Mining the FDA Adverse Events Databases for Predictive Toxicology

Jeffrey Wiseman and Matthew Clark, Pharmatrope, USA

The adverse events databases maintained by drug regulatory agencies such as the FDA are the most extensive compilations of chemical toxicity with respect to the numbers of patients exposed. They also comprise the only open public databases of human toxicity and as such are a key resource for validating predictions of toxicity in humans. The databases, however, contain more than eight million records that show links between drugs and adverse events but do not necessarily validate a cause and effect relationship. We have developed a resource that pre-compiles the statistically significant relationships in database form and made these available through the OpenTox framework as a resource for data mining. We present results for drug-cardiac adverse event associations that illustrate the utility of the resource by analyzing the association of long QT syndrome with cardiac failure. Finally, we consider the application of the drug adverse events to the SEURAT program, which is faced with the daunting task of extending our understanding of drug toxicity to new chemical classes -  cosmetics, environmental chemicals, and industrial chemicals - outside the pharmaceutical chemical space.