Reliable stability prediction to manage research or marketed vaccines and pharmaceutical products. "Avoid any doubt for the end-user of vaccine compliance at time of administration"

Abstract

A major challenge for the pharmaceutical/vaccine industry is to anticipate and test/control product stability, regardless of the time/temperature profile of the product, from release to administration. Current empirical stability protocols performed to ensure product stability remain limited to the prediction of product stability in a thermal excursion (cold chain break) during their long-term storage. As recently recommended by the World Health Organization, mathematical models can be used for shelf-life and stability predictions. Therefore, various approaches have been published with good performance for simple chemical reactions. However, for biomolecules/vaccines, more complex reaction profiles require more complex models to predict their stability with a good level of confidence. This complexity constitutes a real scientific challenge because the number of model parameters increases with model complexity and need to be balanced with the limited number and quality of the available experimental data. We have developed a dedicated method/software based on different vaccines/pharmaceutical case studies. This predictive method considers phenomenological models, five levels of model confidence assessment, predictive quality value and simulated designs of experiment to improve and define the limits within which the prediction models can be used, and to increase model/prediction confidence to the required regulatory and scientific levels. This artificial intelligence system should help to avoid any doubt of stability at time of vaccine injection.

Keywords: Confidence levels; Modeling; New software; Prediction; Regulatory requirements; Vaccine stability DoE.