Quantitative Screening for Benzodiazepines in Blood by Dual-Column Gas Chromatography and Comparison of the Results with Urine Immunoassay*

Quantitative Screening for Benzodiazepines in Blood by Dual-Column Gas Chromatography and Comparison of the Results with Urine Immunoassay* Ilpo Rasanen, Ilkka Ojanpedi, and Erkki Vuori Department of Forensic Medicine, P.O. Box 40, F/N-O0014 Universi~/ of Helsinki, Finland [ Abstract [ A dual-column retention index method is described for quantitative gas chromatographic (GC) screening of 26 benzodiazepine drugs and metabolites in the blood using DB-5 and DB-17 capillary columns and electron capture detection. The method involves a one-step, smau-scale liquid-liquid extraction with ethyl acetate and derivatization with N-methyI-N-(tert-bulyldimethylsilyl)trifluoroacetamide with 1% tert-butyldimethylsilyl chloride. The results from the GC screening of 514 postmortem blood samples were compared to those obtained from urine immunoassay (Syva ETSplus with a 200-ng/mL cutoff). Both methods gave a negative result in 284 cases and a positive result in 149 cases. In 48 cases, urine was negative by immunoassay but blood was positive by GC. The opposite situation (blood negative, urine positive) was detected only in four cases. In 29 cases, an invalid result was obtained by urine by immunoassay: 26 blood samples of those cases were negative and three samples positive by GC. In postmortem forensic toxicology, the present GC method seems to be a good alternative to the common combination of urinary immunoassay followed by quantitative analysis of blood by chromatography. Introduction In urinary drug testing, the presence of benzodiazepines is usually revealed by screening urine samples by immunochemical techniques and confirming positive findings by chromatographic methods. In postmortem toxicology and in drugs and driving investigations, it is also necessary to perform a quantitative benzodiazepine determination of the blood in order to estimate the degree of toxicity. Although several methods, especially gas chromatography-mass spectrometry (GC-MS), have been published to detect benzodiazepines in urine (I-15), there are fewer methods for a broad range quantitative screening analysis in the whole blood. These methods ' Presented in pan at the SOFT-TIAFT meeting, AJl~Jquerque, New Mexico, October 1998. mainly involve gas chromatography (GC) (16-20) or liquid chromatography (HPLC) (21-24). By using GC methods, benzodiazepines are usually analyzed separately from other drugs because of their distinctive chromatographic properties. Benzodiazepines are fairly polar compounds with low volatility and low blood concentrations, and an electron withdrawing group makes them especially amenable to electron capture (EC) detection. Because of their polarity, many benzodiazepines and particularly their metabolites are difficult to analyze at low levels without derivatization of the active hydroxyl or amide groups. The present paper describes a quantitative dual-column GC retention index (RI) method for screening benzodiazepines in the blood and compares the performance of the GC blood analysis with a urine immunoassay in 514 successive postmortem cases. Materials and Methods Reagents N-MethyI-N-(tert-butyldimethylsilyl)trifluoroacetamide (MTBSTFA) with 1% tert-butyldimethylsilyl chloride fib- DMSCI) was from Aldrich (Steinheim, Germany). AquaSil T~ siliconizing fluid was from Pierce (Rockford, IL). [~-Glu" curonidase from E. coli K 12 was from Boehringer MannheiW (Mannheim, Germany). Alprazolam, 1-hydroxyalprazolam, tri" azolam, and 1-hydroxytriazolam were from Upjohn (Kala" mazoo, MI). Bromazepam, desmethyldiazepam, lorazepam, Iormetazepam, and prazepam were from Sigma (St. LouiS, MO). Brotizolam was from Boehringer Ingelheim (Ingelhei~ am Rhein, Germany). Clobazam and norclobazam were fro~ Hoechst AG (Frankfurt am Main, Germany). Clonazepam, 7- I aminoclonazepam, chlordiazepoxide, demoxepam, fluni" trazepam, desmethylflunitrazepam, flurazepam, midazolaro, and 1-hydroxymidazolam were from Hoffman-La Roche (Basel, Switzerland). Diazepam, nitrazepam, oxazepam, and temazepam were from Orion (Espoo, Finland). Phenazepa~ 46 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.

was a donation from the Republican Centre of Forensic Medicine (Moscow, Russia). The R-series RI standards (Figure 1) were synthesized in the authors' laboratory (25). The compounds R2, R6, R8, R12, R14, and R16 were used as qualitative standards for calculating the retention index values of unknown peaks. R2, structurally closest to the benzodiazepine drugs, was also used as a quantitative internal standard. F Figure I. The structures of the R-series retention index standards. R = alkyl (R2: R = ethyl, R6: R = hexyl, elc.) R. The stock standard solutions were prepared by weighing each of the pure substances into a 4-rnL vial and adding methanol (acetonitrile for norclobazarn) in order to obtain a concentration of 1 mg/ml. The stock solutions were further diluted with methanol (norclobazam solution with acetonitrile) to obtain working solutions of 100, 10, and 1 IJg/mL. Sample preparation for GC and GC-MS The centrifuge tubes were silanized before use by dipping them into a 0.2% solution of Aquasil in water, rinsing with water, and allowing to dry at room temperature for 24 h. The working blood standards were prepared by adding the standard solutions into centrifuge tubes (three or four compounds per one tube) and evaporating the solvent to dryness under a gentle stream of nitrogen. Whole blood (bovine blood for working blood standards) (1 g) was transferred into a centrifuge tube (10-ram i.d.). The internal standard 1,3-dihydro-l-ethyi-7-fluoro-5-(4-fluorophenyl)-2H-1,4-benzodiazepin-2-one (R2) (0.3 Ilg) in methanol (50 I~L) and saturated ammonium chloride solution (0.5 ml) were added to obtain a ph of 7.4, and the mixture was shaken. The sample was extracted with ethyl acetate (0.5 ml) in a vortex mixer for 5 rain and centrifuged. An aliquot of Table I. Retention Indices, Linearity, Limit of Detection (LOD), and Limit of Quantitation (LOQ) RI' RI' Range r 2 r 2 LOD LOQ ~ Working blood Substance DB-5 DB-17 (ng/g) DB-5 DB-17 (ng/g) (ng/g) standards (ng/g*) Diazepam 442 642 20-2000 0.9985 0.9978 10 20 200, 800, 1500 Desmethyldiazepam.TBDMS 519 453 20-2000 0.9988 0.9998 8 20 200, 1000, 2000 Clobazam 613 836 80-2000 0.9959 0.9987 50 75 100, 500, 1000 Midazolam 655 841 50-2000 0.9985 0.9999 30 50 100, 500, 1000 Norclobazam-TBDMS 667 685 30-2000 0.9930 0.9990 20 30 200, 1000, 2000 Flunitrazepam 669 879 5-300 0.9991 0.9981 3 6 10, 20, 50 Prazepam 705 860 20-I 000 0.9963 0.9961 I0 15 I0, 50, I00 Brornazepam.TBDMS 719 800 20-1000 0.9912 0.9987 15 30 100, 500, 1000 Chlordiazepoxide1.TBDMS 766 793 4w 4~ 50 500, 2000, 5000 Chlordiazepoxide2.TBDMS 865 800 50 Phenazepam-TBDMS 793 817 10-500 0.9977 0.9990 5 I0 50, 200, 500 Desmethylflunitrazepam.TBDMS 800 811 10-500 0.9977 0.9934 5 10 10, 20, 50 Nitrazepam-TBDMS 819 836 20-1000 0.9913 0.9997 10 20 50, 300, 500 Oxazepam-TBDMS 822 691 10-2000 0.9981 0.9945 5 10 200, 1000, 2000 Demoxepam-TBDMS 837 927 50-2000' 0.9901 0.9983 10 50 200, 1000, 2000 Flurazepam 838 939 20-I 000 0.9978 0.9988 I0 20 20, 50, 200 Temazepam-TBDMS 871 964 10-2000 0.9993 0.9994 5 I0 200, 800, 1500 7"Aminoclonazepam.TBDMS 880 1063 30-300 4w 0.9950 20 60 20, 50, 200 Clonazepam.TBDMS 937 993 30-500 0.9968 0.9979 10 20 20, 50, 200 l-ormetazepam.tbdms 940 1053 10-500 0.9972 0.9979 5 8 10, 20, 50 lorazepam-tbdms 946 862 20-2000 0.9972 0.9980 5 8 100, 500, 1000 Alprazolam 1003 1340 30-500 0.9907 0.9997 20 20 20, I00, 200 I "Hydroxymidazolam-TBDMS 1044 1131 15-2000 0.9980 0.9999 8 15 50, 100, 300 Triazolam 1109 1445 10-I000 0.9968 0.9990 3 I0 I0, 50, I00 Brolizolam 1120 1476 10-500 0.9952 0.9985 5 I0 I0, 50, I00 1-Hydroxyalprazolam_TBDMS 1326 1509 10-1000 0.9967 0.9996 5 10 50, 100, 200 1-Hydroxytriazolam.TBDMS 1435 1634 20-500 0.9946 0.9954 I0 15 I0, 30, 50 El values are delermined T with a new,',akr hof columns and may slightly e differ from those in Table III and Figure 3. V~lues are based on the EURACHEM approach and are measured wilh a different set of samples thus slightly differing from lhe lower limit of linearity. ~r~ ~ u. are based on ~'-.e~r... ty, u..:~dl~: L... u,i~,,. i~,,.j,:.~ n~ u,,,,~r,k~,4 u- g, and r ea blood concentrallons " in forenstc ' ' cases...~ aljlative result only. 47

the organic phase (100 ~tl) was transferred to an autosampler vial (07-CPV(A), Chromacol, Trumbull, CT) and evaporated to dryness in order to eliminate the traces of water in ethyl acetate that disturb the silylation procedure. The residue was reconstituted with 100 IJL of ethyl acetate, and 50 IlL of MTBSTFA with 196 TBDMSCI was added. The vial was capped with a crimp cap (8-AC5, Chromacol) and heated at 85~ for 30 rain. Urine samples were hydrolyzed by adding 10 pl of [3-glucuronidase to 1 ml of urine and heating for 40 min at 40~ The rest of the sample preparation procedure was identical to the procedure for blood samples. GC The GC was a Micromat HRGC 412 (HNU-Nordion, Helsinki, Finland) with two EC detectors. The fused silica capillary columns were DB-5 (15 m x 0.32-mm i.d., 0.1-~tm film thickness) and DB-17 (15 m x 0.32-ram i.d., 0.15-1am film thickness, J&W Scientific, Folsom, CA). The columns entered a single injector through a two-hole ferrule. The Grob-type split/splitless injector was operated in the splitless mode. Silanized glass wool was used inside the liner. Automated injections were performed with a CTC A200S autosampler (CTC Analytics, Zwingen, Switzerland) using a 1.5-pL apparent injection volume. The autosampler was set to take the RI standards in ethyl acetate from a separate vial prior to the sample injection. The carrier gas was helium with a flow rate of approximately 3.0 ml/min for DB-5 and 2.0 mumin for DB-17 (130~ The split flow rate was 15 ml/min, and the septum purge flow rate 3 ml/min. The detector makeup gas (5% CH4 in At) flow rate was 25 ml/min. The injector and detector temperatures were 250~ and 300~ respectively. The splitless time was 0.6 rain. The oven temperature was initially held at 130~ for 0.6 rain, increased by 25~ to 220~ and then increased by 5~ to 290~ which was held for 2 rain. GC-MS GC--MS was performed with a Hewlett-Packard (Wilmington, DE) 5972 mass selective detector coupled to a HP 5890 series I1 GC equipped with a HP-1 (12 m x 0.20-ram i.d., 0.33-pro film thickness) capillary column. The GC-MS was operated by Chemstation software and the GC was used in the splitless mode. The injector port temperature was 250~ and the transfer line temperature was 280~ The oven temperature was initially held at 130~ for 0.5 rain and then increased by 15~ to 300~ which was held for 2 min. Data processing in GC Data processing was performed with MS Windows-based SC- WorkStation 3.0 software (Sunicom, Helsinki, Finland). The program was set to automatically identify the internal RI standards by pattern recognition, to calculate the temperature programmed (linear polygonal) RI values of all detected peaks, and to compare them with library data (substance windows). A part of the SC-Workstation, the SC-Compare 1.11 software, first identifies all substances in the library that could match a peak, separately on each column. It then compares the data from the single columns and reports only those substances that were identified with both columns in an easily interpretable form (26). Urine immunoassay Urine samples were centrifuged and the supernatants were analyzed by Emit d.a.u. TM (ETSplus) benzodiazepine assay (Syva, San Jose, CA) using a 200-ng/mL cutoff value. Postmortem specimens For the blood-urine comparison study, 514 successive postmortem cases in which both peripheral blood and urine was sent for toxicological analysis by the medical examiners were used. Results The GC RI values of 26 benzodiazepine drugs and metabolites and the linearity of quantitation, limit of detection, and limit of quantitation in the blood are presented in Table I. The linearity was good on both columns, the r 2 exceeding 0.99, except for chlordiazepoxide, which decomposed producing several peaks. All the compounds could be detected at the therapeutic level and most of them at the low therapeutic level. A concentration in the blood that resulted in a signal three times the baseline noise was used as the criterion for the limit of detection. The EURACHEM approach with 10% precision was used for the calculation of the limit of quantitation. In the EURACHEM approach, samples of decreasing amounts of the analyte are injected six times. The calculated relative standard deviation is plotted against the analyte amount, and the amount that corresponds to the previously defined required precision is equal to the limit of quantitation (27). The extraction recovery was over 80% for each compound. The recoveries were determined by comparing values obtained with unextracted and extracted standards. An unextracted standard was used as a 100% recovery reference. Three blood samples containing the middle concentration of working blood Table II. Quantitative Precision of Selected Compounds Extracted from Blood Theoretical Concentration Mean* Substance (pg/g) (pg/~ cv% Desmethyldiazepam-TBDMS 0.20 0.19 5.5 1.00 0.91 2.4 Diazepam 0.20 0.23 6.4 1.00 1.00 8.2 Temazepam-TBDMS 0.20 0.24 10.3 1.00 1.02 13.6 Midazolam 0.10 0.13 11.2 0.50 0.50 9.0 * Values are mean values from two columns and are based on four determination: during a four-week period. 48

standards for each compound (Table I) were extracted normally except that the internal standard was added after extraction before derivatization. The results of the three experiments were then averaged. The long-term quantitative (concentrations) and qualitative (retention indices) precision of selected compounds was good on both columns (Tables II and III). Ch.l: DB-5 Table III. Precision of Retention Indices Concentration Substance (pg/g) Column Mean RI* CV%* Desmethyldiazepam-TBDMS 0.2 DB-5 520.2 0.09 DB-17 450.2 0.14 Diazepam 0.2 DB-5 442.6 0.17 D8-17 641.5 0.14 Temazepam-TBDMS 0.2 DB-5 872.2 0.04 DB-17 966.5 0.16 Midazolam 0.1 DB-5 653.7 0.14 DB-17 845.6 0.19 * Values are based on twelve runs during a lwelve-week period. Time (mln) Ch.2:DB-17 t; I I ~ Ch.l: DB-5 19 IT,,I 1 19 11 ii 9 ~ 19 Tlme (mln) i i " A 2, Ch2:DB-17 ' f,,i iii i ' i Time (mln) Figure 2. The separation of the drugs (except chlordiazepoxide) and RI standards on DB-5 and DB-17 columns. Peak identification: 1,112; 2, diazepam; 3, desmethyldiazepam-tbdms; 4, R6; 5, clobazam; 6, midaolam; 7, norclobazam-tbdms; 8, flunitrazepam; 9, prazepam; 10, romazepam-tbdms; II, phenazepam-tbdms; 12, 118; 13, de smethylflunitrazepam.tbdms; 14, nitrazepam-tbdms; 15, oxazepam- TBDMS; 16, demoxepam-tbdms; 17, flurazepam; 18, temazepam- ~ BDMS; 19, 7-aminoclonazepam-TBDMS; 20, clonazepam-tgdms; 21, ormetazepam.tbdms; 22, lorazepam-tbdms; 23, alprazolam; 24, 1-hydro 25, triazolam; 26, brotizolam; 27, R12; 28, 1- hydroxyalprazolam-tbdms; 29, R14; 30, 1-hydroxytriazolam-TBDMS; and 31,1116. M Time (mln) Figure 3. A dual-column chromatogram obtained from an autopsy blood sample. The R series RI standards were co-injected with the sample. *** 9C-C~I~'e 9~m~:t [Vs:sl~ 1.30 *** Data rile : g827tthd.dta Ibt~xl : blmso:11,~ Ds~e C--eared : h~ A~g 29 1998 st; 00:29:33 Date Ana11,sed : hi Ik~ 4 1998 st 12:02:39 r cqz. ~ AleUt zdpaurs Dtt! J4x~ Aaount ~/g 1. ml~ndszd 92/n1"Z.9 1 1' 4.407 200.00 0.@OO 1207495 0.500 2 1' 5*980 200.00 0.000 246929 0.000 duaet.bylr -*m 1 11 9.99T 519.12 1.749 1513T5 0.070 2 O T,999 447.37 0,59T 59419 0.095 2. e~'~=~j M 1 15' 5.973 900.00 0.000 149947 O.00O 2?.0 t 9.042 500.00 0.000 182429 0.000 (I~11 IlellMm 1 10 5.385 444.40 0,320 979570 0.101 2 11 1.370 439.28 0.917 73799 0.113 ox~epaa-'11da 1 19 7.503 820.99 1.509 29959 0.00T 2 12 9.757 995.74 1,410 99979 0.0221 0. ztanda.cd U 1 19' 7.202 900.00 0.00O 140900 0.000 2 14* 9.T19 900,09 0,000 99994 0.OO0... dasmm;~m*11:16 1 20 7,959 935.93 0.931 167961 0.051 2 15 11.055 925.72 1.490 149229 0.099 tasslm~a*t]lqm 1 22 T.99T 8"/0.92 1.643 94999 0.011 9 19 11.452 954.05 2.994 59714 O,OOI 4. a ~ 11t2 1 25* 11.155 1200.00 0.0O0 124959 0.0O0 2 17' 14.9TT 1200.00 0.000 8990 0.000 slp:ss01sa 1 23 9.310 1007.94 2.078 120514 0.149 2 22 15.598 1344.15 1.381 125093 0.192 0. e~"~'~'4 P.14 1 29" 19.460 1400.00 0.000 104708 0.000 2 25* 10.227 1400.00 0.000 T9299 0.0O0 1-h,~lpzuollm- 1 29 12.415 1326.30 0.413?ST2 0.009 9 24 17.487 1511.26 2.790 15441 0.009 5. et.aj~illrd 1110 1 30* 15.705 1900.00 O.0O0 135095 0.000 2 25* 18.492 1600.00 0.00O 110455 0.000 Figure 4. The SC-Compare report for the case in Fisure 3. 49

Figure 2 shows the separation of all compounds studied (except chlordiazepoxide) together with the co-injected R! standards on DB-5 and DB-17 columns. Figures 3 and 4 show the dual-column chromatogram from an actual autopsy blood sample and the SC-Compare analysis report for the case, respectively. Table IV. Compounds Tested for Interference Compound RI (DB-5) RI (DB-17) Amobarbital-TBDMS 196" ND ~ Brallobarbital-TBDMS 397' 193 Brompheniramine ND ND Butalbital-TBDMS 194 ND Carbamazepine-TBDMS 480 600 Carbinoxamine ND ND Chlometiazole ND ND Chlormezanone 200 497 Chloroquine-TBDMS 659 622 Chlorpheniramine ND ND Chlorpromazine 515 585 Chlorpropamide ND ND Chlorprothixene 522 590 Chlorzoxazone ND ND Citalopram 402, 600 723, 737 Clobutinol ND ND Clomipramine ND ND Clozapine 921 1154 Cyclobarbital-TBDMS 478' 229* Diltiazem 1021 1301 Felodipine 727 935 Fenfluramine ND ND Flecainide-TBDMS 735 627 Flumazenile 600 839 Fluoxetine-TBDMS 498 485 Fluvoxamine-TBDMS 507, 587 324, 341 Haloperidol 995 1120 Hexobarbital-TBDMS ND ND 10-Hydroxycarbamazepine-TBDMS 600 666 Hydroxychloroquine-TB DMS 1187 1117 Hydroxyzine-TBDMS 1269 1255 Ketamine ND ND Lamotrigine-TBDMS 681,910' 800, 882* Meclozine 091 1200 Melperone ND ND Metoklopramide 695 847 Moclobemide 198 244 Moperone 881 1005 Nifedipine 646 911 Oxcarbazepin-TBDMS 381,903 828, 881 Parathion ND ND Pentobarbital-TBDMS 221 * ND Phenobarbital-TBDMS 444' 245* Phenytoin-TBDMS 694 618 Secobarbital-TBDMS 269' ND 5ertraline-TBDMS 1074 1264 Tiopental-TBDMS 223, 545, 731 222, 240 Trazodone 1374 1760 Vinbarbital-TBDMS 236' ND Zopiclone 1158 1548 * Fronting peak, which may disturb the integration of peaks eluting before the compound. Not detected inside the elution range of benzodiazepines. A selection of compounds that could be detected by ECD was tested for interference (Table IV). Although some of the compounds might be confused with benzodiazepines in a single column, none of them interfered with benzodiazepines in the present dual-column procedure. The results from the comparison study between the GC blood analysis and urine immunoassay are summarized in Table V. The following drugs were detected in those cases where the urine was negative and the blood positive: diazepar~ (18 times: range 0.02-0.7 rag/l, desmethyldiazeparn (18: 0.02-0.7 mg/l), oxazepam (13:0.01-0.4 mg/l), temazepa~ (19:0.01-0.3 rag/l), alprazolam (2:0.02 rag/l), and lorazepam (1:0.01 rag/l). In the cases where the urine was positive and the blood negative, both oxazepam and temazepam were found in urine in three cases, and in one case only oxazepam was found, as analyzed by GC-MS. The GC-MS analysis was performed by selected-ion monitoring, using three ions for each compound (oxazepam-tbdms: 457, 458, 459 and temazepam" TBDMS: 357, 359, 283) (5). The detailed examination of the two groups of cases is presented in Tables VI and VII. Discussion The dual-column technique using DB-5 and DB-170I columns has been proven to improve the reliability of the identification of unknown substances in GC drug screenin~l compared to a single-column technique (26). In the present study, the DB-17 column was used instead of DB-1701 be. cause the day-to-day precision of RIs on DB-1701 was poor. This is interesting because in our earlier studies with under" ivatized benzodiazepines the precision of RIs on a similar sta" tionary phase was very good (25). Depending on the compound, the RI values increased or decreased drastically during the 12-week period on DB-1701, unlike on DB-17, where the variation of RI values was smaller and more randor0. Also the detection limits of some compounds, especially 7" aminoclonazepam-tbdms, got higher during the 12-week pe. riod on DB-1701. Possibly, the excess of the silylation reagenl in the injection solvent reacted with the cyanopropyl groups of DB-1701 and changed the chemical composition of the sta' tionary phase. Table V. Comparison of Urine Immunoassay and Blood GC Results Urine (El'S) Blood (GC) No. of cases negative negative 284 positive positive 149 negative positive 48 positive negative 4 invalid' negative 26 invalid' positive 3 Total 514 ' No result obtained by ETS. J _.# 50

Although a few benzodiazepine compounds co-elute on one column (Figure 2), there is no pair of compounds that coelute on both columns. In practice, the only possible combination of benzodiazepines that may cause identification problems is the simultaneous presence of midazolam, oxazepam and nitrazepam. Oxazepam-TBDMS and nitrazepam- TBDMS co-elute on DB-5 and nitrazepam-tbdms and midazolam nearly co-elute on DB:17. Small amounts of nitrazepam may thus be missed. Of nonbenzodiazepine drugs, lamotrigine and some barbiturates, especially cyclobarbital, brallobarbital, and phenobarbital, produce big fronting peaks or a group of peaks and may thus disturb the integration of peaks eluting before them (Table IV). A disadvantage of the method is the fact that the amino Table Vl. Cases Where Urine was Negative by Immunoassay and Blood Positive by GC Case no. GC (blood)(ijg/g) Case history ----... 679 diazepam 0.02 683 desmethyldiazepam 0.03 715 desmethyldiazepam 0.02 716 desmethyldiazepam 0.05, oxazepam 0.01 742 diazepam 0.04, desmethyldiazepam 0.05 750 oxazepam 0.02 774 diazepam 0.02, desmethyldiazepam 0.02 778 temazepam 0.07 781 alprazolam 0.02 784 diazepam 0.1, desmethyldiazepam 0.02, oxazepam 0.02 794 desmethyldiazepam 0.04 795 oxazepam 0.2 808 temazepam 0.05 818 desmethyldiazepam 0.05 829 d iazepam 0.04, desmethyldiazepam 0.1 837 diazepam 0.04, desmethyldiazepam 0.1, temazepam 0.02 871 desmethyldiazepam 0.02 938 diazepam 0.4, desmethyldiazepam 0.1, temazepam 0.01 941 diazepam 0.07, desmethyldiazepam 0.06, temazepam 0.2 947 temazepam 0.06, oxazepam 0.01 972 temazepam 0.01 976 diazepam 0.1, desmethyldiazepam 0.1, oxazepam 0.01, temazepam 0.01, lorazepam 0.01 988 oxazepam 0.4 998 temazepam 0.02 1000 diazepam 0.03 I019 diazepam 0.08 1030 diazepam 0.7, desmethyldiazepam 0.7, oxazepam 0.05, temazepam 0.04 1095 oxazepam 0.03 1106 temazepam 0.3 1129 temazepam 0.01 1132 temazepam 0.01 1147 temazepam 0.04 1210 diazepam 0.03, desmethyldiazepam 0.04 1232 oxazepam 0.02 1242 temazepam 0.03 1243 oxazepam 0.08 1268 diazepam 0.04 1275 temazepam 0.1 1286 temazepam 0.03 1304 temazepam 0.01 1322 diazepam 0.15, desmethyldiazepam 0.01 1337 desmethyldiazepam 0.05 1381 oxazepam 0.02 1384 diazepam 0.3, temazepam 0.01 1391 diazepam 0.02 1400 oxazepam 0.03 1421 alprazolam 0.02 1431 diazepam 0.02,,... died during an operation, resuscitated? hypothermia alcoholic, died in hospital suicide by poisoning (amitriptyline), died in hospital resuscitated suicide (train) traffic accident fracture of the skull, died abroad, embalmed heart disease fracture of the skull, treated 11 h in hospital bolus, ethanol 2.2 g/l, levomepromazine + fluoxetine died in health centre, ethanol 3.6 gll died while drinking alcohol, ethanol 3.5 gll unconscious, survived 12 h in hospital problem drinker, citalopram 4.7 mgll. (blood) suicide by poisoning (doxepin, propranolol), died in hospital heart disease heart disease heart disease bolus in psychiatric hospital suicide by hanging suicide by suffocation unconscious, died on the way to hospital, resuscitated?, drug user, poisoning (mianserin, propranolol, zopiclone, ethanol) suicide by poisoning (carbon monoxide), ethanol 2.8 g/l ethanol 2.6 g/l died in sauna, ethanol 3.5 g/l drowned diabetes, ethanol 3.4 g/l poisoning by ethylene glycol suicide by jumping from a balcony suicide by hanging suicide by shooting suicide by poisoning (citalopram, zopiclone, levomepromazine) suicide by poisoning (amitriptyline) suicide by hanging traffic accident, died in hospital, ethanol 3.0 gll, epilepsy, mentally handicapped suicide by shooting 51

groups of the amino metabolites of clonazepam and flunitrazeparn do not silylate in the present silylation procedure, and the detection limits of 7-arnino rnetabolites are high. 7-Aminoclonazeparn is included, but 7-aminoflunitrazeparn excluded from the screening. The detection of 7-aminonitrazeparn is not possible by EC detector because of the lack of an electron withdrawing group. No GC or GC-MS methods have been reported for the detection of all three 7-amino rnetabolites in the blood. However, HPLC methods are available (23,28). The quantitative precision results (Table II) are based on four determinations during a four-week period using a threepoint calibration and without recalibration during that time. To maintain the performance of the method, quantitative and qualitative calibration should be done at least once a month and the injector liner should be changed once a week. In addition, a few centimeters at the beginning of the column may be cut, if necessary, to maintain a good peak shape of triazoloor irnidazobenzodiazepines. One pair of columns can be used for the analysis of approximately 1400 blood extracts (four months of daily use in our laboratory). Although a few GC methods for screening benzodiazepines in blood have been published earlier, the dual-column approach with two different stationary phases has only been used once (19), and two similar columns with two different detectors (electron capture, nitrogen selective) were used twice (18,20). The quantitative precision of these methods has been generally well documented, and interfering compounds were tested in three papers (18-20), but the qualitative precision, implying the reliability of identification, is hardly discussed at all. The identification of compounds in two published methods was based on the relative retention time (19,20) and on the sole absolute retention time in three methods (16-18), but none of the methods used retention indices as the identification parameter. Although derivatization is widely used in GC-MS to improve the chromatographic properties of benzodiazepines, it has not been used in GC screening procedures in the blood. However, there is a target analysis for alprazolarn and rnetabolites after acylation (29). In the blood-urine comparison study, some of the bloodpositive and urine-negative results (Table VI) can be explained by a recently given dose during resuscitation. In most cases, however, the differences are due to different concentrations in the respective specimens and the performance of the analysis technique. It has been reported that enzymatic hydrolysis of glucuronide conjugates is required to ensure adequately sensitive detection of oxazeparn by EMIT d.a.u. (30) and oxazeparn, Table VII. Cases Where Urine was Positive by Irnrnunoassay and Blood Negative by GC Case no. GC-MS (urine) Case history 658 oxazepam, temazepam alcoholic, 718 oxazepam, temazepam 1118 oxazepam, temazepam homicide 1379 oxazepam jail death temazepam, and lorazepam by other immunoassays (31,32). In Table VI the blood benzodiazepine concentrations are, in all cases, at the therapeutic or subtherapeutic level and obviously have no contribution to the death. However, in crime-related cases, all drug findings may be significant. The four positive urine results in Table VII can be explained by the longer presence of the glucuronide conjugates in the urine than the parent drug in the blood. The results suggest that in postmortem forensic toxicology, benzodiazepines are more reliably detected in the blood by advanced GC methods than in urine by irnmunoassay. Furthermore, the GC method used allows simultaneous quantitation of the drugs. The blood drug concentrations are fairly low in many cases, and consequently, derivatization is required for high sensitivity. The results also reveal the widespread usage of benzodiazepines: 39.7% of the postmortem cases received for drug screening were positive for benzodiazepines. References I. H. Maurer and K. Pfleger. Identification and differentiation of benzodiazepines and their metabolites in urine by computerized gas chromatography-mass spectrometry. J. Chromatogr. 422: 85-101 (1987). 2. S.J. Mule and G.A. Gasella. Quantitation and confirmation of the diazolo- and triazolobenzodiazepines in human urine by gas chromatography/mass spectrometry. J. Anal. Toxicol. 13:179-184 (1989). 3, P.H. Dickson, W. Markus, J. McKernan, and H.C. Nipper. Urinalysis of (z-hydroxyalprazolam, mhydroxytriazolam and other benzodiazepine compounds by GC/EIMS. J. Anal. Toxicol. 16" 67-71 (1992). 4. W.A. Joern. Confirmation of low concentrations of urinary benzodiazepines, including alprazolam and triazolam by GC/MS: an ex, tractive alkylation procedure. ]. Anal. ToxicoL 16:363-367 (1992). 5. R.E. West and D.P. Ritz. GCIMS analysis of five common benzodiazepine metabolites in urine as tert-butyl-dimethylsilyl derivatives. J. Anal. Toxicol. 17:114-I 16 (I 993). 6. R.L. Fitzgerald, D.A. Rexin, and D.A. Herold. Benzodiazepine analysis by negative chemical ionization gas chromatography/mass spectrometry. ]. Anal. Toxicol. 17" 342-347 (I 993). 7. D.A. Black, G.D. Clark, V.M. Haver, J.A. Garbin, and A.J. Saxon. Analysis of urinary benzodiazepines using solid-phase extraction and gas chromatography-mass spectrometry. J. Anal. Toxicol, 18:185-188 (1994). 8. L.E. Edinboro and A. Poklis. Detection of benzodiazepines and tribenzazolams by TRIAGETM: confirmation by solid-phase extraction utilizing SPEC microcolumns and GC-MS. ]. Anal. Toxicol. 18:312-316 (1994). 9. R. Meatherall. GC-MS confirmation of urinary benzodiazepine metabolites. ]. Anal. Toxicol. 18:369-381 (I 994). 10. R.L. Fitzgerald, D.A. Rexin, and D.A. Herold. Detecting benzo" diazepines: immunoassays compared with negative chemical ionization gas chromatography/mass spectrometry. Clin. Chem. 40:373-380 (I 994). 11. C. Moore, G. Long, and M. Marr. Confirmation of benzodiazepine~ in urine as trimethylsilyl derivatives using gas chromatog" raphy/mass spectrometry. J. Chromatogr. B 655:132-I 37 (I 994). 12. S.B. Needleman and M. Porvaznik. Identification of parent ben" zodiazepines by gas chromatography/mass spectroscopy (GC/MS) from urinary extracts treated with l]-glucuronidase. Forensic 5ci. Int. 73:49-60 (I 995). 52

13. J.W. King and L.J. King. Solid-phase extraction and on-disc derivatization of the major benzodiazepines in urine using enzyme hydrolysis and Toxi-Lab VC MP3 column. J. Anal. Toxicol. 20: 262-265 (1996). 14. J.L. Valentine, R. Middleton, and C. Sparks. Identification of urinan/benzodiazepines and their metabolites: comparison of automated HPLC and GC-MS after immunoassay screening of clinical specimens. J. Anal Toxicol. 20:416-424 (1996). 15. O. Beck, Z. Lin, K. Brodin, S. Borg, and P. Hjemdahl. The online screening technique for urinary benzodiazepines: comparison with EMIT, FPIA, and GC-MS. J. Anal. Toxicol. 21:554-557 (1997). 16. H.W. Peel and B.J. Perrigo. Toxicological analysis of benzodiazepine-type compounds in post-mortem blood by gas chromatography. J. Anal Toxicol. 4:105-113 (1980). 17. ].M.F. Douse. Trace analysis of benzodiazepine drugs in blood using deactivated amberlite XAD-7 porous polymer beads and silica capillary column gas chromatography with electron-capture detection. J. Chromatogr. 301: 137-154 (1984). 18. p. Lillsunde and T. Sepp~l~i. Simultaneous screening and quantitative analysis of benzodiazepines by dual-column gas chromatography using electron-capture and nitrogen-phosphorus detection. J. Chromatogr. Biomed. Appl. 553:97-110 (1990). 19. H. Gjerde, E. Dahlin, and A.S. Christophersen. Simultaneous determination of common benzodiazepines in blood using capillary gas chromatography. J. Pharm. Biomed. Anal. 10:317-322 (1992). 20. y. Galliard, J.-P. Gay-Montchamp, and M. Ollagnier. Simultaneous screening and quantitation of alpidem, zolpidem, buspirone and benzodiazepines by dual-channel gas chromatography using electron-capture and nitrogen-phosphorus detection after solid-phase extraction. J. Chromatogr Biomed. Appl. 622:197-208 (1993). 21. p. Mura, A. Piriou, P. Fraillon, Y.Papet, and D. Reiss. Screening procedure for benzodiazepines in biological fluids by high-performance liquid chromatography using a rapid-scanning multichannel detector. J. Chromatogr. Biomecl. Appl. 416:303-310 (1987). 22. E Mul~hoff and T. Daldrup. A rapid solid-phase extraction and HPLC/DAD procedure for the simultaneous determination and quantification of different benzodiazepines in serum, blood and post-mortem blood. Int. J. Leg. Med. 105:105-109 (1992). 23. I.A. Mclntyre, M.L. Syrjanen, K. Crump, S. Horomidis, A.W. Peace, and O.F. Drummer. Simultaneous HPLC analysis of 15 benzodiazepines and selected metabolites in postmortem blood. J. Anal. Toxicol. 17:202-207 (1993). 24. W. He, N. Parissis, and T. Kiratzidis. Determination of benzodiazepines in forensic samples by HPLC with photo-diode array detection. J. Forensic Sci. 43:1061-1067 (1998). 25. I. Rasanen, I. Ojanper~i, E. Vuori, and T.A. Hase. An homologous series of benzodiazepine retention index standards for gas chromatography. J. High Res. Chromatogr. 16:495-500 (1993). 26. I. Rasanen, I. Ojanper~i, J. Vartiovaara, E. Vuori, and P. Sunila. The advantage of dual-column approach and retention indices combined with refined reporting in gas chromatographic drug screening. J. High. Res. Chromatogr. 19:313-321 (1996). 27. L. Huber. Validation of analytical methods: review and strategy. LC-GC Int. 11(2): 96-105 (1998). 28. M.D. Robertson and O.H. Drummer. High-performance liquid chromatographic procedure for the measurement of nitrobenzodiazepines and their 7-amino metabolites in blood. J. Chromatogr. Biomed. Appl. 667:179-184 (1995). 29. W.A. Joern and A.8. Joern. Detection of alprazolam (Xanax) and its metabolites in urine using dual column, dual nitrogen detector gas chromatography. J. Anal. Toxicol. 11: 247-251 (1987). 30. O. Beck, P. Lafolie, G. Odelius, and L.O. Boreus. Immunological screening of benzodiazepines in urine: improved detection of oxazepam intake. Toxicol. Lett. 52:7-14 (1990). 31. P. Simonsson, A. Lid~n, and S. Lindberg. Effect of l~-glucuronidase on urinary benzodiazepine concentrations determined by fluorescence polarization immunoassay. Clin. Chem. 41:920-923 (1995). 32. R. Meatherall. Benzodiazepine screening using EMIT ~ I1 and TDxe: urine hydrolysis pretreatment required. ]. Anal To 18: 385-390 (1994). Manuscript received February 1, 1999; revision received April 26, 1999. 53