Multivariate analysis of cell culture bioprocess data--lactate consumption as process indicator

Area of Science:

• Biotechnology
• Bioprocess Engineering
• Data Science in Manufacturing

Background:

• Multivariate analysis of cell culture bioprocess data can reveal hidden characteristics and improve process understanding.
• Analyzing time-series data from large-scale manufacturing offers insights into factors influencing process performance.

Purpose of the Study:

• To investigate the predictive power of multivariate methods on bioprocess outcomes.
• To identify key process parameters influencing final antibody and lactate concentrations.
• To explore opportunities for metabolic intervention to enhance bioprocess productivity and robustness.

Main Methods:

• Kernel-based support vector regression (SVR) and partial least square regression (PLSR) were applied to time-series data.
• 134 process parameters were analyzed from inoculum train and production bioreactors across 243 runs.
• Prediction models were developed using early-stage and inoculum train data.

Main Results:

• Accurate prediction of final antibody titer and lactate levels was achieved using early production-scale data.
• Inoculum train data provided lower prediction accuracy but indicated significant influence of culture history.
• Lactate metabolism and cell viability parameters were most critical for prediction accuracy.
• Lactate consumption emerged as a key factor for high productivity, independent of glucose/lactate levels.

Conclusions:

• Pattern recognition techniques, such as SVR and PLSR, are valuable Process Analytical Technologies (PAT).
• Early-stage bioprocess data can reliably predict final outcomes, enabling proactive process control.
• Understanding and intervening in lactate metabolism, particularly lactate consumption, offers a pathway to enhance productivity and robustness.
• Findings support the application of Quality by Design (QbD) principles for improved biomanufacturing processes.