Epigenome-wide Association Studies and the Interpretation of Disease -Omics

Abstract

Epigenome-wide association studies represent one means of applying genome-wide assays to identify molecular events that could be associated with human phenotypes. The epigenome is especially intriguing as a target for study, as epigenetic regulatory processes are, by definition, heritable from parent to daughter cells and are found to have transcriptional regulatory properties. As such, the epigenome is an attractive candidate for mediating long-term responses to cellular stimuli, such as environmental effects modifying disease risk. Such epigenomic studies represent a broader category of disease -omics, which suffer from multiple problems in design and execution that severely limit their interpretability. Here we define many of the problems with current epigenomic studies and propose solutions that can be applied to allow this and other disease -omics studies to achieve their potential for generating valuable insights.

Fig 1. An overview of considerations in designing and interpreting epigenome-wide association studies (EWAS).

In (a) we explicitly define the cellular hypothesis being tested in EWAS: that there are changes in epigenetic mediators of transcriptional regulation (denoted by gains or losses of methyl groups) that distinguishes a canonical cell type in individuals with a phenotype (green) from those without the phenotype (blue). The EWAS is frequently performed to address the idea that the epigenetic dysregulation is occurring as a response to a cellular exposure or stress. In a situation of reverse causation, the reason the epigenetic change is observed in association with a phenotype is because the phenotype induces the epigenetic change, rather than the other way around. In (b) we describe three study designs: the typical cross-sectional design comparing individuals with and without the phenotype of interest, and two designs that reduce the effects of genetic polymorphism, which are studies of monozygotic twins discordant for a phenotype, and longitudinal studies of people before and after they develop the phenotype.