(299e) Quality by Design and Mechanistic Models of Pharmaceutical Unit Operations | AIChE

(299e) Quality by Design and Mechanistic Models of Pharmaceutical Unit Operations


Developing pharmaceutical product formulations in a timely manner and ensuring quality is a complex process. Within the spirit of the QbD initiative it requires a systematic, science based approach and collection and combination of information from various categories, including properties of drug substance and excipients, interactions between materials, unit operations, and equipment. Advances in measurement tools in combination with quantitative analysis techniques (e.g. statistical tools, empirical correlations or mechanistic models) enable engineers to design and scale-up pharmaceutical unit operations and processes with more confidence and reliability. Manufacture of most drugs employs different (usually batch) unit operations of the drug substance manufacture or the drug product manufacture which can be categorized as follows: A/ Unit operations of the drug substance manufacture (reactions, separations, crystallization of the active pharmaceutical ingredient, API) B/ Unit operations of the sterile liquid products manufacture (mixing, dissolution of powders, filtration, lyophilization) C/ Unit operations of the solid dosage products manufacture (milling, blending, conveying, granulation, tabletting, encapsulation) Mathematical modelling of unit operations of the group A is common in chemical engineering community and commercial software exists for batch reactors, separators or crystallizers. Mathematical modelling of unit operations of group B is less common. Commercial software is not available but general level of modeling and understanding is already at satisfactory level. Mathematical modeling and understanding of unit operations of group C is less rigorous and more challenging than for groups A and B. A key reason is still unsatisfactory understanding of multi-phase gas-solid flow, particularly for highly concentrated small, non-spherical or cohesive particles, crystals, powders or granules. Depending on the model complexity, different approaches of the model implementation and dissemination must be used. The goal is to develop user-friendly models, preferentially using the environment common for collection of experimental data and allowing equation-solving and visualization of simulation results (Excel, Dynochem, Aspen, Visual Fortran, Fluent). Goal of the presentation is to share our experience how to use models for development and scale-up and to initiate discussion about the current practice and strategy of applications of mechanistic models within the QbD process: 1/ if, when and how to employ models in the QbD filing for the product registration? 2/ what level of the model verification and validation is needed for its accreditation by FDA? 3/ how to extend the model use from development to manufacture and supply?