Performance of a Microtiter Plate ELISA for Screening of Postmortem Blood for Cocaine and Metabolites

Transcription

1 Technical Note I Performance of a Microtiter Plate ELISA for Screening of Postmortem Blood for Cocaine and Metabolites Vina Spiehlerl, *, Daniel S. Isenschmid 2, Parrish Matthews 3, Philip Kemp 4, and Tom Kupiec 3 1Spiehler & Associates, Newport Beach, California; 20ffice of the Wayne County Medical Examiner, Detroit, Michigan; 3Analytical Research Laboratories, Inc., Oklahoma City, Oklahoma; and 40ffice of the Medical Examiner, Oklahoma City, Oklahoma Abstract I The object of this study was to evaluate the suitability of the Neogen Corporation microtiter plate enzyme-linked imrnunoassay (ELISA) for cocaine and melabolites for screening of postmortem blood. Sixly-five poslmortem whole blood specimens were obtained from drug-involved deaths that had been screened and confirmed positive for cocaine and/or benzoylecgonine (BE). Fifty-eight negative specimens were obtained from noncocaine-involved deaths. Specimens were tested using the Neogen TM Cocaine/BE microtiter plate ELISA assay. No matrix effects were found for whole blood in this assay. The effect of dilutions of the whole blood specimens of 1:5 through 1:50 was studied. A dilution of 1:5 was chosen to correspond to that used for other Neogen microtiter plate assays for drugs in whole blood. True positives, true negatives, false positives, and false negatives were determined and graphed for the ELISA results against gas chromatography-mass spectrometry (GC-MS), GC-nitrogen-phosphorus detection, and case histories. From these graphs and the receiver operating characteristic curves, the optimal cutoff for the Neogen Cocaine/BE ELISA was found to be 5 ng/ml BE equivalents at a 1:5 dilution. The optimum cutoff for a 1:50 dilution was 50 ng/ml BE equivalents. The Neogen Cocaine/BE ELISA had a sensitivity of 3.8% 2.% and a specificity of 6.6% 2.4% versus GC-MS at a cutoff of 5 ng/ml BE equivalents (1:5 dilution) and a sensitivity of 100% 0.5% and specificity of 8.3% 1.7% versus GC-MS at a 50 ng/ml BE equivalents cutoff (1:50 dilution). Introduction Cocaine is one of the most frequent analytes detected in medical examiner/coroner offices, particularly in major metropolitan areas. The forensic toxicology laboratory of the Office of the Wayne County Medical Examiner (WCMEO) in Detroit, Michigan handles approximately 3200 cases per year, of which approximately 2700 cases a year are tested for cocaine. Between 185 and 187, a fivefold increase in cocaine-related deaths occurred at the WCMEO (1). From 12 to 2000, cocaine or co- "Author to whom correspondence should be addressed: Vina Spiehler, 422 Tustin Ave., Newport Beach, CA Spiehleraa@aol.com. caine metabolites were the most common positive finding, excluding ethyl alcohol, and were detected in 10-15% of the cases tested. Toward the end of the 180s, the incidence of cocaine deaths began to plateau and during the 10s remained fairly constant. Only a 12% increase in the incidence of cocaine in medical examiner cases occurred between 16 and 2000, compared with a 143% increase for heroin (2). The forensic toxicology laboratory of the Office of the Chief Medical Examiner in Oklahoma (OCMEO) handles approximately 3400 cases per year. In 2001, 46 (4.3%) of the 1068 cases tested by immunoassay were positive for cocaine and/or benzoylecgonine (BE). The presence of cocaine in blood suggests recent cocaine exposure, most likely within the last 4--6 h (3). Cocaine metabolite (BE) in blood suggests cocaine use within the last 24 h (3). Although the determination of BE in blood is important to lend context, it tells more about past cocaine use than acute use. Ethyl cocaine, or cocaethylene, formed during simultaneous cocaine and ethanol use, is indicative of both recent cocaine and ethanol use. Ethyl cocaine is psychoactive and may be at least as toxic as cocaine. It is important to appreciate that the cocaine concentrations in postmortem cases bear little correlation to toxicity. That is, low concentrations of cocaine may precipitate a fatal event in a sensitized individual, whereas high concentrations may not play a significant role in the death. Witnessed behavior, cardiovascular pathology, and patient history are important factors when interpreting the role of cocaine in a medical examiner case. Traditionally, immunoassays have been used to screen for cocaine metabolite in urine. However, the presence of cocaine and its metabolites in urine do not provide data that is pharmacologically significant. Thus, it is important for a medical examiner's office to have a method to screen for cocaine and its metabolites in blood. Traditional radiommunoassays have been used for this purpose because blood can be assayed directly without any sample preparation. More recently, enzyme-linked immunoassays (ELISA) have been able to serve in this capacity without the need for any radioactive waste disposal issues. In addition, ELISA is readily amenable to automation. The object of this study was to evaluate the suitability of the Neogen Corporation microtiter plate ELISA for cocaine and metabolites for screening of postmortem blood. The results for specimens Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 587

2 tested using the microtiter plate ELISA assay were compared to the drugs identified and/or quantitated by gas chromatography-mass spectrometry (GC-MS). The optimum cutoff of the ELISA assay was determined by receiver operating characteristic (ROC) analysis and the sensitivity, specificity, and predictive value of the assay calculated. Methods Specimen selection Sixty-five postmortem specimens were selected, which were screened positive by the laboratory's current immunoassay. For specimens submitted by the WCMEO, screening was performed using a BE-specific radioimmunoassay with a cutoff of 100 ng/ml (Immunalysis, Pomona, CA). For specimens submitted by the OCMEO, screening was performed using microtiter plate enzyme immunoassay, also using a cutoff of 100 ng/ml (Orasure, Table I. GC-MS Results for Positive Cocaine Whole Blood Specimens Bethlehem, PA). All positive screens were confirmed by GC-MS using selected ion monitoring. Quantitative results are summarized in Table I. The WCMEO utilized a liquid-liquid extraction procedure followed by derivatization of BE to a t-butylsilyl derivative with N-methyl-N-(tertbutyldimethylsilyl)trifluoroacetamide with 1% tertbutyldimethylchlorosilane (Standard Operating Procedure, Wayne County Medical Examiner's Office, Detroit, MI, 2001). The limit of quantitation (LOQ) for the assay was 10 ng/ml for cocaine, 25 ng/ml for BE, and 10 ng/ml for cocaethylene. The OCMEO also utilized a liquid-liquid extraction procedure followed by derivatization of BE with N,N-dimethylformamide and N,N-dimethylformamide dipropylacetal to n-propylcocaine as described by Isenschmid et al. (4) and Spiehler et al. (5). The LOQ was 25 ng/ml for cocaine and BE and 12.5 ng/ml for cocaethylene. An additional 58 specimens were selected that were negative for cocaine, cocaethylene, and BE by immunoassay and by a gas-liquid chromatography-nitrogen-phosphorus detection basic screen, some of which had been found positive for other drugs. Cocaine Cocaine Case ID (mg/l) BE (mg/t) CE (rag/l) Source Case ID (rag/l) BE (rag/l) CE (mg/t) Source OK WC OK ,071 WC OK WC OK WC OK WC OK WC OK WC OK WC OK WC O OK WC OK WC OK WC Pos OK WC Pos OK WC OK WC WC WC WC WC ,2 WC WC < < WC , WC WC WC WC WC WC WC WC WC < WC ,43 WC WC WC WC WC WC WC WC WC WC ,02 1 WC WC WC WC <0, <0.025 WC WC WC WC * Abbreviations: WC = Wayne County, MI; OK = Oklahoma City, OK; and CE = cocaethylene. 588

3 Specimen dilution To determine the effect of dilution, whole blood standards were run diluted 1:5, 1:8, 1:11, 1:15, 1:20, 1:25, and 1:50 with ELISA buffer from Neogen (phosphate buffered saline with bovine serum albumin). Both 1:5 and 1:50 dilutions of the postmortem specimens were used for specimen analysis: 50 IJL whole blood to 200 ]JL ELISA buffer and 50 pl whole blood to 2.45 ml ELISA buffer. The BE equivalents were estimated from the calibration curve using the ratio of the mean absorbance of the specimen to the mean absorbance of the zero standard. Whole blood standards were prepared by fortifying an appropriate amount of BE in solution into a certified negative whole blood pool. The standard concentrations were 0, 5, 10, 25, 50, 75, and 100 ng/ml BE. Standard curves were assayed at dilutions of 1:5 and 1:50 with ELISA buffer to match the dilution of the postmortem specimens in that run. Neogen microtiter plate assays The microtiter plate-based Neogen Cocaine/BE ELISA (Neogen Corporation, Lexington, KY) was evaluated. In addition to the fortified whole blood calibration standards and specimens, the manufacturer's ELISA standard and negative and positive urine-based controls were run on each plate. Twenty microliters of neat and diluted samples were pipetted manually into the microtiter plate wells in duplicate. The ELISAs were then run according to the manufacturer's instructions. Diluted drug-enzyme conjugate (180 pl) was added to the microtiter plate wells and the mixture incubated at room temperature for 45 rain. After incubation, the plate was washed five times with wash buffer (300 pl phosphate buffer with 1~veen 20) to remove any unbound sample or drug-enzyme conjugate employing an ELx 50 Auto Strip Washer (BioTek Instruments, Highland Park, Winosky, VT). K-blue substrate [150 IJL tetramethylbenzidine (TMB) plus hydrogen peroxide] was added, and after a 30-rain substrate incubation, the reaction was halted with the addition of 50 ljl of Neogen's Red Stop solution (a nonacid peroxidase stop solution). The test was read using an ELx 800 Universal Microplate Reader equipped with a 650-nm filter (BioTek Instruments). Calibration curves were plotted as log concentration versus the logit of the ratio of the mean absorbance at each concentration divided by the mean absorbance of the zero standard (B/Bo). Assay precision Intra-assay precision was determined from 40 replicate analysis of the 1:50 dilution of the 50-ng/mL BE standard and 16 replicate analyses of the 1:5 dilution of the 5-ng/mL standard. Interassay precision was calculated from the absorbance ratios for the standards and controls run at the beginning and end of each microtiter plate experiment. The targets for the urine controls were 0 and 2 ng/ml BE. Probability analysis The number of true positives, false negatives, false positives, and true negatives was determined for nine possible cocaine cutoff concentrations ranging from I to 200 ng/ml BE equivalents by comparison to GC-MS (for cocaine- and metabolite-positive specimens), GC, and case histories (negative specimens). Sensitivity, specificity, and predictive value Sensitivity, the true-positive rate, was calculated from the tally of true positives and false negatives determined as: Sensitivity = TP/(TP + FN). Specificity, the true-negative rate, was calculated as Specificity = TN/(TN + FP). Because sensitivity and specificity are probabilities, the standard error (SEp) is equal to SEp = square rootp(1 -p)/n. ROC curves were obtained by plotting the sensitivity at each putative cutoff concentration versus 1 - Specificity at that cutoff value (6). The positive predictive value was calculated as fp/[fp + (1 - f)(1 - q)], where fis the prevalence in the population to be tested, p is the sensitivity, and q is the specificity. Results Figure I shows the calibration curves for spiked whole blood diluted 1:5, 1:8, 1:11, 1:20, 1:25, and 1:50 for the Neogen Cocaine ELISA. Calibration curves appeared linear and parallel within the precision of the assay for all the whole blood dilutions tested (Figure 1). From plate to plate, the average error in the backfit estimation of the 5-ng/mL standard (1:5 dilution) was 31.5% (interassay), and the average error in the backfit estimation of the 50-ng/mL standard (1:50 dilution) was 58.4%. The percent error in backfit of the calibration values from the regression lines ranged from 4.6% at the 5-ng/mL standard to 86% at the 100-ng/mL standard (1:50 dilution) and from 5% at the 100-ng/mL standard to 80% at the 50-ng/mL standard (1:5 dilution). Because the 50% displacement (IC50) of the plates is so low, dilutions of 1:50 blood to assay buffer were needed to bring the expected whole blood concentrations in fatal cases within the range of the IC50 of the assay. Precision The intra-assay precision of the Neogen Cocaine/BE ELISA for 40 replicates of spiked whole blood at 50 ng/ml BE (1:50 dilution) calculated from the absorbance was mean OD SD for a %CV of The intra-assay precision of the B/Bo ratio at 50 ng/ml was mean (5.22 %CV) and for estimated concentration, the mean was 53.6 ng/ml 3.2 ng/ml (6.1%CV). For the 1:50 dilution, the interassay precision of the ratio of the absorbance of the 50 ng/ml calibrator to that of the zero whole blood calibrator (B/Bo) was mean SD or a % CV of 11.8 (n = 6). The interassay coefficients of variation calculated from the B/Bo ratios of the whole blood " "J X 10 : - x x 1:5 dil 1:8 dil 1:11 dil x 1:15 dil x 1:20 dil 1:25 dil 1:50 dil Log of ConcemnlUon (ng/mlj Figure 1. Effect of dilution: calibration curves for spiked whole blood diluted 1:5, 1:8, 1:11, 1:20, 1:25, and 1:50. 58

4 calibrators run in each assay ranged from 6.4% to 21.2%. The intra-assay precision of the Neogen Cocaine/BE ELISA for 16 replicates of spiked whole blood at 5 ng/ml BE (1:5 dilution) calculated from the absorbance was mean OD SD for a %CV of For the 1:5 dilution, the interassay precision of the ratio of the absorbance of the 5-ng/mL calibrator to that of the zero whole blood calibrator (B/Bo) was mean SD or a %CV of 17 (n = 5). The interassay precision calculated from the B/Bo ratios of the urine controls run in each assay was (2.4 %CV) for the negative control and (14.6 %CV) for the positive control. The precision of the estimated concentration of the positive control was mean ng/ml 16 ng/ml, range ng/ml, n =, and 15.6 %CV. The positive control was always positive, and the negative urine control was always negative. Case results The GC-MS values for cocaine and metabolites in the cocaine-related medical examiner cases are shown in Table I. By comparing the ELISA result to the GC-MS results as refer- I m Cutoff (ng/ml) BE equivalents Figure 2. True positives (TP), false negatives (FN), false positives (FP), and true negatives (TN) for cocaine and metabolites for cutoffs ng/ml BE equivalents: 1:50 dilution of whole blood specimens ~ ~.~ 100 i" o '~ ml '~ m m m mm m: Cutoff (ng/mk) BE equivalents Figure 3. True positives (TP), false negatives (FN), false positives (FP), and true negatives (TN) for cocaine and metabolites for cutoffs ng/ml BE equivalents: 1:5 dilution of whole blood specimens $ - ence, the true-positive, false-negative, false-positive, and truenegative values at putative cutoffs of ng/ml BE equivalents were calculated. These are shown in Figure 2 for the 1:50 dilution and in Figure 3 for the 1:5 dilution. From the true-positive and false-negative rates, the sensitivity of each assay was determined at each cutoff concentration. From the true-negative and the false-positive rates, the specificity was calculated at each cutoff concentration. The sensitivity was plotted versus 1 - specificity for each cutoff concentration value to give an ROC curve. The ROC curves for the Neogen Cocaine/BE ELISA on postmortem whole blood at different whole blood dilutions are shown in Figure 4. From the curve in Figure 4 or the bar graphs in Figures 2 and 3, it can be seen that the optimum lution and 5 ng/ml BE at the 1:5 dilution. This is because of the cutoff concentration is 50 ng/ml BE equivalents at the 1:50 di- extremely high sensitivity of the coated microtiter plates for cocaine and the very high specificity of the assay. The IC50 for the drug in buffer are reported by the manufacturer to be 0.3 ng/ml for BE, 0.4 ng/ml for cocaine, and 0.0 ng/ml for cocaethylene (7). The cross-reactivities of cocaine and metabolites relative to Table II. Cross-Reactivity of the Neogen Cocaine/BE ELISA Compound Percent Cross-Reactivity Cocaethylene 144% B E 100% Cocaine 75% p-hydroxycocaine 37.5% AIIopseudococaine 1% N-Norcocaine 1% Ecgonine methyl ester 0.11% Ecgonine 0.66% Procaine < 0.01% Lidocaine < 0.01% Atropine < 0.01% Cotinine < 0.01% G-Procaine < 0.01% Procainamide < 0.01% Table III. Sensitivity and Specificity of the Neogen Cocaine/BE ELISA 1:5 Dilution, 5-ng/mL cutoff Result by GC-MS + Sensitivity = 61/65 = 3.8% 2.% Specificity = 56/58 = 6.6% 2.4% + ELISA Results Specifclty 0.6 -~ 1:50 Dilution 1:5 Dilution Figure 4. ROC curves for Neogen Cocaine/BE ELISA for whole blood at 1:5 and 1:50 dilutions. 50 1:50 Dilution, 50-ng/mL cutoff Result by GC-MS + Sensitivity = 65/65 = 100% + 0.5% Specificity = 57/58 = % ELISA Results

5 BE for the assay are shown in Table II. At a cutoff of 50 ng/rnl and whole blood dilution of 1:50, the Neogen Cocaine/BE ELISA has a sensitivity of 100% and specificity of 8.3%. At a cutoff of 5 ng/rnl (1:5 dilution), the sensitivity was 3.8%, and the specificity was 6.6%. The calculation of these values with their standard errors is shown in Table III. The four specimens that were positive by GC-MS, but negative by ELISA at 5 ng/ml (1:5 dilution), were cases 215, 302, 326, and 334. Although using a dilution of 1:5 causes some loss of accuracy (lower area under the ROC curve) for the cocaine ELISA, the convenience of using one dilution for all drug screening by microtiter plates outweighs this loss. Discussion The assays identified the presence of cocaine and metabolites in whole blood specimens. The specimens from decomposed bodies showed non-zero results, but none were above the 5-ng/mL cutoff at a dilution of 1:5. False positives were encountered in two cases of polydrug use, including illegal stimulants. These persons may have also used cocaine, although it was not confirmed by the laboratory. In these cases, the ELISA may be more sensitive than the GC-MS using a 50-ng/mL cutoff value for cocaine or BE. The precision was similar to other immunoassays (8-11). The intra-assay and interassay precision was good for absolute absorbance and the BAIo ratio (intra-assay %CV less than 10% and interassay %CV less than 20%). The precision and accuracy, as measured by the percent error of backfit, was best at the IC50 midpoint, where the B/Bo ratio was 50%, and decreased with both increasing and decreasing concentrations as the values became further from the IC50. This is because of the S-shaped dose-response curve of all competitive binding assays, of which imrnunoassays are a subclass. For this reason, a limit of detection is not as useful a measure of accuracy in imrnunoassay as it is in chromatographic or colorimetric methods. Also, the high percent error in the backfit of the calibration regression line for the individual calibration standards other than the one that was nearest to the IC50 indicates that these ELISAs would not be useful in quantitation of cocaine and cocaine metabolite levels except as an approximate guide to dilutions for confirmation analyses. These imrnunoassays are useful in screening specimens to eliminate negative specimens and determine which specimens shouid be confirmed for the presence of cocaine and/or cocaine metabolites. The best measure of the screening performance of an imrnunoassay is the true-positive rate (sensitivity) and true-negative rate (specificity) (8,10). The cross-reactivities for Neogen Cocaine ELISA shown in Table II were confirmed by the results of the study. The ELISA response was greater when cocaine and cocaethylene were present. ~o of the four false negatives at the 5-ng/mL cutoff (1:5 dilution) contained only BE. The positive predictive value of a positive cocaine result with the ELISA, using 1:5 dilution with a 5-ng/mL BE cutoff, or the probability that cocaine was involved in the death with a positive result in the Neogen Cocaine/BE ELISA for a prevalence of 50% was 6.5% and for a prevalence of 15% (the prevalence of cocaine-involved deaths in Wayne County) the positive predictive value was 83.2%. The negative predictive value, or probability that cocaine was not involved in the death when the assay was negative, at 50% prevalence was 4.1%, and the negative predictive value at the prevalence of 15% cocaine (or 85% not cocaine) was 8.%. Conclusions In conclusion, for forensic purposes, a positive result with this imrnunoassay should be confirmed on an independent aliquot by a method based on different chemical or physical properties of cocaine and its rnetabolites. A negative result in these populations need not be confirmed. References 1. L Hood, D. Ryan, J. Monforte, and J. Valentour. Cocaine in Wayne County Medical Examiner's cases. J. Forensic ScL 35: (10). 2. D.S. Isenschmid, B.R. Hepler, and S. Kanluen. Patterns in drugs of abuse deaths in metropolitan Detroit. Presented at the American Academy of Forensic Sciences, Seattle, WA, 2001 (updated data). 3. D.S. Isenschmid, M.W. Fischman, R.W. Foltin, and Y.H. Caplan. Concentration of cocaine and metabolites in plasma of humans following intravenous administration and smoking of cocaine. J. Anal ToxicoL 16: (12). 4. D.S. Isenschmid, B.S. Levine, and Y.H. Caplan. A method for the simultaneous determination of cocaine, benzoylecgonine, and ecgonine methyl ester in blood and urine using GC/EIMS with derivatization to produce high mass molecular ions. J. Anal. Toxicol. 12: (188). 5 V.R. Spiehler and D. Reed. Brain concentrations of cocaine and benzoylecgonine in fatal cases. J. Forensic ScL 30(4): (185). 6. M.H. Zweig and G. Campbell. Receiver-operating characteristic (ROC) plots; a fundamental evaluation tool in clinical medicine. Clin. Chem. 3: (13). 7. Neogen Corporation. CocaineJBenzoylecgonine Enhanced ELISA Kit. Package insert, Neogen Corporation, Lexington, KY, S.D. Ferrara, L. Tedeschi, G. Frison, G. Brusini, R Castagna, B. Bernardelli, and D. Soregaroli. Drugs of abuse testing in urine: statistical approach and experimental comparison of immunochemical and chromatographic techniques. J. Anal ToxicoL 18: (14).. B.J. Perrigo and B.R Joynt. Use of ELlSA for the detection of common drugs of abuse in forensic whole blood samples. Can. Soc. Forensic Sci. J. 28(4): (15). 10. V. Spiehler, RR. Sedgwick, I. Collison, S. Perez, S.D. Le, and D.A. Farnin. Validation of an automated enzyme immunoassay for drug screening of postmortem blood. J. Anal. Toxicol. 22: (18). 11. S. Kerrigan and W.H. Phillips, Jr. Comparison of ELISA's for opiates, methamphetamine, cocaine metabolite, benzodiazepines, phencyclidine and cannabinoids in whole blood and urine. Clin. Chem. 47(3)" (2001). Manuscript accepted August 1, 2002 revision accepted February 4,