Hierarchical Bayesian approach applied to the formulation of sustained-release suppositories and dissolution profile modeling

Abstract

The hierarchical Bayesian modeling approach was used to select the appropriate empirical kinetics model of sustained release and to optimize the in vitro dissolution rate of the sustained-release suppository by controlling the composition of Eudragit L-100 and Eudragit S-100 in the experimental mixture. Thirteen formulations of suppositories were prepared with 2 g (10%) mixture of Eudragit® R-100 and S-100 according to a personalized mixture experimental design. The cumulative release of active ingredient was measured at five times (20, 50, 80, 160, and 235 minutes). The best model was selected using Rsq(Adjust) and akaike information criterion for standard method and by using the weight of widely applicable information criterion (WAIC) and leave-one-out (LOO) cross-validation for the Bayesian approach. Frequentist approach gave three best model depending on the formulation. Compared to this, the Bayesian method was able to define a single model, which is the first-order model. The relative probability of this model is 0.97, 0.99 based on the WAIC, and LOO, respectively. The relationship between K1 (Release rate constant) and the quantities of the two Eudragits is quadratic, for Eudragit_L, Qrelease (%) = 0.0031X2−0.0026 X + 0.0069 and X is the Eudragit L100 and K1 (Rate release) = 0.41 minutes−1. The Bayesian method allowed finding the most adequate model among several models that can be generated by the standard frequentist approach.