Computational Estimation of Biliary Excretion of Compounds and the Role of Transporters

Biliary excretion is one of the main elimination pathways for drugs and/or their metabolites. Therefore, an insight into the structural profile of cholephilic compounds through accurate modelling of the biliary excretion is important for the estimation of clinical pharmacokinetics in early stages of drug discovery. The aim of this project was to develop Quantitative Structure-Activity Relationships(QSAR) as computational tools for the estimation of biliary excretion. In addition, the structural requirements for biliary excretion were investigated in relation to the structural requirements for binding to uptake and efflux transporter proteins that are involved in hepatobiliary elimination.

The study used three datasets; 1. percentage of dose excreted intact into bile in rat for 217 compounds, 2. P-gp inhibition constants for 219 compound, 3. percentage inhibition of OATP transporters, OATP1B1, OATP1B3 and OATP2B1. Statistical techniques were stepwise regression analysis, Classification and Regression Trees (C&RT), Chi-square Automatic Interaction Detector (CHAID), Boosted trees (BT), Random Forest (RF) and Multivariate Adaptive Regression Splines (MARS) models.

The study resulted in QSARs for the prediction of biliary excretion, P-gp binding constants and percentage inhibition of OATPs, along with QSARs incorporating predicted P-gp and OATP inhibition values for the prediction of biliary excretion. Simple regression tree models were of similar accuracy to the boosted trees model in the estimation of the percentage of bile excretion of compounds. Molecular descriptors selected by these models indicated a higher biliary excretion for relatively hydrophilic compounds especially if they have acid/base dissociation, and a large molecular size above 348 Da.

The major role of OATPs in biliary excretion was indicated using interactive decision tree models with OATP1B1 binding being the most successful predictor of biliary excretion amongst the three OATP subfamilies. In contrast, predicted P-gp binding parameters were not successful in the prediction of biliary excretion. This may be due to problems in extrapolating the in vitro P-gp binding data to the in vivo situation, or due to the difference in the chemical spaces of the P-gp and biliary excretion datasets which may lead to the compounds in biliary excretion dataset to fall outside the applicability domain of the P-gp models.