Toxicants Associated with Spontaneous Abortion in the Comparative Toxicogenomics Database (CTD)

Abstract

BackgroundUp to 70% of all pregnancies result in either implantation failure or spontaneous abortion (SA). Many events occur before women are aware of their pregnancy and we lack a comprehensive understanding of high-risk SA chemicals. In epidemiologic research, failure to account for a toxicants impact on SA can also bias toxicant-birth outcome associations. Our goal was to identify chemicals with a high number of interactions with SA genes, based on known toxicogenomic responses.\n\nMethodsWe used reference SA (MeSH: D000022) and chemical gene lists from the Comparative Toxicogenomics Database in three species (human, mouse, and rat). We prioritized chemicals (n=25) found in maternal blood/urine samples or in groundwater, tap water, or Superfund sites. For chemical-disease gene sets of sufficient size (n=13 chemicals, n=20 comparisons), chi-squared enrichment tests and proportional reporting ratios (PRR) were calculated. We then cross-validated enrichment results. Finally, among the SA genes, we assessed enrichment for gene ontology biological processes and for chemicals associated with SA in humans, we visualized specific gene-chemical interactions.\n\nResultsThe number of unique genes annotated to a chemical ranged from 2 (bromacil) to 5,607 (atrazine), and 121 genes were annotated to SA. In humans, all chemicals tested were highly enriched for SA gene overlap (all p<0.001; parathion PRR=7, cadmium PRR=6.5, lead PRR=3.9, arsenic PRR=3.5, atrazine PRR=2.8). In mice, highest enrichment (p<0.001) was observed for naphthalene (PRR=16.1), cadmium (PRR=12.8), arsenic (PRR=11.6), and carbon tetrachloride (PRR=7.7). In rats, we observed highest enrichment (p<0.001) for cadmium (PRR=8.7), carbon tetrachloride (PRR=8.3), and dieldrin (PRR=5.3). Our findings were robust to 1,000 permutations each of gene sets ranging in size from 100 to 10,000. SA genes were overrepresented in biological processes: inflammatory response (q=0.001), collagen metabolic process (q=1x10-13), cell death (q=0.02), and vascular development (q=0.005).\n\nConclusionWe observed chemical gene sets (parathion, cadmium, naphthalene, carbon tetrachloride, arsenic, lead, dieldrin, and atrazine) were highly enriched for SA genes. Exposures to chemicals linked to SA, thus linked to probability of live birth, may deplete fetuses susceptible to adverse birth outcomes. Our findings have critical public health implications for successful pregnancies as well as the interpretation of environmental pregnancy cohort analyses.