From a regulatory perspective, over the last three decades, there has been a shift toward subcutaneous products due to their self-administration and patient-friendly nature. Cornelius Pompe, Chief Development Officer at Leukocare, explained that his company uses data science and machine learning to rationalise excipient selection and optimise formulation screening. Leukocare’s technology enhances subcutaneous delivery by eliminating the need for clinical dose preparations and the administration of products via healthcare professionals, such as in IV settings.
Pompe stressed that data science plays a critical role throughout the formulation process, from using amino acid sequences and target product profiles to guide excipient selection, to applying machine learning models like random forests to predict excipient suitability. The company has developed internal databases by scraping and structuring data from public sources, therefore enabling more informed and efficient formulation decisions.
Shifting gears a little, Pompe detailed how molecular modelling and sequence alignment tools help identify chemical liabilities and outlier properties in new molecules. These insights are used to rationalise excipient choices and streamline development. For example, a case study involving an antibody with exposed tyrosine residues showed how cyclodextrin excipients improved solubility and reduced viscosity.
Pompe also discussed the use of advanced kinetic modelling (AKM) for shelf-life prediction, demonstrating that reliable stability forecasts can be made within two months, thus significantly accelerating decision-making. He presented successful case studies, including the reformulation of Herceptin to reduce viscosity and aggregation, and the conversion of a lyophilised intravenous biosimilar into a stable, high-concentration SC liquid formulation. To wrap up, Pompe reiterated that the integration of formulation expertise with data science not only enhances productivity but also enables the development of patient-centric, high-quality drug products.
