2024 AIChE Annual Meeting

(485e) Application of a Computational Methodology for the Prediction of Mab Produced By CHO Cell Clones in pH Heterogeneous Bioreactors

Authors

Venkatarama Reddy, J., University of Delaware
Attfield, L., GlaxoSmithKline (GSK)
Kedge, O., University of Delaware
Kotidis, P., Imperial College London
Finka, G., GSK Medicines Research Centre
Marchese, G., GlaxoSmithKline (GSK)
Yang, O., GlaxoSmithKline (GSK)
Talwar, S., Duquesne University
Hu, Z., GlaxoSmithKline (GSK)
Ierapetritou, M., University of Delaware
Monoclonal antibodies (mAbs) have become ubiquitous within the biopharmaceutical space. CHO cell-based monoclonal antibody production be challenging to scale up, and scale up optimization can begin in the early stages of cell line development. Previously, we have presented on a computational methodology for the prediction of monoclonal antibodies in large scale bioreactors, where heterogeneities in important physical and chemical parameters can be expected1.2. This methodology links computational fluid dynamics (CFD) modelling with kinetic models of metabolism and of a critical quality attribute called glycosylation to capture the evolution of heterogeneities while capturing their impact on cell metabolism, mAb production, and product quality.

Furthering this work, collaborators at GlaxoSmithKline (GSK) have conducted scale down simulator experiments3 at the 250mL scale for multiple CHO cell lines, exposing them to oscillations in pH and dissolved oxygen. Data such as viable cell density, mAb titer, glucose, amino acids, lactate, and glycosylation patterns are collected in these experiments to regress the metabolic and glycosylation model parameters. By running scale down simulator experiments, we also gain information about the cell culture shifts resulting from the oscillations. Attributes which characterize the patterns of physiochemical oscillations in the experiments are incorporated into the models to account for the impact of the oscillations on metabolism and glycosylation.

Utilizing the modified kinetic models and a CFD model of a production scale bioreactor of interest, we can use the computational methodology described in previously1,2 to iterate between CFD predictions of oscillation patterns and models of metabolism and glycosylation to predict culture dynamics, mAb production, and glycosylation patterns. Furthermore, production and product quality at the large scale can be optimized through adjusting culture conditions, specifically impeller speed, sparge composition and sparge rate, and feeding strategies.

References

  1. Raudenbush, Reddy, Attfield, Kotidis, Kedge, Barodiya, Finka, Marchese, Yang, Talwar, Hu, Ierapetritou. AIChE Annual Meeting 2023.
  2. Raudenbush, Thomas, Papoutsakis, Ierapetritou. AIChE Annual Meeting 2022.
  3. Zakrzewski, Lee, Lye. Biotechnology Progress 2022. DOI: 10.1002/btpr.3264