Diabetes Status Modifies the Association Between Different Measures of Obesity and Heart Failure Risk Among Older Adults: A Pooled Analysis of Community-Based NHLBI Cohorts

Supplemental Digital Content is available in the text. Background: Obesity and diabetes are associated with a higher risk of heart failure (HF). The interrelationships between different measures of adiposity—overall obesity, central obesity, fat mass (FM)—and diabetes status for HF risk are not well-established. Methods: Participant-level data from the ARIC study (Atherosclerosis Risk in Communities; visit 5) and the CHS (Cardiovascular Health Study; visit 1) cohorts were obtained from the National Heart, Lung, and Blood Institute Biologic Specimen and Data Repository Information Coordinating Center, harmonized, and pooled for the present analysis, excluding individuals with prevalent HF. FM was estimated in all participants using established anthropometric prediction equations additionally validated using the bioelectrical impedance-based FM in the ARIC subgroup. Incident HF events on follow-up were captured across both cohorts using similar adjudication methods. Multivariable-adjusted Fine-Gray models were created to evaluate the associations of body mass index (BMI), waist circumference (WC), and FM with risk of HF in the overall cohort as well as among those with versus without diabetes at baseline. The population attributable risk of overall obesity (BMI≥30 kg/m2), abdominal obesity (WC>88 and 102 cm in women and men, respectively), and high FM (above sex-specific median) for incident HF was evaluated among participants with and without diabetes. Results: The study included 10 387 participants (52.9% ARIC; 25.1% diabetes; median age, 74 years). The correlation between predicted and bioelectrical impedance-based FM was high (R2=0.90; n=5038). During a 5-year follow-up, 447 participants developed HF (4.3%). Higher levels of each adiposity measure were significantly associated with higher HF risk (hazard ratio [95% CI] per 1 SD higher BMI=1.15 [1.05, 1.27], WC=1.22 [1.10, 1.36]; FM=1.13 [1.02, 1.25]). A significant interaction was noted between diabetes status and measures of BMI (P interaction=0.04) and WC (P interaction=0.004) for the risk of HF. In stratified analysis, higher measures of each adiposity parameter were significantly associated with higher HF risk in individuals with diabetes (hazard ratio [95% CI] per 1 SD higher BMI=1.29 [1.14–1.47]; WC=1.48 [1.29–1.70]; FM=1.25 [1.09–1.43]) but not those without diabetes, including participants with prediabetes and euglycemia. The population attributable risk percentage of overall obesity, abdominal obesity, and high FM for incident HF was higher among participants with diabetes (12.8%, 29.9%, and 13.7%, respectively) versus those without diabetes (≤1% for each). Conclusions: Higher BMI, WC, and FM are strongly associated with greater risk of HF among older adults, particularly among those with prevalent diabetes.

O besity affects more than one-third of adults in the United States and is a significant risk factor for diabetes and heart failure (HF). 1 The obesity epidemic is associated with substantial morbidity, mortality, and health care expenditures totaling nearly $150 billion annually. Several factors contribute to the obesity-associated risk of cardiovascular disease (CVD), particularly HF, including the higher downstream burden of CVD risk factors and direct effects of higher body mass index (BMI) on cardiac structure and function. [2][3][4] However, BMI-the standard metric used for assessing obesity-does not adequately capture the heterogeneity of adiposity and regional adiposity measures. 5 Alternative measures of regional adiposity, like visceral adiposity and waist circumference (WC), can identify specific adipose tissue locations and are associated with HF risk. 6,7 Similarly, adiposity measures such as fat mass (FM) have distinct associations with subclinical cardiac phenotypes. 8 Body composition varies considerably across metabolic states with a high prevalence of abdominal obesity in diabetes. 9 However, it is unclear if the association of different measures of adiposity with HF risk is modified by the presence of metabolic abnormalities, particularly diabetes.
Accordingly, we evaluated the independent contributions of different adiposity measures-overall obesity (BMI), central obesity (WC), and FM-toward HF risk in a cohort of older adults and assessed whether these relationships were modified by diabetes status. On the basis of previous study findings examining the relationships among anthropometric measures of adiposity with risk of HF across different study populations, we hypothesized that these anthropometric measures would be significantly associated with a greater risk of HF, particularly among those with diabetes. [10][11][12]

METHODS
The present analysis used deidentified data from the ARIC study (Atherosclerosis Risk in Communities) and the CHS (Cardiovascular Health Study) obtained with approval from the National Heart, Lung, and Blood Institute Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC). Data analyzed in the present study will not be made available for reproducing the study results and can be obtained from BioLINCC by submitting a research proposal.

Study Design and Participants
Participants from ARIC and CHS were included in the present study. Briefly, ARIC is an epidemiological cohort study that enrolled 15 792 adults who were 45 to 64 years of age from 1987 to 1989 in 4 communities across the United States (Forsyth County, NC; Jackson, MS; Minneapolis, MN; Washington County, MD). 13,14 Between 2011 and 2013, participants were asked to return for ARIC visit 5 in which they completed questionnaires and underwent testing with blood sampling, assessment of body composition, and echocardiography. 8 For the present analysis, ARIC visit 5 is considered the baseline visit. CHS is a cohort study of older adults (≥65 years of age) from 4 communities: Forsyth County, NC; Pittsburgh, PA; Sacramento County, CA; and Washington County, MD. 15,16 Between 1989 and 1990, 5201 participants were enrolled in examination 1 of CHS and completed standardized questionnaires, underwent anthropometric measurements, and had blood drawn. ARIC and CHS participants with available data from BioLINCC were considered for inclusion (ARIC: n=5952; CHS: n=5201). Participants were excluded from the present study if they had BMI <18.5 kg/m 2 , history of HF, missing baseline race/ethnicity, HF, or outcome data (n=766). Each study site institutional review board approved the study protocol. All study participants provided written informed consent.

Primary Exposure Variables of Interest
The primary exposure variables of the present study were BMI, WC, and estimated FM. Trained personnel measured

Clinical Perspective
What Is New?
• Among older adults, diabetes status modifies the association of body mass index and waist circumference for the risk of heart failure. • Higher measures of overall obesity (body mass index), central adiposity (waist circumference), and fat mass are more strongly associated with risk of heart failure among older adults with diabetes than those without diabetes.
What Are the Clinical Implications?
• Comorbid diabetes and obesity (overall and central) may identify individuals at high risk for developing heart failure. • Patients with coexisting diabetes and obesity may benefit from effective heart failure prevention therapies, including sodium-glucose cotransporter 2 inhibitors and intentional weight loss. height, weight, and WC using standardized protocols. 8,16 Standing height was measured using a stadiometer. Weight was assessed while participants were shoeless on a scale. BMI was calculated by dividing weight in kilograms by height in meters. 2 In ARIC, WC was measured to the nearest centimeter along a horizontal plane at the uppermost lateral border of the ilium using a tape measure after a regular expiration. In CHS, WC was measured at the level of the umbilicus. FM assessed by the bioelectrical impedance method was available only among participants from ARIC, but not CHS. 8 In this pooled analysis, FM was estimated using established anthropometric equations that incorporate demographics (sex, race/ ethnicity, age) and commonly assessed anthropometric measures (height, weight, WC) as detailed in the Supplemental Methods. 17 Sex-specific FM prediction equations were previously validated using dual-energy x-ray absorptiometry data from the general population (R 2 =0.90-0.93) plus a cohort of participants with diabetes (R 2 =0.87). 11,17 Previous studies have demonstrated that these estimated measures of FM are associated with CVD events. 11,18 These prediction equations were validated with the bioelectrical impedance-based FM measures among ARIC participants in the present study. Bioelectrical impedance-based FM data from ARIC were included in sensitivity analyses.

Clinical Covariates, Echocardiographic Parameters, and Cardiac Biomarkers
Demographic and clinical characteristics were assessed during ARIC visit 5 and CHS examination 1 using standard protocols described previously. 8,[13][14][15][16] Age, sex, race, and medical history were self-reported. Systolic and diastolic blood pressure were measured according to standardized protocols. Hypertension was defined on the basis of self-reported antihypertensive medication use, systolic blood pressure ≥140 mm Hg, or diastolic blood pressure ≥90 mg Hg. Triglycerides and total and high-density lipoprotein cholesterol were measured using standard assays. In the pooled cohort, standardized categories were used to harmonize individual-level data of participants from ARIC and CHS. Diabetes was defined in both cohorts on the basis of the presence of any of the following: self-reported physician diagnosis, antihyperglycemic medication use, or fasting plasma glucose ≥126 mg/dL. 19,20 Participants without diabetes who had a fasting plasma glucose ≥100 mg/dL were classified as having prediabetes. Euglycemia was defined as an absence of prediabetes and diabetes. Because of the lack of hemoglobin A1c data in CHS, only fasting plasma glucose was considered in the definitions for diabetes, prediabetes, and euglycemia, consistent with a previous pooled analysis. 21 Among participants from the ARIC cohort (visit 5) and CHS cohort, echocardiographic images were obtained using standardized image acquisition protocols as previously described. 22,23 Left ventricular mass was calculated using the Devereaux formula and indexed to body surface area according to standardized protocols described further in the Supplemental Methods. 24 NT-proBNP (N-terminal pro-B-type natriuretic peptide) levels, a measure of subclinical cardiac stress, were measured from stored blood samples obtained from participants during ARIC visit 5 and CHS examination 1 as detailed in the Supplemental Methods. 25,26 hs-cTnT (high-sensitivity cardiac troponin T) data, a measure of subclinical myocardial injury, were available in participants from the ARIC cohort but not among CHS participants in the National Heart, Lung, and Blood Institute BioLINCC data repository.

Clinical Outcomes of Interest
The primary outcome of interest was incident HF. The adjudication processes and definitions of incident HF were similar in ARIC and CHS. In ARIC, incident HF was defined as the first hospitalization for HF on the basis of International Classification of Diseases-Ninth Revision (428) and International Classification of Diseases-Ninth Revision codes (I50) on the discharge list. 25,27,28 A committee adjudicated HF events on the basis of the review of medical records including relevant HF signs, symptoms, diagnostics, and medications. In CHS, during semiannual visits, cases of HF were initially identified on the basis of self-report of a physician diagnosis. 16,29 The CHS Events Committee reviewed medical records for signs, symptoms, and imaging findings. The presence of HF was determined on the basis of HF diagnosis plus treatment. ARIC and CHS outcome data were available until December 31, 2017, and December 31, 2011, respectively. Outcome data were censored at 5 years of follow-up.

Statistical Analysis
Standardized adiposity measures, diabetes strata, and incident HF events were used to harmonize individual-level data and facilitate pooling ARIC and CHS into a multicohort analysis. Baseline characteristics were reported as median (interquartile interval, 25th percentile-75th percentile) for continuous and number (percentage) for categorical variables. All relevant covariates had <10% missingness. Missing values were imputed using random forest imputation. 30 The adjusted associations among continuous measures of adiposity with risk of overall HF were evaluated using Fine-Gray models accounting for competing risk of death. Separate models were created for each exposure variable of interest (BMI, WC, FM) with sequential adjustment for the following potential confounders: Model 1 = age, sex, race, alcohol use, smoking history, study cohort; Model 2 = Model 1 covariates plus history of coronary heart disease, history of hypertension, systolic blood pressure, history of diabetes, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride level, and statin use. Multiplicative interaction terms were included in the most adjusted models to evaluate if diabetes status modified the associations between the adiposity measures and risk of overall HF (diabetes status X each exposure variable [BMI, WC, FM], in respective models).
Stratified analyses were performed according to diabetes status separately among those with versus without diabetes at baseline. Among participants with and without diabetes, baseline characteristics of those with high versus low categories for BMI (obesity: BMI ≥30 kg/m 2 versus no obesity), WC (abdominal obesity: >88 cm for women and >102 cm for men versus no abdominal obesity), and FM (high versus low defined as above versus below the sex-specific median, respectively) were compared using the Kruskal-Wallis test for continuous variables and the chi-square test for categorical variables. Cumulative incidence of HF was assessed among participants stratified by diabetes status (no diabetes versus diabetes) plus adiposity measures (high versus low categories). Adjusted Patel et al Obesity, Diabetes, and HF Risk Fine-Gray models (including covariates described previously in Models 1 and 2) and restricted cubic splines (including Model 2 covariates) were also constructed to evaluate the association of continuous measures of adiposity (BMI, WC, FM) with the risk of overall HF among individuals with and without diabetes. Sensitivity analyses were performed examining FM assessed by the bioelectrical impedance method in the ARIC cohort. Across diabetes categories (no diabetes versus diabetes), the population attributable risk percentage of overall obesity (BMI ≥30 kg/m 2 ), abdominal obesity (WC >88 cm for women, >102 cm for men), and high FM (above sex-specific median) for incident HF was evaluated. Population attributable risk percentage was calculated as (1-[(1-S 0 (t))/(1-S(t))] such that if the adiposity measure were eliminated from the study population, then S 0 (t) is the counterfactual survival function and S(t) is the factual survival function.
Subgroup analyses were performed among participants without diabetes stratified into euglycemic and prediabetes categories.
Across diabetes strata, multivariable-adjusted linear regression models were created to evaluate the associations of continuous adiposity measures with left ventricular mass index (pooled cohort), subclinical myocardial stress (NT-proBNP, in pooled cohort), and injury (hs-cTnT, in ARIC subgroup only) using the most adjusted models as described above. Because of skewed distributions, hs-cTnT and NT-proBNP were log-transformed.
Two-sided P value <0.05 was considered statistically significant. Analyses were performed using R version 3.6.3 (R Foundation for Statistical Computing).

RESULTS
Among the 10 387 participants included in the pooled cohort (52.9% ARIC; median age 74 years), 57.5% were women, and 13.1% were of self-reported Black race. Baseline characteristics of participants included from each of the 2 pooled cohorts are shown in Table S1. The prevalence of diabetes and prediabetes was 25.1% and 36.0%, respectively. In the overall cohort, 25.8% and 55.9% of participants had obesity and abdominal obesity, respectively. Among 5038 ARIC participants, predicted and noninvasive assessment of FM were highly correlated (R 2 =0.90) with minimal bias on the basis of Bland-Altman analysis (1.82 kg; 95% limits of agree-ment=1.71-1.95 kg; Figure S1).

Adiposity and Risk of HF
Across diabetes categories, participants with obesity were younger and had a greater burden of hypertension and cardiometabolic abnormalities, including higher triglycerides and lower high-density lipoprotein cholesterol, compared with those who did not have obesity (Table 1). A similar pattern of baseline characteristics was observed among participants across diabetes subgroups and further stratified by abdominal obesity and estimated FM categories (Tables S2 and S3).

Adiposity and Risk of HF Among Individuals With Versus Without Diabetes
The risk of HF was higher in participants with diabetes (event rate per 1000 person-years=14.8) versus those without diabetes (event rate per 1000 person-years=7.5). The cumulative incidence of HF was highest in participants with diabetes and the higher measure of adiposity (obesity versus no obesity, abdominal obesity versus no abdominal obesity, high versus low estimated FM; Figure 1). For example, among participants with diabetes, 7.5% with obesity and 6.1% without obesity developed HF during a 5-year follow-up. In the no diabetes subgroup, the 5-year cumulative incidence of HF was 3.5% and 3.4% for those without and with obesity, respectively.
In adjusted analysis, among participants with diabetes, higher BMI was significantly associated with a higher risk of HF after adjusting for demographics and traditional CVD risk factors (HR per 1 SD higher BMI, 1.29 [95% CI, 1.14-1.47]; Model 2; Table 2, Figure 2 Figure 2). In contrast, among participants without diabetes, BMI, WC, and estimated FM were not significantly associated with risk of HF after adjustment for potential confounders (Table 2, Figure 2). In sensitivity analysis restricted to ARIC participants with bioelectrical impedance-based measures of FM, results were similar compared with the primary analysis such that higher FM was significantly associated with higher risk of HF among participants with diabetes (HR per 1 SD higher FM, 1 Figure S2). The population attributable risk percentage of each adiposity measure (BMI, WC, and estimated FM) for incident HF was higher among participants with versus without diabetes. In the diabetes subgroup, the population attributable risk percentage of overall obesity (BMI ≥30 kg/m 2 ), abdominal obesity (WC >88 cm in women and WC >102 cm in men), and high estimated FM (above sex-specific median) for incident HF was 12.8%, 29.9%, and 13.7%, respectively. In the nondiabetes subgroup, the population attributable risk percentage of each adiposity measure was less than or equal to 1% (generalized obesity: 1.0%; abdominal obesity: -0.9%; high estimated FM: 0.5%).

Adiposity and Risk of HF Among Individuals With Prediabetes and Euglycemia
In subgroup analysis stratifying participants without diabetes into prediabetes and euglycemia categories, the risk of developing HF was similar in the prediabetes (event rate per 1000 person-years=7.8) and euglycemia subgroups (event rate per 1000 person-years=7.2). Across both euglycemia and prediabetes strata, there was no significant association between any adiposity measures (BMI, WC, and estimated FM) and risk of HF ( Table 3).

Association of Adiposity Measures With Cardiac Parameters
In the pooled cohort, in adjusted analysis accounting for potential confounders, higher BMI, WC, and FM were significantly associated with greater left ventricular mass index. The strength of the associations was qualitatively greater among participants with versus without diabetes (Table 4). Higher levels of each adiposity measure were significantly associated with lower NT-proBNP among participants without diabetes. In the diabetes subgroup, higher FM, but not BMI or WC, was significantly associated with lower NT-proBNP. In the ARIC subgroup, higher BMI and WC were significantly associated with higher hs-cTnT after accounting for demographics and CVD risk Continuous variables are presented as median (interquartile intervals: 25th percentile-75th percentile) and compared using the Kruskal-Wallis test. Categorical variables are presented as number (percentage) and compared using the chi-squared test. BMI indicates body mass index; BP, blood pressure; CHD, coronary heart disease; GFR, glomerular filtration rate; hs-cTnT, high-sensitivity cardiac troponin T; LVMi, left ventricular mass index; N/A, not applicable; and NT-proBNP, N-terminal pro-B-type natriuretic peptide.
*Data for hs-cTnT were available only among participants from the Atherosclerosis Risk in Communities study.
Patel et al Obesity, Diabetes, and HF Risk factors irrespective of diabetes status. In unadjusted and adjusted analysis, FM was not significantly associated with hs-cTnT (Table S3, Table 4).

DISCUSSION
In the present study, we observed several important findings. Among older adults, higher BMI, WC, and FM measures were each associated with higher risk of HF. However, the association of these measures of adiposity with the risk of HF was modified by diabetes status. Thus, higher BMI, WC, and FM were significantly associated with greater risk of HF in individuals with diabetes but not among those without diabetes. Furthermore, the population attributable risk of generalized obesity, abdominal obesity, and high FM for HF risk was higher among in-dividuals with versus without diabetes, suggesting that these measures of adiposity may be more relevant for downstream risk of HF in the presence of diabetes. These findings highlight the contribution of diabetes to the adiposity-associated risk of HF. Diabetes is a well-established risk factor for HF. Previous studies from multiple cohorts have identified approximately double the risk of HF in those with versus without diabetes. 21,31 The relative importance of obesity for HF risk in participants with diabetes has also been evaluated in previous studies. Rawshani et al demonstrated that elevated BMI out of the normal range was 1 of the strongest predictors of HF in individuals with diabetes. 32 Similarly, higher BMI and intentional changes in body weight have been identified as important predictors of HF among patients with dia-A C B Figure 1. Cumulative incidence of heart failure (HF) among participants stratified by diabetes status and adiposity measure. A, Participants are stratified by diabetes status and obesity status (obesity vs no obesity). B, Participants are stratified by diabetes status and abdominal obesity status (abdominal obesity vs no abdominal obesity). C, Participants are stratified by diabetes status and fat mass categories (above sex-specific median fat mass vs below sex-specific median fat mass).

Patel et al
Obesity, Diabetes, and HF Risk betes in previous studies. 33,34 In the present study, we extend these observations by demonstrating the prognostic importance of different measures of obesity for the risk of HF in individuals with diabetes. Specifically, overall obesity, central obesity, and high FM contributed nearly 13%, 30%, and 14% risk of incident HF. Separate multivariable-adjusted Fine-Gray models were constructed for each adiposity measure and heart failure with sequential adjustment for the following covariates: Model 1 = age, sex, race, alcohol use, smoking history, and study cohort; Model 2 = Model 1 covariates plus history of coronary heart disease, history of hypertension, systolic blood pressure, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride level, and statin use. HR indicates hazard ratio. The adjusted model included the following covariates: age, sex, race, alcohol use, smoking history, study cohort, history of coronary heart disease, history of hypertension, systolic blood pressure, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride level, and statin use.

Patel et al
Obesity, Diabetes, and HF Risk The mechanisms through which obesity may predispose to incident HF include greater antecedent development of HF risk factors and direct effects of obesity on cardiac structure and function. [2][3][4] Among individuals without diabetes, obesity predisposes individuals to metabolic dysregulation, insulin resistance, and a higher risk of diabetes. 35 In the present study, we demonstrated that diabetes modified the risk of HF associated with higher levels of different measures of obesity-BMI, WC, and FM. Obesity was associated with a higher risk of HF only in the presence of diabetes. In contrast, the risk of HF among individuals with obesity without coexisting diabetes was comparable with healthier weight individuals. These findings highlight the important contribution of metabolic dysregulation, particularly diabetes, to obesityrelated HF risk.
Previous studies have demonstrated that metabolically healthy individuals with overweight and obesity have a lower or comparable risk of HF (versus metabolically healthy normal-weight individuals), with a higher risk noted only among metabolically unhealthy individuals irrespective of the BMI range. 12,36,37 We confirmed and extended these observations across different measures of obesity-overall obesity, central obesity, and high FM-among community-dwelling older adults. We further extend the previous observations by focusing on a clinically relevant subgroup of metabolically unhealthy individuals with prevalent diabetes. These findings have important clinical implications considering the recent emergence of novel pharmacotherapies such as sodium glucose cotransporter-2 inhibitors as effective strategies for lowering HF risk in high-risk individuals with diabetes. 38 Recent stud- Separate multivariable-adjusted Fine-Gray models were constructed for each adiposity measure and heart failure with sequential adjustment for the following covariates: Model 1 = age, sex, race, alcohol use, smoking history, and study cohort; Model 2 = Model 1 covariates plus history of coronary heart disease, history of hypertension, systolic blood pressure, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride level, and statin use. HR indicates hazard ratio. β estimate represents the change in the outcome per 1 SD higher exposure variable while keeping other covariates fixed. Separate models were created for each continuous adiposity measure and each specified outcome with adjustment for the following covariates: age, sex, race, alcohol use, smoking history, study cohort, history of coronary heart disease, history of hypertension, systolic blood pressure, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride level, and statin use. hs-cTnT indicates high-sensitivity cardiac troponin T; LVMi: left ventricle mass index; and NT-proBNP, N-terminal pro-B-type natriuretic peptide.
*Data for hs-cTnT were available only among participants from the Atherosclerosis Risk in Communities study.
Patel et al Obesity, Diabetes, and HF Risk ies have demonstrated greater absolute risk reduction with the use of sodium glucose cotransporter-2 inhibitors among patients with diabetes and obesity. 39 Future efforts should be directed at aggressive implementation of such therapeutic strategies in individuals with diabetes and obesity at the highest risk of developing HF. The effect modification of prevalent diabetes on the risk of HF associated with obesity may also have implications for the therapeutic efficacy of weight loss strategies. Along these lines, effective weight loss interventions, such as bariatric surgery, have been shown to be associated with greater reduction in HF risk among patients with obesity and diabetes. 40 Similarly, weight loss therapies such as liraglutide have demonstrated trends toward a greater CVD risk reduction among individuals with cardiometabolic risk factor clustering. 41 Mechanistic studies suggest that weight loss interventions may lead to reverse cardiac remodeling with benefits to both left and right ventricular function. 42,43 Future randomized controlled trials are needed to determine if targeting effective weight loss approaches toward individuals with obesity and diabetes may effectively lower the risk of HF.
Several mechanisms may underlie the observed effect modification of the association between adiposity parameters and risk of HF by diabetes. First, myocardial energy use may be impaired among individuals with diabetes with lower glucose and higher fatty acid use. 44 This alteration in myocardial energy substrate use may make the myocardial tissue more vulnerable to damage by stressors such as ischemia, pressure overload, low fitness, and high adiposity. Second, there may also be greater sympathetic nervous system activation and myocardial triglyceride content among individuals with obesity and diabetes, which are important determinants of the downstream risk of HF. 45,46 Third, the mechanisms through which adiposity and diabetes predispose patients to increased HF risk may also be related to adverse cardiac remodeling and development of subclinical diabetic cardiomyopathy phenotype. 3,47 As observed in the present study, higher measures of adiposity were associated with greater left ventricular mass. Taken together, the clustering of metabolic risk factors may provide a fertile ground for the manifestations of adverse effects of excess adiposity on cardiac structure and function and contribute to the greater downstream risk of HF.
In contrast with observations from the present study, previous cohort studies in younger individuals with longer-term follow-up have demonstrated an increased risk of HF associated with higher BMI among individuals with and without diabetes at baseline. [48][49][50] These differences may be related to differences in the study population and follow-up duration for HF risk assessment. It is plausible that older individuals -such as those included in the present study-with overall or abdominal obesity without diabetes represent an early stage of cardiometabolic dysfunction and thus may have a lower risk of HF during a shorter (5-year) follow-up. Obesity without coexisting diabetes in younger or middle-aged individuals may predispose to HF on long-term follow-up through interval development of diabetes and cardiometabolic dysregulation. Along these lines, investigators in the Multi-Ethnic Study of Atherosclerosis have demonstrated that progression from metabolically healthy to unhealthy obesity was associated with an increased risk of HF. 37 Also, weight loss strategies have been shown to favorably modify downstream burden of CVD and its risk factors irrespective of diabetes status. [51][52][53] Future studies with long-term follow-up are needed to determine if interval development of cardiometabolic dysregulation/diabetes may mediate the risk of HF in older age among young/ middle-aged metabolically healthy individuals.
The present study has several notable limitations. First, the present study included an older study population, and there is potential for selection bias. Measures of adiposity change with aging, including loss of lean mass and increased FM, and diabetes and HF risk increases over time. Thus, the present study findings may not be generalizable to younger individuals. Also, participants from both ARIC and CHS were included in the present study with the latter cohort examined more than 2 decades earlier. Approaches to risk factor modification have evolved over time with slight differences in baseline characteristics and HF event rates across both cohorts. In addition, adult participants from ARIC were initially enrolled and subsequently followed for >2 decades, which makes the present study susceptible to survival bias. Participants who were alive in later years might not attend follow-up encounters, suggesting a possible visit attendance bias. Second, estimated measures of FM were included in pooled analyses rather than direct measures. Advanced measures of adiposity assessed by bioelectrical impedance were not available in the entire study cohort. However, anthropometric prediction equations used in the present study were validated and demonstrated high correlation and minimal bias compared with bioelectrical impedance assessment of FM. In addition, measurement errors were likely nondifferential given the prospective study design and would have biased the study results toward the null. Third, data on regional adiposity depots such as visceral and subcutaneous adipose tissue were not available, limiting our ability to characterize the relationship between adiposity phenotypes, diabetes, and risk of HF. Last, the study design was observational, and we cannot exclude the potential for residual measured or unmeasured confounding.
In conclusion, among older adults, higher BMI, WC, and FM are associated with a higher risk of HF, particularly among those with prevalent diabetes. Future studies are needed to understand the mechanisms underlying the interrelationships of adiposity measures and diabetes with the risk of HF in older and younger individuals and evaluate whether targeting this cluster of cardiometabolic factors is an effective strategy for HF prevention.