Applicability of NAMs to support a read-across based hazard assessment workflow for systemic toxicity: a conazole case study.

Abstract

This work explores the applicability of new approach methodologies (NAMs) using a read-across approach to predict systemic toxicity with liver as the critical target organ, and derive a point of departurefor the target compound (TC) difenoconazole. Eight conazoles were used as source compounds (SCs), alongside metabolites hydroxyitraconazole and 1,2,4-triazole. The case study was structured according to the Alternative Safety Profiling Algorithm (ASPA) workflow developed within the RISK-HUNT3R project. Hierarchical clustering using chemical fingerprints identified similarity of different SCs to the TC, supporting their inclusion in the read-across analysis. A broad screening panel of NAMs, including nuclear receptor reporter assays, ToxProfiler, Cell Painting PLUS, transcriptomics in primary human hepatocytes and H295R steroidogenesis assays, was used to characterize biological activity of the TC and SCs and to substantiate the structure-based read-across. In a first prediction round, the nearest neighbour quantitative read-across approach (based on allometrically-scaled human lowest-observed-adverse-effect levels (LOAELs) of the SC) was used to predict 6.57 mg/kg/day as the LOAEL of the TC. This closely aligned with the in vivo-based value of 4.88 mg/kg/day. In a second approach, data from NAMs were used to predict human equivalent dose (HED), and the respective LOAELs of SC and TC, based on physiologically based kinetic modelling with quantitative in vitro to in vivo extrapolation. The HED derived from NAMs was approximately ten-fold lower than the LOAELs from traditional approaches. This case study illustrates suitable (ASPA-guided) approaches as well as the toxicological applicability of a NAM-based read-across assessment to obtain safe/conservative estimates of toxicity.