Before testing and reporting results, laboratories validate assays. The general requirement for this process includes providing accuracy, precision, limits, stability, recovery, selectivity, calibration model (linearity) and robustness. The recommended deviations from expected values for all these components of validation differ for screening and confirmation tests.
To establish adequate assay response, cutoff concentrations are used for screening tests. The assay should have inequity around the cutoff. Positive samples should be 90% (with concentrations 20% above the cutoff concentration) and the negative control samples should have the same percentage as positive samples (with concentrations 20% below the cutoff concentrations). Though the required number of control samples to determine this process is not formally established, however it’s a preferred practice to have smaller number of samples per run and multiple runs rather than a large number of samples in one run (11). More than 5 different batches are required to be tested on different days for the control samples. Accuracy, precision, limits and linearity of a screening assay are satisfied by adopting this manner of testing of the control samples. Revalidating an assay in response to each change in calibrator or reagent lot or with instrumentation changes is considered a fine laboratory practice. Comparison of the new assay should be done with the previously validating assays in current use. Acceptable limits for differences between the current and new analyses should be illustrated in the laboratory SOP.
Since selectivity for screening assays is pivotal, regulated workplace drug testing uses immunoassays to screen urine for drugs or their metabolites. Though assays should ideally be tested only for compound of interest (unlike other applications where detection of a broad range of drugs is preferable), most immunoassays cross-react with similar compounds existing in the urine; a case example of the case is point is of amphetamine. Sympathomimetic amines such as pseudoephrine and phenylpropanolamine along with amphetamine and methamphetamine are also detected in many marketed assays. Identification and minimization of the number of interfering compound through a choice of assays or a combination of assays is important in validation. To determine that combination of screening and confirmation tests of a specimen is positive only for the drug or metabolite of interest is critical.
The laboratories must ensure that assays remain viable throughout the test; however, there are no specific standards for stability and robustness. Performed on a neat specimen, recovery is not an issue with immunoassays or most screening tests. Executed on high-volume analyzers, most initial tests have automated pipetting systems. Demonstrating that there is no carryover of one specimen to the nest during the next is also important. This can be done by following a drug free sample after analyzing a number of specimens with high concentration of drug.
Gas chromatography-mass spectrometry methods (GC-MS), a particular validation requirement for confirmation assays, along with other validation requirements are more quantifiable than for screening tests. Analysis of several (not specified) control specimens prepared using the test matrix establishes the accuracy and impression. Within-assays, between-assay and total impression should be established. Testing controls in various hatches on different days is important for validation as mentioned in the Screening section. Generally, each batch should have 3 to 5 controls and more than 5 batches should be analyzed. 20% of expected values should be quantified and coefficients of variation ±20% for repeated measures near the cutoff concentration. At least 40% of the cutoff concentration must constitute the lower limit of quantification (LOQ). Retention time for all these measures must be ±2% and the dual ratios of monitored ions ±20% amongst those in the calibrator(s). Being the limiting factor in the detection, the ion ratio requirement usually makes the limit of detection (LOD) equivalent to the LOQ. Since LOD is the cutoff concentration used in retests, it is a crucial feature of the assay and is usually cited as the limit of contamination and carryover in drug-free specimens. The upper limit of linearity (ULOL) must be determined and though a specific linear range does not exist, the drug contamination is considered invalid if it’s measured above this limit without specimen dilution or reanalysis.
Analogous to screening assays, analyzing structurally similar compounds in the test matrix and determining potential interference for GC-MS must be done by laboratories. To challenge the assay, the concentrations of potentially interfering substances should be in the upper range of expected physiologically relevant values. Ability to remove the interference – if detected- and properly identifying and quantify the compound of interest must be demonstrated by the laboratories. Correctly quantifying the compound of interest at the cutoff level is required and many laboratories also ensure that the potential interfering compounds and not quantified in the drug-free samples. Testing a few actual specimens along with established assays is a good practice when the new assays are put into practice for the first time. Detecting unusual interfering substances, which may not have included in controlled studies, is made easier this way for the drug-testing laboratories.
Specific requirements for recovery do not exist as deuterated internal standards are used depending on their availability; for all SAMHSA-specified drugs in the workplace urine testing program, they are not employed. Measured concentrations for loss of analyte during extraction, injection and analysis process are corrected through the internal standards. A specified range for specimen-specific recovery requiring the abundance of the internal standard ion used for quantification of each specimen to be 50-200% has been recently introduced for the calibrator or QC samples. Adding to internal standards, this requirement acts as a check for errors.
Other than continuous monitoring during all subsequent testing, assay robustness and stability have no designated requirements. Features of assay such as stability of analytical solutions, hydrolysis conditions, lot-to-lot changes, and derivatization conditions, stability of analyte derivatives, changes of instruments / analysts and changes in laboratory environment are included in monitoring. Monitoring QC specimens in each analytical run and usage of the limits of these controls determine the limits of acceptability. Having certifying scientists to evaluate actual specimens for any unusual findings is included in robustness, e.g. occasional sample specific interferences. Most laboratories maintain performance standards for months or years, even though specific requirements for robustness do not exist, and adopt procedures that are not sensitive to change in analysts, instruments or conditions.
QA follows once the assays are validated and placed into service. There must be a QC sample with no drug, in each initial testing latch, with one of it targeting 75% of the cutoff concentration and the other one targeting 125% of the cutoff concentration. Both the batches should have a minimum of 10% QC samples and 1% blind QC i.e. appearing as a donor specimen to the analyst. A QC with drug above the cutoff concentration must screen positive and those below must screen negative to be acceptable. Bracketed by open QC samples that have correct results is a requirement for each specimen result. The entire testing batch has to be aliquoted again and retested if the blind specimen is incorrectly identified.
Though the number of QC samples for confirmation testing bears no requirements, in each batch there must be a drug-free control, one targeted at 125% of the cutoff concentration, one targeted at 40% or below the cutoff concentration and one to test the effectiveness of any hydrolysis procedures. Results for controls containing drug must be ±20 of expected concentration and the drug-free control must have an assay response below the LOD to be acceptable. Two unique lots of reference material must be prepared for controls and calibrators in the test matrix. As occasionally with ∆9-tetrehydrocannabinol-9-carboxylic acid (THCCOOH), when unique lots are not feasible they should be prepared from two different stock solutions.
Many laboratories have carryover controls in all batches, however during validation the assays can be checked for carryover of drug from one specimen to the other. This often includes a high concentration sample followed by a drug-free sample. Though special QC requirements, such as those for isomer analysis do exist, the controls should, in accordance to the general principle, challenge the criteria for distinguishing positive and negative results.
Testing new and old calibrators and controls in parallel to certifying the new materials is a specific requirement causing QA dilemma. Though it is a good practice, the old material is often replaced due to its produced results that are outside tolerance or trending towards the limits of acceptability, which raises the concern, “Should we really certify the new material by comparing it to a previously certified sample that is known to be deteriorating?” The answer is negative, however a variety of solutions are there to address this problem. Designing the QA system such that it identifies trends early before certified material expires can be one possible solution. The new material should be certified through independent reference material with known concentrations of drug, once the certified calibrators or controls fail unexpectedly.