Beyond the limits in early ADME prediction to boost v-HTS

CINF 78

Jacques R. Chretien1, Han van de Waterbeemd2, Nadege Piclin3, Christophe Wechman4, and Marco Pintore3. (1) BioChemics Consulting, Innovation Center, 16 L. de Vinci, 45074 Orleans cedex 2, France, (2) PDM, Department of Drug Metabolism, Pfizer Global Research and Development, IPC 664, Sandwich, CT13 9NJ, United Kingdom, (3) BioChemics Consulting SAS, Innovation Center, 16 L. de Vinci, 45074 Orleans cedex 2, France, (4) LBLGC / CBI, UPRES EA 1207, University of Orleans, Faculty of Sciences, 45067 Orleans Cedex 2, France
Appropriate pharmacokinetic properties are important for the success of a drug discovery program. There is a need to incorporate ADME considerations already in the first phases of the drug discovery, more particularly in virtual design and virtual screening (v-HTS). Such procedures will be able to predict ADME properties of any molecule beyond the limits of the Lipinski rules. Recently, new computational methods based on Genetic Algorithms and Fuzzy Logic have been developed by us allowing to develop a number of early ADME predictors [1]. In this contribution, their application to the main pharmacokinetic properties, i.e. oral absorption, bioavailability, volume of distribution and clearance, will be discussed. All models generated were validated by cross-validation, test set and Y-sampling procedures, and most of them were able to predict correctly ADME properties with prediction rates higher than 65-70%. Moreover, the proposed techniques showed robustness and a prediction power higher than those derived from other comparable methods.

References: [1] Pintore M, van de Waterbeemd H, Piclin N, Chrétien J., Prediction of oral bioavailability by adaptive fuzzy partitioning, Eur J Med Chem (2003), 38, 427-431.