MS Screening

Anal. Chem. 2004, 76, 6365-6373

Information-Dependent Acquisition-Mediated LC-MS/MS Screening Procedure with Semiquantitative Potential Tineke N. Decaestecker,† Sofie R. Vande Casteele,† Pierre E. Wallemacq,‡ Carlos H. Van Peteghem,§ Dieter L. Defore,| and Jan F. Van Bocxlaer*,†

Laboratory of Medical Biochemistry and Clinical Analysis, Laboratory of Toxicology, and Laboratory of Pharmaceutical Biotechnology, Ghent University, Harelbekestraat 72, B-9000 Ghent, Belgium, and Laboratory of Toxicology and Special Chemistry, Clinical Hospital St.-Luc, Hippocrate Avenue, B-1200 Brussels, Belgium

The development of a LC-MS/MS general unknown screening procedure for toxicologically relevant substances in blood samples by means of informationdependent acquisition on a Q-TOF is reported. IDA is an artificial intelligence-based product ion scan mode providing automatic “on-the-fly” MS to MS/MS switching. By performing information-dependent scanning at two different fragmentation energies, two collision-induced dissociation product ion spectra for each of the detected compounds are generated. As such, information-rich MS/ MS spectra are obtained from precursor ions not known beforehand. In addition, limitation of the MS/MS acquisition time to an acceptable minimum resulted in an almost instantaneous switch back to the MS mode. As such, this approach provided MS chromatograms that still could be of use for semiquantitative purposes. Since the switching intensity threshold, unequivocally related to the background noise, proved a critical parameter, the solid-phase extraction procedure, the liquid chromatographic conditions, and the mass spectrometric parameters all were optimized to the advantage of information-dependent acquisition. Finally, the screening procedure we developed was benchmarked, on one hand, qualitatively against the results obtained from traditional GUS approaches in a number of routine toxicological laboratories (20 samples) and, on the other hand, quantitatively with respect to its potential against established LC-MS/MS methods (7 samples). The procedure performed very well from a qualitative point of view; almost all of the drugs detected by the conventional techniques were identified, as well as additional drugs that were not previously reported. The procedure proved well-suited for an initial semiquantitative assessment, as is customary in, for example, forensic toxicology before accurate intoxication levels are determined using targeted analytical analyses. * To whom correspondence should be addressed. Phone: +32 9 264 81 31. Fax: +32 9 264 81 97. E-mail: [email protected]. † Laboratory of Medical Biochemistry and Clinical Analysis, Ghent University. ‡ Clinical Hospital St.-Luc. § Laboratory of Toxicology, Ghent University. | Laboratory of Pharmaceutical Biotechnology, Ghent University. 10.1021/ac0492315 CCC: $27.50 Published on Web 09/25/2004

© 2004 American Chemical Society

Systematic toxicological analysis (STA) in forensic toxicology comprises general unknown screening (GUS) procedures, whether restricted to well-defined subgroups, as well as specific confirmation and quantitation of individual compounds. GUS procedures can be defined as analytical techniques or combinations of analytical techniques aimed at detecting and identifying unknown compounds in biological fluids.1 Even though, in some forensic cases, the implicated substance involved is known or its involvement strongly suspected, the possibility that other toxic compounds may have contributed to the observed biological effect cannot be excluded. Efficient and extensive screening strategies are therefore indispensable. Nowadays, when the usual course of a STA is followed, forensic samples will initially be directed to immunochemical techniques, including noninstrumental on-site and instrumental formats, e.g., enzyme immunoassays, typically enzyme multiplied immunotechniques (EMIT) and fluorescence polarization immunoassays.2,3 These preliminary immunochemical screening procedures mainly concern rapid-response analytical tools providing a binary yes/no response, which indicates whether the target analytes are present above a preset concentration threshold.4 In a next stage, the samples that provided a “yes” response to one or more compound classes or target analytes are subjected to a confirmation method that is at least as sensitive as the screening test and that provides a higher level of confidence in the result. Obviously, this two-step strategy can only be followed provided that first the drugs or poisons to be determined are specified in one way or the other, e.g., a “black list” dictated by law, and second that the particular immunoassays are available on the market. If these demands are not met, the screening strategy must be more extensive, considering the huge amount of potentially hazardous drugs and poisons.5 Most often, STA is performed by GC/MS6,7 and traditional HPLC-DAD8,9 methods. Several reviews on GC and LC techniques used to screen (1) Marquet, P. Ther. Drug Monit. 2002, 24, 125-133. (2) Ferrara, S. D.; Tedeschi, L.; Frison, G.; Brusini, G.; Castagna, F.; Benardelli, B.; Soregaroli, D. J. Anal. Toxicol. 1994, 18, 278-291. (3) Hino, Y.; Ojanpera¨, I.; Rasanen, I.; Vuori, E. Forensic Sci. Int. 2003, 148155. (4) Valca´rcel, M.; Ca´rdenas, S.; Gallego, M. Trends Anal. Chem. 1999, 18 (11), 685-694. (5) Pfleger, K.; Maurer, H. H.; Weber, A. Mass Spectral and GC Data of Drugs, Poisons, Pesticides, Pollutants and their Metabolites; Wiley: Weinheim, 1999. (6) Maurer, H. H. J. Chromatogr. 1992, 580, 3-41.

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 6365

biological specimens for the presence of drugs regarding STA have recently been published.10-14 GC/MS is most often used. However, despite its many advantages, GC/MS does have limitations in terms of the scope of analytes amenable to analysis. HPLCDAD alleviates many of the chromatographic constraints of GC, although at the cost of separation efficiency and most of all detection power. DAD clearly is not as specific as MS and can fail to detect molecules with no or little UV absorbance. All of these considerations have led to the exploration of LCMS. In the past decade, the number of LC-MS applications in forensic toxicology has increased remarkably.15 Regarding STA, several attempts have been made to render LC-MS suitable to this purpose.1,16 Review of the literature published during the last 5 years regarding STA using LC-MS, shows that three different strategies can be followed to develop an efficient LC-MS GUS procedure. One approach is single mass spectrometry, whether or not coupled to in-source collision-induced dissociation (CID).17,18 In general, in single MS mode, quadrupole-based mass spectrometry is used. Only Gergov et al.19 have made an attempt to use LC-time-of-flight (TOF) for a GUS procedure, based on the higher mass resolution of TOF detectors. Ion trap MS has apparently not been used in the single MS mode for STA. Despite the fact that up until now single MS has been extensively investigated as a possible technique to develop a GUS procedure, it still has major drawbacks. MS spectra are produced that are not resolved from background ions or, even worse, from ions originating from an overlapping chromatographic peak. In addition, poor interlaboratory reproducibility of in-source CID mass spectra with both electrospray ionization (ESI) and atmospheric pressure chemical ionization sources has been reported.20 Alternatively, the choice is tandem mass spectrometry (MS/MS)21,22 or MS/MS with information-dependent acquisition (IDA).23,24 Tandem mass spectrometry produces very clean spectra, devoid of contaminant ions and thus easily amenable to library searching. However, it requires beforehand knowledge of the expected compounds in order to (7) Polettini, A.; Groppi, A.; Vignali, C.; Montagna, M. J. Chromatogr., B 1998, 713 (1), 265-279. (8) Lambert, W. E.; Van Bocxlaer, J. F.; De Leenheer, A. P. J. Chromatogr., B 1997, 689 (1), 45-53. (9) Gaillard, Y.; Pepin, G. J. Chromatogr., A 1997, 763, 146-163. (10) Maurer, H. H. J. Chromatogr., B 1999, 733, 3-25. (11) Drummer, O. H. J. Chromatogr., B 1999, 733, 27-45. (12) Polettini, A. J. Chromatogr., B 1999, 733, 47-63. (13) Maurer, H. H. Comb. Chem. High Throughput Screening 2000, 3, 467-480. (14) Drummer, O. H.; Gerostamoulos, J. Ther. Drug Monit. 2002, 24 (2), 199209. (15) Marquet, P. Ther. Drug Monit. 2002, 24, 255-276. (16) Clauwaert, K. M.; Van Bocxlaer, J. F.; Lambert, W. E.; Van den Eeckhout, E. G.; Lemiere, F.; Esmans, E. L.; De Leenheer, A. P. Anal. Chem. 1998, 70, 2236-2344. (17) Marquet, P.; Venisse, N.; Lacassie, E.; Lachaˆtre, G. Analusis 2000, 925934A. (18) Venisse, N.; Marquet, P.; Duchoslav, E.; Dupuy, J. L.; Lachaˆtre, G. J. Anal. Toxicol. 2003, 27, 1-8. (19) Gergov, M.; Boucher, B.; Ojanpera¨, I.; Vuori, E. Rapid Commun. Mass Spectrom. 2001, 15, 521-526. (20) Bogusz, M. J.; Maier, R. D.; Kru ¨ ger, K. D.; Webb, K. S.; Romeril, J.; Miller, M. L. J. Chromatogr., A 1999, 844, 409-418. (21) Weinmann, W.; Svoboda, M. J. Anal. Toxicol. 1998, 22, 319-328. (22) Gergov, M. Robson, J. N.; Duchoslav, E.; Ojanpera¨, I. J. Mass Spectrom. 2000, 35, 912-918. (23) Fitzgerald, R. L.; Rivera, J. D.; Herold, D. A. Clin. Chem. 1999, 45 (8), 1224-1234. (24) Decaestecker, T. N.; Clauwaert, K. M.; Van Bocxlaer, J. F.; Lambert, W. E.; Van den Eeckhout, E. G.; Van Peteghem, C. H.; De Leenheer, A. P. Rapid Commun. Mass Spectrom. 2000, 14, 1787-1792.

6366

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

apply classic multiple-reaction monitoring (MRM) or product ion scan methods. The latter is not compatible with the objective of a GUS technique, i.e., the detection of unknowns. Therefore, we present here a complete LC-MS strategy to circumvent these limitations. The procedure is based upon IDA, an artificial intelligence-based product ion scan mode. This laboratory introduced as first, in a preliminary communication, the concept of IDA-based LC-MS/MS drug screening using a Q-TOF.24 A Q-TOF is especially suited for this purpose in view of its nonscanning thus flash mass analysis capability. In contrast to triple quadrupole instruments and ion trap instruments, it does not scan the mass range one mass at a time but allows one to achieve continuous production and simultaneous detection of ions across the full mass range. By performing automatic “on-the-fly” MS to MS/MS switching, even simultaneously for up to three or four coeluting precursor molecules, a complete set of qualitative and semiquantitative data could be acquired. Major advantages of this autoadaptive MS/MS product ion scan mode are its high specificity and selectivity, as the spectra recorded derive from a single precursor ion. Additionally, advance precursor ion knowledge about the xenobiotics that could be responsible for a certain intoxication and for which a product ion spectrum must be obtained is no longer necessary. The present study presents the complete analytical approach and the application of an IDA-mediated LC-MS/MS screening methodology in full extent, qualitatively as well as quantitatively. We present here the potential and pitfalls of the practical application in real biological (forensic) samples of IDA by benchmarking its performance against classic GUS approaches. EXPERIMENTAL SECTION Materials. The compounds studied for optimization of mass spectrometric parameters, LOD determinations and quantitative evaluation were morphine, benzoylecgonine, methylenedioxymethamphetamine (MDMA), codeine, strychnine, ethylmorphine, nalorphine, cocaine, lidocaine, bromazepam, methaqualone, diazepam, triazolam, methadone, trazodone, haloperidol, oxazepam, and butorphanol (internal standard). This representative test set was made from drug standards from different sources, available from our laboratories’ collection. Methanol and acetonitrile were all of HPLC grade (Biosolve, Valkenswaard, The Netherlands). High-purity water was provided from a Synergy 185 system (Millipore Corp., Bedford, MA). Acetic acid (purity minimum 99.7%) and ammonium acetate (purity minimum 98%) were supplied by Sigma-Aldrich (Steinheim, Germany), and ammonia solution 25% was purchased from Merck-Eurolab (Leuven, Belgium). The apolar C8 solid-phase extraction (SPE) columns (sorbent mass 100 mg, 1 mL) were provided by International Sorbent Technology (IST, Hengoed, Mid Glamorgan, U.K.). Blood samples were collected from autopsies and used after consent. The procedure does not require any specific safety measures, besides the normal precautions to be taken when working with biological (whole) blood samples, in particular when drug abuse is involved. Instrumentation. Reproducible, automated SPE was performed on a Zymark RapidTrace Solid-Phase Extraction Workstation (Zymark, Hopkington, MA) equipped with one singleextraction module. The IDA experiments were performed on a Waters Alliance 2790 separation module integrated with a Q-TOF instrument (Waters, Milford, MA).

Chromatographic Conditions. Chromatography was conducted on an Xterra MS C18 column (100 × 2.1 mm, 3.5-µm particle size; Waters). The flow rate was set to 0.3 mL/min, and the column oven temperature was held at 40 °C. Gradient elution was performed, starting at 100% of a mixture of water/methanol/ acetonitrile (80:10:10, v/v) containing 5 mM ammonium acetate (solvent A), programmed linearly, within 7 min, to 50% of a mixture of water/methanol/acetonitrile (20:40:40, v/v), again containing 5 mM ammonium acetate (solvent B), holding for 7 min. To remove late-eluting substances, a step gradient to 100% solvent B was included for 1.5 min. Subsequently, the system was programmed to regain its initial conditions over 0.5 min, followed by a 8-min reequilibration prior to the next injection. The injection volume was 25 µL, and the entire column effluent was directed into the mass spectrometer. Sample Preparation. An important obstacle that was encountered during field testing of the procedure was the difficult detection of signals of toxicological interest, among background noise, and the difficult setting of a switching intensity threshold24 due to the intense and, above all, highly variable background noise produced by extracts of (whole blood) biological samples. An appropriate sample cleanup proved highly important. An SPE method was fully optimized to suit the particular demands of an IDA screening strategy.25-27 Prior to the SPE cleanup procedure, the whole blood samples were pretreated as follows. After fortifying the blood samples with 50 µL of the internal standard (4 µg/mL), they were vortex mixed for 30 s, equilibrated, and ultrasonicated for 15 min. The blood samples were then diluted with 1 mL of 60 mM ammonium acetate buffer (pH 9.0), whereupon a fixed period of mixing, ultrasonicating, and centrifugation (2000g) followed. Before application of the supernatants obtained, the SPE C8 columns were conditioned with 3 mL of methanol and 3 mL of 60 mM ammonium acetate buffer (pH 9.0) at 1 mL/min. Samples were slowly applied at a rate of 0.5 mL/min. Interferences were washed off the columns with, successively, 5 mL of the ammonium acetate buffer containing 15% of methanol and 1 mL of water at 5 mL/min. Elution was performed with 1 mL of methanol containing 1% of acetic acid. Elution flow was set at 0.5 mL/min. The eluates were evaporated to dryness under a gentle stream of nitrogen, and the residues were redissolved in 200 µL of solvent A, 25 µL of which was injected into the LC-MS system. Mass Spectrometric Analysis and IDA. Detection of the compounds was performed in the positive electrospray ionization (ESI+) mode using IDA, generating a survey scan, single MS spectra with molecular mass information, product ion spectra, and extracted ion fragmentograms (XIC). Reference 24 details the whole IDA procedure from a MS point of view and corresponding experimental settings. However, because of the use of whole blood for this study and as a consequence of the fully optimized SPE procedure, specifically for the benefit of IDA, some parameter settings needed readjustment. The maximum number of components eventually fragmented at the same time was reduced from (25) Decaestecker, T. N.; Coopman, E. M.; Van Peteghem, C. H.; Van Bocxlaer, J. F. J. Chromatogr., B 2003, 789, 19-25. (26) Decaestecker, T. N.; Lambert, W. E.; Van Peteghem, C. H.; Deforce, D.; Van Bocxlaer, J. F. Abstracts of HTC-8, February 4-6, Bruges, Belgium, 2004. (27) Decaestecker, T. N.; Lambert, W. E.; Van Peteghem, C. H.; Deforce, D.; Van Bocxlaer, J. F. J. Chromatogr., A, in press.

four to three, since the fourth MS/MS channel was rarely used. As such, the blind spot, created in the MS trace, could be limited to 9 s instead of 12 s, i.e., a 20% fraction of the average LC peak width. An almost instant switch back to the MS mode was ensured. The influence of the switching intensity threshold also needed more carefully tuning. To that end, two threshold values have been compared, i.e., 400 and 100 counts/s. Finally, by including a collision energy profile in the IDA strategy, two MS/MS spectra were recorded per compound, one at low collision energy (15 eV), and another at a higher setting (25 eV). Mass Spectral Library. The generation of the mass spectral library was performed using the library software tool, a feature included in the MassLynx software. The library was created by introducing solutions of over 300 drugs and toxicants, available from our laboratories’ collection, into the electrospray ion source of the instrument in either one of the following ways, depending upon the experiment: (a) by flow injection directly into the source or (b) by actual LC-MS(/MS) analysis (Xterra MS C18 column). The first approach was used to generate MS/MS spectra at five different collision energies (15, 20, 25, 30, and 35 eV), while relative retention data were gathered by running a mixture of each compound with the internal standard (butorphanol) via the second approach. As such, a library was constructed to include, per compound, five MS/MS spectra, the molecular formula, and its molecular mass, together with the relative retention time. The mass spectral library comprises a total of 1600 MS/MS spectra. Switching Intensity Threshold and Limit of Detection. Since the limit of detection (LOD) is inevitably related to the switching intensity threshold, it cannot be determined, for example, as stated by IUPAC.28 A compound was considered as detected when its precursor ion signal exceeded the switching intensity threshold and thus its MS/MS spectra were recorded. Consequently, within our analytical system, LOD is defined as the lowest concentration present in a biological sample which still allows detection, i.e., the generation of a product ion spectrum. The LOD was determined in spiked whole blood samples for 17 drugs and toxicants. To that end, a set of spiked whole blood samples of decreasing concentration was injected for all drugs. Analyses were performed in 5-fold. Quantitative Performance. Calibration standards in whole blood were prepared by spiking stock solutions of drug mixtures (of the above 17 drugs) in a concentration range of 0.08-16 µg/ mL, resulting in a set of calibrators with the following concentrations: 4.0, 8.0, 16.0, 40.0, 60.0, 80.0, 160.0, 400.0, 600.0, and 800.0 ng/mL blood. These calibrators were analyzed using our IDA procedure, and the results were used to construct calibration curves based on the relationship between concentration and peak area ratio versus the internal standard for the extracted ion fragmentograms of the respective precursor ions. Qualitative Benchmarking. To evaluate the qualitative characteristics of the IDA approach, 20 whole blood samples were selected for this benchmarking study. As it was our intention to assess the IDA screening technique within a systematic toxicological analysis setting, the procedure was compared to the conventional GUS approach consisting of a combination of several methods, including EMIT, RIA, HPLC-DAD,8 GC/MS, and GC(28) Thompson, M.; Ellison, S. L. R.; Wood, R. Pure Appl. Chem. 2002, 74 (5), 835-855.


6367

NPD. Two established toxicological laboratories participated in this part of the study, which was run in double-blind conditions. Since the standard operating procedures of the participating laboratories consisted of analyzing urine samples prior to blood samples, due to the relatively higher amount of drugs present, a urine sample was provided in 18 out of 20 cases and a guiding analysis of urine was permitted. Application of the IDA-mediated GUS procedure does not require such a preanalysis. Quantitative Benchmarking. After the qualitative evaluation, a quantitative exploration of the IDA approach was pursued. Out of the above 20 cases, 7 whole blood samples were selected for this investigation. The above-constructed calibration curves were used to quantify the particular drug(s) found in the different benchmarking samples. The laboratory of Toxicology and Special Chemistry from the Clinical University St.-Luc (Brussels, Belgium) was prepared to act as reference laboratory. Their instrumentation consisted of a Quattro micro tandem mass spectrometer (Waters), operated in ESI+ mode and coupled to a Waters 2795 Alliance HPLC system. Three different extraction procedures together with eight different MRM methods have been applied to quantitate the whole blood samples. The MRM methods included procedures to determine successively opiates, amphetamines, cocaine, methadone, and its metabolite EDDP, antidepressants (2 methods), benzodiazepines, and one method for atropine, noscapine, and papaverine.29,30 RESULTS AND DISCUSSION Assay Performance. (1) Solid-Phase Extraction. The developed SPE procedure fully meets the criteria of a LC-MSbased GUS procedure using IDA, as a compromise had been reached in that the substances of interest were isolated at a yield as high as possible and the interfering substances from the biological matrix were removed. The optimization and complete characteristics of this extraction procedure by means of experimental design have extensively been reported in a separate paper.27 (2) Mass Spectrometric Analysis and IDA. The information, gathered by the IDA approach, as well as the efficient and straightforward way in which this information can be retrieved from the data set, is clearly illustrated in Figure 1. The left side of the figure describes, top to bottom, the MS total ion chromatogram (TIC) or survey chromatogram and the MS extracted ion chromatogram of methadone (here taken as an example), its molecular structure, and the MS/MS spectra (recorded at 15 and 25 eV). The TICs, obtained in the MS/MS mode (3 channels) are displayed on the right side of the figure. In the MS/MS mode, a TIC consists of a number of points, at each of which two MS/ MS spectra were recorded. The software connects the different points with one another through straight lines, and as such, strange-shaped TICs are created. Jumping from inflection point to inflection point provides easy access to relevant points of the chromatographic run.24 All this information was acquired in one single injection. (3) Mass Spectral Library Search Performance. The mass spectral library was successfully used to identify the “unknowns”. (29) Wallemacq, P. E.; Vanbinst, R. SQBC-CSCC Joint Meeting, June 1-5, Que´bec, Canada, 2003. (30) Wallemacq, P. E.; Vanbinst, R.; Asta, S.; Cooper, D. P. Clin. Chem. Lab. Med. 2003, 41 (7), 921-925.

6368


The library search process proved to be specific enough to lead to accurate and coherent results. All compounds appearing in the different toxicological blood samples were positively identified by the library search, except for one. LC-MS/MS libraries have long been a source of discussion in the field of LC-MS GUS procedures. LC-MS/MS spectra were considered too variable to be able to create universal libraries, thus making a library only valid for the instrument it was created with. Very recently, Gergov et al. have presented an interesting report regarding the exchangeability of MS/MS libraries.31 They showed that spectral data exchange between instruments is possible. Taking this into consideration, the MS/MS spectra of the only unidentified peak have been matched, in the form of a peak list, to the library created by Gergov et al. in their analytical environment. A positive identification of the compound as cetirizine (a substance not present in our library) was obtained. Clearly, this proves the potential of the library searching within LC-MS and the power of IDA-based LC-MS/MS GUS, in a broader perspective. The availability of five different library entries was deemed necessary, since variations in the MS/MS spectra were nevertheless observed. This can be attributed to instrumental variations. Although generally regarded as independent from each other, certainly on a theoretical basis, ionization conditions, e.g., variations between biological samples, does seem to influence the product ions formed through CID. Variations in the MS/MS spectra as observed could be related to dirty extracts with coeluting compounds, despite the “filtering” effect of the first quadrupole. (4) Switching Intensity Threshold and LOD. Table 1 gives an overview of the results of the LOD determinations in whole blood for the test panel of 17 drugs, with the switching intensity threshold set to 100 and 400 counts/s. As can be seen, setting the threshold value to 100 counts/s results in a significant sensitivity gain. On the other hand, setting a low switching intensity threshold will increase the number of interferences detected, which will, as a consequence, complicate the interpretation of the data set. Figure 2 illustrates the effect of the switching intensity threshold value on the total number of precursor ions tagged by IDA for CID. This analysis represents a real blood sample. Obviously, the whole set of MS/MS events includes those for the toxicologically relevant compounds present as well as those originating from interference-related ions. By setting the switching intensity threshold to a lower value, the MS/MS channels will increasingly be “occupied”. Although decreasing the LOD, theoretically this also increases the chance that all channels are at a given point in time occupied by interference-derived ions, allowing important compounds to pass by unnoticed. (5) Quantitative Performance. It is clear, however, that the more time spent to record MS/MS spectra and identify unknowns, the less time is left to record the MS survey chromatogram. As a consequence, the peak shape of the XICs can easily be deformed, as shown in Figure 3. Here, the XICs of benzoylecgonine are shown. The upper one represents the XIC in the case where no switching occurs and consequently no blind spots are generated in the peak slope. In the other cases, a lower and higher threshold value, respectively, was used. Each dot represents a data point describing the peak curvature. (31) Gergov, M.; Weinmann, W.; Meriluoto, J.; Uusitalo, J.; Ojanpera¨, I. Rapid Commun. Mass Spectrom. 2004, 18, 1039-1046.

Figure 1. TIC in the MS mode (A), XIC of methadone (B), 2 MS/MS spectra (C, D) of methadone, and TICs of the three MS/MS channels (E).

As such, the switching intensity threshold indirectly determines the peak shapes obtained in the XICs, which undoubtedly affects the quantitative characteristics of this IDA approach. Additionally, while being an excellent instrument for qualitative purposes in many domains, a Q-TOF as such is clearly less performing from a quantitative point of view. To evaluate all of this, 10 calibrators of a mixture of 17 drugs have been prepared in whole blood. These calibrators have been injected using three different switching intensity threshold settings. For the first and second MS methods, the threshold was respectively set to 100 and 400 counts/s, while for the third, no switching was allowed. As such, a MS chromatogram, free of blind spots, was recorded,

which could act as a 100% reference. Depending on the LOD of the compounds, the number of calibrators taken into account varied per compound. Table 2 shows the average, taken of all calibrators, of the normalized deviation of the peak area (normalized against the 100% reference) for both thresholds. Significantly higher deviations are obtained when a lower threshold is applied. As can be deduced from this table, for (semi-)quantitative purposes, the use of a higher threshold is preferred. In a next phase, calibration curves for these 17 drugs were constructed in whole blood. Based on a visual examination of the curves and analysis of the residuals, quadratic regression curves, after logarithmic transformation of both axes, gave the best fit Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

6369

Figure 2. Overview of the total number of switches performed for two different threshold settings. Each inflection point represents a MS to MS/MS switch event. A higher order MS/MS channel is only activated when at a given time more than one precursor ion exceeds the switching threshold. Table 1. LODs (in Whole Blood) Determined for 17 Drugs Applying Two Different Threshold Values

Table 2. Influence of MS/MS Threshold on the Peak Area

LOD (ng/mL)

morphine benzoylecgonine XTC codeine strychnine ethylmorphine nalorphine cocaine bromazepam methaqualone oxazepam triazolam haloperidol lidocaine methadone trazodone diazepam

% deviation

100 counts/s

400 counts/s

60 16 8 40 16 16 40 8 80 8 80 40 40 8 8 16 40

160 60 40 80 60 60 160 16 400 40 400 60 160 16 16 60 80

for all compounds. A calibration curve of methadone is depicted in Figure 4. An explanation for this nonlinearity can probably be found in the limited linear dynamic range of a time-of-flight instrument. Accordingly to Guilhaus, this can be attributed to a limited upper dynamic range of the time-to-digital converter, the ion counting device used of the TOF.32 By trying to calibrate on a large calibration interval, mandatory in, for example, forensic toxicology where concentration of the compounds can vary enormously in a single sample, the underlying physical process (32) Guilhaus, M. J. Mass Spectrom. 1995, 30, 1519-1532.

6370 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

morphine benzoylecgonine XTC codeine strychnine ethylmorphine nalorphine cocaine bromazepam methaqualone oxazepam triazolam haloperidol lidocaine methadone trazodone diazepam

100 counts/s

400 counts/s

9.51 36.01 20.82 17.43 41.45 15.57 17.08 50.10 24.96 54.98 21.32 32.37 11.35 12.61 32.38 27.30 24.44

8.30 14.64 8.45 7.88 5.26 6.61 5.54 6.14 22.06 11.34 12.79 12.81 17.34 9.79 14.42 12.52 11.23

is clearly ill-represented by a linear function. Fitting a linear function thus produces worse results compared to using a more complicated higher order function. Taking everything into consideration regarding the quantitative nature of our procedure, it should be concluded that the quantitative results are additional to the valuable qualitative information obtained. This fits perfectly into the picture of screening analytical approaches. They all provide a qualitative result directing further analysis, often qualitative confirmation and certainly full quantitation, and generally also provide a first idea of the quantity of toxicant present. The latter should be confirmed in a dedicated

Figure 3. A zoomed-in region of the chromatogram (XIC constructed for benzoylecgonine) at different settings of the switching threshold illustrating potential peak shape deformation observed with IDA.

Figure 4. Calibration curve of methadone after logarithmic transformation of both axes.

quantitative analysis but also provides a quick reference to the severity of an intoxication, for example, in emergency toxicology. They level with those quantitative data as obtained from most immunoassays and should therefore be labeled as semiquantitative. Benchmarking. (1) Qualitative Benchmarking. Twenty whole blood samples of autopsy cases and of drug addicts were analyzed by the IDA LC-MS/MS method and by the conventional GUS techniques (Table 3). In all cases, the LC-MS screening was benchmarked against the results obtained from a comprehensive STA, including EMIT, RIA, HPLC-DAD,8 GC/MS, and

GC-NPD techniques. Again, a comparison has been made between switching intensity threshold values of 100 and 400 counts/s. By analyzing Table 3, one can conclude that, from a qualitative point of view, a threshold setting of 100 counts/s is more ideal. Five caffeine findings by EMIT and four cotinine findings by RIA were missed by this IDA-based screening. This can be ascribed to the lower LODs of the dedicated immunoassays for these compounds; moreover, the other chromatographic screening techniques have also missed these compounds, except for one (case 8). In three cases, benzodiazepines, namely, bromazepam and diazepam (cases 1, 5, and 20), were missed, most likely due to the relatively high Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

6371

Table 3. Qualitative Benchmarking of IDA Approacha IDA LC-MS/MS no. 1

2

3

4

STA

400 counts/s

nd

ranitidine

ranitidine

caffeine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) cotinine (RIA) benzodiazepines (EMIT) bromazepam (HPLC-DAD) desalkylflurazepam (HPLC-DAD) quinine (HPLC-DAD) nd

caffeine

caffeine

ndb

nd

nd desalkylflurazepam quinine lidocaine

nd nd

nd nd caffeine (EMIT) tricyclic antidepressants (EMIT)

propranolol ephedrine nd

propranolol ephedrine nd

nd nd nd caffeine (EMIT) nd tricyclic antidepressants (EMIT) nd nd caffeine (EMIT/GC-MS/ GC-NPD) cotinine (RIA/GC-MS/GC-NPD) cocaine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) nd nd

amitriptyline nortriptyline midazolam nd ranitidine

amitriptyline nortriptyline midazolam nd ranitidine

amitriptyline midazolam caffeine

nd midazolam caffeine

cotinine cocaine

cotinine cocaine

cocaethylene benzoylecgonine nd

cocaethylene benzoylecgonine nd

methadone MDMA

methadone MDMA

nd caffeine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) cotinine (RIA/GC-MS/ GC-NPD) benzodiazepines (EMIT) nd

ranitidine caffeine

nd nd

cotinine

cotinine

nordazepam

nd

diazepam (HPLC-DAD)

nd

nd

nd nd

trazodone piroxicam

trazodone piroxicam

16

caffeine (EMIT) cotinine (RIA) amfetamines (EMIT)

nd cotinine

nd cotinine

17

nd

MDMA

MDMA

18

methylecgonine (GC-MS/ GC-NPD) nd nd 5

6

IDA LC-MS/MS

100 counts/s

quinine lidocaine

no. 8

9

10 11

12

13

14

15

100 counts/s

400 counts/s

caffeine (EMIT/GC-MS/ GC-NPD) cotinine (RIA/GC/MS/ GC-NPD) benzodiazepines (EMIT) nd diazepam (HPLC-DAD) nd

nd

nd

cotinine

cotinine

nordazepam diazepam nordazepam

nd diazepam nd

nd caffeine (EMIT/GC/MS/ GC-NPD) cotinine (RIA) olanzapine (HPLC-DAD) nd caffeine (EMIT/HPLC-DAD/ GC/MS/GC-NPD) cotinine (RIA) caffeine (EMIT) cotinine (RIA) nd nd nd cotinine (RIA) bromazepam (HPLC-DAD) nd

atropine caffeine

atropine caffeine

nd olanzapine alprazolam caffeine

nd olanzapine alprazolam caffeine

nd caffeine nd lormetazepam lidocaine atropine cotinine bromazepam lidocaine

nd caffeine nd nd lidocaine atropine cotinine bromazepam lidocaine

nd nd

atropine lormetazepam

atropine nd

cotinine (RIA) tricyclic antidepressants (EMIT) nd

cotinine nd

cotinine nd

benzoylecgonine

nd

nd caffeine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) trazodone (HPLC-DAD) thioridazine (HPLC-DAD)

cetirizine caffeine

cetirizine caffeine

trazodone thioridazine

trazodone thioridazine

thioridazine metabolite atropine caffeine

thioridazine metabolite atropine caffeine

venlafaxine

venlafaxine

olanzapine caffeine

olanzapine caffeine

cotinine nd chlorpheniramine caffeine

cotinine nd chlorpheniramine caffeine

cotinine

cotinine

nd nd

nd nd

thioridazine

thioridazine thioridazine metabolite trazodone atropine benzoylecgonine cocaine methadone MDMA cocaethylene nd nd

thioridazine metabolite (HPLC-DAD) nd caffeine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) venlafaxine (GC-MS/ GC-NPD) nd caffeine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) cotinine (RIA) caffeine (EMIT) nd

19

caffeine (EMIT/HPLC-DAD/ GC-MS/GC-NPD) cotinine (RIA/GC-MS/ GC-NPD) nicotine (GC/MS/GC-NPD) tricyclic antidepressants (EMIT) thioridazine (HPLC-DAD)

20

thioridazine metabolite (HPLC-DAD) trazodone (GC/MS) nd benzoylecgonine (GC/MS)

benzodiazepines (EMIT)

7

STA

nordazepam (HPLC-DAD) diazepam (HPLC-DAD)

nordazepam diazepam

nordazepam diazepam

caffeine (EMIT/GC-MS/ GC-NPD) cotinine (RIA)

caffeine

caffeine

cotinine

cotinine

morphine (RIA) benzodiazepines (EMIT) bromazepam (HPLC-DAD)

nd

morphine

bromazepam

bromazepam

nordazepam (HPLC-DAD/ GC-MS/GC-NPD) nd nd nd nd nd benzoylecgonine (GC-MS/ GC-NPD)

nordazepam

nordazepam

cocaine (GC/MS)

thioridazine metabolite trazodone atropine benzoylecgonine cocaine

diazepam oxazepam noscapine codeine papaverine benzoylecgonine

diazepam nd nd nd papaverine benzoylecgonine

methadone (GC/MS) MDMA (GC/MS) nd methylecgonine (GC/MS) bromazepam (GC/MS)

methadone MDMA cocaethylene nd nd

a STA comprises EMIT, RIA, HPLC-DAD,7 GC/MS, and GC-NPD analyses. EMIT analyses were performed to detect the following compound classes or individual representatives: amphetamines, cocaine, opiates, methaqualone, methadone, propoxyphene, phencyclidine, caffeine, benzodiazepines, tricyclic antidepressants, digoxine. RIAs detected LSD, morphine, fentanyl, cotinine, digoxine and buprenorphine. b nd, not detected.

6372 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

limits of detection observed for this group of compounds. Clearly, the developed procedure touches upon its limitations in the field of benzodiazepine detection. When a higher switching intensity threshold is applied, the number of false negatives increases correspondingly. On the other hand, a number of drugs have been detected by the IDA approach that were not by any of the other participants reported, including ranitidine (cases 1, 3, and 5), propranolol (case 1), piroxicam (case 5), papaverine (case 7), atropine (cases 8, 11, 12, 14, and 19), the benzodiazepines midazolam (cases 2 and 3) and alprazolam (case 9), and several others. This illustrates one of the major advantages of the IDA approach, i.e., the potential to detect all of the basic and neutral compounds present, not being restricted to a limited number of selected compounds screened for, common in traditional LC-MS/MS screening methods. Its advantage over traditional GC/MS screening procedure lies in its robustness and the fact that the obtained information content is much higher. This proves very advantageous for the identification of unknowns. In that respect, the major benefit comes from the uncomplicated nature of the product ion spectra one obtains. In contrast to GC/MS they are devoid of any interfering ions and thus easily interpretable or library searchable. (2) Quantitative Benchmarking. To evaluate the (semi-)quantitative performance of the IDA screening strategy, 7 samples were selected out of the above 20 samples. Results of this quantitative benchmarking study are summarized in Table 4. Comparison of the quantitative results obtained by both IDA and MRM confirms the semiquantitative potential of the IDA procedure. Nevertheless, the results compare favorably in the majority of the cases. On close inspection, a distinct difference in quantitative results was observed for cocaine and benzoylecgonine. This is attributed to the difference in time on which the IDA and MRM analyses were performed. Cocaine is indeed subject to postmortem breakdown to its metabolite, benzoylecgonine.33,34 As mentioned, our procedure equals the performance characteristics of most screening procedures. A good indication of the degree of intoxication is provided, but a follow-up dedicated quantitative analysis is often required. Of course, the latter is also governed by the clinical application or goal. CONCLUSIONS By assessing the potential and pitfalls of the IDA-mediated LCMS/MS screening approach, its feasibility within a forensic toxicological setting was demonstrated. Regarding its qualitative characteristics, it can be stated that the strategy performed very well and that it can be considered as a valuable alternative to the traditional GUS procedures of drugs and toxic compounds. When (33) Moriya, F.; Hashimoto, Y. J. Forensic Sci. 1996, 41, 612-616. (34) Warner, A.; Norman, A. B. Ther. Drug Monit. 2000, 22, 266-270.

Table 4. Quantitative Benchmarking of IDA Approacha concn (ng/mL) no. 1

2

3 4 5 6 7 a

methadone MDMA cocaine benzoylecgonine bromazepam nordazepam diazepam morphine papaverine benzoylecgonine lidocaine atropine atropine trazodone diazepam methadone benzoylecgonine methadone

IDA

MRM

99 1532 545 1795 1235 1749 57 281 20 382 30 30 23 173 110 121 148 20

70 1479 173 2706 1278 3086 47 250 12 49 ND 15 18 132 77 123 130 24

nd, not detected; MDMA, methylenedioxymethamphetamine.

no semiquantitative indication is required, the switching intensity threshold can be set relatively low, and as a consequence, almost all of the drugs detected by the conventional techniques were identified, as well as additional drugs that were not previously reported, and this by applying one single method. However, as is customary in forensic toxicology, before accurate intoxication levels are determined using targeted analytical analysis, semiquantitative results are required. To that end, an increase of the switching intensity threshold was inevitable. Because of this a loss of sensitivity was registered, but for forensic toxicological screening purposes still acceptable LODs were obtained. Therefore a higher switching intensity threshold was preferred. Finally, its semiquantitative potential was benchmarked against MRM methods. It proved to be more than suitable within the application field of toxicological analysis. ACKNOWLEDGMENT We gratefully thank Mr. Wim Goeteyn and Ing. Sofie Vande Casteele for their practical assistance in performing respectively the solid-phase extractions and the mass spectrometric analyses. This work was supported by Grant GOA99-120501.99 (Bijzonder OnderZoeksFonds) and Grant FWO 1.5.097.99 of the FSRFlanders (FWO-Vlaanderen). Received for review May 26, 2004. Accepted August 3, 2004. AC0492315


6373

MS Screening

Recommend Documents