(518e) Nonlinear Model Predictive Control of the Monoclonal Antibody Glycosylation in Cell Culture

Glycosylation, the enzymatic addition of sugar molecules to organic compounds, plays an important role in the effectiveness of monoclonal antibodies (mAb), the leading entities in the biopharmaceutical sector [1]. Consequently, the pattern of glycosylation is deemed a crucial quality attribute (CQA), necessitating precise control within a defined range to maintain the quality of the product [2]. Given the intricate nature of mAb glycosylation, employing nonlinear model predictive control (NMPC) is advantageous for ensuring consistent product quality, aligning with the quality-by-design (QbD) framework [3].

The successful application of NMPC to mAb glcyosylation relies on the availability of a high-quality glycosylation model, and a robust and efficient optimization technique to solve the underlying dynamic optimization problem. Recently, a multiscale glycosylation model incorporating mechanisms from intracellular level to bioreactor level was shown to have good predictability [4]. However, the glycosylation model consists of large-scale and stiff partial differential algebraic equations (PDAE), and dynamic optimization based on such model is intractable at present, barring its use in NMPC. Therefore, a novel algorithm is needed to solve the PDAE-based dynamic optimization problem reliably and efficiently.

In this work, we proposed an algorithm that decoupled the PDAE for the multiscale model into a series of differential algebraic equation (DAE) systems after introducing several mild assumptions. This reduced the computational time for the PDAE solution by one to two orders of magnitude compared to solving the PDAE using the finite difference method. Moreover, detailed analysis and computational results indicated that the solution from the decomposition algorithm had negligible accuracy loss. Based on the decomposition-based PDAE solution method for the multiscale model, an NMPC algorithm with soft constraints were then developed and applied to the mAb production from Chinese hamster ovary (CHO) cells. The comparison between open-loop dynamic optimization and NMPC under different extents of model-plant mismatch and production requirements shows that NMPC can drive the process to generate more qualified products robustly, while open-loop dynamic optimization may cause serious constraint violation.

(1) Walsh, G. Biopharmaceutical Benchmarks 2018. Nat. Biotechnol. 2018, 36 (12), 1136–1145. https://doi.org/10.1038/nbt.4305.

(2) Batra, J.; Rathore, A. S. Glycosylation of Monoclonal Antibody Products: Current Status and Future Prospects. Biotechnol. Prog. 2016, 32 (5), 1091–1102. https://doi.org/10.1002/btpr.2366.

(3) Rathore, A. S.; Winkle, H. Quality by Design for Biopharmaceuticals. Nat. Biotechnol. 2009, 27 (1), 26–34. https://doi.org/10.1038/nbt0109-26.

(4) Kotidis, P.; Jedrzejewski, P.; Sou, S. N.; Sellick, C.; Polizzi, K.; del Val, I. J.; Kontoravdi, C. Model-Based Optimization of Antibody Galactosylation in CHO Cell Culture. Biotechnol. Bioeng. 2019, 116 (7), 1612–1626. https://doi.org/10.1002/bit.26960.