Can acute oral in vivo toxicity testing for EU REACH be fully replaced by a QSAR method? Evaluation of the CATMoS model using chemical industry data.

Abstract

The assessment of acute oral toxicity is a fundamental endpoint for health hazard and risk assessment and a standard information requirement under REACH. The European Commission plans to update the standard information requirements, focusing on integrating non-animal-based information, proposing CATMoS as the preferred in silico tool. We evaluated the ability of CATMoS to predict GHS classification for acute oral toxicity using 860 REACH-registered chemicals. The output parameters of CATMoS were combined with an expert judgement of data quality and nearest neighbor analysis to assign a reliability category (high, moderate, or low). In a subset of 20 chemicals, predictions showed about one third each with high, moderate, and low reliability. High reliability predictions matched the experimentally determined GHS category or an adjacent one, aligning with the variability of in vivo rat acute LD50 data. Solely relying on CATMoS output could result in accepting predictions differing by two or more hazard categories, emphasizing the need for expert judgement. In conclusion, CATMoS only in combination with expert judgement is suitable as a replacement for acute oral toxicity studies under REACH if a prediction with high reliability is available and all available information is analyzed and reported as described in the QSAR assessment framework.