The Laboratory of Process Analytical Technology (LPPAT) focuses on the implementation of PAT systems in innovative pharmaceutical production processes and therefore has always worked in close collaboration with the Laboratory of Pharmaceutical Technology of Prof. Dr. C. Vervaet and is part of the QbD and PAT Sciences Network.
Manufacturing processes of interest:
- Continuous wet granulation
- Production of solid oral dosage forms
- Continuous melt granulation
Implementation of PAT systems:
- The development and implementation of process analyzers in the process stream allowing real-time collection of critical process and (intermediate) product information.
- Data-analysis methods (chemometrics) allowing to extract useful information from the large datasets that process analyzers supply. Process analyzers are only valuable if they provide the desired information with sufficient accuracy. Being able to build accurate and robust models to reliably translate the data (e.g., obtained spectra) into process or product knowledge is crucial.
- Design of Experiments (DoE) to maximize the information content from experimental series while keeping the number of experiments low. As the process (step) endpoints and the intermediate or end product properties (e.g., product solid state, chemical properties, physical properties,…) are influenced by numerous process and formulation variables, appropriate experimental design approaches must be applied to find out which variables and variable interactions significantly influence processes and product properties.
- Statistical process control and visualization: a final aim of implementing PAT systems in pharmaceutical production processes is complete process control. Based on process knowledge and process models, the information obtained in real-time should be used for guiding the process to its desired state, possibly allowing real-time release. Early warnings should be given when a process is moving into an unwanted direction and the process models should allow to determine how process settings must be adapted by operators to lead the process to its desired state, thereby reducing batch rejection.
- Mechanistic modelling: empirical modelling is based on historical data and as such they are of limited use in new applications outside the experimental space studied. Apart from cause-and-effect between variables, not much else is required in terms of process knowledge. Mechanistic modelling is based on the fundamental understanding of the underlying physics and chemistry governing the behaviour of the process. Hence, mechanistic modelling does not require much data for model development, and hence is not subject to the idiosyncrasies in data. Mechanistic modelling forces to fundamentally and completely understand processes. The different steps of mechanistic modelling are mentioned in appendix. This new approach of modelling of pharmaceutical processes should allow making useful process simulations and process predictions.