Epigenome-wide association study of global cortical volumes in generation Scotland: Scottish family health study

Abstract

A complex interplay of genetic and environmental risk factors influence global brain structural alterations associated with brain health and disease. Epigenome-wide association studies (EWAS) of global brain imaging phenotypes have the potential to reveal the mechanisms of brain health and disease and can lead to better predictive analytics through the development of risk scores.We perform an EWAS of global brain volumes in Generation Scotland using peripherally measured whole blood DNA methylation (DNAm) from two assessments, (i) at baseline recruitment, ~6 years prior to MRI assessment (N = 672) and (ii) concurrent with MRI assessment (N=565). Four CpGs at baseline were associated with global cerebral white matter, total grey matter, and whole-brain volume (Bonferroni p≤7.41×10^-8, β_range = -1.46x10^-6 to 9.59 × 10^-7). These CpGs were annotated to genes implicated in brain-related traits, including psychiatric disorders, development, and ageing. We did not find significant associations in the meta-analysis of the EWAS of the two sets concurrent with imaging at the corrected level.These findings reveal global brain structural changes associated with DNAm measured ~6 years previously, indicating a potential role of early DNAm modifications in brain structure. Although concurrent DNAm was not associated with global brain structure, the nominally significant findings identified here present a rationale for future investigation of associations between DNA methylation and structural brain phenotypes in larger population-based samples.

Keywords: DNA methylation; cortical volumes; epigenome-wide association study; generation Scotland.

Figure 1.

Manhattan plots showing the results from EWASs of cerebral white matter (1A), total grey matter (1B), and whole-brain volume (1 C), using baseline DNAm data (N = 672). The black line defines the threshold for epigenome-wide significance (p ≤ 7.41x10⁻⁸) and the dotted line defines CpG sites at p ≤ 1x10⁻⁵. Epigenome-wide significant hits for each phenotype are labelled on the graph.

Figure 2.

Manhattan plots showing meta-analysis of EWAS of cerebral white matter (2A), total grey matter (2B), and whole-brain volume (2 C), across the 2 concurrent sets ^(Nset 1^{=331; N}set 2^{=234; N}total⁼⁵⁶⁵⁾. The black line defines the threshold for epigenome-wide significance (p ≤ 6.5x10⁻⁸) and the dotted line defines p ≤ 1x10⁻⁵. CpGs that met a significance of p ≤ 1x10⁻⁵ are labelled on the graph.

Figure 3.

Power curves for cerebral white matter, total grey matter, and whole-brain volume calculated separately for set 1 and set 2. The x-axis indicates how many participants would be needed to detect an effect with 60%, 80%, 90%, 95% or 99% power at p < 6.51x10⁻⁸ (set 1 (W1)) and p < 6.53x10⁻⁸ (set 2 (W2)) with 36 regression coefficients included in the linear model. Effect sizes were calculated based on the largest effect size obtained in EWAS for each phenotype at baseline.