Drug product dissolution is a key input to Physiologically Based Biopharmaceutics Models (PBBM) to be able to predict in vivo dissolution. The integration of product dissolution in PBBMs for immediate release drug products should be mechanistic, i.e. allow to capture the main determinants of the in vitro dissolution experiment, and extract product batch specific parameter(s). This work focussed on the Product Particle Size Distribution (P-PSD), which was previously shown to integrate the effect of dose, volume, solubility (pH), size and concentration of micelles in the calculation of a batch specific input to PBBMs, and proposed new hydrodynamic (HD) models, which integrate the effect of USP2 apparatus paddle rotation speed and medium viscosity on dissolution. In addition, new models are also proposed to estimate the quantitative impact of formulation and drug sedimentation or “coning” on dissolution. Model “HDC-1” predicts coning in the presence of formulation insoluble excipients and “HDC-2” predicts the sedimentation of the drug substance only. These models were parameterized and validated on 166 dissolution experiments and 18 different drugs. The validation showed that the HD model average fold errors (AFE) for dissolution rate prediction of immediate release formulations, is comprised between 0.85 and 1.15, and the absolute average fold errors (AAFE) are comprised between 1.08 and 1.28, which shows satisfactory predictive power. For experiments where coning was suspected, the HDC-1 model improved the precision of the prediction (defined as ratio of “AAFE-1”values) by 2.46 fold compared to HD model. The calculation of a P-PSD integrating the impact of USP2 paddle rotation, medium viscosity and coning, will improve the PBBM predictions, since these parameters could have an influence on in vitro dissolution, and could open the way to better prediction of the effect of prandial state on human exposure, by developing new in silico tools which could integrate variation of velocity profiles due to the chyme viscosity.
Read full abstract