Sex-Dependent Effects of Body Fat on Cardiac Function and Outcomes in Community-Dwelling Older Adults

Abstract

Background

Among older adults, sex hormones may exert differential effects on body fat and cardiac health.

Objectives

The authors aimed to study the effects of sex on body fat composition and aging-related diastolic function, as well as their longitudinal changes.

Methods

Community-dwelling older adults without cardiovascular disease underwent prospective baseline same-day anthropometry for body fat mass (BFM), percentage body fat (PBF), and waist hip ratio (WHR) quantification by bioimpedance and transthoracic Doppler echocardiography, with measurements repeated at 5 years. Study outcomes included diastolic grade, ratio of early to late diastolic inflow velocities (E/A), ratio of mitral early diastolic inflow velocity to the average of septal and lateral early diastolic annular tissue velocities (E/e') ratios, left ventricle mass index, left atrial volume index and clinical outcomes. Correlations between baseline or interval changes (Δ) of WHR, BFM and PBF with follow-up or Δ values of outcomes were analyzed, with multivariable linear regression model adjustments for significant variables.

Results

Among 409 older adults (mean age 61 ± 12.7 years, 214 women) with median follow-up of 4.9 (4.5, 5.2) years, baseline BFM predicted echocardiographic diastolic dysfunction at follow-up in women (OR: 1.059; 95% CI: 1.051-1.115; P = 0.002). Baseline PBF was associated with ΔE/A (β: 0.010; P = 0.028) among women and ΔE/e' (β: −0.069; P = 0.020) among men, with significant interactions between sex and PBF in predicting ΔE/A (P < 0.05). ΔBFM (β: −0.007; P = 0.05) and ΔPBF (β: −0.005; P = 0.047) were associated with ΔE/A independent of baseline BFM/PBF among women. Baseline (HR: 1.141; 95% CI: 1.027-1.158; P = 0.005) and changes in WHR (HR: 0.857; 95% CI: 0.790-0.931; P < 0.001) predicted cardiovascular outcomes among women.

Conclusions

Baseline and interval increase in fat quantity via bioimpedance defined differential trajectories of longitudinal cardiac functional deteriorations between sexes.

Central Illustration

Obesity is a major public health concern and is associated with a range of complications, including cardiovascular complications.^1,2 Methods of accurately measuring and tracking changes in obesity are crucial to estimating and reducing obesity-related morbidity and mortality. Body mass index (BMI), although still widely used as a screening tool for obesity, has been shown to be inaccurate in quantification of body fat and controversial in its prognosticating value.³ This is especially so in older adults whose BMI readings may be spuriously reduced due to age-related sarcopenia.⁴ Other biometrics of obesity, such as waist circumference (WC) and waist hip ratio (WHR), which are markers of central obesity, are demonstrated to be superior to BMI in prognostication.^5,6 In particular, the quantitation of body fat via bioimpedance analysis (BIA), a noninvasive and low-cost method of objectively measuring body fat, has shown promise as a potential risk assessment tool.⁷ Although body fat has been shown to be correlated with cardiac dysfunction on cross-sectional analysis,⁶ longitudinal assessment of body fat in association with cardiac function is lacking. In addition, body fat composition diverges with increasing age in men vs women due to hormonal changes and menopausal status in women.⁸ The significance of these variations in body fat mass (BFM) with age and sex on cardiac function have yet to be fully elucidated. Furthermore, data in a multiethnic Asian population are scarce. We thus aim to evaluate the cardiac effect of body fat on aging adults in an Asian population, hypothesizing that sex would exert differential effects on progression of aging-related cardiac diastolic dysfunction.

Methods

Study population

The subjects were recruited from the CAS (Cardiac Aging Study),⁹ a prospective study initiated in 2014 that examines characteristics and determinants of cardiovascular function in elderly adults. CAS participants were recruited from the prospective, population-based cohort, the Singapore Chinese Health Study¹⁰ as well as directly from the local community. The current study sample consisted of men and women who participated in the baseline CAS 2014 to 2017 examination who underwent prospective same-day transthoracic echocardiography as well as body fat quantitation by BIA, at baseline and 5 years. All participants had no self-reported history of physician-diagnosed cardiovascular disease (such as coronary heart disease or atrial fibrillation), stroke, or cancer. Written informed consent was obtained from each participant on enrollment. The SingHealth Centralised Institutional Review Board (CIRB/2014/628/C) had approved the study protocol.

Data acquisition

All participants were examined and interviewed on at least 2 study visits by trained study coordinators. Participants completed a standardized questionnaire that included medical history and coronary risk factors. Hypertension was defined by the current use of antihypertensive drugs or physician-diagnosed hypertension. Diabetes mellitus was defined by the current use of antidiabetic agents or physician-diagnosed diabetes mellitus. Dyslipidemia was defined by the current use of lipid-lowering agents or physician-diagnosed dyslipidemia. Smoking history was defined as ever smokers (former or current smokers) or never smokers. Cardiovascular risk category was classified into 0/1/2/3 based on the number of risk factors of hypertension, dyslipidemia, and smoking history. Clinical outcomes were defined as a composite of all-cause mortality or hospitalization during follow-up, which were collected via review of medical records and verified by telephone at follow-up.

Echocardiography was performed using ALOKA α10 with a 3.5- MHz probe. In each subject, standard echocardiography, which included 2-D, M-mode, pulse Doppler, and tissue Doppler imaging, was performed in the parasternal and apical (apical 4-chamber, apical 2-chamber, and apical long) views, and 3 cardiac cycles were recorded. The transmitral flow E and A waves with the sample volume position at the tip of the mitral valve leaflets from the apical 4-chamber view were recorded by Doppler echocardiography. Pulsed wave tissue Doppler imaging was performed with the sample volume at the septal and lateral annuli from the apical 4-chamber view. The frame rate was between 80 and 100 frames per second. The tissue velocity patterns were recorded and expressed as e', and a'. All measurements were measured by the same operator and the measurements were averaged over 3 cardiac cycles. The measurements of interest in this study were markers of cardiac diastolic function, namely, the ratio of early diastolic mitral inflow velocity to the average of the septal and lateral early diastolic mitral annular tissue velocities (E/e′) and¹¹ the ratio of early to late diastolic inflow mitral velocities (E/A).¹¹ E/A ratio had previously been shown to be prognostic of clinical outcomes in community-dwelling older adults.¹² Other potential markers of left ventricle (LV) diastolic function, including left ventricle mass index (LVMI) and left atrial volume index (LAVI) were also explored. A diastolic grade (0, I, II, or III) was assigned to each patient according to the latest 2025 guidelines from the American Society of Echocardiography guidelines.¹³ The echocardiography readers were blinded to the obesity status of the participants.

Fat quantification was assessed using a multifrequency tetrapolar bioelectrical impedance method (Inbody 370). BFM was measured as the mass of body fat in kilograms, and percentage body fat (PBF) was calculated as BFM divided by total body mass in kilograms. WC was measured 2.5 cm above the umbilicus. Hip circumference was obtained at the widest part of the hips. WHR was defined as WC divided by hip circumference.

Statistical analysis

Clinical characteristics are presented as means and SDs for continuous data and frequency and percentage for categorical data, unless otherwise stated. The Student's t-test was used for continuous data and the χ² test was used for categorical data. Multivariate logistic regression was used to determine the association between baseline and interval changes (Δ) of PBF, BFM, and WHR measurements with the presence of diastolic dysfunction function at follow-up (defined as either diastolic grading I/II/II), presented as OR with corresponding 95% CI. Multivariable linear regression analysis was performed to ascertain the relationship of follow-up or Δ of E/A, E/e′, LVMI, and LAVI, to baseline or Δ of WHR, BFM, and PBF measurements (defined as follow-up values minus baseline values), presented as regression coefficients with 95% CIs. HRs for clinical outcomes and their 95% CI were calculated via Cox proportional hazards models. All models were adjusted for significant covariates on univariate analysis, which can be found in Supplemental Tables 1 and 2. Models examining follow-up in fat quantity markers with cardiac function were also adjusted for baseline values. Sex-stratified analysis was performed to determine any interactions between sex and obesity markers in predicting follow-up and ΔE/A, E/e′. Outlier analysis was performed by removing subjects with a value >3 SD away from the mean. All statistical analyses were performed using STATA 15. For all analyses, a 2-tailed P value of <0.05 was considered statistically significant.

Results

Population characteristics

In total, 409 adults with 214 (52%) women, 406 (99.3%) Chinese, 1 (0.2%) Malay, and 2 (0.5%) other ethnicities and a mean age of 61 ± 12.7 years at baseline were included in the present study. The overall cohort demonstrated deteriorations in ΔE/A of −0.1 ± 0.3, ΔE/e' of 1.8 ± 2.8, ΔLAVI of 19.9 ± 6.7, and ΔLVMI of −5.1 ± 31.3 over a median interval of 4.9 (4.5, 5.2) years. Women demonstrated ΔE/A of −0.15 ± 0.4 and ΔE/e' of 2.0 ± 3.0, whereas men demonstrated ΔE/A of −0.09 ± 0.3 and ΔE/e′ of 1.6 ± 2.5 over the follow-up duration. Baseline characteristics, coronary risk factors, and echocardiographic parameters are summarized in Table 1.

Table 1.

Baseline Clinical and Echocardiographic Parameters of Study Cohort

	All (N = 409)	Women (n = 212)	Men (n = 197)	P Value
Clinical
Age, y	61 ± 12.7	60 ± 13.3	63 ± 11.7	0.0026
Ethnicity, Chinese/Malay/Others	406/1/2	212/0/0	194/1/2	0.11
Body mass index, kg/m2	24 ± 3.3	23 ± 3.3	24 ± 3.2	0.0002
Systolic blood pressure, mm Hg	135 ± 18.5	130 ± 18.9	139 ± 17.0	<0.0001
Diastolic blood pressure, mm Hg	75 ± 11.5	71 ± 11.4	79 ± 10.2	<0.0001
Pulse, beats/min	72 ± 11.3	71 ± 10.5	74 ± 12.0	0.015
Waist circumference, cm	82 ± 10.4	78 ± 10.1	86 ± 9.2	<0.0001
Hypertension, n (%)	126 (30.8%)	50 (23.6%)	76 (38.6%)	0.001
Dyslipidaemia, n (%)	147 (35.9%)	69 (32.6%)	78 (39.6%)	0.15
Diabetes mellitus, n (%)	43 (10.5%)	17 (8.0%)	26 (13.2%)	0.11
Smoking, n (%)	61 (14.9%)	7 (3.3%)	54 (27.4%)	<0.0001
Cardiovascular risk category 0/1/2/3	197/109/84/19	126/48/36/2	71/61/48/17
Obesity markers
Body fat mass, kg	19.4 ± 6.3	20.2 ± 6.4	18.5 ± 6.0	0.0062
Body fat mass, %	30.8 ± 7.4	34.5 ± 6.5	26.8 ± 6.1	<0.0001
Waist hip ratio	1.0 ± 1.7	1.0 ± 2.3	0.9 ± 0.1	0.0036
Echocardiographic parameters
E/A	1.2 ± 0.5	1.3 ± 0.5	1.1 ± 0.4	0.0012
E/A at follow-up	1.0 ± 0.4	1.1 ± 0.4	1.0 ± 0.4	0.0015
E/A change	−0.12 ± 0.3	−0.15 ± 0.4	−0.09 ± 0.3	0.2
E/e′	8.7 ± 2.4	8.9 ± 2.3	8.5 ± 2.4	0.027
E/e' at follow-up	10.5 ± 3.1	10.9 ± 3.1	10.1 ± 3.0	0.0021
E/e' change	1.8 ± 2.8	2.0 ± 3.0	1.6 ± 2.5	0.26
Left ventricular mass index at baseline	74.4 ± 30.9	69.6 ± 18.1	78.1 ± 37.6	0.018
Left ventricular mass index at follow-up	65.6 ± 15.8	61.0 ± 12.7	70.5 ± 17.4	<0.0001
Left ventricular mass index change	−5.1 ± 31.3	−5.0 ± 20.1	−5.2 ± 37.8	0.46
Pulmonary artery systolic pressure (mm Hg) at baseline	24.5 ± 6.1	24.2 ± 6.1	24.7 ± 6.1	0.32
Pulmonary artery systolic pressure (mm Hg) at follow-up	23.6 ± 4.9	23.1 ± 4.5	24.1 ± 5.4	0.069
Pulmonary artery systolic pressure (mm Hg) change	−1.3 ± 5.7	−1.5 ± 5.8	−1.2 ± 5.6	0.99
Left atrial volume index (mL/m2) at baseline	19.9 ± 6.7	19.8 ± 6.3	20.0 ± 7.1	0.95
Left atrial volume index (mL/m2) at follow-up	26.3 ± 6.8	26.1 ± 6.9	26.4 ± 6.8	0.61
Left atrial volume index (mL/m2) change	6.4 ± 7.8	6.3 ± 7.6	6.5 ± 7.9	0.5
Left ventricular ejection fraction at baseline (%)	73.3 ± 7.8	74.1 ± 7.7	72.6 ± 7.9	0.039
Diastolic grade 0/I/II/III	153/82/112/19	81/36/56/15	72/46/56/4
Time interval between 2 measurements, y	4.9 ± 0.4	5.0 ± 0.4	4.8 ± 0.5	0.0005

Overview and comparison of baseline clinical and echocardiographic characteristics between men and women. Bold values indicate P < 0.05.

E/A = ratio of early to late diastolic inflow velocities.

Effect of obesity markers on cardiac diastolic function

The overall cohort had a mean baseline BFM of 19.4 ± 6.3 kg, PBF of 30.8% ± 7.4%, and WHR of 1.0 ± 1.7.

Baseline BFM was associated with the presence of adverse diastolic function at follow-up in the overall cohort (OR: 1.043; 95% CI: 1.004-1.083; P = 0.031), a result primarily driven by women (OR: 1.059; 95% CI: 1.051-1.115; P = 0.002) and not men (OR: 1.007; 95% CI: 0.954-1.062; P = 0.806). Baseline PBF and WHR was associated with adverse diastolic function at follow-up in women on univariate analysis but this was attenuated on multivariate analysis (PBF: OR: 1.039; 95% CI: 0.987-1.095; P = 0.146; WHR: OR: 17.59; 95% CI: 0.048-6,770.76; P = 0.345) (Table 2).

Table 2.

Effect of Body Fat on Cardiac Diastolic Function

	Diastolic Dysfunction at Follow-Up
	Simple Logistic Regression			Multiple Logistic Regression
	OR	95% CI	P Value	OR	95% CI	P Value
All
Baseline BFM	1.039	1.003-1.077	0.03	1.043a	1.004-1.083	0.031
Baseline PBF	1.015	0.987-1.045	0.299
Baseline WHR	32.68	0.921-1,159	0.055
Women
Baseline BFM	1.082	1.029-1.138	0.002	1.059b	1.005-1.115	0.031
Baseline PBF	1.068	1.018-1.122	0.008	1.025	0.987-1.095	0.146
Baseline WHR	315.1	1.244-79,827	0.042	17.59	0.046-6,771	0.345
Men
Baseline BFM	1.007	0.955-1.061	0.806
Baseline PBF	1.033	0.981-1.087	0.224
Baseline WHR	3.11	0.032-300	0.626

Multiple logistic regression models of the effect of body fat metrics on diastolic grading on echocardiography. Bold values indicate P < 0.05.

BFM = body fat mass; PBF = percentage body fat; WHR = waist hip ratio.

^aAdjusted for age, pulse rate, sex, diabetes status, cardiovascular risk category.

^bAdjusted for age, systolic and diastolic blood pressures, cardiovascular risk category.

In the overall cohort, baseline BFM, PBF, and WHR did not predict follow-up or ΔE/A or E/e′ after adjustments for covariates, including baseline E/A and E/e′, respectively (Table 3A). Sex-specific analysis suggests that among women, higher baseline BFM and PBF were associated with deteriorations in ΔE/A (BFM: β 0.009; 95% CI: 0.001-0.018; P = 0.042; PBF: β: 0.010; 95% CI: 0.001-0.018; P = 0.028) independent of covariates but not ΔE/e' (BFM: β: 0.007; 95% CI: −0.058 to 0.723; P = 0.822; PBF: β: 0.014; 95% CI: −0.051 to 0.078; P = 0.430) (Table 3B). In contrast, among men, higher baseline BFM and PBF were associated with deteriorations in ΔE/e′ (BFM: β: −0.073; 95% CI: −0.131 to −0.014; P = 0.016; PBF: β: −0.069; 95% CI: −0.126 to 0.011; P = 0.020) independent of covariates but not ΔE/A (BFM: β: −0.001; 95% CI: −0.008 to 0.005; P = 0.685; PBF: β: −0.002; 95% CI: −0.009 to 0.005; P = 0.568) (Table 3C). Baseline WHR did not predict interval changes in or follow-up E/A or E/e′ in both sexes (P > 0.05 for all analyses). Baseline PBF or BFM did not have any significant association between follow-up or Δ values of LAVI or LVMI (P > 0.05 for all analyses) (Supplemental Table 3).

Table 3.

Effect of Baseline Body Fat on E/A and E/e'

	Simple Linear Regression								Multiple Linear Regression
	E/A Change		E/e' Change		E/A at Follow-Up		E/e' at Follow-Up		E/A Change		E/e' Change		E/A at Follow-Up		E/e' at Follow-Up
	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value
All
Baseline BFM	0.005	0.092	−0.028	0.213	−0.009	0.009	0.023	0.352					−0.003a	0.187
Baseline PBF	0.002	0.321	−0.002	0.916	−0.008	0.009	0.061	0.003					−0.002a	0.342	−0.018b	0.397
Baseline WHR	−0.005	0.619	0.144	0.084	−0.01	0.424	0.148	0.106
Women
Baseline BFM	0.011	0.01	0.007	0.822	−0.016	0.001	0.035	0.3	0.009c	0.042			−0.003d	0.357
Baseline PBF	0.127	0.003	0.014	0.43	−0.021	<0.001	0.042	0.202	0.010c	0.028			−0.001d	0.621
Baseline WHR	−0.004	0.71	0.143	0.11	−0.012	0.363	0.14	0.13
Men
Baseline BFM	−0.001	0.685	−0.079	0.009	−0.004	0.422	−0.01	0.783			−0.073e	0.016
Baseline PBF	−0.002	0.568	−0.6	0.044	−0.012	0.011	0.051	0.145			−0.069e	0.02	−0.004f	0.227
Baseline WHR	−0.4	0.893	−4.88	0.065	0.149	0.728	−2.75	0.381

Simple and multiple linear regression models of the effect of baseline body fat metrics on cardiac function in women and men. Variables significant on univariate analysis were adjusted for other significant clinical variables. Bold values indicate P < 0.05.

Abbreviations as in Table 2.

^aAdjusted for age, sex, pulse rate, diabetes status, cardiovascular risk category, baseline E/A ratio.

^bAdjusted for age, sex, diabetes status, cardiovascular risk category, baseline E/e' ratio.

^cAdjusted for age, pulse rate, systolic blood pressure.

^dAdjusted for age, diabetes status, cardiovascular risk category, baseline E/A or E/e' ratio.

^eAdjusted for age.

^fAdjusted for age, pulse rate, diabetes status, cardiovascular risk category, baseline E/A ratio.

There were significant interactions between sex and baseline PBF in predicting both ΔE/A (P = 0.007) (Figure 1) and follow-up E/A (P = 0.020) (Figure 2). However, there was no significant interaction between baseline PBF and sex in predicting either ΔE/e' (P = 0.098) or follow-up E/e' (P = 0.86) (Supplemental Figures 1 and 2).

Figure 1.

Interaction Between Baseline Body Fat Percentage and Sex in Predicting Interval Changes in E/A Ratio

Graph plotting E/A change against baseline body fat percentage, stratified by sex. E/A change and baseline body fat percentage is positively correlated in women (r = 0.21; P = 0.003) but not significantly correlated in men (P = 0.57). There is significant interaction between sex and baseline body fat percentage (P = 0.007). E/A = ratio of early to late diastolic inflow velocities.

Figure 2.

Interaction Between Baseline Body Fat Percentage and Sex in Predicting E/A Ratio at Follow-Up

Graph plotting E/A ratio at follow-up against baseline body fat percentage, stratified by sex. E/A at follow-up is negatively correlated with baseline body fat percentage in both women (r = −0.29; P < 0.001) and men (r = −0.18; P = −0.011). There is significant interaction between sex and baseline body fat percentage (P = 0.02). Abbreviations as in Figure 1.

Effect of interval changes obesity markers on cardiac function

The overall cohort had mean ΔBFM of 1.3 ± 4.7 kg, ΔPBF of 2.6 ± 7.1%, and ΔWHR of −0.12 ± 1.7.

Interval increases in PBF over time, but not BFM, was significantly associated with reductions in ΔE/A in the overall cohort (PBF: β: −0.005; 95% CI: −0.0100 to −0.0001; P = 0.047; BFM: β: −0.005; 95% CI: −0.013 to 0.002; P = 0.137), particularly among women (PBF: β: −0.007; 95% CI: −0.014 to −0.001; P = 0.040; BFM: β: −0.011; 95% CI: −0.022 to −0.001; P = 0.043) but not men (PBF: β: −0.002; 95% CI: −0.009 to 0.004; P = 0.448; BFM: β: −0.003; 95% CI: −0.012 to 0.006; P = 0.469), independent of other covariates (Table 4, Figure 3). The association between ΔPBF with ΔE/A in women was attenuated after removal of outliers (β: −0.004; 95% CI: −0.011 to 0.001; P = 0.118 (Supplemental Figure 3). There was neither significant association between ΔPBF nor ΔBFM with ΔE/e' or follow-up E/e' in men or women (P > 0.05 for all analyses). Interval increases in WHR was significantly associated with greater reductions in ΔE/A ratio in men (β: −0.519; 95% CI: −0.937 to −0.102; P = 0.015) although there were no significant covariates to adjust for (Table 4).

Table 4.

Effect of Changes in Body Fat on E/A and E/e' Ratio

	Simple Linear Regression								Multiple Linear Regression
	E/A Change		E/e' Change		E/A at Follow-Up		E/e' at Follow-Up		E/A Change		E/e' Change		E/A at Follow-Up		E/e' at Follow-Up
	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value	β	P Value
All
Change in BFM	−0.008	0.04	−0.008	0.797	0.002	0.601	−0.035	0.318	−0.006a	0.137
Change in PBF	−0.006	0.023	0.01	0.623	−0.001	0.781	0.005	0.816	−0.005a	0.047
Change in WHR	0.005	0.644	−0.158	0.057	0.009	0.475	−0.16	0.08
Women
Change in BFM	−0.12	0.04	−0.037	0.442	0.01	0.126	−0.048	0.328	−0.011b	0.043
Change in PBF	−0.008	0.036	0.001	0.975	0.003	0.416	−0.004	0.878	−0.007b	0.04
Change in WHR	0.005	0.687	−0.149	0.098	0.011	0.377	−0.142	0.125
Men
Change in BFM	−0.003	0.469	0.027	0.525	−0.005	0.412	−0.015	0.757
Change in PBF	−0.002	0.448	0.02	0.479	−0.007	0.093	0.014	0.676
Change in WHR	−0.519c	0.015	−2.6	0.171	−0.355	0.247	−4.12	0.066

Simple and multiple linear regression models of the effect of changes in body fat metrics on cardiac function in women and men. Variables significant on univariate analysis were adjusted for other significant clinical variables. Bold values indicate P < 0.05.

Abbreviations as in Table 2.

^aAdjusted for baseline BFM/PBF, pulse rate, systolic blood pressure.

^bAdjusted for baseline BFM/PBF, age, pulse rate, systolic blood pressure.

^cNo significant covariates.

Figure 3.

Interaction Between Interval Changes in Body Fat Percentage and Sex

Graph plotting change in E/A ratio against change in body fat percentage, stratified by sex. E/A change is negatively correlated with change in body fat percentage in women (r = −0.14; P = 0.036) but not significantly correlated in men (P = 0.45). There is no significant interaction between sex and change in body fat percentage (P = 0.29). Abbreviations as in Figure 1.

Longitudinal changes in PBF and BFM were significantly associated with LAVI at follow-up (ΔPBF: β: −0.113; 95% CI: −0.204 to −0.022; P = 0.015; ΔBFM: β: −0.170; 95% CI: −0.316 to 0.024; P = 0.072), independent of baseline body fat or LAVI. Interval increases in BFM and PBF was associated with LVMI at follow-up in the overall cohort on univariate analysis, but this was attenuated on multivariate analysis (ΔBFM: β: −0.440; 95% CI: −0.939 to −0.059; P = 0.083; ΔPBF: β: −0.116; 95% CI: −0.486 to 0.153; P = 0.307) (Supplemental Table 3).

Interval changes in body fat had no significant association with the presence of impaired diastolic grading on echocardiography at follow-up in either men or women (P > 0.05) (Supplemental Table 4).

Clinical outcomes

At the follow-up assessment, 78 patients had at least 1 hospitalization, among which 18 of them were cardiovascular admissions. There were no mortalities recorded.

Both baseline and changes in BFM and PBF did not predict clinical outcomes in the overall cohort and in either men or women (Supplemental Table 5). Higher baseline WHR predicted clinical outcomes in the overall cohort (HR: 1.087; 95% CI: 1.022-1.155; P = 0.008), and particularly among women (HR: 1.091; 95% CI: 1.027-1.158; P = 0.005), but was nonsignificant in men (HR: 0.480; 95% CI: 0.006-39.265; P = 0.744). ΔWHR was not associated with clinical outcomes in the overall cohort and either sex. Baseline and ΔWHR was significantly associated with cardiovascular hospitalizations in the overall cohort (Baseline WHR: HR: 1.141; 95% CI: 1.069-1.217; P < 0.001; ΔWHR: HR: 0.876; 95% CI: 0.821-0.935; P < 0.001) and women (Baseline WHR: HR: 1.165; 95% CI: 1.073-1.265; P < 0.001; ΔWHR: HR: 0.857; 95% CI: 0.790-0.931; P < 0.001) but not men (P > 0.05 for all analysis) (Table 5).

Table 5.

Effect of Waist-Hip Ratio on Clinical Outcomes

	All-Cause Hospitalizations						Cardiovascular Hospitalizations
	Univariate			Multivariate			Univariate			Multivariate
	HR	95% CI	P Value	HR	95% CI	P Value	HR	95% CI	P Value	HR	95% CI	P Value
All
Baseline WHR	1.085	1.023-1.151	0.007	1.087a	1.022-1.155	0.008	1.141	1.069-1.217	<0.001	1.18b	1.099-1.266	<0.001
Change in WHR	0.921	0.869-0.977	0.006	0.759c	0.865-0.978	0.793	0.876	0.821-0.935	<0.001	0.847b	0.790-0.909	<0.001
Women
Baseline WHR	1.091d	1.027-1.158	0.005				1.165d	1.074-1.265	<0.001
Change in WHR	0.917d	0.863-0.973	0.004				0.857d	0.790-0.931	<0.001
Men
Baseline WHR	0.48	0.006-39.164	0.744				0.882	0.001-2,376.287	0.975
Change in WHR	0.98	0.038-25.022	0.99				8.092	0.018-3,603.649	0.502

Simple and multiple linear regression models of the effect of body fat metrics on clinical outcomes. Variables significant on univariate analysis were adjusted for other significant clinical variables. Bold values indicate P < 0.05.

Abbreviations as in Table 2.

^aAdjusted for age, diabetes status, cardiovascular risk category, E/e' ratio.

^bAdjusted for gender, cardiovascular risk category, diabetes status.

^cAdjusted for baseline WHR, age, diabetes status, cardiovascular risk status, E/e' ratio.

^dNo significant co-variate.

Discussion

In this longitudinal cohort study of community-dwelling older adults from a multiethnic Asian population free of overt cardiovascular disease, baseline and changes in body fat quantity were predictive of baseline and longitudinal alterations in left ventricular diastolic function over a period of 5 years, most notably in women. BFM and PBF exerted significantly different sex-dependent effects on long-term cardiac diastolic function. Higher central adiposity, as represented by WHR, was independently associated with long-term clinical and cardiovascular outcomes in women (Central Illustration).

Central Illustration.

Sex-Dependent Effects of Changes in Body Fat

Overview of the effects of changes in body fat metrics, as assessed via bioimpedance, on cardiac function and clinical outcomes in a cohort of community-dwelling adults free of cardiovascular disease. Significant sex-dependent effects on cardiac function and clinical outcomes were demonstrated between men and women. E/A = ratio of early to late diastolic inflow velocities; E/e' = ratio of mitral early diastolic inflow velocity to the average of septal and lateral early diastolic annular tissue velocities; LAVI = left atrial mass index.

Although obesity is an established cardiovascular risk factor and predictor of adverse clinical outcomes,¹⁴ the use of additional metrics to characterize obesity in aging is evolving¹⁵ as the field recognizes that older adults have fat-free and fat-dependent alterations with aging.¹⁶ BMI, the most common method for defining obesity, is a cost-effective and proven marker of cardiovascular risk and cardiac dysfunction.17, 18, 19 It is the definition that has been the most commonly applied in prior studies on the effects of obesity on cardiac function,20, 21, 22 despite its several limitations, including lack of specificity for fat mass, as well as sex-, age-, and race-related differences.^3,8 Although metrics such as WC and WHR have also been associated with cardiovascular risk^23,24 and cardiac dysfunction,⁶ they are indirect markers of central adiposity.

BIA is a noninvasive, safe, easy, and convenient method for measuring fat and lean body mass.²⁵ By measuring resistance to an electrical current applied to the skin, lean and fat body mass can be estimated.²⁶ Refinements in BIA technology over the past 2 decades have established it as a recognized method of monitoring body composition.²⁶ BIA metrics, such as PBF, have been demonstrated to be superior to BMI and WC in prognosticating future cardiovascular events, and are potential methods for classifying obesity in the future.⁷ Although the effects of longitudinal changes in BMI and WHR across adult life on cardiac function have previously been evaluated,²⁷ there is a lack of data regarding how baseline and longitudinal changes in BIA metrics affect cardiac diastolic function, an indicator of subclinical cardiac disease,⁶ which is critical for early detection and prevention.

Baseline body fat predicted the development of diastolic dysfunction, especially among women. This is consistent with prior data on the association of obesity, female sex, and diastolic dysfunction.^28,29 However, our work illustrates the impact of a variety of body fat metrics beyond BMI, on progressive cardiac risks over time. Among women, higher baseline BFM and PBF were independently associated with worsening myocardial relaxation as reflected by reductions in E/A ratio over time (BFM: β: 0.009; P = 0.029; PBF: β: 0.01; P = 0.028). Conversely, in men, higher baseline BFM and PBF predicted increases in E/e' (BFM: β: −0.073; P = 0.016; PBF: β: −0.069; P = 0.02), a surrogate of elevated LV filling pressures. Significant sex-specific interactions exist between baseline BFM and PBF on long-term deteriorations in myocardial relaxation.

Interval increases in body fat were significantly associated with specific markers of diastolic function including changes in E/A (β: −0.005; P = 0.047) and LAVI (β: −0.113; P = 0.015), suggestive of progressive impairments in myocardial relaxation and increasing left atrial pressures. A longer follow-up may be insightful. Interval increases in PBF was significantly associated with worsening of E/A ratio in women (β: −0.005; P = 0.029), independent of baseline PBF. Although this relationship was attenuated after removing outliers, the overall trend highlights the need for interval assessments of body fat when evaluating cardiovascular risk in women. Tracking changes in PBF thus appears to carry greater clinical significance in women compared to men, which may possibly be explained by hormonal differences. Postmenopausal hormonal shifts promote central adiposity, which is more strongly associated with cardiovascular risk.³⁰ Increases in total fat mass in older women may reflect concurrent increases in central adiposity, contributing to the observed sex-specific associations. Future work using segmental body composition analysis and charting specific cardiac changes over time for both sexes will be insightful.

Our study also affirms the association between central adiposity, as represented by WHR, and an increased risk of all-cause and cardiovascular hospitalizations. Sex-stratified analysis suggests that this relationship is significantly stronger in women than men. The relationship between central adiposity and increased cardiovascular risk is well known and linked via multiple direct and indirect pathophysiological mechanisms.³¹ Baseline BFM and percentage did not display significant association with any overt cardiovascular events in the present Asian cohort, in contrast to the findings in larger European cohorts.⁷ This difference may thus be related to the limitations of a smaller cohort size, shorter follow-up duration, or represent ethnic-related differences in body fat characteristics.³² The present study does not suggest any relationship between baseline WHR with cardiovascular risk in men although WHR has been shown to have a stronger association with cardiovascular mortality compared to BMI or WC.³³ This may be related to the narrower range of WHR (0.9 ± 0.1) in men compared to women (1.0 ± 2.3) in the present cohort, which resulted in insufficient sensitivity to detect the effect of changes in WHR in men. Future work defining body fat distribution may more accurately define cardiovascular risk.

Study Limitations

There are several limitations in the present study. The participants enrolled represent a predominantly Asian population and may not be generalizable to cohorts of other ethnicities. Body fat and echocardiographic data collection was performed at 2 time points with a 5-year interval, and fluctuations in body fat during this period cannot be accounted for. Although we used BIA in the current study, we acknowledge that other modalities such as dual-energy X-ray absorptiometry and magnetic resonance imaging may be more specific. However, these modalities are less practical and carry inherent radiation risks especially with dual-energy X-ray absorptiometry and are thus not feasible for a cohort of otherwise healthy community-dwelling adults. Moreover, BIA has been previously shown to correlate well with magnetic resonance imaging body composition.²⁵ Fluid status has been previously suggested to affect BIA accuracy, such as in heart failure patients.³⁴ Although hydration status of the current cohort was not described, participants were generally neither dehydrated nor volume overloaded due to heart or kidney diseases. Variations in fluid intake in otherwise healthy older adults have been shown to have negligible effect on BIA readout.³⁵

Mitral E/A ratio, although an important marker of LV diastolic function in myocardial aging, is related with it via a U-shaped relationship, which could be elevated in pseudonormalization. However, one of the key features of cardiac aging is impaired myocardial relaxation.³⁶ The pattern of pseudonormalization is less likely in this cohort who had normal left atrial structures. Despite the limitations of E/A, we demonstrated significant relationships between body fat parameters with the American Society of Echocardiography grading of diastolic function,¹³ and other metrics of diastolic function such as LAVI.

Lastly, our findings remain observational, which prohibits causal inferences. Nevertheless, our prospective longitudinal study design with detailed annotations is rigorous and novel. The use of repeated longitudinal assessments over a period of 5-years is a significant strength of this work.

Conclusions

In this prospective, multiethnic cohort of community-dwelling older Asian adults without overt cardiovascular disease, body fat quantitation via bioimpedance differentiated trajectories of cardiac diastolic function over time. Sex-specific differences between men and women, underscore the need for sex-tailored approaches to risk stratification based on body fat. Central adiposity, as represented by WHR, emerged as a key predictor of adverse clinical outcomes in women. Refining cardiovascular risk based on quantitative sex-specific fat targets for older adults may be a strategy in the future, especially with growing use of BIA. Future studies incorporating segmental fat distribution and sex-specific longitudinal trajectories are warranted to refine cardiovascular risk prediction in aging populations.

Perspectives.

COMPETENCY IN MEDICAL KNOWLEDGE: Obesity is a well-known risk factor for cardiovascular disease. There exist significant interactions between sex and baseline body fat in predicting interval changes in cardiac diastolic function. In particular, changes in body fat in women appear to carry the greatest effect on predicting changes in cardiac function.

COMPETENCY IN PATIENT CARE: Efforts should be made to educate and promote the importance of weight loss in preventing deteriorations in cardiac function, especially in women with no history of prior cardiovascular disease.

TRANSLATIONAL OUTLOOK: Bioimpedance analysis is a relatively simple yet accurate estimate of body fat without corresponding risks of radiation. Routine use of bioimpedance in older adults represent opportunities for cardiovascular risk prediction based on quantitative sex-specific fat targets.

Funding support and author disclosures

Dr Koh has received grant funding from the National Medical Research Council (MOH-001200, MOH-000153). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.