Segmental-Dependent Intestinal Drug Permeability: Development and Model Validation of In Silico Predictions Guided by In Vivo Permeability Values

Abstract

The goal of this work was to develop an in silico model that allows predicting segmental-dependent permeability throughout the small intestine (SI). In vivo permeability of 11 model drugs in 3 SI segments (jejunum, mid-SI, ileum) was studied in rats, creating a data set that reflects the conditions throughout the SI. Then, a predictive model was developed, combining physicochemical drug properties influencing the underlying mechanism of passive permeability: Log p, polar surface area, MW, H-bond count, and Log fu, with microenvironmental SI conditions. Excellent correlation was evident between the predicted and experimental data (R2 = 0.914), with similar predictability in each SI segment. Log p and Log fu were identified as the major determinants of permeability, with similar contribution. Total H-bond count was also a significant determinant, followed by polar surface area and MW. Leaving out any of the model parameters decreased its predictability. The model was validated against 5 external drugs, with excellent predictability. Notably, the model was able to predict the segmental-dependent permeability of all drugs showing this trend experimentally. Model predictability was better in the high-permeability versus low-permeability range. Overall, our approach of constructing a straightforward in silico model allowed reliable predictions of segmental-dependent intestinal permeability, providing new insights into relative effects of drug-related factors and gastrointestinal environment on permeability.