Data-Driven Analysis of Monoclonal Antibody Charge Variants in Perfusion Cultures

Charge variant profiles constitute a critical quality attribute in monoclonal antibody (mAb) production. Controlling charge variant profiles generated in upstream processes can help reduce the burden on downstream separation units. This work investigates the role of different process parameters, such as nutrient levels and culture duration on controlling the generated charge variant profiles in perfusion cultivation. A data-driven model was developed to predict charge variant concentrations at different operating conditions. The model was developed using partial least squares (PLS) based on a set of four 2 L scale experiments in perfusion mode culture. A global sensitivity analysis was conducted to determine the most influential parameters. Probability maps were generated in two case studies to visualize the impact of variations in the identified critical process parameters on the resulting charge variant profile. The results showed the strong influence of culture duration and glutamate concentrations on the generated charge variant concentrations. The probability maps identified trade-offs between the conditions required to maintain the acidic and basic charge variant concentrations within the thresholds set in the case study. Alternative conditions were suggested to prolong the potential culture duration without violating the set concentration thresholds. The model-based analysis could isolate and investigate the impact of nutrient-related parameters on the charge variant profiles. This approach can be extended to other influential process parameters, leading to better process understanding and the creation of comprehensive operating design spaces for optimal production processes.