Improving the Safety Assessment of Chemicals and Drug Candidates by the Integration of Bioinformatics and Chemoinformatics Data.

The application of read-across and in silico tools for regulatory decision-making has been limited for pharmaceutical compounds to the assessment of genotoxic impurity. In contrast, the broad availability of toxicity data for industrial chemicals has triggered regulatory frameworks for read-across (e.g. ECHA Read-Across Assessment Framework), software tools and public databases for an automated process of gap filling in the context of safety assessment. This MiniReview provides an overview of the currently existing in silico and read-across approaches for chemicals together with recent developments for pharmaceutical compounds in these areas. It also highlights the differences and commonalities in the in silico safety assessment of industrial chemicals and drug candidates. Whereas toxicity data collection and sharing is now common practice for chemicals falling under the European REACH regulation (Registration, Evaluation, Authorisation and Restriction of Chemicals), the biggest hurdle for establishing preclinical safety databases for pharmaceutical compounds is the unwillingness to share proprietary data and lack of published data sets. In a recent consortium approach, thirteen pharmaceutical companies, eleven academic partners and six small to medium size enterprises (SMEs) of the bioinformatics sector joined forces over the last 7 years within the European Innovative Medicines Initiative project eTOX ('electronic toxicity') to design and implement a strategy for leveraging these preclinical data for small molecules and sharing them across project partners. The eTOX database has evolved as the largest preclinical toxicity database for drugs and drug candidates and currently contains more than 1900 different chemical structures and more than 8000 in vivo toxicity study data sets. It can be foreseen that the development and application of such databases for drugs or drug candidates will in the future also cross-fertilize the read-across and the in silico assessment of industrial or consumer chemicals particularly as soon as human safety data from clinical trials are integrated too.