Myocardial infarction patients had decreased methylation at four growth differentiating factor-15 (GDF-15) related CpG sites (cg13033858, cg16936953, cg17150809, and cg18608055). These sites had not been studied for their association with cardiovascular disease (CVD) deaths. Thus, we aimed to assess the associations independent of genes, shared environment, and traditional CVD risk factors. Nineteen white, male, monozygotic twin pairs discordant for CVD deaths were included from the National Heart, Lung and Blood Institute Twin Study (NHLBI) initiated in 1969. Data on vital status was collected through December 31, 2014. Methylation of buffy coat DNA at exam 3 (1986–87) was measured using the Illumina HumanMethylation450 BeadChip. Principal component analysis was used to generate a score representing blood leukocyte composition and baseline CVD risk factors and predominated with natural killer cells, CD4+ T cells, and Framingham risk score. Conditional logistic regression demonstrated that methylation at the four CpG sites was not associated with CVD deaths before (all p > 0.05, bootstrapped p > 0.05) and after adjustment for the score (all p > 0.05). Joint influences of cg16936953 and the score were statistically significant (p < 0.05). In conclusion, joint influences of methylation at the site cg16936953 and the score are prospectively associated with CVD deaths independent of germline and common environment.ClinicalTrials.gov Identifier for NHLBI Twin Study: NCT00005124.

Background: Growth differentiation factor- 15 (GDF-15) is positively associated with the risk of mortality from acute coronary syndrome and chronic heart failure. A prior study reported decreased methylation at four promising GDF-15 related CpG sites tended to be associated with myocardial infarction, however, little is known of the association of methylation levels at these sites with the risk for cardiovascular disease (CVD) death. Objective: To evaluate whether methylation levels at the four GDF-15 CpG sites (site A: cg13033585, site B: cg16936953, site C: cg17150809, and site D: cg18608055) are associated with death from CVD, independent of genes and shared environmental factors. Method: We included 19 male monozygotic twin pairs discordant for death from CVD through December 31, 2014 from the National Heart, Lung, and Blood Institute (NHLBI) Twin Study initiated in 1969-1973. Buffy coat DNA samples were collected in exam 3 (1986-87). The vital status was followed up through December 31, 2014. Genome wide DNA methylation levels were quantified using the Illumina Infinium HumanMethylation450 (450K) BeadChip. Conditional logistic models were used to estimate hazard ratio (HR). Known baseline CVD risk factors were adjusted. Results: The twins’ mean baseline age was 50.4 years with standard deviation of 2.4. The crude HR was 0.01 (95% CI: 0.00, 2854.46), 2038.89 (95%CI: 0.01, 3.84 X 10 8 ), 0.12 (95% CI: 0.00, 55.99), and 2.08 (95% CI: 0.00, 1.26 X 10 6 ) for sites A, B, C, and D, respectively, suggesting that our sample size was small to test these sites. After adjustment for body mass index, years of education, and Framingham risk scores, HR was 0.03 (95% CI: 0.00, 45281.06) for site A, 700.96 (95% CI: 0.00, 2.32 X10 8 ) for site B, 0.00 (95% CI: 0.00, 7.35) for site C, and 0.81 (95% CI: 0.00, 1.44 X 10 6 ) for site D. Further adjustment for white blood cell subtypes dramatically changed HRs and/or largely widened 95% CIs, suggesting potential overadjustment bias: 0.01 (95% CI:0.00, 1.89 X 10 10 ) for site A, 2.78 X 10 8 (95% CI: 0.00, 7.84 X 10 21 ) for site B, 0.00 (95% CI: 0.00, 620.94) for site C, and 1.69 (95% CI: 0.00, 4.18 X 10 13 ) for site D. Conclusion: DNA methylation levels at the GDF-15 CpG sites are not associated with death risk from cardiovascular disease independent of genes and shared environment.