Mechanistic Modeling: The core technology behind smart downstream process development

Mechanistic Modeling: The core technology behind smart downstream process development

Mechanistic Modeling: The core technology behind smart downstream process development

Karlsruhe, November 2019. For the fifth time, the international conference devoted to high-throughput process development (HTPD) gathered the biopharma community in an inspiring environment. This time "Smart PD" was officially added to the title of the event. The community showcased major advances, remarkably many of them based on mechanistic modeling:

  • Michael Chinn presented Genentech’s workflow for model calibration, relying only on robotic experiments.
  • Tobias Hahn from GoSilico showed that it is possible to predict column runs from filter plates.
  • Peter Hagwall from GE addressed the robustness of processes in the light of QbD.
  • The award for the best poster went to a case study by Felix Wittkopp from Roche pRED, supported by ChromX.

HTPD and mechanistic modeling

– alternatives or perfect match? The perfect interplay of high-throughput experiments and mechanistic modeling is not easy to achieve, but several contributions showed that the effort is worth it. Mechanistic chromatography modeling relies on understanding the system dynamics from very few experiments, each fulfilling a certain purpose, e.g. characterizing the fluid dynamics or the adsorption. High-throughput experiments that change process parameters in a narrow range are often not well-suited to determine the unknown model parameters with high certainty and build a predictive model. An ideal type of experiment would be a long gradient elution that, using robotic columns, can only be approximated with a pipetting robot.

GoSilico had shown that it is possible to use RoboColumn experiments for model calibration at a previous HTPD Meeting, but a long step-wise gradient was needed to generate enough UV data for curve fitting. This year, Genentech’s Michael Chinn presented a workflow relying only on robust step elutions. Instead of UV data, Genentech used only attributes of the elution fraction, such as product yield and impurity levels to calibrate the model, circumventing the above-mentioned problem arising from the availability of only few UV data points.

Optimum use of Kp screenings

Apart from RoboColumns, a widely used HT format is filter plates. These are often used in early stage process development to find appropriate elution conditions that separate the product from impurities by measuring the partition coefficient Kp. After having identified a promising condition, all these dozens of measurements remain unused. This was the starting point of a study conducted by Boehringer Ingelheim and GoSilico to interpret Kp screenings mechanistically. In fact, Tobias Hahn could show in his presentation that it is possible to predict column runs from filter-plate experiments and get an impression of the process design space before doing any further lab-scale experiments.

GE Healthcare addresses the robustness of processes

Miniaturization is an essential part of HTPD and naturally comes with lots of advantages. But working with smaller chromatography columns has the side effect that only very few resin lots are encountered during process development, in the extreme case only a single one. Peter Hagwall from GE Healthcare discussed in his talk on “QbD and smart PD join forces” that much more robust processes can be developed through better understanding the impact of material attributes, for example the resin capacity.

To this end, GE Healthcare has established small batch production processes for its key materials on the market that are capable of producing average but also extreme cases within the specified attribute range. At GoSilico, we are looking forward to including experiments done with different resin lots to our workflows for model-based process characterization. Resin capacity is almost always identified as a potentially critical material attribute requiring special attention.

Poster award for a case study supported by ChromX

Finally, we want to highlight the work that received this year’s award for the best poster presentation. The team around Felix Wittkopp from Roche pRED presented a case study on early stage process development for a complex new molecule format.

Especially in such early stage industrial projects, the sample complexity is high and the available sample amount very limited. We are delighted that the pRED team used our software ChromX for their work and was able to fulfill their project goals with a perfect combination of experimental design, simulation software and optimized workflows. We congratulate them on this well-deserved win.

The next HTPD Meeting will be held in autumn 2021. For those who cannot wait and are interested in the chromatography part of it, save the date for the upcoming Chromatography Modeling Days in 2020, May 18-20th .