Comparison of stochiometric and colloidal adsorption models for mechanistic modeling of parvovirus clearance by anion exchange chromatography

Anion exchange chromatography (AEX) is one of the downstream unit operations that is most frequently claimed for its capability to remove viruses. However, the impact of various process parameters on virus removal by AEX is still not fully understood. Mechanistic modeling could be a promising way to approach this knowledge gap. These models leverage physical and chemical principles to simulate a wide range of experimental conditions based on a limited number of wet lab calibration experiments. Especially the reduced need for resource-intensive virus spiking studies makes them a valuable tool to improve mechanistic understanding of viral clearance in silico. We compared a stoichiometric and a colloidal adsorption model of parvoviral mock virus particles (MVP) on Q Sepharose FF to assess their abilities to predict LRVs of MVP over a pH range from 6.1 - 7.7 and sodium chloride concentrations from 0 - 400 mM. A general rate column model including pore diffusion was applied based on the finding that MVP can access ∼3 % of the pore volume. While the stoichiometric model needed three pH-dependent parameters to describe the pH range, we were able to calibrate a colloidal model with only the charge being pH-dependent. Validation at the edge-of-failure showed good and comparable predictive power for both models. The colloidal particle adsorption (CPA) model showed a better fit to a validation dataset generated with process intermediates containing mAbs at higher volumetric loads. Since no residual MVPs were detected at low ionic strength (e.g. < 140 mM NaCl at pH 6.1 or < 220 mM NaCl at pH 7.7) the behavior of the adsorption models at these conditions could not be verified. This study contributes to the applicability of in silico methods to virus removal by AEX and could aid knowledge-based platform validation of virus clearance by chromatographic methods.