Six ELISA Validation Mistakes That Slip Through Peer Review

ELISA validation looks straightforward on paper. You run your accuracy, precision, and linearity experiments, you calculate your CVs, and you write it up. But a surprising number of validation packages that arrive at contract laboratories for method transfer contain the same small gaps. Gaps that are easy to miss during peer review and expensive to find later.

Forgetting to validate in the actual study matrix ¶

The most common problem is validating in a surrogate matrix. Phosphate-buffered saline, for example. When the study samples will be serum or plasma. Matrix effects in biological fluids are real and variable. Proteins, lipids, and haemolysis can all suppress or enhance signal in ways that a PBS-based validation will not detect.

The fix is straightforward: run your accuracy and precision experiments in the matrix you will actually use, spiked at three concentrations across the dynamic range. If your matrix is difficult to source in a blank form, a charcoal-stripped or immunodepleted version is usually acceptable, but document the rationale.

Treating the standard curve as a formality ¶

Standard curves are often run once, accepted if the R-squared value looks reasonable, and then used as the reference for all subsequent calculations. But a single curve run on a single day does not tell you whether the curve is stable across analysts, reagent lots, or ambient temperature variation.

Run your standard curve on at least three separate days, with at least two different reagent lots if possible. Calculate the inter-assay CV for each standard point. If any point shows a CV above 15%, investigate before the method goes into routine use.

Undocumented incubation conditions ¶

Incubation time and temperature are almost always specified in a written protocol. What is less often specified is whether the plate should be covered, whether it should be on a rocker or shaker, and at what speed. These details are frequently passed on verbally and never written down.

In a method-transfer context, this is where reproducibility problems start. The receiving laboratory follows the written protocol exactly and gets different results from the originating laboratory, because the originating laboratory was doing something the protocol did not say. Ask the originating scientist to walk through the protocol verbally and compare it to the written version step by step.

Skipping freeze-thaw stability ¶

Most study samples are not run fresh. They are collected, frozen, and run in batches. Sometimes after two or three freeze-thaw cycles. If your validation does not include a freeze-thaw stability assessment, you do not know whether your method is measuring what you think it is measuring by the time the samples reach the plate.

A three-cycle assessment at minus 20 and minus 80 degrees Celsius, with percentage recovery calculated against a freshly prepared reference, is sufficient for most applications. It adds a day to the validation timeline and removes a significant source of uncertainty from the study data.

None of these gaps are difficult to close once you know they are there. The challenge is that they are easy to overlook when you are close to the work. A second pair of eyes on a validation package before it goes into a method-transfer brief is usually worth the time.