Forensic Applications of the Determination of Benzodiazepines in Blood Samples by Microcolumn Cleanup and High-Performance Liquid Chromatography with Reductive Mode Electrochemical Detection J.B.F. Lloyd and D.A. Parry Home Office Forensic Science Laboratory, Gooch Street North, Birmingham B5 6QQ, United Kingdom l Abstract I Recently described microcolumn cleanup and highperformance liquid chromatography with electrochemical reductive mode detection have facilitated the determination of benzodiazepines in contaminated and degraded blood samples. Examples from an extensive application of the procedure to British forensic science casework are given here. Introduction Recently introduced procedures based on high-performance liquid chromatography (HPLC) with electrochemical reductive mode detection and on microcolumn cleanup enable the determination of benzodiazepines and their metabolites in small samples of degraded and contaminated bloods, which are typical of forensic science work (l). Depending on the actual benzodiazepine, detection limits and recovery values are in the respective ranges of 0.3-12 ng / ml and 75-95 070, and selectivities greatly exceed those of the HPLC techniques usually applied to this type of sample. Examples drawn from eighteen months' experience with these procedures are given in this report. Experimental The procedures and the usual validating data are given in detail elsewhere (1). Briefly, centrifugally filtered blood samples (usually, 100/zL), diluted with 250 #L of 0.2070 w/v sodium octyl sulfate in aqueous 0.1M disodium hydrogen phosphate to suppress protein binding, were extracted with 5 mg of loose Porapak-T (Millipore-Waters). The extractant was transferred to a microcolumn prepacked with an equal amount of clean Porapak-T. From the column the benzodiazepines were eluted in 60 #L water/acetonitrile (ca. 1:2 v/v, depending on the activity of the adsorbent). The eluate was chromatographed on ODS-Hypersil (Shandon), 3/~m, 150 x 4.5 mm, at 40~ in deoxygenated methanol/l-propanol/aqueous phosphate (0.02M, ph 6), 100:8:80 v/v/v (1). Detection was at a pendent mercury drop electrode (EG&G PARC; model 310, modified (2)), which was held, usually, at - 1.2 V vs. Ag/AgCI. The injection volume of the deoxygenated microcolumn eluate was 10 #L. Coefficients of the within-day variation of the retention times were in the range 0.5-1.0~ questioned peaks were identified with reference to standards included in accompanying chromatograms. The variation in retention times apparent in the figures reflects the variation seen in 18 months' work with several columns (each prepared from the same batch of packing material). The eluent composition was varied slightly to accommodate changes in column selectivity with age (1). For work within the quoted detection limits or to reduce the setup time, the mercury electrode was preceded by a coulometric detector (Environmental Sciences Associates, model 5010). The porous carbon working electrodes were held at 0 V (proprietary reference electrode). This arrangement removed the more readily reduced traces of impurities in the effluent stream and allowed oxidative detection of other compounds. We strongly emphasize that, along with some other straightforward precautions (l), the most efficient and practicable way to deoxygenate an HPLC eluent for electrochemical detection is to maintain the eluent reservoir under a slow reflux while the chromatograph is in use. Quantitative analyses were calibrated against spiked hemolyzed human blood. The internal standard was either a benzodiazepine selected to occupy a vacant position in the chromatograms or 6-nitroquinoline when the use of a benzodiazepine was impracticable, for example, in the examination of previously uncharacterized samples or a collection of samples in which a considerable variety of benzodiazepines and metabolites was present. Results and Discussion To summarize the results and to place them in perspective, details from 18 months' casework are given in Table I. The quoted concentrations of parent drug include the equivalent amount of the common metabolites, which are also detected. Benzodiazepines were found in 82 of the 118 samples that were analyzed. The total of the cases column, 143, exceeds the number of samples because more than one benzodiazepine had been ingested in some cases. Most of the samples (73~ were taken post-mortem; the others were mainly from subjects involved Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 163
in motoring offenses. Except for temazepam the two sets of data do not significantly differ, and all have been tabulated together, therefore. The means of the respective temazepam concentrations are 2878 ng/ml (N = 26) and 438 ng/ml (N = 9). Two of the postmortem results exceed 10 #g/ml. Alcohol and other drugs were also involved in these deaths. In most instances the tabulated results exceed the range expected from therapeutic plasma levels (3) because of the influence of overdose cases. The differences are particularly marked if account of the hemolyzed state of the samples is taken. The conventionally quoted plasma concentrations include benzodiazepines extensively bound to serum albumin (4) together with any remaining unbound drug. The liberation of the erythrocyte contents, which contains negligible amounts of bound drug, lowers the measured concentrations of hemolyzed samples. The techniques were designed to give distinguishable peaks on a single chromatogram of all the benzodiazepines and their metabolites important to British casework. A chromatogram of these compounds, together with the 6-nitroquinoline used as an internal standard, is shown in Figure 1. The compounds are listed in the figure caption. Common to all of these benzodiazepines, and to the benzodiazepines in general with the exception of clobazam (unencountered in any casework), is an azomethine group which, occasionally with some others, is responsible for the reductive response. Straightforward modification of the chromatography conditions permits application to the benzodiazepines apart from those specifically considered here (1). In contrast to clinical samples, the blood samples encountered in forensic science are often severely degraded, and in addition to the benzodiazepines, their metabolites, and their degradation products, exogenous contaminants may be present. An example from a road traffic case is shown in Figure 2 (chromatogram A), where there are six peaks assigned to the benzodiazepines and metabolites with concentrations varying from 45 ng/ml (oxazepam) to 398 ng/ml (diazepam). The chromatogram includes a peak thought to be due to an uncharacterized metabolite (seen only in such cases), an internal standard peak, and a peak due to contamination from the particular type of septum fitted to the sample container. This last peak often Qccurs but does not interfere with any of the benzodiazepines. The other relatively weak and early emerging peaks occur in various amounts from blood samples generally. A negative chromatogram from a sample taken post-mortem is also shown in Figure 2 (chromatogram B). As well as an examination of the variation of the HPLC response with the detector electrode potential (1), the identifications may be confirmed by direct mass spectrometric examination of the microcolumn eluates (5), or the presence of benzodiazepines generally may be detected by a broadly specific radioimmunoassay (RIA) screening procedure (6). A comparison between the R1A and the HPLC results from a variety of cases positive by both techniques (21 samples) is shown in double logarithmic coordinates in Figure 3, for which the HPLC results have been converted to the diazepam units of the RIA by means.of cross-reactivity factors. Although the RIA gives only a semiquantitative indication of the presence of a drug (6), which is shown by the substantial scatter from a one-to-one relationship (broken line) in Figure 3, there is a clearly evident correlation between the two sets (r = 0.865, N = 21). This is particularly so in cases involving diazepam (r = 0.985, N = 6), with which the RIA is calibrated for screening purposes. With the inclusion of negative results from either technique, 28 comparisons were made. There was only one instance in which HPLC revealed no benzodiazepine in an RIA-positive sample (70 ng/ml diazepam equivalents). Possibly the RIA result was due to residual uncharacterized metabolites, because for most benzodiazepines the techniques are comparable in detection limits. In the original comparison (6) the applied HPLC (ultraviolet absorbance detection) and gas chromatography (electron capture and alkali flame ionization detection) failed to confirm 9 f na Table I. Summary of Results of Analyses for Benzodiazepines in 118 Casework Blood Samples Concentration in blood Parent (nglml)* Therapeutic range in bonzodlazeplne Cases Mean Range plasma (nglml)*" Chlordiazepoxide 6 2274 293-7350 370-3200 Clonazepam 1 5 13-72 Oiazepam 28 820 69-3862 500-2500 Flurazepam 1 1870 13-190 Lorazepam 12 203 12-479 17-70 Nitrazepam 19 845 32-2084 30-100 0xazepam 3 607 297-983 100-1400 Temazepam 35 2172 58-13210 600-900 Triazolam 2 37 36-38 3-20 None found 36 9 Includes metabolites expressed as parent compound *9 Based on ref. 3. 8 12 t t 1 1 1 1 1 o 9 rain. Figure 1. Chromatogram of a standard mixture: 1 = 7=aminonitrazepam; 2 = 7-acetamidonitrazepam; 3 = 6-nitroquinoline; 4 = demoxepam; 5 = nitrazepam; 6 = triazolam; 7 = Iorazepam; 8 = oxazepam; 9 = Ioprazolam; 10 = temazepam; 11 = Iormetazepam; 12 = chlordiazepoxide; 13 = desmethyldiazepam; 14 = diazepam. Each peak is due to a 10-ng amount, except for peak 3 (5 ng). 164
Journal of Analytical Toxicology, Vol 13, May/June 1989 the presence of benzodiazepines in 19 of 55 RIA-positive samples. A direct comparison of!;he performance of the reductive and the UV detection techniques in the HPLC of 14 samples has demonstrated that the UV technique may often be vitiated by irrelevant sample components, whereas the reductive technique is unaffected (l). Examples of a positive HPLC result in the face of a negative or weak response in the RIA have occurred only in nitrazepam cases. In these the major chromatographic peak was usually due to 7-aminonitrazepam, which exhibits negligible cross-reactivity in the RIA. In Table I the concentrations given as equivalent to nitrazepam are largely due to the amino compound. For these 19 cases the mean nitrazepam and aminonitrazepam levels were 57 and 704 ng/ml respectively, the nitrazepam figure mainly being due to two high values, 241 and 512 ng/ml. The latter was from a sample taken post-mortem, and the other was from a defendant in a road traffic case. Usually little or no nitrazepam was present. Examples are given in Figure 4. Chromatogram A shows the presence of the aminonitrazepam (525 ng/ml) and the complete absence of nitrazepam. The RIA result was negative, but there was independent evidence of the ingestion of nitrazepam. Chromatogram B, which exhibits an offscale peak of the aminonitrazepam (1500 ng/ml), is from a similar case. Chromatogram C is from a case in which both the aminonitrazepam and nitrazepam were found (460 and 512 ng/ml respectively); the corresponding R1A response was augmented considerably by the diazepam and the desmethyldiazepam also present (5 and 116 ng/ml respectively; this part of the chromatogram is not included in C). The effect of a less cathodic potential on the peaks of the last-mentioned sample is illustrated in D. The result is characteristic of the peak assignments (1). From clinical work, the major metabolites of nitrazepam are the 7-amino and 7-acetamido derivatives, which occur in blood at levels roughly comparable to that of any remaining nitrazepam (7). That only the amino compound was consistently present in the samples of interest here is apparently due to reactions occurring post-mortem or during storage. This was shown in some brief experiments conducted with a variety of drugfree blood samples to which nitrazepam and its metabolites were individually added. It was found that at room temperatures over several days nitrazepam was lost and replaced in various amounts with the 7-amino derivative. The effect was suppressed in samples containing fluoride preservative or stored at 4 ~ The 7-acetamido compound was similarly lost, but the 7-amino compound was stable under the conditions. Our previous paper shows that the determination of 7-aminonitrazepam by HPLC 3 6 8 9 t na 7 A J, O 10 Figure 2. Casework example of a positive blood sample from a defendant in a road traffic case (A) and of a negative blood sample taken post-mortem (8) 1 = 7-aminonitrazepam; 2 = uncharacterized metabolite; 3 = 6-nitro- min. B r I I I I I I I I I I 0 10 quinoline (internal standard); 4 = nitrazepam; 5 = oxazepam; 6 = temazepam; 7 = septum component; 8 = desmethyldiazepam; 9 = diazepam 165
) 100 lop 1P N J" L // /J L J~ J,J JJ f I 1 10./ D" T," I. N RIA ng ml "I 9 0"" D J T j," M / I s "N T D 0 T D, I I 100 1000 Figure 3. Comparison of HPLC and RIA casework results. The HPLC results have been converted to the diazepam equivalents of the RIA. Each point indicates the component responsible for the main RIA crossreactivity. D = diazepam; L = Iorazepam; M = desmethyldiazepam; N = nitrazepam; 0 = oxazepam; T = temazepam. The broken line indicates the response required by a one-to-,~ne relationship; correlation coefficients are given in the text. T with UV detection is subject to severe interference by components of degraded blood samples (1). This, coupled with the present results and with the insensitivity of the RIA screening to the aminonitrazepam, explains the low detection rate sometimes seen in earlier nitrazepam casework. Apparently the circumstances of temazepam cases in forensic science also give rise to differences from the clinical results, which indicated that the desmethyl metabolite (oxazepam) should appear only in small amounts (8). Thus, in 35 temazepam cases (Table I), none of which showed evidence of separate oxazepam ingestion or that the temazepam was the product of diazepam metabolism, the level of oxazepam, expressed as temazepam equivalents relative to the remaining temazepam, varied over the range < 0.2-59~ Chromatograms exemplifying extremes of the range are given in Figure 5. In these cases the respective concentrations of temazepam and oxazepam were 760 and 430 ng/ml (chromatogram A) and 5.7/zg/mL and 50 ng/ml (chromatogram B). To date, no common factor within cases at either extreme has emerged. Both compounds were shown to be stable in the samples received. Flunitrazepam has not occurred in any material to which the new techniques have been applied. In view of the recently reported difficulty of detecting this compound and its metabolites (9), however, some experiments were made on spiked blood samples. We concluded that this group of compounds may be [ t t 2! JL_ I i I 270 sec. Figure 4. Three nitrazepam cases (A-C) and an example (D) of the effect of a modification of the detector potential from - 1.2 V to - 1.0 V vs. Ag/AgCI in case C. The same scale expansion was used in C and D; otherwise it was arbitrarily varied. 1 --- 7-aminonitrazepam; 2 = 6-nitroquinoline (internal standard); 3 = nitrazepam. Concentrations are given in the text. 166
[ Irl.8 na 1 1!! 420 sec. I I I! 420 sec. Figure 5. Chromatograms from blood samples taken post-mortem in two temazepam overdoses (A,B). 1 = 6-nitroquinoline (internal standard); 2 = oxazepam; 3 = temazepam. Concentrations are given in the text. Figure 6. Chromatograms from blood spiked with flunitrazepam and two metabolites (A) and unspiked blood (B). 1 = 7-aminoflunitrazepam (20 ng/ml); 2 --= 6-nitroquinoline (internal standard); 3 = desmethylflunitrazepam (10 ng/ml); 4 = flunitrazepam (10 ng/ml). The eluent given in the experimental was modified by the omission of the l-propanol: determined similarly to the other benzodiazepines. An example is given in Figure 6. Hemolyzed human blood was spiked at 10 ng/ml with flunitrazepam and desmethylflunitrazepam and at 20 ng/ml with aminoflunitrazepam (chromatogram A). The unspiked blood is also shown (chromatogram B). Both samples included the usual internal standard. To remove flunitrazepam from the nitrazepam position, the eluent was modified by the omission of 1-propanol. For the interpretation of results from actual casework, account of degradation processes similar to those of nitrazepam cases may be needed. The new cleanup and detection techniques are likely to be relevant to other drugs and toxic materials. An obvious example is the alkyl nitrate vasodilators and their metabolites. These techniques were developed and are routinely used for this type of compound--in the form of traces of explosives and firearm residues (10). Polychlorinated compounds are another example, an instance of which occurred in a benzodiazepine case, shown in Figure 7. In addition to temazepam (1200 ng/ml) and oxazepam (280 ng/ml), the sample gave a peak due to trichloroethanol (3 #g/ml) from the metabolism of chloral hydrate. Other possible applications may be found in the extensive literature of polarography. Used with other detection techniques, the cleanup procedure could be widely applicable. (It should be noted that the detection technique exploited here is not a polarographic technique: each chromatogram is detected at a single mercury drop held at a constant potential.) References 1. J.B.F. Lloyd and D,A. Parry. Detection and determination of common benzodiazepines and their metabolites in blood samples of forensic science interest: Microcolumn cleanup and high- 3 na i! 0 sec. 420 Figure 7. Chromatogram from the postmortem blood of a subject prescribed chloral hydrate and temazepam. 1 = trichloroethanol; 2 = 6-nitroquinoUne (internal standard); 3 = oxazepam; 4 = temazepam. Concentrations are given in the text. 167
performance liquid chromatography with reductive electrochemical detection at a pendent mercury drop electrode. J. Chromatogr. 449:281-97 (1988). 2. J.B.F. Lloyd. High-performance liquid chromatography of organic explosives components with electrochemical detection at a pendent mercury drop electrode. J. Chromatogr. 257:227-36 (1983). 3. M.D. Osselton. Toxicological tables: A compendium of pharmacological, therapeutic and toxicological data on 136 drugs and chemicals in humans. Bull. InL Assoc. Forensic Toxicol. 17: 16-33 (1983). 4. E.M. Sellers, C.A. Naranjo, V. Khouw, and D.J. Greenblatt. Binding of Benzodiazepines to Plasma Proteins. In Pharmacology of Benzodiazepines, E. Usdin, P. Skolnick, J.F, Tallman, D. Greenblatt, and S.M. Paul, Eds. MacMillan, London, 1982, pp. 271-84. 5. S.R. Blunt, J.B.F. Lloyd, and D.A. Parry. Unpublished work. 6. C.P. Goddard, A. Stead, P.A. Mason, B. Law, and A.C. Moffat. An iodine-125 radioimmunoassay for the direct detection of benzodiazepines in blood and urine. Analyst 111:525-29 (1986). 7. K. Tada, A. Miyahira, and T. MorojL Liquid chromatographic assay of nitrazepam and its main metabolites in serum, and its application to pharmacokinetic study in the elderly. J. Liq. Chromatogr. 10:465-76 (1987). 8. H.J. Schwarz. Pharmacokinetics and metabolism of temazepam in man and several animal species. Br. J. C/in. Pharmacol. 8: 23S-29S (1979). 9. B. Heyndrickx. Fatal intoxication due to flunitrazepam. J. Anal. Toxicol. 11 : 278 (t987). 10. J.B.E Lloyd. Liquid chromatography with electrochemical detection of explosives and firearms propellant traces. Anal. Proc. 24: 239-40 (1987). 168