(532c) Model Assisted Development for Downstream Processing of Therapeutic Proteins

In the last decade, many model-based approaches describing the development, robustness or other aspects of bioprocess unit operations have been developed, applied, and presented. These tools, relying on statistical or mechanistic principles, tend to concentrate on certain facets of single steps, and do not necessarily represent the reality of the multi-step bioprocess. Over the last year, we have focused on combining many of these models to enable process development for the entirety of the downstream process.

In this talk, we will discuss the creation and application of a new model assisted development strategy for the downstream process. By understanding the chemical properties of the different buffer solutions, as well as the interactions between proteins and chromatographic surfaces, we can predict, and with minimal experimental work, confirm the performance of protein A and ion exchange chromatography steps. These tools are presented along with design and validation experiments. In the second part of our talk, we describe the combination of these models into a single entity. This allows us to fully reproduce and optimize the multi-step downstream process as a whole and in silico. The robustness and sensitivity to input changes are then analyzed in terms of product quality, productivities, and process costs.