Application of online MVDA during process development of an antibody producing CHO cell line with batch to batch variability

In this work we present an application of multivariate data analysis (MVDA) online process monitoring and control to aid process development of an antibody producing CHO cell line which exhibits batch to batch variability. For this program, a fed-batch production process was first developed in which the amount of feed was determined by offline metabolite measurements. With this process, batch to batch variability was observed with two fold differences in product titer at harvest. By implementing a fixed feeding schedule based on historic process performance rather than feedback control, the variability in harvest titer was mitigated but the variability in other cell culture metrics persisted. Spent media analysis identified depletion of key metabolites for fast growing cultures. Online process monitoring using MVDA was implemented and demonstrated the ability to successfully describe the variability in cell culture performance. Based on these findings, a multivariate feedback control scheme was trained and implemented to control the timing and amount of feed based on multiple on-line and off-line measurements. This method was successfully demonstrated to reduce titer variability compared to the original process. Additionally, spent media analysis revealed the new MVDA based control scheme maintained residual levels of key metabolites which were found to be depleted in the previous control scheme.