Genome-wide association study of blood mercury in European pregnant women and children

Abstract

BackgroundMercury (Hg) is a toxic heavy metal which humans are most commonly exposed to through food chain contamination, especially via fish consumption. Even low-level exposure can be harmful because of the poor clearance rate, particularly for methylmercury. It is likely that genetic variation modifies exposure through changes in the absorption, metabolism, and/or removal of mercury. Associations have been reported between Hg and variants at multiple genetic loci, but in many cases these results are not yet replicated. MethodsThis study included two populations: pregnant women from the Avon Longitudinal Study of Parents and Children (ALSPAC, n=2,893) and children from the Human Early Life Exposome (HELIX, n=1,042). Genome-wide testing by cohort was performed by fitting linear regressions models on whole blood Hg levels and Haplotype Reference Consortium imputed single-nucleotide polymorphisms (SNPs). SNP heritability was estimated using linkage disequilibrium (LD)-score regression, and the biological functions of the top variants were investigated using resources which aggregate prior literature. ResultsHg SNP heritability was estimated to be 24.0% (95% CI: 16.9% to 46.4%) for pregnant women. The number of genetic variants independently associated with whole blood mercury levels above a suggestive p-value threshold (P < 1x10-5) was 16 for pregnant women and 21 for children. However, none were replicated in both populations, nor did any pass a stronger genome-wide significant threshold (P < 5x10-8). Several suggestive variants had possible biological links to Hg such as rs146099921 in metal transporter SLC39A14, and two variants (rs28618224, rs7154700) in potassium voltage-gated channels genes. DiscussionThere was evidence for a considerable proportion of Hg variance being attributed to genome-wide variation in pregnant women. However, results between pregnant women and children were highly discordant which could reflect differences in metabolism and a gene-age interaction with Hg levels. There were a large number of SNPs suggestively associated with Hg levels, which likely include both true associations and false positives. These interim findings will be expanded following collaboration with additional study groups.