Validated Method for the Determination of the Ethanol Consumption

Jul 18, 2006 - 4-Chloro-6-pyrimidinylferrocene modified silica gel: A novel multiple-function stationary phase for mixed-mode chromatography ... A ful...
0 downloads 10 Views 188KB Size
Anal. Chem. 2006, 78, 5884-5892

Validated Method for the Determination of the Ethanol Consumption Markers Ethyl Glucuronide, Ethyl Phosphate, and Ethyl Sulfate in Human Urine by Reversed-Phase/Weak Anion Exchange Liquid Chromatography-Tandem Mass Spectrometry Wolfgang Bicker,† Michael La 1 mmerhofer,*,† Thomas Keller,‡ Rainer Schuhmacher,§ Rudolf Krska,§ and Wolfgang Lindner†

Christian Doppler Laboratory for Molecular Recognition Materials, Department of Analytical and Food Chemistry, University of Vienna, Wa¨hringer Strasse 38, A-1090 Vienna, Austria, IFFB Department of Forensic Medicine and Forensic Neuropsychiatry, University of Salzburg, Ignaz-Harrer-Strasse 79, A-5020 Salzburg, Austria, and Christian Doppler Laboratory for Mycotoxin Research, Department IFA Tulln, University of Natural Resources and Applied Life Sciences, Konrad-Lorenz-Strasse 20, A-3430 Tulln, Austria

A liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for the determination of the ethanol consumption markers ethyl glucuronide (EtG), ethyl phosphate (EtP), and ethyl sulfate (EtS) in human urine was developed. A reversed-phase/weak anion exchange type stationary phase demonstrated particular suitability for the analysis of these highly polar acids. Sample preparation was minimized to centrifugation and dilution of urine prior to injection. The method was validated in the range of 5-750 µg‚L-1 with 1:20 and 1:1000 diluted urine, which corresponds to actual concentration ranges of 0.1-15 and 5-750 mg‚L-1 in undiluted samples. Method validation was carried out using six different lots of human urine. Over the entire calibration range intraday and interday precision (each n ) 5, three concentration levels per dilution factor) adopted values between 0.6 and 4.7% and 0.8 and 12.1%, relative standard deviation, respectively. Corresponding accuracy values ranged between 94.2 and 113.5% and 86.6 and 110.9%, respectively. Matrix effects (absolute/relative) were found to be present in minor extent (∼ -30% to +15% MS/MS signal alterations) and were well corrected by the employed isotopically labeled internal standards. The validated assay was applied to urine samples of a drinking study as well as postmortem specimens. It was possible to assess the principal potential of EtP as ethanol consumption marker. Elevated concentration levels were found in real samples; however, EtP seems to be less sensitive compared to the previously known ethanol phase II conjugates EtG and EtS.

* Corresponding author. Tel.: +43-1-4277-52323. Fax: +43-1-4277-9523. E-mail: [email protected]. † University of Vienna. ‡ University of Salzburg. § University of Natural Resources and Applied Life Sciences.

5884 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

Alcohol abuse is a sociological phenomenon all over the world, and the research efforts on ethanol consumption markers are still a growing field in forensic toxicology.1-3 Due to the potentially severe legal and personal consequences, two critical issues need to be addressed in this context: The first is the choice of the marker, and the second its reliable determination in the biological matrix. Ethanol phase II conjugates, viz. ethyl glucuronide (EtG) and ethyl sulfate (EtS) (Figure 1), attracted, due to their high specificity and sensitivity for indicating recent ethanol consumption, attention during the past years.3 Their concurrent analysis in urine can be regarded as an important advancement to provide security for confirming ethanol intake. However, while for example in a recent study, in ∼98% of clinical cases a good qualitative agreement between the two markers could be obtained, still in ∼2% only one of the conjugates could be identified.4 Hence, the determination of further ethanol consumption markers would be desirable for valid conclusions. An early study reported the formation of ethyl phosphate (EtP; Figure 1) in liver of rats pretreated with high doses of ethanol.5 The authors speculated that EtP may originate from ethanolysis of endogenous phosphate esters, which might qualify it as another marker for ethanol exposure. However, EtP was, except of recent preliminary data,6 not yet considered as an ethanol consumption marker in humans. Methods for both EtG and EtS can be found in the literature, based on either liquid chromatography-(tandem) mass spectrom(1) Musshoff, F. J. Chromatogr., B 2002, 781, 457-480. (2) Helander, A. J. Neural Transm. Suppl. 2003, 66, 15-32. (3) Wurst, F. M.; Tabakoff, B.; Alling, C.; Aradottir, S.; Wiesbeck, G. A.; MuellerSpahn, F.; Pragst, F.; Johnson, B.; Javors, M.; Ait-Daoud, N.; Skipper, G. E.; Spies, C.; Nachbar, Y.; Lesch, O.; Ramskogler, K.; Hartmann, S.; Wolfersdorf, M.; Dresen, S.; Weinmann, W.; Hines, L.; Kaiser, A.; Lu, R. B.; Ko, H. C.; Huang, S. Y.; Wang, T. J.; Wu, Y. S.; Whitfield, J.; Snell, L. D.; Wu, C.; Hoffman, P. L. Alcohol.: Clin. Exp. Res. 2005, 29, 1268-1275. (4) Helander, A.; Beck, O. J. Anal. Toxicol. 2005, 29, 270-274. (5) Tomaszewski, M.; Buchowicz, J. Biochem. J. 1972, 129, 183-186. (6) Halter, C.; Dresen, S.; Lauer, J.; Wurst, F. M.; Weinmann, W. T + K 2005, 72, 33 (poster abstract). 10.1021/ac060680+ CCC: $33.50

© 2006 American Chemical Society Published on Web 07/18/2006

and EtS experienced increasing popularity. The simultaneous determination of EtG and EtS was previously performed in reversed-phase (RP) mode using porous graphite4 or polarembedded RP phases.7 Adequate retention was achieved only under highly aqueous conditions,4,7 and postcolumn addition of an organic modifier was required to achieve adequate MS sensitivity.7 Yet, such a chromatography may not be selective enough and, hence, may lead to coelution of the only weakly retained analytes and matrix components which may cause MS signal adulterations.20 The objective of the present study was to provide a LC-MS/ MS method for the simultaneous determination of EtG, EtP, and EtS in human urine samples employing stable isotope-labeled internal standards (IS) for each solute. Method validation was carried out according to the guidelines developed by the U.S. Food and Drug Administration21 laying emphasis on the evaluation of matrix effects. The chromatography made use of a recently developed mixed-modal stationary phase with particular suitability for polar acidic solutes.22,23 By analyzing urine specimens from a drinking study and autopsied corpses, it was elucidated whether EtP is a valuable complement to EtG and EtS as urinary ethanol consumption marker.

Figure 1. Chemical structures of the ethanol consumption markers under investigation, viz. ethyl glucuronide (EtG), ethyl phosphate (EtP), and ethyl sulfate (EtS), together with stable isotope-labeled analogues, employed as IS.

etry (LC-MS(/MS)),4,7-12 LC with pulsed electrochemical detection,13 gas chromatography (GC)/MS,14-16 capillary electrophoresisindirect UV detection,17,18 or an enzyme-linked immunosorbent assay.19 On the other hand, no analytical method was as yet reported for EtP. With the advent of LC-MS(/MS) and the accompanied reduction in sample preparation expenditure, the analysis of EtG (7) Politi, L.; Morini, L.; Groppi, A.; Poloni, V.; Pozzi, F.; Polettini, A. Rapid Commun. Mass Spectrom. 2005, 19, 1321-1331. (8) Weinmann, W.; Schaefer, P.; Thierauf, A.; Schreiber, A.; Wurst, F. M. J. Am. Soc. Mass Spectrom. 2004, 15, 188-193. (9) Stephanson, N.; Dahl, H.; Helander, A.; Beck, O. Ther. Drug Monit. 2002, 24, 645-651. (10) Nishikawa, M.; Tsuchihashi, H.; Miki, A.; Katagi, M.; Schmitt, G.; Zimmer, H.; Keller, T.; Aderjan, R. J. Chromatogr., B 1999, 726, 105-110. (11) Dresen, S.; Weinmann, W.; Wurst, F. M. J. Am. Soc. Mass Spectrom. 2004, 15, 1644-1648. (12) Schneider, H.; Glatt, H. Biochem. J. 2004, 383, 543-549. (13) Kaushik, R.; LaCourse, W. R.; Levine, B. Anal. Chim. Acta 2006, 556, 267274. (14) Janda, I.; Alt, A. J. Chromatogr., B 2001, 758, 229-234. (15) Alt, A.; Wurst, F. M.; Seidl, S. Blutalkohol 1997, 34, 360-365. (16) Schmitt, G.; Aderjan, R.; Keller, T.; Wu, M. J. Anal. Toxicol. 1995, 19, 9194. (17) Krivankova, L.; Caslavska, J.; Malaskova, H.; Gebauer, P.; Thormann, W. J. Chromatogr., A 2005, 1081, 2-8. (18) Esteve-Turrillas, F. A.; Bicker, W.; Laemmerhofer, M.; Keller, T.; Lindner, W. Electrophoresis, in press. (19) Zimmer, H.; Schmitt, G.; Aderjan, R. J. Anal. Toxicol. 2002, 26, 11-16.

EXPERIMENTAL SECTION Reagents. EtG and d5-ethyl glucuronide. (d5-EtG) were purchased from Medichem World (Steinenbronn, Germany). EtS sodium salt was obtained from TCI Europe (Zwijndrecht, Belgium) and ethyl dichlorophosphate from Acros (Geel, Belgium). EtP and 18O2-ethyl phosphate (18O2-EtP) were prepared by hydrolysis of ethyl dichlorophosphate with either water or 18Owater (95 atom % 18O) (Aldrich, Vienna, Austria). The aqueous solutions were diluted to a final concentration of 10 g‚L-1 (EtP) and 2.8 g‚L-1 (18O2-EtP), respectively, and frozen at -20 °C for one week (start of validation study). d5-Ethyl sulfate (d5-EtS) was prepared by esterification of sulfuric acid with d5-ethanol (99.5 atom % D) (Aldrich).24 Mobile phase constituents, such as acetonitrile (ACN) (Fisher Scientific, Loughborough, UK), water (Aldrich), acetic acid (HAc) (Fluka, Buchs, Switzerland), aqueous ammonium hydroxide solution (Fluka), were of HPLC grade. Preparation of Stock and Working Solutions. Individual stock solutions of EtG, EtP, and EtS were prepared in water at 3000 mg‚L-1 and those of the IS, i.e., d5-EtG, 18O2-EtP, and d5-EtS at 1000 mg‚L-1. From the individual analyte stocks, multicomponent working solutions were prepared in water at eight concentration levels (1-150 mg‚L-1). A multicomponent aqueous IS spiking solution was prepared at 1.25 mg‚L-1. All solutions were aliquoted and stored at -20 °C. The number of freeze-and-thaw cycles per aliquot was a maximum of five. (20) Matuszewski, B. K.; Constanzer, M. L.; Chavez-Eng, C. M. Anal. Chem. 2003, 75, 3019-3030. (21) U.S. Food and Drug Administration, Guidance for Industry on Bioanalytical MethodValidation,May2001;http://www.fda.gov/cder/guidance/4252fnl.pdf (cited May, 31, 2006). (22) Bicker, W.; Laemmerhofer, M.; Genser, D.; Kiss, H.; Lindner, W. Toxicol. Lett. 2005, 159, 235-251. (23) Bicker, W.; Laemmerhofer, M.; Lindner, W. J. Chromatogr., B 2005, 822, 160-169. (24) Nalesso, A.; Frison, G.; Favretto, D.; Maietti, S.; Ferrara, S. D. Rapid Commun. Mass Spectrom. 2005, 19, 3612-3614.

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

5885

Preparation of Spiked Urine Samples. Blank spot urine was obtained from six healthy volunteers between 25 and 39 years of age (three male and female each; creatinine 8-14 mmol‚L-1 determined by Birkmayer Laboratories, Vienna, Austria) with ethanol abstinence of >72 h prior to collection. Pooled urine was prepared from the individual samples. Both the individual urine lots and the pooled urine were used for the validation study and stored at -20 °C. After thawing unassisted at room temperature, spiked urine samples were prepared freshly following a standard procedure: 50 µL of the respective multicomponent working solution was added to 450 µL of blank urine. After vortexing, 50 µL of the fortified urine was placed in a glass autosampler vial containing 50 µL of the multicomponent IS solution, and the mixture was diluted with 900 µL of ACN/water (50:50; v/v). These 1:20 diluted samples with a final concentration range of 5-750 µg‚L-1 (corresponds to a concentration range of 0.1-15 mg‚L-1 in undiluted urine) were then analyzed by LC-MS/MS. In a similar manner, blank urines (individual lots, pooled), diluted 1:50 with water, were processed as above, thereby obtaining calibration samples for the range 5-750 mg‚L-1 in undiluted urine (total dilution factor of 1000). Collection of Postmortem Urine Samples. Urine samples from 52 autopsy cases that gave a positive result in the routinely performed headspace GC-flame ionization detection screening for ethanol were analyzed for EtG, EtP, and EtS. All specimens were stored at -20 °C in tightly sealed polypropylene containers until analysis. Drinking Study. Two healthy volunteers (female, age 24 years, 62 kg; male, age 26 years, 85 kg) gave informed consent to participate in a pilot drinking experiment. A total of 0.2 g‚kg-1 body weight ethanol was consumed as sparkling wine (7.5% (v/ v) ethanol) within 15 min shortly after dinner. Immediately before and 2, 4, 6, 12, 18, and 24 h after the ethanol intake, the totally excreted amount of urine was measured and an aliquot each was collected for later determination of creatinine and ethanol consumption markers. During the collection time frame, any further ethanol intake was strictly avoided. Urine samples were stored for 4 days at -20 °C and then subjected to analysis. Sample Preparation for Real Samples. Real samples were thawed unassisted at room temperature and centrifuged (8000 rpm, 2 min) when necessary. Analysis was carried out in 1:20 dilution (50 µL of IS solution + 50 µL of urine + 900 µL of ACN/ water (50:50; v/v) in an autosampler vial), as well as in 1:1000 dilution (1:50 dilution, i.e., mixing of 20 µL of urine with 980 µL of water, preceding the 1:20 dilution step). Urine samples collected from the drinking study were solely analyzed in 1:20 dilution. LC-MS/MS Method. (1) Instrumentation. Analytical runs were performed on a LC-MS/MS system consisting of an Agilent (Vienna, Austria) HP1100 liquid chromatograph hyphenated with an Applied Biosystems/MDS Sciex (Thornhill, Canada) 4000 Q Trap and interfaced with a Turbo V ion spray source. Data processing was performed using the Analyst 1.4.1 software. (2) Liquid Chromatography. A mixed-modal RP/weak anion exchange (WAX) stationary phase based on N-(10-undecenoyl)3-aminoquinuclidine23,25,26 bonded to thiol-functionalized silica

particles (5 µm, 100 Å, selector density 360 µmol‚g-1) was employed (Figure 2). The 50 × 2 mm i.d. stainless steel columns were packed with this material by VDS Optilab (Berlin, Germany). LC runs were carried out at 25 °C, and the injection volume was set to 15 µL. Isocratic elution was performed from 0 to 3.5 min at 0.5 mL‚min-1 with eluent A, i.e., 25 mM HAc in ACN/water (77:23; v/v) at pHa 7.4 (adjusted with 12% aqueous ammonium hydroxide solution). At 3.5 min, the flow direction was reversed from column head to column end (use of a six-port switching valve in back-flush mode). At the same time, the flow was increased to 1.0 mL‚min-1 and the mobile phase was changed to the stronger eluent B, which was composed of 50 mM HAc in ACN/water (77:23; v/v) at pHa 8.2. After 4.5 min, the eluent was changed back to A while keeping the flow rate at 1.0 mL‚min-1. The analytical run was completed after 10 min. For the next injection, the flow rate was decreased to 0.5 mL‚min-1 and the flow direction was reversed to yield the starting conditions. (3) Mass Spectrometry. Parameters for MS detection with the 4000 Q Trap system were optimized using the quantitative optimization tool of the software with manual fine-tuning. The compounds, dissolved in mobile phase at 1 mg‚L-1, were infused separately at 600 µL‚h-1 employing a syringe pump (Harvard Apparatus, Holliston, MA). Ionization was performed in negative ion mode at -4500 V with the source temperature kept at 600 °C and the flow of curtain, nebulizer, and heater gas at 15, 40, and 60 psi, respectively. Detection was carried out in multiple reaction monitoring (MRM) mode using nitrogen as collision gas (CAD set to “high”) with 75-ms dwell time for each dissociation pathway (10-ms pause between). For each compound, one quantifier and one qualifier transition was selected and each pair (analyte + IS) was detected in a distinct period applying the respective optimized conditions (Table 1). To keep contamination of the ion optics with matrix components minimal, a postcolumn 10-port switching valve was used to divert only the respective analyte-containing effluents to the MS system (see Table 1). Method Validation. (1) Selectivity. Analytes and IS were initially checked for purity by controlling the absence of signals in the elution time frame and MS/MS channels of the other compounds. In analogy, each of six urine lots as well as pooled urine was analyzed for absence of interferences and assay selectivity was assured at 0.1 mg‚L-1 (intended lower limit of quantification, LLOQ). Selectivity of the LC-MS/MS method was further guaranteed by a dedicated optimization of the eluent composition as to result in sufficient chromatographic separation of EtG, EtS, and EtP from some major urinary matrix components, viz. chloride, creatinine, sulfate, urea, and uric acid. In addition,

(25) Nogueira, R.; Laemmerhofer, M.; Lindner, W. J. Chromatogr., A 2005, 1089, 158-169.

(26) Nogueira, R.; Lubda, D.; Leitner, A.; Bicker, W.; Laemmerhofer, M.; Lindner, W. J. Sep. Sci. 2006, 29, 966-978.

5886 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

Figure 2. Structure of the employed mixed-mode RP/WAX stationary phase.

Table 1. Elution Times (te), Switching Times, and MS/MS Detection Parameters of Analytes and Internal Standards compd EtS

te (min)a

effluent to MS (min)

MRM pairb

1.30

1.0-1.8

2.55

2.1-3.0

4.85

4.2-6.0

125/97 125/80 130/98 130/80 221/75 221/85 226/75 226/85 125/79 125/63 129/83 129/65

d5-EtS EtG d5-EtG EtP 18O

2-EtP

DPc

EPd

CEe

CXPf

-45

-10

-50

-3

-45

-10

-24 -44 -24 -44 -22 -24 -22 -24 -18 -78 -18 -78

-15 -11 -15 -11 -11 -13 -11 -13 -11 -9 -11 -9

a Variation of (3% was deemed acceptable. b Italics indicates quantifier pairs. c Declustering potential (V). d Entrance potential (V). e Collision energy (eV). f Collision cell exit potential (V).

it was checked for MS/MS channel cross-talk by analyzing blank urine samples spiked either with analytes at the highest concentration level in the absence of IS or with IS at the IS spiking level in the absence of analytes (zero samples) and by monitoring all MRM transitions used for quantification and qualification purposes. (2) Linearity. Two eight-point calibration sets were measured. Calibrants for calibration range 1 (1:20 dilution, corresponds to 0.1-15 mg‚L-1 in undiluted urine) were prepared in water, pooled blank human urine, and six different lots of human urine. On the other hand, for calibration range 2 (corresponds to 5-750 mg‚L-1 in undiluted urine), the blank urine specimens were diluted 1:50 with water prior to spiking with the working solutions (total 1:1000 dilution). Calibration curves (analyte/IS peak area ratio vs nominal spiking concentration) were obtained by 1/x2 weighted linear leastsquares regression. (3) Precision and Accuracy. Intraday and interday precision as well as accuracy were assessed across the entire calibration range. Intraday precision (n ) 5), expressed as percentage relative standard deviation (RSD, %), and accuracy (%), defined as the agreement of the calculated concentration with the nominal concentration, were examined at three levels (5, 62.5, 750 µg‚L-1) each with pooled and 1:50 diluted pooled urine, i.e., corresponding to six concentrations in undiluted urine (0.1, 1.25, 15 mg‚L-1 for calibration range 1; 5, 62.5, 750 mg‚L-1 for calibration range 2). Interday precision and accuracy were determined for the same concentrations by analyzing spiked urine samples (individual lots, pooled; each for both calibration ranges) prepared freshly on a daily basis on five different days. (4) Matrix Effects. Absolute matrix effects20 were investigated by comparing validation data obtained from purely aqueous solutions with those obtained from corresponding spiked urine samples and by postcolumn infusion experiments. For the latter, individual solutions of EtG (20 mg‚L-1 dissolved in eluent A), EtP (40 mg‚L-1, eluent B), and EtS (20 mg‚L-1, eluent A) were infused at 600 µL‚h-1 using a stainless steel mixing tee. Alterations of the respective MS/MS transition intensities were monitored after injecting water, 1:20 diluted urine, and 1:1000 diluted urine. Relative matrix effects20 were characterized by comparing absolute

peak areas of IS obtained in the linearity study for the individual urine lots, by statistical evaluation of differences in the calibration line slopes calculated for each lot of urine (calibration range 1), and by carrying out the interday precision and accuracy study with six human urine lots of different origin. (5) Storage Stability. Storage stability was examined with solutions of the analytes spiked at 0.2 and 100 mg‚L-1, respectively, to water and to pooled urine at pH 4.5 (adjusted with formic acid), pH 6.0 (unadjusted), and pH 7.5 (adjusted with aqueous ammonium hydroxide solution), respectively. Aliquots were stored for 15 days at either room temperature (22), +6, and -20 °C with three freeze-and-thaw cycles or -20 °C in tightly sealed polypropylene vials. Together with these samples, aliquots of five postmortem urine specimens with an ethanol content of 1.504.16 g‚L-1 were each stored under analogous conditions. Degradation was monitored in terms of changes in the ratio of analytes to the IS, which were added right before analysis, compared to freshly prepared samples. RESULTS AND DISCUSSION Method Development. (1) Liquid Chromatography. Owing to the highly polar acidic nature of EtG (pKa ) 2.84 ( 0.70), EtS (pKa ) -3.14 ( 0.15), and EtP (pKa ) 1.91 ( 0.1, 6.39 ( 0.30, all data calculated with ACD software27), and the presence of a large number of inorganic and organic anionic species in urine with similar extraction or chromatographic properties, an effective sample cleanup and analyte enrichment from urinary matrix was deemed to be difficult and presumably lengthy. To end up with a high-throughput analysis method, direct injection after dilution of urine was chosen. This strategy requires great care in making the chromatography sufficiently selective in terms of separation of the target solutes from major urinary matrix components. We propose herein as an advanced separation concept the use of a mixed-mode RP/WAX phase based on N-(10-undecenoyl)-3aminoquinuclidine.23,25,26 This stationary phase allows tuning of retention and selectivity by both hydrophobic/hydrophilic interactions and anion exchange mechanisms as well. Satisfactory retentivity of the target solutes due to anion exchange and a great flexibility in the optimization of the elution conditions are the result. Thus, in the first place, a number of mobile-phase variables with prime importance for the involved RP or hydrophilic interaction liquid chromatography (HILIC) as well as ion exchange retention mechanisms were investigated. To illustrate the retention behavior on RP/WAX, Figure 3 shows the dependency of the retention factors of EtG and EtS on ACN content and pHa of the eluent (25 mM HAc, counterion was ammonium) (an analogous figure for the urine matrix components chloride and uric acid is given in Supporting Information). It is seen from Figure 3 that, independently from the modifier concentration, an increase in pHa led to a decrease in retention of EtG and EtS. This behavior can readily be explained by a lowering of the anion exchange capacity through reduced protonation of the RP/WAX selector. On the other hand, within the investigated modifier range, the retention factors always rose when the ACN fraction was increased, in accordance with a HILIC (27) ACD/Labs 7.00 Release, Advanced Chemistry Development Inc., Toronto, Canada.

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

5887

Figure 3. Dependency of the retention factor of EtG and EtS on the ACN content and pHa of the hydroorganic eluent using a RPWAX stationary phase. For further experimental details, see text.

retention mechanism. While the effect was moderate for EtS, it was more pronounced for EtG as indicated by a steeper slope of the response surface. This leads to a reversal of the elution order of EtG and EtS at ∼60% ACN (Figure 3). The multitude of hydroxyl groups in the EtG molecule appears to be responsible for developing strong hydrophilic interaction forces with the RPWAX moiety in ACN-rich mobile phases. An eluent composed of 25 mM HAc in ACN/water (77:23; v/v) with pHa adjusted to 7.4 (eluent A) provided sufficient selectivity for the separation of EtG (k ) 9.5) and EtS (k ) 4.1) from the urinary macrocomponents chloride (k ) 5.5), sulfate (k ) 7.5), and uric acid (k ) 6.1). Urea and creatinine eluted with the void volume under these conditions. This mobile-phase composition, however, precluded the elution of EtP within a reasonable time (k ) 128). Attempts to decrease retention of EtP by increasing the ionic strength and the pHa both showed the expected effect, but were accompanied by a loss of resolution of EtG and EtS toward chloride, sulfate, and uric acid. To end up with a reasonably fast and selective LC method, the strong retention of EtP was exploited to the advantage: By use of eluent A, EtG and EtS were eluted isocratically, while EtP remained trapped essentially on the column head. At 3.5-min run time, the elution strength of the mobile phase was stepwise increased (change from pHa 7.4 to 8.2 and from 25 to 50 mM HAc, eluent B), and concomitantly, the flow direction of the eluent through the column was reversed using a six-port switching valve (back-flush mode); simultaneously, the flow rate was increased from 0.5 to 1.0 mL‚min-1 (schematic figure is given in Supporting Information). With this back-flush strategy, EtP could be eluted with an adequate peak shape within 1.5 min after switching. The total run time (including reconditioning of the column) could thereby be reduced to 10 min. (2) Mass Spectrometry. To meet the common criteria for confirmatory analysis in forensic toxicology,28,29 detection was performed in MRM mode by monitoring two transitions per 5888 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

analyte and IS, respectively (Table 1). Product ion scans of EtG, EtP, and EtS are available as Supporting Information. Structure proposal for EtG and EtS product ions can be found elsewhere.8,11 EtP ([M - H]- m/z 125) experienced fragmentation putatively to [H2PO3]- (m/z 79; quantifier dissociation) and [PO2]- (m/z 63; qualifier dissociation). For 18O2-EtP ([M - H]- m/z 129) m/z 83 and 65 as the corresponding product ions were observed. It is noted that both EtS and EtP generate pseudomolecular ions with the same m/z. Thus, in case of using single ion monitoring of [M - H]- for quantification of EtS,4,30 selective chromatographic conditions are mandatory to separate EtP from EtS. Method Validation. (1) Assay Selectivity. The optimized LC-MS/MS method proved to be selective for the determination of EtG, EtP, and EtS in human urine. Only in the EtP MRM traces was a minor peak with same elution time as EtP consistently identified in the six blank urine lots, although well below the intended LLOQ of 0.1 mg‚L-1. It remains unclear whether this is due to a normal endogenous level of this solute. None of the IS did contribute to the analyte MS/MS signals and vice versa (absence of channel cross-talk). At a spiking level of 0.1 mg‚L-1 the analytes could be well distinguished (S/N > 6) from the background noise (Figure 4). (2) Matrix Effect. Matrix-induced alterations of MS(/MS) signals caused by ion suppression or enhancement processes may be referred to as “absolute”, i.e., signal differences between analytes spiked to neat solutions and to matrix and, “relative”, i.e., biofluid lot-to-lot dependency of the analyte signal.20 Such matrix effects may impair method reliability considerably,20,31-34 but it appears that often no proper attention is paid to this issue during method validation. A “semiquantitative” measure of the influence of the urinary matrix on the analyte MRM traces was gained by a postcolumn infusion study of the individual analytes and injection of water as well as 1:20 and 1:1000 diluted pooled urine (see Supporting Information for figure). While close to the void volume a severalfold ion suppression was present, especially with the 1:20 diluted urine, relevant signal alterations were not observed in the retention time window of EtS and EtG. For EtP, a minor ion suppression (