Model-assisted approaches for continuous chromatography: Current situation and challenges

Abstract

Continuous bioprocessing is a promising trend in biopharmaceutical production, and multi-column continuous chromatography shows advantages of high productivity, high resin capacity utilization, small footprint, low buffer consumption and less waste. Due to the complexity and dynamic nature of continuous processing, traditional experiment-based approaches are often time-consuming and inefficient. In this review, model-assisted approaches were focused and their applications in continuous chromatography process development, validation and control were discussed. Chromatographic models are useful in describing particular process performances of continuous capture and polishing with multi-column chromatography. Model-assisted tools showed powerful ability in evaluating multiple operating parameters and identifying optimal points over the entire design space. The residence time distribution models, model-assisted process analytical technologies and model-predictive control strategies were also developed to reveal the propagation of disturbances, enhance real time monitor and achieve adaptive control to match the transient disturbances and deviations of continuous processes. Moreover, artificial neural networks and machine learning concepts were integrated into modeling approaches to improve data treatment. In general, further development in research and applications of model-assisted approaches for continuous chromatography are needed urgently to support the continuous manufacturing.