(580c) Application of Multivariate Data Analysis to Cell Culture Development

Application of Multivariate data analysis to Cell Culture Development
John S. Bowers, Senior Principal Scientist, BPD Balrina Gupta, Associate Principal Scientist, BPD T. Craig Seamans, Principal Scientist, BPD
Louis Obando, Principal Scientist, Chemistry, MMD PAT.
A large amount of data on a number of variables are collected to understand the behavior CHO cell culture processes, like the process for the manufacture of monoclonal antibodies. In the early development environment, statistical analysis based on set points, averages, or visual review of trends, have typically been used to understand results. This approach does not fully utilize the hundreds of data points taken per batch. In order to gain better understanding of a monoclonal antibody cell culture process, Multivariate data analysis (MDVA), using SIMCA from Umetrics, has been applied to the initial process ranging studies performed in BioProcess Development in New Jersey.
MDVA evaluations on control runs confirmed process robustness. Variations in parameters during control runs had weak effects on performance and quality attributes. Evaluations on DOE experiments using pH, temperature, feed timing and amount, and elevated pCO2 as the perturbed parameters confirmed conclusions from average based analysis, but MDVA highlighted when the variations on parameters were most important. Conditions when most of the antibody was produced, during the production phase, controlled quality far more than conditions in the cell growth phase. Conversely, variations in the growth phase impacted performance more. Short transients, less than 1 day, of operating parameters had very little impact on quality. MDVA analysis was better at describing continuous behavior, such as cell growth and death during the batch, than average measurements since the entire time based behavior is included in the analysis.