A case study with benzoic acid demonstrating how integration of In Silico Tools, mechanistic In Vitro NAM, and toxicological data improves definition of point of departure for a chemically defined class in read-across assessments

Benzoic acid and its derivatives are widely used preservatives in cosmetic and food products. This study quantitatively assesses the suitability of benzoic acid-related substances for read-across (RAX) within a Next Generation Risk Assessment framework, aimed at defining a Point of Departure (POD) for the chemical class. Using in silico Matched Molecular Pair and Quantitative Similarity approaches, we delineated sub-groups within the broader benzoic acid derivative category, further refined by metabolism and reactivity data to characterise the class boundaries. Pharmacological profiling, ToxCast, and transcriptomics confirmed the low bioactivity of benzoic acid analogues and identified compounds outside the defined class exhibiting distinct reactivity profiles. A composite TotalScore of similarity across metabolism, reactivity, and physicochemical properties guided analogue selection, with a scale ranging from 0.0 (indicating no similarity in any attribute) to 3.0 (indicating a similarity of 1 for each attribute). Highest scoring analogues encompassed sodium, potassium, calcium, magnesium, and ammonium benzoate salts along with m- and p-toluic acids, based on their analogous metabolism. Physiologically Based Kinetic modelling aided interpretation of discordant in vivo toxicity data related to administration routes and corroborated the lack of bioactivity seen in in vitro systems with this class of compounds. This case study illustrates a structured approach combining in silico, mechanistic, and toxicological data to define POD for chemical classes of low systemic toxicity.