Monoclonal antibodies (mAb) inherently contain charge variants that can be partially separated in preparative-scale cation-exchange chromatography. Different models for the multi-component system are available but parameter estimation for such a system remains non-trivial. Moreover, the popular models sometimes fail to capture the elution details at both low and saturated protein loadings for the same set of model parameters. In this work, we have developed a multi-component colloidal isotherm within the well-established general rate model for a mAb on a Fractogel SO3- resin. The colloidal isotherm addresses the protein-surface and protein-protein interaction parameters independently, which makes their estimation straightforward. Protein-protein exclusion and surface diffusion are limiting factors for the mAb to reach saturated load levels on the resin. Both of these factors were successfully estimated using multi-component mAb breakthrough data at different ionic strength. The exclusion factor (protein-protein repulsion) and surface diffusion can lead to similar features in a breakthrough curve, which makes the independent estimation tricky. To overcome this we have identified patterns in the curve that allow protein-protein interaction and surface diffusion effects on the shape of the curve to be distinguished. We also used a trained artificial neural network to identify the parameters, which can autonomously evaluate the different features of the breakthrough curve. This was corroborated by a parameter search using an adaptive parallel tempering algorithm. The model developed was used to predict elution curves with high accuracy, wherein the target charge variants were identified as independent species or similar species lumped together. The model was also capable of predicting peak shape in a wide variety of gradient and step elution experiments at different flow rates and protein loads. The model was also used to propose the best method to separate charge variant and to optimize load conditions that approach the static capacity of the resin. The in silico methodology used in this work for parameter estimation along with minimum experimental data can help the industry to adopt model-based optimization and control of preparative cation-exchange chromatography with the highest accuracy.

Multi-component model for prediction of elution of monoclonal antibody in ion exchange chromatography: Parameter estimation and peak shape analysis

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.