Dissolution Profiles Comparison Using Conventional and Bias Corrected and Accelerated f2 Bootstrap Approaches with Different Software's: Impact of Variability, Sample Size and Number of Bootstraps.

Dissolution profiles comparison is an important element in order to support biowaivers, scale-up and post approval changes and site transfers. Highly variable dissolution can possess significant challenges for comparison and f2 bootstrap approach can be utilized in such cases. However, availability of different types of f2 and confidence intervals (CI) methods indicates necessity to understand each type of calculation thoroughly. Among all approaches, bias corrected and accelerated (BCa) can be an attractive choice as it corrects the bias and skewness of the distribution. In this manuscript, we have performed comparison of highly variable dissolution data using various software's by adopting percentile and BCa CI approaches. Diverse data with different variability's, number of samples and bootstraps were evaluated with JMP, DDSolver, R-software, SAS and PhEq. While all software's yielded similar observed f2 values, differences in lower percentile CI was observed. BCa with R-software and JMP provided superior lower percentile as compared to other computations. Expected f2 recommended by EMA has resulted as stringent criteria as compared to estimated f2. No impact of number of bootstraps on similarity analysis was observed whereas number of samples increased chance of acceptance. Variability has impacted similarity outcome with estimated f2 but chance of acceptance enhanced with BCa approach. Further, freely available R-software can be of attractive choice due to computation of various types of f2, percentile and BCa intervals. Overall, this work can enable regulatory submissions to enhance probability of similarity through appropriate selection of number of samples, technique based on variability of dissolution data.