Bridging the gap in chemical risk assessment: Leveraging metabolite similarity for enhanced read-across applications

Read-across is frequently used in chemical risk assessment to assess the toxicological properties of data-poor compounds, termed target compounds. The selection of relevant source compounds (SCs) is one of the main challenges, particularly for endpoints such as chronic and developmental toxicity, where knowledge on the mechanisms leading to the observed apical findings is frequently lacking. In this study, chemical and biological similarity versus metabolite similarity were tested for improved read-across strategies in support of chemical safety assessment. To this effect, existing data from two well-known pesticide classes of 27 triazoles and 8 triazinyl-sulfonylureas as reference groups were integrated for chemical, biological and metabolite similarity within SC identification to assess the performance of the three approaches and their modular combinations. Our findings indicate that utilizing metabolite similarity enhances SC identification, yielding high positive predictive values (PPV) of 85-100% while sensitivity remained lower (12-65%). Observed and predicted metabolites yielded comparable results. Chemical and biological similarity alone was less effective to discriminate SC from irrelevant other compounds. Modular approaches, combining e.g. structural and biological similarity, enhanced SC selection (PPV up to 100%). This research underscores the potential of metabolite data to strengthen read-across justifications, thereby contributing to regulatory compliance and safety evaluations. The proposed framework integrates metabolite similarity into existing EU-ToxRisk workflows, aiming to enhance their reliability in chemical safety assessments.