Leveraging prior knowledge for process parameter classification in mAb Protein A chromatography

Protein A (ProA) affinity chromatography plays an essential role in purifying monoclonal antibodies (mAbs) and their analogues by reducing impurities like residual host cell proteins (HCPs), residual DNA, process additives, and potential viral contaminants. Decades of mAb process development and commercialization efforts have built extensive prior knowledge in the Protein A process. The prior knowledge facilities streamlined process development and minimized the need for extensive process characterization studies to inform manufacturing control strategies. This manuscript presents a comprehensive prior knowledge package, consolidating process parameter characterization data from ten molecules developed by Eli Lilly and Company using the Protein A chromatography process. Results from multiple Design of Experiment (DOE) studies on these molecules demonstrated that no process parameters significantly impacted critical quality attributes when operated within platform ranges. Additionally, a Bayesian hierarchical model was applied to analyze historical data and predict the effects of process parameters, further confirming that parameter effects were insignificant across the platform ranges for the Protein A process. By leveraging this historical data package, we directly supported the classification of ProA process parameters for new therapeutic antibodies, effectively replacing the need for product-specific process characterization evaluations. This approach has been positively received by global regulatory agencies during the market authorization filings for two Lilly's products.