A high degree of concordance was observed between the Online and FPIA. (fluorescence polarization immunoassay) systems when analyzing. 138 randomly ...
Journal of Analytical Toxicology, Vol. 21, November/December 1997
The Online ScreeningTechniquefor Urinary Benzodiazepines:Comparisonwith EMIT, FPIA,and GC-MS* Olof Beck 1,+, Zhen Lin 1, Kerstin Brodin 1, Stefan Borg 2, and Paul Hjemdahl 1
1Department of Clinical Pharmacology, Karolinska Hospital, S- 171 76, Stockholm, Sweden and 2Department of Clinical Neuroscience, Karolinska Institute, Psychiatry Section at St. G6rans Hospital, Stockholm, Sweden.
Abstract Three commercial immunoassaysystems(EMIT, FPIA, Online) for the screening of benzodiazepines in urine were evaluated using authentic patient sampleswith gas chromatography-mass spectrometry (GC-MS) as the reference method. The Online system (kinetic interaction of microparticles in solution) gained in performance by applying a 100-ng/mL cutoff limit and by incorporating 13-glucuronidasetreatment, which could be automated on the Cobas Mira Plus instrument. When using enzymatic hydrolysis, all three immunoassaysystemshad high levels of sensitivity, including samples containing only flunitrazepam and nitrazepam metabolites. A high degree of concordance was observed between the Online and FPIA (fluorescence polarization immunoassay)systemswhen analyzing 138 randomly selected patient samples.The EMIT II and EMIT d.a.u. (enzyme multipled immuno technique) systemsgave a higher number of positive results, but the presence of benzodiazepines could not be verified by GC-MS in a substantial number of these cases. The rate of unconfirmed positive results was increased when enzyme hydrolysiswas incorporated in the EMIT II assay.Although differences in the performances of the investigated assaysystems were observed, they all seem appropriate for clinical use in detecting benzodiazepine intake in drug abusers when enzymatic hydrolysis is included.
zodiazepines (1). Existing immunological assays for benzodiazepines generally suffer from a lack of sensitivity with regard to several relevant compounds among the broad range of parent drugs and metabolites excreted in urine following intake. This has led to a situation in which individual laboratories have modified existing technologies to better meet the clinical need of method performance (2,3). Several benzodiazepines and metabolites are glucuronidated before being excreted into urine (4), and these glucuronidated metabolites are often poorly detected by the assays (1,3,5-7). Therefore, many laboratories routinely include manual enzyme hydrolysistreatment of urine before the immunoassay (5-8). It is important to evaluate new assays and their modifications. In this study, we evaluated the Roche Online (Roche Diagnostic Systems, Branchburg, NJ) system (KIMS, kinetic interaction of microparticles in solution) and compared it with two EMIT (enzyme multipled immuno technique, Syva, San Jose, CA) assays and our existing routine procedure, which is based on a modified FPIA assay (fluorescence polarization immunoassay, Abbott Laboratories, Abbott Park, IL) (6). Another aim of this study was to extend our previous study on healthy volunteers (6) by using samples from patients being treated for drug abuse.
Experimental
Introduction
Benzodiazepines are valuable therapeutic drugs, but they can also produce dependence and abuse, which are often denied by the patient. Urinalysis offers an effective instrument for the detection and treatment of benzodiazepine-dependent patients. The clinical value of a laboratory diagnostic test depends on its sensitivity and specificity.Methodologicalproblems have made it difficultto find reliable urine screening assay systems for ben-
Urine samples
The urine samples were randomly selected from the flow of clinical routine samples from drug abusers treated in the Stockholm County Methadone Program, which were sent to our laboratory for benzodiazepine screening. Because of the low frequency of lorazepam use in these patients, samples were also taken from a previous study of samples obtained after lorazepam intake (3.75-mg dose) in healthy volunteers (6). Analytical procedures
* A preliminaryreportof this work was presentedat the 4th InternationalCongressof Therapeutic Drug Monitoring and Clinical Toxicology,Vienna,Austria,September1995. *Author to whom correspondenceshould be addressed.
554
Assays by FPIAwere performed with ADx instruments using reagents from Abbott. The procedure was modified by lowering
Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.
Journalof AnalyticalToxicology,Vol. 21, November/December1997
the cutoff limit to 100 ng/mL (calibrator nordiazepam) and by including manual preincubation with ~-glucuronidase from E. coli, supplied in 50% glycerol (Boehringer-Mannheim, Mannheim, Germany) (6). Assays by EMIT II and EMIT d.a.u, were performed on a Hitachi 717 instrument (Boehringer-Mannheim)with reagents from Syva,and according to the instrumental protocol supplied by Syva.The EMITII reagent was also applied on a Cobas Mira Plus instrument (Roche Diagnostics)with automated preincubation with l~-glucuronidase.The applied cutoff limit for both EMIT reagents was 100 ng/mL (calibrator, oxazepam). Online assays were performed on a Cobas Mira Plus instrument with reagents from Roche. An optimized procedure, which enabled automated preincubation with E. coli glucuronidase,was kindly suppliedby Dr. AlanJ. McNallyat Roche Diagnostic Systems. The modified program of the Cobas Mira used 5 ]JL (1 U) of ~-glucuronidaseto a samplevolume of 18 IJL. The effectiveincubation (at 37~ delay time was 200 s (eight machine cycles). The following instrument settings on the Table I. Comparison of Different Cutoff Limits using the Online System With and Without Enzyme Hydrolysis* Positiverate (%)
Online cutoff (ng/mt)
no hydrolysis
with hydrolysis
100 200 300
68 58 48
90 84 68
Cobas Mira were used: sample cycle, 1; volume, 18.0 ]JL; diluent, 77.0 pL; reagent cycle, 1; volume, 100 IJL;start reagent, 1; cycle, 2; volume, 5.0 pL; diluent, 3.0 IJL; start reagent, 2; cycle, 8; volume, 30.0 pL; diluent, 40.0 pL; number of steps, 1; calculated step A endpoint; readings first 7, readings last 16. Unlessotherwise stated, the cutoff limit appliedwas 100 ng/mL (calibrator, nordiazepam). The detection limit of the assay was 17.7 ng/mL, according to the manufacturer's specifications. Identification and quantitation (verificationassay) of benzodiazepines in the samples were performed by quadrupole gas chromatography-massspectrometry(GC-MS),employingenzymatic pretreatment of the urine (the same as in the screening assay), solid-phaseextraction,and the formationof trimethylsilyl derivatives (partly described in reference 9). The following analyteswere identifiedand quantitated using selectedion monitoring at two or three mass numbers: oxazepam, diazepam, nordiazepam, temazepam, lorazepam, and internal standard (oxazepam-ds); 7-amino-nitrazepam, 7-amino-flunitrazepam, and internal standard (7-amino-l-methylclonazepam);or alprazolam, r162 and internal standard ((zhydroxyalprazolam-ds).The cutoff limits for identification and quantitation were between 50 and 72 ng/mL (0.2 IJmol/L)for all analytes. The within-day coefficients of variation at 2.5 times these cutoff limits were between 2 and 8%.
Results Evaluation of the Online screening system
The effects of applying different cutoff limits and enzyme hydrolysisin the Online screening system were evaluatedusing
* The study was done with 50 patient samples that were positive for benzodiazepines with our routine FPIAscreening (5).
Table II. Sensitivity* of Screening Techniquest in Detecting Benzodiazepines in Patient Urine* 7-Amino-flunitrazepam
Concentration > 300
7-Amino-nitrazepam
% positive
N
% positive
N
69 31 54 62 100 46
13 13 13 13 11 13
100 75 75 100 66 100
4 4 4 4 3 4
90 90 95 I00 I00 100
20 20 20 13 13 13
91 73 82 91 100 73
11 11 11 11 8 11
Oxazepam, nordiazepam, temazepam
Multi-components
N
% positive
N
100 27 73 100 100 73
11 11 11 11 8 II
100 66 66 66 100 66
3 3 3 3 3 3
100 92 97 95 97 95
39 39 39 39 35 39
100 85 100 100 100 100
13 13 13 13 6 13
% positive
ng/ml.
FPIAw(modified) Online Onlinew EMIT II EMIT II~ EMITd.a.u.
Concentration> 300 ng/ml FPIAw(modified) Online Onlinew EMIT II EMIT II~ EMITd.a.u.
* Percent with positive results. t Applied with 100-nglmL cutoff. * All samples contained benzodiazepines or metabolites according to GC-MS. w With enzyme hydrolysis.
555
Journalof Analytical Toxicology,Vol. 21, November/December1997
50 randomly selected patient samples that were positive in our routine FPIAscreening. The positive rate was clearly increased by incorporating hydrolysis and applying a lower cutoff limit (Table I). The hydrolysis could be automated on the Cobas Mira instrument. The automated procedure was compared with manual incubation in 50 patient samples. This showed that more effectivehydrolysis was achieved with the automated procedure, as evidenced by an increased number of positive samples. Reliable quantitative results (greater than two times the detection limit, 17.7 ng/mL) were obtained in 26 samples using the Cobas Mira instrument either with or without hydrolysis, and these results were used to estimate the effectiveness of the automated hydrolysis. Compared with manual hydrolysis, the mean immunoresponse increased from 118 to 141 ng/mL (+19%) (p < 0.001, paired t-test) and, compared with unhydrolyzed urine, the mean immunoresponse in these samples increased from 77 to 141 ng/mL (+83%) (p < 0.001, paired t-test). The automated hydrolysis procedure was further validated by analyzing 25 negative clinical samples (blank readings in FPIA). In 22 of those samples, responses from the Online system were below 17.7 ng/mL (detection limit), and the remaining three samples had responses below 25.
of positive outcome was obtained in samples with concentrations lower than 300 ng/mL. Because no patient samples containing lorazepam were found, samples from healthy volunteers were used to study this compound. Results for the various screening systems using samples co]lected after intake of 3.75 mg lorazepam were as follows (given as percent positive): modified FPIA, 90%; Online, 0%; Online with enzyme hydrolysis, 85%; EMITII, 0%; EMIT II with enzyme hydrolysis, 85%; EMIT d.a.u., 0% (N = 20). The screening systems were also compared using 138 randomly selected patient samples without knowledge of the FPIA or GC-MS results. The results are presented in Table III and show a high degree of concordance between FPIA and Online when hydrolysis was incorporated into the assay. Benzodiazepines were detected by GC-MS in 1 of 3 samples that were positive with Online but negative with FPIAand in 3 of 4 samples that were positive with FPIA but negative with Online. Enzymatic hydrolysis made 20 Emit II negative samples turn positive, but benzodiazepines were detected by GC-MS in only nine of those samples. Comparison of FPIA and Online with EMIT II including enzyme hydrolysis showed 35 additional positive results, but benzodiazepines were detected by GC-MS in only 15 of them. Also, unhydrolyzed samples which were posComparisonof the FPIA, EMIT, and Online systems itive with EMIT II or EMIT d.a.u, but negative with FPIA had significant numbers of unconfirmed positives (35 and 57%, The sensitivities (rate of detecting true positives) of several respectively). Positive results were obtained with all immunoscreening assays were evaluated using patient samples found to assay systems in 36 of the samples. The presence of benzocontain benzodiazepines by GC-MS. The samples were divided into categories according to concentration and type of drug diazepines was verified by GC-MS in al] but four of these samples. Negative results were obtained with all immunoassay found by GC-MS (Table II). This study also showed that enzyme hydrolysis increased the ability of the Online system to detect systems in 56 of the samples, and they were not further invessamples containing benzodiazepines or their metabolites. The tigated by GC-MS. Thus, the concordance between all screenpositive rate was increased by incorporating enzymatic hydroling systems was 67%. Benzodiazepines were detected in 48% of ysis with the EMIT II system. Overall, the different systems the nonconcordant samples. showed high degrees of detectability for all groups of comThe sensitivities and specificities of the various screening pounds at concentrations greater than 300 ng/mL. A lower rate systems in detecting benzodiazepines, which are based on results obtained in the two sets of clinical samples, are shown in Figure 1. The sensitivities were calculated using all data [ - ~ Sensitivity [ ~ Specificity presented in Table II, and the specificities )0 -~ ............................. -9-2- . . . . . . . . ~ ................. were calculated using samples with posi89 86 87 82 85 o,~ tive outcomes as presented in Table III. [ ] Results from GC-MS analyses were used ...... ~ ...... ~6 .......... '5 as references. ---
i0
!5
0
:,2:
..........
. . . . . . . . . . . . . . . .
1
~ ......
~
.......
~
. . . . . . . .
. . . . . . . . . . . . . . . . .
~
~ ......
~
.......
~
. . . . . . . . .
Discussion
_
FPIA*
Online
Online*
EMIT II
EMIT I1"
EMITd.a.u.
Figure 1. Summary of estimated sensitivities and specificities of the various screening techniques (applied with cutoff, 100 ng/mL) for the detection of benzodiazepines in urine. Sensitivities were calculated from Table IJ, and specificities were determined by comparing positive results presented in Table III with results obtained by GC-MS. Asterisks indicate that enzyme hydrolysis was included in the assay procedure.
556
The benefit of hydrolyzing conjugated benzodiazepines and their metabolites before analysis with the Online reagent system is in agreement with earlier studies with FPIA or EMIT systems (1,5-8). This is most likely due to a common failure of the antibodies used in existing methods to bind conjugated forms. The enzymatic hydrolysis can be
Journal of Analytical Toxicology, Vol. 21, November/December1997
Table III. Comparison of Screening Techniques for Detection of Benzodiazepines in Authentic Patient Samples FPIA
Online* EMIT II EMITII* EMITd.a.u.
POS NEG POS NEG POS NEG POS NEG
Online *
POS
NEG
POS
NEG
42 4 40 6 45 1 41 5
3 89 20 72 35 57 14 78
41 4 45 0 42 3
19 74 35 58 13 80
EMIT II
POS
NEG
60
20 58 4 74
0
50 10
* Includesenzymehydrolysis.
automated and become integrated in the immunoassay procedure using the Cobas Mira instrument, on which both Online and EMIT reagents can be used. Our study using authentic patient samples demonstrates a good overall ability to detect a broad range of benzodiazepines for all three systems studied, even if differencesexist between the systems. The observationthat lorazepam cannot be detected by any of the studied techniques without hydrolysisis in agreement with previous results (5,7). In addition, a significant increase in the detectability of oxazepam, temazepam, diazepam, chlordiazepoxide, nitrazepam, flunitrazepam, alprazolam, and triazolam intake is obtained by performing hydrolysis (5,7-8). Hydrolysis did not increase the response for flurazepam, but sensitivity for this compound was high with the EMIT II and FPIA immunoassay systems (7). In discussing the benefit of enzymatic hydrolysis, it is pertinent to stress that different [3-glucuronidaseenzymes occur commercially.Differencesare likely to exist in the ability to hydrolyze glucuronidated benzodiazepine and metabolites, but this has not been studied in detail. Therefore, studies aimed at optimizing the hydrolysis procedure by considering different enzymes and substrates (glucuronidatedbenzodiazepinesand metabolites)might prove profitable. Another important feature of this study was the documentation of sensitivitiesand specificitiesof the systems in detecting drug intake by studying authentic patient samples.The ingested drug was not prescribed in the majority of the cases. Therefore, the present work extends our previous study with controlled intake of therapeutic doses by healthy volunteers (5). Most importantly, the results indicate that the modified assays are reliable in detecting benzodiazepines actually chosen by drug abusers. Conversely,without modification of, for example,the Online system, we would have missed a substantial number of the positive samples. It has previously been pointed out that immunoassay systems lack sufficient sensitivity to detect certain commonly prescribed benzodiazepines (e.g., nitrazepam, flunitrazepam, and lorazepam) (10-12). Apart from performing enzymatic hydrolysis,it is also important to apply a low cutoff limit. Both the FPIA and Online systems performed well with
EMIT I1'
POS
NEG
the 100-ng/mL cutoff, but the EMIT system showed a tendency towards poorer specificity. However,this could be acceptable if reliable verification procedures are available. Therefore, the 100-ng/mL cutoff limit should still be generally applied in the screening assay.
Conclusion 54 26
I 57
In summary, the present study demonstrates the possibility of performing automated enzyme hydrolysisto obtain a reliablescreeningof urine for benzodiazepinesin authentic clinical samples.
References I. A.D. Fraser.The benzodiazepines: a controversial class of drugs in the 1990s. AACC In-$en4ce Training and continuing Education. Washington, D.C. 13(15): 7-I 3 (I 992). 2. R.L. Boeckx. User modification of commercial immunoassay kits. Clin. Chem. 38:1402-1403 (1992). 3. O. Beck, P. Lafolie, and P. Hjemdahl. Modification of commercial assay kits is justified. Clin. Chem. 39:1352-53 (1993). 4. S. Kaplan and M. lack. Metabolism of the benzodiazepines: pharmacokinetic and pharmacodynamic considerations. In The Benzodiazepines: From Molecular Biology to Clinical Practice. E. Costa, Ed. Raven Press, New York, NY, 1983, pp 173-99. 5. O. Beck, R Lafolie, G. Odelius, and L. Bor(~us. Immunological screening of benzodiazepines in urine: improved detection of oxazepam intake. Toxicol. Lett. 52:7-14 (1990). 6. O. Beck, P. Lafolie, R Hjemdahl, S. Borg, G. Odelius, and R Wirbing. Detection of benzodiazepine intake in therapeutic doses by immunoanalysis of urine: two techniques evaluated and modified for improved performance. Clin. Chem. 38:271-75 (1992). 7. R. Meatherall. Benzodiazepine screening using EMIT II and TDx: urine hydrolysis pretreatment required. J. Anal. Toxicol. 18: 385-90 (1994). 8. R Simonsson, A. Liden, and S. Lindberg. Effect of 13-glucuronidase on urinary benzodiazepine concentration determined by fluorescence polarization immunoassay. Clin. Chem. 41:920-23 (1995). 9. Z. Lin and O. Beck. Procedure for verification of flunitrazepam and nitrazepam intake by gas chromatographic-mass spectrometric analysis of urine. J. Pharm. Biomed. Anal. 13:719-22 (1995). 10. E. Minder, R. Schaubhut, and F. Simmler. Toxicological screening for benzodiazepines in urine: EMI'PTM versus high-performance liquid chromatography with photo-diode array detection. ToxicoL Lett. 45:93-99 (1989). 11. I. B~umler. Der analytische Nachweis von akuten Vergiftungen. MTAJ. 7:360-64 (1985). 12. H. SchCitz. Modern screening strategies in analytical toxicology with special regard to new benzodiazepines. Z. Rechstmed. 100: 19-37 (1988). Manuscript received September 18, 1996; revision received February 24, 1997.
557
