Should medicinal chemists consider the pharmacokinetic (PK) profile of drug candidates? Absolutely. PK concerns can kill even the most promising drug candidates. Classical computational estimators of PK, which are based on assumptions of linearity, are widely used in this arena. We show that these are sufficient to predict multiple dosing profiles at steady state, but only for limited classes of drugs. Significant deviations from this predicted behavior are common, and may remain undiscovered until late in the drug discovery and development cycle. Those cases are often blamed on “nonlinearities in PK” – but what does that really mean? What specific factors contribute to these different outcomes? Using the Advanced Compartmental Absorption and Transit™ (ACAT™) model, we have performed simulation studies on several drugs that show deviations from linear pharmacokinetics with multiple dosing. For example, a linear model underestimates the steady state PK of digoxin dosed at 24h intervals because it does not take into account intricacies of its colonic absorption. On the other hand, a linear model overestimates the steady state PK of carbamazepine due to too optimistic assessment of its solubility and dissolution. Full-scale mechanistic oral absorption/PBPK simulations can account for these mechanisms, but are not well-suited for large scale discovery projects. We have developed a fast in silico tool for conducting oral absorption/compartmental PK simulations that are faster than full-blown PBPK models and more accurate than simple linear models for initial dose estimations in large discovery compound libraries. Our work concludes with a set of recommendations for predicting steady state PK in lead optimization.

Beyond IC50 and simple PK models – considerations for discovery chemists

253rd National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.