Abstract
We designed a three-step statistical approach to transfer bioanalytical assays (ELISA and Biacore) which evaluates the (1) average equivalence between the two labs (2) concordance in individual sample results between the two labs, and (3) long-term stability of assay performance. Each experimental design evaluated the contribution of four critical variables to the overall variability. Two lots of each variable were examined in a controlled experiment. The variables tested for ELISA were analyst, plate washer, biotinylated-therapeutic protein, and streptavidin-horseradish peroxidase; and for Biacore were analyst, instrument, chip lot, and conjugation chemistry reagent lots. Equivalence in the mean signal to noise (S/N) or mean relative units (RU) between the two labs was established through statistical evaluation of the assay performance characteristics across multiple assay variables. Concordance between the two labs in the individual sample results was subsequently verified both quantitatively and qualitatively. The long-term maintenance of assay stability was monitored by performance testing of a predefined set of samples which were prepared in sufficient quantities to last several years. The process of method validation for biomarker testing in clinical trials is to analyze the variability of the assay performance. However, different factors contribute to this variability and need to be evaluated when the method is transferred to another site/lab. Lack of understanding the critical variables can potentially result in unexpected problems and delays. The three-step statistical approach of assay transfer provides a robust process for transferring complex biological assays.