November 7 - 8, 2001. Obernai, France, ACD European Users' Meeting

The Design of Drug-Like Properties

Christopher A. Lipinski, Pfizer

Abstract

Aqueous solubility and permeability data must be provided to chemistry as early as possible to avoid oral absorption problems. The minimum acceptable solubility for a drug depends on its permeability and projected clinical potency. The logic for an experimental turbidimetric solubility assay depends on marked differences in the way solubility is invoked in an early discovery as opposed to development setting. The solubility ranking of collections of chemical compounds can be estimated either from data mining, e.g. using the "rule of five" or from internal experimental solubility measurements or from calculations from among the many commercially available solubility programs. Although these programs may have limited utility in structure activity relationships on individual compounds, they are useful in categorizing collections of compounds and in ranking real or virtual chemical libraries. Fairly reliable algorithms such as polar surface area are also available to rank intestinal permeability. In terms of resource allocation there is no excuse for not understanding the likely solubility and permeability rankings of collections of compounds. These points will be illustrated by a discussion of how differences in research approach between Merck and Pfizer, Groton have led to differences in the physical property trends of clinical candidates.

With regard to combinatorial chemistry, poor solubility or poor permeability chemistry space can be illustrated by constructing a two dimensional grid of calculations estimating the aqueous solubility and the intestinal permeability of a variety of chemical libraries. There is a quadrant in chemistry space containing libraries predicted to have substantial solubility and permeability problems. The screening dose in Caco-2 cell culture permeability assays in a pharmaceutical company setting clearly points out the problem caused by poor solubility chemistry space. It is frequent to see presentations on pharmaceutical industry Caco-2 screening in which the Caco-2 screening dose is 10 or 20 uM. One third of compounds screened at 10 uM were insoluble (or erratically soluble) in an aqueous medium. Assay reproducibility becomes a major issue. Solubility problems are a widely distributed industry wide problem and are found in commercially available combinatorial libraries.

Download the presentation in PDF format (260 Kb) or MS Power Point 97/2000 format (210 Kb ZIP file).