Model-based optimization for an industrial cation exchange step to purify antibodies
Antibody purification is commonly based on platform processes. This guarantees both fast process development and a reduction in time-to-market. At the same time, platform based approaches do not exploit the full potential of optimization. To address this issue, modeling tools are attracting more and more interest.
Unknown feed compositions complicate the optimization. Industrial downstream processes typically show very heterogeneous mixtures like antibody compounds or entirely unknown feed compositions. Both the complexity and uncertainties greatly impede modeling approaches, since they increase the number of unknown model parameters.
Inverse method to determine model parameters. With steadily increasing time-to-market pressure, screening a large parameter space is not practicable. In the present case study, unknown model parameters could successfully be estimated from few column experiments and fraction analyses. By means of an inverse peak fitting approach, ChromX systematically altered the model parameters until the model prediction matched the measured chromatograms. No prior knowledge of molar feed concentrations was necessary, but UV signals could be utilized instead. The inverse method allowed an arbitrary number of unknown parameters to be estimated at the same time. In this way, even crude feedstocks and complex compounds could be identified and incorporated into the model.
Time and resource-efficient process optimization. Once the model was calibrated from few UV-data, it could be utilized for further process optimization or characterization. In the present case study, a model-based optimization was then applied to identify optimal process conditions.