(565e) Identification of Key/Non-Key Process and Model Parameters of Polishing Chromatography, Primary Containers and Autoinjectors Using Mechanistic Models

Identifying key/non-key process parameters is an essential step in the Quality-by-Design methodology. It demonstrates true process understanding and it enables risk-based decision making. Typically, this is done by FMEA and DOE studies which have some important limitations. First-principles and mechanistic models are being used to advance the vision of rational drug substance development via a rigorous and quantitative Quality-by-Design approach. In this work, we use three mechanistic models to study the factor/response of drug/process/device systems of practical interest to industry. We employ well-established global sensitivity analysis techniques (Sobol indices) to address the so-called factor screening, factor prioritization and factor fixing setting [Saltelli et al., 2008]. This enables a rigorous analysis of key/non-key process and model parameters to accelerate design iterations, manage/reduce risk, and ultimately assure quality via a probabilistic in-silico QbD design space. This new methodology can work as a complement (or a replacement) of more traditional and experiential-based FMEA approaches. This is the first time such a rigorous analysis has been published for a polishing chromatography system. The application to the primary container case-study reveals that only half of the factors of this process are non-essential. For the autoinjector device two sub-components happen to be the most relevant to system variability.