These themes resonate with ADDoPT's mission to help the pharmaceutical industry define a system for top-down, knowledge-driven Digital Design and Control for drug products and their manufacturing processes. It is therefore good to report that the project will be well represented at this high-profile international gathering of those working on and thinking about crystallization in the industrial context.
A presentation arising from collaboration between ADDoPT partners GSK and PSE will describe work addressing the recognised limitations of the statistical / Design of Experiments (DoE) approach generally used to develop process understanding and a design space in crystallization. Whilst this approach describes how process parameters affect the critical quality attributes (CQAs) of the manufactured product, it does not answer the more fundamental question of why they affect the CQAs, and so such models may have poor predictive capabilities. In an alternative approach, crystallization processes can be described via mechanistic modelling, which uses kinetic expressions to describe dynamic changes in the system. This can offer a more fundamental understanding of the system, and develop models of greater predictive power. George Taylor of GSK will be reporting on the successful application of PSE's mechanistic modelling tools to understand and predict the crystallization of a late stage development product, where complex agglomeration behaviour had otherwise made statistical model based predictions of particle size distribution very difficult.
The theme of the advantageous use of mechanistic modelling with be further developed by Yunes Salman from the CP3 Centre for Doctoral Training at the University of Leeds. His presentation entitled "Application of mechanistic models for the online control of crystallisation processes" is the result of tripartite collaboration between Leeds, PSE and Perceptive Engineering, and will cover exciting work to harness mechanistic modelling alongside a Model Predictive Control (MPC) approach to describe, model and control the seeded batch cooling crystallisation process of L-glutamic acid from aqueous solutions.
MPC is an established industrial technology but has only recently been applied in the pharmaceutical industry, where it has been used successfully in batch and continuous crystallisation processes to optimise control of super-saturation, delivering tight control of final particle properties at various scales, better yields, and more consistent product. To date however, MPC applications have characterised the crystallisation process using statistical models and data-driven techniques, and this requires experimental time, money and effort: for batch systems, the product generated during these tests is typically discarded, and furthermore, a subset of the tests must be repeated during scale up as key results (the limits of the meta-stable zone) are both product and process dependent. Combining information from validated mechanistic models into the MPC system promises to reduce experimental testing time during initial development and scale-up.
In presenting the current work, Yunes will outline the application of PSE's advanced process modelling tool, gCRYSTAL to describe the seeded batch cooling crystallization process of L-glutamic acid from aqueous solutions and lab and pilot scale, and integration of the validated mechanistic model with Perceptive Engineering's PharmaMV Advanced Process Control system. With this approach, the validated mechanistic model was utilised to drive the successive control steps to achieve a target product crystal size distributions (CSD), defined by the D10, D50 and D90 of the final product CSDs.
Finally, Caiyun Ma, Senior Research Fellow at the School of Chemical and Process Engineering, University of Leeds, will be presenting a paper entitled Morphological Population Balance (MPB) modelling for simulating crystal size and shape evolution in pharmaceutical crystallization processes. Based on crystal morphology, MPB identifies the individual faces and defines their normal distances to the crystal centre as MPB dimensions. By solving the formed multiple dimensional MPB equations over time, the evolution of individual crystal faces and their growth can be obtained as a function of system operating conditions and processing time. Combining this with crystal morphology information re-constructs the shape distribution of all crystals over time, generating more accurate crystal size distributions, and hence the evolution of size/shape distributions. CaiYun will use results for α-para amino benzoic acid (α-pABA), crystallized from ethanol, to illustrate the ability of MPB to provide a true simulation, optimisation and control tool for the digital design of crystallization processes in order to manufacture precise crystals of desired size and shape distribution..
These presentations will be complemented by several research poster presentations from researchers working on various aspects of crystallization related processes, primarily within ADDoPT WP4, and the broader conference programme also includes several other presentations highly relevant to the industrial uptake of digital design. We hope to see you in Dublin to tell you more about the excellent progress being made.