Validating your ELISA kit

Would your ELISA kit meet the validation criteria for a commercial assay?

You’ve titred your antibodies and have a good calibration curve, but can you be confident in the performance of your new ELISA test? Here are some of the key validation steps commercial diagnostic companies take before launching a new ELISA test kit.

Stress testing of components

Most ELISA kits can be stored refrigerated. Freezing is generally only recommended for patient samples, and sometimes controls and calibrators. Repeated freezing and thawing of components is discouraged and the user is instructed to aliquot and freeze calibrators and controls where necessary.

Unlike in-house ELISA reagents and plates which are often used the same day as they are made, commercial ELISA kits need to have long shelf lives and show tolerance to the extremes of temperature and humidity that might be encountered during transportation to distributors and customers. All components are stress tested through repeated freeze-thaw cycles, high humidity and raised temperatures at 37C for prolonged periods to simulate storage in adverse environments. Components will also be screened for microbial contamination and growth throughout their shelf life. If a significant deterioration in performance is observed then the kit will be passed back to the R&D department for reformulation. The most common issue is with elevated temperatures and freeze thaw cycles on ready to use enzyme conjugates and the effect of high humidity on the stability of the coated plate. Humidity can be controlled through storing the plate after drying in a sealed pouch with desiccant packs. The effect of humidity and temperature on the coated plate stability can also be addressed using additives in the coating solution or plate blocking solution.

Specimen collection

Some blood tests require plasma samples where a specific type of anticoagulant, for example, edta, citrate or heparin may be specified whilst others will require serum collected in a plain tube or one with  clotting activator. Manufacturers often provide a table of results in the instructions for use illustrating the recoveries of analytes with different anticoagulants and serum.

Haemolysed, lipaemic and samples with high levels of bilirubin or rheumatoid factor may also interfere with the assay and the manufacturer will provide information on whether these types of samples can be used in the assay.
ELISA kits that use cell culture supernatants or urine samples may require the user to centrifuge or 0.2µm filter the samples before use to remove cells, debris and other particulates that might interfere with the analysis. Repeated freeze-thaw cycles should be avoided and the sample mixed well before adding to the ELISA plate well.

Sample stability

Patient samples are often stored frozen. To show sample stability, patient samples are freeze-thawed multiple times and the recovery of analyte is calculated after each cycle.

Calculation of results

A standard curve is constructed from the absorbance or rate of increase of absorbance in each well at the end of the assay and the concentration of the calibrator. Readings from patient samples can be interpolated from the standard curve. The manufacturer will identify an appropriate formula for curve fitting and specify suitable graph paper for plotting the standard curve.

Analytical sensitivity

The limit of detection is frequently defined as the analyte concentration resulting in an absorbance reading slightly higher (mean plus two standard deviations) compared with the zero calibrator.

Reproducibility

Intra-assay precision

Multiple replicates of samples are run in a single ELISA plate and the coefficient of variation (%CV) of the interpolated concentration of analyte is calculated. If samples are required to be run in duplicate, then replicates of duplicates will be run and the duplicates averaged before calculating the CV. For example, twenty replicates of duplicate samples would use a total of 40 wells on the ELISA plate. Whether the magnitude of the CV is acceptable will depend on the clinical application of the test but generally the CV should be less than 10%. Precision is usually measured at three clinically relevant levels across the working range of the standard curve.

Inter-assay precision

Assay to assay precision within a laboratory using different plates, for example duplicates from 20 separate runs, are calculated in a similar way to the intra-assay precision. Here, the CV’s are often slightly larger than with intra-assay as the between run errors need to be taken into account.

Spike and recovery

To show the suitability of the antigen preparation from which the calibrator sets are made, concentrated analyte is spiked into normal patient samples with three different but known levels of analyte. The recovery of the spiked material can then be calculated. If the observed levels are significantly different to expected then the suitability of the antigen preparation should be questioned.

Parallelism and dilution

The matrix with which commercial calibrator sets are manufactured may not be the same as the sample under investigation. There are many reasons why a manufacturer would use a different matrix to that of the sample under investigation; not only availability, cost and stability but also to ensure that the calibrator matrix is free from endogenous analyte. Practically, patient samples, containing high levels of analyte are diluted in the zero calibrator matrix and the observed levels compared with expected levels of analyte and parallelism with the calibration curve. Manufacturers will ensure that both spike recovery and parallelism are acceptable before release of the ELISA kit.

Specificity

Samples with known levels of analyte are spiked with relevant cross-reactants such as therapeutic drugs, and potential cross-reacting substances. The detection of cross reactivity, if any, will be documented in the instructions for use. If the cross-reactivity is unacceptable, the choice of antibodies may need to be reviewed.

High dose hook effect

Although the high dose hook effect occurs mostly in immunometric assays where the labelled antibody, analyte and coated antibody are incubated simultaneously, it is a simple procedure and worthwhile checking anyway irrespective of the assay format.

Calibrator matrix can be spiked with a very high level of analyte, higher than would be normally expected to be found in a patient sample, and then serially diluted in calibrator matrix in to the working range of the assay. The high dose hook effect generally is seen as a bell shaped curve. If the level of analyte causing the effect is within the range expected for patient samples then the assay is usually passed back to R&D for reformulation.

Expected values

Expected values from normal healthy donors should be determined but care needs to be taken to ensure that they are representative of the reference population with which the tests will ultimately be used.

Clinical sensitivity and specificity

Clinical studies using the ELISA kit can help identify appropriate cut-offs and an indication of the usefulness of the test as a clinical diagnostic tool. Measurements of samples from healthy subjects, free from disease, will be taken and compared with measurements from patients with known disease.

Sensitivity, the ability of a test to correctly classify individuals as having a specific disease, can be calculated by the true positives (TP) divided by the sum of true positives (TP) and false negatives (FN), Specificity, the ability of a test to correctly classify individuals as free from disease, can be calculated as the true negatives (TN) divided by the sum of the true negatives (TN) and false positives (FP).

Method comparison

If possible, it is good practice to compare the results from patient samples run in your ELISA kit with a commercial kit which will help give reassurance that your ELISA has equivalent performance. As above, the results can be presented in terms of sensitivity and specificity versus the ‘gold standard’ diagnostic. A graph showing the relationship between the two tests together with an equation of the straight line relationship and correlation coefficient can be helpful in deciding whether to continue with an in-house assay or investment in a commercial kit.

Summary

Every ELISA test is unique and ultimately, the requirements for validation equally so. It is not uncommon for commercial immunoassays, during development, to fail at the last validation step. The order in which you progress the validation steps will depend on the specific application of your assay. Changes to the formulations will have to be made and the validation process repeated until the performance of the ELISA kit is acceptable. Even if your assay is only used in your lab for your own research project, knowing the key performance characteristics can help reproducibility and confidence in your results.

If you would like any further information on validating your ELISA test please contact JRBiomedical