Epigenome-wide Association Study of Change in Blood Metabolite Levels From Young- to Middle Adulthood in the Northern Finland Birth Cohort 1966

摘要

Several associations between metabolite levels and DNA methylation (DNAm) have been reported. However, the relationship between longitudinal changes in metabolite levels and differential DNAm is underexplored. We assessed associations between epigenome-wide blood DNAm and change in blood metabolomics-based variables. For 595 non-diabetic individuals from the Northern Finland Birth Cohort 1966 for whom nuclear magnetic resonance-based metabolomics data were available at both ages 31 (Tl) and 46 (T2) as well as concurrent blood DNAm data at T2, we calculated for each of the 228 metabolomics-based variables the average change in level per year between Tl and T2. We used our methylSCOPA software, which enables both longitudinal and multi-phenotype epigenome-wide association studies (EWAS), for single-phenotype EWAS of change residuals - corrected for sex - for each metabolomic variable versus the degree of DNAm for 832,569 markers on the Illumina (San Diego, CA) MethylationEPIC BeadChip. We quality-controlled, residualized, and normalized the DNAm data, and mapped genomic locations to CGCh37/hgl9. We detected 67 epigenome-wide significant (P < 1 x 10~-7) associations between a DNAm site and change in the level of a metabolomic variable, involving 28 unique DNAm sites and 53 unique metabolomic variables. Nine DNAm sites associated significantly with change in the levels of multiple metabolomic variables, and change in the levels of five metabolomic variables associated with multiple DNAm sites. When looked together, effects of these 28 DNAm sites formed four robust clusters of effects on metabolite levels. Using a novel powerful methylSCOPA approach, we demonstrated that longitudinal changes in blood metabolite levels are associated with DNAm.