Dr. Mark Cronin welcomed hundreds of attendees in a packed to capacity session Monday morning devoted to presentations relating to the use of computational approaches to predict toxicity from cosmetic ingredients. Dr. Cronin described the European SEURAT-1 Research Initiative (€50 million Euro initiative funded by EU and Cosmetics Europe), which launched in 2011 and is intended to find alternatives to animal testing. The COSMOS Project is one of six SEURAT-1 research projects and supports in silico approaches to predict toxicity from cosmetics ingredients. Such approaches are required under EU’s Recast regulations, which ban animal testing for ingredients used exclusively in cosmetic ingredients.
Chihae Yang’s presentation introduced the European COSMOS database, which contains more than 81,600 chemical structures, including 19,000 cosmetics-related substances, with links to COSING and SCCS opinions. The Threshold of Toxicological Concern (TTC) approach was discussed, along with the practical application of TTC to cosmetics, along with challenges associated with the use of the TTC to assess the safety of cosmetics.
Interesting fact: Yang noted that a female eats one kg of lipstick over a lifetime!
Elena Fioravanzo spoke about the role of bioavailability in the risk assessment of cosmetics ingredients. A challenge to using the large dataset of oral route data to predict the toxicity of cosmetics ingredients is the extrapolation to dermal exposure scenarios.
Mark Cronin discussed the application of adverse outcome pathways (AOPs) to characterize a chemical’s toxicity. Computational prediction is needed to characterize the toxicity of the 19,000 cosmetics ingredients in the COSMOS database. Use of a Molecular Initiating Event (MIE) to link chemistry to adverse outcome is a new paradigm to predict toxicity. Describing the MIE will inform the development of structural alerts that can be checked against data and further refined to support toxicity prediction. The AOP provides the framework for gathering evidence to support category membership and possibly provide a quantitative prediction.
Kirk Arvidson of US FDA discussed the US FDA’s CERES Knowledgebase (Chemical Centric Food Additives Knowledgebase). CERES provides data (toxicity data and use data), structure analog searching and data retrieval mode of action QSAR models, and TTC evaluations. Tools highlighted included ChemoTyper, which is a free software tool that allows for searching and highlighting chemical chemotypes in datasets of molecules. ChemoTyper includes the ToxPrint public set of chemotypes developed by Altamira LLC for FDA’s CERES project. ToxPrint contains three basic subsets: generic structural fragments, Ashby-Tennant genotoxic carcinogen rules, and cancer TTC categories.
Ann Richard from the US EPA’s National Center for Computational Toxicology spoke about the US EPA’s ToxCast, Tox21, and the EU COSMOS projects. The Tox21/ToxCast chemical overlap with COSMOS is significant (e.g., Tox21 contains 8,300 chemicals with 1,478 similar structures in the COSMOS database). Widespread adoption of computational approaches requires sufficiently large datasets and public availability of databases and tools, as well as transparency of methods. Ideally, a public chemotype knowledge base should be built to inform to prioritize chemical selection and safety assessments.