Rapid advances in process data acquisition has enabled a large number of parameters to be monitored at any given time throughout the cell culture process. Modern large-scale cell culture facilities are fit with digital data-acquisition and aggregation systems that allow to scientists and engineers to analyze data. Although enormous amounts of data are collected through these systems in real-time, substantial effort is placed to extract meaningful information from the data. Recently, strategies have been developed to provide a more holistic view of the process, allowing analysts to extract useful process information that may not be captured via analysis of individual parameters. By unearthing these interactions, process improvement opportunities may arise that were not apparent prior to interrogating the data. Recently, multivariate analysis (MVA) has been used to observe and compare complex datasets for troubleshooting or characterization.
In this case study, a closed-loop process was utilized to implement process changes for increasing process yield and robustness. First, principles of MVA were applied to online and offline data from a 25,000L fed-batch CHO process to identify parameters that contribute to overall process variability. From this analysis, several parameters were identified as key contributors to variability in culture titer and glycosylation levels of the recombinant fusion protein produced from the culture. Results from this MVA were confirmed with appropriate experiments using a representative 5-L scale down model of the 25,000L bioreactor. Results from the scale-down model and MVA were used to implement at-scale process changes to reduce culture performance variability and increase overall process yield. Thus, the combined experimental and modeling approach enabled actionable changes to large-scale process, yielding increases in overall process yield.
See more of this Group/Topical: Pharmaceutical Discovery, Development and Manufacturing Forum