Background and Aims Adiponectin, an adipose-secreted protein that has been linked to insulin sensitivity, plasma lipids, and inflammatory patterns, is an established biomarker for metabolic health. Despite clinical relevance and high heritability, the determinants of plasma adiponectin levels remain poorly understood. Methods and Results We conducted the first epigenome-wide cross-sectional study of adiponectin levels using methylation data on 368,051 cytosine-phosphate-guanine (CpG) sites in CD4+ T-cells from the Genetics of Lipid Lowering Drugs and Diet Network (GOLDN, n= 991). We fit linear mixed models, adjusting for age, sex, study site, T-cell purity, and family. We have identified a positive association (regression coefficient ± SE= 0.01 ± 0.001, P=3.4x10−13) between plasma adiponectin levels and methylation of a CpG site in CPT1A, a key player in fatty acid metabolism. The association was replicated (n=474, P=0.0009) in whole blood samples from the Amish participants of the Heredity and Phenotype Intervention (HAPI) Heart Study as well as White (n=592, P=0.0005) but not Black (n=243, P=0.18) participants of the Bogalusa Heart Study (BHS). The association remained significant upon adjusting for BMI and smoking in GOLDN and HAPI but not BHS. We also identified associations between methylation loci in RNF145 and UFM1 and plasma adiponectin in GOLDN and White BHS participants, although the association was not robust to adjustment for BMI or smoking. Conclusion We have identified and replicated associations between several biologically plausible loci and plasma adiponectin. These findings support the importance of epigenetic processes in metabolic traits, laying the groundwork for future translational applications.

Two genetic variants (rs3798220 and rs10455872) in the apolipoprotein (a) gene (LPA) have been implicated in cardiovascular disease (CVD), presumably through their association with lipoprotein (a) [Lp(a)] levels. While Lp(a) is recognized as a lipoprotein with atherogenic and thrombogenic characteristics, it is unclear whether or not the two Lp(a)-associated genetic variants are also associated with markers of thrombosis (i.e., plasminogen levels and fibrinolysis). In the present study, we genotyped the two genetic variants in 2919 subjects of the Old Order Amish (OOA) and recruited 146 subjects according to the carrier and noncarrier status for rs3798220 and rs10455872, and also matched for gender and age. We measured plasma Lp(a) and plasminogen levels in these subjects, and found that the concentrations of plasma Lp(a) were 2.62- and 1.73-fold higher in minor allele carriers of rs3798220 and rs10455872, respectively, compared with noncarriers (P = 2.04 × 10−17 and P = 1.64 × 10−6, respectively). By contrast, there was no difference in plasminogen concentrations between carriers and noncarriers of rs3798220 and rs10455872. Furthermore, we observed no association between carrier status of rs3798220 or rs10455872 with clot lysis time. Finally, plasminogen mRNA expression in liver samples derived from 76 Caucasian subjects was not significantly different between carriers and noncarriers of these two genetic variants. Our results provide further insight into the mechanism of action behind two genetic variants previously implicated in CVD risk and show that these polymorphisms are not major modulating factors for plasma plasminogen levels and fibrinolysis.