Summary

Lack of efficacy continues to be a problem in drug development. Piecemeal rules of thumb such as Lipinski’s Rule of Five [1] help avoid compounds likely to be poorly absorbed, but many otherwise “allowed” property combinations can also lead to problems.  Good and bad interactions can be identified using physiologically-based pharmacokinetic (PBPK) simulation programs like GastroPlus™. Such programs are used to analyze pre-clinical and clinical data, and so have to be able to take variations in subject sex, weight, and disease state into account. They also need to incorporate subtle transporter and metabolic effects. Such details are less relevant in discovery and lead optimization, where the goal is to quickly survey likely problems with a class of chemistry and to differentiate them from ones that are specific to a particular compounds. Rodent PK is especially relevant early on.  We have created a streamlined, high-throughput version of the GastroPlus gastrointestinal absorption model for ADMET Predictor™ that enables users to rapidly estimate percentage absorbed (%Fa) and percentage available after first pass metabolism (%Fb) for a typical rat or human subject (Figures 1 & 2).  Experimental property values can be used if available, but the validation results shown here were obtained using only in silico property predictions from ADMET Predictor. Experimental deviations from those predictions can provide important insight into complicating factors that are best addressed early in a project.

22nd European Symposium on Quantitative Structure-Activity Relationships, September 16-20, 2018, Thessaloniki, Greece

By Robert D Clark