(665b) Bi-Directional Control of Glycosylation of Antibodies

Monitoring and control of critical quality attributes (CQAs) is important for biologics manufacturing. CQAs need to be controlled within desirable ranges in order to manufacture a clinically effective product. These ranges can be particularly tight for biosimilars, because similarity to the innovator molecule must be demonstrated. To address these challenges, new flexible methods of monitoring and modulating CQAs are required. Multivariate model predictive control (MPC) can be used to target CQAs within desirable ranges to ensure the delivery of high quality biologic products. In this study, MPC was applied to control the CQA of interest (CQA1) of a recombinant antibody by using two media additives and a cell culture parameter as modulation levers. A partial factorial experiment (n=25) was first conducted in bench-top bioreactors to understand the impact of these levers on CQA1. Both media additives were able to modulate CQA1 but in opposite directions. The cell culture parameter alone was able to modulate CQA1 in both directions. After these levers were identified and characterized, an empirical multivariate model was created to predict and control CQA1 during the cell culture process. Next, the robustness of this control strategy was tested in 3 L bench-top and 100 L pilot-scale bioreactors. For the test, CQA1 was purposefully deviated in either direction (high or low) away from the target using the cell culture parameter as a lever. Our results indicate that the model effectively predicted the amount of media additive required to bring CQA1 back to the target. The target CQA1 range was met for both scales at the end of the cell culture. Thus, the use of media additives to achieve bi-directional model predictive control of CQA1 has been achieved. This study demonstrates that MPC is a robust tool to modulate CQAs within desired targets for cell culture processes.