High-throughput microscale chromatography for process characterization and control strategy development in biologics manufacturing

Biopharmaceutical manufacturers rely on high throughput, microscale models to accelerate bioprocess development and deepen process understanding. These tools enable more efficient process development in an increasingly demanding biopharmaceutical landscape that includes various modalities such as monoclonal antibodies, bispecific antibodies, antibody drug conjugates, among others. Technical and organizational hurdles have, however, continued to restrain use of these microscale models for late-stage process characterization and validation due to a perceived risk accompanying a change from well-established scale down models. Of relevance to this work, a rigorous body of literature supporting microscale models (such as Robocolumn® models and methods) has reached a threshold to strongly support more widespread adoption. This work provides a case study to more clearly detail the practical application of Robocolumn scale-down models for implementation for process characterization and validation activities. Deeper insight into study execution and data analysis is shared to provide increased confidence in the development of robust process control strategies based off Robocolumn data. Process characterization for a bispecific antibody was performed in replicate on both traditional bench-scale chromatography models and microscale Robocolumn models, followed by a rigorous statistical comparison of the predictive capability of the methods for manufacturing scale performance. Three different chromatography steps were characterized and evaluated for their impact on seven process performance and product quality attributes. Performance of Robocolumn scale down models was found to be in excellent agreement with manufacturing scale at target operating conditions and were aligned with bench-scale outcomes from multivariate studies. Finally, process parameter classifications derived from separate scale-down models were consistent, indicating that either model could be used to develop a robust process control strategy.