The model produced a good correlation (r2 = 0.531, n = 506) and reasonable prediction (error = 0.862). The predictive ability of the model was assessed using the non-parametric approach. A multiple regression analysis was performed with the combination of Sd and Kad values and the pharmacokinetic parameter values as the dependent variables.
The pharmacokinetic parameters of the 506 drugs determined from in vivo data (Cmin, CL, and Vd) were used as the empirical descriptor of the permeability. The model was composed of two parameters: the fraction of drug dissolution by gastrointestinal fluids (Sd) and the elimination/absorption coefficient (Kad), which can be readily obtained from in vitro dissolution data. Therefore, we used 506 drugs whose bioavailability from in vitro dissolution was previously evaluated as a descriptor of their bioavailability in vivo. This could be due to the possibility that the dissolution data are not always available for physicochemical evaluation of drug properties. However, the mechanistic relation between the dissolution and the bioavailability has not been satisfactorily understood in general. The dissolution-based model of absorption is a simple and useful tool for predicting the bioavailability of poorly water-soluble drugs.
In drug development, the bioavailability is often compromised owing to undesired drug-dissolution interactions that induce the drug to dissolve in the gastrointestinal media more slowly than intended. In general, drug clearance and drug distribution in the body are the two principal determinants of the drug bioavailability. In vitro dissolution is a predictive tool for the in vivo behavior of a drug. The Challenge of GendebPrediction of in vivo drug clearance from in vitro dissolution data of drugs having a poor bioavailability.