Aim: We aimed for an epigenome-wide identification of vitamin D-associated CpG sites in leukocyte DNA.
Materials & methods: Infinium HumanMethylation450BeadChip measurements in 402 Caucasian older men were evaluated for significant association with 25-hydroxy-vitamin (25(OH)D) using Spearman's correlation and median regression to adjust for confounding variables. A cross-validation approach as well as a bootstrapping procedure were implemented to determine the replicability of significant associations. Multiple testing was corrected for by Benjamini-Hochberg or Bonferroni.
Results: Although in the screening subcohorts significant associations of DNAm with 25(OH)D were observed in the validation cohorts these associations were not replicated after adjustment for potential confounders. At none of the 361,945 CpGs a significant association of DNAm with 25(OH)D was found in all 100 random bootstrap samples, but in comparison at 462 CpGs for the well-established association with age.
Conclusion: Leukocyte DNAm was not associated with 25(OH)D levels after validation and consideration of confounders.
Keywords: 25(OH)D; DNA methylation; bootstrapping; cohort study; cross-validation; elderly; epigenetics; humans; vitamin D.