Towards a Common Regulatory Framework for Computational Toxicology: Current Status and Future Perspectives

In order to efficiently and effectively assess the risks of large numbers of existing chemicals and new chemical entities, there is an increasing emphasis in the regulatory setting on the use of so-called ¿non-testing¿ methods, either as a supplement to, or as a substitute for, traditional testing methods. In particular, alternatives to animal methods are being developed to reduce the need for animal testing in pharmacology and toxicology. Non-testing methods are based on the premise that the properties (including physicochemical properties and biological activities) of a chemical depend on its intrinsic nature and can be directly predicted from its molecular structure or inferred from the properties of similar compounds whose activities are known. Non-testing methods include a range of predictive approaches, including Structure-Activity Relationships (SARs), Quantitative Structure Activity Relationships (QSARs), chemical grouping and read-across, and computer-based tools based on the use of one or more of these approaches. The main question for the assessor when applying non-testing methods for regulatory purposes concerns the usefulness of the approach, which can be broken down into the practical applicability of the method and the adequacy of the predictions. Considerable progress has been made at the EU and international levels to develop a harmonised framework for assessing and documenting non-testing methods and their predictions. Exactly how this framework is applied in practice will depend on the provisions of the specific legislation (e.g. chemicals, pesticides, biocides, cosmetics) and the context in which the non-testing data are being used (including, for example, whether a traditional testing method is being replaced, whether additional, supporting data are available, and the consequences of making an inaccurate prediction). The general framework leaves largely open the difficult question of how to determine the adequacy of predicted data, and there is a considerable need to develop detailed guidance on how the predictions generated by non-testing methods can be translated into regulatory decisions. This chapter introduces the conceptual basis of SARs and QSARs, collectively referred to as (Q)SARs, as well as the related approach of chemical grouping (category formation) and read-across within chemical groups (categories). The current international framework for (Q)SAR models and predictions is then described (a similar framework has been developed for category and read-across approaches). The practical applicability of this framework. is illustrated by focussing on a checklist of 10 key questions, with respect to some well known software tools and their predictions of genotoxicity of two case study compounds. The purpose of these case studies is to highlight some of the scientific issues that need to be considered, as well the difficulties encountered. This leads into a discussion of what is needed to provide further guidance on the assessment of prediction adequacyJRC.I.5-Systems Toxicolog