A two-phase approach optimizing productivity for a mAb-producing CHO cell culture process using dynamic response surface methodology models

Fed-batch cell culture processes are common in the biomanufacturing industry due to their ease of development and simplicity of execution in the cGMP environment. One challenge of fed-batch operation is that the physiochemical conditions within the bioreactor do not remain constant throughout a batch but change over time as the viable cell density changes. It is well accepted that fed-batch cultures are multi-phasic, divided into growth, stationary, and death phases. Yet, using empirical modeling methods such as response surface models (RSMs) for optimization is still common. While RSMs can identify local performance maxima, they are limited by the time-invariant factors (input parameters) they use. These factors do not change throughout the run. Dynamic RSM (DRSM) models predict the time-dependent impact of the process inputs allowing for the identification of optimized process operations that change with time. Starting with the data sets generated via a traditional Design of Experiments (DoE) design, the DRSM model successfully optimized the peak cell count during the growth phase of culture and then identified a second set of process parameters that optimized productivity during the stationary phase. The resulting process achieved a harvest titer improvement of 28% relative to the base process condition.

Automated summary

Fed-batch cell culture processes are commonly used in biomanufacturing due to their simplicity and applicability in cGMP environments. However, the challenge lies in the changing physiochemical conditions within the bioreactor as the cell density changes. Traditional response surface models (RSMs) are commonly used for optimization but are limited by their use of time-invariant factors. Dynamic RSM (DRSM) models can predict the time-dependent impact of process inputs, allowing for optimization of process operations that change over time.