Predictive ADME-Tox London, UK, 27 -- 28 April 2005.
The Predictive ADME-Tox conference brought together scientists responsible for innovative in vitro and in silico screens to predict particular chemical liabilities. Oral bioavailability without adverse effects requires good absorption, distribution, metabolism and excretion (ADME) and minimal toxicity. Traditionally, these properties are measured late in development, when sufficient material is available for classical assays. Although these assays remain the gold standard, the conference focused on novel methods to predict liabilities early in the development pipeline, when a course change has less impact in the development timeline. Emerging themes from the meeting were i) integration - of data from a number of predictive assays into a 'profile' of a new class, and of safety scientists into discovery teams, in order to guide interpretation of the profile and identify tests to include in the hit-to-lead process; and ii) collaboration. As in silico predictive models are increasingly accepted, from reactivity and genetic toxicity to models for hERG inhibition, kinase crossreactivity and cytochrome P450 reactivity, increasing amounts of high-quality data will be needed to 'feed' the models. Such data will soon be beyond the capabilities of individual organisations to generate, and methods for pooling data are needed to advance the generation of predictive models for new toxicities. In addition, collaboration to apply novel methods such as metabonomics profiling to withdrawn drugs may permit the identification of new safety biomarkers to prevent such costly failures.