A Bayesian statistical approach to continuous qualification of a bioassay.

A validated bioassay is used to measure the potency of commercial lots, and as such, must be accurate, precise, and fit for its intended purpose. Regulatory guidelines for the validation of a bioassay include a characterization of features, such as accuracy, precision, linearity, and range. The journey of a validated bioassay typically starts in a development lab, where it may be eventually be qualified to measure the potency of clinical lots. Prior to validation, as the bioassay is transferred to new laboratories, the bioassay is studied and assessed across a series of small experiments. In this work, instead of considering each study as a separate, independent report, it is proposed that, beginning with the qualification study, the accuracy and precision of the bioassay be continuously characterized, with each subsequent study result building from the ones that preceded. We call this continuous qualification. Such a proposition is naturally carried out using Bayesian statistical methods in which the historical study data is used to construct prior knowledge that is blended with the current study data. By doing so, bioassay accuracy and precision may be assessed with high confidence well ahead of commercial manufacturing. Further, by following the total-variance approach, the method also allows for the construction of control limits for potency.