Race-ethnic and sex differences in left ventricular structure and function: the Coronary Artery Risk Development in Young Adults (CARDIA) Study.

BACKGROUND: We investigated race-ethnic and sex-specific relationships of left ventricular (LV) structure and LV function in African American and white men and women at 43 to 55 years of age. METHODS AND
RESULTS: The Coronary Artery Risk Development in Young Adults (CARDIA) Study enrolled African American and white adults, age 18 to 30 years, from 4 US field centers in 1985-1986 (Year-0) who have been followed prospectively. We included participants with echocardiographic assessment at the Year-25 examination (n=3320; 44% men, 46% African American). The end points of LV structure and function were assessed using conventional echocardiography and speckle-tracking echocardiography. In the multivariable models, we used, in addition to race-ethnic and gender terms, demographic (age, physical activity, and educational level) and cardiovascular risk variables (body mass index, systolic blood pressure, diastolic blood pressure, heart rate, presence of diabetes, use of antihypertensive medications, number of cigarettes/day) at Year-0 and -25 examinations as independent predictors of echocardiographic outcomes at the Year-25 examination (LV end-diastolic volume [LVEDV]/height, LV end-systolic volume [LVESV]/height, LV mass [LVM]/height, and LVM/LVEDV ratio for LV structural indices; LV ejection fraction [LVEF], Ell, and Ecc for systolic indices; and early diastolic and atrial ratio, mitral annulus early peak velocity, ratio of mitral early peak velocity/mitral annulus early peak velocity; ratio, left atrial volume/height, longitudinal peak early diastolic strain rate, and circumferential peak early diastolic strain rate for diastolic indices). Compared with women, African American and white men had greater LV volume and LV mass (P<0.05). For LV systolic function, African American men had the lowest LVEF as well as longitudinal (Ell) and circumferential (Ecc) strain indices among the 4 sex/race-ethnic groups (P<0.05). For LV diastolic function, African American men and women had larger left atrial volumes; African American men had the lowest values of Ell and Ecc for diastolic strain rate (P<0.05). These race/sex differences in LV structure and LV function persisted after adjustment.
CONCLUSIONS: African American men have greater LV size and lower LV systolic and diastolic function compared to African American women and to white men and women. The reasons for these racial-ethnic differences are partially but not completely explained by established cardiovascular risk factors.

Introduction

Heart failure (HF) is more common in the US African American population, affecting ≈3% of all African American adults. This reflects a 50% greater incidence of HF among African Americans than seen in the remaining population.[1] In addition, African American men represent a vulnerable population for longer HF hospitalization.[2-3] The Multi‐Ethnic Study of Atherosclerosis (MESA) study suggested that African American populations are exposed to subsequent HF because of a higher rate of cardiovascular risk factors.[4] The Coronary Artery Risk Development in Young Adults (CARDIA) study prospectively assessed a young adult biracial cohort and reported predictors of incident HF before the age of 50 years including diastolic blood pressure, body mass index (BMI), and high‐density lipoprotein cholesterol, which are more prominent in African American than in white participants.[5]

Myocardial deformation assessed by speckle‐tracking echocardiography (STE) may be an earlier indicator of cardiac dysfunction compared to traditional echocardiography measurements.[6-8] Moreover, a detailed study of race–ethnic differences in cardiac function has not been performed in younger individuals. Two‐dimensional (2D) STE, as well as tissue Doppler imaging, have evolved as novel tools to accurately quantify both global and regional myocardial function. Strain is a measure of deformation expressed as a fractional or percentage change from an object's original dimension and has recently been used as an additional method to assess left ventricular (LV) function.[9] 2D‐STE is an angle‐independent method for deformation assessment that enables strain measurement in the longitudinal, circumferential, and radial directions based on conventional echocardiographic images. The echocardiographic method has been validated for measurements of early subclinical cardiac dysfunction by assessing myocardial deformation against tagged magnetic resonance imaging and sonomicrometry.[9-10]

In the MESA study, racial differences in regional LV systolic function in a large cohort study of adults have been previously reported.[11] However, the association of sex and race ethnic‐specific differences in younger adults among African Americans from a large community‐based population has not been reported. Based on what is known about the earlier onset of clinical heart disease among African American men, we hypothesize that (1) white and female CARDIA participants will have better LV systolic and diastolic myocardial function, compared to African Americans and male participants, respectively; and (2) these differences in function are partially or completely explained by body size and exposure to cardiovascular risk factors. We propose to investigate such relationships in the CARDIA study, a large community‐based population including African American and white men and women who have been followed for 25 years.

Methods

Subjects

The CARDIA study is a National Institutes of Health/National Heart, Lung, and Blood Institute‐sponsored multicenter study designed to investigate the development and determinants of coronary disease risk factors in young adults. Initially, 5115 African American and white men and women 18 to 30 years of age at the time of enrollment (1985–1986) were recruited and examined at 4 CARDIA Field Centers in Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California. Echocardiography was performed in the cohort in the follow‐up Year‐25 examination. The overall design and objectives of the CARDIA study have been presented elsewhere.[12] The institutional review board at each of the study sites approved the study protocols, and written informed consent was obtained from all participants.

Of the 3499 participants attending the Year‐25 examination, 3475 underwent echocardiography. For this study, we included 3474 participants with echocardiographic assessment at the CARDIA examinations for Year‐25 (2010–2011). Of these, 1 changed sex, and 154 missed at least 1 of the clinical covariates used in the study. The remaining 3320 patients were included in our analytic cohort.

Echocardiography

Doppler echocardiography and 2D‐guided M‐mode echocardiography were performed with the Artida cardiac ultrasound scanner (Toshiba Medical Systems, Otawara, Japan) by trained sonographers using standardized protocols across all field centers. In a central laboratory, experienced sonographers made measurements from digitized images using a standard software off‐line image‐analysis system (Digisonics, Inc, Houston, TX). Using 2D‐guided M‐mode echocardiography, LV end‐diastolic volume (LVEDV), LV end‐systolic volume (LVESV), and left atrial volume (LAV) was measured from an apical 4‐chamber view based on the American Society of Echocardiography (ASE).[13] LVEDV, LVESV, LAV, and LV mass (LVM) were indexed to body height (LVEDV/ht, LVESV/ht, LAV/ht, and LVM/ht). Peak early diastolic velocity, peak late diastolic velocity, early diastolic and atrial ratio, isovolumetric relaxation time, and early peak diastolic mitral annular velocity (e') were measured from pulsed‐Doppler echocardiographic recordings of transmitral flow. Using tissue Doppler imaging, e' was measured at the septal mitral annulus.[14]

Two‐Dimensional STE Analysis

STE images for myocardial strain and strain rate measurements were analyzed in a 16‐segment basis for LV midwall layer, using Wall Motion 2D Tracking software (Toshiba Medical Systems). Three cardiac cycles from each view were recorded for offline analyses.[15] Strain was calculated as the change in segment length relative to its end‐diastolic length from peak systolic values. Longitudinal strain and strain rate curves were assessed from 4‐chamber views. Circumferential strain and strain rate were assessed from the short‐axis view at midventricular level. Global strain values were calculated as the average of segmental peak strains. Global strain rate values were also calculated from the average of segmental peak values for each phase (in sec−1). The STE image set in each view was excluded if more than 3 segments were improperly tracked. Excluded cases for poor image quality for both the echo and STE measurements included 392 cases for longitudinal strain and strain rate assessed from the 4‐chamber view and 353 cases for circumferential strain and strain rate assessed from the short‐axis view. STE indices of systolic cardiac deformation at the Year‐25 examination included 4‐chamber longitudinal peak strain (Ell) and circumferential peak strain (Ecc), and 4‐chamber longitudinal systolic strain rate (Ell_SRs), and circumferential systolic strain rate (Ecc_SRs). Diastolic STE indices were as follows: peak early diastolic strain rate in the 4‐chamber longitudinal (Ell_SRe) and circumferential (Ecc_SRe).

Covariates

Standardized protocols were used to measure height, weight, heart rate, blood pressure, cholesterol, glucose, smoking, educational level, and physical activity at baseline and at the Year‐25 examination.[12,16] Sex and ethnicity were self‐reported. Weight and height were measured with the use of a standard balance‐beam scale. Weight was measured to the nearest 0.1 kg with the participants wearing light clothing. Height was measured to the nearest 0.5 cm without shoes. BMI was calculated as weight (kg) divided by height squared (m2). We used the average of the second and third of 3 blood pressure measurements performed at 1‐minute intervals after the participant had been sitting quietly for 5 minutes in a quiet room. The presence of diabetes was assessed at each examination based on a combination of medication use for diabetes (Years‐0 and ‐25), fasting plasma glucose ≥126 mg/dL (Years‐0 and ‐25), 2‐hour glucose ≥200 mg/dL (Year‐25), or HbA1c ≥6.5% (Year‐25). Information on demographic characteristics, smoking status, educational level, and medication use was collected at the time of clinical interview.

Statistical Analysis

Descriptive statistics were displayed using means and SDs for continuous variables. Categorical variables were present as numbers and percentages. Paired or unpaired t tests, χ2 tests, F‐tests, Bonferroni adjustment for multiple comparison, and Tukey–Kramer tests were used to compare the difference in prevalence of various risk factors among race–ethnicity and sex subgroups. LV structural parameters were defined as LVEDV/ht (height), LVESV/ht, LVM/ht, and LVM/LVEDV ratio; LV systolic function was defined as LV ejection fraction (LVEF), Ell, and Ecc; and LV diastolic function was defined as E/A ratio, e, LAV/ht, Ell_SRe, and Ecc_SRe. We conducted univariate linear regression analysis to assess the association of race/sex groups with LV structure and function. Multivariable linear regression models were used to assess the association of race/sex groups with LV structure and function. Model 1: LV structure or LV function=demographic variables (age)+BMI at Year‐25; Model 2: LV structure or LV function=Model 1+Year‐25 risk factors (systolic blood pressure [mm Hg], diastolic blood pressure [mm Hg], heart rate [bpm], presence of diabetes [yes/no], use of antihypertensive medications [yeas/no], number of cigarettes/day, physical activity [exercise unit], and educational level [years]); and Model 3: LV structure or LV function=Model 2+Year‐0 risk factors (BMI, systolic blood pressure, diastolic blood pressure, heart rate, presence of diabetes, use of antihypertensive medications, number of cigarettes/day, physical activity, and educational level). BMI at Year‐25 was included in Model 1 and Model 2. By including BMI at Year‐0, Model 3 suggested that BMI at Year‐0 could have additional influence on the current LV structural and functional parameters. The correlation between BMI at Year‐0 and BMI at Year‐25 was 0.69 (P<0.0001). On the potential collinearity between BMI at Year‐0 and Year‐25 for the LV structural and functional parameters, the variance inflation factor of the Year‐0 and Year‐25 was <10, suggesting that the collinearity between BMI at Year‐0 and Year‐25 was not too high (Table 1). Since BMI changed between Year‐0 and Year‐25 for many study participants, incorporating the BMI values at both time points in Model 3 provided additional information on the influence of BMI on the LV structural and functional parameters. A 2‐sided P value of <0.05 was considered to be statistically significant. All statistical analyses were performed using JMP (version 10.0 for Windows, SAS Institute Inc, Cary, NC) and STATA (version 12.0, Stata Corp, College Station, TX).

Table 1.

Potential Collinearity Assessment Between BMI at Year‐0 and Year‐25 for the LV Structural and Functional Parameters

	n	R 2	Y‐25 BMI		Y‐0 BMI
			β (SE)	VIF	β (SE)	VIF
Model 1
Structural indices
LVEDV/height, mL/m	3078	0.07	0.61 (0.04)*	1.00
LVM/height, g/m	2980	0.20	1.88 (0.07)*	1.00
Systolic indices
LVEF, %	3077	0.001	−0.03 (0.02)	1.00
Ell, %	2908	0.04	0.07 (0.01)*	1.00
Diastolic indices
E/e' ratio	3239	0.06	0.08 (0.01)*	1.00
Ell_SRe, sec−1	2896	0.01	−0.004 (0.001)*	1.00
Model 2
Structural indices
LVEDV/height, mL/m	3078	0.08	0.32 (0.06)*	1.89	0.61 (0.08)*	1.89
LVM/height, g/m	2980	0.22	0.09 (0.09)*	1.92	1.42 (0.14)*	1.92
Systolic indices
LVEF, %	3077	0.002	0.01 (0.03)	1.89	−0.10 (0.04)*	1.89
Ell, %	2908	0.04	0.07 (0.01)*	1.86	−0.003 (0.01)	1.86
Diastolic indices
E/e' ratio	3239	0.06	0.08 (0.01)*	1.91	0.003 (0.01)	1.91
Ell_SRe, sec−1	2896	0.02	−0.002 (0.001)	1.85	−0.006 (0.001)*	1.85

Model 1 is adjusted for BMI at Year‐25. Model 2 is adjusted for BMI at Year‐25 and Year‐0. BMI indicates body mass index; E/e', ratio of mitral early peak velocity/mitral annulus early peak velocity; Ell, longitudinal peak systolic strain; Ell_SRe, longitudinal peak early diastolic strain rate; LV, left ventricular; LVEDV, LV end‐diastolic volume; LVEF, LV ejection fraction; LVM, LV mass; VIF, variance inflation factor.

P<0.05.

Results

Descriptive statistics for the cohort (n=3320) at Years‐0 and ‐25 are presented in Table 2. The study population was 43.6% male and 46.3% African American with a mean age of 50.2±3.6 at the Year‐25 examination. BMI for all groups increased over 25 years with an increased prevalence of overweight and obesity. African American women had higher mean BMI, higher prevalence of hypertension, and lower educational level and physical activity compared with white women at Year‐0. African Americans had higher mean BMI, higher prevalences of hypertension and diabetes, and lower educational level and physical activity at Year‐25 compared to whites. The echocardiographic parameters at the Year‐25 examination are shown in Table 2. African American men had the highest LVM/ht, LVEDV/ht, and LVESV/ht compared with other race/sex groups (Table 3). For LV systolic function, LVEF, Ell, Ecc, Ell_SRs, and Ecc_SRs in African American men were the lowest among all groups (Table 3). For LV diastolic function, e', Ell_SRe, and Ecc_SRe in African American men were also the lowest among all groups (Table 3).

Table 2.

Participant Characteristics at the CARDIA Year‐0 and Year‐25 Examinations (n=3320)

Variable	African American Men (n=616, 18.6%)		African American Women (n=920, 27.7%)		White Men (n=830, 25.0%)		White Women (n=954, 28.7%)
	Year‐0	Year‐25	Year‐0	Year‐25	Year‐0	Year‐25	Year‐0	Year‐25
Age, y	24.3 (3.8)*	49.4 (3.8)**	24.6 (3.8)*	49.6 (3.8)**	25.6 (3.3)	50.7 (3.4)*	25.6 (3.4)	50.8 (3.4)*
BMI, kg/m2	24.6 (4.3)*	30.4 (6.8)**	25.9 (6.4)*	33.5 (8.1)**	24.3 (3.5)*	28.8 (5.0)**	22.9 (4.1)	28.0 (7.1)*
BSA, m2	1.94 (0.18)*	2.12 (0.24)**	1.75 (0.21)*	1.95 (0.23)**	1.95 (0.16)*	2.09 (0.20)**	1.68 (0.15)	1.82 (0.21)*
Heart rate, beats/min	63.9 (9.4)*	67.1 (10.8)*	71.4 (10.6)	68.2 (11.0)**	67.0 (10.1)*	64.7 (10.7)**	71.9 (10.9)	66.3 (9.9)*
SBP, mm Hg	115.4 (10.1)*	126.0 (15.1)**	108.1 (10.0)*	123.3 (17.5)**	114.2 (10.1)*	119.2 (13.6)**	104.5 (9.2)	112.3 (14.4)*
DBP, mm Hg	70.2 (10.0)*	78.5 (11.0)**	67.3 (9.1)*	78.2 (11.2)**	70.8 (9.4)*	73.9 (10.0)**	66.2 (8.3)	70.0 (10.5)*
Hypertension, n (%)	23 (3.7)	274 (44.5)**	49 (5.3)*	471 (51.2)**	26 (3.1)	218 (26.3)**	15 (1.6)	186 (19.5)*
Diabetes, n (%)	0 (0)	109 (17.7)**	6 (0.7)	157 (17.1)*	3 (0.3)	67 (8.1)*	4 (0.4)	56 (5.9)*
Number of cigarettes, cig/day	4.4 (7.0)*	2.9 (6.2)**	3.9 (7.0)*	1.6 (4.3)*	5.9 (10.0)*	2.0 (6.1)*	5.8 (9.4)	1.2 (4.2)*
Using antihypertensive medication, n (%)	9 (1.5)	210 (34.1)**	38 (4.1)*	380 (41.3)**	11 (1.3)	166 (20.0)**	10 (1.1)	136 (14.3)*
Education level, y	13.1 (1.9)*	13.8 (2.5)**	13.2 (1.8)*	14.3 (2.4)**	14.9 (2.4)	15.9 (2.7)*	14.8 (2.2)	16.0 (2.5)*
Physical activity, EU	525 (345)*	371 (295)**	273 (223)*	228 (222)**	514 (295)*	431 (287)**	414 (267)	343 (261)*

BMI indicates body mass index; BSA, body surface area; CARDIA, Coronary Artery Risk Development in Young Adults; DBP, diastolic blood pressure; EU, exercise unit; SBP, systolic blood pressure.

P<0.05: vs Year‐0 white women.

P<0.05: vs Year‐25 white women.

P<0.05: vs Year‐0 among the same race/gender group.

Table 3.

Conventional Echocardiographic Parameters and Speckle‐Tracking Echocardiographic Parameters at the Year‐25 Examination

Variables	African American Men (n=616, 18.6%)		African American Women (n=920, 27.7%)		White Men (n=830, 25.0%)		White Women (n=954, 28.7%)
	n	Year‐25	n	Year‐25	n	Year‐25	n	Year‐25
Conventional echocardiography
LVEDV/height, mL/m	577	73.2 (18.3)*	842	62.2 (13.6)*	784	71.4 (16.5)*	875	57.9 (11.3)
LVESV/height, mL/m	577	30.2 (12.4)*	842	23.6 (8.3)*	784	28.3 (10.8)*	874	22.1 (6.0)
LVSV/height, mL/m	577	43.0 (9.7)*	842	38.6 (8.5)*	784	43.1 (9.2)*	874	35.8 (7.4)
LVEF, %	577	59.5 (7.8)*	842	62.5 (7.5)	784	60.8 (6.9)*	874	62.0 (6.2)
PW, mm	553	9.7 (1.8)*	848	9.0 (1.7)*	710	9.1 (1.5)*	871	8.0 (1.4)
IVS, mm	577	10.0 (2.0)*	848	9.1 (1.7)*	710	9.3 (1.5)*	873	8.2 (1.4)
LVM, g	552	202.3 (58.7)*	848	161.2 (47.0)*	709	186.6 (44.9)*	871	137.0 (36.1)
LVM/height, g/m	552	114.0 (32.9)*	848	98.1 (28.1)*	709	104.8 (24.7)*	871	83.1 (21.8)
RWT	551	0.37 (0.08)*	848	0.37 (0.08)*	706	0.34 (0.07)*	871	0.33 (0.07)
LVM/volume ratio	523	1.61 (0.52)*	792	1.63 (0.52)*	678	1.53 (0.49)*	815	1.48 (0.44)
E, cm/s	612	75.7 (15.8)*	893	83.1 (16.9)	822	73.9 (14.5)*	939	80.9 (14.9)
A, cm/s	612	61.5 (14.5)*	893	70.3 (18.3)*	822	57.5 (14.0)*	939	64.6 (15.9)
E/A ratio	612	1.28 (0.35)	893	1.24 (0.37)*	822	1.34 (0.34)	939	1.32 (0.38)
LA volume/height, mL/m	611	30.9 (9.6)*	902	30.5 (9.6)*	824	29.2 (8.4)*	943	26.5 (7.9)
e', cm/s	603	8.7 (2.3)*	902	9.1 (2.5)*	820	9.3 (2.2)*	936	9.8 (2.4)
E/e' ratio	601	9.2 (2.9)*	890	9.7 (3.1)*	817	8.4 (2.3)*	931	8.7 (2.6)
Speckle‐tracking echocardiography
Ell, %	540	−14.1 (2.3)*	781	−14.9 (2.5)*	739	−15.0 (2.3)*	848	−16.0 (2.3)
Ecc, %	560	−14.4 (2.8)*	812	−15.5 (3.0)	739	−15.2 (2.7)*	856	−15.8 (2.7)
Ell_SRs, sec−1	540	−0.63 (0.13)*	781	−0.65 (0.12)*	739	−0.66 (0.11)*	848	−0.69 (0.12)
Ecc_SRs, sec−1	560	−0.69 (0.16)*	812	−0.70 (0.16)	724	−0.70 (0.15)	856	−0.71 (0.14)
Ell_SRe, sec−1	537	0.74 (0.23)*	776	0.82 (0.26)*	737	0.79 (0.23)*	846	0.91 (0.27)
Ecc_SRe, sec−1	560	0.71 (0.30)*	809	0.82 (0.33)	723	0.78 (0.32)*	855	0.84 (0.32)

Values are mean (SD). LVEDV, LVESV, LVSV, LAV, and LVM were indexed to body height (m). A indicates mitral late peak velocity; E, mitral early peak velocity; E/A, early diastolic and atrial ratio; E/e', ratio of mitral early peak velocity/mitral annulus early peak velocity; e', mitral annulus early peak velocity; Ecc, circumferential peak systolic strain; Ecc_SRe, circumferential peak early diastolic strain rate; Ecc_SRs, circumferential peak systolic strain rate; Ell, longitudinal peak systolic strain; Ell_SRe, longitudinal peak early diastolic strain rate; Ell_SRs, longitudinal peak systolic strain rate; IVS, interventricular septum thickness; LA, left atrium; LVEDV, left ventricular end‐diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end‐systolic volume; LVM, left ventricular mass; LVSV, left ventricular stroke volume; PW, posterior wall thickness; RWT, relative wall thickness.

P<0.05: vs white women as the referent group among race–ethnicity and sex groups.

Race–Ethnicity and Sex Variations in LV Size and Structure

Being an African American man was significantly associated with larger LV structural parameters (Table 4). BMI at Year‐25 was a significant determinant of differences in LVEDV/ht and LVESV/ht comparing African American men and white women in Model 1 (Table 4). The differences remained consistent in Models 2 and 3. BMI, heart rate, education level, systolic and diastolic blood pressure, smoking, and antihypertensive medications at Year‐25 attenuated the differences in LVM/ht between race–ethnicities in Model 2. Additionally, the magnitude of differences in LVM/ht remained after adjusting for Year‐0 risk factors in Model 3.The LVM/LVEDV ratio was not different between women and men in univariate analyses. BMI, physical activity, diastolic blood pressure, antihypertensive medications at Year‐25 and BMI, number of cigarettes per day, and diabetes at Year‐0 attenuated the differences in the LVM/LVEDV ratio between race–ethnicities in Models 2 and 3.

Table 4.

Relationship of Race–Ethnicity and Sex Group to LV Structural and Functional Parameters at the Year‐25 Examination

	n	R 2	African American Men	African American Women	White Men	White Women
			β (SE)	β (SE)	β (SE)	β (SE)
Structural indices
LVEDV/height, mL/m
Unadjusted	3078	0.15	15.27 (0.79)‡	4.33 (0.71)‡	13.50 (0.73)‡	Ref
Model 1	3078	0.23	13.63 (0.77)‡	0.54 (0.72)	12.89 (0.69)‡	Ref
Model 2	3078	0.25	12.35 (0.84)‡	0.09 (0.75)	12.00 (0.75)‡	Ref
Model 3	3078	0.26	11.06 (0.91)‡	−0.18 (0.77)	10.79 (0.80)‡	Ref
LVESV/height, mL/m
Unadjusted	3077	0.11	8.11 (0.50)‡	1.51 (0.45)†	6.20 (0.46)‡	Ref
Model 1	3077	0.15	7.26 (0.50)‡	−0.31 (0.47)	5.92 (0.45)‡	Ref
Model 2	3077	0.25	6.73 (0.55)‡	−0.45 (0.50)	5.67 (0.50)‡	Ref
Model 3	3077	0.26	6.33 (0.60)‡	−0.51 (0.51)	5.17 (0.53)‡	Ref
LVM/height, g/m
Unadjusted	2980	0.15	30.88 (1.45)‡	14.94 (1.29)‡	21.68 (1.35)‡	Ref
Model 1	2980	0.33	26.30 (1.31)‡	4.97 (1.21)‡	20.11 (1.20)‡	Ref
Model 2	2980	0.37	19.50 (1.41)‡	−0.14 (1.25)	17.35 (1.28)‡	Ref
Model 3	2980	0.39	18.15 (1.52)‡	−0.21 (1.27)	15.46 (1.36)‡	Ref
LVM/LVEDV ratio*
Unadjusted	2808	0.02	14.0 (3.00)‡	15.00 (2.00)‡	6.00 (3.00)*	Ref
Model 1	2808	0.05	10.97 (2.75)‡	8.100 (2.56)†	5.02 (2.51)*	Ref
Model 2	2808	0.07	2.45 (3.03)	0.69 (2.71)	2.52 (2.74)	Ref
Model 3	2808	0.07	3.90 (3.31)	1.18 (2.80)	2.30 (2.94)	Ref
Systolic indices
LVEF, %
Unadjusted	3077	0.02	−2.51 (0.40)‡	0.52 (0.34)	−1.17 (0.35)†	Ref
Model 1	3077	0.03	−2.25 (0.38)‡	0.94 (0.36)†	−1.11 (0.35)†	Ref
Model 2	3077	0.04	−2.31 (0.43)‡	0.80 (0.38)*	−1.25 (0.38)†	Ref
Model 3	3077	0.04	−2.48 (0.47)‡	0.73 (0.39)	−1.23 (0.41)†	Ref
Ell, %
Unadjusted	2908	0.07	1.91 (0.13)‡	1.11 (0.12)‡	0.97 (0.12)‡	Ref
Model 1	2908	0.10	1.82 (0.13)‡	0.82 (0.12)‡	0.92 (0.12)‡	Ref
Model 2	2908	0.15	1.30 (0.14)‡	0.50 (0.12)‡	0.62 (0.12)‡	Ref
Model 3	2908	0.16	1.37 (0.15)‡	0.50 (0.13)‡	0.70 (0.13)‡	Ref
Ecc, %
Unadjusted	2952	0.03	1.39 (0.15)‡	0.25 (0.14)	0.55 (0.14)‡	Ref
Model 1	2952	0.04	1.24 (0.15)‡	−0.01*(0.14)	0.51 (0.14)‡	Ref
Model 2	2952	0.05	0.94 (0.17)†	−0.12 (0.15)	0.35 (0.15)†	Ref
Model 3	2952	0.06	1.02 (0.19)‡	−0.26 (0.16)	0.39 (0.16)†	Ref
Diastolic indices
E/A ratio
Unadjusted	3266	0.01	−0.04 (0.02)	−0.07 (0.02)‡	0.02 (0.02)	Ref
Model 1	3266	0.08	−0.03 (0.02)	−0.04 (0.02)†	0.03 (0.02)	Ref
Model 2	3266	0.21	0.04 (0.02)*	0.02 (0.02)	0.05 (0.02)†	Ref
Model 3	3266	0.21	0.03 (0.02)	0.02 (0.02)	0.05 (0.02)†	Ref
e', cm/s
Unadjusted	3264	0.03	−1.10 (0.12)‡	−0.70 (0.11)‡	−0.55 (0.11)‡	Ref
Model 1	3264	0.12	−1.14 (0.12)‡	−0.58 (0.11)‡	−0.53 (0.11)‡	Ref
Model 2	3264	0.21	−0.50 (0.13)‡	−0.12 (0.11)	−0.21 (0.12)	Ref
Model 3	3264	0.23	−0.67 (0.13)‡	−0.15 (0.12)	−0.36 (0.12)†	Ref
E/e' ratio
Unadjusted	3239	0.05	0.51 (0.14)‡	1.02 (0.13)‡	−0.37 (0.13)†	Ref
Model 1	3239	0.10	0.49 (0.14)†	0.80 (0.13)‡	−0.40 (0.13)†	Ref
Model 2	3239	0.17	−0.29 (0.15)	0.29 (0.14)*	−0.82 (0.14)‡	Ref
Model 3	3239	0.18	−0.08 (0.17)	0.37 (0.15)†	−0.68 (0.15)‡	Ref
LA volume/height, mL/m
Unadjusted	3280	0.04	4.34 (0.46)‡	3.94 (0.41)‡	2.69 (0.42)‡	Ref
Model 1	3280	0.16	3.41 (0.43)‡	1.60 (0.40)‡	2.35 (0.39)‡	Ref
Model 2	3280	0.20	2.82 (0.47)‡	1.36 (0.42)†	2.07 (0.42)‡	Ref
Model 3	3280	0.22	1.96 (0.51)‡	1.34 (0.43)†	1.19 (0.45)†	Ref
Ell_SRe, sec−1
Unadjusted	2896	0.06	−0.17 (0.01)‡	−0.09 (0.01)‡	−0.12 (0.01)‡	Ref
Model 1	2896	0.09	−0.18 (0.01)‡	−0.08 (0.01)‡	−0.12 (0.01)‡	Ref
Model 2	2896	0.12	−0.13 (0.02)‡	−0.05 (0.01)‡	−0.09 (0.01)‡	Ref
Model 3	2896	0.12	−0.12 (0.02)‡	−0.05 (0.01)†	−0.08 (0.01)‡	Ref
Ecc_SRe, sec−1
Unadjusted	2947	0.02	−0.13 (0.02)‡	−0.02 (0.02)	−0.06 (0.02)‡	Ref
Model 1	2947	0.02	−0.12 (0.02)‡	−0.01 (0.02)	−0.06 (0.02)‡	Ref
Model 2	2947	0.03	−0.10 (0.02)‡	0.002 (0.02)	−0.04 (0.02)‡	Ref
Model 3	2947	0.03	−0.09 (0.02)‡	0.0002 (0.02)	−0.03 (0.02)	Ref

Model 1 is adjusted for age and BMI at Year‐25. Model 2 is further adjusted for Model 1+Year‐25 risk factors (educational level, physical activity, heart rate, presence of diabetes, systolic blood pressure, diastolic blood pressure, use of antihypertensive medications, and number of cigarettes/day). Model 3 is further adjusted for Model 2+Year‐0 risk factors (BMI, educational level, physical activity, heart rate, presence of diabetes, systolic blood pressure, diastolic blood pressure, use of antihypertensive medications, and number of cigarettes/day). BMI indicates body mass index; E/A, early diastolic and atrial ratio; E/e', ratio of mitral early peak velocity/mitral annulus early peak velocity; e', mitral annulus early peak velocity; Ecc, circumferential peak systolic strain; Ecc_SRe, circumferential peak early diastolic strain rate; Ell, longitudinal peak systolic strain; Ell_SRe, longitudinal peak early diastolic strain rate; LA, left atrial; LV, left ventricular; LVEDV, LV end‐diastolic volume; LVEF, left ventricular ejection fraction; LVESV, LV end‐systolic volume; LVM, LV mass.

*P<0.05; †P<0.01; ‡P<0.001: vs white women as the referent group among race–ethnicity and sex groups.

LVM/LVEDV was multiplied by 100.

Race–Ethnicity and Sex Differences in LV Systolic Function

Being an African American man was significantly associated with worse LV systolic function (Table 4). There were significant differences in LVEF after adjustment for demographics and BMI in Model 1 (Table 4). BMI, age, and heart rate at Year‐25 and BMI at Year‐0 were significant determinants of attenuation of the differences in LVEF between race–ethnicities. However, men maintained lower LVEF than women after adjustments in Models 2 and 3. African American men had the least Ell and Ecc in Models 1 to 3 compared to white women. These sex/race–ethnic differences in Ell persisted after adjustment for risk factors at both Year‐0 and 25. For Ecc, there were no differences between African American women and white women in univariate and multivariable analyses. The β‐coefficient of systolic parameters for the African American men was almost 2.0 times larger than the β‐coefficient for the white men.

Race–Ethnicity and Sex Differences in LV Diastolic Function

Being an African American man was significantly associated with worse LV diastolic function (Table 4). BMI and age were significant determinants that attenuated the differences in E/A ratio among race and sex in Model 1 among conventional LV diastolic parameters, (Table 4). For the e', BMI, age, heart rate, education level, physical activity, diastolic blood pressure, diabetes, and antihypertensive medications at Year‐25 attenuated the differences between race–ethnicities. The differences were still present after adjusting for Year‐0 risk factors. BMI, heart rate, education level, physical activity, systolic and diastolic blood pressure at Year‐25 and BMI, heart rate, educational level, physical activity, and systolic and diastolic blood pressure at Year‐0 were significant attenuators of the differences between men and women among white CARDIA participants in Model 3. In STE parameters, age, diastolic blood pressure, and antihypertensive medications were significant determinants of attenuation of the differences in Ell_SRe between African American men and white men in Model 2. Additionally, diastolic blood pressure and number of cigarettes per day at Year‐0 were significant factors in the attenuation of differences in Ell_SRe between African American men and white men. Ecc_SRe was not different between African American and white women in univariate analyses. However, BMI and heart rate at Year‐25 attenuated the differences in Ecc_SRe between African American men and white men after adjustment for cardiovascular risk factors at Year‐25. The differences remained after adjusting for Year‐0 risk factors. The association of BMI at Year‐25 as risk factor to LV structure and function was of a higher magnitude than BMI at Year‐0 (Table 5).

Table 5.

Relationship of BMI to LV Structural and Functional Parameters at the Year‐25 Examination

	n	R 2	Year‐25 BMI	Year‐0 BMI
			β (SE)	β (SE)
Structural indices
LVEDV/height, mL/m
Model 1	3078	0.23	0.68 (0.04)*
Model 2	3078	0.25	0.73 (0.04)*
Model 3	3078	0.27	0.52 (0.06)*	0.42 (0.08)*
LVM/height, g/m
Model 1	2980	0.33	1.89 (0.07)*
Model 2	2980	0.37	1.76 (0.07)*
Model 3	2980	0.39	1.27 (0.09)*	1.04 (0.13)*
Systolic indices
LVEF, %
Model 1	3077	0.03	−0.05 (0.02)*
Model 2	3077	0.04	−0.06 (0.02)*
Model 3	3077	0.04	−0.01 (0.03)	−0.10 (0.04)
Ell, %
Model 1	2908	0.10	0.06 (0.01)*
Model 2	2908	0.15	0.04 (0.01)*
Model 3	2908	0.16	0.04 (0.01)*	−0.01 (0.01)
Diastolic indices
E/e' ratio
Unadjusted	3239	0.05
Model 1	3239	0.10	0.06 (0.01)*
Model 2	3239	0.17	0.05 (0.01)*
Model 3	3239	0.18	0.05 (0.01)*	−0.003 (0.01)
Ell_SRe, sec−1
Unadjusted	2896	0.06
Model 1	2896	0.09	−0.01 (0.001)*
Model 2	2896	0.12	−0.001 (0.001)*
Model 3	2896	0.12	−0.0003 (0.001)	−0.003 (0.001)

Model 1 is adjusted for race–ethnicity and sex group, age, and BMI at Year‐25. Model 2 is further adjusted for Model 1+Year‐25 risk factors (educational level, physical activity, heart rate, presence of diabetes, systolic blood pressure, diastolic blood pressure, use of antihypertensive medications, and number of cigarettes/day). Model 3 is further adjusted for Model 2+Year‐0 risk factors (BMI, educational level, physical activity, heart rate, presence of diabetes, systolic blood pressure, diastolic blood pressure, use of antihypertensive medications, and number of cigarettes/day). BMI indicates body mass index; E/e', ratio of mitral early peak velocity/mitral annulus early peak velocity; Ell, longitudinal peak systolic strain; Ell_SRe, longitudinal peak early diastolic strain rate; LV, left ventricular; LVEDV, LV end‐diastolic volume; LVEF, left ventricular ejection fraction; LVM, LV mass.

P<0.05.

Discussion

This study reports that differences in cardiac remodeling and myocardial function between women and men, as well as between African American and white individuals, are present by midlife and are partially derived from known cardiovascular risk factors and socioeconomic variables.

Race–Ethnicity and Sex Differences in LV Size and Structure

African American men have the highest values of LVM among the 4 race/sex groups in our study, after adjustment for demographics and cardiovascular risk factors at Year‐25. Among middle‐aged and elderly populations, race–ethnicity impacts cardiac remodeling after adjustment for risk factors, with African American men having higher LVM than white individuals.[17-20] In the MESA study, LVM and LV volume were greater in men than in women.[21-23] In prospective analyses, variations of LV structure after adjusting for Year‐0 risk factors yielded similar results to those found in the Year‐25 cross‐sectional relationships. In Year‐0, the magnitude of risk factors in African American men was significantly greater compared with white women. Sex differences in LV structure were in agreement with previous studies.[13,18,20,22,24] Both CARDIA and the Bogalusa Heart Study investigated the relation of LVM to risk factors in younger individuals using 5‐year follow‐up data in a biracial cohort.[25-26] These reports indicated that high BMI and blood pressure have a marked impact on increasing LVM over 5 years in childhood and during young adulthood. More recently, Gidding et al reported that higher blood pressure, worsening obesity, tobacco use, and long‐standing diabetes are associated with significant worsening of LV remodeling in the CARDIA cohort.[27] This study suggests that exposure to cardiovascular risk factors attenuates but does not fully explain differences in LV structure by gender and race–ethnicity.

This study shows greater differences among the 4 race/gender groups in myocardial strain than LVEF. Men have lower LVEF than women after adjusting for risk factors and other potential confounders. However, exposure to cardiovascular risk factors (BMI, heart rate, and age) attenuates the race–ethnic differences in LVEF, but not between women and men. Sex differences found in our study were consistent with previous evaluations.[13,23] There were significant differences in Ell among the 4 race/sex groups, even after adjusting for risk factors and potential confounders. African American men had the lowest level for longitudinal and circumferential strains. Conversely, white women had the largest magnitude of myocardial systolic strain. In the MESA study, using tagged magnetic resonance imaging, African American groups had the lowest Ecc compared to 3 other racial/ethnic groups, consistent with our findings.[11] The proportion of the variance explained by cardiovascular risk factors at both Year‐0 and Year‐25 in Ell (R2=0.16) was higher than in LVEF (R2=0.04). Our findings suggest that LV systolic parameters are affected by still unknown cardiovascular risk factors potentially including genetic and environmental factors. In addition, LVEF is a poor indication of early LV systolic dysfunction or cardiovascular risk compared to STE parameters.[28]

In conventional Doppler and novel STE diastolic measures, African American men had lower LV diastolic function compared to African American women and white men and women. The contribution of cardiovascular risk factors was higher in African American men and women than in white men and women at the Year‐25 examination. LV diastolic dysfunction may reflect chronic exposure to cardiovascular risk factors such as diabetes, hypertension, and increased BMI.[29] However, racial differences in diastolic dysfunction are not entirely explained by cardiovascular risk factors. LV diastolic as well as systolic function may be affected by genetic as well as environmental factors over the human life course.[30-31] The MESA study, which included participants aged 45 to 84 years, showed that racial/ethnic differences in myocardial dysfunction remained significant, adjusting for cardiovascular risk factors and socioeconomic variables.[11] In an older population (Cardiac Abnormalities and Brain Lesions study), no race–ethnic differences in LV diastolic function were observed.[32] This could have resulted from progression of atherosclerosis and adverse LV remodeling among all race–ethnic and sex groups or due to attrition among the group of African American male Cardiac Abnormalities and Brain Lesions study participants.

Clinical Implications

A previous CARDIA study showed that African American ethnicity had a strong relationship with subsequent development of HF over a 20‐year follow‐up period.[5] Furthermore, the MESA study also reported that the risk of developing HF was greater among African Americans compared with whites.[4] LV volume enlargement, higher LVM, lower LVEF, and worse conventional LV diastolic dysfunction were associated with the development of HF. Global systolic and diastolic strain parameters have recently been reported as predictors of incident HF.[7,33] Our findings suggest that African American men have echocardiographic abnormalities that may represent antecedents of subsequent HF.[5] Risk‐factor control to forestall LV functional deterioration should begin at an early age with a particular emphasis on achieving these factor controls among African American men.

Study Limitations

Observed race/sex differences in LV structure and function may reflect exposure to genetic and environmental factors that have yet to be identified. Additionally, our study suggests that both types of determinants may be implicated in different cardiac remodeling pathways starting early in life, as suggested in other studies.

Conclusions

In this biracial community‐based population of middle‐aged adults, our findings indicate race/sex differences in LV structure and function. African American men have greater LV mass and reduced LV systolic and diastolic function among the 4 race/sex groups analyzed in this study before and after adjustment for cardiovascular risk factors. These racial‐ethnic and sex differences are only partially explained by age, BMI, heart rate, blood pressure, and diabetes.

Acknowledgments

The Coronary Artery Risk Development in Young Adults Study (CARDIA) is supported by contracts HHSN268201300025C, HHSN268201300026C, HHSN268201300027C, HHSN268201300028C, HHSN268201300029C, and HHSN268200900041C from the National Heart, Lung, and Blood Institute (NHLBI), the Intramural Research Program of the National Institute on Aging (NIA), and an intra‐agency agreement between NIA and NHLBI (AG0005).