Bioanalytical Separations [volume 4]

Citation preview

Preface "Nothing tends so much as to the advancement of knowledge, as the application of a new instrument. The native intellectual powers of men in different times are not so much the causes of the different success of their labour, as the peculiar nature of the means and artificial resources in their possession"

Sir Humphrey Davy, 1840 In the context of this volume "bioanalysis" has a very specific meaning. It is the analysis of drugs and their metabolites in biological fluids. There is probably no field which has benefited from the introduction of new instrumentation and ways of analysing samples as much as this type of analysis. Today the laboratories of pharmaceutical companies, and the contract houses that serve them, are filled with sophisticated HPLC-MSMS systems devoted to the analysis of compounds at concentrations unachievable by previous generations of analysts. Thus, detection and quantification at concentrations below 1 nanogram per millilitre are now commonplace, and the pace of innovation seems still to be increasing. The bulk of these advances in bioanalysis have resulted from the development and implementation of robust and sensitive HPLC-MS interfaces. The high capital cost of much of this instrumentation has also led to increasing pressures on improving the efficiency of method development and instrument usage. The application of these new HPLC-MS-based methods is covered in this volume in chapters on forensic bioanalysis and the role of this way of analysis in drug discovery. However, despite the success of such devices, all of the problems of bioanalysis have not been solved by their introduction, and there is a continuing need for sustained innovation. In particular the low concentrations, and the presence of large amounts of endogenous interferences in biological fluids and tissues, has meant that sample preparation techniques remain of prime importance to the bioanalyst seeking the highest sensitivities and specificity. This activity is reflected in several contributions on the theme of sample preparation. Chromatography is still important, even when the detector is as sensitive and specific as the mass spectrometer is claimed to be, and new phases of the types described in this volume, providing chiral separations or improvements in conventional chromatography, are always needed. In addition the linking of spectrometers other than mass spectrometers has been an important area of innovation, and this is reflected in contributions on HPLC-NMR and HPLC-ICPMS. Finally, it has always to be remembered that, because of the importance of bioanalytical data in the generation of regulatory submissions, such work is subject to

vi

Preface

very precise and challenging regulatory control. The needs of the regulators for bioanalytical separations are therefore also covered in the final contribution to the volume. I would like to thank all of the authors who have contributed to this work for their time, patience and expertise. The collected wisdom and scholarship that these chapters reflect have been a source of considerable pleasure and education to me as I have put the volume together. Ian D. Wilson

AstraZeneca Macclesfield, U.K.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations,Vol. 4 92003 Elsevier Science B.V.All rights reserved

CHAPTER 1

New developments in integrated sample preparation for bioanalysis M.W.J. van Hout l'*, H.A.G. NiederRinder ~, R.A. de Zeeuw 1 and G.J. de Jong 2 Department of Analytical Chemistry and Toxicology, University Centrefor Pharmacy, University of Groningen, A. Deusinglaan 1, 9713 AV Groningen, The Netherlands 2 Department of Biomedical Analysis, Faculty of Pharmacy, University of Utrecht, P.O. Box 80082, 3508 TB Utrecht, The Netherlands * Current address: Pharma Bio-Research Group B.V., Laboratories Assen, Westerbrink 3, 9405 BJ Assen, The Netherlands

1.1 INTRODUCTION Increasing knowledge of the working mechanisms of drugs has led to the development of very potent drugs. Hence, the administered dosages are small, and consequently, the concentration levels in biological fluids are decreasing. Furthermore, biological samples are very complex, because they contain many endogenous substances. Blood fluids, such as serum and plasma, represent an extra problem due to the presence of proteins. Protein binding may affect the extractability of the analytes. Deproteinisation techniques can help to overcome this problem. It may, however, also give rise to even more difficulties, since analytes can be co-precipitated with the proteins. Thus, sample pretreatment techniques are required that retain the analyte(s) of interest, at the same time efficiently removing the endogenous interferences. The most common systems exist of an extraction step prior to separation and detection. A considerable gain in sensitivity and selectivity can be obtained during the extraction, as the analytes of interest are usually concentrated and separated from the matrix. An ideal extraction method should be rapid, simple, inexpensive, and give reproducible and high recoveries without the possibility of degradation of the analytes. Furthermore, the extraction method should not generate large amounts of chemical waste [ 1]. Sample pretreatment used to be a long step in the analysis of biological samples. Since the numbers of samples to be analysed is increasing, very rapid, but still selective and sensitive systems are required. In modern systems using advanced sample handling, the separation step may be more time-consuming. However, with the introduction of short columns in liquid chromatography (LC) and the selectivity of the mass spectrometer (MS), throughput of samples is again more and more limited by the time References pp. 39-44

2

Chapter 1

required for sample pretreatment. This is especially the case in off-line systems, which may also require extensive manual work. Therefore, various systems have been developed in order to integrate sample pretreatment with the separation and detection technique (Fig. 1.1) [2].

Sample pretreatment

!i~,,~iiii!ii~ ~,~ ~i!~ii!ii~i

Separation/Detection

Off-line

At-line

On-line

In-line

Fig. 1.1. Schematic presentation of various integration methods of the sample pretreatment step with the separation and detection technique. (Reprinted from [2, modified], with permission from Elsevier Science).

New developments in integrated sample preparation for bioanalysis Basically, three possibilities have been proposed for integrated sample pretreatment in the analytical procedures, i.e. (1) at-line; (2) on-line; and (3) in-line. The at-line coupling involves sample preparation by a robotic device and an autoinjector to inject the extracts into the analytical instrument. No direct stream of liquid between extraction unit and analysing unit is present. Moreover, not the entire extract is transferred to the analysing instrument. Disadvantages as observed with off-line extractions, i.e. collection of the extract, evaporation and reconstitution, are not eliminated. An example of an at-line system is the 96-well plate design for solid-phase extraction (SPE). Samples can be extracted simultaneously, thus increasing the sample throughput, provided that the separation and detection can be performed very rapidly or by using simultaneous analytical instruments. With on-line systems, there is a direct transport of the entire extract to the analysing technique, and the latter is receiving the entire extract. Samples can be processed in series, i.e. samples are pretreated and analysed one after the other, or in parallel, in which one sample is being analysed while another is being extracted. The latter system offers a high sample throughput. A very prominent advantage of on-line systems is that some error-prone steps of the extraction procedure, such as evaporation and reconstitution are eliminated, hereby increasing precision and accuracy. In-line systems exist of sample pretreatment fully incorporated into the separation system, hereby creating a new device. In contrast to on-line procedures, application of in-line systems imply the direct injection of the sample into the analytical instruments. Various approaches for in-line SPE-capillary electrophoresis have been reported [2]. It should be noted that the differences in interfacing are often not as clear as mentioned above. For example, the extraction can be performed manually (off-line) or by robot (at-line), but the final step of the extraction, i.e. the desorption of the analytes may be performed on-line with the analytical step. Furthermore, dividing systems into on-line and in-line techniques is very disputable. These systems are usually closely related to each other and a distinct difference can often not be made. Therefore, in this chapter on-line and in-line systems will be considered as similar. The goal of this chapter is to show the current status of modem sample pretreatment techniques such as SPE, solid-phase microextraction (SPME) and membrane-based extraction systems, and to outline novel trends in the bioanalytical area with regard to integrated sample preparation. It will focus not only on pretreatment techniques integrated with chromatographic separation systems, but also on their direct coupling to MS. SPE was originally designed for off-line purposes [3-5], but is now routinely used in on-line systems with LC [6-9]. The combination of SPE on-line with gas chromatography (GC) is less common, especially in the bioanalytical field. The current state of SPE-GC will be discussed here. Since an LC column can also be used as cleanup prior to GC analysis [10-13], on-line LC-GC applications without any further sample pretreatment will also be presented. Turbulent-flow chromatography (TFC) is to a certain extent similar to SPE. The use of high flow-rates offers new possibilities for sample pretreatment [14-17]. Therefore, the current state in TFC will be presented. SPME was originally designed for the analysis of volatile compounds with GC [18-22]. However, nowadays SPME is also coupled with LC for analysis of less-volatile compounds. The applicability of these SPME-LC systems in bioanalysis will be shown. Membrane-based techniques are, like SPME, diffusion-based sample pretreatment References pp. 39-44

4

Chapter 1

techniques. Dialysis is a more mature membrane method for sample pretreatment [23-25]. However, non-porous membranes provide new challenges for clean-up of biological samples. Therefore, the focus of membrane-based techniques for sample clean-up will be on the latter type.

1.2 CHROMATOGRAPHY-BASED EXTRACTION TECHNIQUES 1.2.1 Solid-phase extraction - gas chromatography

1.2.1.1 General aspects of SPE

A very common and powerful sample clean-up and concentration technique is SPE. It was originally developed for off-line purposes, but due to the demand for speed and the growing numbers of samples, at-line (including 96-well designs [9,26,27]) and on-line systems, such as the Prospekt, have been developed for coupling with LC [6,7,28,29]. On-line SPE is a very attractive sample pretreatment technique since the entire process of activation, conditioning, extraction, washing, and elution takes place in an enclosed circuit, which eliminates error-prone steps like evaporation and reconstitution. Also, the entire eluate is usually injected into the analytical instrument. Therefore, better precision and sensitivity may be observed when compared to off-line SPE. The most common on-line coupling of SPE is with LC, since similar solvents are used and virtually no modifications have to be made to the instruments. As this technique has already evolved and matured, the on-line coupling of SPE-LC will not be discussed here in detail. Worth mentioning, however, is the growing interest for high-throughput systems based on short-column LC coupled with MS, or even direct coupling of SPE and MS [30-39]. In such systems the extraction and detection should offer both sensitivity and selectivity in order to be able to detect low quantities of analytes in biological fluids. It should be noted that many applications applying little or no separation prior to MS may have to deal with ion suppression effects [31,40-43], clearly showing that the SPE eluates are not always free of matrix compounds. Numerous applications on off-line SPE-GC have been reported for the analysis of biological samples, and various reviews have appeared to which the reader can be referred [3-5]. Also, at-line systems, e.g. PrepStation [44,45] and ASPEC [46,47], will not be considered here. With the latter systems the extraction is performed automatically, but the eluate is collected in vials and subsequently the eluate is, usually only partially, injected into the GC. Numerous applications have been reported about the usefulness of off-line SPE combined with gas chromatography for analysis of biological samples. However, on-line SPE-GC appears to be emanating as a rather unexplored, yet promising technique. The on-line coupling of SPE with GC implies injection of large volumes of solvent into the GC, thus requiring modification of the injection system. A number of interfaces have been proposed for this purpose [48], i.e.: (1) on-column injection; (2) loop-type injection; and (3) programmed temperature vaporiser (PTV). With on-column injection, solvent is introduced at a speed above the evaporation rate and at temperatures below the boiling point of the solvent, ensuring wetting of the retention gap [49]. Solvent is

New developments in integrated sample preparation for bioanalysis evaporated in the retention gap and eliminated via the solvent vapour exit (SVE). An extra retaining pre-column enables refocusing of the analyte prior to transfer to the actual analytical column. A second way to allow large-volume injection (LVI) is the loop-type interface, originally designed for on-line LC-GC [50]. Injection is performed by filling a loop and flushing the contents to the retention gap inside the GC, which is slightly above the boiling point of the solvent. Formation of vapour results in pressure build-up and prevents further penetration of the solvent into the retention gap. Finally, a PTV injector was designed by Vogt et al. [51,52]. The injector strongly resembles a conventional split/splitless injector. The main difference is injection of solvent at temperatures 30-40~ below its boiling point on a packed liner. The liner packing acts as a liquid reservoir. A high purge flow ensures evaporation of the solvent, while analytes are retained on the liner packing. After almost complete evaporation of the solvent, the analytes are thermally released from the packing and transferred to the GC column. The latter is still at low temperatures, allowing refocusing of the analytes.

1.2.1.2 Applications of SPE-GC A selective SPE-GC method was described by Farjam et al. [53], who coupled immunoaffinity sample pretreatment with GC. A column with immobilised antibodies was used for the extraction of [3-19-nortestosterone from urine. A reversed phase reconcentration column and a retention gap were used for interfacing the extraction and the GC. Desorption from the antibody-column was performed with methanol-water (95:5, v/v), and after subsequent dilution with water, the analytes were trapped on the reconcentration column. Elution from this column was performed with 75 p~l ethyl acetate. The high selectivity during trapping made it possible to analyse large urine samples (5-25 ml), with good sensitivity for all investigated steroid hormones (LOD about 0.1 ng/ml). The total analysis time was still 40 min, mainly due to the long GC analysis. Benzodiazepines were determined in plasma by on-line dialysis-SPE-GC [54]. Clean-up was based on performing dialysis for 7 min, and subsequently, the diffused analytes were trapped on a PLRP-S pre-column. After drying elution was performed with 275 p~l ethyl acetate, which was injected into the GC via a loop-type interface. The SPE step in this procedure was more a reconcentration step than an actual extraction process. Nonetheless, this on-line dialysis-SPE-GC system showed sufficient selectivity (Fig. 1.2). The authors claim that the benzodiazepines could be detected at therapeutic levels (5-25 ng/ml) and that extra selectivitity could be obtained by acidification of plasma prior to extraction (Fig. 1.2C and D), but these claims are not substantiated by their figures. The PTV injector is an interesting injection system, as it allows LVI and thus on-line LC-GC and SPE-GC. The possibilities of PTV-GC in combination with SPE for plasma samples have already been demonstrated [55]. Now that integrated, automatic instruments have become commercially available, on-line SPE-PTV/GC will be facilitated. Moreover, the PTV injector also offers possibilities for thermal desorption [10,56-60]. Thus, no solvent is introduced into the GC, hereby eliminating some difficulties observed with LVI-GC, such as introduction of large volumes of solvent and its evaporation within the GC system. However, the reports available so far all deal with References pp. 39-44

Z

Time (min)

15

20

QJ

E

Time (rain)

25

Fig. 1.2. On-line dialysis-SPE-GC-NPD of (A) untreated blank plasma, (B) untreated plasma spiked with 1 la,g ml -J of nitrazepam, (C) acidified blank plasma and (D) acidified plasma with 1 lu,g ml -~ of medazepam. (Reprinted from [54], with permission from Elsevier Science).

50 mVI

E

O",

New developments in integrated sample preparation for bioanalysis environmental samples. Finally, the PTV injector has also been used for direct injection of plasma samples [61 ]. In this set-up, acidification and subsequent ultrafiltration were the only sample pretreatment steps. About 50 ~1 of the ultrafiltrate were injected onto a packed liner. No interference of the matrix was observed (see Fig. 1.3) and the chromatographic system was not damaged. The only drawbacks were occasional memory effects and the necessity to change the liner after 20 injections. The latter is probably due to the injection of ultrafiltrates which still contain some proteins, causing adsorption to the liner packing and GC column. The quantitation limit for ropivacaine was down to 300 pg/ml.

1.2.1.3 Remarks regarding the applicability of SPE-GC New devices have been developed for the coupling of miniaturised SPE with GC [62,63], enlarging the possibilities for incorporation of on-line SPE-GC into routine analysis. Until now, the number of applications of on-line SPE-GC in bioanalysis is very limited, in contrast to the numerous reports of this technique in the analysis of surface and drinking water [10-12,64,65]. This is probably due to the complexicity of biological matrices in comparison with water samples. Furthermore, various compounds in biomedical and pharmaceutical studies cannot be analysed with GC due to thermolability of the compounds. Nevertheless, the applicability of SPE-GC with LVI, and in particular the PTV injector, for biological samples seems worth further exploring.

1.2.2 Liquid chromatography - gas chromatography

1.2.2.1 General aspects of LC-GC A similar approach to SPE-GC is the coupling of LC on-line with GC in which the LC column functions as a sample pretreatment technique. Only the fractions of interest will be transferred to the GC (heart-cutting). The LC column is merely used for clean-up purposes and the GC column is used for the actual separation. As with on-line SPE-GC, on-line LC-GC also implies the introduction of relatively large liquid volumes into the GC, so that LVI must be used. As discussed with on-line SPE-GC, several approaches have been proposed in order to allow injections up to 1 ml into the GC. Nearly all online LC-GC applications involve normal phase (NP) LC, since the introduction of volatile elution solvents into the GC is more easily achieved than that of aqueous solvents [10]. However, direct analysis of biological, i.e. aqueous, samples in NPLC is not possible. Therefore, a separate sample pretreatment step, e.g. LLE or SPE, is always required. Consequently, no applications in the bioanalytical field have been reported for on-line NPLC-GC with direct injection of the sample. It is more common to use reversed phase (RP) LC in biomedical and pharmaceutical analysis. Coupling RPLC with GC implies introduction of large volumes of aqueous and ionogenic solutions into the GC. Water is very disadvantegous for GC analysis due to its high boiling point, high surface tension, poor wetting characteristics and aggressive hydrolytic reactivity, whereas non-volatile buffers (i.e. its ions) are also non-compatible

References pp. 39-44

0

0

1oo

0

100

. . . .

I

. . . .

0:40

m/l:131

ml z : 126

"

. . . .

. . . .

Time

9:00

I

. . . .

(min)

'

I .... 9:20

9653 i,I n94,~.

""~'''1

Blank plasma

9:40

r

....

. ~04 S.4ab

g*04 2.67S

Z+04 1.222

lO:OO

-

-

o

10

40

60

80

ZOO

o

20

40

60

80

100

0

2O

4O

100

....

,~

)37 11:47 .14944

/i

47780

8 : 4, 0. . . . . . . .

m/-"-o.13.1

n / 9 912S

n/z:lli

351

....

..

,oo':'- . . . . . . .

" ' '

8:$4 254)6 40097

Time

(min)

,2-:'o

2t$1S! 4S4~37

I:|4

)TI

~I$14$ z2~2o

)|1 9:15

'

Plasma sample

,r .

.

J : 4' 0

.

.

.

. ......

.

| lO:O0

1.119

9 g-iS t.4t0

Fig. 1.3. Chromatograms obtained from extracted blank plasma and from plasma spiked with analytes (400 nM each) and I.S.: (A) PPX (m/z: 84), (B) ropivacaine (m/z: 126) and (C) internal standard ropivacaine-D7 (m/z: 133). (Reprinted from [61], with permission from Elsevier Science).

(c)

(B)

(A)

:r

New developments in integrated sample preparation for bioanalysis with GC. Nonetheless, several techniques have been proposed for the analysis with online RPLC-GC [10-13,66-69]. Basically, they can be divided in two groups: (1) direct introduction of aqueous RPLC fractions using special GC-injection systems; and (2) phase-switching techniques, by which the analyte is transferred to an organic solvent prior to introduction into the GC. In the latter situation RPLC is coupled to GC via online LLE. Both approaches have found their way in environmental analyis [11-13], but the number of applications in bioanalysis is very limited. In a few cases RPLC has been coupled on-line with GC for the analysis of biological samples using an off-line sample preparation, i.e. LLE [70] or SPE [71] was applied prior to injection into the LC system.

1.2.2.2 Applications of RPLC-GC An example of the first approach, i.e. direct introduction, was described by Duquet and co-authors [72], who coupled ~RPLC on-line with GC using a aminopropyltriethoxysilane-deactivated retention gap. Diazepam was determined in urine by transferring only 2-~1 methanol-water (80:20 v/v) fractions from the LC into the GC. Goosens et al. [73-75] also applied a retention gap to transfer eluents from the RPLC to the GC for drug analysis. Up to 200 ~1 of eluent (acetonitrile-water) could be introduced into a Carbowax-deactivated retention gap by using an on-column interface and SVE [76]. The presence of remaining water after azeotropic evaporation was found to deteriorate the analysis. Thereofore, prior to introduction into the GC, addition of 10% acetonitrile to the LC eluent was performed resulting in an azeotropic acetonitrile-water mixture (84:16 v/v). By these means, the maximum amount of water remaining after evaporation was never exceeded. In order to inject ion-free fractions into the GC an anion-exchange micromembrane was inserted between the LC and GC parts [74]. Methanesulphonic acid was efficiently removed (99.9%) from the eluent acetonitrilewater, allowing the reproducible analysis of the potential drug eltoprazine, i.e. the coefficient of variation was found to be 3% (n = 5, 150 ~g/ml). A different approach, i.e. phase switching, was applied by Wessels et al. [77] and Ogorka et al. [78]. A phase switch was performed by using an LLE interface between the LC and GC part (loop-type interface). The set-up of this system is depicted in Fig. 1.4. Use of a GC-MS system allowed the identification of various unknown impurities in pharmaceutical products. Even though the instrumental set-up is rather complicated, reliable results were obtained for the quantitative determination of [3blockers in human serum and urine [79,80]. A total analysis time of 45 min. was required for the selective removal of the matrix compounds and efficient and repeatable LLE-GC analysis. Hy6tyl~inen et al. [81] applied a similar system to determine morphine and its analogues in urine. After LC separation, a phase switch was applied using an elevated temperature for the eluent and the extraction coil, which resulted in increased recoveries. After phase-switching, the analytes were derivatised on-line with N,O-bis(trimethylsilyl)trifluoroacetamide prior to GC-FID analysis. The total analysis time was less than 60 rain. As can be seen in Fig. 1.5, the LC clean-up procedure was effective as no matrix compounds were observed. Extra peaks in the chromatogram are due to the excess of derivatisation reagent. In principle this is a very powerful system. However, although the above studies show interesting approaches, the utility of the References pp. 39-44

Separat ion

Column

--

EarLy Vapor EJ~iL

I~etaintng Retention Precolu~n Gao

GC

kC Colux.n {llest r ict ion)

:. ,~l~]

1]]]]s

2

|O-Portvalye

~ . / V V V V " 5 o o I J | sample L~176

--

Flow Pressure Regulator Regulator

waste

Resl~rictor

L ~

_ _

~ ~l Lo~_W~_ e

Va!ve Restrtctor

NeeOle

T Cooljng Bath w| th j Extract ion Coil Sandwich TyDe i Phase Separat~ I Detect~ i Restrictor ]

---

-~Waste

Fig. 1.4. Instrumentation for on-line coupled reversed phase LC-GC-MS. ( 9 John Wiley & Sons Inc. Reproduced from [77] with permission).

MS

i ,

-

1

Extraction i i li$ olvent

I HPLC-Pu~ 1-~)~.=~(- - -u

~

Ll)w Pressure valVeFlP.Column

il,,l

New developments in integrated sample preparation for bioanalysis

11

135

2 80

3

mV 1

5

26

L

135

80 mV

26

o

. . . . . ~o

20 Time (rain)

30

40

Fig. 1.5. GC chromatograms of (top) urine spiked with the opiates and (bottom) blank urine. Peaks: (1) dihydrocodeine, (2) codeine, (3) ethylmorphine, (4) morphine and (5) heroin. Concentration of the analytes was 3 ~g/ml. (Reprinted from [81], with permission from Elsevier Science).

resulting systems sometimes remain questionable. For example, in the final application of morphine and analogues in urine, it should be noted that the concentrations of the analytes are very high (3 ~g/ml), and the absence of interferences is not clearly demonstrated by Fig. 1.5. Furthermore, most of the analytes are metabolised prior to excretion into urine. It is thus more interesting to analyse the metabolites. Also, the analytes can be easily determined by SPE-HPLC-UV, allowing a faster and easier analysis without the requirement of phase switching and an extra derivatisation step.

1.2.2.3 Remarks regarding the applicability of RPLC-GC The results achieved with R P L C - G C are still not very remarkable, as analysis times are long and poor detection limits (> ~g/ml) are obtained. The possibility of LC-MS analysis without derivatisation and the availability of similar but better sample

References pp. 39-44

12

Chapter 1

pretreatment techniques such as SPE will most certainly imply that RPLC-GC and NPLC-GC will not be applied on a routine basis.

1.2.3 Turbulent-flow chromatography 1.2.3.1 General aspects of TFC

Various techniques for SPE automation in combination with LC have been developed, e.g. ASPEC (Gilson), PrepStation (Hewlett Packard), Prospekt (Spark-Holland). Even though these systems greatly facilitate sample handling, the analysis time is usually still long. The analysis time can be significantly reduced when chromatography is performed with high flow-rates, e.g. under turbulent flow conditions. The latter was introduced by Quinn and Takarewski [82] in 1997 as a fast method for sample analysis. In this approach typical flow-rates of 3-5 ml/min are applied using a 1.0 mm i.d. column. These high flow-rates can be applied due to the low column back pressure associated with the use of large porous particles (typically 30-60 l~m) [14-17]. The solvent front profile is shaped like a plug rather than a parabolic profile as observed under laminar flow conditions. The high flow-rate and the plug flow profile increase the effective diffusion rates within the pores of the stationary phase. Consequently, the flow regime cannot be described by the Van Deemter equation. As a result, plate heights are significantly lower compared to predictions based on that equation. However, these conditions result in a considerable reduction of the chromatographic analysis time [14-17,83-86]. The typical TFC procedure, usually applied for the direct-injection analysis of crude plasma, basically consists of four stages, similar to conventional SPE: sample clean-up (extraction), analyte elution, LC separation and system re-equilibration [87]. It should be noted that the turbulent flow conditions can be used during each step of the procedure, but that some steps, e.g. the separation, can be performed using a laminar flow. During the purification step the analytes have to be separated from the matrix and should be retained by the stationary phase. Removal of plasma proteins from drugs is achieved by size exclusion (pore size 60,~) and slow diffusion of proteins into the pores. Since large particles are used, the technique also allows the use of large end-column frits (10-40 I~m) [ 15]. As a result, large protein molecules in the plasma sample can easily pass through these columns without clogging of the frits due to precipitation. Elution is performed using a steep gradient of organic solvent followed by chromatographic separation on a second column. Only limited separation is achieved on the analytical column due to the high percentage of organic solvent. Finally, the extraction and separation column are equilibrated with suitable solvents for subsequent analysis. A basic set-up of TFC is depicted in Fig. 1.6. As described above, TFC strongly resembles SPE, with the use of high flow-rates instead of normal flow-rates, and larger particles in the extraction column. As a consequence, the set-up for TFC (Fig. 1.6) is similar to that for SPE-LC systems. TFC is applicable for the analysis of drugs that bind to a high extent to proteins. In general, recoveries of 70-100% are obtained [87-89]. Only with very strongly bound drugs modification of the TFC procedure is required, e.g. low-flow sampling (0.5 ml/min) or acidification of the sample prior to extraction [15].

13

New developments in integrated sample preparation for bioanalysis From Pump B y

To Mass Spectrometer

\ Analytical Column

Autosampler To Waste From

Pumps A Extraction Column

Configuration (a): Loading sample, extraction and equilibration

From Pump B ....

oMass

Spectrometer Analytical Column Autosampler

To Waste From

Pumps A Extraction Column

Configuration (b): Eiution Fig. 1.6. Schematic representation of the flow path during the on-line extraction, elution and equilibration. ( 9John Wiley & Sons Inc. Reproduced from [91, modified] with permission).

1.2.3.2 Applications of TFC The first application of TFC-MS for direct-injection analysis of plasma was reported by Ayrton et al. [85] in 1997. In successive years, more applications have been developed, mainly for the analysis of low molecular-weight drugs in plasma [14-16,83,84,86-94] and serum [14,95]. Nonetheless, Cass and co-authors [95] reported the successful determination of vancomycine, a 1448-Da peptide, in urine by means of a fully automated TFE-LC-MS/MS system with a detection limit of 1 ng/ml. Most applications of TFC focus on the determination of a single compound in plasma in a single run. Jemal et al. [92] reported the simultaneous determination of simvastatin

References pp. 39-44

14

Chapter 1

and simvastatin acid in human plasma by direct-injection LC-MS/MS. The possibilities for multi-component determination in a single run were further explored by Wu and coauthors [15]. This study was set up for high-throughput pharmacokinetic screening using LC-MS/MS and a turbulent-flow column-switching system by which ten compounds had to be analysed simultaneously. The set-up of the system was similar to the design presented in Fig. 1.6. A 4-ml/min flow was used for 1 min. during which sampling and purification was performed. Then, the trapped analytes were eluted in a back-flush mode from the extraction column towards the analytical column using a flow of 0.4 ml/min. Elution was completed within two minutes. Flushing and equilibration of the extraction column was performed during the separation of the analytes on the analytical column. Good separation and peak shapes (Fig. 1.7) were achieved within a run time of 10 min. including the extraction time. A dynamic range of 1-2500 ng/ml was obtained, with a limit of quantitation (LOQ) of 1 ng/ml. Using the highest concentration, i.e. 2500ng/ml, a carry-over of 0.14+0.07% was observed. One extraction column could be used for 200-300 plasma sample injections without causing significant back-pressure increase. It should be noted, however, that the simultaneous analysis of the analytes in this study is of limited interest for pharmacokinetic screening. Instead, analytes and their metabolites or co-administered drugs should have been chosen as target compounds. A ternary-column system was introduced by Xia et al. [88] for high-throughput direct-injection analysis of plasma. Basically, the system consisted of two extraction columns in parallel and one analytical column. In this way, one column was equilibrated while on the other column the extraction of 10 txl plasma was performed. Thus, the equilibration step does not add extra time to the injection cycle time. The on-line purification step lasted for only 0.3 min. and the total run time was 1.6 min. The extraction recovery of the guanidine-type drug was > 95%. Using the sensitivity and selectivity of the mass spectrometer by operating it in the SRM mode, an LOQ of 1 ng/ ml was obtained. Good intraday and interday precision (< 6.6%) was achieved in the range of 3-1000 ng/ml. Ayrton and co-authors [90] applied ultra-high flow-rate capillary LC with MS/MS (SRM mode) for the direct determination of an isoquinoline drug in plasma. The extraction column had an internal diameter of 0.18 mm, thus allowing a flow-rate of 130 ~l/min for turbulent flow. Upon injection of only 2.5 Ixl plasma (diluted 1 : 1 with an aqueous standard solution), an LOQ of 0.5 ng/ml was obtained within a total run time of two minutes.

1.2.3.3 Remarks regarding the applicability of TFC Comparing the potentials of TFC with results obtained with LLE and automated 96-well SPE, it can be noted that similar results are obtained in terms of dynamic range, LOQ, accuracy and precision [15,83,85]. As turbulent flows reduce time-consuming steps, the speed of the system is superior to more conventional on-line SPE systems. The limited concentration potential is a disadvantage of TFC. Although this can be compensated by using capillary LC, the load capacity is then reduced, allowing less injection of the sample. The rate of solvent consumption of TFC is high in comparison to conventional LC. However, the total volumes per analytical run are similar for turbulent-flow LC-MS

New developments in integrated sample preparation for bioanalysis

15

r~

.,..~

a ,d

,d

.~

,d

o

b~ o

r..)~ r~

~

J

s~ d

,e-

8 ~

~

9,..~

F

8

References pp. 39-44

$

X'~

,~.~ r.~

0

16

Chapter 1

and for SPE followed by conventional LC-MS [15]. Up till now, TFC is not rapidly expanding, probably due to the mainly size-exclusion-based principle for sample cleanup. As a result, virtually all applications are analyses of plasma or similar fluids, since the analytes of interest can be easily separated from proteins and other large matrix compounds. Handling a urine matrix is much more complicated as small drugs cannot be easily separated from small urine matrix components. Even though the speed is advantageous, TFC will probably not become a universal and widely applicable sample pretreatment technique. Some applications use turbulent flows only during the first step, i.e. purification, with subsequent separation on an LC column using laminar flows. Therefore, in such cases the term 'turbulent-flow chromatography' (TFC) should possibly be replaced by 'turbulent-flow extraction' (TFE), and TFC should only be used when the separation is performed under turbulent flow conditions. Furthermore, it should be noted that the sample pretreatment is actually LC modified into an SPE procedure. This can also be seen in Fig. 1.6, as the set-up for TFC is similar to SPE-LC. The technique might also be referred to as modified (ultra-high flow) LC-LC with subsequent detection. Another critical point was raised by Ayrton et al. [86]. Careful consideration of commonly used flows (3-5 ml/min) and particle sizes (30-60 txm) showed that the actual flows are not turbulent. Therefore, the term ultra-high flow-rate LC with direct sample injection might be preferred, as is also used now in other systems [34].

1.3 DIFFUSION-BASED EXTRACTION TECHNIQUES 1.3.1 Solid-phase microextraction 1.3.1.1 General aspects of SPME

At the introduction of SPME [96], it was designed for gas chromatographic analysis by direct sampling of liquids. SPME integrates sampling, extraction, concentration and sample introduction into a single solvent-free step [21,97]. Originally, SPME was performed with a modified syringe with a stainless steel needle in which a thinly-coated fused silica fiber (100 Ixm in diameter) could be moved up and down via the plunger [19,20]. The fiber is coated with a suitable stationary phase, which is usually a polymeric phase of 7- to 100-1xm thickness. The fiber is immersed into the sample (direct immersion SPME (DI/SPME)) or into the head-space of the sample (HS/SPME). The advantage of HS/SPME is that, because the fiber is not inserted into the sample itself, relatively dirty samples can be analysed while obtaining clean extracts. Another advantage is the relatively high speed, i.e. short equilibrium times in comparison to DI/ SPME. Furthermore, more aggressive sample preparation can be applied, e.g. extremely low or high pH values, without the risk of damaging the coating [21 ]. After equilibrium or a well-defined time, the fiber is transferred to undergo liquid desorption, usually followed by LC analysis. For GC analysis the analytes are thermally desorbed. For LC a special desorption chamber is used. SPME is an equilibrium technique and is based on the partition of the analyte between the stationary phase and the matrix. As a result, SPME is a non-exhaustive

New developments in integrated sample preparation for bioanalysis

17

extraction method. The amount of analyte extracted (n) is proportional to the initial sample concentration (Co) and sample volume (Vs), the volume of the stationary phase (Vf), and the fiber coating/sample distibution constant (Kf3 [22,98]: n = (Kf~•

Vf x V s x

Co)/(Kf~ •

Vf -[- Vs)

(1)

Extensive studies have been performed to describe the theoretical fundamentals of SPME [20,97,99,100,101]. Basically, the processes in SPME can be divided in thermodynamics and kinetics [20,102]. Because of the physicochemical properties of, for example, polydimethylsiloxane (PDMS, glass transition temperature-123 to -126~ [20,99]), which is the most commonly used phase in SPME, the extraction may be described as liquid-liquid extraction. Octanol-water partitioning coefficients (K~ may be used to estimate the extraction behaviour of an analyte towards PDMS, although it should be noted that octanol and PDMS differ severely with regard to chemical properties. PDMS is generally considered to be an absorptive phase [103], although adsorption effects may occur as well [100,104,105]. The diffusion coefficients of the analyte towards the coating and in the coating determine the time that is required to reach equilibrium. Agitation will increase the kinetic processes since diffusion in the sample is no longer limiting. In an agitated solution the diffusion towards the stationary phase through a static water layer around the fiber is the limiting process in DI/SPME [98,100,106]. Various factors can influence the extraction and desorption efficiency. If available, the coating of the fiber is usually chosen based on the principle 'like dissolves like'. Adjustment of the pH of the sample may affect the yield because non-ionised species are better absorbed by the commonly used fiber coatings. Addition of salt increases the ionic strength and favors salting-out, which often results in an increase of the yield. Increasing the sorption temperature has a dualistic effect. On the one hand, diffusion coefficients are higher at higher temperatures, thus leading to a decrease in time to reach equilibrium. On the other hand, higher temperatures lead to lower partition coefficients in the stationary phase, thus decreasing the extraction yield [99]. During desorption with an aqueous solvent, a pH shift may be applied to cause ionisation. Also, the addition of a suitable organic solvent may be helpful to speed up the desorption process [ 107]. All factors have been studied thoroughly [ 19,20,99,107] and will therefore not be discussed in detail here. Initially, a lot of work has been done on the environmental and pesticide residue analysis [108,109] with SPME-GC. Nowadays, many applications of SPME-GC with either HS/SPME or DI/SPME have been described in various fields, like the analysis of food [108,110,111], explosives [112], biological and pharmaceutical samples [18-20,108,113]. For blood samples, it is possible to determine the degree of proteinbinding of drugs, since SPME usually extracts the free fraction of the drug only [114,115]. Hence, the protein binding can easily be determined if the equilibrium between bound and free drug is not disturbed in the SPME procedure. Another possibility for the estimation of the degree of protein binding is to dilute the sample and determine the extracted amount for different dilution factors [ 116]. When, on the other hand, DI/SPME is applied to urine, matrix interference may be quite severe.

References pp. 39-44

18

Chapter 1

In this chapter we will not focus on SPME-GC as it has already been extensively described elsewhere [18-22,108]. Instead, we will show the current status of the less common coupling of SPME to LC and the direct coupling of SPME to MS via an L C MS interface for the analysis of biological samples. Though the coupling of SPME with CE has been described [ 117-121 ], the application to biological samples has only been performed off-line [119] and will therefore not be discussed here.

1.3.1.2 Applications of SPME-LC In 1995 the first paper on SPME-LC appeared [122]. Basically, two formats of SPME have been coupled to LC, i.e. fiber-based SPME and in-tube SPME. Fiber-based SPME is similar to that applied for GC analysis. Only a specially designed LC desorption chamber [122,123], which acts as a chromatographic tee with a six-way valve, is required to allow desorption with solvent. This chamber can be simply interfaced with the LC. The extraction process is mainly passive though agitation and/or temperature elevation may be applied. With in-tube SPME the sample is aspirated and pushed back by means of a syringe through a capillary coated with the stationaire phase, thus creating an active extraction [124]. This aspiration-push back sequence can be repeated at will. The set-up (Fig. 1.8) is similar to open-tubular SPE and has been known for many years. An overview of bioanalytical applications using SPME-LC is presented in Table 1.1. Since the first application of fiber-based SPME combined with LC, some reports [107,127,130,131,135-137] have been made on the applicability to the field of bioanalysis. Volmer et al. [135] were the first to report the combination of SPME and LC-MS (SIM mode) in the bioanalytical field. The method was developed to analyse 11 corticosteriods and 2 steroid conjugates in urine. Sorption times of 10 to 60 min. were observed, and only 5 min. desorption was satisfactory with regard to carry-over. The detection limits were 4-25 ng/ml due to low yields (1.7-15%). Kumazawa and co-authors [137] described the determination of phenothiazines in whole blood and urine by fiber-based SPME and LC-MS/MS with electrospray ionisation, showing the problem of low yields versus the potential for selective and sensitive analysis. Before SPME was performed, whole blood was deproteinised by adding perchloric acid, centrifugated, and the pH was adjusted to about 8. With urine samples only pH adjustment was performed. A sorption time of 60 min. was applied, which ensured that equilibrium had been reached for all compounds, and subsequently, desorption was performed within 10 min. Even though the SIM mode was applied during single-MS analysis, severe matrix interference was observed for both urine and whole blood (Fig. 1.9a). Applying MS/MS and selected reaction monitoring (SRM) eliminated the visibly interfering matrix (Fig. 1.9b). The extraction yields were very low for whole blood (0.0002-0.12%), resulting in detection limits of 0.2-200 ng/ml. In the case of urine, better results were obtained, as less interference of matrix was observed. The yields were 2.6-39.8%, with detection limits of 4-22 pg/ml. Most probably, the low extraction efficiencies in blood are not due to the equilibrium nature of the extraction as claimed by the authors. It is more likely that a co-precipitation of proteins and bound drug due to acidifcation with perchloric acid occurred.

%

I

.

"%

x"k

*'~176176

9%

,,

,o

/

\

Solvent vial

capillary

Buffer tublnq

Septum piercing needle Sample vial

Vial retainer arm

I Adjustable needle guide/depth gauge

Adjustable capillary guide/depth gauge

LOAD position

INJECT position

Stx-port valve

Analytical column

HPLC Pump

Fig. 1.8. Instrumental setup of the new on-line SPME-HPLC interface based on an in-tube SPME capillary technique. A piece of GC column (in-tube SPME) hosts in the position of the former needle capillary. The aqueous sample is frequently aspirated from the sample vial through the GC column and dispensed back to the vial (INJECT position) by movement of the syringe. After the extraction step, the six-port valve is switched to the LOAD position for the desorption of the analytes from the in-tube SPME by flushing 100% methanol from another vial through the SPME capillary. The volume is transferred to the loop. After switching the Valco valve to the INJECT position, an isocratic separation using a mixture of 60:40 acetonitrile/water was performed. A detailed view of the in-tube SPME capillary is included at the left side of the figure. (Reprinted from [124]. Copyright 1997 Am. Chem. Soc.).

\

/

.-.........---~... Waste o0

Dispenser

~,,i.

t~

t~

Urine Urine Serum Urine Urine Urine Serum Urine Urine Urine Serum Ufine Serum Urine Urine Urine Urine Urine Urine Blood Urine

Amphetamines Amphetamines, [3-blockers

N.D.: no data available; DI: direct immersion. a: unless stated otherwise. b: molecularly imprinted polymer (MIP) coating. c: relative to aqueous samples.

Ranitidine

Carnitines Corticosteroids + conjugates Flavonoids Lidocaine Lidocaine Phenothiazines

[3-blockers

Benzodiazepines Brombuterol + analogues [3-blockers

Antidepressants (tricyclic) Antidepressants (tricyclic) Benzodiazepines

Sample

Compound

(DI) (DI) (DI) (DI) (DI) (DI)

In-tube

Fiber Fiber Fiber Fiber Fiber Fiber

In-tube

Fiber (DI) Fiber b (DI) In-tube

Fiber (DI) In-tube In-tube

In-tube In-tube

Type

TABLE 1.1 BIOANALYTICAL APPLICATIONS OF SPME-LC

0.38-0.82 0.1-1.2

81-98 c 0.7-16 N.D. N.D. 3.0-9.2 N.D. 45 88-110 c 70-109 c 84-113 c 71-11U N.D. 5-45 56-86 ~ 22 6.5 4-40 0.0002-0.120 60

11;4 11;4

180; 30 10; 5 11;4

180; 30 45; 2 x 5 11;4 10; 4 60; 2 15; 5 10; 6 45; 2 x 10 5;4 6O; 10 11;4

LC-ESUMS LC-ESI/MS ~LC-UV p~LC-UV LC-ESUMS (tx)LC-ESI/MS LC-ECD LC-ESI/MS LC-ESI/MS ESI/MS LC-MS LC-ESI/MS LC-UV APCI/MS/MS LC-MS/MS LC-ESI/MS

0.2-12.8 4-30 2.7-25.4 pg/ml 25 0.40 4-22 pg/ml 0.2-200 1.4

1-6 10 3-36 (LOQ) 9-43 (LOQ) 0.1-1.2

3-40 N.D. 0.024-2.00

LOD (ng/ml) a

Yield (%)a

Sorption; desorption time (min)a

System

[138]

[137]

[104]

[134] [135] [136] [107]

[1331

[130] [131] [1321

[127] [128] [1291

[125] [126]

Ref

tO O

Z

%

blood

. . . m,'= ;36~

m/z :'f/c

;434

m/z:43~

] Tdflu~

m/z:408

~/z

......

m/z:438

-

PUr.:44~

. . . . . . . . .

.

3 : 00

,

.

.

.

.

~

.

.

.

.

.

.

.

.

,~

.

~...,.,,.~-~

_J ~ L _ _

,~'

A

.

.,~

.

.

9:00

~-.

~

.

/k.~._

.

t 2 .?00

. . . . . . . . . .

_ _ a~___~_P_ . . . . . . .

. . . . . . . . . .

6 : O0

;

__

.

. . . . .

~ X . , . , _ . . . _ , . . _ ,

.

r-'-

-

, . . . . . . . . .

Thioproperazine

m.l Z : 4 4 7

Clospirazine

.

~

9

~

--

-

i

E-~04

~z t

Z'03

.

.

,

_J

0

I

'

""--- --

Thiethylperaz

1''''~

- m.,'z : 374,

ioridazme

. . . .

m/z-3"/l

mlz : 366

n . / = : 34C

extract

,e,O4

15 ;%0

--

E*04

m - -

~

~

.

i

.

;. 4 ~ 0

0

Flup~ine

-m/~ 94~.A

Flup~bxol

1

100

....

......

---

, ....

~

, . . . . . . . . . . . . . . .~.: .1.0 3 : 00

Thioproperazine

]- m/z:44"/

.....

Clospirazine

,oo1.,,,,, o

100

~,,

P

--

_

loo

'

j~

~

-

A

_IL_I~

_

T

11

:100

-

'

i

"" 15

A

~;+0~ 5.8 a :O0

/_i~.

z. ~9,

z ,sos

~"

'~

~.

~ s Ar,.o, ,... _

00~

I"~ ~ )

,~-~

.~::~

r '~ ~_~_ z. ~s2

, 9,~

~'--

~,~. ;~

~.~

r':;'~'~,~

E,C,4

F...,.,~3 3 6~ 8

:,~~~~

~

xA ~.__.

..

.

" "~''~"~'--~----]

__

_

____

..

---=.,.r~

,~ J'.--

- -----A- ~---

/ ~

,

9

P

.

s ....

.

Trifluoperazino /k~ _ _ ~ ...... -- . . . . ~ . - " " ' ~ --"

i/ll

Perphenazine

o '----,

o

,

I00--a/=, 43S

100

C,

"

, _ s %_ _

1~176 ":~'~176~S~.-,.~-~,

0 100

100 -

D

a00

Urine

[,. %~ ~o'1":'~ , ~

A a, r: ,, _

-

__

~. ~80

A P"~

v-

.,- E - 0 3 _ . j . , ~ ~ , . 420

Fig. 1.9. (A) SIM of S P M E - L C / M S for the 11 phenothiazine derivatives extracted from human whole blood and urine. The amounts of each drug spiked into 1 ml whole blood and urine were 0.5 ~g and 0.5 ng, respectively. The arrows show the locations where drug peaks should appear. The left vertical axes represent the relative percentage intensity of the peaks, the fight vertical axes show the absolute intensities of peaks recorded by the mass spectrometer and the horizontal axes indicate the LC running time (min).

0

0 1CO

iCO

.

Thiethylperazine

m/z:400

o ] Flu~ine

iCO

.

extract

Tnioridazilqe

.

Perazine

~,,.' ~ ; } 4 0

Prochloperazine

t

-1

o ] Flupentixoi

iOO

0

ZOO

o

C lOG

100

o

ZOO

0

zoo

Whole

40).'.0~,

399.;0>

DA

O1~

ot

of

. . . .

,

. . . .

..

446.70)

I : " "

3:~)0

]Thioprope~azine

IRXC,:, ~

4r

r

o~ 4 3 4 . ' 0 >

o razJne

~tC:>

-OA

. k

.

.

__

" ' "

"

"

441'.70

4r

4]8.?0-

&~5.?O

408"/0

"

l

. . . .

(~:O0

. . . . . . . . . 404,70

40:~.'/0

..

"

__

/L

A

.

.

L

"

.

.

.

.

.

.

....

.

.

.

'

.

.

.

.

.

.

.

li % o

.

.......

]. L .

~t

I ,,o~

.

jl~

j~

.

.

.

9

_

..

....

........

.....

,,

1

E-D1 3.4)8

.

I;.110

|,02

9 Ir 3 .SSO

E*02 ? .8o1

.oo

.

1,371

E * 0 4

Z*03

F,*02

P""

r [1.381

.

4

8..1

~

g*O~ .

-- E * 0 1

5

I2.82S

Z-02

9

RZ~-~' DA

]PernzJ~

0

0s

of

nf 17%.4~,~

3")0.4;0-',.

I74,K0

DA

o1~ r

Thiethylperazine 404,.70

. . . . . IRZ.',=> DA of 399. "/0> 4 0 0 . 7 0

--RZ"~=~'

--

] ProchlOl~mzine

-

RZ2m~, D?,

] TNo~dazlne

RZ2-~.

)71.1;C

,,.,o-3~2"i:"

)40.6C

RICm> 1D~. of

IDA

ga.

13'7170,,

" -~'~;--

: !- ' ' ~ " 3:00

' . . . . .I S : O0

o f 44S.60~ 4 4 6 . i ; 0

*~176 ] Thiolxopemzi n, 0

4:38.?0

435.70

'." . . . . .

of

t$&.'/O*

o ]Closmrszine

RI~,*

~xC-~

] Flupentixol

]

100 -

o

100

o

0

10o "1 ,.:c.. = o, ,o', ",0. ,o,7:,o Triflu o J ~ r

0 :1,00

D 100

100

100

o,

of' :339.60;"

0 ]Proper~azine

Ioo - "~-'" ~

0

100

Urine extract

k

-

"

FL.m

. . . . . .

;'-; : 9 . . . . . l 9:00

k

J•..

~'--"

.

k

.....

12

I " "'" " " : O0

....

Is

.

_

.

_

_

.

.

.

.

s 8. 922

E- 0 2 U, 0 4 "x

]. 6C4

E+02

9.858

E*02

It.01 t. 3 5 9

6.9S9

g-02

1.722

E - 0 3

3. 7 4 5

E*03

?.'754

~*22

_ E-32 8. 5 7 2

Fig. 1.9. (B) SRM with SPME-LC/ES-MS/MS for the 11 phenothiazine derivatives extracted from human whole blood and urine. The amounts of each drug spiked into 1 ml of whole blood and urine were 0.5 ~g and 0.5 ng, respectively. Axes as in (A). ( 9 John Wiley & Sons Inc. Reproduced from [137] with permission).

0

100

0

' glC=)

_

JL

.......... k

-3";&.60

.........

3"~3.60,

of 4C?.:0>

O~

of

of

o~ 3"~.::).(C'- 3"/L.60

tFlul~nli~ol

'~176

o

3 4 0 . 4 ~ 0

. . . . . . 365 .(0> 3 6 6 . 6 0

339.110>

]Tfifluopecazine

1~C-) O~

DA

0 --" i00 I~1:C-> ~

100

I~.1C->

DA

P,I~->

o

n~

,

RIC-,

DA

0

_

r

o!

rid~ine

RIC=> DA

RIC=)

-

D&

Perazine

IRIC-.-

0

0

100

I00

Whole blood extract

New developments in integrated sample preparation for bioanalysis

23

With the introduction of in-tube SPME [124] it became possible to automate the analytical procedure, and consequently, a number of applications in the bioanalytical field have been reported [125,126,128,129,132,133,138]. Most applications show great similarity, and therefore only one application will be discussed here. The results of other in-tube SPME-LC studies are summarised in Table 1.1. Kataoka et al. [ 126] developed an automated in-tube SPME-LC-ESI/MS method for the simultaneous determination of 5 amphetamines and 9 [3-blockers in biological samples using single-MS (SIM mode). Applying a flow-rate of 100 l~l/min proved to be optimal [22,124], and varying the number of draw/eject cycles of 35-1~1 samples resulted in the extraction-time profile (Fig. 1.10). As can be seen, equilibrium was not reached after 20 cycles. However, when the extraction was stopped after 15 cycles (taking about 11 min.), excellent detection limits (0.1-1.2 ng/ml) and linearities (R> 0.998) were obtained. Similar effects of type of matrix were observed as with fiber-based SPME, i.e. extraction from urine resulted in higher yields than extraction from serum. Worth mentioning is the differences in extraction yield and kinetics of drugs within the same class. This phenomenon has also been observed with fiber-based SPME. This may be explained by differences in diffusion coefficients of the drugs, but this is not expected. Surprisingly, the optimal pH during sorption was 8.5, ensuring that all compounds (pK, 9-10) were protonated to a large extent. This is contrasting to most applications in which non-protonated analytes were more easily absorbed than the protonated species. An explanation of the use of this pH value might be that the capillary coating was stripped at pH 10, resulting in lower yields than with pH 8.5.

1,20E+07 --O--

1.00E+07

MA

--,.--MDMA --e--- Pi ndo|ol

8.00E+06

. ~

c:

Propranolol

;~-- hO~.butolol

o

~a 6.00E+06 t13

4.01~*'06

2.00E+06

0.00E+00 = 0

- .............

' 5

.

10

.

. 15

.

.

. 20

Number of draw/eject cycle

Fig. 1.10. Extraction-timeprofile of drugs with Omegawax 250 capillary. SPME conditions: drugs, 0.5 ~g/ ml; sample pH, 8.5 (Tris-HC1); draw/eject cycles, 15; draw/eject volume, 30 or 35 ~1; draw/eject rate, 100 i~l/min; desorption, mobile phase. AM = ( + )-amphetamine, MA = ( + )-methamphetamine, MDMA= (=)-3,4-methylenedioxymethamphetamine. ( 9 John Wiley & Sons Inc. Reproduced from [126] with permission).

References pp. 39-44

24

Chapter 1

1.3.1.3 Applications of SPME-MS SPME is hardly ever used in high-throughput systems due to the relatively long sorption and desorption times. Even when SPME is followed by chromatographic separation, the sorption and desorption are usually the time-limiting steps. This also holds for fiberbased SPME systems and/or systems in which no separation is performed, i.e. the desorption device is directly coupled to the detector. P6rbs and co-authors [ 139] coupled HS/SPME directly to MS by thermal desorption and a short GC transfer-line for the determination of volatiles in cheese using only 10 min. sorption. A typical desorption profile of HS/SPME-MS is shown in Fig. 1.11. Orzechowska et al. [140] determined cocaine by direct coupling of HS/SPME with ion mobility spectrometry. Camitine, an essential factor in the fatty acid metabolism of organisms, was determined by SPMEESI/MS [134]. Severe matrix interference during detection and a long sorption time were the major drawbacks of this method. A possibility to enhance the throughput is to use non-equilibrium SPME, which is more commonly applied in in-tube SPME than in fiber-based SPME. Yet, such an approach may put extra pressure on the sensitivity and reproducibility of the method. Ai [141 ] described a theoretical model for fiber-based SPME allowing quantitation before equilibrium is reached. In a challenging practical application, Van Hout et al. [104] performed ultra-rapid non-equilibrium fiber-based SPME and coupled it directly to APCI/MS/MS for the determination of lidocaine in urine. Direct immersion of the 100 ~m PDMS fiber in urine was applied for 5 rain. Subsequently, the fiber was desorbed for 4 min., after which 1 min. analysis was performed. No matrix interference was observed. The method had not yet been optimised for quantitation, but initial results on repeatability and reproducibility looked promising. Even without an internal standard, coefficients of variation below 15% (LOQ) were observed at a level of 2 ng/ ml. The detection limit was 0.4 ng/ml and good linearity was observed. This approach clearly showed that the long equilibrium times of SPME are disadvantageous for rapid analysis, but that use of non-equilibrium SPME can help to overcome this problem.

1.3.1.4 Remarks regarding the applicability of SPME SPME has clearly demonstrated its utility for bioanalysis, and in particular for the analysis of urine samples. The number of applications for the analysis of plasma, serum and blood samples is more limited. So far, no reports have been found about ion suppression in SPME-LC-MS systems. This might be due to the clean extracts that can be obtained with SPME due to the low yields for both analytes and matrix compounds and differences in diffusion coefficients. However, it may also be due to the lack of attention for these effects. A major drawback of SPME is the small number of commercially available stationary phases, especially for fiber-based SPME, hereby limiting the choice for selectivity. The main commercially available coatings for fiber-based SPME are PDMS, polyacrylate and some mixed phases, e.g. PDMS/carboxen, PDMS/divinylbenzene (DVB), Carbowax/DVB and Carbowax/templated resin. From this list it is obvious that the choice for selective extraction is limited. Moreover, ion exchange phases are not available. For intube SPME, a piece of capillary GC column is commonly used, thus providing more

25

New developments in integrated sample preparation for bioanalysis 9 9

o

.H o

8 ,p

o

o~

,.~,< 0

,I1 "

'

%

~ ! ~

0

o

(T w ~ l ~ u m m v

,p

~.~ ;;~

.

. ,,...q

o

o

o

~~

0

2~ o

.

/

~s 0

C !

A

9

.=_

g

..... ;_

o

U3 0

),unoo UOl lelOJ.

References pp. 39-44

I--

I

o

~

.,..~

9

,.~

9

r.~

.~

26

Chapter 1

diversity in stationary phases than in fiber-based SPME. In addition, phases for more polar analytes are available. Yet, with the development of more stationary phases for both fiber-based SPME [131,142-144] and in-tube SPME [132] the potential of SPME may be improved even further. The advantage of in-tube SPME over fiber-based SPME is the shorter sorption times, as can also be seen in Table 1.1. This is due to the fact that in-tube SPME is usually stopped before equilibrium is reached. However, as described above, non-equilibrium fiber-based SPME can be performed as well. With the latter, transfer of the fiber with the trapped analytes to the desorption chamber is required after sorption, which is not necessary with in-tube SPME. The latter system also allows the use of 100% organic solvent for desorption of the analytes. The use of high percentages of organic solvent is more complicated with fiber-based SPME, since the coating can be stripped from the silica. Samples containing particular matter should be filtered prior to extraction with intube SPME to prevent clogging of the capillary, which cannot cause problems in fiber-based SPME. With regard to automation, in-tube SPME has more to offer at this moment. Automation is more easily established with this system, since no transfer of the stationary phase to a desorption chamber is required. The availability of an autosampler for use with fiber-based SPME-LC is still being awaited [21]. Thus, both fiber-based and in-tube SPME have their own advantages and disadvantages. To some extent, the systems can be considered complementary and the best choice will depend on the sample and on the particular requirements of the individual applications. A recently introduced SPME device is the stir bar sorptive extraction [145], which has a larger volume of stationary phase than fiber-based SPME and in-tube SPME. Although it has not been applied for bioanalysis, it has already shown its potential for the coupling with LC and determination of (semi-) volatiles in aqueous samples.

1.3.2 Membrane-based sample preparation techniques 1.3.2.1 General aspects of membrane-based techniques As an alternative to SPE and/or LLE, membranes may be used as a sample preparation technique [23,25,146-153]. When using the latter, it is essential to differentiate between porous and non-porous membranes. Sample pretreatment with porous membranes is based on the principle of size exclusion to differentiate between substances, whereas non-porous membranes utilise the difference in partition coefficients of substances, thus being an actual extraction technique. An overview of the various techniques is given in Table 1.2. The porous membrane techniques (PMTs) drew major attention in the late 1980s and early 1990s, whereas the applicability of non-porous membranes for sample pretreatment in the biomedical field is being explored more recently. In PMTs, the liquids on each side are physically connected through pores. Transport through the membranes is based on size-exclusion, i.e. sufficiently small molecules can permeate through the pores, whereas larger molecules cannot. This can result in an efficient clean-up from large matrix molecules, but no distinction can be made between small molecules. The latter is only possible to some extent with electrodialysis, for which an ion-exchange membrane is used. Now, large molecules and molecules with a

New developments in integrated sample preparation for bioanalysis

27

TABLE 1.2 OVERVIEWOF MEMBRANE-BASEDSAMPLEPREPARATIONTECHNIQUES Driving force

Mainly combined with

Dialysis Porous Electrodialysis P o r o u s

Concentrationdifference Potentialdifference

LC CE

Filtration Membrane extraction

Pressure difference Concentration difference

LC LC, GC, CE

Technique

Membrane Principle type Size-exclusion Size-exclusion and selective ion transport Porous Size-exclusion Non-porous Differencein partition coefficient

given charge will be excluded. Strictly taken, PMT is not an extraction technique, but a filtration process. Extensive descriptions of the principles of porous-membrane techniques were given by Van de Merbel [23] and co-authors [24,25], and will therefore not be discussed here. Non-porous membrane techniques (NPMTs) employ an organic or polymeric (solid or liquid) layer, placed between two other liquid phases. The analyte must actually be extracted from the donor phase, dissolve into the membrane in order to be able to pass through, and then be released in the acceptor phase. The behavior of the analytes largely depends on partition coefficients between the different parts of the membrane system. Only analytes that are easily extracted from the donor phase and, in addition, are easily released from the membrane into the acceptor phase will be transported. Thus, the separation is based on the same principles as LLE with back-extraction. It is thus possible to separate molecules of similar size, yet with different physicochemical properties [146,152]. The non-porous membrane technique can be subdivided into four main groups: (I) Supported liquid membrane extraction (SLME), (II) Microporous membrane liquid-liquid extraction (MMLLE), (III) Polymeric membrane extraction (PME) and (IV) Membrane extraction with a sorbent interface (MESI). SLME is the most widely used non-porous membrane technique [146], but various applications of MMLLE, PME and MESI have been reported as well [23,25,146-150,152]. It should be noted that MMLLE is considered to be NPMT, even though a membrane with micropores is used. All NPMTs utilise a membrane unit constructed from two blocks of inert material with a machined groove in each. A membrane is placed between the blocks and the total unit is clamped together. Hence, two flow channels are formed, one being the donor channel, the other being the acceptor channel. In principle, SLME utilises a pH shift between the donor phase, in which the analyte is uncharged, and the acceptor phase, in which the analyte is protonated, thus ensuring that no back-extraction in the (organic) membrane can occur. MMLLE is performed with organic solvent as the acceptor phase in the micropores of the organic membrane, and can therefore be compared with a single liquid extraction. MMLLE is mainly used for the analysis of hydrophobic compounds that cannot be extracted from an organic membrane into an aqueous acceptor solvent,

References pp. 39-44

28

Chapter 1

as is the case with SLME. PME is similar to SLME, with the exception that a polymeric membrane is used. Due to this membrane it is also possible to use organic solvent in the donor and/or acceptor phase. However, the composition of the membrane is fixed, limiting further chemical tuning. Furthermore, low diffusion coefficients and slow mass transfer may lead to slow extraction. MESI differs from the previous techniques in that a solid polymeric membrane is used. MESI was mainly developed for the combination with GC, thus in order to use a gaseous acceptor phase [146,148,152], while the donor phase is aqueous or gaseous. Obviously, MESI works best for the analysis of volatile and relatively non-polar compounds [152]. Most applications of MESI are in the environmental field for the analysis of aqueous samples [146,154-156]. Both PMTs and NPMTs usually use the terms efficiency and/or enrichment. The efficiency is defined as the ratio between number of moles input to the system during the extraction and the amount collected in the acceptor, and can be directly measured [ 150]. Efficiency should not be confused with the term recovery, which is commonly used with extraction techniques. Recovery is (or should be) relatively constant under the selected conditions, and should, therefore, not affect the accuracy of the system if the response is corrected. With membrane-based techniques, the efficiency is usually not allowed to become 100%, because of time dependence. It is obvious that efficiency may be sacrificed for speed if sensitivity is not of major concern. As a result, efficiency is not always constant, as the time for sample preparation can be varied. Moreover, various factors, e.g. the composition of the donor phase, acceptor phase, the membrane and the sample, can affect the efficiency of the system. The most often observed side effects, i.e. binding to matrix proteins and adsorption to the membrane, with consequent carry-over, have been described in various studies and these effects will be pointed out later in this chapter. Besides efficiency the term enrichment is also often mentioned, especially with PMTs. Enrichment is the accumulated amount of analyte in the acceptor phase during a given time. In membrane techniques the efficiency decreases with increasing donorflow. Contrary, the enrichment increases with increasing donor-flow (Fig. 1.12). At very low donor-flows the enrichment is close to zero, as the extracted analyte is diluted immediately in the acceptor phase. With high donor-flows, the efficiency is decreasing due to incomplete diffusion of the analyte into the acceptor phase, but the enrichment is increasing with increasing donor-flow. However, high donor-flows imply large consumption of sample, and can therefore only be applied if sufficient sample is available [ 146,150].

1.3.2.2 Porous membrane techniques As shown in Table 1.2, three major types of PMTs can be distinguished. Although a porous membrane is also used in microdialysis, the latter is a rather different technique, being mainly applied for in vivo studies [157-162] and will therefore not be discussed here. The applications of on-line filtration are limited to fermentation broths [23]. Electrodialysis has been coupled on-line to CE for the determination of inositol phosphates in plasma [163-165]. Only one report is available for the on-line coupling with LC [164]. Despite a high efficiency (95%), ephedrine could only be determined in serum three times before the membrane faltered. The studies on electrodialysis were merely explorative and no applications in routine analysis have been reported yet. The

New developments in integrated sample preparation for bioanalysis 1

.

.

.

.

.

.

.

0.8

29

120 100 8O

0.6 W

60 0.4 40 0.2

20

0 0

0.05

r

0.1

0 0.15

Fig. 1.12. Extractionefficiency E and enrichment factor Ee (arbitrary units) as functions of the reduced flow parameter % Note: q~ is the volumetric flow divided by the membrane area. (Reprinted from [150], with permission from Elsevier Science).

most common PMT is dialysis, employing a cellulose-based membrane and an aqueous donor and acceptor phase. Though some reports have been made about the on-line coupling of dialysis with GC [54,166] and CE [2,167,168], most bioanalytical applications couple dialysis on-line with LC. An overview of the results obtained with the latter system is presented in Table 1.3. The major disadvantages of dialysis are the typically low efficiency and the long dialysis time. Several approaches have been applied for concentration, e.g. automated sequential trace enrichment of dialysates (ASTED) [25,176,202] using a reversed phase enrichment column. Selective trapping has also been performed [177,182]. Furthermore, the use of membranes might result in adsorption of the analytes by the membrane [ 182]. Covering of the active sites by adding a surfactant can minimise adsorption [ 192,196]. The binding of analytes to macromolecules in the matrix such as proteins has an even more pronounced effect on the dialysis efficiency. Only the unbound fraction can diffuse through the membrane. As a consequence, the efficiency may be substantially lower than with aqueous solutions. This will especially be the case for compounds that bind to a high degree to proteins. This allows the determination the free fraction of analyte [174,176,185,188]. A number of ways have been suggested to release the bound drug from the protein. The simplest way to reduce protein binding is to dilute the sample. Another possibility is to change the pH of the sample, hereby changing the structure of the macromolecule and/or the charge on the analytes, causing the release of the analytes of interest [174-176,184,187,192]. Denaturation of the proteins can be performed by addition of a strong acid. However, this may also result in loss of bound drug that coprecipitates with proteins. A more selective and elegant strategy for releasing the analytes from the proteins is the addition of competing agents or displacers for the protein binding sites [24,174,176,180,187], e.g. fatty acids with appropriate chain

References pp. 39-44

Amino acids Barbiturates Enoximone + metabolite

Salicylic acid

Mitomycin C

Anticonsulvants Theophylline Corticoids Phenobarbitone Phenytoin Phenylbutazone Theophylline Warfarin Azidothymidine Des-enkephalin~/-endorphin Oxytetracycline

1985 1985 1986

1986

1987

1987

Oxolinic acid Flumequine Iopentol

Pholcodine

Benzodiazepines Rogletimide Antiviral drug* + metabolite#

1992

1992

1992 1992 1993

1991

1991

1990 1990

1988 1988

C ompound

Year

Plasma Whole blood Plasma Whole blood Plasma Plasma Plasma

Whole blood Plasma Whole blood plasma

Plasma Plasma

Serum Plasma Serum

Plasma Urine Serum

Serum

Serum Serum Serum

S ample

UV

Detection

UV MS/MS UV

Flu

UV

UV

UV

UV Flu

UV UV

UV

Salicylate electrode UV

TABLE 1.3 APPLICATIONS OF DIALYSIS ON-LINE WITH LC IN BIOANALYSIS

7.6 14 90%) and excellent peak shapes. As well as urine and plasma, a variety of other different matrices have been used in TFC, including brain homogenate, undiluted plasma, and microsome mixtures. There was no significant difference found between the various matrices, Fig. 4.17b. Currently

Turbulent flow chromatography in bioanalysis

121 intensity: 2759 cps

217

100" 90" 80' 70' 60' 50' 40' 30' 20' 10' 0

' 161 0:21

Ik

261 0:41

361 1:02

401 1:22

501 1:43

66f" 761 2:03 2:24

-'-

801 2:44

A

. . . .1 ~ . . . . . 901 1 0 0 1 1101 Scan 3:05 3:25 3:46 Time

.t,_..,~,~.

Fig. 4.16. Chromatogramsof 10 ng/mL of taxol spiked in human plasma.

the technology is being refined to allow the analysis of whole blood with no sample pretreatment.

4.3.2.2 Forensic applications Grant [56] developed an assay for the analysis of cortisol and a metabolite in urine. Cortisol is a naturally occurring hormone which influences metabolism, inflammation, electrolyte and water balance. Their synthetic derivatives are used therapeutically for their anti-inflammatory and immunosuppressive actions. They are used in certain sports to improve the performances of the athletes (euphoria, motor activity). The ability to distinguish between the active components and some of the metabolites formed within the body is important for the accurate determination of the amount of initial drug taken. A focus method was employed which allowed urine samples to be prepared and analysed in less than five minutes. The data presented in Table 4.2 shows that cortisol

TABLE 4.2 RESULTS OBTAINED BY GRANT [57] FOR THE ANALYSIS OF CORTISOL OVER THE RANGE 0.25 ng/ml-250 ng/mL Concentration/ng/mL 0.25 25 100 250

References pp. 127-128

Average response (n = 3)

%RSD

2.74E2 1.96E4 7.86E4 2.33E5

13.7 7.4 5.2 8.9

Chapter 4

122

Deprenyl 300~ :1

~ 2ooo.

2

~

" Time

4|

2

5'

;

~

.......

Time(rain)

(rain)

20001800"

Haloperidol

K252a

1600~., 1400~ " 6000 ,..., :~" 1200" "~ ~

v

I000-

..~"~ 4500

81)03000 600-

J~

400I

200" i

0 _ 2

.

.

9

4

2

s

Time(rain) Time

(rain)

4000~

CEP-1347 300OO"

2

1000~

02

3

4

5

0 2

Time(rain)

~ 3

4

Time(rain)

Fig. 4.17. (a) Example chromatograms of 6 compounds, all at 5 ng/mL, with differing hydrophobicites all

analysed directly from rat plasma using one standard method.

can be detected precisely over the 0.25 to 250 ng/mL range. This approach has also been used for a wide range of opiate analysis in urine. In particular the focus method has been successfully used in the analysis of codeine, dihydrocodeine, morphine, and heroin. As a comparison, the previous approach to analysing these types of compounds is using an automated solid phase extraction technique followed by blowing the subsequent extract down, then reconstitution in a suitable mobile phase and injection onto a GC-MS. HPLC-MS can be used but sample preparation is still required before the sample can be analysed. The whole sample preparation time takes about 30 minutes

Turbulent flow chromatography in bioanalysis 7oooo-

123

Plasma

Urine

7OOOO" 6OOO0" 5O0OO

4OOOO

~ 4oooo-~

j

3oooo. 2OO0010000-

lOOOO-

~2

;

~

J

o

-

2 Tin~

Time (mln)

?0000"

7OOOO"

Brain

(mill)

Intestine

6OOOO5~000,

5OOO0

I=1.,1(}000.

40ooo

21)00010000-

10000-

0-

0-

2

2

Time (min)

Time (mill)

CSF 300OO

~9

20000

10000"

o

. 2

.

.

.

.

.

" ;

, 4

5

Time (rain)

Fig. 4.17. (b) 50ng/mL K252a in rat plasma, urine, brain homogenate, liver homogenate, intestinal perfusates and cerebral spinal fluid.

with the analysis time being about 10 minutes. The sample preparation can be parallelised, substantially reducing sample preparation time but the samples still have to be analysed after they have been prepared.

4.3.3 Practical issues in bioanalytical TFC During the development of turbulent flow chromatography to the analysis of biological samples, there have been several issues that have been raised that gave analysts cause

References pp. 127-128

124

Chapter 4

for concern regarding the technique. Some of the major issues are discussed in the next section, with solutions to them supplied. 4.3.3.1 Carryover

Carryover is a problem that has been reported by some authors, although it is compound dependent. The investigations by Chassaing [48] revealed that there was substantial carryover, more than 1%, for the compounds being investigated. Several sources for the carryover were determined including the autosampler, the column, and the valves. The majority of the observed carryover, about 40%, was from the autosampler. It was discovered that overfilling the sample loop on the autosampler resulted in a gradual build up of proteins on the porous rotor seal of the autosampler. A partial loopfill, and a careful rinse procedure of both the syringe and the injection valve, substantially reduced carryover, by almost a third, to acceptable levels. Carryover attributed to the column was initially identified as being due to an ion pairing mechanism rather than a pH mechanism. Thus, the addition of 0.01% TFA was sufficient to remove the carryover. However, Chassaing noted that to make the column cleaning more genetic a range of different solvents were used in the column cleaning step. In particular a high pH cleaning step was also added (0.1% NH3, pH 8), as well as a strongly elutropic organic solvent such as THF/acetonitrile. The final source of carry-over appeared to be the rotor seals, and the Tefzel stators in the Rheodyne six port valves. Use of one valve reduced the carryover slightly, but compromised the overall performance of the system. Instead a series of valve switching was incorporated to ensure that the various channels were cleaned thoroughly. This resulted in an increase in the time of analysis but meant that the method became genetic and applicable to a wide range of compounds. The use of multiple valves results in a greater potential for contamination of the system, resulting in carryover. There are several mechanisms to alleviate this problem. The approach taken by Chassaing has proven successful, but it is possible to reduce carryover from the valves by changing the material of the rotors and stators. Grant et al. [57] have demonstrated that use of PAEK substantially reduces carryover. 4.3.3.2 Pressure build up

Although using a turbulent mobile phase has been shown to improve the lifetime of a column when performing bioanalysis, it was discovered that centrifugation [43] of the sample prior to analysis improved the longevity of the column substantially. It is now standard procedure to centrifuge all biological samples prior to analysis with turbulent flow chromatography. This has been shown to remove excess particulates from the sample, which cause the premature column failure. Another source found for the premature failure of columns was the storage temperature. Work performed by Takarewski et al. [43] clearly demonstrates that storing the samples at 5~ as opposed to room temperature does affect the lifetime of the columns since the samples deteriorate at the higher temperatures quicker than at the lower temperature.

Turbulentflow chromatography in bioanalysis

125

4.3.3.3 Protein binding A significant finding from the work of Chassaing et al. [48] was that although the three compounds analysed were highly protein bound (>90%), there was no significant reduction in analyte response when comparing an injection of analyte spiked into plasma with that of an identical concentration in aqueous solution. Acidification of the load solvent ensures that the proteins are softly denatured and that the analyte under investigation can be trapped by the turboflow column. This has also been observed by other authors [51,52].

4.3.4 Environmental applications of TFC Work performed at the University of Leipzig [58,59] has shown that substantial benefits can be achieved with the use of turbulent flow chromatography in the analysis of a range of water samples, including river and lake samples. Traditional methods of water analysis for organic contaminants, such as pesticides and herbicides, involves a lengthy analysis procedure with many different steps. Turbulent flow chromatography has allowed the analysis of environmental samples without the need of complex sample preparation. In the method developed it is possible to determine concentrations of contaminants in river water for all of the compounds listed between 1-125 ng/litre in an analysis time of about 15 minutes. TFC has been successfully coupled to other analytical techniques to obtain greater sensitivity within the environmental field. Using the turboflow column as an on-line

500000 -

R2 = 0.9865

a 2 = 0.9844

R2 = 0.999

R2 = 0.9765

R2

9 Diuron 9 Chlortoluron

400000- R 2 = 0 . 9 9 8 9

m "~ r == 300000

9Isoproturon

R2 = 0.999

600000 - R 2 = 0 . 9 9 9 2

x Simazin Atrazin

R2 = 0.993

9 Terbutylazin

= 0.9986

+ Prometn]n

R2 = 0.9993 =E 5n, 2O0000 " R 2 = 0 . 9 9 9 5

. Chlorfenvinphos - Chlorpyrifos 9 Alachlor

100000

9 Tdfluralin

0

20

40

60

80

100

120

140

C o n c e n t r a t i o n (spiked) [ngll]

Fig. 4.18. Calibrationdata for 12 compounds extracted from clean water, comparable data was also obtained for fiver water samples.

References pp. 127-128

126

Chapter 4

solid phase extraction cartridge allows larger sample volumes to be injected, which is critical for analysis of pollutants within an aqueous media. Typically the detection limits required for organic pollutants in water are in the range of pg/mL, a factor of 1000 lower than typically required for bioanalysis. In order to obtain such sensitivity solid phase extraction or liquid/liquid extraction has been used to pre-concentrate the sample volume from several hundred millilitres to a few hundred microlitres. The use of turboflow chromatography allows the sample to be loaded onto the extraction column substantially faster than with solid phase extraction and since all the sample is being analysed, less initial sample is required. Results from Asperger et al. [58] have shown that this approach can yield high recoveries, and very low limits of detection. Using a single valve method these authors were able to detect l pg/mL of a pesticide mixture spiked into clean water, using an initial sampling volume of 10mL. However, this technique is not applicable to non-polar compounds such as PAHs, chlorobenzenes, and chloronitrobenzenes, as these compounds do not ionise well using the LC-MS interface. Large volume injection using a programmable temperature vaporizing (PTV) injector for GC-MS is another technique that has also been utilised to increase the sensitivity of the analytical technique for GC analysis. It relies on the compounds of interest being dissolved in a more volatile solvent, typically the eluant from solid phase extraction. The analytes are trapped onto a sorbent, whilst most of the solvent is evaporated before the compounds of interest are swept into the column. This is becoming a routine tool within environmental laboratories. It is however, possible to couple this technology to TFC. Results from Bahl et al. [59] show that coupling of the two techniques is feasible with data being presented on apolar compounds not applicable for direct analysis using TFC-LC-MS.

4.3.5 Capillary turboflow chromatography Ayrton et al. [60] noted that using smaller ID columns would reduce the solvent consumption. Using specially prepared 180 ~m diameter column, packed with 30 ~m particles, enabled flow rates to be used that allowed direct introduction into the mass spectrometer with no requirement to split the flow. This results in a lowering of the detection limits, and validation data was presented showing concentrations from 0.5 ng/ ml to 500 ng/ml with only 5 p~l sample injected (as with the previous work the sample was diluted with internal standard 1 : 1, equivalent to 2.5 ~1 injection of neat plasma). Capillary turbulent flow chromatography has also been applied to the analysis of methadone and its metabolites from serum. Souverain et al. [61] used a single valve technique with 180 ~m i.d. column packed with 30 ~zm particles. The linearity was shown to be good, as was the limit of quantification which was reported as 10 ng/mL using a single quadrupole mass spectrometer as the detector. The LOQ of the method was found to be improved by the addition of a second analytical column, which allowed for some focusing of the analytes and also allowed for chromatographic separation of the various components.

Turbulent flow chromatography in bioanalysis

127

4.4 CONCLUSIONS There is a substantial amount of evidence to indicate that TFC is applicable for the analysis of a wide range of compounds. The lack of sample preparation means that method development time is substantially reduced in comparison to standard approaches. It is a technique that is geared towards high throughput laboratories that have a sample throughput problem. However, it is not a panacea, just as with other sample preparation techniques such as SPE, protein precipitation and liquid/liquid extraction there will be compounds that will be difficult to analyse, however, as a front line approach to bio-analysis, it is probably the quickest and easiest to use. A fuller understanding of the mechanism of turbulent flow chromatography is being developed, and research groups are coupling the predictive power of computational fluid dynamics with classical chromatographic theory. This will give a greater insight into the effects of fluid dynamics on a chromatography system, which will help in the design of superior stationary phases. It may lead to a situation where laminar flow chromatography becomes redundant, and the only form of chromatography is based around chaotic fluid dynamics.

4.5 REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

H.M. Quinn and J.J. Takarewski, International Patent Number WO 97/16724. J.J. van Deemter, EJ. Zuiderweg and A. Klinkenberg, Chem. Eng. Sci., 5 (1956) 271. G. Guiochon, Anal. Chem., 52 (1980) 2002. J.H. Knox and M. Saleem, J. Chromatogr. Soc., 7 (1969) 614. R.A. Hartwick and D.D. Dezaro, in: E Kucera (Ed.), Microcolumn High Performance Liquid Chromatography, Elsevier, Amsterdam, 1984. J.H. Knox, J. Chem. Soc., (1961) 433. J.C. Giddings, Unified Separation Science, Wiley-InterScience publications, 1991, p. 283. J.C. Giddings, Anal. Chem., 34 (1962) 1186. J.H. Knox, J. Chromatogr., A, 831 (1999) 3. J.H. Knox and R.EW. Scott, J. Chromatogr., 282 (1983) 297. G.I Taylor, Proc. Royal Soc., A219 (1953) 186. J.C. Giddings, Anal. Chem., 35 (1963) 1338. J.J. van Deemter, 2nd Informal Symposium of the Gas Chromatography Discussion Group Cambridge, 1957, (quoted in Gas Chromatography, J.H. Purnell (Ed.) Wiley London, 1962, p. 128.). J.C. Giddings, Dynamics of Chromatography - part 1, Marcel Dekker, New York, London, 1965. M. Golay, Gas Chromatography, D. Desty (Ed.), 1958, Butterworth, London, p. 35. R.B. Bird, W.E. Steward and E.N. Lightfoot, Transport Phenomena, Wiley, New York 1960. R. Aris, Proc. Roy. Soc., A235 (1956) 67. L.J. Tichacek, C.H. Barkelew and T. Baron, A.I.Ch.E.J., 3,439, (1957). R. Aris, Proc. Roy. Soc., A252 (1959) 538. T. Pretorius and T.W. Smuts, Anal. Chem., 38 (1966) 1. J.C. Giddings, W.A. Manwaring and M.N. Myers, Science, 154 (1966) 146. C.J. Oberhauser, A.E. Niggebrugge, D. Lachance, J.J. Takarewski, M.M. Pegram and H.M. Quinn, LCGC. 18 (2000) 716. H.A. Kusch, J.M. Ottino and D.M. Shannon, Ind. Eng. Chem. Res., 28 (1989) 302. J. Baldyga, J.R. Bourne and S.J. Hearne, Chem. Eng. Sci., 52 (1997) 457. M.C. Jones, R.D. Nassimbene, J.D. Wolfe and N.V. Fredrick, Chem. Eng. Sci., 51 (1996) 1009. G. Taylor, Proc. Roy. Soc., A219 (1953) 186.

128 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61

Chapter 4 A.A. Barres, Chem. Eng. Sci., 5 (1997) 807. O. Reynolds, Roy. Soc. Phil. Trans., 35 (1883) 84. C.O. Bennett and J.E. Myers, "Momentum, Heat, and Mass Transfer", 3rd Ed., McGraw-Hill, NY (1982) pp. 202-209. E Moin and J. Kim, Scientific American, 276 (1997) 62-68. J. Baldyga and J.R. Bourne, Chem. Eng. Commun., 28 (1984) 231. S. Ergun, Chem. Eng. Prog., 48 (1952) 89. F.A.L. Dullen, Chem. Eng. J., 10 (1975) 1. I.E Macdonald, M.S. E1-Sayed, K. Mow and EA.L. Dullen, Ind. Eng. Chem. Fundam., 18 (1979) 199. T. Farkas, G. Zhong and G. Guichon, J. Chrom. A., 849 (1999) 35. J.C. Giddings, Unified Separation Science, J.Wiley & Sons (1991) p. 65. EC. Blake, Trans. Amer. Soc. Chem. Engrs., 14 (1922) 415. EC. Carman, Trans. Inst. Chem. Eng., 15 (1937) 150. S.P. Burke and W.B. Plummer, Ind. Eng. Chem., 20 (1928) 196. A. Edge, 11th Nordic Mass Spectrometry Conference, Loen, Norway, 18-21 Aug. 2001. A.E. Fick, Annalen der Physik (Leipzig), 170 (1855) 59. EW. Atkins, Physical Chemistry 2nd Edition, Oxford University Press, p. 879. J. Takarewski, D. Mageira and H.M. Quinn, 47th ASMS Conference, Dallas, Texas, 1999. B. Biddlecombe, 25th Annual meeting British Mass Spectrometry Society, Cambridge, U.K., 19-20 Dec 2001. W.D. van Dongen, R. Ramaker, F. van Schalk, B. Ooms and E. Koster. 25th Annual meeting British Mass Spectrometry Society, Cambridge, U.K., 19-20 Dec 2001. J. Ayrton, Rapid Commun. Mass Spec., 11 (1997) 1953. J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallet and R.S. Plumb. Rapid Commun. Mass Spectrom., 11 (1999) 1953. C. Chassaing, P. Macrae, E Wright, A. Harper, J. Luckwell, K. Saunders and R. Venn, Chromatographia, 53 (2001) 122. W.N. Kennedy, H.M. Quinn, J.J. Takarewski, C.J. Oberhauser and D.J. Malpas, J. Chromatogr. Submitted. D.Zimmer, V. Packard, W. Czembor and C. Muller, J. Chrom. A., 854 (1999) 22. M. Jemal and W. Yuan-Qing, Rapid Commun Mass Spectrom., 12 (1998) 1389. J-T. Wu, H. Zeng, M. Qian, B.L. Brogdon and S.E. Unger, Anal. Chem., 72 (2000) 61. N. Brignol, R. Bakhtiar, L. Dou, T. Majumdar and F.L.S. Tse, Rapid Commun. Mass Spectrom., 14 (2000) 141. W.D. van Dongen, TFC user group meeting, Holland, Oct. 2001,. J.L. Herman, Rapid Comm. Mass Spectrom., 16 (2002) 1-6. R. Grant, 15th IMSC, Barcelona, Aug. 2000. R. Grant, C. Cameroon, S. Mackenzie and M. Young, Rapid Commun. Mass Spectrom., 16 (2002) 1785-1792. Asperger, J. Efer and W. Engewald, 10th Symposium on Handling of Environmental and Biological Samples in Chromatography, April 2001, Mainz, Germany. K. Bahl, J. Efer, A. Asperger and W. Engewald, HTC 2002, Bruges, Belgium. J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallet and R.S. Plumb, Rapid Commun. Mass Spectrom., 13 (1999) 1657. S. Souverain, D. Ortelli, S. Rudaz, J.L. Veuthey and E. Varesio, J.E 17th Montreux symposium on LCMS, SFC-MS, CE-MS, and MS-MS, Switzerland, Nov. 2000.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations,Vol. 4 92003 Elsevier Science B.V. All rights reserved

129

CHAPTER 5

Chiral bioanalysis D.M. Wallworth and J.T. Lee Advanced Separation Technologies Ltd, Blake Street, Congleton, U.K.

5.1 INTRODUCTION Although the importance of enantiomeric drugs and the science of analytical chiral chromatography have become well established [1,2], new challenges have presented themselves in the last few years as single enantiomeric drugs have progressed through the development stages into pre-clinical and clinical studies. Chiral analysis of biological samples introduces new demands not previously experienced in chiral chromatography. In the past, achieving baseline separation of two enantiomers within 30-40 minutes under any mobile phase conditions possible was acceptable, if not considered quite an achievement. For bioanalysis, not only is a good enantioseparation now required, but also speed becomes far more critical. For the high throughput of many hundreds of samples under the commercial time pressures of a clinical trial, every minute saved is crucial. Mobile phase choice also becomes more significant because of the resulting requirements of sample preparation, including its speed and ability to be automated. Aqueous chiral methods are often preferred over normal phase systems, allowing the possibility of simple or on-line sample preparation techniques, and for reasons of better solubility of polar compounds, the use of less costly solvents and, most importantly, easier interfacing of HPLC with tandem mass spectrometry detection (HPLC-MS-MS). In addition, lower therapeutic levels have increased the demands on sensitivity and on the chromatography itself. There is an increasing drive to select either single isomer or racemate early in the development process. Early screening usually defines the pharmacokinetic profile, but there is always the possibility of in vivo chiral inversion (prochiral to chiral, chiral to achiral, chiral to diastereomer, chiral to chiral) when administering a single enantiomer. Potential enantioselective absorption, distribution, metabolism and excretion define a range of bioanalytical studies to be completed. For these studies, HPLC-MS-MS is increasingly being used for optimum selectivity and sensitivity, especially where there is the possible presence of chiral metabolites that may need to be monitored. References pp. 180-184

130

Chapter 5

5.1.1 Scope and aim This review covers the developments in chiral bioanalysis since 1995, focusing mainly on the techniques used in HPLC but also providing a brief introduction to bioanalytical applications in GC, CE and SFC. It will include a review of the principle types of chiral stationary phases (CSPs), their mechanism of action, typical mobile phase choices and applicability. A method of choice must be sufficiently robust to withstand large numbers of biological samples. It must be capable of method transfer to other research laboratories, such as a contract laboratory facility, as the drug development process progresses. New chemical entity submissions to the FDA and other organisations around the world also need to incorporate validated data using freely available methods. Direct HPLC methods (rather than the use of achiral stationary phases with chiral additives or derivatisation of the solute with chiral reagents) are generally preferred in pharmaceutical research. It is mainly for these reasons that HPLC (and in particular HPLC-MS-MS) will be a main focus for this chapter. It is also for these reasons that the chapter will only include commercially available and proven technologies. It is not intended as a comprehensive review of all CSPs currently available for all types of applications.

5.1.2 The mechanism of chiral recognition and choice of CSP In nature, chiral recognition by amino acids, proteins, enzymes, receptors is commonplace. This capability has been utilised for chiral discrimination in HPLC or GLC, by utilising such selectors in a CSE The protein phases such as Chiral-AGP or the amino acid w-acid/base phases are two examples. Many CSPs initially developed from natural products have been modified to increase robustness and applicability to chiral bioanalysis. Often, more than one CSP may appear to be suitable for a particular chiral assay and the choice may well be made on peak shape (for best sensitivity), elution order (for very low enantiomeric ratios in a single enantiomer racemisation study) or retention times for speed. The basic concept of a chiral interaction in HPLC is the formation of a dynamic diastereomeric complex between the CSP and the analyte. The complex forms by utilising one or more of different types of interactions such as hydrogen bonding, -rr-w interactions, inclusion complexing, dipole-dipole and ionic interactions, to name just a few. Some types of interaction are specific to certain CSPs and some are specific to certain mobile phases. Many CSPs operate under several different types of interaction and most also have the capability for non-chiral interactions. It is often possible to look for useful points of interaction between the CSP and the racemate and then to choose the best CSP and mobile phase to enhance that effect, aiming to get the maximum strength for the differential chiral interactions of the two enantiomers. Non-chiral interactions need to be kept at a minimum but are generally useful for anchoring the racemate. Along with lowered solubility and lowered mass transfer, they are often responsible for peak tailing observed in chiral HPLC. In practice, automated and genetic screening techniques are frequently used to determine the CSP of choice from a range that includes those best for the compound

Chiral bioanalysis

131

TABLE 5.1 CHIRAL INTERACTIONSBETWEENCSP AND SOLUTE Interaction

Mobile Phase Type for the Interaction to Dominate

Ionic Ion-dipole interactions Hydrogen bonding Dipole-dipole Steric interactions Hydrophobic/inclusion interactions ~r-w interaction

Reversed-phase, polar organic mode Reversed-phase Polar organic mode, normal phase, reversed-phase Normal phase, polar organic mode Normal phase, polar organic mode Reversed-phase Normal phase

class of interest, or those best for the mobile phase of choice (for reasons of solubility or detection). Chiral method development kits are also available from many chiral column suppliers (for example, Regis, Advanced Separation Technologies, Akzo Nobel) for speed in determining the optimum CSE An indication only of the most suitable CSP is often the result of such a screen, optimisation being conducted on the single (often longer) column separately. Table 5.1 shows the types of chiral interactions that occur in many CSPs and in which mobile phase they are most enhanced. If the functional groups close to the chiral (stereogenic) centre have specific interaction capabilities (such as w-deficiency), this may well influence the initial choice of CSE There have been many attempts over the years to classify and characterise the wealth of stationary phases that are available, but perhaps the simplest is to categorise CSPs in terms of their main interaction. Table 5.2 provides details of the most commonly used types of CSP, their main mode of interaction and commercial notation.

5.1.3 Mobile phase types Reversed-phase solvent systems are generally composed of an aqueous portion, which may or may not be buffered, and a miscible organic portion. A typical response is to observe an increase in retention and resolution as the aqueous portion is increased, although this may also depend on pH. For the macrocyclic antibiotic and cyclodextrin phases, however, an increase in retention and resolution has been observed [3] at both high and low compositions of the organic component. Typical chiral interactions in reversed-phase chromatography are generally ionic, hydrogen bonding and inclusion complexation. The usual components for normal phase eluants are hexane (or isohexane, heptane) with 2-propanol (or ethanol, isobutanol) as the lesser component. As the concentration of the polar component is increased, retention and resolution decrease. Predominant interactions are in this case -rr-~r interactions and hydrogen bonding. The polar organic mode is a novel mobile phase originally developed by Armstrong for use with the cyclodextrin type CSPs. It was shown [4] to be highly advantageous for obtaining efficient separations on these phases and has subsequently References pp. 180-184

Chapter 5

132 TABLE 5.2 PRINCIPAL COMMERCIALLY AVAILABLE CHIRAL STATIONARY PHASES Chiral Stationary Phase Type

Commercial CSP

Typical Mobile Phase Conditions

Charge transfer (Pirkle TM type) phases (Hbonding, w-complex)

Regis Kromasil Sumichiral Chirex

NP (some RP)

Cyclodextrins (H-bonding, inclusion) Cyclodextrin Derivatives (as above, plus w-complex for Cyclobond SN, RN, DMP)

CYCLOBOND TM ChiraDexTM NucleodexTM

RP and POM NP, RP and POM

Chiral Polymers (H-bonding, hydrophobic, plus w-complexation or ionic interaction where indicated) 1. Proteins: o~-acid glycoprotein bovine serum albumin human serum albumin ovamucoid cellobiohydrolase 2. Cellulose/Amylose (w-complexation) 3. Cellulose/Amylose, Aqueous

CHIRAL-AGPT M ResolvosilTM CHIRAL-HSAT M Ultron ES-OVM CHIRAL-CBHT M Chiralcel and Chiralpak TM

RP RP RP RP RP NP Only RP Only

Astec CLC Nucleosil Chiral 1 Chiralpak WH/WM/WE

Aqueous Cu complex

Crownpak CR

Aqueous perchloric acid

CHIROBIOTIC T M

NP, RP and POM

Ligand Exchange (ionic complex)

Crown Ethers (ionic)

Macrocyclic Glycopeptides (multiple)

been used extensively with the macrocyclic antibiotics. It is composed of methanol (or acetonitrile/methanol mixtures when used with cyclodextrin CSPs) with added acid and base and is therefore essentially anhydrous, providing long term stability for most of these CSPs. It is simply the ratio of acid to base that mainly controls enantioselectivity, with ionic interactions playing a role when the polar organic mode is used with the macrocyclic antibiotics. One main advantage of this mobile phase is that there are few requirements of the analyte for separation to be effective: there must be at least two hydrogen bonding functional groups on the chiral molecule, only one of which must be on or near the stereogenic centre. This low requirement means that this mobile phase system has the potential for very broad applicability in chiral HPLC. Typical functional groups for hydrogen bonding are: Halogen: I > Br > C1 > F Amine: 3 ~ ~ 2 ~ > 1~ Carbonyl: -COOH, -CHO, -C=O, -COOR Sulpho-, phospho- and hydroxyl groups Methoxy groups

Chiral bioanalysis

133

The organic solvent primarily controls the retention and the ratio of acid to base controls the resolution. The type of acid and base can also be chosen for interfacing with HPLCMS, and it is in this area that its use is seen to be growing strongly. The polar organic mode operates through hydrogen bonding, dipole-dipole interactions and/or ionic interactions (when used with the macrocyclic antibiotic CSPs). Whilst this mobile phase is essentially a 'normal phase' solvent, it is most commonly known as the 'polar organic mode', or 'polar ionic mode' when used with the macrocyclic antibiotic CSPs. More recently, a modified version of the polar organic mode has been used with the cellulosic and amylosic CSPs, although typically either acid or base are added, rather than both

[5]. 5.1.4 Direct vs. indirect chiral separations Alternative methods for chiral separations include the use of derivatisation techniques whereby a chiral derivatisation reagent is used to form a diastereomer of the chiral solute. An achiral (frequently a C18) column is then used for the separation since diastereomers now have different physical properties that can be utilised in an HPLC method. Whilst this is a feasible technique, it tends not to be practical for several reasons. Firstly, it relies on optimum chiral purity for the derivatisation reagent, on an assumption that each enantiomer will react with the reagent to an identical extent, that the derivatisation reagent is highly selective and has non-detectable or non-interfering by-products, and that the reaction conditions are simple and fast enough for high throughput clinical assays. In most cases, the increased validation issues alone are sufficient to discount achiral methods as a method of choice. In addition, indirect methods have largely been superseded by technological improvements in stationary phases, but are always an alternative where no direct method can be found.

5.1.5 Achiral-chiral column switching techniques There are many instances of methods involving achiral-chiral column switching techniques being incorporated into biological assays. Where the chiral drug has active chiral metabolites, all solutes of interest will need to be quantified. Often it is necessary to combine the separating capabilities of both traditional reversed-phase and chiral stationary phases to achieve this. This can be accomplished either on-line (where a column switching valve directs the fractions containing chiral moieties onto a coupled chiral column), or off-line (collecting fractions for the drug and any active chiral metabolites and evaporating to dryness before injecting onto the CSP in a separate assay). Instances of the former are prevalent in the literature, despite the increased difficulties of optimisation and choice of compatible mobile phases.

5.1.6 HPLC-MS Technological advances over the past five years have led to a rapidly increasing use of MS detection in drug research. Sensitivity and specificity are often increased over alternative methods of detection and where characterisation of the molecule is required, References pp. 180-184

134

Chapter 5

MS is particularly attractive. When combined with HPLC, MS has gained in popularity for the trace analysis of drugs and their metabolites in biological fluids. Whilst there are a large number of pharmaceutical applications already developed using normal phase solvents, it is sometimes beneficial to re-develop such methods to reversed-phase ones or to take additional safety precautions when drug development requires a bioassay and the high sensitivity of HPLC-MS. Utilisation of normal phase methods have a potential for an unstable ion current and poorer sensitivity in HPLC-MS and combined with high voltage electrospray ion source, it presents a potential explosion hazard. Post-column mixing with an aqueous solvent can be used to circumvent these problems [6,7]. It is generally considered that the rigorous use of nitrogen as nebulising and solvent drying gas, plus the post-column addition of water are essential for the safe use of normal phase solvents in HPLC-MS. If an alternative method is available that utilises volatile organic buffers, especially with high organic modifier content, then this may be preferred. Electrospray ionisation (ESI) and atmospheric pressure chemical ionisation (APCI) are two of the most successful HPLC-MS interfaces, producing little or no fragmentation of the solute. ESI does have some limitations in its low tolerance of salts, detergents and inorganic buffers which give signal suppression. It is also concentration sensitive such that the best results are achieved at relatively low flow rates, requiring the use of narrow bore (typically 2 mm internal diameter) columns, or the use of standard (4.6 mm internal diameter) columns with post-column split flow, allowing a portion of the eluent to be transferred to the MS interface. Recent new technologies such as turboassisted ion-spray (TSIP) allow for flow rates that are typically 1.0 to 1.5 ml/min. APCI offers a wider linear dynamic range than ESI and is relatively more tolerant of higher buffer concentrations, providing best sensitivity by the addition of an electrolyte. Typically, no sensitivity gains are achieved with lower internal diameter columns or lower flow rates for APCI. In designing a chiral assay for HPLC-MS, it is important that inorganic buffers and high aqueous mobile phases are avoided as these can potentially lead to ion suppression. Where unavoidable (as in the protein phases), success with HPLC-MS can be achieved by changing buffer type, and ammonium acetate has been successfully used in many cases. When using flammable solvents such as hexane in normal phase assays, safety measures are required when using ESI or APCI interfaces. In a recent study of verapamil, sotalol, doxazosin and oxybutynin using API-MS-MS [8], a 2 mm internal diameter Chiralcel AD CSP was used successfully in normal phase mode using hexane, 2-propanol and diethylamine as mobile phase. The heated nebuliser system of the SCIEX system used nitrogen as both the nebuliser and auxiliary gas to displace any oxygen and post-column addition of an aqueous reagent provided the ion concentrations required to sustain ionisation in the API source and to dilute out the hexane. A mixture of 25% 0.025 M aqueous ammonium acetate and 75% 2-propanol was found to be miscible with mobile phases of up to 95% hexane, without compromising chromatographic resolution. The column flow rates were maintained below 0.2 ml/min and the reagent added at a ratio of 4:1 to the mobile phase. Although this technique potentially introduces sensitivity problems by virtue of dilution, it has been used in many other assays such as the normal phase separation of omeprazole [9] and Org4428 (an antidepressant) [ 10].

Chiral bioanalysis

135

For reversed-phase chiral HPLC/MS bioanalysis, API-compatible solvents such as methanol, acetonitrile and ethanol enhance the formation of ions and assist in the nebulisation and desolvation processes to help achieve optimum sensitivity. Of the reversed-phase CSPs available, the macrocyclic antibiotics appear to be extremely broad ranging and versatile and to be cost-effective for HPLC-MS bio-analysis [ 11 ], with no separation deterioration observed for a plasma assay of methylphenidate after approximately 2500 injections. Other similar examples have been reported [e.g. 12,13]. Although electrospray is well suited to polar conditions, it shows poor sensitivity in the presence of triethylamine (TEA) [ 14]. Plasmaspray ionisation HPLC-MS interfaces, however, can use TEA to advantage. Suprofen and p-chlorowarfarin were used [14] to demonstrate the use of a Cyclobond I 2000 RN column in the polar organic mode in conjunction with MS detection as a viable method for the analysis of those compounds having no UV chromophore, or having UV chromophores absorbing below the UV cutoff of this mobile phase (ca. 240 nm).

5.1.7 Temperature It is well documented that a reduction in the temperature at which chiral chromatography is performed generally increases enantioselectivity since in the majority of cases chiral separations are enthalpy dominated. However, there have been a few cases reported where elution order is entropy-governed and therefore is temperature dependant such that there is a 'crossover' point where elution order reversal occurs. In a study on the chiral separation of sotalol on Chiral-CBH [ 15] it was found that the (R)-enantiomer eluted before the (S)- at 5-15~ but that the isomers eluted in the opposite order at 40~ The crossover point occurred between 25 and 30~ and was found to be mobile phase dependant. Although relatively rare, there have been other reports of this phenomenon, e.g. for R,S prominal on a cyclodextrin column [16], for mosapride on Chiral-AGP [ 17] and for various compounds on Chiralcel OD [18]. These examples emphasise the need for temperature control in chiral HPLC.

5.1.8 Validation For bioanalytical methods, there are many different issues associated with validation. It is important that the measured concentrations of each enantiomer replicates the in vivo levels and is not affected by in vitro degradation or interconversion. This encompasses a wide range of issues from sample collection (the effect of anti-coagulants and/or acidification on enantiomeric ratios), sample storage, sample preparation (any effect of protein precipitation or extraction), and the effect of an achiral-chiral column switching technique or the effect of derivatisation, if used. The success of the enantiomeric separation in terms of degree of resolution will also affect calibration, limits of quantification (LOQ) and linearity of the method. All of the above are highlighted in a useful review article [ 19]. See also Hill, this volume. References pp. 180-184

136

Chapter 5

5.1.9 Gas liquid chromatography (GLC) Although just 25 references for chiral bioanalysis by GLC were published for the period covered by this review, by far the majority were determinations of small molecules such as biomarkers, or of drugs of abuse. Because of the inherent sensitivity of GLC detection systems, chiral GLC provides high sensitivity, often to ng or pg concentrations. It therefore is useful as an alternative detector for use with LC methods by heart cutting the analyte of interest, evaporating to dryness and re-injecting onto a GLC system. In general, however, most drug molecules are insufficiently volatile to enable direct GLC methods, requiring derivatisation to acquire a degree of volatility. GLC often also requires more complex sample preparation. For these reasons, and the rapid growth in the use of highly sensitive H P L C - M S - M S methods, chiral GLC is not frequently used in pharmaceutical research and so will not be covered further here.

5.1.10 Capillary electrophoresis (CE) Method development in CE is often to be fast, it is a versatile technique and consumable costs are low. Even so, there is a reluctance to use the technique for bioanalysis, perhaps for reasons of available expertise, generally lower sensitivity or for difficulties experienced in method transfer. A comprehensive and valuable review of the use of CE in bioanalysis has recently been published and offers excellent coverage of the topic

[2o]. 5.1.11 Supercritical fluid chromatography (SFC) As chiral HPLC has matured, interest in the use of SFC has been renewed for this area. SFC offers several advantages over HPLC, such as the lower pressure drop across the column, allowing for increased flow rates, and the possible use of multiple columns or column coupling. Supercritical CO2 has a polarity similar to that of pentane, so must be used in conjunction with a polar modifier such as methanol, acetonitrile or dichloromethane. Acidic and basic modifiers can also be added. For this reason, it is generally considered as possible replacement for normal phase chiral applications [21 ]. Samples that require aqueous conditions are not generally amenable to SFC. SFC can be performed using either a capillary column (5-100 Ixm i.d.) or a packed column (designated pcSFC), using columns designed for HPLC. Whilst many CSPs have been used successfully in pcSFC, including celluloses, charge transfer, cyclodextrins [22] and macrocyclic antibiotics [23], there is no guarantee that a separation that works well by HPLC will also do so in SFC. However, sharper peaks, faster equilibration times and reduced analyses times have been reported [21], making it seemingly ideal for bioanalysis. In a study of a range of 44 different drug types, 70% separated on Chiralpak AD, 66% on Chiralcel OD, 50% on Chirobiotic T, and 40% on Chirobiotic V [24]. However, no true clinical applications were found at this time. It is possible that the introduction of new instrumentation, pcSFC may well find its place in many laboratories over the next few years.

137

Chiral bioanalysis

5.2 C H I R A L STATIONARY PHASES 5.2.1 Macromolecular or polymeric CSPs Cellulose and amylose are the most common of naturally occurring optically active polymers and as such offer great potential as chiral selectors. In their natural state, however, they are not practical as CSPs. Once the hydroxyl groups of the glucose moieties are derivatised, and they are immobilised on to a silica support, they become very valuable phases. As early as 1973, Hesse and Hagel demonstrated the efficacy of crystalline cellulose triacetate [25]. It was later shown that the chiral selectivity would not be lost by dissolving the polysaccharide in organic solvents [26] and coating onto silica gel. Since the enantioselectivity of these phases is strongly influenced by their tertiary structure [27], it is difficult to predict chiral recognition purely from the structure of the glucose monomer unit. Chiral recognition appears to be higher when the glucose hydroxyl group is derivatised with an aromatic group to form the benzoate, especially if the aromatic group has alkyl or other electron donating functional groups. This has led to a wide range of CSP's varying by the type of derivative (Fig. 5.1). The resulting increase in carbonyl oxygen electron density plays a significant role in enantioselectivity through increased participation in dipole-dipole interactions with a carbonyl group, or in hydrogen bonding with an amine or hydroxyl on the analyte. The effect is most

o~

Amylose derivatives

Cellulose derivatives

o

R,:

.,, CHIRALCEL OA

RI:

CHIRALCEL OO CHIRALCEL OD.R CHIRALCEL OD-RH

RI:

o

c.,

"':

N

CHIRALPAK AD CHIRALPAK AD-RH

R=:

o

CHIRAI.CEL OB

.~ Rt: ~ ,

O CHIRALCEL OJ

CHIRALCEL OJ-R ~ 0

CH~ CHIRALPAK AS

CHIRALPAKAS-RH

CHIRALCELOC CHIRAt.CELOK Fig. 5.1. Commerciallyavailable cellulosic and amylosic derivatised chiral stationary phases. References pp. 180-184

138

Chapter 5

marked for alkyl substituted phenylcarbamate derivatives. The position of the phenyl functional group is also critical in that it will affect the formation of an ordered tertiary structure. The resulting interactions- hydrogen bonding, 7r-Tr stacking and dipoledipole interactions, are more effective under normal phase conditions. The structures of the phenylcarbamate cellulosic phases have been determined [28] and shown to have a chiral helical groove with the polar carbamate residues lying along the main chain. If a racemate can enter this groove, chiral selectivity can be accomplished. Chiralcel OD, prepared from tris(3,5-dimethylphenylcarbamate) cellulose, is one of the most successful of the cellulosic range of CSPs. It is stable in mixtures of hexane and 2-propanol where it offers a wide range of chiral separations: out of a range of 510 compounds, 62% were resolved on this CSP [28]. Its amylose equivalent also exhibits high enantioselectivity for a wide range of structures. Of all of the amylose derivatives formed, it appears that the tris(3,5-dimethylphenylcarbamate), Chiralcel AD, is the most useful CSP and is complementary to the cellulosic form. Together with the tris(4methylbenzoate) cellulose CSP (OJ) and tris((5)-l-phenylethylcarbamate) (AS), these form the most effective columns out of the range. They have good capacity in normal phase, and are often used for preparative HPLC. These CSPs, however, are not covalently attached to the silica support, but are coated on to a wide pore silica that has first been silanised [29]. Consequently, some caution must be exercised with types of mobile phase solvents used. In normal phase separations, trifluoroacetate (TFA) or diethylamine (DEA) are often added to improve peak shape and it is general practice to dedicate a column for the use of acid/base modifiers because of their strong affinity for the stationary phase. Columns are stored in hexane after use. Since there are no ionic functional groups to utilise for enantiomeric interactions with charged analytes, reversed-phase conditions have been little used in the past. However, the addition of a counter ion to an aqueous solvent system can effectively neutralise an ionised analyte to enable its separation on polysaccharide phases under reversed-phase conditions. Specially prepared versions of the cellulosic and amylosic phases were introduced to enable the use of reversed-phase solvents. For Chiralcel OD-RH, OJ-R and Chiralpak AD-RH (R- designates a reversed-phase format, H - designates a higher performance (5 txm) version), this enables their easier use in bioanlysis. Chiralcel O D R can be operated in 100% acetonitrile, 100% methanol, 100% ethanol and 0-100% water, maintaining the pH between 2 to 7 for maximum column life. Enantioselectivity is controlled through the choice of buffer salt and its pH, maintaining analyte neutrality. A charged analyte may also have a high degree of solvation, limiting its direct interaction with this type of CSP [30], resulting in poor separation. For neutral compounds, a simple water-organic modifier mix can be used and no beneficial effect is generally seen when buffers, acids or bases are added [30]. In general, acetonitrile exhibits a lower retention time than methanol: there is no rule for choosing one over the other in optimising peak shape. For acidic analytes, ion suppression is effected at pHs between 2 and 4, while basic compounds often require the addition of a suitable counter ion, since high pH ion suppression is not possible for silica gel supports. The concentration and type of ion pair reagent are critical for cellulosic CSPs. Generally, the following series of counter ions decreases in retentivity [31 ] in the order:

Chiral bioanalysis

139

PF66> BF44> C104 > SCN > I > NO~ > H2PO4 > Br > C1 > AcOIn contrast, the amylose phases show no such effect and require basic pH rather than added counter ions to effect a separation under reversed-phase conditions. Although it is expected that hydrogen bonding plays a lesser role in reversed-phase separations, many molecules separate under both reversed- and normal phase conditions. Of a series of propanolol analogues [32], approximately 70% separated on the polysaccharide CSPs in both normal and reversed-phase modes, although only 10% showed any reversal of elution order in going from one to the other. Examples of the use of Chiralcel OD-R for sample enantiomeric drugs such as benzoin and indapamide has been reported [33]. Neutral molecules were separated with water-acetontrile mixtures, while acidic and basic compounds required the addition of 0.05% trifluoroacetic acid to the mobile phase. Resolutions varying from 1.78 to 7.51 were obtained, indicating their applicability to bioanalysis, although such assays were not included in this study.

5.2.2 Protein phases Bonded protein phases are employed widely for bioanalysis and are used in reversedphase mode only, using a combination of aqueous buffers and small amounts of organic solvents. Method development tends to be fast, but because of the large size of the chiral selector, protein-based CSPs have the lowest capacity of all CSPs available- a feature that fortunately does not affect their use in bioanalysis. Although all proteins are by their nature stereoselective, only a few have been discovered to have enantioselectivity over a wide range of solutes. Of these, ~ - a c i d glycoprotein (AGP), is the most broad ranging and is an extremely stable protein. It can tolerate organic solvents in high concentrations and can be used over a reasonably wide range of pH without the protein being denatured. The protein is immobilised by a crosslinking covalent attachment to surface modified silica. The carbohydrate portion of AGP contains 14 sialic acid residues, making the protein very acidic in character: when used over the pH range of 4 to 7, it retains a negative charge. Both amines and weak acids separate well on this phase. In most cases, the enantioselectivity for acidic compounds increases with decreasing pH as the charge on the AGP protein and the consequential repulsion decrease. For amines, a pH range of 6-7 is used, with organic modifiers such as 2-propanol or acetonitrile being used to control retention. Neutral analytes also show enantioselectivity that is affected strongly by changing pH. Hydrogen bonding, hydrophobic interactions and steric interactions with the tertiary structure of the AGP also play roles. It is thought that the amount and concentration of the organic modifier effect a separation by competing with the solute for the hydrophobic portions of the protein molecule, or with hydrogen bonding sites, depending on the nature of the organic solvent. It is also known that the concentration of the organic modifier will induce a temporary change in the tertiary structure of the protein, making some interacting sites more, or less, available. For some separations, the addition of a charged modifier, such as N,N-dimethyloctylamine or octanoic acid, is References pp. 180-184

140

Chapter 5

necessary for enantioselectivity. Once a protein column has been used with such charged modifiers, however, its structure and selectivity may be permanently altered and therefore a column needs to be dedicated for this purpose. Commercially known as Chiral-AGP, this CSP has been used extensively for bioanalysis, both alone and when coupled with a reversed-phase column. Although phosphate buffer is used routinely, quite often this can be replaced with ammonium acetate for interfacing with HPLC-MS detection. Of the other protein phases available, cellobiohydrolase (available as Chiral-CBH), human serum albumin (Chiral-HSA and Hypersil HSA), bovine serum albumin (Resolvosil BSA) and ovomucoid (Ultron ES-OVM) all have their place in bioanalysis. Cellobiohydrolase is a very stable enzyme that separates a variety of basic solutes when immobilised onto silica, while HSA is primarily used for the separation of weak and strong hydrophilic acids and zwitterionic solutes. However, cellobiohydrolase is easily poisoned by metals [34] and must therefore be used with a pre-column and with EDTA in the mobile phase to prevent deactivation. Ovomucoid, purified from chicken egg white, provides enantioselectivity for amines and carboxylic acids principally and is similar in behaviour to AGE A review of other, less commonly used, protein phases is available [35].

5.2.3 Cyclodextrins Cyclodextrins are cyclic oligomers of ct-(1,4)-linked glucose, produced by the enzymatic coupling of glucose units following the action of cyclodextrin glycosyltransferase on starch. They form into crystalline, homogeneous toroidal structures of different molecular sizes. Although many forms are known, it is the 6, 7 and 8 glucose unit structures (denoted et, [3 and ~/, respectively) that have proved useful for chiral HPLC. In aqueous or hydro-organic solutions, cyclodextrins are known to form hostguest complexes with a variety of molecules, ions and solvents [36]. The internal cavity of the cyclodextrin molecule is composed of glycoside oxygens and methylene hydrogens giving it an apolar character, whilst the surface is hydrophilic as a result of the 2,3 and 6-position hydroxyl groups. As a consequence, non-polar molecules (or the most hydrophobic portions of a molecule) tend to prefer to reside inside the cavity where they can bind through dipole-dipole interactions, hydrogen bonding or London dispersion forces with the outer edge. The orientation of the aromatic ring is selective due to the electron sharing of the aromatic methylene groups with those of the glucoside oxygens. The other prominent characteristic of the cyclodextrin phases is that the cyclodextrin ring provides a chiral environment of its own. [3-cyclodextrin, for example, has 35 stereogenic centres. Based on the evidence of x-ray crystallographic data, the [3 and ~/structures appear to have quite rigid and inflexible structures and are stable to a wide variety of aqueous and organic solvents. In contrast, the bonds of the oL-cyclodextrin appear to be weaker and capable of stretching, a fact that can be utilised by using high aqueous buffer conditions to encourage deformation of the cavity to enable inclusion of a molecule. Typically, substituted phenyl, naphthyl and biphenyl tings can be separated on [3-

Chiral bioanalysis

141

cyclodextrin, smaller molecules on [3, and molecules with three to five rings in their structure are best separated on ~/-cyclodextrin. Some of the primary (6-position) hydroxyl groups have been used to covalently bond the cyclodextrin to the surface of silica or other media. When this was first accomplished in the early 1980s [37], it was, along with the protein phases, among the first of the commercially available reversed-phase CSPs. In determining whether a separation is possible in the reversed-phase mode, the most hydrophobic portion of the molecule would preferentially have halogen, nitrate, sulphate, phosphate or hydroxyl (depending on the pH) functional groups for inclusion to dominate. Following inclusion of this part of the molecule, the aim is for the part of the molecule containing the chiral centre to interact with rim of the cavity (Fig. 5.2). Carbonyl, carboxyl and amine moieties prefer to hydrogen bond, so these will interact with the mouth of the cavity in preference to inclusion. Enantioselectivity is then optimised by varying buffer type and concentration, pH, organic modifier, flow rate and temperature. The secondary hydroxyl groups can be derivatised selectively, generally using those in position 2 first and then in position 3, changing the physical and chemical properties of the cyclodextrin and extending the separation capabilities. Fig. 5.3 shows those derivatives that have been produced commercially. One of the derivatives, the naphthylethyl carbamate, introduces ~vcomplexing capabilities through the aromatic ring and can therefore also be used very effectively with normal-phase solvents (when interactions are surface phenomena only), as well as in the reversed-phase mode where inclusion complexing predominates. Each of the six derivative types incorporates different possible mechanisms and are therefore capable of different ranges of applications. All of them have in common the fact that they can be used in both reversed-phase and in polar solvents with different mechanisms and consequential increased opportunities for enantioseparation: in the case of the naphthylethyl carbamate and 3,5-dimethyl phenylcarbamate, normal phase chromatography is additional to this list. Changing back and forth from each solvent system does not damage the column, provided buffers are washed out correctly and intermediate, miscible solvents are used in the case of normal phase. NIl 2

A H NH. "-,

H

o/

Fig. 5.2. Inclusioncomplexingschematic for the enantioselectivemechanismon cyclodextrinCSPs. References pp. 180-184

Chapter 5

142

i

Silica Gel

CYCLOBOND I 2000 SUFFIX

R=

DM (dknethylated)

OCH 3

AC**

COCH3

(acetylated)

OH

SP or RSP (hydroxypropyl ether)

I

~CH2CHCH 3

'S

CH3 ~CONHCH

CH3 ~CONH--~ CH3

RNorSN (naphthylethyt carbarrete)

DMP (3,5-dimethylphenyl c a rba fret e)

Fig. 5.3. Commercially available derivatised cyclodextrin chiral stationary phases.

The usefulness of the cyclodextrin range of CSPs has been further extended by using the polar organic mode (described in section 5.1.3), when the interaction of the solute with the CSP occurs with the mouth of the cavity (inclusion does not occur). For the cyclodextrin-based CSPs, a non-hydrogen bonding, polar-organic solvent (such as acetonitrile) is used as the main component of the mobile phase. The acetonitrile tends to occupy the cyclodextrin cavity resulting in an emphasis on hydrogen bonding interactions between the solute and the hydroxyl groups at the mouth of the cyclodextrin cavity, or with the appended carbamate, acetate, or hydroxypropyl functional groups. A hydrogen bonding solvent (such as methanol) can be added to decrease the retention of highly retained compounds. In addition, small amounts of glacial acetic acid and anhydrous triethylamine are added to control protonation of the analyte. It is the ratio

Chiral bioanalysis

143

of these that is used to effect the separation. An effective ratio of acid to base (when using the AcOH-TEA combination) has been found in the range 4:1 to 1:4, the typical average being 1.5 : 1.0 and the concentrations used vary over the range 0.002% to 2.5%. This ratio range changes to 2:1 to 5 : 1 when acetic acid and ammonia are used (for MS compatibility). Method development techniques generally start with a 9 5 : 5 : 0 . 3 : 0 . 2 acetonitrile-methanol-glacial acetic acid-triethylamine composition, and optimisation achieved by varying the acid/base ratio and/or organic composition. The ratio of acetonitrile to methanol can also be adjusted to optimise retention (affecting resolution also, but to a much lesser extent). This mobile phase has not only led to a wider range of applications, but has produced highly efficient, faster separations not previously possible on these phases. It also has other advantages. Compounds that exist as hydrochloride salts can be separated in this mobile phase but not generally in normal phase. When using achiral-chiral coupled column systems, fractions from the reversed-phase column can be switched directly into the cyclodextrin CSE The demands on the structure of the analyte for potential chiral separation are also much lower than with other CSPs in different mobile phases. Cyclodextrin CSPs were first introduced commercially under the Cyclobond TM nomenclature, Cyclobond I being the [3 cyclodextrin, Cyclobond II the ~/and Cyclobond III the ot version. The derivatives are named as in Fig. 5.3. Other available CSPs include Nucleodex| available in oL, [3 and ~/native cyclodextrin and in a permethylated version, where all of the secondary hydroxyl groups have been methylated, so that this phase operates via inclusion and proton acceptance. The ChiraDex| range offers a native [3 and ~/cyclodextrin phase and differs in the choice of more polar spacer arm used to attach the cyclodextrin to the silica media, giving it different characteristics to other cyclodextrin technologies.

5.2.4 Macrocyclic antibiotics Macrocyclic antibiotics are the newest class of chiral selectors but have already proved to be invaluable for a wide range of applications in chiral HPLC, SFC and CE, and especially in bioanalysis. There are several closely related oligophenolic glycopeptides (Fig. 5.4). Of these, Vancomycin, Teicoplanin and Ristocetin (denoted Chirobiotic V, T and R, respectively) have been used for a wide variety of pharmaceutical applications [38]. Vancomycin is an amphoteric glycopeptide containing 18 chiral centres surrounding three shallow inclusion pockets, bridged by five aromatic ring structures. Hydrogen donor and acceptor sites lie close to the ring structures, and two sugar moieties are positioned at the edge of the structure. Twenty chiral centres are present in teicoplanin, together with four inclusion pockets, seven aromatic rings, three sugar molecules and a multitude of hydrogen donor and acceptor sites. Uniquely, one of the sugars is a glucosamine that has a N-acyl hydrocarbon chain, making teicoplanin considerably more surface active than the other glycopeptides. Ristocetin was the latest to be developed and is the most complex. It has 38 chiral centres, four pockets and six sugar molecules. The peptide chain and additional ionisable groups give this structure the complexity and diversity to separate a wide range of analytes. All of the three References pp. 180-184

o.

c.~o. o.

3

CH 3

-H NH 2

OH

HO ~

.N

O

.o.

..

~,

CH~)H

~

~'~0 ~

o~~

o

~N

,~j~

CI

H

,~

O

O OH

,

CH~cH O

~o

HO OH

CHIROBIOTIC T (Teicoplanin)

~~"~--7~

O

E

NH 2

--.o

OH

Fig. 5.4. Commercially available macrocyclic glycopeptide chiral stationary phases.

HO

HO

HO ~

~:_.j

"

~o, ~ N " ~ ~

HO HO ~ _ ~ _

o~-""h'-u-,,--.L

"~~, n~ ,r - - O ~ o

,o,..< ~'o

NH~ ~

H 3C

CHIROBIOTIC V (Vancomycin)

HOHOH~

HO "~- ~

~

~'--"~-- OH

~'O

OH

o,. ~O..o

o

.o

~ ~ CH

c ~

~..~OH

OH

OH

-o ~ -"y-"3

o.

HO

O

H HO~,~ W

H

H

CI

B H N

O

H O

i

H

O

O C H H I

CI

O I

.-NH 2

CHIROBIOTIC TAG (Teicoplanin Aglycone)

OH

.o

C H I R O B I O T I C R (Ristocetin A)

L~

Chiral bioanalysis

145

glycopeptides are covalently bonded to silica using multiple linkages, making them extremely stable for both analytical and preparative LC applications. Each of the three CSPs operate in three different solvent modes - reversed-phase, polar organic and normal p h a s e - and enantioselectivity can be different in each mode. Strong interactions are available from w-~r complexation, hydrogen bonding and ionic interactions. Dipole-dipole interactions are less strong and inclusion or steric interactions weak. The mobile phase can therefore be chosen to enhance the most favourable interaction for enantioselectivity. By using intermediate solvents (such as ethanol when moving from the polar organic to normal phase), a single column can be switched between all three solvents systems without any detrimental effect. The three CSPs also appear to be complementary in that an increase in selectivity is often seen when moving from one column to another, using the same mobile phase. In the reversed-phase mode, typical solvent systems comprise a buffer with methanol, tetrahydrofuran or (occasionally) acetonitrile. Interestingly, enantioselectivity has been observed in both high aqueous and high organic content. Like the cyclodextrin columns, the macrocyclic antibiotics can also be used in the polar organic mode. In this case, 100% methanol is used in combination with acid and base. Again, it is the acid/base ratio that governs enantioselectivity: retention is controlled by changing the total concentration of the acid and base. Like the cyclodextrin columns, the choice of acid and base for the polar organic mode can be varied, such that volatile ones can be used for bioanalysis with HPLC-MS. Fast HPLC-MS bioanalysis using single salts such as ammonium acetate or trifluoroacetate have also been achieved (Table 5.3 is a summary of recent HPLC-MS clinical studies using the polar organic mode). A recent development in this type of CSP is the formation of the aglycone derivative of the teicoplanin bonded phase [39]. Having removed the sugar molecules, the resulting Chirobiotic TAG column displays especially enhanced c~ values for native, synthetic and some N-blocked amino acids. Interestingly, neutral molecules, such as oxazolidinones, hydantoins and benzodiazepines often separate well on the Chirobiotic TAG CSP in 100% of a single polar organic solvent.

5.2.5 ~-Complex CSPs Some of the earliest phases of development for chiral chromatography were based on the principle of 'charge transfer'. The chiral selector in these types contains either a wacid or w-base, aiming for interaction with a w-basic or w-acidic analyte, respectively. Other simultaneous interactions must also be present for enantioselectivity to occur and these include hydrogen bonding, steric repulsion and/or dipole-dipole interactions, with hydrogen bonding taking precedence [40]. Since these interactions are strongest in normal phase, solvents systems for enantioselectivity are often hexane-ispropanol mixtures. The length and geometry of the tether connecting the chiral selector to the silica surface is found to be critical [40]. These CSPs are the simplest to understand mechanistically, but can be highly specific to certain compound classes. A large array of ~r-acid or ~r-base type systems have been developed since the pioneering work of Pirkle [41] that started with the bonding of the 3,5-dinitrobenzoyl References pp. 180-184

CHIROBIOTIC T

CHIROBIOTIC V

CHIROBIOTIC V

CHIROBIOTIC V

CHIROBIOTIC T

CHIROBIOTIC T

Ritalinic acid [ 188]

Methylphenidate [ 171 ]

Fluoxetine [ 188]

Nicardepine [ 188]

Metoprolol [ 188]

Salbutamol [45]

* ATFA is ammonium trifluoroacetate.

Column

Compound

250 • 4.6 mm

150 • 4.6 mm

50 z 4.6 mm

50 • 4.6 mm

150 • 4.6 mm

50 • 4.6 mm

Column size

6.2, 6.8 min 1.1, 1.6 min 7.0, 7.8 min 3.2, 3.7 min (metabolite at 2.0)

1.0 ml/min 1.0 ml/min 1.2 ml/min 2.0 ml/min

100 MeOH/0.03% ATFA* 100 MeOH/0.03% ATFA* 100/0.5/0.1 MeOH/AcOH/NH4OH

100 MeOH/0.03% ATFA* 100 MeOH/0.03% ATFA*

5.0, 10.4 min 6.1, 7.2 min

1.2 ml/min 1.0 ml/min

100 MeOH/0.03% ATFA*

tR enantiomers

Flow rate

Mobile phase

TABLE 5.3 S U M M A R Y OF H P L C / M S C L I N I C A L STUDIES IN THE POLAR O R G A N I C M O D E

r

4~

Chiral bioanalysis

147

derivatives of some amino acids, noteably phenylglycine (denoted DNPBG). The only shortcoming of these types was the need to have an analyte structure with a complementary ~r-acid or w-base functionality, often making the CSP very specific for certain classes of structures. If it did not have such a functionality, derivatisation would be necessary. The results of the degree of enantioselectivity achieved with a wide range of phases developed were used to prepare 'reciprocal' phases and to model the interactions occurring [41]. In addition, chiral selectors have been designed for certain 'target' molecules of interest. For example, the [3-GEM-1 CSP (Regis) was originally designed for the separation of non-steroidal anti-inflammatory drugs (NSAIDs), but required that the carboxylic acid group be converted to the anilide derivative. Subsequent research led to the Whelk O-1 CSP (Regis) that resolved underivatised NSAIDs and also a wide range of other molecular types [40]. This was the first of a series of [3-complex CSPs that have been designed to contain both [3-acid and [3-base groups. Consequently, they are the most broad ranging of this type of chiral phase. More recently, two CSPs based on tartaric acid derivatives that have been polymerised and crosslinked have been developed (Kromosil CHI, Eka Nobel). The retention and selectivity are mainly dependant on the hydrogen bonding capabilities of the analyte and so are mostly used in normal phase solvents. They are advantageous for preparative chiral HPLC, and no reports to date have been noted for their use in bioanalysis. Table 5.4 shows the variety of [3-complex CSPs that are available commercially. Application areas for enantioselectivity have not been added for reasons of complexity but are available from individual manufacturers data. 5.3 A P P L I C A T I O N S OF C H I R A L H P L C IN BIOANALYSIS

5.3.1 [~-Adrenergic agonists The enantiomers of terbutaline in urine samples have been determined by utilising achiral-chiral column-switching techniques [42]. Following silica SPE extraction, endogenous compounds were separated from terbutaline and betaxolol (internal standard) on a silica HPLC column. A column-switching valve equipped with a silica pre-column enabled concentration of the terbutaline fraction before its normal phase chiral separation on a Sumichiral OA-4900 CSE Detection was by fluorescence (excitation 276 and emission at 306 nm) provided an assay sensitivity of 0.3 ng/ml with linearity from 1 to 250 ng/ml. A Chirobiotic T column was evaluated for the enantiomeric resolution of clenbuterol in the method development study looking at the side effects of the drug [43]. Liquidliquid extraction and an internal standard of practolol were used in the method. The CSP was used in the polar ionic mode with a mobile phase of methanol-acetonitrile-acetic acid-triethylamine (70:30:0.3:0.2 v/v/v/v). The enantiomers of both clenbuterol and the internal standard were separated in under 13 minutes: it was noted that there was no variation in peak retention times over the 1000 samples of the study. Clenbuterol has also been separated on the w-donor column, Chirex 3022 [44]. Normal phase conditions provided a resolution of 4.2 in 15 minutes, and a LOQ of 0.1 nmol by way of UV detection at 254 nm. References pp. 180-184

w-acceptor/w-donor rr-acceptor/w-donor w-acceptor/w-donor w-acceptor/w-donor w-acceptor w-acceptor w-acceptor w-donor w-acceptor w-acceptor

1-(3,5-dinitrobenzamido)-tetrahydrophenanthrene

1-(3,5-dinitrobenzamido)-tetrahydrophenanthrene

3,5-dintrobenzoyl-diphenylethylenediamine 3,5-dintrobenzoyl- 1,2-diaminocyclohexane 3-(3,5-dintrobenzamido)-4-phenyl-[3-1actam Dimethyl N-3,5-dintrobenzoyl-oL-amino-2,2-dimethyl-4-pentenylphosphonate N-3,5-dintrobenzoyl-3-amino-3-phenyl-2-( 1,1-dimethyl)-propanoate N-(1-naphthyl)leucine 3,5-Dinitrobenzoylphenylglycine

3,5-Dinitrobenzoylleucine 1R,3R-chrysanthemate-R-phenylglycine 3,5-Dinitrobenzoyl-R-naphthylglycine 3,5-Dinitroanaline-S-valine 3,5-Dinitroanaline-S-tert-leucine 3,5-Dinitroanaline-R-phenylglycine S- 1-(ot-naphthyl)ethylamine-S-valine R- 1-(o~-naphthyl)ethylamine-S-valine S- 1-(a-naphthyl)ethylamine-S-proline R- 1-(ot-naphthyl)ethylamine-S-proline S- 1-(ot-naphthyl)ethylamine-S-tert-leucine R- 1-(oL-naphthyl)ethylamine-S-tert-leucine S- 1-(ot-naphthyl)ethylamine-S-indoline-2-carboxylic acid R- 1-(ot-naphthyl)ethylamine-S-indoline-2-carboxylic acid O,O'-bis(3,5-dimethylbenzoyl)-N,N'-diallyl-L-tartardiamide O,O'-bis(4-tert-butylbenzoyl)-N,N'-diallyl-L-tartardiamide w-acceptor w-acceptor w-acceptor w-acceptor w-donor w-donor w-donor w-donor w-donor w-donor w-donor w-donor w-donor w-donor

Type

Bonding chemistry

TABLE 5.4 RANGE OF w-COMPLEX CSPs AVAILABLE

chemistry for > stability to hydrolysis) (Regis) ULMO (Regis) DACH-DNB (Regis) Pirkle 1-J (Regis) R- and S-a-Burke 2 (Regis) R,R- and S,S-[3-GEM 1 (Regis) D- and L-Naphthylleucine (Regis) D- and L-Phenylglycine (Regis) Chirex 3001, Sumichiral OA-2000 D- and L-Leucine (Regis) Sumichiral OA-2200 Chirex 3005 Chirex 3010, Sumichiral OA-3100 Chirex 3011, Sumichiral OA-3200 Chirex 3012, Sumichiral OA-3300 Chirex 3014, Sumichiral OA-4000 Sumichiral OA-4100 Chirex 3017, Sumichiral OA-4400 Chirex 3018, Sumichiral OA-4500 Chirex 3019, Sumichiral OA-4600 Chirex 3020, Sumichiral OA-4700 Sumichiral OA-4800 Chirex 3022, Sumichiral OA-4900 Kromasil CHI-DMB Kromasil CHI-TBB

R,R- and S,S-Whelk-01 (Regis) R,R- and S,S-Whelk-02 (uses trifunctional bonding

Commercial name

t~

Chiral bioanalysis

149

Salbutamol (albuterol) is a safe and effective short-acting bronchodilator, considered one of the best of its type. It is administered as the racemate, even though it is wellknown that its clinical activity is mainly associated with the (-)-R isomer. Whilst many methods have been developed, many suffer from lack of sensitivity and so can be suitable only for defining the pharmacokinetics of salbutamol after oral and intravenous dosing [45]. These authors described a method that was sufficiently sensitive for inhaled doses using H P L C - M S - M S and on-line robotic 96-well plate extraction. A total assay time of less than 5 minutes was achieved for the simultaneous separation of the racemic parent drug and both enantiomers of the 4-O-sulphate metabolite in plasma and urine (Fig. 5.5). A Chirobiotic T column in an HPLC-MS compatible polar ionic mode solvent of methanol, acetic acid and ammonia ( 1 0 0 0 : 5 : 1 ) was used for the assay, ^)

;-$allmmmol

250000

m/z 24.0 -,+ 1411

2OOOOO 15OOOO

I00000 R .

.

.

.

.

.

.

.

. . . . . . .

-

.

.

.

.

.

B)

.

. . . . .

~:

;-

:

.

+

,]-SaPmm~ol ~243

--~15|

l .k

.

_

,

.

i

_

.

_

,

_

li|

. . . . .

R

C) m/z 342 --+244

1.0

2.0

3.0 T~m

4.0

5.0

(mi~)

Fig. 5.5. HPLC-MS/MS SRM chromatograms of parent drug and its O-sulphate metabolite on Chirobiotic T column resulting from a 50 ~1 injection of extracted urine following inhalation of racemic Salbutamol [45]. References pp. 180-184

150

Chapter 5

splitting 200 I~1 of the column effluent to the ionspray interface. This avoided the difficulty of using normal phase solvents with APCI. A sensitivity of 25 pg/ml was achieved and the method was applied to some 4000 samples. A similar method was developed by Fried et al. using fluorescence detection at 230 excitation and 310 nm emission, enabled concentrations down to 125 pg/ml to be detected from a 1 ml plasma sample [46]. In this case, a mobile phase of methanol, acetonitrile, acetic acid and diethylamine provided a 13-minute assay time. During the cross-validation study, an interfering endogenous substance was noted in both human and canine plasma. To accommodate this, the separation was further optimised by reducing the amount of methanol by 10%. This increased resolution from 1.5 to 1.8 although the enantioselectivity and k' were largely unchanged. The method was applied to samples from a single-dose inhalation of racemic salbutamol and also to a canine inhalation study of the single R-(-)-enantiomer. No chiral inversion of the R-(-)-isomer was noted in the dog. A further method for albuterol has been developed using a Chiral-AGP column [47] but the authors reported only an 80% resolution. Salbutamol had been previously assayed with clenbuterol in plasma and urine using a normal phase method on the w-donor, Chirex 3022 [48]. This separation, when combined with fluorescence detection, provided an LOQ 10-fold less than the current HPLC-MS methods for the parent drug enantiomers. Chiral-AGP was the method of choice for the determination in urine of the R,R and S,S isomers of formoterol, a long acting [3-adrenoceptor agonist [49]. Complete pharmacokinetic data had previously been lacking due to the absence of a sensitive chiral method. Electrochemical detection was used with this method for single inhalation doses in human studies, facilitated by using a mobile phase of 2-propanol and phosphate buffer with added 1 mM KC1 and EDTA (Fig. 5.6). The limits of detection for the R,R and S,S isomers were 60 and 75 pmol/L respectively.

5.3.2 [3-Adrenergic blockers An achiral-chiral column-switching technique was used for the quantification of metoprolol in human urine [50]. A silica HPLC column was first used to separate metoprolol and the internal standard from interfering urinary compounds with the analytes then switched on to the chiral column via a silica trap column. Chiralcel OD in normal phase mode and fluorescence detection provided a limit of quantification of 25 ng/ml for each enantiomer. Direct and indirect methods for the determination of metoprolol were compared for the R(+) and S(-) enantiomers in human plasma [51]. The direct method used a Chiralpak AD (using solid-phase sample preparation) or a Chiralcel OD-H column, while the indirect method used derivatisation with S-(-)-menthyl chloroformate followed by separation of the resulting diastereomers on a reversed-phase C8 column. It was found that the direct, Chiralpak AD, method provided the highest sensitivity, although it did not measure the o~-hydroxy metabolites. A further study [52] did achieve this, enabling the separation of the parent drug and the four enantiomers of the othydroxy metabolites in 13 minutes (Fig. 5.7). A Chirobiotic T column in the polar ionic

Chiral bioanalysis

151

RR

$s

IS

r

Inject

Fig. 5.6. Separation of the RR and SS enantiomers of formoterol and its diastereomer as internal standard on Chiral-AGP using electrochemical detection ( + 0.63V vs. Ag/AgC1) [49].

mode was used. Unusually, the mobile phase also contained dichloromethane (to which this CSP is stable), added to assist in the separation of the closely eluting metabolite isomers. The analytes were detection by fluorescence and the lower limit of quantification was repoted as 0.5 ng/ml for metoprolol and 1.0 ng/ml for oL-hydroxy metoprolol. A fully automated assay for oxprenolol in human plasma was developed that utilised on-line dialysis (using an ASTED system) through a cellulose acetate membrane for sample preparation [53]. Since the [3-adrenergic blocking activity of this compound is mainly dependant on the S-(-) enantiomer, it was considered important to measure the plasma concentrations of the individual enantiomers. Dialysis was followed by a precolumn clean-up and trace enrichment: injection onto a Chiralcel OD-R column in a mobile phase of acetonitrile and phosphate buffer containing sodium perchlorate provided an assay time of about 12 minutes. The influence of perchlorate concentration on the separation was studied and it was found that increased concentrations, up to a limit of 0.45 M, gave rise to an improvement in enantioselectivity and resolution. It was postulated that this was due to an ion-paring effect. References pp. 180-184

Chapter 5

152 o

9

~ w 6"I =

.o

u.~ s

....

"~.~

......

;:F. ,~_~ 9

~

o ~0

-9~

0

m m 0"s i ~

_

-

-

"

_

~

-

-

0

9

J ,

.

_J

_

0

~.=

0 0

>.,.=.

~16"I

1

,

.

ii

ii

-

. . . . i .i i . .

-

~

-

E r

~J

0"~

E~

Chiral bioanalysis

153

The enantiomers of atenolol were separated using a Chiral-CBH column equipped with a silica pre-column, added to compress the peaks and improve sensitivity [54]. The mobile phase comprised 2-propanol, 50 ~M cellobiose and 50 ~M EDTA and the method was validated for use with CSE A sensitive assay for all four enantiomers of labetolol was developed using a Chirex 3022 ~r-donor CSP in normal phase with fluorescence detection [55]. The retention times were 33, 37, 43 and 51 minutes and each was linear over a concentration range of 2.5 to 125 ng/ml. Differing ratios of the enantiomers were observed in plasma compared to urine: a ratio of 41 : 25 : 25 : 9 was found for the S,S (inactive), S,R, R,S and R,R (active) enantiomers, respectively in plasma ( 5 8 : 1 4 : 4 : 2 4 for urine), even though the active R,R was only 2% of the total dosage. Pindolol was determined in human serum using a Chiralcel OD-R CSP in a mobile phase of acetonitrile and perchlorate [56]. Fluorescence detection at 310 nm (excitation 270 nm) enabled detection limits of 1.2 and 21 ng/ml for the R and S enantiomers in serum, and 4.3 and 76 ng/ml for the same in urine. The widely used blocker, propanolol, has been determined in plasma and urine using racemic alprenolol as internal standard [57]. It was confirmed that the concentrations of the enantiomers were dose dependant and that the concentrations of the (S)-isomer were always higher than those of the (R)-form. The method used Cyclobond I 2000 in the polar organic mode.

5.3.3 Alcohol deterrent drugs The knowledge that disulfiram binds selectively to the single free cysteine on HSA was used to develop a method for the analysis of this drug in plasma on a Chiral-HSA CSE Low concentrations of disulfiram as a mobile phase modifier improved the performance of this CSP for disulfiram and a range of co-administered drugs, and was useful for studying their interactions [58].

5.3.4 Amino acids The concentrations of amino acids in blood serum are known to be influenced by hepatic diseases. Whilst extensive studies have previously looked at the levels of L-amino acids, a recent study [59] looked at the levels of D-serine as a measure of hepatic function, since nutritional serine is thought to be metabolised by D-amino acid oxidase, especially in the liver and kidney. Total serine was measured as its benzofurazan derivative using achiral, ODS chromatography, collecting the NBD-serine fraction and re-chromatographing on a Sumichiral OA-4700, a w-donor Pirkle-type CSE The method was used to measure serum concentrations of D-serine in liver transplant patients who had been suffering from biliary atresia: it was found that D-serine levels in blood serum gradually decreased with the recovery of hepatic function. In the past, many amino acid determinations in biological samples have been carried out utilising derivatisation techniques, principally to provide enhanced detection. One References pp. 180-184

Chapter 5

154

such study [60], used the fluorescent chiral tagging reagent, R(-)- or S(+)-4(3-isothiocyanatopyrrolidin- 1-yl)-7-(N,N-dimethylaminosulphonyl)-2,1,3-benzoxadiazole, to study the amounts of both hydrophilic and hydrophobic D-amino acids in urine. The resulting diastereomeric compounds could be separated on a C18 column and detected by fluorescence at 5 5 0 n m (excitation at 460nm) with identification by ESI-MS. The separation of amino acids for bioanalysis without derivatisation on a Chirobiotic T column relies on a free COOH on the amino acid which interacts ionically with a primary amine on the CSE A comprehensive study of methods suitable for a wide range of amino acids, including enantiomers, positional isomers and proteinogenic chiral amino acids such as DOPA and ot-ABA was carried out [61]. An isocratic method in combination with the high specificity of H P L C - M S - M S was used to increase the number of underivatised amino acids that could be separated simultaneously (Fig. 5.8). Post-column addition of acetic acid was used to increase the MS response, enabling sensitivities that were 10 to 100-fold greater than achieved by fluorescence detection. Although the method was developed for the analysis of meteorite samples, it was

I u

_

!

Fig. 5.8. Simultaneous isocratic analysis of 15 underivatised amino acids by HPLC-ICP/MS/MS on Chirobiotic T [61].

Chiral bioanalysis

155

reported that the developed method would be suitable for therapeutic amino acids and peptides currently under investigation for cancer and schizophrenia.

5.3.5 Analgesic drugs (narcotics) Methadone is a central acting analgesic with a high affinity for Ix-opiod receptors that has been used to treat opiate dependence and also cancer pain. Although used therapeutically as the racemic mixture, the R-(-) isomer is about 25-50 times more potent an analgesic than its S-( + ) counterpart, due to both pharmacodynamic effects and to enenatioselective pharmacokinetics. A method was reported [62] that enabled the simultaneous assay of the enantiomers of the parent compound, its major metabolite, 2-ethylidine-l,5-dimethyl-3,3-diphenylpyrrolidine (EDDP) and an internal standard (Estazolam) for the first time on Cyclobond I 2000 RSE Previous methods were available on Chiral AGP and on Chiralcel OJ but they were reported to give overlap for the metabolites. More importantly, the method reported no interference from most common drugs of abuse or commonly administered compounds such as the benzodiazepines and caffeine. The assay utilised a reversed-phase mobile phase of acetonitrile/0.08% triethylamine acetate buffer (pH 4.5). Linearity for both serum and urine over a 0.05 to 2.0 ixg/ml concentration range was observed, with an intra-day and inter-day precision of

1

83,840

259/213

cr

SIL (R)-

0.00

2.00

4.s Time (min)

SlL (S)-kt 6.00

Fig. 5.12. HPLC/MS/MS chromatographic profiles for ketoprofen and stable isotope labelled (SIL) ketoprofen on Chirex 3005. A: Human plasma spiked with 50 ng/ml SIL ketoprofen; B: Calibration standard containing 0.1 ng/ml ketoprofen and 50 ng/ml SIL ketoprofen [ 131 ].

A column switching technique was used for a study of the potential for chiral inversion of a morpholine derivative, 2-(S)-[3,5-bis(trifluoromethyl)benzyloxyl]4-[3-(5-oxo-lH,4H-1,2,4-triazolo)methyl]-3-(S)-phenylmorpholine, having 2 chiral centres [137]. The achiral separation was timed to switch a segment from 26 to 38 minutes directly on to the chiral column, a Chiralcel OD-H and a mobile phase of nhexane-methyl-t-butyl ether-methanol was used to separate the 4 enantiomers over a 50-minute period. UV at 210 nm was used to monitor the separation. The study showed that there was no in vivo inversion at either of the chiral centres in dogs. Chiralpak AD was the CSP of choice for the resolution of sulindac, using hexaneethanol containing 0.05% TFA [138]. Urinary samples were injected either directly or after dilution onto an ODS column and the endogenous component-free eluate collected for analysis by the chiral method. The method was used in a study of urinary excretion of free and conjugated sulindac in human volunteers.

Chiral bioanalysis

169

5.3.21 Antiischaemic drugs Column switching techniques were used for the clean up of gerbil plasma samples for subsequent analysis of a potential antiischaemic agent, dichlorophenyl ketoamino acid, on a CrownPak CR ( + ) CSP [139]. Following C18 SPE extraction, the samples were loaded on to a cyano HPLC column. The fraction eluting between 13 and 40 minutes was switched onto the CSP and the enantiomers eluted with a mobile phase of perchlorate and methanol. Detection at 257 nm UV gave an LOQ of 20 ng/ml.

5.3.22 Antineoplastics The four enantiomers of a potent inhibitor of human leukocyte elastase were determined in plasma [140 After liquid-liquid extraction, a normal phase separation of the enantiomers was carried out on Chiralcel OD-H. Fluorescence detection provided an LOQ of 50 ng/ml, compared with 10 ng/ml for an achiral method carried out on the same samples. Rogletimide has effective antitumor activity in cases of breast carcinoma, and exhibits fewer side effects than its analogous predecessor, aminoglutethamide. A sensitive and stereospecific assay for its enantiomers was developed and validated using a Chiralcel OJ-R column, using aminoglutethamide as internal standard and UV detection at 257 n m [ 141]. When optimising the mobile phase composition 80:20 (v/v) sodium perchlorate-acetonitrile was found to give the best peak shape and retention (6-8 minutes) for the enantiomers. The LOQ was 100 ng/ml for each enantiomer, identical to a corresponding method developed using CE [ 142]. Cyclophosphamide in serum was separated on a Chiral-AGP equipped with a C1 guard column: detection at 195 nm gave an LOQ of 1.25 mg/ml [143].

5.3.23 Antiparkinsonian agents The increasing use of LC/MS using cyclodextrin phases is exemplified by the separation of the anti-parkinsonian drug, trihexylphenidyl (THP) [144] (Fig. 5.13). THP belongs to a group of synthetic anticholinergic drugs that compete with the neurotransmitter acetylcholine at its receptor sites. It does, however, show a range of side effects. While achiral analysis employed GC methods, a chiral HPLC method was required. HPLCESI-MS detection was used because of poor UV absorption. Because this molecule has an aromatic group and two hydrogen bonding groups (one on the stereogenic centre), it was considered suitable for the polar organic mode. A native cyclodextrin column, Cyclobond 1 2000, was used in a 2.0 mm internal diameter format. The optimal mobile phase was found to be 9 5 : 1 : 0 . 5 : 0 . 3 (v/v/v/v) acetonitrile-methanol-acetic acidtriethylamine. It was found that the ratio of acid to base had a dramatic effect on both resolution and retention and this was used in the optimisation process. Calibration curves showed good linearity over the range of 1.3 to 132.3 ng/ml, and also at THP concentrations 48 times higher than the maximum therapeutic plasma levels of 55 ng/ ml. References pp. 180-184

~- - ' - " ' ~ ' ~ " " ' - " ' " ' "

2.50

....

9 ....

, .r ~ . . . . .

" J';:"" i " ""

~ i:;"

";

5: ....

I:':-':;"

I " " "";"U"

5.00

7.50

10,00

12.50

w---'-'i;';"~'i:';"i:v~"~"i"';"'i-;":v,"'"":':"i":'v';",;"'~'"'~

~..

15.00

" ' :'~ : "

....

r ....

"; ; ' i

20,00

22.50

:~-,i---

qD

' ; ; ; i ;:" ' ":-i-;:"

|;---~---~i

;-;:~r ; ' i " ; ' " :

17.50

;:; i'W;:-:':i

'"

,

25.00

~v;-;;T"tll~

'W . . . .

Fig. 5.13. HPLC-ESI-MS chromatograms ofTHP enantiomers (lower trace) and diphenidol (internal standard) (upper trace) in human serum on Cyclobond 1 2000 in the polar organic mode [ 144].

0.00

O'r--',,--,-w

%

100" 5.16e3

0

100" 8.34e3

c~

"-..3 0

Chiral bioanalysis

171

5.3.24 Antipsychotic agents Remoxipride is a benzamide that has demonstrated stereoselective affinity for central dopamine D2-receptors from the S-(-) configuration. The first direct chiral HPLC method for the enantiomers in human plasma was reported [145] and used to investigate the systemic inversion of administered S-(-) remoxipride in healthy human volunteers. A Chiralcel OD-R CSP was used in a mobile phase of acetonitrile with potassium hexafluorophosphate at pH 3.5 and detection made at 214 nm. Plasma samples underwent extensive liquid-liquid extraction, enabling a sensitivity of 20 ng/ml, with a precision of 10.6 and 10.4% for the R-( + )- and S-(-)-enantiomers, respectively. It was noted that despite the fact that the column required conditioning with unextracted standards, after storage in methanol for two years, the column exhibited the same retention times, peak response and resolution in a subsequent study. The R-( + ) isomer is inactive, and showed higher toxicity than its counterpart but was not found in any of the plasma samples in this study. The four enantiomers of amisulpride, a benzamide derivative, were separated on a Chiralpak AS in hexane/ethanol with added diethylamine [ 146] and the method applied to pharmacokinetic studies in human plasma.

5.3.25 Antivirals A series of synthetic homo-isoflavonoids, chloro-substituted rac-3-benzylchroman4-ones, were studied for possible enantiomeric anti-rhinovirus activity [ 147]. Two CSPs were compared for their analysis. The paper reports the separation of the enantiomers of 3-(4-chlorobenzyl)chroman-2-one and of its 6-chloro derivate on an R,R Whelk-01 column. A further chlorinated derivative that was unresolved on the Whelk column, was partially separated on a Chiralcel OD-H CSP.

5.3.26 Antiulcerative drugs The pharmacokinetics of pantoprazole, a long-acting proton pump inhibitor was investigated in rats [148]. Gradient elution using a mobile phase of sodium perchlorate and acetonitrile was used to separate the enantiomers on a Chiralcel OJ-R CSE Detection at 290 nm was sufficient to show that serum concentrations of both enantiomers declined rapidly with time. Chiral inversion was noted for intravenous feeding of the racemate but not for feeding of the single enantiomer in rats. In an earlier method, gradient elution was used to separate the enantiomers of pantoprazole sodium sesquihydrate on a Chiralcel OJ-R: a gradient of acetonitrile/perchlorate was used, increasing to 100% acetonitrile over 15 minutes at 40~ [149]. The column was backflushed with the starting mobile phase at the end of the cycle. Chemometric techniques were used for the optimisation of the chiral separation of omeprazole and its metabolites on Chiral-AGE providing a separation within 15 minutes [150] Column temperature and acetonitrile concentration were found to be the most important variables. Detection was by UV at 302 nm. References pp. 180-184

172

Chapter 5

Good retention, resolution and stability were reported over 400 runs of a human serum assay of lansoprazole using Chiral-AGP as the CSP [151]. A mobile phase of 2-propanol and phosphate buffer at pH 7 combined with UV detection at 283 nm were used.

5.3.27 Anxiolytics A combination of UV and circular dichroism detectors was used in a pharmacokinetic study of lorazepam in plasma [152]. The separation of the enantiomers was compared for both a Chiralcel OD column in normal phase and on a Shodex [3-cyclodextrin column under reversed-phase conditions. UV detection provided a sensitivity limit of 3 ng/ml and a circular dichroism (CD) detector was used for identification. Reversal of elution order was observed between the two methods (R- eluting before S- for the cyclodextrin method). In a previous study [ 153], a method developed on ChiralPak AS using hexane-2-propanol-ethanol was developed a tested with spiked plasma. Three analogues of the 5-HT receptor agonist, 8-hydroxy-(di-n-propylamino)tetralin, were synthesised and screened for pharmacological activity as potential orally-active CNS drugs [154]. A method was developed for all the analogues on Chiral-AGP and factorial design techniques used to determine the importance on selectivity of all variables. This was used to support an in vitro metabolism and pharmacokinetic study, in which the analyst replaced the guard column and reconditioned the analytical column every 40-50 samples. Interestingly, sensitivity was found to double by the application of a short ammonium acetate gradient instead of the original isocratic phosphate buffer method. A chiral method for diazepam, and its chiral and achiral metabolites in spiked plasma was developed using reversed-phase conditions and a Chiralcel OD-R CSP [ 155]. It was reported to give good recoveries and detection limits (25 ng/ml at 210 nm).

5.3.28 Biochemical markers For the determination of D- and L-lactate in rat serum, it was necessary to derivatise with 4-(N,N-dimethylaminosulphonyl)-7-piperazino-2,1,3-benzoxadiazole in the presence of triphenylphosphine and 2.2'-dipyridyl disulphide in acetonitrile for 40 minutes, followed by acetylation of the hydroxyl group of the lactate. The resulting derivatives were separated on a Chiralcel OD-RH column in methanol and detected by fluorescence at 560 nm (excitation at 450 nm) [156]. The levels of D- and L-lactic acid were determined from calf serum by using a ligand exchange technique [157]. Ultrafiltrates of the sample were injected onto a short (50 x 4.6 mm) ChiralPak MA column using acetonitrile and 2 mM CuSO4 as mobile phase. The acid was detected at 236 nm. 2-Hydroxyglutaric acid is a chiral polar aliphatic dicarboxylic acid excreted in very small amounts in mammalian urine. D-2-hydroxyglutaric aciduria and L-2-hydroxyglutaric aciduria are two distinct inherited metabolic diseases: accurate diagnosis of the

Chiral bioanalysis

173

disease relies on the determination of the configuration of the enantiomer excreted in excess in urine. The two enantiomers have been separated using a Chirobiotic R CSP interfaced with an ESI source, with detection in negative ion mode MS/MS [ 158]. Using a mobile phase of triethylamine acetate (pH 7.0) and methanol (9:1), almost baseline separation of the two enantiomers was achieved in less than 6 minutes. Glyceric acid is another marker whose increased excretion is an indicator for two other metabolic diseases, D- and L-glyceric aciduria. The separation of the enantiomers was achieved using a narrow bore Chirobiotic R CSP in reversed-phase mode interfaced to MS/MS with an ESI source in the negative ion mode [159]. A flow rate of 0.3 ml/min enabled direct connection to the MS without a split. The isomers were separated at 3.6 and 4.5 minutes and the method used for confirmation of the disease in three patients. Pipecolic acid is an important biochemical marker for the diagnosis of peroxisomal disorders. In this case, a Chirobiotic T CSP interfaced directly to HPLC-MS/MS in a simple mobile phase of methanol/water, the L-enantiomer eluting before the D- (11.7 minutes) at 7 minutes [160]. The method was validated using an internal standard of phenylalanine and showed a linear range of 0.5 to 80 Ixmol~.

5.3.29 Calcium channel blockers 4-Aryl-1,4-dihydropyridines of the nifedipine type are the most studied class (Fig. 5.14) of organic calcium channel modulators and have been used extensively for the treatment of cardiovascular diseases such as hypertension, cardiac arrythmias or angina. A detailed structure-activity study [161] established that calcium channel modulation is dependant on the absolute configuration C4 (R- or S-configuration), where the orientation of the C4-aryl group acts as a 'molecular switch' between antagonist and agonist activity. Due to these opposing pharmacological effects, the development of single enantiomers is critical. A study [162] was carried out of 29 different dihydropyridine structures having the important structural features of biologically active analogues. In all but one case, baseline resolution was achieved (R > 1.26) on at least one of the eight CSPs selected. A strong influence of the substitution on the C4 group was noted for Chiralcel OD-H in normal phase and this CSP separated 82% of the dihydropyridines. The cyclodextrin types, as exemplified by ChiraDex, separated 76%, and was dependant on the bulkiness of the aromatic moiety. The most selective for this type of heterocycle was found to be Chirobiotic T and Chirobiotic V with 92% of the structures separated, mostly at low k' values. Interestingly, the two that did not separate were conformationally restricted dihydropyridines that readily separated on the other columns. The other commercial column tested was the w-acidic Whelk-01 and this separated 42% of the racemates. One of the most often used calcium channel blocking agents, nicardepine, also has potent oral vasodilating activity. A sensitive UV assay was developed [ 163] to determine the ( + ) enantiomer, which was found to be approximately three times more potent that the (-) isomer. Sensitivity in previous methods had been hampered by broad peaks. This method utilised a w-donor Sumichiral OA-4500 column in the normal phase solvent hexane-l,2-dichloroethane-ethanol-trifluroacetic acid. A rapid combined solid phase References pp. 180-184

~g

No2

N~

(-3

[ ~ u No2

nisoldipine

H

,3c" -,.- "c.3

~.HN02

CH3OOC,~COOCH2CH ~ 2e5]

'~c"Ce

le51 ,

=

,

--

ii

I

I

5.0

6.0

i

Time, min

Fig. 8.5. LC-ESI-MS-MS of buprenorphine glucuronide (1), norbuprenorphine (2), norbuprenorphine-d3, buprenorphine (4), and buprenorphine-d3 (5) in plasma. From ref. [25] with permission of Elsevier Science.

References pp. 267-270

248

Chapter 8

hydroxybenzphetamine, hydroxybenzylamphetamine, methamphetamine and amphetamine were isolated from urine with SPE, separated using an alkaline mobile phase, and detected with ESI-MS (SIM). The LODs ranged from 0.3 to 10 p~g/1 urine [32]. Dimethylamphetamine and its metabolites dimethylamine-N-oxide, methamphetamine and amphetamine were determined in urine by LC-MS (SIM) after isolation with SPE with LODs of 5-50 p~g/1 urine [33]. Kataoka et al. [34] applied SPME to the isolation of amphetamine, methamphetamine, MDMA, MDEA and MDA from urine. The drugs were desorbed by mobile phase flow and detected with ESI-MS (SIM), with a LOD below 2 ~g/1 urine. MDMA, MDEA and MDA were determined in whole blood, urine and vitreous humor of rabbits using HPLC with fluorometric and MS/MS detection and a very good correlation between both methods was found [35]. Wood et al. [36] applied LC-ESI-MS-MS for determination of amphetamines (amphetamine, methamphetamine, MDMA, MDA, MDEA, ephedrine) in saliva. The collected saliva samples (50 p~l) were diluted with 200 p~l methanolic of a solution of a mixture of internal standards (deuterated analogues) and centrifuged. The supernatant was then subjected to HPLC using an ACN-ammonium acetate eluent. The drugs were detected using transitions of protonated quasi-molecular ions in optimized conditions. Selegiline, an inhibitor of monoamine oxidase-B used in the treatment of Parkinson's disease, is metabolised to methamphetamine and amphetamine. Katagi et al. [7] determined this drug as well as the specific metabolites selegiline-N-oxide, methamphetamine and amphetamine in human urine. All four compounds were isolated from urine using SPE on C18 cartridges, separated on a cation exchange column and subjected to ESI-MS detection in full scan acquisition mode. In urine samples collected from selegiline patients, only metabolites were detected. The method may serve for differentiation between selegiline and methamphetamine intake.

8.3.1.3 Cocaine

The advent of LC-MS brought substantial progress in the determination of cocaine metabolites, which may be detected without any derivatization. Bogusz et al. used LCAPCI-MS for examination of various drugs of abuse, among them cocaine, BZE and EME in biological fluids [31] with LODs ranging from 0.2 to 0.5 p~g/1. In a study by Jeanville et al. LC-MS was applied to automated urinalysis [37] with centrifuged urine samples injected into an LC-MS-MS system equipped with an on-line extraction unit. The total analysis time was less than 4 min. with LODs for EME, BZ and cocaine of 0.5, 2.0 and 0.5 p~g/1,respectively. Skopp et al. carried out a study on the stability of cocaine and metabolites at different temperatures using LC-ESI-MS-MS [38,39]. For each compound one product ion originating from the protonated quasi-molecular ion was monitored. Only ecgonine appeared to be stable at room temperature. The conversion of cocaine to the final metabolite ecgonine was stoichiometric. A validated LC-APCIMS-MS procedure for cocaine and benzoylecgonine in human plasma was published by Lin et al. [40]. The transitions m/z 304---~182 and 2 9 0 ~ 168 were monitored for cocaine and benzoylecgonine, respectively with an LOQ of 2.5 ~g/1 for both compounds.

Use of liquid chromatography-mass spectrometry in acute human toxicology

249

8.3.1.4 Cannabinoids

In the study of Mireault [41], blood or urine samples were extracted by SPE cartridge and analysed on a C8 column using a methanol-ammonium acetate mobile phase. An ion-trap instrument equipped with APCI source was used in positive ion MS-MS mode. A detection limit of 1 ~g/1 was achieved for THC, 11-OH-THC and THC-COOH. In a second study the same group [42] used an APCI- triple quadrupole mass spectrometer. In this investigation an LOQ of 0.25 ~g/1 was reported for all three compounds (Fig. 8.6). In biological extracts, the limit of detection was 10-40 times lower for the quadrupole instrument than the ion trap. Breindahl and Andreasen have applied ESI-LC-MS to the determination of T H C COOH in urine [43]. Urine was subjected to basic hydrolysis and solid phase extraction. THC-COOH and its deuterated analogue were analyzed with HPLC-ESI-MS, using C8 reversed-phase column, gradient elution with acetonitrile-formic acid and SIM detection (positive ions). In-source collision induced dissociation was applied and protonated quasi-molecular ions as well as two fragment ions were monitored. An LOD of 15 ~g/1 was obtained. Tai and Welch [44] also determined THC-COOH in urine with HPLC-ESI-MS (negative ions). The drug was extracted from urine with SPE and subjected to isocratic separation on an ODS column using a methanol-ammonium acetate mobile phase. Only the deprotonated quasi-molecular ions of the drug and its deuterated analog were monitored with a reported LOD of 0.5 ~g/1 urine. Weinmann et al. have used L C - M S - M S for the simultaneous determination of T H C COOH and its glucuronide in urine [45]. In this method the cleavage of conjugates was omitted. THC-COOH and its glucuronide were extracted from urine with ethyl acetate/ ether (1 : 1) and separated on a C8 column with a gradient of ammonium formate buffer and ACN. ESI-MS-MS was used for detection using protonated quasi-molecular ions as precursor ions and two fragment ions for each drug as product ions. The specificity of the method was checked using enzymatic hydrolysis of THC-COOH-glucuronide. In another study, Weinmann et al. [46] developed a fast method for determination of T H C COOH in urine using automated solid phase extraction after alkaline hydrolysis. HPLC was done using gradient elution on a short ODS column. THC-COOH was detected with APCI-MS/MS in negative ionisation mode. Three product ions, originating from the deprotonated quasi-molecular ion were used for identification and quantification achieving an LOD of 2.0 ~g/1 and an LOQ of 5.1 ~g/1. 8.3.1.5 LSD

LSD (lysergic acid diethylamide) is an extremely potent hallucinogenic drug. Its single dose ("trip") ranges from 30 to 100 ~g. The drug is rapidly and extensively metabolised, and only a very small fraction is excreted unchanged in the urine [47]. Webb et al. [48] developed an immunoaffinity extraction of LSD from urine followed by LC-ESI-MS. Methysergide was used as internal standard for quantification. The protonated molecular ion and two fragments were monitored and an LOD of 0.5 ~g/1 was achieved using 5 ml of urine. In the next study of the same group, the methysergide was replaced as internal standard by a tri-deuterated LSD analog [49]. This was associated with much better accuracy and precision, however, the LOD (0.5 ~g/1) was not improved. Kanel et al. References pp. 267-270

0

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

A9_THC

.

.........

.

.

0--~-~ 0,0

~ ~ ~ I ~ ~"~ 0,5

.

.

.

.

.

~ .

.

.

.

i '

" 2,5 Time (rain)

315=~193

.

.

~ ~ 'I ~ ~'~ 3,0

315=~259

~ I ~ ~ ~" ~"I ~ ~ ~ ~ 1,0 1,5 2,0

1oO.o . 9 THe 43

0

.

A9-THC-COOH .

.

~oi A9-THC 1~176

0

50

0

--4

.

r

~ ~ i ~ ~ ~ b '[ 4,0 4,5

RT: 4,69

RT: 4,71

RT:-4,70

345=~327

345=~299

331 ~295

~ i ~-~ 3,5

.

331~313

5,0

Fig. 8.6. L C - A P C I - M S - M S of THC, 1 1 - O H - T H C and T H C - C O O H extracted from blood at concentration level of 0.25 txg/1. From ref. [42] with permission of the authors,

'4=

x~ 50

e-"

.

A9-THC-OH

..............

A9_THC_OH

~~176 i~o A9-THC-COOH

~'100

~ O

!

0 ~.

,o

100

o-~

5o ~

1oo~q

oo

o

Use of liquid chromatography-mass spectrometry in acute human toxicology

251

[50] extracted LSD and its demethylated metabolite from 1 ml of blood, serum, plasma, and urine with automated mixed-phase SPE. The drugs were detected with L C - E S I MS-MS using a phenyl column for separation with an LOD of 0.025 txg/1 for both substances reported. In the study of Bodin et al. [51], LSD was extracted from urine with organic solvent, back extracted to an acetate buffer and subjected to LC-ESI-MS. An LOD of 0.020 txg/1 was achieved by this procedure. The window of detection of LSD in urine following use is not longer than 12-22 h [52]. Poch et al. [53] demonstrated the importance of determination of 2-oxo3-hydroxy-LSD, a prevalent metabolite of LSD excreted via the urine. This metabolite, as well as LSD, nor-LSD, and iso-LSD, were determined in urine with LC-APCI-MS (ion trap). The concentrations of 2-oxo-3-hydroxy-LSD were distinctly higher than those of the parent drug and other metabolites (Fig. 8.7). In the next study Poch et al. [54] compared three detection methods for LSD and metabolites: LC-APCI-MS, L C APCI-MS-MS (ion trap) and GC-MS. The latter method was used only for the parent drug. Very good agreement between both LC-MS methods was found. According to the authors, the detection window for LSD use may be significantly increased by determination of the metabolite. Sklerov et al. [55] determined LSD and 2-oxo3-hydroxy-LSD in blood and urine. The analytes were isolated from urine with alkaline solvent extraction whilst for blood a solvent extraction followed by SPE was used. L C ESI-MS was applied with in-source collision-induced dissociation and monitoring of three ions for each compound. LODs of 0.1 txg/1 for LSD and 0.4 txg/1 for the metabolite in urine were obtained. In authentic cases, 2-oxo-3-hydroxy-LSD was found in high concentrations in urine but was not present in blood samples. An L C - E S I - M S - M S technique was applied by Canezin et al. [56] for determination of LSD, iso-LSD in plasma with LODs of 0.02 txg/1 for both compounds. In urine also metabolites were also detected including 2-oxo-3-hydroxy-LSD, nor-LSD, nor-iso-LSD, 13- and 14-OH-LSD, lysergic acid ethylamide, trioxydated-LSD and lysergic acid ethyl-2hydroxyethylamide. The method was applied in two cases of drug abuse.

8.3.2 Therapeutic drugs 8.3.2.1 Benzodiazepines

Midazolam in a short-acting benzodiazepine used for induction of anesthesia. This drug and its active hydroxylated metabolite were extracted from serum with etherisopropanol (98:2) at alkaline pH and separated on an ODS column (Nucleosil C18, 150x lmm). The drugs were determined with ESI-LC-MS in SIM mode. The protonated quasi-molecular ions and fragments of both compounds were monitored with an LOQ for both compounds of 0.5 txg/1 [57]. In another study, where midazolam and hydroxymidazolam as well as triazolam and hydroxytriazolam were used as internal standards, the analytes were extracted from plasma with Oasis HLB cartridges and determined with L C - A P C I - M S - M S [58]. All drugs were eluted within 2 rain. and the LOQ was 0.1 p~g/1 using multiple reaction monitoring. Flunitrazepam, a potent hypnotic drug, is of particular toxicological importance due to high toxicity in combination with ethyl alcohol and subsequent misuse in drugReferences pp. 267-270

70

50

5

6

5.31 5.75

7

6.96 7.71

8

O-H-LSD

8.73

9

10

iii

BBL-

13

~

14

~I

14.10

Nor-LSD 12.60 12.99

11 12 T i m e (min)

10.08 11.15

O - H - L A M P A (IS)

15

___

LSD

SPECIMEN 47

16

.95

LAMPA

17

17.00

18

17.27

Iso-LSD

Fig. 8.7. LC-APCI-MS-MS (ion trap) of LSD and metabolites extracted from urine sample. From ref. [53] with permission of Elsevier Science.

10

20

30

r ~ 40 n~

.=.. ,.e-.,,e

"0 t-- 6O .Q
324

7.38

1300

5"

" i

.

.

.

.

.

.

9

Time, rain Fig. 8.8. Comparison of simultaneous detection of polythiazide in urine by UV at 220 nm (top), LC-ESI-MS (middle) and LC-ESI-MS-MS (bottom). From ref. [77] with permission of Elsevier Science.

Use of liquid chromatography-mass spectrometry in acute human toxicology

257

gradient. An abundant ammonia adduct (M + NH4) + and much less abundant protonated quasi molecular ion ( M + H ) + were observed in ESI. In APCI the substances decomposed to give many fragments. Time-scheduled selected reaction monitoring was applied, using (M + NH4) + ions as precursor ions and one fragment for each compound as product ion. The recovery ranged from 11% to 64% with LODSs from 0.05 to 1 Ixg/1. The method was applied for the TDM of cardiac glycosides. Lacassie et al. [79] described a non-fatal case of intoxication with cardiac glycosides from foxglove (see Section 3.3.2.) A feasibility of separation of 17 cardiac glycosides was demonstrated (Fig. 8.9). 8.3.2.6 Muscle relaxants

Cisatracurium, one of ten isomers contained in atracurium besylate, is an intermediateacting neuromuscular blocking agent. The drug and its metabolites were identified in urine and bile using LC-ESI-MS coupled to an on-line radioactivity monitor [80]. Rocuronium, neuromuscular blocking agent used widely during general anesthesia, was determined in plasma of patients. The drug and internal standard (verapamil) were extracted from plasma with dichloromethane and separated on an ODS column by gradient elution with ACN-0.1% TFA. Protonated quasi molecular ions for both compounds were monitored. The LOQ was 25 Ixg/1 [81]. Ballard et al. [82] developed an L C - Q - T O F - M S - M S method for the determination of pancuronium, vecuronium, tubocurarine, rocuronium, and succinylcholine in postmortem tissues. These drugs were isolated using a combination of solvent extraction followed by an ion pairing SPE on C 18 cartridges. The detection based on exact mass measurement of product ions. 8.3.2.7 Antidiabetics

Ramos et al. [83] published a fast L C - A P C I - M S - M S method for determination of glibenclamide in human plasma after acetonitrile precipitation, using deuterated analog as internal standard. The retention time of the drug was 3 min. and the LOQ 1 Ixg/1. The sulfonylurea antidiabetics tolbutamide, chlorpropamide, glibenclamide and glipizide were detected and quantified in serum using LC-ESI-MS by Magni et al [84]. SPE on C18 cartridges gave much cleaner extracts than acidic toluene extraction. The drugs were separated with a methanol-acetic acid gradient and detected by ESI-MS in full scan mode (positive ions, m/z 265-510). For quantification the protonated quasimolecular ions were used. The LOQ was 10 Ixg/1 for all drugs (Fig. 8.10). Identification, detection and quantification of insulin in blood are very important in clinical and forensic toxicology. The presence of bovine or porcine insulin in the blood of a non-diabetic person may serve as evidence of poisoning. Since insulin and Cpeptide are released from proinsulin, the determination of both products may be of relevance. Very high level of human insulin without elevation of the C-peptide indicates exogenous administration. When both insulin and C-peptide are present in high concentrations, the presence of pancreatic tumor or administration of insulin-releasing drugs, like sulfonylureas, is possible. A straightforward LC-ESI-MS method for determination of human, bovine and porcine insulin, as well as C-peptide, was published by Darby et al. [85]. The analytes were isolated from acidified plasma with References pp. 267-270

o%

.5 .Oe4.

5.0e

1.0e5

5000

'

___

.

.

.

O,GO•

.

.

9

.

.

i' 5

5

i 10

10

DIGOXIGF:: NIN B~S DIGITOXOSIDE

.... ;10

II

H

rain

,~

,:,,?o?,~,,,,

ALPHA ACETYL

rain

Time,

116

rain

LANATOS IDE C 3.74

J

~

Time,

15

, ,~

.

-~

~'IO0x'N

.

l o,%o?,~

Time,

115

STANOARD 0.e~

.

. ;0

j

210

/

2%

.

.

.

";"

DIGITOXIGENIN MONO DIGITOXOSIDE

.....

.

. ;5

i

2 5

/i

G , T~Lo~OX, N

2'5

IN

o,,.OXN

OLEANDR 25.55

~

310

310

310

)%

~--

Fig. 8.9. L C - E S I - M S - M S of an extract of serum sample spiked with the 17 cardiac glycosides to the concentration of 50 I~g/l each. From ref. [79]. Copyright ASTM, reprinted with permission.

!

~

.

INTERNAL t.m

259

Use of liquid chromatography-mass spectrometry in acute human toxicology

1001 m/z:271 1

100- m/z:277 100---m/z:446 loft m/z:494 6:00

~

" 8:00

~10:00

min

100i m/z:271

.9v,,,~~ 100- m/Z:T77 "~

-

_

~

~

~

0'3

9

,~ 100- in/z:446

.

.

.

.

.

.

~ v==,~

0 ;>" 100--m/z:494 "-

/~

o 10:00

15:00

1001m/z:271,

min

___j~

100-] m]z: 446 l

20:00

~ j ~

,

-

1~176 z:494 6:00

7:00

. . . . . . . . . . . . .

8:00

9:00

10:00 min

Fig. 8.10. LC-ESI-MS of tolbutamide (m/z 271), chlorpropamide (m/z 277), glipizide (m/z 446) and glibenclamide(m/z 494). Upper trace: methanolstandards, middle trace: toluene serum extract, lower trace: serum SPE extract. Fromref. [84] with permissionfromAcademicPress.

References pp. 267-270

260

Chapter 8

C18 SPE cartridges, separated on a 150x 2.1 mm C18 column and detected with ESIMS (ion trap). Multiply charged molecular ions (M+3H) +3 and ( M + 4 H ) +4 were monitored. Both ions were selected for quantification, since the intensity ratio was variable between runs. The LOQ was 1 txg/1 and a comparison with radioimmunoassay showed full agreement of values. The stability of insulin from blood stored in various conditions was studied. Zhu et al. [86] developed a L C - E S I - M S - M S assay for quantification of human insulin, its analogue, and their catabolites in plasma. The compounds were isolated through precipitation followed by SPE. The quantification range was 1-500 txg/1.

8.3.2.8 Hormones Fiori et al. [87] published an LC-APCI-MS procedure for identification of the main corticosteroids: beclamethasone, betamethasone, dexamethasone, fludrocortisone, flumethasone, hydrocortisone, methylprednisolone, prednisolone, prednisone, and tramcinolone in milk replacers. The method, however, may also be adapted for human biofluids. Solid phase extraction was applied for isolation, the LOD was about 12 txg/1 of reconstituted milk. Corticosterone is an indicator of stress in laboratory animals. This compound was extracted from 0.1 ml of rat plasma with ethyl ether and subjected to ESI-LC-MS in SIM mode [88]. The hormone was separated on an octyl column using a methanolwater acetic acid mobile phase, with a run time of 7 min. The sodium adducts (M + Na) + of the analyte and internal standard were monitored giving an LOD of 3 pg of corticosterone on-column (the calibration curve began at 2 Ixg/1 plasma). In experimental samples of stressed animals, high cortisone levels and the presence of 18-hydroxydeoxycorticosterone were demonstrated. Melatonin (N-acetyl-5-methoxytryptamine) is a pineal gland hormone, which is used for the treatment of sleep and immunological disorders. This drug and its hydroxylated metabolite were determined in plasma after enzymatic hydrolysis and solvent extraction, using LC-ESI-MS. The LODs were 2 Ixg/1 for both substances [89]. Catecholamine metabolites metanephrine and normetanephrine were determined in post-mortem fluids and organs using SPE followed by L C - E S I - M S - M S (QTOF) [90]. The analysis was applied in a case of a serial killer involving over 30 victims in several hospitals.

8.3.2.9 Other drugs of toxicological relevance Methylphenidate is a central stimulant, with a history of abuse, which is used for the treatment of narcolepsy. The drug has two chiral centers and is marketed as a racemic mixture. Ramos et al. have described two LC-APCI-MS-MS procedures for the determination of the racemate [91] and enantiomers [92]. For isolation of the racemate a 96-well plate solvent extraction was utilized. The LOQ was below 100 ng/1 plasma. Scopolamine and the internal standard atropine was extracted from 0.2 ml serum using Oasis HLB cartridges in an automated procedure [93]. The drugs were separated on a C18 column by gradient elution with a mobile phase consisting of ACN, ammonium acetate and formic acid. Detection was with ESI-MS-MS (positive ions).

Use of liquid chromatography-mass spectrometry in acute human toxicology

261

Protonated quasi-molecular ions of scopolamine and atropine were used as parent ions and two product ions for each drug were monitored. The method, which was used for pharmacokinetic studies, had an LOQ of 0.05 Ixg/1. Nicotine and its seven metabolites were determined in urine by L C - E S I - M S - M S after enzymatic hydrolysis and solvent extraction. For quantification, deuterated or 13Clabeled analogs were used for each compound. The LODs ranged from 5 to 38 ng/1. 3-OH-cotinine was identified as a most dominant metabolite [94]. Weinmann et al. [95] have described the post-mortem detection and identification of sildenafil (Viagra) and its metabolites by LC-ESI-MS and LC-ESI-MS-MS. The drug and three metabolites were isolated from urine and organs of the highly putrefied body of an 80-year-old man, who died due to coronary sclerosis. Analytical results were compared with experimental data obtained from a volunteer who ingested 25 mg sildenafil. Sildenafil may be also used as a doping substance for racehorses, due to its protective action against exercise induced pulmonary hemorrhage. Rudy et al. [96] have presented a LC-ESI-MS/MS method for identification of sildenafil and metabolites in equine plasma and urine. A liquid/liquid alkaline extraction was applied to the sample, followed by separation on CN-column and Q-TOF detection. Product ions in the range m/z 100-500 from four parent ions, corresponding to sildenafil and its three metabolites, were monitored.

8.3.3 Environmental poisons and natural compounds 8.3.3.1 Pesticides LC-ESI-MS in combination with GC-MS was used for the determination of 61 pesticides in human biological fluids. Solid phase extraction with polymeric supports was used for isolation with LC-MS for polar pesticides (carbamates, benzimidazoles). The method was applied in acute poisonings and in post-mortem analysis [97]. Turcant et al. [98] reported a case of self-poisoning with metobromuron, a phenylurea derivative used as a herbicide. The patient was admitted into an emergency unit 17 h after ingestion of 250 ml of "Patoran" (50% metobromuron) and 3 1 of beer. Metabolic acidosis and 80% MetHb was observed. Metobromuron and its three metabolites were identified in blood, whilst in urine eight further metabolites were found. The drugs were extracted with ethyl ether at alkaline pH and determined with HPLC-DAD, L C - E S I MS and LC-ESI-MS-MS. High Met-Hb was explained by the postulated metabolism of metobromuron to bromoaniline. Warfarin enantiomers were extracted from plasma with diethyl ether in acidic conditions and separated on a [3-cyclodextrin column in ACN-acetic acid-TEA (1000:3:2.5) in less than 10 min. The enantiomers were determined with ESI-MS-MS (negative ions) using MRM for both drug and the internal standard p-chlorowarfarin. The limit of quantification was 1 ng/ml [99].

8.3.3.2 Plant and bacterial toxins Lacassie et al. [79] reported a case of a 36-year-old female who ingested a concoction of foxglove leaves (Digitalis purpurea) and was admitted to emergency unit some hours

References pp. 267-270

262

Chapter 8

later. She developed a sinus bradycardia, abdominal pain, nausea and vomiting. After five days of treatment she was discharged without symptoms. Cardiac glycosides: acetyldigitoxin, convallatoxin, deslanoside, digitoxigenin, digitoxin, digoxin, gitaloxin, gitoxin, latanoside C, methyldigoxin, oleandrin, proscilardin and strophantidin were determined in blood and urine samples from the patient with LC-ESI-MS. The analytes were extracted with an organic solvent mixture after acetonitrile precipitation and separated on an ODS column using a gradient of ACN-ammonium formate pH 3.0. For each analyte the protonated quasi molecular ion (M + H) + and one or two fragment ions were monitored. The recovery was 67.8 to 98.6%, and the limit of detection 1 to 10 ~g/1. This sensitivity was adequate for acute poisoning cases but not for therapeutic drug monitoring. Glycoside levels in blood were monitored from 8 to 100 hours after intoxication. In a case reported by Gaillard and Pepin [100], veratridine and cevadine (toxins present in Veratrum album) were identified and quantified in the blood of two persons found in a mountain lake. In the stomachs, seeds of Veratrum were identified. LC-ESI-MS was used. Measured blood concentrations were 0.17 and 0.40 p~g/1 for veratridine and 0.32 and 0.48 ~g/1. for cevadine. The cardiac glycosides oleandrin, odoroside, neritaloside and the aglycone oleandrigenin, present in the extract of Nerium Oleander (Anvirzel) were analyzed with a QqTOF-MS. CID mass spectra were obtained with mass accuracy greater than 5 ppm. The LOD for oleandrin was 20 pg injected. The method was applied to the determination of oleander glycosides in human plasma after intramuscular injection of Anvirzel [101]. A 45-year-old female took Nerium oleander in an attempted suicide, showing nausea, vomiting, cardiovascular shock and sinus bradycardia. Oleandrin was measured in plasma at 1.1 ng/ml using ESI-LC-MS. Oleandrin was also identified in urine [ 102]. Mass spectral data for 18 taxanes were obtained from Taxus baccata and Taxus brevifolia preparations. The mass spectral library was used for rapid identification of taxanes in mixtures [ 103]. An L C - A P I - M S - M S method was developed for the identification of Podophyllum emodi based on the profile of lignan marker products [ 104]. The structural data reported for this herb were matched with the experimental data. In this way the author differentiated Podophyllum emodi from closely related species. Maurer et al. [105] determined or- and [3-amanitine in urine and plasma of patients intoxicated with Amanita mushrooms. The toxins were isolated with immunoaffinity extraction columns and determined with LC-ESI-MS, with an LOD of 2.5 p~g/1. Azaspiracid and its two analogs are toxic compounds present in mussels, which caused poisoning in the Netherlands in 1995 and in Ireland in 1997. An LC-MS method was developed for the determination of these compounds with a detection limit of 50 pg [ 106,107]. Draisci et al. [ 108] extracted azaspiracid from mussel meat with acetone and separated it on a C18 column with a mobile phase of ACN-H20 (85:15) containing 0.03% TFA. L C - E S I - M S - M S was performed, using the protonated molecular ion as precursor ion and three dehydrated fragments as product ions. The detection limit was 20 pg. LC-MS and LC-MS-MS have also been found for the identification of acidic analogues of the marine toxin pectenotoxin-2 [109]. This compound is present in toxic marine phytoplankton Dinophysis acuta and is responsible for diarrheic shellfish

Use of liquid chromatography-mass spectrometry in acute human toxicology

263

poisoning. The substances were isolated from bulk phytoplankton and analysed using flow injection-LC/MS. The diarrheic shellfish-poisoning toxins okadaic acid, dinophysistoxin and pectenotoxin-6 were determined by LC-ESI-MS in scallops and mussels and the profile of toxins in organisms living in the Mutsu Bay was assessed [110]. An amnestic shellfish-poisoning toxin, domoic acid, was found in shellfish samples collected in 1998/1999 in Scotland [111]. The compound was detected in various shellfish species with LC-UV diode array detection and LC-ESI-MS with the latter method showing more specific results.

8.3.3.3 Inorganic compounds Inorganic arsenic compounds, which are the most toxic form of arsenic, are methylated after ingestion to monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA). These compounds, which are much less toxic than inorganic arsenic, are excreted with the urine. The kinetics of elimination of inorganic As and its methylated metabolites may be useful for the clinician to monitor the efficiency of the chelation therapy after acute intoxication. Le Bouil et al. [112] developed a L C - M S - M S method for the determination of MMA and DMA in urine. The substances were extracted with chloroform and separated on a 33 x 4.6 mm ODS column in less than 1 min. (Fig. 8.11). An attempt to use direct injection of urine was unsuccessful, due to severe signal suppression by urine matrix.

8.3.4 Screening procedures for multiple compounds 8.3.4.1 General screening for various groups of drugs Weinmann et al. [113] used tuning compounds for the standardisation of in-source collision-induced fragmentation. Four drugs: haloperidol, paracetamol, metronidazole and metamizole were selected as tune compounds for LC-ESI-MS. Comparative experiments were performed using two LC-MS instruments with different construction of interface (Sciex API 365 and Agilent 1100 MSD). Very similar fragmentation patterns were observed after adjustment of fragmentor voltages of both instruments suggesting that the establishment of a generally applicable library of mass spectra obtained with different LC-MS instrument is possible, when the fragmentation energy has been adjusted using selected tune compounds. Rittner et al. [114] established a library of mass spectra of 70 various psychoactive drugs and metabolites with LC-ESI-MS. In the preliminary study, the efficiency of various solid phase extraction methods and HPLC columns was tested. The best results were obtained with C18 SPE cartridges and C18 columns. Chromatographic separation was performed using an ACN-water-methanol-formic acid gradient with mass spectra recorded at two levels of fragmentation energy in full scan mode (m/z 100-650, positive ions). For many drugs, sodium, potassium, and ACN adducts were observed. The usefulness of the screening procedure was checked on 140 serum samples taken from road traffic offenders. In 9.8% of cases various drugs, mostly benzodiazepines, were detected.

References pp. 267-270

Chapter 8

264

2.21e3 cps

TIC of +MRM 20001500 lOOO-

r

I,u

500L r

I

0.5

Time (rain) ,

1.0

,,..

, ,t

2.74e2 cps

141.0 1 91.1

'~

200,

MMA

r

lOO md

j ~

....

!

I

0.5

1.0

I

Time (rain) .,

.,

,

139.0 1 91.1

..

2.18o3 cps

2000

~ ~J

DMA

1500-

lOOO

5OO

4

j A

_

.

_

,

i

u O.5

1.'0 Time (rain)

Fig. 8.11. LC-ESIMS/MS of monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA) extracted from urine of a patient intoxicated with arsenic trioxide. From ref. [ 112] with permission of Preston Publ.

Lips et al. [ 115] performed a study on the applicability of LC-ESI-MS (positive ions) for the development of a mass spectral library based upon in-source collision-induced fragmentation. The influence of mobile phase composition on the reproducibility of

Use of liquid chromatography-mass spectrometry in acute human toxicology

265

mass spectra of drugs was tested, and data obtained with two instruments of the same brand but different type were compared. The breakdown curves (i.e. fragmentation profiles related to fragmentation energy) of selected drugs were compared with the data of other groups. The authors stated that the concentration of organic modifier, pH and molarity of the buffer exerted negligible influence on the mass spectra. This observation is in agreement with the previous finding of Bogusz et al. [116]. In order to obtain reproducible mass spectra, the fragmenter voltage was dynamically ramped based on the mass of substance. The efficiency of identification was tested on over 40 extracts from plasma containing various acidic and basic drugs. All analytes except phenobarbital were correctly recognized. Phenobarbital was not detected at all because of positive ionization applied. With experiments performed in negative ionization mode, acidic drugs were properly detected. Gergov et al. [ 117] presented a very straightforward approach for application of LCESI-TOF-MS in toxicological screening. The features of TOF-MS were utilized, i.e. high mass accuracy and high sensitivity over the full spectrum. On this basis a library was established containing 433 toxicologically relevant compounds (parent drugs and their metabolites) simply by the calculation of their monoisotopic masses. The mass range for compounds included in the library extended from 105 to 734 Da. This library was used for identification of drugs in urine extracts, using gradient elution with ACNammonium acetate buffer at pH 3.2. The retention data of drugs were stored, but not used for identification in this study. The results of LC-MS-TOF screening were in concordance with the results of GC and TLC screening run in parallel. According to authors, the method is very promising and allows screening for compounds with known formula even without reference compounds. This is of practical value, particularly for metabolites, due to the frequent lack of available reference standards. An interlaboratory reproducibility of mass spectra obtained with similar or different instruments is of primary relevance for establishing and use of mass spectra library. Weinmann, Gergov and Goemer [118] compared mass spectra, obtained with two identical and one different LC-MS-MS instruments in three laboratories. Methadone, benzoylecgonine and diazepam were used as test substances. Product ion spectra of protonated quasi-molecular were similar for all laboratories. Also, in-source CID spectra showed good similarity with the product ion spectra. The last point is very important, because in the first step of LC-MS screening, the in-source generated mass spectra are collected and eventually confirmed by MS-MS. 8.3.4.2 Group screening for substances belonging to the same therapeutic class An automated LC-ESI-MS/MS screening/confirmation method for 16 beta-blocking agents in urine samples has been described [119]. The drugs were tentatively identified on the base of retention time and (M + H) + value. For confirmation, any qualified compounds were automatically subjected to fragmentation and the product ions were compared with the library data. The limits of identification ranged from 0.02 to 1.2 mg/1. The same group developed an LC-ESI-MS-MS procedure for screening and quantification of 18 antihistamine drugs in blood [ 120]. Since the drugs belong to acidic and basic compounds, two solvent extractions were performed at pH 3 and 11 and the References pp. 267-270

266

Chapter 8

extracts were combined. The chromatography was done in 5 min. using an acetonitrileammonium acetate gradient (50 to 100% ACN). For preliminary identification, the protonated quasi-molecular ion, the most intense fragment and retention time of each peak were used. Confirmation of identity was done in a second run using the whole product ion spectrum. The limits of identification for each drug were below the lowest therapeutic concentration. Thieme et al. [77] established a method for screening and quantification of 32 diuretic compounds and their metabolites in urine. The method was developed for doping control and was based on LC-ESI-MS-MS. The drugs were extracted from urine using XAD columns and separated on C8 column in gradient of ACN-ammonium acetate. The library of mass spectra was developed using positive or negative ionization and optimized fragmentation conditions for particular compounds. Since diuretics may belong to acidic and basic drugs, the authors recommended two subsequent chromatographic runs (in positive and negative mode) in screening procedure. This was superior to using polarity switching in a single run. Lacassie et al. [97] have published a procedure for the determination of 61 pesticides of various classes (organophosphates, carbamates, organochlorines, benzimidazoles) in serum. A total of 47 compounds were determined with GC/MS after SPE on Oasis HLB cartridges, whereas LC-ESI-MS was applied for 14 thermolabile and polar pesticides, such as carbamates and benzimidazoles. These substances were isolated with Oasis MCX cation exchange cartridges. LC-MS was applied for clinical diagnostics in carbofuran and aldicarb self-poisonings.

8.4 CONCLUSIONS AND PERSPECTIVES Existing needs and observed trends of development allow the prediction of the main pathways of progress in LC-MS applied for clinical toxicology. This progress may be divided into three categories: 9With regard to the separation part of LC-MS, a trend to shortening of analysis time will be observed. This may be achieved either through introduction of columns assuring faster and more efficient separation (short, small particle size or monolithic columns) or through the application of fast elution (fast gradient, high percentage of organic solvent in mobile phase, high flow rate). 9Detection part of LC-MS, i.e. the mass spectrometer itself, will become smaller without compromising quality. Bench-top tandem mass spectrometers, at an affordable price, already appeared on the market and will definitely become more widespread in clinical toxicological laboratories. 9The application area of LC-MS will cover practically the whole spectrum of compounds of toxicological relevance. The use of LC-MS as a tool for general screening procedures will be common, and the libraries of mass spectra will be available with any purchased instrument.

Use of liquid chromatography-mass spectrometry in acute human toxicology

267

8.5 ABBREVIATIONS USED IN THE TEXT ACN LC-APCI-MS LC-ESI-MS LC-ESI-MS-MS LC-ESI-TOF/MS LC-MS LOD LOQ ODS SIM SPE TFA

acetonitrile liquid c h r o m a t o g r a p h y - a t m o s p h e r i c p r e s s u r e c h e m i c a l ionization mass spectrometry liquid c h r o m a t o g r a p h y - e l e c t r o s p r a y ionization m a s s spectrometry liquid c h r o m a t o g r a p h y - e l e c t r o s p r a y ionization t a n d e m mass spectrometry liquid c h r o m a t o g r a p h y - e l e c t r o s p r a y ionization time-of-flight mass spectrometry liquid c h r o m a t o g r a p h y - m a s s s p e c t r o m e t r y ( a t m o s p h e r i c p r e s s u r e ionization) limit of detection limit of quantitation octadecylsilica selected ion m o n i t o r i n g solid p h a s e extraction trifluoroacetic acid

8.6 REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14

15 16 17 18

D. Temesi and B. Law, LC-GC International, 12 (1999) 175. W. Naidong, W. Shou, Y.L. Chen and X. Jiang, J. Chromatogr. B, 754 (2001) 387. W. Naidong, J.W. Lee, X. Jiang, M. Wehlin, J.D. Hulse and P.P. Lin, J. Chromatogr. B, 735 (1999) 255. W.Z. Shou, X. Jiang, B.D. Beato, W. Naidong, Rapid Commun. Mass Spectrom., 15 (2001) 466. W. Naidong, X. Jiang, K. Newland, R. Coe, E Lin and J. Lee, J. Pharm. Biomed. Anal., 23 (2000) 697. M. Katagi, M. Nishikawa, M. Tatsuno, A. Miki and H. Tsushihashi, J. Chromatogr. B, 751 (2001) 177. M. Katagi, M. Tatsuno, A. Miki, M. Nishikawa, K. Nakajima and H. Tsushihashi, J. Chromatogr. B, 759 (2001) 125. J. Zweigenbaum, K. Heinig, S. Steinborner, T. Wachs and J. Henion, Anal. Chem., 71 (1999) 2294. Y.E Cheng, Z. Lu and U. Neue, Rapid Comm. Mass Spectrom., 15 (2001) 141. K. Cabrera, D. Lubda, H.M. Eggenweiler, H. Minakuchi and K. Nananishi, J. High Resol. Chromatogr., 23 (2000) 93. J.T. Wu, H. Zeng, Y. Deng and S.E. Unger, Rapid Commun. Mass Spectrom., 15(2001) 1113. W.M.A. Niessen, J. Chromatogr. A, 794 (1998) 407. W.M.A.Niessen, J. Chromatogr. A, 856 (1999) 179. K.M. Clauwaert, J.E Van Bocxlaer, H.J. Major, J.A. Claereboudt, W.E. Lambert, E.M. Van den Eeckhout, C.H. Van Peteghem and A.P. De Leenheer, Rapid Commun. Mass Spectrom., 13 (1999) 1540. H. Zhang and J. Henion, J. Chromatogr. B, 757 (2001) 151. M.J. Bogusz (Ed.), Forensic Science. Handbook of Analytical Separations, Volume II, (Series Ed. R.M. Smith), Elsevier, Amsterdam, 2000. M.J. Bogusz: Opiate Agonists, in: M.J. Bogusz (Ed.), Forensic Science. Handbook of Analytical Separations, Volume II, (Series Ed. R.M. Smith), Elsevier, Amsterdam, 2000, pp. 3-65. S. Pichini, I. Altieri, M. Pellegrini, P. Zuccaro and R. Pacifici, Mass Spectrom., Rev., 18 (1999) 119.

268

Chapter 8

19 20

G. Schanzle, S. Li, G. Mikus and U. Hofmann, J. Chromatogr. B, 721 (1999) 65. A. Dienes-Nagy, L. Rivier, G. Giroud, M. Augsburger and E Mangin, J. Chromatogr. A, 854 (1999) 109. M. Blanchet, G. Bru, M. Guerret, M. Bromet-Petit and N. Bromet, J. Chromatogr. A, 854 (1999) 93. M.H. Slawson, D.J. Crouch, D.M. Andrenyak, D.E. Rollins, J.K. Lu and EL. Bailey, J. Anal. Toxicol., 23 (1999) 116. A. Cailleux, A. Le Bouil, B. Auger, G. Bonsergent, A. Turcant and P. Allain, J. Anal. Toxicol., 23 (1999) 620. J.M. Gaulier, E Marquet, E. Lacassie, J.L. Dupuy and G. Lachatre, J. Forensic Sci., 45 (2000) 226. A. Polettini and M.A. Huestis, J. Chromatogr. B, 754 (2001) 447. T. Breindahl and K. Andreasen, J. Chromatogr. B, 736 (1999) 103. J.O. Svensson, J. Sawe and A. A1-Shurbaji, Ther. Drug Monit., 23 (2001) 399. A. Ceccato, E Vanderbist, J.Y. Pabst and B. Streel, J. Chromatogr. B, 748 (2000) 65). S.J. Juzwin, D.C. Wang, N.J. Anderson and EA. Wong, J. Pharm. Biomed. Anal., 22 (2000) 469. M.J. Bogusz, K.D. Kruger and R.D. Maier, J. Anal. Toxicol., 24 (2000) 77. M.J. Bogusz, J. Chromatogr. B, 748 (2000) 3. M. Sato, T. Mitsui and H. Nagase, J. Chromatogr. B, 751 (2001) 277. M. Katagi, M. Tatsuno, A.S. Miki, M. Nishikawa and H. Tsushihashi, J. Anal. Toxicol., 24 (2000) 354. H. Kataoka, H.L. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 257. K.M. Clauwert, J.E VanBocxlaer, E.A. De Letter, S. Van Calenbergh, W.E. Lambert and A.E De Leenheer, Clin. Chem., 46 (2000) 1968. M. Wood, D. De Boeck, N. Samyn, D. Cooper and M. Morris, Poster presented on 49th ASMS Conference, Chicago, 2001. P.M. Jeanville, E.S. Estape, I. Torres-Negron and A. Marti, J. Anal. Toxicol., 25 (2001) 69. G. Skopp, A. Klingmann, L. Potsch and R. Matern, Ther. Drug Monit., 23 (2001), 174. A. Klingmann, G. Skopp and R. Aderjan, J. Anal. Toxicol., 25 (2001) 425. S.N. Lin, D.E. Moody, G.E. Bigelow and R.L. Foltz, J. Anal. Toxicol., 25 (2001) 497. P. Mireault. Analysis of Ag-tetrahydrocannabinol (THC) and its two major metabolites by APCI-LC/ MS. Poster at the 46th ASMS conference, Orlando, 1998. E Picotte, P. Mireault and G. Nolin. A rapid and sensitive LC/APCI/MS/MS method for the determination of A9-tetrahydrocannabinol and its metabolites in human matrices. Poster at the 48th ASMS conference, Long Beach, 2000. T. Breindahl and K. Andreasen, J. Chromatogr. B, 732 (1999) 155. S.S. Tai and M.J. Welch, J. Anal. Toxicol., 24 (2000) 385. W. Weinmann, S. Vogt, R. Goerke, C. Muller and A. Bromberger, Int. J. Legal Med., 114 (2001) 252. W. Weinmann, M. Goerner, S. Vogt, R. Goerke and S. Pollak, Forensic Sci. Int., 121 (2001) 103. R.C. Baselt, Disposition of Toxic Drugs and Chemicals, in Man. Foster City, 2000, pp. 486-489. K.S. Webb, P.B. Baker, N.P. Cassells, J.M. Francis, D.E. Johnson, S.H. Lancaster, P.S. Minty, G.D. Reed and S.A. White, J. Forensic Sci., 41 (1996) 938. S.A. White, A.S. Kidd and K.S. Webb, J. Forensic Sci., 44 (1999) 375. J. de Kanel, W.E. Vickery, B. Waldner, R.M. Monahan and EX. Diamond, J. Forensic Sci., 43 (1998) 622. K. Bodin and J.O. Svensson, Ther. Drug Monit., 23 (2001) 389. C.C. Nelson and R.L. Foltz, J. Chromatogr., 580 (1992) 97. G.K. Poch, K.L. Klette, D.A. Hallare, M.G. Manglicmot, R.J. Czarny, L.K. McWhorter and C.J. Anderson, J. Chromatogr. B, 724 (1999) 23. G.K. Poch, K.L. Klette and C. Anderson, J. Anal. Toxicol., 24 (2000) 170. J.H. Sklerov, J. Magluilo, K.K. Shannon and M.L. Smith, J. Anal. Toxicol., 24 (2000) 543. J. Canezin, A. Cailleux, A. Turcant, A. LeBouil, P. Harry and P. Allain, J. Chromatogr. B, 765 (2001) 15. E Marquet, O. Baudin, J.M. Gaulier, E. Lacassie, J.L. Dupuy, B. Francois and G. Lachatre, J. Chromatogr. Biomed. Appl., 734 (1999) 137.

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

Use of liquid chromatography-mass spectrometry in acute human toxicology 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76

77 78 79 80 81 82

83 84 85 86 87 88 89 90

91 92

269

M.L. Ware and B.J. Hidy, LC/MS/MS determination of midazolam and 1-hydroxymidazolam in human plasma. Poster at the 49th ASMS Conference, Chicago, 2001. M.J. Bogusz, R.D. Maier, K.D. Kruger and W. Fruchtnicht, J. Chromatogr. B, 713 (1998) 361. M.A. LeBeau, M.A. Montgomery, J.R. Wagner and M.L. Miller, J. Forensic Sci., 45 (2000) 1133. J. Darius and E Banditt, J. Chromatogr. B, 738 (2000) 437. H. Yuan, Z. Mester, H. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 718. J.M. Hoskins, A.S. Gross, G.M. Shenfield and L.E Rivory, J. Chromatogr. B, 754 (2001) 319. EC. Sutherland, D. Badenhorst, A.D. de Jager, T. Scanes, H.K. Hundt, K.J. Swart and A.E Hundt, J. Chromatogr. A, 914 (2001) 45. M. Berna, R. Shugert and J. Mullen, J. Mass Spectrom., 33 (198) 1003. M.J. Bogusz, K.K. Kruger, R.D. Maier, R. Erkwoh and E Tuchtenhagen, J. Chromatogr. B, 732 (1999) 257. M. Aravagiri and S.R. Marder, J. Pharm. Biomed. Anal., 26 (2001) 301. M. Aravagiri and S.R. Marder, J. Mass Spectrom., 35 (2000) 718). K.E. Goeringer, I.M. McIntyre and O.H. Drummer, Forensic Sci. Int., 121 (2001) 70. T. Kumazawa, H. Seno, K. Suzuki-Watanabe, H. Hattori, A. Ishii, K. Sato and O. Suzuki, J. Mass Spectrom., 35 (2000) 1091. C. Muller, S. Vogt, R. Goerke, A. Kordon and W. Weinmann, Forensic Sci. Int., 113 (2000) 415. E Lachatre, P. Marquet, S. Ragot, J.M. Gaulier, E Cardot and J.L. Dupuy, J. Chromatogr. Biomed. Appl., 738 (2000) 281. G.L. Lensmeyer and M.A. Poquette, Ther. Drug Monit., 23 (2001) 239. T.E. Murdter, J. Coller, A. Claviez, E Schonberger, U. Hofmann, E Dreger and M. Schwab, Clin. Chem., 47 (2001) 1437. S. Steinborner and J. Henion, Anal. Chem., 71 (1999) 2340. L. Hogge, M. Powell, S. Ambrose and G. McKay. The development of an LC/MS/MS method for the determination of carboplatin in plasma. Poster presented on the 49th ASMS Conference, Chicago, 2001. D. Thieme, J. Grosse, R. Lang, R.K. Mueller and A. Wahl, J. Chromatogr. B, 757 (2001) 49. E Guan, A. Ishi, H. Seno, K. Suzuki-Watanabe, T. Kumazawa and O. Suzuki, Anal. Chem., 71 (1999) 4034. E. Lacassie, E Marquet, S. Martin-Dupont, J.M. Gaulier and G. Lachatre, J. Forensic Sci., 45 (2000) 1154. G.J. Dear, J.C. Harelson, A.E. Jones, T.E. Johnson and S. Pleasance, Rapid Commun. Mass Spectrom., 9 (1995) 1457. C. Farenc, C. Enjalbal, E Sanchez, E Bressolle, M. Audran, J. Martinez and J.L. Aubagnac, J. Chromatogr. A, 910 (2001) 61. K.D. Ballard, W.V. Vickery, R.D. Citrino, EX. Diamond and E Rieders. Analysis of quarternary ammonium neuromuscular blocking agents in a forensic setting using LC-MS/MS with internal standardization on a Q-TOE Poster presented on 49th ASMS Conference, Chicago, 2001. L. Ramos, R. Bakhtiar and E Tse, Rapid Commun. Mass Spectrom., 13 (1999) 2439. E Magni, L. Marazzini, S. Pereira, L. Monti and M. Galli Kienle, Anal. Biochem., 282 (2000) 136. S.M. Darby, M.L. Miller, R.O. Allen and M. LeBeau, J. Anal. Toxicol., 25 (2001) 8. Y. Zhu, D. Meyer, Z. Lam and B. Chien, Simultaneous quantitation of human insulin, its analogue, and their catabolites in plasma. Poster presented on 49th ASMS Conference, Chicago, 2001. M. Fiori, E. Pierdominici, E Longo and G. Brambilla, J. Chromatogr. A, 807 (1998) 219. A. Marwah, P. Marwah and H. Lardy, J. Chromatogr. B, 757 (2001) 333. S. Hartter, S. Morita, K. Bodin, C. Ursing, G. Tybring and L. Bertilsson, Ther. Drug Monit., 23 (2001) 282. W.E. Vickery, K.D. Ballard, R.D. Citrino and E Rieders. Analysis of the catecholamine metabolites metanephrine and normetanephrine in forensic tissue by LC-MS/MS with isotope dilution on a QTOE Poster presented on the 49th ASMS Conference, Chicago, 2001. L. Ramos, R. Bakhtiar and EL. Tse, Rapid Commun. Mass Spectrom., 14 (2000) 740. L. Ramos, R. Bakhtiar, T. Majumdar and EL. Tse, Rapid Commun. Mass Spectrom., 13 (1999) 2054.

270

Chapter 8

93 94

R. Oertel, K. Richter and W. Kirch, J. Chromatogr. B, 750 (2001) 121. Y. Xia, J.E. McGuffey, L. Wang, C.S. Sosnoff and J.T. Bernert, Analysis of urinary nicotine metabolite profiles by LC atmospheric pressure ionization tandem mass spectrometry. Poster presented at 49th ASMS Conference, Chicago, 2001. W. Weinmann, M. Bohnert, A. Wiedeman, M. Renz, N. Lehmann and S. Pollak, Forensic Sci. Int., 113 (2000) 381. J.A. Rudy, C.E. Uboh, L. Soma, E Guan, E. Birks, M. Kahler, D. Tsang, A. Watson and D. Teleis, Identification of sildenafil and metabolites in equine plasma and urine by LCMSMS. Poster presented at 49th ASMS Conference, Chicago, 2001. E. Lacassie, E Marquet, J. Gaulier, M. Dreyfuss and G. Lachatre, Forensic Sci. Int., 121 (2001) 116. A. Turcant, A. Cailleux, A. LeBouil, E Allain, E Harry and A. Renault, J. Anal. Toxicol., 24 (2000) 157. W. Naidong, ER. Ring, C. Midtlien and X. Jiang, J. Pharm. Biomed. Anal., 25 (2001) 219. Y. Gaillard and G. Pepin, J. Anal. Toxicol., 25 (2001) 485. X. Wang, J.B. Plomley, R.A. Newman and A. Cisneros, Anal. Chem., 72 (2000) 3547. A. Tracqui, E Kintz, E Branche and B. Ludes, Int. J. Leg. Med., 111 (1998) 32. E.H. Kerns, K.J. Volk, S.E. Hill and M.S. Lee, J. Nat. Prod., 57 (1994) 1391. S.K. Wong, S.K. Tsui, S.Y. Kwan, X.L. Su and R.C. Lin, J. Mass Spectrom., 35 (2000) 1246. H.H. Maurer, C.J. Schmitt, A.A. Weber and T. Kraemer, J. Chromatogr. B, 748 (2000) 125. K. Ofuji, M. Satake, Y. Oshima, T. McMahon, K.J. James and T. Yasumoto, Nat. Toxins, 7 (1999) 247. E. Ito, M. Satake, K. Ofuji, N. Kurita, T. McMahon, K. James and T. Yasumoto, Toxicon, 38 (2000) 917. R. Draisci, L. Palleschi, E. Feretti, A. Furey, K.J. James, M. Satake and T. Yasumoto, J. Chromatogr. A, 871 (2000) 13. K.J. James, A.G. Bishop, R. Draisci, L. Palleschi, C. Marchiafava, E. Feretti, M. Satake and T. Yasumoto, J. Chromatogr. A, 844 (1999) 53. T. Suzuki and T. Yasumoto, J. Chromatogr. A, 874 (2000) 199. E Hess, S. Gallacher L.A. Bates, N. Brown and M.A. Quilliam, J.A.O.A.C., 84 (2001) 1657. A. Le Bouil, A. Cailleux, A. Turcaint and E Allain, J. Anal. Toxicol., 23 (1999) 257. W. Weinmann, M. Stoertzel, S. Vogt and J. Wendt, J. Chromatogr. A, 926 (2001) 199. M. Rittner, E Pragst, W.R. Bork and J. Neumann, J. Anal. Toxicol., 25 (2001) 115. A.G. Lips, W. Lameijer, R.H. Fokkens and N.M. Nibbering, J. Chromatogr. B, 759 (2001) 191. M.J. Bogusz, R.D. Maier, K.D. Kruger, K.S. Webb, J. Romeril and M.L. Miller, J. Chromatogr. B, 844 (1999) 409. M. Gergov, B. Boucher, I. Ojanpera and E. Vuori, Rapid Commun. Mass Spectrom., 15 (2001) 521. W. Weinmann, M. Gergov and M. Goerner, Analusis, 28 (2000) 934. M. Gergov, J.N. Robson, E. Duchoslav and I. Ojanpera, J. Mass Spectrom., 35 (2000) 912. M. Gergov, J.N. Robson, I. Ojanpera, O.P. Heinonen and E. Vuori, Forensic Sci. Int., 121 (2001) 108.

95 96

97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 ll7 118 119 120

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 92003 Elsevier Science B.V. All rights reserved

271

CHAPTER 9

HPLC-MS(MS) for bioanalysis in drug discovery and development Brian Law AstraZeneca Pharmaceuticals, Department of Drug Metabolism and Pharmacokinetics, Mereside, Alderley Park, Macclesfield, Cheshire, SKIO 4TG, U.K.

9.1 INTRODUCTION The process of drug discovery within the pharmaceutical industry changed dramatically during the 1990s. There were significant developments in terms of synthetic chemistry, with the application of robotic methods, multi-parallel synthesis and combinatorial chemistry. With the exception of the last of these, which has never fulfilled its promise, the others are in everyday use in most medicinal chemistry laboratories and are capable of efficiently generating large numbers of potential drug compounds in a short space of time. During the same period, high-throughput screening (HTS) against pharmacological targets benefited massively from application of automated pipetting and assays systems allowing several thousands of compounds to be screened in a week or even a day. Many companies have built up large collections of compounds, often in excess of one million for the bigger companies such that it is not unusual now for a screening campaign against a particular therapeutic target to produce many hundreds or even many thousands of hits, i.e. molecules with activity against an enzyme. Even with some form of in silico analysis or analysis on paper, many of these hits will require experimental evaluation in terms of their physicochemical, biopharmaceutical, drug metabolism and pharmacokinetic properties. The above developments have conspired to significantly increase the potential workload of the analyst in the drug discovery environment. The process of drug development has also undergone significant changes during the same period. The driver in this case has been the need to minimise the time in development and hence maximise the return on investment whilst the compound is still under patent. Every extra day of patent life can mean an extra $100 million in sales for a blockbuster drug. This has necessitated a somewhat different, though related approach References pp. 291-292

272

Chapter 9

to drug analysis, in this case to allow the efficient support of multiple, large-scale clinical studies, with a rapid turnaround of data. Whilst the demands on the analyst have been increasing, the tools to do the job have also undergone a radical change. Although high-performance liquid chromatography (HPLC) has continued to be the main analytical technique, the means of detection has shifted over the past decade. The use of mass-spectrometry (MS) detection with either single ion monitoring (SIM) or multiple reaction monitoring (MRM) using triple quadrupole (MS-MS) instruments has virtually replaced ultra-violet (UV) detection for bioanalysis in the laboratories of major pharmaceutical companies and contract laboratories. Most MS instruments used for quantitative bioanalysis employ either electrospray ionisation (ESI) or atmospheric pressure chemical ionisation (APCI). Both these approaches to molecular ionisation are considered 'soft' techniques in that they give very little fragmentation, thus giving very good sensitivity if the parent molecular ion is monitored.. Thus, the pharmaceutical analyst has not only had to adapt to a massively increased workload and faster turnaround times, but at the same time embrace a new detection technology. MS detection, used in the fight way, offers many advantages over UV and has allowed many of the analytical problems such as sensitivity, selectivity, throughput and speed of turnaround to be addressed. However, MS detection is far more complex and quite unlike UV in many respects and requires a different way of working. Despite this complexity and the relatively high cost of MS equipment, this approach has been so successful that not only has it replaced HPLC-UV analysis but it has impacted markedly on the use of immunoassay techniques, which for many years were the mainstay of clinical drug analysis. The following discussion will consider how analysts have embraced this new technology and have been able to meet the problem of increased demand for their services. This will be considered primarily from the perspective of the scientist in the drug discovery environment where many of the advances have been made, although problems in the drug development area will also be addressed. For convenience of discussion, five main approaches will be considered, these are: 9The elimination of method development through the use of genetic gradient HPLCMS methods 9The pooling of samples prior to analysis and the dosing of compounds as cocktails, both in vivo and in vitro 9The use of short columns with fast isocratic elution in clinical analysis 9The use of gradient elution with high flow rates 9The dilution of sample prior to introduction into HPLC to maximise sample loading. Needless to say all five approaches have their respective advantages and disadvantages which will be given due consideration. 9.2 THE USE OF GENERIC METHODS

In drug discovery, many compounds are only ever analysed once. If the outcome of that analysis indicates that the compound will be unsuitable as a drug or drug candidate the

HPLC-MS(MS) for bioanalysis in drug discovery and development

273

compound may never be analysed again. If a new method has to be developed for each compound or even compound series, then bioanalytical support for drug discovery would be highly inefficient. It is necessary therefore to eliminate the method development phase and the application of genetic methods, which can be successfully applied to a wide range of compounds, is a particularly attractive option. The ideal generic method should involve a single column and eluent and be capable of chromatographing a wide range of compounds with a high degree of success. The method obviously needs to be applicable to acids, bases and neutrals, furthermore, all compounds should chromatograph with good peak shape and efficiency and no particular compound type should be over or under retained. The resulting conditions, eluent make-up, flow rate, etc. should also be compatible with MS detection. Whilst this is a demanding list of attributes, through careful evaluation of all the essential parameters it is possible to develop a method that comes very close to this ideal. Not surprisingly, this is based on reversed-phase HPLC (RP-HPLC) with gradient elution and MS detection using electrospray ionisation. The discussion below will focus on the steps we have gone through in developing this methodology, starting with column selection and then moving onto eluent selection and optimisation. At all times it is necessary to consider not only the chromatographic performance but also how the chromatographic parameters affect MS performance. 9.2.1 Column selection

The first step in setting up any method is the selection of a suitable stationary phase or column. Our approach has been to determine the efficiency, symmetry and column retentivity with a large diverse set of test probes [ 1]. This test set has consisted primarily of bases since these have tended to be the most difficult to chromatograph successfully. These include both strong and weak bases, with hindered and unhindered nitrogens and compounds with multiple basic centres. The situation with regard to the chromatography of bases has certainly improved in recent years and modern base-deactivated materials based on high purity silicas give much improved performance compared to the materials used a decade ago. The ultimate aim of the evaluation work is to identify unequivocally column(s) which give good all-round performance with neutral, acidic and basic solutes. Whilst it is easy to generate the data, in order to obtain a clear outcome from the work, careful manipulation and presentation of the data is essential. The easiest way to grasp this approach is to consider the peak asymmetry [1]. The observed asymmetry (ASobs) for any compound can be considered to be made up of two components. Firstly, the intrinsic asymmetry of the column (Asidoa~) which amongst other things is a function of how well the column has been packed. And secondly, As~i~, which is the contribution to the asymmetry due to the silanol interactions, although other factors such as interactions with metals may also be involved. These parameters are related as shown below. ASobs = AsideaI + Assil Since ASobs is obtained by measurement and Asidea~is obtained from the mean asymmetry shown by a set of simple neutral analytes (phenol, anthracene, toluene, etc), it is thus

References pp. 291-292

274

Chapter 9

possible to extract the contribution to asymmetry due to the silanols interacting with the test analyte. In this way it is thus possible to focus the evaluation of the column on the quality of the stationary phase (with respect to the interaction with bases) and not how well the column has been packed. It is however important to consider the mean AS~dea~ and if this is high i.e. > 1.5 then consideration needs to be given to whether the column is suitable for evaluation. Rather than just creating frequency histograms for each column showing the distribution of As~, the comparison of which can be very subjective, the use of cumulative frequency histograms are recommended. Figure 9.1 shows a plot of cumulative frequency against raw As~ for 4 different columns evaluated previously [ 1]. As well as allowing good visual assessment of the data and clear comparison between columns, it is also possible to extract a single quantifiable parameter that gives an overall measure of the properties of each stationary phase. For instance, the percentage of compounds giving a predefined As~, or the As~ shown by 90% of the compounds studied (As~ 90%) can be easily read off the plots. These parameters make the comparison of column performance with a range of test solutes more objective and measurable. As measured by As,, 90%, the performance of the columns decreases in the order IV > II> III> I. Although the outcome is clear using the cumulative frequency plots (Fig. 9.1) the conclusion is ambiguous if the histograms of the raw As or As~ are merely considered. Figure 9.2 shows the four histograms for As~, where it is clearly more difficult to distinguish between columns IV, II and III. Hopefully, the conclusions are clear: use a wide range of test compounds, extract the contribution due to the silanols (As~j) and present the data in the form of cumulative frequency plots to obtain valid conclusions with regard to the best overall stationary phase.

loo.o% 9o.o%

80.0%

oc -

70.0%

o

60.0%

=

0"

>

Column I

50.0%

- I - - Column II

40.0%

Column III

_~

30.0%

"5 o

20.0%

)(

Column IV

10.0% .0% 0

2

4

6

8

10

12

As sil

Fig. 9.1. Cumulative frequency profiles for A~ determined with four columns using a panel of 24 test compounds.

275

H P L C - M S ( M S ) f o r bioanalysis in drug discovery and development

"r fA m

e~

U)

,
, r e,3

tr3 r r

%

314

Chapter 10

Time (min) 60-

40-

20-

0

-60.0

"

"

'"

60

"

I

-60.5

"

"

"

"

"

I

-61.0

.

.

.

.

I

-61.5

.

.

.

.

I

-62.0

.

.

.

.

ppm

Fig. 10.5. 470 MHz continuous flow JgF NMR-detected HPLC chromatogram from rat urine following administration of 2-bromo-4-trifluoromethylaniline.The J9F NMR chemical shift is along the horizontal axis, the HPLC retention time is on the vertical axis and the eluting peak intensity is shown in a contour plot.

studies have been performed on 4-bromoaniline [90]. H P L C - N M R - M S with concomitant radioactivity detection has been used to investigate the metabolism of ~3C-acetyl labelled [3-blocker practolol [91 ]. Whilst the bulk of the reported studies have involved the use of either tH or 19FH P L C - N M R , some limited work has been done using other nuclei. Thus, some preliminary work has been carried out to determine the possibility of using 2H NMR spectroscopic detection in HPLC-NMR. To this end, the metabolism of dimethylformamide-dT, DMF-d7, in the rat was investigated using 2H NMR spectroscopy [92]. 2H NMR detection of the deuterated metabolites from D M F - d 7 in rat urine was attempted in continuous-flow H P L C - N M R [14]. These experiments were carried out at a 2H observation frequency of 92.1 MHz (which corresponds to ~H at 600 MHz). Based on chemical shifts and the known metabolism of DMF-d7, the parent compound was the major drug-related material present. The next two most abundant species had very similar retention times, with each containing one type of methyl group. One of these arises from dimethylamine-d6 whilst the other was one rotational form of Nhydroxymethyl-N-methylformamide-d6. In addition, one peak was seen which was detected in the 0-8 h urine only and which was not then assigned. This compound also shows a formyl deuteron resonance and possibly arises from a demethylated product such as N-methylformamide-d4. It is clear that 2H NMR spectroscopy is not likely to be of major use in drug metabolism, but because the 2H nucleus has a rapid NMR relaxation time, data can be acquired rapidly and for small molecules the linewidths are reasonably sharp. This means that for equal numbers of nuclei (i.e. the same level of isotopic substitution), it has approximately the same sensitivity as ~3C NMR spectroscopy although the chemical shift range and hence spectral dispersion is much higher for the latter. As well as deuterium 3~p NMR spectroscopy offers useful possibilities for H P L C NMR even though relatively few drugs contain phosphorus. Probably the most

Biomedical applications of directly-coupled chromatography

315

(IV) a)

'

I

'

'

F

-60.5

'

-'

'

'

- 61.0

I

'

'

'

- 61.5

b)

I

''"

'

- 62.0

1

'

'

'

- 62.5

1

ppm

HI )O

4j = 2 Hz t

3

J 7.55

"-

~

'

I

7.5

'

'

~

'

1

.

.

.

.

7.50

1

7.0

.

.

.

7.45

.

[

6.5

ppm

.

.

.

solvent impurity

.

6.0

[

.

.

.

.

I

5.5

.

.

.

.

I

5.0

.

.

.

.

I

4.5

.

.

.

.

I

4.0

'

'

ppm

CF.,

[M-D]1oo

.~.~~

3

336 338

% ^.

0 .... , .... , .Trm,19,8211, . . . . . . . . . . . . . . . ,r,,

255-25~"258 ' 29~.-294 .

. I' . "1 . .... . "[")'. 141 . ' ''

!.

.

.

. ~l . ........

339

a~],lh/ ~! ....

I ....

I ....

3583833.85 1'"1"~;1""1

....

416~,!8 I " ~ l " " r

....

I ....

438 ! ....

1 ....

458 I ....

~ ....

476 I ....

I ....

496 I'tI'1

Fig. 10.6. NMR and MS data for the major metabolite of 2-bromo-4-trifluoromethylaniline, namely 2-amino3-bromo-5-trifluoromethylphenyl sulfate, in rat urine obtained by stop-flow H P L C - N M R - M S . (a) 470 MHz J')F NMR spectrum, (b) 500 MHz ~H NMR spectrum showing resolution-enhanced expansion of the region of the aromtic protons (* denotes unresolved meta spin coupling), and (c) negative ion electrospray MS of the molecule with all exchangeable hydrogens replaced by deuterium.

References pp. 325-329

316

Chapter 10

important class are those related to cyclophosphamide which are used in chemotherapy. The presence of phosphorus in such a molecule does provide the opportunity for using 31p NMR as a specific method for detection of metabolites. The toxicity of the anticancer drug ifosfamide has been studied using extracts from the urine of patients on ifosfamide therapy [15] and this study was combined with investigations of ifosfamide metabolism in the rat. In this case, using urine samples which had been freeze-dried and reconstituted at a 12-fold concentration, it was possible to locate the drug-related substances using 3~P-detected HPLC-NMR in the continuous flow mode and subsequently to use stop-flow methods to characterise the metabolites using ~H NMR. Although 31p NMR spectroscopy is much less sensitive than ~H or 19F NMR spectroscopy, it proved useful for the identification of the chromatographic retention times of phosphorus-containing species in this study, particularly since the metabolites of ifosfamide have poor UV absorption characteristics. Under the HPLC conditions used, the parent drug and three metabolites were identified, ifosforamide mustard, 4-hydroxyifosfamide and 2-de(chloroethyl)ifosfamide.

10.8.4 Application to in vitro metabolism studies In vitro techniques, such as the use of tissue slices and cell suspensions, are being used

increasingly to obtain insights into the metabolism of new compounds in both animals and man. Directly coupled 750 MHz HPLC-~H NMR spectroscopy has been applied to the characterisation of low level metabolites of 3-nitro-2-(2-fluorophenoxy)pyridine and 3-amino-2-(2-fluorophenoxy)pyridine produced by rat microsomes [93]. In stop-flow HPLC-NMR mode, the direct injection of microsomal extracts enabled the separation and characterisation of minor metabolites. Unequivocal identification of the metabolites was achieved without the use of radiolabel or synthetic standards. GW1370U87, 1-ethyl-phenoxathiin-10,10-dioxide, was intended for use as a monoamine oxidase-A inhibitor and its metabolism has been studied in human liver microsomes [94]. The supernatant from the cells was collected, concentrated and examined using 600 MHz ~H HPLC-NMR in stop-flow mode. The UV-detected chromatogram was uncomplicated with most, but not all, of the UV peaks being due to GW1370U87 or its metabolites. Chromatography was stopped at the top of each peak and a IH NMR spectrum obtained. In all, six GW1370U87-related HPLC peaks were characterised. In addition, the metabolism of the multi-drug resistance inhibitor LY335979 [82] and 7-ethoxycoumarin [95] have been studied in human liver microsome incubations.

10.8.5 Application to drug metabolite reactivity Many drugs containing carboxylate groups form [3-1-O-acyl glucuronides as major metabolites. Such ester glucuronides are potentially reactive due to the susceptibility of the acyl group to nucleophilic reactions and they can undergo hydrolysis, acyl migration

Biomedical applications of directly-coupled chromatography

317

and covalent adduct formation. The acyl migration reactions result in positional isomers and anomers as shown below and these may be reactive towards serum proteins with toxicological consequences. The acyl group migrates successively to the 2-, 3- and 4-hydroxyl groups of the glucuronic acid moiety, thereby allowing the formation of both oL- and [3-anomers of the positional isomers (see Fig. 10.7). Synthetic fluorobenzoic acid and trifluoromethylbenzoic acid glucuronide conjugates were chosen as model compounds of carboxylate group-containing drugs and an HPLC method has been developed for the simultaneous determination of the 1-, 2-, 3- and 4-positional isomers of the acyl glucuronides, and their c~- and [3-anomers for 2-, 3- and 4-fluorobenzoic acids together with the aglycones formed via hydrolysis. A typical result is shown in Fig. 10.8 which depicts the continuous flow 750 MHz ~H HPLCNMR characterisation of the glucuronides from an equilibrium mixture of transacylated glucuronides of 4-fluorobenzoic acid, measured in the continuous flow mode [96]. The ~H NMR frequency is on the horizontal axis and the chromatographic retention time is on the vertical axis. Each of the glucuronide isomers is eluted separately and can be identified from its NMR spectrum. It has been noted that in general the elution order of transacylated glucuronides is [3-4-O-acyl-, oL-4-O-acyl-, oL-3-O-acyl-, [3-3-O-acyl-, [32-O-acyl- and oL-2-O-acyl- irrespective of the nature of the carboxylic acid-containing moiety. This directly-coupled HPLC-NMR method has been used to investigate the acyl migration kinetics of individual isomers of 2-, 3- and 4-fluoro-, and 2- and 3-trifluoromethylbenzoyl-D-glucopyranuronic acid separated from an equilibrium mixture of the [3-1-O-acyl isomer, the e~- and [3-2-O-acyl isomers, the oL- and [3-3-O-acyl isomers and the oL- and [3-4-O-acyl isomers at pH 7.4 and 25~ [96-99]. Both continuous-flow HPLC-NMR at 750 MHz and stop-flow methods have been used. For detailed kinetic studies, each isomer was separated using reversed-phase HPLC and then led into an NMR flow probe in a 600 MHz NMR spectrometer. The flow was stopped and sequential ~H NMR spectra collected, thus allowing the direct observation of the appearance of the glucuronide positional isomers of that particular glucuronide isomer which had been isolated. This is illustrated in Fig. 10.9 which shows the build-up of other products following the introduction of the [3-4-O-acyl-glucuronide of 2-fluorobenzoic acid into the NMR probe after HPLC separation. The rate constants for the decomposition of the various isomers were determined and the acyl migration reactions were simulated using a mathematical model of the kinetics of the glucuronide rearrangement (incorporating 9 first-order rate constants determining acyl migration reactions and 6 first-order rate constants describing the mutarotation of the 2-, 3- and 4-positional isomers. The acyl migration of the glucuronide metabolite of the model drug 6,11 dihydro11-oxo-dibenz (b,e) oxepin-2-acetic acid has also been investigated in pH 7.4 buffer using and urine directly-coupled 600 and 750MHz stop-flow HPLC-~H NMR spectroscopy [100,101]. Other substances studies using this approach include the glucuronide of a novel retinoid known as CD271 [102], the glucuronides of enantiomeric 2-phenyl propionic acids [103] and S-naproxen glucuronide [104,105]. More recently, the presence of the oL-1-O-acyl isomer has been confirmed demonstrating that the back reaction from the oL-2-O-acyl isomer is possible. This is References pp. 325-329

Chapter 10

318 glucuronic acid HO HOOC

O

HO HOOC

0

+ RCOOH

OH 13-1-O-acyl

13

0 HO HOOC

O

OH

1L HO HOOC

~o.

HO HOOC,

O

R

C

cz-2-O-acyl

HO HOOC

O

~ 13-3-O-acyl

HOOC,

OH

O

OH OH

RCO0

O

OCOR

13-2-O-acyl

HO HOOC

~

0

(~-3-O-acyl

RCOO HOOC,

I OH

O

OH 13-4-O-acyl

(z-4-O-acyl

Fig. 10.7. The reaction scheme depicting acyl migration in glucuronide conjugates.

OH

Biomedical applications of directly-coupled chromatography

319

._o C p

('Ulm) emil UOlleJedes olqdeJw Ill ,p.

0 ,r--

9 ~D

O

u') o

!

.

.

,

;

~

-~

.....

t Q~,,

~N . ,..~

,.o 9 9

ea.

e~

~

4~ 9 ~D

9

N

9

N

zz

9

~)

-

|

8

~ ~

ZZ N ZZ

0 tt~

~D ,.a

0 |

References pp. 325-329

l

r

(=) 3-.o-a,'~

,

w

!

(P)

i,

,

5.1

i

i

II

,

I

)~

!

I

i tt

r

!:!I' I;

2,2',([~)

,

,

(~p,) 2-O-acyl

4,4',((x, i3) II

,

r

I

, l

i

4.6

I

'

'

,

'

''

/

3,5',(ct, p)

4.

(o0 3-o-acyl

]

,

II

. 4,5',(=)

'

II

Ox) 2-O-acyl

|

~

I

'

.

d

'

i

]

3.6

i

,

....

4,2',(I])

~O-aWl

(=,p) 2,0-acyl

2,4',(~p)

4,2~,(o0

4,5',(~) 4,3',(p) . 2 4 3' (a) I h '7' ,--.'z3'4"((xJ' ' 'i Ir"-'t"1 2,s',(p)

~--

!

ppm

i

2O

(Hours)

3O Reaction time domain

Fig. 10.9. 600 MHz ]H NMR spectrum obtained in stop-flow mode after the [3-4-O-acyl-glucuronide of 2-fluorobenzoic acid had been isolated in the NMR flow cell. This shows the successive formation of 3-O-acyl and 2-O-acyl isomers with time.

(~

i

I

3,3',((x)

3,1',(=)I

4,1',(o0 I

"~

Biomedical applications of directly-coupled chromatography

321

reflected in the reaction scheme shown in Fig. 10.7 and increases the number of kinetic parameters to be determined [ 106]. It is clear that directly-coupled HPLC-NMR spectroscopy offers a unique analytical approach to obtain structural information of inter-converting compounds in a complex mixture of isomers. This method will be of value in the elucidation of the reactivity of drug glucuronides in terms of acyl migration and enable an investigation of the potential for protein binding. Furthermore, this HPLC-NMR approach to the study of glucuronide acyl migration reactions allows unique kinetic information to be obtained relating to glucuronide reactivity and this approach will be useful in future structureactivity studies on the toxicity of drug ester glucuronides. The metabolism of paracetamol proceeds through a reactive intermediate known as NAPQI which can react with glutathione at the 2, 3, or ipso carbons. This reaction has been investigated using directly-coupled HPLC-NMR at 500 MHz [107]. The two reactants, (NAPQI and glutathione, GSH) were mixed and the product mixture separated using directly-coupled HPLC-NMR. In fact all three isomers were shown to be produced with the ipso derivative being most abundant and the 2'-isomer the least. By holding the ipso-NAPQI-GSH adduct in the flow probe of the NMR spectrometer for one hour the rate at which it decomposed to the other isomers and other species could be monitored.

10.9 FUTURE DEVELOPMENTS 10.9.1 Automation and informatics

While this review demonstrates the wide applicability of HPLC-NMR achieved thus far, it is possible to identify areas where rapid advances are to be expected that will further increase the utility of HPLC-NMR. One area for example, is automation, where the ability for improved analysis times in coupled separation-NMR techniques will be a major factor in enabling the more widespread application of the technology. Indeed, it is not difficult to envisage a time when HPLC-NMR combined with in-line HPLC-MS will, automatically, provide the complete qualitative and quantitative metabolic fate of xenobiotics in a single chromatographic run. In order to analyse and extract all information from the huge amounts of data likely to be generated from such systems, novel methods of data analysis are also required. Pattern recognition for example has been combined with flow injection NMR spectroscopy to provide analysis of high throughput biochemical screening [108]. Principal components analysis (PCA) has also recently been applied to HPLC-NMR to differentiate between regio-isomers in a less than optimal HPLC separation [109]. By applying PCA to the on-flow pseudo-2D HPLC-NMR spectrum, it was possible to determine the number and identity of 3 unresolved isomers. In another study, ~H NMR spectra of rat urine samples from subpopulations of normal laboratory rats showed characteristic spectral differences when analysed using PCA. This statistical approach allowed the direction of HPLC-NMR experiments for identification of the endogenous species responsible for the differences [110].

References pp. 325-329

322

Chapter 10

Other new technical developments are also occurring which, in the foreseeable future, will provide greatly increased NMR sensitivity, reduced solvent usage or additional analytical information. These developments include the use of higher magnetic field strengths and hence observation frequencies. In addition the development of NMR probes and preamplifiers cooled with cryogenic liquids will provide lower detection limits and higher sensitivities to a degree surpassing any arising from increases in magnetic field [ 111 ]. It is also possible to envisage new areas of application for HPLC-NMR-MS, and for example recently HPLC-NMR spectroscopy was used to separate and characterise lipoproteins fractions from human blood plasma [112].

10.9.2 Miniaturisation in separations coupled to NMR Miniaturisation of analytical equipment is occurring with respect to both the NMR detection systems, and the initial chromatographic mode. The improvement of methods for the rapid analysis of multiple samples has led to the development of NMR probes containing multiple detection coils [113]. By placing up to four coils within the same cell, such 'Multiplex NMR' significantly reduces sample analysis time. Different approaches exist for the detection of each sample, and more importantly, the elimination of cross-talk [ 114,115]. Microcoil probes have also been demonstrated for use in HPLC-NMR applications [ 116]. The union of capillary HPLC with microcoil NMR has been demonstrated for the detection of terpenoids. With a 1.1 ~1 observation volume, it was possible to detect 37 ng of oL-pinene [117]. This approach with its consequent low solvent usage allows fully deuterated NMR solvents to be employed at reasonable cost and alleviates the solvent suppression problems referred to earlier. It has been found that detection of low nanogram quantities of material can be achieved in 3-4 minutes under stop-flow conditions and 2-dimensional NMR spectra are therefore possible also [116,118]. It is expected that new eluent systems for HPLC which are advantageous for NMR and MS detection will be developed and some preliminary results have been given on the use of superheated D20 in this respect [ 119,120]. Capillary electrophoresis (CE) coupled to NMR has been shown to be a very powerful addition to the armoury of analytical methods. The technique is very simple experimentally, with all that is required being a length of fused silica capillary with an optical window to enable detection, a detector (UV, fluorescence or mass spectrometry), a high voltage source, two electrode assemblies and buffer solutions in suitable reservoirs. The technique has been shown to provide very high separation efficiencies but the small injection volume (a few nl) means that high sensitivity can only be achieved if concentrations of the analyte in the sample are high. The use of NMR spectroscopy for detection in CE has been demonstrated [121-124]. This has an active volume of ~-5 nl and limits of detection using ~H NMR in the ng range for acquisition periods of the order of 1 minute [125]. The limit of detection in concentration terms is about that of HPLC-NMR but in mass terms represents about two orders of magnitude less.

Biomedical applications of directly-coupled chromatography

323

min D

1

60

III

(!1i) II

(ll)

It

~l)

t

t,i t'

30

8.0

6.0

4.0

~pm

Fig. 10.10. 600 MHz on-flow CEC-NMR spectrum of paracetamol metabolites separated from an extract of human urine. The use of CE-NMR, and the related technique CEC-NMR at an observation frequency of 600 MHz, has been applied to the detection and characterisation of paracetamol metabolites found in human urine [126,127], a crude synthesised dinucleotide and its related by-products [ 128], and aspartame and caffeine [129]. As an illustration, the on-flow CEC-NMR spectrum of paracetamol metabolites from a human urine extract sample is shown in Fig. 10.10. The sample was a solid phase extract of urine from a human volunteer after administration of 500 mg of paracetamol. The 600 MHz NMR spectrum was acquired with 8 scans per row (i.e. about 10 s acquisition time) and the contours seen correspond to the two major metabolites, namely the glucuronide and sulfate conjugates of paracetamol. Also visible is the spectrum of the endogenous species hippurate. Extracted individual rows corresponding to these three substances from this on-flow NMR-detected separation are shown in Fig. 10.11. This on-flow detection is of about 300 ng of paracetamol glucuronide [ 126,127]. A more recent application is the use of capillary isotachophoresis (cITP) prior to NMR detection [130]. cITP is a sample focusing method. It uses a leading and terminating electrolyte, where the leading electrolyte (with a high electrophoretic mobility) forms the front zone and the terminating electrolyte (with a low electrophoretic mobility) the rear zone, with the sample in between the zones. When an electric field is applied, the components separate into discrete bands, with the sample components focused as a function of the ion concentration of the leading electrolyte. The work demonstrated a 100-fold increase in NMR signal-to-noise ratio when comparing non-focused samples with those that were focused using cITE

10.9.3 Hypernation Further hyphenation of other spectroscopic techniques for analyte identification ('hypernation') such as infra-red spectroscopy has recently been demonstrated [131].

References pp. 325-329

Chapter 10

324

HOD CH2 H4

(c)

H2/6

H3/5

Hippurate

NCH 3 Sulfate conjugate (b)

[ NCH 3 H2/6

H3/5

,

I

H2, H4,

i Glucuronide conjugate

(a)

'........8.0" .... 7 ' I 0 ' i ~ 0 " 5 . 0 "

. . . . . . . .

4'.0

I

. . . . . . . .

310

1 " ' I ' 1

. . . .

I

ppm

Fig. 10.11. Single rows extracted from the on-flow chromatogram shown in Fig. 10.10. The identification of the two paracetamol metabolites and the endogenous substance hippurate are as shown. For the paracetamol metabolites, H2/6 are the aromatic protons ortho to the N-acetyl group, H3/5 are meta. H1 '-H5' denote the protons of the glucuronide ring and NCH3 denotes the N-acetyl resonance. For hippurate the numbering scheme has the side-chain at C 1.

H P L C - U V - I R - N M R - M S was employed for the analysis of a mixture of NSAIDs, allowing almost complete structural characterisation to be performed [132]. A time of flight (TOF) mass spectrometer was used in this system enabling accurate mass (and therefore atomic composition) to be obtained. While such a system undoubtedly provides a large amount of structural information however, the degree of complexity of such a system increases rapidly with the addition of each new technique. It is necessary, for example, to ensure solvents (and modifiers) are compatible with all techniques [ 133]. A slightly modified system using flow injection rather than HPLC along with UV, IR, N M R and MS has also recently been reported [134], and the same system coupled to superheated water chromatography has been applied to the analysis of a mixture of pharmaceuticals [135] and a series of plant extracts [136]. In addition, it is expected that publications on the use of other types of mass spectrometric detection in H P L C - N M R - M S , such as time-of-flight (TOF) and ion-

Biomedical applications of directly-coupled chromatography

325

cyclotron r e s o n a n c e MS, w h i c h allow accurate masses and h e n c e e m p i r i c a l m o l e c u l a r f o r m u l a e to be d e t e r m i n e d , will be f o r t h c o m i n g

10.10 CONCLUSIONS

The progress that has b e e n m a d e in the coupling of H P L C and related techniques with N M R s p e c t r o s c o p y over the last decade has b e e n very rapid. The n u m e r o u s applications of the t e c h n i q u e in b i o m e d i c a l analysis are e v i d e n c e of its utility and the n u m b e r of reports of the use of H P L C - N M R in this area continues to increase m a r k i n g the transition of H P L C - N M R f r o m a research technique to a routine analytical m e t h o d o l o g y . F u r t h e r d e v e l o p m e n t s leading to the use of m u l t i p l e h y p h e n a t i o n (hypernation) with H P L C - N M R - M S and h i g h e r concatenations have also b e g u n to generate applications. With further t e c h n o l o g i c a l advances in the areas of miniaturised flow probes, for use with capillary separations, and c r y o p r o b e s leading to l o w e r sample r e q u i r e m e n t s the usefulness of N M R c o u p l e d to separations will u n d o u b t e d l y increase.

10.11 REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

J.C. Lindon, J.K. Nicholson and I.D. Wilson, Prog. NMR Spectrosc., 29 (1996) 1. K. Albert, J. Chromatogr. A, 703 (1995) 123. S.H. Smallcombe, S.L. Patt and EA. Keifer, J. Magn. Reson., 117 (1995) 295. M. Liu, X. Mao, C. Ye, H. Huang, J.K. Nicholson and J.C. Lindon, J. Magn. Reson., 132 (1998) 125. K. Albert, U. Braumann, L.-H. Tseng, G. Nicholson, E. Bayer, M. Spraul, M. Hofmann, C. Dowle and M. Chippendale, Anal. Chem., 66 (1994) 3042. L. Griffiths, Anal. Chem., 67 (1995)4091. L. Griffiths, Magn. Reson. Chem., 35 (1997) 257. B.C. Sweatman, R.D. Farrant, EN. Sanderson, I. Philippe, S.R. Salman, J.K. Nicholson and J.C. Lindon, J. Magn. Reson. Anal., 1 (1995) 9. N. Wu, L. Webb, T.L. Peck and J.V. Sweedler, Anal. Chem., 67 (1995) 3101. B. Behnke, G. Schlotterbeck, U. Tallarek, S. Strohschein, L.-H. Tseng, T. Keller, K. Albert and E. Bayer, Anal. Chem., 68 (1996) 1110. T.D.W.Claridge, High-resolution NMR Techniques in Organic Chemistry, Pergamon Oxford, 1999. ES. Pullen, A.G. Swanson, M.J. Newman and D.S. Richards, Rapid Comm. Mass Spec., 12 (1998) 1732. S. Taylor, B. Wright, E. Clayton and I.D. Wilson, Rapid Comm. Mass. Spec., 12 (1998) 1732. R.D. Farrant, B.C. Cupid, J.K. Nicholson and J.C. Lindon, J. Pharmaceut. Biomed. Anal., 16 (1997) 1. EJ.D. Foxall, E.M. Lenz, J.C. Lindon, G.H. Neild, I.D. Wilson and J.K. Nicholson, Ther. Drug Monit., 18 (1996) 498. L. Griffiths and R. Horton, Mag. Res. Chem., 36 (1998) 104. J.E Shockcor, S.E. Unger, I.D. Wilson, EJ.D. Foxall, J.K. Nicholson and J.C. Lindon, Anal. Chem., 68 (1996) 4431. K.I. Burton, J.R. Everett, M.J. Newman, ES. Pullen, D.S. Richards and A.G. Swanson, J. Pharm. Biomed. Anal., 15 (1997) 1903.

326

Chapter 10

19

R.M. Holt, M.J. Newman, ES. Pullen, D.S. Richards and A.G. Swanson, J. Mass Spectr., 32 (1997) 64. G.J. Dear, R.S. Plumb, B.C. Sweatman, J. Ayrton, J.C. Lindon, J.K. Nicholson and I.M. Ismail, J. Chromatogr. B, 748 (2000) 281. J.C. Lindon, R.D. Farrant, EN. Sanderson, EM. Doyle, S.L. Gough, M. Spraul and M. Hofmann, Mag. Reson. Chem., 33 (1995) 857. J. Chin, J.B. Fell, M. Jarosinski, M.J. Shapiro and J.R. Wareing, J. Org. Chem., 63 (1998) 386. International Conference on Harmonisation (ICH 2), Document on Impurities in New Drug Substances. J.K. Roberts and R.J. Smith, J. Chromatogr. A, 677 (1994) 385. N. Mistry, I.M. Ismail, M.S. Smith, J.K. Nicholson and J.C. Lindon, J. Pharmaceut. Biomed. Anal., 16 (1997) 697. B.C.M. Potts, K.E Albizati, M.O. Johnson and J.P. James, Magn. Reson. Chem., 37 (1999) 393. S.X. Peng, B. Borah, R.L.M. Dobson, Y.D. Liu and S. Pikul, J. Pharmaceut. Biomed. Anal., 20 (1999) 75. W. Feng, H. Liu, G. Chen, R. Malchow, E Bennett, E. Lin, B. Pramanik and T. Chan, J. Pharmaceut. Biomed. Anal., 25 (2001) 545. C. Yang, Y.K. Si and W.Y. He, Chinese J. Anal. Chem., 29 (2001) 796. B.E. Segmuller, B.L.Armstrong, R. Dunphy and A.R. Oyler, J. Pharmaceut. Biomed. Anal., 23 (2000) 927. I.D. Wilson, L. Griffiths, J.C. Lindon and J.K. Nicholson, Efficient characterisation of minor components in pharmaceutical materials using HPLC-NMR and related hyphenated NMR methods, in: S. G6r6g (Ed.), Identification and Determination of Impurities in Drugs (pp. 299-322). Elsevier, Amsterdam, The Netherlands (2000). N. Mistry, A.D. Roberts, G.E. Tranter, E Francis, I. Barylski, I.M. Ismail, J.K. Nicholson and J.C. Lindon, Anal. Chem., 71 (1999) 2838. L.H. Tseng, U. Braumann, M. Godejohann, S.S. Lee and K. Albert, J. Chinese Chem., Soc 47 (2000) 1231. S. Johnson, E.D. Morgan, I.D. Wilson, M. Spraul and M. Hofmann, J. Chem. Soc. Perkin, 1 (1994) 1499. J.-L. Wolfender, K. Ndjoko and K. Hostettmann, Current Org. Chem., 2 (1998) 575. K. Hostettmann, O. Potterat and J.-L. Wolfender, Chimia, 52 (1998) 10. J.-L. Wolfender, K. Ndjoko and K. Hostettmann, Phytochem. Anal., 12 (2001) 2. K. Hostettmann and J.-L. Wolfender, Pesticide Sci., 51 (1997) 471. S. Rodriguez, J.-L. Wolfender, K. Hostettmann, H. Stoeckli-Evans and M.E Gupta, Helv. Chim. Acta, 81 (1998) 1403. J.-L. Wolfender, S. Rodriguez and K. Hostettmann, J. Chromatogr. A, 794 (1998) 299. E. Garo, J.-L. Wolfender, K. Hostettmann, W. Hiller, S. Antus and S. Mavi, Helv. Chim. Acta, 81 (1998) 754. J.R. Ioset, J.-L. Wolfender, A. Marston, M.E Gupta and K. Hostettmann, Phytochem. Anal., 10 (1999) 137. A. Cavin, O. Potterat, J.-L. Wolfender, K. Hostettmann and W. Dyatmyko, J. Nat. Products, 61 (1998) 1497. K. Ndjoko, J.-L. Wolfender, E. Roder and K. Hostettmann, Planta Med., 65 (1999) 562. G. Bringmann, M. Ruckert, W. Saeb and V. Mudogo, Magn. Reson. Chem., 37 (1999) 98. J. Schlauer, M. Ruckert, B. Wiesen, M. Harderich, L. Assi, R. Haller, S. Bar, K.-U. Frohlich and G. Bringmann, Arch. Biochem. Biophys., 350 (1998) 87. I.D. Wilson, E.D. Morgan, R. Lafont and B. Wright, J. Chromatogr. A, 799 (1998) 333. K. Albert, J. Chromatogr. A, 856 (1999) 199. T. Renukappa, G. Roos, I. Klaiber, B. Vogler and W. Krauss, J. Chromatogr. A, 847 (1999) 109. B. Vogler, I. Klaiber, G. Roos, C.U. Walter, W. Hiller, E Sandor and W. Krauss, J. Nat. Prod., 61 (1998) 175. S.H. Hansen, A.G. Jensen, C. Cornett, I. Bjornsdottir, S. Taylor, B. Wright and I.D. Wilson, Anal. Chem., 71 (1999) 5235.

20 21 22 23 24 25 26 27 28 29 30 31

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51

Biomedical applications of directly-coupled chromatography 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81

327

A. Lommen, M. Godejohann, D.P. Venema, EC.H. Hollman and M. Spraul, Anal. Chem., 72 (2000) 1793. I.D. Wilson, E.D. Morgan, R. Lafont, J.P. Shockcor, J.C. Lindon, J.K. Nicholson and B. Wright, Chromatographia, 49 (1999) 374. G. Bringmann, K. Messer, M. Wohlfarth, J. Kraus, K. Dumbuya and M. Ruckert, Anal. Chem., 71 (1999) 2678. G. Bringmann, M. Wohlfarth, H. Rischer, M. Heubes, W. Saeb, S. Diem, M. Herderich and J. Schlauer, Anal. Chem., 73 (2001) 2571. S.C. Bobzin, S. Yang and T.E Kasten, J. Chromatogr. B, 748 (2000) 259. M. Sandvoss, L.H. Pham, K. Levsen, A. Preiss, C. Mugge and G. Wunsch, Eur. J. Org. Chem., 7 (2000) 1253. M. Sandvoss, A. Weltring, A. Preiss, K. Levsen and G. Wuensch, J. Chromatogr. A, 917 (2001) 75. C.B.L. Abel, J.C. Lindon, D. Noble, B.A.M. Rudd, EJ. Sidebottom and J.K. Nicholson, Anal. Biochem., 270 (1999) 220. R.J.P. Cannell, T. Rashid, I.M. Ismail, EJ. Sidebottom, A.R. Knaggs and ES. Marshall, Xenobiotica, 27 (1997) 147. G.B. Scarfe, B. Wright, E. Clayton, S. Taylor, I.D. Wilson, J.C. Lindon and J.K. Nicholson, Xenobiotica, 28 (1998) 373. I.D. Wilson, J.K. Nicholson, M. Hofman, M. Spraul and J.C. Lindon, J. Chromatogr., 617 (1993) 324. M. Hofmann, P. Dvortsak, J.K. Nicholson and I.D. Wilson, Anal. Chem., 65 (1993) 327. E. Clayton, S. Taylor, B. Wright and I.D. Wilson, Chromatographia, 47 (1998) 264. M. Spraul, M. Hofmann, I.D. Wilson, E. Lenz, J.K. Nicholson and J.C. Lindon, J. Pharmaceut. Biomed. Anal., 11 (1993) 1009. J.P. Shockcor, S.E. Unger, E Savina, J.K. Nicholson and J.C. Lindon, J. Chromatogr. B, 748 (2000) 269. J.R. Bales, P.J. Sadler, J.K. Nicholson and J.A. Timbrell, Clin. Chem., 30 (1984) 1631. J.R. Bales, J.K. Nicholson and EJ. Sadler, Clin. Chem., 31 (1985) 757. M. Spraul, M. Hofmann, J.C. Lindon, J.K. Nicholson and I.D. Wilson, NMR in Biomedicine, 7 (1994) 295. A.W.Nicholls, R.D. Farrant, J.E Shockcor, S.E. Unger, I.D. Wilson, J.C. Lindon and J.K. Nicholson, J. Pharmaceut. Biomed. Anal., 15 (1997) 901. U . G . Sidelmann, U. Braumann, M. Hofmann, M. Spraul, J.C. Lindon, J.K. Nicholson and S.H. Hansen, Anal. Chem., 69 (1997) 607. U . G . Sidelmann, I. Bj~rnsdottir, J.P. Shockcor, S.H. Hansen, J.C. Lindon and J.K. Nicholson, J. Pharmaceut. Biomed. Anal., 24 (2001) 569. L.W. Frick, L. St John, L.C. Taylor, G.R. Painter, P.A. Furman, D.C. Liotta, E.S. Furfine and D.J. Nelson, Antimicrob. Agents & Chemother., 37 (1993) 2285. J.E Shockcor, R.W. Wurm, L.W. Frick, P.N. Sanderson, R.D. Farrant, B.C. Sweatman and J.C. Lindon, Xenobiotica, 26 (1996) 189. J. Skilling, Quantified Maximum Entropy, in: EE Fougere (Ed.), Maximum Entropy and Bayesian Methods, Kluwer, Dordrecht, The Netherlands, 1990. A.W. Nicholls, S.T. Caddick, I.D. Wilson, R.D. Farrant, J.C Lindon and J.K. Nicholson, Biochem. Pharmacol., 49 (1995) 1155. A.W. Nicholls, J.C. Lindon, R.D. Farrant, J.P. Shockcor, I.D. Wilson and J.K. Nicholson, J. Pharmaceut. Biomed. Anal., 20 (1999) 865. A.E. Mutlib, J. Shockcor, S.Y. Chen, R.J. Espina, D.J. Pinto, M.J. Orwat, S.R.Prakash and L.S. Gan, Chem. Res. Tox., 15 (2002) 48. A. Mutlib, J. Shockcor, S.Y. Chen, R. Espina, J.R. Lin, N. Graciani, S. Prakash and L.S. Gan, Drug Metab. Disp., 29 (2001) 1296. A.E. Mutlib, J. Shockcor, R. Espina, N. Graciani, A. Du and L.S. Gan, J. Pharmacol. Exp. Therapeut., 294 (2000) 735. A.E. Mutlib, J.T. Strupczewski and S.M. Chesson, Drug Metab. Disp., 23 (1995) 951.

328

Chapter 10

82

W.J. Ehlhardt, J.M. Woodland, T.M. Baughman, M. Vandenbranden, S.A. Wrighton, J.S. Kroin, B.H. Norman and S.R. Maple, Drug Met. Dispos., 26 (1998) 42. A.E. Mutlib, S. Diamond, J. Shockcor, R. Way, G. Nemeth, L. Gan and D.D. Christ, Xenobiotica, 30 (2000) 1091. M. Dachtler, H. Handel, T. Glaser, D. Lindquist, R.M. Hawk, C.N. Karson, R.A. Komoroski and K. Albert, Magn. Reson. Chem., 38 (2000) 951. G.J. Dear, J. Ayrton, R. Plumb, B.C. Sweatman, I.M. Ismail, I.J. Fraser and EJ. Mutch, Rapid Comm. Mass Spectr., 12 (1998) 2023. G.J. Dear, R.S. Plumb, B.C. Sweatman, P.S. Parry, A.D. Roberts, J.C. Lindon, J.K. Nicholson and I.M. Ismail, J. Chromatogr. B, 748 (2001) 295. G.B. Scarfe, B. Wright, E. Clayton, S. Taylor, I.D. Wilson, J.C. Lindon and J.K. Nicholson, Xenobiotica, 29 (1999) 77. G.B. Scarfe, J.C. Lindon, J.K. Nicholson, B. Wright, E. Clayton and I.D. Wilson, Drug Metab. Disp., 27 (1999) 1171. G.B. Scarfe, E. Clayton, I.D. Wilson and J.K. Nicholson, J. Chromatogr. B, 748 (2000) 311. J.K. Nicholson, J.C. Lindon, G.B. Scarfe, I.D. Wilson, E Abou-Shakra, A.B. Sage and J. Castro-Perez, Anal. Chem., 73 (2001) 1491. G.B. Scarfe, J.C. Lindon, J.K. Nicholson, E Martin, B. Wright, S. Taylor, E. Lenz and I.D. Wilson, Xenobiotica, 30 (2000) 717. R.D. Farrant, S.R. Salman, J.C. Lindon, B.C. Cupid and J.K. Nicholson, J. Pharmaceut. Biomed. Anal., 11 (1993) 687. O. Corcoran, M. Spraul, M. Hofmann, I.M. Ismail, J.C. Lindon and J.K. Nicholson, J. Pharmaceut. Biomed. Anal., 16 (1997) 481. J.E Shockcor, I.S. Silver, R.W. Wurm, P.N. Sanderson, R.D. Farrant, B.C. Sweatman and J.C. Lindon, Xenobiotica, 26 (1996) 41. M.B. Fisher, D. Jackson, A. Kaerner, S.A. Wrighton and A.G. Borel, Drug Metab. Disp., 30 (2002) 270. U. Sidelmann, C. Gavaghan, H.A.J. Carless, R.D. Farrant, J.C. Lindon, I.D. Wilson and J.K. Nicholson, Anal. Chem., 67 (1995) 3401. U.G. Sidelmann, S.H. Hansen, C. Gavaghan, H.A.J. Carless, J.C. Lindon, R.D. Farrant, I.D. Wilson and J.K. Nicholson, Anal. Chem., 68 (1996) 2564. U.G. Sidelmann, C. Gavaghan, H.A.J. Carless, M. Spraul, M. Hofmann, J.C. Lindon, I.D. Wilson and J.K. Nicholson, Anal.Chem., 67 (1995) 4441. U.G. Sidelmann, A.W. Nicholls, E Meadows, J. Gilbert, J.C. Lindon, I.D. Wilson and J.K. Nicholson, J. Chromatogr. A, 728 (1996) 377. U. Sidelmann, E.M. Lenz, EN. Sanderson, M. Hofmann, M. Spraul, J.C. Lindon, I.D. Wilson and J.K. Nicholson, Anal. Chem., 68 (1996) 106. E.M. Lenz, D. Greatbanks, I.D. Wilson, M. Spraul, M. Hofmann, J.C. Lindon, J. Troke and J.K. Nicholson, Anal. Chem., 68 (1996) 2832. R. Ruhl, R. Thiel, T.S. Lacker, S. Strohschein, K. Albert and H. Nau, J. Chromatogr. B, 757 (2001) 101. K. Akira, H. Hasegawa, Y. Shinohara, M. Imachi and T. Hashimoto, Biol. & Pharmaceut. Bull., 23 (2000) 506. R.W. Mortensen, O. Corcoran, C. Cornett, U.G. Sidelmann, J. Troke, J.C. Lindon, J.K. Nicholson and S.H. Hansen, J. Pharmaceut. Biomed. Anal., 24 (2001) 477. R.W. Mortensen, O. Corcoran, C. Cornett, U.G. Sidelmann, J.C. Lindon, J.K. Nicholson and S.H. Hansen, Drug Metab. Disp., 29 (2001) 375. O. Corcoran, R.W. Mortensen, S.H. Hansen, J. Troke and J.K. Nicholson, Chem. Res. Tox., 14 (2001) 1363. W. Chen, J.P. Shockcor, R. Tonge, A. Hunter, C. Gartner and S.D. Nelson, Biochemistry, 38 (1999) 8159. M. Spraul, M. Hofmann, M. Ackermann, A.W. Nicholls, S.J.E Damment, J.N. Haselden, J.E Shockcor, J.K. Nicholson and J.C. Lindon, Anal. Comm., 34 (1997) 339. C.Y. Airiau, H. Shen and R.G. Brereton, Anal. Chim. Acta, 447 (2001) 199.

83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

101 102 103 104 105 106 107 108 109

Biomedical applications of directly-coupled chromatography 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136

329

C.L. Gavaghan, J.K. Nicholson, S.C. Connor, I.D. Wilson, B. Wright and E. Holmes, Anal. Biochem., 291 (2001) 245. T.M. Logan, M. Murali, G.S. Wang and C. Jolivet, Magn. Reson. Chem., 37 (1999) 512. C.A. Daykin, O. Corcoran, S.H. Hansen, I. Bjornsdottir, C. Cornett, S.C. Connor, J.C. Lindon and J.K. Nicholson, Anal. Chem., 73 (2001) 1084. T. Hou, J. Smith, E. MacNamara, M. Macnaughtan and D. Raftery, Anal. Chem., 73 (2001) 2541. Y. Li, A.M. Wolters, EV. Malawey, J.V. Sweedler and A.G. Webb, Anal. Chem., 71 (1999) 4815. T. Hou, E. MacNamara and D. Raftery, Anal. Chim. Acta, 400 (1999) 297. R. Subramanian, W.P. Kelly, P.D. Floyd, Z.J. Tan, A.G. Webb and J.V. Sweedler, Anal. Chem., 73 (1999) 5335. M.E. Lacey, Z.J. Tan, A.G. Webb and J.V. Sweedler, J. Chromatogr. A, 922 (2001) 139. M.E. Lacey, R. Subramanian, D.L. Olson, A.G. Webb and J.V. Sweedler, Chem. Rev., 99 (1999) 3133. R.M. Smith, O. Chienthavorn, I.D. Wilson and B. Wright, Anal. Comm., 35 (1998) 261. R.M. Smith, O. Chienthavorn, I.D. Wilson, B. Wright and S.D. Taylor, Anal. Chem., 71 (1999) 4493. N. Wu, T.L. Peck, A.G. Webb, R.L. Magin and J.V. Sweedler, J. Am. Chem. Soc., 116 (1994) 7929. N. Wu, T.L. Peck, A.G. Webb, R.L. Magin and J.V. Sweedler, Anal. Chem., 66 (1994) 3849. K. Albert, Angew. Chem. Int. Ed. Engl.. 34 (1995) 641. P. Gfrorer, L.H. Tseng, E. Rapp, K. Albert and E. Bayer, Anal. Chem., 73 (2001) 3234. D.L. Olson, M.E. Lacey, A.G. Webb and J.V. Sweedler, Anal. Chem., 71 (1999) 3070. K. Pusecker, J. Schewitz, P. Gfrorer, L.-H.Tseng, K. Albert, E. Bayer, I.D. Wilson, N.J. Bailey, G.B. Scarfe, J.K. Nicholson and J.C. Lindon, Anal. Comm., 35 (1998) 213. J. Schewitz, E Gfrorer, K. Pusecker, L.-H.Tseng, K. Albert, E. Bayer, I.D. Wilson, N.J. Bailey, G.B. Scarfe, J.K. Nicholson and J.C. Lindon, The Analyst, 123 (1998) 2835. J. Schewitz, K. Pusecker, P. Gfrorer, U. Gotz, L.H. Tseng, K. Albert and E. Bayer, Chromatographia, 50 (1999) 333. E Gfrorer, J. Schewitz, K. Pusecker, L.H. Tseng, K. Albert and E. Bayer, Electrophoresis, 20 (1999) 3. R.A. Kautz, M.E. Lacey, A.M. Wolters, E Foret, A.G. Webb, B.L. Karger and J.V. Sweedler, J. Am. Chem. Soc., 123 (2001) 3159. M. Ludlow, D. Louden, A. Handley, S. Taylor, B. Wright and I.D. Wilson, J. Chromatogr. A, 857 (1999) 89. D. Louden, A. Hadley, S. Taylor, E. Lenz, S. Miller, I.D. Wilson and A. Sage, Anal. Chem., 72 (2000) 3922. I.D. Wilson, J. Chromatogr. A, 892 (2000) 315. E. Lenz, S. Taylor, C. Collins, I.D. Wilson, D. Louden and A. Handley, J. Pharm. Biomed. Anal., 27 (2002) 191. E. Lenz, S. Taylor, C. Collins, I.D. Wilson, D. Louden and A. Handley, J. Pharmaceut. Biomed. Anal., 27 (2002) 191. D. Louden, A. Handley, R. Lafont, S. Taylor, I. Sinclair, E. Lenz, T. Orton and I.D. Wilson, Anal. Chem., 74 (2002) 288.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 92003 Elsevier Science B.V. All rights reserved

331

CHAPTER 11

Ultra-sensitive detection of radiolabelled drugs and their metabolites using accelerator mass spectrometry Graham Lappin and R. Colin Garner Xceleron Ltd., York Biocentre, York, YOIO 5NY, U.K.

11.1 I N T R O D U C T I O N Our current understanding of the metabolism of endogenous and xenobiotic compounds has come in the main from the use of molecules labelled with radioisotopes. The compound under investigation is synthesised so that selected atoms in the molecule are enriched with the desired radioisotope. The resulting radiolabelled compound, which is chemically identical to its non-radioactive counterpart, is then administered to a test system and radioactivity is followed over time. This enables the passage of the compound and its metabolites to be distinguished from the myriad of endogenous substances. Test systems have ranged from isolated organelles, cells, plants, animals, soils, water and ecosystems. Radiotracer studies are an essential part of the registration dossier for pharmaceuticals, pesticides and some veterinary products and the techniques used today are essentially the same as those used by Hans Krebs to elucidate the TCA cycle in the 1950s. Radiotracer studies in animals, otherwise known as absorption, distribution, metabolism and excretion (ADME) studies typically follow the radioactive compound and its metabolites through organs, tissues, blood, bile, urine and faeces. For pharmaceuticals, radiotracer studies can be performed in volunteers or in patients (with their informed consent) although for ease of collection samples for analysis are limited to blood and excreta and, rarely, biopsy samples. A number of radioisotopes have been used in ADME studies but since most pharmaceuticals are organic, the most frequently used tracer is 14C, a low energy [3emitter. Central to the conduct of ADME studies are methods for the detection and quantification of radioactivity. For [3-emitters, the most commonly used methods of detection are liquid scintillation counting (LSC) or radioluminography. In LSC, the radioactive sample is dissolved in a scintillation cocktail where the kinetic energy of the References pp. 347-349

332

Chapter 11

[3-particles is converted into bursts of photons which are detected by a photomultiplier. For a review of radiotracer LSC see [1 ]. The main drawback of LSC is one of sensitivity. A [3-particle is emitted when a 14C atom decays to ~4N. The frequency of decay is dependent on the half-life (tl/2) of the radioisotope; the longer the t~/2, the fewer decay events occur in any given time. For ~4C, with a tl/2 of 5730 years, only 0.012% of the atoms decay in a year and therefore to measure one disintegration per minute (dpm) by LSC, 4.3 • 109 ~4C atoms have to be present in the sample (see Section 11.4). Because of this inherent lack of analytical sensitivity, situations can arise where LSC is inadequate and a more sensitive method of detection is required. Examples of such situations are listed below. (i) There may be constraints on the maximum attainable specific activity of the test compound due to the route of synthesis. This can be a particular problem with products such as proteins or secondary metabolites produced during fermentations ie those molecules derived from biological sources. (ii) The specific activity of the test compound may be limited by autoradiolysis. (iii) The amount of radioactivity that can be administered to a test system may be limited by the physical dose that can be given. If the dose is low, so is the amount of radioactivity. (iv) The amount of radioactivity that can be administered to humans is limited by international regulations that control human radiation exposures particularly for volunteer studies where the individual gains no therapeutic benefit. (v) Biological samples may contain low levels of radioactivity because of poor absorption or low bioavailability. (vi) For certain sample preparations such as protein, DNA or isolated cell types, uptake of radioactivity may be very low. (vii) Problems arising from any of the above may be exacerbated if radiochromatographic analysis is required. In the case of (iv), the amount of radioactivity that can be administered to human volunteers per annum, according to the World Health Organisation (WHO) guidelines, is limited to 1 mSv, but more commonly 0.5 mSv is used [2]. This value does vary from country to country with some countries such as Japan not permitting radioactive studies in human volunteers. The Sievert (Sv) is a measure of radiation exposure (the integral of dose and duration of exposure) which is caused by the ionising radiation and can be determined on a compound specific basis from dosimetry studies performed in animals (typically pigmented rats). The radioactive dose is a product of the energy of the radioemission, the time resident in the body and the susceptibility of certain tissues to radiation damage, particularly the GI tract in the case of oral dosing [3]. Each tissue is graded by a factor relating to the likelihood of DNA in that tissue becoming mutated; the gonads have a higher susceptibility factor than the brain for example. Typically, a 0.5 mSv dose in the human is equivalent to 1.5-3 MBq, or about 20 KBq/kg body weight. This is about 100-fold less than the dose that can be used in laboratory animal studies. Moreover, for drugs with long plasma half-lives, or those that bind to protein or melanin, the residence time in the body can be prolonged and so the amount of radioactivity that can be administered may have to be significantly reduced. Further to

Ultra-sensitive detection of radiolabelled drugs

333

these difficulties, subjects receiving the maximum allowable dose cannot participate in further radiolabelled experiments for 1 year. This puts severe restrictions on radiolabelled cross-over studies. The regulatory perspective of using radiolabels in human studies is given in [4]. In 1977 two papers appeared, coincidentally, in the journal Science, reporting the development of the accelerator mass spectrometer [5,6]. Accelerator mass spectrometry (AMS) measures the number of 12C, ~3C and ~4C atoms in a sample; the result of AMS sample analysis is expressed as an isotope ratio. AMS therefore, does not rely on the detection of infrequent decay events to detect 14C and consequently it is about a million times more sensitive than LSC. AMS was originally developed for carbon dating and it was some 13 years later that the first papers reporting its use for biomedical research were published (26A1 measurements [7] and DNA binding [8]). At the time of writing there are about 40 AMS facilities in the World, mostly involved in environmental, archaeological or geological research. There is only one company dedicated to commercial biomedical AMS analysis in existence (Xceleron Ltd, York U.K.) [9].

11.2 INSTRUMENTATION Figure 11.1 shows a schematic of a typical AMS. The sample, which has to be specially prepared (see Section 11.3) is placed into a caesium "sputter" ion source (A). The ion source is held under vacuum at about 10-6 Yorr, whilst the vacuum in the rest of the instrument is around 10-9 Torr. On bombardment with caesium vapour, atoms within the sample form negative ions that are extracted by a series of plates held at several thousand volts more positive than the ion source. The negative ion beam enters an injection magnet (B) where ions are selected depending upon their mass to charge ratio (m/z). For carbon analysis the injection magnet resolves to 1 m/z allowing the 12C-, ~3Cand ~4C-ions to pass as a series of pulses in sequence by virtue of the electric field being 'bounced' to permit only the selected atoms through. At this stage, molecular ions within the defined mass range are also present, such as 91 and ~2CH~ and these also pass through the injection magnet. The pulsed ion beam passes into a tandem electrostatic Van de Graaff particle accelerator (C) where the negative ions "drop" through a very high potential difference towards a positive terminal held at typically 1-5 million volts. The charge is delivered by accelerator chains, which are surrounded with pressurised SF 6 as an insulating gas to prevent sparking from the high voltage terminal. As the ions travel towards the terminal they attain high velocities, dependant upon the voltage of the accelerator. At 5 MV, their velocity is about 3% c (9,000 km/s). The highenergy ion beam is focused to collide with argon gas molecules or a thin carbon foil, 0.02-0.05 txm thick (D) in a collision cell. This has the effect of stripping the outer valency electrons and, with the loss of electrons, the charge on the ions switches from negative to positive. The extent of the electron stripping depends upon the energy of the ions, which is dependent upon the voltage of the accelerator. At 1 MV volts, the most abundant ions are C ~+ and C 2+ at about 40% each. There is insufficient energy at this voltage to form ions with higher charges above about 5% abundance. At 5 MV, over References pp. 347-349

334

Chapter 11

i

Fig. 11.1. Schematic of an accelerator mass spectrometer. Key: A Negative ion sputter source B Injection magnet C Tandem electrostatic Van de Graft particle accelerator D Electron stripper E High energy analysing magnet F Faraday cups G Quadrupole H Electrostatic cylindrical analyser I Gas ionisation detector

60% of the carbon ions are C 4* and about 30% are C 3*, ions with lower charges are unstable at this energy. Molecular charged ions such as ~3CH- and ~2CH2 do not survive the electron stripping process and are converted to atomic species. The positive ion beam is repelled by the positive high terminal voltage and exits the accelerator with a velocity of about 6.4% c (19,300 km/s) at 5 MV. The term tandem accelerator is derived from this two-stage "pull-push" effect. The ion beam passes into a high-energy analysing magnet (E) where ~2C, ~3C and 14C are separated according to their mass momentum charge state ratio. For maximum sensitivity the analysing magnet is tuned to the most abundant ions (C 4* at 5 MV). ~2C and ~3C are relatively abundant compared to ~4C and their current can be measured with Faraday cups (F). The ~4C beam is focused by a quadrupole (G) and electrostatic cylindrical analyser (H) and the atoms are counted in a gas ionisation detector (I). The instrument measures the 14C/12C/13C ratio and thereby the amount of ~4C per mg carbon (see Section 11.4). Along with carbon ions, other atomic species can also be formed in the AMS that can be potentially interfering; these are known as isobars. Whereas molecular species such as 13CH- and 12CH~ are destroyed during ion stripping, isobars are transmitted all the

Ultra-sensitive detection of radiolabelled drugs

335

way through to the gas detector. The principal isobar formed alongside carbon is believed to be lithium ((7Li2)4+) although its origin is somewhat open to question [9]. Although the lithium isobar enters the gas detector, at the very high energies involved in AMS (ca 20-100 MeV) it can be separated from 14C and eliminated from the counting procedure [10]. 14N is not an interfering isobar of ~4C as negative nitrogen ions decay in about 5 • 10-~4 s and so do not reach the accelerator. Prior to the development of AMS, cyclotron accelerators were used for isotope separation. These instruments however, were based on positive ion sources and so ~4N caused major interference with 14C. In addition to carbon, a range of other isotopes can be measured using AMS. The isotope must form a stable negative ion; the instrument must operate at sufficiently high voltages for the electron stripping process to occur for that particular isotope and the energy of the resulting ions must be sufficient to separate the analyte isotope from any isobars. Instruments operating at 5 MV can be used to analyse a range of isotopes including 4~Ca, 3~'C1,3H, 57Fe and 26A1 amongst many others [reviewed in 11]. Although there are important biomedical applications involving these isotopes, this article will concentrate on the use 14C.

11.3 SAMPLE PREPARATION At the present time, there are limited ways in which samples can be introduced into the AMS ion source. Carbon can be introduced as CO2 gas [12] but more commonly samples for ~4C analysis are placed in the ion source in the form of graphite [13]. The chemistry behind sample preparation is shown schematically in Fig. 11.2. A sample of the biological material is placed in a glass tube containing copper oxide, which is sealed under vacuum and heated at 900~ for 2 hours, where the carbon in the sample is oxidised to CO2. The sample tube is connected to one arm of a Y-manifold. To the other arm is attached a glass graphitisation tube containing titanium hydride and zinc powder as reductants and cobalt as a catalyst. The sample tube is dipped into dry ice/isopropanol and the graphitisation tube into a bath of liquid nitrogen. The CO2 in the sample tube

Carbon in biological sample CuO2

9~176176 r

Cu

Oxidation

CO2 500~ withCo Catalyst

H2from Till2 + Zn

Reduction H20

Graphite Fig. 11.2. A schematic of the graphitisation process.

References pp. 347-349

Chapter 11

336

is cryogenically transferred to the graphitisation tube, which is heat-sealed under vacuum. The tube is heated at 500-550~ for about 10 hours to reduce CO2 to graphite. The graphite/cobalt mix is taken out of the tube and pressed into hollow cathodes before being placed into the AMS ion source. The minimum sample size for AMS is governed by the amount that can be physically graphitised and pressed into the sample cathodes. Practically, this is between 0.5 and 2 mg graphite. The size of the biological sample taken for analysis should therefore be sufficient to generate this amount of carbon following the graphitisation process. Blood is about 10% w/v carbon, plasma is about 4% w/v and urine about 1% w/v and so the sample size taken for analysis is adjusted accordingly. If the biological sample is not of sufficient size to produce an appropriate amount of carbon, then it can be mixed with carbon carrier (see Section 11.6). Currently there are no direct LC-AMS interfaces available. For HPLC analysis, the eluate is collected as series of fractions, the mobile phase solvent is removed and carbon carrier added to each fraction prior to oxidation and reduction to graphite. A similar process has been described for GC, whereby compounds eluted from a GC were cold trapped [14]. A direct sample interface is being developed [15]. Further direct, or semidirect, sample introduction techniques are considered a priority for the development of AMS technology (see Section 11.7).

11.4 DATA ANALYSIS

AMS instruments for ~4C analysis have traditionally been used for carbon dating and the commercial software tends to be biased to this application. In particular, one of the output values is a percentage of modem carbon (pMC). This term is somewhat confusing for the biomedical researcher and so it warrants a brief explanation. All living entities contain ~4C in equilibrium with the natural abundance in the atmosphere. A level of J4C referred to as "100% modem" (100pMC) corresponds to 1 ~4C atom per 1.18 • 10 ~2 atoms of carbon, or 97.6 attomole ~4C per mg carbon. AMS standards with precisely known pMC values are available as instrument checks and to normalise data if necessary. The two most widely used are standard oxalic acid from the U.S. National Institute of Standards Technology (NIST) and a crop of sugar harvested in Australia in the 1960s and certified by the Australian National University (ANU). The NIST oxalic acid standard has a pMC of 95 and ANU sugar has a pMC of 150.61 (the latter standard was harvested during a period where radioactive fallout from atomic weapons was still relatively high, hence the pMC value is > 100. At the time of writing, the atmosphere has a pMC of about 110). As stated in Section 11.1, the number of decay events for any radioisotope is governed by its t~/2. The longer the tl/2, the fewer decay events occur in any given time. The number of decay events per minute for a given amount of radioisotope is defined by equation (1).a dN

~ This equation is derived from the fundamental equation defining half life: - - = - K N . dt

Ultra-sensitive detection of radiolabelled drugs In 2 dpm=--xN t~/2

337 (1)

where In 2 is the natural log of 2 (0.6932), tl/2 is the half-life in minutes and N is the number of atoms of the radioisotope, calculated by multiplying the number of moles by Avrogadro's number (6.0225 x 1023). From equation (1), 1 mole ~4C, with a tl/2 of 5730 years (3.0138 x 10 9 minutes) equals 1.3852 z 10 j4 dpm. b Thus, 1 dpm equals 7.2192 z 10-~5 mole, which equals 4.3477 z 10 9 atoms. As stated above, 100 pMC is equivalent to 97.6 attomole ~C per mg carbon. When converted to units of radioactivity this is equal to 13.56 dpm per g carbon. It is important to understand that AMS provides a dpm value per g carbon based on the isotope ratio and not absolute values such as dpm per g of biological sample. This is a fundamental difference in the output of AMS compared to LSC. AMS measures numbers of atoms and not radioactive decay events. It is only to provide units familiar to biomedical researchers that the isotope ratio is converted to dpm values. To calculate dpm per g sample, the proportion of carbon in the sample must be known and this is measured using a suitable C,H,N analyser.

11.5 APPLICATIONS There are a number of published accounts where results from LSC and AMS were directly compared and correlation coefficients of 0.999 were achieved [16-18]. The upper limit for LSC is hundreds of thousands or millions of dpm, whereas the upper limit for the AMS gas detector, before it becomes saturated, is around 50 dpm, and ideally no more than 10 dpm. In order to compare LSC and AMS values therefore samples measured by LSC need to be diluted up to 1000 fold prior to AMS analysis. There are a number of reviews on the application of AMS in biomedical research [9,18-28]. The first published accounts of biomedical AMS examined the covalent binding of animal carcinogens to DNA. Prior to the advent of AMS, detecting the presence of radiolabelled adducts on macromolecules such as DNA was very challenging. Even using potent carcinogens, only a very small proportion of an administered radiolabelled compound forms adducts [29] and hence the isolated DNA has a very low specific activity. The first results using AMS to detect radioactive adducts on DNA showed that the limit of detection was in the region of 1 adduct per 10 ~ bases, an order of magnitude greater sensitivity than 32P-post labelling [8]. The macromolecular binding of heterocyclic amines, such as those formed in the cooking of protein-rich foods have been studied [reviewed in 30]. In many cases, dietary relevant doses were administered and tissues showing pathologies were analysed in rodents and humans [8,30-36]. The macromolecular binding of benzene at doses as low as 700 pg/ kg, (118 kBq/kg) was successfully investigated and showed detection levels in pg The maximum theoretical specific activity of a compound containing one J4C per molecule, is therefore 1.3852 z 10j4 dpm/mole or 2.3 GBq/mmole.

b

References pp. 347-349

338

Chapter 11

benzene equivalents/g [37,38]. DNA binding of trichloroethylene [39], benzo[a]pyrene [36,40] tamoxifen and toremifene [41,42] have also similarly been investigated. In such studies, 6 adducts in 1012 nucleotides were routinely measured. The potential for (~4C)nicotine to form adducts with protein and DNA in mice at a level equivalent to a single cigarette (125 Ixg/kg bodyweight) has also been investigated [43]. For DNA adducts to be chemically characterised, isolated DNA is hydrolysed (enzymatically or by acid) and the individual nucleotides or nucleosides are separated chromatographically. Hydrolysed adducted DNA has been analysed by HPLC-AMS and adducts detected down to the fg level [31,34,39,41]. Typically, nucleotides were separated using a MVC-18 column (250 x 4.6 mm, 3 txm particle size) eluted with 0.1% (v/v) trifluoracetic acid (solvent A) and acetonitrile (solvent B) at 1 mL/min, using a gradient of 10% to 35% B over 35 min. In addition to studying adduct formation, the metabolism of the heterocyclic amines, 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) and 2-amino-l-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) have been studied in rodents and humans [33,44,45,46]. Following the administration of ~4C-PhIP at levels as low as 50 ng/kg, the distribution of PhIP and its metabolites was followed in tissues and the milk of female rats, as well as distribution of radioactivity in the tissues of the pups [47]. Although MeIQx and PhIP are potential carcinogens, because the human dose administered was as low as 300 ng/kg, a level lower than that encountered from everyday exposure to eating cooked meat, it was ethically possible to dose volunteers. Urine from PhIP-dosed mice and humans analysed by HPLC-AMS showed that there were marked differences in the metabolic profiles between species. The principal metabolite in the mouse was 4'-PhIP-sulphate whereas in the human the principal metabolite was the N-OH-PhIP-N2-glucuronide. Typically, metabolites were separated on a C18 250x4.6 mm column (3 txm particle size) eluted with 0.1% (v/v) trifluoracetic acid (solvent A) and acetonitrile (solvent B) at 1 mL/min, using a gradient of 5% to 30% B over 30 min. Knowledge of the action of potential carcinogens and other toxic chemicals have been largely based on animal models due to the ethical constraints of administering such substances to humans. With the advent of AMS however, it is possible to administer extremely small doses of compound, below the levels that would be considered to have harmful effects. It is now possible therefore, to directly compare carcinogen metabolism in humans with that of laboratory animals and challenge some of the risk assumptions made on the basis of animal models alone. In order to study the metabolism of certain endogenous compounds, it is necessary to monitor turn-over for prolonged periods of time. Prior to the advent of AMS, it was very difficult to do this in humans as the long exposure times placed severe constraints on the amounts of radioactivity that could be dosed. Using AMS however, the long term metabolism of fat was investigated by the analysis of exhaled ~4CO2 over a 320 day period after administering as little as 74 kBq ~4C-triolein precursor [48]. Folic acid metabolism has been studied in human volunteers following the ingestion of just 3.7 kBq pteroyl-[~4C(U)]-glutamic acid for up to 202 days. The limits of detection were 0.4, 0.04 and 0.12 fmol J4C-folate/mL in plasma, urine and faeces respectively [49,50]. The metabolism of ~4C-[3-carotene was studied in an adult volunteer following the

Ultra-sensitive detection of radiolabelled drugs

339

administration of 7.4 kBq [51]. Radioprofiling of plasma by HPLC, fraction collection and AMS allowed the kinetics of retinyl acids, retinol and [3-carotene to be followed over 202 days. Compounds were separated using an XDB-C18 column (3 x 150 mm, 3-5 ~m particle size) fitted with a guard column with the same packing. The column was eluted isocratically with acetonitrile/methanol/2-propanal with 0.1% ammonium acetate (49:21:30 v/v/v) at 1 mL/min. The dermal absorption of atrazine was studied in human volunteers [52]. Atrazine was administered in doses down to 167 ~g, 239 kBq in the form of a skin patch. Urine was analysed by HPLC-AMS after injecting approximately 10 dpm on column and showed a complex series of metabolites, six of which were tentatively identified on the basis of co-elution with standards. Metabolites were separated using a YMC 303 S-5 120A ODS-AQ column and guard column (250x 4.6 mm) eluted with 0.1% (v/v) aqueous acetic acid (solvent A) and 0.1% acetic acid (v/v) in acetonitrile (solvent B) at 1 mL/min. The gradient was 100% A to 100% B over 55 min. The absorption kinetics of trichloroethylene through human skin, measured in vitro, has also been studied using very small ~4C-labelled doses, followed by AMS analysis [53]. More recently, AMS has been applied to the study of pharmaceutical metabolism, primarily at our Centre. A pharmaceutical under development was dosed to four human volunteers using a radioactive dose of 1.85 kBq per subject [54]. Rates of excretion were investigated in urine and faeces, along with plasma pharmacokinetics. Metabolite profiles were studied by HPLC-AMS in plasma and extracts of faeces. At the Cm,x, plasma contained just 2 dpm/mL. To put the dose of 1.85 kBq per subject into context, the average human contains about 3.7 kBq naturally occurring ~4C [27]. This study was therefore conducted using essentially what was a non-radioactive dose. In another study, how the particle size of formulated Daflon 500 | effected absorption was investigated in 12 human volunteers administered just 925 Bq [55]. This radioactive dose was so low, that conventional tabletting methods were used for formulation. AMS was used to measure the ~4C-content of urine and faeces and quantitative recoveries of radioactivity were achieved. An ADME study was performed with (~4C)-GI1817771 on six male volunteers using a dose of 121 Bq per subject, equivalent to a radioactive exposure of 0.06 ~Sv [56]. Radioactivity was measured in serum, urine and faeces. Approximately 99.2% of the administered dose was recovered in faeces over the five day study period. Four healthy male volunteers were dosed with 1.85 kBq Rl15777, a farnesyl transferase inhibitor and plasma and excreta were analysed by AMS [57]. Urine was analysed by HPLC-AMS before and after treatment with [3-glucuronidase. The pharmacokinetic data are shown in Fig. 11.3 and the chromatogram for urine is shown in Fig. 11.4. Metabolites were separated using a Luna C-18 column (250x 4.6 mm, 5 ~m particle size) eluted with ammonium acetate buffer (pH 5.5)/methanol/acetonitrile (10:10:80 v/v/v). The gradient was 100% A to 40% A over 5 min., held for 15 min. then to 20% A over 10 min. The flow rate started at 1 mL/min, for 10 min. then at 0.5 mL/min. Fractions were collected every 60 seconds. The maximum level of radioactivity in the urine chromatograph was ca 3.3 dpm/fraction. One of the first human experiments conducted in the development of a new pharmaceutical is a dose escalation and tolerability study. In one such study, a small References pp. 347-349

m

0.5

1

0

20

40

0

i- + -

0

0.5

1

1.5

2

Hours post dose

80

10

100

15

Volunteer I - , - Volunteer 2 - * - Volunteer 3 ~

60

5

20

T

140

25

160

the samples contain less than 2.5 dpm/mL.

Volunteer 4 1

120

r

Expanded for flint 24 houm

Fig. 11.3. A typical pharmacokinetics curve generated from plasma samples analysed by AMS. Note that at C . . . . Reproduced with permission [57].

"O

a.

E

.J

Q.

W

E 1.5

,.51

2.5

~-~

Ultra-sensitive detection of radiolabelled drugs

341

3.5

2.5 c 0

=(.1

2

tl

E 1.5

O. "0

i

0.5

|1

~

,

_

0

~

~

............

10

20

_.

.

.

.

30

~

40

50

60

T ime (min) I

.

.

.

.

Untreated B - g l u c u r o n i d a s e treated]

Fig. 11.4. HPLC-AMSchromatogramof urine metabolites before and after treatment with [3-glucuronidase. Reproduced with permission [57]. dose of 14C-labelled compound (1.85-18.5 kBq) was administered along with an unlabelled drug, which represented an exposure of less than 10 IxSv [58,59]. The concentration of radioactivity at the Cmax was less than 1 dpm/mL but by injecting just 0.35 dpm on the HPLC column a complex series of metabolites was revealed (Fig. 11.5). Metabolites were separated using a C18 (20 x 3.9mm 5 Ixm particle size) column eluted with 20 mM aqueous ammonium acetate, adjusted to pH 4.6 with acetic acid (solvent A) and acetonitrile (solvent B). The gradient was 95% A to 0% A over 85 min. The flow rate was 1 mL/min, and the column temperature was 40~ The studies described in [58,59] required regulatory approval from the Swiss authorities in order to administer the radioactive dose to humans. In some other countries it is feasible to administer these very low levels of radioactivity to humans without regulatory approval, thus negating the need for dosimetry studies. This opens up the possibility of combining early human experiments with the radiolabelled study thereby avoiding volunteer recruitment, specifically for this purpose, at a later stage. For certain life-saving pharmaceuticals such as cancer and HIV treatments, the first human studies are in patients. Given that there is already one account of using conventional tabletting machines to incorporate very low levels of '4C [55], this opens the possibility of conducting ADME studies with patients at this very early stage. 11.6 L I M I T S OF D E T E C T I O N AND QUANTIFICATION Since AMS counts individual atoms, it can be argued that the absolute limit of detection (LOD) for the technique is essentially zero. In order to obtain a statistically significant References pp. 347-349

Chapter 11

342

br~ O0 br~ 9 c~ c~

o 0 0

9

0

d

=

O~

9

o -

0

=

Qo

9

(

0

.~

d

"0 ,+.,

9

c-

0

v

O3 Q.

E

brb

@

.m

0.0

d

0

d

~0

O0

c~ 9

o d 9

o d

"o n

9

o i

i

i

i

0 ~0 ~0

0 ~ ~0

0 ~ ~0

0 OQ ~-0

d

d

d

d

i

i

~

i

o

0 0 ~0

0 CO 0 0

0 CO 0 0

0 ~I" 0 0

0 OQ 0 0

0 0 0 0

d

d

o

d

d

d

CtO 9

uop,oeJlltudp 9 0

U~

Ultra-sensitive detection of radiolabelled drugs

343

count however, about one thousand atoms are counted, which puts the statistical limit of detection in the zeptomole range (zepto = 10-2~). In practice, the LOD for AMS, as for any analytical technique, is defined by the signal to noise ratio. The "signal" depends upon the amount of J4C from isotopically enriched drug and the "noise" depends upon the amount of endogenous ~4C. Since the ratio of ~4C to ~2C is fixed in all living entities (see Section 11.4) then the amount of endogenous ~4C increases proportionally with the total amount of carbon in the sample. Some biological samples (eg faeces) contain a higher proportion of carbon than others (eg serum) and so the LOD in faeces is somewhat higher than in serum. There are a number of ways of reducing the amount of endogenous carbon in a sample, whilst not significantly lowering the contribution of ~4C from the drug. For example, proteins can be precipitated by the addition of water-miscible organic solvents such as acetonitrile. Protein is then removed by centrifugation. Although it is possible to graphitise faeces directly, the amount of endogenous ~4C can be significantly reduced, thus improving the LOD, by including a solvent extraction step in the analysis. This technique does of course, rely on quantitative extraction efficiencies. Aliquots of the solvent extracts are dried to remove the solvent and then graphitised for AMS analysis. The LOQ for a series of matrices was recently determined [56] and are shown in Table 11.1. The LOQ in this case was defined as five times the standard deviation of the background. Note that the LOQ for serum was improved nearly 100-fold following solvent extraction. Experiments undertaken by the authors, whereby plasma was spiked with decreasing amounts of radioactivity demonstrated that, as a general guide, 0.1 dpm/mL above background could be reliably quantified and 0.06 dpm/mL could be reliably detected (Fig. 11.6). Probably the most efficient method of separating endogenous carbon from that of the drug is HPLC. Under ideal conditions, the drug and its metabolites are eluted from the HPLC column in a highly pure form and endogenous carbon is removed entirely, leaving only drug related J2C and ~4C. LOD's for HPLC are therefore, in theory, extremely low. In practice, the amount of carbon contained in a HPLC fraction is too low to be handled in the graphitisation process and therefore "carrier carbon" has to be

TABLE 11.1 LOQs FOR AMS DETERMINED IN A NUMBER OF BIOLOGICALMATRICES. LOQ IS DEFINEDAS FIVE TIMES THE STANDARD DEVIATION OF THE BACKGROUND VALUE. TBT - GLYCEROL TRIBUTANOATE. LP- LIQUID PARAFFIN Sample type Faeces Urine Serum Serum extract HPLC faction using TBT HPLC faction using LP

References pp. 347-349

LOQ (dpm/mL) 0.86 0.20 0.43 0.005 0.036 0.018

0.3000

0.0000

0.0500

0.1000

0.1500

0.2000

0

10

20

Percent dilution

30

40

50

60

Fig. 11.6. A calibration line generated from the serial dilution of plasma. The LOQ in this case was 0.1 dpm (3 x SD) and the LOD was 0.06 dpm above background. (Plasma was used for dilution thus maintaining a constant carbon background for each data point). Error bars are standard error.

"0

"~ 0.2500

J

0.3500

0.4000

0.4500

0.5000

~

Ultra-sensitive detection of radiolabelled drugs

345

added to bring the quantity to a manageable amount (Section 11.3). The apparent dichotomy of first removing background carbon and then adding it back again is overcome by using ~4C-depleted carriers originating from immensely old carbon sources from the petrochemicals industry. One such commonly used carrier is glycerol tributanoate (TBT). TBT however, still contains some ~4C and where the lowest limits of detection are required substances such as liquid paraffin are used as the carrier. The LOQ for HPLC fractions shown in Table 11.1 compares those using TBT and liquid paraffin carriers. Under certain circumstances the use of carbon carrier can lead to errors. Providing the actual amount of carbon in the sample is vanishingly small, It can be assumed to be zero and only the carbon in the carrier is taken into account when calculating the dpm/mL value from the isotope ratio. This is potentially a source of error, as the sample will always make some contribution to the carbon level, albeit potentially very small. As a general rule therefore, where the carbon content is measurable by a conventional C,H,N analyser it should be taken into account. Only if it is below the sensitivity of the analyser should it be assumed to be zero, in which case the error is very minor and probably within the precision of the measurement and can be ignored. Lower LODs than those shown in Table 11.1 have been achieved using HPLC-AMS (unpublished data). As little as 0.024 dpm of a plasma extract was injected onto a HPLC. Fractions were collected, liquid paraffin added as a carbon carrier, graphitised and analysed by AMS. The resulting chromatogram can be seen in Fig. 11.7. The LOQ was estimated at 0.0008 dpm/fraction (0.0016 dpm/mL) based on 3 times the standard deviation for the background. The peak eluting at 40-41 minutes was shown to be endogenous ~4C eluting from the column and was not drug related. The chromatogram in Fig. 11.7 illustrates that when operating at such low levels of detection, endogenous carbon coincidentally eluting from the column can be misleading and control samples (ie pre-dose) should always be run alongside samples so that these spurious peaks can be eliminated. The sensitivity of AMS is both its biggest advantage and its biggest drawback as even minute amounts of contamination can result in anomalous results. If experiments are conducted in laboratories where "conventional" studies are performed, contamination is a real danger. The AMS laboratory has to have controlled entry where "high" levels of radioactivity are excluded. To ameliorate this difficulty, some companies have opted to build special facilities for conducting AMS work [56].

11.7 CONCLUSIONS AND THE FUTURE Although several groups are working on interfaces which will allow the direct or semidirect introduction of biological sample into the AMS instrument [ 12], no such interface is readily available at the time of writing. The current technique of sample preparation by graphitisation, is labour intensive. We are developing methods to permit samples to be directly inserted into the ion source thus eliminating the graphitisation process. In contrast to the sample preparation time, AMS run times per sample can be as short as 100 seconds and still give good precision. AMS is the most sensitive analytical method References pp. 347-349

0.0025 !

5

10

15

Fraction collection

/

21

26

Retention time (min)

18.5

LOQ estimated at 0.0008 dpm/fraction (background + 3xSD) .

Parent drug

Fig. 11.7. HPLC-AMS chromatogram produced from 0.024 dpm injected on column. The peak eluting at eluting from the column.

0 . 0 0 0 0 -+ 0

i

L

O.OOO5 ~.t

O.0010 -

a. 0.0015 ? E

L_

,_m 0.0020-I

c o

0.0030 i

0.0035 -

ca

36

41

41 min. is not drug related but background carbon

31

/

j 1

Background peak

r~

C~

Ultra-sensitive detection of radiolabelled drugs

347

ever developed and a range of studies, particularly in humans, which were previously impossible, can now be performed. Regulatory authorities are increasingly aware of the technique and are becoming less likely to grant exemptions from performing certain studies on the grounds that it is not feasible to detect ~4C-tracer at very low levels. Up to the present time, AMS has been used largely as a problem-solving tool but there is increasing interest in using the technology strategically as part of the drug registration process. It is possible to combine radiolabelled studies in humans with other volunteer studies, thus saving time and cost. With AMS, radiolabelled studies can be performed with patients and even with children [60]. There is also great potential for acquiring data on human metabolism at a much earlier stage which can assist in compound selection or in the selection of the most appropriate species for toxicology studies. The textbook approach to radiotracer studies, particularly those involving humans has been overturned by the arrival of AMS in biomedical research. The use of AMS has the potential to eliminate the need for dosimetry studies altogether and to free human ADME work from the regulatory constraints that surround the use of radiolabelling. The extreme sensitivity of AMS allows researchers to venture into new territories, following the radiotracer into individual cell types, or binding to biomolecules and receptors. The applications for AMS in biomedical research have, in reality, only just started.

11.8 R E F E R E N C E S 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

M.E L'Annunziata and M.J. Kessler, in: M.E L'Annunziata (Ed.), Handbook of Radioactivity Analysis, Academic Press, New York, 1998. International Commision on Radiological Protection in Biomedical Research. ICRP Publication 62 Ann. ICRP 22(3) (1992) 1-18. G.W. Dolphn and I.S. Eve, Health Physics, 12 (1966) 163. J.G. Dain, J.M. Collins and W.T. Robinson, Pharmaceut. Res., 11 (6) (1994) 925. C.L. Bennett, R.E Beukens, M.R. Glover, H.E. Gove, R.B. Liebert, A.E. Litherland, K. H. Purser and W.E. Sondheim, Science, 198 (1977) 508.. D.E. Nelson, R.G. Korteling and W.R. Stott, Science, 198 (1977) 507.. J. Barker, J.E Day, T.W. Aitken, T.R. Charlesworth and R.C. Cunningham, Nucl. Instr. Meth. Phys. Res., B, 52 (1990) 540. K.W. Turtletaub, J.S. Felton, B.L. Gledhill, J.S. Vogel, J.R. Southon, M.W. Caffee, R. C. Finkel, D.E. Nelson, I.D. Proctor and J.C. Davis, Proc. Natl. Acad. Sci. USA, 87 (1990) 5288. J. Barker and R.C. Garner, Rapid Commun. Mass Spec., 13 (1999) 285. U. Zoppi, M. Suter and H-A. Synal, Nucl. Instr. Meth. Phys. Res., B, 89 (1994) 262. M. Hotchkis, D. Fink, C. Tuniz and S. Vogt, Appl. Radiat. and Isotopes, 53 (2000) 31. D.A. Mucciarone and R.B. Dunbar, J. Sed. Petrol., 62 (1992) 731. J.S. Vogel, Radiocarbon, 34 (1992) 344. L.A. Currie, T.I. Eglinton, B.A. Benner Jr andA. Pearson, Nucl. Instr. Meth. Phys. Res., B, 123 (1997) 475. D.J.W. Mous, W. Fokker, R van den Broek, R. Koopmans, C.B. Ramsey and R.E.M. Hedges, Radiocarbon, 40 (1998) Abstract 35. R.C. Garner, J. Barker, C. Flavell, J.V. Garner, M. Whattam, G.C. Young, N. Cussans, S. Jezequel and D. Leong, J. Pharm. Biomed. Anal., 24 (2000) 197. S.D.Gilman, S.J. Gee and B.D.Hammock, J.S. Vogel, K. Haack, A. Buchholz, P.H.T. Freeman, R.C. Wester, X. Hui and H.I. Maibach, Anal. Chem., 70 (1998) 3463. B. Kaye, R.C. Garner, R.J. Mauthe, S.E H.T. Freeman and K.W. Turteltaub, J. Pharm. Biomed. Anal., 16 (1997) 541.

348 19 20 21 22 23 24 25 26 27 28

29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

46 47 48 49 50

Chapter 11 J.S. Vogel and K.W. Turteltaub, Nucl. Instr. Meth. Phys. Res., B, 92 (1994) 445. J.S. Vogel, K.W. Turteltaub, R. Finkel and D.E. Nelson, Anal. Chem., 67 (1995) 353A. S.E H.T. Freeman and J.S. Vogel, Int. J. Mass Spectrom. Ion Processes, 143 (1995) 247. C. Tuniz, J.R. Bird, D. Fink and G.E Herzog, Accelerator Mass Spectrometry Ultrasensitive Analysis for Global Science, CRC Press, 1998. B.C. Cupid and R.C. Garner, Accelerator Mass Spectrometry - A new tool for drug metabolism studies, in: Gooderham, N. (Ed,), Drug Metabolism: Towards the New Millennium, IOS Press (1998). R.C. Garner, European Pharmaceutical Contractor, Technomark Consulting Services November 1999. R.C. Garner and D. Leong, Nucl. Instr. Meth. Phys. Res., B, 172 (2000) 892. R.C. Garner, Curr. Drug Metab., 1 (2000) 205. K.W. Turteltaub and J.S. Vogel, Curr. Pharmaceut. Design, 6 (2000) 991. R.C. Garner, Ultrasensitive analysis of isotopes in drug discovery and development using accelerator mass spectroscopy, in: U. Pleiss and R Voges (Eds.), Synthesis and Applications of Isotopically Labelled Compounds, Vol. 7, John Wiley & Sons Ltd., 2001 W.K. Lutz, Mutation Res., 65 (1979) 289. K.W. Turteltaub, J.S. Vogel, C.E. Frantz and E. Fultz, IARC Sci. Publ., 124 (1993) 293. K.W. Turteltaub, R.J. Mauthe, K.H. Dingley, J.S. Vogel, C.E. Frantz, R.C. Garner and N. Shen, Anal. Res., 376 (1997) 243. T. Su, W. He, J. Gu, T.W. Lipinskas and X. Ding, Amer. Soc. Pharmacol. Exp. Ther., 26 (1998) 825. K.W. Turteltaub, K.H. Dingley, K.D.Curtis, M.A. Malfatti, R.J. Turesky, R.C. Garner, J.S. Felton and N.E Lang, Cancer Lett., 143 (1999) 149. R.J. Mauthe, K.H. Dingley, S.H. Leveson, S.E H.T. Freeman, R.J. Turesky, R.C. Garner and K.W. Turteltaub, Int. J. Cancer, 80 (1999) 539. K.H. Dingley, K.D. Curtis, S. Nowell, J.S. Felton, N.E Lang and K.W. Turteltaub, Cancer Epidemiol. Biomarkers Prev., 8 (1999) 507. T.J. Lightfoot, J.M. Coxhead, B.C. Cupid, S. Nicholson and R.C. Garner, Mutation Res., 4721 (2000) 119. M.R. Creek, C. Mani, J.S. Vogel and K.W. Turtletaub, Carcinogenesis, 18 (1997) 2421. C. Mani, S. Freeman, D.O. Nelson, J.S. Vogel and K.W. Turteltaub, Toxicol. Appl. Pharmacol., 159 (1999) 83. A. Kautiainen, J.S. Vogel and K.W. Turteltaub, Chem. Biol. Interactions, 106 (1997) 109. R. Goldman, B.W. Day, T.A. Carver, R.J. Mauthe, K.W. Turteltaub and EG. Shields, Chem. Biol. Interactions, 126 (2000) 171. E.A. Martin, E Carthew, I.N.H. White, R.T. Heydon, M. Gaskell, R.J. Mauthe, K.W. Turteltaub and L.L. Smith, Carcinogenesis, 18 (1997) 2209. I.N.H. White, E.A. Martin, R.J. Mauthe, J.S. Vogel, K.W. Turteltaub and L.L. Smith, Chem. Biol. Interactions, 106 (1997) 149. C. Mani and K.W. Turteltaub, Drug Met. Rev., 32 (supp. 2) (2000) abstract 129 from the 10th North American ISSX Meeting, Indianapolis, USA. K.W. Turteltaub, J.S. Vogel, C.E. Frantz and N. Shen, Cancer Res., 52 (1992) 4682. R.C. Garner, T.J. Lightfoot, B.C. Cupid, D. Russell, J.M. Coxhead, W. Kutschera, A. Priller, W. Rom, E Steier, D.J. Alexander, S.H. Leveson, K.H. Dingley, R.J. Mauthe and K.W. Turteltaub, Cancer Lett., (1999) 161. N.P. Lang, S. Nowell, M.A. Malfatti, K.S. Kulp, M.G. Knize, C. Davis, J. Massengill, S. Williams, S. MacLeod, K.H. Dingley, J.S. Felton and K.W. Turteltaub, Cancer Lett., 143 (1999) 135. R.J. Mauthe, E.G. Snyderwine, A. Ghoshal, S.EH.T. Freeman and K.W. Turteltaub, Carcinogenesis, 19 (1998) 919. K. Stenstr6m, S. Leide-Svegborn, B. Erlandsson, R. Hellborg, S. Mattsson, L-E. Nilsson, B. Nosslin, G. Skog and A. Wiebert, Appl. Radiat. Isot., 47 (1996)417. B.A. Buchholz, A. Arjomand, S.R. Dueker, ED. Schneider, A.J. Clifford and J.S. Vogel, Anal. Biochem., 269 (1999) 348. A.J. Clifford, A. Arjomand, S.R. Dueker, ED. Schneider, B.A. Buchholz and J.S. Vogel, Mathematical Modeling in Experimental Nutrition, Clifford and Mtiller (eds) Plenum Press New York 1998.

Ultra-sensitive detection of radiolabelled drugs 51 52 53 54 55 56 57 58 59 60

349

S.R. Dueker, Y. Lin, B.A. Buchholz, ED. Schneider, M.W. Lame, H.J. Segall, J.S. Vogel and A.J. Clifford, J. Lipid Res., 41 (2000) 1790. B.A. Buchholz, E. Fultz, K.W. Haak, J.S. Vogel, S.D. Gilman, S.J. Gee, B.D.Hammock, X. Hui, R.C. Wester and H.I. Maibach, Anal. Chem., 71 (1999) 3519. K.T. Bogen, G.A. Keating, S. Meissner and J.S. Vogel, J. Expos. Anal. Environ. Epidemiol., 8 (1998) 253. J. Dain, J. Warsheski, R.C. Garner and V. Fischer, Drug Metabolism Reviews 32 (supp. 2) (2000) abstract 23 from the 10th North American ISSX Meeting, Indianapolis, USA. R.C. Garner, J.V. Garner, S. Gregory, M. Whattam, A. Calam and D. Leong, J. Pharm. Sci., 91 (2002) 32. G. Young, W. Ellis, J. Ayrton, E. Hussey and B. Adamkiewicz, Xenobiotica, 31 (2001) 619. R.C. Garner, I. Goris, A.A.E. Laenen, E. Vanhoutte, W. Meuldermans, S. Gregory, J.V. Garner, D. Leong, M. Whattam, A. Calam and C.A.W. Snel, Drug Metab. and Disposition 30 (2002) 823-830. E Waldmeier and C. Garner, Drug Met. Rev., 133 (supp. 1) (2001) abstract 239 from the 6th International ISSX Meeting (2000), Munich, Germany. E Waldmeier and R.C. Garner, J. Labelled Comp. Radiopharm., 44 (2001) 973. S. Leide-Svegborn, K. Stenstrom, M. Olofsson, S. Mattsson, L-E. Nilsson, B. Nosslin, K. Pau, L. Johansson, B. Erlandsson, R. Hellborg and G. Skog, Europ. J. Nucl. Med., 26 (1999) 573.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 92003 Elsevier Science B.V. All rights reserved

351

CHAPTER 12

Biomedical applications of inductively coupled plasma mass spectrometry (ICP-MS) as an element specific detector for chromatographic separations Fadi R. A b o u - S h a k r a GV Instruments, Crewe Road, Manchester M23 9BE, U.K.

12.1 AN I N T R O D U C T I O N TO I C P - M S Inductively coupled plasma mass spectrometry (ICP-MS) is an instrumental analytical technique based on the use of a high temperature ionisation source (ICP) coupled to a mass spectrometer. The technique evolved from the work of Gray and Houk in the early 1980s [ 1]. Since then it has rapidly matured into the technique of choice for routine ultra trace-element analysis. This is mainly due to its advantages in terms of detection limits, relative freedom from interference and speed of analysis.

12.1.1 Inductively coupled plasma as an ion source

Inductively coupled plasmas were developed in the early 1960s [2]. They are flame-like discharges that can reach up to 10,000 K in temperature and are formed in a stream of argon, though other gases have been used, flowing through an electromagnetic field. Typically in an ICP the gas flows into three concentric tubes. These tubes are assembled together in what is commonly referred to as the plasma torch (Fig. 12.1). The flowing gas in the outer tube, which is typically in the range of 13-17 L/min, is often referred to as the cool gas and its primary role is to ensure that the high temperature of the plasma does not melt the torch. In addition to this primary function, the cool gas also plays a role in giving the plasma its distinct shape. The flowing gas in the central tube, which is usually in the range of 1 L/min, is commonly referred to as the plasma gas and is actually the gas that gets ionised to form the plasma. Finally, the gas flowing in the References pp. 370-371

352

Chapter 12

/

z

/

. / \,

/

Fig. 12.1. A schematic of the drawing of an ICP torch. innermost ring is called the carrier gas and is used to punch the plasma whilst carrying an aerosol or particulate form of the sample to be analysed. As shown in Fig. 12.1, one end of the torch is encircled by a coil. A radio-frequency current flows through the coil to create an electromagnetic field in which the argon flows. To light the plasma a high voltage discharge (spark) is passed through the argon. When passing through the electromagnetic field, electrons from the discharge gain energy. Some of these electrons transfer this energy through collisions to the argon atoms in the gas thereby ionising them. This will create sets of ion-electron pairs that are in turn energised in the presence of the electromagnetic field and are made to participate in this cascading method of energy transfer from the coil to the gas. The outcome of this cascade is the formation of a steady-state plasma that is maintained as long as the radio-frequency current is upheld at a sufficient intensity and the gas flow is preserved. The fireball-like argon plasma is characterised by a bluish white emission which is a combination of emissions from the line spectrum of atomic argon and the continuous spectrum of ion-electron recombinations taking place within the plasma. When formed, the plasma has the shape of a prolate spheroid, and the rapid expansion and acceleration of the argon gas inside it makes it difficult for sample introduction. However, by punching a central channel inside the plasma, its shape is changed into an annular "doughnut" form with the outer "plasma" gas virtually shielding the central "carrier" gas and with very little mixing taking place between the two. As a result, samples can be introduced in a gaseous or aerosol form along this central channel without significantly disturbing the plasma or changing its composition. However, during its stay in this channel, the sample will be efficiently desolvated, atomised, excited and ionised. Hence any chemical species injected into the plasma will be broken down into its constituting elements which are then ionised independently of their original form, but depending mainly on their ionisation energies as predicted by the Saha Equation (see Appendix 1). In fact, most elements in the periodic table will be efficiently converted into singly charged ions. As shown in Fig. 12.2 the efficiency of ionisation, decays as the ionisation energy of the element approaches that of argon (15.8 eV). Hence, elements such as F, Ne and He are so poorly ionised that the technique cannot be used for direct trace level determinations of these elements.

,..q

"..,I

r~

%

0.0%

20.0%

40.0%

60.0%

5.0

6.0

................ , 7.0

, ................. 8.0

,

X

X XX

Xx x@ x X

9.0

...... , .........

X

X

X

X X

_.

, ........

IE (eV)

10.0

X

X

X

11.0

, .......... 12.0

X

X

14.0

X . . . . . X ..... ~

13.0

X

15.C

............

!

...........................................................................................................................................................................................................................

Fig. 12.2. A representation of the efficiency of ionisation of elements in an ICP as a function of their ionisation energy.

C 0, .m

0 ,.,.,,

I::

80.0%

100.0%

120.0%

L~h

354

Chapter 12

12.1.2 Interfacing the ICP to a mass spectrometer

Mass spectrometers require a low operating pressure in order to avoid high voltage discharges and to ensure that the mean free path of the ions is long enough to avoid collisions with the background atmosphere in the system. The transfer of ions from the ICE which operates at atmospheric pressure, into a mass spectrometer was the focal point of early research into the feasibility of the technique. In most modern ICP-MS systems, the plasma is interfaced to the mass spectrometer via a series of vacuum chambers confined within extraction cones with small apertures. The design in terms of shape and aperture size of these cones plays an important role in defining the analytical performance of the system. Common to all ICP-MS systems is the presence of the socalled sampler and skimmer cones. The former is the interface between the ICP and the first vacuum chamber. This cone is subject to an intense heat as the plasma impinges on its surface and is therefore in need of efficient cooling. As a result, these cones are usually made of a material that has good resistance to heat and is a good thermal conductor. The most popular material used for the design of these cones is nickel, although dependent on the application of interest A1, Cu and Pt have been used for this purpose. As the plasma makes contact with the cooled surface of the sampler cone, it cools down rapidly forming a boundary layer between the two. In that layer, ions tend to recombine forming a plethora of molecular species, such as ArC1 + and ArO+, that are not normally observed in bench chemistry. It is crucial that the ions entering the first vacuum chamber are sampled from the central channel of the plasma and not this boundary layer. Hence the aperture of the sampler cone has to be larger than a critical diameter (usually 0.4 mm), but must not be too large as to admit gases from outside the central channel of the ICE Once inside the first vacuum chamber, the sampled gas, which at this stage consists of a mixture of ions, atoms and electrons, expands rapidly. The role of the skimmer cone is to ensure that a representative population of this mixture is extracted into the next stage. In order to achieve this, skimmer cones are designed to have sharper edges and smaller apertures than sampler cones (see Fig. 12.3).

12.1.3 The building blocks of an I C P - M S

Although the ICP and the interface arguably play the central role in ICP-MS, research investigating and characterising different areas of the technique are still being carried out. Figure 12.4 shows a schematic of the building blocks of an ICP-MS. Sample introduction is a much-researched area and often called the "Achilles heel" of ICP-MS [3]. When dealing with liquid samples, it is crucial that the sample introduction unit reproducibly delivers to the ICP a representative portion of the sample. Typically, the sample introduction unit consists of a pump, a nebuliser and a spray chamber. The pump ensures that a constant flow of the sample reaches the nebuliser and is also used to pump out any sample waste from the spray chamber. The nebuliser is used to convert the stream of liquid sample into a fine aerosol and the spray chamber acts as a trap that

Biomedical applications of inductively coupled plasma mass spectrometry

355

Load Coil u

I......... ................................. i Optics /

/

/

//

L ~/ i

Skimmer

........,

"'. Sampler

Torch

I i.i ................... ............ j

To pump Fig. 12.3. Schematic of an ICP-MS interface, the sampled region of the plasma expands behind the sampler whilst a representative portion flows through the skimmer and to the ion optics region.

filters out large droplets. Obviously, it is desirable that the sample introduction unit delivers a very large proportion of the introduced sample to the plasma. However, most nebuliser/spray chamber systems have a very poor efficiency of sample delivery (ca. 2%). This is mainly due to the wide spectrum of droplet sizes delivered by the nebuliser. However, the ICP does not tolerate the disturbances caused by large droplets. Furthermore, since the residence time of the aerosol in the plasma is independent of the aerosol droplet size, the products of small desolvated droplets will have more time to atomise and ionise than those from large droplets (because larger droplets take longer

Sample Intoduction

i ......................................................................................

Mass Spectrometer

Ion Optics

~,0 ~ ~

~F...

-,.-ii~{........

~1 Sample In.-J | Nebuliser .....................

i ~-Spray Chamber .....Waste out Turbomolecular pump

Turbomolecular pump

!

IRotary Pump I Fig. 12.4. A schematic of the building blocks of an ICP-MS.

References pp. 370-371

356

Chapter 12

to desolvate in the plasma). Hence, the plasma is more efficient when handling small aerosol droplets. Therefore, the spray chamber is designed to ensure that only the finest of the aerosol droplets produced by the nebuliser ( < 10 Ixm) reach the ICE which in turn means that most of the sample is lost to waste. Improving the efficiency of the sample introduction system has been achieved through the use of efficient nebulisation techniques such as the use of ultrasonic transducers to shatter the aerosol. With proper design, these so-called ultrasonic nebulisers could provide a nearly 100% efficiency. However, it was observed that with such a high transfer rate, the solvent loading of the ICP becomes excessive and causes changes to the plasma composition/characteristics. This problem was overcome by introducing a desolvating stage prior to the aerosol reaching the plasma. Another approach to improving efficiency is the use of low sample flow rates. High efficiency low flow nebulisers have been used extensively in ICP-MS during the past few years. They enabled the coupling of sample flow rates as low as 10 txl/min. The efficiency of these nebulisers in producing an aerosol with small size droplets is extremely high, however, since the sample consumption rate is low, there is little overloading of the plasma or indeed gain in sensitivity when using these nebulisers. Apart from efficiency, another important analytical issue strongly associated with the sample introduction, is stability. In order to achieve high quality analytical data, instruments must be stable both in the short term and the long term. Short term instabilities are often associated with the pulsating nature of nebulisers. The spray chamber's secondary role is to smooth out these pulses. A typical short-term stability figure for ICP-MS is < 2% relative standard deviation (RSD) over 10 minutes. Longterm instabilities are on the other hand associated with partial blockages of the nebulisers and changes in the physical parameters affecting the performance of the spray chamber namely, temperature and surface wetting. Nebuliser blockages can be overcome by the careful choice of the type of nebuliser suitable for the application. For example, when dealing with clean samples or solvents, standard concentric nebulisers are often the right choice since it is possible to match these nebulisers to the required flow rate. With samples or solvents containing high levels of salts, more salt tolerant nebulisers are recommended. These nebulisers are widely commercially available and the most popular of them are the cross flow nebulisers and the Burgener nebulisers. Finally when dealing with high levels of suspended particulates, V-groove type nebulisers offer the ability to handle these samples or solvents but often at the expense of a 30-50% reduction in sensitivity. Thermally isolating the spray chamber and maintaining it at a constant temperature has now become a standard practice in order to avoid temperature associated long-term drift. This can be achieved either by using a thermoelectric "Peltier" cooling device to control the temperature or by "jacketing" the spray chamber with a continuous flow of water or a cooling fluid flowing from a chiller/recirculator unit. In addition to minimising drift, this approach can be used to selectively reduce the plasma loading from volatile solvents such as alcohol by maintaining the spray chamber temperature between -5~ and - 10~ The choice of material for the manufacture of spray chambers plays an important role in ensuring their compatibility with various solvents and the stability of their

Biomedical applications of inductively coupled plasma mass spectrometry

357

performance. Most common spray chambers are made of quartz. However, inert spray chambers made of perfluoroalkoxy polymers (PFA) or polypropylene are also widely used. As described in Sections 12.1.1 and 12.1.2, the aerosol passing through the spray chamber, is ionised in the ICP and introduced into the vacuum through the interface. After the interface, the ions are guided through a set of ion lenses into the mass spectrometer region. These traditionally electrostatic lenses are designed to steer the ions without changing the composition of the ions beam exiting the skimmer. The magnitude of this undertaking is often underestimated. The difficulty arises from the fact that, as the positive ions are focused, electrons exit the ion beam rendering it rich in positively charged ions that tend to repel each other (columbic effects). Different approaches have been adopted with some success by different commercial ICP-MS manufacturers and they vary from accelerating the ion beam [4] to reducing the ion current by off axis extraction of the ions. Most of these approaches however, involve the use of an increased number of lenses. Although this can complicate the issue of instrument tuning, automated optimisation is often offered since these lenses are all computer controlled. A secondary role of the ion optics is to prevent neutral species and photons from reaching the detectors, since this will cause an elevated background and, as such, poorer detection limits. Different configurations that address this issue are currently in use. They include the use of either photon stops or off axis lenses. More recently, reaction/collision cells have been used as ion guides in ICP-MS. They offer more tolerance for the ions energy spread and interactions than electrostatic lenses. Furthermore, these cells offer the potential of improving detection limits by reducing interferences from other species. The origins of these interferences will be discussed in Section 12.1.4. The most popular form of mass filters used in ICP-MS are quadrupoles. They offer speed of analysis, robustness, stability and ease of use. Other mass filters used include magnetic sector, time-of-flight and ion trap. Each of these mass filters offer advantages and disadvantages over quadrupole systems. Magnetic sector ICP-MS systems offer the potential of resolving interferences from overlapping peaks. However, they tend to be larger and more expensive that quadrupole systems. Time of flight (TOF) ICP-MS offer the potential of detecting whole spectra instantaneously, which could be useful when handling short transient bursts of signals (e.g. ablation of inclusion fluids in rocks). However, the current TOF-ICP-MS systems tend to have lower sensitivity than quadrupole ICP-MS and are yet to be widely accepted in the market [5]. Ion traps are the most recent addition to the ICP-MS family [6]. They offer the capabilities of isolating, storing and, if required, conducting ion-molecule reactions on the ions of interest. The greatest advantage of ion trap ICP-MS is that high efficiency can be obtained from continuous weak beams, through the use of long accumulation times. This advantage is however, not very beneficial when dealing with short transient signals. The main disadvantage of ion trap ICP-MS is that the trap has a limited storage capacity (ca. 106 ions). This is a major limitation for ICP-MS since large beams of Ar +, O +, and other molecular argon ions are often encountered. One means of overcoming this limitation is the use of larger ion traps and the rapid elimination of the abundant ions References pp. 370-371

358

Chapter 12

either by applying a resonant frequency to eject them or by reacting them out with an appropriate reaction gas.

12.1.4 Analytical capabilities of ICP-MS There are a number of analytical features that distinguishes ICP-MS from other analytical techniques. It is a very sensitive instrumental technique for elemental determination with detection limits that compare favourably with, and in most cases exceed those of, other leading techniques for ultra trace element analysis namely graphite furnace atomic absorption (GF-AAS) and neutron activation analysis (NAA). However, unlike GF-AAS, ICP-MS is a multi-element technique that can handle a continuous flow of samples, and unlike NAA, ICP-MS is rapid with total determinations taking only a few minutes whilst in some cases NAA analysis can take months. In terms of speed of analysis, ICP-MS has a similar sample throughput as other rapid analytical techniques such as atomic absorption spectroscopy (AAS) or inductively coupled plasma optical emission spectroscopy (ICP-OES), but outdoes both techniques by nearly three orders of magnitude on detection limits. Typical detection limits for ICP-MS are in the low to sub ng/L range. This applies for most of the elements in the periodic table with the exception of halogens where the high ionisation energy of these elements makes them difficult to ionise in the ICP. Other analytical characteristics to consider when evaluating the technique are interferences and robustness. Interferences can be generally classified as spectroscopic and non-spectroscopic interferences. Spectroscopic interferences are due to overlapping signals. These overlaps can be either due to peaks that share the same nominal mass or due to tailing from adjacent peaks. Peaks that share the same nominal mass with the ion of interest can be due to either elemental ions or to ionised molecular species. An example of elemental ion interferences is the overlap of 5aFe+ and 58Ni+ and an example of molecular ion interferences is the overlap of 56Fe+ and 4~ Whilst the origin of the interfering elemental ions is in most cases elements in the sample that are ionised in the plasma, molecular ions are usually due to recombinations taking place in the interface region (since as described earlier the plasma cools down rapidly in that region) and involving the major components of the sample such as O, N, C combining either with each other or with Ar from the plasma gas. Identifying the presences of these interferences is normally straightforward since these interferences tend to have their own isotopic patterns which, when present, distort the isotopic pattern of the element of interest. Overcoming these interferences can be achieved either by using an isotope which is free of interference (e.g. in the case of the iron interference on 58Ni+, 6~ can be used) or by adopting a mathematical correction that takes into account the isotopic distribution of the interfering species to correct for its contribution (e.g. in the case of the 5~Fe+ interference on 58Ni+, the contribution from 58Fe+ can be calculated as the Signal for 56Fe+ multiplied by the ratio of the abundances of 5SFe/56Fe). In extreme cases, where there is no appropriate alternative isotope to use and where the magnitude of interference is so much larger than the signal of interest such that the uncertainty from the interference correction can lead to erratic results, high resolution ICP-MS

Biomedical applications of inductively coupled plasma mass spectrometry

359

(HR-ICP-MS) can be used to physically separate the two peaks. Typically the separation power of an HR-ICP-MS system is reported as the ratio of the mass of the ion of interest over the mass difference between the two ions, and its working range is from 400 up to 10,000. The higher the resolution figure the better the separation. However, this is done at the expense of sensitivity. For example in the case of 58Fe+ interference on 58Ni+, the mass difference between the two ions is 5 7 . 9 3 5 3 4 7 9 - 57.9332805 = 0.002067 and the required resolution to separate them is 57.9353479/0.002067 = 28,023, which is outside the range of a HR-ICP-MS. Whereas the resolution required to separate 56Fe+ and 4~ is 55.9349421/0.022356 = 2502, which is well within the capabilities of the system. A more recent approach for addressing these spectroscopic interferences is the use of collision/reaction cell instruments where one of the interfering species is reacted out [7,8]. For example, with a reaction cell instrument, the addition of hydrogen in the cell will react out the ArO § ions to form ArO neutrals and H +, H 3+ species which do not interfere with 56Fe+. Tailing from adjacent peaks is dependent on the abundance sensitivity of the system. Most ICPMS instruments have an abundance sensitivity of < 10-6 on the low mass side and < 10-7 on the high mass side. In other words, if the Signal of 56Fe+ is 100,000,000 counts per second (cps), there will be < 100 cps contribution from this signal of at mass 55 and < 10 cps at mass 57. This problem is not often a major issue with ICP-MS. However, if needed, either H R - I C P - M S or collision/reaction cell ICP-MS can be used to overcome it. Non-spectroscopic interferences are usually manifested in the form of a response change for the element of interest as a result of influence from the sample matrix or some of its components. Typical forms of non-spectroscopic interferences observed in ICP-MS are: (a) reduction in sensitivity as a result of partial blockages of the nebuliser or the cones; (b) changes in sensitivity due to changes in nebulisation characteristics as the matrix/solvent changes; (c) Signal suppression from high levels of easily ionisable elements in the sample (e.g. > 1000 ppm Na) which can increase the electron density in the central channel of the plasma, thereby lowering the ionisation efficiency (see Appendix 1); and (d) Selective loss of sensitivity for light elements in the presence of high levels of heavy elements (e.g. > 100 ppm U) due to columbic repulsion in the interface and the early elements of the ion optics. The magnitude of these effects does vary with different designs but in general, careful selection of the nebulisation system and cone apertures will minimise them dramatically. Nebuliser blockages can be eliminated by selecting more tolerant nebulisers whist the effect of the solvent change is often non observable [9].

12.2 ICP-MS AS AN ELEMENT SPECIFIC DETECTOR FOR CHROMATOGRAPHIC SEPARATIONS 12.2.1 Coupling an HPLC to ICP-MS Coupling HLPC to ICP-MS is relatively simple to achieve. The flow rates from HPLC systems, which can vary between 1 ixl/min up to 2-3 ml/min, are directly compatible References pp. 370-371

360

Chapter 12

with the sample introduction flow rates in ICP-MS. As such HPLC-ICP-MS especially ion chromatography (IC) ICP-MS was used as a tool for identifying elemental species since as far back as the late 1980s [10]. However, it rapidly became obvious after the initial euphoria that HPLC-ICP-MS as an integrated technique cannot be effected by simply linking the outlet of the LC to the inlet of the ICE Considerations related mainly to the mobile phases in use had to be taken seriously. The high levels of salts used in most IC mobile phases caused blockages at the nebuliser tip and cone apertures. Furthermore, mobile phases consisting of a high percentage of an organic solvent (such as those used mainly in reversed phase HPLC) cause rapid blockages of the cones due to the deposition of carbon. In order to address the issue of salt, on line dilution as well as exchange membranes, whereby Na, K and Ca ions are exchanged with hydrogen ions, were used. With regard to handling organic solvents, the problem arises from the fact that carbon has a relatively high ionisation energy (11.3 eV), which as a result leads to a lower ionisation efficiency in the plasma. Hence with high levels of organic solvents there will be a large population of carbon atoms in the plasma. These atoms tend to deposit on the apertures of the sampler and skimmer cones, thereby blocking them. In order to address this problem, the levels of organic solvents reaching the plasma have to be reduced. This is achieved by using low flow rates (< 200 txl/min) and desolvating the aerosol generated by the nebuliser prior to it reaching the plasma through cooling the spray chamber to sub-zero temperatures (typically -5~ In addition, oxygen is bled through the plasma in order to assist in burning the excess carbon. However, this process can change the electrical impedance of the ICE Therefore, unless the RF circuitry of the instrument is designed to cope with this impedance shift, the plasma will extinguish. Purpose made desolvating nebulisers are currently commercially available and are often used to couple HPLC to ICP-MS. With these nebulisers, the generated aerosol, is first heated, then in one case partially desolvated through passing the hot mixture of gas and aerosol into a thermally controlled chamber kept at sub-zero temperatures, and finally desolvated through passing the mixture in a heated semi porous membrane with a counter flow of Ar on the outside. Although, these nebulisers provide an ideal way for linking HPLC to ICP-MS, volatile species can be lost through the membrane.

12.2.2 Coupling GC to ICP-MS GC-ICP-MS has potentially superior detection power to HPLC-ICP-MS. This is mainly because, GC has a superior resolution, and the low gas flows from the GC can be directly injected into the plasma thereby providing a nearly 100% efficiency in terms of the transport of the analytes. In their paper in 1986, Van Loon et al., were the first to report on the coupling of a GC to ICP-MS for the purpose of detecting organotin compounds [11]. This was quickly followed by a number of papers by various groups. With the advance of capillary GC, Kim et al. reported on the use of a heated fused silica capillary transfer line to affect the GC-ICP-MS coupling [12]. Although commercial GC-ICP-MS transfer lines are currently available, most researchers developed their own in-house systems [ 13,14]. The designs are basically simple with a deactivated but uncoated fused

Biomedical applications of inductively coupled plasma mass spectrometry

361

Pe~k A-ea 100000 -

C8 C 12 C 14 C 16 C 18 C20 C 22

O

75000

e,O Ill q

A

m

cN

~.14[,~54,

C 26

2.56E*'04

9. . . . .

o.

N

2, I'~E,N)4 3.9TE'H~4 3.79E,04 7,96E.H)4 4,05E,~, 4,17E~)4 3,93E*04

,,,,., ,,,, , , , . . . , .

,.,,,.,.,,...... ,.,..

'r

50000

25000

0

.

.

.

.

.

L____

.aL

.

.

.

.

Ill-

I

600

700

11

300

4OO

500

-

J

Retention time [s] Fig. 12.5. Separationof an n-paraffin standard with C16 added directly into the solvent [15].

silica running inside a heated tube that extends between the GC and the tip of injector in the plasma torch. Heated argon is made to flow inside the heated transfer line around the fused silica. This makes up the flow required to punch the plasma and also ensure uniform heating of the fused silica. Some of the limitations observed with current GC-ICP-MS transfer lines were discussed recently by Glindemann et al. [15]. They related to signal repeatability, a potential carbon deposit from high organic solvents, and the limitations of current designs to handle high boiling analytes. The authors proposed a newer interface, based on rapid dilution of the effluent gases with hot argon and solvent peak elimination by periodic flow reversal in the transfer line. Using this interface, the authors could separate C26 n-paraffin (412~ at only 140~ temperature in the transfer line (Fig. 12.5).

12.2.3 Coupling CE to ICPMS Olesik et al. first reported on the coupling of CE to ICP-MS in 1995 [ 16]. It was clear from their work that the interface must, in addition to providing a stable electrical current across the capillary, prevent the introduction of a laminar flow through the capillary. The issue arises from the special nebulisers needed to handle the low flows of CE. These nebulisers tend to be self-aspirating. This will create a suction effect that causes a laminar flow through the capillary and compromises the efficiency of the electrophoretic separation.

References pp. 370-371

362

Chapter 12

Most CE-ICP-MS publications to-date focus on the design of the CE interface. The most popular interface is based on a micro-concentric nebulizer, and uses a coaxial electrolyte sheath flow (make-up) to establish the electrical connection to the analyte. Recently Cetac technologies (Omaha, USA) released a commercial CE interface that employs a micro-concentric nebuliser and with the nebuliser uptake rates minimised to allow for minimal dilution (make up flow rate of < 10 ~l/min) without affecting the separation. In addition to these interfaces, direct injection nebulisers (DIN) have also been investigated as a possible CE-ICP-MS interfacing tool [17]. Although these nebulisers can offer nearly 100% efficiency in terms of sample delivery, their performance is offset by their high cost. A more affordable version of these nebulisers called the direct injection high efficiency nebuliser (DIHEN), has recently been made commercially available (J.E. Meinhard Associates, Inc., Santa Ana, CA, USA). These nebulisers connect directly to the back of the plasma and are capable of handling extremely low flows (< 10 pA/min) whilst maintaining a very efficient sample delivery rate. Majidi et al. reported that using this type of nebuliser reduced peak broadening and resulted in sharp symmetrical peaks and improved signal to noise performance [18]. Similarly, Bendahl el al [19] used DIHEN to separate Se species by CE-ICP-MS and reported detection limits in the range of 25-125 fg (0.3-1.5 fmol) of selenium.

12.3 APPLICATIONS OF ICP-MS IN THE BIOMEDICAL FIELD 12.3.1 Detection of metabolites

The full characterization of the chemical, pharmacological and toxicological properties of any biologically active substance, including its metabolites, is a prerequisite that most current regulatory authorities demand. Hence, the detection and identification of impurities or metabolites structurally related to these substances are of the utmost importance. ICP-MS as an element specific detector can play a major role in this process. Traditionally, the determination of the metabolic fate of drugs and other xenobiotics in experimental animals and man required the radiochemical labelling of compounds. In the absence of a suitable label, the detection of metabolites can be a difficult exercise, especially if they are significantly different in structure from the parent. Furthermore, even when the detection and identification of metabolites is possible using traditional MS and UV detectors, quantification is more challenging as the MS and UV response of the metabolites may significantly differ from that of the parent. Since a large number of these biologically active materials contain heteroatoms, the specificity of ICP-MS for detection of those atoms, coupled with its detection power leads to much simpler chromatograms and rapid identification of the retention times of the compounds of interest. Knowing these retention times, it is possible then to use molecular mass spectrometry such as electrospray ionization mass spectrometry (ESIMS) in order to identify them. Furthermore, the sensitivity of ICP-MS is element specific but compound independent therefore, by knowing the concentration of one

Biomedical applications of inductively coupled plasma mass spectrometry

363

compound in the detected chromatogram, it is possible to accurately calculate the concentration of the rest of the detected species. This information will be invaluable for "mass balance" studies where the amount of starting material is compared to the total amount of detected impurities, degradation products or metabolites found together with the remaining starting material. In the case of substances bereft of heteroatoms, tags can be used to provide these atoms. ICP-MS can then be used to detect and quantify the complexes containing these tags. A number of papers have recently reported the use of ICP-MS for tracing the metabolic fate of xenobiotics and drugs. In a short communication, Nicholson et al. investigated the fate of 4-bromoaniline in the urine of rats [20]. Their results highlighted the potential of the technique as a simple, rapid and specific method for the detection of metabolites. This paper was rapidly followed be a series of investigations from the group profiling Dichlofenac (using C1 and Sulfur detection) [21] and 2-bromo4-trifluoromethylaniline [22,23]. In that later work, parallel analysis of the chromatographic eluent was carried out using ICP-MS, ESI-MS (with a time of flight mass analyser) and diode array detection. This combination proved to be a powerful tool for immediate identification and quantification of the metabolites (see Fig. 12.6). A similar approach was adopted by the group of Marshall et al. in their investigation of bradykinin metabolism in human and rat plasma [24,25]. Bradykinin is a bioactive peptide produced from the actions of kininogenases on kininogens and exerts its role by activating B2 receptors. However, in order to detect bradykinin by ICP-MS, the peptide had to be labelled with an appropriate heteroatom. The group adopted what they called a "labelled ligand" approach in which bromine was used as a tag. The structures of bradykinin and its tagged product are shown in Fig. 12.7. Another interesting work on metabolite profiling by HPLC-ICP-MS was carried out by Duckett et al. [26]. The

100-

A

'

.|

9

I

i

!

~

100: o..,,

~:

B .:

~" "~

' I00,

CI Bll"./~D A\; l

,E I

C

Fig. 12.6. (A) UV-DAD, (B) bromine-detectedICPMS total ion current trace for the bromine isotpes 79 and 81, and (C) oa-TOF total ion current traces obtained for the reversed-phase gradient HPLC separation of the 0-8 h postdose rat urine samples. References pp. 370-371

364

Chapter 12

an

Arg -P ro=-P ro 3-G ly 4- P h e s-S e r6- P ro 7_p h e 8-A rg 9 SN2_Arg -Pro=-Pro3-Gly4-p-BrPheS-Ser6-ProT-Phe8-Argm

b-

Fig. 12.7. (a) Structure of bradykinin, (b) structure of bromo-bradykinin used in the work of Marshall et al. [24].

authors investigated the profile produced by the earthworm Eisenia veneta following exposure to 2-fluoro-4-iodoaniline. The ICP-MS profile was produced by tracing the 127I signal, and the authors used a post column injection of a known concentration of iodine to work out the amount of metabolite found in the various parts of the worm as a percentage of the initial dose of the drug. Hywel Evans et al. [27], used ICP-MS to detect impurities in Cimetidine based upon monitoring the presence of sulphur in the eluents following chromatographic separation of a solution containing the drug. Their data indicated the presence of compounds that were not detected in earlier studies using ESI-MS. The higher detection power of the ICP-MS is also highlighted in a recent work by Smith et al. [28]. In this work, blood samples from animals treated with a platinum-based anticancer drug were analysed for the presence of the drug and its biologically active compounds. This study provided a direct comparison between ICP-MS and triple quadrupole atmospheric pressure ionisation mass spectrometry (API-MS-MS). Although API-MS-MS is acknowledged as the standard technique for conducting such investigations, the result of the study showed that ICP-MS did out-perform it in terms of quantification limits (0.1 ng/mL for HPLC-ICP-MS compared to 5 ng/mL for the HPLC-MSMS method) and linear dynamic range (over five orders of magnitude for HPLC-ICP-MS compared to typically three orders of magnitude for the HPLC-MSMS method. Axelsson et al. [29] reported on the potential of the technique for general organic compounds related to the pharmaceutical industry. The paper discussed a variety of compounds including iodine containing X-ray contrast substances and gadolinium metal complexes. Metal complexes can be used as tags for specific antibodies that can then target antigens of interest and make them detectable using ICP-MS. In addition, the paper discussed the detection of phospholipids using phosphorus as the target element. Furthermore, the paper addressed the issue of phosphorylation which is again a major potential application field of ICP-MS.

12.3.2 Phosphorylation detection by ICP-MS Phosphorylation is a key process in the regulation of protein activity and has long been appreciated as an essential mechanism for the control of cellular function [30]. Phosphorylation reactions are associated with numerous biological processes and as such they attract at lot of interest from biomedical researchers. This research does not just aim at identifying the presence of phosphorylation but also at pinpointing the exact position of the phosphorylated amino acid. Typical analytical techniques used for this purpose involve the digestion of proteins followed by HPLC-MS-MS and the tracing of

365

Biomedical applications of inductively coupled plasma mass spectrometry

the indicative PO3 ion at mass 79 that the phosphorylated peptides yield under collisionally induced dissociation. Alternatively, proteins are separated using gel electrophoresis, and this step is followed either by the analysis of the gel using MALDIMS or by excising specific spots and analyzing them using HPLC-MS-MS. However, with both techniques, the ionization efficiency is compound-dependent, resulting, for example, in reduced ionization efficiency for phosphopeptides compared to unmodified peptides. As a result, information about the phosphorylation status of a protein using these methods is only qualitative and large phosphopeptides with poor fragmentation characteristics (mol wt > 2500) may escape their recognition as such. Other techniques involve the incorporation and subsequent detection of radioactive phosphorus (e.g. 32p, 33p) or the use of anti-phosphopeptide antibodies, but again each technique does suffer from specific limitations. Axelsson et al. [29] used HPLC-ICP-MS to distinguish between mono-phosphorylated and di-phosphorylated peptides. Their results showed that the technique was highly selective since no non-phosphorylated peptides were detected whilst it was easy to identify mono- and di-phosphorylated peptides that were at similar concentrations. Furthermore, a tryptic digest of [3-caesin was analysed using the technique and the resulting chromatogram as compared to UV detection is shown in Fig. 12.8. The detection capability for the HPLC-ICP-MS method in this study was however limited 100-

!4. 11,

.

4

- -, 38.56

loo-

3.49 34.23 22,12 28,53 32,55 I~. t t )I 2 7 0 2 -~ 1~ J l

.

8,70 .84

11.32 ' 15.32 /t ~,

40.33

,..,m.w~,~.,=,=, I .

....

I"

--;

....

5.00

T ....

I ....

10.00

I""

"'!

.....

15.00

r''~'T

'''''r

20.00

....

! ....

25.00

I ....

I ....

30.00

I ....

,

r"

--"

35.00

~lr ....

1

. . . .

40.00

~

. . . .

I

r ~ r ~

4S,O0

Fig. 12.8. LC-ICP-MS(top) and LC-UV(bottom) traces of a tryptic digest of [3-casein. The LC-ICP-MS trace corresponds to the presence of phosphorus in the eluting components (SIR of m/z = 31). References pp. 370-371

366

Chapter 12

due to background interferences on mass 31 from NOH and NO species, that resulted in a relatively elevated baseline. The use of high resolution ICP-MS which can provide a mass spectrometric separation between phosphorus and the above mentioned ions, as well as the use of reaction cell-based ICP-MS, which offers the potential to selectively react some of the ions extracted from the ICP have led to an improvement in the detection power of the technique. Wind et al. [31 ] used capillary LC coupled to high resolution ICP-MS to detected low levels of P in a mixture of synthetic phosphopeptides, a tryptic digest of activated MAP kinase and protein kinase A catalytic subunit. The data using this technique compared well with LC-ESI-MS. The same author then published work using both phosphorus and sulfur detection for accurately determining the degree of phosphorylation for cysteine- and methionine-containing proteins and peptides [32]. Marshall et al. [33] used a laser ablation ICP-MS to investigate phosphorylation. In this system shots from a UV-laser (213 nm) were used to ablate fine spots (ca. 120 ~m in diameters) from the surface of gels and gel blots and the ablated material was swept directly into the ICP-MS. The authors reported some success with the gel blots being able to easily measure low pico-molar levels of [3-caesin. However, with electrophoresis gels, the authors reported the presence of a very high baseline that affected the analytical performance of the technique

12.3.3 Other applications ICP-MS has also been coupled with chromatographic separation to evaluate the biological effects of some metal/semi metal elements that enter the human body through either the food chain, environmental activities or a combination of both. One element that attracted such an attention is arsenic. In nature, arsenic is often found in sulphidic ores in the form of metal oxides. However, the commercial use of arsenic compounds as arsenical pesticides, herbicides and crop desiccants and the use of arsenic as an additive to livestock feed, particularly for poultry has raised its environmental levels and exposed humans to relatively higher levels of this element. HPLC-ICP-MS has been used to investigate the fate of arsenic in humans through determining the distribution of arsenic species in human/animal urine [34], and hair [35]. The technique has also been applied for the analysis of water [36] and seafood products [37]. More recently arsenic trioxide, a well-known carcinogen, has been used for the treatment of leukemia in humans [38]. This report agrees fully with data from our laboratory where we observed only very small amounts of dimethylarsinic acid (DMA) in the urine of control subjects from china, whereas the urine of patients from the same region treated for leukemia showed very high levels of DMA as well as monomethylarsonic acid (MMA) and inorganic arsenic(V) ions (Fig. 12.9). Furthermore, in order to highlight the role of diet, we show in Fig. 12.9 a chromatogram of the urine from a control subject on a rich seafood diet. The chromatogram shows extremely high levels of arsenobetaine, which is rich in various types of fish. Other forms of arsenic that could be taken through diet are arsenosugars and arsenocholine, which exists at very high levels in seaweeds [39].

Biomedical applications of inductively coupled plasma mass spectrometry

367

3600 Control Subject 2700

(a)

1800 900

~IL U,L~B~,.~L~"-..... J' '"' ~'' . ~ . J i a l ~ , l.~_ ,fl_l t",~,v-~'~'T .... ,-m""~rl,rr~ 1',~ .l'f

~]I~~iI~,1~~1~

80000

Treated patient 60000

(b) DMA

40000 MMA 20000 As(V) J .

200000

.

.

.

.

.

.

.

.

.

.

.

.

.

, r . . . . . . . .

.

I

......

11

Subject on seafood diet

150000

(c)

AsB

100000 50000 .,

0

5

-- ~

....

A

.

.

.

10

.

.

.

.

.

~- . . . . . . . . . . . . . .

15

,

20

Time (min}

Fig. 12.9. Arsenic species in the urine of: (a) a control subject with extremely low levels of As in the form of DMA observed; (b) a treated patient with elevated levels of DMA, MMA, As(V) as a result of intake of arsenic based drugs; and (c) a control subject on seafood diet with very high levels of Arsenobetaine (AsB) the most common form of As in some types of fish. Another element that has been investigated intensely is selenium. Selenium is best known for its role at the active center of the enzyme glutathione peroxidase, which protects against oxidative stress. It is an essential trace element required for the functioning of a number of different enzymes and its deficiency has been associated with several disease states, including coronary heart disease and cancers. As a result, selenium supplementation is nowadays a very c o m m o n practice. This in turn has led to a plethora of papers attempting to identify the various species of selenium observed in food products, food supplements and human and animal excretions. A good review of these papers is the technical report by the "commission on microchemical techniques and trace analysis" from the Analytical Chemistry Division at IUPAC (International Union of Pure and Applied Chemistry) [40]. Traditionally, the analysis of selenium by I C P - M S at low levels has been limited by isobaric interferences from ArAr molecules. However, the use of collision/reaction cell based I C P - M S systems, has enabled such analysis to be carried out more efficiently. Using such a system M a r c h a n t e - G a y o n

References pp. 370-371

368

Chapter 12

reported on the speciation of Se in urine and quote detection limits in the range of 30-80 pg/ml (1.5-4 pg on column) [41 ]. These values are an order of magnitude lower than the best previously reported data of 50-218 pg by Lafuente et al. [42]. More recently, Gammelgaard et al. [43], used a traditional ICP-MS coupled to a low flow ion pairing chromatographic system and managed to obtain detection limits of 800-1700pg/ml. However, since with this chromatographic system the injection volume is only 3 txl as compared to 50 txl by Marchante-Gayon, the absolute detection limits (pg on column) of the two methods are comparable. Therefore, one could safely assume that using this ion pairing method coupled to a collision/reaction cell instrument, the detection limits of Se will be pushed even lower making the potential of detecting low femtomolar concentrations of these compounds a real possibility. Toxic organic species of tin and mercury have also been analysed by ICP-MS using GC separations. Organotin compounds, are introduced into the environment through their use as antifouling paints on ships and also as wood preservatives, fungicides, biocides, and polymer additives. Detection of these species was reported in the very first GC-ICP-MS paper by Van Loon et al. [11]. More recently, Wahlen [44] published a comparison between GC-ICP-MS and HPLC-ICP-MS for the analysis of tin species. The paper concluded that whilst GC-ICP-MS is a more sensitive technique (3 pg of tributyltin detection limits by HPLC-ICP-MS compared to 0.03 pg by GC-ICP-MS), complex sample preparation procedures meant that HPLC-ICP-MS is a cheaper and much easier technique to use for the analysis of large sample batches. Human exposure to mercury is mainly due to the accumulation of methyl mercury in fish. In humans, methyl mercury exhibits neurotoxic effects, hence there is an interest in monitoring its levels especially at source, that is in fish. A number of scientists have used GC-ICP-MS for this purpose with typical detection limits reported to be in the region 0.02-0.08 pg [45,46]. Finally, mixed mode HPLC has been coupled to ICP-MS in order to identify some of the protein-bound trace elements in human serum [47]. The separation allowed proteins/large size molecules to elute in the first 3 minutes whilst small molecules such as free ions or amino acids eluted at times > 3.5 min. Based on the data shown in Fig. 12.10, the authors concluded that Cu, Zn, and Fe were bound to proteins such as albumin ceruloplasmin and transferrin. Se, Mg, Ca, Sr, Ba and I, on the other hand could be found in both proteins and small molecules, whilst Mo, Br, and C1 were only observed as free ions or bound to small molecules. This technique could be a valuable tool as a rapid screening method for the distribution of trace elements in biological fluids.

12.4 SUMMARY Inductively coupled plasma mass spectrometry is a powerful analytical technique that is recently finding popularity as an element specific detector for chromatographic separations. This has opened the door for novel research in the biomedical field investigating the metabolic fate of drugs, phosphorylation reactions, the role of essential trace elements in human health, and the mechanisms in which toxic elements affect the

"-,,i

t~ --q

%

o

|

t.,u

a6Zrl

2

,1

.

3

.

.

.

I,

8

9

1o0~r

I !0000c~.,

4 5 6 7 Retention time / rain

~k..~ __

j

(B)

-~ 0.05

(A)

110

.,,..,,

0

i

l

i

Br _ i

i

2

_

'-'sB~_ __J

I

i

i

i

i

i

I

i

i

i

i

8

i

i

9

i

I 100000 cps

3 4 5 6 7 Retention time / rain

i

A