Skip to main content
Download PDF

Associations between DNA methylation and BMI vary by metabolic health status: a potential link to disparate cardiovascular outcomes

Clinical Epigenetics volume 13, Article number: 230 (2021) Cite this article

Abstract

Background

Body mass index (BMI), a well-known risk factor for poor cardiovascular outcomes, is associated with differential DNA methylation (DNAm). Similarly, metabolic health has also been associated with changes in DNAm. It is unclear how overall metabolic health outside of BMI may modify the relationship between BMI and methylation profiles, and what consequences this may have on downstream cardiovascular disease. The purpose of this study was to identify cytosine-phosphate-guanine (CpG) sites at which the association between BMI and DNAm could be modified by overall metabolic health.

Results

The discovery study population was derived from three Women’s Health Initiative (WHI) ancillary studies (n = 3977) and two Atherosclerosis Risk in Communities (ARIC) ancillary studies (n = 3520). Findings were validated in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort (n = 1200). Generalized linear models regressed methylation β values on the interaction between BMI and metabolic health Z score (BMI × MHZ) adjusted for BMI, MHZ, cell composition, chip number and location, study characteristics, top three ancestry principal components, smoking, age, ethnicity (WHI), and sex (ARIC). Among the 429,566 sites examined, differential associations between BMI × MHZ and DNAm were identified at 22 CpG sites (FDR q < 0.05), with one site replicated in MESA (cg18989722, in the TRAPPC9 gene). Three of the 22 sites were associated with incident coronary heart disease (CHD) in WHI. For each 0.01 unit increase in DNAm β value, the risk of incident CHD increased by 9% in one site and decreased by 6–10% in two sites over 25 years.

Conclusions

Differential associations between DNAm and BMI by MHZ were identified at 22 sites, one of which was validated (cg18989722) and three of which were predictive of incident CHD. These sites are located in several genes related to NF-kappa-B signaling, suggesting a potential role for inflammation between DNA methylation and BMI-associated metabolic health.

Background

Obesity rates continue to rise with obesity occurring in more than 41.1% of women in the USA in 2016 [1]. While obesity is most typically defined as body mass index (BMI) > 30 kg/m2, limitations in the use of BMI have been noted, including variation in associations with health outcomes by race/ancestry, physical activity, and age [2, 3], as well as some reports finding no association between higher-risk categories (overweight and middle obesity) with mortality [4, 5]. These conflicting reports have motivated several studies to examine whether differential phenotypes of obesity exist and whether examining BMI in isolation of additional metabolic health parameters is a sufficient metric of overall health.

A growing body of evidence has found heterogeneity in obesity, with some phenotypes exhibiting differential risk for cardiovascular outcomes. Metabolically healthy obesity (MHO) has been defined as obesity with less than two or three metabolic health risk factors. Some but not all studies have found MHO to be associated with reduced risk of cardiovascular outcomes compared to metabolically unhealthy obesity (MUO) [6,7,8,9,10,11]. In a recent systematic review, MHO had a higher risk of cardiovascular events than metabolically healthy, normal weight participants (risk ratio [RR] 1.45, 95% CI 1.20–1.70, reference metabolically healthy, normal weight), but had lower risk to metabolically unhealthy normal weight (RR 2.07, 95% CI 1.62–2.65, reference metabolically healthy, normal weight) and MUO individuals (RR 2.31, 95% CI 1.99–2.69, reference metabolically healthy, normal weight) [12]. These findings suggest that metabolic health status may differentially influence the relationship between BMI and health outcomes. Examining the molecular underpinnings of this phenotype may guide our understanding of this epidemiological phenomenon by identifying the biological mechanisms which may be leading to a reduction in risk of health outcomes associated with obesity. Additionally, identifying biomarkers of MHO, particularly if they can identify individuals more likely to remain in MHO, would be advantageous for more targeted interventions.

Epigenetic mechanisms, such as DNA methylation (DNAm), are important biological features to examine in the context of chronic diseases such as obesity and metabolic health. Changes to DNAm can induce changes in gene expression in causal disease pathways potentially mediating or modifying differential health outcomes [13]. Obesity has been widely examined and shown to associate with prolific methylation changes in the blood and adipose tissue [14,15,16]. Similarly, metabolic syndrome and metabolic health risk factors have been found to associate with differential methylation [17,18,19,20,21]. Indeed the mouse model which is used to represent MHO is developed from deletion of the BRD2 gene, which is a primary epigenetic regulator of histone acetylation [22]. However, no studies have integrated these phenotypes to examine how BMI-associated methylation varies by metabolic health status. Particularly since DNAm has been reported to mediate the relationship between obesity and increased cardiovascular outcomes [23], evaluating the epigenome may provide insight into pathways contributing to the differences in outcomes. Thus, the purpose of this study is to examine whether BMI associates with methylation differentially according to metabolic health status (Fig. 1).

Fig. 1
figure 1

Conceptual framework research question. Abbreviations: BMI, body mass index

Full size image

Results

A summary of the methods is included in Fig. 2.

Fig. 2
figure 2

Summary of the analyses. Abbreviations: WHI, Women’s Health Initiative; EMPC, Epigenetic Mechanisms of Particulate Matter-Mediated Cardiovascular Disease; BAA23, the Integrative Genomics for Risk of Coronary Heart Disease and Related Phenotypes in WHI cohort; AS311, Bladder Cancer and Leukocyte Methylation; ARIC, Atherosclerosis Risk in Communities study; BMI, body mass index; CHD, coronary heart disease; MESA, Multi-Ethnic Study of Atherosclerosis; GTP, Grady Trauma Project

Full size image

Demographic characteristics of the cohorts are described in Table 1 for the three ancillary studies from the Women’s Health Initiative (WHI) including Epigenetic Mechanisms of Particulate Matter-Mediated Cardiovascular Disease (EMPC, aka AS315), the Integrative Genomics for Risk of Coronary Heart Disease and Related Phenotypes in WHI cohort (BAA23), and Bladder Cancer and Leukocyte Methylation (AS311) and the two ancillary studies from the Atherosclerosis Risk in Communities study (ARIC) including European Americans (EA) and African Americans (AA). Overall, 7497 participants were included in the discovery analysis. To examine the differential impact of metabolic health status on BMI, linear regression models were used regressing the methylation β value on the interaction term for BMI and metabolic health status, adjusting for each higher-level variable (BMI and metabolic health) and covariates. We conducted two epigenome-wide assocation study (EWAS) with metabolic health status defined dichotomously (BMI × MH) and continuously (BMI x metabolic health Z-score [MHZ]). We identified no statistically significant differential associations between cytosine and guanine nucleotide pair (CpG) methylation and BMI by dichotomized metabolic health status (BMIxMH). When metabolic health status was examined continuously (MHZ), 22 CpG sites were associated with BMIxMHZ (false discovery rate [FDR] q value < 0.05, Table 2; Fig. 3). For ease of interpretation, we described the direction of effect in the 22 significant sites in the models examining BMI × MH  (Additional file 1: Table S1). In 13 of the 22 sites, an increase in BMI was associated with an opposite direction of effect in the coefficient in metabolically healthy vs unhealthy individuals. In the replication analysis in the Multi-Ethnic Study of Atherosclerosis (MESA), cg18989722 associated with BMIxMHZ (p < 0.05 in a consistent direction, Additional file 1: Table S2). cg18989722 inversely associated with BMI × MHZ. When examining BMI × MH in this site, every unit increase in BMI was associated with increased DNAm in metabolically healthy individuals and decreased methylation in metabolically unhealthy individuals. We examined a gene ontology analysis of the 22 significant sites. However, no pathways were significantly enriched.

Table 1 Demographic characteristics of each ancillary study in the Women’s Health Initiative (WHI) and the Atherosclerosis Risk in Communities (ARIC)
Full size table
Table 2 Significant sites associated with body mass index * metabolic health Z score interaction (BMIxMHZ) including the effect size for the higher-level variables
Full size table
Fig. 3
figure 3

Manhattan plot of the association between the interaction of BMI and metabolic health Z score and DNA methylation. Significant sites identified as those above the red line (FDR < .05; p < 2 × 10–6)

Full size image

Given the known relationships between obesity and metabolic status and cardiovascular disease, we examined whether DNAm taken at baseline predicted incident myocardial infarction over 25 years in the WHI. After excluding individuals from WHI with a history of cardiovascular disease, 3746 individuals remained (BAA23 n = 1823, EMPC n = 1775, AS311 n = 148). In WHI, there were 714 events (number of events = 585 [BAA23], 113 [EMPC], 16 [AS311]) with an average follow up of 14–15 years (mean follow up = BAA23 14.0 years, EMPC 15.69 years, AS311 15.85 years). When predicting incident coronary heart disease (CHD), we initially examined whether the interaction between BMI × MHZ taken at baseline was associated with incident CHD, adjusting for BMI, MHZ, case–control status, age, smoking status, and ethnicity. BMIxMHZ was significantly associated with incident CHD (hazard ratio [HR] 1.02, 95% CI 1.004, 1.03, p value = 0.005, Fig. 4A). Then we added the DNA methylation at cg18989722 as a predictor and found that it was not associated with incident CHD (Additional file 1: Table S3). However, when examining the 22 sites from the discovery analysis, three sites were associated with incident CHD adjusted for the reduced set of covariates (p < 0.05, Fig. 4B–D; Table 3). In models adjusting for the full set of covariates, two sites were significantly associated with CHD, cg16461485 and cg16543390.

Fig. 4
figure 4

Probability of incident CHD by tertile of the interaction between body mass index (BMI) and metabolic health Z score (BMI × MHZ) (A) and CpG site methylation of cg16461485 (B), cg20210586 (C), and cg02851049 (D) over 25 years in the Women’s Health Initiative (WHI)

Full size image
Table 3 Significant CpG sites associated with incident coronary heart disease (CHD) in the Women’s Health Initiative (WHI) over 25 years
Full size table

For the replicated site and the three sites associated with incident CHD in the reduced set of covariates, we examined whether they were associated with differential gene expression in the blood in two cohorts: Grady Trauma Project (GTP) and MESA (Additional file 1: Table S4). None of the sites were reported to be associated with gene expression in cis. cg18989722 (in chromosome 8) was associated with differential expression of PTGS1 in chromosome 9 and cg16461485 (in chromosome 3) was associated with differential expression of TNFRSF13B in chromosome 17 representing trans associations.

Table 4 Adult treatment panel (ATP) III clinical identification of metabolic syndrome
Full size table

In sensitivity analyses, we examined the change in the association of BMI × MHZ with the 22 significant CpG sites in the WHI and ARIC when adjusted for physical activity. After including physical activity as a covariate, BMI × MHZ was no longer significantly associated with any CpG site. However, this may be in part due to a reduction in power as the effect sizes did not change significantly in the significant sites (correlation in effect sizes from main analysis compared to analysis adjusted for physical activity = 0.92). When examining the change in the effect size when sequentially leaving out each of the metabolic health parameters one at a time from the MHZ score, omission of high-density lipoprotein cholesterol (HDL) most potently attenuated the effect estimate (effect estimate correlation = 0.89, Additional file 1: Table S5). In the 22 sites, effect sizes and Z scores changed minimally when adjusted for lipid, hypertension, and glycemic medications (all correlations in effect size = 0.97 and all correlations in Z scores = 0.95). We examined the influence of individual ancillary studies on the results by examining the change in significance and effect size when ancillary studies (BAA23, EMPC, AS311, ARIC AA, and ARIC EA) were individually excluded from the analysis. Significance changed moderately with exclusion of each study with 30, 20, 25, 20 and 27 significant sites when BAA23, EMPC, AS311, ARIC AA, and ARIC EA were excluded, respectively. Differences in effect size were minor (correlation with main analysis = 0.99) in all studies except with exclusion of ARIC AA (correlation with main analysis = 0.86, Additional file 1: Figures S1–S5). We additionally examined the residuals of the four sites described above in the EWAS models to assess violations of a non-normal distribution (Additional file 1: Figures S6–S13). These plots appear to be consistent with a normal distribution, with the exception of ARIC EA. When ARIC EA is excluded from the analysis, all 22 remain significant and Z scores tend to be higher as compared to the full analysis.

Discussion

In this study, we found 22 CpG sites were associated with BMIxMHZ in the WHI and ARIC cohorts, with one site replicating in a consistent direction in MESA. Among the 22 sites, two CpG sites inversely and one CpG site positively associated with a change in incidence of CHD over 25 years in the WHI cohort.

One site replicated in MESA in a consistent direction, cg18989722 located in the body of the TRAPPC9 gene. TRAPPC9 has been shown to play a role in NF-kappa-B signaling by activating NF-kappa-B through increased phosphorylation of the IKK complex [24]. TRAPPC9 encodes NIBP, which binds to IKK/NIK to enhance NF-kappa-B activation. TRAPPC9 has recently been identified as an imprinted gene primarily expressing the maternal allele (70% of transcripts in the brain expressed maternal allele). TRAPPC9 knock-out mice exhibit a rare intellectual disability accompanied by an increase in fat mass and body weight [25], suggesting that expression of this gene may protect against obesity. Several CpG sites in TRAPPC9 have been identified in EWAS of childhood adiposity [26, 27]. As gene body methylation has often been cited as an indicator of an active gene [28], our findings are in alignment with previous reports of protection against obesity since individuals with lower BMIxMHZ had higher methylation in this site.

This site was also associated with increased gene expression of the PTGS1 gene. PTGS1 (also known as COX1) catalyzes the conversion of arachinodate to prostaglandin protein and is inhibited by anti-inflammatory drugs. In our sensitivity analysis, when adjusted for lipid medication use including peripheral vasodilators such as aspirin, the effect size moderately changed (− 8.36 × 10–5 in unadjusted models and − 5.25 × 10–5 in adjusted models). However, the Z score was smaller (− 6.54 in unadjusted models and − 2.55 in adjusted models). This suggests some attenuation in the relationship between BMI × MHZ and DNAm in this site is potentially modified by medication use.

Methylation in three sites was associated with incident CHD over 25 years in the WHI cohort: cg16461485 located in the body of SELT, cg02851049 located in the body of POLR3K, and cg20210586 in the body of TRIM39. None of these sites have been identified in previous EWAS. We also found that cg16461485 associated with reduced gene expression of TNFRSF13B, which encodes the tumor necrosis factor (TNF) receptor superfamily member 13B, also known as the transmembrane activator and CAML interactor (TACI). This protein activates NFAT, AP1, and NF-kappa-B [29]. TACI knock-out mice were protected against high-fat-diet-induced inflammation and dysglycemia, which may be mediated by a shift in adipose tissue macrophages from M1 to M2, which tend to promote a phenotype of insulin sensitivity [30]. These findings further support the role that methylation in cg16461485 exhibiting a protective effect.

Given the molecular functions of these genes, differential inflammatory mechanisms (potentially mediated by the NF-kappa-B pathway) may account for the observed differences in health outcomes by BMIxMHZ. Obesity and several of the metabolic health parameters associate with excess inflammation [31]. However, some studies have posited that MHO may be due to an uncoupling of obesity and insulin resistance due to differential inflammatory mechanisms as MHO has been associated with lower inflammatory markers including C-reactive protein (CRP), TNF-α, interleukin-6 (IL-6), and plasminogen activator inhibitor-1 [32, 33]. As the CpG sites identified in this study were adjusted for cell composition which uses surrogate measures from six cell types [34], we may be identifying unique immune cell subsets associated with these disease exposures not captured by this method which may drive differences in outcomes. Additionally, several studies have also observed a unique relationship between inflammatory markers and adiposity in individuals of African descent, where these markers do not appear to be as sensitive to adiposity compared to individuals of European descent [35, 36]. This may explain the significant differences observed when we exclude the ARIC AA cohort in sensitivity analyses.

When we adjusted for physical activity, no sites were significantly associated with BMI × MHZ. We included this analysis as several studies have touted that MHO may be the product of increased fitness in this population [36, 37]. This may be due to true confounding by physical activity. However, there may also be collinearity, as physical activity is highly associated with BMI (p = 2.5e−17) and metabolic health (p = 2.94e−12), which would lead to a reduction in significance in the identified variables. Future studies could further explore these associations in physical activity interventions in these populations.

There are several important limitations in this study. Given the cross-sectional design, we cannot determine any causal association and may be at risk of reverse causality, if methylation is contributing to changes in BMI or metabolic risk factors. Moreover, metabolic risk factors may also be a product of duration of obesity, since several studies have found MHO to be a transitory state [6, 7, 37]. However, understanding the methylomic differences in these populations would still be advantageous to identify biological mechanisms that may be driving the differences in outcomes. Another limitation includes the potential for confounding by cell composition. While we found unique relationships between three CpG sites and CHD, none of these sites replicated in an external population suggesting that other confounding factors may be causing this association. Nevertheless, the limited replication may be due in part to limited power as the replication analysis had the power to detect effect sizes as low as 0.01 and the effect sizes from our discovery EWAS were much lower (Additional file 1: Table S6). Additionally, the results may have diverged between ARIC and WHI populations since the WHI includes only women. While we adjusted for sex in ARIC, we may be identifying signals in WHI that are differential in women versus men. A strength of this study is examining the unique interaction between BMI and metabolic health in three population-based cohort studies and examining their impact on gene expression and CHD outcomes.

Overall, we found four CpG sites which may have a unique relationship with BMI in metabolically healthy vs unhealthy individuals. Our study findings may align with several studies suggesting that differential inflammatory mechanisms may account for differences in metabolic risk factors associated with increasing BMI. Future research studies could benefit from examining longitudinal changes in methylation associated with change in metabolic health status to determine the direction of effect and single cell epigenomic signatures of obesity and metabolic health to examine how individual cell profiles influence this relationship.

Methods

Study population

Two cohorts were used in the discovery phase: the WHI and the ARIC. Data from three WHI ancillary studies were included: EMPC, aka AS315, the BAA23, and AS311. EMPC assessed epigenetic mechanisms underlying associations between ambient particulate matter air pollution and cardiovascular disease within the WHI Clinical Trials (CT, n = 2200). BAA23 was a case–control study assessing predictors of coronary heart disease (CHD) within the WHI CT (n = 1664) and observational study (OS, n = 442), where cases were identified using eight biomarkers of CHD. AS311 is a matched case–control study of bladder cancer among women within the WHI CT (n = 405) and OS (n = 455) [38].

ARIC included data from two ancillary studies of AA and EA. ARIC is an ongoing prospective cohort study investigating the etiology of CHD in four US communities: Forsyth County, NC; Jackson, MS; Minneapolis, MN; Washington County; MD. Participants were aged 45–64 and followed up in each community over 30 years with 7 study visits [39, 40]. DNAm was measured in 2879 AA and 1100 EA participants from ARIC in visit 2 (1990–1992) or visit 3 (1993–1995).

The replication cohort derived from the MESA study. MESA is a longitudinal, population cohort study designed to examine risk factors for and the progression of CHD. Participants aged 45–84 years without clinically apparent CHD were recruited between July 2000 and August 2002 from six regions in the USA: Winston-Salem, NC; Northern New York, NY; Baltimore, MD; St. Paul, MN; Chicago, IL; and Los Angeles, CA. DNAm was derived from peripheral blood mononuclear cell samples at Exam 1 or Exam 5 in a random sample of 1200 non-Hispanic white, AA, Hispanic, and Chinese American participants [41, 42].

Measurements

In WHI, weight, height, waist circumference, and blood pressure (BP) were measured at the physical exam. In ARIC, these measurements were taken at Visit 2 or 3. BMI was calculated as weight (kg)/height (m)2. Waist circumference was measured to the nearest 0.5 cm. Systolic/diastolic BP was measured twice and three times in WHI and ARIC, respectively, with the average of the two (WHI) or last two (ARIC) measurements used. Biochemical measurements were analyzed in blood samples collected after a 12-h fast. These include triglycerides (TG), HDL, and fasting glucose.

Metabolic health exposures

Metabolic health was examined in two ways, dichotomously and continuously. Metabolic risk was dichotomously defined by the presence of three or more components of metabolic syndrome using the Adult Treatment Panel III (ATP III) criteria (Table 4). Thus, MUO and MHO referred to the presence of three or more and less than three components, respectively. Metabolic health was also examined continuously as a Z score of the clinical measures used in the ATP III criteria. For each metabolic parameter, for example TG, the Z score for TG was created by (TG − mean TG)/standard deviation (TG) of the population. Then all the clinical parameter Z scores were pooled to define a MHZ. For HDL, the inverse of HDL was used as a higher MHZ is indicative of poorer health. We examined the HDL variable for normality (Additional file 1: Figure S14–S15). We examined BMI continuously. Individuals were excluded if metabolic health parameters and DNAm were not measured within the same year.

Covariates

Age, race/ethnicity (White, AA, Hispanic/Latino, Asian, American Indian, other in WHI and EA and AA in ARIC), and smoking status (current/former or never) were self-reported. Physical activity was measured by the Baecke questionnaire in ARIC [43] and a self-administered questionnaire in WHI [44] and expressed as total energy expended from light, moderate, or vigorous intensity recreational physical activity which includes walking, mild, moderate, and strenuous physical activity in kcal/week/kg (MET-hours/week).

DNA methylation

In the all cohorts, DNA was extracted from peripheral blood leucocytes collected at visit-specific fasting blood draws [45]. In the WHI and ARIC cohorts, DNAm was measured using the Illumina HM450K Infinium Methylation BeadChip. In the MESA cohort, DNAm was measured via the Illumina MethylationEPIC BeadChip array. DNAm was estimated as the proportion of methylated beads relative to combined unmethylated and methylated beads for a specific CpG site defined as the β value (ranging from 0 [unmethylated] to 1 [methylated]). All methylation data were normalized using beta-mixture quantile normalization [46]. Technical covariates included chip and row to adjust for batch effects and cell composition, which was estimated using the reference-based Houseman method [34]. Additional quality control procedures for each of the studies has been included in the Additional file 1: Methods. After quality control, 428,278 probes remained in all cohorts and were examined.

Statistical analysis

We used R (https://www.r-project.org/) for all analyses. We calculated means and standard deviations or counts and proportions for study population characteristics. In the EWAS, all models were stratified by cohort (EMPC, BAA23, AS311 in WHI) or race (AA and EA in ARIC) and pooled using inverse-variance weighted fixed effect meta-analysis. BMI was examined continuously. To examine the differential impact of metabolic health status on BMI, linear regression models were used regressing the methylation β value on the interaction term for BMI and metabolic health status, adjusting for each higher-level variable (BMI and metabolic health) and covariates. We conducted two EWAS with metabolic health status defined dichotomously (BMIxMH) and continuously (BMIxMHZ). Covariates in all models included cell composition, the top three principal components of genetic relatedness, race/ethnicity (WHI), sex (ARIC), smoking status, row, and age. WHI study-specific covariates included trial study and randomization arm (EMPC, BAA23, AS311) and case–control status (BAA23, AS311). To adjust for batch effects, the DNAm array was included as a random effect for each BeadChip in our model. Significant CpG sites were identified by the interaction p value at a FDR q value < 0.05. We examined the correlation between BMI and MHZ to test for collinearity (Additional file 1: Figure S16). These variables were not highly correlated.

Results identified in the discovery cohorts were replicated in the MESA cohort using linear regression models as previously described. Significant CpG sites were examined using the same linear regression model as above examining BMI × MHZ. Models were adjusted for DNAm array number and row location, cell composition, principal components of genetic relatedness, race/ethnicity, age, sex, alcohol consumption, and smoking. Significant replication was defined at p < 0.05 and a consistent direction of effect.

Outcomes analyses

In significant sites identified through EWAS, DNAm at CpG sites was examined as a predictor of incident CHD in the WHI. CHD was defined by incident myocardial infarction or CHD death. Acute, hospitalized myocardial infarction was identified in medical records on the basis of cardiac pain, electrocardiogram, and biomarker data, and then physician-adjudicated. Further details regarding the review, classification, and adjudication of CHD in WHI have been described [47].

Multivariate Cox proportional hazard ratios were used to examine whether significant sites identified through EWAS (exposure) were associated with incident CHD in WHI. Individuals with a history of (or incident) myocardial infarction or coronary revascularization (angioplasty; stent; bypass) before measurement of DNAm were excluded. Covariates included age, race/ethnicity, smoking status, case–control status (BAA23 and AS311), DNAm array, row, and cell composition in the reduced model. In the full model, we adjusted for the covariates in the reduced model as well as physical activity and diet quality. Diet quality was measured using the Alternative Healthy Eating Index-2010 score [48] derived from food frequency questionnaires in the WHI. Significant sites were defined by p < 0.05.

Gene expression

To elucidate the potential functional implications of the identified CpG sites, we examined gene expression information using previously published significant gene expression quantitative trait methylation loci (eQTMs) summary statistics in blood from MESA and the GTP [49]. We examined the summary statistics for the four CpG sites identified in validation and in secondary CHD analyses. This population from MESA had minimal overlap with the MESA population examined in replication analyses.

Sensitivity analyses

As metabolic health status is constructed from several metabolic parameters, differences in methylation may be driven by individual metabolic parameters. To assess the degree that individual metabolic parameters influence methylation at significant sites, we reanalyzed associations between BMI × MHZ status and methylation excluding individual metabolic parameter in the MHZ score and compared the effects to the original estimates obtained through EWAS. For the significant sites, we also examined changes in effect size when adjusting for lipid, hypertension and glycemic medication use. We additionally repeated the primary EWAS analysis adjusting for physical activity. We additionally examined the four sites identified in replication and secondary analyses for normality and heteroskedasticity by examining the QQ plots and residual plots from the discovery analysis.

Availability of data and materials

The data for all studies are publicly available and more information can be found at the individual cohort websites: WHI (www.whi.org), ARIC (https://sites.cscc.unc.edu/aric/desc), and MESA (www.mesa-nhlbi.org).

Abbreviations

BMI:

Body mass index

DNAm:

DNA methylation

CpG:

Cytosine-phosphate-guanine

WHI:

Women’s Health Initiative

ARIC:

Atherosclerosis Risk in Communities

MHZ:

Metabolic health Z score

ATP:

Adult treatment panel

BMI × MHZ:

Interaction between body mass index and metabolic health Z score

FDR:

False discovery rate

CHD:

Coronary heart disease

MHO:

Metabolically healthy obesity

MUO:

Metabolically unhealthy obesity

RR:

Risk ratio

EMPC:

Epigenetic Mechanisms of Particulate Matter-Mediated Cardiovascular Disease

BAA23:

Integrative Genomics for Risk of Coronary Heart Disease and Related Phenotypes in WHI cohort

AS311:

Bladder Cancer and Leukocyte Methylation

CT:

Clinical trial

OS:

Observational study

AA:

African American

EA:

European American

MESA:

Multi-ethnic Study of Atherosclerosis

BP:

Blood pressure

TG:

Triglycerides

HDL:

High-density lipoprotein cholesterol

EWAS:

Epigenome-wide association study

eQTM:

Expression quantitative trait methylation loci

GTP:

Grady Trauma Project

HR:

Hazard ratio

TNF:

Tumor necrosis factor

TACI:

Transmembrane activator and CAML interactor

CRP:

C-reactive protein

IL-6:

Interleukin 6

References

  1. Hales CM, Fryar CD, Ogden CL. Prevalence of obesity among adults and youth: United States, 2015–2016. 2017.

  2. Winter JE, MacInnis RJ, Wattanapenpaiboon N, Nowson CA. BMI and all-cause mortality in older adults: a meta-analysis. Am J Clin Nutr. 2014;99:875–90.

    Article  CAS  PubMed  Google Scholar 

  3. Evans EM, Rowe DA, Racette SB, Ross KM, McAuley E. Is the current BMI obesity classification appropriate for black and white postmenopausal women? Int J Obes. 2006;30:837–43.

    Article  CAS  Google Scholar 

  4. Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA. 2013;309:71–82.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Xu H, Cupples LA, Stokes A, Liu C-T. Association of obesity with mortality over 24 years of weight history: findings from the Framingham Heart Study. JAMA Netw Open. 2018;1:e184587–e184587.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Mongraw-Chaffin M, Foster MC, Kalyani RR, Vaidya D, Burke GL, Woodward M, Anderson CAM. Obesity severity and duration are associated with incident metabolic syndrome: evidence against metabolically healthy obesity from the multi-ethnic study of atherosclerosis. J Clin Endocrinol Metab. 2016;101:4117–24.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Mongraw-Chaffin M, Foster MC, Anderson CAM, Burke GL, Haq N, Kalyani RR, Ouyang P, Sibley CT, Tracy R, Woodward M, Vaidya D. Metabolically healthy obesity, transition to metabolic syndrome, and cardiovascular risk. J Am Coll Cardiol. 2018;71:1857–65.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Stefan N, Häring HU, Schulze MB. Metabolically healthy obesity: the low-hanging fruit in obesity treatment? Lancet Diabetes Endocrinol. 2018;6:249–58.

    Article  PubMed  Google Scholar 

  9. Hamer M, Stamatakis E. Metabolically healthy obesity and risk of all-cause and cardiovascular disease mortality. J Clin Endocrinol Metab. 2012;97:2482–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Guo F, Garvey WT. Cardiometabolic disease risk in metabolically healthy and unhealthy obesity: Stability of metabolic health status in adults. Obesity (Silver Spring). 2016;24:516–25.

    Article  Google Scholar 

  11. Roberson LL, Aneni EC, Maziak W, Agatston A, Feldman T, Rouseff M, Tran T, Blaha MJ, Santos RD, Sposito A, Al-Mallah MH, Blankstein R, Budoff MJ, Nasir K. Beyond BMI: The “Metabolically healthy obese” phenotype & its association with clinical/subclinical cardiovascular disease and all-cause mortality – a systematic review. BMC Public Health. 2014;14:14.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Eckel N, Meidtner K, Kalle-Uhlmann T, Stefan N, Schulze MB. Metabolically healthy obesity and cardiovascular events: a systematic review and meta-analysis. Eur J Prev Cardiol. 2015;23:956–66.

    Article  PubMed  Google Scholar 

  13. Petronis A. Epigenetics as a unifying principle in the aetiology of complex traits and diseases. Nature. 2010;465:721–7.

    Article  CAS  PubMed  Google Scholar 

  14. Wang X, Zhu H, Snieder H, Su S, Munn D, Harshfield G, Maria BL, Dong Y, Treiber F, Gutin B, Shi H. Obesity related methylation changes in DNA of peripheral blood leukocytes. BMC Med. 2010;8:87.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  15. Dick KJ, Nelson CP, Tsaprouni L, Sandling JK, Aïssi D, Wahl S, Meduri E, Morange P-E, Gagnon F, Grallert H, Waldenberger M, Peters A, Erdmann J, Hengstenberg C, Cambien F, Goodall AH, Ouwehand WH, Schunkert H, Thompson JR, Spector TD, Gieger C, Trégouët D-A, Deloukas P, Samani NJ. DNA methylation and body-mass index: a genome-wide analysis. The Lancet. 2014;383:1990–8.

    Article  CAS  Google Scholar 

  16. Sayols-Baixeras S, Subirana I, Fernández-Sanlés A, Sentí M, Lluís-Ganella C, Marrugat J, Elosua R. DNA methylation and obesity traits—an epigenome-wide association study: The REGICOR study. Epigenetics. 2017;12:909–16.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Barres R, Zierath JR. DNA methylation in metabolic disorders. Am J Clin Nutr. 2011;93:897S-900S.

    Article  CAS  PubMed  Google Scholar 

  18. Samblas M, Milagro FI, Martínez A. DNA methylation markers in obesity, metabolic syndrome, and weight loss. Epigenetics. 2019;14:421–44.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Braun KVE, Dhana K, de Vries PS, Voortman T, van Meurs JBJ, Uitterlinden AG, Hofman A, Hu FB, Franco OH, Dehghan A, Consortium B. Epigenome-wide association study (EWAS) on lipids: the Rotterdam Study. Clin Epigenetics. 2017; 9:15.

  20. Dhana K, Braun KVE, Nano J, Voortman T, Demerath EW, Guan W, Fornage M, van Meurs JBJ, Uitterlinden AG, Hofman A, Franco OH, Dehghan A. An epigenome-wide association study of obesity-related traits. Am J Epidemiol. 2018;187:1662–9.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Walaszczyk E, Luijten M, Spijkerman AMW, Bonder MJ, Lutgers HL, Snieder H, Wolffenbuttel BHR, van Vliet-Ostaptchouk JV. DNA methylation markers associated with type 2 diabetes, fasting glucose and HbA1c levels: a systematic review and replication in a case-control sample of the Lifelines study. Diabetologia. 2018;61:354–68.

    Article  CAS  PubMed  Google Scholar 

  22. Wang F, Deeney JT, Denis GV. Brd2 gene disruption causes “metabolically healthy” obesity: epigenetic and chromatin-based mechanisms that uncouple obesity from type 2 diabetes. Vitam Horm. 2013;91:49–75.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Mendelson MM, Marioni RE, Joehanes R, Liu C, Hedman ÅK, Aslibekyan S, Demerath EW, Guan W, Zhi D, Yao C, Huan T, Willinger C, Chen B, Courchesne P, Multhaup M, Irvin MR, Cohain A, Schadt EE, Grove ML, Bressler J, North K, Sundström J, Gustafsson S, Shah S, McRae AF, Harris SE, Gibson J, Redmond P, Corley J, Murphy L, Starr JM, Kleinbrink E, Lipovich L, Visscher PM, Wray NR, Krauss RM, Fallin D, Feinberg A, Absher DM, Fornage M, Pankow JS, Lind L, Fox C, Ingelsson E, Arnett DK, Boerwinkle E, Liang L, Levy D, Deary IJ. Association of Body Mass Index with DNA methylation and gene expression in blood cells and relations to cardiometabolic disease: a mendelian randomization approach. PLoS Med. 2017;14:e1002215–e1002215.

    Article  PubMed  PubMed Central  Google Scholar 

  24. TRAPPC9 Gene.

  25. Liang ZS, Cimino I, Yalcin B, Raghupathy N, Vancollie VE, Ibarra-Soria X, Firth HV, Rimmington D, Farooqi IS, Lelliott CJ, Munger SC, O’Rahilly S, Ferguson-Smith AC, Coll AP, Logan DW. Trappc9 deficiency causes parent-of-origin dependent microcephaly and obesity. PLoS Genet. 2020;16:e1008916.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Huang RC, Garratt ES, Pan H, Wu Y, Davis EA, Barton SJ, Burdge GC, Godfrey KM, Holbrook JD, Lillycrop KA. Genome-wide methylation analysis identifies differentially methylated CpG loci associated with severe obesity in childhood. Epigenetics. 2015;10:995–1005.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Rzehak P, Covic M, Saffery R, Reischl E, Wahl S, Grote V, Weber M, Xhonneux A, Langhendries J-P, Ferre N, Closa-Monasterolo R, Escribano J, Verduci E, Riva E, Socha P, Gruszfeld D, Koletzko B. DNA-methylation and body composition in preschool children: epigenome-wide-analysis in the European childhood obesity project (CHOP)-Study. Sci Rep. 2017;7:14349.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  28. Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nat Rev Genet. 2012;13:484–92.

    Article  CAS  PubMed  Google Scholar 

  29. TNFRSF13B Gene.

  30. Liu L, Inouye KE, Allman WR, Coleman AS, Siddiqui S, Hotamisligil GS, Akkoyunlu M. TACI-deficient macrophages protect mice against metaflammation and obesity-induced dysregulation of glucose homeostasis. Diabetes. 2018;67:1589–603.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Ferrante AW Jr. Obesity-induced inflammation: a metabolic dialogue in the language of inflammation. J Intern Med. 2007;262:408–14.

    Article  CAS  PubMed  Google Scholar 

  32. Karelis AD, Faraj M, Bastard JP, Pierre DH, Brochu M, Prud’homme Rabasa-Lhoret DR. The metabolically healthy but obese individual presents a favorable inflammation profile. J Clin Endocrinol Metab. 2005;90:4145–50.

    Article  CAS  PubMed  Google Scholar 

  33. Phillips CM, Perry IJ. Does inflammation determine metabolic health status in obese and nonobese adults? J Clin Endocrinol Metab. 2013;98:E1610–9.

    Article  CAS  PubMed  Google Scholar 

  34. Houseman EA, Molitor J, Marsit CJ. Reference-free cell mixture adjustments in analysis of DNA methylation data. Bioinformatics (Oxford, England). 2014;30:1431–9.

    Article  CAS  Google Scholar 

  35. Doumatey AP, Lashley KS, Huang H, Zhou J, Chen G, Amoah A, Agyenim-Boateng K, Oli J, Fasanmade O, Adebamowo CA, Adeyemo AA, Rotimi CN. Relationships among obesity, inflammation, and insulin resistance in African Americans and West Africans. Obesity. 2010;18:598–603.

    Article  CAS  PubMed  Google Scholar 

  36. Hakim O, Bello O, Ladwa M, Peacock JL, Umpleby AM, Charles-Edwards G, Amiel SA, Goff LM. The link between obesity and inflammatory markers in the development of type 2 diabetes in men of Black African and White European Ethnicity. Nutrients. 2020;12:3796.

    Article  CAS  PubMed Central  Google Scholar 

  37. Power C, Thomas C. Changes in BMI, duration of overweight and obesity, and glucose metabolism: 45 years of follow-up of a birth cohort. Diabetes Care. 2011;34:1986–91.

    Article  PubMed  PubMed Central  Google Scholar 

  38. Design of the Women's Health Initiative clinical trial and observational study. The Women's Health Initiative Study Group. Controlled Clin Trials. 1998;19:61–109.

  39. Sharrett AR. The Atherosclerosis Risk in Communities (ARIC) Study: introduction and objectives of the hemostasis component. Ann Epidemiol. 1992;2:467–9.

    Article  CAS  PubMed  Google Scholar 

  40. Investigators TA. The atherosclerosis risk in communities (ARIC) study: design and objectives. Am J Epidemiol. 1989;129:687–702.

    Article  Google Scholar 

  41. Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Folsom AR, Greenland P, Jacob DR Jr, Kronmal R, Liu K, Nelson JC, O’Leary D, Saad MF, Shea S, Szklo M, Tracy RP. Multi-ethnic study of atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871–81.

    Article  PubMed  Google Scholar 

  42. Bild DE, Bluemke DA, Burke GL, Detrano R, DiezRoux AV, Folsom AR, Greenland P, Jacobs DR, Kronmal R, Liu K, Nelson JC, O’Leary D, Saad MF, Shea S, Szklo M, Tracy RP. Multi-ethnic study of atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871–81.

    Article  PubMed  Google Scholar 

  43. Richardson MT, Ainsworth BE, Wu H-C, Jacobs DR, Leon AS. Ability of the Atherosclerosis Risk in Communities (ARIC)/Baecke Questionnaire to assess leisure-time physical activity. Int J Epidemiol. 1995;24:685–93.

    Article  CAS  PubMed  Google Scholar 

  44. Meyer A-M, Evenson KR, Morimoto L, Siscovick D, White E. Test-retest reliability of the Women’s Health Initiative physical activity questionnaire. Med Sci Sports Exerc. 2009;41:530–8.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Gondalia R, Baldassari A, Holliday KM, Justice AE, Méndez-Giráldez R, Stewart JD, Liao D, Yanosky JD, Brennan KJM, Engel SM, Jordahl KM, Kennedy E, Ward-Caviness CK, Wolf K, Waldenberger M, Cyrys J, Peters A, Bhatti P, Horvath S, Assimes TL, Pankow JS, Demerath EW, Guan W, Fornage M, Bressler J, North KE, Conneely KN, Li Y, Hou L, Baccarelli AA, Whitsel EA. Methylome-wide association study provides evidence of particulate matter air pollution-associated DNA methylation. Environ Int. 2019;2019:104723.

    Article  CAS  Google Scholar 

  46. Teschendorff AE, Marabita F, Lechner M, Bartlett T, Tegner J, Gomez-Cabrero D, Beck S. A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA methylation data. Bioinformatics (Oxford, England). 2013;29:189–96.

    Article  CAS  Google Scholar 

  47. Curb JD, McTiernan A, Heckbert SR, Kooperberg C, Stanford J, Nevitt M, Johnson KC, Proulx-Burns L, Pastore L, Criqui M, Daugherty S. Outcomes ascertainment and adjudication methods in the Women’s Health Initiative. Ann Epidemiol. 2003;13:S122–8.

    Article  PubMed  Google Scholar 

  48. Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang M, Stampfer MJ, Willett WC. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142:1009–18.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Kennedy EM, Goehring GN, Nichols MH, Robins C, Mehta D, Klengel T, Eskin E, Smith AK, Conneely KN. An integrated-omics analysis of the epigenetic landscape of gene expression in human blood cells. BMC Genomics. 2018;19:476.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

Download references

Acknowledgements

The authors would like to acknowledge and thank all participants from the WHI, ARIC, and MESA.

Funding

The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C. The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute (NHLBI), National Institutes of Health, Department of Health and Human Services (contract numbers HHSN268201700001I, HHSN268201700002I, HHSN268201700003I, HHSN268201700004I, and HHSN268201700005I). The authors thank the staff and participants of the ARIC study for their important contributions. Funding was also supported by 5RC2HL102419 and R01NS087541. MESA TOPMed Methylation data for the Trans-Omics in Precision Medicine (TOPMed) program was supported by the NHLBI. MESA TOPMed Methylation data for “Genome-wide methylation for NHLBI TOPMed: Multi-Ethnic Study of Atherosclerosis (MESA)” (phs001416.v1.p1) were performed at Keck Molecular Genomics Core Facility (HHSN268201600034I). Centralized read mapping and genotype calling, along with variant quality metrics and filtering were provided by the TOPMed Informatics Research Center (3R01HL-117626-02S1, contract HHSN268201800002I). Phenotype harmonization, data management, sample-identity QC, and general study coordination were provided by the TOPMed Data Coordinating Center (R01HL-120393; U01HL-120393; contract HHSN268201800001I). We gratefully acknowledge the studies and participants who provided biological samples and data for TOPMed. WLD was supported in part by the Nalini and Ravi Saligram Scholarship.

Author information

Authors and Affiliations

  1. Nutrition and Health Sciences Program, Laney Graduate School, Emory University, 1518 Clifton Rd, Atlanta, GA, 30322, USA

    Whitney L. Do

  2. Herbert Wertheim School of Public Health and Human Longevity Science, University of California, San Diego, CA, USA

    Steve Nguyen & Aladdin Shadyab

  3. Department of Pediatrics, The Institute for Translational Genomics and Population Sciences, The Lundquist Institute for Biomedical Innovation at Harbor-UCLA Medical Center, Torrance, CA, USA

    Jie Yao, Xiuqing Guo & Jerome I. Rotter

  4. Departments of Epidemiology and Medicine, University of North Carolina, Chapel Hill, NC, USA

    Eric A. Whitsel

  5. Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN, USA

    Ellen Demerath

  6. Center for Public Health Genomics, University of Virginia, Charlottesville, VA, USA

    Stephen S. Rich

  7. Division of Biomedical Informatics & Personalized Medicine, School of Medicine, Colorado University Anschutz Medical Campus, Aurora, CO, USA

    Leslie Lange

  8. Gerontology and Geriatric Medicine, School of Medicine, Wake Forest, Winston-Salem, NC, USA

    Jingzhong Ding

  9. Department of Population and Public Health Sciences, University of Southern California, Los Angeles, CA, USA

    David Van Den Berg

  10. Duke Molecular Physiology Institute, Duke University, Durham, NC, USA

    Yongmei Liu

  11. Center for Biomedical and Translational Informatics, Geisinger, Wilkes-Barre, PA, USA

    Anne E. Justice

  12. Division of Biostatistics, School of Public Health, University of Minnesota, Minneapolis, MN, USA

    Weihua Guan

  13. Department of Human Genetics, University of California, Los Angeles, Los Angeles, CA, USA

    Steve Horvath

  14. Department of Medicine, Stanford University, Stanford, CA, USA

    Themistocles L. Assimes

  15. Cancer Control Research, BC Cancer, Vancouver, BC, Canada

    Parveen Bhatti

  16. Department of Epidemiology, University of Washington, Seattle, WA, USA

    Kristina Jordahl

  17. College of Medicine, University of Arizona Cancer Center, Tucson, AZ, USA

    Celina I. Valencia

  18. Hubert Department of Global Health, Rollins School of Public Health, Emory University, Atlanta, GA, USA

    Aryeh D. Stein, Lisa R. Staimez & K. M. Venkat Narayan

  19. Department of Gynecology and Obstetrics, School of Medicine, Emory University, Atlanta, GA, USA

    Alicia Smith

  20. Department of Human Genetics, Emory University, Atlanta, GA, USA

    Karen Conneely

Authors
  1. Whitney L. Do
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Steve Nguyen
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jie Yao
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Xiuqing Guo
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Eric A. Whitsel
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Ellen Demerath
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Jerome I. Rotter
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Stephen S. Rich
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Leslie Lange
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Jingzhong Ding
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. David Van Den Berg
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Yongmei Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Anne E. Justice
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Weihua Guan
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Steve Horvath
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Themistocles L. Assimes
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Parveen Bhatti
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Kristina Jordahl
    View author publications

    You can also search for this author in PubMed Google Scholar

  19. Aladdin Shadyab
    View author publications

    You can also search for this author in PubMed Google Scholar

  20. Celina I. Valencia
    View author publications

    You can also search for this author in PubMed Google Scholar

  21. Aryeh D. Stein
    View author publications

    You can also search for this author in PubMed Google Scholar

  22. Alicia Smith
    View author publications

    You can also search for this author in PubMed Google Scholar

  23. Lisa R. Staimez
    View author publications

    You can also search for this author in PubMed Google Scholar

  24. Karen Conneely
    View author publications

    You can also search for this author in PubMed Google Scholar

  25. K. M. Venkat Narayan
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

WLD, KMVN, and KC conceived of the study. The methodology was developed by WLD, KMVN, ADS, AS, and LRS. The data for WHI were curated by EAS, YL, SH, TLA, PB, and KJ. The data for ARIC were curated by ED, WG, and AEJ. The data for MESA were curated by XG, JIR, SSR, LL, JD, and DVDB. All analyses were completed by WLD, SN, and JY. The manuscript was written by WLD. Editorial and content expertise was provided by all authors. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Whitney L. Do.

Ethics declarations

Ethics approval and consent to participate

Ethics approval was obtained for all the studies, and all participants provided consent to participate in individual studies. More information can be found for the cohorts at the following websites: WHI (www.whi.org), ARIC (https://sites.cscc.unc.edu/aric/desc), and MESA (www.mesa-nhlbi.org).

Consent for publication

Not applicable.

Competing interests

The authors declare they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1.

Supplemental Figures 1-16. Supplemental Tables 1-6.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Do, W.L., Nguyen, S., Yao, J. et al. Associations between DNA methylation and BMI vary by metabolic health status: a potential link to disparate cardiovascular outcomes. Clin Epigenet 13, 230 (2021). https://doi.org/10.1186/s13148-021-01194-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13148-021-01194-3

Keywords

Clinical Epigenetics

ISSN: 1868-7083

Contact us