Physicochemical property space of hepatobiliary transport and computational models for predicting rat biliary excretion.

Biliary excretion (BE) is a major elimination pathway, and its prediction is particularly important for optimization of systemic and/or target-site exposure of new molecular entities. The objective is to characterize the physicochemical space associated with hepatobiliary transport and rat BE and to develop in silico models. BE of 123 in-house compounds was obtained using the bile-duct cannulated rat model. Human and rat hepatic uptake transporters (hOATP1B1, hOATP1B3, hOATP2B1, and rOatp1b2) substrates (n = 183) were identified using transfected cells. Furthermore, the datasets were extended by adding BE of 163 compounds and 97 organic anion transporting polypeptide (OATP) substrates from the literature. Approximately 60% of compounds showing percentage of BE (%BE) ≥ 10 are anions, with mean BE of anions (36%) more than 3-fold higher than that of nonacids (11%). Compounds with %BE ≥ 10 are found to have high molecular mass, large polar surface area, more rotatable bonds, and high H-bond count, whereas the lipophilicity and passive membrane permeability are lower compared with compounds with %BE < 10. According to statistical analysis and principal component analysis, hOATPs and rOatp1b2 substrates showed physicochemical characteristics that were similar to those of the %BE ≥ 10 dataset. We further build categorical in silico models to predict rat BE, and the models (gradient boosting machine and scoring function) developed showed 80% predictability in identifying the rat BE bins (%BE ≥ 10 or < 10). In conclusion, the significant overlap of the property space of OATP substrates and rat BE suggests a predominant role of sinusoidal uptake transporters in biliary elimination. Categorical in silico models to predict rat BE were developed, and successful predictions were achieved.