How should assay linearity be evaluated when validating IVD immunoassay calibrators and working ranges? Best Practices

Linearity evaluation is not a mere checkbox; it is the foundational validation that confirms an assay's ability to produce results directly proportional to the concentration of the analyte. The primary method involves creating a dilution series from a high-concentration clinical specimen, covering the entire analytical measurement range from the upper limit down to below the lowest non-zero calibrator. You then plot the measured values against the expected, diluted values, statistically verifying that the relationship remains linear without significant bias.

The core purpose of evaluating linearity in IVD immunoassays is to provide documented evidence that the assay system accurately recovers the target analyte throughout the working range. For systems relying on reduced calibration models, this verification is critical to guarantee that a two-point calibration curve can reliably interpolate patient results across all reportable concentrations.

The Foundational Protocol for Linearity Assessment

The core objective is to challenge the assay’s signal-to-concentration relationship across its entire reportable range. This process directly validates the accuracy of the calibrators and the integrity of the working range that depends on them.

Selecting the Correct Test Samples

Using the right biological material is the single most important factor. Matrix integrity must be preserved.

High-concentration clinical specimens containing endogenous analyte are the gold standard. Serial dilutions must be made using a low-concentration or analyte-free serum matrix, never water or simple buffers. Using a non-biological diluent introduces a matrix effect, altering protein content and potentially disrupting the antibody-antigen binding kinetics that the calibrators are designed to reflect.

Alternatively, a set of five or more calibrators can be prepared in an analyte-free serum matrix. This allows a direct comparison of the assay's signal against a set of known, artificially created standards. However, this method primarily validates the pipetting and the signal response and may not expose matrix-specific biases from endogenous analytes.

Executing the Dilution Protocol

A systematic approach ensures coverage of the full analytical measurement range (AMR) and the critical medical decision points.

You must start with a sample near the upper limit of quantification (ULOQ). Serially dilute this high-value sample in a stepwise manner to create 5–7 concentrations that descend past the lower limit of quantification (LLOQ) and below the lowest non-zero calibrator. This "bracketing" of the calibrated range directly tests the limits of interpolation.

For dilutional linearity, especially for high-dose hook checks, you must push beyond the upper limit. Fortify a clinical specimen above the ULOQ and then use validated dilution factors (e.g., 2-, 5-, 10-fold) to bring it back into range. A diluent matched to the test matrix or a validated analyte-free surrogate is essential here.

Analyze each dilution level in triplicate to assess precision at each point. The plot of mean observed concentration against the expected, dilution-corrected concentration must be linear. Any deviation, particularly a "hockey stick" shape at the high end, immediately signals a high-dose hook effect or signal saturation.

Statistical Confirmation of a Linear Relationship

A visual plot is a good start, but objective acceptance requires rigorous statistical checks. The goal is to prove that the relationship is not just a good fit, but the simplest possible linear fit.

The Percent Recovery Method

This is the most intuitive and widely accepted assessment of individual points across the range.

Calculate the recovery for each dilution level as (Observed Mean / Expected Value) × 100%. The generally accepted criterion is that all recovery values must fall strictly within the 95% to 105% range. A systematic drift in recovery from low to high concentrations can reveal subtle matrix effects or non-specific binding.

Polynomial Regression and Residual Analysis

This evaluates the fit of the entire dataset to mathematically confirm linearity.

Perform polynomial regression to test for a significant non-linear coefficient. A linear (first-order) model should be the best fit. The statistical test determines if adding a quadratic (second-order) or cubic (third-order) term significantly improves the fit. If it does, linearity is compromised.

A runs test on the residuals is equally critical. This test checks for randomness in the error distribution. A non-random pattern—such as a clustering of positive and then negative residuals—visually confirms a curve even when the raw data plot looks convincing. Randomly distributed residuals are the hallmark of a true linear relationship.

Understanding the Trade-offs and Pitfalls

Objectivity requires a clear-eyed view of what can go wrong. The most common failures stem from shortcuts in sample preparation.

Using water or buffer as a diluent is a primary source of false non-linearity. The resultant protein crash alters sample density and can introduce a systematic bias that is a matrix artifact, not a flaw in the calibrator system. This generates a non-linear recovery pattern that disappears when the test is correctly repeated with a serum matrix diluent.

Another critical pitfall is the calibration model itself. Two-point calibration is mathematically dependent on perfect linearity through the coordinate origin. If a degree of inherent, acceptable non-linearity exists at the extremes, a two-point model will amplify that error during interpolation. Validating linearity against a full, multi-point calibration curve is often necessary to create a safe "working range" before locking in a two-point model.

Making the Right Choice for Your Validation Goal

Your specific validation objective will dictate the rigor of the protocol. The approach must be tailored to the question you are answering.

• If your primary focus is validating a new assay’s calibrators and AMR: Use a high-concentration native patient pool serially diluted with an analyte-free serum matrix. Subject the data to both percent recovery criteria (95–105%) and polynomial residual analysis to ensure the calibrators truly reflect the biological analyte’s behavior across the entire range.

• If your primary focus is verifying a dilution protocol for high patient samples: Fortify a sample above the ULOQ and dilute it with your validated diluent until it falls within the linear range. Accept the results only if the dilution-corrected mean bias is within ±15% and the coefficient of variation (CV) is below 15%, thus proving the assay’s ability to safely extend its reportable range.

A deliberate, matrix-aware linearity study is the definitive proof that your assay does not just work with standards but delivers truth for patients, from the very bottom to the very top of its stated range.

Summary Table:

Ready to optimize your assay performance and linearity validation? CamelBio provides diagnostic manufacturers, labs, and research institutes with one-stop access to premium IVD raw materials, technical services, and expert consulting—covering every stage from concept to clinic. Accelerate your IVD development with our end-to-end solutions — contact us today!