Multivariate data analysis of CHO cell line development data to identify cell line selection criteria

This paper presents a multivariate data analysis (MVDA) of GS-CHO cell line development (CLD) data to uncover new insights for decision-making in CLD cell line selection. Developing a monoclonal antibody (mAb) using Chinese Hamster Ovary (CHO) cells entails a resource intensive CLD process to select an optimal, clonally derived CHO cell line with consistent mAb productivity, viability, and quality. The aim was to identify process variables in the CLD data provide predictive power for cell line performance and understand the intercorrelations across the CLD stages. Early-CLD stages involve single-cell cloning (SCC) and clone expansion in static plates and shake flasks, analysed and triaged based on mAb productivity and cell growth. Late CLD assesses clones for manufacturing suitability and stable mAb productivity using the Ambr15TM. The MVDA revealed SCC screening data using the BeaconTM does not correlate with late CLD performance, whilst scale-up data from static plates and shake flasks do. High-performing cells were characterised by minimal cell growth, suggesting a potential novel criterion for cell line selection. Ammonium and lactate appeared to play a key role in stably highly productive cell lines, emphasizing the importance of monitoring these during CLD stages, with implications for early-stage monitoring as well. Early CLD data predicts late CLD cell line performance in productivity and cell growth but not stability (i.e. stable productivity over >70 generations).