Rapid development of a transferable Raman model using high‐throughput cell culture for monitoring monoclonal antibody titer

Abstract Monoclonal antibody (mAb) titer monitoring is a key capability during process development and optimization, enabling timely decision making and increasing the speed of development. Raman spectroscopy is a prominent process analytical technology (PAT), but resource‐efficient calibration strategies for the development of transferable models are limited. This work demonstrates the development and successful transfer of a calibration model for monoclonal antibody concentration between two different cell lines with varied metabolic profiles expressing different antibodies. The root mean square error of prediction (RMSEP) for titer in the source cell line (0.266 g L−1) was comparable to that of the target cell line (0.325 g L−1). The transferable model was achieved by conducting a spiking study in a high‐throughput parallel bioreactor system. Different experimental approaches with models trained on spiked versus native samples were compared. This analysis revealed that model transferability was influenced by the degree of correlation of lactate with antibody titer in the source process, emphasizing the importance of process knowledge in the development of Raman calibration models. Overall, the study presents evidence of the feasibility of transferable titer models, marking a significant advancement in process monitoring capabilities for high‐throughput cell culture as well as introducing a generic methodology for calibration of transferable models in the context of upstream bioprocessing.