Evaluating dissolution profiles of an anti-HIV agent using ANOVA and non-linear regression models in JMP software.

A powerful statistical method was designed using JMP software to detect factors contributing to differences in the dissolution process of an antiviral drug delivered in an oral dosage form. Due to the large number of dissolution media available for solid dosage forms, a statistical method to choose the appropriate medium is critical for testing solid dosage forms. We have developed an analysis of variance model to analyze the overall dissolution profile obtained from the various media. In vitro tests were performed using a standard USP basket apparatus (Vankel Inc., Cary, NC), and the analysis used the restricted/residual maximum likelihood method (JMP software) to partition the variance due to media (pH 1.2 and 6.8, +SDS, water alone and at pH 1.2 with pepsin), time (repeated measure) and capsule (random effect). This allowed correct standard error estimates to be used to compare dissolution in different media using planned linear contrasts. The model provided us with statistically powerful criteria to identify significant differences in capsule dissolution across time and to quantify capsule-to-capsule population variance estimate. The time specific linear contrasts showed the largest sum of square values (SS) occurred at 180 min (SS=0.268) for the simulated SIF (pH 6.8) versus SGF (pH 1.2) comparison (DF=166, MSE=3.92 x 10(-3)). The dissolution processes were further characterized using a non-linear regression fit of a power law function to the data for each capsule. This resulted in a method to statistically differentiate between the dissolution processes of the capsules in different media.