Incorporating epigenetic information in a Mendelian randomization framework.

<p>(A) Alcohol exposure is associated with risk of head and neck squamous cell carcinoma (HNSCC) and this may be mediated by altered DNA methylation (CpG). The relationship between alcohol exposure and HNSCC is potentially confounded by factors such as socio-economic position, which correlate with both exposure and disease. A common variant in ADH1B can be used as an unconfounded, genetic proxy for alcohol exposure, and if this SNP is associated with CpG (either locally or more widely across the genome), it would lend support to the hypothesis that alcohol intake causally influences DNA methylation. However, showing associations of these epigenetic measures with HNSCC does not demonstrate causality of either alcohol or CpG on HNSCC, as either or both associations (alcohol→HNSCC and CpG→HNSCC) could be confounded or alcohol could influence HNSCC through another pathway (dashed line). (B) To investigate this, another Mendelian randomization experiment could be undertaken using an SNP known to have a <i>cis</i> influence on loci-specific DNA methylation. If an association were observed between this SNP and both CpG and HNSCC, this would support a role for DNA methylation in the causation of HNSCC.