You are here:
Towards replacing the two-year bioassay with short-term assays: gene expression thresholds can predict rat liver tumorigens
Citation:
Corton, Jon. Towards replacing the two-year bioassay with short-term assays: gene expression thresholds can predict rat liver tumorigens. Society of Toxicology 59th Annual Meeting, Anaheim, CA, March 16 - 19, 2020. https://doi.org/10.23645/epacomptox.24535009
Impact/Purpose:
Presentation to the Society of Toxicology 59th Annual Meeting March 2020. Traditional data sources for cancer risk assessment are resource-intensive, retrospective, and not feasible for the vast majority of environmental chemicals. Incorporation of quantitative genomic data from short-term rodent studies may adequately define protective thresholds for potential tumorigens as a bridge to move from current testing to greater reliance on in vitro assays. We hypothesized that gene expression biomarkers that measure the activation of the major molecular initiating events (MIEs) in rodent liver cancer adverse outcome pathways exhibit chemical-independent thresholds beyond which cancer occurs and the thresholds could be used to predict liver cancer. The hypothesis was tested by defining thresholds of gene expression biomarkers of liver cancer MIEs using training sets from the 77 and 86 chemicals in the TG-GATES and DrugMatrix datasets, respectively and testing them in a number of contexts.
Description:
Traditional data sources for cancer risk assessment are resource-intensive, retrospective, and not feasible for the vast majority of environmental chemicals. Incorporation of quantitative genomic data from short-term rodent studies may adequately define protective thresholds for potential tumorigens as a bridge to move from current testing to greater reliance on in vitro assays. We hypothesized that gene expression biomarkers that measure the activation of the major molecular initiating events (MIEs) in rodent liver cancer adverse outcome pathways exhibit chemical-independent thresholds beyond which cancer occurs and the thresholds could be used to predict liver cancer. The hypothesis was tested by defining thresholds of gene expression biomarkers of liver cancer MIEs using training sets from the 77 and 86 chemicals in the TG-GATES and DrugMatrix datasets, respectively and testing them in a number of contexts. The biomarkers tested consisting of 7-113 genes included those that predict genotoxicity, cytotoxicity, and activation of AhR, CAR, ER, or PPARα. Thresholds were calculated as the maximum values derived from exposures that do not lead to liver cancer. For all of the biomarkers, clear threshold values could be identified that were consistent across training and test sets. While thresholds derived from the TG-GATES study were not very predictive of liver tumorigens in the DrugMatrix study (77-81%), thresholds derived from the DrugMatrix study were very predictive in the TG-GATES study (84-100%). The DrugMatrix derived thresholds were most predictive when applied to test sets of 7d and 14d treatments (100% and 99%, respectively). In addition, thresholds derived from just 12 genes (2/biomarker) exhibited high predictive accuracy (up to 94%). These findings support the idea that early genomic changes can be used to establish threshold estimates or “molecular tipping points” that are predictive of later-life outcomes. (This abstract does not reflect EPA policy.)
URLs/Downloads:
DOI: Towards replacing the two-year bioassay with short-term assays: gene expression thresholds can predict rat liver tumorigens
CORTON_SOT_SYMPOSIUM_02_14_20.PDF (PDF, NA pp, 1269.147 KB, about PDF)