Abstract
AIMS: Environmental factors can alter DNA methylation (DNAm) levels, influence gene expression, and then change fasting glucose (FG) or hemoglobin A1c (HbA1c). METHODS: Through analyzing DNAm data of 2366 Taiwan Biobank individuals aged between 30 and 70 years, I evaluated the role of DNAm in mediating the associations of seven non-genetic factors (BMI, chronological age, sex, smoking, drinking alcoholic beverages, education, and regular exercise) with FG and HbA1c. RESULTS: Among 846,232 Cytosine-phosphate-Guanine (CpG) sites, the single-mediator model explored that 21, 15, 10, 3, 3, and 1 CpGs significantly mediated (p < 6.6E-9) the BMI-HbA1c, BMI-FG, sex-FG, age-FG, age-HbA1c, and drinking-HbA1c associations, respectively. The multiple-mediator model considered all significant mediators and selected the model with the smallest Akaike Information Criterion, and identified 8 CpGs that linked exposures (BMI, sex, age, and drinking) to diabetes indicators. Seven out of the 8 CpGs have been reported to be associated with diabetes, FG, HbA1c, or insulin resistance in previous epigenome-wide association studies. CONCLUSION: Four of the 8 CpGs (cg19693031, cg04816311, cg00574958, and cg11024682) were associated with the expression of genes implicated in diabetes and metabolism, including the TXNIP, GPR146, CPT1A, and SREBF1 genes. These findings highlight the underlying epigenetic mechanism linking non-genetic factors with diabetes.