(224c) A Computational Methodology for the Optimization of Mab Produced in Dissolved Oxygen Heterogeneous Bioreactors

Scale up of biopharmaceutical manufacturing is complex and time-intensive process, yet it is crucial for the large scale production of high-demand biopharmaceutical products. Scale up induced heterogeneities in physical parameters such as dissolved oxygen (DO) and pH can lead to performance losses, as optimal culture conditions cannot be maintained spatially across large scale, and extended mixing times in large scale bioreactors prolong the exposure to the suboptimal conditions.

We have developed a computational methodology to model the evolution of scale up related heterogeneities in the bioreactor and their impact on cell metabolism and biologics production. The methodology iterates between computational fluid dynamics (CFD) and a kinetic model of metabolism which has been functionalized to account for oscillations in key physical parameters in large scale bioreactors. We have applied this methodology for prediction of pH heterogeneities’ effects on GSK proprietary cell clones^1,2 as well as DO heterogeneities’ effects on a ‘typical’ Chinese hamster ovary (CHO) cell based on aggregated literature data³. Compelling new experimental data has become available in the literature for a particular CHO cell line affected strongly by DO heterogeneities.⁴ Using this data, we’ve adapted our computational methodology to predict CHO cell culture performance in bioreactors with DO heterogeneities. Machine learning has been applied to reduce the complexity of computational fluid dynamics model, allowing us to extend the model for optimization of biologics titer through optimal feeding strategies, media compositions and dissolved oxygen control strategies.

Fed batch bioreactor feeding strategy including frequency and media compositions for novel biologics often follow an existing platform process, with a common basal media and higher concentration feed media provided in particular volume boluses at a pre-determined frequency. Using our in silico model for scale up, we explore the effects of varying the feed media concentrations, feeding volumes, and feeding frequencies in DO heterogeneous bioreactors. Optimization is applied to select the feeding variables which maximize the predicted biologics production, improving upon standard platform approaches.

Similarly, we have explored the optimization of DO control methods for reducing DO heterogeneities impact on culture performance. DO is controlled through gas sparging and can be manipulated by the total gas sparge rate, oxygen concentration within the sparged gas, and impeller rotational speed. Each of these terms affect the properties of the DO heterogeneities present in the large bioreactor. Our model is used to optimize the DO control methodology which will maximize biologics production through reduced heterogeneities.

In this talk, we will explore relationships between oscillation characteristics and cell metabolism, make predictions for various bioreactor operating conditions and bioreactor designs, and modify culture conditions to optimize biologics production.

References:

1. Raudenbush, Reddy, Attfield, Kotidis, Kedge, Finka, Marchese, Yang, Talwar, Hu, Ierapetritou. AIChE Annual Meeting 2024.
2. Raudenbush, Reddy, Attfield, Kotidis, Kedge, Barodiya, Finka, Marchese, Yang, Talwar, Hu, Ierapetritou. AIChE Annual Meeting 2023.
3. Raudenbush, Ierapetritou. NAMF Mixing XXVIII 2024.
4. Gaugler, Hofmann, Schluter, Takors. Biotechnol Bioeng 2024.