Objective criteria for statistical assessments of cross validation of bioanalytical methods

To improve the integrity and comparability of pharmacokinetic data in clinical trials by refining the statistical assessment methodology for cross validation of bioanalytical methods across multiple laboratories.

Cross validation assessments were conducted using statistical tools recommended by International Council for Harmonization M10 guidance, including Bland-Altman plots, Deming regression, and Lin's Concordance. Recognizing limitations in Deming regression and Lin's Concordance for interpreting cross validation results, we introduced a combined approach: Bland-Altman plots with equivalence testing. The acceptance threshold was defined such that the 95% confidence interval of the mean log10 difference between laboratories must fall within boundaries based on method validation criteria.

The proposed methodology was validated across diverse bioanalytical methods. Unlike conventional approaches that impose strict constraints on Deming regression parameters, our framework accommodates practical assay variability. This approach provided consistent and credible cross validation outcomes in real-world scenarios.

Integrating Bland-Altman plots with equivalence boundaries offers a robust, statistically sound framework for cross validation in bioanalytical studies. This method enhances the quality and consistency of pharmacokinetic data, supporting more reliable clinical trial endpoints.