Author + information
- Received September 15, 2016
- Revision received October 11, 2016
- Accepted October 14, 2016
- Published online December 26, 2016.
- Yashashwi Pokharel, MD, MSCRa,b,c,
- Wensheng Sun, MPHa,
- Salim S. Virani, MD, PhDa,b,d,
- Vijay Nambi, MD, PhDa,b,d,
- Ron C. Hoogeveen, PhDa,b,
- Patricia P. Chang, MD, MHSe,
- Chiadi E. Ndumele, MD, MHSf,
- Scott D. Solomon, MDg,
- Biykem Bozkurt, MD, PhDa,d,
- Elizabeth Selvin, PhDf,
- Christie M. Ballantyne, MDa,b and
- Anita Deswal, MD, MPHa,d,∗ ()
- aSection of Cardiology, Baylor College of Medicine, Houston, Texas
- bCenter for Cardiovascular Disease Prevention, Houston Methodist DeBakey Heart and Vascular Center, Houston, Texas
- cSaint Luke’s Mid-America Heart Institute, University of Missouri-Kansas City, Kansas City, Missouri
- dSection of Cardiology, Michael E. DeBakey Veterans Affairs Medical Center, Houston, Texas
- eDivision of Cardiology, University of North Carolina, Chapel Hill, North Carolina
- fJohns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland
- gDepartment of Medicine, Harvard Medical School, Boston, Massachusetts
- ↵∗Reprint requests and correspondence:
Dr. Anita Deswal, Michael E. DeBakey VA Medical Center and Baylor College of Medicine, VAMC (111B), 2002 Holcombe Bouldevard, Houston, Texas 77030.
Objectives This study sought to determine whether pre-heart failure (HF) myocardial injury explains the differential mortality after HF across weight categories.
Background Obesity is a risk factor for HF, but pre-HF obesity is associated with lower mortality after incident HF. High-sensitivity cardiac troponin T (hs-cTnT) is a sensitive marker of myocardial injury, and predicts incident HF and mortality.
Methods Stratifying 1,279 individuals with incident HF hospitalizations by their pre-HF hs-cTnT levels (< and ≥ 14 ng/l), we examined the association of pre-HF body mass index (BMI) with mortality after incident HF hospitalization in the ARIC (Atherosclerosis Risk In Communities) study.
Results Mean age at HF was 74 years (53% women, 27% black). Individuals with pre-HF hs-cTnT ≥14 ng/l had higher mortality after incident HF (hazard ratio [HR]: 1.46; 95% confidence interval [CI]: 1.18 to 1.80) compared to individuals with hs-cTnT <14 ng/l in an adjusted model including BMI. Compared with normal weight subjects, the mortality was lower in overweight (HR: 0.69, 95% CI 0.48-0.98) and obese individuals (HR: 0.50; 95% CI: 0.35 to 0.72) with hs-cTnT <14 ng/l; and in those with hs-cTnT ≥14 ng/l (overweight HR: 0.50; 95% CI: 0.30 to 0.83; obese HR: 0.56; 95% CI: 0.34 to 0.91; interaction: p = 0.154 between BMI and hs-cTnT). The lower mortality risk in obese and overweight subjects remained similar when log hs-cTnT was added as a continuous variable to a multivariable model, and in sensitivity analyses after further adjusting for left ventricular hypertrophy or high-sensitivity C-reactive protein.
Conclusion Although greater pre-existing subclinical myocardial injury was associated with higher mortality after incident HF hospitalization, it did not explain the obesity paradox in HF, which was observed irrespective of subclinical myocardial injury. (Atherosclerosis Risk In Communities [ARIC]; NCT00005131)
Obesity and being overweight affect approximately two-thirds of the U.S. population (1). Obesity is associated with an increased risk of heart failure (HF) (2,3). Paradoxically, once HF develops, obesity is associated with better prognosis (4–7). Using the ARIC (Atherosclerosis Risk In Communities) study cohort, we previously showed that individuals who were overweight and obese (using pre-HF body mass index [BMI]) before the onset of incident HF hospitalization also had lower mortality after HF development than those with normal BMI, independent of demographics and comorbidities, including cancer, smoking, and diabetes (8). Our findings suggested that weight loss from advanced HF did not completely explain the survival benefit associated with higher BMI in HF patients. One potential explanation for this obesity paradox is that the HF patients who are able to preserve their weight may represent a noncatabolic subgroup with different neurohormonal, inflammatory, and metabolic profiles (9). However, there may also be heterogeneity in the degree of myocardial injury in obese individuals who present with the constellation of symptoms and signs such as shortness of breath and fluid retention, leading to a diagnosis of HF.
Cardiac troponin T, measured using a high-sensitivity assay (hs-cTnT), is a sensitive marker of subclinical myocardial injury (10). We have shown that hs-cTnT levels predicted incident HF in the ARIC study (11). Other studies of community dwelling and otherwise asymptomatic individuals have also demonstrated similar findings (12,13). Furthermore, higher hs-cTnT levels were also associated with increased cardiac and all-cause mortality (11–13) and a higher degree of myocardial dysfunction (13). We postulated that hs-cTnT measurements used before incident HF would be helpful in identifying overweight and obese individuals at various levels of risk for adverse outcomes after the development of the clinical HF syndrome. Undetectable or low levels of hs-cTnT, therefore, may identify “cardiac biomarker-healthy” obese individuals with lower degree of myocardial injury in whom symptoms of HF such as shortness of breath and edema are driven mostly by noncardiac obesity-related mechanisms and who would therefore have a good prognosis contributing to the obesity paradox. In contrast, “cardiac biomarker-unhealthy” obese, in whom HF is driven to a greater extent by myocardial injury/dysfunction as indexed by hs-cTnT measurements would be expected to have a poor prognosis. We postulated that the obesity paradox would be observed in “biomarker-healthy obese” individuals but not in the “biomarker-unhealthy obese,” as stratified by concentrations of hs-cTnT prior to incident HF. Accordingly, we investigated the risk of mortality associated with pre-HF BMI and hs-cTnT levels in individuals who subsequently had incident HF hospitalization in the ARIC study.
The ARIC study is a prospective cohort study of 15,792 individuals enrolled from the following 4 U.S. communities: Washington County, Maryland; Jackson, Mississippi; Forsyth County, North Carolina; and suburbs of Minneapolis, Minnesota (14). The baseline evaluation (visit 1) of individuals 45 to 64 years of age took place between 1987 and 1989. Participants were subsequently examined at 3 follow-up visits at approximately 3-year intervals, and after an extended interval, a fifth study visit was recently completed between 2011 and 2013. The design, recruitment, and examination protocols for the ARIC study have been described in detail previously (14). The institutional review boards at each site approved all study protocols, and informed consent was provided by all study participants. The authors are solely responsible for the design, conduct, and analyses of the study and the drafting, editing, and preparation of the final version of the manuscript. Visit 4 (1996 to 1998), at which hs-cTnT was measured in all participants, was the baseline for the current analysis. Of 11,656 visit 4 participants, 1,279 individuals who had an incident HF hospitalization at least 6 months after visit 4, were eligible for this analysis after exclusions as detailed in Figure 1.
Measurement of hs-cTnT
Details regarding the hs-cTnT measurements have been previously published (11). Briefly, hs-cTnT levels were measured in 2010, from blood samples collected during ARIC visit 4, using a high-sensitivity assay, Elecsys Troponin T (Roche Diagnostics, Basel, Switzerland) with an automated Cobas e411 analyzer (Roche). The 99th percentile value for the hs-cTnT measured in a healthy reference population (20 to 70 years of age) was 14 ng/l according to the manufacturer (Roche).
Measurements were taken in standard scrub attire and no shoes. Weight was measured using a scale that was zeroed daily and calibrated quarterly. Body mass index assessment was performed at the time of hs-cTnT measurement. Pre-HF BMI was defined as the BMI measurement from ARIC visit 4, which was at least 6 months before the date of incident HF hospitalization (8). Patients with HF were divided by pre-HF BMI into normal weight (18.5 to <25 kg/m2), overweight (25 to <30 kg/m2), and obese (≥30 kg/m2) groups. Patients in the underweight category (BMI <18.5 kg/m2; n = 83) were excluded because of small numbers and the possible presence of other comorbidities leading to lower BMI (8).
Ascertainment of demographics and comorbidities at each study visit has been described previously (14). Age was assessed at the time of incident HF hospitalization, and sex, race, and education levels were obtained at ARIC visit 1 from interviewer-administered questionnaires. Physical activity was assessed at ARIC visit 3 by using the sport index of the Baecke questionnaire (15). Baecke sport activity scores range from 1 to 5 with a higher score indicating higher physical activity (16). Comorbidities such as hypertension, diabetes, cancer, history of coronary heart disease (CHD), stroke, and transient ischemic attacks were assessed as present if the conditions were documented to be present at any pre-HF ARIC visit. All other baseline variables were obtained from visit 4. Diabetes was defined as fasting blood glucose level ≥126 mg/dl, nonfasting blood glucose level ≥200 mg/dl, or self-reported physician diagnosis of or treatment for diabetes. Total cholesterol and high-density lipoprotein cholesterol were determined by enzymatic methods. Estimated glomerular filtration rate was computed using the Modification of Diet in Renal Disease study equation (17). Blood pressure was measured twice, and the average of the 2 measurements was used. Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure BP ≥90 mm Hg measured with random-zero mercury manometers or antihypertensive medications. Details of ascertainment of CHD and stroke have been described previously (18–20).
Ascertainment of HF and all-cause mortality
ARIC investigators conducted continuous, comprehensive surveillance of all cardiovascular (CV) disease-related hospitalizations and deaths in the 4 communities. Incident HF hospitalization was determined using annual interviews with study participants regarding interim hospitalizations (response rate: 93% to 96%), review of discharge lists from local hospitals and survey of health department death certificate files and the national death index. In the ARIC cohort, incident HF was defined as the first occurrence of either a hospitalization that included an International Classification of Diseases-9th revision-Clinical Modification (ICD-9-CM) discharge diagnosis code for HF beginning with “428” (i.e., 428.0 to 428.9) in any position or a death certificate ICD-9 code beginning with “428” (HF) or ICD-10 code “I50” (HF or I50.0 to I50.9) in any position (8,21). Deaths were confirmed by review of hospital discharge records for inpatient deaths and death certificates for deaths outside the hospital.
The outcome of interest was all-cause mortality in individuals who had incident HF hospitalization. For this study, only incident HF ascertained ≥6 months after ARIC visit 4 until December 31, 2011, was included because this was the last date for which follow-up was available at the time the study was designed. The last HF hospitalization occurred on December 29, 2011. Eligible study participants with incident HF hospitalization were followed for death up to 5 years after incident HF hospitalization or until December 31, 2011. Categorical variables were compared between groups, using chi-square tests, and continuous variables were compared using the analysis of variance (ANOVA) test. To accommodate the wide range of follow-up duration, we calculated survivor function in the overall population and within BMI and hs-cTnT results categories by using survival estimates from Cox proportional hazard model. The survivor function estimates the proportion of patients surviving at any time point. We assessed the associations between pre-HF BMI categories (normal weight, overweight, and obese) and all-cause mortality after HF in a model adjusted for age, sex, race, education level, health insurance, systolic blood pressure, physical activity, total cholesterol, current smoking, current alcohol use, estimated glomerular filtration rate, diabetes mellitus, hypertension, history of CHD, stroke/transient ischemic attack, and cancer, using Cox proportional hazards model (model 1). Subsequently, we added log-transformed hs-cTnT levels as a continuous variable to the model above. We selected covariates a priori based on previous knowledge of potentially important variables (8). We also examined hazard ratios (HRs) for all-cause mortality by pre-HF BMI and hs-cTnT level, with hs-cTnT levels as 2 (< or ≥ 14 ng/l) categories. This cutoff for hs-cTnT level was chosen because hs-cTnT level equal to 14 ng/l corresponds to the 99th percentile in a healthy reference population as provided by the assay manufacturer and has been used in prior analyses (11,20,22). We examined the interaction between BMI and hs-cTnT categories and between BMI category and hs-cTnT levels as a continuous variable (log-transformed). We also assessed the HR for death in individuals with pre-HF hs-cTnT ≥14 ng/l compared to hs-cTnT <14 ng/l using model 1 plus BMI (treated as an ordinal variable, i.e., normal weight, overweight, and obese). The proportional hazards assumption was examined using time-dependent covariates and likelihood ratio statistics. When the proportionality assumption was violated, a time-dependent term, the product of BMI/hs-cTnT group and log time was added to the model (8). In addition, we performed some sensitivity analyses. We developed additional models where we further adjusted model 1 with left ventricular hypertrophy as determined by Cornell voltage electrocardiographic criteria (data available for 976 of 1,279 individuals in the study cohort) and with high-sensitivity C-reactive protein concentration (data available for 1,176 individuals). All presented tests were 2-tailed, and a p value <0.05 was considered statistically significant. All analyses were performed using SAS version 9.3 software (Cary, North Carolina) and STATA version 12 (STATA, College Station, Texas).
The final study cohort consisted of 1,279 individuals with incident HF hospitalization whose BMI and hs-cTnT levels were measured at least 6 months before incident HF. Both the pre-HF BMI and hs-cTnT levels were measured at a median of 8.3 (25th, 75th percentile: 5.0, 11.4) years before the HF diagnosis. The mean age was 74 years old at the time of incident HF hospitalization, 53% of the participants with HF were women, 27% were black, 17% had normal weight, 35% were overweight, and 48% were obese.
Compared to individuals with lower pre-HF hs-cTnT levels, those with higher hs-cTnT levels developed HF at a slightly younger age, were more often men, had more comorbidities such as hypertension, diabetes, history of myocardial infarction, CHD, left ventricular hypertrophy, and cancer; had lower levels of estimated glomerular filtration rate; total and high-density lipoprotein cholesterol but higher levels of triglycerides; high-sensitivity C-reactive protein and N-terminal pro-B-type natriuretic peptide (NT-proBNP); and had a shorter time to development of HF; however, fewer of them were smokers (Table 1). There were no differences in hemoglobin levels, heart rate, and physical activity according to hs-cTnT categories. The mean BMI was similar in both of the hs-cTnT groups. Baseline characteristics by pre-HF BMI groups are described in Online Table S1. Briefly, individuals in higher BMI categories developed HF at a relatively younger age, were more likely to be black, more often had diabetes and hypertension, were less likely to be current smokers and alcohol users, had higher levels of pre-HF high-sensitivity C-reactive protein concentration but lower levels of pre-HF NT-proBNP, but showed no differences in the time to incident HF and in the prevalence of pre-existing CHD and history of myocardial infarction.
A total of 579 of 1,279 individuals (45.3%) who developed HF died within 5 years after incident HF hospitalization (survival estimate: 0.48; 95% CI: 0.45 to 0.51). Overall, overweight and obese groups had better survival over 5 years after HF (Online Figure 1, Online Table S2) compared with normal weight group. Similarly, individuals with hs-cTnT level <14 ng/l had better survival than those with hs-cTnT level ≥14 ng/l (Online Table S2).
After incident HF hospitalization, overweight and obese individuals had better survival compared with normal-weight people in groups with either hs-cTnT <14 or ≥14 ng/l (Figure 2, Online Table S2). In individuals with hs-cTnT concentration ≥14 ng/l, the Kaplan-Meier survival curves for obese and overweight groups overlapped eventually; however, the normal weight group curve was separated throughout follow-up.
After adjusting for demographics and CV risk factors/disease, overweight and obese HF individuals had lower mortality than normal weight HF cohort members in both the group with hs-cTnT <14 and the group with hs-cTnT ≥14 ng/l (p = 0.154 for interaction between BMI and hs-cTnT) (Table 2). For example, obese individuals with hs-cTnT <14 ng/l had 50% lower relative mortality risk than those with normal weight and hs-cTnT <14 ng/l (HR: 0.50; 95% CI: 0.35 to 0.72); and obese individuals with hs-cTnT ≥14 ng/l had 44% lower relative mortality risk than those with normal weight and hs-cTnT ≥14 ng/l (HR: 0.56; 95% CI: 0.34 to 0.91). When all 6 of the possible groups resulting from BMI and hs-cTnT categories were compared with normal weight and the hs-cTnT group with <14 ng/l as the reference, the highest mortality was observed in those with normal weight and hs-cTnT levels ≥14 ng/l (HR: 1.50), and the lowest mortality rate was seen in those who were obese and had a hs-cTnT level <14 ng/l (HR: 0.50) (Figure 3). In the current data, we again observed that premorbidly overweight and obese individuals who subsequently developed HF had reduced mortality risk after HF compared with normal weight individuals after adjusting for demographics and CV risk factors/disease (Online Table S3, Model 1). Results were similar when we further adjusted the model with hs-cTnT (log-transformed) as a continuous variable (p = 0.635 for interaction between BMI and log hs-cTnT levels) (Online Table S3, Model 2) with no appreciable differences in HRs between models, which did or did not include hs-cTnT level. In additional analyses, individuals with hs-cTnT ≥14 ng/l had greater risk for death (HR: 1.46; 95% CI: 1.18 to 1.80) than individuals with hs-cTnT <14 in model 1 plus BMI categories.
Results of sensitivity analyses were similar to those of the main analyses (Online Tables S4 and S5). When the models examining risk of mortality based on BMI categories within the higher and lower troponin groups were further adjusted for either left ventricular hypertrophy on electrocardiogram (n = 976) or high-sensitivity C-reactive protein levels (n = 1,176), the survival benefit for overweight/obese compared to normal weight individuals was still observed within the 2 troponin groups.
In this study, we found that despite an increased risk for mortality with greater pre-existing subclinical myocardial injury, as assessed by hs-cTnT the mortality risk was lower in overweight or obese individuals compared with normal weight individuals, regardless of pre-existing subclinical myocardial injury. Furthermore, individuals with normal pre-HF weight and more advanced pre-HF subclinical myocardial injury (i.e., hs-cTnT ≥14 ng/l) had the highest mortality after HF development.
Association of pre-HF hs-cTnT levels and heart failure
Data from the ARIC cohort was recently used to show that higher BMI was independently associated with a linear increase in myocardial injury as assessed using hs-cTnT levels (22). Furthermore, individuals with highest hs-cTnT levels who were also severely obese had a HR of 9.2 for incident HF hospitalization compared to that of those with normal weight and undetectable hs-cTnT levels (22). Other studies have also demonstrated that obesity is independently associated with an increased risk of incident HF (2,3). Our results are in keeping with those observations by demonstrating that overweight and obese individuals with HF have higher level of subclinical myocardial injury several years before the development of HF (Online Table S1).
Association of pre-HF hs-cTnT level and mortality in heart failure
Prior studies have demonstrated that increasing myocardial injury, as indexed by hs-cTnT levels was associated with higher cardiac and all-cause mortality (11–13) and higher degree of myocardial dysfunction (13). In the current analysis, we have further demonstrated that higher myocardial injury several years before incident HF hospitalization is also associated with higher post-HF mortality. Among individuals with incident HF hospitalization, those with pre-HF hs-cTnT levels ≥14 ng/l had 46% relative increase in mortality risk compared to those with hs-cTnT <14 ng/l in a model adjusting for BMI and other baseline variables.
pre-HF myocardial injury and the obesity paradox
Once HF develops, obese individuals have lower mortality risk than normal weight individuals, (i.e., the obesity paradox) (4–8). Obesity and overweight measurements before the onset of HF (pre-morbid BMI) have also been associated with lower mortality risk than normal weight individuals (8), suggesting that weight loss/cachexia after HF does not completely explain the lower mortality risk seen in obese HF patients, as was replicated in the current study. We had hypothesized that the obesity paradox may only be observed in individuals with lesser myocardial injury before HF where noncardiac factors may contribute to the volume overload presenting as the HF syndrome in obese individuals (biomarker healthy obese); but the paradox would not be seen in a more homogenous group with higher pre-HF myocardial injury. However, our analysis found that the risk for death after HF decreased in overweight and obese individuals compared to normal weight individuals regardless of pre-existing subclinical myocardial injury in models stratified by hs-cTnT levels, as well as in models with hs-cTnT level as a continuous variable. These results remained the same in sensitivity analyses which adjusted for pre-HF biomarkers of inflammation (high-sensitivity C-reactive protein) and for ventricular structure (left ventricular hypertrophy by electrocardiogram). Taken together, our findings suggest that the obesity paradox in HF may not be explained by pre-HF “biomarker-healthy” or “biomarker-unhealthy” obesity, using the biomarkers examined in this study. Previously, higher levels of tumor necrosis factor-alpha and adiponectin and higher systolic pulmonary arterial pressure have been associated with lower BMI compared to higher BMI in HF patients (23). Future studies will need to explore whether these and other cardiac and noncardiac factors associated with obesity such as presence of greater metabolic reserve, attenuated neurohormonal activation, or other biomarkers, such as higher levels of soluble tumor necrosis factor receptors available to bind harmful circulating tumor necrosis factor, or just a lead time bias with earlier presentation of clinical HF, contribute to the obesity paradox as noted in HF as well as in other chronic conditions (24–27).
Although differential levels of pre-HF hs-cTnT levels do not explain the obesity paradox, the higher incidence of HF, and higher mortality after HF that is associated with higher levels of troponin measured several years before incident HF, suggests that this biomarker could be used to target individuals for aggressive preventive strategies to decrease the incidence of HF and mortality.
Study strengths and limitations
The findings of this study should be interpreted in the context of several strengths and limitations. The study used a well-characterized cohort of 1,279 individuals with incident HF hospitalizations from the ARIC study, which has rigorous data collection and surveillance process. Both hs-cTnT levels and BMI were measured remotely prior to HF diagnosis and were therefore not affected by overt myocardial dysfunction or clinically manifest HF/volume overload, which could otherwise confound the levels of hs-cTnT and BMI, respectively. On the other hand, measurements of their values too remotely may not be an accurate proxy of recent myocardial injury. Although only hospitalized incident HF cases were included in the current analysis because of lack of consistent data for outpatient HF through the study period, the sensitivity and positive predictive value of ICD-9 code 428.x for HF classified by subsequent medical record review by the ARIC criteria were 0.95 and 0.77, respectively for combined acute decompensated HF and chronic HF (in comparison to 0.83 and 0.78, respectively, by Framingham criteria) (21). Furthermore, there is a possibility of differential misclassification because HF diagnosis in obese patients may be less specific than in normal-weight patients. We were also not able to account for the treatment after the onset of HF. In addition, HF types (preserved or reduced ejection fraction) were not assessed for the entire study period. Finally, despite rigorous adjustment, residual confounding could still be possible because of nonrandomized nature of the study.
The presence of pre-existing subclinical myocardial injury (i.e., before incident HF) could not explain the lower mortality risk after incident HF observed in obese and overweight individuals compared with normal weight individuals. Future research is needed to explore other factors to explain the observed obesity paradox in HF even using the pre-HF obesity status.
COMPETENCY IN MEDICAL KNOWLEDGE: Although greater subclinical myocardial injury (assessed by high-sensitivity cardiac troponin-T levels) before heart failure increases risk of death after heart failure diagnosis, individuals who were obese and overweight prior to heart failure had lower risk of death than normal weight individuals after development of heart failure, regardless of the presence or absence of pre-heart failure subclinical myocardial injury.
TRANSLATIONAL OUTLOOK: Future studies should evaluate other cardiac and noncardiac factors including neurohormonal activation, metabolic reserve, and other biomarkers to explain the reasons for the obesity paradox in heart failure.
The authors thank the staff and participants of the ARIC study for their important contributions. Reagents for high-sensitivity troponin T were donated by Roche Diagnostics.
For supplemental tables and figure, please see the online version of this article.
The Atherosclerosis Risk in Communities Study is a collaborative study supported by National Institutes of Health/National Heart, Lung, and Blood Institute contracts HHSN268201100005C, HHSN268201100006C, HHSN268201100007C, HHSN268201100008C, HHSN268201100009C, HHSN268201100010C, HHSN268201100011C, and HHSN268201100012C. Dr. Pokharel was supported by NIH/National Heart, Lung, and Blood Institute Award T32HL110837. Dr. Selvin is supported by grants from the NIH. Dr. Virani has received honoraria from American College of Cardiology as Associate Editor for Innovations. Dr. Nambi is the holder of a patent assigned to Roche and a compensated adjudicator for studies conducted by Siemens. Dr. Hoogeveen has received research support from Denka Seiken, Co. Dr. Selvin has served on an advisory board for Roche Diagnostics. Dr. Ballantyne is a consultant for and holder of a patent assigned to Roche. 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
- coronary heart disease
- cardiac troponin T measured using high-sensitivity assay
- heart failure
- N-terminal pro-B-type natriuretic peptide
- Received September 15, 2016.
- Revision received October 11, 2016.
- Accepted October 14, 2016.
- Mozaffarian D.,
- Benjamin E.J.,
- Go A.S.,
- et al.
- Loehr L.R.,
- Rosamond W.D.,
- Poole C.,
- et al.
- Horwich T.B.,
- Fonarow G.C.,
- Hamilton M.A.,
- MacLellan W.R.,
- Woo M.A.,
- Tillisch J.H.
- Shah R.,
- Gayat E.,
- Januzzi J.L. Jr..,
- et al.
- Khalid U.,
- Ather S.,
- Bavishi C.,
- et al.
- Saunders J.T.,
- Nambi V.,
- de Lemos J.A.,
- et al.
- ↵(1989) The Atherosclerosis Risk in Communities (ARIC) study: design and objectives. The ARIC investigators. Am J Epidemiol 129:687–702.
- Baecke J.A.,
- Burema J.,
- Frijters J.E.
- Rathore S.S.,
- Hinn A.R.,
- Cooper L.S.,
- Tyroler H.A.,
- Rosamond W.D.
- Pokharel Y.,
- Sun W.,
- de Lemos J.A.,
- et al.
- Rosamond W.D.,
- Chang P.P.,
- Baggett C.,
- et al.
- Ndumele C.E.,
- Coresh J.,
- Lazo M.,
- et al.
- Kalantar-Zadeh K.,
- Streja E.,
- Kovesdy C.P.,
- et al.
- Lavie C.J.,
- Alpert M.A.,
- Arena R.,
- Mehra M.R.,
- Milani R.V.,
- Ventura H.O.