Addressing the challenges of acute fish toxicity hazard classification using a non-animal defined approach.

Abstract

Acute fish toxicity is an ecotoxicological endpoint that provides important information about a chemical's potential to elicit (an) adverse effect(s) on fish. These effects are typically studied using in vivo tests, but for animal welfare reasons as well as the quest for increased species relevance, biological coverage, mechanistic understanding of effects, and throughput, there have been significant efforts in recent years to reduce or replace the use of animals in (eco)toxicological hazard assessment by developing defined approaches (DA) or integrated approaches to testing and assessment. To this end, a novel score-based DA has been developed as a proof-of-concept, which integrates three in silico predictions from freely available (quantitative) structure activity relationship models: the VEGA Fish (KNN-Read-Across) and Fathead Minnow (KNN-IRFMN) models and the United States Environmental Protection Agency ECOSAR Fish 96-h LC50 model, along with in vitro RTgill-W1 data. The DA provides a categorical output aligned with the United Nations Globally Harmonized System of Classification and Labelling framework (Acute Category 1, Acute Category 2, Acute Category 3, or Not Classified) with an overall accuracy of 80%, offering a reliable alternative to traditional in vivo testing methods for acute fish toxicity.