Efficient determination of developable protein drug candidates and stable solution conditions is a key challenge in industrial drug development. Protein aggregation is difficult to predict and can lead to challenges in manufacturing, storage, and patient safety. In this work, stability of four monoclonal antibodies (MAbs) were studied at a wide range of solution conditions and incubation temperatures intended to systematically evaluate attributes that can influence aggregation rates. The studies were conducted as a function of pH, ionic strength, MAb concentrations, and incubation temperatures that were representative of industrial stability studies. Results were analyzed in the contexts of conformational stability and net self-interactions. Interpretable machine learning models were applied to parse and quantify the phenomena relevant to high-concentration aggregation rates, with emphasis on refrigerated conditions representative of common storage conditions for MAb products. The results indicated that the aggregation rates were non-Arrhenius, and stability studies at 30 to 50 °C were broadly misleading with respect to the stability rankings of the different formulations and MAbs in comparison to the stability rankings at refrigerated storage conditions. For this set of MAbs and formulation conditions, the net valence was the most significant predictor of aggregation rates at refrigerated storage conditions.
Forder et al. (Thu,) studied this question.