Affiliation:
1. 1Department of Epidemiology and Biostatistics, School of Public Health, Peking University, Beijing, China
2. 2Qingdao Center for Disease Control and Prevention, Qingdao, China
3. 3Zhejiang Center for Disease Control and Prevention, Hangzhou, China
4. 4Jiangsu Center for Disease Control and Prevention, Nanjing, China
5. 5Sichuan Center for Disease Control and Prevention, Chengdu, China
6. 6Heilongjiang Center for Disease Control and Prevention, Harbin, China
Abstract
Investigators of previous cross-sectional epigenome-wide association studies (EWAS) in adults have reported hundreds of 5′-cytosine-phosphate-guanine-3′ (CpG) sites associated with type 2 diabetes mellitus (T2DM) and glycemic traits. However, the results from EWAS have been inconsistent, and longitudinal observations of these associations are scarce. Furthermore, few studies have investigated whether DNA methylation (DNAm) could be modified by smoking, drinking, and glycemic traits, which have broad impacts on genome-wide DNAm and result in altering the risk of T2DM. Twin studies provide a valuable tool for epigenetic studies, as twins are naturally matched for genetic information. In this study, we conducted a systematic literature search in PubMed and Embase for EWAS, and 214, 33, and 117 candidate CpG sites were selected for T2DM, HbA1c, and fasting blood glucose (FBG). Based on 1,070 twins from the Chinese National Twin Registry, 67, 17, and 16 CpG sites from previous studies were validated for T2DM, HbA1c, and FBG. Longitudinal review and blood sampling for phenotypic information and DNAm were conducted twice in 2013 and 2018 for 308 twins. A cross-lagged analysis was performed to examine the temporal relationship between DNAm and T2DM or glycemic traits in the longitudinal data. A total of 11 significant paths from T2DM to subsequent DNAm and 15 paths from DNAm to subsequent T2DM were detected, suggesting both directions of associations. For glycemic traits, we detected 17 cross-lagged associations from baseline glycemic traits to subsequent DNAm, and none were from the other cross-lagged direction, indicating that CpG sites may be the consequences, not the causes, of glycemic traits. Finally, a longitudinal mediation analysis was performed to explore the mediation effects of DNAm on the associations of smoking, drinking, and glycemic traits with T2DM. No significant mediations of DNAm in the associations linking smoking and drinking with T2DM were found. In contrast, our study suggested a potential role of DNAm of cg19693031, cg00574958, and cg04816311 in mediating the effect of altered glycemic traits on T2DM.
Funder
National Natural Science Foundation of China
Publisher
American Diabetes Association
Subject
Endocrinology, Diabetes and Metabolism,Internal Medicine