Abstract

Rational Bioavailability Design: Global Sensitivity Analysis of Physiologically-Based Pharmacokinetics (GSA of PBPK)

Eric Martin, Director, Computational Chemistry, Novartis

Medicinal chemists are eager for guidance on how to improve bioavailability (%F) for their compound series. From simple rules of thumb they can suggest over a dozen compound properties to adjust: solubility, logP, logD, Peff, CLint, PPB, RBP, MW, efflux, PSA, flexibility,pKa1, pKa2, pKb1, etc. Since the difficulty of multi-parameter optimization grows exponentially with the number of variables, efficient %F optimization requires identifying the 2 or 3 most influential properties for their specific medchem series.

These very properties are also the principal inputs to physiologically-based pharmacokinetics (PBPK) models, which simulate the movement of a drug through an animal. Usually, PBPK simulations are carefully tuned for a few advanced drug candidates, using a combination of in vitro, in vivo and calculated inputs, to estimate human doses for clinical trials, or to prioritize compounds for expensive animal studies. In this work, GastroPlus PBPK models are instead customized for entire medchem series based on 15 to 20 rat PK studies. The key innovation was building a local QSAR for effective intrinsic clearance. All inputs are subsequently computed from structure alone, so the models can be applied to candidate molecules in advance of synthesis.

Beyond that, performing 1000s of PBPK calculations over the active property ranges for their series characterizes that series’ “bioavailability landscape”. Analyzing that landscape with PLS-based global sensitivity analysis (GSA) identifies the most influential properties affecting %F. A critical innovation here was employing a large database of drug-like compounds to eliminate incompatible property combinations. The resulting analysis identifies for the chemists the important few properties to adjust for their series as a whole, as well as more specific advice for optimizing %F around particular compounds of interest.