Gaussian Process Modeling for Dissolution Curve Comparisons

Abstract

Dissolution studies are an integral part of pharmaceutical drug development, yet standard methods for analysing dissolution data are inadequate for capturing the true underlying shapes of the dissolution curves. Methods based on similarity factors, such as the f 2 statistic, have been developed to demonstrate comparability of dissolution curves, however, this inability to capture the shapes of the dissolution curves can lead to substantial bias in comparability estimators. In this article, we propose two novel semi-parametric dissolution curve modeling strategies for establishing the comparability of dissolution curves. The first method relies upon hierarchical Gaussian process regression models to construct an f 2 statistic based on continuous time modeling that results in significant bias reduction. The second method uses a Bayesian model selection approach for creating a framework that does not suffer from the limitations of the f 2 statistic. Overall, these two methods are shown to be superior to their comparator methods and provide feasible alternatives for similarity assessment under practical limitations. Illustrations highlighting the success of our methods are provided for two motivating real dissolution data sets from the literature, as well as extensive simulation studies.

Statistical modeling approaches for the comparison of dissolution profiles

“…Finally, we would like to point out that all of the inferences and analyses conducted in this paper were carried out under a frequentist paradigm. Although not explored in this paper, there are many Bayesian alternatives to dissolution profile comparisons that provide viable alternatives (for example, see Novick et al, 9 Mockus and LeBlond, 39 Pourmohamad et al, 11 and the references therein) that allow for, say, calculating probabilities for the associated equivalence hypotheses which frequentist procedures cannot do.…”

“…Many statistical methods have been proposed for deciding whether or not two dissolution profiles are similar. Common approaches to dissolution profile comparison are to use either model-dependent methods [7][8][9][10][11][12][13] or modelindependent methods. 10,[14][15][16] Model-dependent methods make either the explicit assumption of an underlying statistical distribution for the dissolution measurements or assume a particular parametric form for the dissolution profiles.…”