Improving extracellular vesicles production through a Bayesian optimization-based experimental design.

Area of Science:

• Biotechnology
• Process Engineering
• Biopharmaceutical Manufacturing

Background:

• Optimizing biopharmaceutical production is challenging due to multiple objectives like yield, activity, and purity.
• Accelerated production with minimal resource use is critical for diverse biological drugs.
• Existing methods like Design of Experiments (DoE) and One-Factor-At-a-Time (OFAT) can be resource-intensive.

Purpose of the Study:

• To develop and apply a novel Multi-Objective Batch Bayesian Optimization (MOBBO) algorithm.
• To optimize the production of extracellular vesicles (EVs) from mesenchymal stem cells (MSCs).
• To simultaneously maximize vesicle-to-protein ratio, CD73 enzymatic activity, and minimize calregulin impurities.

Main Methods:

• Implementation of a Multi-Objective Batch Bayesian Optimization (MOBBO) algorithm.
• Application to a 3D cell culture system for mesenchymal stem cell (MSC)-derived extracellular vesicle (EV) production.
• Evaluation of four process parameters: microcarrier concentration, seeding density, centrifugation time, and impeller speed.

Main Results:

• Achieved optimal process parameters with only 32 experiments, comparable to or fewer than DoE and OFAT.
• Demonstrated adaptive sampling to exclude unfavorable regions, minimizing experimental requirements.
• Successfully balanced multiple objectives: maximizing vesicle-to-protein ratio and CD73 activity while minimizing calregulin impurities.

Conclusions:

• MOBBO offers an efficient strategy for optimizing complex bioprocesses with multiple objectives.
• The algorithm significantly reduces the number of experiments needed for process optimization.
• The findings support applications in Quality by Design, process monitoring, control, and scale-up for biopharmaceuticals.