Epigenome-wide meta-analysis of prenatal vitamin D insufficiency and cord blood DNA methylation

Abstract

Low maternal vitamin D concentrations during pregnancy have been associated with a range of offspring health outcomes. DNA methylation is one mechanism by which the maternal vitamin D status during pregnancy could impact offspring's health in later life. We aimed to evaluate whether maternal vitamin D insufficiency during pregnancy was conditionally associated with DNA methylation in the offspring cord blood. Maternal vitamin D insufficiency (plasma 25-hydroxy vitamin D $\le$ 75 nmol/L) during pregnancy and offspring cord blood DNA methylation, assessed using Illumina Infinium 450k or Illumina EPIC Beadchip, was collected for 3738 mother-child pairs in 7 cohorts as part of the Pregnancy and Childhood Epigenetics (PACE) consortium. Associations between maternal vitamin D and offspring DNA methylation, adjusted for fetal sex, maternal smoking, maternal age, maternal pre-pregnancy or early pregnancy BMI, maternal education, gestational age at measurement of 25(OH)D, parity, and cell type composition, were estimated using robust linear regression in each cohort, and a fixed-effects meta-analysis was conducted. The prevalence of vitamin D insufficiency ranged from 44.3% to 78.5% across cohorts. Across 364,678 CpG sites, none were associated with maternal vitamin D insufficiency at an epigenome-wide significant level after correcting for multiple testing using Bonferroni correction or a less conservative Benjamini-Hochberg False Discovery Rate approach (FDR, p > 0.05). In this epigenome-wide association study, we did not find convincing evidence of a conditional association of vitamin D insufficiency with offspring DNA methylation at any measured CpG site.

Keywords: DNA methylation; EWAS; PACE; Vitamin D; Vitamin D insufficiency; epigenetics; pregnancy.

Figure 1.

Meta-analysis of the association between maternal mid-pregnancy vitamin D and offspring DNA methylation, adjusted for fetal sex, maternal smoking, maternal age, maternal pre-pregnancy BMI, maternal education, gestational age at measurement, parity, cell type composition, principal components of ancestry, and, where appropriate, technical covariates and study design covariates. A total of 364,678 CpG sites were included in this analysis. No sites were epigenome-wide significantly associated with maternal vitamin D insufficiency using either a Bonferroni correction for multiple testing (solid line) or a Benjamini–Hochberg correction corresponding to a false discovery rate of 0.05.

Figure 2.

Meta-analysis of the association between maternal mid-pregnancy vitamin D and offspring DNA methylation, adjusted for the season of measurement, fetal sex, maternal smoking, maternal age, maternal pre-pregnancy BMI, maternal education, gestational age at measurement, parity, cell type composition, principal components of ancestry, and, where appropriate, technical covariates and study design covariates. No sites were epigenome-wide significantly associated with maternal vitamin D using either a Bonferroni correction for multiple testing (solid line) or a Benjamini–Hochberg correction corresponding to a false discovery rate of 0.05.