Impact of changes made to a product needs to be assessed to ensure product performance. Comparison of dissolution profiles is used to evaluate similarity in product performance. Similarity factor (f2) has been widely used in the pharmaceutical industry under regulatory guidances. However, the f2 method is not suitable for highly variable cases. Therefore, it is critical to identify a right tool/method in order to make meaningful assessment for product quality.
At AbbVie, we have tackled this important problem by considering the fundamentals in statistical methods, formulation design/processing/manufacturing, and dissolution methods. Several statistical methods have been explored and compared through real case studies.
In this poster, a process decision tree is provided to guide method selection, data analysis, and conclusion for regulatory use. The proposed decision tree assures good statistical power with adequate type I error control that ensures product quality, as well as ensuring passing results display a high level of confidence.
An R shiny tool based on such decision tree is presented, exemplifying typical analysis results from several suitable statistical methods.
|