EMA: Expectations for bootstrapping to calculate the 90% confidence interval for the f2 similarity factor when comparing dissolution profiles with high variability
The new PKWP/BSWP Question and Answer 3.11 deals with conducting and reporting bootstrap methodology for calculating 90% confidence interval for the f2 similarity factor when comparing dissolution profiles
Appendix I of the Guideline on the Investigation of Bioequivalence (CPMP/EWP/QWP/1401/98 Rev. 1) defines the f2 similarity factor as the method for testing the similarity of dissolution profiles. The guidance also specifies, that f2 similarity factor is not appropriate, when relative standard deviations (RSD) of any product is higher than 20% at early time-points (up to 10 minutes), and/or higher than 10% at other time points. The guidance does not prescribe any method in detail for such cases when high variability is observed in dissolution tests.
However, PKWP/BSWP Question and Answer 3.9 clarified that the Mahalanobis Distance (MD) metric cannot be supported as a preferred methodological approach to decide upon similar dissolution, even in situations where the f2 statistic should not be used in the way outlined in the CHMP bioequivalence guideline.
As of February 2022, a new PKWP/BSWP Question and Answer 3.11 was published that deals with expectations for bootstrapping to calculate the 90% confidence intervals for the f2 similarity factor. In it, bootstrap methodology is described, how it should be conducted and reported. Also, requirements for the software for conducting the analysis are stated.
The expectations for conducting and reporting bootstrap methodology are available at Question and Answer 3.11 here: https://www.ema.europa.eu/en/human-regulatory/research-development/scientific-guidelines/clinical-pharmacology-pharmacokinetics/clinical-pharmacology-pharmacokinetics-questions-answers#3.-bioequivalence-(general)-section