You are here:
Predicting developmental toxicity using the ToxCast library and pluripotent embryonic stem cell assays
Citation:
Zurlinden, T., S. Hunter, K. Saili, N. Baker, AND T. Knudsen. Predicting developmental toxicity using the ToxCast library and pluripotent embryonic stem cell assays. Presented at FutureTox IV, Crystal City, VA, November 14 - 16, 2018. https://doi.org/10.23645/epacomptox.7752080
Impact/Purpose:
20181114 - To utilize pluripoentent stem cell assays in conjunction with high throughput screening data to characterize potential developmental toxicants. (FutureTox IV)
Description:
ToxCast chemicals were profiled for developmental toxicity using the devTOX quickPredict platform from Stemina (STM). This embryonic stem cell assay utilizes human pluripotent H9 stem cells and measures the ornithine to cysteine (O/C) ratio, a ratio previously linked to developmental toxicity. Using the results from screening 1065 ToxCast chemicals, coupled with ToxRefDB developmental studies, model performance for devTOX predictions was determined for 432 compounds, of which 187 showed evidence of developmental toxicity from prenatal rat and rabbit developmental toxicity (lowest effect level ≤200 mg/kg/day). Model performance for the STM devTOX model compared to a benchmark set of 42 compounds (BM42) was 78.5% accuracy (0.65 sens, 1.0 spec, 0.79 f1) and 61.3% accuracy (0.31 sens, 0.84 spec, 0.41 f1) for the ToxRefDB compounds. To characterize the biological pathways underpinning the STM hit calls and to utilize pathway-based predictions of developmental toxicity, 337 enzymatic and receptor signaling assays in the ToxCast NovaScreen dataset (NVS) were trained on the STM hit call data to determine gene associations with STM positive and STM negative responses. The addition of NVS predictions resulted in BM42 accuracy increasing to 83.3% (0.77 sens, 0.88 spec, 0.83 f1) while accuracy in the full set of compounds was 60% (0.55 sens, 0.64 spec, 0.54 f1). These findings demonstrate the foundation for a new approach methodology for predictive developmental toxicity based on high throughput in vitro data and in silico models. [Does not reflect US EPA policy].
URLs/Downloads:
DOI: Predicting developmental toxicity using the ToxCast library and pluripotent embryonic stem cell assays
FUTURETOX_ZURLINDEN_FINAL_M.PDF (PDF, NA pp, 2548.206 KB, about PDF)