| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Laboratorium voor Toxicologie, Universiteit Gent, Harelbekestraat 72, B-9000 Gent, Belgium
2 Laboratorium voor Fysiologie, Biochemie en Biometrie, Universiteit Gent, Salisburylaan 133, B-9820 Merelbeke, Belgium
aauthor for correspondence: fax 32-9-2648197, e-mail Andre.DeLeenheer{at}rug.ac.be
Screening for benzodiazepines is routinely performed by immunochemical analysis, such as enzyme immunoassay. However, the wide range of therapeutic and toxic concentrations of the different benzodiazepines and their extensive metabolism and conjugation hamper the design of an immunoassay protocol that can detect all benzodiazepines and their corresponding urinary metabolites. Multiple commercially available immunoassay methods exist, but false-negative (FN) results are still encountered, even with some frequently prescribed benzodiazepines.
Several earlier investigations focused on determining how a particular benzodiazepine reacts in the different types of immunoassays, but only a few investigators have tried to improve the detection of some benzodiazepines, either by incorporating an enzymatic hydrolysis step before screening (1)(2) or by adding ß-glucuronidase to the immunoassay reagents (3). These studies were performed with supplemented samples or with samples collected after controlled intake of therapeutic doses of one single benzodiazepine by healthy volunteers. We evaluated the relevance of hydrolyzing urine samples before screening with the Emit® d.a.u.TM Benzodiazepine Assay by analyzing a large number (n = 530) of authentic patient urine samples, reflecting actual practice.
All urine samples were screened before (Emit) and after (Emit-H) enzymatic hydrolysis and subsequently analyzed by a sensitive gas chromatographic–mass spectrometric (GC-MS) method specifically developed for the analysis of benzodiazepines as a reference method (4)(5). The Emit d.a.u. Benzodiazepine Assay reagent sets were from Syva Co. (Dade Behring) and were used on a Cobas Mira Instrument (Roche) according to the instructions of the manufacturer.
The applied hydrolysis, solid-phase extraction, and derivatization procedures were validated for alprazolam, 
-hydroxyalprazolam, 4-hydroxyalprazolam, flunitrazepam, 7-aminoflunitrazepam, desmethylflunitrazepam, flurazepam, hydroxyethylflurazepam, N-desalkylflurazepam, ketazolam, lorazepam, lormetazepam, oxazepam, triazolam, and -hydroxytriazolam. Because oxazepam and nordiazepam are common metabolites of many benzodiazepines, identification of the corresponding parent compounds was based on MS detection of more specific metabolites or the parent compound itself. The recoveries for bromazepam, diazepam, prazepam, and temazepam determined (n = 4) at 0.005 (selected-ion monitoring), 0.1, and 1.0 mg/L were 85–93% (CV <12%), 81–96% (CV <11%), 82–93% (CV <14%), and 86–95% (CV <10%), respectively, indicating that the optimized GC-MS procedure allows reliable detection of these benzodiazepines as well.
Briefly, 2 mL of 0.2 mol/L sodium acetate buffer (pH 4.5) and 5500 U of Helix pomatia ß-glucuronidase were added to 1 mL of urine. The tubes were mixed and incubated at 56 °C for 1 h. After centrifugation at 1121g for 10 min, 100 µL of the supernatant was used for Emit-H analysis. The internal reference (N-methylclonazepam) was added to the remainder of the supernatant, which was brought to pH 6.8 by the addition of sodium hydroxide and phosphate buffer before solid-phase extraction on phenyl-type cartridges. The dried extracts were derivatized by acetylation and reconstituted in 20 µL of ethyl acetate. On-column injection of 1-µL aliquots was performed with a HP 7683 autosampler on a RESTEK hydroguard guard column [5 m x 0.32 mm (i.d.)] coupled to a SGE (Achrom) BP1 capillary column [30 m x 0.25 mm (i.d.); 0.25-µm film thickness] with a Universal angled press-tight connector (RESTEK). The GC-MS instrument consisted of a HP 6890 Series Gas Chromatograph coupled to a HP 5973 mass selective detector, used in the electron impact scan mode. The limit of detection for all compounds ranged from 0.013 to 0.030 mg/L.
Of the 530 samples analyzed, 174 were positive for benzodiazepines by GC-MS analysis. The following compounds were identified (with the corresponding number of samples in parentheses): alprazolam (n = 24), bromazepam (n = 48), diazepam (n = 24), flunitrazepam (n = 2), flurazepam (n = 8), lormetazepam (n = 44), oxazepam (n = 74), prazepam (n = 2), and temazepam (n = 34). Classification of the samples as true positives (TPs) and true negatives (TNs) was based on these GC-MS results.
The specificity and sensitivity of the two Emit screening tests were first evaluated at the cutoff value of 0.2 mg/L oxazepam as specified in the Syva package insert. The 95% confidence intervals (CIs) were calculated based on the binomial distribution, and the specificity and sensitivity of the two screening tests were compared by the Fisher exact test (6). Of the 174 GC-MS-positive samples, 117 and 152 yielded a positive screening result at the cutoff value of 0.2 mg/L oxazepam in the Emit and Emit-H tests, respectively. This implies that the sensitivity increased from 67% (95% CI, 60–74%) for the Emit test to 87% (95% CI, 81–92%) for the Emit-H test, which is a highly significant gain of 20% (P <0.0001). The 57 FN samples generated by Emit contained the following benzodiazepines (measured GC-MS concentrations, 0.02–3.86 mg/L; corresponding number of samples in parentheses): lormetazepam (n = 20), bromazepam (n = 17), oxazepam (n = 9), temazepam (n = 5), alprazolam (n = 1), and flurazepam (n = 1). In four FN samples, two different benzodiazepines were identified. The 22 FN samples generated by Emit-H contained the following benzodiazepines (measured GC-MS concentrations, 0.02–0.51 mg/L; corresponding number of samples in parentheses): lormetazepam (n = 3), bromazepam (n = 9), oxazepam (n = 1), temazepam (n = 5), alprazolam (n = 1), and flurazepam (n = 1). In two FN samples, two different benzodiazepines were identified. Enzymatic pretreatment of the urine samples before Emit screening thus seems to be most advantageous for the detection of lormetazepam, bromazepam, and oxazepam. The increase in Emit absorbance on hydrolysis was dependent on the identity of the ingested drug, and the largest increase was for samples containing flurazepam. Of the 356 GC-MS-negative samples, all were negative in the Emit test, whereas 341 were negative in the Emit-H test. This implies that the specificity decreased from 100% (95% CI, 99–100%) for the Emit test to 96% (95% CI, 93–98%) for the Emit-H test, which is a small but significant loss of 4% (P <0.001).
To study the effect of changing the cutoff value on sensitivity and specificity, the ROC curves for the two screening tests were derived and compared (Fig. 1
) (7)(8). The ROC graph is a plot of all the sensitivity/specificity pairs resulting from continuously varying the cutoff value over the entire range of results observed. The x axis represents (1 - specificity), or the false-positive (FP) fraction, defined as the following ratio: (number of FP results)/(number of TN + number of FP results). The y axis represents the sensitivity, or the TP fraction, defined as the following ratio: (number of TP results)/(number of TP + number of FN results). The closer the plot is to the upper left corner, where the TP fraction is 100% and the FP fraction is 0%, the higher the overall accuracy of the test. By comparing both ROC curves, it is clear that at a fixed specificity, the corresponding sensitivity is systematically higher for the Emit-H test. For the urine samples screened in this study, the recommended cutoff value of 0.2 mg/L for the Emit is optimal with respect to specificity (100%) but corresponds to a sensitivity of only 67%, and it would be advantageous to decrease the cutoff value. In the case of the Emit-H test, the corresponding values are 96% for specificity and 87% for sensitivity, and the gain in sensitivity is therefore only marginal when the cutoff value is decreased.
|
In conclusion, enzymatic pretreatment of urine samples before Emit screening for benzodiazepines yields a highly significant gain in sensitivity of 20% at the cutoff value of 0.2 mg/L oxazepam. From the ROC curves, it is clear that Emit-H outperforms Emit; for a fixed specificity, the corresponding sensitivity is systematically higher for the Emit-H test. Our results are in accordance with those reported by Meatherall (2) and Ropero-Miller et al. (3) but in marked contrast to the results reported by Beck et al. (1). However, by analyzing this large number of authentic patient samples; confirming the results for all samples, not just the presumptive positives, with the optimized GC-MS procedure; and performing a detailed statistical evaluation, we produced more convincing evidence that Emit-H is the recommended screening procedure for urinary benzodiazepines in routine practice.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
