Validation Report for the Neogen Fentanyl Kit for ELISA Screening of Whole Blood and Urine Specimens This document describes the validation of a Neogen Fentanyl kit for the semi-quantitative analysis of whole blood and urine specimens by enzyme linked immunosorbant assay (ELISA). A Dynex DSX Automated ELISA System was used for all validation steps. Blood was diluted 1:5 with EIA buffer before being loaded onto the instrument. Urine was diluted 1:2 with EIA buffer by the instrument. This validation included the evaluation of: Validation Parameter Acceptance Criteria (1) Validation Results Sensitivity/Limit of Detection (LOD) Repeatability/ Precision Specificity/Selectivity Carryover 1. The LOD must be verified by three sources of blank matrix samples fortified at decreasing concentrations analyzed in duplicate for at least three runs. 2. Reproducible instrument response with a signal that is greater than or equal to three times the standard deviation of the signal from the negative samples. 1. Evaluate at least three concentration levels (below, at, and above the decision/cutoff point). 2. At least 3 replicates from 5 separate runs should be evaluated for each level. 3. The coefficient of variation (CV) should be within 20%. 4. The mean plus or minus two standard deviations for each concentration must not overlap. 1. No interference from compounds that are commonly identified in whole blood and/or urine case samples. 2. No interference from a high concentration of other compounds in the control mix the target drug will be added to. 1. A blank matrix sample must be analyte free when run after a standard prepared at or above the highest calibrator concentration. 1. The LOD was verified by three sources of blank matrix samples fortified at decreasing concentrations analyzed in duplicate for three runs. 2. All reproducible instrument response was greater than three times the standard deviation of the signal from the negative samples. 1. Evaluated three concentration levels (below, at, and above decision points). 2. Three replicates from 5 separate runs were evaluated for each level. 3. The coefficient of variation (CV) was less than 6%. 4. The mean plus or minus two standard deviations for each concentration did not overlap. 1. There was no interference from compounds that are commonly identified in whole blood and/or urine case samples. 2. There was no interference from a high concentration of other compounds in the control mix the target drug was added to. 1. The blank matrix sample was analyte free when run after a standard prepared at 500 ng/ml. Page 1 of 12
Drift Ruggedness/Robustness Case Sample Comparison 1. No statistically significant difference between twentyfour intra-run replicate measurements of the cutoff control using a student s t-test to compare each group of eight samples to the cutoff calibrator. 2. The coefficient of variation (CV) should be within 20%. 1. Validation studies will be performed over multiple days. 1. Any case samples that have been previously determined to contain the target analyte must have identical qualitative results. 2. Any case samples that have been previously determined to be analyte free must have identical qualitative results. 1. There was no statistically significant difference between twenty-four intrarun replicate measurements of the cutoff control using a student s t-test when each group of eight samples was compared to the cutoff calibrator. 2. The coefficient of variation (CV) was within 5%. 1. The validation was performed over multiple days and demonstrated repeatable results. 1. Case samples that had been previously determined to contain the target analyte had identical qualitative results. 2. Case samples that have been previously determined to be analyte free had identical qualitative results. Dilution Integrity 1. Not applicable. 1. Not applicable. Stability 1. Not applicable. 1. Not applicable. Validation Steps: Step #1: Method Development 6/12/14 7/3/14 1. Prepare controls. 2. Analyze standards at successively lower levels (50, 10, 5, 2, 1, 0.5, and 0.1 ng/ml and a matrix blank) to evaluate the analytical curve of the assay. [8 tests] 3. Determine suitable levels for the low, mid/cutoff, and high control. If no suitable control levels are found then appropriately modify the protocol and repeat #2. Page 2 of 12
Step #2: Sensitivity & Carryover 6/13/14 7/7/14 1. Analyze standards in duplicate at the blank, low, mid/cutoff, and high control levels. Follow with 8 blank controls to evaluate the theoretical limit of detection (LOD), a control at 10 times the concentration of the high control (or other suitable concentration), and 3 blank controls to evaluate carryover. [20 tests] Step #3: Sensitivity 6/16/14 7/8/14 1. After the theoretical LOD has been determined, analyze a series of 4 sample concentrations in duplicate from the proposed mid/cutoff to the theoretical LOD using three separate matrix sources in three separate runs, to determine the experimental LOD. If necessary, adjust the concentration of the low control to more accurately reflect the experimentally determined sensitivity of the assay. Note: Setting a control concentration at the LOD may not be appropriate to obtain repeatable results. [24 tests] Step #4: Repeatability 6/17/14 7/7/14 1. If necessary, prepare new controls. 2. Analyze 3 replicates at a low concentration level (no more than 50% below the cutoff/decision point) over 5 different runs to evaluate precision of the assay below the cutoff point (CV of 20% is acceptable at this level). [15 tests] 3. Analyze 3 replicates at the cutoff level over five different runs to evaluate precision and drift at the cutoff concentration for the assay (CV of 20% is acceptable at this level). [15 tests] 4. Analyze 3 replicates at a high concentration level (no more than 50% above the cutoff/decision point) over five different runs to evaluate precision above the cutoff of the assay (generally a CV of 20% is acceptable at this level). [15 tests] 5. Additional replicates may be necessary to evaluate repeatability at the low control level, if it will be used as a decision point. Page 3 of 12
Step #5: Drift and Nicholas Tiscione 6/17/14 7/10/14 1. Analyze standards in duplicate at the blank, low, and mid/cutoff control levels, along with 24 replicates (3 strips) at the cutoff control level to evaluate drift at the cutoff concentration for the assay. [30 tests] Step #6: Specificity 7/1/14 7/10/14 1. Evaluate cross-reactivity for similar compound, using a series of concentrations to yield a response that is within the analytical curve (low, mid/cutoff, high) to determine the approximate cross-reactivity. 2. Analyze two replicates of high control for each matrix for other ELISA tests to evaluate crossreactivity. [4 tests] Step #7: Specificity and Case Comparison / Evaluation 7/3/14 7/10/14 1. Analyze at least 10 case samples for each matrix according to the procedure that will be used normally (with blank, low, mid/cutoff, and high). At least 5 of these case samples should be positive. If positive case samples are not available at the time of validation then positive unknowns of varying concentrations may be prepared and analyzed. [~10 tests] 2. Confirm positive and an appropriate number of negative results by GC/MS and/or LC-MSMS. Depending on case sample volume, confirmation of negative results may not be possible and/or appropriate. Page 4 of 12
Results 1. Method Development Urine Two fentanyl working controls with concentrations of 20 ng/ml and 1000 ng/ml were prepared in saline. The 50 ng/ml, 10 ng/ml and 5 ng/ml fentanyl-spiked samples in urine were prepared using the 1000 ng/ml working control and analyzed by ELISA. The 2 ng/ml, 1 ng/ml, 0.5 ng/ml and 0.1 ng/ml fentanyl-spiked samples in urine were prepared using the 20 ng/ml control and analyzed by ELISA. Spiked samples were analyzed with no dilution and 1:2, 1:10, and 1:50 dilutions performed by the instrument. Analytical curves were evaluated for each dilution. The 1:2 and 1:10 dilution curves had the best optical density spread and linearity near the desired decision points of 1 and 5 ng/ml. Suitable control levels were chosen to be 1 ng/ml for the low control, 5 ng/ml for the mid/cutoff control and 50 ng/ml for the high control. Steps 1-5 were performed using a 1:10 dilution but the results did not meet the validation requirements for repeatability around the decision points. The steps were performed again with a 1:2 dilution and the results did meet the validation requirements. The urine validation was performed with a 1:2 dilution performed by the instrument. Conclusion: Control levels were chosen to be 1 ng/ml for the low control, 5 ng/ml for the mid/cutoff control and 50 ng/ml for the high control. The urine samples were run with a 1:2 dilution performed by the instrument. Blood Two fentanyl working controls with concentrations of 20 ng/ml and 1000 ng/ml were prepared in saline. The 50 ng/ml, 10 ng/ml and 5 ng/ml fentanyl-spiked samples in blood were prepared using the 1000 ng/ml working control and analyzed by ELISA. The 2 ng/ml, 1 ng/ml, 0.5 ng/ml and 0.1 ng/ml fentanyl-spiked samples in blood were prepared using the 20 ng/ml control and analyzed by ELISA. Spiked samples were diluted 1:5 with EIA buffer before being loaded onto the instrument. The analytical curve was evaluated and suitable control levels were chosen to be 0.5 ng/ml for the low control, 1 ng/ml for the mid/cutoff control and 50 ng/ml for the high control. Conclusion: Control levels were chosen to be 0.5 ng/ml for the low control, 1 ng/ml for the mid/cutoff control and 50 ng/ml for the high control. The blood samples were diluted 1:5 with EIA buffer before being loaded onto the instrument. 2. Sensitivity Urine The theoretical LOD was calculated to be 0.03 ng/ml by using three times the standard deviation of the eight replicate blank responses from Step #2. Additional fentanyl-spiked samples at 5 ng/ml in three different urine sources were prepared using the 1000 ng/ml working control, and 1, 0.5, and 0.1 ng/ml fentanyl-spiked samples in three different urine sources were prepared using the 20 ng/ml working control and analyzed by ELISA. The experimental LOD was determined to be at least 0.5 ng/ml. The concentrations for the controls were set at 1 ng/ml for Page 5 of 12
the low control (near the experimental LOD), 5 ng/ml for the mid/cutoff control and 50 ng/ml for the high control. Conclusion: The theoretical LOD was calculated to be 0.03 ng/ml. The experimental LOD was determined to be 0.5 ng/ml. The concentrations for the controls were set at 1 ng/ml for the low control (near the experimental LOD), 5 ng/ml for the mid/cutoff control and 50 ng/ml for the high control and were evaluated in subsequent steps. Blood The theoretical LOD was calculated to be 0.18 ng/ml by using three times the standard deviation of the eight replicate blank responses from Step #2. Additional fentanyl-spiked samples at 1, 0.75, 0.5, and 0.25 ng/ml in three different blood sources were prepared using the 20 ng/ml working control and analyzed by ELISA. The experimental LOD was determined to be 0.25 ng/ml. The concentrations for the controls were set at 0.5 ng/ml for the low control (near the experimental LOD), 1 ng/ml for the mid/cutoff control and 50 ng/ml for the high control. Conclusion: The theoretical LOD was calculated to be 0.18 ng/ml. The experimental LOD was determined to be 0.25 ng/ml. The concentrations for the controls were set at 0.5 ng/ml for the low control (near the experimental LOD), 1 ng/ml for the mid/cutoff control and 50 ng/ml for the high control and were evaluated in subsequent steps. 3. Carryover Three matrix matched blank samples were analyzed following a positive control at 500 ng/ml for both blood and urine to check for carryover. The response from the three matrix matched blanks were compared to the response from the blank calibrators. No carryover was observed in the matrix matched blank samples. Conclusion: No carryover was observed in a matrix matched blank sample analyzed following a 500 ng/ml positive control in both blood and urine. 4. Repeatability / Precision Urine To analyze the precision at the mid control decision point of 5 ng/ml, 3 replicates of standards at the mid control level and at ± 50% of the concentration of the mid control level (7.5 ng/ml, 5 ng/ml and 2.5 ng/ml) were analyzed over 5 separate runs. To analyze the precision at the low control decision point of 1 ng/ml, 3 replicates of standards at the low control level and at ± 50% of the concentration of the low control level (1.5 ng/ml, 1 ng/ml and 0.5 ng/ml) were analyzed over 5 separate runs. The repeatability was determined by calculating the mean, standard deviation, and coefficient of variation (CV) for each set of inter-run concentration replicate measurements. The mean plus or minus two standard deviations for each concentration did not overlap. The CV was less than 6% for concentrations at, above, and below both decision points. Page 6 of 12
Table 1: Repeatability / Precision at 5 ng/ml Concentration 7.5 ng/ml 5 ng/ml 2.5 ng/ml Mean Optical Density 0.371 0.488 0.719 Standard Deviation (S.D.) 0.008 0.027 0.024 CV 2.092 5.603 3.356 Mean + 2 S.D. 0.386 0.542 0.768 Mean - 2 S.D. 0.355 0.433 0.671 Table 2: Repeatability / Precision at 1 ng/ml Concentration 1.5 ng/ml 1 ng/ml 0.5 ng/ml Mean Optical Density 0.984 1.148 1.451 Standard Deviation (S.D.) 0.020 0.027 0.042 CV 2.062 2.348 2.909 Mean + 2 S.D. 1.024 1.201 1.535 Mean - 2 S.D. 0.943 1.094 1.366 Conclusion: The observed repeatability / precision is suitable for the semi-quantitative identification of fentanyl by ELISA in urine at 1 and 5 ng/ml. Blood To analyze the precision at the mid control decision point of 1 ng/ml, 3 replicates of standards at the mid control level and at ± 50% of the concentration of the mid control level (1.5 ng/ml, 1 ng/ml and 0.5 ng/ml) were analyzed over 5 separate runs. For the low control decision point of 0.5 ng/ml, 3 replicates of standards at the low control level and at ± 50% of the concentration of the low control level (0.75 ng/ml, 0.5 ng/ml and 0.25 ng/ml) were analyzed over 5 separate runs. The repeatability was determined by calculating the mean, standard deviation, and coefficient of variation (CV) for each set of inter-run concentration replicate measurements. The mean plus or minus two standard deviations for each concentration did not overlap. The CV was less than 3% for concentrations at, above, and below both decision points. Table 3: Repeatability / Precision at 1 ng/ml Concentration 1.5 ng/ml 1 ng/ml 0.5 ng/ml Mean Optical Density 1.196 1.379 1.663 Standard Deviation (S.D.) 0.029 0.029 0.036 CV 2.425 2.103 2.165 Mean + 2 S.D. 1.254 1.437 1.735 Mean - 2 S.D. 1.138 1.321 1.591 Page 7 of 12
Table 4: Repeatability / Precision at 0.5 ng/ml Concentration 0.75 ng/ml 0.5 ng/ml 0.25 ng/ml Mean Optical Density 1.528 1.663 1.848 Standard Deviation (S.D.) 0.026 0.036 0.029 CV 1.702 2.165 1.569 Mean + 2 S.D. 1.580 1.735 1.906 Mean - 2 S.D. 1.476 1.591 1.790 5. Drift Conclusion: The observed repeatability / precision is suitable for the semi-quantitative identification of fentanyl by ELISA in blood at 0.5 and 1 ng/ml. Urine Twenty-four intra-run replicate measurements of fentanyl at 5 ng/ml (mid/cutoff control) were analyzed by ELISA to determine if drift was observed across a plate. A student s t-test was done for each group of eight samples compared to the mid/cutoff calibrator. If the calculated t value is less than the tabulated value at the stated confidence level (CL), then there is no significant difference in the replicates and the calibrator at the stated confidence interval. No significant difference was observed at the 95% confidence level. No significant drift was observed. Table 5: Drift Sample ID Optical Density Sample ID Optical Density Sample ID Optical Density Mid 1 0.493 Mid 9 0.457 Mid 17 0.483 Mid 2 0.463 Mid 10 0.472 Mid 18 0.478 Mid 3 0.454 Mid 11 0.476 Mid 19 0.457 Mid 4 0.469 Mid 12 0.463 Mid 20 0.48 Mid 5 0.469 Mid 13 0.441 Mid 21 0.453 Mid 6 0.483 Mid 14 0.464 Mid 22 0.46 Mid 7 0.479 Mid 15 0.474 Mid 23 0.465 Mid 8 0.474 Mid 16 0.482 Mid 24 0.473 Table 6: Statistical Analysis t (Mid 1-8) 95% CL t (Mid 9-16) 95% CL t (Mid 17-24) 95% CL 2.099 2.365 0.464 2.365 1.147 2.365 Conclusion: No statistically significant drift was observed in the analysis of twenty-four intrarun replicate measurements of fentanyl at 5 ng/ml (mid/cutoff control) after a student s t-test was done for each group of eight samples compared to the mid/cutoff calibrator. Page 8 of 12
Blood Twenty-four intra-run replicate measurements of fentanyl at 1 ng/ml (mid/cutoff control) were analyzed by ELISA to determine if drift is observed across a plate. A student s t-test was done for each group of eight samples compared to the mid/cutoff calibrator. If the calculated t value is less than the tabulated value at the stated confidence level (CL), then there is no significant difference in the replicates and the calibrator at the stated confidence level. No significant difference was observed at the 95% confidence level. No significant drift was observed. Table 7: Drift Sample ID Optical Density Sample ID Optical Density Sample ID Optical Density Mid 1 1.478 Mid 9 1.402 Mid 17 1.454 Mid 2 1.457 Mid 10 1.420 Mid 18 1.431 Mid 3 1.429 Mid 11 1.773 Mid 19 1.461 Mid 4 1.451 Mid 12 1.433 Mid 20 1.438 Mid 5 1.408 Mid 13 1.449 Mid 21 1.447 Mid 6 1.417 Mid 14 1.462 Mid 22 1.446 Mid 7 1.469 Mid 15 1.441 Mid 23 1.424 Mid 8 1.475 Mid 16 1.444 Mid 24 1.432 Table 8: Statistical Analysis t (Mid 1-8) 95% CL t (Mid 9-16) 95% CL t (Mid 17-24) 95% CL 0.630 2.365 0.844 2.365 0.084 2.365 Conclusion: No statistically significant drift was observed in the analysis of twenty-four intrarun replicate measurements of fentanyl at 1 ng/ml (mid/cutoff control) after a student s t-test was done for each group of eight samples compared to the mid/cutoff calibrator. 6. Specificity Two related compounds, norfentanyl and acetyl fentanyl, were prepared in a series of concentrations and analyzed by ELISA. Norfentanyl was prepared at 1000, 500, and 250 ng/ml in urine and blood. Acetyl fentanyl was prepared at 25, 10, and 5 ng/ml in urine and blood. The percent cross reactivity was calculated and the results are presented below. Table 9: Specificity Drug Concentration (ng/ml) Urine Observed % Cross Reactivity Blood Observed % Cross Reactivity Stated % Cross Reactivity (by Neogen) Acetyl Fentanyl 25 59% 35% 42% (at 0.78 ng/ml) Acetyl Fentanyl 10 51% 29% 42% (at 0.78 ng/ml) Acetyl Fentanyl 5 56% 29% 42% (at 0.78 ng/ml) Norfentanyl 1000 0.05% 0.04% 0.01% (at 4870 ng/ml) Norfentanyl 500 0.10% 0.08% 0.01% (at 4870 ng/ml) Norfentanyl 250 0.20% 0.17% 0.01% (at 4870 ng/ml) Page 9 of 12
Two replicates of the high controls (multi and ABO) for each matrix for other ELISA tests were analyzed to evaluate cross-reactivity. Cross-reactivity was not observed with either control. The fentanyl working standards used for preparing the low, mid and high controls will be combined with the multi controls for blood and urine. To ensure that fentanyl does not interfere with other ELISA kits used for urine and blood screening, the cross reactivity of fentanyl was evaluated with the assays. Fentanyl standards were prepared at 50 ng/ml in blood and urine and analyzed using the ELISA kits. ELISA assays evaluated for urine specimens included buprenorphine, oxycodone/oxymorphone, barbiturates, benzodiazepines, carisoprodol, cocaine/benzoylecgonine, opiates and cannabinoids. ELISA assays evaluated for blood specimens included amphetamines, buprenorphine, oxycodone/oxymorphone, barbiturates, benzodiazepines, carisoprodol, cocaine/benzoylecgonine, methamphetamines, opiates, and cannabinoids. No cross reactivity of the other assays to fentanyl was observed. Five fentanyl-negative urine and blood case samples along with matrix matched negative and positive (low, mid/cutoff and high) controls were analyzed to evaluate specificity. The urine and blood cases contained other drugs listed in Table 10. No matrix interferences were observed in the case samples and the UTAK blank urine and blood negative for fentanyl. Three urine and four blood unknowns were prepared with known amounts of fentanyl and analyzed by ELISA. All spiked samples in blood and urine detected the presence of fentanyl. Table 10: Specificity Drug Number of Cases Positive for the Drug Alprazolam 6 α-oh-alprazolam 4 THCA 2 Diphenhydramine 1 Tramadol 2 6-MAM 1 Morphine 5 Hydromorphone 3 Codeine 1 Buprenorphine 3 Cocaine 2 Cocaethylene 1 Benzoylecgonine 2 Diazepam 2 Nordiazepam 2 Temazepam 2 Oxazepam 2 Hydrocodone 3 Oxycodone 1 Oxymorphone 1 Page 10 of 12
Carisoprodol 2 Meprobamate 2 Mirtazapine 1 Lorazepam 2 Clonazepam 1 Midazolam 1 Conclusion: Cross reactivity to acetyl fentanyl was observed. No cross reactivity was observed for the compounds in the high control mixes by the fentanyl assay or for fentanyl by the other ELISA assays. Fentanyl will be combined in the working standards used to prepare the low, mid and high multi controls. No matrix interferences were observed in five urine and five whole blood fentanyl- negative case samples that contained various other drugs and metabolites commonly identified in case samples. All positive results will be confirmed by GC/MS or LC-MSMS. 7. Case Comparison / Evaluation Seven urine and six blood case samples and three urine and four blood unknowns along with matrix matched negative and positive (low, mid/cutoff and high) controls were analyzed. The results are presented below. Table 11: Urine Case Comparison / Evaluation Case / Unknown ELISA Fentanyl Intended Result Comments 1 POS POS 50 ng/ml 2 POS POS 25 ng/ml 3 POS POS 10 ng/ml 14-010830 POS POS Norfentanyl confirmed in specimen. Low fentanyl with interference from quetiapine metabolite not confirmed. 14-083792 POS POS Fentanyl and norfentanyl confirmed in specimen. 14-077947 NEG NEG 14-069063 NEG NEG 14-080703 NEG NEG 14-011069 NEG NEG 14-011072 NEG NEG Page 11 of 12
Table 12: Blood Case Comparison / Evaluation Case / Unknown ELISA Fentanyl Intended Result Comments 1 POS POS 2 ng/ml 2 POS POS 5 ng/ml 3 POS POS 10 ng/ml 4 POS POS 25 ng/ml 14-010848 POS POS or DET Fentanyl < 1 ng/ml (~ 0.9 ng/ml) 14-010902 NEG NEG 14-010909 NEG NEG 14-071076 NEG NEG 14-090122 NEG NEG 14-090742 NEG NEG Conclusion: In five urine and five blood fentanyl-positive samples the presence of fentanyl was appropriately detected. In five urine and five blood fentanyl-negative samples fentanyl was not detected. All positive case samples were confirmed by GC/MS or LC-MSMS. 8. Ruggedness / Robustness This analysis was performed on multiple days and gave repeatable results. Approved by: Date: 8/27/14 Chemistry/Toxicology Manager Page 12 of 12