Modelling the dissolution of diazepam in lactose interactive mixtures

The dissolution of diazepam (1-10%) in lactose interactive mixtures was studied using the USP paddle method over rotational speeds of 50-200 rpm to elucidate the mechanism of dissolution and to model the process. Dissolution of diazepam was concentration dependent and occurred rapidly, i.e., greater than 95% dissolved within 10 and 20 min for the 1 and 10% mixtures, respectively. Conventional dissolution models like the Hixson-Crowell Cube Root Law, Wagner's log probability plot and monoexponential equations provided unsuccessful data fits. The hypothesis that dissolution occurred from both dispersed and aggregated particles was tested using a biexponential model. Data treatment using the Marquardt-Levenberg non-linear least squares algorithm (Peakfit(R)) provided good fits. The rate constants for diazepam dissolution from the dispersed and aggregated particles and their initial concentrations were estimated and was consistent with the observed concentration-dependent behaviour. Laser diffraction was used to demonstrate the existence of dispersed particles and aggregates during the dissolution process. Studies in surfactant solution provided further indirect evidence of aggregate formation.