Dynamic Optimization for Monoclonal Antibody Production

TL;DR

This paper develops a dynamic optimization approach for monoclonal antibody production in a perfusion reactor, significantly increasing yield and exploring optimal feeding strategies through an extended model.

Contribution

It introduces a general modeling methodology for simulation and optimization of mAb production, including a glucose inhibition term, and analyzes multiple optimal control solutions.

Findings

Optimal operation increases mAb yield by up to 52%.

Multiple optimal feeding trajectories exist.

Full glucose utilization can be achieved without reducing mAb production.

Abstract

This paper presents a dynamic optimization numerical case study for Monoclonal Antibody (mAb) production. The fermentation is conducted in a continuous perfusion reactor. We represent the existing model in terms of a general modeling methodology well-suited for simulation and optimization. The model consists of six ordinary differential equations (ODEs) for the non-constant volume and the five components in the reactor. We extend the model with a glucose inhibition term to make the model feasible for optimization case studies. We formulate an optimization problem in terms of an optimal control problem (OCP) and consider four different setups for optimization. Compared to the base case, the optimal operation of the perfusion reactor increases the mAb yield with 44% when samples are taken from the reactor and with 52% without sampling. Additionally, our results show that multiple optimal feeding trajectories exist and that full glucose utilization can be forced without loss of mAb formation.