(513e) Leveraging Molecular Modeling in the Selection of Protein-Drug Candidates for Late-Stage Developability
AIChE Annual Meeting
2022
2022 Annual Meeting
Pharmaceutical Discovery, Development and Manufacturing Forum
Advances in New Modalities: Predictive Modeling Technologies
Wednesday, November 16, 2022 - 1:54pm to 2:15pm
Understanding and controlling the stability of protein solutions is a particularly relevant and current issue in the biopharmaceutical pipeline. Proteins are marginally stable and are susceptible to undergo multiple degradation routes such as aggregation, phase separation and oxidation, which poses a significant challenge for the development and manufacturing of biotherapeutics. Traditionally, these instabilities are mitigated by minimizing protein interactions via changes in processing and formulation conditions (e.g., adjusting pH, addition of excipients). While this strategy has been proven successful for advancing biotherapeutics to clinical trials and subsequently to help patients, it generally requires large amount of resources (both time and material) to identify suitable processing and phase-appropriate solution conditions. In this regard, there is a need for developing complementary approaches that reduce the time and cost for the advancement of biologics and accelerate bringing drug products to patients. This talk explores one of such approaches that relies on the use of computational tools for selecting biologic drug candidates with minimized risks for formulation and process development. Specifically, this work aims to guide the early large-molecule development to modulate not only their potency and efficacy but also their interactions to enhance their physical and chemical stability. This talk illustrates the implementation of different fundamental and statistical approaches (e.g., molecular modeling and machine learning) to identify the presence of interacting regions or âpatchesâ on the protein surface, which are related to instability problems in biotherapeutics. Moreover, the results show different strategies based on statistical modeling and atomistic and coarse-grained simulations to identify key residues in the protein surface, which yield the highest impact in the distribution of protein surface patches. These results are contrasted against experimental biophysical and biochemical characterization of the thermodynamic stability and aggregation propensity of multiple point-mutations of a model monoclonal antibody.