Extension of the EFSAshort forEuropean Food Safety Authority Pesticides Genotoxicity Database

The results of the project on evaluation of the performance of the existing (Q)SAR models and read across methodologies for prediction of genotoxicity of pesticides finalised in March 2019 showed that: for the prediction of point mutation (Ames test) in silico tools “[…] generated statistically significant predictions, comparable with the experimental variability of the test […] whereas the reliability of the predictions for the other endpoints appears to be far from the optimal to be used for regulatory purpose”. Besides gene mutations in bacteria, tests for clastogenicity/aneugenicity in mammalian cells (e.g. chromosomal aberration and micronucleus tests) are required by regulation. However, data requirements for metabolites and impurities are less stringent than for active substances. According to Regulation (EC) 283/2013 “supplementary studies, where they relate to substances other than the active substance, are not a routine requirement”. Therefore, in silico tools remain a valuable alternative or a complementing tool for the evaluation of the clastogenic/aneugenic potential of chemicals (other than the active substance) but need improvement in order to be routinely used in the regulatory risk assessment. One of the issues are the availability and quality of the data used for the development and testing of these tools. The current project aims to i) extend the genotoxicity database with curated experimental data for relevant endpoints with a particular interest in the in vitro micronucleus (MN) test (OECD TG 487, 2016) while, ii) also include data extracted by BfRshort forGerman Federal Institute for Risk Assessment for the bacterial reverse mutation test (Ames test, OECD TG 471) and; iii) validate the database considering the latest changes in the experimental protocol and evaluation criteria of the results. Therefore, the proposed project would provide additional curated data for further analysis of the reasons of low performance of the in silico tools for endpoints different than Ames mutagenicity.

In 2012 the EFSAshort forEuropean Food Safety Authority Panel on Plant Protection Products and their Residues, the Panel pointed out to the need for a well‐curated database of genotoxicity data to support reliable use of in silico methods for predicting genotoxicity of chemicals in this domain. Subsequently, EFSAshort forEuropean Food Safety Authority developed a database of genotoxicity extracted from data submitted in the course of regulatory approval of pesticides, and this database was used to test the performance of existing tools for predicting genotoxicity. It was concluded that, at present, QSAR works well for Ames mutagenicity, but not for other endpoints. BfRshort forGerman Federal Institute for Risk Assessment has developed its own internal database of curated genotoxicity data extracted from original study reports, focusing initially on Ames and subsequently expanded to include in vitro micronucleus (MN) results. To facilitate extension the EFSAshort forEuropean Food Safety Authority database with Ames and in vitro MN data from the BfRshort forGerman Federal Institute for Risk Assessment database, a workflow for the migration of the data in the BfRshort forGerman Federal Institute for Risk Assessment Genotoxicity Database to a IUCLID format was developed and applied. After migration to IUCLID, there were 349 files for Ames and 183 for MN, one for each substance per endpoint. In the course of performing the data migration to the IUCLID/OHT template, several changes to the structure of the OHT 70 (Genetic toxicity in vitro) which would improve its usefulness were recognized. The BfRshort forGerman Federal Institute for Risk Assessment genotoxicity database, available in https://zenodo.org/communities/efsa-kj, represents a valuable extension, provides additional dose‐response information and might offer opportunities to validate the EFSAshort forEuropean Food Safety Authority genotoxicity database.