Abstract
FDA's experience to date has shown that completion of stability data requirements is one of the most observed challenges for applicants of New Drug Applications (NDAs) with an expedited review designation. Since NDAs submitted under these expedited pathways often have limited available real-time stability data from the primary batches, Modeling Approaches to Reimagine Stability (MARS) have been proposed to support establishment of tentative retest periods of the drug substance and/or expiration dating period (shelf-life) of the drug product. MARS incorporate statistical principles and available tools as a part of the predictive models. In this study, a data mining exercise has been conducted with regulatory submissions of Investigational New Drug (IND) Applications, NDAs, and Abbreviated New Drug Applications (ANDAs) containing MARS data. The case studies presented herein demonstrate how MARS data has been applied to regulatory scenarios involving prediction of retest and/or shelf-life, bridging major development changes, and confirming that no degradation has been observed or predicted. Using the assumption of a linear time trend for those cases that do not display sufficient degradation to conduct MARS for projection of an expiration date, an analysis of covariance (ANCOVA) model is developed and described herein to test the hypothesis of zero slope by a p-value method. Our results show that the application of MARS adequately supported establishment of a tentative commercially viable retest date/shelf-life, thus enabling earlier access to critical drugs for patients with unmet medical needs.