Gene Expression Thresholds Derived From Short-term Exposures Identify Rat Liver Tumorigens. (30th June 2020)

• Record Type: Journal Article
• Authors: Hill, Thomas; Rooney, John; Abedini, Jaleh; El-Masri, Hisham; Wood, Charles E; Corton, J Christopher
• Is Part Of: Toxicological sciences. Volume 177:Number 1(2020)
• Page Start: 41
• Page End: 59
• Publication Date: 2020-06-30
• Subjects: toxicogenomics -- carcinogenicity -- biomarkers -- adverse outcome pathway -- molecular initiating event -- key event -- constitutive activated receptor -- transcript profiling -- liver cancer -- peroxisome proliferator-activated receptor alpha; Toxicology -- Periodicals; Toxicology -- Periodicals; Toxicology; Periodicals; 615.9
• DOI: 10.1093/toxsci/kfaa102
• Languages: English
• ISSNs: 1096-6080

Abstract

Abstract: Traditional methods for cancer risk assessment are resource-intensive, retrospective, and not feasible for the vast majority of environmental chemicals. In this study, we investigated whether quantitative genomic data from short-term studies may be used to set protective thresholds for potential tumorigenic effects. We hypothesized that gene expression biomarkers measuring activation of the key early events in established pathways for rodent liver cancer exhibit cross-chemical thresholds for tumorigenesis predictive for liver cancer risk. We defined biomarker thresholds for 6 major liver cancer pathways using training sets of chemicals with short-term genomic data (3–29 days of exposure) from the TG-GATES ( n = 77 chemicals) and DrugMatrix ( n = 86 chemicals) databases and then tested these thresholds within and between datasets. The 6 pathway biomarkers represented genotoxicity, cytotoxicity, and activation of xenobiotic, steroid, and lipid receptors (aryl hydrocarbon receptor, constitutive activated receptor, estrogen receptor, and peroxisome proliferator-activated receptor α). Thresholds were calculated as the maximum values derived from exposures without detectable liver tumor outcomes. We identified clear response values that were consistent across training and test sets. Thresholds derived from the TG-GATES training set were highly predictive (97%) in a test set of independent chemicals, whereas thresholds derived from the DrugMatrix study were 96%–97% predictive for the TG-GATES study. Threshold values derived from an abridged gene list (2/biomarker) also exhibited high predictive accuracy (91%–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 health outcomes.