(435a) Reaction Kinetic Model Application to Speed up Development
AIChE Annual Meeting
2020
2020 Virtual AIChE Annual Meeting
Pharmaceutical Discovery, Development and Manufacturing Forum
Predictive Scale-Up/Scale-Down for Production of Pharmaceuticals and Biopharmaceuticals I
Monday, November 16, 2020 - 8:00am to 8:15am
To initially confirm the kinetic model provided by UCB (both the UCB and Hovione used Dynochem for modelling) it was necessary to execute experimental work that can provide enough variability regarding the reaction profiles. At Hovione, we have compiled a workflow/set of experimental conditions that is suitable to a wide range of reactions types and can capture most critical physical and chemical transformation rates during a reaction (e.g. product converted into impurities, reagents reacting with other species (aside from starting raw material), mass transfer limitations). This set of experiments was used to confirm the kinetic model and afterwards we were able to explore the design space using a design of experiments approach.
The DoE was a Plackett-Burman design with 4 targeted parameters namely reagent equivalents (upper and lower boundaries set by economic and quality targets), solvent volumes, reaction temperature and acid equivalents. In order to minimize for scale-up issues, the DoE used heating rates similar to the ones used at production scale. Additionally, a maximum reaction time of 33h was used, which enabled to reach a 99% yield for all experimental conditions of the DoE. The kinetic model was able to accurately describe all the experiments carried out during the DoE and was used to define the critical variables proven acceptable ranges (PAR) and normal operating ranges (PAR). The isolated product showed a 99.0 % w/w assay as demonstration trial in the lab, while at a >100 kg scale it was possible to achieve a 99.7 % w/w assay.
It was proven that by resorting to a mechanistic modelling approach and a reduced number of experiments we were able to save resources (time and materials) and generate more knowledge than the traditional approach and have a more straightforward way to reach production scale with Right-First-Time batches.