Qualitative Screening for Drugs of Abuse in Hair Using GC-MS

Qualitative Screening for Drugs of Abuse in Hair Using GC-MS Susan Paterson*, Neil McLachlan-Troup, Rosa Cordero, Martin Dohnal, and Suzannah Carman ~ Toxicology Unit, Imperial College School of Medicine, St. Dunstan's Road, London, United Kingdom, W6 8RP I Abstract [ A previously described method for the analysis of hair has been modified to include analysis for amphetamines, benzodiazepines, cocaine and its metabolites, methadone and its metabolite, and phencyclidine in addition to opiates on a sample of hair. The samples of hair were washed twice with dichloromethane and cut into l-ram segments prior to extraction with methanol at 45~ for 18 h. The extracts were split into two parts; both were evaporated to dryness. One half of the extract was derivatized using MBTFA for analysis of amphetamines, and the other half was derivatized using MTBSTFA for analysis of the remaining drugs. The extracts were analyzed using electron impact gas chromatography-mass spectrometry operating in selected ion monitoring mode. In total, 18 drugs of abuse/metabolites could be detected. The method was used to screen 20 hair samples from patients attending a methadone-maintenance clinic. Introduction During recent months the Toxicology Unit has been approached by clinicians from various specialties suggesting collaborative research projects that would involve the screening for all classes of drugs of abuse on a sample of hair. Among the suggested projects was collaboration with the Department of Psychiatry at Imperial College of Science, Technology, and Medicine on a number of studies, including an investigation of the extent, nature, and clinical impact of substance use in schizophrenics. For example, illicit drug use is one of the foremost environmental factors implicated in psychotic breakdown (1), but research has been blighted by the unreliable nature of self-reports and the difficulties in obtaining urine samples covering the period immediately prior to hospital admission. Because of its retrospective time window, analysis of hair could solve these problems. A further example is covert substance misuse among outpatients receiving methadone-maintenance therapy at a central London drug-dependency unit. As drug- * Author to whom all correspondence should be addressed. E-maik s.paterson@ic.ac.uk. "~ Priory Research Fellow, Imperial College School of Medicine. treatment programs are open to abuse, it is usual practice to screen urine samples to check patient compliance. However, screening urine is of limited value because drugs, with the exception of cannabis, are only detected in urine for 1-3 days. In order to be certain of a patient's compliance, frequent urinary drug testing is necessary, but this is prohibitively expensive. Analyzing hair, which gives the drug history for weeks or months rather than days, may help solve these problems. Other suggested studies include looking at substance misuse in patients on remand. All the projects involved investigating the value of analysis of hair and comparing it with the more conventional approach of analysis of urine. In order to undertake these studies it was necessary first to develop methods for the analysis of urine and hair which detect all major classes of drugs of abuse simultaneously. A method has been developed for the analysis of urine for all major classes of drugs of abuse using mixed-mode solidphase extraction followed by gas chromatography-mass spectrometry (GC-MS) (2). A GC-MS method that could carry out similar analysis on a sample of hair was also developed and is described here. Moeller et al. (3), Goldberger et al. (4), and Kauert and Rohrich (5) have described methods for broad spectrum screening for drugs of abuse in hair, although neither Goldberger et al. (4) nor Kauert and Rohrich (5) included amphetamines, and none of the methods included benzodiazepines. Because benzodiazepine abuse is so widespread, particularly among patients receiving methadone maintenance, this class of drug has been included. The method presented here, which is a modification of the work of Kintz et al. (6), uses a simple methanolic extraction. The method follows the same wash and extraction procedures. The extract is then split into two parts: one half is derivatized using MBTFA and used for the analysis of amphetamines, the other half is derivatized using MTBSTFA and used for the remaining drugs. A GC-MS in selected ion monitoring (SIM) mode is used to detect and identify the drugs. Using Hewlett-Packard ChemStation software, at least 18 drugs/metabolites were looked for in each hair sample. The results of the analysis of hair samples from 20 patients attending a methadone-maintenance clinic are reported and compared with the prescribed drug therapy. Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 203

Experimental Materials Dichloromethane and methanol, both analytical reagent grade, were obtained from Merck (Poole, U.K.). All drug standards, deuterated phencylidine (PCP-ds), and MBTFA were obtained from Sigma-Aldrich (Poole, U.K.). Deuterated methylenedioxyamphetamine (MDA-ds) and deuterated oxazepam (oxazepam-ds) were both obtained from Promochem (Welwyn Garden City, U.K.). MTBSTFA was obtained from Pierce & Warriner (Chester, U.K.). Drug-free hair was obtained from laboratory personnel. Preparation of internal standards MDA-ds, oxazepam-ds, and PCP-d5 were used as the internal standards. Standard solutions were diluted to give a single solution containing all three internal standards at a concentration of 0.1 mg/ml in methanol. MDA-d5 was used as the internal standard for amphetamines because, like all amphetamines, it is volatile and it forms a TFA derivative. PCP-d5 was used as the internal standard for all the other compounds. Oxazepam-ds was used to check the TBDMS derivatization and to check the retention times of the late eluting peaks Extraction/derivatization procedure A sample of hair, typically 3 cm in length, was washed by soaking in 5 ml of dichloromethane for 10 min. The sample was dried, then soaked in another 5 ml of dichloromethane for 10 min. The sample of hair was then cut while still wet over a piece of aluminum foil into l-ram segments and allowed to dry. When dry, the segments of hair were transferred into a glass screw-cap tube, and the sample weight was recorded. Two milliliters of methanol and 10 ]JL of internal standard solution were added; the tube was sealed with a PTFE-lined screw cap and was left in a heating block at 45~ for 18 h. The solvent was then transferred, dividing it equally between two microvials. Both were left to evaporate to dryness at room temperature. To one vial, 25 pl of MTBSTFA was added, and 25 ~L of MBTFA was added to the other. The microvials were then capped, vortex mixed, and placed in a heating block at 90~ for 1 h. After cooling, 1 pl from each vial was injected into the GC-MS. Instrumental analysis A Hewlett-Packard (HP) model 5973 mass selective detector with a 6890 plus series GC fitted with a split/splitless injector port, an HP 7683 series automatic sampler, and the HP Enhanced Productivity MS Chemstation software were used. The analytical column was an HP-5 MS (crosslinked 5% phenyl methyl siloxane, 30 m x 0.25-ram i.d., 0.25-1Jm film thickness) fitted with an HP retention gap (uncoated, deactivated) (1 m x 0.25-ram i.d., 0-~m film thickness). Temperature conditions were as follows: (1) for the extract derivatized with MTBSTFA, the initial temperature was 60~ for 2 min, increased to 290~ at 6~ and held for 5 rain and (2) for the extract derivatized with MBTFA, the initial temperature was 60~ for 2 min, increased to 290~ at 16~ and held for 5 rain. The flow of the carrier gas (helium) was maintained at 1.0 ml/min in constant flow mode. The MS was operated in SIM mode, 220eV above the relative autotune value; an autotune was routinely performed before each run. The ion source and quadrupole temperatures were maintained at 230~ and 150~ respectively. The injector port was set at 280~ The GC-MS was programmed to perform a 1-pL splitless injection. Data acquisition was performed using an HP Vectra Kayak XA computer running HP 1701BA Enhanced ChemStation software (version B.01.00). A macro that performed target compound analysis on each sample for the 18 drugs/metabolites was written in-house. Method validation Recovery was experimentally determined by preparing two sets of samples of 1 ng/mg hair. One set was prepared by adding 50 pl ofa i-ng/pl solution in methanol of the 18 drug/metabolites shown in Table I to 50 mg of washed and dried drug-free hair. Two milliliters of methanol was added, and the spiked hair solutions were placed in tubes, sealed with a PTFE-lined screw cap, and left in a heating block at 45~ for 18 h. An external standard, 10 pl of a 0.1-mg/mL solution of PCP-ds, MDA-ds, and oxazepam-d5 in methanol, was then added. The solutions were then split into two parts, left to evaporate to dryness, derivatized, and run on the GC-MS as normal. The second set was prepared in exactly the same manner, but the heating stage was omitted. The ratio of drug to MDA-d 5 in the heattreated samples was compared to the ratio of drug to MDA-ds in the non-heat-treated samples. By comparing these ratios, the Table I. Validation Data Limit of %Recovery RSD of detedion ng/mg hair recoveries Drug (ng/mg) tinearity (n = 6) % Amphetamine-TFA* 0.02 0.9933 96.2 11.6 Methamphetamine-TFA 0.03 0.9942 97.9 9.5 MDA-TFA 0.06 0.9979 102.9 3.9 MDMA-TFA 0.07 0.9990 97.3 2.4 Cocaine 0.12 0.9985 79.6 7.1 Ecgonine methylester-tbdms 0.07 0.9976 87.5 11.9 Benzoylecgonine- TBDMS 0.09 0.9991 76.8 5.2 Cocaethylene 0.16 0.9977 81.8 6.9 Methadone 0.15 0.9958 83.9 9.0 EDDP 0.27 0.9988 93.8 23.9 Dihydrocodeine 0.25 0.9969 73.8 9.7 Morphine-TBDMS 0.40 0.9904 73.6 15.3 6-Monoacetylmorphine-TBDMS 0.06 0.9984 80.2 9.9 Diazepam 0.11 0.9979 79.2 11.6 Desmethyldiazepam-TBDMS 0.21 0.9859 80.2 7.4 Oxazepam-TBDMS 0.11 0.9969 75.9 4.7 Temazepam-TBDMS 0.30 0.9909 77.5 5.19 Phencyclidine 0.17 0.9981 81.3 6.4 * Abbreviations: TFA, trifluoroacetamide derivative; TBDMS, tert-butyldimethylsilyl derivative; 2TBDMS, 2tert-butyldimethylsilyl derivative; MDA, melhylenedioxyamphetamine; MDMA, methylenedioxyrnethamphetamine; and EDDP, 2.-ethylidine 1,5-climethyl-3,3~iphenylpyrrolidone (methadone metabolite). 204

percent recovery was calculated for the amphetamine, methamphetamine, MDA, and MDMA. The ratio of drug to PCP-ds in the heat-treated samples was compared to the ratio of drug to PCPd5 in the non-heat-treated samples. By comparing these ratios, the percent recovery was calculated for the remaining drugs/rnetabolites. The precision of the assay is shown by the relative standard deviations (RSD) of the recoveries. In order to check recovery across a concentration range, standard curves were prepared. A standard solution of amphetamine, metharnphetarnine, MDA, MDMA, cocaine, co- caethylene, benzoylecgonine, ecgoninemethylester, methadone, EDDP, morphine, 6-rnonoacetylrnorphine, dihydrocodeine, diazeparn, desrnethyldiazeparn, oxazeparn, ternazepam, and phencyclidine at 20 ng/pl was prepared in methanol. Standard curves were prepared by adding 0, 1, 2.5, 10, 15, and 25 pl of this solution to 50 rng of blank hair to give drug concentrations of 0, 0.4, 1, 4, 6, and 10 ng/rng hair. A reagent blank was also prepared. Both inter- and intra-assay reproducibility were monitored by running the same 6 ng/mg standard every 10th injection 5500000! soooooo! t 4SO0000 i 40000OO t ei :o~ TIC: PAT3BC-S25.D 12 t 500ooo 10 J IO00000,! sooooo i 1 il i 6 J 1 111 14 ~5! 18 :, i*! 4 i*,~ ~ u ~li i '3f I] 16:~ i' 19.00 20.00 21.100 22.00 23,00 24.00 25.00 26.00 27.00 28: Time (min) Figure 1. Total ion chromatogram of a l O-ng/mg hair standard. Peak identification: 1, ecgonine methyl ester-tbdms; 2, phencyclidine-ds; 3, phencyclidine; 4, EDDP; 5, methadone; 6, cocaine; 7, cocaethylene; 8, dihydrocodeine; 9, diazepam; I0, desmethyldiazepam-tbdms; 11, benzoylecgonine-tbdms; 12, endogenous compound; 13, dihydrocodeine-tbdms; 14, morphine-tbdms; 15, oxazepam-tbdms; 16, temazepam-tbdms; 17, 6-monoacetylmorphine- TBDMS; and 18, rnorphine-2tbdms. 10c~C oo TIC: PAT3B-BLK.D 40oo00of i oooo 1 250OOOO] 2ooooooj 15000001, t 10000001 19.00 20.00 2"~,00.~.00"-'2"3,~)~24.~"25!00 ~,00 27100" ~',5,~'0()" 29~]0 30~()0 "31!0()' 32~00" ~00 34.00"35,~ 36.00 37~00 38.00~'-"~1.00"~42!(]'0 ~ ~,3:0~0~'44.'00 ~ Figure 2. Total ion chromatogram of a blank hair standard. Time (rain) ii" t 205

throughout each run. Analytes were identified on the basis that retention time and relative abundance of each target ion in the sample matched with the standard. The limit of detection was determined by estimating the minimum concentration equivalent to, or greater than, three times the background noise while still allowing detection of all target ions. Both washes from the analysis of the first 10 samples of hair were analyzed and were negative for all 18 drugs/metabolites. Hair samples Hair was collected from 20 patients attending a methadone maintenance clinic. The hair was cut as close to the scalp as possible, and the root end was carefully labeled. Measuring from the root end, 3-cm lengths were cut, representing approximately 3 months growth, and the exact weight was noted. The drug therapy the patient was receiving was also noted. Table II. Retention Times and Mass Spectral Data Drug Ions (m/z) RT ion for other ions for identification (min) quantitation 1 2 3 4 Amphetamine-TFA* 8.54 118 140' 91 65 Methamphetarnine- TFA 9.42 154 t 110 118 91 65 MDA-ds-TFA 11.10 136' 167 280 MDA-TFA 11.13 162 1354" 140 MDMA-TFA 11.95 162 154' 135 110 Ecgonine methylester-tbdms 23.82 313 298 256 182' Phencyclidine 25.75 243 242 200* Phencyclidine-d5 25.71 246 248 205* EDDP 27.52 277* 276 262 220 Methadone 29.86 294 223 165 72* Cocaine 30.19 303 272 198 182' Cocaethylene 31.62 317 272 196 82* Dihydrocodeine 32.78 301* 244 284 Diazepam 34.23 256* 283 284 221 Desmethyldiazepam-TBDMS 34.77 384 327* 329 383 Benzoylecgonine- TBDMS 35.01 403 346 282* Dihydrocodeine- TBDMS 36.61 358 415 282 315' Morphine-TBDMS 37.27 399 342* 229 285 Oxazepam-TBDMS 37.91 514 459 457* Oxazepam-ds-TBDMS 38.25 519 464 462* Temazepam-TBDMS 38.71 359 357* 385 6-Monoacetylmorphine-TBDMS 38.79 342* 384 382 441 Morphine-2TBDMS 40.59 513 413' 456 335 * Abbreviations: TFA, trifluoroacetamide derivative; TBDMS, tert-butyldimethylsilyl derivative; 2TBDMS, 2tert-butyldimethylsilyl derivative; MDA, methy]enedioxyamphetamine; MDMA, methy]enedioxymethamphetamine; and EDDP, 2-ethylidine 1,5-dimethyl-3,3-diphenylpyrrolidone (methadone metabolite). t Most abundant ion Results and Discussion The MS was run in SIM mode, and Figures 1-3 show typical chromatograms for a standard, a blank, and a patient sample. Table II shows the retention times and mass spectral data used to perform target compound analysis and produce extracted ion chromatograms (EIC). A printout was issued for each sample showing a chromatogram of the selected mass-to-charge ratio ions for each compound. If the EIC indicated the presence of a particular compound by showing that the ions were present and in approximately the correct ratio and the retention time matched, the data were checked manually with the spectra from standards of the same run. Although standard curves were prepared to validate the method, the method was designed for use as a simple screening procedure. When used as such, only a high (10 ng/mg of hair) and a low (0.4 ng/mg of hair) standard were run. Even for qualitative work, the three deuterated internal standards were added to the samples to check that the extraction and derivatization procedures had been completed satisfactorily. The eluent was selectively derivatized with MBTFA, which formed trifluoroacetamide derivatives of primary and secondary amines. To form the silyl derivatives, MTBSTFA was used in place of the more usual MSTFA + 1% TMCS because it is an exceptionally strong silylating reagent. Derivatives are reported to be up to 10,000 more stable to hydrolysis than their corresponding TMS derivatives, yet they produce easy-to-interpret mass spectra (7). Some drugs gave more than one derivatization product. These included dihydrocodeine, which had both an underivatized and a TBDMS peak, and morphine, which had both a TBDMS and a 2TBDMS peak. For each of these drugs, the major peak was used to estimate the limit of detection and calculate percent recovery. These were the underivatised peak for dihydrocodeine and the TBDMS peak for morphine. As no clean-up stage was used, the extracts could be relatively dirty, and in order to cope with this, maintenance of the mass spectrometer had to be vigorous. A retention gap was used and it was changed every 200 injections. The column was changed every 500 injections, and the ion source was cleaned after every 100 injections. Table I shows the validation data including linearity covering the concentration range 0-10 ng/mg hair and percent recovery for hair spiked at 1.0 ng/mg hair for each of the 18 drug/metabolites of interest. The results show good linearity for all drugs for the range 0.4 to 10 ng/mg of hair. The method has the potential to be used quantitatively. Recovery is acceptable for all drugs. Table III shows the results of the analysis of hair samples taken from 20 patients attending a methadone-maintenance clinic. The drugs the patients had been receiving for the three months prior to the samples being taken are also shown. All patients were receiving maintenance doses of methadone of between 26 and 200 mg/day. Methadone was detected in all hair samples, but its metabolite, EDDP, was detected in only two samples. Of the five patients prescribed benzodiazepines in addition to methadone, only three showed positive for these drugs 206

in hair. The possible reasons for this are either the benzodiazepines are present in the hair but in concentrations below the limit of detection, they are not readily extracted out of hair by this method, or the patients are not complying with treatment. Simultaneous analysis of the urine would determine whether the patients were taking their prescribed therapy and would confirm the reliability of the hair results. Although the method was validated using 50-mg samples of 4 TIC: PAT5B-209.D 55OOOOO 50000001 45ooooo1 4oooooo I 3000000 I 15Oo000! i 1000000} sooooo'~ ~r ": 6 8 i :i i il~,. ^,'~,,,Ji.,..:I.!, Ti J~ 0 26.0o 27.00 28.00 2~.00 30.00 31.00 32.00 "~3.0o 34.00 ~5.00 ~.00 37.00 3s.oo 39.0040.00,uoo 42.Q04a.oo 4400 Time (mln) Figure 3. Total ion chromatogram of a patient's hair sample. Peak identification: I, ecgonine methyl ester-tbdms; 2, phencyclidine-ds; 3, EDDP; 4, methadone; 5, cocaine; 6, cocaethylene; 7, desmethyldiazepam-tbdms; and 8, benzoylecgonine-tbdms. Table III. Drugs Detected in Hair Samples from 20 Patients Attending a Methadone-Maintenance Clinic* Patient Daily medication (Hair wt, Meth t Diaz Tmz Drugs found in hair mg) (mg) (mg) (mg) Meth EDDP Cocaine BE Desdiaz Temaz Amphet DHC Morph 6-MAM 1 (12.5) 95 10 2 (99.1) 70 10 3 (22.8) 165 4 (26.9) 100 5 (9.7) 200 20 6 (13.7) 45 7 (21.6) 60 8 (23.5) 70 9 (33.5) 120 10 10 (19.8) 30 11 (53.0) 100 12 (25.2) 55 13 (39.2) 80 40 40 14 (20.6) 26 15 (25.6) 190 16 (20.9) 140 17 (24.7) 120 18 (24.1) 85 19 (18.2) 150 20 (21.3) 60 * All samples were negative for methamphetamine, I-MDMA, MDA, ecgonine methyl ester, phencyclidine, cocaethylene, diazepam, and oxazepam. t Abbreviations: Meth, methadone; Tmz, temazepam; EDDP, 2-ethylidine 1,5-dimethyl-3,3-diphenylpyrrolidone (methadone metabolite); BE, benzoylecgonine; Desdiaz, desmethyldiazepam; Amphet, amphetamine; DHC, dihydrocodeine; Morph, morphine; and 6-MAM, 6-monoacetylmorphine. 207

hair spiked with drugs, the weight of hair the clinicians submitted for analysis varied from 9.7 to 99.1 rag. For 13 out of the 20 patient samples, the amount of hair analyzed was less than 25 rag. The results showed that the method was robust and sensitive enough to detect drug/metabolites even in these small samples. All classes of drugs were detected in the hair samples, including methadone and its metabolite EDDP, amphetamine, desmethyldiazepam, temazepam, cocaine and its metabolite benzoylecgonine, dihydrocodeine, morphine, and 6- monoacetylmorphine. This shows that the method is suitable for broad-spectrum screening of hair for drugs of abuse using a sample of hair. A larger, more comprehensive study is planned to evaluate just how useful screening of hair is for drug-treatment centers. Such a screen would be particularly useful for patients when they first attend a drug-treatment clinic for verifying their drug abuse and for providing the clinician with a history of their drug exposure during the preceding weeks. As hair analysis has a greater retrospective time window than urine, less frequent analysis is necessary. For patients, giving hair rather than urine is a more dignified procedure, and hair is less susceptible to tampering. If hair could replace or complement urine analysis for some patients, it could be a saving on resources and may help the patient/clinician relationship. References I. J. Smith and S. Hucker. Schizophrenia and substance abuse. Br. J. Psychiatry 164:13-21 (1994). 2. S. Paterson, R. Cordero, S. McCulloch, and P. Houldsworth. Analysis of urine for drugs of abuse using mixed-mode solid-phase extraction and gas chromatography/mass spectrometry. Ann. Clin. Biochem. 37:690-700 (2000) 3. M.R. Moeller, P. Fey, and R. Wennig. Simultaneous determination of drugs of abuse (opiates, cocaine and amphetamine) in human hair by GC/MS and its application to a methadone treatment program. Forensic $ci. Int. 63:185-206 (I 993). 4. B.A. Goldberger, A.G. Darraj, Y.H. Caplan, and E.J. Cone. Detection of methadone, methadone metabolites, and other illicit drugs of abuse in hair of methadone-treatment subjects. J. Anal. Toxicol. 22:526-530 (1998). 5. G. Kauert and J. Rohrich. Concentrations of delta 9-tetrahydrocannabinol, cocaine and 6-monoacetylmorphine in hair of drug abusers. Int. J. Legal Med. 108(6): 294-299 (1996). 6. R Kintz, R Bundeli, R. Brenneisen, and B. Ludes. Dose-concentration relationships in hair from subjects in a controlled heroinmaintenance program. J. Anal Toxicol. 22:231-236 (1998). 7. Derivatisation Reagents. In Regis Technologies, Inc. Chromatography Guide. From Phase Sep U.K., pp 37-49. Manuscript received May 19, 2000; revision received October 30, 2000. 208