Development and analytical performance of an automated screening method for cannabinoids on the Dimension clinical chemistry system

A fully automated, random access method for the determination of cannabinoids (UTHC) was developed for the Dimension AR and XL clinical chemistry systems. The method utilizes Abuscreen ONLINE reagents and a multianalyte liquid calibrator containing 11-nor-Δ9-THC-9-carboxylic acid. Within-run and total reproducibility, determined using NCCLS protocol EP5- T2, was less than 0.6% and 1.6% CV, respectively, at all concentrations. Calibration stability was retained for at least 30 days. An extensive evaluation of non-structurally related drugs and various physiological substances indicated lack of interference in the method. No sample carry-over was observed following a specimen containing 1886 ng/ml 11-nor-Δ9-THC-9-carboxylic acid. A 99.1% agreement (N = 445 samples) was found between an EMIT based method on the aca discrete clinical analyser and the Dimension UTHC method. Dimension clinical chemistry system and aca discrete clinical analyser are registered trademarks of Dade International.


Introduction
Laboratory support has been recognized for many years to be an integral part of the hospital service for patients with drug abuse problems. There continues to be laboratory demand for high-quality screening immunoassays for all of the abused drugs including cannabinoids. To provide workstation consolidation and rapid turnaround-times, assays for the detection of cannabinoids need to be available on a high throughput analyser as part of an overall drug of abuse panel. This paper describes the development and performance of a cannabinoid method (UTHC) for the Dimension (R) AR and XL clinical chemistry systems. The method provides a high quality, rapid screening approach for measuring cannabinoids on these general chemistry analysers. Stable liquid reagents, calibrators and controls are utilized to avoid reagent preparation steps. Since the Dimension @ clinical chemistry system is used for the measurement of many other analytes [1], the benefits of workstation consolidation are realized.

Reagents
The Dimension @ UTHC method utilizes Abuscreen ONLINE reagents (ONLINE is a registered trademark of Roche Diagnostic Systems, Inc Assay principle Each test is initiated by the addition of 90 gl of diluent and l15gl of monoclonal antibody to the reaction cuvette. After a 60s incubation period, 26 gl of sample and 59 gl of water are added to the reaction. After 78 s, 60 gl of microparticle reagent, with cannabinoid derivative bound to the particle surface, are added. Cannabinoids, if present in the sample, compete with the microparticle reagent for the antibody, thereby decreasing the rate of aggregation. Thus, the rate of aggregation is inversely proportional to the concentration of cannabinoids present in the sample. The rate of particle aggregation, as measured by absorbance changes at 510 and 700 nm, is determined over the next 143 s. The absorbance change at the cut-off (50ng/ml) calibrator is normalized during calibration to provide a reference numerical value arbitrarily set to 1000. These 0142-0453/97 $12.00 (C) 1997 Taylor & Francis Ltd units are referred to as 'normalized qual units'. Negative results are less than 1000 normalized qual units and positive results are greater than or equal to 1000 normalized qual units. The instrument automatically calculates and prints the results of the UTHC analysis in normalized qual units, as well as the interpretation of the result as positive or negative.

Assay optimization
Antibody, particle reagent, diluent and sample volumes were determined using simultaneous variable co-optimization (RS/Discover by Bolt, Beranek and Newman, Inc.) and linear optimization techniques. Diluent volume had the most pronounced effect on the rate of particle aggregation, especially with drug-free samples. Sample size was the most significant factor to influence the shape of the calibration curve (figure 1). A 26 gl sample size was selected to provide optimal precision around the 50 ng/ml cut-off, to provide a reasonable calibration curve up to 100ng/ml and to minimize false positive and false negative results with clinical samples.

Reproducibility
For within-run and total reproducibility experiments, the method was calibrated on the first day of the study. Calibrator, quality controls and urine pools were then analysed in duplicate for 20 non-consecutive days without recalibration. The within-run and total coefficients of variation (% CV), expressed as normalized qual units, were calculated using analysis of variance (NCCLS protocol EP5-T2). The results, shown in table 1, indicate that within-run reproducibility is less than 0.6% and total reproducibility is less than 1-6% at all levels tested. Expressed in concentration units (ng/ml), the within-run and total reproducibility at the 50 ng/ml cut-off are 2.1% and 6.5%, respectively. In addition, the within-run reproducibility of the method was further tested using drug-free urine samples spiked with 40 and 60 ng/ml 11-nor-Ag-THC-9-carboxylic acid.
These samples were tested with 20 replicates per level on two days across five analysers. All of the 400 results were  Sample carry-over was evaluated by interleaving five replicate groups of drug-free urine samples and a sample containing 1886 ng/ml 11-nor-Ag-THC-9-carboxylic acid. No detectable sample carry-over (<2 ng/ml) was observed in any of the drug-free samples.

Stability
To determine the calibration interval for the UTHC method, the method was calibrated on day with the 50 ng/ml cut-off calibrator and samples containing 25, 60 and 100 ng/ml 11-nor-A-THC-9-carboxylic acid were tested (figure 2). The normalized qual units at each of the levels were converted to analyte concentration (ng/ml) using a logit equation and a linear regression analysis performed (y =ng/ml THC, x day). The slopes of these regression equations were -0.007, 0.155, and   Figure 3. Long term stability" the mean results of 10 replicate sets per level were obtained from reagents calibrated on day 1 of the study.
From these regression slopes, the analyte concentrations had changed less than 10% after 30 days. Based on these results, and results from multiple reagent lots, calibration of the UTHC method was determined to be stable for at least 30 days.
The long-term reagent stability of the UTHC method was determined by testing urine controls containing 25, 50, 60 and 75 ng/ml 11-nor-Ag-THC-9-carboxylic acid for 394 days. The means of the normalized qual units (N 10 on each test date) at each of these levels is shown in figure 3.

Specificity
The specificity of the UTHC method to various cannabinoids and cannabinoid metabolites was determined by spiking each compound into drug-free human urine and determining the approximate quantity needed to give a response equivalent to that of the 50ng/ml 11-nor-A -THC-9-carboxylic acid cut-off calibrator (table 2).
To additionally characterize the method's specificity, 79 non-structurally related drugs were added to aliquots of pooled drug-free normal human urine at a concentration of 100 000 ng/ml and found to be negative in the UTHC method (table 3).
Potential interferences for the UTHC method were tested by adding various physiological substances and other compounds to a human urine pool containing 50ng/ml l-nor-A-THC-9-carboxylic acid. All responses from these samples corresponded to concentrations between 45 and 55 ng/ml ll-nor-A-THC-9-carboxylic acid (  [2]. In addition, the US Department of Health and Human Service Guidelines for forensic urine drug testing now specify a 50 ng/ml cut-off. Within-run and total reproducibility, determined using a 20 day NCCLS protocol (EP5-T2) were excellent for the UTHC method. While authors often compare reproducibility for drug of abuse methods in terms of absorbance units, it is more appropriate to transform absorbances to concentrations and to compare reproducibility in concentration units [3]. The normalized qual units as shown in table were converted to concentration units. At the 50 ng/ml cut-off the within-run and total reproducibility were 2-1 and 6.5% respectively.
The calibration stability results presented in figure 2 clearly indicate that the method retains acceptable calibration for at least one month. Results from the long- Table 3. lO0000ng/ml of these compounds were added to drug-free human urine and determined to give a negative result in the UTHC method.  Table 4. These compounds were added at the concentrations shown to a human urine pool containing 50ng/ml ll-nor-A 9-THC-9-carboxylic acid and tested in the UTHC method.  figure 3, demonstrate that the flex reagents are stable for at least one year after being manufactured.
The specificity of the UTHC method to various cannabinoids and cannabinoid metabolites is shown in table 2.
The single most abundant marijuana metabolite found in urine after smoking or ingestion is 11-nor-Ag-THC-9 carboxylic acid [4]. Therefore, a monoclonal antibody having good reactivity to this metabolite was utilized in the method. As 11-nor-Ag-THC-9-carboxylic acid is used to calibrate this method, samples containing greater than 50ng/ml l-nor--THC-9-carboxylic acid, or other metabolites having reactivity equivalent to a 50ng/ml calibrator, will produce a positive result in the method.  [7]. Variation of detection rates between immunoassay systems is probably due to differences among individual drug users' metabolism along with differences in antibody specificity [8,9].
The clinical performance of the Dimension (R) UTHC method was demonstrated by comparison to an EMIT based method on the aca (R) analyser using 445 samples (table 3). A 99-1% agreement between these methods was observed. Three samples, positive on the aca (R) and negative on Dimension(R), with immunochemical reactivity near the cut-off of 50 ng/ml, contained concentrations of 12-17 ng/ml by GC/MS. For the population of samples evaluated in this study, aca (R) (EMIT (R) reagents) and Dimension (R) (ONLINE (R) reagents) detected 99% and 97%, respectively, of samples confirmed positive by GC/ MS. All positive cannabinoid results were confirmed as positive by GC/MS utilizing a cut-off of 10 ng/ml. These differences in cannabinoid detection rates are typical.

RIA (Roche Abuscreen) and ONLINE (R) reagents (on a
Hitachi 717 analyser) detected 99% of a population of 353 GC/MS confirmed positive samples, whereas EMIT (R) II reagents (on a Hitachi 717 analyser) detected 88% and TDx (R) (TDx is a registered trademark of Abbott Laboratories) detected 95% of these same samples [5]. In another study of 100 EMIT (R) positive THC samples, the Triage (R) Panel for Drugs of Abuse (Triage is a registered trademark of Biosite Diagnostics, Inc.) detected 93% [6]. Using the Hitachi 717 analyser with EMIT (R) II, CEDIA (R) (CEDIA is a registered trademark of Microgenics Corporation) and ONLINE (R) reagents, In conclusion, the cannabinoid method developed for use on the Dimension (R) AR and XL chemistry systems provides a precise, accurate and rapid screening approach for measuring cannabinoids on these general chemistry analysers. Since the Dimension (R) is used for the measurement of many other analytes in a true random access mode, the benefits of workstation consolidation may be realized.