Certification of Phencyclidine in Lyophilized Human Urine Reference Materials

Transcription

1 Certification of Phencyclidine in Lyophilized Human Urine Reference Materials S. S.-C. Tai 1,*, R.G. Christensen 2, K. Coakley 3, P. Ellerbe 1, T. Long 1, and M.J. Welch 2 1College of American Pathologists, Northfield, IL 60093, 2Analytical Chemistry Division, National Institute of Standards and Technology, Gaithersburg, MD 20899, and 3Statistical Engineering Division, National Institute of Standards and Technology, Boulder, CO Abstract The National Institute of Standards and Technology (formerly the National Bureau of Standards), in cooperation with the College of American Pathologists (CAP), has certified the concentrations of phencyclidine (PCP) in two new reference materials (RMs). One of these materials is Standard Reference Material (SRM) 1511, Multidrugs of Abuse in Freeze-dried Urine, and the other material is a CAP PCP RM. In order to minimize the possibility of undetected bias, two independent analytical methods, employing gas chromatography-mass spectrometry and liquid chromatography-mass spectrometry, were used to certify PCP in these materials. Results from the two methods were in good agreement and were statistically combined to yield certified values of 23.8 ng/ml for PCP in SRM 1511 and 11.9, 23.4, and 49.5 ng/ml for three levels of PCP in the CAP RM. A round-robin study of SRM 1511 among five military laboratories demonstrated the suitability of the SRM for its intended purpose. Introduction The growth of illegal drug use has stimulated widespread testing for substance abuse. In an effort to provide standards for the drug testing community to validate the accuracy of their testing methods, the National Institute of Standards and Technology (NIST), in cooperation with the College of American Pathologists (CAP), has developed human urine-based Standard Reference Materials (SRMs) for 11-nor-Ag-tetrahydrocannabinol-9-carboxylic acid (1), cocaine and benzoylecgonine (2), morphine and codeine (3), morphine glucuronide, and a CAP RM for amphetamines (4). Most of these SRMs are multilevel materials with one level near the National Institute on Drug Abuse (NIDA) specified cutoff level and two other levels. NIST has recently developed a new SRM, Multidrugs of Abuse in Freeze-dried Urine (SRM 1511), with concentrations near the cutoffs for the seven compounds specified in the NIDA guidelines, namely morphine, codeine, benzoylecgonine, am- * Author to whom correspondence should be addressed. phetamine, methamphetamine, phencyclidine (PCP), and 11- nor-ag-tetrahydrocannabinol-9-carboxylic acid. All analytes except amphetamine and methamphetamine were certified (5). CAP has also developed a new three-level lyophilized urine material for PCP. This paper describes the certification of PCP concentrations in SRM 1511 and the new CAP PCP RM. 1-(1-Phenylcyclohexyl)-piperidine, commonly known as PCP, was developed in the 1950s as an anesthetic agent. It was withdrawn from medical use when severe postsurgical psychotic reactions were noted in some patients. PCP has been one of the most commonly abused drugs, either alone or in combination with other drugs, especially marijuana. Its persistence in street drugs is probably because of cheap and easy syntheses of PCP and its derivatives. Detection of PCP use is currently based on the analysis for the parent drug. The drug abuse program of the NIDA includes PCP as a targeted drug for detection and quantitation. For drug-testing of federal employees, the cutoff concentration of PCP was set at 25 ng/ml by NIDA (6). SRM 1511 contains a single concentration of PCP near the cutoff, along with six other compounds specified in the NIDA guidelines. The new CAP RM consists of samples with three concentrations of PCP. The middle concentration (level II) is near the cutoff concentration. The low concentration (level I) is approximately half of the cutoff, and the high concentration (level III) is approximately double the cutoff. In addition, a urine sample with no detectable concentration of the analyte is included as a blank (level 0). A variety of methods have been reported for the determination of PCP in biological samples using gas chromatography-mass spectrometry (GC-MS) (7-12). These methods involve addition of a deuterated internal standard to the urine, followed by liquid-liquid or solid-phase extraction (SPE) to isolate the PCP and the internal standard from the urine. The underivatized PCP is then analyzed by GC-MS. A variety of liquid chromatographic methods have also been reported. These methods involve paired-ion, reversed-phase liquid chromatography (LC) with ultraviolet or fluorescence detection (13-16). NIST has developed two independent methods to certify the Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 43

2 concentrations of PCP in SRM 1511 and the new CAP RM. The first method uses SPE to isolate the PCP from the urine, followed by isotope-dilution GC-MS for measurement. The second method employs a different SPE, followed by isotopedilution LC-thermospray mass spectrometry (LC-MS). The results of an interlaboratory comparison of the measurements of PCP in SRM 1511 conducted among five military laboratories involved in urine drug testing are reported to verify that the material is appropriate for analysis using the methods employed in their large-scale drug-testing program. Materials and Methods Materials PCP as the hydrochloride salt was obtained from a commercial supplier. The purity of this reference compound was evaluated at NIST by direct-probe MS, GC-MS, and differential scanning calorimetry (DSC). Based on these measurements, we judged that the purity was 99 1%; therefore, no corrections were made for impurities. The moisture content of this compound was determined to be % by Karl Fischer analysis at NIST, and no corrections were made. Methanolic solutions of Table I. GC-MS Measurements (ng/ml) of PCP in SRM 1511 Set Vial Ion trap MSD* , , , , , , , , , , 22.9 Set mean SD* CV+ t , , , , , , , , , 25.7 Set mean SD CV , , , , , , , , , , 26.0 Set mean SD CV Overall mean * MSD = Mass selective detector. "l- Standard deviation (SD) of the single measurements within a set. + CV = Coefficient of variation. Given as a percent. Table II. GC-MS Measurements (ng/ml) of PCP in the CAP RM Set Vial/aliquot Ion trap MSD' Level I 1 1/1 12.1, , /1 12.5, , /1 E + E Set mean SD* CVw /1 12.6, , /1 12.0, , /1 12.5, , 12.3 Set mean SD CV Overall mean Level II 1 1/1 24.9, , /1 22.9, , /1 24.3, , 24.0 Set mean SD CV /1 23.2, , /2 24.0, , /I 23.0, , /2 22.7, , /1 23.2, , /2 E E Set mean SD CV Overall mean Level III 1 1/1 52.1, , /1 48.2, , /1 E E Set mean SD CV /1 52.5, , /2 51.7, , /I 52.3, , /2 50.0, , /I 48.9, , /2 49.8, , 53.6 Set mean SD CV Overall mean * MSD = Mass selective detector. ~r An error occurred during processing. r Standard deviation (SD) of the single measurements within a set. w CV = Coefficient of variation. Given as a percent. 44

3 the isotopically labeled internal standard, PCP-d5 hydrochloride, were also obtained from a commercial supplier. The ion-pairing agent, heptanesulfonic acid, was prepared at NIST from a commercially available sodium salt. Sodium heptanesulfonate was suspended in ethanol, and hydrogen chloride gas was bubbled through it. Sodium chloride precipitated out, yielding heptanesulfonic acid in ethanol. The excess hydrogen chloride was allowed to evaporate, and the ethanol was removed. Residual ethanol must be minimized or esterification will gradually occur, rendering the ion-pairing agent ineffective. Bond-Elut Certify TM SPE cartridges were obtained from Analytichem International (Harbor City, CA). Clean Screen DAU SPE cartridges were obtained from World Wide Monitoring Co. (Horsham, PA). An Ultrasphere C18 column (15 cm x 4.6 ram; 5-1Jm particle diameter) was obtained from Beckman (Fullerton, CA). A DB-5 MS fused-silica capillary column (30 m x 0.25-ram i.d.; Jm film thickness) and a DB-1701 fused-silica capillary column (60 m x 0.25-ram i.d.; Jm film thickness) were obtained from J&W Scientific (Rancho Cordova, CA). Solvents used for LC-MS measurements were high-performance liquid chromatographic grade, and all other chemicals were reagent grade. Vials of SRM 1511 and the CAP PCP RM were prepared by commercial suppliers. For each material, a single urine pool was sterile filtered and fortified with appropriate amounts of the analyte. Aliquots were dispensed into vials and lyophilized. GC-MS method The first method used for the determination of PCP was a GC-MS method. For SRM 1511, samples were prepared in three sets, each containing five vials. For the CAP PCP RM, samples were prepared in two sets, each set consisting of three vials each of the three levels. The vials were reconstituted with distilled water (25.0 ml for SRM 1511 and 10.0 ml for the CAP RM) and allowed to stand for 30 rain with occasional swirling to ensure complete dissolution. For SRM 1511, a single 5.0-mL aliquot was taken from each vial. For the CAP RM, duplicate 4.0-mL aliquots were taken from each vial for levels II and III of set 2, and a single aliquot was taken for the rest of the vials. Table III. tc-ms Measurements (ng/ml) of PCP in SRM 1511 Sample Concentration (ng/ml) , , , , , , 23.3 Mean 23.1 SD* 0.80 CV* 3.5 * Standard deviation (SD) of the single measurements within a set. t CV = Coefficient of variation. Given as a percent. Each aliquot was spiked with known amounts of a methanolic solution of the labeled internal standard, PCP-d5, at a concentration of approximately 5 IJg/mL to give approximately 1:1 ratios of analyte to internal standard. Bond-Elut Certify cartridges, which utilize a mixed-mode retention mechanism of ion-exchange and reversed-phase interactions, were used to isolate the PCP from the urine. Samples were processed as recommended in the product literature, except that the final residue was dissolved in 40 ]JL of ethyl acetate. Absolute recoveries of analytes by this extraction averaged 73%. Because the isotope-labeled internal standards added to the samples should be recovered to the same degree as the analytes, no corrections for recoveries were necessary. Standard solutions of PCP were prepared as follows: I~vo independently weighed stock solutions of PCP, at concentrations of approximately 3-4 IJg/mL in methanol, were prepared. Five standards were prepared for each sample set by mixing appropriate amounts of the unlabeled stock and labeled internal standard solutions to obtain the weight ratios of unlabeled-tolabeled compounds ranging between 0.6 and 1.3. The mixtures were dried under nitrogen at 40~ then reconstituted Table IV. LC-MS Measurements (ng/mt) of PCP in the CAP RM Concentration Level Vial (ng/mt) Mean SD* CV ~ 11.3, 11.4, , , 11.5, , , 11.0, , 10.8, II , 21 =3, 22.4 II , 22.7, 23.8 II , 22.9 II , 24.6, 23.4 II , 22.7, 21.9 II , 21.7,21.4 Mean 22.7 SD 1.02 CV 4.5 III , 47.5, 47.9 III , 51.2, 47.6 III , 47.1, 45.8 III , 47.4, 45.1 III , 42.4, 45.8 III , 48.5, 46.1 Mean 46.7 SD 1.92 CV 4.1 * Standard deviation (SD) of the single measurements within a set. f CV = Coefficient of variation. Given as a percent. 45

4 with 40 IJL ethyl acetate for the GC-MS analyses. No derivatization was involved in the GC-MS method. Each sample was analyzed on two different mass spectrometers equipped with two different GC columns. The first mass spectrometer was a Finnigan ITMS ion-trap mass spectrometer with a 60-m DB-1701 fused-silica capillary column (a moderately polar phase, 14% (cyanopropylphenyl)-methylpolysiloxane) connected directly to the ion source. The GC injector and transfer line were heated to 200~ Helium was used as the carrier gas, and the head pressure was set at 100 kpa (15 psi). The relays were set for a splitless injection, opening the split valve at 1 rain. The temperature of the column was initially held at 99~ for 0.5 min and then programmed to 280~ at a rate of 40~ The mass spectrometer was operated in the electron-ionization mode at 70 ev under automatic gain control (AGC) conditions. The temperature of the housing was set at 120~ The sensitivity parameter B was set at 4000, and the scan mode was full scan at m/z at 0.25 s/scan. The multiplier voltage was 1900 V, and the AGC background mass was 197. The ion chromatograms for the fragment ions at m/z 200 and corresponding ions at m/z 205 were integrated for PCP because they were the most intense of the higher mass ions, and no interferences were evident. The second mass spectrometer used was a Hewlett-Packard 5971 mass selective detector (MSD) with a 30-m DB-5 MS fused-silica capillary column (a nonpolar phase, 95% dimethyl polysiloxane-5% phenylmethyl polysiloxane) connected directly to the ion source. The GC injector and transfer lines were heated to 200~ Helium was used as the carrier gas, and the head pressure was set at 100 kpa (15 psi). The relays were set for a splitless injection, opening the split valve at 0.75 min. The temperature of the column was initially held at 140~ for 0.5 min and then programmed to 280~ at a rate of 30~ The mass spectrometer was operated in the electron-ionization mode with an ionization energy of 70 ev. As with the measurements using ITMS ion trap, the same fragment ions at m/z 200 and 205 were measured for PCP. The mass spectrometer was operated in the selected-ion monitoring mode, dwelling on each ion for 0.1 s per cycle. For each set of samples run on each mass spectrometer, the following measurement protocol was used. A single analysis of each of the five standards in order of increasing weight ratio was run first. Subsequently, duplicate analyses of each sample were run. Finally, the five standards were run again in reverse order. By combining the data of the standards run before and after the samples, a composite linear regression was calcu- lated; it was used to convert the measured intensity ratios of analytes to weight ratios. The weight ratios were then used along with the amounts of the internal standard added to calculate analyte concentrations. LC-MS method The second method for the determination of PCP was based on LC-MS. For SRM 1511, six vials were prepared in one set. For the CAP PCP RM, samples were prepared in three sets, each set consisting of six vials of one of the three levels. Each vial was reconstituted as previously described. For SRM 1511, a single 10.0-mL aliquot was taken from each vial and spiked with a known amount of the internal standard (PCP-ds) to a concentration approximating the expected level of the analyte in the sample (i.e., so that weight ratios of unlabeled-to-labeled compounds were approximately 1:1). For the CAP RM, the entire contents of each vial were spiked with PCP-ds. Clean Screen DAU cartridges, where retention is based on a mechanism similar to the Bond-Elut Certify cartridges as previously described, were used to extract the PCP from the Urine. Each sample was adjusted to a ph of approximately 6.0 by adding 4.0 ml 0.1M potassium phosphate buffer (ph 6.0). Samples were processed as i'ecommended in the product literature, exceptthat the PCP in the CAP RM was eluted from the cartridge with a solvent of ethyl acetate-concentrated ammonium hydroxide (98:2) instead of the solvent of dichloromethane-2-propanol-concentrated ammonium hydroxide (78:20:2) to eliminate interference peaks coeluting with PCP. The eluents were dried under nitrogen at approximately 35~ and reconstituted with 250 IJL of the mobile phase for the LC-MS analyses. Standard solutions of PCP were prepared from three independently weighed stock solutions at concentrations of approximately 2.5 I~g/mL in methanol. For SRM 1511, calibration mixtures were made by spiking 10 ml of water with appropriate amounts of the unlabeled stock and labeled internal solutions to give weight ratios of unlabeled-to-labeled compounds ranging between 0.8 and 1.3. For the CAP PCP RM, calibration mixtures were made by spiking the contents of the reconstituted blank urine with appropriate amounts of the unlabeled stock and labeled internal solutions to give weight ratios of unlabeled-to-labeled compounds ranging between 0.8 and 1.2. Two aliquots from each of the three unlabeled stock solutions were taken, yielding six standards. Each standard was then subjected to the DAU SPE described above for sample preparation. The residue was reconstituted in the mobile phase. Table V. Comparison of PCP Results from the Methods and Certified Values for SRM 1511 Mean results ion trap MSD versus versus GC-ion trap GC-MSD LC-MS LC LC Certified value (ng/ml) (ng/ml) (ng/ml) (%) (%) (ng/ml) overall mean difference = +3.9% 46

5 Aliquots of IJL of standard or extract were separated by LC on a Beckman Ultrasphere C18 column with an isocratic mobile phase consisting of 1raM heptanesulfonic acid, 20raM ammonium acetate, and 3.5% glacial acetic acid in methanolwater (55:45). The flow rate was I mumin. The Vestec model 201 thermospray interfaced to a quadrupole mass spectrometer was used to monitor the (M+H) ions at m/z 244 and 249 for PCP and PCP-ds,.respectively. Thermospray ionization was used (i.e., discharge and electron ionization turned off). The block temperature was 275~ and the thermospray control setting was optimized for sensitivity and stability. For SRM 1511, each sample was analyzed on 2 days. Standards were run alternately with samples throughout each day's analyses, resulting in one replicate of each sample and one to two replicates of each standard. For the CAP RM, each level was analyzed in I day. Standards were run alternately with samples, resulting in two to three replicates of each sample and two to three replicates of each standard for each level. A mean response factor calculated from all the standards run on a given day was used together with the measured ratio for each sample to calculate the analyte concentration. The limit of detection for PCP with this method is estimated to be 0.5 ng/ml. GC-MS methods of the five laboratories All laboratories were affiliated with the Department of Defense Drug Testing Program. GC-MS methods for the determination of PCP were used by all five laboratories. The methods involved addition of a deuterated internal standard to the reconstituted urine, followed by liquid-liquid extraction to isolate the PCP from the urine. The underivatized PCP was then analyzed by GC-MS. Results and Discussion The concentrations of PCP in SRM 1511 and the CAP RM were certified at NIST, in cooperation with the CAP. Certifica- tion is the process of determining the concentrations of anaiytes in reference materials and evaluating the analytical data to assign uncertainties. The approach used to certify these two reference materials involved measurements by two different analytical methods. Agreement among these methods minimizes the likelihood of undetected bias. GC-MS is the most commonly used method to positively identify and quantitate drugs and their metabolites in urine, and therefore, it was employed as the primary method for the determination of PCP. PCP was selectively isolated from urine components using an SPE column with a mixed-mode bonded phase and then determined by an isotope-dilution GC-MS method. Samples were prepared and analyzed in three independent sets for SRM 1511 and two sets for the CAP PCP RM. Each sample was analyzed on two different GC-MS instruments, an ion-trap mass spectrometer with a moderately polar GC column and an MSD with a nonpolar column. The data for the GC-MS measurements of PCP are shown in Tables I and II for SRM 1511 and the CAP RM, respectively. Good precision was obtained for both materials, and coefficients of variation (CVs) of the single measurements within a set ranged from 1.6 to 5.8%. The resuits from the two instruments with two different columns agreed to within a few percent. The second method used in the determination of PCP concentration was LC-MS. Samples were processed using SPE columns, which have a retention mode similar to the type used for the GC-MS method. The PCP was determined by isotope-dilution MS with thermospray ionization. The results for the LC-MS measurements of PCP are shown in Tables III and IV for SRM 1511 and the CAP RM, respectively. Good precision was obtained for both materials, and the CV of the single measurements within a set ranged from 3.5 to 4.6%. The mean results for PCP in SRM 1511 from two different methods are summarized and compared in Table V. The data in this table show that the LC-MS and GC-MS results were in good agreement, and there was an overall mean relative difference of 3.9%. The small differences observed were generally Table Vl. Comparison of PCP Results from the Methods, Certified Values, and Components of Variance for PCP in the CAP RM Mean results ion trap MSD versus versus Certified Components of variance (%) GC-ion trap GC-MSD LC-M5 LC LC value method set vial random Level (ng/ml) (ng/ml) (ng/ml) (%) (%) (ng/mt) CV CV CV error CV l ,2 +10, * II * III * ' ' 2.7 Mean ignificant component (effect) at alpha equal to overall mean difference = +8.4% 47

6 Table VII. Round-Robin Results for PCP in SRM 1511 among Five Military Laboratories* Laboratory Concentration (ng/ml) Mean 23.1 SD t 1.2 * Certified value = nglml. "I- One standard deviation (SD) of the single reported results from each of the five military laboratories. The urine blank that is included as part of the CAP PCP RM is not certified for the concentration of PCP. However, samples from the reconstituted blank were extracted and analyzed using both GC-MS and LC-MS methods. No PCP was detected in the urine blank at limits of detection of 0.3 and 0.5 ng/ml for GC-MS and LC-MS, respectively. No urine blank is available for SRM Conclusion The agreement among the results from our two methods of analysis provided evidence for the accuracy of the certified values. The agreement of results from the military laboratories verified that the SRM can be analyzed accurately in the field using routine drug-testing methodologies. within the imprecision of these methods. The certified values for PCP as the free base form for SRM 1511 are also listed in Table V. For GC-MS, there were three sets of data. For each set, the ion-trap and MSD data were pooled to yield one set mean. For LC-MS, all the data within the set were averaged to yield one set mean. Based on these four set means, a two-sided, 95% confidence interval was computed based on the t distribution and the estimated standard error. The mean results for PCP in the CAP RM from two different methods were summarized and compared in Table VI. The data in this table show that the GC-MS and LC-MS results were in acceptable agreement with an overall mean relative difference of 8.4%. For level II, the level closest to the NIDA cutoff of 25 ng/ml, the agreement was particularly good, and there was a mean difference of 4.8% between the two methods. Overall, the LC-MS results tended to be slightly lower than either the ion-trap or MSD results for reasons that are not understood. Nested analyses of variance models were used to budget the variance into the respective factors: method, set, vial, and random errors. Table VI lists these components of variance and indicates tliose factors that are statistically significant. For levels I and III,almost all the error was due to the method differences. The set factor for level II and the vial factor for level III were marginally significant and likely to be due to random error or the modeling techniques. Finally, 95% confidence intervals were constructed around the overall means incorporating the method differences. These confidence intervals encompassed any methodology, set, and vial differences for each level. The intervals, based upon the standard error of the mean, reflect the quality of the certified values and imply that with 95% confidence the true concentration will be contained in the interval. Table VI also lists the certified values for PCP as the free base form and their 95% confidence intervals for the CAP RM. A round-robin analysis of PCP in SRM 1511 was conducted among five military laboratories involved in urine drug testing to verify the usefulness of this material in field laboratories. A variety of GC-MS methods were used by the five laboratories, and the results are given in Table VII. They were in good agreement with the NIST certified values. Acknowledgments We gratefully acknowledge the support of S.S.-C. Tai and P. Ellerbe by the College of American Pathologists. We thank Dr. S. Margolis for the Karl Fisher analyses and Ms. R. Parris for the DSC analyses. Note: Certain commercial equipment, instruments, and materials are identified in this paper to adequately specify the experimental procedure. Such identification does not imply recommendation or endorsement by NIST nor does it imply that the equipment, instruments, or materials are necessarily the best available for the purpose. References I. N.E. Craft, G.D. Byrd, and L.R. Hilpert. Preparation and certification of Standard Reference Material 1507: 11-nor-delta-9- tetrahydrocannabinol-9-carboxylic acid in freeze-dried urine. Anal. Chem. 61' (I 989). 2. P. Ellerbe, S.S.-C. Tai, R.G. Christensen, R. Espinosa-Leniz, R.C. Paule, L.C. Sander, L.T. Sniegoski, M.J. Welch, and E. White, V. The certification of cocaine and benzoylecgonine in a human urine Standard Reference Material. J. Anal. Toxicol. 16: (1992). 3. S.S.-C. Tai, R.G. Christensen, R.C. Paule, L.C. Sander, and M.J. Welch. The certification of morphine and codeine in a human urine Standard Reference Material. J. Anal. Toxicol. 18" 7-12 (I 994). 4. P. Ellerbe, T. Long, and M.J. Welch. The determination of amphetamine and methamphetamine in a lyophilized human urine reference material. J. Anal. Toxicol. 17: (I 993). 5. M.J. Welch, P. Ellerbe, S.S.-C. Tai, R.G. Christensen, L.T. Sniegoski, L.C. Sander, and C.S. Phinney. NIST reference materials to support accuracy in drug testing. Fresenius Z. Anal. Chem. 350: (1994). 6. Department of Health and Human Services. Mandatory guidelines for federal workplace drug testing programs. Fed. Regist. 53(69): (1988). 7. Y.H. Caplan and B. Levine. Abbott phencyclidine and barbiturates abused drug assays: evaluation and comparison of ADx FPIA, TDx FPIA, EMIT, and GC/MS methods. J. Anal. Toxicol. 13: (1989). 48

7 8. S.J. Mul~ and G.A. Casella. Confirmation of marijuana, cocaine, morphine, codeine, amphetamine, methamphetamine, phencyclidine by GC/MS in urine following immunoassay screening. J. Anal. Toxicol. 12: (1988). 9. B.R. Kuhnert, B.G. Bagby, and N.L. Golden. Measurement of phencyclidine and two hydroxylated metabolites by selected ion monitoring. ]. Chromatogr. 276: (1983). 10. K.E. Brooks and N.B. Smith. Versatile, efficient system for extracting drugs from urine for gas chromatographic-mass spectrometric analysis. Clin. Chem. 35: (1989). 11. E.J. Cone, W. Buchwald, and D. Yousefnejad. Simultaneous determination of phencyclidine and monohydroxylated metabolites in urine of man by gas chromatography-mass fragmentography with methane chemical ionization. J. Chromatogr. 223: (1981). 12. D. Berberich and M. Uhrich. Phencyclidine: identification and quantitation with the ITS40 (application data sheet no. 39). Finnigan, San Jose, CA, K.G. Rao and S.K. Soni. Separation and identification of phency- clidine and some of its analogs. J. Assoc. Off. Anal. Chem. 66: (1983). 14. L.A. Jones, R.W. Beaver, and T.L. Schmoeger. Optimization of a paired-ion reversed-phase liquid chromatographic separation of synthetic phencyclidine mixtures. Anal. Chem. 54: (1982). 15. H. Derendorf and E.R. Garrett. High-performance liquid chromatographic assay of methadone, phencyclidine, and metabolites by postcolumn ion-pair extraction and on-line fluorescent detection of the counterion with application. J. Pharm. Sci. 72: (1983). 16. T. Nabeshima, K. Yamaguchi, H. Fukaya, M. Hiramatsu, H. Furukawa, and T. Kameyama. Simultaneous determination of phencyclidine and its major metabolites in biological samples by high-performance liquid chromatography. Res. Commun. Substance Abuse 6:65-78 (1985). Manuscript received January 24, 1995; revision received May 9,