While high-resolution design space screening based on Design-of-Experiments (DoE) works for most of the established molecular scaffolds, it requires a high sample amount, elaborate offline analytics, and is restricted to the calibrated experimental range. Mechanistic modeling of biopharmaceutical processes has been in the focus of academic research for a long time. Nevertheless, industry failed to adapt these technologies, mainly due to the complexity of the underlying technology. At current time, mechanistic modeling building is mostly performed by highly skilled engineers with a scientific background in both biotechnology and mathematical modeling. This contribution presents an industrial chromatography development challenge for which the traditional DoE-based process development carried out by an industrial partner failed to identify an economically feasible process. In order to prevent a delay of the biologically promising antibody, a mechanistic model was built for a cation-exchange step based on only three experimental chromatograms. To quickly arrive at a calibrated model for in silico optimization, the process of model building was automated to a large extend. The log book of the lab scale chromatography system was parsed to extract all relevant events. Conductivity, pH, and temperature traces were used to deduce molar salt concentration from the given dominant salt species. Thereafter, all UV signals were treated with a Savitsky–Golay smoother to remove noise without losing information. Automatic peak detection and deconvolution was performed based on polynomial-exponential modified Gaussian peaks. Using well-established correlations between elution behavior and isotherm parameters, the search space for molecule-specific parameters could be reduced to a minimum. Effectively the process of model calibration succeeded with minimal interaction. Afterwards, a much larger range of different process options compared to the initial DoE was evaluated by means of simulations. This in silico process development identified a feasible chromatographic process by varying the column aspect ratio and the load density in particular.

Towards automated model-based process development

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.