Stability modeling methodologies to enable earlier patient access.

Abstract

Over recent years, confidence has been gained that predictive stability modeling approaches using statistical tools, prior knowledge and industry experience enable, in many instances, a robust and reliable shelf-life/expiry or retest period prediction for medicinal products. These science and risk-based approaches can compensate for not having a complete real-time stability data set to be included in regulatory applications at the time of initial submission and, thereby, accelerate the availability of new medicines. Examples of predictive stability modeling include accelerated stability assessment procedure (ASAP), advanced kinetic modeling (AKM), and novel modeling approaches that involve the use of Bayesian statistics and Artificial Intelligence (AI) applications such as Machine Learning (ML), with applicability to both synthetic and biological molecules. For biologics, product-specific and platform prior knowledge could be used to overcome model limitations known for non-quantitative stability indicating attributes. A successful ongoing verification approach by comparing the predicted data with real-time stability data would be an appropriate risk management approach which is intended to address regulatory concerns, and further build confidence in the robustness of these predictive modelling approaches with regulatory agencies. Global regulatory acceptance of stability modeling could allow patients to receive potential life-saving medications faster without compromising quality, safety or efficacy.