University of Groningen On-line coupling of sample ... · On-line coupling of sample pretreatment...

University of Groningen

On-line coupling of sample pretreatment with chromatography or mass spectrometry for high-throughput analysis of biological samplesHout, Mischa Willem Johannes van

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2003

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Hout, M. W. J. V. (2003). On-line coupling of sample pretreatment with chromatography or massspectrometry for high-throughput analysis of biological samples. Groningen: s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 23-06-2020

https://www.rug.nl/research/portal/en/publications/online-coupling-of-sample-pretreatment-with-chromatography-or-mass-spectrometry-for-highthroughput-analysis-of-biological-samples(04b52303-30f2-4097-8547-efa748eff200).html



On-line coupling of sample pretreatmentwith chromatography or mass

spectrometry for high-throughputanalysis of biological samples

Rijksuniversiteit Groningen

On-line coupling of sample pretreatmentwith chromatography or mass

spectrometry for high-throughputanalysis of biological samples

Proefschrift

ter verkijging van het doctoraat in deWiskunde en Natuurwetenschappenaan de Rijksuniversiteit Groningen

op gezag van deRector Magnificus, Dr. F. Zwarts,in het openbaar te verdedigen op

vrijdag 31 januari 2003om 14.15 uur

door

Mischa Willem Johannes van Hout

geboren op 15 september 1975te Alphen aan den Rijn

Promotores: Prof. Dr. G.J. de JongProf. Dr. R.A. de Zeeuw

Beoordelingscommissie: Prof. Dr. R.P.H. BischoffProf. Dr. U.A.Th. BrinkmanProf. Dr. A.P. De Leenheer

ISBN-nummer: 90-367-1714-0

Paranimfen: Naomi HagemanMichaël Knol

Printed by: Ridderprint BV (Ridderkerk, The Netherlands)

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

The financial support of

- Pharma Bio-Research Group- Xendo Laboratories- Organon- Spark Holland- ATAS GL- Groningen University Institute for Drug Exploration (GUIDE)

for publication of this thesis is gratefully acknowledged.

Table of contents

1. Introduction ............................................................................................. 91.1 General remarks................................................................................. 111.2 Scope of the thesis ............................................................................. 151.3 References ......................................................................................... 18

2. New developments in integrated sample preparationfor bioanalysis .......................................................................................... 212.1 Introduction ....................................................................................... 242.2 Flow-through sample preparation techniques.................................... 272.3 Diffusion-based sample preparation techniques ................................ 392.4 Concluding remarks........................................................................... 582.5 References ......................................................................................... 62

3. Integration of solid-phase extraction and gas chromatography.......... 693.1 Evaluation of the programmed temperature vaporiser for

large-volume injection of biological samplesin gas chromatography....................................................................... 71

3.2 Coupling device for desorption of drugs from solid-phaseextraction – pipette tips and on-line gas chromatographic analysis .. 87

3.3 Feasibility of the direct coupling of solid-phase extraction –pipette tips with a programmed temperature vaporiser forgas chromatographic analysis of drugs in plasma ............................. 97

3.4 Solid-phase extraction – thermal desorption – gas chromatographywith mass selective detection for the determination of drugsin urine............................................................................................... 111

4. Coupling of solid-phase extraction and mass spectrometry ................ 1234.1 On-line coupling of solid-phase extraction with mass spectrometry

for the analysis of biological samples. Determination ofclenbuterol in urine using multiple-stage mass spectrometry inan ion-trap mass spectrometer ........................................................... 125

4.2 Ion suppression in the determination of clenbuterol in urine bysolid-phase extraction – mass spectrometry ...................................... 143

4.3 On-line coupling of solid-phase extraction with mass spectrometryfor the analysis of biological samples. Determination ofprednisolone in serum........................................................................ 157

5. Coupling of solid-phase microextraction and mass spectrometry ...... 1695.1 Non-equilibrium solid-phase microextraction coupled directly to

ion-trap mass spectrometry for rapid analysis of biologicalsamples .............................................................................................. 171

5.2 Ultra-rapid non-equilibrium solid-phase microextraction atelevated temperatures and direct coupling to ion-trapmass spectrometry for the analysis of biological samples................. 183

6. General conclusions and future perspectives ........................................ 197

Samenvatting ............................................................................................... 205List of publications ...................................................................................... 211Dankwoord................................................................................................... 213

11INTRODUCTION

The

General remarks

11

1.1 General remarks

The bioanalysis of drugs and related substances has always been aspecialism dealing with rather complex samples. The most frequently analysedbiological samples are urine and blood, plasma or serum, but analyses of saliva,hair, sweat, cerebrospinal fluid, vitreous humour and tissue homogenates arealso performed. Even though urine is mainly an aqueous sample, it can varybetween individuals in viscosity, salt contents and presence of other compoundsdue to diseases, drink and food consumption as well as by administered drugs.As urine is considered to be a major waste disposal of the body, one can expectmany substances to be present in this fluid. Blood is the transporter of manyvital substances and nutrients for the entire body and thus contains manyendogenous and exogenous compounds in different concentrations. An extraproblem posed by blood samples is the presence of proteins, which can lead toprotein binding of the analyte.

The importance of bioanalysis is most appreciated in various fields ofpharmaceutical sciences. During the development of new drugs, extensivestudies are performed in the pre-clinical and clinical stages. At the pre-clinicalstage, fluids from animals are analysed. In the clinical stage, human samplesneed to be examined. Virtually all aspects of a new drug will be investigated.The toxicological and therapeutic concentrations of the parent drug and itsmetabolites must be determined, in combination with the pharmacodynamic andpharmacokinetic properties of the potential drug. Finally, the formulation of thedrug must be optimised, which also relies to a large extent on bioanalyticalevaluation. The sooner a drug can be placed on the market, the better for allinvolved, thus rapid development is preferred. This means that many samplesshould be analysed in a rather short time. Once a drug is on the market,therapeutic drug monitoring can be a vital aspect. Other important applicationsof bioanalysis are, amongst others, the control of residues in food andfood-producing animals, drug abuse testing, clinical and forensic toxicology andenvironmental control. Thus, in various application fields many biologicalsamples need to be analysed. Furthermore, the knowledge of workingmechanisms of drugs is increasing. Consequently, more potent andendogenous-like drugs are developed, allowing the administration of lowerdosages of drugs, which results in lower concentrations of the compound and/orits metabolites in blood and/or urine samples. Thus, the number of complexsamples is increasing, whereas the concentrations of the analyte(s) aredecreasing. As a result, there is a strong demand for highly sensitive andselective systems that can be used in high-throughput analysis.

Biological samples can normally not be injected directly into theanalysing system without sample preparation. Sample pretreatment is thus ofutmost importance for the adequate analysis of drugs. However, as sample

Chapter 1 – Introduction

12

pretreatment can be a time-consuming process, this can limit the samplethroughput. The proper selectivity can be obtained during the samplepreparation, the separation and the detection. A major differentiation betweenthe analyte(s) of interest and the other compounds is often made during the firststep. Sensitivity is to a large extent obtained by the detector. However,pre-concentration of the analytes during sample preparation will also add tothis. Thus, sample pretreatment is required for achieving sufficient sensitivityand selectivity, whereas the time should be kept to a minimum in order to obtainadequate speed. Therefore, there is a clear trend towards integration of samplepretreatment with the separation and the detection.

Numerous sample preparation techniques have been developed forbioanalytical purposes. The most classical sample preparation technique isliquid-liquid extraction (LLE) [1]. The broad polarity range of solvents and itsgeneral applicability made this technique popular. However, from anenvironmental point of view, the use of large amounts of organic and oftenchlorinated solvents is unfavourable. Furthermore, LLE has often limitedseparation efficiency. Various techniques, like protein precipitation,membrane-based techniques (dialysis, ultrafiltration, supported liquidmembrane extraction), supercritical fluid extraction (SFE), solid-phaseextraction (SPE) and solid-phase microextraction (SPME), have been developedfor sample pretreatment in order to replace LLE or to introduce new approachesto sample preparation. Nowadays, in particular SPE is replacing LLE due to thevariety of available stationary phases, allowing either non-selective extractionor selective extraction of only a single compound or class of compounds.Generally less solvent is required for SPE than for LLE [1,2]. Despite thesedevelopments, the sample pretreatment was often considered a weak andtime-consuming link in a bioanalytical system up to ten years ago. Thus, mucheffort was put in exploring the possibilities of miniaturisation and automation ofthe extraction procedures to minimise or eliminate the limitations of the samplepreparation.

With the integration of sample pretreatment, separation and detection,coupled techniques were developed, which were linked by hyphens in writtenlanguage. Consequently, the term ‘hyphenated technique’ was introduced.Nowadays, hyphenation refers to systems that combine separation and detectionoffering structural information. The most important and routinely applieddetection in such systems in biopharmaceutical analysis is mass spectrometry(MS). Initially, MS was mainly used for identification of compounds. With theintroduction of atmospheric pressure ionisation interfaces like electrospray andatmospheric pressure chemical ionisation, (LC-) MS has gained tremendousinterest, and has shown good potential for quantitation as well as identification.Furthermore, multi-stage MS (MSn with n≥2) increased the potential forselective detection. Currently, the interest for bioanalysis using MS is somewhat

General remarks

13

dualistic. On the one hand, MS can offer extreme selectivity, hereby enhancingthe possibilities for more rapid and simple sample pretreatment and separation.Incomplete separation of the analytes from interfering compounds can beovercome if the components have different m/z values. However, the presenceof such compounds may still lead to interferences during ionisation, e.g., by ionsuppression or enhancement [3,4], when the interfering compounds and theanalyte of interest co-elute. Furthermore, elution of substances that do notoverlap with the analyte may also pose problems due to contamination ofvarious parts of the MS. Thus, careful considerations should still be made aboutreducing sample preparation and separation. On the other hand, MS can offerhigh sensitivity, especially if the selectivity is enhanced as well. This is ofparticular importance for extractions in which little pre-concentration takesplace and/or in which the recoveries or yields are low, e.g., in SPME. Theoverall effect of the application of MS in the biopharmaceutical field is fasterseparation. Consequently, the sample preparation has become critical again.Good selectivity is required to minimise the risk of detection problems, andrapid extraction should be performed in order to maintain high-throughputanalysis. Thus, the sample pretreatment needs to be further accelerated andautomated.

To minimise the required volumes of solvent as well as sample, SPE isavailable in a miniaturised format such as the SPE disk and the SPE pipette tip(SPEPT) [5,6]. Generally, SPE disks contain a smaller bed with a morehomogeneous particle size distribution than conventional cartridges. The disk istypically a membrane-like bed with a thickness of about 0.5 mm and a diameterof down to 4 mm, which can be easily adapted to an at-line 96-well approach,thus decreasing the handling time per sample. The SPEPTs are modified pipettetips with a small disk (about 4 mg, 4 mm diameter, thickness including fritsabout 1.5 mm) of stationary phase positioned in the point of the tip. The formatsallow more rapid flushing of sample and smaller volumes of various solventsthan in conventional SPE [5]. However, even though the miniaturisationenhances the sample handling speed, the various steps in the SPE procedure,that is activation, conditioning, sampling, washing, drying and elution, are stillnecessary [2,6]. Generally, a disadvantage of miniaturisation is that smallersample volumes have to be used, hereby possibly decreasing the sensitivity if asufficient amount of sample is available. On the other hand, the miniaturisationof SPE enhances the potential for automation and on-line coupling with variousseparation systems. SPE has been primarily coupled on-line with liquidchromatography (LC) [7,8], since aqueous solvents may be flushed directly to areversed phase (RP) LC-column. The simplest on-line arrangement forSPE-RPLC is the use of switching valves and commercial pre-columns [7,9,10].On-line SPE-LC is now a well-established technique. With the automation ofthe various steps of the SPE procedure, the LC may now become the


14

time-limiting step, but this could be counteracted by the use of MS. Thesimultaneous increase in selectivity and sensitivity of MS allowed the use ofshort-column LC [11-13], hereby reducing the time-limitations. Even the directcoupling of SPE with MS has been applied [14-17]. However, as already statedabove, systems based on SPE-MS will put extra pressure on the speed of theextraction as well as on the selectivity of the extraction and the detection, sinceno real separation is performed. Thus, the possibilities and limitations of bothsteps must be carefully explored and possibly modified with regard to those ofSPE-LC-MS systems.

The on-line coupling of SPE with gas chromatography (GC) is importantfor toxicological and drug abuse analysis, since various recommended and/orlegally accepted procedures require the use of GC. However, integrating SPEwith GC is more complicated than SPE-LC, and care should be taken to avoidthe introduction of water and other polar solvents into the GC. This implies thatthe SPE stationary phase should be dried extensively prior to the elution. Theelution poses another key problem, since the elution volume is usually too largefor direct injection into the GC. To increase the compatibility of integrated SPEand GC, a special injection procedure allowing large-volume injection (LVI)must be applied, hereby possibly eliminating the error-prone evaporation/reconstitution processes. With LVI the entire eluate, or a major portion there of,is introduced into the GC, which can result in increased sensitivity. Yet, notonly a large amount of solvent is introduced, but also an equivalent amount ofimpurities that are co-extracted with the analytes. Thus, special attention shouldbe paid to the selectivity of the extraction and detection. A number of interfaceshave been proposed for LVI [18], i.e., on-column injection [19], loop-typeinjection [20] and the programmed temperature vaporiser (PTV). The PTVinjector was designed by Vogt et al. [21,22], and resembles a conventionalsplit/splitless injector. The main solvent is injected at temperatures 30-40ºCbelow its boiling point into a packed liner. After almost complete evaporation ofthe solvent, the analytes are thermally released from the packing and transferredto the GC column. Critical factors in the use of the PTV for LVI are themaximum injection volume, which depends on the flow-rate during theinjection, the injected solvent and the temperature of the injector [23].Furthermore, the inertness of the liner packing must be closely monitored.Besides in LVI, the PTV injector can also be used for solid-phase extraction –thermal desorption (SPETD), since the injector can be heated very rapidly (upto 16°C/s). After the extraction the analytes are thermally desorbed from thesorbent inside the PTV injector and, subsequently, conventional GC isperformed. In SPETD with the PTV, a suitable sorbent must be used inside theliner to obtain retention of the analytes. Furthermore, the sorbent must be cleanand thermostable. Obviously, regardless of the approach, the analytes must haveadequate thermostability to withstand the high temperatures in both the injector

Scope of the thesis

15

and the GC. Generally, the use of MS is required for adequate selectivity inSPE-GC systems for bioanalysis. The investigation of the possibilities andlimitations of integrating SPE with GC will also be a major part of this thesis.

Another interesting approach of miniaturised sample preparation is SPME[1,24]. In contrast to the flow-through sampling of SPE, a passivediffusion-based process is used in SPME. The set-up is rather simple, as acoated fiber is used in a syringe-like device. Excellent papers can be found onthe principles, theory, practical aspects and applications of SPME [24-29]. Thelatter refer mainly to SPME combined with GC, as this type of extraction wasoriginally designed for GC analysis. Most applications are in the field ofenvironmental analysis, but more recently the utility of SPME for bioanalysishas also been demonstrated [25,26]. The number of SPME-LC applications isstill limited, but nowadays the possibilities of SPME followed by liquiddesorption and LC are being explored more and more in various areas, allowingthe analysis of thermolabile compounds. A simple valve-like interface with adesorption chamber is used for the desorption of the analytes using anappropriate solvent composition [30,31]. The three main limiting factors inSPME-based systems are correlated to the SPME principles. The extraction isdiffusion-driven and non-exhaustive, which can result in low yields, especiallyin comparison with SPE. The diffusion also limits the throughput due to thelong times required to reach equilibrium. The third factor, which is only validfor SPME-LC, is the static desorption, which is also based on an equilibriumprocess. Thus, incomplete desorption of the analyte from the coating may alsooccur, resulting in carry-over. Non-equilibrium SPME can be used to decreasethe extraction time, which also allows the direct coupling of SPME with MS.

1.2 Scope of the thesis

The objectives of this thesis are to develop fast, sensitive, integratedsystems for high-throughput analysis in the biopharmaceutical field. Since thepossibilities of integrated SPE-LC and SPME-GC are reasonably well known,also with regard to automation, our major focus was on SPE-GC and on thedirect coupling of SPE and SPME with MS. SPE-GC has hardly been applied inthe biopharmaceutical field so far. Generally, the limitations of SPE-GC are thedesorption volume and the time-consuming drying step. Therefore, thepossibilities and limitations of novel types of integration of SPE with GC havebeen investigated and the application for biological samples has beendemonstrated.

Though SPE-LC is a well-established technique, its use inhigh-throughput systems is limited by the time required for both the extractionand the separation. The direct coupling of SPE with MS (i.e. omitting the major


16

separation step) by a suitable interface is very interesting for more rapidbioanalysis. Investigations with regard to the potentials and pitfalls of SPE-MSwere performed.

Another appealing system is the direct coupling of SPME with MS. Theuse of SPME in high-throughput systems is more limited than for SPE-basedsystems due to the long sorption times. An approach based on non-equilibriumSPME was developed that allowed the use of SPME directly coupled with MSfor bioanalysis while high sample throughput and adequate sensitivity isobtained.

In Chapter 2 an overview is given over various sample pretreatmenttechniques in bioanalysis with a focus on integration of sample pretreatment andseparation and/or detection. New developments and less common combinationsare highlighted, whereas well-established techniques such as SPE-LC are notdiscussed in detail. Firstly, the principles of the techniques and special aspectsare presented. Secondly, bioanalytical applications are described. Finally, thepotentials and limitations of the techniques are discussed, with a focus on theobtained selectivity and sensitivity. The versatility, the ease of operation and thecomplexity of instrumentation and possible automation are also included in thediscussion.

The integration of SPE with GC for bioanalytical purposes is described inChapter 3, focusing on the miniaturisation of SPE and possible automation ofthe SPE-GC system. Integration of SPE with GC implies the injection of largevolumes of solvents into the GC. Section 3.1 shows the results of the evaluationof the PTV for LVI of extracts of biological samples. Special attention was paidto the solvent purity as well as the liner packing (maximum injection volumeand packing inertness). Also the selectivity obtained by the total procedureproved to be critical. Extra selectivity was obtained by the use of a massselective detector (MSD) in the selected ion monitoring (SIM) mode. The finalsystem allowed injection of the entire eluate of SPE (100 µl), and goodsensitivity was obtained, i.e. a limit of detection (LOD) down to 250 pg/ml forthe test drugs lidocaine, phenobarbital, secobarbital and diazepam in plasma.Section 3.2 deals with the use of SPEPTs, which allows smaller desorptionvolumes, and thus eliminates the need for evaporation and reconstitution of theeluate. An exterior coupling device was developed by which the SPEPTs couldbe coupled to the PTV injector. By these means, rapid off-line extraction wasfollowed by on-line desorption. Even further integration was performed byapplying a so-called internal coupling device (Section 3.3). A shortened packedliner and a shortened SPEPT were coupled inside the injector after off-lineextraction of 200 µl plasma. The desorption was performed in-line by theinjection of ethyl acetate on top of the disk of the SPEPT. This coupling deviceis an important step towards the integration of miniaturised and rapid SPE withPTV/GC. Another approach is that the extraction is followed by thermal

Scope of the thesis

17

desorption. This SPETD approach is discussed in Section 3.4. Tenax wasinserted into a fritted liner after which off-line extraction and subsequentlyin-line thermal desorption was performed. A small amount of stationary phase(5 mg) allowed rapid extraction (8 min, including 5 min drying). With thisSPETD-GC-MSD system (SIM mode) and by extracting only 50 µl urine,LODs below 500 pg/ml were obtained for the test drugs lidocaine anddiazepam.

Chapter 4 deals with the direct coupling of SPE via an LC-MS interfacewith an ion-trap MS to obtain a rapid screening system. In Section 4.1,clenbuterol was determined in urine at a sub-ng/ml level applying SPE with apolymeric stationary phase and multiple-stage MS, that is MS3. Although cleanchromatograms were obtained, some ion suppression was observed. InSection 4.2, a post-cartridge continuous infusion system was applied todetermine ion suppression and an investigation was performed to check whichurine compounds were causing the ion suppression during the determination ofclenbuterol. A comparison was made between the selectivity and ionsuppression effects after extraction by the polymeric phase and a more selectivephase, that is a molecularly imprinted polymer (MIP). The former showedsignificant interferences of urine matrix compounds, whereas with MIPcartridges the bleeding of the template molecule caused the ion suppression.Section 4.3 describes the use of an SPE-MS2 system for the determination ofprednisolone in serum. Due to unfavourable fragmentation of the analyte a highdetection limit (5 ng/ml) was observed when using an ion-trap MS. Due tomatrix interference, no improvement was observed with a triple-quadrupole MSwith atmospheric pressure chemical ionisation or atmospheric pressure photoionisation. The total analysis time was less than 5 min.

Chapter 5 describes the direct coupling of SPME with MS as ahigh-throughput system for bioanalysis. Lidocaine was extracted from urine bynon-equilibrium SPME with a 100-µm polydimethylsiloxane (PDMS) coatedfiber at room temperature (Section 5.1). The total analysis time was about10 min (including desorption) and an LOD in the sub-ng/ml range was obtained,even though the total yield was only about 7%. Acceptable reproducibility(interday relative standard deviations < 14%) was observed. In an SPME-MSn

system the sorption is time-limiting. In order to speed up the system it is usefulto discern the limiting factors during the sorption. Using elevated temperaturesduring the sorption and desorption, the diffusion processes could besignificantly enhanced. In Section 5.2, the effect of the temperature on thesorption and the desorption is described, and an optimised SPME-MS2 system ispresented for determination of lidocaine in urine. When applying a 30-µmPDMS coated fiber with only 1 min sorption at 60°C, 1 min desorption at 55°Cand 1 min MS (i.e. a turn-around time of some 3 min), reproducible results were


18

obtained in the sub-ng/ml range. This system clearly shows the potential ofnon-equilibrium SPME at elevated temperatures coupled directly to MS.

In Chapter 6 some general conclusions and future perspectives arepresented. The potentials and limitations of the various systems with integratedsample preparation, which are presented in this thesis are discussed and, wherepossible, compared with each other.

1.3 References

[1] Z.E. Penton. Advances in Chromatogr. 37 (1997) 205.[2] J.P. Franke, R.A. de Zeeuw. J. Chromatogr. B 713 (1998) 51.[3] K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng. Anal. Chem. 70 (1998)

882.[4] D.L. Buhrman, P.I. Price, P.J. Rudewicz. J. Am. Soc. Mass Spectrom. 7 (1996)

1099.[5] R.E. Majors. LC-GC Intern. May, 1998, S8-S15.[6] J.S. Fritz, M. Macka. J. Chromatogr. A 902 (2000) 137.[7] M.C. Hennion. J. Chromatogr. A 856 (1999) 3.[8] D.T. Rossi, N. Zhang. J. Chromatogr. A 885 (2000) 97.[9] D.A. McLoughlin, T.V. Olah, J.D. Gilbert. J. Pharm. Biomed. Anal. 15 (1997)

1893.[10] M.W.F. Nielen, A.J. Valk, R.W. Frei, U.A.Th. Brinkman. J. Chromatogr. 393

(1987) 69.[11] A.C. Hogenboom, P. Speksnijder, R.J. Vreeken, W.M.A. Niessen, U.A.Th.

Brinkman. J. Chromatogr. A 777 (1997) 81.[12] A.C. Hogenboom, W.M.A. Niessen, U.A.Th. Brinkman. J. Chromatogr. A 794

(1998) 201.[13] W.A. Minnaard, A.C. Hogenboom, U.K. Malmqvist, P. Manini, W.M.A.

Niessen, U.A.Th. Brinkman. Rapid Commun. Mass Spectrom. 10 (1996) 1569.[14] A. Schellen, B. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de

Lagemaat, E. Verheij. Rapid Commun. Mass Spectrom. 14 (2000) 230.[15] J. Ding, U.D. Neue. Rapid Commun. Mass Spectrom. 13 (1999) 2151.[16] C.H.P. Bruins, C.M. Jeronimus-Stratingh, K. Ensing, W.D. van Dongen, G.J. de

Jong. J. Chromatogr. A 863 (1999) 115.[17] G.D. Bowers, C.P. Clegg, S.C. HughesC, A.J. Harker, S. Lambert. LC•GC 15

(1997) 48.[18] H.G.J. Mol, H.-G. Janssen, C.A. Cramers, J.J. Vreuls, U.A.Th. Brinkman.

J. Chromatogr. A 703 (1995) 277.[19] K. Grob, J.-M. Stoll. J. High Resolut. Chromatogr. 9 (1986) 518.[20] K. Grob Jr., G. Karrer, M.-L. Riekkola. J. Chromatogr. 334 (1985) 129.[21] W. Vogt, K. Jacob, H.W. Obwexer. J. Chromatogr. 174 (1979) 437.[22] W. Vogt, K. Jacob, A.-B. Ohnesorge, H.W. Obwexer. J. Chromatogr. 186 (1979)

179.[23] W. Engewald, J. Teske, J. Efer. J. Chromatogr. A 856 (1999) 259.[24] H.L. Lord, J.B. Pawliszyn. J. Chromatogr. A 885 (2000) 153.[25] G. Theodoridis, E.H.M. Koster, G.J. de Jong. J. Chromatogr. B 745 (2000) 49.[26] S. Ulrich. J. Chromatogr. A 902 (2000) 167.

References

19

[27] N.H. Snow. J. Chromatogr. A 885 (2000) 455.[28] H.L. Lord, J.B. Pawliszyn. J. Chromatogr. A 902 (2000) 17.[29] R. Eisert, J. Pawliszyn. Crit. Rev. Anal. Chem. 27 (1997) 103.[30] J. Chen, J.B. Pawliszyn. Anal. Chem. 67 (1995) 2530.[31] A.A. Boyd-Boland, J.B. Pawliszyn. Anal. Chem. 68 (1996) 1521.

22NEW DEVELOPMENTS IN INTEGRATED

SAMPLE PREPARATION FORBIOANALYSIS*

The pieces

*: M.W.J. van Hout, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de Jong. Accepted for publication inHandbook of Analytical Separations: Pharmaceuticals / Veterinary – Toxicology / Metabolism (Ed. I.Wilson).

Summary

23

Summary

Biological samples are commonly considered to be complex matrices. Theanalysis of such samples requires an adequate sample preparation step prior tothe separation, detection and quantitation. Since the sample pretreatment is atime-consuming procedure, various systems have been developed to couplesample preparation with the separation/detection/quantitation step in order tospeed up the procedure and to eliminate error-prone handling. In this chapter,the focus will be on new developments in integrated sample preparation andseparation systems. Less common integrations offering high-throughputpotential such as SPE-GC, SPME-LC, SPE directly coupled to massspectrometry (MS) and turbulent-flow chromatography (or extraction) will behighlighted. Furthermore, some challenging approaches, e.g. non-porousmembrane-based extractions will be described. The systems and typicalapplications will be shown. The possibilities and limitations of the techniquesfor bioanalytical purposes will be discussed and compared.

Chapter 2 – New trends in integrated sample preparation for bioanalysis

24

2.1 Introduction

Increasing knowledge of the working mechanisms of drugs has led to thedevelopment 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 manyendogenous substances. Blood fluids, such as serum and plasma, represent anextra problem due to the presence of proteins. Protein binding may affect theextractability of the analytes. Deproteinisation techniques can help to overcomethis problem. It may, however, also give rise to even more difficulties, sinceanalytes can be co-precipitated with the proteins. Thus, sample pretreatmenttechniques are required that retain the analyte(s) of interest, at the same timeefficiently removing the endogenous interferences. The most common systemsexist of an extraction step prior to separation and detection. A considerable gainin sensitivity and selectivity can be obtained during the extraction, as theanalytes of interest are usually concentrated and separated from the matrix. Anideal extraction method should be rapid, simple, inexpensive, and givereproducible and high recoveries without the possibility of degradation of theanalytes. Furthermore, the extraction method should not generate large amountsof chemical waste [1].

Sample pretreatment used to be a long step in the analysis of biologicalsamples. Since the number of samples to be analysed is increasing, very rapid,but still selective and sensitive systems are required. In modern systems usingadvanced 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 isagain more and more limited by the time required for sample pretreatment. Thisis especially the case in off-line systems, which may also require extensivemanual work. Therefore, various systems have been developed in order tointegrate sample pretreatment with the separation and detection technique(Fig. 1) [2].

Basically, three possibilities have been proposed for integrated samplepretreatment 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 deviceand an auto-injector to inject the extracts into the analytical instrument. Nodirect 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 theextract, evaporation and reconstitution, are not eliminated. An example of anat-line system is the 96-well plate design for solid-phase extraction (SPE).

Introduction

25

Fig. 1: Schematic presentation of various integration methods of the samplepretreatment step with the separation and detection technique [2, modified].

Samples can be extracted simultaneously, thus increasing the samplethroughput, provided that the separation and detection can be performed veryrapidly or by using simultaneous analytical instruments. With on-line systems,there is a direct transport of the entire extract to the analysing technique, and thelatter receives the entire extract. Samples can be processed in series, i.e.,samples are pretreated and analysed one after the other, or in parallel, in whichone sample is being analysed while another is being extracted. The latter systemoffers a higher sample throughput. A very prominent advantage of on-line

Off-line

At-line

On-line

In-line

Samplepretreatment

Separation/Detection


26

systems is that some error-prone steps of the extraction procedure, such asevaporation and reconstitution are eliminated, hereby increasing precision andaccuracy. In-line systems exist of sample pretreatment fully incorporated intothe separation system, hereby creating a new device. In contrast to on-lineprocedures, application of in-line systems implies the direct injection of thesample into the analytical instruments. Various approaches for in-lineSPE-capillary electrophoresis have been reported [2]. It should be noted that thedifferences in interfacing are often not as clear as mentioned above. Forexample, 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 analytesmay be performed on-line with the analytical step. Furthermore, dividingsystems into on-line and in-line techniques is very disputable. These systemsare usually closely related to each other and a distinct difference can often notbe made. Therefore, in this chapter on-line and in-line systems will beconsidered similarly.

The goal of this chapter is to show the current status of modern samplepretreatment techniques such as SPE, solid-phase microextraction (SPME) andmembrane-based extraction systems, and to outline novel trends in thebioanalytical area with regard to integrated sample preparation. It will focus notonly on pretreatment techniques integrated with chromatographic separationsystems, but also on their direct coupling to MS. SPE was originally designedfor 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 lesscommon, especially in the bioanalytical field. The current state of SPE-GC willbe discussed here. Since an LC column can also be used as clean-up prior to GCanalysis [10-13], on-line LC-GC applications without any further samplepretreatment will also be presented. Turbulent-flow chromatography (TFC) is toa certain extent similar to SPE. The use of high flow-rates offers newpossibilities for sample pretreatment [14-17]. Therefore, the current state inTFC will be presented. SPME was originally designed for the analysis ofvolatile compounds with GC [18-22]. However, nowadays SPME is alsocoupled with LC for analysis of less-volatile compounds. The applicability ofthese SPME-LC systems in bioanalysis will be shown. Membrane-basedtechniques are, like SPME, diffusion-based sample pretreatment techniques.Dialysis is a more mature membrane method for sample pretreatment [23-25].However, non-porous membranes provide new challenges for clean-up ofbiological samples. Therefore, the focus of membrane-based techniques forsample clean-up will be on the latter type.

Although subdividing the various sample pretreatment techniques intocategories may be questionable, in this chapter we have chosen to considerSPE-GC, LC-GC and TFC-LC as techniques in which the major part of thesample handling involves flow-through of the sample in the stationary phase,

Flow-through sample preparation techniques

27

during which the analyte needs to be sorbed by this phase. With SPME andmembrane-based techniques, diffusion of the analyte is the key principle.Therefore, these sample preparation techniques will be further mentioned asdiffusion-based techniques.

2.2 Flow-through sample preparation techniques

2.2.1 Solid-phase extraction – gas chromatography

2.2.1.1 General aspects of SPE

A very common and powerful sample clean-up and concentrationtechnique is SPE. It was originally developed for off-line purposes, but due tothe demand for speed and the growing numbers of samples, at-line (including96-well designs [9,26,27]) and on-line systems, such as the Prospekt, have beendeveloped for coupling with LC [6,7,28,29]. On-line SPE is a very attractivesample 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, theentire eluate is usually injected into the analytical instrument. Therefore, betterprecision and sensitivity may be observed when compared to off-line SPE.

The most common on-line coupling of SPE is with LC, since similarsolvents are used and virtually no modifications have to be made to theinstruments. As this technique has already evolved and is mature, the on-linecoupling of SPE-LC will not be discussed here in detail. Worth mentioning,however, is the growing interest for high-throughput systems based onshort-column LC coupled with MS, or even direct coupling of SPE and MS[30-39]. In such systems the extraction and detection should offer bothsensitivity and selectivity in order to be able to detect low quantities of analytesin biological fluids. It should be noted that many applications using little or noseparation prior to MS will 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 theanalysis of biological samples, and various reviews have appeared to which thereader can be referred [3-5]. Also, at-line systems, e.g. PrepStation [44,45] andASPEC [46,47], will not be considered here. With the latter systems theextraction is performed automatedly, but the eluate is collected in vials andsubsequently the eluate is, usually only partially, injected into the GC.Numerous applications have been reported about the usefulness of off-line SPEcombined with gas chromatography for analysis of biological samples.


28

However, on-line SPE-GC appears to be emerging as a rather unexplored, yetpromising technique.

The on-line coupling of SPE with GC implies injection of large volumesof solvent into the GC, thus requiring modification of the injection system. Anumber of interfaces have been proposed for this purpose [48], i.e.,(1) on-column injection, (2) loop-type injection and (3) programmedtemperature vaporiser (PTV). With on-column injection, solvent is introduced ata speed above the evaporation rate and at temperatures below the boiling pointof the solvent, ensuring wetting of the retention gap [49]. Solvent is evaporatedin the retention gap and eliminated via the solvent vapour exit (SVE). An extraretaining pre-column enables refocusing of the analyte prior to transfer to theactual analytical column. A second way to allow large-volume injection (LVI)is the loop-type interface, originally designed for on-line LC-GC [50]. Injectionis performed by filling a loop and flushing the contents to the retention gapinside the GC, which is slightly above the boiling point of the solvent.Formation of vapour results in pressure built-up and prevents furtherpenetration of the solvent into the retention gap. Finally, a PTV injector wasdesigned by Vogt et al. [51,52]. The injector strongly resembles a conventionalsplit/splitless injector. The main difference is injection of solvent attemperatures 30-40ºC below its boiling point on a packed liner. The linerpacking acts as a liquid reservoir. A high purge flow ensures evaporation of thesolvent, while analytes are retained on the liner packing. After almost completeevaporation of the solvent, the analytes are thermally released from the packingand transferred to the GC column. The latter is still at low temperatures,allowing refocusing of the analytes.

2.2.1.2 Applications of SPE-GC

A selective SPE-GC method was described by Farjam et al. [53], whocoupled immunoaffinity sample pretreatment with GC. A column withimmobilised antibodies was used for the extraction of ß-19-nortestosterone fromurine. A reversed phase reconcentration column and a retention gap were usedfor interfacing the extraction and the GC. Desorption from the antibody-columnwas performed with methanol-water (95:5, v/v), and after subsequent dilutionwith water, the analytes were trapped on the reconcentration column. Elutionfrom this column was performed with 75 µl ethyl acetate. The high selectivityduring trapping made it possible to analyse large urine samples (5-25 ml), withgood sensitivity for all investigated steroid hormones (limit of detection (LOD)about 0.1 ng/ml). The total analysis time was still 40 min, mainly due to thelong 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,


29

the diffused analytes were trapped on a PLRP-S pre-column. After drying,elution was performed with 275 µl ethyl acetate, which was injected into theGC via a loop-type interface. The SPE step in this procedure was more areconcentration step than an actual extraction process. Nonetheless, this on-linedialysis-SPE-GC system showed sufficient selectivity (Fig. 2). The authorsclaim that the benzodiazepines could be detected at therapeutic levels(5-25 ng/ml) and that extra selectivity could be obtained by acidification ofplasma prior to extraction (Figs. 2C and D), but these claims are notsubstantiated by their Figures.

Fig. 2: On-line dialysis-SPE-GC-NPD of (A) untreated blank plasma, (B) untreatedplasma spiked with 1 µg·ml-1 of nitrazepam, (C) acidified blank plasma and(D) acidified plasma with 1 µg·ml-1 of medazepam [54].

The PTV injector is an interesting injection system, as it allows LVI andthus on-line LC-GC and SPE-GC. The possibilities of PTV/GC in combinationwith SPE for plasma samples have already been demonstrated [55]. Now thatintegrated, automated instruments have become commercially available, on-lineSPE-PTV/GC will be facilitated. Moreover, the PTV injector also offerspossibilities for thermal desorption [10,56-60]. Thus, no solvent is introducedinto the GC, hereby eliminating some difficulties observed with LVI/GC, suchas introduction of large volumes of solvent and its evaporation within the GCsystem. However, the reports available so far all deal with environmentalsamples. Finally, the PTV injector has also been used for direct injection ofplasma samples [61]. In this set-up, acidification and subsequent ultrafiltration


30

were the only sample pretreatment steps. About 50 µl of the ultrafiltrate wereinjected onto a packed liner. No interference of the matrix was observed (seeFig. 3) and the chromatographic system was not damaged. The only drawbackswere occasional memory effects and the necessity to change the liner after 20injections. The latter is probably due to the injection of ultrafiltrates, which stillcontain some proteins, causing adsorption to the liner packing and GC column.The quantitation limit for ropivacaine was down to 300 pg/ml.

Fig. 3: Chromatograms obtained from extracted blank plasma and from plasma spikedwith 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) [61].

2.2.1.3 Remarks regarding the applicability of SPE-GC

New devices have been developed for the coupling of miniaturised SPEwith GC [62,63], enlarging the possibilities for incorporation of on-lineSPE-GC into routine analysis. Until now, the number of applications of on-lineSPE-GC in bioanalysis is very limited, in contrast to the numerous reports ofthis technique in the analysis of surface and drinking water [10-12,64,65]. Thisis probably due to the complexity of biological matrices in comparison with


31

water samples. Furthermore, various compounds in biomedical andpharmaceutical studies cannot be analysed with GC due to thermolability of thecompounds. Nevertheless, the applicability of SPE-GC with LVI, and inparticular the PTV injector, for biological samples seems worth furtherexploring.

2.2.2 Liquid chromatography – gas chromatography

2.2.2.1 General aspects of LC-GC

A similar approach to SPE-GC is the coupling of LC on-line with GC inwhich the LC column functions as a sample pretreatment technique. Only thefractions of interest will be transferred to the GC (heart-cutting). The LCcolumn is merely used for clean-up purposes and the GC column is used for theactual separation. As with on-line SPE-GC, on-line LC-GC also implies theintroduction of relatively large liquid volumes into the GC, so that LVI must beused. As discussed with on-line SPE-GC, several approaches have beenproposed in order to allow injections up to 1 ml into the GC. Nearly all on-lineLC-GC applications involve normal phase (NP) LC, since the introduction ofvolatile elution solvents into the GC is more easily achieved than that ofaqueous solvents [10]. However, direct analysis of biological, i.e. aqueous,samples in NPLC is not possible. Therefore, a separate sample pretreatmentstep, e.g., LLE or SPE, is always required. Consequently, no applications in thebioanalytical field have been reported for on-line NPLC-GC with directinjection of the sample.

It is more common to use reversed phase (RP) LC in biomedical andpharmaceutical analysis. Coupling RPLC with GC implies introduction of largevolumes of aqueous and ionogenic solutions into the GC. Water is verydisadvantageous for GC analysis due to its high boiling point, high surfacetension, poor wetting characteristics and aggressive hydrolytic reactivity,whereas non-volatile buffers (i.e. its ions) are also non-compatible with GC.Nonetheless, several techniques have been proposed for the analysis withon-line RPLC-GC [10-13,66-69]. Basically, they can be divided in two groups,(1) direct introduction of aqueous RPLC fractions using special GC-injectionsystems, and (2) phase-switching techniques, by which the analyte is transferredto an organic solvent prior to introduction into the GC. In the latter situationRPLC is coupled to GC via on-line LLE. Both approaches have found their wayin environmental analyis [11-13], but the number of applications in bioanalysisis very limited. In a few cases RPLC has been coupled on-line with GC for theanalysis 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.


32

2.2.2.2 Applications of RPLC-GC

An example of the first approach, i.e. direct introduction, was describedby Duquet and co-authors [72], who coupled µRPLC on-line with GC using anaminopropyltriethoxysilane-deactivated retention gap. Diazepam wasdetermined in urine by transferring only 2-µl methanol-water (80:20 v/v)fractions from the LC into the GC. Goosens et al. [73-75] also applied aretention gap to transfer eluents from the RPLC to the GC for drug analysis. Upto 200 µl of eluent (acetonitrile-water) could be introduced into a Carbowaxdeactivated retention gap by using an on-column interface and SVE [76]. Thepresence of remaining water after azeotropic evaporation was found todeteriorate the analysis. Therefore, prior to introduction into the GC, addition of10% acetonitrile to the LC eluent was performed resulting in an azeotropicacetonitrile-water mixture (84:16 v/v). By these means, the maximum amountof water remaining after evaporation was never exceeded. In order to injection-free fractions into the GC an anion-exchange micromembrane was insertedbetween the LC and GC parts [74]. Methanesulphonic acid was efficientlyremoved (99.9%) from the eluent acetonitrile-water, allowing the reproducibleanalysis of the potential drug eltoprazine, i.e. the coefficient of variation wasfound to be 3% (n=5, 150 µg/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 LLEinterface between the LC and GC part (loop-type interface). The set-up of thissystem is depicted in Fig. 4.

Fig. 4: Instrumentation for on-line coupled reversed phase LC-GC-MS [77].


33

Use of a GC-MS system allowed the identification of various unknownimpurities in pharmaceutical products. Even though the instrumental set-up israther complicated, reliable results were obtained for the quantitativedetermination of ß-blockers in human serum and urine [79,80]. A total analysistime of 45 min was required for the selective removal of the matrix compoundsand efficient and repeatable LLE-GC analysis. Hyötyläinen et al. [81] applied asimilar system to determine morphine and its analogues in urine. After LCseparation, a phase switch was applied using an elevated temperature for theeluent and the extraction coil, which resulted in increased recoveries. Afterphase-switching, the analytes were derivatised on-line with N,O-bis(trimethylsilyl)trifluoroacetamide prior to GC-flame-ionisation detection(FID) analysis. The total analysis time was less than 60 min.

Fig. 5: GC chromatograms of (top) urine spiked with the opiates and (bottom) blankurine. Peaks: (1) dihydrocodeine, (2) codeine, (3) ethylmorphine, (4) morphineand (5) heroin. Concentration of the analytes was 3 µg/ml [81].

As can be seen in Fig. 5, the LC clean-up procedure was effective as no matrixcompounds were observed. Extra peaks in the chromatogram are due to theexcess of derivatisation reagent. In principle this is a very powerful system.


34

However, although the above studies show interesting approaches, the utility ofthe resulting systems sometimes remain questionable. For example, in the finalapplication of morphine and analogues in urine, the concentrations of theanalytes are very high (3 µg/ml), and the absence of interferences is not clearlydemonstrated by Fig. 5. Furthermore, most of the analytes are metabolised priorto excretion into urine. It is thus more interesting to analyse the metabolites.Also, the analytes can be easily determined by SPE-LC-UV, allowing a fasterand easier analysis without the requirement of phase switching and an extraderivatisation step.

2.2.2.3 Remarks regarding the applicability of RPLC-GC

The results achieved with RPLC-GC are still not very remarkable, asanalysis times are long and poor detection limits (> µg/ml) are obtained. Thepossibility of LC-MS analysis without derivatisation and the availability ofsimilar but better sample pretreatment techniques such as SPE will mostcertainly imply that RPLC-GC and NPLC-GC will not be applied on a routinebasis.

2.2.3 Turbulent-flow chromatography

2.2.3.1 General aspects of TFC

Various techniques for SPE automation in combination with LC havebeen 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 significantlyreduced when chromatography is performed with high flow-rates, e.g., underturbulent flow conditions. The latter was introduced by Quinn and Takarewski[82] in 1997 as a fast method for sample analysis. In this approach typicalflow-rates of 3-5 ml/min are applied using a 1.0 mm i.d. column. These highflow-rates can be applied due to the low column back-pressure associated withthe use of large porous particles (typically 30-60 µm) [14-17]. The solvent frontprofile is shaped like a plug rather than a parabolic profile as observed underlaminar flow conditions. The high flow-rate and the plug flow profile increasethe effective diffusion rates within the pores of the stationary phase. The flowregime cannot be described by the Van Deemter equation. As a result, plateheights are significantly lower compared to predictions based on that equation.These conditions result in a considerable reduction of the chromatographicanalysis time [14-17,83-86].

The typical TFC procedure, usually applied for the direct-injectionanalysis of crude plasma, basically consists of four stages, similar to


35

conventional SPE: sample clean-up (extraction), analyte elution, LC separationand system re-equilibration [87]. It should be noted that the turbulent flowconditions 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 thepurification step the analytes have to be separated from the matrix and shouldbe retained by the stationary phase. Removal of plasma proteins from drugs isachieved by size exclusion (pore size 60 Å) and slow diffusion of proteins intothe pores. Since large particles are used, the technique also allows the use oflarge end-column frits (10-40 µm) [15]. As a result, large protein molecules inthe plasma sample can easily pass through these columns without clogging ofthe frits due to precipitation. Elution is performed using a steep gradient oforganic solvent followed by chromatographic separation on a second column.Only limited separation is achieved on the analytical column due to the highpercentage of organic solvent. Finally, the extraction and separation columnsare equilibrated with suitable solvents for subsequent analysis. A basic set-up ofTFC is depicted in Fig. 6.

Fig. 6: Schematic representation of the flow path during the on-line extraction, elutionand equilibration [91, modified].

As described above, TFC strongly resembles SPE, with the use of highflow-rates instead of normal flow-rates, and larger particles in the extractioncolumn. As a consequence, the set-up for TFC (Fig. 6) is similar to that for


36

SPE-LC systems. TFC is applicable for the analysis of drugs that bind to a highextent to proteins. In general, recoveries of 70-100% are obtained [87-89]. Onlywith 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 toextraction [15].

2.2.3.2 Applications of TFC

The first application of TFC-MS for direct-injection analysis of plasmawas reported by Ayrton et al. [85] in 1997. In successive years, moreapplications have been developed, mainly for the analysis of low-molecularweight 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 TFC-LC-MS/MSsystem with a detection limit of 1 ng/ml.

Most applications of TFC focus on the determination of a singlecompound in plasma per run. Jemal et al. [92] reported the simultaneousdetermination of simvastatin and simvastatin acid in human plasma bydirect-injection LC-MS/MS. The possibilities for multi-componentdetermination in a single run were further explored by Wu and co-authors [15].This study was set up for high-throughput pharmacokinetic screening usingLC-MS/MS and a turbulent-flow column-switching system by which tencompounds had to be analysed simultaneously. The set-up of the system wassimilar to the design presented in Fig. 6. A 4-ml/min flow was used for 1 minduring which sampling and purification was performed. Then, the trappedanalytes were eluted in a back-flush mode from the extraction column towardsthe analytical column using a flow of 0.4 ml/min. Elution was completed within2 min. Flushing and equilibration of the extraction column was performedduring the separation of the analytes on the analytical column. Good separationand peak shapes (Fig. 7) were achieved within a run time of 10 min includingthe extraction time. A dynamic range of 1-2500 ng/ml was obtained, with alimit of quantitation (LOQ) of 1 ng/ml. Using the highest concentration, i.e.2500 ng/ml, a carry-over of 0.14±0.07% was observed. One extraction columncould be used for 200-300 plasma sample injections without causing significantback-pressure increase. It should be noted, however, that the simultaneousanalysis of the analytes in this study is of limited interest for pharmacokineticscreening. Instead, analytes and their metabolites or co-administered drugsshould have been chosen as target compounds.

A ternary-column system was introduced by Xia et al. [88] forhigh-throughput direct-injection analysis of plasma. Basically, the systemconsisted of two extraction columns in parallel and one analytical column. Inthis way, one column was equilibrated while on the other column the extraction


37

of 10 µl plasma was performed. Thus, the equilibration step does not add extratime to the injection cycle time. The on-line purification step lasted for only0.3 min and the total run time was 1.6 min. The extraction recovery of theguanidine-type drug was >95%. Using the sensitivity and selectivity of the massspectrometer by operating it in the selected reaction monitoring (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.

Fig. 7: Total ion chromatogram and MRM chromatograms of the LC-MS/MS assayfor the ten compounds with 100 µl direct injection of a 250 ng/ml chimpanzeeplasma standard. A 2.0 × 150 mm C18 column (Symmetry, Waters Corp.) wasused as the analytical column, which was operated in the gradient mode with aflow-rate of 0/4 ml/min [15].

Ayrton and co-authors [90] applied ultra-high flow-rate capillary LC withMS/MS (SRM mode) for the direct determination of an isoquinoline drug inplasma. The extraction column had an internal diameter of 0.18 mm, thusallowing a flow-rate of 130 µl/min for turbulent flow. Upon injection of only2.5 µl plasma (diluted 1:1 with an aqueous standard solution), an LOQ of0.5 ng/ml was obtained within a total run time of two min.


38

2.2.3.3 Remarks regarding the applicability of TFC

Comparing the potential of TFC with results obtained with LLE andautomated 96-well SPE, it can be noted that similar results are obtained in termsof dynamic range, LOQ, accuracy and precision [15,83,85]. As turbulent flowsreduce time-consuming steps, the speed of the system is superior to moreconventional on-line SPE systems. The limited concentration power is adisadvantage of TFC. Although this can be compensated by using capillary LC,the load capacity is then reduced, allowing the injection of less sample. The rateof solvent consumption of TFC is high in comparison to conventional LC.However, the total volumes per analytical run are similar for turbulent-flowLC-MS and for SPE followed by conventional LC-MS [15]. Up till now, TFC isnot rapidly expanding, probably due to the mainly size-exclusion basedprinciple for sample clean-up. As a result, virtually all applications are analysesof plasma or similar fluids, since the analytes of interest can be easily separatedfrom proteins and other large matrix compounds. Handling a urine matrix ismuch more complicated as small drugs cannot be easily separated from smallurine matrix components. Even though the speed is advantageous, TFC willprobably not become a universal and widely applicable sample pretreatmenttechnique.

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 TFCshould only be used when the separation is performed under turbulent flowconditions. Furthermore, it should be noted that the sample pretreatment isactually LC modified into an SPE procedure. This can also be seen in Fig. 6, asthe set-up for TFC is similar to SPE-LC. The technique might also be referredto as modified (ultra-high flow) LC-LC with subsequent detection. Anothercritical point was raised by Ayrton et al. [86]. Careful consideration ofcommonly used flows (3-5 ml/min) and particle sizes (30-60 µm) showed thatthe actual flows are not turbulent. Therefore, the term ‘ultra-high flow-rate LCwith direct sample injection’ might be preferred, as is also used now in othersystems [34].

Diffusion-based sample preparation techniques

39

2.3 Diffusion-based sample preparation techniques

2.3.1 Solid-phase microextraction

2.3.1.1 General aspects of SPME

At the introduction of SPME [96], it was designed for gaschromatographic analysis by direct sampling of liquids. SPME integratessampling, extraction, concentration and sample introduction into a singlesolvent-free step [21,97]. Originally, SPME was performed with a modifiedsyringe with a stainless steel needle in which a thinly-coated fused silica fiber(100 µm in diameter) could be moved up and down via the plunger [19,20]. Thefiber is coated with a suitable stationary phase, which is usually a polymericphase of 7- to 100-µm 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 notinserted into the sample itself, relatively dirty samples can be analysed whileobtaining clean extracts. Another advantage is the relatively high speed, i.e.short equilibrium times in comparison to DI/SPME. Furthermore, moreaggressive sample preparation can be applied, e.g., extremely low or high pHvalues, without the risk of damaging the coating [21]. After equilibrium or awell-defined time, the fiber is transferred to undergo liquid desorption, usuallyfollowed 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 theanalyte between the stationary phase and the matrix. As a result, SPME is anon-exhaustive extraction method. The amount of analyte extracted (n) isproportional to the initial sample concentration (C0) and sample volume (Vs),the volume of the stationary phase (Vf), and the fiber coating/sampledistribution constant (Kfs) [22,98]:

n = (Kfs×Vf×Vs×C0)/(Kfs×Vf+Vs) (1)

Extensive studies have been performed to describe the theoreticalfundamentals of SPME [20,97,99,100,101]. Basically, the processes in SPMEcan be divided in thermodynamics and kinetics [20,102]. Because of thephysicochemical properties of, for example, polydimethylsiloxane (PDMS,glass transition temperature -123 to -126ºC [20,99]), which is the mostcommonly used phase in SPME, the extraction may be described asliquid-liquid extraction. Octanol-water partitioning coefficients (Kow) may beused to estimate the extraction behaviour of an analyte towards PDMS,although it should be noted that octanol and PDMS differ severely with regard


40

to chemical properties. PDMS is generally considered to be an absorptive phase[103], although adsorption effects may occur as well [100,104,105]. Thediffusion coefficients of the analyte towards the coating and in the coatingdetermine the time that is required to reach equilibrium. Agitation will increasethe kinetic processes since diffusion in the sample is no longer limiting. In anagitated solution the diffusion towards the stationary phase through a staticwater layer around the fiber is the limiting process in DI/SPME [98,100,106].

Various factors can influence the extraction and desorption efficiency. Ifavailable, the coating of the fiber is usually chosen based on the principle ‘likedissolves like’. Adjustment of the pH of the sample may affect the yieldbecause non-ionised species are better absorbed by the commonly used fibercoatings. Addition of salt increases the ionic strength and favours salting-out,which often results in an increase of the yield. Increasing the sorptiontemperature has a two-fold effect. On the one hand, diffusion coefficients arehigher at higher temperatures, thus leading to a decrease in time to reachequilibrium. On the other hand, higher temperatures lead to lower partitioncoefficients in the stationary phase, thus decreasing the extraction yield [99].During desorption with an aqueous solvent, a pH shift may be applied to causeionisation. Also, the addition of a suitable organic solvent may be helpful tospeed 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 pesticideresidue analysis [108,109] with SPME-GC. Nowadays, many applications ofSPME-GC with either HS/SPME or DI/SPME have been described in variousfields, like the analysis of food [108,110,111], explosives [112], biological andpharmaceutical samples [18-20,108,113]. For blood samples, it is possible todetermine the degree of protein-binding of drugs, since SPME usually extractsthe free fraction of the drug only [114,115]. Hence, the protein binding caneasily be determined if the equilibrium between bound and free drug is notdisturbed in the SPME procedure. Another possibility for the estimation of thedegree of protein binding is to dilute the sample and determine the extractedamount for different dilution factors [116]. When, on the other hand, DI/SPMEis applied to urine, matrix interference may be quite severe.

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


41

2.3.1.2 Applications of SPME-LC

In 1995 the first paper on SPME-LC appeared [122]. Basically, twoformats of SPME have been coupled to LC, i.e., fiber-based SPME and in-tubeSPME. Fiber-based SPME is similar to that applied for GC analysis. Only aspecially designed LC desorption chamber [122,123], which acts as achromatographic tee with a six-way valve, is required to allow desorption withsolvent. This chamber can be simply interfaced with the LC. The extractionprocess is mainly passive though agitation and/or temperature elevation may beapplied. With in-tube SPME the sample is aspirated and pushed back by meansof a syringe through a capillary coated with the stationary phase, thus creatingan active extraction [124].

Fig. 8: Instrumental set-up of the new on-line SPME-HPLC interface based on anin-tube SPME capillary technique. A piece of GC column (in-tube SPME)hosts in the position of the former needle capillary. The aqueous sample isfrequently aspirated from the sample vial through the GC column anddispensed back to the vial (INJECT position) by movement of the syringe.After the extraction step, the six-port valve is switched to the LOAD positionfor the desorption of the analytes form the in-tube SPME by flushing 100%methanol from another vial through the SPME capillary. The volume istransferred to the loop. After switching the Valco valve to the INJECTposition, an isocratic separation using a mixture of 60:40 acetonitrile/water wasperformed. A detailed view of the in-tube SPME capillary is included at theleft side of the figure [124].


42

This aspiration-push back sequence can be repeated at will. The set-up (Fig. 8)is similar to open-tubular SPE and has been known for many years. Anoverview of bioanalytical applications using SPME-LC is presented in Table 1.

Table 1: Bioanalytical applications of SPME-LC.

Compound Sample Type§ System Sorption;desorptiontime (min)

Yield (%)a LOD(ng/ml)a

Ref.

Amphetamines Urine In-tube LC-ESI/MS 11; 4 81-98c 0.38-0.82 [125]Amphetaminesß-Blockers

UrineSerum

In-tube LC-ESI/MS 11; 4 0.7-16 0.1-1.2 [126]

Antidepressants(tricyclic)

Urine Fiber µLC-UV 180; 30 N.D. 3-40 [127]

Antidepressants(tricyclic)

Urine In-tube µLC-UV 10; 5 N.D. N.D. [128]

Benzodiazepines UrineSerum

In-tube LC-ESI/MS 11; 4 3.0-9.2 0.024-2.00 [129]

Benzodiazepines Urine Fiber µLC-ESI/MS 180; 30 N.D. 1-6 [130]Brombuterol +analogues

Urine Fiberb LC-ECD 45; 2×5 45 10 [131]

ß-Blockers UrineSerum

In-tube LC-ESI/MS 11; 4 88-110c

70-109c3-36#

9-43#[132]

ß-Blockers UrineSerum

In-tube LC-ESI/MS 10; 4 84-113c

71-112c0.1-1.2 [133]

Carnitines Urine Fiber ESI/MS 60; 2 N.D. 0.2-12.8 [134]Corticosteroids +conjugates

Urine Fiber LC-MS 15; 5 5-45 4-30 [135]

Flavonoids Urine Fiber LC-ESI/MS 10; 6 56-86c 2.7-25.4± [136]Lidocaine Urine Fiber LC-UV 45; 2×10 22 25 [107]Lidocaine Urine Fiber APCI/MS2 5; 4 6.5 0.40 [104]Phenothiazines Urine

BloodFiber LC-MS2 60; 10 4-40

0.0002-0.1204-22±

0.2-200[137]

Ranitidine Urine In-tube LC-ESI/MS 11; 4 60 1.4 [138]N.D.: no data available.§: Fiber-based applications all apply direct immersion.#: Limit of quantitation.±: pg/mla: unless stated otherwise.b: molecularly imprinted polymer (MIP) coating.c: relative to aqueous samples.

Since the first application of fiber-based SPME combined with LC, somereports [107,127,130,131,135-137] have been made on the applicability to thefield of bioanalysis. Volmer et al. [135] were the first to report the combinationof SPME and LC-MS (SIM mode) in the bioanalytical field. The method was


43

developed to analyse eleven corticosteroids and two steroid conjugates in urine.Sorption times of 10 to 60 min were observed, and only 5 min desorption wassatisfactory with regard to carry-over. The detection limits were 4-25 ng/ml dueto low yields (1.7-15%).

Fig. 9: (A) SIM of SPME-LC-MS for the eleven phenothiazine derivatives extractedfrom human whole blood and urine. The amounts of each drug spiked into 1 mlwhole blood and urine were 0.5 µg and 0.5 ng, respectively. The arrows showthe locations where drug peaks should appear. The left vertical axes representthe relative percentage intensity of the peaks, the right vertical axes show theabsolute intensities of peaks recorded by the mass spectrometer and thehorizontal axes indicate the LC running time (min). (B) SRM withSPME-LC-ESI/MS/MS for the eleven phenothiazine derivatives extractedfrom human whole blood and urine. The amounts of each drug spiked into 1 mlof whole blood and urine were 0.5 µg and 0.5 ng, respectively. Axes as in (A)[137].

A

B


44

Kumazawa and co-authors [137] described the determination ofphenothiazines in whole blood and urine by fiber-based SPME and LC-MS/MSwith electrospray ionisation (ESI), showing the problem of low yields versusthe potential for selective and sensitive analysis. Before SPME was performed,whole blood was deproteinised by adding perchloric acid, centrifuged, and thepH was adjusted to about 8. With urine samples only pH adjustment wasperformed. A sorption time of 60 min was applied, which ensured thatequilibrium had been reached for all compounds, and subsequently, desorptionwas performed within 10 min. Even though the SIM mode was applied duringsingle-MS analysis, severe matrix interference was observed for both urine andwhole blood (Fig. 9A). Applying MS/MS (SRM mode) eliminated thenoticeable interference (Fig. 9B). The extraction yields were very low for wholeblood (0.0002-0.12%), resulting in detection limits of 0.2-200 ng/ml. In case ofurine, 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 ofthe extraction as claimed by the authors. It is more likely that a co-precipitationof proteins and bound drug due to acidification with perchloric acid occurred.

With the introduction of in-tube SPME [124] it became possible toautomate the analytical procedure, and consequently, a number of applicationsin the bioanalytical field have been reported [125,126,128,129,132,133,138].Most applications show great similarity, and therefore only one application willbe discussed here. The results of other in-tube SPME-LC studies aresummarised in Table 1. Kataoka et al. [126] developed an automated in-tubeSPME-LC-ESI/MS method for the simultaneous determination offive amphetamines and nine ß-blockers in biological samples using single MS(SIM mode). Applying a flow-rate of 100 µl/min proved to be optimal [22,124],and varying the number of draw/eject cycles of 35-µl samples resulted in theextraction-time profile (Fig. 10). As can be seen, equilibrium was not reachedafter 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 aswith fiber-based SPME, i.e. extraction from urine resulted in higher yields thanextraction from serum. Worth mentioning is the differences in extraction yieldand kinetics of drugs within the same class. This phenomenon has also beenobserved with fiber-based SPME. This may be explained by differences indiffusion coefficients of the drugs, but this is not expected. Surprisingly, theoptimal pH during sorption was 8.5, ensuring that all compounds (pKa 9-10)were protonated to a large extent. This is contrasting to most applications inwhich non-protonated analytes were more easily absorbed than the protonatedspecies. An explanation of the use of this pH value might be that the capillarycoating was stripped at pH 10, resulting in lower yields than with pH 8.5.


45

Fig. 10: Extraction-time profile of drugs with Omegawax 250 capillary. SPMEconditions: drugs, 0.5 µg/ml; sample pH, 8.5 (Tris-HCl); draw/eject cycles, 15;draw/eject volume, 30 or 35 µl; draw/eject rate, 100 µl/min; desorption, mobilephase. AM = (+)-amphetamine, MA = (+)-methamphetamine,MDMA = (±)-3,4-methylenedioxymethamphetamine [126].

2.3.1.3 Applications of SPME-MS

SPME is hardly ever used in high-throughput systems due to therelatively long sorption and desorption times. Even when SPME is followed bychromatographic separation, the sorption and desorption are usually thetime-limiting steps. This also holds for fiber-based SPME systems and/orsystems in which no separation is performed, i.e. the desorption device isdirectly coupled to the detector. Pérès and co-authors [139] coupled HS/SPMEdirectly to MS by thermal desorption and a short GC transfer-line for thedetermination of volatiles in cheese using only 10 min sorption. A typicaldesorption profile of HS/SPME-MS is shown in Fig. 11. Orzechowska et al.[140] determined cocaine by direct coupling of HS/SPME with ion mobilityspectrometry. Carnitine, an essential factor in the fatty acid metabolism oforganisms, was determined by SPME-ESI/MS [134]. Severe matrix interferenceduring detection and a long sorption time were the major drawbacks of thismethod.


46

Fig. 11: (a) Desorption peak of a sample of cru2 Camembert cheese obtained bySPME-MS. The extraction was carried out by placing an SPMECarboxen/PDMS fiber in the headspace at 20ºC for 10 min. (b) Averagespectrum of 0-3 min [139].

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 andreproducibility of the method. Ai [141] described a theoretical model forfiber-based SPME allowing quantitation before equilibrium is reached. In achallenging practical application, Van Hout et al. [104] performed ultra-rapidnon-equilibrium fiber-based SPME and coupled it directly to atmosphericpressure chemical ionisation (APCI)/MS/MS for the determination of lidocainein urine. Direct immersion of the 100 µm PDMS fiber in urine was applied for 5min. Subsequently, the fiber was desorbed for 4 min, after which 1 min analysiswas performed. No matrix interference was observed. The method had not yetbeen optimised for quantitation, but initial results on repeatability andreproducibility 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. Thisapproach clearly showed that the long equilibrium times of SPME aredisadvantageous for rapid analysis, but that use of non-equilibrium SPME canhelp to overcome this problem.


47

2.3.1.4 Remarks regarding the applicability of SPME

SPME has clearly demonstrated its utility for bioanalysis, and inparticular for the analysis of urine samples. The number of applications for theanalysis of plasma, serum and blood samples is more limited. So far, no reportshave been found about ion suppression in SPME-LC-MS systems. This mightbe due to the clean extracts that can be obtained with SPME due to the lowyields for both analytes and matrix compounds and differences in diffusioncoefficients. However, it may also be due to the lack of attention that was paidto these effects.

A major drawback of SPME is the small number of commerciallyavailable stationary phases, especially for fiber-based SPME, hereby limitingthe choice for selectivity. The main commercially available coatings forfiber-based SPME are PDMS, polyacrylate and some mixed phases, e.g.,PDMS/carboxen, PDMS/divinylbenzene (DVB), Carbowax/DVB andCarbowax/templated resin. From this list it is obvious that the choice forselective extraction is limited. Moreover, ion exchange phases are not available.For in-tube SPME, a piece of capillary GC column is commonly used, thusproviding more diversity in stationary phases than in fiber-based SPME. Inaddition, phases for more polar analytes are available. Yet, with thedevelopment of more stationary phases for both fiber-based SPME[131,142-144] and in-tube SPME [132] the potential of SPME may beimproved even further.

The advantage of in-tube SPME over fiber-based SPME is the shortersorption times, as can also be seen in Table 1. This is due to the fact that in-tubeSPME is usually stopped before equilibrium is reached. However, as describedabove, non-equilibrium fiber-based SPME can be performed as well. With thelatter, transfer of the fiber with the trapped analytes to the desorption chamber isrequired after sorption, which is not necessary with in-tube SPME. The lattersystem also allows the use of 100% organic solvent for desorption of theanalytes. The use of high percentages of organic solvent is more complicatedwith fiber-based SPME, since the coating can be stripped from the silica.Samples containing particular matter should be filtered prior to extraction within-tube SPME to prevent clogging of the capillary, which cannot causeproblems in fiber-based SPME. With regard to automation, in-tube SPME hasmore to offer at this moment. Automation is more easily established with thissystem, since no transfer of the stationary phase to a desorption chamber isrequired. The availability of an autosampler for use with fiber-based SPME-LCis still being awaited [21]. Thus, both fiber-based and in-tube SPME have theirown advantages and disadvantages. To some extent, the systems can beconsidered complementary and the best choice will depend on the sample andon the particular requirements of the individual applications. A recently


48

introduced SPME device is the stir bar sorptive extraction [145], which has alarger volume of stationary phase than fiber-based SPME and in-tube SPME.Although it has not been applied for bioanalysis, it has already shown itspotential for the coupling with LC and determination of (semi-) volatiles inaqueous samples.

2.3.2 Membrane-based sample preparation techniques

2.3.2.1 General aspects of membrane-based techniques

As an alternative to SPE and/or LLE, membranes may be used as asample preparation technique [23,25,146-153]. When using the latter, it isessential to differentiate between porous and non-porous membranes. Samplepretreatment with porous membranes is based on the principle of size exclusionto differentiate between substances, whereas non-porous membranes utilise thedifference in partition coefficients of substances, thus being an actual extractiontechnique. An overview of the various techniques is given in Table 2. Theporous membrane techniques (PMTs) drew major attention in the late 80’s andearly 90’s, whereas the applicability of non-porous membranes for samplepretreatment in the biomedical field is being explored more recently.

Table 2: Overview of membrane-based sample preparation techniques.

Technique Membrane

type

Principle Driving force Mainly

combined

withDialysis Porous Size-exclusion Concentration

differenceLC

Electrodialysis Porous Size-exclusion andselective iontransport

Potentialdifference

CE

Filtration Porous Size-exclusion Pressuredifference

LC

Membraneextraction

Non-porous Difference inpartition coefficient

Concentrationdifference

LC, GC, CE

In PMTs, the liquids on each side are physically connected through pores.Transport through the membranes is based on size-exclusion, i.e., sufficientlysmall molecules can permeate through the pores, whereas larger moleculescannot. This can result in an efficient clean-up from large matrix molecules, butno discrimination can be made between small molecules. The latter is onlypossible to some extent with electrodialysis, for which an ion-exchangemembrane is used. Now, large molecules and molecules with a given charge


49

will be excluded. Strictly taken, PMT is not an extraction technique, but afiltration process. Extensive descriptions of the principles of porous-membranetechniques were given by Van de Merbel [23] and co-authors [24,25], and willtherefore not be discussed here.

Non-porous membrane techniques (NPMTs) employ an organic orpolymeric (solid or liquid) layer, placed between two other liquid phases. Theanalyte must actually be extracted from the donor phase, dissolve into themembrane in order to be able to pass through, and then be released in theacceptor phase. The behaviour of the analytes largely depends on partitioncoefficients between the different parts of the membrane system. Only analytesthat are easily extracted from the donor phase and, in addition, are easilyreleased 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 isthus possible to separate molecules of similar size, yet with differentphysicochemical properties [146,152]. The non-porous membrane techniquecan be subdivided into four main groups: (I) Supported liquid membraneextraction (SLME), (II) Microporous membrane liquid-liquid extraction(MMLLE), (III) Polymeric membrane extraction (PME) and (IV) Membraneextraction with a sorbent interface (MESI). SLME is the most widely usednon-porous membrane technique [146], but various applications of MMLLE,PME and MESI have been reported as well [23,25,146-150,152]. It should benoted that MMLLE is considered to be NPMT, even though a membrane withmicropores is used.

All NPMTs utilise a membrane unit constructed from two blocks of inertmaterial with a machined groove in each. A membrane is placed between theblocks and the total unit is clamped together. Hence, two flow channels areformed, one being the donor channel, the other being the acceptor channel. Inprinciple, SLME utilises a pH shift between the donor phase, in which theanalyte 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 themicropores of the organic membrane, and can therefore be compared with asingle liquid extraction. MMLLE is mainly used for the analysis of hydrophobiccompounds that cannot be extracted from an organic membrane into an aqueousacceptor solvent, as is the case with SLME. PME is similar to SLME, with theexception that a polymeric membrane is used. Due to this membrane it is alsopossible 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 toslow extraction. MESI differs from the previous techniques in that a solidpolymeric membrane is used. MESI was mainly developed for the combinationwith GC, thus in order to use a gaseous acceptor phase [146,148,152], while the


50

donor phase is aqueous or gaseous. Obviously, MESI works best for theanalysis of volatile and relatively non-polar compounds [152]. Mostapplications of MESI are in the environmental field for the analysis of aqueoussamples [146,154-156].

Both PMTs and NPMTs usually use the terms efficiency and/orenrichment. The efficiency is defined as the ratio between number of moles putinto the system during the extraction and the number collected in the acceptor,and can be directly measured [150]. Efficiency should not be confused with theterm 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 tobecome 100%, because of time dependence. It is obvious that efficiency may besacrificed for speed if sensitivity is not of major concern. As a result, efficiencyis not always constant, as the time for sample preparation can be varied.Moreover, various factors, e.g., the composition of the donor phase, acceptorphase, the membrane and the sample, can affect the efficiency of the system.The most often observed side effects, i.e., binding to matrix proteins andadsorption to the membrane, with consequent carry-over, have been describedin various studies and these effects will be pointed out later in this chapter.Besides efficiency the term enrichment is also often mentioned, especially withPMTs. Enrichment is the accumulated amount of analyte in the acceptor phaseduring a given time. In membrane techniques the efficiency decreases withincreasing donor-flow. Contrary, the enrichment increases with increasingdonor-flow (Fig. 12).

Fig. 12: Extraction efficiency E and enrichment factor Ee (arbitrary units) as functionsof the reduced flow parameter ϕ. Note: ϕ is the volumetric flow divided by themembrane area [150].


51

At very low donor-flows the enrichment is close to zero, as the extractedanalyte is diluted immediately in the acceptor phase. With high donor-flows, theefficiency is decreasing due to incomplete diffusion of the analyte into theacceptor phase, but the enrichment is increasing with increasing donor-flow.However, high donor-flows imply large consumption of sample, and cantherefore only be applied if sufficient sample is available [146,150].

2.3.2.2 Porous membrane techniques

As shown in Table 2, three major types of PMTs can be distinguished.Although a porous membrane is also used in microdialysis, the latter is a ratherdifferent technique, being mainly applied for in vivo studies [157-162] and willtherefore not be discussed here. The applications of on-line filtration are limitedto fermentation broths [23]. Electrodialysis has been coupled on-line to CE forthe determination of inositol phosphates in plasma [163-165]. Only one report isavailable for the on-line coupling with LC [164]. Despite a high efficiency(95%), ephedrine could only be determined in serum three times before themembrane became too highly contaminated. The studies on electrodialysis weremerely exploratory and no applications in routine analysis have been reportedyet. The most common PMT is dialysis, employing a cellulose-based membraneand an aqueous donor and acceptor phase. Though some reports have beenmade 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. Anoverview of the results obtained with the latter system is presented in Table 3.

The major disadvantages of dialysis are the typically low efficiency andthe 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 beenperformed [177,182]. Furthermore, the use of membranes might result inadsorption of the analytes by the membrane [182]. Covering of the active sitesby adding a surfactant can minimise adsorption [192,196]. The binding ofanalytes to macromolecules in the matrix such as proteins has an even morepronounced effect on the dialysis efficiency. Only the unbound fraction candiffuse through the membrane. As a consequence, the efficiency may besubstantially lower than with aqueous solutions. This will especially be the casefor compounds that bind to a high degree to proteins. This allows thedetermination of the free fraction of analyte [174,176,185,188]. A number ofways have been suggested to release the bound drug from the protein. Thesimplest way to reduce protein binding is to dilute the sample. Anotherpossibility is to change the pH of the sample, hereby changing the structure ofthe macromolecule and/or the charge on the analytes, causing the release of theanalytes of interest [174-176,184,187,192]. Denaturation of the proteins can be


52

performed by addition of a strong acid. However, this may also result in loss ofbound drug that co-precipitates with proteins. A more selective and elegantstrategy for releasing the analytes from the proteins is the addition of competingagents or displacers for the protein binding sites [24,174,176,180,187], e.g.,fatty acids with appropriate chain length [187]. The latter showed to be efficienteven for highly protein-bound compounds.

Table 3: Applications of dialysis on-line with LC in bioanalysis.

Year Compound Sample Detection Dialysistime(min)

Efficiency(%)

LODa Ref.

1985 Amino acids Serum Flu 1 2 6 nM [169]1985 Barbiturates Serum UV 2 30 N.D. [170]1986 Enoximone +

metaboliteSerum UV 4 17

481015

[171]

1986 Salicylic acid Serum Electr.± 2 86-105 N.D. [172]1987 Mitomycin C Plasma

UrineUV 4 25 1 [173]

1987 AnticonvulsantsTheophylline

Serum UV 3 50 50-170 [174]

1988 Corticoids Serum UV 3 85-90b 30 nM [175]1988 Phenobarbitone

PhenytoinPhenylbutazoneTheophyllineWarfarin

PlasmaSerum

UV 12 68 N.D. [176]

1990 Azidothymidine Plasma UV 17 40 20 [177]1990 Des-enkephalin-γ-

endorphinPlasma Flu 15 25 10 [178]

1991 Oxytetracycline Blood§

PlasmaUV 7.3 60 50 [179]

1991 Oxolinic acidFlumequine

Blood§

PlasmaUV 7.3 60-69b 50 [180]

1992 Iopentol PlasmaBlood§

UV 7.15 5047

500c [181]

1992 Pholcodine PlasmaBlood§

Flu 8.25 60 40 [182]

1992 Benzodiazepines Plasma UV 7.6 37-50 20-25 [24]1992 Rogletimide Plasma MS/MS 14 33 5 [183]1993 Antiviral drug* +

metabolite#Plasma UV <20 10-20 0.2

nmol/mlc[184]

1993 Phenytoin Plasma UV 7.5 3 0.6 µg/ml [185]1994 Antiviral drug* +

metabolite#Urine UV 4 20 2.5 nmol/lc [186]

1995 NSAIDs Plasma UV 9.6 40-65 0.01-2µg/ml

[187]


53

Table 3: Applications of dialysis on-line with LC in bioanalysis.(cont.)

Year Compound Sample Detection Dialysistime(min)

Efficiency(%)

LODa Ref.

1995 Antiepileptics Plasma UV 10 3-10 (total)22-23 (free)

0.1-0.8µg/ml

[188]

1996 Verapamil +Norverapamil

Plasma Flu 10 7376

1.31.4

[189]

1997 Sildenafil (Viagra)+ metabolite

Plasma UV 6 30 1.00c [190]

1997 Levosimendan SerumPlasma

UV 6.6 40b 5c [191]

1997 Clozapine Plasma UV 8.5 50 15 [192]1997 Oxprenolol Plasma UV 10 81 17 [193]1998 Lamotrigine Serum UV N.D. 97 1 µM [194]1998 Amphetamines Plasma

SerumFlu 6 52-54 10 [195]

1998 Antidepressants Plasma UVFlu

12.8 52-65 5-12 [196]

1999 Quinolones Serum Flu 5-15 N.D. N.D. [197]2000 Iodixanol Plasma UV 5.25 55 130 [198]2000 Albendazole +

metabolites (2)Plasma UV 6.5 65-70 2 [199]

2001 Local anaesthetics Plasma UV 11.8 67-72 11-15 [200]2001 Sotalol Plasma UV 8 60 2.5 [201]N.D.: No data available.Flu: Fluorescence detection.a: ng/ml unless stated otherwise.b: relative to aqueous samples.c: Limit of quantitation (LOQ).

§: Whole blood.±: Salicylate electrode.*: 1-(ß-D-arabinofuranosyl)-5-(1-propynyl)-uracil#: 5-propynyluracil

When the dialysis time is in the same order as the analysis time of aconventional LC separation (10-20 min), serial analysis of samples can beachieved, thus preparing one sample while analysing another. However, inbioanalysis conventional LC is more and more replaced by LC-MS/MS.Consequently, the separation time is decreasing, since the MS will make up fora lower LC separation efficiency, allowing analysis times of just a few minutes.Therefore, dialysis is beginning to disappear as a sample preparation techniqueand 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 bioanalyticalfield.


54

2.3.2.3 Non-porous membrane techniques

NPMTs show a more versatile use in on-line coupling to separationtechniques. Besides the more common coupling to LC [152,203-207], GC[152,208-212] and CE [152,213-216] (see also Table 2), NPMTs have also beencoupled to atomic absorption spectrophotometry [217-219], electrochemicalinstruments [220,221] and flow-injection systems with UV detection [222,223].Both PME and MESI have not been applied to the analysis of biologicalsamples, and therefore, these techniques will not be discussed in more detail.SLME and MMLLE are already applied to bioanalysis and their applicationswill be discussed below.

The advantage of NPMTs in comparison to PMTs is the higher degree ofselectivity for any type of sample. High enrichment factors can be achievedsimultaneously with high selectivity. Another advantage is the relatively smallsolvent consumption in comparison with other sample preparation techniques.As shown above, the possibility of on-line coupling to various analyticalinstruments and the ease of automation make SLME and MMLLE attractive forbioanalysis. As with PMTs, protein binding can decrease the extractionefficiency. Other critical factors are the short-term and the long-term stability ofthe membranes. Especially for SLME and MMLLE, the pressure differencesover the membrane must be low enough to hold the organic solvent in the poresof the hydrophobic membrane and to prevent the acceptor phase from leakinginto the donor phase and vice versa. Also, the chemical stability may be critical,especially if more polar membranes are used. Carry-over effects are usuallyovercome by appropriate washing of the membrane. Finally, the extractionprocess of SLME and MMLLE is slower than conventional techniques like SPEand LLE, thus limiting the throughput of samples. However, use of parallelmembrane 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 SLMEand MMLLE is to provide selectivity in the extraction of drugs from thecomplex biological matrix. An example of the selectivity that can be achievedby SLME was presented by Jönsson et al. [204]. SLME employing a porouspoly(tetrafluoroethene) (PTFE) membrane was followed by ion-pairchromatography with variable wavelength UV detection. The membrane wassoaked in the membrane liquid (10% tri-n-octyl phosphine oxide in di-n-hexylether) for 30 min. The pH of the donor phase was adjusted to 9.5, whereas theacceptor phase was an acidic buffer at pH 2.5. As can be seen in Fig. 13, nointerfering peaks were observed from the urine matrix. This allowed simpleisocratic chromatographic analysis, and an LOD of 2-18 nM was achieved forropivacaine and its metabolites. The extraction time was similar to thechromatographic run, thus allowing sequential sampling and analysis. The total


55

enrichment factors were 6-136 with efficiencies of 3-73%. Polar compoundswere less efficiently extracted due to the apolarity of the membrane.

Fig. 13: Chromatograms of a water solution (a) and a urine sample (b), both spikedwith 3-OH-PPX (1; 1.0 µM), 4-OH-Ropivacaine (2; 0.80 µM),3-OH-Ropivacaine (3; 0.83 µM), PPX (4; 1.0 µM), Iso-PPX (5; 0.84 µM) andRopivacaine (6; 0.90 µM) [204].


56

Lindegård and co-authors [205] used SLME-LC-UV (detection at265 nm) to determine amperozide in plasma. A PTFE membrane was soaked for15 min in di-n-hexyl ether prior to extraction. The acceptor phase contained2.5 mM H2SO4, giving a pH of 2.5. In order to obtain satisfactory enrichmentthe sample (containing 12.5 mM EDTA, pH 10) was extracted three times.Despite this repeated sampling the efficiency was only 13% for aqueoussolutions. As amperozide is highly bound to proteins (97%), the expectedrecovery from plasma relative to water was 6% (as the plasma was diluted 1:1with donor buffer). To enhance release from the protein, the pH of the samplewas 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 inefficiency.

A final example worth mentioning is the result of a combinedSLME-µLC-CE system for the analysis of bambuterol [216]. This basic drugwas extracted from plasma using SLME and was introduced into a micro-LCcolumn. A heart-cut was transferred to the CE, in which enantiomer separationwas performed. A total enrichment factor of 40000 was observed, giving anLOD of 0.15 nM for each bambuterol enantiomer with simple UV detection.The main enrichment did not originate from the SLME procedure. Thesignificance 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 theseenrichments by analyte focussing on the LC column and from double stackingin CE.

The coupling of SLME to GC was investigated for the determination ofamines in urine [208]. A good LOD (1 ng/ml) and repeatability (3.5-4%) wereobtained for the analysis of more than 600 urine samples. As the final extract isan aqueous solution, SLME seems more appropriate for coupling with LC thanwith GC.

MMLLE, which has the advantage that the extract ends up in an organicsolvent, can more easily be combined with GC [150,209]. Even thoughMMLLE is different from SLME, similar problems were observed for theanalysis of local anaesthetics in plasma [209], i.e., matrix interference decreasedthe efficiency of the extraction and adsorption to the membrane causedcarry-over. As only one liquid-liquid extraction step is being performed, a lowselectivity may be obtained with the MMLLE procedure. Thus, the use of aselective detector is important. An advantage of MMLLE is, however, thatdilution 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 ofmethamphetamine [215]. Liquid-liquid-liquid microextraction (LLLME) wascoupled to CE. Urine and plasma samples were adjusted to pH 13 and,


57

subsequently, by use of a polypropylene hollow fiber the analyte was extractedthrough a thin phase of 1-octanol inside the pores of the fiber and finally in25 µl acceptor phase (0.1 M HCl) inside the fiber. The acceptor phase wasanalysed by capillary zone electrophoresis. A schematic diagram of the LLLMEunit is presented in Fig. 14. Effective clean-up was obtained as large molecules,acidic compounds and neutral compounds were not extracted, and an extractionefficiency 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.Fig. 14: Diagram of the LLLME extraction unit (not to scale) [215].

2.3.2.4 Remarks regarding the applicability ofmembrane-based techniques

Membrane-based techniques have three critical factors. The first problemis 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 studieshave shown that simple precautions can help to minimise or even eliminatethese problems. The positive aspects of membrane-based techniques, and inparticular NPMTs, are good selectivity in the extraction from complexbiological matrices, ease and versatility of automation and compatibility withanalytical instruments. The enrichment or efficiency can be sacrificed for morerapid analysis in cases where sensitivity is not a major issue. More sensitiveanalysis can be obtained by allowing the extraction process to take more time.As already mentioned before, porous membranes seem to have lost interest inthe bioanalytical field. NPMTs have shown potential and are now awaited to beaccepted as a new complementary technique for sample preparation. Thecommercial availability of suitable instrumentation will be of great importance.


58

2.4 Concluding remarks

The previous sections of this chapter have clearly shown that the varioussample pretreatment techniques each have their advantages and disadvantagesfor the clean-up and concentration of biological samples in integrated analyticalsystems. The various sample pretreatment techniques are compared in Table 4.

The set-up for SPE-GC and SPE-LC is somewhat complex, as a specialextraction device with several switching valves and a solvent delivery unit isrequired. For SPE-GC, a special GC injector is also required. The latter isrequired to allow the injection of the entire eluate by means of LVI.Membrane-based techniques are also more complicated due to various solventpumps and the membrane device. The set-up for fiber-based SPME is simple,especially in combination with GC. In-tube SPME involves the use of a lesssimple extraction device, but it can easily be set up, interfaced and automated.

Table 4: Comparison of sample pretreatment techniques.

SPE SPME Membrane(non-porous)

Simplicity + + + + + +Automation possibilities + + + +/+ + +a + +Choice of extraction phase + + + +b +Robustness + + + + +Versatility Samples

Coupling with LCCoupling with GC

+ + ++ + +

+

++ +

+ + +

+ ++ + +

+Speed + + + + +/–c + + +/–c

Recovery/Yield/Efficiency + + + –/+ +c –/+ +c

Carry-over + + +d + ++ + +: 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.

At present, the number of sorbents is largest for SPE, thus allowing agood choice for selectivity. However, with the advent of SPME the number oftypes of sorbents is also increasing. Despite their limited selectivities withregard 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-basedextraction principle in SPME and NPMTs, in which the small drug moleculesusually have larger diffusion or partition coefficients towards the stationaryphase than the matrix compounds.

Concluding remarks

59

Inversely related to the selectivity of the extraction is the requiredselectivity of the detector. The use of MS offers the possibility to sacrificeseparation efficiency when speeding up the analysis. Because of its selectivity,MS – either in the single mode or especially in the multiple modes (MSn, n≥2) –is a formidable tool to allow direct analysis of drugs in low quantities in acomplex biological matrix. However, it should be noted that poor separation ofthe analytes of interest from matrix compounds may result in either severereduction of the signal (ion suppression) or increase of the signal (ionenhancement). Systems in which a chromatographic separation step is omitted,i.e. direct coupling of sample pretreatment to MS, offer tremendous possibilitiesfor high-throughput analysis of biological samples, but the matrix may affectthe detection and quantitation even more. Therefore, careful attention should bepaid to the selectivity of extraction. The use of MSn may enhance the detectionselectivity even further, yet even then the selectivity of the sample pretreatmentwill remain to play an essential role in bioanalysis with these systems.

The robustness of SPME is still critical. The fiber as well as the coatedcapillary is fragile, and the coating can easily be stripped. NPMTs are not veryrobust as the stability of the membrane is a limiting factor. SPE is the mostestablished technique and offers very good robustness. Furthermore, the lattertechnique can easily be interfaced with both LC and GC, although at present thelatter is less common and more complex. SPE can be applied to a large varietyof samples such as urine, blood, plasma, serum, saliva, vitreous humour,cerebrospinal fluid, tissue homogenates, etc. Drugs are usually released fromproteins in SPE, thus allowing determination of the total concentration of thedrug of interest. SPME was originally designed for GC analysis, but thecoupling with LC proved also suitable. SPME is mainly used for the analysis ofurine samples, although examples of analysis of blood have been reported aswell. SPME offers the possibility to determine the free fraction of the drug inblood, whereas the analysis of the total concentration of drugs is morecomplicated. 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 recoverieswith SPE can be advantageous for sensitive analysis. However, also matrixcompounds can be extracted to a high extent, hereby limiting the selectivity ofthe system. In principle, SPME and NPMTs are slower than SPE because ofdiffusion limitations. However, with both SPME and NPMT the yield orefficiency is inversely related to the speed of the extraction. High speed can beobtained at the cost of sensitivity, i.e., low yields or efficiencies. Contrary, ifvery sensitive analysis is required, the speed can be lowered to achieve betteryields or efficiencies. Finally, carry-over is usually no problem with SPE asoften a new stationary phase is used for each extraction. However, on-line SPEalso allows the re-use of cartridges, thus increasing the risk of carry-over. With


60

SPME and NPMTs the same extraction phase is used for many analyses. As aresult, carry-over effects occur more often than with SPE.

The applicability of SPE-GC seems promising and its routine use isawaited. The latter system will be complementary to, but will not replaceSPME-GC. SPME-LC is a novel approach that can be used as a complementarytechnique to the well-established SPE-LC methods. Both fiber-based SPME andin-tube SPME can be routinely applied in bioanalysis. NPMTs are beingconsidered with growing interest and these systems seem to offercomplementary possibilities for sample pretreatment in bioanalysis. Thecoupling of LC-GC resembles SPE-GC. Although LC may give somechromatographic separation of the drugs of interest from the matrix compounds,it does not really add new possibilities to sample pretreatment with regard toSPE prior to GC analysis. The complexity of the system still limits its routineapplication.

TFC (or TFE)-LC is similar to SPE-LC, yet, with a high flow-rate duringextraction and the use of large particles in the extraction column. The extractionis partly size-exclusion based, and therefore mainly applied to plasma samples.PMTs have some potential to separate small analyte molecules from largematrix molecules such as proteins. Yet, it seems that they have found onlylimited use in practice. Both TFC (or TFE) and PMTs are mainly based onsize-exclusion extraction and may be useful for the analysis of plasma, serumand blood. Yet, protein binding can limit the applicability of these extractionsystems.

Integrated sample preparation has now been introduced in manybioanalytical laboratories. The experiences of routine laboratories greatlydetermine the acceptance and implementation of new techniques. Therefore,open communication channels between developers and users of sample clean-uptechniques are of utmost importance. In addition, lack of standardisation and/orsuitable automation appears to limit the acceptance of new techniques in routinepractice.

Acknowledgements

This research was supported by the Technology Foundation STW, appliedscience division of NWO and the technology programme of the Ministry ofEconomic Affairs.

List of Abbreviations

61

List of abbreviations

APCI Atmospheric pressure chemical ionisationDI/SPME Direct-immersion solid-phase microextractionESI Electrospray ionisationFID Flame-ionisation detection / flame-ionisation detectorFlu Fluorescence detector / fluorescence detectionGC Gas chromatography / gas chromatographHS/SPME Head-space solid-phase microextractionLC Liquid chromatography / liquid chromatographLLE Liquid-liquid extractionLLLME Liquid-liquid-liquid microextractionLOD Limit of detectionLOQ Limit of quantitationLVI Large-volume injection / large-volume injectorMESI Membrane extraction with a sorbent interfaceMIP Molecularly imprinted polymerMMLLE Microporous membrane liquid-liquid extractionMRM Multiple reaction monitoringMS Mass spectrometry / mass spectrometerMSn Multiple-stage mass spectrometryNPLC Normal phase liquid chromatographyNPMT Non-porous membrane techniquePDMS PolydimethylsiloxanePME Polymeric membrane extractionPMT Porous membrane techniquePTFE Poly(tetrafluoroethene)PTV Programmed temperature vaporiserRPLC Reversed phase liquid chromatographyRSD Relative standard deviationSIM Selected ion monitoringSLME Supported liquid membrane extractionSPE Solid-phase extractionSPME Solid-phase microextractionSRM Selected reaction monitoringSVE Solvent vapour exitTFC Turbulent-flow chromatographyUV Ultraviolet


62

2.5 References

[1] S.B. Hawthorne. Anal. Chem. 62 (1990) 633A.[2] J.R. Veraart, H. Lingeman, U.A.Th. Brinkman. J. Chromatogr. A 856 (1999)

483.[3] M. Dressler. J. Chromatogr. 165 (1979) 167.[4] R.E. Majors, H.G. Barth, C.H. Lochmüller. Anal Chem. 56 (1984) 300R.[5] Z.E. Penton. Adv. Chromatogr. 37 (1997) 205.[6] E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman. Chromatographia 29 (1990)

415.[7] E.H.R. van der Wal, E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman.

Chromatographia 39 (1994) 239.[8] D.A. McLoughlin, T.V. Olah, J.D. Gilbert. J. Pharm. Biomed. Anal. 15 (1997)

1893.[9] M. Jemal, D. Teitz, Z. Ouyang, S. Khan. J. Chromatogr. B 732 (1999) 501.[10] E.C. Goosens, D. de Jong, G.J. de Jong, U.A.Th. Brinkman. Chromatographia 47

(1998) 313.[11] A.J.H. Louter, J.J. Vreuls, U.A.Th. Brinkman. J. Chromatogr. A 842 (1999) 391.[12] J.J. Vreuls, A.J.H. Louter, U.A.Th. Brinkman. J. Chromatogr. A 856 (1999) 279.[13] K. Grob. J. Chromatogr. A 892 (2000) 407.[14] M. Jemal, Y.-Q. Xia, D.B. Whigan. Rapid Commun. Mass Spectrom. 12 (1998)

1389.[15] J.-T. Wu, H. Zeng, M. Qian, B.L. Brogdon, S.E. Unger. Anal. Chem. 72 (2000)

61.[16] C. Chassaing, J. Luckwell, P. Macrae, K. Saunders, P. Wright, R. Venn.

Chromatographia 53 (2001) 122.[17] M. Jemal. Biomed. Chromatogr. 14 (2000) 422.[18] N.H. Snow. J. Chromatogr. A 885 (2000) 455.[19] G. Theodoridis, E.H.M. Koster, G.J. de Jong. J. Chromatogr. B 745 (2000) 49.[20] S. Ulrich. J. Chromatogr. A 902 (2000) 167.[21] G.A. Mills, V. Walker. J. Chromatogr. A 902 (2000) 267.[22] H.L. Lord, J.B. Pawliszyn. J. Chromatogr. A 885 (2000) 153.[23] N.C. van de Merbel. J. Chromatogr. A 856 (1999) 55.[24] N.C. van de Merbel, J.M. Teule, H. Lingeman, U.A.Th. Brinkman. J. Pharm.

Biomed. Anal. 10 (1992) 225.[25] N.C. van de Merbel, J.J. Hageman, U.A.Th. Brinkman. J. Chromatogr. 634

(1993) 1.[26] J.P. Allanson, R.A. Biddlecombe, A.E. Jones, S. Pleasance. Rapid Commun.

Mass Spectrom. 10 (1996) 811.[27] W.Z. Shou, M. Pelzer, T. Addison, X. Jiang, W. Naidong. J. Pharm. Biomed.

Anal. 27 (2002) 143.[28] R.W. Frei, K. Zech (eds). Selective sample handling and detection in high-

performance liquid chromatography (part A), Elsevier Science Publishers,Amsterdam, The Netherlands 1988.

[29] E.R. Brouwer, D.J. van Iperen, I. Liska, H. Lingeman, U.A.Th. Brinkman. Intern.J. Environ. Anal. Chem. 47 (1992) 257.

[30] C.H.P. Bruins, C.M. Jeronimus-Stratingh, K. Ensing, W.D. van Dongen, G.J. deJong. J. Chromatogr. A 863 (1999) 115.

References

63

[31] G.P. Bowers, S.C. Clegg, S.C. Hughes, A.J. Harker, S. Lambert. LC•GC 15(1997) 48.

[32] J.A. Ooms, G.S.J. Haak, O. Halmingh. Intern. Lab. 27 (1997) 18A.[33] M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de

Jong. Rapid Commun. Mass Spectrom. 14 (2000) 2103.[34] A. Schellen, J.A. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de

Lagemaat, E. Verheij. Rapid Commun. Mass Spectrom. 14 (2000) 230.[35] A.C. Hogenboom, P. Speksnijder, R.J. Vreeken, W.M.A. Niessen, U.A.Th.


(1998) 201.[37] J. Ding, U.D. Neue. Rapid Commun. Mass Spectrom. 13 (1999) 2151.[38] W.A. Minnaard, A.C. Hogenboom, U.K. Malmqvist, P. Manini, W.M.A.

Niessen, U.A.Th. Brinkman. Rapid Commun. Mass Spectrom. 10 (1996) 1569.[39] B.J. Hall, B. Goolsby, J.S. Brodbelt. Appl. Spectrosc. 53 (1999) 1361.[40] K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng. Anal. Chem. 70 (1998)

882.[41] I. Fu, E.J. Woolf, B.K. Matuszewski. J. Pharm. Biomed. Anal. 18 (1998) 347.[42] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle. Rapid Commun. Mass Spectrom.

19 (1999) 1175.[43] D.L. Buhrman, P.I. Price, P.J. Rudewicz. J. Am. Soc. Mass Spectrom. 7 (1996)

1099.[44] A. Namera, M. Yashiki, K. Okada, Y. Iwasaki, M. Ohtani, T. Kojima.

J. Chromatogr. B 706 (1998) 253.[45] A. Namera, M. Yashiki, Y. Iwasaki, M. Ohtani, T. Kojima. J. Chromatogr. B 716

(1998) 171.[46] A.J.H. Louter, E. Bosma, J.C.A. Schipperen, J.J. Vreuls, U.A.Th. Brinkman.

J. Chromatogr. B 689 (1997) 35.[47] A.J.H. Louter, R.A.C.A. van der Wagt, U.A.Th. Brinkman. Chromatographia 40

(1995) 500.[48] H.G.J. Mol, H.-G. Janssen, C.A. Cramers, J.J. Vreuls, U.A.Th. Brinkman.

J. Chromatogr. A 703 (1995) 277.[49] K. Grob, J.-M. Stoll. J. High Resolut. Chromatogr. 9 (1986) 518.[50] K. Grob Jr., G. Karrer, M.-L. Riekkola. J. Chromatogr. 334 (1985) 129.[51] W. Vogt, K. Jacob, H.W. Obwexer. J. Chromatogr. 174 (1979) 437.[52] W. Vogt, K. Jacob, A.-B. Ohnesorge, H.W. Obwexer. J. Chromatogr. 186 (1979)

179.[53] A. Farjam, J.J. Vreuls, W.J.G.M. Cuppen, U.A.Th. Brinkman, G.J. de Jong.

Anal. Chem. 63 (1991) 2481.[54] R. Herráez-Hernández, A.J.H. Louter, N.C. van de Merbel, U.A.Th. Brinkman.

J. Pharm. Biomed. Anal. 14 (1996) 1077.[55] M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong. J. Chromatogr. B

729 (1999) 199.[56] M.P.M. van Lieshout, H.-G. Janssen, C.A. Cramers, G.A. van den Bos.

J. Chromatogr. A 764 1997) 73.[57] A.J.H. Louter, J. van Doornmalen, J.J. Vreuls, U.A.Th. Brinkman. J. High

Resolut. Chromatogr. 19 (1996) 679.[58] H.G.J. Mol, H.-G. Janssen, C.A. Cramers, U.A.Th. Brinkman. J. High Resolut.

Chromatogr. 16 (1993) 459.


64

[59] J.J. Vreuls, U.A.Th. Brinkman, G.J. de Jong, K. Grob, A. Artho. J. High Resolut.Chromatogr. 14 (1991) 455.

[60] J.J. Vreuls, R.T. Ghijsen, G.J. de Jong, U.A.Th. Brinkman. J. Microcol. Sep. 5(1993) 317.

[61] M. Abdel-Rehim, K.A. Svensson, Y. Askemark, K.-J. Pettersson. J. Chromatogr.B 755 (2001) 253.

[62] M.W.J. van Hout, R.A. de Zeeuw, G.J. de Jong. J. Chromatogr. A 858 (1999)117.

[63] M.W.J. van Hout, W.M.A. van Egmond, J.P. Franke, R.A. de Zeeuw, G.J. deJong. J. Chromatogr. B 766 (2001) 37.

[64] R. Sasano, T. Hamada, M. Kurano, M. Furuno. J. Chromatogr. A 896 (2000) 41.[65] T. Hankemeier, U.A.Th. Brinkman. Chromatogr. Sci. 86 (2001) 155.[66] K. Grob. J. Chromatogr. A 703 (1995) 265.[67] T. Hyötyläinen, M.-L. Riekkola. J. Chromatogr. A 819 (1998) 13.[68] F. Munari, P.A. Colombo, P. Magni, G. Zilioli, S. Trestianu, K. Grob.

J. Microcol. Sep. 7 (1995) 403.[69] L. Mondello, P. Dugo, G. Dugo, A.C. Lewis, K.D. Bartle. J. Chromatogr. A 842

(1999) 373.[70] I.A. Mostert, K.A. Ramsteiner. J. Chromatogr. 477 (1989) 359.[71] V. Giansello, E. Brenn, G. Figini, A. Gazzaniga. J. Chromatogr. 473 (1989) 343.[72] D. Duquet, C. Dewaele, M. Verzele, S. McKinley. J. High Resolut. Chromatogr.,

Chromatogr. Commun. 11 (1988) 924.[73] E.C. Goosens, D. de Jong, J.H.M. van den Berg, G.J. de Jong, U.A.Th.

Brinkman. J. Chromatogr. 552 (1991) 489.[74] E.C. Goosens, I.M. Beerthuizen, D. de Jong, G.J. de Jong, U.A.Th. Brinkman.

Chromatographia 40 (1995) 267.[75] E.C. Goosens, K.H. Stegman, D. de Jong, G.J. de Jong, U.A.Th. Brinkman.

Analyst 121 (1996) 61.[76] E.C. Goosens, D. de Jong, G.J. de Jong, U.A.Th. Brinkman. J. Microcol. Sep. 6

(1994) 207.[77] P. Wessels, J. Ogorka, G. Schwinger, M. Ulmer. J. High. Resolut. Chromatogr.

16 (1993) 708.[78] J. Ogorka, G. Schwinger, G. Braut, V. Seidel. J. Chromatogr. 626 (1992) 87.[79] T. Hyötyläinen, T. Andersson ,M.-L. Riekkola. J. Chromatogr. Sci. 35 (1997)

280.[80] T. Hyötyläinen, R. Pilvio, M.-L. Riekkola. J. High Resolut. Chromatogr. 19

(1996) 439.[81] T. Hyötyläinen, H. Keski-Hynnilä, M.-L. Riekkola. J. Chromatogr. A 771 (1997)

360.[82] H.M. Quinn, J.J. Takarewski. International Patent Number WO97/16724 (1997).[83] D. Zimmer, V. Pickard, W. Czembor, C. Müller. J. Chromatogr. A 854 (1999)

23.[84] N. Brignol, R. Bakhtiar, L. Dou, T. Majumbar, F.L.S. Tse. Rapid Commun. Mass

Spectrom. 14 (2000) 141.[85] J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallett, R.S. Plumb. Rapid Commun.

Mass Spectrom. 11 (1997) 1953.[86] J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallett, R.S. Plumb. J. Chromatogr. A

828 (1998) 199.

References

65

[87] M. Jemal, Z. Ouyang, Y.-Q. Xia, M.L. Powell. Rapid Commun. Mass Spectrom.13 (1999) 1462.

[88] Y.-Q. Xia, D.B. Whigan, M.L. Powell. M. Jemal, Rapid Commun. MassSpectrom. 14 (2000) 105.

[89] Y. Hsieh, M.S. Bryant, G. Gruela, J.-M. Brisson, W.A. Korfmacher. RapidCommun. Mass Spectrom. 14 (2000) 1384.

[90] J. Ayrton, R.A. Clare, G.J. Dear, D.N. Mallett, R.S. Plumb. Rapid. Commun.Mass Spectrom. 13 (1999) 1657.

[91] M. Jemal, M. Huang, X. Jiang, Y. Mao, M.L. Powell. Rapid Commun. MassSpectrom. 13 (1999) 2125.

[92] M. Jemal, Z. Ouyang, M.L. Powell. J. Pharm. Biomed. Anal. 23 (2000) 323.[93] D. Zimmer, V. Pickard, W. Czembor, C. Müller. Chromatographia 52 (2000)

S26.[94] H.K. Lim, K.W. Chan, S. Sisenwine. J.A. Scatina. Anal. Chem. 73 (2001) 2140.[95] R.T. Cass, J.S. Villa, D.E. Karr, D.E. Schmidt. Rapid Commun. Mass Spectrom.

15 (2001) 406.[96] C.L. Arthur, J.B. Pawliszyn. Anal. Chem. 62 (1990) 2145.[97] Z. Zhang, M.J. Yang, J.B. Pawliszyn. Anal. Chem. 66 (1994) 844A.[98] D. Louch, S. Motlagh, J.B. Pawliszyn. Anal. Chem. 64 (1992) 1187.[99] H.L. Lord, I.B. Pawliszyn. J. Chromatogr. A 902 (2000) 17.[100] H.A.G. Niederländer, V. Jas, M.W.J. van Hout. Submitted to Anal. Chem.[101] W.H.J. Vaes, C. Hamwijk, E.U. Ramos, H.J.M. Verhaar, J.L.M. Hermens. Anal.

Chem. 68 (1996) 4458.[102] R. Eisert, J. Pawliszyn. Crit. Rev. Anal. Chem. 27 (1997) 103.[103] T. Górecki, X. Yu, J.B. Pawliszyn. Analyst 124 (1999) 643.[104] M.W.J. van Hout, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de Jong.

Analyst 127 (2002) 355.[105] Y.Yang, S.B. Hawthorne, D.J. Miller, Y. Liu, M.L.Lee. Anal. Chem. 70 (1998)

1866.[106] J.B. Pawliszyn. Solid Phase Microextraction – Theory and Practice, Wiley, New

York, 1997.[107] E.H.M. Koster, N.S.K. Hofman, G.J. de Jong. Chromatographia 47 (1998) 678.[108] J. Beltran, F.J. López, F. Hernández. J. Chromatogr. A 885 (2000) 389.[109] T. Kumazawa, O. Suzuki. J. Chromatogr. B 747 (2000) 241.[110] H. Kataoka, H.L. Lord, J.B. Pawliszyn. J. Chromatogr. A 880 (2000) 35.[111] A. Namera, M. Yashiki, N. Nagasawa, Y. Iwasaki, T. Kojima. Forensic Sci. Int.

88 (1997) 125.[112] K.G. Furton, J.R. Armirall, M. Bi, J. Wang, L.Wu. J. Chromatogr. A 885 (2000)

419.[113] P. Okeyo, S.M. Rentz, N.H. Snow. J. High Resolut. Chromatogr. 20 (1997) 171.[114] W.H.J. Vaes, E.U. Ramos, H.J.M. Verhaar, W. Seinen, J.L.M. Hermens. Anal.

Chem. 68 (1996) 4463.[115] W.H.J. Vaes, E.U. Ramos, C. Hamwijk, I. Van Holsteijn, B.J. Blaauboer, W.

Semen, H.J.M. Verhaar, J.L.M. Hermens. Chem. Res. Toxicol. 10 (1997) 1067.[116] E.H.M. Koster, C. Wemes, J.B. Morsink, G.J. de Jong. J. Chromatogr. B 739

(2000) 175.[117] D. Figeys, A. Ductre, J.R. Yates, R. Aebersold. Nat. Biotechnol. 14 (1996) 1579.[118] C.-W. Whang, J.B. Pawliszyn. Anal. Commun. 35 (1998) 353.[119] S. Li, S. Weber. Anal. Chem. 69 (1997) 1217.


66

[120] K. Jinno, Y. Han, H. Sawada, M. Taniguchi. Chromatographia 46 (1997) 309.[121] K. Jinno, H. Sawada, Y. Han. Biomed. Chromatogr. 12 (1998) 126.[122] J. Chen, J.B. Pawliszyn. Anal. Chem. 67 (1995) 2530.[123] A.A. Boyd-Boland, J.B. Pawliszyn. Anal. Chem. 68 (1996) 1521.[124] R. Eisert, J. Pawliszyn. Anal. Chem. 69 (1997) 3140.[125] H. Kataoka, H.L. Lord, J. Pawliszyn. J. Anal. Toxicol. 24 (2000) 257.[126] H. Kataoka, H.L. Lord, S. Yamamoto, S. Narimatsu, J. Pawliszyn. J. Microcol.

Sep. 12 (2000) 493.[127] K. Jinno, M. Kawazoe, M. Hayashida. Chromatographia 52 (2000) 309.[128] Y. Saito, M. Kawazoe, M. Hayashida, K. Jinno. Analyst 125 (2000) 807.[129] H. Yuan, Z. Mester, H. Lord, J. Pawliszyn. J. Anal. Toxicol. 24 (2000) 718.[130] K. Jinno, M. Taniguchi, M. Hayashida. J. Pharm. Biomed. Anal. 17 (1998) 1081.[131] E.H.M. Koster, C. Crescenzi, W. den Hoedt, K. Ensing, G.J. de Jong. Anal.

Chem. 73 (2001) 3140.[132] J. Wu, H.L. Lord, J. Pawliszyn, H. Kataoka. J. Microcol. Sep. 12 (2000) 255.[133] H. Kataoka, S. Narimatsu, H.L. Lord, J. Pawliszyn. Anal. Chem. 71 (1999) 4237.[134] M. Möder, H. Löster, R. Herzschuh, P. Popp. J. Mass Spectrom. 32 (1997) 1195.[135] D.A. Volmer, J.P.M. Hui. Rapid Commun. Mass Spectrom. 11 (1997) 1926.[136] M. Satterfield, D.M. Black, J.S. Brodbelt. J. Chromatogr. B 759 (2001) 33.[137] T. Kumazawa, H. Seno, K. Watanabe-Suzuki, H. Hattori, A. Ishii, K. Sato, O.

Suzuki. J. Mass Spectrom. 35 (2000) 1091.[138] H. Kataoka, H.L. Lord, J. Pawliszyn. J. Chromatogr. B 731 (1999) 353.[139] C. Pérès, C. Vaillon, J.-L. Berdagué. Anal. Chem. 73 (2001) 1030.[140] G.E. Orzechowska, E.J. Poziomek, V. Tersol. Anal. Letters 30 (1997) 1437.[141] J. Ai. Anal. Chem. 69 (1997) 1230.[142] Y. Liu, M.L. Lee, K.J. Hageman, Y. Yang, S.B. Hawthorne. Anal. Chem. 69

(1997) 5001.[143] Y. Liu, Y.F. Shen, M.L. Lee. Anal. Chem. 69 (1997) 190.[144] F. Mangani, R. Cenciarini. Chromatographia 41 (1995) 678.[145] E. Baltussen, P. Sandra, F. David, C. Cramers. J. Microcol. Sep. 11 (1999) 737.[146] J.Å. Jönsson, L. Mathiasson. J. Chromatogr. A 902 (2000) 205.[147] B.N. Cordero, J.L.P. Pavón, C.G. Pinto, M.E.F. Laespada, R.C. Martínez, E.R.

Gonzalo. J. Chromatogr. A 902 (2000) 195.[148] J.Å. Jönsson, L. Mathiasson. J. Sep. Sci. 24 (2001) 495.[149] J.Å. Jönsson, L. Mathiasson. Chromatographia 52 (2000) S-8.[150] J.Å. Jönsson, L. Mathiasson. Trends Anal. Chem. 18 (1999) 318.[151] J.Å. Jönsson, L. Mathiasson. Trends Anal. Chem. 18 (1999) 325.[152] J.Å. Jönsson, L. Mathiasson. Adv. Chromatogr. 41 (2001) 53.[153] M. Gilar, E.S.P. Bouvier, B.J. Compton. J. Chromatogr. A 909 (2001) 111.[154] G. Matz, G. Kibelka, J. Dahl, F. Lenneman. J. Chromatogr. A 830 (1999) 365.[155] B. Hauser, P. Popp. J. High Resolut. Chromatogr. 22 (1999) 205.[156] S. Mitra, N. Zhu, X. Zhang, B. Kebbekus. J. Chromatogr. A 736 (1996) 165.[157] A. Chen, C.E. Lunte. J. Chromatogr. A 691 (1995) 29.[158] K.M. Steele, C.E. Lunte. J. Pharm. Biomed. Anal. 13 (1995) 149.[159] B.H.C. Westerink. Trends Anal. Chem. 11 (1992) 176.[160] C.E. Lunte, D.O. Scott, P. Kissinger. Anal. Chem. 63 (1991) 773A.[161] E.H. Kerns, K.J. Volk, S.E. Klohr, M.S. Lee. J. Pharm. Biomed. Anal. 20 (1999)

115.[162] T.-H. Tsai, Y.-F. Chen, I.-F. Chen, C.-F. Chen. J. Chromatogr. B 729 (1999) 119.

References

67

[163] B.A.P. Buscher, U.R. Tjaden, J. van der Greef. J. Chromatogr. A 764 (1997) 135.[164] A.J.J. Debets, W.Th. Kok, K.-P. Hupe, U.A.Th. Brinkman. Chromatographia 30

(1990) 361.[165] B.A.P. Buscher, A.J.P. Hofte, U.R. Tjaden, J. van der Greef. J. Chromatogr. A

777 (1997) 51.[166] P. González, C.A. Fente, C. Franco, B. Vázquez, E. Quinto, A. Cepeda.

J. Chromatogr. B 693 (1997) 321.[167] J.R. Veraart, J. van Hekezen, M.C.E. Groot, C. Gooijer, H. Lingeman, N.H.

Velthorst, U.A.Th. Brinkman. Electrophoresis 19 (1998) 2944.[168] J.R. Veraart, M.C.E. Groot, C. Gooijer, H. Lingeman, N.H. Velthorst, U.A.Th.

Brinkman. Analyst 124 (1999) 115.[169] D.C. Turnell, J.D.H. Cooper. J. Autom. Chem. 7 (1985) 177.[170] D.C. Turnell, J.D.H. Cooper. J. Autom. Chem. 7 (1985) 181.[171] J.D.H. Cooper, D.C. Turnell. J. Chromatogr. 380 (1986) 109.[172] Q. Chang, M.E. Meyerhoff. Anal. Chim. Acta 186 (1986) 81.[173] U.R. Tjaden, E.A. de Bruijn, R.A.M. van der Hoeven, C. Jol, J. van der Greef, H.

Lingeman. J. Chromatogr. 420 (1987) 53.[174] D.C. Turnell, J.D.H. Cooper. J. Chromatogr. 395 (1987) 613.[175] D.C. Turnell, J.D.H. Cooper. B. Green, G. Hughes, D.J. Wright. Clin. Chem. 34

(1988) 1816.[176] J.D.H. Cooper, D.C. Turnell, B. Green, F. Verillon. J. Chromatogr. 456 (1988)

53.[177] H. Irth, G.J. de Jong, H. Lingeman, U.A.Th. Brinkman. Anal. Chim. Acta 236

(1990) 165.[178] D.S. Stegehuis, U.R. Tjaden, J. van der Greef. J. Chromatogr. 511 (1990) 137.[179] T. Agasøster, K.E. Rasmussen. J. Chromatogr. 570 (1991) 99.[180] T. Agasøster, K.E. Rasmussen. J. Chromatogr. 564 (1991) 171[181] A.T. Andresen, P.B. Jacobsen, K.E. Rasmussen. J. Chromatogr. 575 (1992) 93.[182] A.T. Andresen, M. Krogh, K.E. Rasmussen. J. Chromatogr. 582 (1992) 123.[183] E. van Bakergem, R.A.M. van der Hoeven, W.M.A. Niessen, U.R. Tjaden, J. van

der Greef, G.K. Poon, R. McCague. J. Chromatogr. 598 (1992) 189.[184] A.R. Buick, C.T.C. Fook Sheung. J. Chromatogr. 617 (1993) 65.[185] A.T. Andresen, K.E. Rasmussen, H.E. Rugstad. J. Chromatogr. 621 (1993) 189.[186] J.D.H. Cooper, C.T.C. Fook Sheung, A.R. Buick. J. Chromatogr. B 652 (1994)

15.[187] R. Herráez-Hernández, N.C. van de Merbel, U.A.Th. Brinkman. J. Chromatogr.

B 666 (1995) 127.[188] K. Johansen, M. Krogh, A.T. Andresen, A.S. Christophersen, G. Lehne, K.E.

Rasmussen. J. Chromatogr. B 669 (1995) 281.[189] A. Ceccato, P. Chiap, Ph. Hubert, B. Toussaint, J. Crommen. J. Chromatogr. A

750 (1996) 351.[190] J.D.H. Cooper, D.C. Muirhead, J.E. Taylor, P.R. Baker. J. Chromatogr. B 701

(1997) 87.[191] M. Karlsson, T. Korkolainen, T. Wikberg. Biomed. Chromatogr. 11 (1997) 54.[192] K. Johansen, M. Krogh, K.E. Rasmussen. J. Chromatogr. B 690 (1997) 223.[193] A. Ceccato, B. Toussaint, P. Chiap, Ph. Hubert, J. Crommen. J. Pharm. Biomed.

Anal. 15 (1997) 1365.[194] G. Higgins, A. Highes, J. Dutton, N.B. Roberts. Ann. Clin. Biochem. 35 (1998)

534.


68

[195] F. Sadeghipour, J.-L. Veuthey. J. Pharm. Biomed. Anal. 17 (1998) 801.[196] K. Johansen, K.E. Rasmussen. J. Pharm. Biomed. Anal. 16 (1998) 1159.[197] T. Zupanèiè, B. Pihlar. J. Chromatogr. A 840 (1999) 11.[198] P.B. Jacobsen. J. Chromatogr. B 749 (2000) 135.[199] P. Chiap, B. Evrard, M.A. Bimazubute, P. de Tullio, Ph. Hubert, L. Delattre,

J. Crommen. J. Chromatogr. A 870 (2000) 121.[200] P. Chiap, B. Boulanger, L. Fotsing, Ph. Hubert, J. Crommen. Chromatographia

53 (2001) 678.[201] P. Chiap, A. Ceccato, B. Miralles Buraglia, B. Boulanger, Ph. Hubert,

J. Crommen. J. Pharm. Biomed. Anal. 24 (2001) 801.[202] D.C. Turnell, J.D.H. Cooper. J. Autom. Chem. 8 (1986) 151.[203] J. Trocewicz, Z. Suprynowicz, J. Markowicz. J. Chromatogr. B 685 (1996) 129.[204] J.Å. Jönsson, M. Andersson, C. Melander, J. Norberg, E. Thordarson,

L. Mathiasson. J. Chromatogr. A 870 (2000) 151.[205] B. Lindegård, H. Björk, J.Å. Jönsson, L. Mathiasson, A.-M. Olsson. Anal. Chem.

66 (1994) 4490.[206] E. Thordarson, S. Pálmarsdóttir, J.Å. Jönsson, L. Mathiasson. Anal. Chem. 68

(1996) 2559.[207] J. Norberg, J. Emnéus, J.Å. Jönsson, L. Mathiasson, E. Burestedt, M. Knutsson,

G. Marko-Varga. J. Chromatogr. B 701 (1997) 39.[208] G. Audunsson. Anal. Chem. 60 (1988) 1340.[209] Y. Shen, J.Å. Jönsson, L. Mathiasson. Anal. Chem. 70 (1998) 946.[210] B. Lindegård, J.Å. Jönsson, L. Mathiasson. J. Chromatogr. 573 (1992) 191.[211] J.Å. Jönsson, L. Mathiasson, B. Lindegård, J. Trocewicz, A.-M. Olsson.

J. Chromatogr. A 665 (1994) 259.[212] Y. Shen, L. Mathiasson, J.Å. Jönsson. J. Microcol. Sep. 10 (1998) 107.[213] S. Pálmarsdóttir, E. Thordarson, L.-E. Edholm, J.Å. Jönsson, L. Mathiasson.

Anal. Chem. 69 (1997) 1732.[214] S. Pálmarsdóttir, L. Mathiasson, J.Å. Jönsson, L.-E. Edholm. J. Chromatogr. B

688 (1997) 127.[215] S. Pedersen-Bjergaard, K.E. Rasmussen. Anal. Chem. 71 (1999) 2650.[216] S. Pálmarsdóttir, L. Mathiasson, J.Å. Jönsson, L.-E. Edholm. J. Cap.

Electrophoresis 3 (1996) 255.[217] M. Papantoni, N.-K. Djane, K. Ndung’u, J.Å. Jönsson, L. Mathiasson. Analyst

120 (1995) 1471[218] F. Malcus, N.-K. Djane, L. Mathiasson, G. Johansson. Anal. Chim. Acta 327

(1996) 295.[219] N.-K. Djane, I.A. Bergdahl, K. Ndung’u, A. Schütz, G. Johansson, L.

Mathiasson. Analyst 122 (1997) 1073.[220] N.-K. Djane, S. Armalis, K. Ndung’u, G. Johansson, L. Mathiasson. Analyst 123

(1998) 393.[221] S. Armalis, I. Kriksciuniene, E. Kubilene, N.-K. Djane, K. Ndung’u,

L. Mathiasson. Intern. J. Environm. Anal. Chem. 74 (1999) 233.[222] E. Luque-Peréz, A. Ríos, M. Valcárcel, L.-G. Danielsson, F. Ingman. Lab.

Autom. Inform. Managem. 34 (1999) 131.[223] R.G. Melcher. Anal. Chim. Acta 214 (1988) 299.[224] J.Å. Jönsson, L. Mathiasson. Trends Anal. Chem. 11 (1992) 106.

33INTEGRATION OF SOLID-PHASE

EXTRACTION AND GASCHROMATOGRAPHY

The pieces have

Evaluation of the PTV for LVI of biological samples in GC

71

3.1

Evaluation of the programmed temperaturevaporiser for large-volume injection of biologicalsamples in gas chromatography*

Summary

The use of a programmed temperature vaporiser (PTV) with a packed liner wasevaluated for the injection of large volumes (up to 100 µl) of plasma extracts ina gas chromatograph. Solvent purity, which is essential when large volumes areinjected into the GC system, was determined. Special attention was paid to thepurity of the solvents used for the solid-phase extraction (SPE) procedure. Forthis SPE method, ethyl acetate was used as the extraction and reconstitutionsolvent, and thus the purity of the ethyl acetate was critical, especially when anon-selective GC detector was applied. The liquid capacity and inertness ofdifferent packed liners were investigated. The liner packed with ATAS "A"(modified Chromosorb-based material with special treatment) was found to bethe most suitable for the analysis of the tested drugs. Good linearity in responsefor variations in volume and concentration was observed. A comparison wasmade between the applicability of flame-ionisation detection (FID) and massselective detection (MSD). When 50-µl volumes of plasma extracts wereinjected with the PTV, the detection limits for secobarbital, lidocaine,phenobarbital, and diazepam were about 50 times lower than when 1-µlvolumes were injected. The detection limits of the tested compounds in plasmafor injection of 50-100 µl plasma extract are 5-10 ng/ml for GC-FID whereasplasma concentrations of 250 pg/ml can be detected using the selected ionmonitoring (SIM) mode of an MSD. For non-selective GC-FID, the backgroundfrom a 50-µl injection was substantially larger than with 1-µl injection as aresult of co-injected plasma matrix components and solvent impurities. Thesebackground effects were less with GC-MSD in the total ion current (TIC) modeand virtually absent with GC-MSD in the SIM mode.

*: M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong. J. Chromatogr. B 729 (1999) 199-210.

Chapter 3 – Integration of solid-phase extraction and gas chromatography

72

3.1.1 Introduction

Increasing knowledge of the working mechanism of biologically activesubstances has led to the development of potent drugs. Hence, lower dosagescan be administered to produce a therapeutic effect and, consequently, drugconcentrations in biological samples often are much lower than before. For thedetermination of these lower levels in biological samples, analytical techniqueswith much higher sensitivity are needed. A way to increase the sensitivity is toincrease the amount of sample injected into the analytical system.

In gas chromatography (GC) several techniques are available to performlarge-volume injections (LVIs) [1]. On-column injection with the use ofso-called retention gaps is currently the most common technique [1]. A secondpossibility for LVI is the loop-type interface [2], originally designed for thecoupling of liquid chromatography (LC) and GC. The main advantage of thesetechniques is that the complete sample is introduced into the GC column.However, this may also become a disadvantage since all impurities areintroduced into the GC system as well. A third option to allow LVI in GC is touse a programmed temperature vaporiser (PTV). Despite good results obtainedby Vogt and co-workers [3,4] in the late seventies, only recently PTV injectionwas applied as a routine technique for environmental analysis [1].

Besides conventional split/splitless injection, the PTV can be used forseveral modes of LVI. The coupling of LC and GC using the PTV was reviewedby Grob [5], and recently interesting publications appeared on the same subject[6,7]. The PTV is often applied for this purpose because the packed linergenerally has a larger liquid storage capacity than a retention gap. In addition,wetability is not very critical for the liquid retention and packing materials aremore water-resistant than retention gaps with a silica backbone. The packing ismore easily and rapidly heated than a retention gap [5]. Main reasons to coupleLC with GC are that LC provides better resolution than more conventionaltechniques of sample preparation, and secondly, the possibility of automationthrough on-line coupling, which reduces or eliminates manual samplepreparation work and, therefore, reduces analysis time and improves accuracyand precision [5,7]. The use of a PTV as the interface between LC and GC hasbeen demonstrated for the analysis of olive oil and for environmental analysis[6,7]. The PTV is also used for thermal desorption-pyrolysis of solidgeochemical samples (characterisation of oil and kerogens in source rocks) [8],and for on-line solid-phase extraction-thermal desorption (SPETD) of methylesters of the C10-C26 carboxylic acids, pesticides, chlorobenzenes andchlorophenols in aqueous samples [9-11].

Most applications of LVI are in the analysis of environmental aqueoussamples [1,9-13]. Pesticides were determined in aqueous samples after SPE ofsamples of 200 ml with concentrations between 0.2 and 5 ng/l by Steen et al.


73

[12], whereas Teske et al. [13] determined triazines like atrazine, propazine,ametryne and simazine in water after in-vial liquid-liquid extraction and directinjection of the extracts with detection limits as low as 0.01 µg/l, andpolychlorinated biphenyls (PCBs) and polycyclic aromatic hydrocarbons(PAHs) at the ppt-level [13]. Another application of the PTV is the residueanalysis of 385 pesticides down to the 0.01-ppm concentration level in plantfoodstuff [14].

The purpose of the present work is to investigate the possibilities of thePTV coupled to GC for the analysis of plasma extracts to provide lowerdetection limits for drugs. Special attention was paid to the impact of solventimpurities in view of the larger solvent volumes injected, to the liquid capacityand inertness of the PTV liners, and to the degree of selectivity provided byflame-ionisation detection (FID) and mass selective detection (MSD).

3.1.2 Experimental

3.1.2.1 Instrumentation

Gas chromatographic analyses were performed with a Hewlett-PackardHP 5890 series II with FID or a GC-MSD system (HP 5971 series). A HP-530 m×0.32 mm capillary column with 0.25 µm film thickness was used for theanalyses with FID, whereas analyses with MSD were performed using a HP-5MS 30 m×0.25 mm column with 0.25 µm film thickness. The PTV injectionsystem was an OPTIC 2 (ATAS International, Veldhoven, The Netherlands),which was equipped with 80 mm×3.4 mm i.d. liners obtained from ATASInternational. The liners were packed with either ATAS "A" packing (amodified Chromosorb-based material with special treatment, ATASInternational), silanised glass wool (research grade, Serva, Feinbiochemica,Heidelberg, Germany), or disposable capillaries for thin-layer chromatography(TLC) (nine capillaries of 10 µl and two of 2 µl, cut at a length of 2 cm).

Plasma extractions were performed using Bond Elut Certify cartridges(Varian, Harbor City, CA, USA), column type LRC of 10 ml with 130 mgsorbent. A Visiprep system (Supelco, Bellefonte, PA, USA) was used to applyvacuum during the extraction.

3.1.2.2 Chemicals

Acetonitrile and methanol (Lab Scan, Dublin, Ireland) were of HPLCquality. Acetone, hexane, acetic acid glacial 100% (v/v), ammonia solution25%, and KH2PO4 were all of analytical-reagent quality (Merck, Darmstadt,Germany). Ethyl acetate (Reinst and Suprasolv - for organic residue analysis)


74

was obtained from Merck (Darmstadt, Germany). Ethyl acetate Ultraresi-analysed (for organic residue analysis) was purchased from MallinckrodtBaker (Deventer, The Netherlands). Water used during SPE was ultra pure(Elgastat maxima, Salm en Kipp, Breukelen, The Netherlands). Secobarbital,phenobarbital (both BP quality, Siegfried, Zofingen, Switzerland), lidocaine(Eur. Ph., Holland Pharmaceutical Supply, Alphen A/D Rijn, The Netherlands),and diazepam (Centrafarm, Etten-Leur, The Netherlands) were used as testcompounds (Fig. 1) and dissolved in ethyl acetate (for organic residue analysis,Mallinckrodt Baker). Stock solutions of 1 mg/ml were stored in the dark at 4°C.Stock solutions were mixed and then diluted with ethyl acetate (for organicresidue analysis, Mallinckrodt Baker). The compounds of the referenceRI-mixture [15] were dissolved in ethyl acetate:methanol (1:1) (1 mg/ml).

Fig. 1. Structures of the test compounds: (A) secobarbital, (B) lidocaine,(C) phenobarbital, (D) diazepam.

3.1.2.3 Methods

The carrier gas for GC-FID and GC-MSD was helium. The sametemperature program was used for both methods. The starting temperature was40°C, and after 3 min the temperature was raised at 20°C/min to 215°C,followed by an increase at 5°C/min to 230°C and a final increase at 25°C/min to290°C. This final temperature was maintained for 5-10 min. The detectortemperature was 300°C. A column flow of 1.35 ml/min was used duringanalysis with GC-FID and 0.48 ml/min with GC-MSD. The injector was set at40°C and 10 s after the evaporation of the solvent (delay time) the temperaturewas raised with 5°C/s to 290°C. The end time was set at a time equal to the totalrun time of one analysis. Other PTV settings are presented in Table 1.

During analysis performed with GC-MSD in the total ion current (TIC)mode an m/z range of 50-300 was monitored. Using the selected ion monitoring(SIM) mode, the monitored m/z values were 86.0, 167.0, 204.0, and 256.0,which were corresponding to the most intense fragment of lidocaine,secobarbital, phenobarbital and diazepam, respectively.


75

SPE was performed as described previously [16] with some minormodifications. The SPE column was activated with 2 ml methanol (2 ml/min),followed by conditioning of the SPE column with 2 ml 0.1 M of K2HPO4 bufferpH 6 (2 ml/min). Subsequently, 1 ml plasma, diluted with 4 ml K2HPO4 buffer,was brought on the column during approximately 1 min. Then the SPE columnwas washed with 1 ml water and 0.5 ml 1 M acetic acid (1.5 ml/min). Thecolumn was dried under vacuum for 4 min, after which 50 µl of methanol werepassed through to remove remaining traces of water. The column was driedunder vacuum for 1 min. The tips of the Visiprep system were dried and tubeswere inserted for the collection of the eluate. The acidic fraction was elutedwith 1 ml ethyl acetate-acetone (1:1) (0.8 ml/min), followed by the elution ofthe alkaline fraction with 0.5 ml acetonitrile-ammonia (98:2) (0.5 ml/min). Thefractions were evaporated until almost dry and reconstituted in 100 µl ethylacetate (for organic residue analysis, Mallinckrodt Baker). Finally, 50-100 µl ofthese extracts were injected into the GC system.

Table 1: PTV settings.

GC-FID GC-MSDVent flow (ml/min) 150 150Split flow (ml/min) 57.4 57.4Purge flow (ml/min) 2.32 2.32Purge press (p.s.i.*) 8.0 4.0Transfer press (p.s.i.*) 14.0 4.0Transfer time (min:s) 2:45 2:45Initial press (p.s.i.*) 8.0 2.0Final press (p.s.i.*) 8.0 2.0Vent mode AUTO AUTOSplit open time (min:s) 2:30 2:30Threshold 20 20

*: 1 p.s.i. = 6894.76 Pa

3.1.3 Results and discussion

The basic set-up of a PTV injector strongly resembles a conventionalsplit/splitless injector. The main difference is that a (packed) liner system isapplied which is temperature-controlled. Injections up to about 150 µl can occurat once, whereas larger volumes must be injected at a controlled rate. In theinjection mode the temperature of the liner is set at 30-40°C below the boilingpoint of the used solvent. A high vent flow ensures selective evaporation ofsolvent via the split line, whereas less volatile solutes are retained in the liner.After evaporation of almost all the solvent, rapid transfer of the lattercomponents to the column is performed, using the splitless mode, by rapidlyheating the liner, or, optionally, by a high transfer pressure. During the transfer


76

of the components, the GC column is maintained at a low temperature (samestarting temperature as the injector), thus leading to a refocusing of componentsat the front of the column. Further analysis is performed with normaltemperature-programmed GC.

3.1.3.1 Purity of solvents and chemicals

Since much larger solvent volumes are injected into the PTV/GC system,the impact of solvent impurities was checked. In the plasma extractionprocedure, ethyl acetate is being used as extraction and reconstitution solvent[16]. Various qualities of ethyl acetate were tested, and some analytical resultsare presented in Fig. 2. Fig. 2A shows that a brand, which was found acceptablefor 1-µl injections, contained far too many impurities when 100 µl was injectedinto the PTV/GC system. The best results were obtained when using 100-µlinjections with the quality “for organic residue analysis” (Mallinckrodt Baker)as shown in Fig. 2B. Therefore, this latter quality was used in furtherexperiments. The purity of methanol, acetonitrile, ammonia and KH2PO4 werefound to be acceptable in that the quantities used in the present procedure didnot introduce major impurities. Attempts to purify ethyl acetate by a C18-LCcolumn or a C18 cartridge were not successful.

Fig. 2. GC-FID chromatogram of injection of 100 µl ethyl acetate (A) Reinst (Merck),(B) For organic residue analysis (Baker).


77

3.1.3.2 Liquid capacity of the liners

A liner must have a relatively large liquid capacity (Vmax) to allowinjection of 50-100 µl sample at once (no speed-controlled injection). The liquidcapacity of a liner can be easily determined by removing the column from theinjector without turning off the carrier gas. Then, 200 µl solvent are injectedrapidly and the injector outlet is checked for solvent droplets. The amount ofsolvent injected is reduced until no droplets are observed which reflects Vmax. InTable 2 the measured Vmax for ethyl acetate is given for several liners. Allpackings were of similar dimensions (height 2.5 cm, 1.8 cm from the columnside).

Table 2: Liquid capacity of liners.

Liner packing Vmax (µl)ATAS-1* 150Glass wool** 65Glass capillaries*** 40

*: Total amount of 80-85 mg ATAS ‘A’ packing in fritted liner.**: Total amount of 144 mg glass wool was inserted into a open liner.***: 9 TLC capillaries of 10 µl and 2 TLC capillaries of 2 µl were cut at a length of 2 cm, andinserted into a fritted liner. A small plug of glass wool was placed under and above the capillaries.

The glass wool and the ATAS "A" liner appeared to be best suitable forthe injection of large volumes ethyl acetate (>50 µl), and the ATAS "A" linercan even be used for samples larger than 100 µl. Mol et al. [17] found a Vmax of115 µl for a liner packed with glass wool instead of 65 µl. This difference isprobably due to the major problem with glass wool, that is, inserting thepacking into the liner in a reproducible way [17,18].

3.1.3.3 Inertness of the liner packings

Comparison of packingsIt was necessary to investigate the inertness of liner packing materials for

biological samples since a high recovery of the analytes must be obtainedduring consecutive sample injections. Several ATAS "A" liners (Nos. 1-5), aglass wool packed liner (No. 6) and a liner packed with open capillaries (No. 7)were tested. Cutting glass capillaries can create active places at the cutting site,and glass wool is known to have a limited inertness [17]. The ATAS "A" linerwas originally designed for the analysis of pesticides and mineral oils. Theinertness of the ATAS "A" liner can be influenced by a high temperature sincethe packing has a Tmax of 325°C. Higher temperatures will degrade the packing,


78

which will have a negative effect on the inertness of the liner. Analysis of highboiling compounds (>325°C) is therefore not possible.

Prior to testing the inertness, liners 1 and 2 were used for many injectionsof standard solution and plasma extracts. The colour of liner 1 had changedfrom white to completely brown, and the upper half of the packing of liner 2had changed to brown with the lower half still white. Liners 3-7 were not usedbefore. Injection of 1 µl of 250 µg/ml test compounds (secobarbital, lidocaine,phenobarbital and diazepam) into a liner with a glass frit without packingproduced the reference chromatogram, that is, since no active packing waspresent, the response of the compounds was set at 100%. Injections of 1 µl of250 µg/ml or 100 µl 2.5 µg/ml test compounds into liners 3, 4, and 5 producedthe same responses, thus the ATAS "A" liner can be considered to be inert if theliner packing is not previously used for analysis.

In Table 3 the responses are tabulated for liners 1, 2, 6 and 7, as comparedto liners 3, 4 and 5. Liners 1, 2, 6 and 7 showed adsorption activity forphenobarbital, liner 1 and 6 being the most active. Injection of 1 µgphenobarbital into liner 1 produced even no peak. For secobarbital, liner 1 isvery active whereas liners 2, 6 and 7 are much less active. For lidocaine, liners1 and 2 showed limited activity and the same was observed for diazepam. Fromthese results, it appears that adsorption losses are more pronounced in the orderdiazepam/lidocaine<secobarbital<phenobarbital. The change in colour seems agood parameter to indicate adsorption activity of the ATAS "A" liner forbarbiturates whereas it has hardly any effect on the response of lidocaine anddiazepam.

Table 3: Recovery (%) of 250 ng of compounds compared with average response ofliner 3, 4, and 5 (set at 100%). Liners: 1-5 = ATAS ‘A’ (No.1 completelybrown, No. 2 upper half brown, No. 3-5 white), 6 = silanized glass wool,7 = glass capillaries.

Liner Secobarbital Lidocaine Phenobarbital Diazepam1 3 86 0 872 96 92 59 963,4,5 100 100 100 1006 81 99 17 1007 88 97 49 100

Glass wool packed liners were shown to adsorb fatty acids from 42 to100% [17]. In this work glass wool also showed adsorption activity for theweakly acidic compounds secobarbital and phenobarbital, but the materialappears to be better suitable for weakly basic drugs. Thus, a glass wool liner canbe used for some of the drugs investigated in this work whereas the glass woolliner used by Mol et al. [17] was unsuitable for the analysis of all tested


79

compounds. Biedermann et al. [19] pointed out that the deactivation ofvaporising chambers can also play an important role. Deactivating of the glasssurface was performed before the packing material was inserted into the liner.In order to investigate if there are other classes of drugs for which theATAS "A" packing is not inert, a reference mixture [15] was analysed usingliner 1, 2 and 5. The results are presented in Fig. 3.

Fig. 3: Effect of packing inertness on response of compounds of RI-mix [15].¨ = liner 1, = liner 2, n = liner 5. Compounds: 1 = amphetamine,2 = ephedrine, 3 = benzocaine, 4 = methylphenidate, 5 = diphenhydramine,6 = tripelenamine, 7 = methaqualone, 8 = trimipramine, 9 = codeine,10 = nordazepam, 11 = prazepam, 12 = papaverine, 13 = haloperidol,14 = strychnine.

The percentages given below are recoveries with the response of the compoundduring analysis with liner 5 set at 100%. No effect on the response was foundfor amphetamine, ephedrine, diphenhydramine, tripelenamine and trimipramine.Liner 1 showed a small decrease in recovery for benzocaine (90%),methaqualone (96%), nordazepam (95%), prazepam (85%) and haloperidol(80%). Liner 2 was not active for these substances. The same was observed withliner 2 for strychnine, but liner 1 was active for this compound (46%). Theeffect of liner activity for methylphenidate (liner 2 and 1, 82 and 5%,respectively), codeine (90 and 28%), and papaverine (38 and 5%) wascomparable with that for phenobarbital. Thus, the ATAS "A" liner can becomeless inert to substances due to injection of plasma extracts and/or standardsolution. The results confirm the suggestion that a beginning brown colouration


80

can be used as an indication that the liner is starting to loose its inertness andthat it needs replacement. From the results with both the test compounds as wellas the compounds of the RI-mixture, it can be concluded that barbiturates andopium alkaloids are particularly prone to recovery losses. On the other hand,even a completely brown ATAS "A" liner has little effects on the recoveries ofbenzodiazepines.

Long-term useThe ATAS "A" liner and the glass wool liner have a sufficient liquid

capacity to allow injections up to 50 µl. However, the glass wool liner appearedto be less inert for some type of compounds. Therefore, only the inertness of theATAS "A" liners as a function of the number of injections was alsoinvestigated. Two new ATAS "A" liners were used. For one liner only astandard solution (50 µl of 2.5 µg/ml) was injected 35 times (Fig. 4). The linerappeared to remain inert under these conditions since no loss in response wasobserved for all test compounds. For the second liner, 50 µl of 2.5 µg/mlstandard solution with two injections of plasma extract (one alkaline and oneacidic fraction) between subsequent injections of the standard solution wereanalysed. The response of the test compounds (secobarbital, phenobarbital,lidocaine and diazepam) in the standard solution is plotted as a function of thenumber of plasma injections (Fig. 4).

Since up to 35 injections of standard solution had no influence on thestability of the liner packing, a decrease in response must have been caused bythe plasma extracts influencing the inertness of the packing material. Injectionof plasma extracts introduces a large decrease in inertness of the packing forphenobarbital starting with a slightly variable response at 14 injections ofplasma extracts, and a definite loss in response after 20 injections. Forsecobarbital, after 20 injections of plasma extracts only a small increase inactivity of the ATAS "A" packing is observed. Injection of 32 plasma extractshas no effect on the response of diazepam and lidocaine (lidocaine not shown).An increase in activity of the packing was again found to correlate with thecolour of the packing material. Inert ATAS "A" material is white but thisbecomes brown with increasing activity. After some 10-15 injections of plasmaextracts the colour of the ATAS "A" packing started to change from white tobrown, and the brown colour became more apparent on continued analysis ofplasma extracts. The decrease in inertness and change in colour of theATAS "A" packing when plasma extracts are injected is probably due to thedegradation of matrix components that are not desorbed from the liner onheating.


81

Fig. 4: Stability of ATAS-1 packing. Response of 125 ng test compound in 50 µlstandard solution of (A) diazepam, (B) secobarbital, (C) phenobarbital;u = response for injections of only the standard solution, n = response forinjections of the standard solution with two plasma extract injections betweensubsequent injections of the standard solution (number of injectionscorresponds with the amount of plasma extract injections).


82

Carry-overCarry-over was checked by injecting large amounts of the test compounds

or RI-mixture (up to 2.5 µg), followed by a second injection in which a blank,i.e. 100 µl ethyl acetate, was introduced. No carry-over was observed for eitherthe test compounds or the components of the RI-mixture when using anATAS "A" liner.

3.1.3.4 Linearity with PTV/GC-FID for standard solutions

Since the ATAS "A" liner has a relatively large liquid capacity, is inertfor the tested compounds and no carry-over occurs, the ATAS "A" liner issuitable for LVI of bioanalysis. Linearity in response for lidocaine, diazepam,secobarbital and phenobarbital was determined for variation in volume andconcentration. For the determination of the linearity in response when thevolume is varied, volumes varying from 20 to 100 µl of a 1.0 µg/ml standardsolution were injected on a new ATAS "A" liner. It was found that allcompounds showed a good linearity (Table 4).

The linearity of response versus concentration was determined byinjections of 100 µl standard solution over a concentration range of 5 to2000 ng/ml. As can be observed in Table 4, a good linearity was obtained. Oneshould note that no internal standard was applied to correct for injection volumeand signal drift.

Table 4: Linearity (coefficient of correlation (R)) of secobarbital, lidocaine,phenobarbital, and diazepam with variation of volume (20-100 µl of 1 µg/ml)and concentration (5-2000 ng/ml).

R (volume) R (concentration)Secobarbital 0.9989 0.9974Lidocaine 0.9937 0.9965Phenobarbital 0.9979 0.9923Diazepam 0.9976 0.9982

3.1.3.5 PTV/GC-FID and PTV/GC-MSD of plasma extracts

Linearities as well as the detection limits for lidocaine, diazepam,secobarbital, and phenobarbital in plasma extracts were determined using FIDand MSD. With LVI a large amount of matrix components and solventimpurities is injected. Therefore peak identification can become difficult whennon-selective detectors are used. Use of a mass selective detector may help toovercome this problem. With the mass selective detector analyses wereperformed in both the TIC and the SIM mode.


83

Linearity was determined for concentrations ranging from the detectionlimit up to 100 ng using 1 ml of spiked plasma extract. The results are listed inTable 5. For all detectors linearity was found to be better for the alkalinefraction than the acidic fraction, except for diazepam. For this compound,coefficients of correlation are comparable for both fractions. The acidic fractionanalysed with MSD shows a relatively low linearity for phenobarbital. Thismight be due to an increase in activity of the GC system during analysis (seeSection 3.1.3.3).

Table 5: Linearity (coefficient of correlation) of secobarbital, lidocaine, phenobarbital,and diazepam in acidic and alkaline SPE fraction of plasma; range: detectionlimit (see text below) to 100 ng/ml.

RFID(acidic)

FID(alkaline)

TIC(acidic)

TIC(alkaline)

SIM(acidic)

SIM(alkaline)

Secobarbital 0.9963 0.9981 0.9940 0.9976 0.9997 0.9998Lidocaine 0.9936 0.9992 0.9986 0.9962 0.9991 0.9997Phenobarbital 0.9947 0.9996 0.9993 0.9958 0.9719 0.9992Diazepam 0.9975 0.9962 0.9969 0.9953 0.9991 0.9994

The detection limits were determined at a signal-to-noise ratio of 3 forboth FID and MSD. It should, however, be mentioned that when using FID andMSD in the TIC mode, interfering matrix compounds and solvent impuritiescan make the determination of the detection limit laborious as blank peaks havea negative influence. Blank plasma was extracted and the extracts were spikedwith the test compounds and detection limits for plasma extracts werecalculated assuming a 100% recovery of the test substances in the SPEprocedure. The actual recoveries using this SPE method were found to be80-100% [16].

Chromatograms of 40-45 ng compounds in the alkaline fraction analysedwith FID, TIC and SIM are presented in Fig. 5. Using FID, the detection limitusing 1 ml of plasma is 5-10 ng for all compounds in both the acidic andalkaline fraction. The detection limits observed using the TIC mode of the MSDare 4-5 ng. Therefore, a small gain in detection limit can be achieved. This ismainly due to the fact that with TIC a positive identification can be given forthe peaks present in a sample since reliable mass spectra are obtained with goodcorrespondence with library spectra. Acidic plasma extracts analysed in the SIMmode give a detection limit of 0.5 ng for each test compound whereas thealkaline fraction gives a detection limit of 0.25 ng.

The gain in sensitivity when compared with FID or TIC can be explainedby the enhanced selectivity since detection occurs only at four m/z values. If thecomplete sample is injected, plasma concentrations as low as 250 pg/ml could


84

be detected in the alkaline fraction. With conventional GC, that is injection of1 µl plasma extract, also a detection limit of 0.25 ng for the test compounds isfound. As a consequence, the corresponding plasma concentration is 50 timeshigher than with injection of 50 µl, i.e. a concentration of 12.5 ng/ml can bedetected.

Fig. 5. Chromatograms of 50 µl of the alkaline fraction of plasma extracts analysedwith (A) GC-FID (41.7 ng), (B) GC-MSD; TIC mode (45.5 ng), (C) GC-MSD;SIM mode (45.5 ng). The monitored m/z values were 86.0, 167.0, 204.0 and256.0 for lidocaine (L), secobarbital (S), phenobarbital (P) and diazepam (D),respectively. Note: the peak (*) eluting just after the diazepam peak (D) iscaused by an impurity with an m/z value similar to phenobarbital.


85

3.1.4 Conclusions

It was demonstrated that the PTV is potentially suitable to inject largevolumes of extracts of biological samples in GC. In this way detection limitscan be improved considerably. In order to be able to apply LVI in GC it isnecessary to select an appropriate liner packing. As for the liner packings testedin this work the ATAS "A" packing was found to be the most suitable since thismaterial has a large liquid capacity and is relatively inert. However, the packingcan become active after a number of injections of plasma extracts for certaintypes of compounds. It is therefore recommended to carefully monitor theinertness of the packing material by injection of a few suitable compounds andto monitor the colour of the liner packing.

Injection of large volumes implies that an equivalent amount of impuritiesis injected. Therefore, it is essential to use very pure solvents and chemicalsduring the work-up procedure. However, not only solvent impurities areinjected. Also matrix components that are co-extracted with the analytes makeidentification and quantitation difficult when a non-selective detector is used.The use of a selective detector is essential to overcome this problem. Using amass selective detector, a 100 times gain in concentration-sensitivity can beachieved if 100 µl of a plasma extract instead of 1 µl is injected.

The present system can be used as a routine technique in research andclinical laboratories. However, further evaluation of the system for variouspurposes (including other matrices) and different types of compounds is needed.The on-line coupling of SPE and GC for bioanalysis will also be investigated inthe near future in our laboratory.

Acknowledgements

Jan Henk Marsman and Ronald Veenhuis (Department of ChemicalEngineering, University of Groningen) are gratefully acknowledged for the useof the GC-MSD system and their assistance. This research was supported by theTechnology Foundation STW, applied science division of NWO and thetechnology programme of the Ministry of Economic Affairs.

3.1.5 References

[1] H.G.J. Mol, H.-G. M. Janssen, C.A. Cramers, J.J. Vreuls, U.A.Th. Brinkman.J. Chromatogr. A 703 (1995) 277.

[2] K. Grob, J.-M. Stoll. J. High Resolut. Chromatogr. Commun. 9 (1986) 518.[3] W. Vogt, K. Jacob, H.W. Obwexer. J. Chromatogr. 174 (1979) 437.


86

[4] W. Vogt, K. Jacob, A.-B. Ohnesorge, H.W. Obwexer. J. Chromatogr. 186 (1979)197.

[5] K. Grob. J. Chromatogr. A 703 (1995) 265.[6] P. Van Zoonen, G.R. van der Hoff. LC-GC Int. 16 (1998) 240.[7] T. Hyötyläinen, M.-L. Riekkola. J. Chromatogr. A 819 (1998) 13.[8] M.P.M. van Lieshout, H.-G. Janssen, C.A. Cramers, G.A. van den Bos.

J. Chromatogr. A 764 (1997) 73.[9] J.J. Vreuls, U.A.Th. Brinkman, G.J. de Jong, K. Grob, A. Artho. J. High Resolut.

Chromatogr. 14 (1991) 455.[10] J.J. Vreuls, G.J. de Jong, R.T. Ghijsen, U.A.Th. Brinkman. J. Microcol. Sep. 5

(1993) 317.[11] A.J.H. Louter, J. Van Doornmalen, J.J. Vreuls, U.A.Th. Brinkman. J. High

Resolut. Chromatogr. 19 (1996) 679.[12] R.J.C.A Steen, I.L. Freriks, W.P.Cofino, U.A.Th. Brinkman. Anal. Chim. Acta

353 (1997) 153.[13] J. Teske, J. Efer, W. Engewald. Chromatographia 47 (1998) 35.[14] H.J. Stan, M. Linkerhagner. J. Chromatogr. A 750 (1996) 369.[15] R.A. de Zeeuw, J.P. Franke, H.H. Maurer, K. Pfleger. Gas chromatographic

retention indices of toxicologically relevant substances on packed and capillarycolumns with dimethylsilicone stationary phases, third edition, VCH, Weinheim,1992.

[16] X.-H. Chen, J. Wijsbeek, J.P. Franke, R.A. de Zeeuw. J. Forensic Sci. 37 (1992)61.

[17] H.G.J. Mol, P.J.M. Hendriks, H.-G. Janssen, C.A. Cramers, U.A.Th. Brinkman.J. High Resolut. Chromatogr. 18 (1995) 124.

[18] H.G.J. Mol, H.-G. Janssen, C.A. Cramers, U.A.Th. Brinkman. J. High Resolut.Chromatogr. 18 (1995) 19.

[19] M. Biedermann, K. Grob, M. Wiedmer. J. Chromatogr. A 764 (1997) 65.

Coupling device for desorption of drugs from SPEPTs into GC

87

3.2

Coupling device for desorption of drugs fromsolid-phase extraction – pipette tips and on-linegas chromatographic analysis*

Summary

Solid-phase extraction – pipette tips (SPEPTs) were used for micro solid-phaseextraction of lidocaine and diazepam. Off-line desorption was done after in-vialcollection for reference purposes, whereas with on-line desorption the eluatewas directly introduced in the gas chromatograph. With both methods the totaleluate (100 µl) was introduced into the GC, which was equipped with aprogrammed temperature vaporiser (PTV) for large-volume injection. Foron-line desorption a laboratory-made coupling device was developed to connectthe pipette tips with the injector of the PTV. The coupling device was appliedsuccessfully since no leakage occurred at the connection of the coupling deviceand the pipette tip. No significant differences in recovery of lidocaine anddiazepam and in presence of impurities were observed between chromatogramsobtained with either off-line or on-line desorption. Preliminary experimentswith standard solutions showed recoveries of about 75 % for a concentrationlevel of 1 µg/ml. The system seems particularly suitable for high-throughputanalysis.

*: M.W.J. van Hout, R.A. de Zeeuw, G.J. de Jong. J. Chromatogr. A 858 (1999) 117-122.


88

3.2.1 Introduction

High sample throughput is becoming increasingly important in variousareas of bioanalysis. Therefore, it is essential to reduce analysis time includingsample pre-treatment. A very popular sample pre-treatment technique issolid-phase extraction (SPE), which was originally developed as an off-linesample clean-up and pre-concentration procedure [1-7]. In order to obtain highsample throughput, there is a growing interest in SPE with at-line and on-lineliquid chromatography [8-11] and, more recently, gas chromatography (GC)[12-19]. In addition, it has been claimed that the various steps in on-line SPEcan be carried out with greater precision than in off-line SPE, resulting in morereliable data. Another advantage of on-line SPE-GC is that the total eluate canbe analysed, which was not possible with off-line SPE. However, withlarge-volume injection (LVI) of extracts of biological samples in gaschromatography via a retention gap [13,19] or a programmed temperaturevaporiser (PTV) [20], it is possible to inject nearly the total eluate of off-lineextractions. Yet, with LVI, special attention must be given to solvent purity andselectivity of the extraction procedure [20], since with the injection of largevolumes an equivalent amount of impurities is also injected into the analysingequipment.

Another interesting development is the miniaturisation of analyticalsystems. The first attempts to miniaturise SPE were done using SPE disksinstead of conventional SPE cartridges [5,6,9-11,21-28]. Generally, SPE diskscontain a smaller bed with smaller particles and a more homogeneous particlesize distribution than conventional cartridges. An advantage of SPE disks overSPE cartridges is the possibility to use smaller desorption volumes (50-400 µlvs. 1-6 ml) [21,23]. Another benefit besides the use of less solvent is that ifLVI/GC is applied in combination with miniaturised SPE no evaporation andreconstitution of the extracts is required, which eliminates an error-prone step inthe extraction procedure, and thus increases reliability and reduces samplepreparation time. Moreover, solvent purity is less critical than when LVI isapplied in combination with conventional SPE, because less desorption solventis used.

Further miniaturisation has led to micro-SPE, which can be performedusing solid-phase extraction – pipette tips (SPEPTs) [7,29]. Extractions can becarried out more easily and rapidly than with conventional SPE or with SPEdisks simply by using a pipettor and pipette tips with extraction material insidethe tip [29]. An interesting aspect of extractions with pipette tips is thatbi-directional flow and cycling, that is aspirating and dispensing, can be applied[7,29]. However, micro-SPE may imply that smaller sample volumes have to beused which leads to higher concentration detection limits. This loss insensitivity may be overcome by applying LVI/GC, which may even lead to an


89

increase in sensitivity when compared to conventional injection volumes of 1 to2 µl [20].

In order to reduce the analysis time and increase the reliability of theresults, an on-line coupling of micro-SPE with GC appears to be attractive. TheGC system must be equipped with a special injector, e.g., a PTV, so that LVI ispossible. In this study a coupling device for the connection of pipette tips andthe PTV injector was developed and evaluated for micro-SPE-GC, that is, theextraction was carried out off-line but the desorption and subsequent GCanalysis was done on-line.

3.2.2 Experimental

3.2.2.1 Apparatus and chromatographic conditions

Gas chromatographic analyses were performed with a Hewlett-PackardHP 5890 series II instrument with flame-ionisation detection (FID). Thecapillary column was a HP-5 30 m×0.32 mm with 0.25 µm film thickness.Helium was used as carrier gas. The following temperature program was usedfor the GC. The starting temperature was 40°C and after 3 min the temperaturewas raised at 20°C/min to 215°C, followed by a raise of 5°C/min to 230°C anda final raise of 25°C/min to 290°C. This final temperature was maintained for 5to 10 min. The detector temperature was set at 300°C, and a column flow of1.1 ml/min was used during analysis.

The PTV injection system was an OPTIC 2 (ATAS International,Veldhoven, The Netherlands), which was equipped with a 80 mm×3.4 mm i.d.liner obtained from ATAS International. The liner was packed with ATAS "A"packing (a modified Chromosorb-based material with special treatment). Theinjector was set at 40°C in the vent mode and evaporation of the solventoccurred using the “AUTO vent mode” with a vent flow of 150 ml/min. Afterthe evaporation of the solvent the valve was switched to the splitless mode andafter 10 s the temperature was raised with 5°C/s to 290°C. This finaltemperature was maintained during the analysis. The splitless mode was appliedfor 2.50 min and, subsequently, the valve was switched to the split mode. Theused split flow was 57.4 ml/min, whereas the purge flow and pressure were2.32 ml/min and 2.5 p.s.i., respectively (1 p.s.i.=6894.76 Pa). A transferpressure of 14.0 p.s.i. was applied for 2.75 min. During the analysis the initialand final pressure were maintained at 8.0 p.s.i.


90

3.2.2.2 Chemicals

Methanol (Lab Scan, Dublin, Ireland) was of HPLC quality. KH2PO4 wasof analytical-reagent quality (Merck, Darmstadt, Germany). Ethyl acetate Ultraresi-analysed (for organic residue analysis) was purchased from MallinckrodtBaker (Deventer, The Netherlands). Water used during SPE was ultra pure(Elgastat maxima, Salm en Kipp, Breukelen, The Netherlands). Lidocaine(Eur. Ph., Holland Pharmaceutical Supply, Alphen A/D Rijn, The Netherlands)and diazepam (Centrafarm, Etten-Leur, The Netherlands) were used as testcompounds and dissolved in ethyl acetate (for organic residue analysis,Mallinckrodt Baker) or in phosphate buffer pH 8.0. Stock solutions of 1 mg/mlwere stored in the dark at 4°C.

3.2.2.3 Methods

Micro-SPE was performed using pipette tips (SPEC•PLUS•PT) with aC18-AR stationary phase (Ansys Diagnostics, Lake Forest, CA, USA). The SPEprocedure was carried out by connecting a 10-ml gas-tight plastic syringe(Omnifix syringe, B. Braun, Melsungen, Germany) (Fig. 1(A)) to the pipette tip(B). The SPE disk (C) in the pipette tip was activated with ca. 200 µl methanolfollowed by conditioning of the disk two times with ca. 100 µl of 0.1 MK2HPO4 buffer (pH 8.0). Subsequently, 200 µl phosphate buffer spiked with1 µg/ml lidocaine and diazepam were extracted on the disk. Then the disk waswashed with ca. 100 µl water and, subsequently, dried by pushing air throughthe disk (10×10 ml). For the desorption of the analytes from the disk 100 µlethyl acetate were used. The desorption occurred in-vial or on-line. The solventsand samples were drawn into the pipette tip until the fluid had gone through thedisk and then the fluid was completely pushed back out of the pipette tip(bi-directional flow). The fluids did not enter the plastic syringe, thus thesyringe could be used for subsequent extractions.

On-line desorption was possible by connecting the pipette tip to the PTVinjector via a laboratory-made coupling device as depicted in Fig. 1. Thecoupling device (D) was made by replacing the glass from a gas-tight GCsyringe by a piece of PTFE (E). All other parts (F-I) of the coupling device arealso present in conventional GC syringes with a removable needle (I). Thedimensions of the PTFE piece were 5.5 mm×9 mm o.d. (2.0 mm i.d.) (upperhalf) and 6 mm×6 mm o.d. (0.5 mm i.d.) (lower half, inside the screw thread).


91

Fig. 1. Scheme of the coupling device of the solid-phase extraction-pipette tip and thePTV injector. Parts: A = plastic syringe, B = SPEC•PLUS•PT pipette tip,C = SPE disk, D = coupling device, E = PTFE piece, F = standard metal screwthread, G = vulcanised rubber, H = standard needle nut, I = needle.


92


The use of pipette tips offered the possibility of bi-directional flow andcycling (aspirating and dispensing) [7,29]. Visualisation of these phenomenawas described by Blevins and Hall [29]. A plastic syringe was used in ourapproach instead of a conventional pipettor since the solvents and samplescould be better drawn into and pushed back out of the pipette tip with the plasticsyringe. However, bi-directional flow should not be applied with air, since arapid flow of air in upward direction may dislodge the disk. Therefore, upondrying the disk with air, only dispensing is possible. Generally, aspirating anddispensing solvents was carried out gently to make sure that not too much airwent through the disk with the exception of the final part of the dispension inthe elution step. Ethyl acetate was pushed back completely followed by air toremove as much ethyl acetate as possible from the disk.

During the development of the coupling device for on-line desorption thedimensions of the PTFE piece were chosen so that it ensured to fit just right inplace of the removed glass and that approximately half of the tip of the pipettetip could be inserted tightly into the coupling device. When desorption wascarried out on-line, the needle of the coupling device was completely insertedinto the PTV injector. Subsequently, ethyl acetate was drawn into the pipettetip, and then the pipette tip was immediately placed on top of the couplingdevice and the ethyl acetate was pushed back through the disk directly into thePTV injector. The coupling device and pipette tip were removed at the sametime, so that no leakage of carrier gas and solvent vapour occurred through theneedle of the coupling device.

In order to be able to inject the extract on-line into the GC systemsubstantial pressure must be applied during the push-back since a highback-pressure is present when the pipette tip and coupling device are attached tothe PTV injector. The back-pressure can be lowered slightly by decreasing thepurge pressure of the PTV injector. However, lowering the purge pressure leadsto an increase of vent time, and thus to an increase of analysis time. Theback-pressure is mainly caused by the small inner diameter of the needle(0.1 mm) of the coupling device. Despite the back-pressure no leakage occurredwhen desorption was carried out on-line. Due to the back-pressure the diskcould not be dried completely. With SPEPTs, bi-directional flow is appliedwhich means that the desorption solvent that flows through the disk first, leavesthe pipette tip last (“first in, last out-principle”). Hence, the ethyl acetate thatremains in the disk may contain a part of the analytes. Thus, if the disk cannotbe dried completely with on-line desorption, this may cause a decrease in therecovery of the analytes. The decrease in recovery due to the ethyl acetate thatremained in the disk was found to be 5%. Thus, due to mixing of the desorptionsolvent inside the pipette tip the loss caused by the “first in, last out principle” is


93

only small. In order to be able to quantify reliably the analytes present in theextracts an internal standard can be added to the ethyl acetate to correct for theloss of analytes that remain in the disk during on-line desorption. Usinguni-directional flow during desorption, that is only dispensing with desorptionsolvent, might prevent the possible loss of analytes with the “first in, last outprinciple”. However, this has the disadvantages that the pipettor and pipette hasto be disconnected and that desorption solvent can be present in the upper partof the pipette tip which has to be reconnected prior to injection and, so, cancontaminate the pipettor.

Chromatograms of in-vial and on-line desorption are presented in Fig. 2.No significant differences in peak heights of lidocaine and diazepam wereobserved between an analysis performed with in-vial desorption (referencepurposes) and one in which on-line desorption took place. With in-vialdesorption somewhat more impurities were present than with on-linedesorption. Optimisation of the extraction procedure with regard toconditioning, activation and washing solvents and volumes may producecleaner extracts. Besides solvent impurities the presence of impurities in theextracts might also be due to interferences that are being leached from thesorbent material or the sorbent holder. During some extractions white spotsappeared inside the pipette tip during the drying step and disappeared whenethyl acetate was drawn into the pipette tip. The white spots, which probablyoriginate from the stationary phase, seem to produce clusters of peaks in thechromatograms.

Fig. 2. Chromatograms of extracts of 200 µl phosphate buffer containing 200 nglidocaine (L) and diazepam (D) after desorption with 100 µl ethyl acetate:(A) in-vial desorption, (B) on-line desorption.

Recoveries of lidocaine and diazepam were about 75%. It should benoticed that the SPE procedure has still to be optimised, which may result inhigher recoveries. After elution of a disk with ethyl acetate, the carry-over from


94

the coupling device was checked by injection of 100 µl ethyl acetate via apipette tip without SPE disk and was found to be negligible.

3.2.4 Conclusions and perspectives

It is possible to perform an extraction off-line with SPEPTs and desorbthe analytes with on-line GC analysis with a laboratory-made coupling device toconnect the pipette tip with the PTV injector. This allows the injection of thecomplete eluate (100 µl). Since no significant difference between in-vial andon-line desorption and negligible carry-over from the coupling device isobserved, it is advantageous to desorb the analytes on-line, because thisincreases the rapidity and reliability of the extraction procedure.

The coupling device needs further optimisation and evaluation withregard to robustness of the system. The dimensions of the PTFE piece can bechanged so that a more optimal connection between the coupling device and apipette tip can be obtained. The inner diameter of the needle can be increasedwhich will result in a lower back-pressure when the eluent is pushed backthrough the coupling device into the PTV injector. However, increase of theinner diameter of the needle will also give an enhanced flow through the needle,which might result in leakage of eluate at the connection of the coupling deviceand the pipette tip.

Preliminary experiments have shown that the total set-up can also be usedfor plasma samples. The SPE procedure has still to be optimised with regard tosolvent and sample volumes and, if available, other SPE stationary phases,which might produce cleaner extracts and higher recoveries. Another possibilityto reduce the interference of impurities in the solvents and plasma extracts is theuse of more selective detectors, such as mass spectrometry [20] or anitrogen-phosphorous detection.

Acknowledgements

SPEC•PLUS•PT pipette tips were kindly provided by Ansys Diagnostics,Inc. This research was supported by the Technology Foundation STW, appliedscience division of NWO and the technology programme of the Ministry ofEconomic Affairs.


95

3.2.5 References

[1] M. Dressler. J. Chromatogr. 165 (1979) 167.[2] A. Lagana, B.M. Petronio, M. Rotatori. J. Chromatogr. 198 (1980) 143.[3] R.E. Majors, H.G. Barth, C.H. Lochmüller. Anal Chem. 56 (1984) 300R.[4] X.H. Chen, J.P. Franke, R.A. de Zeeuw. Forensic Sci. Review 4 (1992) 147.[5] Z.E. Penton. Advances in Chromatogr. 37 (1997) 205.[6] J.P. Franke, R.A. de Zeeuw. J. Chromatogr. B 713 (1998) 51.[7] R.E. Majors. LC-GC Int. May (1998) 8.[8] R.W. Frei, K.Zech (Eds.). Selective Sample Handling and Detection in

High-Performance Liquid Chromatography, Part A, Elsevier, Amsterdam 1988.[9] E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman. Chromatographia 29 (1990)

415.[10] E.R. Brouwer, D.J. van Iperen, I. Liska, H. Lingeman, U.A.Th. Brinkman. Intern.

J. Environ. Anal. Chem. 47 (1992) 257.[11] E.H.R. van der Wal, E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman.

Chromatographia 39 (1994) 239.[12] E.C. Goosens, D. de Jong, G.J. de Jong, U.A.Th. Brinkman. Chromatographia 47

(1998) 313.[13] A.J.H. Louter, E. Bosma, J.C.A. Schipperen, J.J. Vreuls, U.A.Th Brinkman.

J. Chromatogr. B, 689 (1997) 35.[14] A. Namera, M. Yashiki, Y. Iwasaki, M. Ohtani, T. Kojima. J. Chromatogr. B 716

(1998) 171.[15] K.K. Verma, A.J.H. Louter, A. Jain, E. Pocurull, J.J. Vreuls, U.A.Th. Brinkman.

Chromatographia 44 (1997) 372.[16] D. Jahr. Chromatographia 47 (1998) 49.[17] A. Namera, M. Yashiki, K. Okada, Y. Iwasaki, M. Ohtani, T. Kojima.

J. Chromatogr. B 706 (1998) 253.[18] P. Enoch, A. Putzler, D. Rinne, J. Schlüter. J. Chromatogr. A 822 (1998) 75.[19] A.J.H. Louter, R.A.C.A. van der Wagt, U.A.Th. Brinkman. Chromatographia 40

(1995) 400.[20] M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong. J. Chromatogr. B

729 (1999) 199.[21] A. Koole, A.C. Jetten, Y. Luo, J.P. Franke, R.A. de Zeeuw. J. Anal. Toxicol. 23

(1999) 632.[22] H. Lingeman, S.J.F. Hoekstra-Oussoren. J. Chromatogr. B 689 (1997) 221.[23] D.A. Wells, G.L. Lensmeyer, D.A. Wiebe. J. Chromatogr. Sci. 33 (1995) 386.[24] S. Rudaz, W. Haerdi, J.L. Veuthey. Chromatographia 44 (1997) 283.[25] K. Hartonen, M.L. Riekkola. J. Chromatogr. B 676 (1996) 45.[26] K. Ensing, J.P. Franke, A. Temmink, X.H. Chen, R.A. de Zeeuw. J. Forensic Sci.

37 (1992) 460.[27] D.F. Hagen, C.G. Markel, G.A. Schmitt, D.D. Blevins. Anal. Chim. Acta 236

(1990) 157.[28] P.J.M. Kwakman, J.J. Vreuls, U.A.Th. Brinkman, R.T. Ghijsen.

Chromatographia 34 (1992) 41.[29] D.D. Blevins, D.O. Hall. SPEC NEWS 3-1 (1998) 1.

Feasibility of the direct coupling of SPEPTs with PTV/GC

97

3.3

Feasibility of the direct coupling ofsolid-phase extraction – pipette tips with aprogrammed temperature vaporiser forgas chromatographic analysis of drugs in plasma*

Summary

Solid-phase extraction – pipette tips (SPEPTs) were used for micro solid-phaseextraction of lidocaine and diazepam from plasma. Off-line extraction wasfollowed by on-line desorption. On-line desorption was carried out by directcoupling of the SPEPTs with the liner of the programmed temperaturevaporiser. This coupling only required shortening of the liner by maximally 16mm, cutting the SPEPT, and equipping the remaining part with two O-rings.Due to the heating of the injector the SPEPTs were heated as well, whichresulted in a significant amount of impurities. Pre-heating and pre-washing wasperformed prior to the extraction to reduce the impurity level. The internalcoupling device was applied successfully for the analysis of plasma sampleswith gas chromatography (GC) and mass selective detection. Detection limits of0.75 ng/ml and 2.5 ng/ml were obtained for lidocaine and diazepam,respectively, using 200 µl plasma. Recoveries for both compounds were about80%. Although it is possible, the internal coupling device was not developed tobe used as such. The main goal of this coupling was to show the feasibility ofthe integration of SPEPTs with GC and to realise an important step to newautomated SPE-GC systems.

*: M.W.J. van Hout, W.M.A. van Egmond, J.P. Franke, R.A. de Zeeuw, G.J. de Jong. J. Chromatogr. B 766(2002) 37-45.


98

3.3.1 Introduction

Increasing knowledge of working mechanisms of drugs leads to thecoming on the market of more potent drugs. As a result, administered dosagesare decreasing. In order to be able to determine these drugs in lowconcentrations in biological matrices, more sensitive techniques are required. Apossibility to increase the sensitivity in a gas chromatographic (GC) system is toincrease the injected sample volume. Several techniques are available toperform large-volume injection (LVI) in GC [1]. On-column injection withretention gaps is a common technique [1]. A second option to allow LVI is theloop-type interface [2], which was originally designed for the coupling of liquidchromatography and GC. A third possibility for LVI is to use a programmedtemperature vaporiser (PTV). The PTV has been mainly applied forenvironmental analysis [1,3-6], although its potential for the analysis ofbiofluids has been explored as well [7]. With LVI, special attention must begiven to solvent purity and selectivity of the extraction procedure. Largevolumes imply the injection of an equivalent amount of impurities into theanalysing equipment [7].

Biological samples cannot be introduced directly into the GC.Furthermore, the decreasing concentrations of drugs in biological samplesrequire pre-concentration. For these purposes solid-phase extraction (SPE) isvery suitable. Originally, SPE was developed as an off-line sample clean-up andpre-concentration procedure [8-12]. In order to obtain high sample throughput,SPE can be coupled at-line and on-line to GC [13-18]. With on-lineSPE-LVI/GC, the aim is to introduce the total eluate of the SPE system into theGC system, thus increasing the sensitivity of the system. Furthermore, greaterprecision is obtained than with off-line SPE, since an error-prone step in theextraction procedure is eliminated. However, the critical aspect remains theamount of eluate that the GC system is able to accept. Moreover, the mainlimitation of the present on-line systems is the long drying step, which typicallytakes 10-30 min.

Miniaturisation of SPE has led to the development of SPE disks.Generally, SPE disks contain a small bed with small particles and have ahomogeneous particle size distribution [12,19-21]. An advantage of SPE disksis the possibility to use smaller solvent volumes during the several steps of theSPE procedure [19,22]. The use of smaller desorption volumes in combinationwith LVI/GC implies that no evaporation and reconstitution of the extracts isrequired, which eliminates critical and time-consuming steps in the extractionprocedure. Further miniaturisation of SPE has led to the development ofsolid-phase extraction – pipette tips (SPEPTs). An interesting aspect of SPEPTsis that bi-directional flow and cycling, i.e., aspirating and dispensing, can beapplied [22-24]. A disadvantage of micro-SPE may be that smaller sample


99

volumes have to be used, which leads to higher concentration detection limits(expressed in plasma concentrations). However, the injection of relatively largevolumes into a GC system can considerably increase the sensitivity. A firstattempt to perform on-line GC-analysis with SPEPTs was carried out using anexterior coupling device [22]. Extractions were performed off-line, andconsecutive desorption and GC analysis were performed on-line. The systemwas not applied to the analysis of biological samples.

The purpose of the present work was to investigate the further integrationof SPEPT and the PTV injector. The SPEPT was inserted into the injector ontop of the liner of the PTV to perform on-line desorption, i.e. an internalcoupling device. Special attention was paid to the impurity levels introduced bythe coupling of SPEPT and the liner. Also, several aspects regarding theextraction properties of the stationary phase of the SPEPT were investigated.The system was applied to the analysis of lidocaine and diazepam in plasma. Acomparison between the exterior coupling device [22] and the internal couplingdevice will be made. Both coupling devices, and in particular the internalcoupling device, can be considered as an intermediate step to the developmentof new, miniaturised, and automated SPE-GC systems. Therefore, a completeoptimisation and validation of the SPE procedure was not performed.

3.3.2 Experimental

3.3.2.1 Equipment and chromatographic conditions

SPEC•PLUS•PT pipette tips were obtained from ANSYS Diagnostics(Lake Forest, CA, USA). The extraction disk (4 mg) consisted of C18-ARstationary phase. The pipette tip had an inner diameter of 4.0 mm and an outerdiameter of 5.0 mm.

The PTV injection system was an OPTIC 2 (ATAS International,Veldhoven, The Netherlands), equipped with a shortened liner (64 mm×3.4 mmi.d.×5.0 mm o.d.). The liner was packed with ATAS “A” packing (a modifiedChromosorb-based material). The behaviour of this packing was previouslyinvestigated for the analysis of drugs in plasma, as well as for the settings of thePTV [7]. The injector was set at 50°C and evaporation of the solvent occurredusing the “AUTO vent mode” with a vent flow of 150 ml/min. After theevaporation of the solvent the valve was switched to the splitless mode and after10 s the temperature was increased at 10°C/s to 250°C. This final temperaturewas maintained during the analysis. The splitless mode was applied for1.75 min and, subsequently, the valve was switched to the split mode. The usedsplit flow was about 57 ml/min, whereas the septum purge flow and pressurewere 1.2 ml/min and 6.0 p.s.i. (1 p.s.i.=6894.76 Pa), respectively. A transfer


100

pressure of 12.0 p.s.i. was applied for 1.90 min. During the analysis the initialand final pressure were maintained at 8.0 p.s.i.

Gas chromatographic analyses were performed with a Hewlett-PackardHP 5890 series II instrument with a flame-ionisation detection (FID) or aGC-mass selective detection (MSD) system (HP 5972 series). The capillarycolumn was a HP-5 30 m×0.32 mm with 0.25 µm film thickness for analysiswith FID, whereas analyses with MSD were performed using a HP-5 MS30 m×0.25 mm with 0.25 µm film thickness. Helium was used as carrier gas.The column flow-rates were 1.1 and 0.5 ml/min for analysis with FID andMSD, respectively. The following temperature program was used for the GCsystem. The starting temperature was 40°C and after 3 min the temperature wasraised with 20°C/min at 215°C, followed by an increase of 5°C/min to 230°Cand a final rate of 25°C/min to 290°C. This final temperature was maintainedfor 3 min. The temperatures of the FID and mass selective detection systemswere set at 300°C and 280°C, respectively.

During analysis performed with GC-MSD in the total ion current (TIC)mode an m/z range of 50-350 was monitored. Using the selected ion monitoring(SIM) mode, an m/z value of 86, being the most intense fragment of lidocaine,was monitored from the start of the run to 16 min. From 16 min to the end ofthe run the m/z values 256 and 283 were monitored, corresponding to the mostintense fragment and the parent ion, respectively, of diazepam.

3.3.2.2 Chemicals

Methanol (Lab Scan, Dublin, Ireland) was of HPLC quality. K2HPO4 wasof analytical-reagent grade quality (Merck, Darmstadt, Germany). Ethyl acetateUltra resi-analysed (for organic residue analysis) was purchased fromMallinckrodt Baker (Deventer, The Netherlands). Water used during SPE wasultra pure (Elgastat maxima, Salm en Kipp, Breukelen, The Netherlands).Lidocaine (Eur. Ph., Holland Pharmaceutical Supply, Alphen A/D Rijn, TheNetherlands) and diazepam (Centrafarm, Etten-Leur, The Netherlands) wereused as test compounds and dissolved in ethyl acetate (for organic residueanalysis, Mallinckrodt Baker) or in phosphate buffer pH 8.0. Stock solutions of1 mg/ml were stored in the dark at 4°C.

3.3.2.3 SPE procedure and coupling to PTV

The SPE procedure was carried out by connecting a 10-ml gas tightplastic syringe (Omnifix syringe, Melsungen, Germany) to the upper end of thepipette tip. Liquid transport through the disk was done by applyingbi-directional flow with the exception of the desorption solvent. Air was only


101

applied in the downward direction. Drawing air into the pipette tip through thedisk implies the possibility of dislodging the disk.

The SPEPTs were pre-heated in an oven at 145°C for 2 h. The SPE diskin the pipette tip was pre-washed with five times 300 µl ethyl acetate. The diskwas then activated with ca. 200 µl methanol followed by conditioning with twotimes ca. 100 µl of 0.1 M K2HPO4 buffer (pH 8.0). Subsequently, 200 µl spikedphosphate buffer or 200 µl spiked plasma diluted with 200 µl blank phosphatebuffer were extracted on the disk. The sample was drawn into and pushed out ofthe tip twice. Then the disk was washed twice with ca. 100 µl water and,subsequently, dried by pushing air through the disk (10×10 ml).

Fig. 1: Internal coupling device: integration of SPEPT and the liner of the PTV.Parts of a conventional PTV injector: (1) septum, (2) carrier gas, (3) split flow,(4) septum purge, (5) cooling pipe, (6) power supply, (7) thermocouple,(8) heating unit, (9) capillary column, (10) oven wall.


102

The SPEPT was cut at the tip (4 mm) and the barrel (28 mm). Two O-ringswere placed around the remaining part of the pipette tip, one around the upperpart and a smaller one around the remainder of the tip. The silicone O-rings haddimensions of 6.0 mm and 1.3 mm, respectively. After opening the injector, theSPEPT system was inserted into the liner house on top of the shortened(16 mm) liner, and subsequently the injector was closed. For the desorption ofthe analytes from the disk 250 µl ethyl acetate was used. The ethyl acetate wasinjected through the septum of the GC injection port with a conventional GCsyringe with a shortened needle, so the ethyl acetate was injected into thepipette tip. The desorption occurs inside the injector. Due to the pressure andgas flow in the injector the ethyl acetate is transferred through the disk to theliner packing. The rest of the analysis is performed like conventional LVI/GCwith a PTV injector. The total system of the shortened liner and the cut pipettetip with the O-rings will be further mentioned as internal coupling device. Theset-up of the internal coupling device is depicted in Fig. 1. For each extraction anew SPEPT was used; the O-rings were used multiple times.

After the pipette tip was inserted into the injector, ethyl acetate as eluentcould be injected on top of the disk. To ensure that the injection needle did notperforate the disk, the needle of a conventional GC syringe was shortened to23 mm, positioning the tip of the needle about 1 mm above the disk of thepipette tip. If standard solutions were analysed, the fluids should not be injectedon top of a SPE disk. Therefore, a cut pipette tip without the disk assembly wasinserted into the injector, which mimicked the system with a disk inside thepipette tip.


3.3.3.1 Development of internal coupling device

To be able to perform on-line SPEPT-GC, solid-phase extraction− pipettetips should be directly attached to the injection system of the PTV. In a previouspaper [22], we described an exterior coupling device in which the SPEPT unitremained outside the injector. Though this device worked well, it had somedisadvantages, such as a high back-pressure during desorption and a loss ofabout 5% of analyte due to bi-directional flow during desorption. Furthermore,the system was not very robust.

Therefore, a further integration of SPEPT and GC was developed byinserting the pipette tip into the liner house (see Fig. 1). Since the outerdiameter of the pipette tip and the liner were similar, no changes had to be madeto the injector. Yet, to be able to accommodate the pipette tip into the linerhouse and install it on top of the liner, the 80-mm liner had to be shortened. If


103

the injector was set at 250°C, a temperature distribution was observed as shownin Fig. 2. This implicates that the liner packing is at 250°C, and that there is atemperature drop to both ends of the liner. This allows insertion of the SPEPTinto the injection system. The plastic holder of the SPEPT was found to melt atabout 160°C. Using 150°C as maximum temperature (TMAX) to which theplastic of the SPEPT should be exposed, it can be concluded from Fig. 2 that theSPEPT can be inserted for maximally 18 mm. Shortening the liner bymaximally 16 mm and cutting off 4-5 mm from the tip of the SPEPT ensuresthat TMAX will never be exceeded. Cutting off about 28 mm (± 1 mm) from thebarrel of the SPEPT allows accommodation of the pipette tip into the linerhouse, thus replacing the removed part of the liner. The cutting should ensurethat maximum lengths are not exceeded. The shortened liner and the cut PTshould have a combined length of 80-85 mm. Slightly shorter or not very evencutting of the PT does not affect the performance of the system.

Fig. 2: Temperature distribution (A) over the range of the entire liner of PTV. Themaximum temperature of SPEPTs defines the allowable position of the tip ofSPEPTs (B, dashed line) inside the injector and the shortened liner (C, dottedline).

An O-ring was placed around the barrel of the PT. This ring wasnecessary to prevent gas to go from the carrier gas line to the split flow linewithout going through the disk of the pipette tip and the liner packing. A secondO-ring was placed around the tip of the PT. This ring was essential to preventleakage of elution solvent and carrier gas. Without the latter O-ring carrier gas


104

could flow from the carrier gas line through the disk to the split flow line. TheO-rings and the combination of SPEPT and liner, which is slightly longer than aconventional unshortened liner of 80 mm, ensured a gas-tight connectionbetween PT and liner. No fluids came into contact with the O-ring. The totaltime required for the cutting and installation of the PT is less than two min,whereas the extraction time is less than 4 min.

3.3.3.2 Effect of pre-heating and pre-washing

Upon heating the liner packing, the temperature at the position of thepipette tip also increased. This heating of the internal coupling device resultedin a significant amount of compounds in the front of the chromatogram (up to16 min). Thus, (semi-) volatile compounds were released by this heatingprocess. The impurities probably originate from both the disk and thePT-housing. After 2 h pre-heating at 145°C in an oven, less than 1% of theinitial amount of impurities was still present. No visible changes of the diskwere observed. An important drawback of pre-heating is that the extractionproperties changed. Extracting lidocaine and diazepam from buffer afterpre-heating the pipette tip resulted in lower recoveries (from about 80% to 40%)with more variation.

Cutting of the pipette tip and insertion of the remaining part of the pipettetip into the injector and subsequent injection of ethyl acetate on top of apre-heated pipette tip resulted in a significant amount of medium- andless-volatile impurities which were observed in the chromatograms as clustersof peaks (Fig. 3A). These impurities interfered with the determination oflidocaine and diazepam and, therefore, had to be removed. Removal of theimpurities was done by applying a pre-wash step prior to the actual SPEprocedure. After pre-washing the disk with 1.5 ml ethyl acetate usingbi-directional flow, over 99% of the impurities were removed (Fig. 3B). Therecoveries of lidocaine and diazepam were not effected by this pre-wash step.This means that the properties of the stationary phase are not changed by thewashing.

Since the (semi-) volatile impurities did not interfere with thedetermination of lidocaine and diazepam, no pre-heating was applied prior touse of the SPEPTs for further experiments. Furthermore, since the pre-washingremoved interfering impurities and did not affect the extraction process, thisstep was applied during further experiments. Each SPEPT was used for onlyone extraction, since the heating of the injector during analysis could also affectthe properties of the stationary phase.


105

Fig. 3: Effect of pre-washing SPEPTs on the impurities in the chromatogram:(A) without pre-washing (offset signal: 2000); (B) 1.5 ml ethyl acetate.Note: for both chromatograms pre-heating is applied.

3.3.3.3 Application of internal coupling device

Practical aspectsInjection of ethyl acetate on the disk and consecutive evaporation resulted

in some tailing of the peak of the organic solvent. This was caused by theincomplete transfer of ethyl acetate from the disk to the liner packing. Uponheating the injector, the remaining ethyl acetate was evaporated from the diskand transferred to the GC column. The tailing of the solvent peak did notinterfere with the analysis of lidocaine and diazepam.

With slow injection of ethyl acetate (100 µl in 7-10 s), only a part of thedisk was moistened with ethyl acetate, which resulted in low recoveries.Injecting the organic solvent rapidly, i.e. injection of 100 µl in less than asecond, produced higher recoveries of lidocaine and diazepam and smallervariations in recovery. Upon rapid injection of ethyl acetate, a reservoir ofsolvent was formed on top of the disk and this ensured that the entire disk wasmoistened by the solvent. Desorption of lidocaine and diazepam with 250 µlethyl acetate resulted in still higher recoveries (increase from about 60 to 80%)than desorption with 100 µl ethyl acetate. However, the liquid capacity (Vmax)of the ATAS “A” liner is only 150 µl [7]. If Vmax is exceeded solvent will enterthe column. Therefore, the injection of 250 µl ethyl acetate was performed in athree-step injection. First, 100 µl ethyl acetate was rapidly injected on top of thedisk. After 2.25 min, a second portion of 75 µl ethyl acetate was injectedrapidly, and after 4.25 min the remaining 75 µl ethyl acetate was injected


106

rapidly. The total evaporation time was about 6.25 min. This injectionprocedure ensured that the next amount of ethyl acetate was never injected afterthe previous amount was completely evaporated. Too late injection of the nextamount of solvent implies injection while the actual GC analysis has alreadybeen started. The three-step injection procedure also ensured that Vmax of theliner was not exceeded at any time.

Injection of 100 µl ethyl acetate at a pressure of 2.5 p.s.i. instead of6.0 p.s.i. resulted in a small loss in recovery. The evaporation time was alsolonger (about 0.3 min) than the three-step injection of 250 µl ethyl acetate at6.0 p.s.i. Therefore, in the present work desorption was performed using 250 µlethyl acetate while the pressure was maintained at 6.0 p.s.i.

Analytical dataIn previous experiments the use of MSD proved to be necessary to

determine analytes in low concentrations after extraction from buffer andespecially from biological samples if LVI/GC was applied [7,22]. With thepresent system, i.e. the internal coupling device, extraction of lidocaine anddiazepam from buffer already showed the necessity of MSD, since aninterfering peak was observed for lidocaine.

The injection of 100 µl of standard solutions using the internal couplingdevice without SPE disk resulted in good linearity (R>0.998, range fromdetection limit to 250 ng/ml), demonstrating the reliability of the device. Thedetection limit (LOD), which was determined as S/N 3 or three times the blankpeak, was 10 ng/ml for both lidocaine and diazepam using the TIC mode. TheLOD was decreased to 0.5 ng/ml for both compounds if the SIM mode wasused. For the determination of lidocaine and diazepam the m/z values 86 and283 were used, respectively. Many silica-based compounds have a fragmentwith m/z 86. The m/z value of 283 corresponds with compounds fromsilicone-based materials [25]. These compounds probably originate from thesepta used on top of the vials in which the samples were stored and/or from thesilicone O-ring of the internal coupling that was used for the connection of thepipette tip and the liner.

The analysis of 200-µl plasma samples was performed using both the TICand SIM mode. Correlation coefficients (R) and LODs are presented in Table 1.Both scan modes showed good linearity over the entire concentration range.Also for plasma recoveries of about 80% were observed. Use of the SIM moderesulted in a lower LOD as compared with the TIC mode. A similar differencein LOD between TIC and SIM was observed for plasma extracts as comparedwith standard injections. If the SIM mode was used the background peaks werereduced, but not completely eliminated. This is partly due to the fact that nopre-heating was applied and due to the low m/z value of lidocaine. Moreover, as


107

with standard injections m/z 86 and 283 were present due to the silica-basedcompounds and the silicone-based compounds, respectively.

Table 1: Detection limits (LODs) and correlation coefficients (R) for lidocaine anddiazepam in 200 µl plasma with GC-MS analysis using the internal couplingdevice for the coupling with SPE (desorption with 250 µl ethyl acetate).

TIC SIMLOD

(ng/ml)R* LOD

(ng/ml)R*

Lidocaine 15 0.9990(n=4)

0.75 0.9999(n=7)

Diazepam 60 0.9998(n=3)

2.5 0.9995(n=6)

*: Ranges from LOD to 250.0 ng/ml.

In Fig. 4, representative chromatograms are depicted for the TIC mode(Figs. 4A and B) and the SIM mode (Figs. 4C and D) of blank and spikedplasma extracts, respectively. In Fig. 4D, the peaks of silicone-basedcompounds are lower than in Fig. 4C. This can be explained by the fact that thesilicone O-ring between SPEPT and liner was used before. Furthermore, whileclosing the injector the SPEPT and liner are pressed together. The tighter theseparts are pressed together, the greater the risk of tearing the ring, thus the O-ringcan release more compounds.

3.3.4 Conclusions and perspectives

The direct coupling of SPEPTs to the liner of the PTV injector couldeasily be established. The SPEPTs, and thus the extracts, can be purified bysimple pre-heating and pre-washing. However, the extraction properties of thestationary phase of the SPEPTs can be altered during the pre-heating, which isonly needed for the analysis of relatively volatile compounds. Other compoundspresent in the extracts originate from the biological matrix and solventimpurities. With LVI the use of selective detection, such as MSD, is essential todetermine drugs at low concentrations in biological samples.

The system is better applicable than the exterior coupling device [22],since the pipette tips and the liner are directly connected, which makes thesystem more robust. Furthermore, no carry-over is observed, and no leakage ofdesorption solvent can occur. The exterior coupling device has the advantage ofbi-directional flow during desorption, which allows in principle use of lessdesorption solvent. However, with this device 5% loss of analyte was observedbecause of incomplete desorption caused by the high back-pressure.


108

Fig. 4: GC-MS chromatograms of extracts of 200 µl plasma: (A) blank (TIC mode),(B) 62.5 ng/ml (TIC mode), (C) blank (SIM mode), (D) 5.5 ng/ml (SIMmode). L = lidocaine, D = diazepam, C = caffeine, * = silicone-basedcompounds.


109

The necessity of high sample-throughput requires automation. On-linesystems for SPE-GC already exist, but the main limitation of these systems isthe relatively long drying of the SPE cartridge, which typically takes 10-30 min.The use of SPEPTs enables decrease of the extraction time and especially thedrying step, which now only takes about 30 s. Automation of a system based onSPEPT and GC can be obtained more easily with the internal coupling devicethan with the exterior coupling device. A pipette robot based on a 96-well platedesign can simultaneously perform the extractions at-line in pre-cut SPEPTswith O-rings and a metal cap at the top. A recently developed liner exchanger[26] can be used to combine the SPEPT with the PTV liner, after which theinjector is closed automatically via a pneumatic system. Subsequently, thedesorption solvent can be injected on top of the disk. For automated injectionthe injection height should to be adjusted, or a shorter needle must be used inorder to prevent perforation of the disk. In such an approach, high-throughputanalysis is performed by means of a combination of miniaturised SPE and GCusing at-line extraction and on-line desorption. In the future, an even furtherintegration of SPE and GC might be obtained by performing the extraction inthe liner of the GC system.

Acknowledgements

SPEC•PLUS•PT pipette tips were kindly provided by ANSYSDiagnostics (Lake Forest, CA, USA). Jan Henk Marsman and Ronald Veenhuis(Department of Chemical Engineering, University of Groningen, Groningen,The Netherlands) are gratefully acknowledged for the use of their GC-MSDsystem and their assistance. This research was supported by the TechnologyFoundation STW, applied science division of NWO and the technologyprogramme of the Ministry of Economic Affairs.

3.3.5 References

[1] H.G.J. Mol, H.-G.M. Janssen, C.A. Cramers, J.J. Vreuls, U.A.Th. Brinkman.J. Chromatogr. A 703 (1995) 277.

[2] K. Grob, J.-M. Stoll. J. High Resolut. Chromatogr. Commun. 9 (1986) 518.[3] J.J. Vreuls, U.A.Th. Brinkman, G.J. de Jong, K. Grob, A. Artho. J. High Resolut.

Chromatogr. 14 (1991) 455.[4] J.J. Vreuls, G.J. de Jong, R.T. Ghijsen, U.A.Th. Brinkman. J. Microcol. Sep. 5

(1993) 317.[5] A.J.H. Louter, J. van Doornmalen, J.J. Vreuls, U.A.Th. Brinkman. J. High

Resolut. Chromatogr. 19 (1996) 679.


110

[6] R.J.C.A. Steen, I.L. Freriks, W.P. Cofino, U.A.Th. Brinkman. Anal. Chim. Acta353 (1997) 153.

[7] M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong. J. Chromatogr. B729 (1999) 199.

[8] M. Dressler. J. Chromatogr. 165 (1979) 167.[9] A. Lagana, B.M. Petronio, M. Rotatori. J. Chromatogr. 198 (1980) 143.[10] R.E. Majors, H.G. Barth, C.H. Lochmüller. Anal Chem. 56 (1984) 300R.[11] X.H. Chen, J.P. Franke, R.A. de Zeeuw. Forensic Sci. Rev. 4 (1992) 147.[12] Z.E. Penton. Adv. Chromatogr. 37 (1997) 205.[13] E.C. Goosens, D. de Jong, G.J. de Jong, U.A.Th. Brinkman. Chromatographia 47

(1998) 313.[14] A.J.H. Louter, E. Bosma, J.C.A. Schipperen, J.J. Vreuls, U.A.Th. Brinkman.

J. Chromatogr. B 689 (1997) 35.[15] A. Namera, M. Yashiki, Y. Iwasaki, M. Ohtani, T. Kojima. J. Chromatogr. B 716

(1998) 171.[16] P. Enoch, A. Putzler, D. Rinne, J. Schüter. J. Chromatogr. A 822 (1998) 75.[17] A.J.H. Louter, C.A. van Beekvelt, P.Cid Montanes, J. Slobodník, J.J. Vreuls,

U.A.Th. Brinkman. J. Chromatogr. A 725 (1996) 67.[18] R. Sasano, T. Hamada, M. Kurano, M. Furuno. J. Chromatogr. A 896 (2000) 41.[19] D.A. Wells, G.L. Lensmeyer, D.A. Wiebe. J. Chromatogr. Sci. 33 (1995) 386.[20] E.H.R. van der Wal, E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman.

Chromatographia 47 (1994) 239.[21] K. Hartonen, M.L. Riekkola. J. Chromatogr. B 676 (1996) 45.[22] M.W.J. van Hout, R.A. de Zeeuw, G.J. de Jong. J. Chromatogr. A 858 (1999)

117.[23] R.E. Majors. LC•GC Int. May (1998) 8.[24] D.D. Blevins. D.O. Hall. SPEC News 3-1 (1998) 1.[25] M. Spiteller, G. Spiteller. Massenspektrensammlung von Lösungsmitteln,

Verunreinigungen, Säulenbelegmaterialien und Einfachen AliphatischenVerbindungen, Springer-Verlag, Vienna, 1973, S70.

[26] ATAS International, Veldhoven, The Netherlands, http://www.ATAS-INT.com

SPETD-GC-MSD for determination of drugs in urine

111

3.4

Solid-phase extraction – thermal desorption –gas chromatography with mass selective detectionfor the determination of drugs in urine*

Summary

Solid-phase extraction (SPE) was combined with thermal desorption (TD) andgas chromatographic (GC) analysis to determine drugs in urine. The extractionwas performed inside a fritted GC liner using about 5 mg Tenax that wasinserted into the liner on top of the frit. After the extraction the liner was placedinto the injector of the GC and the analytes were thermally desorbed by usingprogrammed temperature vaporiser. Several stationary phases were investigatedfor the applicability of SPETD-GC analysis. Tenax proved to be the mostsuitable extraction phase, since hardly any interferences were observed andacceptable absolute recoveries (73 and 74%) were obtained for lidocaine anddiazepam. A mass selective detector (MSD) in the selected ion monitoringmode allowed the detection of lidocaine and diazepam down to 0.5 ng/ml using50 µl urine. The use of 5 mg stationary phase allowed a rapid extractionprocedure, while a 10-m GC column provided a fast chromatographic system.As a result, the total analysis time was less than 20 min, including 5 min dryingof Tenax and 5 min thermal desorption. Thus, SPETD-GC-MSD appears to be apowerful tool for the rapid analysis of biological samples.

*: M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong. Accepted for publication in Chromatographia.


112

3.4.1 Introduction

In bioanalysis, the numbers of samples are increasing, thus requiringrapid analytical systems. Not only the separation must be performed rapidly, butalso the extraction may not be time-limiting. Consequently, the extraction iscoupled more and more on-line with the separation step. For example,solid-phase extraction (SPE) is now coupled on-line with liquidchromatography (LC) on a routine basis [1-11], and on-line SPE-gaschromatography (GC) is also gaining interest [12-20]. Besides the approach ofon-line extraction and separation, another possibility is miniaturisation of theextraction [21,22]. Some time-limiting steps can then be shortened or evenomitted. On-line SPE-GC implies the desorption of the analytes with relativelylarge volumes of eluent, and thus the injection of large volumes of solvent intothe GC. This can be achieved by using a retention gap or a programmedtemperature vaporiser (PTV). The elution and injection of large solvent volumescan be critical with regard to flow-rate during elution and evaporation of thesolvent [22,23].

Another option of the PTV is thermal desorption, since the PTV allowsrapid heating of the injector. When applying solid-phase extraction – thermaldesorption (SPETD), the extraction procedure is basically the same as foron-line SPE up to and including the drying step. Subsequently, the analytes arethermally desorbed instead of using liquid desorption. Thus, no injection andevaporation of the eluate in the injector is required. In order to perform SPETD[17,18,20,24-31], the stationary phase of the extraction unit is critical. Thephase must have good extraction properties and must also be thermostable inorder to prevent interference of the phase components during analysis.Polydimethylsiloxane (PDMS) [29,30] and Tenax (a porous polymer basedupon 2,6-diphenyl-p-phenylene oxide) [24,27,28,31] have been claimed topossess these characteristics. Another important aspect is the drying of thephase prior to thermal desorption. In previous studies the drying step was oftenvery time-consuming (up to 30 min) and/or high gas flow-rates (>100 ml/min)were required [24,26,27,29,30,32]. Minimising the amount of extraction phasecan reduce the length of the drying step.

The goal of the present study was to explore the possibilities of SPETDfor the analysis of biological samples, setting up a system with good sensitivity,selectivity and speed. The amount of the extraction phase was kept to aminimum. In order to obtain an integrated design, the extraction was carried outoff-line in the liner of the GC, after which the liner was put into the GC andon-line thermal desorption took place.


113

3.4.2 Experimental

3.4.2.1 Chemicals

Methanol (Lab Scan, Dublin, Ireland) was of HPLC quality. KH2PO4

(Merck, Darmstadt, Germany) was of analytical-reagent grade quality. Ethylacetate Ultra resi analysed (for organic residue analysis) was purchased fromMallinckrodt Baker (Deventer, The Netherlands). Water used during SPE wasultra pure (ElgaStat Maxima, Salm & Kipp, Breukelen, The Netherlands).Lidocaine (Eur. Ph., Holland Pharmaceuticals Supply, Alphen a/d Rijn, TheNetherlands) and diazepam (Centrafarm, Etten-Leur, The Netherlands) wereused as test compounds and dissolved in ethyl acetate or phosphate buffer(0.1 M, pH 8.0). Stock solutions of 1 mg/ml were stored in the dark at 4ºC.

Stationary phases that were investigated for thermal desorption wereResin GP, PLRP-S and Styrene-divinylbenzene (all from Spark Holland,Emmen, The Netherlands), Tenax, Hayesep Q, ATAS Focus Trap and PDMS(all from Varian-Chrompack, Middelburg, The Netherlands), Bond Elut LMSand Abselut Nexus from Varian Sample Preparation Products (Harbor City, CA,USA), and C18 SPEC disks (ANSYS Diagnostics, Lake Forest, CA, USA).

3.4.2.2 Equipment and chromatographic conditions

The PTV injection system was an OPTIC 2 (ATAS International,Veldhoven, The Netherlands), with a fritted liner (80 mm×3.4 mm i.d.×5.0 mmo.d., frit 15 mm from the bottom, ATAS International) to hold the stationaryphase. The extraction was also performed inside these liners. A Visiprep system(Supelco, Bellefonte, PA, USA) was used to apply vacuum under the linersduring the extraction. After extraction the liner was inserted into the injector,which was set at 50°C. Then the temperature of the injector was increased at15°C/s to 350°C. This final temperature was maintained during the analysis.The splitless mode was applied for 5.10 min and, subsequently, the valve wasswitched to the split mode. The split flow was about 57 ml/min, whereas theseptum purge flow and pressure were 1.1 ml/min and 1.5 p.s.i. (1 p.s.i. =6894.76 Pa), respectively. A transfer pressure of 1.5 p.s.i. was applied for5.25 min. During the analysis the initial and final pressure were maintained at1.0 p.s.i.

Gas chromatographic analyses were performed with a Hewlett-PackardHP 5890 series II instrument with flame-ionisation detection (FID) or aHP 6890 series GC- mass selective detection (MSD) system (HP 5972 series).The capillary column was a HP-5 10 m×0.32 mm with 0.25 µm film thickness.Helium was used as carrier gas. The column flow was 2.2 ml/min. The startingtemperature was 50°C and after 5.25 min the temperature was raised with 30 or


114

50°C/min for FID or MSD, respectively, to 325°C. This final temperature wasmaintained for 0.50 min. The temperature of the FID and MSD was set at300°C and 280°C, respectively.

During analysis performed with GC-MSD in the total ion current (TIC)mode an m/z range of 50-350 was monitored. Using the selected ion monitoring(SIM) mode, an m/z value of 86, the most intense fragment of lidocaine, wasmonitored from the start of the run to 8.50 min. From 8.50 min to the end of therun the m/z values 283 and 284 were monitored, corresponding to the parent ionand its protonated form, respectively, of diazepam.

3.4.2.3 SPETD procedure

About 5.0 mg stationary phase was brought into the fritted liner. Thelatter was tapped carefully to pack the stationary phase more tightly on top ofthe frit, resulting in about a 2-mm height of the phase. The C18 SPEC disks(4.0 mg) were inserted as such on top of the frit. The extraction unit, existing ofthe liner and the stationary phase, was then placed onto the Visiprep systemafter which 250 µl ethyl acetate was flushed through the unit as pre-wash step.Subsequently, 100 µl methanol was pipetted into the liner to remove theremaining ethyl acetate prior to conditioning the stationary phase with 100 µlphosphate buffer (0.1 M, pH 8.0). Then 100 µl sample was applied. The latterexisted of 50 µl untreated calf urine (blank or spiked), diluted with 50 µlphosphate buffer, unless stated otherwise. The sample was introduced on top ofthe stationary phase inside the liner and then passed through the phase by meansof gravity, which took about 15-20 s. This was followed by washing of theextraction unit with 150 µl water. Then vacuum (-15 mm Hg) was applied underthe extraction unit for 5 min at room temperature to remove the remainingwater. Finally, the extraction unit was inserted into the PTV injector, afterwhich thermal desorption took place at 350°C for 5 min. The liners werere-used but the stationary phase was replaced after each analysis.


3.4.3.1 Selection of the stationary phase

Several stationary phases were investigated. In order to determine theirsuitability for thermal desorption, about 5 mg was filled into the liner afterwhich the liner was inserted into the injector and subsequently, heating at 300°Ccombined with GC was performed. Various well-known LC phases wereselected because of their proven extraction properties. These phases could notwithstand high temperatures and, consequently, a high background was


115

observed. In a previous study [22] it was shown that the (volatile) impuritiescould be removed by pre-heating the SPEC disk. However, the extractionproperties also changed. Therefore, the LC phases were not further consideredfor SPETD purposes.

Subsequently, some typical GC phases were treated in the same way.ATAS Focus Trap proved to be clean upon heating. Yet, its extractionproperties were very poor, i.e., no retention of lidocaine and diazepam wasobserved upon extraction of buffer solutions. The impurity level of PDMS wasalso considerable. The suitability of this phase for thermal desorption has beenclaimed [29], though it was also noticed that disturbing siloxane breakdowncould also occur [17,18]. Pre-washing of about 5 mg PDMS with 750 µl ethylacetate and subsequent pre-heating at 300°C for 24 hours decreased theimpurity level, but severe interferences remained visible, even with massspectrometric detection in the SIM mode. Therefore, PDMS was notinvestigated further. Another GC stationary phase, Tenax, also showed asignificant amount of interferences. In several reports, a pretreatment of Tenaxwas applied prior to analysis to clean the phase. Vreuls et al. performedwashing of Tenax with acetone [28], whereas pre-heating Tenax for a certainperiod of time [27] or repeatedly pre-heating the phase in the extraction unit(2-5 times) prior to using it for the actual analysis [24] was also applied. In thepresent set-up, pre-heating Tenax (at 300°C for 15 min) in the liner eliminatedmost of the interferences, but pre-washing was also suitable. The latter waspreferred as this could be integrated into the extraction procedure more easily.About 5 mg Tenax in the liner was pre-washed with 250 µl ethyl acetate prior tothe extraction, after which only a few interferences were observed. As thisphase had shown acceptable extraction properties with typical recoveries of80-110% [24,26,27], this phase was selected for further exploration using theliner as the extraction tube.

3.4.3.2 Optimisation of SPETD procedure

Initially, 5 mg Tenax was inserted into the liner and, after pre-washing thephase with 250 µl ethyl acetate, 100 µl methanol and subsequently 150 µlphosphate buffer were used to condition the extraction phase. Then 400 µlsample (100 µl urine diluted with 300 µl buffer pH 8.0) was extracted, and theextraction unit was washed with 500 µl water. Vacuum (-15 mm Hg) at roomtemperature was applied to dry the phase during 0.5 min, which was thenfollowed by insertion of the extraction unit into the injector and 0.5 min thermaldesorption at 300°C and analysis with GC-FID.

Increasing the desorption temperature proved favourable for bothlidocaine and diazepam. For both compounds a gain in signal was observed upto 350°C, and no higher signals were observed with higher desorption


116

temperatures. Furthermore, the signals of both compounds also increased bylengthening the desorption time from 0.5 min to 5 min. No higher signals wereobserved with desorption times longer than 5 min. Therefore, the desorptiontemperature was set at 350°C with a desorption time of 5 min.

At pH 8.0 diazepam is in its neutral form, whereas lidocaine is partiallyprotonated. When using Tenax it might be favourable for the recovery to havethe compounds in their non-protonated form. Increasing the pH to 10 ensuresthat lidocaine is also non-protonated. However, no gain in recovery wasobserved.

Then, the focus was on the time required for the entire extraction/dryingprocess. It is essential to avoid water entering the GC, so, the drying should becomplete. In Fig. 1 the drying time was varied, and the amount of remainingwater in the stationary phase was calculated by weighing the extraction unit.After 1 or 2 min drying both the liner and the stationary phase were still visiblywet. After 3 min, the liner appeared to be dry, but as can be seen in Fig. 1, itstill retained a substantial amount of water. The drying was near complete atabout 5 min, thus shortening of the extraction time was not possible. Anyremaining water traces after 5 min drying did not interfere with the analysis. Itshould be noted that in many other studies [24,26,29,30,32] 10-30 min of dryingis recommended. In our study, the drying time can be kept relatively short dueto the limited amount of stationary phase in the extraction. Furthermore, in thepresent set-up, only vacuum is applied at room temperature. In other systemsheating of the SPE device and a high gas flow-rate during the drying step areneeded [24,26], which can be determinative in the final analysis.

Fig. 1: Remaining amounts of water in the stationary phase at different drying times;calculated by weighing of the total extraction unit.

02468

10121416

0 2 4 6 8 10

Drying time (min)

Rem

ain

ing

wat

er in

st

atio

nar

y p

has

e (m

g)


117

Using the optimised conditions as described above, the system wasapplied to the analysis of 400 µl spiked phosphate buffer prior to analysis ofurine. Both lidocaine and diazepam could be analysed with GC-FID withoutinterference of compounds originating from the stationary phases or solventsused during the extraction (Fig. 2A).

Fig. 2: (A) SPETD-GC-FID of blank buffer (lower line) and spiked buffer (200 ng/ml;upper line), (B) SPETD-GC-FID of blank urine/buffer (1:3; lower line) andspiked urine/buffer (1:3; 100 ng/ml lidocaine in urine; upper line),(C) SPETD-GC-FID of blank (lower line) and spiked urine/buffer (1:3;100 ng/ml diazepam in urine; upper line).


118

When extracting urine, it was necessary to dilute with buffer to ensure properflowing of the sample through the stationary phase. However, if 100 µl spikedurine was diluted with 300 µl buffer and the sample was extracted and thenanalysed with GC-FID, severe interference of the urine matrix was observed.Only enlarging the time scale revealed the peaks of lidocaine and diazepam(Figs. 2B and C). Using 100 µl urine, the absolute recovery was in the order of56% for lidocaine and 65% for diazepam at a level of 200 ng/ml, and a limit ofdetection (LOD) of about 50 ng/ml could be obtained for both compounds. Therecoveries for urine were similar to those observed upon the extraction frombuffer.

The low recovery may be caused by occupation of sorption sites bymatrix components or by chromatographic breakthrough. Therefore, 50 µl urinewas diluted with 50 µl buffer, resulting in a 100-µl sample. The recoveries nowincreased to 74 and 73% for lidocaine and diazepam, respectively, at aconcentration of 200 ng/ml. To obtain even higher recoveries, the amount ofstationary phase was increased to 10 mg. The recoveries were now about 90%for both compounds, but the peaks became very broad. This is probably due tolarger amounts of water being retained by the stationary phase after drying.Consequently, the broader peaks led to higher LODs and the water can damagethe chromatographic system. The latter could be overcome by lengthening thedrying time. However, in order to keep the extraction time as short as possible,the amount of stationary phase was kept at 5 mg, accepting the slightly lowerrecoveries.

3.4.3.4 Application of SPETD-GC-MS

To improve the applicability of SPETD-GC for biological samples, anMSD was used. Now, 50 µl spiked urine diluted with 50 µl buffer was extractedwithin 8 min, including 5 min drying, and using 5 mg Tenax. Thermaldesorption was performed at 350°C for 5 min, after which the GC temperaturewas increased rapidly. The analysis time could also be kept very short, since acapillary GC column of only 10 m was used. The total GC-MSD time, including5 min thermal desorption, took about 11 min, thus a total analysis time of about19 min was obtained. In the TIC mode, significant interference of the samplematrix was observed. By extracting the desired m/z values, the LODs(signal-to-noise ratio 3), 10 ng/ml, were lower than with FID (50 ng/ml). Evenbetter results were obtained using the SIM mode due to the increased sensitivityand selectivity (LOD about 0.5 ng/ml), but as can be seen in Fig. 3, theinterference was not completely eliminated. The difference between the blank(Fig. 3A) and spiked urine (Fig. 3B) between 6.0 and 8.5 min may be due to themanual packing of the stationary phase. The latter may result in a difference infirmness of the packing and thus in a change in the filtration properties of the


119

stationary phase. However, no adverse effect on the determination of lidocaineand diazepam was observed.

Fig. 3: SPETD-GC-MSD in the SIM mode: (A) Blank urine (50 µl; diluted withbuffer 1:1), (B) 0.5 ng/ml lidocaine and diazepam in urine (50 µl; diluted withbuffer 1:1).

The results of this SPETD-GC-MSD system are presented in Table 1.Even though only 50 µl urine was used, acceptable LODs (sub-ng/ml) wereobtained for both compounds with good linearity over a range of 0.5-200 ng/ml.A quantitation limit of about 1 ng/ml could be obtained. The absolute recoveriesas well as the reproducibilities (n=8) were also satisfactory. The use of asuitable internal standard may help to obtain even better reproducibilities.


120

Table 1: Analytical data of urine analysis using SPETD-GC-MSD.

Lidocaine DiazepamAbsolute recovery ± RSDa (%) 74 ± 6.5 (n=8) 73 ± 5.9 (n=8)LOD (ng/ml) – SIM mode 0.5 0.5R 0.9999 0.9971Linear range (ng/ml) 0.5-200 0.5-200

a: at 10 ng/ml.

3.4.4 Conclusions

The applicability of solid-phase extraction combined with thermaldesorption strongly depends on a useful, clean stationary phase with goodextraction properties and thermostability. Most stationary phases seem to lackone or more of these characteristics, as the phases are usually developed foreither LC or GC purposes. In this study, Tenax was a suitable compromise interms of acceptable recoveries and impurity levels, but the ideal stationaryphase for SPETD purposes seems at yet unavailable.

In this study, only 5 mg of stationary phase was required, herebydecreasing the time-consuming drying step in the extraction procedure. As theextraction was carried out in a GC liner, the thermal desorption was very easy toperform. Despite the small sample volume (50 µl urine + 50 µl buffer), goodsensitivity was obtained by using MS detection. If a higher sensitivity isimportant, more Tenax and probably also more urine can be used. Heating ofthe SPE unit during the drying might help to minimise the time required fordrying of the stationary phase. However, this will also make the extraction unitmore complex. Moreover, it is important to investigate if the stationary phasecan be used more than once for thermal desorption.

So far, the extraction was performed off-line with on-line thermaldesorption. By using a suitable robotic system, the extraction can be performedat-line and the liner can be installed and removed by the liner exchanger. In thisway automation of the entire extraction procedure can be obtained and thenSPETD seems a powerful approach for combination of SPE and GC for theanalysis of biological samples.

Acknowledgements

Wil van Egmond (ATAS International) is gratefully acknowledged for theuse of the MSD. The authors thank René de Nijs (Varian-Chrompack), BertOoms (Spark Holland), Dennis Blevins (ANSYS Diagnostics) and NigelSimpson (Varian Sample Preparation Products) for their donation of the


121

stationary phases. This research was supported by the Technology FoundationSTW, applied science division of NWO and the technology programme of theMinistry of Economic Affairs.

3.4.5 References

[1] O.V. Olesen, P. Plougmann, K. Linnet. J. Chromatogr. B 746 (2000) 233.[2] N.G. Knebel, S. Grieb, S. Leisenheimer, M. Locher. J. Chromatogr. B 748 (2000)

97.[3] M.C. Hennion. J. Chromatogr. A 856 (1999) 3.[4] J. Solà, J. Pruñonosa, H. Colom, C. Peraire, R. Obach. J. Liq. Chromatogr. Relat.

Technology 19 (1996) 89.[5] A. Pastoris, L. Cerutti, R. Sacco, L. De Vecchi, A. Sbaffi. J. Chromatogr. B 664

(1995) 287.[6] G. García-Encina, R. Farrán, S. Puig, M.T. Serafini, L. Martinez. J. Chromatogr.

B 670 (1995) 103.[7] D. Barrón, J. Barbosa, J.A. Pascual, J. Segura. J. Mass Spectrom. 31 (1996) 309.[8] A. Marchese, C. McHugh, J. Kehler, H. Bi. J. Mass Spectrom. 33 (1998) 1071.[9] M. Yritia, P. Parra, J.M. Fernández, J.M. Barbanoj. J. Chromatogr. A 846 (1999)

199.[10] F. Beaudry, J.C.Y. Le Blanc, M. Coutu, N.K. Brown. Rapid. Commun. Mass

Spectrom. 12 (1998) 1216.[11] J.A. Pascual, J. Sanagustín. J. Chromatogr. B 724 (1999) 295.[12] R. Herráez-Hernández, A.J.H. Louter, N.C. van de Merbel, U.A.Th. Brinkman.

J. Pharm. Biomed. Anal. 14 (1996) 1077.[13] A. Farjam, J.J. Vreuls, W.J.G.M. Cuppen, U.A.Th. Brinkman, G.J. de Jong.

Anal. Chem. 63 (1991) 2481.[14] A.J.H. Louter, E. Bosma, J.C.A. Schipperen, J.J. Vreuls, U.A.Th. Brinkman.

J. Chromatogr. B 689 (1997) 35.[15] K.K. Verma, A.J.H. Louter, A. Jain, E. Pocurull, J.J. Vreuls, U.A.Th. Brinkman.

Chromatographia 44 (1997) 372.[16] T. Hankemeier, U.A.Th. Brinkman. Chromatogr. Sci. 86 (2001) 155.[17] J.J. Vreuls, A.J.H. Louter, U.A.Th. Brinkman. J. Chromatogr. A 856 (1999) 279.[18] A.J.H. Louter, J.J. Vreuls, U.A.Th. Brinkman. J. Chromatogr. A 842 (1999) 391.[19] R. Sasano, T. Hamada, M. Kurano, M. Furuno. J. Chromatogr. A 896 (2000) 41.[20] E.C. Goosens, D. de Jong, G.J. de Jong, U.A.Th. Brinkman. Chromatographia 47

(1998) 313.[21] M.W.J. van Hout, R.A. de Zeeuw, G.J. de Jong. J. Chromatogr. A 858 (1999)

117.[22] M.W.J. van Hout, W.M.A. van Egmond, J.P. Franke, R.A. de Zeeuw, G.J. de

Jong. J. Chromatogr. B. 766 (2001) 37.[23] K. Grob, M. Biedermann. J. Chromatogr. A 750 (1996) 11.[24] J.J. Vreuls, R.T. Ghijsen, G.J. de Jong, U.A.Th. Brinkman. J. Microcol. Sep. 5

(1993) 317.


122

[25] M.P.M. van Lieshout, H.-G. Janssen, C.A. Cramers, G.A. van den Bos.J. Chromatogr. A 764 (1997) 73.

[26] A.J.H. Louter, J. van Doornmalen, J.J. Vreuls, U.A.Th. Brinkman. J. HighResolut. Chromatogr. 19 (1996) 679.

[27] H.G.J. Mol, H.-G. Janssen, C.A. Cramers, U.A.Th. Brinkman. J. High Resolut.Chromatogr. 16 (1993) 459.

[28] J.J. Vreuls, U.A.Th. Brinkman, G.J. de Jong, K. Grob, A. Artho. J. High Resolut.Chromatogr. 14 (1991) 455.

[29] E. Baltussen, F. David, P. Sandra, H.-G. Janssen, C.A. Cramers. J. Microcol.Sep. 11 (1999) 471.

[30] E. Baltussen, F. David, P. Sandra, H.-G. Janssen, C. Cramers. J. High Resolut.Chromatogr. 21 (1998) 645.

[31] S. Müller, J. Efer, W. Engewald. Chromatographia 38 (1994) 694.[32] E. Baltussen, F. David, P. Sandra, H.-G. Janssen, C.A. Cramers.

J. Chromatogr. A 805 (1998) 237.

44COUPLING OF

SOLID-PHASE EXTRACTIONAND MASS SPECTROMETRY

The pieces have been

SPE-MS3 for the determination of clenbuterol in urine

125

4.1

On-line coupling of solid-phase extraction withmass spectrometry for the analysis of biologicalsamples. Determination of clenbuterol in urine usingmultiple-stage mass spectrometry in an ion-trapmass spectrometer*

Summary

Solid-phase extraction (SPE) was coupled to ion-trap mass spectrometry todetermine clenbuterol in urine. For SPE a cartridge exchanger was used, andafter extraction, the eluate was directly introduced into the mass spectrometer(MS). For two types of cartridges, i.e., C18 and polydivinylbenzene (PDVB), thetotal SPE procedure (including injection of 1 ml urine, washing, and desorption)has been optimised. The total analysis, including SPE, elution, and detection,took 8.5 min with PDVB cartridges, while an analysis time of 11.5 min wasobtained with C18 cartridges. A considerable amount of matrix was present afterextraction of urine over C18 cartridges, resulting in significant ion suppression.With PDVB cartridges, the matrix was less prominent, and less ion suppressionwas observed. For single MS, a detection limit (LOD) of about 25 ng/ml wasfound with PDVB cartridges. With C18 cartridges an LOD of only about50 ng/ml could be obtained. Applying tandem mass spectrometry (MS/MS) didnot lead to an improved LOD, because of an interfering compound. However, aconsiderable improvement in the LOD could be obtained with MS3. Theselectivity and sensitivity were increased by the combination of efficientfragmentation of clenbuterol and reduction of the noise. Detection limits of2 and 0.5 ng/ml were obtained with C18 and PDVB cartridges, respectively. Theion suppression was 4 to 45% (concentration range: 250 to 1.0 ng/ml) afterextraction of urine using PDVB cartridges, and up to 70% ion suppression wasobserved using C18 cartridges. With MS4, no further improvement in selectivityand sensitivity was achieved, due to inefficient fragmentation of clenbuterol andno further reduction of the noise.

*: M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, G.J. de Jong. Rapid Commun. Mass Spectrom. 14(2000) 2103-2111.

Chapter 4 – Coupling of solid-phase extraction and mass spectrometry

126

4.1.1 Introduction

The necessity of high-throughput analysis of biological samples isincreasing, since, due to stricter policies for the registration and use of drugs,more samples have to be analysed in less time. Moreover, the development ofmore potent drugs results in lower concentrations of these drugs in biologicalsamples. Thus, fast and sensitive analytical techniques are required. A solutionfor these seemingly incompatible requirements is presented by systems basedon on-line liquid chromatography-mass spectrometry (LC-MS) withatmospheric pressure ionisation interfaces. Using the specificity of the MS,analytical systems have been developed in which only short LC columns (25 to50 mm) are required for the analysis of biological samples [1-3], thus leading toshort analysis times. However, the time-limiting step is usually the samplepreparation, especially if carried out manually. A powerful technique forclean-up and preconcentration of biological samples is solid-phase extraction(SPE) [4,5]. With increasing numbers of samples, automation of SPE is needed.Consequently, there is a growing interest for on-line and at-line (including96-well SPE [6,7]) coupling of SPE to LC [8-11] and, more recently, to gaschromatography (GC) [12-18]. On-line SPE is a very attractive samplepretreatment technique since the entire process of activation, conditioning,sampling, washing, and elution takes place in an enclosed circuit, whicheliminates error-prone steps like evaporation and reconstitution. Therefore, ingeneral, better precision and sensitivity are observed when compared withoff-line SPE.

A very rapid system for the analysis of complex samples is obtained bydirect coupling of automated SPE (“short-column LC”) to MS [19-26]. A truechromatographic separation step is not performed, which enhances speed andreduces costs. However, as a result, the SPE procedure must be performed veryefficiently to ensure that clean extracts are obtained, which can be difficult ifcomplex biological samples have to be analysed. The presence of matrixcompounds in the extract can cause ion suppression [27-30], resulting in loss ofreliability and accuracy of the analytical data obtained with an SPE-MS system.Multiple MS (MSn with n≥2) may give higher selectivity and thus highersensitivity, but the effect of ion suppression will persist.

Inspection for the abuse of illegal compounds, like most growthpromoters, requires high-throughput analysis of biological samples, anddetection of those compounds at low concentrations. The combination of SPEand MS, as described above, shows high potential for such applications. Apopular growth promoter is clenbuterol, which is administered to cattle toincrease the deposition of protein while reducing fat accretion. The effect isgrowth of muscle tissue and a weight gain [31]. Consumption of meat thatcontains clenbuterol can cause acute poisoning. Due to the severe effects on


127

human health [32-33], the use of clenbuterol as a growth promoter is prohibitedin most countries. Previous efforts to develop a fast and cost-effective screeningmethod for clenbuterol in urine resulted in a method with a three-step samplepretreatment, including use of immobilised antibodies, followed bychromatographic separation and electrochemical detection [34]. The limit ofdetection (LOD) was about 4 ng/ml.

Another system coupled SPE directly to a triple-quadrupole massspectrometer using C18 and mixed mode cartridges for sample pretreatment[19]. About 50% ion suppression was observed, as well as the formation ofadducts of clenbuterol and creatinine if mixed mode cartridges and electrosprayionisation (ESI) were used. Experiments with C18 cartridges and atmosphericpressure chemical ionisation (APCI) showed no interference from creatinine.Selected reaction monitoring (SRM) resulted in an increase of the signal-to-noise ratio (S/N) by a factor of 5, compared to MS/MS full-scan detection withextraction of the m/z 203 signal. The LOD was about 20 and 2 ng/ml for the ESIsource and APCI source, respectively. The total analysis time was still long,about 13.5 min, including sampling (4 min) and sample pretreatment.

In this paper we present the continuation of our research [19] concerningthe potential of SPE-MS as a screening method for clenbuterol in urine. Insteadof a triple-quadrupole mass spectrometer, a relatively cheap ion-trap instrumentwas used. A gain in sensitivity had to be achieved by adjusting the MS mode,i.e., single MS vs. MSn (n≥2). The clean-up is more critical with an ion-trapmass spectrometer than with a triple-quadrupole one. Besides ion suppression[27-30] a possible loss of signal can be caused by the limited capacity for ionsof the ion-trap. Only an APCI source was used, since the best results had beenobtained with this source in previous experiments [19]. The purpose was toproduce cleaner extracts by use of another stationary phase in the SPEcartridges. Furthermore, ways to reduce the analysis time were investigated.

4.1.2 Experimental

4.1.2.1 Chemicals and instrumentation

All on-line experiments were performed with a Prospekt (Spark, Emmen,The Netherlands) using one six-port valve, the cartridge-switching device, and asolvent delivery unit (SDU). Activation, conditioning, sampling, trapping, andwashing were done using the SDU. The effluent was connected to waste duringthese steps. All steps of the SPE procedure were carried out using aforward-flush mode, by which some chromatographic separation was obtained.It was possible to replace a cartridge with a new cartridge in the cartridge holderafter single or multiple use. The elution was performed using a Hewlett-Packard


128

gradient pump Series 1100 (Hewlett-Packard, Waldbronn, Germany), whichwas connected to an LCQ ion-trap mass spectrometer (Thermoquest, San Jose,CA, USA). The mass spectrometer was equipped with an APCI source.Experiments with UV detection were carried out with a Hewlett-Packard diodearray detector, Series 1100. The Prospekt cartridges used in this work wereHySphere–9 (C18, 10×2 mm, particle size 7 µm) and HySphere Resin GP(10×2 mm, particle size 10 µm), a spherical polymeric phase ofpolydivinylbenzene (PDVB).

Methanol was of HPLC grade (Lab Scan, Dublin, Ireland). Theammonium acetate was of pro-analysis quality (Merck, Darmstadt, Germany).The 5 mM ammonium acetate buffer was adjusted to pH 8 using 2.5% ammonia(pro-analysis quality). Pure water was obtained from an Elgastat maximasystem (ultra pure water, Salm and Kipp, Breukelen, The Netherlands).Aqueous solutions were filtered over a 0.45-µm RC 55 membrane filter(Schleicher & Schuell, Dassel, Germany) prior to use. Clenbuterol (SigmaAldrich, Dorset, United Kingdom) was dissolved in methanol (1 mg/ml) andstored in the dark at -20ºC. Samples were prepared by diluting the clenbuterolstock solution with 5 mM ammonium acetate buffer (pH 8), or with humanurine.

4.1.2.2 SPE procedure

The SPE procedure for C18 cartridges was as follows: activation wasperformed with 2.5 ml methanol, and conditioning with 2.5 ml 5 mMammonium acetate buffer (pH 8) at a flow-rate of 2.5 ml/min. The 1-ml samplewas loaded onto the cartridge at 1.0 ml/min, followed by a washing step with4.0 ml ammonium acetate buffer (5 mM, pH 8) at a flow-rate of 2.0 ml/min.Elution was performed with a gradient of ammonium acetate buffer (5 mM,pH 8) and methanol. In 1.0 min the methanol percentage was increased from0 to 10 %, which was followed by a subsequent increase to 40% in 2.5 min.This percentage was maintained for 2.5 min. A flow-rate of 1.0 ml/min wasused during elution.

For the polymeric cartridges, activation, conditioning, and sampling wereperformed with the same procedure as for C18 cartridges. The cartridge waswashed with 4.0 ml of a mixture of ammonium acetate buffer (5 mM, pH 8) andmethanol (1:1 v/v) at a flow-rate of 2.0 ml/min. Clenbuterol was eluted from thecartridge using a gradient of ammonium acetate buffer (5 mM, pH 8) andmethanol. The methanol percentage was increased from 50 to 70% in 0.5 min.This percentage was maintained for 2.0 min. A flow-rate of 1.0 ml/min wasused during elution.


129

4.1.2.3 Mass spectrometry

Using the ion-trap mass spectrometer with an APCI source, the vaporisertemperature was set at 450ºC. The sheath gas and auxiliary gas (both nitrogen)were 30 and 3 (arbitrary units), respectively. The discharge current was set at3.00 µA and the capillary voltage was 8.00 V. The temperature of the heatedcapillary was 200ºC, and the tube lens offset was set at 20.00 V. All scans wererecorded in full-scan mode with 3 microscans over the range of m/z 185 to 285using positive-ion mode. The maximum injection time was set at 200 ms.Helium was applied as cooling gas and collision gas. Extracted ionchromatograms in all MS modes were obtained for [M+H]+ or fragment ions± 0.5 Th. The isolation width during MSn experiments was 2.5 Th. Collisionenergies were optimised for individual MSn steps.



During the optimisation of the settings of the mass spectrometer, whichcan be performed automatedly or manually, the temperature of the heatedcapillary proved to be a critical factor. The highest response was found with theheated capillary set at 170ºC. However, at this temperature a significantmemory effect was observed for clenbuterol. Therefore, the heated capillarywas set at 200ºC, which resulted in a loss in sensitivity of about 25%, but nomemory effect was observed.

The fragmentation of protonated clenbuterol ([M+H]+) could bemonitored very precisely. In MS/MS mode hardly any fragments other thanm/z 259 were formed from the precursor ion m/z 277 (loss of H2O). Consecutivefragmentation of m/z 259 (MS3) produced an ion with m/z 203 as the onlyfragment (loss of the isobutene group). In MS4 experiments the ion withm/z 203 was fragmented even further. This resulted in the formation of fivefragments: m/z 186 (loss of NH3), m/z 168 (loss of Cl•), m/z 167 (loss of HCl),m/z 132 (loss of Cl• and HCl), and m/z 131 (loss of two HCl) [35]. Thecomplete fragmentation pathway is depicted in Fig. 1.

The collision energies required for the formation of fragments (Fig. 1)could be optimised in such a way that the precursor ion was almost completelyfragmented to just one daughter ion with the exception of MS4. The optimumcollision energy (arbitrary units) was 24, 30, and 35% for MS2, MS3, and MS4,respectively. The overall collision induced dissociation (CID) efficiency (ECID),which includes both fragmentation efficiency (EF) and collection efficiency(EC), was about 90, 70, and still 33% for MS2, MS3, and MS4, respectively. This


130

means that the overall ECID is 63% for MS3 and 21% for MS4. The overall ECID

in the ion-trap mass spectrometer is very high in comparison to that of atriple-quadrupole one. The EF is somewhat higher in the ion-trap massspectrometer. However, the main difference is due to the EC, since thisefficiency is up to 100% in the ion-trap and only 10-50% in a triple-quadrupole[36-37]. This difference can be explained by the fact that MS/MS in the ion-trapmass spectrometer is “tandem-in-time”, whereas with a triple-quadrupole massspectrometer “tandem-in-space” is performed. With a triple-quadrupole, theparent ion must be transmitted through a collision cell (a quadrupole andassociated lenses) with mass resolution. Scattering of ions in the collision cellresults in rather low transmission efficiency. With the ion-trap massspectrometer, all MSn processes occur in the same spatial region. Therefore, notransmission losses can occur [36].

Fig. 1: Fragmentation pathway of clenbuterol. Numbers in parentheses indicate thecollision energy required for consecutive fragmentation.


131

Ion-trap MS offers the possibility to perform MSn using wide-bandactivation. This technique involves the fragmentation of a precursor ion with achosen m/z value, and the (m/z-18) ion will automatically be fragmented aswell. For clenbuterol this implied the simultaneous fragmentation of m/z 277and 259 to form the ion with m/z 203, which is comparable to the fragmentationunder multiple collision conditions in a triple-quadrupole mass spectrometer[19]. The required collision energy (40%) was higher than for conventionalMS/MS using the ion-trap (24%). The ECID using wide-band activation wasabout 60%. This efficiency provides no improvement as compared to MS3

without wide-band activation. Therefore, wide-band activation was consideredto be advantageous only if the ions [M+H]+ and [M+H-18]+ were both present atsimilar abundances. In general, wide-band activation is 10-40% less effectivethan single frequency resonant excitation, i.e. the fragmentation of a singlem/z value [38]. Since the efficiency of MS3 analysis was rather high in this case,wide-band activation was not further investigated.

4.1.3.2 Optimisation of SPE

C18 cartridgesChromatograms obtained with diode array detection (DAD) and single

MS in the total ion current mode (TIC) showed a significant amount of urinematrix (Fig. 2A) if C18 cartridges were used. With C18 cartridges, it was notpossible to wash the cartridge with a small percentage of methanol withouteluting clenbuterol [19]. Clenbuterol is a relatively polar compound with a lowretention on a C18 stationary phase. This is very unfavourable for separatingclenbuterol from polar matrix compounds. As a result, efficient removal of theurine matrix could not be achieved. The absence of a proper wash proceduremay be unfavourable for the detection limit, since co-eluted compounds caninterfere, for example, via ion suppression [27-30] (see below), during analysisby MS.

Using the optimised conditions it was not possible to determineclenbuterol in urine below about 1 µg/ml with DAD (Fig. 2B), whichcorresponds to earlier published results [19]. Using the C18 cartridges the totalanalysis time (including sample pretreatment) was about 11.5 min. Thisprocedure was slightly shorter than previous experiments with C18 cartridges[19], in which about 13.5 min were required for a single analysis. The gain intime was mainly caused by an increase of the sampling flow to 1 ml/min. Theincrease in flow did not influence the peak shape or the sorption of clenbuterol.

The use of C18 cartridges also led to a deposit of matrix compounds on thespray shield and (the front of) the heated capillary of the ion-trap massspectrometer. Also, a decrease of the response of clenbuterol was observed over


132

one day. The system (from spray shield to skimmer) had to be cleaned at leastonce per day, which is time-consuming in routine analysis.

Fig. 2: UV detection at 247 nm of SPE eluate after extraction with a C18 cartridge of1.0 ml of (A) blank urine and (B) urine spiked with 0.9 µg/ml clenbuterol. Thechromatograms are shown from the start of the sampling step (elution startsafter 4 min).

An interesting phenomenon was observed when C18 cartridges were used.Clenbuterol eluted about 0.5 min later after extraction from urine, as comparedto extraction from buffer. Furthermore, after extraction of clenbuterol frombuffer the peak width (at the base) was about 0.8 min, while extraction fromurine produced a peak with a width of only about 0.5 min. A possibleexplanation for this difference in peak broadening and elution time might bethat the urine matrix interacts with the C18 stationary phase, thus changing theproperties of the cartridge.

Polydivinylbenzene cartridgesIn order to improve the selectivity a polymeric phase was studied. Using

cartridges with the polymeric phase, it proved to be possible to decrease theamount of matrix observed in the chromatogram. The polymeric phase chosenwas polydivinylbenzene. Apart from its more hydrophobic characteristics, thearomatic rings of the stationary phase also permit π-π interactions between thesorbent and the analyte and, therefore, an increase in retention. As a result, thepolymeric phase was expected to be more retentive in comparison with the C18

cartridges [39]. It was possible to wash more rigorously, using a mixture ofammonium acetate buffer and a relatively high percentage of methanol, withouteluting clenbuterol from the cartridge. Nevertheless, if the polymeric phase isused there is still some matrix present after extraction of blank urine (Fig. 3A).


133

Fig. 3: UV detection at 247 nm of SPE eluate after extraction with a PDVB cartridgeof 1.0 ml of (A) blank urine and (B) urine spiked with 0.9 µg/ml clenbuterol.The chromatograms are shown from the start of the sampling step (elutionstarts after 4 min).

For the cartridges with the polymeric phase, a mixture of ammoniumacetate buffer and methanol could be used during the wash step. If a mixture of60:40 v/v ammonium acetate buffer (5 mM, pH 8)/methanol was used forwashing the cartridge, after injection of 1 ml of urine, a significant amount ofmatrix was still present in the eluate. As with C18 cartridges, this resulted in adeposit of matrix compounds on the spray shield and (the front of) the heatedcapillary and a decrease in response of clenbuterol. The system had to becleaned at least once per day. Furthermore, the response of clenbuterol wassignificantly higher when standard solutions were analysed, as compared withurine samples. This indicates that washing with 40% methanol could notprevent ion suppression when urine was analysed.

During the analysis of standard solutions, washing with a mixture of50:50 v/v ammonium acetate buffer (5 mM, pH 8)/methanol was found to leadto a similar response to washing with 40% methanol. This indicates that nobreakthrough occurred. In contrast, analysis of urine samples washed with 50%methanol produced responses that were twice as high (concentration range50-250 ng/ml) as those observed when washing urine-sampled cartridges with40% methanol, indicating that less interference of matrix, i.e. ion suppression,occurred. Moreover, by loading the cartridge with urine and consecutivewashing with 50% methanol, no deposit of matrix on the spray shield and theheated capillary was observed. For precautionary reasons the system wascleaned once per week. However, no decrease of response was observed duringa week.

For further experiments 50% methanol was used during the washprocedure whenever the cartridges with the polymeric phase were used. Using


134

an even higher percentage of methanol during the wash step resulted inbreakthrough of clenbuterol. The elution of clenbuterol occurred using agradient of methanol starting with 50:50 v/v ammonium acetate buffer (5 mM,pH 8)/methanol (i.e. the mixture of the wash step), and increasing thepercentage methanol to 70% within 0.5 min. A good peak shape was obtained,but it was not possible to determine clenbuterol in urine below about 1 µg/mlwith DAD (Fig. 3B). However, it should be noted that the total analysis timewas now reduced to 8.5 min including sample pretreatment. This is mainly dueto the fact that it was possible to wash with a relatively high percentage ofmethanol. As a consequence, a faster gradient with a smaller range could beused during elution, and only about 2.5 min data acquisition was required.

4.1.3.3 Use of the total system

Comparison of ion-suppression effectsThe reliability of quantitative results provided by an LC-MS system may

not be absolute. Co-eluting ‘unseen’ endogenous compounds may interfere withthe accurate determination of the analyte of interest. The undetected matrixcomponents may reduce the ion intensity of the analyte and affect thereproducibility and accuracy of the LC-MS method [27-30]. Ion suppression isinherently independent of the MS mode, as ionisation and MS analysis areseparated in time and space [27]. Under routine analysis conditions, using adeuterated version of the analyte, or a close homologue, as an internal standard,suppression effects can be corrected for. However, even then it is necessary tostudy the cause and amount of suppression.

Quantitative information about ion suppression can be obtained bycomparing the response of the analyte in the biological matrix with the responsein the mobile phase. It is assumed that the difference in response is caused bycomponents of the extract that are not present in the mobile phase. Anotheruseful experiment is to extract a blank sample, which gives qualitativeinformation about the interfering compounds. A post-column infusion ofanalyte combined with extraction of a blank sample will also providequantitative information about the ion suppression [28]. Polar compounds aremore sensitive to ion suppression than non-polar components [28]. Using anSPE-MS system to determine a relatively polar compound, e.g. clenbuterol, ionsuppression should be carefully monitored. A comparison of ion-suppressioneffects, using external calibration, at different concentrations with C18 cartridgesand PDVB cartridges is presented in Table 1. The ion suppression was studiedusing the MS3 mode, since this mode offered the best selectivity and sensitivity(see below).


135

Table 1: Ion suppression (%) of different concentrations of clenbuterol extracted fromurine using a PDVB cartridge and a C18 cartridge measured with MS3.

93 ng/ml 45 ng/ml 9 ng/mlPDVB 4 27 32C18 37 69 No data

As can be observed from the data in Table 1, the ion suppression washigher at lower concentrations with both cartridges, which can be explained bythe less favourable matrix/analyte ratio. Also, ion suppression was moreprominent after extraction of clenbuterol from urine with C18 cartridges, whichis due to the fact that the cartridge could not be washed with methanol withouteluting clenbuterol. Therefore, many matrix compounds were still present in theeluate, causing ion suppression. Despite high ion suppression, only a smallrelative standard deviation (<6%) was obtained. This indicates that the ionsuppression is almost constant at a certain concentration in the set-up usedduring these experiments. However, using other urine samples will mostprobably show dissimilar ion suppression, as other interfering compounds canbe present and at different concentration levels.

Analytical dataUsing both the C18 and the PDVB cartridges, linearities and LODs were

determined for the extraction of clenbuterol from buffer and urine. Several MSmodes were used. The results are presented in Table 2.

Table 2: Detection limit, linear range and regression coefficient (R) using PDVBcartridges and C18 cartridges and the MS, MS2 and MS3 mode.

PDVB C18

LOD(ng/ml)

Range(ng/ml)

Ra LOD(ng/ml)

MSBuffer 2.5 5-250 0.9950 (n=6) 2.5Urine 25 50-250 0.9910 (n=3) 50MS2

Urine 50 ---b ---b ---b

MS3

Urine 0.5 1.0-250 0.9988 (n=8) 2.0a: Weighted regression (1/x)b: Not measured


136

The use of single MS did not provide the required selectivity to determinelow levels of clenbuterol in urine (Table 2). A large amount of matrix wasobserved in all TIC chromatograms. Extracting the molecular ion (m/z 277)gave a reduction of the visible interfering matrix (Fig. 4), but total eliminationof the matrix effect by extracting any specific m/z value proved to beimpossible.

Fig. 4: Extracted ion chromatograms at m/z 277 after extraction of 50 ng/mlclenbuterol in urine with SPE-single MS using (A) a C18 cartridge and(B) a PDVB cartridge.

The matrix interference is probably due to endogenous compounds, normallypresent in urine, e.g. amines, urea, lipids, and proteins. Alternatively, theseinterferences may also result from compounds originating from digested foodand drinks, or administered drugs and their metabolites. It should be noted thatthe difference in selectivity of the stationary phases also affected the presenceof the interfering compounds. With C18 cartridges m/z 248, 300 and 414 werethe most prominent ions, whereas with PDVB cartridges m/z 248, 274 and 276appeared to interfere the most with the accurate determination of clenbuterol.None of the interfering compounds could be positively identified.

Generally, sensitivity can be improved by operating the mass analyser ina higher MS mode (MSn, where n≥2), leading to a more significant reduction ofthe noise than the signal. However, with clenbuterol, MS2 led to a loss ofsensitivity (Table 2). This is due to the presence of an interfering compound thatalso produces a fragment with m/z 259 (Fig. 5). Most probably, the precursorion of this fragment (m/z 276) is an ammonium adduct, which loses NH3 in thefirst fragmentation step. Alternatively, as many endogenous compounds presentin urine contain an amine group, the fragment may originate from a primaryamine, also losing NH3 in the first fragmentation step.


137

Fig. 5: Extracted ion chromatogram at m/z 259 after extraction of 50 ng/ml clenbuterolin urine with a PDVB cartridge and detection by MS/MS.

Fig. 6: Extracted ion chromatogram at m/z 203 after extraction of 1.0 ng/mlclenbuterol in urine with a PDVB cartridge and detection by MS3.


138

In contrast, for MS3, the noise could be reduced more than the clenbuterolfragment signal at m/z 203. This resulted in an associated increase in sensitivity.The LOD was 0.5 ng/ml and 2.0 ng/ml with PDVB and C18 cartridges,respectively (Table 2). A representative MS3 chromatogram is depicted inFig. 6. Applying MS4 did not lead to any further improvement in sensitivity, i.e.the LOD was 10 ng/ml clenbuterol in urine. This is mainly due to inefficientfragmentation of the ion at m/z 203 to a total of five fragments (Fig. 7).

In all MS modes presented in Table 2, linearity (R) was always betterthan 0.99. As expected, the LODs for standard solutions in buffer were aboutthe same in all the MS modes.

Fig. 7: Mass spectrum of clenbuterol after SPE-MS4 analysis(m/z 277 à m/z 259 à m/z 203 à full scan [m/z 185-285]).

4.1.4 Conclusions

Direct coupling of SPE with ion-trap MS offers the possibility todetermine drugs in biological samples with a short analysis time. Ion-trap MSoffers the possibility of very selective and sensitive detection, especially if MSn

(n≥2) is used. However, special attention must also be paid to the SPEprocedure to ensure that clean extracts are obtained. This is essential for reliableanalysis, since the presence of matrix compounds can increase the backgroundsignal and also leads to reduction of the signal of the analyte due to ionsuppression during the formation of the ions in the LC-MS interface.Summarising, development of an SPE-MS method implies considering the


139

required selectivity at the front of the analysis system (SPE) as well as at theend (MS).

At the front of the system, the stationary phase material proves to be animportant parameter. Using C18 cartridges, an LOD of 2 ng/ml was obtained forSPE-MS3 using an ion-trap mass spectrometer. The use of PDVB cartridgesresulted in cleaner extracts, a lower detection limit (0.5 ng/ml), and a shorteranalysis time (8.5 min). The cleaner extracts are probably caused by π-πinteractions between the phenyl ring of clenbuterol and the phenyl groups of thestationary phase, so that more efficient washing of the cartridge can be realised.The use of this polymeric phase can possibly be expanded to screening for otherdrugs with an aromatic functionality, e.g. fingerprinting of cocaine oramphetamines. In the future, more selective SPE, e.g. using molecular imprintsand immobilised antibodies, will be investigated to obtain even cleaner extracts.

The use of ion-trap MS is relatively new for applications in bioanalysis.To our knowledge, this type of MS was not previously used in an SPE-MSsystem. From the results in this work it can be concluded that the ion-trap MScan be used for analysis without a real chromatographic separation step. If cleanextracts can be obtained, the use of an SPE-(ion-trap) MS system offers thepossibility to perform an analysis, including sample pretreatment, in less than10 min. Thus, high sample throughput can be achieved; however, the SPEprocess needs to be further optimised in order to obtain sufficient compatibilitywith the speed of the MS step. The SPE might be shortened even further byfaster activation, conditioning, and washing of the cartridge or by reduction ofthe volumes in each step in the sample pretreatment procedure. However, thiscan also negatively influence the sorption and desorption characteristics of thestationary phase. A set-up using two cartridges parallel to each other [20] willalso help to increase the sample throughput.

In conclusion, on-line SPE-(ion-trap) MSn is very promising and thesystem will be tested for other compounds and matrices in the near future.

Acknowledgements

The authors are very grateful to Spark (Emmen, The Netherlands) forproviding a Prospekt system. This research was supported by the TechnologyFoundation STW, applied science division of NWO and the technologyprogramme of the Ministry of Economic Affairs.


140

4.1.5 References

[1] M.L. Constanzer, C.M. Chavez, B.K. Matuszewski, J. Carlin, D. Graham.J. Chromatogr. B 693 (1997) 117.

[2] N.C. van de Merbel, A.P. Tinke, W.D. van Dongen, B. Oosterhuis,J.H.G. Jonkman, Ph. Ladure, C.Puozzo. J. Chromatogr. B 708 (1998) 113.

[3] D.A. McLoughlin, T.V. Olah, J.D. Gilbert. J. Pharm. Biomed. Anal. 15 (1997)1893.

[4] R.E. Majors, H.G. Barth, C.H. Lochmüller. Anal. Chem 56 (1984) 300R.[5] R.E. Majors. LC•GC Int. May (1998) 8.[6] M. Jemal, D. Teitz, Z. Ouyang, S. Khan. J. Chromatogr. B 732 (1999) 501.[7] J.P. Allanson, R.A. Biddlecombe, A.E. Jones, S. Pleasance. Rapid Commun.

Mass Spectrom. 10 (1996) 811.[8] R.W. Frei, K.Zech (eds). Selective sample handling and detection in high-

performance liquid chromatography (part A), Elsevier Science Publishers,Amsterdam, The Netherlands (1988).

[9] E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman. Chromatographia 29 (1990)415.


[11] E.H.R. van der Wal, E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman.Chromatographia 39 (1994) 239.

[12] E.C. Goosens, D. de Jong, G.J. de Jong, U.A.Th. Brinkman. Chromatographia 47(1998) 313.

[13] A.J.H. Louter, E. Bosma, J.C.A. Schipperen, J.J. Vreuls, U.A.Th. Brinkman.J. Chromatogr. B 689 (1997) 35.

[14] A. Namera, M. Yashiki, Y. Iwasaki, M. Ohtani, T. Kojima. J. Chromatogr. B 716(1998) 171.

[15] K.K. Verma, A.J.H. Louter, A. Jain, E. Pocurull, J.J. Vreuls, U.A.Th. Brinkman.Chromatographia 44 (1997) 372.

[16] D. Jahr. Chromatographia 47 (1998) 49.[17] P. Enoch, A. Putzler, D. Rinne, J. Schlüter. J. Chromatogr. A 822 (1998) 75.[18] A.J.H. Louter, R.A.C.A. van der Wagt, U.A.Th. Brinkman. Chromatographia 40

(1995) 400.[19] C.H.P. Bruins, C.M. Jeronimus-Stratingh, K. Ensing, W.D. van Dongen, G.J. de

Jong. J. Chromatogr. A 863 (1999) 115.[20] A. Schellen, B. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de




Niessen, U.A.Th. Brinkman. Rapid Commun. Mass Spectrom. 10 (1996) 1569.[25] G.D. Bowers, C.P. Clegg, S.C. Hughes, A.J. Harker, S. Lambert. LC•GC 15

(1997) 48.[26] J.A. Ooms, G.S.J. Haak, O. Halmingh. Intern. Lab. 27 (1997) 18A.


141

[27] K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng. Anal. Chem. 70 (1998)882.

[28] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle. Rapid Commun. Mass Spectrom.19 (1999) 1175.

[29] D.L. Buhrman, P.I. Price, P.J. Rudewicz. J. Am. Soc. Mass Spectrom. 7 (1996)1099.

[30] I. Fu, E.J. Woolf, B.K. Matuszewski. J. Pharm. Biomed. Anal. 18 (1998) 347.[31] M. Lafontan, M. Berlan, M. Prud’Hon. Reprod. Nutr. Dev. 28 (1988) 61.[32] G. Brambilla, T. Cenci, F. Franconi, R. Galarini, A. Macri, F. Rondoni,

M. Strozzi, A. Loizzo. Toxicol. Lett. 114 (2000) 47.[33] V. Sporano, L. Grasso, M. Esposito, G. Oliviero, G. Brambilla, A. Loizzo. Vet.

Hum. Toxicol. 40 (1998) 141.[34] A. Koole, J. Bosman, J.P. Franke, R.A. de Zeeuw. J. Chromatogr. B 726 (1999)

149.[35] G. Biancoto, R. Angeletti, R.D.M. Piro, D. Favretto, P. Traldi. J. Mass Spectrom.

32 (1997) 781.[36] J.V. Johnson, R.A. Yost, P.E. Kelley, D.C. Bradford. Anal. Chem. 62 (1990)

2162.[37] B.A. Mansoori, E.W. Dyer, C.M. Lock, K. Bateman, R.K. Boyd, B.A. Thomson.

J. Am. Soc. Mass Spectrom. 9 (1998) 775.[38] R.E. March, J.F.J. Todd (Eds.). Practical Aspects of Ion Trap Mass Spectrometry.

Part III: Chemical, Environmental, and Biomedical Applications, CRC Press,Boca Raton, FL, USA (1995) p.167.

[39] E.M. Thurman, M.S. Mills, J.D. Winefordner (Eds). Solid-Phase ExtractionPrinciples and Practice, John Wiley & Sons, Inc., New York, NY, USA (1998)p.36.

Ion suppression in the determination of clenbuterol in urine by SPE-MS3

143

4.2

Ion suppression in the determination ofclenbuterol in urine bysolid-phase extraction – mass spectrometry*

Summary

Ion suppression effects were observed during the determination of clenbuterolin urine with solid-phase extraction – multiple-stage mass spectrometry(SPE-MS3), in which a polymeric stationary phase (polydivinylbenzene) wasapplied. Post-cartridge infusion of analyte to the SPE eluate after the extractionof blank urine was performed to obtain a profile of the suppression. Single andmultiple-stage MS was performed to get insight in the suppressing compounds.The ion suppression was mainly ascribed to two m/z values, but still noidentification of the compounds was achieved from the multiple-stage MS data.No ionisable and non-ionisable complexes and/or precipitation of clenbuterolwith matrix compounds were observed. A concentration dependence of thepercentage of suppression was observed. Up to 70% of the signal wassuppressed upon post-cartridge infusion of 0.22 µg/ml (at 5 µl/min) clenbuterolinto the eluate, and this decreased to about 4% at infusion of 22 µg/mlclenbuterol. Molecularly imprinted polymers were used to enhance theselectivity of the extraction. Although matrix components were still presentafter extraction, no interference of these compounds with the analyte wasobserved. However, the bleeding of the imprint from the polymer (brombuterol)caused significant ion suppression.

*: M.W.J. van Hout, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de Jong. Submitted to Rapid Commun. MassSpectrom.


144

4.2.1 Introduction

In the biopharmaceutical field, fast, selective and sensitive analyticalsystems are required. These seemingly incompatible requirements may be metby systems based on on-line liquid chromatography (LC) and massspectrometry (MS) with atmospheric interfaces [1,2]. A key factor inbioanalysis is, however, the sample pretreatment. A powerful technique forclean-up and preconcentration of biological samples is solid-phase extraction(SPE). When carried out off-line, the extraction procedure can be time-limiting.Automation of SPE and on-line coupling with LC-MS implies that theseparation step becomes the most time-consuming step. Consequently, short LCcolumns (25 to 50 mm) are applied [3-5], or the actual LC separation wasomitted, i.e. SPE directly coupled to MS [6-12]. Both approaches resulted insystems in which the extraction was again limiting the sample throughput.

The application of an SPE-MSn system for bioanalytical purposes impliesthat the SPE procedure must be performed very rapidly for high-throughputaims, whereas it should also be efficient to ensure that clean extracts areobtained. The selectivity of the MS can eliminate the signals of co-eluting(matrix) compounds, but their presence in the eluate may still cause ionsuppression [13-17], resulting in the loss of reliability and accuracy of theobtained data.

The effect of ion suppression in LC-MS systems has been recognisedlately and a number of reports can be found in literature [10,11,18-32]. Thecauses of ion suppression with electrospray ionisation (ESI) can be a decreasein evaporation of the solution or an increase in surface tension due to (highconcentrations of) matrix compounds [16,17]. Another pathway is binding toand (co-)precipitation of the analyte with non-volatile materials [17].Alternatively, a competition between compounds for the charge in the liquidphase may result in ion suppression as well [18]. A final possible cause isgas-phase neutralisation processes [17]. The latter, as well as theco-precipitation, is of particular importance with atmospheric pressure chemicalionisation (APCI). With APCI, gas-phase basicity is the key. Only the strongestgas-phase base (positive mode) will be ionised [19].

A general trend can be observed in LC-MS applications. Particularly ESIis prone to ion suppression [10,18-20,23-32], although ion suppression withAPCI has also been described [11,22]. The stronger effect for ESI can be partlyexplained by the higher tendency towards adduct formation with ESI [28].Furthermore, with ESI an upper limit of the total number of ions (typically10-5 M) is observed, which depends on the surface area of all droplets beingformed during ESI [33]. Thus, a saturation of the ionisation can occur. In allapplications, problems are observed when analysing relatively polarcompounds, i.e., compounds with a high gas phase acidity or basicity, in the


145

presence of even more polar interfering matrix compounds [33]. Thus, for theelimination of the ion suppression, one should attempt to decrease the amountof polar matrix compounds present after the sample preparation step.

Various reports describe the presence of ion suppression after SPE with aC18 stationary phase [11,23,28,29,32], but the effects are also observed afterextraction with an ion-exchange phase [30] or with liquid-liquid extraction [20].Various solutions have been presented to correct for and/or minimise the ionsuppression. The use of a suitable internal standard, e.g. a deuterated analogueor a structural analogue of the analyte, may be used, provided that a similarsuppression effect is observed for the internal standard [24,26,28,30]. Extractinga smaller volume of the sample [20] may also be useful to reduce the ionsuppression due to reduction of the amount of interfering compounds. Otherapproaches for reducing or eliminating ion suppression can be the use of LC/LC[20,22,23] or a change in sample preparation [19,25,29,32]. The latter ispreferred to differ in selectivity from the separation step (if applied) [25]. Anadditional step in the extraction, such as SPE followed by preparative LC andsubsequent LC-MS analysis has also been applied [32]. All reports onlymention the presence of ion suppression, but no identification of thesuppressing compounds was described.

A few methods have been proposed for determining interferences ofmatrix components [15]. Extracting a blank sample with subsequent separationand/or detection will give qualitative information about the interferingcompounds. A post-column infusion of analyte to the SPE eluate after theextraction of a blank sample will provide quantitative information about the ionsuppression. With this system, a time-profile of the suppressing compounds canalso be obtained. Another method that provides quantitative information iscomparing the response of the analyte after extraction from the biologicalmatrix with the response after extraction from a standard solution. It is assumedthat the difference in response is caused by components of the extract that arenot present in the standard solution, but one should be aware that any losses inthe extraction procedure are also included in this approach. With off-line SPE,this can be encountered by spiking the eluate after extraction. However, in anon-line system spiking the eluate is not possible.

From a previous study [11], it was observed that the best results for thedetermination of clenbuterol in urine with an SPE-MSn system with an APCIinterface were obtained using polydivinylbenzene (PDVB) cartridges. However,some ion suppression (about 4%) was observed at high concentrations ofclenbuterol, and more ion suppression (up to 40%) was observed at lower levels(<10 ng/ml) of clenbuterol. These results were the basis for the current study.The ion suppression will be studied with regard to the origin and the amount ofthe suppression. More selective extraction, by using molecularly imprintedpolymers (MIPs), will be applied in order to investigate the possibilities of


146

MIPs in an SPE-MSn system with respect to the extraction selectivity and theion suppression.

4.2.2 Experimental


All on-line SPE-MS experiments were performed with a Prospekt samplehandler (Spark, Emmen, The Netherlands) using one six-port valve, thecartridge-switching device, and a solvent delivery unit (SDU). Activation,conditioning, sampling, trapping, and washing were done using the SDU. Theeffluent was connected to waste during these steps. All steps of the SPEprocedure were carried out using a forward-flush mode. Cartridges wereautomatedly replaced after single use. A second flow stream was used for theelution applying a gradient pump Series 1100 (Hewlett-Packard, Waldbronn,Germany), which was connected to an LCQ ion-trap mass spectrometer(Thermoquest, San Jose, CA, USA) via the cartridge. The mass spectrometerwas equipped with an APCI source. Experiments with UV detection werecarried out with a Hewlett-Packard diode array detector Series 1100. TheProspekt cartridges were HySphere Resin GP (Spark, 10×2 mm, particle size10 µm), a spherical polymeric phase of polydivinylbenzene (PDVB), or MIPs(MIP Technologies, Lund, Sweden, brombuterol imprinted, 10×2 mm, polymerparticle size 25-38 µm).

Methanol and acetonitrile were of HPLC grade (Lab Scan, Dublin,Ireland). Ammonium acetate and acetic acid were of analytical-reagent grade(Merck, Darmstadt, Germany). Trifluoroacetic acid (TFA, spectrophotometricgrade 99+%) was obtained from Aldrich Chemical (Milwaukee, WI, USA). The5 mM ammonium acetate buffer was adjusted to pH 8 using 2.5% ammonia(analytical-reagent grade). Water was obtained from an Elgastat maxima system(ultra pure water, Salm and Kipp, Breukelen, The Netherlands). Aqueoussolutions were filtered over a 0.45-µm RC 55 membrane filter (Schleicher &Schuell, Dassel, Germany) prior to use. Clenbuterol.HCl (Sigma Aldrich,Dorset, United Kingdom) was dissolved in methanol (1 mg/ml) and stored inthe dark at -20ºC. Human male urine (samples #1 and #2) and female urine(sample #3), as well as calf (sample #4) and bovine (sample #5) urine wereused. Spiking of the urine samples was performed by addition of a smallvolume of the stock solution that is adequately diluted with buffer to urine.


147

4.2.2.2 Procedures

The SPE procedure with PDVB cartridges was based on a previouslydescribed method [11], but higher flow-rates were used during the various stepsof the extraction procedure in order to obtain a rapid SPE-MSn system. Thefollowing procedure was used: conditioning was performed with 2.5 ml 5 mMammonium acetate buffer (pH 8) at a flow-rate of 5.0 ml/min. The 1-ml samplewas then loaded onto the cartridge with an excessive amount of buffer during1 min at 3.0 ml/min, followed by a washing step with 5.0 ml buffer/methanol(1:1 v/v) at a flow-rate of 5.0 ml/min. The elution was performed at a flow-rateof 1.0 ml/min with a gradient of ammonium acetate buffer (5 mM, pH 8) andmethanol. In 0.25 min the methanol percentage was increased from 50 to 70%.The latter percentage was maintained for 1.75 min.

Based upon a previously described method [34], the following extractionprocedure was used with the MIP cartridges. Conditioning with 1.25 ml waterwas applied. The 1-ml sample was loaded onto the cartridge within 1 min at aflow-rate of 3.0 ml/min. The cartridge was then washed at 5.0 ml/min with5.0 ml acetonitrile containing 1% acetic acid. The elution (1 ml/min) wasstarted with a mixture of water/methanol (60/40 v/v) containing 5 mM TFA.Within 1 min this was replaced by acetonitrile/methanol (80/20 v/v, 5 mMTFA). This mixture was maintained for 2 min.

The continuous-infusion method was applied to determine the profile ofthe ion suppression. Unless stated otherwise, clenbuterol (0.5 µg/ml inmethanol) was infused post-cartridge at 5 µl/min. Data acquisition with the MSwas always started 0.5 min prior to the start of the elution.


During the experiments with the ion-trap mass spectrometer with anAPCI source, the vaporiser temperature was set at 450ºC. The sheath gas andauxiliary gas (both nitrogen) were 65 and 10 (arbitrary units), respectively. Thedischarge current was set at 5.00 µA and the capillary voltage was 24.00 V. Thetemperature of the heated capillary was 200ºC, and the tube lens offset was setat 25.00 V. All scans were recorded in the full-scan mode with 3 microscansover the range of m/z 70 to 500 using the positive-ion mode. The maximuminjection time was set at 200 ms. Protonated clenbuterol ([M+H]+, m/z 277) wasmonitored during the single MS mode, whereas m/z 259 and 203 weremonitored in the MS2 and MS3 mode, respectively. Extracted ionchromatograms in all MS modes were obtained for [M+H]+ or fragment ions± 0.5 Th. The isolation width during MSn experiments was 1.8 Th. Helium wasapplied as cooling gas and collision gas. The collision energies were 25, 28 and35% during MS2, MS3 and MS4 experiments, respectively.


148

4.2.3 Results and Discussion

4.2.3.1 Extraction with PDVB

Suppression profileThe changes in the SPE procedure (see Section 4.2.2.2) in comparison to

a previously described method [11] implied that the total analysis time wasdecreased to 4 min. Analysis of blank urine with single MS showed a largematrix peak (Fig. 1A).

Fig. 1: (A) SPE-MS of blank urine applying PDVB cartridges (total ionchromatogram). (B) SPE-MS3 of blank urine with post-cartridge infusion ofclenbuterol (0.5 µg/ml, 5 µl/min); extracted ion chromatogram (XIC) m/z 203.(C) SPE-MS3 of clenbuterol in urine (100 ng/ml); XIC m/z 203.


149

Numerous m/z values with varying abundances and elution times were found inthis large peak originating from sample #1. Four other urine samples of human(male and female), bovine and calf origin were also monitored using single MS,and similar results were obtained with regard to elution times and m/z valuesfound in blank urine. Upon post-cartridge continuous infusion of clenbuterolwhile extracting blank urine, two suppression regions, that is from 0.4-1.2 andfrom 1.6-2.0 min, were observed (Fig. 1B). Only the first region interfered withthe determination of clenbuterol, since the elution window of clenbuterol(0.9-1.5 min, Fig. 1C) and the first suppression region partially overlap. Thus,the focus was on the matrix compounds in this overlapping elution window(0.9-1.2 min).

Elimination of the ion suppression may be accomplished by changing thepH in the wash step. When increasing the pH from 8.0 to 9.5, no differences inthe presence of matrix compounds and the elution of clenbuterol was observed.Applying an acidic wash step, at pH 4.5, clenbuterol eluted from the cartridgeduring the wash step. Thus, in this approach, changing the pH was not useful.Therefore, the cause of the ion suppression was further investigated for theoriginal system.

Complex formation / precipitationThe ion suppression can be due to various processes. One of those is the

formation of a (non-ionisable) complex of the matrix component and the analyte[10], which subsequently deposits [17] on the surface of the spray shield andheated capillary. Another possibility of a (ionised) complex of clenbuterol withcreatinine has been observed after extraction on an ion-exchange cartridge andanalysis with ESI/MS [10], but in the present set-up no such complex was seen.The formation of such a complex after extraction on PDVB cartridges and/orprecipitation was checked by extraction of about 100 ng/ml clenbuterol frombuffer, buffer/urine (4:1 and 1:1) and urine. After five extractions of one of thefour solutions mentioned above the spray shield and heated capillary werewashed, prior to extraction of another solution, with 5 ml methanol and 5 mlbuffer pH 8, hereby cleaning the MS and breaking a possible complex of matrixcompounds with clenbuterol. All fluids were collected, and subsequentlytriplicate injections into the MS were performed. About 1% of the summedamount of clenbuterol of the five extractions was found in the wash solventsregardless of the samples being extracted. Thus, the ion suppression was notcaused by the formation of a non-ionisable complex and subsequentprecipitation. This suggests that the suppression is caused by a competition ofthe analyte and the matrix components for the charge in the gas phase.


150

Concentration dependenceIn the previous study [11] a concentration dependence in the ion

suppression was observed upon the comparison of the extractions from spikedstandard solutions and from spiked urine. Below 10 ng/ml (or <6 ng/min with apeak width of 0.6 min and injecting 1 ml), the suppression was about 40%(absolute) and this gradually decreased to only 4% at higher concentrations(>100 ng/ml, or >60 ng/min). The continuous infusion method with varyingconcentrations was now also used to study this effect. The observedsuppression, averaged over the elution window of clenbuterol, was about 70%with the infusion of 0.22 µg/ml clenbuterol into the eluate at 5 µl/min (i.e.1.1 ng/min) and decreased to about 4% during the infusion of 22 µg/ml at5 µl/min (110 ng/min). The decreasing percentage of ion suppression withincreasing concentrations is probably due to a more favourable analyte/matrixratio at higher concentrations. The absolute amount that was suppressedincreased with increasing infusion concentrations (0.22 to 1.0 µg/ml). Withhigher concentrations, the suppressed amount reached a plateau and remainedalmost constant at about 3 ng. This indicates that there is indeed a competitionfor the charge in the gas phase between the analyte and the matrix molecules.The latter can suppress the analyte until the matrix components are fullycharged. This explains the increase and subsequent stabilisation of thesuppressed amount of clenbuterol if the infused concentration is increased.

Structural information about suppressing compoundsThree m/z values, i.e., m/z 257, 274 and 276, were considered to be

possibly linked to the ion suppression effects of clenbuterol, since compoundsgiving these m/z values were present in the elution window of interest.However, m/z 257 was related to m/z 274 by the loss of NH3, thus the number ofm/z values of interest was reduced to only two. The m/z values 274 and 276were present in all urine samples, and are thus considered to be common urinecomponents. More structural information of the compounds at m/z 274 and 276was obtained by performing MS2, MS3 and MS4. A scheme of the fragmentationpattern of these compounds is presented in Fig. 2. Despite the fact that manyurine compounds are known [35], and even though some functional groups ofthe components could be specified, so far no positive identification could bemade. This is partly due to the complexity of the urine matrix, since manysubstances can be present in this sample type, which originate from variousexogenous and endogenous sources. Furthermore, several metabolismprocesses, e.g. hydroxylation, methylation, acetylation, glucuronidation,sulfonation and combinations of these reactions, can result in the presence ofmany compounds besides the parent molecules. The unknown compounds aremost likely not sulfonated, since no isotope peaks of sulfur were observed.


151

Furthermore, glucuronidation and methylation are not likely either, as thesegroups would immediately be lost in MS/MS analysis [36-39].

Fig. 2: Fragmentation pattern and possible functional groups of two main interferingm/z values.


152

4.2.3.2 Extraction with MIPs

More selective extraction of clenbuterol could possibly be obtained with(brombuterol) MIP cartridges. A representative chromatogram is presented inFig. 3A. The extraction of blank buffer already showed severe interference ofthe brombuterol imprint (peak #1, 0.5-1.2 min). Upon the extraction of urine, asecond peak (peak #2, 1.2-2.0 min) was observed, which originated from theurine matrix. The m/z values of the matrix compounds (peak #2) present afterextraction with the MIP cartridges were different from those present afterextraction with the PDVB cartridges, and m/z 274 and 276 were not present.

Fig. 3: (A) SPE-MS of blank urine applying (brombuterol) MIP cartridges (total ionchromatogram). (B) SPE-MS3 of blank urine with post-cartridge infusion ofclenbuterol (0.5 µg/ml, 5 µl/min); XIC m/z 203. (C) SPE-MS3 of clenbuterol inurine (100 ng/ml); XIC m/z 203.


153

Continuous infusion of clenbuterol (0.5 µg/ml at 5 µl/min) showed that no ionsuppression was observed at peak #2. Hardly any ionisation of clenbuterol wasobserved from 0.5-0.7 min at the infused analyte concentration (Fig. 3B).Peak #1 was confirmed to be due to bleeding of brombuterol from the polymer.Thus, the significant amount of brombuterol, which co-eluted with clenbuterol(Fig. 3C), caused the ion suppression. More extensive washing of the polymerbefore use may reduce this effect. However, due to time-limitations, no suchexperiments were conducted.

When comparing the results of the extraction of spiked urine with aPDVB or a MIP cartridge (Figs. 1B and 1C versus 3B and 3C, respectively), itcan be concluded that the suppression (percentage) of clenbuterol is moreprominent due to the bleeding of brombuterol from the polymer (MIPextraction) than due to matrix compounds (PDVB extraction), since thedecrease of the clenbuterol signal is larger (Figs. 1B and 3B) and the peak area(or abundance) is much higher after extraction with PDVB cartridges (Figs. 1Cand 3C).

4.2.4 Conclusions

Ion suppression is a critical aspect in SPE-LC-MS systems forbioanalysis, and it becomes even more important to monitor this effect when theLC column length is reduced or the LC separation step is omitted. Post-columncontinuous infusion of analyte to the SPE eluate after the extraction of a blanksample can be used to get insight into the extent of the ion suppression.Subsequent single and multiple MS analysis of blank urine will then provideinformation about the interfering compounds. However, the use of an LCseparation of the interfering compounds (which should also be capable ofhandling a large polarity range) as well as high-resolution MS may be useful toincrease the chance for identification of the interfering compounds. In an ionsuppression study, samples from different sources should be used. However, thecurrent study showed that common urine compounds seem to be the suppressingcomponents.

Ion suppression should be an important consideration during thevalidation of a method. If ion suppression is observed, the selectivity of theextraction should be optimised by changing the washing or elution step in theprocedure of the extraction or by the choice of another stationary phase. MIPswere considered to be very suitable in order to prevent ion suppression, but thebleeding of the imprint from the polymer was shown to be significant andseverely suppressed the ionisation of clenbuterol. The bleeding of the MIPsneeds to be minimised, otherwise this type of extraction might not be suitablefor direct coupling to MS. One should thus always be alert that any compound,


154

matrix and non-matrix related (e.g. the internal standard), can cause ionsuppression when present in a relatively high concentration. However, incombination with a real separation step the selectivity obtained with the MIPextraction principle is often different from the LC separation, thus this type ofextraction combined with a short LC column and MS detection can more easilyresult in the prevention of ion suppression than the combination of a commonreversed-phase extraction and a similar separation step. Furthermore, otherselective SPE materials, such as immobilised antibodies, seem very promisingin this respect.

Acknowledgements

A.P. Bruins is acknowledged for helpful discussions. The authors are verygrateful to Spark (Emmen, The Netherlands) for providing a Prospekt system.MIP cartridges were kindly donated by MIP Technologies (Lund, Sweden).This research was supported by the Technology Foundation STW, appliedscience division of NWO and the technology programme of the Ministry ofEconomic Affairs.

4.2.5 References

[1] A.P. Bruins, T.R. Covey, J.D. Henion. Anal. Chem. 59 (1987) 2642.[2] T. Yasuda, M. Tanaka, K. Iba. J. Mass Spectrom. 31 (1996) 879.[3] D.A. McLoughlin, T.V. Olah, J.D. Gilbert. J. Pharm. Biomed. Anal. 15 (1997)

1893.[4] M.L. Constanzer, C.M. Chavez, B.K. Matuszewski, J. Carlin, D.Graham.

J. Chromatogr. B 693 (1997) 117.[5] N.C. van de Merbel, A.P. Tinke, W.D. van Dongen, B. Oosterhuis,

J.H.G. Jonkman, Ph. Ladure, C. Puozzo. J. Chromatogr. B 708 (1998) 113.[6] W.A. Minnaard, A.C. Hogenboom, U.K. Malmqvist, P. Manini, W.M.A.

Niessen, U.A.Th. Brinkman. Rapid Commun. Mass Spectrom. 10 (1996) 1569.[7] A. Schellen, B. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de



(1998) 201.[10] C.H.P. Bruins, C.M. Jeronimus-Stratingh, K. Ensing, W.D. van Dongen, G.J. de

Jong. J. Chromatogr. A 863 (1999) 115.[11] M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, G.J. de Jong. Rapid

Commun. Mass Spectrom. 14 (2000) 2103.[12] J. Ding, U.D. Neue. Rapid Commun. Mass Spectrom. 13 (1999) 2151.


155

[13] K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng. Anal. Chem. 70 (1998)882.



[16] I. Fu, E.J. Woolf, B.K. Matuszewski. J. Pharm. Biomed. Anal. 18 (1998) 347.[17] R. King, R. Bonfiglio, C. Fernandez-Metzler, C. Miller-Stein, T. Olah. J. Am.

Soc. Mass Spectrom. 11 (2000) 942.[18] B. Law, D. Temesi. J. Chromatogr. B 748 (2000) 21.[19] F.M. Lagerwerf, W.D. van Dongen, R.J.J.M. Steenvoorden, M. Honing, J.H.G.

Jonkman. Trends Anal. Chem. 19 (2000) 418.[20] B.K. Choi, D.M. Hercules, A.I. Gusev. J. Chromatogr. A 907 (2001) 337.[21] S.D. Clarke, H.M. Hill, T.A.G. Noctor, D. Thomas. Pharm. Sci. 2 (1996) 203.[22] E. Dijkman, D. Mooibroek, R. Hoogerbrugge, E. Hogendoorn, J.-V. Sancho,

O. Pozo, F. Hernández. J. Chromatogr. A 926 (2001) 113.[23] R. Pascoe, J.P. Foley, A.I. Gusev. Anal. Chem. 73 (2001) 6014.[24] B.K. Choi, A.I. Gusev, D.M. Hercules. Anal. Chem. 71 (1999) 4107.[25] P.L. Ferguson, C.R. Iden, A.E. McElroy, B.J. Brownawell. Anal. Chem. 73

(2001) 3890.[26] K. Heinig, F. Bucheli. J. Chromatogr. B 769 (2002) 9.[27] C. Müller, P. Schäfer, M. Störtzel, S. Vogt, W. Weinmann. J. Chromatogr. B 773

(2002) 47.[28] D.J. Borts, L.D. Bowers. J. Mass Spectrom. 35 (2000) 50.[29] T. Ternes, M. Bonerz, T. Schmidt. J. Chromatogr. A 938 (2001) 175.[30] R. Kitamura, K. Matsuoka, E. Matsushima, Y. Kawaguchi. J. Chromatogr. B 754

(2001) 113.[31] X. Tong, I.E. Ita, J. Wang, J.V. Pivnichny. J. Pharm. Biomed. Anal. 20 (1999)

773.[32] T.K. Majumbar, L.L. Martin, D. Melamed, F.L.S. Tse. J. Pharm. Biomed. Anal.

23 (2000) 745.[33] A.P. Bruins. J. Chromatogr. A 794 (1998) 345.[34] C. Berggren, S. Badyoudh, D. Sherrington, K. Ensing. J. Chromatogr. A 889

(2000) 105.[35] Wissenschaftliche Tabellen Geigy, Teilband Körperflüssigkeiten. CIBA-GEIGY,

Basle, Switzerland, 1977 p51-104.[36] M. Claeys, H. van den Heuvel, S. Chen, P.J. Derrick, F.A. Mellon, K.R. Price.

J. Am. Soc. Mass Sspectrom. 7 (1996) 173.[37] D.C. van Setten, G.J. ten Hove, E.J.H.J. Wiertz, J.P. Kamerling, G. van de

Werken. Anal. Chem. 70 (1998) 4401.[38] P. Waridel, J.-L. Wolfender, K. Ndjoko, K.R. Hobby, H.J. Major,

K. Hostettmann. J. Chromatogr. A 926 (2001) 29.[39] J. Wittig, M. Herderich, E.U. Graefe, M. Veit. J. Chromatogr. B 753 (2001) 237.

SPE-MS2 for the determination of prednisolone in serum

157

4.3

On-line coupling of solid-phase extraction withmass spectrometry for the analysis of biologicalsamples. Determination of prednisolone in serum*

Summary

Solid-phase extraction (SPE) was directly coupled to mass spectrometry (MS)for the rapid determination of prednisolone in serum. A C18 stationary phaseallowed washing of the cartridge with 25% methanol. A rapid increase of thepercentage of methanol (25 to 50% within 0.1 min) was applied during theelution. The high flow-rates during the extraction (5.0 ml/min) combined withMS detection resulted in a total analysis time of 4 min. Some matrixinterference was still observed with a triple-quadrupole MS, even in themultiple reaction monitoring mode. This resulted in a detection limit (LOD) ofabout 10 ng/ml. The matrix interference and the LOD were similar foratmospheric pressure chemical ionisation and atmospheric pressure photoionisation. Applying an ion-trap MS in the MS/MS mode resulted in cleanerchromatograms. Due to extensive fragmentation of prednisolone, the LOD wasnot lower than about 5 ng/ml prednisolone in serum, and a limit of quantitationof about 10 ng/ml (relative standard deviation <15%) was observed.

*: M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, A.P. Bruins, R.A. de Zeeuw, G.J. de Jong.Submitted to J. Chromatogr. B.


158

4.3.1 Introduction

Prednisolone is a glucocorticoid derived from hydrocortisone. Its maintherapeutic application is due to its immunosuppressive effects [1]. For thisreason, prednisolone is often administered prior to, during and after organtransplants in order to decrease the risk of organ rejection. Prednisolone is ahormone-like compound and its side-effects are therefore also hormone-related.A balance must be found between the side effects and the chance of a successfulorgan transplant. Thus, the concentration of prednisolone should be carefullymonitored. Since immediate action should be undertaken upon blood levels thatare too high or too low, the rapid determination of prednisolone in serum atconcentrations down to the low ng/ml level is required. Most methods used forthe determination of prednisolone in serum or plasma apply liquid-liquidextraction [2-5] or off-line solid-phase extraction (SPE) [6,7]. Such techniquesare time-consuming and error-prone steps such as evaporation andreconstitution of the eluate are required. Modern developments in the couplingof liquid chromatography with mass spectrometry (MS) [8,9] have offeredtremendous potential for high-throughput analysis. On-line coupling of SPEwith LC-MS is well established [10-13]. Furthermore, the potential of SPEcoupled directly with MS was also shown [14,15]. For the determination ofclenbuterol in urine, good sensitivity and selectivity were obtained by applyingMS3, whereas the total analysis time was about 8.5 min.

In this study, we investigated the potential of SPE-MSn for the rapidanalysis of prednisolone in serum down to the low ng/ml level. Steroids areeasily fragmented to various fragments simultaneously by thermal degradationas well as by collision-induced dissociation (CID) [16-21], which may result inlimited sensitivity, and the latter aspect should thus be carefully investigated. Toachieve low-ng/ml levels, an ion-trap MS and a triple-quadrupole MS werecompared. The latter type of MS was used applying atmospheric pressurechemical ionisation (APCI) and atmospheric pressure photo ionisation (APPI).A detailed description of the APPI mechanism can be found in literature [22].Basically, a dopant is ionised through photoionisation and reacts with solventmolecules, after which proton transfer to the analytes takes place. APPI mayionise analytes, and in particular hormones, more efficiently than APCI [22-24],thus potentially allowing the determination of lower concentrations ofprednisolone.


159

4.3.2 Experimental


All on-line SPE-MS experiments were performed with a Prospekt samplehandler (Spark, Emmen, The Netherlands) using one six-port valve, thecartridge-switching device, and a solvent delivery unit (SDU). Activation,conditioning, sampling, trapping, and washing were done using the SDU. Theeffluent was connected to waste during these steps. All steps of the SPEprocedure were carried out using a forward-flush mode. A cartridge wasreplaced after single use. A second flow stream from a Series 1100 gradientpump (Hewlett-Packard, Waldbronn, Germany) was used for the elution, whichwas connected to the mass spectrometer via the cartridge. During theoptimisation of the SPE procedure two polymeric (PLRP-S and Resin GP) and aC18 stationary phase (all from Spark-Holland, Emmen, The Netherlands) wereused. After optimisation, HySphere–9 (C18, 10×2 mm, particle size 7 µm)cartridges were applied.

Methanol was of HPLC grade (Lab Scan, Dublin, Ireland). Glacial aceticacid was of analytical-reagent grade (Merck, Darmstadt, Germany). Water wasobtained from an Elgastat Maxima system (Salm and Kipp, Breukelen, TheNetherlands). Aqueous solutions were passed through a 0.45-µm RC 55membrane filter (Schleicher & Schuell, Dassel, Germany) prior to use.Prednisolone (Ph. Eur., Genfarma, Maarssen, The Netherlands) was dissolvedin methanol (1 mg/ml) and stored in the dark at -20ºC. Spiking of samples wasperformed by addition of a small volume of the stock solution that is adequatelydiluted with buffer to foetal calf serum (PAA Laboratories, Linz, Austria).

4.3.2.2 SPE procedure

The final SPE procedure for the C18 cartridges was as follows: activationwas performed with 2.5 ml methanol, and conditioning with 3.75 ml dilutedacetic acid (0.5%, pH about 3) at a flow-rate of 5.0 ml/min. A 500-µl samplewas loaded onto the cartridge with diluted acetic acid (0.75 min at 2.0 ml/min)to minimise possible carry-over, followed by a washing step with 3.75 ml of25:75 methanol:diluted acetic acid (0.5%) at a flow-rate of 5.0 ml/min. Theelution was started with 25:75 methanol:acetic acid (0.5%), and subsequentlywithin 0.1 min the methanol percentage was increased to 50%. This percentagewas maintained for 1.15 min. A flow-rate of 1.0 ml/min was used duringelution. Quantitation was performed by the use of external calibration.


160


An LCQ Classic ion-trap MS (Thermoquest, San Jose, CA, USA)equipped with an APCI source was used. The vaporiser temperature was set at350ºC. The sheath gas and auxiliary gas settings (both nitrogen) were 34 and 3(arbitrary units), respectively. The discharge current was set at 5.00 µA and thecapillary voltage was 15.00 V. The temperature of the heated capillary was170ºC, and the tube lens offset was set at 40.00 V. All scans were recorded inthe full-scan mode with 3 microscans over the range of m/z 295 to 370 using thepositive-ion mode. The maximum injection time was set at 300 ms. Helium wasapplied as cooling gas and collision gas. Extracted ion chromatograms in all MSmodes were obtained for [M+H]+ (m/z 361) or fragment ions ± 0.5 Th. Theisolation width during MSn experiments was 2.0 Th. The collision energyapplied during MS/MS experiments was 20%.

An API3000 triple-quadrupole MS (MDS-Sciex, Concord, Ontario,Canada) was used with both an APCI and an APPI [22] source. The settingsused during single-MS analysis and multiple-reaction monitoring (MRM) arepresented in Table 1. In the single-MS mode a Q1-scan was performed fromm/z 100 to 400. In the MRM mode m/z 361.1 was fragmented and the productsat m/z 307.0 and 325.0 were monitored. When applying APCI, nitrogen wasused as curtain gas and auxiliary gas, and zero air was used as nebulising gas.During APPI experiments, only nitrogen was used. The lamp protection gas wasset at about 1 l/min, and a lamp current of 0.75 mA was used. Toluene was usedas dopant, which was added to the auxiliary gas line via a T-piece. A flow-rateof 50 µl/min was used.

Table 1: Settings of the triple-quadrupole MS during Q1-scan and MRM experiments.

Nebulizer gas (arb) 15 Q0 (V) -10.00Curtain gas (arb) 10 IQ1 (V) -11.00CAD (arb) 0 (4a) ST (V) -16.00Needle current (µA) 2.00b RO1 (V) -11.0Temperature (°C) 450 IQ2 (V) -18.0Orifice (V) 20 RO2 (V) -100 (-25.0a)Ring (V) 50 ST3 (V) -120 (-45.0a)CEM 2500 RO3 (V) -102 (-27.0a)

DF (V) -100a: Numbers between brackets are settings for MRM experimentsb: For APPI experiments – source offset voltage = 2000 V.


161


4.3.3.1 Optimisation of the SPE procedure

When applying SPE, steroid- or hormone-like compounds are commonlyextracted from the sample by use of an apolar extraction phase [25-27].Therefore, the apolar C18 phase and two polymeric phases (PLRP-S andResin GP) were investigated for the extraction of prednisolone. The sample wasloaded onto the cartridge after which the stationary phase was washed withbuffer at pH 8.5, 7 or 3. Subsequently, a 10-min gradient from 0 to 100%methanol, buffered at the same pH, was used for the elution. Diode-arraydetection was applied during the optimisation of the SPE procedure. The peakshape and the position for the analyte with regard to the matrix were the criteriafor the selection of an appropriate stationary phase. With the polymeric phases,hardly any separation between the analyte and the matrix could be obtained.

With the C18 stationary phase, no separation was observed at pH 8.5.Decreasing the pH to 7 or 3 did not move the prednisolone peak due to the factthat prednisolone is a neutral species. However, the retention of the co-extractedmatrix components was increased, resulting in more distinction betweenprednisolone and the matrix compounds. With pH 3, the best results wereobtained, since most of the matrix compounds eluted at more than 50%methanol. Some matrix compounds eluted in front of prednisolone and someco-elution was observed. It was possible to wash the cartridge with 25:75methanol:diluted acetic acid (0.5%) at a flow-rate of 5.0 ml/min withoutbreakthrough of the analyte from the extraction phase. This ensured that theearly eluting matrix compounds, i.e. polar components, were removed and a100% recovery was obtained for prednisolone.

Table 2: Optimised SPE procedure for C18 cartridges; injected sample volume is 500 µl.

Flow-rate(ml/min)

Volume(ml)

Time(min)

Activation 5.0 2.5 0:00 – 0:50Conditioning 5.0 3.75 0:50 – 1:25Sampling 2.0 1.5 1:25 – 2:00Washing 5.0 3.75 2:00 – 2:75Elution 1.0 1.25 2:75 – 4:00

Subsequently, the elution was performed starting at 25% methanol andincreasing this to 50% within 0.1 min. No further increase in the percentage ofmethanol was used so that most of the co-extracted and apolar matrixcomponents that were not removed during the wash step were retained by thestationary phase, and could therefore not interfere with the MS detection. After


162

each analysis the cartridge was discarded. The final extraction procedure ispresented in Table 2. The total analysis time was about 4 min.

4.3.3.2 SPE-MS system

Use of ion-trap MSAn ion-trap MS was applied for the SPE-MS system. The vaporiser

temperature was set at 350°C, which gave adequate evaporation and no memoryeffect was observed. However, extensive fragmentation was observed due to thethermolability of the analyte and/or the easy CID of prednisolone. Thefragmentation pattern and a mass spectrum in the MS mode are shown inFigs. 1A and B. The fragmentation of prednisolone implied that the parent ion

Fig. 1: (A) Structure and fragmentation pathway of prednisolone in the ion-trap MS;(B) mass spectrum (MS mode); (C) mass spectrum (MS/MS mode,fragmentation of m/z 361).


163

m/z 361 was only about 54% of the total abundance. With MS/MS experiments,no formation of the fragments with m/z 329 and 301 was observed, suggestingthat these ions were indeed formed by thermal degradation. Three fragments,i.e., m/z 343, 325 and 307, were formed after CID of the parent ion [M+H]+

(Fig. 1C). Fragment m/z 343 gave the highest signal, but the summation of theextracted ions of m/z 307 and 325 resulted in the best signal-to-noise (S/N)ratio, even though only about 15% of the signal of the product ion wasconverted into these fragments.

The use of the ion-trap MS in the single-MS mode (Fig. 2A) showedsevere matrix interferences and a rather high limit of detection (LOD; 50 ng/ml;three times the blank level) was obtained. The application of MS/MS resulted inclean chromatograms after extraction of blank serum (Fig. 2B) and an improvedLOD (5 ng/ml; determined as three times the level of spikes in thechromatogram) was observed (Table 3). A good reproducibility and linearitywere obtained. A representative chromatogram of spiked serum is depicted inFig. 2C. Good linearity and reproducibilities were observed over theinvestigated concentration range. The limit of quantitation was about 10 ng/ml(relative standard deviation <15%). Comparing the signals after extraction frombuffer and from serum showed higher signals for the latter. At 10 ng/ml, thesignal after extraction from serum was about 2 times as high as the signal afterextraction from buffer. This decreased to a factor of about 1.5 at 20 ng/ml andwas about 1.2 at higher concentrations (30-550 ng/ml). A post-cartridgecontinuous infusion of the analyte while extracting blank serum [28] showed asimilar ion enhancement effect.

Table 3: Analytical data of the SPE-MS2 systems for the determination of prednisolonein serum using an ion-trap MS (MS/MS mode) and a triple-quadrupole MS(MRM mode).

Triple-quadrupole MSIon-trap MSAPCI APCI APPI

LOD (ng/ml) 5 10 10Linearity* (R) 0.9944 0.9962 0.9959Range (ng/ml) 10-550# 10-275# 10-275#

RSD (%)± 8.8 9.9 10.2*: weighted regression (1/x).#: maximum concentration investigated.±: at 20 ng/ml, n=6.


164

Fig. 2: (A) SPE-MS with an ion-trap MS using the APCI interface for extraction ofblank serum (total ion count, m/z 295-370); (B) SPE-MS/MS of blank serumand summation of extracted ions m/z 307 and 325, (C) SPE-MS/MS of serumspiked with 10 ng/ml prednisolone.


165

Use of triple-quadrupole MSAnother type of MS, a triple-quadrupole instrument, was used with an

APCI and APPI source. The vaporiser temperature was set at 450°C foradequate evaporation of the eluate. With lower temperatures, memory effectswere observed. The high temperature caused severe fragmentation ofprednisolone due to thermal degradation. Similar chromatograms as with theion-trap MS were obtained in the single-MS mode (Q1 scan) upon analysis ofserum samples (Fig. 3A). However, applying MS/MS (MRM mode),monitoring the same fragments as with the ion-trap MS (that is m/z 325 and307), did not completely eliminate the matrix interference (Fig. 3B). The LODwas now about 10 ng/ml (Table 3). A representative chromatogram is depictedin Fig. 3C. With APPI, similar results were observed as with APCI with respectto sensitivity, linearity and reproducibility (Table 3). The matrix interferencewas at the same level and no improved sensitivity was observed. An increase insensitivity can probably only be obtained if electronic noise is limiting thesensitivity. In case of chemical noise, i.e. background caused by matrixcompounds, such an increase is less likely to be obtained. This was confirmedby comparing the LODs of APCI and APPI in the MRM mode for theextraction of prednisolone from buffer, in which no chemical noise interferedwith the determination. Then, about a factor of five more sensitivity wasobserved with APPI than with APCI. These results were in accordance with theresults of other studies [22,23].

The determination of the LOD (three times the blank level) afterextraction from serum and detection with the triple-quadrupole MS was ratherambiguous due to the presence of the matrix. In this study, constant signalswere obtained with the blank serum. However, with real-life samples, often noblank sample is available, thus making detection at such low levels moreunpredictable. The presence of matrix interference from blank serum may bedue to similar fragmentation pathways of co-extracted compounds, since manyendogenous compounds such as steroid hormones have similar fragmentationpatterns [16,17]. The loss of H2O is rather easy to establish from anyhydroxylated compound. With hormone-like compounds, the loss of two ormore H2O molecules is very common [16,20].

The difference in matrix interference between the ion-trap MS (MS/MSmode) and the triple-quadrupole MS (MRM mode) is probably due to thedifference of the MS/MS principles. In the ion-trap MS, only the precursor ionis accelerated sufficiently to be fragmented. Fragments are readily stabilised inthe centre of the trap, hereby decreasing the potential for further fragmentation.In the triple-quadrupole MS, more fragmentation is commonly observed[17,18], since all ions are accelerated. This implies that an ion that is alreadyfragmented will still be accelerated towards the end of the second and the thirdquadrupole. Thus, a formed fragment (product ion) may thereby collide and


166

fragment further in the second and the third quadrupole. This has theconsequence that not only the analyte is more easily fragmented (as observedduring MS/MS experiments), but also the matrix components. In atriple-quadrupole MS, the higher level of interference can thus be due toproduct ions (ions should already be formed in the source or in the firstquadrupole to m/z 361) and their consecutive fragments, which is less commonin an ion-trap MS.

Fig. 3: (A) SPE-MS with a triple-quadrupole MS using the APCI interface forextraction of blank serum (Q1 scan m/z 100-400); (B) SPE-MS/MS of blankserum (MRM mode, transition of m/z 361 to m/z 325 and 307);(C) SPE-MS/MS of serum spiked with 10 ng/ml prednisolone (offset 500 cps).


167

4.3.4 Conclusions

The separation of prednisolone from endogenous compounds present inserum was difficult to obtain, and some matrix interference was still observed inthe MRM mode. The present SPE-MS/MS system applying the ion-trap MSshowed an increase in selectivity with regard to the triple-quadrupole MS. TheLOD is in the low-ng/ml range, but is still rather high due to the thermaldegradation and easy CID of prednisolone. The total analysis time was about4 min, which was due to high flow-rates (up to 5 ml/min, except for the elution)during the SPE procedure. The use of even higher flow-rates (up to 10 ml/min),as well as the application of two cartridges in parallel may further enhance thethroughput.

In general, SPE-MSn has shown good potential for high-throughputbioanalysis. The development of a rapid SPE-MS system implies rapidextraction, while obtaining good selectivity at the front of the analysis (SPE) aswell as at the end (MS). One should always carefully monitor the effects ofmatrix, and in particular those compounds that co-elute with the analyte ofinterest. The use of an internal standard may help to improve the reproducibilitywhen required.

Acknowledgements

The authors are very grateful to Spark (Emmen, The Netherlands) forproviding a Prospekt system. This research was supported by the TechnologyFoundation STW, applied science division of NWO and the technologyprogramme of the Ministry of Economic Affairs.

4.3.5 References

[1] T.M. Brody, J. Larner, K.P. Minneman. Human Pharmacology. Molecular toClinical, Mosby-Year Book, St. Louis, MI, USA, 1998, 626.

[2] F.J. Frey, B.M. Frey, L.Z. Benet. Clin. Chem. 25 (1979) 1944.[3] M.H. Cheng, W.Y. Huand, A.I. Lipsey. Clin. Chem. 34 (1988) 1897.[4] W.J. Jusko, N.A. Pyszczynski, M.S. Bushway, R. D’Ambrosio, S.M. Mis.

J. Chromatogr. 658 (1994) 47.[5] S.A. Döppenschmitt, B. Scheidel, F. Harrison, J.P. Surmann. J. Chromatogr. 674

(1995) 237.[6] H. Hirata, T. Kasama, Y. Sawai, R.R. Fike. J. Chromatogr. 658 (1994) 55.[7] N. Shibata, T. Hayakawa, K. Takada, N. Hoshino, T. Minouchi, A. Yamaji.

J. Chromatogr. B 706 (1998) 191.[8] A.P. Bruins, T.R. Covey, J.D. Henion. Anal. Chem. 59 (1987) 2642.[9] T. Yasuda, M. Tanaka, K. Iba. J. Mass Spectrom. 31 (1996) 879.


168

[10] N.G. Knebel, S. Grieb, S. Leipenheimer, M. Locher. J. Chromatogr. B 748(2000) 97.

[11] R.A.M. van der Hoeven, A.J.P. Hofte, M. Frenay, H. Irth, U.R. Tjaden, J. van derGreef, A. Rudolphi, K.-S. Boos, G. Marko Varga, L.E. Edholm. J. Chromatogr.A 762 (1997) 193.


[13] E.H.R. van der Wal, E.R. Brouwer, H. Lingeman, U.A.Th. Brinkman.Chromatographia 39 (1994) 239.

[14] M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, G.J. de Jong. RapidCommun. Mass Spectrom. 14 (2000) 2103.

[15] G.P. Bowers, S.C. Clegg, S.C. Hughes, A.J. Harker, S. Lambert. LC•GC 15(1997) 48.

[16] M. Mulholland, T.J. Whelan, H. Rose, J. Keegan. J. Chromatogr. A 870 (2000)135.

[17] M. Fiori, E. Pierdominici, F. Longo, G. Brambilla. J. Chromatogr. A 807 (1998)219.

[18] J.-P. Antignac, B. le Bizec, F. Monteau, F. Poulain, F. André. Rapid Commun.Mass Spectrom. 14 (2000) 33.

[19] M. van de Wiele, K. de Wasch, J. Vercammen, D. Courtheyn, H. de Brabander,S. Impens. J. Chromatogr. A 904 (2000) 203.

[20] P.W. Tang, W.C. Law, T.S.M. Wan. J. Chromatogr. B 754 (2001) 229.[21] K. Fluri, L. Rivier, A. Dienes-Nagy, C. You, A. Maître, C. Schweizer, M. Saugy,

P. Mangin. J. Chromatogr. A 926 (2001) 87.[22] D.B. Robb, T.R. Covey, A.P. Bruins. Anal. Chem. 72 (2000) 3653.[23] D.B. Robb, A.P. Bruins, T.R. Covey. Atmospheric pressure photoionization

(APPI): a new ionization technique for LC-MS. Proceedings of 48th Conferenceof Mass Spectrometry and Allied Topics, 2000, Long Beach, CA, USA.

[24] D.B. Robb, A.P. Bruins, H.A.M. Peters, P.L. Jacobs. Atmospheric pressurephotoionization (APPI) for high sensitivity LC-MS in bioanalysis. Proceedings of48th Conference of Mass Spectrometry and Allied Topics, 2000, Long Beach,CA, USA.

[25] R. Draisci, L. Palleschi, E. Ferreti, L. Lucentini, P. Cammarata. J. Chromatogr. A870 (2000) 511.

[26] X.-Y. Xiao, D.V. McCalley, J. McEvoy. J. Chromatogr. A 923 (2001) 195.[27] S. Noé, J. Böhler, E. Keller, A.W. Frahm. J. Pharm. Biomed. Anal. 18 (1996)

471.[28] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle. Rapid Commun. Mass Spectrom.

19 (1999) 1175..

55COUPLING OF

SOLID-PHASE MICROEXTRACTIONAND MASS SPECTROMETRY

The pieces have been put

Non-equilibrium SPME coupled directly to MS2 for rapid bioanalysis

171

5.1

Non-equilibrium solid-phase microextractioncoupled directly to ion-trap mass spectrometry forrapid analysis of biological samples*

Summary

To determine sub-ppb levels of drugs in biological samples selective, sensitiveand rapid analytical techniques are required. The present work shows thepossibilities for high-throughput analysis of solid-phase microextraction(SPME) directly coupled to an ion-trap mass spectrometer equipped with anatmospheric pressure chemical ionisation source. As no chromatographicseparation is performed, the SPME procedure is the time-limiting step.Direct-immersion SPME under non-equilibrium conditions permits thedetermination of lidocaine in urine within 10 min. After a 5-min sorption timewith a 100-µm polydimethylsiloxane-coated fiber, the extraction yield oflidocaine from urine is about 7%. When applying 4 min desorption, using amixture of ammonium acetate buffer (pH 4.5) and acetonitrile (85+15 v/v),about 10% of the analyte is retained on the fiber. An extra cleaning step of thefiber is therefore used to prevent carry-over. By use of tandem MS, no matrixinterference is observed. The detection limit for lidocaine is about 0.4 ng/ml,and intraday repeatability and interday reproducibility are within 14% over aconcentration range of 2-45 ng/ml.

*: M.W.J. van Hout, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de Jong. Analyst 127 (2002) 355-359.

Chapter 5 – Coupling of solid-phase microextraction and mass spectrometry

172

5.1.1 Introduction

High-throughput analysis is becoming increasingly important owing tothe large numbers of samples that have to be analysed. However, throughputshould not negatively affect the sensitivity and selectivity of the systems. As theconcentrations of interest are often in the (sub) ppb range, the systems mustallow fast, selective and sensitive analysis of samples. Most systems that offerthese possibilities combine sample preparation on-line with a separationtechnique and mass spectrometry. A commonly used coupled technique issolid-phase extraction (SPE) combined with liquid chromatography (LC) andmass spectrometry (MS) 1,2]. Sensitivity is obtained by the mass spectrometerand the large sample volumes in the on-line SPE procedure. The SPE proceduremight take too much time, but this can be counteracted by the use of highflow-rates and small volumes during the various steps of the procedure.Moreover, using a system that offers the possibility to perform dual SPE [3,4]may also increase the throughput. The second limiting factor is the time neededfor the separation step, which may be influenced by using short LC columns[1,5,6] or even by eliminating the separation step, i.e. direct coupling of SPE toMS [3,7-11].

Solid-phase microextraction (SPME), based on extraction by a coatedfiber, was originally designed to be combined with gas chromatography, but canalso be combined with LC. Direct immersion SPME is carried out, followed bydesorption in a desorption chamber filled with a suitable solvent to remove theanalytes from the fiber. Until now, SPME has hardly ever been used inhigh-throughput systems owing to the seemingly disadvantageouscharacteristics of SPME for such systems. Despite the various factors thatpositively influence the sorption [12-16], it is usually time consuming becauseof the slow equilibrium process. The maximum extraction yield is achievedafter equilibrium is reached between the sample and the fiber [15,17,18].Although desorption can be performed faster, this step is also an equilibriumprocess. Other limiting factors of SPME are the relatively low extraction yieldsand possible carry-over [12]. Furthermore, not as many stationary phases as forSPE are available yet, thereby limiting the choice for selectivity. However,SPME has some clear advantages [12] such as easy handling, little use ofsolvents, and little requirements regarding instruments.

Some reports [13-19] have been published concerning the use ofnon-equilibrium sorption conditions aiming for reduced analysis times.Reproducible GC analysis under non-equilibrium conditions using eitherheadspace or direct-immersion SPME has proved to be possible, provided thatthe agitation conditions and sorption time were held constant [19]. An increasein the throughput may also be achieved by direct coupling of SPME with MS,for example by using thermal desorption and direct introduction, across a short


173

GC transfer line, into the mass spectrometer [20,21]. This particular set-up wasapplied to determine volatiles of cheese with headspace SPME and MS analysis[20]. In another study the desorption profiles of aromatic hydrocarbons wereinvestigated [21].

No reports are available concerning the direct coupling of SPME with MSusing an LC interface for relatively involatile compounds. The goal of thepresent work was to develop such a system that can meet the requirements ofhigh-throughput analysis. The potential of direct coupling of SPME with anion-trap mass spectrometer, equipped with an LC interface, was investigated.The required selectivity can be obtained by using multiple-stage MS (MSn withn≥2). Since the sorption and desorption processes are key factors in thethroughput, these parameters were studied and optimised for a rapid analysis ofa model substance (lidocaine) in urine.

5.1.2 Experimental

5.1.2.1 Chemicals and Instrumentation

Polydimethylsiloxane (PDMS) coated (7 and 100 µm) SPME fibers wereobtained from Supelco (Bellefonte, PA, USA). The SPME holder and fiberwere compatible with the SPME-LC desorption chamber (Supelco) [17]. AHewlett-Packard (Waldbronn, Germany) HPLC gradient pump series 1100 wasused for desorption and transport of the solvent to the mass spectrometer. Themass spectrometer was an LCQ Classic ion-trap (Thermoquest, San Jose, CA,USA) and it was equipped with an atmospheric pressure chemical ionisation(APCI) or electrospray ionisation (ESI) source. Ultrapure water was obtainedusing an Elga Maxima ultrapure water purification system (Salm & Kipp,Breukelen, The Netherlands).

Acetonitrile (Lab-Scan, Dublin, Ireland) was of HPLC quality. Aceticacid, boric acid, and sodium chloride (all of analytical-reagent grade) wereobtained from Merck (Darmstadt, Germany) and ammonium acetate (99.99%)was from Sigma-Aldrich (St. Louis, MO, USA). Stock standard solutions oflidocaine hydrochloride (USP, Holland Pharmaceutical Supply, Alphen a/dRijn, The Netherlands) were prepared in acetonitrile. Ammonium acetate buffer(10 mM) was prepared by dissolving ammonium acetate in ultrapure water andadjusting the pH to 4.5 with acetic acid (10% v/v). Buffer solutions pH 10(0.2 M) were prepared by dissolving boric acid in ultrapure water and adjustingthe pH with 1 M sodium hydroxide. Triplicate injections (50 µl) of controlsamples [24.81 ng/ml lidocaine in buffer (pH 4.5)-acetonitrile (85+15 v/v)]were used to correct for day-to-day variations in the sensitivity of the mass


174

spectrometer. Analyses performed to investigate the reproducibility weresubjected to two-way analysis of variance (ANOVA) with replications.

5.1.2.2 SPME procedure

A previously optimised SPME procedure for the determination oflidocaine in urine [15] was used as a starting point for the present work. Minormodifications to the system included the change from phosphate buffer(25 mM, pH 4.0) to ammonium acetate buffer (10 mM, pH 4.5) to make thesystem compatible with MS. Furthermore, the pH of the extraction buffer wasincreased from 9.5 to 10.

Samples were prepared by diluting calf urine with 0.2 M borate buffer(pH 10) in a ratio of 1:1. Sodium chloride was added to the urine-buffer solutionto give a final salt concentration of 0.3 g/ml. The pH was readjusted to pH 10with 10% sodium hydroxide. The coated part of the fiber was completelyimmersed into 1.25 ml of sample and subsequent extraction was performedwhile continuously stirring the sample using a 7×2 mm stirring rod and amagnetic stirrer (Metrohm, Herisau, Switzerland). Desorption was performedby placing the fiber into the desorption chamber (Supelco), which was filledwith 70 µl ammonium acetate buffer (10 mM, pH 4.5)-acetonitrile (85+15 v/v).After a certain time of (static) desorption the chamber was flushed to the MSwith the same solvent at a flow-rate of 0.5 ml/min.


Using the ion-trap mass spectrometer with an APCI source, the vaporisertemperature was set at 400ºC. During analysis with the APCI source the sheathgas and auxiliary gas (both nitrogen) were 48 and 3 (arbitrary units),respectively, and the corona discharge current was set at 5.0 µA. If ESI wasused the sheath gas and auxiliary gas were 90 and 5 (arbitrary units),respectively, and a spray voltage of 5.0 kV was used. With both sources theheated capillary was set at 140ºC and a capillary voltage of 46.0 V was applied.The tube lens offset was 45.0 V. The first octapole offset, the inter-octapoleoffset, and the second octapole offset were -2.0 V, -26.0 V, and -6.0 V,respectively. All scans were recorded in the full-scan mode with threemicroscans over the range m/z 75-300, using the positive-ion mode. Themaximum injection time was set at 300 ms. Helium was applied as cooling andcollision gas. Extracted ion chromatograms in single-MS mode were obtainedfor [M+H]+ (m/z 235 ± 0.5). For MS/MS the fragment ion m/z 86 ± 0.5 wasmonitored. The isolation width during MS/MS experiments was 2.0 Th, using acollision energy of 34%.


175


5.1.3.1 Setting up the SPME-MS system

A fast method, based on direct-immersion SPME-MS, was obtained byconnecting the desorption chamber to the LC-MS interface. After staticdesorption a pump was used to flush the contents of the desorption chamber tothe MS. Sorption was done at pH 10, which ensured that lidocaine is in itsneutral form, thereby increasing the yield [15]. A further increase of the yieldwas obtained by the addition of 0.3 g/ml NaCl. Equilibrium was more rapidlyreached by stirring of the sample [15,16]. The reproducibility of the extractionwas increased by dilution of urine with buffer (pH 10) and addition of saltalmost to saturation. An existing method [15] for lidocaine in urine involvingSPME-LC and diode-array detection used phosphate buffer (pH 4) duringdesorption (2×10 min). Since this buffer is not compatible with MS, a morevolatile ammonium acetate buffer at a similar pH was used. Desorption atpH 4.5 allowed lidocaine to become redissolved as the protonated base, sincethe pKa value of lidocaine is 7.9. To enhance the desorption of lidocaine fromthe fiber, 15% acetonitrile was added to the desorption solvent [15]. Moreacetonitrile can be advantageous for analysis by MS, since the desolvatationwill be more efficient [23,24]. However, the amount of acetonitrile is limited,since too much acetonitrile implies the possibility of stripping the fiber uponwithdrawal from the desorption chamber.

Both APCI and ESI are suitable for the analysis of lidocaine. ESI behavesas a concentration-sensitive detector, thus lowering the flow-rate will lead tolarger peak areas. Decreasing the flow-rate from 0.7 ml/min to 0.1 ml/minresulted in larger peak areas. However, as expected, no increase in peak heightswas observed, and there was no gain in sensitivity (signal-to-noise ratio, S/N).Moreover, the peaks were very broad at low flow-rates. Use of APCI provided afactor 4-5 better sensitivity (S/N) than ESI. Therefore, further experiments wereperformed with APCI. The minimum flow-rate for this type of source is0.5 ml/min. Although narrower peaks were obtained at higher flow-rates, theexpected increase in sensitivity (S/N) was not observed, as the noise increasedproportionally with peak height. Therefore, a flow-rate of 0.5 ml/min wasmaintained to minimise contamination of the ion source owing to involatilesubstances in the eluate [11].

5.1.3.2 Selectivity of the system

As no chromatographic separation was performed, the required selectivityof the system had to be obtained by the extraction procedure and by the massspectrometer. With respect to the extraction, the advantages and disadvantages


176

of SPE and SPME must be taken in consideration. SPE is an exhaustiveextraction, whereas SPME has non-exhaustive properties. Despite the fact thatSPE has the advantage that it allows some chromatographic separation ofcompounds, severe matrix interference can be obtained with an SPE-MS system[11]. With SPE the aim is to eliminate breakthrough, thus the sorbent must havea high capacity. With SPME breakthrough is not an issue, thus allowing a betterchoice for selectivity of the sorbent. In SPME the amount extracted by the fibercoating is determined by the ratio of sample volume and coating volume owingto the non-exhaustive character of SPME. The sample volume (≥1 ml) is usuallymuch larger than the coating (0.628 µl for the 100-µm PDMS coating), so largepartition constants (between matrix and sorbent) are required for extraction[14,15]. In exhaustive extractions, i.e. SPE, the sample is forced through thesorbent, and thus the ratio of void volume and sorbent volume is decisive. Sincethese volumes are usually similar, only small partition constants are required fortrapping of compounds [15]. Hence, matrix compounds will also be trappedmore easily with SPE, thus decreasing the selectivity. Even more, especiallyunder non-equilibrium sorption conditions in SPME, diffusion plays a moreprominent role in the sorption process than with SPE. The analyte of interest,which is usually a small molecule, has a diffusion coefficient towards PDMSthat is commonly higher than the diffusion coefficient of the larger matrixcomponents [24]. This may also increase the selectivity of the extractionmethod.

However, the PDMS-coated fiber did not show sufficient selectivity forthe analysis of urine. Applying 45 min sorption and 10 min desorption withsingle-MS detection resulted in a significantly interfering peak (Fig. 1A) atm/z 235, which is also the m/z value of [M+H]+ of lidocaine. Despite the matrixinterference no ion suppression was observed after extraction of blank urine,using continuous infusion of analyte into the desorption solvent in front of theMS interface [25,26]. The absence of ion suppression indicates that SPME mayindeed be more selective than SPE, which resulted in significant ionsuppression if directly coupled to MS for urine analysis [11]. Other matrixinterferences could be avoided by performing MS/MS. If MS/MS is applied,protonated lidocaine will be fragmented to the product ion at m/z 86 (Fig. 2)with a fragmentation efficiency of about 61%. The chromatogram of blankurine analysed with SPME-MS/MS showed no interference of the urine matrix(Fig. 1B).


177

Fig. 1: Extracted ion chromatograms of blank urine samples using SPME (45 minsorption, 10 min desorption) and analysed (A) in the MS mode and (B) in theMS/MS mode.

Fig. 2: Fragmentation pathway of lidocaine.


178

5.1.3.3 Sorption and desorption conditions

A characteristic feature of SPME is the time-sorption curve. Such curvesshow the equilibrium time and also the maximum yield under the experimentalconditions [27]. Since this work combined SPME with the sensitivity of the MSlow yields may not be critical. A 7-µm and a 100-µm PDMS coating were usedin order to investigate the possibilities of an SPME-MS system for thedetermination of the model substance lidocaine. The data points were fittedaccording to time-sorption equations described by Ai [19]. The equilibriumtime, i.e. the time required to reach 95% of the maximum extraction yield, wasabout 1.1 min for the 7-µm coating (Fig. 3A), whereas the 100-µm coating hadan equilibrium time of about 31 min (Fig. 3B). The maximum extraction yieldfor the 7-µm coated fiber (0.15%) is significantly less than predicted [13,14](0.7%) based on the difference in volume (i.e. capacity) of the stationary phaseand the yield obtained with the 100-µm coated fiber (19%). This may beexplained by the fact that the 7-µm coating is highly cross-linked PDMS,whereas the 100-µm coating is only partially cross-linked [28]. As a result,though both fibers are chemically similar (i.e., build up from the same monomerand cross-linker), the physico-chemical properties governing extractionthermodynamics may differ considerably.

Fig. 3: Time-sorption curves using (A) a 7-µm and (B) a 100-µm PDMS-coated fiber.Urine samples (diluted 1+1 with buffer) spiked with 50 ng and 10 ng lidocainein (A) and (B), respectively.

Applying equilibrium extraction, the sensitivity of the SPME-MS systemwill be higher using the 100-µm coating, but the analysis time will be muchshorter with the 7-µm coating. However, the yield after 1 min sorption with the100-µm coating (about 1.7%) is still higher than the yield of the 7-µm coatingunder equilibrium conditions. Therefore, further experiments were done usingthe 100-µm coated fiber. Nevertheless, the 7-µm coated fiber may still be usefulif the reproducibility of the system is more important and the sensitivity is notthe main issue.


179

Desorption is also an equilibrium process. Changes in pH and solventcomposition are used to desorb the analytes rapidly from the fiber. After asorption time of 5 min at pH 10, which resulted in a yield of 7.2%, staticdesorption was performed by applying a pH change and adding an organicmodifier to the desorption solvent [85% buffer (pH 4.5)-15% acetonitrile].Using a desorption time of 4 min, about 10% of the analyte was still retained bythe fiber. Therefore, after 4 min desorption, an extra 4 min clean-up step with afresh portion of the desorption solvent was required, to remove the remaininglidocaine form the fiber prior to using it for the next analysis. The solvent fromthe first desorption was introduced into the mass spectrometer with an analysistime of about 1 min.

5.1.3.4 Analytical data

The present SPME-MS/MS system was developed to obtain a shortanalysis time per sample by use of non-equilibrium sorption conditions. Therelatively low yield was compensated for by the sensitivity of the MS. The limitof detection (LOD, S/N = 3) of this set-up, using a 100-µm PDMS-coated fiberand a 5 min sorption time, was about 0.4 ng/ml lidocaine in urine. Arepresentative chromatogram is shown in Fig. 4.

Fig. 4: Extracted ion chromatograms at m/z 86 using SPME-MS/MS (5 min sorption,4 min desorption) for the determination of 0.4 ng/ml lidocaine in urine.

Since the urine volume was 0.625 ml, diluted to 1.25 ml with buffer, a totalyield of 6.5% (sorption yield of 7.2% and desorption efficiency of 90%) wasobtained. This corresponds to an absolute amount of about 16 pg (or 70 fmol)lidocaine. A limit of quantitation (RSD = 15%) of about 2 ng/ml was obtained.


180

Good linearity (correlation coefficient (R) = 0.9968, weighted regression 1/x,n=8) was observed over the concentration range 0.4-80 ng/ml.

The reproducibility of the non-equilibrium system was investigated atthree concentration levels (Table 1), using the control samples to correct forday-to-day variation in the sensitivity of the mass spectrometer. The intraday(repeatability) and interday RSD (reproducibility) were always <14%, even forthe lowest concentration. At the highest concentration, the interday RSD wasabout 5%. It should be noted that the interday RSD of the system underequilibrium SPME extraction conditions is 4.8% for a lidocaine concentrationof 6.2 ng/ml. It can therefore be concluded that the RSD is slightly increasedowing to non-equilibrium sorption conditions. This increase in RSD withnon-equilibrium SPME can be expected as the sorption is carried out at thesteep part of the time-sorption curve. Slight variations in experimentalconditions may thus have a larger impact on the reproducibility and the amountof analyte being trapped in the PDMS coating. Nonetheless, the reproducibilitywith non-equilibrium SPME is still very acceptable.

Table 1: Repeatability and reproducibility [relative standard deviation (RSD)] of theSPME-MS/MS system for the determination of lidocaine in urine using a5-min sorption and a 4-min desorption time (n=6)

Concentration(ng/ml)

Intraday RSD(%)

Interday RSD(%)

2.3 13.7 12.88.7 3.9 12.3

43.7 2.5 5.7

5.1.4 Conclusions

The developed method shows the potential for high-throughput analysis,using SPME directly coupled to MS. The present work was carried out with anon-selective coating on the fiber, but MS/MS provided clean chromatograms.The speed of analysis of an SPME-MS/MS system is determined by the SPMEprocedure. However, using non-equilibrium SPME drastically reduced theanalysis time to about 10 min per sample. For lidocaine as a model substance inurine, an LOD of 0.4 ng/ml can be obtained with 5 min sorption and 4 mindesorption. A simple clean-up step of the fiber was applied to eliminatecarry-over. The reproducibility of the non-equilibrium SPME system wassatisfactory over the entire concentration range investigated. Although internalstandards are often used to improve reproducibility, it should be noted thatnon-equilibrium SPME requires standards with diffusion coefficients that areclosely comparable to that of the analyte [13]. It may be interesting to perform


181

non-equilibrium SPME using a deuterated internal standard, thereby alsoensuring that the time-sorption curves of the analyte of interest and the internalstandard are equal.

The use of elevated temperatures during sorption and desorption is apossibility to enhance the sensitivity while maintaining a short extraction time.This also allows further reduction of sorption and desorption times as will bedemonstrated in a subsequent paper. An understanding of the fundamentalprocesses at the fiber-sample boundary and inside the fiber might lead to bettercontrol of the processes, so improving the development of a reproducible,sensitive, and fast SPME procedure. The effects of coating thickness,temperature during sorption and desorption and limiting factors in the sorptionkinetics are currently being investigated in our laboratory. The throughput canfurthermore be enhanced by the use of multiple fibers simultaneously forsorption, desorption and cleaning of the fiber. Since the mass spectrometer willbe the cost-limiting part of an SPME-MS system, the set-up can easily beexpanded by using multiple fibers and desorption chambers. Automation ofSPME-MS can be achieved by use of robotics.

Acknowledgements

Jan Brands from Sigma-Aldrich is gratefully acknowledged for supplyingthe SPME fibers. This research was supported by the Technology FoundationSTW, applied science division of NWO and the technology programme of theMinistry of Economic Affairs.

5.1.5 References

[1] D.A. McLoughlin, T.V. Olah, J.D. Gilbert. J. Pharm. Biomed. Anal 15 (1997)1893.

[2] M. Jemal, D. Teitz, Z. Ouyang, S. Khan. J. Chromatogr. B 732 (1999) 501.[3] A. Schellen, B. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de

Lagemaat, E. Verheij. Rapid Commun. Mass Spectrom. 14 (2000) 230.[4] A.C. Hogenboom, M.P. Hofman, D.A. Jolly, W.M.A. Niessen, U.A. Th.

Brinkman. J. Chromatogr. A 885 (2000) 377.[5] M.L. Constanzer, C.M. Chavez, B.K. Matuszewski, J. Carlin, D. Graham.


J.H.G. Jonkman, Ph. Ladure, C. Puozzo. J. Chromatogr. B 708 (1998) 113.[7] A.C. Hogenboom, P. Speksnijder, R.J. Vreeken, W.M.A. Niessen,

U.A.Th. Brinkman. J. Chromatogr. A 77 (1997) 81.[8] A.C. Hogenboom, W.M.A. Niessen, U.A.Th. Brinkman. J. Chromatogr. A 794

(1998) 201.


182

[9] J. Ding, U.D. Neue. Rapid Commun. Mass Spectrom. 13 (1999) 2151.[10] W.A. Minnaard, A.C. Hogenboom, U.K. Malmqvist, P. Manini, W.M.A.

Niessen, U.A.Th. Brinkman. Rapid Commun. Mass Spectrom. 10 (1996) 1569.[11] M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, G.J. de Jong. Rapid

Commun. Mass Spectrom. 14 (2000) 2103.[12] S. Ulrich. J. Chromatogr. A 902 (2000) 167.[13] H. Lord, J. Pawliszyn. J. Chromatogr. A 902 (2000) 17.[14] H. Lord, J. Pawliszyn. J. Chromatogr. A 885 (2000) 153.[15] E.H.M. Koster, N.S.K. Hofman, G.J. de Jong. Chromatographia 47 (1998) 678.[16] N.H. Snow. J. Chromatogr. A 885 (2000) 445.[17] A.A. Boyd-Boland, J. Pawliszyn. Anal. Chem. 64 (1996) 1521.[18] B.J. Hall, M. Satterfield-Doerr, A.R. Parikh, J.S. Brodbelt. Anal. Chem. 70

(1998) 1788.[19] J. Ai. Anal. Chem. 69 (1997) 1230.[20] C. Pérès, C. Viallon, J.-L. Berdagué. Anal. Chem 73 (2000) 1030.[21] J.J. Langenfeld, S.B. Hawthorne, D.J. Miller. J. Chromatogr. A 740 (1996) 139.[22] A.P. Bruins. J. Chromatogr. A 794 (1998) 345.[23] M.G. Ikonomou, A.T. Blades, P. Kebarle. Anal. Chem. 62 (1990) 957.[24] E.L. Cussler. Diffusion, Mass Transfer in Fluid Systems, Cambrige University

Press, Cambrigde, 1984.[25] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle. Rapid Commun. Mass Spectrom.

13 (1999) 1175.[26] R. King, R. Bonfiglio, C. Fernandez-Metzler, C. Miller-Stein, T. Olah. J. Am.

Soc. Mass Spectrom. 11 (2000) 942.[27] D. Louch, S. Motlagh, J. Pawliszyn. Anal. Chem. 64 (1992) 1187.[28] Supelco, personal communications.

Ultra-rapid non-equilibrium SPME at elevated temperatures

183

5.2

Ultra-rapid non-equilibriumsolid-phase microextraction at elevatedtemperatures and direct coupling to ion-trap massspectrometry for the analysis of biological samples*

Summary

Solid-phase microextraction (SPME) has been directly coupled to an ion-trapmass spectrometer (MS) for the determination of lidocaine in urine. Thethroughput of samples has been increased using non-equilibrium SPME withpolydimethylsiloxane (PDMS) fibers. The effect of the temperature on thesorption and the desorption was studied. Elevated temperatures during sorption(65ºC) and desorption (55ºC) had a considerable influence on the speed of theextraction. The desorption was carried out with home-made desorption chamberallowing thermostating. Only 1 min sorption and 1 min desorption wereperformed, after which MS detection took place, resulting in a total analysistime of 3 min. Detection limits below 1 ng/ml could be obtained despite yieldsof only 2.1 and 1.5% for a 100 and a 30-µm PDMS-coated fiber, respectively.Furthermore, the determination of lidocaine in urine had acceptablereproducibilities, i.e. relative standard deviations (RSDs) below 10%. A limit ofquantitation (RSD<20%) of about 1 ng/ml was obtained. No extra wash step ofthe extraction fiber was required after desorption if a 30-µm coating was used,whereas not all the analyte was desorbed from the 100-µm coating in a singledesorption. Therefore, the SPME-MS/MS system with a 30-µm PDMS-coatedfiber for rapid non-equilibrium SPME at elevated temperatures is the mostsuitable for high-throughput analysis of biological samples.

*: M.W.J. van Hout, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de Jong. Submitted to Anal. Chem.


184

5.2.1 Introduction

In many areas of bioanalysis there is a strong call for high-throughputsystems. Besides fast analysis, the systems must also be selective and sensitiveas the analytes of interest should be determined in the sub-ng/ml range in acomplex matrix. Traditionally, solid-phase extraction (SPE) was performedoff-line [1-4], after which the eluate could be analysed by liquidchromatography (LC) or gas chromatography (GC). In these approaches, theSPE procedure was often the time-limiting step. Automation and coupling ofSPE and LC, which was more easily established than SPE-GC, allowed fasterextraction prior to separation [5,6]. By these means, systems were created inwhich the separation was the time-limiting step. To overcome the latterproblem, short LC columns were used [6-8], or the real separation step waseven omitted, thus SPE was directly coupled with mass spectrometry (MS)[9-14]. These systems make the extraction time more critical. This may beproblematic due to the various steps in the SPE procedure, i.e., activation,conditioning, sampling, washing and elution. Increasing the extraction speedmay result in less selectivity during the extraction, which subsequently canresult in detection problems, e.g., ion suppression [14-18]. Reduction of thematrix influence may be obtained by performing multiple-stage MS (MSn, withn≥2).

Solid-phase microextraction (SPME) is a simple technique. AlthoughSPME was originally designed for GC analysis [19-22], it is now also combinedwith LC. Usually, immersion of a coated fiber into the sample is carried out,after which the fiber is withdrawn from the sample and transferred to thedesorption chamber. Here, the analyte is released from the coating using a pHchange and/or an organic solvent. The main disadvantage of SPME is the slowsorption process, due to diffusion limitations. Fundamental studies [23,24] ofthe sorption processes pointed out that in an agitated sample mainly a staticwater layer around the fiber is limiting the sorption. Besides the time factor, theextraction yields with SPME are relatively low [21]. Normally, the extraction isstopped when equilibrium is established between analyte concentrations in thesample and in the coating of the fiber, that is, if 95% of the maximum yield isestablished. In some studies, non-equilibrium SPME has been performed[25,26], in which the extraction was stopped before equilibrium was reached.

Möder et al. [27] coupled SPME directly to electrospray (ESI)/MS usingthe selected ion monitoring mode. A suitable desorption chamber was used. A60-min sorption time was applied after which direct detection was performed,i.e. no separation step was used. This study clearly showed the time-limitationof the SPME procedure. Moreover, despite the use of MS in the selected ionmonitoring mode, severe matrix interference due to poor selectivity deterioratedthe reproducibility of the system (relative standard deviations (RSDs) >15% at


185

ng/ml-levels). SPME was also directly coupled to ESI-high field asymmetricwaveform ion mobility spectrometry-MS for the determination ofamphetamines in urine [28]. High sensitivity (down to 200 pg/ml) andselectivity were obtained. However, the sorption time was still 12 min (with ayield of about 1%), and the static desorption took about 3 min. Furthermore, a2-min cleaning of the coating (at 250°C) was performed to prevent carry-over.Thus, the SPME procedure still hampered the sample throughput. In a previousstudy of our group [29] lidocaine was determined in urine by SPME directlycoupled to MS, thus eliminating the separation step. Good selectivity wasobtained by the SPME procedure and by the use of MS/MS. A 100-µmpolydimethylsiloxane (PDMS) coated fiber was used for extraction at roomtemperature. A 0.5 ml/min flow-rate after desorption proved to be optimal fordetection and peak shape. The sample throughput was enhanced by performingnon-equilibrium SPME, applying only 5 min sorption and 4 min desorption.However, despite the short SPME procedure, this remained the time-limitingstep, as the MS detection took only about 1 min. Furthermore, a washing of thefiber was required after desorption to eliminate remaining amounts of analyte inthe coating (about 10%) after a single desorption. The washing took anadditional 4 min. The limit of detection (LOD) was about 0.4 ng/ml and theRSDs were below 14%. This system showed the potential of non-equilibriumSPME, and the results of this study were the starting point for the currentinvestigation of the applicability of ultra-rapid non-equilibrium SPME at anelevated sorption and desorption temperature.

5.2.2 Experimental section

5.2.2.1 Chemicals and Instrumentation

Acetic acid, boric acid, and sodium chloride (all of analytical-reagentgrade) were obtained from Merck (Darmstadt, Germany) and ammoniumacetate (99.99%) was from Sigma-Aldrich (St. Louis, MO, USA). Acetonitrile(Lab-Scan, Dublin, Ireland) was of HPLC quality. Stock solutions of lidocainehydrochloride (USP, Holland Pharmaceutical Supply, Alphen a/d Rijn, TheNetherlands) were prepared in acetonitrile. Ammonium acetate buffer (10 mM)was prepared by dissolving ammonium acetate in ultrapure water and adjustingthe pH to 4.5 with acetic acid (10% v/v). Buffer solutions pH 10 (0.2 M) wereprepared by dissolving boric acid in ultrapure water and adjusting the pH with1 M sodium hydroxide. Water was obtained using an Elga Maxima ultrapurewater purification system (Salm & Kipp, Breukelen, The Netherlands).

PDMS-coated (30 and 100 µm) SPME fibers were obtained from Supelco(Bellefonte, PA, USA). The SPME holder and fiber were compatible with a


186

modified desorption chamber. Heating during sorption was performed using amagnetic heater (Metrohm, Herisau, Switzerland). A Hewlett-Packard(Waldbronn, Germany) HPLC gradient pump series 1100 was used fordesorption and transport of the solvent to the mass spectrometer. The massspectrometer was an LCQ Classic ion trap (Thermoquest, San Jose, CA, USA)equipped with an atmospheric pressure chemical ionisation (APCI) source.

Triplicate injections (50 µl) of control samples (24.81 ng/ml lidocaine inbuffer pH 4.5-acetonitrile (85:15 v/v)) were used to correct for day-to-dayvariations in the sensitivity of the mass spectrometer. Analyses performed toinvestigate the reproducibility were subjected to two-way analysis of variance(ANOVA) with replications.

5.2.2.2 SPME procedure

The method was similar to the one described previously [29]. Sampleswere prepared by diluting blank or spiked calf urine with 0.2 M borate buffer(pH 10) in a ratio of 1:1. Sodium chloride was added to the urine-buffer solutionto give a final salt concentration of 0.3 g/ml. The pH was readjusted to 10 with10% sodium hydroxide. The coated part of the fiber was completely immersedinto 1.25 ml sample and extraction was performed for a particular time, whilecontinuously stirring the sample using a 7×2 mm stirring rod and a magneticstirrer (Metrohm, Herisau, Switzerland). Desorption was performed by placingthe fiber into the desorption chamber, which was filled with 70 µl ammoniumacetate buffer (10 mM, pH 4.5):acetonitrile (85:15 v/v). After a certain time ofstatic desorption, the contents of the chamber were flushed to the MS with thesame solvent at a flow-rate of 0.5 ml/min.

Elevated sorption temperatures were obtained by placing the sample ontothe heater (Metrohm) and allowing the sample (about 5 min) to obtain thedesired temperature (about 65°C) prior to extraction under continuous stirring.Elevated temperatures during desorption (about 55°C) could be achieved bymodifying a standard SPME desorption chamber (Supelco). Four heatingelements were inserted into the stainless steel housing of the unit (Fig. 1). Byusing a thermocouple, the temperature of the desorption solvent could becontrolled during desorption. A PTFE housing was used for isolation of thedesorption unit.


Using the ion-trap mass spectrometer with an APCI source, the vaporisertemperature was set at 400ºC. During analysis the sheath gas and auxiliary gas(both nitrogen) were 48 and 3 (arbitrary units), respectively, and the coronadischarge current was set at 5.0 µA. The heated capillary was set at 140ºC and a


187

capillary voltage of 46.0 V was applied. The tube lens offset was 45.0 V. Thefirst octapole offset, the inter-octapole offset, and the second octapole offsetwere - 2.0 V, - 26.0 V, and - 6.0 V, respectively. All scans were recorded in thefull-scan mode with three microscans over the range m/z 75-300, using thepositive-ion mode. The maximum injection time of the ion-trap was set at300 ms. Helium was applied as dampening and collision gas. Extracted ionchromatograms in the single-MS mode were obtained for [M+H]+

(m/z 235 ± 0.5) and for MS/MS the fragment ion m/z 86 ± 0.5 was monitored.The isolation width during MS/MS experiments was 2.0 Th using a collisionenergy of 34%.

Fig. 1: Schematic presentation of (A) the modified desorption chamber withpossibility to heat the system, (B) the bottom view of modified desorptionchamber. 1 = SPME fiber holder, 2 = needle guide, 3 = needle,4 = compression unit, 5 = ferrule, 6 = SPME fiber, 7 = solvent desorptionchamber, 8 = thermocouple, 9 = PTFE housing (isolation), 10a-d = heatingelements, 11 = temperature control unit (not to scale).


188


5.2.3.1 Effect of temperature on sorption

In our previous study [29], all experiments were carried out at 25ºC andwith the 100-µm PDMS coating. In order to further enhance the sensitivity, thespeed of the sorption process, i.e. the diffusion, can be increased by usingelevated temperatures during extraction. Increasing the temperature from 25ºCto about 65ºC during sorption resulted in a decrease of the equilibrium time, thetime at which 95% of the maximum yield is reached, from about 90 min(Fig. 2A) to 28 min (Fig. 2B) for the 100-µm coating. Both time-sorption curvesin Fig. 2 were fitted according to equations described by Ai [26]. It should benoted that the equilibrium time in this study is longer than in our previousstudies [29,30] due to a different geometry of the experimental set-up, whichwas necessary to allow temperature control. Typical yields after 1 and 5 minwith sorption at 25ºC are 0.7 and 3.3%, respectively. At these sorption times,but with 65ºC, the yields change into 2.1 and 8.7%. Thus, in non-equilibriumsituations the yield is significantly higher at 65ºC, whereas at equilibrium onlyminor differences in the yield are observed at the temperatures investigated.

Fig. 2: Effect of sorption temperature (Ts) on yield and equilibrium time.Time-sorption curve of (A) 100-µm coating, Ts = 25ºC; (B) 100-µm coating,Ts = 65ºC. Concentration of 16 ng/ml lidocaine in urine.

The decrease in viscosity of the sample leads to more rapid diffusion [24,31],resulting in a shorter equilibrium time. In non-equilibrium situations the yield ishigher at elevated temperatures due to the increased diffusion through a static


189

water layer around the fiber. Based on thermodynamics, higher temperatureswould result in a decrease in the partition constant of the analyte and thus inlower yields [25]. However, this effect seems to be restricted, which is inaccordance with the results of others [31-34]. This may be explained by the factthat the sorption yield is the overall result of a series of processes, where anincrease in the sorption temperature may lead to opposite effects. Physicalchanges of the polymeric phase, as suggested by the results from differentialscanning calorimetry of PDMS [32], may have a positive effect on the partitionconstant.

Decreasing the volume of the coating implies a more rapid establishmentof the equilibrium. If the coating thickness decreases by a factor of three (i.e.,volume and equilibrium extracted amount decrease a factor of five), the timerequired to establish equilibrium is thought to decrease similarly [23]. However,the surface area of the coating also decreases, hereby reducing the effect of thesmaller volume. This decrease in area is counteracted due to a thinner staticwater layer around a smaller coating [32]. Thus, although various factorsinfluence the equilibrium time, the overall effect is a considerable decrease inequilibrium time with a smaller coating [24,32]. For the 30-µm coating, anequilibrium time of 3.5 min was achieved at 65ºC with a yield of 2.2%. So, theequilibrium time with the 30-µm coating is shorter than predicted using thedifference in volume between the 100-µm and the 30-µm coating. After 1 minsorption at 65ºC, a yield of 1.5% was obtained, which is already about 65% ofthe maximum yield.

5.2.3.2 Effect of temperature on desorption

The time-desorption curves presented in Fig. 3, using the 100-µm coating,have been obtained after 5 min sorption at 25ºC (¡ and o) or 65ºC (l and n),followed by desorption. Sorption at 25ºC and subsequent desorption at25ºC (¡) resulted in incomplete desorption, even after 10 min desorption. Afterabout 4 min, the desorption level had reached its plateau and no more analytecould be desorbed. This is due to the fact that the desorption is also anequilibrium process. A similar effect was observed for sorption at 65ºCcombined with desorption at 25ºC (l). The maximum desorption level wasreached after about 4 min. No more analyte was released from the coating, eventhough about 25% of the sorbed amount was still present in the coating.Lidocaine is protonated upon release from the coating. A change in propertiesof the PDMS coating during the high sorption temperature may result in astronger retention by the stationary phase.

A high desorption temperature (o and n) proved to be advantageous interms of completeness of the desorption regardless the sorption temperature.The overall effect of the higher desorption temperature is a shift of the


190

equilibrium of the desorption process towards the desorption solvent. The bestdesorption was observed at an elevated temperature after sorption at roomtemperature (o). However, the low sorption yield is disadvantageous for theoverall sensitivity of the system. Desorption at an elevated temperature ensuredthat the desorption was not only more complete, but the process was also muchfaster at this temperature. This effect was most prominently observed whencomparing the results of experiments in which similar sorption and desorptiontemperatures, i.e., both at either low (¡) or at high temperatures (n) wereapplied. Despite the larger amount of analyte sorbed due to the higher sorptiontemperature (n), the desorption was more complete. Currently, we are stillinvestigating the fundamental aspects involved during the desorption.

Fig. 3: Time-desorption curves after 5 min sorption using a 100 µm PDMS-coatedfiber. Temperature during sorption (Ts) and desorption (Td): Ts = 25°C,Td = 25°C (¡); Ts = 65°C, Td = 25°C (l); Ts = 25°C, Td = 55°C (o); Ts = 65°C,Td = 55°C (n). Concentration of 16 ng/ml lidocaine in urine.

Using the 100-µm PDMS-coated fiber, after 4 min desorption at anelevated temperature about 5% of the sorbed analyte was still retained by thecoating. Thus, an additional wash step still remained to prevent carry-over, or alonger desorption time should be applied, since after 10 min desorption theamount of retained lidocaine is negligible. Both approaches are disadvantageous


191

for high-throughput analysis. With the 30-µm coating, a negligible amount oflidocaine was retained after desorption at 55ºC for 1 min.

5.2.3.3 Application

The SPME-MS/MS system was applied to the determination of lidocainein urine. The 100 and 30-µm coatings were both used to investigate theirpotentials towards ultra-rapid non-equilibrium SPME. For the 100-µm coating,using a 5-min sorption time at 65ºC combined with 4 min desorption at 55ºCresulted in an LOD (signal-to-noise ratio of 3) of about 0.16 ng/ml and a goodlinearity was obtained (Table 1). A sorption yield of 8.7% was found underthese conditions. However, after the desorption about 5% of the extractedlidocaine was still retained by the fiber, which required removal by anadditional wash step. Shorter sorption and desorption times were also applied tothis fiber, i.e. both processes were performed for 1 min. A slightly higher LODis obtained due to the lower yield (about 2%). During the 1 min desorption alsoabout 95% of the extracted lidocaine was removed from the fiber. Thus, anextra washing step, with the same solvent, was still required to preventcarry-over. Since the incompleteness of the desorption was disadvantageous,only a small concentration range was tested.

Table 1. Conditions and corresponding analytical data of SPME-MS/MS applyingnon-equilibrium SPME with PDMS-coated fibers at elevated temperatures(sorption at 65°C; desorption at 55°C).

100 µm 100 µm 30 µmSorption time (min) 5 1 1Desorption time (min) 4 1 1Sorption yield (%) 8.7 2.1 1.5Desorption yield (%) 95 95 100Limit of detection (ng/ml) 0.16 0.40 0.50Correlation coefficient 0.9998 0.9960 0.9986Range (ng/ml) 0.16-400 0.40-16* 0.5-225

*: maximum concentration studied

To obtain an ultra-rapid system with good sensitivity and full desorptionof the analyte without an extra wash step, the 30-µm coating was used underextreme non-equilibrium conditions with elevated temperatures, i.e., 1 minsorption at 65ºC and 1 min desorption at 55ºC. Even though the yield was lowerthan with the 100-µm coating, the LOD (0.50 ng/ml; determined as three timesthe level of occasional spikes in the chromatogram) was still in the sub-ng/mlrange (Table 1). Since negligible amounts of analyte were retained by thecoating after the desorption, no extra wash step was required. Good linearitywas obtained over a large range. Chromatograms of blank and spiked urine are


192

presented in Figs. 4A and 4B, respectively. Some slight tailing of the peaks wasobserved due to the direct coupling of SPME and MS.

Fig. 4: Extracted ion chromatograms (m/z 86) of SPME-MS/MS with the 30-µmPDMS-coated fiber after 1 min sorption at 65°C and 1 min desorption at 55°C:(A) blank urine; (B) 1.6 ng/ml lidocaine in urine.

The repeatabilities (intraday RSD) and reproducibilities (interday RSD)obtained with the 30-µm coating are presented in Table 2. Both the intraday andthe interday RSDs were less than 10%, except near the LOD. At high


193

concentrations, i.e. 218 ng/ml, the reproducibilities were substantially better. Alimit of quantitation (LOQ; RSD <15%) of about 1 ng/ml could be obtained.The results show a slight improvement of the reproducibility in comparisonwith our previous study [29], in which rapid non-equilibrium SPME at roomtemperature was applied (intraday and interday RSD lower than 14%). The useof a deuterated standard or a homologue of lidocaine as internal standard mayhelp to obtain even better reproducibility.

Table 2. Repeatability and reproducibility of the SPME-MS/MS system with the 30-µmPDMS coating using elevated temperatures during sorption and desorption(both 1 min) for the determination of lidocaine in urine.

Concentration (ng/ml) 0.9 12 44 218Intraday RSD (%) 18.6 8.5 9.1 1.2Interday RSD (%) 18.5 8.1 9.2 2.5

Note: n = 3 for 0.9 ng/ml; n = 6 for other concentrations.

5.2.4 Conclusions

The current study has clearly shown the potential of ultra-rapid SPME forhigh-throughput systems. Even though no chromatographic step wasincorporated, the use of MS/MS provided adequate selectivity. The sensitivityof the combined SPME-MS/MS system was acceptable with LODs in thesub-ng/ml range and an LOQ of about 1 ng/ml. To obtain goodreproducibilities, the conditions of ultra-rapid non-equilibrium SPME should becarefully controlled. Elevated temperatures during sorption and desorptionincreased the diffusion, and thus the sensitivity and/or speed of the system, andcontrolling the temperatures improved the reproducibility. A high desorptiontemperature is advantageous for complete desorption and should thus be used incombination with any sorption temperature. If a high sorption temperature isused, a high desorption temperature is even more required to overcomeretention of analyte in the coating, otherwise an additional wash step is needed.The 30-µm coated PDMS fiber was most suitable for the rapid determination oflidocaine in urine, since an acceptable yield was obtained with goodreproducibility, and a negligible amount of analyte was retained by the coatingafter a single desorption.

The throughput of the current system, applying 1 min sorption, 1 mindesorption and 1 min detection, can be increased by using two fibers and twodesorption chambers. When the first fiber is in the sorption phase, the secondfiber can be desorbed, while the eluate of a third sample can be analysed by MS.In this way, 3 analyses can be performed within 3 min, increasing thethroughput to 1 sample/min, provided that all samples have been brought to


194

65°C prior to starting the sorption process. Obviously, when applyingultra-rapid SPME-MS/MS, attention must be paid to the performance of the MSover time, since matrix components may precipitate on parts of the MS.Possible memory effects by the fiber after repeated use should also bemonitored. Negligible amounts of analyte retained by the coating after onedesorption may build up to a larger amount, especially after analysis of highconcentrations of analyte, and may finally cause carry-over. The performanceand reproducibility of different batches of fibers should also be investigated.The latter aspect will be more important when multi-fiber systems are used.

Acknowledgements

Jan Brands from Sigma-Aldrich is gratefully acknowledged for supplyingthe SPME fibers. This research was supported by the Technology FoundationSTW, applied science division of NWO and the technology programme of theMinistry of Economic Affairs.

5.2.5 References

[1] M. Dressler. J. Chromatogr. 165 (1979) 167.[2] A. Lagana, B.M. Petronio, M. Rotatori. J. Chromatogr. 198 (1980) 143.[3] R.E. Majors. LC-GC Intern. May (1998) S8.[4] Z.E. Penton. Advances in Chromatogr. 37 (1997) 205.[5] M. Jemal, D. Teitz, Z. Ouyang, S. Khan. J. Chromatogr. B 732 (1999) 501.[6] D.A. McLoughlin, T.V. Olah, J.D. Gilbert. J. Pharm. Biomed. Anal. 15 (1997)

1893.[7] M.L. Constanzer, C.M. Chavez, B.K. Matuszewski, J. Carlin, D.Graham.


J.H.G. Jonkman, Ph. Ladure, C. Puozzo. J. Chromatogr. B 708 (1998) 113.[9] A. Schellen, B. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de

Lagemaat, E. Verheij. Rapid Commun. Mass Spectrom. 14 (2000) 230.[10] A.C. Hogenboom, P. Speksnijder, R.J. Vreeken, W.M.A. Niessen,

U.A.Th. Brinkman. J. Chromatogr. A 77 (1997) 81.[11] A.C. Hogenboom, W.M.A. Niessen, U.A.Th. Brinkman. J. Chromatogr. A 794


Niessen, U.A.Th. Brinkman. Rapid Commun. Mass Spectrom. 10 (1996) 1569.[14] M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer, G.J. de Jong. Rapid

Commun. Mass Spectrom. 14 (2000) 2103.[15] K. Matuszewski K, M.L. Constanzer, C.M. Chavez-Eng. Anal. Chem. 70 (1998)

882.


195



[18] I. Fu, E.J. Woolf, B.K. Matuszewski. J. Pharm. Biomed. Anal. 18 (1998) 347.[19] S. Ulrich. J. Chromatogr. A 902 (2000) 167.[20] G. Theodoridis, E.H.M. Koster, G.J. de Jong. J. Chromatogr. B 745 (2000) 49.[21] N.H. Snow. J. Chromatogr. A 885 (2000) 445.[22] C.L. Arthur, J.B. Pawliszyn. Anal. Chem. 62 (1990) 2145.[23] D. Louch, S. Motlagh, J.B. Pawliszyn. Anal. Chem. 64 (1992) 1187.[24] J.B. Pawliszyn, Solid Phase Microextraction – Theory and Practice, Wiley, New

York, 1997.[25] H. Lord, J.B. Pawliszyn. J. Chromatogr. A 902 (2000) 17.[26] J. Ai. Anal. Chem. 69 (1997) 1230.[27] M. Möder, H. Löster, R. Herzschuh, P. Popp. J. Mass Spectrom. 32 (1997) 1195.[28] M.A. McCooeye, Z. Mester, B. Ells, D.A. Barnett, R.W. Purves, R. Guevremont.

Anal. Chem. 74 (2002) 3071.[29] M.W.J. van Hout, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw, G.J. de Jong.

Analyst 127 (2002) 355.[30] E.H.M. Koster, N.S.K. Hofman, G.J. de Jong. Chromatographia 47 (1998) 678.[31] K. Jinno, M. Taniguchi, M. Hayashida. J. Pharm. Biomed. Anal. 17 (1998) 1081.[32] H.A.G. Niederländer, V. Jas, M.W.J. van Hout. Submitted to Anal. Chem.[33] K. Jinno, M. Kawazoe, M. Hayashida. Chromatographia 52 (2000) 309.[34] M. Satterfield, D.M. Black, J.S. Brodbelt. J. Chromatogr. B 759 (2001) 33.

66

GENERAL CONCLUSIONS ANDFUTURE PERSPECTIVES

The pieces have been put together

Conclusions and perspectives: SPE-GC

199

In this thesis several coupled techniques have been evaluated forintegrated sample preparation with subsequent separation and/or detection.Although the used techniques differ substantially, all systems were developedwith a similar objective, i.e., the systems should allow high-throughput,sensitive and selective analysis of biological samples by using a samplepretreatment step coupled with the separation and/or detection step.

Although some applications in the environmental field could be found inthe literature about the use of the programmed temperature vaporiser (PTV) forlarge-volume injection (LVI) into gas chromatography (GC), the use of the PTVfor the bioanalysis of drugs was unexplored at the beginning of the project asdescribed in this thesis. Even though the PTV allows LVI/GC, the maximuminjection volume is still too limited when conventional solid-phase extraction(SPE) was performed. Therefore, miniaturised SPE in the pipette-tip (SPEPT)format was applied in order to inject the entire eluate of 100 µl. Since only asmall desorption volume was used, on-line desorption could be performed,hereby increasing the degree of integration of SPE and GC. The highestintegration of SPE applying liquid desorption and GC was obtained by insertingthe SPEPT into the injector. Liquid desorption could be performed on-line bydirect injection of the desorption solvent onto the SPE disk and the packed liner.Another interesting option of the PTV is thermal desorption (TD). This wasperformed in a final system in which the extraction was carried out off-lineinside a liner and, after drying of the extraction unit, the liner was inserted intothe injector and thermal desorption was performed. The main problem was toobtain a clean and thermostable stationary phase with good extractionproperties. Although the ideal phase was not found, Tenax was shown to besuitable for SPETD of biological samples. It should be noted that thermaldesorption overcomes the problems encountered with LVI, such as the inertnessof the liner packing and injection of large amounts of impurities. Yet, SPETD isonly applicable to analytes that are thermostable in the liner. Finding newsuitable stationary phases will be of utmost importance for further optimisation.

In general, miniaturised SPE allows a rapid extraction, and the use of ashort GC column also helps to decrease the total analysis time. Thus, rapidanalysis of biological samples can be obtained by integrated SPE-PTV/GCapplying either liquid or thermal desorption. A disadvantage of miniaturisingthe sample preparation may be that a smaller sample volume can only beapplied, hereby possibly limiting the concentration sensitivity. This may beovercome by the use of sensitive detection systems. The selectivity of theextraction and detection should be carefully investigated, regardless the type ofdesorption. Integrated SPE-GC is a good alternative for and preferable toliquid-liquid extraction-GC. Since GC offers high separation efficiency,SPE-GC is a very strong combination if the analytes of interest are

Chapter 6 – General conclusions and future perspectives

200

thermostable. The use of GC is still required in various fields such as drugabuse testing and working place analysis. Furthermore, GC is often a useful toolfor compounds that cannot be analysed by LC with UV detection due to the lackof UV absorption.

At this moment, an on-line SPE-PTV/GC system has already beendescribed and is commercially available. Such a system uses an SPE cartridgethat has originally been developed for on-line SPE-LC. A disadvantage of theseclassical cartridges is the long drying time (typically 15-30 min). Furthermore,the elution requires a considerable amount of desorption solvent, whichsubsequently needs to be evaporated inside the injector. This thesis providessome interesting solutions to these problems. A possible design for miniaturisedextraction is visualised in Fig. 1A, whereas the combination of SPE and GCwith various desorption modes is depicted in Fig. 1B-D.

Fig. 1: SPE device with a syringe for extraction (A) and a possible set-up for liquiddesorption with injection on a packed liner (B) or applying the sweeping balltechnique (C) and a possible set-up for thermal desorption (D). 1 = syringe;2 = extraction unit with cap; 3 = shortened liner without cap; 4 = shortenedliner with cap; extraction unit without cap.

The extraction unit may be a modified SPEPT. After rapid extraction, e.g. bythe use of a pipettor (unit without a cap) or a syringe (Fig. 1A), the extractionunit is transferred (by a robotic system) and inserted into the liner.Subsequently, liquid desorption can be applied by injection of solvent on top ofthe SPE phase, and the transfer of the solvent to the packed liner can take placeby gas pressure above the SPE phase (Fig. 1B). Subsequently, evaporation ofthe eluate is performed through the vent opening. In order to overcome possibleinertness problems of the liner packing the ‘sweeping ball’-technique may be

Conclusions and perspectives: SP(M)E-MS

201

used (Fig. 1C). Solvent stops flowing into the column at the point whereequilibrium is established between the carrier gas pressure and the solventvapour pressure. The ball will prevent further leakage of solvent into thecolumn as well. Subsequently, the solvent vapour is vented, and finally heatingof the injector transfers the concentrated analytes towards the GC column.Alternatively, the extraction unit can also be placed onto the bottom of theinjector (Fig. 1D, part 5), after which a shortened liner (part 4) is placed on topof the SPE device, hereby creating the possibility for thermal desorption.

In the applications of on-line SPE-LC a tendency can be observedtowards shorter columns to obtain high-throughput systems. These SPE-(shortcolumn)-LC systems usually apply MS as the detection system. For bothSPE-LC-MS and SPE-GC-MS, the ultimate step is the direct coupling of SPEwith MS via a suitable interface, that is an LC-MS or GC-MS interface. For anSPE-MS system, the SPE procedure is the time-limiting step, since the MSdetection will take only about 1-1.5 min. High flow-rates (up to 5 ml/min)during the various steps of the SPE procedure may enhance the samplethroughput, resulting in a total analysis time less than 5 min. The use of evenhigher flow-rates (up to 10 ml/min) as well as the parallel analysis of twosamples (Prospekt II, Spark Holland, Emmen, The Netherlands) are otherapproaches to increase the sample throughput.

SPE-MS is a powerful system for rapid analysis, but not only the speedshould be maximised, but it is also essential to obtain good selectivity in boththe extraction and detection. For the latter, multiple-stage MS (MSn with n≥2)should be applied. Obviously, this reduces only the detection of matrixcompounds, but these compounds may still be present. The extraction is thusthe most important step here for the actual removal of matrix components.Co-elution of the analyte of interest with matrix compounds can lead todetection problems such as ion suppression, which may deteriorate the analysis.A combination of experiments is recommended to obtain qualitative andquantitative information about the ion suppression. Investigating the causes ofion suppression and the approaches to eliminate ion suppression will becomemore important in the near future, in particular if SPE-MSn systems are appliedfor high-throughput bioanalysis. Additional selective extraction techniques,such as affinity phases or molecularly imprinted polymers, may help to decreaseion suppression. Expanding the utility of SPE-MSn system to other fields likeproteomics can also be of interest. In this field, a suitable affinity phase can beused for the selection of a certain class of proteins. Such systems can offerinformation on both the biological activity and the chemical structure.

Solid-phase microextraction (SPME) suffers from limited choice ofselectivity in comparison with SPE, since only a few stationary phase types are

Chapter 6 – General conclusions and future perspectives

202

available for the former technique. However, a different selectivity may beobserved, when compared to SPE, due to the diffusion-based extraction. WithSPE, a new SPE cartridge is generally used for each extraction. With SPME, thecoated fibers are still re-used, which implies a significant risk of carry-over.Finally, with SPE recovery and speed of the extraction are normally notcorrelated. With SPME, the highest yield is obtained after a long extraction, i.e.,speed and sensitivity are inversely related. Therefore, up till now, SPME has notbeen considered to be of interest for high-throughput systems. Nonetheless, inthis thesis we investigated to what extent SPME-MSn, coupled directly by asuitable interface, could be converted into an ultra-rapid analytical system forthe bioanalysis of drugs, while suitable sensitivity is obtained.

In order to develop rapid SPME-based systems, non-equilibrium sorptionconditions were applied. Applying an elevated temperature duringnon-equilibrium SPME increased the diffusion and higher yields were observed.By use of a similar desorption temperature with a special custom-made unit, thespeed of the desorption was enhanced as well. Finally, non-equilibrium SPMEwith a 30-µm polydimethylsiloxane coating and elevated temperatures duringthe sorption and the desorption allowed the ultra-rapid (within 3 min)determination of lidocaine in urine. Good reproducibility (relative standarddeviation �10%) and sensitivity (sub-ng/ml range) were obtained. To obtainbetter reproducibility, the use of a suitable internal standard may be suitable.Only a negligible amount of analyte was retained after a single desorption. Forthe present application, no interference from the urine matrix was observed(with MS2 detection), which may be due to the non-exhaustive principle of theextraction procedure.

The automation of the SPME procedure (sorption and desorption) caneasily be obtained by the use of commercially available robotic devices. Untilnow, a single-fiber approach was used. A multi-fiber system might beenvisaged, applying two fibers and two desorption chambers. One fiber is in thesorption stage, the second fiber can be desorbed, while the third stage is thedetection. In this way, the sample throughput may easily be increased to about1 sample/min.

In the present approach, only one test drug was used to examine thepossibilities of SPME. SPME-MSn can also be applied to other drugs and otherbiological matrices or in other fields such as proteomics or environmentalanalysis. In the future, the selectivity and sensitivity of this set-up should becompared with SPE-MSn, as well as with other SPME devices such as thestir-bar extraction and in tube-SPME. Since SPME was originally developed forcombination with thermal desorption and GC analysis, it also seems interestingto investigate the potential of SPME with thermal desorption coupled to a shortGC column or even also directly coupled to MS.

Conclusions and perspectives: final remarks

203

In conclusion, the systems developed in the second part of this thesis arepowerful techniques based on the on-line coupling of sample pretreatment withselective detection, i.e. MS, offering high-throughput potential for bioanalyticalpurposes. It should be noted, however, that all system set-ups were tested withbiological samples spiked with the analyte(s) and real-life samples were not yetanalysed. The use of these systems for the routine analysis of real-life sampleswill require additional and appropriate validation.

Samenvatting

205

Samenvatting

In-lijn koppeling van monstervoorbewerking metchromatografie of massaspectrometrie voor

hoge-doorvoer analyse van biologische monsters

Bioanalyse van geneesmiddelen, drugs en andere, gerelateerdeverbindingen is een specialisatie die zich richt op complexe monsters. De meestgeanalyseerde monsters zijn urine en bloed, plasma of serum. Via de urineworden veel (afval)stoffen uitgescheiden, al dan niet in gemetaboliseerde vorm.Bloed transporteert naast geneesmiddelen veel essentiële bouwstoffen enbioregulatoren door het gehele lichaam, en bevat daardoor veel exogene enendogene verbindingen.

Het belang van bioanalyse komt tot uiting in vele toepassingsgebiedenvan het farmaceutisch onderzoek. Nieuwe geneesmiddelen worden in diverseontwikkelingsfasen onderworpen aan diverse testen. In de preklinische faseworden monsters van dierlijke oorsprong geanalyseerd, terwijl in de klinischefase met name monsters van humane vrijwilligers en patiënten worden gemeten.In deze fase worden het potentiële geneesmiddel en diens metabolieten bepaald,en worden in combinatie met farmacokinetiek en -dynamiek de therapeutischeen toxische concentraties bepaald. De gewenste formulering van het potentiëlegeneesmiddel, welke voor een belangrijk deel ook steunt op de bioanalyse, dienteveneens bestudeerd te worden. Als het geneesmiddel op de markt komt is‘therapeutic drug monitoring’ van belang. Door de toegenomen kennis van dewerking van geneesmiddelen worden potentere stoffen ontwikkeld, waardoor dedosering steeds lager kan worden. Hierdoor wordt ook de concentratie van hetgeneesmiddel en/of diens metabolieten steeds lager. Bovendien moet deontwikkeling van een geneesmiddel zo weinig mogelijk tijd kosten. Andereapplicaties van bioanalyse waarbij veel monsters moeten worden geanalyseerdzijn o.a. de controle van voeding op residuen van geneesmiddelen engevaarlijke stoffen, testen op drugsmisbruik en de klinische en forensischetoxicologie. Kortom, het aantal complexe biologische monsters neemt toe,terwijl de te bepalen concentraties van de componenten afnemen. Hierdoor iseen grote vraag ontstaan naar gevoelige en selectieve systemen met een hogemonsterdoorvoer. Gebruikelijke systemen in de bioanalyse bestaan uit eenmonstervoorbewerking, een scheidingsstap en uiteindelijk de detectie, waarbijkwantificering en zonodig identificatie plaatsvindt.

Samenvatting

206

Diverse monstervoorbewerkingstechnieken zijn ontwikkeld voor debioanalyse. De meest klassieke vorm is vloeistof-vloeistof extractie (LLE).Tegenwoordig wordt vaker vaste-fase extractie (SPE) gebruikt. Deze techniekkan worden gebruikt voor algemene screening op geneesmiddelen, maar ookvoor de selectieve extractie van een component of een groep verbindingen.Ondanks de ontwikkeling van nieuwe technieken bleef monstervoorbewerkingvaak een zwakke schakel in een bioanalytisch systeem. Daarom is veel werkverricht om de mogelijkheden en beperkingen van deze methoden, ook tenaanzien van automatisering en/of miniaturisering, in kaart te brengen. Parallelaan de ontwikkeling van betere monstervoorbereidingstechnieken liep deontwikkeling van selectieve detectiesystemen. Met name massaspectrometrie(MS) levert veel selectiviteit en gevoeligheid met mogelijkheden totidentificatie en kwantificering van de analiet. De ontwikkeling van interfacesdie bij atmosferische druk werken, hebben gezorgd voor grote interesse invloeistofchromatografie (LC)-MS, en dit is momenteel de belangrijkste,routinematig toegepaste techniek in de bioanalyse. Door de toename inselectiviteit in de detectie werd het mogelijk om in voorkomende gevallen deselectiviteit in de scheidingsstap te verminderen. Dit had tot gevolg dat snellescheidings- en detectiesystemen ontstonden, waardoor de monster-voorbewerking wederom snelheidsbepalend werd. Daarom moet deze stapverder versneld, geminiaturiseerd en/of geautomatiseerd worden.

Er zijn diverse mogelijkheden om tot miniaturisering te komen. SPE kanworden verkleind door minder stationaire fase te gebruiken. Hierdoor wordenalle te gebruiken volumina kleiner en kan SPE sneller worden uitgevoerd. Deintegratie van SPE met LC is al langer succesvol gerealiseerd, aangezien heteluaat relatief eenvoudig rechtsreeks in het LC systeem kan wordengeïnjecteerd (zelfs zonder miniaturisering van SPE). Het kleinere elutie-volume(tot 100 µl) maakt het ook mogelijk SPE te integreren met gaschromatografie(GC), met dien verstande dat de SPE fase goed gedroogd moet worden voor deelutie en een geschikte injector moet worden gebruikt, die groot-volume injectie(LVI) in GC mogelijk maakt. In dit onderzoek is gebruikt gemaakt van de‘geprogrammeerde-temperatuur verdamper’ (PTV). Deze injector maakt hetmogelijk om naast vloeistofdesorptie ook thermische desorptie toe te passen.Voor dit laatste moet wel een geschikte stationaire fase worden gevonden metgoede extractie eigenschappen en thermostabiliteit.

Een andere interessante vorm van geminiaturiseerde extractie isvaste-fase microextractie (SPME). De extractie is gebaseerd op eendiffusieproces van de analiet uit het monster naar een gecoate fiber.Oorspronkelijk is SPME ontwikkeld om te worden gecombineerd metthermische desorptie en GC, maar de mogelijkheden voor combinatie metvloeistof-desorptie en LC hebben inmiddels ook veel aandacht gekregen. Degrootste beperkingen van SPME zijn de lage opbrengst en lange extractietijd.

Samenvatting

207

De diffusie beperkt de monsterdoorvoer, doordat het lang duurt voordatevenwicht is bereikt tussen analiet in het monster en de fiber. Bovendien is destatische desorptie bij SPME-LC, evenals de sorptie, een evenwichtsproces,waardoor de kans op carry-over bestaat.

Voor de ontwikkeling van snelle, selectieve en gevoelige systemen diegebaseerd zijn op on-line monstervoorbewerking en analyse zijn demogelijkheden en beperkingen van SPE-GC en de directe koppeling van SPE enSPME met MS onderzocht.

In Hoofdstuk 1 staan de uitgangspunten en doelen van het verrichteonderzoek beschreven. Hoofdstuk 2 geeft een overzicht van nieuwe trends inintegratie van monstervoorbewerking en een scheidings- en/of detectiesysteem.De mogelijkheden en beperkingen van de diverse systemen worden besprokenen vergeleken.

Hoofdstuk 3 bevat het eerste deel van het onderzoek, namelijk deintegratie van (geminiaturiseerde) SPE met GC. Sectie 3.1 toont de resultatenvan de evaluatie van de PTV voor LVI/GC van extracten van biologischemonsters. De zuiverheid van het oplosmiddel en de linerpakking (maximuminjectie-volume en inertheid) zijn bestudeerd. Daarnaast blijkt de selectiviteitvan de detectie (MS) kritisch. Na indampen en reconstitutie kan het volledigeextract (100 µl) worden geïnjecteerd. Een detectielimiet van 0.25 ng/ml isbehaald voor lidocaine, fenobarbital, secobarbital en diazepam in plasma.

In Sectie 3.2 zijn SPE-pipetpunten (SPEPTs) toegepast, waardoor eenkleiner desorptievolume kan worden gebruikt. Een uitwendig koppelstukjemaakt on-line desorptie mogelijk. Een verdergaande integratie van SPEPTs enGC (Sectie 3.3) is mogelijk door inkorten van de SPEPT en de liner. Hierdoor iseen in-lijn koppeling gecreëerd. Off-line extractie van 200 µl plasma kan zogecombineerd worden met in-lijn desorptie door injectie van desorptievloeistofop de SPE disk in de injector. De detectielimieten zijn 0.75 en 2.5 ng/ml voorrespectievelijk lidocaine en diazepam in plasma.

Het gebruik van SPE met thermische desorptie (SPETD) staat beschrevenin Sectie 3.4. Alhoewel de ideale stationaire fase nog niet is gevonden, blijktTenax voldoende thermostabiel te zijn en kunnen eveneens goede extractie-opbrengsten worden verkregen (recoveries voor lidocaine en diazepamrespectievelijk 73 en 74%). Slechts 5 mg Tenax is in een liner gedaan waarinook de (off-line) extractie is uitgevoerd. Na drogen is de liner in de injectorgezet en is thermische desorptie toegepast. Door de kleine hoeveelheidstationaire fase is de totale extractietijd slechts 8 min., inclusief 5 min. drogen.Hoewel slechts 50 µl urine is gebruikt, is een bepalingslimiet (relatievestandaard deviatie (RSD) <15%) behaald van 1.0 ng/ml voor beidecomponenten. De in dit hoofdstuk beschreven systemen zijn een belangrijke

Samenvatting

208

stap in de richting van nieuwe integratie- en automatiseringsmogelijkheden voorSPE – GC.

Hoofdstuk 4 beschrijft het tweede experimentele gedeelte van hetproject, te weten SPE-MS. Clenbuterol is bepaald in urine (Sectie 4.1) doorextractie over een polymeerfase en detectie met een ion-trap MS in de MS3

modus. Schone chromatogrammen zijn verkregen, maar er is wel sprake vanionsuppressie. De bepalingslimiet is ongeveer 1.0 ng/ml bij gebruik van 1 mlurine. Er is een goed lineair verband gevonden over een grootconcentratiebereik (1.0-250 ng/ml).

De ionsuppressie-effecten zoals die werden waargenomen bij de bepalingvan clenbuterol zijn verder onderzocht (Sectie 4.2). Met behulp van continupost-cartridge infusie van de analiet is ook een concentratie-afhankelijkheid vande suppressie-effecten aangetoond. Verder zijn de twee stationaire fasen(polydivinylbenzeen [PDVB] en molekulair geïmprinte polymeren [MIP])vergeleken. Bij de PDVB fase is de ionsuppressie door de urinematrixveroorzaakt, terwijl bij de MIPs de bleeding van het imprintmolekuul(broombuterol) tot sterke suppressie van het clenbuterol-signaal leidt.

In Sectie 4.3 staat een SPE-MS2-methode beschreven voor de bepalingvan prednisolon in serum. Ongunstige fragmentatiepatronen en matrix-interferentie zorgen voor een detectielimiet van ca. 10 ng/ml bij gebruik van eentriple-quadrupole MS. SPE-MS2 met een ion-trap zorgt voor een goedeselectiviteit, met als gevolg dat een detectielimiet van ongeveer 5 ng/ml. Debepalingslimiet is ca. 10 ng/ml, en een goede lineariteit over het bereik10-550 ng/ml is verkregen. Tevens is een goede reproduceerbaarheidaangetoond (RSD <10% bij 20 ng/ml). Voor deze applicatie met dit systeem isgeen ionsuppressie waargenomen.

Het laatste deel van het proefschrift met betrekking tot experimentenbeschrijft de directe koppeling van SPME met MS (Hoofdstuk 5). Lidocaine isuit urine geëxtraheerd met een 100-µm polydimethylsiloxaan (PDMS) gecoatefiber (Sectie 5.1). De experimenten zijn bij kamertemperatuur uitgevoerd. Deevenwichtstijd is ongeveer 45 min. Aangezien door de directe koppeling met deMS de detectietijd slechts 1 min is, is overgegaan op niet-evenwicht SPMEtoegepast. De sorptie is na 5 min. gestopt, waarna er 4 min. is gedesorbeerd. Dedesorptievloeistof is vervolgens naar de MS getransporteerd. De totaleanalysetijd wordt nu 10 min., met een detectielimiet van 0.4 ng/ml en eenbepalingsgrens van ongeveer 2 ng/ml, waarbij een acceptabelereproduceerbaarheid (RSD <15%) wordt verkregen. Na een enkele desorptie isechter nog ongeveer 10% van de gesorbeerde analiet in de coating aanwezig.

Een van de beperkende factoren tijdens de sorptie is de diffusie door eenstatisch waterlaagje rond de fiber. Een hogere sorptietemperatuur kan dediffusie door het laagje versnellen. In Sectie 5.2 is derhalve een hogesorptietemperatuur (65°C) toegepast. Het bleek nu mogelijk om met een 30-µm

Samenvatting

209

fiber goede selectiviteit en gevoeligheid te verkrijgen met slechts 1 min. sorptie,1 min. desorptie en 1 min. detectie. De bepalingsgrens (RSD <15%) is ongeveer1 ng/ml. Het blijkt essentieel te zijn de desorptie ook bij hoge temperatuur(55°C) uit te voeren om de analiet sneller en vollediger uit de coating te krijgen,en hiermee ook carry-over te elimineren. Verbeterde temperatuurcontrole m.b.v.een speciaal ontwikkelde desorptie-unit zorgt tevens voor een goedereproduceerbaarheid (RSD <10%) over een relatief groot lineair bereik(5-220 ng/ml). Dit laatste systeem toont heel duidelijk de mogelijkheden voorhoge monsterdoorvoer-analyse op basis van niet-evenwicht SPME bij hogeresorptie- en desorptietemperatuur en directe koppeling met MS.

Het laatste deel van dit proefschrift (Hoofdstuk 6) bevat enkele algemeneconclusies en een vergelijking van de onderzochte systemen. Daarnaast wordende mogelijkheden en beperkingen van deze systemen besproken en wordenenkele ideeën voor toekomstig onderzoek gegeven.

List of publications

211

List of Publications

Chapter 2 M.W.J. van Hout, H.A.G. Niederländer, R.A. de Zeeuw,G.J. de Jong.New developments in integrated sample preparation forbioanalysis.Accepted for publication in “Handbook of AnalyticalSeparations: Pharmaceuticals / Veterinary – Toxicology /Metabolism” (Ed. I. Wilson).

Section 3.1 M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong.Evaluation of the programmed temperature vaporiser for large-volume injection of biological samples in gas chromatography.J. Chromatogr. B 729 (1999) 199-210.

Section 3.2 M.W.J. van Hout, R.A. de Zeeuw, G.J. de Jong.Coupling device for desorption of drugs from solid-phaseextraction-pipette tips and on-line gas chromatographicanalysis.J. Chromatogr. A 858 (1999) 117-122.

Section 3.3 M.W.J. van Hout, W.M.A. van Egmond, J.P. Franke,R.A. de Zeeuw, G.J. de Jong.Feasibility of the direct coupling of solid-phase extraction-pipette tips with a programmed-temperature vaporiser for gaschromatographic analysis of drugs in plasma.J. Chromatogr. B 766 (2001) 37-45.

Section 3.4 M.W.J. van Hout, R.A. de Zeeuw, J.P. Franke, G.J. de Jong.Solid-phase extraction-thermal desorption-gas chromatographywith mass selective detection for determination of drugs inurine.Accepted for publication in Chromatographia.

Section 4.1 M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer,G.J. de Jong.On-line coupling of solid-phase extraction with massspectrometry for the analysis of biological samples. II.Determination of clenbuterol in urine using multiple-stagemass spectrometry in an ion-trap mass spectrometer.Rapid Commun. Mass Spectrom. 14 (2000) 2103-2111.

List of publications

212

Section 4.2 M.W.J. van Hout, H.A.G. Niederländer, R.A. de Zeeuw,G.J. de Jong.Ion suppression effects in the determination of clenbuterol inurine by solid-phase extraction-mass spsectrometry.Submitted to Rapid. Commun. Mass Spectrom.

Section 4.3 M.W.J. van Hout, C.M. Hofland, H.A.G. Niederländer,A.P. Bruins, R.A. de Zeeuw, G.J. de Jong.On-line coupling of solid-phase extraction with massspectrometry for the analysis of biological samples. III.Determination of prednisolone in serum.Submitted to J. Chromatogr. B.

Section 5.1 M.W.J. van Hout, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw,G.J. de Jong.Non-equilibrium solid-phase microextraction coupled directlyto ion-trap mass spectrometry for rapid analysis of biologicalsamples.Analyst 127 (2002) 355-359.

Section 5.2 M.W.J. van Hout, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw,G.J. de Jong.Ultra-rapid non-equilibrium solid-phase microextraction atelevated temperatures and direct coupling to massspectrometry for the analysis of biological samples.Submitted to Anal. Chem.

Additional papers on related subjectsM.W.J. van Hout, C.M. Hofland, V. Jas, H.A.G. Niederländer, R.A. de Zeeuw,G.J. de Jong.High-throughput analysis of biological samples by direct coupling of solid-phase (micro-)extraction and mass spectrometry.Chromatographia 55 (2002) S23-S24.

H.A.G. Niederländer, V. Jas, M.W.J. van Hout.Solid Phase Micro-Extraction Using PDMS Coated Fibers; NewConsiderations Concerning the Kinetics of Absorption and Desorption.Submitted to Anal. Chem.

Dankwoord

213

Dankwoord

Tijdens mijn onderzoek leek het af en toe erop dat zoeken naar een speld(of naald?) in een hooiberg makkelijker zou gaan, gezien de lage concentratiesdie we wilden bepalen in de moeilijke biologische matrices met zo eenvoudigmogelijke technieken. Deze laatste verschilden dan ook nog weer zodanig vanelkaar, waardoor het soms moeilijk was de grote lijn te blijven zien. Echter,beetje bij beetje bleken de puzzelstukjes in elkaar te passen en aan het eind zijnde spelden (of: naalden) dan toch van de biologische hooiberg gescheiden. Ditbetekent dat mijn vier jaar als AiO (of beter: OiO) er nu op zitten, en rest enkeldit nog: ‘collega’s, vrienden, familie en anderen die op welke wijze dan ook bijmijn promotie betrokken zijn geweest: BEDANKT’. Sommigen hebben tochmeer betekenis gehad dan anderen, en hen wil ik dan ook bij naam noemen.

Ad en Rokus, jullie boden mij de mogelijkheid om direct na mijndoctoraal bij farmacie aan een project te gaan werken waarbij ik mijn interessevoor de analytische chemie verder kon uitdiepen. Ruim vier jaar verder, heteindstadium van het onderzoek, was het af en toe wat lastig om jullie snel tebereiken vanwege de overstap naar Utrecht van Ad en emeritus-aangelegenheden van Rokus. Desondanks is het jullie gelukt om door deniet-afnemende stapel manuscripten heen te worstelen die ik jullie in dat laatstestadium voorschotelde, hetgeen uiteindelijk toch bijzonder vlot heeft geleid tothet verschijnen van dit proefschrift. Dank daarvoor.

Jan Piet en Andries, ondanks dat jullie inbreng vaak indirecte invloedhadden op het project heb ik jullie ideeën en discussies toch altijd als zeerwaardevol beschouwd. Harm (N.), je bent pas halverwege mijn project erbijbetrokken geraakt, maar altijd waren er weer ideeën voor nieuw onderzoek.Helaas heeft het door tijdgebrek niet kunnen leiden tot een extra hoofdstuk inmijn proefschrift. Ik hoop in de toekomst nog mee te kunnen blijven kijken innieuwe projecten zoals die nu beschreven staan.

Aangezien mijn project door STW gefinancierd werd, betekende dit dat erook een halfjaarlijks gebruikerscommissie-overleg was. Alle afzonderlijkeleden hiervan hartelijk dank voor jullie feedback. De mensen van ATAS wil ikextra hartelijk bedanken voor de keren dat ik bij jullie in het bedrijf te gastmocht zijn om wat te experimenteren met jullie apparatuur, en voor hetbeschikbaar stellen van de MSD gedurende de laatste fase van mijn project.

Hoewel het aantal stagiaires en bijvakkers niet overweldigend is dat aanmijn project heeft gewerkt, is hun bijdrage dat zeker wel. Corry, als ervarenanalist van de ziekenhuisapotheek kwam jij je bijvak doen en moest ik proberenom je nog iets nieuws te leren. Ik denk dat dat uiteindelijk prima is gegaan, endat je je goed thuis voelde binnen de groep mag blijken uit het feit dat je alweer

Dankwoord

214

geruime tijd als collega werkzaam bent. Ik leerde van jou ook twee belangrijkedingen, namelijk het aangename van ‘s ochtends vroeg beginnen op het lab enhet plezier van vaak en ver op vakantie gaan. Vincent kwam als stagiair vanuithet HLO om een jaartje onderzoek te verrichten, maar had er zoveel plezier enkunde in dat hij graag als vakantiekracht nog een aantal keer terug kon en wildekomen. Vincent en Corry, beiden bijzonder veel dank voor jullie inzet.

Alle (ex-) collega’s van de werkgroepen Farmaceutische Analyse,Bioanalyse & Toxicologie en Massaspectrometrie, bedankt voor de prettige tijddie ik samen met jullie in en rond het lab heb mogen doorbrengen. Daarnaastheb ik het genoegen gehad om met een groot aantal kamergenoten (Anneke,William, Alex, Lourens, Lutea, Roelof, Widia, Vincent, Laurent) meer ofminder tijd zin en onzin te mogen delen tussen alle werkzaamheden door. Dankvoor jullie aanspraak en gezelligheid.

Mijn motto was ‘mentale ontspanning en versterking door fysiekeinspanning’. Dit heb ik gelukkig samen met Albert kunnen bewerkstelligen.Eerst tijdens de volleybal, en daarna tijdens onze fitness-sessies, waarbij weregelmatig in onze strijd ‘niet voor elkaar onder te willen doen’ de spiermassatot het uiterste hebben geteisterd.

Michaël, al ruim negen jaar goede vriend, was als een van de weinigeechte ‘buitenstaanders’ van mijn project in staat om enigszins te begrijpen waarik mee bezig was. Dank je dat je nu mijn paranimf wilt zijn.

Tot slot ben ik bijzonder veel dank aan de personen voor wie hetinhoudelijke van mijn onderzoek vaak abracadabra was. Dat schrikte papa,mama, Jaschenka en Eliza echter niet af, en hun onvoorwaardelijke steun (zelfsvanuit de onbereikbare gebieden van Bosnië-Herzegovina) heeft het mogelijkgemaakt om naast de academische ontwikkeling ook op het persoonlijke vlaksuccessen te boeken.

En dan natuurlijk nog Naomi. Jouw aanwezigheid, liefde en steun op allefronten zorgden voor een zeer aangename en ontspannen laatste fase van demeest hectische periode van mijn promotie door het moeten afronden van depromotie, het beginnen met mijn nieuwe baan en het opknappen van de nieuwewoning. Dankjewel dat je ook nu naast me staat als paranimf tijdens mijnallerlaatste stadium als promovendus.

Allemaal bedankt

Notities

215

Date post:	15-Jun-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

University of Groningen On-line coupling of sample ... · On-line coupling of sample pretreatment...

Documents