent to which a person may put themselves in stressful situations) contain significant gene-environment correlation (rGE), which captures the influence of genetically-driven variability in behavior as a precipitator of specific experiential triggers (e.g., heritable difficult temperament, resulting in harsh parenting, which in turn are both correlated with later antisocial behavior). Gene-environment correlation occurs when exposure to environmental conditions is dependent upon one’s genotype. For example, the correlation between an “environmental” risk factor such as harsh parenting and aggression may actually reflect a genetic pathway (mothers who are harsh may pass on genes to their children that increase the likelihood that they are aggressive). Thus, some G x E interaction studies may be biased by rGE in that many of the “experiences” are exposures that could be, to some degree, precipitated by the individual and thus contingent on their genotype ( Jaffee and Price 2007; Jaffee 2011). That is some G x E interaction studies may actually be finding latent G x G interactions. Several recent G x E interaction studies address this rGE issue through designs including behavior genetic approaches (Reiss and Leve 2007). For example, adoption studies can separate the extent to which parents and children share genotype (Leve et al. 2010), and twin designs can isolate ways in which experiences differentially affect individuals with the same genotype ( Jaffee et al. 2005). Studies have also addressed rGE by capitalizing on natural disasters (Kilpatrick et al. 2007) and natural experiments (Costello, Compton, Keeler, and Angold 269
2003) to explore G x E effects where individual genotypes are unlikely to be related to the experiential phenomenom (e.g., a hurricane). Studies using experimental manipulation in humans (King and Liberzon 2009) and non-human primates (Bennett et al. 2002), as well as randomized treatment designs (Brody, Beach, Philibert, Chen, and Murry 2009), are also able to exclude rGE through experimental manipulation of experience.
Three-Way Interactions Third, although G x E research alone has increased the depth and complexity of our understanding of factors influencing the etiology of psychopathology, it is certain that even greater complexity exists in the form of G x E x E and G x G x E (Kaufman et al. 2004; Wenten et al. 2009; Rutter and Dodge 2011). For example, in an interesting G x E x E study, the authors report that the 5-HTTLPR genotype x maltreatment interaction predicting depressive symptoms originally reported by Caspi and colleagues (2003) was further moderated by social support wherein only short homozygotes with a history of childhood maltreatment and low social support showed increased depressive symptoms (Kaufman et al. 2004). These results emphasize the complex and multifaceted nature of the relationship between genes, experiences, and behavior, in which some experiences exacerbate risk (maltreatment), while others are protective (high social support). This specific finding is interesting when considering the neural effects of the 5-HTTLPR in that a recent study by our group has shown that the much replicated positive correlation between amygdala reactivity to threat and trait anxiety is moderated by the level of social support (Hyde, Manuck, and Hariri 2011), further emphasizing the complexity inherent in understanding gene-behavior links. Consistently replicating such increasingly complex interactions requires sample sizes and statistical power present in few current studies, particularly when analyzing interactions using canonical approaches that involve identifying first the main effects of each variable (McClelland and Judd 1993). In this approach, the interaction is limited in power by inherent distributional properties of the interaction term in non-experimental studies and by accounting for main effects before examining interactions. This limitation in power can be further compounded by the frequency of the minor allele of a polymorphism (i.e., the less common allele at a polymorphic locus), the rate at which individuals are exposed to a given trigger (and severity of the exposure; Weder et al. 2009) and the frequency (and error of measurement) of possible dichotomous psychiatric diagnosis
(Caspi et al. 2010; though see Eaves, Silberg, and Erkanli 2003; Uher 2011) (see McClelland and Judd 1993 for a discussion of approaches that may yield more power and note that experimental studies have much greater power to detect interactions).
The Importance of Development Fourth, G x E interaction studies have typically ignored development and the possibility that findings of G x E interactions may vary depending on age or developmental stage (Vrieze, Iacono, and McGue 2012; Hyde, 2015). Thus G x E x development (G x E x D) studies are critical to understanding how these interactions unfold across development (Hyde, 2015). For example, harsh parenting may only be a potent moderator of certain genotypes when measured in early childhood and when the behavioral outcome (e.g., antisocial behavior) is measured in adolescence when rates of the outcome are higher (Choe et al. 2013). In contrast, interactions between genotype and peer experiences may only be significant in predicting behavioral outcomes when peer experiences and outcomes are measured in adolescence, when peers have the greatest effect on behavior. Studies that test G x E interactions longitudinally across multiple developmental periods are likely to help uncover these more complex interactions (Banaschewski 2012; Korosi et al. 2012).
The Consistency and Specificity of G x E Effects Fifth, in light of concerns regarding the lack of consistent replication in G x E interaction research, it is important to be cautious of spurious findings, and for replication studies and meta-analyses to be conducted (Button et al. 2013). Controversial reports on the interaction of the 5-HTTLPR and life stress predicting depression underscore this point. After initial findings (Caspi et al. 2003), replication attempts produced conflicting findings (e.g., Gillespie, Whitfield, Williams, Heath, and Martin 2005; Kendler, Kuhn, Vittum, Prescott, and Riley 2005), with recent meta-analyses reaching opposing conclusions (Duncan and Keller 2011; Karg et al. 2011). This controversy emphasizes the complexity inherent in testing, interpreting, and understanding these studies. Even at the meta-analysis stage, the question is not as simple as whether the interaction is significant or not. For example, an early meta-analysis suggested no reliable effect of this interaction on depression diagnosis (Risch et al. 2009), but was subsequently criticized for a biased selection of included studies. Specifically, authors noted that
2 7 0 part v : I maging G enetics and the E nvironment
included studies were characterized by relatively poor stress measurement (Uher and McGuffin 2010) and an emphasis on dichotomous outcomes, which lower statistical power (Uher 2011). Beyond the idea that meta-analyses themselves may vary based on inclusion or exclusion criteria, it is also important for studies to explore reasons that may contribute to conflicting findings, rather than simply arguing over whether the interaction is “reliable” or not (Byrd and Manuck 2013). For example, in the case of the 5-HTTLPR x life stress interaction, there has been evidence that the type of measure of stress (e.g., semi-structured interviews with contextual ratings versus self-report) may affect the findings with only some measures (e.g., interview), yielding consistent significant interactions (Uher and McGuffin 2010). Even if variables are measured in an ideal way, there may be further considerations affecting the consistency of the interaction having to do with what outcomes and what predictors we expect this interaction to extend to. For example, within 5-HTTLPR x stress interaction predicting depression studies, depression itself may be a heterogeneous outcome, and stress effects may vary based on the timing and severity of the stressor (Uher and McGuffin 2010). Consistent with the kindling model of depression (Post 1992), empirical research suggests that stressful life events are predictive of early depressive episodes (Bogdan et al. 2013), but less predictive of the future recurrence of depression (Kendler, Thornton, and Gardner 2000). In other words, this interaction may predict some types or patterns of depression, but not others, as “depression” is not a single and simple outcome. Studies can address this issue by exploring subtypes of disorders (e.g., child versus adult onset depression) or phenotypes within a disorder (e.g., anhedonia within depression), by narrowing criteria for a disorder (e.g., only those with recurrent rather than a single depressive episode), or by exploring specific symptoms within a disorder (e.g., focusing on symptoms of depression that have to do with negative cognition). Moreover, development may have a role: the 5-HTTLPR x early life adversity interaction more strongly predicts depression than the interaction between 5-HTTLPR and stressful life events that occur during adulthood (Karg et al. 2011; Bogdan et al. 2013), indicating that earlier and more severe experiences may be more predictive of depression than those experienced (and/or assessed) later in life. Thus one G x E interaction may actually yield high complexity with substantial nuance present when examining in which contexts the interaction will be significant, with factors such as the measurement of variables, timing chapter 1 8 : I ncorporating the E nvironment
of experiences, and heterogeneity within the outcome all affecting (i.e., moderating) expected results.
Statistical Approaches to Strengthen G x E Research Sixth, beyond these issues of measurement and development, applying certain statistical approaches may also help researchers to minimize the possibility of false positives that emerge from the way statistical interactions are tested and interpreted. Specifically, Eaves (2006) recommends using transformed continuous data as opposed to dichotomized variables to minimize false-positive results in regressions containing these interactions (Eaves 2006) (for more details see Kendler 2011; Uher 2011)).
Further Moderators and Potential Confounds Finally, beyond issues of measurement, demographic variables such as age (Lenroot and Giedd 2011) and gender (Sjöberg et al. 2006; Uher and McGuffin 2010), as well as race/ethnicity (Serretti, Kato, De Ronchi, and Kinoshita 2006; Widom and Brzustowicz 2006; Munafò et al. 2009) and possible genetic substructure (Cardon and Palmer 2003), are all likely to influence findings and require careful control and examination as additional moderators. In general, the issue of possible genetic substructure (differential allele frequencies in those of different ancestry; Shriver and Kittles 2004; Fujimura and Rajagopalan 2011) is a major issue in G x E research, as this research needs to address replication of effects across more diverse populations (Freedman et al. 2004; Helgason, Yngvadottir, Hrafnkelsson, Gulcher, and Stefánsson 2004). Moreover, as we discuss later in this chapter, sampling in both G x E interaction studies and in neuroimaging studies is critical to building a corpus of knowledge that is replicable and consistent across studies. Biases in samples can introduce artifacts that may lead to false positives, and race and culture may moderate findings (Falk et al. 2013). In sum, G x E interaction research has provided a more nuanced understanding of the interplay between biology and environment in shaping risk for psychopathology. However, G x E alone has not revealed the specific biological mechanisms for this risk (Blakely and Veenstra-Vanderweele 2011). Ultimately, for a genetic or environmental variable to affect behavior, it must “get under the skin” (Taylor, Repetti, and Seeman 1997; Hertzman and Boyce 2010; Kendler 2011; Meyer-Lindenberg 2011; Rutter and Dodge 2011; Boyce, Sokolowski, and Robinson 2012). G x E must be instantiated in the brain if it is to affect behavior and the etiology of psychopathology. 271
IM AGING GEN ETI CS As described in other chapters in this volume, linking common genetic polymorphisms to variability in brain structure, function, and connectivity is the foundation of imaging genetics (Hariri et al. 2002; Pezawas et al. 2005; Hariri et al. 2006) and fits within the broader framework of a neurogenetics approach (Hariri 2009; Bogdan, Hyde, and Hariri 2012; Hyde, Swartz, and Hariri 2014). This foundation is important for several reasons. First, by connecting genetic variation to an intermediate biological phenotype (the brain), a plausible mechanism is provided through which genes affect behavior (see Figure 18.1). For example, several studies have demonstrated a link between the short allele of the 5-HTTLPR and increased amygdala reactivity to threat (Hariri et al. 2002; Hariri et al. 2006), as well as increased functional connectivity between the amygdala and prefrontal regions (Pezawas et al. 2005). Given links between increased amygdala reactivity and anxiety and depression (Fakra et al. 2009; Price and Drevets 2010), these studies address possible mechanisms through which variation in the 5-HTTLPR might affect risk for these psychopathologies. Second, evidence supporting functional effects of a target polymorphism (e.g., altered gene transcription) strengthens its use as a proxy for individual differences in underlying brain chemistry, offering putative molecular mechanisms through which differences in brain function arise at a molecular (e.g., neurotransmitter) level. For example, in the case of the 5-HTTLPR, the short allele has been linked to decreased transcription of the serotonin transporter (Lesch et al. 1996), which affects clearance of extracellular serotonin from the synapse. Third, neural and genetic variables of interest allow for more effective synergy with animal models (e.g., transgenic mouse models, optogenetics), which in turn can advance the detailed understanding of molecular and cellular events, ultimately linking genetic variation to brain to behavior (Holmes 2008; Caspi et al. 2010; Blakely and Veenstra-Vanderweele 2011). In addition, neurogenetics approaches that use imaging genetics in the context of multimodal PET/fMRI (Fisher, Meltzer, Ziolko, Price, and Hariri 2006) and pharmacological fMRI designs (Bigos et al. 2008) have the potential to further illuminate critical molecular pathways mediating genetic effects on brain function (Hariri 2009; Caspi et al. 2010). Fourth, by focusing on dimensional and relatively objective intermediate phenotypes (e.g., regional brain activation to specific stimuli), analyses are not limited by broad nosological definitions (e.g., DMS-IV diagnoses) that are
often plagued by heterogeneity in symptoms/behaviors or inherent biases in self-report (e.g., Andreasen 2000), which is more consistent with recent shifts to a Research Domain Criteria (RDoC) approach, as emphasized by the National Institute of Mental Health (Insel et al. 2010). Moreover, by using a biological phenotype (i.e., behaviorally relevant brain structure and function) more proximal to the direct functional effects of genetic variants, imaging genetics likely gains power relative to research with more distal behavioral phenotypes, which are presumably the result of multiple interacting neural pathways. As genetically informed neurobiological pathways are identified through imaging genetics, these pathways can in turn be targeted in association studies with behavioral and/or clinical phenotypes (Hasler and Northoff 2011). In sum, imaging genetics offers new insight into psychopathology by mapping predictive links between genes, brain, and behavior, furthering our understanding of the etiology of disorders at the genetic and molecular level (see other chapters in this volume). CHALLENGES TO PROGRESS AND POSSIBLE ADVANCES IN IMAGING GENETICS RESEARCH
As in G x E interaction studies, imaging genetics studies have contributed to our understanding of psychopathology, but some major issues are worth noting.
Drawing Links from Gene to Brain to Behavior First, a majority of imaging genetics studies have established links between genetic polymorphisms and brain, but have failed to link these variables directly to meaningful differences in behavior (e.g., Hariri et al. 2002; Pezawas et al. 2005). Recently, imaging genetics studies have begun to establish such meaningful links by modeling indirect pathways from genes to behavior via the brain (Furmark et al. 2008; Fakra et al. 2009). For example, in a study by our group (Fakra et al. 2009), we examined the impact of common functional variation in the gene coding for the serotonin 1A receptor (HTR1A). Building on in vitro and in vivo studies demonstrating that the −1019G allele of HTR1A blocks transcriptional repression, leading to increased expression of this autoreceptor, which provides negative feedback on the system (Lemonde et al. 2003), and a PET study demonstrating that the density of serotonin 1A autoreceptors accounts for 30%–44% of variability in amygdala reactivity in healthy adults (Fisher et al. 2006), we examined the impact of this genetic variant on amygdala reactivity and subsequent trait anxiety in
2 7 2 part v : I maging G enetics and the E nvironment
(a) G x E Genetic variation
1A
Gene X Environment
1B Behavior & psychopathology
Environments
Modeling Further Complexity within Imaging Genetics
1C
(b) Imaging Genetics 2D
Neural functioning 2A Genetic variation
2B 2C
models specifying the brain as a mechanism between gene and behavior emphasize the importance of using statistical approaches that can model indirect (mediated) pathways (Preacher, Rucker, and Hayes 2007). Like G x E studies, imaging genetics studies demonstrate that there are important relationships between genes and behavior even when large direct relationships are not evident.
Behavior & psychopathology
Figure 18.1 A conceptual and statistical diagram of G x E and imaging
genetics studies A. A G x E framework: Genes and environments might each have a “main effect” on behavior (paths 1A and 1C), but the focus of these studies is on the interaction term, which is modeled as a product of the two variables. B. An ideal imaging genetics framework: Genetic variation in individuals leads to individual variability in neural functioning (path 2A), individual variability in neural functioning leads to differences in behavior or psychopathology (path 2B). Genetic variation might or might not have a direct impact on distal complex behavior (path 2C). Genetic variation has an indirect or mediated effect on behavior via its effect on neural functioning (arrow 2D—note that this path is not actually modeled statistically but is provided for conceptual clarity; this effect can be modeled as an indirect pathway of genes to behavior via neural functioning). Adapted from Figure 1 in Hyde LW, Bogdan R, Hariri AR. (2011). Understanding risk for psychopathology through imaging gene–environment interactions. Trends Cogn Sci. 15(9): 417–427.
a sample of healthy adults. Consistent with these findings, the −1019G allele, independent of 5-HTTLPR genotype, was associated with significantly decreased threat-related amygdala reactivity (presumably via increased negative feedback on this system). Importantly, though the main effect of this gene on trait anxiety was of small effect size and was not statistically significant, a path analysis revealed an indirect effect from this polymorphism to trait anxiety via its effect on amygdala reactivity. In other words, G carriers had decreased trait anxiety due to their relatively decreased amygdala reactivity. This study illustrates how imaging genetics studies can build effectively on other neurogenetics approaches (e.g., PET, molecular genetics assays) and that imaging genetics studies can find indirect pathways between gene and behavior through the brain, when no strong direct gene-behavior link exists. Moreover, these chapter 1 8 : I ncorporating the E nvironment
A second critical challenge within imaging genetics is to model even greater complexity of genetic effects on the brain. G x E studies clearly demonstrate the importance of considering environmental experience when estimating the effects of genetic variation on behavior, and thus we argue that the environmental variables should be modeled in future studies (see IG x E description in a later section of this chapter). Beyond issues of the environment, just as in G x E, the issue of epistasis (i.e., G x G interactions) and the likely small effect of any single polymorphism highlight the need for novel analytic approaches such as investigating these G x G interactions (Buckholtz et al. 2007; Talkowski et al. 2008), constructing cumulative genetic profiles (Nikolova, Ferrell, Manuck, and Hariri 2011), attempting hypothesis-free imaging GWAS (Liu et al. 2010) as has been done with G x E (Ege et al. 2011) (though greater application of GWAS to G x E interaction and neuroimaging research are both needed), examination of rare gene or copy number variants (Blakely and Veenstra-Vanderweele 2011), and novel statistical approaches to quantitatively integrate multiple genes into models (Hizer, Wright, and Garcia 2004; Gruenewald, Seeman, Ryff, Karlamangla, and Singer 2006). Beyond interaction effects (G x E, G x G), future studies that incorporate complementary neurogenetics techniques (e.g., neuroreceptor PET, pharmacologic challenge, animal models) or approach modeling neural reactivity in novel ways, such as through machine learning (Pereira, Mitchell, and Botvinick 2009; Yang, Liu, Sui, Pearlson, and Calhoun 2010) or graph theory (Astolfi et al. 2007; Davis et al. 2013), will better capture the molecular mechanisms mediating genetic effects on brain (Hariri 2009; Ressler et al. 2011).
The Need for a Greater Emphasis on Development As with G x E, there has been a lack of focus on development within imaging genetics (Hyde, 2015). It will be important to examine how development may moderate imaging genetics findings via longitudinal studies of individuals with repeated fMRI scans (Hyde et al. 2014). Moreover, imaging genetics findings in adults may not 273
apply to children and may be moderated by developmental stage given the large changes that occur in brain structure and function across childhood and adolescence (Paus 2012; Hyde, Shaw, and Hariri 2013).
of the predictor variable on an intermediate variable. This intermediate variable may serve as the mechanism linking the independent and dependent variable. Similarly, indirect effects denote the extent to which the independent variable affects the dependent variable through the independent variable’s effect on the mediator (and the mediators effect on the dependent variable). (Note that consistent with others [Preacher et al. 2007], we use the terms “mediation” and “indirect effects” interchangeably in this chapter and thus do not imply that direct effects must be present between independent and dependent variables in order to find indirect effects.) Here, as previously (Hyde, Bogdan, and Hariri 2011), we advocate for an integration of these approaches within a broad neurogenetics approach to help understand conditional mechanisms through which genes, environments, and the brain interact to predict behavior and risk for psychopathology. We have termed this integrative strategy “imaging gene-environment interactions” (IG x E) (see Figure 18.2). Several recent reviews (Casey et al. 2009; Caspi et al. 2010; Meyer-Lindenberg 2011) have demonstrated possible IG x E by combining findings from research in animal models,
IM AGING G X E Both G x E and imaging genetics research examines potential relationships between genetic variation and individual differences in behavior and risk for psychopathology. In G x E interaction studies, the relationship is conditional (i.e., statistical moderation) on experiences that are necessary to unmask genetic effects (or vice versa). This type of statistical moderation occurs when a “moderator” variable affects the direction and/or strength of the relationship between a predictor variable and a dependent variable. In imaging genetics, a biological mechanism can be specified (i.e., statistical mediation/indirect effects) in which variability in brain links genes and behavior. This type of statistical mediation/indirect effects occur when the link between a predictor and dependent variable is dependent upon the effects IG x E
Neural functioning
b c
Genetic variation
e d Gene X Environment
rGE
e a
b
a Behavior & psychopathology Environments
a
Figure 18.2 A conceptual model of IG x E
To understand how IG x E might be modeled conceptually and statistically, we demonstrate the relationships of the variables by highlighting traditional G x E and imaging genetics paths as well as new paths possible in IG x E studies. The a paths model typical G x E relationships; b paths model the paths from an ideal imaging genetics study. The c path models the direct effect of the environment on neural functioning demonstrated in epigenetic studies. The d path models a gene-environment interaction predicting neural functioning (IG x E effect). In this interaction, a gene would be more predictive or have a greater effect on a neural phenotype in some environments but not others (or the reverse: the environment would be predictive of neural functioning for those with one genetic variant but not another). The e path represents another interaction: the possibility that genetic variation or an environmental variable could interact with neural functioning to predict behavior. Interactions involving the environment could be between gene and environment predicting neural function (d path), or between gene and neural functioning predicting behavior (e path), but in typical G x E studies both of these interactions would be equivalent even though these interactions are likely to be due to very different mechanisms. Note that indirect and mediated pathways can be connected between many of the variables (e.g., G x E to behavior through neural functioning) and thus an ideal IG x E finding would be that the G x E interaction term predicts behavior through neural functioning. Finally, within an SEM modeling a continuous interaction, the covariance between a genetic variant and an environment can be modeled, which reflects the rGE between the specific genetic variant and specific environment. Adapted from Figure 2 in Hyde LW, Bogdan R, Hariri AR. (2011). Understanding risk for psychopathology through imaging gene–environment interactions, Trends Cogn Sci. 15(9) 417–427.
2 7 4 part v : I maging G enetics and the E nvironment
G x E interaction studies and imaging genetics studies to explain the interactions of genetic variants with environmental variables to predict learning, memory, and psychopathology. Though these reviews are exciting, empirical studies are only just beginning to test components of IG x E directly (Gerritsen et al. 2011; Kohli et al. 2011) and thus we explore how these conditional mechanisms can be specified conceptually and statistically in single human neuroimaging studies and review studies that test components of a IG x E interactions. CONCEPTUAL MODELS
Statistically, the concept of IG x E can be modeled by a moderated mediation framework (also called conditional indirect effects; Preacher et al. 2007) in which mediated/ indirect effects are moderated by a third variable. In this framework, any or all paths within a mediation framework (gene to brain, brain to behavior, gene to behavior via brain) may differ depending on the level of a moderator variable (e.g., presence or absence of childhood abuse). As seen in Figure 18.2, there are multiple ways in which genetic, neural, environmental, and behavioral variables could interact, and each model yields answers to slightly different questions (see also Preacher et al. 2007). A particularly intuitive IG x E model is a G x E in which the interaction term predicts behavior through its effect on brain function (Figure 18.2, path d). In this case, there are direct effects of both genetic and environmental variables on brain function, but their interaction predicts non-additive unique variance, which in turn predicts behavior. For example, the 5-HTTLPR polymorphism predicts increased amygdala reactivity (Hariri et al. 2002), as do experiences like early environmental deprivation (Tottenham et al. 2011), maltreatment (McCrory et al. 2013), or lower SES in childhood (Gianaros et al. 2008; Gianaros et al. 2011), and individuals with both this genetic variation and environmental experience could show a synergistic increase in amygdala reactivity, which then predicts increased anxiety symptoms. Alternatively, a positive environment such as parental warmth could negate any relationship between genetic variation in serotonin signaling and amygdala reactivity, and this lowered amygdala reactivity could then predict average levels of anxiety symptoms (see Figure 18.3). This particular interaction (G x E predicting brain function) also underlies much of the potential of IG x E approaches. By combining the power of proximal intermediate phenotypes and the potential of G x E to clarify such relationships, IG x E may provide further insight into the conundrum of hidden heritability. chapter 1 8 : I ncorporating the E nvironment
For example, if a genetic variant has no association with a neural or behavioral phenotype in most circumstances, but has a robust association in relatively rare environments (e.g., physical abuse), IG x E may be able to detect this association, particularly with more proximal neural phenotypes. Finally, it is important to note that within an IG x E model, we also specify paths in which genes or experience may moderate the brain-behavior link, based on studies supporting this statistical moderation. For example, those with a variant in a gene affecting endocannabinoid signaling show greater correlation between reward-related brain reactivity and a measure of impulsivity (Hariri et al. 2009). Additionally, those with low social support have a greater relationship between their threat-related neural reactivity and trait anxiety (Hyde, Manuck, et al. 2011). These moderated pathways likely reflect interaction between functioning in multiple brain areas but are important to test with the notion that brain-behavior links may not be static across those with different experiences and environmental resources, nor across those with different genetic backgrounds. Such relationships can be tested using path or structural equation modeling (SEM) software, or even via add-ons to common statistical packages like SAS, SPSS, and R (Hayes, 2012). As in G x E studies, the way these relationships are tested (and graphed) can affect the interpretation of the results. The statistical model itself does not necessarily specify which independent variable is the “moderator” and which is the “predictor.” For example, following from imaging genetics models, the environment could be seen in IG x E studies as the moderator of the paths in an imaging genetics analysis (see Figure 18.4). However, this approach privileges genetic factors as the “direct” predictors of neural activity, even though there is evidence that experience can affect the brain in direct and causal ways (see epigenetics section later in this chapter) and G x E studies often model genes as the moderator. Alternatively, consistent with models of differential susceptibility, genotype can be seen as a factor that makes individuals more or less sensitive to experience effect on brain and behavior, making genotype the moderator and experience the main effect on brain and behavior. This subtle point is important because within regressions and SEMs that test interactions, the equation does not specify which variable is the “moderator.” Thus the choice to graph or describe one variable (i.e., genetic polymorphism) versus another (i.e., experiential variable) is a conceptual decision, rather than a testable hypothesis, but one that will lead readers (and the public) to have a very different understanding of the role of gene and experience on brain function. 275
(a) Genetic variant (e.g., 5-HTTLPR)
Environmental experience (e.g., abuse)
Short allele carriers have less 5-HT clearance which leads to increased amygdala reactivity to threat
vit cti ea o r ala a t gd dal t my myg rea a a l th es as ng ta re itizi en c s m n e i en on us by s nvir e Ab
Two separate mechanisms cause very high chronic levels of amygdala reactivity and broader cortico-limbic reactivity
y
Above average level of anxiety symptoms
(b)
Genetic variant (e.g., 5-HTTLPR)
Environmental experience (e.g., high levels of socialsupport)
Short allele carriers have less 5-HT clearance which leads to increased amygdala reactivity to threat
Amygdala reactivity initially increased through greater 5-HT signaling Prefrontal cortex is strengthened through learning coping and down-regulates amygdala reactivity
rn lea ith ild al w arn h le s c de elp to hild g t h sms ps c enin r o i t l pp an he ea su ch ts, thr ial me ugh not c So ping tho are co iety uli x m an sti
Average level of anxiety symptoms
Figure 18.3 Possible biological models of IG x E interactions within the brain
As research suggests there are plausible causal biological mechanisms through which experience affects transcriptional effects of genes on neural functioning, it is helpful to specify how genes and environments might interact at a conceptual level to bring out the statistical relationships that could be found in IG x E studies. A. A synergistic model: Both genes and environments directly act on one brain area on similar mechanisms at the synapse. For example, SS carriers of 5-HTTLPR could have increased 5-HT signaling in the amygdala (Hariri et al. 2002), leading to greater reactivity to threat, and abuse or extreme neglect could increase the transcription of non-individually varying sequences in genes that affect amygdala function (McCrory et al. 2013; Tottenham et al. 2011), causing parallel increases in amygdala reactivity to threat. Thus the amygdala could have two pushes toward being more reactive to threat and show a multiplicatively exaggerated response. B. A buffering model: While an SS carrier of the 5-HTTLPR has increased amygdala 5-HT signaling, high levels of social support cause changes in areas of the prefrontal cortex, which are able to down-regulate amygdala reactivity, leading to normal reactivity to threat (alternatively abuse could affect the prefrontal cortex, diminishing its ability to regulate the amygdala; Roth & Sweatt, 2011). In both A and B, interactions could occur within the same brain area or across multiple brain areas within a related circuitry (e.g., a cortio-limbic circuitry). Note: it is also important to keep in mind that all of these relationships are probabilistic, not deterministic, and thus these models offer possibilities as a way of understanding IG x E. Adapted from Figure 3 in Hyde LW, Bogdan R, Hariri AR. (2011). Understanding risk for psychopathology through imaging gene–environment interactions. Trends Cogn Sci. 15(9): 417–427.
AN EXAMPLE OF A STUDY TESTING COMPONENTS OF IG X E MODEL
Approaches testing a “full” IG x E model in which a G x E interaction predicts brain function, which in turn predicts behavior through a mediated pathway, are exciting but only just beginning to emerge (e.g., Funderburk et al. 2013).
However, several studies have been published testing G x E interactions that predict brain function, a critical first step in this emerging field (Cousijn et al. 2010; Gerritsen et al. 2011; Ursini et al. 2011; Drabant et al. 2012). In the first study testing portions of an IG x E model, Canli and colleagues (2006) reported that 5-HTTLPR genotype
2 7 6 part v : I maging G enetics and the E nvironment
interacted with life stress to predict resting state activity in the amygdala. More specifically, this study found that short allele carriers, who are more susceptible to the “depressogenic” effects of stress (Karg et al. 2011), had elevated amygdala activity at rest, but only for those who had experienced more life stress, providing a neural mechanism through which short allele carriers may be more susceptible to the environment at the neural level. In other words, the short allele was only related to amygdala activity at rest for those with greater experience of negative life events, demonstrating an interaction between genotype and environment predicting brain function. In another example, Bogdan and colleagues showed, in a relatively large sample of children and adolescents (n = 279), that genetic variation affecting hypothalamic-pituitary-adrenal (HPA) axis function moderates the association between childhood emotional neglect and threat-related amygdala reactivity (Bogdan, TABLE 18.1
Williamson, and Hariri 2012) (see Chapter 23 within this volume for a more detailed description of this study). Thus, studies are beginning to demonstrate that gene effects on brain are moderated by experience (or vice versa: that experience effects on the brain are moderated by genotype), a major component of an IG x E model. CHALLENGES TO PROGRESS AND POSSIBLE ADVANCES IN THE EMERGING FIELD OF IG X E
The above promise of IG x E, like that of its parent strategies, is not without challenges. First, the challenges noted in the G x E and imaging genetics sections generally apply to IG x E models (see Table 18.1 and Figure 18.4). Second, IG x E models test statistical correlations in humans, specifying possible relationships, and thus need to be paralleled by work in animal models or with experimental designs (e.g.,
AREAS FOR PROGRESS IN IG X E AND NEUROGENETICS STUDIES
GENES Challenges
• Single polymorphisms are of small effect. • Issues such as epistasis and developmental regulation of genes have not been addressed in most studies.
Solutions
• Genetic risk profiles representing the cumulative impact of multiple functional polymorphisms within a system (e.g., dopamine)
and statistical models combining polymorphisms within and between systems (recursive partitioning, regression trees) can identify small genetic effects and their interactions. • Longitudinal studies of genetic effects in animals and humans can inform when and how each genetic variant might affect brain
and behavior. • More complex models including additional moderators such as gender, race/ethnicity, and age may uncover gene associations
present in only some groups Outstanding Questions
• When and how do most genes of interest have their effect on brain and behavior? Are there development x G x E effects that we
are missing? • Are there more complex mechanisms or organized ways in which genes interact across development? • When are certain genes having their effect across development? Are there sensitive periods for their effect?
ENVIRONMENTS Challenges
• Many G x E studies have relied on self-report or other measures with substantial error (e.g., retrospective reports). • For many environmental variables, it is not clear when certain experiences might have their effect on brain or behavior. • For many experiences, G x E studies have not paid attention to whether it is the objective account or the subjective report that matters.
Solutions
• Observational measures and multiple well-validated measures of the same construct can help decrease error of measurement,
as can modeling latent constructs of these variables to reduce error. • Prospective longitudinal studies can address developmental cascades and determine “sensitive periods” during which certain
experiences might have the greatest impact. • Studies with multiple informants and methods can compare the impact of subjective versus objective accounts of experiences.
Outstanding Questions
• Are there certain experiences that have an impact no matter when they occur? • Are there experiences that interact differently with genetic polymorphisms depending on when they occur? • Are there experiences for which objective or subjective reporting is more important? • How can we begin to identify the molecular mechanisms linking these experiences to biological embedding at the neural level?
BRAIN Challenges
• Much imaging genetics research focuses on a single brain region or the simple relationships between two regions, while behavior
reflects complex interactions within and across multiple brain regions. • fMRI studies are relatively indirect measures of cellular activity.
(continued)
chapter 1 8 : I ncorporating the E nvironment
277
TABLE 18.1
(CONTINUED)
Solutions
• Exploratory statistical techniques such as machine learning, factor analysis, and graph theory analysis use a data•driven
approach to identify complex circuit function and whole-brain network organization. • Multi-modal human and animal studies can help address cellular and molecular mechanisms underlying brain activity. Mediation
analyses can be used in multi-modal studies to provide plausible pathways (e.g., does brain structure mediate gene•brain function relationship? Do receptor levels, assayed with PET, mediate the gene-brain function relationship?). • Additionally, experimental studies (e.g., experiments that manipulate the environment, pharmacological studies to manipulate
neural chemistry) and animal studies can address mechanisms from a causal perspective. Outstanding Questions
• Are studies finding relationships between single brain areas (e.g., the amygdala) and behavior the result of more complex inter-
actions between multiple brain structures that we do not yet understand? • How does the interaction between brain regions map onto behavior? • To what extent do genetic variants affect behavior through their influence on function versus structure versus connectivity in
the brain? • How can we improve our measurement of brain structure and function to better capture changes that occur over time with experience?
COMPLEX BEHAVIOR AND PSYCHOPATHOLOGY Challenges
• Our conceptualization and resulting measurement of complex behavior and psychopathology are still rudimentary and are based
on observable behavior, which can lead to increased error in diagnosis. • Dichotomous diagnoses limit statistical and inferential power, and miss the likely dimensional nature of most behavior and psy-
chopathology. Solutions
• Imaging genetics and IG x E provide intermediate continuous phenotypes, which might be more objectively measured and more
powerful. • Continuous and hierarchical models of broad psychopathology can increase power and model the high comorbidity found in stud-
ies of psychopathology. • Examining subgroups within diagnostic category may identify underlying heterogeneity that has obscured the brain-outcome rela-
tionship in past studies. • Observational methods of behavior can provide more reliable measures, particularly when combined in latent constructs with
multiple converging self-report measures Outstanding Questions
• How do we account for the high levels of comorbidity across most psychopathology? Or the context specificity of many types of
complex behavior? • Can intermediate phenotypes and, ultimately, the genetic polymorphisms by which they are predicted usefully inform diagnosis
and treatment? • How can we define and delineate subgroups within broad diagnostic categories (e.g., antisocial personality disorder) that
express more homogeneous alterations in behavior and, by extension, brain dysfunction? • Will certain patterns of brain activity be related to broad, latent risk for psychopathology, and/or will other patterns be related to
specific psychopathologies (e.g., specific to depression versus anxiety)?
drug treatment protocols, adoption studies) that can infer causality (Meaney 2010). Moreover, as we discuss later in this chapter, these models should be guided by biologically plausible relationships between variables. Third, these complex models require substantially larger samples than those typically available to have acceptable levels of power. Moderated mediation models require starting sample sizes in the range of 500–1000 subjects to examine the expected small to moderate effects of each variable (Preacher et al. 2007). Moreover, this estimate does not include issues such as low minor allele frequencies and environmental exposure rates, which could necessitate even larger samples. Though samples of this size might sound untenable in neuroimaging, several large-scale neuroimaging studies have emerged by expertly piecing together smaller convenience samples (Fennema-Notestine et al.
2007; Yan, Craddock, Zuo, Zang, and Milham 2013), scanning larger and larger samples of individuals (Butterworth, Cherbuin, Sachdev, and Anstey 2012; Satterthwaite et al. 2012; Nikolova, Singhi, Drabant, and Hariri 2013), using data sharing and open access data (Marcus et al. 2007; Jack et al. 2008; Potkin and Ford 2009; Yan et al. 2013), consortium models (Paus 2010; Schumann et al. 2010; Fjell et al. 2012; Thyreau et al. 2012; Toga, Clark, Thompson, Shattuck, and Van Horn 2012), and neuroimaging metaanalysis (Hedman et al. 2012; Jahanshad et al. 2013), all of which could be leveraged going forward to test IG x E models. However, single studies with standard protocols (or multi-site projects with standardized protocols), representative sampling, and coverage of key genes are likely to be the most effective in testing these questions (Paus 2010; Falk et al. 2013).
2 7 8 part v : I maging G enetics and the E nvironment
Neural Reactivity Targets: brain, function, structure and connectivity Novel approaches: factor analysis, machine learning and graph theory to identify distributed circuitries
Behavior
Genetic Variability
Targets: personality, mood and affect, psychopathology Novel approaches: latent factors to identify underlying constructs across measures; Using bifactor models to identify “P” factor. Observational & experimental methods
Targets: genotypes with demonstrated functional effects Novel approaches: cumulative genetic profiles; modeling cascades of multiple interacting genes and molecular pathways, imaging GWAS
rGE
Environmental Experiences Targets: experiences that precipitate individual variability in response; Experiences linked to psychopathology; Experiences linked to changes in biology or gene expression; Extreme evironments likely to have larger effects on brain and behavior Novel approaches: observational methods, longitudinal measurement of environments, experimental manipulation or intervention studies Figure 18.4 Targets for future IG x E studies
Specific targets for G x E, imaging genetics, and IG x E studies. Importantly, novel approaches across each domain are needed to help progress understanding across all models. Moreover, similar to Figure 18.2, this model emphasizes the interaction between the environment and biology (genes, neural reactivity), as these variables predict behavior. More transparent arrows signify links made in traditional research. Bolded arrows represent newly proposed paths specific to IG x E models. Adapted from Figure 4 in Hyde LW, Bogdan R, Hariri AR. (2011). Understanding risk for psychopathology through imaging gene–environment interactions. Trends Cogn Sci. 15(9): 417–427.
Fourth, it is important to understand that development plays a large role in the unfolding of gene-environmentbrain-behavior relationships (Hyde, 2015). For example, many studied genetic variants (e.g., MAOA, 5-HTTLPR) likely have their functional affect in utero or very early in development (Sibille and Lewis 2006; Fowler et al. 2007; Jedema et al. 2010). Moreover, environmental experiences differ in their impact, depending on the developmental stage of the individual (e.g., types of stressors might differ for a child than an adult) (Sroufe and Rutter 1984; Kendler and Myers 2009) and epigenetic studies demonstrate that certain experiences might have a greater biological impact during “sensitive periods” of development (Meaney, 2010). As emphasized above, IG x E findings may also vary by developmental stage, as experiences (e.g., parenting) will change forms, and their impact on biology and the impact of genotype may be affected by the progression of development (e.g., changes associated with puberty). Moreover, as chapter 1 8 : I ncorporating the E nvironment
the brain is changing across development, its relationship to gene and experience is likely to be moderated. Thus, some IG x E findings may be specific to certain developmental periods but without greater emphasis on development (e.g., Hyde, 2015), and using IG x E methods in children and adolescents, we are unable to test these hypotheses (Hyde et al. 2014). Fifth, as would be applicable to G x E interaction studies and imaging genetics studies, sampling is a major issue that may be under-appreciated within this field, especially within neuroscience. Most of the broad neuroimaging literature on which imaging genetics and IG x E are built is based on samples of convenience (e.g., college students in subject pools, community volunteers who respond to a flier), which are likely to vary in a number of important dimensions that could affect the consistency and replicablity of findings and distort the relationship between individual differences in brain and behavior (Chiao and Cheon 279
2010; Henrich, Heine, and Norenzayan 2010). Moreover, we know very little about how brain-behavior links may be moderated by culture, and even less about how IG x E links may vary based on culture (as well as ancestry; Chiao and Cheon 2010; Chiao et al. 2010). Thus, IG x E studies that are thoughtful about sampling and who the study will generalize to are mostly likely to lead to “true” findings and additional work examining whether culture moderates of these effects (Chiao et al. 2010) will be important going forward (for more details on this point, see Falk et al. 2013). Sixth, beyond sampling, thoughtfulness about phenotype is critical to imaging genetics, G x E, and IG x E studies, as the precision needed to find smaller or complex effects is high. Thus, investigators interested in IG x E models of complex behavior and psychopathology should consider examining specific, as well as broader phenotypes. Neuroimaging has begun to demonstrate that for some phenotypes (e.g., antisocial behavior), heterogeneity within those high on a certain behavior may mask subgroups of individuals with different or even opposite etiologies. For example, in the study of youth antisocial behavior and violence, emerging fMRI studies are beginning to delineate two different sets of youth with opposite neural profiles (Viding, Fontaine, and McCrory 2012; Viding, Sebastian, et al. 2012; Hyde et al. 2013). Thus if researchers ignore possible subgroups within a specific outcome group (e.g., those high on antisocial behavior), they may find weak or non-significant effects because they are mixing heterogeneous groups of individuals together. In contrast, studies examining the structure of psychopathology (and personality) have demonstrated that bifactor models or models that specify the covariation between disorders fit observed patterns of psychopathology the best (Krueger and Markon 2006; Lahey et al. 2012; Caspi et al. 2013). If researchers model the hierarchical structure of psychopathology, specifying one latent and overarching factor (a “p factor” of general psychopathology liability), as well as specific factors that distinguish individual disorders, they may gain better traction in modeling genes, brain, experience, and their interaction as predictors of both general and specific psychopathology dimensions (Ofrat and Krueger 2012). Moreover, implicit within this point is that IG x E models will be better powered if they examine dimensional outcomes, rather than dichotomous diagnoses, especially as most psychiatric disorders have been shown to be dimensional, rather than categorical in nature (Plomin, Haworth, and Davis 2009; Krueger and Markon 2011). Finally consistent with an “RDoC” approach, understanding more basic dimensions (e.g., emotion regulation, anhedonia) that underlie specific diagnoses (and cut across diagnoses) from the neural level may help
to explain the high comorbidity seen in psychiatric disorders and model their more “basic” components (Insel et al. 2010). Finally, just as G x E and imaging genetics studies have required researchers to bridge several areas and/or work in multidisciplinary teams, IG x E studies require even greater knowledge and collaboration. We hope that the conceptual models introduced in IG x E models will garner even greater appreciation for the work of colleagues in disparate fields (e.g., animal neurophysiology, biostatistics, epidemiology, developmental psychology). PLAU S IB LE B IOLOGIC AL ME C H ANIS MS Imaging genetics studies in humans and non-human primates (e.g., Bennett et al. 2002; Jedema et al. 2010), as well as studies of strain differences in laboratory mice (e.g., Holmes 2008), convincingly link inter-individual genetic variability to differences in brain and behavior. What about the environment—does it alter biology in ways that affect brain and behavior? For many biologists, including neuroscientists, the obvious answer might be “yes,” but given the “nature-nurture” debates in some areas of psychology (Meaney 2010), it is important to specify models whereby experiences are transduced into functional biological signals that affect brain function and subsequent behavior. A fundamental example of such transduction comes from molecular studies demonstrating that learning is supported by long-term changes (i.e., long-term potentiation and depression) in synaptic physiology, which are mediated by changes in gene expression (Kandel 1991; Tada and Sheng 2006). Thus, activity-dependent gene regulation drives changes in protein expression and adaptations in the molecular machinery for neurons and neuronal circuits supporting behavior. Importantly, such environmentally induced changes ultimately manifest in the reorganization of brain circuits and their functional responses (Kandel 1991; Tada and Sheng 2006). Another fundamental mechanism governing the transduction of experience into changes in biology and behavior is epigenetics (Meaney 2010; Zhang and Meaney 2010; Mill 2011). Reviews of epigenetic regulation of brain and behavior are available elsewhere (e.g., Meaney 2010; Zhang and Meaney 2010; Mill 2011; Roth and Sweatt 2011). Briefly, epigenetic regulation refers broadly to the local (i.e., cell-specific) modifications of gene expression by way of altering the DNA-histone complex and the resulting accessibility of specific genes for transcription. Studies have demonstrated that early experience can alter epigenetic
2 8 0 part v : I maging G enetics and the E nvironment
markers and subsequent patterns of transcription in a way that affects brain structure and function as well as behavior (Meaney 2010). Chief among studies of epigenetic regulation of behavior are those conducted by Meaney and colleagues demonstrating that in rats, maternal care of offspring affects later adult behavior through epigenetic regulation of hypothalamicpituitary-adrenal (HPA) axis reactivity to stress (Weaver et al. 2004; McGowan et al. 2009; Zhang and Meaney 2010). Specifically, higher levels of maternal licking and grooming and arched-back nursing (LG-ABN) of rat pups during the first week of life leads to increased serotonin levels, which drive the expression of nerve growth factor inducible protein A (NGFI-A). Increased NGFI-A, in turn, leads to decreased methylation and increased acetylation of the promoter region of the glucocorticoid receptor (GR) gene in hippocampal neurons. This pattern of decreased methylation and increased acetylation results in increased gene expression and higher GR numbers in the hippocampus, which mediate negative feedback regulation of the HPA axis response to stress. These changes persist throughout the life span and promote adult behavior that is characterized by relative stress resilience and increased subsequent maternal care. Thus, through this epigenetic mechanism, high LG-ABN mothers beget relatively stress-resilient pups that become high LG-ABN mothers by experience-dependent mechanisms (Weaver et al. 2004; Meaney 2010). In these and similar studies, early experience affects epigenetic modifications triggering a cascade of changes in cellular signaling (particularly in the brain), which shape adult behaviors. In a compelling extension of this research to humans, a study of postmortem hippocampal tissues from individuals who committed suicide (compared to others who had accidental deaths) found increased methylation of the human GR promoter and decreased GR mRNA (McGowan et al. 2009). However, this difference was only observed in a subset of suicide completers who had been abused as children and not in completers without a history of abuse. Thus, there may be a remarkable conservation of epigenetic mechanisms regulating brain and behavior across species, which gives us confidence in developing plausible biological models of IG x E in humans based on findings in animal models (McGowan et al. 2009). Similar epigenetic effects have been documented in other genes and brain regions associated with complex behavior and psychopathology (Tsankova, Renthal, Kumar, and Nestler 2007; Roth and Sweatt 2011). Collectively these studies suggest that the environment has a very direct and long-lasting effect on biology at the epigenetic and neural level and that these effects translate chapter 1 8 : I ncorporating the E nvironment
into differences in behaviors, thus emphasizing that the environment can have a direct biological effect on gene translation, and that G x E is the rule rather than the exception when understanding variability in behavior (Meaney 2010). Trying to parse main effects of genetic versus environmental variables is to ignore that the genome and environment are in constant interaction (Meaney 2010): the biological primacy of gene-environment interactions is apparent from the realization that transcription factors can be and often are controlled by environmental signals (Zhang and Meaney 2010). Thus, these biological mechanisms indicate that the impact of genetic variation on relative risk and resilience for psychopathology will be experience and context dependent (Masten 2001). It is unclear, however, if such changes can be examined in the context of human IG x E research because it is not known if epigenetic markers in peripheral human tissue (e.g., blood cells) are faithful proxies for changes in the brain (Fraga et al. 2005; Mill 2011; Roth and Sweatt 2011; though see Tsankova et al. 2007), which is not surprising given that epigenetics is a major mechanism by which different cell types develop different functions. Moreover, future studies are needed that examine the impact of epigenetic mechanisms on genetic polymorphisms, especially promoter variants, to test true epigenetic G x E relationships (Falkenberg, Gurbaxani, Unger, and Rajeevan 2011; Roth and Sweatt 2011). For example, a recent study has demonstrated relationships between stress-related methylation of variability in catechol-O-methyltransferenace (COMT) gene, prefrontal neural reactivity, and working memory performance (Ursini et al. 2011). This study suggests that epigenetic markers may be an additional moderator to IG x E mechanisms. LOOK ING F OR WAR D With the emergence of detailed measures for both genes and brain, IG x E research is poised to accelerate the pace of scientific discovery by fueling novel exchanges between studies in humans and those in animals. Specific brain substrates (e.g., amygdala reactivity), environments (e.g., childhood neglect), and genes (e.g., 5-HTT) identified through human IG x E research can generate the next set of targets in animal models that can delve into the detailed molecular mechanisms that link these larger elements. Likewise, research in animals, especially studies that identify novel molecular and genetic factors in the regulation of brain and behavior, can generate targets for human research, which can model these factors through common polymorphisms in the genes of interest and fMRI probes of the relevant brain circuits. 281
Dynamic exchanges across human studies and animal models promise to elucidate tractable biological mechanisms that can inform the etiology and pathophysiology of psychopathology. Of course, one challenge inherent in this exchange is in specifying equivalent behaviors and experiences across human and animal studies (e.g., what does antisocial behavior look like in a rodent? Does licking and grooming in a rodent equate to nurturant parenting in humans?). Within human studies, an IG x E approach connects the pieces of the puzzle—whereas G x E interaction studies have implied that the mechanism through which G x E interactions affect behavior is through the brain, and whereas imaging genetics studies have missed the interaction of biology with experience, IG x E studies can elucidate conditional mechanisms through which genes and experience interact to affect neural structure and function and ultimately behavior and psychopathology. Specifying these models through careful statistical and methodological approaches in well-characterized samples is critical for the ability of IG x E to inform our understanding of psychopathology. The treatment implications of such work are critical as medicine moves toward greater personalization (Willard and Ginsburg 2009). For example, IG x E studies could lead to intervention and prevention trials that target those at specific genetic and/or environmental risk (Brody et al. 2009; Meaney 2010) by identifying more homogenous subgroups of individuals within the same diagnosis and those who may respond differently to treatment or need treatment tailored in a different way (Dadds and Rhodes 2008; Mehta et al. 2011; Almirall, Compton, Gunlicks-Stoessel, Duan, and Murphy 2012; Lanza and Rhoades 2013). Thus, future IG x E research might inform the development of frameworks for determining when and for whom certain treatments will work (e.g., which environments could sabotage the treatment process, which genes could predict treatment success, which combinations of genes and environments could be the targets of early preventative intervention projects) and might help to refine diagnostic criteria. Overall, IG x E can provide a more nuanced and complex model of human health and disease by extending beyond nature-nurture debates and revealing specific mechanisms through which the constantly interacting environment and genome can be understood at the level of brain function and behavior. AC K NOW L EDG MEN TS This manuscript is largely based on Hyde LW, Bogdan R, Hariri AR. (2011). Understanding risk for psychopathology
through imaging gene-environment interactions. Trends Cogn Sci. 15: 417–427. We would like to thank Patrick M. Fisher for insightful comments on a draft of this manuscript.
R E F E R E NC E S Almirall D, Compton SN, Gunlicks-Stoessel M, Duan N, Murphy SA. (2012). Designing a pilot sequential multiple assignment randomized trial for developing an adaptive treatment strategy. Stat Med. 31: 1887–1902. Andreasen NC. (2000). Schizophrenia: the fundamental questions. Brain Res Rev. 31: 106–112. Astolfi L, De Vico Fallani F, Cincotti F, Mattia D, Marciani M, Bufalari S, et al. (2007). Imaging functional brain connectivity patterns from high resolution EEG and fMRI via graph theory. Psychophysiology. 44: 880–893. Banaschewski T. (2012). Editorial: can we dissect the interplay of genes and environment across development? J Child Psychol Psychiatry. 53: 217–218. Belsky J, Jonassaint C, Pluess M, Stanton M, Brummett B, Williams R. (2009). Vulnerability genes or plasticity genes? Mol Psychiatry. 14: 746–754. Belsky J, Pluess M. (2009). Beyond diathesis stress: differential susceptibility to environmental influences. Psychol Bull. 135: 885–908. Belsky J, Pluess M, Widaman KF. (2013). Confirmatory and competitive evaluation of alternative gene-environment interaction hypotheses. J Child Psychol Psychiatry. 54: 1135–1143. Bennett A, Lesch K, Heils A, Long J, Lorenz J, Shoaf S, et al. (2002). Early experience and serotonin transporter gene variation interact to influence primate CNS function. Mol Psychiatry. 7: 118–122. Bigos KL, Pollock BG, Aizenstein HJ, Fisher PM, Bies RR, Hariri AR. (2008). Acute 5-HT reuptake blockade potentiates human amygdala reactivity. Neuropsychopharmacology. 33: 3221–3225. doi: npp200852 [pii] 10.1038/npp.2008.52. Blakely R, Veenstra-Vanderweele J. (2011). Genetic indeterminism, the 5-HTTLPR, and the paths forward in neuropsychiatric genetics. Arch Gen Psychiatry. 68: 457–458. Bogdan R, Agrawal A, Gaffrey MS, Tillman R, Luby JL. (2014). Serotonin transporter-linked polymorphic region (5-HTTLPR) genotype and stressful life events interact to predict preschool onset depression: a replication and developmental extension. J Child Psychol Psychiatry. 55: 448–457. Bogdan R, Hyde L, Hariri A. (2012). A neurogenetics approach to understanding individual differences in brain, behavior, and risk for psychopathology. Mol Psychiatry. 18: 288–299. Bogdan R, Williamson DE, Hariri AR. (2012). Mineralocorticoid receptor iso/val (rs5522) genotype moderates the association between previous childhood emotional neglect and amygdala reactivity. Am J Psychiatry. 169: 515–522. Boyce WT, Sokolowski MB, Robinson GE. (2012). Toward a new biology of social adversity. Proc Natl Acad Sci. 109: 17143–17148. Brody GH, Beach SRH, Philibert RA, Chen Y, Murry VMB. (2009). Prevention effects moderate the association of 5-HTTLPR and youth risk behavior initiation: Gene× Environment hypotheses tested via a randomized prevention design. Child Devel. 80: 645–661. Buckholtz J, Sust S, Tan H, Mattay V, Straub R, Meyer-Lindenberg A, et al. (2007). fMRI evidence for functional epistasis between COMT and RGS4. Mol Psychiatry. 12: 893–895. Burt SA. (2008). Genes and popularity evidence of an evocative gene-environment correlation. Psychol Sci.19: 112–113. Butterworth P, Cherbuin N, Sachdev P, Anstey KJ. (2012). The association between financial hardship and amygdala and hippocampal
2 8 2 part v : I maging G enetics and the E nvironment
volumes: results from the PATH through life project. Soc Cogn Affect Neurosci. 7: 548–556. doi: 10.1093/scan/nsr027. Button KS, Ioannidis JP, Mokrysz C, Nosek BA, Flint J, Robinson ES, et al. (2013). Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci. 14: 1–12. Byrd AL, Manuck SB. (2014). MAOA, childhood maltreatment, and antisocial behavior: meta-analysis of a gene-environment interaction. Biol Psychiatry. 75: 9–17. doi: http://dx.doi.org/10.1016/j. biopsych.2013.1005.1004. Canli T, Qiu M, Omura K, Congdon E, Haas BW, Amin Z, et al. (2006). Neural correlates of epigenesis. Proc Natl Acad Sci. 103: 16033–16038. Cardon LR, Palmer LJ. (2003). Population stratification and spurious allelic association. Lancet. 361: 598–604. Casey BJ, Glatt CE, Tottenham N, Soliman F, Bath K, Amso D, et al. (2009). Brain-derived neurotrophic factor as a model system for examining gene by environment interactions across development. Neuroscience. 164: 108–120. Caspi A, Hariri AR, Holmes A, Uher R, Moffitt TE. (2010). Genetic sensitivity to the environment: the case of the serotonin transporter gene and its implications for studying complex diseases and traits. Am J Psychiatry. 8: 398–416. doi: appi.ajp.2010.09101452 [pii] 10.1176/appi.ajp.2010.09101452. Caspi A, Houts RM, Belsky DW, Goldman-Mellor SJ, Harrington H, Israel S, et al. (2014). The p factor: one general psychopathology factor in the structure of psychiatric disorders? Clin Psychol Sci. 2: 119–137. Caspi A, McClay J, Moffitt TE, Mill J, Martin J, Craig IW, et al. (2002). Role of genotype in the cycle of violence in maltreated children. Science. 29:, 851–854. doi: 10.1126/science.1072290297/5582/851 Caspi A, Moffitt TE. (2006). Gene–environment interactions in psychiatry: joining forces with neuroscience. Nat Rev Neurosci. 7: 583–590. Caspi A, Moffitt TE, Cannon M, McClay J, Murray R, Harrington HL, et al. (2005). Moderation of the effect of adolescent-onset cannabis use on adult psychosis by a functional polymorphism in the catechol-O-methyltransferase gene: longitudinal evidence of a gene X environment interaction. Biol Psychiatry. 57: 1117–1127. Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H, et al. (2003). Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science. 301: 386–389. doi: 10.1126/ science.1083968. Chiao JY, Cheon BK. (2010). The weirdest brains in the world. Behav Brain Sci. 33: 88–90. Chiao JY, Hariri AR, Harada T, Mano Y, Sadato N, Parrish TB, et al. (2010). Theory and methods in cultural neuroscience. Soc Cogn Affect Neurosci. 5: 356–361. Choe DE, Shaw DS, Hyde LW, Forbes EE. (2014). Interactions between MAOA and punitive discipline in African American and Caucasian men’s antisocial behavior. Clin Psychol Sci. 2: 591–601. Costello EJ, Compton SN, Keeler G, Angold A. (2003). Relationships between poverty and psychopathology. JAMA. 290: 2023–2029. Cousijn H, Rijpkema M, Qin S, van Marle HJ, Franke B, Hermans EJ, et al. (2010). Acute stress modulates genotype effects on amygdala processing in humans. Proc Natl Acad Sci. 107: 9867–9872. Dadds MR, Rhodes T. (2008). Aggression in young children with concurrent callous-unemotional traits: can the neurosciences inform progress and innovation in treatment approaches? Philos Trans Roy Soc London B. 363: 2567–2576. doi: N269U824637LP003 [pii] 10.1098/rstb.2008.0029. Davis FC, Knodt AR, Sporns O, Lahey BB, Zald DH, Brigidi BD, et al. (2013). Impulsivity and the modular organization of resting-state neural networks. Cereb Cortex. 23: 1444–1452. Drabant EM, Ramel W, Edge MD, Hyde LW, Kuo JR, Goldin PR, et al. (2012). Neural mechanisms underlying 5-HTTLPR-related sensitivity to acute stress. Am J Psychiatry. 169: 397–405. Duncan LE, Keller MC. (2011). A critical review of the first 10 years of candidate gene-by-environment interaction research in psychiatry. Am J Psychiatry. 168: 1041–1049. chapter 1 8 : I ncorporating the E nvironment
Eaves L. (2006). Genotype x environment interaction in psychopathology: factor or artifact? Twin Res Human Genet. 36: 74–87. Eaves L, Silberg J, Erkanli A. (2003). Resolving multiple epigenetic pathways to adolescent depression. J Child Psychol Psychiatry. 44: 1006–1014. Ege MJ, Strachan DP, Cookson WOCM, Moffatt MF, Gut I, Lathrop M, et al. (2011). Gene-environment interaction for childhood asthma and exposure to farming in Central Europe. J Allergy Clin Immun. 127: 138–144. Ellis BJ, Boyce WT. (2011). Differential susceptibility to the environment: Toward an understanding of sensitivity to developmental experiences and context. Devel Psychopathol. 23: 1–5. Fakra E, Hyde L, Gorka A, Fisher P, Munoz K, Kimak M, et al. (2009). Effects of HTR1A C (-1019) G on amygdala reactivity and trait anxiety. Arch Gen Psychiatry. 66: 33–40. Falk EB, Hyde LW, Mitchell C, Faul J, Gonzalez R, Heitzeg MM, et al. (2013). Neuroscience meets population science: what is a representative brain? Proc Natl Acad Sci. 110: 17615–17622, doi: 10.1073/ pnas.1310134110. Early edition: www.pnas.org/cgi/doi/10.1073/ pnas.1310134110. Falkenberg VR, Gurbaxani BM, Unger ER, Rajeevan MS. (2011). Functional genomics of serotonin receptor 2A (HTR2A): interaction of polymorphism, methylation, expression and disease association. Neuromol Med. 13: 66–76. Fennema-Notestine C, Gamst AC, Quinn BT, Pacheco J, Jernigan TL, Thal L, et al. (2007). Feasibility of multi-site clinical structural neuroimaging studies of aging using legacy data. Neuroinformatics. 5: 235–245. Fisher PM, Meltzer CC, Ziolko SK, Price JC, Hariri AR. (2006). Capacity for 5-HT1A–mediated autoregulation predicts amygdala reactivity. Nat Neurosci. 9: 1362–1363. Fjell AM, Walhovd KB, Brown TT, Kuperman JM, Chung Y, Hagler DJ, et al. (2012). Multimodal imaging of the self-regulating developing brain. Proc Natl Acad Sci. 109: 19620–19625. Fowler JS, Alia-Klein N, Kriplani A, Logan J, Williams B, Zhu W, et al. (2007). Evidence that brain MAO A activity does not correspond to MAO A genotype in healthy male subjects. Biol Psychiatry. 62: 355–358. Fraga MF, Ballestar E, Paz MF, Ropero S, Setien F, Ballestar ML, et al. (2005). Epigenetic differences arise during the lifetime of monozygotic twins. Proc Natl Acad Sci U S A. 102: 10604–10609. Franklin TB, Mansuy IM. (2010). Epigenetic inheritance in mammals: evidence for the impact of adverse environmental effects. Neurobiol Disease. 39: 61–65. Freedman ML, Reich D, Penney KL, McDonald GJ, Mignault AA, Patterson N, et al. (2004). Assessing the impact of population stratification on genetic association studies. Nat Genet. 36: 388–393. Fujimura JH, Rajagopalan R. (2011). Different differences: the use of ‘genetic ancestry’ versus race in biomedical human genetic research. Social Stud Sci. 41: 5–30. Funderburk SC, Michalski LJ, Carey CE, Gorka AX, Drabant EM, Bogdan R, et al. (2013). Ventral striatum reactivity and coping strategies indirectly link a PDYN haplotype to alcohol use. Paper presented at the 21st International Society of Psychiatric Genetics Meeting, Boston, MA. Furmark T, Appel L, Henningsson S, Åhs F, Faria V, Linnman C, et al. (2008). A link between serotonin-related gene polymorphisms, amygdala activity, and placebo-induced relief from social anxiety. J Neurosci. 28: 13066–13074. Gerritsen L, Tendolkar I, Franke B, Vasquez A, Kooijman S, Buitelaar J, et al. (2011). BDNF Val66Met genotype modulates the effect of childhood adversity on subgenual anterior cingulate cortex volume in healthy subjects. Mol Psychiatry. ePub online; doi: 10.1038/ mp.2011.51. Gianaros PJ, Horenstein JA, Hariri AR, Sheu LK, Manuck SB, Matthews KA, et al. (2008). Potential neural embedding of parental social standing. Soc Cogn Affect Neurosci. 3: 91–96. 283
Gianaros PJ, Manuck SB, Sheu LK, Kuan DCH, Votruba-Drzal E, Craig AE, et al. (2011). Parental Education Predicts Corticostriatal Functionality in Adulthood. Cereb Cortex. 21: 896–910. Gillespie NA, Whitfield JB, Williams B, Heath AC, Martin NG. (2005). The relationship between stressful life events, the serotonin transporter (5-HTTLPR) genotype and major depression. Psychol Med. 35: 101–111. Gruenewald TL, Seeman TE, Ryff CD, Karlamangla AS, Singer BH. (2006). Combinations of biomarkers predictive of later life mortality. Proc Natl Acad Sci U S A. 103: 14158–14163. Hariri AR. (2009). The neurobiology of individual differences in complex behavioral traits. Ann Rev Neurosci. 32, 225–247. doi: 10.1146/ annurev.neuro.051508.135335. Hariri AR, Drabant EM, Weinberger DR. (2006). Imaging genetics: perspectives from studies of genetically driven variation in serotonin function and corticolimbic affective processing. Biol Psychiatry. 59: 888–897. Hariri AR, Gorka A, Hyde LW, Kimak M, Halder I, Ducci F, et al. (2009). Divergent effects of genetic variation in endocannabinoid signaling on human threat- and reward-related brain function. Biol Psychiatry. 66: 9–16. doi: 10.1016/j.biopsych.2008.10.047. Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D, et al. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science. 297: 400. Hasler G, Northoff G. (2011). Discovering imaging endophenotypes for major depression. Mol Psychiatry. 1: 1–16. Hayes AF. (2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. http://www.afhayes.com/public/process2012.pdf. Hedman AM, van Haren NE, Schnack HG, Kahn RS, Pol H, Hilleke E. (2012). Human brain changes across the life span: a review of 56 longitudinal magnetic resonance imaging studies. Hum Brain Mapp. 33: 1987–2002. Heiming RS, Sachser N. (2010). Consequences of serotonin transporter genotype and early adversity on behavioral profile: pathology or adaptation? Front Neurosci. 4: 187. doi: 10.3389/fnins.2010.00187. Helgason A, Yngvadottir B, Hrafnkelsson B, Gulcher J, Stefánsson K. (2004). An Icelandic example of the impact of population structure on association studies. Nat Genet. 37: 90–95. Henrich J, Heine SJ, Norenzayan A. (2010). The weirdest people in the world. Behav Brain Sci. 33: 61–83. Hertzman C, Boyce T. (2010). How experience gets under the skin to create gradients in developmental health. Ann Rev Public Health. 31: 329–347. Hizer SE, Wright TM, Garcia DK. (2004). Genetic markers applied in regression tree prediction models. Animal Genet. 35: 50–52. Holmes A. (2008). Genetic variation in cortico-amygdala serotonin function and risk for stress-related disease. Neurosci Biobehav Rev. 32: 1293–1314. doi: S0149-7634(08)00042-0 [pii] 10.1016/j.neubiorev.2008.03.006. Hyde, L.W. (2015). Developmental psychopathology in an era of molecular genetics and neuroimaging: A developmental neurogenetics approach. Development and Psychopathology. 27: 587–613. Hyde LW, Bogdan R, Hariri AR. (2011). Understanding risk for psychopathology through imaging gene-environment interactions. Trends Cogn Sci. 15: 417–427. Hyde LW, Manuck SB, Hariri AR. (2011). Social support moderates the link between amygdala reactivity and trait anxiety. Neuropsychologia. 49: 651–656. Hyde LW, Shaw DS, Hariri AR. (2013). Neuroscience, developmental psychopathology and youth antisocial behavior: review, integration, and directions for research. Devel Rev. 33: 168–223. Hyde LW, Swartz JR, Waller, R., Hariri AR. (2014). Neurogenetics approaches to mappying pathways in developmental psychopathology. In Cicchetti D, ed., Wiley’s Developmental psychopathology (3rd ed., Vol. II). Hoboken, NJ: Wiley and Sons.
Insel TR, Cuthbert BN, Garvey MA, Heinssen RK, Pine DS, Quinn KJ, et al. (2010). Research domain criteria (RDoC): toward a new classification framework for research on mental disorders. Am J Psychiatry.167: 748–751. Jack CR, Bernstein MA, Fox NC, Thompson P, Alexander G, Harvey D, et al. (2008). The Alzheimer’s disease neuroimaging initiative (ADNI): MRI methods. J Magn Reson Imaging. 27: 685–691. Jaffee SR. (2011). Genotype-environment correlations: definitions, methods of measurement, and implications for research on adolescent pscyhopathology. In Kendler KS, Jaffee SR, Romer D, eds., The dynamic genome and mental health (pp. 79–102). New York: Oxford. Jaffee SR, Caspi A, Moffitt TE, Dodge KA, Rutter M, Taylor A, et al. (2005). Nature X nurture: genetic vulnerabilities interact with physical maltreatment to promote conduct problems. Devel Psychopathol. 17: 67–84. Jaffee SR, Price TS. (2007). Gene–environment correlations: a review of the evidence and implications for prevention of mental illness. Mol Psychiatry. 12: 432–442. Jahanshad N, Kochunov P, Sprooten E, Mandl RC, Nichols TE, Almassy L, et al. (2013). Multi-site genetic analysis of diffusion images and voxelwise heritability analysis: a pilot project of the ENIGMA–DTI Working Group. Neuroimage. 81: 455–469. doi: 10.1016/j.neuroimage.2013.1004.1061. Jedema HP, Gianaros PJ, Greer PJ, Kerr DD, Liu S, Higley JD, et al. (2010). Cognitive impact of genetic variation of the serotonin transporter in primates is associated with differences in brain morphology rather than serotonin neurotransmission. Mol Psychiatry. 15: 512–522. Kandel ER. (1991). Cellular mechanisms of learning and the biological basis of individuality. In Kandel ER, Schwartz JH, Jessell TM, eds., Principles of neural science (4th ed., pp. 1247–1279). New York: McGraw-Hill. Karg K, Burmeister M, Shedden K, Sen S. (2011). The serotonin transporter promoter variant (5-HTTLPR), stress, and depression meta-analysis revisited: evidence of genetic moderation. Arch Gen Psychiatry. 68: 444–454. Kaufman J, Yang BZ, Douglas-Palumberi H, Houshyar S, Lipschitz D, Krystal JH, et al. (2004). Social supports and serotonin transporter gene moderate depression in maltreated children. Proc Natl Acad Sci U S A. 101: 17316–17321. Kendler KS. (2011). A conceptual overview of gene-environment interaction and correlation in a developmental context. In Kendler KS, Jaffee SR, Romer D, eds., The dynamic genome and mental health (pp. 5–28). New York: Oxford. Kendler KS, Kuhn JW, Vittum J, Prescott CA, Riley B. (2005). The interaction of stressful life events and a serotonin transporter polymorphism in the prediction of episodes of major depression: a replication. Arch Gen Psychiatry. 62: 529–535. Kendler KS, Myers J. (2009). A developmental twin study of church attendance and alcohol and nicotine consumption: a model for analyzing the changing impact of genes and environment. Am J Psychiatry. 166: 1150–1155. Kendler KS, Thornton LM, Gardner CO. (2000). Stressful life events and previous episodes in the etiology of major depression in women: an evaluation of the “kindling” hypothesis. Am J Psychiatry. 157: 1243–1251. Kilpatrick DG, Koenen KC, Ruggiero KJ, Acierno R, Galea S, Resnick HS, et al. (2007). The serotonin transporter genotype and social support and moderation of posttraumatic stress disorder and depression in hurricane-exposed adults. Am J Psychiatry. 164: 1693–1699. King AP, Liberzon I. (2009). Assessing the neuroendocrine stress response in the functional neuroimaging context. Neuroimage. 47: 1116–1124. Kohli MA, Lucae S, Saemann PG, Schmidt MV, Demirkan A, Hek K, et al. (2011). The neuronal transporter gene SLC6A15 confers risk to major depression. Neuron. 70: 252–265.
2 8 4 part v : I maging G enetics and the E nvironment
Korosi A, Naninck E, Oomen C, Schouten M, Krugers H, Fitzsimons C, et al. (2012). Early-life stress mediated modulation of adult neurogenesis and behavior. Behav Brain Res. 227: 400–409. Krueger RF, Markon KE. (2006). Reinterpreting comorbidity: a model-based approach to understanding and classifying psychopathology. Ann Rev Clin Psychol. 2: 111–133. Krueger RF, Markon KE. (2011). A dimensional-spectrum model of psychopathology: progress and opportunities. Arch Gen Psychiatry. 68: 10. Lahey BB, Applegate B, Hakes JK, Zald DH, Hariri AR, Rathouz PJ. (2012). Is there a general factor of prevalent psychopathology during adulthood? J Abnorm Psychology. 121: 971–977. Lanza ST, Rhoades BL. (2013). Latent class analysis: an alternative perspective on subgroup analysis in prevention and treatment. Prev Sci. 14: 157–168. Lemonde S, Turecki G, Bakish D, Du L, Hrdina PD, Bown CD, et al. (2003). Impaired repression at a 5-hydroxytryptamine 1A receptor gene polymorphism associated with major depression and suicide. J Neurosci. 23: 8788. Lenroot RK, Giedd JN. (2011). Annual research review: developmental considerations of gene by environment interactions. J Child Psychol Psychiatry. 52: 429–441. Lesch KP, Bengel D, Heils A, Sabol SZ, Greenberg BD, Petri S, et al. (1996). Association of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory region. Science. 274: 1527–1531. Leve LD, Kerr DCR, Shaw D, Ge X, Neiderhiser JM, Scaramella LV, et al. (2010). Infant pathways to externalizing behavior: evidence of Genotype× Environment interaction. Child Devel. 81: 340–356. Liu X, Akula N, Skup M, Brotman MA, Leibenluft E, McMahon FJ. (2010). A genome-wide association study of amygdala activation in youths with and without bipolar disorder. J Amer Acad Child Adolesc Psychiatry. 49: 33–41. Maher B. (2008). Personal genomes: the case of the missing heritability. Nature. 456: 18–21. Manuck SB. (2009). The reaction norm in gene × environment interaction. Mol Psychiatry. 15: 881–882. Manuck SB, McCaffery JM. (2010). Genetics of stress: gene-stress correlation and interaction. In Steptoe A, ed., Handbook of behavioral medicine: methods and applications (pp. 454–478). New York: Springer. Marcus DS, Wang TH, Parker J, Csernansky JG, Morris JC, Buckner RL. (2007). Open Access Series of Imaging Studies (OASIS): cross-sectional MRI data in young, middle aged, nondemented, and demented older adults. J Cogn Neurosci. 19: 1498–1507. Masten AS. (2001). Ordinary magic: resilience processes in development. Am Psychologist. 56: 227–238. McClelland GH, Judd CM. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychol Bull. 114: 376–390. McCrory EJ, De Brito SA, Kelly PA, Bird G, Sebastian CL, Mechelli A, et al. (2013). Amygdala activation in maltreated children during pre-attentive emotional processing. Br J Psychiatry. epub ahead of print; doi: 10.1192/bjp.bp.1112.116624. McGowan PO, Sasaki A, D’Alessio AC, Dymov S, Labonté B, Szyf M, et al. (2009). Epigenetic regulation of the glucocorticoid receptor in human brain associates with childhood abuse. Nat Neurosci. 12: 342–348. Meaney MJ. (2010). Epigenetics and the biological definition of gene × environment interactions. Child Devel. 81: 41–79. Mehta D, Gonik M, Klengel T, Rex-Haffner M, Menke A, Rubel J, et al. (2011). Using polymorphisms in FKBP5 to define biologically distinct subtypes of posttraumatic stress disorder: evidence from endocrine and gene expression studies. Arch Gen Psychiatry. epub online, doi: 10.1001/archgenpsychiatry.2011.50. Meyer-Lindenberg A. (2011). Neurogenetic mechanisms of gene-environment interactions. In Dodge KA, Rutter M, ed., Gene-environment interactions in developmental psychopathology (pp. 71–87). New York: Guilford.
chapter 1 8 : I ncorporating the E nvironment
Mill J. (2011). Epigenetic effectson gene function and their role in mediating gene-environment interactions. In Kendler KS, Jaffee SR, Romer D, eds., The dynamic genome and mental health (pp. 145–171). New York: Oxford. Moffitt TE, Caspi A, Rutter M. (2005). Strategy for investigating interactions between measured genes and measured environments. Arch Gen Psychiatry. 62: 473–481. Munafò MR, Durrant C, Lewis G, Flint J. (2009). Gene environment interactions at the serotonin transporter locus. Biol Psychiatry. 65: 211–219. Munafò MR, Freimer NB, Ng W, Ophoff R, Veijola J, Miettunen J, et al. (2009). 5-HTTLPR genotype and anxiety-related personality traits: a meta-analysis and new data. Am J Med Genet B. 150: 271–281. Nikolova YS, Ferrell RE, Manuck SB, Hariri AR. (2011). Multilocus genetic profile for dopamine signaling predicts ventral striatum reactivity. Neuropsychopharmacology. 36: 1940–1947. doi:10.1038/ npp.2011.82. Nikolova YS, Singhi EK, Drabant EM, Hariri AR. (2013). Reward‐ related ventral striatum reactivity mediates gender‐specific effects of a galanin remote enhancer haplotype on problem drinking. Genes Brain Behav. 12: 516–524. doi: 10.1111/gbb.12035. O’Connor TG, Deater-Deckard K, Fulker D, Rutter M, Plomin R. (1998). Genotype–environment correlations in late childhood and early adolescence: antisocial behavioral problems and coercive parenting. Devel Psychology. 34: 970–981. Ofrat S, Krueger RF. (2012). How research on the meta-structure of psychopathology aids in understanding biological correlates of mood and anxiety disorders. Biol MoodAnxiety Disord. 2: 13. Paus T. (2010). Population neuroscience: why and how. Hum Brain Mapp. 31: 891–903. Paus T. (2012). Some thoughts on the relationship of developmental science and population neuroscience. Int J Devel Sci. 6: 9–11. Pereira F, Mitchell T, Botvinick M. (2009). Machine learning classifiers and fMRI: a tutorial overview. Neuroimage. 45: S199–S209. Pezawas L, Meyer-Lindenberg A, Drabant EM, Verchinski BA, Munoz KE, Kolachana BS, et al. (2005). 5-HTTLPR polymorphism impacts human cingulate-amygdala interactions: a genetic susceptibility mechanism for depression. Nat Neurosci. 8: 828–834. Plomin R. (2005). Finding genes in child psychology and psychiatry: when are we going to be there? J Child Psychol Psychiatry. 46: 1030–1038. Plomin R, Haworth C, Davis O. (2009). Common disorders are quantitative traits. Nat Rev Genet. 10: 872–878. Pluess M, Belsky J. (2013). Vantage sensitivity: individual differences in response to positive experiences. Psychol Bull. 139: 901–916. Post RM. (1992). Transduction of psychosocial stress into the neurobiology of affective disorder. Am J Psychiatry. 149: 999–1010. Potkin, S. G., & Ford, J. M. (2009). Widespread cortical dysfunction in schizophrenia: the FBIRN imaging consortium. Schizophrenia Bull. 35: 15–18. Preacher KJ, Rucker DD, Hayes AF. (2007). Addressing moderated mediation hypotheses: theory, methods, and prescriptions. Multivar Behav Res. 42: 185–227. Price JL, Drevets WC. (2010). Neurocircuitry of mood disorders. Neuropsychopharmacology. 35: 192–216. doi: npp2009104 [pii] 10.1038/ npp.2009.104. Reiss D, Leve LD. (2007). Genetic expression outside the skin: clues to mechanisms of genotype × environment interaction. Development and Psychopathology. 19: 1005–1027. Ressler KJ, Mercer KB, Bradley B, Jovanovic T, Mahan A, Kerley K, et al. (2011). Post-traumatic stress disorder is associated with PACAP and the PAC1 receptor. Nature. 470: 492–497. Risch N, Herrell R, Lehner T, Liang KY, Eaves L, Hoh J, et al. (2009). Interaction between the serotonin transporter gene (5-HTTLPR), stressful life events, and risk of depression. JAMA. 301: 2462–2471. Roth TL, Sweatt JD. (2011). Annual research review: epigenetic mechanisms and environmental shaping of the brain during sensitive periods of development. J Child Psychol Psychiatry. 52: 398–408. 285
Rutter M, Dodge KA. (2011). Gene-environment interactions: the state of science. In Dodge KA, Rutter M, eds., Gene-environment interaction in developmental psychopathology (pp. 87–101). New York: Guilford. Satterthwaite TD, Ruparel K, Loughead J, Elliott MA, Gerraty RT, Calkins ME, et al. (2012). Being right is its own reward: load and performance related ventral striatum activation to correct responses during a working memory task in youth. Neuroimage. 61: 723–729. Scarr S, McCartney K. (1983). How people make their own environments: a theory of genotype → environment effects. Child Devel. 54: 424–435. Schumann G, Loth E, Banaschewski T, Barbot A, Barker G, Büchel C, et al. (2010). The IMAGEN study: reinforcement-related behaviour in normal brain function and psychopathology. Mol Psychiatry. 15: 1128–1139. Serretti A, Kato M, De Ronchi D, Kinoshita T. (2006). Meta-analysis of serotonin transporter gene promoter polymorphism (5-HTTLPR) association with selective serotonin reuptake inhibitor efficacy in depressed patients. Mol Psychiatry. 12: 247–257. Shriver MD, Kittles RA. (2004). Genetic ancestry and the search for personalized genetic histories. Nat Rev Genet. 5: 611–618. Sibille E, Lewis DA. (2006). SERT-ainly involved in depression, but when? Am J Psychiatry, 163: 8–11. Sjöberg RL, Nilsson, KW, Wargelius HL, Leppert J, Lindström L, Oreland L. (2006). Adolescent girls and criminal activity: role of MAOA-LPR genotype and psychosocial factors. Am J Med Genet B. 144: 159–164. Sroufe LA, Rutter M. (1984). The domain of developmental psychopathology. Child Devel. 55: 17–29. Tada T, Sheng M. (2006). Molecular mechanisms of dendritic spine morphogenesis. Curr Opin Neurobiol. 16: 95–101. Talkowski ME, Kirov G, Bamne M, Georgieva L, Torres G, Mansour H, et al. (2008). A network of dopaminergic gene variations implicated as risk factors for schizophrenia. Human Mol Genet. 17: 747–758. Taylor SE, Repetti RL, Seeman T. (1997). Health psychology: what is an unhealthy environment and how does it get under the skin? Ann Rev Psychology. 48: 411–447. Thyreau B, Schwartz Y, Thirion B, Frouin V, Loth E, Vollstädt-Klein S, et al. (2012). Very large fMRI study using the IMAGEN database: sensitivity–specificity and population effect modeling in relation to the underlying anatomy. Neuroimage. 61: 295–303. Toga AW, Clark KA, Thompson PM, Shattuck DW, Van Horn JD. (2012). Mapping the human connectome. Neurosurgery. 71: 1–5. doi: doi: 10.1227/NEU.0b013e318258e9ff. Tottenham N, Hare T, Millner A, Gilhooly T, Zevin J, Casey B. (2011). Elevated amygdala response to faces following early deprivation. Devel Sci. 14: 190–204. Tsankova N, Renthal W, Kumar A, Nestler EJ. (2007). Epigenetic regulation in psychiatric disorders. Nat Rev Neurosci. 8: 355–367. Tuvblad C, Grann M, Lichtenstein P. (2006). Heritability for adolescent antisocial behavior differs with socioeconomic status: gene-environment interaction. J Child Psychol Psychiatry.
47: 734–743. doi: JCPP1552 [pii] 10.1111/j.1469-7610.2005.015 52.x Uher R. (2011). Gene-environment interactions. In Kendler KS, Jaffee SR, Romer D, eds., The dynamic genome and mental health (pp. 29–58). New York: Oxford. Uher R, McGuffin P. (2010). The moderation by the serotonin transporter gene of environmental adversity in the etiology of depression: 2009 update. Mol Psychiatry. 15: 18–22. Ursini G, Bollati V, Fazio L, Porcelli A, Iacovelli L, Catalani A, et al. (2011). Stress-related methylation of the catechol-O-methyltransferase val158 allele predicts human prefrontal cognition and activity. J Neurosci. 31: 6692–6698. Viding E, Fontaine NMG, McCrory EJ. (2012). Antisocial behaviour in children with and without callous-unemotional traits. J Roy Soc Med. 105: 195–200. Viding E, Sebastian CL, Dadds MR, Lockwood PL, Cecil CAM, De Brito SA, et al. (2012). Amygdala response to preattentive masked fear in children with conduct problems: the role of callous-unemotional traits. Am J Psychiatry. 169. 1109–1116. Vrieze SI, Iacono WG, McGue M. (2012). Confluence of genes, environment, development, and behavior in a post Genome-Wide Association Study world. Devel Psychopathol. 24: 1195–1214. Weaver ICG, Cervoni N, Champagne FA, D’Alessio AC, Sharma S, Seckl JR, et al. (2004). Epigenetic programming by maternal behavior. Nat Neurosci. 7: 847–854. Weder N, Yang BZ, Douglas-Palumberi H, Massey J, Krystal JH, Gelernter J, et al. (2009). MAOA genotype, maltreatment, and aggressive behavior: the changing impact of genotype at varying levels of trauma. Biol Psychiatry. 65: 417–424. Wenten M, Gauderman WJ, Berhane K, Lin PC, Peters J, Gilliland FD. (2009). Functional variants in the catalase and myeloperoxidase genes, ambient air pollution, and respiratory-related school absences: an example of epistasis in gene-environment interactions. Am J Epidemiol. 170: 1494–1501. Widaman KF, Helm JL, Castro-Schilo L, Pluess M, Stallings MC, Belsky J. (2012). Distinguishing ordinal and disordinal interactions. Psychological Methods. 17: 615–622. Widom CS, Brzustowicz LM. (2006). MAOA and the “cycle of violence:” childhood abuse and neglect, MAOA genotype, and risk for violent and antisocial behavior. Biol Psychiatry. 60: 684–689. Willard HF, Ginsburg GS. (2009). Essentials of genomic and personalized medicine. San Diego, CA: Academic Press. Yan C-G, Craddock RC, Zuo X-N, Zang Y-F, Milham MP. (2013). Standardizing the intrinsic brain: towards robust measurement of inter-individual variation in 1000 functional connectomes. Neuroimage. 80: 246–262. doi: 10.1016/j.neuroimage.2013.1004.1081. Yang H, Liu J, Sui J, Pearlson G, Calhoun D. (2010). A hybrid machine learning method for fusing fMRI and genetic data: combining both improves classification of schizophrenia. Front Human Neurosci. 4: 192. Zhang TY, Meaney MJ. (2010). Epigenetics and the environmental regulation of the genome and its function. Ann Rev Psychology. 61: 439–466.
2 8 6 part v : I maging G enetics and the E nvironment
PA RT V I . IMAGING GENETICS AND COGNITION
19. IMAGING GENETICS OF EPISODIC MEMORY Björn Rasch, Susanne Erk, Andreas Papassotiropoulos, and Dominique J.-F. de Quervain
INTRODUCTI O N Human memory is a cognitive trait that is influenced by both genetic and environmental factors. Twin studies have estimated that genetic factors account for approximately 50% of the variability in human memory capacity (McClearn et al. 1997), indicating that naturally occurring genetic variations must have a significant impact on this cognitive ability. In support of this assumption, several recent behavioral and imaging genetics studies have successfully identified and characterized genetic variations significantly associated with human memory performance. In this chapter, we will focus on episodic memory, but similar developments are taking place for other forms of memory, such as working memory (see Chapter 20 of this volume). In 2003, two genetic factors associated with episodic memory in healthy humans were identified: the Val66Met polymorphism in the gene encoding the brain-derived neurotrophic factor (BDNF) (Egan et al. 2003) and the His452Tyr polymorphism in the gene encoding the serotonin 2A receptor (HTR2A) receptor (de Quervain et al. 2003). Following these two studies, several other polymorphisms related to individual differences in episodic memory performance have been recognized (e.g., polymorphisms in COMT, de Frias et al. 2004; GRM3, Egan et al. 2004; PRNP, Papassotiropoulos et al. 2005b; CHRFAM7A, Dempster et al. 2006; KIBRA, Papassotiropoulos et al. 2006; CAMTA, Huentelman et al. 2007; CPEB3, Vogler et al. 2009; PDYN, Kolsch et al. 2009; CTNNBL1, Papassotiropoulos et al. 2013; BCL1, Ackermann et al. 2013), and the investigation of the genetic basis of human episodic memory is developing in an exciting and rapidly growing research field (see Papassotiropoulos and de Quervain 2011). In addition to and in combination with
purely behavioral measures, several groups have started to use neuroimaging methods to investigate the neural underpinnings of genotype-dependent differences in human episodic memory performance. In this review, we describe how functional imaging technology can be used to validate and extend findings in the field of cognitive genetics of episodic memory, and discuss advantages and limitations of different methodological approaches. We start by briefly reviewing some major neuroanatomical findings with regard to episodic memory. IMAGING E PIS ODIC ME MOR Y Episodic memory refers to memory for particular events or autobiographical episodes in a life of an individual, which includes information about the content of the experience and the spatial and temporal context in which it occurred (Tulving 1983). Episodic memories, together with non-contextual semantic memories, are termed “declarative” or “explicit” memories, because they can be intentionally, consciously, and in most cases verbally assessed (Squire and Zola 1996). Consequently, testing episodic memory usually involves asking the participant to describe a past event or stimulus presented at a certain time in the experimental procedure using free- or cued-recall procedures. Recognition paradigms are sometimes also used, although performance in recognition tests is confounded by correct answers given based on non-episodic feelings of familiarity (Yonelinas 2001; Tulving 1993). Several functional imaging studies have investigated cerebral activation during memory tasks. This large body of evidence suggests that in particular the medial temporal lobe (MTL) but also the frontal and parietal brain areas are involved in episodic memory processes (Squire et al. 2004;
289
Moscovitch et al. 2006; Cabeza et al. 2008; Spaniol et al. 2009). Within the medial temporal lobe, numerous lesion studies indicate that the functional integrity of the hippocampal complex (encompassing the CA fields, the dentate gyrus, the subiculum, and the parahippocampal gyrus) is critical for encoding, as well as for most retrieval processes of episodic memories (Milner 1972; Squire 1992; Nadel et al. 2000; Squire et al. 2004; Moscovitch et al. 2006). Brain activity in the hippocampal complex, as recorded by positron emission tomography (PET) or functional magnetic resonance imaging (fMRI), has been consistently associated with memory encoding and retrieval processes also in the healthy brain (Eichenbaum and Lipton 2008; Schacter and Wagner 1999; Cohen et al. 1999; Cabeza and Nyberg 2000). In particular, higher activity in medial temporal lobe regions during the encoding of episodic memories is typically associated with better memory for the encoded events, as shown by comparing encoding-related brain activity of subsequently remembered versus subsequently forgotten items (subsequent memory effect) (Rugg et al. 2002; Spaniol et al. 2009). During the retrieval phase, successful recollection of old versus correctly rejected new memories is similarly accompanied by increased activity in hippocampal and parahippocampal areas, whereas recognition processes based on feelings of familiarity are less dependent on the hippocampus (Eichenbaum et al. 2007). In addition to MTL regions, lesion studies as well as imaging studies provide evidence for an involvement of medial and lateral prefrontal regions in episodic memory encoding and retrieval (Rugg et al. 2002; Spaniol et al. 2009). The prefrontal cotex (PFC) plays a putative role in memory storage ( Jung et al. 2008), but is also involved in cognitive control processes accompanying episodic memory formation, like stimulus selection and organization, attention direction, and semantization during the encoding and initialization of retrieval search, as well as maintaining and monitoring the search result during the retrieval process (Fletcher and Henson 2001). Activity in prefrontal regions (in particular in the dorsolateral and ventrolateral cortices) is strongly related to successful memory encoding and retrieval (Spaniol et al. 2009). However, higher activity in these regions might also reflect increased task difficulty or recruitment of additional cognitive control resources to achieve a certain performance level. For example, aging is often accompanied by impaired memory performance, but age-related increases as well as decreases in memory-related PFC activity have been reported (Rajah and D’Esposito 2005; Prvulovic et al. 2005; Han et al. 2009). Furthermore, changes in lateralization and reduced hemispheric asymmetry frequently occur in the elderly (Hedden and Gabrieli
2005). Consequently, age-related changes need to be considered in imaging genetics studies of episodic memory, and age should be routinely included as a covariate in the analyses (Papassotiropoulos et al. 2005a). Recently, several authors have implicated parietal regions in episodic memory processes, highlighting the importance of attention–memory interaction for forming and retrieving memory traces (Cabeza et al. 2008; Uncapher and Wagner 2009). Considering emotional aspects, increased activity in the amygdala, orbitofrontal cortex, and insula is associated with the robust memory enhancement induced by emotional stimuli (Cahill and McGaugh 1998; Phelps 2004; LaBar and Cabeza 2006). In sum, a network of brain regions encompassing the medial temporal lobe, prefrontal cortex, and parietal and limbic regions underlies episodic memory encoding and retrieval. It is primarily in this network that we expect genotype-dependent differences in brain activity during episodic memory tasks. IMAGING GE NOT YPE - DE PE NDE NT DIF F E R E NC E S IN E PIS ODIC ME MOR Y The main rationale for combining behavioral and neuroimaging methods in studies investigating genetic polymorphisms of episodic memory is to validate and extend purely behavioral studies by providing insight into the genetic differences in memory processes at the level of neural circuits. There are mainly two strategies of subject selection with regard to memory performance used in imaging genetics studies, which are detailed in the following sections (see Tables 19.1, 19.2, and 19.3 for examples of imaging genetics studies on episodic memory using fMRI). MAT C H E D ME MOR Y PE R F OR MANC E One strategy is to select a subpopulation of participants for fMRI from a larger sample used for behavioral genetics in a way that the genotype groups of the fMRI subpopulation are exactly matched for memory performance. The rationale for this matching procedure is to avoid measuring genotype-unrelated performance effects on brain activations, but to instead capture solely genotype-dependent differences in brain activation patterns. We used this approach in an imaging genetics study of KIBRA (Papassotiropoulos et al. 2006). In that genome-wide association study, we identified a single nucleotide polymorphism (an intronic C to T substitution) in the gene
2 9 0 part V I : I maging G enetics and C ognition
EXAMPLES OF IMAGING GENETIC STUDIES OF EPISODIC MEMORY USING fMRI IN HEALTHY PARTICIPANTS WITH MATCHED MEMORY PERFORMANCE TABLE 19.1
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Egan et al., 2003 Cell
BDNF Val66Met (SNP rs6265) Met/Met homozygotes associated with worse story recall
20 Val/Val 10 Val/Met (two cohorts, mostly European American)
N-Back Working memory task
Matched (for working memory performance)
Val/Met > Val/Val
Bilateral Hippocampus
Abnormal hippocampal activation during a working memory task in Met allele carriers
Huentelmann et al., 2007 Hum. Mol. Gen.
CAMTA1 (rs4908449) T allele associated with poorer immediate verbal memory (words, 5 min)
22 TT/TC 13 CC (European)
Retrieval of faceprofession associations
Matched
TT/TC > CC for cued recall > baseline task
R Hippocampus
Neural compensation in T allele carriers
Egan et al., 2004 PNAS
GRM3 GG higher list learning than A carriers (including schizophrenics, mostly europeanamerican)
32 AA 32 GA/GG (mostly European American)
Encoding and immediate recognition of scenes
Matched
GA/GG > AA
Encoding: R Hippocampus
Greater activity reflects more efficient encoding in G allele carriers
Papassotiropoulos et al., 2006 Nature
KIBRA (rs17070145) T-allele associated with better immediate verbal memory (words, 5 min)
15 TT/TC; 15 CC (European)
Retrieval of faceprofession associations
Matched
CC > TT/TC for cued recall > baseline task
Retrieval: R Hippocampus R Parahippocampus Medial frontal gyrus (BA 6,8,9) Parietal cortex (BA 40)
Neural compensation in CC
Kauppi et al., 2011 J. Neurosci
KIBRA (rs17070145) T-allele associated with better immediate verbal memory (words, 5 min)
51 TT/TC; 32 CC (European, 55–60 years)
Encoding and retrieval of face-name associations
Matched and Unmatched (same results)
TT/TC > CC for cued recall > baseline task
Encoding: –/–
Improved hippocampal functioning in T-allele carriers
PRNP Met129Val Met allele associated with better verbal long-term memory (words, 24 h; Papassotiropoulos et al., 2005)
12 Met/Met, 12 Val/Met, 12 Val/Val (European)
Buchmann et al., 2008 Neuropsychologia
Retrieval: –/–
62 TT/TC; 51 CC (European, 65–75 years)
Retrieval: Bilateral Hippocampus No result in participants aged > 65 years.
Recognition of verbs learned 30 min or 24 h earlier
Matched
Val/Val / Val/ Met > Met/ Met for Hits > false alarms (pair wise comparisons)
Prefrontal cortex Middle temporal gyrus L Hippocampus L Parahippocampus Parietal cortex (BA 40) (effects more pronounced for 30 min recognition than 24 h)
Neural compensation in Val carriers
EXAMPLES OF IMAGING GENETIC STUDIES OF EPISODIC MEMORY USING fMRI IN HEALTHY PARTICIPANTS WITH UNMATCHED MEMORY PERFORMANCE TABLE 19.2
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Hariri et al., 2003 J. Neurosci.
BDNF Val66Met (SNP rs6265) Met /Met homozygotes associated with worse story recall (Egan et al., 2003)
14 Val/Val 14 Val/Met / Met/Met (mostly European American)
Encoding and immediate recognition of neutral scenes
Unmatched, Val/Val better than Met carriers
Val/Val > Met carriers for Encoding vs. Rest and Recognition vs. Rest
Encoding: R Hippocampus R Parahippocampus
Better hippocampal functioning / stronger memory trace in Val/Val
BDNF Val66Met
17 Val/Val 22 Val/Met 12 Met/Met (Asian)
Encoding and immediate recognition of neutral scenes
Unmatched, no significant differences
Number of Val alleles for Encoding vs. Rest and Recognition vs. Rest
Encoding: Bilateral Hippocampus
39 Val/Val 19 Val/Met / Met/Met (European, healthy, at high risk for schizophrenia)
Encoding and immediate recognition of single words
Unmatched, No significant differences
Met carriers > Val/Val for Encoding vs. Rest and correct recognition vs. correct rejection
Encoding: L Inferior occipital gyrus
32 Val/Val 15 Val/Met / Met/Met (European)
Encoding and recognition of male neutral faces
Unmatched, No significant differences
Met carriers > Val/Val for remembered vs. forgotten faces during encoding and recognition
Encoding: Bilateral amygdala (only in males)
Hashimoto et al., 2008 Neurosci. Res.
Baig et al. 2010 Psychiatric Res.
Van Wingen et al. 2010 Neuroimage
BDNF Val66Met
BDNF Val66Met
Retrieval: Bilateral hippocampus Bilateral Parahippocampus
Retrieval: –/–
Better hippocampal functioning with Val allele (genedose dependent) none
Retrieval: Cingulate Gyrus R DLPFC Posterior parietal cortex R hippocampus (trend)
Retrieval: L inferior frontal gyrus Posterior cingulate cortex (only in males)
Compensatory effect in the amygdala in Met carriers to reach memory performance of Val/Val; effect is genderdependent
Sambataro et al. 2010 Mol. Psychiatry
BDNF Val66Met
80 Val/Val 45 Val/Met / Met/Met (Caucasian, 19–85 years)
Encoding and immediate recognition of neutral scenes
Unmatched, No significant differences
Age-related correlation for Encoding vs. Rest and Recognition vs. Rest
Encoding + retrieval: Hippocampus: Stronger age-related decline in Met carriers
Val/Val more resilient to agerelated changes in declarative memory
Dennis et al., 2011 Hippocampus
BDNF Val66Met
11 Val/Val 11 Val/Met / Met/Met (Caucasian + Asian)
Task 1: Encoding and immediate recognition of neutral scenes
Unmatched, No significant differences
Met carriers > Val/Val for Encoding vs. Baseline and retrieval vs. Baseline
Encoding R hippocampus (Task 1 + 2) L parahippocampus (T1)
Met carriers require more MTL activity to reach memory performance of Val/Val
Task 2: Memory for facescene pairings
Met carriers > Val/Val for subsequently remembered vs. forgotten pairs
Retrieval: R post. Hippocampus (T1 + 2) Bilateral ant. Hippocampus (T2)
Alternative: Increased MTL functioning in Met carriers in spite of similar performance
TABLE 19.2
CONTINUED
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Kauppi al., 2013 Neuropsychologia
BDNF Val66Met
122 Val/Val 72 Val/Met / Met/Met (European, 55–75 years)
Encoding and cued recall of Face-Name Paired associates
Unmatched, No significant differences
Val/Val > Met carriers for Encoding vs. Baseline and retrieval vs. Baseline
Encoding: Bilateral Parahippocampus (age-related decline not replicated) Retrieval: –/–
Better MTL functioning in Val/Val But: no differences in episodic memory in behavioral sample (N = 2229)
Fera et al., 2013 Plos One
BDNF Val66Met
11 Val/Val 14 Val/Met / Met/Met (European, healthy, additional groups with multiple sclerosis reported in the paper)
Encoding and immediate recognition of neutral scenes
Unmatched, No significant differences
Met carriers > Val/Val for Encoding vs. Baseline and retrieval vs. Baseline
Encoding + retrieval L posterior hippocampus Bilateral parahippocampus L posterior cingulate
Increased MTL functioning in healthy Met carriers (opposite results for patients with multiple sclerosis)
Mascetti et al., 2013 J. Neurosci.
BDNF Val66Met
14 Val/Val 15 Val/Met / Met/Met (European)
Encoding and delayed recognition of neutral faces (delay 1: 1 h; delay 2: 16 h including sleep)
Unmatched, Val/Val better recognition than Met allele carriers at delay 1 + 2 (larger at delay 2)
Met carriers > Val/Val during delayed testing 1 + 2 for remembered hits vs. know / guess hits
Delayed retrieval 1: R intraparietal sulcus
Val/Val carriers show more system consolidation during sleep as Met carriers
57 G/G 43 G/A 16 A/A (EuropeanAmerican)
Encoding and immediate recognition of neutral and aversive scenes
Unmatched, no significant differences
A/A > G carriers for encoding of aversive pictures
Encoding: Bilateral hippocampus
–/–
60 G/G 41 G/A 9 A/A (European)
Encoding, free recall and recognition of faceprofession pairs
Unmatched, no significant differences
G/G > A carriers for Memory processes vs. control condition
Recall: Bilateral hippocampus perigenual anterior cingulate cortex Middle / sup. temporal gyrus (no effects during encoding and recognition)
Impairment of hippocampal functioning in risk allele carriers important for bipolar disorders.
Bigos et al., 2010 Arch Gen Psychiatry
CACNA1C Rs1006737 (Risk A-allele associated with affective disorders and schizophrenia)
Erk et al., 2011 Arch Gen Psychiatry
CACNA1C Rs1006737
Change from delay 1 to 2: Stronger increases in Val/Val L angular gyrus Bilateral occipital gyri Biltateral intraparietal sulcus
Connectivity Increased coupling between hippocampi (continued)
TABLE 19.2
CONTINUED
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Krug et al., 2014 Eur Arch Psychiatry Clin Neurosci
CACNA1C Rs1006737
Sample 1: 43 G/G 51 A-allele carriers
Encoding and recognition of neutral faces
Unmatched, Sample 1: no significant differences
G/G > A carriers for Encoding Vs. Rest And Recognition Vs. Rest
Encoding: R Hippocampus (S1) Anterior Cingulate (S1 +; S2 -) …several inconsistent results between Sample 1 & 2)
Rs1006737 associated with hippocampal memory system in healthy subjects
Sample 2: 58 G/G 53 A-allele carriers (European)
Sample 2: Better recognition for G/G vs. A carriers
Retrieval: R Hippocampus (S1 & S2) Bertolino et al., 2006 Biological Psychiatry
Schott et al., 2006 J. Neurosci.
COMT Val158Met Met/Met associated with better episodic & semantic memory, no effect for recognition (de Frias et al., 2004)
9 Met/Met 9 Val/Met 9 Val/Val (European)
COMT Val158Met (VNTR in DAT1 also investigated)
17 Met/Met 17 Val/Met 15 Val/Val (European)
Encoding and immediate recognition of neutral scenes
Unmatched, Val/Val worse than Met allele carriers
Encoding and free recall of words
Unmatched, no significant differences
Number of Met alleles for Encoding Vs. Rest And Recognition Vs. Rest
Encoding: L Hippocampus (+) R VLPFC (BA44) (-)
Val/Val > Met/Met for remembered vs. forgotten words (subsequent memory effect)
R Prefrontal (BA 6,47) Cingulate cortex (BA 32) L Fusiform (BA 19) Medial occipital (BA 17,18)
Retrieval: L Hippocampus (+) R VLPFC (BA 44) (-)
Better hippocampal functioning / less involvement of VLPFC with Met allele (genedose dependent)
Compensatory mechanism in frontal regions in Val/Val; higher efficiency and prefrontal-hipp. coupling Met/Met
Functional coupling between L Hippocampus & Prefrontal cortex (stronger in Met/Met) Bertolino et al., 2008 Biological Psychiatry
COMT Val158Met VNTR in DAT 3’
21 Met/Met 43 Val/Met 18 Val/Val (Caucasian) 51 9 repeat carriers 31 10/10 repeat
Encoding and immediate recognition of neutral scenes
Unmatched, no significant differences
Number of Met alleles for Encoding Vs. Rest 10/10 > 9 repeat carriers for Encoding Vs. Rest
Encoding: Main effect COMT Bilateral Hippocampus (+) R VLPFC (BA44) (-) Main effect DAT L anterior hippocampus (+) Interaction COMT x DAT Bilateral Hippocampus R VLPFC (BA44)
Better hippocampal functioning with number of Met alleles only in DAT 9 repeat carriers, opposite in DAT 10/10.
TABLE 19.2
CONTINUED
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
DiGiorgio et al., 2011 Psychological Medicine
COMT Val158Met
7 Met/Met 20 Val/Met 6 Val/Val (Caucasian, healthy, 28 patients with schizophrenia also reported)
Encoding and immediate recognition of neutral scenes
Unmatched, no significant differences
Number of Met alleles for Encoding Vs. Rest
Encoding: Bilateral parahippocampus (+)
Better parahippocampal functioning with Met allele (genedose dependent) Opposite result in patients
Erk et al., 2011 J. Neurosci
CLU / APOJ (rs1111360000)
36 CC 52 CT 21 TT (European)
Free recall and recognition of faceprofession pairs.
Unmatched no significant differences
Number of C alleles with BOLD activity during recall / recognition vs. baseline as well as hippocampal coupling.
BOLD contrasts: No significant differences
Risk allele associated with reduced hippocampal-DLPF coupling relevant for memory retrieval
290 AA 29 AG 3 GG (European)
Encoding and free recall of neutral and emotional scenes
Unmatched no significant differences
Number of A alleles for subsequently remembered vs. forgotten pictures (all valences)
Encoding: R parahippocampal gyrus Medial frontal gyrus
Greater activity associated with better memory functioning in A allele carriers
Encoding L middle frontal (BA 11) Cuneus (BA 17)
Higher effort / different cognitive strategies to compensate for mild deficits in risk allele carriers
Risk C allele associated with Alzheimer’s disease
Papassotiropoulos et al., 2013 Mol. Psychiatry
Thimm et al., 2010 Human Brain Mapping
Goldberg et al., 2006 Neuropsychopharmacology
CTNNBL1 (rs7363432) Number of G alleles associated with improved episodic memory in 3 independent samples
DTNBP1 (rs1018381) Relation to episodic memory performance unknown
G72 SNP 10 TT homozygote schizophrenics have poorer verbal memory performance
Coupling: Reduced coupling between hippocampus and R DLPFC
AA > G-allele carriers revealed similar results 29 risk allele carriers, 55 non carriers
Encoding and recognition of faces
Unmatched, no significant differences
Risk > no Risk
Retrieval R Medial / inferior frontal (BA 9) R Parietal (BA 40) 7 TT 7 AA/AT (mostly European American)
Encoding and immediate recognition of aversive and neutral scenes
Unmatched no significant differences
AA/AT > TT
Encoding (neutral): L Parahippocampus Encoding (aversive) L Parahippocampus R Hippocampus
Greater activity associated with better memory functioning in A allele carriers
Retrieval (aversive): R Parahippocampus (continued)
TABLE 19.2
CONTINUED
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Heck et al., 2011 PlosOne
KCNH5 & KCNB2 (rs243146 & rs7006287)
9 A/A 11 A/G 3 G/G (European, all G/G homocygotes of KCNB2)
Encoding and free recall of neutral and emotional scenes
Unmatched, no significant differences
Number of A alleles during encoding of scenes vs. scrambled pictures
Encoding: L Parahippocampus Superior temporal gyrus Inferior / medial frontal cortex Medial occipital gyrus
Interacting genetic variants relate to memory performance and parahippocampal functioning
34 T/T 32 T/C 28 C/C (European)
Encoding and recognition of neutral faces
Unmatched, no significant differences
Number of C alleles during encoding vs. baseline and recognition vs. baseline
Encoding: R Cingulate gyrus Fusiform gyrus Middle frontal gyrus Middle occipital gyrus R sup. Temporal gyrus
NRG1 has an influence on brain activations in key areas of episodic memory processes
Improved memory with number of A-alleles of KCNH5 only in G/G homocygotes of KCNB2 in 2 independent samples Krug et al., 2010 Neuroimage
NRG1 (rs35753505) NRG1 associated with schizophrenia
Retrieval: Middle occipital gyrus Krug et al., 2013 Schizophrenia Bulletin
NRGN (rs12807809) NRGN associated with schizophrenia
67 T/T 27 C allele carriers (European)
Encoding and recognition of neutral faces
Unmatched, Better recognition in T/T
TT > C carriers during encoding vs. baseline and recognition vs. baseline (performance included as covariate)
Encoding: L lingual gyrus Anterior cingulate cortex Retrieval: L precentral gyrus R cingulate gyrus L insula
NRGN does not influence hippocampal but cortical areas associated with memory processes.
EXAMPLES OF IMAGING GENETIC STUDIES OF EMOTIONAL EPISODIC MEMORY USING fMRI IN HEALTHY PARTICIPANTS WITH UNMATCHED MEMORY PERFORMANCE TABLE 19.3
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Rasch et al., 2009 PNAS
ADRA2B Carriers of the deletion variant associated with better emotional memory (de Quervain et al., 2007)
30 deletion carriers 27 non carriers (European)
Encoding and free recall of neutral and emotional scenes
Unmatched, no significant differences
Deletion carriers > non-carriers for Negative > Neutral
Encoding: R Amygdala L Insula L Inf. parietal (BA 40) R Sup. Temporal (BA 38)
Higher amygdala responsivity and amygdala-insula interaction lead to stronger influence of emotional arousal on memory
Increased amygdala-insula coupling
TABLE 19.3
CONTINUED
Examples of research studies
Polymorphism and its relation to memory performance
Number of subjects
Task
Memory performance
Key contrast
Key regions
Interpretation with regard to memory
Urner et al., 2011 Human Brain Mapping
ADRA2B
28 deletion carriers 23 non carriers (European)
Encoding and recognition of positive and negative scenes (no neutral scenes)
Unmatched, no significant differences
Deletion carriers > non-carriers for remembered vs. forgotten scenes (encoding) or hits vs. misses (recognition) (irrespective of valence)
Encoding: L Amygdala L inferior frontal gyrus
ADRA2B genotype modulates emotional memory formation during encoding
234 AA 139 AG 21 GG (European)
Encoding and free recall of neutral and emotional scenes
Unmatched, Better memory with higher number of A alleles
Number of A alleles for subsequently remembered vs. forgotten negative pictures
Encoding: Middle frontal gyrus Inferior frontal gyrus Superior frontal gyrus
De Quervain et al., 2012 PNAS
PRKCA (rs 7496904) Number of A alleles associated with better memory for pictures in 2 independent samples and with increased risk for PTSD
Retrieval –/–
(no effects for positive and neutral pictures)
Higher prefrontal activity implicated in better memory (and increased risk for PTSD) with number of A alleles, without any differences in hippocampus or amygdala
encoding KIBRA as significantly associated with recall of words in three independent samples (Figure 19.1A). The association of this SNP with episodic memory performed has now been confirmed in several other studies (see Milnik et al. 2012 for a meta-analysis). In a subsequent fMRI Study, 15 carriers of the T allele, which was associated with better episodic memory in the three large samples, and 15 non-carriers were selected, so that genotype groups were matched according to their memory performances (p= 1). During the retrieval phase in the fMRI, non-carriers of the T allele exhibit increased activity in the hippocampus and parahippocampal gyrus (Figure 19.1B), as well as in the medial frontal gyrus and inferior parietal gyrus. All these regions are part of the network involved in memory retrieval. Therefore, these findings suggested that non-carriers of the T allele need more activation in these memory retrieval–related brain regions to reach the same level of retrieval performance as T allele carriers. Contrasting with this conclusion (and in spite of a replication of the behavioral effects), reduced hippocampal activation in non-carriers of the T allele during episodic memory retrieval was revealed in older participants (aged 55–65 years), and both matching and not matching the groups for differences in memory performance did not alter the results (Kauppi et al. 2011). No genotype-dependent differences in brain activity were observed in participants aged > 65 years in this
study, suggesting a strong age dependency of the effects of genetic variations in KIBRA on memory-related brain activity. Huentelman and coauthors (2007) also applied a matching approach. After identifying an association between a polymorphism in the gene encoding calmodulin-binding transcription activator 1 (CAMTA1) and episodic memory performance in two independent samples, the additional fMRI study in genotype groups matched for memory performance revealed increased activity in medial temporal lobe regions during retrieval testing in carriers of the allele associated with poorer memory performance. A third example is the study by Buchmann et al. (2008), who investigated differences in retrieval-related brain activity associated with genetic variations of the prion protein gene (PRNP). Participants were selected from a previous study that reported a link between these polymorphisms and recall of word pairs after 24 hours, but not after 5 minutes, suggesting a role for PRNP in the long-term consolidation of episodic memories (Papassotiropoulos et al. 2005b). After matching genotype groups for 24-hour recall, carriers of the genetic variant associated with poorer episodic memory retrieval had increased brain activity in the hippocampus, middle temporal gyrus, and middle and inferior frontal gyrus, suggesting that also in that study, carriers of the genetic variant associated with poorer performance needed more activity to achieve the same level of performance.
chapter 1 9 : I maging G enetics of E pisodic M emory
297
12
Episodic memory recall
10
(b)
SNP rs17070145 (KIBRA) **
L
R
** CC CT/TT
8
6 5 4
**
6
3 4
H
2
P
2
t value
(a)
1 0
0 Cohort I
Cohort II
Cohort III
Figure 19.1 KIBRA and episodic memory.
(A) A single-nucleotide polymorphism (SNP) in the KIBRA gene was significantly associated with episodic memory performance in a genome-wide association analysis (Cohort I). The effect was replicated in two independent cohorts, II and III. (B) After matching subjects for performance, non-carriers of the T allele, who had poorer episodic memory in the large samples, exhibited increased activity in the hippocampus (H) and parahippocampal gyrus (P), as compared to T allele carriers, during memory retrieval. This finding suggests that non-carriers need more activation in these memory-related brain regions to reach the same level of memory performance as T allele carriers. Adapted from Papassotiropoulos et al. (2006).
In sum, comparing brain activity during retrieval in genotype groups matched for memory performance consistently reveals increased activity in memory-related brain regions for the genotypes associated with poorer episodic memory (but see Egan et al. 2004 for discrepant results and interpretation). A careful matching of the genotype groups according to memory performance is a prerequisite for this approach. The advantage of such a matching procedure is that differences in brain activity can be fully attributed to genetic differences (i.e., genotype-unrelated performance effects on brain activations are avoided). The disadvantage is that the matching procedure prevents measuring genotype-dependent differences in brain activity that are actually related to the genotype-dependent differences in behavior. For that purpose, using performance-unmatched genotype groups is advantageous. UNM ATC HED MEMO RY PERFO RMAN CE Several imaging genetics studies examined memory-related brain activity without matching genotype groups for performance. For example, Hariri and colleagues (2003) recorded brain activity in 28 healthy participants during encoding and recognition of complex visual scenes and investigated the role of the memory-related Val66Met polymorphism in the gene encoding BDNF (Egan et al. 2003). On the behavioral level, Val homozygotes (as compared to Met allele carriers) were significantly more accurate at recognizing encoded scenes. With fMRI, the
authors observed increased brain activity in the posterior hippocampal complex during encoding as well as recognition testing for Val homozygotes as compared to Met allele carriers. The interaction between the Val66Met polymorphism and hippocampal activity accounted for 25% of the variance in recognition performance. Using basically the same paradigm, Hashimoto and others (2008) examined dose-dependent effects of the Val66Met polymorphism in 58 Japanese subjects. They similarly reported a significant positive association with the number of Val alleles in the right parahippocampus and bilateral hippocampus during encoding, although with more anterior peak activations. In contrast to the previous study, no reliable genotype effect on medial-temporal activity during retrieval and no genotype effect on memory performance were observed. In further studies, Val/Val homozygotes showed a reduced age-related decline in hippocampal activity during encoding and retrieval (Sambataro et al. 2010) and exhibited stronger activity changes in memory-related cortical areas after a consolidation period including sleep (Mascetti et al. 2013). In a recent study including 194 healthy participants aged 55–75 years, Val/Val homocygotes exhibited an increased parahippocampal activation only during encoding, while the increased resilience against an age-related decline in hippocampal activity was not replicated (Kauppi et al. 2013). The observed activity increases in hippocampal areas in Val homozygotes during encoding might indicate a generally higher sensitivity and responsivity of the medial temporal lobe network to external cues, which is in line with the known role of BDNF in hippocampal plasticity
2 9 8 part V I : I maging G enetics and C ognition
(Poo 2001). In contrast to this notion, other fMRI studies reported either no effect or even obtained an opposite results pattern, that is, increased activation in MTL and/ or other memory-related brain regions in Met allele carriers (Baig et al. 2010; van Wingen et al. 2010; Dennis et al. 2011; Fera et al. 2013). However, the latter studies contained either relatively small study samples or examined healthy participants at high genetic risk for schizophrenia (Baig et al. 2010). In view of the inconsistent results, the understanding of the effects of the Val66Met BDNF polymorphism on activity in memory-related brain regions and its relation to episodic memory performance require further examination. Other examples for genotype-dependent activity differences in the memory network in performance-unmatched fMRI samples are provided with regard to genetic variations in the dopaminergic system. Dopamine availability is modulated by the activity of the catalyzing enzyme catechol-Omethyltransferase (COMT), and the gene encoding COMT harbors a functional Val158Met polymorphism that alters enzyme activity (Lachman et al. 1996). In a cognitive genetics study in 286 healthy subjects, Met homozygotes (having the low activity enzyme and putatively higher dopamine availability) had higher episodic and semantic memory scores as compare to Val allele carriers (de Frias et al. 2004). In an independent fMRI study (n = 27) (Bertolino et al. 2006), the Met allele was associated with increased recognition performance and with increased activity within the hippocampal complex during encoding and recognition of complex visual scenes as compared to Val allele carriers, together with reduced activity in ventrolateral prefrontal regions. Furthermore, task-independent coupling between the hippocampal complex and the ventro lateral prefrontal cortex over time was less negative in Met allele carriers, and the strength of negative coupling predicted recognition performance. Another fMRI study (n = 49) also observed changes in prefrontal activity as well as functional coupling between hippocampal and prefrontal regions depending on COMT Val158Met genotype (Schott et al. 2006). In contrast, neither genotype-dependent increases in hippocampal activity nor differences in recognition performance were observed in this study. Two other studies replicated the general increase in hippocampal or parahippocampal activity during encoding with the number of Met alleles (Bertolino et al. 2008; Di Giorgio et al. 2011), one of them specifying that this effect was particularly pronounced in carriers of the 9-repeat variant in the DAT 3 polymorphism (Bertolino et al. 2008). The genotype-dependent differences in MTL activation and hippocampal-prefrontal coupling together with the reduced prefrontal activity in Met allele carriers might
reflect reduced effort to encode and retrieve the information, possibly indicating a higher efficiency of memory formation processes in Met allele carriers of the COMT Val158Met polymorphism. Using unbiased genome-wide screening, we recently identified CTNNBL1 as a gene significantly associated with human memory (Papassotiropoulos et al. 2013). The SNP rs1698690 in CTNNBL1 (encoding beta-catenin-like protein 1) was significantly associated with performance in a verbal episodic memory task in two independent samples (n = 1198 and n = 524), revealing improved memory for freely recalled words with increased number of G alleles. The number of G alleles was also significantly associated with memory for pictures in a subsample (n = 872). An additional fMRI study including 322 participants revealed increased activation in the right parahippocampus with increased number of G alleles during successful encoding of pictures (subsequently remembered vs. forgotten pictures). A similar result pattern was observed in the medial frontal cortex/anterior cingulate. Although the sample size was remarkably large for MR standards, the genotype-dependent behavioral difference in episodic memory observed in the even larger behavioral sample was not significant in the fMRI sample, possibly due to a lack of statistical power (see further discussion later in this chapter). Similarly inspired by genome-wide association studies conducted not on episodic memory but psychiatric diseases, Erk and colleagues (2010) have examined the role of a polymorphism (rs1006737) in CACNA1C for hippocampal functioning during episodic memory processes. CACNA1C variants have been robustly associated with multiple psychiatric disorders, including bipolar disorder, major depression, and schizophrenia (Ferreira et al. 2008; Liu et al. 2011; Smoller et al. 2013), and CACNA1Cmediated L-type voltage-dependent calcium channels have been associated with processes of synaptic plasticity in the hippocampus underlying episodic memory (Moosmang et al. 2005). In a total of 110 healthy volunteers, Erk and colleagues (2010) reported a pronounced reduction in hippocampal activity during recall of face-profession pairs in carriers of the risk allele (A allele). Risk allele carriers had also reduced activity in the perigenual anterior cingulate cortex, a region involved in monitoring and selecting retrieval information for task-appropriate responses and inhibiting inappropriate information (Nieuwenhuis and Takashima 2011). These findings had been replicated by Erk et al. in an independent sample of 179 healthy subjects (Erk et al. 2014). Furthermore, functional coupling between the left and right hippocampus was significantly diminished in risk allele carriers. No effects were observed during encoding and
chapter 1 9 : I maging G enetics of E pisodic M emory
299
recognition. Contrasting these findings, Bigos et al. (2010) observed a trend for increased bilateral hippocampal activation in risk allele carriers during encoding and immediate recognition of aversive and neutral scenes. Recently, Krug and colleagues (2014) confirmed the findings by Erk and colleagues (2010, 2014) and reported a consistent reduction in hippocampal activity in A allele carriers during recognition of neutral faces. Importantly, Krug et al. (2014) also used a replication approach, reporting imaging results from two independent fMRI samples (n = 94 and n = 111). While the effect on hippocampal activity during recognition was consistently observed in both samples, several other genotype-dependent differences in memory-related brain activity differed between the two samples and were possibly false positives. The result pattern underlines the importance of including replication samples in imaging genetics studies. Interestingly, genotype-dependent differences in memory performance were only observed in one of these two samples. Genetics can also affect memory by influencing modulators of episodic memory. For example, emotional arousal is well known to enhance episodic memories, which has obvious adaptive value in evolutionary terms, but can also be maladaptive in the context of aversive experiences as a starting point for the development of anxiety disorders, such as post-traumatic stress disorder (PTSD) (de Quervain et al. 2009). In a recent study, a functional polymorphism in the gene encoding the alpha-2B-adrenergic receptor (ADRA2B) was significantly associated with strength of emotional memories in healthy humans and with the strength of traumatic memories in war victims (de Quervain et al. 2007). A subsequent fMRI study (n = 57, unmatched) revealed that carriers of the genetic variant associated with enhanced emotional memories exhibited increased amygdala activation in response to negative pictures (Rasch et al. 2009). In addition, functional coupling between the amygdala and the insula, brain regions that are known to be hyperactive in PTSD (Liberzon & Martis 2006), was significantly enhanced in this genotype group. Furthermore, deletion carriers also exhibited increased amygdala activation during successful memory encoding of negative and positive pictures (i.e., subsequently remembered vs. forgotten pictures, n = 51), but not during memory retrieval (Urner et al. 2011). In both studies, however, emotional memory performance did not differ significantly between genotype groups. The ADRA2B genotype–dependent amygdala responsivity and interaction with the insula during the acquisition of memories may affect the strength of emotional and traumatic memories in these genotype groups, which might have important implications for PTSD.
Recently, we directly examined whether genetic variation in protein kinases are implicated in processes of emotional and traumatic memories (de Quervain et al. 2012). The SNP rs4790904 in PRKCA (encoding protein kinase Cα) was significantly associated with memory for emotional pictures in a behavioral sample of 723 healthy volunteers, revealing better free recall performance with an increasing number of A alleles. In addition, the number of A alleles was significantly correlated with PTSD-related symptoms as well as risk for PTSD in a separate sample of survivors of the Rwandan civil war (n = 347). In a separate fMRI experiment (n = 394), the number of A alleles of SNP rs4790904 was significantly associated with increased activation in lateral and medial prefrontal brain regions during successful encoding of aversive pictures. No genotype-dependent differences were observed in the amygdala or hippocampus. The genotype-dependent differences in brain activity were accompanied by significantly better memory for emotional pictures with an increasing number of A alleles, replicating the results observed in the behavioral sample. Taken together, numerous imaging genetics studies— using either performance-matched or unmatched genotype groups—observe significant differences in brain activity as well as functional coupling between brain regions involved in memory processing. While genotype-dependent activity differences in MTL regions in performance-matched studies are generally interpreted as compensatory activity, similar increases in unmatched genotype groups are often interpreted as improved processing underlying memory formation, even though these findings are not always paralleled by significant increases in performance. GE NE T IC C OMPLE X IT Y OF E PIS ODIC M E M O R Y It is important to emphasize at this point that the described findings capture only a very small part of the genetic complexity underlying human episodic memory. Human memory is a polygenic trait and depends on gene ´ gene and gene ´ environment interactions, which requires moving from single SNP analyses to more sophisticated methodology. Some studies have started to include more than one genetic variant in fMRI studies of human memory and to investigate their genetic interaction (e.g., Schott et al. 2006, 2014). In a recent study we applied a compound genetic analysis (de Quervain and Papassotiropoulos 2006), which goes beyond the traditional SNP by SNP comparison. In a first step, we identified a cluster of seven genetic variations (i.e., polymorphisms and haplotypes) that were highly significantly associated with episodic memory performance
3 0 0 part V I : I maging G enetics and C ognition
in 302 healthy subjects (Figure 19.2A) and generated an individual’s compound genetic score, which correlated positively with individual memory performance. In an independent fMRI experiment, the individual genetic score also correlated positively with brain activity in the hippocampus and parahippocampal gyrus during the encoding phase (Figure 19.2B and C). Using genetic scores instead of single SNP comparisons may greatly increase the sensitivity and power of imaging genetic studies, because this approach accounts better for the high genetic complexity of human episodic memory. Similar approaches have been successfully applied in the context of Alzheimer’s disease (Papassotiropoulos et al. 2005c; Reiman et al. 2005; Reiman 2007). In addition to these analytical tools, exciting new high-throughput technologies allow for the initiation of genome-wide association analyses, which represent an unbiased approach of identifying association with a certain phenotype, for example hippocampal activity during memory processing, at high genomic resolution (Potkin et al. 2009). However, large study samples are required in order to reliably apply this method.
the coupling between neural activity and the blood-oxygenlevel dependent (BOLD) response. Second, morphological changes in certain brain regions may account for differences in the strength and extent of a signal. Third, activity differences may be due to increased variance or noise of the BOLD signal depending on the genotype. Samanez-Larkin and D’Esposito (2008) have recommended using interaction or parametric designs instead of simple comparisons of activity, as well as including appropriate control tests (e.g., short, simple visual or motor tasks) to control for possible hemodynamic differences between groups. Morphological differences should be analyzed and accounted for by selecting the best normalizing algorithm and individual adjustment of regions of interest. Testing for variance equality as well as reporting the time course of activation can help to clarify the contribution of increased noise or changes in shape, height, or latency of the BOLD signal to the observed differences between genotypes.
GENERAL PROBLEMS OF GROUP COMPARISONS IN GENETIC IMAGING STUDIES
In addition to the general problems of group comparisons, the interpretation of genotype-dependent activity differences in the context of memory processing should be done with caution. Increased activity in the medial temporal lobe during memory tasks can either reflect deeper and improved encoding of events or compensatory activity due to recruitment of additional neural resources or slower reaction times. Also, in the prefrontal cortex, higher activity is
Differences in brain activity between genotype groups (or any other groups) may be confounded by several factors (see Samanez-Larkin and D’Esposito 2008; Han et al. 2009, for reviews). First, brain activity differences between groups may be due to changes in hemodynamics altering (a)
NE U R AL C OMPE NS AT ION V E R S U S E NC ODING E F F IC IE NC Y
(b)
(c) 1.5
20
MTL
7 6
15
5 4
10
3 2
5
1 0
Response at [24, –12, –20]
Cluster P = 0.00008
T value
% contribution to gene cluster
25
1 0.5 0 –0.5
PRKCA
GRM3
GRIN2A
GRIN2B
CAMK2G
PRKACG
ADCY8
0 –1 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 IMAGS
Figure 19.2 Prediction of memory performance and brain activity by a gene score.
(A) A cluster of seven genetic variations (i.e., polymorphisms and haplotypes) was significantly associated with episodic memory performance. The seven-SNP cluster was used for the calculation of an individual’s memory-related genetic score, termed individual memory-associated genetic score (IMAGS). (B) Regression analysis revealed a significant positive correlation between the IMAGS and learning-induced brain activations in the medial temporal lobe (MTL), including the hippocampus and parahippocampal gyrus. (C). Scatter plot illustrating the positive correlations between IMAGS and learning-induced brain activations in the hippocampus at coordinate position [24 12 20]. Adapted from de Quervain and Papassotiropoulos (2006).
chapter 1 9 : I maging G enetics of E pisodic M emory
301
related to successful encoding, but might reflect increased cognitive control, which is indicative of less efficient processing. Thus, taking into account genotype-dependent differences in memory performance is absolutely crucial for reliably interpreting the results of imaging genetics studies in the context of memory. In the case of genotype groups matched for performance, activity increases for the low-memory genotype can be interpreted as compensatory activity to achieve the same level of performance as the high-memory group. In theory, such an activity difference should disappear or go in the opposite direction when the same genotypes are compared without prior matching. In contrast, increased activity for the high-memory genotype is more difficult to explain in performance-matched groups: Is this activity unrelated to memory? Or does it reflect deeper processing, which does not translate in differences in performance? The use of reaction time analysis in addition to recall success might be one way to increase the sensitivity of memory measures on the behavioral level. Also, long-term (> 24 h) as compared to short-term recall could lead to more pronounced differences between genotypes with regard to episodic memory performance, because of the reduced influence of short-term/working memory processes on successful retrieval. In the case of significant performance differences between genotype groups examined with fMRI, one would expect higher activity in memory-related brain regions in the high-memory genotype. Here, the question remains whether increased activity is actually related to the differences in genetics, or is due to the differences in performance between genotype groups. If sufficient participants are available, a reanalysis using post hoc matching according to performance, or including performance as a covariate, might help to answer this question (Murphy and Garavan 2004). Compensatory increases in brain activity might also occur in groups with different performance levels. For example in the elderly, more extended MTL activation and/ or recruitment of additional prefrontal regions are observed during memory tasks in spite of lower memory scores as compared to young controls (Rajah and D’Esposito 2005; Han et al. 2009). However, decreased memory-related activity in the elderly, paralleling the age-related changes in performance, have been similarly reported (Prvulovic et al. 2005). To account for these apparent discrepancies, it has been proposed that activity of a less efficient processing unit can only increase until its maximum processing capacity is reached. Thus, compensatory increases occur only when task difficulty is low and sufficient “reserve” capacity is available. When processing capacity is reduced (e.g., by
degeneration) or task difficulty is too high, decreased brain activity occurs (see Prvulovic et al. 2005). As a consequence, tasks with various difficulty levels should be used in imaging genetics studies of episodic memory to map the course of compensatory activity and its limits more accurately. The most problematic case for interpretation occurs when genotype-dependent activity differences are reported in performance-unmatched groups, which are accompanied by non-significant differences in memory performance. We will turn to this problem in the next section. T H E C ONS E QU E NC E S OF DIF F E R E NT S E NS IT IV IT Y OF IMAGING V E R S U S B E H AV IOR AL GE NE T IC S S T U DIE S Interestingly, the number of subjects used in imaging genetics studies that found significant genotype-dependent differences in brain activity typically lies between 20 and 60 subjects, whereas behavioral genetics studies usually use hundreds of subjects to consistently produce significant results (Figure 19.3). A possible explanation for this observation is that biological phenotypes like neural activity are more proximate to the direct effects of functional genetic polymorphisms on gene products and their function, and might therefore be more sensitive in estimating genotype-dependent differences in mental processing (Hariri et al. 2006; Mattay et al. 2008). Thus, the size of the effects of genetic variations on brain activity seems to be much larger as compared to behavioral measures. Consequently, in imaging genetics studies of episodic memory, fewer participants are required to achieve enough statistical power to detect a significant genotype effect than in behavioral genetics studies. Although the increased sensitivity of neuroimaging parameters for gene effects is an advantage, it can also cause problems. While small sample sizes may be sufficient to detect significant genotype effect on brain activity, they are insufficient in reliably replicating genotype-effects on memory performance in the same sample. Thus, it is very likely that significant genotype-dependent differences in brain activity are reported in small fMRI samples, which are accompanied by non-significant differences in memory performance. Statistical considerations imply that the probability of detecting a certain effect (i.e., obtain a significant result) depends on the effect size, the sample size, and the chosen significance level (Cohen 1988). The lower the effect size, the more subjects are required to detect the effect. For example, the size of the effect of the BDNF Val66Met polymorphism on memory performance (as estimated
3 0 2 part V I : I maging G enetics and C ognition
Recall
> 100 subjects needed to reliably detect significant genotype–dependent differences in memory performance(a)
Memory performance 20–60 subjects needed to reliably detect significant genotype–dependent differences in brain activity(b)
Brain circuits Naturally occuring genetic variations, e.g., single nucleotide polymorphisms (SNPs)
Genes Figure 19.3 Levels of analysis in genetic studies of episodic memory.
At the level of genes, subjects are genotyped with regard to naturally occurring genetic variations in the human genome. On the level of brain circuits, neuroimaging is used to examine genotype-dependent differences in brain activity or functional coupling between brain regions; 20–60 subjects are typically sufficient to detect differences for certain memory-related polymorphisms (a). On the level of memory performance, usually hundreds of subjects are required to consistently produce significant results, suggesting that genotype effects are smaller on the level of performance as compared to brain circuits (b). Reprinted from NeuroImage, 53, Rasch B, Papassotiropoulos A, Quervain, D-F de, Imaging genetics of cognitive functions: Focus on episodic memory, 870–877, Copyright (2010), with permission from Elsevier.
from behavioral genetics studies) is approximately η2 = 7%1 (i.e., 7% of the sample variance in memory performance is explained by differences in BDNF genotype; Egan et al. 2003; Dempster et al. 2005; Miyajima et al. 2008). How high is the chance to replicate the BDNF effect of η2 = 7% on memory behavior with 50 subjects? As indicated in Figure 19.4, the probability of detecting such an effect is as low as 37% (see, e.g., software G*Power3, Faul et al., 2007, for calculation of effect sizes and statistical power). Importantly, these probabilities are calculated under the assumption that a true genotype effect of η2 = 7% on memory performance exists in the population. Effects of single genetic variations on behavioral measures of memory are 1 η2 = (Sum of squareseffect / Sum of squarestotal). η2 indicates the amount of explained variance of the effect under investigation relative to the total variance on the level of the sample. The interpretation is similar to the effect size r2 used in regression analyses. We recommend using η2 as compared to partial η2p offered by several statistical software packages, because η2p—values can strongly overestimate the effect when more than one factor is included in the ANOVA; Levine and Hullet (2002). For t-test or one-factorial ANOVAs, η2 and η2p do not differ.
typically even smaller (e.g., η2 = 2% for COMT de Frias et al. 2004). The described statistical phenomenon can be readily observed in the literature of imaging genetics of episodic memory. In spite of numerous reports of genotype-dependent activity differences in memory-related brain regions (see Tables 19.1, 19.2, and 19.3), only very few of these studies also report differences in memory performance (e.g., (Hariri et al. 2003; Bertolino et al. 2006; Krug et al. 2013a). In fact, Krug et al. (2014) included two independent fMRI samples in their paper and reported a consistent increase in hippocampal activity during memory retrieval, while memory performance only differed significantly in one of these samples. According to the statistical considerations detailed earlier, these inconsistencies regarding the behavioral results are not surprising. Consistent results for both behavior and brain activity can only be expected when the number of participants is sufficiently high to reliably detect both genotype-dependent effects
chapter 1 9 : I maging G enetics of E pisodic M emory
303
η2 = 0.07
Statistical power (1-β)
(a) 1.0 0.8 0.6
α = 0.05 α = 0.1
0.4 0.2 0.0 0
50
100
200
250
300
250
300
η2 = 0.25
(b) Statistical power (1-β)
150
1.0 0.8 0.6
α = 0.001 α = 0.01
0.4 0.2 0.0
0
50
100
150
200
Total sample size N Figure 19.4 Statistical power to detect an effect of η2 = 0.07.
(A) and η2 = 0.25 (B) depending on the total sample size N and the significance level α. An fMRI-sample of 50 subjects may have sufficient statistical power (70%) to detect a large genotype difference in brain activity (B), but has low statistical power (37%) to detect small genotype effects on memory behavior (A). Statistical power is calculated here under the assumption that an effect of η2 = 0.07 of genotype on memory performance exists in the population, although these effects can also be smaller. Thus, significant genotype effects on brain activity are likely to be paired with non-significant genotype effects on behavioral measures in fMRI studies with small sample sizes. Power calculations were done with G*Power3 (Faul et al., 2007). Reprinted from NeuroImage, 53, Rasch B, Papassotiropoulos A, Quervain D-F de, Imaging genetics of cognitive functions: Focus on episodic memory, 870–877, Copyright (2010), with permission from Elsevier.
on behavior and underlying neural activation (see, e.g., de Quervain et al. 2012). How high is the chance to replicate the genotypedependent activation difference in an fMRI study? For example, the highest MTL activity difference between BDNF genotypes during memory retrieval reported by Hariri et al. (2003) had an estimated effect size of approximately η2 = 25%. Applying the logic of univariate statistics, the chance of replicating this effect is 70% with 50 subjects when using an uncorrected threshold of p = 0.001 (Figure 19.4). However, estimates of effect sizes and statistical power in imaging studies are much more complex due to the high number of examined inter-correlated voxels (see, e.g., Fox et al. 2001; Desmond & Glover 2002; Murphy & Garavan 2004; Reiman 2007 for further discussion on this topic). Recently, Hayasaka and colleagues (2007) have proposed a framework to calculate power and sample size
maps for fMRI data. Here, effect sizes estimates are based on spheres around voxels to account for data smoothing. It might be useful to report such effect sizes in future fMRI studies instead of only peak activations to better estimate sample sizes required for replication studies (see toolbox PowerMap, Joyce and Hayasaka 2012; or fMRIpower, Mumford and Nichols 2008). In sum, non-significant effects of genetic variants on episodic memory performance are very likely in small fMRI samples, while genotype-dependent differences in brain activation are detected more reliably. However, how can we interpret significant genotype-dependent effect on brain activity, if these effects do not translate in significant differences in episodic memory performance? Do they really reflect increased responsivity or efficiency of the memory-related brain regions, which underlies or at least partially explains genotype-dependent difference in memory processing? How are activity differences in the memory network related to the differential episodic memory performance in genotype groups? To fully answer these important questions, future imaging genetic studies will require larger sample sizes to achieve sufficient statistical power to reliably detect genetic effects on both brain activity as well as behavioral measures of memory. Notably, sample sizes have remarkably increased in recent imaging genetics studies, now sometimes including several hundreds of participants (e.g., Papassotiropoulos et al. 2013; de Quervain et al. 2012). These large samples allow the investigation of both performance-unmatched and performance-matched genotype groups, to measure both genotype-dependent differences in brain activity that are related to the genotypedependent differences in behavior and genotype-dependent differences in brain activity under exclusion of behavioral differences (see, e.g., Kauppi et al. 2011). Finally, genetic imaging studies of episodic memory should be replicated or include replication samples, as it is common in behavioral genetic studies to avoid reporting of false positives (see, e.g., Krug et al. 2014; Erk et al. 2010, 2014). C ONC LU S IONS At this point, numerous imaging genetics studies of episodic memory have shown that it is feasible to measure genotype-dependent differences in brain activity using fMRI. Although the field of imaging genetics of episodic memory is still young, it has already become clear that imaging methods have a large potential to enhance our understanding of the neural mechanisms that underlie genetic differences in memory functions and to help validate results
3 0 4 part V I : I maging G enetics and C ognition
obtained from behavioral genetic studies. However, as with behavioral genetics studies, replications of the existing results are absolutely necessary for an appropriate evaluation of the genetic effects on brain activity. Regarding the different sensitivity of imaging genetics studies and behavioral genetics studies, we recommend indicating effect sizes and statistical power estimations to avoid misinterpretations and inconsistencies due to statistical reasons. In spite of the potential, there are also important limitations: imaging genetics studies using functional imaging techniques do not inform us about the underlying molecular mechanisms related to gene effects on memory, because recording of imaging data captures effects only on a broad system level of neural networks. Furthermore, the site of the largest genotype-dependent difference in brain activity does not necessarily indicate the major site of the gene effect on the molecular or cellular level. Finally, it is important to emphasize that studies examining genetic differences in memory performance or brain activity are correlational in nature and do not allow causal interpretations. Several important questions remain unanswered. For example, what does it mean in terms of memory functioning that similar effects on memory performance and memory-related brain activity are associated with different genetic variations? How do the effects of individual SNPs relate to each other? Do they simply add up, or do they interact (see, e.g., Bertolino et al. 2008; Heck et al. 2011)? Can the reported genetic effect on memory be generalized to other memory tasks or even other memory systems? By investigating these and other questions, future imaging genetics studies of episodic memory will have a great potential to increase our understanding of the genetics of human episodic memory, and may also influence and reshape our neurobiological concepts underlying memory functions.
Ackermann S, Heck A, Rasch B, Papassotiropoulos A, Quervain DJ de (2013). The BclI polymorphism of the glucocorticoid receptor gene is associated with emotional memory performance in healthy individuals. Psychoneuroendocrinology. 38: 1203–1207.
Baig BJ, Whalley HC, Hall J, McIntosh AM, Job DE, CunninghamOwens DG, Johnstone EC, Lawrie SM. (2010). Functional magnetic resonance imaging of BDNF val66met polymorphism in unmedicated subjects at high genetic risk of schizophrenia performing a verbal memory task. Psychiatry Res. 183: 195–201. Bertolino A, Di GA, Blasi G, Sambataro F, Caforio G, Sinibaldi L, Latorre V, Rampino A, Taurisano P, Fazio L, Romano R, Douzgou S, Popolizio T, Kolachana B, Nardini M, Weinberger DR, Dallapiccola B. (2008). Epistasis between dopamine regulating genes identifies a nonlinear response of the human hippocampus during memory tasks. Biol. Psychiatry. 64: 226–234. Bertolino A, Rubino V, Sambataro F, Blasi G, Latorre V, Fazio L, Caforio G, Petruzzella V, Kolachana B, Hariri A, Meyer-Lindenberg A, Nardini M, Weinberger DR, Scarabino T. (2006). Prefrontal-hippocampal coupling during memory processing is modulated by COMT val158met genotype. Biol.Psychiatry. 60: 1250–1258. Bigos KL, Mattay VS, Callicott JH, Straub RE, Vakkalanka R, Kolachana B, Hyde TM, Lipska BK, Kleinman JE, Weinberger DR. (2010). Genetic variation in CACNA1C affects brain circuitries related to mental illness. Arch Gen Psychiatry. 67: 939–945. Buchmann A, Mondadori CR, Hanggi J, Aerni A, Vrticka P, Luechinger R, Boesiger P, Hock C, Nitsch RM, Quervain DJ de, Papassotiropoulos A, Henke K. (2008). Prion protein M129V polymorphism affects retrieval-related brain activity. Neuropsychologia. 46: 2389–2402. Cabeza R, Nyberg L. (2000). Neural bases of learning and memory: functional neuroimaging evidence. Curr Opin Neurol. 13: 415–421. Cabeza R, Ciaramelli E, Olson IR, Moscovitch M. (2008). The parietal cortex and episodic memory: an attentional account. Nat Rev Neurosci. 9: 613–625. Cahill L, McGaugh JL. (1998). Mechanisms of emotional arousal and lasting declarative memory. Trends Neurosci. 21: 294–299. Cohen J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Cohen NJ, Ryan J, Hunt C, Romine L, Wszalek T, Nash C. (1999). Hippocampal system and declarative (relational) memory: summarizing the data from functional neuroimaging studies. Hippocampus. 9: 83–98. Dempster E, Toulopoulou T, McDonald C, Bramon E, Walshe M, Filbey F, Wickham H, Sham PC, Murray RM, Collier DA. (2005). Association between BDNF val66 met genotype and episodic memory. Am J Med Genet B Neuropsychiatr Genet. 134, 73–75. Dempster EL, Toulopoulou T, McDonald C, Bramon E, Walshe M, Wickham H, Sham PC, Murray RM, Collier DA. (2006). Episodic memory performance predicted by the 2bp deletion in exon 6 of the “alpha 7-like” nicotinic receptor subunit gene. Am J Psychiatry. 163: 1832–1834. Dennis NA, Cabeza R, Need AC, Waters-Metenier S, Goldstein DB, Labar KS. (2011). Brain-derived neurotrophic factor val66met polymorphism and hippocampal activation during episodic encoding and retrieval tasks. Hippocampus. 21: 980–989. Desmond JE, Glover GH. (2002). Estimating sample size in functional MRI (fMRI) neuroimaging studies: statistical power analyses. J Neurosci Methods. 118: 115–128. Di Giorgio A, Caforio G, Blasi G, Taurisano P, Fazio L, Romano R, Ursini G, Gelao B, Bianco LL, Papazacharias A, Sinibaldi L, Popolizio T, Bellomo A, Bertolino A. (2011). Catechol-O-methyltransferase Val(158)Met association with parahippocampal physiology during memory encoding in schizophrenia. Psychol Med. 41: 1721–1731. Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, Zaitsev E, Gold B, Goldman D, Dean M, Lu B, Weinberger DR. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 112: 257–269. Egan MF, Straub RE, Goldberg TE, Yakub I, Callicott JH, Hariri AR, Mattay VS, Bertolino A, Hyde TM, Shannon-Weickert C, Akil M, Crook J, Vakkalanka RK, Balkissoon R, Gibbs RA, Kleinman JE, Weinberger DR. (2004). Variation in GRM3 affects cognition,
chapter 1 9 : I maging G enetics of E pisodic M emory
305
AC K NOW L E D G MEN TS The authors gratefully acknowledge the permission to reproduce the copyrighted material (i.e., text extracts and figures from Rasch B, Papassotiropoulos A, de Quervain DJ-F, Neuroimage 53 [2010]: 870–877, Elsevier Inc.), in this chapter.
R EF ERENCES
prefrontal glutamate, and risk for schizophrenia. Proc Natl Acad Sci U S A. 101: 12604–12609. Eichenbaum H, Lipton PA. (2008). Towards a functional organization of the medial temporal lobe memory system: role of the parahippocampal and medial entorhinal cortical areas. Hippocampus. 18: 1314–1324. Eichenbaum H, Yonelinas AP, Ranganath C. (2007). The medial temporal lobe and recognition memory. Annu Rev Neurosci. 30: 123–152. Erk S, Meyer-Lindenberg A, Schnell K, von Opitz BC, Esslinger C, Kirsch P, Grimm O, Arnold C, Haddad L, Witt SH, Cichon S, Nothen MM, Rietschel M, Walter H. (2010). Brain function in carriers of a genome-wide supported bipolar disorder variant. Arch Gen Psychiatry. 67: 803–811. Erk S, Meyer-Lindenberg A, Linden DE, Lancaster T, Mohnke S, Grimm O, Degenhardt FA, Holmans P, Pocklington A, Schmierer P, Haddad L, Mühleisen T, Mattheisen M, Witt SH, Romanczuk-Seiferth N, Tost H, Schott B, Cichon S, Nöthen M, Rietschel M, Heinz A, Walter H. (2014). Replication of brain function effects of a genome-wide supported psychiatric risk variant in the CACNA1C gene and new multi-locus effects. Neuroimage. 94: 147–154. Faul F, Erfelder E, Lang A-G, Buchner A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods. 39: 175–191. Fera F, Passamonti L, Cerasa A, Gioia MC, Liguori M, Manna I, Valentino P, Quattrone A. (2013). The BDNF Val66Met polymorphism has opposite effects on memory circuits of multiple sclerosis patients and controls. PLoS One. 8: e61063. Ferreira MA, O’Donovan MC, Meng YA, Jones IR, Ruderfer DM, Jones L, Fan J, Kirov G, Perlis RH, Green EK, Smoller JW, Grozeva D, Stone J, et al. (2008). Collaborative genome-wide association analysis supports a role for ANK3 and CACNA1C in bipolar disorder. Nat Genet. 40: 1056–1058. Fletcher PC, Henson RN. (2001). Frontal lobes and human memory: insights from functional neuroimaging. Brain. 124: 849–881. Fox NC, Crum WR, Scahill RI, Stevens JM, Janssen JC, Rossor MN. (2001). Imaging of onset and progression of Alzheimer’s disease with voxel-compression mapping of serial magnetic resonance images. Lancet. 358: 201–205. Frias CM de, Annerbrink K, Westberg L, Eriksson E, Adolfsson R, Nilsson LG. (2004). COMT gene polymorphism is associated with declarative memory in adulthood and old age. Behav Genet. 34: 533–539. Han SD, Bangen KJ, Bondi MW. (2009). Functional magnetic resonance imaging of compensatory neural recruitment in aging and risk for Alzheimer’s disease: review and recommendations. Dement Geriatr Cogn Disord. 27: 1–10. Hariri AR, Drabant EM, Weinberger DR. (2006). Imaging genetics: perspectives from studies of genetically driven variation in serotonin function and corticolimbic affective processing. Biol Psychiatry. 59: 888–897. Hariri AR, Goldberg TE, Mattay VS, Kolachana BS, Callicott JH, Egan MF, Weinberger DR. (2003). Brain-derived neurotrophic factor val66met polymorphism affects human memory-related hippocampal activity and predicts memory performance. J Neurosci. 23: 6690–6694. Hashimoto R, Moriguchi Y, Yamashita F, Mori T, Nemoto K, Okada T, Hori H, Noguchi H, Kunugi H, Ohnishi T. (2008). Dose-dependent effect of the Val66Met polymorphism of the brain-derived neurotrophic factor gene on memory-related hippocampal activity. Neurosci Res. 61; 360–367. Hayasaka S, Peiffer AM, Hugenschmidt CE, Laurienti PJ. (2007). Power and sample size calculation for neuroimaging studies by non-central random field theory. NeuroImage. 37: 721–730. Heck A, Vogler C, Gschwind L, Ackermann S, Auschra B, Spalek K, Rasch B, Quervain DJ de, Papassotiropoulos A. (2011). Statistical epistasis and functional brain imaging support a role of voltage-gated potassium channels in human memory. PLoS One. 6: e29337.
Hedden T, Gabrieli JDE. (2005). Healthy and pathological processes in adult development: new evidence from neuroimaging of the aging brain. Curr Opin Neurology. 18: 740–747. Huentelman MJ, Papassotiropoulos A, Craig DW, Hoerndli FJ, Pearson JV, Huynh KD, Corneveaux J, Hanggi J, Mondadori CR, Buchmann A, Reiman EM, Henke K, Quervain DJ de, Stephan DA. (2007). Calmodulin-binding transcription activator 1 (CAMTA1) alleles predispose human episodic memory performance. Hum.Mol.Genet., 16, 1469–1477. Joyce KE, Hayasaka S. (2012). Development of PowerMap: a software package for statistical power calculation in neuroimaging studies. Neuroinformatics. 10: 351–365. Jung MW, Baeg EH, Kim MJ, Kim YB, Kim JJ. (2008). Plasticity and memory in the prefrontal cortex. Rev Neurosci. 19: 29–46. Kauppi K, Nilsson LG, Adolfsson R, Eriksson E, Nyberg L. (2011). KIBRA polymorphism is related to enhanced memory and elevated hippocampal processing. J Neurosci. 31: 14218–14222. Kauppi K, Nilsson LG, Adolfsson R, Lundquist A, Eriksson E, Nyberg L. (2013). Decreased medial temporal lobe activation in BDNF (66)Met allele carriers during memory encoding. Neuropsychologia. 51(12): 2462–2468 Kolsch H, Wagner M, Bilkei-Gorzo A, Toliat MR, Pentzek M, Fuchs A, Kaduszkiewicz H, van den BH, Riedel-Heller SG, Angermeyer MC, Weyerer S, Werle J, Bickel H, Mosch E, Wiese B, Daerr M, Jessen F, Maier W, Dichgans M. (2009). Gene polymorphisms in prodynorphin (PDYN) are associated with episodic memory in the elderly. J Neural Transm. 116: 897–903. Krug A, Krach S, Jansen A, Nieratschker V, Witt SH, Shah NJ, Nothen MM, Rietschel M, Kircher T. (2013a). The effect of neurogranin on neural correlates of episodic memory encoding and retrieval. Schizophr Bull. 39: 141–150. Krug A, Witt SH, Backes H, Dietsche B, Nieratschker V, Shah NJ, Nothen MM, Rietschel M, Kircher T. (2014). A genome-wide supported variant in CACNA1C influences hippocampal activation during episodic memory encoding and retrieval. Eur Arch Psychiatry Clin.Neurosci. 264: 103–110. Labar KS, Cabeza R. (2006). Cognitive neuroscience of emotional memory. Nat Rev Neurosci. 7: 54–64. Lachman HM, Papolos DF, Saito T, Yu YM, Szumlanski CL, Weinshilboum RM. (1996). Human catechol-O-methyltransferase pharmacogenetics: description of a functional polymorphism and its potential application to neuropsychiatric disorders. Pharmacogenetics. 6: 243–250. Levine TR, Hullet CR. (2002). Eta squared, partial eta squared, and misreporting of effect size in communication research. Human Comm Res. 28: 612–625. Liberzon I, Martis B. (2006). Neuroimaging studies of emotional responses in PTSD. Ann N Y Acad Sci. 1071, 87–109. Liu Y, Blackwood DH, Caesar S, Geus EJ de, Farmer A, Ferreira MA, Ferrier IN, Fraser C, Gordon-Smith K, Green EK, Grozeva D, Gurling HM,et al. (2011). Meta-analysis of genome-wide association data of bipolar disorder and major depressive disorder. Mol Psychiatry. 16: 2–4. Mascetti L, Foret A, Schrouff J, Muto V, Dideberg V, Balteau E, Degueldre C, Phillips C, Luxen A, Collette F, Bours V, Maquet P. (2013). Concurrent synaptic and systems memory consolidation during sleep. J Neurosci. 33: 10182–10190. Mattay VS, Goldberg TE, Sambataro F, Weinberger DR. (2008). Neurobiology of cognitive aging: insights from imaging genetics. Biol Psychol. 79: 9–22. McClearn GE, Johansson B, Berg S, Pedersen NL, Ahern F, Petrill SA, Plomin R. (1997). Substantial genetic influence on cognitive abilities in twins 80 or more years old. Science. 276: 1560–1563. Milner B. (1972). Disorders of learning and memory after temporal lobe lesions in man. Clin Neurosurg. 19; 421–446. Milnik A, Heck A, Vogler C, Heinze HJ, Quervain DJ de, Papassotiropoulos A. (2012). Association of KIBRA with episodic and working memory: a meta-analysis. Am J Med Genet B Neuropsychiatr Genet. 159: 958–969.
3 0 6 part V I : I maging G enetics and C ognition
Miyajima F, Ollier W, Mayes A, Jackson A, Thacker N, Rabbitt P, Pendleton N, Horan M, Payton A. (2008). Brain-derived neurotrophic factor polymorphism Val66Met influences cognitive abilities in the elderly. Genes Brain Behav. 7: 411–417. Moosmang S, Haider N, Klugbauer N, Adelsberger H, Langwieser N, Muller J, Stiess M, Marais E, Schulla V, Lacinova L, Goebbels S, Nave KA, Storm DR, Hofmann F, Kleppisch T. (2005). Role of hippocampal Cav1.2 Ca2+ channels in NMDA receptor-independent synaptic plasticity and spatial memory. J Neurosci. 25: 9883–9892. Moscovitch M, Nadel L, Winocur G, Gilboa A, Rosenbaum RS. (2006). The cognitive neuroscience of remote episodic, semantic and spatial memory. Curr Opin Neurobiol. 16: 179–190. Mumford JA, Nichols TE. (2008). Power calculation for group fMRI studies accounting for arbitrary design and temporal autocorrelation. NeuroImage. 39: 261–268. Murphy K, Garavan H. (2004). An empirical investigation into the number of subjects required for an event-related fMRI study. NeuroImage. 22: 879–885. Nadel L, Samsonovich A, Ryan L, Moscovitch M. (2000). Multiple trace theory of human memory: computational, neuroimaging, and neuropsychological results. Hippocampus. 10: 352–368. Nieuwenhuis ILC, Takashima A. (2011). The role of the ventromedial prefrontal cortex in memory consolidation. Behav Brain Res. 218: 325–334. Papassotiropoulos A, Henke K, Aerni A, Coluccia D, Garcia E, Wollmer MA, Huynh KD, Monsch AU, Stahelin HB, Hock C, Nitsch RM, Quervain DJ de (2005a). Age-dependent effects of the 5-hydroxytryptamine-2a-receptor polymorphism (His452Tyr) on human memory. Neuroreport. 16: 839–842. Papassotiropoulos A, Quervain DJ de (2011). Genetics of human episodic memory: dealing with complexity. Trends Cogn Sci. 15: 381–387. Papassotiropoulos A, Stefanova E, Vogler C, Gschwind L, Ackermann S, Spalek K, Rasch B, Heck A, Aerni A, Hanser E, Demougin P, Huynh KD, Luechinger R, Klarhofer M, Novakovic I, Kostic V, Boesiger P, Scheffler K, Quervain DJ de (2013). A genome-wide survey and functional brain imaging study identify CTNNBL1 as a memory-related gene. Mol Psychiatry. 18: 255–263. Papassotiropoulos A, Stephan DA, Huentelman MJ, Hoerndli FJ, Craig DW, Pearson JV, Huynh KD, Brunner F, Corneveaux J, Osborne D, Wollmer MA, et al. (2006). Common Kibra alleles are associated with human memory performance. Science. 314: 475–478. Papassotiropoulos A, Wollmer MA, Aguzzi A, Hock C, Nitsch RM, Quervain DJ de (2005b). The prion gene is associated with human long-term memory. Hum Mol Genet. 14: 2241–2246. Papassotiropoulos A, Wollmer MA, Tsolaki M, Brunner F, Molyva D, Lutjohann D, Nitsch RM, Hock C. (2005c). A cluster of cholesterol-related genes confers susceptibility for Alzheimer’s disease. J Clin Psychiatry. 66: 940–947. Phelps EA. (2004). Human emotion and memory: interactions of the amygdala and hippocampal complex. Curr Opin Neurobiol. 14: 198–202. Poo MM. (2001). Neurotrophins as synaptic modulators. Nat Rev Neurosci. 2: 24–32. Potkin SG, Turner JA, Guffanti G, Lakatos A, Torri F, Keator DB, Macciardi F. (2009). Genome-wide strategies for discovering genetic influences on cognition and cognitive disorders: methodological considerations. Cogn Neuropsychiatry. 14: 391–418. Prvulovic D, van de Ven V, Sack AT, Maurer K, Linden DEJ. (2005). Functional activation imaging in aging and dementia. Psychiat Res. 140: 97–113. Quervain DJ de, Aerni A, Schelling G, Roozendaal B. (2009). Glucocorticoids and the regulation of memory in health and disease. Front Neuroendocrinol. 30: 358–370. Quervain D-F de, Henke K, Aerni A, Coluccia D, Wollmer MA, Hock C, Nitsch RM, Papassotiropoulos A. (2003). A functional genetic variation of the 5-HT2a receptor affects human memory. Nat Neurosci. 6: 1141–1142.
Quervain DJ-F de, Kolassa IT, Ackermann S, Aerni A, Boesiger P, Demougin P, Elbert T, Ertl V, Gschwind L, Hadziselimovic N, Hanser E, Heck A, et al. (2012). PKCalpha is genetically linked to memory capacity in healthy subjects and to risk for posttraumatic stress disorder in genocide survivors. Proc Natl Acad Sci U S A. 109: 8746–8751. Quervain DJ-F de, Kolassa IT, Ertl V, Onyut PL, Neuner F, Elbert T, Papassotiropoulos A. (2007). A deletion variant of the alpha2b-adrenoceptor is related to emotional memory in Europeans and Africans. Nat Neurosci. 10: 1137–1139. Quervain DJ-F de, Papassotiropoulos A. (2006). Identification of a genetic cluster influencing memory performance and hippocampal activity in humans. Proc Natl Acad Sci U S A. 103: 4270–4274. Rajah MN, D’Esposito M. (2005). Region-specific changes in prefrontal function with age: a review of PET and fMRI studies on working and episodic memory. Brain. 128: 1964–1983. Rasch B, Papassotiropoulos A, Quervain D-F de (2010). Imaging genetics of cognitive functions: Focus on episodic memory. NeuroImage. 53: 870–877. Rasch B, Spalek K, Buholzer S, Luechinger R, Boesiger P, Papassotiropoulos A, Quervain DJ-F de (2009). A genetic variation of the noradrenergic system is related to differential amygdala activation during encoding of emotional memories. Proc Natl Acad Sci U S A. 106: 19191–19196. Reiman EM. (2007). Linking brain imaging and genomics in the study of Alzheimer’s disease and aging. Ann N Y Acad Scie. 1097: 94–113. Reiman, E. M., Chen, K., Alexander, G. E., Caselli, R. J., Bandy D, Osborne D, Saunders AM, Hardy J. (2005). Correlations between apolipoprotein E epsilon4 gene dose and brain-imaging measurements of regional hypometabolism. Proc Natl Acad Sci U S A. 102: 8299–8302. Rugg MD, Otten LJ, Henson RN. (2002). The neural basis of episodic memory: evidence from functional neuroimaging. Philos Trans R Soc Lond B Biol Sci. 357: 1097–1110. Samanez-Larkin GR, D’Esposito M. (2008). Group comparisons: imaging the aging brain. Soc Cogn Affect Neurosci. 3: 290–297. Sambataro F, Murty VP, Lemaitre HS, Reed JD, Das S, Goldberg TE, Callicott JH, Weinberger DR, Mattay VS. (2010). BDNF modulates normal human hippocampal ageing [corrected]. Mol Psychiatry. 15: 116–118. Schacter DL, Wagner AD. (1999). Medial temporal lobe activations in fMRI and PET studies of episodic encoding and retrieval. Hippocampus. 9: 7–24. Schott BH, Seidenbecher CI, Fenker DB, Lauer CJ, Bunzeck N, Bernstein HG, Tischmeyer W, Gundelfinger ED, Heinze HJ, Duzel E. (2006). The dopaminergic midbrain participates in human episodic memory formation: evidence from genetic imaging. J Neurosci. 26: 1407–1417. Schott BH, Assmann A, Schmierer P, Soch J, Erk S, Garbusow M, Mohnke S, Pohland L, Romanczuk-Seiferth N, Barman A, Wustenberg T, Haddad L, Grimm O, Witt S, Richter S, Klein M, Schutze H, Muhleisen TW, Cichon S, Rietschel M, Noethen MM, Tost H, Gundelfinger ED, Duzel E, Heinz A, Meyer-Lindenberg A, Seidenbecher CI, Walter H. (2014) Epistatic interaction of genetic depression risk variants in the human subgenual cingulate cortex during memory encoding. Transl Psychiatry 4, e372. Smoller JW, Craddock N, Kendler K, Lee PH, Neale BM, Nurnberger JI, Ripke S, Santangelo S, Sullivan PF. (2013). Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis. Lancet. 381: 1371–1379. Spaniol J, Davidson PS, Kim AS, Han H, Moscovitch M, Grady CL. (2009). Event-related fMRI studies of episodic encoding and retrieval: meta-analyses using activation likelihood estimation. Neuropsychologia. 47: 1765–1779. Squire LR. (1992). Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans. Psychol Rev. 99: 195–231. Squire LR, Stark CE, Clark RE. (2004). The medial temporal lobe. Annu Rev Neurosci. 27: 279–306.
chapter 1 9 : I maging G enetics of E pisodic M emory
307
Squire LR, Zola SM. (1996). Structure and function of declarative and nondeclarative memory systems. Proc Natl Acad Sci U S A. 93: 13515–13522. Tulving E. (1983). Elements of episodic memory. New York: Oxford: University Press. Tulving E. (1993). What is episodic memory? Curr Direct Psychol Sci. 2: 67–70. Uncapher MR, Wagner AD. (2009). Posterior parietal cortex and episodic encoding: insights from fMRI subsequent memory effects and dual-attention theory. Neurobiol Learn Mem. 91: 139–154. Urner M van Wingen, Franke B, Rijpkema M, Fernandez G, Tendolkar I. (2011). Genetic variation of the alpha2b-adrenoceptor affects neural
correlates of successful emotional memory formation. Hum Brain Mapp. 32: 2096–2103. van Wingen G, Rijpkema M, Franke B, van Eijndhoven, Tendolkar I, Verkes RJ, Buitelaar J, Fernandez G. (2010). The brain-derived neurotrophic factor Val66Met polymorphism affects memory formation and retrieval of biologically salient stimuli. Neuroimage. 50: 1212–1218. Vogler C, Spalek K, Aerni A, Demougin P, Muller A, Huynh KD, Papassotiropoulos A, Quervain DJ de (2009). CPEB3 is associated with human episodic memory. Front Behav Neurosci. 3: 4. Yonelinas AP. (2001). Components of episodic memory: the contribution of recollection and familiarity. Philos Trans R Soc Lond B Biol Sci. 356: 1363–1374.
3 0 8 part V I : I maging G enetics and C ognition
20. IMAGING GENETICS OF WORKING MEMORY Tristram A. Lett, Eva J. Brandl, Daniel J. Müller, Andreas Heinz, and Henrik Walter
W
orking memory deficits are a core feature of many psychiatric disorders, most prominently schizophrenia. Imaging genetics provides the unique opportunity to link genetic risk variants to brain alterations that underlie cognitive dysfunction. In this regard, imaging genetics can describe how a specific gene variant can impact on components of working memory (Egan et al. 2001). This expanded knowledge may, in turn, lead to novel treatment options. In this chapter, we first discuss some of the theoretical and neurobiological mechanisms of working memory. Then, we review the impact of the most important and best replicated candidate genes and genome-wide variants on neuroimaging phenotypes relevant to working memory functioning. Finally, we will discuss clinical relevance, limitations, and future directions in the field of imaging genetics of working memory. DEF INITI O N O F WO RKI N G MEMO RY Working memory refers to the temporary storage and manipulation of information to guide goal-directed behavior in the absence of the original input. It has been proposed in the seminal theoretical work by Alan Baddeley that working memory includes a central executive to control attention and information flow between verbal and short-term memory buffers (Baddeley 1992). Working memory storage functions can be fractionalized into specialized systems that serve as buffers for the storage and manipulation of information. The contents of working memory are constitutively updated, monitored, and manipulated in response to immediate processing demands (Norman and Shallice 1983; Baddeley 1986). Working memory prolongs the impact of experience beyond immediately accessible information to enable the incorporation of information from long-term memory, lexical labels, and other events into goal-oriented behavior (Mesulam 1998).
Studies in non-human primates suggest that lesions to the prefrontal cortex (PFC) cause marked reduction in working memory function, and that subdivisions of the PFC may represent multiple working memory domains, each having its own specialized processing (Petrides 1996) or content-specific storage (Goldman-Rakic 1996). The PFC also contributes to working memory functioning by exerting top-down control through filtering and strategic reorganization of information (Postle 2006). Therefore, working memory performance depends on efficient communication to the PFC and its capacity to inhibit extraneous information. WOR K ING ME MOR Y AND NE T WOR K C ONNE C T IV IT Y The dorsolateral prefrontal cortex (DLPFC) has been identified as a crucial region for working memory function in healthy adults (D’Esposito et al. 2000). It has been suggested that the PFC acts as a flexible hub by which frontal connectivity is adjusted according to task demands (Miller and Cohen, 2001). Recent advances in applying graph theory (i.e., network theory) to functional magnetic resonance imaging (fMRI) data underline the importance of the DLPFC as a global hub. High global brain network connectivity to the DLPFC robustly correlates with working memory performance as well as general fluid intelligence (Cole et al. 2012). Working memory seems to be particularly dependent on synchronous activations between the DLPFC and the inferior parietal lobe (i.e., frontoparietal connectivity) and between the DLPFC and the medial temporal lobe (i.e., frontotemporal connectivity). Electroencephalogram (EEG) studies measuring neural oscillation show that phase coupling of theta oscillations (4–8 Hz) between the prefrontal and parietal cortices 309
increases with the complexity of working memory manipulation (Sauseng et al. 2005), load (Payne and Kounios 2009), and performance (Kopp et al. 2006). Theta phase synchrony between the prefrontal and temporal cortices occurs during the maintenance phase of working memory (Sarnthein et al. 1998; Serrien et al. 2004) in addition to encoding and retrieval (Sauseng et al. 2004). Furthermore, precision of theta-gamma coupling in the hippocampus predicts working memory performance (Axmacher et al. 2010). Phase synchronization between higher-order sensory, frontal, and temporal cortices and the hippocampus provides a mechanism for working memory maintenance by which activity in different brain regions is sustained in the absence of direct sensory input (Axmacher et al. 2008; Sarnthein et al. 1998; Sauseng et al. 2004; Serrien et al. 2004). The disruption of working memory in neuropsychiatric disorders has been linked to DLPFC functioning. In particular, patients with schizophrenia have cognitive impairments manifesting in the prodromal period. Typically, patients suffer from deficits in working memory with no clear differences across domains or tasks (Lee and Park 2005; Forbes et al. 2009). Compared to healthy individuals, schizophrenia patients apparently need to engage more prefrontal activation to match working memory performance, and patients with reduced activation have poorer performance (Callicott et al. 2003). Schizophrenia patients also tend to have abnormal DLPFC functional connectivity. During working memory tasks, patients tend to have increased activation in the thalamus, anterior cingulate cortex, and temporal cortices (Walter et al. 2007) and decreased cerebellar activation compared to control subjects (Meyer-Lindenberg et al. 2001; Schlosser et al. 2003). Furthermore, subjects with good working memory performance show frontoparietal activation matching task demand, whereas high working memory demand in poor performers leads to increased thalamic and striatal activity (Wolf and Walter 2005). It has been suggested that these stronger activations in deeper regions may be a compensatory mechanism, (e.g., in the thalamus [Minzenberg et al., 2009] and the anterior cingulate cortex [Bor et al. 2011]). Using dynamic causal modeling, Schlagenhauf et al. observed that working memory-dependent effective connectivity from prefrontal to parietal cortex is reduced in all schizophrenia patients, independent of performance (Schlagenhauf et al. 2013; Deserno et al., 2012). Also, it has been described that patients with schizophrenia are characterized by aberrant frontotemporal connectivity (Meyer-Lindenberg et al. 2005b; Wolf et al., 2009). These findings suggest that schizophrenia patients are unable to engage working memory networks efficiently.
H E R ITAB ILIT Y OF WOR K ING ME MOR Y Working memory appears to have a molecular genetic component, providing clues into how brain networks can be disrupted, causing dysfunction relevant to psychiatric disorders. Indeed, it is clear that many complex behaviors and psychiatric disorders have a high degree of heritability, that is, the component of the total variance in a trait that is explained by genetic variation (Burmeister et al. 2008). However, with the exception of Alzheimer’s disease, no causative genes have been associated with the majority of psychiatric disorders. This “missing heritability” has a number of potential explanations, including heterogeneous phenotypes, low penetrance, genetic heterogeneity, polygenic inheritance, epigenetic regulation, and environmental effects (Maher 2008). Working memory dysfunction is common among multiple disorders, including schizophrenia, autism spectrum disorders (ASD), alcoholism (Charlet et al. 2013), and attention deficit hyperactivity disorders (ADHD), and is present in first-degree relatives. Furthermore, working memory dysfunction is relatively stable, readily measurable, and a disease-related phenotype that carries a high degree of heritability, suggesting that it may be a useful endophenotype (Gottesman and Gould 2003). The Consortium on the Genetics of Schizophrenia (COGS) has reported that in 183 nuclear families, approximately 40% of the variance in the letter-number sequencing test (LNS performance), a measure of working memory, was explained by genetics (Greenwood et al. 2007). Working memory may also be a partially heritable endophenotype in ADHD (Castellanos and Tannock 2002) and ASD (Viding and Blakemore 2007). It is likely that the genetic contribution to working memory is mediated by changes in brain activity. A study of 60 healthy twin pairs revealed that performance on the n-back working memory task during fMRI was significantly influenced by genetics (explained variance by heritability h2 = 55%–70%) with a smaller degree of variance explained by activation in the left medial frontal gyrus (Blokland et al. 2008). A subsequent update of the study with a sample of 319 healthy twin pairs reported that task-related brain activation was significantly heritable, with higher estimates (40%–65%) in the inferior, middle, superior frontal gyrus, and inferior parietal lobe (Blokland et al. 2011). Moreover, recent work suggests that greater than 40% of the variance in resting-state network connectivity is explained by genetic factors (Glahn et al. 2010). Cortical structures relevant to working memory are also explained, in part, by genetics. Frontal gray matter volume was found to be under significant genetic influence,
3 1 0 part vi : I maging G enetics and C ognition
particularly near the DLPFC (Brodmann areas [BA] 9 and 46), an area strongly associated with working memory (Thompson et al. 2001). It should be stressed that cortical volumes are composed of surface areas and cortical thickness that are partly heritable, with heritability [h2] explaining between 40%–95% of the variance, and they appear to be under distinct genetic influences. In schizophrenia, mean cortical thickness reductions were partly heritable, although the effect size in the siblings of patients was limited (Goldman et al. 2009). Diffusion tensor imaging of 705 twins shows that genetic variation explained over 80% of the variance in frontal fractional anisotropy (FA) in subjects with above-average IQ, but only 40% in individuals with low IQ (Chiang et al. 2011). Interestingly, the authors state that gene by environment effects play a significant role in predicting white matter FA (Chiang et al. 2011). In a study of 467 subjects from extended families, it was reported that all working memory task administered, gray matter regions, and white matter FA tracts had a significant genetic component (h2 > 40%, except for spatial-delayed response), although only spatial delayed response was associated with superior longitudinal fasciculus FA (frontoparietal tract) (Karlsgodt et al. 2010). However, longitudinal studies in pediatric and adults subjects show that the influence of genetic factors on any particular brain structure varies significantly with development and age (Finch 2002; Gilmore et al. 2010a; Gilmore et al. 2010b; Gilmore et al. 2012). Taken together, brain structure and function relevant to working memory are likely under genetic influence. Working memory is partly heritable, a stable trait, and dysfunction is present in first-degree relatives and is commonly reported in neuropsychiatric disorders. Therefore, working memory represents an intermediate phenotype that, like brain structure and connectivity, has garnered considerable attention (Meyer-Lindenberg and Weinberger 2006; Meyer-Lindenberg 2009; Glahn et al. 2010; Heinz et al. 2011; Tost et al. 2012). The following sections review the most promising results from imaging genetic studies on working memory, and results from fMRI studies are summarized in Table 20.1.
Blockade of the glutamate-mediated excitatory neurotransmission by NMDA receptor antagonists mimics positive and negative symptoms, as well as cognitive deficits
in schizophrenia. These findings suggest that enhancing NMDA receptor neurotransmission may reverse cognitive deficits (Malhotra et al. 1997; Moghaddam 2004). Furthermore, NMDA receptor ablation on GABA interneurons impairs hippocampal theta rhythm, leading to impaired working memory (Korotkova et al. 2010). NMDA receptor activation also subserves persistent DLPFC neuronal firing during working memory, suggesting that glutamate function and connectivity are integral to working memory performance (Wang et al. 2013). Variation in the dysbindin gene (DTNBP1) has been associated with schizophrenia and cognitive deficits resembling characteristics of schizophrenia (Hallmayer et al. 2005), and with spatial working memory in healthy individuals and patients with schizophrenia (Donohoe et al. 2007; Wolf et al. 2011). Dysbindin is expressed in axon terminals of glutamatergic pyramidal neurons, and putatively influences glutamatergic signaling via vesicular trafficking (Talbot et al. 2004). Mice carrying a null mutation in the dtnbp1 gene had impairments in spatial working memory compared with wild-type controls ( Jentsch et al. 2009). These impairments were mediated by reduction in paired pulse facilitation, and aberrant post-synaptic currents in neocortex pyramidal cells, suggesting that dysbindin potentially regulates glutamatergic signaling that is directly related to working memory ( Jentsch et al. 2009). Furthermore, dtnbp1 mutant mice have reduced NMDA receptor subunit NR1 expression and decreased NMDA-evoked currents in prefrontal pyramidal neurons that correlate with impaired spatial working memory performance (Karlsgodt et al. 2011). A haplotype within the DTNBP1 gene has been associated with increased risk for schizophrenia, decreased mRNA expression, poorer cognitive performance, and early sensory processing deficits (Williams et al. 2004). Using voxel-based morphometry (VBM) in 38 patients with schizophrenia, carriers of the risk haplotype were reported to have reductions in occipital and prefrontal gray matter volume (Donohoe et al. 2010a). In 57 healthy controls undergoing fMRI, carriers of the schizophrenia risk allele of the rs1018381 marker had increased activation in the middle frontal gyrus (BA 9) during the n-back working memory task, although activation and genotype were not associated with 2-back WM performance (Markov et al. 2010). In 86 healthy controls, the risk variant was also reported to be associated with greater FA in PFC, suggesting that this marker may alter prefrontal connectivity engaged in working memory but may not necessarily affect capacity (Nickl-Jockschat et al. 2012). In 56 healthy controls, carriers of the G-allele of rs1047631 had increased
chapter 2 0 : I maging G enetics of W orking M emory
311
S IGNA L ING SYSTEMS AN D CAN D I DAT E GE NE S GLUTAMATE
TABLE 20.1
FUNCTIONAL MRI STUDIES EXAMINING CORTICAL ACTIVATION DURING WORKING MEMORY TASKS
Gene
Imaging Method
Marker
Population (MRI)
Main Imaging Findings
References
APOE
fMRI, n-back
ε3/ε4
n = 35 HC
ε4 allele carriers had greater activation in medial frontal and parietal regions.
(Wishart et al. 2006)
BDNF
fMRI, n-back; MRI spectroscopy
rs6265
n = 30 HC
Met allele was associated with poorer episodic memory, increased hippocampal activation assayed with fMRI, and lower hippocampal n-acetyl aspartate (NAA)
(Egan et al. 2003)
CACNA1C
fMRI, n-back
rs1006737
n = 316 HC
A-allele homozygotes (risk genotype) had greater PFC activation compared to G-allele carriers.
(Bigos et al. 2010)
CACNA1C
fMRI, n-back
rs1006737
n = 94 HC
A-allele homozygotes had decreased DLPFC activation, and increased frontohippocampal connectivity compared to G-allele carriers.
(Paulus et al. 2013)
COMT
fMRI, n-back
rs4860
n = 11 SCZ, n = 27 siblings
Val carriers had worse executive functioning and increased prefrontal activation.
(Egan et al. 2001)
COMT
Meta-analysis; fMRI, n-back and others
rs4860
20 studies
Val allele was robustly associated with PFC activation (Cohen’s D = 0.73).
(Mier et al. 2010)
CYP2D6
fMRI, n-back
CYP2D6 enzymatic activity alleles
n = 114 HC
CYP2D6 activity was positively correlated with activation in the fusiform gyrus and precuneus.
(Stingl et al. 2012)
DTNBP1
fMRI, n-back
rs1018381
n = 57 HC
No difference in performance; risk carriers had increased activation in middle frontal gyrus (BA 9).
(Markov et al. 2010)
DTNBP1
fMRI, emotional working memory task
rs1047631
n = 56 HC
G-allele carriers had increased expression, working memory capacity, and activation in the hippocampus, temporal, and frontal cortex.
(Wolf et al. 2011)
GAD1
fMRI, N-back
rs7557793
n = 102 HC
C-allele carriers had lower activation in the DLPFC.
(Straub et al. 2007)
GRM3
fMRI, N-back
rs6465084
n = 65 HC
A-allele associated with poorer verbal list learning and fluency, lower glutamate, and increased activation in BA 46 during n-back.
(Egan et al. 2004)
MAOA
fMRI, N-back
Promoter VNTR
n = 40 HC, Male
High-activity variant carriers had increased activation in the ventrolateral PFC.
(Cerasa et al. 2008)
MIR137
fMRI, SIRP
rs1625579
n = 48 HC; n = 48 SCZ
T-allele homozygotes had increased activation in the DLPFC in healthy controls and patients with schizophrenia.
(van Erp et al. 2013)
NRG1
fMRI, n-back
rs3575305
n = 85 HC
Number of C-alleles associated with increased PFC activation (BA 10).
(Krug et al. 2008)
SNAP25
fMRI, Encoding
rs363039
n = 79 HC
A-allele carriers had greater deactivation of the posterior cingulate cortex during the working memory task.
(Soderqvist et al. 2010)
TPH2
fMRI, N-back
rs4570625
n = 49 HC
T-allele homozygotes had increased activation in the DLPFC.
(Reuter et al. 2008)
ZNF804A
fMRI, N-back
rs1344706
n = 115 HC
A-allele (psychosis risk) was associated with decreased connectivity between the right and left DLPFC, and increased right DLPFC-left hippocampus connectivity.
(Esslinger et al. 2009)
ZNF804A
fMRI, N-back
rs1344706
n = 153 HC; n = 171 A-allele modulated connectivity from the right SCZ siblings; n = 78 DLPFC to the hippocampus in healthy controls, SCZ siblings, and schizophrenia patients.
(Rasetti et al. 2011)
(continued)
TABLE 20.1 (CONTINUED)
Gene
Imaging Method
Marker
Population (MRI)
Main Imaging Findings
References
Studies examining statistical epistasis among genetic variants COMT X AKT1
fMRI, n-back; sMRI
rs4680, rs1130233
n = 46 HC; n = 68 HC
A-allele carriers (AKT1) had increased DLPFC activation. Greater effects were observed in Val homozygotes (COMT) with increased activation and reduced DLPFC gray matter.
(Tan et al. 2008)
COMT X DAOA
fMRI, n-back
rs4680, rs1421292
n = 82 HC
Val homozygotes (COMT) and T-allele homozygotes (GRM3) had greater DLPFC activation than other groups.
(Nixon et al. 2011)
COMT X GRM3
fMRI, n-back
rs4680, rs64650844
n = 29 HC
Val COMT homozygotes and A-allele homozygotes (GRM3) had lower DLPFC-inferior parietal connectivity.
(Tan et al. 2007)
NRG1 X ERBB4 X AKT1
fMRI, n-back
rs10503929, rs1026882, rs2494734
n = 172 HC
2- and 3-way interactions were observed with risk variants predicting less efficient DLPFC processing.
(Nicodemus et al. 2010)
AKT1 = v-akt murine thymoma viral oncogene homolog 1; APOE = apolipoprotein E; BDNF = brain-derived neurotrophic factor; CACNA1C = calcium channel, voltage-dependent, L type, alpha 1C subunit; COMT, = catechol-O-methyltransferase; CYP2D6 = cytochrome P450, family 2, subfamily D, polypeptide 6; DLPFC = dorsolateral prefrontal cortex; DTNBP1 = dystrobrevin binding protein 1; ERBB4 = v-erb-a erythroblastic leukemia viral oncogene homolog 4 (avian); fMRI = functional magnetic resonance imaging; GAD1 = glutamate decarboxylase 1; GRM3 = glutamate receptor, metabotropic 3; HC = healthy controls; MAOA = monoamine oxidase A; NRG1 = neuregulin 1; PFC = prefrontal cortex; SCZ = patients with schizophrenia; SIRP = Sternberg item recognition paradigm; sMRI = structural magnetic resonance imaging; SNAP25 = synaptosomal-associated protein, 25kDa; TPH2 = tryptophan hydroxylase 2; ZNF804A = zinc finger protein 804A.
DTNBP1 expression and higher working memory capacity for happy faces during an emotional working memory task (Wolf et al. 2011). Furthermore, G-allele carriers had increased activation across emotions in the hippocampus and temporal and frontal cortex (Wolf et al. 2011). The glutamate receptor, metabotropic (GRM3) gene was also associated with poorer working memory. The A-allele of the rs6465084 marker was associated with poorer verbal list learning and fluency and lower prefrontal N-acetylaspartate, a marker of neuronal integrity associated with glutamate. Furthermore, in 65 healthy controls, A-allele carriers had greater DLPFC activation (BA 46) during the n-back. These convergent findings suggest that GRM3 genotype alters prefrontal glutamate transmission integral to working memory (Egan et al. 2004).
The dopamine D1 receptor (D1R) is the most abundant receptor in the PFC, and determines the cellular effects of dopamine on this region (Guillin et al. 2007). Increased or decreased stimulation of the D1R in the PFC impairs spatial working memory producing an “inverted-U” dose response (Vijayraghavan et al. 2007). Local injection of D1R antagonists into the monkey PFC was reported to induce behavior indicative of working memory deficits, but had no effect on sensory and motor functions (Sawaguchi and Goldman-Rakic 1991; Goldman-Rakic et al. 2004). Increased PFC D1 receptor expression has been repeatedly reported in schizophrenia, may reflect a compensatory
up-regulation due to reduced PFC D1 release, and has been directly associated with poor working memory performance (Okubo et al. 1997; Abi-Dargham et al. 2002; Abi-Dargham and Moore 2003). The catechol-O-methyltransferase gene (COMT) has been associated with PFC dopamine turnover and working memory performance (Bertolino et al. 2004; MeyerLindenberg et al. 2005a; Heinz and Smolka, 2006).The Met allele of a missense mutation in COMT (rs4680; Val158Met) has been associated with a 3- to 4-fold reduction in enzymatic activity, leading to higher dopamine in the PFC (Chen et al. 2004). In a landmark study by Egan and colleagues, Val carriers had worse executive functioning and increased prefrontal activation during the n-back working memory task (Egan et al. 2001). In a meta-analysis of 20 fMRI studies examining the association between prefrontal activation and rs4680 genotype, the Val allele was robustly associated with higher prefrontal activation (Cohen’s D = 0.73) (Mier et al. 2010). However, there are reports of opposite effects of COMT on functional activation elicited by affective stimuli in prefrontal and limbic brain areas (Smolka et al. 2005; Smolka et al. 2007; Mier et al. 2010). The Met allele has been strongly, although not consistently, associated with enhanced working memory, whereas the Val allele conferred better emotional stimuli processing (Mier et al. 2010). Considerable research suggests that COMT may lead to greater risk to prefrontal functioning and working memory through statistical epistatic interactions with other putative schizophrenia risk genes. For example, COMT and GRM3 modulated frontoparietal functional
chapter 2 0 : I maging G enetics of W orking M emory
313
D O PA M I N E
connectivity, in which Val homozygotes (COMT, rs4680) and A-allele homozygotes (GRM3, rs6465084) had lower DLPFC–inferior parietal connectivity, and Met/G-allele carriers had greater functional connectivity (Tan et al. 2007). Conversely, the effect of COMT and GRM3 genotype was in the opposite direction for VLPFC-inferior parietal connectivity, suggesting that this statistical epistatic interaction may affect prefrontal engagement differently for executive versus emotional stimuli. The D-amino acid oxidase activator (DAOA; G72) gene encodes a protein that degrades D-serine (potent activator of NMDAR) and that is associated with schizophrenia. Statistical epistasis between COMT and DAOA may increase risk of schizophrenia (Nicodemus et al. 2007) and alters efficient prefrontal engagement in the DLPFC (Nixon et al. 2011). The AKT1 gene (encodes the RAC-alpha serine/ threonine-protein kinase) has been implicated as a downstream effector protein of dopamine signaling by indirectly influencing pre-synaptic dopamine transporters (Wei et al. 2007). Carrying the A-allele of AKT1 rs1130233 has been associated with reduced AKT1 levels and increased prefrontal activation during the n-back working memory task, and this effect was predominantly observed in Val homozygotes of the COMT rs4680 marker (Tan et al. 2008). Furthermore, this statistical epistasis was reported between AKT1 and COMT in decrease gray matter volume of the right PFC (BA 47) (Tan et al. 2008). However, a cortical assessment of gene-gene interactions with respect to COMT genotypes and processing of affective stimuli suggested over-fitting of data with a failure to adequately control for multiple testing (Smolka et al., 2007; Puls et al. 2009). γ-AMINOBUTYRIC ACID (GABA)
GABAergic inhibitory neurotransmission in the DLPFC is altered in schizophrenia (Lewis et al. 2004) and is integral to organizing gamma oscillations associated with working memory load (Howard et al. 2003). The neuregulin 1 (NRG1) gene and its receptor ERBB4 have been identified to be integral in GABA circuitry development. ERBB4 promotes the formation of axo-axonic inhibitory synapses over pyramidal cell mediated through NRG1. Thus, ERBB4 regulates the formation of excitatory synapses on GABA interneurons (Fazzari et al. 2010). Knockout of the erbB4 receptor specifically in pyramidal interneurons of mice leads to increased hyperactivity and impaired working memory, suggesting that NGR1 regulates the activity of pyramidal interneurons by promoting GABA release from GABA interneurons. In a sample of 429 healthy controls,
the NRG1 SNP8NRG221533 (rs3575305) marker was not associated with verbal or spatial working memory; however, in the fMRI subsample (n = 85) undergoing the n-back, the number of schizophrenia risk alleles was associated with hyperactivation of the superior frontal gyrus (BA 10) (Krug et al. 2008). Variants in the NRG1, ERBB4, and AKT1 gene may have an epistatic effect on DLPFC activation during working memory. Using a machine learning algorithm in a large neuroimaging sample of healthy controls (n = 172), Nicodemus and colleagues reported an interaction between NRG1 5' and 3' markers in which carriers showed inefficient processing in the DLPFC. Further, there were both two-way and three-way epistatic interactions among NRG1, ERBB4, and AKT1, with a large effect size observed in the three-way interaction (odds ratio = 27.1[3.3–223.0]). These results suggest that NRG1 acts in concert with a complex network of genes for proper working memory function. The major determinant of GABA in the neocortex is glutamic acid decarboxylase-67 (GAD67; encoded by the GAD1 gene). One of the most consistent findings in schizophrenia is down-regulation of GAD1 mRNA and protein in the PFC (Torrey et al. 2005). DNA methylation profile of GAD1 in the PFC shows an 8-fold increase in the promoter region, leading to repressed expression (Huang and Akbarian 2007). Genetic variation in the 5' promoter and untranslated region of GAD1 was associated with child-onset schizophrenia, and increased rate of cortical gray matter loss over a 2- to 8-year period (Addington et al., 2004). Straub and colleagues reported that variation in GAD1 influences multiple cognitive domains, including declarative memory, attention, and working memory in families with schizophrenia. Variants in the promoter region were associated with decreased GAD1 expression in postmortem PFC of schizophrenia patients (Straub et al. 2007). Moreover, in healthy controls, major allele homozygotes had increased and hence potentially inefficient activation during the 2-back working memory paradigm (Straub et al. 2007). OT H E R C ANDIDAT E GE NE S Besides the most important genes and pathways described in the preceding sections, imaging-genetic studies of working memory have investigated several other candidate genes. This section summarizes examples involved in different signaling systems and pathways. The implication of the serotonergic system in working memory function has led to studies on key players in serotonin metabolism. Cerasa et al. (2008) investigated the
3 1 4 part vi : I maging G enetics and C ognition
gene encoding mono-amine oxidase A (MAOA), which is important for the breakdown of serotonin. Participants of the study were 30 healthy males matched for clinical and genetic variables (such as the COMT Val158Met polymorphism) (Cerasa et al. 2008). The authors found increased activity of the right ventrolateral PFC in carriers of the high-activity variants of a variable number of tandem repeats (VNTR) in the promoter region of the gene during an n-back task. Similarly, a study in 49 healthy volunteers investigating the −703G/T (rs4570625) variant in the promoter region of the tryptophan hydroxylase 2 (TPH2) gene, involved in serotonin synthesis, reported increased activity in prefrontal and parietal brain regions in T-allele homozygotes during the n-back test (Reuter et al. 2008). The authors concluded that T-allele homozygotes, who performed worse in previous studies on executive function, may compensate for these deficits by activation of larger brain areas during working memory tasks. In summary, both of these studies support the role of the serotonin system in working memory function. The ε4 variant of the apolipoprotein E (APOE) gene is associated with increased risk for Alzheimer’s disease and cognitive impairment. In healthy subjects without cognitive impairment, it was demonstrated that heterozygous carriers of the ε4 allele had greater activation in the medial frontal and parietal regions compared to ε3 homozygotes during the n-back task, which may reflect a compensatory mechanism for impaired functioning prior to symptom onset of cognitive impairment (Wishart et al. 2006). The synaptosomal-associated protein 25 (SNAP25) plays an important role in pre-synaptic vesicle fusion with the membrane and is involved in neuronal maturation. A study by Soderqvist et al. (2010) investigated several gene variants previously associated with ADHD in two samples of healthy children and young adults. The authors reported a robust association between the A-allele of the SNAP25 rs363039 polymorphism and better visuospatial working memory performance in children and young adults. The polymorphism also influenced parietal grey matter maturation, and the A-allele carriers had greater deactivation or the posterior cingulated cortex, an area supposedly involved in attention networks, during the working memory task. Brain-derived neurotrophic factor (BDNF) is one of the key regulators of neuroplasticity, memory function, and consolidation. The Met allele of the Val66Met (rs6265) polymorphism was associated with abnormal hippocampal activation and poorer episodic memory in schizophrenia patients, unaffected siblings, and healthy controls during a cognitive test battery including the Wechsler Memory Scale, the California Verbal Learning Test, and the Wisconsin
Card Sorting Test (Egan et al. 2003). The authors suggested impaired intracellular trafficking and activity-dependent BDNF-secretion as the underlying mechanism. CYP2D6, a member of the cytochrome P450 family enzyme that is responsible for the metabolism of most medications (Gerretsen et al. 2009), is highly expressed in the brain. Stingl et al. (2012) demonstrated genetically predicted CYP2D6 activity was associated with performance in an n-back working memory task and in an implicit emotional face matching task in a sample of 114 healthy, drug free subjects. Individuals with higher CYP2D6 activity showed higher activation in the fusiform gyrus and the precuneus during the n-back task, and during the face matching task increased activation in the cuneus was observed. In summary, these examples of a variety of candidate genes suggest genetic influence on working memory function through the modification of different pathways and circuits, and they underline the complexity of the genetic interplay influencing working memory and executive function.
chapter 2 0 : I maging G enetics of W orking M emory
315
GE NE S IDE NT IF IE D B Y GE NOME - W IDE AS S OC IAT ION S T U DIE S Technological advances have led to relatively inexpensive genome-wide technology, allowing for hypothesis-free testing of upward of a million common variants (markers with a minor allele frequency greater than 1%) for association with disease traits. Genome-wide association studies (GWAS) on working memory performance have not yielded results reaching the required statistical threshold (for example, see Need et al. 2009). Nevertheless, there are many variants that have repeatedly been associated with psychiatric disorders with working memory dysfunction (Cross-Disorder Group of the Psychiatric Genomics et al. 2013). Since these variants are also associated with the disorder, they are particularly useful in examining intermediate imaging phenotypes. ZNF804A
The zinc-finger 804A gene (ZNF804A) encodes the zinc-finger protein 804A that is expressed broadly in the brain, especially in the developing hippocampus and cortex, as well as the adult cerebellum (Donohoe et al. 2010b). The rs1344706 variant has been implicated in schizophrenia in several GWAS, and results became more significant when patients with bipolar disorder were included, even though the odds ratio was still only slightly above one
(p = 9.96 x 10-9; OR = 1.12) (O’Donovan et al. 2008). Furthermore, the variant may be particularly associated with psychotic symptoms in bipolar disorder (Lett et al. 2011). The conserved region around the rs1344706 variant is a potential binding site for transcription factors, Myt1L zinc-finger protein and POU3F1/Oct-6, which are involved in oligodendrocyte differentiation and proliferation (Riley et al. 2010). Alternatively, the mouse homolog of ZNF804A, zfp804a, is a target for HOXC8, suggesting that the ZNF804Agene may be involved in early neurodevelopment; however, the exact function of ZNF804A is unknown (Chung et al. 2010). Both potential functional mechanisms of rs1344706 are consistent with the proposed mechanism of abnormal myelination and neurodevelopment in bipolar disorder and schizophrenia (McIntosh et al. 2009; Rujescu and Collier 2009; Gill et al. 2010). In an fMRI imaging genetics study employing the n-back working memory paradigm, healthy individuals (n= 115) carrying the ZNF804A risk genotypes exhibited no changes in regional activity, although there was a pronounced gene dosage-dependent alteration in functional connectivity (Esslinger et al. 2009). That is, risk allele carriers had reduced connectivity between the right and left DLPFC and increased connectivity between right DLPFC and left hippocampus. Evidence for prefrontal-hippocampal connectivity as an intermediate phenotype for schizophrenia comes from an independent replication study where it was shown that the ZNF804A risk variant modulated connectivity in healthy controls (n = 153), healthy siblings (n = 178), and patients with schizophrenia (n = 78) (Rasetti et al. 2011). Moreover, the impact of the ZNF804A risk variant on prefrontal-hippocampal connectivity was specific to the working memory task, whereas right-left DLPFC connectivity was modulated by the risk variants during working memory as well as during a face-matching paradigm and during resting state (Esslinger et al. 2011). Indeed, during a theory of mind task (a measure of social cognition), the risk variant was also associated with altered connectivity between medial PFC and the angular gyrus (BA 39) (Walter et al. 2011). Further, healthy controls homozygous for the risk variant were reported to have reduced cortical thickness in the left posterior cingulate cortex, left superior temporal gyrus, and right anterior cingulate cortex, all regions involved in attentional control and working memory (Voineskos et al. 2011). Together, these results support the notion that ZNF804A is conferring risk on basic brain processes involved in proper cognitive function. ZNF804A may also impact heterogeneity within schizophrenia. In a two-stage cognitive study examining cognitive function in schizophrenia patients (n = 297; n = 165)
and controls (n = 165; n = 1475), there was a significant gene-by-diagnosis interaction in both episodic and working memory (Walters et al. 2010). In both samples, schizophrenia patients with the ZNF804A risk genotype performed worse in multiple working memory and episodic memory tasks, although no effect was observed in healthy controls. Moreover, this finding was stronger in patients with lower IQ. Notably, these findings have been independently replicated in a Japanese sample (Hashimoto et al. 2010). This suggests that, first, intermediate phenotypes may be sensitive to subtle effects of GWAS variants; and second, ZNF804A may act in concert with other schizophrenia risk variants to impact working memory function. CACNA1C
The CACNA1C gene encodes the alpha subunit of the L-type voltage-dependent calcium channel CAv1.2. The rs1006737 variant of CACNA1C was associated with bipolar disorder in a meta-analysis of several large independent GWAS (p = 7.0x10-8, OR = 1.18) (Ferreira et al. 2008). In addition, the CACNA1C gene has been reported to be associated with schizophrenia and unipolar depressive disorder (Green et al., 2010; Nyegaard et al. 2010). Furthermore, recent results from the Cross-Disorder Group of the Psychiatric Genomics Consortium show that two markers in genes involved in calcium regulation, CACNA1C and CACNB2, reached genome-wide significance across five disorders (ASD, ADHD, bipolar disorder, major depressive disorder, and schizophrenia) (Cross-Disorder Group of the Psychiatric Genomics et al. 2013). At a lower significance threshold (p < 10-3), 20 of the 67 calcium-active genes were associated with these disorders, suggesting that calcium channels may have a pleiotropic effect on psychopathology. The risk allele of the CACNA1C rs1006737 marker has been associated with increased CACNA1C expression in the DLPFC, increased hippocampal activity during emotional processing, and increased PFC activity during the n-back working memory task (Bigos et al. 2010). During reward and fear processing, healthy controls and first-degree relatives had increased amygdala activation, while bipolar patients were had reduced ventrolateral PFC activation. Healthy risk variant carriers also showed reduced activation of the hippocampus and the subgenual PFC during an episodic memory task (Erk et al. 2010). Most recently, it was found in healthy controls during the n-back working memory task that the risk allele was associated with decreased activation in the DLPFC and increased functional coupling between the DLPFC and the medial temporal lobe (Paulus et al. 2013). Furthermore, in
3 1 6 part vi : I maging G enetics and C ognition
schizophrenia patients and healthy controls, the risk allele was associated with poor working memory performance, whereas the association was in the opposite direction in bipolar patients (Zhang et al. 2012). In summary, these results indicate that CACNA1C may be a modulator of prefrontal function and cortical connectivity, although the direction of effects and their localization showed significant variance across subjects and tasks. Considering the inconsistent direction of effect of the rs1006737 variant, more research is necessary to understand how CACNA1C may impact working memory in different disease populations. MIR137
was no different from healthy control subjects on these neuroimaging measures (Lett et al. 2013). In summary, MIR137 may impact on brain structure and function that are important in engaging working memory, and the effects may be more pronounced in patients with schizophrenia. Considering that MIR137 is involved in networks of genes involved in schizophrenia risk and adult neuroplasticity, future imaging genetics studies may provide key insights into pleiotropic effects of MIR137 in conjunction with other genes on brain functioning during working memory. GE NE - E NV IRONME NT INT E R AC T ION
One of the top GWAS findings from the Psychiatric Genomics Consortium is the association between the microRNA 137 (MIR137) gene and schizophrenia (Schizophrenia Psychiatric Genome-Wide Association Study 2011). The top variant, rs1625579, is located in the intronic region of the MIR137HG gene (MIR137 host gene, non-protein coding; includes MIR137). MIR137 has also been shown to specifically regulate genes with replicated genome-wide significant evidence for a role in schizophrenia, most notably CACNA1C, ZNF804A and TCF4 (Kwon et al. 2011; Kim et al. 2012). Furthermore, microRNA-137 serves as a regulator of adult neural stem cell maturation and migration (Smrt et al. 2010; Szulwach et al. 2010; Sun et al., 2011) in the subventricular zones in proximity to the lateral ventricles and the subgranular zone of the hippocampus. Potkin and colleagues (2010) found that networks of genes regulated by miR-137 reached genome-wide significance for association with prefrontal activation during the Sternberg item recognition working memory paradigm in two independent samples using gene set enrichment analysis (GSEA); subsequently, it was shown that the rs1625579 MIR137 risk genotype predicted elevated DLPFC activation (van Erp et al. 2013). In individuals with high genetic risk for schizophrenia and bipolar disorder, the MIR137 risk genotype was associated with greater activation in the right medial frontal gyrus (BA 6) during the item completion paradigm, suggesting that the variant may be associated with PFC efficiency (Whalley et al. 2012). Finally, MIR137 risk genotype strongly predicted an earlier age at onset of psychosis across four independently collected samples of patients with schizophrenia (Lett et al. 2013a). In an imaging genetics subsample, MIR137 risk genotypes had reduced white matter integrity throughout the brain as well as smaller hippocampi, and larger lateral ventricles (Lett et al. 2013). Furthermore, the brain structure of patients who were carriers of the protective allele
The importance of gene-environment interactions in psychiatric disorders has repeatedly been demonstrated during the past years. These interactions are likely to impact on working memory performance, and imaging phenotypes may be mediated by epigenetic mechanisms (such as histone binding or DNA methylation) or via direct effects (such as activation of monoaminergic neurotransmission on the hypothalamicpituitary-adrenal axis) (Heinz et al. 2012). However, imaging epigenetics are limited by several methodological issues, most importantly by tissue- and region-specific differences in epigenetic modifications, which are hampering the applicability of epigenetic findings obtained in blood or saliva samples to the brain. Postmortem studies found differential histone methylation in patients with psychiatric disorders (Akbarian and Huang 2009). For example, differences in GAD1 methylation, a gene highly relevant to working memory, led to reduced postmortem prefrontal GAD1-mRNA levels (Huang and Akbarian 2007). The only imaging epigenetics study to date on working memory was published by Ursini and colleagues (2011). The authors found methylation in COMT Val allele homozygote individuals negatively correlated with lifetime stress and was positively association with working memory performance. Prefrontal activity during a working memory task was modulated by lifetime stress in a way that cortical efficiency was reduced with higher stress and lower methylation. Although the authors could demonstrate that COMT methylation in the PFC of rats was correlated with COMT methylation in peripheral blood mononuclear cells, the interpretation of the study is limited by the use of peripheral blood cells. Despite this limitation, the findings provide first evidence for the modification of a gene relevant to working memory performance and prefrontal activity by environmental factors and suggest that the rather high heritability rates of working memory performance may
chapter 2 0 : I maging G enetics of W orking M emory
317
partially be influenced by epigenetic mechanisms. Future studies should validate and extend these findings to provide further insight into epigenetic factors of working memory. C L INICA L R E LEVAN CE Imaging genetics findings of the past years significantly contributed to the understanding of underlying genetic, anatomical, and functional mechanisms of working memory dysfunction. Moreover, it may improve treatment for working memory deficits by identification of potential novel drug targets. In addition, response to working memory treatments has a high inter-individual variability (Lett et al. 2013b). Imaging genetics may help identify individuals with expected good or poor response prior to treatment start, which may improve treatment outcomes. Currently, there are very limited pharmacological and non-pharmacological options for working memory improvement available, and all of these have shown only moderate benefits for the patients. Antipsychotic medication has been demonstrated to only slightly improve working memory deficits, and second-generation antipsychotics are no longer considered to be superior compared to first-generation antipsychotics to improve working memory function (Lett et al. 2013b). There is increasing evidence that antipsychotic medication contributes to decreases in frontal volume independent of disease severity (Ho et al. 2011). Nonpharmacological treatments include cognitive remediation therapy that consists of practice exercises, teaching strategies, and compensatory trainings to improve different cognitive domains, and, more recently, repetitive transcranial magnetic stimulation (Lett et al. 2014). Cognitive training is particularly promising, as working memory performance can be trained and the effects can be detected via functional changes in brain activation (Heinzel et al. 2013). Due to these limited options, novel therapeutic approaches for working memory deficits are required. Research has been conducted on substances influencing various targets relevant to working memory function. Among these are D-serine, modifying prefrontal NMDA receptors (Fossat et al. 2012), the GABA-A agonist MK-0777 (Lewis et al. 2004; Buchanan et al. 2011), modafinil (Scoriels et al. 2012), and pregnenolone (Ritsner et al. 2010). However, some of those substances may carry the risk of abuse and even addiction (Heinz et al. 2013). Very few imaging studies to date have investigated the effects of novel substances on brain areas relevant to working memory function. The dopamine D1 receptor agonist dihydrexidine (DAR-0100) has been demonstrated to improve prefrontal perfusion in
schizophrenia subjects after application of a single dose (Mu et al. 2007). However, no working memory improvement after a single dose of dihydrexidine was observed in another study (George et al. 2007). Apud et al. (2006) investigated differential effects of tolcapone, a COMT inhibitor, on cognitive function and cortical information processing by COMT genotype in healthy subjects. The authors found improved functional efficiency in the DLPFC and working memory in Val homozygotes (with supposedly higher COMT function and lower PFC DA) after the administration of tolcapone, while Met homozygotes did worse. In summary, imaging genetics may help identify novel treatment approaches for working memory deficits, and may improve treatment outcome by predicting treatment response. However, only a few studies combining imaging, genetics, and pharmacogenetics have been performed, and more research is required. Considering the important impact of CYP2D6 activity on the metabolism of most psychotropic drugs (for a review, see Kirchheiner et al. 2004) and on working memory function and brain perfusion (Penas et al. 2009; Stingl et al. 2012; Stingl et al. 2013), future genetics studies on medication for the treatment of working memory may benefit from considering CYP2D6 status. LIMITAT IONS OF IMAGING GE NE T IC S T U DIE S OF WOR K ING ME MOR Y There are several limitations to current studies on the imaging genetics of working memory. First, sample sizes are generally low, leading to low power and inflated effect size of genetic association in discovery samples (Button et al. 2013). For example, Munafo et al. (2009) demonstrate that the effect of the 5-HTTLPR (serotonin-transporterlinked polymorphic region) on amygdala activation in the discovery samples are usually much higher than in any replication. Second, rare copy number variants, deletions, or duplications of large areas of the genome may influence imaging correlates of working memory but were not investigated, considering the cost and sample size necessary to detect these variants. Third, currently there are only few studies linking genetic variants to altered gene expression, cell biology, and brain structure, which then led to working memory dysfunction. However, in the near future this is likely to be the standard in imaging genetics studies. Finally, most studies focused on how changes of activation mediate differences in working memory performance, and not vice versa. This led to the primary investigation of brain activation, which may not be directly related to working memory processes.
3 1 8 part vi : I maging G enetics and C ognition
S UM M A RY, FUTURE O UTLO O K, A ND DIRECTI O N
Abi-Dargham A, Mawlawi O, Lombardo I, Gil R, Martinez D, Huang Y, Hwang DR, Keilp J, Kochan L, Van Heertum R, Gorman JM, Laruelle M. (2002). Prefrontal dopamine D1 receptors and working memory in schizophrenia. J Neurosci. 22: 3708–3719. Abi-Dargham A, Moore H. (2003). Prefrontal DA transmission at D1 receptors and the pathology of schizophrenia. Neuroscientist. 9: 404–416. Addington A, Gornick M, Duckworth J, Sporn A, Gogtay N, Bobb A, Greenstein D, Lenane M, Gochman P, Baker N. (2004). GAD1 (2q31. 1), which encodes glutamic acid decarboxylase (GAD67), is associated with childhood-onset schizophrenia and cortical gray matter volume loss. Mol Psychiatry. 10: 581–588. Akbarian S, Huang H-S. (2009). Epigenetic regulation in human brain—focus on histone lysine methylation. Biol Psychiatry. 65: 198–203.
Apud JA, Mattay V, Chen J, Kolachana BS, Callicott JH, Rasetti R, Alce G, Iudicello JE, Akbar N, Egan MF. (2006). Tolcapone improves cognition and cortical information processing in normal human subjects. Neuropsychopharmacology. 32: 1011–1020. Axmacher N, Henseler MM, Jensen O, Weinreich I, Elger CE. Fell J. (2010). Cross-frequency coupling supports multi-item working memory in the human hippocampus. Proc Natl Acad Sci U S A. 107: 3228–3233. Axmacher N, Schmitz DP, Wagner T, Elger CE, Fell J. (2008). Interactions between medial temporal lobe, prefrontal cortex, and inferior temporal regions during visual working memory: a combined intracranial EEG and functional magnetic resonance imaging study. J Neurosci. 28: 7304–7312. Baddeley A. (1992). Working memory. Science. 255: 556–559. Baddeley AD. (1986). Working memory, Oxford: Oxford University Press. Bertolino A, Caforio G, Blasi G, De Candia M, Latorre V, Petruzzella V, Altamura M, Nappi G, Papa S, Callicott JH, Mattay VS, Bellomo A, Scarabino T, Weinberger DR, Nardini M. (2004). Interaction of COMT (Val(108/158)Met) genotype and olanzapine treatment on prefrontal cortical function in patients with schizophrenia. Am J Psychiatry. 161: 1798–1805. Bigos KL, Mattay VS, Callicott JH, Straub RE, Vakkalanka R, Kolachana B, Hyde TM, Lipska BK, Kleinman JE, Weinberger DR. (2010). Genetic variation in CACNA1C affects brain circuitries related to mental illness. Arch Gen Psychiatry. 67: 939. Biswal BB, Mennes M, Zuo XN, Gohel S, Kelly C, Smith SM, Beckmann CF, Adelstein JS, Buckner RL, Colcombe S, Dogonowski AM, Ernst M, Fair D, et al. (2010). Toward discovery science of human brain function. Proc Natl Acad Sci U S A. 107: 4734–4739. Blokland GA, Mcmahon KL, Hoffman J, Zhu G, Meredith M, Martin NG, Thompson PM, De Zubicaray GI, Wright MJ. (2008). Quantifying the heritability of task-related brain activation and performance during the N-back working memory task: a twin fMRI study. Biol Psychology. 79: 70–79. Blokland GA, Mcmahon KL, Thompson PM, Martin NG, De Zubicaray GI, Wright MJ. (2011). Heritability of working memory brain activation. J Neurosci, 31: 10882–10890. Bor J, Brunelin J, Sappey-Marinier D, Ibarrola D, D’amato T, Suaud-Chagny MF, Saoud M. (2011). Thalamus abnormalities during working memory in schizophrenia. An fMRI study. Schizophr Res, 125, 49–53. Buchanan RW, Keefe RS, Lieberman JA, Barch DM, Csernansky JG, Goff DC, Gold JM, Green MF, Jarskog LF, Javitt DC, Kimhy D, Kraus MS, Mcevoy JP, Mesholam-Gately RI, Seidman LJ, Ball MP, Mcmahon RP, Kern RS, Robinson J, Marder SR. (2011). A randomized clinical trial of MK-0777 for the treatment of cognitive impairments in people with schizophrenia. Biol Psychiatry. 69: 442–449. Burmeister M, Mcinnis MG, Zollner S. (2008). Psychiatric genetics: progress amid controversy. Nat Rev Genet. 9: 527–540. Button KS, Ioannidis JP, Mokrysz C, Nosek BA, Flint J, Robinson ES, Munafò MR. (2013). Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci. 14(5): 365–376. Callicott JH, Mattay VS, Verchinski BA, Marenco S, Egan MF, Weinberger DR. (2003). Complexity of prefrontal cortical dysfunction in schizophrenia: more than up or down. Am J Psychiatry. 160: 2209–2215. Castellanos FX, Tannock R. (2002). Neuroscience of attention-deficit/ hyperactivity disorder: the search for endophenotypes. Nat Rev Neurosci. 3: 617–628. Cerasa A, Gioia MC, Fera F, Passamonti L, Liguori M, Lanza P, Muglia M, Magariello A, Quattrone A. (2008). Ventro-lateral prefrontal activity during working memory is modulated by MAO A genetic variation. Brain Res. 1201: 114–121. Charlet K, Beck A, Jorde A, Wimmer L, Vollstädt-Klein S, Gallinat J, Walter H, Kiefer F, Heinz A. (2014). Increased neural activity during high working memory load predicts low relapse risk in alcohol dependence. Addict Biol. 19(3): 402–414.
chapter 2 0 : I maging G enetics of W orking M emory
319
In summary, despite the aforementioned limitations, imaging genetics studies have provided promising insights into how genetics may influence brain structure and function, as well as how that might in turn affect working memory performance. Indeed, almost all of the investigated genes are related to neuroplasticity, such as BDNF, MIR137, and NRG1, or to neurotransmitter signaling in the brain, such as CACNA1C, GAD1, and COMT. GWAS-identified genes may be of especially high importance, since these schizophrenia risk variants might provide deeper insight into genetic mechanisms contributing to working memory deficits in the context of psychiatric disorders. Future imaging genetics studies of working memory will benefit from the novel genetic methods and imaging technologies that are continuously being developed. For instance, whole genome sequencing becomes more affordable. The Human Connectome Project (http://www.human connectomeproject.org/) will provide the connection between brain structure and function with unprecedented detail to address fundamental questions about brain variability and its relationship to working memory (Biswal et al. 2010). Furthermore, as sample sizes continue to increase, future studies will be sufficiently powered to employ advanced statistical methods such as structural equational modeling. The novel CLARITY method for the examination of detailed brain structures using light microscopy might also help to gain deeper knowledge of brain organization in psychiatric disorders (Chung et al. 2013). Taken together, imaging genetics can improve insight into genetics, anatomy, and pathophysiology of working memory deficits and may ultimately lead to better treatment options. R EF ERENCES
Chen J, Lipska BK, Halim N, Ma QD, Matsumoto M, Melhem S, Kolachana BS, Hyde TM, Herman MM, Apud J, Egan MF, Kleinman JE, Weinberger DR. (2004). Functional analysis of genetic variation in catechol-O-methyltransferase (COMT): effects on mRNA, protein, and enzyme activity in postmortem human brain. Am J Hum Genet. 75: 807–821. Chiang M-C, Mcmahon KL, De Zubicaray GI, Martin NG, Hickie I, Toga AW, Wright MJ, Thompson PM. (2011). Genetics of white matter development: a DTI study of 705 twins and their siblings aged 12 to 29. Neuroimage. 54: 2308–2317. Chung HJ, Lee J-Y, Deocaris CC, Min H, Kim SH, Kim MH. (2010). Mouse homologue of the schizophrenia susceptibility gene ZNF804A as a target of Hoxc8. J Biomed Biotechnol. 2010: 231708. Chung K, Wallace J, Kim S-Y, Kalyanasundaram S, Andalman AS, Davidson TJ, Mirzabekov JJ, Zalocusky KA, Mattis J, Denisin AK. (2013). Structural and molecular interrogation of intact biological systems. Nature. 497(7449): 332–337. Cole MW, Yarkoni T, Repovs G, Anticevic A, Braver TS. (2012). Global connectivity of prefrontal cortex predicts cognitive control and intelligence. J Neurosci. 32: 8988–8999. Cross-Disorder Group of the Psychiatric Genomics, Smoller JW, Craddock N, Kendler K, Lee PH, Neale BM, Nurnberger JI, Ripke S, Santangelo S, Sullivan PF. (2013). Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis. Lancet. 381: 1371–1379. D’esposito M, Postle BR, Rypma B. (2000). Prefrontal cortical contributions to working memory: evidence from event-related fMRI studies. Exp Brain Res. 133: 3–11. Deserno L, Sterzer P, Wustenberg T, Heinz A, Schlagenhauf F. (2012). Reduced prefrontal-parietal effective connectivity and working memory deficits in schizophrenia. J Neurosci. 32: 12–20. Donohoe G, Frodl T, Morris D, Spoletini I, Cannon DM, Cherubini A, Caltagirone C, Bossu P, Mcdonald C, Gill M, Corvin AP, S palletta G. (2010a). Reduced occipital and prefrontal brain volumes in dysbindin-associated schizophrenia. Neuropsychopharmacology. 35: 368–373. Donohoe G, Morris DW, Clarke S, McGhee KA, Schwaiger S, Nangle JM, Garavan H, Robertson IH, Gill M, Corvin A. (2007). Variance in neurocognitive performance is associated with dysbindin-1 in schizophrenia: a preliminary study. Neuropsychologia. 45: 454–458. Donohoe G, Morris DW, Corvin A. (2010b). The psychosis susceptibility gene ZNF804A: associations, functions, and phenotypes. Schizophr Bull. 36: 904–909. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, Goldman D, Weinberger DR. (2001). Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci. 98: 6917–6922. Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, Zaitsev E, Gold B, Goldman D, Dean M, Lu B, Weinberger DR. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 112: 257–269. Egan MF, Straub RE, Goldberg TE, Yakub I, Callicott JH, Hariri AR, Mattay VS, Bertolino A, Hyde TM, Shannon-Weickert C, Akil M, Crook J, Vakkalanka RK, Balkissoon R, Gibbs RA, Kleinman JE, Weinberger DR. (2004). Variation in GRM3 affects cognition, prefrontal glutamate, and risk for schizophrenia. Proc Natl Acad Sci U S A. 101: 12604–12609. Erk S, Meyer-Lindenberg A, Schnell K, Opitz Von Boberfeld C, Esslinger C, Kirsch P, Grimm O, Arnold C, Haddad L, Witt SH, Cichon S, Nothen MM, Rietschel M, Walter H. (2010). Brain function in carriers of a genome-wide supported bipolar disorder variant. Arch Gen Psychiatry. 67: 803–811. Esslinger C, Kirsch P, Haddad L, Mier D, Sauer C, Erk S, Schnell K, Arnold C, Witt SH, Rietschel M, Cichon S, Walter H Meyer-Lindenberg A. (2011). Cognitive state and connectivity
effects of the genome-wide significant psychosis variant in ZNF804A. Neuroimage. 54: 2514–2523. Esslinger C, Walter H, Kirsch P, Erk S, Schnell K, Arnold C, Haddad L, Mier D, Opitz Von Boberfeld C, Raab K, Witt SH, Rietschel M, Cichon S, Meyer-Lindenberg A. (2009). Neural mechanisms of a genome-wide supported psychosis variant. Science. 324: 605. Fazzari P, Paternain AV, Valiente M, Pla R, Luján R, Lloyd K, Lerma J, Marín O, Rico B. 2010. Control of cortical GABA circuitry development by Nrg1 and ErbB4 signalling. Nature. 464: 1376–1380. Ferreira MAR, O’Donovan MC, Meng YA, Jones IR, Ruderfer DM, Jones L, Fan J, Kirov G, Perlis RH, Green EK, et al. (2008). Collaborative genome-wide association analysis supports a role for ANK3 and CACNA1C in bipolar disorder. Nature Genetics. 40: 1056–1058. Finch CE. (2002). Neurons, glia, and plasticity in normal brain aging. Adv Gerontol. 10: 35–39. Forbes NF, Carrick LA, Mcintosh AM, Lawrie SM. (2009). Working memory in schizophrenia: a meta-analysis. Psychol Med. 39: 889–905. Fossat P, Turpin FR, Sacchi S, Dulong J, Shi T, Rivet J-M, Sweedler JV, Pollegioni L, Millan MJ, Oliet SH. (2012). Glial D-serine gates NMDA receptors at excitatory synapses in prefrontal cortex. Cerebral Cortex. 22: 595–606. George MS, Molnar CE, Grenesko EL, Anderson B, Mu Q, Johnson K, Nahas Z, Knable M, Fernandes P, Juncos J, Huang X, Nichols DE, Mailman RB. (2007). A single 20 mg dose of dihydrexidine (DAR-0100), a full dopamine D1 agonist, is safe and tolerated in patients with schizophrenia. Schizophr Res. 93: 42–50. Gerretsen P, Muller DJ, Tiwari A, Mamo D, Pollock BG. (2009). The intersection of pharmacology, imaging, and genetics in the development of personalized medicine. Dialogues Clin Neurosci. 11. 363–376. Gill M, Donohoe G, Corvin A. (2010). What have the genomics ever done for the psychoses? Psychol Medicine. 40: 529–540. Gilmore JH, Kang C, Evans DD, Wolfe HM, Smith JK, Lieberman JA, Lin W, Hamer RM, Styner M, Gerig G. (2010a). Prenatal and neonatal brain structure and white matter maturation in children at high risk for schizophrenia. Am J Psychiatry. 167: 1083. Gilmore JH, Schmitt JE, Knickmeyer RC, Smith JK, Lin W, Styner M, Gerig G, Neale MC. (2010b). Genetic and environmental contributions to neonatal brain structure: a twin study. Hum Brain Mapp. 31: 1174–1182. Gilmore JH, Shi F, Woolson SL, Knickmeyer RC, Short SJ, Lin W, Zhu H, Hamer RM, Styner M, Shen D. (2012). Longitudinal development of cortical and subcortical gray matter from birth to 2 years. Cerebral Cortex. 22: 2478–2485. Glahn DC, Winkler AM, Kochunov P, Almasy L, Duggirala R, Carless MA, Curran JC, Olvera RL, Laird AR, Smith SM, Beckmann CF, Fox PT, Blangero J. (2010). Genetic control over the resting brain. Proc Natl Acad Sci U S A. 107: 1223–1228. Goldman-Rakic PS. (1996). Regional and cellular fractionation of working memory. Proc Natl Acad Sci U S A. 93: 13473–13480. Goldman-Rakic PS, Castner SA, Svensson TH, Siever LJ, Williams GV. (2004). Targeting the dopamine D1 receptor in schizophrenia: insights for cognitive dysfunction. Psychopharmacology (Berl). 174: 3–16. Goldman AL, Pezawas L, Mattay VS, Fischl B, Verchinski BA, Chen Q, Weinberger DR, Meyer-Lindenberg A. (2009). Widespread reductions of cortical thickness in schizophrenia and spectrum disorders and evidence of heritability. Arch Gen Psychiatry. 66: 467. Gottesman II, Gould TD. (2003). The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry. 160: 636–645. Green EK, Grozeva D, Jones I, Jones L, Kirov G, Caesar S, Gordon-Smith K, Fraser C, Forty L, Russell E, Hamshere ML, Moskvina V, Nikolov I, Farmer A, Mcguffin P, Wellcome Trust Case Control, Holmans PA, Owen MJ, O’Donovan MC, Craddock N. (2010). The bipolar disorder risk allele at CACNA1C also confers risk of recurrent major depression and of schizophrenia. Mol Psychiatry. 15: 1016–1022. Greenwood TA, Braff DL, Light GA, Cadenhead KS, Calkins ME, Dobie DJ, Freedman R, Green MF, Gur RE, Gur RC, Mintz J,
3 2 0 part vi : I maging G enetics and C ognition
Nuechterlein KH, Olincy A, et al. (2007). Initial heritability analyses of endophenotypic measures for schizophrenia: the consortium on the genetics of schizophrenia. Arch Gen Psychiatry. 64: 1242–1250. Guillin O, Abi-Dargham A, Laruelle M. (2007). Neurobiology of dopamine in schizophrenia. Int Rev Neurobiol. 78: 1–39. Hallmayer JF, Kalaydjieva L, Badcock J, Dragovic M, Howell S, Michie PT, Rock D, Vile D, Williams R, Corder EH, Hollingsworth K, Jablensky A. (2005). Genetic evidence for a distinct subtype of schizophrenia characterized by pervasive cognitive deficit. Am J Hum Genet. 77: 468–476. Hashimoto R, Ohi K, Yasuda Y, Fukumoto M, Iwase M, Iike N, Azechi M, Ikezawa K, Takaya M, Takahashi H, Yamamori H, Okochi T, Tanimukai H, Tagami S, Morihara T, Okochi M, Tanaka T, Kudo T, Kazui H, Iwata N, Takeda M. (2010). The impact of a genome-wide supported psychosis variant in the ZNF804A gene on memory function in schizophrenia. Am J Med Genet B Neuropsychiatr Genet. 153B: 1459–1464. Heinz A, Kipke R, Muller S, Wiesing U. (2013). True and false concerns about neuroenhancement: a response to ‘Neuroenhancers, addiction and research ethics’, by D M Shaw. J Med Ethics. 40(4): 286–287. Heinz A, Smolka MN. (2006). The effects of catechol O-methyltransferase genotype on brain activation elicited by affective stimuli and cognitive tasks. Rev Neurosci. 17: 359–367. Heinz AJ, Beck A, Meyer-Lindenberg A, Sterzer P, Heinz A. (2011). Cognitive and neurobiological mechanisms of alcohol-related aggression. Nat Rev Neurosci. 12: 400–413. Heinzel S, Schulte S, Onken J, Duong QL, Riemer TG, Heinz A, Kathmann N, Rapp MA. (2013). Working memory training improvements and gains in non-trained cognitive tasks in young and older adults. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn. 21(2): 146–173. Ho BC, Andreasen NC, Ziebell S, Pierson R, Magnotta V. (2011). Long-term antipsychotic treatment and brain volumes: a longitudinal study of first-episode schizophrenia. Arch Gen Psychiatry. 68: 128–137. Howard MW, Rizzuto DS, Caplan J.B, Madsen JR, Lisman J, Aschenbrenner-Scheibe R, Schulze-Bonhage A, Kahana MJ. (2003). Gamma oscillations correlate with working memory load in humans. Cereb Cortex. 13: 1369–1374. Huang HS, Akbarian S. (2007). GAD1 mRNA expression and DNA methylation in prefrontal cortex of subjects with schizophrenia. PLoS One. 2: e809. Jentsch JD, Trantham-Davidson H, Jairl C, Tinsley M, Cannon TD Lavin A. (2009). Dysbindin modulates prefrontal cortical glutamatergic circuits and working memory function in mice. Neuropsychopharmacology. 34: 2601–2608. Karlsgodt KH, Kochunov P, Winkler AM, Laird AR, Almasy L, Duggirala R, Olvera RL, Fox PT, Blangero J, Glahn DC. (2010). A multimodal assessment of the genetic control over working memory. J Neuroscience. 30: 8197–8202. Karlsgodt KH, Robleto K, Trantham-Davidson H, Jairl C, Cannon TD, Lavin A, Jentsch JD. (2011). Reduced dysbindin expression mediates N-methyl-D-aspartate receptor hypofunction and impaired working memory peformance. Biol Psychiatry. 69: 28–34. Kim AH, Parker EK, Williamson V, McMichael GO, Fanous AH, Vladimirov VI. (2012). Experimental validation of candidate schizophrenia gene ZNF804A as target for hsa-miR-137. Schizophr Res. 141: 60–64. Kirchheiner J, Nickchen K, Bauer M, Wong ML, Licinio J, Roots I, Brockmoller J. (2004). Pharmacogenetics of antidepressants and antipsychotics: the contribution of allelic variations to the phenotype of drug response. Mol Psychiatry. 9: 442–473. Kopp F, Schroger E, Lipka S. (2006). Synchronized brain activity during rehearsal and short-term memory disruption by irrelevant speech is affected by recall mode. Int J Psychophysiol. 61: 188–203. Korotkova T, Fuchs EC, Ponomarenko A, Von Engelhardt J, Monyer H. (2010). NMDA receptor ablation on parvalbumin-positive interneurons impairs hippocampal synchrony, spatial representations, and working memory. Neuron. 68: 557–569.
Krug A, Markov V, Eggermann T, Krach S, Zerres K, Stocker T, Shah NJ, Schneider F, Nothen MM, Treutlein J, Rietschel M, Kircher T. (2008). Genetic variation in the schizophrenia-risk gene neuregulin1 correlates with differences in frontal brain activation in a working memory task in healthy individuals. Neuroimage. 42: 1569–1576. Kwon E, Wang W, Tsai LH. (2011). Validation of schizophrenia-associated genes CSMD1, C10orf26, CACNA1C and TCF4 as miR-137 targets. Mol Psychiatry. 18(1): 11-12. Lee J, Park S. (2005). Working memory impairments in schizophrenia: a meta-analysis. J Abnorm Psychology. 114: 599. Lett TA, Chakravarty MM, Felsky D, Brandl EJ, Tiwari AK, Gonçalves VF, Rajji TK, Daskalakis ZJ, Meltzer HY, Lieberman JA. (2013). The genome-wide supported microRNA-137 variant predicts phenotypic heterogeneity within schizophrenia. Mol Psychiatry. 18(4): 443-450. Lett TA, Voineskos AN, Kennedy JL, Levine B, Daskalakis ZJ. (2014). Treating working memory deficits in schizophrenia: a review of the neurobiology. Biol Psychiatry. 75(5): 361–370. Lett TA, Zai CC, Tiwari AK, Shaikh SA, Likhodi O, Kennedy JL, Muller DJ. (2011). ANK3, CACNA1C and ZNF804A gene variants in bipolar disorders and psychosis subphenotype. World J Biol Psychiatry. 12: 392–397. Lewis DA, Volk DW, Hashimoto T. (2004). Selective alterations in prefrontal cortical GABA neurotransmission in schizophrenia: a novel target for the treatment of working memory dysfunction. Psychopharmacology (Berl). 174: 143–150. Maher B. (2008). Personal genomes: the case of the missing heritability. Nature. 456: 18–21. Malhotra AK, Pinals DA, Adler CM, Elman I, Clifton A, Pickar D, Breier A. (1997). Ketamine-induced exacerbation of psychotic symptoms and cognitive impairment in neuroleptic-free schizophrenics. Neuropsychopharmacology. 17: 141–150. Markov V, Krug A, Krach S, Jansen A, Eggermann T, Zerres K, Stocker T, Shah NJ, Nothen MM, Treutlein J, Rietschel M, Kircher T. (2010). Impact of schizophrenia-risk gene dysbindin 1 on brain activation in bilateral middle frontal gyrus during a working memory task in healthy individuals. Hum Brain Mapp. 31: 266–275. Mcintosh AM, Hall J, Lymer GKS, Sussmann JED, Lawrie SM. (2009). Genetic risk for white matter abnormalities in bipolar disorder. Int Rev Psychiatry. 21: 387–393. Mesulam MM. (1998). From sensation to cognition. Brain. 121(Pt 6): 1013–1052. Meyer-Lindenberg A. (2009). Neural connectivity as an intermediate phenotype: brain networks under genetic control. Hum Brain Mapp. 30: 1938–1946. Meyer-Lindenberg A, Kohn PD, Kolachana B, Kippenhan S, Mcinerney-Leo A, Nussbaum R, Weinberger DR, Berman KF. (2005a). Midbrain dopamine and prefrontal function in humans: interaction and modulation by COMT genotype. Nat Neurosci. 8: 594–596. Meyer-Lindenberg A, Poline JB, Kohn PD, Holt JL, Egan MF, Weinberger DR, Berman KF. (2001). Evidence for abnormal cortical functional connectivity during working memory in schizophrenia. Am J Psychiatry. 158: 1809–1817. Meyer-Lindenberg A, Weinberger DR. (2006). Intermediate phenotypes and genetic mechanisms of psychiatric disorders. Nat Rev Neurosci. 7: 818–827. Meyer-Lindenberg AS, Olsen RK, Kohn PD, Brown T, Egan MF, Weinberger DR, Berman KF. (2005b). Regionally specific disturbance of dorsolateral prefrontal-hippocampal functional connectivity in schizophrenia. Arch Gen Psychiatry. 62: 379–386. Mier D, Kirsch P, Meyer-Lindenberg A. (2010). Neural substrates of pleiotropic action of genetic variation in COMT: a meta-analysis. Mol Psychiatry. 15: 918–927. Miller EK, Cohen JD. (2001). An integrative theory of prefrontal cortex function. Ann Rev Neurosci. 24: 167–202. Minzenberg MJ, Laird AR, Thelen S, Carter CS, Glahn DC. (2009). Meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia. Arch Gen Psychiatry, 66: 811–822.
chapter 2 0 : I maging G enetics of W orking M emory
321
Moghaddam B. (2004). Targeting metabotropic glutamate receptors for treatment of the cognitive symptoms of schizophrenia. Psychopharmacology (Berl). 174: 39–44. Mu Q, Johnson K, Morgan PS, Grenesko EL, Molnar CE, Anderson B, Nahas Z, Kozel FA, Kose S, Knable M, Fernandes P, Nichols DE, Mailman RB, George MS. (2007). A single 20 mg dose of the full D1 dopamine agonist dihydrexidine (DAR-0100) increases prefrontal perfusion in schizophrenia. Schizophr Res. 94: 332–341. Munafo MR, Freimer NB, Ng W, Ophoff R, Veijola J, Miettunen J, Jarvelin MR, Taanila A, Flint J. (2009). 5-HTTLPR genotype and anxiety-related personality traits: a meta-analysis and new data. Am J Med Genet B Neuropsychiatr Genet. 150B: 271–281. Need AC, Attix DK, Mcevoy JM, Cirulli ET, Linney KL, Hunt P, Ge D, Heinzen EL, Maia JM, Shianna KV, Weale ME, Cherkas LF, Clement G, Spector TD, Gibson G, Goldstein DB. (2009). A genome-wide study of common SNPs and CNVs in cognitive performance in the CANTAB. Hum Mol Genet. 18: 4650–4661. Nickl-Jockschat T, Stocker T, Markov V, Krug A, Huang R, Schneider F, Habel U, Zerres K, Nothen MM, Treutlein J, Rietschel M, Shah NJ, Kircher T. (2012). The impact of a Dysbindin schizophrenia susceptibility variant on fiber tract integrity in healthy individuals: a TBSS-based diffusion tensor imaging study. Neuroimage. 60: 847–853. Nicodemus KK, Kolachana BS, Vakkalanka R, Straub RE, Giegling I, Egan MF, Rujescu D, Weinberger DR. (2007). Evidence for statistical epistasis between catechol-O-methyltransferase (COMT) and polymorphisms in RGS4, G72 (DAOA), GRM3, and DISC1: influence on risk of schizophrenia. Hum Genet. 120: 889–906. Nicodemus KK, Law AJ, Radulescu E, Luna A, Kolachana B, Vakkalanka R, Rujescu D, Giegling I, Straub RE, McGee K, Gold B, Dean M, Muglia P, Callicott JH, Tan HY, Weinberger DR. (2010). Biological validation of increased schizophrenia risk with NRG1, ERBB4, and AKT1 epistasis via functional neuroimaging in healthy controls. Arch Gen Psychiatry. 67: 991–1001. Nixon DC, Prust MJ, Sambataro F, Tan HY, Mattay VS, Weinberger DR, Callicott JH. (2011). Interactive effects of DAOA (G72) and catechol-O-methyltransferase on neurophysiology in prefrontal cortex. Biol Psychiatry. 69(10): 1006–1008. Norman DA, Shallice T. (1983). Attention to action: willed and automatic-control of behavior. Bull Psychonom Soc. 21: 354–354. Nyegaard M, Demontis D, Foldager L, Hedemand A, Flint TJ, Sørensen KM, Andersen PS, Nordentoft M, Werge T, Pedersen CB, Hougaard DM, Mortensen PB, Mors O, Børglum AD. (2010). CACNA1C (rs1006737) is associated with schizophrenia. Mol Psychiatry. 15: 119–121. O’Donovan MC, Craddock N, Norton N, Williams H, Peirce T, Moskvina V, Nikolov I, Hamshere M, Carroll L, Georgieva L, Dwyer S, Holmans P, Marchini JL, Spencer CCA, Howie B, et al. (2008). Identification of loci associated with schizophrenia by genome-wide association and follow-up. Nat Genetics. 40: 1053–1055. Okubo Y, Suhara T, Suzuki K, Kobayashi K, Inoue O, Terasaki O, Someya Y, Sassa T, Sudo Y, Matsushima E, Iyo M, Tateno Y, Toru M. (1997). Decreased prefrontal dopamine D1 receptors in schizophrenia revealed by PET. Nature. 385: 634–636. Paulus FM, Bedenbender J, Krach S, Pyka M, Krug A, Sommer J, Mette M, Nothen MM, Witt SH, Rietschel M, Kircher T, Jansen A. (2013). Association of rs1006737 in CACNA1C with alterations in prefrontal activation and fronto-hippocampal connectivity. Hum Brain Mapp. 35(4): 1190–1200. Payne L, Kounios J. (2009). Coherent oscillatory networks supporting short-term memory retention. Brain Res. 1247: 126–132. Peñas-LLedó EM, Dorado P, Pacheco R, González I, LLerena A. (2009). Relation between CYP2D6 genotype, personality, neurocognition and overall psychopathology in healthy volunteers. Pharmacogenomics. 10: 1111–1120. Petrides M, Baddeley A. (1996). Specialized systems for the processing of mnemonic information within the primate frontal cortex. Philos Trans R Soc Lond B Biol Sci, 351: 1455–1461; discussion 1461–1462.
Postle BR. (2006). Working memory as an emergent property of the mind and brain. Neuroscience. 139: 23–38. Potkin SG, Macciardi F, Guffanti G, Fallon JH, Wang Q, Turner JA, Lakatos A, Miles MF, Lander A, Vawter MP, Xie X. (2010). Identifying gene regulatory networks in schizophrenia. Neuroimage. 53: 839–847. Puls I, Mohr J, Wrase J, Vollstadt-Klein S, Lemenager T, Vollmert C, Rapp M, Obermayer K, Heinz A, Smolka MN. (2009). A model comparison of COMT effects on central processing of affective stimuli. Neuroimage. 46: 683–691. Rasetti R, Sambataro F, Chen Q, Callicott JH, Mattay VS, Weinberger DR. (2011). Altered cortical network dynamics: a potential intermediate phenotype for schizophrenia and association with ZNF804A. Arch Gen Psychiatry. 68: 1207–1217. Reuter M, Esslinger C, Montag C, Lis S, Gallhofer B, Kirsch P. (2008). A functional variant of the tryptophan hydroxylase 2 gene impacts working memory: A genetic imaging study. Biol Psychology. 79: 111–117. Riley B, Thiselton D, Maher BS, Bigdeli T, Wormley B, Mcmichael GO, Fanous AH, Vladimirov V, O’Neill FA, Walsh D, Kendler KS. (2010). Replication of association between schizophrenia and ZNF804A in the Irish Case-Control Study of Schizophrenia sample. Mol Psychiatry. 15: 29–37. Ritsner MS, Gibel A, Shleifer T, Boguslavsky I, Zayed A, Maayan R, Weizman A, Lerner V. (2010). Pregnenolone and dehydroepiandrosterone as an adjunctive treatment in schizophrenia and schizoaffective disorder: an 8-week, double-blind, randomized, controlled, 2-center, parallel-group trial. J Clin Psychiatry. 71: 1351–1362. Rujescu D, Collier DA. (2009). Dissecting the many genetic faces of schizophrenia. Epidemiologia Psichiatr Soc. 18: 91–95. Sarnthein J, Petsche H, Rappelsberger P, Shaw GL, Von Stein A. (1998). Synchronization between prefrontal and posterior association cortex during human working memory. Proc Natl Acad Sci U S A. 95: 7092–7096. Sauseng P, Klimesch W, Doppelmayr M, Hanslmayr S, Schabus M, Gruber WR. (2004). Theta coupling in the human electroencephalogram during a working memory task. Neurosci Lett. 354: 123–126. Sauseng P, Klimesch W, Schabus M, Doppelmayr M. (2005). Fronto-parietal EEG coherence in theta and upper alpha reflect central executive functions of working memory. Int J Psychophysiol. 57: 97–103. Sawaguchi T, Goldman-Rakic PS. (1991). D1 dopamine receptors in prefrontal cortex: involvement in working memory. Science. 251: 947–950. Schizophrenia Psychiatric Genome-Wide Association Study. (2011). Genome-wide association study identifies five new schizophrenia loci. Nat Genet. 43: 969–976. Schlagenhauf F, Rapp MA, Huys QJ, Beck A, Wustenberg T, Deserno L, Buchholz HG, Kalbitzer J, Buchert R, Bauer M, Kienast T, Cumming P, Plotkin M, Kumakura Y, Grace AA, Dolan RJ, Heinz A. (2013). Ventral striatal prediction error signaling is associated with dopamine synthesis capacity and fluid intelligence. Hum Brain Mapp. 34: 1490–1499. Schlosser R, Gesierich T, Kaufmann B, Vucurevic G, Hunsche S, Gawehn J, Stoeter P. (2003). Altered effective connectivity during working memory performance in schizophrenia: a study with fMRI and structural equation modeling. Neuroimage. 19: 751–763. Scoriels L, Barnett JH, Soma PK, Sahakian BJ, Jones PB. (2012). Effects of modafinil on cognitive functions in first episode psychosis. Psychopharmacology (Berl). 220: 249–258. Serrien DJ, Pogosyan AH, Brown P. (2004). Influence of working memory on patterns of motor related cortico-cortical coupling. Exp Brain Res. 155: 204–210. Smolka MN, Buhler M, Schumann G, Klein S, Hu XZ, Moayer M, Zimmer A, Wrase J, Flor H, Mann K, Braus DF, Goldman D, Heinz A. (2007). Gene-gene effects on central processing of aversive stimuli. Mol Psychiatry, 12: 307–317. Smolka MN, Schumann G, Wrase J, Grusser SM, Flor H, Mann K, Braus DF, Goldman D, Buchel C, Heinz A. (2005).
3 2 2 part vi : I maging G enetics and C ognition
Catechol-O-methyltransferase val158met genotype affects processing of emotional stimuli in the amygdala and prefrontal cortex. J Neurosci. 25: 836–842. Smrt RD, Szulwach KE, Pfeiffer RL, Li X, Guo W, Pathania M, Teng ZQ, Luo Y, Peng J, Bordey A, Jin P, Zhao X. (2010). MicroRNA miR-137 regulates neuronal maturation by targeting ubiquitin ligase mind bomb-1. Stem Cells. 28: 1060–1070. Soderqvist S, McNab F, Peyrard-Janvid M, Matsson H, Humphreys K, Kere J, Klingberg T. (2010). The SNAP25 gene is linked to working memory capacity and maturation of the posterior cingulate cortex during childhood. Biol Psychiatry. 68: 1120–1125. Stingl JC, Brockmoller J, Viviani R. (2013). Genetic variability of drug-metabolizing enzymes: the dual impact on psychiatric therapy and regulation of brain function. Mol Psychiatry. 18: 273–287. Stingl JC, Esslinger C, Tost H, Bilek E, Kirsch P, Ohmle B, Viviani R, Walter H, Rietschel M, Meyer-Lindenberg A. (2012). Genetic variation in CYP2D6 impacts neural activation during cognitive tasks in humans. Neuroimage. 59: 2818–2823. Straub R, Lipska B, Egan M, Goldberg T, Callicott J, Mayhew M, Vakkalanka R, Kolachana B, Kleinman J, Weinberger D. (2007). Allelic variation in GAD1 (GAD67) is associated with schizophrenia and influences cortical function and gene expression. Mol Psychiatry. 12: 854–869. Sun G, Ye P, Murai K, Lang MF, Li S, Zhang H, Li W, Fu C, Yin J, Wang A, Ma X, Shi Y. (2011). miR-137 forms a regulatory loop with nuclear receptor TLX and LSD1 in neural stem cells. Nat Commun. 2: 529. Szulwach KE, Li X, Smrt RD, Li Y, Luo Y, Lin L, Santistevan NJ, Li W, Zhao X, Jin P. (2010). Cross talk between microRNA and epigenetic regulation in adult neurogenesis. J Cell Biol. 189: 127–141. Talbot K, Eidem WL, Tinsley CL, Benson MA, Thompson EW, Smith RJ, Hahn CG, Siegel SJ, Trojanowski JQ, Gur RE, Blake DJ, Arnold SE. (2004). Dysbindin-1 is reduced in intrinsic, glutamatergic terminals of the hippocampal formation in schizophrenia. J Clin Invest. 113: 1353–1363. Tan HY, Chen Q, Sust S, Buckholtz JW, Meyers JD, Egan MF, Mattay VS, Meyer-Lindenberg A, Weinberger DR, Callicott JH. (2007). Epistasis between catechol-O-methyltransferase and type II metabotropic glutamate receptor 3 genes on working memory brain function. Proc Natl Acad Sci. 104: 12536–12541. Tan HY, Nicodemus KK, Chen Q, Li Z, Brooke JK, Honea R, Kolachana BS, Straub RE, Meyer-Lindenberg A, Sei Y, Mattay VS, Callicott JH, Weinberger DR. (2008). Genetic variation in AKT1 is linked to dopamine-associated prefrontal cortical structure and function in humans. J Clin Invest. 118: 2200–2208. Thompson PM, Cannon TD, Narr KL, Van Erp T, Poutanen V-P, Huttunen M, Lönnqvist J, Standertskjöld-Nordenstam C-G, Kaprio J, Khaledy M. (2001). Genetic influences on brain structure. Nat Neurosci. 4: 1253–1258. Torrey EF, Barci BM, Webster MJ, Bartko JJ, Meador-Woodruff JH, Knable MB. (2005). Neurochemical markers for schizophrenia, bipolar disorder, and major depression in postmortem brains. Biol Psychiatry. 57: 252–260. Tost H, Bilek E, Meyer-Lindenberg A. (2012). Brain connectivity in psychiatric imaging genetics. Neuroimage. 62: 2250–2260. Ursini G, Bollati V, Fazio L, Porcelli A, Iacovelli L, Catalani A, Sinibaldi L, Gelao B, Romano R, Rampino A, Taurisano P, Mancini M, Di Giorgio A, Popolizio T, Baccarelli A, De Blasi A, Blasi G, Bertolino A. (2011). Stress-related methylation of the catechol-Omethyltransferase Val 158 allele predicts human prefrontal cognition and activity. J Neurosci. 31: 6692–6698. Van Erp TG, Guella I, Vawter MP, Torri F, Ford JM, Lim KO, Bustillo J, Belger A, Preda A, Nguyen D, Turner JA, Mathalon DH, Macciardi F, Potkin SG. (2013). Abstracts of the 14th International Congress on Schizophrenia Research (ICOSR). April 21-15, 2013. Orlando Grande Lakes, Florida, USA. Schizophr Bull. 39(Suppl 1): S1–376.
Viding E, Blakemore S-J. (2007). Endophenotype approach to developmental psychopathology: implications for autism research. Behav Genetics. 37: 51–60. Vijayraghavan S, Wang M, Birnbaum SG, Williams GV, Arnsten AF. (2007). Inverted-U dopamine D1 receptor actions on prefrontal neurons engaged in working memory. Nat Neurosci. 10: 376–384. Voineskos AN, Lerch JP, Felsky D, Tiwari A, Rajji TK, Miranda D, Lobaugh NJ, Pollock BG, Mulsant BH, Kennedy JL. (2011). The ZNF804A gene: characterization of a novel neural risk mechanism for the major psychoses. Neuropsychopharmacology. 36: 1871–1878. Walter H, Schnell K, Erk S, Arnold C, Kirsch P, Esslinger C, Mier D, Schmitgen MM, Rietschel M, Witt SH, Nothen MM, Cichon S, Meyer-Lindenberg A. (2011). Effects of a genome-wide supported psychosis risk variant on neural activation during a theory-of-mind task. Mol Psychiatry, 16: 462–470. Walter H, Vasic N, Hose A, Spitzer M, Wolf RC. (2007). Working memory dysfunction in schizophrenia compared to healthy controls and patients with depression: evidence from event-related fMRI. Neuroimage. 35: 1551–1561. Walters JT, Corvin A, Owen MJ, Williams H, Dragovic M, Quinn EM, Judge R, Smith DJ, Norton N, Giegling I. (2010). Psychosis susceptibility gene ZNF804A and cognitive performance in schizophrenia. Arch Gen Psychiatry. 67: 692. Wang M, Yang Y, Wang CJ, Gamo NJ, Jin LE, Mazer JA, Morrison JH, Wang XJ, Arnsten AF. (2013). NMDA receptors subserve persistent neuronal firing during working memory in dorsolateral prefrontal cortex. Neuron, 77, 736–749. Wei Y, Williams JM, Dipace C, Sung U, Javitch JA, Galli A, Saunders C. (2007). Dopamine transporter activity mediates amphetamine-induced inhibition of Akt through a Ca2+/ calmodulin-dependent kinase II-dependent mechanism. Mol Pharmacol. 71: 835–842. Whalley HC, Papmeyer M, Romaniuk L, Sprooten E, Johnstone EC, Hall J, Lawrie SM, Evans KL, Blumberg HP, Sussmann JE, Mcintosh AM. (2012). Impact of a microRNA MIR137 susceptibility variant on brain function in people at high genetic risk of schizophrenia or bipolar disorder. Neuropsychopharmacology. 37: 2720–2729. Williams NM, Preece A, Morris DW, Spurlock G, Bray NJ, Stephens M, Norton N, Williams H, Clement M, Dwyer S, Curran C, Wilkinson J, Moskvina V, Waddington JL, Gill M, Corvin AP, Zammit S, Kirov G, Owen MJ, O’Donovan, M. C. (2004). Identification in 2 independent samples of a novel schizophrenia risk haplotype of the dystrobrevin binding protein gene (DTNBP1). Arch Gen Psychiatry. 61: 336–344. Wishart H, Saykin A, Rabin L, Santulli R, Flashman L, Guerin S, Mamourian A, Belloni D, Rhodes C, McAllister T. (2006). Increased brain activation during working memory in cognitively intact adults with the APOE ε4 allele. Am J Psychiatry. 163: 1603–1610. Wolf C, Jackson MC, Kissling C, Thome J, Linden DE. (2011). Dysbindin-1 genotype effects on emotional working memory. Mol Psychiatry. 16: 145–155. Wolf RC, Vasic N, Sambataro F, Hose A, Frasch K, Schmid M, Walter H. (2009). Temporally anticorrelated brain networks during working memory performance reveal aberrant prefrontal and hippocampal connectivity in patients with schizophrenia. Prog Neuropsychopharmacol Biol Psychiatry. 33: 1464–1473. Wolf RC, Walter H. (2005). Evaluation of a novel event-related parametric fMRI paradigm investigating prefrontal function. Psychiatry Res. 140: 73–83. Zhang Q, Shen Q, Xu Z, Chen M, Cheng L, Zhai J, Gu H, Bao X, Chen X, Wang K, Deng X, Ji F, Liu C, Li J, Dong Q, Chen C. (2012). The effects of CACNA1C gene polymorphism on spatial working memory in both healthy controls and patients with schizophrenia or bipolar disorder. Neuropsychopharmacology. 37: 677–684.
chapter 2 0 : I maging G enetics of W orking M emory
323
PA RT V I I . IMAGING GENETICS OF COGNITIVE AGING
21. NEUROBIOLOGY AND GENETICS OF COGNITIVE AGING I N SI GH T S F ROM NEU ROI M AGING STUDIE S
John C. Muse, Milap A. Nowrangi, Daniel R. Weinberger, and Venkata S. Mattay
INTRODUCTI O N As individuals advance in age, the ability to maintain an independent and good quality of life is dependent not only on their health status but also on their level of cognitive function. While cognitive decline was once thought to be an inevitable consequence of the normal aging process, research suggests that the onset of age-associated cognitive impairment could potentially be delayed (Whalley et al. 2004). The mechanisms by which we might intervene in this process and potentially halt its progression, however, have not been completely explored. While understanding the mechanisms underlying cognitive aging is complex and challenging, it also presents opportunities. For example, identifying the factors that have allowed some individuals to retain their cognitive abilities may be an important strategy to discover mechanisms or strategies to curb the decline in other individuals. Additionally, identifying risk factors could help target individuals for focused preventative strategies. There is an age-related increase in cognitive heterogeneity whereby individual differences in one’s level of cognitive function, as well as the rate at which cognitive changes occur, increase with increasing age. Twin and familial studies have shown that the majority of individual cognitive ability is inherited and that the impact of an individual’s genes on cognition is still significant in older individuals (Lee, Henry, et al. 2010). Additionally, multiple studies have confirmed that the human life span itself is moderately heritable and that maintaining cognitive ability into older age is a characteristic of extremely long-lived humans, which may be passed on to their descendants (Schupf, Costa, et al. 2004; Barral, Cosentino, et al. 2012; Brooks-Wilson 2013; Sanders, Minster, et al. 2014). This evidence suggests that identifying the genetic variants
associated with maintaining cognitive abilities into one’s later years and those associated with increased risk for early decline may be a useful strategy to uncover the molecular pathways involved in age-related cognitive decline. Over the last couple of decades, advances in neuroimaging techniques have facilitated the investigation of changes in brain structure and function that occur with increasing age, thereby helping uncover some of the neuronal substrates underlying cognitive aging. This research has begun to answer the question of where in the brain changes that contribute to cognitive decline occur with increasing age, and has also shown the remarkable functional plasticity of the aging human brain to compensate for declines. Neuroimaging findings have also helped generate theories about cognitive aging and have begun the process of understanding, at the level of brain structure and function, how some individuals maintain cognitive function throughout their life spans. More recently, imaging genetics, a form of genetic association approach in which neuroimaging measures are the phenotypes of interest, is being increasingly used to study not only study complex genetic diseases such as schizophrenia, autism, and depression, but also cognitive aging (Hariri and Weinberger 2003; Mattay et al. 2008; Petrella et al. 2008). The neuroimaging measures used are considered intermediate phenotypes in that they are a quantifiable risk factor for the complex disease or phenotype of interest that is presumably less complex and thus easier to detect in genetic association studies. The potential utility of using intermediate phenotypes rests on the principle that genetic effect on a particular phenotype will be stronger the closer it is to the level of the gene’s action. The goal of this chapter is to present current knowledge about the genetic influences on cognitive aging and the utility of imaging genetics of cognitive aging. Toward 327
this end, we will structure this chapter in the following sections: age-related cognitive changes; neuronal substrates of cognitive aging; genetics underlying individual variability in cognitive aging; and imaging genetics of cognitive aging: principles and examples.
AGE- R EL ATE D CO G N I TI VE CHAN G ES WHAT COGNITIVE CHANGES OCCUR WITH INCREASING AGE?
To date there is a consistent body of literature supporting a general pattern of age-related changes in cognitive function in which crystallized abilities (general knowledge and acquired skills) are relatively preserved with increasing age, while fluid abilities (reasoning, pattern recognition, and novel problem-solving) gradually decline after they peak, typically in one’s twenties or thirties (Salthouse 2009; Vance 2012 Harada, Natelson Love, et al. 2013). For the purposes of this chapter, we will narrow our focus in this section to specific cognitive domains that are typically investigated in regard to normal age-related changes: namely, cognitive (psychomotor) speed, attention, executive functions, and memory. Normal aging is accompanied by a gradual reduction in the rate at which cognitive tasks are executed (processing speed). This is thought to contribute, at least in part, to age-related declines in other cognitive functions such as working and episodic memory. In fact, older individuals tend to do worse on tasks that have some time constraint, regardless of which specific ability is being tested (Ardila 2007), and they have also been shown to exhibit prolonged completion time (lower performance) on processing-speed tasks in the absence of any sensory or motor dysfunction (Bowie and Harvey 2006). With increasing age, it also becomes more difficult to selectively maintain one’s attention in situations with multiple competing stimuli, which has in part provided evidence for the “inhibitory deficit theory of cognitive aging” (Oberauer 2001; Healey, Campbell, et al. 2008; Drag and Bieliauskas 2010; Pettigrew and Martin 2014). This ability to divide one’s attention processes is even more age-sensitive than selective attention (McDowd and Craik 1988; Glisky 2007; Drag and Bieliauskas 2010) and attention processing may play a role in many of the other age-sensitive cognitive functions (Fernandes and Moscovitch 2000; Adam 2011). Executive function, the higher order cognitive domain with many sub-processes (including working memory, cognitive control, regulation of goal-directed behavior, and the real-time manipulation/processing of information) has also
been shown to be age-sensitive, with declines starting in the twenties (Arnsten and Li 2005; Drag and Bieliauskas 2010; Albert 2011). Further, multiple studies have shown that as working memory task complexity increases, there is a greater performance gap between older and younger individuals (Bopp and Verhaeghen 2005; Verhaeghen, Cerella, et al. 2006; Oosterman, Boeschoten, et al. 2014). Due to its association with dementia, the domain of memory has been extensively investigated in aging research. It involves aspects that are age-resistant as well as those that are age-sensitive. For example, long-term semantic memory and procedural memory typically are maintained until at least very late in one’s life. This is in contrast to episodic memory, which is more age-sensitive and typically starts to decline around age sixty. Even within the domain of episodic memory, there are various sub-processes that may or may not be the primary cause of decline for any given individual. For example, a failure to recall information could be due to a failure or faulty encoding of that information, or it could be the result of dysfunctional retrieval of such information. There is evidence suggesting that older individuals typically perform worse on tasks of free recall, both immediate and delayed, but tend to do almost as well as younger individuals on tasks of recognition (Glisky 2007; Drag and Bieliauskas 2010; Albert 2011; Harada, Natelson Love, et al. 2013). TO WHAT EXTENT ARE THERE INDIVIDUAL D I F F E R E N C E S I N C O G N I T I V E A B I L I T Y, A N D D O THEY CHANGE WITH INCREASING AGE?
An age-related increase in cognitive heterogeneity, that is, an increase in inter-individual differences, is commonly reported as a characteristic of older populations (Mattay, Goldberg, et al. 2008; Rabbitt 2011). Age-related increases in performance variability have been reported for a variety of tasks (Ardila 2007; McArdle and Plassman 2009). For example, Ardila (2007) investigated age-related changes in average performance, as well as score variability, on the various sub-tests included in the WAIS-III across different age groups and reported that the oldest age group had the lowest mean scores for all tests and that dispersion increased with age for all tests except for digit span. Interestingly, the magnitude of score dispersion (range from 20% to over 200%) was test-specific. While Salthouse (2011) has argued that individual differences in rates of cognitive aging may be relatively small, Rabbitt (2011) and Raz and Linderberger (2011) present evidence to support the claim that there is an age-related increase in inter-individual variation in some cognitive abilities, especially those that are most sensitive to the effects of aging.
3 2 8 part vii : I maging G enetics of C ognitive A ging
W H AT A C C O U N T S F O R T H E VA R I A B I L I T Y, AND WHY DOES IT MATTER?
An older individual’s cognitive ability assessed at any one point in time is going to be related to a number of factors, including his or her previous peak level of ability, the magnitude of change that has taken place since that peak, how long it has taken for that change to take place, and occasion-specific error (Harris and Deary 2011). Besides the occasion-specific error, the remainder of these factors are all likely to be determined by both environmental and genetic influences. To date there have been a plethora of environmental influences shown to be important predictors of cognitive ability and/or cognitive change at older ages including, but not limited to, one’s level of physical fitness (Kramer, Colcombe, et al. 2005; Deary, Corley, et al. 2009; Gomez-Pinilla and Hillman 2013), early life psychiatric issues (Brown 2010; Tschanz, Pfister, et al. 2013; van den Kommer, Comijs, et al. 2013) and living environment (Fors, Lennartsson, et al. 2009), level of educational attainment (Anstey and Christensen 2000; Deary, Corley, et al. 2009; Reuser, Willekens, et al. 2011; Lam, Eng, et al. 2013), occupational complexity (Finkel, Andel, et al. 2009); and age of retirement (Rohwedder and Willis 2010). In addition to environmental influences, familial and twin studies have indicated that heritability may account for up to 50% of the observed variance in cognitive abilities among individuals, including general cognitive ability (Hariri and Weinberger 2003; Mattay, Goldberg, et al. 2008; Deary, Johnson et al. 2009; Payton 2009; Harris and Deary 2011; Lee and Sachdev 2014). Furthermore, the influence of genetics on an individual’s cognitive abilities does not appear to remain constant throughout the life span, as evidenced by changes in heritability estimates with age. A review of twin studies investigating genetic influences on cognitive abilities in older individuals reported that heritability estimates for level of processing speed and executive functioning were positively correlated with age, whereas there was a negative correlation in measures of verbal, spatial, and general ability (Lee, Henry, et al. 2010). The rate of the observed cognitive changes also seems to be influenced by genetic variation (Mattay and Goldberg 2007; Mattay et al. 2010; Papenberg et al. 2013), either independently or via epistasis/gene-environment interaction (Payton 2009; Lee, Henry, et al. 2010). Taken together, there appears to be ample evidence to suggest that genetics plays a major role in determining an individual’s level of cognitive ability at any one point in his or her life, and may possibly play a modest role in the rate at which this ability changes over time.
T H E NE U RONAL S U B S T R AT E S OF C OGNIT IV E AGING Neuoimaging methods can be (and have been) used extensively to better understand the neurobiological changes that take place, which are thought to cause and/or precede the changes in cognitive ability observed via behavioral/ cognitive assessment. We will now describe the current state of neuroimaging research aimed at uncovering the neurobiological mechanisms of cognitive aging, prefaced with a brief survey of the structural and functional changes that occur at the cellular level, which will provide a necessary background from which we can interpret the neuroimaging measures. CELLULAR ASPECTS OF COGNITIVE AGING
Contrary to initial reports that were compromised by poor stereological techniques, studies using newer quantitative neuroanatomical methods for counting neurons suggest that, in contrast to Alzheimer’s disease (AD) and other neurodegenerative disorders where there is extensive neuronal loss, normal aging is not associated with a significant neuronal loss (for a review, see Pannese 2011). On the other hand, evidence from studies in humans and animal models suggest that normal aging is associated with a decrease in the number and length of dendrites, loss of dendritic spines, and decrease in the number of axons that undergo segmental demyelination followed by re-myelination and a significant loss of synapses (for a review, see Pannese 2011). Additionally, in neocortical neurons there is an age-related decline in perikaryon volume, but not nuclear volume (Stark, Toft, et al. 2007) or mitochondrial volume (Bertoni-Freddari, Fattoretti, et al. 2007). In general these studies seem to suggest that age-related cognitive decline is more likely to be associated with alterations in neuronal architecture and synaptic connectivity than with neuronal loss (Morrison and Baxter, Nat Rev Neurosci 2012). Evidence from electron microscopy studies also suggests that the molecular mechanisms of synaptic aging are not global but tend to exhibit regional heterogeneity. In particular, there is a difference in the nature of synaptic aging in the prefrontal cortex, which subserves working memory, and the hippocampus, which underlies declarative memory function, the two cognitive processes that are most vulnerable to aging (for a review, see Morrison and Baxter 2012). In the prefrontal cortex there is extensive loss of axospinous processes, with a decrease in the thin plastic dendritic spines by almost 50%, while other spine classes are relatively stable (Dickstein, Weaver, et al. 2013). Aging
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
329
in the hippocampus, on the other hand, is characterized by a loss of previously potentiated perforated synapses, which are critical for induction and maintenance of long-term potentiation. While the changes in synaptic architecture provide a structural basis for altered neuronal communication, electrophysiological recordings provide direct physiological evidence of age-related changes in neuronal communication. Extensive evidence from studies in animals suggests that there is a decline in the intrinsic excitability of hippocampal pyramidal neurons located within the cornu ammonis (CA) 1 region with increasing age, which is linked to declines in learning and memory (Wu, Chan, et al. 2004; Disterhoft and Oh 2007; Oh, Oliveira, et al. 2010). This decline in excitability is driven primarily by an increase in the slow after-hyperpolarization current (sAHP), which normally is involved in spike frequency adaptation by limiting neuronal firing during high frequency periods (Wu, Chan, et al. 2004). Additionally, there is an age-related increase in the density of L-type voltage-gated calcium channels (L-VGCC) in hippocampal pyramidal neurons (Oh, Oliveira, et al. 2010). While alterations in hippocampal biophysical properties have been associated with age-related declines in learning and memory, a study by Wang, Gamo, et al. (2011) in non-human primates provides evidence for the electrophysiological basis for age-related declines in working memory. An important sub-process in working memory tasks is the ability to maintain goal-relevant information over a delay period during which the stimulus is not available. The cellular basis for this cognitive ability has been attributed to the persistent firing of prefrontal cortex neurons during this delay period. Wang et al. (2011) found that a decline in persistent firing during the delay period, which begins in middle age and becomes progressively worse with increasing age, accounts for the decline in working memory performance in non-human primates. Furthermore, the authors linked the decline in firing to an increase in cyclic AMP (cAMP) signaling, which resulted in the opening of HCN and KCNQ potassium (K+) channels. One possible explanation for this signaling cascade would be a reduction in alpha 2A adrenergic receptors that have been observed in the PFC of older monkeys (Wang, Gamo, et al. 2011). MACROSCOPIC STRUCTURAL CHANGES
The human cerebral cortex undergoes dynamic changes, globally and regionally, throughout an individual’s lifetime. Brain morphometric studies with neuroimaging in general report results that are in line with postmortem studies by
Critchley (1931), namely that there are declines in global gray matter and white matter, as well as an increase in cerebral spinal fluid volume, with increasing age (Good 2001; Raz, Rodrigue, et al. 2007; Drag and Bieliauskas 2010). In their seminal review of age-related changes in brain morphometry, Raz and Rodrigue (2006) indicate that there is regional variation in the rate of cortical decline, with the lateral prefrontal cortex exhibiting a linear decline with age of about 5% per decade after the age of 20 and the medial temporal lobe, including the hippocampus, showing a non-linear change with age, with the largest declines generally occurring after 60 years. While volumes of regions like the primary visual cortex have been found to be relatively stable across the life span, evidence from both cross-sectional and longitudinal studies shows that striatal volume declines with age, with the caudate showing the fastest rate of decline of up to 0.83% per year (Gunning-Dixon, Head, et al. 1998; Raz et al. 2003). To address discrepant findings across studies possibly resulting from differences in segmentation procedures and selection of regions, Walhovd, Westlye, et al. (2011) examined structural MRI data from five independent samples, with the same segmentation tools, and reported that the results were largely replicable across samples. To address potential confounds arising from cross-sectional studies, in a recent study Pfefferbaum et al. (2013) examined data from subjects who had longitudinal MRI data (2 –6 MRIs over a 1–8 year interval) and reported that while the volume of the pre-central and post-central cortices showed a linear decline with age, the volumes of temporal, calcarine, occipital, and subcortical structures such as the thalamus, caudate, putamen, and amygdala showed non-linear changes, suggesting accelerated changes at older age. Over the last decade, while automated voxel based morphometric (VBM) analysis of high-resolution structural MRI has been a commonly used approach to assess brain morphometric changes associated with normal aging (Good, Ashburner, et al. 2001; Matsuda 2013), approaches that allow the estimation of surface-based measures such as cortical thickness and surface area present a viable alternative to assess subtle cortical changes. Using such an approach, Lemaitre, Goldman, et al. (2012) reported a linear decline with advancing age in prefrontal cortical thickness and volume, but in the parietal region, only cortical thickness but not volume changed with age, suggesting that these different morphometric measures—gray matter volume, cortical thickness, and surface area—may show differential regional sensitivity to age-related anatomical change. Age-related vulnerability of the prefrontal cortex to cortical thinning has also been reported in a large
3 3 0 part vii : I maging G enetics of C ognitive A ging
multi-sample analysis (Fjell, Westlye, et al. 2009). This report additionally noted that the anterior temporal lobe and anterior cingulate showed the greatest stability with increasing age. Importantly, these age-related trends of greater cortical thinning in the anterior/posterior cortices, with a lesser effect being seen in the temporal/occipital cortices, have been reported in a longitudinal analysis as well (Thambisetty, Wan, et al. 2010). One report investigated the similarities and differences between volumetric analysis and cortical thickness, studying age-related changes, and found that volumetric measures contain a greater amount of variability than cortical thickness measures, suggesting that cortical thickness may be a more sensitive measure (Hutton, Draganski, et al. 2009; Fjell and Walhovd 2010). Importantly, the authors noted that both methods may provide independent information and could therefore be considered complementary. Changes in other measures of cortical morphology have also been reported to accompany advanced age. For example, Liu, Wen, et al. (2010) found that the sulci become less curved, and become wider and shallower with age, as estimated by a global sulcal index (g-SI), with the superior frontal sulcus being most affected. The authors note that g-SI is in contrast to a relative preservation of gyrification index (GI) with increasing age, which is the ratio of total and exposed cortical surface. With regard to any regional differences in the volume of white matter adjacent to cortical regions of interest, Salat, Greve, et al. (2009) reported that while the majority of regions exhibited some degree of age-related decline, the bilateral fusiform, orbital frontal, superior frontal, and the middle/inferior temporal gyri regions of interest (ROI) exhibited the greatest decline. The authors also report that most of the regions followed a nonlinear path, suggesting that increasing age accelerates the rate of decline. One region that is consistently studied but has yielded inconsistent results is the medial temporal lobe (MTL), including the hippocampus. Part of the inconsistencies may be due to the regional susceptibility of this area to agerelated changes (Sullivan, Marsh, et al. 1995; Rodrigue and Raz 2004; Mueller and Weiner 2009; Rajah, Kromas, et al. 2010; Ta, Huang, et al. 2011). There is evidence to suggest regional heterogeneity of age-related changes within the hippocampus with an anterior-posterior age-related gradient whereby the more anterior areas show greater age-related reductions than posterior areas (Chen, Chuah et al. 2010). Although these differences may seem subtle, they may prove to be important as the anterior and posterior aspects of the hippocampal region differ greatly in their inputs and outputs to cortical and other subcortical
areas (Eichenbaum 2000; Eichenbaum, Yonelinas, et al. 2007; Fanselow and Dong 2010). Overall, while the in vivo morphometric measures are generally in line with postmortem data, there are inconsistencies, possibly reflecting the susceptibility of MRI-based measurements to changes in vascular compartments and other non-neural changes that contribute to overall tissue volume measures. Many of these macroscopic age-related changes have been correlated with performance changes on various cognitive tests (Fjell and Walhovd 2010; Kaup, Mirzakhanian, et al. 2011). While morphometric changes in prefrontal areas have typically been linked with performance changes on tasks related to executive functions (Picq, Aujard, et al. 2010; Burzynska, Nagel, et al. 2011), performance changes on an episodic memory task have been correlated with morphometric changes in the medial temporal region as well as the prefrontal cortex (Head, Rodrigue, et al. 2008; Chen, Chuah, et al. 2010; Burgmans, van Boxtel, et al. 2011; Shing 2011). In general, the trend in older individuals tends to be that “bigger is better,” since most reports find that a decline in performance is correlated with a decline in structural abundance as well. WHITE MATTER INTEGRITY
Over the last decade, diffusion weighted imaging (DWI) and diffusion tensor imaging (DTI) have been increasingly used to assess white matter integrity, structure, and connectivity (Basser and Pierpaoli 1996; de Figueiredo, Borgonovi, et al. 2011). The DTI method, in particular, has been applied extensively in the study of cognitive aging and has been reviewed in depth elsewhere (Salat, Tuch, et al. 2005; Sullivan and Pfefferbaum 2006; Minati, Grisoli, et al. 2007; Madden, Bennett, et al. 2009; Madden, Bennett, et al. 2012; Sexton, Walhovd, et al. 2014). In general there is a decline in fractional anisotropy (FA), a measure of orientational coherence that represents fiber integrity, with increasing age, which follows an anterior to posterior gradient. This is in parallel to an increase in mean diffusivity (MD), which is higher in less ordered tissue and is thought to represent an increase in water content. In addition to FA and MD, additional measures such as radial diffusivity (RD), which reflects the degree of myelination, and axial diffusivity (AD), which is a measure of fiber coherence, also show regional susceptibility to change with increasing age. Using these DTI measures, Burzynska, Preuschhof, et al. (2010) reported five distinct region-specific patterns of age-related changes and suggested that these different patterns may be attributed to distinct biological processes such as demyelination, Wallerian degeneration, gliosis, and
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
331
severe fiber loss with advancing age. Longitudinal studies of age-related alteration in white matter report an acceleration of change with age with significant annual decrease in FA and increase in AD, RD, and MD, particularly in the superior brain regions beginning in the fifth decade (Sexton, Walhovd, et al. 2014). Age-related changes in these DTI measures have also been correlated with performance on various cognitive tests. Ziegler, Piguet, et al. (2010) found a positive correlation between cognitive control tasks and frontal lobe FA, as well as a positive correlation between episodic memory tasks and posterior FA values. Kennedy and Raz (2009) reported regional variation in the correlation between age-related white matter changes and cognitive performance, with anterior changes accounting for declines in processing speed, posterior changes accounting for declines in performance on task of inhibition and task switching, and central changes accounting for declines in episodic memory. In another study, Yassa, Muftuler, et al. (2010), using ultra-high sub-millimeter resolution DTI, reported a decline in diffusion signal in the perforant pathway with age and an associated decrease in delayed recall performance. Magnetization transfer ratio (MTR), a measure derived from magnetization transfer imaging (MTI), reflects the efficiency of magnetization transfer between protons of unbound water molecules in tissue to those bound by macromolecules in the same tissue (Rovaris, Iannucci, et al. 2003; Filippi and Rocca 2007; Inglese, Ge, et al. 2007). As tissue organization decreases, so does the ratio. In a study including both DTI and MTI, Venkatraman, Aizenstein, et al. (2011) found that both the values of white matter FA and MTR peaked independently in older adults and explained variance in processing speed performance, suggesting that these measures were complementary. Age-related tissue changes can affect the MRI contrast properties of neural structures. Using T1-weighted images, Salat, Lee, et al. (2009) investigated age-related changes in T1-weighted gray matter and white matter tissue intensities throughout the cortex and found that gray matter intensity declined with age throughout most of the cortical mantel, with the strongest effects being observed in the medial frontal, anterior cingulate, and inferior temporal areas. White matter intensity also declined most prominently in the superior/medial frontal, cingulum, and medial/lateral temporal regions. The ratio of gray matter to white matter intensity was also found to be altered by age throughout the cortical mantle with the superior/inferior frontal, lateral parietal, superior temporal, and precuneus regions being most age-sensitive. Interestingly, these observed age effects were significantly stronger than changes in cortical
thickness. The authors speculate that the changes in signal intensity are most likely due to changes in myelination, as T1-weighted images are sensitive to macromolecules such as fats. In another study, Saito, Sakai, et al. (2009) report both age- and region-specific differences in gray matter and white matter T1 signal. Most notably, the authors found that throughout the life span, anterior regions have a longer gray matter T1 and shorter white matter T1 than posterior regions, possibly suggesting a higher level of myelination in the prefrontal regions. In another study, Westlye, Walhovd et al. (2010) used MRI signal properties to differentiate the phenomenon of cortical thinning during the process of maturation versus that of aging. The authors found a posterior to anterior maturation gradient in intracortical T1 signal with T2* signal following a similar trend but later in life. An age-related increase in the correlation between cortical thickness and gray matter signal intensity was also reported, which the authors suggest indicates that measures of cortical thickness become more sensitive with increasing age. It is important that these age-related changes in tissue signal properties be kept in mind when interpreting results that are at least partially dependent on tissue classification based on signal intensity. There is an age-related increase in T1 white matter signal intensity, which could result in it being misclassified as gray matter ( Jernigan 2001). Additionally, measures such as cortical thickness rely at least in some part on signal contrast differences to delineate the boundary between white matter and gray matter (Dale, Fischl, et al. 1999). Indeed, it has been found that including signal intensity measures as covariates in morphometric analysis can increase the sensitivity of these measures of interest such as cortical thickness (Westlye, Walhovd, et al. 2009). Overall there is ample evidence to show that MRI has become a powerful tool to investigate in vivo changes in brain structure that occur with increasing age and how these changes relate to changes in cognitive function. One problem with these studies, though, is that the underlying biological meaning is not fully understood for many of these MRI-based measures. Unfortunately the technology is not yet at a point to detect structural changes at the cellular level, and so we must rely on postmortem analysis, as well as animal studies, to gather this information. MOLECULAR AND NEUROMETABOLIC ASPECTS OF COGNITIVE AGING
There is a great deal of evidence supporting age-related changes in most neurotransmitter systems, including
3 3 2 part vii : I maging G enetics of C ognitive A ging
the cholinergic (Court, Lloyd, et al. 1997; Picciotto and Zoli 2002; Dumas and Newhouse 2011; van Waarde, Ramakrishnan et al. 2011), histamine (Yanai, Watanabe, et al. 1992; Terao, Steininger, et al. 2004), GABAergic (Sundman-Eriksson and Allard 2006; Rissman, De Blas, et al. 2007; Duncan, Webster, et al. 2010), cannabinoid (Liu, Bilkey, et al. 2003; Van Laere, Goffin, et al. 2008; Canas, Duarte, et al. 2009), glutamatergic (Ikonomovic 1999; Liu, Smith, et al. 2008; Zahr, Mayer, et al. 2008; Canas, Duarte, et al. 2009; Magnusson, Brim, et al. 2010), adenosine (Meyer, Elmenhorst, et al. 2007; Canas, Duarte, et al. 2009), and opioid (Hiller, Fan, et al. 1993; Kawamura, Kimura, et al. 2003; Gerald, Howlett, et al. 2008; van Waarde, Ramakrishnan, et al. 2011) systems. Human in vivo studies with positron emission tomography (PET) and single photon emission computerized tomography substantiate this evidence. Nearly all aspects of the dopaminergic system are sensitive to the effects of increasing age, including disruptions in normal dopamine synthesis and breakdown (Burchinsky 1984). Additionally, human postmortem studies have found age-related declines in both D1-like (D1) and D2-like (D2) receptors (Burchinsky 1984; Bäckman, Lindenberger, et al. 2010). Neuroimaging modalities have validated these results by showing age-related declines in regional dopamine (DA) D1 binding potentials (Iyo and Yamasaki 1993; Karlsson, Nyberg, et al. 2009; Bäckman, Karlsson, et al. 2011; Rieckmann, Karlsson, et al. 2011; Rieckmann, Karlsson, et al. 2011) and D2 binding potentials (Antonini, Leenders, et al. 1993; Kaasinen, Vilkman, et al. 2000). Additionally, declines in midbrain dopamine transporter (DAT) density (Volkow 1994; Erixon-Lindroth, Farde, et al. 2005) and dopamine storage as assessed using 6-[18F] fluorodopamine (Kumakura, Vernaleken, et al. 2005; Kumakura, Vernaleken, et al. 2010) have been reported. This evidence indicates that there are clearly widespread significant age-related changes in the dopamine system, which is of great interest given the known relationship between this system and higher order cognitive functions. Widespread age-related changes in the serotonin system have also been reported (Nobler, Mann, et al. 1999). Most studies have found that increasing age is associated with a decline in 5-HT2 receptor availability (Marcusson, Morgan, et al. 1984; Moses-Kolko, Price, et al. 2011; Uchida, Chow, et al. 2011) and affinity (Arranz, Eriksson, et al. 1993). Both age and sex differences in both pre- and post-synaptic 5-HT1 receptors were reported by Moses-Kolko, Price, et al. (2011). While there are reports of a 20%–30% decline in these receptors with age (Shiroma, Geda, et al. 2010), there are also reports of no change with age (Rabiner, Messa,
et al. 2002). One study found that 5-HT4 receptors were relatively stable with increasing age (Madsen, Haahr, et al. 2011), while another report found a significant decline (Marner, Gillings, et al. 2010). Interestingly, training on a Pavlovian/instrumental auto-shaping task has been shown to be associated with reduced mRNA expression of 5-HT6 receptor in the prefrontal cortex, hippocampaus, and striatum (Huerta-Rivas, Pérez-García, et al. 2010). Consistent with this, another animal study showed that blockade of 5-HT6 receptors can restore performance in aged mice on memory tasks (Da Silva Costa-Aze, Dauphin, et al. 2011), suggesting that the 5-HT6 receptor may be a useful therapeutic target for age-related cognitive deficits. The serotonin transporter (SERT) is primarily responsible for synaptic serotonin re-uptake, thereby controlling synaptic serotonin activity as a function of level and duration (Daws and Gould 2011). To date, numerous studies have provided evidence for regional age-related declines in SERT availability (Pirker, Asenbaum, et al. 2000; van Dyck, Malison, et al. 2000; Yamamoto, Suhara, et al. 2002; Hesse, Barthel, et al. 2003), with one study noting that this decline may not be apparent until after 30 years of age (Buchert, Schulze, et al. 2006). Altogether, these results suggest that the serotonergic system is vulnerable to the effects of increasing age. While these are just a few illustrations of the in vivo changes in neurotransmission associated with advancing age, the significance of this approach to cognitive aging is supported by work reporting correlations between measures of alterations in cognition, behavior, motor coordination, and PET measures of neuroreceptor function associated with advancing age (Breier et al. 1998; Volkow et al. 1998). PET is also used to study changes associated with pathological aging, such as the level of amyloid beta (AB) deposition through the use of various radioligands including the [11C] Pittsburgh Compound-B (PIB) (Vallabhajosula 2011) and Florbetapir (Choi et al. 2009; Rodrigue 2012). AB deposits are considered to be a hallmark trait of AD pathology (Hardy 2009), but it has long been known that cognitively normal individuals can have cerebral AB burden levels similar to individuals with a clinical diagnosis of AD (Rodrigue, Kennedy, et al. 2009). Indeed, PM analysis have indicated that 25%–45% of cognitively normal individuals have clinical levels of AD pathology, while in vivo PET studies have reported that 10%–30% of cognitively normal individuals are PIB-positive and the pattern of regional binding is similar to that seen in AD with the prefrontal cortex, lateral/medial parietal, lateral temporal cortices, and striatum being primarily affected (Rabinovici and Jagust 2009). The authors suggest that the presence of AB may precede cognitive change and could potentially
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
333
prove to be an indicator of future neuronal dysfunction, although its association with cognitive scores was inconclusive in healthy older individuals. Alternately, these findings could suggest that the cognitively normal individuals who are PIB-positive have a higher cognitive reserve and use compensatory mechanisms to counter the ill effects of increased amyloid deposition. In fact, the cognitive reserve hypothesis was born out of an attempt to make sense of the fact that the brains of cognitively normal individuals can have AB levels equal to that of severely demented individuals. Studying this subpopulation in greater depth with genomics, epigenomics, and multi-modal neuroimaging may enable the identification of mechanisms that confer higher cognitive reserve and facilitate the discovery of strategies to slow cognitive aging. Another method of assessment, 18-fluorodeoxyglucose (18-FDG) PET, is commonly used to assess regional glucose utilization via the uptake of radiolabeled glucose molecules (Herholz 2011). This method is commonly used in AD research as it has been commonly reported that AD patients have regional hypometabolism at rest, with the most common regions cited being bilateral parietal, temporal, frontal, and posterior cingulate areas (Mosconi 2005). In a study using optimized VBM and resting-state 18-FDG PET with correction for partial volume effects, Kalpouzos, Chételat, et al. (2009) reported an age-related decline in 18-FDG uptake in the bilateral superior frontal areas, while there was no such change observed in the anterior hippocampus and posterior cingulate, the regions shown to be affected early in AD.
use creatine as an internal standard should be interpreted with caution, as it has been reported that there is a 10% increase in total creatine per decade in the frontal lobes (Gruber, Pinker, et al. 2008) and that there is also an increase in total creatine in white matter with increasing age (Saunders, Howe, et al. 1999; Charlton, McIntyre, et al. 2007). NAA is only observed in neuronal tissue and has been implicated as a marker of neuronal integrity, density, and/ or viability and its synthesis has been coupled to mitochondrial activity (Gujar, Maheshwari, et al. 2005; Ross, Sachdev, et al. 2005). Several studies have reported regional declines in absolute NAA, NAA/choline, and NAA/creatine with increasing age (Gruber, Pinker, et al. 2008; Zimmerman, Pan, et al. 2008). This reduction had been interpreted as a possible reduction in neuronal metabolism, as NAA is only present in neuronal cells and is correlated with oxygen consumption (Minati, Grisoli, et al. 2007). There have also been studies that report no age-related changes in NAA or any NAA ratio with increasing age (Saunders, Howe, et al. 1999) but that do find a correlation between NAA level and performance on tests of executive function (Valenzuela, Sachdev, et al. 2000; Charlton, McIntyre, et al. 2007) and processing speed (Salem et al. 2008). Carbon 13 MRS (13C-MRS) has been used to study the glutamate-glutamine cycle as well as other metabolic fluxes (Rothman, Behar, et al. 2003). Using this technique, Boumezbeur, Mason, et al. (2009) reported a 30% reduction in glutamate-glutamine cycle flux in older adults, which correlated with reductions in the level of NAA and glutamate. BRAIN FUNCTION AND COGNITIVE AGING
NEUROCHEMICAL ASPECTS OF COGNITIVE AGING
Magnetic resonance spectroscopy (MRS) can be used as a means to quantify absolute and/or relative concentration of neuronal metabolites/chemicals in vivo, as well as their distribution throughout the brain (Gujar, Maheshwari, et al. 2005). Using this technique, investigations have been typically limited to a handful of molecules, including choline, creatine/phosphocreatine, N-acetyl-aspartate (NAA), myoinositol, lactate, and glutamate/glutamine (Glx) (Gujar, Maheshwari, et al. 2005; Minati, Grisoli, et al. 2007). Choline is a cell membrane constituent and may be a marker of membrane turnover, as alterations in choline concentrations have been linked to cellular proliferation as well as membrane breakdown (Gujar, Maheshwari, et al. 2005). Creatine and phosphocreatine are markers of brain energy metabolism, and their combined measure is most commonly used as an internal standard (Gujar, Maheshwari, et al. 2005). However, MRS studies investigating age-related changes that
Functional brain imaging studies using PET and fMRI have consistently shown alterations in brain activity and connectivity with age. In general, investigators have reported regional increased and/or decreased activation, recruitment of additional cortical areas, as well as different patterns of activity during many different cognitive tasks (Whalley, Deary, et al. 2004; Rajah and D’Esposito 2005; Grady 2008; Raz 2009). These activation differences that are commonly observed have led to many interesting cognitive aging models, such as the Hemispheric Activity Reduction in Older Adults (HAROLD), which accounts for the general finding of less lateralized cortical activation when compared to younger individuals (Cabeza 2001; Cabeza 2002; Li, Moore, et al. 2009). This model shares some features with the Scaffolding Theory of Cognitive Aging (STCA) put forth by Park and Reuter-Lorenz in 2009, as well as the brain reserve model of cognitive aging (Stern 2009), which state that activation differences in older adults are
3 3 4 part vii : I maging G enetics of C ognitive A ging
the manifestation of compensatory mechanisms at work to shore up existing structures or to create new ones (Park and Reuter-Lorenz 2009). These models are in contrast to the notion that alterations in neural function with aging could reflect de-differentiation of neural processing, leading to inefficient processing (Logan et al. 2002). Another model proposed includes the posterior-anterior shift in aging (PASA) model in which there is decreased response in the posterior regions and a greater response in the anterior regions in older subjects when compared to younger individuals (Davis et al. 2008). To address the relationship between the reported age-related regional alterations in cortical activity during functional imaging studies and cognitive performance, Eyler et al. (2011) did a comprehensive review of 550 articles (published from 1989 to 2009), and reported that the majority of studies found a positive correlation between cortical activity and cognitive performance. While this was more often observed in the frontal cortex and parietal and temporal cortices, the findings in the occipital cortex and medial temporal lobe were more mixed. Using advanced analytical tools, alterations in functional connectivity of brain regions within networks have also been documented with advancing age. Functional connectivity of brain regions reflects functional communication and information transfer between regions and is measured based on how synchronized or temporally correlated the activity of different brain regions within a network is. A variety of analysis techniques including graph-theoretical, non-linear, independent component analysis and correlational approaches can be used to measure the functional connectivity of brain regions. Using these approaches, in general both decreased as well as increased functional connectivity across brain regions have been observed with aging (for reviews, see (Ferreira and Busatto 2013; Antonenko and Floel 2014). Decreased connectivity of brain regions with increasing age within specific functional networks underlying different cognitive domains has been associated with worse cognitive performance. Increased connectivity of regions across networks, which could be interpreted as a reduction in specificity of brain networks with more diffuse and less specialized patterns of functional connections with aging, is consistent with the functional de-differentiation theory of aging proposed by Park et al. (2009). INDIVIDUAL VARI ABI LI TY I N CO G N I TIV E AGING A ND THE RO LE O F G EN ETI CS In the previous two sections we have presented evidence for cognitive changes with age, and the neurobiology
underlying these changes. Evidence suggests that the trajectories of age-related changes in cognition and the underlying neurobiological changes as determined by neuroimaging techniques significantly vary across individuals (Mattay, Goldberg 2007). These observable inter-individual differences in cognitive aging trajectories may be the result of both genetic and non-genetic factors, and we are only beginning to understand how genetic variation accounts for these differences. It is important to note that most inherited aspects of the human aging process do not follow a simple Mendelian inheritance pattern. Instead, age-related phenotypes, including longevity and cognitive aging, are complex genetic traits that follow a polygenetic inheritance pattern whereby many small gene effects may account for a significant portion of inherited variance (Christensen, Johnson, et al. 2006; Mattay, Goldberg, et al. 2008). In the context of aging and changes in cognitive ability, genetic variation may be protective by resisting decline or may be detrimental by amplifying decline (Goldberg and Mattay 2007). At the neuronal level, protective genetic variation may act to maintain the structural/functional integrity of existing circuitry and/or promote compensatory activity through plasticity mechanisms. Alternatively, detrimental genetic variation may act to increase the rate at which normal age-related neuronal damage occurs and/or may impair compensatory mechanisms. In addition to genetics, there are several other factors that add to the inter-individual variance in cognitive aging. These include gender, IQ, education, social and environmental factors, medical and psychiatric history, and lifestyle choices ranging from profession, physical exercise, diet, smoking, and drug and alcohol use. Investigation into these and other factors and their interactions with genetics is necessary to have a complete understanding of the variability in cognitive aging (Mattay, Goldberg, et al. 2008). What genetic analysis can potentially offer, however, is that through a better understanding of the genetic architecture of cognitive aging, specific genes or gene pathways may be identified as being responsible for a significant proportion of the variance in cognitive decline, which in turn will lead to the novel therapeutic targets as identified by these genes products to slow cognitive aging. IMAGING GE NE T IC S IN T H E C ONT E X T OF C OGNIT IV E AGING Imaging genetics is a form of a genetic association study in which the phenotype of interest is a measure such as
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
335
brain structure (volume), chemistry, or function (e.g., physiological response of the brain during information processing) derived from a neuroimaging modality (Hariri and Weinberger 2003; Mattay, Goldberg, et al. 2008). The general principle of this approach in cognitive aging is, as in other areas of neuropsychiatry and neuropsychology, the biological impact of a variation in a gene traverses an increasingly divergent path, from subtle molecular
alterations at the cellular level, to alterations in neural systems, to variability in cognition and behavior with aging (Figure 21.1). Imaging genetics allows for the estimation of these genetic effects at the level of neural systems or brain information processing, which represents a more proximate biological link to genes and serves as an obligatory intermediate to age-related changes in cognition and behavior.
GENOME
MOLECULAR ALTERATIONS
IMAGING GENETICS IN COGNITIVE AGING
CELLULAR STRUCTURE AND FUNCTION
BRAIN STRUCTURE AND FUNCTION
AGE-RELATED COGNITIVE CHANGES
Figure 21.1 Principle of imaging genetics in cognitive aging.
As in other areas of neuropsychiatry and neuropsychology, the biological impact of a variation in a gene traverses an increasingly complex path from alterations at the molecular and cellular level to alterations in neural systems, which lead to variability in cognition and behavior with aging. Imaging genetics operates on the theory that the biological impact of a variation in a gene is greatest on the phenotypes that are at a level closer to the gene product itself. Neuroimaging phenotypes related to cognitive aging, such as alterations in brain morphometry and functional neural systems, are a step closer to the effect of genes and molecular/cellular function and are less heterogeneous than clinical, behavioral, and cognitive phenotypes. They are therefore deemed to have greater detection power (effect size), making imaging genetics a useful tool in the study of genetic influences on cognitive aging. Genome: ©iStock.com/freestylephoto Molecular Alterations: ©iStock.com/Eraxion Cellular Structure and Function: (1) Republished with permission of John Wiley & Sons, Inc., from Peters A, Sethares C, Aging and the myelinated fibers in prefrontal cortex and corpus callosum of the monkey. Journal of Comparative Neurology. 442. Copyright © 2002, John Wiley & Sons. (2) Republished with permission of John Wiley & Sons, Inc., from Dickstein DL, Kabaso D, Rocher AB, Luebke JI, Wearne SL, Hof PR, Changes in the structural complexity of the aged brain. Aging Cell. 6(3). Copyright © 2007, John Wiley & Sons. (3) Republished with permission of Taylor and Francis Group LLC Books, from Kumar A, Foster TC, Neurophysiology of old neurons and synapses. In: Riddle DR, ed., Brain aging: models, methods, and mechanisms, Copyright © 2007; permission conveyed through Copyright Clearance Center, Inc. Age-Related Cognitive Changes: Republished with permission of Annual Review of Psychology, Annual Reviews, from Park DC, Reuter-Lorenz P, The adaptive brain: aging and neurocognitive scaffolding. Annual Review of Psychology. 60(1). Copyright © 2009, Annual Reviews.
3 3 6 part vii : I maging G enetics of C ognitive A ging
Compelling replicated evidence from research in recent years clearly illustrates the significant advantages of neuroimaging techniques over traditional behavioral strategies in the study of the role of genetics in cognition and behavior in health and disease. It is well documented now that brain imaging methods allow the exploration of gene effects at the level of neuronal circuitry that are closer to the biologic effects of genes. In addition, whole brain imaging methods allow the study of many individual processes, in contrast to traditional behavioral measures, which report a single summary measure that is a product of multiple interactive processes. Further, the quantitative and qualitative changes in brain structure and function measured through neuroimaging often predate or precede observable cognitive and behavioral changes. Although promising, the identification of genes associated with cognitive aging from over 20,000 genes expressed in the human brain, many of which have alternative splicing, is a monumental task (Goldberg and Weinberger 2004; Mattay, Goldberg, et al. 2008). This issue is not unique to cognitive aging research, and there has been a great effort to generate and implement various strategies to identify genes that are associated with a particular phenotype. These strategies range from the targeted candidate gene association strategies to data-driven searches of hundreds of thousands of polymorphisms throughout the genomethe so-called genome-wide association (GWA) approach (Mattay, Goldberg, et al. 2008; Marian 2012). Candidate gene approaches investigate potential genes pre-selected by investigators based on their known biological importance, function, and/or association with disease. In addition, the candidate gene approach also allows exploring the effects of new risk variants identified through GWA studies for their effects at the neural systems level on cognition and behavior, as well as their association with risk for neurodegenerative disorders and other age-related medical diseases. In the context of cognitive aging, researchers can choose from a long list of genes known to be associated with neurodegenerative/dementia diseases, longevity, cardiovascular and other systemic diseases, neuronal plasticity and stress response, as well as those known to be associated with cognitive ability (Deary, Wright, et al. 2004; Mattay, Goldberg, et al. 2008). In contrast, the unbiased data-driven approach used in GWA studies involves the testing of hundreds of thousands of polymorphisms in the genome for a significant association with a particular phenotype with a caveat that the identified genetic markers may not be causative and may be in linkage disequilibrium (LD) with the true causal variants.
IMAGING GENETICS IN COGNITIVE AGING: EXAMPLES
We will now review past and recent cognitive aging research using the imaging genetics approach. This survey will include studies that used the candidate gene, as well as those that used the GWA approach, and will be divided into sections that encompass similar biological areas, including age-related disorders associated with cognitive dysfunction with a focus on Alzheimer’s disease (AD), neurotropic signaling, neurotransmitter systems, and a novel memory-related protein, KIBRA. These studies highlight the advantages and limitations of using an imaging genetics approach to studying the genetics of cognitive aging.
Alzheimer’s Disease Genetics: Uncovering the Molecular Basis of Increased Risk for Dementia across the Life span Given that a clinical diagnosis for Alzheimer’s disease (AD) requires the presence of significant memory decline, there has been extensive investigation into the relationship between genetic variants associated with AD and age-related cognitive change. The classification of the Apolipoprotein E (APOE) gene as an AD risk gene began with linkage analysis in the early 1990s, which identified a genomic region on chromosome 19 as being associated with late onset AD (Pericak-Vance, Bebout, et al. 1991). This is in contrast to early onset AD, which occurs before the age of 60 and is considered the familial version of AD (FAD), as it follows a Mendelian pattern of inheritance (Bird 1993). Mutations in the amyloid precursor protein (APP), presenilin 1 and 2 (PSEN1, PSEN2) genes have been associated with FAD (St George-Hyslop, Tanzi, et al. 1987; Schellenberg, Bird, et al. 1992; Benzinger et al. 2013; Ridgway et al. 2013). Currently, along with chronological age, genetic variation in the APOE gene is a well-established predictor of being at increased risk for developing late onset AD (referred to as AD for the rest of the text unless otherwise stated). Three different allelic variants of APOE are known to exist in humans—E2, E3, and E4—resulting in a possibility of six distinct genotypes. Carriers of just one E4 allele have a three to four times greater risk of developing AD when compared to non-carriers (Corder, Saunders, et al. 1993; Strittmatter, Saunders, et al. 1993; Bu 2009). While the E4 allele of the APOE gene is often considered the risk allele for developing AD, in reality the E4 allele confers risk of developing AD at an earlier age than individuals without the E4 allele (Corder, Saunders. et al. 1993; Meyer 1998),
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
337
as well as a shorter survival time once diagnosed (Dal Forno 2002). Not surprisingly, the E4 allele has also been shown to decrease the probability of survival into very old age (Schachter, Faure-Delanef, et al. 1994; Christensen, Johnson, et al. 2006). In fact, GWA studies of human longevity have confirmed that the APOE locus influences longevity (Deelen, Beekman, et al. 2011; Nebel, Kleindorp, et al. 2011; Sebastiani, Solovieff, et al. 2012). At the amino acid level, the three isoforms differ in containing either cysteine or arginine at positions 112 and 158, which results in different structural and functional properties (Hatters, Peters-Libeu, et al. 2006). Also, there is evidence to suggest that the E2 allele has a protective effect when compared to the E3 or E4 allele, although the rare E2 allele has also been associated with premature cardiovascular disease (Hatters, Peters-Libeu, et al. 2006). The APOE gene encodes a protein that is involved in numerous neurological/and cardiovascular processes, suggesting that it has pleiotropic effects (Smith 2000; Hatters, Peters-Libeu, et al. 2006; Bu 2009). One notable process in the context of AD is its involvement in the removal of amyloid-B (AB), a molecule whose increased aggregation is considered the hallmark feature of AD (Hardy and Allsop 1991; Karran, Mercken, et al. 2011). As a candidate gene for cognitive aging studies, this gene has proven to be quite informative in studying one of the many biological mechanisms responsible for cognitive aging, as it has been shown to influence structural, functional, and behavioral measures across the life span, and its underlying biology is relatively well understood. At the cognitive level, multiple studies have shown that APOE E4 carriers tend to have lower levels of performance on tests of global cognition, episodic memory, and executive function (Small, Rosnick, et al. 2004; Raz, Rodrigue, et al. 2009; Wisdom, Callahan, et al. 2011). Furthermore, a recent meta-analysis of 77 studies including over 40,000 cognitively healthy adults reported that there was a larger performance gap between non-E4 carriers and E4 allele carriers in tests of episodic memory and global cognition in favor of the non-carriers (Wisdom, Callahan, et al. 2011). Similar results have also been found with longitudinal analysis. A decline in general cognitive ability from age 11 to age 79 was found only in the E4 carriers (Deary, Whiteman, et al. 2002; Harris and Deary 2011), and a similar finding was also reported for change in verbal declarative memory performance from age 79 to age 87 (Harris and Deary 2011; Schiepers, Harris, et al. 2012). Along with these behavioral measures, there have been reports of this polymorphism influencing measures of brain structure and function. A review article by Cherbuin,
Leach, et al. (2007) found consistent results across studies reporting increased rates of atrophy in the hippocampus, amygdala, and temporal lobe in E4 carriers. Additionally, the authors note that the E4 allele was also associated with decreased cerebral blood flow and glucose metabolism globally. Another study found that reductions in frontal lobe white matter and cortical volumes were related to slower processing and increase in systolic blood pressure in E4 carriers compared to E3 homozygotes. The authors suggest that even in healthy E4 carriers, clinically unremarkable increase in vascular risk may be associated with reduced frontal volumes and impaired cognitive functions (Bender and Raz 2012). Longitudinal analyses have also shown that the E4 allele is associated with greater rates of medial temporal lobe cortical thinning (Donix, Burggren, et al. 2010) and decline in regional cerebral blood flow in frontal, parietal, and temporal cortices (Thambisetty, Beason-Held, et al. 2010). Studies have also found thinner frontal cortical thickness and decreased gray matter volume in middle-aged E4 carriers, suggesting that the polymorphism could be associated with morphologic change several years before the potential onset of cognitive symptoms (Fennema-Notestine, Panizzon, et al. 2011; Alexander, Bergfield, et al. 2012) In one study, thicker frontal lobes were associated with the E2 allele, supporting its protective role (Fennema-Notestine, Panizzon, et al. 2011). There is evidence to suggest that white matter also is regionally influenced by this polymorphism, with the E4 allele conferring reduced integrity across the life span in areas such as the corpus callosum, superior longitudinal fasciculus, and parahippocampal gyrus (Honea, Vidoni, et al. 2009; Heise, Filippini, et al. 2011). Interestingly, Raz, Yang, et al. (2012) recently reported that when compared to E3 and E4 carriers, individuals carrying the rarer E2 allele had greater frontal white matter hyperintensity volumes in the frontal lobe only. Although the E2 allele has been viewed as protective against AD, the results by Raz et al. (2012) are in agreement with previous findings that implicate the E2 allele with white matter disease (Lemmens 2007). Brain activation differences associated with APOE allele differences have also been observed in tasks of working memory (Wishart, Saykin, et al. 2006), episodic memory (Mondadori, de Quervain, et al. 2007; Filippini, MacIntosh, et al. 2009; Adamson, Hutchinson, et al. 2011), and during rest (Filippini, MacIntosh, et al. 2009; Westlye, Lundervold, et al. 2011; Trachtenberg, Filippini, et al. 2012). In a study comparing fMRI activation between E4 carriers and non-carriers in a group of healthy young adults and healthy older adults during an encoding memory task,
3 3 8 part vii : I maging G enetics of C ognitive A ging
Filippini et al. (2011) report that while young E4 carriers showed increased activation compared to non-carriers of the same age, this association was reversed in the older group. Since there was no memory impairment in the older group, the authors suggest that the decreased activity in the E4 carriers may be representative of the early stages of neuronal dysfunction, which may later manifest as memory impairment. Interestingly, overactivity is a common phenomenon reported in cognitive aging literature when comparing older adults to younger adults at the same level of performance (considered compensatory activity), suggesting the possibility that the overactivity observed in this study may indicate the early onset of a cognitive aging phenotype reflected as cortical inefficiency in the E4 carriers. These results are further supported by Brown et al. (2011), who similarly showed that E4 carriers demonstrate accelerated age-related loss of local interconnectivity when assessed by DTI measures. Regional variation included loss of interconnectivity within the precuneus, orbitofrontal cortex, and lateral parietal cortex, as well as mean global loss of interconnectivity (Brown, Terashima, et al. 2011). PET imaging of AB has been used to explore the possible impact of genetic variation in APOE on AB burden. As expected, given the increased AB deposition in individuals with AD, being a carrier of the E4 allele has been reported to be associated with increased AB binding when compared to non-carriers, suggesting that the E4 allele is a risk factor for increased amyloid deposition in cognitively healthy individuals (Morris 2010; Rodrigue, Kennedy, et al. 2012). While APOE gene has received a great deal of attention because it is robustly reproducible as a risk gene for late-life dementia, other novel genes associated with risk for developing late onset AD have been identified through case-control GWA analysis, and investigators have begun to explore their impact on cognitive aging using the imaging genetics approach as well. In two separate GWA casecontrol studies, a polymorphism in the clusterin (CLU) gene showed genome-wide significance indicating that allelic differences in this gene conferred increased risk of developing AD (Harold 2009; Lambert 2009). Additional independent reports have confirmed that the C allele of this polymorphism is associated with increased risk for AD when compared to the T allele (Bertram and Tanzi 2010; Carrasquillo, Belbin, et al. 2010; Corneveaux, Myers, et al. 2010). As was the case with variation in the APOE gene, allelic differences in CLU have been reported to be associated with cognitive differences in non-demented older individuals with individuals carrying the risk-associated C allele exhibiting poorer cognitive function when compared
to those with the T allele in a dose-dependent manner (Mengel-From, Christensen, et al. 2011). In an fMRI study in healthy young adults, Erk, MeyerLindenberg, et al. (2011) report an additive effect of the risk allele on prefrontal-hippocampal activity coupling during an episodic memory task. The authors noted that altered coupling between the prefrontal cortex and hippocampal region has been observed in individuals diagnosed with AD, suggesting that connectivity dysfunction between these two regions may be an intermediate phenotype for AD. In a recent study that included longitudinal regional cerebral blood flow (rCBF) measurements using PET, Thambisetty, Beason-Held, et al. (2013) showed that CLU risk allele carriers were associated with increased longitudinal regional cerebral blood flow (rCBF) changes in brain regions intrinsic to memory processes over time (up to 8 annual visits). In cognitively normal participants who eventually converted to mild cognitive impairment or AD, CLU risk carriers showed faster rates of cognitive decline relative to non-carriers, suggesting that CLU may be important in monitoring disease progression in at-risk elderly subjects. Finally, a study by Braskie, Jahanshad, et al. (2011) found that the C allele, the risk variant, was associated with lower white matter integrity, as assessed by fractional anisotropy, in multiple brain regions known to be affected by AD in a sample of healthy young adults. While these studies need to be independently replicated, and researchers should aim to study the effects of this polymorphism across the life span, evidence to date suggests that variation in CLU may influence an individual’s level of cognitive ability at very old ages, as well as modulate brain structure and function in early adulthood. So far we have provided examples of genes with AD risk-associated variants in APOE and CLU that were discovered in very different ways. APOE was found in the pregenome era through linkage analysis, only to be later further confirmed with GWA studies and shown to impact brain structure and function using the candidate gene approach. CLU, on the other hand, was discovered through GWA studies, and now a candidate gene approach is being used to understand its impact on cognition as well as brain structure and function. Another method that is being tested and has produced some significant results is that of using neuroimaging measures as the phenotype for GWA analysis. Indeed, AD researchers have applied this method to the various biomarkers associated with having AD or being at increased risk for developing it, including CSF levels of AB 1-42, t-tau, and p-tau (Kim, Swaminathan, et al. 2011), as well as morphometric measures of the prefrontal cortex,
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
339
hippocampus, and entorhinal cortex (Shen, Kim, et al. 2010). In a GWA study that included AD patients, individuals with mild cognitive impairment (MCI), and elderly normal controls, Stein, Hua, et al. (2010) using temporal lobe volume as the phenotype of interest report a positive association between temporal lobe volume and variation in the GRIN2B gene that encodes the NR2B subunit of the NMDA receptor. This gene could be another promising candidate for cognitive aging research as the NMDA receptor is involved in many age-sensitive functions such as synaptic plasticity (Carroll and Zukin 2002; Lau and Zukin 2007), hippocampal dependent memory (Nakazawa 2002; Liu 2004), and neuronal viability (Leveille 2008), and exhibits expression differences across the life span (Magnusson, Brim, et al. 2010). Overall these studies, which include a wide range of imaging measures, provide evidence on how risk-associated gene variants can be used to explore their impact on brain structure and function across the life span, thereby leading to greater insight into the biological mechanisms that eventually lead to cognitive dysfunction later on in life.
Genetic Variation in the Neurotrophin Family: Using the BDNF Val66Met Polymorphism to Link Basic Cellular Mechanisms to Brain Structure and Function across the Life Span While there is little evidence of neuronal loss across the lifespan, there is evidence suggesting a decline in neuronal plasticity with aging (Burke and Barnes 2006), which has been associated with age-related cognitive change (Rosenzweig and Barnes 2003; Burke and Barnes 2010; Schimanski and Barnes 2010). Neurotrophins are a class of molecules that are important for neuronal survival, growth, maintenance, and plasticity (McAllister, Katz, et al. 1999; Poo 2001; Lu 2004; Lu, Pang, et al. 2005). Within this family of molecules, brain-derived neurotropic factor (BDNF) has been shown to be involved in activity-dependent synaptic plasticity and regulation of cell survival, proliferation, and synaptic growth during development, and is critical for maintaining synaptic changes in the hippocampus following long-term potentiation (LTP) (Korte, Kang, et al. 1998; Gottschalk 1999; Lu 2003; Pang and Lu 2004; Bekinschtein 2007; Lipsky and Marini 2007). The role of BDNF in molecular processes known to be important for acquiring new memories as well as maintaining memories make it a strong candidate gene for cognitive aging studies (Mizuno, Yamada, et al. 2003; Mattson, Maudsley, et al. 2004; Pang and Lu 2004). Additionally, there is evidence that the BDNF gene and its receptors, TrkB and p75, are dynamically regulated
across cortical regions as well as throughout the life span (Lein, Hohn, et al. 2000; Romanczyk, Weickert, et al. 2002; Webster, Weickert, et al. 2002; Silhol, Bonnichon, et al. 2005; Webster, Herman, et al. 2006). The exact mechanism driving these age-related changes in expression is still unresolved, although the notion that it may be in part due to epigenetic mechanisms, some of which could be the result of early life experiences, has some support (Martinowich, Hattori, et al. 2003; Farah 2011; Roth and Sweatt 2011). The role of activity-dependent secretion of BDNF and its downstream targets on brain structure and function has been studied extensively (Egan 2003; Hariri, Goldberg, et al. 2003; Hajek, Kopecek, et al. 2012; Molendijk, Bus, et al. 2012). A common functional polymorphism at codon position 66 of the BDNF gene results in an amino acid change from valine to methonine and has been associated with psychiatric and neurological disease (Sklar, Gabriel, et al. 2002; Ventriglia 2002; Pezawas 2004; Martinowich, Manji, et al. 2007). This Val-Met substitution results in dysfunctional activity-induced secretion of BDNF in Met carriers due to disrupted intracellular packaging and localization (Egan 2003; Chen 2004). Knowledge of this genetic variation in BDNF-signaling pathway processes has been used to interpret the result of behavioral and neuoimaging studies that consistently find differences between allele carriers. In healthy adult individuals, the Met allele has been associated with lower performance on multiple cognitive tests, including those of episodic (Egan, Kojima, et al. 2003) and working memory (Richter-Schmidinger et al. 2010; Brooks et al. 2014). Contrary to what may be assumed from these previous studies, it has been reported that the Met allele confers an advantage on tasks of response inhibition (Beste, Baune, et al. 2010). We should note, however, that there have been studies reporting no effect of this polymorphism on cognitive abilities (Gong, Zheng, et al. 2012; Mandelman and Grigorenko 2012). The reported genetic effect is present at the older end of the aging spectrum as well. In older adults, the Met allele is associated with lower ability in cognitive processes such as episodic memory, processing speed, working memory, and executive function (Erickson, Kim, et al. 2008; Miyajima, Ollier, et al. 2008; Raz, Rodrigue, et al. 2009), as well as reduced transfer effects from cognitive training programs (Colzato, van Muijden, et al. 2011). It has also been shown that the allele variants can impact cognitive performance in an age-dependent manner with performance differences only being manifested at older ages (Li, Chicherio, et al. 2009) or an epistatic-like manner with the gene for BDNF exaggerating allelic differences in COMT Val158Met polymorphism on task performance (Nagel,
3 4 0 part vii : I maging G enetics of C ognitive A ging
Chicherio, et al. 2008). Another study reported that the Met allele is associated with better performance in old age on a memory-based task-switching cognitive paradigm (Gajewski, Hengstler, et al. 2011). In addition to associations at the cognitive/behavioral level, neuroimaging methods have shown how genetic variation in the BDNF gene influences brain structure and function across the life span. Younger Met allele carriers exhibit decreased hippocampal activity during tasks of episodic memory (Egan 2003; Hariri, Goldberg, et al. 2003; Hashimoto, Moriguchi, et al. 2008) and spatial navigation (Banner, Bhat, et al. 2011) when compared to the valine allele homozygotes. The Val66Met polymorphism has also been shown to influence neuronal activity in other cortical areas during behavioral adaptation (Beste, Kolev, et al. 2010), selective attention (Schofield, Williams, et al. 2009), and memory (van Wingen, Rijpkema, et al. 2010) tasks. Individual differences in regional brain structure, including the prefrontal cortex and hippocampus, are also associated with genetic variation in this gene (Pezawas, Verchinski et al. 2004). In one recent study, the BDNF Val66Met SNP was associated with allelic differences in whole brain and right hippocampal volume reduction over 2 years in a large sample of elderly subjects aged 55–90 years (Honea, Cruchaga, et al. 2013). Additionally, there have been reports of lower hippocampal NAA levels in Met carriers (Egan 2003; Stern, Savostyanova, et al. 2008; Gruber, Hasan, et al. 2012). A recent study reported that even though performance differences were non-significant, Val homozygote healthy young male individuals performing a spatial navigation training paradigm had an increase in hippocampal NAA, which correlated with performance (Lövdén, Schaefer, et al. 2011). The authors suggest that this may be due to the impaired activity-induced secretion of BDNF in Met allele carriers and that, while Val homozygotes are able to efficiently utilize hippocampal dependent spatial processes, the Met carriers may have to rely more on strategy to perform well on spatial navigation tasks, a view that has been supported by other investigators as well (Banner, Bhat, et al. 2011). Although BDNF is often associated with hippocampal-dependent memory, there is evidence suggesting that BDNF signaling plays a significant role in developing prefrontal circuitry as well as maintaining its functional and structural integrity (Galloway, Woo, et al. 2008). The effect of the Val66Met polymorphism on neuroimaging measures extends into old age (and in some cases may be exaggerated). A decline in white matter integrity with increasing age for only Met carriers has been observed (Kennedy, Rodrigue, et al. 2009), as well as greater age-related declines in bilateral DLPFC gray matter volumes in Met-carriers when
compared to Val homozygotes (Nemoto, Ohnishi, et al. 2006). Using fMRI, Sambataro, Murty, et al. (2010) investigated the impact of the BDNF Val66Met polymorphism on age-related changes in neuronal function during an episodic memory task. Similar to the cortical morphology results, the age-related decline in hippocampal activity during both the encoding and retrieval sessions was significantly steeper in Met carriers when compared to performance-matched Val homozygotes. Additionally, the authors noted that this result remained significant even after controlling for hippocampal volume, suggesting that the BDNF genotype-related difference in age-related decline in hippocampal function as measured with BOLD fMRI was not solely due to alteration in neuronal morphology alteration with age, but rather with age-related alteration in the functional activity of the neuronal circuitry itself. Taken together, these studies investigating the impact of a well-characterized functional polymorphism in the BDNF gene provide evidence of how a gene’s product impacts on cognitive function through its effect on brain structure and function across the life span. DISRUPTED COMMUNICATION IN COGNITIVE AGING: GENETIC VARIATION IN NEUROTRANSMITTER SYSTEMS
Many of the cognitive abilities that decline with age have been associated with various neurotransmitter systems. Importantly, for many of these neurotransmitter systems there is evidence of age-related changes (see previous section discussing age-related changes in molecular imaging), which may be in part due to temporal patterns of gene expression (Colantuoni, Lipska, et al. 2011; Kang, Kawasawa, et al. 2011). Dopaminergic activity, with its known importance for many cognitive abilities that decline with age and well-established neural circuits known to be impacted by normal aging, has been identified as a possible culprit for cognitive decline, and there is compelling evidence to suggest that this is indeed the case (Li, Lindenberger, et al. 2001; Bäckman, Nyberg, et al. 2006; Düzel, Schütze, et al. 2008; Mizoguchi, Shoji, et al. 2009; Rollo 2009; Bäckman, Lindenberger, et al. 2010; Fischer, Nyberg, et al. 2010). Dopamine flux is regulated throughout the mammalian brain by many pathways, but in the prefrontal cortex converging evidence suggests that it may be primarily regulated by the catechol-O-methyl-transferase (COMT) enzyme (Weinberger, Egan, et al. 2001; Tunbridge, Harrison, et al. 2006; Käenmäki, Tammimäki, et al. 2010). A common single nucleotide substitution in exon 4 of the gene for COMT results in a valine to methionine substitution at codon 158, leading to functional differences in COMT activity.
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
341
Quantitatively, the polymorphism results in a three to four times greater level of enzymatic action in Val homozygotes when compared to Met homozygotes, with the heterozygotes exhibiting an intermediate level of activity (Weinberger, Egan, et al. 2001). As a result of these functional differences in enzyme activity, Met carriers have greater extra-synaptic DA levels, which have been used to explain the observed performance advantage in individuals with the Met allele on cognitive tasks that are dependent on the prefrontal cortex (Egan 2001; Malhotra, Kestler, et al. 2002; Goldberg and Weinberger 2004; Wishart, Roth, et al. 2011; Farrell, Tunbridge, et al. 2012; Bellander, Backman, et al. 2014; Das, Tan, et al. 2014). This polymorphism has proven to be especially fruitful in the context of understanding how dopaminergic activity can modulate neuronal activity related to cognitive performance through optimal dopamine levels following an inverted U function (Mattay, Goldberg, et al. 2003; Amy 2011). In younger adults, the COMT Val158Met genotype has been associated with individual differences in neuronal activity during tasks involving working memory (Egan, Goldberg, et al. 2001; Meyer-Lindenberg, Nichols, et al. 2006; Tan, Chen, et al. 2007; Bellander, Backman, et al. 2014), episodic memory (Bertolino, Rubino, et al. 2006), cognitive control (Ettinger, Kumari, et al. 2008), reward processing (Schmack, Schlagenhauf, et al. 2008), and emotion processing (Drabant, Hariri, et al. 2006). Additionally, allelic differences have been associated with differences in brain structure (Ehrlich, Morrow, et al. 2009; Honea, Verchinski, et al. 2009), midbrain dopamine synthesis (Meyer-Lindenberg, Kohn, et al. 2005), and frontal D1 receptor binding (Slifstein, Kolachana, et al. 2008), but not with D2 receptor binding in the frontal cortex or striatum (Hirvonen, Någren, et al. 2010). These behavioral and cognitive differences associated with COMT Val158Met have also been reported in older adults. The Met-allele advantage has been observed on tasks of episodic memory (de Frias, Annerbrink, et al. 2004; Raz, Rodrigue, et al. 2009), visuospatial abilities (Frias, Annerbrink, et al. 2005), processing speed (Starr, Fox, et al. 2007; Raz, Rodrigue, et al. 2009), attention (Liu, Hong, et al. 2008), working memory (Nagel, Chicherio, et al. 2008), and inhibition (Raz, Rodrigue, et al. 2009). In addition, there have been reports of the Val allele conferring an advantage at older age ranges (O’Hara, Miller, et al. 2006), as well as reports that at an older age, heterozygotes perform better than individuals homozygous for either allele (Harris, Wright, et al. 2005). To further confuse interpretation of this polymorphism’s impact on cognitive abilities throughout the life
span, there is evidence to suggest that gender interacts with this genotype and age (O’Hara, Miller, et al. 2006; Solis-Ortiz, Perez-Luque, et al. 2010). For example, Solis-Ortiz, Perez-Luque, et al. (2010) reported that in a group of middle-aged women, Val homozygotes performed better on an executive function task. This finding is interesting because women in the age range reported by the authors would be going through menopause or would be post-menopausal, and female sex hormones have been shown to regulate COMT gene expression ( Jiang, Xie, et al. 2003; Elizabeth 2010). Evidence for the impact of the COMT Val158Met on neuroimaging measures in old age is relatively sparse. But there are a few studies reporting age differences in the gene effects on neuroimaging measures. Using high-resolution structural MRI, Rowe, Hughes, et al. (2010) found that Val homozygotes have greater global gray matter volumes, but only for young adults, and that the genotype difference disappeared with increasing age. Age by genotype interactions were also observed regionally, with the Val homozygotes having greater gray matter density in the premotor area only in older age. Additionally, Met homozygotes had smaller bilateral insula gray matter volume at younger ages but did not show signs of decline with age, whereas Val-allele carriers did. From a study of a sample of individuals aged 18–35 years, Zinkstok, Schmitz, et al. (2006) reported that Val carriers have regional increases in both gray matter and white matter densities with increasing age, whereas Met homozygotes were found to have regional decreases in both gray and white matter densities with increasing age. Interestingly, the gene effect was only observed in females, and the authors speculate that COMT Val158Met may alter cortical maturation rates in females through female sex hormone regulation of COMT gene expression. Importantly for our discussion, this difference in maturation rates may account for differences in cognitive ability later on in life, as carriers of one allele demonstrate the age-related cortical change earlier than the other, placing them at increased risk for cognitive impairment earlier in life even though their actual rates of decline were not necessarily different. In another morphometric study with MRI, Gennatas, Cholfin, et al. (2012) recently reported that the COMT Val158Met polymorphism influenced neurodegeneration within dopamine-innervated brain regions. The authors show a dose-dependent correlation with increasing Val allele dosage correlating with decreased gray matter in the regions of the ventral tegmental area (VTA), ventromedial prefrontal cortex, bilateral dorsal midinsula, left dorsolateral prefrontal cortex, and right ventral striatum, and that these volume reductions correlated with cognitive and behavioral deficits.
3 4 2 part vii : I maging G enetics of C ognitive A ging
There has also been evidence of this polymorphism influencing neuronal activity during a low-level working memory task from early adulthood to older age. Using BOLD fMRI during working memory, Sambataro, Reed, et al. (2009) report a main effect of COMT Val158Met, as well as evidence for it interacting with age on both task-related activity and network connectivity. Older Val homozygotes were found to have increased DLPFC activity when compared to Met homozygotes of a similar age. The authors also found evidence of the polymorphism regionally influencing connectivity strength between prefrontal and parietal regions, and that this COMT genotype–based difference in network activity changes as a function of age. Specifically, within the prefrontal-parietal network, Val homozygotes had stronger DLPFC connectivity with other regions, which increased with age, when compared to Met homozygote individuals. These findings were interpreted by the authors to reflect an exaggerated effect of COMT Val158Met polymorphism with increasing age. Outside the prefrontal cortex, other molecules in the dopaminergic system play a much greater role in regulating dopaminergic flux. The dopamine transporter (DAT), encoded by the SLC6A3 gene, is the primary determinant of mesostriatal dopaminergic activity through synaptic re-uptake, thereby regulating the level and duration of dopamine signaling (Kawarai, Kawakami, et al. 1997; Bannon, Michelhaugh, et al. 2001). Numerous aspects of the midbrain DAT system are sensitive to the effects of aging. The number of DAT producing neurons (Ma, Ciliax, et al. 1999), mRNA expression (Bannon and Whitty 1997), and membrane transporter expression (Volkow 1994; Salvatore, Apparsundaram, et al. 2003) declines with increasing age and has been linked to individual differences in various cognitive abilities and neuronal activity (Mozley, Gur, et al. 2001; Erixon-Lindroth, Farde, et al. 2005). The SLC6A3 gene contains a 40-base pair variable number of tandem repeat (VNTR) polymorphism, with the 9- and 10-repeat variants being the most common (Vandenbergh, Persico, et al. 1992). There have been reports showing that the 10-repeat allele conferred higher levels of transport expression when compared to the 9-repeat allele (Mill, Asherson, et al. 2002; Brookes, Neale, et al. 2007), which would theoretically confer increased dopamine transporter membrane expression and thus a shorter duration of dopmaminergic signaling in the 10-repeat carriers compared to those with the 9-repeat, but other studies found the opposite effect on DAT transcript levels as well as regional variation in the VNTR effect (Shumay, Chen, et al. 2011). Investigations into the VNTR effect on transporter membrane expression have yielded similar inconsistencies in results (van Dyck,
Malison, et al. 2005; VanNess, Owens, et al. 2005; Willeit and Praschak-Rieder 2010; Costa, Riedel, et al. 2011). While further analysis must be carried out to understand the functional impact of this gene variant at the molecular level, there is substantial evidence that genetic differences in SLC6A3 modulate the neuronal circuitry of numerous cognitive abilities, such as episodic memory (Schott, Seidenbecher, et al. 2006; Bertolino, Di Giorgio, et al. 2008), working memory (Bertolino, Blasi, et al. 2006), cognitive flexibility (Garcia-Garcia, Barcelo, et al. 2010), and novelty detection (Garcia-Garcia, Barceló, et al. 2010). Less is known about the impact of this genetic variant on cognitive functioning in older adults, with one study reporting a modest gene effect on older individuals’ startle response, with the 9-repeat allele carriers showing a smaller startle response than the 10-allele homozygotes (Armbruster, Mueller, et al. 2011). Using fMRI, Brehmer, Westerberg, et al. (2009) investigated the impact of this VNTR on working memory plasticity in a sample of young adults split into groups of 9/10 allele carriers and 10/10 allele carriers. There was no significant difference in performance between the two groups at the onset of the task, but there was a significant effect of allele status on performance at week 4 of the training, with the 9-allele carriers showing greater training-related gains. The authors concluded that DAT may be important for modulating training-related performance gains, which they labeled as working memory plasticity, and that genetic variation in DAT may account for differences in this but not necessarily baseline white matter performance. Although this study did not include older individuals, its findings are important for cognitive aging research, as they suggest that training-related neuronal plasticity differences between allelic carriers that are apparent at younger ages may place those individuals with reduced functional plasticity at risk for cognitive decline earlier on in life. The studies highlighted in this section provide evidence for the impact of dopaminergic function on brain function and structure and how using knowledge of the molecular impact of genetic variation in genes within the dopaminergic pathway, such as COMT and DAT, can be used to highlight the impact of dopaminergic alterations on cognitive function as one ages. THE SEARCH FOR NEW CLUES: GENOME-WIDE ASSOCIATION STUDIES FOR AGE-SENSITIVE COGNITIVE ABILITIES
Uncovering a complete picture of the molecular architecture underlying cognitive aging will undoubtedly require the discovery of novel molecules as well as their pathways—the
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
343
search for which is a difficult task. The data-driven GWA approach is ideal for this (Papassotiropoulos and de Quervain 2011), but unfortunately the highly complex nature of cognitive changes that occur with increasing age makes it difficult to identify a suitable phenotype to use for genetic association. Until a reliable and valid cognitive phenotype emerges, positive results from GWA studies in healthy young adults on cognitive abilities that are age-sensitive can be used as candidate genes to investigate their impact on cognitive aging. To date there have been studies reporting significant results for short-term memory (Papassotiropoulos, Henke, et al. 2011) and episodic memory (Papassotiropoulos, Stephan, et al. 2006; Need 2009; Papassotiropoulos, Stefanova, et al. 2011). One of the episodic memory–related genes has attracted a great deal of attention since its discovery and its impact on the neuronal substrates of episodic memory has been investigated throughout adult life. The brain and kidney protein KIBRA (Kremerskothen, Plaas, et al. 2003) was labeled a memory-related molecule following a GWA study that reported polymorphisms in KIBRA as well as CLSTN2 to be significant predictors for differences in episodic memory (Papassotiropoulos, Stephan, et al. 2006). The significant WWC1 polymorphism, rs17070145, results in a common T → C substitution within the ninth intron. The authors reported that individuals carrying the rarer T allele performed significantly better than C homozygotes on both 5-minute and 24-hour free recall, and this association was not found for performance on the immediate recall portion of the episodic memory task or on tasks of executive functions, working memory, or attention (Papassotiropoulos, Stephan, et al. 2006). The authors also found increased neuronal activity in memory-related brain areas, including the hippocampus, frontal cortex, and parietal cortex, during the recognition phase of an episodic memory task in C-allele homozygotes when compared to the favorable T-allele carriers. Additionally, these activation differences were not due to performance differences and were not observed during the encoding phase or during a working memory task. The authors concluded that the effect of this polymorphism is most likely specific to the neuronal process involved in retrieval of information, rather than in the actual memory formation. Following this initial study, there have been multiple replications confirming the effect of this polymorphism on episodic memory performance in both young and older adults (Schaper, Kolsch, et al. 2008; Bates, Price, et al. 2009; Preuschhof, Heekeren, et al. 2010; Vassos, Bramon, et al. 2010; Kauppi, Nilsson, et al. 2011), but there have also been studies reporting no effect (Need, Attix, et al. 2008; Wersching, Guske, et al. 2011; Sédille-Mostafaie, Sebesta,
et al. 2012), and the opposite effect on memory performance (Nacmias, Bessi, et al. 2008) and on hippocampal activity during the recognition phase of an episodic memory task (Kauppi, Nilsson, et al. 2011). A recent meta-analysis (Milnik, Heck, et al. 2012) confirmed the significant association of rs17070145 with both episodic (r = 0.068, p = 0.001) and working memory performance (r = 0.035, p = 0.018). The authors indicated that the polymorphism explained 0.5% of the variance in episodic memory tasks and 0.1% of the variance in working memory tasks. The effect of the polymorphism on hippocampal volume was recently shown by Palombo, Amaral, et al. (2013), who reported larger hippocampal CA1 volume in T-allele carriers relative to non-carriers linking a potential neural mechanism for the effects of KIBRA on episodic memory performance. Additionally, there have been mixed reports of this polymorphism’s impact on risk for late-life dementia, with two studies reporting non-T-allele carriers having increased risk (Corneveaux, Liang, et al. 2008; Burgess, Pedraza, et al. 2011), one reporting no effect on risk (Sédille-Mostafaie, Sebesta, et al. 2012), and one reporting increased risk for T-allele carriers (Rodríguez-Rodríguez, Infante, et al. 2009). Recent work by Muse, Emery, et al. (2013) examined the effect of this WWC1 polymorphism on hippocampal activity, as assessed by fMRI during an episodic memory task in adults across the life span (18–89 years of age). Results from this study showed that increasing age was associated with an exaggerated effect of the polymorphism on measures of recollection memory (as assessed by neuropsychological testing). Additionally, the authors report a clear negative association between hippocampal engagement during both the encoding and retrieval phase of the task and increasing age in CC individuals, but not in the advantageous T-allele carriers. Clearly continued cognitive and neuroimaging research is needed to better understand this polymorphism’s impact on cognition, as well as the molecular genetics and biochemical characterization of this gene and its downstream products, which has already been initiated following the initial GWA finding ( Johannsen, Duning, et al. 2008). Using in situ hybridization assays and immunohistological staining of human and rodent (rat) brain tissue, Johannsen, Duning, et al. (2008) reported that KIBRA exhibited spatial-temporal expression patterns, with expression highest in the hippocampus and cortex, and lower expression levels in the cerebellum and hypothalamus. A link further solidifying KIBRA’s role in memory comes from a recent study that reported in vivo evidence that KIBRA interacts with protein interacting with kinase 1 (PICK1) and influences AMPA receptor (AMPAR) trafficking following neuronal activity (Makuch, Volk, et al.
3 4 4 part vii : I maging G enetics of C ognitive A ging
2011). Kibra knockout mice were also generated, and the authors report that adult knockout out mice had impaired LTP and LTD, whereas the young knockout mice did not. Adult Kibra knockout mice also had deficits in the learning and retention of previously trained materials. These recent molecular studies provide a biological basis from which the genetic association studies of the KIBRA polymorphism can be interpreted, something that is critical if we are to use GWA-identified genes to understand their mechanistic importance in cognition. We have focused in this chapter on just a handful of genes and their associated imaging phenotypes. There are, however, many other genes that have been related to cognitive decline and the underlying changes at the neural substrate level assessed with neuroimaging across the adult life span. While not completely inclusive, we will now briefly introduce several other genes that have been recently implicated in cognitive aging and their pertinent neuroimaging associations. First, iron metabolism has been long considered a potential mechanism of increased oxidative stress and cell death in several neurodegenerative conditions that also impact cognition such as AD (Gregory and Hayflick 1993). Genes that regulate iron expression include hemochromatosis H63D (HFE H63D) and transferrin C2 (TfC2), which have been associated with higher brain iron levels in older individuals. Recent studies have investigated the interaction between memory performance, neuroimaging measures, and gene carrier status and have suggested significant associations between cognitive decline and presence of the gene (Bartzokis, Lu, et al. 2011; Gebril, Kirby, et al. 2011). Second, in addition to genes well-known to be associated with AD (APOE, PICALM, APP, CLU), the CR1 gene has recently been similarly associated with AD (Hazrati, Van Cauwenberghe, et al. 2012; Jin, Li, et al. 2012; Thambisetty, An, et al. 2013). One study (Bralten, Franke, et al. 2011), found smaller local gray matter volume in the entorhinal cortex of the risk allele carriers, further raising the possibility for increased risk of AD. Finally, there have been a number of other genes and their polymorphisms that have been related to cognitive decline in aging. Because the literature is much more limited for these, we mention the following here for completeness and for further reference: SCN1A (Meier, Demirakca, et al. 2012) and GOLM1 (Inkster, Rao, et al. 2012).
C ONCL US IO N S Improving older individuals’ quality of life through the alleviation of cognitive decline holds promise, but achieving
this goal will require gaining a better understanding of the mechanisms responsible for cognitive aging. Genetic as well as non-genetic factors and their interaction throughout the life span are responsible for the level of an older individual’s cognitive ability and the rate at which that ability changes with increasing age. We began with a discussion of the way in which brain structure and function change with age and how these changes account for individual differences in cognitive level and change. Using genetic association studies has been suggested as a means to further our understanding the molecular pathways that are involved in cognition and how cognitive ability changes with increasing age. To date there have been many exciting findings, but there has also been a lack of consistency in these findings, especially at the behavioral level, which may be a reflection of how complex cognitive change is with increasing age, as well as the heterogeneous nature of its change across individuals. Neuroimaging measures, which can identify the neuronal correlates of the many sub-processes necessary for performing complex cognitive tasks in genetic association studies, show promise to provide a more powerful approach to uncover the molecular determinants of cognitive aging. We have provided examples of how the imaging genetics approach has been used to study cognitive aging through the use of both candidate gene and whole genome studies. While progress in accomplishing the goal of discovering new therapeutic targets and intervention strategies has been slow and will take time to truly reach fruition, the examples we outline in this chapter illustrate how imaging genetics can help in identifying some of the genetic mechanisms underlying the variability in cognitive aging. Replications in independent samples are, however, necessary to further validate the reported gene-imaging phenotype associations. F U T U R E DIR E C T IONS The application of molecular genetics and neuroimaging technologies to the study cognitive aging has already in its short history greatly enhanced our knowledge of the many different variables that may be at work to cause individual differences in the degree to which cognitive abilities change over time. However, the work so far has only begun to scratch the surface, and we are still far from a level of understanding what will be required if we are to implement effective strategies that will improve older individuals’ quality of life. In the spirit of moving this research forward toward its goal of improving the quality of life in older years, we will end this chapter with some current topics within the field that should be addressed to improve the quality of research,
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
345
as well as future directions that may provide new avenues for discovery. Study design, be it longitudinal or cross-sectional, is a major source of disagreement among cognitive aging researchers because studies often yield differing results as to when cognitive decline begins. This argument has been commented on in detail with respect to how age influences cognitive abilities and so we direct the reader to the following commentaries (K. Warner 2009; Lise 2009; Nilsson, Sternäng, et al. 2009; Timothy 2009). The continuous technological evolvution of neuroimaging methods places limitations on the potential use of longitudinal analysis. That being said, only longitudinal analysis can form causative associations between changes in brain structure/function and cognition associated with aging. Cross-sectional studies are typically conducted in two different ways: those that compare data acquired in extreme age groups (younger vs. older cohorts), and those that study a phenotype across the life span. The advantage of the latter approach is that since aging is a continuous process, it allows defining trajectories of age-related changes in brain structure and function across the life span. That being said, results from such cross-sectional life-span studies are often interpreted as rates of change. It has been argued that this interpretation is inappropriate because cross-sectional studies produce a single point in time value, which should not be used to derive an actual rate of change measurement. Interestingly, a few heritability studies have reported that point-in-time measurements and rate-of-change measurements have their own unique variance, suggesting that both study designs can be used to gain a better understanding of the genetic influences on cognitive aging. The majority of the studies reviewed are cross-sectional studies with one or two neuroimaging approaches (structural, functional, and/or metabolic). Systematic multi-modal longitudinal studies will facilitate a better understanding of the mechanisms underlying the effects of genes and environment on cognitive aging. R EF ERENCES Adam G. (2011). Influence of early attentional modulation on working memory. Neuropsychologia 49(6): 1410–1424. Adamson MM, Hutchinson JB, Shelton AL, Wagner AD, Taylor JD. (2011). Reduced hippocampal activity during encoding in cognitively normal adults carrying the APOE ɛ4 allele. Neuropsychologia. 49(9): 2448–2455. Albert MS. (2011). Changes in cognition. Neurobiol Aging. 32(Suppl 1): S58–63. Alexander GE, Bergfield KL, Chen K, Reiman EM, Hanson KD, Lin L, Bandy D, Caselli RJ, Moeller JR. (2012). Gray matter network associated with risk for Alzheimer’s disease in young to middle-aged adults. Neurobiol Aging. 33(12): 2723–2732.
Amy FTA. (2011). Catecholamine influences on dorsolateral prefrontal cortical networks.” Biol Psychiatry. 69(12): e89–e99. Anstey K, Christensen H. (2000). Education, activity, health, blood pressure and apolipoprotein E as predictors of cognitive change in old age: a review. Gerontology. 46(3): 163–177. Antonenko D, Floel A (2014). Healthy aging by staying selectively connected: a mini-review. Gerontology. 60(1): 3–9. Antonini A, Leenders KL, Reist H, Thomann R, Beer HF, Locher J. (1993). Effect of age on D2 dopamine receptors in normal human brain measured by positron emission tomography and 11C-raclopride. Arch Neurol. 50(5): 474–480. Ardila A. (2007). Normal aging increases cognitive heterogeneity: analysis of dispersion in WAIS-III scores across age. Arch Clin Neuropsychol. 22(8): 1003–1011. Armbruster D, Mueller A, Strobel A, Lesch K-P, Kirschbaum C, Brocke B. (2011). Variation in genes involved in dopamine clearance influence the startle response in older adults. J Neural Trans. 118(9): 1281–1292. Arnsten AFT, Li B-M. (2005). Neurobiology of executive functions: catecholamine influences on prefrontal cortical functions. Biol Psychiatry. 57(11): 1377–1384. Arranz BN, Eriksson A, Mellerup E, Plenge P, Marcusson J. (1993). Effect of aging in human cortical pre- and postsynaptic serotonin binding sites. Brain Res. 620(1): 163–166. Bäckman L, Karlsson S, Fischer H, Karlsson P, Brehmer Y, Rieckmann A, MacDonald SWS, Farde L, Nyberg L. (2011). Dopamine D1 receptors and age differences in brain activation during working memory. Neurobiol Aging. 32(10): 1849–1856. Bäckman L, Lindenberger U, Li S-C, Nyberg L. (2010). Linking cognitive aging to alterations in dopamine neurotransmitter functioning: recent data and future avenues. Neurosci Biobehav Rev. 34(5): 670–677. Bäckman L, Nyberg L, Lindenberger U, Li S-C, Farde L. (2006). The correlative triad among aging, dopamine, and cognition: current status and future prospects. Neurosci Biobehav Rev. 30(6): 791–807. Banner H, Bhat V, Etchamendy N, Joober R, Bohbot VD. (2011). The brain-derived neurotrophic factor Val66Met polymorphism is associated with reduced functional magnetic resonance imaging activity in the hippocampus and increased use of caudate nucleus-dependent strategies in a human virtual navigation task. Eur J Neurosci. 33(5): 968–977. Bannon MJ, Michelhaugh SK, Wang J, Sacchetti P. (2001). The human dopamine transporter gene: gene organization, transcriptional regulation, and potential involvement in neuropsychiatric disorders. Eur Neuropsychopharmacol. 11: 449–455. Bannon MJ, Whitty CJ. (1997). Age-related and regional differences in dopamine transporter mRNA expression in human midbrain. Neurology 48(4): 969–977. Barral S, Cosentino S, Costa R, Matteini A, Christensen K, Andersen SL, Glynn NW, Newman AB, Mayeux R. (2012). Cognitive function in families with exceptional survival. Neurobiol Aging. 33(3): 619.e1–7. Bartzokis G, Lu PH, Tingus K, Peters DG, Amar CP, Tishler TA, Finn JP, Villablanca P, Altshuler LL, Mintz J, Neely E, Connor JR. (2011). Gender and iron genes may modify associations between brain iron and memory in healthy aging. Neuropsychopharmacology. 36(7): 1375–1384. Basser PJ, Pierpaoli C. (1996). Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn Reson B. 111(3): 209–219. Bates TC, Price JF, Harris SE, Marioni RE, Fowkes FG, Stewart MC. (2009). Association of KIBRA and memory. Neurosci Lett. 458: 140–143. Bekinschtein P. (2007). Persistence of long-term memory storage requires a late protein synthesis- and BDNF-dependent phase in the hippocampus. Neuron 53: 261–277. Bellander M, Backman L, Liu T, Schjeide BM, Bertram L, Schmiedek F, Lindenberger U, Lovden M. (2014). Lower baseline performance but
3 4 6 part vii : I maging G enetics of C ognitive A ging
greater plasticity of working memory for carriers of the val allele of the comt val158met polymorphism. Neuropsychology. 29(2): 247–254. Bender AR, Raz N. (2012). Age-related differences in memory and executive functions in healthy APOE varepsilon4 carriers: the contribution of individual differences in prefrontal volumes and systolic blood pressure. Neuropsychologia. 50(5): 704–714. Benzinger TL, et al. (2013). Regional variability of imaging biomarkers in autosomal dominant Alzheimer’s disease. Proc Natl Acad Sci U S A. 110(47): E4502–4509. Bertolino A, Blasi G, Latorre V, Rubino V, Rampino A, Sinibaldi L, Caforio G, Petruzzella V, Pizzuti A, Scarabino T, Nardini M, Weinberger DR, Dallapiccola B. (2006). Additive effects of genetic variation in dopamine regulating genes on working memory cortical activity in human brain. J Neurosci. 26(15): 3918–3922. Bertolino A, Di Giorgio A, Blasi G, Sambataro F, Caforio G, Sinibaldi L, Latorre V, Rampino A, Taurisano P, Fazio L, Romano R, Douzgou S, Popolizio T, Kolachana B, Nardini M, Weinberger DR, Dallapiccola B. (2008). Epistasis between dopamine regulating genes identifies a nonlinear response of the human hippocampus during memory tasks. Biol Psychiatry. 64(3): 226–234. Bertolino A, Rubino V, Sambataro F, Blasi G, Latorre V, Fazio L. (2006). Prefrontal-hippocampal coupling during memory processing is modulated by COMT val158met genotype. Biol Psychiatry. 60: 1250–1258. Bertoni-Freddari C, Fattoretti P, Giorgetti B, Grossi Y, Balietti M, Casoli T, Di Stefano G, Perretta G. (2007). Synaptic and mitochondrial morphometry provides structural correlates of successful brain aging. Ann N Y Acad Sci. 1097(1): 51–53. Bertram L, Tanzi RE. (2010). Alzheimer disease: new light on an old CLU. Nat Rev Neurol. 6(1): 11–13. Beste C, Baune BT, Domschke K, Falkenstein M, Konrad C. (2010). Paradoxical association of the brain-derived-neurotrophic-factor val66met genotype with response inhibition. Neuroscience. 166(1): 178–184. Beste C, Kolev V, Yordanova J, Domschke K, Falkenstein M, Baune BT, Konrad C. (2010). The role of the BDNF Val66Met polymorphism for the synchronization of error-specific neural networks. J Neurosci. 30(32): 10727–10733. Bird, TD. (1993). Early-Onset Familial Alzheimer Disease. In: R. A. Pagon, M. P. Adam, H. H. Ardinger et al. (eds), GeneReviews(R). Seattle (WA). Bopp KL, Verhaeghen P. (2005). Aging and verbal memory span: a meta-analysis. J Gerontol B Psychol Sci Soc Sci. 60(5): P223–233. Boumezbeur F, Mason GF, de Graaf RA, Behar KL, Cline GW, Shulman GI, Rothman DL, Petersen KF. (2009). Altered brain mitochondrial metabolism in healthy aging as assessed by in vivo magnetic resonance spectroscopy. J Cereb Blood Flow Metab. 30(1): 211–221. Bralten J, Franke B, Arias-Vasquez A, Heister A, Brunner HG, Fernandez G, Rijpkema M. (2011). CR1 genotype is associated with entorhinal cortex volume in young healthy adults. Neurobiol Aging. 32(11): 2106 e2107–2111. Braskie MN, Jahanshad N, Stein JL, Barysheva M, McMahon KL, Zubicaray GI, Martin NG, Wright MJ, Ringman JM, Toga AW, Thompson PM. (2011). Common Alzheimer’s disease risk variant within the CLU gene affects white matter microstructure in young adults. J Neurosci. 31(18): 6764–6770. Brehmer Y, Westerberg H, Bellander M, Fürth D, Karlsson S, Bäckman L. (2009). Working memory plasticity modulated by dopamine transporter genotype. Neurosci Lett. 467(2): 117–120. Brookes KJ, Neale BM, Sugden K, Khan N, Asherson P, D’Souza UM (2007). Relationship between VNTR polymorphisms of the human dopamine transporter gene and expression in post-mortem midbrain tissue. Am J Med Genet B Neuropsychiat Genet. 144B(8): 1070–1078. Brooks-Wilson AR. (2013). Genetics of healthy aging and longevity. Hum Genet. 132(12): 1323–1338. Brooks SJ, Nilsson EK, Jacobsson JA, Stein DJ, Fredriksson R, Lind L, Schioth HB. (2014). BDNF polymorphisms are linked to poorer
working memory performance, reduced cerebellar and hippocampal volumes and differences in prefrontal cortex in a Swedish elderly population. PLoS One. 9(1): e82707. Brown JA, Terashima KH, Burggren AC, Ercoli LM, Miller KJ, Small GW, Bookheimer SY. (2011). Brain network local interconnectivity loss in aging APOE-4 allele carriers. Proc Natl Acad Sci U S A. 108(51): 20760–20765. Brown MT. (2010). Early-life characteristics, psychiatric history, and cognition trajectories in later life. Gerontologist. 50(5): 646–656. Bu G. (2009). Apolipoprotein E and its receptors in Alzheimer’s disease: pathways, pathogenesis and therapy. Nat Rev Neurosci. 10(5): 333–344. Buchert R, Schulze O, Wilke F, Berding G, Thomasius R, Petersen K, Brenner W, Clausen M. (2006). Is correction for age necessary in SPECT or PET of the central serotonin transporter in young, healthy adults? J Nucl Med. 47(1): 38–42. Burchinsky SG. (1984). Neurotransmitter receptors in the central nervous system and aging: pharmacological aspect (review). Exper Gerontol. 19(4): 227–239. Burgess JD, Pedraza O, Graff-Radford NR, Hirpa M, Zou F, Miles R, Nguyen T, Li M, JLucas JA, Ivnik RJ, Crook J, Pankratz VS, Dickson DW, Petersen RC, Younkin SG, Ertekin-Taner N. (2011). Association of common KIBRA variants with episodic memory and AD risk. Neurobiol Aging. 32(3): 557.e551, 7–9. Burgmans S, van Boxtel MPJ, van den Berg KEM, Gronenschild EHBM, Jacobs HIL, Jolles J, Uylings HBM. (2011). The posterior parahippocampal gyrus is preferentially affected in age-related memory decline. Neurobiol Aging. 32(9): 1572–1578. Burke SN, Barnes CA. (2006). Neural plasticity in the ageing brain. Nat Rev Neurosci. 7(1): 30–40. Burke SN, Barnes CA. (2010). Senescent synapses and hippocampal circuit dynamics. Trends Neurosci. 33(3): 153–161. Burzynska AZ, Nagel IE, Preuschhof C, Gluth S, Bäckman L, Li S-C, Lindenberger U, Heekeren HR. (2011). Cortical thickness is linked to executive functioning in adulthood and aging. Human Brain Mapp. 33(7): 1607–1620. Burzynska AZ, Preuschhof C, Bäckman L, Nyberg L, Li SC, Lindenberger U, Heekeren HR. (2010). Age-related differences in white matter microstructure: region-specific patterns of diffusivity. NeuroImage. 49(3): 2104–2112. Cabeza R. (2001). Cognitive neuroscience of aging: contributions of functional neuroimaging. Scand J Psychol. 42(3): 277–286. Cabeza R. (2002). Hemispheric asymmetry reduction in older adults: the HAROLD model. Psychol Aging. 17(1): 85–100. Canas PM, Duarte JM, Rodrigues RJ, Kofalvi A, Cunha RA. (2009). Modification upon aging of the density of presynaptic modulation systems in the hippocampus. Neurobiol Aging. 30(11): 1877–1884. Carrasquillo, M. M., O. Belbin, T. A. Hunter, L. Ma, G. D. Bisceglio, F. Zou, Crook JE, Pankratz VS, Dickson DW, Graff-Radford NR, Petersen RC, Morgan K, Younkin SG. (2010). Replication of CLU, CR1, and PICALM associations with alzheimer disease. Arch Neurol. 67(8): 961–964. Carroll RC, Zukin RS. (2002). NMDA-receptor trafficking and targeting: implications for synaptic transmission and plasticity. Trends Neurosci. 25: 571–577. Charlton RA, McIntyre DJ, Howe FA, Morris RG, Markus HS. (2007). The relationship between white matter brain metabolites and cognition in normal aging: the GENIE study. Brain Res. 1164: 108–116. Chen KHM, Chuah LYM, Sim SKY, Chee MWL. (2010). Hippocampal region-specific contributions to memory performance in normal elderly. Brain Cognition. 72(3): 400–407. Chen ZY. (2004). Variant brain-derived neurotrophic factor (BDNF) (Met66) alters the intracellular trafficking and activity-dependent secretion of wild-type BDNF in neurosecretory cells and cortical neurons. J Neurosci. 24: 4401–4411. Cherbuin N, Leach LS, Christensen H, Anstey KJ. (2007). Neuroimaging and APOE genotype: a systematic qualitative review. Dement Geriatr Cogn Disord. 24(5): 348–362.
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
347
Choi SR, Golding G, Zhuang Z, Zhang W, Lim N, Hefti F, Benedum TE, Kilbourn MR, Skovronsky D, Kung HF. (2009). Preclinical properties of 18F-AV-45: a PET agent for Abeta plaques in the brain. J Nucl Med. 50(11): 1887–1894. Christensen K, Johnson TE, Vaupel JW. (2006). The quest for genetic determinants of human longevity: challenges and insights. Nat Rev Genet. 7(6): 436–448. Colantuoni C, Lipska BK, Ye T, Hyde TM, Tao R, Leek JT, Colantuoni EA, Elkahloun AG, Herman MM, Weinberger DR, Kleinman JE. (2011). Temporal dynamics and genetic control of transcription in the human prefrontal cortex. Nature. 478(7370): 519–523. Colzato LS, van Muijden J, Band GPH, Hommel B. (2011). Genetic modulation of training and transfer in older adults: BDNF Val66Met polymorphism is associated with wider useful field of view. Front Psychol. 2: 199. Corder E, Saunders A, Strittmatter W, Schmechel D, Gaskell P, Small G, Roses A, Haines J, Pericak-Vance M. (1993). Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261(5123): 921–923. Corneveaux JJ, Liang WS, Reiman EM, Webster JA, Myers AJ, Zismann VL. (2008). Evidence for an association between KIBRA and late-onset Alzheimer’s disease.” Neurobiol Aging. 31: 901–909. Corneveaux JJ, Myers AJ, Allen AN, Pruzin JJ, Ramirez M, Engel A. (2010). Association of CR, CLU and PICALM with Alzheimer’s disease in a cohort of clinically characterized and neuropathologically verified individuals. Hum Mol Genet. 19: 3295–3301. Costa A, Riedel M, Müller U, Möller H-J, Ettinger U. (2011). Relationship between SLC6A3 genotype and striatal dopamine transporter availability: a meta-analysis of human single photon emission computed tomography studies. Synapse 65(10): 998–1005. Court JA, Lloyd S, Johnson M, Griffiths M, Birdsall NJM, Piggott MA, Oakley AE,Ince PG, Perry EK, Perry RH. (1997). Nicotinic and muscarinic cholinergic receptor binding in the human hippocampal formation during development and aging. Develop Brain Res. 101(1–2): 93–105. Da Silva Costa-Aze V, Dauphin F, Boulouard M. (2011). Serotonin 5-HT6 receptor blockade reverses the age-related deficits of recognition memory and working memory in mice. Behav Brain Res. 222(1): 134–140. Dal Forno G. (2002). APOE genotype and survival in men and women with Alzheimer’s disease. Neurology. 58: 1045–1050. Dale AM, Fischl B, Sereno MI. (1999). Cortical surface-based analysis. I. Segmentation and surface reconstruction. NeuroImage. 9(2): 179–194. Das D, Tan X, Bielak AA, Cherbuin N, Easteal S, Anstey KJ. (2014). Cognitive ability, intraindividual variability, and common genetic variants of catechol-O-methyltransferase and brain-derived neurotrophic factor: a longitudinal study in a population-based sample of older adults. Psychol Aging. 29(2): 393–403. Davis SW, et al. (2008). Que PASA? The posterior-anterior shift in aging. Cereb Cortex. 18(5): 1201–1209. Daws LC, Gould GG. (2011). Ontogeny and regulation of the serotonin transporter: Providing insights into human disorders. Pharmacol Therapeut. 131(1): 61–79. de Figueiredo EHMSG, Borgonovi AFNG, Doring TM. (2011). Basic concepts of MR imaging, diffusion MR imaging, and diffusion tensor imaging. Mang Res Imag Clin N Am. 19(1): 1–22. de Frias CM, Annerbrink K, Westberg L, Eriksson E, Adolfsson R, Nilsson L-G. (2004). COMT gene polymorphism is associated with declarative memory in adulthood and old age.” Behav Genet. 34(5): 533–539. Deary I, Johnson W, Houlihan L. (2009). Genetic foundations of human intelligence. Human Genetics 126(1): 215–232. Deary IJ, Corley J, Gow AJ, Harris SE, Houlihan LM, Marioni RE, Penke L, Rafnsson SB, Starr JM. (2009). Age-associated cognitive decline. Br Med Bull. 92(1): 135–152. Deary IJ, Whiteman MC, Pattie A, Starr JM, Hayward C,Wright AF. (2002). Cognitive change and the APOE epsilon 4 allele. Nature. 418: 932.
Deary IJ, Wright AF, Harris SE, Whalley LJ, Starr JM. (2004). Searching for genetic influences on normal cognitive ageing. Trends Cogn Sci. 8(4): 178–184. Deelen J, Beekman M, Uh H-W, Helmer Q, Kuningas M, Christiansen L, Kremer D, van der Breggen R, Suchiman HED, Lakenberg N, … Slagboom PE. (2011). Genome-wide association study identifies a single major locus contributing to survival into old age; the APOE locus revisited. Aging Cell. 10(4): 686–698. Dickstein DL, Weaver CM, Luebke JI, Hof PR. (2013). Dendritic spine changes associated with normal aging. Neuroscience. 251: 21–32. Disterhoft JF, Oh MM. (2007). Alterations in intrinsic neuronal excitability during normal aging. Aging Cell. 6(3): 327–336. Donix M, Burggren AC, Suthana NA, Siddarth P, Ekstrom AD, Krupa AK, Jones M, Rao A, Martin-Harris L, Ercoli LM, Miller KJ, Small GW, Bookheimer SY. (2010). Longitudinal changes in medial temporal cortical thickness in normal subjects with the APOE-4 polymorphism. NeuroImage. 53(1): 37–43. Drabant EM, Hariri AR, Meyer-Lindenberg A, Munoz KE, Mattay VS, Kolachana BS, Egan MF, Weinberger DR. (2006). Catechol O-methyltransferase Val158Met genotype and neural mechanisms related to affective arousal and regulation. Arch Gen Psychiatry. 63(12): 1396–1406. Drag LL, Bieliauskas LA. (2010). Contemporary review 2009: cognitive aging. J Geriatr Psychiat Neurol. 23(2): 75–93. Dumas JA, Newhouse PA. (2011). The cholinergic hypothesis of cognitive aging revisited again: Cholinergic functional compensation. Pharmacol Biochem Behav. 99(2): 254–261. Duncan CE, Webster MJ, Rothmond DA, Bahn S, Elashoff M, Shannon Weickert C. (2010). Prefrontal GABAA receptor α-subunit expression in normal postnatal human development and schizophrenia. J Psychiatric Res. 44(10): 673–681. Düzel S, Schütze H, Stallforth S, Kaufmann J, Bodammer N, Bunzeck N, Münte TF, Lindenberger U, Heinze H-J, Düzel E. (2008). A close relationship between verbal memory and SN/VTA integrity in young and older adults. Neuropsychologia. 46(13): 3042–3052. Egan M. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 112: 257–269. Egan MF. (2001). Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci U S A. 98: 6917–6922. Egan MF. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 112: 257–269. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM,Straub RE. (2001). Effect of comt val108/158 met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci U S A. 98: 6917–6922. Egan MF,Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A. (2003). The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 112: 257–269. Ehrlich S, Morrow EM, Roffman JL, Wallace SR, Naylor M, Bockholt HJ. (2009). The COMT val108/158met polymorphism and medial temporal lobe volumetry in patients with schizophrenia and healthy adults. American College of Neuropsychopharmacology. 5;53(3): 992–1000. Eichenbaum H. (2000). A cortical-hippocampal system for declarative memory. Nature Rev Neurosci. 1: 41–50. Eichenbaum H, Yonelinas AP, Ranganath C. (2007). The medial temporal lobe and recognition memory. Annu Rev Neurosci. 30: 123–152. Elizabeth MT. (2010). The catechol-O-methyltransferase gene: its regulation and polymorphisms. Int Rev Neurobiol. 95: 7–27. Erickson KI, Kim JS, Suever BL, Voss MW, Francis BM, Kramer AF. (2008). Genetic contributions to age-related decline in executive function: a 10-year longitudinal study of COMT and BDNF polymorphisms. Front Human Neuroscience 2: 11.
3 4 8 part vii : I maging G enetics of C ognitive A ging
Erixon-Lindroth N, Farde L, Robins Wahlin T-B, Sovago J, Halldin C, Bäckman L. (2005). The role of the striatal dopamine transporter in cognitive aging. Psychiatry Res Neuroimag. 138(1): 1–12. Erixon-Lindroth N, Farde L, Wahlin TBR, Sovago J, Halldin C, Bäckman L. (2005). The role of the striatal dopamine transporter in cognitive aging. Psychiatry Res Neuroimag. 138(1): 1–12. Erk S, Meyer-Lindenberg A, Opitz von Boberfeld C, Esslinger C, Schnell K, Kirsch P, Mattheisen M, Mühleisen TW, Cichon S, Witt SH, Rietschel M, Nöthen MM, Walter H. (2011). Hippocampal function in healthy carriers of the CLU Alzheimer’s disease risk variant. J Neurosci. 31(49): 18180–18184. Ettinger U, Kumari V, Collier DA, Powell J, Luzi S, Michel TM, Zedomi O, Williams SCR. (2008). Catechol-O-methyltransferase (COMT) Val158Met genotype is associated with bold response as a function of task characteristic. Neuropsychopharmacology. 33(13): 3046–3057. Eyler LT, et al. (2011). A review of functional brain imaging correlates of successful cognitive aging. Biol Psychiatry. 70(2): 115–122. Fanselow MS, Dong HW. (2010). Are the dorsal and ventral hippocampus functionally distinct structures? Neuron. 65: 7–19. Farah DL. (2011). Epigenetic gene regulation in the adult mammalian brain: Multiple roles in memory formation. Neurobiol Learn Mem. 96(1): 68–78. Farrell SM, Tunbridge EM, Braeutigam S, Harrison PJ. (2012). COMT Val(158)Met genotype determines the direction of cognitive effects produced by catechol-O-methyltransferase inhibition. Biol Psychiatry. 71(6): 538–544. Fennema-Notestine C, Panizzon MS, Thompson WR, Chen CH, Eyler LT, Fischl B, Franz CE, Grant MD, Jak AJ, Jernigan TL, Lyons MJ, Neale MC, Seidman LJ, Tsuang MT, Xian H, Dale AM, Kremen WS. (2011). Presence of ApoE epsilon4 allele associated with thinner frontal cortex in middle age. J Alzheimers Dis. 26(Suppl 3): 49–60. Fernandes MA, Moscovitch M. (2000). Divided attention and memory: evidence of substantial interference effects at retrieval and encoding. J Exp Psychol Gen. 129: 155–176. Ferreira LK, Busatto GF. (2013). Resting-state functional connectivity in normal brain aging. Neurosci Biobehav Rev. 37(3): 384–400 Filippi M, Rocca M. (2007). Magnetization transfer magnetic resonance imaging of the brain, spinal cord, and optic nerve. Neurotherapeutics. 4(3): 401–413. Filippini N, Ebmeier KP, MacIntosh BJ, Trachtenberg AJ, Frisoni GB, Wilcock GK, Beckmann CF, Smith SM, Matthews PM, Mackay CE. (2011). Differential effects of the APOE genotype on brain function across the lifespan. NeuroImage. 54(1): 602–610. Filippini N, MacIntosh BJ, Hough MG, Goodwin GM, Frisoni GB, Smith SM. (2009). Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A. 106: 7209–7214. Finkel D, Andel R, Gatz M, Pedersen NL. (2009). The role of occupational complexity in trajectories of cognitive aging before and after retirement. Psychol Aging. 24(3): 563–573. Fischer H, Nyberg L, Karlsson S, Karlsson P, Brehmer Y, Rieckmann A, MacDonald SWS, Farde L, Bäckman L. (2010). Simulating neurocognitive aging: effects of a dopaminergic antagonist on brain activity during working memory. Biol Psychiatry. 67(6): 575–580. Fjell AM, Walhovd KB. (2010). Structural brain changes in aging: courses, causes and cognitive consequences. Rev Neurosci. 21(3): 187–221. Fjell AM, Westlye LT, Amlien I, Espeseth T, Reinvang I, Raz N, Agartz I, Salat DH, Greve DN, Fischl B, Dale AM, Walhovd KB. (2009). High consistency of regional cortical thinning in aging across multiple samples. Cereb Cortex. 19(9): 2001–2012. Fors S, Lennartsson C, Lundberg O. (2009). Childhood living conditions, socioeconomic position in adulthood, and cognition in later life: exploring the associations.” J Gerontol B Psychol Sci Soc Sci. 64B(6): 750–757. Frias, CMD, Annerbrink K, Westberg L, Eriksson E, Adolfsson R, Nilsson L-G. (2005). Catechol O-methyltransferase Val158Met
polymorphism is associated with cognitive performance in nondemented adults. J Cogn Neurosci. 17(7): 1018–1025. Gajewski PD, Hengstler JG, Golka K, Falkenstein M, Beste C. (2011). The Met-allele of the BDNF Val66Met polymorphism enhances task switching in elderly. Neurobiol Aging. 32(12): 2327.e7–19. Galloway EM, Woo NH,Lu B. (2008). Persistent neural activity in the prefrontal cortex: a mechanism by which BDNF regulates working memory? Progress Brain Res. 169: 251–266. Garcia-Garcia M, Barcelo F, Clemente IC, Escera C. (2010). The role of the dopamine transporter DAT1 genotype on the neural correlates of cognitive flexibility. Eur J Neurosci. 31: 754–760. Garcia-Garcia M, Barceló F, Clemente IC, Escera C. (2010). The role of DAT1 gene on the rapid detection of task novelty. Neuropsychologia. 48(14): 4136–4141. Gebril OH, Kirby J, Savva G, Brayne C, Ince PG. (2011). HFE H63D, C282Y and AGTR1 A1166C polymorphisms and brain white matter lesions in the aging brain. J Neurogenet. 25(1–2): 7–14. Gennatas ED, Cholfin JA, Zhou J, Crawford RK, Sasaki DA, Karydas A, Boxer AL, Bonasera SJ, Rankin KP, Gorno-Tempini ML, Rosen HJ, Kramer JH, Weiner M, Miller BL, Seeley WW. (2012). COMT Val158Met genotype influences neurodegeneration within dopamine-innervated brain structures. Neurology. 78(21): 1663–1669. Gerald TM, Howlett AC, Ward GR, Ho C, Franklin SO. (2008). Gene expression of opioid and dopamine systems in mouse striatum: effects of CB1 receptors, age and sex. Psychopharmacology (Berl). 198(4): 497–508. Glisky EL. (2007). Changes in cognitive function in human aging. In: Riddle DR, editor. Brain Aging: Models, Methods, and Mechanisms. Boca Raton (FL): CRC Press; 2007. Chapter 1. Frontiers in Neuroscience. Goldberg TE, Weinberger DR. (2004). Genes and the parsing of cognitive processes. Trends Cogn Sci. 8(7): 325–335. Gomez-Pinilla F, Hillman C. (2013). The influence of exercise on cognitive abilities. Compr Physiol. 3(1): 403–428. Gong P, Zheng Z, Chi W, Lei X, Wu X, Chen D, Zhang K, Zheng A, Gao X, Zhang F. (2012). An association study of the genetic polymorphisms in 13 neural plasticity-related genes with semantic and episodic memories. J Mol Neurosci. 46(2): 352–361. Good CD. (2001). A voxel-based morphometric study of ageing in 465 normal adult human brains. NeuroImage. 14: 21–36. Good CD, Ashburner J, Frackowiak RS. (2001). Computational neuroanatomy: new perspectives for neuroradiology. Rev Neurol (Paris). 157(8–9 Pt 1): 797–806. Gottschalk WA. (1999). Signaling mechanisms mediating BDNF modulation of synaptic plasticity in the hippocampus. Learn Mem. 6: 243–256. Grady CL. (2008). Cognitive neuroscience of aging. Ann N Y Acad Sci. 1124(1): 127–144. Gregory A, Hayflick S. (1993). Neurodegeneration with brain iron accumulation disorders overview. GeneReviews. Pagon RA, Adam MP, Bird TD, et al. Seattle, WA. In: Pagon RA, Adam MP, Ardinger HH, Wallace SE, Amemiya A, Bean LJH, Bird TD, Dolan CR, Fong CT, Smith RJH, Stephens K, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2015. 2013 Feb 28 [updated 2014 Apr 24]. Gruber O, Hasan A, Scherk H, Wobrock T, Schneider-Axmann T, Ekawardhani S, Schmitt A, Backens M, Reith W, Meyer J, Falkai P. (2012). Association of the brain-derived neurotrophic factor val66met polymorphism with magnetic resonance spectroscopic markers in the human hippocampus: in vivo evidence for effects on the glutamate system. Eur Arch Psychiat Clin Neurosci. 262(1): 23–31. Gruber S, Pinker K, Riederer F, Chmelík M, Stadlbauer A, Bittšanský M, Mlynárik V, Frey R, Serles W, Bodamer O, Moser E. (2008). Metabolic changes in the normal ageing brain: consistent findings from short and long echo time proton spectroscopy.” Eur J Radiol. 68(2): 320–327.
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
349
Gujar SK, Maheshwari S, Bjorkman-Burtscher I, Sundgren PC. (2005). Magnetic resonance spectroscopy. J Neuroophthalmol. 25(3): 217–226. Hajek T, Kopecek M, Hoschl C. (2012). Reduced hippocampal volumes in healthy carriers of brain-derived neurotrophic factor Val66Met polymorphism: meta-analysis. World J Biol Psychiatry. 13(3): 178–187. Harada CN, Natelson Love MC, Triebel KL. (2013). Normal cognitive aging. Clin Geriatr Med. 29(4): 737–752. Hardy J. (2009). The amyloid hypothesis for Alzheimer’s disease: a critical reappraisal. J Neurochemistry. 110(4): 1129–1134. Hardy J, Allsop D. (1991). Amyloid deposition as the central event in the aetiology of Alzheimer’s disease. Trends Pharmacol Sci. 12: 383–388. Hariri AR, Goldberg TE, Mattay VS, Kolachana BS, Callicott JH, Egan MF, Weinberger DR. (2003). Brain-derived neurotrophic factor val66met polymorphism affects human memory-related hippocampal activity and predicts memory performance. J Neurosci. 23(17): 6690–6694. Hariri AR, Weinberger DR. (2003). Imaging genomics. Br Med Bull. 65: 259–270. Harold D. (2009). Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer’s disease. Nature Genet. 41: 1088–1093. Harris SE, Deary IJ. (2011). The genetics of cognitive ability and cognitive ageing in healthy older people. Trends Cogn Sci. 15(9): 388–394. Harris SE, Wright AF, Hayward C, Starr JM, Whalley LJ, Deary IJ. (2005). The functional COMT polymorphism, Val158Met, is associated with logical memory and the personality trait intellect/imagination in a cohort of healthy 79 year olds. Neurosci Lett. 385(1): 1–6. Hashimoto R, Moriguchi Y, Yamashita F, Mori T, Nemoto K, Okada T, Hori H, Noguchi H, Kunugi H, Ohnishi T. (2008). Dose-dependent effect of the Val66Met polymorphism of the brain-derived neurotrophic factor gene on memory-related hippocampal activity. Neurosci Res. 61(4): 360–367. Hatters DM, Peters-Libeu CA, Weisgraber KH. (2006). Apolipoprotein E structure: insights into function. Trends Biochem Sci. 31(8): 445–454. Hazrati LN, Van Cauwenberghe C, Brooks PL, Brouwers N, Ghani M, Sato C, Cruts M, Sleegers K, St George-Hyslop P, Van Broeckhoven C, Rogaeva E. (2012). Genetic association of CR1 with Alzheimer’s disease: a tentative disease mechanism. Neurobiol Aging. 33(12): 2949.e5–2949.e12. Head D, Rodrigue KM, Kennedy KM, Raz N.(2008). Neuroanatomical and cognitive mediators of age-related differences in episodic memory. Neuropsychology. 22(4): 491–507. Healey MK, Campbell KL, Hasher L. (2008). Cognitive aging and increased distractibility: costs and potential benefits. Progr Brain Res. 169: 353–363. Heise V, Filippini N, Ebmeier KP, Mackay CE. (2011). The APOE [epsiv]4 allele modulates brain white matter integrity in healthy adults. Mol Psychiatry. 16(9): 908–916. Herholz K. (2011). Perfusion SPECT and FDG-PET. Int Psychogeriatr. 23(Suppl S2): S25–S31. Hesse S, Barthel H, Murai T, Muller U, Muller D, Seese A, Kluge R, Sabri O. (2003). Is correction for age necessary in neuroimaging studies of the central serotonin transporter? Eur J Nucl Med Mol Imaging. 30(3): 427–430. Hiller JM, Fan LQ, Simon EJ. (1993). Alterations in delta opioid receptor levels in discrete areas of the neocortex and in the globus pallidus of the aging guinea pig: a quantitative autoradiographic study. Brain Res. 614(1–2): 86–98. Hirvonen M, Någren K, Rinne J, Pesonen U, Vahlberg T, Hagelberg N, Hietala J. (2010). COMT Val158Met genotype does not alter cortical or striatal dopamine d2 receptor availability in vivo. Mol Imag Biol. 12(2): 192–197. Honea R, Verchinski BA, Pezawas L, Kolachana BS, Callicott JH, Mattay VS, Weinberger DR, Meyer-Lindenberg A. (2009). Impact of
interacting functional variants in COMT on regional gray matter volume in human brain. NeuroImage. 45(1): 44–51. Honea R, Vidoni E, Harsha A, Burns J. (2009). Impact of APOE on the healthy aging brain: a voxel-based MRI and DTI Study. J Alzheimer’s Disease. 18: 553–564. Honea RA, Cruchaga C, RPerea RD, Saykin AJ, Burns JM, Weinberger DR, Goate AM. (2013). Characterizing the role of brain derived neurotrophic factor genetic variation in Alzheimer’s disease neurodegeneration. PLoS One. 8(9): e76001. Huerta-Rivas A, Pérez-García G, González-Espinosa C, Meneses A. (2010). Time-course of 5-HT6 receptor mRNA expression during memory consolidation and amnesia. Neurobiol Learn Memory. 93(1): 99–110. Hutton C, Draganski B, Ashburner J, Weiskopf N. (2009). A comparison between voxel-based cortical thickness and voxel-based morphometry in normal aging. NeuroImage. 48(2): 371–380. Ikonomovic MD. (1999). Distribution of glutamate receptor subunit NMDAR1 in the hippocampus of normal elderly and patients with Alzheimer’s disease. Exp Neurol. 160: 194–204. Inglese M, Ge Y, Grossman RI. (2007). Magnetization transfer imaging neurodegeneration in multiple sclerosis. In Filippi M, Rovaris M, Comi G, eds., Neurodegeneration in Multiple Sclerosis (pp. 47–53). Milan: Springer. Inkster B, Rao AW, Ridler K, Filippini N, Whitcher B, Nichols TE, Wetten S, Gibson RA, Borrie M, Kertesz A, Guzman DA, Loy-English I, Williams J, Saemann PG, Auer DP, Holsboer F, Tozzi F, Muglia P, Merlo-Pich E, Matthews PM. (2012). Genetic variation in GOLM1 and prefrontal cortical volume in Alzheimer’s disease. Neurobiol Aging. 33(3): 457–465. Iyo M, Yamasaki T. (1993). The detection of age-related decrease of dopamine D1, D2 and serotonin 5-HT2 receptors in living human brain. Progr Neuro-Psychopharm Biol Psychiatry. 17(3): 415–421. Jernigan TL. (2001). Effects of age on tissues and regions of the cerebrum and cerebellum. Neurobiol Aging. 22: 581–594. Jiang H, Xie T, Ramsden DB, Ho SL. (2003). Human catechol-Omethyltransferase down-regulation by estradiol. Neuropharmacology. 45(7): 1011–1018. Jin C, Li W, Yuan J, Xu X, Cheng Z. (2012). Association of the CR1 polymorphism with late-onset Alzheimer’s disease in Chinese Han populations: a meta-analysis. Neurosci Lett. 527(1): 46–49. Johannsen S, Duning K, Pavenstädt H, Kremerskothen J, Boeckers TM. (2008). Temporal-spatial expression and novel biochemical properties of the memory-related protein KIBRA. Neuroscience. 155(4): 1165–1173. Kaasinen V, Vilkman H, Hietala J, Någren K, Helenius H, Olsson H, Farde L, Rinne JO. (2000). Age-related dopamine D2/D3 receptor loss in extrastriatal regions of the human brain. Neurobiol Aging. 21(5): 683–688. Käenmäki M, Tammimäki A, Myöhänen T, Pakarinen K, Amberg C, Karayiorgou M, Gogos JA, Männistö PT. (2010). Quantitative role of COMT in dopamine clearance in the prefrontal cortex of freely moving mice. J Neurochem. 114(6): 1745–1755. Kang HJ, Kawasawa YI, Cheng F, Zhu Y, Xu X, Li M, Sousa AMM, Pletikos M, Meyer KA, Sedmak G, Guennel T, Shin Y, Sestan N. (2011). Spatio-temporal transcriptome of the human brain. Nature. 478(7370): 483–489. Karlsson S, Nyberg L, Karlsson P, Fischer H, Thilers P, MacDonald S, Brehmer Y, Rieckmann A, Halldin C, Farde L, Bäckman L. (2009). Modulation of striatal dopamine D1 binding by cognitive processing. NeuroImage. 48(2): 398–404. Karran E, Mercken M, Strooper BD. (2011). The amyloid cascade hypothesis for Alzheimer’s disease: an appraisal for the development of therapeutics. Nat Rev Drug Discov. 10(9): 698–712. Kaup AR, Mirzakhanian H, Jeste DV, Eyler LT. (2011). A review of the brain structure correlates of successful cognitive aging. J Neuropsychiat Clin Neurosci. 23(1): 6–15. Kauppi K, Nilsson L-G, Adolfsson R, Eriksson E, Nyberg L. (2011). KIBRA polymorphism is related to enhanced memory and elevated hippocampal processing.” J Neurosci. 31(40): 14218–14222.
3 5 0 part vii : I maging G enetics of C ognitive A ging
Kawamura K, Kimura Y, Tsukada H, Kobayashi T, Nishiyama S, Kakiuchi T, Ohba H, Harada N, Matsuno K, Ishii K, Ishiwata K. (2003). An increase of sigma1 receptors in the aged monkey brain. Neurobiol Aging. 24(5): 745–752. Kawarai T, Kawakami H, Yamamura Y, Nakamura S. (1997). Structure and organization of the gene encoding human dopamine transporter. Gene 195: 11–18. Kennedy KM, Raz N. (2009). Aging white matter and cognition: differential effects of regional variations in diffusion properties on memory, executive functions, and speed. Neuropsychologia. 47(3): 916–927. Kennedy KM, Rodrigue KM, Land SJ, Raz N. (2009). BDNF val66met polymorphism influences age differences in microstructure of the corpus callosum. Frontiers in Human Neuroscience. 3: 19. Kim S, Swaminathan S, Shen L, Risacher SL, Nho K, Foroud T, Shaw LM, Trojanowski JQ, Potkin SG, Huentelman MJ, Craig DW, DeChairo BM, Aisen PS, Petersen RC, Weiner MW, Saykin AJ. (2011). Genome-wide association study of CSF biomarkers Abeta1-42, t-tau, and p-tau181p in the ADNI cohort. Neurology. 76(1): 69–79. Korte M, Kang, H Bonhoeffer T, Schuman E. (1998). A role for BDNF in the late-phase of hippocampal long-term potentiation. Neuropharmacology. 37: 553–559. Kramer AF, Colcombe SJ, McAuley E, Scalf PE, Erickson KI. (2005). Fitness, aging and neurocognitive function. Neurobiol Aging. 26(Suppl 1): 124–127. Kremerskothen J, Plaas C, Büther K, Finger I, Veltel S, Matanis T, Liedtke T, Barnekow A. (2003). Characterization of KIBRA, a novel WW domain-containing protein. Biochem Biophys Res Comm. 300(4): 862–867. Kumakura Y, Vernaleken I, Buchholz H-G, Borghammer P, Danielsen E, Gründer G, Heinz A, Bartenstein P, Cumming P. (2010). Age-dependent decline of steady state dopamine storage capacity of human brain: An FDOPA PET study. Neurobiol Aging. 31(3): 447–463. Kumakura Y, Vernaleken I, Grunder G, Bartenstein P, Gjedde A, Cumming P. (2005). PET studies of net blood-brain clearance of FDOPA to human brain: age-dependent decline of [lsqb]18F[rsqb]fluorodopamine storage capacity. J Cereb Blood Flow Metab. 25(7): 807–819. Lam M, Eng GK, Rapisarda A, Subramaniam M, Kraus M, Keefe RS, Collinson SL. (2013). Formulation of the age-education index: measuring age and education effects in neuropsychological performance. Psychol Assess. 25(1): 61–70. Lambert JC. (2009). Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer’s disease. Nature Genet. 41: 1094–1099. Lau CG, Zukin RS. (2007). NMDA receptor trafficking in synaptic plasticity and neuropsychiatric disorders. Nat Rev Neurosci. 8(6): 413–426. Lee T, Henry JD, Trollor JN, Sachdev PS. (2010). Genetic influences on cognitive functions in the elderly: A selective review of twin studies. Brain Res Rev. 64(1): 1–13. Lee T, Sachdev P. (2014). The contributions of twin studies to the understanding of brain ageing and neurocognitive disorders. Curr Opin Psychiatry. 27(2): 122–127. Lein E, Hohn A, Shatz C. (2000). Dynamic regulation of BDNF and NT-3 expression during visual system development. J Comp Neurol. 420: 1–18. Lemaitre H, Goldman AL, Sambataro F, Verchinski BA, Meyer- Lindenberg A, Weinberger DR,Mattay VS. (2012). Normal age-related brain morphometric changes: nonuniformity across cortical thickness, surface area and gray matter volume? Neurobiol Aging. 33(3): 617.e1–9. Lemmens, R., et al. (2007). Association of apolipoprotein E epsilon2 with white matter disease but not with microbleeds. Stroke 38(4): 1185–1188.
Leveille F. (2008). Neuronal viability is controlled by a functional relation between synaptic and extrasynaptic NMDA receptors. FASEB J. 22: 4258–4271. Li S-C, Chicherio C, Nyberg L, von Oertzen T, Nagel IE, Papenberg G, Sander T, Heekeren HR, Lindenberger U, Bäckman L. (2009). Ebbinghaus revisited: influences of the BDNF Val66Met polymorphism on backward serial recall are modulated by human aging.” J Cogn Neurosci. 22(10): 2164–2173. Li S-C, Lindenberger U, Sikström S. (2001). Aging cognition: from neuromodulation to representation. Trends Cogn Sci. 5(11): 479–486. Li Z, Moore AB, Tyner C, Hu X. (2009). Asymmetric connectivity reduction and its relationship to “HAROLD” in aging brain. Brain Res. 1295: 149–158. Lipsky RH, Marini AM. (2007). Brain-derived neurotrophic factor in neuronal survival and behavior-related plasticity. Ann N Y Acad Sci 1122: 130–143. Lise A. (2009). Exploring the generality of retest effects: commentary on “When does age-related cognitive decline begin?”. Neurobiol Aging. 30(4): 525–527. Liu L. (2004). Role of NMDA receptor subtypes in governing the direction of hippocampal synaptic plasticity. Science. 304: 1021–1024. Liu M-E, Hong C-J, Liou Y-J, Tsai Y-L, Hsieh C-H, Tsai S-J. (2008). Association study of a functional catechol-O-methyltransferase polymorphism and executive function in elderly males without dementia. Neurosci Lett. 436(2): 193–195. Liu P, Bilkey DK, Darlington CL, Smith PF. (2003). Cannabinoid CB1 receptor protein expression in the rat hippocampus and entorhinal, perirhinal, postrhinal and temporal cortices: regional variations and age-related changes. Brain Res. 979(1–2): 235–239. Liu P, Smith PF, Darlington CL. (2008). Glutamate receptor subunits expression in memory-associated brain structures: regional variations and effects of aging. Synapse 62(11): 834–841. Liu T, Wen W, Zhu W, Trollor J, Reppermund S, Crawford J, Jin JS, Luo S, Brodaty H, Sachdev P. (2010). The effects of age and sex on cortical sulci in the elderly. NeuroImage. 51(1): 19–27. Logan JM, et al. (2002). Under-recruitment and nonselective recruitment: dissociable neural mechanisms associated with aging. Neuron 33(5): 827–840. Lövdén M, Schaefer S, Noack H, Kanowski M, Kaufmann J, Tempelmann C, Bodammer NC, Kühn S, Heinze H-J, Lindenberger U, Düzel E, Bäckman L. (2011). Performance-related increases in hippocampal n-acetylaspartate (NAA) induced by spatial navigation training are restricted to bdnf val homozygotes. Cereb Cortex. 21(6): 1435–1442. Lu B. (2003). BDNF and activity-dependent synaptic modulation. Learn Mem. 10: 86–98. Lu B. (2004). Acute and long-term regulation of synapses by neurotrophins. Prog Brain Res. 146: 137–150. Lu B, Pang PT, Woo NH. (2005). The yin and yang of neurotrophin action. Nat Rev Neurosci 6(8): 603–614. Ma SY, Ciliax BJ, Stebbins G, Jaffar S, Joyce JN, Cochran EJ, Kordower JH, Mash DC, Levey AI, Mufson EJ. (1999). Dopamine transporter-immunoreactive neurons decrease with age in the human substantia nigra. J Compar Neurol. 409(1): 25–37. Madden DJ, Bennett IJ, Burzynska A, Potter GG, Chen N-K, Song AW. (2012). Diffusion tensor imaging of cerebral white matter integrity in cognitive aging. BBA Mol Basis Dis 1822(3): 386–400. Madden DJ, Bennett IJ, Song AW. (2009). Cerebral white matter integrity and cognitive aging: contributions from diffusion tensor imaging. Neuropsychol Rev. 19(4): 415–435. Madsen K, Haahr MT, Marner L, Keller SH, Baare WF, Svarer C, Hasselbalch SG, Knudsen GM. (2011). Age and sex effects on 5-HT4 receptors in the human brain: a [lsqb]11C[rsqb]SB207145 PET study. J Cereb Blood Flow Metab. 31(6): 1475–1481. Magnusson KR, Brim BL, Das SR. (2010). Selective vulnerabilities of N-methyl-D-aspartate (NMDA) receptors during brain aging. Front Ag Neurosci. 2(11): 1–15.
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
351
Makuch L, Volk L, Anggono V, Johnson RC, Yu Y, Duning K, Kremerskothen J, Xia J, Takamiya K, Huganir RL. (2011). Regulation of AMPA receptor function by the human memory-associated gene KIBRA. Neuron 71(6): 1022–1029. Malhotra AK, Kestler LJ, Mazzanti C, Bates JA, Goldberg T, Goldman D. (2002). A functional polymorphism in the COMT gene and performance on a test of prefrontal cognition. Am J Psychiatry. 159(4): 652–654. Mandelman SD, Grigorenko EL. (2012). BDNF Val66Met and cognition: all, none, or some? A meta-analysis of the genetic association. Genes Brai Behav. 11(2): 127–136. Marcusson JO, Morgan DG, Winblad B, Finch CE. (1984). Serotonin-2 binding sites in human frontal cortex and hippocampus. Selective loss of S-2A sites with age. Brain Res. 311: 51–56. Marian AJ. (2012). Molecular genetic studies of complex phenotypes. Transl Res. 159(2): 64–79. Marner L, Gillings N, Madsen K, Erritzoe D, Baaré WFC, Svarer C, Hasselbalch SG, Knudsen GM. (2010). Brain imaging of serotonin 4 receptors in humans with [11C]SB207145-PET. NeuroImage 50(3): 855–861. Martinowich K, Hattori D, Wu H, Fouse S, He F, Hu Y. (2003). DNA methylation-related chromatin remodeling in activity-dependent BDNF gene regulation. Science 302: 890–893. Martinowich K, Manji H, Lu B. (2007). New insights into BDNF function in depression and anxiety. Nat Neurosci. 10: 1089–1093. Matsuda H. (2013). Voxel-based morphometry of brain MRI in normal aging and Alzheimer’s disease.” Aging Dis. 4(1): 29–37. Mattay VS, Goldberg TE, Fera F, Hariri AR, Tessitore A, Egan MF. (2003). Catechol O-methyltransferase val158-met genotype and individual variation in the brain response to amphetamine. Proc Natl Acad Sci U S A. 100: 6186–6191. Mattay VS, Goldberg TE, Sambataro F, Weinberger DR. (2008). Neurobiology of cognitive aging: insights from imaging genetics. Biol Psychology. 79: 9–22. Mattay VS, Goldberg TE, Sambataro F, Weinberger DR. (2008). Neurobiology of cognitive aging: Insights from imaging genetics. Biol Psychology. 79(1): 9–22. Mattson MP, Maudsley S, Martin B. (2004). BDNF and 5-HT: a dynamic duo in age-related neuronal plasticity and neurodegenerative disorders. Trends Neurosci. 27: 589–594. McAllister AK, Katz LC, Lo DC. (1999). Neurotrophins and synaptic plasticity. Annu Rev Neurosci. 22: 295–318. McArdle JJ, Plassman BL. (2009). A biometric latent curve analysis of memory decline in older men of the NAS-NRC twin registry. Behav Genet. 39(5): 472–495. McDowd JM, Craik FI. (1988). Effects of aging and task difficulty on divided attention performance. J Exper Psychology. 14(2): 267–280. Meier S, Demirakca T, Brusniak W, Wolf I, Liebsch K, Tunc-Skarka N, Nieratschker V, Witt SH, Matthaus F, Ende G, Flor H, Rietschel M, Diener C, Schulze TG. (2012). SCN1A affects brain structure and the neural activity of the aging brain. Biol Psychiatry. 72(8): 677–683. Mengel-From J, Christensen K, McGue M, Christiansen L. (2011). Genetic variations in the CLU and PICALM genes are associated with cognitive function in the oldest old. Neurobiol Aging. 32(3): 554.e7–11. Meyer MR. (1998). APOE genotype predicts when—not whether—one is predisposed to develop Alzheimer disease. Nature Genet. 19: 321–322. Meyer PT, Elmenhorst D, Boy C, Winz O, Matusch A, Zilles K, Bauer A. (2007). Effect of aging on cerebral A1 adenosine receptors: a [18F] CPFPX PET study in humans. Neurobiol Aging. 28(12): 1914–1924. Meyer-Lindenberg A, Kohn PD, Kolachana B, Kippenhan S, McInerney-Leo A, Nussbaum R. (2005). Midbrain dopamine and prefrontal function in humans: interaction and modulation by COMT genotype. Nat Neurosci 8: 594–596.
Meyer-Lindenberg A, Nichols T, Callicott JH, Ding J, Kolachana B, Buckholtz J. (2006). Impact of complex genetic variation in COMT on human brain function. Mol Psychiatry 11: 867–877. Mill J, Asherson P, Browes C, U DS,Craig I. (2002). Expression of the dopamine transporter gene is regulated by the 3[prime] utr vntr: evidence from brain and lymphocytes using quantitative rt-pcr. Am J Med Genetics. 114: 975–979. Milnik A, Heck A, Vogler C, Heinze HJ, de Quervain DJ, Papassotiropoulos A. (2012). Association of KIBRA with episodic and working memory: a meta-analysis. Am J Med Genet B Neuropsychiatr Genet. 159B(8): 958–969. Minati L, Grisoli M, Bruzzone MG. (2007). MR spectroscopy, functional MRI, and diffusion-tensor imaging in the aging brain: a conceptual review.” J Geriatr Psychiatr Neurol. 20(1): 3–21. Miyajima F, Ollier W, Mayes A, Jackson A, Thacker N, Rabbitt P, Pendleton N, Horan M, Payton A. (2008). Brain-derived neurotrophic factor polymorphism Val66Met influences cognitive abilities in the elderly. Genes Brain Behav. 7(4): 411–417. Mizoguchi K, Shoji H, Tanaka Y, Maruyama W, Tabira T. (2009). Age-related spatial working memory impairment is caused by prefrontal cortical dopaminergic dysfunction in rats. Neurosci. 162(4): 1192–1201. Mizuno M, Yamada K, He J, Nakajima A, Nabeshima T. (2003). Involvement of BDNF receptor TrkB in spatial memory formation. Learn Mem. 10: 108–115. Molendijk ML, Bus BA, Spinhoven P, Kaimatzoglou A, Oude Voshaar RC, Penninx BW, van IMH, Elzinga BM. (2012). A systematic review and meta-analysis on the association between BDNF val(66) met and hippocampal volume—a genuine effect or a winners curse? Am J Med Genet B Neuropsychiatr Genet. 159b(6): 731–740. Mondadori CRA, de Quervain DJ-F, Buchmann A, Mustovic H, Wollmer MA, Schmidt CF, Boesiger P, Hock C, Nitsch RM, Papassotiropoulos A, Henke K. (2007). Better memory and neural efficiency in young apolipoprotein e ε4 carriers. Cereb Cortex. 17(8): 1934–1947. Morris JC. (2010). APOE predicts amyloid-β but not tau Alzheimer pathology in cognitively normal aging. Ann Neurol. 67: 122–131. Morrison JH, Baxter MG (2012). The ageing cortical synapse: hallmarks and implications for cognitive decline. Nat Rev Neurosci 13(4): 240–250. Mosconi L. (2005). Brain glucose metabolism in the early and specific diagnosis of Alzheimer’s disease. FDG-PET studies in MCI and AD. Eur J Nucl Med Mol Imaging. 32(4): 486–510. Moses-Kolko EL, Price JC, Shah N, Berga S, Sereika SM, Fisher PM, Coleman R, Becker C, Mason NS, Loucks T, Meltzer CC. (2011). Age, sex, and reproductive hormone effects on brain serotonin-1A and serotonin-2A receptor binding in a healthy population. Neuropsychopharmacology. 36(13): 2729–2740. Mozley LH, Gur RC, Mozley PD, Gur RE. (2001). Striatal dopamine transporters and cognitive functioning in healthy men and women. Am J Psychiatry. 158: 1492–1499. Mueller SG, Weiner MW. (2009). Selective effect of age, Apo e4, and Alzheimer’s disease on hippocampal subfields. Hippocampus. 19: 558–564. Muse J, Emery M, Sambataro F, Lemaitre H, Tan H, Chen Q, Kolachana B, Das S, Callicott JH, Weinberger DR, Mattay VS. (2013). WWC1 genotype modulates age-related decline in episodic memory function across the adult life span. Biol Psychiatry. 75(9): 693–700. Nacmias B, Bessi V, Bagnoli S, Tedde A, Cellini E, Piccini, Sorbi S, Bracco L. (2008). KIBRA gene variants are associated with episodic memory performance in subjective memory complaints. Neurosci Lett. 436(2): 145–147. Nagel IE, Chicherio C, Li S-C, Von Oertzen T, Sander T, Villringer A, Heekeren HR, Bckman L, Lindenberger u. (2008). Human aging magnifies genetic effects on executive functioning and working memory. Front Human Neurosci. 2: 1. Nakazawa K. (2002). Requirement for hippocampal CA3 NMDA receptors in associative memory recall. Science. 297: 211–218.
3 5 2 part vii : I maging G enetics of C ognitive A ging
Nebel A, Kleindorp R, Caliebe A, Nothnagel M, Blanché H, Junge O, Wittig M, Ellinghaus D, Flachsbart F, Wichmann H-E, Meitinger T, Nikolaus S, Franke A, Krawczak M, Lathrop M, Schreiber S. (2011). A genome-wide association study confirms APOE as the major gene influencing survival in long-lived individuals. Mech Ageing Devel. 132(6–7): 324–330. Need AC. (2009). A genome-wide study of common SNPs and CNVs in cognitive performance in the CANTAB. Hum Mol Genet. 18: 4650–4661. Need AC, Attix DK, McEvoy JM, Cirulli ET, Linney KN, Wagoner AP. (2008). Failure to replicate effect of Kibra on human memory in two large cohorts of European origin. Am J Med Genet B Neuropsychiatr Genet. 147B: 667–668. Nemoto K, Ohnishi T, Mori T, Moriguchi Y, Hashimoto R, Asada T, Kunugi H. (2006). The Val66Met polymorphism of the brain-derived neurotrophic factor gene affects age-related brain morphology. Neurosci Lett. 397(1–2): 25–29. Nilsson L-G, Sternäng O, Rönnlund M, Nyberg L. (2009). Challenging the notion of an early-onset of cognitive decline. Neurobiol Aging. 30(4): 521–524. Nobler MS, Mann JJ., Sackeim HA. (1999). Serotonin, cerebral blood flow, and cerebral metabolic rate in geriatric major depression and normal aging. Brain Res Rev. 30(3): 250–263. O’Hara R, Miller E, Liao C-P, Way N, Lin X, Hallmayer J. (2006). COMT genotype, gender and cognition in community-dwelling, older adults. Neurosci Lett. 409(3): 205–209. Oberauer K. (2001). Removing irrelevant information from working memory: a cognitive aging study with the modified Sternberg Task. J Exp Psychol. 27(4): 948–957. Oh MM, Oliveira FA, Disterhoft JF. (2010). Learning and aging related changes in intrinsic neuronal excitability. Front Aging Neurosci. 2: 2. Oosterman JM, Boeschoten MS, Eling PA, Kessels RP, Maes JH. (2014). Simple and complex rule induction performance in young and older adults: contribution of episodic memory and working memory. J Int Neuropsychol Soc. 20(3): 333–341. Palombo DJ, Amaral RS, Olsen RK, Muller DJ, Todd RM, Anderson AK, Levine B. (2013). KIBRA polymorphism is associated with individual differences in hippocampal subregions: evidence from anatomical segmentation using high-resolution MRI. J Neurosci. 33(32): 13088–13093. Pang PT, Lu B. (2004). Regulation of late-phase LTP and long-term memory in normal and aging hippocampus: role of secreted proteins tPA and BDNF. Ageing Res Rev. 3: 407–430. Pannese E. (2011). Morphological changes in nerve cells during normal aging. Brain Struct Funct. 216(2): 85–89. Papenberg G, et al. (2013). Dopaminergic gene polymorphisms affect long-term forgetting in old age: further support for the magnification hypothesis. J Cogn Neurosci. 25(4): 571–579. Papassotiropoulos A, de Quervain DJF. (2011). Genetics of human episodic memory: dealing with complexity. Trends Cogn Sci. 15(9): 381–387. Papassotiropoulos A, Henke K, Stefanova E, Aerni A, Muller A, Demougin P, Vogler C, Sigmund JC, Gschwind L, Huynh KD, Coluccia D, … de Quervain DJF. (2011). A genome-wide survey of human short-term memory. Mol Psychiatry. 16(2): 184–192. Papassotiropoulos A, Stefanova E, Vogler C, Gschwind L, Ackermann S, Spalek K, Rasch B, Heck A, Aerni A, Hanser E, Demougin P, Huynh KD, Luechinger R, Klarhofer M, Novakovic I, Kostic V, Boesiger P, Scheffler K, de Quervain DJF. (2011). A genome-wide survey and functional brain imaging study identify CTNNBL1 as a memory-related gene. Mol Psychiatry. 18(2): 255–263. Papassotiropoulos A, Stephan DA, Huentelman MJ, Hoerndli FJ, Craig DW, Pearson JV. (2006). Common KIBRA alleles are associated with human memory performance. Science. 314: 475–478. Park DC, Reuter-Lorenz P. (2009). The adaptive brain: aging and neurocognitive scaffolding. Annu Rev Psychol. 60: 173–196. Payton A. (2009). The impact of genetic research on our understanding of normal cognitive ageing: 1995 to 2009. Neuropsychol Rev. 19(4): 451–477.
Pericak-Vance MA, Bebout JL, Gaskell PC, Jr, Yamaoka LH, Hung WY, Alberts MJ, Walker AP, Bartlett RJ, Haynes CA, Welsh KA, et al. (1991). Linkage studies in familial Alzheimer disease: evidence for chromosome 19 linkage. Am J Hum Genet. 48(6): 1034–1050. Pettigrew C, Martin RC. (2014). Cognitive declines in healthy aging: evidence from multiple aspects of interference resolution. Psychol Aging. 29(2): 187–204. Pezawas L. (2004). Interaction of SERT & BDNF: Susceptibility for depression is reflected in morphometric changes of critical limbic circuits in humans. Neuropsychopharmacology. 29: S200. Pezawas L, Verchinski BA, Mattay VS, Callicott JH, Kolachana BS, Straub RE, Egan MF, Meyer-Lindenberg A, Weinberger DR. (2004). The brain-derived neurotrophic factor val66met polymorphism and variation in human cortical morphology. J Neurosci. 24(45): 10099–10102. Picciotto MR, Zoli M. (2002). Nicotinic receptors in aging and dementia. J Neurobiol. 53(4): 641–655. Picq J-L, Aujard F, Volk A, Dhenain M. (2010) Age-related cerebral atrophy in nonhuman primates predicts cognitive impairments. Neurobiol Aging. 33(6): 1096–1109. Pirker W, Asenbaum S, Hauk M, Kandlhofer S, Tauscher J, Willeit M, Neumeister A, Praschak-Rieder N, Angelberger P, Brucke T. (2000). Imaging serotonin and dopamine transporters with 123I-beta-CIT SPECT: binding kinetics and effects of normal aging. J Nucl Med. 41(1): 36–44. Poo MM. (2001). Neurotrophins as synaptic modulators. Nature Rev Neurosci. 2: 24–32. Preuschhof C, Heekeren HR, Li S-C, Sander T, Lindenberger U, Bäckman L. (2010). KIBRA and CLSTN2 polymorphisms exert interactive effects on human episodic memory. Neuropsychologia. 48(2): 402–408. Rabbitt P. (2011). Between-individual variability and interpretation of associations between neurophysiological and behavioral measures in aging populations: Comment on Salthouse (2011). Psychol Bull. 137(5): 785–789. Rabiner EA, Messa C, Sargent PA, Husted-Kjaer K, Montgomery A, Lawrence AD, Bench CJ, Gunn RN, Cowen P, Grasby PM. (2002). A database of [11C]WAY-100635 binding to 5-HT1A receptors in normal male volunteers: normative data and relationship to methodological, demographic, physiological, and behavioral variables. NeuroImage 15(3): 620–632. Rabinovici GD, Jagust WJ. (2009). Amyloid imaging in aging and dementia: Testing the amyloid hypothesis in vivo. Behav Neurol. 21(1): 117–128. Rajah MN, D’Esposito M. (2005). Region-specific changes in prefrontal function with age: a review of PET and fMRI studies on working and episodic memory. Brain 128(9): 1964–1983. Rajah MN, Kromas M, Han JE, Pruessner JC. (2010). Group differences in anterior hippocampal volume and in the retrieval of spatial and temporal context memory in healthy young versus older adults. Neuropsychologia. 48(14): 4020–4030. Raz N. (2009). Decline and compensation in aging brain and cognition: promises and constraints. Neuropsychol Rev. 19(4): 411–414. Raz N, Rodrigue KM. (2006). Differential aging of the brain: patterns, cognitive correlates and modifiers. Neurosci Biobehav Rev. 30: 730–748. Raz N, Rodrigue KM, Haacke EM. (2007). Brain aging and its modifiers: Insights from in vivo neuromorphometry and susceptibility weighted imaging. Ann NY Acad Sci. 1097: 84–93. Raz N, Rodrigue KM, Kennedy KM, Land S. (2009). Genetic and vascular modifiers of age-sensitive cognitive skills: effects of COMT, BDNF, ApoE, and hypertension. Neuropsychology. 23: 105–116. Raz N, Yang Y, Dahle CL, Land S. (2012). Volume of white matter hyperintensities in healthy adults: contribution of age, vascular risk factors, and inflammation-related genetic variants. BBA Mol Basis Dis. 1822(3): 361–369. Reuser M, Willekens F, Bonneux L. (2011). Higher education delays and shortens cognitive impairment: a multistate life table analysis of the US Health and Retirement Study. Eur J Epidemiol. 26(5): 395–403.
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
353
Rieckmann A, Karlsson S, Fischer H, Bäckman L. (2011). Caudate dopamine D1 receptor density is associated with individual differences in frontoparietal connectivity during working memory. J Neurosci. 31(40): 14284–14290. Rieckmann A, Karlsson S, Karlsson P, Brehmer Y, Fischer H, Farde L, Nyberg L, Bäckman L. (2011). Dopamine D1 receptor associations within and between dopaminergic pathways in younger and elderly adults: links to cognitive performance. Cereb Cortex. 21(9): 2023–2032. Rissman RA, De Blas AL, Armstrong DM. (2007). GABAA receptors in aging and Alzheimer’s disease. J Neurochem. 103(4): 1285–1292. Rodrigue K, Kennedy K, Park D. (2009). Beta-amyloid deposition and the aging brain. Neuropsychol Rev. 19(4): 436–450. Rodrigue, KM, Kennedy KM, Devous MD, Sr, Rieck JR, Hebrank AC, Diaz-Arrastia R, Mathews D, Park DC. (2012). beta-Amyloid burden in healthy aging: Regional distribution and cognitive consequences. Neurology. 78(6): 387–395. Rodrigue KM, Raz N. (2004). Shrinkage of the entorhinal cortex over five years predicts memory performance in healthy adults. J Neurosci. 24(4): 956–963. Rodríguez-Rodríguez E, Infante J, Llorca J, Mateo I, Sánchez-Quintana C, García-Gorostiaga I, Sánchez-Juan P, Berciano J, Combarros O. (2009). Age-dependent association of KIBRA genetic variation and Alzheimer’s disease risk. Neurobiol Aging. 30(2): 322–324. Rohwedder S, Willis RJ. (2010). Mental retirement. J Econ Perspect. 24(1): 119–138. Rollo C. (2009). Dopamine and aging: intersecting facets. Neurochem Res. 34(4): 601–629. Romanczyk TB, Weickert CS, Webster MJ, Herman MM, Akil M, Kleinman JE. (2002). Alterations in trkB mRNA in the human prefrontal cortex throughout the lifespan. Eur J Neurosci. 15(2): 269–280. Rosenzweig ES, Barnes CA. (2003). Impact of aging on hippocampal function: plasticity, network dynamics, and cognition. Prog Neurobiol. 69: 143–179. Ross AJ, Sachdev PS, Wen W, Valenzuela MJ, Brodaty H. (2005). Cognitive correlates of 1H MRS measures in the healthy elderly brain. Brain Res Bull. 66(1): 9–16. Roth TL, Sweatt JD. (2011). Epigenetic marking of the BDNF gene by early-life adverse experiences. Horm Behav. 59(3): 315–320. Rothman DL, Behar KL, Hyder F, Shulman RG. (2003). In vivo nmr studies of the glutamate neurotransmitter flux and neuroenergetics: implications for brain function. Ann Rev Physiol. 65(1): 401–427. Rovaris M, Iannucci G, Cercignani M, Sormani MP, De Stefano N, Gerevini S, Comi G, Filippi M. (2003). Age-related changes in conventional, magnetization transfer, and diffusion-tensor MR imaging findings: study with whole-brain tissue histogram analysis. Radiology. 227(3): 731–738. Rowe JB, Hughes L, Williams-Gray CH, Bishop S, Fallon S, Barker RA, Owen M. (2010). The val158met COMT polymorphism’s effect on atrophy in healthy aging and Parkinson’s disease. Neurobiol Aging. 31(6): 1064–1068. Saito N, Sakai O, Ozonoff A, Jara H. (2009). Relaxo-volumetric multispectral quantitative magnetic resonance imaging of the brain over the human lifespan: global and regional aging patterns. Magn Reson Imaging. 27(7): 895–906. Salat DH, Lee SY, van der Kouwe AJ, Greve DN, Fischl B, Rosas HD. (2009). Age-associated alterations in cortical gray and white matter signal intensity and gray to white matter contrast. NeuroImage. 48(1): 21–28. Salat DH, Tuch DS, Greve DN, van der Kouwe AJW, Hevelone ND, Zaleta AK, Rosen BR, Fischl B, Corkin S, Rosas HD, Dale AM. (2005). Age-related alterations in white matter microstructure measured by diffusion tensor imaging. Neurobiol Aging. 26(8): 1215–1227. Salvatore MF, Apparsundaram S, Gerhardt GA. (2003). Decreased plasma membrane expression of striatal dopamine transporter in aging. Neurobiol Aging. 24(8): 1147–1154.
Sambataro F, Murty VP, Lemaitre HS, Reed JD, Das S, Goldberg TE, Callicott JH, Weinberger DR, Mattay VS. (2010). BNDF modulates normal human hippocampal ageing. Mol Psychiatry. 15(2): 116–118. Sambataro F, Reed JD, Murty VP, Das S, Tan HY, Callicott JH, Weinberger DR, Mattay VS. (2009). Catechol-O-methyltransferase Valine158Methionine polymorphism modulates brain networks underlying working memory across adulthood.” Biol Psychiatry. 66(6): 540–548. Sanders JL, Minster RL, Barmada MM, Matteini AM, Boudreau RM, Christensen K, Mayeux R, Borecki IB, Zhang Q, Perls T, Newman AB. (2014). Heritability of and mortality prediction with a longevity phenotype: the healthy aging index. J Gerontol A Biol Sci Med Sci. 69(4): 479–485. Saunders DE, Howe FA, van den Boogaart A, Griffiths JR, Brown MM. (1999). Aging of the adult human brain: in vivo quantitation of metabolite content with proton magnetic resonance spectroscopy. J Magn Reson Imaging. 9(5): 711–716. Schachter F, Faure-Delanef L, Guenot F, Rouger H, Froguel P, Lesueur-Ginot L. (1994). Genetic associations with human longevity at the APOE and ACE loci. Nat Genet. 6: 29–32. Schaie, KW (2009). “When does age-related cognitive decline begin?” Salthouse again reifies the “cross-sectional fallacy.” Neurobiol Aging. 30(4): 528–529. Schaper K, Kolsch H, Popp J, Wagner M, Jessen F. (2008). KIBRA gene variants are associated with episodic memory in healthy elderly. Neurobiol Aging. 29: 1123–1125. Schiepers OJG, Harris SE, Gow AJ, Pattie A, Brett CE, Starr JM, Deary IJ. (2012). APOE E4 status predicts age-related cognitive decline in the ninth decade: longitudinal follow-up of the Lothian Birth Cohort 1921. Mol Psychiatry. 17(3): 315–324. Schimanski LA, Barnes CA. (2010). Neural protein synthesis during aging: effects on plasticity and memory. Front Aging Neurosci. 2(pii): 26. Schmack K, Schlagenhauf F, Sterzer P, Wrase J, Beck A, Dembler T, Kalus P, Puls I, Sander T, Heinz A, Gallinat J. (2008). Catechol-Omethyltransferase val158met genotype influences neural processing of reward anticipation. NeuroImage. 42(4): 1631–1638. Schofield PR, Williams LM, Paul RH, Gatt JM, Brown K, Luty A, Cooper N, Grieve S, Dobson-Stone C, Morris C, Kuan SA, Gordon E. (2009). Disturbances in selective information processing associated with the BDNF Val66Met polymorphism: evidence from cognition, the P300 and fronto-hippocampal systems. Biol Psychol. 80(2): 176–188. Schott BH, Seidenbecher CI, Fenker DB, Lauer CJ, Bunzeck N, Bernstein H-G, Tischmeyer W, Gundelfinger ED, Heinze H-J, Düzel E. (2006). The dopaminergic midbrain participates in human episodic memory formation: evidence from genetic imaging. J Neurosci. 26(5): 1407–1417. Schupf N, Costa R, Tang M-X, Andrews H, Tycko B, Lee JH, Mayeux R. (2004). Preservation of cognitive and functional ability as markers of longevity. Neurobiol Aging. 25(9): 1231–1240. Sebastiani P, Solovieff N, DeWan AT, Walsh KM, Puca A, Hartley SW, Melista E, Andersen S, Dworkis DA, Wilk JB, Myers RH, Steinberg MH, Montano M, Baldwin CT, Hoh J, Perls TT. (2012). Genetic signatures of exceptional longevity in humans. PLoS One. 7(1): e29848. Sédille-Mostafaie N, Sebesta C, Huber K, Zehetmayer S, Jungwirth S, Tragl K, Fischer P, Krugluger W. (2012). The role of memory-related gene polymorphisms, KIBRA and CLSTN2, on replicate memory assessment in the elderly. J Neural Trans. 119(1): 77–80. Sexton CE, Walhovd KB, Storsve AB, Tamnes CK, Westlye LT, Johansen-Berg H, Fjell AM. (2014). Accelerated changes in white matter microstructure during aging: a longitudinal diffusion tensor imaging study.” J Neurosci. 34(46): 15425–15436. Shen L, Kim S, Risacher SL, Nho K, Swaminathan S, West JD, Foroud T, Pankratz N, Moore JH, Sloan CD, MHuentelman MJ, Craig DW, DeChairo BM, Potkin SG, Jack CR, Jr, Weiner MW, Saykin AJ. (2010). Whole genome association study of brain-wide
3 5 4 part vii : I maging G enetics of C ognitive A ging
imaging phenotypes for identifying quantitative trait loci in MCI and AD: A study of the ADNI cohort. NeuroImage. 53(3): 1051–1063. Shing YL. (2011). Hippocampal subfield volumes: age, vascular risk, and correlation with associative memory. Front Aging Neurosci. 3: 2. Shiroma PR, Geda YE, Mrazek DA. (2010). Pharmacogenomic implications of variants of monoaminergic-related genes in geriatric psychiatry. Pharmacogenomics. 11(9): 1305–1330. Shumay E, Chen J, Fowler JS, Volkow ND. (2011). Genotype and ancestry modulate brain’s DAT availability in healthy humans.” PLoS One. 6(8): e22754. Silhol M, Bonnichon V, Rage F, Tapia-Arancibia L. (2005). Age-related changes in brain-derived neurotrophic factor and tyrosine kinase receptor isoforms in the hippocampus and hypothalamus in male rats. Neuroscience. 132(3): 613–624. Sklar P, Gabriel SB, McInnis MG, Bennett P, Lim YM, Tsan G. (2002). Family-based association study of 76 candidate genes in bipolar disorder: BDNF is a potential risk locus. Brain-derived neutrophic factor. Mol Psychiatry. 7: 579–593. Slifstein M, Kolachana B, Simpson EH, Tabares P, Cheng B, Duvall M. (2008). COMT genotype predicts cortical-limbic D1 receptor availability measured with [lsqb]11c[rsqb]nnc112 and pet. Mol Psychiatry. 13: 821–827. Small BJ, Rosnick CB, Fratiglioni L, Bäckman L. (2004). Apolipoprotein E and cognitive performance: a meta-analysis. Psychol Aging. 19(4): 592–600. Smith JD. (2000). Apolipoprotein E4: an allele associated with many diseases. Ann Med. 32(2): 118–127. Solis-Ortiz S, Perez-Luque E, Morado-Crespo L, Gutierrez-Munoz M. (2010). Executive functions and selective attention are favored in middle-aged healthy women carriers of the Val/Val genotype of the catechol-o-methyltransferase gene: a behavioral genetic study. Behav Brain Funct. 6(1): 67. St George-Hyslop P, et al. (1987). Search for the familial Alzheimer's disease gene. J Neural Transm Suppl. 24: 13–21. Stark AK, Toft MH, Pakkenberg H, Fabricius K, Eriksen N, Pelvig DP, Møller M, Pakkenberg B. (2007). The effect of age and gender on the volume and size distribution of neocortical neurons. Neuroscience. 150(1): 121–130. Starr JM, Fox H, Harris SE, Deary IJ, Whalley LJ. (2007). COMT genotype and cognitive ability: a longitudinal aging study. Neuroscience Letters 421(1): 57–61. Stein JL, Hua X, Morra JH, Lee S, Hibar DP, Ho AJ, Leow AD, Toga AW, Sul JH, Kang HM, Eskin E, Saykin AJ, Shen L, Foroud T, Pankratz N, … Thompson PM. (2010). Genome-wide analysis reveals novel genes influencing temporal lobe structure with relevance to neurodegeneration in Alzheimer’s disease. NeuroImage. 51(2): 542–554. Stern Y. (2009). Cognitive reserve. Neuropsychologia. 47(10): 2015–2028. Stern AJ, Savostyanova AA, Goldman A, Barnett AS, van der Veen JWC, Callicott JH, Mattay VS, Weinberger DR, Marenco S. (2008). Impact of the brain-derived neurotrophic factor Val66Met polymorphism on levels of hippocampal N-acetyl-aspartate assessed by magnetic resonance spectroscopic imaging at 3 Tesla.” Biol Psychiatry. 64(10): 856–862. Strittmatter WJ, Saunders AM, Schmechel D, Pericak-Vance M, Enghild J, Salvesen GS, Roses AD. (1993). Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proc Natl Acad Sci. 90(5): 1977–1981. Sullivan EV, Marsh L, Mathalon DH, Lim KO, Pfefferbaum A. (1995). Age-related decline in MRI volumes of temporal lobe gray matter but not hippocampus. Neurobiol Aging. 16(4): 591–606. Sullivan EV, Pfefferbaum A. (2006). Diffusion tensor imaging and aging. Neurosci Biobehav Rev. 30(6): 749–761. Sundman-Eriksson I, Allard P. (2006). Age-correlated decline in [3H] tiagabine binding to GAT-1 in human frontal cortex. Aging Clin Exp Res. 18(3): 257–260. Ta AT, Huang S-E, Chiu M-J, Hua M-S, Tseng W-YI, Chen S-HA,Qiu A. (2011). Age-related vulnerabilities along the hippocampal longitudinal axis. Human Brain Mapp. 33(10): 2415–2427.
Tan H-Y, Chen Q, Goldberg TE, Mattay VS, Meyer-Lindenberg A, Weinberger DR, Callicott JH. (2007). Catechol-O-methyltransferase Val158Met modulation of prefrontal–parietal–striatal brain systems during arithmetic and temporal transformations in working memory. J Neurosci. 27(49): 13393–13401. Terao A, Steininger TL, Morairty SR, Kilduff TS. (2004). Age-related changes in histamine receptor mRNA levels in the mouse brain. Neurosci Lett. 355(1–2): 81–84. Thambisetty M, An Y, Nalls M, Sojkova J, Swaminathan S, Zhou Y, Singleton AB, Wong DF, Ferrucci L, Saykin AJ, Resnick SM. (2013). Effect of complement CR1 on brain amyloid burden during aging and its modification by APOE genotype. Biol Psychiatry. 73(5): 422–428. Thambisetty M, Beason-Held L, An Y, Kraut MA, Resnick SM. (2010). APOE [epsiv]4 genotype and longitudinal changes in cerebral blood flow in normal aging. Arch Neurol. 67: 93–98. Thambisetty M, Beason-Held LL, An Y, Kraut M, Nalls M, Hernandez DG, Singleton AB, Zonderman AB, Ferrucci L, Lovestone S, Resnick SM. (2013). Alzheimer risk variant CLU and brain function during aging. Biol Psychiatry. 73(5): 399–405. Thambisetty M, Wan J, Carass A, An Y, Prince JL, Resnick SM. (2010). Longitudinal changes in cortical thickness associated with normal aging. NeuroImage. 52(4): 1215–1223. Timothy AS. (2009). When does age-related cognitive decline begin? Neurobiol Aging. 30(4): 507–514. Trachtenberg AJ, Filippini N, Ebmeier KP, Smith SM, Karpe F, Mackay CE. (2012). The effects of APOE on the functional architecture of the resting brain. NeuroImage 59(1): 565–572. Tschanz JT, Pfister R, Wanzek J, Corcoran C, Smith K, Tschanz BT, Steffens DC, Ostbye T, Welsh-Bohmer KA, Norton MC. (2013). Stressful life events and cognitive decline in late life: moderation by education and age: The Cache County Study. Int J Geriatr Psychiatry. 28(8): 821–830. Tunbridge EM, Harrison PJ, Weinberger DR. (2006). Catechol-omethyltransferase, cognition, and psychosis: Val158Met and beyond. Biol Psychiatry. 60(2): 141–151. Uchida H, Chow TW, Mamo DC, Kapur S, Mulsant BH, Houle S, Pollock BG, Graff-Guerrero A. (2011). Effects of aging on 5-HT2AR binding: a HRRT PET study with and without partial volume corrections. Int J Geriatr Psychiatry. 26(12): 1300–1308. Valenzuela MJ, Sachdev PS, Wen W, Shnier R, Brodaty H, Gillies D. (2000). Dual voxel proton magnetic resonance spectroscopy in the healthy elderly: subcortical-frontal axonal n-acetylaspartate levels are correlated with fluid cognitive abilities independent of structural brain changes.” NeuroImage. 12(6): 747–756. Vallabhajosula S. (2011). Positron emission tomography radiopharmaceuticals for imaging brain beta-amyloid. Sem Nucl Med. 41(4): 283–299. van den Kommer TN, Comijs HC, Aartsen MJ, Huisman M, Deeg DJ, Beekman AT. (2013). Depression and cognition: how do they interrelate in old age?” Am J Geriatr Psychiatry. 21(4): 398–410. van Dyck CH, Malison RT, Jacobsen LK, Seibyl JP, Staley JK, Laruelle M, Baldwin RM, Innis RB, Gelernter J. (2005). Increased dopamine transporter availability associated with the 9-repeat allele of the SLC6A3 gene. J Nucl Med. 46(5): 745–751. van Dyck CH, Malison RT, Seibyl JP, Laruelle M, Klumpp H, Zoghbi SS, Baldwin RM, Innis RB. (2000). Age-related decline in central serotonin transporter availability with [123I]β-CIT SPECT. Neurobiol Aging. 21(4): 497–501. Van Laere K, Goffin K, Casteels C, Dupont P, Mortelmans L, de Hoon J, Bormans G. (2008). Gender-dependent increases with healthy aging of the human cerebral cannabinoid-type 1 receptor binding using [(18)F]MK-9470 PET. NeuroImage. 39(4): 1533–1541. van Waarde A, Ramakrishnan NK, Rybczynska AA, Elsinga PH, Ishiwata K, Nijholt IM, Luiten PGM, Dierckx RA. (2011). The cholinergic system, sigma-1 receptors and cognition. Behav Brain Res. 221(2): 543–554.
chapter 2 1 : N eurobiology and G enetics of C ognitive A ging
355
van Wingen G, Rijpkema M, Franke B, van Eijndhoven P, Tendolkar I, Jan Verkes R, Buitelaar J, Fernández G. (2010). The brain-derived neurotrophic factor Val66Met polymorphism affects memory formation and retrieval of biologically salient stimuli. NeuroImage. 50(3): 1212–1218. Vandenbergh DJ, Persico AM, Uhl GR. (1992). Human dopamine transporter gene (DAT1) maps to chromosome 5p15.3 and displays a VNTR. Genomics. 14: 1104–1106. VanNess S, Owens M, Kilts C. (2005). The variable number of tandem repeats element in dat1 regulates in vitro dopamine transporter density. BMC Genetics. 6: 55. Vassos E, Bramon E, Picchioni M, Walshe M, Filbey FM, Kravariti E. (2010). Evidence of association of KIBRA genotype with episodic memory in families of psychotic patients and controls. J Psychiatr Res. 44: 795–798. Venkatraman V, Aizenstein HJ, Newman AB, Yaffe K, Harris T, Kritchevsky S, Ayonayon H. Rosano C. (2011). Lower digit symbol substitution score in the oldest old is related to magnetization transfer and diffusion tensor imaging of the white matter. Front Aging Neurosci. 3:11. Ventriglia M. (2002). Association between the BDNF 196 A/G polymorphism and sporadic Alzheimer’s disease. Mol Psychiatry. 7: 136–137. Verhaeghen P, Cerella J, Basak C. (2006). Aging, task complexity, and efficiency modes: the influence of working memory involvement on age differences in response times for verbal and visuospatial tasks. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn. 13(2): 254–280. Volkow ND. (1994). Decreased dopamine transporters with age in health human subjects. Ann Neurol. 36: 237–239. Walhovd KB, Westlye LT, Amlien I, Espeseth T, Reinvang I, Raz N, Agartz I, Salat DH, Greve DN, Fischl B, Dale AM, Fjell AM. (2011). Consistent neuroanatomical age-related volume differences across multiple samples. Neurobiol Aging. 32(5): 916–932. Wang M, Gamo NJ, Yang Y, Jin LE, Wang X-J, Laubach M, Mazer JA, Lee D, Arnsten AFT. (2011). Neuronal basis of age-related working memory decline. Nature. 476(7359): 210–213. Webster MJ, Herman MM, Kleinman JE, Shannon Weickert C. (2006). BDNF and trkB mRNA expression in the hippocampus and temporal cortex during the human lifespan. Gene Express Patt. 6(8): 941–951. Webster MJ, Weickert CS, Herman MM, Kleinman JE. (2002). BDNF mRNA expression during postnatal development, maturation and aging of the human prefrontal cortex. Brain Res Dev Brain Res. 139: 139–150. Weinberger DR, Egan MF, Bertolino A, Callicott JH, Mattay VS, Lipska BK, Berman KF, Goldberg TE. (2001). Prefrontal neurons and the genetics of schizophrenia. Biol Psychiatry. 50(11): 825–844. Wersching H, Guske K, Hasenkamp S, Hagedorn C, Schiwek S, Jansen S, Witte V, Wellmann J, Lohmann H, Duning K, Kremerskothen J, Knecht S, Brand E, Floel A. (2011). Impact of common KIBRA allele on human cognitive functions. Neuropsychopharmacology. 36(6): 1296–1304. Westlye ET, Lundervold A, Rootwelt H, Lundervold AJ, Westlye LT. (2011). Increased hippocampal default mode synchronization during rest in middle-aged and elderly APOE ε4 carriers: relationships with memory performance.” J Neurosci. 31(21): 7775–7783. Westlye LT, Walhovd KB, Dale AM, Bjørnerud A, Due-Tønnessen P, Engvig A, Grydeland H, Tamnes CK, Østby Y, Fjell AM. (2010). Differentiating maturational and aging-related changes of the cerebral cortex by use of thickness and signal intensity. NeuroImage. 52(1): 172–185.
Westlye LT,Walhovd KB, Dale AM, Espeseth T, Reinvang I, Raz N, Agartz I, Greve DN, Fischl B, Fjell AM. (2009). Increased sensitivity to effects of normal aging and Alzheimer’s disease on cortical thickness by adjustment for local variability in gray/white contrast: a multi-sample MRI study. NeuroImage. 47(4): 1545–1557. Whalley LJ, Deary IJ, Appleton CL, Starr JM. (2004). Cognitive reserve and the neurobiology of cognitive aging. Ageing Res Rev. 3(4): 369–382. Willeit M, Praschak-Rieder N. (2010). Imaging the effects of genetic polymorphisms on radioligand binding in the living human brain: a review on genetic neuroreceptor imaging of monoaminergic systems in psychiatry. NeuroImage. 53(3): 878–892. Wisdom NM, Callahan JL,Hawkins KA. (2011). The effects of apolipoprotein E on non-impaired cognitive functioning: a meta-analysis. Neurobiol Aging 32(1): 63–74. Wishart HA, Roth RM, Saykin AJ, Rhodes CH, Tsongalis GJ, Pattin KA, Moore JH, McAllister TW. (2011). COMT Val158Met genotype and individual differences in executive function in healthy adults. J Int Neuropsychol Soc. 17(01): 174–180. Wishart HA, Saykin AJ, Rabin LA, Santulli RB, Flashman LA, Guerin SJ. (2006). Increased brain activation during working memory in cognitively intact adults with the APOE epsilon4 allele. AmJ Psychiatry. 163: 1603–1610. Wu WW, Chan CS, Disterhoft JF. (2004). Slow afterhyperpolarization governs the development of NMDA receptor–dependent afterdepolarization in ca1 pyramidal neurons during synaptic stimulation. J Neurophysiol. 92(4): 2346–2356. Wu WW, Chan CS, Disterhoft JF (2004). Slow afterhyperpolarization governs the development of nmda receptor-dependent afterdepolarization in CA1 pyramidal neurons during synaptic stimulation. J Neurophysiol. 92(4): 2346–2356. Yamamoto M, Suhara T, Okubo Y, Ichimiya T, Sudo Y, Inoue M, Takano A, Yasuno F, Yoshikawa K, Tanada S. (2002). Age-related decline of serotonin transporters in living human brain of healthy males. Life Sciences. 71(7): 751–757. Yanai K, Watanabe T, Meguro K, Yokoyama H, Sato I, Sasano H, Itoh M, Iwata R, Takahashi T. Ido T. (1992). Age-dependent decrease in histamine H1 receptor in human brains revealed by PET. Neuroreport. 3(5): 433–436. Yassa MA, Muftuler LT, Stark CE. (2010). Ultrahigh-resolution microstructural diffusion tensor imaging reveals perforant path degradation in aged humans in vivo. Proc Natl Acad Sci U S A. 107(28): 12687–12691. Zahr NM, Mayer D, Pfefferbaum A, Sullivan EV. (2008). Low striatal glutamate levels underlie cognitive decline in the elderly: evidence from in vivo molecular spectroscopy. Cereb Cortex. 18(10): 2241–2250. Ziegler DA, Piguet O, Salat DH, Prince K, Connally E, Corkin S. (2010). Cognition in healthy aging is related to regional white matter integrity, but not cortical thickness. Neurobiol Aging. 31(11): 1912–1926. Zimmerman ME, Pan JW, Hetherington HP, Katz MJ, Verghese J, Buschke H. (2008). Hippocampal neurochemistry, neuromorphometry, and verbal memory in nondemented older adults. Neurology. 70: 1594–1600. Zinkstok J, Schmitz N, van Amelsvoort T, de Win M, van den Brink W, Baas F, Linszen D. (2006). The COMT val158met polymorphism and brain morphometry in healthy young adults. Neurosci Lett. 405(1–2): 34–39.
3 5 6 part vii : I maging G enetics of C ognitive A ging
22. NEUROIMAGING GENETICS OF ALZHEIMER’S DISEASE Theresa M. Harrison, Alison C. Burggren, and Susan Y. Bookheimer
INTRODUCTI O N When the human genome project was completed in 2003, many believed it marked the beginning of an era of genomic medicine. The genetic basis of highly heritable neurological disorders like Alzheimer’s disease (AD), autism, and schizophrenia would be discovered and applied to the development of new therapies. Some went one step further, predicting that personalized genomic medicine would soon follow the elucidation of disease-specific genetic fingerprints. These predictions and hopes have yet to come to fruition, and in their stead an appreciation for the complexity of polygenic brain disorders has been steadily growing. Current estimates indicate that both autism and schizophrenia are each associated with hundreds of genes and even more single nucleotide polymorphisms (SNPs) (SFARI Gene Database; Ripke et al. 2014). Similarly, the genetics of AD have proven to be a complex problem despite the early promise of the apolipoprotein E (APOE) gene. More than 20 years ago, Corder and colleagues (1993) discovered that a single copy of the ε4 allele of APOE increased an individual’s risk of getting AD 4- to 5-fold, and two copies increased AD risk up to 15-fold. Another way to measure the risk conferred by APOE is to examine what proportion of AD heritability can be accounted for by APOE. Twin studies reveal that the heritability of AD is 60%–80% and that APOE accounts for about 50% of the variation in heritability (Bergem, Engedal, and Kringlen 1997; Gatz et al. 2006; Waring and Rosenberg 2008). Current understanding of human genetics allows one to appreciate that APOE has a relatively huge effect on AD risk. Still, roughly half of the variation in heritability is presumably due to other genetic risk factors, of which there are now greater than 20 validated candidates (Lambert et al. 2013).
The increasing number of risk loci associated with AD and other neurological disorders is the result of a recent surge in the efficiency of genomic technologies combined with data-sharing efforts that have allowed researchers to identify genomic loci associated with disease, even with very low effect sizes. Low effect size associations require extremely large cohorts in order to provide sufficient power for detection. Recently, a consortium of AD researchers created a large data set of ~70,000 subjects known as the International Genomics of Alzheimer’s Project (IGAP) (http://www.wikigenes.org/e/art/e/258. html). Analyses performed on this large data set were able to reveal 11 new AD risk loci, in addition to confirming previously identified loci (Lambert et al. 2013). Based on this success and the success of other consortiums, there continues to be interest in ever larger cohorts. However, as sample sizes continue to rise, “significant” effect sizes get smaller and smaller, and there may be a risk of being statistically over-powered. Too much statistical power could lead to the identification of spurious risk loci that are not actually associated with the disease phenotype. The process of parsing out true and spurious associations will certainly be a topic of research in the coming years. Today, the major challenge is understanding the clinical and pathological roles played by each of the AD risk genes and their products. Neuroimaging has emerged as one popular method for characterizing genetic AD risk factors in humans. Progress on this front will be reviewed in subsequent sections. Neuroimaging methods, including magnetic resonance imaging (MRI) and positron emission tomography (PET) imaging, are often used to study AD in humans. Recent advances in MRI hardware and pulse sequence development have enabled characterization of brain structure and function at improved spatial and temporal resolutions with higher signal-to-noise ratio and better
357
tissue-type contrast. Structural MRI (sMRI) is now able to reliably delineate subregions of the hippocampus, while functional MRI (fMRI) has reached spatial resolution high enough to resolve functional units in visual cortex (Wisse et al. 2014; Yacoub, Harel, and Ugurbil 2008). Improved tissue contrast in sMRI allows for consistent delineation of white and gray matter, making cortical thickness and volumes estimates more robust (Fujimoto et al. 2014). fMRI sequences are utilizing fast, multiband acquisition techniques that minimize signal dropout without sacrificing signal-to-noise ratios (Boyacioğlu et al. 2014). In addition, diffusion weighted imaging (DWI) acquisition and analysis are constantly evolving and improving. Gradient strengths and the number of directions in which pulses are applied are increasing as equipment improves, allowing for better estimation of water diffusion at every voxel (Alexander et al. 2007). In addition, spatial resolution continues to increase, with recent studies boasting sub-millimeter in-plane resolution (Shi et al. 2014). In PET imaging, the development of 18 F radiotracers, such as florbetapir, has made it easier for more sites to acquire amyloid deposition data (a measure of AD pathology in the brain) than before, when only the more volatile 11C tracers were available (Wong et al. 2010). In addition, there are now preliminary data on tau-specific tracers, including one 18F tracer called T807, which, when analyzed alongside Aβ-specific tracers, will help to elucidate the temporal dynamics of Aβ and tau deposition in AD pathogenesis (Xia et al. 2013). Together, these advances in neuroimaging will allow for more sensitive and accurate biomarker detection and longitudinal monitoring, both of which are important to help identify individuals who are at increased risk for AD. This chapter focuses on AD, the genetic basis of the disease and how genetic markers have been studied in combination with neuroimaging methods to help elucidate the effects of genetic risk for AD on the structure and function of the brain. After a brief review of the clinical and pathological features of AD, we begin by discussing the rare autosomal dominant forms of AD that result from a mutation in one of three genes: APP, PSEN1, or PSEN2. Next, we cover how neuroimaging has helped to shed light on the still-murky relationship between AD and Down syndrome, a condition in which individuals have a third copy of chromosome 21 and, therefore, an extra copy of APP. The next portion of the chapter focuses on less penetrant, statistical genetic risk factors. Sections are organized to reflect the strength of the risk conferred by each locus, beginning with the discovery of the APOE ε4 risk allele in the early 1990s, followed by the recent discoveries of strong AD associations with TOMM40 and TREM2
variants, and finally the identification of AD-associated loci through genome-wide association (GWA) studies. We will also cover the growing literature linking the neurotrophic factor BDNF with AD. Next, genetic risk factors identified through studies of AD-related neuroimaging biomarkers and endophenotypes will be discussed. In each section, relevant work in structural and functional imaging will be reviewed. In addition, challenges in the field of neuroimaging genetics of AD will be discussed, alongside possible future directions for the field. Finally, the relevance and impact of neuroimaging genetics research in the fight against AD will be examined. PAT H OLOGIC AL AND C LINIC AL F E AT U R E S OF ALZ H E IME R ’ S DIS E AS E More than 100 years ago, at a meeting of psychiatrists in Germany, Alois Alzheimer presented a case study of a woman who had suffered from progressive memory loss and other cognitive and psychiatric symptoms (Alzheimer et al. 1995; Goedert and Spillantini 2006). The most enduring aspect of his presentation was his description of abnormal deposits that he discovered after silver staining the patient’s brain tissue. These deposits, called plaques and tangles based on their morphology under the microscope, remain the defining neuropathological features of AD today. At the molecular level there are two main proteins that accumulate abnormally in AD: beta-amyloid (Aβ) and tau. Soluble Aβ oligomers collect to form extracellular neuritic plaques, while hyperphosphorylated tau proteins form intracellular inclusions called neurofibrillary tangles. The gene that transcribes Aβ, as well as two genes that transcribe enzymes involved with regulating Aβ isoforms, are each the site of many mutations that give rise to dominantly inherited, familial AD. In addition, the link between Down syndrome and AD, which will be described in a subsequent section, can be traced back to Aβ. The connection is based on the fact that Aβ is transcribed from a gene on chromosome 21, the chromosome that is in triplicate in Down syndrome. It is believed that the third copy of this gene results in Aβ overexpression, which may be the cause of the near 100% incidence of AD in adults with Down syndrome. All of this evidence helped lead to the belief that Aβ is the primary pathology in AD and to the “amyloid cascade hypothesis” (Hardy and Higgins 1992; Hardy and Selkoe 2002). This hypothesis states, generally, that the ineffective clearance of Aβ leads to the deposition of plaques, and that this is the first in a cascade of molecular events that eventually cause neuronal death and, in some cases, vascular damage.
3 5 8 part vii : I maging G enetics of C ognitive A ging
However, it has been shown that Aβ pathology does not correlate with clinical symptoms of AD, while tau pathology does (Bennett et al. 2004). The formation of neurofibrillary tangles results from the polymerization of hyperphosphorylated tau. Under normal circumstances, tau is a major component of neuronal cytoskeleton, but when tau becomes hyperphosphorylated it tends to accumulate into neurofibrillary tangles. These tangles are associated with the marked loss of synapses in AD, which is putatively caused, at least in part, by a breakdown in cytoskeletal maintenance (Pooler, Noble, and Hanger 2014). These findings have led to increased attention on tau and a convincing counterargument to the amyloid cascade hypothesis. Now it appears that other proteins may also play a role in AD, which may be further evidence that Aβ is not a trigger, but just one of several pathologies in AD. Very recent work has shown that two proteins known to play a role in other neurodegenerative diseases are, in fact, also associated with AD. For example, TDP-43 inclusions, a major feature of frontotemporal lobar degeneration (FTLD), have been shown to occur at a higher rate in those with AD compared to healthy controls ( Josephs et al. 2014). In addition, in mice it has been shown that decreased progranulin, another protein associated with FTLD, encourages Aβ deposition (Minami et al. 2014). Finally, the amyloid cascade hypothesis is also called into question based on data from PET imaging studies. PET scans acquired with Aβ specific tracers have revealed that individuals can be positive for Aβ in the brain and have no clinical symptoms. This state can persist for years. So while it is undeniable that Aβ pathology is necessary for a diagnosis of AD, it remains unclear whether it is sufficient to initiate the “cascade” of events that leads to full-blown AD, or whether it is just one of many factors that together cause AD. The emerging neuropathological picture of AD is complex. Hopefully, the development of additional, specific PET tracers will resolve some of the complexity by increasing our understanding of the temporal and spatial dynamics of each type of proteinaceous inclusion associated with AD. Clinically, dementia is defined as the loss of cognitive ability that interferes with activities of daily living. Dementia can be caused by many conditions. Differential diagnosis, based on clinical symptoms, neuroimaging biomarkers, and other criteria, is necessary in order to identify the cause of a dementia syndrome. AD is the most frequent cause of dementia, accounting for 60%–80% of cases (Fratiglioni et al. 1999; Mendez and Cummings 2004). The next most common cause is vascular dementia, which accounts for about 10% of cases (Mendez and Cummings 2004). It should be noted, however, that in about half of
AD cases, there is also concomitant vascular pathology discovered at autopsy (Fernando and Ince 2004). The most common presentation for AD involves a slow and steady decrease in episodic memory function, as well as dysfunction in at least one other cognitive domain. In 2011, the diagnostic criteria for AD got a much-needed update. The National Institute on Aging worked alongside the Alzheimer’s Association to form a working group of experts who reviewed the then-current criteria, first published in 1984 (McKhann et al. 1984). The 1984 guidelines were primarily based on clinical presentation and focused on memory impairment. This meant that in order to receive a diagnosis of AD, clinical symptoms needed to already be present and interfering with daily activities (Mendez and Cummings 2004). However, research over the past 20 years has shown beyond a doubt that AD pathogenesis is a process that begins long before the emergence of cognitive impairment. This early, pre-symptomatic stage is estimated to last as long as 15–20 years in some patients (Morris 2005). In order to better account for the long and slow progression of AD during the preclinical phase, the new criteria describe three stages of AD: preclinical AD, mild cognitive impairment (MCI) due to AD, and dementia due to AD (Sperling et al. 2011). Another change to the criteria incorporates biomarker data in the process of diagnosis (McKhann et al. 2011). Possible biomarkers for AD include levels of Aβ and tau analytes in cerebrospinal fluid (CSF), deposition of Aβ in the brain detected with PET imaging, and hippocampal atrophy measured by MRI. It has been shown that decreased levels of Aβ in the CSF, increased levels of phosophorylated tau in the CSF, Aβ positivity as measured with PET tracers (cutoff ratios used to define positive or negative designations are still an active area of research), and hippocampal thickness or volume loss are all indications of possible AD. More work is needed before biomarker testing can be used to conclusively diagnose AD. The continued collection and development of these biomarker measurements are especially important to efforts to identify individuals who are in the early stages of AD, before clinical symptoms emerge. These are the individuals who would benefit most from any available interventions or therapies. In addition, biomarkers provide critical benchmarks for monitoring the success of experimental treatments. A key concept in AD pathophysiology is that clinical symptoms emerge only after there has been a long period of degeneration, resulting in substantial neuronal loss ( Jack et al. 2010). Indeed, this is true of all age-related neurodegenerative disorders. The superstructure of neuronal circuitry is complex and the result of both genetic and environmental effects compounded over one’s lifetime.
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
359
Thus, the prospect of regeneration after extensive cell death seems more like science fiction than a reasonable therapeutic goal. Therefore, investigators are increasingly focusing on the phase of AD progression that occurs before clinical symptoms can be detected. This phase, called preclinical or pre-symptomatic AD, is marked by progressive neuronal loss in the brain, especially the hippocampus, and by Aβ accumulation. It has been posited that some AD treatment trials have failed to show a positive effect because the drugs were given too late in the course of the disease (Sperling et al. 2011). Preclinical AD is a hypothetical state (subjects may die before progressing to clinical AD) during which intervention aimed at halting disease progression may be most effective. Since its proposed addition to the AD diagnostic framework in 2011, many studies have attempted to focus on preclinical AD, which is still unfortunately difficult to define in the absence of longitudinal data. There are, however, exceptions to this definition problem. For example, the preclinical AD phase can be reliably identified in subjects who carry genetic mutations that cause dominantly inherited forms of AD. Therefore, families with these mutations are extremely valuable to the AD research community. NEUROIM AGI N G HI G HLY PEN ETRAN T GENETIC C A USES O F ALZHEI MER’S DIS EA S E In the introduction or background sections of most papers written about AD, the authors will list statistics detailing the incidence and prevalence of the disease. If the authors are interested in genetics, they may also cite studies that describe the heritability of AD and lifetime risk given certain genetic risk factors, including family history of the disease. In these cases, the authors are invariably referring to sporadic, late-onset AD, which is, by far, the most common form of the disease. There are, however, people who develop AD as a result of an identifiable, underlying genetic cause. These genetic causes fall into two categories: familial AD and Down syndrome. The following sections will cover these highly penetrant forms of AD and how neuroimaging genetics findings in these unique cohorts can inform the study of AD in general. For a summary of the literature reviewed in this portion of the chapter see Table 22.1. FA M I L I A L A D
Early-onset AD is the clinical manifestation of the disease before the age of 65. In the vast majority of cases, early-onset AD is caused by a rare, autosomal dominant form of the
disease, characterized by a mutation in one of three genes. Together, patients who have one of these mutations are diagnosed with what is called familial AD (FAD). The three genes with known mutations that cause FAD are amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2) (Goate et al 1991; Schellenberg et al. 1992; Rogaev et al. 1995). More than 50 specific deleterious mutations in APP have been discovered, but these only account for less than 10% of early-onset AD cases. In contrast, more than 175 specific deleterious mutations in PSEN1 have been identified, and these account for up to 70% of FAD cases. Less than 5% of cases are caused by PSEN2 mutations, of which just over a dozen have been identified (Cruts, Theuns, and Van Broeckhoven 2012). Average age of onset varies across the three genes, with PSEN1 mutations leading to earlier onsets (~42 years old), followed by APP (~52 years old) and PSEN2 (~57 years old). Since the discovery of these highly penetrant AD genes over 20 years ago, studies of their function, both in normal and mutated conditions, have provided strong support to the so-called “amyloid cascade hypothesis” (Hardy and Selkoe 2002). APP, PSEN1, and PSEN2 functionally converge on the production of Aβ, the peptide that aggregates to form extracellular amyloid plaques, one of the two primary neuropathological features of AD. Specifically, APP encodes the protein precursor of the pathogenic Aβ oligomer. PSEN1 and PSEN2 encode peptides that are components of secretase complexes, enzymes that modify proteins at specific cleavage sites. The protein presenilin 1 is a proteolytic subunit of a complex called gamma-secretase, which is perhaps best known for its role in cleaving the amyloid precursor protein, the protein product of the APP gene. Mutations in PSEN1 result in an overproduction of the pathogenic Aβ peptide. Mutations in PSEN2 cause a very similar effect, altering presenilin 2 such that gamma-secretase activity is disrupted, leading to over-production of pathogenic Aβ. Interestingly, recent work has indicated that there may also be mutations in these genes that decrease risk for AD. In a recent study led by Kari Stefansson and colleagues, a unique Swedish population was found to harbor a protective mutation in APP that actually decreased risk for AD in carriers ( Jonsson et al. 2012). The exact mechanism of this protection remains unknown but, presumably, the mutation causes a decrease either in amyloid precursor protein levels or, more downstream, results in a decrease of the amyloidogenic oligomers. A clinical feature of FAD that is useful in research is that specific mutations are associated with a relatively precise age of onset of disease. Because of this, researchers can stage the preclinical phase of a mutation carrier based on
3 6 0 part vii : I maging G enetics of C ognitive A ging
TABLE 22.1
GENETIC CAUSES OF AND RISK GENES FOR AD: NEUROIMAGING MODALITIES IN THE LITERATURE AND REPRESENTATIVE REFERENCES
Genetic Anomaly
Gene
First Associated with AD
sMRI
DWI
t-fMRI
rs-fMRI
PET
Point Mutation
APP
Goate et al. 1991
Schott et al. 2003; Ridha et al. 2006; Knight et al. 2011; Apostolova et al. 2011; Bateman et al. 2012; Cash et al. 2013; Lee et al. 2013; Scahill et al. 2013; Thordardottir et al. 2014
Ringman et al. 2007
Braskie et al. 2012; Braskie et al. 2013
Chhatwal et al. 2013; Thomas et al. 2014
Villemagne et al. 2009; Bateman et al. 2012;
PSEN1
Schellenberg et al. 1992
Schott et al. 2003; Ridha et al. 2006; Knight et al. 2011; Apostolova et al. 2011; Bateman et al. 2012; Cash et al. 2013; Lee et al. 2013; Quiroz et al. 2013; Ryan et al. 2013; Scahill et al. 2013; Thordardottir et al. 2014
Ringman et al. 2007; Ryan et al. 2013
Quiroz et al. 2010; Braskie et al. 2012; Braskie et al. 2013; Sala-Llonch et al. 2013
Chhatwal et al. 2013; Sala-Llonch et al. 2013; Thomas et al. 2014
Koivunen et al. 2008; Villemagne et al. 2009; Bateman et al. 2012
PSEN2
Rogaev et al. 1995
Bateman et al. 2012; Cash et al. 2013
Chhatwal et al. 2013; Thomas et al. 2014
Bateman et al. 2012
Duplication
APP
Sleegers et al. 2006
Trisomy 21
APP
Jervis 1948
Kesslak et al 1994; Krasuski et al. 2002; Teipel 2003; White et al. 2003; Teipel et al. 2004; Beacher et al. 2010
Powell et al. 2014
Poly-T repeat
TOMM40
Roses et al. 2010
Johnson et al. 2011; Ferencz et al. 2013
Lyall et al. 2014
Single Nucleotide Polymor phisms
APOE
Corder et al. 1993
Jak et al. 2007; Shaw et al. 2007; Burggren et al. 2008; Cherbuin et al. 2008; Espeseth et al. 2008; Mueller et al. 2008; Donix et al. 2010; Liu et al. 2010; Lu et al. 2011; Pievani et al. 2011; Richter-Schmidinger et al. 2011; Spampinato et al. 2011; Alexander et al. 2012; O’Dwyer et al. 2012; Samuraki et al. 2012; Hostage et al. 2013; Khan et al. 2014; Dean et al. 2014; Knickmeyer et al. 2014; Manning et al. 2014; Matura et al. 2014; Taylor et al. 2014;
Brown et al. 2011; Heise et al. 2011; Westlye et al. 2012; Westlye, Hodneland, et al. 2012; Kljajevicet al. 2014; Matura et al. 2014
Petrini et al. 1997; Landt et al. 2011; Sabbagh et al. 2011 Liu et al. 2014 Bookheimer et al. 2000; Bondi et al. 2005; Borghesani et al. 2008; Filipinni et al. 2009; Xu et al. 2009; Suthana et al. 2010; Adamson et al. 2011; Green et al. 2014;
Filipinni et al. 2009; Sheline et al. 2010; Machulda et al. 2011; Westlye et al 2011; Damoiseaux et al. 2012; Trachtenberg et al. 2012; Heise et al. 2014;
Reiman et al. 1996; Mosconi et al. 2004; Reiman et al. 2004; Drzezga et al. 2005; Pike et al. 2007; Morris et al. 2010; Fleisher et al. 2011; Fleisher et al. 2013; Ossenkoppele et al. 2013; Knopman et al. 2014; Lehmann et al. 2014; Lowe et al. 2014; Sheinin et al. 2014; (continued)
TABLE 22.1
Genetic Anomaly
(CONTINUED) Gene
First Associated with AD
sMRI
DWI
t-fMRI
rs-fMRI
Braskie et al. 2011
Erk et al. 2011; Green et al. 2014
Zhang et al. 2014
TREM2
Jonsson et al. 2013; Guerreiro et al. 2013
Rajagopalan et al. 2013; Luis et al. 2014
CLU
May et al. 1990(Harold et al. 2009; Lambert et al. 2009)
Bralten et al. 2011; Stevens et al. 2014
PICALM
Harold et al. 2009
Biffi et al. 2010; Bralten et al. 2011; Furney et al. 2011;
CR1
Melchoir et al. 2010
Biffi et al. 2010; Bralten et al. 2011
BIN1
Jun et al. 2010
Biffi et al. 2010
ABCA7
Hollingworth et al. 2011
Hughes et al. 2014
EphA1
Hollingworth et al. 2011; Naj et al. 2011
Hughes et al. 2014
CD33
Tebar et al. 1999; Morgan and Carrasquillo 2013
Bradshaw et al. 2013
Zhang et al. 2014
A mark indicates that there is published work exploring the relationship of a given genetic mutation or risk factor and a given neuroimaging modality. Citations of studies referred to in the text are given.
PET
Hohman et al. 2013
Hohman et al. 2013
Another group of investigators, led by Bradford Dickerson, used Freesurfer, a computational neuroanatomy software suite, to test for differences between preclinical PSEN1 mutation carriers and non-carriers in so-called AD-signature regions of cortex. These regions were based on previous studies comparing the cortical thickness of sporadic, late-onset AD patients and controls (Dickerson et al. 2009) (Figure 22.1A). In other words, the AD-signature regions represent cortical regions that are particularly vulnerable to atrophy in AD. In the PSEN1 carriers, they found that AD-signature regions as a whole were thinner (a)
e
f
d
h
i
g
a
c
b
*
*
*
S PR G EC UN
3.5
L
4
Controls PSEN1 Carriers
*
3 2.5 2 1.5 1 0.5
ig -S
PV C
AD
FG M
SF G
SP
AG
TP
M
TL
0 ITG
(b) Mean thickness in the AD-signature ROIs (mm)
the age of onset of a parent or family member who carried the same mutation (Lopera 1997). Thus, biomarker data from carriers of different mutations can be pooled according to preclinical stage, represented by years to expected onset. Pooling data across specific rare mutation types is essential in order to assemble large cohorts for research. The Dominantly Inherited Alzheimer Network (DIAN) is a worldwide network of FAD research centers based out of Washington University in St. Louis, which has spearheaded much of the relatively large cohort neuroimaging research in FAD mutation carriers (DIAN webpage, http://dian-info. org/default.htm). sMRI-based measurements indicate that the rate of change of hippocampal volume is higher in FAD mutation carriers than in age-matched non-carriers (Schott et al. 2003; Ridha et al. 2006). Hippocampal thinning or shrinkage is a feature of AD, but it is a tricky biomarker candidate because hippocampal volume loss is also associated with normal aging. It is now believed that it is the rate of that loss that is important, with slow changes indicating normal aging and a faster trajectory indicating AD. A significant difference between FAD mutation carriers and non-carriers in the rate of change of hippocampal volume loss is evident ~2–5 years before the expected onset of disease (Schott et al. 2003; Ridha et al. 2006; Bateman et al. 2012). Measuring rate of change, though, requires longitudinal data, which is not ideal for diagnostic use in a clinical setting. Cross-sectionally, it has been shown that FAD mutation carriers have decreased hippocampal volume bilaterally compared to non-mutation-carrier controls up to 15 years before the expected onset of disease (Bateman et al. 2012). This is a potentially very early biomarker, but it is difficult to assess on an individual basis since size and shape and, thus, volume of hippocampi vary in healthy populations as well as disease populations (Ringman, Pope, and Salamon 2010). In addition to hippocampal changes, atrophy in the cortex in FAD mutation carriers has been studied using sMRI. One study, completed on preclinical FAD mutation carriers and non-carriers in a Swedish cohort, found that the mutation carriers had decreased gray matter volume in the left precuneus, superior temporal gyrus, and fusiform gyrus (Thordardottir et al. 2014). Another large study showed that there was a significant difference in gray matter volumes between mildly symptomatic (Clinical Dementia Rating scale = 0.5) carriers and healthy non-carriers in the thalamus and putamen, as well as in cortical regions, including the temporal lobe, precuneus, and the cingulate gyrus (Cash et al. 2013). The authors observed the same differences, with a greater magnitude, in moderate to severely symptomatic carriers.
Figure 22.1 Comparison of presenilin 1 (PSEN1) FAD mutation carriers and
non-carriers in cortical thickness from a priori ROIs that compose the “AD-signature” regions. Figure reproduced from Quiroz YT, et al. (2013), Journal of Neurology, Neurosurgery and Psychiatry. A. AD-signature regions: (a) Medial temporal lobe (MTL), (b) inferior temporal gyrus (ITG), (c) temporal pole (TP), (d) angular gyrus (AG), (e) superior frontal gyrus (SFG), (f) superior parietal lobule (SPL), (g) supramarginal gyrus (SG), (h) precuneus (Precun), (i) medial frontal gyrus (MFG), primary visual cortex (PVC). B. Bar graphs show mean cortical thickness within each ROI in the PSEN1 mutation carriers and non-carriers, averaged across hemispheres (p < 0.005). AD-signature (AD-sig) regions combined cortical thickness is also compared. Error bars show 1 SE of the mean. Reprinted with permission from BMJ Publishing Group Ltd.
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
363
when compared to non-carriers (Quiroz et al. 2013). Upon further analysis, the authors found that differences in the angular gyrus, superior parietal lobule, and precuneus were driving this effect (Figure 22.1B). This is in agreement with other evidence that the precuneus is one of the earliest cortical regions to begin to atrophy in FAD mutation carriers (Knight et al. 2011). Another study that used tensor-based morphometry found significant differences in cortical regions only when demented FAD subjects were compared to non-carriers (Lee et al. 2013). However, the lack of differences between pre-symptomatic carriers and non-carriers might be related to low sample size, as other studies of relatively small cohorts also were unable to detect differences (Apostolova et al. 2011). Interestingly, there is some evidence to indicate that specific FAD mutations, especially in different genes, may have distinct effects on atrophy patterns and rates across the cortex and in the hippocampus (Scahill et al. 2013). Other studies have found no differences between mutation gene types when tested explicitly (Bateman et al. 2012). Still, one must consider the validity of combining individuals from families with different mutations in a single cohort based on preclinical stage. This approach certainly increases sample sizes and statistical power, which is likely a worthwhile trade-off of combining multiple genetic mutation types in carrier groups. There has been relatively little work published examining white matter structure in FAD mutation carriers. One study showed that in carriers, white matter volume is decreased in areas including the fornix and the cingulum, two important fiber bundles connecting the hippocampus and limbic areas (Cash et al. 2013). Another study examining white matter using tensor modeling found that symptomatic mutation carriers showed increased diffusivity (especially radial diffusivity) and reduced fractional anisotropy (FA) in the fornix, cingulum, and the corpus callosum (Ryan et al. 2013). The reduced FA phenotype in preclinical FAD mutation carriers has also been reported elsewhere (Ringman et al. 2007). FA is an index that ranges from 0 to 1 that indicates, in a given voxel, the preference of water molecules to diffuse along the principal axis of diffusion. Reduced FA in a given region may indicate a breakdown of white matter, often referred to as decreased white matter integrity. Reduced FA is often, but not always, accompanied by increased diffusivity, an inversely related measure of how freely water molecules diffuse in a given voxel. FA and diffusivity are common metrics calculated based on tensor modeling of DWI data, and will be referred to in subsequent sections. It has been posited that functional changes in the brain measured by fMRI might be candidates for early AD biomarkers (Sperling 2011). Task-based and resting fMRI have
both been used to examine possible differences between FAD mutation carriers and non-carriers. Several task-based studies found that FAD mutation carriers show reduced BOLD activity compared to non-carriers in regions normally associated with the task (Braskie et al. 2013; Sala-Llonch et al. 2013). Specifically, in one study, there was reduced activity in the hippocampus, inferior parietal cortex, precuneus, and posterior middle temporal gyrus during the retrieval phase of a memory task (Braskie et al. 2013). The authors also looked at behavior, noting that in no scenario did higher activity in mutation carriers correlate with better performance. This is important because, as will be discussed in subsequent sections, the fMRI literature in genetic risk for sporadic, late-onset AD is contradictory, which leads to contradictory interpretations of results. One possible interpretation of higher activity in high genetic risk groups is that it is a compensatory mechanism and a sign of early disease (Bookheimer et al. 2000). However, if there is no correlation between higher activity and behavioral performance, this interpretation is difficult to support with evidence. Another study in preclinical FAD mutation carriers found that activity in carriers increased as a function of preclinical stage (in other words, there was an inverse relationship between activity and years to expected onset) in the middle temporal gyri and fusiform (Braskie et al. 2012). The authors suggest that this increasing activity phenotype, which was not observed in non-carriers, could be related to early AD processes. They did not test for an association of the increased activity with behavior. Another study that controlled for behavior during the functional task (as a variable of non-interest) found increased activation in the right anterior hippocampus during encoding in a group of PSEN1 mutation carriers versus non-carriers (Quiroz et al. 2010). Resting state fMRI, or task-free fMRI, has only recently gained traction in the FAD literature. Based on spatial coherency, resting state fMRI data can be used to identify specific functional networks in the brain. As expected from work in sporadic, late-onset AD, the default mode network (DMN) is the principal network that is disrupted in FAD. A study led by Reisa Sperling and colleagues used resting state data from 83 FAD mutation carriers and found that overall DMN functional connectivity decreased in mutation carriers as their Clinical Dementia Rating score increased (higher scores indicate greater severity of dementia symptoms) (Chhatwal et al. 2013). This relationship was especially pronounced in the precuneus, posterior cingulate, and the parietal cortices. These same regions also showed significantly decreased DMN functional connectivity in mutation carriers when compared directly to
3 6 4 part vii : I maging G enetics of C ognitive A ging
non-carriers. These results are supported by other studies that suggest DMN dysfunction is a feature of preclinical FAD (Sala-Llonch et al. 2013; Thomas et al. 2014). Perhaps some of the most important work being performed with FAD mutation carriers is the characterization of metabolic changes in the brain and amyloid deposition during preclinical AD using PET tracers. The precuneus, which is a region known to be affected by Aβ deposition and cortical thinning relatively early in AD, shows decreased glucose metabolism in FAD mutation carriers up to 10 years before the expected onset of symptoms (Bateman et al. 2012). Metabolism is measured by using fluorodeoxyglucose (FDG)-PET, which is a radioactive glucose analog that gets taken up by tissues actively using glucose for energy. Lower uptake of the FDG tracer indicates reduced metabolic function, a relatively well-characterized feature of preclinical AD. Aβ imaging using Pittsburgh compound B (PiB)-PET reveals differences between FAD mutation carriers and non-carriers in the precuneus up to 15 years before expected onset of disease (Bateman et al. 2012). In addition, imaging studies using PiB-PET identified one of the major pathological differences between FAD and sporadic, late-onset AD. Specifically, FAD mutation carriers have much higher levels of Aβ in the striatum than patients with sporadic, late-onset AD, even in the preclinical phase (Villemagne et al. 2009). As discussed above, FAD provides a unique opportunity to learn about AD progression during the pre-symptomatic phase. The genetic mutations that cause FAD and many of the imaging findings discussed in this section seem to support an amyloid-centric view of AD pathogenesis. However, as discussed briefly earlier, there are problems with the amyloid cascade hypothesis, which points to amyloid as the catalytic pathology in AD. Namely, it appears that being positive for Aβ in the brain is a necessary component of AD, but it is not sufficient to cause the disease. Research shows that 20%–40% of cognitively healthy elderly adults are positive for Aβ and can remain healthy in that state for years ( Jack et al. 2010). Of course, these subjects do not have an FAD mutation, but work linking FAD to the much more common sporadic, late-onset AD is promising (Quiroz et al. 2013). However, while there are certainly many similarities between FAD and sporadic, late-onset AD, it is still open for debate whether or not the findings in FAD carriers will be directly applicable to developing treatments for sporadic, late-onset AD patients. One notable difference between these two types of AD that was elucidated via neuroimaging studies is that FAD mutation carriers have early Aβ deposition and volume loss in deep brain structures including the thalamus, caudate, and
putamen (Koivunen et al. 2008; Lee et al. 2013; Ryan et al. 2013). The pattern of Aβ deposition and atrophy in sporadic, late-onset AD does not include significant involvement of these structures. To close this section on autosomal dominant AD, let us consider another unique genetic event that can cause FAD. In addition to mutations in the APP, PSEN1, and PSEN2 genes, a duplication of APP, first identified in a Dutch sample, also leads to highly penetrant AD (Sleegers et al. 2006). While there is no neuroimaging work to review on these subjects, this is an appropriate segue into the next section that will focus on another genetic mechanism of presumed APP overexpression: Down syndrome. DOWN SYNDROME
Over 50 different mutations in APP are known to cause familial, autosomal dominant AD. APP is also implicated as a causative gene in the development of AD in individuals with Down syndrome (DS). DS results from an extra copy of chromosome 21. In ways not fully understood, trisomy 21 causes intellectual disability and increases the risk for many medical conditions, including congenital heart defects, hearing and vision impairment, and endocrine dysfunction. Individuals with DS have an average life expectancy of 55 years and suffer from age-related cognitive decline after the age of 40 (Zigman, Silverman, and Wisniewski 1996; Alexander et al. 1997). The APP gene is located on the long arm of chromosome 21 at position 21.3. Compared to healthy individuals with two copies of chromosome 21, there is a dose-dependent increase in the amount of Aβ produced in the brains of individuals with DS. The connection between DS and AD was first described in English in the late 1940s ( Jervis 1948; Lott and Head 2001). The original paper describes the cognitive decline of three older DS patients. Postmortem neuropathological examination of these three patients and countless others in the intervening decades have consistently revealed the presence of amyloid plaques and neurofibrillary tangles in the brains of individuals with DS over the age of 30 (Alexander et al. 1997). Thus, there is assumed to be a connection between the increased expression of APP and the invariable and early appearance of amyloid pathology in the brain. The fact that tau pathology in the form of intracellular neurofibrillary tangles is also present in the brains of middle-aged persons with DS may support the theory that amyloid aggregation is the trigger in a cascade of physiological changes that lead to clinical AD. The relatively limited neuroimaging work in adults with DS has primarily been focused on characterizing
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
365
brain structure using sMRI. Many studies are designed to compare non-demented DS subjects to demented DS subjects. Because differences in these studies are likely attributable to the advanced disease state of the demented subjects, we choose to not review those studies here. Instead we focus on reports that compare non-demented elderly DS subjects to younger DS subjects. Genetics is not the major factor in group differences, but these studies may shed light on features of preclinical AD. Because there are other developmental effects of DS, directly comparing a DS group to a healthy, control group makes it difficult to parse apart developmental differences versus changes due to preclinical AD. That said, the hippocampus and entorhinal cortex, key structures affected early in AD, are reported as reduced in volume in non-demented adults with DS in several studies (Kesslak et al. 1994; Krasuski et al. 2002; Teipel 2003). Also, as expected, with advancing age there is a decrease in volume of medial temporal lobe structures in individuals with DS (Teipel 2003). Because nearly all DS patients will develop AD, it is assumed that this decrease in volume is representative of a disease process rather than normal aging. In the cerebral cortex, one study found that age is correlated with atrophy in regions of frontal and parietal cortices as well as parahippocampal gyrus (Teipel et al. 2004). Another, using cross-sectional data, found a steeper age-related decrease in the volume of frontal, parietal and temporal lobes when compared with age-matched healthy controls (Beacher et al. 2010). A third examined DS brain morphology compared to healthy controls and found decreases in volume in the cingulate gyrus, left medial frontal lobe, and regions of the right temporal lobe (White, Alkire, and Haier 2003). Aside from these three studies, there is a dearth of publications that employ modern volumetric analysis techniques, such as voxel-based morphometry, tensor-based morphometry, or cortical thickness measurements with Freesurfer, in aged DS cohorts. There are also almost no studies that use DWI as a method to interrogate the putative preclinical AD phase in DS. The exception is a very interesting recent study that used tensor modeling with DWI to calculate FA across the brain in healthy older DS subjects. Their findings included a positive correlation between decreasing scores on a global functioning measure and FA in specific frontal ROIs, which indicates that late-myelinating white matter tracts may be particularly vulnerable in older individuals with DS (Powell et al. 2014). Functional imaging in DS is limited to a handful of PET studies. To our knowledge, there are no fMRI studies in older or aging DS subjects. As mentioned previously,
individuals with DS usually are positive for Aβ in the brain at, or soon after, age 30. Thus, the focus of much PET imaging work in DS has been to quantify this deposition in vivo. The first Aβ-specific tracer, PiB, was first published in 2004 (Klunk et al. 2004). In 2011, a study to test the utility and safety of PiB in DS subjects was completed (Landt et al. 2011). The major findings of this study were that the tracer was successful in measuring Aβ plaque load and that age and a clinical diagnosis of AD were positive predictors of amyloid positivity (Landt et al. 2011). Also in 2011, there was an exhaustive case study published in which a 55-year-old DS subject with AD received a PET scan with florbetapir, another amyloid tracer (Sabbagh et al. 2011). At death, the subject’s brain was donated and neuropathological analysis was completed. In general, the pattern of amyloid deposition matched the pattern found in sporadic, late-onset AD and was corroborated by the neuropathological findings. Furthermore, results from another study show that the reduction in glucose metabolism (as measured by FDG-PET) that has been observed years before the onset of sporadic, late-onset AD is recapitulated in non-demented, older DS subjects (Pietrini et al. 1997). Findings like these have helped to motivate the study of older DS subjects because there appear to be many similarities between AD in DS and sporadic, late-onset AD. Concomitant with the increasing interest in the preclinical phase of AD, there is recent, renewed interest in the connection between DS and AD. However, because DS results from an extra copy of an entire chromosome, there could be as yet undiscovered aging-related genes on chromosome 21 that help to influence life span, aging, and AD in individuals with DS, affecting the interpretability of results (Zigman, Silverman, and Wisniewski 1996). NE U ROIMAGING GE NE T IC R IS K F OR ALZ H E IME R ’ S DIS E AS E In the vast majority of cases, AD presents with no clear, underlying genetic cause. This so-called sporadic version of the disease usually affects patients later in life, with an average age of onset roughly 20 or 30 years later than FAD or AD associated with DS, respectively (Zigman, Silverman, and Wisniewski 1996; Cruts, Theuns, and Van Broeckhoven 2012; Thambisetty, An, and Tanaka 2013). The following sections will detail the genetic risk loci that have been associated with sporadic, late-onset AD and neuroimaging findings related to these risk factors. For a summary of the literature reviewed in this portion of the chapter, see Table 22.1.
3 6 6 part vii : I maging G enetics of C ognitive A ging
APOE
The explosion of the neuroimaging genetics field is due largely to the recent rapid identification of novel risk factors for sporadic diseases. In AD, these risk factors can be genotyped in healthy human subjects, allowing genetic risk for AD to be studied in a highly generalizable way in the population at large, rather than small restricted groups of individuals with a highly penetrant genetic mutations or DS. This makes recruitment of large numbers of subjects much more feasible, increasing statistical power. After age, genetic risk factors, such as APOE, are the strongest predictors of sporadic, late-onset AD currently available (Bertram and Tanzi 2012). Because sporadic AD accounts for ~99% of the diagnosed cases of AD, a better understanding of this disease is essential to the development of prevention and treatment strategies. Sporadic AD, hereafter referred to simply as AD, is unique among polygenic human diseases because there is a well-validated, non-causative genetic risk factor, APOE, which accounts for a relatively large portion of the variation in heritability. Specifically, twin studies reveal that the heritability of AD may exceed 60%–80% and APOE genotype accounts for about 50% of the variation in heritability (Bergem, Engedal, and Kringlen 1997; Gatz et al. 2006; Waring and Rosenberg 2008). A single APOE ε4 (APOEε4) allele increases lifetime risk for AD 4- to 5-fold, and two copies of the allele confer at least a 10-fold increase (Corder et al. 1993; Bertram and Tanzi 2012). APOE was identified as a susceptibility gene for AD over 20 years ago and has been studied extensively since (Corder et al, 1993; Schmechel et al. 1993; Strittmatter et al. 1993). The APOE gene is localized on chromosome 19 and has three common alleles (e2, e3, and e4) determined by polymorphisms at two SNP sites, rs429358 and rs7412. Combinations of these three alleles result in six possible genotypes in the general population. APOE is a lipid transport protein that is believed to play a fundamental role in cell maintenance and repair (Leduc et al. 2011). It has also been implicated as a regulator of normal cell metabolism, as well as other functions (Leduc et al. 2011). In the years since the discovery of the association between APOE genotype and AD, fMRI has progressed from a novel, infant technology to one of the most popular methods in human neuroimaging research. The strength of the disease risk conferred by APOE, as well as the co-maturation of the fields of AD genetics and fMRI acquisition and analysis, led to the first study combining a genetic risk factor for a disease and neuroimaging (Bookheimer et al. 2000). This study, which found putative compensatory increases in activity in APOEε4 carriers, as well as others published shortly thereafter, helped to
expand the horizons of neuroimaging genetics, a new subfield of neuroscience and the topic of this book. To date, there have been hundreds of publications focusing on neuroimaging the genetic risk for AD conferred by APOE. Because it is impossible to cover every aspect of this dense literature, it is worth noting that there are excellent reviews available to complement the information included in this section (Cherbuin et al. 2007; Bookheimer and Burggren 2009; Donix, Small, and Bookheimer 2012; Trachtenberg, Filippini, and Mackay 2012). We will summarize the key elements of this body of work, focusing on new and emerging research. Due to the heterogeneity of cohorts across the literature, we have divided our summary of imaging findings into three subsections: healthy older adult cohorts, young healthy cohorts and, finally, MCI and AD cohorts.
Healthy, Older Adult Cohorts In healthy older adults, hippocampal volumes have been shown to be smaller in APOEε4 carriers compared to non-carriers (Lu et al. 2011; Taylor et al 2014). Hippocampal atrophy rates are also higher in APOEε4 carriers ( Jak et al. 2007; Donix et al. 2010b). There is evidence that hippocampal volumes vary in an allele dosedependent manner, but most studies’ recruitment efforts conclude before they can amass enough homozygous APOEε4 carriers to consider them separately (Hostage et al. 2013). In addition to whole hippocampal volume, sMRI can be used to measure structural changes within specific areas of the hippocampus. High-resolution, partial field of view sMRI allows for the segmentation of the hippocampal complex into specific subregions, such as the subiculum, the entorhinal cortex, and the CA subfields. Using this approach, several labs have reported smaller or thinner subregions in healthy APOEε4 carriers. Specifically, healthy APOEε4 carriers have been found to have thinner entorhinal cortex and subiculum compared to non-carriers (Burggren et al. 2008). Two additional studies, each using MR images acquired at 4T, found thinner CA3 and dentate gyrus subfields in APOEε4 carriers (Mueller et al. 2008; Mueller and Weiner 2009). However, not all studies that examined hippocampal volumetric differences between healthy APOEε4 carriers and non-carriers found significant differences, although these are certainly in the minority (Cherbuin et al. 2008). There are very few reports of differences in cerebral cortex volume or thickness in healthy older adults based
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
367
on APOE genotype, so a consensus is difficult to develop. This may be because neutral results are not published as often as results showing significant differences. One published study that examined cortical volumetric differences between healthy APOEε4 carriers and non-carriers found no significant differences (Cherbuin et al. 2008). However, another study found that APOEε4 carriers had thicker cortex in bilateral frontal and temporal regions, but a steeper longitudinal atrophic trajectory across the cortex (Espeseth et al. 2008). This points, again, to an emerging theme that individuals with at least one copy of the APOEε4 allele experience an acceleration of the volume loss seen in normal aging. A caveat of volumetric, structural findings in the APOE literature is that atrophy or volume loss is often seen as an indication of disease processes, while increased volumes or decreased atrophy rates are not. Thus, it is likely that intuitive results, for example where APOEε4 carriers have lower or smaller volumetric measurements, are favored in the published literature. A lack of such a biasing intuition in fMRI may partially explain why the results in the APOEfMRI literature are more contradictory, as will be discussed in the following text. Because APOE is a lipoprotein that transports endogenous lipids, there is interest in better understanding its relationship with myelin, which needs lipids for maintenance and repair. In neuroimaging, investigators can use DWI to examine the potential relationship between APOE and myelination, using “white matter integrity” measured by FA as a proxy for myelin health. White matter integrity in the medial temporal lobe, but not entorhinal thickness, has been shown to be associated with improved performance on a verbal memory task (Westlye et al. 2012). There is also evidence for a general decrease in FA in APOEε4 carriers (Heise et al. 2011). Diffusion tensor imaging (DTI) allows mathematical concepts from the field of graph theory to be applied to structural brain imaging data. In a study by Brown and colleagues (2011), graph theory was used to measure global integration and local interconnectivity in healthy, older subjects. APOEε4 carriers had an age-related decrease in local interconnectivity that may indicate different aging trajectories in APOEε4 carriers and non-carriers. The application of graph theory to sMRI data as well as resting state fMRI data may help to elucidate the local and global network properties that change during early AD, but more research is needed in this area before such measures can be considered as potential biomarkers or endophenotypes of AD. A quick review of the task-based fMRI-APOE literature reveals a frustratingly complex picture. Some studies
have reported increased, putatively compensatory, activity in APOEε4 carriers (Bookheimer et al. 2000; Bondi et al. 2005). Others have reported decreased activity, putatively caused by a loss of function due to disease processes (Borghesani et al. 2008; Xu et al. 2009). Part of the complexity stems from the heterogeneity of task designs (Trachtenberg, Filippini, and Mackay 2012). Differences can be stark. For example, it may be hard to compare results from a semantic memory task and a visuospatial memory task (Bookheimer et al. 2000; Borghesani et al. 2008). Other potentially confounding factors in task design can be more subtle. A task described as a “paired associates” memory task can actually vary widely on several factors including, but not limited to, method of presentation of stimuli (audio, visual, or both), types of stimuli (images, words, etc.), and instructions (“pay attention” versus “remember these pairs”) (Trachtenberg, Filippini, and Mackay 2012). There are also many studies in the literature in which investigators used non-episodic-memory–based tasks, complicating interpretation because there is evidence that APOEε4 exerts a specific effect on episodic memory systems (Rogalski et al. 2011). In contrast to the whole-brain approach of the studies cited here, the results from studies that examined BOLD activity in the hippocampus as an ROI are more cohesive. One study, which acquired data using a high-resolution fMRI sequence, found decreased activity in APOEε4 carriers in the CA2, CA3, and dentate gyrus subregions of the hippocampus (Suthana et al. 2010). Another found decreased hippocampal activity during encoding in APOEε4 carriers (Adamson et al. 2011). Results from resting state fMRI work in healthy older APOEε4 carriers present a more unified picture. There appears to be a convergence on the DMN and connectivity therein, by which APOEε4 carriers and non-carriers differ. In a very recent study, connectivity between the posterior cingulate cortex and the hippocampus, two major nodes of the DMN, was found to be diminished in APOEε4 carriers (Heise et al. 2014). Another study, focusing on female APOEε4 carriers, reported significantly reduced DMN connectivity compared to female non-carriers (Damoiseaux et al. 2012). Finally, decreased DMN connectivity and increased connectivity of another, opposing cognitive network, the salience network, have been described (Machulda et al. 2011; Sheline et al. 2010). One theory explaining the DMN dysfunction reported in APOEε4 carriers states that the genetic vulnerability for AD may cause a loss of appropriate hippocampal decoupling from cortical DMN regions during activity, for example when completing a task (Westlye et al. 2011). This theory is supported by the discovery of a negative
3 6 8 part vii : I maging G enetics of C ognitive A ging
correlation between hippocampus-DMN synchronization and performance on a memory test (Westlye et al. 2011). More work is needed to explicitly test this theory. PET imaging has helped elucidate the relationship between APOE and Aβ. Today, we understand that while the relationship is still far from fully understood, the protein products of APOE play a role in Aβ clearance, with APOEε4 performing this task less well than the ε3 or ε2 alleles (Ledue et al. 2011). This idea is supported by PET imaging studies in which the relationship between Aβ (measured with PiB or florbetapir) and APOEε4 carrier status is examined. The majority of these studies report that healthy, older APOEε4 carriers have increased amyloid load compared to non-carriers (Morris et al. 2010; Fleisher et al. 2011; Fleisher et al. 2013; Sheinin et al. 2014). There are also metabolic differences between healthy APOEε4 carriers and non-carriers. A very large study with 806 cognitively normal, PiB negative subjects recently showed that glucose metabolism in APOEε4 carriers is lower than noncarriers in the posterior cingulate, precuneus, lateral parietal, and inferior temporal regions (Knopman et al. 2014) (Figure 22.2A). The magnitude of this difference was small, but commensurate with differences observed between cognitively normal and MCI APOEε4 carriers (Figure 22.2B). There was also an overall negative correlation between FDG uptake and age across the whole cohort, with the posterior cingulate and precuneus exhibiting a particular vulnerability to both age and APOEε4 carrier status (Knopman et al. 2014). This work is supported by previous studies that also reported hypometabolism in AD-vulnerable regions in healthy APOEε4 carriers (Reiman et al. 1996). However, a recent study of 600 cognitively normal older subjects found no FDG-PET metabolism differences in APOEε4 carriers and non-carrriers (Lowe et al. 2014). This discrepancy may be based on the inclusion of PiB positive subjects in the latter report, who were stratified based on tracer uptake. Perhaps when subjects are binned by amyloid burden, the power to detect APOEε4 differences in metabolism is diminished.
Young, Healthy Cohorts There is a burgeoning literature focusing on the effects of the APOEε4 allele in younger people, from middleaged adults to young adults to infants. While there are few uncontested results, it is evident that APOEε4 carrier status affects brain structure and function well before old age. One thorough study of the effect of APOEε4 measured by various imaging modalities in young adults only found differences in fMRI activity, despite also acquiring
and analyzing DWI for tensor modeling and sMRI for VBM in the same subjects (Matura et al. 2014). Other studies have also found no differences in hippocampal volume (Richter-Schmidinger et al. 2011; Khan et al. 2014). However, there is some evidence that hippocampal volume differs between APOEε4 carriers and non-carriers. In a small study of 44 subjects, the authors found decreased hippocampal volume in the group of 22 APOEε4 carriers compared to non-carriers (O’Dwyer et al. 2012). Small sample size may be one reason that this finding does not fall in line with the others that interrogated hippocampal volume. In the cerebral cortex, reduced gray matter volume in AD-signature regions, including the lateral parietal, temporal, and cingulate cortices, has been detected in young adult APOEε4 carriers (Alexander et al. 2012). Another study found no differences in gray matter volume in young APOEε4 carriers and non-carriers (Samuraki et al. 2012). The contradictions in these structural findings will hopefully be resolved as larger data sets of young adults are being genotyped for larger numbers of SNPs (perhaps even undergoing whole genome sequencing). The expanding genetic data available may include AD risk factors that were previously unlikely to be included in large data collection efforts focused on young adults. In contrast to sMRI, DWI appears to be a relatively sensitive imaging modality for uncovering differences between young APOEε4 carriers and non-carriers. A study of 203 subjects found a diffuse and widespread increase in mean diffusivity (MD) in APOEε4 carriers (Westlye et al. 2012). Another found a general reduction in FA along with increased MD in carriers (Heise et al. 2011). More work is needed to establish alterations in DTI metrics as a potential biomarker of early APOE-mediated neural differences in young adults. Contradictory results in fMRI experiments comparing APOEε4 carriers to non-carriers are not limited to older adult cohorts. Functional studies in young adults have reported decreased task-related activity in APOEε4 carriers that may indicate a blunted recruitment of the neural machinery necessary to complete the task efficiently (Green et al. 2014). However, there is also evidence that hippocampal activation during memory tasks is higher in young APOEε4 carriers (Filippini et al. 2009). Greater activation could be indicative of a compensatory mechanism in order to maintain performance. This theory does not appear to be supported by the APOE literature in young adults. In fact, differences in activity and cognitive performance suggest that APOEε4 carriers have better attention and memory function than non-carriers (Mondadori 2007; Rusted et al. 2013). The latter findings and others
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
369
Cognitively Normal
(b)
Elderly Carrier
2.4
2.4
2.2
2.2
FDG Ratio (PVCY) AD-signature meta-ROI
FDG Ratio (PVCY) AD-signature meta-ROI
(a)
2.0 1.8 1.6
2.0 1.8 1.6 1.4
1.4
1.2 Non-Carrier
Carrier
CN
MCI
AD dementia
Figure 22.2 Boxplots of the FDG to reference region ratio (partial volume corrected; PVCY) in AD-signature meta-ROI. Figure adapted from Knopman DS, et al. (2014), Neurobiology of Aging.
A. FDG binding in CN APOEε4 carriers and non-carriers of all ages. B. FDG binding across elderly (age ≥ 70 years) cognitively normal (CN), MCI, and AD cohorts. Metabolism decreases as clinical disease severity, as represented by clinical diagnosis, increases. The magnitude of the differences between cognitively normal APOEε4 carriers and non-carriers is similar to the difference between CN and MCI groups, which were matched for APOEε4 carrier status in addition to age and sex. Reprinted with permission from Elsevier.
have led to a moderately popular theory of antagonistic pleiotropy, still only tenuously supported, in which the APOEε4 allele confers some beneficial advantage in young people, only to then predispose older people to AD. One reason that this theory has gained some traction is that is may help explain why, despite the negative effects of the APOEε4 allele, it remains a relatively common variant, with 20%–25% of the population carrying at least one copy (Corder et al. 1993). The argument is that an allele with deleterious effects would not be so common unless there were early life benefits. The counterargument to this evolutionary reasoning is that the human life span has only been long enough to experience the negative effects of the APOEε4 allele for a relatively brief epoch of our evolutionary history. Furthermore, even in APOEε4 carriers, AD usually manifests at the end of or after the reproductive phase of life, which would minimize selection pressure against APOEε4 carriers. Using resting state fMRI, there is evidence of altered DMN function in young adults with the APOEε4 allele (Filippini et al. 2009). This mirrors what has been discovered in older healthy adults and, as will be discussed later, in MCI and AD. There is also evidence that alterations in resting state networks mediated by the APOEε4 allele may not be tightly linked to risk for AD. In a study by Trachtenberg and colleagues (2012), resting state networks that differed between APOEε4 carriers and APOEε3 homozygotes (including bilateral hippocampal networks,
the auditory network, the left frontal-parietal network and the lateral visual network) also differed between APOEε2 carriers and APOEε3 homozygotes (Figure 22.3). The APOEε2 allele has been shown to be protective against AD (Corder et al. 1994). Therefore, the authors reason, these findings would indicate that the differences between APOEε4 carriers and non-carriers were not a reflection of increased AD risk or early AD-related changes, but rather point to a role for APOE in neurodevelopment. Certainly, this study provides a compelling rationale for including APOEε2 allele carriers as an additional experimental group in future studies that aim to elucidate early AD-related changes in the brain. As a final note on functional imaging of APOE genotype in young adults, a seminal FDG-PET study in 2004 found that in a small cohort of young adults, APOEε4 carriers were hypometabolic compared to non-carriers in the posterior cingulate cortex, parietal, temporal, and frontal lobes (Reiman et al. 2004). These regions are particularly vulnerable to AD and are the sites of marked hypometabolism early in the disease. Unfortunately, there are very few studies using PET imaging in young adults, so these results have not been properly reproduced. We expect the number of studies that report on cohorts of adolescents, children, and infants and the APOE gene will increase in the coming years. Today, there are several interesting published reports that are sure to inspire follow-up and further experiments. A study completed on 239 children
3 7 0 part vii : I maging G enetics of C ognitive A ging
AHN
ε4 vs. ε3/ε3
ε2ε3 vs. ε3/ε3 2.0 ***
–4.0
ε4 vs. ε3/ε3
10.0
ε4 vs. ε3/ε3
*
*
ε2ε3 ε3ε3 ε3ε4 ε4ε4
15.0 10.0
*
*
**
5.0 ε2ε3 ε3ε3 ε3ε4 ε4ε4
ε2ε3 vs. ε3/ε3 40.0 Connectivity
Connectivity
*
–6.0
0
ε2ε3 ε3ε3 ε3ε4 ε4ε4
30.0 20.0 * 10.0
*
–4.0
20.0 Connectivity
Connectivity
20.0
0
*
*
30.0 20.0
ε4 vs. ε3/ε3
ε2ε3 vs. ε3/ε3
–
**
*
10.0 0
ε2ε3 ε3ε3 ε3ε4 ε4ε4
ε2ε3 ε3ε3 ε3ε4 ε4ε4
2.0 *** Connectivity
IFPN
ε2ε3 vs. ε3/ε3
*
0 –2.0
–8.0
ε2ε3 ε3ε3 ε3ε4 ε4ε4
30.0
0
AUN
*
–8.0 –12.0
PHN
*
*
Connectivity
Connectivity
0
0 –2.0 –4.0 –6.0
* ε2ε3 ε3ε3 ε3ε4 ε4ε4
Figure 22.3 The effects of APOE genotype on connectivity of several resting state networks, including the anterior hippocampal network (AHN), the
posterior hippocampal network (PHN), the auditory network (AUN), and the left frontal-parietal network (lFPN). The effects of APOE genotype on connectivity of several resting state networks, including the anterior hippocampal network (AHN), the posterior hippocampal network (PHN), the auditory network (AUN), and the left frontal-parietal network (lFPN). Figure reproduced from Trachtenberg AJ, et al. (2012), NeuroImage.
The left column shows the results of voxel-wise comparisons between APOEε4 carriers and APOEε3 homozygotes. The right column shows the results of voxel-wise comparisons between APOEε2/ε3 heterozygotes and APOEε3 homozygotes. Images are thresholded at z > 2.3 and corrected for multiple comparisons using a corrected cluster significance threshold of p < 0.05. Bar graphs show the ROI analyses on the significant regions from the voxelwise comparisons. Error bars denote standard error of the mean. * Significantly different from APOEε3/ε3 (p < 0.05); **significantly different from APOEε3/ε3 and APOEε3/ε4 (p < 0.05); ***significantly different from APOEε3/ε3, APOEε3/ε4, and APOEε4/ε4 (p < 0.05). Reprinted with permission from Elsevier.
and adolescents found that APOEε4 carriers had significantly thinner entorhinal cortex than non-carriers (Shaw et al. 2007). Moving to even younger subjects, a recent study led by Eric Reiman scanned 162 infants from 2 to 25 months old and found that APOEε4 carriers had lower gray matter volume in the precuneus, the cingulate, lateral temporal cortex, and medial fusiform gyrus (Dean et al. 2014). Support for these findings can be found in a paper examining the effects of psychiatric risk genes in prenatal development. Specifically, the APOEε4 allele was related to decreased gray matter volume in bilateral hippocampus, parahippocampus, fusiform, and temporal gyri (Knickmeyer et al. 2014). Together, these studies
suggest that there is a role for APOEε4 in development. This begs the question, is APOEε4 risk for AD a developmental susceptibility or a direct interaction with diseases processes? It has been shown that APOEε4 risk is specific to amnestic dementia (Rogalski et al. 2011). Studies on extremely young subjects might help to uncover if there are developmental clues as to the mechanism of this specificity. To date, there are no studies examining the functional consequences, measured with fMRI or PET, of APOE risk in infants. fMRI studies are likely to be completed soon as safety concerns for infants in fMRI experiments are minimal and motion correction techniques are constantly improving. PET studies in infants are extremely
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
371
unlikely due to the required radiation exposure via the use of the radioactive tracer.
MCI and AD Cohorts Similar to findings in healthy, older adults, APOEε4 carriers with AD or MCI have higher rates of hippocampal atrophy compared to non-carriers with AD or MCI, respectively (Pievani et al. 2011; Manning et al. 2014). Related to the increased rates of atrophy, AD and MCI subjects who carry at least one copy of APOEε4 have reduced hippocampal volume or more severe hippocampal thinning (Liu et al. 2010; Pievani et al. 2011). Earlier, it was mentioned that clinical symptoms of AD generally follow marked atrophy, and synaptic and neuronal loss, so it is not surprising that volumetric loss has occurred in these symptomatic cohorts. However, these studies are highlighting that the APOEε4 allele mediates more severe disease phenotypes, even in clinically affected patients. In studies of the cerebral cortex, APOEε4 carriers with MCI who then progress to AD show decreased gray matter volume in the temporal and parietal lobes (as well as decreased hippocampal volume), while APOEε4 non-carriers showed no gray matter volume changes over the same time elapsed (Spampinato et al. 2011). Taking together these sMRI findings and the many studies reporting that APOEε4 is associated with an earlier age of onset of AD, it is clear that APOEε4 is associated with a more rapid disease progression (Thambisetty, An, and Tanaka 2013). Alterations in DTI metrics based on APOEε4 carrier status appear to be a feature of young, healthy subjects and the preclinical phase of AD but not of symptomatic cohorts. A study that calculated FA and MD in two groups, AD patients and healthy controls, found that MD was significantly greater in APOEε4 carriers compared to noncarriers in the healthy control group, but not in the group of AD subjects (Kljajevic et al. 2014). In fact, there were no differences in DTI metrics mediated by APOE genotype in the AD group. At the time of preparation of this chapter there were no published studies that employed either task-based or resting state fMRI to study differences in symptomatic populations based on APOE genotype. There is, however, a well-established literature in which PET tracers are used to assess differences in MCI and AD cohorts based on APOEε4 carrier status. With FDG-PET, AD patients who are APOEε4 carriers present with more dramatic metabolic reductions in the regions that are normally hypometabolic in AD, including the lateral parietal lobe, posterior cingulate, precuneus, and temporal lobes
(Mosconi et al. 2004). The spatial extent of the hypometabolic regions is also greater in APOEε4 carriers (Drzezga et al. 2005). PiB-PET studies that test for an association between Aβ deposition and APOEε4 carrier status in AD patients find that APOEε4 carriers have higher tracer uptake across diffuse regions of the cortex (Ossenkoppele et al. 2013; Lehmann et al. 2014). Another PiB-PET study, this one in MCI patients, found that of the 61% of MCI subjects who were positive for Aβ, 80% of them were APOEε4 carriers (Pike et al. 2007). This indicates that APOEε4 is associated with an increased risk for amyloid positivity in subjects with MCI. Taking this one step further, it would indicate that individuals with MCI who are also carriers of the APOEε4 allele are more likely to convert to AD than non-carriers because amyloid positivity is a good predictor of progression (Hatashita and Yamasaki 2013). As an interesting aside, let us consider APOE from another, extremely rare, perspective. Recently, it was discovered that a man with a severe form of dysbetalipoproteinemia was completely missing the APOE gene. A case study was published detailing his neurological status based on cognitive testing, MR imaging, and CSF analytes (Mak et al. 2014). He was found to have no neurological deficits or structural abnormalities of the nervous system. The story of this remarkable patient has led to a resurgence of attention on APOE as a potential therapeutic target. If the absence of APOE does not negatively affect neurological function, perhaps the APOEepsilon4 allele product can be silenced, thus eliminating APOEepsilon4-mediated AD risk. Of course, this therapy would need to be targeted to the CNS, as a lack of APOE throughout the body results in excessively high cholesterol as well as other clinical problems. TOMM40
In 2010, a non-coding region on chromosome 19 located just upstream from APOE was identified as a strong genetic risk locus for AD (Roses et al. 2010). This stretch of DNA, called TOMM40 for translocase of outer mitochrondrial membrane 40, varies with respect to the length of a poly-T polymorphism. Longer length poly-T variants were found to be associated with increased risk for AD, as well as a lower age of onset (Roses et al. 2010). The authors of these initial findings contended that the discovery was important because Tom40, the protein encoded by this region, is crucial to healthy mitochondrial function. The Tom40 protein forms a channel in the outer mitochondrial membrane that is used to import proteins (Hong et al. 2010). Since these initial findings, there has been much disagreement in the
3 7 2 part vii : I maging G enetics of C ognitive A ging
field as to whether or not TOMM40 is specifically associated with AD. Some believe that TOMM40 is in such close linkage disequilibrium with APOE that any signal at the TOMM40 locus in an AD association study is driven by the very strong APOE signal (Carrasquillo et al. 2012). These investigators postulate that the TOMM40 polymorphism is “behaving as a surrogate for the well-established AD risk allele, APOEε4.” Subsequent to this, three groups attempted to replicate Roses and colleagues’ original (2010) findings. One group found that longer lengths of the TOMM40 polyT repeat did in fact increase risk for AD, but only in the absence of APOEε4 (Maruszak et al. 2012). Another found no correlation between TOMM40 poly-T repeat length and age of onset of AD (Chu et al. 2011). Finally, a third group found that the TOMM40 poly-T polymorphism association did not replicate, but reported another TOMM40 polymorphism associated with increasing risk of AD in APOEε3 homozygotes, but in the opposite of the expected direction (i.e., increasing length was associated with a lower risk of AD) (Cruchaga et al. 2011). Given the various directions of the findings, the true implications of the TOMM40 polymorphism remain to be determined. One group attempted to elucidate the complicated regulation of the relatively small haplotype block that encompasses APOE and TOMM40. The authors investigated the effect of putative cis-regulatory haplotypes on in vitro expression driven by TOMM40 and APOE promoters, and their results suggest that genetic variation at the TOMM40 locus may indeed be associated with late-onset AD, independently of APOE (Bekris, Lutz, and Yu 2012). More recently, neuroimaging evidence that TOMM40 is not merely a marker of APOE genotype has come to light (Ferencz et al. 2013). Specifically, TOMM40 was found to have an additive and separable effect on the association between hippocampal volume and memory performance on a free recall task (Ferencz et al. 2013). Another study of healthy older adults who were not carriers of APOEε4 found a dose-dependent effect of high-risk TOMM40 alleles correlated to decreasing performance on retrieval in a verbal memory task ( Johnson et al. 2011). The authors also reported a dose-dependent, high-risk allele correlation with decreasing gray matter volume in the ventral posterior cingulate cortex and medial ventral precuneus, both regions that are implicated early in AD pathophysiology. Finally, in a cohort of generally healthy older adults assessed with DWI, the authors found significant and independent effects of both the APOEε4 and TOMM40 “short” alleles on specific tracts, independent of age, gender, vascular disease and childhood intelligence (Lyall et al. 2014). For TOMM40, these tracts were the left uncinate fasciculus,
left rostral cingulum, and left ventral cingulum. It remains unclear why specific tracts show significant deleterious effects of genetic variation at the APOE or TOMM40 loci, but one hypothesis is that these late-myelinating tracts are particularly susceptible to injury or pathology (Benitez et al. 2014). Clearly, a better understanding of the regulatory mechanisms affecting TOMM40 and APOE will be necessary to tease apart their relationships to AD risk. In addition, continuing to measure variant-driven pathological differences in the brain will help resolve whether the two genes have an additive effect on disease-related changes. Perhaps by investigating the rare variants within each of the genes that are not in linkage disequilibrium but that do associate with AD, it could be determined whether the genes exert their effects independently. Only a next generation sequencing method would support the large-scale effort required to implement such an approach. TREM2
Triggering receptor expressed on myeloid cells 2 (TREM2) is a gene that has very recently been implicated in AD. Two independent studies were published in 2013 that linked a SNP (rs75932628) located within TREM2 to AD (Guerreiro et al. 2013; Jonsson et al. 2013). The first studies quantifying the risk conferred by this TREM2 variant have indicated that it could be as strong or stronger than the APOEε4 allele, which is, however, much more common. In one study, the allelic odds ratio for the TREM2 variant was over 11 (Finelli et al. 2014). For comparison, risk loci identified in GWA studies for AD have odds ratios up to 1.5, while the APOE loci is often near or above 3 (see next section) (Lambert et al. 2009). The strength of the association to AD in these initial reports has given rise to a burst of interest in the neurobiological underpinnings of the relationship. The substitution of a C to a T base-pair results in the substitution of a histidine for arginine in the TREM2 protein and has been associated with low cell-surface expression of the protein and an increased risk for AD (Guerreiro et al. 2013). TREM2 is an immune receptor responsible for regulating microglial cytokine production and phagocytosis of neuronal elements, neuritic debris, and bacteria (Finelli et al. 2014). This protein is expressed in microglia, and low cell-surface expression of TREM2 in transgenic mouse models of AD is associated with reduced phagocytic functions (Melchior et al. 2010; Lue et al. 2014). Recent evidence supports the notion that TREM2 is capable of phagocytosing Aβ, and mutations in TREM2 may lead to reduced clearance of protein aggregates in the
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
373
brain (Melchior et al. 2010). Additionally, up-regulation of TREM2 has been shown to alleviate neuropathological symptoms of AD in a transgenic mouse model ( Jiang et al. 2014). Interestingly, TREM2 variants have also been associated with other neurodegenerative diseases such as Parkinson’s disease, frontotemporal dementia, and amyotrophic lateral sclerosis (Rayaprolu et al. 2013; Cady et al. 2014; Ruiz et al. 2014). Despite its association with numerous neurodegenerative diseases, no systematic description of the clinical and neuropsychological features of the TREM2 variant has been determined, in part due to the rarity of the mutation (Reitz and Mayeux 2013). The infrequency of the variant has limited the number of neuroimaging investigations dedicated to investigating the clinical presentation and patterns of gray and white matter morphology specific to TREM2. One cross-sectional study used voxel-based morphometric analysis to investigate regional patterns of gray and white matter loss associated with the at-risk variant (Luis et al. 2014). The authors found gray matter volume loss was largely restricted to frontobasal regions including orbitofrontal cortex and anterior cingulate cortex. In another tensor-based morphometry study, Rajagopalan and colleagues (2013) reported 1.4%–3.3% annual rates of increased volume loss of the medial temporal lobe in at-risk TREM2 subjects. However, no comprehensive whole-brain sMRI study of gray and white matter differences in TREM2 variant carriers exists as of yet. Functional studies, especially PET imaging work exploring the relationship between TREM2 and Aβ deposition, are likely forthcoming. GWA-STUDY-IDENTIFIED ALZHEIMER’S DISEASE RISK GENES
Beginning in 2008, several large-scale genome-wide association (GWA) studies examining genetic association with AD were published (Bertram et al. 2008; Harold et al. 2009; Lambert et al. 2009; Hollingworth et al. 2011). These studies confirmed previously identified risk factors (APOE, CLU), and identified new putative genetic risk factors for AD (PICALM, CR1, BIN1, and others). Several of the GWAidentified loci have been examined in subsequent studies of quantitative measures of cognitive decline and biomarkers for AD. These phenotypes can include metrics of cognitive performance, functional and structural imaging biomarkers, PET tracer uptake, and CSF analytes. One purpose of genotype-driven phenotype studies is to better understand the role of GWA-identified genes and their protein product(s) in AD pathogenesis. Neuroimaging data are acquired at a resolution much lower than the scale on which proteins act, so
these data can provide only limited insight into pathogenic processes at the molecular or cellular level. Instead, what neuroimaging can do very well is help assess the clinical utility of AD-associated genetic variants. In other words, in a living patient, can low effect size genetic risk factors be combined, along with biomarkers, to improve predictions about conversion to MCI and, subsequently, to AD? If so, this clinical utilization of GWA-study-identified AD risk genes could allow clinical trial enrollment to be a more rigorous and specific process, with the ultimate goal of including only those individuals who are most likely to respond to the treatment, increasing statistical power to show an effect. In 2013, the International Genomics of Alzheimer’s Project (IGAP) consortium published their first GWA effort, the largest ever on AD (Lambert et al. 2013). Using a uniquely large cohort of 74,046 subjects amassed from four smaller data consortia, the authors were able to detect 11 new AD risk loci, in addition to confirming previously identified loci (Figure 22.4). Using reference haplotype data from the 1000 Genomes Project for imputation and a predetermined genome-wide significance level of p < 5 X 10-8, the stage 1 analysis resulted in 15 genomic regions that showed an association to AD. These regions included 10 previously identified AD genetic risk factors, including APOE, and 5 newly implicated loci. The 9 previously identified loci were CR1, BIN1, CD2AP, EPHA1, CLU, MS4A6A, PICALM, ABCA7, and CD33. All available neuroimaging genetics findings for these loci will be reviewed in the following text. The 6 new loci were HLA-DRB5HLA-DRB1, PTK2B, SORL1, SLC24A4, RIN3, and DSG2. After the stage 2 replication analyses, there were 7 additional loci that reached statistical significance for association: INPP5D, MEF2C, NME8, ZCWPW1, CELF1, FERMT2, and CASS4. Notably, in the stage 2 replication analyses there were two loci from stage 1 that did not reach statistical significance: CD33 (a previously identified risk locus) and DSG2 (a newly identified locus). This left a total of 9 fully replicated, previously identified risk loci, including APOE, as well as 11 newly identified potential risk loci (Figure 22.4). While this landmark study further complicated the genetic landscape of AD, it has helped to provide a more complete picture of the underlying genetics that lead to AD. These findings are very exciting to investigators taking an imaging genetics approach because this larger set of genetic markers will help to elucidate the role of the field in the future of AD research. Genetic risk markers identified in early GWA studies are featured in a growing number of imaging genetics publications. Unfortunately, the new 11 loci described by Lambert and colleagues have not yet been
3 7 4 part vii : I maging G enetics of C ognitive A ging
APOE
40
–log10(P)
30
20
10
20 21 22
17 18 19
15 16
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0 Chromosome Figure 22.4 Manhattan plot depicting genome-wide associations with AD (17,008 cases and 37,154 controls). Figure reproduced from Lambert JC, et al. (2013), Nature Genetics.
Genes that have been identified in previous GWA studies are in shown in black and newly associated genes are shown in red. Red diamonds indicate loci with the smallest p-values in the overall analysis. Reprinted with permission from Nature Publishing Group.
queried using a neuroimaging genetics approach. Surely, these studies are underway. In several years, it is likely that these newly identified risk loci will also be represented in the imaging genetics literature. The remainder of this section will cover the published neuroimaging genetics findings for replicated GWA-identified risk factors, including a brief background on each gene.
CLU The gene clusterin (CLU) is ubiquitously expressed, and its protein product has been implicated in a plethora of cellular functions, which seem to converge on CLU’s role as a chaperone protein (Morgan and Carrasquillo 2013). Before the advent of the GWA approach, CLU was already implicated in AD. Its potential relationship to the disease was first described by May and colleagues in 1990 when they found that CLU expression was significantly increased in the hippocampi of AD patients compared to controls (May et al. 1990). However, the coincident implication of CLU in two independent GWA studies published in 2009 reignited the interest in CLU and its role in AD (Harold et al. 2009; Lambert et al. 2009). Furthermore, since those two initial reports, the association of a single SNP within
the CLU gene has been replicated several times, making it a GWA study replication success story (Corneveaux et al. 2010; Jun et al. 2010; Seshadri et al. 2010). One caveat that is especially important to molecular biologists is that the associated SNP (rs11136000) is intronic and therefore is not expected to have an effect on protein function (Harold et al. 2009). The search for the exonic, coding variants that are the true causative polymorphisms that underlie the association is ongoing. Compared to other GWA-study-identified genetic risk factors, CLU has been afforded the most attention in the neuroimaging genetics world. This may be because there was a preexisting literature linking CLU and AD and, thus, investigators felt that the CLU association was “real” and represented a signal strong enough to be picked up in neuroimaging approaches. The relatively large number of CLU studies may also be due to the highly reproduced nature of the genome-wide association with a single locus, again indicating a strong, “real” association. There is no evidence that CLU genotype is associated with volumetric measures of medial temporal lobe structures. In healthy young adults, CLU was specifically shown to not be associated with hippocampal or entorhinal cortex volume (Bralten et al. 2011). However, in a study of
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
375
PICALM The gene encoding phosphatidylinositol binding clathrin assembly protein (PICALM) was identified as an AD risk factor in 2009 (Harold et al. 2009). The implicated
Functional coupling during recall
Coupling with Right Hippocampus 1.0
5
T
2
Size of effect at [32 52 20]
(a)
0.8 0.6 0.4 0.2 0.0 –0.2 –0.4 –0.6
CC
CT
TT
Genotype
(b)
Functional coupling during recognition
Coupling with Left Hippocampus 1.5
5 2
T
Size of effect at [36 34 24]
young, healthy adults, the CLU risk allele was associated with poorer working memory performance and, further, this relationship was mediated by the gray matter volume of a region of the parietal lobe (Stevens et al. 2014). The authors also tested for a relationship between APOE, working memory, and gray matter morphology, but found no significant association (Stevens et al. 2014). Using DTI, the CLU risk variant has been associated with lower FA in several white matter regions, including the fornix, the splenium of the corpus callosum, and the cingulum, which are all tracts that contribute directly to the structural connectivity of the medial temporal lobe (Braskie et al. 2011). Decreases in FA have emerged in the APOE literature as a possible early indication of disease-susceptibility. More work is needed to ascertain whether or not there is an additive effect of risk genes on FA, but preliminary efforts are promising (Kohannim et al. 2012). In contrast to findings in gray matter volume described in the preceding text, an fMRI experiment that tested for an effect of CLU and/or APOE on brain activity during an executive attention task found that the effect of the genes was additive (Green et al. 2014). Specifically, the authors found that as genetic risk across the two genes increased (represented by the number of risk alleles), the activity associated with executive attention decreased in the medial temporal lobe, as well as other regions (Green et al. 2014). On its own, the CLU AD-risk variant mediated connectivity differences in another fMRI study (Erk et al. 2011). Healthy carriers of the CLU risk variant showed decreased coupling of the hippocampus and prefrontal cortex during memory retrieval tasks (recall and recognition) (Erk et al. 2011) (Figure 22.5). Finally, the functional connectivity of the hippocampus and the relationship of this measure to the CLU polymorphism was recently reported in a study of resting state fMRI data. Compared to carriers of the protective allele, subjects who were homozygous for the CLU risk allele had the same general pattern of positive and negative functional connectivity, but the magnitude of the connectivity was stronger in both the positive and negative directions (Zhang et al. 2014). Taken together, these studies indicate that the BOLD signal as measured by fMRI may be modulated by CLU genotype. More studies are needed to confirm the association and define the dynamics of the modulation.
1.0 0.5 0.0 –0.5 –1.0
CC
CT
TT
Genotype Figure 22.5 Functional coupling of the hippocampus and frontal cortex in
three CLU genotype groups. Figure reproduced from Erk S, et al. (2011), The Journal of Neuroscience.
A. Carriers of the risk allele (C at rs11136000) show decreased, allele dose-dependent coupling of the right DLPFC and the right hippocampus during recall (Z = 5.06; p < 0.05, family-wise error corrected for multiple testing across the whole brain). Each red dot represents the effect size for one subject and reflects connectivity between right DLPFC and the right hippocampal seed region. B. Carriers of risk allele exhibit significantly decreased allele dosage-dependent coupling of the right DLPFC with the left hippocampus during recognition (Z = 4.73; p < 0.05, family-wise error corrected for multiple testing across the whole brain). Each red dot represents the effect size for one subject and reflects connectivity between right DLPFC and left hippocampal seed region. Reprinted with permission from The Society for Neuroscience.
region (rs3851179) was located upstream from PICALM, but subsequent studies have not only replicated this finding but also identified AD-risk SNPs within the PICALM gene itself (Naj et al. 2011). PICALM, like CLU, has widespread expression in the brain. It is involved in many cellular processes, especially the trafficking of proteins and lipids via clathrin mediated endocytosis (Tebar, Bohlander, and Sorkin 1999). Lately, this process, essential to synaptic transmission, has received increased attention in the study of AD, in part because of the strong association of PICALM to AD uncovered in GWA studies (Morgan 2011). In terms of reproducibility in GWA studies, PICALM ranks third after the APOE/TOMM40 locus and CLU ( Jun et al. 2010; Seshadri et al. 2010; Naj et al. 2011; Lambert et al. 2013). Perhaps because of this highly reproduced association, PICALM is fairly well represented in the neuroimaging genetics of AD literature.
3 7 6 part vii : I maging G enetics of C ognitive A ging
A study on older adults who ranged from cognitively healthy to diagnosed with AD reported a significant association between PICALM (rs3851179) and hippocampal volume such that carriers of the PICALM risk variant had lower hippocampal volume (Biffi et al. 2010). The authors also described a similar relationship between PICALM risk and entorhinal cortex thickness. This finding has been replicated in another study that found that the PICALM risk allele is associated with a thinner entorhinal cortex 191. However, in young adults, PICALM was not associated with either hippocampal or entorhinal cortex volume (Bralten et al. 2011). The functional connectivity of the hippocampus and the relationship of this measure to PICALM was recently reported in a resting state fMRI experiment. Compared to subjects who were homozygous for the protective allele, risk allele carriers showed weaker negative functional connectivity of the hippocampus to many regions (Zhang et al. 2014). This finding is preliminary and needs to be replicated. Lastly, a study of amyloid deposition as measured by florbetapir-PET found an epistatic effect involving PICALM and BIN1, another AD risk gene (Hohman, Koran, and Thornton-Wells 2013). This study is described later in this chapter (see BIN1).
CR1 Unlike CLU and PICALM, expression of complement component (3b/4b) receptor 1 (CR1) is likely to be low in the brain (Singhrao et al. 1999). The CR1 protein’s function is complex and varies by cell type, but it is generally involved in the regulation of the complement cascade, a major component of the innate immune system that helps to amplify the response of the immune system to potential targets. The CR1 protein is involved in transporting opsonized immune complexes through the circulatory system for removal (Morgan and Carrasquillo 2013). Neuroinflammation has been associated with AD for many years, but has often been dismissed as a consequence, not a cause, of the disease (Hensley 2010). This view is beginning to change, and inflammatory processes are being studied as potential pathogenic processes in AD (Morgan 2011). One reason that interest in neuroinflammation and AD has been renewed is the identification of an association between a polymorphism in CR1 and AD in a 2009 GWA study (Lambert et al. 2009). The CR1 risk variant has been shown to be associated with thinner entorhinal cortex in healthy older adults (Biffi et al. 2010). Interestingly, there is also evidence that CR1 is associated with lower entorhinal cortex volume in young
healthy adults, a finding that was confirmed in two independent cohorts (Bralten et al. 2011). Additional research is needed to assess whether or not this relationship between CR1 and a potential endophenotype of AD is reproducible in larger samples.
BIN1 Bridge integrator 1 (BIN1) was conclusively reported as a risk gene for AD in 2010, after borderline associations were reported in one of the large 2009 GWA studies (Harold et al. 2009; Seshadri et al. 2010). Like PICALM, BIN1 is associated with the intracellular trafficking of lipids and proteins. BIN1 encodes a protein that has at least 10 isoforms (Morgan and Carrasquillo 2013). Each isoform has specific domains that influence the function of the protein. The isoform of BIN1 that is specific to the brain contains the clathrin-associated protein-binding (CLAP) domain, which plays a role in clathrin-mediated endocytosis (Pant et al. 2009). Clathrin-mediated endocytosis is an essential process in synaptic vesicle recycling, which is a crucial component of efficient synaptic transmission. It is interesting that both PICALM and BIN1 are implicated as molecular components of this general cellular process, and it suggests that variability in synaptic transmission efficiency may contribute to AD pathology, especially during the early phase characterized by synaptic loss and neuronal death (Morgan 2011). The convergence of PICALM and BIN1 function led one group to test for epistatic genetic effects between the risk loci for each gene identified in GWA studies. Hohman and colleagues analyzed florbetapir-PET scans from older adults to test for a possible interaction effect of BIN1 and PICALM on amyloid deposition (Hohman, Koran, and Thornton-Wells 2013). The authors found that there was indeed an interaction and that this interaction was reproducible in a second data set. The BIN1 risk variant was related to higher levels of amyloid burden, but only in individuals who were carriers of the PICALM protective variant. This study is a simple illustration of the weakness of candidate gene studies because both BIN1 and PICALM are not related to amyloid deposition when tested on their own. There is, however, evidence that BIN1 genotype is directly associated with neuroimaging biomarkers of AD. Based on the preliminary, non-significant evidence from the 2009 GWA studies that BIN1 was associated with AD, Biffi and colleagues (2010) tested for an association of the BIN1 risk variant and a number of neuroimaging phenotypes. The authors found that the BIN1 risk variant is associated with
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
377
thinner temporal pole and entorhinal cortex in healthy older adults (Biffi et al. 2010). The same year that paper was published, a BIN1 locus reached genome-wide significance in a new AD GWA study (Seshadri et al. 2010).
EphA1 is associated with the likelihood of being amyloid positive, as measured by PiB-PET. In contrast to the findings for ABCA7, the authors described a decreasing risk of amyloid positivity for each C allele of EphA1 (rs11767557) (Hughes et al. 2014).
ABCA7 ATP-binding cassette, subfamily A, member 7 (ABCA7) is one gene in a group of highly conserved transmembrane transporters that participate in active transport of various substrates across membranes, both cellular and organelle membranes (Vasiliou et al. 2009). ABCA transporters have been linked to cholesterol and lipid homeostasis, and appear to work directly with APOE by transporting lipids out of the cell to APOE for clearance (Hirsch-Reinshagen et al. 2004). This coordination with APOE may hint at the mechanism of the association between ABCA7 and AD. Still, ABCA7 was first directly linked to AD through a GWA study in 2011 (Hollingworth et al. 2011). The ABCA7 locus has been implicated in only one very recent neuroimaging study in which the authors were interested in the relationship between cholesterol levels and amyloid deposition (Hughes et al. 2014). Hughes and colleagues (2014) described an over 2-fold increased risk of amyloid positivity in carriers of the ABCA7 (rs3752246) risk variant.
EphA1 EphA1 was originally named after the cell line it was discovered in, erythropoietin-producing human hepatocellular carcinoma (Morgan and Carrasquillo 2013). EphA1 is a member of a superfamily of proteins called the receptor tyrosine kinases and is expressed widely in multiple tissues including the brain (Hirai et al. 1987; Chen, Fu, and Ip 2012). The Eph-ephrin family of receptors and ligands are all membrane-bound proteins that are involved in cell adhesion and cell-cell contact mediated signaling, for example in axonal guidance during development (Chen, Fu, and Ip 2012). The link between EphA1 and AD was first described in two GWA studies in 2011 (Hollingworth et al. 2011; Naj et al. 2011). There are few clues as to the neurobiological processes that link this gene to AD. It is known that EphA receptors, as a class, are highly expressed in the hippocampus, but the expression and function of specifically EphA1 in the hippocampus is not well understood (Nakamura-Hirota et al. 2012). Neuroimaging genetics results describing EphA1 are limited to a single study. As they did for the ABCA7 genetic locus, Hughes and colleagues (2014) found that
CD33 Sialic acid binding immunoglobulin-like lectin-3 (CD33) is a membrane-bound receptor expressed on immune cells (Crocker, Paulson, and Varki 2007). It plays an important role in the differentiation of immature immune cells and the signaling of mature immune cells in the innate and adaptive immune system (Crocker, Paulson, and Varki 2007). Despite strong evidence from several GWA studies that CD33 is associated with AD, this association was not fully replicated in the second stage of the IGAP consortium GWA study (Hollingworth et al. 2011; Naj et al. 2011; Lambert et al. 2013). This may cast some doubt on the strength and reproducibility of this gene’s association with AD. Perhaps the association is specific to certain regions and genetic backgrounds. In any case, there is some preliminary work using neuroimaging to measure neural substrates of CD33 risk. A single study examined the relationship between CD33 genotype and a neuroimaging phenotype. Bradshaw and colleagues (2013) found that the risk variant of CD33 was associated with greater, diffuse amyloid deposition as measured with PiB-PET imaging. BDNF
Brain-derived neurotrophic factor (BDNF) is a growth factor that is widely expressed in the brain, including nearly all cortical areas. Through signaling with its main receptor tropomyosin-related kinase receptor type B (TRKB), BDNF is involved in regulating and supporting essential physiological functions in the adult brain. The literature supporting this assertion is vast and varied. For example, BDNF has been shown to modulate synaptic plasticity and dendritic spine dynamics and morphology (Tanaka et al. 2008). BDNF also supports long-term potentiation in the hippocampus, a process that is essential for learning and memory (Figurov et al. 1996). Furthermore, it has been suggested that BDNF plays an important role in energy homeostasis, providing a link between peripheral glucose metabolism and the brain that might be responsible for mediating the cognitive sequelae of physical exercise and a lack thereof (Marosi and Mattson 2014). On account of the role of BDNF in these crucial processes, therapeutic uses
3 7 8 part vii : I maging G enetics of C ognitive A ging
of synthetic BDNF are actively being explored in a variety of neurological and psychiatric disorders (Nagahara and Tuszynski 2011). For example, in several animal models of AD, ranging from mice to non-human primates, treatment with BDNF has been shown to recover lost synapses, promote normal cell signaling, and alleviate learning and memory deficits (Nagahara et al. 2009). Thus, despite the fact that BDNF has not been directly associated with AD incidence, there is certainly evidence that BDNF may mediate neuroprotective processes that may slow, or even reverse, some aging and AD-related neuronal changes (Von Bohlen und Halbach 2010). There is a common polymorphism of the gene that encodes BDNF that results in a methionine substitution for a valine at codon 66 of the protein (Val66Met). The BDNF Met variant is expressed at normal levels, but the secretion of the protein from neurons is decreased (Chen et al. 2008). A possible relationship between this polymorphism and neuroimaging endophenotypes of AD have been explored in a small number of studies. In a report by Lim and colleagues, healthy Met carriers who had high Aβ, as measured by PiB-PET, had higher rates of hippocampal volume loss over 3 years than Val/Val homozygotes (Lim et al. 2013). In addition, the Met carriers with high Aβ showed more dramatic decline in cognition, including measures of executive functioning and episodic memory. There were no differences between healthy Met carriers and Val/ Val homozygotes in the low Aβ group. These findings indicate that Met carrier status may help to predict decline in individuals who may be in the preclinical phase of AD (Lim et al. 2013). Another study found that Val66Met, as well as other SNPs within the BDNF gene, were associated with hippocampal and cortical atrophy over 2 years in a mixed cohort of healthy adults and patients with MCI and AD (Honea et al. 2013). The authors concluded that though BDNF was not associated with AD diagnosis, it is a factor in AD-related neurodegeneration measured by neuroimaging (Honea et al. 2013). In a study of healthy adults from age 19 to 82, Val66Met interacted with age to predict cortical thinning in the entorhinal cortex and adjacent temporal areas (Voineskos et al. 2011). The authors of this study also found an Val66Met–age interaction that predicted decreased FA in temporal white matter tracts, indicating a loss of white matter integrity in these regions (Voineskos et al. 2011). These studies, and others, indicate that BDNF variants may modulate the severity of AD-related changes to cortical and hippocampal morphology (Hashimoto et al. 2009). There is also some evidence that the Val66Met variant is related to glucose metabolism measured by FDGPET. One study found decreased metabolism in healthy
older Met carriers in the right parahippocampal gyrus and the superior temporal gyrus (Xu et al. 2010). The authors also report increased metabolism in healthy Met carriers in frontal regions. This pattern of differences was also observed in patients with MCI. Further studies examining BDNF variants and FDG-PET imaging are needed to replicate these results, as well as to help elucidate whether proposed connections between BDNF and peripheral metabolism extend to central glucose metabolism (Marosi and Mattson 2014). Finally, a recent study by Adamczuk and colleagues showed an interesting relationship between APOEε4 and Val66Met, such that APOEε4 carriers who were also Met carriers had higher Aβ load than APOEε4 carriers who were Val/Val homozygotes (Adamczuk et al. 2013). However, in APOEε4 non-carriers, there was no association between Aβ load and the Val66Met variant. These findings point to a potential interaction between APOEε4 and BDNF Met variant. Better characterization of this relationship, as well as the other findings described in this section, will be essential to developing potential BDNF-based therapies for use in the AD. NEUROIMAGING-IDENTIFIED ALZHEIMER’S DISEASE RISK GENES
In addition to APOE, TOMM40, TREM2, and GWAstudy-identified AD-risk loci, there are genes that have been identified as potential risk factors for AD through the use of human neuroimaging. This represents a reversal of the types of studies we have covered thus far in this chapter, in which genetic associations are discovered via epidemiological studies, molecular biology experiments, linkage analyses, or GWA studies of a disease state, and then the effects of these genetic variants are studied in living humans using neuroimaging. In contrast, the following studies use neuroimaging and creative experimental design in order to search for genetic associations with disease biomarkers, such as hippocampal atrophy. These studies are examples of some of the most exciting work materializing from the field of neuroimaging genetics because they are an example of true cooperation of human genetics methods and neuroimaging biomarkers/ endophenotypes. In a creative experiment by Nho and colleagues (2013), a small set of subjects from the Alzheimer Disease Neuroimaging Initiative (ADNI) who had experienced either extreme or very little hippocampal atrophy over 2 years were selected and their exomes were sequenced. Often, the loci identified in genetic association studies are not coding variants, but exome sequencing ensures that any
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
379
associations one finds will be a potentially functional variant. The authors isolated 57 SNP variants that were found in all rapid atrophy subjects and in none of the slow atrophy subjects. Next, they used these SNPs and performed a quantitative trait analysis on a separate, larger cohort of subjects homozygous for the APOEε3 allele. Two genes, PARP1 and CARD10, were associated with the rate of hippocampal atrophy in the larger validation group. While further research is required to assess the reproducibility and clinical utility of these results, this cross-discipline design utilizes large cohort enrollment efforts by selecting for extreme cases, advancing human genetics methods for exome sequencing and neuroimaging for measurement of a potential biomarker. We believe that experiments like this will become more common as neuroimaging genetics methods are further integrated into mainstream biomedical research. A study by Shen and colleagues (2010) also used ADNI data in order to perform many GWA studies using a different neuroimaging measure as the target phenotype in each case. The authors used voxel-based morphometry to define cortical gray matter volume, as well as the volume of 43 ROIs in each hemisphere. Freesurfer was also used to calculate cortical thickness and volume measures for ROIs across the cortical mantle. After an iterative GWA study was performed for each of these potential endophenotypes, it was not surprising that APOE and TOMM40 were associated with several ROIs, including bilateral hippocampus and amygdala, volume of right cerebral cortex, and cortical thickness of left cerebral cortex. Several additional genes were implicated, including EphA4, TP63, and NXPH1. Further analyses focused on the locus proximal to NXPH1 showed that subjects homozygous for the putative risk allele had significantly reduced bilateral hippocampal gray matter density (Shen et al. 2010). Another area of interest in neuroimaging genetics involves testing for the association of functional pathways with neuroimaging measures (Meda et al. 2012; Silver et al. 2012). In other words, based on previous understanding of protein interactions and signaling, genes that encode proteins in a given biological pathway can be assessed for an aggregate association with a phenotype. Putative gene pathways are available in a variety of databases, including the Molecular Signatures Database (http://www.broad institute.org/gsea/msigdb/index.jsp). In a study by Silver and colleagues (2012), the authors present a new statistical method for testing biological pathway associations called sparse reduced-rank regression. Using this method and a voxel-wise measurement of structural change intended to maximize the difference in trajectories between normal
controls and AD patients, the authors found that insulin signaling, vascular smooth muscle contraction, and focal adhesion pathways were associated with AD-related changes. They then took their analyses a step further and tried to identify single SNPs that might be driving the association of these pathways and found nearly 10 candidates. The authors also found that APOE, TOMM40, and CR1 were associated with their voxel-wise endophenotype, suggesting that it captures disease-related structural changes in AD (Silver et al. 2012). Thus far, genetic risk factors for AD have predominantly been identified in large cohorts of white American or European individuals, with relatively little effort made to replicate findings in different ethno-racial groups. In a 2012 study by Melville and colleagues, a specific aim of their research was to identify SNPs that associate with neuroimaging phenotypes in both Caucasian and African American cohorts. The neuroimaging endophenotypes they chose to query were total cerebral volume, hippocampal volume, and the volume of white matter hyperintensities. SNP associations were evaluated in a 2-stage reproducible analysis, a common study design in GWA studies. The unique feature of this study was that the stage 1 cohorts were Caucasian cohorts and the stage 2 cohort was composed of African American individuals. Thus, they reported only genes that were associated with a given phenotype in these two different ethno-racial groups. The authors found that loci within APOE, as well as F5/SELP, LHFP, and GCFC2, were associated with hippocampal volume. In addition, they reported that two different SNPs, both in the PICALM gene, were associated with hippocampal volume in the Caucasian and African American cohorts (Melville et al. 2012). This finding in PICALM is very interesting because it indicates that specific risk variants may differ between ethnic or racial groups, even within the same risk gene. Recently, a new, potential dementia-associated gene, SPON1, was shown to be associated with the density of structural connections between the left posterior cingulate gyrus and the left superior parietal lobe, such that two copies of the minor allele were associated with increased structural connectivity ( Jahanshad et al. 2013). Before this relationship was uncovered, the authors used their cohort of young adult twins to assess the heritability of a structural connectome created using tractography and a common cortical parcellation. This was an important step because it indicated that portions of the structural connectome were heritable enough to perform genome-wide association scanning. After SPON1 was implicated as a genetic factor mediating structural connectivity between two nodes, the authors tested the relationship between the putatively
3 8 0 part vii : I maging G enetics of C ognitive A ging
protective SPON1 variant and the morphology of the brain, finding associations with larger posterior cingulate cortex volume and smaller ventricular size. Furthermore, the minor allele was associated with milder dementia as measured by the Clinical Dementia Rating scale ( Jahanshad et al. 2013). The publications discussed in this section are examples of neuroimaging genetics studies that aim to identify genetic risk loci associated with neuroimaging biomarkers or endophenotypes. This requires authors to perform interdisciplinary work that spans the fields of human genetics and neuroimaging. It is likely that neuroimaging genetics will move further into this interdisciplinary space in the future, with fewer “candidate gene” studies and more experiments like those reviewed here. In addition, efforts to develop new methods to statistically test the association of many genetic risk factors as a single polygenic risk score or metric are sure to be a major focus of the neuroimaging genetics field moving forward. Unfortunately, the sample sizes of these kinds of studies are lagging behind those that have been achieved in AD GWA studies. Although there is evidence that GWA studies of neuroimaging endophenotypes are more statistically efficient than GWA studies based on behavior, sample sizes still need to grow in order to ensure generalizability and reproducibility (Rose and Donohoe 2013). The smaller sample sizes are due in part to difficulties in combining neuroimaging datasets caused by differences in acquisition parameters, processing, and inclusion/exclusion criteria that may produce a confounding effect in a given neuroimaging measure. However, data-sharing efforts that focus on resolving these issues with standardized protocols are gaining ground. One effort that is specifically focused on neuroimaging genetics is the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) project (Thompson et al. 2014). The aim of the ENIGMA project is to increase the ability of neuroimaging genetics researchers to share their own data, as well as access others’ data in order to increase statistical power in their own studies. FA M I LY H I S T O R Y O F A L Z H E I M E R ’ S D I S E A S E
Thus far, we have discussed neuroimaging genetics approaches that focus on either a known genetic cause of AD, a known genetic risk factor for AD, or identifying new genetic associations. Another approach in neuroimaging genetics of AD is to study the effect of a family history of the disease. A positive family history of AD may indicate that a subject carries genetic risk factors for AD perhaps even above and beyond the known susceptibility loci that were covered in previous sections.
It can be thought of as a composite genetic risk factor that may reflect susceptibility conferred by both known and unknown risk genes (Donix et al. 2012). Indeed, a family history of AD confers a strong predisposition to the disease, doubling one’s chances of developing AD (Canadian Study 1994). In addition, the risk for AD captured by family history cannot be explained by APOE alone. Several studies have found additive effects of family history and APOE (Donix et al. 2012). This supports the idea that family history is a composite measure and cannot be explained or supplanted by even the strong genetic association of APOE to AD. A positive family history of AD is associated with higher rates of thinning in the hippocampus, especially in the entorhinal cortex and subiculum, in cognitively normal, older subjects (Donix et al. 2010a). Family history has also been linked to hippocampal volume in middle-aged adults, specifically in the left hippocampus (Okonkwo et al. 2012). Furthermore, cognitively normal, older subjects with a family history of AD show more severe whole-brain gray matter volume loss than subjects with a negative family history (Honea et al. 2011). Interestingly, gray matter volume loss was significant in the precuneus, parahippocampal gyrus, and the hippocampus when subjects with only a maternal history of AD were compared to the negative family history group or to the paternal family history group (Honea et al. 2011). This indicates that phenotypic differences ascribed to subjects with a family history of AD might actually be driven by subjects with a maternal history of AD. Evidence in the literature supports the theory that a maternal history of AD results in more severe changes in structural endophenotypes than a paternal history of AD (Honea et al. 2010; Berti et al. 2011). There is also evidence that a maternal history of AD is related to Aβ load. PiB tracer uptake in people with a maternal history of AD reveals significantly more Aβ in parietal cortex, precuneus, posterior cingulate, and sensorimotor cortex when compared to subjects with a paternal history of AD (Honea et al. 2012; Mosconi et al. 2013). The mechanism of this relationship between maternal family history of AD and risk for the disease is not known. However, it has been posited that maternal-lineage inheritance of mitochondrial DNA may play an important role (Mosconi et al. 2011). In fact, there is preliminary evidence that adult children whose mothers have AD show mitochondrial dysfunction and reduced cytochrome oxidase activity compared to subjects with an affected father and to controls with a negative family history (Mosconi et al. 2011).
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
381
In a recent study by Mosconi and colleagues (2014), subjects with a family history of AD were further subdivided into groups of individuals who had a maternal history of AD, a paternal history of AD, or both a maternal and paternal history of AD. These three groups were then each compared to a reference group of age and sex-matched subjects who had a no family history of AD. The authors found that the subjects who had a history of AD in both their maternal and paternal lineages showed more severe alterations in all of the neuroimaging phenotypes they measured. Specifically, these subjects had higher retention of the PiB tracer, indicative of higher Aβ load, as well as lower FDG uptake, indicative of hypometabolism. In addition, subjects with a history of AD on both sides of their family had more severe gray matter volume reductions across the cortex (Mosconi et al. 2014). When the authors examined the subjects with only a maternal history of AD, they found intermediate phenotypes, followed by subjects with only a paternal history. Their findings are in line with others that reported more severe changes in neuroimaging endophenotype measures in people with a maternal family history of AD. The discovery that there is an additional additive effect of a maternal and paternal history of AD is a novel finding that will certainly be important in designing studies of family history of AD in the future. As we have discussed, DMN connectivity is altered in FAD and in APOEε4 carriers of varying ages. There is also evidence that DMN connectivity is modulated by family history of AD (Fleisher et al. 2009; Wang et al. 2012). One study found that subjects with a family history of AD (they did not specify maternal or paternal lineage) had reduced connectivity between the posterior cingulate and the medial temporal lobe (Wang et al. 2012). Another paper describes a direct comparison of the ability of task-based fMRI and resting state fMRI to differentiate between two groups stratified by AD risk: subjects with a family history of AD and at least one copy of APOEε4 versus subjects with no family history and no APOEε4 alleles (Fleisher et al. 2009). Comparing DMN average connectivity between the two groups was the best differentiator, accounting for 62% of the variance due to riskgroup membership compared to only 25% accounted for when comparing task-related fMRI activations (Fleisher et al. 2009). The consistency of the DMN disruption findings across different cohorts representing preclinical AD or risk for AD indicates that DMN functional connectivity may be a uniquely viable fMRI-based endophenotype for AD.
F U T U R E DIR E C T IONS AND C H ALLE NGE S The challenges associated with the neuroimaging genetics field apply to neuroimaging genetics of AD with one major exception. As described earlier, APOE accounts for a large amount of the genetic variance of AD, more than any single genetic locus in another human polygenic neurological disorder. Because of this, there is a particularly large body of work using APOE in “candidate gene”–type neuroimaging experiments. The sheer volume of these studies, although not all in agreement, is indicative of the unique position of APOE in the field of human genetics. In theory, because APOE is soaking up a good portion of heritability variance in AD, it is possible that polygenic risk modeling will be easier in AD than in other common polygenic diseases. This makes AD an attractive disease for neuroimaging genetics researchers. Indeed, attempts to model multiple genetic risk factors in neuroimaging studies of AD have showed promise, predicting conversion from MCI to AD as well as cortical thickness changes in AD-vulnerable regions (Sabuncu et al. 2012; RodríguezRodríguez 2013). The causal variants that give rise to the APOEε4 allele, which then confers increased risk for AD, are known polymorphisms at rs429358 and rs7412. Variants at these sites alter the structure and function of the APOE protein (Mahley, Weisgraber, and Huang 2009). In fact, so-called APOEε4 “structure correctors,” which make APOEε4 behave like the more common APOEε3, are currently being developed as a possible treatment for AD (Chen, Liu, et al. 2012). In contrast, many of the GWA-identified AD risk loci (CLU, BIN1, ABCA7, EphA1) are located in intronic (CLU, ABCA7) or intragenic (BIN1, EphA1) regions, with no evidence that variants affect protein structure or function. In general, an intragenic region may play some regulatory function, but in the cases of EphA1 and BIN1 there is little evidence of conservation of these intragenic regions, therefore making a regulatory role in genetic expression unlikely (Morgan and Carrasquillo 2013). Thus, the search for the causal variants is still ongoing for these genes and, also, for genes implicated in other common disease by GWA studies (Cirulli and Goldstein 2010). Ostensibly, the causal variant for one of these genetic risk loci will be a polymorphism in high linkage disequilibrium with the GWA locus that affects the function of the gene’s protein product in some way. It is possible that the polymorphisms driving the signal of these GWA study associations are rare variants occurring in less than 5% of people (minor allele frequency < 0.05) (Cirulli and Goldstein 2010). If this is the case, large sample sizes in GWA studies will increase our ability to detect rare
3 8 2 part vii : I maging G enetics of C ognitive A ging
polymorphisms associated with disease. Still, it remains to be seen if the underlying genetics of a common disease like AD will be best described as the coincidence of several strongeffect rare variants or of many low-effect common variants. In either case, until the casual variants of GWA-study-identified risk loci are identified, the utility of these risk factors in targeted efforts like drug development is minimal. From the perspective of neuroimaging, there are advantages and disadvantages to the rare-variant or common-variant theory of AD genetics. Obviously, because rare variants occur in so few individuals, it would be difficult to amass a large enough cohort of carriers to produce statistically significant results. However, the field is moving fast toward larger and larger data sets through data-sharing efforts and multi-center study designs. Access to ever-expanding reservoirs of data may mean that reasonably sized samples of individuals with specific rare variants may be plausible. The great advantage of studying rare variants with neuroimaging is that the effect size of these rare variants is likely to be much larger than common variants, likely making differences between carrier groups easier to detect, even at smaller sample sizes. In contrast, methods for modeling multiple genetic risk factors in a single experiment are actively being developed and may help to exploit the synergistic predictive power of many low-effect-size common variants. As a final note on GWA-study-identified genetic risk factors for AD, it is important to recognize that the 20 loci discussed in this chapter were identified using large cohorts of Caucasian European or American subjects. There are many reasons the genetic loci implicated by these studies might fail to replicate in a cohort of subjects from a different ethnic background, including population-specific variants, differing patterns of linkage disequilibrium, or even a heterogeneous genetic basis of AD in different ethnic groups. As an illustrative example, many small GWA studies have tried to replicate the association of CLU with AD in non-Caucasian cohorts. The results of these studies tell us that there appears to be an association between CLU and AD in Chinese cohorts, but not in cohorts of non-white Americans or Europeans ( Jun et al. 2010; Yu et al. 2010; Chen, Kao, et al. 2012). Clearly, this is a limitation of the published large GWA studies in AD, and a greater effort must be made to amass comparably large samples of different ethnoracial groups for study. It is possible that this effort may result in the identification of certain genes that are associated with AD regardless of genetic background, and that these genes could then be the focus of increased research resources due to their greater generalizability. In addition, further exploration of the genetic basis of AD
in people of African and Hispanic descent may help elucidate epidemiological differences observed in these ethnic groups, including higher incidence and earlier onset of AD (Gurland et al. 1999). Another challenge in the field is the predominant use of cross-sectional experimental designs in trying to elucidate the pathophysiological trajectory of AD. Given the importance of early detection in neurodegenerative diseases, as well as the published associations of various AD risk genes with differences in brain structure and function in young people (even children and infants), it is clear that longitudinal mapping of disease progression is essential in the fight against AD. A better understanding of how the disease manifests in individuals, each with his or her own unique genetic and environmental risk profile, would help clinicians detect preclinical AD. As described earlier, preclinical AD, or the phase of AD before changes in behavior and symptoms emerge, is believed to provide the best opportunity for treatment, especially with a progression-halting drug. Such a drug is not available yet, but accurate definition of preclinical AD will be crucial to the success of any candidates. So how do we design experiments to study AD risk and preclinical AD? Overwhelmingly, inferences about the trajectory of AD are made from cross-sectional studies in which data are collected from each subject only once, and all the subjects are randomly distributed across the age range under investigation. This approach is problematic because cross-sectional studies are excellent at confounding between-subject and within-subject variation (Thompson, Hallmayer, and O’Hara 2011). In other words, in a cross-sectional study, one loses the ability to separate differences mediated by normal variation in a given subject from variation across two different subjects, or cohorts of subjects. Thus, drawing longitudinal conclusions based on cross-sectional evidence, even based on many cross-sectional studies, is precarious and should be done cautiously (Kraemer et al. 2000). Longitudinal designs are better for making inferences about disease trajectory but they are difficult in practice. Still, multi-cohort longitudinal designs are feasible in today’s pro-collaboration atmosphere because many sites can collect longitudinal data on a relatively small number of subjects and then, assuming that proper standardization and oversight is in place, these subjects can be combined to create a much larger cohort. Indeed, ADNI is a good example of this type of effort in neuroimaging genetics of AD. We believe that future efforts to define the pathophysiological trajectory of AD through neuroimaging genetics should follow this example of a multi-cohort longitudinal design.
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
383
R EL EVA NC E AN D I MPACT Alzheimer’s disease affects more than 13% of individuals aged 65 years and older, a subset of the population that is rapidly growing across the world. According to a recent report by the US Census Department, by the benchmark year of 2050 there will be nearly 84 million people aged 65 years and older, leaving at least 13 million individuals suffering from AD (Herbert et al. 2003; Census Projections 2012). The number is likely to be higher, as the risk for AD grows proportionately with age and the fraction of the above-65 set who are the oldest old (say, over 95 years, when over 40% of individuals have AD) is continuously growing as people live longer and longer lives (Fratiglioni et al. 1999). Although the focus of this chapter is on neuroimaging genetics of AD, age is the strongest risk factor for AD, so demographic trends in aging are an important factor to weigh in understanding the future need for AD research. AD is a debilitating disease that, especially when combined with other age-related health struggles, can require years of part- or full-time care for a single patient. The economic impact of this looming need for elder care providers is difficult to fully grasp. There is, of course, the high cost of professional care, either in the home or in an institution, which is prohibitive for many older Americans. There is also the economic burden that families will take on to care for aging relatives. A large proportion of elder care will be provided wage-free by adult children who, in order to be available to ailing parents, may be forced to leave jobs or dip into their own savings. The predictions for the future of AD in the United States, and indeed across the world, paint a dark picture in which advancing medical care, public health interventions and programs, and improved healthcare literacy will lead to more and more people reaching extreme old age with nothing to protect them from the reality of high AD incidence. Imagine a scenario where, in the future, an individual could undergo a battery of non-invasive tests, including cognitive testing, a blood draw for genetic profiling, and MR and/or PET imaging, in order to generate a report or panel detailing the likelihood that he or she will go on to develop AD. What if that report could estimate with a high certainty the age of onset? Of course, these revelations in the absence of effective treatment would lead to a situation similar to the current state of Huntington’s disease (HD) diagnosis. Many at risk for HD choose to learn what their genetic fate is, but many do not (Dufrasne et al. 2011). The latter group sees no benefit to knowing their fate in advance,
especially if that fate is to die of a terrible degenerative disease. Genetic counseling is a crucial aspect of these difficult decisions. But even with genetic counseling and support, learning that you will one day develop an unpreventable and untreatable disease can hardly be argued as universally empowering knowledge. However, the ability to make such predictions and then adjust the predictions when they fall short is an important part of clinical trial design. In other words, the better we are at estimating, on an individual basis, AD risk or AD clinical stage, the better we will be at enrolling a homogenous group in treatment trials (see Figure 22.6 for schematic). Phase 3 AD treatment trials in humans have almost exclusively failed, even after very promising data in model organisms and in earlier trial phases (Doody et al. 2014; Salloway et al. 2014). A potential reason for this high failure rate is the highly heterogenous nature of the subjects being enrolled in these trials. One problem is neuropathological variation. For example, of the people with AD who come to autopsy, up to 75% of those patients also have vascular pathology severe enough to have contributed to their dementia syndrome (Toledo et al. 2013). Before neuropathological processing, it is extremely difficult to differentiate so-called “pure AD” from mixed AD and vascular disease. Furthermore, a clinical diagnosis of AD corresponds to a neuropathological diagnosis of AD (pure or mixed pathology) about 80% of the time (Struyfs et al. 2014). These leaves 20% of clinically diagnosed AD subjects who in fact had another disease entirely, like frontotemporal lobar degeneration (FTLD) or corticobasal degeneration (CBD). It is not unreasonable to assume that subjects with each of these diseases, from pure AD and mixed AD pathology to FTLD and CBD, will respond differently (if at all) to treatments that target a single molecular species, like Aβ oligomers or plaques. Therefore, a concerted effort must be made to minimize incorrect clinical-pathological diagnoses in subjects enrolled in clinical trials. Unfortunately, the only way to make a pathological diagnosis is by examining the brain tissue at autopsy. Luckily, PET imaging allows clinicians and researchers to shed some light on what is inside the black box. Using PET imaging of Aβ and tau as a prescreening technique in clinical trials, while expensive, may increase our ability to amass a pathologically homogeneous cohort. Indeed, neuropathological prescreening using PET imaging has just been implemented for the first time as part of the Anti-Amyloid Treatment in Asymptomatic AD (A4) trial, which is requiring a positive Aβ florbetapir-PET scan for enrollment into the treatment arm of the trial (Sperling et al. 2014). In addition to heterogeneity of neuropathology in clinical trial subjects, we must also consider the heterogeneity of
3 8 4 part vii : I maging G enetics of C ognitive A ging
Clinical Trial Design
Characterization of Preclinical AD
Neuroimaging Genetics: • Preclinical participants • Prescreening techniques → Neuropathology → Genetic Risk • Non-cognitive endpoints • Minimize placebo effect
Neuroimaging Genetics: • Genetic risk quantification • Longitudinal volumetric changes in hippocampus • Amyloid and tau PET imaging • Signature of DMN dysfunction Increase Sensitivity
Match treatment strategies with design and screening procedures for well-powered and controlled trials
Treatment Strategies • Novel drug development • Drug repurposing • Nonpharmacological interventions Increase Effect Sizes
Effective Prevention
Figure 22.6 Neuroimaging genetics in the path to Alzheimer’s disease prevention.
Neuroimaging genetics research characterizing the preclinical phase of AD is aimed at improving our understanding of the pathophysiology of AD, specifically the prodrome that precedes memory loss (and other cognitive symptoms). Neuroimaging genetics will empower clinical trials by informing enrollment procedures, increasing the ability to enroll participants who have preclinical AD, thus increasing AD incidence in the cohort. When potential treatments come to clinical trial, neuroimaging genetics approaches will play a critical role in prescreening procedures and end point (outcome) definition.
the underlying genetics in each individual subject (Figure 22.6). It is not a particularly new idea that genetic variation can predispose individuals to respond well or less well to pharmacological treatments (Roden and George 2002). One well-validated example of this is the connection between genetic variation and the efficacy of antidepressants, a class of drugs targeted to the brain. Specifically, variants in the FKBP5 gene have been linked to better response to antidepressants, especially when given in combination (Kirchheiner et al. 2008). FKBP5 is a co-chaperone of the glucocorticoid receptor, which is involved in hypothalamic-pituitary-adrenal axis activation, a pathogenic state for depression (Kirchheiner et al. 2008). Given these findings, it is reasonable to assume that there may be variation in drug response in trial participants with different genetic risk factors for AD. In order to control for as many of these genetic variables as possible, clinical trials should consider implementing genetic prescreening measures that select for participants that have certain genetic risk factors for AD. An interesting study by Kohannim and colleagues (2013) tested the theory that a genetic prescreening protocol would decrease the sample size necessary to detect a treatment effect. In other words, they were interested in understanding how homogenizing the genetic risk profile of trial participants in favor of higher risk would affect the statistical power of a hypothetical trial. Specifically, the authors ranked 394 cognitively healthy and MCI ADNI subjects in order of decreasing genetic risk score, calculated
based on multiplying risk alleles for APOE, CLU, CR1, and PICALM by the logarithm of the odds ratios reported for each gene in GWA studies. They found that by selecting only the top 15% of subjects in order of highest genetic risk, the required sample size to show differences in temporal lobe atrophy decreased from 142 to 69 (Kohannim et al. 2013). This provides excellent preliminary evidence that genetic prescreening would increase statistical power in trials. Binning participants by genetic risk may very well be the next frontier in AD clinical trial design. Let us return to the hypothetical report detailing one’s AD risk and prognosis. The measures that will comprise that report are minimally invasive cognitive testing, genetic sequencing, CSF analysis, and neuroimaging. These are the tools we have available when studying living humans. The field of neuroimaging genetics of AD is performing research that explicitly combines two of these data types, and often incorporates the others. This research has and will continue to lead to the insights needed in order to prescreen participants for clinical trials, increasing the ability of those trials to detect the effect of a useful drug. Another important role for neuroimaging genetics in the fight against AD is the development of hard (non-cognitive) endpoints to assess treatment efficacy in clinical trials (Figure 22.6). Most AD trials to date have used soft endpoints, like paper-and-pencil memory measures or composite dementia severity scores (Doody et al. 2014; Salloway et al. 2014). These soft endpoints are particularly vulnerable to confounding effects,
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
385
such as the placebo effect and the within-subject variance (the “good day, bad day” phenomenon). The proximal goal of an individualized AD risk report based on genetics, neuroimaging biomarkers, and other measures is the pre-selection of clinical trial participants (Figure 22.6). The distal goal is to provide detailed prognoses in the clinic, combined with effective treatment. Neuroimaging genetics research in AD will play an essential role in achieving these goals.
R EF ERENCES Adamczuk K, De Weer A-S, Nelissen N, et al. (2013). Polymorphism of brain derived neurotrophic factor influences β amyloid load in cognitively intact apolipoprotein E ε4 carriers. NeuroImage Clin. 2: 512–520. doi: 10.1016/j.nicl.2013.04.001. Adamson MM, Hutchinson JB, Shelton AL, Wagner AD, Taylor JL. (2011). Reduced hippocampal activity during encoding in cognitively normal adults carrying the APOE ɛ4 allele. Neuropsychologia. 49(9): 2448–2455. doi: 10.1016/j.neuropsychologia.2011.04.022. Alexander GE, Bergfield KL, Chen K, et al. (2012). Gray matter network associated with risk for Alzheimer’s disease in young to middle-aged adults. Neurobiol Aging. 33(12): 2723–2732. doi: 10.1016/j.neurobiolaging.2012.01.014. Alexander AL, Lee JE, Lazar M, Field AS. (2007). Diffusion tensor imaging of the brain. Neurotherapeutics. 4(3): 316–329. doi: 10.1016/j. nurt.2007.05.011. Alexander GE, Saunders AM, Szczepanik J, et al. (1997). Relation of age and apolipoprotein E to cognitive function in Down syndrome adults. Neuroreport. 8(8): 1835–1840. Available at: http://www. ncbi.nlm.nih.gov/pubmed/9223061. Accessed October 1, 2014. Alzheimer A, Stelzmann RA, Schnitzlein HN, Murtagh FR. (1995). An English translation of Alzheimer’s 1907 paper, “Uber eine eigenartige Erkankung der Hirnrinde”. Clin Anat. 8(6): 429–431. doi: 10.1002/ ca.980080612. Apostolova LG, Hwang KS, Medina LD, et al. (2011). Cortical and hippocampal atrophy in patients with autosomal dominant familial Alzheimer’s disease. Dement Geriatr Cogn Disord. 32(2): 118–125. doi: 10.1159/000330471. Bateman RJ, Xiong C, Benzinger TLS, et al. (2012). Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N Engl J Med. 367(9): 795–804. doi: 10.1056/NEJMoa1202753. Beacher F, Daly E, Simmons A, et al. (2010). Brain anatomy and ageing in non-demented adults with Down’s syndrome: an in vivo MRI study. Psychol Med. 40(04): 611–619. Available at: http://journals.cambridge. org/abstract_S0033291709990985 (accessed October 1, 2014). Bekris LM, Lutz F, Yu C-E. (2012). Functional analysis of APOE locus genetic variation implicates regional enhancers in the regulation of both TOMM40 and APOE. J Hum Genet. 57(1): 18–25. doi: 10.1038/jhg.2011.123. Benitez A, Fieremans E, Jensen JH, et al. (2014). White matter tract integrity metrics reflect the vulnerability of late-myelinating tracts in Alzheimer’s disease. NeuroImage Clin. 4: 64–71. doi: 10.1016/j. nicl.2013.11.001. Bennett DA, Schneider JA, Wilson RS, Bienias JL, Arnold SE. (2004). Neurofibrillary tangles mediate the association of amyloid load with clinical Alzheimer disease and level of cognitive function. Arch Neurol. 61(3): 378–384. doi: 10.1001/archneur.61.3.378. Bergem AL, Engedal K, Kringlen E. (1997). The role of heredity in late-onset Alzheimer disease and vascular dementia. A twin study.
Arch Gen Psychiatry. 54(3): 264–270. Available at: http://www.ncbi. nlm.nih.gov/pubmed/9075467 (accessed August 26, 2014). Bertram L, Lange C, Mullin K, et al. (2008). Genome-wide association analysis reveals putative Alzheimer’s disease susceptibility loci in addition to APOE. Am J Hum Genet. 83(5): 623–632. doi: 10.1016/j. ajhg.2008.10.008. Bertram L, Tanzi RE. (2012). The genetics of Alzheimer’s disease. Prog Mol Biol Transl Sci. 107: 79–100. doi: 10.1016/B978-012-385883-2.00008-4. Berti V, Mosconi L, Glodzik L, et al. (2011). Structural brain changes in normal individuals with a maternal history of Alzheimer’s. Neurobiol Aging. 32(12): 2325.e17–26. doi: 10.1016/j.neurobiolaging.2011.01.001. Biffi A, Anderson CD, Desikan RS, et al. (2010). Genetic variation and neuroimaging measures in Alzheimer disease. Arch Neurol. 67(6): 677–685. doi: 10.1001/archneurol.2010.108. Bondi MW, Houston WS, Eyler LT, Brown GG. (2005). fMRI evidence of compensatory mechanisms in older adults at genetic risk for Alzheimer disease. Neurology. 64(3):501–508. doi: 10.1212/01. WNL.0000150885.00929.7E. Bookheimer S, Burggren A. (2009). APOE-4 genotype and neurophysiological vulnerability to Alzheimer’s and cognitive aging. Annu Rev Clin Psychol. 5: 343–362. doi: 10.1146/annurev.clinpsy.032408.153625. Bookheimer SY, Strojwas MH, Cohen MS, et al. (2000). Patterns of brain activation in people at risk for Alzheimer’s disease. N Engl J Med. 343(7): 450–456. doi: 10.1056/NEJM200008173430701. Borghesani PR, Johnson LC, Shelton AL, et al. (2008). Altered medial temporal lobe responses during visuospatial encoding in healthy APOE*4 carriers. Neurobiol Aging. 29(7): 981–991. doi: 10.1016/j. neurobiolaging.2007.01.012. Boyacioğlu R, Schulz J, Müller NCJ, Koopmans PJ, Barth M, Norris DG. (2014). Whole brain, high resolution multiband spin-echo EPI fMRI at 7 T: a comparison with gradient-echo EPI using a color-word Stroop task. Neuroimage. 97:142–150. doi: 10.1016/j. neuroimage.2014.04.011. Bradshaw EM, Chibnik LB, Keenan BT, et al. (2013). CD33 Alzheimer’s disease locus: altered monocyte function and amyloid biology. Nat Neurosci. 16(7): 848–850. doi: 10.1038/nn.3435. Bralten J, Franke B, Arias-Vásquez A, et al. (2011). CR1 genotype is associated with entorhinal cortex volume in young healthy adults. Neurobiol Aging. 32(11): 2106.e7–11. doi: 10.1016/j.neurobiolaging.2011.05.017. Braskie MN, Jahanshad N, Stein JL, et al. (2011). Common Alzheimer’s disease risk variant within the CLU gene affects white matter microstructure in young adults. J Neurosci. 31(18): 6764–6770. doi: 10.1523/JNEUROSCI.5794-10.2011. Braskie MN, Medina LD, Rodriguez-Agudelo Y, et al. (2012). Increased fMRI signal with age in familial Alzheimer’s disease mutation carriers. Neurobiol Aging. 33(2): 424.e11–21. doi: 10.1016/j.neurobiolaging.2010.09.028. Braskie MN, Medina LD, Rodriguez-Agudelo Y, et al. (2013). Memory performance and fMRI signal in presymptomatic familial Alzheimer’s disease. Hum Brain Mapp. 34(12): 3308–3319. doi: 10.1002/hbm.22141. Brown JA, Terashima KH, Burggren AC, et al. (2011). Brain network local interconnectivity loss in aging APOE-4 allele carriers. Proc Natl Acad Sci U S A. 108(51): 20760–20765. doi: 10.1073/ pnas.1109038108. Burggren AC, Zeineh MM, Ekstrom AD, et al. (2008). Reduced cortical thickness in hippocampal subregions among cognitively normal apolipoprotein E e4 carriers. Neuroimage. 41(4): 1177–1183. doi: 10.1016/j.neuroimage.2008.03.039. Cady J, Koval ED, Benitez BA, et al. (2014). TREM2 variant p.R47H as a risk factor for sporadic amyotrophic lateral sclerosis. JAMA Neurol. 71(4): 449–453. doi: 10.1001/jamaneurol.2013.6237. Canadian Study of Health and Aging. (1994). The Canadian Study of Health and Aging: risk factors for Alzheimer’s disease in Canada. Neurology. 44(11): 2073–2073. doi: 10.1212/WNL.44.11.2073.
3 8 6 part vii : I maging G enetics of C ognitive A ging
Carrasquillo MM, Morgan K. (2012). Commentary on “Functional analysis of APOE locus genetic variation implicates regional enhancers in the regulation of both TOMM40 and APOE.” J Hum Genet. 57(1): 3–4. doi: 10.1038/jhg.2011.135. Cash DM, Ridgway GR, Liang Y, et al. (2013). The pattern of atrophy in familial Alzheimer disease: volumetric MRI results from the DIAN study. Neurology. 81(16): 1425–1433. doi:10.1212/ WNL.0b013e3182a841c6. Census Projections by Age Group. Available at: http://www.census.gov/ population/projections/data/national/2012/summarytables.html. Chen Z-Y, Bath K, McEwen B, Hempstead B, Lee F. (2008). Impact of genetic variant BDNF (Val66Met) on brain structure and function. Novartis Found Symp. 289: 180–188; discussion 188–195. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=27 35856&tool=pmcentrez&rendertype=abstract (accessed October 19, 2014). Chen Y, Fu AKY, Ip NY. (2012). Eph receptors at synapses: implications in neurodegenerative diseases. Cell Signal. 24(3): 606–611. doi: 10.1016/j.cellsig.2011.11.016. Chen LH, Kao PYP, Fan YH, et al. (2012). Polymorphisms of CR1, CLU and PICALM confer susceptibility of Alzheimer’s disease in a southern Chinese population. Neurobiol Aging. 33(1): 210.e1–7. doi: 10.1016/j.neurobiolaging.2011.09.016. Chen H-K, Liu Z, Meyer-Franke A, et al. (2012). Small molecule structure correctors abolish detrimental effects of apolipoprotein E4 in cultured neurons. J Biol Chem. 287(8): 5253–5266. doi: 10.1074/ jbc.M111.276162. Cherbuin N, Anstey KJ, Sachdev PS, et al. (2008). Total and regional gray matter volume is not related to APOE*E4 status in a community sample of middle-aged individuals. J Gerontol A Biol Sci Med Sci. 63(5): 501–504. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/18511754 (accessed October 2, 2014). Cherbuin N, Leach LS, Christensen H, Anstey KJ. (2007). Neuroimaging and APOE genotype: a systematic qualitative review. Dement Geriatr Cogn Disord. 24(5): 348–362. doi: 10.1159/000109150. Chhatwal JP, Schultz AP, Johnson K, et al. (2013). Impaired default network functional connectivity in autosomal dominant Alzheimer disease. Neurology. 81(8): 736–744. doi: 10.1212/ WNL.0b013e3182a1aafe. Chu SH, Roeder K, Ferrell RE, et al. (2011). TOMM40 poly-T repeat lengths, age of onset and psychosis risk in Alzheimer disease. Neurobiol Aging. 32(12): 2328.e1–9. doi: 10.1016/j.neurobiolaging.2011.06.016. Cirulli ET, Goldstein DB. (2010). Uncovering the roles of rare variants in common disease through whole-genome sequencing. Nat Rev Genet. 11(6): 415–425. doi: 10.1038/nrg2779. Corder EH, Saunders AM, Risch NJ, et al. (1994). Protective effect of apolipoprotein E type 2 allele for late onset Alzheimer disease. Nat Genet. 7(2): 180–184. doi: 10.1038/ng0694-180. Corder EH, Saunders AM, Strittmatter WJ, et al. (1993). Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science. 261(5123): 921–923. Available at: http://www. ncbi.nlm.nih.gov/pubmed/8346443 (accessed August 13, 2014). Corneveaux JJ, Myers AJ, Allen AN, et al. (2010). Association of CR1, CLU and PICALM with Alzheimer’s disease in a cohort of clinically characterized and neuropathologically verified individuals. Hum Mol Genet. 19(16): 3295–3301. doi: 10.1093/hmg/ddq221. Crocker PR, Paulson JC, Varki A. (2007). Siglecs and their roles in the immune system. Nat Rev Immunol. 7(4): 255–266. doi: 10.1038/ nri2056. Cruchaga C, Nowotny P, Kauwe JSK, et al. (2011). Association and expression analyses with single-nucleotide polymorphisms in TOMM40 in Alzheimer disease. Arch Neurol. 68(8): 1013–1019. doi: 10.1001/archneurol.2011.155. Cruts M, Theuns J, Van Broeckhoven C. (2012). Locus-specific mutation databases for neurodegenerative brain diseases. Hum Mutat. 33(9): 1340–1344. doi: 10.1002/humu.22117.
Damoiseaux JS, Seeley WW, Zhou J, et al. (2012). Gender modulates the APOE ε4 effect in healthy older adults: convergent evidence from functional brain connectivity and spinal fluid tau levels. J Neurosci. 32(24): 8254–8262. doi: 10.1523/JNEUROSCI.0305-12.2012. Dean DC, Jerskey BA, Chen K, et al. (2014). Brain differences in infants at differential genetic risk for late-onset Alzheimer disease: a cross-sectional imaging study. JAMA Neurol. 71(1): 11–22. doi: 10.1001/jamaneurol.2013.4544. DIAN webpage. Available at: http://dian-info.org/default.htm. Dickerson BC, Bakkour A, Salat DH, et al. (2009). The cortical signature of Alzheimer’s disease: regionally specific cortical thinning relates to symptom severity in very mild to mild AD dementia and is detectable in asymptomatic amyloid-positive individuals. Cereb Cortex. 19(3):497–510. doi: 10.1093/cercor/bhn113. Donix M, Burggren AC, Suthana NA, et al. (2010a). Family history of Alzheimer’s disease and hippocampal structure in healthy people. Am J Psychiatry. 167(11): 1399–1406. doi: 10.1176/appi. ajp.2010.09111575. Donix M, Burggren AC, Suthana NA, et al. (2010b). Longitudinal changes in medial temporal cortical thickness in normal subjects with the APOE-4 polymorphism. Neuroimage. 53(1): 37–43. doi: 10.1016/j.neuroimage.2010.06.009. Donix M, Small GW, Bookheimer SY. (2012). Family history and APOE-4 genetic risk in Alzheimer’s disease. Neuropsychol Rev. 22(3): 298–309. doi: 10.1007/s11065-012-9193-2. Doody RS, Thomas RG, Farlow M, et al. (2014). Phase 3 trials of solanezumab for mild-to-moderate Alzheimer’s disease. N Engl J Med. 370(4): 311–321. doi: 10.1056/NEJMoa1312889 Drzezga A, Grimmer T, Riemenschneider M, et al. (2005). Prediction of individual clinical outcome in MCI by means of genetic assessment and (18)F-FDG PET. J Nucl Med. 46(10): 1625–1632. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16204712 (accessed October 3, 2014). Dufrasne S, Roy M, Galvez M, Rosenblatt DS. (2011). Experience over fifteen years with a protocol for predictive testing for Huntington disease. Mol Genet Metab. 102(4): 494–504. doi: 10.1016/j. ymgme.2010.12.001. Erk S, Meyer-Lindenberg A, Opitz von Boberfeld C, et al. (2011). Hippocampal function in healthy carriers of the CLU Alzheimer’s disease risk variant. J Neurosci. 31(49): 18180–18184. doi: 10.1523/ JNEUROSCI.4960-11.2011. Espeseth T, Westlye LT, Fjell AM, Walhovd KB, Rootwelt H, Reinvang I. (2008). Accelerated age-related cortical thinning in healthy carriers of apolipoprotein E epsilon 4. Neurobiol Aging. 29(3): 329–340. doi: 10.1016/j.neurobiolaging.2006.10.030. Ferencz B, Laukka EJ, Lövdén M, et al. (2013). The influence of APOE and TOMM40 polymorphisms on hippocampal volume and episodic memory in old age. Front Hum Neurosci. 7: 198. doi: 10.3389/ fnhum.2013.00198. Fernando MS, Ince PG. (2004). Vascular pathologies and cognition in a population-based cohort of elderly people. J Neurol Sci. 226(1–2): 13–17. doi: 10.1016/j.jns.2004.09.004. Figurov A, Pozzo-Miller LD, Olafsson P, Wang T, Lu B. (1996). Regulation of synaptic responses to high-frequency stimulation and LTP by neurotrophins in the hippocampus. Nature. 381(6584): 706–709. doi: 10.1038/381706a0. Filippini N, MacIntosh BJ, Hough MG, et al. (2009). Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A. 106(17): 7209–7214. doi: 10.1073/ pnas.0811879106. Finelli D, Rollinson S, Harris J, et al. (2014). TREM2 analysis and increased risk of Alzheimer’s disease. Neurobiol Aging. 36(1): 546e9–13. doi: 10.1016/j.neurobiolaging.2014.08.001. Fleisher AS, Chen K, Liu X, et al. (2011). Using positron emission tomography and florbetapir F18 to image cortical amyloid in patients with mild cognitive impairment or dementia due to Alzheimer
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
387
disease. Arch Neurol. 68(11): 1404–1411. doi: 10.1001/archneurol.2011.150. Fleisher AS, Chen K, Liu X, et al. (2013). Apolipoprotein E ε4 and age effects on florbetapir positron emission tomography in healthy aging and Alzheimer disease. Neurobiol Aging. 34(1):1–12. doi: 10.1016/j. neurobiolaging.2012.04.017. Fleisher AS, Sherzai A, Taylor C, Langbaum JBS, Chen K, Buxton RB. (2009). Resting-state BOLD networks versus task-associated functional MRI for distinguishing Alzheimer’s disease risk groups. Neuroimage. 47(4): 1678–1690. doi: 10.1016/j.neuroimage.2009.06.021 Fratiglioni L, De Ronchi D, Agüero-Torres H. (1999). Worldwide prevalence and incidence of dementia. Drugs Aging. 15(5): 365–375. doi : 10.2165/00002512-199915050-00004. Fujimoto K, Polimeni JR, van der Kouwe AJW, et al. (2014). Quantitative comparison of cortical surface reconstructions from MP2RAGE and multi-echo MPRAGE data at 3 and 7 T. Neuroimage. 90: 60–73. doi: 10.1016/j.neuroimage.2013.12.012. Furney SJ, Simmons A, Breen G, et al. (2011). Genome-wide association with MRI atrophy measures as a quantitative trait locus for Alzheimer’s disease. Mol Psychiatry. 16(11): 1130–1138. doi: 10.1038/ mp.2010.123. Gatz M, Reynolds CA, Fratiglioni L, et al. (2006). Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry. 63(2): 168–174. doi: 10.1001/archpsyc.63.2.168. Goate A, Chartier-Harlin MC, Mullan M, et al. (1991). Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer’s disease. Nature. 349(6311): 704–706. doi: 10.1038/349704a0. Goedert M, Spillantini MG. (2006). A century of Alzheimer’s disease. Science. 314(5800): 777–781. doi: 10.1126/science.1132814. Green AE, Gray JR, Deyoung CG, et al. (2014). A combined effect of two Alzheimer’s risk genes on medial temporal activity during executive attention in young adults. Neuropsychologia. 56: 1–8. doi: 10.1016/j.neuropsychologia.2013.12.020. Guerreiro R, Wojtas A, Bras J, et al. (2013). TREM2 variants in Alzheimer’s disease. N Engl J Med. 368(2): 117–127. doi: 10.1056/ NEJMoa1211851. Gurland BJ, Wilder DE, Lantigua R, et al. (1999). Rates of dementia in three ethnoracial groups. Int J Geriatr Psychiatry. 14(6): 481–493. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10398359 (accessed October 9, 2014). Hardy JA, Higgins GA. (1992) Alzheimer’s disease: The amyloid cascade hypothesis. Science. 256(5054): 184–185. doi: 10.1126/science.1566067 Hardy J, Selkoe DJ. (2002). The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science. 297(5580): 353–356. doi: 10.1126/science.1072994. Harold D, Abraham R, Hollingworth P, et al. (2009). Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer’s disease. Nat Genet. 41(10): 1088–1093. doi: 10.1038/ng.440. Hashimoto R, Hirata Y, Asada T, et al. (2009). Effect of the brain-derived neurotrophic factor and the apolipoprotein E polymorphisms on disease progression in preclinical Alzheimer’s disease. Genes Brain Behav. 8(1): 43–52. doi: 10.1111/j.1601-183X .2008.00440.x. Hatashita S, Yamasaki H. (2013). Diagnosed mild cognitive impairment due to Alzheimer’s disease with PET biomarkers of beta amyloid and neuronal dysfunction. PLoS One. 8(6): e66877. doi: 10.1371/journal.pone.0066877. Hebert LE, Scherr PA, Bienias JL, Bennett DA, Evans DA. (2003). Alzheimer disease in the US population: prevalence estimates using the 2000 census. Arch Neurol. 60(8): 1119–1122. doi: 10.1001/archneur.60.8.1119. Heise V, Filippini N, Ebmeier KP, Mackay CE. (2011). The APOE ɛ4 allele modulates brain white matter integrity in healthy adults. Mol Psychiatry. 16(9): 908–916. doi: 10.1038/mp.2010.90.
Heise V, Filippini N, Trachtenberg AJ, Suri S, Ebmeier KP, Mackay CE. (2014). Apolipoprotein E genotype, gender and age modulate connectivity of the hippocampus in healthy adults. Neuroimage. 98: 23–30. doi: 10.1016/j.neuroimage.2014.04.081. Hensley K. (2010). Neuroinflammation in Alzheimer’s disease: mechanisms, pathologic consequences, and potential for therapeutic manipulation. J Alzheimers Dis. 21(1): 1–14. doi: 10.3233/ JAD-2010-1414. Hirai H, Maru Y, Hagiwara K, Nishida J, Takaku F. (1987). A novel putative tyrosine kinase receptor encoded by the eph gene. Science. 238(4834): 1717–1720. Available at: http://www.ncbi.nlm.nih. gov/pubmed/2825356 (accessed October 9, 2014). Hirsch-Reinshagen V, Zhou S, Burgess BL, et al. (2004). Deficiency of ABCA1 impairs apolipoprotein E metabolism in brain. J Biol Chem. 279(39): 41197–41207. doi: 10.1074/jbc.M407962200. Hohman TJ, Koran ME, Thornton-Wells T. (2013). Epistatic genetic effects among Alzheimer’s candidate genes. PLoS One. 8(11): e80839. doi: 10.1371/journal.pone.0080839. Hollingworth P, Harold D, Sims R, et al. (2011). Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP are associated with Alzheimer’s disease. Nat Genet. 43(5): 429–435. doi: 10.1038/ng.803. Honea RA, Cruchaga C, Perea RD, et al. (2013). Characterizing the role of brain derived neurotrophic factor genetic variation in Alzheimer’s disease neurodegeneration. PLoS One. 8(9): e76001. doi: 10.1371/ journal.pone.0076001. Honea RA, Swerdlow RH, Vidoni ED, Burns JM. (2011). Progressive regional atrophy in normal adults with a maternal history of Alzheimer disease. Neurology. 76(9): 822–829. doi: 10.1212/ WNL.0b013e31820e7b74. Honea RA, Swerdlow RH, Vidoni ED, Goodwin J, Burns JM. (2010). Reduced gray matter volume in normal adults with a maternal family history of Alzheimer disease. Neurology. 74(2): 113–120. doi: 10.1212/WNL.0b013e3181c918cb. Honea RA, Vidoni ED, Swerdlow RH, Burns JM. (2012). Maternal family history is associated with Alzheimer’s disease biomarkers. J Alzheimers Dis. 31(3): 659–668. doi: 10.3233/JAD-2012-120676. Hong M-G, Alexeyenko A, Lambert J-C, Amouyel P, Prince JA. (2010). Genome-wide pathway analysis implicates intracellular transmembrane protein transport in Alzheimer disease. J Hum Genet. 55(10): 707–709. doi: 10.1038/jhg.2010.92. Hostage CA, Roy Choudhury K, Doraiswamy PM, Petrella JR. (2013). Dissecting the gene dose-effects of the APOE ε4 and ε2 alleles on hippocampal volumes in aging and Alzheimer’s disease. PLoS One. 8(2): e54483. doi: 10.1371/journal.pone.0054483. Hughes TM, Lopez OL, Evans RW, et al. (2014). Markers of cholesterol transport are associated with amyloid deposition in the brain. Neurobiol Aging. 35(4): 802–807. doi:10.1016/j.neurobiolaging.2013.09.040. IGAP Website. Available at: http://www.wikigenes.org/e/art/e/258. html. Jack CR, Knopman DS, Jagust WJ, et al. (2010). Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade. Lancet Neurol. 9(1):119–128. doi: 10.1016/S1474-4422(09)70299-6. Jahanshad N, Rajagopalan P, Hua X, et al. (2013). Genome-wide scan of healthy human connectome discovers SPON1 gene variant influencing dementia severity. Proc Natl Acad Sci U S A. 110(12): 4768–4773. doi: 10.1073/pnas.1216206110. Jak AJ, Houston WS, Nagel BJ, Corey-Bloom J, Bondi MW. (2007). Differential cross-sectional and longitudinal impact of APOE genotype on hippocampal volumes in nondemented older adults. Dement Geriatr Cogn Disord. 23(6): 382–389. doi: 10.1159/000101340. Jervis GA. (1948). Early senile dementia in mongoloid idiocy. Am J Psychiatry. 105(2): 102–106. Available at: http://www.ncbi.nlm.nih. gov/pubmed/18876430 (accessed October 2, 2014). Jiang T, Tan L, Zhu X-C, et al. (2014). Upregulation of TREM2 ameliorates neuropathology and rescues spatial cognitive impairment in a
3 8 8 part vii : I maging G enetics of C ognitive A ging
transgenic mouse model of Alzheimer’s disease. Neuropsychopharmacology. doi: 10.1038/npp.2014.164. Johnson SC, La Rue A, Hermann BP, et al. (2011). The effect of TOMM40 poly-T length on gray matter volume and cognition in middle-aged persons with APOE ε3/ε3 genotype. Alzheimers Dement. 7(4): 456–465. doi: 10.1016/j.jalz.2010.11.012. Jonsson T, Atwal JK, Steinberg S, et al. (2012). A mutation in APP protects against Alzheimer’s disease and age-related cognitive decline. Nature. 488(7409): 96–99. doi: 10.1038/nature11283. Josephs KA, Murray ME, Whitwell JL, Parisi JE, Petrucelli L, Jack CR, Petersen RC, Dickson DW. (2014). Staging TDP-43 pathology in Alzheimer’s disease. Acta Neuropathol. 127: 441–450. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=394479 9&tool=pmcentrez&rendertype=abstract [Accessed August 7, 2015]. Jonsson T, Stefansson H, Steinberg S, et al. (2013). Variant of TREM2 associated with the risk of Alzheimer’s disease. N Engl J Med. 368(2): 107–116. doi: 10.1056/NEJMoa1211103. Jun G, Naj AC, Beecham GW, et al. (2010). Meta-analysis confirms CR1, CLU, and PICALM as alzheimer disease risk loci and reveals interactions with APOE genotypes. Arch Neurol. 67(12): 1473–1484. doi: 10.1001/archneurol.2010.201. Keage HA, Hunter S, Matthews FE, et al. (2014) TDP-43 pathology in the population: prevalence and associations with dementia and age. Journal of Alzheimer’s Disease. 42(2): 641-650. doi: 10.3233/JAD-132351 Kesslak JP, Nagata SF, Lott I, Nalcioglu O. (1994). Magnetic resonance imaging analysis of age-related changes in the brains of individuals with Down’s syndrome. Neurology. 44(6): 1039–1045. Available at: http:// www.ncbi.nlm.nih.gov/pubmed/8208396 (accessed October 1, 2014). Khan W, Giampietro V, Ginestet C, et al. (2014). No differences in hippocampal volume between carriers and non-carriers of the ApoE ε4 and ε2 alleles in young healthy adolescents. J Alzheimers Dis. 40(1): 37–43. doi: 10.3233/JAD-131841. Kirchheiner J, Lorch R, Lebedeva E, et al. (2008). Genetic variants in FKBP5 affecting response to antidepressant drug treatment. Pharmacogenomics. 9(7): 841–846. doi: 10.2217/14622416.9.7.841. Kljajevic V, Meyer P, Holzmann C, et al. (2014). The ε4 genotype of apolipoprotein E and white matter integrity in Alzheimer’s disease. Alzheimers Dement. 10(3): 401–404. doi: 10.1016/j. jalz.2013.02.008. Klunk WE, Engler H, Nordberg A, et al. (2004). Imaging brain amyloid in Alzheimer’s disease with Pittsburgh Compound-B. Ann Neurol. 55(3): 306–319. doi: 10.1002/ana.20009. Knickmeyer RC, Wang J, Zhu H, et al. (2014). Common variants in psychiatric risk genes predict brain structure at birth. Cereb Cortex. 24(5): 1230–1246. doi: 10.1093/cercor/bhs401. Knight WD, Kim LG, Douiri A, Frost C, Rossor MN, Fox NC. (2011). Acceleration of cortical thinning in familial Alzheimer’s disease. Neurobiol Aging. 32(10): 1765–1773. doi: 10.1016/j.neurobiolaging.2009.11.013. Knopman DS, Jack CR, Wiste HJ, et al. (2014). 18F-fluorodeoxyglucose positron emission tomography, aging, and apolipoprotein E genotype in cognitively normal persons. Neurobiol Aging. 35(9): 2096–2106. doi: 10.1016/j.neurobiolaging.2014.03.006. Kohannim O, Hua X, Rajagopalan P, et al. (2013). Multilocus genetic profiling to empower drug trials and predict brain atrophy. NeuroImage Clin. 2: 827–835. doi: 10.1016/j.nicl.2013.05.007. Kohannim O, Jahanshad N, Braskie MN, et al. (2012). Predicting white matter integrity from multiple common genetic variants. Neuropsychopharmacology. 37(9): 2012–2019. doi: 10.1038/npp.2012.49. Koivunen J, Verkkoniemi A, Aalto S, et al. (2008). PET amyloid ligand [11C]PIB uptake shows predominantly striatal increase in variant Alzheimer’s disease. Brain. 131(Pt 7): 1845–1853. doi: 10.1093/ brain/awn107. Kraemer HC, Yesavage JA, Taylor JL, Kupfer D. (2000). How can we learn about developmental processes from cross-sectional studies, or can we? Am J Psychiatry. 157(2): 163–171. Available at: http:// www.ncbi.nlm.nih.gov/pubmed/10671382 (accessed October 11, 2014).
Krasuski JS, Alexander GE, Horwitz B, Rapoport SI, Schapiro MB. (2002). Relation of medial temporal lobe volumes to age and memory function in nondemented adults with Down’s syndrome: implications for the prodromal phase of Alzheimer’s disease. Am J Psychiatry. 159(1): 74–81. Available at: http://www.ncbi.nlm.nih. gov/pubmed/11772693 (accessed September 15, 2014). Lambert JC, Heath S, Even G, et al. (2009). Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer’s disease. Nat Genet. 41(10): 1094–1099. doi: 10.1038/ng.439. Lambert JC, Ibrahim-Verbaas CA, Harold D, et al. (2013). Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet. 45(12): 1452–1458. doi: 10.1038/ng.2802. Landt J, D’Abrera JC, Holland AJ, et al. (2011). Using positron emission tomography and Carbon 11-labeled Pittsburgh Compound B to image Brain Fibrillar β-amyloid in adults with down syndrome: safety, acceptability, and feasibility. Arch Neurol. 68(7): 890–896. doi: 10.1001/archneurol.2011.36. Leduc V, Domenger D, De Beaumont L, Lalonde D, Bélanger-Jasmin S, Poirier J. (2011). Function and comorbidities of apolipoprotein e in Alzheimer’s disease. Int J Alzheimers Dis. 2011: 974361. doi: 10.4061/2011/974361. Lee GJ, Lu PH, Medina LD, et al. (2013). Regional brain volume differences in symptomatic and presymptomatic carriers of familial Alzheimer’s disease mutations. J Neurol Neurosurg Psychiatry. 84(2): 154–162. doi: 10.1136/jnnp-2011-302087. Lehmann M, Ghosh PM, Madison C, et al. (2014). Greater medial temporal hypometabolism and lower cortical amyloid burden in ApoE4-positive AD patients. J Neurol Neurosurg Psychiatry. 85(3): 266–273. doi: 10.1136/jnnp-2013-305858. Lim YY, Villemagne VL, Laws SM, et al. (2013). BDNF Val66Met, Aβ amyloid, and cognitive decline in preclinical Alzheimer’s disease. Neurobiol Aging. 34(11): 2457–2464. doi: 10.1016/j.neurobiolaging.2013.05.006. Liu X, Bai F, Yue C, et al. (2014). The association between TOMM40 gene polymorphism and spontaneous brain activity in amnestic mild cognitive impairment. J Neurol. 261(8): 1499–1507. doi: 10.1007/ s00415-014-7368-x. Liu Y, Paajanen T, Westman E, et al. (2010). Effect of APOE ε4 allele on cortical thicknesses and volumes: the AddNeuroMed study. J Alzheimers Dis. 21(3): 947–966. doi: 10.3233/JAD-2010-100201. Lopera F. (1997). Clinical features of early-onset alzheimer disease in a large kindred with an E280A presenilin-1 mutation. JAMA. 277(10): 793. doi: 10.1001/jama.1997.03540340027028. Lott IT, Head E. (2001). Down syndrome and Alzheimer’s disease: a link between development and aging. Ment Retard Dev Disabil Res Rev. 7(3): 172–178. doi: 10.1002/mrdd.1025. Lowe VJ, Weigand SD, Senjem ML, et al. (2014). Association of hypometabolism and amyloid levels in aging, normal subjects. Neurology. doi: 10.1212/WNL.0000000000000467. Lu PH, Thompson PM, Leow A, et al. (2011). Apolipoprotein E genotype is associated with temporal and hippocampal atrophy rates in healthy elderly adults: a tensor-based morphometry study. J Alzheimers Dis. 23(3): 433–442. doi: 10.3233/JAD-2010-101398. Lue L-F, Schmitz CT, Sorrano G, Sue LI, Beach TG, Walker DG. (2014). TREM2 protein expression changes correlate with Alzheimer’s disease neurodegenerative pathologies in postmortem temporal cortices. Brain Pathol. doi: 10.1111/bpa.12190. Luis EO, Ortega-Cubero S, Lamet I, et al. (2014). Frontobasal gray matter loss is associated with the TREM2 p.R47H variant. Neurobiol Aging. doi: 10.1016/j.neurobiolaging.2014.06.007. Lyall DM, Harris SE, Bastin ME, et al. (2014). Alzheimer’s disease susceptibility genes APOE and TOMM40, and brain white matter integrity in the Lothian Birth Cohort 1936. Neurobiol Aging. 35(6): 1513.e25–33. doi:10.1016/j.neurobiolaging.2014.01.006. Machulda MM, Jones DT, Vemuri P, et al. (2011). Effect of APOE ε4 status on intrinsic network connectivity in cognitively normal elderly subjects. Arch Neurol. 68(9): 1131–1136. doi: 10.1001/archneurol.2011.108.
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
389
Mahley RW, Weisgraber KH, Huang Y. (2009). Apolipoprotein E: structure determines function, from atherosclerosis to Alzheimer’s disease to AIDS. J Lipid Res. 50 Suppl: S183–188. doi: 10.1194/jlr. R800069-JLR200. Mak ACY, Pullinger CR, Tang LF, et al. (2014). Effects of the Absence of Apolipoprotein E on Lipoproteins, Neurocognitive Function, and Retinal Function. JAMA Neurol. doi: 10.1001/jamaneurol.2014.2011. Manning EN, Barnes J, Cash DM, et al. (2014). APOE ε4 is associated with disproportionate progressive hippocampal atrophy in AD. PLoS One. 9(5): e97608. doi: 10.1371/journal.pone.0097608. Marosi K, Mattson MP. (2014). BDNF mediates adaptive brain and body responses to energetic challenges. Trends Endocrinol Metab. 25(2): 89–98. doi: 10.1016/j.tem.2013.10.006. Maruszak A, Pepłońska B, Safranow K, Chodakowska-Żebrowska M, Barcikowska M, Zekanowski C. (2012). TOMM40 rs10524523 polymorphism’s role in late-onset Alzheimer’s disease and in longevity. J Alzheimers Dis. 28(2): 309–322. doi: 10.3233/JAD-2011-110743. Matura S, Prvulovic D, Jurcoane A, et al. (2014). Differential effects of the ApoE4 genotype on brain structure and function. Neuroimage. 89: 81–91. doi: 10.1016/j.neuroimage.2013.11.042. May PC, Lampert-Etchells M, Johnson SA, Poirier J, Masters JN, Finch CE. (1990). Dynamics of gene expression for a hippocampal glycoprotein elevated in Alzheimer’s disease and in response to experimental lesions in rat. Neuron. 5(6): 831–839. Available at: http://www. ncbi.nlm.nih.gov/pubmed/1702645 (accessed October 7, 2014). McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. (1984). Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology. 34(7): 939–939. doi: 10.1212/WNL.34.7.939. McKhann GM, Knopman DS, Chertkow H, et al. (2011). The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7(3): 263–269. doi: 10.1016/j.jalz.2011.03.005. Meda SA, Narayanan B, Liu J, et al. (2012). A large scale multivariate parallel ICA method reveals novel imaging-genetic relationships for Alzheimer’s disease in the ADNI cohort. Neuroimage. 60(3): 1608–1621. doi: 10.1016/j.neuroimage.2011.12.076. Melchior B, Garcia AE, Hsiung B-K, et al. (2010). Dual induction of TREM2 and tolerance-related transcript, Tmem176b, in amyloid transgenic mice: implications for vaccine-based therapies for Alzheimer’s disease. ASN Neuro. 2(3): e00037. doi: 10.1042/ AN20100010. Melville SA, Buros J, Parrado AR, et al. (2012). Multiple loci influencing hippocampal degeneration identified by genome scan. Ann Neurol. 72(1): 65–75. doi: 10.1002/ana.23644. Mendez MF, Cummings JL. (2003). Dementia: a clinical approach. Butterworth-Heinemann: 654. Available at: http://books.google. com/books?id=LzM4v_OruZ8C&pgis=1 (accessed September 29, 2014). Minami SS, Min S-W, Krabbe G, et al. Progranulin protects against amyloid β deposition and toxicity in Alzheimer’s disease mouse models. Nat Med. 2014;advance on. doi:10.1038/nm.3672. Molecular Signatures Database. Available at: http://www.broadinstitute.org/gsea/msigdb/index.jsp. Mondadori CRA, de Quervain DJ-F, Buchmann A, et al. (2007). Better memory and neural efficiency in young apolipoprotein E epsilon4 carriers. Cereb Cortex. 17(8): 1934–1947. doi: 10.1093/cercor/bhl103. Morgan K. (2011). The three new pathways leading to Alzheimer’s disease. Neuropathol Appl Neurobiol. 37(4): 353–357. doi: 10.1111/j.1 365-2990.2011.01181.x. Morgan K, Carrasquillo MM. (2013). Genetic variants in Alzheimer’s disease. New York: Springer Science + Business Media. Morris JC. (2005). Early-stage and preclinical Alzheimer disease. Alzheimer Dis Assoc Disord. 19(3):163–165. Available at: http://
www.ncbi.nlm.nih.gov/pubmed/16118535 (accessed September 29, 2014). Morris JC, Roe CM, Xiong C, et al. (2010). APOE predicts amyloid-beta but not tau Alzheimer pathology in cognitively normal aging. Ann Neurol. 67(1): 122–131. doi: 10.1002/ana.21843. Mosconi L, de Leon M, Murray J, et al. (2011). Reduced mitochondria cytochrome oxidase activity in adult children of mothers with Alzheimer’s disease. J Alzheimers Dis. 27(3): 483–490. doi: 10.3233/ JAD-2011-110866. Mosconi L, Murray J, Tsui WH, et al. (2014). Brain imaging of cognitively normal individuals with 2 parents affected by late-onset AD. Neurology. 82(9): 752–760. doi: 10.1212/WNL.0000000000000181. Mosconi L, Nacmias B, Sorbi S, et al. (2004). Brain metabolic decreases related to the dose of the ApoE e4 allele in Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 75(3): 370–376. Available at: http://www. pubmedcentral.nih.gov/articlerender.fcgi?artid=1738980&tool=p mcentrez&rendertype=abstract (accessed October 3, 2014). Mosconi L, Rinne JO, Tsui WH, et al. (2013). Amyloid and metabolic positron emission tomography imaging of cognitively normal adults with Alzheimer’s parents. Neurobiol Aging. 34(1): 22–34. doi: 10.1016/j.neurobiolaging.2012.03.002. Mueller SG, Schuff N, Raptentsetsang S, Elman J, Weiner MW. (2008). Selective effect of Apo e4 on CA3 and dentate in normal aging and Alzheimer’s disease using high resolution MRI at 4 T. Neuroimage. 42(1): 42–48. doi: 10.1016/j.neuroimage.2008.04.174. Mueller SG, Weiner MW. (2009). Selective effect of age, Apo e4, and Alzheimer’s disease on hippocampal subfields. Hippocampus. 19(6): 558–564. doi: 10.1002/hipo.20614. Nagahara AH, Merrill DA, Coppola G, et al. (2009). Neuroprotective effects of brain-derived neurotrophic factor in rodent and primate models of Alzheimer’s disease. Nat Med. 15(3): 331–337. doi: 10.1038/nm.1912. Nagahara AH, Tuszynski MH. (2011). Potential therapeutic uses of BDNF in neurological and psychiatric disorders. Nat Rev Drug Discov. 10(3): 209–219. doi: 10.1038/nrd3366. Naj AC, Jun G, Beecham GW, et al. (2011). Common variants at MS4A4/MS4A6E, CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer’s disease. Nat Genet. 43(5): 436–441. doi: 10.1038/ng.801. Nakamura-Hirota T, Kadoyama K, Takano M, Otani M, Matsuyama S. (2012). The expression changes of EphA3 receptor during synaptic plasticity in mouse hippocampus through activation of nicotinic acetylcholine receptor. Neuroreport. 23(13): 746–751. doi: 10.1097/ WNR.0b013e3283565144. Nho K, Corneveaux JJ, Kim S, et al. (2013). Whole-exome sequencing and imaging genetics identify functional variants for rate of change in hippocampal volume in mild cognitive impairment. Mol Psychiatry. 18(7): 781–787. doi: 10.1038/mp.2013.24. O’Dwyer L, Lamberton F, Matura S, et al. (2012). Reduced hippocampal volume in healthy young ApoE4 carriers: an MRI study. PLoS One. 7(11): e48895. doi: 10.1371/journal.pone.0048895. Okonkwo OC, Xu G, Dowling NM, et al. (2012). Family history of Alzheimer disease predicts hippocampal atrophy in healthy middle-aged adults. Neurology. 78(22): 1769–1776. doi: 10.1212/ WNL.0b013e3182583047. Ossenkoppele R, van der Flier WM, Zwan MD, et al. (2013). Differential effect of APOE genotype on amyloid load and glucose metabolism in AD dementia. Neurology. 80(4): 359–365. doi: 10.1212/ WNL.0b013e31827f0889. Pant S, Sharma M, Patel K, Caplan S, Carr CM, Grant BD. (2009). AMPH-1/ Amphiphysin/Bin1 functions with RME-1/Ehd1 in endocytic recycling. Nat Cell Biol. 11(12): 1399–1410. doi: 10.1038/ncb1986. Pietrini P, Dani A, Furey ML, et al. (1997). Low glucose metabolism during brain stimulation in older Down’s syndrome subjects at risk for Alzheimer’s disease prior to dementia. Am J Psychiatry. 154(8): 1063–1069. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/9247390 (accessed October 1, 2014).
3 9 0 part vii : I maging G enetics of C ognitive A ging
Pievani M, Galluzzi S, Thompson PM, Rasser PE, Bonetti M, Frisoni GB. (2011). APOE4 is associated with greater atrophy of the hippocampal formation in Alzheimer’s disease. Neuroimage. 55(3): 909–919. doi: 10.1016/j.neuroimage.2010.12.081. Pike KE, Savage G, Villemagne VL, et al. (2007). Beta-amyloid imaging and memory in non-demented individuals: evidence for preclinical Alzheimer’s disease. Brain. 130(Pt 11): 2837–2844. doi: 10.1093/ brain/awm238. Pooler AM, Noble W, Hanger DP. (2014). A role for tau at the synapse in Alzheimer’s disease pathogenesis. Neuropharmacology. 76(Pt A): 1–8. doi: 10.1016/j.neuropharm.2013.09.018. Powell D, Caban-Holt A, Jicha G, et al. (2014). Frontal white matter integrity in adults with Down syndrome with and without dementia. Neurobiol Aging. 35(7): 1562–1569. doi: 10.1016/j.neurobiolaging.2014.01.137. Quiroz YT, Budson AE, Celone K, et al. (2010). Hippocampal hyperactivation in presymptomatic familial Alzheimer’s disease. Ann Neurol. 68(6): 865–875. doi: 10.1002/ana.22105. Quiroz YT, Stern CE, Reiman EM, et al. (2013). Cortical atrophy in presymptomatic Alzheimer’s disease presenilin 1 mutation carriers. J Neurol Neurosurg Psychiatry. 84(5): 556–561. doi: 10.1136/ jnnp-2012-303299. Rajagopalan P, Hibar DP, Thompson PM. (2013). TREM2 and neurodegenerative disease. N Engl J Med. 369(16): 1565–1567. doi: 10.1056/NEJMc1306509#SA3. Rayaprolu S, Mullen B, Baker M, et al. (2013). TREM2 in neurodegeneration: evidence for association of the p.R47H variant with frontotemporal dementia and Parkinson’s disease. Mol Neurodegener. 8: 19. doi: 10.1186/1750-1326-8-19. Reiman EM, Caselli RJ, Yun LS, et al. (1996). Preclinical evidence of Alzheimer’s disease in persons homozygous for the epsilon 4 allele for apolipoprotein E. N Engl J Med. 334(12): 752–758. doi: 10.1056/ NEJM199603213341202. Reiman EM, Chen K, Alexander GE, et al. (2004). Functional brain abnormalities in young adults at genetic risk for late-onset Alzheimer’s dementia. Proc Natl Acad Sci U S A. 101(1): 284–289. doi: 10.1073/pnas.2635903100. Reitz C, Mayeux R. (2013). TREM2 and neurodegenerative disease. N Engl J Med. 369(16): 1564–1565. doi: 10.1056/NEJMc1306509#SA1. Richter-Schmidinger T, Alexopoulos P, Horn M, et al. (2011). Influence of brain-derived neurotrophic-factor and apolipoprotein E genetic variants on hippocampal volume and memory performance in healthy young adults. J Neural Transm. 118(2): 249–257. doi: 10.1007/s00702-010-0539-8. Ridha BH, Barnes J, Bartlett JW, et al. (2006). Tracking atrophy progression in familial Alzheimer’s disease: a serial MRI study. Lancet Neurol. 5(10): 828–834. doi: 10.1016/S1474-4422(06)70550-6. Ringman JM, O’Neill J, Geschwind D, et al. (2007). Diffusion tensor imaging in preclinical and presymptomatic carriers of familial Alzheimer’s disease mutations. Brain. 130(Pt 7): 1767–1776. doi: 10.1093/brain/awm102. Ringman JM, Pope W, Salamon N. (2010). Insensitivity of visual assessment of hippocampal atrophy in familial Alzheimer’s disease. J Neurol. 257(5): 839–842. doi: 10.1007/s00415-009-5436-4. Ripke S, Neale BM, Corvin A, et al. (2014). Biological insights from 108 schizophrenia-associated genetic loci. Nature. 511(7510): 421–427. doi:10.1038/nature13595. Roden DM, George AL. (2002). The genetic basis of variability in drug responses. Nat Rev Drug Discov. 1(1): 37–44. doi: 10.1038/nrd705 Rodríguez-Rodríguez E, Sánchez-Juan P, Vázquez-Higuera JL, et al. (2013). Genetic risk score predicting accelerated progression from mild cognitive impairment to Alzheimer’s disease. J Neural Transm. 120(5): 807–812. doi: 10.1007/s00702-012-0920-x. Rogaev EI, Sherrington R, Rogaeva EA, et al. (1995). Familial Alzheimer’s disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer’s disease type 3 gene. Nature. 376(6543): 775–778. doi: 10.1038/376775a0.
Rogalski EJ, Rademaker A, Harrison TM, et al. (2011). ApoE E4 is a susceptibility factor in amnestic but not aphasic dementias. Alzheimer Dis Assoc Disord. 25(2): 159–163. doi: 10.1097/ WAD.0b013e318201f249. Rose EJ, Donohoe G. (2013). Brain vs behavior: an effect size comparison of neuroimaging and cognitive studies of genetic risk for schizophrenia. Schizophr Bull. 39(3): 518–526. doi: 10.1093/schbul/ sbs056. Roses AD, Lutz MW, Amrine-Madsen H, et al. (2010). A TOMM40 variable-length polymorphism predicts the age of late-onset Alzheimer’s disease. Pharmacogenomics J. 10(5): 375–384. doi: 10.1038/ tpj.2009.69. Ruiz A, Dols-Icardo O, Bullido MJ, et al. (2014). Assessing the role of the TREM2 p.R47H variant as a risk factor for Alzheimer’s disease and frontotemporal dementia. Neurobiol Aging. 35(2): 444.e1–4. doi: 10.1016/j.neurobiolaging.2013.08.011. Rusted JM, Evans SL, King SL, Dowell N, Tabet N, Tofts PS. (2013). APOE e4 polymorphism in young adults is associated with improved attention and indexed by distinct neural signatures. Neuroimage. 65: 364–373. doi: 10.1016/j.neuroimage.2012.10.010. Ryan NS, Keihaninejad S, Shakespeare TJ, et al. (2013). Magnetic resonance imaging evidence for presymptomatic change in thalamus and caudate in familial Alzheimer’s disease. Brain. 136(Pt 5): 1399–1414. doi: 10.1093/brain/awt065. Sabbagh MN, Fleisher A, Chen K, et al. (2011). Positron emission tomography and neuropathologic estimates of fibrillar amyloid-β in a patient with Down syndrome and Alzheimer disease. Arch Neurol. 68(11): 1461–1466. doi: 10.1001/archneurol.2011.535. Sabuncu MR, Buckner RL, Smoller JW, Lee PH, Fischl B, Sperling RA. (2012). The association between a polygenic Alzheimer score and cortical thickness in clinically normal subjects. Cereb Cortex. 22(11): 2653–2661. doi: 10.1093/cercor/bhr348. Sala-Llonch R, Fortea J, Bartrés-Faz D, et al. (2013). Evolving brain functional abnormalities in PSEN1 mutation carriers: a resting and visual encoding fMRI study. J Alzheimers Dis. 36(1): 165–175. doi: 10.3233/JAD-130062. Salloway S, Sperling R, Fox NC, et al. (2014). Two phase 3 trials of bapineuzumab in mild-to-moderate Alzheimer’s disease. N Engl J Med. 370(4): 322–333. doi: 10.1056/NEJMoa1304839. Samuraki M, Matsunari I, Chen W-P, et al. (2012). Glucose metabolism and gray-matter concentration in apolipoprotein E ε4 positive normal subjects. Neurobiol Aging. 33(10): 2321–2323. doi: 10.1016/j. neurobiolaging.2011.11.020. Scahill RI, Ridgway GR, Bartlett JW, et al. (2013). Genetic influences on atrophy patterns in familial Alzheimer’s disease: a comparison of APP and PSEN1 mutations. J Alzheimers Dis. 35(1): 199–212. doi: 10.3233/JAD-121255. Schellenberg G, Bird T, Wijsman E, et al. (1992). Genetic linkage evidence for a familial Alzheimer’s disease locus on chromosome 14. Science (80-). 258(5082): 668–671. doi: 10.1126/science.1411576. Schmechel DE, Saunders AM, Strittmatter WJ, et al. (1993). Increased amyloid beta-peptide deposition in cerebral cortex as a consequence of apolipoprotein E genotype in late-onset Alzheimer disease. Proc Natl Acad Sci U S A. 90(20): 9649–9653. Available at: http://www. pubmedcentral.nih.gov/articlerender.fcgi?artid=47627&tool=pmc entrez&rendertype=abstract (accessed August 26, 2014). Schott JM, Fox NC, Frost C, et al. (2003). Assessing the onset of structural change in familial Alzheimer’s disease. Ann Neurol. 53(2): 181–188. doi: 10.1002/ana.10424. Scheinin NM, Wikman K, Jula A, et al. (2014). Cortical 11C-PIB uptake is associated with age, APOE genotype, and gender in “healthy aging.” J Alzheimers Dis. 41(1):193–202. doi: 10.3233/JAD-132783. Seshadri S, Fitzpatrick AL, Ikram MA, et al. (2010). Genome-wide analysis of genetic loci associated with Alzheimer disease. JAMA. 303(18): 1832–1840. doi: 10.1001/jama.2010.574. SFARI Gene Database Webpage. Available at: https://gene.sfari.org/ autdb/HG_Home.do.
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
391
Shaw P, Lerch JP, Pruessner JC, et al. (2007). Cortical morphology in children and adolescents with different apolipoprotein E gene polymorphisms: an observational study. Lancet Neurol. 6(6): 494–500. doi: 10.1016/S1474-4422(07)70106-0. Sheline YI, Morris JC, Snyder AZ, et al. (2010). APOE4 allele disrupts resting state fMRI connectivity in the absence of amyloid plaques or decreased CSF Aβ42. J Neurosci. 30(50): 17035–17040. doi: 10.1523/JNEUROSCI.3987-10.2010. Shen L, Kim S, Risacher SL, et al. (2010). Whole genome association study of brain-wide imaging phenotypes for identifying quantitative trait loci in MCI and AD: A study of the ADNI cohort. Neuroimage. 53(3): 1051–1063. doi: 10.1016/j.neuroimage.2010.01.042. Shi X, Ma X, Wu W, Huang F, Yuan C, Guo H. (2014). Parallel imaging and compressed sensing combined framework for accelerating high-resolution diffusion tensor imaging using inter-image correlation. Magn Reson Med. 73(5): 1775–1785. doi: 10.1002/mrm.25290. Silver M, Janousova E, Hua X, Thompson PM, Montana G. (2012). Identification of gene pathways implicated in Alzheimer’s disease using longitudinal imaging phenotypes with sparse regression. Neuroimage. 63(3): 1681–1694. doi: 10.1016/j.neuroimage.2012.08.002. Singhrao SK, Neal JW, Rushmere NK, Morgan BP, Gasque P. (1999). Differential expression of individual complement regulators in the brain and choroid plexus. Lab Invest. 79(10): 1247–1259. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10532588 (accessed October 8, 2014). Sleegers K, Brouwers N, Gijselinck I, et al. (2006). APP duplication is sufficient to cause early onset Alzheimer’s dementia with cerebral amyloid angiopathy. Brain. 129(Pt 11): 2977–2983. doi: 10.1093/ brain/awl203. Spampinato MV, Rumboldt Z, Hosker RJ, Mintzer JE. (2011). Apolipoprotein E and gray matter volume loss in patients with mild cognitive impairment and Alzheimer disease. Radiology. 258(3): 843–852. doi: 10.1148/radiol.10100307. Sperling R. (2011). Potential of functional MRI as a biomarker in early Alzheimer’s disease. Neurobiol Aging. 32(Suppl 1): S37–43. doi: 10.1016/j.neurobiolaging.2011.09.009. Sperling RA, Aisen PS, Beckett LA, et al. (2011). Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7(3): 280–292. doi: 10.1016/j.jalz.2011.03.003. Sperling RA, Rentz DM, Johnson KA, et al. (2014). The A4 study: stopping AD before symptoms begin? Sci Transl Med. 6(228): 228fs13. doi: 10.1126/scitranslmed.3007941. Stevens BW, DiBattista AM, William Rebeck G, Green AE. (2014). A gene-brain-cognition pathway for the effect of an Alzheimer׳s risk gene on working memory in young adults. Neuropsychologia. 61: 143–149. doi: 10.1016/j.neuropsychologia.2014.06.021. Strittmatter WJ, Weisgraber KH, Huang DY, et al. (1993). Binding of human apolipoprotein E to synthetic amyloid beta peptide: isoform-specific effects and implications for late-onset Alzheimer disease. Proc Natl Acad Sci U S A. 90(17): 8098–8102. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcg i?artid=47295&tool=pmcentrez&rendertype=abstract (accessed August 26, 2014). Struyfs H, Molinuevo JL, Martin J-J, De Deyn PP, Engelborghs S. (2014). Validation of the AD-CSF-index in autopsy-confirmed Alzheimer’s disease patients and healthy controls. J Alzheimers Dis. 41(3): 903–909. doi: 10.3233/JAD-131085. Suthana NA, Krupa A, Donix M, et al. (2010). Reduced hippocampal CA2, CA3, and dentate gyrus activity in asymptomatic people at genetic risk for Alzheimer’s disease. Neuroimage. 53(3):1077–1084. doi: 10.1016/j.neuroimage.2009.12.014. Tanaka J-I, Horiike Y, Matsuzaki M, Miyazaki T, Ellis-Davies GCR, Kasai H. (2008). Protein synthesis and neurotrophin-dependent structural plasticity of single dendritic spines. Science. 319(5870): 1683–1687. doi: 10.1126/science.1152864.
Taylor JL, Scanlon BK, Farrell M, et al. (2014). APOE-epsilon4 and aging of medial temporal lobe gray matter in healthy adults older than 50 years. Neurobiol Aging. 35(11): 2479–2485. doi: 10.1016/j. neurobiolaging.2014.05.011. TDP-43 prelim findings in AD from AAIC. Available at: http://www.alz. org/aaic/releases_2014/wed-830am-research-from-aaic.asp. Tebar F, Bohlander SK, Sorkin A. (1999). Clathrin assembly lymphoid myeloid leukemia (CALM) protein: localization in endocytic-coated pits, interactions with clathrin, and the impact of overexpression on clathrin-mediated traffic. Mol Biol Cell. 10(8): 2687–2702. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2550 0&tool=pmcentrez&rendertype=abstract. Accessed October 9, 2014. Teipel SJ. (2003). Relation of corpus callosum and hippocampal size to age in nondemented adults with down’s syndrome. Am J Psychiatry. 160(10): 1870–1878. doi: 10.1176/appi.ajp.160.10.1870. Teipel SJ, Alexander GE, Schapiro MB, Möller H-J, Rapoport SI, Hampel H. (2004). Age-related cortical grey matter reductions in non-demented Down’s syndrome adults determined by MRI with voxel-based morphometry. Brain. 127(Pt 4): 811–824. doi: 10.1093/ brain/awh101. Thambisetty M, An Y, Tanaka T. (2013). Alzheimer’s disease risk genes and the age-at-onset phenotype. Neurobiol Aging. 34(11): 2696. e1–5. doi: 10.1016/j.neurobiolaging.2013.05.028. Thomas JB, Brier MR, Bateman RJ, et al. (2014). Functional connectivity in autosomal dominant and late-onset Alzheimer disease. JAMA Neurol. 71(9): 1111–1122. doi: 10.1001/jamaneurol.2014.1654. Thompson PM, Stein JL, Medland SE, et al. (2014). The ENIGMA Consortium: large-scale collaborative analyses of neuroimaging and genetic data. Brain Imaging Behav. 8(2): 153–182. doi: 10.1007/ s11682-013-9269-5. Thompson WK, Hallmayer J, O’Hara R. (2011). Design considerations for characterizing psychiatric trajectories across the lifespan: application to effects of APOE-ε4 on cerebral cortical thickness in Alzheimer’s disease. Am J Psychiatry. 168(9): 894–903. doi: 10.1176/appi. ajp.2011.10111690. Thordardottir S, Ståhlbom AK, Ferreira D, et al. (2014). Preclinical cerebrospinal fluid and volumetric magnetic resonance imaging biomarkers in Swedish familial Alzheimer’s disease. J Alzheimers Dis. doi:c10.3233/JAD-140339. Toledo JB, Arnold SE, Raible K, et al. (2013). Contribution of cerebrovascular disease in autopsy confirmed neurodegenerative disease cases in the National Alzheimer’s Coordinating Centre. Brain. 136(Pt 9): 2697–2706. doi: 10.1093/brain/awt188. Trachtenberg AJ, Filippini N, Ebmeier KP, Smith SM, Karpe F, Mackay CE. (2012). The effects of APOE on the functional architecture of the resting brain. Neuroimage. 59(1): 565–572. doi: 10.1016/j.neuroimage.2011.07.059. Trachtenberg AJ, Filippini N, Mackay CE. (2012). The effects of APOE-ε4 on the BOLD response. Neurobiol Aging. 33(2): 323–334. doi: 10.1016/j.neurobiolaging.2010.03.009. Vasiliou V, Vasiliou K, Nebert DW. (2009). Human ATP-binding cassette (ABC) transporter family. Hum Genomics. 3(3): 281–290. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi ?artid=2752038&tool=pmcentrez&rendertype=abstract (accessed September 26, 2014). Villemagne VL, Ataka S, Mizuno T, et al. (2009). High striatal amyloid beta-peptide deposition across different autosomal Alzheimer disease mutation types. Arch Neurol. 66(12): 1537–1544. doi: 10.1001/ archneurol.2009.285. Voineskos AN, Lerch JP, Felsky D, et al. (2011). The brain-derived neurotrophic factor Val66Met polymorphism and prediction of neural risk for Alzheimer disease. Arch Gen Psychiatry. 68(2): 198–206. doi: 10.1001/archgenpsychiatry.2010.194. Von Bohlen und Halbach O. (2010). Involvement of BDNF in age-dependent alterations in the hippocampus. Front Aging Neurosci. 2. doi: 10.3389/fnagi.2010.00036.
3 9 2 part vii : I maging G enetics of C ognitive A ging
Wang L, Roe CM, Snyder AZ, et al. (2012). Alzheimer disease family history impacts resting state functional connectivity. Ann Neurol. 72(4): 571–577. doi: 10.1002/ana.23643. Waring SC, Rosenberg RN. (2008). Genome-wide association studies in Alzheimer disease. Arch Neurol. 65(3): 329–334. doi: 10.1001/ archneur.65.3.329. Westlye ET, Hodneland E, Haász J, Espeseth T, Lundervold A, Lundervold AJ. (2012). Episodic memory of APOE ε4 carriers is correlated with fractional anisotropy, but not cortical thickness, in the medial temporal lobe. Neuroimage. 63(1): 507–516. doi: 10.1016/j.neuroimage.2012.06.072. Westlye ET, Lundervold A, Rootwelt H, Lundervold AJ, Westlye LT. (2011). Increased hippocampal default mode synchronization during rest in middle-aged and elderly APOE ε4 carriers: relationships with memory performance. J Neurosci. 31(21): 7775–7783. doi: 10.1523/ JNEUROSCI.1230-11.2011. Westlye LT, Reinvang I, Rootwelt H, Espeseth T. (2012). Effects of APOE on brain white matter microstructure in healthy adults. Neurology. 79(19): 1961–1969. doi: 10.1212/WNL.0b013e3182735c9c. White NS, Alkire MT, Haier RJ. (2003). A voxel-based morphometric study of nondemented adults with Down Syndrome. Neuroimage. 20(1): 393–403. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/14527599 (accessed October 1, 2014). Wisse LEM, Biessels GJ, Heringa SM, et al. (2014). Hippocampal subfield volumes at 7T in early Alzheimer’s disease and normal aging. Neurobiol Aging. 35(9):2039–2045. doi: 10.1016/j.neurobiolaging.2014.02.021. Wong DF, Rosenberg PB, Zhou Y, et al. (2010). In vivo imaging of amyloid deposition in Alzheimer disease using the
radioligand 18F-AV-45 (florbetapir [corrected] F 18). J Nucl Med. 51(6): 913–920. doi: 10.2967/jnumed.109.069088. Xia C-F, Arteaga J, Chen G, et al. (2013). [(18)F]T807, a novel tau positron emission tomography imaging agent for Alzheimer’s disease. Alzheimers Dement. 9(6): 666–676. doi: 10.1016/j. jalz.2012.11.008. Xu G, McLaren DG, Ries ML, et al. (2009). The influence of parental history of Alzheimer’s disease and apolipoprotein E epsilon4 on the BOLD signal during recognition memory. Brain. 132(Pt 2): 383–391. doi: 10.1093/brain/awn254. Xu C, Wang Z, Fan M, et al. (2010). Effects of BDNF Val66Met polymorphism on brain metabolism in Alzheimer’s disease. Neuroreport. 21(12): 802–807. doi: 10.1097/WNR.0b013e32833ccaf4. Yacoub E, Harel N, Ugurbil K. (2008). High-field fMRI unveils orientation columns in humans. Proc Natl Acad Sci U S A. 105(30): 10607–10612. doi: 10.1073/pnas.0804110105. Yu J-T, Li L, Zhu Q-X, et al. (2010). Implication of CLU gene polymorphisms in Chinese patients with Alzheimer’s disease. Clin Chim Acta. 411(19–20): 1516–1519. doi: 10.1016/j.cca.2010.06.013. Zhang P, Qin W, Wang D, et al. (2014). Impacts of PICALM and CLU variants associated with Alzheimer’s disease on the functional connectivity of the hippocampus in healthy young adults. Brain Struct Funct. doi: 10.1007/s00429-014-0738-4. Zigman W, Silverman W, Wisniewski HM. (1996). Aging and Alzheimer’s disease in Down syndrome: Clinical and pathological changes. Ment Retard Dev Disabil Res Rev. 2(2): 73–79. doi: 10.1002/(SICI)1098-2779(1996)2:23 .0.CO;2-Y.
chapter 2 2 : N euroimaging G enetics of A lzheimer ’ s D isease
393
PA RT V I I I . IMAGING GENETICS AND MULTI-LOCUS MODELS
23. IMAGING GENETICS OF THE HYPOTHALAMICPITUITARY-ADRENAL AXIS I M P L I CAT I ONS F OR P SYCH OPATHOLOGY
Nadia S. Corral-Frías, Lindsay J. Michalski, Christina R. Di Iorio, and Ryan Bogdan
S
tress, which occurs when perceived resources are unable to meet perceived demands, is among the most potent predictors of mental illness (Heim and Nemeroff 2001; Bogdan et al. 2013b). Its predictive power transcends diagnostic categories and may account for comorbidity across psychiatric disorders, as well as other forms of disease (e.g., cardiovascular disease, autoimmune disease, cancer; Dong et al. 2004; Kittleson et al. 2006; Dube et al. 2009; Keinan-Boker et al. 2009; Miller et al. 2011; Penninx et al. 2013). Stress triggers a biological-behavioral cascade governed by a corticolimbic neural network that drives neuroendocrine, autonomic, and immune systems to confront environmental challenges (Herman et al. 2003; Ulrich-Lai and Herman 2009). While these systems typically encourage adaptive responses to environmental challenges, many forms of psychopathology (e.g., depression, anxiety, post-traumatic stress disorder, schizophrenia) are characterized by dysregulation within them (Kathol et al. 1989; Yehuda et al. 1991; McEwen 1998; Ehlert et al. 2001; Chrousos 2009). The neuroendocrine hypothalamic-pituitary-adrenal (HPA) axis is a central coordinator of stress responsiveness that both readies the body to respond to challenge and facilitates its return to homeostasis after the stressor has passed (Herman and Cullinan 1997; Herman et al. 2003). In addition to well-documented associations with various forms of psychopathology (Ehlert et al. 2001; Stetler and Miller 2011), emerging research suggests that abnormal HPA axis function plays a causal role in the development of psychiatric illness. Consistent with non-human animal models demonstrating that HPA axis manipulation (e.g., gene knockout, pharmacologic challenge) can induce anxiety- and depressive-like behaviors (e.g., Müller and Keck 2002; Kolber and Muglia 2009; Marques et al. 2009),
pharmacologic challenge and prospective epidemiological treatment outcome studies (in which corticosteroids are administered; e.g., asthma treatment) conducted in humans indicate that chronic HPA axis stimulation can induce severe psychiatric symptoms that remit when treatments ends (Patten and Neutel 2000; Wada et al. 2000; Fardet et al. 2012). Moreover, successful treatment of stress-related psychiatric disorders normalizes HPA axis function (Ising and Holsboer 2007), prompting hope that novel treatments targeting this system may be therapeutically efficacious (de Kloet et al. 2007; Thomson and Craighead 2007; Millan 2009; Otte et al. 2010). Encouraged by these putative causal links, a logical step toward developing a more comprehensive mechanistic understanding of stress-related psychopathology is to use multiple methodologies to explore potential pathways through which variability in HPA axis function emerges and how these differences may confer vulnerability to adversity. Such an approach has the potential to not only identify individuals who may be more vulnerable to stress, but also to identify novel treatment targets. Imaging genetics, known more broadly as neurogenetics, which integrates the fields of genetics, neuroscience, psychology, and psychiatry, attempts to link variability in the genome (i.e., differences in DNA sequence and/or signatures of epigenetic regulation) to brain circuit structure, function, and connectivity, with the ultimate goal of understanding the mechanisms through which differences in human behavior and risk for psychopathology arise. The application of this approach to the HPA axis is in relative infancy but has already expanded our understanding of how genetic and environmental factors contribute to stress-related psychopathology. However, the field is met with many challenges. In addition to difficulties familiar 397
to genetics, neuroscience, psychology, and psychiatry (e.g., small effects, the absence of detailed mechanisms, and the need to translate findings into clinically useful information; for a review of these challenges in the context of imaging genetics research, see Bogdan et al., 2013a), HPA axis imaging genetics research faces several unique challenges, including (1) the integration of the environment (through measurement or laboratory manipulation), (2) the examination of mediating neuroendocrine mechanisms, and (3) heightened replication concerns in light of the inclusion of environmental and genetic measures. In this chapter, we review recent interdisciplinary HPA-axis imaging genetics research, which is beginning to inform our understanding of the risk factors for stress-related psychopathology, and discuss approaches to confront its unique challenges. THE HPA A XIS The HPA axis operates primarily through a three-step hormonal cascade (for a detailed review, see Ulrich-Lai and Herman 2009). Upon the perception of stress, afferent projections from the medial prefrontal cortex (mPFC), hippocampus, amygdala, and brain stem stimulate the release of corticotropin-releasing hormone (CRH) from the paraventricular nucleus (PVN) of the hypothalamus. CRH binds to CRH type 1 (CRHR1) and type 2 (CRHR2) receptors in the anterior pituitary gland, triggering the secretion of adrenocorticotropic hormone (ACTH), which then binds to melanocortin receptors in the adrenal gland, facilitating the release of cortisol. Cortisol functions through a binary corticosteroid receptor system consisting of mineralocorticoid and glucocorticoid receptors. Because of its relatively high affinity for cortisol, the mineralocorticoid receptor (MR) is typically occupied with cortisol throughout the circadian cycle and provides a stable excitatory tone in the hippocampus, which, in turn, inhibits the HPA axis under basal and stressful circumstances (Reul et al. 2000; de Kloet et al. 2005). As such, cortisol-MR binding constrains the onset and progression of the stress response and governs initial behavioral reactions to stress exposure, such as appraisal (de Kloet et al. 2005; Joëls et al. 2009). In contrast, the glucocorticoid receptor (GR) has a relatively low affinity for cortisol; it becomes occupied following only large spikes in cortisol, such as those associated with circadian rhythms (e.g., the awakening response) or exposure to environmental stress. Within the hippocampus, the GR functions to inhibit continued HPA axis activity (though cortisol-GR binding in other regions, such as the amygdala, can potentiate the
HPA axis response; Kolber et al. 2008) and to facilitate a return to homeostasis after a stressor has passed. Generally, CRH and ACTH signaling primarily stimulate HPA axis activity, while cortisol-MR binding constrains the initial HPA axis response, and cortisol-GR binding returns the body to homeostasis. Cortisol binding to intracellular corticosteroid receptors is especially important because, once bound, these receptors can translocate to the nucleus. There, they function as transcription factors which can bind to sections of DNA, known as glucocorticoid response elements, to enhance or suppress the transcription of a wide range of genes (de Kloet et al. 2005). This mechanism allows for the HPA axis to extend its reach to a diverse array of proteins and likely contributes to the pleiotropic effects of stress, including subsequent vulnerabilty to psychopathology (Arloth et al. 2015). Notably, receptors for CRH, ACTH, and cortisol are expressed diffusely throughout the brain, and binding in regions outside the HPA axis has important consequences for stress-related behavior (Chrousos 2009; Joëls and Baram 2009). Moreover, while there are many additional factors that influence HPA axis function and are appealing candidates for further study (e.g., urocortins, vasopressin; Hauger et al. 2006), they have yet to be extensively investigated in imaging genetics research and hence are beyond the scope of this chapter. Variability at any step of the HPA axis cascade, from onset (e.g., CRH, ACTH, MR binding) to recovery (e.g., GR binding), can alter this highly regulated cycle and confer risk for the development of psychopathology (for a review of HPA axis associations with psychopathology, see de Kloet et al. 2005). For example, a wealth of research suggests that depression is characterized by a flattened diurinal cortisol response (i.e., blunted cortisol upon awakening and elevated cortisol throughout the day; Vreeburg 2009; Knorr et al. 2010). A blunted HPA axis response to stress coupled with impaired negative feedback results in prolonged—and potentially pathological—activation of the system, rendering it unable to adequately respond to acute challenges (Burke et al. 2005). ORIGINS OF VARIABILITY: GENETICS AND THE ENVIRONMENT
HPA axis function varies widely across individuals and is relatively stable over time (Márquez et al. 2005; Fox et al. 2006; Kudielka et al. 2009), suggesting that trait-like variation in HPA axis function may contribute to stable differences in stress response and thus vulnerability to stress-related psychopathology. Research across species has
3 9 8 part vii I : I maging G enetics and M ulti - L ocus M odels
demonstrated that both life experience, particularly in early age, and genetic architecture contribute to individual differences in diurnal and stress-evoked HPA axis function (Heim and Nemeroff 2001; Weaver et al. 2004; Tarullo and Gunnar 2006; McGowan et al. 2009; Kudielka and Wüst 2010; Miller et al. 2011; Heim and Binder 2012). Building upon twin research documenting that diurinal and stress-evoked HPA axis activity is moderately to largely heritable (Federenko et al. 2004), recent genetic association studies have begun to link genetic variation within the HPA axis cascade to individual differences in the system’s function (for a review, see DeRijk et al. 2006; DeRijk 2009; supplemental tables provided online: bogdanlab.com/ HPAaxisTables.html). Research has further documented associations between these same genetic variants and psychopathology, but typically only in the context of recent stressful life events or early-life adversity (Heim and Binder 2012; Mehta and Binder 2012). Supporting these gene x environment associations, rodent and non-human primate research has shown that stressful experiences can have long-lasting effects on HPA axis function (Lupien et al. 2009) that may arise from epigenetic regulation (Weaver et al. 2004; for a brief review on epigenetic regulation, see Chapter 18 in this volume). For example, in rodents, chronic early-life adversity (e.g., maternal separation) results in increased CRH receptor expression in the pituitary (Anisman et al. 1998) and a reduction of GR binding sites in the hippocampus (Liu 1997). Given the potentiating effect of CRH binding and the inhibitory role of hippocampal cortisol-GR binding on HPA axis activity, it is unsurprising that animals exposed to chronic early adversity have heightened basal HPA axis output and display atypical diurnal cortisol and ACTH output patterns (McEwen 2000; Solberg et al. 2001). Providing evidence for conservation across species, similar associations (i.e., elevated ACTH and cortisol and reduced GR expression in the hippocampus) have been found in non-human primates and humans exposed to adversity early in life (Heim and Nemeroff 2001; van der Vegt et al. 2009; Yehuda 2009; Tyrka et al. 2013), which may be the result of epigenetic mechanisms (McGowan et al. 2009). Importantly, these effects are most pronounced when adversity occurs earlier in life (i.e., prenatally until young adulthood) and in the context of enduring (as opposed to acute) adversity, although chronic later-life adversity experiences show modifications similar to those observed in early-life (Bogdan and Hariri 2012; Burghy et al. 2012; Gee et al. 2013; Zhang et al. 2013). Recent research provides evidence that through its effects on brain function, structure, and connectivity, HPA axis function mediates the
association between life experience (e.g., stress) and genetic differences with risk for pychopathology (Heim and Binder 2012; Mehta and Binder 2012). ASSOCIATIONS WITH BRAIN STRUCTURE, FUNCTION, AND CONNECTIVITY
A vast array of research has shown that stress is associated with individual differences in brain circuit structure, function, and connectivity that mirror those observed in psychopathology. For example, consistent with non-human animal research (Lupien et al. 2009; Rodrigues et al. 2009; Ulrich-Lai and Herman 2009), early-life adversity is associated with structural enlargement and increased threat-related activation of the amygdala, a subcortical structure critical for behavioral vigilance (Tottenham and Sheridan 2009; Tottenham et al. 2011; Bogdan et al. 2012). Notably, amygdala enlargement (Frodl et al. 2002; MacMillan et al. 2003; Barrós-Loscertales et al. 2006) and increased threat-related reactivity are found across anxiety and mood disorders (Siegle et al. 2002; Phan et al. 2006) and mediate anxiety-related behavior, such as eye gaze (Holmes et al. 2006; Fox et al. 2007). Despite many studies linking stress to individual differences in the brain, comparatively less research has examined how variation in HPA axis activity plays a role in such relationship and, furthermore, might mediate associations between stress and brain function in humans. A few notable exceptions compliment the rich foundation of non-human animal literature (e.g., Woolley et al. 1990; Magariños et al. 1996; Conrad et al. 1999; Kim and Diamond 2002; Vyas et al. 2002; Heine et al. 2004; Vyas et al. 2004; Rosenkranz et al. 2010; for review, see de Kloet et al. 2005; Dedovic et al. 2009a; Dedovic et al. 2009b; McEwen and Gianaros 2010). First, research has linked Cushing’s syndrome, which is characterized by excessive secretion of cortisol, to increased threat-related amygdala reativity (Maheu et al. 2008). Second, dysregulated HPA axis function in healthy adults has been associated with heightened amygdala and reduced ventromedial prefrontal cortex (vmPFC) activation during the regulation of negative affect; this provides an intriguing putative mechanism (i.e., deficient emotion regulation) through which HPA axis dysfunction may contribute to psychopathology (Urry et al. 2006). Third, a recent study in females suggests that greater early-life adversity is associated with heightened cortisol during childhood, which predicts reduced amygdala-vmPFC resting-state functional coupling and subsequent risk for stress-related disorders in young adulthood (Bogdan and Hariri 2012; Burghy et al. 2012) (Figure 23.1). Differences in HPA axis function that
chapter 2 3 : T he H ypothalamic P ituitary A drenal A xis
399
Childhood cortisol Hypothalamus
Anterior Pituitary
Anxiety and depression Cortisol Adrenal cortex
vmPFC Amygdala
Maternal stress during infancy
Amygdala-vmPFC resting-state functional connectivity
Figure 23.1 Biological pathways linking early life stress to later psychopathology.
In girls, a history of maternal stress during infancy is associated with heightened basal cortisol during childhood, which predicts reduced resting-state amygdala-vmPFC functional connectivity during adolescence, which, in turn, mediates the association between childhood cortisol and adolescent symptoms of anxiety and depression. Adapted from Bogdan R, Hariri AR (2012), Neural embedding of stress reactivity. Nat Neurosci. 15: 1605–1607, with permission from Nature Publishing Group.
originate in childhood can modify brain function in adolescence, effectively setting the stage for the development of psychopathology.
the mechanisms through which variability in HPA axis function affects brain circuit function, structure, and connectivity, ultimately shedding light on the etiology of stressrelated psychopathology.
SUMMARY
Collectively, research suggests that stressful experiences, particularly when they occur early in life, are associated with a hyperactive HPA axis, characterized by both blunted diurinal variation and impaired regulation. In turn, this dysregulation predicts individual differences in neural structure, function, and connectivity, mirroring disruptions found in psychopathology and those exposed to chronic adversity. Notably, recent research suggests that these individual differences in neural cicutry may mediate associations between HPA axis dysregulation and risk psychopathology. Emerging research indicates that genetic variation confers individual differences in HPA axis function and moderates the effects of stress on HPA axis integrity, as well as risk for stress-related psychopathology. By incorporating measures and manipulations of the environment, imaging genetics research is uniquely positioned to improve our understanding of the biological and environmental factors through which individual differences in HPA axis function emerge. Most important, the field is poised to elucidate
IMAGING GE NE T IC S S T U DIE S OF H PA AX IS POLYMOR PH IS MS Given well-documented associations between stress exposure and psychopathology, as well as dysregulation of the HPA axis across psychiatric disorders, understanding how genetic variation within the HPA axis and environmental exposure impact neural correlates of psychopathology holds transdiagnostic promise. However, to date, only a handful of studies have investigated how genetic variation in the HPA axis predicts brain circuit structure, function, and connectivity in humans, and even fewer have incorporated the effect of environmental exposure (Enoch et al. 2008; Chen et al. 2010; Bogdan et al. 2011; Bogdan et al. 2012; Hsu et al. 2012; Ridder et al. 2012; White et al. 2012; El-Hage et al. 2013; Fani et al. 2013) (see Tables 23.1 and 23.2). The majority of this work has focused on amygdala function and structure, which we highlight here (see Table 23.1). A brief summary of other neural phenotypes
4 0 0 part vii I : I maging G enetics and M ulti - L ocus M odels
TABLE 23.1
HPA AXIS GENETIC VARIANTS AND DIFFERENCES IN AMYGDALA STRUCTURE, FUNCTION, AND CONNECTIVITY
Gene and Polymorphism
Study and Methods
Results
Genetic profile including 19 single-nucleotide polymorphisms (SNPs) that moderate the effects of GR expression: rs10837328, rs11587682, rs12086155, rs12611262, rs12620091, rs1805059, rs2027349, rs2269799, rs2302585, rs2395891, rs2422008, rs4724213, rs4838884, rs4845391, rs4902033, rs7826635, rs8106959, rs921320, rs9858280
(Arloth et al. 2015) Sample: College students (n = 276) Neural phenotype: Amygdala reactivity during an emotional face-matching task Environmental phenotype: Self-reported childhood life adversity measured by the Childhood Trauma Questionnaire.
Genetic main effects: None Interactive effects: Higher genetic risk profile was associated with reduced right centromedial amygdala reactivity at high levels of childhood trauma. However, at low levels of childhood trauma, higher scores were associated with relative elevated amygdala reactivity.
Mineralocorticoid Receptor (NR3C2) rs5522
(Bogdan et al. 2012) Sample: Children (n = 279) Neural phenotype: Amygdala reactivity during an emotional face-matching task. Environmental phenotype: Self-reported childhood emotional neglect measured by the Childhood Trauma Questionnaire (Bernstein et al. 2003)
Genetic main effects: Prior childhood emotional neglect and the Val risk allele were associated with greater threat-related amygdala reactivity. Interactive effects: Val allele carriers had elevated amygdala reactivity relative to Iso allele homozygotes only at relatively low levels of emotional neglect. Iso allele homozygotes had a positive association between amygdala reactivity and prior childhood emotional neglect that was absent in Val allele carriers.
Genetic profile including 10 SNPs: Corticotropin Releasing Hormone Receptor 1 (CRHR1) rs4792887, rs110402, rs242941, rs242939, rs1876828; Mineralocorticoid Receptor (NR3C2) rs5522; Glucocorticoid Receptor (NR3C1) rs41423247, rs10482605, and rs1005295; FK506 binding protein 5 (FKBP5) rs1360780
(Pagliaccio et al. 2014) Sample: Children (n = 120) Neural phenotype: Amygdala and hippocampal volume in adolescence Environmental phenotype: In-person assessment interviews with participants and their parents/guardians (Preschool-Age Psychiatric Assessment, Childhood and Adolescent Psychiatric Assessment; Egger et al. 2006; Angold & Costello 2000) Environment manipulation: Experimental stress (Laboratory Temperament Assessment Battery; Goldsmith & Rothbart 1992)
Genetic main effects: Higher genetic risk profile scores predicted elevated cortisol response to acute experimental stress. Interactive effects: Higher profile scores in interaction with stressful life events during early childhood and predicted differences in left amygdala and hippocampal volume in adolescence. Cortisol response in early childhood mediated the interaction of profile scores and early-life events on amygdala volume.
Glucocorticoid Receptor (NR3C1) rs33389, rs6195, rs41423247, rs1005297, and rs4986593; Corticotropin Releasing Hormone Receptor 1 (CRHR1) rs1876831 and rs242938
(Ridder et al. 2012) Samples: Healthy participants (n = 60 and n = 52) Neural phenotype: Amygdala and rostrolateral prefrontal cortex reactivity and connectivity to classical aversive delayed cue fear-conditioning task
Genetic main effect: Minor alleles for rs33389, rs41423247, rs4986593 were collectively associated with increased left amygdala reactivity and connectivity with the left PFC during fear acquisition. Participants with at least two minor alleles of NR3C1 and CRHR1 SNPs had increased left amygdala and left PFC coupling during extinction.
FK506 binding protein 5 (FKBP5) rs7748266, rs1360780, rs9296158, rs3800373, rs9470080, and rs9394309
(White et al. 2012) Sample: Healthy European-American youth (n = 183) Neural phenotype: Amygdala reactivity during an emotional face-matching task Environmental phenotype: Self-reported childhood emotional neglect measured by the Childhood Trauma Questionnaire (Bernstein et al. 2003)
Genetic main effect: None. Interactive effects: Risk allele carriers (i.e., rs1360780 T allele, rs9296158 A allele, rs9470080 T allele, rs3800373 G allele, rs7748266 T allele, rs9394309 G allele, as well as haplotypes composed of risk alleles) had increased dorsal amygdala reactivity in the context of relatively elevated emotional neglect.
investigated within an imaging genetics HPA axis framework are summarized in Table 23.2. AMYGDALA
The amygdala is a central structure of a corticolimbic circuit coordinating physiological and behavioral vigilance to changing environmental contingencies (LeDoux 2000,
2007); thus, it is unsurprising that amygdala structure, function, and connectivity are linked to stress-related psychopathology (Phillips et al. 2003; Hariri 2009; Price and Drevets 2009). The amygdala drives HPA axis response to environmental demands. Further, it contains an abundance of CRH, melanocortin, and corticosteroid receptors through which the HPA axis can reciprocally regulate amygdala function, making it a prime candidate
chapter 2 3 : T he H ypothalamic P ituitary A drenal A xis
401
TABLE 23.2
HPA AXIS GENETIC VARIANTS AND DIFFERENCES IN STRUCTURE, FUNCTION, AND CONNECTIVITY IN OTHER NEURAL REGIONS
Gene and Polymorphism
Study and Methods
Results
Corticotropin Releasing Hormone Receptor 1 (CRHR1) rs12938031
(Bogdan et al. 2011) Sample: Healthy female participants (n = 84) Neural phenotype: feedback-related negativity while performing a reward learning task Environment manipulation: acute experimental stress (threat of shock)
Genetic main effects: None Interactive effects: A allele carriers demonstrated compromised response bias during acute stress and had enhanced and delayed feedback-related negativity and decreased ACC reactivity to reward receipt.
Corticotropin Releasing Hormone Receptor 1 (CRHR1) rs4076452, rs242937, rs242936, rs242947, rs110402, rs242924, rs1396862, rs878886, and rs878887
(Chen et al. 2010) Sample: Caucasians (n = 1049) case-control (alcohol dependence; n = 472) Neural phenotype: P3 amplitude during an information-processing task
Genetic main effects: Several polymorphisms (rs110402, rs242924, rs1396862, rs878886, and rs878887) were associated with P3 amplitude. The most common haplotype (including all 9 SNPs) was associated with decreased P3 amplitude.
Glucocorticoid Receptor (NR3C1) rs41423247, rs6198, and rs1866388
(El-Hage et al. 2013) Sample: Healthy Caucasians (n = 90) Neural phenotype: dorsolateral prefrontal cortex reactivity during working memory task
Genetic main effects: G allele carriers of rs4142324 had greater dorsolateral prefrontal cortex activation during the working memory task.
Corticotropin Releasing Hormone Binding Protein (CRHBP) rs32897, rs6453267, rs7728378, rs1875999, rs10474485, rs7704995 [now rs1715747], rs3792738, and rs1500
(Enoch et al. 2008) Sample: Plains Indian and Caucasian probands with family history of alcoholism (n = 373) Neural phenotype: resting EEG alpha waves
Genetic main effects: Risk-allele carriers (rs7728378, rs1875999, rs7704995, and rs1500) had reduced resting alpha oscillations in both Caucasian and Plains Indian samples. Haplotypes (rs7728378, rs1875999, rs10474485, rs7704995, and rs1500) for Plains Indians and (rs7728378, rs1875999, rs7704995, and rs1500) for Caucasian group were also associated with decreased alpha.
FK506 binding protein 5 (FKBP5) rs1360780
(Fani et al. 2013) Sample: Trauma-exposed African-American women (n = 36) Neural phenotype: hippocampal morphology and reactivity to a threat-bias task
Genetic main effects: T allele carriers had significant differences in hippocampal morphology (i.e., greater displacement in CA1), as well as increased hippocampal reactivity to threat-bias.
Corticotropin Releasing Hormone Receptor 1 (CRHR1) rs110402
(Hsu et al. 2012) Sample: Case-Control (n = 83 healthy participants; n = 16 unmedicated MDD patients) healthy participants; n = 99; Neural phenotype: ROI analyses of subgenual cingulate cortex, hypothalamus, amygdala, and ventral striatal reactivity during an emotional word-stimulus task
Genetic main effects: MDD A carriers demonstrated reduced reactivity compared to controls in the hypothalamus and left nucleus accumbens. Among G homozygotes, subgenual cingulate cortex reactivity was greater in MDD patients.
Glucocorticoid Receptor (NR3C1) rs41423247
(Montag et al. 2013) Sample: Healthy Caucasians (n = 254) Neural phenotype: hippocampal and amygdala volume Environmental phenotype: inferred solar activity during first trimester of gestation
Genetic main effects: No main effects found. Interaction effect: C allele carriers exposed to high solar activity while in the womb (first trimester) had relatively larger left hippocampal volumes and lower neuroticism scores. No effects were found in amygdala.
for imaging genetics studies of the HPA axis (LeDoux 2007). Recent research has shown that genetic variation associated with functional differences in the HPA axis moderates amygdala structure and threat-related reactivity (Bogdan et al. 2012; Hsu et al. 2012; Ridder et al. 2012; White et al. 2012). MINERALOCORTICOID RECEPTOR (NR3C2)
Inspired by non-human animal research documenting the anxiolytic and antidepressant properties of
mineralocorticoid receptor expression in the amygdala (Mitra et al. 2009), Bogdan et al. (2012) examined a functional missense Iso/Val polymorphism (rs5522) located in exon 2 of the mineralocorticoid receptor (MR)gene (NR3C2). The Val allele at this locus has been associated with a loss of function with regard to cortisol (but not aldosterone), and presumably, reduced MR-cortisol binding which inhibits the HPA axis (DeRijk et al. 2006; van Leeuwen et al. 2010). Consistent with these findings, the Val allele has also been associated with blunted cortisol diurinal variation and heightened stress reactivity, as indexed
4 0 2 part vii I : I maging G enetics and M ulti - L ocus M odels
by endocrine, autonomic, and self-report measures (DeRijk et al. 2006; Bogdan et al. 2010; van Leeuwen et al. 2010). Even after controlling for significant main effects of childhood emotional neglect and Val allele carrier status, an interaction between prior childhood emotional neglect and MR genotype predicted threat-related amygdala reactivity in a sample of 279 children (Figure 23.2). Specifically, there was a positive association between emotional neglect and threat-related amygdala reactivity in Iso allele homozygotes. In contrast, Val-allele carriers had increased amygdala reactivity relative to Iso allele homozygotes, but only in the context of relatively low childhood adversity. Thus, even in the context of low prior adversity, Val allele carriers display neural patterns similar to those of maltreated individuals. This may reflect a physiological ceiling in which the presence of one or more Val alleles prevents maltreatment from further exacerbating threat-related amygdala reactivity through disruption of the HPA axis. Additionally, it suggests that Iso allele homozygotes may be more sensitive to environmental circumstances, including both social adversity and support. However, Val allele carriers may be more vulnerable to the development of stress-related illness, even in the absence of significant environmental stress. Notably, emerging research has shown that, along with rs2070950, rs5522 forms a haplotype that is further predictive of differential MR function and risk for psychopathology (van Leeuwen et al. 2010; Klok et al. 2011b). This haplotype may help explain conflicting reports of the relationship between the rs5522 variant in single SNP analyses of HPA axis function and psychopathology (Klok et al. 2011a; van Leeuwen et al. 2011) and allow future research to more accurately characterize functional consequences of variation in the MR gene.
Right Amygdala Reactivity
2.0 Iso/Iso Val Carrier
1.5 1.0 0.5 0.0
–5
0
5
10
15
CTQ (mean centered) Figure 23.2 Mineralocorticoid receptor rs5522 genotype interacts
with childhood emotional neglect to predict individual differences in threat-related amygdala reactivity. Adapted from Bogdan R, Hyde LW, Hariri AR (2013a), A neurogenetics approach to understanding individual differences in brain, behavior, and risk for psychopathology. Mol Psychiatry. 18: 288–299, with permission from Nature Publishing Group.
FK506 BINDING PROTEIN 5 (FKBP5)
HPA axis imaging genetics studies have also examined polymorphisms within the FK506 binding protein 5 (FKBP5) gene (White et al. 2012) that affect HPA axis function and/or risk for stress-related psychopathology (Binder et al. 2004; Binder et al., 2008; Touma et al. 2011; Menke et al. 2013). Briefly, FKBP5 is a co-chaperone that facilitates cortisol-GR binding and translocation of the GR complex to the nucleus, hence mediating the effects of cortisol on widespread gene transcription (Binder 2009). Elevated FKBP5 protein levels are associated with reduced GR sensitivity to cortisol, leading to decreased regulation (via negative feedback) of the HPA axis and a slower return to homeostasis following the resolution of a stressor (for a review see Binder 2009). Non-human animal research has shown that increased FKBP5 levels are associated with anxietylike behaviors (Attwood et al. 2011), and genetic polymorphisms associated with differences in FKBP5 expression have been linked to risk for depression and PTSD in the context of environmental adversity (Binder et al. 2008). In a study by White and colleagues (2012), several FKBP5 polymorphisms in high linkage disequilibrium (i.e., rs3800373, rs9296158, rs7748266, rs1360780, rs9394309, rs9470080) predicted heightened threat-related dorsal amygdala reactivity in the context of early-life adversity (White et al. 2012). This finding was specific to the dorsal amygdala, which is consistent with animal literature suggesting that stress up-regulates FKBP5 expression in the dorsal, but not ventral, region of the amygdala (Scharf et al. 2011). Unlike the ventral amygdala which primarily receives sensory, hippocampal, and prefrontal input, the dorsal amygdala projects to the brain stem, hypothalamus, and prefrontal cortex. These dorsal afferents drive autonomic and neuroendocrine systems, as well as attention and vigilance, to respond to environmental challenges (Davis and Whalen 2001; LeDoux 2007). Thus, the interaction of FKBP5 genotype and early adversity on dorsal amygdala reactivity may reflect sensitized responses to, rather than sensory perceptions of, threat. Furthermore, preliminary research has replicated this interaction in an independent sample of ethnically-heterogeneous young adults (n = 334; Di Iorio et al. 2013). STUDIES OF VARIANTS ACROSS MULTIPLE GENES
The vast majority of imaging genetics research has examined the association between single polymorphic loci and individual differences in neural phenotypes. In light of small effects typically conferred by single polymorphisms and the proposed utility of multi-locus approaches
chapter 2 3 : T he H ypothalamic P ituitary A drenal A xis
403
in traditional psychiatric research (International Schizophrenia Consortium et al. 2009), imaging genetics research has begun to incorporate multi-locus genetic profiles (Nikolova et al. 2011; Holmes et al. 2012; Bogdan et al. 2013a; Bogdan et al. 2013b; Pagliaccio et al. 2014; Di Iorio et al. submitted). Very broadly, this approach can be qualified into two primary scoring schemes: (1) summation of risk alleles or weighted effects, providing a “risk” score, or (2) a biologically-informed score based upon previously reported associations with gene function or downstream consequences representing the function of a biological system (e.g., the HPA axis). To our knowledge, three imaging genetics studies of the HPA axis have used this approach. Ridder and colleagues (2012) recently examined how amygdala reactivity during fear acquisition and extinction is moderated by several genetic polymorphisms within the glucocorticoid receptor and corticotropin-releasing hormone receptor type 1 genes (NR3C1, CRHR1). In two small independent samples (sample 1, n = 60; sample 2, n = 52), an increasing number of minor alleles across NR3C1 SNPs (rs41423247, rs33389, rs4986593) predicted elevated amygdala reactivity to cues during fear conditioning and increased amygdala-vmPFC connectivity. The authors further found that combining SNPs across NR3C1 (rs6195, rs10052957) and CRHR1 (rs242938, rs1876831) predicted blunted ventral PFC response during fear extinction, potentially reflecting deficiencies in extinction. However, this combined profile did not predict differences in physiological (i.e., skin conductance) or affective (i.e., self-report of emotional valence and arousal) stress responses. Because functional consequences of SNPs included in this profile are unclear, it is not presently possible to speculate on mechanisms underlying these associations. Nonetheless, these data suggest that common genetic variation across the HPA axis collectively influences corticolimbic circuit function, with potentially important consequences for psychopathology risk. A longitudinal study of children (n = 120) has recently shown that an HPA axis genetic risk profile predicts cortisol response to an acute stressor, which in turn mediates a relationship between the genetic profile and later childhood amygdala volume (Pagliaccio et al., 2014). The risk profile comprised 10 SNPs [CRHR1 (rs4792887, rs110402, es242941, rs242939, rs1876828), NR3C2 (rs5522), NR3C1 (rs41423247, rs10482605, rs10052957), and FKBP5 (rs1360780)] associated with HPA axis function and/or depressive phenotypes. Consistent with the literature, the profile predicted elevated cortisol response to acute stress. Interestingly, the profile interacted with stressful life events during early childhood to predict later differences in amygdala (as well as hippocampal)
volume measured during ages 7–12, which is consistent with a cortisol-mediated mechanism predicting later structural differences (Burghy et al. 2012). Finally, Arloth et al. (2015) identified genes that are differentially expressed following pharmacologic glucocorticoid stimulation (likely through transcription factor effects). A profile was comprised of polymorphisms that both moderated the effects of GR stimulation on gene expression and were independently associated with depression at nominally-significant, alpha levels (p < 0.05). This profile predicted individual differences in threat-related amygdala reactivity, as well as risk for mood and anxiety disorders. Specifically, higher genetic-profile scores (indicating increased risk for depression and GR-mediated changes in gene expression) were associated with increased amygdala reactivity to neutral stimuli and increased disorder risk in the context of elevated levels of early-life adversity. Thus suggesting genetic variation that indirectly influences the regulatory effects of HPA axis function may impact corticolimbic function (i.e., lead to overgeneralization) and, consequently, impart risk for psychopathology. It will be important for future studies to examine gene expression changes in response to acute stress manipulations concurrent with the collection of neuroimaging data to more directly investigate potential mediating mechanisms. SUMMARY
Emerging imaging genetics research has begun to document that HPA axis polymorphisms are linked to amygdala function and structure, providing putative neural mechanisms by which psychopathology may arise. Longitudinal and interdisciplinary studies provide critical insights into how differences in cortisol response and GR-related gene expression may affect the interaction between early-life adversity and amygdala structure and function. However, the field still lacks a clear mechanistic understanding of how, precisely, genetically mediated differences in HPA axis function across individuals lead to changes in neural function and structure that may confer risk for psychopathology. C H ALLE NGE S OF IMAGING GE NE T IC S H PA AX IS R E S E AR C H Imaging genetics research holds great promise to identify mechanisms through which individual differences in neural circuit structure, function, and connectivity emerge. However, looking forward, this research must address a host of challenges, including adequately incorporating
4 0 4 part vii I : I maging G enetics and M ulti - L ocus M odels
the environment, evaluating putative mediational mechanisms, and assessing whether effects are replicable. In the following sections, we highlight these challenges and discuss ways in which the field is presently confronting them. INTEGRATING THE ENVIRONMENT
Given the well-documented effects of environmental exposure on HPA axis function, as well as evidence that HPA axis polymorphisms confer risk for psychopathology under certain environmental contexts, it is critical for future research to incorporate environmental measures and manipulations (McCrory et al. 2011). In human research, the assessment of environmental exposure is fraught with difficulties, ranging from memory and information-processing biases (e.g., depression; Monroe et al. 2007; Monroe and Reid 2008) to the use of widely heterogeneous stress phenotypes (e.g., chronic early adversity, specific events within the past 6 months) and assessment methods (e.g., self-report checklist, clinician-administered interview) across studies (Monroe and Reid 2008). For example, there is only a modest correlation between self-reported versus clinician-assessed stressful life events (McQuaid et al. 1992; McQuaid et al. 2000; Monroe et al. 2007; Monroe and Reid 2008). Moreover, in self reports, individuals may endorse events that objective raters would not report as stressful. Numerous studies emphasize the importance of both an objective evaluation of stress (i.e., based on contextual factors) and an individual’s own subjective perception (Simons et al. 1993; McQuaid et al. 2000). Ideally, measures of the environment would be carefully selected to assess various aspects of stress exposure, including the subjective perception of stress and its objective contextual characterization, and, when possible, used within the context of prospective designs that can assess changes over time (Pagliaccio et al. 2014). Recent developments in naturalistic experience sampling methods among self-report measures (Wichers et al. 2009) and biological assesment of arousal (Ertin et al. 2011) may also prove beneficial. While stress manipulation is common in non-human animal models (Van Dijken et al. 1992; Stam et al. 2002; Louvart et al. 2005), there is a dearth of imaging genetics research that directly manipulates either the environment (Pagliaccio et al. 2014; Bogdan et al. 2011) or the HPA axis itself through pharmacologic techniques (Arloth et al. 2015). In addition to understanding how chronic earlylife and recent stressful experiences shape brain function, it is also important to understand the effects of acute stress manipulations and HPA axis stimulation. Emerging research suggesting that these acute manipulations can induce psychopathology-like changes in brain function
(Porcelli et al. 2012) emphasizes the importance of understanding variability in response across individuals. Finally, it is critical to consider theoretical work postulating that polymorphisms typically characterized as risk alleles by the diathesis-stress model may be more accurately envisioned as plasticity variants (Belsky et al. 2009; Pluess and Belsky 2013). In fact, this theory has begun to acquire empirical support: for instance, the short allele of the serotonin-transporter-linked polymorphic region (5-HTTLPR; a region of the SLC6A4 gene), which confers risk to the depressogenic effects of stress, can confer protective effects in the context of positive (or simply benign) enviroments (Bogdan et al., 2014 Hankin et al. 2011; Kochanska et al. 2011), suggesting that environmental exposure can differentially affect complex behavior for better or for worse. Thus, research incorporating environmental measures integrating both adversity and enrichment (e.g., social support) may be particularly useful for investigating individual variation in complex behavioral outcomes. This is underscored by the recently-described stress-buffering effects of positive enrivonmental circumstances (Cohen and Hoberman 1983; Chi 2001; Hyde et al. 2011). EXAMINING MEDIATING NEUROENDOCRINE MECHANISMS
Variability in neurochemical signaling, which reflects the functional consequences of genetic variation and environmental exposure, provides molecular clues to the origin of individual differences in brain and behavior (Hariri 2009). However, our understanding of the ways in which genetic variation and the environment influence neural phenotypes will remain limited until we begin to examine putative mediational mechanisms (Hyde et al. 2015). Individual differences in hormone function (e.g., ACTH, cortisol) and differences in gene expression following stress exposure or pharmacological HPA axis stimulation are proimising intermediary mechanisms. However, with few exceptions (Arloth et al. 2015; Pagliaccio et al. 2014), such data have not been typically examined in imaging genetics research. Despite the potential utility of such mediating constructs, important factors may also limit their explanatory potential. First, assessments of HPA axis function are limited to peripheral (i.e., blood and saliva) biological markers. Importantly, however, research suggests that hormonal outputs of the HPA axis can have regionally specific effects. Furthermore, HPA axis-generated ligand binding in regions outside the HPA axis itself (e.g., CRHR1 binding in limbic and prefrontal regions) may be critically important for psychiatrically relevant phenotypes (Hauger 2006). Our present
chapter 2 3 : T he H ypothalamic P ituitary A drenal A xis
405
ability to understand how these peripheral markers predict region-specific binding in humans is contingent upon the development of PET ligands that can bind to CRH and corticosteroid receptors. In recognition of this need, recent NIH program announcements (e.g., PA-08-137) have specifically encouraged the development of such ligands. Similarly, measuring gene expression differences is limited to measures from peripheral tissues. In light of evidence for regionally specific differences in gene expression and methylation which we are presently unable to examine in live humans, it will be important for convergent non-human animal work and postmortem human studies to examine regionally-specific gene expression. REPLICATION CONCERNS
There have been several robust and reliable associations reported within imaging genetics literature. For instance, influential early research found that threat-related amygdala reactivity depends upon 5-HTTLPR genotype (Hariri et al. 2002; Bertolino et al. 2005; Hariri et al. 2005; Heinz et al. 2007), which has been replicated in subsequent studies, including meta-analyses (Munafò et al. 2009) and non-human animal models (Caspi et al. 2010). However, excitement from initial imaging genetics research has been tempered by sober realizations that common genetic variation confers, at best, only a small effect on brain and behavior. For example, a recent imaging genetics study found that the combined effect of 122,072 variants nominally associated with depression (p < .10) accounted for only 2% of mPFC volume (Holmes et al. 2012). Small effects of single common variants present a major challenge to the field, as such weak pentrance is difficult to detect and likely to result in non-replication, especially in the small samples that compose most current imaging genetics studies. Due to increasing concerns of inadequate statistical power (McClelland and Judd 1993), research incorporating the environment enhances the challenge of small effects, particularly when interactions are driven by the extreme ends of a distribution. Recent reports suggest that GxE studies of psychiatric disorders have a high false-discovery rate and are tainted by both publication bias (i.e., a preference toward publishing positive results and novel findings rather than null results or replications) and low power due to small sample sizes (Duncan and Keller 2011). These concerns have led to increasing demands for large sample sizes to ensure appropriate statistical power, as well as direct replication studies that increase our confidence in novel findings. While recent preliminary work (DiIorio
et al. 2013) has replicated and extended the association between FKBP5 polymorphisms and amygdala function in the context of early-life adversity (White et al. 2012), the majority of HPA axis imaging genetic studies have not been replicated; whether this is the result of unpublished null replication attempts or untested replication is unknown. While inconclusive, some research suggests that effect sizes for intermediate phenotypes (e.g., measures of neural phenotypes or gene expression) are larger than effect sizes for psychiatric phenotypes, suggesting that imaging genetics research may have inherent enhanced power to detect significant associations (Goldman and Ducci 2007). Notably, independent and collaborative imaging genetics studies are beginning to approach sample sizes that will allow for adequate testing within larger samples, and ideally, replication efforts across samples (Duke Neurogenetics Study, www.haririlab.com; IMAGEN, http://www.imageneurope.com/; Brain Genomics Superstruct Project, http:// www.nmr.mgh.harvard. edu/nexus/; Human Connectome Project, http://www.humanconnectomeproject.org/; Neurodevelopmental Genomics Project, http://project reporter.nih.gov/project_info_description.cfm?aid¼794 3007&icde¼922023; Adult Health and Behavior Project, http ://www.chronicle.pitt.edu/media/pcc010924/ AHAB.html; Alcohol Outcomes Study, http://adolescent healthinstitute.com.). S U MMAR Y AND C ONC LU S IONS HPA axis imaging genetics research has begun to contribute to our understanding of how individual differences in HPA axis function emerge and contribute to stress-related psychopathology. Because there are currently only a few HPA axis imaging genetics studies, it will be important for future research to replicate these reports. Moreover, for the true strength of this research to be realized, it will need to integrate environmental measures and manipulations, as well as test mediational pathways that might underlie the effect of gene x environment interaction on brain circuitry. Finally, and like other forms of imaging genetics research, it will be important for this research to use longitudinal assessment to examine whether these individual differences predict the development of psychopathology. If these challenges can be met, HPA axis imaging genetics research promises to inform our etiologic understanding of stress-related psychopathology and, ultimately, contribute to knowledge that may lead to the development of novel and efficacious treatment and prevention methods.
4 0 6 part vii I : I maging G enetics and M ulti - L ocus M odels
R EF ERENCES Angold A & Costello EJ. (2000). The Child and Adolescent Psychiatric Assessment (CAPA). J Am Acad Child Adolesc Psychiatry. 39(1): 39–48. Anisman H, Zaharia MD, Meaney MJ, Merali Z. (1998). Do early-life events permanently alter behavioral and hormonal responses to stressors? Int J Dev Neurosci. 16: 149–164. Arloth J, et al. (2015). Genetic Differences in the Immediate Transcriptome Response to Stress Predict Risk-Related Brain Function and Psychiatric Disorders. Neuron. 86: 1189–1202. Attwood BK, Bourgognon J-M, Patel S, Mucha M, Schiavon E, Skrzypiec AE, Young KW, Shiosaka S, Korostynski M, Piechota M, Przewlocki R, Pawlak R. (2011). Neuropsin cleaves EphB2 in the amygdala to control anxiety. Nature. 473: 372–375. Barrós-Loscertales A, Meseguer V, Sanjuán A, Belloch V, Parcet MA, Torrubia R, Ávila C. (2006). Behavioral inhibition system activity is associated with increased amygdala and hippocampal gray matter volume: a voxel-based morphometry study. NeuroImage. 33: 1011–1015. Belsky J, Jonassaint C, Pluess M, Stanton M, Brummett B, Williams R. (2009). Vulnerability genes or plasticity genes? Mol Psychiatry. 14: 746–754. Bernstein DP, Stein JA, Newcomb MD, Walker E, Pogge D, Ahluvalia T, Stokes J, Handelsman L, Medrano M, Desmond D, Zule W. (2003). Development and validation of a brief screening version of the Childhood Trauma Questionnaire. Child Abuse Negl. 27(2): 169–90. Bertolino A, Arciero G, Rubino V, Latorre V, De Candia M, Mazzola V, Blasi G, Caforio G, Hariri A, Kolachana B, Nardini M, Weinberger DR, Scarabino T. (2005). Variation of human amygdala response during threatening stimuli as a function of 5’HTTLPR genotype and personality style. Biol Psychiatry. 57: 1517–1525. Binder EB, et al. (2004) Polymorphisms in FKBP5 are associated with increased recurrence of depressive episodes and rapid response to antidepressant treatment. Nat Genet. 36: 1319–1325. Binder EB. (2009). The role of FKBP5, a co-chaperone of the glucocorticoid receptor in the pathogenesis and therapy of affective and anxiety disorders. Psychoneuroendocrinology. 34(Suppl 1): S186–195. Binder EB, Bradley RG, Liu W, Epstein MP, Deveau TC, Mercer KB, Tang Y, Gillespie CF, Heim CM, Nemeroff CB, Schwartz AC, Cubells JF, Ressler KJ. (2008). Association of FKBP5 polymorphisms and childhood abuse with risk of posttraumatic stress disorder symptoms in adults. JAMA. 299: 1291–1305. Bogdan R, Agrawal A, Gaffrey MS, Tillman R, Luby JL. (2014). Serotonin transporter linked polymorphic region (5-HTTLPR) genotype and stressful life events interact to predict preschool-onset depression: a replication and developmental extension. J Child Psychol Psychiatry. 55: 448–457. Bogdan R, Hariri AR. (2012). Neural embedding of stress reactivity. Nat Neurosci. 15: 1605–1607. Bogdan R, Hyde LW, Hariri AR. (2013a). A neurogenetics approach to understanding individual differences in brain, behavior, and risk for psychopathology. Mol Psychiatry. 18: 288–299. Bogdan R, Nikolova Y, Pizzagalli D. (2013b). Neurogenetics of depression: a focus on reward processing and stress sensitivity. Neurobiol Dis. 52: 12–23. Bogdan R, Perlis RH, Fagerness J, Pizzagalli DA. (2010). The impact of mineralocorticoid receptor ISO/VAL genotype (rs5522) and stress on reward learning. Genes Brain Behav. 9: 658–667. Bogdan R, Santesso DL, Fagerness J, Perlis RH, Pizzagalli DA. (2011). Corticotropin-releasing hormone receptor type 1 (CRHR1) genetic variation and stress interact to influence reward learning. J Neurosci. 31: 13246–13254. Bogdan R, Williamson DE, Hariri AR. (2012). Mineralocorticoid receptor Iso/Val (rs5522) genotype moderates the association between previous childhood emotional neglect and amygdala reactivity. Am J Psychiatry. 169: 515–522.
Burghy CA, Stodola DE, Ruttle PL, Molloy EK, Armstrong JM, Oler JA, Fox ME, Hayes AS, Kalin NH, Essex MJ, Davidson RJ, Birn RM. (2012). Developmental pathways to amygdala-prefrontal function and internalizing symptoms in adolescence. Nat Neurosci. 15: 1736–1741. Burke HM, Davis MC, Otte C, Mohr DC. (2005). Depression and cortisol responses to psychological stress: a meta-analysis. Psychoneuroendocrinology. 30: 846–856. Caspi A, Hariri AR, Holmes A, Uher R, Moffitt TE. (2010). Genetic sensitivity to the environment: the case of the serotonin transporter gene and its implications for studying complex diseases and traits. Am J Psychiatry. 167: 509–527. Chen ACH, Manz N, Tang Y, Rangaswamy M, Almasy L, Kuperman S, Nurnberger J, O’Connor SJ, Edenberg HJ, Schuckit MA, Tischfield J, Foroud T, Bierut LJ, Rohrbaugh J, Rice JP, Goate A, Hesselbrock V, Porjesz B. (2010). Single-nucleotide polymorphisms in corticotropin releasing hormone receptor 1 gene (CRHR1) are associated with quantitative trait of event-related potential and alcohol dependence. Alcohol Clin Exp Res. 34: 988–996. Chou KL, Chi I. (2001). Stressful life events and depressive symptoms: social support and sense of control as mediators or moderators? Int J Aging Hum Dev. 52: 155–171. Chrousos GP. (2009). Stress and disorders of the stress system. Nat Rev Endocrinol. 5: 374–381. Cohen S, Hoberman HM. (1983). Positive events and social supports as buffers of life change stress. J Appl Soc Psychol. 13: 99–125. Conrad CD, LeDoux JE, Magariños AM, McEwen BS. (1999). Repeated restraint stress facilitates fear conditioning independently of causing hippocampal CA3 dendritic atrophy. Behav Neurosci. 113: 902–913. Davis M, Whalen PJ. (2001). The amygdala: vigilance and emotion. Mol Psychiatry. 6: 13–34. De Kloet ER, DeRijk RH, Meijer OC. (2007). Therapy insight: is there an imbalanced response of mineralocorticoid and glucocorticoid receptors in depression? Nat Clin Pract Endocrinol Metab. 3: 168–179. De Kloet, Joëls M, Holsboer F. (2005). Stress and the brain: from adaptation to disease. Nat Rev Neurosci. 6: 463–475. Dedovic K, D’Aguiar C, Pruessner JC. (2009a). What stress does to your brain: a review of neuroimaging studies. Can J Psychiatry Rev Can Psychiatr. 54: 6–15. Dedovic K, Duchesne A, Andrews J, Engert V, Pruessner JC. (2009b). The brain and the stress axis: The neural correlates of cortisol regulation in response to stress. NeuroImage. 47: 864–871. DeRijk RH. (2009). Single nucleotide polymorphisms related to HPA axis reactivity. Neuroimmunomodulation. 16: 340–352. DeRijk RH, Wust S, Meijer OC, Zennaro M-C, Federenko IS, Hellhammer DH, Giacchetti G, Vreugdenhil E, Zitman FG, de Kloet ER. (2006). A common polymorphism in the mineralocorticoid receptor modulates stress responsiveness. J Clin Endocrinol Metab. 91: 5083–5089. Di Iorio CR, Michalski LJ, Drabant EM, Bogdan R, Hariri AR. (2013, September). FKBP5 SNP rs1360780 and childhood adversity interact to predict amygdala reactivity: a replication. Poster session presented at: Twenty Seventh Annual Meeting of the Society for Research in Psychopathology; Oakland, CA. Dong, Giles WH, Felitti VJ, Dube SR, Williams JE, Chapman DP, Anda RF. (2004). Insights into causal pathways for ischemic heart disease: adverse childhood experiences study. Circulation. 110: 1761–1766. Dube SR, Fairweather D, Pearson WS, Felitti VJ, Anda RF, Croft JB. (2009). Cumulative childhood stress and autoimmune diseases in adults. Psychosom Med. 71: 243–250. Duncan LE, Keller MC. (2011). A critical review of the first 10 years of candidate gene-by-environment interaction research in psychiatry. Am J Psychiatry. 168: 1041–1049. Egger HL, Alaattin E, Keeler G, Potts E, Walter BK, Angold A. (2006). Test-Retest Reliability of the Preschool Age Psychiatric Assessment (PAPA). J Am Acad Child Adolesc Psychiatry. 45(5): 538–49.
chapter 2 3 : T he H ypothalamic P ituitary A drenal A xis
407
Ehlert U, Gaab J, Heinrichs M. (2001). Psychoneuroendocrinological contributions to the etiology of depression, posttraumatic stress disorder, and stress-related bodily disorders: the role of the hypothalamus-pituitary-adrenal axis. Biol Psychol. 57: 141–152. El-Hage W, Phillips ML, Radua J, Gohier B, Zelaya FO, Collier DA, Surguladze SA. (2013). Genetic modulation of neural response during working memory in healthy individuals: interaction of glucocorticoid receptor and dopaminergic genes. Mol Psychiatry. 18: 174–182. Enoch M-A, Shen P-H, Ducci F, Yuan Q, Liu J, White KV, Albaugh B, Hodgkinson CA, Goldman D. (2008). Common genetic origins for EEG, alcoholism and anxiety: the role of CRH-BP. Mansvelder HD, ed. Plos One. 3: e3620. Ertin E, Stohs N, Kumar S, Raij A, al’ Absi M, Shah S. (2011, November). AutoSense: unobtrusively wearable sensor suite for inferring the onset, causality, and consequences of stress in the field. Poster session presented at 9th International Conference on Embedded Networked Sensor Systems; Seattle, WA. Fani N, Gutman D, Tone EB, Almli L, Mercer KB, Davis J, Glover E, Jovanovic T, Bradley B, Dinov ID, Zamanyan A, Toga AW, Binder EB, Ressler KJ. (2013). FKBP5 and attention bias for threat: associations with hippocampal function and shape. JAMA Psychiatry. 70: 392–400. Fardet L, Petersen I, Nazareth I. (2012). Suicidal behavior and severe neuropsychiatric disorders following glucocorticoid therapy in primary care. Am J Psychiatry. 169: 491–497. Federenko IS, Nagamine M, Hellhammer DH, Wadhwa PD, Wüst S. (2004). The heritability of hypothalamus pituitary adrenal axis responses to psychosocial stress is context dependent. J Clin Endocrinol Metab. 89: 6244–6250. Fox E, Mathews A, Calder AJ, Yiend J. (2007). Anxiety and sensitivity to gaze direction in emotionally expressive faces. Emotion. 7: 478–486. Fox HC, Wilker EH, Kreek MJ, Sinha R. (2006). Reliability of salivary cortisol assessments in cocaine dependent individuals. J Psychopharmacol. 20: 650–655. Frodl T, Meisenzahl E, Zetzsche T, Bottlender R, Born C, Groll C, Jäger M, Leinsinger G, Hahn K, Möller H-J. (2002). Enlargement of the amygdala in patients with a first episode of major depression. Biol Psychiatry. 51: 708–714. Gee DG, Gabard-Durnam LJ, Flannery J, Goff B, Humphreys KL, Telzer EH, Hare TA, Bookheimer SY, Tottenham N. (2013). Early developmental emergence of human amygdala-prefrontal connectivity after maternal deprivation. Proc Natl Acad Sci. 110: 15638–15643. Goldman D, Ducci F. (2007). Deconstruction of vulnerability to complex diseases: enhanced effect sizes and power of intermediate phenotypes. ScientificWorld Journal. 7: 124–130. Goldsmith HH & Rothbart MK. (1992). The Laboratory Temperament Assessment Battery (LAB-TAB): Locomotor Version 3.0. University of Oregon, Dept. of Psychology, Eugene, OR, USA. Hankin BL, Nederhof E, Oppenheimer CW, Jenness J, Young JF, Abela JRZ, Smolen A, Ormel J, Oldehinkel AJ. (2011). Differential susceptibility in youth: evidence that 5-HTTLPR x positive parenting is associated with positive affect “for better and worse.” Transl Psychiatry. 1: e44. Hariri AR. (2009). The neurobiology of individual differences in complex behavioral traits. Annu Rev Neurosci. 32: 225–247. Hariri AR, Drabant EM, Munoz KE, Kolachana BS, Mattay VS, Egan MF, Weinberger DR. (2005). A susceptibility gene for affective disorders and the response of the human amygdala. Arch Gen Psychiatry. 62:146–152. Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D, Egan MF, Weinberger DR. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science. 297: 400–403. Hauger RL, Risbrough V, Brauns O, Dautzenberg FM. (2006). Corticotropin releasing factor (CRF) receptor signaling in the central nervous system: new molecular targets. CNS Neurol Disord Drug Targets. 5: 453–479.
Heim C, Binder EB. (2012). Current research trends in early life stress and depression: review of human studies on sensitive periods, gene-environment interactions, and epigenetics. Exp Neurol. 233: 102–111. Heim C, Nemeroff CB. (2001). The role of childhood trauma in the neurobiology of mood and anxiety disorders: preclinical and clinical studies. Biol Psychiatry. 49: 1023–1039. Heine VM, Maslam S, Zareno J, Joëls M, Lucassen PJ. (2004). Suppressed proliferation and apoptotic changes in the rat dentate gyrus after acute and chronic stress are reversible. Eur J Neurosci. 19:.131–144. Heinz A, Smolka MN, Braus DF, Wrase J, Beck A, Flor H, Mann K, Schumann G, Büchel C, Hariri AR, Weinberger DR. (2007). Serotonin transporter genotype (5-HTTLPR): effects of neutral and undefined conditions on amygdala activation. Biol Psychiatry. 61: 1011–1014. Herman JP, Cullinan WE. (1997). Neurocircuitry of stress: central control of the hypothalamo-pituitary-adrenocortical axis. Trends Neurosci. 20: 78–84. Herman JP, Figueiredo H, Mueller NK, Ulrich-Lai Y, Ostrander MM, Choi DC, Cullinan WE. (2003). Central mechanisms of stress integration: hierarchical circuitry controlling hypothalamo–pituitary–adrenocortical responsiveness. Front Neuroendocrinol. 24: 151–180. Holmes A, Richards A, Green S. (2006). Anxiety and sensitivity to eye gaze in emotional faces. Brain Cogn. 60: 282–294. Holmes AJ, Lee PH, Hollinshead MO, Bakst L, Roffman JL, Smoller JW, Buckner RL. (2012). Individual differences in amygdala-medial prefrontal anatomy link negative affect, impaired social functioning, and polygenic depression risk. J Neurosci. 32: 18087–18100. Hsu DT, Mickey BJ, Langenecker SA, Heitzeg MM, Love TM, Wang H, Kennedy SE, Pecina M, Shafir T, Hodgkinson CA, Enoch M-A, Goldman D, Zubieta J-K. (2012). Variation in the corticotropin-releasing hormone receptor 1 (CRHR1) gene influences fMRI signal responses during emotional stimulus processing. J Neurosci. 32: 3253–3260. Hyde LW, Bogdan R, Hariri AR. (2015). Neurogenetics of individual differences in brain, behavior and risk for psychopathology. In (Canli T, ed). The Oxford handbook of molecular psychology (pp. 223–242). NY, Oxford University Press. Hyde LW, Gorka A, Manuck SB, Hariri AR. (2011). Perceived social support moderates the link between threat-related amygdala reactivity and trait anxiety. Neuropsychologia. 49: 651–656. International Schizophrenia Consortium, Purcell SM, Wray NR, Stone JL, Visscher PM, O’Donovan MC, Sullivan PF, Sklar P. (2009). Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature. 460: 748–752. Ising M, Holsboer F. (2007). CRH-sub-1 receptor antagonists for the treatment of depression and anxiety. Exp Clin Psychopharmacol. 15: 519–528. Joëls M, Baram TZ. (2009). The neuro-symphony of stress. Nat Rev Neurosci. 10: 459–466. Joëls M, Krugers HJ, Lucassen PJ, Karst H. (2009). Corticosteroid effects on cellular physiology of limbic cells. Brain Res. 1293: 91–100. Kathol RG, Jaeckle RS, Lopez JF, Meller WH. (1989). Pathophysiology of HPA axis abnormalities in patients with major depression: an update. Am J Psychiatry. 146: 311–317. Keinan-Boker L, Vin-Raviv N, Liphshitz I, Linn S, Barchana M. (2009). Cancer incidence in Israeli Jewish survivors of World War II. J Natl Cancer Inst. 101: 1489–1500. Kim JJ, Diamond DM. (2002). The stressed hippocampus, synaptic plasticity and lost memories. Nat Rev Neurosci. 3: 453–462. Kittleson MM, Meoni LA, Wang N-Y, Chu AY, Ford DE, Klag MJ. (2006). Association of childhood socioeconomic status with subsequent coronary heart disease in physicians. Arch Intern Med. 166: 2356–2361. Klok, Giltay EJ, Van der Does AJW, Geleijnse JM, Antypa N, Penninx BWJH, de Geus EJC, Willemsen G, Boomsma DI, van Leeuwen N, Zitman FG, de Kloet ER, DeRijk RH. (2011a). A common and
4 0 8 part vii I : I maging G enetics and M ulti - L ocus M odels
functional mineralocorticoid receptor haplotype enhances optimism and protects against depression in females. Transl Psychiatry. 1: e62. Klok, Vreeburg SA, Penninx BWJH, Zitman FG, de Kloet ER, DeRijk RH. (2011b). Common functional mineralocorticoid receptor polymorphisms modulate the cortisol awakening response: interaction with SSRIs. Psychoneuroendocrinology. 36: 484–494. Knorr U, Vinberg M, Kessing LV, Wetterslev J. (2010.) Salivary cortisol in depressed patients versus control persons: a systematic review and meta-analysis. Psychoneuroendocrinology. 35: 1275–1286. Kochanska G, Kim S, Barry RA, Philibert RA. (2011). Children’s genotypes interact with maternal responsive care in predicting children’s competence: diathesis-stress or differential susceptibility? Dev Psychopathol. 23: 605–616. Kolber BJ, Muglia LJ. (2009). Defining brain region-specific glucocorticoid action during stress by conditional gene disruption in mice. Brain Res. 1293: 85–90. Kolber BJ, Roberts MS, Howell MP, Wozniak DF, Sands MS, Muglia LJ. (2008). Central amygdala glucocorticoid receptor action promotes fear-associated CRH activation and conditioning. Proc Natl Acad Sci. 105: 12004–12009. Kudielka BM, Hellhammer DH, Wüst S. (2009). Why do we respond so differently? Reviewing determinants of human salivary cortisol responses to challenge. Psychoneuroendocrinology. 34: 2–18. Kudielka BM, Wüst S. (2010). Human models in acute and chronic stress: Assessing determinants of individual hypothalamus–pituitary–adrenal axis activity and reactivity. Stress Int J Biol Stress. 13: 1–14. LeDoux J. (2007). The amygdala. Curr Biol. 17: R868–R874. LeDoux JE. (2000). Emotion circuits in the brain. Annu Rev Neurosci. 23: 155–184. Liu D. (1997). Maternal care, hippocampal glucocorticoid receptors, and hypothalamic-pituitary-adrenal responses to stress. Science. 277: 1659–1662. Louvart H, Maccari S, Ducrocq F, Thomas P, Darnaudéry M. (2005). Long-term behavioural alterations in female rats after a single intense footshock followed by situational reminders. Psychoneuroendocrinology. 30: 316–324. Lupien SJ, McEwen BS, Gunnar MR, Heim C. (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nat Rev Neurosci. 10: 434–445. MacMillan S, Szeszko PR, Moore GJ, Madden R, Lorch E, Ivey J, Banerjee SP, Rosenberg DR. (2003). Increased amygdala: hippocampal volume ratios associated with severity of anxiety in pediatric major depression. J Child Adolesc Psychopharmacol. 13: 65–73. Magariños AM, McEwen BS, Flügge G, Fuchs E. (1996). Chronic psychosocial stress causes apical dendritic atrophy of hippocampal CA3 pyramidal neurons in subordinate tree shrews. J Neurosci. 16: 3534–3540. Maheu FS, Mazzone L, Merke DP, Keil MF, Stratakis CA, Pine DS, Ernst M. (2008). Altered amygdala and hippocampus function in adolescents with hypercortisolemia: a functional magnetic resonance imaging study of Cushing syndrome. Dev Psychopathol. 20: 1177–1189. Marques AH, Silverman MN, Sternberg EM. (2009). Glucocorticoid dysregulations and their clinical correlates: from receptors to therapeutics. Ann N Y Acad Sci. 1179: 1–18. Márquez C, Nadal R, Armario A. (2005). Responsiveness of the hypothalamic-pituitary-adrenal axis to different novel environments is a consistent individual trait in adult male outbred rats. Psychoneuroendocrinology. 30: 179–187. McClelland GH, Judd CM. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychol Bull. 114: 376–390. McCrory E, De Brito SA, Viding E. (2011). The impact of childhood maltreatment: a review of neurobiological and genetic factors. Front Psychiatry. 2. Available at: http://www.frontiersin.org/Child_and_ Neurodevelopmental_Psychiatry/10.3389/fpsyt.2011.00048/ abstract (accessed October 28, 2013). McEwen BS. (1998). Stress, adaptation, and disease: allostasis and allostatic load. Ann N Y Acad Sci. 840: 33–44.
McEwen BS. (2000). The neurobiology of stress: from serendipity to clinical relevance. Brain Res. 886: 172–189. McEwen BS, Gianaros PJ. (2010). Central role of the brain in stress and adaptation: links to socioeconomic status, health, and disease. Ann N Y Acad Sci. 1186: 190–222. McGowan PO, Sasaki A, D’Alessio AC, Dymov S, Labonté B, Szyf M, Turecki G, Meaney MJ. (2009). Epigenetic regulation of the glucocorticoid receptor in human brain associates with childhood abuse. Nat Neurosci. 12: 342–348. McQuaid JR, Monroe SM, Roberts JE, Kupfer DJ, Frank E. (2000). A comparison of two life stress assessment approaches: prospective prediction of treatment outcome in recurrent depression. J Abnorm Psychol. 109: 787–791. McQuaid JR, Monroe SM, Roberts JR, Johnson SL, Garamoni GL, Kupfer DJ, Frank E. (1992). Toward the standardization of life stress assessment: Definitional discrepancies and inconsistencies in methods. Stress Med. 8: 47–56. Mehta D, Binder EB. (2012). Gene x environment vulnerability factors for PTSD: the HPA-axis. Neuropharmacology. 62: 654–662. Menke A, Klengel T, Rubel J, Brückl T, Pfister H, Lucae S, Uhr M, Holsboer F, Binder EB. (2013). Genetic variation in FKBP5 associated with the extent of stress hormone dysregulation in major depression. Genes Brain Behav. 12: 289–296. Millan MJ. (2009). Dual- and triple-acting agents for treating core and co-morbid symptoms of major depression: novel concepts, new drugs. Neurotherapeutics. 6: 53–77. Miller GE, Chen E, Parker KJ. (2011). Psychological stress in childhood and susceptibility to the chronic diseases of aging: moving toward a model of behavioral and biological mechanisms. Psychol Bull. 137: 959–997. Mitra R, Ferguson D, Sapolsky RM. (2009). Mineralocorticoid receptor overexpression in basolateral amygdala reduces corticosterone secretion and anxiety. Biol Psychiatry. 66: 686–690. Monroe SM, Reid MW. (2008). Gene-environment interactions in depression research: genetic polymorphisms and life-stress polyprocedures. Psychol Sci. 19: 947–956. Monroe SM, Slavich GM, Torres LD, Gotlib IH. (2007). Severe life events predict specific patterns of change in cognitive biases in major depression. Psychol Med. 37: 863–871. Montag C, Eichner M, Markett S, Quesada CM, Schoene-Bake J-C, Melchers M, Plieger T, Weber B, Reuter M. (2013). An interaction of a NR3C1 polymorphism and antenatal solar activity impacts both hippocampus volume and neuroticism in adulthood. Front Hum Neurosci. 7. Available at: http://www.frontiersin.org/Human_Neuroscience/10.3389/fnhum.2013.00243/abstract (accessed October 7, 2013). Müller MB, Keck ME. (2002). Genetically engineered mice for studies of stress-related clinical conditions. J Psychiatr Res. 36: 53–76. Munafò MR, Freimer NB, Ng W, Ophoff R, Veijola J, Miettunen J, Järvelin M-R, Taanila A, Flint J. (2009). 5-HTTLPR genotype and anxiety-related personality traits: a meta-analysis and new data. Am J Med Genet Part B. 150B: 271–281. Nikolova YS, Ferrell RE, Manuck SB, Hariri AR. (2011). Multilocus genetic profile for dopamine signaling predicts ventral striatum reactivity. Neuropsychopharmacology. 36: 1940–1947. Otte C, Hinkelmann K, Moritz S, Yassouridis A, Jahn H, Wiedemann K, Kellner M. (2010). Modulation of the mineralocorticoid receptor as add-on treatment in depression: a randomized, double-blind, placebo-controlled proof-of-concept study. J Psychiatr Res. 44: 339–346. Pagliaccio D, Luby JL, Bogdan R, Agrawal A, Gaffrey MS, Belden AC, Botteron KN, Harms MP, Barch DM. (2014). Stress-System Genes and Life Stress Predict Cortisol Levels and Amygdala and Hippocampal Volumes in Children. Neuropsychopharmacology. 39: 1245–1253. Patten SB, Neutel CI. (2000). Corticosteroid-induced adverse psychiatric effects: incidence, diagnosis and management. Drug Saf Int J Med Toxicol Drug Exp. 22: 111–122.
chapter 2 3 : T he H ypothalamic P ituitary A drenal A xis
409
Penninx BW, Milaneschi Y, Lamers F, Vogelzangs N. (2013). Understanding the somatic consequences of depression: biological mechanisms and the role of depression symptom profile. BMC Med. 11: 129. Phan KL, Fitzgerald DA, Nathan PJ, Tancer ME. (2006). Association between amygdala hyperactivity to harsh faces and severity of social anxiety in generalized social phobia. Biol Psychiatry. 59: 424–429. Phillips ML, Drevets WC, Rauch SL, Lane R. (2003). Neurobiology of emotion perception II: Implications for major psychiatric disorders. Biol Psychiatry. 54: 515–528. Pluess M, Belsky J. (2013). Vantage sensitivity: individual differences in response to positive experiences. Psychol Bull. 139: 901–916. Porcelli AJ, Lewis AH, Delgado MR. (2012). Acute stress influences neural circuits of reward processing. Front Neurosci. 6:157. Available at: http://www.frontiersin.org/Decision_Neuroscience/10.3389/ fnins.2012.00157/abstract (accessed October 28, 2013). Price JL, Drevets WC. (2009). Neurocircuitry of mood disorders. Neuropsychopharmacology. 35: 192–216. Reul JM, Gesing A, Droste S, Stec IS, Weber A, Bachmann C, Bilang-Bleuel A, Holsboer F, Linthorst AC. (2000). The brain mineralocorticoid receptor: greedy for ligand, mysterious in function. Eur J Pharmacol. 405: 235–249. Ridder S, Treutlein J, Nees F, Lang S, Diener S, Wessa M, Kroll A, Pohlack S, Cacciaglia R, Gass P, Schütz G, Schumann G, Flor H. (2012). Brain activation during fear conditioning in humans depends on genetic variations related to functioning of the hypothalamic-pituitary-adrenal axis: first evidence from two independent subsamples. Psychol Med. 42: 2325–2335. Rodrigues SM, LeDoux JE, Sapolsky RM. (2009). The influence of stress hormones on fear circuitry. Annu Rev Neurosci. 32: 289–313. Rosenkranz JA, Venheim ER, Padival M. (2010). Chronic stress causes amygdala hyperexcitability in rodents. Biol Psychiatry. 67: 1128–1136. Scharf SH, Liebl C, Binder EB, Schmidt MV, Müller MB. (2011). Expression and regulation of the Fkbp5 gene in the adult mouse brain. Plos One. 6: e16883. Siegle GJ, Steinhauer SR, Thase ME, Stenger VA, Carter CS. (2002). Can’t shake that feeling: event-related fMRI assessment of sustained amygdala activity in response to emotional information in depressed individuals. Biol Psychiatry. 51: 693–707. Simons AD, Angell KL, Monroe SM, Thase ME. (1993). Cognition and life stress in depression: cognitive factors and the definition, rating, and generation of negative life events. J Abnorm Psychol. 102: 584–591. Solberg LC, Olson SL, Turek FW, Redei E. (2001). Altered hormone levels and circadian rhythm of activity in the WKY rat, a putative animal model of depression. Am J Physiol Regul Integr Comp Physiol. 281: R786–794. Stam R, van Laar TJ, Akkermans LMA, Wiegant VM. (2002). Variability factors in the expression of stress-induced behavioural sensitisation. Behav Brain Res. 132: 69–76. Stetler C, Miller GE. (2011). Depression and hypothalamic- pituitary-adrenal activation: a quantitative summary of four decades of research. Psychosom Med. 73: 114–126. Tarullo AR, Gunnar MR. (2006). Child maltreatment and the developing HPA axis. Horm Behav. 50: 632–639. Thomson F, Craighead M. (2007). Innovative approaches for the treatment of depression: targeting the HPA axis. Neurochem Res. 33: 691–707. Tottenham N, Hare TA, Millner A, Gilhooly T, Zevin JD, Casey BJ. (2011). Elevated amygdala response to faces following early deprivation: neurodevelopment and adversity. Dev Sci. 14: 190–204. Tottenham N, Sheridan MA. (2009). A review of adversity, the amygdala and the hippocampus: a consideration of developmental timing. Front Hum Neurosci. 3(68): 1–18. Touma C, Gassen NC, Herrmann L, Cheung-Flynn J, Büll DR, Ionescu IA, Heinzmann J-M, Knapman A, Siebertz A, Depping A-M, Hartmann J, Hausch F, Schmidt MV, Holsboer F, Ising M, Cox MB,
Schmidt U, Rein T. (2011). FK506 binding protein 5 shapes stress responsiveness: modulation of neuroendocrine reactivity and coping behavior. Biol Psychiatry. 70: 928–936. Tyrka, Burgers DE, Philip NS, Price LH, Carpenter LL. (2013). The neurobiological correlates of childhood adversity and implications for treatment. Acta Psychiatr Scand. 128(6): 434–447. Ulrich-Lai YM, Herman JP. (2009). Neural regulation of endocrine and autonomic stress responses. Nat Rev Neurosci. 10: 397–409. Urry HL, van Reekum CM, Johnstone T, Kalin NH, Thurow ME, Schaefer HS, Jackson CA, Frye CJ, Greischar LL, Alexander AL, Davidson RJ. (2006). Amygdala and ventromedial prefrontal cortex are inversely coupled during regulation of negative affect and predict the diurnal pattern of cortisol secretion among older adults. J Neurosci. 26: 4415–4425. Van der Vegt EJM, van der Ende J, Kirschbaum C, Verhulst FC, Tiemeier H. (2009). Early neglect and abuse predict diurnal cortisol patterns in adults A study of international adoptees. Psychoneuroendocrinology. 34: 660–669. Van Dijken HH, Van der Heyden JA, Mos J, Tilders FJ. (1992). Inescapable footshocks induce progressive and long-lasting behavioural changes in male rats. Physiol Behav. 51: 787–794. Van Leeuwen N, Bellingrath S, de Kloet ER, Zitman FG, DeRijk RH, Kudielka BM, Wüst S. (2011). Human mineralocorticoid receptor (MR) gene haplotypes modulate MR expression and transactivation: Implication for the stress response. Psychoneuroendocrinology. 36: 699–709. Van Leeuwen N, Kumsta R, Entringer S, de Kloet ER, Zitman FG, DeRijk RH, Wüst S. (2010). Functional mineralocorticoid receptor (MR) gene variation influences the cortisol awakening response after dexamethasone. Psychoneuroendocrinology. 35: 339–349. Vreeburg SA. (2009). Major depressive disorder and hypothalamic-pituitary-adrenal axis activity: results from a large cohort study. Arch Gen Psychiatry. 66:617–626. Vyas A, Mitra R, Shankaranarayana Rao BS, Chattarji S. (2002). Chronic stress induces contrasting patterns of dendritic remodeling in hippocampal and amygdaloid neurons. J Neurosci. 22: 6810–6818. Vyas A, Pillai AG, Chattarji S. (2004). Recovery after chronic stress fails to reverse amygdaloid neuronal hypertrophy and enhanced anxiety-like behavior. Neuroscience. 128: 667–673. Wada K, Yamada N, Suzuki H, Lee Y, Kuroda S. (2000.) Recurrent cases of corticosteroid-induced mood disorder: clinical characteristics and treatment. J Clin Psychiatry. 61: 261–267. Weaver ICG, Cervoni N, Champagne FA, D’Alessio AC, Sharma S, Seckl JR, Dymov S, Szyf M, Meaney MJ. (2004). Epigenetic programming by maternal behavior. Nat Neurosci. 7: 847–854. White MG, Bogdan R, Fisher PM, Muñoz KE, Williamson DE, Hariri AR. (2012). FKBP5 and emotional neglect interact to predict individual differences in amygdala reactivity. Genes Brain Behav. 11: 869–878. Wichers M, Schrijvers D, Geschwind N, Jacobs N, Myin-Germeys I, Thiery E, Derom C, Sabbe B, Peeters F, Delespaul P, van Os J. (2009). Mechanisms of gene-environment interactions in depression: evidence that genes potentiate multiple sources of adversity. Psychol Med. 39: 1077–1086. Woolley CS, Gould E, McEwen BS. (1990). Exposure to excess glucocorticoids alters dendritic morphology of adult hippocampal pyramidal neurons. Brain Res. 531: 225–231. Yehuda R. (2009). Status of glucocorticoid alterations in post-traumatic stress disorder. Ann N Y Acad Sci. 1179: 56–69. Yehuda R, Giller EL, Southwick SM, Lowy MT, Mason JW. (1991.) Hypothalamic-pituitary-adrenal dysfunction in posttraumatic stress disorder. Biol Psychiatry. 30: 1031–1048. Zhang L, Zhang J, Sun H, Zhu H, Liu H, Yang Y. (2013). An enriched environment elevates corticosteroid receptor levels in the hippocampus and restores cognitive function in a rat model of chronic cerebral hypoperfusion. Pharmacol Biochem Behav. 103: 693–700.
4 1 0 part vii I : I maging G enetics and M ulti - L ocus M odels
INDEX
Aβ. See beta-amyloid ABCA7 (ATP-binding cassette, subfamily A, member 7), 378 ADRA2B,(adrenoceptor alpha 2B), 296–97t, 300 adrenocorticotropic hormone (ACTH), 398, 399 adverse life events. See life stress; lifetime stress; stressful life events age. See also cognitive aging; cognitive changes: age-related imaging genetics and advancing age, 8 SERT imaging and, 38 alcoholism, reward processes and, 178–79 alpha-2B-adrenergic receptor. See ADRA2B Alzheimer, Alois, 358 Alzheimer’s disease (AD), 357–58 family history of, 381–82 genetic causes of and risk genes for, 361–62t genetics of, 337–40 neuroimaging genetic risk for, 366–82 neuroimaging genetics of, 357–58 future directions and challenges, 382–83 relevance and impact, 384–86 neuroimaging highly penetrant genetic causes of, 360–66 pathological and clinical features of, 358–60 Alzheimer’s disease (AD) prevention, neuroimaging in the path to, 384–85, 385f Alzheimer’s disease (AD) risk genes. See also under Alzheimer’s disease; GWA studies: Alzheimer’s disease GWA-study-identified, 374–75 neuroimaging-identified, 379–81 amygdala HPA axis and, 400–402 5-HTTLPR and, 227, 228,270, 272, 275–77, 276f, 318, 406 life stress and, 89, 216 PTSD, ACC, and, 224–25, 228–29, 300 Williams syndrome and, 124–25, 129–31t amygdala-ACC connectivity/circuitry, 33, 84–86, 89–92, 210–12, 224 amygdala structure, function, and connectivity HPA axis genetic variants and, 400–401, 401t amygdala-vmPFC connectivity, 258, 399, 400f amyloid beta (AB), 333–34, 338, 339. See also amyloid-B; beta-amyloid Famyloid cascade hypothesis, 358–60, 365 amyloid precursor protein. See APP
anterior cingulate cortex (ACC), 89. See also amygdala-ACC connectivity/circuitry; subgenual anterior cingulate cortex anxiety disorders and, 224–26t, 227, 229 depression and, 41, 84–87, 170, 210, 211, 216 DLPFC and, 170 dorsal, 228 hippocampus and, 84–85 posterior cingulate cortex (PCC) and, 84, 85, 87 PTSD and, 224–25, 228–29 reward processing and, 170, 175–77 volume of, 216, 228, 229 antidepressant drug response in the brain, 85–87 imaging genetics in the context of variable, 87–91 variability of, 81–82, 91–92 antidepressants, classical, 85–87 antisocial behavior, 251 aversive emotional processing and, 252–53 brain imaging and neurobiology of, 252 function, 252–54 structure, 252 clinical features and assessment of, 251–52 neural mechanisms of genetic risk for, 258–59 reward, motivation, learning, and, 253 selective attention and, 253 serotonin and the genetic architecture of, 254–57 theory of mind (ToM), prosocial concern, and, 253–54 antisocial personality disorder (APD), 252 anxiety brain function in, 224 diagnosis and prevalence of, 223 genetics of, 223–24 imaging genetics of, 225–28 anxiety disorders 5-HTTLPR, 5-HTTLPR, and, 225t, 226–29 anterior cingulate cortex (ACC) and, 224, 225, 225–26t, 227, 229 functional imaging genetics of, 228 fear conditioning and, 229–30 imaging genetics in structural studies, 228–29 imaging genetics studies of brain structure in, 226t SERT binding and, 42
anxiety neuroimaging, meta-analysis of, 224–25 APOE (apolipoprotein E), 315 Alzheimer’s disease and, 242, 315, 337–39, 357, 367–74, 375f, 376, 378, 380–82 working memory and, 312t APOE2, 370 APOE3, 370 APOE E4, 315, 338. APOEepsilon4, (APOE ε4), 315, 357, 358, 367–73, 370f, 379, 382. APOE genotype, 242, 367–68, 370, 371f, 372, 373 effects on connectivity of resting state networks, 370, 371f APOE ε3 315, 367, 373, 380 working memory and, 312t apolipoprotein E gene. See APOE APP (amyloid precursor protein), 337, 358, 360 APP (amyloid precursor protein), 8, 358, 360, 365 Arloth, J., 404 ATK1, 72–74, 103 ATK1, 73, 103, 314 ATP-binding cassette, subfamily A, member 7 (ABCA7), 378 attention deficit hyperactivity disorder (ADHD), imaging response of pharmacological response in, 106 attention networks, cognitive control, and schizophrenia, 190–91 autism spectrum disorders (ASD), 67, 176 autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE) and D1R binding, 50 axial diffusitivity (AD), 331, 332, 339 Baddeley, Alan, 309 basal ganglia disorders, 241. See also specific disorders BDNF (brain-derived neurotrophic factor) Alzheimer’s disease and, 378–79 BDNF signaling in hippocampus, 85 depression and, 85, 213–14 neuroplasticity and, 90, 104, 214, 216, 298–99, 315, 340, 378 BDNF (brain-derived neurotrophic factor), 89–91 5-HTTLPR and, 216 antidepressant effects and, 85, 87, 89–92 BDNF Met allele, 104, 216 brain function and, 214 brain structure and, 214–15 depression and, 81, 104, 105, 213–17 memory and, 8, 298–99, 302–3, 312t, 315
Parkinson’s disease and, 245, 246 variants, 379 BDNF Val66Met (polymorphism), 40, 70, 315 anxiety disorders and, 225t, 226t links basic cellular mechanisms to brain structure and function, 340–41 depression and, 104–5, 214, 215 memory and, 245, 289, 291–93t, 302–3, 315, 340–41 Parkinson’s disease and, 245, 246 beta-amyloid (Aβ), 58, 358–60, 365, 366, 369, 372–74, 379, 381, 384. See also amyloid-B; amyloid beta Bilder, R. M., 173 binding potential (BP), 17, 17f bipolar disorder (BP), 42, 233 genetics and, 233–34 imaging genetics, 235–38 imaging genetics publications in, 142, 144–47t imaging response of pharmacological response in, 105 neuroimaging and, 234–35 Blasi, Giuseppe, 103 blood oxygen level dependent (BOLD) contrast imaging. See BOLDcontrast imaging Bogdan, Ryan, 282 BOLD-contrast imaging, 70, 72, 82, 125, 128, 196–97t, 244f, 301 S allele and, 33 BOLD fMRI, 70–71, 82, 224, 225t, 226, 246, 247, 341, 343, 376 BPND. See D2R binding brain structure, 163–64 common variant influence on, 161–62 crowdsourcing neuroimaging genetics through the ENIGMA Consortium, 162–63 genetic correlation and, 159 genetics and, 157 heritability of, 157–59 somatic variation and mosaicism affecting, 161 brain structure change, monogenic causes of, 159–61 bridge integrator 1 (BIN1), 377–78, 382 Brockmann, H., 211 Brunner, H. G., 254 Buckner, R. L., 9 Buhmann, C., 246 Burzynska, A. Z., 331
411
CACNA1C, 316–17 bipolar disorder and, 105, 195, 234–38 memory and, 293–94t, 299, 312t, 316–17 psychosis and, 193t, 196t schizophrenia and, 316, 317 Callicott, J. H., 190 calmodulin-binding transcription activator 1. See CAMTA1 [11C]AMT, 24 CAMTA1 (calmodulin-binding transcription activator 1 gene), 7, 291t, 297 Canli, T., 216 cannabinoid receptor 1 gene (CNR1), 104 cannabinoid (CB) receptors, 25, 177 cannabinoid (CB) signaling, 25. See also endocannabinoid signaling cannabis use and dependence, 177 Carter, C. S., 10 Caspi, A., 257, 268, 270 [11C]carfentanil binding, 23 CD33, 362t, 374, 378 [11C]DASB, 27, 36–42 Cerasa, A., 314–15 challenge phMRI, 82, 85, 87 [11C]harmine, 26 Chen, P. S., 58 Cheon, K. A., 106 Cleckley, Hervey, 251 CLU (clusterin), 339, 375–76, 376f, 383 [11C]McN5652 BP, 37–39, 41–43 CNR1 (cannabinoid receptor 1), 104 CNTR1, 177 cognition imaging genetics and, 7–8 cognitive ability individual differences in, 328 cognitive aging, 327–28, 345 brain function and, 334–35 cellular aspects of, 329–30 disrupted communication in, 341–43 future directions for, 345–46 genome-wide association studies for agesensitive cognitive abilities, 343–45 imaging genetics in the context of, 335–37, 336f, 341–45 examples, 337–41 individual variability in, and the role of genetics, 335 molecular and neurometabolic aspects of, 332–34 neurochemical aspects of, 334 neuronal substrates of, 329–35 macroscopic structural changes, 330–31 white matter integrity, 331–32 cognitive changes. See also cognitive aging age-related, 328–29 what accounts for variability in, 329 that occur with increasing age, 328 cognitive control, attention networks, and schizophrenia, 190–91 cognitive flexibility, 245 common disease/common variant hypothesis, 162 complement component (3b/4b) receptor 1 (CR1), 377 COMT (catechol-O-methyltransferase), 318, 341 nature of, 23, 212 COMT (catechol-O-methyltransferase), 23, 50, 59, 102, 215–16, 299, 313t, 313–14
bipolar disorder and, 236 brain function and, 212–13 brain structure and, 213–14 COMT-DAT1 interaction, 173 depression and, 212–15 dopamine D2 and D3 receptor availability and, 53–54 genotype, 2, 102, 199, 213, 243–45, 268, 314, 318, 343 methylation, 74–75, 317 Met/Met genotype, 50 nature of, 59, 172 reward processing and, 170–73 polymorphism, 50 Parkinson’s disease and, 242–46 schizophrenia and, 1, 2, 102 single nucleotide polymorphism (SNP) and, 1, 23, 72, 74, 171, 172 Val108/158Met genotype, 1 Val158Met allele, 227 Val158Met genotype, 1, 23, 53–54, 92, 102, 177, 294–95, 299, 342, 343 Val158Met polymorphism, 89 anxiety disorders and, 225t, 226t, 227–29 cognitive aging and, 340–43 dopamine and, 50, 53–55, 242, 299, 313 memory and, 102, 294–95t, 299, 313 Parkinson’s disease and, 59, 240, 242, 244 PTSD and, 228–29 reward processing and, 171–75 Val allele, 1, 2 Val/Met genotype, 1, 2 Val/Val genotype, 50 copy number variants (CNVs), 187, 198–99, 201 Corder, E. H., 357 corpus callosum (CC), 189 cortical volume and thickness, and schizophrenia, 188 corticobasal degeneration (CBD), 384 corticotropin-releasing hormone (CRH), 398, 399 cortisol, 398, 399, 400f, 402–4 [11C] Pittsburgh Compound-B (PIB), 333, 334 [11C]raclopride, 18, 21, 22, 51. See also dopamine D2 receptor (D2R) availability crowdsourcing neuroimaging genetics through the ENIGMA Consortium, 162–63 CTNNBL1, 295t, 299 CYP2D6 (cytochrome P450 2D6), 312t, 315, 318 CYP2D6 activity, 312t, 315 CYP2D6 genetic variation, 107 cytochrome P450 2D6. See CYP2D6 cytosine-adenosine-guanine (CAG), 247 D1R. See dopamine D1 receptor D2 long isoform (D2L), 70 D2R binding (BPND), 51–54, 58, 60. See also dopamine D2 receptor (D2R) availability D-amino acid oxidase activator gene (DAOA; G72), 314 DAOA (D-amino acid oxidase activator) gene, 314 DAT (dopamine transporter), 15 DAT1, 171–72, 178. See also DAT gene; SLC6A3 DAT1 VNTR, 171–74, 177, 178 DAT availability, 56
genes affecting, 56–58 studies on striatal, 56, 57t DAT/COMT interaction, 173 DAT gene, 56–58. See also DAT1; SLC6A3 3’ DAT VNTR, 72 studies on, 56, 57, 57t, 60, 72 DAT VNTR, 56, 57 polymorphism, 55, 72 DCX (doublecortin), 161 default mode network (DMN), 84–86, 364–65, 382 dementia. See also Alzheimer’s disease defined, 359 uncovering the molecular basis of increased risk for, 337–40 depression. See also antidepressant drug response anterior cingulate cortex (ACC) and, 41, 84–87, 170, 210, 211, 216 anxious, 104 depression, imaging genetics of, 209, 217 5-HTT and, 209–13, 215–17 beyond single genes, 215 gene-environment interaction, 216–17 gene-gene interactions, 215–16 polygenic burden, 215, 217 brain-derived neurotrophic factor (BDNF) and, 213–15 COMT, COMT, and, 212–15 future directions for, 217 imaging response of pharmacological response in depression, 103–5 SERT binding and, 41–42 D’Esposito, M., 301 development and gene-environment-brain-behavior relationships, 270, 279 and imaging genetics research, 273–74 diathesis-stress model, 269, 405 diffusion tensor imaging (DTI), 9, 83, 189, 235, 252, 331–32, 368, 369, 372 diffusion-weighted imaging (DWI), 9, 159, 163, 194, 331, 358, 361t, 366, 368, 369 DiGeorge syndrome deletion syndrome (22q11.2DS) functional neuroimaging studies in, 200–201 future neuroimaging studies in rare variants, 201 qualitative neuroimaging studies in, 199 quantitative structural neuroimaging studies in cortical morphology, 200 cross-sectional volumetric studies, 199 longitudinal volumetric studies, 199–200 white matter microstructure, 200 disease, imaging genetics and, 6–7 DNA methylation, 25, 69, 74, 75, 314. See also methylation Domschke, Katarina, 104, 227, 228 DOPA, 58 dopamine. See also under reward processing in human brain prefrontal cortex (PFC) and, 171–73, 242–46, 313 psychiatric disorders and, 50–53, 67–68 schizophrenia and, 68–74, 102, 103, 313 working memory and, 313–14 dopamine clearance genetics, reward processing and, 170–73 dopamine D1 receptor (D1R), 50, 313
dopamine D1 receptor (D1R) availability, 49–50 dopamine D2 receptor (D2R) availability, 50–54, 58, 60. See also D2 studies on, 51, 52t dopamine D2 receptor gene. See DRD2 dopamine D3 receptor (D3R), 60 dopamine D3 receptor (D3R) availability, 50 dopamine D4 receptor gene. See DRD4 dopamine receptor genetics, reward processing and, 168–70 dopamine receptor signaling genetic variation and D2, 69–74 imaging epigenetics of, 74–75 dopamine release, 54–56 dopamine synaptic terminal activity, imaging genetics of, 67–69 dopamine synthesis capacity, 58–59 dopamine transmission parameters, imaging of genetic variation impacting, 49, 60–61 dopamine transporter. See DAT dorsal raphe nucleus (DRN), 84 dorsolateral prefrontal cortex (dlPFC/ DLPFC), 8, 55, 142, 152, 170, 172, 196–97t, 198, 294t, 309–11, 312–13t, 314, 316, 376f doublecortin/doublin. See DCX Down syndrome (DS), 161 Alzheimer’s disease and, 365–66 DRD2 (dopamine D2 receptor), 22, 58 genetic variation and D2 dopamine receptor signaling, 69–74 polymorphisms related to, 51–53, 52t reward processing and, 169–70 schizophrenia and, 102, 103 single nucleotide polymorphism (SNP) and, 22, 55, 58, 69–73, 169, 170 DRD2 TaqIA, 51–53, 55, 58, 59, 70, 72, 103, 174, 242, 246 DRD2 TaqIA A1 allele, 51–52, 59, 60, 176, 246 DRD2 TaqIA A2 allele, 174, 246 DRD2 variation and genotype-genotype interaction, 72–74 DRD3. See dopamine D3 receptor DRD4 (dopamine receptor D4), 55, 106, 169, 170 alcoholism and, 178 reward processing and, 170 DRD4 VNTR, 55, 72, 174–75, 178 drug discovery, imaging genetics and, 4 drug use and addiction, 42. See also alcoholism reward processes and, 177 DTNBP1 (dystrobrevin-binding protein 1), 311, 312t, 313 dual syndrome hypothesis, 245 dynamic casual modeling (DCM), 10 dysbindin gene (DTNBP1), 311, 312t, 313 early life stress (ELS). See also life stress; stressful life events and depression, 212, 216 and later psychopathology, biological pathways linking, 400f ecstasy (MDMA), 42 Egan, M. F., 313 emotional challenge in anxiety disorder patients, functional brain activity during, 225t emotional episodic memory, 296–97t. See also episodic memory
4 1 2 I ndex
endocannabinoids (eCB), 25 endocannabinoid signaling, 175, 177. See also cannabinoid (CB) signaling endophenotypes, 68 Enhancing Neuro Imaging Genetics through Meta-Analysis (ENIGMA) Consortium, 162–63, 381 environment imaging genetics and the, 7 EphA1, 378, 382 epigenetics, 24–25 epilepsy, frontal lobe, 50 episodic memory, 304–5 consequences of different sensitivity of imaging vs. behavioral genetics studies, 302–4 genetic complexity of, 300–301 imaging, 289–90 imaging genotype-dependent differences in, 290 levels of analysis in genetic analysis of, 303f matched memory performance, 290, 297–98 imaging genetic studies using fMRI in healthy participants with, 291t neural compensation vs. encoding efficiency, 301–2 unmatched memory performance, 298–300 imaging genetic studies using fMRI in healthy participants with, 292–97t ethnicity and SERT imaging, 38–39 Eyler, L. T., 335 FAAH (fatty acid amide hydrolase), 25, 175, 177 FAAH 385A allelle, 175 FAAH C385A, 175, 177 A allelle of, 175 FAD (familial Alzheimer’s disease), 337, 360 FAD mutation carriers and noncarriers, 363f, 363–65 false discovery rate (fdr), 147 familial Alzheimer’s disease. See FAD fatty acid amide hydrolase. See FAAH 18-FDG. See 18 fluorodeoxyglucose (18-FDG) PET [18F]-DOPA, 59 fear conditioning and imaging genetics, 229–30 [18F]FDOPA (6-[18F]fluorodopa), 58 18 [ F]FDOPA uptake, 59 6-[18F]fluorodopa ([18F]FDOPA), 58 FKBP5 (FK506 binding protein 5), 385, 403, 406 18 fluorodeoxyglucose (18-FDG) PET (FDG-PET), 334, 365, 369, 370, 370f, 372, 379 fractional anisotropy (FA), 152, 189, 235, 311, 331, 364 Friedel, E., 211 frontal lobe epilepsy, autosomal dominant nocturnal, 50 frontotemporal lobar degeneration (FTLD), 359 F-spondin. See SPON1 functional magnetic resonance imaging (fMRI), 1, 312–13t BOLD fMRI, 70–71, 82, 224, 225t, 226, 246, 247, 341, 343, 376
I ndex
pharmacological MRI (phMRI), 81, 82, 84, 85, 87 structural MRI (sMRI), 234–37, 358, 361–62t, 363, 366–69 fusiform face area (FFA), 117, 122, 125, 126 fusiform gyrus, 117–18, 371 GABA (γ-aminobutyric acid) and working memory, 314 GABA receptors, 176 GABRA2, 176 GABRA2 SNP, 176 GAD1, 312t, 314, 317 Galaburda, A. M., 116 Gamo, N. J., 330 Gasic, G. P., 214 GBA (glucocerebrosidase), 241–42 gender and SERT imaging, 38 gene-environment correlation (rGE), 269–70, 274f, 279f gene-environment interactions (G x E) predicting behavior, 267–68, 317–18. See also imaging geneenvironment interactions (IG x E) research/studies challenges to progress and possible advances in G x E research, 269 consistency and specificity of G x E effects, 270–71 importance of development, 270, 279 moderators and potential confounds, 271 nature of environmental experience, 269–70 range of effects, 269 statistical approaches to strengthen G x E research, 271 three-way interactions, 270 conceptual and statistical diagram of G x E and imaging genetics studies, 273f generalized anxiety disorder (GAD), 223. See also anxiety disorders genes, experience, and the brain, 267 gene-set enrichment analysis (GSEA), 151 genetic discovery, imaging genetics and, 4–6 genetic effects on molecular neuroimaging. See also molecular neuroimaging genetics example mechanisms of, 19–20, 20f genetic imaging studies, problems of group comparisons in, 301 genetic risk score (GRS), 150, 198 genome-wide association (GWA) studies. See GWA studies Georgiou-Karistianis, N., 248 glucocerebrosidase (GBA) gene, 241–42 glucocorticoid receptor (GR), 281, 398, 399, 403, 404 glutamate and working memory, 311, 313 glutamate receptor, metabotropic gene (GRM3), 291t, 312–13t, 313–14 glutamic acid decarboxylase-67, 314. See also GAD1 glycogen synthase kinase-3. See GSK-3 GNB3 (GNβ3), 58 GRM3 (glutamate receptor, metabotropic), 291t, 312–13t, 313–14 group comparisons in genetic imaging studies, problems of, 301 GSEA (gene-set enrichment analysis), 151 GSK-3 (glycogen synthase kinase-3), 105
GSK3-Beta promoter gene variants, 105 GSK-3β (glycogen-synthase-kinase 3β), 72–74 GWA (genome-wide association) approach to gene discovery, 5 GWA studies (GWAS), 91, 141–43, 162, 337, 344. See also mental illness Alzheimer’s disease, 339–40, 373–83 bipolar disorder, 234–37 “crowdsourcing,” 162–63 genetic imaging studies from GWAS era for psychosis risk variants, 193–94t, 196–97t schizophrenia, 192, 197–98 working-memory-related genes identified by, 315–17 gyrification, 116, 117, 122, 188, 200 gyrification index (GI), 200, 331 Haas, B. W., 125 Hariri, Ahmad R., 214, 282 health phenotypes, 43 5-HIAA (5-hydroxyindoleacetic acid), 33, 254, 256, 257 high-affinity state, 18 hippocampal cortex, 121t Williams syndrome and, 123–24 hippocampal neuroplasticity, 85, 136, 214, 298–99 hippocampus anterior cingulate cortex (ACC) and, 84–85 BDNF signaling in, 85 frontal cortex and, 376, 376f Holmes, A. J., 217 HPA (hypothalamic-pituitary-adrenal) axis, 397–98, 406 associations with brain structure, function, and connectivity, 399–400 genetics, environment, and origins of variability of, 398–99 HPA axis genetic variants and differences in structure, function, and connectivity of neural regions, 400–401, 402t HPA axis polymorphisms, imaging genetics studies of, 400–404 studies of variants across multiple genes, 403–4 HPA axis research, imaging genetics challenges of, 404–5 examining mediating neuroendocrine mechanisms, 405–6 integrating the environment, 405 replication concerns, 406 5-HT (5-hydroxytryptamine), 31 reasons for studying, 31–32 5-HT(1A) gene (HTR1A), 19, 103–4 5-HT2A (serotonin 2A), 19, 40 5-HT2A receptor, 19, 25, 40 5-HT2A receptor (HTR2A), 19, 25, 275, 289 5-HT2A receptor binding, 19, 40 5-HT6 receptor, 333 5HT catabolism, 254–55 5-HT clearance, 256. See also SERT HTR1A (5-HT(1A)), 19, 103–4 HTR1b, 256–57 HTR2A, 19, 25, 272, 289 5-HT release, 256–57 5HT synthesis, 257 HTT (huntingtin), 247 5-HTT, 24. See also SERT depression and, 209–13, 215–17
5-HTTLPR amygdala and, 210–11, 227, 228 anxiety disorders and, 226–29 brain function and, 210–11 brain structure and, 211–12 depression and, 209–13, 215–16 gene-gene interactions and, 215–16 S allele of, 226, 228, 229 5-HTTLPR (serotonin-transporter-linked polymorphic region), 2, 32, 88–90, 210–12, 215–16, 226, 256, 258, 273, 276–77 5-HT4 binding and, 24 amygdala and, 270, 272, 275–77, 276f, 318, 406 depression and, 103, 104, 268, 270, 271, 405 5-HTTLPR polymorphism, 211–13, 275 anxiety disorders and, 225t 5-HTTLPR “short” allele, 256, 258, 405 Human Genome Project, 162 huntingtin gene. See HTT Huntington’s disease (HD), 50, 384 D1R and, 50 genetic imaging in, 247–48 Hyde, Luke W., 282 5-hydroxyindoleacetic acid (5-HIAA), 33, 254, 256, 257 5-hydroxytryptamine. See 5-HT hypothalamic-pituitary-adrenal axis. See HPA axis Ikeda, A., 105 IL1B (interleukin 1 beta), 104 imaging gene-environment interactions (IG x E) model(s), 274–75, 279f, 280–82 biological, 275f conceptual, 274f, 275 a study testing components of, 276–77 imaging gene-environment interactions (IG x E) research/studies, 274–75, 279f, 280–82. See also geneenvironment interactions (G x E) predicting behavior areas for progress in IG x E and neurogenetics studies, 277–78t challenges to progress and possible advances in IG x E, 277–80 future of, 281–82 plausible biological mechanisms and, 280–81 targets for future, 279f imaging genetics, 68, 272. See also specific topics and future applications, 9–10 imaging genetics research, challenges to progress and possible advances in, 272 drawing links from gene to brain to behavior, 272–73 modelling further complexity within imaging genetics, 273 need for greater emphasis on development, 273–74 impulsive antisociality, 252 independent component analysis (ICA), parallel, 154 inferior frontal gyrus (IFG), 190–91, 235, 236
413
inferior fronto-occipital fasciculus (IFOF), 125 inhibitory deficit theory of cognitive aging, 328 insular cortex, 131t interleukin 1 beta gene (IL1B), 104 intron 8 VNTR, 58 Jabbi, M., 126, 127 Jensen, K. P., 257 Kato, T., 105 KCNK2. See TREK1 ketamine, 87 KIBRA (kidney and brain expressed protein) gene and episodic memory, 290, 291t, 297, 298t, 344–45 Knöchel, C., 189 Krug, A., 300 Laine, T. P., 58 language networks, social cognition, and schizophrenia, 191 Lee, T., 332 LEP (leptin), 54, 55 life stress, 400f. See also diathesis-stress model; lifetime stress; stressful life events amygdala and, 89, 216 depression and, 212, 216, 268, 270, 271, 277 lifetime stress, 74, 317 likelihood estimation in graphical analysis (LEGA), 36 lissencephaly, 160 Logan, J., 36 Logan graphical analysis, 36, 38–40 low-affinity state, 18 MacDonald, A. W., 191 magnetic resonance imaging (MRI). See also functional magnetic resonance imaging MRI scans from identical twin pairs, 158f magnetic resonance spectroscopy (MRS), 334 magnetization transfer imaging (MTI), 332 magnetization transfer ratio (MTR), 332 major depressive disorder (MDD), 41, 81, 91, 209. See also depression diagnostic criteria for, 83 neurobiology of, 83–85 neuroimaging methods for, 82–83 MAO-A (monoamine oxidase A), 25–26, 254–55 MAO-A activity, 26, 59–60 MAOA gene, 59, 175–76, 257, 258 antisocial behavior and, 257 5HT catabolism and, 254–55 animal research on, 255 human research on, 255–56 MAOA genetic variation, 258 MAOA genotype, 257, 268 MAOA-H, 255–57 MAOA-L, 256–58 MAOA upstream VNTR (MAOA u-VNTR), 255 MAOA variants, 59 MAOA VNTR, 25–26, 59–60, 89, 228 MAO-B, 254–55
MAO-H, 256, 257 MAO-L, 256–58 marijuana. See cannabis use and dependence MCI APOEepsilon4 carriers, 369, 370f, 372 MDMA (3,4-methylenedioxy-methamphetamine), 42 mean diffusitivity (MD), 331, 332 Meaney, M. J., 281 medial frontal gyrus (MFG), 236 medial orbitofrontal cortex (OFC), 126, 224. See also orbitofrontal cortex medial prefrontal cortex (PFC), 178. See also prefrontal cortex medial temporal lobe (MTL), 246–47, 289–90, 298–300, 330–32, 366, 374–76 medical phenotypes, 43 Melville, S. A., 380 memory, 289. See also episodic memory; working memory prefrontal cortex and, 290, 309, 312t, 316 memory networks and schizophrenia, 190 memory performance and brain activity, prediction of by a gene score, 300–301, 301f mental illness, serious. See also psychiatric disorders; specific mental illnesses identifying unanticipated genes and mechanisms in, 141–42, 154–55 background, 142–43 brain circuitry as a quantitative trait, 152 gene network and pathway analysis, 151–52 imaging genetics statistics, 143, 146–51 multivariate methods, 154 parallel independent component analysis (ICA), 154 replication, 152–53 resting state, 153 structural and functioning neuroimaging, 143 mesencephalon, 157 metabotropic glutamate receptor (GRM3) gene, 291t, 312–13t, 313–14 methylation, 24, 25, 280. See also DNA methylation MAO-A and, 60 methylenetetrahydrofolate reductase (MTHFR), 213 methylphenidate, 106 Meyer-Lindenberg, A., 10, 125, 126, 258 microRNA 137 (MIR137 gene), 317 mild cognitive impairment (MCI) and Alzheimer’s disease (AD), 372, 379 mineralocorticoid receptor (MR), 398 mineralocorticoid receptor (MR) gene (NR3C2), 402–3 modulation phMRI, 82, 85 molecular neuroimaging genetics, 15–16, 27–28 examples of, 19–20 opportunities for future research on, 25–27 molecular neuroimaging genetic studies, examples of, 19–25 molecular neuroimaging methodology, 16–17 molecular neuroimaging tools, development of, 25–26 monoamine hypothesis of depression, 84
monoamine oxidase A. See MAO-A Montag, C., 214 Mosconi, L., 382 MRTM (multilinear reference tissue model), 35–36, 38–42 Mueller, S. C., 228 multilinear reference tissue model (MRTM), 35–36, 38–42 multi-locus models, imaging genetics and, 8–9 multimodal neuroimaging genetics, 28 multitracer protocols, 28–29 N-acetyl-aspartate (NAA), 334, 341 neuregulin 1 (NRG1) gene, 296t, 312– 13t, 314 neurochemistry, imaging genetics and, 3–4 neurocognitive disorders, 42 neuroendocrine mechanisms. See under HPA axis research neuroimaging techniques, 1, 3, 9. See also specific techniques neuroplasticity, 84, 268, 269, 327, 335, 337, 405 BDNF and, 90, 104, 214, 216, 298–99, 315, 340, 378 bipolar disorder and, 105 depression and, 89, 104 hippocampal, 85, 136, 214, 298–99 neurotrophins and, 213, 216, 340 schizophrenia and, 317, 319 synaptic, 89, 216, 340, 378 VNTR, 343 working memory, 343 neurotransmitter release, endogenous benefits and limitations of measuring, 20–22 neurotransmitter systems, genetic variation in, 341–43 neurotrophic factor, brain-derived. See BDNF neurotrophin family, genetic variation in, 340–41 neurotrophin hypothesis of depression, 89, 213, 216 neurotrophins, 213, 215, 216, 340 Nho, K., 379–80 nicotine addiction, 177 NMDA receptors, 311, 340 norepinephrine (NE), 254 norepinephrine reuptake inhibitors, 85, 86 NRG1 (neuregulin 1) gene, 296t, 312–13t, 314
O obesity, reward processes and, 176 Oertel, V., 190 one tissue compartment model (1TC), 35–37, 42 opiod neurotransmission, genetic sources of variability in, 22–23 μ-opioid receptor gene (OPRM1), 23, 178 OPRM1 (μ-opioid receptor gene), 23, 178 OPRM1 genotype, 23, 178 orbitofrontal cortex (OFC), 125–27, 169, 177–79, 224 oxytocin gene (OXT), 55–56 oxytocin receptor gene (OXTR), 176 panic disorder (PD), 223. See also anxiety disorders functional imaging genetics of, 227–28 parahippocampal gyrus, 90 parieto-occipital cortex, 119–21t
Park, D. C., 334–35 PARK1-13, 241–42 Parkinson’s disease (PD) COMT polymerism and, 242–45 genetic background to, 241–42 genetic influences on neurocognitive functioning in, 245–47 progression of cognitive deficits in early, 243, 243f serotonin system in, 42 underactivation of frontoparietal network in, 244, 244f pathway polygenic risk score (PPRS), 150–51 penetrance, 268. See also under Alzheimer’s disease PER2 (Period 2) gene, 54 personality research, 42 PET. See positron emission tomography Pezawas, Lukas, 210 P-glycoprotein (P-gp), 106, 107 pharmacological MRI (phMRI), 81, 82, 84, 85, 87 phenotypes, 5, 43, 195, 280. See also under SERT binding; Williams syndrome intermediate, 68 phMRI (pharmacological MRI), 81, 82, 84, 85, 87 PICALM, 376–77, 380 Pinel, Philippe, 251 PINK1 (PTEN-induced putative kinase 1), 242 PINK1 mutation, 246 Pittsburgh Compound-B (PIB), 333, 334 plasticity. See neuroplasticity polygenic burden, 215, 217 polygenic risk score (PRS), 147 polymorphism. See also specific topics genetic, 24 positron emission tomography (PET), molecular neuroimaging with, 16, 333–34 agonist and antagonist radiotracers and, 18, 18f methodology of, 16–17 relation between brain morphology and outcome in, 17–18 posterior-anterior shift in aging (PASA) model, 335 posterior cingulate cortex (PCC) dementia, Alzheimer’s disease, and, 364, 368–70, 372, 374, 380–82 depression and, 84, 85, 87 post-traumatic stress disorder (PTSD). See also anxiety disorders amygdala, ACC, and, 224–25, 228–29, 300 fear conditioning and, 226, 229 genetics of, 223 memory and, 297t, 300 prefrontal cortex (PFC), 90. See also dorsolateral prefrontal cortex; prefrontal cortex (PFC) dopamine; ventromedical prefrontal cortex (vmPFC/ VMPFC) depression and, 41, 42 dopamine and, 171–73, 242–46, 313 memory and, 290, 309, 312t, 316 Parkinson’s disease and, 242–46 psychosis and, 193–94t, 196–97t reward processing and, 171–73, 175–78 schizophrenia and, 314 prefrontal cortex (PFC) dopamine (DA), 242, 243, 245. See also prefrontal cortex: dopamine and
4 1 4 I ndex
presenilin 1 and presenilin 2. See PSEN1 and PSEN2 genes Preuschhof, C., 331 principal component analysis (PCA), 154 prion protein gene (PRNP), 7, 291t, 297 prosencephalon, 157 PSEN1 and PSEN2 genes, 337, 360 PSEN1 FAD mutation carriers and noncarriers, 363, 363f psychiatric disorders. See also mental illness; specific disorders dopamine and, 50–53, 67–68 imaging response of pharmacological response in, 101–2 future directions for, 106–7 psychopathology, 278t. See also mental illness; psychiatric disorders; specific disorders biological pathways linking early-life stress to later, 399–400, 400f psychopathy, 251–52 psychosis risk genes. See also under schizophrenia effects for neuroimaging phenotypes vs. behavioral markers, 195 PTEN-induced putative kinase 1. See PINK1 raclopride. See [11C]raclopride radial diffusitivity (RD), 331, 332 radiotracer binding sensitive to endogenous neurotransmitter release, models of, 20, 21f Raz, N., 330 Reed, J. D., 343 regional cerebral blood flow (rCBF), 224, 225t, 226, 339 Reiman, Eric, 371 Research Domain Criteria (RDoC) approach, 67, 92, 141, 272, 280 “researcher degrees of freedom,” 92 resting state networks and schizophrenia, 191 Reuter-Lorenz, P., 334–35 reward processes in clinical populations, imaging genetics of, 176–79 reward processing in human brain, imaging genetics of, 167–68 dopamine clearance genetics, 170–73 dopamine receptor genetics, 168–70 future disorders and considerations for, 179–80 multi-locus dopamine genetics profiles and, 174–75 non-dopamine genetics and, 175–76 reward system, primate, 167 rhombencephalon, 157 Ringner, M., 154 risk profile score (RPS), 147–48, 150 Rodrigue, K. M., 330 rs4680 (Val158Met), 74, 312–13t, 313, 313f, 314. See also COMT Val158Met genotype; COMT Val158Met polymorphism Salat, D. that., 332 Samanez-Larkin, G. R., 301 Sambataro, F., 343 sampling, 271, 279–80 Sarpal, D., 122, 125 schizophrenia (SZ), 141. See also specific topics bipolar disorder and, 233–37 CACNA1C and, 316–17
I ndex
D1R availability and, 50 D2R and, 51–53 dopamine and, 68–74, 102, 103, 313 dorsolateral prefrontal cortex and, 310 DTNBP1 and, 311 functional imaging for genetic liability for, 190–91 GABA, GAD1, and, 314 how genetic susceptibility for SZ has shaped our understanding of brain structure and function, 195, 197–98 imaging genetics of, 147–53, 187, 201–2 common variant studies, 192–98 corpus callosum, other fiber tracts, and, 189–90 cortical volume and thickness and, 188 family studies, 187–88, 192 hemispheric laterality and, 190 imaging response of pharmacological response in, 101–3 implications for imaging genetics research, 192 publications on, 142, 144–47t qualitative neuroimaging studies, 199 rare variant studies, 198–202 subcortical structures and, 188–89 MIR137 and, 317 polygenic, additive, and interactome models of, 198 applications in genetic imaging, 198 psychosis risk genes and functional neuroimaging abnormalities associated with, 194–95, 196–97t psychosis risk genes and structural abnormalities associated with, 192, 193–94t, 194 rare genetic variants in role of, 198–99 studies of, 198–202 single nucleotide polymorphism (SNP) and, 141, 142, 144–47t, 147, 150–54, 192, 198 ZNF804A and, 315–16 Schulz-Heik, R. J., 228–29 score alleles, 148 selective serotonin reuptake inhibitors (SSRIs), 43, 85–88 serotonin. See 5-HT serotonin hypothesis of depression, 41 serotonin reuptake inhibitors, selective, 43, 85–88 serotonin transporter (5-HTT) coding gene. See SLC6A4 gene serotonin transporter gene. See SERT serotonin-transporter-linked polymorphic region. See 5-HTTLPR SERT (serotonin transporter gene), 87–89, 256. See also SLC6A4 gene SERT (serotonin transporter), 104, 333. See also 5-HTT; SERT binding genetics of, 32–33 neuroreceptor imaging of genetic variation impacting future directions for, 43–44 PET imaging and, 31, 33–37, 42, 43 role in serotonin system, 31–32 SERT binding link between phenotypes of interest and, 40–43 challenges, 43 SLC6A4 variation on, 39
SERT binding estimates, interpreting, 33–37 SERT imaging research findings, 37–40 demographic variables and, 38–39 health-related variables and, 39 seasonality and, 39 SERT ligand comparison, 36–37 SERT modeling considerations, 35–36 SERT outcome measures, 34–35 Shen, L., 380 sialic acid binding immunoglobulin-like lectin-3. See CD33 Silver, M., 380 simplified tissue reference model (STRM), 35, 36 single nucleotide polymorphism (SNP), 23, 257, 300, 380, 401t, 404. See also Alzheimer’s disease; GWA studies AKT1 and, 73, 74 bipolar disorder and, 234–37 COMT and, 1, 23, 72, 74, 171, 172 D3R and, 53, 54 DAT and, 57 depression and, 104, 105 DRD2 and, 22, 55, 58, 69–73, 169, 170 DRD4 and, 169, 170 GSK-3β and, 74 in KIBRA gene, 298f psychosis, GWAS, and, 141–47t, 146–48, 148f, 149f, 150–54 psychosis risk variants and, 193–94t reward processing and, 175–79 schizophrenia and, 141, 142, 144–47t, 147, 150–54, 192, 198 TACR1, 179 SLC6A3 gene, 56–58, 106, 171–72, 343. See also DAT1 gene; DAT gene SLC6A4 gene, 24, 88, 209. See also 5-HTTLPR; SERT neural response to salient stimuli as a function of, 33 SERT binding and, 32, 39, 40, 42, 43 SLC6A4 genotype, 33, 39, 40 SLC6A4 polymorphism, 32, 39 SLC6A4 variation, 32, 33, 39, 42, 43 Smith, S. M., 9 Smolka, M. N., 215–16, 228 SNAP25 (synaptosomal-associated protein 25), 312t, 315 SNP-set enrichment analysis (SSEA), 151, 152 social anxiety disorder (SAD), 223. See also anxiety disorders functional imaging genetics of, 226–28 social cognition, language networks, and schizophrenia, 191 Sperling, Reisa, 364 SPON1 gene, 380–81 SSEA (SNP-set enrichment analysis), 151, 152 Steen, V. M., 105 Stefansson, Kari, 360 stressful life events, 32, 84, 88, 104, 105, 216, 271. See also life stress; lifetime stress structural equation modeling (SEM), 275 structural magnetic resonance imaging (sMRI), 234–37, 358, 361–62t, 363, 366–69 subgenual anterior cingulate cortex (sACC/sgACC), 55, 86, 87, 89, 216 amygdala and, 210, 224 depression and, 84, 210, 211, 216
DLPFC and, 55, 170 PTSD and, 224–25 substance use, 42. See also drug use and addiction synaptic neuroplasticity, 89, 216, 340, 378 synaptosomal-associated protein 25 (SNAP25), 312t, 315 Szobot, C. M., 106 tachykinin receptor 1 gene (TACR1), 178–79 TACR1 (tachykinin receptor 1 gene), 178–79 TaqIA, 69–70, 72, 169, 246. See also DRD2 TaqIA TaqIA polymorphism, 52t, 55, 58, 60, 102–3, 169, 242, 246 TaqIB, 51, 52, 52t, 53 1TC (one tissue compartment model), 35–37, 42 temporal lobe, medial. See medial temporal lobe temporo-occipital cortex, 118–19t theory of mind (ToM), 253 time activity curve (TAC), 16, 17 TNIK (TRAF2 and NCK-interacting kinase), 142 Tom40, 372 TOMM40, 358, 372–73, 380 TPH2 gene, 226–28, 315 TRAF2 and NCK-interacting kinase (TNIK), 142 translocator protein 18 kDa. See TSPO (translocator protein) 18 kDa TREK1 gene, 175 triggering receptor expressed on myeloid cells 2 (TREM2), 373–74 tryptophan hydroxylase (TPH), 257, 315. See also TPH2 gene TSPO (translocator protein) 18 kDa, 23 TSPO binding affinity and binding potential, genetic effects on, 23–24 twins, molecular neuroimaging in, 5, 19, 158 MRI scans from identical twin pairs, 158f uncinate fasciculus (UF), 125 Ursini, G., 317 Val158Met. See COMT Val158Met genotype; rs4680 Val158Met COMT polymorphism. See COMT Val158Met polymorphism van de Giessen, Elsmarieke, 57 variable number of tandem repeat. See VNTR ventromedical prefrontal cortex (vmPFC/ VMPFC), 211, 224, 227, 228, 253 Viding, E., 254 visual cortex, primary, 118–19t VNTR (variable number of tandem repeat), 32, 343. See also 5-HTTLPR intron 8, 58 3’ VNTR, 56, 57, 106 VNTR polymorphism, 88, 106, 169–73, 177, 255, 268, 343 voxel-based morphometry (VBM), 83, 126, 148f, 193–94t, 246, 311 voxels, 16, 82, 83 voxel selection bias, 90, 92 Wang, D, 24 Wang, M., 330
415
Weinberger, Daniel, 242 Williams-Gray, C., 245 Williams syndrome (WS), 113, 136 brain phenotype in, 115–36 clinical phenotype in cognitive profile, 114–15 medical presentation, 114 personality profile, 115 domain-general cognitive systems in, 127–28, 132 genetics of, 113–14 importance of research on, 113
neural systems findings in, 132, 136 neuroimaging literature on neural systems in, 132, 133–35t, 136 socio-emotional neural systems in, 124, 129–31t amygdala, 124–25 insular cortex, 127 orbitofrontal cortex, 125–27 visual neural systems in, 116 hippocampal cortex, 123–24 parietal-occipital cortex, 122–23 primary visual cortex, 116–17
temporo-occipital cortex, 117, 118–21t, 122 working memory, 309, 319 clinical relevance of, 318 definition of, 309 future outlook and direction for, 319 gene-environment interaction and, 317–18 genes associated with genome-wide association studies of, 315–17 heritability of, 310–11
limitations of imaging genetic studies of, 318 network connectivity and, 309–10 signaling systems and candidate genes for, 311, 313–15 working memory tasks, fMRI studies examining cortical activation during, 312–13t zinc-finger 804A gene (ZNF804A), 315–17 Zubieta, J. K., 23
4 1 6 I ndex