Riet Dams,, Constance M. Murphy,*, Robin E. Choo, Willy E. Lambert, Andre P. De Leenheer, and Marilyn A. Huestis

Anal. Chem. 2003, 75, 798-804 LC-Atmospheric Pressure Chemical Ionization-MS/ MS Analysis of Multiple Illicit Drugs, Methadone, and Their Metabolites in Oral Fluid Following Protein Precipitation Riet Dams,, Constance M. Murphy,*, Robin E. Choo, Willy E. Lambert, Andre P. De Leenheer, and Marilyn A. Huestis Chemistry and Drug Metabolism, National Institute on Drug Abuse, 5500 Nathan Shock Drive, Baltimore, Maryland 21224, and Laboratory of Toxicology, Ghent University, Harelbekestraat 72, B-9000 Gent, Belgium A quantitative LC-APCI-MS/MS method for simultaneous determination of multiple illicit drugs, methadone, and their metabolites in oral fluid was developed and validated. Sample pretreatment was limited to acetonitrile protein precipitation. LC separation was performed in 25.5 min, with a total analysis time of 35 min. Identification and quantitation were based on selected reaction monitoring. Calibration by linear regression analysis utilized deuterated internal standards and a weighing factor 1/x. Limits of detection and lower limits of quantitation (LOQ) were established between 0.25 and 5 ng/ ml and 0.5-10 ng/ml, respectively. Linearity was obtained with an average correlation coefficient (R 2 ) of >0.99, over a dynamic range from the LOQ up to maximum 500 ng/ml. The method demonstrated good accuracy, intra- and interbatch precision, recovery, and stability for all compounds. No oral fluid matrix effect was observed throughout the chromatographic run. Protein precipitation provided a fast and simple sample pretreatment, while LC-APCI-MS/MS proved to be a sensitive and rugged quantitative method for multiple illicit and legal drugs in oral fluid. The method proved to be suitable for the evaluation of oral fluid as an alternative matrix to urine for monitoring illicit drug use and for determining oral fluid methadone concentrations in pregnant opiate and/or cocaine addicts. For years, monitoring drug use in drug treatment, criminal justice, and workplace drug-testing programs has been performed by urinalysis. 1 Urine drug testing has become an established, reliable, standardized, relatively inexpensive, and widely available technology. 2 However, tactics for adulteration of test results are well known. Furthermore, urinalysis mainly provides information on drug metabolites while parent drugs are often excreted in low * Corresponding author: (tel) +1 410 550 1815, ext 41; (fax) +1 410 550 2971; (e-mail) cmurphy@intra.nida.nih.gov. National Institute on Drug Abuse. Ghent University. (1) Huestis, M. A.; Cone, E. J.; Wong, C. J.; Umbricht, A.; Preston, K. L. J. Anal. Toxicol. 2000, 24, 509-521. (2) Preston, K. L.; Huestis, M. A.; Wong, C. J.; Umbricht, A.; Goldberger, B. A.; Cone, E. J. J. Anal. Toxicol. 1999, 23, 313-322. concentrations. The disadvantages of urine drug testing have led to the development of alternative drug testing tools to monitor illicit and licit drug use. In response to these issues, the interest in other biological matrixes has increased. 3 Oral fluid is one of the most promising new matrixes. This colorless liquid is excreted into the oral cavity from the submandibular, sublingual, and parotid glands. 4 The major components are water (99%), protein (mostly enzymes, 0.3%), mucin (0.3%), and salts. Oral fluid is a natural ultrafiltrate of plasma with substances transported across epithelial membranes into oral fluid by passive diffusion across a concentration gradient. Although less common, low-molecular-mass compounds also can be transferred into oral fluid by active secretion or diffusion through pores in the membrane. 5,6 As an alternative matrix for drug testing, oral fluid offers some clear advantages. The matrix is relatively clean and readily accessible for sampling. Sample collection is easy, noninvasive, and inexpensive. Additionally, sampling can be better supervised without invasion of privacy, reducing the opportunity for sample adulteration or substitution. Finally, because of the low protein content of oral fluid and because highly protein bound molecules are unlikely to cross cellular membranes, oral fluid testing offers the possibility of direct comparison of unbound, pharmacologically active drug concentrations to the observed effects. 4-6 There are also a number of disadvantages to oral fluid testing. One disadvantage is a shorter detection window in comparison to urine. In general, drug concentrations found in oral fluid are lower than those found in urine. Drug concentrations may be higher than those found in blood in instances where the drugs concentrate in the oral fluid due to ion trapping. Additionally, salivary ph and flow can be highly variable, directly influencing drug transport into the oral fluid. Finally, the possibility of oral contamination, and the smaller sample volume collected, are also drawbacks of oral fluid testing. 4-7 (3) Cone, E. J. Forensic Sci. Int. 2001, 121, 7-15. (4) Rivier, L. Best Pract. Res. Clin. Endocrinol. Metab. 2000, 14, 147-165. (5) Kidwell, D. A.; Holland, J. C.; Athanaselis, S. J. Chromatogr. 1998, 713, 111-135. (6) Pichini, S.; Altieri, I.; Zuccaro, P.; Pacifici, R. Clin. Pharmacokinet. 1996, 30, 211-228. (7) Schramm, W.; Smith, R. H.; Craig, P. A.; Kidwell, D. A. J. Anal. Toxicol. 1992, 16, 1-9. 798 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003 10.1021/ac026111t CCC: $25.00 2003 American Chemical Society Published on Web 01/15/2003

To overcome these limitations and evaluate oral fluid as a viable drug-testing biomatrix, sensitive analytical techniques are needed. In recent years, liquid chromatography combined with atmospheric pressure ionization mass spectrometry (LC-API-MS) has made a substantial contribution to the identification and quantitation of drugs in clinical and forensic toxicology. 8-10 Highperformance liquid chromatography separates a wide range of analytes without derivatization. The combination of chromatographic separation and mass spectrometric detection further allows for universal and highly selective detection of drugs and their related metabolites. Furthermore, when operated in tandem MS mode, LC-MS has improved sensitivity and specificity for the determination of trace amounts of analytes in complex biological fluids. Additionally, LC-MS/MS offers the possibility of reduced and simplified sample pretreatment prior to analysis. To date, there are limited reports applying LC-MS of oral fluid for analysis of drugs of abuse and pharmacotherapeutics. Bentley et al. described an LC-MS/MS method for the determination of nicotine metabolites with a triple quadrupole mass spectrometer equipped with an electrospray ion source. 11 Sample preparation was performed by automated solid-phase extraction. Cotinine was evaluated as a biomarker of low-level environmental tobacco smoke (ETS). Furthermore, a single quadrupole LC-MS, with orthogonal electrospray, was used by Ortelli et al. to evaluate oral fluid as a biological matrix for methadone monitoring. 12 After filtration of the oral fluid samples, a conventional and enantioselective HPLC separation was performed, prior to mass analysis. LC-MS sensitivity permitted measurement of low concentrations of compounds of interest, eliminating the need for preconcentration during sample pretreatment. The results showed a good correlation between the enantiomeric ratios of methadone in oral fluid and serum. Mortier et al. briefly addressed the pitfalls associated with LC-MS/MS in quantitative bioanalysis of drugs of abuse in oral fluid. 13 MS experiments were performed on a QTOF instrument, fitted with a Z-spray electrospray source. Different protein precipitation and ultrafiltration techniques were compared in an effort to overcome the suppressive effect of endogenous matrix components on the ionization of compounds of interest, i.e., matrix effect. An LC-MS/MS method for amphetamines, opiates, and cocaine was briefly described. The authors pointed out that sensitivity and precision of an LC-MS method can be seriously influenced by matrix effect. An LC-APCI-MS/MS method, combined with protein precipitation, for the simultaneous quantitation of 27 compounds, including illicit drugs, methadone, and their metabolites in oral fluid, is presented. To facilitate discussion, the analytes have been divided into three groups: group I (heroin, opiates), group II (methadone, propoxyphene), and group III (cocaine). To the best of our knowledge, this is the first paper reporting simultaneous oral fluid (8) Marquet, P. Ther. Drug Monit. 2002, 24, 125-133. (9) Van Bocxlaer, J. F.; Clauwaert, K. M.; Lambert, W. E.; Deforce, D. L.; Van den Eeckhout, E. G.; De Leenheer, A. P. Mass Spectrom. Rev. 2000, 19, 165-214. (10) Maurer, H. H. J. Chromatogr., B 1998, 713, 3-25. (11) Bentley, M. C.; Abrar, M.; Kelk, M.; Cook, J.; Phillips, K. J. Chromatogr., B 1999, 723, 185-194. (12) Ortelli, D.; Rudaz, S.; Chevalley, A. F.; Mino, A.; Deglon, J. J.; Balant, L.; Veuthey, J. L. J. Chromatogr., A 2000, 871, 163-172. (13) Mortier, K. A.; Clauwaert, K. M.; Lambert, W. E.; Van Bocxlaer, J. F.; Van den Eeckhout, E. G.; Van Peteghem, C. H.; De Leenheer, A. P. Rapid Commun. Mass Spectrom. 2001, 15, 1773-1775. LC-MS/MS analysis for this complex mixture of compounds. The method will be applied to an outpatient clinical study of pregnant heroin or cocaine addicts enrolled in a methadone treatment program, to monitor drug use throughout pregnancy and evaluate oral fluid as an alternative matrix to urine. EXPERIMENTAL SECTION Instrumentation. All experiments were carried out on an LCQ Deca XP ion trap mass spectrometer, equipped with an orthogonal APCI source, and interfaced to a Surveyor HPLC system (ThermoFinnigan, San Jose, CA). All data were acquired and analyzed using Xcalibur Software, version 1.2. Ultracentrifugation of the samples was performed in an Eppendorf centrifuge 5415C (Hamburg, Germany). Evaporation under nitrogen was conducted in a TurboVap LV evaporator from Zymark (Hopkinton, MA). Reagents. A complete list of the reference and internal standards is given in Table 1. All analytes were purchased from Cerilliant (Austin, TX) with the exception of methadone, papaverine, noscapine, ecgonine, norcodeine, m- and p-hydroxycocaine, benzoylnorecgonine, and p-hydroxybenzoylecgonine obtained from Sigma Chemical Co. (St. Louis, MO); cocaine-d 8 from High Standard Products Corp. (Westminster, CA); and acetylcodeined 3 from Lipomed (Arlesheim, Switzerland). Reagent grade ammonium formate and formic acid were purchased from Sigma Chemical Co. All solvents were of HPLC grade or higher. A pool of drug-free oral fluid was obtained from healthy volunteers. Study specimens were collected from participants in an outpatient IRB-approved study investigating in utero drug exposure in a population of heroin- and/or cocaine-addicted pregnant women enrolled in a methadone treatment program. Volunteers provided informed consent and were paid for their participation. Oral fluid specimens were collected and stored in the freezer at -20 C until use. Preparation of Standard Solutions. Individual stock solutions of 100 µg/ml for the reference compounds were prepared in methanol or acetonitrile, according to the solubility of the solute, and stored in the dark at -20 C until use. Working solutions, ranging from 10 ng/ml to 10 µg/ml, were prepared by dilution with methanol/acetonitrile (50:50, v/v). Blank oral fluid samples were fortified with the working solutions to make oral fluid standards at concentrations ranging from 0.5 to 500 ng/ml. For the deuterated internal standards, stock solutions of 10 µg/ml were prepared in methanol or acetonitrile and stored in the dark at -20 C until use. A 2 µg/ml working solution of the complete mix was prepared in methanol/acetonitrile (50:50, v/v). Ten microliters of this working solution was added to each sample, giving a final deuterated internal standard concentration of 100 ng/ml. Procedures. Sample Preparation. Sample pretreatment of fortified oral fluid samples and clinical specimens was performed by acetonitrile protein precipitation. A 200-µL sample of oral fluid was combined with 10 µl of the internal standard working solution. The sample was vortexed for 10 s. After addition of 600 µl of acetonitrile, the sample was vortexed for 30 s and centrifuged at 14 000 rpm for 10 min. The supernatant was evaporated to dryness under nitrogen at 45 C. The dried sample was reconstituted in 100 µl of a mixture of mobile phases A and B (97:3, v/v) and vortexed before injection on the LC-MS system. Analytical Chemistry, Vol. 75, No. 4, February 15, 2003 799

Table 1. LC-APCI-MS/MS Method Parameters compound a precursor ion (m/z) collision energy (V) product ion(s) b (m/z) Rt (min) variation Rt (%, n ) 30) segment ecgonine-d 3 189.3 30 171.2 1.8 0.80 I ecgonine 186.3 30 168.3 1.8 0.68 I EME-d 3 203.2 30 185.2 2.3 1.25 I EME 200.2 30 182.2 2.3 1.41 I EEE 214.3 30 196.2 3.1 3.02 I normorphine 272.3 30 254.2; 229.1; 211.2 5.8 0.83 II AEM 182.3 35 151.1; 122.1 6.5 0.94 II morphine-d 6 292.4 35 274.3; 229.1; 201.2 7.5 0.39 II morphine 286.3 35 268.3; 229.1; 211.2 7.6 0.75 II p-hydroxybenzoylecgonine 306.3 40 168.2; 186.2 8.5 0.59 III m-hydroxybenzoylecgonine 306.2 30 168.2 9.2 0.41 III benzoylnorecgonine 276.2 30 154.1 9.4 0.36 III norcodeine 286.3 30 268.2; 243.3; 225.3 9.7 0.40 III benzoylecgonine-d 8 298.3 30 171.2 10.1 0.33 III benzoylecgonine 290.3 30 168.2 10.2 0.48 III codeine-d 6 306.4 35 288.3; 218.2; 228.3 11.1 0.19 III codeine 300.3 35 282.3; 243.2; 215.2 11.2 0.31 III acetylmorphine-d 6 334.4 35 211.3; 271.3; 193.2 13.2 0.17 IV 6-acetylmorphine 328.3 35 268.3; 211.2; 193.2 13.3 0.26 IV p-hydroxycocaine 320.2 30 182.2 15.4 0.31 IV m-hydroxycocaine 320.3 30 182.2 15.9 0.27 IV heroin-d 9 379.4 40 316.3; 335.3; 272.2 18.0 0.10 V heroin 370.3 40 310.2; 328.2; 268.3 18.1 0.18 V acetylcodeine-d 3 345.4 35 285.3; 225.2 18.3 0.14 V acetylcodeine 342.3 35 282.3; 225.2 18.3 0.17 V norcocaine-d 3 293.2 30 171.1 18.8 0.12 V norcocaine 290.2 30 168.1; 136.2 18.8 0.14 V cocaine-d 8 312.3 30 185.2 19.0 0.12 V cocaine 304.3 30 182.1 19.1 0.16 V papaverine 340.3 40 202.2; 325.3; 243.1 19.8 0.11 VI norcocaethylene 304.2 30 182.1; 136.1 20.2 0.16 VI cocaethylene-d 8 326.3 30 204.3 20.7 0.09 VI cocaethylene 318.3 30 196.2 20.7 0.13 VI noscapine 414.2 30 220.2 20.8 0.09 VI methadol 312.3 30 223.0; 171.2; 105.2 23.1 0.21 VII propoxyphene-d 11 351.2 30 277.3 23.5 0.16 VII propoxyphene 340.1 40 266.1 23.6 0.24 VII methadone-d 9 319.3 30 268.3 24.9 0.18 VII methadone 310.9 30 265.2 25.0 0.21 VII EDDP-d 3 281.4 40 249.2 25.0 0.20 VII EDDP 278.0 40 249.2 25.1 0.21 VII a Abbreviations: EME, ecgonine methyl ester; EEE, ecgonine ethyl ester; AEME, anhydroecgonine methyl ester mesylate; EDDP, 2-ethyl-1,5- dimethyl-3,3-diphenylpyrrolinium. b Ion sequence according to descending abundance. Liquid Chromatography. Chromatographic separation was performed on a Synergi Polar RP column (150 2.0 mm, 4 mm), fitted with a guard column with identical packing material (4 2.0 mm) (Phenomenex, Torrance, CA). The column oven was maintained at 25 C, and 50 ml of each sample was injected. Gradient elution with (A) 10 mm ammonium formate in water, 0.001% formic acid (ph ) 4.5) and (B) acetonitrile, at a flow rate of 300 ml/min, was applied. The initial gradient conditions were 5% B, increasing to 26% B in 13 min, with a final composition of 90% B in 9 min. The column was flushed for 2 min at 90% B. Initial gradient conditions were reestablished in 3 min, and the column was equilibrated for an additional 7 min. A divert valve was set to direct the LC flow initially to waste for 0.5 min, subsequently to the mass spectrometer for 25.5 min, and then back to the waste for the remaining 9.5 min of the analysis. Mass Spectrometry. All MS data were collected in positive ion mode. The following APCI-MS parameter settings were applied: corona discharge needle voltage, 4.5 kv; vaporizer temperature, 450 C; sheath gas (high-purity nitrogen) pressure, 70 psi; no auxiliary gas; and transfer capillary temperature, 220 C. The electron multiplier voltage was set at 850 ev. Identification and quantitation was based on selected reaction monitoring (SRM). To ensure adequate quantitation, the chromatographic run was divided into seven segments. Each segment incorporated a fullscan experiment followed by a number of SRM scan events, optimized for the compounds eluting within the time segment. The scan range for the parent scan was 150-500 atomic mass units (amu), and each scan consisted of three microscans with a maximum ion inject time of 50 ms. Each SRM scan consisted of one microscan and a 50-ms inject time. A complete overview of the SRM transitions, collision energy, retention time, and corresponding segment used for each analyte is given in Table 1. Data Analysis. Calibration, using internal standardization, was done by linear regression analysis over a maximum concentration range from 0.5 to 500 ng/ml. For each standard curve, a minimum of six different concentrations was used. In the absence of commercially available stable isotopes of a limited number of compounds, the deuterated form of a structural analogue of the same group, eluting close to the compound, was used as the 800 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

Table 2. LODs, LLOQs, and Calibration Results compound internal standard LOD (ng/ml) LOQ linear dynamic (ng/ml) equation R 2 range (ng/ml) group I normorphine morphine-d 6 0.5 1.0 Y )-0.0077 + 0.016X 0.992 1-500 morphine morphine-d 6 0.5 1.0 Y ) 0.0012 + 0.0123X 0.996 1-500 norcodeine codeine-d 6 0.5 1.0 Y )-0.0021 + 0.0101X 0.995 1-500 codeine codeine-d 6 0.5 1.0 Y )-0.0046 + 0.0102X 0.993 1-500 6-acetylmorphine acetylmorphine-d 6 0.5 1.0 Y )-0.0106 + 0.0103X 0.996 1-500 heroin heroin-d 9 0.5 1.0 Y )-0.0079 + 0.0107X 0.993 1-500 acetylcodeine acetylcodeine-d 3 0.5 1.0 Y )-0.0006 + 0.0112X 0.993 1-500 papaverine cocaethylene-d 8 0.5 1.0 Y ) 0.0050 + 0.0214X 0.995 1-500 noscapine cocaethylene-d 8 0.5 1.0 Y )-0.0051 + 0.0082X 0.992 1-500 group II methadol methadone-d 9 1.0 5.0 Y )-0.0034 + 0.0007X 0.992 5-500 methadone methadone-d 9 0.5 1.0 Y ) 0.0053 + 0.0088X 0.993 1-500 EDDP EDDP-d 3 0.5 1.0 Y ) 0.0086 + 0.0084X 0.994 1-400 propoxyphene propoxyphene-d 11 5.0 10.0 Y )-0.0435 + 0.0070X 0.986 10-500 group III ecgonine ecgonine-d 3 1.0 5.0 Y ) 0.0878 + 0.0074X 0.994 5-500 EME EME-d 3 5.0 10.0 Y ) 0.0870 + 0.0029X 0.991 10-500 EEE EME-d 3 5.0 10.0 Y ) 0.0077 + 0.0103X 0.991 10-500 AEME benzoylecgonine-d 8 5.0 10.0 Y )-0.0003 + 0.0001X 0.991 10-500 p-hydroxybenzoylecgonine benzoylecgonine-d 8 0.5 1.0 Y )-0.0033 + 0.0054X 0.995 1-500 m-hydroxybenzoylecgonine benzoylecgonine-d 8 0.5 1.0 Y )-0.0034 + 0.0066X 0.992 1-500 benzoylnorecgonine benzoylecgonine-d 8 0.5 1.0 Y ) 0.0103 + 0.0056X 0.996 1-500 benzoylecgonine benzoylecgonine-d 8 0.25 0.5 Y )-0.0029 + 0.0099X 0.993 0.5-500 p-hydroxycocaine norcocaine-d 3 0.5 1.0 Y )-0.0023 + 0.0055X 0.993 1-500 m-hydroxycocaine norcocaine-d 3 0.5 1.0 Y )-0.0052 + 0.0091X 0.991 1-500 norcocaine norcocaine-d 3 0.5 1.0 Y )-0.0063 +0.0118X 0.992 1-500 cocaine cocaine-d 8 0.5 1.0 Y )-0.0004 + 0.0106X 0.995 1-500 norcocaethylene cocaethylene-d 8 0.5 1.0 Y ) 0.0027 + 0.0123X 0.996 1-500 cocaethylene cocaethylene-d 8 0.5 1.0 Y )-0.0010 + 0.0105X 0.995 1-500 internal standard (see Table 2). Peak area ratios of target analytes and their respective internal standards were calculated for each concentration by Xcalibur s LCQuan software (version 1.2). The data were fit to a linear least-squares regression curve with a weighing factor of 1/x. Validation. The following criteria were used to evaluate the method: sensitivity, linearity, intra- and interbatch precision, accuracy, recovery, and matrix effect. Specificity was achieved by a unique combination of retention time, precursor, and fragment ion. The sensitivity of the method was evaluated by determining the limit of detection (LOD) and limit of quantitation (LOQ). LOD was defined as the concentration with a signal-to-noise ratio of at least 3, while LOQ was the lowest standard with a signal-to-noise ratio of at least 10 and acceptable precision and accuracy (relative standard deviation (RSD) and percent difference (% Diff), respectively, less than 23.0%). Both parameters were determined empirically by analysis of a series of decreasing concentrations of the drug-fortified oral fluid in multiple replicates. The linearity of the method was investigated by calculation of the regression line by the method of least squares and expressed by the correlation coefficient (R 2 ).A1/x weighing factor was applied to reduce heteroscedasticity. Linearity of each of the compounds was determined with at least six concentration levels not including the blank matrix. Precision and accuracy were evaluated over the linear dynamic range, at three different concentration levels, i.e., low, medium, and high. The specific concentrations tested for each group of compounds are given in Table 3. Intrabatch precision was assessed by five determinations per concentration in 1 day. Interbatch precision was measured for five replicates per concentration, on 4 consecutive days (n total ) 20). Precision was expressed as RSD. Accuracy was determined by comparison of the mean result for five analyses to the nominal concentration value. Accuracy was expressed as the percent difference from the expected value. The recovery or extraction efficiency (%) for each analyte was also determined at low, medium, and high concentration levels (see Table 4). The internal standard working solution was added to fortified oral fluid samples before and after sample pretreatment, and the resultant peak area ratios were compared. Stability of the stock solutions was tested monthly by injection of freshly prepared working solutions. Stability experiments for the method were performed on oral fluid fortified with the analytes of interest at one concentration level (10 ng/ml for groups I and II; 100 ng/ml for group III). The freeze-and-thaw stability of the analytes was determined after five freeze-and-thaw cycles of fortified oral fluid. Short-term temperature stability was evaluated for fortified oral fluid kept at room temperature for 24 h. Analyte stability during pretreatment was determined at the following stages of protein precipitation: acetonitrile extract, 24 h at room temperature; dry extract, 24 h at room temperature, 4 C, -80 C, and 4 days at room temperature. Matrix effect was evaluated by the experimental setup developed by King et al. 14 Simultaneous with injection of a blankpretreated oral fluid sample, postcolumn infusion of a morphine solution (10 µg/ml) at a flow of 5 µl/min was performed. Comparison of the total ion chromatogram obtained with and without injection of blank oral fluid shows the areas of the (14) King, R.; Bonfiglio, R.; Fernandez-Metzler, C.; Miller-Stein, C.; Olah, T. J. Am. Soc. Mass Spectrom. 2000, 11, 942-950. Analytical Chemistry, Vol. 75, No. 4, February 15, 2003 801

Table 3. Precision and Accuracy Data intrabatch precision (% RSD, n ) 5) interbatch precision (% RSD, n ) 20) accuracy (% Diff, n ) 5) compound low medium high low medium high low medium high group I 5 ng/ml 50 ng/ml 400 ng/ml 5 ng/ml 50 ng/ml 400 ng/ml 5 ng/ml 50 ng/ml 400 ng/ml normorphine 14.4 10.6 8.6 16.3 12.0 6.0 12.7 9.9 5.7 morphine 14.3 6.4 6.0 21.0 9.2 6.5 13.4 4.7 5.1 norcodeine 12.4 7.6 5.1 12.9 8.3 4.6 8.9 6.5 4.0 codeine 10.2 5.9 4.4 16.4 8.1 6.3 8.9 4.3 6.7 6-acetylmorphine 9.8 7.0 3.8 12.3 6.1 4.9 8.1 6.5 3.9 heroin 15.1 7.0 6.4 11.0 9.4 6.0 12.0 6.1 5.9 acetylcodeine 15.5 7.3 5.3 10.9 7.9 4.6 12.9 7.0 5.2 papaverine 12.6 3.4 8.9 11.1 7.9 8.8 14.3 5.0 6.1 noscapine 14.4 10.6 8.6 16.4 15.4 6.9 12.7 15.8 5.7 group II 1 ng/ml 50 ng/ml 250 ng/ml 1 ng/ml 50 ng/ml 250 ng/ml 1 ng/ml 50 ng/ml 250 ng/ml methadol 6.8 a 7.2 3.5 7.1 a 12.0 7.5 20.8 a 7.1 5.0 methadone 14.1 3.6 8.9 15.2 6.4 6.3 11.0 10.4 6.4 EDDP 3.4 7.3 4.3 21.6 7.1 5.4 12.5 4.7 7.2 propoxyphene 15.8 a 13.6 8.9 16.6 a 14.2 7.2 19.6 a 17.2 3.9 group III 5 ng/ml 250 ng/ml 500 ng/ml 5 ng/ml 250 ng/ml 500 ng/ml 5 ng/ml 250 ng/ml 500 ng/ml ecgonine 11.8 9.4 4.9 22.8 8.6 7.5 16.1 8.4 6.8 EME 13.2 a 9.8 4.8 13.8 a 9.1 8.8 10.4 a 7.0 8.5 EEE 21.9 a 4.2 8.0 13.5 a 12.5 14.2 8.9 a 6.4 7.7 AEME 2.8 a 10.3 5.1 5.7 a 17.4 15.3 5.1 a 12.8 4.3 p-hydroxybenzoylecgonine 11.0 9.6 5.0 12.3 8.9 8.0 11.5 9.3 6.3 m-hydroxybenzoylecgonine 13.3 12.1 9.2 10.5 11.0 14.3 13.8 12.7 6.3 benzoylnorecgonine 10.3 7.3 5.5 15.7 8.2 8.8 19.3 5.7 6.7 benzoylecgonine 10.0 5.0 9.9 9.8 5.6 6.7 9.4 3.4 7.3 p-hydroxycocaine 10.7 5.0 10.7 17.4 7.8 7.8 14.2 5.2 8.9 m-hydroxycocaine 10.1 7.6 11.9 9.3 9.0 8.3 7.3 6.5 10.3 norcocaine 15.8 3.7 10.7 15.4 6.2 7.3 16.3 3.2 10.0 cocaine 8.5 3.5 7.6 15.3 4.7 6.5 9.5 2.3 7.2 norcocaethylene 14.9 5.4 5.4 13.1 4.7 4.2 12.9 4.3 5.2 cocaethylene 12.7 4.0 5.8 12.2 5.0 5.5 11.5 2.9 5.2 a QC < LOQ, calculations made at LOQ level. chromatographic run where the matrix effect takes place. The experiment was performed in triplicate to ensure its validity. Safety Considerations. The method demands no specific safety precautions. Universal precautions for the handling of chemicals and biofluids were applied. RESULTS AND DISCUSSION LC-MS Method Development. The separation of 27 compounds and their respective internal standards was achieved in 25.5 min. After a column wash and reequilibration period of 9.5 min, the next sample was injected. The stability of the LC method was evaluated by calculation of the variation of retention times. RSD, calculated from retention times obtained over 30 injections, proved to be less than 3.0% for all compounds, indicating good chromatographic stability (Table 1). The precursor and product ion(s) for each analyte of interest was determined by direct infusion of single-analyte solutions (10 µg/ml in water/methanol (50:50, v/v)). After optimization of the LC separation and selection of the unique precursor-product ion combination for each drug, a quantitative LC-APCI-MS/MS method was developed based on SRM. To ensure maximum sensitivity of the MS analysis, the chromatographic run was divided into seven segments. Each segment was optimized for the compounds of interest eluting within a given time period. The following mass spectrometric parameters were specified within each segment: transfer capillary voltage, tube lens voltage, ion optic voltages, collision energy, and SRM scan events. Table 1 shows an overview of the MS parameters including SRM transitions, collision energy, retention time, and different segments for all the analytes. Figure 1 shows the SRM chromatograms for the target analytes obtained from the injection of a standard mixture (10 ng on column). Calibration and Validation. Calibration using internal standardization with deuterated analogues of the drugs was performed. Stable isotope internal standards were employed in order to minimize the effects on the ionization potential of the compounds by interferences from the biological matrix or the method. In the absence of commercially available stable isotopes for all compounds, the deuterated form of a structural analogue eluting close to the compound was selected as the internal standard (see Table 2). To prevent cross-talk interference caused by isotopic contributions to the target ion or by contribution of nondeuterated impurities in the standard, high-purity analogues with the highest possible extent of deuteration and highest purity were selected. 15 For acetylcodeine, 2-ethyl-1,5-dimethyl-3,3-diphenylpyrrolinium (EDDP), ecgonine methyl ester (EME), ecgonine, and norcocaine only the d 3 analogue was commercially available. All other compounds had a deuteration of at least 6. All standards had a purity of at least 99.0%, as reported by the manufacturer. To (15) Bogusz, M. J. J. Anal. Toxicol. 1997, 21, 246-247. 802 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

Table 4. Relative Recovery Results relative recovery (%, n ) 5) compound low medium high group I 5 ng/ml 50 ng/ml 400 ng/ml normorphine 86.3 98.1 93.9 morphine 85.5 97.1 92.1 norcodeine 103.3 94.5 94.2 codeine 99.5 96.3 95.3 6-acetylmorphine 104.1 101.5 93.9 heroin 100.7 101.9 95.6 acetylcodeine 114.6 102.6 91.3 papaverine 91.8 106.6 94.3 noscapine 107.5 95.9 95.8 group II 1 ng/ml 50 ng/ml 250 ng/ml methadol <LOQ 101.1 95.4 methadone 105.6 99.6 97.6 EDDP 102.4 107.1 100.0 propoxyphene <LOQ 107.5 97.3 group III 5 ng/ml 250 ng/ml 500 ng/ml ecgonine 98.5 90.4 100.1 EME <LOQ 99.7 103.0 EEE <LOQ 105.0 98.0 AEME <LOQ 99.1 100.5 p-hydroxybenzoylecgonine 101.5 100.5 103.3 m-hydroxybenzoylecgonine 101.7 102.6 102.8 benzoylnorecgonine 94.7 96.3 102.9 benzoylecgonine 101.6 95.4 100.7 p-hydroxycocaine 95.1 96.3 101.8 m-hydroxycocaine 104.3 96.0 98.2 norcocaine 99.9 99.8 98.1 cocaine 99.1 96.6 96.7 norcocaethylene 96.9 93.8 99.0 cocaethylene 92.2 93.4 100.5 evaluate cross-talk interference, 10 µl of the deuterated internal standard working solution was added to a blank oral fluid sample (known to be negative for the target compounds). The oral fluid sample was pretreated by protein precipitation and analyzed by the developed method. No signal was detected for any of the target compounds. The complete method was evaluated according to the criteria described in the Experimental Section. Table 2 provides the LODs, LOQs, and calibration results for all the analytes. LOD and LOQ were established between 0.25 and 5.0 and 0.5-10.0 ng/ml, respectively. The majority of the analytes (n ) 20) had LODs and LOQs of 0.5 and 1.0 ng/ml, respectively. One compound, benzoylecgonine, had lower LOD and LOQ values, while six compounds had higher LODs and LOQs. Linearity was obtained with an average correlation coefficient (R 2, weighing factor, 1/x) of>0.99, over a dynamic range from the LOQ value up to 500 ng/ml for each of the analytes, except for EDDP (Table 2). This primary methadone metabolite was linear up to 400 ng/ml. Precision and accuracy of the method were evaluated at three concentrations over the linear dynamic range (low, medium, high). Table 3 includes the concentrations tested and results for both validation parameters. For methadol, EME, ecgonine ethyl ester (EEE), anhydroecgonine methyl ester (AEME), and propoxyphene, the low QC concentration was lower than the LOQ. Low-level precision and accuracy data for these compounds were calculated at their respective LOQ. Intrabatch precision for all compounds proved to be less than 15.8% at low, medium, and high QC concentrations, except for EEE, which had a coefficient of variation of 21.9%. Interbatch precision was less than 20% for all compounds, except for ecgonine, EDDP, and morphine, which all had a slightly higher result for the low QC. Accuracy, calculated as the percent difference from the target value, was less than 20.8% at all concentrations. The recovery of the analytes with the protein precipitation procedure was also determined at three concentrations (low, medium, high). No recovery data could be obtained for methadol, EME, EEE, AEME, and propoxyphene at the lowest concentration due to their higher LOQ. An overview of the concentrations tested and the results obtained is given in Table 4. Recoveries ranged from 85.5 to 114.5%. Under the stated conditions, stock solutions proved to be stable for at least 6 months. Analyte recoveries in the stability experiments were within the variability range obtained for precision and accuracy. No significant loss or deterioration for any of the compounds of interest was observed. Analytes were stable in oral fluid and during sample pretreatment, under conditions identical to study conditions. Losses of less than 20% were measured. The last parameter to be validated was the presence of matrix effect. An evaluation of this parameter has become highly recommended in bioanalytical method development by LC-MS with the API ion sources. 16 Although no fixed procedure for the validation of matrix effect has been established, we used the postcolumn infusion experiment first described by King et al. 14 The effect of the sample matrix on the ionization of the compound in the ion source can be easily visualized by injection of a blankpretreated oral fluid sample in the LC system with the simultaneous postcolumn infusion of the analyte of interest. In our case, morphine was infused. We detected no suppressive effect throughout the chromatographic run, indicating that the LC-APCI-MS/ MS method was not influenced by matrix effect. Finally, the method will be applied to oral fluid specimens from a clinical study of pregnant heroin or cocaine addicts enrolled in a methadone treatment program. Our purpose is to determine the most effective method to monitor maternal drug use and to determine potential correlations between daily methadone dosage and oral fluid methadone levels. The method fulfilled our analytical standard criteria. SRM provided high specificity for all of the compounds, and no crosstalk interference with the deuterated internal standards was observed. The sensitivity and linear dynamic range of the method were clinically relevant to monitor drug use by oral fluid analysis. 7 The method achieved precise and accurate oral fluid measurements of the compounds of interest. There was no significant loss of any of the 27 compounds during sample pretreatment, indicating acceptable recovery. Analytes proved to be stable in stock solutions as well as during analysis. No significant matrix effect was observed, and the method was applicable to clinical specimens. LC-MS proved to be a viable analytical tool for oral fluid analysis of licit and illicit drugs. CONCLUSION Protein precipitation combined with LC-APCI-MS/MS provided an efficient, precise, and sensitive method for the quantitation of a wide variety of illicit drugs, methadone, and their (16) (FDA) Food and Drug Administration, Guidance for Industry: Bioanalytical Method Validation 2001; pp 1-22. Analytical Chemistry, Vol. 75, No. 4, February 15, 2003 803

Figure 1. SRM chromatograms of the analytes of interest obtained for the injection of a standard mixture (10 ng on column). metabolites in oral fluid. Sample pretreatment by protein precipitation was characterized by its ease of use and speed. It provided a sufficient cleanup of the oral fluid samples prior to LC-MS analysis and showed no significant loss of any of the analytes of interest during sample handling. APCI proved to be a stable ionization technique, exhibiting no ion suppression under the simplified sample preparation conditions. SRM-based LC-MS/ MS proved to be a sensitive and specific technique, allowing for the quantitative analysis of 27 compounds of interest within the expected clinical range with a limited volume of 200 µl of oral fluid. 7 The method will enable us to support a clinical study monitoring maternal drug use throughout pregnancy and to assess the usefulness of oral fluid as an alternative matrix for drug monitoring. ACKNOWLEDGMENT R.D. gratefully acknowledges the travel grant V 4.037.01N from the F.S.R.-Flanders (Belgium). Received for review September 5, 2002. Accepted December 4, 2002. AC026111T 804 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003