Author + information
- Received October 11, 2016
- Revision received December 26, 2016
- Accepted December 28, 2016
- Published online February 27, 2017.
- Vivek G. Patel, BSEa,b,∗ (, )
- Deepak K. Gupta, MD, MScia,b,
- James G. Terry, MSa,b,
- Edmond K. Kabagambe, PhDa,b,c,
- Thomas J. Wang, MDa,b,
- Aldolfo Correa, MDd,
- Michael Griswold, PhDd,
- Herman Taylor, MDd,e and
- John Jeffrey Carr, MDa,b
- aVanderbilt University School of Medicine, Nashville, Tennessee
- bVanderbilt Translational and Clinical Cardiovascular Research Center, Nashville, Tennessee
- cJackson Heart Study Vanguard Center at Vanderbilt University, Nashville, Tennessee
- dUniversity of Mississippi Medical Center, Jackson, Mississippi
- eMorehouse School of Medicine, Atlanta, Georgia
- ↵∗Address for correspondence:
Dr. Vivek G. Patel, Vanderbilt University School of Medicine, 3200 Long Boulevard, Unit 1, Nashville, Tennessee 37232.
Objectives This study sought to assess whether body mass index (BMI) was associated with subclinical left ventricular (LV) systolic dysfunction in African-American individuals.
Background Higher BMI is a risk factor for cardiovascular disease, including heart failure. Obesity disproportionately affects African Americans; however, the association between higher BMI and LV function in African Americans is not well understood.
Methods Peak systolic circumferential strain (ECC) was measured by tagged cardiac magnetic resonance in 1,652 adult African-American participants of the Jackson Heart Study between 2008 and 2012. We evaluated the association between BMI and ECC in multivariate linear regression and restricted cubic spline analyses adjusted for prevalent cardiovascular disease, conventional cardiovascular risk factors, LV mass, and ejection fraction. In exploratory analyses, we also examined whether inflammation, insulin resistance, or volume of visceral adipose tissue altered the association between BMI and ECC.
Results The proportions of female, nonsmokers, diabetic, and hypertensive participants rose with increase in BMI. In multivariate-adjusted models, higher BMI was associated with worse ECC (β = 0.052; 95% confidence interval: 0.028 to 0.075), even in the setting of preserved LV ejection fraction. Higher BMI was also associated with worse ECC when accounting for markers of inflammation (C-reactive protein, E-selection, and P-selectin), insulin resistance, and volume of visceral adipose tissue.
Conclusions Higher BMI is significantly associated with subclinical LV dysfunction in African Americans, even in the setting of preserved LV ejection fraction.
Obesity disproportionately affects African-American individuals, with approximately 50% having a body mass index (BMI) >30 kg/m2 (1). BMI is an established risk factor for cardiovascular disease (CVD), including heart failure (2). Obesity is associated with cardiac remodeling and hemodynamic changes collectively referred to as obesity cardiomyopathy (3–6). Although BMI was previously characterized in European Americans, whether higher BMI is associated with left ventricular (LV) systolic dysfunction in African Americans is not well understood (7).
Advances in noninvasive cardiac imaging enable quantification of myocardial mechanics, which can be used to characterize subtle changes in cardiac motion. Evaluation of LV function can be performed by measuring myocardial strain (i.e., deformation). Strain has been demonstrated to be a more sensitive method than left ventricular ejection fraction (LVEF) for the detection of subclinical cardiac dysfunction (8). Evidence from small echocardiographic studies suggest that obesity may be associated with subclinical LV systolic dysfunction, as measured by strain imaging, even in the setting of preserved or increased LVEF (8–10). Tagged cardiac magnetic resonance (CMR) is an alternative to echocardiography for the measurement of LV myocardial mechanics that may be particularly informative in understanding the relationship between BMI and cardiac structure function, as it is less susceptible to poor imaging windows in obese individuals than transthoracic echocardiography (11).
Despite the high prevalence of obesity in African-American individuals, the relationship between BMI and myocardial mechanics remains understudied in this population. The JHS (Jackson Heart Study) cohort of African-American adults who participated in the CMR examination (2008 to 2012) provided a unique opportunity to test the hypothesis that higher BMI was associated with subclinical LV systolic dysfunction, as measured by global circumferential strain (ECC).
The JHS is a prospective observational cohort study designed to study the causes of CVD in African Americans. Study design and protocols have been previously described (12). Briefly, between 2000 and 2004, adult African-American men and women, 35 to 84 years of age (n = 5,301), were recruited from metropolitan Jackson, Mississippi, to participate in the baseline examination. The JHS cardiac magnetic resonance (CMR) component began in 2008 at the end of visit 2 and continued through visit 3, ending in December 2012. Exclusion from the CMR component included pregnancy, contraindication to CMR (implanted electrical devices, pacemaker, history of metal around orbit, and so forth), or claustrophobia or were unable to fit in the MRI machine. A total of 1,672 JHS participants completed CMR. Participants in whom Eularian circumferential systolic strain (ECC) was missing (n = 20) were excluded, resulting in a final sample of 1,652 participants (n = 253 from visit 2 and n = 1399 from visit 3). There were no other exclusion criteria. The study protocol was approved by the Institutional Review Boards of Jackson State University, Tougaloo College, and University of Mississippi Medical Center, and all study participants provided written informed consent.
Cardiac magnetic resonance was performed using a large-bore (70-cm), high-gradient, 1.5-T magnet CMR machine (Espree with TIM cardiac software, Siemens Medical Solutions USA, Inc., Malvern, Pennsylvania), using a multichannel matrix surface coil to obtain short- and long-axis electrocardiogram-gated steady-state free precession CINE images of the heart, using a standardized protocol for measurement of cardiac structure and function developed in conjunction with the CMR examination, performed as part the National Heart, Lung, and Blood Institute MESA (Multi-Ethnic Study of Atherosclerosis) trial CMR protocol at examinations 2 and 3. Protocols were standardized for potential future comparison studies with JHS and MESA. Cardiac magnetic resonance data for the present analysis was not shared with MESA, and the strain data from the JHS cohort have not been previously published. Sequences for functional assessment were obtained during a short breath-hold, using white blood sequences, fast imaging with steady procession (TrueFISP or TRUFI; sequence variant Tfi2d1_18; Siemens) with the following parameters: field of view: 400 mm; slice thickness: 8 mm; matrix: 109 × 192 mm; repetition time: 45.5 ms; echo time: 1.1 ms; flip angle: 78° to 82°. Additional single breath-hold, electrocardiogram-gated, short-axis sequences located at the cardiac base and mid and apex were obtained using a CINE radiofrequency grid-tagging sequence with the following parameters: field of view: 400 mm; slice thickness: 8 mm; 192 × 256 mm matrix; repetition time: 60 s; echo time: 4 s; flip angle: 12° (sequence: Tl2d1r5; Siemens). Radiofrequency pulses were used to “tag” the myocardium during CMR.
Global circumferential strain and LV structure and function
Cardiac magnetic resonance scans were centrally analyzed at Wake Forest University School of Medicine (Winston-Salem, North Carolina). Cardiac Image Modeller software (Auckland UniServices Ltd., Auckland, New Zealand) was used to measure LV mass, end-diastolic volume (EDV), and end-systolic volume (ESV), from which stroke volume (SV), LVEF, and LV remodeling index [LVRI = mass/EDV] were calculated.
Harmonic phase software (Diagnasoft, Research Triangle Park, Morrisville, North Carolina) was used to measure ECC strain. Tagged tissues in the midmyocardium were tracked through systole as an assessment of ECC (11). Harmonic phase software uses the k-space data and specifically the phase information to create a binary image that facilitates semiautomated detection of the tag lines throughout the cardiac cycle and has been previously used to determine subclinical LV dysfunction (13–15). Global peak ECC was determined to be the average of midwall strain in the base, mid, and apical LV regions from short-axis images.
Regional myocardial circumferential strain was defined by the Lagrangian formula: [e = (L − Lo)/Lo], where e is strain, Lo is the baseline length, and L is the instantaneous length at the time of measurement. Consequently, negative strain occurs if the measured segment length is shorter than its original length in the circumferential direction. Therefore, global circumferential systolic strain values closer to zero (less negative) are indicative of worse LV function.
Reproducibility of CMR measures using harmonic phase software was assessed on 96 scans blind duplicate examinations that had their DICOM headers relabeled and were inserted into the workflow as newly received exams. Interclass correlation coefficients for global ECC analysis was 0.78. Interclass correlation coefficients for EDV, ESV, and LV mass, were 0.95, 0.88, and 0.96, respectively.
BMI and covariates
Body mass index and covariates were ascertained from the closest visit preceding the date of acquisition of CMR data. Body mass index was defined as kg/m2. Fasting blood samples were collected according to standardized procedures, and assessments of glucose, lipids, insulin, creatinine, C-reactive protein (CRP), P-selectin, and E-selectin were processed at a central laboratory, as previously described (12,16). Blood pressure was taken at the baseline examination as the average of 2 measurements recorded at 5-min intervals while the patients was seated.
Hypertension was defined by a self-reported diagnosis of hypertension, blood pressure >140/90 mm Hg, or treatment with antihypertensive medications. Diabetes mellitus was defined by fasting plasma glucose concentration ≥126 mg/dl, an HbA1c level of ≥6.5%, or treatment with insulin or oral hypoglycemic agent. Smoking status was determined by self-report and categorized as ever or never. Cardiovascular disease status was considered present if the participant reported a history of myocardial infarction; had an abnormal stress test result or had undergone coronary artery bypass graft surgery, or coronary angioplasty; or had a history of stroke at any visit. Estimated glomerular filtration rate (eGFR) at examination 3 (n = 1,389) was calculated from serum creatinine level, using the chronic kidney disease epidemiology collaboration equation (17). Chronic kidney disease was defined as an eGFR concentration of ≤60 ml/min/1.73 m2. Plasma E-selectin and P-selectin levels were quantified by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, Minnesota). The interassay coefficients of variation for the E-selectin and P-selectin detection methods were 9.78% and 5.14%, respectively (18). C-reactive protein level was determined using the immunoturbidimetric CRP Latex assay (Kamiya Biomedical Co., Tukwila, Washington) following the manufacturer’s high-sensitivity protocol. The interassay coefficients of variation for control samples repeated in each assay were 4.5% and 4.4% at CRP concentrations of 0.45 mg/l and 1.56 mg/l, respectively. The reliability coefficient for masked quality control replicates was 0.95 for the CRP assay (19).
Visceral adipose tissue (VAT) was measured at visit 2 (n = 263), using multidetector computed tomography imaging (Lightspeed 16 Pro, GE Healthcare, Milwaukee, Wisconsin) as previously described (20). Briefly, 24 continuous 2.5-mm-thick image slices were acquired, covering 60 mm above the level of S1 (vertebral level). The volumetric quantification of VAT was performed using volume analysis software (Advantage Windows [Microsoft, Redmond, WA], GE Healthcare), and the sum of VAT pixels over 24 slices were calculated as the volumes of VAT.
Descriptive statistics of demographic, clinical, and CMR variables were stratified by BMI according to categories defined by World Health Organization, as follows: lean BMI: <25 kg/m2; overweight: 25 to <30 kg/m2; class I obese: 30 to <35 kg/m2; and class II or III obese: ≥35 kg/m2. Results are mean ± SD or percentages, as appropriate. One-way ANOVA was performed for continuous variables and chi-squared test for categorical variables. Correlations between BMI and ECC were examined using Spearman rank correlation coefficients.
Sequential multivariate-adjusted linear regression models were used to examine the association between BMI on a continuous scale and ECC. Both BMI and ECC were normally distributed and did not require transformation. Covariates were chosen a priori based on previous reports of factors associated with myocardial mechanics (6,14,21) and included age, sex, diabetes mellitus status, smoking, hypertension, low-density lipoprotein cholesterol (LDL-C), triglycerides, heart rate, eGFR, LVEF, and LV mass. Interaction terms were included in the multivariate-adjusted models to assess whether the association between BMI and ECC differed in pre-specified subgroups of age, sex, BMI, hypertension, diabetes, CVD history, and chronic kidney disease. Multivariate-adjusted restricted cubic spline analysis with 4 knots was performed to graphically display and evaluate for nonlinear associations between BMI on a continuous scale with ECC or LVEF. The number of knots was selected based upon the model that produced the lowest Akaike information criterion.
In exploratory analyses, potential mechanisms for the association between BMI and ECC were examined with additional models adjusted for inflammatory markers (E-selectin, P-selectin, and CRP); insulin resistance, measured by the homeostasis model of insulin resistance (HOMA-IR) among nondiabetics (n = 915); and VAT (n = 263) as the primary measurement of fat depot.
A sensitivity analysis was performed to account for the heterogeneity in time difference between CMR and the clinical visit based on the following subgroups: <1 year (n = 1,076), <6 months (n = 1,011), and <1 month (n = 597). All statistical analyses were conducted using STATA version 12.0 software (STATA Statistical Software, College Station, Texas). A 2-tailed level of significance was set at a p level of <0.05.
Baseline characteristics and cardiac structure and function
Obesity was common, present in 55% of participants (Table 1). Age was similar across the spectrum of BMI. The proportion of females was higher, whereas the frequency of smoking was lower with higher BMI. Increased BMI was also associated with higher prevalence of diabetes and hypertension and with higher heart rate and blood pressure measurements. History of CVD and lipid levels were similar across the range of BMI. Unadjusted, LV volume, mass, LVEF, and LVRI were higher with increased BMI (Table 2).
Association between BMI and global circumferential systolic strain: Multivariate analyses
Higher BMI was associated with reduced LV systolic function as shown by less negative ECC values, unadjusted (Table 2) and remained significant with adjustment for renal function, LVEF, and LV mass (Table 3). The association of higher BMI and LV systolic dysfunction was consistent between visits and across multiple clinically relevant subgroups, including age, sex, and conventional cardiovascular risk factors (Figure 1).
The relationships between BMI (range: 15 to 69 kg/m2) and ECC or LVEF were examined using multivariate-adjusted restricted cubic spline analysis (Figure 2). Higher BMI was related to higher LVEF (β = 0.19; 95% confidence interval [CI]: 0.094 to 0.29; p < 0.0001). In contrast, across the range of BMI, greater BMI was associated with worsening of LV systolic function (ECC), even in the setting of preserved and increasing LVEF.
Potential mechanisms for subclinical LV dysfunction
In exploratory analyses, multivariate-adjusted linear regression models including available inflammatory biomarkers, computed tomography-measured VAT, and HOMA-IR did not significantly attenuate the association between BMI and subclinical LV dysfunction (Table 4).
In this large, community-based cohort of middle-aged African-American individuals, we found that higher BMI across a range of BMI from 15 to 69 kg/m2 was significantly associated with subclinical LV systolic dysfunction as measured by CMR midwall circumferential strain. The relationship between BMI and LV dysfunction was independent of prevalent CVD, conventional cardiometabolic risk factors, and LVEF and was consistent in subgroup analyses. The significant association between BMI and ECC persisted even after adjustment for inflammatory markers, insulin resistance, and abdominal visceral fat.
Several studies have reported that obesity is independently associated with alterations in cardiac structure and mechanics; however, those studies were conducted in predominantly white populations, using echocardiography and did not include individuals with BMI >40 kg/m2 (8–10). Transthoracic echocardiography can be particularly challenging in obese patients due to acoustic window constraints from excess chest wall fat. Cardiac magnetic resonance, like echocardiography, is a noninvasive procedure, but it overcomes several of the limitations of echocardiography related to obesity (22,23). Our findings of impaired LV systolic function with higher BMI are consistent with those of several previous studies examining the relationship between obesity subgroups (classes I, II, and III) and LV strain patterns (7,10,24). However, there have been conflicting reports of alterations in myocardial mechanics for individuals with milder degrees of obesity (9,25–27). Our study suggests that the association of BMI and subclinical LV dysfunction is nearly linear across the BMI range of 27 to 50 kg/m2 in African Americans. Possible explanations for the differences between our study and prior studies may be related to less sensitive techniques used to detect preclinical measurements of LV dysfunction, lack of power in past study designs, categorization of obesity subgroups as opposed to analyzing BMI on a continuous spectrum, and predominantly Caucasian populations.
Our findings further suggest that the association between BMI and ECC was not explained by confounding from conditions predisposing to myocardial impairment. The association of higher BMI with LV dysfunction was consistent across multiple clinically relevant subgroups including age, sex, hypertension, diabetes, CVD, and chronic kidney disease. Previous reports showed that comorbid conditions with obesity including hypertension and diabetes in asymptomatic patients are characterized by improved circumferential strain, which has been proposed as a compensatory mechanism to preserve LVEF (21,28–30). However, our findings suggest that higher BMI is associated with impaired circumferential strain independent of these comorbidities. This discrepancy suggests that obesity may contribute to subclinical LV dysfunction through a variety of mechanisms not limited to changes in loading conditions, insulin resistance, and coronary artery disease.
Although LVEF is the most commonly used measurement of systolic function in clinical practice, it is highly load-dependent and insensitive to subtle changes in cardiac function (31,32). Across the entire clinical range of BMIs measured in our study population, LVEF was largely within the normal range (55% to 65%) and positively related to BMI after multivariate adjustment. As demonstrated in Figure 2, above a BMI of 35 kg/m2, the relationship between BMI and LVEF is effectively flat and in the normal range. In contrast, ECC worsens in a nearly linear relationship across the entire spectrum of BMI despite a normal range of LVEF, supporting the concept that higher BMI is associated with altered myocardial mechanics.
Proposed mechanisms by which higher BMI may associate with adverse cardiac remodeling and dysfunction include inflammation, insulin resistance, and myocardial metabolic changes. Systemic release of inflammatory mediators from excess adipose tissue results in myocardial injury and subsequent cardiac structural alterations (2,33–35). Phosphorus-31 (31P) CMR spectroscopy has shown a myocardial energy deficit exists in obesity, likely resulting in impaired systolic function (36). Some studies suggest insulin resistance leads to these alterations in myocardial substrate metabolism, which contributes to the LV dysfunction associated with obesity (24,37,38). These metabolic derangements are characterized by increased myocardial fatty acid metabolism resulting in myocardial injury via lipotoxicity and concurrently decreased glucose metabolism (39–41). In addition, BMI may be a surrogate for the distribution of body fat and muscle, with some studies implicating visceral adiposity as a greater cardiovascular risk factor than BMI (42). Visceral adiposity may better represent both myocardial triglyceride content and LV mass, which was shown to partially attenuate the association of BMI and ECC in our models. We examined whether circulating markers of inflammation, insulin resistance, or VAT mitigated the association between BMI and worsened ECC. Although insulin resistance and visceral adiposity attenuated the magnitude of the relationship, these factors only partially accounted for the relationship between BMI and ECC, suggesting alternate unmeasured factors may also be involved.
We examined a large, well-phenotyped adult African-American population 35 to 84 years of age and spanning a BMI range of 15 to 69 kg/m2 to quantify the relationship between BMI and subclinical LV systolic dysfunction, using tagged CMR. To our knowledge, our study is the first to examine the association of BMI on a continuous scale and subclinical LV dysfunction using CMR in African Americans. Our study is unique in that we were able use the large-bore Espree CMR machine (Siemens) to include JHS participants with BMI >50 kg/m2, which was not possible in previous studies such as the MESA trial.
We could not determine causal relationships between BMI and LV dysfunction given the cross-sectional study design. Measurements of longitudinal systolic strain, which may be a more sensitive measure than ECC for the detection of changes in myocardial deformation, were not available in the JHS population but are a future direction. Cardiac magnetic resonance studies were not obtained on the same day that BMI and clinical covariates were ascertained. Differing myocardial preload, afterload, and contractile state values at the time of CMR examination could affect ECC and confound the analysis. We accounted for preload and afterload with heart rate and LVRI, which were taken at the time of CMR. Blood pressure was not measured concurrently with the CMR examinations; however, controlling for another measurement of afterload, LVRI, the relationship between BMI and ECC was still significant. In addition, adjusting for the time differences between imaging and covariate measurements did not change the inference. Body mass index may not account for distributions of body fat or other proxy measurements of obesity, such as waist circumference or waist-to-hip ratio, which might have provided different information. However, waist circumference and and waist-to hip ratios have greater measurement error and are less commonly used in routine clinical practice than BMI; therefore, they were not the focus of this analysis. Despite accounting for inflammation, insulin resistance, and VAT, the precise mechanisms underlying the association between BMI and subclinical LV dysfunction remain unknown.
In a community-based cohort of African-American individuals, higher BMI was significantly associated with subclinical LV dysfunction independent of conventional cardiometabolic risk factors, renal function, cardiac structural characteristics, and LVEF. Body mass index may reflect not only obesity but also pathologic processes that result in impaired myocardial mechanics. Better understanding of the pathophysiology of how higher BMI relates to cardiomyopathy should inform strategies to prevent and potentially reverse detrimental subclinical LV alterations.
COMPETENCY IN MEDICAL KNOWLEDGE: Obesity disproportionately affects the African-American population and is associated with adverse cardiac remodeling and hemodynamic changes. Body mass index is an established risk factor for CVD and heart failure, but the relationship between BMI and myocardial mechanics remains understudied in this population. We found that higher BMI is significantly associated with subclinical LV systolic dysfunction in African-American individuals.
TRANSLATIONAL OUTLOOK: Future studies to clarify the pathophysiology of how higher BMI relates to cardiomyopathy are warranted to inform strategies to prevent and potentially reverse detrimental subclinical LV alterations.
The authors thank the participants and data collection staff of the Jackson Heart Study.
The Jackson Heart Study was supported by contracts HHSN268201300046C, HHSN268201300047C, HHSN268201300048C, HHSN268201300049C, and HHSN268201300050C from the National Heart, Lung, and Blood Institute and the National Institute on Minority Health and Health Disparities. Research reported in this paper was supported by National Institutes of Health grants K12 HL109019, K23 HL128928-01A1, and R01-HL-102780. The views expressed in this manuscript are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; the National Institutes of Health; or the U.S. Department of Health and Human Services. Dr. Taylor is a consultant. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- body mass index
- cardiac magnetic resonance
- global circumferential strain
- homeostasis model of insulin resistance
- left ventricular
- left ventricular ejection fraction
- left ventricular remodeling index
- Received October 11, 2016.
- Revision received December 26, 2016.
- Accepted December 28, 2016.
- 2017 American College of Cardiology Foundation
- Mozaffarian D.,
- Benjamin E.J.,
- Go A.S.,
- et al.
- Ndumele C.E.,
- Coresh J.,
- Lazo M.,
- et al.
- Russo C.,
- Jin Z.,
- Homma S.,
- et al.
- Ballo P.,
- Motto A.,
- Mondillo S.,
- Faraguti S.A.
- Wong C.Y.,
- O'Moore-Sullivan T.,
- Leano R.,
- Byrne N.,
- Beller E.,
- Marwick T.H.
- Buckberg G.,
- Hoffman J.I.,
- Mahajan A.,
- Saleh S.,
- Coghlan C.
- Pascual M.,
- Pascual D.A.,
- Soria F.,
- et al.
- Fernandes V.R.,
- Polak J.F.,
- Edvardsen T.,
- et al.
- Kishi S.,
- Armstrong A.C.,
- Gidding S.S.,
- et al.
- Edvardsen T.,
- Detrano R.,
- Rosen B.D.,
- et al.
- Fuqua S.R.,
- Wyatt S.B.,
- Andrew M.E.,
- et al.
- Penman A.,
- Hoadley S.,
- Wilson J.G.,
- Taylor H.A.,
- Chen C.J.,
- Sobrin L.
- Jiang K.,
- Yu X.
- Yoneyama K.,
- Gjesdal O.,
- Choi E.Y.,
- et al.
- Peterson L.R.,
- Herrero P.,
- Schechtman K.B.,
- et al.
- Mizuguchi Y.,
- Oishi Y.,
- Miyoshi H.,
- Iuchi A.,
- Nagase N.,
- Oki T.
- Aurigemma G.P.,
- Zile M.R.,
- Gaasch W.H.
- Carabello B.A.
- Cherian S.,
- Lopaschuk G.D.,
- Carvalho E.
- Rider O.J.,
- Francis J.M.,
- Ali M.K.,
- et al.
- Park S.Y.,
- Cho Y.R.,
- Kim H.J.,
- et al.
- Kim G.,
- Jo K.,
- Kim K.J.,
- et al.