(327c) Modeling and Dynamic Simulation of Cultivation in Monoclonal Antibody Production Considering Cell Metabolic Shifts

The market for biopharmaceuticals has been rapidly growing, particularly for monoclonal antibody (mAb) drugs. Chinese hamster ovary (CHO) cells are the most frequently used host cells for mAb production processes. Cell cultivation is known to be one of the most expensive and time-consuming steps. To increase the efficiency of mAb production, developments have been made to cell cultivation, including host cell lines, control strategies, and operating modes. Also, in addition to productivity, product quality is also an important indicator. The generation of process-related impurities (e.g., host cell proteins (HCPs)) from dead cells affects product quality [1] and increases the load on downstream purification units. Therefore, considering quality earlier in the design of cultivation processes is necessary. However, cell metabolic shifts, such as from lactate production to consumption are challenging to follow with current fundamental kinetic models [2], particularly for high-performance cell lines. This is because the models usually assume constant coefficients (e.g., specific lactate production rate [mmol cell^–1 h^–1]). In this work, to overcome this challenge, hybrid modeling [3], combining data-driven and fundamental kinetic models, was used to obtain insights into cell cultivation phenomena and to develop new representative and robust models.

For the data analysis and model development, experimental data using a newly developed high-performance cell line, CHO-MK cell line, were obtained from a pilot-scale research facility. The approach consists of three steps. First, variations in the online measurements (e.g., dissolved oxygen, pH) were used to gain insights into the parameters to be varied in the kinetic model. Principal components analysis followed by regression (PCR) [4] and data clustering [5] were used for the analysis of the gap between process measurements and the calculated metabolite profiles using the available kinetic models. The analysis was used to identify the key factors correlated with changes in cell phases and metabolic behavior. The clustering and PCR enabled the identification of the important factors for differentiating cell phases, as well as the resolution of the gap observed between experimental and calculation results from previous kinetic models. Second, for each identified phase, dynamic models were developed to follow cell metabolic shifts and replace the constants used in the kinetic model. This study focused on the specific lactate production/consumption rate in particular as a modeling target. Third, the new dynamic parameter models were integrated into the overall mass balance equations in the fundamental kinetic models. The new models were then used to evaluate the impacts of changes in operating strategies on production efficiency and product quality represented by the final concentrations of mAb and HCP, respectively.

The modeling results showed that the approach enabled highly accurate modeling and simultaneous simulation of nine factors, including products and impurities associated with the newly developed cell line. Process evaluation results showed that, for the explored range, a compromise must be made to maintain the productivity and quality targets. The developed model provides more precise time and cost estimations for process design purposes and can be used for early screening of process alternatives. The evaluation results thus contribute to the robust design of integrated processes, including downstream purification units. The model would also help creating better control models and strategies for process operation.

[1] Goey, C. H., Alhuthali, S., & Kontoravdi, C. (2018). Host Cell Protein Removal from Biopharmaceutical Preparations: Towards the Implementation of Quality by Design. Biotechnology Advances, 36 (4), 1223–1237. https://doi.org/10.1016/j.biotechadv.2018.03.021

[2] Badr, S., Okamura, K., Takahashi, N., Ubbenjans, V., Shirahata, H., & Sugiyama, H. (2021). Integrated Design of Biopharmaceutical Manufacturing Processes: Operation Modes and Process Configurations for Monoclonal Antibody Production. Computers & Chemical Engineering, 153, 107422. https://doi.org/10.1016/j.compchemeng.2021.107422

[3] Narayanan, H., Sokolov, M., & Morbidelli, M. (2019). A New Generation of Predictive Models: The Added Value of Hybrid Models for Manufacturing Processes of Therapeutic Proteins. Biotechnology and Bioengineering, 116(10), 2540–2549. https://doi.org/10.1002/bit.27097

[4] Okamura, K., Badr, S., Murakami, S., Sugiyama, H. (2022) Hybrid Modeling of CHO Cell Cultivation in Monoclonal Antibody Production with an Impurity Generation Module. Industrial & Engineering Chemistry Research, 61, 14898–14909. https://doi.org/10.1021/acs.iecr.2c00736

[5] Badr, S., Oishi, K., Okamura, K., Murakami, S., Sugiyama, H. Hybrid modelling and data-driven parameterization of monoclonal antibody cultivation processes: Shifts in cell metabolic behavior. Computer Aided Chemical Engineering, accepted