A Novel Lack-of-Fit Assessment as a System Suitability Test for Potency Assays

Abstract

Bioassay data analysis is used to determine the potency of protein therapeutics. To properly determine potency, the experimental data need to be fitted to a model that adequately describes the observed dose-response relationship. Typical models include 4-parameter logistic curve fits, 5-parameter logistic curve fits or parallel line analysis. Lack-of-fit assessment can be used as a measure of potency assay system suitability to ensure appropriate closeness of the chosen model fit to the experimental data. We present a novel lack-of-fit approach that overcomes the shortcomings of previously described lack-of-fit tests, such as the conventional analysis of variance (ANOVA) F-test and the lack-of-fit sum of squares test. Simulation studies and examples are used to assess the performance of the new lack-of-fit test. The results show that the described lack-of-fit approach can effectively reject poorly fitted data while retaining well-fitted data, and has advantages in potency assay applications where instrument-to-instrument variability in absolute readout is expected.

LAY ABSTRACT: Potency assays are analytical procedures used for characterization as well as release and stability analysis in drug development and for approved products. Dose-response data generated from a drug sample and a well-characterized reference standard are evaluated to determine the potency of the drug sample relative to the reference standard. In order to obtain a potency determination, dose-response data need to be fitted to a proper model that adequately describes the observed dose-response relationship. There are different options described to assess the goodness-of-fit of the data. One approach is the goodness-of-fit assessment based on F-test. This approach compares the lack-of-fit error (representing the discrepancy between observed data and fitted curve) to the pure error (representing the random noise between replicate measurement) to determine if the observed lack-of-fit error can be attributed to random noise. A limitation of goodness of fit assessments via F-test lies in its propensity to penalize precise data (small lack-of-fit error can be considered significantly high if the assay has exceptionally low pure error) and accept undesirable noisy data (large undesirable lack-of-fit error can be considered insignificant due to large pure error). An alternative approach based on lack-of-fit sum of squares is only applicable to certain types of assays where the magnitude of measurements is consistent across different instruments given that the lack-of-fit sum of squares will increase when the magnitude of the assay signal measurements increase, even if the relative magnitude of assay data versus fitted curve remains the same. We introduce here a novel approach that overcomes the limitations of F-test and sum of squares-based approaches. This new approach will effectively reject poor data and retain good data, and it is independent of differences in absolute readout across instruments.