Bioanalytical Separations, Volume 4 (Handbook of Analytical Separations)

Bioanalytical Separations H A N D B O O K OF A N A L Y T I C A L S E P A R A T I O N S Series Editor: ROGER M. SMITH ...

Author: Ian D. Wilson

82 downloads 1434 Views 23MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Bioanalytical Separations

H A N D B O O K OF A N A L Y T I C A L S E P A R A T I O N S Series Editor: ROGER M. SMITH

In this series:

Vol. 1: Separation Methods in Drug Synthesis and Purification Edited by K. Valk6 Vol. 2: Forensic Science Edited by M.J. Bogusz Vol. 3: Environmental Analysis Edited by W. Kleib6hmer Vol. 4: B ioanalytical Separations Edited by I.D. Wilson

HANDBOOK

OF

ANALYTICAL

SEPARATIONS

-

VOLUME

4

9 l y " tlcal Separations " B~oana

Edited

by

IAN D. WIL S ON AstraZeneca Macclesfield, U.K.

2003 ELSEVIER Amsterdam - Boston - Heidelberg - London - New York - Oxford - Paris San Diego - San Francisco - Singapore - Sydney - Tokyo

ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 EO. Box 211, 1000 AE Amsterdam, The Netherlands 9 2003 Elsevier Science B.V. All rights reserved. This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier's Science & Technology Rights Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail: [email protected]. You may also complete your request on-line via the Elsevier Science homepage (http://www.elsevier.com), by selecting 'Customer Support' and then 'Obtaining Permissions'. In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: ( + 1) (978) 7508400, fax: ( + 1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 207 631 5555; fax: (+44) 207 631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier's Science & Technology Rights Department, at the phone, fax and email addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. First edition 2003 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for. British Library Cataloguing in Publication Data A catalogue record from the British Library has been applied for.

ISBN: 0-444-50658-6 ~) The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.

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.

vii

Contents P refa c e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 1.

1.1 1.2

1.3

1.4 1.5 1.6 1.7

New developments in integrated sample preparation for bioanalysis

M.W.J. van Hour, H.A.G. Niederl~inder, R.A. de Zeeuw and G.J. de Jong . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chromatography-based extraction techniques . . . . . . . . . . . . . . . . 1.2.1 Solid-phase e x t r a c t i o n - gas chromatography . . . . . . . . . . . . 1.2.1.1 General aspects of SPE . . . . . . . . . . . . . . . . . . . 1.2.1.2 Applications of S P E - G C . . . . . . . . . . . . . . . . . . 1.2.1.3 Remarks regarding the applicability of S P E - G C . . . . . 1.2.2 Liquid c h r o m a t o g r a p h y - gas chromatography . . . . . . . . . . . 1.2.2.1 General aspects of L C - G C . . . . . . . . . . . . . . . . . 1.2.2.2 Applications of R P L C - G C . . . . . . . . . . . . . . . . . 1.2.2.3 Remarks regarding the applicability of R P L C - G C . . . . 1.2.3 Turbulent-flow chromatography . . . . . . . . . . . . . . . . . . . 1.2.3.1 General aspects of TFC . . . . . . . . . . . . . . . . . . 1.2.3.2 Applications of TFC . . . . . . . . . . . . . . . . . . . . 1.2.3.3 Remarks regarding the applicability of TFC . . . . . . . . Diffusion-based extraction techniques . . . . . . . . . . . . . . . . . . . . 1.3.1 Solid-phase microextraction . . . . . . . . . . . . . . . . . . . . . 1.3.1.1 General aspects of SPME . . . . . . . . . . . . . . . . . 1.3.1.2 Applications of S P M E - L C . . . . . . . . . . . . . . . . . 1.3.1.3 Applications of S P M E - M S . . . . . . . . . . . . . . . . 1.3.1.4 Remarks regarding the applicability of S P M E . . . . . . . 1.3.2 Membrane-based sample preparation techniques . . . . . . . . . . 1.3.2.1 General aspects of membrane-based techniques . . . . . . 1.3.2.2 Porous membrane techniques . . . . . . . . . . . . . . . 1.3.2.3 Non-porous membrane techniques . . . . . . . . . . . . . 1.3.2.4 Remarks regarding the applicability of membrane-based techniques . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 4 4 4 5 7 7 7 9 11 12 12 13 14 16 16 16 18 24 24 26 26 28 32 35 35 38 38 39

viii

Contents

Chapter 2. Solid-phase extraction on molecularly imprinted polymers 2.1 2.2

2.3

2.4

2.5 2.6

Lars I. A n d e r s s o n and L e i f S c h w e i t z . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I m p r i n t preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 R e m o v a l of t e m p l a t e m o l e c u l e s . . . . . . . . . . . . . . . . . . . 2.2.2 C h o i c e of t e m p l a t e . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3 F o r m a t of p o l y m e r . . . . . . . . . . . . . . . . . . . . . . . . . . M I S P E m e t h o d d e v e l o p m e n t strategies . . . . . . . . . . . . . . . . . . . 2.3.1 Non-specific adsorption . . . . . . . . . . . . . . . . . . . . . . . 2.3.2 S o l v e n t switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.3 E l u t i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.4 T e m p l a t e b l e e d i n g . . . . . . . . . . . . . . . . . . . . . . . . . . S o l i d - p h a s e extraction applications . . . . . . . . . . . . . . . . . . . . . 2.4.1 O n - l i n e extraction s y s t e m s . . . . . . . . . . . . . . . . . . . . . . 2.4.2 Extraction systems with direct detection . . . . . . . . . . . . . . . 2.4.3 Off-line extraction s y s t e m s . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 3.

3.1 3.2 3.3

3.4 3.5 3.6 3.7 3.8

4.1 4.2

Techniques for sample preparation using solid-phase extraction

U w e Dieter Neue, C l a u d e R. Mallet, Ziling Lu, Y u n g - F o n g C h e n g and Jeffrey R. M a z z e o . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D e s c r i p t i o n of the sorbents . . . . . . . . . . . . . . . . . . . . . . . . . . Off-line M e t h o d s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 1-D R e v e r s e d - p h a s e solid p h a s e extraction of biological s a m p l e s 3.3.2 2-D R e v e r s e d - p h a s e solid p h a s e extraction of biological s a m p l e s C a t i o n - e x c h a n g e solid p h a s e extraction . . . . . . . . . . . . . . . . . . . A n i o n - e x c h a n g e solid p h a s e extraction . . . . . . . . . . . . . . . . . . . On-line methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 4.

45 45 46 48 48 50 52 53 53 53 55 55 55 63 63 68 69

. .

73 73 74 75 76 77 82 84 85 89 89

Turbulent flow chromatography in bioanalysis

Tony E d g e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Band broadening processes . . . . . . . . . . . . . . . . . . . . . 4.2.2 Theoretical interpretation . . . . . . . . . . . . . . . . . . . . . . 4.2.3 D e s c r i p t i o n of the van D e e m t e r constants . . . . . . . . . . . . . . 4.2.3.1 A term . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3.2 B t e r m . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3.3 C term . . . . . . . . . . . . . . . . . . . . . . . . . . .

91 91 92 92 93 94 94 96 96

ix

Contents

4.2.4

4.3

4.4 4.5

D e v e l o p m e n t of turbulent flow c h r o m a t o g r a p h y m o d e l . . . . . . . 4.2.4.1 Definition of turbulent flow . . . . . . . . . . . . . . . . 4.2.4.2 Definition of turbulence . . . . . . . . . . . . . . . . . . 4.2.5 O v e r c o m i n g the p r o b l e m of pressure drop . . . . . . . . . . . . . . 4.2.6 Practical investigation . . . . . . . . . . . . . . . . . . . . . . . . 4.2.7 Mass transfer into pores . . . . . . . . . . . . . . . . . . . . . . . 4.2.8 C o m b i n i n g mass transfer and pressure drop . . . . . . . . . . . . . Applications of turbulent flow c h r o m a t o g r a p h y . . . . . . . . . . . . . . . 4.3.1 A p p l y i n g the m o d e l . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1.1 Single valve m e t h o d . . . . . . . . . . . . . . . . . . . . 4.3.1.2 Q u i c k elute m o d e . . . . . . . . . . . . . . . . . . . . . . 4.3.1.3 F o c u s m o d e . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.2 A p p l i c a t i o n areas . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.2.1 D r u g m e t a b o l i s m and p h a r m a c o k i n e t i c ( D M P K ) studies . 4.3.2.2 F o r e n s i c applications . . . . . . . . . . . . . . . . . . . . 4.3.3 Practical issues in bioanalytical T F C . . . . . . . . . . . . . . . . 4.3.3.1 Carryover . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.3.2 Pressure build up . . . . . . . . . . . . . . . . . . . . . . 4.3.3.3 Protein binding . . . . . . . . . . . . . . . . . . . . . . . 4.3.4 E n v i r o n m e n t a l applications of T F C . . . . . . . . . . . . . . . . . 4.3.5 Capillary turboflow c h r o m a t o g r a p h y . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 5. 5.1

5.2

97 99 102 102 106 107

110 111 111 112

112 115 115 115 121 123 124 124 125 125 126 127 127

Chiral bioanalysis

D.M. W a l l w o r t h and J.T. L e e . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Scope and aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2 The m e c h a n i s m of chiral r e c o g n i t i o n and choice of C S P . . . . . 5.1.3 M o b i l e p h a s e types . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.4 Direct versus indirect chiral separations . . . . . . . . . . . . . . 5.1.5 A c h i r a l - c h i r a l c o l u m n switching techniques . . . . . . . . . . . . 5.1.6 HPLC-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.7 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.8 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.9 Gas L i q u i d c h r o m a t o g r a p h y ( G L C ) . . . . . . . . . . . . . . . . 5.1.10 Capillary electrophoresis (CE) . . . . . . . . . . . . . . . . . . . 5.1.11 Supercritical fluid c h r o m a t o g r a p h y (SFC) . . . . . . . . . . . . . Chiral stationary phases . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 M a c r o m o l e c u l a r or p o l y m e r i c CSPs . . . . . . . . . . . . . . . . 5.2.2 Protein phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Cyclodextrins . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4 M a c r o c y c l i c antibiotics . . . . . . . . . . . . . . . . . . . . . . . 5.2.5 ~r-Complex C S P s . . . . . . . . . . . . . . . . . . . . . . . . . .

129 129 130 130 131 133 133 133 135 135 136 136 136 137 137 139 140 143 145

Contents 5.3

A p p l i c a t i o n s o f chiral H P L C in b i o a n a l y s i s . . . . . . . . . . . . . . . . .

147

5.3.1

[3-Adrenergic a g o n i s t s

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

147

5.3.2

[3-Adrenergic b l o c k e r s

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

150

5.3.3

Alcohol deterrent drugs . . . . . . . . . . . . . . . . . . . . . . .

153

5.3.4

A m i n o acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

153

5.3.5

A n a l g e s i c d r u g s (narcotics) . . . . . . . . . . . . . . . . . . . . .

155

5.3.6

Analgesic drugs (non-narcotic) . . . . . . . . . . . . . . . . . . .

156

5.3.7

A n e s t h e t i c drugs ( i n t r a v e n o u s ) . . . . . . . . . . . . . . . . . . .

157

5.3.8

Anorexic drugs . . . . . . . . . . . . . . . . . . . . . . . . . . .

157

5.3.9

Anthelmintic agents . . . . . . . . . . . . . . . . . . . . . . . . .

157

5.3.10

Antiarrhythmic agents

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

158

5.3.11

Antibacterial drugs . . . . . . . . . . . . . . . . . . . . . . . . .

160

5.3.12

Anticoagulants

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

160

5.3.13

Anticonvulsants . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.3.14

Antidepressants . . . . . . . . . . . . . . . . . . . . . . . . . . .

161 161

5.3.15

Antiemetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

164

5.3.16

Antifungals . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

164

5.3.17

Antihistamines

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

164

5.3.18

Antihyperlipoproteinemics . . . . . . . . . . . . . . . . . . . . .

165

5.3.19

Antihypertensives . . . . . . . . . . . . . . . . . . . . . . . . . .

166

5.3.20

Antiinflammatory drugs . . . . . . . . . . . . . . . . . . . . . . .

166

5.3.21

Antiischaemic drugs

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

169

5.3.22

Antineoplastics

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

169

5.3.23

Antiparkinsonian agents

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

169

5.3.24 5.3.25

Antipsychotic agents . . . . . . . . . . . . . . . . . . . . . . . . Antivirals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

171 171

5.3.26

Antiulcerative drugs . . . . . . . . . . . . . . . . . . . . . . . . .

171

5.3.27

Anxiolytics

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

172

5.3.28 5.3.29

Biochemical markers . . . . . . . . . . . . . . . . . . . . . . . . Calcium channel blockers . . . . . . . . . . . . . . . . . . . . . . C h o l i n e s t e r a s e inhibitors . . . . . . . . . . . . . . . . . . . . . .

172 173 175

5.3.30 5.3.31

CNS Stimulants . . . . . . . . . . . . . . . . . . . . . . . . . . .

175

5.3.32

Gastroprokinetic agents . . . . . . . . . . . . . . . . . . . . . . .

176

5.3.33

Hallucinogenics . . . . . . . . . . . . . . . . . . . . . . . . . . .

176

5.3.34

H I V p r o t e a s e inhibitors . . . . . . . . . . . . . . . . . . . . . . .

177

5.3.35

Natriuretics . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

177

5.3.36

Leukotriene antagonists . . . . . . . . . . . . . . . . . . . . . . .

177

5.3.37

Mucolytics

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

178

5.3.38

Radiosensitisers . . . . . . . . . . . . . . . . . . . . . . . . . . .

178

5.3.39

Sedative/hypnotics

178

5.3.40

Serotonin uptake inhibitors . . . . . . . . . . . . . . . . . . . . .

178

5.3.41

Thyromimetic agents . . . . . . . . . . . . . . . . . . . . . . . .

179

5.3.42

Vasodilators (cerebral)

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

179

5.3.43

Vitamins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

179

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

xi

Contents

5.4 5.5

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 6.

6.1

6.3 6.4 6.5

7.1 7.2

7.3

7.4 7.5

185 185 188 192 200 213 214

Immobilized enzyme reactors in liquid chromatography: On-line bioreactors for use in synthesis and drug discovery

Nektaria M a r k o g l o u and Irving W. Wainer . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Immobilized enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1 G e n e r a l a p p r o a c h e s . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.2 E n z y m e i m m o b i l i z a t i o n on c h r o m a t o g r a p h i c supports . . . . . . . 7.2.3. Effect of i m m o b i l i z a t i o n on e n z y m e stability . . . . . . . . . . . . 7.2.4 Effect of i m m o b i l i z a t i o n on e n z y m e kinetics . . . . . . . . . . . . 7.2.5 Effect of i m m o b i l i z a t i o n on the thermal stability of an e n z y m e . . 7.2.6 The effect of i m m o b i l i z a t i o n on the e n z y m e ' s r e s p o n s e to p H . . . On-line i m m o b i l i z e d e n z y m e reactors ( I M E R S ) . . . . . . . . . . . . . . 7.3.1 B i o c h r o m a t o g r a p h y . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2 On-line I M E R s . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2.1 On-line M i c h a e l i s - M e n t e n kinetics using an L C - I M E R format . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2.2 A p p l i c a t i o n of I M E R s to on-line enantiospecific synthesis and purification . . . . . . . . . . . . . . . . . 7.3.2.3 O n - L i n e study of c o m p l e x biological systems using coupled IMERs . . . . . . . . . . . . . . . . . . . . . . . I m m o b i l i z e d e n z y m e s and I M E R s in drug discovery . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 8.

8.1 8.2

Method development in reversed-phase chromatography

U w e Dieter Neue, Eric S. G r u m b a c h , Jeff R. M a z z e o , K i m V a n Tran and Diane M. W a g r o w s k i - D i e h l . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Tools for the m e a s u r e m e n t of selectivity differences and the quality of a separation . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2 M e a s u r i n g selectivity differences . . . . . . . . . . . . . . . . . . M e t h o d d e v e l o p m e n t strategy . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 7.

180 180

215 215 216 217 217 218 219 220 222 222 222 224 225 226 228 233 233

Use of liquid chromatography-mass spectrometry in acute human toxicology

M.J. B o g u s z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M e t h o d i c a l considerations . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.1 O p t i m i z a t i o n of c h r o m a t o g r a p h i c analysis . . . . . . . . . . . . . . 8.2.2 Use of different ionization sources. Use of single- and triple q u a d r u p o l e instruments . . . . . . . . . . . . . . . . . . . . . . . . 8.2.3 Use of various mass analyzers . . . . . . . . . . . . . . . . . . . .

235 235 236 236 240 240

Contents

xii 8.3

8.4 8.5 8.6

Applications of L C - M S in clinical toxicological analysis . . . . . . . . . 8.3.1 Illicit drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1.1 Opiate agonists . . . . . . . . . . . . . . . . . . . . . . . 8.3.1.2 A m p h e t a m i n e s . . . . . . . . . . . . . . . . . . . . . . . 8.3.1.3 C o c a i n e . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1.4 C a n n a b i n o i d s . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1.5 L S D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2 T h e r a p e u t i c drugs . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2.1 B e n z o d i a z e p i n e s . . . . . . . . . . . . . . . . . . . . . . 8.3.2.2 Antidepressants and antipsychotics . . . . . . . . . . . . 8.3.2.3 I m m u n o s u p r e s s a n t s and antineoplastic drugs . . . . . . . 8.3.2.4 Diuretics . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2.5 Cardiac g l y c o s i d e s . . . . . . . . . . . . . . . . . . . . . 8.3.2.6 M u s c l e relaxants . . . . . . . . . . . . . . . . . . . . . . 8.3.2.7 Antidiabetics . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2.8 H o r m o n e s . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2.9 Other drugs of toxicological relevance . . . . . . . . . . . 8.3.3 E n v i r o n m e n t a l poisons and natural c o m p o u n d s . . . . . . . . . . . 8.3.3.1 Pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.3.2 Plant and bacterial toxins . . . . . . . . . . . . . . . . . 8.3.3.3 Inorganic c o m p o u n d s . . . . . . . . . . . . . . . . . . . . 8.3.4 S c r e e n i n g p r o c e d u r e s for multiple c o m p o u n d s . . . . . . . . . . . 8.3.4.1 G e n e r a l screening for various groups of drugs . . . . . . 8.3.4.2 G r o u p screening for substances b e l o n g i n g to the s a m e therapeutic class . . . . . . . . . . . . . . . . . . . . . . C o n c l u s i o n s and perspectives . . . . . . . . . . . . . . . . . . . . . . . . A b b r e v i a t i o n s used in the text . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

240 242 242 246 248 249 249 251 251 253 254 255 255 257 257 260 260 261 261 261 263 263 263 265 266 267 267

Chapter 9. HPLC-MS(MS) for bioanalysis in drug discovery and development 9.1 9.2

9.3

9.4 9.5

Brian Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T h e use of generic m e t h o d s . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.1 C o l u m n selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.2 E l u e n t selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.2.1 Organic modifier . . . . . . . . . . . . . . . . . . . . . . 9.2.2.2 Buffer and p H modifier . . . . . . . . . . . . . . . . . . 9.2.2.3 Buffer c o n c e n t r a t i o n . . . . . . . . . . . . . . . . . . . . 9.2.3 Effects of eluent p H and c o m p o u n d type . . . . . . . . . . . . . . S a m p l e p o o l i n g and cocktail dosing . . . . . . . . . . . . . . . . . . . . . 9.3.1 S a m p l e pooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.2 Cocktail dosing . . . . . . . . . . . . . . . . . . . . . . . . . . . . Short c o l u m n H P L C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S a m p l e introduction onto H P L C . . . . . . . . . . . . . . . . . . . . . . .

271 271 272 273 277 277 278 278 280 281 281 283 283 287

xiii

Contents

9.6 9.7 9.8 9.9

The use of gradient elution with high flow rates . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 10.

10.1 10.2

10.3 10.4 10.5 10.6 10.7 10.8

10.9

10.10 10.11

11.1 11.2 11.3 11.4

Biomedical applications of directly-coupled chromatography-nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS)

John C. Lindon, Nigel J.C. Bailey, Jeremy K. Nicholson and Ian D. Wilson . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Technical developments in H P L C - N M R and H P L C - N M R - M S . . . . . 10.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2.2 The requirement for high dynamic range in N M R spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2.3 Avoidance of compromised chromatographic resolution . . . . 10.2.4 The need for high N M R sensitivity . . . . . . . . . . . . . . . 10.2.5 Additional considerations for double coupling of N M R and MS to HPLC . . . . . . . . . . . . . . . . . . . . . . . . . . . Operational methods in H P L C - N M R and H P L C - N M R - M S . . . . . . Applications in combinatorial chemistry . . . . . . . . . . . . . . . . . Application to drug impurities . . . . . . . . . . . . . . . . . . . . . . Chiral H P L C - N M R and H P L C - C D for pharmaceutical mixtures . . . . Application to natural products . . . . . . . . . . . . . . . . . . . . . . Application to drug metabolism . . . . . . . . . . . . . . . . . . . . . . 10.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.8.2 A summary of human metabolism studies . . . . . . . . . . . . 10.8.3 Animal metabolism studies of pharmaceuticals and model compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.8.4 Application to in vitro metabolism studies . . . . . . . . . . . 10.8.5 Application to drug metabolite reactivity . . . . . . . . . . . . Future developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.9.1 Automation and informatics . . . . . . . . . . . . . . . . . . . 10.9.2 Miniaturisation in separations coupled to N M R . . . . . . . . . 10.9.3 Hypernation . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 11.

288 291 291 291

293 293 294 294 294 295 296 296 298 300 301 302 304 306 306 307 310 316 316 321 321 322 323 325 325

Ultra-sensitive detection of radiolabelled drugs and their metabolites using accelerator mass spectrometry

Graham Lappin and Introduction . . . . . . . . . Instrumentation . . . . . . . Sample preparation . . . . . Data analysis . . . . . . . .

R. Colin . . . . . . . . . . . . . . . . . . . .

Garner . . . . . . . . . . . . . . . . . . . .

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

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

331 331 333 335 336

xiv 11.5 11.6 11.7 11.8

Contents Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L i m i t s of detection and quantification . . . . . . . . . . . . . . . . . . . C o n c l u s i o n s and the future . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 12.

12.1

12.2

12.3

12.4 12.5 12.6

13.1 13.2

13.3

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 . . . . . . . . . . . . . . . . . . . . . . . . A n i n t r o d u c t i o n to I C P - M S . . . . . . . . . . . . . . . . . . . . . . . . 12.1.1 Inductively c o u p l e d p l a s m a as an ion source . . . . . . . . . . . 12.1.2 Interfacing the I C P to a m a s s s p e c t r o m e t e r . . . . . . . . . . . . 12.1.3 T h e building blocks of an I C P - M S . . . . . . . . . . . . . . . . 12.1.4 A n a l y t i c a l capabilities of I C P - M S . . . . . . . . . . . . . . . . I C P - M S as an e l e m e n t specific detector for c h r o m a t o g r a p h i c separations 12.2.1 C o u p l i n g an H P L C to I C P - M S . . . . . . . . . . . . . . . . . . 12.2.2 C o u p l i n g G C to I C P - M S . . . . . . . . . . . . . . . . . . . . . 12.2.3 C o u p l i n g C E to I C P M S . . . . . . . . . . . . . . . . . . . . . . A p p l i c a t i o n s of I C P - M S in the b i o m e d i c a l field . . . . . . . . . . . . . . 12.3.1 D e t e c t i o n of m e t a b o l i t e s . . . . . . . . . . . . . . . . . . . . . . 12.3.2 P h o s p h o r y l a t i o n detection by I C P - M S . . . . . . . . . . . . . . 12.3.3 O t h e r applications . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 13.

337 341 345 347

351 351 351 354 354 358 359 359 360 361 362 362 364 366 368 370 371

Chromatography in a regulated environment

H.M. Hill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R e g u l a t o r y issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.1 Regulatory environment . . . . . . . . . . . . . . . . . . . . . . 13.2.2 C o m p l i a n c e with G L P ? . . . . . . . . . . . . . . . . . . . . . . 13.2.3 I n s t r u m e n t qualification and validation . . . . . . . . . . . . . . B i o a n a l y t i c a l validation process . . . . . . . . . . . . . . . . . . . . . . 13.3.1 Full validation . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.1.1 P r e - s t u d y p h a s e . . . . . . . . . . . . . . . . . . . . . 13.3.2 A p p l i c a t i o n of a validated analytical m e t h o d . . . . . . . . . . . 13.3.2.1 S y s t e m suitability . . . . . . . . . . . . . . . . . . . 13.3.2.2 D i s p o s i t i o n of standards, Q C s and samples in a batch 13.3.2.3 C h r o m a t o g r a p h i c a c c e p t a n c e . . . . . . . . . . . . . . 13.3.2.4 R e i n t e g r a t i o n of c h r o m a t o g r a p h i c peaks . . . . . . . . 13.3.2.5 Standard curve a c c e p t a n c e . . . . . . . . . . . . . . . 13.3.2.6 Quality control a c c e p t a n c e criteria . . . . . . . . . . . 13.3.2.7 S a m p l e assay repeat criteria . . . . . . . . . . . . . .

373 373 375 376 376 378 378 379 379 392 393 394 395 396 397 397 397

Contents 13.3.3

XV

P o s t v a l i d a t i o n issues . . . . . . . . . . . . . . 13.3.3.1 M e t a b o l i t e s in safety testing ( M I S T ) 13.3.3.2 Cross validation . . . . . . . . . . . 13.3.3.3 M e t h o d transfer . . . . . . . . . . . . 13.3.3.4 Partial v a l i d a t i o n . . . . . . . . . . . 13.3.3.5

13.4 13.5

Limit assays

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

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

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

Study documentation . . . . . . . . . . . . . . . . . . . . . . . . . . Statistical c o n s i d e r a t i o n s . . . . . . . . . . . . . . . . . . . . . . . . 13.5.1 R a t i o n a l e b e h i n d the c o n s e n s u s statistics . . . . . . . . . . . 13.5.2 I n t e r b a t c h a n d i n t r a b a t c h p r e c i s i o n . . . . . . . . . . . . . . 13.5.3

. . . .

. . . .

398 398 400 400 401 402 403 404 404 405

Standard curves . . . . . . . . . . . . . . . . . . . . . . . . . .

406

T h e future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

406

13.6.1

Ethical implications . . . . . . . . . . . . . . . . . . . . . . . .

406

13.6.2

Instrumentation quantification and validation . . . . . . . . . . .

407

13.7

13.6.3 Biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

407

13.8

13.7.1 Regulatory changes . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4O8 408

13.6

Subject index

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

407

413

This Page Intentionally Left Blank

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations,Vol. 4 9 2003 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

O",

E

E

50 mVI

QJ Z

15

Time (min)

20

25

Time (rain)

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).

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

Blank plasma

Plasma sample

:r

n/z:lli

(A) -

100

Z+04 1.222

100

351

8:$4 254)6 40097

4O 2O

0

0

ml z : 126

(B)

9653 i,I n94,~.

-

1oo

g*04 2.67S

n / 9 12S 9

)|1 9:15

~I$14$ z2~2o

100

9g - i S t.4t0

80 60 40 20 0

,~

m/l:131

o

..

m/-"-o.13.1

,r .

.

.

.

.

.

" ' ' )TI

(c)

I:|4

. ~04 S.4ab

2t$1S! 4S4~37

ZOO

1.119

80 60 )37 11:47 .14944

40

47780

/i

10 0

. . . .

I

. . . .

0:40

"

. . . .

I

. . . .

9:00

Time

'

. . . .

(min)

I .... 9:20

""~'''1

9:40

r

....

.... .

lO:OO

o

....

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

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

,2-:'o

Time

'

J : 4' 0

......

| lO:O0

(min)

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).

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

il,,l

~

-~Waste

Ll)w Pressure valVeFlP.Column

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

:. ---

NeeOle

Va!ve Restrtctor

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

Extraction i i li$ olvent 1

-

,~l~]

--

kC Colux.n {llest r ict ion)

1]]]]s

~ ~l Lo~_W~_

Restrictor ]

GC e

i ,

_ _

L ~

MS

2

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

waste Separat ion

Column

--

|O-Portvalye

I~etaintng Retention Precolu~n Gao EarLy Vapor EJ~iL

Resl~rictor

Flow Pressure Regulator Regulator

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

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. ( 9 John 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

TIC 5.05

7 I

Phentolamine

1

Haloperidol

K

4.22

1

Carbamazepine

I \ Temazepam

n

d 2.00 . 4 0 0 Fig. 1.7. Total ion chromatogram and MRM chromatograms of the LC-MS-MS assay for the 10 compounds with 100 )*.1 direct injection of a 250 nglml chimpanzee plasma standard. A 2.0 x 150 mm C18 column (Symmetry, Waters Corp.) was used as the analytical column, which was operated in the gradient mode with a flow rate of 0/4 mumin. (Reprinted from [ 1.51 Copyright 2000 Am. Chem. Soc.).

New developments in integrated sample preparation for bioanalysis

References pp. 39-44

Puromycin

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.

% t~

Dispenser

HPLC Pump t~ ~,,i.

Buffer tublnq

Stx-port valve

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

/

I

Analytical column

INJECT position LOAD position

\ x"k

capillary

Adjustable capillary guide/depth gauge I Adjustable needle guide/depth gauge Vial retainer arm

\

/

,o

.

"% 9%

Solvent vial

Septum piercing needle Sample vial

,, *'~176176

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.).

tO O TABLE 1.1 BIOANALYTICAL APPLICATIONS OF SPME-LC Compound

Sample

Type

System

Sorption; desorption time (min)a

Yield (%)a

LOD (ng/ml) a

Ref

Amphetamines Amphetamines, [3-blockers

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

In-tube In-tube

LC-ESUMS LC-ESI/MS

11;4 11;4

81-98 c 0.7-16

0.38-0.82 0.1-1.2

[125] [126]

Fiber (DI) In-tube In-tube

~LC-UV p~LC-UV LC-ESUMS

180; 30

N.D. N.D. 3.0-9.2

3-40 N.D. 0.024-2.00

[127] [128] [1291

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

(tx)LC-ESI/MS LC-ECD LC-ESI/MS

180; 30 45; 2 x 5 11;4

LC-ESI/MS

10; 4

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

[130] [131] [1321

In-tube Fiber Fiber Fiber Fiber Fiber Fiber

ESI/MS LC-MS LC-ESI/MS LC-UV APCI/MS/MS LC-MS/MS

60; 2 15; 5 10; 6 45; 2 x 10 5;4 6O; 10

LC-ESI/MS

11;4

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

Antidepressants (tricyclic) Antidepressants (tricyclic) Benzodiazepines Benzodiazepines Brombuterol + analogues [3-blockers [3-blockers Carnitines Corticosteroids + conjugates Flavonoids Lidocaine Lidocaine Phenothiazines Ranitidine

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

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

In-tube

10; 5 11;4

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

[1331 [134] [135] [136] [107]

[104] [137] [138]

%

Whole zoo

blood

extract

Urine

~,,.' ~ ; } 4 0

-1

9

Perazine 0

.

ZOO

Z

. . . m,'= ;36~

~ .

-

,

.

.

A

~

.

.

.

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

.

.

.

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

.

v-

. . . . .

i

,~'

Tnioridazilqe

~z t

o m/z :'f/c

100

--

,~

Prochloperazine

t

Thiethylperazine C lOG

.

.

.

.

.

.

.

.

.

.

.

.

~

.

~

-

.

0

.

m - -

,

~

~

"

s .... ..

--

, _ s %_ _

.

;. 4 ~ 0

P _

__

0

1

m/~

~

,e,O4

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

__

/k.~._

m.l Z : 4 4 7

, A_ a, r: ,, --

, . . . . . . . . .

3 : 00

E*04

100 ;

. . . . . . . . . .

6 : O0

~-.

9:00

. . . . . . . . . .

t 2 .?00

15 ;%0

Clospirazine

.....

]- m/z:44"/

---

, ....

Thioproperazine ....

......

T

11

, 9,~

~.__.

j~

J'.--

-

~'-.~::~

r '~ ~_~_ z. ~s2

~

,~-~

~.

I"~ ~

A

)

00~

'~

z ,sos

,oo1.,,,,, o

r':;'~'~,~

:,~~~~

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

,~

-4~.A 9

_

~,~. ;~

~

xA

____

~,,

Perphenazine i/ll

..

Flup~ine

PUr.:44~

. . . . . . . . .

1''''~

i

~.~

E,C,4

9

Flup~bxol

Thioproperazine 0

'

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

100

Clospirazine 0 1CO

""--- --

o '----,

m/z:438

-

. .

.

P

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

o ] Flu~ine iCO

I

o

.

ioridazme . .

- m.,'z : 374,

0

.

I00--a/=, 43S

_J

o ] Flupentixoi iCO

0 100

.

m/z-3"/l

C,

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

......

100 -

100

m/z:43~

iOO

mlz : 366

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

m/z:408

] Tdflu~

D

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

Thiethylperaz

__

;434

o

ZOO

.

~. ~80

.,~

_J ~ L _ _ r-'-

~/z

E-~04

A P"~

m/z:400

a00

Z'03

-

extract n . / = : 34C

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

~ s Ar,.o, ,... z. ~9,

,

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

/ ~

---=.,.r~

_

_

'

-

i

/_i~.

~;+0~

5.8 a

loo

_IL_I~

:100

'

"" 15

:O0

A

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).

~"

Urine extract

Whole blood extract IRIC-.-

I00

D&

o!

339.110>

r

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

Perazine RIC=> DA

0

3 4 0 . 4 ~ 0

.....

JL

.

.

.

.

.

.

,

g*O~

RIC=)

DA

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

0

RIC-,

n~

of

3"~3.60,

P,I~->

DA

of

399.;0>

40:~.'/0

I~.1C->

DA

O~

40).'.0~,

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

o

100

-

1~C-) O~

.........

. k

_

-3";&.60

of 4C?.:0>

408"/0

o~ 4 3 4 . ' 0 >

&~5.?O

' glC=)

-OA

0

~tC:>

DA

O1~

E-D1

.

.

.

.

_

.

r [1.381

j~

P""

F,*02

Z*03

.

.

]. L . 1,371

.

4]8.?0-

__

..

..

4r

4r

446.70)

441'.70

__

.

..

IRXC,:, ~

of

]Thioprope~azine 0

. . . .

,

. . . .

I : " "

3:~)0

" ' "

"

"

.

A

"

l

. . . .

(~:O0

"

/L

I ,,o~ "

.

.

.

.

'

.

.

.

.

li % o

.

9

....

,,

]PernzJ~

RZ~-~' DA

0

9

o,

100

of

RZ2-~.

] TNo~dazlne 0

100

D 100

0 :1,00

-

RZ2m~, D?,

--

nf

.

100

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

.

RICm> 1D~.

17%.4~,~

o1~ r

of

] Flupentixol

100 -

~xC-~

ga.

of

-

"

404,.70

_

t$&.'/O*

-

k

435.70

.

IDA

13'7170,,

g-02

6.9S9

It.01 t. 3 5 9

E*02

9.858

E+02

4:38.?0

_

E- 0 2 U, 0 4 "x

.

s 8. 922

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

*~176 ] Thiolxopemzi n, " -~'~;--

E - 0 3

1.722

]. 6C4

o ]Closmrszine 0

_

'." . . . . .

RI~,*

E*03

3. 7 4 5

.

DA

~*22

. . . . . .

I74,K0

--RZ"~=~'

]

.

)71.1;C

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

_ E-32 8. 5 7 2

?.'754

k

3")0.4;0-',.

Thiethylperazine

9Ir 3 .SSO

.oo

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

] ProchlOl~mzine

o

|,02

0s

J•..

)40.6C

0 ]Proper~azine

E*02 ? .8o1

I;.110

of' :339.60;"

Ioo - "~-'" ~

o

E * 0 4

.

o razJne 100

4

3.4)8

jl~ .

r

'~176

1

8..1

~t

ot

~

-- E * 0 1

L

tFlul~nli~ol o

........

.......

.......... k

]Tfifluopecazine

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

.

....

0

100

100

. 5

rid~ine _

Z-02 I2.82S

: !- ' ' ~ " 3:00

FL.m

' . . . . .I S : O0

;'-; :

k

9 . . . . .l 9:00

.

~'--"

.....

12

I " "'" " " : O0

....

Is

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).

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

New developments in integrated sample preparation for bioanalysis

References pp. 39-44 Fig. 1.11. (a) Desorption peak of a sample of cru2 Camembert cheese obtained by SPME-MS. The extraction was carried out by placing an SPME Carboxed PDMS fiber in the headspace at 20°C for 10 min. (b) Average spectrum of 0-3 min. (Reprinted from [139]. Copyright 2001 Am. Chem. SOC.).

p3

VI

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

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

C ompound

S ample

1985 1985 1986

Amino acids Barbiturates Enoximone + metabolite

Serum Serum Serum

1986

Salicylic acid

Serum

1987

Mitomycin C

1987

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

Plasma Urine Serum

1988 1988

1990 1990 1991 1991 1992

Oxolinic acid Flumequine Iopentol

1992

Pholcodine

1992 1992 1993

Benzodiazepines Rogletimide Antiviral drug* + metabolite#

Detection

Dialysis time (min)

UV Salicylate electrode UV UV

Efficiency (%)

LOD a

Ref

2 30 17 48 86-105

6 nM N.D. 10 15 N.D.

[1691 [1701 [171]

25

1

[173]

50

50-170

[174]

[172]

Serum Plasma Serum

UV UV

3 12

85-90 b 68

30 nM N.D.

[1751 [1761

Plasma Plasma

UV Flu

17 15

40 25

20 10

[1771 [178]

Whole blood Plasma Whole blood plasma

UV

7.3

60

50

[1791

UV

7.3

60-69 b

50

[180]

UV

7.15

500 c

[1811

Flu

8.25

50 47 60

40

[182]

UV MS/MS UV

7.6 14 <20

37-50 33 10-20

20-25 5 0.2 nmol/ml c

[24] [1831 [184]

Plasma Whole blood Plasma Whole blood Plasma Plasma Plasma

%

q~

TABLE 1.3 CONTINUED Year

Compound

Sample

Detection

Dialysis time (min)

Efficiency (%)

LOD a

Ref

1993 1994

Plasma Urine

UV UV

7.5 4

3 20

0.6 ~g/ml 2.5 nmol/1 c

[185] [186]

1995 1995

Phenytoin Antiviral drug* + metabolite# NSAIDs Antiepileptics

Plasma Plasma

UV UV

9.6 10

0.01-2 ~zg/ml 0.1-0.8 ~zg/ml

[187] [188]

1996

Verapamil + Norverapamil

Plasma

Flu

10

Sildenafil (Viagra) + Metabolite Levosimendan

Plasma

UV

6

1.3 1.4 1.00 c

[189]

1997

40-65 3-10 (total) 22-23 (free) 73 76 30

Serum

UV

6.6

40 b

5c

[191]

UV UV UV Flu

8.5 10 N.D. 6

50 81 97 52-54

15 17 1 p~M 10

[192] [193] [194] [195]

12.8

52-65

5-12

[196]

Plasma Plasma

UV Flu Flu UV UV

5-15 5.25 6.5

N.D. 55 65-70

N.D. 130 2

[197] [198] [199]

Plasma Plasma

UV UV

11.8 8

67-72 60

11-15 2.5

[200] [201]

1997 1997 1997 1998 1998

Clozapine Oxprenolol Lamotrigine Amphetamines

1998

Antidepressants

1999 2000 2000

Quinolones Iodixanol Albendazole + metabolites (2) Local Anaesthetics Sotalol

2001 2001

N.D.: No data available. Flu: Fluorescence detection. a: ng/ml unless stated otherwise. b: relative to aqueous samples. c: Limit of quantitation (LOQ). *: 1-( [3-D-arabinofuranosyl)-5-( 1-propynyl)-uracil #: 5-propynyluracil

Plasma Plasma Plasma Serum Plasma Serum Plasma Serum

~,~~

[190]

~,~~

ta~

32

Chapter 1

length [187]. The latter showed to be efficient even for highly protein-bound compounds. When the dialysis time is in the same order as the analysis time of a conventional LC separation (10-20 min.), serial analysis of samples can be achieved, thus preparing one sample while analysing another. However, in bioanalysis conventional LC is more and more replaced by LC-MS-MS. Consequently, the separation time is decreasing, since the MS will make up for a lower LC separation efficiency, allowing analysis times of just a few minutes. Therefore, dialysis is beginning to disappear as a sample preparation technique and is being succeeded by batch-like techniques, e.g. 96-well plate SPE [23]. Hence, fewer applications of dialysis are to be expected in the bioanalytical field.

1.3.2.3 Non-porous membrane techniques NPMTs show a more versatile use in on-line coupling to separation techniques. Besides the more common coupling to LC [152,203-207], GC [152,208-212] and CE [152,213-216] (see also Table 1.2), NPMT have also been coupled to atomic absorption spectrophotometry [217-219], electrochemical instruments [220,221 ] and flow-injection systems with UV detection [222,223]. Both PME and MESI have not been applied to the analysis of biological samples, and therefore, these techniques will not be discussed in more detail. SLME and MMLLE are already applied to bioanalysis and their applications will be discussed below. The advantage of NPMTs in comparison to PMTs is the higher degree of selectivity for any type of sample. High enrichment factors can be achieved simultaneously with high selectivity. Another advantage is the relatively small solvent consumption in comparison with other sample preparation techniques. As shown above, the possibility of on-line coupling to various analytical instruments and the ease of automation make SLME and MMLLE attractive for bioanalysis. As with PMTs, protein binding can decrease the extraction efficiency. Other critical factors are the short-term and the longterm stability of the membranes. Especially for SLME and MMLLE, the pressure differences over the membrane must be low enough to hold the organic solvent in the pores of the hydrophobic membrane and to prevent the acceptor phase from leaking into the donor phase and vice versa. Also, the chemical stability may be critical, especially if more polar membranes are used. Carry-over effects are usually overcome by appropriate washing of the membrane. Finally, the extraction process of SLME and MMLLE is slower than conventional techniques like SPE and LLE, thus limiting the throughput of samples. However, use of parallel membrane systems may be a suitable option [146,151,152,224]. In bioanalytical applications enrichment factors of 30-70 can be achieved. Although the enrichment may not be extraordinary, the main focus for SLME and MMLLE is to provide selectivity in the extraction of drugs from the complex biological matrix. An example of the selectivity that can be achieved by SLME was presented by J6nsson et al. [204]. SLME employing a porous poly(tetrafluoroethene) (PTFE) membrane was followed by ion-pair chromatography wih variable wavelenght UV detection. The membrane was soaked in the membrane liquid (10% tri-n-octyl phosphine oxide in di-nhexyl ether) for 30 min. The pH of the donor phase was adjusted to 9.5 whereas the acceptor phase was an acidic buffer at pH 2.5. As can be seen in Fig. 1.13, no interfering

33

New developments in integrated sample preparation for bioanalysis ?~OOO

O.

4OOO4

30000

i 2OOOO

2

5

4

10000

~.A_ 2

lk

/k 4

6

4

I0

|

-Ioo~

12

14

111

|l

2~0

Tim, Imin)

5OOOO

3

40000

b

3OOOO

2

20000

5

10000

of

2

4

6

II

10

12

6

14

16

18

20

-10000 j

Fig. 1.13. Chromatograms of a water solution (a) and a urine sample (b), both spiked with 3-OH-PPX (1; 1.0 ~zM), 4-OH-Ropivacaine (2; 0.80 p~M), 3-OH-Ropivacaine (3; 0.83 p~M), PPX (4; 1.0 ~M), Iso-PPX (5; 0.84 ~M) and Ropivacaine (6; 0.90 ~M). (Reprinted from [204], with permission from Elsevier Science).

References pp. 39-44

34

Chapter 1

peaks were observed from the urine matrix. This allowed simple isocratic chromatographic analysis, and an LOD of 2-18 nM was achieved for ropivacaine and its metabolites. The extraction time was similar to the chromatographic run, thus allowing sequential sampling and analysis. The total enrichment factors were 6-136 with efficiencies of 3-73%. Polar compounds were less efficiently extracted due to the apolarity of the membrane. Lindeghrd and co-authors [205] used S L M E - L C - U V (detection at 265 nm) to determine amperozide in plasma. A PTFE membrane was soaked for 15 min. in di-nhexyl ether prior to extraction. The acceptor phase contained 2.5 mM H2SO4, given a pH of 2.5. In order to obtain satisfactory enrichment the sample (containing 12.5 mM EDTA, pH 10) was extracted three times. Despite this repeated sampling the efficiency was only 13% for aqueous solutions. As amperozide is higly bound to proteins (97%), the expected recovery from plasma relative to water was 6% (as the plasma was diluted 1:1 with donor buffer). To enhance release from the protein, the pH of the sample was changed to 13. As a result, the efficiency in plasma samples was about 30% relative to aqueous solutions. The use of a displacer (ammonia, triethylenetetramine or piperazine) did not result in a noticeable increase in efficiency. A final example worth mentioning is the result of a combined SLME-IxLC-CE system for the analysis of bambuterol [216]. This basic drug was extracted from plasma using SLME and was introduced into a micro-LC column. A heart-cut was transferred to the CE, in which enantiomer separation was performed. A total enrichment factor of 40000 was observed, giving an LOD of 0.15 nM for each bambuterol enantiomer with simple UV detection. The main enrichment did not originate from the SLME procedure. The significance of the SLME is that no matrix peaks interfered with the detection. Due to this high degree of selectivity of SLME it was possible to obtain these enrichments by analyte focusing on the LC column and from double stacking in CE. The coupling of SLME to GC was investigated for the determination of amines in urine [208]. A good LOD (1 ppb) and repeatability (3.5-4%) were obtained for the analysis of more than 600 urine samples. As the final extract is an aqueous solution, SLME seems more appropriate for coupling with LC than with GC. MMLLE, which has the advantage that the extract ends up in an organic solvent, can more easily be combined with GC [150,209]. Even though MMLLE is different from SLME, similar problems were observed for the analysis of local anaesthetics in plasma [209], i.e. matrix interference decreased the efficiency of the extraction and adsorption to the membrane caused carry-over. As only one liquid-liquid extraction step is being performed, a low selectivity may be obtained with the MMLLE procedure. Thus, the use of a selective detector is important. An advantage of MMLLE is, however, that dilution by elution from a pre-column is prevented. Several reports deal with the coupling of SLME with CE for bioanalysis [152,213,214]. A new design, based on SLME, was presented for the analysis of methamphetamine [215]. Liquid-liquid-liquid microextraction (LLLME) was coupled to CE. Urine and plasma samples were adjusted to pH 13 and, subsequently, by use of a polypropylene hollow fiber the analyte was extracted through a thin phase of 1-octanol inside the pores of the fiber and finally in 25 Ixl acceptor phase (0.1 M HC1) inside the fiber. The acceptor phase was analysed by capillary zone elecrophoresis. A schematic

New developments in integrated sample preparation for bioanalysis

Injection of acceptor solution~~..t[

i].....ii

i

35

Collection of acceptor solution

Sample solution (Donor solution) Porous hollow fiber Magnetic stirrer

Fig. 1.14. Diagramof the LLLMEextraction unit (not to scale). (Reprinted from [215]. Copyright 1999 Am. Chem. Soc.).

diagram of the LLLME unit is presented in Fig. 1.14. Effective clean-up was obtained as large molecules, acidic compounds and neutral compounds were not extracted, and an extraction efficiency of 75% was achieved within 45 min. Detection limits of 5 ng/ml were obtained in both urine and plasma and no adverse matrix effects were observed.

1.3.2.4 Remarks regarding the applicability of membrane-based techniques Membrane-based techniques have three critical factors. The first problem might be the adsorption of drugs and/or proteins, which may give rise to carry-over. Secondly, analytes that are protein-bound are more difficult to analyse. Finally, the stability of the membranes is a limiting factor. However, several studies have shown that simple precautions can help to minimise or even eliminate these problems. The positive properties of membrane-based techniques, and in particular NPMTs, are good selectivity in the extraction from complex biological matrices, ease and versatility of automation and compatibility with analysing instruments. The enrichment or efficiency can be sacrificed for more rapid analysis in cases where sensitivity is not a major issue. More sensitive analysis can be obtained by allowing the extraction process to take more time. As already mentioned before, porous membrane seems to have lost interest in the bioanalytical field. NPMTs have shown potential and are now awaited to be accepted as a new complementary technique for sample preparation. The commercial availability of suitable instrumentation will be of great importance.

1.4 CONCLUDING REMARKS From the previous sections of this chapter it has become clear that the various sample pretreatment techniques each have their advantages and disadvantages for the clean-up

References pp. 39-44

36 TABLE

Chapter 1 1.4

COMPARISON

OF SAMPLE

PRETREATMENT

TECHNIQUES

SPE

SPME

Membrane (non-porous)

+ +

+ + +

+

Automation possibilities

Simplicity

+ + +

+/+ + + a

+ + +

Choice of extraction phase

+ + +

+ b

Robustness

+ + +

+

+

VersatilityqSamples

+ + +

+

+ +

Coupling with LC

+ + +

+ +

+ + +

Coupling with GC

+ + +

+ + + + + +/_c

+ + + +/_c

+ + +

_/+ + c

_/+ + c

+ + + d

+

+

Speed Recovery/Yield/Efficiency Carry-over + + + : excellent; + + : good; + : acceptable; -: poor. a: Fiber-based ( + ), in-tube ( + + + ). b: Still increasing.

c: Speed and efficiency are inversely related: fast - poor efficiency, or slow - good efficiency. d: With re-use of cartridges the risk of carry-over is increased.

and concentration of biological samples in integrated analytical systems. The various sample pretreatment techniques are compared in Table 1.4. The set-up for SPE-GC and SPE-LC is somewhat complex, as a special extraction device with several switching valves and a solvent delivery unit is required. For SPEGC, a special GC injector is also required. The latter is required to allow the injection of the entire eluate by means of LVI. Membrane-based techniques are also more complicated due to various solvent pumps and the membrane device. The set-up for fiber-based SPME is the simplest, since only a desorption chamber is required. In-tube SPME involves the use of a less simple extraction device, but it is still more easily set up and interfaced than SPE or NPMTs. Related to the simplicity of the system is also the possibility for automation. The latter is more easily obtained with SPE and in-tube SPME than with fiber-based SPME. At present, the number of extraction phases is largest for SPE, thus allowing a good choice for selectivity. However, with the advent of SPME the number of types of phases is also increasing. Despite their limited selectivities with regard to choice of stationary phase and membrane for SPME and NPMT, respectively, these techniques have shown to produce cleaner extracts than SPE, even with complex blood samples. This is probably due to the diffusion-based extraction principle in SPME and NPMTs, in which the small drug molecules usually have larger diffusion or partition coefficients towards the stationary phase than the matrix compounds. Inversely related to the selectivity of the extraction is the required selectivity of the detector. The use of MS offers the possibility to sacrifice separation efficiency when speeding up the analysis. Because of their selectivity, MS - either in the single mode or especially in the multiple mode (MS n) - is a formidable tool to allow direct analysis of

New developments in integrated sample preparation for bioanalysis

37

drugs in low quantifies in a complex biological matrix. However, it should be noted that poor separation of the analytes of interest from matrix compounds may result in either severe reduction of the signal (ion suppression) or increase of the signal (ion enhancement). Systems in which a chromatographic separation step is omitted, i.e., direct coupling of sample pretreatment to MS, offer tremendous possibilities for high throughput of biological samples, but the matrix may affect the detection and quantitation even more. Therefore, careful attention should be paid to the selectivity of extraction. The use of MS n may enhance the detection selectivity even further, yet even then the selectivity of the sample pretreatment will remain to play an essential role in bioanalysis with these systems. The robustness of SPME is still critical. The fiber as well as the coated capillary are fragile, and the coating can easily be stripped. NPMTs are not very robust as the stability of the membrane is a limiting factor. SPE is the most established technique and offers very good robustness. Furthermore, the latter technique can easily be interfaced with both LC and GC, although at present the latter is less common and more complex. SPE can be applied to a large variety of samples such as urine, blood, plasma, serum, saliva, vitreous humor, cerebrospinal fluid, tissue homogenates, etc. Drugs are usually released from proteins in SPE, thus allowing determination of the total concentration of the drug of interest. SPME was originally designed for GC analysis, but the coupling with LC proved also suitable. SPME is mainly used for the analysis of urine samples, although examples of analysis of blood have been reported as well. SPME offers the possibility to determine the free fraction of the drug in blood, whereas the analysis of the total concentration of drugs is more complicated. This is also observed with NPMTs. The SPE procedure exists of various steps, i.e. activation, conditioning, sampling, washing and elution, thus limiting the speed. The high recoveries with SPE can be advantageous for sensitive analysis. However, also matrix compounds can be extracted to a high extent, hereby limiting the selectivity of the system. In principle SPME and NPMTs are slower than SPE because of diffusion limitations. However, with both SPME and NPMT the yield or efficiency is inversely related to the speed of the extraction. High speed can be obtained at the cost of sensitivity, i.e. low yields or efficiencies. Contrary, if very sensitive analysis is required, the speed can be lowered to achieve better yields or efficiencies. Finally, carry-over is usually no problem with SPE as often a new stationary phase is used for each extraction. However, on-line SPE also allows the re-use of cartridges, thus increasing the risk of carry-over. With SPME and NPMTs the same extraction phase is used for many analyses. As a result, carry-over effects occur more often than with SPE. The applicability of SPE-GC seems promising and its routine use is awaited. The latter system will be complementary to, but will not replace SPME-GC. SPME-LC is a novel approach that can be used as a complementary technique to the well-established SPE-LC methods. Both fiber-based SPME and in-tube SPME can be routinely applied in bioanalysis. NPMTs are being considered with growing interest and these systems seem to offer complementary possibilities for sample pretreatment in bioanalysis. The coupling of LC-GC resembles SPE-GC. Although LC may give some chromatographic separation of the drugs of interest from the matrix compounds, it does not really add

References pp. 39--44

38

Chapter 1

new possibilities to sample pretreatment with regard to SPE prior to GC analysis. The long analysis times and poor LODs limit its routine application. TFE-LC is similar to SPE-LC, yet, with a high flow-rate during extraction and the use of large particles in the extraction column. The extraction is partly size-exclusion based, and therefore mainly applied to plasma samples. PMTs have some potential to separate small analyte molecules from large matrix molecules such as proteins. Yet, it seems that they have found only limited use in practice. Both TFE and PMTs are mainly based on size-exclusion extraction and may be useful for the analysis of plasma, serum and blood. Yet, protein binding can limit the applicability of these extraction systems. Integrated sample preparation has now been introduced in many bioanalytical laboratories. The experiences of routine laboratories greatly determine the acceptance and implementation of new techniques. Therefore, open communication channels between developers and users of sample clean-up techniques is of utmost importance. In addition, lack of standardisation and/or suitable automation appears to limit the acceptance of new techniques in routine practice.

1.5 ACKNOWLEDGMENTS

This research was supported by the Technology Foundation STW, applied science division of NWO and the technology programme of the Ministry of Economic Affairs.

1.6 LIST OF ABBREVIATIONS

APCI DI/SPME ECD ECD ESI FID Flu GC HS/SPME LC LLE LLLME LOD LOQ LVI MESI MIP MMLLE MRM MS

Atmospheric pressure chemical ionisation Direct-immersion solid-phase microextraction Electrochemical detection/electrochemical detector (with LC) Electron-capture detection/electron-capture detector (with GC) Electrospray ionisation Flame-ionisation detection/flame-ionisation detector Fluorescence detector/fluorescence detection Gas chromatography/gas chromatograph Head-space solid-phase microextraction Liquid chromatography/liquid chromatograph Liquid-liquid extraction Liquid-liquid-liquid microextraction Limit of detection Limit of quantitation Large-volume injection/large-volume injector Membrane extraction with a sorbent interface Molecularly imprinted polymer Microporous membrane liquid-liquid extraction Multiple reaction monitoring Mass spectrometry/mass spectrometer

New developments in integrated sample preparation for bioanalysis MS n NPLC NPMT PDMS PME PMT PTFE PTV RPLC RSD SIM SLME SPE SPME SRM SVE TFC UV

39

M u l t i p l e - s t a g e mass s p e c t r o m e t r y N o r m a l p h a s e liquid c h r o m a t o g r a p h y Non-porous membrane technique Polydimethylsiloxane P o l y m e r i c m e m b r a n e extraction Porous membrane technique Poly(tetrafluoroethene) P r o g r a m m e d t e m p e r a t u r e vaporiser R e v e r s e d p h a s e liquid c h r o m a t o g r a p h y Relative standard deviation S e l e c t e d ion m o n i t o r i n g S u p p o r t e d liquid m e m b r a n e extraction S o l i d - p h a s e extraction Solid-phase microextraction S e l e c t e d reaction m o n i t o r i n g S o l v e n t v a p o u r exit Turbulent-flow c h r o m a t o g r a p h y Ultraviolet

1.7 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

S.B. Hawthorne, Anal. Chem., 62 (1990) 633A. J.R. Veraart, H. Lingeman and U.A.Th. Brinkman, J. Chromatogr. A, 856 (1999) 483. M. Dressier, J. Chromatogr. 165 (1979) 167. R.E. Majors, H.G. Barth and C.H. Lochmtiller, Anal Chem. 56 (1984) 300R. Z.E. Penton, Advances in Chromatogr., 37 (1997) 205. E.R. Brouwer, H. Lingeman and U.A.Th. Brinkman, Chromatographia, 29 (1990) 415. E.H.R. van der Wal, E.R. Brouwer, H. Lingeman and U.A.Th. Brinkman, Chromatographia, 39 (1994) 239. D.A. McLoughlin, T.V. Olah and J.D. Gilbert, J. Pharm. Biomed. Anal., 15 (1997) 1893. M. Jemal, D. Teitz, Z. Ouyang and S. Khan, J. Chromatogr. B, 732 (1999) 501. E.C. Goosens, D. de Jong, G.J. de Jong and U.A.Th. Brinkman, Chromatographia, 47 (1998) 313. A.J.H. Louter, J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr. A, 842 (1999) 391. J.J. Vreuls, A.J.H. Louter and U.A.Th. Brinkman, J. Chromatogr. A, 856 (1999) 279. K. Grob, J. Chromatogr. A, 892 (2000) 407. M. Jemal, Y.-Q. Xia and D.B. Whigan, 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. C. Chassaing, J. Luckwell, E Macrae, K. Saunders, E Wright and R. Venn, Chromatographia, 53 (2001) 122. M. Jemal, Biomed. Chromatogr., 14 (2000) 422. N.H. Snow, J. Chromatogr. A, 885 (2000) 455. G. Theodoridis, E.H.M. Koster and G.J. de Jong, J. Chromatogr. B, 745 (2000) 49. S. Ulrich, J. Chromatogr. A, 902 (2000) 167. G.A. Mills and V. Walker, J. Chromatogr. A, 902 (2000) 267. H.L. Lord and J.B. Pawliszyn, J. Chromatogr. A, 885 (2000) 153. N.C. van de Merbel, J. Chromatogr. A, 856 (1999) 55. N.C. van de Merbel, J.M. Teule, H. Lingeman and U.A.Th. Brinkman, J. Pharm. Biomed. Anal., 10 (1992) 225.

40 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 58 59

Chapter 1 N.C. van de Merbel, J.J. Hageman and U.A.Th. Brinkman, J. Chromatogr., 634 (1993) 1. J.E Allanson, R.A. Biddlecombe, A.E. Jones and S. Pleasance, Rapid Commun. Mass Spectrom., 10 (1996) 811. W.Z. Shou, M. Pelzer, T. Addison, X. Jiang and W. Naidong, J. Pharm. Biomed. Anal., 27 (2002) 143. R.W. Frei and K. Zech (Eds.), Selective sample handling and detection in high-performance liquid chromatography (part A), Elsevier Science Publishers, Amsterdam, The Netherlands, 1988. E.R. Brouwer, D.J. van Iperen, I. Liska, H. Lingeman and U.A.Th. Brinkman, Intern. J. Environ. Anal. Chem., 47 (1992) 257. C.H.E Bruins, C.M. Jeronimus-Stratingh, K. Ensing, W.D. van Dongen and G.J. de Jong, J. Chromatogr. A, 863 (1999) 115. G.E Bowers, S.C. Clegg, S.C. Hughes, A.J. Harker and S. Lambert, LC. GC, 15 (1997) 48. J.A. Ooms, G.S.J. Haak and O. Halmingh, Intern. Lab., 27 (1997) 18A. M.W.J. van Hout, C.M. Hofland, H.A.G. Niederl~inder, R.A. de Zeeuw and G.J. de Jong, Rapid Commun. Mass Spectrom., 14 (2000) 2103. A. Schellen, J.A. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de Lagemaat and E. Verheij, Rapid Commun. Mass Spectrom., 14 (2000) 230. A.C. Hogenboom, E Speksnijder, R.J. Vreeken, W.M.A. Niessen and U.A.Th. Brinkman, J. Chromatogr. A, 777 (1997) 81. A.C. Hogenboom, W.M.A. Niessen and U.A.Th. Brinkman, J. Chromatogr. A, 794 (1998) 201. J. Ding and U.D. Neue, Rapid Commun. Mass Spectrom., 13 (1999) 2151. W.A. Minnaard, A.C. Hogenboom, U.K. Malmqvist, E Manini, W.M.A. Niessen and U.A.Th. Brinkman, Rapid Commun. Mass Spectrom., 10 (1996) 1569. B.J. Hall, B. Goolsby and J.S. Brodbelt, Appl. Spectroscopy, 53 (1999) 1361. K. Matuszewski, M.L. Constanzer and C.M. Chavez-Eng, Anal. Chem., 70 (1998) 882. I. Fu, E.J. Woolf and B.K. Matuszewski, J. Pharm. Biomed. Anal., 18 (1998) 347. R. Bonfiglio, R.C. King, T.V. Olah and K. Merkle, Rapid Commun. Mass Spectrom., 19 (1999) 1175. D.L. Buhrman, EI. Price and EJ. Rudewicz, J. Am. Soc. Mass Spectrom., 7 (1996) 1099. A. Namera, M. Yashiki, K. Okada, Y. Iwasaki, M. Ohtani and T. Kojima, J. Chromatogr. B, 706 (1998) 253. A. Namera, M. Yashiki, Y. Iwasaki, M. Ohtani and T. Kojima, J. Chromatogr. B, 716 (1998) 171. A.J.H. Louter, E. Bosma, J.C.A. Schipperen, J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr. B, 689 (1997) 35. A.J.H. Louter, R.A.C.A. van der Wagt and U.A.Th. Brinkman, Chromatographia, 40 (1995) 500. H.G.J. Mol, H.-G. Janssen, C.A. Cramers, J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr. A, 703 (1995) 277. K. Grob and J.-M. Stoll, J. High Resolut. Chromatogr., 9 (1986) 518. K. Grob Jr., G. Karrer and M.-L. Riekkola, J. Chromatogr., 334 (1985) 129. W. Vogt, K. Jacob and H.W. Obwexer, J. Chromatogr., 174 (1979) 437. W. Vogt, K. Jacob, A.-B. Ohnesorge and H.W. Obwexer, J. Chromatogr., 186 (1979) 179. A. Farjam, J.J. Vreuls, W.J.G.M. Cuppen, U.A.Th. Brinkman and G.J. de Jong, Anal. Chem., 63 (1991) 2481. R. Herrfiez-Hernfindez, A.J.H. Louter, N.C. van de Merbel and U.A.Th. Brinkman, J. Pharm. Biomed. Anal., 14 (1996) 1077. M.W.J. van Hout, R.A. de Zeeuw, J.E Franke and G.J. de Jong, J. Chromatogr. B, 729 (1999) 199. M.EM. van Lieshout, H.-G. Janssen, C.A. Cramers and G.A. van den Bos, J. Chromatogr. A, 764 1997) 73. A.J.H. Louter, J. van Door nmalen, J.J. Vreuls and U.A.Th. Brinkman, J. High Resolut. Chromatogr., 19 (1996) 679. H.G.J. Mol, H.-G. Janssen, C.A. Cramers and U.A.Th. Brinkman, J. High Resolut. Chromatogr.,16 (1993) 459. J.J. Vreuls, U.A.Th. Brinkman, G.J. de Jong, K. Grob and A. Artho, J. High Resolut. Chromatogr., 14 (1991) 455.

New developments in integrated sample preparation for bioanalysis 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 93 94 95 96 97 98 99

41

J.J. Vreuls, R.T. Ghijsen, G.J. de Jong and U.A.Th. Brinkman, J. Microcol. Sep., 5 (1993) 317. M. Abdel-Rehim, K.A. Svensson, Y. Askemark and K.-J. Pettersson, J. Chromatogr. B, 755 (2001) 253. M.W.J.van Hout, R.A. de Zeeuw and G.J. de Jong, J. Chromatogr. A, 858 (1999) 117. M.W.J.van Hout, W.M.A. van Egmond, J.E Franke, R.A. de Zeeuw and G.J. de Jong, J. Chromatogr. B, 766 (2001) 37. R. Sasano, T. Hamada, M. Kurano and M. Furuno, J. Chromatogr. A, 896 (2000) 41. T. Hankemeier and U.A.Th. Brinkman, Chromatogr. Sci., 86 (2001) 155. K. Grob, J. Chromatogr. A, 703 (1995) 265. T. Hy6tyl~iinen and M.-L. Riekkola, J. Chromatogr. A, 819 (1998) 13. E Munari, EA. Colombo, E Magni, G. Zilioli, S. Trestianu and K. Grob, J. Microcol. Sep., 7 (1995) 403. L. Mondello, E Dugo, G. Dugo, A.C. Lewis and K.D. Bartle, J. Chromatogr. A, 842 (1999) 373. I.A. Mostert and K.A. Ramsteiner, J. Chromatogr., 477 (1989) 359. V. Giansello, E. Brenn, G. Figini and A. Gazzaniga, J. Chromatogr., 473 (1989) 343. D. Duquet, C. Dewaele, M. Verzele and S. McKinley, J. High Resolut. Chromatogr., Chromatogr. Commun., 11 (1988) 924. E.C. Goosens, D. de Jong, J.H.M. van den Berg, G.J. de Jong and U.A.Th. Brinkman, J. Chromatogr., 552 (1991) 489. E.C. Goosens, I.M. Beerthuizen, D. de Jong, G.J. de Jong and U.A.Th. Brinkman, Chromatographia, 40 (1995) 267. E.C. Goosens, K.H. Stegman, D. de Jong, G.J. de Jong and U.A.Th. Brinkman, Analyst, 121 (1996) 61. E.C. Goosens, D. de Jong, G.J. de Jong and U.A.Th. Brinkman, J. Microcol. Sep., 6 (1994) 207. E Wessels, J. Ogorka, G. Schwinger and M. Ulmer, J. High. Resolut. Chromatogr., 16 (1993) 708. J. Ogorka, G. Schwinger, G. Braut and V. Seidel, J. Chromatogr., 626 (1992) 87. T. Hy6tyl~iinen, T. Andersson and M.-L. Riekkola, J. Chromatogr. Sci., 35 (1997) 280. T. Hy6tyl~iinen, R. Pilvio and M.-L. Riekkola, J. High Resolut. Chromatogr., 19 (1996) 439. T. Hy6tyl~iinen, H. Keski-Hynnil~i and M.-L. Riekkola, J. Chromatogr. A, 771 (1997) 360. H.M. Quinn and J.J. Takarewski, International Patent Number WO97/16724 (1997). D. Zimmer, V. Pickard, W. Czembor and C. Mtiller, J. Chromatogr. A, 854 (1999) 23. N. Brignol, R. Bakhtiar, L. Dou, T. Majumbar and EL.S. Tse, Rapid Commun. Mass Spectrom., 14 (2000) 141. J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallett and R.S. Plumb, Rapid Commun. Mass Spectrom., 11 (1997) 1953. J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallett and R.S. Plumb, J. Chromatogr. A, 828 (1998) 199. M. Jemal, Z. Ouyang, Y.-Q. Xia and M.L. Powell, Rapid Commun. Mass Spectrom., 13 (1999) 1462. Y.-Q. Xia, D.B. Whigan, M.L. Powell and M. Jemal, Rapid Commun. Mass Spectrom., 14 (2000) 105. Y. Hsieh, M.S. Bryant, G. Gruela, J.-M. Brisson and W.A. Korfmacher, Rapid Commun. Mass Spectrom., 14 (2000) 1384. J. Ayrton, R.A. Clare, G.J. Dear, D.N. Mallett and R.S. Plumb, Rapid. Commun. Mass Spectrom., 13 (1999) 1657. M. Jemal, M. Huang, X. Jiang, Y. Mao and M.L. Powell, Rapid Commun. Mass Spectrom., 13 (1999) 2125. M. Jemal, Z. Ouyang and M.L. Powell, J. Pharm. Biomed. Anal., 23 (2000) 323. D. Zimmer, V. Pickard, W. Czembor and C. Mtiller, Chromatographia, 52 (2000) $26. H.K. Lim, K.W. Chan, S. Sisenwine and J.A. Scatina, Anal. Chem., 73 (2001) 2140. R.T. Cass, J.S. Villa, D.E. Karr and D.E. Schmidt, Rapid Commun. Mass Spectrom., 15 (2001) 406. C.L. Arthur and J.B. Pawliszyn, Anal. Chem., 62 (1990) 2145. Z. Zhang, M.J. Yang and J.B. Pawliszyn, Anal. Chem., 66 (1994) 844A. D. Louch, S. Motlagh and J.B. Pawliszyn, Anal. Chem., 64 (1992) 1187. H.L. Lord and I.B. Pawliszyn, J. Chromatogr. A, 902 (2000) 17.

42 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146

Chapter 1 H.A.G. Niederl~inder, M.W.J. van Hout and V. Jas, in preparation. W.H.J. Vaes, C. Hamwijk, E.U. Ramos, H.J.M. Verhaar and J.L.M. Hermens, Anal. Chem., 68 (1996) 4458. R. Eisert and J. Pawliszyn, Crit. Rev. Anal. Chem., 27 (1997) 103. T. G6recki, X. Yu and J.B. Pawliszyn, Analyst, 124 (1999) 643. M.W.J. van Hout, V. Jas, H.A.G. Niederl~inder, R.A. de Zeeuw and G.J. de Jong, Analyst, 127 (2002) 355. Y. Yang, S.B. Hawthorne, D.J. Miller, Y. Liu and M.L. Lee, Anal. Chem., 70 (1998) 1866. J.B. Pawliszyn, Solid Phase Microextraction - Theory and Practice, Wiley, New York, 1997. E.H.M. Koster, N.S.K. Hofman and G.J. de Jong, Chromatographia, 47 (1998) 678. J. Beltran, EJ. L6pez and E Hern~indez, J. Chromatogr. A, 885 (2000) 389. T. Kumazawa and O. Suzuki, J. Chromatogr. B, 747 (2000) 241. H. Kataoka, H.L. Lord and J.B. Pawliszyn, J. Chromatogr. A, 880 (2000) 35. A. Namera, M. Yashiki, N. Nagasawa, Y. Iwasaki and T. Kojima, Forensic Sci. Int., 88 (1997) 125. K.G. Furton, J.R. Armirall, M. Bi, J. Wang and L. Wu, J. Chromatogr. A, 885 (2000) 419. P. Okeyo, S.M. Rentz and N.H. Snow, J. High Resolut. Chromatogr., 20 (1997) 171. W.H.J. Vaes, E.U. Ramos, H.J.M. Verhaar, W. Seinen and J.L.M. Hermens, Anal. Chem., 68 (1996) 4463. W.H.J. Vaes, E.U. Ramos, C. Hamwijk, I. Van Holsteijn, B.J. Blaauboer, W. Semen, H.J.M. Verhaar and J.L.M. Hermens, Chem. Res. Toxicol. 10 (1997) 1067. E.H.M. Koster, C. Wemes, J.B. Morsink and G.J. de Jong, J. Chromatogr. B, 739 (2000) 175. D. Figeys, A. Ductre, J.R. Yates and R. Aebersold, Nat. Biotechnol., 14 (1996) 1579. C.-W. Whang and J.B. Pawliszyn, Anal. Commun., 35 (1998) 353. S. Li and S. Weber, Anal. Chem., 69 (1997) 1217. K. Jinno, Y. Han, H. Sawada and M. Taniguchi, Chromatographia, 46 (1997) 309. K. Jinno, H. Sawada and Y. Han, Biomed. Chromatogr., 12 (1998) 126. J. Chen and J.B. Pawliszyn, Anal. Chem., 67 (1995) 2530. A.A. Boyd-Boland and J.B. Pawliszyn, Anal. Chem., 68 (1996) 1521. R. Eisert and J. Pawliszyn, Anal. Chem. 69 (1997) 3140. H. Kataoka, H.L. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 257. H. Kataoka, H.L. Lord, S. Yamamoto, S. Narimatsu and J. Pawliszyn, J. Microcol. Sep., 12 (2000) 493. K. Jinno, M. Kawazoe and M. Hayashida, Chromatographia, 52 (2000) 309. Y. Saito, M. Kawazoe, M. Hayashida and K. Jinno, Analyst, 125 (2000) 807. H. Yuan, Z. Mester, H. Lord and J. Pawliszyn, J. Anal. Toxicol. 24 (2000) 718. K. Jinno, M. Taniguchi and M. Hayashida, J. Pharm. Biomed. Anal., 17 (1998) 1081. E.H.M. Koster, C. Crescenzi, W. den Hoedt, K. Ensing and G.J. de Jong, Anal. Chem., 73 (2001) 3140. J. Wu, H.L. Lord, J. Pawliszyn and H. Kataoka, J. Microcol. Sep., 12 (2000) 255. H. Kataoka, S. Narimatsu, H.L. Lord and J. Pawliszyn, Anal. Chem., 71 (1999) 4237. M. M6der, H. L6ster, R. Herzschuh and P. Popp, J. Mass Spectrom., 32 (1997) 1195. D.A. Volmer and J.P.M. Hui, Rapid Commun. Mass Spectrom., 11 (1997) 1926. M. Satterfield, D.M. Black and J.S. Brodbelt, J. Chromatogr. B, 759 (2001) 33. T. Kumazawa, H. Seno, K. Watanabe-Suzuki, H. Hattori, A. Ishii, K. Sato and O. Suzuki, J. Mass Spectrom., 35 (2000) 1091. H. Kataoka, H.L. Lord and J. Pawliszyn, J. Chromatogr. B, 731 (1999) 353. C. P6rbs, C. Vaillon and J.-L. Berdagu6, Anal. Chem., 73 (2001) 1030. G.E. Orzechowska, E.J. Poziomek and V. Tersol, Anal. Letters, 30 (1997) 1437. J. Ai, Anal. Chem., 69 (1997) 1230. Y. Liu, M.L. Lee, K.J. Hageman, Y. Yang and S.B. Hawthorne, Anal. Chem., 69 (1997) 5001. Y. Liu, Y.E Shen and M.L. Lee, Anal. Chem., 69 (1997) 190. E Mangani and R. Cenciarini, Chromatographia, 41 (1995)678. E. Baltussen, E Sandra, E David and C. Cramers, J. Microcol. Sep., 11 (1999) 737. J.A. J6nsson and L. Mathiasson, J. Chromatogr. A, 902 (2000) 205.

New developments in integrated sample preparation for bioanalysis 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

43

B.N. Cordero, J.L.E Pav6n, C.G. Pinto, M.E.E Laespada, R.C. Martfnez and E.R. Gonzalo, J. Chromatogr. A, 902 (2000) 195. J.A. J6nsson and L. Mathiasson, J. Sep. Sci., 24 (2001) 495. J.,~,. J6nsson and L. Mathiasson, Chromatographia, 52 (2000) S-8. J./~. J6nsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 318. J./k. J6nsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 325. J.~,. J6nsson and L. Mathiasson, Adv. Chromatogr., 41 (2001) 53. M. Gilar, E.S.E Bouvier and B.J. Compton, J. Chromatogr. A, 909 (2001) 111. G. Matz, G. Kibelka, J. Dahl and E Lenneman, J. Chromatogr. A, 830 (1999) 365. B. Hauser and E Popp, J. High Resolut. Chromatogr., 22 (1999) 205. S. Mitra, N. Zhu, X. Zhang and B. Kebbekus, J. Chromatogr. A, 736 (1996) 165. A. Chen and C.E. Lunte, J. Chromatogr. A, 691 (1995) 29. K.M. Steele and C.E. Lunte, J. Pharm. Biomed. Anal., 13 (1995) 149. B.H.C. Westerink, Trends Anal. Chem., 11 (1992) 176. C.E. Lunte, D.O. Scott and E Kissinger, Anal. Chem., 63 (1991) 773A. E.H. Kems, K.J. Volk, S.E. Klohr and M.S. Lee, J. Pharm. Biomed. Anal., 20 (1999) 115. T.-H. Tsai, Y.-F. Chen, I.-E Chen and C.-E Chen, J. Chromatogr. B, 729 (1999) 119. B.A.P. Buscher, U.R. Tjaden and J. van der Greef, J. Chromatogr. A, 764 (1997) 135. A.J.J. Debets, W.Th. Kok, K.-P. Hupe and U.A.Th. Brinkman, Chromatographia, 30 (1990) 361. B.A.P. Buscher, A.J.E Hofte, U.R. Tjaden and J. van der Greef, J. Chromatogr. A, 777 (1997) 51. P. Gonz~ilez, C.A. Fente, C. Franco, B. V~izquez, E. Quinto and A. Cepeda, J. Chromatogr. B, 693 (1997) 321. J.R. Veraart, J. van Hekezen, M.C.E. Groot, C. Gooijer, H. Lingeman, N.H. Velthorst and U.A.Th. Brinkman, Electrophoresis, 19 (1998) 2944. J.R. Veraart, M.C.E. Groot, C. Gooijer, H. Lingeman, N.H. Velthorst and U.A.Th. Brinkman, Analyst, 124 (1999) 115. D.C. Tumell and J.D.H. Cooper, J. Autom. Chem., 7 (1985) 177. D.C. Turnell and J.D.H. Cooper, J. Autom. Chem., 7 (1985) 181. J.D.H. Cooper and D.C. Tumell, J. Chromatogr., 380 (1986) 109. Q. Chang and M.E. Meyerhoff, Anal. Chim. Acta, 186 (1986) 81. U.R. Tjaden, E.A. de Bruijn, R.A.M. van der Hoeven, C. Jol, J. van der Greef and H. Lingeman, J. Chromatogr., 420 (1987) 53. D.C. Turnell and J.D.H. Cooper, J. Chromatogr., 395 (1987) 613. D.C. Turnell, J.D.H. Cooper, B. Green, G. Hughes and D.J. Wright, Clin. Chem., 34 (1988) 1816. J.D.H. Cooper, D.C. Tumell, B. Green and E Verillon, J. Chromatogr., 456 (1988) 53. H. Irth, G.J. de Jong, H. Lingeman and U.A.Th. Brinkman, Anal. Chim. Acta, 236 (1990) 165. D.S. Stegehuis, U.R. Tjaden and J. van der Greef, J. Chromatogr., 511 (1990) 137. T. AgasCster and K.E. Rasmussen, J. Chromatogr., 570 (1991) 99. T. AgasCster and K.E. Rasmussen, J. Chromatogr., 564 (1991) 171 A.T. Andresen, EB. Jacobsen and K.E. Rasmussen, J. Chromatogr., 575 (1992) 93. A.T. Andresen, M. Krogh and K.E. Rasmussen, J. Chromatogr., 582 (1992) 123. E. van Bakergem, R.A.M. van der Hoeven, W.M.A. Niessen, U.R. Tjaden, J. van der Greef, G.K. Poon and R. McCague, J. Chromatogr., 598 (1992) 189. A.R. Buick and C.T.C. Fook Sheung, J. Chromatogr., 617 (1993) 65. A.T. Andresen, K.E. Rasmussen and H.E. Rugstad, J. Chromatogr., 621 (1993) 189. J.D.H. Cooper, C.T.C. Fook Sheung and A.R. Buick, J. Chromatogr. B, 652 (1994) 15. R. Hemiez-Hem~indez, N.C. van de merbel and U.A.Th. Brinkman, J. Chromatogr. B, 666 (1995) 127. K. Johansen, M. Krogh, A.T. Andresen, A.S. Christophersen, G. Lehne and K.E. Rasmussen, J. Chromatogr. B, 669 (1995) 281. A. Ceccato, E Chiap, Ph. Hubert, B. Toussaint and J. Crommen, J. Chromatogr. A, 750 (1996) 351. J.D.H. Cooper, D.C. Muirhead, J.E. Taylor and P.R. Baker, J. Chromatogr. B, 701 (1997) 87. M. Karlsson, T. Korkolainen and T. Wikberg, Biomed. Chromatogr., 11 (1997) 54. K. Johansen, M. Krogh and K.E. Rasmussen, J. Chromatogr. B, 690 (1997) 223.

44 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224

Chapter 1 A. Ceccato, B. Toussaint, R Chiap, Ph. Hubert and J. Crommen, J. Pharm. Biomed. Anal., 15 (1997) 1365. G. Higgins, A. Highes, J. Dutton and N.B. Roberts, Ann. Clin. Biochem., 35 (1998) 534. E Sadeghipour and J.-L. Veuthey, J. Pharm. Biomed. Anal., 17 (1998) 801. K. Johansen and K.E. Rasmussen, J. Pharm. Biomed. Anal., 16 (1998) 1159. T. Zupan~i~ and B. Pihlar, J. Chromatogr. A, 840 (1999) 11. P.B. Jacobsen, J. Chromatogr. B, 749 (2000) 135. E Chiap, B. Evrard, M.A. Bimazubute, E de Tullio, Ph. Hubert, L. Delattre and J. Crommen, J. Chromatogr. A, 870 (2000) 121. E Chiap, B. Boulanger, L. Fotsing, Ph. Hubert and J. Crommen, Chromatographia, 53 (2001) 678. E Chiap, A. Ceccato, B. Miralles Buraglia, B. Boulanger, Ph. Hubert and J. Crommen, J. Pharm. Biomed. Anal., 24 (2001) 801. D.C. Turnell and J.D.H. Cooper, J. Autom. Chem., 8 (1986) 151. J. Trocewicz, Z. Suprynowicz and J. Markowicz, J. Chromatogr. B, 685 (1996) 129. J.,~. J6nsson, M. Andersson, C. Melander, J. Norberg, E. Thordarson and L. Mathiasson, J. Chromatogr. A, 870 (2000) 151. B. Lindeg~d, H. Bj6rk, J.,~. J6nsson, L. Mathiasson and A.-M. Olsson, Anal. Chem., 66 (1994) 4490. E. Thordarson, S. Pfilmarsd6ttir, J./k. J6nsson and L. Mathiasson, Anal. Chem., 68 (1996) 2559. J. Norberg, J. Emndus, J./~. J6nsson, L. Mathiasson, E. Burestedt, M. Knutsson and G. Marko-Varga, J. Chromatogr. B, 701 (1997) 39. G. Audunsson, Anal. Chem., 60 (1988) 1340. Y. Shen, J./k. J6nsson and L. Mathiasson, Anal. Chem., 70 (1998) 946. B. Lindegfrd, J.A. J6nsson and L. Mathiasson, J. Chromatogr., 573 (1992) 191. J./k. J6nsson, L. Mathiasson, B. Lindeg~d, J. Trocewicz and A.-M. Olsson, J. Chromatogr. A, 665 (1994) 259. Y. Shen, L. Mathiasson and J.,~. J6nsson, J. Microcol. Sep., 10 (1998) 107. S. Pfilmarsd6ttir, E. Thordarson, L.-E. Edholm, J.A. J6nsson and L. Mathiasson, Anal. Chem., 69 (1997) 1732. S. Pfilmarsd6ttir, L. Mathiasson, J.,~. J6nsson and L.-E. Edholm, J. Chromatogr. B, 688 (1997) 127. S. Pedersen-Bjergaard and K.E. Rasmussen, Anal. Chem., 71 (1999) 2650. S. Pfilmarsd6ttir, L. Mathiasson, J./k. J6nsson and L.-E. Edholm, J. Cap. Electrophoresis, 3 (1996) 255. M. Papantoni, N.-K. Djane, K. Ndung'u, J.*. J6nsson and L. Mathiasson, Analyst, 120 (1995) 1471 E Malcus, N.-K. Djane, L. Mathiasson and G. Johansson, Anal. Chim. Acta, 327 (1996) 295. N.-K. Djane, I.A. Bergdahl, K. Ndung'u, A. Schtitz, G. Johansson and L. Mathiasson, Analyst, 122 (1997) 1073. N.-K. Djane, S. Armalis, K. Ndung'u, G. Johansson and L. Mathiasson, Analyst, 123 (1998) 393. S. Armalis, I. Kriksciuniene, E. Kubilene, N.-K. Djane, K. Ndung'u and L. Mathiasson, Intern. J. Enviro nm. Anal. Chem., 74 (1999) 233. E. Luque-Perdz, A. Rfos, M. Valcfircel, L.-G. Danielsson and F. Ingman, Lab. Autom. Inform. Managem., 34 (1999) 131. R.G. Melcher, Anal. Chim. Acta, 214 (1988) 299. J.A. J6nsson and L. Mathiasson, Trends Anal. Chem., 11 (1992) 106.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 Elsevier Science B.V.All rights reserved

45

CHAPTER 2

Solid-phase extraction on molecularly imprinted polymers Lars I. Andersson ~ and Leif Schweitz 2

2

Research DMPK, AstraZeneca R&D S6dertiilje, S-151 85 SOdertiilje, Sweden Analytical Development, AstraZeneca R&D MOlndal, S-431 83 MOlndal, Sweden

2.1 INTRODUCTION Despite the recent progress in mass spectrometry, providing both high separation power and impressive detection sensitivity, selective sample preparation and efficient chromatographic separation will remain indispensable to a robust analytical method. The extent of sample pre-treatment and separation required depends on the complexity of the analytical problem where, in particular, trace and ultra-trace analysis of complex biological and environmental matrices relies on efficient sample enrichment prior to a selective assay. Also, for high-concentration level analysis a more selective sample clean-up simplifies the total analytical protocol. Solid-phase extraction (SPE) is continuously growing in importance [1-5], and various formats of SPE are currently routine sample preparation techniques employed in numerous environmental and bioanalytical applications. Many current SPE sorbents retain, however, not only the target analytes but also other matrix components, and often a considerable amount of method development work is spent on optimising the extraction protocol. Whereas for most materials the separation is based on physicochemical retention on the functionalised surface, more selective SPE materials, such as immunosorbents [6-7] and molecularly imprinted polymers (MIPs) [8-11 ], rely on affinity interactions. Characteristic for both affinity sorbents are their high ligand selectivity and affinity, which potentially yield a higher degree of sample cleanup. Selectivity can be pre-determined for a particular analyte or structural class of compounds by, respectively the choice of antigen used for antibody generation and the choice of template used for polymer preparation. Recent years have seen an increasing research activity into molecular-imprint based solid-phase extraction (MISPE). In parallel with genetic research into methodology development an increasing number of investigations into the use of MIPs for analysis of drugs and pollutants in biological and environmental samples are being reported. This References pp. 69-71

46

Chapter 2

review discusses the potential scope and some pitfalls of using MIPs for solid-phase extraction as well as some fundamental aspects of the molecular imprinting technology for such applications.

2.2 IMPRINT PREPARATION

Imprinting of molecules occurs by the polymerisation of functional and cross-linking monomers in the presence of a template (Fig. 2.1) [8-11]. The monomer system is chosen such that pre-polymerisation complexes of one or several functional monomers per template molecule are formed. During the polymerisation reaction these complexes become spatially fixed in the highly cross-linked polymer network and subsequent removal of the template molecules leaves behind imprints, or "memory sites", in the solid polymer. These imprints possess a topological (size and shape) and chemical (spatial arrangement of complementary functionality) memory for the template, and enable the polymer to selectively rebind the imprint species from a mixture. Two principally different approaches to molecular imprinting may be distinguished: The noncovalent, or self-assembly, protocol [12] where complex formation is the result of non-covalent or metal ion coordination interactions, and the covalent, or pre-organised, protocol [13], which employs reversible covalent bonds, usually involving a prior chemical synthesis step to link the monomers to the template. While the non-covalent strategy is the one being by far the most widely employed for applications in the analytical field, both approaches have their strengths and weaknesses. Whereas it is generally perceived that covalent imprinting yields more defined and more homogenous binding sites, non-covalent imprinting is more flexible in terms of the range of chemical functionalities that can be targeted and thus the range of templates that can be used. The latter is also much easier practically, since complex formation occurs on mixing template and monomers in solution prior to polymerisation. Recent developments have seen the advent of the covalent sacrificial spacer imprinting with non-covalent recognition (semi-covalent) protocol [14,15] and the stoichiometric non-covalent imprinting protocol [16,17]. Both techniques combine the advantages of well-defined imprint formation characteristic of covalent imprinting with rapid association/ dissociation kinetics of non-covalent interaction-based recognition. Although, at present, they suffer from the same limitation as the covalent protocol of "individualised" monomer design with a consequent lack of general applicability, both protocols show great promise to improve on current imprinting efficiency. In analytical chemistry, MIPs are increasingly being employed in application areas as diverse as liquid chromatography [18-20], capillary electrochromatography and capillary electrophoresis [21-23], pseudo-immunoassay [24,25], chemical sensing [26-29] and solid-phase extraction [25,30-35], where each application area has its own specific demands on the properties and formats of the MIP preparations used. The following discussion will, however, concentrate on issues critical for MIP preparation for SPE applications, such as removal of template molecules, choice of template, and format of the polymer. A discussion on imprint preparation generally is beyond the

%

~,,,i ~

,,,,.i

t%

t,,~o

~,~~ t%

Fig. 2.1. Schematic depiction of a molecular imprinting protocol. The imprint molecule is mixed with monomers, which have the ability to interact non-covalently with chemical functionality of the template, and cross-linking monomer in an apolar organic solvent. Typically one or several of functional monomers, such as methacrylic acid, trifluoromethylacrylic acid, 2- or 4-vinylpyridine, methacrylamide, or hydroxyethylmethacrylate, are used together with a cross-linking monomer, such as ethylene glycol dimethacrylate, trimethylolpropane trimethacrylate or divinylbenzene. Following addition of an azobis-nitrile initiator, the polymerisation is conducted either by elevation of temperature or by irradiation of UV-light. Extraction removes the imprint substance and leaves behind imprints, which have a chemical and topological memory for the original template.

48

Chapter 2

scope of this review, instead consultation of the excellent reviews by Sellergren [ 12] and Wulff [ 13] is recommended.

2.2.1 Removal of template molecules MIPs are made in the presence of high concentrations of template and despite exhaustive washing, trace amounts of the imprint species may remain in the MIP and may later leak during use, so-called template bleeding. While being less critical for most chromatographic and other continuous-flow applications, near-quantitative removal of the imprint species is crucial for a sensitive SPE application, where even a small amount of bleeding may greatly affect the assay result. Template bleeding has repeatedly been observed [36-43]. Generally, a more thorough extraction yields a MIP where more of the high-avidity sites are free, leading to a material better equipped to bind analyte from highly diluted samples and less prone to bleed template molecules at use. Complete extraction requires extensive washing using solvents with strong ability to swell the MIP, such as chlorinated solvents, and solvents with strong elution power, such as aqueous methanol or ethanol, containing acid or base. Alternating acid and base washings may be beneficial. Heat-treatment of the polymer accompanied with excessive washes with strong eluents has been tried and this treatment was claimed to greatly reduce or even eliminate bleeding of template from the MIP phase [44]. Since in this study relatively high sample concentrations were extracted it remains, however, to be seen how useful this procedure will be for trace analysis. In a comprehensive investigation Ellwanger et al. studied the influence of various post-polymerisation treatments, such as thermal annealing, microwave assisted extraction, Soxhlet extraction and supercritical fluid desorption, on the level of bleeding [45]. While microwave assisted extraction using trifluoroacetic acid or formic acid was found to be the most efficient extraction technique, also polymer degradation and loss of selectivity were observed. Neither of the treatments eliminated template bleeding completely and it was concluded that, at present, the bleeding problem therefore appears to be best solved by the use of an analogue of the analyte as the template [45].

2.2.2 Choice of template The obvious choice is, of course, to use the target analyte as the template and this approach is employed in the vast majority of molecular imprinting studies. In at least two situations, however, a different approach may be required: (i) when the template is insoluble in the polymerisation solution and (ii) when template bleeding causes problems (see above). To circumvent these problems an alternative template, which is a structural analogue of the analyte(s) of interest, can be used for the MIP preparation. The analyte and the alternative template should possess common structural features such that the template used gives rise to imprints that have the ability to bind the target analyte(s) (Fig. 2.2). The wealth of data already available on template structure-imprint selectivity relationships will often aid the selection of a suitable structural analogue. A

%

~,,~~ "-3

~..,d.

~,,~o

t~

@

Fig. 2.2. Imprinting of a close structural analogue generates imprints with a chemical complementarity for the target analyte.

4~

50

Chapter 2

typical example of difficult templates is peptides, which due to their low solubility in the pre-polymerisation mixture normally are not amenable to non-covalent molecular imprinting. In order to circumvent the limited solubility Leu-enkephalin anilide has been substituted for the target ligand for the preparation of a Leu-enkephalin selective MIP [46]. Similarly, amino acids are insoluble in organic solvents and L-phenylalanine anilide has been used for the preparation of a MIP, which recognized the enantiomers of the parent amino acid phenylalanine [47]. In both instances, conversion of the carboxylic acid functionality into an anilide improved solubility while the sub-site formed for the amide functionality also bound a carboxylic acid. Hence, the imprints were able to rebind the parent structure with a free acid in this position. The resultant MIPs were used, respectively for fundamental investigations into molecular recognition of enkephalin peptides and for capillary electrochromatographic separation of the enantiomers of phenylalanine. For solid-phase extraction purposes, where even low levels of template bleeding would interfere with the quantification, the alternative-template species approach is often used for preparation of the MIE Should template bleeding occur, the subsequent analytical separation will resolve the template from the analyte(s) and the quantification will still be accurate. This was first demonstrated by the preparation of a MIP for extraction of sameridine using a close structural analogue [36]. Another example is the imprinting of trialkylmelamines for the preparation of MIPs capable of recognition of the related triazine structures [48]. In the LC mode these MIPs retained atrazine and related triazines in a group selective manner, whereas structural unrelated agrochemicals were not adsorbed to the imprinted polymer, and in a subsequent study the MIP was used for selective SPE of atrazine from a mixture of herbicides [49]. The alternative template protocol has been used in many more investigations (Table 2.1), however, the availability at reasonable cost of a suitable analogue has to be evaluated on a case-bycase basis.

2.2.3 Format of polymer Bulk polymerisation, followed by grinding and particle sizing is the most commonly used technique for preparation of non-covalent MIPs [12]. The grinding process produces, however, irregular particles as well as a considerable quantity of fine particles which have to be removed, for instance by sedimentation. Typically about 20-60% of the ground polymer is recovered as useable particles. While the procedure is simple and requires no specialist knowledge or speciality equipment, and works well for MIP synthesis in the research laboratory, it is probably not suitable for preparation uniform, high quality particles on a large scale. Furthermore, ground particles are irregular and non-uniform in size, which renders them technically difficult to pack into columns. Means to improve particle shape have included dispersion polymerisation, which starts with a homogenous solution of monomers and template, where, as the polymerisation proceeds, the growing polymer becomes insoluble and precipitates out as particles. This technique has been used for the preparation of MIP particles for pentamidine extraction [50]. Monodisperse imprinted beads can be made by a two-step swelling technique [51 ].

51

Solid-phase extraction on molecularly imprinted polymers

TABLE 2.1 SOME EXAMPLES OF MIPs PREPARED USING A STRUCTURAL ANALOGUE AS THE TEMPLATE Analyte

Template

H2N~ ~ / N

H Atrazine

Reference

~N

NH~ 2-Aminopyridine

4-Aminopyridine

N

MIP system j

L N~

N"

~ N ~ N ~ NH

[76]

MAA/EGDMA, suspension polymerisation

[49]

"N

H~ Dibutylmelamine

H

I

MAA/EGDMA

.~.

4: ~ Bupivacaine OH H2

[41,57]

OH H2N

8r

Brombuterol

CI

Clenbuterol

MAA/EGDMA on silica fiber

[79]

2-VPy/EGDMA

[91 ]

4-VPy/EGDMA, multistep swelling polymerisation and modification of surface with hydrophilic layer

[42]

MAA/EGDMA

[36]

O II O--P--OH

O II O-P--OH

Diphenylphosphate

~

MAA/EGDMA

Pentycaine

Ditolylphosphate

OH

~ o ~ ~ O

S-Ibuprofen

OH

S-Naproxen

L.. Sameridine

\ N,N-dimethyl-analogue of sameridine

MAA, methacrylic acid; 2-VPy, 2-vinylpyridine;4-VPy, 4-vinylpyridine;EGDMA, ethylene glycol dimethacrylate

References pp. 69-71

52

Chapter 2

Aqueous two-step swelling employs a suspension of latex seed particles, which are first swelled with solvent and then with monomers and template to the desired particle size prior to polymerisation. In a more recent study, the surface was made hydrophilic through a second polymerisation of a mixture of glycerol mono- and dimethacrylate onto the beads. Chromatographic evaluation showed that, in contrast to the 10% recovery for the unmodified surface, bovine serum albumin was not retained on the hydrophilic beads and was recovered quantitatively, and the internal chiral recognition sites for S-naproxen remained unchanged [52]. This material was used in a coupledcolumn system for direct injection of rat plasma [42]. Potential problems with aqueous and polar solvent suspension media are their partial distribution into the organic droplets where they interfere with monomer-template complex formation, as well as the risk of less lipophilic monomers and templates distribute into the suspension phase. Both processes were attenuated in a novel suspension polymerisation protocol by which small droplets of imprinting mixture were polymerised in a continuous phase of an inert liquid perfluorocarbon [53]. Like water the perfluorocarbon phase is immiscible with the monomers and solvent used, however, should some partition occur the perfluorocarbons do not participate in hydrogen bonding and therefore leave the monomer-template complexes intact. A promising recent approach is the grafting of thin films of imprinted polymer in the pores of preformed well-defined silica particles [54], which may be useful in particular for online extraction using coupled-column systems.

2.3 MISPE METHOD DEVELOPMENT STRATEGIES

Efficient MISPE method development requires an understanding of the selective and non-specific binding modes to the MIE Generally, a MIP is best characterised as being a mixed-mode separation material containing, in addition to the imprinted affinity sites, both polar and lipophilic surface functionality. Hence, under normal-phase conditions non-specific physicochemical retention is due mainly to polar interactions and under reversed-phase conditions to hydrophobic interaction. Also, the selectivity of the imprint-analyte interactions is tuned by the solvent properties of the surrounding medium. For each compound, analyte as well as all other components of the sample, the observed retention on a MIP column is the sum of selective and non-specific retention modes through interaction with imprints and polymer surface, respectively. If the nonspecific retention mechanisms dominate, the analyte will be retained mainly through adsorption to the polymer surface and any selectivity shown by the imprints may remain undetected. The optimisation of the sample loading conditions and of the wash and elution steps of the MISPE method should be based on an understanding how the strength and nature of imprint-analyte and polymer surface-analyte interactions, respectively, vary with the type of solvent or buffer employed. The following paragraphs discuss problems encountered with the use of MIPs for SPE and highlight some issues relevant to MISPE method development.

Solid-phase extraction on molecularly imprinted polymers

53

2.3.1 Non-specific adsorption Due to the hydrophobic nature of the polymer, extraction of aqueous samples often results in moderately and highly lipophilic compounds being non-specifically adsorbed to the MIE This hydrophobic driven adsorption can be reduced by the addition of an organic modifier [55,56], such as ethanol, methanol or acetonitrile, or a detergent to the sample. The upper limit of the organic modifier content is, however, in part dependent on the type of sample, as protein precipitation may occur for plasma samples. Detergents tested and found useful include Tween 20, Triton X-100 and Brij 35 [57]. Also, buffer pH may influence the extent of non-specific adsorption, with an increase with increased pH being seen for adsorption of basic compounds to MAA-type polymers [55,56] and a decrease with increased pH for adsorption of acidic compounds to VPy-type polymers [58]. Non-specific adsorption may also be reduced by the use of small amounts of MIP, thereby reducing the polymer surface area available for lipophilic adsorption. For off-line extractions using columns or cartridges polymer amounts ranging from 2 g down to 15 mg have been used. As most applications deal with trace analysis, binding capacity does not seem to be a problem.

2.3.2 Solvent switch Following extraction of aqueous samples analyte molecules are retained both imprintbound and adsorbed non-specifically through hydrophobic interactions, and a wash step is required to improve MIP selectivity and to remove all other adsorbed sample components. A solvent switch, e.g. to dichloromethane or acetonitrile, changes the retention conditions to the normal-phase mode, which leads to re-distribution of the analyte to the imprinted sites and washing off of non-related structures (Fig. 2.3). In apolar solvents the selective imprint-analyte binding, which is due mainly to hydrogen bonding and electrostatic interactions, is strong and non-specific hydrophobic adsorption is weak. For environmental analysis hydrophobic, non-specific adsorption can be employed for capturing the analyte from a large volume of water passed through the column. A subsequent solvent switch assures a selective MISPE method. This protocol was first introduced by Takeuchi and co-workers [59] and later used by others [60-63].

2.3.3 Elution Due to the often, strong imprint-analyte affinity, difficulties in effecting quantitative elution of the analyte have been encountered in some cases [36,41,64-65]. This is most pronounced for extraction of strong bases, such as compounds with amino functionalities, on MAA type MIPs where typically eluents, consisting of acetonitrile or methanol containing (sometimes high percentages of) acetic acid, TFA or TEA are used [37,40-41,44,64,66,67]. Sometimes very harsh elution conditions are required, an example of this is the batch-mode extraction of sameridine which used a mixture of 5M References pp. 69-71

4~

OO

9

Q

Fig. 2.3. Loading of an aqueous sample results in the analyte, as well as other compounds, are retained through hydrophobic adsorption to the polymer surface, reversed-phase mode retention. A solvent switch to an organic solvent changes the retention conditions to the normal-phase mode, which leads to re-distribution of the analyte into imprinted sites and removal of hydrophobically bound contaminants. A second switch to a more polar solvent elutes the analyte.

t,o

Solid-phase extraction on molecularly imprinted polymers

55

sodium hydroxide, ethanol and heptane [36]. For neutral compounds, weak acids and weak bases, however, complete elution may occur simply by treatment with polar solvents or mixtures of polar solvents and water [42,49,59,60,68-71 ].

2.3.4 Template bleeding Each method development must include a confirmation that template bleeding does not interfere with the assay and gives rise to poor accuracy and precision. Control over template bleeding is particularly important when dealing with trace analysis. The risk is most severe for off-line extractions protocol using fresh material for each extraction, whereas for on-line protocols the MIP column is constantly washed by the continuous flow.

2.4 SOLID-PHASE EXTRACTION APPLICATIONS MISPE has been demonstrated in a number of proof of concept studies as well as been applied to pre-concentration of drug compounds in plasma and urine samples (Table 2.2), pollutants in environmental water and soil samples (Table 2.3) and controlled substances in tissue and urine (Table 2.4). The following paragraphs summarise publications according to the experimental set-up used.

2.4.1 On-line extraction systems On-line systems reported include both systems where the sample is injected directly on the MIP column and those where the MIP column is placed after a trapping column, where the latter system uses a pre-column to capture the analyte from an aqueous sample and transfer it to a solvent in which MIP-analyte binding is selective. An ,example of the second type is a MIP-C18 coupled-column system described by Bjarnason et al. for detection of four triazine herbicides, simazine, atrazine, propazine and terbuthylazine [71 ]. Samples, consisting of urine, apple extract or water containing humic acid, were spiked with the triazines and injected into a C18 pre-column which t~rapped the analytes from the aqueous sample. Elution with acetonitrile transferred the analytes into the MIP column, which under these conditions selectively bound the triazines. Subsequent elution with water and C18-based analytical reversed phase liquid chromatography completed the separation. Whereas humic acid was efficiently removed, urine and apple extract had some tendency to be retained by the MIP column. Enrichment, with enrichment factors of up to 100, was observed in all cases. Koeber et al. used an analogous solvent-switch approach for determination of nine triazines in river water samples [72]. Following injection of the water sample on a restricted access column, elution with acetonitrile transferred the analytes into the MIP column. A restricted access material (RAM) combines size exclusion and adsorption chromatography [73]. The material has a hydrophilic outer surface, which permits high molecular References pp. 69-71

TABLE 2.2 BIOANALYTICAL AND PHARMACEUTICAL APPLICATIONS OF MISPE Analyte

Type of sample

Off-line mode extraction Atenolol Aqueous standards Bupivacaine Human plasma Brombuterol

Caffeine

Human urine Buffer standards Acetonitrile standards Human plasma

Chlorphenamine Darifenacin

Cola beverage Water Plasma

Hydroxycoumarin Nicotine and oxidation products Phenytoin Propranolol

Sameridine Tamoxifen

Urine Chewing gum Plasma Dog plasma Rat bile Human urine Aqueous standards Toluene standards Human plasma Human plasma Human urine Acetonitrile standards

Treatment prior to MISPE

None Dilution with citrate buffer pH 5 None

Analytical separation and detection

GC-NPD LC-electrochemical detection

Protein precipitation with methanol + acetonitrile and dilution with water Dilution with water. None Protein precipitation with acetonitrile None Extraction in ethyl acetate

LC-UV

None None

LC-UV LC-UV

Extraction in heptane and addition of ethanol None

LC-UV LC-APCI-MS-MS CZE-UV diode array LC-UV diode array

GC-NPD LC-UV

Concentration range

Reference

[83]

10 I~g/mL 160-1000 nmol/L

[41,571

100 ng/mL

[79]

10 ixg/L

[82]

0.02-1 txmol/L - 50 ~zg/mL, 25-1600 pg/mL 10-50 ~g/mL 0.04 mg/mL 2.5-40 txg/mL 2.5 txg/mL

[84] [39,85] [86] [44]

2.5 ~g/mL

[87] [38,64,65, 67,80,81]

2.5 ~g/mL 2.5-20 ixg/mL 10 txg/mL 8-120 nmol/L

[36]

500-1000 ng/mL

[37,88]

t,o

% % t% r~

, ,Iq

r~

TABLE 2.2 CONTINUED Analyte

Type of sample

Treatment prior to MISPE

Analytical separation and detection

Concentration range

Reference

Rat plasma Rat plasma Human serum Human plasma

None None None Extraction on RAM-SPE and elution with acetonitrile

LC-UV LC-UV LC-UV LC-fluorescence LC-UV

0.2-50 ~g/mL n.d. 0.5-100 p~g/mL 10 ng/mL 10 ~g/mL

[42] [42] [75] [74]

Extraction in chloroform Extraction in methanol -0.1N NaOH (1:1) and dilution with methanol Dilution with phosphate buffer pH 5 and acetonitrile Extraction in chloroform

UV UV

2.5-100 txg/mL 1.8-1000 Ixg/mL

[76] [77]

UV

10-60 nM

[50]

UV

0.25-1000 ~g/mL

[68-70]

On-line mode extraction

Ibuprofen Naproxen Propranolol Tramadol

%-,

Extraction with direct detection

4-Aminopyridine Nicotine

Human serum Tobacco

Pentamidine

Urine

Theophylline

Human serum Chloroform standards

--.,I

TABLE 2.3 ENVIRONMENTAL APPLICATIONS OF MISPE Analyte

Off-line mode extraction Atrazine Bentazone Chlorophenoxyacetic acids Chlorotriazine pesticides Chlorotriazine pesticides

Type of sample

Treatment prior to MISPE

Analytical separation and detection

Aqueous standards Aqueous standards pH 5 River water

None

LC-UV LC-UV CZE-UV

Ground water Sediment samples Tap and groundwater Soil Corn

Diphenylphosphate Diquat Nerve agent degradation products Phenylurea herbicides

Methanol standard Drinking water Agricultural soil Human serum

Water Soil

Simazine

Aqueous standards

On-line mode extraction Aquacheck samples Chlorotriazine River water pesticides

Acidification to pH 4

LC-UV diode array Soxhlet extraction with methanol Extraction on PS-DVB SPE disks and re-dissolution in toluene Extraction in acetone and re-dissolution in toluene Extraction in acetonitrile and re-dissolution in toluene None None Suspension in water Extraction in acetonitrile

Extraction on PS-DVB SPE disks and re-dissolution in toluene Extraction in acetone and re-dissolution in toluene None Extraction on C18 RAM-SPE and elution with acetonitrile

MEKC-UV diode array

Concentration range

Reference

0.1 Ixg/mL 10 mg/L 2-10 tzg/L

[49] [89] [62]

20 ~g/L

[60]

0.1-0.5 p~g/L

[90]

100 ~g/L 100 ~g/L 2.1 ~g/mL 3-18 Ixmol/L

[91] [92]

0.2-10 ~xg/g

[93]

LC-UV diode array

1 ~g/mL

[94]

LC-UV diode array

0.1 lxg/mL

[591

0.1-2 ng/mL

[72]

LC-ESI-MS Differential pulse voltammetry CE-UV

LC-APCI-MS

% r~

I

~..~~

TABLE 2.3 CONTINUED Analyte

Type of sample

Treatment prior to MISPE

Analytical separation and detection

4-Nitrophenol Triazine herbicides

River water Water containing humic acid Apple extract

Acidification to pH 2.5 Extraction C 18 SPE column and elution with acetonitrile Extraction in methanol, re-dissolution in buffer, extraction on C18 SPE column and elution with acetonitrile Extraction C18 SPE column and elution with acetonitrile

LC-UV LC-UV

None

Differential pulse voltammetry

Concentration range

Reference

10 txg/L 0.5 ng/mL

[61,63] [71]

t...,.

Urine

20 ng/mL

20 ng/mL

Extraction with direct detection

Pirimicarb

Tap water, spring water, fiver water and sea water

71.5 ~g/L

[78]

TABLE 2.4 AGRICULTURAL AND FOOD CONTROL APPLICATIONS OF MISPE Analyte

Type of sample

Treatment prior to MISPE

Analytical separation and detection

Concentration range

Reference

LC-UV ELISA LC-electrochemical detection LC-UV diode array

0.005-0.5 ppm 0.005-0.5 ppm 100 ng/mL

[66]

5 ng/mL 5 ng/mL 20 ng/g 5 ng/g liver

[95]

LC-UV

250 Ixg/L

[96]

LC-UV

8.8 mg/L

[97]

Off-line mode extraction Atrazine

Beef liver homogenate

Extraction in chloroform

Clenbuterol

Calf urine

None

Clenbuterol

Milk replacer, Bovine urine Bovine liver Liver samples

Chloroform standard

Extraction on Extrelut 20 and residue resuspended in phosphate buffer (pH 3.4)-acetonitrile (3:7; v/v) Matrix solid-phase dispersion on C18-sorbent and elution with acetonitrile- 1% acetic acid None

Merlot red wine

None

Clenbuterol

Indoleacetic acid Quercetin

LC-electrochemic al detection

[40]

[43]

r

t,~

Solid-phase extraction on molecularly imprinted polymers

61

weight humic substances and large proteins to flow though the column without retention, while smaller molecules, such as triazines, are retained on the hydrophobic inner surface. The RAM column reduces the concentration of matrix components and the MIP column selectively retains the triazines whereas the residual matrix molecules are not retained and separated completely. The cleaned and enriched extract is subsequently eluted with methanol-water-acetic acid to a C18 column and analysed by LC-MS. The accuracy of the RAM-MIP-LC-MS system was checked using a certified reference material (Aquacheck). The deviation from the certified value was below 9% and the relative standard deviation was better than 8%. The applicability of the method to the clean-up of real samples was demonstrated by injection of contaminated fiver water samples and the results gave good agreement with previous determinations. Boos and Fleischer used a very similar system for tramadol analysis [74]. Serum samples were injected on a RAM column, transferred with acetonitrile into the MIP column and eluted with buffer at low pH for the final reversed phase analytical separation. Detection with fluorescence or UV revealed that all biological matrix components were eliminated. The purpose of the water-to-solvent switch is to first quantitatively trap the analyte from the aqueous sample and then change the solvent to one in which the MIP binds the analyte in a highly selective manner, and in which non-specific MIP-analyte adsorption is weak or absent. The solvent switch can also follow the (non-specific) adsorption of the analyte on the MIP, thus permitting direct injection of the sample onto the MIP column. This was first demonstrated by Takeuchi and co-workers for the extraction of triazines in the off-line mode [59] (see below) and later employed by Masque et al. for the selective on-line MISPE of nitrophenol [61,63]. Spiked river water samples were acidified to pH 2.5 and applied to the MIP-column. An intermediate dichloromethane wash removed non-specifically bound compounds, including other phenolic structures, and strengthened the selective irnprint-nitrophenol binding. Finally, the analyte was eluted and transferred to the analytical column by acetonitrile containing 1% acetic acid. Three different MIPs were evaluated: two prepared by the non-covalent approach using either methacrylic acid or 4-vinylpyridine as monomer and one prepared by the semicovalent approach where the monomer was 4-nitrophenyl methacrylate. The 4-VPy-MIP gave best recovery and was the most selective one of the three. The MIP was compared with a commercially available highly cross-linked poly(styrenedivinylbenzene) resin (LiChrolut EN) for extraction of fiver water and the former yielded cleaner extracts (Fig. 2.4). Haginaka and co-worker developed a column-switching system, consisting of a RAM-MIP and a conventional C18-column, for direct injection of serum and determination of ibuprofen and naproxen [42]. The RAM-MIP was prepared by an initial multistep swelling and thermal polymerisation protocol using 4-vinylpyridine as the functional monomer, followed by coating of the polymer outer surface with an external hydrophilic layer by a second polymerisation of an equimolar mixture of glycerol monomethacrylate and glycerol dimethacrylate. Plasma samples could be injected directly on the RAM-MIP column, which previously had been equilibrated with a mixture of phosphoric acid pH 2.2 and 20% acetonitrile, and proteinaceous components washed away. Elution and transfer of the analyte to the analytical reversed References pp. 69-71

62

Chapter 2

18000

2

3-4

IIEK)O0

14000 8..9

6

12000

7' ^

10.

10000

8OOO

6OOO i

10

.

.

.

.

,

2O

-

(n~)

......

3O

Fig. 2.4. LC-UV chromatogramsobtained by on-line SPE with a 4-nitrophenol-MIP(a, c) and LiChrolut EN (b, d) of 10 mL of Ebro fiver water spiked at 10 Ixg/L with 11 phenolic compounds. (a, b) with a washing step using dichloromethane and (c, d) without a washing step. Peak 2 is 4-nitrophenol. Reproduced with permission from [61].

phase column was effected with a mixture of phosphate buffer pH 7.3 and 25% acetonitrile. Leakage of imprint molecules prevented accurate determination of the drug, a problem that could be overcome by the use of a RAM-MIP prepared against naproxen for determination of ibuprofen. By this approach ibuprofen was determined in rat plasma with good precision and accuracy in the concentration range of 0.2-50 I~g/mL, which was sufficient for pharmacokinetic measurements. The column could be used for up to 500 injections. An automated and on-line MIP solid-phase micro extraction (SPME) method has been developed by Mullett et al., and its versatility was demonstrated by the determination of propranolol in biological fluids [75]. The system consisted of a propranolol MIP packed in a fused-silica capillary column, preconditioned with

Solid-phase extraction on molecularly imprinted polymers

63

acetonitrile, in which the serum sample was repeatedly drawn and ejected 10 to 20 times through the capillary for extraction. The extracted analytes were directly desorbed from the capillary by water-methanol containing 0.2% trifluoroacetic acid for transport to the LC column. Excellent method reproducibility and column reusability ( > 500 injections) were observed over a fairly wide dynamic range of 0.5-100 ~g/mL in serum samples. The method showed improved selectivity in comparison to alternative in-tube stationary-phase materials, and the MIP extraction was claimed to overcome present selectivity limitations of existing SPME coating materials.

2.4.2 Extraction systems with direct detection Provided the extraction is sufficiently selective, the downstream analytical separation can be omitted and the eluent directed directly to the detector for analyte detection. Sample pre-concentration with direct in-line UV-detection of the analyte following elution from the MIP-column was first shown by Sellergren in a model study on the determination of pentamidine in urine [50]. A mixture of spiked urine, phosphate buffer pH 5 and 70% of acetonitrile was loaded onto a pentamidine MIP-column and elution was effected by lowering the pH to 3, using the same buffer-acetonitrile composition. An enrichment factor of up to 54 was achieved. In a series of papers, Mullett and Lai explored the use of a MIP micro-column and direct in-line UV-detection for determination of theophylline in serum [68-70]. Human serum was extracted with an equal volume of chloroform and an aliquot of the organic layer was injected into the MIP micro-column heated at 60~ Non-specific adsorption of interfering drugs was eliminated by an intermediate wash with a pulse of acetonitrile, followed by quantitative desorption of the bound theophylline by a pulse of methanol and in-line UV determination. The method showed good accuracy and precision over the linear dynamic range of 2-20 ~g/mL, which was concluded adequate for therapeutic monitoring of the drug. This protocol, which was termed MISPE with differential pulsed elution, has also been applied to the determination of 4-aminopyridine in human serum [76] and nicotine in tobacco [77]. Due to its more polar nature and therefore stronger binding to the imprints, acetonitrile could be used as the mobile phase for extraction of nicotine. In this instance, the sample was injected dissolved in methanol, the column was washed with a pulse of methanol and the analyte eluted with 1% TFA in water. Recently, a similar approach was employed for MISPE of the analyte primicarb from water samples, in this instance quantification was done by differential pulse voltammetry [78]. 25 mL of spiked water samples including tap, spring, fiver, and sea water was loaded onto the MIP-microcolumn. Elution with an enrichment factor of 50 was effected with methanol containing 20% water and 10% acetic acid.

2.4.3 Off-line extraction systems Despite great success with on-line coupled-column systems, various off-line SPE techniques are still the more popular approaches with respect to the number of samples processed in routine sample pre-concentration. Likewise, the majority of studies into References pp. 69-71

64

Chapter 2

MISPE have used cartridges or columns off-line from the downstream analytical separation. A study by Andersson addressed mainly problems encountered with direct extraction of bupivacaine from human plasma [41 ]. The template was a structural analogue. The conditions for efficient and quantitative binding of the analyte from the plasma sample, intermediate wash steps and elution were optimised through a series of radioligand binding experiments. The final MISPE protocol consisted of adjusting pH of the human plasma samples by addition of citrate buffer pH 5, containing ethanol and Tween 20, prior to sample loading, washing with 20% methanol in water followed by acetonitrile, and elution of the analyte using 2% TEA in acetonitrile. The low recovery of 65-75% was due mainly to incomplete elution. TFA-acetonitrile based elution did not improve recovery and the eluates were found less pure. A direct comparison with conventional sample pre-treatment methods showed the MISPE method resulted in cleaner chromatographic traces than were obtained both after liquid-liquid extraction and C18based SPE (Fig. 2.5). To identify suitable solvents for loading, washing and elution, Ensing and co-workers undertook a feasibility study into the use of a clenbuterol-MIP for pre-concentration of this compound from calf urine [40]. Neat urine was loaded onto the MIP column, which was washed with 1% acetic acid in acetonitrile prior to elution of the analyte using 10% acetic acid. Freeze-drying of the urine sample and re-dissolution in acetonitrile prior to loading increased recovery up to 100%. Template bleeding was found to prevent accurate determination of trace levels of clenbuterol and it was concluded that future studies should use a MIP made against a structural analogue. Hence, the next study employed a bromoclenbuterol-MIP for determination of residue clenbuterol in liver tissue [43]. The combination of MISPE with matrix solid-phase dispersion (MSPD) was investigated for its potential to simplify post-MSPD sample clean-up treatment. Liver samples were ground in a mortar together with C 18 sorbent, the homogenised mixture packed into an SPE cartridge and placed on top of a MISPE cartridge. Elution with acetonitrile containing 1% acetic acid, which previously was found a good solvent for selective imprint-clenbuterol binding, transferred the analyte to the MIP column. Finally, clenbuterol was eluted with acidic methanol and determined by LC electrochemical detection (LC-ECD) with a recovery for the complete extraction exceeding 90%. Using LC ion trap mass spectroscopy (LC-IT-MS) analysis the method detection limit was <0.1 Ixg/kg, which satisfies regulatory requirements for food control of this substance. Furthermore, MISPE has been employed in the solid-phase microextraction mode with silica fibres coated with a 75-1xm layer of clenbuterol-MIP [79]. Extraction efficiency was evaluated for five structural analogues of clenbuterol, which could all be extracted selectively from acetonitrile and through non-specific adsorption from phosphate buffer pH 7. Brombuterol was extracted by non-specific adsorption from human urine by immersion for 45 minutes, then a solvent switch through immersion for 5 minutes in acetonitrile re-distributed the analyte to the selective imprint sites and removed most matrix components. Finally, the analyte was desorbed by methanol 10% acetic acid and analysed by LC-ECD. Several investigators have evaluated the potential to use MISPE prior to multi-residue analysis of triazine type herbicides in environmental water samples. The group of

% c~ (% r~

C18-SPE

MIP-SPE

=L

% r~ (%

count

counts

8OOO1

I.S. Bupivacaine 1 ! Template

,,,4 6000(

40001

. -9

200~

_

A

_-

.

ivacaine 8000q

;= 600~

4000

2O0O 5

4

6

7

8

(%

__.

9

min

L ...... , _ _ _ _ . ....

4

9. . . . o

5

A i

-

7

6

8

9

min

(%

liquid extraction

Liquid-

SPE on REF

(%

count

I.S. Bupivacaine

count 8O0O1

I.S. Bupivacaine

600~

800~

600(~

400~ 400(~

2000

A 9

4

-

_

_

.

.

5

.

.

.

.

.

6

.

.

.

.

7

.

.

.

.

.

8

.

.

.

.

,

9

min

2000

---

4

9

--"

5

_

_JL

L

6

7

8

9

min

Fig. 2.5. GC-NPD chromatograms of human plasma spiked at 735 nmol/L with bupivacaine and subjected to either MISPE on a pentycaine-MIE SPE on a nonimprinted reference polymer, SPE on a C18 column or liquid-liquid extraction. Reproduced with permission from [57].

Lab

66

Chapter 2

Takeuchi studied the use of an atrazine-MIE prepared by suspension polymerisation, for selective extraction of simazine from water [59]. Of an aqueous mixture of simazine and some structurally unrelated agrochemicals all compounds were found retained on the MIP column. A solvent switch to dichloromethane removed the contaminants whereas non-specifically bound simazine was re-distributed to the selective imprints, and finally methanol effected elution. The same group used dibutylmelamine as a structural analogue to prepare a MIP, which was assumed useful for selective extraction of triazines [49]. MISPE of the above aqueous mixture of herbicide standards, using the same solvent-switch protocol, confirmed that the MIP indeed selectively extracted atrazine with high recovery. Employing essentially the same solvent-switch protocol Barcelo and co-workers used a terbuthylazine MIP for enrichment of six chlorotriazines from natural water and sediment samples [60]. The analytes were captured by passing large volumes of water samples through the cartridge, again a solvent switch to dichloromethane removed non-specifically bound contaminants while the selective imprint-analyte binding was strengthened, and methanol eluted the analytes for subsequent LC-diode array analysis. Recoveries were better than 80% for most chlorotriazines and the limits of detection varied from 0.05 to 0.2 p~g/1. Furthermore, natural sediments samples containing atrazine and diethylatrazine were Soxhlet extracted and analysed by the novel method. The chromatograms recorded for sediment extracts following MISPE were cleaner, with better baselines than were obtained using conventional extraction on C18 cartridges. No significant sample matrix interferences were noticed and it was concluded that an additional clean-up step, which in general is necessary for sediment sample preparation, could be avoided. Turiel et al. applied MISPE to the clean-up of drinking water, ground water and soil prior to determination of six chlorotriazines by micellar eletrokinetic chromatography [90]. An atrazine MIP was evaluated by Muldoon and Stanker for its ability to clean up organic solvent extracts of beef liver [66]. The purification protocol consisted of extraction of beef liver with chloroform, loading of the organic layer, column wash using chloroform and elution using 10% acetic acid in acetonitrile. Purified and unpurified beef liver extracts were analysed by both LC and ELISA, and the MISPE was found to improve accuracy of both methods, and improve precision and lower limit of detection of the LC method. MISPE for the clean-up of chlorinated phenoxy acids from fiver water samples prior to CE analysis was investigated by Baggiani et al. [62]. Large volumes of acidified river water samples, spiked with chlorinated phenoxy acids, were loaded on a 2,4,5-trichlorophenoxyacetic acid-MIE the column was washed with methanol and the analytes eluted with methanol-20% acetic acid. Template bleeding was not detected and the MISPE system was concluded comparable to a conventional C18-SPE in terms of recovery and superior in terms of sample clean-up (Fig. 2.6). Martin et al. investigated the use of a propranolol-MIP for SPE of propranolol and analogues from a variety of aqueous matrices, as well as from toluene [64-65,67,80-81]. Extraction of propranolol from water, bile and urine was found complete, however, some losses of approximately 10% were observed for plasma samples [64]. Adsorption of a range of similar and dissimilar structures in addition to binding of propranolol was seen, and the conditions chosen for elution of the analyte was concluded important to ensure selectivity. Eluents based on methanol-water 1%

67

Solid-phase extraction on molecularly imprinted polymers

oo,ii ii, ,,.....,........'

on.--n ..........

i i - 84 ..............

b 0.008

a

0006

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

~"~'~ ""~0"~

~

~*~'~"~

4

2

0.004

I

0.002 -

0.000

~3

I I

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

4

| ........................ I

5

6

-'.........

I

7

, , 83

I

4

.......

I

5

.......

I

6

................... l " "

7

8

migration time, min

Fig. 2.6. Capillary electropherogram for 100 mL river water spiked at 5 Ixg/L with: (1) 2,4,5-T; (2) 2,4-D; (3) fenoprop; (4) dichlorprop pre-concentrated using either: (a) liquid-liquid extraction; (b) SPE on a C18 column; (c) MISPE on a 2,4,5-T-MIP; and (d) SPE on a non-imprinted reference column. Reproduced with permission from [62].

TEA were found more selective than those based on methanol-water 1% TFA. Template leakage prevented accurate determination of propranolol and the MIP was explored for its ability to extract structural analogues [67]. Adsorption from water and TEA-based elution resulted in class selectivity for a series of structurally very similar compounds, although a slight correlation between extent of retention and structural similarity to propranolol was seen. A similar problem was encountered by the group of Stevenson for extraction of tamoxifen from human plasma and urine using a MIP prepared against the analyte [37]. Clean L C - U V traces were obtained, however, again template leakage prevented reliable quantification of low levels of the drug. Likewise, Venn and Goody pointed out the difficulty of quantitative removal of the template and the consequent effect of bleeding on assay precision and accuracy as a major obstacle to reliable trace

References pp. 69-71

68

Chapter 2

level determination [39]. A darifenacin MIP was found able to extract darifenacin directly from human plasma, protein-precipitated with an equivolume acetonitrile. During washing with acetonitrile, the drug was strongly retained on the MIP whereas early breakthrough was seen for a non-imprinted polymer. Furthermore, the MIP was found selective for sub-structure elements of darifenacin. A phenytoin-MIP was evaluated as a selective sorbent for determination of this substance in plasma [87]. Several washing solvents were studied for their ability to disrupt the non-specific interactions occurring on extraction from plasma and dichloromethane was found most optimal. Phenytoin could be determined in plasma with good precision and accuracy, and at the concentration range studied (2.5--40 Ixg/mL) template bleeding was not noticed. In a recent study, protein precipitated human plasma and diluted cola beverage samples were extracted using a caffeine-MIP followed by analysis of caffeine by HPLC [82]. An investigation into the MISPE conditions found that optimal sample clean-up was obtained through aqueous loading at high pH, followed by a first wash with buffer at high pH, a second wash with acetonitrile 1% TEA and elution with acetonitrile 1% acetic acid. Application of MISPE to selective off-line extraction and pre-concentration of red wine samples was studied by Mizaikoff and co-workers [97]. Imprinting of the flavoinid quercetin yielded a MIR which selectively retained this substance while C18based SPE co-extracted other phenolic compounds.

2.5 CONCLUSIONS The rapid increase in the number of articles published each year testifies to the growing interest in molecular imprinting technology generally, with molecular-imprint based solid-phase extraction, MISPE, arguably being the most advanced sub-area with respect to future use in routine work. As is seen in this review, already several examples of highly selective pre-concentration of biological and environmental samples have been reported. Essential for a successful development of selective MISPE methods is a generic understanding of imprint-analyte binding mechanisms as well as the various physicochemical binding modes to these extraction materials. Compared with immunoaffinity materials MIPs have the clear advantage of being highly stable under a large range of buffer pH, solvent, temperature and pressure conditions, allowing large opportunities for selection of the best experimental conditions for the most efficient use of MIPs. Also, the issue of template bleeding need to be addressed, however, it must be kept in mind that semi-irreversibly trapped template molecules are inherent to the imprinting process. While imprinting of an analyte analogue circumvents the problem rather than solves it, this approach may in many instances be vital to successful use of MISPE in trace analysis. Other urgent areas, and presently focus for intense research, are extension of the types of chemistry available for imprint formation and development of strategies for selection of the best recipe of template, monomers, cross-linkers and polymerisation conditions given a combination of analyte, sample matrix and analytical context. An interesting example is post-polymerisation hydrophilisation of the external surface of monodisperse MIP beads to render them more compatible with proteins in biosamples, thus allowing direct injection of serum samples into a highly selective MIP

Solid-phase extraction on molecularly imprinted polymers

69

p r e - c o l u m n in a c o u p l e d - c o l u m n system. M I S P E with direct d e t e c t i o n of the analyte f o l l o w i n g elution is a p p e a l i n g as it e l i m i n a t e s the n e e d for an analytical c o l u m n , w h i c h w o u l d simplify the overall separation s y s t e m and increase speed of analysis. T h e direct detection a p p r o a c h involves, however, the d e m a n d i n g r e q u i r e m e n t for c o m p l e t e r e m o v a l of all s a m p l e c o n t a m i n a n t s , and, secondly, a p r e s e n t limitation m a y be the large elution v o l u m e s often required. N o t w i t h s t a n d i n g , these p r o b l e m s are addressable, and given the high selectivity of i m p r i n t e d materials M I S P E with direct d e t e c t i o n will likely develop into a viable alternative for e n v i r o n m e n t a l and b i o - s a m p l e analysis.

2.6 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 27 28 29 30 31

C.E Poole and I.D. Wilson (Eds.), Special issue on solid-phase extraction, J. Chromatogr. A, 885 (2000). M.-C. Hennion, J. Chromatogr. A, 856 (1999) 3. D. Martinez, M.J. Cugat, E Borull and M. Calull, J. Chromatogr. A, 902 (2000) 65 N. Masque, R.M. Marc6 and E Borull, Trends Anal. Chem., 17 (1998) 384. I. Ferrer and D. Barcel6, Trends Anal. Chem., 18 (1999) 180. V. Pichon, M. Bouzige, C. Miege and M.-C. Hennion, Trends Anal. Chem., 18 (1999) 219. D. Stevenson, J. Chromatogr. B, 745 (2000) 39. B. Sellergren (Ed.), Molecularly imprinted polymers. Man-made mimics of antibodies and their applications in analytical chemistry. Elsevier, Amsterdam, The Netherlands, 2001. R.A. Bartsch and M. Maeda (Eds.), Molecular and ionic recognition with imprinted polymers. American Chemical Society, Washington, D.C., USA. ACS Symposium Series 703, (1998) G. Wulff, Angew. Chem. Int. Ed. Engl., 34 (1995) 1812. K. Mosbach and O. Ramstrrm, Bio/Technology, 14 (1996) 163. B. Sellergren, in: B. Sellergren (Ed.), Molecularly imprinted polymers. Man-made mimics of antibodies and their applications in analytical chemistry (p. 113). Elsevier, Amsterdam, The Netherlands, 2001. G. Wulff and A. Biffis, in: B. Sellergren (Ed.), Molecularly imprinted polymers. Man-made mimics of antibodies and their applications in analytical chemistry (p. 71). Elsevier, Amsterdam, The Netherlands, 2001. M. Ltibke, M.J. Whitcombe and E.N. Vulfson, J. Am. Chem. Soc., 120 (1998) 13342. J.U. Klein, M.J. Whitcombe, E Mulholland and E.N. Vulfson, Angew. Chem. Int. Ed. Engl., 38 (1999) 2057. G. Wulff, T. Gross and R. Schrnfeld, Angew. Chem. Int. Ed. Engl., 36 (1997) 1962. C. Ltibke, M. Ltibke, M.J. Whitcombe and E.N. Vulfson, Macromolecules, 33 (2000) 5098. B. Sellergren, J. Chromatogr. A, 906 (2001) 227. V.T.Remcho and Z.J. Tan, Anal. Chem., 71 (1999) 248A. T. Takeuchi and J. Haginaka, J. Chromatogr. B, 728 (1999) 1. L. Schweitz, E Spegel and S. Nilsson, Electrophoresis, 22 (2001) 4053. ET. Vallano and V.T. Remcho, J. Chromatogr. A, 887 (2000) 125. L. Schweitz, L.I. Andersson and S. Nilsson, J. Chromatogr. A, 817 (1998) 5. B. Sellergren and L.I. Andersson, Methods, 22 (2000) 92. L.I. Andersson, J. Chromatogr. B, 739 (2000) 163. K. Haupt and K. Mosbach, Chem. Rev., 100 (2000) 2495. S. A1-Kindy, R. Badia, J.L. Suarez-Rodriguez and M.E. Diaz-Garcia, Critical Rev. Anal. Chem., 30 (2000) 291. S.A. Piletsky and A.EE Turner, Electroanalysis, 14 (2002) 317. EL. Dickert and O. Hayden, Fresenius J. Anal. Chem., 364 (1999) 506. E Lanza and B. Sellergren, Chromatographia, 53 (2001) 599. N. Masque, R.M. Marce and E Borrull, Trends Anal. Chem., 20 (2001) 477.

70

Chapter 2

32 33 34 35 36 37 38

J. Olsen, R Martin and I.D. Wilson, Anal. Commun., 35 (1998) 13H. A. Martin-Esteban, Fresenius J. Anal. Chem., 370 (2001) 795. B. Sellergren, Trends Anal. Chem., 18 (1999) 164. K. Ensing, C. Berggren and R.E. Majors, LCGC, 19 (2001) 942. L.I. Andersson, A. Paprica and T. Arvidsson, Chromatographia, 46 (1997) 57. B.A. Rashid, R.J. Briggs, J.N. Hay and D. Stevenson, Anal. Comm., 34 (1997) 303. E Martin, I.D. Wilson, G.R. Jones and K. Jones, in: E. Reid, H.M. Hill & I.D. Wilson (Eds.), Drug development assay approaches including molecular imprinting and biomarkers (pp. 21-27). The Royal Society of Chemistry, Cambridge, U.K. (1998). R.E Venn and R.J. Goody, Chromatographia, 50 (1999) 407. C. Berggren, S. Bayoudh, D. Sherrington and K. Ensing, J. Chromatogr. A, 889 (2000) 105. L.I. Andersson, Analyst, 125 (2000) 1515. J. Haginaka and H. Sanbe, Anal. Chem., 72 (2000) 5206. C. Crescenzi, S. Bayoudh, EA.G. Cormack, T. Klein and K. Ensing, Anal. Chem., 73 (2001) 2171. A, Zander, E Findlay, T. Renner, B. Sellergren and A. Swietlow, Anal. Chem., 70 (1998) 3304. A. Ellwanger, C. Berggren, S. Bayoudh, C. Crecenzi, L. Karlsson, EK. Owens, K. Ensing, E Cormack, D. Sherrington and B. Sellergren, Analyst, 126 (2001) 784. L.I. Andersson, R. Mtiller, G. Vlatakis and K. Mosbach, Proc. Natl. Acad. Sci. USA, 92 (1995) 4788. J.-M. Lin, T. Nakagama, K. Uchiyama and T. Hobo, J. Pharm. Biomed. Anal., 15 (1997) 1351. J. Matsui, K. Fujiwara and T. Takeuchi, Anal. Chem., 72 (2000) 1810. J. Matsui, K. Fujiwara, S. Ugata and T. Takeuchi, J. Chromatogr. A, 889 (2000) 25. B. Sellergren, Anal. Chem., 66 (1994) 1578. K. Hosoya, K. Yoshizako, Y. Shirasu, K. Kimata, T. Araki, N. Tanaka and J. Haginaka, J. Chromatogr. A, 728 (1996) 139. J. Haginaka, H. Takehira, K. Hosoya and N. Tanaka, J. Chromatogr. A, 849 (1999) 331. A.G. Mayes and K. Mosbach, Anal. Chem., 68 (1996) 3769. C. Sulitzky, B. Rtickert, A.J. Hall, E Lanza, K. Unger and B. Sellergren, Macromolecules, 35 (2002) 79. L.I. Andersson, Anal. Chem., 68 (1996) 111. J. Karlsson, L.I. Andersson and I.A. Nicholls, Anal. Chim. Acta, 435 (2001) 57. L.I. Andersson, M. Abdel-Rehim, L. Nicklasson, L. Schweitz and S. Nilsson, Chromatographia, 55 (2002) S-65. K. Haupt, A. Dzgoev and K. Mosbach, Anal. Chem., 70 (1998) 628. J. Matsui, M. Okada, M. Tsuruoka and T. Takeuchi, Anal. Commun., 34 (1997) 85. I. Ferrer, E Lanza, A. Tolokan, V. Horvath, B. Sellergren, G. Horvai and D. Barcelo, Anal. Chem., 72 (2000) 3934. N. Masque, R.M. Marce, E Borrull, EA.G. Cormack and D.C. Sherrington, Anal. Chem., 72 (2000) 4122. C. Baggiani, C. Giovannoli, L. Anfossi and C. Tozzi, J. Chromatogr. A, 938 (2001) 35. E. Caro, N. Masque, R.M. Marc6, E Borrull, EA.G. Cormack and D.C. Sherrington, J. Chromatogr. A, 963 (2002) 169. E Martin, I.D. Wilson, D.E. Morgan, G.R. Jones and K. Jones, Anal. Commun., 34 (1997) 45. J. Olsen, E Martin, I.D. Wilson and G.R. Jones, Analyst, 124 (1999) 467. M.T. Muldoon and L.H. Stanker, Anal. Chem., 69 (1997) 803. E Martin, I.D. Wilson and G.R. Jones, J. Chromatogr., 889 (2000) 143. W.M. Mullett and E.EC. Lai, Anal. Chem., 70 (1998) 3636. W.M. Mullett and E.EC. Lai, Microchem.J., 61 (1999) 143. W.M. Mullett and E.EC. Lai, J. Pharm. Biomed. Anal., 21 (1999) 835. B. Bjamason, L. Chimuka and O. Ramstr6m, Anal. Chem., 71 (1999) 2152. R. Koeber, C. Fleischer, E Lanza, K.-S. Boos, B. Sellergren and D. Barcelo, Anal. Chem., 73 (2001) 2437. K.-S. Boos and C.-H. Grimm, Trends Anal. Chem., 18 (1999) 175. K.-S. Boos and C.T. Fleischer, Fresenius J. Anal. Chem., 371 (2001) 16. W.M. Mullett, E Martin and J. Pawliszyn, Anal. Chem., 73 (2001) 2383.

39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75

Solid-phase extraction on molecularly imprinted polymers 76 77 78 79 80 81 82 83 84 85

86 87 88

89 90 91 92 93 94 95 96 97

71

W.M. Mullett, M.E Dirie, E.EC. Lai, H. Guo and X. He, Anal. Chim. Acta, 414 (2000) 123. W.M. Mullett, E.EC. Lai and B. Sellergren, Anal. Commun., 36 (1999) 217. M.L. Mena, E Martinez-Ruiz, A.J. Reviejo and J.M. Pingarron, Anal. Chim. Acta, 451 (2002) 297. E.H.M. Koster, C. Crescenzi, W. Den Hoedt, K. Ensing and G.J. de Jong, Anal. Chem., 73 (2001) 3140. P. Martin, I.D. Wilson and G.R. Jones, Chromatographia Suppl., 52 (2000) S-19. ED.Martin, T.D. Wilson, I.D. Wilson and G.R. Jones, Analyst, 126 (2001) 757. G. Theodoridis and E Manesiotis, J. Chromatogr. A, 948 (2002) 163. D. Stevenson, Trends Anal. Chem., 18 (1999) 154. W. Chen, E Liu, X. Zhang, K.A. Li and S. Tong, Talanta, 55 (2001) 29. R.E Venn and R.J. Goody, in: E. Reid, H.M. Hill & I.D. Wilson (Eds.), Drug development assay approaches including molecular imprinting and biomarkers (pp. 13-20). The Royal Society of Chemistry, Cambridge, U.K. (1998). M. Walshe, J. Howarth, M.T. Kelly, R. OI~ennedy and M.R. Smyth, J. Pharm. Biomed. Anal., 16 (1997) 319. A. Bereczki, A. Tolokan, G. Horvai, V. Horvath, E Lanza, A.J. Hall and B. Sellergren, J. Chromatogr. A, 930 (2001) 31. D. Stevenson, R. J. Briggs, J. Hay and B. Rashid, in: E. Reid, H. M. Hill & I. D. Wilson (Eds.), Drug development assay approaches including molecular imprinting and biomarkers (pp. 49-51). The Royal Society of Chemistry, Cambridge, U.K. (1998). C. Baggiani, E Trotta, G. Giraudi, C. Giovannoli and A. Vanni, Anal. Commun., 36 (1999) 263. E. Turiel, A. Martin-Esteban, P. Fernandez, C. Perez-Conde and C. Camara, Anal. Chem., 73 (2001) 5133. K. M611er, U. Nilsson and C. Crescenzi, J. Chromatogr. A, 938 (2001) 121. B.B. Prasad and S. Banerjee, Chromatographia, 55 (2002) 171. Z.-H. Meng and Q. Liu, Anal. Chim. Acta, 435 (2001) 121. A. Martin-Esteban, E. Turiel and D. Stevenson, Chromatographia, 53 (2001) $434. G. Brambilla, M. Fiori, B. Rizzo, V. Crescenzi and G. Masci, J. Chromatogr. B, 759 (2001) 27. A. Kugimiya and T Takeuchi, Anal. Chim. Acta, 395 (1999) 251. A. Molinelli, R. Weiss and B. Mizaikoff, J. Agric. Food Chem., 50 (2002) 1804.

This Page Intentionally Left Blank

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 Elsevier Science B.V. All rights reserved

73

CHAPTER 3

Techniquesfor sample preparation using solid-phase extraction U w e Dieter Neue, Claude R. Mallet, Ziling Lu, Yung-Fong Cheng and Jeffrey R. M a z z e o Waters Corporation, 34 Maple St, Milford, MA 01757, USA

3.1 I N T R O D U C T I O N The analysis of drugs and metabolites in body fluids is a challenging task, comparable literally to finding the proverbial needle in a haystack. However, it is a task that is carried out successfully day in and day out at high throughput in the pharmaceutical industry. We have been involved in this type of assays for several years, and we have seen the advances in technology that have moved this analysis to higher sensitivity and to faster speeds. Without question, the most important factor has been the impact of mass spectrometry as the dominant detection tool. The specificity of the MS detection enables us to find the "needle", but we still need good tools to remove a lot of the "hay" which can mask the analytes. In the course of the last few years, we have developed specific tools of sample preparation that aid in the clean-up of biological samples. Most of our focus was on the treatment of plasma samples, but other matrices such as urine, organ homogenates or cell cultures have been encountered as well [ 1-7]. We have developed specific generic techniques that can be used for a range of different analytes, and we describe them in this article. The techniques were originally used off-line, as true sample preparation methods that are independent of the subsequent analysis. More recently, we have investigated the use of the same or similar approaches on-line in combination with MS/ MS or LC/MS/MS [6,8]. The following techniques will be discussed in this article. The 1-D reversed-phase sample clean-up method is a simple technique designed to remove major interferences such as plasma proteins or polar compounds. The 2-D reversed-phase solid-phase extraction method provides much cleaner backgrounds, but the development of the method is more complex. The combination of hydrophobic interaction and ionexchange is used for another more specific sample clean-up technique. This combination can also be used for specific fractionations or the removal of specific References pp. 89-90

74

Chapter 3

contaminants. In addition, we have been able to couple some of these techniques with on-line analysis. Here we describe the principles of the different approaches, together with applications.

3.2 DESCRIPTION OF THE SORBENTS

All of the methods described in this article and the specific applications developed use packings of a particular sorbent group, the Oasis | family of packings. The parent packing is the Oasis | HLB sorbent, a hydrophobic, but water wettable material [9,10]. It is a balanced copolymer of N-vinylpyrrolidone, the hydrophilic component, and divinylbenzene, which is the lipophilic component and responsible for reversed-phase retention. The N-vinylpyrrolidone content of the packing provides the wettability in water or buffers. A good wettability in water distinguishes the Oasis | HLB packing from other reversed-phase sorbents such as silica-based C~8 packings or polymeric styrene-divinylbenzene-based packings. If purely hydrophobic packing materials partially dry during the sample preparation process before the sample is loaded onto the packing, a loss of adsorption capacity is found [9]. This does not occur with the Oasis | HLB packing due to the specific composition of the polymer. This favorable wetting property simplifies the preparation procedures for aqueous samples, such as plasma samples or environmental water samples, especially when parallel sample preparation techniques in multi-well plates are used. This property also provides the name of the packing: HLB stands for Hydrophilic-Lipophilic Balance. A consequence of these properties of the Oasis HLB packing is a rather universal applicability and the good retention of polar compounds [ 11 ]. Due to the fact that the packing is a polymeric sorbent, it can be used without difficulty over the entire pH range [9-11 ]. As we will see below, this property can be used with advantage in method development. In addition, special preparation techniques ensure that the bleed from the packing is negligible, which is an important property for a broad-based sample preparation tool. Mixed-mode ion-exchange sorbents have been created from the basic hydrophobic Oasis | HLB packing. The Oasis | MCX packing (Mixed-mode Cation eXchanger) is a partially sulfonated material with a controlled ion-exchange capacity of 1 meq/g. The cation-exchange capacity allows for the specific retention of positively charged analytes, in conjunction with the hydrophobic properties of the parent packing. The Oasis | MAX packing (Mixed-mode Anion eXchanger) contains quaternary ammonium functions and provides the same properties for negatively charged analytes. The optimum ionexchange capacity was found to be at 0.3 meq/g. The functionality is a butyldimethylammonium group attached via a methylene bridge to the divinylbenzene backbone of the matrix. The structure of both packings is shown in Fig. 3.1. The HPLC instruments used throughout were Alliance | instruments from Waters corporation. For detection in the UV-visible range, we used the Waters 996 Photodiode Array Detector. The Micromass Quattro Ultima triple-quadrupole mass spectrometer (MS) was used for HPLC/MS and HPLC/MS/MS analysis. Chromatography was performed on a variety of column types including Symmetry | SymmetryShield | and

'Techniques for sample preparation using solid-phase extraction

y

75

~'-'C 4H9

so 3

Fig. 3.1. Chemical structure of the Oasis| MAX (left) and the Oasis| MCX (right) sorbents. XTerra | (Waters). The specific properties of each column type are described elsewhere [12-17]. All applications utilized Oasis | packings (Waters) which are available in a broad range of configurations, from syringe-barrel-style cartridges to 96-well plates to prepacked cartridge columns for on-line sample preparation. In addition, a range of different volumes are available for the different devices. For example, the 96-well plate configuration contains 5 mg, 10 mg, 30 mg and 60 mg of sorbents. The techniques described in this article are general, and can be applied to different configurations without difficulty, but the actual sample volumes or the wash volumes have been optimized for the specific configurations used in each application.

3.3 OFF-LINE METHODS The first part of this article focuses on sample preparation methods that are carried out outside an HPLC instrument. These are the off-line sample preparation techniques. In general, off-line methods provide a larger flexibility, since they do not suffer from the constraint that the final elution conditions from the SPE cartridge need to be compatible with the HPLC analysis. In addition, other suitable means, for example gas chromatography (GC) or GC/MS, can be used for the final analysis. On the other hand, on-line techniques, as discussed later, may be more efficient, especially for highthroughput analysis. At the same time, they require a larger instrumental complexity, and a greater experimental ingenuity as well. In the following sections that examine the off-line methods the 1-D reversed-phase solid-phase extraction method is explained together with the 2-D reversed-phase solid-

References pp. 89-90

Chapter 3

76

phase extraction method. In addition, methods using both mixed-mode cation and anion exchangers are described.

3.3.1 1-D Reversed-phase solid phase extraction of biological samples The 1-dimensional reversed-phase sample preparation techniques described here are widely used for plasma or urine samples and similar principles have been used for sample preparation methods outside the pharmaceutical industry as well The term "l-dimensional" refers to the number of variables varied during sample preparation. In this case only the amount of the organic solvent for clean-up of the sample or for elution of the sample from the surface of the sorbent is varied. In the 2-dimensional technique described in the next paragraph, the concentration of the organic solvent and the pH are varied. Let us first consider the preparation of plasma samples, either from animal sources or from humans. Typically only a small sample volume is available, and the size of the extraction cartridge or plate used depends on the sample volume. Table 3.1 gives the size of the cartridges or 96-well plates in dependence of the sample volume. The 1-dimensional sample preparation technique is very straightforward. The sample is prepared by adding an internal standard. Most often, it is diluted about 1:1 with the internal standard solution. This reduces the viscosity of the plasma sample. Often, a small amount of acid or base is added to break the interaction of the analyte with the plasma proteins. The sorbent itself is prepared by wetting it with an organic solvent, e.g. methanol, then water or buffer of the same pH as the sample. Then the sample is applied to the sorbent. The reason for wetting the sorbent prior to the addition of a plasma sample lies primarily in the preconcentration effect that a plasma or serum sample encounters on a dry sorbent. Large molecular weight components of the sample are excluded from the pore space. This increases the interstitial concentration of the proteins present in such samples, thus increasing the viscosity of the sample quite significantly and slowing down the flow of the samples through the cartridges or wells. If the sorbent is wetted first with water or buffer, the preconcentration of proteins is avoided and the sample flows more freely through the prewetted cartridge. Once the sample is adsorbed on the cartridge, it is first washed with 5% methanol to remove proteins and other polar interferences such as carbohydrates. The pore size TABLE 3.1 USE OF OASIS CARTRIDGESOR 96-WELLEXTRACTION PLATES Amount of Sorbent Per Cartridge or Well 5 mg 10 mg 30 mg 60 mg

Maximum Mass

Typical Sample

Elution

Capacity

Volume

Volume

0.15-1 mg 0.35-2 mg 1-5 mg 2-10 mg

10-100 txL 50-200 lxL 100 IxL-1 mL 200 ixL-2 mL

< 150 IxL < 250 ixL <400 IxL <800 lxL

Techniques for sample preparation using solid-phase extraction

77

distribution of the Oasis sorbent is designed to minimize the adsorption of large amounts of protein [9]. Subsequently, the analytes are reextracted from the cartridge with an organic solvent such as methanol. Very hydrophobic compounds, such as steroids, however, may require a stronger extraction solvent. A simple solution is a mixture of 50% methanol and 50% methylene chloride. Such a solvent mixture can be evaporated easily in the subsequent step. Other water-miscible solvents useful for extraction are acetone with an increased elution strength and a high vapour pressure, acetonitrile, tetrahydrofuran or ethanol and the propanols with an increased elution strength over methanol. Generally, good results are obtained with this simple technique [18-27]. This has been true both for the use of extraction cartridges and 96-well plates. The recovery is always good, nearly independent of the analyte. Internal standards are recovered with the same efficiency as the analyte, and one is fairly free in the choice of the standard. The most important components of a plasma sample are removed, and sufficient lifetime of the analytical column is obtained. Users have commented that "this technique is simple and problem-free, with no system clogging problems". As an example of this method, enalapril and enalaprilate were analyzed in human plasma at levels down to 0.5 ng/mL with standard deviations under 4% (Table 3.2, data courtesy of PPD Pharmaco). In this case, the actual analysis was performed with LC/MS/MS as the analytical tool. TABLE 3.2 ENALAPRIL/ENALAPRILATEDETERMINATIONIN HUMANPLASMA Enalapril Plasma Concentration 0.5 ng/mL 5 ng/mL 80 ng/mL

Enalaprilate

% Recovery (n=6)

Precision (% CV)

% Recovery (n=6)

Precision (% CV)

110.9 112.2 107.0

3.6 4.3 1.6

103.6 109.4 104.5

3.0 3.2 1.2

3.3.2 2-D Reversed-phase solid phase extraction of biological samples The 1-dimensional technique described above allows the removal of proteins, carbohydrates and other polar contaminants of a biological sample. However, this approach does nothing for the removal of low molecular weight interferences that coelute in the same retention window as the analytes of interest. This is not a problem if a highly specific detection technique such as fluorescence is used in the final analysis, but with less specific methods such as UV detection one often encounters coelution of the analytes of interest and endogeneous compounds resulting in interference. With electrospray mass spectrometry as the detection tool, ion suppression may occur from

References pp. 89-90

Chapter 3

78

endogenous materials in the samples. In all such cases, a better sample preparation technique is required that permits the removal of these interferences and makes a problem-free high-sensitivity analysis possible. To solve problems of this type, the 2-dimensional reversed-phase SPE technique was developed. In this technique, which will be described in detail in this section, advantage is taken of the differences in the elution profile of analytes and interferences as a function of both the pH of the wash and the concentration of the elution solvents. In the following approach an estimate of the difference in solvent composition required for equal retention of the charged and the uncharged form of an ionizable analyte is made. This is a good generic test of the validity of the technique. HA designates the protonated form of the analyte, for example the acid form of an acidic analyte or the ammonium form of an amine. A- is the deprotonated form, for example the anion of the acid or the deprotonated amine. ln(kna) = ln(ko,HA) -- BHA " d~

(1)

ln(kA- ) = ln(ko,A-) - BA-. dO

(2)

kHA and kz- are the retention factors at the solvent composition ~b, and

kO,HA and k0~- are the retention factors in a fully aqueous mobile phase. The slopes of the relationship of the logarithm of the retention factor with solvent composition, BI4 z and BA-, are often the same (or at least similar) for the charged and the uncharged forms of the analyte. One can then estimate the difference in solvent composition A+ required for equal retention both forms of the analyte, i.e. when k14z and k A- a r e equal.

1

Ad)-B" ln\k0~ -

(3)

Typical values might be a factor of 10 difference in the retention factor and a value of 6 for the slope B for a very small molecule or a factor of 30 difference in the retention with a value of 9 for B for a typical pharmaceutical analyte. In both cases, the difference in the solvent composition for equal retention is roughly:

zx+=0.35 This difference is large enough to be used in efficient sample preparation techniques to obtain cleaner fractions than is possible with the 1-dimensional technique described above. To understand how this difference in retention can be used in sample preparation, let us look at the elution profile of an ionizable analyte in its charged form and its neutral form as a function of the concentration of the elution solvent (Fig. 3.2). The charged form of the analyte exhibits less retention and is completely eluted at a low solvent composition, while the non-ionic form of the analyte is still completely retained. We can therefore create a sample wash protocol that washes the adsorbed sample at a pH and solvent composition that still retains the uncharged form of the analyte. This elutes all contaminants that are more hydrophilic than the analyte of interest. Then the pH value

79

Techniques for sample preparation using solid-phase extraction ~ 1 7. 6 .

C 0 I_

LL

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

0.9 0.8 0.7

J

0.6

0.5 ,~: 0.4 c

!

J

"0 0.3 = 0.2 LLI 0.1 m

~

0

0.2

0.4

0.6

0.8

1

Volume Fraction of Methanol Fig. 3.2. Elutionprofile of a charged analyte (dotted line) and an uncharged analyte (solid line) as a function of the solvent composition

at the same solvent composition is changed to a pH where the analyte is ionized, and eluted rather specifically, while more hydrophobic contaminants remain retained on the SPE cartridge. The only remaining interferences are compounds that have the same charge pattern and the same hydrophobicity as the target compound. This simple experimental approach reduces the number of possible interferences quite dramatically. This sample preparation technique requires a knowledge of the retention characteristics of the analyte of interest at both acidic and alkaline pH. A generic protocol has been worked out that allows a rapid determination of a suitable washing and elution procedure. The analyte of interest is adsorbed on 20 solid-phase extraction cartridges or on 20 wells in a 96-well extraction plate. The cartridges or wells are washed with a fixed volume of an extraction solvent with increasing concentration of organic solvent at acidic and basic pH. The organic concentration is varied in 10% increments from 0% to 90% methanol. For adjustment of the pH, acetic acid or formic acid are used at acidic pH, and ammonia at alkaline pH. Using short HPLC columns and rapid elution protocols (e.g. [5]), the entire procedure can be carried out in about 2 hours. If the complete elution protocol is executed, one obtains the elution characteristics of the analyte under both acidic and basic conditions. For analytes with either a clear acidic or basic character, it is sufficient to carry out the study at the pH at which the analyte is not charged. One then selects the highest organic solvent concentration at which the analyte is still completely retained on the sorbent at one pH value for the washing protocol, and changes the pH at this solvent composition to elute the analyte. This saves additional time in the evaluation protocol, but requires knowledge of the pKa of the analytes and a high confidence in the procedure.

References pp. 89-90

Chapter 3

80

,owo..G___,.G o%

High pH

5%

,,G 65%

~ooo/; MeOH

,,@

Fig. 3.3. Standardwash and elution protocol for a plasma sample. The analyte is verapamil. After the sample is loaded (step 1), the sample is washed at acidic (step 2) and basic pH (step 3) with a solvent containing 5% methanol. Then another wash is carried out at alkaline pH at a high concentration of organic solvent (step 4). The concentration is just below the concentration where a breakthrough of the analyte occurs. Then the pH is changed at the same organic concentration to elute verapamil at 65% methanol and acidic pH.

As an example of the implementation of the 2-D sample preparation method, a typical wash and elution protocol for a plasma sample is shown in Fig. 3.3. The analyte is verapamil, a basic compound. The pH of a plasma sample was adjusted to an acidic pH with an acid (phosphoric, acetic or formic acid) to break the possible interaction of the analyte with the plasma proteins. The sample was then adsorbed onto the solid-phase extraction cartridge or well. Then it was washed with 5% organic solvent, commonly methanol, first at acidic pH and then at basic pH. This removed all polar interferences, such as plasma proteins. Next, the organic concentration was increased at basic pH to 65% methanol, just below the concentration that would show elution of verapamil at basic pH. This washing step eluted all contaminants with a higher polarity than the analyte. Finally, the pH is changed to acidic pH at the same organic concentration, 65% methanol. This step elutes the analyte, while leaving more hydrophobic interferences on the SPE cartridge. A rather clean chromatogram is obtained from a plasma sample after this protocol. This represents a significant improvement in the background compared to the simple washing protocol used in the 1-D method (Fig. 3.4). In this comparison of a sample prepared by the 1-D method with a sample from the 2-D method described here, one can clearly see the reduction in interferences due to the use of the 2-D protocol. A range of different analytes has been determined by us in serum or plasma samples using this 2-D protocol: naltrexone, verapamil and norverapamil [3], doxepin and nordoxepin, amitriptyline and nortriptyline [5], imipramine, trimipramine, trazodone, and methadone [4]. We have also used the same protocol for the determination of amphetamine and methamphetamine, methadone and its metabolite EDDP, diphenhydramine, propranolol, metoprolol and oxprenolol in urine samples. Thus, the 2-D principle can be applied successfully to many analytes and to multiple sample matrices. For plasma samples, the 1-D protocol is always simpler, but if the removal of interferences is important, the 2-D protocol always gives superior results.

2

1

3

Sample

[J:, 0

2

2

1

\ 3

,

, 4

)(,

.

, 6

,

,

min.

,

, 8

,

$A

Sample ,B,,

anK

,

10

Fig. 3.4. Comparison of results from the I-D (top) and the 2-D method (bottom). The analyte is verapamil. Column: Symmetry Shield RP8 5 pm, 3.9 x 150 mm. Mobile Phase: 50 mM Phosphate pH 7 : acetonitrile : methanol, 41 : 37 :22. Temperature: 30°C. Flow Rate: 1.0 mLImin. Injection Volume: 40 pL. Detection: UV at 230 nm. Peak Identification: Peak 1: Norverapamil; Peak 2: Verapamil; Peak 3: Methoxyverapamil (I.S.)

82 3.4 C A T I O N - E X C H A N G E

Chapter 3 SOLID PHASE EXTRACTION

The partial sulfonation of Oasis creates a packing with both hydrophobic character and cation-exchange character. Negatively charged and neutral compounds are retained via hydrophobic properties alone, positively charged analytes are retained by both hydrophobic interaction and ion-exchange interaction [7,28]. The packing is used predominantly for the clean-up of samples containing positively charged analytes, such as pharmaceuticals with amino groups. The sample is acidified to protonate the compounds of interest and loaded onto the cartridge. It is also possible - and sometimes preferable- to adsorb the analyte in its neutral form by hydrophobic interaction. Now all polar interferences can be removed under acidic conditions with a wash with dilute acid, maybe in the presence of a small amount of organic solvent. In the case that the sample was loaded under neutral conditions, the acidic wash protonates the analyte and locks it onto the ion-exchanger. Then all more hydrophobic contaminants that are either neutral or negatively charged are extracted using an organic solvent, potentially with some acetic or formic acid added to ensure continued retention of the positively charged analytes. Next, the pH is changed to alkaline using ammonia in methanol. The analytes are deprotonated and elute from the cartridge. If necessary, a more selective elution can be carried out by increasing the water content of the extraction solvent, thereby promoting hydrophobic retention of interferences with the same charge pattern as the analytes. While such a sample clean-up is highly effective, it requires a knowledge of the elution condition of the analyte as a function of the organic solvent composition. A set of experiments similar to the preparation for the 2-D protocol using the Oasis HLB sorbent is needed for carrying out such a selective procedure. Alternatively, a wash at alkaline pH conditions with a high water content can be used to remove polar interferences of the same charge as the analytes, followed by an elution of the compounds of interest at high pH and high organic concentration. As one can see, the combination of hydrophobic and ion-exchange properties of the packing gives a broader range of options for sample clean-up. Since both the concentration of the organic solvent and the pH can be used for clean-up, elution or fractionation, the mixed-mode ion-exchange packing provides at least as much flexibility as the 2-D technique described above for the Oasis HLB sorbent. In general, good results have been obtained with simple protocols, with both plasma samples and with urine samples. An example is the determination of codeine and its metabolite codeine-6-glucuronide in human urine and porcine plasma [7]. The same simple protocol was used in both cases: the acidified sample was loaded onto the SPE cartridge, first washed with 0.1 M hydrochloric acid, then washed with methanol. The analytes were eluted with 5% ammonium hydroxide and 95% methanol. The UV chromatogram of a urine sample is shown in Fig. 3.5. The published study [7] contains the clearance results for codeine and the metabolite after a single oral dose of codeine. Even simpler extraction protocols have been applied, when HPLC/MS/MS is used for the final analysis. Due to the high selectivity of the mass spectrometer, there is less concern about interferences, and the important element of sample preparation is the removal of plasma proteins from the sample to prevent clogging of the HPLC column, or contamination of the ion source of the mass spectrometer. A typical protocol

Techniques for sample preparation using solid-phase extraction

83

0.040

0.030

2 AU

3

0.020

1 0.010

le O.O00f ,M",

Blank | ' ~

' ~ I '

2.00

~

~ I ' s

~'

I

~

~'

~

I

~

'

L

'

I

' ~ ' ~ I ~ ' ~ ' I ~

6.00

L'

I ' ~

~

I

'

~

10.00

'

Fig. 3.5. Determination of codeine (peak 3) and its metabolite codeine-6-glucuronide (peak 2) in human urine. The internal standard (peak 1) is ranitidine. Column: Symmetry Cls, 3.5 I~m, 2.1 x 150 mm with a Symmetry C,s Sentry guard column, 2.1 x 10 mm. Mobile phase: 0.05% TFA : acetonitrile : methanol 90: 5:5. Flow rate: 0.3 mL/min. Temperature 30~ Detection: UV at 220 nm.

consisted of loading the sample under acidic conditions, a single wash with 2% hydrochloric acid to remove protein, and then elution of the analyte in methanol under basic conditions. The protocol was carried out with a range of analytes in rat plasma, oxybutynin, omeprazole, propranolol, salbutamol, chlorpromazine, ranitidine and trimethoprim, at plasma concentrations down to 1 ng/mL. The HPLC conditions varied, but the HPLC column was typically a 2.1 x 30 mm XTerra MS C~8 column. If the simple protocol described here results in ion suppression, it is rather straightforward to fall back to a more sophisticated wash protocol as described earlier. The dual retention mechanism of the Oasis MCX column also allows a fractionation of analytes. Neutral and acidic analytes are retained by hydrophobic interaction, and bases by ion-exchange. One can elute neutral and acidic analytes with methanol, and then basic analytes with methanol and 5% ammonium hydroxide in a second step. For example, acetaminophen and barbiturates would elute in the methanol fraction, while the base amphetamine would elute in the basic methanol fraction. An interesting highly selective elution scheme was developed by Iraneta [29]. The strength of the base used in the elution scheme determines which types of basic analytes

References pp. 89-90

Chapter 3

84

/

./V

a

V

I

,

0

,

,

*

,

,

,

,

,

l

~

,

~

,

,

,

,

l

l

,

2

,

V

,

,

,

4

,

,

~

l

,

,

,

,

,

,

,

,

,

6

Minutes

Fig. 3.6. Determination of amphetamine (1) and methamphetamine (2) in human urine. Internal standard: phentermine, peak 3. Top: blank without TEA wash; center: blank with TEA wash; bottom: spiked urine sample after TEA wash. Column: 3.5 Ixm SymmetryShield RPs, 4.6 x 75 mm with Sentry guard column. Temperature: 37~ Mobile Phase: 5/95 methanol/20 mM K2PO4, pH 7. Flow Rate: 2 mL/min. Injection volume: 10 IxL. Detection: UV at 214 nm. TM

can be extracted. Thus, one can separate tertiary and aromatic amines from primary and secondary amines using a 5% methanolic solution of triethylamine (TEA). The weaker amines are eluted, while the primary and secondary amines are retained. The stronger amines can be eluted subsequently with 5% ammonia in methanol. An application of this principle to biological samples is the determination of methamphetamine and amphetamine in urine, shown in Fig. 3.6. In this case, the selective TEA wash was used to remove interferences. The chromatogram at the bottom shows the determination of the analytes, peaks 1 and 2, and the internal standard phentermine vs. the blank urine sample in the center chromatogram. The top chromatogram shows a blank without the TEA wash. Multiple interferences, most likely weakly basic compounds, were eluted at the same time as the analytes of interest but were removed by the highly specific TEA wash.

3.5 ANION-EXCHANGE SOLID PHASE EXTRACTION The Oasis mixed-mode anion exchanger has both hydrophobic properties and anionexchange properties. This means that retention can be controlled in a similar way as described in the previous paragraph for the cation exchanger. Consequently the wash and elution protocols are similar to the protocols used for the cation exchanger, with a reversal of the pH. In the standard protocol, the sample is loaded onto the Oasis anion exchanger from an aqueous solution such as urine or plasma. Either an acid or a base

Techniques for sample preparation using solid-phase extraction

85

may be added to the sample to break any interaction of the analyte with proteins in a plasma sample. The drugs can be either ionized or neutral during the loading procedure. After loading, the Oasis cartridge or well is washed with an aqueous solution of 2% ammonia. This step is designed to lock the analyte onto the ion-exchange portion of the packing. Now the adsorbed sample is washed with methanol containing 2% ammonia, or simply methanol with elution accomplished by washing the cartridge with methanol acidified with formic acid. As is true for the Oasis MCX methods, an additional simplification can be made if the final method is executed using an LC/MS system. Diclofenac, chlorsoxazone, and warfarin have been determined in rat plasma with a single wash of the sorbed analyte with water and 2% ammonia, followed by an elution with methanol or methanolacetonitrile 1:1, both acidified with 5% formic acid. The lowest concentrations determined were between 5 and 10 ng/mL of plasma sample.

3.6 ON-LINE METHODS

The methods described above as sample preparation methods prior to LC analysis can be used also for the direct analysis of samples, if a sufficiently specific detector such as a fluorescence detector or a mass spectrometer is employed. Several scenarios can be applied to such analyses. In the simplest form, a short column, e.g. a 20 mm • 2.1 mm 25 ~m Oasis HLB extraction column, is used alone [5,6,30]. In more sophisticated schemes [6,31 ], the sample preparation column is combined with an analytical column. To reduce analysis time, a combination of two sample preparation columns can be combined with two analytical columns. While the analytical separation is carried out with one combination, the second Oasis column performs the sample preparation procedure while the second analytical column is reequilibrated with the starting mobile phase of the analytical gradient. Other, more elaborate and faster or more sensitive techniques have been developed as well [8]. Another approach designed for ultra-rapid sample throughput combines multiple chromatography streams with a single mass spectrometer [32], with a sample throughput exceeding 100 samples per hour. The simplest procedure is the use of the Oasis HLB extraction column alone. In general, the sample is loaded onto a short 2 cm column packed with 25 ~m Oasis HLB packing and washed free of polar interferences and proteins. Then, a rapid gradient is executed, and the effluent is directed towards the mass spectrometer for detection and quantitation of the analyte. Protocols of this type [30] can be carried out in a fashion that allows a cycle time of just over one minute per sample. An example is a protocol developed for the rapid analysis of drugs in rat plasma [6]. The instrument used was a Waters Alliance 2790 HPLC system supported with a Waters 515 HPLC pump and a Rheodyne LabPro 10-port switching valve, and a Micromass Ultima Triple Quadrupole mass spectrometer. The set-up uses two 2.1 mm • 20 mm 25 ~m Oasis HLB columns operated in parallel. While the sample pretreatment and washing protocol is executed on one Oasis column, the elution of the analyte into the mass spectrometer is carried out using the other. The entire process works as follows: the rat plasma is centrifuged and either acidified with phosphoric acid (20 ~L per mL

References pp. 89-90

Chapter 3

86

of plasma) or treated with ammonium hydroxide (same concentration), depending on whether or not the drug is acidic or basic. The treatment with acid or base disrupts the interaction of the drug with plasma proteins and ensures good retention of the drug on the Oasis HLB column during the sample loading step. Then 500 lxL of the pretreated rat plasma are diluted with 400 txL of internal standard dissolved in water and 200 IxL of this solution are loaded (at 4 mL/min) onto the Oasis HLB column. The loaded sample is washed free of proteinaceous interferences for 1 minute at this high flow rate. Then the valve is switched, and a 1-minute gradient from 5% to 95% acetonitrile in the presence of 0.5% formic acid at 0.4 mL/min is used to elute the analyte and internal standard into the mass spectrometer. While the gradient is executed on one Oasis column, the second Oasis column is loaded with the next sample and washed free of protein. A high flow rate is used during the wash cycle, allowing for a rapid flushing of the column and elution of the interferences such as proteins and polar compounds. During the elution cycle, the flow rate is reduced to a value that is compatible with the mass spectrometer. This permits a complete elution of the entire analyte into the mass spectrometer, and therefore a high sensitivity. Stream splitting before the mass spectrometer, which is effectively a waste of analyte, is avoided. Good results together with good column life were obtained for a range of samples. For example, diflunisal was analysed in rat plasma using this protocol down to concentrations of 5 ng/mL of plasma, with a standard deviation of as little as 1.5% at this concentration (Table 3.3). The same protocol was used for clemastine (Table 3.3) down to a plasma concentration of 1 ng/mL with a relative standard deviation of 2.1%. The calibration curve of clemastine is shown in Fig. 3.7. We have found that the lifetime of the Oasis sample preparation column is in excess of 250 plasma samples in these protocols. This result is based on a 200 IxL injection of a plasma sample diluted 1:1 with water. If a protein precipitation is used before the injection of the sample, the column lifetime exceeds 400 injections. This set-up has been combined with the use of an analytical column during the elution step. Good results have been obtained with and without analytical column. Thus,

TABLE 3.3 CALIBRATION CURVES FOR THE DIRECT DETERMINATION OF CLEMASTINEAND DIFLUNISAL IN RAT PLASMA (n = 6) Clemastine Concentration [ng/mL] 1.0 2.5 5.0 10.0 100 200 250

Diflunisal

Response [ng/mL]

%RSD

0.98 2.56 5.25 9.50 101.4 201.1 247.1

2.1 4.4 2.3 2.7 2.6 1.6 0.6

Concentration [ng/mL] 5.0 20.0 50.0 200 250 500 1000

Response [ng/mL]

%RSD

5.1 19.3 51.1 201.5 243.8 496.5 1005

1.5 4.6 2.8 1.5 3.3 2.8 1.1

Techniques for sample preparation using solid-phase extraction 18.5

CalibrationC~

87

curve:~ Determination: 0.999397

//~:

O

Oil ~ ....

0.0

I ....

I ....

25.0

I ....

I ....

50.0

I ....

I ....

75.0

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I'''~1

100.0 125.0 150.0 175.0 200.0 225.0 250.0 Concentration (ng/nl~,)

Fig. 3.7. Calibrationcurve of clemastine for the on-line analysis of plasma samples

Fig. 3.8 shows the chromatograms obtained for clemastine in rat plasma in an arrangement that combines two Oasis sample preparation columns with one XTerra analytical column. The concentrations are 2.5 ng/mL and 250 ng/mL. Similar results have been obtained with diflunisal. The combination of the analytical column with two Oasis sample preparation columns allows an efficient use of all devices. Three switching valves are required to execute this protocol [6]. The sample is loaded onto one of the two precolumns and washed free of polar interferences such as plasma proteins, while a gradient elutes the previously cleaned sample from the second precolumn onto the analytical column and into the mass spectrometer. The analytical column used, an XTerra MS C,8 column, is compatible with alkaline pH. Therefore, the wash protocol or the elution protocol can be carried out either with formic acid or ammonia as additives for the control of the pH. Typically both additives are used at a concentration of 0.5%. It is also possible to accomplish the wash at one pH and the elution at another pH [8]. The combination of two precolumns and an analytical column improves the speed of the procedure. While the total cycle time is 3 minutes for each sample, the combination of two precolumns and one analytical column reduces the analysis time down to 2 minutes per sample. However, the benefit of an additional analytical column is arguable. Off-line sample preparation protocols have traditionally been used to clean a sample before the final analysis via an HPLC method. In the on-line analysis, a key element is the use of the

References pp. 89-90

Chapter 3

88 10

2

Clemastine 250 ng/mL "'f"

'

"

'

~t

,I

'~"l

....

I ........

II

'1

....

I"

'1'

"t'l"l

'l

't

. . . . . . . .

I

"r

....

r ....

10

J ....

l''"l

I'

~1'

"

'"

r'"FI

....

'

['"'l'J~rr

I

I

2.25

S

"'f'

'

"

P'

'l

....

l'"~t

~''

'*'r'

'1

"I

l'

'l

"t'

"l

'1

"J

"

I''"1

'r""r

'"I""f'"'F'

10

'1

'"

2.44 C]emastine

2.5 n g / m L '"1

....

I ....

I

P....

I ....

r ....

f ....

I ....

I ....

I ....

I ....

I ....

] ....

t ....

l'

'1 ....

i'-v~T

TT'ITV"~i~VZ'I'~''~'/'-'''/

[ ....

I ....

I ....

I '"

1

IS

0"'1

....

I ....

0.20

r ....

t ....

0.40

l''"r

....

0.60

I ....

! ....

0.80

I"~"l

....

1.00

I ....

I""t

1.20

....

I ....

1.40

t ....

I '''~

1.60

....

r''t

1.80

F .... 1 .... i'' i "l""l"r'['""r ;~, "'!'"' ,~, Time 2.00 2.20 2.40 2.60 2.80

....

Fig. 3.8. Determination of clemastine in rat plasma. Top 2 chromatograms: clemastine at 250 ng/mL and internal standard. Bottom 2 chromatograms: clemastine at 2.5 ng/mL and internal standard. Clemastine signal: MRM of 2 channels ES +, 344.1 > 214.9; internal standard: MRM of 2 channels ES +, 340 > 202. mass spectrometer as a highly specific detector. Less demands are placed on the separation system. Nevertheless, a few minimal requirements must be met. MS detection can suffer from ion suppression [33,34], caused by interferences in the sample that coelute with the analyte. However, tests can be designed that quantify, if there is a concern for ion suppression [35]. To carry out such a test, the normal on-line sample preparation protocol is executed, using a blank plasma sample without analyte. The column effluent is mixed just before the mass spectrometer with a stream that carries a known amount of analyte, and the MS system monitors the stream at the mass/charge ratio used to quantify the analyte. If the signal declines, ion suppression is present. If the suppression occurs outside the elution window of the analyte in the standard protocol, the analysis can be executed without concern. If indeed ion suppression interferes with the quantification, the elution protocol may need to be modified. In our case, a change of the pH of the elution gradient is often a fast solution. Such a change is enabled through the use of Oasis columns, which are stable over the entire pH range. While the MS sensitivity for analyte and interferences may depend on the pH as well, the drastic shift in the elution pattern may move the remaining plasma interferences away from the elution window of the analyte. An alternative solution might be the ionization mode of the MS system. Generally, the APCI mode is less prone to ion suppression than the ESI

Techniques for sample preparation using solid-phase extraction

89

mode. A third possibility is the use of a standard analytical column that modifies the elution pattern of analyte and interferences. This is possible only, if the interaction between the analyte and the reversed-phase analytical column is as least as strong or stronger than the interaction between the analyte and the precolumn. Fortunately, it is possible to modify the interaction of the analyte with the analytical column by either changing the pH or adding water as a diluent between the pre-column and the analytical column. In both of these cases, the sample is retained and focused again at the top of the analytical column. The subsequent elution from the analytical column is effectively a second dimension that is useful for the removal of interferences. One should also compare the ion suppression of the analyte and the internal standard. If one is using a closely related compound as the internal standard, such as a deuterated form of the analyte, the LC retention pattern can be identical. Under these circumstances, the ion suppression may also be identical, and a change in the elution pattern away from interferences that cause the ion suppression may not be necessary. It is always worthwhile to determine the ion suppression for the analyte as well as for the internal standard, but in the case of identical ion suppression a correction of the LC retention pattern may be avoidable. However, to maximize sensitivity it is always better to have a clean separation between the analyte and the background. The parallel on-line pre-cleaning of samples together with a direct LC/MS analysis shows high promise. Alternative sample pretreatment schemes have been explored recently [8] that combine Oasis mixed-mode ion-exchange columns with an XTerra analytical column. Very low background and therefore very high sensitivity have been achieved with a more sophisticated procedure. We anticipate also that the combination of multiple sample preparation and analytical columns with a single mass spectrometer beyond the set-up just described will result in an additional shortening of the analysis time per sample. Our current efforts aim into this direction.

3.7 C O N C L U S I O N S Here a range of useful methods for the sample preparation of samples of biological origin (plasma and urine) prior to HPLC or HPLC/MS/MS analysis have been described with the focus on the general principles of each method. Each method has been carried out with multiple analytes and can be adapted to related sample preparation problems without difficulties. However, occasionally, the particular properties of a sample matrix, an analytical technique or even simply the analytes require departures from the details of the approaches described here [36,37]. Both off-line and on-line SPE techniques have advantages depending upon the particular application and we anticipate that additional progress will be made in the future with respect to the speed and efficiency of both approaches.

3.8 REFERENCES 1 2

Y.-ECheng, D.J. Phillips and U. Neue, Chromatographia, 44 (1997) 187.. Y.-ECheng, D.J. Phillips, U.D. Neue and L. Bean, J. Liq. Chrom. & Rel. Technol., 20 (1997) 2461.

90

Chapter 3

3 4 5 6 7

Y.-E Cheng, U.D. Neue and L. Bean, J. Chromatogr., A, 828 (1998) 273. Y.-E Cheng, U.D. Neue and L.L. Woods, J. Chromatogr. B, 729 (1999) 19. J. Ding and U.D. Neue, Rapid Commun. Mass Spectrom., 13 (1999) 2151. C.R. Mallet, J.R. Mazzeo and U.D. Neue, Rapid Commun. Mass Spectrom., 15 (2001) 1075. Y.-E Cheng, U.D. Neue, R. Bonin, E. Block and L. Bean, J. Liq. Chrom. & Rel. Tech., 24(2001) 1353. C.R. Mallet, Z. Lu, J. Mazzeo and U. Neue, Rapid Commun. Mass Spectrom., 16 (2002) 805. E.S.E Bouvier, D.M. Martin, RC. Iraneta, M Capparella, Y.-E Cheng and D.J. Phillips, LC-GC. 15 (1997) 152. E.S.E Bouvier, P.C. Iraneta, U.D. Neue, ED. McDonald, D.J. Phillips, M. Capparella and Y.-E Cheng, in: Current Trends and Developments in Sample Preparation, LC-GC, (1998) $53. E Martin and I.D. Wilson, J. Pharm. Biomed. Anal., 17 (1998) 1093. U.D. Neue, D.J. Phillips, T.H. Walter, M. Capparella, B. Alden and R.E Fisk, LC-GC, 12 (1994) 468. J. O'Gara, B. Alden, T.H. Walter, C. Niederl~inder and U.D. Neue, Anal., Chem., 67 (1995) 3809. J.E. O'Gara, D.E Walsh, B.A. Alden, E Casellini and T.H. Walter, Anal. Chem., 71 (1999) 2992. U.D. Neue, Y.-E Cheng, Z. Lu, B.A. Alden, EC. Iraneta, C.H. Phoebe and K. Tran, Chromatographia, 54 (2001) 169. J.E. O'Gara, D.E Walsh, C.H. Phoebe, B.A. Alden, E.S. P Bouvier, E C. Iraneta, M. Capparella and T.H. Walter, LC-GC, 19 (2001) 632. Y.-E Cheng, T.H. Walter, Z. Lu, E C. Iraneta, B.A. Alden, C. Gendreau, U.D. Neue, J.M. Grassi, J.L. Carmody, J.E. O'Gara and R. P Fisk, LC-GC, 18 (2000), 1162. Y.-E Cheng, Z. E1 Fallah, U.D. Neue and D.J. Phillips, Waters Colunm VI, (1997) 10. D.J. Phillips, L.J. Bean, E.S.E Bouvier, M. Capparella, Y.-E Cheng, P.C. Iraneta, U.D. Neue and ED. McDonald, Fast and Easy Solid-phase Extraction with a New Reversed-Phase Sorbent, Waters Column International VI (1997) 1. E.J. Woolf and B.K. Matuszewski, J. Chromatogr., A, 828 (1998) 229. M.A. Zemaitis and P.D.Kroboth, J. Chromatogr., B, 716 (1998) 19. K.M. van Rij, D. Compas, E.L. Swart, EN. EC. de Goede and D.J. Youw, Ther. Drug Monit., 21 (1999) 416. K.A. Goerka, V.E Samanidou and I.N. Papadoyannis, J. Liq. Chrom. & Rel. Tech. 22 (1999) 2975. M. Mabuchi, Y. Kano, T. Fukuyama and T. Kondo, J. Chromatogr. B, 734 (1999) 145. L.K. SCrensen, L.K. Snor, T. Elkaer and H. Hansen, J. Chromatogr. B, 734 (1999) 307. J.T. Wu, H. Zeng, Y. Deng and S.E. Unger, Rapid Commun. Mass Spec., 15(2001) 1113. R.E. Aarnoutse, C.E W.G.M. Verweij - van Wissen, W.J.M. Underberg, J. Kleinnijenhuis, Y.A. Hekster and D.M. Burge, J. Chromatogr. B, 764 (2001) 363. K. Richter and R. Oertel, J. Chromatogr. B, 724 (1999) 109. EC. Iraneta, E.S.E Bouvier, D.J. Laviolette, E.K. Brown, D.J. Phillips, J.J. Lee, Y.-E Cheng, R.A. Collamati and D.E Walsh, paper no. 384 at PittCon '99, Orlando J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallet and R.S. Plumb, J. Chromatogr. A, 828 (1998) 199. M. Jemal, Y.Q. Xia and D.B. Whigan, Rapid Commun. Mass Spectrom., 12 (1998) 1389. M.K. Bayliss, D. Little, D.N. Mallet and R.S. Plumb, Rapid Commun. Mass Spec., 14 (2000) 2039. J.M. Poirier, N Radembino and E Jaillon, Ther. Drug Monit., 21 (1999) 120 R. King, R. Bonfiglio, C. Fernandez-Metzler, C. Miller-Stein and T. Olah, J. Am. Soc. Mass Spectrom., 11 (2000) 942. C. Miller-Stein, R. Bonfiglio, T.V. Olah and R.C. King, Am. Pharm. Rev., 3 (2000) 54. R. Ricciarello, S. Pichini, R. Pacifici, I. Altieri, M. Pellegrini, A. Fattorossi and E Zuccaro, J. Chromatogr. B, 707 (1998) 219. J.M. Poirier, N Radembino, E Robidou and P. Jaillon, Ther. Drug Monit., 22 (2000) 465.

8 9 10 11 12 13 14 15 16 17

18 19

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 O 2003 Elsevier Science B.V. All rights reserved

91

CHAPTER 4

Turbulent flow chromatography in bioanalysis Tony Edge Cohesive Technologies, 17 Cochran Close, Crownhill, Milton Keynes, MK8 0AJ, U.K.

4.1 I N T R O D U C T I O N Sample analysis has often been limited by the incompatibility of the initial sample to the analytical technique being performed. This is particularly the case with chromatography, where lengthy sample preparation procedures are routinely used, to ensure the integrity of the analysis system. This can be either due to potential impurities within the sample matrix, which can influence the final detection of the compounds of interest, or due to the stability of the chromatographic media, which can be compromised by the introduction of the original sample. Within the field of bioanalysis, such an occurrence is observed, where the analyte under investigation is very different in physical and chemical properties to the majority of the matrix. However in this scenario the matrix, which comprises of water, ionic salts and an assortment of proteins, can also interact with the stationary phase causing premature failure of the column. This can either be in the form of a gradual deterioration in the chromatographic performance of the system or physical failure of the column due to pores and interstitial spaces within the packing material becoming blocked by the adsorption of large protein molecules, thus reducing the porosity of the stationary phase. Fouling of the chromatographic media can also result in the sensitivity of the detection system being compromised due to a gradual build up of matrix material in the detection cell. The introduction of mass spectrometry as a universal and specific detector has highlighted this issue. The use of turbulent flow chromatography allows direct injection of biological fluids onto a standard LC system [1], without a significant loss in chromatographic performance, over many injections, typically approaching one thousand. This is at least an order of magnitude better than conventional HPLC columns, used in a standard chromatographic environment. The difference in performance is associated with the fluid dynamics of the mobile phase, and the subsequent interaction with the stationary phase. To obtain a better understanding of the phenomena it is important to understand some of the theory associated with traditional laminar flow HPLC. References pp. 127-128

92

Chapter 4

4.2 BACKGROUND Early attempts to decrease the analysis time in chromatography have been thwarted by the existence of an apparent optimum mobile phase velocity, and consequently an apparent fixed analysis time [2-6]. Increasing the mobile phase velocity above this optimum reduces the performance of the chromatographic system, which is not always acceptable [2]. Experimental evidence and a sound theoretical understanding of the band broadening processes occurring within a chromatographic system have meant that these ideas were never seriously challenged. This left chromatographers with a very limited range of flow rates that could be utilized around an optimum linear flow velocity. To better understand why there existed an apparent optimum flow rate to perform a separation within a chromatographic environment, it is necessary to identify the varying band broadening processes that can occur during a chromatographic separation. As early as 1956 van Deemter 2 identified the three major contributing factors, which resulted in band broadening in chromatography. These factors, which resulted in poorer resolution of components due to broader peak shapes, were encapsulated within the Van Deemter equation.

4.2.1 Band broadening processes The ideal chromatographic process is one in which the components of a mixture form discrete narrow bands as they move down the column. The narrowness of the band or peak, relative to the time spent in the column, is a measure of the efficiency of the process whilst the resolution is assessed by the ability to separate individual components of the original sample. This value can be quantified by a parameter referred to as the plate number or number of theoretical plates, N. If the resulting peak is Gaussian in form, then the following equation is a good approximation of the efficiency of the column.

N=5.54 ~ tr t 2

(1)

where

t r - retention time of compound. W l / 2 - peak width measured at half the peak height. N - number of theoretical plates.

Any skewness, tailing or fronting, in the peak, generally caused by secondary interactions with the stationary phase, will to a certain extent make interpretation of the efficiency of the column difficult.

Turbulent flow chromatography in bioanalysis

93

However the dimensionality of N means that it is difficult to compare columns directly using this term. A more effective method of comparing the efficiency of columns is to use a dimensionless form, h - the reduced plate height, given by: L h =--

(2)

where: L - column length d p - particle diameter N - number of theoretical plates A plot of the reduced plate height vs. the flow rate produces a parabola with a minimum at the best smallest reduced plate height, which equates to the most efficient separation, Fig. 4.1. A minimum reduced plate height of 2 is generally accepted as being very good, with values up to 10 being acceptable, for a standard HPLC column. Better efficiencies can be achieved using different chromatographic techniques but they rely on different methods of driving the mobile phase through the column.

4.2.2 Theoretical interpretation The van Deemter equation is a theoretical treatment of the peak broadening within a chromatographic column. The equation, which describes the band broadening processes, is given by equation (3).

B

h =A + - + Cv

(3)

V

Van Deemter plot

.r z~O) q) J= Q) Q. "0 4) (.1

Reduced velocity,v Fig. 4.1. Van Deemter plot showing optimum reduced velocity, giving minimal band broadening which occurs at lower reduced velocities. References pp. 127-128

94

Chapter 4

where h - reduced plate height, a dimensionless measure of the band broadening. A, B, C are constants. v - reduced velocity, given by: . dp v - - - ~Lmp Om

where: L Ix,,,p= - average mobile phase velocity t0 t0- time for unretained totally permeating peak. D m - diffusion coefficient of analyte in the mobile phase. This equation is clearly dependent on the reduced velocity through the column with an optimum value being obtained when: dh

--=O-Cdv

B

--

(4)

V2

which on rearrangement and assigning suitable boundary conditions produces the following equation [7]:

(5) This assumes that the three constants are independent of the reduced velocity. There have been numerous investigations into the band broadening processes involved within a chromatographic system. Although many authors have advanced the understanding beyond that of the original work of van Deemter, this does provide an ideal starting point to further the understanding of how the hydrodynamics of the mobile phase affect the performance of the chromatographic system.

4.2.3 Description of the van Deemter constants 4.2.3.1 A term

The A term is generally referred to as the tortuosity factor and relates to the various paths that are available through a packed bed [8,9]. It accounts for the variable, unequal pathways, around the stationary phase particles or support material. Fig. 4.2 clearly shows this concept. This is greatly affected by the particle size with smaller more uniform particles producing less band broadening. The A term can be reduced by better packing processes and by using smaller uniform particles, as these give less variation in

%

Number

Of nolecules

\

Residence time in c o l u m n Fig. 4.2. Schematic representation of A term in van Deemter equation, showing two of the many different pathways through a HPLC column.

96

Chapter 4

the pathlength. Because of this factor many HPLC column manufacturers have concentrated their efforts in producing packed columns with smaller particle size, and more uniform shape and size distribution throughout the column. This approach has greatly improved the performance of columns, with respect to their ability to resolve chemically similar components. However, one consequence of reducing the particle size is that the pressure drop across the column per unit length is also increased; hence high performance liquid chromatography has also been referred to as high pressure liquid chromatography. This is not beneficial for the generation of turbulence as will be seen later. 4.2.3.2 B term

The B term relates to the diffusion of the molecules in a longitudinal direction and describes the band broadening processes due to the random motion of the solute molecules in the mobile phase as they travel through the column [ 10]. The effect of the B term is flow dependent and also dependent on the nature of the mobile phase and the analyte under study. At lower flow rates the effect of longitudinal diffusion is more pronounced, resulting in a greater band broadening of the peak. This effect has a greater significance in gas chromatography since the diffusion coefficients in liquid chromatography are an order of magnitude smaller than in gas chromatography. Since this term is dominant at low flow rates it is generally not considered a limitation within liquid chromatography, as most analysts would like shorter analysis times.

t2

t3

t4

Fig. 4.3. Schematic representation of B term in van Deemter equation, showing band broadening due to diffusional processes, h-4 represent different snapshots in time as the initial band proceeds down the open tube.

4.2.3.3 C term

The C term relates to the mass transfer of the analyte through the mobile phase and to the stationary phase, and any subsequent diffusion on the surface. Much theoretical work has been performed on the exact determination of this phenomenon [ 11-14]. In the original text, van Deemter proposed that this was a single term related solely to the mass transfer to the stationary phase. Subsequent investigations have shown that this term can

Turbulent flow chromatography in bioanalysis

ii

i

f

97

Stationaryphase rticle

Fig. 4.4. Schematicrepresentation of the C term in the van Deemter equation. Under laminar flow in an open tubular system, flow at the sides of the tube is slower due to greater frictional forces, resulting in a parabolic flow profile and hence band broadening.

be split into two distinct processes, mass transfer through the mobile phase and mass transfer from the mobile phase to the stationary phase. Under laminar flow conditions, diffusional processes dominate the radial mass transfer through the mobile phase, since there are no other transport mechanisms available. Unless there is a dynamic equilibrium between the radial movement of fluid and the movement of fluid longitudinally through the column, a parabolic flow profile for an open tubular system, given in Fig. 4.4, will exist. Thus, fluid at the surface of the open tubular system will be effectively stationary, with the velocity of the fluid increasing as the central flow path is approached, due to the drag effect of the many parallel fluid paths. Since diffusion is a slow process, increasing the fluid flow will exaggerate any discrepancies in the flow profile. This results, in a chromatographic environment, in peak tailing at higher laminar flow rates. Thus, although the components will elute quicker at higher flow rates, because of the hydrodynamics, peak tailing will result. Incorporation of obstacles in the fluid path does affect the dispersion of mass by creation of a greater number of fluid paths through the system, but movement between these individual fluid paths is still limited through diffusion. The analyte will be better dispersed through a packed bed than through an open tubular system under the same flow conditions, but at high laminar flow rates this dispersion of analyte molecules will not be as effective as in turbulent conditions. The transport of the analyte molecule through the mobile phase and to the surface of the stationary phase, and ensuring that this process is homogeneous is fundamental to improving the chromatographic process. If this is not achieved then discrepancies will result in residence times for identical chemical components due solely to the fluid flow.

4.2.4 Development of turbulent flow chromatography model To understand the basic concepts of turbulent flow chromatography, an open tubular system will be investigated and then the concepts transferred to a packed bed system. Golay's equations [15] define implicitly the behaviour of an open tubular chromatographic system under laminar flow conditions.

References pp. 127-128

Chapter 4

98

Under these conditions a parabolic radial velocity profile can be assumed. Diffusional processes dominate mass transfer within the mobile phase. These fundamental concepts form the basis of Golay's treatment of band broadening within an open tubular system. However under turbulent conditions the radial velocity profile is no longer parabolic but becomes velocity dependent [ 16], and radial transfer is dependent on convection due to the spontaneous formation of eddies, although diffusional processes still do occur within this environment. There have been several studies of band dispersion in turbulent flow [17-19]. Early work indicated how turbulence affected the mass transfer [20] and although computational models have developed substantially since these early days, they are still applicable and give a good understanding of how mass transfer effects are substantially reduced by moving to a turbulent flow regime. The following treatment takes Golay's equations for dispersion within a chromatographic system and examines the effects of the incorporation of a turbulent term to improve mass transfer. Golay's general equation, which is used for dispersion in an open tubular system (e.g. capillary GC) is given by: B

h=-+[Cm+Cs]v

(6)

P

where: h is the reduced plate height, a measure of the spreading of the analyte component in the column. B is a diffusionally related constant. Cm- is a mass transfer constant relating to effects occurring in the mobile phase. C~ - is a mass transfer constant relating to the effects occurring on or near the stationary phase. There has been much discussion relating to the exact nature of these terms [14-17], although it is generally agreed the above forms a reasonable representation of events in a laminar and in a turbulent flow open tubular system [21 ]. However, for development of the discussion it is necessary to expand the two mass transfer terms to incorporate turbulent effects.

Cm=

2[(11, - 21,2 + 1,3) + 2(I,, - Ii2)k q- I, lk2] ( 1 + k2)

2k 2s C~ = ~3(1 + k) d/ 9

Ds~;,

(7)

(8)

where:

I~j, Ii2, 1i3 are integrals relating to the parabolic radial flow profile of the mobile phase. k - capacity factor, a dimensionless retention time.

Turbulentflow chromatography in bioanalysis

99

Their exact forms are:

I~, =

112

--

2r~(r)

fro [fro r, r r] _

_ _ _

2r~,(r)

Ij 3

r2

r~

I rO r 3 . dr

j0

2r4t~(r)

where: t~(r) is a function that determines the dispersion coefficient for a particular radial position relative to a nominal value. For this analysis, dispersion includes all forms of mass transfer, either due to turbulence (eddy formation) or due to molecular diffusion. ~(r) is a function that determines the velocity for a particular radial position relative to the mean velocity. These equations are equally valid for turbulence however, unlike a laminar flow system where it is relatively easy to define an exact equation for the parabolic flow profile, indeed the integrals are a well-known solution [15], in a turbulent system this is more difficult to do. Herein lies one of the foundations to the understanding of turbulent flow chromatography. As the flow rate increases, so the flow profile changes. The flow profile describes variation in the dispersion coefficients of the analyte radially, it also describes the variation in velocity profile radially. In a turbulent flow system the velocity profile is much flatter, as is the dispersion variation. This results in much more even mass transfer within the mobile phase, which reduces band broadening within the mobile phase. It does not affect the mechanism of band dispersion due to stationary phase effects, and so for turbulent flow chromatography to be effective, stationary phase interactions must be kept small [22]. This will be discussed later with reference to the most common application area of turbulent flow chromatography, namely bioanalysis.

4.2.4.1 Definition of turbulent flow The concept of turbulence has been introduced but has not been defined. The characterisation of fluid flow, and mass transfer within it, is something that has been extensively studied [23-27]. A dimensionless parameter that is often quoted to characterise the flow is the Reynolds number [28]. This is defined as the ratio of the inertial to viscous forces present in a fluid system. Dependent on the system being investigated slightly different mathematical definitions are used to express this non-

References pp. 127-128

0 0

..........

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

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

'

(a)

,. . . . . . . . . . . . . . . . . . . . . . . . . - .... 2._IL .

,,,,,,,,,

.21

(b~

Fig. 4.5. (a) L a m i n a r flow w i t h poor m a s s distribution across the channels. (b) Turbulent flow with i m p r o v e d m a s s distribution across c h a n n e l s due to formation o f eddies.

Turbulent flow chromatography in bioanalysis

101

dimensionless term. This is primarily due to the need to identify an appropriate length scale by which to characterise the flow. Thus the definition of the Reynolds number in an open tube is different to that obtained for a fluid flowing in a packed bed. In an open tubular system the Reynolds number is defined as follows: jxo . l Re = - -

(9)

where: l - is a characteristic length scale, which for an open tubular system is the diameter of the pipe. Ix0- is the mean linear velocity of the fluid through the column, superficial velocity xl - is the dynamic viscosity of the mobile phase. In a packed bed, i.e. a HPLC column, the characteristic length scale used is related to the particle diameter and the external porosity of the column, whereas the characteristic length scale used in an open tubular system is the tube diameter. For an open tubular system, it is generally agreed that the onset of turbulence occurs at approximately a critical Reynolds number of 2000, whereas in a packed bed of uniform spheres it is generally agreed that the critical Reynolds value above which turbulence occurs is between 1 and 10 [29]. The lack of certainty associated with the onset of turbulence within a packed bed is due to the nature of the particles. Completely smooth spherical particles will have less resistance to the flow than irregular particles. Equation (10) gives the modified Reynolds number, Re', and incorporates the particle diameter and an external porosity term. Re'=

txodp

(10)

qq(1 - ~0) where:

dp- average particle diameter. e0 - external porosity of column Ix0- superficial velocity Turbulence or inertial flow occurs in many kinds of fluid flows and is used in many processes for positive purposes [30]. Laminar flow, although more deterministic, has the disadvantage of poor mass transfer, due to the lack of convective motion. In a chromatographic environment the limitation of poor mass transfer through the mobile phase is overcome by reducing the interstitial fraction, so there is less fluid for molecules to diffuse through between the particles. Also the flow rate is chosen to give an optimal value in accordance with the observations of van Deemter. There are however practical issues with high inertial flows. Primarily within a standard HPLC column the backpressures required to obtain an inertial flow and thus generate non-deterministic flows would be prohibitively high, this is particularly the case with the development of new stationary phases, which use smaller average particle diameters.

References pp. 127-128

102

Chapter 4

4.2.4.2 Definition of turbulence Turbulence is an eddying motion that exists at high Reynolds numbers. Turbulence has a wide spectrum of eddy sizes with a corresponding spectrum of fluctuation frequencies. Turbulence has prevailing rotational motion that can be thought of as a tangle of vortex elements with highly unsteady vorticity vectors that are aligned in all directions. The largest eddies have sizes on the same order of magnitude as the flow domain, have low frequencies, and are effected by the boundaries and the mean flow. The smallest eddies, on the other hand, are determined by the viscosity of the fluid and have high frequency fluctuations. As the Reynolds number of a given flow increases, the width of the spectrum, or the difference between the largest and smallest eddies, increases. The large eddies extract kinetic energy from the mean motion and feed it to the largescale turbulent motion. The eddies may be considered as vortex elements that stretch each other. Due to this vortex stretching, energy is passed down the cascade to smaller and smaller eddies until viscosity causes the dissipation of the eddies. The rate of energy dissipated is determined by the large-scale motion, although dissipation occurs at the smallest scales [31 ]. It is important to note that viscosity does not determine the amount of dissipated energy, but only the scale at which dissipation occurs. Since the size of the large eddies is on the order of the flow domain, their motion strongly depends on the boundary conditions of a problem. The preferred flow direction of the mean flow is imposed on the large-scale turbulent motion, which makes the flow strongly anisotropic. With the cascading process, the direction sensitivity of the flow is diminished, and at high Reynolds numbers, the small-scale dissipative motion is isotropic. This allows for movement of the fluid across the column ensuring that there is good mass distribution. Within a packed bed the bulk flow is still anisotropic, although the introduction of spherical particulates will result in some transfer of individual fluid streams as they are split and recombined. However, the prevalent energy scale dominates radial flow within the fluid streams. In viscous dominated flows, diffusional processes dominate the flow mass transfer. As the inertial forces increase above a critical modified Reynolds number, eddy formation results within the individual fluid channels, resulting in greater dissipation of mass and energy throughout the fluid system.

4.2.5 Overcoming the problem of pressure drop Initial studies into the use of turbulent flow chromatography highlighted several practical implications of using a turbulent mobile phase [14,20]. In particular solvent consumption and pressure drop across the column were noted as major hurdles to the uptake of the technology. This was primarily as a result of the development of high performance liquid chromatography, which has concentrated on improving the chromatographic resolution by combining the effects of reducing the particle size, and designing better stationary phases to reduce secondary interactions with the analyte under investigation. The use of smaller particle sizes is prohibitive for the use of high flow rates, required to generate turbulence. However, increasing the size of the particles, and hence going

Turbulent flow chromatography in bioanalysis

103

against the industrial trend, results in both a reduction in the pressure drop across the column and also results in an increase in the inertial forces within a column, resulting in eddy formation and turbulence (equation (10)). However, as has already been noted it is difficult merely by application of the Reynolds number to characterise the flow to ensure that eddy formation is occurring within the column. However, there are other methods to identify the occurrence of turbulence within a fluid system, which rely on monitoring other properties of the fluid. These will be used to show that using a specific HPLC column turbulence can be generated and thus turbulent flow chromatography can become realised on a practical basis. The Ergun model [32] describes flow of a fluid through a porous packed bed relating the pressure drop to the superficial velocity of the fluid. The model identifies two components relating to two distinct flow regimes, namely laminar flow and turbulent flow. The full equation is given by: Viscous

Inertial

AP An(1 - a)21xo B(1 - a)pl.t 2 --= +

L

(11)

godp

where" AP - pressure drop across column L - length of column ~q - fluid viscosity a0 - external porosity of column !~0-superficial velocity d , - particle diameter p - density of mobile phase A , B - empirically derived constants

=:

E

_=

--------- Laminar part

0 0

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

0 i._ 0 (11

/"

~.............. ...... .....'. . . .

'..............

Turbulent part - Overall pressure drop t. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

o

"0

0L_ t~ i,,, n

Linear velocity

Fig. 4.6. Plot of linear velocity of fluid vs. resulting pressure drop across a packed column showing the transition between laminar and turbulent flow, Ergun equation.

References pp. 127-128

Chapter4

104

There are some assumptions in this model that relate to a uniform particle size and also to a spherical particle, although suitable scaling factors can be used to transform from different particle types. The two numerical constants are derived experimentally from the two regimes where the equation has been derived, viscous and inertial. The values obtained for B lie between 1.5 and 4 [33], and thus the value used by Ergun of 1.75 is an acceptable one. However, values quoted for A vary considerably [32-34]. Ergun used a value of 150, which appeared as a best fit for the data in his study. In early studies it was assumed that the packed bed comprised of identical smooth, non-porous, spherical particles. This results in a derived empirical value for A of 150. However in a chromatographic column this is not the case. There is a particle size distribution, and the particles are porous. This results in an increase in the value of A. Independent data has shown that a value of 270 is more appropriate for modem chromatographic materials [35,36]. At low flow rates, the viscous component, the left hand side of the equation, dominates, whereas at higher flow rates, the inertial component, the right hand side of the equation, is the dominant form. The left part of the equation is referred to as the Blake-Kozeny [37] or Kozeny-Carmen [38] equation. The right hand side of the equation is referred to as the Burke-Plummet [39] equation and is valid for turbulent flows. Combination of the two equations leads to the Ergun equation, which provides a very good approximation of events at low, intermediate and at high flows. The importance of this equation is that it provides a simple method for the determination of the existence of inertially dominated fluid dynamics, turbulence, within a chromatographic column packed with porous media. If the pressure drop across the column is found to be directly proportional to the flow rate then the Ergun equation predicts that that the mobile phase is laminar. However if there is a quadratic dependency between the flow rate and the pressure drop, then fluid flow is turbulent. Although the Ergun equation is very applicable to analysis of data generated from monitoring the pressure drop as the flow rate is varied, it is not easy to compare different column types since it is a dimensional equation. To allow comparisons between different columns and different column configurations it is necessary to use a different method of comparing data. In general a dimensionless term referred to as the friction factor is used to do this. Since this is a dimensionless term, it allows for comparison between different column configurations. The friction factor, which is a measure of the resistance to the flow, can be defined in a variety of ways, but can be derived by a simple normalisation of the Ergun equation with respect to the turbulent part. Thus the modified friction factor is given:

Ape3dp

A-q(1- e) m

L(1 - e)p~2

p$odp

a -B=

Re'

-B

(12)

Experimental data taken from a 50 ~m average diameter particle packed into a 1 mm i.d. column is shown in Fig. 4.7 [1]. The resultant plot, which shows the effect of modified Reynolds number vs. modified friction factor, is a common way of representing the Ergun equation. The two regimes of viscous flow and inertial flow can be seen, and the data generated from the pressure drop investigation clearly shows that there is a change in the friction factor at a transitional modified Reynolds number, as

M o d i f i e d Friction F a c t o r vs Re'

% I00 -

1,1 0

o~

r

~...~

0

~.

*mm *m

lO-

[m

*m

0

l 1

I0

E r g u n (theor)

o

Re'

T u r b o c 1 8 1.0 x 50 . . . . . .

I00

Kozeny ....

1,000

Burke-Plummer

I

Fig. 4.7. Data generated from pressure drop studies on Turbo C18 column, showing the deviation from a linear response as the fluid moves from a viscous into an inertial regime.

106

Chapter 4

predicted by Ergun. Increasing the flow and hence the modified Reynolds number above this point results in a non-linear dependency of the modified friction factor as the modified Reynolds number is increased.

4.2.6 P racti cal investigation

To determine if the chromatographic theory can be substantiated with practical experiences, it is necessary to perform a series of experiments. An organic compound is injected into a highly eluting mobile phase at varying flow rates and the efficiency plotted vs. the flow rate. In order to cope with the higher flow rates required to obtain turbulence, a large diameter particle is used as the packing within the column. The column internal diameter is 1 mm, the column length 50 mm, with the size of the spherical packing material being 50 Ixm. The laminar optimum for this size of column is traditionally quoted as 50 txl/min. The resulting plot, obtained by use of equations (1), (2) and (11), clearly demonstrates the effect of turbulent flow chromatography. The curve initially exhibits classical van Deemter characteristics, with an optimum flow rate occurring at about 50 ixl/min. Increasing the flow rate above this value results in an increase in the plate height, in accordance with the theory discussed earlier. However, as the transitional flow region is reached, between viscous and inertial flow, the plate height is reduced. Subsequent increases in the flow result in a significant reduction in the plate height, due to the increased mass transfer within the mobile phase [40]. The reduction in the mass transfer term associated with turbulent flow can only be utilised, in an isocratic system, if the retention, capacity, factor (k') is small, since it does not readily affect mass transfer within the stationary phase, only within the mobile phase. Higher retention factors imply that the compound is spending more time on the stationary phase and increases in the mass transferred to the vicinity of the surface are no longer beneficial [22]. 1~

-

A 0,5-

0.01

I

I

0.1

1

'1

10

Flow rate/mls/min

Fig. 4.8. Experimental data obtained from injection of organic compound on 1 x 50 mm HPLC column, 50 txm particles, showing the effect of varying flow rate on the band broadening process.

Turbulent flow chromatography in bioanalysis

107

The model developed by Pretorius and Smuts [21] assumes that the stationary phase is non-porous, an assumption which is mutually exclusive to fully porous particles. The majority of modem HPLC columns are packed with porous silica. This is an important difference and brings in the second component for a full understanding of how turbulent flow chromatography works. Data presented by Quinn et al. [1] shows that the molecular mass of the analyte affects the chromatographic performance of the system. In a series of experiments Quinn injected compounds of different masses at different flow rates and then plotted the resulting data as a plot of reduced plate height vs. modified Reynolds number. The data clearly shows that despite analytes being within the same fluid flow regime the effects of analyte band dispersion are affected by the mass of the analyte as shown in Fig. 4.9. The data can, however, be re-plotted using the reduced velocity instead of the modified Reynolds numbers. This gives a much clearer picture showing that there is a reduced velocity value, about 5000, above which increases in the reduced velocity result in the chromatographic performance of the system improving. Investigation of the reduced velocity, equation (3), reveals that there is a diffusion term in the denominator and this term, acting in conjunction with the turbulence in the bulk fluid, produces the desired beneficial performance.

4.2.7 Mass transfer into pores

Examination of the van Deemter equation [2] shows that at higher flow rates, under laminar flow conditions, band broadening is increased due to mass transfer effects. This term in the van Deemter equation can be expanded to include two distinct phenomena. The transport of mass across the fluid flow, and, the diffusion into and out of the pores, at the surface of the stationary phase. It has already been shown that under turbulent conditions there is an increase in the ability of the mobile phase to transport mass radially. Experimental evidence has shown that this effect is observable under experimental conditions, and that band broadening can be reduced with increasing flow. Next, the diffusion of the various sample components into the pores will be investigated. This will be considered as a separate process to the transport within the bulk of the mobile phase in turbulent conditions. Since this is a diffusional process, bulk flow turbulence will not be felt within the pores and adjacent to the surface of the stationary phase. In this case the driving force for mass transfer will be the concentration gradient coupled with the molecular mass of the compound diffusing across the concentration gradient. According to Fick's law [41 ] of diffusion the number of molecules moving across the concentration gradient in a radial direction, r, is given by Jr=-D--

dN dr

where: J is the flux of molecules to the surface of the stationary phase References pp. 127-128

(13)

Chapter 4

108

h v s . Re' 9O 80

/ m ............Wt~..............M .......................M .........................

70 6O 5O 9C 40 3O 20 10

/

0

1 ......

0.0

0.5

1.0

1.5

~. . . . . . . . .

2.0

2.5

i

3.0

Re' [--~ 2~ ~

2o,0oo ......m.......50,000

90,000 ~ 2 o o , o o o

--.-~,0o0. ]

hvs. v 90 80 70 60 .C

50

40 30 20 10

04 100

l ~

1,000

2500

:.. 20,000 ......t .....50,000

10,000

100,000

90,000 --e-- 200,000 ---&-- 650,000 ]

Fig. 4.9. Data obtained by Quinn, showing the effect of: (a) modified Reynolds number on the modified friction factor; (b) reduced velocity on the modified friction factor.

Turbulent flow chromatography in bioanalysis

109

D is the diffusion coefficient dN is the concentration gradient in the vicinity of the stationary phase. dr -

-

and where a reasonable approximation of the diffusion coefficient is given by [42]: D=--

k

(14)

where: k is a constant incorporating temperature effects and a collision cross section. However this term does not incorporate any bonding interactions. M is the mass D is the diffusion coefficient. It can be seen from the flow profile, for a turbulent flow, that the concentration close to the surface varies considerably, and that there will be a significant concentration gradient. Since this is one of the driving forces for the diffusion of molecules to the surface of the stationary phase, initial thoughts may conclude that under turbulent flow conditions, not only is there efficient mass transfer in the bulk flow, due to the existence of turbulent eddies, but that the increase in concentration gradient at the surface also ensures that there is better mass transfer at the surface and within the pores. Increasing the mass will reduce the flux of molecules across the boundary layer and into the pores. Thus, small molecules, like pharmaceutical drugs, will cross the boundary layer quicker than large molecules such as proteins. Optimal choice of the hydrodynamic conditions will ensure that there is sufficient residence time within the column for diffusion of smaller molecules to occur into the pores and to the surface, whereas much larger molecules will not have the time to diffuse across the boundary layer into the pores, before eluting from the column. This is an explanation of the phenomena observed by Quinn, where large molecular mass molecules exhibit the improved phenomenon at lower eddy concentrations of the mobile phase than that required for smaller molecules. This can readily be applied to the analysis of small pharmaceutical molecules in biological matrices. Under laminar flow conditions there is not enough mass transfer within the mobile phase to ensure high recoveries. Increasing the flow rate under laminar flow will reduce the interactions of proteins with the surface of the stationary phase, due to a reduction in residence time, however the poor mass transfer within the mobile phase will also result in a reduction in the amount of analyte molecules being trapped [43]. This is certainly the case with traditional materials used within HPLC columns, i.e. small spherical particles, generally derived from silica (see Fig. 4.10). However, the work by Mallet and Biddlecombe [44] has shown that monolithic materials can be successfully used to separate proteins from small drug-like molecules using high mobile phase flow rates. No reported comparison has yet been done between turbulent flow and monolithic columns. The different material design may aid mass transfer to the surface thus improving the chromatographic performance. It would be References pp. 127-128

110

Chapter 4

Loading flow rate / mL/min

Fig. 4.10. Effect of increasing loading flow rate on recovery of morphine sulfate from plasma. Obtained using a single column arrangement.

intuitive, however, to suggest that monoliths would suffer from significantly reduced capacity. However, it is not only diffusion that can be used to limit the amount of proteins interacting with the surface of the stationary phase. Proteins comprise of many different types of active components and hence binding sites. As a result they readily dissolve in aqueous solutions so can be considered hydrophilic. Their multitude of binding sites ensures that they will also bind very strongly to the surface of the stationary phase. Since the majority of binding sites exist within the pores of the stationary phase, it is possible to utilise a size exclusion process, proteins are substantially larger than most of the molecules under investigation. This is achieved by reducing the size of the pores to less than the size of a typical protein molecule.

4.2.8 Combining mass transfer and pressure drop A combination of increasing the flow rate to obtain a high linear mobile phase velocity and optimising solvent choice ensures that the analyte molecule can be effectively separated from the bulk of the matrix. Dongen [45] showed the effect of increasing the Reynolds number for the recovery of proteins under aqueous loading conditions. A mixture of proteins was diluted down, by a factor of 200, so that analysis could be performed by UV detection at 279 nm. The results showed that the amount of protein removal substantially increased as the flow was moved from laminar to turbulent flow (Fig. 4.11). They suggested that for the columns used this transition occurred between Reynolds numbers of 2-7.

111

Turbulent flow chromatography in bioanalysis 90

S"

~J

% 80

10 mL/mm 70 60 50

40 30 0

i

i

!

i

5

10

15

20

Re-number

Fig. 4.11. Plasma protein removal as function of flow rate 100 pJ plasma injection (200 times diluted). Increasing the flow rate, and hence Reynolds number results in less of the proteins being retained on the turboflow column.

4.3 APPLICATIONS OF T U R B U L E N T F L O W C H R O M A T O G R A P H Y Turbulent flow chromatography is an ideal separation technique for the separation of widely differing components. There are many samples that require extensive sample preparation before analysis can be performed. In particular biological samples, such as plasma and urine are traditionally analysed after extensive pre-treatment to remove large proteins from the sample. This can be achieved in a variety of differing techniques such as solid phase extraction, liquid-liquid extraction, protein precipitation and column switching techniques. All of these techniques when applied to bioanalysis concentrate on separating two very different classes of compounds, namely small pharmaceutical drugs and large proteins. These procedures can take time, using on-line techniques :reduces sample handling and also reduces the overall analysis time.

4.3.1 Applying the model With the development of mass spectrometers, the need to chromatographically separate a multitude of components has diminished. However within the field of bioanalysis, it is important that the detector is not contaminated with matrix, as this will affect sensitivity either due to ion suppression or due to a gradual lowering of signal intensity as the source becomes contaminated. To ensure that this does not occur it is necessary, with most on-line techniques, to :switch the flow between waste and a suitable detector. This requires at least one

,References pp. 127-128

112

Chapter 4

switching valve, which allows the use of a second pump to elute the samples from the column. To ensure that optimal performance is achieved, low flow rates to the detector have to be used. This has meant that either the flow is split, with the majority of the flow going to waste, so affecting sensitivity of the technique; or reducing the flow rate to one that is applicable to the flow rate required for optimal MS performance. The use of higher flow rates can reduce the analysis time, but can also affect the chromatographic performance of the system. It is possible to improve the chromatographic performance either by back-flushing the column, so reducing analyte-column interaction or by utilisation of a gradient, to improve the peak shape. There are many different configurations that have been used to allow the use of turbulent flow chromatography. The three most common configurations are described in detail below.

4.3.1.1 Single valve method There are three basic steps to use of this method: load, elute and re-equilibration of the columns. These are very comparable to the three basic steps involved in solid phase extraction (condition, load, elute) and serve comparable purposes. The sample is loaded onto the extraction column under turbulent flow conditions. The initial loading conditions are chosen to produce a minimal retention of the proteins and other matrix components. This is achieved by loading the sample in an aqueous mobile phase; typically this will be acidified to reduce protein binding. The aqueous mobile phase ensures that organic components within the original sample are retained on the column, whilst proteins and other hydrophilic compounds are eluted very quickly. This ensures that protein-stationary phase interactions are minimised so alleviating the problems of column blockage. The next stage is the elute step, where the components retained after the initial loading step are eluted from the column with a high elutropic mobile phase. The configuration of the valves is such that a second pump can be used in this step to flush out the autosampler and associated tubing. The final phase is to re-equilibrate the system ready for the next injection. This whole process takes about one to four minutes, but the big advantage is that the sample has not been manipulated in any fashion and there is no sample preparation involved.

4.3.1.2 Quick elute mode As with the single valve method there are three basic steps: load, elute and reequilibration of the columns. The use of two valves allows greater flexibility and makes the method more applicable to a wider range of samples The sample is loaded onto the extraction column in exactly the same manner as for the single valve system. The advantage of the two valves is that the column can be backflushed with the loading pump, using the same mobile phase as previously used. The elution step involves flushing the column with a highly elutropic mixture to remove the components of interest. Re-equilibration of the column within the quick elute method can be designed to be an incredibly thorough procedure. It has been shown that this can affect the applicability

%

Single valve one pump Load

Re-equilibrate

Elute

t,...

Prol.du debris

~!"~!~ix~,~:~i~

t'-. ~ii;i~!' ~

/-i~i ~i~ii~iil

~,i~84 ~| c~

t.,.. ~

Fllgho,ga~

~has,

t,.~.

Single valve two pump Load

~i

Fig. 4.12. Diagram of single valve using one and two pumps.

Elute

ii~iii~!

Re-equilibrate

~~

ta.>

Q u i c k Elute M o d e Elute

e l e t a ~ ~e w a u e

tl~Vt~)

i

....

Fig. 4.13. The three basic steps to using the Quick Elute method.

~i~~

Re-equilibration

~J~

Turbulentflow chromatography in bioanalysis

ll5

of TFC to a wider range of analytes. Ensuring that the cleaning stage is thorough will reduce carryover as well as increasing the sensitivity of the technique.

4.3.1.3 Focus mode To achieve chromatographic resolution another method was developed which allowed for a separation of isobaric compounds or reduction in ion suppression effects. There are four simple steps to using this particular technique. The first step is exactly the same as with the quick elute mode, sample is loaded onto the column under turbulent flow conditions which allows separation of the two basic components, namely that of matrix and analytes. The second stage is to transfer the analytes onto a second column, using an organic solvent plug. Both pumps are running aqueous mobile phases and so the result is that the analytes will be retained on the second column as the solvent plug is diluted in the internal 'T'. Careful selection of the flow rates will ensure that the analytes are focused at the top of the column, ready for the third stage. The third stage is to elute the compounds of interest from the top of the second column into the detector. A laminar flow gradient is used here to ensure that optimal separation is obtained. Laminar flow is used as a consequence of the detector requirements. Whilst the compounds are eluting the turboflow column is being cleaned and the sample loop refilled with organic solvent. Finally, the system is re-equilibrated ready for the next injection.

4.3.2 Application areas

Turbulent flow chromatography coupled to tandem mass spectrometry has been used in a wide range of applications from analysis of pharmacokinetic samples, to forensic applications and environmental samples, providing fast, sensitive and specific analysis for the determination of compounds in a range of matrices. The following section will highlight some of these application areas.

4.3.2.1 Drug metabolism and pharmacokinetic (DMPK) studies The first publications based on turbulent flow chromatography, after the original issuing of the patent, applied to the analysis of biological fluids in a commercial environment, were by Ayrton [46,47]. The technique was successfully used in the analysis of a novel pyridine-isoquinolone compound and its deuterated analogue, under development by Glaxo-Wellcome at the time of the research. The compound was found to be strongly protein bound, in excess of 95%. Using a single valve technique, the authors were able to demonstrate that accurate and precise data could be obtained over the range 5-1000 ng/ml with a 10 ~1 sample of plasma spiked at the appropriate levels. The flow was split prior to the mass spectrometer by 10:1, which obviously affects sensitivity, but also reduces possible fouling of the mass spectrometer source. It was noted that using a silica-based column that there was no appreciable pressure rise with an overnight run of 100 samples, with the injection volume raised to 50 ~1 (comprising

References pp. 127-128

Focus Mode Load

v.o,,. , . ~

~~~~":~

Transfer

9~"

! -

~~,~

! I

~.

,.oz~

I [

'h! %%

%

-

Aqucousm~il,~ ph~c

Elute

~,~~-:~

refocused

Aqueous rrmhil~ph~.~

Re-equilibration

i~ ~ ~ ....J ~.............

linear ~rldi

l

....

ch Aquco~ r,~c~d~ilc ld~.~

Fig. 4.14. The four basic steps to using the Focus Mode method.

4~

Turbulent flow chromatography in bioanalysis

117

25 ~l of plasma and 25 p~l internal standard mix) and also that there was no significant fouling of the mass spectrometer interface, a common problem with protein precipitation. The total run time was quoted as 2.5 minutes. Chassaing et al. [48] developed a TFC-MS-MS method which was successfully used and validated for the determination of three basic compounds, doxazosin, CP-122,288 and dofetilide in dog plasma in a 96-well format. Chassaing used a quick elute configuration, utilising 2 Rheodyne 6 port switching valves as well as two standard Agilent 1100 HPLC pumps. The extraction column was a Cohesive turboflow column (C 18, 50 mm x 1 mm 50 ~m average particle diameter). Samples were prepared by adding 200 p~L of a solution of internal standard, made in 90% water, 10% methanol) to the 200 ~L plasma. These were prepared in a 96 well plate format, centrifuged at 2000g for 1 hour at 4~ prior to injection. This process ensured that no particulates reached the system, increasing the column lifetime. The assay showed an excellent linearity across the range of 5 to 500 ng/ml for all compounds. Studies also showed that the analysis was repeatable over a period of time as the interday assay results clearly show. To ensure the validity of the data generated using TFC, 22 real samples from 2 studies were re-analysed using an old solid phase extraction sample pre-treatment followed by analysis of the extract using HPLC-MS-MS. The experiments showed that there was excellent agreement between the plasma concentrations measured by TFC-MS-MS and those measured with the SPE-LC-MS-MS method (see Fig. 4.15). Although most TFC is employed as a precursor to detection by MS or MS-MS, Kennedy [49] used TFC coupled to UV detection to determine ciprofibrate in human plasma. Ciprofibrate is a fibric acid derivative hypolipidaemic agent, used for the treatment of primary hyperlipidaemias resistant to appropriate dietary management including hypercholesterolaemia, hypertriglyceridaemia and combined hyperlipidaemias. A range of concentrations between 1 and 30 ~g/mL were determined. Human plasma was initially diluted with an equal volume of 0.1M hydrochloric acid, with 40 ~L of the resulting mixture being injected onto the turbo flow system. A single valve method was employed for the analysis with the total analysis time being 2 minutes. It was noted that a further 1.3 minutes was spent waiting for the autosampler to prepare for the subsequent injection. The reported data showed excellent linearity over the specified range, with no observable carryover for this particular compound. Calculation of the limit of quantification (LOQ) using the guidelines of the International Conference on Harmonisation (ICH) gave a value of 58 ng/mL. The reported validation data was as good as if not better than data generated using a SPE method. The columns lasted in excess of two hundred injections of diluted plasma, however it was noted that using the developed procedures the columns were not used to their full extent. Kennedy also reported that the performance of the column did not deteriorate with increasing accumulation of backpressure. Zimmer [50], using a quick elute method, demonstrated that TFC could be successfully applied to the analysis of two internal compounds named as compound A and compound B, over a dynamic range of 1 to 500 ng/mL. The validation data presented clearly showed that this technique could be successfully employed in the routine analysis of biological fluids. Cross-validation of the results using data generated References pp. 127-128

TABLE 4.1 INTRA- AND INTER-DAY IMPRECISION AND INACCURACY OBTAINED FROM CHASSAING [49] Intra-day

(ng/ml)

Inter-day

Mean observed concentration (ng/ml)

n

SD

CV (%)

Inaccuracy (%)

Mean observed concentration (ng/ml)

5.50 16.18 249.23 399.12

7 7 7 7

0.36 0.94 6.34 4.94

6.51 5.84 2.54 1.24

9.98 7.86 -0.31 -0.22

5.21 15.72 248.48 399.47

5.20 14.79 240.84 380.24

7 7 7 7

0.23 0.37 3.60 5.78

4.44 2.53 1.50 1.52

4.06 -1.39 -3.66 -4.94

5.32 14.20 241.70 384.58

7 7 7 7

0.96 1.31 5.33 11.8

18.08 9.22 2.20 3.06

6.48 -5.33 -3.32 -3.86

SD

CV (%)

Accuracy (%)

20 21 21 21

0.68 1.30 5.31 8.39

13.1 8.29 2.14 2.10

4.28 4.80 -0.61 -0.13

5.02 14.73 238.50 376.11

20 21 21 21

0.48 0.59 3.96 10.1

9.63 4.01 1.66 2.68

0.37 -1.77 -4.60 -5.97

4.87 14.51 237.48 383.78

20 21 21 21

0.33 0.86 5.82 6.00

6.81 5.96 2.45 1.56

-2.56 -3.29 -5.01 -4.06

Doxazosin level 5 15 25O 4OO CP-122,288 level 5 15 250 400 Dofetilide level 5 15 250 400

4~

Turbulent flow chromatography in bioanalysis 1000 -

o

119

TFC

100

g~

o

lO

0

I

1

I

I

I

2

4

6

8

10

time (hour)

Fig. 4.15. Cross validation data for real study samples.

by a liquid-liquid extraction technique gave very good correlation for both the drags analysed, less than 5% standard deviation for 42 samples for both compounds. One of the most significant findings of the research was the dramatic reduction in analysis time, for one batch of 96 samples, the sample analysis time was reduced to less than a third of the original liquid-liquid extraction method (16 hours). It was noted by the author that this could be further reduced, by reducing the autosampler cycle time. This could be achieved either by reducing the number of washes or more effectively by having the autosampler perform the majority of its cycle during the previous analysis. Jemal [51 ] used a single valve method for the analysis of a variety of compounds with mass spectrometry as the detection system. The initial sample was diluted with an equal amount of an aqueous solution of the internal standard, with 50 b~l of the resulting mixture injected onto the system. Samples were placed in 2 mm vials, vortexed and then centrifuged to allow separation of any particulates within the sample. Two quick elute methods were described for the analysis of compounds in rat plasma and in human plasma. The total analysis time was 4 minutes, with 1 minute for loading of the sample and elution of the matrix, 2 minutes for the elution of the drug, and one minute for reequilibration of the system ready for the next injection. The second method used an analytical column in-line with the extraction column during the elution step to provide chromatographic resolution between the two isomeric compounds. This method had a total run time of 5 minutes with slightly longer loading and elution steps than the previous method. The results presented are very good over the specified range (1-2000 ng/mL), with accuracy ( < 10% across the range) and precision ( < 4% at the lower limit of quantification) (LLOQ) well within regulatory levels. Wu et al. [52] were able to analyse up to 10 compounds simultaneously from cassette dosing experiments using a single valve method with two columns, extraction column and an analytical column. In this study two pumps were used, one to load the sample onto the extraction column, the other to elute the compounds onto the mass spectrometer. Comparisons with on-line SPE revealed that separation efficiency and, dynamic range, accuracy and precision were comparable between the two methods over References pp. 127-128

120

Chapter 4

the range investigated, 1-1000 ng/mL. These studies revealed that pressure build-up using on-line extraction columns under laminar conditions was a problem with the columns failing after 80-100 injections. Using turbulent flow chromatography, with the specially developed columns gave substantially better column lifetimes Injection of 100 ~1 of sample plus internal standard (approximately 85% plasma) was used in the study. A tandem mass spectrometer was used as the detection, with a gradient applied across the extraction column after the initial extraction step had been finished to elute the compounds of interest from the column to the detector. It was found that turbulent flow chromatography was applicable to poorly water-soluble compounds as well as highly protein bound compounds. Brignol et al. [53] investigated the analysis of terbinafine, an antifungal agent, from human and pig plasma. Two procedures were utilised an off-line protein precipitation method followed by LC-MS-MS and the on-line technique. The off-line analysis was able to reach a LLOQ of 0.0678 ng/mL in human plasma, with acceptable precision and accuracy values. Pre-concentration of the precipitated plasma meant that the limit of quantification was lower than using an on-line technique, however it also meant that more plasma was required, which is not always desirable. The dynamic range was from the LLOQ to 89.9 ng/mL. Using the off-line technique recoveries varied across the range from 53.4% to over 70%. A focus method was used for the analysis using TFC. Although the limit of quantitation was higher (0.117 ng/mL), primarily since only 50 ~L was used as opposed to 200 ~L, the precision values were better than quoted for the off-line technique (ranging from 1.15% to 6.68%). Also, since the sample did not have to blown down and then reconstituted in a suitable mobile phase, the sample preparation time was substantially less. Taxol is widely known for its therapeutic effect in the treatment of some cancers. Dongen [54] developed a method for the analysis of this compound in human plasma. Example chromatograms are shown in Fig. 4.16, with the statistics from the validation study shown in Table 4.2. Within most industries there are pressures to produce data more quickly, and using turbulent flow chromatography, Dongen was able to reduce method development time down to two days. Other authors have very generic methods, which require virtually no method development, and they become applicable to a wide range of compounds. Using such a genetic method, Hermann [55] was able to analyse over 1000 different compounds. Samples were prepared by taking 100 p~L of plasma and adding 200 p~L of acetonitrile containing an internal standard. The sample was then centrifuged for ten minutes at 10,000 rpm and the supernatant placed into an autosampler vial. The samples were then analysed using the generic focus method. To illustrate the applicability of the TFC approach, six compounds, with a large range of hydrophobicites, were chosen and analysed in rat plasma. The chromatogram for each compound at 5 ng/mL in rat plasma is shown in Fig. 4.17a. All the compounds have good sensitivities, high recoveries (> 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~

0-

0 2

3

4

5

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

400000- R 2 = 0 . 9 9 8 9

m "~ r == 300000

R2

9 Isoproturon

R2 = 0.999

600000 - R 2 = 0 . 9 9 9 2

9Diuron 9Chlortoluron x Simazin Atrazin

R2 = 0.993

9Terbutylazin

= 0.9986

+ Prometn]n

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

. Chlorfenvinphos - Chlorpyrifos 9Alachlor

100000

9Tdfluralin

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 9 2003 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 ChiralpakTM

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

CHIRAI.CEL OB

.~ Rt: ~ ,

O CHIRALCEL OJ

CHIRALCEL OJ-R ~ 0

c.,

"':

N

CHIRALPAK AD CHIRALPAK AD-RH

R=:

o

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

CHIROBIOTIC V (Vancomycin)

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

H 3C

.o

NH~ ~

HO HO ~ _ ~ _

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

,o,..< ~'o

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

HO

"

~:_.j

E

c ~

OH

3

HO "~- ~

HO OH

~o

.o.

~,

O

.N

O

,

~'O

~ ~

OH

CH

-o ~ -"y-"-

3

OH

NH 2 HOHOH~

CHIROBIOTIC T (Teicoplanin)

CH~cH O

~..~OH

--.o .o

-H

OH

o

OH

CH 3

o.

o.

O

~~"~--7~

HO

o,. ~O..o

~o, ~ N " ~ ~

HO ~

c.~o. o.

~

OH

CHIROBIOTIC TAG (Teicoplanin Aglycone)

,~j~

,~

CI

O

H HO~,~ W

NH 2

..

H O

CI

CI

o

~'--"~-- OH

H

O

B H N

O C i

O

H H I

O I

.-NH 2

~N H

HO ~

O OH

H

~'~0 ~

CH~)H

o~~

HO

H

~

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

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

4~

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 Flow rate

tR enantiomers

100 MeOH/0.03% ATFA*

1.2 ml/min

5.0, 10.4 min

100 MeOH/0.03% ATFA*

1.0 ml/min

6.1, 7.2 min

50 • 4.6 mm

100 MeOH/0.03% ATFA*

1.0 ml/min

6.2, 6.8 min

50 z 4.6 mm

100 MeOH/0.03% ATFA*

1.0 ml/min

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

Compound

Column

Column size

Mobile phase

Ritalinic acid [ 188]

CHIROBIOTIC T

50 • 4.6 mm

Methylphenidate [ 171 ]

CHIROBIOTIC V

150 • 4.6 mm

Fluoxetine [ 188]

CHIROBIOTIC V

Nicardepine [ 188]

CHIROBIOTIC V

Metoprolol [ 188]

CHIROBIOTIC T

150 • 4.6 mm

100 MeOH/0.03% ATFA*

1.2 ml/min

Salbutamol [45]

CHIROBIOTIC T

250 • 4.6 mm

100/0.5/0.1 MeOH/AcOH/NH4OH

2.0 ml/min

* ATFA is ammonium trifluoroacetate.

r

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

TABLE 5.4 RANGE OF w-COMPLEX CSPs AVAILABLE Bonding chemistry

Type

Commercial name

1-(3,5-dinitrobenzamido)-tetrahydrophenanthrene 1-(3,5-dinitrobenzamido)-tetrahydrophenanthrene

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

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

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

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

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-acceptor w-donor w-donor w-donor w-donor w-donor w-donor w-donor w-donor w-donor w-donor

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

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) ~243

.

. . . . .

~:

;-

:

.

+

,]-SaPmm~ol

--~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

Fig. 5.7. Typical chromatograms of standards and unknown from a clinical assay for metoprolol and a-hydroxymetoprolol on Chirobiotic T in the polar ionic mode [52].( I ) blank, (2) standard containing 2 nglml each isomer, (3) standard containing 50 nglml each isomer, (4) post dose sample from a typical patient.

Chapter 5

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 <3.0% and < 4.5%, respectively. It was noted that the column was in operation for about a year without any decrease in resolution or retention any of the solutes. The method was adopted for routine therapeutic monitoring and toxicopharmacokinetic studies. The same method was used with a different internal standard (diphenhydramine) in a recent study [63] that provided a sensitivity of 1.5-1.6 ng/ml for the parent drug and 2.5 ng/ml for the metabolite. A Chiral-AGP method for methadone has been described which used a coupled column technique with a short (30 x 4.6 mm) C8 column and an ion-pairing reagent [64]. Using a mobile phase of dimethyloctylamine, acetonitrile and phosphate buffers, the authors reported a linear range of 0.02 to 2.5 txM for methadone and its two cyclic N-desmethyl metabolites using UV detection at 200 nm. The combined columns could be used for over 1000 injections. The same method had previously been used in combination with an SPE disc extraction technique [65]. A combination of indirect and direct LC methods were used to separate the ecstasy analogue, (R,S)-N-ethyl-3,4-methylenedioxyamphetamine (I) from its metabolites, O-glucuronyl-(R,S)-N-ethyl-3,4-methylenedioxyamphetamine (II) and (R,S)-N-3,4methylenedioxyamphetamine (III) [66]. The major O-glucuronyl-metabolite was separated directly on Chiral-CBH while the separation of the other two compounds was carried out separately using [3-cyclodextrin in the mobile phase on a reversed-phase column. A combination of detection methods was also u s e d - fluorescence at 322 nm for compounds I and III, and electrochemical detection at + 0.6 V for the metabolite II. The LOQs were 16 and 1.2 ng/ml for II and III, respectively. A plasma sample study from six healthy volunteers following oral administration of 140 mg of racemic I revealed that the R-isomer for the parent compound predominated whereas it was the S-isomer that exhibited higher plasma concentrations for the two metabolites. An indirect LC References pp. 180-184

156

Chapter 5

method was developed for the enantiomers of pentazocine that incorporated sulphated [3-cyclodextrin in a phosphate mobile phase on an ODS column with UV detection at 220 nm [67].

5.3.6 Analgesic drugs (non-narcotic) Different but complementary pharmacological mechanisms are responsible for the different analgesic effects of the two enantiomers of tramadol, a centrally acting, medium potency, analgesic that acts as an opiate agonist. It is metabolised in the liver mainly to O-demethyltramadol, mono-N-demethyltramadol and di-N,O-demethyltramadol, although only the first one is pharmacologically active with a ( + ) isomer that is approximately 200-fold more active than its counterpart. A recent study [68] utilised a reversed-phase, 3 txm C8 column (Hypersil BDS) in series with a Chiralcel OD(R) column and a mobile phase of acetonitrile and phosphate buffer to determine the amounts of the enantiomers of both tramadol and O-desmethyl-tramadol. Fluorescence detection at 301 nm (excitation 199 nm) provided a sensitive method, down to an average of 0.04 ng/ml, for the four compounds which was used for a pharmacokinetic study in human volunteers. It was reported that thirty samples could be assayed in a 24 h period. This method had been previously compared with alternative methods on Chiradex and Ultron OVM [69] but was found to be the most suitable. An alternative method was developed for normal phase chromatography, also using the cellulosic CSP, Chiralcel OD [70]. Following automated SPE extraction using ethyl silica cartridges, the samples were evaporated down and reconstituted in the LC mobile phase of diethylamine-ethanol-hexane. Tramodol and O-demethyltramadol in extracts from human plasma were separated on a Chiralpak AD and detected by APCI HPLCMS-MS in the positive ion mode. Precisions of less than 6% were realised for both enantiomers and the method was also found to be selective for the other two metabolites, mono-N-desmethyltramadol and di-N,O-desmethyltramadol. A study of the pharmacokinetics and the effect of gastrointestinal permeability of ketorolac was carried out in rats [71]. Oral doses of the racemate and each of the enantiomers showed that the pharmacokinetics of ketorolac is enantioselective, the inactive R-enantiomer producing higher plasma concentrations when compared with the active S-isomer: it was reported that each enantiomer underwent little or no chiral inversion. Sucrose was used as the GI permeability marker, the S-isomer being found to be less toxic on the upper GI tract but having no advantage in GI toxicity. A normal phase method was used for determination of quantities in plasma and urine (ChiralPak AD), the samples being injected directly after centrifugation at 1800g. Alternatively, ketorolac enantiomers can be separated on Chiral-AGE using naproxen as internal standard, after liquid-liquid extraction [72]. Dimethyloctylamine was used as an ionpairing additive to the 2-propanol/phosphate buffer and detection was made at 320 nm. The method was linear over a wide range of concentrations (20 to 20,000 ng/ml). Alternatively, Chiral-AGP can be used with 2-propanol and phosphate buffer at pH 5.5 containing 2.5 mM sodium azide [73], and this method was found to be comparable with an indirect method.

Chiral bioanalysis

157

5.3.7 Anesthetic drugs (intravenous) A Chiral-AGP column was used for the determination of thiamylol enantiomers following liquid-liquid extraction [74]. A mobile phase of 2-propanol and phosphate buffer enabled the separation of the enantiomers in less than 15 minutes; the metabolite (secobarbital) did not interfere. The limit of detection was 1.5 ng (injected) and good correlation of the method with an achiral method was observed for the sera from seven patients. Intra-assay RSDs of 1.35 to 3.01% (5 samples) were reported. Following intravenous administration of thiopentone, the pharmacokinetics of the enantiomers was studied, determining the concentrations in human plasma using a Chiral-AGP column [75]. Racemic ketamine was used as the internal standard and UV detection at 280 nm provided an LOQ of 10 ng/ml. The same method was used for measuring thiopentone enantiomers in sheep plasma, switching detection from 220 nm to 287 nm after 6 minutes [76]. Following a double SPE extraction, the enantiomers of ketamine and of norketamine were well resolved on a Chiral-AGP column with methanol and phosphate buffer at 40~ [77]. UV detection at 220 nm gave a linear response from 10 to 320 ng/ml.

5.3.8 Anorexic drugs Diethylpropion racemises in plasma with a half-life of 23-25 minutes for each isomer [78]. The effect of the addition of cyclodextrins on stability was investigated: spiked plasma was incubated and tested over time using a Chiracel OD column operated at 5~ in normal phase mode. The rates of racemisation in the presence of a cyclodextrin for each enantiomer did not differ and sulphobutyl ether or methylated [3-cyclodextrin were both successful in retarding the racemisation process.

5.3.9 Anthelmintic agents A novel enantioselective assay for metrifonate has been developed using normal phase chromatography and a mobile phase of heptane-ethanol containing 0.25% water with two 250x 2 mm Chiralpak AD columns connected in series (Fig. 5.9) [79]. This compound is under development for the treatment of Alzheimer's disease and the method was developed for low level determination in blood and brain tissue. The method provides a successful alternative to previous GC methods that suffered from thermal decomposition of the drug to a molecule that is also present as one of the metabolites. This on-line HPLC-MS-MS method used two Chiralpak AS (250 x 2 mm) columns in series at 40~ with an isocratic mobile phase of n-heptane and ethanol containing 2 mM ammonium acetate and 1% distilled water: this provided a run time of 11 minutes. An ionspray interface in the positive-ion mode at 100~ using nitrogen as both auxiliary and nebuliser gas, was used. The method was successfully applied to more than 2000 samples in pre-clinical and toxicological studies. Albendazole and its sulphoxide metabolite were separated on a Chiral-AGP column using 2-propanol and References pp. 180-184

Chapter 5

158 ~

11~~

Mouse

~Y

W

qJm

z ~2'1~ ~ 4is

iJ'A' 9

m. ~ql t, ~se't ~2:Doub 445

Rat IP

tn~d~: ~ q m

BAY z 7216

TO

gO

BAY z 7217

SO 40 30 |o to

~.4 .,L T,l

t1|1-*

~;1 .ha-._ 4f ,

q..

~,,,n

a,n

iX,,_

_

Fig. 5.9. MRM traces from coupled Chiralpak AS columns in normal phase after oral doses of metrifonate in whole blood extracts of mouse and rat [79].

phosphate buffer as mobile phase [80]. The calibration was linear from 0.05 to 1.5 lxg/ ml of both enantiomers. In an alternative method, a Chiralcel OB-H column equipped with a cyano pre-column and a mobile phase of hexane-methanol was selected for the analysis of this drug in plasma [81]. Fluorescence detection at 320 nm (280 excitation) enabled determination of the drug over a wide linear range of 50 to 5000 ng/ml. A method for the enantiomers of praziquantel, was reported on Chiralcel O F - R in perchlorate and acetonitrile [82] and was found to have no interference from trans4-hydroxypraziquantel. Ondansetron was used as the internal standard.

5.3.10 Antiarrhythmic agents Although used clinically as the racemate, propafenone exhibits enantioselective clearance, the R-isomer being cleared faster than its S-counterpart. It is extensively metabolised, and the major metabolite, 5-hydroxypropafenone, also possesses antiarrhythmic properties. A method was developed that enabled the assay of both parent and major metabolite using a Chiral-AGP column with a mobile phase of ammonium acetate and propanol. Previous methods required two separate assays using Chiralcel

Chiral bioanalysis

159

AD and OD [83]. The method, which used ESI-MS in the positive ion mode with SRM, demonstrated sufficient sensitivity for comprehensive pharmacokinetic studies of propafenone. It did, however, require two separate injections under slightly different conditions for each analyte (1-propanol for the parent compound, 2-propanol for the metabolite) following prior separation of parent and metabolite on an achiral column: effluents from the column were collected and injected. An alternative method used UV detection and a Chiralcel OD-R column with a mobile phase of acetonitrile and perchlorate buffer [84] A Chiralpak AD column was also used to simultaneously separate the enantiomers of disopyramide and its mono-N-dealkyl metabolite in plasma and urine [85]. The 4 enantiomers were separated with ethanol-hexane (9/91) containing 0.1% diethylamine at 270 nm within 20 minutes with an LOQ of 12.5 ng/ml. Although administered as a racemate, S-(-)-verapamil is 10-20 times more potent that the R-(+) enantiomer and the S-(-) isomer undergoes extensive first pass metabolism, resulting in higher concentrations of the R-( + ) isomer in plasma: protein binding is also enantioselective. The enantiomer and its seven metabolites have been separated using a Chiral-AGP column operated with a mobile phase of acetonitrile/ 10 mM phosphate (pH 7), 1:9 and fluorescence detection at 310nm (excitation 276 nm), although only some of the metabolites could be separated simultaneously [86]. A tandem HPLC-MS method was also developed on Chiralpak AD in normal phase mode using a short, narrow bore column (100x 2.1 mm) [87]. Post-column addition of an aqueous solvent phase was performed to enable direct coupling of the chiral column to the MS despite the normal phase chromatography (comprehensive details of the technique are provided by the authors). The enantiomers of both verapamil and norverapamil were simultaneously determined by this method. This method was also used for sotalol, for the antihypertensive doxazosin and the anticholinergic oxybutynin. In a previous study, identification of all of the metabolites of verapamil was carried out by collecting fractions following their separation on an Ultron OVM column with M S MS used to confirm that six of the metabolites were present predominantly in the (R)-form with just one mainly as the (S)-form [88]. An indirect method using chiral derivatisation was developed for the enantioselective analysis of N-hydroxymexiletine glucuronide in human plasma [89]. Fluorescence detection gave an LOQ of 50 ng/ml. The major metabolite of aprindine was isolated from rat faeces by TLC and its enantiomeric ratio determined on a ChiralPak AD: enantiomeric ratios of 5 in faeces and 15 in human urine were found [90]. Pindolol was used as a test compound for an on-line extraction method developed in conjunction with a Chiral-AGP column [91]. Serum proteins were removed using a restricted access cartridge, the drug was separated from endogenous material using a gradient on a C 18 column and heart-cutting of the pindolol peak was carried out in preparation for the chiral separation. The method was also applied to other antiarrhythmics. Both the racemate and the pharmacologically active L-enantiomer of Simendan are novel positive inotropic drug candidates for congestive heart failure. An analytical chiral method was developed to determine the possibility of metabolic inversion of the L-form in humans and animals [92]. The method used a Cyclobond I 2000 column and a References pp. 180-184

160

Chapter 5

reversed-phase mobile phase of triethylamine acetate in methanol, and it was concluded that having the chiral centre in between two w-systems assisted the separation. Although protein binding was found to be slightly enantioselective, no chiral inversion was observed, justifying the use of an achiral method for bioanalysis following administration of the L-enantiomer. Detection at 380 nm provided an LOQ of 10 ng/ml for the racemate: in dog plasma 20 ng/ml of the D isomer could be detected in 1000 ng/ml of the L form, while in human plasma the corresponding detection limit was 5 in 500 ng/ ml. An indirect method was validated for the determination of the metabolites of mexiletine; hydroxymethylmexiletine and p-hydroxymexiletine [93]. The enantiomers were monitored using circular dichroism

5.3.11 Antibacterial drugs Two potential methods for the enantioselective separation of the antibacterial drug, temafloxacin, were developed for its clinical investigation as a broad-spectrum antimicrobial agent [94]. An indirect method, employing a lengthy 2-stage derivatisation with the chiral reagent, (S-(-)-N-l-(2-naphthyl sulphonyl)-2-pyrollidine carbonyl chloride (L-NSPC), provided a separation in about 20 minutes. The direct method, however, also required that the molecule was N-acetylated and esterified before complete separation could be achieved on an OVM CSE

5.3.12 Anticoagulants A sensitive method for the separation of acenocoumarol and phenprocoumon in plasma was developed, using S-warfarin as internal standard [95]. A Whelk-01 column in normal phase mode and detection at 310 nm provided a detection limit of 5 Ixg/ml for all isomers with a linear range over 15 to 2000 txg/ml. (R)- and (S)-Warfarin were determined in human plasma using oxybenzone as internal standard and a Cyclobond I 2000 column in the polar organic mode for the analysis. This internal standard was shown to be a better choice than naproxen as previously used [96], as it was considered less likely to be present in plasma in significant amounts, extracted similarly to warfarin and eluted much earlier. The mobile phase consisted of acetonitrile-acetic acid-triethylamine (100:3:2.5 v/v/v/) and the three components eluted at 7.7, 6.9 and 4.4 minutes, respectively. For higher throughput, late-eluting peaks from endogenous components were diverted on to a 50 x 4.6 mm precolumn containing the same CSP by utilising a column-switching device, enabling 96 injections to be made in a 24-hour period. Liquid-liquid extraction was used for sample preparation with UV detection at 320 nm. The overall RSD was reported to be 6%. Analytical column-to-column variation was found to be good, although some variability was noted in the pre-columns used. A Whelk-01 column was also reported to be effective for this assay [97], using reversed-phase solvents and UV detection at 313 nm. This method was linear over a

Chiral bioanalysis

161

range of 0.25 to 1.5 ~g/ml for both enantiomers. A normal phase method for warfarin and 7-hydroxy warfarin using a Chiralcel OD column was the method of choice in a study by Takahashi et al. [98]. UV detection at 312 nm for the parent compound and fluorescence at 415 nm (excitation 320 nm) enabled detection limits of 20 and 40 ng/ml for R and S warfarin, 2.5 and 4.5 ng/ml for R and S 7-hydroxy warfarin.

5.3.13 Anticonvulsants A single dose of mephenytoin was used as a probe for the important drug metabolising enzyme CYP2C19 activity in a study of genetic polymorphism in healthy human volunteers [99]. This was carried out in conjunction with the administration of dextromethorphan as a probe for CYP2D6. An S,S-Whelk-01 with a normal phase mobile phase of hexane-2-propanol (83:17) was used to separate the enantiomers, but each was separately quantified by subsequent achiral GC-MS analysis to obtain an LOQ of 0.1 p~g/ml. The enantiomers of a candidate anticonvulsant, PNU-83894 (3,4-dichloro-N[(1R,2S)-2-(methylamino)cyclohexyl]benzamide), and its N-desmethyl metabolite, PNU-83892, were separated on the [3-cyclodextrin column, Cyclobond 1 2000, with UV detection at 230 nm [100]. Following pre-treatment of the plasma samples using a phenyl bonded SPE column, a mobile phase of acetonitrile-triethylamine acetate buffer was used to elute the isomers. The method was used to evaluate the pharmacokinetics of the compound in the dog. Another anti-convulsant compound, losigamone, was studied in plasma [101]. The method employed used a reversed-phase system (acetonitrile-methanol-phosphate buffer at pH 4.2) on a Chiradex cyclodextrin column. By using a C18 saturator column between the pump and guard column, a column lifetime of 300 injections was achieved. The method showed no interference from carbamazepine and phenytoin or their metabolites, or from other common anti-epileptic drugs. A fully automated method for the simultaneous determination of mephobarbital and phenobarbital used on-line extraction with a LiChrospher ADS pre-column, backflushing the extract directly onto the chiral column, a Chiralcel OJ-R coupled with a DIOL guard column [102]. The method had a within-day RSD averaging 1.7 to 2.2% (n = 6) and a day-to-day RSD of 5.5 to 5.7% (three days). The RR and SS enantiomers of the dihydro metabolite of carbamazepine were separated on a Chiralcel OJ CSP after liquid-liquid extraction and detected at 210 nm [103].

5.3.14 Antidepressants A study was carried out on a newly developed selective monoamine oxidase-A inhibitor, E2011 [104]. Based on an oxazolidinone structure (Fig. 5.10), the compound has two chiral centres and separates on a Chiralcel AD with a normal phase eluent (hexane2-propanol-ethanol). After solid phase extraction on a C18 cartridge and evaporation, References pp. 180-184

162

Chapter 5

(t)

(n)

(,t)

Otto

Fig. 5.10. Structures of monoamine oxidase A inhibitor compounds I and II, and of internal standard, III [104].

the rat plasma samples were reconstituted in isopropanol and 50 txl injected onto the CSP. Chiral inversion, at the oL-carbon of the benzothiazole ring, was found to occur enzymatically in vivo. Citalopram is one of a group of selective serotonin reuptake inhibitors, now prescribed more frequently than tricyclic antidepressants. Two different chemometric modelling programs were used to develop the chiral method on a Cyclobond 1 2000 Ac [105]. A method was required that enabled the assay of the isomers of the parent drug and of the two metabolites formed by N-demethylation - desmethylcitalopram and didesmethylcitalopram (both less potent than citalopram itself). In the end, a compromise had to be reached to achieve reasonable separation of the 6 components and a reversed-phase solvent with a composition of 55:45 methanol-10 mM buffer (pH 6.3) was chosen. A previous method had used Cyclobond I 2000 with methanol and 1% diethylamine buffer to separate citalopram from its desmethylayed metabolites in plasma [106]. Chiral-AGP has been used to provide in an alternative method for citalopram [107]. Ion pairing reagents were added to phosphate buffer (pH 6.5) at 32.7~ to optimise the separation. An earlier method used a Chiralcel OD CSP in normal phase mode to detect between 2 and 15 ng/ml of the enantiomers of citalopram, its desmethylated and proprionic acid metabolites in human plasma by fluorescence [108].

163

Chiral bioanalysis

Following liquid-liquid extraction in a 96-well plate format, chiral analysis by positive ion APCI enabled an LOQ of 2 ng/ml for fluoxetine to be achieved using a Chirobiotic V (250x 4.6 mm, 5 Ixm) column [109]. With a simple mobile phase of methanol, containing 0.075% (by weight) of ammonium trifluoroacetate, a resolution of 1.17 and an assay time of less than 10 minutes were achieved. This mobile phase uses this CSP in the polar ionic mode, utilising its capabilities for strong hydrogen bonding and ionic interactions. The ruggedness of the assay was measured using a 3-day validation. Standard calibration samples were prepared in triplicate: in addition, six replicates of four QC concentrations within the anticipated concentration range (2.0-1000 ng/ml) were analysed on each day. The inter- and intra-assay precision was < 13.6%, within the acceptance criteria, and the correlation coefficient was > 0.990. A method was developed and validated for mirtazepine in human plasma and was used for a study of eleven patients taking daily doses of 30-45mg [110]. A Chiralpak AD column with an internal standard of imipramine and UV detection at 290 nm provided a suitable method for the range of 10 to 100 ng/ml. The benzodiazepine tricyclic antidepressant, trimipramine, is extensively metabolised in rabbits, dogs and humans, forming three major metabolites by desmethylation and hydroxylation [111]. A method was developed which combined a reversed-phase separation on Chiralcel OD-R with mixed-mode disc solid phase extraction and UV detection at 210 nm. The method was found to be linear over the range 15-250 ng/ml. The desmethyltrimipramine enantiomers could be simultaneously separated under the same conditions, but the other two metabolites required a different mobile phase to effect a satisfactory resolution (Fig. 5.11).

D

I :

t~

0

o r

/i

A

Fa t~ H

Fig. 5.11. Typicalchromatogramsfor blank serum (left) and serum spikedwith 50 ng/ml of trimipramine and its metabolites on Chiralcel OD-R. A: unresolvedracemeic 2-hydroxydesmethyltrimipramine,B, C: Partially resolved 2-hydroxytrimipramine enantiomers, D: IS, E, F: desmethyltrimipramine enantiomers, G: Rtrimipramine, H: S-trimipramine [111]. References pp. 180-184

164

Chapter 5

Mianserin and its desmethyl metabolite were assayed sequentially on a Chiral-AGP CSP in slightly differing mobile phases after first separating the two compounds achirally on an analytical silica column [112]. The fractions were collected, evaporated to dryness and injected in mobile phase (acetonitrile and phosphate buffer).

5.3.15 Antiemetics Ondansetron and its three hydroxy metabolites in human serum can be separated enantioselectively on a Chiralcel OD-R column in under 22 minutes [113]. SPE extraction and detection at 210 nm gave an LOQ of 7 ng/ml. A method validated for the primary metabolite of dolasetron mesylate in human plasma used an Ultron ES-OVM column, ammonium acetate with acetonitrile as organic modifier and fluorescence detection [ 114].

5.3.16 Antifungals An evaluation of the chiral inversion of a development antifungal compound, SCH 56592, in animals and humans was carried out [115]. The samples were analysed on a Chiralcel OD column in normal phase mode with fluorescence detection at 270 nm. The RRRS and RRSR isomers were subsequently also analysed using a Chiralpak AD with a hexane-ethanol-diethylamine mobile phase using the same detection method. The LOQ in animal serum was 0.1 txg/ml and the method had an average RSD of just 6%. An earlier triazole antifungal compound, SCH39304, had used Cyclobond I 2000 in water-methanol-acetonitrile combined with UV detection at 205 nm [116].

5.3.17 Antihistamines The enantiomers of chlorpheniramine, an effective and low-cost H~-antihistamine, and its two metabolites, monodesmethylchlorpheniramine and didesmethylchlorpheniramine, were separately determined using a coupled achiral-chiral assay [117]. After discarding the early eluting interferences from a cyanopropyl column to waste, the parent drug and internal standard were automatically switched to the Chiralcel AD chiral column. The same procedure was used for the analysis of the metabolites. No mobile phase composition was found that would enable the analysis of all analytes in one run with reasonable retention times. The method avoided the long retention times and shortened column life associated with direct injection of the extracted biological matrix onto the chiral CSE The cyanopropyl column was found to deteriorate after 1000 injections and was replaced. A human pharmacokinetic study involving administration of 8 mg of racemic chlorpheniramine was carried out using this assay. More recently, a

Chiral bioanalysis

165

direct method was developed to determine lower plasma concentrations of chlorpheniramine as well as several of the metabolites by single ion MS detection [118]. A Cyclobond I 2000 column was used and a reversed-phase solvent of 85:7.5:7.5 aqueous diethylamine (0.25%, pH 4.4)-acetonitrile-methanol. The method was validated and applied to samples from two clinical studies. The second generation H1 histamine receptor agonist, cetirizine, was studied in rat plasma using a Chiral-AGP column with UV detection at 230 nm [119]. The method, which covered the concentration range of 2.5-200 ~g/ml in plasma, did not detect any pharmacokinetic differences between the two enantiomers. Terfenadine can be determined using an Ultron ES-OVM column using a mobile phase of phosphate buffer and 2-propanol with fluorescence detection [120]. Gradient elution was used to separate the parent.

5.3.18 Antihyperlipoproteinemics R-(-)-Carnitine plays an important role in fatty acid metabolism, by facilitating their transport a cross mitochondrial membranes. Conversely, its (S)-isomer has been found to have considerable toxic effects due to competitive inhibition of carnitine acetyltransferase, causing depletion of the levels of the (R)-enantiomer [ 121 ]. It is also a pancreatic secretion stimulant. Carnitine deficiencies, as exemplified in disease states such as acute and chronic myocardium ischaemia, are therefore treated with pure (R)carnitine and its esters. In the past, enantioselective methods for the determination of pharmaceutical purity and for PK studies were difficult and slow, involving derivitisation with enantiopure reagents. The first direct method was recently reported [122] on a Chirobiotic T column. A mobile phase of 90:10 ethanol-0.05 M ammonium acetate was used with evaporative light scattering detection (ELSD) to give a 20 minute assay time. For carnitine and all of its O-acetyl derivatives, the (R) isomer eluted before the (S) isomer while the retention of the derivatives decreased with increasing acetyl chain length. The effect of the organic modifier was also studied, comparing the effect of 2-propanol, ethanol and methanol and it was found that a 3 5 : 3 5 : 3 0 mobile phase of ethanol-methanol-0.05 M ammonium acetate gave higher resolution for some derivatives. The esters, amide or nitrile C1 terminated carnitine derivatives, however, did not separate on this column, suggesting that a free carboxyl group plays a key role in enantioselectivity for these molecules. Fluvastatin inhibits the rate-limiting enzyme in cholesterol biosynthesis and has two asymmetric carbons in the heptenoic side chain: it is administered as the 3S, 5R and 3R, 5S racemic mixture. An on-line SPE system was used for sample clean-up which provided recoveries of greater than 90% [ 123]. Since fluvastatin is highly protein bound, lower flow rates during extraction were required to achieve the optimum recovery, especially important because it was thought that the 3S, 5R binds more strongly to plasma proteins. The samples were then injected on to a Chiralcel OD-R CSP and eluted in a mobile phase of acetonitrile-phosphate buffer (pH 2.5) at 10~ For optimum detection, a photochemical reactor (Beam Boost, ICT, Vienna) was placed in between References pp. 180-184

166

Chapter 5

the chiral column and the UV detector. Irradiation over a 10 m Teflon coil produced a 10-fold increase in sensitivity.

5.3.19 Antihypertensives Terazosin is a selective e~-adrenoreceptor anatagonist used as an antihypertensive vasodilator. A method for the bioanalysis of terazosin was developed [124] using ESIMS. Sensitive quantification of both enantiomers to 62.5 pg/ml in human plasma resulted. The column used was a narrow bore Chiralpak AD (100 x 2.1 mm) with a hexane-2-propanol-diethylamine mobile phase with a flow rate of 0.15 ml/min. A postcolumn solvent addition of 2-propanol/ammonium acetate at 0.3 ml/min was used to overcome the problems of interfacing normal phase eluents with LC-MS. Although the internal standard, prazosin, did not fully separate from the first eluting terazosin enantiomer, it did not contribute to the signal. A fluorescence detection method was also developed using a standard size column, but in this case the same internal standard could not be used, and it did not provide sufficient sensitivity for the lower level enantiomeric differences during elimination. A Chiralcel OJ column was used to separate the enantiomers of the dihydropyridine structure, isradipine, in human serum [ 125]. A mobile phase of hexane and 2-propanol, a column temperature of 39~ and UV detection at 240nm were used. For quantification, two fractions from 9 to 16 minutes and 16 to 22 minutes were collected from the Chiralcel OJ column, evaporated to dryness, reconstituted in toluene and injected onto an achiral GC column. This provided the required LOQ of 0.26 ng/ml using nitrogen selective detection. The arylethanolamine structure, bufuralol, and its l'-oxidised metabolites were separated on an Ultron ES-OVM column using gradient elution with ammonium acetate and acetonitrile [ 126]. Both UV and NMR were used for detection. Possible in vivo racemisation of amlodipine was studied by individual administration of each of the enantiomers and of the racemate [ 127]. To obtain the optimum sensitivity for the drug of 0.1 ng/ml, a column switching technique was used, using firstly a semipreparative Chiral-AGP column at 30~ with a mobile phase of pH 4.5 acetate buffer. Each enantiomer was then switched from the CSP onto a C8 column for quantification. Chloroamlodipine was used as internal standard. A coupled column method has also been developed for amlodipine [128]. The enantiomers were separated on a Chiral-AGP column, then trapped and compressed on a phenyl guard column, and quantified by elution from an analytical phenyl column, adding EDTA and TBAHS to assist amperometric detection at + 0.95 V.

5.3.20 Antiinflammatory drugs Ibuprofen, a non-steroidal anti-inflammatory drug, is generally administered as the racemate. In vivo chiral inversion of the drug from the inactive R-enantiomer to the active S-isomer is well known [129] and leads to a majority of dextrorotatory

Chiral bioanalysis

167

metabolites, no matter whether racemic or single enantiomer ibuprofen was administered. A sequential achiral-chiral LC method was developed for the stereospecific analysis of the two major urinary metabolites of ibuprofen, hydroxyibuprofen and carboxyibuprofen, the latter having two chiral centres [ 130]. Following the assay of the parent drug on an achiral silica column with a hexane-ethanol mobile phase, column eluates containing the two metabolites were separately collected, and concentrated by evaporation to dryness and reconstitution in the chiral mobile phase. Chiralcel AD in hexane/ethanol with trifluoroacetic acid provided a separation of all six isomers within 50 minutes with detection at 220 nm. A new study looking at the individual enantiomer concentrations following topical administration of ketoprofen, a potent non-steroidal anti-inflammatory drug, required the development of a new assay to enable the much lower levels of detection required. An LOQ of 0.05 ng/ml was reached [ 131 ], compared with the 25-50 ng/ml achieved in previous methods [ 132] developed for oral administration. L C - M S - M S (ESI) was used to monitor the ketoprofen isomers eluting from the ~r-acceptor Chirex 3005 column under reversed-phase conditions (95:5 methanol-ammonium acetate, pH3.5) (Fig. 5.12). Although a better resolution was also available on a Chiralpak AD column it used normal phase conditions and a buffered mobile phase was preferred as it was more easily compatible with HPLC-MS. Because the developed method was linear over a very wide range of concentrations (0.05 to 2500 ng/ml) it was suitable for the analysis of both topical and oral samples without changing conditions. Automated sample preparation and an analysis time of just 6.5 minutes allowed the assay of 96 samples in less than 10.5 hours. As a result, the assay was used successfully to support two clinical trials involving several hundred plasma samples. A method for ketoprofen on Chiralpak AD was used for pharmacokinetic studies in rat and human, using UV detection at 254 nm [133]. The quantification limit was 0.25 ~g/ml from 1 ml human plasma samples, or 0.1 ml rat plasma. A further method for ketoprofen was developed using a Chiral-HSA column in combination with an on-line restricted access sample preparation column [134]. This provided the capability of direct injection of plasma samples without prior treatment. Elution from the pre-column in 0.01 M phosphate buffer containing 6% 2-propanol enabled direct transfer of the analytes on to the CSP, which used the same mobile phase with the addition of 5 mM octanoic acid. Flurbiprofen plasma concentrations were measured down to 10-30 fmol (injected) using a Chiralcel OD-R column [135]. This did, however, require derivatisation of the drug with 4-[NN-dimethylamino)-sulphonyl]-7-piperazino-2,1,3-benzoxadiazole (DBD-PZ), or its N-hydrazinoformyl derivative over a 5 hour incubation time to introduce a fluorophore. The resulting derivative was chromatographed on an ODS column, the flurbiprofen peak collected, evaporated to dryness and injected on to the cellulosic CSE The pharmacokinetics and metabolic chiral inversion of the enantiomers of tiaprofenic acid were investigated in vivo in rats [136]. The Chiral-AGP method determined that, following administration of the R- isomer, there was a time delay before the (S)- isomer was formed; after 2 hours, the concentration of the (S)enantiomer exceeded that of the (R)- isomer. References pp. 180-184

168

Chapter 5

(A) Plasma spiked with SIL kt 100t 2551209 ~

625

.__

~ 91 ~ 259/213

; ...........

S ) , L(R -

(S)-kt 66,555

_,,. . . . . . .

(B) Plasma spiked with 0.1 ng/mL kt and SIL kt 1O0t 255/209 755

o~ E f,,.

.

.

.

>

cr

1

.

.

83,840

259/213

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

"-..3 0

100" 8.34e3

0

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

9 ....

~..

, .r ~ . . . . .

" J';:""

i " ""

~ i:;"

";

5: ....

I:':-':;"

I " " "";"U"

" ' :'~ : "

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

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

"; ; ' i

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

100" 5.16e3

'"

'W . . . .

,

qD

%

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

0.00

....

2.50

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

5.00

7.50

10,00

12.50

15.00

....

r ....

17.50

|;---~---~i

20,00

:~-,i---

22.50

~v;-;;T"tll~

25.00

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].

c~

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

..q

o2

~g

No2

CI ~N02 Cl COOC2H 5 CH3OCH2CH2OOC~COOCH ((

CH3OOC~-~COOCH(CH3)2 CH3OOC~COOC2H5 CH300C NC" "N" "CH3 H3C" "N" "CH3 H3C" "N" "CH3 H H H ni Ivadipine nitrendipine felodipine ~u

N~

c.~ii~,~::i H

(-3

H

nimodipine

[ ~ u No2

'c"~"-"S"-~c,~II:Z~::I_O..jO

manidiptne

H3C" "N," "CH3

H

benidipine

~.HN02

CH3OOC,~COOCH2CH <(

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

nisoldipine

Fig. 5.14. Structures of typical dihydropyridine calcium antagonists. (The chiral centre is marked by an asterisk). [Y Tokuma and H Noguchi, J Chromat A, 694 (1995) 181].

q r~

Chiral bioanalysis

175

extraction and column switching technique was also developed to ensure removal of plasma interferences. The cycle time for the assay was 38 minutes with a resolution of 1.2 and sensitivity of 2.5 ng/ml. 5.3.30 Cholinesterase inhibitors

Donepezil is administered as a racemate for the treatment of Alzheimer's disease and it is known that the enantiomers have differing extents of inhibition against acetylcholinesterase in vivo and in vitro. An avidin protein column, Bioptick AV-1 (150x 2.1 mm), was used, eluting the isomers with formic acid and methanol [164]. ESI/MS/MS detection provided a linear range of 0.0206 to 50.0 ng/ml in plasma and the method was used for a human, oral administration study that showed higher plasma concentrations of the S-enantiomer were achieved compared with the R-enantiomer. 5.3.31 CNS Stimulants

The enantiomers of amphetamine and methamphetamine in urine were simultaneously separated on a Chiralcel OD-RH (150 x 2 mm) column following derivatisation of the compounds with 4-(4,5-diphenyl-lH-imidazol-2-yl)benzoyl chloride [165]. Acetonitrile and 0.1 M sodium hexafluorophosphate at 0.1 ml/min were used as the mobile phase and detection carried out by fluorescence (330 nm excitation, 440 nm emission). An LOQ for R and S-methamphetamine of 5 and 4 fmol, respectively, was obtained, while for both isomers of amphetamine, the LOQ was 1 fmol. A review of alternative methods using both HPLC and GLC has been published [166] and showed the simultaneous separation of the enantiomers of ephedrine, amphetamine, metamphetamine and pseudoephedrine on a chiral capillary GLC column, Chiraldex G-PN column. A comparison of HPLC and CE for this separation was also made. It was also noted that the use of the fluorescent tagging reagent, 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC) with amphetamine and Cyclobond I 2000 DMP in reversed-phase mode provided an increase in sensitivity of three orders of magnitude. An earlier study [167] studied the kinetics of methamphetamine, amphetamine and the hydroxylated metabolites p-hydroxymethamphetamine and p-hydroxyamphetamine in rat urine using an indirect method (derivatisation with (-)-l-(9-fluoroenyl)ethyl chloroformate, FLEC, a reagent that reacts with primary amines at room temperature to produce stable, fluorescent diastereomers). The detection limit was 5 ng/ml. Marfey's reagent has been used as an alternative derivatisation method that provides LOQ of 0.16 and 0.4 p~g/ml for methaphetamine and amphetamine, respectively [168]. The L enantiomers eluted before the D in each case. A method to distinguish the intake of (-)-deprenylin Parkinsonian patients from metamphetamine abuse was developed [169]. (S)-( + ) amphetamine and (S)-( + )-methamphetamine could be detected using an Ultron ES column in abusers of amphetamine whereas only R-(-) forms could be detected as metabolites in Parkinsonian patients. GC-MS was used for confirmation of identities. The enantiomers of amphetamine and norephedrine, but not of metamphetamine, were reported to separate on a Crownpak References pp. 180-184

176

Chapter 5

C R ( + ) column. D-amphetamine could be detected down to 0.1% in metamphetamine in forensic samples [ 170]. A study of D,L-threo-methylphenidate has been carried out using a Chirobiotic V column [ 171 ]. The method was developed primarily for use with H P L C - M S - M S and a simple mobile phase of 100% methanol containing 0.05% ammonium acetate was used for APCI in the positive-ion mode. The calibration was linear from 0.129 to 25.09 ng/ml with an LOQ of 87 pg/ml for each enantiomer and a 150 x 4.6 mm column was used to provide a total assay time of just less than 8 minutes. It was noted that no column deterioration was seen after 2500 injections of extracted plasma. The pharmacokinetics of modafinil, a novel CNS stimulant having a sulphur atom at its chiral centre, have been studied in human plasma using a cyclodextrin based CSR ChiraDex [ 172]. Phosphate buffer and acetonitrile provided elution of the enantiomers at 19 and 22 minutes, without any interference from endogenous compounds. The method was chosen because of its ruggedness and the good batch-to-batch reproducibility of the columns (retention times were within 10% over three batches).

5.3.32 Gastroprokinetic agents Stereoselective kinetics were discovered in a study of cisapride in human healthy volunteers [173]. The analytical method developed for the study used a Chiralcel OJ column with a mobile phase of diethylamine-ethanol-hexane (1:70: 129) with UV detection and gave a linear range of 5 to 125 ng/ml for each enantiomer. Intra-day and inter-day RSD values were not greater than 15% at all concentrations over 12.5 ng/ml; below this it was 20%. Mosapride and its des-4-benzyl metabolite were successfully analysed on a ChiralAGP column fitted with a 1 cm AGP guard column [ 174]. Gradient elution was used in this case, using a combination of phosphate buffer, citric acid and methanol. Detection by fluorescence gave a linear range of 10 to 5000 ng/ml for the parent drug and 50 to 5000 ng/ml for the metabolite. The optimum pH was investigated and found to be 4.5 for mosapride and 5 for the metabolite.

5.3.33 Hallucinogenics Another common drug of abuse, 3,4-methylenedioxy-N-methylamphetamine (MDMA, ecstasy), widely reported to cause an increasing number of deaths and acute clinical toxicological problems, also shows enantioselective toxicity. Moreover, illicit tablets often contain other phenylethylamines such as methylenedioxyamphetamine (MDA). S( + ) MDMA is a more potent neurotoxin than the R-(-) isomer, while both MDA isomers cause long term serotonin neurotoxicity. These compounds are detectable for 24 hours in blood and for 2-4 days in urine after single dose ingestion. A method that enabled the simultaneous enantioseparation of MDMA, MDA and two other derivatives was reported [ 175] on Cyclobond 12000 and Cyclobond 1 2000 RSP using fluorescence detection and reversed-phase conditions (phosphate buffer with 5% acetonitrile) (Fig. 5.15). Of the two columns, the RSP was found to be more rugged than the native

Chiral bioanalysis

177

2

3

'I . . . . . . . . . . . .

15.0

4

's . . . . . . . .

20.0

~ .......

25.0

-

....... !

:

30.0

,-

35.0

Fig. 5.15. Stereoselectiveseparation of: 1,2: MDA; 3,4: MDMA, 5,6: MDEA and 7,8: MBDB on Cyclobond 1 2000 [175]. cyclodextrin column, partly because of the high pH used. The assay range was 50 and 3000 ng/ml with an LOD of 11-24 ng/ml and LOQ of 37-80 ng/ml.

5.3.34 HIV protease inhibitors The pharmacokinetics of a candidate pyrone HIV protease inhibitor was studied in rat, dog and human [ 176]. After protein precipitation, the samples were applied to an SPE cartridge and eluted with hexane and ethanol. A Whelk-01 and a mobile phase of hexane/ethanol containing 0.1% acetic acid was used for the separation. Time profiles of plasma enantiomeric ratios were found to be species dependant for single doses but not for multiple dose administration.

5.3.35 Natriuretics A new natriuretic hormone, 2,7,8-trimethyl-2-([~-carboxyethyl)-6-hydroxy chroman, LLU-oL, thought to originate from the oxidative metabolism of ~-tocopherol, required a highly sensitive method to determine the very low levels found in urine. Using fluorescent labelling of the carbonyl with DBD-PZ, followed by acylation of the phenolic group (thought to be responsible for the degradation of the derivative), a method was developed using a column switching technique [177]. This isolated the enantiomers from a phenyl column and transferred them on to a Chiralcel OD-RH column, using a simple mobile phase of methanol and acetonitrile (95:5) that provided a resolution of 1.4. For the isolation and identification of the isomers, a w-acceptor, Sumichiral OA-3100, column was used in the reversed-phase mode.

5.3.36 Leukotriene antagonists Montelukast is a potent and selective leukotriene D4 (cysLT1) receptor agonist, studied for the treatment of chronic asthma [178]. Using protein precipitation and column References pp. 180-184

178

Chapter 5

switching, the enantiomers were resolved on Chiral-AGP with detection by fluorescence. The column switching technique employed a Chromaspher Biomatrix column for on-line extraction. The method was used to evaluate possible bioinversion of montelukast in plasma, but none was detected.

5.3.37 Mucolytics A fluorescent chiral tagging reagent was used to determine the enantiomers of erdosteine and its chiral metabolite in plasma on an ODS column [179]. The sensitivity of the method was improved by labelling one thiol and two carboxy groups of the analyte with 4-(N,N-dimethylaminosulphonyl)-7-fluoro-2,1,3-benzoxadiazole (DBD-F) and its 7-(3-aminopyrrolidine-l-yl) derivative (DBD-APy), respectively. The detection limits achieved were 0.37 pmol for erdosteine and 0.22 pmol for its metabolite.

5.3.38 Radiosensitisers The in vivo chiral inversion of a radiosensitiser, PD 146923, and its pro-drug, CI-1010 was studied [ 180]. A Chiralpak AS column in normal phase was found to provide the best separation for these compounds. The biggest challenge for this assay was the development of an SPE method, required for compatibility with the mobile phase but difficult because of the high reactivity of both compounds. The developed method, however, provided an LOQ of 200 ng/ml for each enantiomer and 2% of the inverted isomer could be detected.

5.3.39 Sedative/hypnotics Bromoisovalerylurea is a sedative/hypnotic given orally as a racemic mixture and frequently used in overdose in suicide cases. Serum samples from 16 overdose cases were analysed a Chirobiotic V, Cyclobond I 2000 or a urea derivative bonded phase [181 ] and found to contain amounts of the ( + ) isomer in all cases. In testing serum and saliva [ 182] only one overdose case had equivalent concentrations of the two isomers. The conclusion was that adsorption from the gastrointestinal tract was nonstereoselective, but that elimination was stereoselective. A rapid method for Thalidomide was developed on Chiral-AGP, achieving optimum separation at pH 7 (ammonium acetate) containing 0.3% THF [183]. Peak tailing was found to be directly related to changes in pH and to THF concentration that were investigated in this study. The method was validated and used for an in vivo study in New Zealand male rabbits.

5.3.40 Serotonin uptake inhibitors Fluoxetine (Prozac) is a potent serotonin reuptake inhibitor used for the treatment of major depression. Both fluoxetin and its demethylated metabolite, norfluoxetine, are

Chiral bioanalysis

179

chiral, with the S-enantiomers in each case having the greater potency. Following elution with dichloromethane-isopropanol-ammonia from SPE tubes, evaporation and reconstitution, a Cyclobond I 2000 Ac column was used with a mobile phase of methanol and 0.3% triethylamine buffer at pH 5.6 to determine plasma, urine and vitreous humour concentrations from rabbit studies [184]. After 24 hours, only Rnorfluoxetine was detectable in plasma: the vitreous humour samples had no detectable concentrations at any time.

5.3.41 Thyromimetic agents An HPLC-MS-MS study of a development compound, an ester, CGS 26214, being evaluated for its cholesterol lowering activity in rats, dogs and monkeys was carried out [ 185]. Since clinical doses were of 1 mg or less, the sensitivity of the assay was critical. The enantiomers of the compound appear to have similar biological activities and rapidly hydrolyse to the free acid in vivo. Sample preparation involved conversion of all of the analytes to the free acid followed by C18 SPE extraction. Evaporation and reconstitution in 20% methanolic water, analysis on a polymeric C18 column and column switching of the 7 to 9.5 minute window onto a microbore Chiral-AGP column completed the method. By using a mobile phase of 0.03% ammonium acetate containing 4% n-propanol and a flow rate of 40 pA/min, direct usage of the tandem (ESI) MS system was possible. Negative ion mode was required to obtain sufficient sensitivity and sample preparation was found to be critical for the minimisation of ion suppression and maximisation of column lifetime. Although the method was successfully applied to generate plasma concentration-time curves in humans, it was noted that the method was limited in sample throughput capacity.

5.3.42 Vasodilators (cerebral) Concentration-time profiles of nimodipine were developed following intravenous and oral administration [ 186]. Total parent drug was determined using achiral reversed phase HPLC and enantiomeric ratios then measured by utilising an Ultron ES-OVM column in a mobile phase of ethanol/phosphate buffer. The method was linear over the range of 5 to 200 ng/ml of nimodipine.

5.3.43 Vitamins A combination of two CSPs was utilised to determine the quantities of RRR- and SRRoL-tocopherols and their quinones in rat plasma and adrenal glands [ 187]. One Chiralcel OD-H and one Sumichiral OA 4100 were connected in series and a mobile phase of hexane-2-propanol used with a flow rate of 0.3 ml/min. The concentrations of the SRRoL-tocopherol was higher than the RRR- enantiomer in the first 6 hours after oral administration, but they became similar after 24 hours. In vivo formation of oLtocopherol isomers had no effect on the enantiomeric ratio present. References pp. 180-184

Chapter 5

180

5.4 CONCLUSIONS From the data published in the period covered by this review, it is clear that greater demands have been placed on the chromatographer for higher sensitivity in recent years and that much of the method development process is directed towards improving this as far as possible. There is a preponderance of fluorimetric methods, while MS has become the preferred choice of detection, making reversed-phase and polar organic or ionic methods more attractive. Cellulosic and protein phases predominate, perhaps partly because they have been available for the longest time and were probably used in the discovery process for drugs that are now going through clinical trials. Of these, over 60% of the cellulosic and amylosic separations used normal phase methods. Protein phases have been used in bioanalysis as often as the cellulosic CSPs, but more often with UV or fluorescence detection rather than with MS because of the prevalent use of phosphate buffer with small concentrations of organic modifier rather than more volatile choices. Cyclodextrin based columns, used mostly in reversed-phase mode are the next most commonly used, while the Chirobiotic CSPs have recently become valuable for their speed in the polar ionic mode. Since only normal phase conditions were used for the w-complex type CSPs, these were the least used of all phases in bioanalysis. It seems that speed, broad applicability and ease of sample preparation have been the main driving forces for chiral bioanalytical method development and it is anticipated that the use of HPLC-MS methods will increase considerably over the next few years.

5.5 REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

M.R. Islam, J.G. Mahdi and I.D. Bowen, Drug Safety, 17 (1997) 149. M. L~immerhofer and W. Lindner, Separation Methods in Drug Synthesis and Purification, 1 (2000) Chapter 9, K. Valk6 (Ed.). D.W. Armstrong, Y. Tang, S. Chen, Y. Zhou, C. Bagwill and J-R. Chen, Anal. Chem., 66 (1994) 1473. G.L. Reid, S. Chen, C.D. Chang and D.W. Armstrong, Trends in Anal. Chem., 12 (1993) 144. K. Tachibana and A. Ohnishi, J. Chromatogr. A, 906 (2001) 127. T. Alebic-Kolbah and A.E Zavitsanos, J. Chromatogr. A, 759 (1997) 65. P. Zavitsanos and T. Alebic-Kolbah, J. Chromatogr. A, 794 (1998) 45. T. Alebic-Kolbah and A.E Zavitsanos, J. Chromatogr. A, 759 (1997) 65. H. Stenhoff, A. Blomqvist and EO. Lagerstrom, J. Chromatogr. B, 734 (1999) 191. J.E. Paanakker, L. de Jong, J.M.S.L. Thio and H.J.M. Hal, J. Pharm. Biomed. Anal., 16 (1998) 981. R. Bakhtiar, L. Ramos and EL.S. Tse, Chirality, 13 (2001) 63. K.B. Joyce, A.E. Jones, R.J. Scott, R.A. Biddlecombe and S. Pleasance, Rapid Commun. Mass Spec., 12 (1998) 1899. R. Bakhtiar and EL.S. Lee, Rapid Commun. MS, 13 (1999) 2054. D.S. Richards, S.M. Davidson and R.M. Holt, J. Chromatogr. A, 746 (1996) 9. K. Fulde and A.W. Frahm, J. Chromatogr. A, 858 (1999) 333. K. Cabera and D. Ludba, J. Chromatogr. A, 666 (1999) 433. J. Hermansson and A. Grahn, J. Chromatogr. A, 687 (1994) 45. K. Balmer, EO. Lagerstrom, B.A. Persson and G. Schill, J. Chromatogr., 592 (1992) 331. J. Ducharme, C. Fernandez, E Gimenez and R. Farinotti, J. Chromatogr. B, 686 (1996) 65. M.R. Hadley, E Camilleri and A.J. Hutt, Electrophoresis, 21 (2000) 1953.

Chiral bioanalysis 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 58 59 60 61 62 63 64 65 66

181

K.W.Phinney, Anal. Chem, (2000) 205A. K.L. Williams, L.C. Sander and S.A. Wise, J. Pharm. & Biomed. Anal., 15 (1997) 1789. Y. Lui and D.A. Armstrong, in press. A. Medvedovici, P. Sandra, L. Toribio and E David, J. Chromatogr. A, 785 (1997) 159. G. Hesse and R. Hagel, Chromatographia, 6 (1973) 227. Y. Okamoto, M. Kawashima, K. Yamamoto and K. Hatada, Chem. Lett. (1984) 739. T. Shibata, T. Shei, H. Nishimura and K. Deguchi, Chromatographia, 24 (1986) 552. U. Vogt and P. Zugenmaier, Ber. Bunsenges Phys. Chem., 89 (1985) 1217. Y. Okamoto, M. Kawashima K. Yamamoto and K. Hatade, Chem. Lett. (1984) 739. K. Tachibana and O. Ohnishi, J. Chromatogr. A, 906 (2001) 127. A. Ishikawa and T. Shibata, J. Liq. Chromatogr., 16 (1993) 859. H.A. Aboul-Enain, L.I. Abou-Basha and S.A. Bakr, Chirality, 8 (1996) 1011. K.V. Penmetsa, C.D. Reddick, S.W. Fink, B.L. Kleintop, G.C. DiDonato, K.J. Volk and S.E. Klohr, J. Liq. Chromatogr. & Rel. Technol., 23 (2000) 831. Chromtech Application Handbook # 19. J. Haginaka, J. Chromatogr. A, 906 (2001) 253. W.L.Hinze, Sep. Purif. Meth., 10 (1981) 159. D.W.Armstrong, W. DeMond, J. Chromatogr. Sci., 22 (1984) 411. Chirobiotic Handbook, Advanced Separation Technologies Inc, 4th Edition, 2002. A. Berthod, X. Chen, D.W. Armstrong, E Gasparrini, C. Villani and A. Corotti, Anal. Chem., 72 (2000) 1767. C.J. Welch, J. Chromatogr. A, 666 (1994) 3. W.H. Pirkle and J.M. Finn, J. Org. Chem., 46 (1981) 2935. K.H. Kim, H.J. Kim, J.H. Kim and S.D. Shin, J. Chromatogr. B, Biomed. Appls., 751 (2001) 69. H.Y.Aboul-Enein and V. Serignese, Biomed. Chromatogr., 13 (1999) 520. L.I. Abou-Basha and H.Y. Aboul-Enein, Biomed. Chromat., 10 (1996) 69. K.B. Joyce, A.E. Jones, R.J. Scott, R.A. Biddlecombe and S. Pleasance, Rapid Commun. Mass Spec., 12 (1998) 1899. K.M. Fried, P. Koch and I.W. Wainer, Chirality, 10 (1998) 484. H.Y.Aboul-Enein and V. Serignese, Chirality, 7 (1995) 158. L.I. Aboul-Basha and H.Y. Aboul-Enein, Biomed. Chromatogr., 10 (1996) 69. J.J. Butter, B.T.J. van den Berg, E.J.G. Portier, G. Kaizer and C.J. Boxtel, J. Liq. Chrom. Rel. Tech., 19 (1996) 993. K.H. Kim, H.J. Kim, J.S. Kang and W.C. Mar, J. Pharm. Biomed. Anal., 22 (2000) 377. V.L. Lanchote, P.S. Bonato, P.M. Cerqueira, V.A. Pereira and E.J. Cesarino, J. Chromatogr. B, Biomed. Appl., 738 (2000) 27. B. Mistry, J.L. Leslie and N.D. Eddington, J. Chromatogr. B, 758 (2001) 153. A. Ceccato, B. Toussaint, P. Chiap, Ph. Hubert and J. Crommen, J. Pharm. & Biomed. Anal., 15 (1997) 1365. T. Fornstedt, A.M. Hesselgren and M. Johansson, Chirality, 9 (1997) 329. J.M. Dakers, D.W. Boulton and J.P. Fawcett, J. Chromatogr. B, Biomed. Appl., 704 (1997) 215. H.L. Zhang, J.T. Stewart and M. Ujhelyi, J. Chromatogr. B, Biomed. Appl., 668 (1995) 309. C. Pham-Huy, B. Radenen, A. Sahui-Gnassi and J-R. Claude, J. Chromatogr. B, 665 (1995) 125. C. Bertucci, V. Andrisano, R. Gotti and V. Cavrini, Chromatographia, 52 (2000) 319. S. Okubo, E Mashige, Y. Hashimoto, K. Nakahara, H. Kanazawa, Y. Matsushima, T. Fukushima, Y. Huang and K. Imai, Chromatogr., 21 (2000) 43. D.G. Jin, T. Miyahara, T. Oe and T. Toyo'oka, Anal. Biochem., 269 (1999) 124. K. Petritis, A. Valleix, C. Elfakir and M. Dreux, J. Chromatogr. A, 913 (2001) 331. C. Pham-Huy, N. Chikhi-Chorfi, H. Galons, N. Sadeg, X. Laqueille, N. Aymard, E Massicot, J.-M. Warnet and J-R. Claude, J. Chromatogr. B, 700 (1997) 155. D.W. Boulton and C.L. DeVane, Chirality, 12 (2000) 681. H.R.Angelo, N. Beck and K. Kristensen, J. Chromatogr. B, Biomed. Appl., 724 (1999) 35. S. Rudaz and J.L. Veuthey, J. Pharm. Biomed. Anal., 14 (1996) 1271. M. Brunnenberg and K.A. Kovar, J. Chromatogr. B, Biomed. Appl., 751 (2001) 9.

182

Chapter 5

67 68 69 70 71 72

E. Ameyibor and J.T. Stewart, J. Chromatogr. B, 703 (1997) 273. M.A. Capanero, B. Calahorra, M. Valle, I.E Troconiz and J. Honorato, Chirality, 11 (1999) 272. A. Ceccato, P. Chiap. R Hubert and J. Crommen, J. Chromatogr. B, Biomed. Appl., 698 (1997) 161. A. Ceccato, E Vanderbist, J.Y. Papst and B. Streel, J. Chromatogr. B, Biomed. Appl., 748 (2000) 65. E Jamali, R. Lovlin, B.W. Corrigan, N.M. Davies and G. Aberg, Chirality, 11 (1999) 201. M.A. Campanero, A. Lopez-Ocariz, E. Garcia-Quetglas, B. Sadaba and J.R. Azanza, Chromatographia, 48 (1998) 203. I. Tsina, Y.L. Tam, A. Boyd, C. Rocha, I. Massey and T. Tarnowski, J. Pharm. Biomed. Anal., 15 (1996) 403. M. Sueyasu, K. Fufito, K. Makino, H. Shuto, Y. Kataoka and R. Oishi, J. Chromatogr. B, Biomed. Appl., 723 (1999) 307. D.J. Jones, K.T. Nguyen, M.J. McLeish, D.P. Crankshaw and D.J. Morgan, J. Chromatogr. B, Biomed. Appl., 675 (1996) 174. J.L. Huang, L.E. Mather and C.C. Duke, J. Chromatogr. B, Biomed. Appl., 673 (1995) 245. J.O. Svensson and L.L. Gustavsson, J. Chromatogr. B, Biomed. Appl., 678 (1996) 373. B. Mey, H. Paulus, E. Lamparter and G. Glaschke, Chirality, 10 (1998) 307. D. Zimmer, V. Muschalek and C. Muller, Rapid Comm. in Mass Spec., 14 (2000) 1425. J.J. Garcia, E Bolas-Fernandez and J.J. Torrado, J. Chromatogr. B, Biomed. Appl., 723 (1999) 265. EO. Paias, V.L. Lanchote, O.M. Takayanagui and RS. Bonato, Chirality, 9 (1997) 722. J. Liu and J.T. Stewart, J. Chromatogr. B, Biomed. Appl., 692 (1997) 141. D. Zhong and X. Chen, J. Chromatogr. B, 721 (1999) 67. C.M. deGaitani, V.L. Lanchote and RS. Bonao, J. Chromatogr. B, Biomed. Appl., 708 (1998) 177. R. Bortocan, V.L. Lanchote, E.J. Cesarino and RS. Bonato, J. Chromatogr. B, Biomed. Appl., 744 (2000) 299. E. Bransteterova and I.W. Wainer, J. Chromatogr. B, Biomed. Appl., 732 (1999) 395. T. Alebic-Kolbah and A.P. Zavitsanos, J. Chromatogr. A, 759 (1997) 65. S.M. Lankford and S.A. Bai, J. Chromat. B, Biomed. Appl., 663 (1995) 91. V.L.Lanchote, V.J. Santos, E.J. Cesarino, S.A.C. Dreossi, Y. Mere and S. Santos, Chirality, 11 (1999) 85. M. Shimizu, K. Takatori, L. Kajiwara and H. Ogata, Biol. Pharm. Bull., 21 (1998) 530. E Mangani,, G. Luck, C. Fraudeau and E. Verette, J. Chromatogr. A, 762 (1997) 235. T. Wikberg, T. Korkolainen and M. Karlsson, Chirality, 8 (1996) 511. V.L. Lanchote, RS. Bonato, E.J. Cesarinoa, Y.A. Mere, S.R. Cavani, J. Santos and C. Bertucci, Chirality, 9 (1997) 732. M. Matsuoka, K. Banno and T. Sato, J. Chromatogr. B, 676 (1996) 117. K.M. Rentsch, U. Gutteck-Amsler, R. Buehrer, K.E. Fattinger and D.J. Vonderschmitt, J. Chromatogr. B, Biomed. Appl., 742 (2000) 131. A.S. Prangle, T.A.G. Noctor and W.J. Lough, J. Pharm. Biomed. Anal., 16 (1998) 1205. K.R. Henne, A. Gaedigk, G. Gupta, J.S. Leeder and A.E. Rettie, J. Chromatogr. B, Biomed. Appl., 710 (1998) 143. H. Takahashi, T. Kashima, S. Kimura, N. Muramoto, H. Nakahata, S. Kubo, Y. Shimoyama, M. Kajiwara and H. Echizen, J. Chromatogr. B, Biomed. Appl., 701 (1997) 71. M. Britzi, M. Bialer, L. Arcavi, A. Shachbari, J. Kapitulnik and S. Kimron, Ther. Drug Monit., 22 (2000) 510. W.Z. Zhong and M.G. Williams, J. Chromatogr. A, 871 (2000) 201. C.D. Torchin, W.D. Yonekawa, I.M. Kapetanovic and H.J. Kupfemerg, J. Chromatogr. B, Biomed. Appl., 724 (1999) 101. A. Caccato, B. Boulanger, P. Chiap, P. Hubert and J. Crommen, J. Chromatogr. A, 819 (1998) 143. S. Eto, N. Tanaka, H. Noda and A. Noda, J. Chromatogr. B, Biomed. Appl., 677 (1996) 325. T. Naitoh, M. Kakiki, S. Kawaguchi, Y. Kagei and T. Horie, J. Chromatogr. B, 694 (1997) 153. B. Carlsson and B. Norlander, Chromatographia, 53 (2001) 266. B. Rochat, M. Amey and P. Baumann, Ther. Drug Monit., 17 (1995) 273. D. Haupt, J. Chromatogr. B, Biomed. Appl., 685 (1996) 299. B. Rochat, M. Amey, H. Van-Gelderen, B. Testa and R Baumann, Chirality, 7 (1995) 389.

73 74 75 76 77 78 79 80 81 82 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

Chiral bioanalysis 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150

183

Z. Shen, S. Wang and R. Bakhtiar, Rapid Commun. Mass Spectrom., 16 (2002) 332. S. Dodd, G.D. Burrows and T.R. Norman, J. Chromatogr. B, Biomed. Appl., 748 (2000) 439. J. Liu and J.T. Stewart, J. Chromatogr. B, 700(1997) 175. G. Tybring, K. Otani, S. Kanekop, K. Mihara, Y. Fukushima and L. Bertilsson, Ther. Drug Monit., 17 (1995) 516. J. Liu and J.T. Stewart, J. Chromatogr. B, Biomed. Appl., 684 (1997) 179. J.S. McElvain, V.J. Vandiver and L.S. Eichmeier, J. Pharm. Biomed. Anal., 15 (1997) 513. H. Kim, C.C. Lin, M. Laughlin, R. Lovey, A. Saksena, L. Heinmark and A.A. Noeir, Chirality, 12 (2000) 590. H. Kim and C.C. Lin. J. Pharm. Biomed. Anal., 13 (1995) 1415. B.T. Hiep, V. Khanh, N.K. Hung, A. Thuillier and E Gimenez, J. Chromatogr. B, 707 (1998) 235. K.M. Fried, A.E. Young, S.U. Yasuda and I.W. Wainer, J. Pharm. Biomed. Anal., 27 (2002) 479. S.O. Choi, S.H. Lee, H.S. Kong, E.J. Kim and H.Y.E Choo, J. Chromatogr. B, Biomed. Appl., 744 (2000) 201. A. Terhechte and G. Blaschke, J. Chromatogr. A, 694 (1995) 219. H. Jung, K. Jung, H.E Kleber; Adv. in Biochem. Eng. Biotech., Ed. A. Fiechter, Springer, Berlin, 1993. I. D'Acquarica, E Gasparrini, D. Misiti, C. Villani, A. Carotti, S. Cellamare and S. Muck, J. Chromatogr. A, 857 (1999) 145. H. Toreson and B.M. Eriksson, Chromatographia, 45 (1997) 29. A.E Zavitsanos and T. Alebic-Kolbah, J. Chrom. A, 794 (1998) 45. H.S. Rask, H.R. Angelo and H.R. Christensen, Chirality, 10 (1998) 808. S. Ikegawa, K. Matsuura, T. Sata, N. Isriyanthi, T. Niwa and S. Miyairi, J. Pharm. Biomed. Anal., 17 (1998) 1. J. Luksa, D. Josic, M. Kremser, Z. Kopitar and S. Milutinovic, J. Chromatogr. B, Biomed. Appl., 703 (1997) 185. M. Josefsson and B. Norlander, J. Pharm. Biomed. Anal., 15 (1996) 267. W.J. Wechter, D.G. Loughhead, R.J. Reischer, G.J. Van Giessen and D.G. Kaiser, Biochem. Biophys. Res. Commun., 61 (1974) 833. S.C. Tan, S.H.D. Jackson, C.G. Swift and A.J. Hutt, J. Chromatogr. B, 701 (1997) 53. T.H. Eichhold, R.E. Bailey, S.L. Tanguay and S.H. Hoke, J. Mass Spectrom., 35 (2000) 504. N. Rifai, M. Lafi, M. Sakamoto and T. Law, Ther. Drug Monit., 19 (1997) 175. R. Lovlin, M. Vakily and E Jamali, J. Chromatogr. B, Biomed. Appl., 679 (1996) 196. W.R.G. Baeyens, G. Van der Weken, J. Haustraete, H.Y. Aboul-Enain, S. Corveleyn, J.P. Remon, A.M. Garcia-Campana and P. Deprez, J. Chromatogr. A, 872 (2000) 153. T. Fukushima, T. Santa, H. Homma, S.M. A1-Kindy and K. Imai, Anal. Chem, 69 (1997) 1793. K. Erb, R. Brugger, K. Williams and G. Geisslinger, Chirality, 11 (1999) 103, and J. Chromatogr. B, 675 (1996) 77. J. Zagrobelny and B.K. Matuszewski, Enantiomer., 2 (1997) 37. A. Slovakova, X. Freiin von Maltzan, B.K. Patel, A.E Drake and A.J. Hutt, Chromatographia, 48 (1998) 369. M. Breda, S. Sarati, G. Basileo and P. Dostert, Chirality, 9 (1997) 133. J. Zagrobelny, B.K. Matuszewski, W.E Kline and S.H. Vincent, J. Pharm. Biomed. Anal., 17 (1998) 1057. M.M. Hefnawy, J. Liq. Chrom. & Rel. Technol., 23 (2000) 781. M.M. Hefnawy and J.T. Stewart, J. Liq. Chrom. & Rel. Technol., 23 (2000) 791. S.A. Corlett and H. Chrystyn, J. Chromatogr. B, Biomed. Appl., 682 (1996) 337. V. Capka, Y. Xu and Y.H. Chen, J. Pharm. Biomed. Anal., 21 (1999) 507. M.E. DePuy, J.L. Demetraides, D.G. Musson and J.D. Rogers, J. Chromatogr. B, 700 (1997) 165. V. Ascalone, M. Ripamonti and B. Malavasi, J. Chromatogr. B Biomed. Appl., 67 (1996) 95. M.G. Quaglia, N. Desideri, E. Bossu, R. Sgro and C. Conti, Chirality, 11 (1999) 495. N. Masubuchi, H. Yamazaki and M. Tanaka, Chirality, 10 (1998) 747. M. Tanaka and H. Yamazaki, Anal. Chem, 68 (1996) 1513. A. Karlsson and S. Hermansson, Chromatographia, 44 (1997) 10.

184

Chapter 5

151 152

K. Borner, E. Borner and H. Lode, Chromatographia, 47 (1998) 171. H. Kanazawa, Y. Kunito, Y. Matsushima, S. Okubo and E Mashige, J. Chromatogr. A, 871 (2000) 181. H. Kanazawa, Y. Konishi, Y. Matsushima and T. Takahashi, J. Chromatogr. A, 797 (1998) 227. S.A. Wood, S.N. Pegg, R.J. Simmonds and D. Stevenson, J. Pharm. and Biomed. Anal, 14 (1996) 1591. J. Liu and J.T. Stewart, Anal. Lett., 30 (1997) 1555. H. Ichihara, T. Fukishima and K. Imai, Anal. Biochem., 269 (1999) 379. O.O. Omole, D.R. Brocks, G. Nappert, J.M. Naylor and G.S.A. Zello, J. Chromatogr. B, Biomed. Appl., 727 (1999) 23. M.S. Rashed, M. A1-Amoudi and H.Y. Aboul-Enein, Biomed. Chromatogr., 14 (2000) 317. M.S. Rashed, H.Y. Aboul-Enein, M. A1-Amoudi, M. Jakob, L.Y. A1-Ahaideb, A. Abbad, S. Shabib and E. A1-Jishi, Biomed. Chromatogr., 16 (2002) 191. M.S. Rashed, L.Y. A1-Ahaideb, H.Y. Aboul-Enein, M. A1-Amoudi and M. Jakob, Clin. Chem., 47 (1002) 2124. D.J. Triggle and S. Padmanabhan, Chemtracts: Org. Chem., 8 (1995) 191. O.E Kleidernigg and C.O. Kappe, Tetrahedron.: Asymmetry, 8 (1997) 2057. T. Uno, T. Ohkubo and K. Sugawara, J. Chromatogr. B, 698 (1997) 181. K. Matsui, Y. Oda, H. Nakata and T. Yoshimura, J. Chromatogr. B, 729 (1999) 147. O.Y. A1-Dirbashi, M. Wada, N. Kuroda, M. Takahashi and K. Nakashima, J. Forensic Science, 45 (2000) 708. D.W. Armstrong, K.L. Rundlett and U.B. Nair, Curr. Seps., 15 (1996) 57. J. Sukbuntherng, A. Hutchaleelaha, H-H. Chow and M. Mayersohn, J. Anal. Toxicol., 19 (1995) 139. B.S. Foster, D.D. Gilbert, A. Hutchaleelaha and M. Mayersohn, J. Anal. Toxicol., 22 (1998) 265. Y. Maina and Y. Suzuki, Jpn. J. Toxicol., 42 (1996) 433. Y. Makino, S. Ohta and M. Hirobe, Forensic Sci. Int., 78 (1996) 65. L. Ramos, R. Bakhtiar, T. Majumdar, M. Hayes and E Tse, Rapid Commun. Mass Spectrom., 13 (1999) 2054. S.H. Gorman, J. Chromatogr. B, 730 (1999) 1. Z. Desta, N.V. Soukhova, A. Morocho, J. Park, S.K. Mahal and D.A. Flockhart, J. Chromatogr. B, Biomed. Appl., 744 (2000) 263. I. Yokoyama, Y. Mizuki, T. Yamaguchi and T. Fujii, J. Pharm. Biomed. Anal., 15 (1997) 1527. E Sadeghipour and J-L. Veuthey, Chromatographia, 47 (1998) 285. W.Z. Zhong, M.G. Williams, M.T. Borin and G.E. Padbury, Chirality, 10 (1998) 210. A. Hattori, T. Fukushima and K. Imai, Anal. Biochem., 281 (2000) 209. L. Liu, H. Cheng, J.J. Zhao and J.D. Roberts, J. Pharm. Biomed. Anal., 15 (1997) 631. N. Muramatsu, T. Toyo'oka, K. Yamaguchi and S. Kobayashi, J. Chromatogr. B, Biomed. Appl., 719 (1998) 177. J.R. Kagel, D.T. Rossi, K.L. Hoffman, B. Leja and C.D. Lathia, J. Pharm. Biomed. Anal., 21 (1999) 527. T. Nishikawa, Y. Kamijo, R. Kondo, H. Sugie, K. Kurihara, T. Okuda, N. Matsumoto, Y. Okada and H. Ohtani, J. Anal. Toxicol., 24 (2000) 691. T. Nishikawa, H. Ohtani, Y. Kamijo, Y. Ohtani, R. Kondo, H. Takeuchi and T. Okuda, Biomed. Chromatogr., 14 (2000) 243. C. Alvarez, J.A. Sanchez-Brunete, S. Torrado-Santiago, R. Cadorniga and J.J. Torrado, Chromatographia, 53 (2000) 455. L. Yee, S.H.Y. Wong and V.A. Skrinska, J. Anal. Toxicol., 24 (2000) 651. T.K. Majumdar, L.L. Martin, D. Melamed and EL.S. Tse, J. Pharm. Biomed. Anal., 23 (2000) 745. H. Wanner-Olsen, EB. Gaarskaer, E.O. Mikkelsen, E Jakobsen and B. Voldby, Chirality, 12 (2000) 660. C. Kiyose, K. Kaneko, R. Muramatsu, T. Ueda and O. Ogarashi, Lipids, 34 (1999) 415. R. Bakhtiar and E L. S. Lee, Rapid Commun. in MS, 14 (2000) 1128.

153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 Elsevier Science B.V. All rights reserved

185

CHAPTER 6

Method development in reversed-phase chromatography U w e Dieter Neue, Eric S. Grumbach, Jeff R. Mazzeo, KimVan Tran and Diane M. Wagrowski-Diehl Waters Corporation, 34 Maple St, Milford, MA, USA

6.1 I N T R O D U C T I O N The efficient development of new reversed-phase HPLC methods has been a topic of discussion since the early times of the technology. The reader may recall with awe the magical suggestion of the experienced chromatographer, who looked at the structure of the compounds that one needed to analyze and then quickly suggested "35% methanol, phosphate buffer pH 2.5". This suggestion was often a very good beginning, and brought the method development process to a rapid start. This was followed by a lengthy tinkering with the mobile phase composition and the pH until finally a satisfactory HPLC method emerged. However, significantly more rational method development strategies are possible, and have been developed during the course of HPLC history. In recent years, the specificity of the detection with mass spectrometers has simplified the method development process: a resolution of all compounds in the chromatographic dimension is not needed anymore. However, there still remains the question of ion suppression due to matrix interferences, and some method development is still needed. In this article, we will primarily discuss the classical HPLC method development, but we will also address the situation where the target is the separation of a few analytes from matrix interferences, such as the analysis of parent drug and metabolites in plasma or urine samples. The classical method development approach was based on the selection of mixtures of three solvents, methanol, acetonitrile, tetrahydrofuran (THF), with water or buffer [1-5]. This method development approach was later refined by Schoenmakers et al. [6,7]. The principle employed here was the fact that the elution pattern changes as a function of the solvent used. The organic solvents in reversed-phase chromatography are not only the strong eluents, but also selective eluents. Thus, solvent selectivity was the only tool in early methods development. If one expands the solvent selectivity triangle into a third dimension by adding the pH of the mobile phase as a method development References p. 214

Chapter 6

186 THF

MeOH

MeCN

THF

.........................\\~ .. .............High . pH " MeOH

MeCN

Fig. 6.1. Method development prism: methods are developed using three organic modifiers of different selectivity at two different pH values. Modifiers: methanol, acetonitrile, tetrahydrofuran. Buffers: example 1: phosphate pH 2 and phosphate pH 7; example 2: ammoniumformate, pH 3.75 and ammonium bicarbonate, pH 9 to pH 10.

tool, one ends up with a solvent selectivity prism. An efficient method development strategy based on the solvent selectivity prism (Fig. 6.1) was developed by E1 Fallah and Neue [8]. The efficiency of this strategy stems from a sequential exploration of the effects of pH and solvent selectivity on the separation. In more recent times, several publications by Dolan and Snyder [9,10] covered the use of temperature as an efficient method development tool. The solvent selectivity triangle uses three convenient solvents: methanol, acetonitrile and tetrahydrofuran (THF). These three solvents were chosen based on their relevant characteristics for method development. Methanol is a proton donor, while acetonitrile and THF are proton acceptors. In addition, the hydrophobicity of these three solvents is significantly different. These different solvent properties result in differences in the selectivity of a separation. One starts off with a gradient from water to a high concentration of the organic solvent for all three solvents. Alternatively, a solution of an acid in water or a buffer may be used for control of the ionization of the analytes. From the elution times of the analytes under the gradient conditions, one determines suitable isocratic elution conditions with all three solvents. The targeted isocratic solvent compositions have approximately equal elution strength. The next step in the method development consists now of a systematic exploration of the solvent selectivity triangle formed by these three isocratic solvent compositions. Binary mixtures and a ternary mixture of the solvent compositions in the corner of the selectivity triangle are explored first. This results in a rather complete knowledge of the separation characteristics within this experimental realm. The final steps involve a fine-tuning of the best separation in the binary or ternary mixture space. This approach ignored the use of pH as an additional tool in method development. This was quite understandable, since the packing materials available at that time often exhibited significant peak tailing at neutral pH due to the influence of silanols on the surface of reversed-phase packings. Silanol-induced tailing could often be suppressed at acidic pH. Thus, the constraint to a single pH value was not seen as a drawback. In the

Method development in reversed-phase chromatography

187

1990s, packings based on high-purity silicas became available. With these packings, very little tailing was observed at neutral pH. Now one was able to fully exploit the use of the pH of the mobile phase in method development. The solvent selectivity triangle expanded into a solvent selectivity prism (Fig. 6.1). E1 Fallah and Neue [8] recognized that the choice of the pH value creates the strongest selectivity effects for ionizable compounds, which are the typical analytes encountered in the pharmaceutical industry. Therefore they created a method development strategy that employed an exploration of the shift in selectivity via pH manipulation first, followed by a fine-tuning of the separation through the use of solvent selectivity. One would first run two gradients at two different pH values, e.g. pH 2 and pH 7 with a phosphate buffer. One examines the primary gradient separations for their usefulness for isocratic method development. Criteria could be the resolution of most compounds or the compactness of the separation as an indication if an isocratic separation is at all possible. One then runs a second gradient at the selected pH with a different gradient slope. The retention times obtained in both gradients are then used to calculate a suitable isocratic solvent composition. This isocratic condition is run next. If the first isocratic separation is unsatisfactory, one now calculates an isoeluotropic mobile phase in a second solvent [7,8]. An isoeluotropic mobile phase provides a retention window of similar retention times. For example, if the first gradients were run using methanol, resulting in an initial isocratic separation employing methanol, one would now explore acetonitrile, and the acetonitrile composition resulting in approximately equal retention times is calculated. Quite often, a shift in the elution order is observed when the isocratic isoeluotropic chromatograms are compared to each other. One can now employ the solvent selectivity differences to fine-tune the separation. If this is not sufficient, one can explore THF as the third solvent option, including the possibility of creating ternary solvent mixtures. This sequential use of pH and solvent selectivity resulted in very rapid and efficient method development schemes. Recently, Dolan and Snyder have explored the simultaneous use of temperature and the gradient profile in the development of methods for complex samples [9,10]. Both parameters cause rather subtle changes in the selectivity of a separation, which makes this approach ideally suited for a fine-tuning of a complex separation, i.e. a separation of 15 to 20 compounds or more. The rather small selectivity effects are also most appropriate for exploration with method development software. A complete exploration of the experimental space without the aid of such software is otherwise a rather daunting task. However, the software will quickly guide the experiment to an optimal set of conditions, and the efficiency of method development improves drastically. The authors also compared the effect of using two solvents, methanol and THE instead of the rather subtle manipulation of selectivity via temperature and gradient duration. For the complex mixtures explored, both approaches were found to be equally effective. This rather surprising result was interpreted by the authors as originating in the simultaneous use of two variables in the optimization process. Another interpretation is the fact that complex sample mixtures leave little space in the chromatogram to explore. As a consequence, there are many different chromatographic conditions where an overlap of some of the sample bands occurs. Only in the intermediate conditions, a complete resolution of all bands is possible. References p. 214

188

Chapter 6

Another interesting method development problem is the separation of a few analytes (for example parent compound and one or two metabolites) from an ocean of interferences (for example components of a plasma sample). Often, this problem can be resolved with sample preparation techniques [11], but a good method development approach should be able to deal with such a situation as well. The general method development strategy discussed here can not only deal with the optimization of the separation of standards, but also with a very complex sample matrix. In completely automated method development, one would like to have a good measure of the quality of the separation. The generally accepted criterion found in the literature is the resolution of the least resolved peak pair [ 1,12]. However, other criteria have been proposed as well. A good review of some of the functions can be found in [12]. E1 Fallah [8] also discusses some of the criteria. A more detailed discussion of previous work is outside the scope of this article.

6.2.1 Tools for the measurement of selectivity differences and the quality of a separation A key factor in the decision-making process in method development is the knowledge of which experimental variables have a large effect on the separation, and which variables can be ignored. It is therefore desirable to have a tool at hand that allows us to measure selectivity differences quantitatively. The experimentalist will typically choose a visual comparison of different chromatograms. The "more different" they are, the larger are the selectivity differences between the different conditions. This thought process can be translated into a simple quantitative measure. We can plot the retention times (or retention factors) obtained under one set of experimental conditions against the retention times (or retention factors) from a different set. If one obtains a straight line with a correlation coefficient, r 2, of 1.0, then there are no selectivity differences between the different chromatographic conditions. On the other hand, if a large scatter is observed, significant selectivity differences have been found. A large scatter results in a small correlation coefficient. Thus, we can use the parameter s 2 s2= 1 - r 2

(1)

as a simple, quantitative measure of the selectivity difference between different experimental conditions. If the correlation between the retention times under the different separation conditions is good, the selectivity difference measured by the selectivity parameter s 2 approaches 0. If there is no correlation, the selectivity parameter s 2 approaches 1. In practice, we can expect that excellent selectivity differences within a single chromatographic technique like reversed-phase chromatography will yield values around 0.5. [Footnote: The justification of Equation 1 is rather simple. The correlation coefficient is defined as the ratio of the variance explained by the curve fit to the total variance of the data. Our value for the selectivity difference is therefore nothing but the ratio of the unexplained variance to the total variance. The "explanation" is similarity of the retention mechanism.]

Method development in reversed-phase chromatography

189

An example of this is shown in Fig. 6.2. We have measured the retention times under gradient conditions for 11 diuretics on two columns and at two different pH values. On the left you see the comparison of the retention times obtained on an XTerra | RP18 column plotted against the retention times obtained on an XTerra | RP8 column at fixed mobile phase conditions. A high correlation is observed, with a correlation coefficient of 0.99. Therefore the selectivity difference is small, with a selectivity parameter of 0.01. On the fight you see the plot of the retention times for the same compounds obtained at pH 3.05 and pH 9.30 on a single column. A significant scatter is observed. The correlation coefficient between both experimental conditions is 0.55. Thus, the selectivity parameter s 2 is large, s 2 - 0.45. Therefore the selectivity difference between both pH values is large. This demonstrates that the selectivity parameter described above is a simple but very useful measurement tool for the quantitative comparison of different experimental conditions. Another question that needs to be addressed is the measurement of the overall quality of the separation. Such a measure should include the fact whether or not all peaks of interest are separated from each other and from interferences, or if the analysis is accomplished in a target time frame. The measure should be "soft", which means that one should be able to see the direction of improvement, if the conditions are not completely satisfactory. On the other hand, if satisfactory conditions have been achieved, further "improvements" should have no additional influence on the measure. In addition, it should be possible to combine the different criteria with each other to form an overall measure of separation quality. In the following discussion, we propose such a measure of separation quality, Qs- The value of the separation quality Qs shall be 0, if the separation is unsatisfactory, and 1, if the separation is satisfactory. We also would like to be able to modify the steepness of the transition between an unsatisfactory condition and a satisfactory condition. A suitable parameter can be formulated as follows: n

Q~ = 1-I

C l+C n

(2)

C is the criterion under question, and n is a power that influences the steepness of the transition of the function between 0 and 1. A small value of the criterion will result in a small value of the separation quality, and a large value will yield a separation quality approaching 1. The separation quality will have a value of 0.5, if the criterion reaches the critical value. For example, a criterion could be to achieve a separation time of approximately 10 minutes or faster. The criterion C would then be a value of l O/tL, where tL is the retention time of the last peak in the chromatogram. If we achieve a faster separation, the criterion will have values exceeding 0.5, if we have a slower separation, the value will be below 0.5. How important the transition is from a separation under 10 minutes to a separation over 10 minutes can be expressed through the power n. High powers emphasize a fast transition, low powers a slow transition. For example, if a separation time of 11 minutes is unsatisfactory, while a separation time of 9 minutes is acceptable, a high power should be chosen. Similarly, we can explore the relationship of the separation quality with the resolution Rs. In this case, the criterion is formed by dividing the resolution obtained by a target

References p. 214

12

all

W

a-lO

bl

@

.m

1_8

~6 I= llm I-, C

0

00

4

$2 =

4.12

s

0

i

t

,

,

2

4

6

8

0.01

2 ..........

0

,

10

R e t e n t i o n Time XTerra | RP 8

12

0

1

2

3

4

5

r2 -

0.55

sz =

0.45

6

7

8

9

R e t e n t i o n Time @ pH 9 . 3

Fig. 6.2. Measurement of the selectivity differences, la. Correlation of retention times obtained on two columns with different chain length. The correlation coefficient is close to 1, and the selectivity coefficient approaches 0. lb. Correlation of retention times obtained at two different pH values on one column. The correlation is poor, and the selectivity coefficient is large.

Method development in reversed-phase chromatography

191

resolution Rscri,. In Equation 3, the separation quality is formulated as the product of all peak pairs in the chromatogram.

Q~-H

-

(3)

In this formulation, a satisfactory separation is achieved, if the resolution for all peaks exceeds the critical resolution appreciably. A reasonable value for the critical resolution would for example be a value of 1, equivalent to a spacing of 4 standard deviations between both peaks. A good separation is achieved, if the actual resolution exceeds the value of the critical resolution appreciably. The power n describes the steepness of the transition between a satisfactory value and an unsatisfactory result. Figure 6.3 shows the separation quality function Q,~for a single peak pair as a function of the resolution between the peaks using Equation 3. The value for the critical resolution was selected to be 1. As we can see, the transition between an unacceptable separation and a good separation depends critically on the steepness value n. If we judge that a 6-sigma separation between adjacent peak pairs adequate, we should select a value of 8 for the power in Equation 3. A value of 4 would drive the separation between adjacent peaks to a value beyond 8 standard deviations. As a generic criterion for good resolution, Equation 3 with a critical resolution value of 1 and a power of 8 appears to be best for most practical cases.

n=8 0.9

-

n=4

--

0.8

c o ,,,.,,, '~' 0.7 r m

~- O.6 t~ J=

0.5

'~

0.4

"~ 0.3 a

0.2 0.1

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Resolution

Fig. 6.3. Separation quality Qs as a function of resolution for a single peak pair. The different curves show an increase in the power of the function: 2, 4, 8.

References p. 214

192

Chapter 6

The separation quality function can include other elements as well. For example, we may decide that we want to have well retained peaks to avoid interferences from extraneous sample constituents eluting close to the void volume. For this purpose, a function of the following nature can be used: /d Qs = - 1+/('

(4)

where k is the retention factor. Of course, one must take care that such a criterion does not conflict with a criterion that specifies a maximum retention time, as discussed above. While it is possible to combine different criteria at will (as shown in equation 1), care must be taken to avoid contradictions. Another, very practical consideration may be to achieve a good peak purity for all peaks whose areas need to be measured. For example, all peaks above a certain threshold value need to be pure, while peaks below this value can be ignored. One can then create a criterion where the peak area of a peak of interest needs to be larger than a critical peak area that is defined by background noise and interferences. A criterion of this type can be defined as follows:

Q, = 1-I

(5)

where A is the peak area measured and A~ is the critical peak area defined by the background in the chromatogram. An appropriate choice for the critical peak area could be 10 times the background noise. To summarize: the general thought process of the criterion of separation quality Q, is the fact that the quality is either good, i.e. it has a value of 1, or it is unsatisfactory, reflected by a value of 0. There is a transition range between both values, and in the transition range the value achieved points into the direction where improvements are possible. Several criteria that follow this same thought process can be created and incorporated into an overall quality criterion. Since the value of a positive result is always 1, one can combine different criteria with each other simply by multiplication. An example is the product of the peak resolution shown in Equation 3. Of course, one needs to realize that while a summary criterion is attractive, an examination of individual criteria is more instructive for making decisions in method development.

6.2.2 Measuring selectivity differences Before we discuss efficient approaches to method development, it is worthwhile to explore first, which experimental parameters are more effective than others. Is it more promising to change the column or to manipulate the mobile phase composition? If we want to explore the mobile phase composition, which experimental variable will result in more powerful effects, and which one will exhibit more subtle influences? In the

Method development in reversed-phase chromatography

193

following we will rate the selectivity differences caused by a change in the column, the solvent, and the pH of the mobile phase. In order to do this, we will explore the selectivity parameter s 2 that was introduced in the last section. We will use a separation of diuretics as our standard. The compounds are shown in Fig. 6.4. The molecular composition of the compounds varies widely. A range of functional groups are found, including acidic and basic functions. The structures vary from a steroidal structure (canrenoic acid, no. 4) to a structure dominated by polar functional groups (e.g. bumetanide, no. 1). A selection of such a broad range of analytes is expected to yield a good overview of selectivity effects. We studied first the selectivity effects caused by the choice of the bonded phase. We chose packings from the XTerra | family (Waters Corporation, Milford, MA) for our study. XTerra | packings are based on a first generation inorganic-organic hybrid particle that exhibits excellent pH stability [13]. It is prepared by a technology similar to the technology of modem high-purity silicas. The difference is the use of a combination of tetraethoxysilane and methyl triethoxysilane. This technique results in an inclusion of a large amount of methyl groups incorporated into the matrix of the packing. The consequence of this is a significant improvement of the pH stability of the packing compared to silica-based packings and a notable reduction in silanol group activity. From the standpoint of our studies, the use of this packing allowed us a freedom of choice in pH that is not possible with classical silica-based packings. Packings of this family have been used at pH 11.5 and 30 ~ C for over 50 days [ 14]. The reversed-phase packings of the XTerra | family comprise trifunctionally bonded C8 and C~8 packings, a difunctionally bonded 2-phenylpropyl packing, and monofunctionally bonded RP8 and RP18 packings with a carbamate group incorporated into the ligand [15]. This family allows first of all a comparison of the influence of the chain length of the ligand on the selectivity of a separation. In addition, the influence of the incorporation of a polar group into the ligand can be studied. Packings of this type have become very popular during the last few years due to the good peak shape for basic analytes caused by the incorporation of the polar group. Phenyl-type ligands have been used traditionally to achieve unique selectivities in a separation. The separation of the 11 diuretics shown in Fig. 6.4 was carried out using a fixed gradient from 0 to 80% methanol at a flow rate of 2 mL/min and a controlled temperature of 30 ~ C. We used short columns, 50 mm x 4.6 mm, packed with 5 Ixm particles. It is generally convenient to use short columns such as these for screening studies in method development. With such a column, a gradient over 30 to 50 column volumes can be carried out in 10 to 15 minutes. Fig. 6.5 shows the separations obtained with four XTerra | columns, the XTerra | RPI8 and RP8 columns with the incorporated polar groups are shown on the top and the XTerra | MS C~8 and MS C8 column are in the lower row of the figure. We have measured the selectivity differences for these columns as well as the XTerra | Phenyl column using the same gradient. The selectivity difference between the XTerra | MS C18 and C8 columns was smallest, s2=0.008. Similarly, the selectivity difference between the XTerra | RP18 and RP8 columns was small as well, 0.016. One can see that selectivity differences achieved by just varying the chain length of a reversed-phase bonded phase are rather small. References p. 214

H2NO2S~COOH C6H50""~

0%//O H2NO2S~S~NH

H2NO2S~S~ONH CII ~ ~ ~

NH(CH2)3CH3

F3c-

N/~CH2SCH2CsH~

1

O TH c , COOH

(CH3CH2CHz)2NO2S 6

OH - - ~

7

9

NH2

SO2NH2

c'y~CON.~N, NH

cHII I N N

~NJ\N~'~NH ~

l0

H2SCH2CH---CH2

CI

O

4

-N~CH2CoHs

O~ ~0

HOOC~

O

v

5

H2NO2cIII; 8

~

H2

CH3CH2Ci ~ O C H 2 C O O H Cl

Cl

Fig. 6.4. Compounds used in the study of selectivity effects. 1. Bumetanide, 2. Benzthiazide, 3. Bendroflumethiazide, 4. Canrenoic acid, 5. Althiazide, 6. Probenecid, 7. Chlorthalidone, 8. Furosemide, 9. Triamterene, 10. Amiloride, 11. Ethacrynic acid.

195

Method development in reversed-phase chromatography .

.~.I~D8

0.0

2.0

4.0

MS C18

......

6.0

8.0

10.0 12.0

I

o,

lo

/ 7582,3

!!4

J(..

j, _jl i!l

li'i

2.0

4.0

.

.

.

.

9

2.0

6.0 8.0 Minutes

10.0 12.0

!

6,11,3

s,B 141

4.0

6.0

8.0

i

10.0 12.0

-MS c8 10

, A )Li!i,\j ~ 9

r .....

,,

j

....

o.o

J

J

q

i

,

0.0

0.0

.

__

2.o

4.o 6.o

a.o

~

10.0 ' 12.0

Minutes

Fig. 6.5. Separationof a sample mixture of the 11 diuretics shown in Fig. 6.3 on four different columns. Top left XTerra| RP~8,top fight XTerra| RP8; bottom left XTerra| MS C~8,bottom fight XTerra| MS C8. All at pH 3.05. In agreement with reports in the literature [16], a larger difference was observed between classical reversed-phase packings and packings with an incorporated polar group, 0.051 and 0.065. For example, peaks 5 and 8, althiazide and furosemide, are well resolved on the classical reversed-phase packings, but coelute on both packings with an incorporated polar group. Differences in the elution pattern are also found at the end of the chromatogram. Peaks 2 and 3, benzthiazide and bendroflumethiazide, elute later on the packings with incorporated polar group. The largest selectivity difference in this group of packings is found between the classical XTerra | MS C~8 packing and the XTerra | RP8 packing with a value of 0.075. If we include the XTerra | Phenyl packing in this comparison, we find a selectivity difference of 0.016 to the classical XTerra | MS C8 packing and a value of 0.047 for the XTerra | RP8 packing. Interestingly, the elution order of benzthiazide and bendroflumethiazide changes on the XTerra | Phenyl column. Such small changes in the elution order can be important in method development. Occasionally, even stronger effects of stationary phase selectivity are found. Phenolic compounds are known to be retained longer on packings with an incorporated polar group [15]. Fig. 6.6 shows the gradient separation of a mixture of catechins as an example of this phenomenon. A large selectivity difference was found between the XTerra | MS C8 packing and the XTerra | RP8 packing. The value of this selectivity parameter was s2-0.183. The largest shifts in the position of the peaks in the chromatogram are pointed out by arrows. In addition, there is a big difference in retentivity. Packings with incorporated amide or carbamate groups retain phenolic compounds better than classical packings.

References p. 214

Chapter 6

196

5.0

10.0

2,~

15.0

3

.

.

~

5.0

,

,

, - ,

~

,

10.0

35.0

40.0

45.0

,~

i'~

.

30.0

9/I

!1 ~,

.

20.0 25.0 Minutes

.

.

.

] .

.

s.o

~i t

. . . . ;

20.0

) ,

T-=

t

,J / ,.

T---

2s.o

Minutes

.

,

.

.

.

30.0

/'

/i

',. .

.

.

3s'.o

.

40.0

i

=-

T

,

4s.o

Fig. 6.6. Separation of a mixture of catechins at pH 2.45. Top: XTerra | MS Cs, 4.6 mm • 150 mm, 5 txm; bottom: XTerra RP8, 4.6 mm x 150 mm, 5 txm. Peak designation: 1. Gallic Acid, 2. Epigallocatechin, 3. Catechin, 4. Caffeine, 5. Epicatechin, 6. Epigallocatechin Gallate, 7. Gallocatechin Gallate, 8. Epicatechin Gallate, 9. Catechin Gallate. Gradient: A: water, B: methanol, C: 50 mM formic acid in water, pH 2.45; gradient time: 45 minutes; initial composition: 84%A: 15%B: 1%C; 10 minutes: 84%A: 15%B: 1%C; 15 minutes: 69%A: 30%B: 1%C; final composition: 69%A: 30%B: 1%C; flow rate: 1.0 mL/min; temperature: 30~

Even larger selectivity differences can be achieved if we compare the XTerra | RP8 packing to the XTerra | Phenyl packing for the same group of compounds (catechins in Fig. 6.6). We achieved a selectivity difference of s2=0.213. On the other hand, the selectivity difference between the XTerra | MS Ca packing with the classical n-octyl ligand and the XTerra | Phenyl packing was only s2=0.026. Several examples are available that demonstrate large selectivity differences between classical packings and packings with an incorporated polar group. The difference in the selectivity of solvents is shown in Fig. 6.7. The value of the selectivity difference between acetonitrile and methanol was s2= 0.188 using a mixture of 9 of the diuretics shown in Fig. 6.4. This value is larger than the selectivity difference between the different columns for the same group of compounds. Quite significant changes in the elution patterns are observed. The position of peak 5, althiazide, changes quite drastically between both solvents. Similarly, peak 4, canrenoic acid, elutes as the last peak with methanol as the organic modifier. With acetonitrile, it elutes close to peak 2, benzthiazide. In addition, there is a change in elution order between peaks 6 and 11, probenecid and ethacrynic acid.

197

Method development in reversed-phase chromatography

i

]

5

MeCN

.......

o

1 0

........ ) , . . . . . . . . . . . .

2.0

3.0

4.0

sio

~

II

if!

1:1

~< ........

6.0

5,8

}II

.. i'~ l'qt_,

:7.0' 8 . 0 ~ 9 . 0

m

6 1

_ _ _ _ _

2.0

4.0

6.0

I MeOH

9

-

/'1

15

i

-!

!

2.0

4.0

8.0' ' io.o .... 12.0 6 1

7 iI ,tt 2

610

~1~,

,i 4 ,,~,

__

0

8 il"

),.........,

I!

~

1.o

~

~

6

i"~ .....

_

8.0

Minutes

lO.0 ~ ~ i 2 . o ~ 4 . :

Fig. 6.7. Separationof a sample of 9 diuretics using acetonitrile (top), methanol (bottom) and a l" 1 mixture of both solvents (center). Analyte designation as in Fig. 6.3.

Unlike stationary phase selectivity, mobile phase selectivity is a continuous variable. We can modify the separation by using solvent mixtures. We measured the selectivity differences between the gradients with the neat solvents methanol and acetonitrile and a 1:1 mixture of both solvents. The values of the selectivity parameter were 0.036 for the difference between acetonitrile and the mixture and 0.069 between methanol and the mixture. Not unexpectedly, both values are smaller than the difference between the neat solvents. In our example, peaks 5 and 8, althiazide and furosemide, and peaks 1 and 4, bumetanide and canrenoic acid, coeluted when the solvent mixture was used. One should also note the difference in the elution strength. Peaks 5 and 8 are widely separated using methanol as the organic modifier. The same is true for peaks 1 and four using acetonitrile. With acetonitrile as the organic modifier, the last peak elutes at 8 minutes. Using methanol, the elution is not complete until about 12 minutes. The mixture of both solvents has an intermediate elution strength. For ionizable compounds, one can expect that the variation of pH is a very powerful tool for manipulating the selectivity of a separation. For monovalent acids and bases, the retention changes by a factor of 10 to 30 between the non-ionic and the ionic form of the analyte [17]. This change in retention is commonly associated with changes in selectivity as well, and the change in retention is not uniform for even closely related analytes such as tricyclic antidepressants. An example is the separation pattern of doxepin, imipramine and nortriptyline [ 17], measured on a XTerra | RP18 column. Under References p. 214

Chapter 6

198

acidic conditions, nortriptyline and imipramine eluted very close to each other. In the alkaline pH range, there was coelution between nortriptyline and doxepin. In the intermediate pH range, all three compounds could be resolved. It is worthwhile to compare quantitatively the selectivity changes caused by pH changes with those originating in the stationary phase and the organic modifier. For this purpose, we explored the selectivity changes caused by pH for our standard set of 11 diuretics (Fig. 6.8). As expected, quite significant changes in the elution pattern are observed. Only the retention of peak 9, triamterene, increases with increasing pH. For all other peaks the retention decreases, but the reduction is not uniform. It is much larger for peaks 5 and 8, althiazide and furosemide, than for peak 7, chlorthalidone. The calculation of the selectivity differences confirms this. For the example shown in Fig. 6.8, the selectivity difference for the XTerra | RP18 column w a s s 2 = 0.30 between the low pH value and the high pH value. For the XTerra | RP8 column (not shown), the selectivity difference was even larger: s2=0.45. Without question, for ionizable compounds a change in pH is the most powerful tool in methods development. The pH value can be varied continuously. However, we need to keep in mind that one needs to have a reasonable buffer capacity for stable retention. Buffers should always be used around the pKa of the buffer components. For the very dilute buffers used today in LC/MS applications, the preferred useful range of a buffer is as little as + 1 pH units around the pKa. With buffers of a higher concentration, as used in classical HPLC with UV detection, one can safely use a buffer up to + 1.5 pH units around the pKa. This constraint makes the variation of pH a semicontinuous variable. For example, phosphate buffers can be used around pH 2 and pH 7, but not at pH 4.5. A citrate buffer, on the other hand, could be used from pH 2 to pH 7.5. However, citrate is rarely used in HPLC. In today's world with MS as the standard detector, formic acid and ammonium formate

6,11 1,4 6 1

5,8 10

2

4,1

10

2

,

! _3

0.00

2.00

4.00

6.00

8.00

1 0 . 0 0 12.00

2.00

4.00

6.00

8.00

10.00

Fig. 6.8. Separation of a sample mixture of the 11 diuretics in Fig. 6.3 at pH 3.05 (left) and pH 9.3 (right) using an XTerra| RPj8column.

Method development in reversed-phase chromatography

199

are used for buffers in the acidic pH range. Formic acid has a pKa of 3.75. For buffering in the more alkaline pH range, ammonium bicarbonate is used, together with ammonia (pKa= 9.24). Carbonate has two pK, values: 10.25 and 6.37. The higher value, together with the pKa of ammonia, gives ammonium carbonate a rather broad buffering range. The lower pK~ value of carbonate can be used as well, with appropriate precautions. If ammonium bicarbonate is titrated to values around pH 7 with a volatile acid such as formic acid or acetic acid, one obtains MS compatible buffers around neutral pH. Of course, due to the possible formation of carbon dioxide, this works best at the rather low buffer concentrations commonly used in LC/MS. In addition degassing methods should be avoided. In this context, one needs to consider that the fine adjustment of pH for achieving a desired selectivity requires simultaneously excellent control of the pH value. The strongest selectivity effects for any analyte are found around + 2 pH units around the pK of the analyte. The reason for the strong selectivity effects is simply a drastic change in retention with pH. While drastic changes are good for selectivity control, they are less desirable for the reproducibility of a separation. If indeed we develop a method that has a significant selectivity shift with small changes in pH, method reproducibility may require strict control of the mobile phase pH. Temperature manipulation is not frequently used as a tool for selectivity manipulation. This is largely due to the fact that the changes in selectivity are rather subtle. However, they are useful for the fine adjustment of a separation [18]. In the literature, temperature variation has been explored for the fine tuning of complex separations [9,10]. An example of such fine tuning is the separation of 14 nitroaromatics (explosives) on an XTerra | Phenyl column. The separation was incomplete at 30~ with an overlap of 2-amino-4,6-dinitrotoluene and 2,4-dinitrotoluene. A change in the temperature to 38~ resolved this peak pair as well, while preserving the remainder of the separation. In summary, we have discussed two different kinds of variables. The mobile phase composition and the temperature are continuous variables. Column selection is obviously not a continuous variable. The pH value is a variable that has both components. It can be varied in a continuous fashion around the pKa of the buffer, but for significantly different pH values, different buffers need to be used. In method development, the stepwise variables such as column selection and pH should be explored first. After a selection of the best separation or separations from this set, one would then employ the continuous variables to fine-tune the separation. The tools for this fine-tuning are the continuous variables, primarily solvent mixtures, fine adjustment of the pH, and temperature. In the next section, we will use these tools in actual method development. The selectivity differences associated with the different variables have been discussed in the examples shown. For ionizable analytes, the strongest selectivity effects have been found through changes in the pH value. For a change from an acidic to a basic pH, we have found selectivity differences as high as s 2 =0.8. More generally, values around s2--0.50 are expected. The choice of the organic modifier has been found to yield selectivity differences between S2=0.1 to s2-'0.7. The general expectation is that the choice of the solvent generates selectivity effects around s 2-- 0.25. The selectivity effects References p. 214

200

Chapter 6

found for the stationary phases examined here ranged from $2:0.05 to S 2-" 0.35. The strongest selectivity differences were found between stationary phases with incorporated polar group and classical stationary phases. For a good choice of stationary phase selectivity, the selectivity differences between carefully chosen stationary phases should be of the order of s2= 0.1. We have chosen such a set of stationary phases for the method development strategy discussed below. Stationary phase selectivity is still a field that is rather empirical, and more results can be expected in the future.

6.3 M E T H O D DEVELOPMENT STRATEGY Before we can start to develop an HPLC method, it is worthwhile to examine the options that are available, and the constraints that one may be facing. The standard HPLC instrument is equipped with a UV detector. Therefore the very first question is the UV response of the compounds of interest. Today, a mass spectrometer is quite commonly used as a detector as well, and we may develop a method using an MS detector, even if we anticipate that the method will use other detection schemes on the long run. If we use a mass spectrometer as a convenient detector in a method development scheme, it is best to use it in APCI mode, since this detection mode does not depend on the ionization state of the analytes in solution. However, even with the more commonly used electrospray mode, the detection of the analyte does not always follow the expected pattern, and sensitivities sufficient for method development can be found at pH values where the analyte is expected to be in a neutral form [19]. Another important element to know up front is the stability of the compound. In our standard method development scheme that we will outline below, the manipulation of the pH is an important tool. Therefore we need to know that the analytes are stable in the pH range of interest. Most of the time this is of little concern, but it is something to think about early on. Another factor that we can assess early on is the complexity of the separation. If we know already that the assay requires the separation of only a few analytes, and that the analytes are all related to each other, this will tell us right away that an isocratic separation should be feasible. On the other hand, a separation of 10 or more compounds of a wide polarity range may make a single isocratic separation impossible, and only multiple isocratic separations or a gradient should be considered. The general strategy of any method development scheme should be to explore first variables that cause a large effect, and then move progressively towards a fine-tuning of the method [8]. Among the variables with a large effect, one should first examine those that can be varied only in a stepwise form. An example of this is the selectivity of an HPLC column. On the other hand, stepwise variables that provide only small selectivity effects are not very useful, since larger effects can be generated via the continuous variables such as solvent mixtures. An example of such a small stepwise selectivity effect is the difference between two columns with the same surface chemistry but different chain length. Therefore, a large amount of work can be saved by eliminating similar columns such as the Cls-C8 or the RPIs-RPs pairs of columns shown above. In the same vein, solvent selectivity effects are quite strong, but they can be varied continuously to achieve the desired selectivity. Temperature is only a small additional

Method development in reversed-phase chromatography

201

variable that can be used if solvent selectivity and the fine adjustment of the pH do not yield a complete separation. The same is true for the buffer concentration. The effects of this latter variable on the selectivity of a separation are commonly so small that an attempt to exploit them are usually a waste of time. Considering these points, a rational method development scheme starts off with exploring several columns of known, large selectivity differences at two pH values with at least two solvents such as acetonitrile and methanol as organic modifiers. This allows for a simultaneous exploration of the stepwise variable, the column chemistry, the intermediate variable, pH, and the continuous variable with the largest selectivity effect, the organic modifier in the mobile phase. This permits a rough exploration of all the major effects on selectivity. These data are not only useful for selecting a starting point for a further fine-tuning of the method, but also for making a decision if an isocratic separation is feasible, or if an optimization of a gradient method is more promising. The final improvement of the method uses the continuous variables solvent composition and temperature, possibly assisted with a fine adjustment of the pH as well. Sometimes, more than one starting point for the fine adjustment of the method can be used, but since this final revision often relies on a direct interactive exploration of the experimental space, it is more effective to select a single condition for the final optimization. In the following, we will explore with a few concrete examples how such a method development strategy can be implemented. We will focus first on the initial selection of the best combination of columns, pH and organic modifier. We will use three XTerra | columns of different selectivity, the XTerra | MS Cls column, with a classical bonded phase, the XTerra | RP~s column with an embedded carbamate group, and the XTerra | Phenyl column. We will examine two pH values, the acidic pH with an ammonium formate buffer at pH 3.64, and the basic pH with an ammonium bicarbonate buffer at pH 9.0. In addition, we will explore the selectivity of two solvents, acetonitrile and methanol. In the first example, we are developing a separation of 6 tricyclic antidepressants. The compounds are nordoxepin, doxepin, imipramine, amitriptyline, nortriptyline and trimipramine. The structures are shown in Fig. 6.9. One can see that the compounds are closely related to each other. Nordoxepin/doxepin and nortriptyline/amitriptyline are pairs of secondary and tertiary amines of the same structure. The pair trimipramine and imipramine differs by a methyl group in the side chain. The difference between the pairs nordoxepin/nortriptyline and doxepin/amitriptyline is just an oxygen in the 7-membered central ring of the main structure. The expectation from these similarities is two-fold: 1. It is anticipated that it will be difficult to achieve the separation; 2. We hope to be able to find isocratic conditions for the separation of all 6 compounds. The general method development scheme requires 4 gradient separations for every column that we want to use: acidic pH and basic pH, each using acetonitrile and methanol as the organic modifier. We used the XTerra | MS C~s column, the XTerra | RP~8 column and the XTerra | Phenyl column. The chromatograms obtained with methanol as the organic modifier are shown in Fig. 6.10. There is a clear difference in the separation achievable at acidic pH and at basic pH. Already in these early gradient chromatograms, a complete separation of all 6 compounds is achievable at alkaline pH, while there are overlaps under acidic conditions. Especially the separation of

References p. 214

Chapter 6

202

CH3 i Hac/N

Doxepin

~'~

CIHa H3c/N-.~

?H3 H/N--

Nordoxepin

~C1-13

Imipramine

I I H3 CH3

Nortriptyline

Amitriptyline

~r )N~CH I 3 CH a

CHa

Trimipramine

Fig. 6.9. Structure of the 6 tricyclic antidepressants used in Example 1.

amitriptyline, nortriptyline and trimipramine appears to be difficult at acidic pH. At alkaline pH, nordoxepin and nortriptyline, the two secondary amines, elute before the group of tertiary amines. The tighter separation of the 6 compounds on the XTerra | Phenyl column holds the promise that a simple isocratic separation will be possible. The good resolution under basic conditions resulted immediately in a final gradient separation. A separation was possible using both the XTerra | MS C,s column and the XTerra | RP~s. We selected the latter column. The gradient was flattened, with a change in methanol from 60% methanol at 2 minutes to 74% at 10 minutes. With this gradient and the short 5 cm XTerra | RP~s column, a complete separation of all compounds and an internal standard was possible. A question arises how one can find a suitable isocratic condition from the elution pattern under gradient conditions. One needs to know the gradient composition at the column outlet at the point of elution of the peaks of interest. In the gradient chromatograms used here for a general exploration of the separation pattern, the analytes will have a retention factor of between 2 and 3 in the solvent composition at the column outlet at which they are eluting. One can then select a solvent composition that is representative for the group of analytes for the first isocratic trial. A fine adjustment of the concentration of the organic modifier may then give the final isocratic chromatogram. The theoretical principle behind this procedure has been outlined by Snyder and coworkers [2]. While our procedure requires only an estimate of the solvent composition at the column outlet, the theory allows for a direct calculation of the isocratic solvent

%

3 2

1

36,5

4~

l

I~__j~tl . . . . . .

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.011.012.0 Minutes

j

5

i

~l -~'

0.00

'

'

I

'

'

'

~

'

'

'

~

'

'

'

I

'

'

7 --s .......

~-~

I

'

46

iilli 4 '

'

I

'

'

'

I

2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 Minutes

3 4 21

,346,5

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.011.012.0 Minutes

l

ii

1

4,6

0.0 1.0 2.0 3.0 4.0 5.0 610 7.0 8.0' 9.0 " ........-10.0~.0~2.0

Minutes

. . . . . . . . . .

0.00 2.00 4.00 6.00 8.00 10.00 12.00 14~00 16.00 Minutes ,.#

~'

1 5

J

I

0.00 '

.......

2.00 '

4.00 ' .

.

.

.

0.00 .

.

.

.

8.00

,,146

t

i6.00 ~210J14.00 16.00

Minutes .

.

Fig. 6.10. Initial gradient separation of 6 tricyclic antidepressants using methanol as the organic modifier. Left: acidic pH; right: basic pH. Columns from top to bottom: XTerra | MS C,~, XTerra | RP~8, XTerra | Phenyl. Peak designation: 1 Nordoxepin, 2 Doxepin, 3 Imipramine, 4 Amitriptyline, 5 Nortriptyline, 6 Trimipramine. Column Dimension: 4.6 • 50 mm 3.5 ~m. Column temperature: 30~ Flow rate: 2.0 ml/min. Detection: 254 nm. Gradient run time: 20 min. Gradient conditions: 0 to 15 minutes linear from 90% A, 0% B, 10% C to 10% A, 80% B, 10% C. Mobile phases: A: water. B: methanol, C: left: 100 mM ammonium formate, pH 3.64; right: 100 mM ammonium bicarbonate, pH 9.0.

t%

0.20-

3

0.15

4

0.10

0.05-

\

0.000.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

10.00

11.00

12.00

Minutes Fig. 6.11. Isocratic separation of 6 tricyclic antidepressants. Column: XTerra | Phenyl; 4.6 x 100 mm, 3.5 p~m. Column Temperature: 30~ Flow rate: 1.4 mL/min. Detection: 254 nm. Mobile phase: 65% methanol; 25% water; 10% 100 mM ammonium bicarbonate, pH 9. Peak designation: 1 Nordoxepin, 2 Doxepin, 3 Imipramine, 4 Amitriptyline, 5 Nortriptyline, 6 Trimipramine.

Method development in reversed-phase chromatography

205

composition [8]. Since there are two unknowns in the equation, two gradients with two different slopes are required for the mathematical procedure. We find that the simple estimate is sufficient to get to a starting point for the isocratic separation. If necessary, a small adjustment in the elution strength of the isocratic composition may be required. Following this thought process, an isocratic separation was accomplished using the XTerra | Phenyl column (Fig. 6.11). The elution of all compounds in a reasonable time was possible with a mobile phase containing 65% methanol and 35% buffer at pH 9.0. Good resolution between all peaks is achieved. The shoulder on the doxepin peak is due to a stereoisomer of doxepin present in the commercial sample. While the separation is possible using a short 5 cm column, we selected a 10 cm column for our final isocratic chromatogram. Due to the excellent resolution, the separation time can still be shortened, if one chooses to do so. Two techniques could be used for this purpose. One can either increase the methanol concentration by a small percentage, or use a higher temperature. We did not pursue any of the options available for a further reduction of the run time. It is worthwhile to examine the quality of the separation achieved in more detail. Ignoring the doxepin stereoisomer, all individual resolution values between each peak pair are larger than 2.5. If we calculate the value of the separation quality described in Equation 3, we reach a value of 1.00 for this separation, using the recommendations in the paragraphs following Equation 3. This means that the separation is good and rugged. This confirms the impression that one gets from a visual inspection of the chromatogram. The second example is a separation of 9 of the diuretics shown in Fig. 6.4. The compounds of interest for this assay were bumetanide (1), benzthiazide (2), canrenoic acid (4), althiazide, (5), probenecid (6), chlorthalidone (7), furosemide (8), triamterene (9) and ethacrynic acid (11). The 12 initial gradient separations are shown in Figs 6.12 and 6.13. Not unexpectedly, significant selectivity shifts are found between acidic pH and alkaline pH, as well as between acetonitrile (Fig. 6.12) and methanol (Fig. 6.13). In general, the acidic pH conditions spread the analytes over a wider elution range in the chromatogram. The basic pH provides a more compact chromatogram. If a final gradient separation is all that is required, several of the gradient chromatograms obtained under acidic conditions give a good starting point. Examples are all the chromatograms obtained with acetonitrile at pH 3.64, or the chromatogram with the XTerra | MS Cls column and methanol as the organic modifier. All chromatograms are amenable for attempts to fine-tune the separation. While a further exploration of mixtures of acetonitrile and methanol might yield an optimal separation, it appears on first glance that solvent selectivity might be too crude a tool. The fine-tuning can be accomplished better via small modifications of the pH, the temperature, or the gradient profile. We calculated the criterion of the quality of the separation shown in Equation 3 for these 4 chromatograms. Using the suggested criteria, we obtained the following values: in acetonitrile: XTerra | MS C~8: 0.002; XTerra | RPIs: 0.047; XTerra | Phenyl: 0.055; in methanol: XTerra | MS C~8: 0.095. The differences are small, but they show a slight advantage for the XTerra | MS C18 column with methanol as the organic modifier. A References p. 214

"-!!-

1-

7

A : ' *

aD - - = =

Chapter 6

Fig. 6.12. Exploratory gradient chromatograms for the separation of 9 diuretics using acetonitrile as the organic modifier. Left: pH 3.64; right: pH 9.0. Top: XTerra" MS C,#;middle: XTerra') RP,,; bottom: XTerraO Phenyl.

%

5,7

9

2,6

9

c~

1

t,,o 4~ )

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

0.0

2.0

4.0

6,11

5,8

8.0

6.0

o.o

9

'

2.~

.

.

.

.

..~ .

.

1, ..~

1

__,,_ '

'

'

8.0 i

.

.

.

~o:o .

.

.

.

.

~2:0

.

.

.

~.io

o.o

4.0

'

.

.

~.; .

.

.

..o "1

'

":

6,11

8

6.6

8.0

Minutes

~010

6.~

'

'

8.;

i

'

i

~o.;

~

~

~2.; i

i

i

~ ' 14.0

5

9

2.0

14.0

1

5

0.0

12.0

6,5

9

",

10.0

~210

~410 0.0

,':i'2: 11,4

8.0

Minutes

14.0

Fig. 6.13. Exploratory gradient chromatograms for the separation of 9 diuretics using methanol as the organic modifier. Left: pH 3.64; right: pH 9.0. Top: XTerra | MS CI~; middle: XTerra | RPjs; bottom: XTerra | Phenyl.

Chapter 6

208

visual examination of the chromatogram shows a good resolution of most peaks. The only difficulty in this chromatogram is the only partial resolution between bumetanide and ethacrynic acid. We decided to explore a modification of the gradient profile of methanol as the organic modifier with the XTerra | MS C~8 column for a final finetuning of the gradient separation. The early part of the chromatogram exhibits good resolution between all peaks. Therefore a steeper gradient can be used in this section without difficulty. On the other hand, an isocratic condition might resolve the peaks in the later part of the chromatogram in which probenecid, ethacrynic acid, bumetanide and canrenoic acid are eluting. Indeed, Fig. 6.14 shows this combination of a steep gradient at the beginning of the chromatogram followed by an isocratic section using an XTerra e MS C~8 column and methanol as the organic modifier at acidic pH. All compounds are resolved without difficulty. To explore the possibility of an isocratic separation, the more compact chromatograms obtained under basic conditions need to be considered. A good starting point is the separation achieved using the XTerra | MS C~8 column at pH 9 with acetonitrile as the organic modifier. Even under gradient conditions, there is only one overlap, benzthiazide and althiazide. Both are separated well with methanol as the organic modifier, but other compounds coelute under these circumstances. There is a good chance that the correct choice of a mixture of acetonitrile and methanol will result in a final isocratic separation of all 9 compounds. One can also anticipate that a good separation can be obtained with acetonitrile as the primary organic modifier, with a small amount of methanol to complete the separation of benzthiazide and althiazide. 5

9

6

2

iii

~

4

J

--T--q--T,

0.0

i

'

1.0

'

'

'

I

'

2.0

'~---Vl

'

3.0

'

''

I

. . . .

4.0

~

. . . .

5.0

I

. . . .

6.0

i"

7.0

'

'

'

I

. . . .

8.0

~

. . . .

9.0

I

. . . .

10.0

T

. . . .

I

11.0 12.0

Minutes

Fig. 6.14. Final gradient separation of the 9 diuretics. Column: 3.5 l~m XTerra| MSCIs, 4.6• Column temperature: 30~ Flow rate: 2.0 mL/min. Detection: 254 nm. Injection Volume: 20 ~1. Mobile phase: A: 100 mM ammoniumformate, pH 3.64. B: water; C: methanol. Gradient: 10% A throughout; 0 min: 90% B; 1 min: 50% B; 3 min. to 10 min: 46% B; 11 min: back to initial conditions.

Method development in reversed-phase chromatography

209

Appropriate isocratic compositions were between 16% acetonitrile and 36% methanol. A few experiments at intermediate compositions revealed that one could indeed achieve a complete isocratic separation of all 9 compounds with a composition of 13% acetonitrile and 4% methanol at pH 9.0, using a 10 mM ammonium carbonate buffer. One needs to realize that ternary solvent mixtures that result in the desired analysis time are not always linear mixtures of the binary solvent mixtures. Often, some fine adjustment in the elution strength is necessary. The final isocratic separation is shown in Fig. 6.15. The separation time is under 15 minutes using the short 5 cm XTerra | MS C18 column packed with 3.5 Ixm particles. The quality of the separation value based on the resolution of all 9 compounds was 0.91, limited by the elution pattern of triamterene, furosemide and chlorthalidone early in the chromatogram. In both cases described here, it was possible to develop both a gradient separation as well as an isocratic separation. Depending on the separation problem, this may not always be possible. If the elution pattern of all analytes of interest stretches over a significant range in all gradient separations, it is unlikely that a single isocratic composition can be found that will resolve and elute all analytes. In some industries, isocratic separations are preferred. The reason for this has been the difficulty in converting gradient separations between instruments from different manufacturers or of different generations. However, there is nothing wrong with a well characterized gradient separation. To transfer gradient separations from instrument to instrument, the

1

2

0.0

t1 1 9

2.0

IIItI~IL 4.0

6.0

itlL

8.0

t

i

II

I

10.0

I

~

I

12.0

1

~

I

~

14.0

I

I

t

16.0

Minutes

Fig. 6.15. Final isocratic separation of the 9 diuretics. Column: XTerra| MS C18,4.6 mm x 50 mm, 3.5 Ixm. Mobile phase: 13% acetonitrile, 4% methanol, 73% water, 10% 100mM ammonium formate, pH 9.0. Detection: 254 nm. References p. 214

210

Chapter 6

gradient delay volume of each instrument must be known. With this knowledge, corrections can be made to the gradient program. Occasionally, difficulties arise also from the accuracy of the gradient generation itself. However, this is similar to difficulties that arise in isocratic separations, when one relies on the instrument to generate the isocratic solvent composition. Today's instruments are better standardized than in the past, and the transfer of a method from one instrument to another should no longer present an insurmountable difficulty. A final application example involves the separation of a small number of compounds from a large body of interferences. Examples of such separation problems are the analysis of parent compounds and metabolites in body fluids such as blood, cerebrospinal fluid and urine, tissue samples or cells. In many cases, the interferences can be removed with efficient sample preparation schemes [ 11 ]. Selective detection such as mass spectrometry can reduce the problem, but higher sensitivities and less ion suppression can still be achieved with a reduced background [20]. However, even with classical detection tools such as UV detection, excellent results can be obtained using the identical method development scheme described for the separation of a complex mixture. To demonstrate this, we used the analysis of verapamil, methoxyverapamil and norverapamil in plasma. The sample was prepared using a simple 1-D sample clean-up method as described in 11. This method removes proteins and polar interferences from a plasma sample, but leaves a large number of low molecular compounds with the analytes. Rat plasma was acidified with concentrated phosphoric acid to a final concentration of 2% H3PO 4. A Waters Oasis cartridge was conditioned with methanol and water, and the acidified plasma sample was loaded onto the cartridge. Proteins and polar interferences were washed off with 5 % methanol in water. The elution was carried out using methanol. The eluent was dried down and reconstituted in water. This is the standard 1-D protocol for plasma samples. The target of method development is now two-fold: 1. the separation of the three analytes from each other and 2. the separation of the analytes from interfering matrix constituents. While the nature of the interferences is unknown, such a problem can be solved in the same way as the separation of standards shown before. The gradient separation of the standards in a plasma sample is obtained using the same gradient method development scheme as shown above for the diuretics. Methanol and acetonitrile gradients are carried out at low pH and at high pH using XTerra RP~s, XTerra MS C~s and XTerra Phenyl. Now we are not only observing, whether the analytes are separated, but we also look for the condition where we obtain the smallest amount of interferences. The results of the method screening for the problem at hand are summarized in Fig. 6.16. The retention times of the three analytes of interest are shown on the y-axis, the different experimental conditions are grouped together on the x-axis. On the left, we see the results obtained under acidic conditions grouped together, and on the right under basic conditions. In both groups, the results obtained with methanol as the organic modifier are shown on the left, and the results with acetonitrile are found on the fight. Below the axis we see the listing of the three columns used for each experimental condition. Only three experimental conditions resulted in a separation of the three standards: acetonitrile at acidic pH using the XTerra MS Cls and the XTerra

Method development in reversed-phase chromatography 14.00

- e - - Verapamil

211

~

-E- Methoxyverapamil, ~' ..... E .

12.00

- ~ - Norverapamil

G}

i"7- 10.00

.... 9

I~

.....

8.00

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

._~

6.00 "4.00

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

pH 3.6

pH 9.0

M~OH

t" ,

MeCN .

"~

f

,

2,,

-" "~

c~ M e o H {

,

MeCN~-~ , ~

,

f

,

,X,

"~

Fig. 6.16. Method screening plot for the separation of verapamil, methoxyverapamil and norverapamil in plasma samples. See text for detailed explanations!

Phenyl columns, and acetonitrile at basic pH using the XTerra Phenyl column. All conditions using methanol as the organic modifier resulted in a partial or complete overlap between verapamil and methoxyverapamil, independent of the pH. On the graph, we also see a notation for the amount of interference from plasma constituents. The rating ranges from "-" to "+ + ". Unfortunately, both of the conditions at acidic pH that yield an at least partial separation of the three analytes also suffer from a large amount of interferences. An example of this condition, marked with a "-", is shown in Fig. 6.17a. The acetonitrile gradient separated all three compounds, but the interference background is high. On the other hand, the condition shown in Fig. 6.17b, acetonitrile at basic pH using an XTerra Phenyl column, resulted in a partial separation of all three compounds and only a small amount of plasma interferences and was rated "+ ". Under the same mobile phase conditions, the separation of the analytes from the plasma interferences was only marginally worse on the XTerra RP18 column (rated "0"), but the separation between verapamil and methoxyverapamil was inferior. The graph shown in Fig. 6.16 gives a quick overview of the merits of the different conditions explored during the method screening process. Based on this information, the chromatogram shown in Fig. 6.17b was therefore selected as the overall best starting point for an isocratic separation. As also shown above, we selected a longer column for the final isocratic optimization, a 10 cm 3.5 Ixm column instead of the 5 cm columns used for the gradient screening approach. The final isocratic chromatogram is shown in Fig. 6.18. A clean separation has been obtained for the three analytes of interest. The baseline is rather free of interferences around the compounds of interest. Only around the norverapamil peak one finds a small level of References p. 214

212

Chapter 6

0.08:

a,

1 3 2

0.06 ~

~./~..~_~J

0.04

o.o / 0.02

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

I0.00 11.00 12.00 13.00 14.00 15.00

Minutes

0.06

bQ I

0.04 <

3

2

0.02

0.00 0.00

2.00

4.00

6.00

8.00 Minutes

*

~ 10.00

'

'12.100'

'

' 14.00 ~ '

Fig. 6.17. Chromatograms of verapamil (1), methoxyverapamil (2) and norverapamil (3) in plasma samples. 17a. Acetonitrile gradient at acidic pH using a 4.6 mm • 50 mm 3.5 I~m XTerra MS C18 column. Gradient: 0% to 80% acetonitrile in 15 minutes. Buffer: 10 mM ammonium formate pH 3.65. Flow rate: 2 mL/min. 17b. Acetonitrile gradient at basic pH using a 4.6 mm x 50 mm 3.5 &m XTerra Phenyl column. Gradient: 0% to 80% acetonitrile in 15 minutes. Buffer: 10 mM ammonium bicarbonate pH 9.0. Flow rate: 2 mL/min. In both cases: Temperature 30~ Detection: 230 nm.

residual interferences. Thus, the method development protocol outlined above for complex mixtures of known standards is also very suitable for the development of a method against a large and complex background of unknowns. In this case, the approach is suitable not only for sample clean-up using standard UV detection, but also for removing interferences that cause ion-suppression if MS detection is used. Of course, if an HPLC/MS system is used for the final method, the small amount of residual interferences are even less likely to create difficulties. As has been shown in reference [ 19], the use of a mobile phase at alkaline pH does not lead to a suppression of the signal for basic analytes in positive ion mode either. Whether one is using an HPLC system with a classical UV detector or a HPLC/MS system, the method screening approach shown here can rapidly lead to a method with desirable optimal results. Of course, faster methods can be developed for LC/MS applications [19], if this is the target of the method development.

213

Method development in reversed-phase chromatography

0.020

< O.OLO

0.000 o.oo

I

.

.

, I

5.00

10.00

15.00

Minutes

Fig. 6.18. Isocratic chromatogram of verapamil (1), methoxyverapamil(2) and norverapamil (3) in plasma. Column: 4.6 mm x 100 mm 3.5 Ixm XTerra Phenyl. 37% acetonitrile, 53% water, 10% 100 mM ammonium bicarbonate pH 9.0. Flow rate: 1.4 mL/min. Temperature 30~ Detection: 230 nm.

6.4 C O N C L U S I O N An efficient starting point for the development of a reversed-phase HPLC method has been outlined. Using this procedure the major selectivity differences between the three main variables pH, solvent and stationary phase can be rapidly evaluated. We have found that the largest selectivity differences for ionizable compounds are caused by pH changes. The second largest selectivity changes are produced by the organic modifier, methanol or acetonitrile. The third group of selectivity changes are due to the stationary phase itself. However, only very small changes are found when stationary phases of different chain lengths are compared. Larger selectivity effects are found when a classical stationary phase is compared to one with an embedded polar group. In our general method development set-up, we have combined the classical Cls and the embedded polar phase with a phenyl phase to maximize the selectivity differences among different packing materials. We have also given some general guidelines on how to fine-tune a method based on the information obtained in the initial screening. For general purposes, the major tool for the fine adjustment of a separation is the solvent selectivity. However, a small modification of the pH or a change in temperature can be used as well. The general starting point for the fine-tuning procedure is the best separation obtained from the initial screening. Even for reasonably complex mixtures, the combination of the parameters used during the initial selectivity study gives an excellent start for the final adjustment of the method. The knowledge on solvent selectivity obtained during the screening study can quickly yield a satisfactory separation of all analytes. This is possible both for multi-component mixtures with known standards and for the analysis of a few known compounds against a vast unknown background. References p. 214

214

Chapter 6

6.5 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 19 20

L.R. Snyder, J. Chromatogr. Sci., 16 (1983) 223. L.R. Snyder, J.W. Dolan and J.R. Gant, J. Chromatogr., 165 (1979) 3. J.W. Dolan, J.R. Gant and L.R. Snyder, J. Chromatogr., 165 (1979) 31. J.L. Glajch, J.J. Kirkland, K.M. Squire and J.M. Minor, J. Chromatogr., 199 (1980) 223. L.R. Snyder, J.L. Glajch and J.J. Kirkland, Practical HPLC Method Development, Wiley-Interscience, New York, 1988. EJ. Schoenmakers, H.A.H. Billiet, R. Tijssen and L. de Galan, J. Chromatogr., 149 (1978) 519. EJ. Schoenmakers, H.A.H. Billiet and L. de Galan, J. Chromatogr., 205 (1981) 13. M.Z. E1 Fallah, in: U.D. Neue (Ed.), HPLC Columns, Theory, Technology, and Practice, Wiley-VCH, New York, 1997. J.W. Dolan, L.R. Snyder, N.M. Djordevic, D.W. Hill and T.J. Waeghe, J. Chromatogr., A 857 (1999), 1-20 J.W. Dolan, L.R. Snyder, N.M. Djordevic, D.W. Hill and T.J. Waeghe, J. Chromatogr. A 857 (1999) 21. U.D. Neue, C.R. Mallet, Z. Lu, Y.-E Cheng and J.R. Mazzeo, Handbook of Analytical Separations, Vol. 4, 83, I. D. Wilson, Ed., Elsevier Science, 2003. J.C. Berridge, Techniques for the automated optimization of HPLC separations, Wiley-Interscience, Chichester, 1986 Y.-E Cheng, T.H. Walter, Z. Lu, EC. Iraneta, B.A. Alden, C. Gendreau, U.D. Neue, J.M. Grassi, J.L. Carmody, J.E. O'Gara and R.P. Fisk, LC-GC, 18 (2000) 1162. U.D. Neue, T.H. Walter, B.A. Alden, Z. Jiang, R.E Fisk, J.T. Cook, K.H. Glose, J.L. Carmody, J.M. Grassi, Y.-E Cheng, Z. Lu and R. Crowley, Amer. Lab., 31 (1999) 36. U.D. Neue, Y.-E Cheng, Z. Lu, B.A. Alden, E C. Iraneta, C.H. Phoebe and K. Tran, Chromatographia, 54 (2001) 169. J.W. Dolan, L.R. Snyder, T. Blanc and L. Van Heukelem, J. Chromatogr. A, 897 (2000) 37. U.D. Neue, C.H. Phoebe, K. Tran, Y.-E Cheng andZ. Lu, J. Chromatogr. A, 925 (2001)49. L.C. Sander and S.A. Wise, J. Sep. Sci., 24 (2001) 910. Y.-E Cheng, Z. Lu and U.D. Neue, Rapid Commun. Mass Spectrom., 15 (2001) 141. C.R. Mallet, Z. Lu, J. Mazzeo and U. Neue, Rapid Comm. Mass Spec., 16 (2002) 805.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 Elsevier Science B.V.All rights reserved

215

CHAPTER 7

Immobilized enzyme reactors in liquid chromatography: On-line bioreactors for use in synthesis and drug discovery Nektaria M a r k o g l o u ~ and Irving W. Wainer ~'2 JDepartment of Experimental Medicine, McGill University, Montreal, Quebec, Canada 2National Institute on Aging, National Institutes of Health, Baltimore, MD, USA

7.1 I N T R O D U C T I O N Enzymes are a class of biopolymers that mediate a variety of in vivo processes including oxidative and conjugative transformations as well as intra-cellular signaling pathways. Due to their wide-range of pharmacological activities, enzymes are often the therapeutic target of a drug discovery program. For example, inhibitors of the angiotensin converting enzyme, ACE inhibitors, are an important class of drugs for the treatment of hypertension while protease inhibitors are key agents in the management of HIV+/ AIDS. Enzymes are complex proteins that accelerate the chemical transformation of a substrate into a product. The catalyzed reactions procede through the formation of enzyme-substrate complexes, which lower the kinetic and energetic barriers associated with the chemical transformation. This process is often described in terms of MichaelisMenten kinetics [1]. The Michaelis-Menten theory of enzyme kinetics assumes that enzymatic reactions are multiple-step processes [2,3]. The simplest form of this mechanism is outlined in Fig. 7.1; where E is the enzyme, S is the substrate, [ES] is the enzyme-substrate complex and P is the product. In this process, k~ and k_~ are the rate constants for the forward and reverse reactions, respectively and k 2 is the rate that the [ES] dissociates to E and E In the Michaelis-Menton approach, it is assumed that k 2 ~ k_l and the Michaelis constant (Kin) is defined as: Km= k_~/k~= [S]-[E]/[ES]

(1)

The Michaelis-Menten approach also defines the velocity of the enzymatic conversion (Vm~x) and these two constants are used to describe the enzyme-substrate interaction. References pp. 233-234

Chapter 7

216

iiiiii!i Substrate

+ Enzyme

Enzyme-Substrate Complex

/

Fig. 1. Idealized depiction of the interaction between an enzyme (E) and a substrate (S) to produce a product (P).

The Michaelis constant, K m, is an equilibrium constant and is conceptually similar to the inhibition constant, K~, used to describe the effect of a reversible inhibitor on the activity of the target enzyme. K m and K~ are similar to the affinity constants, Kd and Ka, used to describe ligand-receptor and protein binding processes. Since affinity chromatographic techniques can be used to determine Kd and Ka values, K m and K~ values should also be measurable using this approach. A key experimental variable is the ability to immobilize the target enzyme in a viable format.

7.2 IMMOBILIZED ENZYMES The standard experimental approach in enzymatic studies is based upon the use of enzymes in their soluble forms [4]. There are numerous disadvantages and limitations to this methodology. Enzymes can be costly, unstable, and difficult to recover from reactions and some are only available in minute amounts. Since enzymes are not altered during the reactions they catalyze, it would be beneficial if they could be reused. In addition, in vivo most enzymes are located within a cellular matrix and can be membrane bound. Consequently, many previously reported in vitro assays that utilize solubilized enzymes are not a true reflection of what is occurring in vivo. With the recent advances in biotechnology there has been an increased interest in the development of immobilized enzymes [5-7]. This interest has been stimulated by the observation that immobilization can stabilize an enzyme without significant loss of

Immobilized enzyme reactors in liquid chromatography I

I

I

217

I

ADSORPTIONl[ ENTRAPMENT11 CROSS.LINKINIGICOVALENTBINDINGI

Fig. 7.2. Experimentalapproaches to enzyme immobilization (ref. 8). enzymatic activity. Thus, an immobilized enzyme system can be reused saving money and time. In addition, enzymes can be immobilized into a membrane-like environment, thus creating a better reflection of in vivo activity.

7.2.1 General approaches Numerous experimental methods and supports have evolved over the years for the immobilization of enzymes and other biopolymers. The various methods are classified into two categories, which are based upon the physical or chemical binding of the enzyme to the chosen support [8]. Methods that have been investigated include adsorption on or covalent binding to a solid support, entrapment and cross-linking. These approaches are summarized in Fig. 7.2. Which of the experimental approaches is preferable is a function of the physical and biochemical properties of the enzyme as well as the intended application. Some key aspects guiding the choice of which method to choose are: (1) knowledge of the molecular composition of the active site, in order to avoid loss of enzymatic activity due to binding of the support to the reactive groups within the active site [8]; (2) biochemical properties of the enzyme such as molecular mass, purity, and stability [8]; (3) functional groups, chemical stability, pore size and particle diameter of the support [9]. There is no ideal method or support that can be used for the immobilization of enzymes. The advantages and disadvantages must be weighed against one another in order to proceed with the ideal method/support for the specified task. It is evident that upon immobilization the possibility of changes in the enzymes' physical and chemical properties may be observed. The effects of immobilization on the stability, kinetic properties and productivity of the enzyme all need to be considered.

7.2.2 Enzyme immobilization on chromatographic supports While a number of useful methods have been utilized in the production of immobilized enzyme reactors (IMERs), the most popular are non-covalent entrapment and covalent attachment. Non-covalent entrapment has been achieved using the immobilized artificial membrane stationary phase (IAM-SP) [ 10]. The IAM-SP is derived from the covalent References pp. 233-234

218

Chapter 7

immobilization of 1-myristoyl-2-[(13-carboxyl)tridecanoyl)]-sn-3-glycerophosphocholine on aminopropyl silica, and resembles one-half of a cellular membrane. In the IAM-SP, the phosphatidylcholine headgroups form the surface of the support and the hydrocarbon side chains produce a hydrophobic interface that extends from the charged headgroup to the surface of the silica. With the IAM interphase, enzymes are embedded within the interphase surroundings (Fig. 7.2). This approach has been used with a variety of enzymes [ 11-16]. Covalent attachment of enzymes to chromatographic stationary phases has been accomplished using Glutaraldehyde-E This packing is a wide pore silica that has been covalently clad with a hydrophilic polymer, polyethleneimine [17]. Immobilization of an enzyme onto the interphase results in formation of a Schiff base linkage which can be reduced using sodium cyanoborohydride [17]. This approach has been used with a variety of cytosolic enzymes [ 11,18-21 ] since it corresponds to a non-membrane bound format. Another method for the covalent immobilization of enzymes on a chromatographic support has been described by Zhang et al. [22]. In this approach, membranes containing the target enzyme are biotinylated and adsorbed onto beads containing immobilized streptavidin. This procedure has been used to immobilize recombinant human N-acetylglucosaminyltransferase V.

7.2.3. Effect of immobilization on enzyme stability

Immobilization can either increase or decrease the enzymatic stability of an enzyme. The magnitude and direction of this effect is dependent upon the method of immobilization and the resulting environment of the enzyme produced by the support. For example, immobilization of lipase on inorganic supports produces higher enzymatic stability than immobilization on organic supports [23]. The environment that the enzyme is subjected to may result in protein unfolding and/or dissociation of the enzymes' subunits. This inevitably decreases the enzyme stability. There are several strategies that can be utilized to improve the stability of immobilized enzymes. They include modifying the enzyme structure, derivatization and the use of stabilizing additives. Chemical modifications are one method of affecting enzymatic stability [8]. For example, cross-linking the subunits of enzymes to form a new quartenary structure can increase enzymatic stability. This is routinely accomplished with the use of polyfunctional macromolecular reagents such as polyaldehydes and polyamines [24]. The large molecular size of these reagents and their ability to reach the different residues makes the connection of the subunits feasible. This was demonstrated with glyceraldehyde-3-phosphate dehydrogenase, where cross-linking with bifunctional reagents increased the enzymes' stability [24]. Negative effects arising from the surface environment can be avoided through modifcation of the properties of the enzyme surface. Changes in charge and hydrophobicity are among some of the commonly used approaches. For example, the introduction of hydrophilic groups on the surface of an enzyme molecule reduces

Immobilized enzyme reactors in liquid chromatography

219

contact between any hydrophobic regions on the enzyme surface and water preventing incorrect refolding after reversible denaturation [25]. Further stabilization of enzymes can be achieved by attaching a spacer group also referred to as an "arm" [26]. The spacer is composed of reactive end groups and forms a bond between the enzyme and the chosen support. The reactive groups are prominently amine groups, carboxylic acid or hydroxyl groups. Increased steric freedom of the immobilized enzyme is achieved utilizing this approach and distancing the enzyme from the support reduces any possible steric hindrance. Several considerations have to be made when utilizing this approach such as the spacer needs to be flexible, inert, hydrophilic and present in small amounts [26]. Dextrans are commonly used as spacers and have been shown to produce stable enzyme-dextransupport bonds [8]. Rennin and protein A have been immobilized with the use of dextrans as spacer arms and were shown to display higher capacities of recognition of their substrates [26].

7.2.4 Effect of immobilization on enzyme kinetics

Michaelis-Menten kinetic parameters of immobilized enzymes can be determined using the same general approaches developed for the study of solubilized enzymes. This includes the effect of temperature and pH on enzymatic activity. However some general considerations must be taken into account since immobilization can introduce new difficulties not associated with the free enzyme. Depending upon the method of immobilization and properties of the support, there may be diffusion-related restrictions associated with the immobilized enzyme [27] and a decrease in enzyme mobility can also affect the mobility of substrates and cofactor. The mass-transfer of substrates and products influences the reaction system and masstransfer resistance can arise due to the location of the enzyme in the support or due to the large particle size of the immobilized enzyme. Under these conditions, the immobilized enzyme operates under diffusion-limiting conditions as opposed to reaction-limiting conditions whereby diffusion layers that form around immobilized enzymes govern its catalytic rate [28]. In this instance, the movement of a substrate from the bulk solution into the unmixed liquid layer surrounding the immobilized enzyme and then through to the active site represents the diffusion layer. A thin diffusion layer in contrast to a thick layer results in limited mass-transfer effects. These include decreasing the particle size of the support, reduction of the enzyme load, and manipulation of the binding of the enzyme to the support [9]. However, flow rates appear to be the key parameter. Mass-transfer resistance has been shown to decrease with increased flow-rates and increased stirring [29]. As such, in the kinetic analysis of immobilized enzymes in flow reactors (IMERs), the investigation of the effect of flow-rate is important in the determination of Km and Vmax of the immobilized enzyme. The effect of flow-rate on enzymatic activity is illustrated by the changes in the observed enzymatic activity of an immobilized D-glyceraldehyde-3-phosphate dehydrogenase enzyme reactor (GAPDH-IMER) [21]. GAPDH catalyzes the oxidative References pp. 233-234

Chapter 7

220

8.00E+08 1 7.00E+08 -~ 6.00E+08 5.00E+08 -1a < 4.00E+08 z (~ 3.00E+08 c).

2.00E+08 1.00E+08

O.OOE+O0

r ............

0

i

i

i

i

0.2

0.4

0.6

0.8

-

-

1

flow rate (ml/min)

Fig. 7.3. The effect of flow-rate on the observed enzymatic activity of glyceraldehyde-3-phosphate dehydrogenase (GADPH) expressed as the area of the NADH produced during the oxidative phosphorylation of D-glyeraldehyde-3-phosphate.

phosphorylation of D-glyceraldehyde-3-phosphate (D-GA3P) to produce 1,3-diphosphoglycerate (1,3-DPGA) and the activity of the enzyme can be monitored by following the production of NADH. The effect of flow rate on the production of NADH was determined using flow rates from 0.1 ml/min to 0.8 ml/min, reflecting substrateenzyme contact times from 8 rain. to about 1 min, respectively. The flow rates between 0.1 and 0.4 ml/min produced the greatest amounts of NADH, Fig. 7.3, which is consistent with a longer reaction time.

7.2.5 Effect of immobilization on the thermal stability of an e n z y m e

Immobilized enzymes have been shown to display increases in thermal stability as illustrated by the effect of temperature on the activity of phenylethanolamine Nmethyltransferase (PNMT) in the immobilized and non-immobilized states, Fig. 7.4 [20]. While the initial enzymatic activity is lower for the immobilized PNMT, relative to the non-immobilized enzyme, the activity remains comparatively stable over the range of experimental temperatures. The increased stability may be a result of the fact that immobilization limits the thermal movement of the enzyme at the higher temperatures. As a result, thermal denaturation may not occur at higher temperatures with an immobilized enzyme. Thermostable enzymes allow for higher reaction rate, lower diffusional restrictions, increased stability and greater yields. Strategies to further improve the thermostability of immobilized enzymes in industrial sectors include the use of genetic manipulation and protein engineering [30,31 ]. For example, with protein engineering techniques thermostable proteases have been produced allowing for their use at higher temperatures offering the added advantage of higher reaction rates, and higher product yields [32]. Other examples

~..,. t~

1.75-

Non-immobilized Immobilized

1.50-

4~

A

0

E r

Ik.

t~

c~

1:IIII;: iiigi~ii

E"

1.25-

0.75-

~~

.......

i .00"

c~ :.E!!!! ::::::: -::::::

EE~!i

~}E}E

E"

C~

:::::;:

0.50-

0.25-

.00 ....

FJ!~':" ...:..:::

:III:II:

t;:l;.'" :..::-.::

22.0

~:

30.0

37.0

45.0

60.0

70,0

Temperature

(oc)

Fig. 7.4. The effect of temperature on the enzymatic activities of immobilized and non-immobilized phenylethanolamine N-methyltransferase (Fig. 3 [ 18]).

222

Chapter 7

include the use of immobilized beta-galactosidase in the dairy industry for the production of lactose-hydrolysed milk [33], thermostable biocatalysts in beer brewing (malting of barley) and in the production of cheese flavor (proteases and peptidases) [341. 7.2.6 The effect of immobilization on the enzyme's response to pH Enzymes are complex proteins composed of charged hydrophilic and neutral hydrophobic constituents. The tertiary structure of an enzyme is produced by intramolecular interactions that include hydrogen bonding, pi-pi stacking, disulfide bridging and electrostatic interactions. The electrostatic interactions are sensitive to the pH of the surrounding environment and pH changes in this environment can result in changes in the pattern of charges on the enzyme, and, consequently, a change in the teritary structure of the enzyme. Finally a change in the tertiary structure of the enzyme can alter the active site producing either an increase or decrease in enzymatic activity. Therefore, the effect of pH on the structure of an enzyme is most often reflected in bell-shaped pHactivity profiles in which the enzyme displays maximal activity at an optimal pH. The shape of the curve and the optimal pH is dependent upon the enzyme. The immobilization of an enzyme on a solid support can radically change the enzyme's microenvironment. This effect has been studied and it has been demonstrated that the use of charged supports can cause shifts in an enzyme's optimum pH [35]. For example, the immobilization of trypsin on a cation-exhange carrier shifted the pH optimum [36]. In addition, the immobilization process itself can alter the pH of the enzyme's microenvironment. The effects of the backbone and immobilization process are illustrated by the pH activity profiles for the enzyme dopamine [3-hydroxylase as the solubilized enzyme, after immobilization through entrapment on an immobilized artificial membrane (IAM) stationary phase and after immobilization by covalent attachment to a glutaraldehyde-P stationary phase [ 11 ]. Each form of the enzyme displayed the standard bell-shaped pHactivity profile, but the pH maxima were shifted, Fig. 7.5. A pH optimum of 5.5 was found for the non-immobilized enzyme while the IAM immobilized enzyme had an optimum pH at 7.0 and immibilization on the glutaraldehyde-P support resulted in a pH optimum of 6.0. There are a number of approaches to the solution of reaction-produced or solid support-produced pH changes. These include changing the type and ionic strength of the immobilization buffer, altering the particle size of the support, and co-immobilization of enzymes that consume any protons generated from the immobilization or enzymatic reaction [9]. 7.3 ON-LINE IMMOBILIZED ENZYME REACTORS (IMERs) 7.3.1 Biochromatography Liquid chromatography and in particular high performance liquid chromatography (HPLC) is often seen as simply a tool for the separation, identification, quantification

223

Immobilized enzyme reactors in liquid chromatography

A 0.75E3 ~

r

0 e,,j

0.50-

0.25-

,.__,

0.00

L

4;s ......

3.5

....

6;5

7;5

8;5

9~5

plt

0.3-

0.2-

l

I=

0

0.1-

0

0.0

,J.

pit

C 0.15 0.10~

0.05 0 0

0.00 -0.05

-

pI-I Fig. 7.5. The effect of pH on the enzymatic activity of dopamine [3-hydroxylase as the (A) non-immobilized enzyme; (B) immobilized on an artificial immobilized membrane stationary phase; (C) immobilized on a gluteraldehyde-P stationary phase (Fig. 2 [26]).

and preparation of chemical substances. However, the basic mechanisms and interactions that govern the chromatographic process resemble mechanisms found in biological processes. For instance, intermolecular interactions such as electrostatic and References pp. 233-234

224

Chapter 7

hydrogen bonding, are common in the biological and chromatographic environments. Thus, within the intimately connected solute-mobile phase and solute-stationary phase interactions lay the ability to utilize HPLC as a probe of biological and pharmacological phenomena. This is particularly true when biomolecules are included in the chromatographic system [37]. When the stationary phase contains an immobilized biopolymer, the chromatographic process has been labeled as biochromatography [37-39]. Biochromatography provides a rapid, simple and precise approach to the investigation of the interactions between small ligands and biomacromolecules. For example, this approach has been utilized to examine serum protein binding [40,41], ligand-receptor [42,43] and ligand-drug transporter interactions [43-45]. The method provides a number of advantages including: (1) the biopolymers remain constant thus limiting the source of experimental error; (2) the immobilized biopolymer stationary phases can be produced using only a small quantity of the target biopolymer.

7.3.2 On-line IMERs

When enzymes are immobilized on chromatographic supports, the resulting IMERs can also be placed in liquid chromatographic (LC) formats. Some of the advantages of an LC-IMER format are: (1) they are easy to automate and control; (2) they are an ideal reflection of biological processes [12]; they can be used for direct synthesis and purification [13-16,46]. An LC-IMER system is usually composed of the IMER connected through a switching valve to an analytical or preparative column. One such system is presented in Fig. 7.6 where an IMER containing immobilized dopamine [3hydroxylase has been coupled to an analytical system containing a phenylboronic acid

Waste 1

v)-4 PJAI

PUMP 1

Recorder

PUMP 2

Detector

ODS

Waste 2

I Waste 3

PUMP 3

Fig. 7.6. Schematic representation of an on-line LC-IMER system containing a dopamine [3-hydroxylaseIMER (DBH-IMER), a phenylboronic acid trapping column (PBA), in-line coupled cynao-C~8columns, an injector (i) and two switching valves (SV) (Fig. 2 [50]).

225

Immobilized enzyme reactors in liquid chromatography B

OCT

~! ..... IS 1

o

5

lO

,

,

15

20

!

Time (Min)

0

-

-

i

5

=

I

10

!

15

. . . . . . .

!

20

Time (Min)

Fig. 7.7. Representative chromatograms from the on-line hydroxylation of tyramine by the dopamine [3hydroxylase-IMER, where: (A) control and (B) experimental run (Fig. 3A,B [50]).

precolumn, for trapping substrate and product, and the precolumn has been connected to cyano and C18 analytical columns coupled in series [ 11 ]. Representative chromatograms from the on-line conversion of tyramine to octopamine, including control, are presented in Fig. 7.7. IMERs can be used either under optimal buffer/pH conditions or investigations into the IMERs activity under varying conditions can be carried out. The acceptable buffer type, pH and solvent conditions will vary amongst the different biopolymers. In general, high percentages of organic solvents render the biopolymers inactive. Similarly the buffer type and pH can alter the IMERs activity. For example, sodium acetate buffer [0.1 M, pH 5.5] renders a dopamine beta-hydroxylase IMER active whereas a sodium phosphate buffer at the same concentration and pH has been shown to inhibit the enzyme [ 11 ]. 7.3.2.1 On-line Michaelis-Menten kinetics using an LC-IMER format

Studies with the IMER-HPLC systems have shown that the activities of the immobilized enzymes reflect the non-immobilized enzymes. Thus, IMER-HPLC systems can be used to carry out standard Michaelis-Menten enzyme kinetic studies and to quantitatively determine enzyme kinetic constants such as K mand Vmax [ 1 1,47]. These References pp. 233-234

Chapter 7

226

TABLE 7.1 KINETIC PARAMETERS FOR NON-IMMOBILIZED (PNMT) AND THE IMMOBILIZED PHENYLETHANOLAMINE N-METHYLTRANSFERASE REACTOR (PNMT-IMER) PNMT

PNMT-IMER

0.109 1.136

0.384 0.292

14.17 1.249

7.31 0.424

Normetanephrine K m (raM) Vm~x (~mol/mg/min)

SAM K m (p~M) Vmax (t, mol/mg/min)

systems can also be used to identify specific inhibitors, to provide information regarding the mode of inhibition and to calculate the K~ of the inhibitor [ 11,12,19,47]. This is demonstrated by the IMER-HPLC system based upon phenylethanolamine Nmethyltransferase (PNMT) which was used to carryout both quantitative and qualitative determinations [47]. The PNMT-IMER was linked to two coupled analytical HPLC columns through a switching valve, similar to the basic system presented in Fig. 7.6, and used for on-line N-methylation of known substrates. The optimal mobile phase for the PNMT-IMER consisted of potassium phosphate buffer [0.1 M, pH 8.30] with a flow rate of 0.2 ml/min. A mobile phase consisting of potassium phosphate buffer [25 mM, pH 2.00] was used for the coupled analytical columns to achieve the desired separation of product from substrate. The PNMT-IMER could be utilized to measure both the affinity (expressed as the Michaelis-Menten constant, Kin) and the enzymatic activity (expressed as maximum velocity, Vmax) of the enzyme, Table 7.1. This is carried out by carrying out injections on a series of substrate/cofactor mixtures. The PNMT-IMER could also be utilized to investigate known inhibitors and to designate the relative affinities of potential inhibitors. The inhibition of the P N M T IMER was carried out using injections of a series of substrate/cofactor/inhibitor mixtures. The inhibitory effect of two PNMT inhibitors, S-adenosyl-L-homocysteine (SAH) and methyldopa was investigated for both PNMT-IMER and non-immobilized enzyme. Fifty percent inhibition was achieved at similar concentrations for both enzyme forms, Table 7.2. The IMER could therefore be utilized for the screening and characterization of potent inhibitors. The inhalation constants (K~'s) of inhibitors can also be determined with the use of IMERs. Alebic-Kolbah et al. have demonstrated this using an IMER-HPLC system based upon alpha-chymotrypsin. The IMER was used to determine K~'s and other kinetic parameters of ACHT inhibitors including the inhibition mechanism [12].

7.3.2.2 Application of lMERs to on-line enantiospecific synthesis and purification IMERs have been used for on-line enantiospecific synthesis and separations. In this approach, the analytical columns in Fig. 7.6 would contain a chiral stationary phase and

Immobilized enzyme reactors in liquid chromatography

227

TABLE 7.2 THE EFFECT OF KNOWN INHIBITORS ON THE ACTIVITY OF NON-IMMOBILIZED PHENYLETHANOLAMINE N-METHYLTRANSFERASE (PNMT) AND ON THE PHENYLETHANOLAMINE N-METHYLTRANSFERASE IMMOBILIZED ENZYME REACTOR (PNMT-IMER) (n = 3) IC5o Inhibitor Methyl-dopa SAH

PNMT

PNMT-IMER

10.4 IxM 40.1 txM

7.6 + 0.2 txM 50.5 + 1.5 IxM

the IMERs would be based upon hydrolytic or co-factor dependent enzyme. When the IMER contains a hydrolytic enzyme, the substrate is usually a racemic mixture. The chiral synthesis is based upon an enantioselective hydrolysis of the substrate. For example, IMERs containing lipase from Candida cylindracea or Candida rugosa (Lipase-IMER) have been coupled to a human serum albumin or a 3,5-dimethylphenylcarbamate chiral stationary phase and used to hydrolyze esters. The results from the enzymatic hydrolysis of racemic-naproxen chloroethyl ester on an L i p a s e - I M E R LC system are presented in Table 7.3 [15]. The enzymatic activity of the L i p a s e - I M E R was compared to the activity of the nonimmobilized enzyme by following the hydrolysis of ibuprofen and ketoprofen methyl esters and phenylethanol chloroacetyl ester, Table 7.4. The results demonstrate that immobilization increased the hydrolytic activity of lipase, which had an affect on the observed enantioselectivity. For phenylethanol chloroacetyl ester, the initial time frame of the study was too long since all of the ester had been hydrolyzed by the 2 h time period. When the flow rate of the substrate through the column was set at 1.0 ml/min, the substrate and enzyme were in contact for less than 10 min. In this period, 56% of the racemic ester was hydrolyzed with an enantioselectivity, expressed as a percentage of enantiomeric excess (ee) of 36% as compared to an 8% conversion with an ee of 30% achieved with the non-immobilized lipase.

TABLE 7.3 DEGREE OF ENZYMATIC HYDROLYSIS OF (R,S)-NAPROXEN CHLOROETHYL ESTER ON A LIPASE IMMOBILIZED ENZYME REACTOR-LC SYSTEM AND THE ENANTIOSELECTIVITY OF THE REACTION AS A FUNCTION OF TIME [ 15] Time (h) 2 3 4 6 7 10 14

References pp. 233-234

Percent Conversion

Enantioselectivity (ee)

18.8 17.7 19.1 33.2 31.1 30.6 46.1

59.2 58.2 54.0 51.9 55.7 56.7 52.4

Chapter 7

228

TABLE 7.4 COMPARISON OF HYDROLYTIC ACTIVITY AS A PERCENTAGE OF CONVERSION (%C) AND ENANTIOSELECTIVITY (EE) OF NON-IMMOBILIZED LIPASE (LP) AND A LIPASE IMMOBILIZED ENZYME REACTOR (LP-IMER). {I.W. Wainer, unpublished data} 2h

6h

24h

Ibuprofen methyl ester LP LP-IMER

%C ee %C ee

6 62 23 93

17 85 50 86

38 92 55 60

%C ee %C ee

8 30 100 0

37 20 100 0

91 2 100 0

%c ee %c ee

1 98 25 97

5 100 27 96

23 97 42 98

Phenylethanol chloroacetylester LP LP-IMER

Ketoprofen methyl ester LP LP-IMER

An IMER containing horse liver alcohol dehydrogenase ( H L A D H - I M E R ) was created through the non-covalent immobilization of the enzyme on the I A M - S P [ 13,46]. H L A D H is a co-factor dependent enzyme that uses NADH to reduce prochiral ketones to chiral alcohols. The H L A D H - I M E R was connected on-line to a column containing a p-methylphenylcarbamate derivatized cellulose chiral stationary phase (OJR-CSP) and the coupled system was used to reduce (R,S)-2-phenyltetrahydropyran-4-one {(R,S)-I, Fig. 7.8} to the corresponding alcohols trans-(2S,4S)-2-phenyl-tetrahydropyran-4-ol {(2S,4S)-2, Fig. 7.8 } and cis-(2R,4S)-2-phenyl-tetrahydropyran-4-ol { (RS,4S)-3, Fig. 7.8 }. The H L A D H - I M E R could also be taken off-line and the mobile phase recirculated through a solution containing glucose-6-phosphate dehydrogenase and glucose6-phosphate. The second enzyme system regenerates the NADH consumed in the H L A D H mediated reduction of the substrate. In this manner, the reaction could be carried out for over 48 h. A chromatogram of the reaction mixture following an 18 h online reaction is presented in Fig. 7.9. The data clearly shows the enantiospecific production of (S,S)-2.

7.3.2.3 On-line study of complex biological systems using coupled IMERs Studies with an IMER containing the NADPH-dependent enzyme horse liver alcohol dehydrogenase (HLADH) indicated that the confluence of cofactor, substrate and supported protein does not pose any difficulties and that on-line co-factor regeneration

229

Immobilized enzyme reactors in liquid chromatography

_OH Fast Reduct~/~

(S)-1

No ~ Reduction

(2S,4S)-2

OH .~

(2S,4R)-3

OH

OH

Reduction

,,~

(2R,4R)-2

Fig. 7.8. The horse liver alcohol dehydrogenasemediatedreduction of (R,S)-2-phenyltetrahydropyran-4-one {(R,S)-I} to the corresponding alcohols trans-(2S,4S)-2-phenyl-tetrahydropyran-4-ol {(2S,4S)-2}, and cis(2R,4S)-2-phenyl-tetrahydropyran-4-ol{(RS,4S)-3}. Reprintedfrom [13]. can be accomplished [13]. The HLADH-IMER LC system contained an NADPH regenerating system. The development of this system demonstrated the applicability of on-line immobilized enzyme reactors in the study of complex enzyme systems. The reproducibility and applications of IMERs are governed by the method of immobilization chosen and the properties of the support and of the enzyme. Thus, IMER LC systems can be expanded to include many of the key metabolic enzymes and systems. Indeed, IMERs based upon non-solubilized rat liver microsomes have been developed [ 14,16]. In addition, solubilized microsomal enzymes on cyanogen bromide-activated Sepharose beads including cytochrome P450 enzymes and UDPglucuronyltransferase [48] or just UDP-glucuronyltransferase [49] have been developed. These systems were active and able to produce glucuronides as well as catalyze the N-demethylation of ethylmorphine and the O-demethylation of p-nitroanisole [ 16,48]. References pp. 233-234

bO

.

i

(s,s)-

(R,S)-3

9. . . . . . . . . . . . . . . . . . . . . .

0

O

10

(R)-I

(S)-1

..

.

.

.

:.:

20

;

30

..........

._.._

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

9

40

, .....

,-,..-,-,-,,..

......

. . . . . . . .

50

•

t

60

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

9

70

Fig. 7.9. The chromatogram of the 18 h incubation of (R,S)-2-phenyltetrahydropyran-4-one { (R,S)- 1 } on a HLADH-IMER followed by separation of the substrate and the products, trans-(2S,4S)-2-phenyl-tetrahydropyran-4-ol { (2S,4S)-2 } and cis-(2R,4S)-2-phenyl-tetrahydropyran-4-ol { (RS,4S)-3 } on the coupled LC column containing a p-methylphenylcarbamate derivatized cellulose chrial stationary phase (OJR-CSP). Reprinted from [ 13].

~ ~t,,,,. -~ "-4

231

Immobilized enzyme reactors in liquid chromatography

Biological processes are complex and involve numerous steps. The analysis of these multi-step processes can prove to be difficult and time-consuming. The use of immobilized enzymes to investigate these processes has proven to be very useful. Immobilizing multiple-enzyme systems have been previously reported. However, these studies dealt with the co-immobilization of a variety of enzymes, i.e. the enzymes were immobilized together on one support [50]. Although this may prove useful in basic research the method would be difficult in the investigation of different enzymes consisting of different concentrations and reaction conditions. An ideal example is the co-immobilization of hexokinase and pyruvate kinase within microcapsules [51 ]. Differences in K m and Vmax were visible when the coimmobilized enzymes were compared to the individual immobilized enzymes. This method of co-immobilization requires the balance of too many variables and makes quantitative determinations difficult. The applicability of a liquid chromatographic system based upon coupled IMERs to organic synthesis, biochemistry and pharmacology has also been investigated [47]. Coupled multiple enzyme systems allow synthetic chemists to add or subtract the enzymes necessary to achieve a particular synthetic goal. Such systems allow for on-line chromatographic purification, structural identification of products and provide a number of approaches to basic research into synthetic and metabolic pathways. More importantly, a coupled system using extremely different enzymes with incompatible cofactors and reaction conditions provides a unique opportunity to explore interrelationships between enzymes, which would not be possible, if they are co-immobilized.

PUMP 2 PUMP 1

ODS

Waste 1

.... Detector

"

I

Recorder ]

Waste 2

-Mp I

Fig. 7.10. (A) Schematic representation of an on-line phenylethanolamine N-methyltransferase immobilized enzyme reactor (PNMT-IMER) LC system. (B) Schematic representation of a dopamine beta-hydroxylase immobilized enzyme reactor (DBH-IMER) and phenylboronic acid (PBA) system that can be incorporated to the existing PNMT-IMER system for the on-line synthesis of epinephrine from dopamine (Fig. 2 [47]). References pp. 233-234

t~

A NE

DA

.

"

0

1i-

5

-

.

.

.

.

i

--

i. . . . . . .

it

. . . . .

i t

-

10

Time (Min)

15

20

0

5

10

15

20

Time (Min)

Fig. 7.11. On-line synthesis of epinephrine (EP) from dopamine (DA) with the intermediate metabolite norpeinphrine (NE) using the on-line DBH-IMER/PNMTIMER system (Fig. 4 [47]).

Immobilized enzyme reactors in liquid chromatography

233

The biosynthetic pathway that transforms dopamine into epinephrine involves the hydroxylation of dopamine by dopamine beta-hydroxylase (DBH), which produces norepinephrine, followed by the phenylethanolamine N-methyltransferase (PMNT) catalyzed methylation of norepinephrine producing epinephrine [47]. Connecting a DBH-IMER and a PNMT-IMER using switching valve technology, Fig. 7.10, produced a coupled IMER system that mimics this pathway. The injection of dopamine onto the DBH-IMER produced epinephrine in the eluent from the analytical column placed at the end of the coupled-IMER system, Fig. 7.11 The system demonstrated the ease at which vastly different enzymes can be united in a single on-line system.

7.4 IMMOBILIZED ENZYMES AND IMERs IN DRUG DISCOVERY

Advances in genomics, proteomics and combinatorial chemistry have produced a vast number of potential new drugs. Indeed, modern drug discovery has been transformed by the automation and industrialization of research techniques, high throughput screening techniques and bioinformatics. The pharmaceutical industries demand innovative technologies that accelerate the development of high-priority compounds through the drug discovery cycle. Immobilized biopolymer-based liquid chromatographic phases have proven to be ideal probes of biochemical and pharmacological properties governing drug-biopolymer interactions. Thus, their importance in modern drug discovery is evident. Acceleration of the drug development process is important for two main reasons. It is primarily important for the rapid discovery of new therapeutic agents to meet the unmet needs of patients with various diseases, and secondly for the obvious economic benefits for the pharmaceutical industry [52]. High-throughput models that can assess and eradicate unfavorable properties from vast quantities of potential drug candidates are vital. However, rapid processes for the analysis and screening of drug-biopolymer interactions are inadequate [52]. Techniques that can characterize these interactions without the need to isolate and dissociate drug/biopolymer interactions are ideal innovations in reducing the drug discovery timeline. In this area, IMERs provide an excellent tool for on-line screening for substrates and inhibitors of a single enzyme or a multiple enzyme system.

7.5 REFERENCES

1 2 3 4 5 6 7 8

M. Dixon and E.C. Webb, Enzymes, 3rd ed. Academic Press New York (1979) pp. 16-23. D.B.Campbell, Psychopharmacology, 100 (1990) 433. G.L.Atkins and I.A. Nimmo, Anal. Biochem., 104 (1980) 1. E. Katchalski-Katzir,Trends Biotechnol., 11 (1993) 471. A.S.Hoffman, Biomater. Artif. Cells Artif. Organs., 18 (1990) 523. J.E Liang,Y.T. Li and V.C. Yang, J. Pharm Sci., 89 (2000) 979. A.P.F Turner, L. Karube and G.S. Wilson, Biosensors Fundamentals and Applications. Oxford University Press, Oxford (1987) pp. 770-775. G.F.Bickerstaff, Immobilization of Enzymes and Cells, Humana Press, Totowa, New Jersey (1997) pp. 1-40.

234 9 10 11 12

13 14

15 16

17 18 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 51 52

Chapter 7 W. Tischer and V. Kasche, Trends Biotechnol., 17 (1999) 326. C. Pidegon, Enzyme Microb. Technol., 12 (1990) 149. N. Markoglou and I.W. Wainer, J. Biochem. Biophys. Methods, 48 (2201) 61. T. Alebic-Kolbah and I.W. Wainer, J. Chromatogr. A, 653 (1993) 122. V. Sotolongo, D.V. Johnson, D. Wahnon and I.W. Wainer, Chirality, 11 (1999) 39. T. Alebic-Kolbah and I.W. Wainer, J. Chromatogr., 646 (1993) 289. X-M. Zhang and I.W. Wainer, Tetrahedron. Lett., 34 (1993) 4731. T. Alebic-Kolbah and I.W. Wainer, Chromatographia, 37 (1993) 608. S.R. Narayanan, S.V. Kakodkar and L.J. Crane, Anal. Biochem., 188 (1990) 278. P. Jadaud, S. Thelohan, G.R. Schonbaum and I.W. Wainer, Chirality, 1 (1989) 38. S. Thelohan, E Jadaud and I.W. Wainer, Chromatographia, 28 (1989) 551. N. Markoglou and I.W. Wainer, Anal. Biochem., 288 (2001) 83. M. Bartolini, V. Andrisano and I.W. Wainer, J. Chromatogr. A, in press. M.M. Zhang, D.C. Palcic, G. Schreimer, M. Alvarez-Manilla, A. Pierce and O. Hindsgaul, Anal. Biochem., 299 (2001) 173. J.M. Moreno and J.V. Sinisterra, J. Mol. Catal., 93 (1994) 357. V.M. Balcao, C. Mateo, R. Fernandez-Lafuente, EX. Malcata and J.M. Guisan, Biotechnol. Prog., 17 (2001) 537. V.V.Mozhaev, Trends Biotechnol., 11 (1993) 88. G. Penzol, E Armisen, R. Fernandez-Lafuente, L. Rodes and J.M. Guisan, Biotechnol. Bioeng., 60 (1998) 518. G. Goldstein, J. Chromatogr., 129 (1976) 61. L. Giorno and E. Drioli, Trends Biotechnol., 18 (2000) 339. J.E Castner and 1.B. Wingard Jr., Biochemistry, 23 (1984) 2203. M. Schulein, Biochem. Biophys. Acta, 1543 (2000) 239. S.J. Taylor, R.C. Brown, P.A. Keene and I.N. Taylor, Bioorg. Ed. Chem., 7 (1999) 2163. B. Van den Burg, G. Vriend, O.R. Veltman, G. Venema and G.H. Eijsink, Proc. Natl. Acad. Sci. USA, 95 (1998) 2056. T.G. Park and A.S. Hoffman, J. Biomed. Mater. Res., 24 (1990) 21. S. Hagedorn and B. Kaphammer, Ann. Rev. Microbiol., 48, (1994) 773. J.D. Chapman and H.O. Hultin, Biotechnol. Bioeng., 17 (1975) 1783. G. Goldstein, J. Chromatogr., 129 (1976) 61. I.W.Wainer, R. Kaliszan and T.A.G. Noctor, J. Pharm. Pharmacol., 45 (Supplement 1) (1993) 367. I.W.Wainer, J. Chromatogr., 666 (1994) 221. R. Kaliszan and I.W. Wainer, in: K. Jinno (Ed.), Chromatographic Separations Based On Molecular Recognition (pp. 273-302). Wiley-VCH, New York, 1997. E. Domenici, C. Bertucci, E Salvadori and I.W. Wainer, J. Pharm. Sci., 80 (1991) 164. T.A.G.Noctor, M.J. Diaz-Perez and I.W. Wainer, J. Pharm. Sci., 82 (1993) 675. Y. Zhang, X. Xiao, K.J. Kellar and I.W. Wainer, Anal. Biochem., 264 (1998) 22. R. Moaddel, L. Lu, M. Baynham and I.W. Wainer, J. Chromatogr. B., 768 (2002) 41. Y. Zhang, E Leonessa, R. Clarke and I.W. Wainer, J. Chromatogr. B., 739 (2000) 33. L. Lu, E Leonessa, R. Clarke and I.W. Wainer, Molec. Pharmacol., 658 (2001) 1. D.V.Johnson, D. Wahnon, V. Sotolongo and I.W. Wainer, Chimica Oggi., 15 (1997) 104. N. Markoglou and I.W.J. Wainer, J. Chromatogr. A, 948 (2002) 249. J.E Lehman, L. Ferrin, C. Fenselau and G.S. Yost, Drug Metab. Dispos., 9 (1981) 15. R.A. Gilissen, J.H. Meerman and G.J. Mulder, Biochem. Pharmacol., 43 (1992) 2661. T.M. Chang, Methods Enzymol., 136 (1987) 67. J. Campbell and J. and T.M. Chang, Biochim. Biophys Acta, 397 (1975) 101. M.T. Cancilla, M.D, Leavell, J. Chow and J.A. Leary, Proc. Natl. Acad. Sci. USA, 97 (2000) 12008.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 Elsevier Science B.V. All rights reserved

235

CHAPTER 8

Use of liquid chromatography-mass spectrometry in acute human toxicology M.J. Bogusz King Faisal Specialist Hospital and Research Centre, Riyadh, Kingdom of Saudi Arabia

8.1 I N T R O D U C T I O N Toxicology may be generally defined as the scientific search for a causal relationship between exposure to a given compound and an observed adverse biological effect. Analytical toxicology provides all the required technical tools necessary for the detection and quantification of toxic compounds, o r - in other words - for the assessment of the exposure. These tools, as well as the whole analytical strategy, may be different in various toxicological disciplines. In clinical toxicology, particularly in acute poisonings, there are some factors of critical importance, like speed of analysis, possibility of detection of active metabolites, and efficient screening in clinically unclear cases of poisoning. LC-MS fulfils at least two of above-mentioned conditions. For this reason, and also due to the gradual introduction of low-cost LC-MS instruments, this technique is now finding an increasingly more important place in clinical toxicology. The main purpose of this review is to present the most important recent applications of LC-MS in clinical toxicology, as well as applications from other disciplines, which may be of clinical relevance. In addition, some general trends of method development, which are pertinent for clinical toxicological analysis, are also discussed here. This chapter is not intended to b e - and cannot b e - an up-to-date catalogue of liquid chromatographic-mass spectrometric procedures used in clinical toxicology. This is due to two reasons: 9 Clinical toxicological analysis is very closely related to other fields where modern analytical techniques have been applied, like, e.g., in forensic or environmental toxicology, in food analysis, or in doping control. Therefore, it should be kept in mind that the LC-MS methods published in some seemingly remote field might find very rapid and direct application in emergency toxicological analysis; References pp. 267-270

236

Chapter 8

9 LC-MS itself is now in the early phase of technical maturity, when more and more laboratories are starting to use LC-MS instruments on a daily basis. Each day therefore brings new applications for numerous toxicologically-relevant compounds. Therefore, the references cited in the text, covering principally the period of 1999-1ate 2001, cannot be regarded as a complete list of all the clinical toxicological applications of LC-MS.

8.2 METHODICAL CONSIDERATIONS Recent years have brought some publications on general aspects of LC-MS, which are of relevance for clinical toxicological analysis. These studies have concerned the optimization of chromatographic analysis, application of various ionization sources and various mass analyzers in toxicology.

8.2.1 Optimization of chromatographic analysis The composition of the mobile phase may greatly affect the sensitivity of LC-MS analysis. Temesi and Law [1] studied the influence of mobile phase on electrospray response in ESI-MS, using 35 various acidic, neutral, and basic drugs as test substances. Methanol and ACN were used as organic modifiers, and ammonium acetate, ammonium formate and TFA were used as electrolytes. Generally, in positive ionization mode, methanol gave stronger signals than ACN. Electrospray response decreased with increasing concentration of ammonium acetate/formate. Very large quantitative individual differences between drugs were observed. The authors concluded that a thorough optimization of all eluent parameters is essential for single analyte analysis. According to Naidong et al. [2] reversed-phase chromatography, although widely used, is less compatible with MS-MS detection than normal-phase chromatography. This is particularly true for highly polar compounds, which are hardly retained on reversed-phase columns, even when the mobile phase is comprised mostly of water. The low percentage of organic phase modifier adversely affects the sensitivity of MS detection due to non-optimal spraying conditions. In contrast, normal-phase columns are used with a mobile phase containing a high percentage of organic solvent, assuring better dispersing and evaporation of electrospray droplets. This was demonstrated in a series of comparative experiments with nicotine, cotinine and albuterol, analyzed with LC-ESI-MS-MS after separation on C18 and silica columns. The mobile phase for reversed-phase chromatography contained 10% ACN, whilst the value for the normalphase eluent was 70%. The sensitivity observed after the normal-phase separation was distinctly higher and the retention times were longer, which also prevented the ion suppression resulting from matrix interference (Fig. 8.1). A similar approach was used by the same group of authors for the analysis of morphine and its glucuronides in serum [3], for fentanyl [4] and for hydromorphone [5]. Other authors have used columns that enable the use of high percentage of organic modifier in the mobile phase. In an automated LC-ESI-MS method for the

Use of liquid chromatography-mass spectrometry in acute human toxicology 6.25e4 cps

XlC of +MRM (4 pairs): fo...

10000

0

=

C G}

20000"

5000

. . . . .

1.40e4 cps

BAM

ca10000 r,3 w q/}

(R r (J

C

_?

2000-

BAM

0

| V

2 Time, rain XIC of +MRM (4 pairs): fo...

C Q

1

1000

!

O

-

2 2.21e3 cps

XIC of +MRM (4 pairs): fo...

"~

5000"

15000

|

1'

Time, rain

XlC of +MRM (4 pairs): fo...

1.34

L_ 1 . 4 7

j

2 Time, rain

C

ALB

C

!

m

~'0

D= U

40000

C

1.17e4 cps

XIC of +MRM (4 pairs): fo...

~'2 ALB

,

237

1 Time, rain 1.70e4 cps

XlC of +MRM (4 pairs): fo...

,,

.N~

g

2 2.42e3 cps

2000

10000

"~

5000

1000

9.44 . . . . . . . !

1 Time, min

2 Time, rain

XIC of +MRM (4 pairs): fo... r

r(.1

1.0e5-

1.36e5 cps

15000- 0. 8

r

,,,.,

GOT

~oooo-

.=. ._ e-

1.71e4 cps

XIC of +MRM (4 pairs): fo...

2.

COT

2

5.0e4

~ ._..

I

2 Time, rain

5000 ~ I

Time, rain

Fig. 8.1. Comparison of LC-MS/MS of albuterol (ALB), bamethan (BAM), nicotine (NIC), and cotinine (COT) on Hypersil silica column (left panel) and Hypersil BDS C18 column (right panel). From ref. [2] with permission of Elsevier Science.

References pp. 267-270

238

Chapter 8

determination of heroin and its metabolites monoacetylmorphine (6-MAM) and morphine in urine, Katagi et al. [6] employed a cation exchange column, using ACNammonium acetate (70:30) as a mobile phase. The use of a mobile phase containing high concentration of ACN was stressed. In a further study the same group [7] applied the cation exchange column for the analysis of selegiline and its metabolites in urine using LC-ESI-MS. Once again a mobile phase with high (70%) acetonitrile content was used for the separation. Studies using rapid, high-throughput, LC-MS analysis are performed mainly in the pharmaceutical industry. However, such studies are of paramount importance not only in drug development, but also in emergency toxicology. Fast separation may be achieved on various ways: 9 through the use of high percentage of organic modifier in mobile phase and use of normal phase or ion exchange columns; 9 through the use of fast and efficient reversed-phase or normal phase columns (short, small particle size or monolithic). Zweigenbaum et al. [8] demonstrated the feasibility of rapid sample preparation and fast analysis by LC-ESI-MS-MS. As model drugs, six benzodiazepine derivatives were used. The drugs were extracted from plasma using 96-well plates and separated on a C18 15 mm x 2.1 mm (particle size 3 Ixm) column in less than 30 s. As a result 1152 samples could be analyzed in less than 12 h. Cheng et al. [9] have studied the possibility of the rapid LC-MS-MS analysis of a mixture of various compounds. Amitriptyline, ibuprofen, lidocaine, naproxen, and prednisolone were used as model drugs. From theoretical considerations, a column capable of ultrafast liquid chromatography should provide a good peak capacity at short gradient run time and high flow rate. The study was done on XTerra MS C18 columns 2 0 m m x 2 . 1 mm (particle size 2.5 txm), 30 mm x 2.1 mm, (particle size 3.5 Ixm), and 50 mm x 2.1 mm (particle size 5.0 txm). The drugs were eluted in ACN-acid or ACN-base gradient. The gradient duration time (5% to 100% ACN) was 0.7 to 1.7 min., and the flow rate was 1.5 ml/min. Both MS-MS and UV were applied for detection. In these conditions the drug mixture was separated in less than one minute (Fig. 8.2). The study of electrospray response at different pH values showed that one of the drugs examined (lidocaine) did not follow the rules of electrospray theory. The authors stated that the optimization for particular drugs should be done checking different solvents, buffer types and pH values, since best results are sometimes achieved outside the expected range. A new approach to high-speed HPLC was presented by Cabrera et al. [10], who developed a new silica-based monolithic type column (Chromolith, Merck). This column is a silica rod with pore volume of 1 ml/g and specific surface area of 300 m2/g. Since Chromolith columns show very low back pressure flow rates up to 10 ml/min, are possible without sacrificing the efficiency of separation. The monolithic column 50 mm x 4.6 mm was used by Wu et al. [11] for separation of plasma extracts of a mixture of temazepam, tamoxifen, fenfluramine, and alprazolam. A baseline separation of drugs was achieved at a flow rate of 6 ml/min, in ACN-water (1:1), with a back pressure of only 61 bar. Certainly, these columns may have a future not only in drug discovery programs, but also in emergency toxicology.

%

o

o

~

o

_

'

.

.

.

.

.

oo.o~1762 lx50mm

.

.

.

....

--

O . O & E

C5 r~

2~

~/~ r%

002~

O.OzO 0.0'~

6

0.010

.,.q

0.000 -0.005

o.oo-o.o6

(:D E

0~ L _

0

<

:I

.

.~o

'

~7-"

.'

.-

c~

~-

.

9

.

-

.."

.

,<

0 .~0 O. r

I

~

O.OSO 0.041~

J

0.040 o 0:36 0.030 0.026 0.0~20 0.0"I 0

$

2 lx30mm 3 5 pm

c~

0 ~

O.OOG

o. oo0

.,,o.oo~.~

0

05

0

10

0

15

0

20

0 25

0

30

0.3~

O,ao

0

4~

.

.

.

.

.

.

.

u

=: 0 . ~ O.Oe~

o"~'.-o 2 l x 2 O m ~ . ~2/~3 ~ 5pm ] O.C]~Q

9

0.0"15 0.010

0.006 O.OOQ -O,~l

~.oo

,,, oo6

-10 1 O

o18

, , o..2o

o

~l

--o3~]

~,, . ~."t~

. . D40

. 046

5

4 O~D

08Jk

~a43

n ~

n 7n

r~ T ~

o mo n'A~

c~ ~

n ~s

9c~

Time (rain) Fig. 8.2. Fast separation of lidocaine (1), prednisolone (2), naproxen (3), amitriptyline (4) and ibuprofen (5) on ODS columns. The particle size is scaled down in proportion to the column length. Run time 1 min., UV detection. From ref. [9] with permission of John Wiley & Sons, Limited.

t,J t~ ~D

240

Chapter 8

8.2.2 Use of different ionization sources. Use of single- and triple-quadrupole instruments A general trend has been observed in the literature towards the use of electrospray ionization in toxicological analysis. Most probably, this is caused by an overwhelming use of this source in the LC-MS of large molecules. However, the choice of ESI or APCI may be of relevance for particular substances. As a rule, the response of ESI for very polar substances is higher than that of APCI. On the other hand, less polar compounds, like a majority of parent drugs, need an active ionization mode for better detectability. It is therefore a good practice to try both ionization sources in the method development stage of the assay. Unfortunately, this approach is not often followed or not reported in the literature. It goes without saying that the use of tandem mass spectrometry (MS-MS) instead of a single stage quadrupole instrument gives obvious advantages in regard to specificity and sensitivity. The main obstacle to the more widespread use of L C - M S - M S is the high capital cost of the instrument. However, more recently some manufacturers have launched bench-top tandem instruments of lower cost, which may be applicable in clinical toxicological laboratories. The reader interested in development trends of L C MS instruments may be referred to reviews of Niessen [12,13].

8.2.3 Use of various mass analyzers Clauwaert et al. [14] applied the L C - E S I - T O F - M S for quantitative measurements of MDMA and its metabolite MDA in body fluids. The LOQ was 1 Ixg/1 and the linear dynamic range extended over four decades. It was concluded that this technique achieves the same linear dynamic range as LC-ESI-MS-MS. Zhang and Henion [15] compared the applicability of L C - E S I - T O F - M S and L C E S I - M S - M S for the quantitative analysis of idoxifene in human plasma. The drug and its deuterated analogue, as internal standard, were isolated from plasma using semiautomated 96-well hexane extraction and separated on 30 mm x 2 mm CN column. This column separated the target compound from the matrix which had not been possible with ODS column. For the TOF-MS instrument, an exact mass of the protonated quasimolecular ion was measured (m/z 524.1441), whereas the MS-MS instrument was used in SRM mode, using m/z 524.2 as a parent ion and m/z 97.9 as a product ion. The comparison showed that LC-MS/MS technique was about 10 times more sensitive than L C - T O F - M S (the LOQs for idoxifene were 0.5 Ixg/1 and 5 Ixg/1, respectively) with both methods showing a satisfactory dynamic range. In conclusion, the authors stated that L C - T O F - M S might be used to complement L C - M S - M S in certain cases (Fig. 8.3).

8.3 A P P L I C A T I O N S OF L C - M S IN C L I N I C A L T O X I C O L O G I C A L ANALYSIS Chromatographic procedures for illicit and therapeutic drugs were reviewed recently in volume II of this Series [ 16]. LC-MS methods were also included there. Therefore, the

% 472

SO1 DoubleBlank 40] m/z 524.2--> m/z 97.9

18.1 components

:_1|

tO0 519.1

B

8O

is

Blank

6O

r

r

9.07e4

IS

B Blank m/z 529.2->m/z 97.9

I r~ r~

m/Z 529.176•

4O

2O

~ o

!c 0---

.-

- - T

....

-.

100

r

q

0j

0

27.0

100]

C

Id

~~1 Loo=o.s ~,m,

xifene

o

~

/~

472

m/z 524.144r

0

" 0

i 0.2

9 -.-A -

j 0.4

A /"V~-~

9 0.6

--e 0.8

-

, 1.0

Retention Time (Min)

9 1.2

! 1.4

/ -

0 0

0.2

0.4

0.6

0.0

1.0

1.2

1.4

Retention Time (Min)

Fig. 8.3. Comparison of the response for idoxifene and idoxifene-d5 (IS) analyzed in the same conditions on LC-ESI-TOF-MS (left panel) and L C - E S I - M S - M S (right panel). From ref. [ 15] with permission of Elsevier Science. t,~ 4~

242

Chapter 8

present review will focus mainly on recent studies published in the years 1999-2001. It should also be stated that the division into illicit and therapeutic drugs is not always sharp. There are many drugs, particularly among narcotic analgesics, stimulants or tranquilizers, which may qualify to be in both categories.

8.3.1 Illicit drugs

8.3.1.1 Opiate agonists The LC-MS procedures used for opiate agonists since 1999 were reviewed by Bogusz [ 17], and by Pichini et al. [18]. In the last two years, several methods for determination of morphine, codeine and its corresponding glucuronide metabolites were published. The methods were generally based on solid phase extraction and ESI-MS or E S I - M S MS detection and are summarized in the Table 8.1 [3,19-22]. Determination of free morphine and both its glucuronides in body fluids may be helpful in the interpretation of a given case from a forensic and clinical point of view. Thus, a high free morphine fraction generally indicates acute poisoning in its very early stages, particularly in a person who did not taken heroin or morphine chronically. Also, the differentiation between pharmacologically active morphine-6-glucuronide (M6G) and the inactive 3-glucuronide (M3G) is of practical importance for interpretation of the severity of poisoning. Cailleux et al. [23] extracted opiate agonists (morphine, 6-monoacetylmorphine (6-MAM), codeine, norcodeine, pholcodine, codethyline) as well as nalorphine and cocaine and its metabolites (benzoylecgonine, ecgonine methyl ester, cocaethylene and anhydromethylecgonine) from blood, plasma or urine with chloroform/isopropanol (95:5) at pH 9. The drugs were separated on an octyl column using a mobile phase composed of ACN-ammonium formate-formic acid. Protonated molecular ions and one fragment for each substance were monitored using ESI-MS-MS. Quantification was done using deuterated internal standards and the LOQs were 10 p~g/1 for opiates and 5 p~g/1 for cocaines and were higher than those reported after solid phase extraction. Katagi et al. [6] developed an automatic method for determination of heroin and its metabolites 6-MAM and morphine in urine. Urine samples were applied on trapping cation exchange column, washed with ammonium acetate and after column switching the drugs were eluted and separated on analytical cation exchange column in ACNammonium acetate (70:30). The detection was done with ESI-MS in full scan or SIM mode. Protonated quasi-molecular ions or acetonitrile adducts (M + H + ACN) § were monitored with LODs ranging from 2 to 30 p~g/1in full scan mode and from 0.1 to 3 p~g/ 1 in SIM (Fig. 8.4). Recently there have also been more publications concerning the determination of synthetic or semisynthetic opiates. Gaulier et al. [24] reported a suicidal poisoning of a 25-year-old male heroin addict with a high dose of buprenorphine. Buprenorphine (BU) and its active metabolite norbuprenorphine (NBU) were determined in body fluids and organs with LC-ESI-MS after deproteinization and SPE. In gastric contents only BU was found, at a concentration of 899mg/1. In selected matrices the following concentrations were found: in blood BU 3.3 rag/l, NBU 0.4 rag/l, in bile BU 2035 rag/l,

% TABLE 8.1 LC-MS METHODS USED FOR DRUGS OF ABUSE t..a

Drug M, M3G, M6G M, 6-MAM, Cod, NorCod, Pholcod M, M3G, M6G, NorM M, M3G, M6G, 6-MAM, Cod, C6G M, M3G, M6G M, M3G, M6G Buprenorphine, BUG, NBUG Ketobemidone, Nor-K 14 Amphetamines and rel. comp. 5 Amphetamines 6 Amphetamines Cocaine, BZE, EME Cocaine, BZE, EME, ECG THC, OH-THC, THCCOOH THC, OH-THC, THCCOOH THCCOOH THCCOOH, THCCOOH-G LSD, Nor-LSD LSD, Nor-LSD LSD, O-H-LSD LSD, O-H-LSD LSD, O-H-LSD

Sample

Isolation

Column, elution conditions

Detection

LOD (Ixg/1)

Ref

Plasma Plasma, urine Serum, urine Serum Serum Plasma Plasma Urine Plasma Urine Saliva Urine Plasma Plasma, urine Plasma, urine Urine Urine Urine Blood, urine Urine Urine Blood, urine

SPE L/1 SPE SPE SPE SPE SPE SPE SPE SPME SPE SPE SPE SPE SPE SPE L/1 SPE SPE L/1 + SPE L/1 + SPE L/1 + SPE

Silica, ACN-HCOOH isocr. C8, ACN-HCOONH4 isocr. C18, ACN-HCOOH grad. C18, ACN-HCOONH4 grad. C18, ACN-HCOONH4 isocr. C 18, ACN-HCOOH isocr. C18, ACN-HCOONH4 g r a d . C8, ACN-HCOOH grad. C18, ACN-HCOONH4 isocr. CN, ACN-CH3COONH4 isocr. C18, ACN-CH3COONH4 isocr. C18, ACN-H20 grad. C8, MeOH/ACN-CH3COONH4 isocr. C8, MeOH-HCOONH4 isocr. C8, MeOH-HCOONH4 isocr. C18, MeOH-CH3COONH4 isocr. C8, ACN-HCOONH4 grad. C18, ACN-CH3COONH4 isocr. Phenyl, ACN-CH3 HCOONH4 C18, ACN-CH3COONH4 isocr. C18, ACN-CH3COONH4 g r a d . C18, MeOH-HCOONH4 grad.

ESI-QQQ, MRM ESI-Q, SIM ESI-QQQ, MRM ESI-Q, SIM ESI-QQQ, MRM ESI-QQQ, MRM ESI-QQQ, MRM ESI-Q, SIM APCI-Q, SIM ESI-Q, SIM ESI-QQQ, MRM ESI-QQQ, MRM ESI-QQQ, MRM APCI-QIT, SIM APCI-QQQ, MRM ESI-Q, SIM ESI-QQQ ESI-Q, SIM ESI-QQQ, MRM APCI-QIT, SIM APCI-QIT, SIM ESI-Q, SIM

0.5-1.0 10 0.3-2.5 0.5-5.0 1.0-5.0 0.25-0.5 0.1 25 1-5 0.4-0.8 1 0.5-2 2.8-4.4 1 0.25 0.5 10 0.5 0.025 0.2 0.4 0.1-0.4

3 6 19 20 21 22 25 26 30 34 36 37 39 41 42 44 45 49 50 53 54 55

I

t.,..

t,,~~

A b b r e v i a t i o n s : M = m o r p h i n e , M 3 G = m o r p h i n e - 3 - g l u c u r o n i d e , M 6 G = m o r p h i n e - 6 - g l u c u r o n i d e , 6 - M A M = 6 - m o n o a c e t y l m o r p h i n e , C o d = codeine, C 6 G = c o d e i n e - 6 - g l u c u r o n i d e , Pholcod = p h o l c o d i n e , B U G = b u p r e n o r p h i n e glucuronide, N o r - K = n o r - k e t o b e m i d o n e , B Z E - b e n z o y l e c g o n i n e , E M E = e c g o n i n e m e t h y l ester, E C G - ecgonine, T H C - t e t r a h y d r o c a n n a b i n o l , THCCOOH-G

= THCCOOH-glucuronide,

O - H - L S D = 2 - o x o - 3 - h y d r o x y - L S D , Q = single stage q u a d r u p o l e , Q Q Q = triple stage q u a d r u p o l e , Q I T = q u a d r u p o l e - i o n trap t..a ta.a

t'.3 4~

TABLE 8.2 LC-MS METHODS USED FOR SELECTED THERAPEUTIC DRUGS AND PESTICIDES Drug Midazolam, 1-OH-midazolam Flunitrazepam, 7-AF, 3-OHF, N - D F Flunitrazepam, 7-AF, N-DF Flunitrazepam Diazepam, N-D, Clonazepam, 7-AF, N-DF Fluoxetine, norfluoxetine Diuretics (32 compounds) Sulfonylureas (5 compounds) Pesticides (14 compounds)

Sample

Isolation

Column, elution conditions

Detection

LOD (txg/l)

Ref

Serum Serum, urine Blood, urine Serum Serum, urine Plasma Urine Serum Serum

L/1 SPE SPE L/1 SPME L/1 SPE SPE SPE

C 18, ACN-HCOONH4 gr. C 18, ACN-HCOONH4 isocr. C 18, MeOH-NH4OH C 18, ACN-H20 isocr. C 18, MeOH-CH3COONH4 isocr C 18, ACN-HCOOH isocr. C8, ACN-H20 grad. C 18, MeOH-CH3COOH grad. C 18, ACN-HCOOH grad.

ESI-Q, SIM APCI-Q, SIM ESI-IT, SIM ESI-IT, SIM ESI-Q, SIM ESI-QQQ, MRM ES I-QQQ, MRM ESI-Q, full scan ESI-Q, SIM

0.2-0.5 0.2-1.0 0.5-1.0 0.2 0.02-2 0.15 1 2 5

56 57 58 59 60 62 74 80 91

Abbreviations: 7-AF = 7-aminoflunitrazepam, 3-OHF = 3-OH-flunitrazepam, N-DF = norflunitrazepam

~z

%

t,,a ,,q

I

"]

--

A

t,o

..

i

.~_

-

2

3

6

m/z

150

m/z

136

rn/z 370 .

.

m/z 342

(a)

m/z 328 rrgz 300

%1""0 ~ . ....

./'v .

.

.

.

.

.

_ .

3

.

_/~6 .

m/z 286

r~ r~

m/z

r5

150

~ m / z 1 3 6

.

m/z 370

_

m/z 342

.

(b)

rm,

m/z 328 ' ----- m/z 300

. . . . . . . .

~

. ,0~

..~,,_~ . . . . . . . ~.~ . . . . . ,;,~. . . . . . . ,.~ . . . . . . . ~ . . . . . . . ,_~ . . . . . . . ,;,~ . . . . . . . ~_~. . . . . . . ~ .

......

3

. . . . . . . ./~

, .

.

.

.

.

.

.

.

l-r

6 m/z 370

~2 . . . . . .

m/z 286

~_.__

m/z 328 m/z 300

(c)

m/z 286

Fig. 8.4. L C - E S I - M S of acetylcodeine (1), monoacetylmorphine (2), codeine (3), methamphetamine (4), amphetamine (5), morphine (6) from urine in the full scan mode (a), serum in the full scan (b) and SIM (c) mode. From ref. [6] with permission of Elsevier Science 9

to

246

Chapter 8

NBU 536 mg/1, in brain BU 6.4 mg/1, NBU 3.9 mg/1. Besides BU and NBU, high concentrations of 7-aminoflunitrazepam were found in blood (1.2 mg/1), urine (4.9 mg/ 1) and gastric contents (28.6 mg/1). Polettini and Huestis [25] have developed an L C - E S I - M S - M S method for the determination of buprenorphine and its metabolites: norbuprenorphine and buprenorphine glucuronide in human plasma (Fig. 8.5). SPE extraction with C18 cartridges and gradient elution was applied to this analysis. For buprenorphine, norbuprenorphine as well as for deuterated analogues used as internal standards, the protonated quasi-molecular ions were monitored whilst for buprenorphine glucuronide the protonated quasi-molecular ion and buprenorphine aglycone were used. The LOQ was 0.1 Ixg/1 for all compounds. On the basis of the transition m/z 590---~414 norbuprenorphine glucuronide was also tentatively detected. The reference standard of this compound was not available so that the identification could not be confirmed. Naidong et al. [5] published an LC-ESI-MS-MS procedure for the determination of hydromorphone in plasma. The drug and deuterated analogue were extracted with solvent and separated from its glucuronide using a 50 mm x 2 mm silica column and a mobile phase consisting of ACN-water-formic acid (80:20:1). The LOQ was 50 ng/1. The same group determined fentanyl in plasma, using automated 96-well solid phase extraction, straight-phase chromatography and ESI-MS-MS [4]. The LOQ for these compounds was 50 ng/ml plasma, based on 0.25 ml sample volume. It should be stressed that the group of Naidong is using straight phase columns and mobile phases containing high concentrations of organic modifier for opiate agonists. Ketobemidone is a synthetic opioid agonist and narcotic analgesic, which is frequently abused, particularly in Scandinavian countries. Breindahl et al. [26] developed a LC-ESI-MS method for determination of ketobemidone and its demethylated metabolite in urine. An approach using mixed-bed SPE cartridges was used for isolation, with a recovery of over 90%. Protonated quasi-molecular ions for both substances as well as three fragments for ketobemidone were monitored with an LOD of 25 lxg/1. Svensson et al. [27] published similar procedure for ketobemidone and its metabolite norketobemidone and reported the LOQ of 0.7 txg/1 for both substances. The enantiomers of tramadol and its active metabolite O-desmethyltramadol were determined with LC-APCI-MS-MS [28]. The substances were isolated from plasma with automated SPE and separated on a Chiralpak AD column in isohexane-ethanoldiethylamine (97:3:0.1) mobile phase. The transitions of protonated quasi-molecular ions of drugs and internal standard (ethyltramadol) to the same ion m/z 58 were monitored. In this study other tramadol metabolites could also be detected. Juzwin et al. [29] determined tramadol N-oxide and its major metabolites in the plasma of experimental animals using LC-ESI-MS-MS. The LOQ was 6 Ixg/1 and the method was applied in preclinical studies.

8.3.1.2 Amphetamines Bogusz et al. [30] isolated fourteen amphetamines and related compounds from biofluids with SPE cartridges and subjected the extracts to LC-APCI-MS examination in SIM mode. The limit of detection ranged from 1 to 5 txg/1. This method was applied in routine casework [31]. Benzphetamine and its metabolites benzylamphetamine,

247

Use of liquid chromatography-mass spectrometry in acute human toxicology XIC of +MRM o

(6 pairs):

for 644.5

/ 468.2

ainu from 042800m018

1.32e3

cps

2.41 e4

cps

1.27e5

r

1000

s~176 1

_

, ""

~

,

I

5.0

6.0

Time, rain

XIC of +MRM (6 pairs)" for 414.2 / 414.2 ainu from 042800m018 = ~ 150001

-

~ '~176176176 1 c

5000

2 !

5.0 XIC of +MRM (6

pairs): for

6.0

Time, rain

417.2 / 417.2 ainu from 042800m018

1.0e

~. r

m

c _= _.r

5.0e41 __._.--.=_ 9

= , i ,

5.0

ii

!

i

6.0

Time, rain

XIC of +MRM (6 pairs): for 468.2 / 468.2 ainu from 042800m018

1.72e4

,~ ~sooo]

>,

10000

C

e

cps

1

5ooo1~ 9

=

|1

.

,'1

5.0

Time, rain

,

.

_

6 I.0

XIC of +MRM (6 pairs): for 472.2 1 472.2 ainu from 042800m018

.

|

2.34e5 cpe

o.o>~ 2e5] '~c"Ce

le51 ,

=

I

5.0

,

--

ii

Time, min

I

6.0

i

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

1oo~q

A9_THC_OH

5o ~ o-~ 100

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

!

A9-THC-OH

,o

0 ~.

.

.

.

.

.

.

331 ~295

.

.

.

.

.

.

.

~~176 i~o A9-THC-COOH ~ O

0

o

331~313

--4

.

.

.

.

.

.

.

.

.

.

345=~299

.

~'100 e-"

x~ 50 '4=

A9-THC-COOH .

0

.

.

.

.

A9_THC

50

.

.

~ .

.

345=~327 .

.

315=~259

RT:-4,70

r

0

~oi A9-THC 1~176 0

.........

1oO.o . 9 THe 43 0--~-~ 0,0

~ ~ ~ I ~ ~"~ 0,5

RT: 4,71

315=~193

i

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

RT: 4,69

'

" 2,5 Time (rain)

~ ~ 'I ~ ~'~ 3,0

~ i ~-~ 3,5

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

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,

oo

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

bO bO

RT: 4.95 - 18.00 8.73

100

O-H-LSD

90 80 r

SPECIMEN 47

70

"0 t-- 6O .Q < 50 .=.. ,.e-.,,e

r ~ 40 n~

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

Iso-LSD

30

LAMPA

LSD

20

17.00

Nor-LSD 10 5.31 5.75

5

6

10.08 11.15

6.96 7.71

7

8

9

10

12.60 12.99 iii

11 12 T i m e (min)

BBL-

~

13

14.10 ~I

14

.95

___

15

16

17.27

17

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.

18

Use of liquid chromatography-mass spectrometry in acute human toxicology

253

facilitated sexual assaults. Since flunitrazepam very quickly disappears from blood, simultaneous determination of its active metabolites is very important. Bogusz et al. [59] have developed an LC-APCI-MS method for determination of flunitrazepam and its metabolites 7-aminoflunitrazepam, N-desmethylflunitrazepam and 3-OH-flunitrazepam in blood or plasma. The comparison of ionization response showed that APCI gave a signal 7-30 times stronger than ESI for all compounds except 7-aminoflunitrazepam. After solid phase extraction on C18 cartridges, the analytes were separated on a C18 column using a mobile phase composed of ACN-ammonium formate buffer (pH 3.0) with detection by APCI-MS. For each compound a protonated quasi-molecular ion was monitored. The LODs ranged from 0.2 to 1 lxg/1. This method has been applied in routine toxicological casework [31]. LeBeau et al. determined flunitrazepam, 7-aminoflunitrazepam and N-desmethylflunitrazepam in blood and urine after SPE with mixed phase cartridges [60]. An L C - E S I - M S - M S (ion trap) procedure was applied. Protonated quasi-molecular ions of the drugs involved were monitored and the identity was confirmed by fragmentation. The LODs in blood were from 0.5 to 1 Ixg/ 1. An ion-trap L C - E S I - M S - M S was also used for determination of flunitrazepam in serum by Darius et al. [61 ]. The drug was extracted with tertiary butylmethyl ether and separated on a C18 column with a ACN-water mobile phase. Product ions of flunitrazepam and the internal standard clonazepam were monitored achieving an LOD of 0.2 Ixg/1. The metabolites were not determined. Yuan et al. has described an automated in-tube solid-phase microextraction combined with LC-ESI-MS for the determination of diazepam, nordiazepam, temazepam, oxazepam, 7-aminoflunitrazepam and N-desmethylflunitrazepam in serum and urine [62]. The isolation technique used allowed solvent-free, automatic extraction in 15 min. for each consecutive sample. The extraction procedure was optimized, using six extraction capillaries, and various extraction conditions. The drugs were detected with L C - E S I MS (full scan m/z 100-400) and SIM and LODs of 0.02 to 2 Ixg/1 were achieved. Drawbacks of the procedure related to the relatively low recovery from serum (below 50%) and some peak broadening caused by automatic desorption. 8.3.2.2 Antidepressants and antipsychotics Hoskins et al. [63] have described a method for the determination of the monoamine oxidase inhibitor moclobemide and its two metabolites in human plasma. The analytes were isolated using Bond Elut C18 SPE cartridges at alkaline pH and separated on a Nova-Pak phenyl column. Protonated quasi-molecular ions of all the compounds of interest and the internal standard were monitored by ESI-MS with LODs of 1 to 5 txg/1. The LOQ for all the analytes was 10 lxg/1. A fast E S I - L C - M S - M S method for the determination of fluoxetine and norfluoxetine was developed by Sutherland et al. [64]. The drugs and internal standard (doxepin) were extracted from plasma with hexane-isoamylalcohol (98:2) and the aqueous phase was frozen and discarded. The organic phase was then back-extracted with 2% formic acid, the aqueous phase was frozen and, after discarding the hexane layer, thawed and injected onto the LC-MS system. The transition of protonated quasi molecular ions of drugs to product ions was monitored. The total chromatographic run time was 2.6 rain. and the LOQ was 0.15 txg/1 for both compounds. References pp. 267-270

254

Chapter 8

Olanzapine, a thienobenzodiazepine, is an antipsychotic agent used broadly for the acute treatment and maintenance of schizophrenia. Due to the concentration-related response and toxicity the therapeutic monitoring of this drug is indicated. Berna et al. [65] published an LC-ESI-MS/MS method for determination of olanzapine in plasma. The drug was extracted with SPE cartridges in single or 96-well format and an LOQ was 0.25 Ixg/1 was obtained. Bogusz et al. [66] applied LC-APCI-MS for determination of olanzapine in serum. SPE extraction on C18 cartridges was applied achieving an LOQ of 1 txg/1. In full-scan LC-MS a postulated olanzapine-10-N-glucuronide was found in urine. Aravagiri and Marder [67] applied L C - E S I - M S - M S for therapeutic drug monitoring of clozapine, clozapine-N-oxide and norclozapine in serum of schizophrenic patients. The drugs were isolated with solvent extraction and separated on an ODS column. The transitions of protonated quasi-molecular ions to single fragments were monitored and an LOQ of 1 Ixg/1 for all substances was reported. The same authors used L C - E S I - M S - M S for the determination of risperidone and 9-hydroxyrisperidone in plasma [68]. Solvent extraction and separation on a phenyl-hexyl column was applied with an LOQ of 0.1 Ixg/1. Venlafaxine, a phenethylamine antidepressant, as well as its O-desmethylated metabolite were determined in postmortem blood and tissues in 12 cases of fatal venlafaxine poisoning [69]. The drug was extracted with butyl chloride and determined by LC-ESI-MS in positive ionization mode. In all cases various other drugs were also detected. Kumazawa et al. [70] applied solid phase microextraction followed by L C - E S I - M S - M S for the determination of eleven phenothiazine derivatives from whole blood and urine. The extraction efficiencies were studied and ranged from 0.0002 to 0.12% for blood and from 2.6 to 39.8% for urine, respectively. The sensitivity in selected reaction monitoring mode was described as satisfactory. LC-ESI-MS and L C - M S - M S were applied to the detection of antidepressants and neuroleptics in the hair of psychiatric patients. The drugs (maprotiline, pipamperone and citalopram) were extracted from powdered hair with methanol and purified with SPE. A mass spectral library search was performed for ESI-CID and MS-MS spectra [71 ].

8.3.2.3 lmmunosupressants and antineoplastic drugs Therapeutic drug monitoring of these drugs is of obvious importance, due to their high toxicity, variable bioavailability and consequences of treatment failure due to too low dosage. The anthracycline antibiotics doxorubicin, daunorubicin, epirubicin and idarubicin as well as the active metabolites doxorubicinol, daunorubicinol and epirubicinol were extracted from serum with C18 BondElut cartridges in automated SPE [72]. The analytes were separated on an ODS column (C18 Symmetry 3.5 Ix, 150 mm x 1 mm) with an ACN-ammonium formate buffer pH 3 mobile phase. Quantification was done with ESI-MS (positive ions) in SIM mode against an internal standard (aclarubicin). The LOQs were between 1 and 2.5 Ixg/1. serum. The method was applied to the TDM of doxorubicin and daunorubicin Lensmeyer and Poquette [73] described an LC-ESI-MS method for the determination of tacrolimus in whole blood. The drug and internal standard (ascomycin) were extracted from precipitated and centrifuged blood samples with Empore disc cartridges. Subsequent chromatography was performed on a C18 column with ACN-water at 75~

Use of liquid chromatography-mass spectrometry in acute human toxicology

255

With these conditions the total run time was 1 rain. The drug was detected with ESI-MS using in-source collision-induced dissociation. Protonated sodium adducts of tacrolimus and ascomycin, as well as fragment ions were monitored. The LOQ was 0.3 txg/1. The method was approximately 60% less expensive and was much more sensitive and specific than the immunoassay. Busulfan, a drug used as conditioner before hematopoetic stem cell transplantation, was determined in plasma by LC-ESI-MS [74]. The drug and its deuterated analog was extracted from 0.2 ml plasma with ethyl ether and separated on a C18 column using an ACN-ammonium acetate gradient. Ammonia adduct ions of drug and internal standard were monitored and an LOD of 2 txg/1 was obtained. Pharmacokinetic data from 4 patients were published Steinborner and Henion have developed a high-throughput method (384 samples per 90 min. per one person) for the determination of methotrexate and its hydroxymetabolite in plasma [75]. The drugs were extracted from 0.2 ml plasma using semi-robotic acetonitrile precipitation followed by chloroform extraction in 96-well plates. The samples were then injected onto a C8 guard cartridge and after 20 s the stream was switched to a C8 analytical column (50 mm x 1 mm). The total run time was 1.2 min. with the retention times of both drug and metabolite below 40 s. Detection was performed with ESI-MS-MS, with monitoring of one product ion for each compound. LODs were 0.05 and 0.1 txg/1 Carboplatin, a potent anti-tumor agent was determined in plasma by Hogge et al. [76]. Since carboplatin is extremely water-soluble, an acetonitrile precipitation of plasma was used for sample pretreatment. In this assay the transition of the protonated molecular ion to one product ion was monitored. The LOQ was 100 txg/1.

8.3.2.4 Diuretics Thieme et al. [77] has 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 a C8 column with a gradient of ACNammonium 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 the screening procedure. This was found to be superior to the alternate polarity switching in one run. An example from this work showing the UV, LC-MS and L C - M S - M S of polythiazide, that highlights the increasingly improved specificity of M S - and MS-MS detection, is shown in Fig. 8.8.

8.3.2.5 Cardiac glycosides Digitalis glycosides: digoxin, digitoxin, deslanoside, digoxigenin and digitoxigenin were determined with E S I - L C - M S - M S [78]. The drugs were extracted from whole blood or urine by SPE using Oasis HLB cartridges at alkaline pH. Chromatographic separation was performed on an ODS column with an ACN-ammonium formate

References pp. 267-270

Chapter 8

256 99

UV-Detector

___~

_

Polythiazide_0-3h

, , ~

c <

~--~ ~ -".i.-

-~-'..

7 . . . .

9.5 XlC of-MRM: 438->438 1.6e4

1

7.38

~_. " 0

0 . 0

~

~

-

-

.

.

.

.

.

_

.

.

.

.

.

. .

.

.

.

.

.

.

_

.

.

.

.

.

.

~

_

_

,~-T_-_-~_

_

1 1300

=

_ _

~

. . . .

~

~ _

~

-

9

XIC of-MRM: 438->324

7.38

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

INTERNAL

STANOARD 0.e~

t.m

. 5000

.

.

.

.

.

.

.

.... ;10

2%

115

Time,

DIGOXIGF:: NIN B~S DIGITOXOSIDE

II

2'5

310

~--

l o,%o?,~

H

1.0e5

~

.....

rain

/

G , T~Lo~OX, N

~

)%

5.0e

, ,~

5

10

~

,~

ALPHA ACETYL

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

-~

OLEANDR 25.55

";"

j

LANATOS IDE C 3.74

J

!

310

DIGITOXIGENIN MONO DIGITOXOSIDE

~'IO0x'N 9

2 5

rain

O,GO•

.5 .Oe4.

i

210

Time,

o%

.

15

/i

IN

o,,.OXN

___ '

i' 5

i 10

.

116

Time,

.

. ;0

.

.

.

.

. ;5

310

rain

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.

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

. ~ 100- m/Z:T77 "~

-

9 v,,,~ 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

1.0

Time (rain) ,

,,..

, ,t

2.74e2 cps

141.0 1 91.1

'~ r

200,

MMA

lOO

md

j ~

....

!

0.5

I

Time (rain)

I

1.0

.,

.,

,

139.0 1 91.1

..

2.18o3 cps

2000

~ ~J

1500-

DMA

lOOO

5OO

4

j A

_

.

_

,

i

u O.5

Time (rain)

1.'0

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: 9 With 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). 9 Detection 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. 9 The 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 9 2003 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: 9 The elimination of method development through the use of genetic gradient HPLCMS methods 9 The pooling of samples prior to analysis and the dosing of compounds as cocktails, both in vivo and in vitro 9 The use of short columns with fast isocratic elution in clinical analysis 9 The use of gradient elution with high flow rates 9 The 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%

=

60.0%

o

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.

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

References pp. 291-292

276

Chapter 9

A similar approach can also be applied to column efficiency (using H, the height equivalent to a theoretical plate), although in this case the manipulation of the data is more complex. As well as extracting the silanol contribution to the column inefficiency, termed H~il 90%, it is also necessary to normalise the data for differences in column efficiency due to extra column band broadening (ECBB) [2]. ECBB results in peaks with low retention (typically retention factors (k)< 4) having poorer efficiency than if they had been eluted with a greater retention. Since many columns can display significant differences in retentivity, selectivity and order of elution it is necessary to take this factor into account in making column comparisons based on efficiency. For this reason, interpretation and evaluation of data from longer columns (> 100 mm) is preferred since these give wider peaks (in volume terms), which are less susceptible to extra column effects. These difficulties can be appreciated by considering the following example. One of the compounds studied previously [1], chlorpheniramine, showed retention factors of 11.7, 1.5, 7.9 and 2.0 when the four columns under study were eluted with the same eluent. Even if all the columns were performing similarly with respect to the silanol interactions, the efficiency of chlorpheniramine would always look worse on the second and last of these columns since it is retained the least and ECBB has a greater effect. This variation in efficiency with respect to retention also makes the comparison of the different compounds on a single column difficult. However, using a series of neutral solutes, covering a reasonably wide range of retention (k 1 to 12) it is possible to correct for these ECBB effects [ 1]. Previously we also looked at retention data since it is a widely accepted that silanol interactions can lead to over-retention for basic compounds. However we now believe that this phenomenon may be beneficial, leading to an improved overall separation, providing efficiency and symmetry are not compromised. In fact there is concern that the elimination of silanol interactions could actually lead to a situation where many columns give very similar separations based purely on a partitioning mechanism. Because of this we would now employ the concept of Discriminating Power [3], which measures how well a given column can separate any two compounds selected at random. This concept has been used extensively to evaluate and compare chromatographic systems employed in drug screening [3]. Methods for column and stationary phase evaluation with special emphasis on basic compounds have been subject to review [4,5]. The method discussed above [1] however, does offer a number of advantages since it is applicable to any stationary phase or particle size, using any eluent and gives an unambiguous and quantifiable outcome. The use of the largest set of test compounds that is practicable is strongly recommended since it is rare for any two columns to respond in the same manner with a given set of test compounds. Furthermore, as evaluations based on either efficiency or asymmetry alone can give different results, the use of both parameters is recommended. The use of data normalisation and appropriate data presentation are also seen as essential to an accurate and unambiguous outcome. Recently, Needham and co-workers (e.g. [6]) have reported on the use of cyanopropyl and pentafluoropropyl stationary phases for use in HPLC-MS using isocratic elution. The rationale for the choice of these phases were that they are highly retentive towards

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

277

common drug compounds (tricyclic antidepressants and [3-blockers in the reference cited) and require elution with organic rich mobile phases such as 90% acetonitrile. Since the authors also report that highly organic mobile phases give enhanced response in electrospray ionisation (but see also Section 9.2.2 below) these could also be a good choice for generic methods. The authors make some limited claims for these columns in terms of chromatographic performance but to-date they do not appear to have undergone full evaluation. It is possible also that any standard C8 or C18 column with a high carbon loading and hence high retentivity would perform similarly. 9.2.2 Eluent selection There are many more restrictions regarding the selection of eluents and eluent components for use with HPLC-MS methods compared with HPLC-UV methods. Essentially the buffer constituents have to be volatile so the usual standby, phosphate, is unacceptable, as are the other common eluent constituents such as ion-pair reagents, organic amines, etc. Whilst the use of these eluent components may give acceptable results (at least in the short term), they would rapidly result in fouling of the source of common atmospheric ionisation interfaces. This would lead to loss in sensitivity and significant down-time as the source was repeatedly cleaned. The use of MS detection therefore necessitates a new way of thinking about HPLC eluents. As mentioned above, it is commonly assumed that organic rich eluents favour increased response in ESI MS, the data to support this however is hard to come by and often contradictory. As part of an extensive study on ESI MS, Kebarle and co-workers [7] showed a significant increase in response for cocaine hydrochloride with increasing methanol concentration. In the analysis of penicillins, Straub and Voyksner [8] showed data which indicated both an increase and a decrease in response with increasing methanol concentration in the HPLC eluent. Needham and co-workers [6] showed significant increases in response, up to 10-fold higher in going from 0 to 99% acetonitrile. This work in the last report however was limited to two very narrow chemical series (tricyclic antidepressants and [3-blockers) and there were significant differences in the magnitude of the effects and the way the response changed as acetonitrile was increased [9]. In gradient HPLC this variation in response is a factor which just has to be accepted since the percentage of organic modifier in the eluent is deliberately changed throughout the run. In our laboratory, we have evaluated both the effect of the organic modifier and the eluent buffer/pH modifier on compound response in both positive and negative ion ESI MS [10]. This work, which is summarised below, involved a panel of 35 drug compounds of varying chemical types (acids, bases, neutrals, peptides) and physicochemical properties. The results varied somewhat with the exact nature of the compounds and the conclusions are based on the overall results. Similar studies have been carried out by Jemal and co-workers [11,12] although these each involved single compounds so the results are less useful.

9.2.2.10rganicmodifier Firstly a decision must be made regarding the use of methanol or acetonitrile. Work carried out in the author's laboratory has indicated methanol to be slightly superior to

References pp. 291-292

278

Chapter 9

acetonitrile with ESI since it gives marginally greater response [10]. Methanol is also preferred since it is reported to give slightly better peak shape for basic compounds in RP-HPLC than acetonitrile [ 13] and it is a better solvent for buffer salts [ 14].

9.2.2.2 Buffer and pH modifier In terms of buffering or pH modification, trifluoroacetic (TFA) acid is rarely used in MS detection since it has a very marked suppressive effect on ionisation in negative mode and to a lesser degree in positive mode. Formic acid (0.1 to 0.2%) can be used as a replacement for TFA, although it cannot be considered a direct replacement. It is difficult to achieve the same low eluent pHs with formic acid, without resorting to high concentrations, and the ion-pairing effects which probably occur with TFA are absent. In an evaluation carried out in the author's laboratory, formic acid was found to be slightly superior to ammonium acetate or ammonium formate in terms of response. In a study employing a number of tetracyclines [15], it was concluded that 1.0% acetic acid was the best in terms of giving the highest positive ion response and 50 mM ammonium hydroxide the best in negative ion mode. The lack of a systematic effect with pH in positive ion mode and our own experience would suggest that these findings are compound or chemical series specific. In order to draw general conclusions, supportive of the development of generic procedures, it is important to employ a diverse compound set. Our preference is to use buffers with a pH near neutrality such as that achieved with ammonium acetate or ammonium formate. The reasons for this are several fold: firstly neutral pH buffers benefit the retention of basic or acidic compounds when on-line sample extraction or on-column pre-concentration are used. Secondly neutral eluents give improved retention and retention distribution compared with eluents with extremes of pH (see below). Where high pH eluent are required then ammonia can be used as a source of OH- ions. This latter approach is becoming increasingly interesting with the development and introduction of HPLC stationary phases stable towards high pH.

9.2.2.3 Buffer concentration The process of ionisation in ESI is saturable [16], hence if the concentration of the buffer constituents are too high they will actually compete with the analytes and effectively suppress the ionisation of the analyte in either positive or negative ESI MS. This effect, which is independent of whether single or triple quadrupole instruments are used, is clearly shown in Fig. 9.3, where two of the analytes tested showed a very marked decrease in response as the buffer concentration was increased from 1 to 100 mM. Although the effect is highly variable and very dependant on the exact nature of the compounds tested, these results are typical and suggest that the concentration of ammonium acetate or formate in the eluent should be kept to a minimum. However, this does need to be balanced against the requirements of the chromatography. Low concentration buffers (1 mM) will have insufficient buffering capacity and at this concentration it is unlikely that good chromatographic performance will be attained. Figure 9.4 shows the variation in chromatographic efficiency (N) for a range of compounds with varying concentrations of ammonium formate in the eluent. In this case

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

279

100 I -

Relative

response

~

Warfarin

80 - -

*=

60 ~

=-

k

Caffeine Tolbutamide Paracetamol

40 20

1

10

100

Concentration (mM) Fig. 9.3. Relative ESI MS response for a series of test compounds with different concentrations of ammonium acetate in the eluent. The compounds were tested using a flow injection method.

40000 30000

[]

10 mM

[]

50 mM

l

100mM

N (plates/m) 20000

10000

!~

0

t

1

n

2

t ~':~

3

ii!

~, l

ii~

i~

;2~

n

4 Compound

t

5

,;i~

n

6

....t

7

8

no.

Fig. 9.4. The variation in chromatographic efficiency for a range of test solutes with varying amounts of ammonium formate in the eluent. The column was a Phenomenex Prodigy (100 • 4.6 mm) eluted with 30% methanol in water.

it is the high concentration, which not unexpectedly gives the best overall performance. The optimal buffer concentration in H P L C - M S therefore is a compromise between chromatographic performance (efficiency and symmetry) and MS response. In practice we have routinely used 50 m M a m m o n i u m acetate although as column performance has improved we have reduced this to 10 mM. These factors are in marked contrast to H P L C - U V analysis where the response of the analyte is totally independent of the eluent buffer concentration and where the concentration of buffer constituents is directed very much to attaining the best chromatographic performance.

References pp. 291-292

Chapter 9

280

9.2.3 Effects of eluent pH and compound type Irrespective of what type of buffer system is used, the exact pH of the eluent is important and to a degree this depends on the acid/base properties of the compounds to be assayed. This can be clearly demonstrated by reference to the generic method used in our laboratories which employs a methanol/water eluent containing a m m o n i u m acetate at a pH of approximately 6.8. This method shows a linear relationship between retention and compound logD (obtained from the ClogP and pKa of the compounds) although as might be expected there is a high degree of scatter in the data. With this method we are able to chromatograph compounds over a relatively wide logD range ( a p p r o x i m a t e l y - 1 to 8), which corresponds to a retention range of around 7 to 14 min. In order to understand the effect of pH we need to simulate a compound collection, consisting of different proportions of acids, bases and neutrals with known logD values. To obtain the logD we have used the ClogP (octanol/water) data for 4012 compounds submitted to one of our enhanced throughput screens. The data was more or less normally distributed with a mean ClogP of 4.04, standard deviation of 1.70 and a median of 4.18. To convert the logP to logD we have used the measured pKa data for a sub-set of acids and bases measured in our laboratories. In both cases the pKa distributions were skewed: for acids the median was 6.7 and for the bases the median was 7.2. If we assume that the HPLC system is incapable of retaining compounds with log D < - 2 (equivalent to retention time of approximately 6 min), then we can study the effect of the compounds type distribution (acids/bases/neutral) and eluent pH on the retention of the simulated data set. The data in Table 9.1 shows how the proportion of compounds with logD < - 2 varies as a function of eluent pH using compounds data sets made up of differing proportions

TABLE 9.1 THE PERCENTAGE OF COMPOUNDS WHICH FALL BELOW THE CUT-OFF EQUIVALENT TO A log D - - 2 (RETENTION TIME OF APPROXIMATELY 6 min) FOR 9 DIFFERENT SETS OF COMPOUNDS (n= 10,000) MADE UP OF DIFFERENT PROPORTIONS OF ACIDS, BASES AND NEUTRALS, CHROMATOGRAPHED AT DIFFERENT pHs. Ratio of acids:bases 9neturals in the data set Eluent pH

33"33"33

10"20"70

15"30"55

20"20"60

100"0"0

0" 100"0

0"0" 100

10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.5 2.0

9.8 7.0 4.9 3.8 3.9 5.6 8.4 11.8 13.7 15.6

3.7 2.9 2.4 2.4 2.6 3.6 5.5 7.7 8.9 10.0

5.0 3.8 2.9 2.8 3.4 5.0 7.6 10.8 12.5 14.1

6.2 4.5 3.4 2.9 2.8 3.7 5.5 7.7 8.9 10.0

27.0 18.2 11.4 6.2 3.3 1.9 1.4 1.2 1.2 1.2

1.2 1.5 2.1 4.0 7.7 13.8 22.7 33.5 39.2 44.9

1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1

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

281

of acids, bases and neutrals. The pH range covered is from 2.5 to 10 which includes that obtained with eluents containing formic acid and ammonia as well as the those where the eluent is buffered with ammonium acetate to obtain a near neutral pH. It can be clearly seen from Table 9.1 that the major problems occur with compound sets made up of a high proportion of bases when acidic eluents are used, or with high proportions of acids when basic eluents are used. In both cases the ionisation of these compounds leads to low logD and hence non-retention on our HPLC system. Thus, in the design of generic methods it is important to have an appreciation of the types of compounds to which the method will be applied. It may also be necessary to make adjustments to the method, modifying the eluent pH to match the types of compounds being produced by the various projects. As a general rule, the data in Table 9.1 suggests that eluents with a pH in the range 6 to 7 are a good compromise, minimising the loss in retention of both polar acids and bases.

9.3 S A M P L E P O O L I N G AND C O C K T A I L DOSING

9.3.1 Sample pooling The idea of mixing or pooling samples prior to analysis is a simple one that potentially can give very significant reductions in analysis time. This is obviously only applicable in the drug discovery phase where every compound to be assayed is different. Using HPLC-MS(MS) it is assumed that even if compounds co-elute, the selectivity of the MS detector will ensure accurate and precise results. Without some form of pre-screening to avoid co-elution, this approach is of limited utility with UV or fluorescence detection. The only drawback with this approach is the potential loss in sensitivity. Mixing n samples together reduces the concentration of each compound in the mixture by the factor n, in practice however this is rarely a problem providing n is kept reasonably small ( < 10). In fact, dilution of samples can be advantageous, bringing concentrations into the linear calibration range of the instrument and minimising the potential for ion suppression. Without any pre-screening of compounds prior to pooling, there is potential for misidentification and the production of inaccurate results. However, adopting the precautions outlined below should allow such problems to be avoided. The initial approach adopted in our laboratory involved the use a single quadrupole MS for in vivo studies, an approach which we believe is still viable today given the improvements in sensitivity with such instruments and their low cost and small size. We carried out the studies without any pre-screening of compounds to determine retention time, their preferred ionisation mode or MS response. Using our optimised generic HPLC system we would simply monitor the (M + H) + and the (M-H)- ions, where M is the monoisotopic molecular mass of the compound of interest, i.e. the molecular weight made up of the elements with the lowest isotopic masses, rather than the average molecular mass, which is generally used. Since biological samples are complex, the ion current chromatogram even when using selected ion monitoring will show several peaks, the largest of which may not References pp. 291-292

282

Chapter 9

actually correspond to the compound of interest. Usually it is possible to identify the analyte peak by reference to the chromatograms for the calibration standards prepared in the same biological matrix. Here we would look for the peak that changes systematically with concentration. Unfortunately, not every compound in a given cocktail will be detected and show up as a peak in the chromatogram, either because it does not ionise under the MS conditions used, it formed an ion which we failed to monitor e.g. (M + 2H) 2+, it was unstable in the sample, it decomposed during the workup procedure or its response was so poor that it was lost in the general background noise. This in itself is not a serious issue, but if combined with another problem it can have serious consequences. This second problem is one of spurious peaks. In certain instances, the calibration standards (and the test sample) may show multiple peaks in the chromatogram due to impurities or decomposition products. Some of these may actually be more responsive than the compound of interest and their peak height will probably also change linearly in parallel with that of the parent compound. It is important to remember that even with electrospray ionisation, a compound does not give a single ion at (M + H)+, but actually a cluster of ions around this parent ion. These additional ions are due to the isotopes of the various elements that make up the molecule. For example a compound with a molecular weight of 259 containing 16 carbon atoms would show a major ion at 260 (the (M + H) + parent ion) but also ions at 261,262 and 263 and these would have intensities of approximately 19%, 2%, and 0.2% of the parent ion respectively. These are due to molecules containing one or more carbon j3 atoms: a natural stable isotope of carbon with an abundance of around 1.1%. It should be possible therefore to see the danger of confusion: some compounds failing to give a response and other compounds giving either multiple peaks or multiple significant ions. It is easier to envisage a situation where two compounds A and B with molecular masses 750 and 755 respectively are pooled. The first A responds and is easily detected, the second B, does not respond, but the isotope ion of A at 755 (2.5% of the parent ion) is misidentified as B. For this reason, when mixing compounds for analysis we always ensure that there is at least 4 Da difference between each compound. The above problem is exacerbated with molecules that contain chlorine, which occurs naturally as two major isotopes with masses 35 and 37 Da, having a distribution of 75:25 respectively. The relatively high proportions of both these isotopes leads to great variation in the intensities of the ions resulting from multiply chlorinated compounds. For example going back to the compound mentioned above, substituting 3 of the hydrogens with chlorine atoms gives an average molecular weigh of 385 and a parent ion (M + H) + of 386. What is interesting now however is that the ion 2 mass units up from the parent ion shows an intensity 98% of the parent and that at 4 mass units up, 33% of the parent. The analyst employing SIM and just looking for the parent ion won't actually see these other very significant ions. Thus, where multiply chlorinated molecules are to be mixed, it may be necessary to increase the mass difference from 4 Da to at least 6 Da to avoid problems of mis-identification. This is an approach that has been successfully employed in our laboratory, where compounds are 'tested' singly but analysed as pools of 6 compounds. Although the benefits of this approach are significant, the literature reports are relatively few [ 17-20]

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

283

and surprisingly, two of these have involved fluorescence [19] and UV [20] detection. In addition, the number of compounds mixed for analysis has been relatively low, always < 6.

9.3.2 Cocktail dosing Cocktail dosing, also known as cassette dosing or n-in-one dosing is where multiple compounds are deliberately introduced into a biological system. This approach was first reported by two groups in 1997 [21-23] and was used initially for in vivo studies but more recently it has also been applied to in vitro work [ 17,24]. The rationale behind this approach is similar to that for sample pooling above, in that it saves on analysis and sample manipulations. But importantly, also in the in vivo situation, it significantly reduces the utilisation of animals and can lead to improved data quality since all the compounds have been dosed at the same time to the same animals. Although there are some concerns surrounding this approach, it has become widespread within pharmaceutical drug discovery. The analytical problems encountered are the same as those described for sample pooling, and when selecting compounds for cocktail dosing the same mass rules outlined in Section 9.3.1 are generally applied. However, there are other factors that make cocktail dosing more complex, necessitating additional care in selecting compounds for a given cocktail. The major complication arises since every biological system will transform or metabolise exogenous compounds to some extent. Where this biological system is a whole animal or a metabolically active preparation such as microsomes or hepatocytes then the degree of metabolism can be very significant. Thus, the resulting samples for analysis will not only contain the compound of interest but potentially a range of metabolites. Typically these metabolites will involve well-characterised changes in mass associated with for example, mono-hydroxylation (+ 16), di-hydroxylation (+ 32) and demethylation (-14), etc. As in the example above it is possible to envisage cases of mis-identification where for example, the mono-hydroxy metabolite of a compound with a molecular weight of 400, is mistaken for another compound in the cocktail with a mass of 416, which gave a low response or did not respond at all. Thus, in addition to the mass rule mentioned above, it is recommended that compounds be excluded from a cocktail where the mass difference between any of them is 14, 16 or 32, etc. In our laboratory we have generated software to automatically carry out this selection and grouping process. The software is flexible enough that other masses can be added to the rules quite easily. However, the use of a large number of exclusion masses can create the situation where from a limited list of compounds it can be difficult to create a cocktail which does not break any of the rules.

9.4 SHORT COLUMN HPLC Mass spectrometry as a detector for HPLC was seen as a major advancement. With either HPLC-MS employing single ion monitoring (SIM) or preferably with HPLCReferences pp. 291-292

284

Chapter 9

MS-MS employing multiple reaction monitoring (MRM) it is sometimes possible to eliminate almost totally the signal from endogenous materials or metabolites, thus obtaining chromatograms showing just the analyte of interest. To many analysts this must have been seemed like a panacea, not only did the MS(MS) offer superb sensitivity (ng/ml or less) but it also gave the ultimate in selectivity. Following the introduction of the atmospheric pressure ionisation interfaces ESI and APCI, many analysts, initially in the drug development or clinical analysis arena, rapidly adopted this approach. Late stage drug development is characterised by the generation of several hundreds or several thousands of samples all containing the same analyte which require analysis in the shortest time possible to support clinical trials or bioavailability studies. Thus, the use of MS and MSMS was highly successful and despite the high initial cost, resulted in the development and application of fast and effective HPLC-MS(MS) methods with a marked reduction in the turnaround of results. Because of the significantly improved selectivity of HPLC-MS(MS) compared with HPLC-UV, it was considered possible to make compromises with regard to both the chromatography and the sample preparation. Why work hard to clean up the sample when the detector is blind to all the junk? Sample preparation frequently became rate limiting for some assay procedures and many laboratories made the transition from complex sample preparation procedures such as liquid-liquid extraction or solid phase extraction to relatively simple protein precipitation. Occasionally this work was carried out in 96-well microtitre plates to further improve throughput and facilitate automation [25]. Similarly, why use a long HPLC column (> 100 mm length) if the separation between the analyte and interferents or analyte metabolites was no longer deemed essential? If the MS detector did not detect the interferent then why bother working to separate it? Thus, workers shifted to using relatively short columns (20-50 mm in length), which reduced retention and ultimately overall analysis times. Because of the lower back pressures generated by these columns they could also be run at higher flow rates thus further reducing the total analysis times. This new approach to bioanalysis; short HPLC columns with limited sample clean-up, is reflected in the significant number of publications in the late 1990s where variants of these methods were applied to clinical samples. Whilst this new methodology undoubtedly worked well in some instances, it was not without some serious flaws, which seem to have passed a number of workers by. The assumption was made that the endogenous materials and metabolites the detector were blind to didn't interfere with the analysis. Unfortunately they can and often do! This problem occurs both with MS and MS-MS detection, since it is a result of changes in the ionisation process itself rather than the separation or detection of the ions produced. As mentioned above, the process of ionisation in an ESI source is a saturatable process [ 16]: there are only so many ions that can be produced in the ionisation source at a given time. Furthermore, some compounds have a greater propensity to ionise than others. Thus, the signal for a given analyte, which happens to co-elute with another component (endogenous, metabolite, formulating agent, etc.), may be suppressed (or occasionally enhanced) and the resulting concentration inaccurately measured. The problems of ion suppression have been clearly demonstrated by a number of workers [26-31]. Temesi and Law [30] using HPLC-MS quantified these effects and showed that the signal for

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

110

63

15

250

1000

285

4000

100 Relative 90 response 80for clenbuterol 7 0 60 50

I

0

2

I

I

I

4 6 8 Injection number

I

10

12

Fig. 9.5. Relative response for a series of injections of clenbuterol with and without the injection of coeluting minoxidil. Clenbuterol was injected at a concentration equivalent to 200 ng/ml of plasma and the minoxidil concentrations were varied from 15 ng/ml (injection 1) to 4000 ng/ml (injection 9). Injections 2, 4, 6, 8, 10 and 12 were controls where clenbuterol alone was injected.

the drug clenbuterol (200 ng/ml) could be reduced by around 30% when it co-eluted with another basic drug, minoxidil (4000 ng/ml) and around 10% with only 63 ng/ml of minoxidil (Fig. 9.5). Because the MS or MS-MS instruments are effectively blind to these other materials when used in SIM or MRM modes respectively, the analyst is usually unaware that the analyte of interest is co-eluting with other materials which could be affecting the signal and hence the measured concentrations. This problem of ion suppression is exacerbated through the use of minimal sample preparation, such as precipitation of proteins with either acetonitrile or methanol, which results in relatively 'dirty' extracts. Additionally, as well as employing short HPLC columns to obtain fast analysis through short retention times, compounds were often chromatographed with short retention factors (k) often < 5 [32-36] and occasionally as low as 2 [32,34]. It is generally accepted that a retention factor of around 5 is a good optimum in terms of resolution and sensitivity. With UV detection for example retention factors less than 5 may pose no real problem since the analyst is aware of what may be co-eluting with the compounds of interest. However, the selectivity of MS and especially MS-MS means that the analyst does not see the other material eluting from the column and with low retention factors the potential for co-elution with polar endogenous materials or metabolites is very high. Romanyshyn and co-workers [31] studied the problem of ion suppression using isocratic methods employing organic-rich eluents (80-95%) with compounds of interest eluting with retention factors of 1 to 4. This work showed a large solvent front, which tailed significantly such that the analyte of interest was overlaid on this background of unresolved polar endogenous materials. In such circumstances ion suppression is likely. Although it could be argued that such effects will be constant between standards and samples and thus whilst sensitivity may be compromised, accuracy will not, we have seen significant differences in the background SIM chromatograms from animals treated and bled in different ways. With

References pp. 291-292

Chapter 9

286

short retention and poor resolution it is also possible that the polar metabolites may be incompletely separated from the parent compound, leading to further potential problems of inaccuracy. Under both circumstances the possibility of ion suppression or enhancement and hence obtaining false results is quite high. The ways of avoiding this situation are several fold. Firstly, during the method development phase it is important to check the chromatographic background fight across the chromatogram in full scan mode, i.e. generating a total ion chromatogram. Whilst this may be of insufficient sensitivity to allow detection of the analyte, it will indicate how well the analyte is separated from endogenous materials and to a degree how effective the clean-up or sample preparation has been. It is probably also necessary to recheck this on a regular basis, and also to compare the biological matrix used to prepare samples with those of the test samples since these could differ significantly. Species differences in HPLC chromatograms of control plasma are commonplace even with H P L C - U V and with MS detection this is no different. It is important therefore not to apply a method developed for one species to samples from another species without carrying out some validation and checking of the background across the chromatogram. When metabolites are identified and isolated their retention should also be checked to guarantee that co-elution does not occur. A useful and interesting approach to evaluating ion suppression was introduced in 1999 [28,29] and is now employed in a number of laboratories. This involves the infusion of an analyte solution into the eluent stream post column and prior to entry into the MS source. Without any injection there is a positive standing response for the analyte. An injection of a blank extract is then made and typically negative responses are seen due to suppression of the analyte signal by materials in the extract. An example of the results of this type of experiment is shown in Fig. 9.6. Not surprisingly these studies [28,29] indicate that extracts produced by protein precipitation cause the most ionsuppression. The data also show that the degree of suppression is very much compound dependent [28].

!

Relative response

:

1

Reference

'

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

__J! .......... '~" . . . . . . . . . . . . . . .

0

'." '~J

|

Biological extract

, ....,,..,_,

Time (min)

Fig. 9.6. Results of an experiment to study the ion suppression effects of plasma extracts. The upper trace is a reference obtained by running the gradient with no sample injection but infusing a solution of analyte post column. The lower trace uses the same gradient and sample infusion but in this case a blank plasma extract has been injected. The negative peaks in the signal indicate suppression of the analyte signal due to endogenous materials in the plasma extract which was prepared by protein precipitation.

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

287

9.5 S A M P L E I N T R O D U C T I O N O N T O H P L C

The application of liquid-liquid or solid phase extraction methods usually generate extracts in volatile solvents which can be easily evaporated and reconstituted in mobile phase or a less elutropic solvent, for injection onto HPLC. Occasionally SPE cartridges can be eluted with the HPLC mobile phase thus avoiding an evaporation stage and further facilitating the injection of the extract onto HPLC. Where protein precipitation is used as the method of sample preparation the situation is more complex. Most workers precipitate proteins using 2 volumes of either acetonitrile or methanol, to one volume of plasma. The subsequent extract therefore contains around 66% organic solvent and 33% water. Whilst such extracts are directly compatible with reversed-phase eluents, it is not possible to make large injections of these extracts. This is due to the fact that they contain a high concentration of organic solvent and they will be analysed using gradient elution with a low starting percentage of organic modifier. Evaporation and redissolution is not always easy since these extracts are nowhere near as simple to evaporate as say an ether or hexane extract. Furthermore, because these extracts are relatively dirty, the residue following evaporation consists of a considerable amount of solids that can make re-dissolution difficult. There is potential for the analyte to be occluded by the solid residue, consequently reducing recovery. The strategy we have adopted in our laboratory to overcome this problem and allow the injection of a large proportion of the extract is to dilute the extract with water prior to injection and then make a large injection (1 ml) of this onto HPLC. Although at first consideration this may seem rather strange, it is a logical approach that has proved highly effective in our hands. The rationale behind this method of sample workup can be understood by considering the data shown in Fig. 9.7. This is taken from published work [37] and the Ph.D. thesis of one of the authors [38]. The figure shows the effect on peak height of indomethacin

50 4.2% MeOH

40

Peak 30 height 20

42% MeOH

10 0

50 100

200 300 Injection volume (ul)

400

500

Fig. 9.7. The effect of sample injection solvent and sample injection volume on analyte peak height using a reversed-phase column (120 x 1.0 mm, Spherisorb 5ODS1) eluted with methanol/phosphate buffer (50/50) at a flow rate of 0.071 ml/min. The test compound was indomethacin which eluted with a retention factor of 4.5.

References pp. 291-292

288

Chapter 9

as a function of the injection volume, employing two different samples. One sample consists of indomethacin dissolved in 42% methanol and the other the same compound dissolved in 4.2% methanol. The peak height on the y-axis could be replaced with efficiency or the reciprocal of peak width at half height. Taking the data for the 42% methanol sample first, then as the injection volume is increased the peak height decreases (as efficiency decreases and peak width increases). This is due to the fact that the sample is being injected in a solvent of high eluotropic strength in relation to the eluent (50/50, methanol/buffer) in which it elutes with a retention factor of 4.5. Consequently, the analyte is effectively smeared across the column as it is injected rather than being concentrated at the head of the column. If we set a limit to how much we can inject based on a loss of peak height of 15% then this would be around 20 txl with this particular sample. If we now consider the sample dissolved 4.2% methanol, then we can see that even with an injection volume of 500 txl we have maintained chromatographic performance and have lost no peak height at all. If the analyte concentration in the 42% methanol sample was 1 txg/ml, then by injecting 20 Ixl we could only inject 20 ng before performance was degraded by 15%. However, if we take that 42% methanol sample and dilute it 10-fold with water to 4.2% methanol (analyte concentration now 0.1 txg/ml) then we are now able to inject > 500 txl or at least 50 ng of analyte and still not lose any chromatographic performance. Hopefully this work indicates a clear strategy for maximising the amount of sample injected onto the column: dilute the sample with water and make a large volume injection. In this way the mass of the analyte injected can be much higher than if only a small volume of the undiluted sample is injected. One criticism of this approach is that in diluting the sample, there is the risk that the analyte will precipitate out of solution and be lost to the walls of the injection vial due to poor solubility in the low organic modifier concentration. One symptom of this phenomenon would be the production of concave calibration curves or calibration lines with negative intercepts. Although we have observed this phenomenon, it is quite rare. Furthermore, should such problems occur then they will be compensated for in the calibration, providing standards and samples are treated the same. The only real consequence will be a loss in assay sensitivity. We have successfully employed the above strategy, involving protein precipitation followed by dilution of the extracts 10 or 20-fold with water and then injection of 1 ml onto the column, for over 5 years.

9.6 THE USE OF GRADIENT ELUTION WITH HIGH FLOW RATES The short column approach described in Section 9.4, although probably originating in the drug development area soon spread to drug discovery. Whilst drug development work involves the analysis of thousands of samples containing the same compound, drug discovery work involves the analysis of thousands of different compounds, often with great diversity of physicochemical properties. So, whilst the principles were the same, isocratic elution was replaced with gradient elution to allow any compound to be chromatographed in a single analytical run. However before considering the

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

289

chromatography per se, it is worth considering the issues surrounding the instrumentation. In drug discovery, particularly in the area of bioanalysis, sensitivity can be a major problem and to this end analysts will often employ narrow-bore HPLC columns (1 to 2 mm ID) especially with ESI MS, which is a concentration sensitive detector, in an endeavour to achieve a greater signal for a given amount of compound injected. When used isocratically these columns presents no problem but, with gradient elution, the HPLC hardware limits the utility of these narrow bore columns. In gradient HPLC the re-establishment of the starting eluent conditions typically involves flushing the column with around 20 column volumes of the starting eluent. Because of the backpressure generated by columns packed with small particles and the limitations with standard HPLC equipment re-equilibration can only be carried out at relatively low flow rates, often < 0.5 ml/min with a 2 mm ID column. With some HPLC equipment, the internal volumes (pump head, mixers, thermal equilabrators, autoinjector, etc) represent a significant volume in relation to the column volume. Thus, on ending a gradient run and commencing the re-equilibration, it is necessary to flush the internal hardware before the starting eluent can reach the head of the column and the re-equilibration proper begin. If this dead volume is 1.5 ml, which is not untypical for some of the older equipment, then it will take 3 min for the starting eluent to reach the head of the column when a 2 mm ID column is being flushed at 0.5 ml/min If a 1 mm ID column is being employed then this could be up to 15 min. Thus, even with very short gradient times of 3 min for example, then the total cycle time could be as high as 19 min, being made up of 3 min actually chromatography time + 1 min re-equilibration time + 15 min flushing of internal volumes. Obviously this situation does not lead to high throughput! Although manufacturers are making efforts to develop instrumentation more suited to high speed gradient applications, the relatively large internal volumes of HPLC equipment still represents a limitation on the use of the narrow bore columns. Returning to the process of separation, then the most interesting aspect of this particular theme lies in the use of very high flow rates (higher than one would intuitively use), which not only give rapid separations but also maintain good resolution and chromatographic performance. This approach, which is based on the theory of gradient elution described by Snyder and co-workers in the late seventies and early eighties [39] was first applied by Mutton [40]. Taking Mutton's approach, then in simple terms, resolution (Rs) in gradient chromatography takes the same form as in isocratic chromatography except that the retention factor k is replaced by k~.,e (Eq. 1) Rs = 0.25 N~

- 1)[kave/(1 + kave) ]

( 1)

When chromatographing mixtures, it is important to maximise and maintain resolution and hence k~veshould be kept high and in the range of approximately 5 to 10. When analysing small molecules, 100 to 500 Da, then k,,ve is simply defined by the following parameters as shown in Eq. 2 k,,ve ~ tg F/L

(2)

where F is the volumetric flow rate, tg is the gradient time and L is the column length. References pp. 291-292

Chapter 9

290

With the objective of reducing analysis time whilst maintaining resolution, then k...... (and hence Rs) can be kept constant by reducing tg, the gradient time, whilst simultaneously increasing the term F/L by the same amount. For example the resolution on a 100 mm column eluted at 1 ml/min with a 50 min gradient will be approximately maintained (see below) through the use of a 33 mm column eluted at 2.2 ml/min with a gradient time of only 7.5 min As indicated above, resolution will only be partially maintained since reducing the column length will result in a small loss in resolution, although in this case this will only by around 40%. It is also assumed that the two columns are packed with small particles (< 5 Ixm) such that the loss in efficiency in running at high flow rates on the short column is relatively small as a consequence of the flat van Deemter curve. Mutton showed data where a 5-fold increase in throughput was achieved at the expense of a halving of resolution. A similar approach based around optimisation using column peak capacity rather than resolution has also been described recently [41 ], although this offers little over Mutton's work. In order to get the best out of such methods, the use of elevated column temperatures is often necessary to minimise backpressure, especially where very high flow rates are employed. This can be particularly important in reducing re-equilibration times, which as discussed above, can constitute a significant proportion of the total cycle time. We have employed this approach in our laboratory and also carried out further investigations. Figure 9.8 shows how mean resolution for a group of ten test markers varies as a function of flow rate on a 30 mm column. To obtain the shortest analysis time and maximal resolution it is necessary to run at a flow rate of 1.2 ml/min. The use of short columns packed with small particles and eluted at high flow rates with steep gradients is now in widespread use in drug discovery laboratories and has been successfully applied by a number of workers to applications such as metabolic inhibition assays [42], biofluid analysis [43,44], combinatorial library analysis [45], metabolic stability [44] and drug transport [44].

6.0 5.0--

Mean R

1.2

4.0--

0.6 Q

3.0 - -

0.3

2.01.0-0.0 0.0

I

!

5.0

10.0

15.0

Analysis cycle time (min) Fig. 9.8. The variation in mean resolution and analysis cycle time as a function of the eluent flow rate (0.3, 0.6 and 1.2 ml/min). The results were obtained for a set of 10 compounds chromatographedon a 30 • 2 mm, 3 txm column eluted with a linear gradient.

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

291

9.7 C O N C L U S I O N S T h e introduction of reliable ESI M S instruments during the 1990s has t r a n s f o r m e d the face of b i o a n a l y s i s in the p h a r m a c e u t i c a l industry. It has given the analyst u n p a r a l l e l e d access to s u b - n g / m l sensitivity with a high degree of selectivity and the possibility of high t h r o u g h p u t t h r o u g h rapid analysis. However, this has necessitated a m a r k e d rethink in how m e t h o d s are d e v e l o p e d and applied. M a n y of the concepts e x p l o i t e d in H P L C with U V detection are no l o n g e r applicable. M S is u n d o u b t e d l y the detector of choice for bioanalytical work. T h e next decade should see a c o n s o l i d a t i o n of this position as the size and cost of instruments c o m e d o w n and their flexibility and versatility increases. At this time, it seems unlikely that H P L C will be totally r e p l a c e d as the m e a n s of separation and introduction into the mass spectrometer.

9.8 ACKNOWLEDGMENTS I wish to thank K a r o l i n e Pitts and Dave Temesi w h o were involved in m u c h of the d e v e l o p m e n t w o r k in our laboratory. I w o u l d also like to thank H e l e n a Toreson of A s t r a Z e n e c a M o l n d a l w h o p r o v i d e d the data on ion suppression.

9.9 REFERENCES 1 2 3 4 5 6 7 8 9 l0 ll 12 13 14 15 16 17 18 19 20 21

B. Law, S.J. Houghton and E Ballard, J. Pharm. Biomed. Anal., 17 (1998) 443. H.H. Lauer and G.P. Rozing, Chromatographia, 14 (1981) 641. K.W.Smalldon and A.C. Moffat, J. Forensic Sci. Soc., 13 (1973) 291. H.A. Claessens, M.A. van Straten, C.A. Cramers, M. Jezierska and B. Buszewski., J. Chromatogr. A., 826 (1998) 135. R.J.M.Vervoort, A.J.J. Debets, H.A. Claessens, C.A. Cramers and G.J. de Jong, J. Chromatogr. A, 897 (2000) 1. S.R. Needham, ER. Brown, K. Duff and D. Bell, J. Chromatogr. A., 869 (2000) 159. M.G. Ikonomou, A.T. Blades and E Kebale, Anal. Chem., 63 (1991) 1989. R.E Straub and R.D.Voyksner, J. Chromatogr. A., 647 (1993) 167. S.R. Needham, ER. Brown and K. Duff., Rapid Commun. Mass Spectrom., 13 (1999) 2231. D. Temesi and B. Law, LC-GC Int., 12 (1999) 175. M. Jemal, Z. Ouyang and D.S. Teitz, Rapid Commun. Mass Spectrom., 12 (1998) 429. M. Jemal and D.J. Hawthorne, Rapid Commun. Mass Spectrom., 13 (1999) 61. D.V.McCalley, J. Chromatogr. A, 738 (1996) 169. B. Law and EE Chan, J. Pharm. Biomed. Anal., 9 (1991) 271. A.M. Kamel, ER. Brown and B. Munson, Anal. Chem., 71 (1999) 968. M.G. Ikonomou; A.T. Blades, E Kerbale, Anal. Chem., 62 (1990) 957. H-Z. Bu, M. Poglod, R.G. Micetich and J.K. Khan, Rapid Commun. Mass Spectrom., 14 (2000) 523. W.A. Korfmacher, K.A. Cox, K.J. Ng, J. Veals, Y. Hsieh, S. Wainhaus, L. Broske, D. Prelusky, A. Nomeir and R.E. White, Rapid Commun. Mass Spectrom., 15 (2001) 335. J.E Atherton, T.J. Van Noord and B-S. Kuo, J. Pharm. Biomed. Anal., 20 (1999) 39. B-S. Kuo, T. Van Noord, M.R. Feng and D.S. Wright, J. Pharm. Biomed. Anal., 16 (1998) 837. D.A. McLoughlin, T.V. Olah and J.D. Gilbert, J. Pharm. Biomed. Anal., 15 (1997) 1893.

292 22 23 24 25 26 27 28 29 30 31 32 32 34 35 36 37 38 39 40 41 42 43 44 45

Chapter 9 T.V.Olah, D.A. McLoughlin and J.D. Gilbert, Rapid Commun. Mass Spectrom., 11 (1997) 17. J. Berman, K. Halm, K. Adkison and J. Shaffer, J. Med. Chem., 40 (1997) 827. C. Tannergren, E Langguth, K.-J. Hoffmann, Pharmazie, 56 (2001) 337. R.A. Biddlecombe and S. Pleasance, J. Chromatogr. B., 734 (1999) 257. D.L. Burhman, EI. Price and EJ. Rudewicz, J. Am. Soc. Mass Spectrom., 7 (1996) 1099-1105. B.K. Matuszewski, M.L. Constanzer and C.M. Chavez-Eng, Anal. Chem., 70 (1998) 882. R. Bonfiglio, R.C. King, T.V. Olah and K. Merkle, Rapid Commun. Mass Spectrom., 13 (1999) 1175. H. Toreson, 16th Montreux LC-MS Symposium, Hilton Head, SC, 1999. B. Law and D. Temesi, J. Chromatogr. B, 748 (2000) 21. L. Romanyshyn, ER. Tiller, R.Alvaro, A. Pereira and C.E.C.A. Hop, Rapid Commun. Mass Spectrom., 15 (2001) 313. A. Xu, J. Havel, K. Linderholm and J. Hulse, J. Pharm. Biomed. Anal., 14 (1995) 33. A. Xu, K. Linderholm, L. Peng and J. Hulse, J. Pharm. Biomed. Anal., 14 (1996) 1675 Y-X. Li, K. Neufeld, J. Chastain, A. Curtis and E Velagaleti, J. Pharm. Biomed. Anal., 16 (1998) 961. Y.N. Li, B.N. Tattam, K.E Brown and J.E Seale, J. Pharm. Biomed. Anal., 16 (1997) 447. M.D.Moyer, T. Johannsen and R.J. Stubbs, J. Pharm. Biomed. Anal., 17 (1998) 45. M.J. Mills, J. Maltas and W.J. Lough, J. Chromatogr. A, 759 (1997) 1. M. Mills, PhD Thesis 1997 in the University of Sunderland. L.R. Snyder, J.W. Dolan and J.R. Gant, J. Chromatogr., 165 (1979) 3. I.M. Mutton, Chromatographia, 47 (1998) 291. Y-E Cheng, Z. Lu and U. Neue, Rapid Commun. Mass Spectrom., 15 (2001) 141. J. Ayrton, R. Plumb, W.J. Leavens, D. Mallett, M. Dickins and G.J. Dear, Rapid Commun. Mass Spectrom., 12 (1998) 217. R. Plumb, G.J. Dear, D. Mallett, I.J. Fraser, J. Ayrton and C. Ioannou,, Rapid Commun. Mass Spectrom., 13 (1999) 865. L. Romanyshyn, ER. Tiller and C.E.C.A. Hop, Rapid Commun. Mass Spectrom., 14 (2000) 1662. E Leroy, B. Presle, E Verillon and E. Verette, J. Chromatogr., Sci. 39 (2001) 487.

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 Elsevier Science B.V.All rights reserved

293

CHAPTER 10

Biomedical applications of directlycoupled chromatography-nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS) John C. Lindon ~, Nigel J.C. Bailey ~, Jeremy K. Nicholson ~ and Ian D. Wilson 2'* Biological Chemistry, Biomedical Sciences, Faculty of Medicine, Imperial College London, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ U.K. 2 Department of Drug Metabolism and Pharmacokinetics, AstraZeneca Pharmaceuticals, Mereside, Alderley Park, Macclesfield, Cheshire SKIO 4TG, U.K.

10.1 I N T R O D U C T I O N High performance liquid chromatography (HPLC) remains one of the principal methods for the separation of chemical mixtures and quantitation of components. However, the conventional detectors used to monitor the separation, usually based on refractive index, UV, fluorescence or electrochemical properties, provide only a limited amount of information on molecular structure. In addition, in studies of drug metabolism, it has generally been necessary to incorporate radioisotopes (such as 3H and ~4C) into the xenobiotic substance under investigation to ensure subsequent detection of compoundrelated material. However, none of these detectors provides sufficient information to allow molecular structural determination. Real advances in on-line minor component structure determination have only resulted from the relatively recent advent of the reliable hyphenation of HPLC and mass spectrometry (MS). This new technology is now widely exploited and there has been a huge growth of applications of HPLC-MS in the pharmaceutical industry, especially in the identification and quantification of drugs and metabolites in biofluids and extracts of tissue and excreta. These advances notwithstanding, MS by itself does not always provide unambiguous structural identification, and NMR spectroscopic data is often needed. * Current address: Scynexis Europe Ltd., Fyfield Business and Research Park, Fyfield Road, Ongar, Essex, UK. References pp. 325-329

294

Chapter 10

However, NMR spectroscopic analysis has, until recently, generally required timeconsuming isolation and purification steps, and sometimes this can conflict with the efficient HPLC-MS approach. The coupling of HPLC with on-line NMR has gone a long way to reducing the time needed to obtain good quality spectra, and it is these advances that are the subject of this article. However, even with such advances it should be noted that, because of sensitivity issues, HPLC-NMR can still be somewhat more time-consuming than HPLC-MS. Despite this caveat there is no doubt that the direct linking of HPLC with NMR spectroscopy has been a remarkable success story [ 1,2] and this has transformed the technique from a research tool to the stage where routine analytical applications are possible. As the number of HPLC-NMR systems have increased in user laboratories, there has been a logical extension of the hyphenation of HPLC-NMR and HPLC-MS into a single combined system for structure elucidation. Here, we describe the operation of HPLC-NMR and describe advances and applications in HPLC-NMR-MS. The operational difficulties of this double hyphenation in practice are discussed together with practical solutions.

10.2 TECHNICAL DEVELOPMENTS IN HPLC-NMR AND H P L C - N M R - M S 10.2.1 Introduction Direct on-line coupling of an NMR spectrometer as a detector for chromatographic separation, analogous to the use of MS for such applications, has required the development of technical features such as flow-probe hardware, efficient NMR solvent suppression pulse sequences and new software. The technical problems that needed resolution for successful practical commercial exploitation are summarised below.

10.2.2 The requirement for high dynamic range in NMR spectroscopy In HPLC-NMR it is necessary to detect signals from low concentrations of analytes in the presence of large ~H NMR signals from the HPLC solvents and this can cause difficulties because both large solvent and small analyte signals have to be detected simultaneously in pulse Fourier transform NMR spectroscopy and the analogue-todigital signal converter has a finite dynamic range. The solution to this problem has been the development of new techniques for either suppressing or not detecting the solvent NMR resonances. These are able to cope with mixed solvents, such as methanol/water, acetonitrile/water and even more complex solvent combinations, including the problems which are associated with eluent proportions changing during a gradient run [3,4]. A related factor which delayed the implementation of practical HPLC-NMR spectroscopy was the earlier need to use relatively expensive deuterated solvents for chromatography to overcome the dynamic range problem. This precluded their widespread use except for studies using microbore separations. Whereas solvents such as CC14 could be used for ~H NMR spectroscopy in normal-phase applications, they are hardly ideal and are now also environmentally unacceptable. Furthermore, this approach would necessitate the use of a probe with a separate external sample compartment containing a deuterated liquid to provide a signal for stabilising the magnetic field. This type of probe has been developed specifically for use with supercritical fluid

Biomedical applications of directly-coupled chromatography

295

chromatography linked to NMR spectroscopy (SFC-NMR) where CO2 is used as the eluent [5]. The solution to this difficulty for reversed-phase HPLC-NMR has come about because modern NMR spectrometers can perform solvent NMR resonance suppression very efficiently, thus reducing or even eliminating the need for deuterated solvents. For practical reasons D20 is still generally used to prepare eluents rather than H20 simply because this makes multiple solvent suppression easier and D20 is inexpensive. However, acetonitrile-d3 is being used increasingly in pharmaceutical laboratories because the cost of this solvent is negligible in relation to the other operating costs of a laboratory and it allows a substantial gain in quality of the results. The most commonly used organic modifiers for reversed-phase HPLC separations are methanol and acetonitrile. Acetonitrile and methanol both give rise to a singlet resonance in the 1H NMR spectrum which can be suppressed easily. However, the 13C satellite peaks also observed with these solvents, caused by the one-bond 1H-~3C spin couplings from the 1.1% of molecules with the naturally abundant ~3C isotope at the methyl carbon remain following suppression of the main peak, and can still cause problems of peak overlap. This is because these satellite peaks are often much larger than the signals for the analytes, and thus must also be suppressed. One approach is to set the suppression irradiation frequency over the central peak and the two satellite peaks in a cyclical fashion. Alternatively, if an inverse geometry probe is used which includes a ~3C coil, then broadband ~3C decoupling is possible, collapsing the satellite peaks under the central peak, enabling conventional single frequency suppression. Commercially available software from NMR instrument manufacturers also provides an automatic method for carrying out solvent suppression. This is particularly important in the case of gradient elution using acetonitrile as the organic modifier where the NMR resonance frequency of the acetonitrile will change during the run as the solvent composition changes. A software routine searches for the 13C satellite peaks in the IH NMR spectrum of the solvent mixture, interpolating to find the main signal and then setting the suppression frequency accordingly. This has the advantages of being nonperturbing to the NMR observation of the analyte signals, being implemented in real-time during the separation and being independent of NMR solvent shifts induced by chromatographic solvent gradients.

10.2.3 Avoidance of compromised chromatographic resolution One of the current problems associated with HPLC-NMR results from the presence of the stray magnetic field from the spectrometer which limits how close the HPLC instrumentation can be placed without compromising the performance of the system. As a result the necessary layout of an HPLC-NMR system requires the use of long columnto-NMR detector transfer times. In addition conventional flow NMR probes use relatively high volume flow cells (by chromatographic standards). These factors were originally considered to be a potential cause of chromatographic peak broadening. For HPLC-NMR probes, a compromise has to be made between the needs of chromatography and NMR for the detection volume of the flow cell. Ideally this should be as low as possible for good chromatography and as high as possible for NMR detection. The

References pp. 325-329

296

Chapter 10

low volume for chromatography can only be compensated for by increasing the filling factor which can be achieved by fixing the RF coil directly to the NMR flow cell wall. This of course makes it impossible to spin the sample, as is performed in conventional probes, to improve magnetic field inhomogeneities. Practically, however, this turns out not to be a problem as field homogeneity with this small volume is good. Further, modern computer-controlled methods for optimising field homogeneity reduce the requirement for spinning. A major factor in determining the sensitivity or peak heights is the observed lineshape as, if the peaks have wide bases, poor signal-noise results because a significant part of the signal intensity is found in this part of the peak. Thus, good magnetic field homogeneity, giving narrow NMR lines, is also a prerequisite for a good signal-noise ratio. A detailed analysis of the flow and NMR requirements for optimum operation of HPLC-NMR has been given recently [6,7].

10.2.4 The need for high NMR sensitivity The relatively low sensitivity of NMR spectroscopy, in relation to the quantities of analytes such as drug metabolites often encountered in the samples to be separated, must be recognised. However, as described later, impressive sensitivities can be achieved using HPLC-NMR in the stop-flow mode and the detection limits of HPLCNMR are continually being revised downwards [8] as new technical advances are made. Recently, these have included the use of high magnetic field strengths (operating at 750 MHz and 800 MHz for JH NMR spectroscopy), the incorporation of digital filtering and oversampling into NMR data acquisition and the introduction of microbore HPLC methods. By combining the use of digital electronics with microbore HPLC [9,10], it appears that detection limits for structural characterisation are in the low ng region for a small molecule analyte even at lower JH NMR observation frequencies of 500 MHz and 600 MHz. In principle, it is possible to effect NMR detection for any of the magnetically-active nuclei, but those of most importance in pharmaceutical studies are ~H, 2H, ~9F, ~3C, ~SN and 3~E In practice, because of the generally small quantities of material in the samples of interest the most sensitive nuclei, ~H and 19F have been used most extensively. The use of J3C NMR in HPLC-NMR can be facilitated through indirect detection of J3C resonances via the much more sensitive ~H NMR signals of attached protons using twodimensional methods such as 1H-~3C heteronuclear single quantum coherence (HSQC) [11]. A major benefit of using 19F NMR spectroscopy for detection of fluorinecontaining molecules is that the background, unlike that for ~H NMR spectroscopy, is usually negligible.

10.2.5 Additional considerations for double coupling of NMR and MS to HPLC HPLC-MS has been employed for many years but only since the advent of modern interfaces such as atmospheric pressure chemical ionisation (APCI) and electrospray ionisation (ESI) has it become a truly robust and routine method for the analysis of mixtures. Now that NMR spectroscopy has been coupled to HPLC, it has become

Biomedical applications of directly-coupled chromatography

297

possible to acquire both NMR and MS data simultaneously from a single chromatographic analysis, as first demonstrated by Pullen et al. [12]. Solvent selection for H P L C - N M R - M S however, can cause complications and has to be a compromise between the ideal requirements of each instrument. Thus, for HPLCNMR the use of inorganic buffers, e.g. sodium phosphate, for pH modification is the best option because no additional signals are introduced to complicate the resulting NMR spectrum. However, inorganic buffers are quite unsuitable additives for HPLCMS solvent systems. An alternative acidic modifier suitable for NMR spectroscopy is trifluoroacetic acid (TFA), which has no protons to cause interferences in the NMR spectrum. Initial experiments, using paracetamol metabolites or propranolol as model analytes [13], showed that 0.1% TFA could be used with HPLC-MS for a limited range of analytes present at high concentration (> 1 p~g on column) in positive ion mode. However, with acidic analytes such as ibuprofen and its metabolites, ion suppression was complete and, even at high sample concentrations, MS data could not be obtained. Formic acid was found to provide a suitable compromise between the needs of MS on the one hand and NMR spectroscopy on the other. The single proton of formic acid, which has a sharp, readily-suppressable NMR singlet near ~8.5, gives minimal interference in the resulting NMR spectra and enables MS data to be acquired for acidic analytes. There are two ways to configure the NMR and mass spectrometers, either in parallel or in series. As NMR spectroscopy is a relatively insensitive technique, large volumes and high concentrations of analytes are used wherever possible to compensate and to reduce analysis time. As a result, generally 4.6 mm HPLC columns are used to avoid problems of overloading. This means flow rates of the order of 0.5-1.0 ml/min are used which meets the requirements of the NMR spectrometer without compromising the chromatography. Such high flow rates can easily be accommodated by modern mass spectrometers. However, MS is concentration dependent, as opposed to mass sensitive, and also destructive. Operating the NMR spectrometer and the MS in parallel, and thus splitting the flow such that a minor fraction goes to the MS, has little effect on sensitivity but enables the bulk of the peak of interest to be collected for further testing if required. It also has benefits such as enhancing source lifetime allowing the mass spectrometer to be operated at optimum sensitivity for longer. If the flow is split prior to the NMR spectrometer, with the length of the capillary to the MS adjusted such that the analyte peak is detected by the MS as it fills the NMR flow cell, the MS can be used to supplement the UV data to direct NMR experiments. Further, splitting in this manner enables the use of stop-flow NMR with minimum degradation of the integrity of the chromatography [ 13]. Running in series, i.e. with the sample destructive MS after the non-destructive NMR spectrometer, allows for the completion of all NMR experiments whether on-flow or stop flow before MS analysis begins, but introduces the possibility of peak dispersion before MS analysis for any peaks trapped between the NMR spectrometer and MS when the flow is stopped. Series operation also causes the NMR flow cell and its connections to be operated at higher pressures than they were designed for, with the consequent possibility that leaks are more likely. Series operation also fails to take advantage of the mass spectrometry ability to flag up peaks of interest quickly.

References pp. 325-329

298

Chapter

10

NMR spectrometer ~:~4 l~ . . . . . ' :: ~=:~ ~'t ~!~. . . . .

~. . . . . . . . . . .

~ ~i

i

...... HPLC, oven, DAD detector and loop collector

:

Fraction collector

~-1 Splitter ! 95%

/

5%1 .i~i

'~"~- "

Mass spectrometer, usually ion trap

Fig. 10.1. A schematic of a typical commercial HPLC-NMR-MS system. Reproduced with permission of Bruker Biospin, Germany.

Given the strength of the magnetic field surrounding an NMR magnet, there is obviously the potential for interference with the operation of the mass spectrometer. There is also the potential for the presence of the MS to interfere with the operation of the NMR spectrometer. Some experiments have been carried out to investigate this in the situation where the mass spectrometer was sited at the ten gauss line with its axis radial to the NMR. Varying the angle of the mass spectrometer to the NMR had no effect on either the data from the mass spectrometer or the NMR. The HPLC, NMR spectrometer and MS were situated approximately at the comers of an equilateral triangle but this was largely dictated by the size of the laboratory and the position of the supplies to the mass spectrometer. Nevertheless, the mass scale needed to be recalibrated and there may also have been a drop in low mass sensitivity (below m/z= 150) but this was not quantified. No effect on the operation of the NMR spectrometer with the MS located in this position was observed [13]. Since these studies were reported, the advent of shielded superconducting NMR magnets has alleviated the potential problems considerably. A schematic of a typical arrangement for combining HPLC with both NMR spectroscopy and mass spectrometry is shown in Fig. 10.1.

10.3 OPERATIONAL METHODS IN H P L C - N M R AND H P L C - N M R - M S The five main options which can be employed for HPLC-NMR using either isocratic or gradient elution are continuous-flow, stop-flow, "time-sliced" stop-flow, peak collection

Biomedical applications of directly-coupled chromatography

299

into capillary loops for post-chromatographic analysis and automatic peak detection with UV-detected triggered NMR acquisition. The simplest of these is continuous-flow detection, but this is usually only practical when using ~H or 19F NMR spectroscopy for detection unless isotopically enriched compounds are available. However, there are a few examples of HPLC-NMR studies using 2H and 3~p NMR detection in the drug metabolism field [14,15]. Where continuous-flow NMR detection is used for gradient elution, the NMR resonance positions of the solvent peaks shift with the changing solvent composition. For effective solvent suppression, these solvent resonance frequencies must be determined as the chromatographic run proceeds. If the retention times of the analytes are known, or there is an efficient method for their detection on-line, such as UV, MS or radioactivity, stop-flow HPLC-NMR spectroscopy becomes a viable option. In the stop-flow technique, all the usual techniques available for high-resolution NMR can be used. In particular, these include valuable techniques for structure determination such as 2-dimensional NMR experiments, which provide correlation between NMR resonances based on mutual spin-spin coupling such as correlation spectroscopy (COSY) or total correlation spectroscopy (TOCSY) [ 11 ]. In practice, it is possible to acquire NMR data on a number of peaks in a chromatogram using a series of stops during elution without on-column diffusion causing an unacceptable loss of chromatographic resolution. There are two further special categories of stop-flow experiment. Firstly fractions eluting from the column can be stored in capillary loops for later off-line NMR study ("peak picking"). Secondly, the flow can be halted at short intervals during the passage of the eluting peak through the NMR flow cell ("time-slicing") in a manner analogous to the use of a diode-array UV detector to obtain spectra from various portions of the peak. This allows chromatographic peak purity to be estimated. Time-slicing is most useful where the separation is poor, or where the compounds under study have weak/no UV chromophores making it difficult to determine the retention times. Stop-flow acquisition may also be performed using pre-concentration of analytes, effected by column trapping. Griffiths and Horton [16] constructed a system whereby analytes were concentrated on a small chromatographic column. The analytes were then re-eluted using a stronger solvent which resulted in a more concentrated peak for subsequent analysis by NMR. Fully automated analysis is also an option wherein the samples are placed in an autosampler and predefined HPLC-NMR experiments are performed. The software allows automatic detection of UV peaks in the chromatogram based on predetermined time-windows or peak intensities. The successful detection of each UV peak triggers the system to stop the flow at an appropriate time to isolate the peak in the NMR flow probe. Then data relating to the peak (intensity, retention time) are transferred to the NMR host computer and used to define the parameters for the automatically-acquired NMR spectrum. This automatic NMR operation includes field homogeneity optimisation, setting and optimisation of all NMR acquisition parameters and the predefinition of the resultant signal-to-noise ratio required in the spectrum. The measurement of 2-dimensional NMR spectra can also be performed. With currently available

References pp. 325-329

300

Chapter 10

commercial software, the automated run can be halted at any time with reversion to manual control if desired. Despite the apparent difficulties of performing multiple stopped flow analysis, a process which seems to defy much chromatographic "common sense" it is not usually necessary to make any compromises in a desired chromatographic procedure to accommodate the various types of HPLC-NMR experiment. It is also noteworthy that the powerful structural elucidation capabilities of NMR spectroscopy often ensure that complete chromatographic separation is not necessary for full characterisation of the peak. In the case of HPLC-NMR-MS experiments there are some additional considerations. So far, the principal MS ionisation method used has been electrospray in either positive or negative ion mode (using either single quadrupole or ion-trap mass spectrometers) and this puts further constraints on the chromatographic solvent systems as outlined earlier. Often in HPLC-NMR studies the initial chromatographic separation has been developed off-line from the NMR using non-deuterated solvents. It is not always simply a matter of replacing non-deuterated solvents with deuterated solvents to reproduce this chromatography for HPLC-NMR or HPLC-NMR-MS as this can give rise to changes in retention times. For this reason it is standard practice to run an initial chromatographic run with a small amount of the sample to check retention time data etc., and then scale up the amount injected for stop-flow NMR, when optimum conditions have been established. We have found that during the initial run it is often possible to acquire a great deal of valuable MS data which can then be used to guide the selection of peaks for study by NMR spectroscopy. Ideally this also allows the second run to be acquired while mixing the eluent just prior to the mass spectrometer with a non-deuterated solvent to back-exchange the deuterium atoms in exchangeable situations (e.g. NH and OH groups) for hydrogens. In this way, if these initial data cannot be readily understood, the number of exchangeable hydrogens in any compound can be counted as it elutes [12]. If this is to be performed successfully however, sufficient time must be allowed for back exchange to occur. As alluded to above, the double hyphenation of NMR and MS to HPLC [12,17-19] brings some additional benefits, particularly, when the analyte reaches the MS before the NMR spectrometer. Thus, MS can be used to search for particular diagnostic groups or fragments (e.g. an increase in m/z of 16 for hydroxylated metabolites, or 196 for a glucuronide) and such data can then be used to direct NMR spectroscopy to particular peaks in the chromatogram [20].

10.4 APPLICATIONS IN COMBINATORIAL C H E M I S T R Y Characterisation of the structure and conformation of small biologically active molecules is part of the standard approach to lead generation in drug design studies. In particular, it is now possible to automatically synthesise many thousands of small molecules and then rapidly measure their effects in a given pharmacological test system. The potential power of such techniques comes from the immense number of compounds which can be generated and screened for activity. Two studies have evaluated HPLC-

Biomedical applications of directly-coupled chromatography

301

NMR in the field, one based on a mixture of 27 closely related tripeptides [21] and the other on two separate mixtures of four aromatic compounds and three pentapeptides [22]. In the first case, the tripeptide application comprised a synthetic mixture of the 27 combinations of the tripeptide formed from alanine (A), methionine (M) and tyrosine (Y), as the C-terminal amide, H2N. CH(R~). CONH. CH(R2). CONH. CH(R3). CONH 2 where R1, R2 and R3 take all combinations of A, M and Y [21]. Based on chemical shifts and peak multiplicities, the on-flow HPLC-NMR characterisation of the majority of the components in the mixture of 27 tripeptides was achieved and it demonstrated that this approach is likely to be an effective method for compound mixtures. The elution positions of all of the alanyl-containing peptides were determined with the exception of A - M - M - N H 2 which may have co-eluted with another peptide or may have been synthesised in a much smaller quantity. The only other tripeptides for which assignments were not obtained were the MYz-NH 2 isomers and two of the three M2Y-NH2 isomers. These elute towards the end of the gradient run and may not be as well resolved under these HPLC conditions. Additionally with changes in the relative chemical shifts of the solvent signals, the intensities of the non-Nterminal oL-CH protons and the methionyl [3-methylene signals from these peptides may have been reduced by the effects of the solvent suppression irradiation of the water and acetonitrile resonances respectively. With further optimisation of the elution conditions, it is possible that all 27 analytes could have been resolved and characterised.

10.5 APPLICATION TO DRUG IMPURITIES

The manufacture of a drug, and its quality control, is regulated by a variety of national authorities. As well as the need to demonstrate drug efficacy, there is also a strong emphasis on the characterisation of the purity of final drug substances and it is necessary to obtain full characterisation and identification of any impurities at the level of > 0.1% of the UV peak area using HPLC analysis [23]. In order to characterise such impurities, it is usually necessary to isolate individual components by preparative HPLC. This work is often time consuming and expensive and yet may not give conclusive identification. Furthermore it is possible for the impurities to be degraded during sample extraction and purification. There is, therefore, a considerable need to develop and validate new methods for determining product purity. One of the first published real applications of HPLC-NMR was concerned with the identification of an impurity in a synthetic drug precursor [24] and a number of examples are now in the literature including characterisation of impurities in a bulk batch of fluticasone propionate [25] an anti-inflammatory drug used for the treatment of the underlying inflammatory component of asthma, a GART inhibitor AG2034 [26] and the degradation products of a HIV protease inhibitor [27]. Directly-coupled HPLC-NMR and HPLC-NMR-MS can greatly enhance the ability to characterise impurities in a pharmaceutical product. As such, this approach is potentially significant as a general tool for purity analyses and would be expected to be important in speeding up production chemistry processes and for regulatory affairs. A

References pp. 325-329

302

Chapter 10

combination of HPLC-NMR and HPLC-MS has been used to study the degradation products of an antifungal drug candidate [28]. Both techniques contributed complementary structural information that aided both the structure elucidation and proposed pathway of formation for the degradation products. An impurity in the synthetic drug naftopidil has been identified using HPLC-NMR spectroscopy as 1-naphthylglycerol [29]. Also, the autoxidation and photodegradation of ethynylestradiol has been studied using the same approach [30]. In another study, Mistry et al. showed that HPLC-NMR spectroscopy could detect and characterise impurities below the 0.1% peak level [25], the relevant limit for submissions to regulatory authorities. However, NMR spectra on the peaks which were at a level of 0.2% or less of the parent drug, required considerable and time-consuming data acquisition to achieve acceptable signal-to-noise ratios. Although this appears to involve a high cost in NMR analysis time it would be justified in cases such as that found with fluticasone propionate where there are few alternative analytical approaches. It may, therefore, be beneficial to concentrate the impurities, if stable, before the HPLCNMR by the application of solid phase extraction chromatography, column switching or concentration/enrichment of the impurities by preparative HPLC. Such techniques can be feasible because of the availability of large amounts of sample from drug production batches. The technique could also be applicable to the investigation of formulated drug substances where impurities often appear as a result of the drug substance reacting with the formulation compound or on samples from degradation studies. These types of adduct can sometimes be unstable during sample extraction and purification. This area is of considerable practical importance in the pharmaceutical industry, and although applications have been reviewed recently [31 ], it is worthy of further study.

10.6 CHIRAL H P L C - N M R AND H P L C - C D FOR PHARMACEUTICAL MIXTURES

Many pharmaceutical products are chiral molecules either as single isomers or more commonly as racemic mixtures. In addition, many formulated products are mixtures of active compounds together with a number of additives such as excipients. For chiral molecules, the pressure to develop single isomer forms as therapeutics in preference to racemic mixtures arises from the fact that one enantiomer is usually more biologically active than the other and also that enantiomers can have very different toxicity profiles. Chiral HPLC on-line with NMR has been performed to demonstrate the application of chiral HPLC-NMR spectroscopy to the separation and characterisation of different isomers present in a drug substance employing, as an example, atracurium besylate, a neuromuscular blocking agent used widely in surgery [32]. Atracurium besylate, (2,2 '-(3,11 -dioxo-4,10-dioxatrideca-methylene)-bis- (2-methyl- 1,2,3,4-tetrahydropapaverinium benzenesulfonate), is prepared from racemic 1,2,3,4-tetrahydropapaverine and has four chiral centres. However, because of the symmetry of the molecule, atracurium has 10 distinct species with the structure given in Fig. 10.2 where the configuration at

%

~..,o

ATRACURIUM BESYLATE

r~

I t~

OMe

5

~..,. c~

4

Me

17

OMe

Me

21

:

8

OMe \

~

_--~

16

= 9

II

20

OMe

22

O

~..,.

O 2 x benzenesulphonate c~

OMe

OMe OMe

OMe

E"

T h e (R,cis; R,cis) i s o m e r is s h o w n Meso

Racemates R,cis; R,cis

S,cis; S,cis

R,cis; S,cis

R,trans; R,trans

S,trans; S,trans

R,trans; S,trans

R,cis; R,trans

S,cis; S,trans

R,cis; S,trans

R,trans; S,cis

Fig. 10.2. The molecular structure of atracurium besylate, showing the 10 isomeric forms. taO

304

Chapter 10

C1 can be R or S. Since the final stage of synthesis is quaternization at N2, the isomers have been distinguished, simply for convenience, by the configuration of the substituents at the C1-N2 bond, such that when a tetrahydroisoquinoline residue has the benzyl group at C1 and, arbitrarily, the 3,11-dioxo-4,10-dioxatridecamethylene chain is in a cis configuration, this is called a "cis" residue. In achiral media, different NMR spectra are expected for each of the four types of enantiomeric pairs and for the two meso compounds. In general because of the synthetic approach, the ratio of cis to trans residues is about three which, assuming that quaternization at one tetrahydroisoquinoline residue does not affect quaternization at the other, leads to proportions of cis-cis, cis-trans and trans-trans isomers in the ratio of 11:6: 1. The preference for cis residues has been proved using NOE NMR measurements and by x-ray crystallography on related substances. The ~H NMR chemical shifts, principally of the H8 proton, can be affected not only by whether the C1-N2 configuration in the residue is cis or trans but also by the configuration of the remote tetrahydroisoquinoline unit. In summary therefore, after on-line chiral HPLC separation, NMR spectroscopy has been used to characterise compounds in terms of the cis and trans isomers and to identify the racemic pairs on the basis of their identical NMR spectra. In addition, HPLC-CD was used to identify the absolute configuration of the enantiomers based on the known CD spectrum of R-laudanosine hydrochloride, a closely related molecule. The resulting chromatograms using both UV and CD detection are shown in Fig. 10.3. In this instance HPLC-NMR spectroscopy was useful for identifying the isomeric configuration at the C1-N2 bond, for identifying the enantiomeric pairs of compounds and for distinguishing them from the meso forms. The HPLC-CD experiments were complementary in that, whilst unable to distinguish the C1-N2 isomers (cis or trans), it was possible to determine the absolute stereochemistry at C1 at each tetrahydroisoquinoline residue as either R/R, S/S or R/S based on the sign of the CD response at a chosen wavelength. A consistent finding was that the S isomers eluted before the R isomers and the trans forms eluted before the cis forms. By these means a full characterization of all of the 10 isomers of atracurium was achieved.

10.7 A P P L I C A T I O N TO NATURAL PRODUCTS Natural products have been and remain a rich source of leads for the pharmaceutical industry and many marketed drugs are either natural products or are modifications of such substances. Hence considerable effort is spent in isolating and characterising chemicals from natural sources which can be tested in a variety of biological screens. Often it is necessary to carry out laborious extraction and purification steps and the advent of directly-coupled HPLC-NMR has been explored as an alternative technique for natural product identification. To evaluate the approach, a model mixture of nine aporphine alkaloids has been used to specifically test the loop storage mode of operation of HPLC-NMR [33]. The first application of HPLC-NMR to natural products was in 1994 on the photochemical reactivity of azadirachtin [34]. The technique has begun to assume

Biomedical applications of directly-coupled chromatography

305

HPLC OF ATRACURIUM (b) CD Detection

. . . .

'

,,,

;

_

,

i

,

,

,

E

;

j

~

~

i

;

9

i

.

.

.

.

~.:

D

(a) UV Detection

I

,,'1 . . . .

5O Time (rain)

!

F

....

I

~

.

.

.

.

.

.

.

J

i

....

t

I O0

Fig. 10.3. (a) The UV-detected chiral HPLC chromatogram of the separation of isomers of atracurium, (b) the corresponding CD-detected chromatogram. The HPLC-NMR spectroscopy identified the isomeric configuration at the C1-N2 bond, the enantiomeric pairs of compounds and distinguished them from the meso forms. The HPLC-CD experiments allowed the determination of the absolute stereochemistry at C 1 at each tetrahydroisoquinoline residue as either R/R, S/S or R/S. Thus, for the atracurium isomer peaks labelled on the figure, A is S-trans/S-trans, B is R-trans/S-trans, C is S-cis/S-trans, D is R-trans/R-trans, F is S-cis/S-cis, H is R-cis/R-trans, J is R-cis/S-cis, K is R-cis/R-cis and E and G are the pair R-cis/S-trans and R-trans/S-cis.

greater prominence in for the identification of natural products, particularly when used in combination with other hyphenated techniques such as H P L C - M S - M S . This area of research has been reviewed by Wolfender [35-37] who has reported numerous applications in this area including the characterisation of polyphenols and bitter components from Gentianaceae species [38], the assignment of stereochemistry at a double bond in a new secoiridoid glycoside, seemannoside [39], antifungal materials from the African plant Swertia calycina [38,39], compounds from the Leguminosae family [40], prenylated flavanones from dichloromethane extracts of Monotes engleri [41], naphthoquinones from Cordia linnaei [42], pyrrolizidine alkaloids from Senecio species [43] and antioxidant compounds from the leaves of Orophea enneandra [44]. Bringmann et al. have identified new naphthylisoquinoline alkaloids from a root extract of Ancistrocladus likoko using directly-coupled H P L C - N M R spectroscopy [45]. H P L C - N M R was also used to determine products from coupling reactions catalysed by enzymes isolated from Ancistrocladus as well as Triphyophyllum species [46] and to

References pp. 325-329

306

Chapter 10

characterise ecdysteroid components in Silene otites [47]. Other studies on natural products using, amongst other techniques, HPLC-NMR include the characterisation of vitamin derivatives [48], saponins from Bacopa monniera Wettst [49], antibacterial sesquiterpene lactones from an extract from Vernonia fastigiata [50]. HPLC-NMR-MS has been used to identify components from Hypericum perforatum L. [51], quercetin glycosides and phloretin glycosides from apple peel [52] and ecdysteroids from Silene otites [53]. Natural products research is also an area where the combined application of HPLCCD and HPLC-NMR has proved advantageous. Bringmann et al. [54] demonstrated the principle by obtaining the structures of chiral compounds in crude plant mixtures from Habropetalum dawei. A more recent paper from the same group demonstrates the application of the 'LC-MS/LC-NMR/LC-CD triad' to the screening of crude plant extracts [55]. In particular, novel alkaloids were identified from extracts of Ancistrocladus griffithii. In addition to applications in phytochemistry, other natural products have been studied. Aptamine was identified as the active component from a marine sponge, Aaptos sp. [56]. The compound was identified by HPLC-NMR spectroscopy from a crude organic extract. Sandvoss et al. used HPLC-NMR-MS to study compounds from the starfish Asterias rubens [57,58]. Previously unreported asterosaponins were identified from subfractionated extracts. The relative complexity of the structures clearly demonstrated the potential of having both MS and NMR available in one HPLC run. Microbial production of secondary metabolites is also an important source of novel therapeutic agents. However, the physiological and biochemical factors that determine the onset of production of a specific secondary metabolite in a particular species are incompletely understood. Generally, a range of analytical techniques, often elaborate, time-consuming and involving extensive sample pre-treatment, have to be developed in order to monitor the details of the metabolic changes and substrate consumption that accompany secondary metabolite production. To provide rapid multi-parametric information about the microbial fermentation process, JH HPLC-NMR has been applied to characterise microbial metabolites directly in the broth supernatants from a wild-type strain of S. citricolor [59]. This species produces aristeromycin, the carbocyclic analogue of adenosine, a secondary metabolite with antibiotic properties. The 600 MHz ~H NMR spectrum of the broth was particularly complex in the chemical shift region between g4.2 and g3.4, since it contains major sugar signals obscuring minor metabolites. Hence an HPLC method with on-line ~H NMR detection was employed to physically separate components which gave signals in this region. HPLC-NMR has also been used to identify biotransformation products of warfarin in cultures of Streptomyces rimosus [60]. 10.8 APPLICATION TO DRUG METABOLISM 10.8.1 Introduction By far, the largest body of work to date using HPLC-NMR and HPLC-NMR-MS is in the field of drug metabolism where the methodology has been used extensively for the

Biomedical applications of directly-coupled chromatography

307

identification of metabolites in studies from clinical trials involving human subjects, the investigation of model drugs in animals in vivo and also through the use of in vitro systems such as liver microsome incubations. The results in the literature are summarised briefly below and one example, the identification of the metabolites of 2-bromo-4-trifluoromethylaniline found in rat urine [61], is given in somewhat more detail since it encapsulates many of the different aspects which comprise HPLC-NMRMS and thus serves to illustrate the various modes of operation of the use of the technology.

10.8.2 A summary of human metabolism studies

Antipyrine has been extensively employed as a probe to investigate the induction of drug metabolism. In man, the main metabolites are conjugates of norantipyrine, 4-hydroxyantipyrine and 3-hydroxymethylantipyrine. Stop-flow HPLC-NMR, on a 500 MHz NMR spectrometer, was used to analyse urine samples obtained from a human volunteer following the oral administration of antipyrine [62]. This experiment enabled the unambiguous determination of the structures of the major antipyrine metabolites, rapidly and without any pre-treatment (other than preconcentration by freeze-drying) of the sample. By this means, it was possible to identify the ether glucuronides of 4-hydroxyantipyrine, norantipyrine-glucuronide and 4-hydroxyantipyrine itself, the latter either excreted as such or produced by degradation of the glucuronide. Norantipyrine tautomerises to give the 5-enol and it is the O-glucuronide which is formed in preference to the corresponding N-glucuronide In addition, a minor component showing signals for both olefinic hydrogen and glucuronide proton resonances was detected, and this probably corresponds to 3-hydroxymethylantipyrine glucuronide. Ibuprofen is a widely used non-steroidal anti-inflammatory drug (NSAID) which is subject to extensive metabolism, via both Phase I (hydroxylation and oxidation) and Phase II (glucuronidation) pathways. The principal Phase I metabolites are hydroxy and carboxy oxidation products. In man, the metabolites of ibuprofen are rapidly excreted in the urine following administration of normal therapeutic doses to a healthy volunteer. The use of HPLC-NMR on urine samples obtained following a normal therapeutic dose of the compound (400 mg) formed the first application of HPLC-NMR to the study of drug metabolism. This work employed gradient elution on freeze-dried urine [63]. From a continuous-flow HPLC-NMR run it could be seen that three ~H NMR signals around ~5.6 were assignable to anomeric protons from glucuronide conjugates in the samples. Confirmation of this identification was obtained via stop-flow measurements. This allowed identification of the glucuronide of the side chain hydroxylated metabolite of ibuprofen, the glucuronide of ibuprofen itself and the glucuronide of the diacid metabolite. Also identified were unconjugated hydroxy-ibuprofen, and the side chain oxidised diacid metabolite. The 2-D TOCSY spectrum (with double solvent suppression) in stop-flow mode was useful in confirming the proposed structures in that the cross peaks due to the glucuronide spin system, and those for the methyl-methine spin system were readily visible. More recently, a comprehensive study employing HPLCReferences pp. 325-329

308

Chapter 10

NMR-MS has been used to confirm the identities of these metabolites [64], and in addition identify a number of additional minor components. A similar application of HPLC-NMR in drug metabolism was the study of the metabolic fate of racemic flurbiprofen ([_+ ]-2-(2-fluoro-4-biphenylyl)propionic acid) [65] which, in the same way as seen for ibuprofen is converted to a number of Phase I and Phase II metabolites. The principal metabolite is 4'-hydroxy-flurbiprofen and 3',4'-dihydroxy flurbiprofen is a minor metabolite, but all metabolites are excreted mainly as the glucuronide conjugates. The presence of fluorine in a drug molecule enables selective detection of drug metabolites for compounds such as flurbiprofen. Apart from fluoride ion, usually present at low concentration in biofluids and giving a single broad resonance, there are virtually no endogenous fluorine-containing compounds and hence 19F NMR spectroscopy provides a rapid diagnostic method for assessing the number and level of metabolites of fluorine-containing drugs in body fluids. The 600 MHz ~H NMR spectrum for a urine sample from a human volunteer, following ingestion of 200 mg of flurbiprofen was complex, showing a multiplicity of signals from endogenous metabolites as well as those due to flurbiprofen-related compounds. This complexity precluded any detailed structural or quantitative analysis demonstrating the limitations of 1H NMR in this case. The corresponding resolution-enhanced 19F NMR spectrum (with ~H decoupling) indicated the presence of four major fluorine containing species. The chemical shifts of the major fluorinated metabolites were all suggestive of modifications to the drug distant to the fluorine-containing phenyl ring whilst a total of some 24 separate fluorinated components were detected in this spectrum altogether. The pseudo-2-D contour plot for the continuous-flow HPLC-NMR experiment, with 19F detection showed four J9F resonances, present as two pairs at retention times of 30.5 and 36.6 min. Of these, the peaks eluting at 30.5 min. corresponded in chemical shift to the largest resonances seen in the ~F NMR spectrum of the whole urine whilst those eluting at 36.6 min. corresponded to the remaining major components. The pairing of the chromatographic peaks was due to the glucuronides being present as diastereoisomers as a result of the conjugation of [3-D-glucuronic acid with the R and S isomers of flurbiprofen or its metabolites. Interestingly the intensities of the signals for the two diastereoisomers in each pair of resonances were not equal indicating either differential excretion or metabolism of the two isomers of the racemic drug or inversion of the chiral centre in vivo. The experiment was repeated using stop-flow JH NMR detection at 600 MHz at the appropriate retention times for the 19F NMR-detected metabolites. The resulting 1H NMR spectra were consistent with the [3-D-glucuronic acid conjugate of the hydroxylated metabolite eluting at 30.5 min. and flurbiprofen glucuronide at 36.6 min. Using the "time slicing" technique, the IH NMR spectrum was obtained for this peak after elution for a further 30 s. This latter spectrum is of only one of the diastereoisomers of flurbiprofen-[3-D-glucuronide, revealing the inhomogeneity of the chromatographic peak. Further studies with ~H NMR in stop-flow mode enabled a further, minor, flurbiprofen metabolite to be identified as the free 4'-hydroxyflurbiprofen. Another fluorinated drug candidate, the HIV-1 reverse transcriptase inhibitor BW935U83 has also recently been studied using hyphenated HPLC-NMR methods

Biomedical applications of directly-coupled chromatography

309

[66]. In this study, the presence of both fluorine and chlorine atoms provided complementary markers beneficial for NMR and MS detection respectively. By obtaining 19F NMR spectra and on-flow HPLC-~9F NMR pseudo-2D spectra, it was possible to determine the number and location of metabolites. This was particularly useful in this case as not all compounds of interest had UV chromophores and would thus have been missed using conventional detection methods. A combination of NMR and MS data acquired simultaneously using HPLC-1H NMR-MS then allowed structural elucidation. Paracetamol, or acetaminophen, is one of the most widely studied of all xenobiotics and its metabolic fate is well documented. A number of ~H NMR studies of paracetamol metabolism in man have been conducted, and the major metabolites, namely, the phenolic ether glucuronide, the sulfate and the product with N-acetylcysteinyl substituted at position C3, together with paracetamol itself were all detected and quantified in urine [67,68]. More recently HPLC-NMR and HPLC-NMR-MS has been used to characterise these metabolites in biofluids including urine [69]. It has also been shown that paracetamol can undergo deacetylation-reacetylation in vivo (futile deacetylation). This was first shown in the rat (see Section 8.3) but a human study has also been undertaken, showing that for the glucuronide metabolite there is about 1% futile deacetylation and for the sulfate metabolite the figure is about 2% [70]. Whilst most HPLC-NMR studies have been undertaken on urine drug metabolites are often also present in other biofluids such as blood plasma (albeit at low concentration). Blood plasma is, physico-chemically, a more complex biofluid than urine, with high concentrations of proteins and lipoproteins with multiphasic elements. The low molecular weight substances present may also bind to the plasma proteins resulting in complications for analysis by ~H NMR methods. However, the determination of drug metabolites in human blood plasma by HPLC-NMR has been demonstrated in a study of plasma from dialysis patients suffering from chronic renal failure [69]. The 750 MHz ~H NMR spectra of plasma samples from these subjects indicated the presence of a paracetamol-like metabolite which subsequently, using ~H HPLC-NMR at 600 MHz, was identified as paracetamol glucuronide. This is a somewhat unusual example as, in most cases, drug metabolites are rapidly eliminated from the plasma. However, for patients with renal failure the reduced ability of the subject to eliminate these compounds results in a build up in the plasma Tolfenamic acid is another NSAID which undergoes oxidation in vivo with the metabolites being conjugated with [3-D-glucuronic acid. This metabolism has been investigated using 800 MHz 1H HPLC-NMR spectroscopy of human urine following oral administration of the drug to a volunteer [71 ]. The stop-flow approach was used and a number of glucuronide conjugates were identified. These included those of the parent compound and of compounds with both methyl group and ring hydroxylation. The NSAID naproxen has also been studied using hyphenated methods, particularly with respect to phase II conjugation [72]. HPLC-MS and HPLC-NMR spectroscopic approaches were employed alongside each other. The combination of on-flow and stopped-flow HPLC-NMR, together with HPLC-MS data allowed determination of the metabolite structures. This paper also reported on work undertaken to study the effects References pp. 325-329

310

Chapter 10

of buffers on the chromatographic separations. Both phosphate and TFA buffers, often used in HPLC-NMR are impractical in HPLC-MS (or HPLC-NMR-MS) experiments due to ion suppression problems. The authors therefore studied the applicability of ammonium formate as a buffer for both systems resulting in effective ionisation for the MS. GW524W91 was intended for use as an anti-HIV infection agent. The human metabolite profile has been determined via a combination of methods including 19F NMR of human urine [73] and urine from a clinical trial has also recently been analysed by HPLC-~H NMR spectroscopy at 500 MHz [74]. In this work the stop-flow technique was used to examine each UV-absorbing peak in the chromatogram. ~H NMR spectroscopy was used to detect the presence of the characteristic doublet resonance (at a chemical shift near ~8) due to the nucleoside base proton. This approach enabled the identification of unchanged GW524W91 as the most abundant of the compound-related material in the sample. In addition, the de-aminated compound, and one of the diastereomeric sulfoxides were also observed. In the case of the glucuronide, identification was possible from the characteristic anomeric proton chemical shift appearing at ~4.51 and spectral assignment was confirmed via the use of a IH-1H 2-D TOCSY experiment, obtained by overnight data acquisition in stop-flow mode. For GW524W91 itself, it was also possible, due to the large amounts of material present in the sample, to obtain a ~H-~3C HMQC spectrum [11 ] from an overnight experiment.

10.8.3 Animal metabolism studies of pharmaceuticals and model compounds In the rat, paracetamol and its metabolites are excreted via the bile as well as in urine. HPLC-NMR has been used to analyse bile from cannulated animals dosed orally with paracetamol using reversed-phase chromatography with gradient elution. The major paracetamol metabolite present in the bile was the phenolic glucuronide with smaller quantities of the sulphate also present. Using HPLC-NMR, characteristic spectra for both of these compounds were obtained without difficulty, using the stop-flow technique [69]. Employing the same HPLC conditions used for the analysis of bile, ~H HPLCNMR spectra of the urine from the paracetamol-dosed rats were obtained, confirming the presence of the sulfate and glucuronide conjugates and paracetamol. The Nacetylcysteinyl conjugate was also detected and a ~H NMR spectrum obtained by stop-flow analysis. This spectrum required about 50 min. to collect and still gave a poor signal-noise ratio as expected for such a minor metabolite. In this type of application it is often necessary to compromise on the data acquisition regime, both in terms of achieved signal-noise ratio and in the spectral digital resolution. This may be the result of the need to preserve sample viability, or because the time scale of events being monitored does not permit long acquisition times. In such instances, advantages, in terms of information content, may be more readily obtained if data are subject to further processing after acquisition and so the maximum entropy technique [75] was applied to the NMR spectrum from the N-acetylcysteinyl metabolite of paracetamol and the optimum lineshape (Lorentzian) and linewidth were both obtained by maximising the probability value of the derived solution. The result clearly showed

Biomedical applications of directly-coupled chromatography

311

the loss of symmetry of the phenyl ring as a result of the formation of the Nacetylcysteinyl adduct with substitution meta to the N-acetyl of paracetamol confirmed by the ~H chemical shifts of the remaining aromatic protons. The non-equivalent methylene and the methine resonances of the cysteine moiety are also present in the spectrum, with chemical shifts consistent with S-substitution. Furthermore, signals could be observed for the N-acetyl protons, albeit somewhat attenuated by the acetonitrile solvent signal irradiation. It has recently been shown using NMR spectroscopy in conjunction with isotopelabelling studies that there is a significant degree of deacetylation-reacetylation (futile deacetylation) of paracetamol metabolites in vivo in the rat [76]. If this also occurs in humans, then it may help to explain the observed incidence of nephrotoxicity of paracetamol in that the process would result in levels of the potent nephrotoxin 4-aminophenol in vivo. Confirmation of the levels of futile deacetylation in individual metabolites of isotopically-labelled paracetamol in man has been achieved using directly-coupled HPLC-NMR spectroscopy at 600 MHz. In this study a solid phase extract of a 0-4 h urine after dosing with paracetamol-d3 was separated using HPLC with a methanol-water gradient elution. Methanol was used instead of the more usual acetonitrile to avoid the large methyl NMR resonance of the latter which would obscure any observation of transacetylated products. Good ~H NMR spectra were obtained from the sulfate and glucuronide conjugates of paracetamol-d3 and quantification of the level of transacetylated products for both of these metabolites was obtained by integration of the observed acetyl peak relative to the peaks from the aromatic protons. In man, it was shown that paracetamol glucuronide underwent transacetylation to an extent of 1% and for the sulfate conjugate the level was 2% [70]. This compares with results for the sulfate in the rat of about 10% [76]. Phenacetin was once in widespread use as an analgesic but, after being implicated as a cause of kidney toxicity, it was withdrawn from the market. Recently, the rat metabolism of phenacetin has been re-investigated using HPLC-NMR and HPLCNMR-MS [77]. This approach showed that the compound is metabolised principally to paracetamol with subsequent conjugation producing paracetamol glucuronide and paracetamol sulfate. N-hydroxyparacetamol was also tentatively identified. Using the same HPLC-NMR approach, the level of futile deacetylation in phenacetin in the rat was much higher than for paracetamol. This has been quantified in the major metabolites, paracetamol sulfate and paracetamol glucuronide, as 30% and 36% respectively using HPLC-NMR spectroscopy. The level of futile deacetylation for paracetamol and a further, tentatively assigned, metabolite, N-hydroxyparacetamol sulfate, was quantified at 32% [77]. The compound DPC423 is extensively metabolised in animals and man and this process has been investigated in detail using a combination of techniques including HPLC-NMR spectroscopy. A number of novel metabolic pathways were identified, including glutamate conjugation [78]. This has been investigated further using a series of analogues of DPC423 to investigate the transfer of glutamate to a drug benzylamine moiety [79]. The pathway has also been studied in relation to the metabolism of paracetamol [80]. References pp. 325-329

312

Chapter 10

There is a number of other HPLC-NMR and HPLC-NMR-MS studies which include the characterisation of xenobiotic metabolites. The metabolites of the potential antipsychotic agent iloperidone have been elucidated. Identification of metabolites in biological fluids from rats, dogs and humans was achieved using HPLC-MS-MS and, for bile in particular, HPLC-NMR was used to identify a number of structures [81]. Metabolites of the multidrug resistance inhibitor LY335979 have been characterised using HPLC-NMR from rat bile and from human liver microsome incubations. An Noxide metabolite was produced from oxidation of a quinoline nitrogen and, in addition, three glucuronide metabolites were identified, formed by conjugation after oxidation in the quinoline ring [82]. Metabolites of the potential anti-thrombotic drug, roxifiban, have been identified [83]. In another study, the metabolites of trifluoroperazine in the rat have been identified using HPLC-NMR spectroscopy [84]. The implementation of HPLC-NMR-MS, connected in series, has been demonstrated and used to identify metabolites of the non-nucleoside HIV reverse transcriptase inhibitor, GW420867. Again the major substance proved to be glucuronide conjugate of a ring-hydroxylated derivative of the compound [85]. Finally, the use of ion-exchange chromatography coupled with MS and NMR spectroscopy has been used to identify a novel metabolite of GW273629 [86]. The metabolic fate and urinary excretion of 2-bromo-4-trifluoromethylaniline has been studied in the rat using 19F NMR spectroscopy and directly-coupled H P L C - N M R MS [61 ]. The 19F NMR spectrum of whole rat urine collected 0-8 hours after i.p. dosing with 50 m g . k g -j of the compound is shown in Fig. 10.4 indicating the number and relative levels of fluorinated molecules in the urine. It was clear that there was very little of the parent compound in the urine from addition of authentic material. After a solidphase extraction step, HPLC-NMR and HPLC-NMR-MS experiments were carried out to identify the three most abundant species seen in Fig. 10.4. Thus, Fig. 10.5 shows the continuous-flow 19F NMR-detected chromatogram as a contour plot and this gives the retention times of the major fluorine-containing species. Subsequent H P L C - N M R - M S experiments with X9Fand ~H NMR spectroscopic detection and negative ion electrospray MS at these retention times demonstrated that the major metabolite (labelled A in Fig. 10.7) was 2-amino-3-bromo-5-trifluoromethylphenylsulfate accounting for 23% of the dose being excreted in the 0-8 hour urine. The spectra corresponding to this metabolite are shown in Fig. 10.6, with Fig. 10.6(a) being the 19F NMR spectrum, Fig. 10.6(b) showing the ~H NMR spectrum with the expected meta-coupled aromatic protons and Fig. 10.6(c) being the negative ion electrospray mass spectrum of the fully deuterated molecule. A similar approach was used to identify Peak B in Fig. 10.4 as 2-bromo4-trifluoromethylphenylhydroxylamine-N-glucuronide (7% of the dose) and Peak C in Fig. 10.4 as 2-amino-3-bromo-5-trifluoromethylphenylglucuronide (1.4% of the dose). In addition MS could be used to detect and identify a number of minor metabolites below the NMR detection limit [61 ]. Similar studies have also been carried out on 2-chloro-4-trifluoromethylaniline [87] and 3-methyl-4-trifluoromethylaniline and its acetanilide [88], whilst HPLC-NMR and HPLC-MS studies have also been performed on 2,3,5,6-tetrafluoro-4-trifluoromethylfluoroaniline [89], and a range of HPLC-NMR-MS and HPLC-ICPMS-TOFMS

"SIN-~IIAIN-DqdH ffu!sn p~g!luzp! szl!loqelztu ~ql mou~p ~) pue fl 'V "p!3e 3!ozu~ql,~ql~tuo:tong!:tl- ~ s e ~ (PlS luI) p~sn pIepuels IetU~lu! ~q,L "~u!I!UeI~ql~tuo-IonH.ul-#-otuoIq-~ jo uo!le.Ils!u.ttupe ie~uol!.t~d-e.Ilu! .t~lje ~u.un lea jo tunal3~ds ~tlNN c/6~ ZHIAI 9Ls "17"01 "ff!d tudd I

~;'E9" I

O'E9[

r 9l

O'I 9I

~09]

0"09 1

. . . .

g6~1

,

P~,S lUl

0

~3

Z>, r e,3

9 ,.,,a

I

~3

tr3 r r

%

V

314 Time (min)

Chapter 10

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

318

Chapter 10 glucuronic acid

HO HOOC

O

HO HOOC

0

OH

0

13-1-O-acyl

+ RCOOH

HO HOOC

13

O

OH

1L HO HOOC

~o.

HO HOOC,

O

~ O

OCOR cz-2-O-acyl

13-2-O-acyl

HO HOOC

R

C

HO HOOC

O

~ 13-3-O-acyl

HOOC,

O

OH OH

RCO0

OH

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 chrornarography

References pp. 325-329

4,1',(o0 I

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

3,1',(=)I

, l

II

3,3',((x)

,

I ,

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

. 4,5',(=)

II

2,2',([~)

I

II

d

4,2~,(o0

4,2',(I])

2,4',(~p)

3,5',(ct, p)

!:!I' I;

3O Reaction time domain (Hours)

r

,

i

2O i tt

I

)~

i

I

i

,

!

I

!

(~p,) 2-O-acyl

(~

l

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

!

] w

(P)

i,

i

i

5.1

,

,

,

r

I 4.6

'

'

'

.

(o0 3-o-acyl ''

/

4.

~

|

'

Ox) 2-O-acyl

'

~O-aWl '

(=,p) 2,0-acyl i

]

3.6

i

,

....

~--

ppm

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.

"~

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.

This Page Intentionally Left Blank

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 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

H20

Reduction

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

2.5

,.51

2

Expanded for flint 24 houm 1.5

m

E 1.5 W

1

Q. .J

E 0.5

a.

"O

1

0 0

5

10

15

20

25

0.5

0

20

40

60

80

100

r

T

120

140

160

Hours post dose

i- + -

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

Volunteer 4 1

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

the samples contain less than 2.5 dpm/mL.

~-~

Ultra-sensitive detection of radiolabelled drugs

341

3.5

2.5 c 0

=(.1 tl

2

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 I I

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

0.5000 0.4500 0.4000

0.3500 0.3000 J

"~ 0.2500 "0

0.2000 0.1500 0.1000 0.0500

0.0000

0

10

20

30

40

50

60

Percent dilution 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.

~

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

C~

Background peak

0.0035 0.0030 i

c o

Parent drug

0.0025 !

,_m 0.0020-I Fraction collection

L_

a. 0.0015 ? E O.0010 -

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

j 1

/

/

O.OOO5 ~.t L i

0 . 0 0 0 0 -+ 0

5

10

15

18.5

21

26

31

36

41

Retention time (min) Fig. 11.7. HPLC-AMS chromatogram produced from 0.024 dpm injected on column. The peak eluting at eluting from the column.

ca

41 min. is not drug related but background carbon

r~

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.

This Page Intentionally Left Blank

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations, Vol. 4 9 2003 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.

% 120.0%

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

r~

"..,I

100.0%

Xx x@ x

,..q

X XX

80.0%

X X

I::

0 ,.,.,, C 0, .m

X X

60.0%

X

X

40.0%

X

_.

X X

20.0% !

X X ................ ,

0.0% 5.0

6.0

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

, 8.0

...... , ......... 9.0

, ........ 10.0

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

, .......... 11.0

12.0

13.0

14.0

X

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

15.C

IE (eV)

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

75000 A

e,O Ill

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

C 26

2.56E*'04

cN

q

m

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

9. . . . .

o.

N

~.14[,~54,

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

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

'r

50000

25000

0

.

.

.

.

.

L____

11

300

.aL

.

.

4OO

.

.

500

-

Ill-

I

600

700

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

loo3.49 .

8,70 .84

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

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 3600

367

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

%

~.~~

t~

t~ --q

"-,,i

(A)

r

0

I

-~ 0.05

A ~..,~

j

a6Zrl

~.

t.,u

,1

(B) .,,..,,

.

~k..~ __

.

.

.

I !0000c~., I,

1o0~r

'-'sB~_ __J

. . . .

Br o

|

2

3

4 5 6 7 Retention time / rain

8

9

110

0

i

i

l

A _ i

_

i

2

i

i

I 100000 cps

i

i

i

i

I

i

i

i

3 4 5 6 7 Retention time / rain

i

i

8

i

i

9

t~

i

10

Fig. 12.10. Element selective chromatograms for trace elements in human serum (A) U V detection, (B, C) ICP-MS data highlighting which elements are bound to large sized molecules (first peak) and those bound to small sized molecules (second peak) [47].

Chapter 12

370

human body. The technique offers selectivity, sensitivity and quantitative determinations and as such it is rapidly gaining acceptance as a routine tool for biomedical and pharmaceutical research.

12.5 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

R.S. Houk, V.A. Fassel, G.D. Flesch, H.J. Svec, A.L. Gray and C.E. Taylor, Anal. Chem., 52 (1980) 2283. T.B. Reed, J. Appl. Phys., 32 (1961) 821. R.E Browner and A.W. Boom, Anal. Chem., 56 (1984) 786A. P.J. Turner, Some observations on mass bias effects occurring in ICPMS systems, in: G. Holland and A.N. Eaton (Eds.), Applications of Plasma Mass Spectrometry (pp. 71-78), Royal Society of Chemistry, 1991. R.E. Sturgeon, J.W.H. Lam and A. Saint, J. Anal. At. Spectrom., 15 (2000) 607. N. Furuta, A. Takeda, J. Zheng and T. Nabeshima, Evaluation of Inductively Coupled Plasma-Ion Trap Mass Spectrometry, in: G. Holland and S.D. Tanner (Eds.), Plasma Source Mass Spectrometry (pp. 9096), Royal Society of Chemistry, 2001. I. Feldmann, N. Jakubowski and D. Stuewer, Fresenius, J. Anal. Chem., 365 (1999) 415. I. Feldmann, N. Jakubowski, C. Thomas and D. Stuewer, Fresenius' J. Anal. Chem., 365 (1999) 422. C.J. Duckett, N.J.C. Bailey, H. Walker, E Abou-Shakra, I.D. Wilson, J.C. Lindon and J.K. Nicholson, Rapid Commun. Mass Spectrom., 16 (2002) 245. D. Beauchemin, M.E. Bednas, S.S. Berman, J.W. McLaren, K.W.M. Siu and R.E. Sturgeon, Anal. Chem., 60 (1988) 2209. J.C. Van loon, L.R. Alcock, W.H. Pinchin and J.B. French, Spectrosc. Lett., 19 (1986) 1125. A.W. kim, M.E. Foulkes, L. Ebdon, S.J. Hill, R.L. Patience, A.G. Barwise and S.J. Rowland, J. Anal. At. Spectrom., 7 (1992) 1147. T. De Smaele, L. Moens and R. Dams, Coupling of GC with ICPMS for trace metal speciation, in: G.Holland and S.D. Tanner (Eds.), Plasma Source Mass Spectrometry (pp. 109-123), Royal Society of Chemistry, 1997. E.M. Krupp, C. Pecheyran, S. Meffan-Main and O.E Donard, Fresenius' J. Anal. Chem., 370 (2001) 573. D. Glindemann, G. Ilgen, R. Herrmann and T. Gollan, J. Anal. At. Spectrom., 17 (2002) 1386. J.W. Olesik, J.A. Kinzer and S.V. Olesik, Anal. Chem., 67 (1995) 1. Y. Liu, V. Lopez-Avila, J.J. Zhu, D.R. Wiederin and W.E Beckert, Anal. Chem., 67 (1995) 2020. V. Majidi, J. Qvarnstrom, Q. Tu, W. Frech and Y. Thomassen, J. Anal. At. Spectrom., 14 (1999) 1933. L. Bendahl, B. Gammelgaard, O. Jr O. Farver and S.H. Hansen, J. Anal. At. Spectrom., 16 (2001) 38. J.K. Nicholson, J.C. Lindon, G. Scarfe, I.D. Wilson, ER. Abou-Shakra, J. Castro-Perez, A. Eaton and S. Preece, Analyst, 125 (2000) 235. O. Corcoran, J.K. Nicholson, E.M. Lenz, ER. Abou-Shakra, J. Castro-Perez, A.B. Sage and I.D. Wilson, Rapid Comm. Mass Spectrom., 14 (2000) 2377. J.K. Nicholson, J.C. Lindon, G.B. Scarfe, I.D. Wilson, ER. Abou-Shakra, A.B. Sage, G. Harland and J. Castro-Perez, Quantification and idntification of 2-bromo-4-trifluoromethylaniline metabolites in rat urine using HPLC-ICP-MS/TOF/MS, in: E. Gelpi (Ed.), Advances in Mass Spectrometry (Vol. 15, pp. 659-661), Chichester, John Wiley and Sons, 2001. J.K. Nicholson, J.C. Lindon, G.B. Scarfe, I.D. Wilson, ER. Abou-Shakra, A.B. Sage and J. CastroPerez, Anal. Chem., 73 (2001) 1491. E Marshall, O. Heudi, S. Mckeown, A. Amour and ER. Abou-Shakra, Rapid Commun. Mass Spectrom., 16 (2002) 220. O. Heudi, C. Ramirez-Molina, E Marshall, A. Amour, S. Peace, S. McKeown and E Abou-Shakra, Journal of Peptide Science, 11 (2002) 591. C.J. Duckett, I.D. Wilson, H. Walker, E Abou-Shakra, J.C. Lindon and J.K. Nicholson, in preparation.

Biomedical applications of inductively coupled plasma mass spectrometry 27 28

371

E. Hywel Evans, J.C. Wolff and C. Eckers, Anal. Chem., 73 (2001) 4722. C.J. Smith, I.D. Wilson, E Abou-Shakra, R. Payne, A.C. Parry, E Sinclair and D.W. Roberts, Anal. Chem., 2003 in press. B.O. Axelsson, M. J6rnten-Karlsson, E Michelsen and ER. Abou-Shakra, Rapid Commun. Mass Spectrom., 15 (2001) 375. T. Hunter, B.M. Sefton, (Eds), Protein Phosphorylation, Parts A and B-Methods in Enzymology, Vols. 200 and 201, Academic Press, San Diego, 1991. M. Wind, M. Edler, N. Jakubowski, M. Linscheid, H. Wesch, and W.D. Lehmann, Anal. Chem., 73 (2001) 29. M. Wind, H. Wesch and W.D. Lehmann, Anal. Chem., 73 (2001) 3006. E Marshall, O. Heudi, S. Bains, H.N. Freeman, E Abou-Shakra and K. Reardon, Analyst, 127 (2002) 459. J. Feldmann, K. John and P Pengprecha, Fres. J. Anal. Chem., 368 (2000) 116-121. A. Shraim, S. Hirano and H. Yamauchi, Anal. Sci., 17 (2001) i1729. Q. Xie, R. Kerrich, E. Irving, K. Liber and E Abou-Shakra, J. Anal. At. Spectrom., 17 (2002) 1037. K.L. Ackley, C. B'Hymer, K.L. Sutton and J.A. Caruso, J. Anal. At. Spectrom., 14 (1999) 845. R. Bentley and T.G. Chasteen, Chem. Educator, 7 (2002) 51. M. Miguens-Rodriguez, R. Pickford, J.E. Thomas-Oates and S.A. Pergantis, Rapid Comm. Mass. Spectrom., 16 (2002) 323. R. Lobinski, J.S. Edmonds, K.T. Suzuki and EC. Uden, Pure Appl. Chem., 72 (2000) 447. J.M. Marchante-Gayon, I. Feldmann, C. Thomas and N. Jakubowski, J. Anal. At. Spectrom., 16 (2001) 457. J.M.G. Lafuente, M. Dlaska, M.L.E Sanchez and A. Sanz-Medel, J. Anal. At. Spectrom., 13 (1998) 423. B. Gammelgaard, L. Bendahl, U. Sidenius and O. JCns, J. Anal. At. Spectrom., 17 (2002) 570. R. Wahlen, LC-GC Europe, 15 (2002) 2. S. Slaets, E Adams, I. Rodrigues Pereiro and R. Lobinski, J. Anal. At. Spectrom., 14 (1999) 851. R.C. Rodriguez Martin-Doimeadios, E. Krupp, D. Amouroux and O.EX. Donard, Anal. Chem., 74 (2002) 2505. K. Inagaki, T. Umemura, H. Matsuura and H. Haraguchi, Anal. Sci., 16 (2000) 787.

29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

12.6

APPENDIX

[M ] = ne ~

-~

j-~exp-k-~o~

"

where: [M +] = the p o p u l a t i o n o f p o s i t i v e l y c h a r g e d ions o f the e l e m e n t M [M] = the p o p u l a t i o n o f a t o m s o f the e l e m e n t M Re

= the e l e c t r o n n u m b e r d e n s i t y in the p l a s m a

me h

= the m a s s o f the e l e c t r o n

L

= the f r e e e l e c t r o n t e m p e r a t u r e

Q+ QO IP Tion

= Planck's constant = the e l e c t r o n i c p a r t i t i o n f u n c t i o n o f the ion = the e l e c t r o n i c p a r t i t i o n f u n c t i o n o f the a t o m -

the i o n i s a t i o n p o t e n t i a l o f the e l e m e n t

= the i o n i s a t i o n t e m p e r a t u r e

References pp. 370-371

This Page Intentionally Left Blank

I.D. Wilson (Ed.), Bioanalytical Separations Handbook of Analytical Separations,Vol. 4 9 2003 Elsevier Science B.V.All rights reserved

373

CHAPTER 13

Chromatography in a regulated environment H.M. Hill Huntingdon Life Sciences, Woolley Road, Alconbury, Huntingdon, Cambridgeshire, PE28 4HS, U.K.

13.1 INTRODUCTION The scope of this chapter is to discuss the regulatory issues surrounding the application of chromatographic techniques to the analysis of drugs in biological fluids. It is not intended to discuss in detail the technical aspects of a wide range of chromatographic techniques, however, where the "how to" implementation of regulations is made clearer by reference to the technical issues this will be done. Technical issues are discussed in more detail in other chapters within this book as well as a range of other articles and books dedicated specifically to analytical methodology. Although predating the widespread application of LC-MS (MS) to bioanalysis, a good general introduction to the subject and issues surrounding bioanalysis is to be found in the book by Chamberlain [1]. More recently, the U.K. Drug Metabolism Discussion Group (U.K.) have produced an excellent primer for bioanalysts [2]. Issues surrounding specific areas are discussed in detail in the series Methodological Surveys In Biochemistry, Ed E Reid [3-16]. Chromatographic techniques now cover a broad spectrum of technologies and are thoroughly reviewed in the above references. While LC-MS-MS predominates, the remaining technologies continue to fill niche roles. Historically chromatographic techniques used in bioanalysis have reflected the developing technologies in other industries, e.g. food, petrochemical, pesticides, environmental chemicals, forensic science, etc. However, the development of atmospheric pressure interfaces in MS detection systems linked to HPLC has been driven by the pharmaceutical industry and the need to quantify drugs in biological fluids at increasingly lower concentrations. In addition other applications of atmospheric pressure MS systems to elucidate the structure of biological polymers and the role of LC-MS in high throughput screening studies have been similarly driven by the pharmaceutical industry. As such this chapter will concentrate on the application of LC-MS based assays to bioanalysis, although in References pp. 406-409

374

Chapter 13

general the issues discussed are applicable to most of the chromatographic techniques in current use. In general, because no International Conference on Harmonization (ICH) process has been developed to cover this aspect of drug development, the FDA guidelines which have been the most thoroughly documented and discussed throughout the world, form the basis of most companies in-house guidelines. This chapter discusses regulatorybased studies, which are subject to these criteria, and does not discuss discovery-based assays. The FDA's Guidance for Industry, Bioanalytical Methods Validation (BMV) [ 17] forms the basis for this chapter and where appropriate, variations from other regulatory bodies will be discussed. The Guidance for Industry at best represents an empirical consensus which reflects both scientific and financial drivers as well as the litigious nature of the society for which they have been developed. Larry Thompson [18] of the FDA, citing the FDA's chief attorney Porter, described this conundrum thus: "The scientist's preference is always to keep options open", adding, "Think about what the regulated industry needs from the agency. The industry needs predictability, and legally if you want to hold people to standards, the standards that you use must be clear and predictable and applied consistently". As a consequence guidelines provide specifications which may or may not have statistical contradictions but which, in the main, err on the side of "safety". The Guidance has been some years in "production", although it has been long foreshadowed by Shah, the senior author of the FDA-BMV [17], who published a paper in 1987 on Bioanalytical Methods Validation [19]. In addition a number of workers [20,21] reviewed their approach to bioanalytical methods validation prior to the 1990 Crystal City Consensus meeting. The Crystal City Consensus Document was published in April 1992 [22]. There then followed a range of meetings that discussed this document and issues associated with its implementation [23-26]. Other critiques discussed the statistical inconsistencies and limitations [27,28] of the document, considered later under the relevant sections with respect to the practical implications and applications of the current guidelines. Some six years later the FDA published the Draft Guidance for Industry, Bioanalytical Methods Validation for Human Studies, in December 1998 [29]. The FDA sought feed back on this document from individuals and corporations. As part of that process an international meeting Bioval 1999, held in London (June 1999) provided some insights into European thoughts, culminating in the January 2000 meeting "Crystal City Revisited- A Decade of Progress", in Crystal City. This meeting discussed the relevance of the Guidance [17] to small molecules, while a further Consensus meeting in March of the same year discussed its relevance to macromolecules. The consensus document was published in December 2000 [30] and the FDA provided the definitive "Guidance for Industry" Bioanalytical Methods Validation in May 2001 [ 17]. While the original draft document was 10 pages, the final document was 22 pages long, and at the behest of the industry, encompasses all species, not just Human Studies, as the draft guidance did, all matrices and methods, be they chemical or ligand assays for "small molecules". A similar FDA guidance document for

Chromatography in a regulated environment

375

macromolecules is yet to appear, although a meeting report was published in July 2001 [31]. Subsequent to the publication of this Guidance, a "How To" meeting sponsored by the AAPS, FDA, RSC, RPSGB and FIE B IOVAL 2002 was held in London in February 2002; while a final report is awaited from this meeting some of those thoughts are incorporated into this chapter.

13.2 REGULATORY ISSUES The application of regulatory guidances/guidelines, which detail specific requirements as to study conduct and documentation for the submission of bioanalytical data should be differentiated from the 'compliance' requirements of GxPs, i.e. GLP, GLP and GME In addition any study submitted to the FDA should comply with 21 CFR Part 11 [32], the intricacies as to how these requirements can be met have been the subject of many meetings/presentations and publications although not all have been directly relevant to bioanalytical studies. Thus, 21 CFR 11, which pertains to electronic records and electronic signatures, should be applied to any system regardless of the implementation of GxPs. While the FDA Guidance provide pivotal documentation with regard to Bioanalytical Methods Validation, the development of best practices and subsequent interpretation by the regulatory authorities provides further insight into the current implementation of this Guidance- it is essential to note that this is a guideline and alternative interpretations may be justified. At Bioval 2002, Dr Shah (FDA) noted, "The Guidance represents the best scientific judgement and current thinking, it is informal and non-binding. It does not create or confer any rights for, or on any person, and does not operate to bind FDA or the public. An alternative approach may be used if such approach satisfies the requirement of the applicable statute, regulation or both". Additional documentation, e.g. the FDA Compliance Manual for in vivo B ioequivalence Studies [33] for internal use by FDA inspectors and reviewers, provides an insight as to how studies submitted to the Divisions of Pharmaceutical Evaluation or the Division of Clinical Pharmacology and Biopharmaceuticals (as part of NDA submission) and for ANDA submissions, by the Division of Generic Drugs should be evaluated. Further interpretations and insights may be obtained from the result of such reviews, and inspections under the freedom of information (FOI), such as Establishment Inspection Reports (EIR) and Form 483s issued as a result of inspection reports detailing deviations from expectation, e.g. from GLP and/or 21 CFR 320. In addition there are reports of FDA meetings, to be found on the FDA website, either alone or in cooperation with other interested parties. Other documents have been developed by the Industry to share best practices. One such document is the Metabolites In Safety Testing (MIST) document [34], which discusses the issues around the relevance in quantifying metabolites in safety studies, another details [35] the FDA's thoughts on the development of Chiral drugs. Both of these documents have a significant impact on the conduct, timing, etc. of bioanalytical method development, validation and References pp. 406-409

376

Chapter 13

study execution. A list of relevant documentation and appropriate websites are to be found in the reference section.

13.2.1 Regulatory environment Perhaps the most overarching document which impacts on all aspects of drug development is that of the electronic data and electronic signatures. For the FDA the requirements are detailed in 21 CFR Part 11. Currently there are no other regulatory bodies providing guidance on the submission of electronic documentation. However, it is likely that other agencies will develop similar requirements. In any case as a consequence of the global nature of the pharmaceutical industry, it is imperative that, where a drug is likely to be submitted to the FDA, 21CFR 11 is implemented. Although this ruling has been in force since 1997, very little commercially-available software is either claimed to be "compliant" with the requirements of 21 CFR 11, or is able to be made 21 CFR 11 compliant, even where claims are made. In the context of chromatographic assays for bioanalytical studies, this largely applies to the collection and processing of chromatographic data as well as subsequent processing and collating of data in a commercial or in-house data base. Similarly the FDA Guidance for Industry, BMV [17], can be implemented in a GLP or non-GLP environment; implementation of this Guidance is not synonymous with GLE Indeed, in the context of the FDA, bioanalysis is an integral part of non-GLP based guidances, e.g. 2 1 C F R 3 2 0 - Bioavailability and Bioequivalence Requirements (Drugs for Human Use) [36]. A similar EU document, CPMP Note for Guidance [37] - has recently been published and has a requirement for bioanalysis to be carried out according to the principles of GLP, it should be noted that this is NOT the same as requiring such studies to be done to GLE It is however expected that study plans/ protocols be prepared, and that audit trails for all processes and data generation should exist.

13.2.2 Compliance with GLP? ICH-53A; Toxicokinetics: Guidance on the assessment of systemic exposure in Toxicity Studies [38]. As the ICH appellation implies it is incorporated into the national legislation of those countries committed to the concept of the ICH including the U.S. Federal Register. This document definitively requires the application of GLP to both the toxicological aspects of the study and to the related bioanalysis. GLP was developed as a consequence of inadequacies in preclinical studies, although as it is now defined, it applies to non-clinical studies. As a consequence it is self-evident that the application of a bioanalytical method to a toxicokinetic study should be carried out in compliance with GLE However, when it comes to the validation of a bioanalytical method the requirement to carry out these aspects in compliance with GLP is debatable, indeed in both the U.K. and Japanese regulations validation is considered a non-GLP activity, although a preference that it should be carried out in a GLP environment is selfevident throughout the industry.

Chromatography in a regulated environment

377

The FDA requirements for B MV [ 17] imply that the analytical laboratory conducting pharmacology/toxicology and other preclinical studies for regulatory submission, should adhere to the FDA's GLP's and to sound principles of quality assurance throughout the testing process. On the other hand bioanalytical methods for human bioavailability (BA), bioequivalence (BE), pharmacokinetic studies (PK) and drug interaction studies must meet the criteria of 21 CFR 320.29 and are essentially not covered by GLP; so what guidelines should apply? The FDA Bioanalytical Methods Validation Guidance [17] provides some insight, i.e. "The analytical laboratory conducting BA and BE studies should "closely adhere" to FDA's Good Laboratory Practices (GLPs) and to sound principles of quality assurance throughout the testing process". The legal claim 'compliance with GLP' is replaced with the term should 'closely adhere to FDA's GLPs'. Many laboratories have developed similar 'statements' which, while not claiming compliance with G L P - claim processes are in operation that are closely related to it, i.e. "This study was carried out in laboratories which are GLP certified", or "This study was carried out in accordance with the principles of GLP". These terminologies are reflected in the CPMP note of Guidance on the Investigation of Bioavailability and Bioequivalence [37], where it states that the bioanalytical part of bioequivalence trials should be conducted according to the applicable principles of GLP, but note, NOT compliant with GLE The FDA 21 CFR Part 320 Bioavailability and Bioequivalent Requirements (Drugs for Human Use) [36], are somewhat less definitive about the bioanalytical requirements, i.e. 21 CFR 320.29, states "The analytical method used in an in vivo bioavailability study to measure the concentration of the active drug ingredient or therapeutic moiety, or its metabolites in body fluids or excretory products or the method used to ensure an acute pharmacological effect, shall be demonstrated to be accurate and of sufficient sensitivity to measure, with appropriate precision the actual concentration (accuracy?) of the active drug ingredient or therapeutic moiety or its metabolites achieved in the body." However, this predates the BMV Guidance [17]. As such, the latter document amplifies these requirements, which are reiterated in the FDA's compliance manual [33] for inspecting BA and BE studies, i.e. "The analytical laboratory should have a written set of standard operating procedures (SOPs) to ensure a complete system of quality control and assurance". The SOPs should cover all aspect of analysis from the time the sample is collected and reaches the laboratory until the results of the analysis are reported. The SOPs should include record keeping, security and chain of sample custody (and integrity). Pragmatically the application of GLP in the context of the bioanalytical laboratory at the operational level should satisfy the requirements of studies from all areas of the drug development process, although compliance is only required and possible for preclinical studies. While it is inappropriate to claim compliance with GLP for clinical studies, it would appear difficult for laboratory procedures to be carried out under different codes. As such, attempts to harmonise the GxPs* at the laboratory level, would appear to be the rational way forward. A major driver in this respect is the implementation of the * GxP is a generic term which covers the commonelements of GME GLP and GCP. References pp. 406-409

378

Chapter 13

FDA's electronic records and signatures requirement under 21CFR 11, which applies equally to GLP, GMP and GCR Similarly a standardised approach to instrument qualification would seem to make practical sense. There already exist some approaches to this issue, in 1994 the FDA's Reviewers Guidance for evaluating/reviewing analytical methods, applies to both bioanalytical and pharmaceutical analytical studies. Recently BARQA (The British Association for Quality Assurance) developed some tentative Good Clinical Laboratory Practice (GCLP) [39] Guidelines, to encompass all laboratory-based activities that support clinical studies in safety clinical chemistries and for the bioanalytical support of pharmacokinetic and pharmacodynamic studies. It would seem appropriate therefore to embrace this trend and to develop requirements for analytical laboratories which cross the GxPs.

13.2.3 Instrument qualification and validation In addition to the implementation of the relevant GxR it is important that the quality of the instrumentation used for both GMP and GLP studies is fit for purpose. Although there are differences in the wording of GLP and GMP requirements as to what is expected, e.g. the U.K.'s MCA Orange Book incorporating the EU GMP Registration Annex 15, page 182 [40] requires equipment, which is critical for the quality of the products, to be subjected to appropriate qualification. The FDA GLP regulations require that equipment, which is used for generation, measurement or assessment of data, is adequately tested, calibrated and/or standardised. There is, however, a developing trend to apply the GMP requirement to qualify equipment more rigorously to GLP studies, although, as detailed above, the GLP guidelines are not rigorously defined. The qualification process is generally divided into a number of stages, i.e. Installation Qualification (IQ), Operation Qualification (OQ) and Performance Qualification (PQ). A pragmatic approach to instrument Qualification has been discussed by Freeman et al. [41 ] in their position paper on the Qualification of Analytical Equipment. Qualification is the responsibility of the end user and is designed to ensure that the equipment is both fit for the purpose and continues to function so, over the 'lifetime' of the equipment. Responsibility cannot be abrogated to the supplier. This does not however, prevent suppliers providing help in the form of protocols and targeted servicing.

13.3 BIOANALYTICAL VALIDATION PROCESS Full method validation is required for a new drug entity and when implementing a developed method for the first time. Importantly the guidance stresses the need to fully validate a method when a "new " metabolite is identified and needs to be quantified to determine its kinetics. Validation of a bioanalytical method comprises two distinct phases:

Chromatography in a regulated environment

379

(1) The pre-study phase in which the characteristics of the assay with respect to stability, specificity (selectivity), accuracy, precision, limits of quantification (sensitivity) and response function are determined. (2) The study phase itself in which the validated bioanalytical method is applied to the actual analysis of samples from the bio-study mainly in order to confirm the stability, accuracy and precision, i.e. the method performs to the same characteristics as the validated method. The Guidance defines two additional types of validation: (1) Partial validation recognises that when modifications are made to a fully validated method, the impact of these changes on the integrity of the methodology should be fully addressed. The practical implication of changing methodology and circumstances necessitating such changes are discussed. (2) Cross Validation is defined as the comparison between two or more different bioanalytical methods, or perhaps more importantly, a comparison between different sites or laboratories. This is essential when data from different sites is generated for the same study.

13.3.1 Full validation

13.3.1.1 Pre-study phase A major prerequisite before starting validation is the need to use fully characterised test articles whose purity is defined. While this is best achieved by use of certified compendial, reference standards, commercially available materials from recognised suppliers and or custom synthesis accompanied by a valid certificate of analysis are also acceptable. For chemical (chromatographic) assays to be validated, the response concentration relationship (calibration or standard curve) should be reproducible and consistent. Its robust establishment together with a well-characterized purity of the standard are essential prerequisites to ensuring the accuracy of an analytical method. In addition the sensitivity selectivity, precision, "recovery" and stability of the analyte spiked into the matrix under investigation must be evaluated.

13.3.1.1.1 Response function (calibration or standard curve). While the standard curve, for a linear function, need only consist of 6 points [17], when establishing an appropriate model and weighting, then more points may be appropriate. Establishing a response function for a specific method can be achieved by plotting the response against concentration, the best fit may be determined using a variety of procedures. Burrows [42], proposed that replicates (n=6) be evaluated at each of up to 10 different concentrations over the standard curve range and the mean unit response of each concentration be plotted against the concentration. Figure 13.1a compares a typical calibration function of y = mx + c (unweighted) where y = response, m = gradient of slope of the curve, x = concentration and c = the intercept. In the y = mx + c model the References pp. 406-409

oc O

(a)

(b)

GC-NP Detector 120.0

$

Response (y) 12 Data

10_

__.

w--1

----.

w = llxx

Apparent Overall Sensitivity (y/x)

*

Data

. . . . Linear ..... Quadratic

100.0

B-W

4)

w = l/xx

_

80.0

i

1

4-

10

1 O0

1000

Concentration (pg/ml)

_

0

I

2

I

4

I

6

Concentration (l~glml)

!

8

10

Linear

y =

Quadratic

y2 = ax + bx 2 + c

B-W

y = a + b x + c x In x

mx + c

Fig. 13.1. (a) Calibration curve indicates little difference between a linear fit using different weightings. (b) Sensitivity plot; unit response of analyte vs. concentration shows relationship is nonlinear and is best fitted. (Permission from J. Burrows).

Chromatography in a regulated environment

381

A p p a r e n t Overall

Sensitivity (y/x)

180 Data 160 140

........ w =

\

1

........ w = l / x x

120 1L 100 80 0.01

I

I

0.1

1

Concentration

10

(pg/ml)

Fig. 13.2. Impact of weighting on calibration curve fit at the lower concentrations. (Permission from J. Burrows).

curve is obviously (sic) linear, however the unit response curves show it to be curvilinear (lb). Figure 13.2 compares different models and weighting using the same data, where 1

the "best visual" fit is obviously a quadratic function with a weighting of ~5- The rationale behind the development and evaluation of the "best model" has been discussed elsewhere. Pateman [43] showed that a minimum of 5 points for a linear standard curve has a statistical basis. The confidence interval of the standard curve changes significantly as the number of points increases to 5 and very little when the number of calibration points increase from 6-10 (and not at all between 10-50 points). Indeed the FDA BMV requires justification of the standard curve model used. Once established this fit should be used for all sample analysis. However, where changes in a detection system necessitate a review of the changing response function, a change may be appropriate, if it can be justified on scientific grounds. In addition, estimation of concentrations above the highest calibration standard, i.e. the upper limit of quantification (ULOQ) or below the lower limit of quantification (LLOQ) by extrapolation is not recommended (this is regulatory speak for "do NOT do it"). The ability to dilute samples those concentrations above the upper limit of quantification (ULOQ) of the method should be demonstrated by determining their accuracy and precision. This can be achieved by preparing a QC sample at a concentration higher than the anticipated maximum concentration expected in test samples, e.g. 10 x mid-point of calibration line) and diluting this sample 10-fold with the same biological matrix prior to analysis. The precision and accuracy of the measurement of this sample should meet the criteria previously discussed. Once the dilution has been validated, the need to incorporate actual within-study over-range QC samples is obviated. References pp. 406-409

382

Chapter 13

13.3.1.1.2 Accuracy. Accuracy as determined by replicate analysis of spiked samples containing known amounts of analyte, should be measured at minimally four concentrations over the entire calibration range, up to 3 x LLOQ(Lo)), middle (Me) and high (Hi) concentrations - as well as at the LLOQ of the method. A minimum of five determinations per concentration is required during a single analytical batch to establish within-batch accuracy. Inter-batch measurements should be determined by analysis of QC samples at the same concentrations on, minimally, three separate occasions. The mean measured concentration should be within 15% of the actual concentration except at the LLOQ, where it should not deviate by more than 20%. The deviation of the mean from the true value being a measure of accuracy. Some laboratories evaluate the ULOQ in this way - this is particularly important where the response function is curvilinear. 13.3.1.1.3 Precision. Precision, in terms of the closeness of agreement between measurements from multiple sampling of the same homogeneous sample, should be determined at a minimum of three concentrations and at the LLOQ of the method, at the concentrations defined for accuracy determination. There is, however, a trend to include the defined ULOQ concentration in this evaluation in order to confirm acceptable accuracy and precision at the ULOQ. A minimum of five determinations at each concentration within a single batch should be undertaken to establish within-batch precision, known variously as, inter-batch (run) precision or repeatability. Measuring precision over time (minimally three occasions), which may involve different analysts, equipment, reagents and laboratories, if appropriate, should be determined (sometimes known as intermediate precision). The precision determined at each concentration should not exceed coefficient of variation (CV) of 15%, except at the LLOQ, where it should not exceed 20%. 13.3.1.1.4 Limit of quantification (sensitivity). In the context of this guidance, sensitivity refers to the lower limit of quantification (LLOQ), which is defined as the lowest calibration standard which meets the acceptance criteria of 20% precision and accuracy. This may be different from the LLOQ as defined by Kaiser [44] where the LLOQ is based on the ratio of the signal to the variability of the noise, i.e. the mean background signal of more than 20 blanks plus 10 times the standard deviation of those blanks. Usually this value is similar to the Guidance LLOQ value but not always, as it is largely dependent on the variability of a large number of blank samples while the Guidance LLOQ is based on the variability of the assay using the same matrix. Dell [45] and Phillips [46] discuss this in more detail. Other criteria are used to define sensitivity, i.e. based on the response of the "system" to a unit of analyte or on the slope of a calibration curve, thus a more sensitive assay will have steeper slope than a less sensitive assay, although this may have no relevance to the LLOQ as defined by the criteria used above. The limit of detection (LOD) has no relevance in determining the pharmacokinetics of a compound except in those situations where no plasma levels are found, in such cases a limit assay would seem more appropriate and is discussed in detail later. Although some pharmacokineticists have justified that data, beyond this limit can be useful in deriving pharmacokinetic parameters, Gabrielsson [47].

Chromatography in a regulated environment

383

13.3.1.1.5 Recovery. While many analysts feel that the determination of this parameter adds little to the validation process, recovery is a requirement to determine the extent of the analyte recovered during sample processing. To be of practical value this parameter should be evaluated prior to validation, although it should be formally determined during the validation process. A high recovery is not essential, as long as it is reproducible, even concentration-related changes might be accounted for by different calibration m o d e l s - as long as they are consistent and reproducible. It is recommended that recovery experiments be performed across the calibration range (at low, medium and high concentrations). Although obtaining meaningful data near the LLOQ presents practical p r o b l e m s - as such, the data should be viewed with caution. The type of recovery discussed here should be differentiated from extractability - where in a liquidliquid extract, the "extractable recovery" is based on the amount of analyte that partitions into the organic or extracting phase while absolute recovery takes into account transfer losses as well as extractability. Recovery should be viewed as a 'nice to know' parameter which can provide useful information for improving the analytical method but does not restrict its applicability. 13.3.1.1.6 Selectivity and specificity. Absolute selectivity, i.e. a specific method is a rare phenomenon and cannot be guaranteed, hence we always deal with degrees of specificity which are defined as the selectivity of the method. On an ongoing basis through the drug development programme, there will be situations which challenge the selectivity of the assay ranging from patient specific issues, e.g. diseased states, coprescribed drugs as well as sample-based issues like contamination introduced through sample processing from collection to extraction such as tube contaminants, preparation of stabilisers or anticoagulants, contaminants in the extracting/processing organic solvents. While the patient specific issues are deemed to be revalidation issues, i.e. partial validation, and are discussed under this heading, the others may be resolved by troubleshooting the contamination source and eliminating it. This does not usually involve changes in methodology however, where contamination has been introduced subsequent to validation changes in the method, e.g. for a cochromatographing peak it may be necessary to use a more selective detector. Ultimately L C - M S - M S detection or where MS is not appropriate or is already the method of choice selective extraction, may be necessary. For example, in a GC NPD method for brompheniramine, plasma levels were found to be 10 times higher than the expected peaks and consistent across all samples, all common materials were evaluated. The contaminant was subsequently shown to be present in the blood collection tubes. Interestingly the flat bottomed tubes used as part of the development process were free from this contamination while the round bottom tubes had consistent and reproducible amounts of contamination. While this resolved the problem for future studies, and indeed the other samples from this same study, it did not resolve the problem for the collected samples. This was eventually resolved by differential extraction of the contaminant. The contaminant unidentified but of different apparent polarity, vis-a-vis the extractability, nevertheless cochromatographed on a GC column regardless of column polarity or temperature. This example illustrates the need to analyse study samples as quickly as possible in order to evaluate any problems References pp. 406--409

384

Chapter 13

between validation/QC samples and real/incurred samples. It also illustrates the point that attempts to identify the source of "problems" in a checklist manner are likely to be compromised by the unexpected- however, checklists while no substitute for scientific troubleshooting, do provide a good staring point. 13.3.1.1.7 Evaluating selectivity. Selectivity from potentially interfering substances (endogenous compounds, metabolites, decomposition products, and concomitant medication) should be established by the analysis of blank samples of the appropriate biological matrix obtained from at least six different sources. Each sample should be checked for interference to ensure selectivity at the LLOQ of the method for each analyte. No absolute criteria are set, except that the analyte response at the LLOQ should ideally be at least five times the response compared to the blank response. In the case of hyphenated mass spectrometry-based methods, tests for interference may be less important. However, matrix effects which may compromise the ionisation of the analyte, should be investigated to ensure that precision, selectivity and sensitivity are not compromised (see Matrix effects). 13.3.1.1.8 Unresolved contamination. A similar issue of contamination but more contentious and difficult to resolve from the analytical perspective, is that of trace amounts of drug in control groups. This is particularly an issue with toxicology studies. In clinical studies this issue rarely occurs. Historically, this has cast doubt on the veracity of the analytical method. If high concentrations are found, then this would be suggestive of misdosing, however, the concern here is with concentrations of drug at or near the LLOQ. This phenomenon has come to be recognised as a real issue to such an extent that an ABPI/EFPIA [48] meeting was held in January 2002 to discuss these issues. The reason for this observation is likely to be contamination, be it at the dosing stage, in the "atmosphere" of the "animal room" or at sampling, or indeed in the bioanalytical laboratory. Adherence to GLP implies that such issues should not arise. However it is the conduct of these practices from a scientific perspective that should be reviewed. Indeed in a GMP environment the problem of cross contamination is well recognised and a separate discipline of cleaning validation has been developed. Whether similar practices in animal dosing rooms and related environs should be implemented, remains to be discussed. From an analytical perspective it behoves the bioanalyst to at least set realistic LLOQs consistent with the observed kinetics. Nevertheless (as always), it is incumbent upon the analyst to reassure the toxicologist that these are indeed real results, i.e. there was no carry-over in the system used to process and analyse the samples. In general this should be done during validation, ensuring carryover is not an issue by the inclusion of blank plasma samples in analytical validation batches. 13.3.1.1.9 Evaluation of carry over. The most usual source of contamination is the injection system, where debris from the sample (especially if dirty) accumulates and holds back small amounts of test article which may be subsequently eluted by other plasma extracts. This problem extends to automated sample preparation systems as well as injection systems. This issue has been addressed by the ISPE (International Society of Pharmaceutical Engineers) in their technology transfer document [49], which

Chromatography in a regulated environment

385

provides a useful basis for determining/validating the performance of automated sample preparation and injection systems. They recommend that a full validation batch be run with samples having similar concentrations to the test samples. Fully processed blank samples are interspersed throughout the batch and acceptance criteria based on the worst case scenario, i.e. peak concentration followed by an LLOQ sample should have a carry over of less than 20% of the LLOQ concentration. Most importantly the probability of carry over creep taking place should be evaluated, i.e. carry over may be acceptable at the start of the batch but might not be acceptable later on. There are a multitude of ways to eliminate this problem, although not always successful, in such cases it may be necessary to inject the batch on a low to high basis. The impact of this can be monitored by running a standard curve at the start of the batch from low to high and one at the batch from high to low. Any divergence of the two curves may be attributable to carry over/adsorption problems although in the case of unstable detectors, e.g. MS, NPD or ECD it could be due to deterioration of the detector response. 13.3.1.1.10 LC MS (MS) specific issues. Only a limited number of issues have been identified as L C - M S - M S specific issues, e.g. use of stable isotopes as internal standards, matrix effects and the need for less rigorous selectivity checks and concomitant changes in sample preparation techniques, which, while not specific to L C MS-MS, have developed into a niche market, i.e. automated 96-well (and larger) plate systems as well as the tendency to nonlinearity in the calibration curves.

13.3.1.1.10.1 Nonlinear standard curves. In LC-MS (MS) there are many practical and theoretical reasons for the standard curve (concentration response function) for LC-MS (MS) methods to be non linear. Kostiainen and Bruins [5] evaluated the effect of solvent on the dynamic range of pneumatically assisted electrospray MS. In addition, adduct formation and aggregation of molecules, both a function of analyte concentration and physicochemistry of the solvents contribute to nonlinearity. The current guidance acknowledges that nonlinear calibration curves exist, i.e. "six to eight concentrations (excluding blank values) can define the standard curve. More standard concentrations may be recommended for nonlinear than for linear relationship. The simplest model that adequately describes the concentration response relationship should be used. Selection of weighting and use of a regression equation should be justified". 13.3.1.1.10.2 Matrix effects. The Guidance states that, "It may be important to consider the variability of the matrix due to the physiological nature of the sample". In the case of LC-MS-MS-based procedures, appropriate steps should be taken to ensure lack of matrix effects throughout the application of the method, especially if the nature of the matrix changes from the matrix used during method validation In terms of LC-MS-MS, matrix effects are sometimes termed ion suppression, although ion suppression could be considered to be one form of matrix effect, indeed the effect is not limited to MS detectors. In a fluorescence assay, the presence of UV absorbing compounds in the presence of a fluorescent analyte can absorb either or both of the incident or fluorescent light resulting in an attenuated signal; in addition the presence of coeluting fluorescent quenchers can similarly reduce the signal. In the field References pp. 406-409

386

Chapter 13

of gas chromatography, the presence of cholesterol or similar high concentration steroid can change the flame characteristics of an NPD or the ion transmission properties of an Electron Capture Detector (ECD). Thus, matrix effects is a generic term for any endogenous matrix component which can attenuate (usually) or accentuate (sometimes) the signal of the analyte in question. In the context of MS detectors, the process has been described in some detail together with some ways of eliminating or at least minimising it [51,52,53]. It should be recognised that it may not be possible to eliminate the effect but it should at least be reproducible from one sample (subject) to another. Whether a matrix effect exists, should be evaluated as soon as chromatographic and sample preparation conditions have been established. The classical format or most widely used format has been described by Bonfiglio et al. [54], in summary it consists of infusing the analyte into the MS at a constant rate post analytical column while make an injection of the blank plasma extract. Any reduction in signal, corresponding to the peak of a suppressing compound(s) which occurs at the retention time of the analyte is likely to cause a problem. At least six samples should be evaluated to determine the extent and magnitude of the effect. If it is regarded as acceptable, or not present, then progress to validation can be made. In cases where matrix effects cannot be eliminated the assay may be acceptable as long as sample to sample variability is acceptable (+ 15%?). However, unacceptable effects should be minimised by one of a number of techniques or a combination thereof, e.g. changing the extraction conditions, changing the chromatographic conditions, modifying the MS detection conditions, and/or moving from, e.g. an electrospray interface to heated nebuliser atmospheric pressure chemical ionisation (APCI) interface. 13.3.1.1.10.3 Internal standard and stable isotope issues. The importance of internal standards in analytical methodologies has been reviewed by a number of workers [55,56,57], as have the limitations and pitfalls of "inappropriate" internal standards. In all cases the importance of having an internal standard whose physicochemical properties are as close as possible to the analyte of interest is emphasised. It is generally considered that for LC-MS assay, this can be best achieved by the incorporation of a stable isotope(s), usually deuterium (2H) or 13C into the molecular structure of the analyte of interest. There should be sufficient atoms of the isotope to present a significant difference between the parent and product ions of the analyte and internal standard so that the likelihood of crosstalk is minimized. As in many cases the ideal is not always achieved, e.g. small differences in retention time are observed, whether they are sufficient to undergo differential response to matrix effects remains to be confirmed. In addition, the location of the stable isotope may impact on the fragmentation of the internal standard resulting in a difference in response between the analyte and internal standard as source conditions are modified [58]. 13.3.1.1.10.4 Automation. Issues in automation largely centre around analyte carry over in such complex systems and has already been discussed under carry over, and GLP issues. Using 96-well plates, it is important to ensure that the configuration of the plates is maintained in the same way so that the identity of each well is not compromised. Thus,

Chromatography in a regulated environment

387

plates should be at least labelled with some unique identifier as well as the study number and markings that ensure that the configuration is assured. Care in labelling is essential to ensure the label (if affixed) does not interfere with automated system. Visually monitoring each stage of the process is essential to ensure the system is operating as required and that the status of the plate is known at all times. MUX technology allows up to 8 columns to be used with one quadrupole detector. While this is widely used in a screening environment, it has only limited application in routine regulatory-based bioanalysis, in such cases the equivalence of performance of each column is essential. In the first instance this is based on system suitability tests and in process on QC performance. 13.3.1.1.11 Endogenous analytes. Endogenous analytes fall into two categories- those compounds that may be dosed as drugs, e.g. hormones, especially steroid hormones or where the endogenous analyte changes in concentration in response to a diseased state, and/or to therapy. In both cases there is a major issue in obtaining "blank" matrix so that a calibration curve can be constructed. These are analytes, which are endogenous to man but are dosed in high concentrations, resulting in circulating levels higher than endogenous levels. These are usually hormone supplements, many of which are steroids, e.g. testosterone and estrogenic replacement therapies. In some cases the compound dosed may be of dietary origin. There are a variety of ways in which a "blank" matrix may be obtained. The most common is to strip them, e.g. by passing through a solid phase extraction column, addition of an adsorbent, e.g. activated charcoal or extraction with an organic solvent. Unfortunately, all of these processes result in a matrix which differs from the original, as such, its comparability should be confirmed with the original. In some cases diseased states may result in loss of the analyte or interest and can be used for preparation of standards and QCs. However, this is rarely possible, as such an alternative is to use a surrogate buffer or synthetic surrogate made of purified plasma components. In such cases (as with stripped matrix), it is essential to show comparability of the "real" matrix (from subjects) and the surrogate matrix used for standards and QCs. This can be achieved by spiking a standard curve in the surrogate matrix and "real" matrix. A comparison of the two curves should be parallel but offset by the amount of endogenous analyte (Burnett [59]) If this is so, then the "surrogate" matrix can be used, otherwise the standard curve must be spiked into a "real" matrix containing endogenous analyte. It may be necessary to screen many samples of "real" matrix to find one with a low level of endogenous analyte. Whatever the concentration of endogenous analyte, it will compromise the sensitivity of the assay. 13.3.1.1.12 Chiral issues. Development of a stereoisomer drug either as a pure enantiomer or as a racemic mixture, presents their own specificity problems. Other stereoisomers and diastereoisomers are considered to have different physical properties, and as such, the isomers should be considered as separate drugs, and in general are not a major consideration of the FDA's Policy statement on the development of new References pp. 406-409

388

Chapter 13

stereoisomeric drugs issued in 1992 [35]. Essentially where a racemic mixture is dosed, it is incumbent on the developer to determine the pharmacokinetics of the separate isomers. Where a single enantiomer is dosed, the developer must show that no significant interconversion to the other enantiomer has taken place. Although the document favours the development of single enantiomers, it recognises "development of racemates may continue to be appropriate" and provides some points to consider where this is the case. As a consequence it is essential that a chiral assay be developed. A major limitation of chiral assays is that (usually) the sensitivity of one of the enantiomers is much greater than the other [60]. Thus, where a racemic mixture is dosed, it may be easier to confirm that the ratio of the analytes does not change by monitoring the ratio at specific points over a pharmacokinetic profile. A partial validation defining the response relationship over a usable range coupled with performance data should be adequate. In some cases where the single enantiomers are not available, the assay may have to be based on the constancy of the determined ratio. This should be validated over a usable range, which although not challenging the LLOQ, should be adequate for the purpose described. If it can be shown that the ratio of the isomers remains constant, i.e. no interconversion takes place or no preferential metabolism takes place, then an achiral assay should be adequate to continue the development process. Where a single enantiomer is dosed the conversion of the dosed analyte need only be evaluated in a range of individuals. If in the former case the ratio is not constant or if interconversion takes place, then a fully validated assay for both enantiomers of sufficient sensitivity to define their kinetics is essential. Historically chiral columns have been thought to lack robustness, as a consequence the use of diastereomeric derivatives which can be separated on conventional columns has been a major area of endeavouralthough subject to further technical and validation issues. The advent of L C - M S - M S coupled to normal phase columns has increased the sensitivity and reliability of chiralbased assays to an extent that they have become routine.

13.3.1.1.13 Stability. The stability is a function of the storage conditions, chemical properties of the drug and the container system and is specific to the matrix and container system - extrapolation to other matrices and container systems is not appropriate. Stability of each analyte in the relevant biological matrix should be confirmed, minimally at low and high concentrations (at least three replicates at each concentration) over the calibration range. NB: the choice of three replicates is presented in the BMV guidance- although the word minimum should not be ignored. Stability should be evaluated during: Sample collection Sample handling (processing) Long term storage Short term storage Bench top Freeze-thaw Analytical process including autosampler stability

Chromatography in a regulated environment

389

Conditions used in stability experiments should reflect situations "likely" to be encountered during actual sample handling and analysis. For short-term storage at room temperature, the storage time should be based on the expected duration that test samples are maintained at this temperature during the intended study (4 to 24 hours). Stability should also be determined after three freeze and thaw cycles and after longer-term storage in the freezer, using identical storage conditions to those of the test samples (i.e. -20~ or-70~ Long term storage should confirm stability over the longest period which test samples are stored, ideally prior to analysis. The stability of each analyte and internal standard stock solution, if appropriate, should be evaluated at room temperature for at least six hours, and if refrigerated or frozen, during a relevant storage period. Stability should be tested by comparing the instrument response of the old solution with that of a freshly prepared solution. Once stability has been confirmed and accuracy of preparation verified, standards and QCs can be prepared from the same spiking stock solution. The stability of processed samples, including residence time in the auto s ampler, should be determined for the length of the anticipated batch size and to cover reinjection of samples, if appropriate. For labile analytes, investigations may include analysis of samples from dosed subjects (incurred samples). The guidance states that concentrations of all stability samples should be compared to the mean of the back-calculated values for the samples, at the appropriate concentrations, measured on the first day of testing. The disadvantage of using this approach, instead of comparing measured values with the theoretical (nominal) spiked concentration, is that error measurements (which can be as much as _+15% for each measurement) may make interpretation difficult.

13.3.1.1.13.1 Evaluating stability. The FDA/BMV suggests that stored samples (at least in triplicate) should be compared with data from freshly prepared samples. They gave little guidance as to how this data can be compared and what are acceptable criteria. Timm et al. [61 ] and Kringle et al. [62] have proposed specific criteria as to how this may be carried o u t - however these do not appear to have gained wide acceptance. The most widely accepted approach is based on the 4 : 6 : 1 5 rule for acceptance of analytical batches. In this case six samples of the stored samples are evaluated and compared with the observed mean at time zero. NB - not the nominal value at time zero. Thus, if at least four of the replicates fall within the + 15% of the mean zero time value then the compound is regarded as being stable. A variation on this approach is to compare the mean of the stored replicates; if this is within + 15% of the mean of the zero time sample then the compound is regarded as stable. Kringle et al. [62] emphasises that there is a high probability that the compound will be regarded as stable when it is not. The Timm et al. [61] approach to some extent, responds to this criticism by comparing replicate determinations of fresh and stored samples, and determining the 90% confidence interval for there to be a true difference between the two sets of samples. However, Kringle points out that for the confidence level to be high, at least 27 samples at each point must be used. In contrast Kringle References pp. 406--409

390

Chapter 13

suggests that a trend or regression approach be used, i.e. nine replicates are analysed at each time point and the fitted regression line calculated. The lower specification limit, based on the calculated (observed) zero time value is set at 0.85, i.e.-15%. Thus, when the regression line crosses this value the sample can be regarded as unstable. This does NOT necessarily invalidate the data, e.g. i s - 1 4 % acceptable a n d - 1 6 % invalid? Interpretation of the data must be carried out in the light of the study objectives. 13.3.1.1.13.2 Stability during sample collection. This is a complex i s s u e - ideally the stability of the analyte in blood, the matrix immediately withdrawn from the subject, should be assessed. However, the assay is usually developed and validated for plasma or serum, not for whole blood. In general whole blood assays require more clean up and "validation" for use as a stability indicating assay. The theoretical way to proceed with the evaluation would be to spike the drug into whole blood and "equilibrate" so that there is a homogeneous distribution, of the drug. However, it is not easy to spike into whole blood and obtain an even distribution, the spiking process may result in localised accumulation of the drug, e.g. sticking to RBCs. In addition the metabolic activity of red blood cells (RBCs) changes with time - how fresh is the blood for spiking? Does the age of the blood influence the "binding"/distribution into RBCs? For example a "fresh" basic drug was spiked into RBCs and concentration in plasma assessed over a six hour time period at time zero. After gentle agitation the plasma levels were high at "zero" time decreasing over two hours before increasing over the next two hours, reaching "equilibrium" over the final two hours. (Fig. 13.3). Spiking blood with drug and using the plasma assay to reflect the stability of the analyte in this matrix, is therefore, likely to lead to erroneous conclusions. Does the plasma level of blood samples taken at different times post dose reflect this profile? Does the drug "in vivo" equilibrate more rapidly? How do you carry out an in vitro stability experiment in blood? One resolution would be to carry out the assay in whole b l o o d - removing any time dependent distribution relationships. 100%

-

I1) t,-0 Q.

50% -

ne

0

2

4 Time / Hours

Fig. 13.3. Drug spiked into blood and the plasma assayed at times post spiking. Profile may reflect distribution/equilibration into red cells and lysis of red cells and/or adsorption to red cells.

Chromatography in a regulated environment

391

Kringle [62] has attempted to resolve this, he recognises that when whole blood pools are spiked with known amounts of compounds, the concentration in plasma is unknown. Additionally, there may be contamination of or binding effect with the collection vessels. Thus, the "nominal" concentration at time zero cannot be assumed to be the nominal spiked amount. Hence, the time zero must be estimated. Kringle uses a regression equivalence over 24 hours to "estimate" stability in blood; however, it is unlikely that the drug once collected will remain in contact with the blood for more than an h o u r - during which time it may exhibit instability or equilibration with the RBCs. Indeed Kringle's own data suggests that over the first six hours the slope has a steeper decline than that seen between 6-24 hours. This suggests that the distribution/ equilibration process takes place over the first six hours. The general trend when evaluating short term ex-vivo stability is to evaluate the analytical matrix, e.g. plasma over six hours at room temperature. Where there may be concerns of distribution into RBC or indeed sticking to red cells, sampling containers, etc. specific experiments should be conducted to evaluate the scale and source of the problem.

13.3.1.1.13.3 Evaluating incurred sample stability. Evaluating the unknown - recognising the problems of "in vivo" metabolism has rarely been proactively evaluated, usually it is in response to problems [63]. However, Knupp presented some pragmatic approaches at the Bioanalytical methods Validation Workshop in January 2000, unfortunately they have not been formally documented. The Canadian authority's requirement to duplicate 15% of the "real" samples in a study is an attempt to confirm the reproducibility of the assay in real samples. However, this may "hide" a reproducible problem. Knapp suggested that incurred samples be exposed to conditions that would favour degradation. This can be accomplished by analysing incurred samples under the validated assay conditions and then repeating the analysis after subjecting them to mild stress, e.g. prolonged storage at room temperature and/or leaving the samples during processing at a "stressed" stage for longer than normal. Additionally, incurred samples may be exposed to "extremes" of temperature, i.e. up to 50~ or extremes of pH prior to processing through the method. Significant differences (-15%) would suggest a problem. How can incurred samples be evaluated? It is unlikely that an ethical committee, be it for animal or human studies, would give permission solely for the purposes of determining the stability of incurred samples. As a consequence such studies should (must) be carried out on samples from studies carried out for other purposes. This poses a problem that the samples will have already been analysed and a result generated. If the sample is further analysed in order to determine stability, another result will be generated- if the results differ- what should be done? One solution to this quandary is to bulk samples from more than one subject- these should be determined at 3-5 time points over the pharmacokinetic profile, minimally at the peak concentration, at 1.5 and 3 times the half life in order to fully evaluate different ratios of parent compound to metabolite(s). These samples should be evaluated as proposed by Knupp above. References pp. 406-409

392

Chapter 13

13.3.1.1.13.4 Stabilizing "labile" samples. Two types of instability exist, the first where the parent compound may be unstable to heat, light, oxygen, pH and or endogenous enzymes such as esterases and that of unstable analytes described above and only found in incurred samples. This is a major unknown and can only be evaluated when samples from subjects dosed with the drug of interest are available. The major concern in these samples is the back conversion of "unstable" metabolites to the parent compounds. These metabolites may be unstable to pH changes or enzymes within the ex vivo sample or to some aspect of sample processing, e.g. N-oxides to "parent". Since the metabolites themselves are of unknown structure or unavailable, most of the investigation must be carried out using ex vivo matrix. In GC, a technique likely to cause thermal degradation, e.g. back conversion of Noxides to parent takes place in the injection port prior to transfer onto the column. In such cases it will not be possible to distinguish the product from the parent compound present in the original extract. In L C - M S - M S systems, thermal degradation is likely to take place in the post column high temperature environment of the mass spectrometer especially when using Turbo Ion Spray. In such cases it is likely that separation of metabolites from the parent compound will have been achieved chromatographically. Where ex vivo stability may be an issue, stability experiments in plasma should be conducted as soon as a "usable" assay is available. If there is a need to stabilize the samples by adding stabilizers, flash freezing collecting in subdued light conditions, etc. this should be established as soon as possible. Dell has reviewed a range of stability issues together with relevant stabilizing processes at Bioval 2002 (Table 13.1) 13.3.1.1.13.5 Sample quality. In some instances samples arrive in the analyst's laboratory prior to the initiation of stability studies or definition of the collection conditions. Even under such circumstances the analyst should record the condition of the sample, the nature of its receipt, labelling, from where/when, was it thawed/frozen, etc. as they may have significant impact on the interpretation of the data when it is generated. This aspect is emphasised in the FDA compliance program [33] re in vivo bioequivalence studies. Prior to the initiation of a study the analyst should be involved in determining the conditions of collection, processing and storage as well as transport. In studies where stabilisation of the analytes is essential and where exact stoichiometric transfers are needed, the analyst may need to be involved in the preparation process or in the training of those involved in sample collection. The importance of ensuring that the quality of the samples cannot be overemphasised- the quality of the data is dependent upon them.

13.3.2 Application of a validated analytical method Once the method has been validated for routine use, its precision and accuracy should be monitored regularly to ensure continued performance. Matrix-based calibration curves, consisting of a minimum of six non-zero standard points (either single or replicate) and quality control (QC) samples at a minimum of three concentrations (low,

Chromatography in a regulated environment

393

TABLE 13.1 STABILISATION OF REACTIVE GROUPS IN PLASMA SAMPLES DURING STORAGE AND/OR SAMPLE PROCESSING AND PREPARATION Substance Type

Reaction

Procedure or Precaution

N-oxides

Revert to parent amines

Presence of antioxidant (e.g. ascorbic acid) to preserve amine can promote reversion Can occur in GC injector, or in MS Avoid heat, exposure to light and high pH

Thermal degradation Sensitive to light and pH Ester Glucuronides

Hydrolyse to parent drug

Acyl group migration

Avoid extremes of pH, cool the sample, denature plasma blood proteins, add esterase inhibitors. pH < 6; in MS avoid high temperatures

N-Glucuronides

Revert to parent drug

Avoid high temperatures and extremes of pH

Alcohols and phenols

Photodegradation, quinone formation

Antioxidants (but quinones produced in vivo can revert back to alcohol); avoid light; cool the sample; denature proteins, acetylate; SPE to stabilise phenols. NB: phenols auto-oxidise usually above pH 10

Lactones

Alcohol: acid adjacent produces lactose

Acid conditions lactose alkali alcohol: acid

middle and high), at least in duplicate, over the calibration range should be analysed with each batch of test samples, or a m i n i m u m of 5% of the batch should be QCs in multiples of three. The response function of the calibration curve, in terms of curve fitting, weighting and goodness of fit, should be the same as used during the validation phase. System suitability is used to ensure optimum operation of the analytical system. Samples should be analysed within their known stability period and, where appropriate, all samples from the same subject, in any one phase of the study, should ideally be analysed within a single batch. This is especially important in definitive bioequivalence studies where small variations can influence the tight statistics that are applied to the interpretation of the data from such studies.

13.3.2.1 System suitability The new B M V Guidance emphasises the need for system suitability tests (SST) without defining specifically what is required. Another FDA d o c u m e n t - Reviewer Guidance published in 1992 for F D A analytical chemistry reviewers provides an overview for bioanalytical and formulation analysis for G M P studies [64]. While the criteria have relevance to the analysis of formulations, for bioanalysis, more method specific criteria, pivotal to the successful implementation of the method, e.g. separation of the analyte from "interferents", viability of the LLOQ, should be defined. Applied SSTs range from the analysis of six processed blanks (which could be from the same or different subjects)

References pp. 406-409

394

Chapter 13

as well as unprocessed LLOQ and/or processed LLOQ samples to the analysis of a blank and a processed LLOQ. Other criteria could include such things as a retention window, separation factors from other peaks, e.g. metabolites, internal standards and identified endogenous interferants. There are a number of variants between these two extremes including the implementation of SSTs at the end of the batch to ensure continued p e r f o r m a n c e - these should be viewed with caution as a batch which passes in all other respects except for these late SSTs is a conundrum. In a regulatory environment, the production of conflicting criteria is not helpful and should be avoided.

13.3.2.2 Disposition of standards, QCs and samples in a batch Where should the standard curve be placed in a b a t c h ? - what concentrations should be used to space the points out within a curve? This has been, and will continue to be, the subject of much discussion. A well established approach is to randomly distribute calibration standards, QCs and samples throughout the batch using a random number generator. This minimises any possible bias and or drift. However, because there is a greater chance of placing samples in incorrect position on the autosampler, i.e. breaking the code can introduce errors, the "consensus" approach, is to place one curve at the beginning of the batch and one at the end. This is an approach that is consistent with Guidance suggestion that "placement of standards and QCs within a run should be designed to detect assay drift". Note however, the Guidance does not suggest how much drift is acceptable. Indeed if the QCs are within the acceptance criteria, invoking a criterion for drift acceptance may introduce a further acceptance criterion at odds with those generally used and should only be used as a diagnostic tool where batches fail. The most widely used approach is to place one curve at the start of the batch in ascending order of concentration and one curve at the end of the batch in descending order of concentration. There are many advantages to this approach if the first curve exhibits poor chromatography linearity sensitivity, then it may be advisable to abort the batch and save the samples to run when the system has been regenerated/repaired, i.e. it can act as a system suitability test. In addition it can provide information on drift in response to other dynamic affects related to chromatography. It does provide a further quandary of, should you set acceptance limits between replicates - the general trend is not to set any. Include all standards in the curve evaluation and review curve acceptance based on the guidance criteria, i.e. 75% of standards (minimum of six) should be within + 15% of the nominal except at the LLOQ where 20% is acceptable. Thus, if you use two curves with 6 points per curve, 3 of those points could fall outside the + 15% limit without rejecting the curve. This approach presents a further advantage in that if one of the LLOQ or ULOQ standards are lost, then the curve range does not need to be truncated. For laboratories which run only one calibration curve per batch, there is a good argument to add an extra standard at the bottom and top concentration so that if one of the points is unacceptable, then a full curve can be retained. However, a priori criteria should be set as to which of the two standards should be used if both are acceptable.

Chromatography in a regulated environment

395

13.3.2.3 Chromatographic acceptance

Obviously the better the chromatography, i.e. the more gaussian the peak shape, the better defined the start and stop areas of the peak, i.e. no interference in these areas and minimal background noise. Without doubt the definitive text on peak integration is that of Dyson [65], although more recently Meyers [66] has evaluated error sources in the determination of chromatographic peak size ratios. Both these articles discuss which is the appropriate choice for quantitation, i.e. peak height or area; this largely depends on whether it is a flow sensitive or a mass sensitive detector. The former detectors do not consume the analyte hence they can be linked in series, i.e. hyphenated. Increased flow rates reduces peak area and vice versa, while peak height remains unchanged. Mass sensitive detectors on the other hand consume the analyte in the response reaction; as such they cannot be linked in series except as the terminal detector. Peak shape and height change with flow, while area remains the same. However, L C - M S - M S systems exhibit mixed flow and mass sensitive characteristics, so the choice of peak height or area as the quantitative response must be made on other grounds. In a regulatory environment, where a choice has to be made, e.g. peak height or area it should be justified on scientific grounds. In the case of MS detection systems other practical considerations come into play. Thus, peak height is robust to changes in baseline and is less influenced by unresolved peaks. On the other hand peak area remains unchanged by column nonlinearities, although it is sensitive to integration start and end times. While small changes in gradient systems cause greater changes in height than area, peak area is more accurate for assymetric peaks. These generalisations may be complicated by incomplete peak resolutions, which make it difficult to integrate or define baselines. Although in LC-MS because of specific nature of the technique other peaks are rarely encountered. The impact of a range of different overlapping peaks on baseline evaluation has been discussed in some detail by Dyson and could provide the basis for a guideline on determining the need for reintegration. Generally, peak area is chosen for LC-MS (MS) systems. This may have something to do with the apparently erratic spikes seen in many LC-MS(MS) chromatographic peaks. In addition to the choice of area or height, the choice of other parameters used to define the peak are also pivotal. Perhaps two of the most important parameters which influence integration are filtering and smoothing which are closely linked and often the terms are wrongly interchanged. Thus, filtering can be seen as a process, which together with the collection rate might be seen to "change" the raw data that is recorded, while smoothing can be carried out post data collection and may be seen as data manipulation. The application of the integration parameters to appropriately smoothed data should not be viewed with suspicion except where it may adversely modify the perceived chromatography, Dyson [66] discusses these issues in detail. On the other hand it can help to better define the integration start and stop points resulting in better more reproducible baseline assignment. Both Dyson [65] and Meyers [66] have reviewed the impact of integration errors, e.g. sampling frequency, threshold and noise effects on quantitation of chromatographic peaks. Meyers [66] emphasises that these are not really "errors" but variables which are a function of the integration process. References pp. 406-409

396

Chapter 13

Dyson [65] provides an overview of parameters used to overcome the integrator's naive logic - individual laboratories could do worse than use this as a basis for building their own log of examples of "unusual" situations which can be used as a training tool. Attempts to provide definitive examples to be followed assiduously should be avoided; on one hand it will not provide a definitive set of scenarios and on the other it will provide a basis for the non-professional to critique the decision of the analyst. Above all consistency is the watch word.

13.3.2.4 Reintegration of chromatographic peaks The essence of consistent application of manual baselines to "difficult" peaks is dependent upon good training, good documentation and above all constant review to ensure maintenance of the established standard amongst a group. Ideally, actual manual assignment should be avoided. This can be done by modifying the established integration parameters and allowing the system to carry out the process. In addition a process should be devised and maintained to ensure compliance with the principle 21CFRll. Providing a scientific rationale for choice of peak area or height is an immediate requirement while ensuring the baseline assignment is accurately executed is essential. Again this can only be accomplished by training the analyst to understand the criteria for integration start and end for the specific system as well as smoothing procedures and optimising data collection points. A good test to evaluate the appropriateness of these procedures is to run a processed LLOQ as a system suitability test (SST), acceptance can be based on a minimum peak area/height and S/N. Where sample concentrations in a study may be low, more than one LLOQ SST can be run to evaluate ruggedness of the integration start and end parameters. Attention to detail in setting up these parameters can minimise the need for sample reintegration, indeed excessive numbers of reintegrations may be an indication of poor peak parameter settings and/or poor chromatography. For high sensitivity assays, where pushing the envelope of the technology is inevitable, excessive reintegration may be the price to be paid. Whatever the reason, reintegration has a cost in time and a potential for increased error. While it is possible to set basic criteria for which peaks should be reintegrated, it is not always easy to provide an all embracing SOP that covers every eventuality. It is suggested that a guideline be provided together with consistent training and awareness, which can be consolidated by peer and or supervisory review. The process commences with a review of the original baseline by the study analyst who identifies those peaks requiring reintegration, these may be peer reviewed by another trained analyst or supervisor and manually reintegrated. Following this, standard curve acceptability based on a SOP or criteria defined in the protocol are determined. If acceptable, the QCs are calculated and reviewed for compliance with the 4 : 6 : 1 5 rule [ 17]. Should there be a need to review the baselines after the standard curve and QCs have been calculated and the batch deemed have failed? Strictly speaking, No, but confirmatory reappraisal would seem sensible to ensure errors of omission have not taken place. But certainly a wholesale re-reintegration process should not be entertained. Thus, the excessive need to reintegrate Standards and QCs over and above the study samples may suggest an extraordinary will to ensure the batches pass.

Chromatography in a regulated environment

397

However, where the control plasma for QCs and standards may be "dirtier" than the pristine study samples there may be a justification for excessive reintegration. In summary, the documentation requirements outlined in the Guidance present little extra work. However, the implementation of an objective integration and reintegration process cannot be limited to a few lines in a guideline, but as always in GLP it is dependent on good training, consistent and objective application of a priori criteria in a fully documented process which is compliant with 21CFR11. 13.3.2.5 Standard curve acceptance

Matrix-based calibration/standard curves for each analyte should consist of a blank sample (matrix sample without internal standard), a zero sample (matrix sample spiked with internal standard) and six to eight non-zero standard points (concentrations), covering the entire concentration range and including the LLOQ. Throughout the guidance various numbers of standard points required to define a standard curve are quoted. In principle, additional points, (n > 6) should be included, particularly for nonlinear relationships. The simplest model that adequately defines the concentrationresponse relationship should be used and the "goodness of fit" in terms of backcalculated responses of the individual concentrations should not deviate by more than 15% from the nominal concentration (20% at the LLOQ). The selection of weighting and the use of a more complex regression model should be justified. At least four out of six non-zero standards should meet the acceptance criteria, including the LLOQ and the highest calibration standard. Excluding standards should not change the model used. If the number of standard concentrations is greater than six, 75% or a minimum of six non-zero standards should be acceptable. 13.3.2.6 Quality control acceptance criteria

Results of the matrix-based QC samples are the basis of accepting or rejecting batches. For a batch to be accepted, at least 67% (four out of six) of the QC samples should be within 15% of their respective theoretical values. This means that 33% of the QC samples may be outside the + 15% criteria, but there must be at least one QC at each concentration level within 15% of its theoretical value. The minimum number of QC samples (in multiples of three) analysed in a batch should represent 5% of the total number of unknown samples or a minimum of six, whichever is the greater. Estimations above the ULOQ or below the LLOQ are not recommended, high concentration samples should be diluted with matrix into the validated calibration range. In the case of multiple analytes, data from only one analyte failing the acceptance criteria should not preclude acceptance of data for the other analytes, which are acceptable. Where samples need to be reanalysed or reintegrated, a guideline or SOP should be in place to explain the rationale for these procedures. 13.3.2.7 Sample assay repeat criteria

In general the need for re-assay fall into two categories: those for which no valid data could be generated and those for which a valid data was generated but in the context of References pp. 406-409

398

Chapter 13

the rest of the data appears anomalous or questionable. In the former case this may include loss of sample due to a laboratory accident, contamination of the sample, misinjection, sample processing issues, etc. i.e. no data can be generated and no assignable reason can be made as to why. On the other hand data may be generated which appears to be "obviously" anomalous and for which there appears to be no analytical reason. Under such circumstances the sample should be repeated in duplicate [17]. The major problem is defining what is obvious and enshrining it in a SOE In general there are number of obvious criteria which do not rely on a priori understanding of the pharmacokinetics of the drug, e.g. measurable concentration in pre-dose samples and in samples where the drug has appeared to have been cleared from the body. Other examples are where an adjacent point may be 100% higher or 50% lower than that point. Such criteria should be reviewed in light of the study design and the dosage form and route. In addition to a SOP identifying anomalous results there should be a SOP/Guideline identifying which of the three results (original and two repeat results should be reported). This can be achieved by incorporating a variety of different decision trees or empirically reporting the median of three sets of data.

13.3.3 Post validation issues

No sooner has the method been applied to routine sample analysis than the need for "revalidation" or additional validation becomes a reality. These situations include changes in the location of the analysis, method of analysis, nature of the samples, changes in instrumentation and or materials, the presence of "new" metabolites and/or possible interferants, be they as a consequence of therapy or diseased state of the subject.

13.3.3.1 Metabolites in safety testing (MIST) While "unknown" metabolites and their stability may cause problems early in the development program, there may be a need to develop assays for new metabolites as the drug development program develops and their identity and/or importance emerges is well recognised. Baillie et al. [34] summarized the deliberations of a multidisciplinary committee sponsored by the Pharmaceutical Research and Manufacturers of America (PhRMA), in their points to consider he summarised the implications of their deliberations thus, "once it is determined that a particular metabolite should be monitored in preclinical and clinical studies, a bioanalytical assay needs to be developed and validated in order to incorporate this metabolite in the exposure assessments". They continue "In general only major human metabolites should be considered for monitoring in selected toxicological studies. Normally, conjugated metabolites are excluded from such analyses unless there is reason to believe them to be chemically reactive, e.g. some acylglucuronides). Moreover, no bioanalytical assay is needed for a metabolite that is major in the toxicology species but minor in humans. However, a metabolite that is major in laboratory animals but minor in humans may need to be

Chromatography in a regulated environment

399

identified if such identification would aid in the interpretation of preclinical safety results. If an important active metabolite circulates in humans, then it should be included as part of the validated bioanalytical assay at least in some of the safety studies in each species evaluated for toxicity and in an appropriate clinical study at the efficacious clinical dose". The bold letterings are mine and serve to emphasise the author's "Points to Consider" contention that definitive guidance is impossible. Emphasis on the imponderables is not meant as a criticism, merely an alert to the analyst that "new", major, toxicologically or pharmacologically-active metabolites may rise to significance at any time during the development process. However, they do define a major metabolite as one that accounts for 25% or more of the exposure to circulating drug-related material, with plenty of (justifiable) caveats together with a sensible decision tree: NB: This decision tree relates to Phase I metabolites and justification to measure Phase II metabolites must be a strong one [34].

Major Metabolite

I..... Structural Alert

I

[

i

~11~

Yes

Quantify

~

Yes

Quantify

No Quantification Fig. 13.4. Decisiontree - when should a metabolite be quantified in clinical studies? Based on reference34.

While the decision tree is logical, the process for ascertaining the presence and identity of major metabolites is less well defined. If significant metabolites are not identified and monitored in toxicology studies and hence exposure not evaluated, serious questions may arise with a concomitant increase in costs. Identifying metabolites at an early stage argues in favour of an early comparison of metabolites in different species, initially, an in vitro comparison and definitively comparative radiolabelled studies in the relevant species (including man). This has two major ramifications - both usually regarded to be of little consequence by the non-bioanalyst. References pp. 406-409

400

Chapter 13

9 Much data could be generated retrospectively if samples were stored appropriately and in sufficient quantities (NB: bioanalysts should develop assays for very small sample volumes) - minimising the need to repeat studies or extrapolate backwards from the current database. 9 A bioanalytical method is NEVER fully validated, additional analytes may change methodology and revalidation requirements at ANY time during the drug development process. It should be noted in the FDA, BMV guidance that full validation of the analytical method is required where a "new" metabolite is to be quantified.

13.3.3.2 Cross validation Historically, this terminology has been used for any validation process which referred back to an original validation, however this has now been defined as either partial validation or cross validation (FDA BMV [17]). In this context, cross validation is restricted to the comparison of different methods used to support the same study, e.g. L C - M S - M S vs. HPLC or to show comparability of the same method where it may be used in two or more different locations. Cross-validation consists of the analysis of both spiked quality control samples and "real" test samples, analysed both ways (by both methods or at both sites/laboratories). Partial validation refers to modifications in the methodology requiring additional validation to confirm the assay has the same characteristics after modification as the original method.

13.3.3.3 Method transfer This refers to a specific form of cross validation where the same method may be used at different sites. Globalization and rationalisation within the pharmaceutical industry has resulted in many labs carrying out the same function on disparate sites across the world. In addition, the need to dramatically increase the number of drugs to market has seen a rise in the use of outsourcing and the concomitant growth in contract research organisations (CROs). As a consequence methods are usually developed at one site and transferred to other (receiving) laboratories. No matter how well defined the analytical method is, differences in philosophy, equipment, reagents, glassware, plasticware, new sources of contamination, ensure that few methods transfer painlessly. The ISPE [49] recognised this as an issue with pharmaceutical formulation methods and developed guidelines with the FDA, with input from the EMEA on Technology Transfer, including method transfer. Thus, although the methodology may be well documented, method transfers are rarely problem free, as such the process of method transfer should be formally established and a process developed. In its simplest form, as much information should be transferred to the receiving laboratory who should evaluate the documentation and seek clarification. This is followed by an attempt to establish the methodology in their own laboratory. At this point a long time taken to establish the method should indicate a problem requiring immediate intervention by the sending laboratory's analyst at the receiving laboratory's facility. It is at this point that differences in working practice and

Chromatography in a regulated environment

401

equipment may be identified which resolves the problem (usually). If not, a definitive problem-solving process should be put into motion in order to resolve the problem. 13.3.3.4 Partial validation

Methods may be modified in order to ensure the method is performing to the same criteria it was validated against as a consequence of changes in materials/equipment. These validations will vary depending on the extent of the modification and the impact on the integrity of the validation data. Partial validation may vary from as little as a one batch intra-assay precision and accuracy determination, for example, when transferring a method between analysts, to almost a full validation, for example, when there is a change in species within a matrix eg rat plasma to mouse plasma. Thus, there is a need to partially validate where samples have been taken in a different anticoagulant or because serum has been taken instead of blood, as changes in status of the patient, i.e. diseased states and coprescribed drugs. In addition the need to carry out in vivo drugdrug interaction study will most likely require confirmation that the two drugs do not interfere with quantitation of each other. Such studies are conducted under the FDA Guidance for Industry, In vivo Drug Metabolism/Drug Interaction Studies - Study Design, Data Analysis and recommendations for dosing and labelling [67] and the EMEA and FDA guidances based on relevant ICH documents E8 [68] E7 [69], E3 [7O]. The study designs for such investigation are diverse and should be based upon the intended clinical use, as such not only the range and sensitivity of the assay may need to be revalidated but certainly the specificity should be fully evaluated. The FDA identifies the drug candidate for submission as the substrate (S) drug and the interacting drug as the interacting drug (I). It is essential for the analyst to understand the study design as well as the dose to ensure that the developed method is adequate for the intended purpose. Thus, where (I) and (S) are dosed chronically, the possibility exists for induction to take place of metabolising enzymes resulting in lower plasma levels of the parent and perhaps higher levels of one or more major metabolites. Concomitant administration of the interacting drug for one or more dosing intervals should be monitored over the dosing period. Dosing of an inhibitor (interacting) drug may result in higher plasma levels of the substrate drug and may require partial validation to extend the range of the curve. Where the intention is to determine whether the substrate drug is metabolised via a specific cytochrome P450, it is usual to dose the substrate with a drug whose metabolism is well known and well defined, i.e. is not metabolised via a multiplicity of enzymes. The FDA recommend that for CYP3A simvastatin or lovastatin be used, theophylline for CYP1A2, warfarin for CYP2C and desipramine or imipramine for CYP2D6. As such it should be part of a laboratories repertoire to have these methods or scientifically valid substitutes [71], "fully" validated and ready to run. However it should be noted that the method CANNOT be fully validated until the assay has been evaluated to take account of the possible interference by the substrate drug and or its metabolites. This may be best achieved by taking plasma from a subject or subjects (n =6?) who have been dosed with the substrate drug only and spiking the matrix with References pp. 406-409

Chapter 13

402

the respective analyte(s) of interest and determining the presence of cochromatographing peaks or matrix-based effects which may interfere with quantitation. While it is possible to determine whether the parent compounds interfere in the respective analytical methods, this is not always possible where metabolites of unknown identity may cochromatograph and hence interfere with certain chromatographic methods, e.g. HPLC-UV, E1CD, FL or GC-NPD, ECD, etc. While this form of interference is unlikely as far as MS detectors are concerned, a far more likely possibility is that metabolites may cochromatograph with the compound of interest and suppress ionisation, thus reducing or eliminating the response. This may be evaluated in the same manner as that by which matrix effects are determined. Changes in methodology as result of the developing needs of the drug development process, or as a consequence of changes in column characteristics requiring changes in the mobile phase, may require the method to be revalidated or, as defined in the guidance, partially validated. The circumstances requiting partial validation have been documented by Dadgar et al. (1995) [23]. Further insight into what may be regarded as acceptable changes, i.e. changes not requiring revalidation, have been detailed by both the USP and EuP, although there are slight differences Table 13.2 shows the limits for assay changes not requiring validation. While these documents have legal relevance for the analysis of pharmaceutical products they have little force for bioanalysis, although their practical relevance may be of value. The bioanalyst is invited to consider it as a guidance or point for discussion. 13.3.3.5 Limit assays

There are many instances where a drug is dosed topically, e.g. dermal or ocular application- including drugs designed to work on the G1 tract, where ideally it should not be absorbed. In other cases a drug may be so rapidly metabolised, i.e. first pass

TAB LE 13.2 HOW M U C H CAN A STANDARD HPLC M E T H O D BE C H A N G E D W I T H O U T THE N E E D TO REVALIDATE?

Variable

Range

Mobile Phase pH

Salts Buffer Concentration Ratio of Components Flow rate Wavelength UV/V is Column Length Column internal diameter Particle size Column Temperature

-+0.2 _+ 10%

minor components + 30% relative or _+2% absolute term max + 10% _+3 nm _+50% _+70%

Reduce by 50% -+ 20~

Published in Pharmacopeial Forum, Vol. 27 (6), Nov/Dec 2001.

Chromatography in a regulated environment

403

effect, or rapidly hydrolysed by esterases within the circulating plasma, that amounts of the parent analyte within the biofluid may be not detectable. Rather than develop a method which quantitates the parent drug over a wide dynamic range, it would appear preferable to develop an assay with a well defined limit of detection (LOD) and or lower limit of quantitation (LLOQ) in order to confirm its absence at a certain limit. This can be achieved by determining the variability of 20-30 blanks and measuring the variability at the retention time of the drug, determining the mean signal and Standard Deviation [44]. The LOD is defined as the mean signal plus 6 times the SD. This can be translated into the amount of drug with which the signal corresponds. Once this is determined, six separate sources of the matrix are spiked with the LOD amount, processed and the peak evaluated to see if it is detectable and may determine whether the peak is measurable; if so this can be used as the LOD. If not, quantifiable increments of drug can be spiked into the matrix, extracted and the peak visually evaluated until the drug can be measured.

13.4 STUDY DOCUMENTATION All validations and sample analyses should adhere to appropriate GxPs. General and specific SOPs and good record keeping are essential to support regulatory submissions. Documentation requirements for method establishment and validation should provide a detailed operational description of the analytical method, including the purity and identity of the reference standards (compound, metabolites, internal standard, if appropriate) used. In addition, a description of all validation experiments and the relevant data obtained in these studies are required including stability studies. Documentation should include legible examples of annotated chromatograms or mass spectrograms, if appropriate. Any deviations from SOPs, protocols and GLPs, if applicable and justifications for deviations. Documentation to support the application of the validated methods should again include the purity and identity of the reference standards (compound, metabolites, internal standard, if appropriate) used. Chain of custody of the samples, in terms of sample identification, collection dates and times, storage conditions prior to and after shipment and their condition and storage prior to analysis should be documented and tabulated. Summary tables of analytical batches should include, batch (run) identification, date and time of analysis, method, analyst, start and stop times, duration, significant equipment and material changes, issues or deviations from the established method. All calibration curve data, including equations used for back-calculation of results, QC sample summary and data on inter-assay accuracy and precision from calibration curves and QC samples used for accepting analytical batches, should be available. Representative complete serial chromatograms of test samples including standards and QC samples, representing 20% of subjects for pivotal bioequivalence studies are required. In other studies, 5% of randomly selected subjects in each study should be included. The selected chromatograms should be defined prior to sample analysis. References pp. 406-409

404

Chapter 13

Reasons for missing samples, repeat analysis of samples and reintegration of data should be documented. Information should include the initial and repeated result, reason for the repeat, requestor and authoriser for the reanalysis. Repeat analysis and reintegration should be undertaken using predefined SOPs. All deviations from the analysis protocol or SOPs, with reasons and justification should be documented. Perhaps one of the major additions is the need to justify reintegration of peaks, obviously the concern is that baselines might be assigned on a self-fulfilling basis. It suggests that a SOP or guideline for sample data reintegration be established, which should include the reason for reintegration and how the reintegration is performed. Although the need for a priori integration criteria is not in the guidance, it would appear logical to establish a priori criteria that are consistent with the reintegration criteria. The BMV Guidance details the reintegration requirements on page 15 of the BMV [17], i.e. "an SOP or guideline for sample data reintegration should be established". This SOP or guideline should explain the reasons for integration and how the reintegration is performed. The rationale for the reintegration should be clearly described and documented. Original reintegration data should be reported, while the requirements for documenting reintegrated data on page 18 of the BMV [ 17] states that "documentation should include the initial and repeat integration results, the method used for reintegration, the reported result, assay run identification, the reason for reintegration, the requestor of the reintegration and the manager authorising reintegration. Reintegration of a clinical or preclinical sample should be performed only under a predefined SOP (or guideline").

13.5 STATISTICAL CONSIDERATIONS 13.5.1 Rationale behind the consensus statistics

Statistics in this context is used in its loosest sense to mean, where do the specifications come from and what are the implications in applying them. The Guidance document is largely based on the consensus document. As such the numbers are those with which the majority of the industry are happy and importantly with which the regulators are comfortable, as such they are empirically-based rather than mathematically-based. The BMV, Guidance for Industry [17] was established primarily with bioavailability and bioequivalence studies in Man in mind. Bioequivalence is an important part of the quality control of product equivalence [72], as such specifications are set for both the statistical evaluation of data [73] and for the bioanalytical methods used to determine the circulating moieties. Study designs are designed to test statistical equivalence. Any latitude in setting such specifications would require strong justification. However there is a case to be made for modifying the specifications when applied to studies other than product-based bioequivalence studies. Humbert et al. [24] concluded that data with poor precision but good accuracy, i.e. CV > 20% can be used to determine parameters associated with the terminal phase such as t l/2~3 and AUC. However this approach should be used only when it is not possible to increase the sensitivity of the bioanalytical assay.

Chromatography in a regulated environment

405

Because the BMV criteria are consensus-based there are likely to be statistical inconsistencies. A range of authors have critiqued the consensus criteria on statistical grounds. Most particularly Kringle [28] has shown that while the assay may meet the validation criteria when used in routine analysis, batches may not meet the ( 4 : 6 : 2 0 ) rule for batch acceptance criteria. Note that this critique is based on the original Crystal City meeting of 1989 - in the latest consensus document the rule was changed to 4 : 6 : 15, even tighter criteria which have been subsequently incorporated into the Guidance document. Thus, Kringle showed that for an assay with accuracy of 5% and precision 10%, then 2% batches (using the 4 : 6 : 2 0 rule) would be rejected even though they meet the validation criteria. At the extreme acceptability of an assay with 15% accuracy and 15% precision there will be a false rejection rate of 50%. Thus, by tightening the 4 : 6 : 2 0 rule to 4 : 6 : 1 5 rule rejection of data which meet the validation criteria is likely to increase - is this bad? The 4:6:15 criteria are specifications designed to maximise the consistency of routine analytical data. Rejection of "good" data presupposes that the validation criteria of 15% precision and accuracy are appropriate. One interpretation is that the validation acceptance criteria are not tight enough. Indeed, most analysts pragmatically set acceptability of a method at 10% for precision and accuracy. This is a major reason why there are fewer rejected batches than Kringle would suggest. Others have taken the opposite perspective (Karnes [26] and Boulanger [74]), i.e. that the validation criteria are correct, i.e. 15% precision and accuracy are acceptable. On that basis they have proposed the Total error criterion be applied to the acceptance criteria. This approach effectively sums the random error (precision + 15%) and the accuracy, in this instance (bias + 15 %) to give a total error of 30%. Thus, the acceptance criteria would be + 30% of the nominal. The inherent objection to the use of 30% (discussed at the 2001 consensus meeting) that these limits are too wide is probably appropriate. Generally, most assays have good accuracy, thus assays with high precision can still lead to minimal rejected batches. It is only where precision and accuracy approach the validation limits of + 15% accuracy, and precision that there is a high percentage of rejected batches.

13.5.2 Interbatch and intrabatch precision Guidance document recommends that intrabatch precision (repeatability?) and the interbatch precision should be + 15%. Horowitz [75] showed some time ago that in general the intrabatch precision data should be 20-30% lower than for interbatch precision, although, his data were based on comparing intermediate precision with reproducibility data, nevertheless a similar argument applies for the comparison between repeatability and intermediate precision. Other acceptance criteria based on the confidence interval approach have been proposed (Karnes [26]) and are deemed acceptable in the guidance. However, the complexity of application has meant that few laboratories have used this approach, indeed the universality and transparency of 4 : 6 : 1 5 acceptance criteria means that it is now incorporated into a number of bioanalytical data based systems, e.g. "WATSON" [76] which allow for automated decision making. References pp. 406-409

406

Chapter 13

After the publication of the first Consensus document, a number of academic and major pharmaceutical companies developed validation "packages "which subjected the derived validation data to statistical evaluation, this process is only likely to gain acceptance where it is available in a readily validatable commercial package which has some flexibility to respond to changing regulatory requirements.

13.5.3 Standard curves Under method development for a standard curve to be acceptable, "at least four out of six non-zero standards should meet the acceptance criteria including the LLOQ and the calibration standard at the highest concentration. NB: acceptable calibration points must be within _+15% of nominal acceptance at LLOQ where it must be within + 20% of nominal - while for routine drug analysis a standard curve is acceptable if 75% or a minimum of six standards including the ULOQ should + 15% of the nominal except at the LLOQ when it should be + 20% of nominal. Certainly for a linear function, 4 out of 6 should be sufficient to define the curve (y = m x + c) while the criteria for routine drug analysis are applicable to nonlinear curve functions.

13.6 THE FUTURE 13.6.1 Ethical implications Pharmacokinetic and toxicokinetic data are integral parts of the drug development decision-making process as well as being an essential part of the of any regulatory submission. Inevitably there is a cost, both financial and in the use of subjects be they laboratory animal or man. Plasma/serum/blood form the basic working materials of the bioanalyst. Their harvesting involves either an invasive procedure or the sacrifice of animals. Each invasive procedure has an element of risk. There are therefore financial and ethical costs, as such it is incumbent upon the analyst to be aware of the changing regulatory and legal issues that surround this hard-won material and to understand it should not be treated as some disposable reagent such as phosphate buffers, etc. The minimal use of physiological materials has both a sound scientific and ethical basis. The more data that can be generated from "less" means fewer animal and humans need be put at risk. The amounts of plasma removed from man is an integral part of the ethical issues reviewed by IRBs for clinical studies and subject to international guidelines set by WHO, while for animal studies the U.K. Home Office set guidelines for animal experiments - a major concern is to minimise any stress to the animals and includes such things as frequency of blood sampling, the method of sampling, the volumes of blood that can be removed from a different species over defined time frames, as well as regulating both the procedures and those that are licensed to carry out those procedures. The need for higher sensitivity and lower sample volumes is a continuing theme.

Chromatography in a regulated environment

407

13.6.2 Instrumentation quantification and validation The GLP requirements for instrument validation and qualification are limited to the need to show they are adequate for the purpose intended- interpretation is open to the user. The current status which has been discussed in Section 13.2.3 of this chapter, which suggests that processes are tending to those outlined in GMP documentation. However, there is strong consensus within the industry that for GLP studies at least, the current process may be excessive, driven by third party interests. "There apparently exists a chasm on the purpose and process of instruments validation between instrument manufacturers and/specialists on the one hand and instrument users on the other". As a consequence the AAPS are holding a joint meeting in early March 2003 cosponsored with FIP (International Pharmaceutical Federation) and ISPE, to discuss analytical instrument validation with the intention of "defining essential parameters and developing procedures for analytical instruments validation".

13.6.3 Biomarkers This is a generic term applied to endogenous biochemicals, be they small molecules or large macromolecules (biopolymers) whose circulating concentrations change in response to disease and may be modified by prescribed therapy. Validation of biomarker assays is discussed in the macromolecules report [31] of the March 2000 Conference on Macromolecules in Washington. It is generally considered that where biomarkers are used in PK/PD modelling [77] they should be validated to the same extent as those moieties used to determine PK parameters. A major issue in validating methodologies for biomarkers is the difficulty in obtaining "analyte" free biological matrix, a problem shared with the quantitation of endogenous analytes (Section 13.3.1.1.11) such as testosterone, hormone replacement therapy (HRT) where estrone and estradiol are measured. Proposals for the development of guidelines for the validation of biomarker assays are currently under consideration, although they are likely to concentrate on macromolecule biomarkers as such the application and validation of ligand binding assays is a major consideration.

13.7 C O N C L U S I O N The latest Guidance has many similarities to the published proceedings from the Consensus meetings held during 1990 and 2000. However, significant changes, are now evident in this document with respect to the minimum numbers of calibration points which define acceptable calibration curves, which together with tighter QC batch acceptance criteria may result in more analytical batches failing, requiring more reanalysis, impacting on the timings and cost of studies. In addition, documentation requirements, as defined in the guidance will have a major impact and may require significant changes in working practices and current SOPs. References pp. 406-409

Chapter 13

408

13.7.1 Regulatory changes Regulatory Guidances are not a substitute for good science, they do however provide a framework that should initiate the questions why and how. Using them only as checklist will inevitably mean obvious errors will be missed. Good Regulatory Guidances should have a timeless quality, which can be applied to as yet unknown technologies just as easily as to the current chromatographic methods. In addition, new guidelines will be developed where bioanalytical methodologies may play a pivotal r o l e - as such their impact on bioanalytical methods validation should be considered. Rather than reinventing Guidance documents, reinterpretation of guidances should be the focus of meetings between the regulators and the industry. Evolution rather than revolution is of benefit to both industry and regulators. Technologies change, although over the last 30 years HPLC has been the dominant force - the advent of MS-MS detectors over the last 10 years has reinforced its position. Is the loss of the analytical column near [78]? Unlikely, selectivity is as important as sensitivity and inextricably linked. Is throughput a major consideration? although there are probably sufficient machines available to service the industry. The need to rapidly turn round studies is still an imperative. Hence the development of automated sample processing systems is likely to continue. Will there be sufficient expertise available in the future? Possibly not! There is continuing growth in the area of Bioanalysis, however like all science areas the influx of newcomers is stagnating - this could be the major driver of new technologies to automate the development, validation and routine operation of bioanalytical methods although there is unlikely to be a major need to revise the validation guidelines - in the near future.

Postscript February 2003 the FDA withdrew their draft Guidance document on 21CFR11. Does this mean that 21CFR11 itself is dead? No, the FDA produced a new DRAFT Guidance "Part 11, Electronic Records, Electronic Signature - Scope and Application"- it states "while this re-examination of Part 11 is underway, we will narrowly interpret the scope of Part 11". It also explains that "we intend to exercise enforcement discretion with respect to certain Part 11 requirements. We will not normally take regulatory action to enforce compliance with the validation, audit trail, record retention, and record copying requirements of Part 11 as explained in this guidance. However, records must still be maintained or submitted in accordance with the underlying predicate rules".

13.8 REFERENCES 1 2

J. Chamberlain, The analysis of Drugs in Biological Fluids, 2nd Edition, CRC Press, Boca Raton, 1995. R.E Venn (Ed.), Principles and Practice of Bioanalysis, Taylor and Francis, London, 2000.

Chromatography in a regulated environment 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

23 24 25

26 27 28

409

E. Reid (Ed.), Assays of Drugs and Other Trace Compounds in Biological Fluids, Wol. 5, North-Holland Publishing Company Amsterdam- New York, Oxford, 1976. E. Reid (Ed.), Blood Drugs and Other Analytical Challenges, Vol 7, Ellis Horwood Ltd., Sussex, England, 1978. E. Reid (Ed.), Trace-Organic Sampling Handling, Vol. 10, Plenum Press, New York and London. E. Reid and J.P. Leppard (Eds.), Drug Metabolite Isolation and Determination, Vol. 12, Plenum Press, New York and London, 1982. E. Reid and I.D. Wilson (Eds.), Drug Determination in Therapeutic and Forensic Contexts, Vol. 14, Methodological Surveys in Biochemicals and Analysis. Plenum Press, New York and London, 1984. E. Reid, B. Scales and I.D. Wilson (Eds.), Bioactive Analytes, Including CNS Drugs, Peptides and Enantiomers, Vol 16, Royal Society of Chemistry, Cambridge, England. E. Reid, J.D. Robinson and I.D. Wilson (Eds.), Bioanalysis of Drugs and Metabolites, Especially AntiInflammatory and Cardiovascular, Vol. 18, Royal Society of Chemistry, Cambridge, England, 1987. E. Reid and I.D. Wilson (Eds.), Analysis for Drugs and Metabolites Including Anti-infective Agents, Vol 20, Royal Society of Chemistry, Cambridge, England, 1990. E. Reid and I.D. Wilson (Eds.), Bioanalytical Approaches for Drugs Including Anti-asthmatics and Metabolites, Vol. 22, Royal Society of Chemistry, Cambridge, England. E. Reid, H.M. Hill and I.D. Wilson (Eds.), Biofluid and Tissue Analysis for Drugs, Including Hypolipidaemics, Vol. 23, Royal Society of Chemistry, Cambridge, England, 1994. E. Reid, H.M. Hill and I.D. Wilson (Eds.), Biofluid Assay for Peptide Related and Other Drugs, Vol. 23, The Royal Society of Chemistry, Cambridge, England, 1996. E. Reid, H.M. Hill and I.D. Wilson (Eds.), Drug Development Assay Approaches Including Molecular Imprinting and Biomarkers, Vol. 25, The Royal Society of Chemistry, Cambridge, England, 1998. E. Reid, H.M. Hill and I.D. Wilson (Eds.), Drug Level Measurement in the Mass Spectrometry Era, Vol 52, Chromatographia suppl., 2000. E. Reid, H.M. Hill and I.D. Wilson (Eds.), Sensitive Bioanalysis in Anti cancer and other Drug Areas, Vol. 55, Chromatographia Suppl., 2002. FDA Guidance. Guidance for Industry, Bioanalytical Method Validation, Food and Drug Administration, Centre for Drug Evaluation and Research (CDER) May, 2001. Science at FDA: The Key to Making the Right Decision. www.fda.gov/fdac/features/ 2000/200_sci.html V.P. Shah, Analytical Methods used in Bioavailability Studies: A Regulatory View Point. Clin. Res. Practices and Drug Reg. Affairs, 5(1) (1987) 51-60. L.J. Phillips, J. Alexander and H.M. Hill, Methodological Surveys in Biochemical analysis, Vol. 20 (1990) pp. 23-36. G.S. Land and R.D. McDowal, Methodological Surveys Biochemical Analysis, Vol. 20 (1990) pp. 49-56. V.P. Shah, K.K. Midha, S. Digh, I.J. McGilveray, J.P. Skelly, A. Yacobi, T. Laylott, C.T. Viswanathan, C.E. Cook, R.D. McDowall, K.A. Pittman and S. Spector, Analytical Methods Validation: Bioavailability, Bioequivalence and Pharmacokinetic Studies, Pharm. Res., 9(4) (1992) 588-592,. D. Dadgar, P.E. Burnett, G.H. Choc, K. Gallicano and J.W. Hooper, J. Pharm. Biomed Anal., 13 (1995) 89-97. A.R. Buick, M.V. Doig, S.C. Jeal, G.S. Land and R.D. McDowall, Method Validation in the Bioanalytical Laboratory, J. Pharm. Biomed. Anal., 8 (1990) 629-637. R.J.N. Tanner, Phamacokinetic and Toxicokinetic Method Validation, in: A.C. Cartwright and B.R. Matthews (Eds.), International Pharmaceutical Product Registration Aspects of Quality Safety and Efficacy, Ellis Horwood, London (1994). H.T. Karnes, G. Shiu and V.P. Shah, Validation of Bioanalytical Methods, Pharm. Res., 8(4) (1991) 421-426. C. Hartmann, D.L. Massart and R.D. McDowall, J. Pharm. Biomed. Anal., 12 (1994) 1337-1343. R.O. Kringle, Pharm. Res., 11 (1994) 556-550.

410 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

Chapter 13 Guidance for Industry, Bioanalytical Method Validation for Human Studies. Food and Drug Administration, (CDER) Draft Guidance, December 1998. V.E Shah, K.K. Midha, J.W. Findlay, H.M. Hill, J.D. Hulse, I.J. McGilveray, G. McKay, K.J. Miller, R.N. Patnaik, M.L. Powell, A. Tonelli, C.T. Viswanathan and A. Yacobi, Pharm. Res.,17(12) (2000) 551-1557. K.J. Miller, R.R. Bowsher, A. Celniker, J. Gibbons, S. Gupta, J.W. Lee, S.J. Swanson, W.C. Smith and R.S. Weiner, Pharm. Res., 18(9) (2001) 1373-1383. 21 CFR 11, Electronic Records and Electronic Signatures, Final Rule FDA, 62, 13430-13466, Washington D.C. 1997. In vivo Bioequivalence, Chapter 48 BioResearch Monitoring Compliance program Guidance Manual. FDA, CDER, Washington. www.fda.gov/ora/compliance_ref/bimo/7348.col 1999. T.A.Baillie, M.N. Cayen, H. Fouda, R.J. Gerson, J.D. Green, S.J. Grossman, L.J. Klunk, B. LeBlanc, D.G. Perkins and L.A. Shipley, Drug Metabolites in Safety Testing, Toxicol. Appl. Pharmacol., 182 (2002) 188-196. FDA's Policy Statement for the Development of New Stereoismeric Drugs. Publication Date May 1, 1992. www.fda.gov/cder/guidance/stereo.htm. 21 CFR 320; Bioavailability and Bioequivalence Requirements (Drugs for Human Use). FDA, Washington, D.C. CPMP Note for Guidance on the Investigation of Bioavailability and Bioequivalence CPMP/EWP/ QWP/140/98 2001. Toxicokinetics: A Guidance for Assessing Systematic Exposure in Toxicology tudies, ICH, Topic 3A CPMP/ICH/384/95. The European Agency for the Evaluation of Medicinal Products, 1995. Good Clinical Laboratory Practice (GCLP) A Quality System for Clinical Laboratories which analyse samples from clinical trials. Draft 5 July 2002. Produced by a working party of the Clinical Committee of the British Association of Research Quality Assurance (BARQA). 2002. Orange Book. Rules and guidance for Pharmaceutical Manufacturers and Distributors Published by TSO (The Stationery Office), Sixth Edition 2002. NB Not to be confused with the FDA Orange Book. M. Freeman, M. Long, D. Morrison and R.E Munden, Position Paper on the Qualification of Analytical Equipment, Pharmaceutical Technology Europe. November 1995. Publication No. 0118. J. Burrows, Personal Communication Bioval 2002 Royal Pharmaceutical Society, Lambeth, London. A.J. Pateman, in: H.H. Blume and K.K. Midha (Eds.), Bio-International, Bioavailability, Bioequivalence and Pharmacokinetic Studies (pp. 399-403), Medapharm, Stuttgart, Germany. H. Kaiser, Two Papers on the Limit of Detection of a Complete Analytical Procedure (Translated by A.C. Menzies, 1968). Published by Adam Hilger Ltd., London. D. Dell, Views on Method Validation, in: Methodological Surveys in Biochemistry and Analysis (Vol. 20, pp. 9-22, 1990). L.J. Phillips, J. Alexander and H.M. Hill, Quantitative Characterisation of Analytical Methods, in: Methodological Surveys in Biochemistry and Analysis (Vol. 20, pp. 23-36, 1990). J. Gabrielsson, Bioval 2002. Royal Pharmaceutical Society, Lambeth, London. ABPI/EFPIA meeting on the Detection of Drug in Control Samples. January 2002. Technology Transfer Guide, Draft for Industry Review November 2001. An ISPE Technical Document. www.ispe.com R. Kostiainen and A.E Bruins, Rapid Comm. in Mass Spectrom., 10 (1996) 1393-1399. D. Thomas, S.D. Clarke, H.M. Hill, T.A.G. Noctor and J.N. Robson, in: E. Reid., H.M. Hill and I.D. Wilson (Eds.), Biofluid Assay for Peptide Related and other drugs (Vol. 24, pp. 230-238, 1996). B.K. Matuszewski, M.L. Constanzer and C.M. Chavez-Eng, Anal. Chem., 70, (1998), 882 L. Fu, Ej. Woolf and B.K. Matuszewski, J. Pharm. B iomed, Anal., 18 (1998) 347. R. Bonfiglio et al., Rapid Commun. Mass Spectrom., 13 (1999) 1175-1185. I.D. Wilson, in: E. Reid and I.D. Wilson (Eds.), Analytes for Drugs and Metabolites Including Antiinfective Agents (Vol. 20, pp. 79-82, 1990). S.H. Curry and R. Whelpton, in: E. Reid (Ed.), Blood Drugs and Other Analytical Challenges (Vol 7, pp. 29-41, 1978). E Haefelfinger, J. Chromatog., 218 (1981) 73-81. -

Chromatography in a regulated environment 58 59 60 61 62 63 64 65 66 67

68 69 70 71 72

73

74 75 76 77 78

411

G. Mackay, November 13; AAPS Annual Meeting and Exposition, Toronto, Canada, 2002. J.E.C. Burnett, I. Thompson and H.M. Hill, in: E. Reid, H.M. Hill and I.D. Wilson (Eds.), Biofluid Assay for Peptide Related and other drugs (Vol. 24. pp. 333, 1996). J. Cazes (Ed.), Optical Resolution by Liquid Chromatography, J. Liq. Chromatog., 9(2&3) (1986). U. Timm, M. Wall and D. Dell, J. Pharm. Sci., 74(9) (1985). R.O. Kringle, D. Hoffman, J. Newton and R. Burton, Drug Inform. J., 35 (2001) 1261-1270. H.M. Hill, I. Smith, I. Fairbrother and K. Tennant, in: E. Reid, H.M. Hill and I.D. Wilson (Eds.), Biofluid and and Tissue Analysis for Drugs Including Hyperlipidemics (Vol. 23, pp. 359-364, 1994). Reviewers Guidance, Validation of Chromatographic Methods, (CDER) FDA, Washington, D.C., 1993. N. Dyson, Chromatographic Integration Methods (Series Ed. R.M. Smith), The Royal Society of Chemistry, Cambridge, U.K. RSC Chromatography Monographic (1990, reprinted 1996). V.R.Meyers, Error Sources in the Determination of Chromatographic Peak Ratios, in: ER. Brown, and E. Grushka (Eds.), Advances in Chromatography (p. 383), Marcel Dekker, New York, 1995. Guidance for Industry - In Vivo Drug Metabolism/Drug Interaction Studies, Study Design, Data Analysis and Recommendation for Dosing and Labelling, November 1999. CDER, FDA fda.gov/cder/ guidance/2635fnl.htm ICH Topic E8 General Consideration for Clinical Studies (December 1997). ICH Topic E7 Studies in Support of Special Populations: Generiatrics (March, 1994) (CPMP/ICH/ 379/95). ICH Topic E3 Structure and Content of Clinical Study Reports (July 1996) (CPMP/ICH/137/95). ER.Frye, G.R. Matzke, A. Adedoyin, J.A. Porter and R.A. Branch, Clin. Pharmacol. and Therapeutics (October 1997) pp. 365-376. P.G.Welling, EL.S. Tse amd S.V. Digh (Eds.), Pharmaceutical Bioequivalence (Vol. 48), of the series Drugs and the Pharmaceutical Sciences. Publisher Marcel Dekker, Inc., New York, Basel, Hong Kong, 1991. M-L. Chen, V. Shah, R. Patnaik, W. Adams, A. Hussain, D. Conner, M. Mehta, H. Milinowski, J. Lazor, S-M. Huang, D. Hare, L. Lesko, D. Spoon and R. Williams, Bioavailability and Bioequivalence: An FDA Regulatory Overview, Pharm. Res., 18(12) (2001) 1645. B. Boulanger, Bioval 2002, March 1, 2, 2002, Royal Pharmaceutical Society, London. W.Horowitz, L.R. Kamps and K.W.J. Boyer, Assoc. of Anal. Chem., 63 (1980) 1344-1354. "WATSON" DMLIMS produced by Innaphase, Pittsburg, USA. A. Yacobi, J.E Skelly, V.E Shah and L.Z. Benet (Eds.), Integration of Pharmacokinetics, Pharmacodynamics and Toxicokinetics in Rational drug Development, Plenum, New York, 1993. J.A. Henion, Nov 13, Maintaining Selectivity in High Throughput LC-MS-MS Analysis. AAPS Annual Meeting and Exposition, Toronto, Ontario, Canada. Nov 10-14, 2002.

This Page Intentionally Left Blank

Subject index

Absorption, distribution, metabolism and excretion (ADME) studies, 331 Accelerator mass spectrometry, 331 ACE inhibitors, 215 Acenocoumarol, 160 Acetaminophen, 309 Acetylcodeine, 245 Acetyldigitoxin, 262 Achiral, 129 Achiral-chiral column switching techniques, 133 oL~-l-Acid glycoprotein (AGP), 139 Aclarubicin, 254 Acute poisoning, 235 [3-1-O-Acyl glucuronides, 316 [3-4-O-Acyl glucuronide of 2-fluorbenzoic acid, 320 Acyl glucuronides, 317 Acyl migration, 316 Acyl migration kinetics, 317 S-adenosyl-L-homcysteine (SAH), 226 [3-Adrenergic agonists, 147 [3-Adrenergic blockers, 150 Alachlor, 125 Albendazole, 31,157 Albuterol, 149, 236, 237 Alcohol deterrent drugs, 153 Alcohols, 392 Aldicarb, 266 Alpha-chymotrypsin, 226 Alprazolam, 15, 238 Alprenolol, 153 Althiazide, 194, 195, 196, 197, 198, 205, 208 Amanita mushrooms, 262 oL-Amantine, 262 [3-Amantine, 262 Amiloride, 194 Amines in urine, 34 Amino acids, 30, 153

2-Amino-3-bromo5-trifluormethylphenylglucuronide, 312 2-Amino-3-bromo5-trifluoromethylphenylsulfate, 312, 315 3-Amino-2-(2-fluorophenoxy)pyridine, 316 2-Amino-4,6-dinitrotoluene, 199 7-Aminoflunitrazepam, 246, 253 Aminoglutethamide, 169 4-aminophenol, 311 4-aminopyridine, 51 Amisulpride, 171 Amitriptyline, 80, 201,202, 204, 238, 239 Amlodipine, 166 Amperozide, in plasma, 34 Amphetamines, 20, 23, 31, 80, 175,243, 245, 246, 248 Anaesthetic drugs (intravenous), 157 Analgesic drugs (narcotics), 155 Analgesic drugs (non-narcotic), 156 Analytical toxicology, 235 Ancistrocladus griffithii, 306 Ancistrocladus likoko, 305 Angiotensin converting enzyme, 215 Anhydromethylecgoine, 242 Anion-exchange solid phase extraction, 84 Anorexic drugs, 157 Anthelmintic agents, 157 Anthracycline antibiotics, 254 Antiarrhythmic agents, 158 Antibacterial drugs, 160 Antibody-column, 5 Anticoagulants, 160 Anticonvulsants, 30, 161 Antidepressants, 31, 161,253 Antidepressants (tricyclic), 20 Antidiabetics, 257 Antiemetics, 164 Antiepileptics, 31 Antifungals, 164

414 Antihistamine drugs, 266 Antihistamines, 164 Antihyperlipopoteinemics, 165 Antihypertensives, 166 Antiinflammatory drugs, 166 Antiischaemic drugs, 169 Antineoplastics, 169, 254 Antiparkinsonian agents, 169 Anti-phosphopeptide antibodies, 365 Antipsychotic agents, 171 Antipsychotics, 253 Antiulcerative drugs, 171 Antivirals, 171 Anxiolytics, 172 Aptamine, 306 Aristeromycin, 306 Arsenic compounds, 366 Arsenic trioxide, 366 Arsenic trioxyde, 264 Arsenical pesticides, 366 Arsenobetaine, 366 Arsenocholine, 366 Arsenosugars, 366 Ascomycin, 254 Aspartame, 323 Asterias rubens, 306 Atomic absorption spectrophotometry, 32 Atomic absortion spectroscopy (AAS), 358 Atracurium besylate, 257 Atracurium besylate (2,2'-(3,11-dioxo4,10-dioxatrideca-methylene)bisphosphonate-(2-methyl1,2,3,4-tetrahydropapverinium benzenesulfonate), 302 Atrazine, 50, 51, 55, 125, 339 Atrazine-MIE 66 Atropine, 260 Automated method development, 188 Automated sequential trace enrichment of dialysates (ASTED), 29 Automation, 386 Azadirachtin, 304 Azaspiracid, 262 Azidothymidine, 30 Bacterial toxins, 261 Bambuterol, 34 Bamethan, 237 Band broadening processes, 92 Barbiturates, 30 BARQA, 378 Beclamethasone, 260 Bendroflumethiazide, 194, 195 Benidipine, 174

Chapter Benzimidazoles, 261,266 Benzo[a]pyrene, 338 Benzodiazepine derivatives, 238 Benzodiazepines, 5, 20, 30, 145, 155, 251, 263 Benzoin, 139 Benzoylecgonine, 242, 248, 265 Benzthiazide, 194, 195, 196, 205, 208 Benzylamphetamine, 246 [3-Blockers, 20, 23,276, 277 [3-Blockers in human serum and urine, 9 Betaxolol, 147 Bioanalysis in drug discovery, 271 Bioanalytical Methods Validation (BMV), 374 Bioanlytical Methods Validation for Human Studies, 374 Bioavailability, 377 Biochemical markers, 172 Biochromatography, 222 Bioequivalence, 377 Biomarkers, 407 Biomedical AMS, 337 2-(S)-[3,5-bisphosphonate(trifluoromethyl)benzyloxyl]-4- [3-(5-oxo- 1H,4H1,2,4-triazolo)methyl]-3-(S)phenylmorpholine, 168 Blake-Kozeny equation, 104 Bovine insulin, 257 Bradykinin, 363 British Association for Quality Assurance, 378 Bromazepam, 15 Brombuterol, 51, 64 Brombuterol + analogues, 20 Bromine-detected ICPMS, 363 Bromoaniline, 261 4-bromoaniline, 314, 363 Bromoclenbuterol-MIE 64 Bromoisovalerylurea, 178 2-Bromo-4-trifluoromethylaniline, 307, 312, 313,315,363 2-Bromo4-trifluoromethylphenylhydroxylamineN-glucuronide, 312 Bufuralol, and its 1'-oxidised metabolites, 166 Bumetanide, 193, 194, 205 Bupivacaine, 51, 64 Buprenorphine, 242, 246 Buprenorphine glucuronide, 246, 247 Buprenorphine-d3,247 Burgener nebulisers, 356 Burke-Plummer equation, 104 Busulfan, 255 Caffeine, 155, 196, 279, 323

Subject index Caffeine-MIE 68 Calcium channel blockers, 173 Candida cylindracea, 227 Candida rugosa, 227 Cannabinoids, 249 Canrenoic acid, 193, 194, 196, 205 Capillary electrochromatography, 46 Capillary electrophoresis, 46 Capillary electrophoresis (CE) coupled to NMR, 322 Capillary flow chromatography, 126 Capillary isotachophoresis (cITP), 323 Carbamates, 261,266 Carbamazepine, 15, 161 Carbaxazepine, 161 Carbofuran, 266 Carboplatin, 255 Cardiac glycosides, 255, 258 Cardiac glycosides from foxglove, 257 Carnitine, 20, 24 R-(-)-Carnitine, 165 [3-Carotene, 338, 339 Carry over, 384 Catechin, 196 Catechin gallate, 196 Catechins, 196 Cation-exchange solid phase extraction, 82 CEC-NMR, 323 CE-ICP-MS, 362 Cellobiohydrolase, 140 Cellulose-based membrane, 29 Ceruloplasmin, 368 Cetirizine, 165 Cevadine, 262 21 CFR 320, 375 21 CFR 320 - Bioavailability and Bioequivalence Requirements, 376 21 CFR Part II, 376 Chemical sensing, 46 Chiral- capillary electrophoresis (CE), 136 Chiral- gas liquid chromatography (GLC), 136 Chiral- supercritical fluid chromatography (SFC), 136 Chiral bioanalysis, 129 Chiral chromatography, 129 Chiral HPLC-NMR, 302 Chiral HPLC in bioanalysis, applications, 147 Chiral interactions - dipole-dipole, 131 Chiral interactions - hydrogen bonding, 131 Chiral interactions - hydrophobic/inclusion interactions, 131 Chiral interactions- ion-dipole interactions, 131

415 Chiral interactions - ionic, 131 Chiral interactions- steric interactions, 131 Chiral interactions - w-~r interaction, 131 Chiral issues, 387 Chiral metabolites, 129 Chiral recognition, 130 Chiral stationary phases (CSPs), 130, 137 Chiral-AGE 130 Chlorbenzenes, 126 Chlorfenvinphos, 125 Chlorinated phenoxy acids, 66 Chlornitrobenzenes, 126 3-(4-chlorobenzyl)chroman-2-one, 171 Chlorotriazines, 66 2-chloro-4-trifluoromethylaniline, 312 p-Chlorowarfarin, 135, 261 Chlorpheniramine, 164, 276 Chlorpromazine, 83 Chlorpropamide, 257, 259 Chlorpyrifos, 125 Chlorsoxazone, 85 Chlorthalidone, 194, 198, 205, 209 Chlortoluron, 125 Cholinesterase inhibitors, 175 Cimetidine, 364 Ciprofibrate, 117 Cis-(2S,4S)-2-phenyl-tetrahydropyran4-o1{ RS,4S-3 }, 229 Cisapride, 176 Cisatracurium, 257 Citalopram, 162, 254 Clemastine, 86, 88 Clemastine in rat plasma, 87 Clenbuterol, 51,147, 284 Clenbuterol-MIE 64 Clinical toxicology, 235 Clonazepam, 244, 253 Clozapine, 31 CNS stimulants, 175 Cocaethylene, 242 Cocaine, 24, 277 Cocaine and its metabolites, 242, 248 Cocktail dosing, 280, 282 Codeine, 11,122, 242, 245 Codeine, in human urine and porcine plasma, 82, 83 Codeine-6-glucuronide, in human urine and porcine plasma, 82, 83 Codethyline, 242 Co-immobilization, 231 Collision/reaction cell ICP-MS, 359 Column selection, 273 Combinatorial library analysis, 290 Continuous-flow HPLC-NMR, 308

416 Contract research organisations (CROs), 400 Convallatoxin, 262 Cordia linnaei, 305 Correlation spectroscopy (COSY), 299 Corticoids, 30 Corticosteroids, 20, 260 Corticosteroids in urine, 18 Corticosterone, 260 Cortisol and a metabolite in urine, 121 Cotinine, 236, 237 Coupled IMERs, 231 Covalent adduct formation, 317 Covalent attachment, 217 Covalent attachment of enzymes to chromatographic stationary phases, 218 Cross Validation, 379, 400 Crystal City Consensus Document, 374 Cyclodextrins as CSPs, 140 Cyclophosphamide, 169, 316 Cytochrome P450 enzymes, 229 2,4-D, 67 D-amino acid oxidase, 153 Darifenacin, 68 Darifenacin MIP, 68 Daunorubicin, 254 Daunorubicinol, 254 2-De(chlorethyl)ifosfamide, 316 Dephenhydramine, 80 Deprenyl, 122 Derivatised cyclodextrin chiral stationary phases, 142 Des-enkephalin-~-endorphin, 30 Deslanoside, 255, 262 Desmethylcitalopram, 162 Desmethyltrimipramine enantiomers, 163 Dextrans, 219 Dextromethorphan, 161 D-glyceradehyde-3-phosphate (D-GA3P), 220 Dialysis, 3, 27 Diarrheic shellfish poisoning, 262 Diastereomer, 129 Diazepam, 253,265 Diazepam, and its chiral and achiral metabolites, 172 Diazepam in urine, 9 Dibytylmelamine, 51 3,4-dichlor-N- {(1R,2S)2-(methylamino)cyclohexyl }benzamide, 161 Dichlorophenyl ketoamino acid, 169 Dichlorprop, 67 Diclofenac, 85, 363 Didesmethylchlorpheniramine, 164

Chapter Didesmethylcitalopram, 162 Diethylpropion, 157 Diflunisal, 87 Digitalis glycosides, 255 Digitalis purpurea, 261 Digitoxigenin, 262 Digitoxin, 255, 262 Digoxigenin, 255 Digoxin, 255, 262 Dihydrocodeine, 11,122 6,11-Dihydro 11-oxo-dibens (b,e) oxepin2-acetic acid, 317 3',4'-Dihydroxy flurbiprofen, 308 2-dimensional NMR experiments, 299 2-dimensional reversed-phase SPE, 78 Dimethylamine-N-oxide, 248 Dimethylamphetamine, 248 Dimethylarsinic acid (DMA), 263,264, 366 Dimethylformamide-d7, 314 Di-N,O-demethyltramadol, 156 2,4-Dinitrotoluene, 199 Dinophysistoxin, 263 Diphenidol, 170 Diphenylphosphate, 51 1,3-diphosphoglycerate (1,3-DPGA), 220 Direct immersion SPME (DI/SPME), 16, 17 Direct injection of biological fluids, 91 Disulfiram, 153 Ditolylphosphate, 51 Diuretics, 193,206, 207, 208, 209, 244, 255, 266 Diuron, 125 Divinylbenzene, 47 DMF-d7, 314 Dofetilide in dog plasma, 117, 118 Dolasetron mesylate, 164 Donepezil, 175 Donophusis acuta, 262 Dopamine, 231,233 Dopamine [3-hydroxylase (DBH), 222, 223, 224, 233 Dopamine [3-hydroxylase immobilized enzyme reactor (DBH-IMER), 231 Doping control, 235 Doxazosin, 117, 118, 134, 159 Doxepin, 80, 197, 201,202, 204, 253 Doxorubicin, 254 Doxorubicinol, 254 Drug impurities, 301 Drug interaction studies, 377 Drug metabolism, 306 Drug metabolite reactivity, 316 Drug transport, 290

417

Subject index

Ecdysteroids from silene otites, 306 Ecgonine methyl ester, 242 Effect of immobilization on enzyme kinetics, 219 Effect of immobilization on enzyme stability, 218 Effect of immobilization on the enzyme's response to pH, 222 Effect of immobilization on the thermal stability of an enzyme, 220 Electrodialysis, 27, 28 Electron capture detector (ECD), 386 Element specific detector, 351 Eltoprazine, 9 Eluent selection, 277 Emergency toxicological analysis, 235 Enalapril, 77 Enalaprilate, 77 Enantiomeric drugs, 129 Enantioseparation, 129 Endogenous analytes, 387 Endogenous biochemicals, 407 Enoximone and metabolite, 30 Envio nmental toxicology, 235 Enviro nmental applications of TFC, 125 Enzphetamine, 246 Enzymatic hydrolysis, 260 Enzyme immobilization on chromatographic supports, 217 Ephedrine, 28 Epicatechin, 196 Epicatechin gallate, 196 Epigallocatechin, 196 Epigallocatechin gallate, 196 Epinephrine, 231,233 Epirubicin, 254 Epirubicinol, 254 Erdosteine and its chiral metabolite, 178 Ergun model, 103 Establishment Inspection Reports (EIR), 375 Estazolam, 15 Ester glucuronides, 392 Estradiol, 407 Estrone, 407 Ethacrynic acid, 194, 196, 205 (R,S)-N-Ethyl3,4-methylenedioxyamphetamine, 155 1-Ethyl-phenoxathiin- 10,10-dioxide, 316 Ethylene glycol dimethacrylate, 47 2-Ethylidine- 1,5-dimethyl3,3-diphenylpyrrolidine (EDDP), 155 Ethylmorphine, 11 7-Ethoxycoumarin, 316 EU GMP Registration Annex 15, 378

Examethasone, 260 Extra column band broadening (ECBB), 276 FDA Compliance Manual for in vivo Bioequivalence Studies, 375 FDA guidelines, 374 Felodipine, 174 Fenatnayl, 236 Fenfluramine, 15,238 Fenoprop, 67 Fermentation broths, 28 Fick's law, 107 Flavonoids, 20 Fludrocortisone, 260 Fluethasone, 260 Flumequine, 30 Flunitrazepam, 244, 251 3-OH-Flunitrazepam, 253 Fluorobenzoic acid, 317 Fluorescent chiral tagging, 154 Fluoxetine (Prozac), 146, 178, 244, 253 Flurbiprofen, 167, 308 Flurbiprofen glucuronide, 308 2-Fluoro-4-iodoaniline, 364 Fluticasone propionate, 301 Fluvastatin, 165 Folic acid metabolism, 338 Forensic applications of TFC, 121 Forensic toxicology, 235 Formoterol, 150 Furosemide, 194, 195, 197, 198, 205,209 Gallic acid, 196 Gallocatechin gallate, 196 Gastroprokinetic agents, 176 GC/MS, 75 GC-ICP-MS, 360 GC-MS, 9, 122 GC-NPD of bupivacaine, 65 Generic methods, 272 Gitaloxin, 262 GLC, 130 Glibenclamide, 257 Glipizide, 257, 259 GLP, 375 Glucose-6-phosphate, 228 Glucose-6-phosphate dehydrogenase, 228 Glucuronide isomers, 317 Glutathione, 321 Gluteraldehyde-P stationary phase, 223 Glyceraldehyde-3-phosphate dehydrogenase, 218 Golay's equations, 98

418 Good Clinical Laboratory Practice (GCLP), 378 Gradient elution with high flow rates, 288 Graphite furnace atomic absorption (GFAAS), 358 Guidance for Industry, 374 2H NMR spectroscopy, 314 Habropetalum dawie, 306 Hallucinogenics, 176 Haloperidol, 15, 122, 263 Head-space SPME, 16 Herbicides, 125 Heroin, 11,122, 238 Hexokinase, 231 High resolution ICP-MS (HR-ICP-MS), 358 High-purity silicas, 187 High-throughput screening, 271 HIV protease inhibitors, 177 HIV/AIDS, 215 Hormone replacement therapy, 407 Hormones, 260 Horse liver alcohol dehydrogenase (HLADHIMER), 228 Horse liver alcohol dehydrogenase mediated reduction, 229 HPLC-CD, 302 HPLC-ICP-MS, 360 HPLC-ICPMS-TOFMS, 312 HPLC-MS, 122, 133 HPLC-MS-MS, 117, 129, 149 HPLC-NMR, 294 HPLC-NMR-MS, 294 HPLC-UV-IR-NMR-MS, 324 HRT, 407 HS/SPME, 16 Human serum albumin (HSA), 140 Human toxicology, 235 Hydrocortisone, 260 Hydromorphone, 236, 246 4-Hydroxyantipyrine, 307 p-Hydroxyamphetamine, 175 Hydroxybenzphetamine, 248 Hydroxybenzylamphetamine, 248 2-Hydroxydesmethyltrimipramine, 163 Hydroxyethylmethacrylate, 47 8-Hydroxy-(di-n-propylamino)tetralin, 172 4'-Hydroxy-flurbiprofen, 308 2-Hydroxyglutaric acid, 172 D-2-Hydroxyglutaric aciduria, 172 L-2-Hydroxyglutaric aciduria, 172 4-Hydroxyifosfamide, 316 p-Hydroxymethamphetamine, 175 3-Hydroxymethylantipyrine, 307

Chapter Hydroxymethylmexiletine, 160 oL-Hydroxymetoprolol, 152 p-Hydroxymexiletine, 160 Hydroxymidazolam, 251 5-Hydroxypropafenone, 158 9-Hydroxyrisperidone, 254 Hydroxytraizolam, 251 2-Hydroxytrimipramine enantiomers, 163 7-Hydroxy warfarin, 161 Hypernation, 323 IAM interphase, 218 Ibuprofen, 166, 238, 239, 307 S-Ibuprofen, 51 Ibuprofen methyl ester, 228 Idoxifene, 240 Ifosfamide, 316 Ifosforamide mustard, 316 Illicit drugs, 240, 242 IMER-HPLC, 225 Imipramine, 80, 197, 198, 201,202, 204 Immobilised antibodies, 5 Immobilised artificial membrane stationary phase (IAM-SP), 217 Immobilised enzymes, 216 Immobilised phenylethanolamine N-methyltransferase reactor (PNMTIMER), 226 Immobilized D-glyceraldehyde-3-phosphate dehydrogenase enzyme reactor (GAPDH-IMER), 219 Immobilized enzyme reactors, 215, 217 Immobilized phenyl-ethanolamine N-methyltransferase reactor (PNMTIMER), 226 Immunoaffinity extraction columns, 262 Immunoaffinity extraction of LSD, 249 Immunoaffinity sample pretreatment, 5 Immunosupressants, 254 Imprint preparation, 46 In vitro metabolism studies, 316 Inclusion complexing, on cyclodextrin CSPs, 141 Indapamide, 139 Inductively coupled plasma mass spectrometry (ICP-MS), 351 Inductively coupled plasma optial emission spectroscopy (ICP-OES), 358 Inorganic arsenic (V), 366 Inorganic compounds, 263 Inositol phosphates, in plasma, 28 Installation Qualification (IQ), 378 Insulin, 257 Insulin and C-peptide, 257

Subject index Integration errors, 395 Interbatch precision, 405 Internal standard, 386 International Conference on Harmonization (ICH), 374 Intrabatch precision, 405 In-tube SPME, 19, 24 Iodixanol, 31 Ion chromatography (IC) ICP-MS, 360 Ion pair chromatography, 32 Ion pairing SPE, 257 Ion suppression effects, 4 Ion suppression effects, of plasma extracts, 286 Ion suppression in MS, 285 Ion trap ICP-MS, 357 Ion-cyclotron resonance MS, 324 Iopentol, 30 Ipso-NAPQI-GSH adduct, 321 Isoproturon, 125 R(-)-4-(3-Isothiocyanatopyrrolidin- 1-yl)7-(N,N-dimethylaminosulphonyl)2,1,3-benzoxadiazole, 154 S( + )-4-(3-Isothiocyanatopyrrolidin-1-yl)7-(N,N-dimethylaminosulphonyl)2,1,3-benzoxadiazole, 154 Isradipine, 166 Ketamine, 157 Keterolac, 156 Ketobemidon, 246 Ketoprofen, 167, 168 Ketoprofen methyl esters, 227, 228 Kozeny-Carmen equation, 104 D-Lactate, ! 72 Lactones, 392 Lactose-hydrolysed milk, 222 Laminar flow, 97 Lamotrigine, 31 Lansoprazole, 172 Latanoside C, 262 LC electrochemical detection (LC-ECD), 64 LC ion trap mass spectroscopy (LC-IT-MS), 64 LC-ESI-TOF-MS, 240 LC-ESI-TOF-MS in toxicological screening, 265 LC-GC, 3, 7, 37 LC-MS in clinical toxicological analysis, 240 LC-MS-MS, 120, 238 Leu-enkephalin anilide, 50 Leukotriene antagonists, 177 Levosimendan, 31

419 Lidocaine, 20, 238, 239 Lidocaine in urine, 24 Ligand-drug transporter interactions, 224 Ligand-receptor, 224 Limit of detection (LOD), 382 Limit of quantification (LOQ), 382 e~-(1,4)-Linked glucose, 140 Lipase, 227 Lipase immobilized enzyme reactor (LPIMER), 228 Lipase on inorganic supports, 218 Lipoproteins, 322 Liquid chromatography-gas chromatography (LC-GC), 7 Liquid scintillation counting (LSC), 331 Liquid-liquid extraction (LLE), 157 Liquid-liquid extraction, in a 96-well plate format, 163 Liquid-liquid-liquid microextraction (LLLME), 34 L-Lactate, 172 Local anaesthetics, 31 Lorazepam, 172 Losigamone, 161 Lower limit of quantification (LLOQ), 380 LSD, 249 Lysergic acid diethylamide, 249, 251 Lysergic acid ethyl-2-hydroxyethylamide, 251 Macrocyclic antibiotics as CSPs, 143 Macromolecular binding, of benzene, 337 Macromolecular or polymeric CSPs, 137 Magnetic sector ICP-MS systems, 357 Manidipine, 174 Maprotiline, 254 Marocyclic glycopeptide chiral stationary phases, 144 Mass transfer, 99 Mass transfer, into pores, 107 Matrix effects, 385 MBDB, 177 MCA Orange Book, 378 MDA, 177 MDEA, 177 MDMA, and its metabolite MDA, 240 Measurement of selectivity differences, 190, 192 Medazepam, 6 Melatonin (N-acetyl-5-methoxytryptamine), 260 Membrane (non-porous), 36 Membrane extraction, with a sorbent interface (MESI), 27

Chapter

420 Membrane-based sample preparation techniques, 26 Mephenytoin, 161 Mephobarbital, 161 Mercury, 368 MESI, 32 Metabolic inhibition assays, 290 Metabolic stability, 290 Metabolites, in safety testing (MIST), 398 Metamizole, 263 Metanephrine, 260 Methacrylamide, 47 Methacrylic acid, 47 Methadone, 80, 155,265 Methadone, and its metabolite EDDP, 80 Methampetamine, 23, 34, 80, 84, 175, 245, 248 Method development, 185 Method development strategy, 200 Method transfer, 400 Methotrexate, 255 Methoxyverapamil, 81, 210, 211, 212, 213 Methyl mercury, 368 Methylarsonic acid (MMA), 366 Methyldigoxin, 262 Methyldopa, 226 N-methylformamide-d4, 314 N-methyltransferase immobilized enzyme reactor (PNMT-IMER), 231 3-Methyl-4-trifluoromethylaniline, 312 (R,S)-N-3,4-Methylenedioxyamphetamine, 155 3,4-Methylenedioxymethamphetamine, 23 3,4-Methylenedioxy-N-methylamphetamine (MDMA, ectasy), 176, 177 Methylphenidate, 146, 260 Methylprednisolone, 260 Methysergide, 249 Metobromuron, 261 Metoprolol, 80, 146, 150, 151, 152 Metrifonate, 157, 158 Metronidazole, 263 Mexiletine, 160 Mianserin and its desmethyl metabolite, 164 Michaelis-Menten kinetics, 215 Michaelis constant (Kin), 215 Microcoil probes, 322 Microporous membrane liquid-liquid extraction (MMLLE), 27 Midazolam, 244, 251 Miniaturisation, NMR detection systems, 322 Mirtazepine, 163 MISPE, 45, 64 MISPE method development, 52

MISPE of nitrophenol, 61 Mitomycin, 30 Mixed-mode ion-exchange sorbents, 74 MMLLE, 32, 34 Moclobemide, 253 Modafinil, 122, 176 Molecular Imprinted Polymers, 45, 46 MIPs - Hydroxycoumarin, 56 M I P s - 4-Aminopyridine nicotine, 57 M I P s - 4-Nitrophenol triazine herbicieds, 59 MIPs Atenolol, 56 MIPs Atrazine, 58, 60 MIPs Bentazone, 58 MIPs Brombuterol, 56 MIPs Bupivacaine, 56 MIPs - Caffeine, 56 M I P s - Chlorophenoxy-acetic acids, 58 M I P s - Chlorotriazine pesticides, 58 MIPs - Chlorphenamine, 56 MIPs choice of template, 48 MIPs - Clenbuterol, 60 MIPs - Darufebacub, 56 MIPs - Diphenylphosphate, 58 MIPs Diquat, 58 MIPs - Ibuprofen, 57 MIPs Indoleacetic acid, 60 MIPs - Naproxen, 57 MIPs Nerve agent degredation products, 58 MIPs Nicotine, 56 M I P s - non-specific adsorption, 53 M I P s - on-line extraction systems, 55 MIPs Pentamidine, 57 MIPs - Phenylure herbicides, 58 MIPs Phenytoin, 56 M I P s - Pirimicarb, 59 MIPs - Propranolol, 56, 57 MIPs Quercetin, 60 M I P s - removal of template molecules, 48 M I P s - Sameridine, 56 MIPs Simazine, 58 M I P s - solvent switch, 53 MIPs - Tamoxifen, 56 M I P s - template bleeding, 50 MIPs - Theophylline, 57 MIPs - Tramadol, 57 MIPSPE applications, 55 Monoacetylmorphine, 245 Monoacetylmorphine (6-MAM), 238, 242 Monodesmethylchlorpheniramine, 164 Monolithic column, 238 Monomethylarsonic acid (MMA), 263, 264 Mono-N-demethyltramadol, 156 Montelukast, 177 Morphine, 11,122, 238, 242, 245 -

-

-

-

-

-

-

-

-

-

-

-

-

-

Subject index

421

Morphine, and its analogues in urine, 9 Morphine, and its glucuronides, 236, 242 Morphine sulfate, from plasma, 110 Morphine-6-glucuronide (M6G), 242 Mosapride, and its des-4-benzyl metabolite, 176 Mucolytics, 178 Multiplex NMR, 322 Muscle relaxants, 257 1-Myristoyl-2- [( 13-carboxyl)tridecanoly)]-sn3-glycerophosphocholine, 218

Norfluoxetine, 178, 244, 253 Norketamine, 157 Norketobemidone, 246 Normetanephrine, 226, 260 Nortriptyline, 80, 197, 198, 201,202, 204 Norverapamil, 80, 81, 159, 210, 211,212, 213 N-oxides, 392 NPLC-GC, 12 NPMT, 32, 35, 36, 37 NSAID, 309 NSAIDs, 31,147, 324

N,N-dimethyl-analogue of sameridine, 51 Nalorphine, 242 Naltrexone, 80 NAPQI, 321 Naproxen, 238, 239, 309 S-Naproxen, 51 S-Naproxen glucuronide, 317 Naproxen chloroethyl ester, 227 Natreuretics, 177 Natriuretic hormone, 177 Natural products, 304 NBD-serine, 153 N-desmethylflunitrazepam, 253 Neritaloside, 262 Nerium Oleander, 262 Neutron activation analysis (NAA), 358 N-Glucuronides, 392 N-Hydroxymexiletine glucuronide, 159 N-Hydroxyparacetamol, 311 Nicardepine, 146, 174 Nicotine, 63,236, 237, 338 Nicotine metabolites, 261 Nifedipine, 173 Nilvadipine, 174 Nimodipine, 122, 174, 179 Nisoldipine, 174 Nitrazepam, 6 Nitrendipine, 174 p-Nitroanisole, 229 Nitroaromatics (explosives), 199 3-Nitro-2-(2-fluorophenoxy)pyridine, 316 Non-covalent entrapment, 217 Non-porous membrane techniques (NPMT), 27, 32 Norantipyrine-glucuronide, 307 Norbuprenorphine (NBU), 242, 246, 247 Norbuprenorphine-d3,buprenorphine, 247 Norcodeine, 242 Nordiazepam, 253 Nordoxepin, 80, 201,202, 204 Norephedrine, 175 Norepinephrine, 233

Oasis@ HLB sorbent, 74 Octopamine, 225 O-Demethyltramadol, 156, 246 Odoroside, 262 Off-line methods, 75 Off-line SPE-GC, 4 O-Glucuronyl-(R,S)-N-ethyl3,4-methylenedioxyamphetamine, 155 Okadaic acid, 263 Olanzapine, 254 Olanzapine- 10-N-glucuronide, 254 Oleandrigenin, 262 Oleandrin, 262 Omeprazole, 83, 171 Ondansetron, 158, 164 On-line bioreactors, 215 On-line coupling of dialysis with GC, 29 On-line dialysis-SPE-GC, 5 On-line dialysis-SPE-GC-NPD, 6 On-line enantiospecific synthesis and purification, 226 On-line filtration, 28 On-line immobilized enzyme reactors (IMERs), 222, 224 On-line LC-GC, 7 On-line LC-GC-MS, 10 On-line Michaekus-Menten kinetics, 225 On-line SPE, 4, 165 On-line SPE-GC, 4, 7 Operation Qualification (OQ), 378 Opiate agonists, 242 Opiate analysis in urine, 122 Optimum mobile phase velocity, 92 Organochlorines, 266 Organophosphates, 266 Organotin, 368 Orophea enneandra, 305 Ovomucoid, 140 Oxalidinones, 145 Oxazepam, 15, 253 2-Oxo-3-hydroxy-LSD, 251 Oxolinic acid, 30

Chapter

422 Oxprenolol, 31 Oxprenolol, in urine samples, 80 Oxybyutynin, 83, 134, 159 Oxytetracycline, 30 3~p detected HPLC-NMR, 316 3~p NMR spectroscopy, 314 PAHs, 126 Pancuronium, 257 Pantoprazole, 171 Paracetamol, 263, 279, 309, 310, 321 Paracetamol glucuronide, 309, 311 Paracetamol sulfate, 311 Pectenotoxin-2, 262 Pectenotoxin-6, 263 Penicillins, 277 Pentamidine, 63 Pentapeptides, 301 Pentazocine, 156 Pentycaine, 51 Pentycaine-MIR 65 Percentage of modem carbon (pMC), 336 Performance Qualification (PQ), 378 Pesticides, 125,244, 261 Pharmacokinetic studies, 377 Phenacetin, 311 Phenobarbital, 161,265 Phenobarbitone, 30 Phenols, 392 Phenothiazines, 20 Phenothiazines, in whole blood and urine, 18 Phenprocoumon, 160 Phentermine, 84 Phentolamine, 15 Phenylalanine, 50 L-Phenylalanine anilide, 50 Phenylboronic acid precolumn, 224 Phenylbutazone, 30 Phenylethanol chloroacetyl ester, 227, 228 Phenylethanolamine N-methyltransferaseimmobilised and non-immobilised, 221 Phenylethanolamine N-methyltransferase (PNMT), 220, 226, 233 (R,S)-2-phenyltetrahydopyran-4-one [(R,S)-I }, 228, 229 2-Phenyl propionic acids, 317 Phenytoin, 30, 31, 161 Phenytoin-MIR 68 Phloretin glycosides, 306 Pholcodine, 242 Pholcodine, 30 Phospholipids, 364 Phosphorylated peptides, 365 Phytochemistry, 306

Pindolol, 153, 159 Pineal gland hormone, 260 oL-Pinene, 322 Pipamperone, 254 Pipecolic acid, 173 Pirkle phases, 145 Plant toxins, 261 PME, 32 PMTs, 32, 38 Podophyllum emodi, 262 Polydimethylsiloxane (PDMS), 17 Polymeric membrane extraction (PME), 27 Polymeric sorbent, 74 Polythiazide, 255 Porcine insulin, 257 Porous membrane techniques (PMT), 26, 27, 28 Practolol, 147, 314 Praziquantel, 158 Precision, 382 Prednisolone, 238, 239, 260 Prednisone, 260 Pressure drop, 102 Primicarb, 63 Principal components analysis (PCA) applied to HPLC-NMR, 321 Probenecid, 194, 196, 205 Prochiral, 129 Programmed temperature vaporiser (PTV), 4 Proinsulin, 257 Prometryn, 125 Propafenone, 158 Propazine, 55 Propranolol, 62, 66, 80, 83, 153 Propranolol analogues, 139 Propranolol-MIR 66 Proscilardin, 262 Prospekt, 4 Protease inhibitors, 215 Protein A, 219 Protein binding, 29 Protein phases as CSPs, 139 Protein-based CSPs, 139 Pseudo-immunoassay, 46 Puromycin, 15 Pyrrolizidine alkaloids, 305 Pyruvate kinase, 231 Quality control acceptance criteria, 397 Quercetin, 68 Quercetin glycosides, 306 Quinolones, 31 Racemate, 129

Subject index Radio-chromatographic analysis, 332 Radioisotopes, 331 Radiolabelled drugs, 331 Radioluminography, 331 Radiosensitisers, 178 Radiotracer studies, 331 RAM-MIP-LC-MS, 61 Ranitidine, 20, 83 Recovery, 382 Regulatory changes, 408 Reintegration, 396 Remoxipride, 171 Rennin, 219 Repeatability, 405 Response function, 379 Restricted access material (RAM), 55, 61 Retinol, 339 Retinyl acids, 339 Reversed-phase CSPs, 141 Reversed-phase HPLC, 185 1-D Reversed-phase solid-phase extraction, 75, 76 2-D Reversed-phase solid phase extraction, 75, 77 Reynold's number, 99, 107 Risperidone, 254 Ristocetin, 143 Ristocetin A, 144 Ritalinic acide, 146 Rocuronium, 257 Rogletimide, 30 Ropivacaine, 7, 32, 33 Roxifiban, 312 RPLC-GC, 12 Saha equation, 352 Salbutamol, 83, 146, 149 Salicylic acid, 30 Sameridine, 50, 51 Sample pooling, 280 Sample preparation for AMS, 335 Saponins, 306 Scopolamine, 260 Screening procedures, for multiple compounds, 263 Secobarbital, 157 Secoiridoid glycoside, 305 Sedative/hypnotics, 178 Seemannoside, 305 Selectivity, 383 Selectivity differences, 188 Selectivity, via pH manipulation, 187 Selectivity, via temperature, 187 Selegiline, 238, 248

423 Selenium, 367 Seligiline-N-oxide, 248 Sensitivity, 382 Separation quality, 189 D-Serine, 153 Serotonin uptake inhibitors, 178 Serum protein binding, 224 SFC-NMR, 295 Silanols, 274 Sildenafil (Viagra), 31, 261 Silene otites, 306 Simazine, 55 Simazon, 125 Simendan, 159 Simvastatin, 13 SLME, 32 SLME to GC, 34 SLME with CE for bioanalysis, 34 SLME-LC-UV, 34 SLME-LC-CE, 34 Solid-phase extraction (SPE), 3, 45, 73 Solid-phase extraction-gas chromatography, 4 Solid-phase microextraction (SPME), 3, 16 Solvent selectivity prism, 187 Solvent selectivity triangle, 186, 187 Sotalol, 31, 134 SPE, 32, 36, 156 SPE-capillary electrophoresis, 3 specificity, 383 SPE-GC, 3, 36, 37 SPE-HPLC-UV, 11 SPE-LC, 36, 37 SPE-LC-MS-MS, 117 SPE-PTV/GC, 5 SPME, 16, 17, 36, 37, 62 SPME-GC, 17, 37 SPME-LC, 3, 18, 37 SPME-LC/ES-MS/MS for phenothiazine derivatives, 22 SPME-LC-ESI/MS, 23 SPME-MS, 24 Stability, 388 Standard curve acceptance, 397 Standard curves, 406 Standard operating procedures (SOPs), 377 Stationary phase evaluation, 276 Steroid conjugates in urine, 18 Steroid hormones, 5 Stir bar sorptive extraction, 26 Stop-flow HPLC-NMR, 307 Streptomyces rimosus, 306 Strophantidin, 262 Study documentation, 403 Succinylcholine, 257

424 Sulfonylurea antidiabetics, 257 Sulfonylureas, 244 Sulindac, 168 Sulphoxides, 145 Support bonds, 219 Supported liquid membrane extraction (SLME), 27 Suprofen, 135 Suspension polymerisation, 66 Swertia calycina, 305 System suitability, 393 2,4,5-T, 67 2,4,5-T-MIE 67 Tacrolimus, 254 Tamoxifen, 67, 238, 338 Taxol, 120 Taxus baccata, 262 Taxus brevifolia, 262 Teicoplanin, 143, 144 Teicoplanin Aglycone, 144 Temafloxacin, 160 Temazepam, 15, 238, 253 Temperature for selectivity manipulation, 199 Terazosin, 166 Terbinafine, 120 Terbutaline, 147 Terbuthylazine, 55, 125 Terbuthylazine MIP, 66 Terfenadine, 165 Terpenoids, 322 Testosterone, 407 2,3,5,6-Tetrafluoro4-trifluoromethylfluoroaniline, 312 1,2,3,4-Tetrahydropapaverine, 302 TFC-MS/MS, 117 TFE, 38 TFE-LC, 38 TFE-LC-MS/MS, 13 Thalidomide, 178 Theophylline, 30, 63 Therapeutic drugs, 240 Thermostable biocatalysts, 222 D,L-Threo-methylphenidate, 176 Thiamylol, 157 Thiopentone, 157 Thyromimetic agents, 179 Tiaprofenic acid, 167 Time of flight (TOF) ICP-MS, 357 Tin, 368 a-Tocopherol isomers, 179 RRR-oL-Tocopherols, 179 SSS-a-tocopherols, 179 Tolbutamide, 257, 259, 279

Chapter

Tolfenamic acid, 309 Toremifene, 338 Total correlation spectroscopy (TOCSY), 299 Toxicity Studies, 376 Toxicokinetics ICH-53A, 376 Toxicology, 235 Traizine type herbicides, 64 Tramadol, 61, 156, 246 Tramadol N-oxide, 246 Tramcinolone, 260 Trans-(2S,4S)-2-phenyl-tetrahydropyran4-o1{2S,4S-2 }, 229 Trans-4-hydroxypraziquantel, 158 Transacylated glucuronides, of 4-fluorbenzoic acid, 317, 319 Transferrin, 368 Trazodone, 80 Trialkylmelamines, 50 Triamterene, 194, 205,209 Triazine herbicides, 55 Triazolam, 251 Trichloroethylene, 338 2,4,5-Trichlorophenoxyacetic acid-MIR 66 Tricyclic antidepressants, 197, 201 Tricyclic antidepressants, 276, 277 Trifluoromethylacrylic acid, 47 Trifluoromethylbenzoic acid, 317 Trifluoroperazine, 312 Trifluralin, 125 Trihexylphenidyl (THP), 169, 170 Trimethoprim, 83 2,7,8-Trimethyl-2-([3-carboxyethyl)-6-hydroxy chroman, LLU-oL, 177 Trimethylolpropane trimethacrylate, 47 Trimipramine, 80, 163, 201,202, 204 Tripeptides, 301 Triprolidine, 15 Tris ((5)-1-phenylethyl carbamate cellulose CSR 138 Tris (4-methlbenzoate cellulose CSP, 138 Tris(3,5-dimethyl phenylcarbamate) cellulose, 138 Tubocurarine, 257 Turbulent flow chromatography coupled to tandem mass spectrometry, 115 Turbulent-flow chromatography (TFC), 3, 12, 91 Turbulent-flow extraction (TFE), 16 Tyramine, 225 UDP-glucuronyltransferase, 229 Ultra-high flow-rate LC, 16 Ultrasonic nebulisers, 356 Upper limit of quantification (ULOQ), 380

Subject index Van Deemter equation, 92, 93, 95, 96, 97 Van Deemter plot, 93 Vancomycin, 13, 143, 144 Vasodilators (cerebral), 179 Vecuronium, 257 Venlafaxine, 254 Verapamil, 81,134, 159, 210, 211,212, 213, 257 S-(-)-verapamil, 159 Verapamil + Norverapamil, 31 Veratridine, 262 Veratrum album, 262 2- or 4- Vinylpyridine, 47

425 Vitamins, 179

Warfarin, 30, 279, 306 (R) - Warfarin, 160 (S) - Warfarin, 160 Warfarin enantiomers, 261 96 well format, 117 96-well hexane extraction, 240 96-well plate extraction, 149 96-well plate solvent extraction, 260 96-well plates, 76, 77, 386 96-well solid phase extraction, 246

',0

oo c~

kll

c~

--,,1 oo

0" k,tl O0 I O~

13:1 Z C~ I

I,,-4