Introduction. Abstract

Transcription

1 A Comparison of Roche Kinetic Interaction of Microparticles in Solution (KIMS ) Assay for Cannabinoids and GC MS Analysis for 11-nor-9-Carboxy- 9 -Tetrahydrocannabinol * Timothy P. Lyons 1,, Catherine K. Okano 1, Judith A. Kuhnle 1, Mark R. Bruins 1, William D. Darwin 2, Eric T. Moolchan 2, and Marilyn A. Huestis 2 1 Forensic Toxicology Drug Testing Laboratory, Tripler Army Medical Center, Honolulu, Hawaii and 2 Chemistry and Drug Metabolism, IRP, NIDA, NIH, 5500 Nathan Shock Drive, Baltimore, Maryland Abstract In this study, we investigated the effectiveness of the Roche Kinetic Interaction of Microparticles in Solution (KIMS) screening assay for cannabinoid metabolites. Urine specimens (N = 1689) were collected during elimination of cannabinoids from 25 subjects with a history of marijuana use. Specimens were analyzed concurrently for cannabinoid metabolites by a customized Department of Defense (DOD) cannabinoid KIMS kit (50-ng/mL cutoff) and for 11-nor-9-carboxy- 9 - tetrahydrocannabinol (THC-COOH) by GC MS (15-ng/mL cutoff). As compared to GC MS results, the sensitivity, specificity, and efficiency of the KIMS assay were 69.7%, 99.8%, and 88.6%, respectively. Many of the false-negative results had GC MS concentrations between 15 and 26 ng/ml (N = 151). The cannabinoid screening results for the DOD samples tested by the laboratory during the same 8-month period were also evaluated. The linear regression analyses of GC MS results in the ng/ml range and KIMS data resulted in regression coefficients of for the research specimens and for DOD specimens. The results suggest that the KIMS cannabinoid screening assay is deficient in detecting positives around the cutoff (15 25 ng/ml THC-COOH). This limitation of the KIMS cannabinoid screening method compromises the identification of true positive specimens, therefore reducing the effectiveness of the assay. The success of the DOD program is dependent on sensitive and specific screening assays; the high prevalence of false-negative cannabinoid results compromises the program s primary objective of drug deterrence. * The views expressed in this manuscript are those of the authors and do not reflect the official policy or position of the Department of the Army, Department of Defense, or the U.S. Government. Author to whom correspondence should be addressed. Timothy.Lyons@haw.tamc.amedd.army.mil. Introduction In a high sample volume environment, a reliable screening procedure is essential for a successful drug-testing program. An effective screening assay identifies negative specimens quickly and inexpensively and minimizes unconfirmed positives. This assay must also have specificity to the targeted compounds and sensitivity to the concentration of the metabolites at the desired cutoff. These performance characteristics determine the efficiency of the assay in identifying true-positive and truenegative results. The Department of Defense (DOD) Drug Testing Program utilized radioimmunoassay (RIA) as its primary screening method prior to January At that time, in an effort to increase productivity and testing capabilities, all DOD laboratories implemented the Roche (Montclair, NJ) Online Kinetic Interaction of Microparticles in Solution (KIMS) immunoassay as the standard screening procedure. This method is still employed as a screening assay method for some drugs of abuse within the DOD program, although a different immunoassay for cannnabinoid testing is now in use. The KIMS procedure uses a drug bound microparticle conjugate that forms a lattice when antibody to the drug is added (1,2). Absorbance increases as the conjugate binds to the antibody. When a urine sample containing the drug of interest is added, free drug in the sample competes with the conjugate for antibody binding sites. This inhibits lattice formation and decreases absorbance proportional to the amount of free drug (1,2). The effectiveness of this assay is also dependent on the established screening and confirmation cutoff values. The screening cutoff in both military and civilian programs is 50 ng/ml for cannabinoid metabolites. Urine analysis samples that equal or exceed the 50-ng/mL cutoff for the screening assay are presumptive positive and confirmed by gas 1 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher s permission. 559

2 chromatography mass spectrometry (GC MS) at a concentration of 15 ng/ml for 11-nor-9 carboxy- 9 -tetrahydocannabinol (THC-COOH). The performance, primarily the sensitivity, of the KIMS cannabinoid immunoassay was questioned after only a few months of use. It was determined that the first generation kit lacked the capability to detect positive samples near the cutoff concentration. Because of these concerns, DOD requested that the cannabinoid kit be reformulated. Reformulation required development of a new monoclonal antibody with broad specificity for cannabinoid metabolites (2). The reformulated assay was introduced in April Since introduction of the reformulated kit, DOD drug laboratories have screened over 6,000,000 urine samples for cannabinoid metabolites. There are no studies that have evaluated the effectiveness of the DOD KIMS assay for cannabinoids. In this study, we investigated the reliability of the current KIMS kit as a screening assay for the detection of cannabinoids at the 50-ng/mL cutoff. We analyzed 1689 urine samples simultaneously for cannabinoids by KIMS (50-ng/mL cutoff) and for THC-COOH by GC MS (15-ng/mL cutoff). Urine specimens were collected in a controlled smoked marijuana administration study. The screening assay s sensitivity, specificity, and efficiency were determined. Regression analyses between screening and confirmation results were conducted and compared for both the research specimens and actual DOD urine analysis program specimens. Materials and Methods Protocol Specimens were collected in conjunction with a research protocol conducted by the Chemistry and Drug Metabolism Section of the Intramural Research Program, NIDA on excretion of cannabinoids in marijuana users. The subjects (n = 25, mean age 26.3 ± 6.4 years) provided informed consent, resided on the secure research unit throughout the study, and were compensated for their participation. Consideration for participation was based on the following inclusion criteria: (1) must be between 21 and 45 years old; (2) must be an experienced, current marijuana user and have smoked or ingested marijuana at least once per week for a minimum of 4 weeks by self-report; and (3) must have a positive cannabinoid urine test prior to admission into the study to confirm self-reported drug use. All urine specimens were collected throughout the study. Once urine cannabinoid values dropped below 20 ng/ml by immunoassay (Microgenics Corporation, Fremont, CA) subjects smoked one or two 2.64% THC cigarettes. Some individuals were medically disqualified or chose to abstain from smoking the second marijuana cigarette. Four hours elapsed between the smoking of the first and second cigarettes. Sample collection terminated when the individual was discharged from the clinical research ward a minimum of 48 h after the last smoked dose. Both food and fluid intake were ad libitum, and both were provided by the research ward. After specimen mixing, volume measurement, and removal of all patient-identifying information, an aliquot was frozen at 20 C for later shipment to the U.S. Army Tripler Forensic Toxicology Drug Testing Laboratory. Specimens were analyzed in a randomized and blind fashion and remained frozen until simultaneous analysis by immunoassay and GC MS for cannabinoid metabolites. Screening procedures The samples were assayed accordingly to manufacturer s instructions for cannabinoid metabolites on the AU 800 analyzer (Olympus, Irving, TX) using the specially formulated DOD Roche Online KIMS kit with a 50-ng/mL cutoff. The results from the analyzer are normalized to a value of 100, which corresponds to the cutoff value of 50 ng/ml. The analyzer was calibrated at the cutoff prior to each analysis. The first two samples of each batch were the high (135% of the cutoff) and low (65% of the cutoff) open quality controls. These open controls were repeated after every 48 specimens. One blind positive (150% of the cutoff) and one blind negative control were also inserted randomly between each set of open controls. A linearity study was conducted for the KIMS assay. Urine quality-control samples with 6, 15, 20, 25, 30, 37, 40, 50, 67, 80, and 100 ng/ml of THC-COOH were assayed in three separate batches. The corresponding normalized KIMS values were averaged for each concentration and plotted. GC MS procedures Reagents/cartridges. All reagents were analytical grade or better. The following chemicals were obtained: methanol, ethyl acetate, hydrochloric acid, and potassium hydroxide were manufactured by Mallinckrodt Baker, Inc. (Paris, KY); hexane, acetic acid, and 1-iodomethane (EM Science, Gibbstown, NJ); dimethylsulfoxide and isooctane (Burdick & Jackson Allied Signal, Muskegon, MI); acetone (Baker, Phillpsburg, NJ); acetonitrile, bovine serum albumin, and sodium azide (Sigma- Aldrich, Milwaukee, WI); and tetramethylammonium hydroxide (Janssen Chemica, Geel, Belgium). The solid-phase extraction (SPE) cartridges (IST Isolute, C8, 100 mg, 1 ml) were manufactured by International Sorbent Technology Ltd. (Hengoed, Mid Glamorgan, U.K.). Standards and controls. Ethanolic solutions (100 µg/ml) of THC-COOH and THC-COOH-d 3 standards were purchased from Research Triangle Institute (Research Triangle Park, NC). Urine standards and controls were prepared using pooled human urine that contained 1 g/l sodium azide and 250 mg/l bovine serum albumin. The standards and controls were stored in the refrigerator at 4 C prior to use. A single-point 15-ng/mL THC-COOH calibrator was used to define the GC MS cutoff. A low quality control (6 ng/ml), blind positive control (20 ng/ml), and blind negative control were also included in each batch to monitor assay performance. All standards and controls were prepared as stock solutions, refrigerated prior to use, and spiked into negative urine at the time of analysis. A 0.3-µg/mL THC-COOH-d 3 internal standard was prepared in methanol. The concentration of the calibrator was verified by analyzing a minimum of two levels of National Institute of Standards and Technology (NIST, Gaithersburg, MD) standards and ensuring quantitation within ± 20% of target concentrations. 560

3 Extraction equipment/batch set-up. The RapidTrace SPE Workstation was used to automate the extraction process (Zymark Corporation, Hopkinton, MA). Five linked extraction modules, each containing 10 sample positions, were used for each batch of specimens; therefore, up to 50 specimens, qualitycontrol samples and deionized water blanks were processed in each batch. The 6- and 20-ng/mL THCCOOH quality-control samples combined to contribute 8 16% of the batch. A lack of carryover up to 500 ng/ml was documented during method validation. Specimens greater than 500 ng/ml were diluted and re-extracted. Extraction procedure. Internal standard (0.15 ml of 0.3 mg/ml THC-COOH-d 3 ) was added to each 3 ml urine specimen in a mm test tube. Alkaline hydrolysis was accomplished with the addition of 0.25 ml of 10N KOH, vortex mixing, and incubation for 15 min at C in a water bath. After the samples were cooled to room temperature, 0.5 ml of glacial acetic acid was added to each sample followed by centrifugation. The specimens were decanted into mm test tubes and placed on the RapidTrace for processing. On the RapidTrace, SPE columns were conditioned with 2 ml of methanol, 2 ml of deionized water, and 2 ml of 0.10M HCl. Hydrolyzed samples were loaded onto the columns and rinsed with 2 ml of water followed by 2 ml of acetonitrile/acetone/hcl (15:15:70). The column was dried for 1 min with nitrogen. THC-COOH was eluted with 2 ml of hexane/ethyl acetate (50:50). Extracts were transferred to 15-mL conical tubes and evaporated to dryness at C using a TurboVapLV (Zymark). Derivatization. A previously described derivatization procedure with minor modifications was used to methylate THC- COOH (3). Tetramethylammonium hydroxide (25% TMAH in methanol) and dimethylsulfoxide (100 µl, 1:20, v/v) were added to reconstitute the residue. This mixture was vortex mixed for 30 s and allowed to stand at room temperature for 2 min. After the addition of 20 µl of 1-iodomethane, vortex mixing, and standing for 5 min, 0.2 ml of 0.1N HCl and 1.0 ml of isooctane were added to the samples. Following additional vortex mixing, the isooctane phase was allowed to separate, and later transferred to a mm test tube. The tubes were placed in a water bath at C and evaporated to dryness under a gentle stream of nitrogen. The residue was dissolved in 40 µl of isooctane and 1-2 µl was injected into the GC MS. GC MS analysis A model 5890 series II Hewlett-Packard GC equipped with a G1513A autosampler coupled to a 5972 mass selective detector (MSD) was used for analysis. The GC was equipped with a Hewlett-Packard Ultra 2 cross-linked 5% phenyl methyl siloxane (12 m 0.20-mm i.d., 0.33-µm film thickness) column. The specimens were analyzed using a split injection (1:15), injector temp, 265 C, and an isothermal oven temperature of 260 C. All specimens that quantitated below 3 ng/ml were reinjected with calibrators and controls from the original batch under splitless conditions using the following conditions: injector temperature 275 C, initial oven temp 200 C, initial time 0.70 min, then a 30 C/min ramp to 280 C. Specimens were quantitated using the THC-COOH-d 3 internal standard. The instrument was operated in the SIM mode monitoring m/z 372, 357, and 313 and m/z 375 and 360 for methyl derivatives of THC-COOH and the internal standard, respectively. Data acquisition and analysis were performed with Hewlett-Packard G1034C Chem Station software in the drug analysis mode. A one-point calibration at 15 ng/ml was obtained from the m/z 372/375 ions. Retention times and mass ion ratios of unknown and quality control specimens had to be within ± 2% and ± 20%, respectively, of the calibrator values to be accepted. In addition, quality control samples were required to quantitate within ± 20% of their target concentrations for batch acceptability. The linearity ranges determined by serial dilution were ng/ml and 1 20 ng/ml for the split and splitless injection conditions, respectively. Data and statistical analysis Performance criteria for the immunoassay were calculated by the following formulas: Sensitivity = [true positives/(true positives + false negatives)] 100 Specificity = [true positives/(true negatives + false positives)] 100 Efficiency = [(true positives + true negatives)/n] 100 True positives are specimens determined to be positive by both immunoassay ( 50 ng/ml) and GC MS ( 15 ng/ml). True negatives are negative by both immunoassay (< 50 ng/ml) and GC MS (< 15 ng/ml). False positives are positive by immunoassay and negative by GC MS. False negatives are negative by immunoassay and positive by GC MS. Regression analysis was also conducted on the screening and GC MS urine results for the research specimens and results from actual testing of DOD specimens during the same time period (Sigmastat, Jandel Scientific, Chicago, IL). Results The laboratory analyzed a total of 1689 urine samples from the NIDA research study simultaneously for cannabinoids by the KIMS immunoassay and by GC MS. The testing occurred over a period of eight months from October 1998 to July During this period, the laboratory continued to fulfill its primary mission of testing specimens for the DOD drug-testing program. Performance characteristics and test results are listed in Tables I and II. Of those samples analyzed, 63% were true negatives and 26% true positives ranging in concentration from 15 to 773 ng/ml by GC MS. The KIMS assay produced 11.3% false-negative tests and only 0.5% false-positive tests. The sensitivity, specificity, and efficiency of the KIMS assay for cannabinoid metabolites were 69.7%, 99.8%, and 88.6%, respectively. We determined that 366 samples had THC-COOH GC MS concentrations of 15 to 42 ng/ml; 190 (51.9%) of these were negative by the KIMS cannabinoid assay (Figure 1). The limits of linearity of the KIMS cannabinoid assay were 6 to 80 ng/ml THC-COOH (Figure 2). The linear regression analysis of the GC MS and KIMS data from the NIDA research specimens for all GC MS results be- 561

4 tween 15 and 42 ng/ml yielded a regression coefficient of (Figure 3A). For comparison, the linear regression analysis on test results from actual DOD specimens tested during the same time period and in the range of ng/ml THC-COOH (Figure 3B) was conducted, yielding a regression coefficient of The regression coefficients for GC MS results in the ng/ml range were and for NIDA and DOD specimens, respectively (Figure 4). A comparison of the regression lines using 95% confidence intervals demonstrated that the lines are similar for the research and DOD samples for both the ng/ml and ng/ml ranges. They also illustrate the tendency for the correlation between the KIMS and GC MS assays to decrease as concentrations exceed the linear range of the KIMS assay. The mean precision for the 6 and 20 ng/ml GC MS qualitycontrol samples included in batches of specimens analyzed in this study were 5.99 (%CV = 6.33, n = 105) and (%CV = 4.63, n = 160), respectively. Discussion The main objective of the immunoassay screen in the DOD drug-testing program is to identify presumptive-positive specimens. This step of the testing program is critical in maintaining a high level of drug deterrence at a reasonable cost. Table I. Performance Characteristics for the Reformulated DOD Roche Online KIMS Immunoassay for Cannabinoids in Urine (N = 1689) Sensitivity % 69.7 Specificity % 99.8 Efficiency % 88.6 Figure 2. Linearity data for reformulated DOD Roche Online KIMS immunoassay for cannabinoids in urine. Data points are means of three separate runs. Table II. Summary of Results for DOD Roche Online KIMS Cannabinoid Immunoassay (50-ng/mL Cutoff) versus Results from GC MS (15-ng/mL THC-COOH Cutoff) GC MS 15 ng/ml + KIMS (TP*) 2 (FP) 50 ng/ml 190 (FN) 1058 (TN) * Abbreviations: TP, true positive; FP, false positive; FN, false negative; TN, true negative. Figure 1. Number of specimens negative by immunoassay (50-ng/mL cutoff) and positive by GC MS (15-ng/mL THC-COOH cutoff). N = 190. Figure 3. Regression analysis of GC MS and immunoassay data: ng/ml range for research samples (A) and ng/ml range for DOD samples (B) from October 1998 to July * Data is normalized, reading of 100 = 50-ng/mL cutoff. 562

5 High assay sensitivity and specificity are needed to ensure that true positive specimens are identified, and, to limit unnecessary testing of specimens that will not confirm at 15 ng/ml by the labor intensive and expensive GC MS assay. Ideally, there would be complete agreement between the two assays at the required cutoffs; however, this is rarely achieved. Minimization of falsepositive (unconfirmed presumptive positive screening results) and false-negative (true positives missed by the screening assay) results is the goal. A strong correlation between the KIMS screening and GC MS confirmation results also is helpful in minimizing false-positive screening tests that can over burden the confirmation section of the laboratory. This study provided a unique opportunity to evaluate the specially-formulated DOD KIMS assay. The relatively large number of individuals and the numerous samples collected per test subject resulted in an uncommonly large sample population. This permitted evaluation of diverse cannabinoid excretion patterns and metabolite profiles. Previous studies have demonstrated that many factors contribute to the metabolism and excretion of cannabinoids (4 7), including drug use history, individual pharmacokinetic variations, and sampling frequency (5,7 10). At least 18 cannabinoid metabolites have been identified in human urine after marijuana smoking; THC-COOH Figure 4. Regression analysis of GC MS and immunoassay data: ng/ml range for research samples (A) and ng/ml range for DOD samples (B) from Oct 1998 to July * Data is normalized, reading of 100 = 50 ng/ml cutoff. and its glucuronide conjugate constitute approximately 27% of total cannabinoid metabolites (11 13). An extensive study is necessary to adequately address these factors and to accurately evaluate an immunoassay s sensitivity and specificity (5). The considerable resources of this high-volume laboratory also facilitated the simultaneous testing of all specimens by both KIMS and GC MS, eliminating potential differences in results due to possible specimen degradation that could occur if samples were tested at different times. In this study, we evaluated the KIMS cannabinoid screening assay and its relationship to THC-COOH GC MS results. After screening and GC MS analysis of 1689 specimens, it was observed that the KIMS cannabinoid method had a limited ability to identify specimens containing between 15 and 25 ng/ml THC-COOH; many cannabinoid true-positive specimens were undetected by the screening test. The number of false-negative tests in this range exceeded 50% and contributed to the low sensitivity of the assay (69.7%), although it is within the range previously reported for other cannabinoid immunoassays (5,14). This trend, over time, could lead to significant numbers of undetected positive specimens throughout the DOD program. In contrast, the KIMS cannabinoid assay was very effective in identifying negative specimens; few specimens produced positive screening tests that were not confirmed by GC MS. The high specificity (99.8%) may be due to selection of an antibody that primarily targets THC-COOH and has lower cross-reactivity to other cannabinoid metabolites (5,7). Other cannabinoid immunoassays were previously reported to have specificities in the range of 95.0 to 98.5% (5). Almost all of the specimens that were subjected to the expensive and labor-intensive GC MS procedures were confirmed positive. As a result, a high-specificity assay is economically advantageous, especially in a highvolume production laboratory. The assay appears to be linear around the cutoff and up to approximately 80 ng/ml; the assay is non-linear above this concentration, indicating a limited dynamic range for the cannabinoid method. This indicates that the assay is optimized for performance around the cutoff concentration of 50 ng/ml, but may be less useful for prediction of required dilutions for GC MS analysis. These data demonstrate that the efficiency of the KIMS cannabinoid assay (88.6%) is similar to efficiencies reported for other cannabinoid immunoassays (5). The sensitivity and specificity of the assay have a direct impact on the efficiency, and subsequently the correlation, between the KIMS and GC MS results (7,14). A reduced efficiency and a resulting weak correlation would translate to the KIMS being a poor predictor of the final GC MS result. The regression coefficients for the ng/ml range were low (0.680 and 0.546) for both the research and DOD specimens, but were comparable to correlation coefficients previously reported for radioimmunoassay and GC MS results (7). The strength of the correlation between the KIMS and GC MS THC-COOH concentrations is primarily dependent on two factors: (1) the specificity of the reagent antibodies and (2) the consistency of cannabinoid metabolite ratios across specimens (7,14). According to the package insert, the KIMS kit is formulated to have antibodies that have cross-reactivity to various cannabinoid metabolites including 11-hydroxy-THC at 563

6 90%, 8-α-hydroxy-THC at 53%, 8-β-11-dihydroxy-THC at 41%, 11-hydroxycannabinol at 10%, and THC at 8% (15). In contrast, the GC MS analysis is specific for the THC-COOH metabolite. These differences in specificity are partially taken into account by requiring a 50-ng/mL screening cutoff for cannabinoid metabolites and a 15-ng/mL GC MS cutoff for THC-COOH. To our knowledge, this is the first study evaluating the performance characteristics of the newly reformulated DOD KIMS cannabinoid assay. Our results suggest that the KIMS cannabinoid screening assay, although as efficient as other cannabinoid screening assays, has a limited capacity to detect positives in the ng/ml THC-COOH range. This limitation compromises the identification of true-positive specimens. A high prevalence of cannabinoid false-negative results compromises the DOD urine drug-testing program s primary objective of drug deterrence. Acknowledgments The clinical research studies were supported by NIDA Intramural Research Program Funds. We would like to thank Shiralen Kawasaki, Paulanne Page, and Alberta Okamoto for their dedicated commitment to this project. References 1. D.A. Armbruster, R.H. Schwarzhoff, B.L. Pierce, and E.C. Hubster. Method comparison of Emit II and Online with RIA for drug screening. J. Forensic Sci. 38: (1993). 2. Roche Diagnostic Systems, Branchberg, NJ. Abuscreen Online pamphlet, B.D. Paul, L.D. Mell, J.M. Mitchell, R.M. Mckinley, and J. Irving. Detection and quantitation of urinary 11-delta-9-tetrahydrocannabinol-9-carboxylic acid, a metabolite of tetrahydrocannabinol, by capillary gas chromatography and electron impact mass fragmentography. J. Anal. Toxicol. 11: 1 5 (1987). 4. L.J. McBurney, B.A. Bobble, and L.A. Sepp. GC/MS and Emit analyses for Tetrahydrocannabinol metabolites in plasma and urine of human subjects. J. Anal. Toxicol. 10: (1986). 5. M.A. Huestis, J.M. Mitchell, and E.J. Cone. Lowering the federally mandated cannabinoid immunoassay cutoff increases true-positive results. Clin. Chem. 40: (1994). 6. P. Kintz, D. Machart, C. Jamey, and P. Mangin. Comparison between GC MS and the Emit II, Abbott ADx, and Roche Online immunoassays for the determination of THCCOOH. J. Anal. Toxicol. 19: (1995). 7. R.H. Liu, C. Edwards, L.D. Baugh, J. Weng, M. Fyfe, and A.S. Walia. Selection of an appropriate initial test cutoff concentration for workplace drug urinalysis cannabis example. J. Anal. Toxicol. 18: (1994). 8. E. Johansson, H. Gillespie, and M. Halldin. Human urinary excretion profile after smoking and oral administration of THC. J. Anal. Toxicol. 14: (1990). 9. M.A. Huestis, J.M. Mitchell, and E.J. Cone. Detection times of marijuana metabolites in urine by immunoassay and GC MS. J. Anal. Toxicol. 19: (1995). 10. M.A. Huestis, J.M. Mitchell, and E.J. Cone. Urinary excretion profiles of 11-nor-9-carboxy- 9 -tetrahydrocannabinol in humans after single smoked doses of marijuana. J. Anal. Toxicol. 20: (1996). 11. M.M. Halldin. Studies on the biotransformation of tetrahydrocannabinol in man and animals. Acta Pharm. Suec. 20: 160 (1983). 12. M.M. Halldin, S. Carlsson, S.L. Kantner, M. Widman, and S. Agurell. Urinary metabolites of 1 -tetrahydrocannabinol in man. Arzneim-Forsch./Drug Res. 32: (1982). 13. M.M. Halldin, L.K.R. Andersson, M. Widman, and L.E. Hollister. Further urinary metabolites of 1 -tetrahydrocannbinol in man. Arzneim-Forsch./Drug Res. 32: (1982). 14. R.H. Liu and D.E. Gadzala. Correlation of immunoassay and GC MS data derived from urine testing for monitoring marijuana exposure. In Handbook of Drug Analysis Applications in Forensic and Clinical Laboratories. American Chemical Society, Washington, D.C., M.E. Wall, D.R. Brine, and M. Perez-Reyes. The metabolism of cannabinoids in man. In Pharmacology of Marijuana, M.C. Braude and S. Szara, Eds. Raven Press, New York, NY, 1976, pp