Introduction

The prevalence of obesity has grown over the past several decades with 68% of US adults currently considered overweight or obese.¹ Although obesity has been associated with increased all-cause mortality, after adjustment for traditional cardiovascular risk factors, body mass index (BMI) and waist circumference (WC), it remains only minimally associated with coronary heart disease and stroke.^1,2,3 There are conflicting data as to whether greater BMI is associated with an increased risk of heart failure.^4,5

See Clinical Perspective

Although many of the mechanisms remain unexplained, a growing body of evidence demonstrates there are detrimental effects of obesity on cardiac structure and function.^6,7,8,9 The majority of studies examining the effects of obesity on the heart used anthropomorphic measures, such as BMI, which do not distinguish between fat and lean tissue and may misclassify individuals with excess fat mass as nonobese, especially in older adults in whom sarcopenia is common. Bioimpedence offers a means of directly measuring body composition and enables body mass to be classified as fat or lean and may help to elucidate the true relationship between excess body fat (BF) and abnormalities of cardiac structure and function. In addition, WC more directly reflects abdominal obesity, which is associated with increased metabolic abnormalities and cardiovascular risk.

We analyzed variations in body composition (percent BF) and traditional anthropomorphic measures (BMI and WC) and their association with cardiac structure and function in an elderly, biracial cohort that was free from clinically overt coronary heart disease or heart failure.

Methods

Study Population

The Atherosclerosis Risk in Communities study (ARIC) is a prospective cohort designed to investigate the causes of atherosclerosis and its related clinical outcomes in 15 792 men and women recruited from 4 US communities (Forsyth County, NC; Jackson, MS; suburbs of Minneapolis, MN; and Washington County, MD¹⁰). Participants were aged 45 to 64 years at the time of the baseline examination (1987–1989) and were subsequently evaluated during 4 follow-up clinical visits (visits 2, 3, 4, and 5) at ≈3-year intervals, except for visit 5 (15-year interval). Annual telephone follow-up of participants has been conducted since the baseline visit. Institutional review boards from each site approved the study, and informed consent was obtained from all participants.

The present investigation is a cross-sectional analysis of 4343 ARIC participants during visit 5 (2011–2013). Of the 6538 participants at visit 5, we first restricted the sample to 5741 who had complete data of echocardiography, BMI, WC, and percent BF. We then excluded participants with prevalent coronary heart disease (n=853) or heart failure (n=480), race other than black or white (n=12), as well as those who were underweight at visit 5 (n=53), yielding a final n of 4343.

Ascertainment and Categorization of Body Composition

Standardized anthropomorphic measurements of weight, height, and WC were obtained, and bioelectric impedance (measured using the Tanita Body Composition Analyzer, TBF-300A) was utilized to calculate percent BF, fat mass, and lean body mass. Obesity was defined using 3 different criteria: (1) BMI (defined as body weight [in kilograms] divided by height [in meters] squared) was categorized as normal (18.5 ≥ BMI <25 kg/m²), overweight (25 ≥ BMI <30 kg/m²), or obese (BMI ≥30 kg/m²). (2) WC (cm) with abdominal obesity categorized defined as WC >102 in men or >88 in women. (3) Percent BF was defined as obese, if >25% in men or 35% in women.^11,12,13

Echocardiographic Methods and Measurements

Echocardiograms were obtained from participants at all 4 sites at visit 5 on identically configured systems using a standardized protocol.¹⁴ Images were digitally transferred to the Cardiovascular Imaging Core Laboratory at Brigham and Women’s Hospital, Boston, MA, for offline analysis. The intraobserver variability (coefficient of variation and intraclass correlation) for key echocardiographic measures has been previously published.¹⁴

Images were obtained in the parasternal long- and short-axis and apical 2- and 4-chamber views. Primary measures of left ventricular (LV) dimensions, volumes, and wall thickness; right ventricular (RV) area; left atrial (LA) dimension, volume, and area; and Doppler measures of mitral inflow, tricuspid regurgitation, and mitral annular relaxation velocities were made in triplicate from the 2-dimensional views in accordance with the recommendations of the American Society of Echocardiography.¹⁵

Calculation of Derived Echocardiographic Variables

LV ejection fraction was calculated as 100×(LV end-diastolic volume- LV end-systolic volume)/LV end-diastolic volume. LV mass was determined according to the ASE recommendations and indexed to height to the power of 2.7. Total LV wall thickness was the sum of anteroseptal and posterior wall thickness, and relative wall thickness (RWT) was calculated as (2×posterior wall thickness)/LV end-diastolic dimension. LV hypertrophy (LVH) was defined as LV mass indexed to >95 g/m² in women and >115 g/m² in men. Normal geometry was classified as RWT ≤0.42 and no LVH; Abnormal geometry was defined as the presence of either concentric remodeling (RWT >0.42 and no LVH), concentric hypertrophy (RWT <0.42 and LVH), or eccentric hypertrophy (RWT ≥0.42 and LVH). RV fractional area change (%) was calculated as 100×(RV end-diastolic area-RV end-systolic area)/RV end-diastolic area. LA volume was indexed to height to the power of 2.7. Abnormalities of diastolic function were assessed using a previously published approach in a community-based cohort.¹⁶ Our population was classified using the following schema: normal diastolic function (deceleration time of E wave >140ms, 0.75<E/A<2 and E/E′<10); mild diastolic dysfunction (declaration time >140 ms, E/A<0.75 and E/E′<10); moderate-to-severe diastolic dysfunction (0.75<E/A<2 and E/E′≥10; or DT <140ms, E/A>2 and E/E′≥10); and unclassifiable if participants did not fall into 1 of these 3 categories. Diastolic function was collapsed into normal diastolic function and abnormal diastolic function (mild or moderate-to-severe diastolic dysfunction) for all analyses.

Statistical Analysis

Data are presented as mean±SD, median (IQR), or n (%). Owing to gender-specific norms for many of the outcome variables and significant interactions between sex and the outcomes, all analyses were performed stratified by sex. Interactions with race were tested for in all outcomes. Intergroup comparisons were compared using nonparametric trend, t test, and χ² tests. Skewed variables were analyzed after log-transformation. Unadjusted and adjusted associations were assessed using linear and logistic regression. Restricted cubic spline models were used to assess all relationships for linearity. χ² tests were used to compare categorical variables. Potential confounders were identified a priori and used to correct all models. Covariates included in the regression models were age, race, heart rate, systolic blood pressure, antihypertensive use, diabetes mellitus, HbA1c, and current smoking. Bonferroni-corrected P values were used to determine statistical significant for all analyses. A corrected P <0.005 was considered statistically significant for the primary echocardiographic measures reported. All reported P values are 2 sided. STATA v12.1 (College Station, TX) was used.

Results

Patient Characteristics

The characteristics of our study sample (n=4343) are shown in Table 1 and Table I in the Data Supplement. Compared with participants with normal BMI, those who were overweight and obese were younger and were more likely to be black. The prevalences of diabetes mellitus and hypertension were high and increased along with BMI and other measures of obesity. Participants with higher BMIs also had significantly larger WCs and fat mass (P for trend <0.001). There was an inverse relationship observed between BMI and current smoking.

LV Dimensions

In both women and men, greater anthropomorphic measures of obesity, BMI, and WC were associated with a significantly increased thickness of the LV walls, larger LV cavity size, and higher LV mass (Table 2). A similar association was seen when percent BF was used as a measure of adiposity in women. There was a significant interaction with race in the relationship between LV dimensions and BF in men that was not observed with BMI or WC; a positive correlation between BF and LV dimensions was not seen in black men, although in white men the findings were consistent with those in women (Figure 1).

• Download figure
• Download PowerPoint

Obese women, but not men, were significantly more likely to exhibit abnormal LV geometry when compared with participants with a BMI <30 kg/m² (P<0.001 women, P=0.49 men; Figure 2).

• Download figure
• Download PowerPoint

LV Systolic Function

The presence of obesity was not associated with LV systolic function as assessed by ejection fraction in either women or men. In women, but not in men, circumferential strain, a more sensitive measure of impaired systolic function, was highly associated with all 3 measures of obesity. Global longitudinal strain was also significantly associated with WC and BF in women, but not in men.

Diastolic Function and the Left Atrium

In women, but not in men, obesity was associated with significantly worse overall diastolic function (Figure 3). In both sexes, left atrial volumes increased significantly in the presence of increasing measures of obesity, even when scaled by height (Table 2).

• Download figure
• Download PowerPoint

RV Structure and Function

The size of the RV as assessed by the RV end diastolic area indexed to height^2.7 was associated positively with BMI, and RV systolic function was negatively associated with BMI in both men and women (P≤0.001). The relationship between RV size and function and measures of obesity remained significant across all measures of body composition in women (P<0.001). In men, the association between BF and RV size and function was significant in white but not black men.

Discussion

In a community-based, biracial population of older adults free of cardiovascular disease, greater metrics of body size and adiposity were associated with subclinical abnormalities in LV structure and function which differed somewhat by sex and race. In women, obesity was associated with a higher LV mass coupled with a disproportionately smaller increase in LV diameter, which resulted in concentric and eccentric hypertrophy with a concurrent decrement in strain and abnormalities of diastolic function. Men also exhibited greater total wall thickness and a positive association between LV diameter and mass in relation to greater BMI and WC although these findings were not significantly accompanied by body size–related impairments of LV function or a particular pattern of remodeling; the relationship with BF was modified by race and among men was found to be significant only in whites.

Similar to the changes that occur in hypertension or as a result of aortic stenosis, obesity is associated with cardiac remodeling. Previous studies using echocardiography have also demonstrated healthy obese men and women are more likely to have increased LV wall thickness and LV mass than their nonobese counterparts which may represent the earliest moment on the path to developing heart failure.^8,9,17 There is emerging evidence from animal models that this change in wall thickness may be mediated by leptin, an adipocyte hormone that promotes collagen synthesis.¹⁸ Despite these changes in LV structure, greater BMI has not previously been associated with a significant decrement in systolic function as assessed by ejection fraction,^8,19 or abnormalities of diastolic function after adjustment for clinical covariates.²⁰

Our results demonstrate that whether assessed via traditional anthropomorphic measures such as BMI and WC, or newer measures of body composition such as BF, obesity is associated with cardiac geometry and function differently in elderly men compared with elderly women. In particular, women are more likely to display primarily LV hypertrophy coupled with diastolic dysfunction and subclinical decrements in systolic function. Less negative circumferential strain has been shown to add incremental prognostic value for the development of incident heart failure²¹ and may partially account for the female predominance of heart failure with preserved ejection fraction. By contrast, more obese men were not observed to exhibit similar patterns of remodeling or subclinical decrements in LV systolic function. This preservation of biventricular systolic function could represent relative protection against the development of clinical heart failure with obesity and a type of obesity paradox in men.

Both sexes and races demonstrated similar decrements in RV function and a positive association between RV size and greater body size. There is a paucity of data regarding the echocardiographic relationships between RV size and obesity, although subclinical dysfunction has been described in association with increasing BMI.⁷ To our knowledge, these data are the first from a large, biracial contemporary cohort to examine RV morphology and function and its relation to obesity. We found that there is a significant decrement in RV systolic function coupled with RV enlargement in overweight and obese individuals that was not explained by comorbid hypertension or diabetes mellitus. These changes are probably multifactorial and may result from increased intravascular volume or sleep-disordered breathing and warrant further study to elucidate the etiologic mechanisms. In addition, as the prevalence of obesity continues to rise, the potential implications of RV dysfunction on long-term survival will be increasingly important.

Several limitations of our study warrant consideration. First, this is an ambulatory elderly cohort and the findings may not be applicable to all obese individuals. Our study is cross-sectional, and therefore a hypothesis generating as to the causal relationship between obesity and heart failure with preserved ejection fraction in women. There may also be a healthy survivor bias observed among the small black male cohort that resulted in their findings differing from the other participants. In addition, there were a small number of participants who were excluded from the analysis because of missing data. These individuals were slightly older, more likely to be black, and slightly more hypertensive than those included. Multiple comparisons were made between 3 indices of obesity and echocardiographic measures. To minimize the possibility of an increased risk of type I error, we used the highly conservative Bonferroni adjustment to account for multiple testing which may have consequentially resulted in limited power. However, this allows us to focus on the most clinically significant differences. Although echocardiography has been well validated for the assessment of cardiac structure and function, magnetic resonance imaging is considered the gold standard for the assessment of RV. However, our results are consistent with those from a prior study that examined the effect of obesity on RV using cardiac magnetic resonance imaging.²²

In conclusion, we observed in a large community-based cohort of elderly individuals that subclinical abnormalities of cardiac structure function are associated with measures of obesity. Both sexes demonstrated similar increases in LV mass and biventricular size as well as decrements in RV function with obesity, although women manifested subclinical LV systolic dysfunction whereas men did not. These sex-specific abnormalities may contribute to different patterns of obesity-related cardiovascular disease.

Acknowledgments

We thank the staff and participants of the ARIC study for their important contributions.

Footnotes