Acute Toxicity Prediction in Multiple Species by Leveraging Mechanistic ToxCast Mitochondrial Inhibition Data and Simulation of Oral Bioavailability

2015 
: There is great interest in assessing the in vivo toxicity of chemicals using nonanimal alternatives. However, acute mammalian toxicity is not adequately predicted by current in silico or in vitro approaches. Mechanisms of acute toxicity are likely conserved across invertebrate, aquatic, and mammalian species, suggesting that dose-response concordance would be high and in vitro mechanistic data could predict responses in multiple species under conditions of similar bioavailability. We tested this hypothesis by comparing acute toxicity between rat, daphnia, and fish and by comparing their respective acute data to inhibition of mitochondria membrane potential (MMP) using U.S. Environmental Protection Agency ToxCast in vitro high-throughput screening data. Logarithmic scatter plots of acute toxicity data showed a clear relationship between fish, daphnia, and intravenous rat but not oral rat data. Similar plots versus MMP showed a well-delineated upper boundary for fish, daphnia, and intravenous data but were scattered without an upper boundary for rat oral data. Adjustments of acute oral rat toxicity values by simulating fractional absorption and CYP-based metabolism as well as removing compounds with hydrolyzable linkages or flagged as substrates for glucuronidation delineated an upper boundary for rat oral toxicity versus MMP. Mitochondrial inhibition at low concentrations predicted highly acutely toxic chemicals for fish and daphnia but not the rat where toxicity was often attenuated. This use of a single high-throughput screening assay to predict acute toxicity in multiple species represents a milestone and highlights the promise of such approaches but also the need for refined tools to address systemic bioavailability and the impact of limited absorption and first pass metabolism.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    37
    References
    14
    Citations
    NaN
    KQI
    []