Advanced Computational Tools To Enhance Continuous Monoclonal Antibody Production

Maria Papathanasiou, Imperial College London
Ana Quiroga Campano, Imperial College London
Athanasios Mantalaris, Imperial College London
Efstratios Pistikopoulos, Texas A&M


Leading pharmaceutical companies invest high percentage of their revenue in the improvement of existing technologies used for the production of monoclonal antibodies (mAbs). Recently, there has been a paradigm shift towards the development of continuous/quasi-continuous purification operations, aiming to reduce capital and operational costs [1]. At the moment, however, there are no standardized methods and/or tools that can be used for global control and monitoring of integrated processes.

Mathematical models and advanced computational tools can be the key for the development of robust, integrated processes, as they can provide valuable insight in the process dynamics and ensure optimal operation [2]. However, such processes are usually characterized by complex mathematical models and periodic operation profiles that result into computationally expensive solutions and challenge the development of global control methods and tools. In this work, we are presenting a novel approach for the development of advanced controllers towards the intensification of mAb production, considering the fed-batch culturing of GS-NS0 cells and the semi-continuous Multicolumn Countercurrent Solvent Gradient Purification (MCSGP) process [3]. The controller development is realized via the application of a generic framework for the development of advanced control strategies (PAROC) [4] that involves: (i) development of a high-fidelity process model, (ii) approximation of the complex, process model, (iii) design of the multi-parametric controller, (iv) ‘closed-loop’, in-silico validation of the controller against the process model. The development of the control policies is based on multi-parametric Model Predictive Control (mp-MPC) policies that reduce the online, computational force of the controller by deriving the control inputs as a set of explicit functions of the system states and can be implemented on embedded devices [5]. One of the main advantages of the proposed framework is the ability to test the controllers ‘in-silico’, against the high-fidelity process model and evaluate their performance before operating them online. The results from this study indicate that optimal operation, under maximum purity and productivity yield can be ensured with the development of advanced computational tools. The control policies are applied both in the upstream and the downstream processing; yielding therefore a fertile ground towards the development of a global control strategy that can ensure continuous operation.