Read‐Across Application for Food or Feed Ingredients

Abstract

Abstract This project evaluated the applicability of existing alternative data, such as chemical, biological and metabolite similarity, to improve the selection of relevant source compound (SC). This information was modularly integrated into read‐across (RAX) case studies addressing systemic toxicity after repeated exposure or developmental toxicity. For this purpose, data‐rich reference classes of pesticides were defined, with propiconazole and iodosulfuron methyl sodium as target compounds (TCs). The combination of chemical and biological similarity for TC propiconazole detected mostly relevant SC from reference class compounds. Biological similarity was calculated using binary hit call from ToxCast dataset, which is highly dependent on the data density. Low data density was used as a measure of uncertainty. In the case of the TC iodosulfuron methyl sodium, ToxCast data confirmed overall low activity. Second case study started with biological similarity calculated from ToxCast dataset. This approach resulted in an overwhelming number of candidate SCs. This indicates that the biological hit call data are relatively unspecific, as they are activated by many compounds. The integration of shared metabolites can efficiently restrict the selection of SCs to the most relevant compounds, coupled with integration of chemical and/or biological similarity. In absence of observed in vivo data, metabolites can be predicted using available tools, which generated comparable results. Based on apical findings from in vivo legacy studies, compound classes were not able to be discerned, primarily due to induced hepatotoxicity observed in about 60% of all repeated dose oral exposure studies. Overall, a RAX assessment framework integrating existing information on metabolites and biological properties to identify SC in a modular approach is recommended. The case studies presented suggest an increased confidence of SC identification using metabolite similarity. This suggestion complements the workflow proposed by EU‐ToxRisk, which focuses on targeted testing and assessment of SC upon their identification.