Computational methods in support of chemical risk assessment

Chemical risk assessment for human health effects is performed in order to establish safe exposure levels of chemicals to which individuals are exposed. The process of risk assessment traditionally involves the generation of toxicological studies from which health based guidance values are derived for a specific chemical. For low level exposures to chemicals, where there are no or limited chemical specific toxicity data, the application of the Threshold of Toxicological Concern (TTC) approach may estimate whether the exposure levels can be considered safe. The TTC approach has recently gained increasing interest as new requirements, under different regulatory frameworks, emerge for the safety assessment of chemicals and to assess chemicals for which testing is not routinely required. The application of TTC relies heavily on computational (in silico) methods. In silico tools are computer implemented models that, based on commonalities in the toxicity of “similar” chemical structures, may predict hazard. In silico methods are rapidly evolving and gaining importance within the context of Integrated Approaches to Testing and Assessment (IATA) and their acceptance for regulatory purposes is expanding. The work presented in this thesis has focused on the use and applicability of a wide range of computational approaches to assist in the application of the TTC concept. In the TTC approach, the identification of genotoxic chemicals is a primary requirement. In silico approaches apply expert knowledge and/or statistical methods to either predict genotoxicity or to identify structural alerts associated with it. This thesis focused, in part, on a group of important environmental pollutants, nitrobenzenes, to assess the applicability of in silico tools to predict genotoxicity. For this purpose a dataset containing 252 nitrobenzenes including Ames test results was compiled. Based on these test results a case study for sodium nitro-guaiacolate, a pesticide active substance, was developed. The case study demonstrated that (Q)SAR and a category approach incorporating read-across, are applicable for the prediction of genotoxicity and supports their use within a weight of evidence approach. Another aspect of the TTC approach is the evaluation of repeat dose, non-cancer endpoints. For that purpose chemicals are separated into groups related to three levels of concern based on the Cramer classification. For each level, namely the Cramer Classes (I, II and III), a safe exposure level has been established. Therefore, as interest to apply TTC expands to new groups of chemicals, the reliability and conservativeness of the established thresholds relative to Cramer Classes for the new chemistries must be established. In this thesis the TTC approach was evaluated for 385 cosmetic ingredients, 77 biocides and 102 compounds classified as reproductive and developmental toxicants. To support the evaluation at different levels, chemical datasets containing toxicological data were utilised and computational tools were applied to compare datasets. The results indicated, that the historical “Munro” dataset is broadly representative for cosmetics and biocides. In addition, that the threshold levels for Cramer Class III are within the range of Munro’s threshold further supports the validity of the TTC approach and its conservativeness for the groups of chemicals analysed. Cramer Class I thresholds were found to be valid only for classified developmental and reproductive toxicants. The results also supported the validity of the classification of chemicals into Cramer class III. It is foreseen that the TTC approach will gain increasing acceptance in the risk assessment of different groups of chemicals. Therefore it is emphasised that the future work should focus on the identification of the limitations of the application of TTC, including the identification of groups of chemicals to which TTC cannot be applied, the expansion of the underlying toxicological datasets, and the development of tools to support the application of TTC so that is transparent and acceptable for regulatory purposes.