Process monitoring is defined as a continuous effort to evaluate manufacturing parameters across multiple lots to ensure commercial manufacturing processes are operating in-control within specifications. As it becomes increasingly important to reduce the cost of goods for cell culture processes, scientists and engineers require digital data-acquisition and aggregation systems to develop a robust process monitoring approach for the entire process. These automated data acquisition tools assist in rapidly contextualizing data over the course of a process lifecycle and can identify correlations across an entire process (cell culture and purification). In particular, advent of digital plant combined with development of informatics tools to generate, acquire, analyze and visualize large data sets for process monitoring has led to increased process understanding, reduced cycle time, and implementation of process improvements based on analysis performed with the tool.
We will present a case study where predictive models of production bioreactor and chromatography steps were used in tandem to identify sources of variability in overall process yield for a 25,000L fed-batch CHO process. From this analysis, bioreactor culture duration was identified to negatively impact glycosylation levels of the recombinant fusion protein produced from the culture. By implementing calibrated, data-driven modifications to the upstream manufacturing schedule, we were able to realize tighter control of upstream protein quality, optimal downstream process performance, and improved overall process yield.