Changes in Left Ventricular Mass and Geometry in the Older Adults: Role of Body Mass and Central Obesity

Abstract

Background:

LV hypertrophy is an independent risk factor for cardiovascular outcomes. There are limited data about modifiable factors associated with progression of LV hypertrophy in the older adults. Our objective is to describe the changes in left ventricular (LV) mass and geometry over time in a predominantly older multi-ethnic cohort and to identify possible predictors of changes over time.

Methods:

We analyzed data from participants in the Northern Manhattan Study (NOMAS) who underwent serial echocardiographic studies, comparing the baseline and the most recent echocardiograms. We recorded changes in LV mass and geometry and correlated them with baseline characteristics using linear regression models.

Results:

There were 826 participants (mean age 64.2 ± 8.0 years) included in the analysis (time between measurements: 8.5 ± 2.7 years). Overall, LV mass index increased from 45.0 ± 12.7 to 50.3 ± 14.6 g/m^2.7 (p < 0.001). There were 548 participants (66.3%) with LV mass increase; 258 subjects (31.2%) showed worsening LV geometry. Multivariable analysis showed that change in LV mass index was independently associated with baseline LV mass index (β estimate: −17.000, [standard error: 1.508], p < 0.001), hypertension (2.094 [0.816], p=0.011), body mass index (0.503 [0.088], p < 0.001) and waist-to-hip ratio (1.031 [0.385], p=0.008).Both waist-to-hip ratio or waist-to-height ratio remained significantly associated with LV mass increase even after adjusting for body mass index (p= 0.008 and p=0.036, respectively)

Conclusions:

Regardless of race/ethnicity, LV mass progressed over time in the older adults. We also observed worsening geometry was frequent. Assessment of central obesity in the older population is important because indicators of central obesity add prognostic value over body mass index for the risk of LV mass increase.

INTRODUCTION

The number of the older adults – those aged 60 years or over – will be more than double by 2050, rising from 962 million globally in 2017 to 2.1 billion in 2050. ¹ Aging and body fat are two of the most powerful risk factors for high blood pressure. ² Pressure overload due to chronic hypertension leads to left ventricular (LV) hypertrophy, which is associated with adverse cardiac sequelae, such as development of heart failure, independent of traditional cardiac risk factors. ³

Despite its public health importance, the progression of LV mass over time in the older adults has not been thoroughly investigated. The published reports were limited to selected groups such as patients with hypertension ⁴, concentric LV geometry (from the Cardiovascular Health Study) ⁵, youths and younger adults ^{6, 7}, including a report from the Coronary Artery Risk Development in Young Adults (CARDIA) Study ⁸, or middle-aged participants from the Framingham Heart Study ^{9, 10}. Those studies showed that female sex, higher body mass index (BMI), higher blood pressure and aging itself were associated with an increase in LV mass over time. Although many patients with heart failure are older adults, the knowledge of the factors that promote adverse LV remodeling in the older population is limited.

Studies investigating an effect of central adiposity on LV remodeling in the older adults are also scarce. Unlike in young individuals, the evaluation of fat mass using BMI alone may not be sufficient in the older adults because BMI does not differentiate between lean and fat mass, and aging is associated with a substantial decrease in lean body mass and an increase in fat mass ¹¹. Indices of central adiposity may be more appropriate in the older adults; however, whether or not central adiposity has additive value in predicting LV mass changes in the older adults is not known.

In the present report, we aimed to investigate the natural course of LV mass and geometry, the factors associated with it, and whether central adiposity has an impact on changes in LV mass in the older adults, beyond that of BMI.

METHODS

Study population

The study cohort was derived from the Northern Manhattan Study (NOMAS), an epidemiological study evaluating the incidence and risk factors for stroke in the population of Northern Manhattan, which enrolled participants between 1993 and 2001 and followed them at yearly intervals. The study design and recruitment details regarding NOMAS have been described previously ¹². Overall, 2,415 out of 3,298 participants of the original cohort underwent a baseline 2-dimensional transthoracic echocardiographic study. Being enrolled in ancillary studies of NOMAS, 891 participants underwent one or more repeat echocardiograms during follow-up. As a myocardial infarction may directly cause adverse LV remodeling, we excluded from the present analysis participants who had a prior history of myocardial infarction and those who experienced one during the follow-up (n=65). The remaining 826 participants constitute the study population of the present report. Written informed consent was obtained from all study participants. The study was approved by the Institutional Review Boards of Columbia University Medical Center and the University of Miami.

Risk factor assessment and anthropometric measures

Hypertension was defined as systolic blood pressure ≥ 140 mm Hg or diastolic blood pressure ≥ 90 mm Hg at the time of the visit (mean of 2 readings obtained in sitting position), or the patient’s self-reported history of use of antihypertensive medications. Diabetes mellitus was defined as fasting blood glucose ≥ 126 mg/dl, tested on multiple occasions, or the patient’s self-reported history use of diabetes medications. Hypercholesterolemia was defined as total serum cholesterol ≥ 240 mg/dl, or the patient’s self-report of hypercholesterolemia with use of lipid-lowering treatment. BMI was calculated as weight in kilograms divided by the square of the height in meters. Waist circumference (WC) was measured at the level of the midpoint between the top of the iliac crest and the lower margin of the last palpable rib in the mid-axillary line. Hip circumference was measured at the largest circumference of the buttocks. Variables such as physical activity, cigarette smoking, and alcohol consumption were based on the patients’ self-report. Given its narrow distribution in the cohort, physical activity was used as a binary variable (regular physical activity or sedentary lifestyle). Cigarette smoking and alcohol consumption were also dichotomized (current user or not; ever or never user).

Echocardiographic assessment

Transthoracic echocardiography was performed using a commercially available system (Sonos 5500 or 7500; iE33; Philips, Andover, Massachusetts) by a trained registered sonographer following a standardized protocol. The LV linear dimensions were measured from a parasternal long-axis view according to the recommendations of the American Society of Echocardiography. In addition, because some patients have sigmoid septa, we also carefully measured the dimensions just distal to the septal bulge so that we would not overestimate LV mass. The LV mass index (LVMI) was calculated with an LV mass derived from the validated formula ¹³ and indexed to height^2.7 rather than to body surface area ¹⁴. As weight is present in the formula of body surface area and change of weight was one of the exposure variables, we choose to divide LV mass by height^2.7, another established method for indexing LV mass by body size.

LV morphology

We defined LV geometry using internal cutoffs because the characteristics of our study cohort (older population; predominantly Hispanic; high frequency of cardiovascular risk factors) are markedly different from the populations on which guidelines for cut-offs of LV mass are based ¹³. LV hypertrophy was defined as an LV mass index greater than the 90^th percentile of the participants without conditions associated with LV hypertrophy, such as hypertension, diabetes, obesity or cardiovascular disease. The cutoffs were 53.6 g/m^2.7 for men and 47.0 g/m^2.7 for women. Relative wall thickness (RWT) was calculated with the formula: 2× (posterior wall thickness at end-diastole /LV end-diastolic dimension).

Cutoffs of 0.52 for men and 0.60 for women were used as the upper limit of normal (again based on the 90^th percentile of the reference, as described above). LV geometry was categorized into four types on the basis of LV mass index and RWT: normal (normal LV mass, normal RWT), concentric remodeling (normal LV mass, abnormal RWT), eccentric hypertrophy (abnormal LV mass, normal RWT), and concentric hypertrophy (abnormal LV mass, abnormal RWT).

When more than two echocardiograms were available during follow up, the most recent one was used for the analysis. Worsening LV geometry was defined as 1) a change from normal geometry to any other category; or 2) a change from concentric remodeling to any LV hypertrophy pattern. The baseline and follow-up echocardiograms were read and quantified by the same experienced cardiologist (MDT). Intra-observer agreements of echocardiogram measurements were excellent (intra-class correlation coefficients >0.90).

Statistical analysis

Data are presented as the mean ± standard deviation (SD) for continuous variables and as proportions for categorical variables. Linear regression models were used to examine the association between change in LVMI and clinical/demographic variables. In all linear regression models, the outcome was the change in LVMI, which is the difference between the last measure of LVMI and the baseline measure of LVMI. The basic bivariate linear models included the variables of interest and log-transformed baseline measure of LVMI. The multivariable linear regression model was built based on the covariates that were significant at the 0.1 level in the basic bivariate model to examine the associations between increase in LV mass index and anthropometric measures (BMI, WC, waist-to-hip ratio [WHR], and waist-to-height ratio [WHtR]). ¹⁵ For all statistical analyses, a two-tailed P < 0.05 was considered significant. Data analysis was conducted SAS software version 9.4 [SAS Institute Inc., Cary, NC].

RESULTS

Table 1 shows the characteristics of the participants at enrollment. Mean age was 64.2 ± 8.0 years; mean interval between echocardiograms was 8.5 ± 2.7 years. The Hispanic subgroup represented over 60% of the total. Although the same participants (N=826) were examined at baseline and follow-up, the information on changes in blood pressure and in weight between baseline and follow-up was missing in 55 patients (Table 1). We observed an overall increase in LV mass index at the follow-up measurement (from 45.0 ± 12.7 g/m^2.7 to 50.3 ± 14.6 g/m^2.7, p < 0.001); 548 (66.3%) participants showed increase in LVMI (Table 2). There were 258 participants (31.2%) with worsening in LV geometry. Among the 564 participants with normal LV mass (either normal LV geometry or concentric remodeling) at baseline, 175 (31.6%) progressed to LV hypertrophy. We observed 67 strokes and 196 deaths during the follow-up. The number of strokes in participants who experienced increased (> 5%), decreased (< 5%) or unchanged (within ±5%) LV mass was 45 (10.3%), 14 (7.6%) and 8 (5.3%), respectively (p = 0.136). The number of deaths was 119 (27.4%), 44 (23.8%) and 33 (21.9%), respectively (p = 0.344).

Table 1.

Characteristics of the participants at the time of enrollment N = 826

Age, year		64.2 ± 8.0
Male		342 (41.4%)
Race/Ethnicity	Hispanic	518 (62.7%)
	Black	157 (19.0%)
	White	131 (15.9%)
	Other	20 (2.4%)
Hypertension		494 (59.8%)
Systolic BP change per year, mmHg/year (n=771)		−0.60±3.53
Diastolic BP change per year, mmHg/year (n=771)		−0.87±1.95
Diabetes		166 (20.1%)
Hypercholesterolemia		383 (46.4%)
LDL cholesterol, mg/dL		130.3 ± 34.1
HDL cholesterol, mg/dL		46.0 ± 14.1
Triglycerides, mg/dL		132.7 ± 72.1
Physical activity, any		467 (56.6%)
Alcohol consumption, ever		652 (78.9%)
Alcohol consumption, current		489 (59.2%)
Cigarette smoking, ever		441 (53.4%)
Cigarette smoking, current		120 (14.5%)
History of coronary artery disease		95 (11.5%)
Body mass index, kg/m2		27.7 ± 4.8
Waist, cm		93.1 ± 11.3
Waist-to-hip ratio		0.90 ± 0.09
Waist-to-height ratio		0.22 ± 0.03
Weight change per year, kg/year (n=771)		−0.09 ± 1.00
Time interval between measurements, years		8.5 ± 2.7

N (%) or mean ± SD.

BP, blood pressure; HDL, high-density lipoprotein; LDL, low-density lipoprotein; LV, left ventricular; LA, left atrium.

Table 2.

Change of echocardiographic variables

	First echo	Last echo	Difference	p-value
LV mass, g	168.9 ± 49.5	189.0 ± 56.3	20.1 ± 45.0	<0.001
LV mass / height2.7, g/m2.7	45.0 ± 12.7	50.3 ± 14.6	5.3 ± 11.9	<0.001
LV mass/body surface area, g/m2	94.2 ± 24.3	105.1 ± 27.8	11 ± 24.6	<0.001
LV systolic diameter, mm	28.2 ± 4.9	28.3 ± 5.2	0.1 ± 4.4	0.814
LV diastolic diameter, mm	43.8 ± 4.8	44.6 ± 4.9	0.9 ± 4.1	<0.001
Fractional Shortening, %	35.1 ± 8.0	36.5 ± 7.4	1.4 ± 8.1	<0.001
LA diameter, mm	35.4 ± 4.6	39.0 ± 5.4	3.7 ± 4.8	<0.001
LV relative wall thickness, mm	0.500 ± 0.089	0.514 ± 0.089	0.01 ± 0.10	<0.001
Septal wall thickness, mm	11.0 ± 2.01	11.7 ± 2.04	0.7 ± 2.0	<0.001
Posterior wall thickness, mm	10.8 ± 1.55	11.4 ± 1.64	0.5 ± 1.7	<0.001
LV geometry
Normal geometry	488 (59.1%)	359 (43.5%)	−15.6%	<0.001
Concentric remodelling	76 (9.2%)	84 (10.2%)	+1.0%
Eccentric hypertrophy	176 (21.3%)	246 (29.8%)	+8.5%
Concentric hypertrophy	86 (10.4%)	137 (16.6%)	+6.2%

LV, left ventricular; LA, left atrium.

Figure 1 illustrates the associations between the changes in LV structure and the change in mean arterial pressure, BMI, and WHtR. Change in mean arterial pressure was positively associated with a change in relative wall thickness (p=0.012, Figure 1B). Moreover, an increase in systolic BP (5% or more over baseline values) was associated with increase in LV mass index (6.78 ± 11.3 g/m^2.7, vs. 4.87 ± 11.1 g/m^2.7 for unchanged SBP and 3.96 ± 12.5 g/m^2.7 for decreased SBP; p = 0.026). However, the relationship between increase in SBP and increase in LV mass was not statistically significant after adjusting for baseline LVMI (Table 3).

Figure 1.

Panel A: relations between the change in LVMI and the change in mean arterial pressure, BMI and waist-to-height ratio. Panel B: relations between the change in LV relative wall thickness and the change in mean arterial pressure, BMI and waist-to-height ratio. The black line represents the population response line over the observed range of each variable.

LVMI, left ventricular mass index; BMI, body mass index; LVMI = LV mass/height^2.7

Table 3.

Associations of demographic and clinical variables with change in LV mass index (N=826, Bivariate model)

Variable	β estimate	Standard error	p value
Age, per 1 year	0.049	0.050	0.326
Male sex	−0.378	0.808	0.640
Hypertension	2.917	0.824	< 0.001
SBP change	0.176	0.115	0.126
DBP change	−0.062	0.209	0.765
Diabetes mellitus	1.881	0.998	0.060
Hypercholesterolemia	−0.166	0.797	0.835
Coronary artery disease	0.709	1.246	0.569
Race/Ethnicity
White	Reference	–	–
Black	1.458	1.350	0.280
Hispanic	1.269	1.126	0.260
Other	−3.754	2.737	0.171
Cigarette smoking, ever	−0.601	0.797	0.451
Cigarette smoking, current	1.265	1.127	0.262
Physical activity	−0.850	0.805	0.291
Alcohol consumption, ever	−1.070	0.975	0.273
Alcohol consumption, current	0.221	0.812	0.786
Time interval between measurements, per 1 year	−0.248	0.149	0.096
LDL cholesterol, per 1 mg/dL	0.011	0.012	0.339
HDL cholesterol, per 1 mg/dL	−0.066	0.028	0.020
Triglycerides, per 1 mg/dL	0.007	0.006	0.178
Body mass index, per 1 kg/m2	0.556	0.086	<0.001
Waist circumference, per SD	2.039	0.403	<0.001
Waist-to-hip ratio, per SD	1.340	0.384	<0.001
Waist-to-height ratio, per SD	2.726	0.410	<0.001
Weight change	0.254	0.409	0.534
Waist circumference change	−0.341	0.691	0.622

LV, left ventricular; SBP, systolic blood pressure; DBP, diastolic blood pressure; LDL, low-density lipoprotein; HDL, high-density lipoprotein.

LV mass index = LV mass/height^2.7

The basic bivariate linear models were adjusted for the variables of interest and the linear effects of log-transformed LV mass index of the baseline.

In the bivariate models (Table 3), the increase in LVMI was associated with hypertension, lower HDL cholesterol, and all indices of body size (BMI, WC, WHR, and WHtR); age at baseline and sex were not associated with increase in LVMI. Changes in blood pressure or body size (weight and WC) during the follow-up were also not associated with the change in LVMI. Table 4 shows the result of multivariate analysis after adjustment for clinical variables (selected by a p-value of < 0.1). Increase in LVMI was independently associated with the baseline LVMI, hypertension and all indices of body size. Table 5 demonstrates the incremental value of central adiposity indices over BMI in predicting the change of LVMI. WHR and WHtR remained significantly associated with increase in LVMI after adjustment for clinical variables and BMI, while WC did not show a significant association.

Table 4.

Associations between increase in LV mass index and indices of body fat (multivariable models)

	Clinical variables +BMI			Clinical variables + WC			Clinical variables + WHR			Clinical variables + WHtR
Clinical variables	β	SE	p value	β	SE	p value	β	SE	p value	β	SE	p value
log-transformed LV mass index of the baseline	−17.0	1.510	<0.001	−15.3	1.480	<0.001	−14.4	1.470	<0.001	−16.7	1.502	<0.0001
Hypertension	2.200	0.818	0.007	2.219	0.830	0.008	2.512	0.829	0.003	1.968	0.823	0.017
Diabetes mellitus	0.854	0.985	0.386	0.909	0.997	0.362	1.095	1.000	0.274	0.694	0.988	0.482
HDL	−0.036	0.028	0.199	−0.030	0.029	0.298	−0.046	0.029	0.112	−0.032	0.028	0.254
Time interval between measurements	−0.224	0.145	0.123	−0.198	0.146	0.176	−0.195	0.147	0.187	−0.201	0.145	0.165
Indices of body fat
BMI, per 1 kg/m2	0.505	0.088	< 0.001	–	–	–	–	–	–	–	–	–
WC, per SD	–	–	–	1.693	0.424	<0.001	–	–	–	–	–	–
WHR, per SD	–	–	–	–	–	–	1.052	0.393	0.008	–	–	–
WHtR, per SD	–	–	–	–	–	–	–	–	–	2.419	0.424	<0.0001

LV mass index = LV mass/height^2.7

SE, standard error; LV, left ventricular; HDL, high-density lipoprotein; SD, standard deviation; BMI, body mass index; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.

Covariates for multivariable models were selected based on their association shown in the basic bivariate model with a threshold set at a p-value of <0.1.

Table 5.

Additional effect of indices of central adiposity over BMI in the prediction of LV mass increase (multivariable model)

	Clinical variables + BMI + WC			Clinical variables + BMI + WHR			Clinical variables + BMI + WHtR
Variables	β	SE	p value	β	SE	p value	β	SE	p value
log-transformed LV mass index of the baseline	−17.000	1.520	< 0.001	−17.000	1.508	< 0.001	−17.200	1.514	< 0.0001
Hypertension	2.199	0.822	0.008	2.094	0.816	0.011	2.011	0.822	0.015
Diabetes mellitus	0.853	0.987	0.388	0.642	0.984	0.515	0.699	0.985	0.478
HDL	−0.036	0.029	0.212	−0.021	0.028	0.470	−0.030	0.028	0.282
Time interval between measurements	−0.223	0.145	0.124	−0.198	0.145	0.171	−0.212	0.145	0.144
BMI, per 1 kg/m2	0.504	0.123	< 0.001	0.503	0.088	< 0.001	0.295	0.133	0.027
WC, per SD	0.009	0.588	0.988	–	–	–	–	–	–
WHR, per SD	–	–	–	1.031	0.385	0.008	–	–	–
WHtR, per SD	–	–	–	–	–	–	1.351	0.642	0.036

LV mass index = LV mass/height^2.7

SE, standard error; LV, left ventricular; HDL, high-density lipoprotein; SD, standard deviation; BMI, body mass index; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio. Covariates for multivariable models were selected based on their association shown in the basic bivariate model with a threshold set at a p-value of <0.1.

DISCUSSION

The present study explored LV remodeling over time in a predominantly older multi-ethnic community cohort of Northern Manhattan and the clinical factors associated with LV mass changes. In our population, LV mass increased over time, and a worsening LV geometry was observed in one-third of participants. Increase in LVMI was independently associated with the baseline LVMI, hypertension, BMI and indices of central obesity (WC, WHR, and WHtR). Moreover, we found that WHR and WHtR remained significantly associated with increase in LV mass index even after adjustment for BMI.

Our analysis confirms that LV remodeling progresses over time in the older adults as was reported in younger cohorts (CARDIA⁸, Framingham Heart Study⁹). Similar to younger cohorts, hypertension and BMI are independently associated with increase in LVMI in the older adults. Moreover, we showed for the first time that central adiposity was significantly associated with increase in LV mass after adjustment for BMI and other clinical variables. We also observed that a higher portion of participants (≥ 30%) experienced progression from normal LV geometry or concentric remodeling to LV hypertrophy; this progression was infrequent in the younger cohorts from CARDIA⁸ and Framingham Heart Study. ¹⁰ Compared to those younger cohorts, we had a higher prevalence of LV hypertrophy, which was mainly due to the high prevalence of hypertension (previously presented in a separate paper ¹⁶), even though we used higher internal cutoffs than the ones reported in the guidelines.

Central adiposity was independently associated with LV mass change in our study. Although a previous cross-sectional study showed central obesity to be associated with LV hypertrophy ¹⁷, the relations between central obesity and the change of LV mass over time was not known before. Our observation is not only in agreement with previous reports showing that central adiposity is an independent predictor of cardiovascular disease, but also adds a possible mechanism for explaining them, since adverse LV remodeling is associated with increased risk of heart failure.

Among longitudinal echocardiographic studies published to date, our study population stands out for the number of participants of Hispanic origin. Previous large longitudinal echocardiographic studies did not include a considerable number of individuals of Hispanic descent; the CARDIA study ⁸ consisted of black and white participants; participants in the Framingham Heart Study were almost exclusively Caucasians ⁹. Describing longitudinal changes in cardiac structure in Hispanics is clinically relevant, as healthy individuals of Hispanic descent show thicker and smaller hearts compared to ASE guidelines-defined reference values ¹⁸. However, our results showed that the change in LVMI over time was not associated with race/ethnicity. (Table 3)

The predominantly older and multi-ethnic cohort of the present study reflects the current demographic landscape of the Western big cities, as a result of aging of the population and immigration patterns. Indeed, racial/ethnic minorities are the fastest growing segment of the older population in many Western countries ¹⁹. In this context, our result is a unique addition to the previous epidemiological studies which showed that central obesity is strong predictors of cardiovascular disease ^20–23. In the older adults, WHtR had greater ability to predict the development of cardiovascular risk factors, such as hypertension or diabetes, than other indices of central adiposity. ^24,25 In racially diverse populations, assessing central adiposity may be particularly important for several reasons. First, race/ethnicity may have an impact on the ratio between fat mass and lean body mass. For example, Whites tend to have a greater fat mass than Blacks at any given BMI because of lower bone mineral and body protein content. ²⁶ Second, in the INTERHEART ²² and the INTERSTOKE ²³, very large case-control studies whose objective was to identify the risk factors for acute myocardial infarction and acute stroke in each region of the globe, WHR predicted cardiovascular events regardless of race/ethnicity, while BMI did not.

The association between central adiposity and increase in LV mass in our study is also biologically plausible. Central adiposity is associated with increased insulin resistance ²⁷ as well as decreased adiponectin levels ²⁸. Increased insulin resistance may lead to vascular endothelial dysfunction, dyslipidemia, hypertension, and vascular inflammation, all of which promote the development of LV remodeling ²⁹. Previous literature showed that decreased adiponectin had an unfavorable effect on LV remodeling. In animal models, adiponectin-deficient mice presented more severe concentric hypertrophy ³⁰; in addition, an epidemiological study confirmed an inverse association between blood adiponectin level and LV mass in humans ³¹.

Finally, sex was not associated with increase in LVMI in our study, while previous studies in younger cohorts showed that women experienced a more pronounced increase in LV mass than men ^{8, 9}. Although the reason for this discrepancy is unclear, the lack of fluctuations in estrogen levels in the women of our older, post-menopausal cohort might provide a possible explanation, as there is evidence that estrogen plays an important role in LV remodeling ³².

Our findings have two implications. First, our results suggest the importance of prevention of LV remodeling at an earlier stage of life ⁹. In our cohort, LV mass at baseline was significantly associated with increase in LVMI over time, while age at baseline was not in itself associated with an increase in LV mass. Taken together with previous findings in younger cohorts, which experienced reverse remodeling through the modification of cardiovascular risk factors ^{4, 33}, this observation may suggest that the prevention of adverse LV remodeling at a younger age is essential to reducing the risk of further adverse remodeling later in life. Second, we confirmed the importance of the assessment of central obesity especially in an older population with diverse race/ethnicity composition because indices of central obesity, such as WHR or WHtR, may provide important prognostic information that is not accounted for by BMI.²²

Strengths and Limitations

There are several limitations to address. First, the observations of our study do not imply a causal relation between the explored variables and change in LVMI, as reverse causation cannot be excluded. For example, participants with the subclinical cardiac disease might develop abdominal adiposity because of an impaired ability to exercise. Second, although we adjusted our analyses for the most pertinent clinical and demographic variables, we cannot exclude the possibility that unmeasured confounders might be involved in the observed associations. For instance, we had no complete data on the change of antihypertensive medication over time. We also had no complete data regarding sleep apnea, which is known to be independently associated with LV remodeling. ³⁴ Third, the particular age and race-ethnicity composition of our cohort may preclude the generalization of our findings to populations with different demographic composition. Fourth, although calculating LVMI using transthoracic echocardiography is a common practice to assess for LV hypertrophy, LV mass calculation by echocardiography tend to be larger than that measured by cardiovascular magnetic resonance, the reference methods for in vivo LVMI assessment to date. ³⁵

However, our study also has strengths. First, this is the largest study and the one with the longest follow-up to track the course of LV mass and geometry in the older adults. Second, the follow-up data contained blood pressure and weight information, which enabled us to account at least in part for the effects of blood pressure/weight changes over time on LV remodeling. Lastly, given its tri-ethnic composition, our cohort was an ideal setting to assess possible race/ethnic influences on the results and included a majority of Hispanic participants, who represent an understudied race/ethnic subgroup.

CONCLUSIONS

In a predominantly older multi-ethnic community, we showed that LV mass progresses over time, and LV geometric changes occur in a sizeable proportion of individuals. Baseline LV mass, hypertension, and BMI are independently associated with an increase in LV mass, as is indices of central obesity (WC, WHR, and WHtR); WHR and WHtR appear to confer additional prognostic information for LV mass increase over BMI alone.

• There is limited data about modifiable factors associated with progression of LV hypertrophy in the elderly.

• In a predominantly elderly multi-ethnic cohort, we investigated the changes in LV mass and geometry over time.

• LV mass increased over time, and the increase was independently predicted by baseline LV mass, hypertension, body-mass index (BMI), and indices of central obesity (waist-to-hip ratio [WHR] or waist-to-height ratio [WHtR]). LV geometry worsened in over 30% of participants.

• WHR or WHtR remained significantly associated with increase in LV mass index after adjustment for BMI. This finding suggests that indices of central obesity provides important prognostic information for LV mass progression in an elderly multi-ethnic population, beyond that provided by BMI.

Abbreviations

LV

left ventricular

LVMI

LV mass index

NOMAS

Northern Manhattan Study

BMI

body-mass index

CARDIA study

the Coronary Artery Risk Development in Young Adults study

WC

waist circumference

WHR

waist-to-hip ratio

WHtR

waist-to-height ratio