[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure.
Association of Individual Biomarkers With Incident Heart Failure With Preserved Ejection Fraction (HFpEF) and Heart Failure With Reduced Ejection Fraction (HFrEF) in Multivariable-Adjusted Analyses
Association of Individual Biomarkers With Incident Heart Failure With Preserved Ejection Fraction (HFpEF) and Heart Failure With Reduced Ejection Fraction (HFrEF) in Multivariable-Adjusted Analyses

Hazard ratios (HRs) and 95% CIs for HFpEF are shown in blue and for HFrEF in grey. Three biomarkers had significantly greater HRs for HFrEF compared with HFpEF, including natriuretic peptides, high-sensitivity troponins (hs-troponins), and C-reactive protein (CRP). IL-6 indicates interleukin 6; LVEF, left ventricular ejection fraction; PAI-1, plasminogen activator inhibitor 1; sST2, soluble suppressor of tumorigenicity; and UACR, urinary albumin to creatinine ratio.

aStatistically significant difference between HRs for HFpEF vs HFrEF (using Lunn-McNeil test).

Table 1.  
Baseline Clinical and Laboratory Covariates by Cohort
Baseline Clinical and Laboratory Covariates by Cohort
Table 2.  
Cohort-Specific Associations of Individual Biomarker With HFpEF and HFrEF
Cohort-Specific Associations of Individual Biomarker With HFpEF and HFrEF
Table 3.  
Multivariable-Adjusted Pooled Associations of Individual Biomarkers With HF Subtypes
Multivariable-Adjusted Pooled Associations of Individual Biomarkers With HF Subtypes
Table 4.  
C Statistics and NRI for Multivariable-Adjusted Models
C Statistics and NRI for Multivariable-Adjusted Models
1.
Lloyd-Jones  DM, Larson  MG, Leip  EP,  et al; Framingham Heart Study.  Lifetime risk for developing congestive heart failure: the Framingham Heart Study.  Circulation. 2002;106(24):3068-3072.PubMedGoogle ScholarCrossref
2.
Ponikowski  P, Anker  SD, AlHabib  KF,  et al.  Heart failure: preventing disease and death worldwide.  ESC Heart Fail. 2014;1(1):4-25.PubMedGoogle ScholarCrossref
3.
Huelsmann  M, Neuhold  S, Resl  M,  et al.  PONTIAC (NT-proBNP selected prevention of cardiac events in a population of diabetic patients without a history of cardiac disease): a prospective randomized controlled trial.  J Am Coll Cardiol. 2013;62(15):1365-1372.PubMedGoogle ScholarCrossref
4.
Ledwidge  M, Gallagher  J, Conlon  C,  et al.  Natriuretic peptide–based screening and collaborative care for heart failure: the STOP-HF randomized trial.  JAMA. 2013;310(1):66-74.PubMedGoogle ScholarCrossref
5.
Ho  JE, Enserro  D, Brouwers  FP,  et al.  Predicting heart failure with preserved and reduced ejection fraction: the international collaboration on heart failure subtypes.  Circ Heart Fail. 2016;9(6):e003116.PubMedGoogle ScholarCrossref
6.
Dawber  TR, Kannel  WB, Lyell  LP.  An approach to longitudinal studies in a community: the Framingham Study.  Ann N Y Acad Sci. 1963;107:539-556.PubMedGoogle ScholarCrossref
7.
Kannel  WB, Feinleib  M, McNamara  PM, Garrison  RJ, Castelli  WP.  An investigation of coronary heart disease in families: the Framingham Offspring Study.  Am J Epidemiol. 1979;110(3):281-290.PubMedGoogle ScholarCrossref
8.
Psaty  BM, Kuller  LH, Bild  D,  et al.  Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study.  Ann Epidemiol. 1995;5(4):270-277.PubMedGoogle ScholarCrossref
9.
Hillege  HL, Fidler  V, Diercks  GF,  et al; Prevention of Renal and Vascular End Stage Disease (PREVEND) Study Group.  Urinary albumin excretion predicts cardiovascular and noncardiovascular mortality in general population.  Circulation. 2002;106(14):1777-1782.PubMedGoogle ScholarCrossref
10.
Bild  DE, Bluemke  DA, Burke  GL,  et al.  Multi-Ethnic Study of Atherosclerosis: objectives and design.  Am J Epidemiol. 2002;156(9):871-881.PubMedGoogle ScholarCrossref
11.
Paulus  WJ, Tschöpe  C.  A novel paradigm for heart failure with preserved ejection fraction: comorbidities drive myocardial dysfunction and remodeling through coronary microvascular endothelial inflammation.  J Am Coll Cardiol. 2013;62(4):263-271.PubMedGoogle ScholarCrossref
12.
Fine  JP, Gray  RJ.  A proportional hazards model for the subdistribution of a competing risk.  J Am Stat Assoc. 1999;94(446):496-509. doi:10.2307/2670170Google ScholarCrossref
13.
Lunn  M, McNeil  D.  Applying Cox regression to competing risks.  Biometrics. 1995;51(2):524-532.PubMedGoogle ScholarCrossref
14.
Pencina  MJ, D’Agostino  RB  Sr, Steyerberg  EW.  Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers.  Stat Med. 2011;30(1):11-21.PubMedGoogle ScholarCrossref
15.
Wang  TJ, Wollert  KC, Larson  MG,  et al.  Prognostic utility of novel biomarkers of cardiovascular stress: the Framingham Heart Study.  Circulation. 2012;126(13):1596-1604.PubMedGoogle ScholarCrossref
16.
Vasan  RS, Sullivan  LM, Roubenoff  R,  et al; Framingham Heart Study.  Inflammatory markers and risk of heart failure in elderly subjects without prior myocardial infarction: the Framingham Heart Study.  Circulation. 2003;107(11):1486-1491.PubMedGoogle ScholarCrossref
17.
Frankel  DS, Vasan  RS, D’Agostino  RB  Sr,  et al.  Resistin, adiponectin, and risk of heart failure: the Framingham Offspring Study.  J Am Coll Cardiol. 2009;53(9):754-762.PubMedGoogle ScholarCrossref
18.
Ho  JE, Liu  C, Lyass  A,  et al.  Galectin-3, a marker of cardiac fibrosis, predicts incident heart failure in the community.  J Am Coll Cardiol. 2012;60(14):1249-1256.PubMedGoogle ScholarCrossref
19.
Ho  JE, Lyass  A, Lee  DS,  et al.  Predictors of new-onset heart failure: differences in preserved versus reduced ejection fraction.  Circ Heart Fail. 2013;6(2):279-286.PubMedGoogle ScholarCrossref
20.
Xanthakis  V, Larson  MG, Wollert  KC,  et al.  Association of novel biomarkers of cardiovascular stress with left ventricular hypertrophy and dysfunction: implications for screening.  J Am Heart Assoc. 2013;2(6):e000399.PubMedGoogle ScholarCrossref
21.
Brouwers  FP, van Gilst  WH, Damman  K,  et al.  Clinical risk stratification optimizes value of biomarkers to predict new-onset heart failure in a community-based cohort.  Circ Heart Fail. 2014;7(5):723-731.PubMedGoogle ScholarCrossref
22.
Brouwers  FP, de Boer  RA, van der Harst  P,  et al.  Incidence and epidemiology of new onset heart failure with preserved vs reduced ejection fraction in a community-based cohort: 11-year follow-up of PREVEND.  Eur Heart J. 2013;34(19):1424-1431.PubMedGoogle ScholarCrossref
23.
Agarwal  SK, Chambless  LE, Ballantyne  CM,  et al.  Prediction of incident heart failure in general practice: the Atherosclerosis Risk in Communities (ARIC) Study.  Circ Heart Fail. 2012;5(4):422-429.PubMedGoogle ScholarCrossref
24.
Seliger  SL, de Lemos  J, Neeland  IJ,  et al.  Older adults, “malignant” left ventricular hypertrophy, and associated cardiac-specific biomarker phenotypes to identify the differential risk of new-onset reduced versus preserved ejection fraction heart failure: CHS (Cardiovascular Health Study).  JACC Heart Fail. 2015;3(6):445-455.PubMedGoogle ScholarCrossref
25.
Kalogeropoulos  A, Psaty  BM, Vasan  RS,  et al; Cardiovascular Health Study.  Validation of the Health ABC heart failure model for incident heart failure risk prediction: the Cardiovascular Health Study.  Circ Heart Fail. 2010;3(4):495-502.PubMedGoogle ScholarCrossref
26.
Bahrami  H, Kronmal  R, Bluemke  DA,  et al.  Differences in the incidence of congestive heart failure by ethnicity: the Multi-Ethnic Study of Atherosclerosis.  Arch Intern Med. 2008;168(19):2138-2145.PubMedGoogle ScholarCrossref
27.
Bahrami  H, Bluemke  DA, Kronmal  R,  et al.  Novel metabolic risk factors for incident heart failure and their relationship with obesity: the MESA (Multi-Ethnic Study of Atherosclerosis) Study.  J Am Coll Cardiol. 2008;51(18):1775-1783.PubMedGoogle ScholarCrossref
28.
AbouEzzeddine  OF, McKie  PM, Scott  CG,  et al.  Biomarker-based risk prediction in the community.  Eur J Heart Fail. 2016;18(11):1342-1350.PubMedGoogle ScholarCrossref
29.
Echouffo-Tcheugui  JB, Greene  SJ, Papadimitriou  L,  et al.  Population risk prediction models for incident heart failure: a systematic review.  Circ Heart Fail. 2015;8(3):438-447.PubMedGoogle ScholarCrossref
30.
Willeit  P, Kaptoge  S, Welsh  P,  et al; Natriuretic Peptides Studies Collaboration.  Natriuretic peptides and integrated risk assessment for cardiovascular disease: an individual-participant-data meta-analysis.  Lancet Diabetes Endocrinol. 2016;4(10):840-849.PubMedGoogle ScholarCrossref
31.
Meijers  WC, de Boer  RA, Ho  JE.  Biomarkers to identify and prevent new-onset heart failure in the community.  Eur J Heart Fail. 2016;18(11):1351-1352.PubMedGoogle ScholarCrossref
32.
de Boer  RA, Daniels  LB, Maisel  AS, Januzzi  JL  Jr.  State of the art: newer biomarkers in heart failure.  Eur J Heart Fail. 2015;17(6):559-569.PubMedGoogle ScholarCrossref
33.
Shah  SJ, Katz  DH, Selvaraj  S,  et al.  Phenomapping for novel classification of heart failure with preserved ejection fraction.  Circulation. 2015;131(3):269-279.PubMedGoogle ScholarCrossref
34.
Senni  M, Paulus  WJ, Gavazzi  A,  et al.  New strategies for heart failure with preserved ejection fraction: the importance of targeted therapies for heart failure phenotypes.  Eur Heart J. 2014;35(40):2797-2815.PubMedGoogle ScholarCrossref
35.
Kitzman  DW, Little  WC, Brubaker  PH,  et al.  Pathophysiological characterization of isolated diastolic heart failure in comparison to systolic heart failure.  JAMA. 2002;288(17):2144-2150.PubMedGoogle ScholarCrossref
36.
Katz  DH, Selvaraj  S, Aguilar  FG,  et al.  Association of low-grade albuminuria with adverse cardiac mechanics: findings from the Hypertension Genetic Epidemiology Network (HyperGEN) Study.  Circulation. 2014;129(1):42-50.PubMedGoogle ScholarCrossref
37.
Brouwers  FP, Asselbergs  FW, Hillege  HL,  et al.  Long-term effects of fosinopril and pravastatin on cardiovascular events in subjects with microalbuminuria: ten years of follow-up of Prevention of Renal and Vascular End-stage Disease Intervention Trial (PREVEND IT).  Am Heart J. 2011;161(6):1171-1178.PubMedGoogle ScholarCrossref
38.
Asselbergs  FW, Diercks  GF, Hillege  HL,  et al; Prevention of Renal and Vascular Endstage Disease Intervention Trial (PREVEND IT) Investigators.  Effects of fosinopril and pravastatin on cardiovascular events in subjects with microalbuminuria.  Circulation. 2004;110(18):2809-2816.PubMedGoogle ScholarCrossref
39.
Wong  YW, Thomas  L, Sun  JL,  et al.  Predictors of incident heart failure hospitalizations among patients with impaired glucose tolerance: insight from the Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research Study.  Circ Heart Fail. 2013;6(2):203-210.PubMedGoogle ScholarCrossref
40.
Katz  DH, Burns  JA, Aguilar  FG, Beussink  L, Shah  SJ.  Albuminuria is independently associated with cardiac remodeling, abnormal right and left ventricular function, and worse outcomes in heart failure with preserved ejection fraction.  JACC Heart Fail. 2014;2(6):586-596.PubMedGoogle ScholarCrossref
41.
Ter Maaten  JM, Damman  K, Verhaar  MC,  et al.  Connecting heart failure with preserved ejection fraction and renal dysfunction: the role of endothelial dysfunction and inflammation.  Eur J Heart Fail. 2016;18(6):588-598.PubMedGoogle ScholarCrossref
42.
van der Velde  AR, Meijers  WC, Ho  JE,  et al.  Serial galectin-3 and future cardiovascular disease in the general population.  Heart. 2016;102(14):1134-1141.PubMedGoogle ScholarCrossref
43.
Meijers  WC, van der Velde  AR, Muller Kobold  AC,  et al.  Variability of biomarkers in patients with chronic heart failure and healthy controls.  Eur J Heart Fail. 2017;19(3):357-365.PubMedGoogle ScholarCrossref
Original Investigation
March 2018

Association of Cardiovascular Biomarkers With Incident Heart Failure With Preserved and Reduced Ejection Fraction

Author Affiliations
  • 1Department of Cardiology, University of Groningen, University Medical Centre Groningen, Groningen, the Netherlands
  • 2Cardiology Division, Department of Medicine, Massachusetts General Hospital, Boston
  • 3Inova Heart and Vascular Institute, Falls Church, Virginia
  • 4Department of Preventive Medicine, Boston University School of Medicine, Boston, Massachusetts
  • 5Cardiovascular Research Center, Massachusetts General Hospital, Boston
  • 6Department of Medicine, Albert Einstein College of Medicine, Bronx, New York
  • 7Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, New York
  • 8Ciccarone Center for the Prevention of Heart Disease, The Johns Hopkins University, Baltimore, Maryland
  • 9Division of Hematology/Oncology, Department of Medicine, University of Vermont Larner College of Medicine, Burlington
  • 10Department of Medicine, Johns Hopkins Medical Institutions, The Johns Hopkins University, Baltimore, Maryland
  • 11Department of Cardiology, Heart and Vascular Institute, Johns Hopkins Medical Institutions, The Johns Hopkins University, Baltimore, Maryland
  • 12Division of Cardiovascular Medicine, Keck School of Medicine of University of Southern California, Los Angeles
  • 13Division of Cardiovascular Medicine, Vanderbilt University Medical Center, Nashville, Tennessee
  • 14Department of Internal Medicine, University of Groningen, University Medical Centre Groningen, Groningen, the Netherlands
  • 15Center for Population Studies, National Heart, Lung, and Blood Institute, Bethesda, Maryland
  • 16Center of Research on Psychology in Somatic Diseases, Department of Medical and Clinical Psychology, Tilburg University, Tilburg, the Netherlands
  • 17Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois
  • 18Framingham Heart Study, Framingham, Massachusetts
  • 19Cardiovascular Medicine Section, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts
  • 20Section of Preventive Medicine and Epidemiology, Boston University School of Medicine, Boston, Massachusetts
  • 21Department of Epidemiology, Boston University School of Public Health, Boston, Massachusetts
  • 22Cardiovascular Health Research Unit, Departments of Medicine, Epidemiology and Health Services, University of Washington, Seattle
  • 23Kaiser Permanente Washington Health Research Institute, Seattle
  • 24Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada
  • 25Department of Biostatistics, University of Washington, Seattle
  • 26Department of Family Medicine and Public Health, University of California, San Diego, La Jolla
  • 27Division of Cardiology, Department of Medicine, Rutgers New Jersey Medical School, Newark
  • 28Division of Cardiology, Northwestern University Feinberg School of Medicine, Chicago, Illinois
  • 29Section on Cardiovascular Medicine, Wake Forest School of Medicine, Winston-Salem, North Carolina
  • 30Division of Public Health Sciences, Department of Epidemiology and Prevention, Wake Forest School of Medicine, Winston-Salem, North Carolina
JAMA Cardiol. 2018;3(3):215-224. doi:10.1001/jamacardio.2017.4987
Key Points

Question  What cardiovascular biomarkers are associated with the development of heart failure with preserved vs reduced ejection fraction?

Findings  Among 22 756 participants enrolled in 4 longitudinal community-based cohorts, several biomarkers of renal dysfunction, endothelial dysfunction, and inflammation, in addition to natriuretic peptides and high-sensitivity troponin, were associated with incident heart failure with reduced ejection fraction. By contrast, only natriuretic peptides and urinary albumin to creatinine ratio were associated with heart failure with preserved ejection fraction.

Meaning  These findings highlight the need for future studies focused on identifying novel biomarkers of heart failure with preserved ejection fraction.

Abstract

Importance  Nearly half of all patients with heart failure have preserved ejection fraction (HFpEF) as opposed to reduced ejection fraction (HFrEF), yet associations of biomarkers with future heart failure subtype are incompletely understood.

Objective  To evaluate the associations of 12 cardiovascular biomarkers with incident HFpEF vs HFrEF among adults from the general population.

Design, Setting, and Participants  This study included 4 longitudinal community-based cohorts: the Cardiovascular Health Study (1989-1990; 1992-1993 for supplemental African-American cohort), the Framingham Heart Study (1995-1998), the Multi-Ethnic Study of Atherosclerosis (2000-2002), and the Prevention of Renal and Vascular End-stage Disease study (1997-1998). Each cohort had prospective ascertainment of incident HFpEF and HFrEF. Data analysis was performed from June 25, 2015, to November 9, 2017.

Exposures  The following biomarkers were examined: N-terminal pro B-type natriuretic peptide or brain natriuretic peptide, high-sensitivity troponin T or I, C-reactive protein (CRP), urinary albumin to creatinine ratio (UACR), renin to aldosterone ratio, D-dimer, fibrinogen, soluble suppressor of tumorigenicity, galectin-3, cystatin C, plasminogen activator inhibitor 1, and interleukin 6.

Main Outcomes and Measures  Development of incident HFpEF and incident HFrEF.

Results  Among the 22 756 participants in these 4 cohorts (12 087 women and 10 669 men; mean [SD] age, 60 [13] years) in the study, during a median follow-up of 12 years, 633 participants developed incident HFpEF, and 841 developed HFrEF. In models adjusted for clinical risk factors of heart failure, 2 biomarkers were significantly associated with incident HFpEF: UACR (hazard ratio [HR], 1.33; 95% CI, 1.20-1.48; P < .001) and natriuretic peptides (HR, 1.27; 95% CI, 1.16-1.40; P < .001), with suggestive associations for high-sensitivity troponin (HR, 1.11; 95% CI, 1.03-1.19; P = .008), plasminogen activator inhibitor 1 (HR, 1.22; 95% CI, 1.03-1.45; P = .02), and fibrinogen (HR, 1.12; 95% CI, 1.03-1.22; P = .01). By contrast, 6 biomarkers were associated with incident HFrEF: natriuretic peptides (HR, 1.54; 95% CI, 1.41-1.68; P < .001), UACR (HR, 1.21; 95% CI, 1.11-1.32; P < .001), high-sensitivity troponin (HR, 1.37; 95% CI, 1.29-1.46; P < .001), cystatin C (HR, 1.19; 95% CI, 1.11-1.27; P < .001), D-dimer (HR, 1.22; 95% CI, 1.11-1.35; P < .001), and CRP (HR, 1.19; 95% CI, 1.11-1.28; P < .001). When directly compared, natriuretic peptides, high-sensitivity troponin, and CRP were more strongly associated with HFrEF compared with HFpEF.

Conclusions and Relevance  Biomarkers of renal dysfunction, endothelial dysfunction, and inflammation were associated with incident HFrEF. By contrast, only natriuretic peptides and UACR were associated with HFpEF. These findings highlight the need for future studies focused on identifying novel biomarkers of the risk of HFpEF.

Introduction

Heart failure (HF) is a major worldwide public health burden. The lifetime risk of HF is substantial, occurring in 20% of both men and women.1,2 With aging populations in the United States and European Union, the overall prevalence of HF is predicted to triple by 2060.2 There is a strong need to improve methods to identify people who are at high risk for developing HF and to implement effective preventive measures. Several previous studies have demonstrated the utility of targeting HF prevention efforts to higher-risk individuals who were identified by single biomarker assessments.3,4 Individual assessment of HF risk, however, is still in its infancy, and the clinical applicability of existing models of assessment of HF risk is limited. Furthermore, half of patients presenting with HF are classified as preserved ejection fraction (HFpEF) vs reduced ejection fraction (HFrEF). This distinction has important therapeutic implications, yet differences in antecedent factors preceding each HF subtype are poorly understood.

Challenges in assessment of HF risk include the relatively low incidence rate in asymptomatic community-dwelling individuals, the heterogeneity of the phenotype, and a lengthy subclinical phase requiring long-term follow-up. To fill this gap in the literature, we established an international collaboration to create and validate risk assessment models for incident HF subtypes.5 We have pooled data from 4 large community-based longitudinal cohorts: the Framingham Heart Study (FHS),6,7 the Cardiovascular Health Study (CHS),8 the Prevention of Renal and Vascular End-stage Disease (PREVEND) cohort,9 and the Multi-Ethnic Study of Atherosclerosis (MESA).10 A previous study has examined differences in clinical factors traditionally associated with HFpEF vs HFrEF, including age, blood pressure, and body weight.5 Whether biomarkers can offer additional insights into risk of future HFpEF vs HFrEF remains unknown. We therefore sought to investigate the association of cardiovascular biomarkers representing several biological pathways, including myocyte stretch, stress, inflammation, and renal function, with the development of future HFpEF vs HFrEF. We leveraged pathway biomarkers that were previously measured across at least 2 cohorts. We hypothesized that biomarker profiles preceding HFpEF would be distinct from those preceding HFrEF given that different predominant pathophysiologic drivers are thought to underlie myocardial remodeling in these 2 entities.11

Methods
Study Sample

We harmonized and pooled individual-level data from 4 prospective, observational community-based cohorts with adjudicated incident HF outcomes as previously described.5 We included 24 803 participants with at least 1 biomarker assessment available from the following baseline examinations: FHS offspring cohort exam 6 (1995-1998), CHS exam 1 (1989-1990; 1992-1993 for supplemental African-American cohort), PREVEND exam 1 (1997-1998), and MESA exam 1 (2000-2002). Individuals with prevalent HF (n = 326), age younger than 30 years (n = 124), missing covariates (n = 1570), or without follow-up available (n = 27) were excluded, leaving 22 756 individuals for analysis. Written informed consent was obtained for all study participants, and institutional review board approval was obtained separately for all 4 cohorts (from Boston University [FHS]; Columbia University [MESA]; Johns Hopkins University [CHS and MESA]; Northwestern University [MESA]; University of California, Davis [CHS]; University of California, Los Angeles [MESA]; University of Groningen [PREVEND]; University of Minnesota [MESA]; University of Pittsburgh [CHS]; and Wake Forest University [CHS and MESA]).

Clinical Assessment

A detailed medical history, physical examination, fasting blood draw, and electrocardiography were performed on all participants at the baseline examinations. Risk factors were evaluated and harmonized across cohorts whenever possible, as described.5 Blood pressure was taken as the mean of 2 seated measurements. Body mass index was calculated as weight in kilograms divided by height in meters squared. Diabetes was defined as a fasting glucose level of 126 mg/dL or more (to convert to millimoles per liter, multiply by 0.0555), random glucose level of 200 mg/dL or more, or the use of hypoglycemic medications. Electrocardiographic left ventricular (LV) hypertrophy was defined based on accepted voltage and ST-segment criteria, as described previously.5

Definition of Incident HF Subtypes

Individuals were followed prospectively for the occurrence of incident HF or death. Outcomes were adjudicated using established protocols by study investigators within each cohort after review of all available outpatient and hospital records. Heart failure was defined using a combination of signs and symptoms, as described previously.5 Each first incident HF event was categorized as HFpEF (LV ejection fraction [LVEF], ≥50%), HFrEF (LVEF, <50%), or unclassified (no LV function assessment available).5 After an incident HF event, participants were censored and were not at risk for the other HF subtype or unclassified HF.

Biomarker Assays

eTable 1 in the Supplement summarizes cohort-specific assay details of the following 12 biomarkers, which were available in at least 2 of 4 cohorts: C-reactive protein, interleukin 6, plasminogen activator inhibitor 1, D-dimer, fibrinogen, galectin-3, N-terminal pro B-type natriuretic peptide (NT-proBNP), renin, aldosterone, high-sensitivity cardiac troponin (hs-Tn) T or I, soluble suppressor of tumorigenicity 2, urinary albumin to creatinine ratio (UACR), and cystatin C. For natriuretic peptides, brain natriuretic peptide was measured in the FHS cohort and NT-proBNP in the other cohorts. Similarly, hs-TnI was measured in the FHS cohort and hs-TnT in the remaining cohorts.

Statistical Analysis

Statistical analysis was performed from June 25, 2015, to November 9, 2017. Baseline clinical characteristics for participants with at least 1 biomarker available were summarized by cohort. Biomarker concentrations were natural log transformed. To account for interassay and cohort-specific factors, we performed direct standardization within each cohort to center mean values to 0 and set each unit change to 1 SD. Individual-level data from each of the 4 cohorts were then pooled for the subsequent analyses. Follow-up time was truncated at 15 years.

We used Fine-Gray proportional subdistribution hazards models to evaluate the association of each individual biomarker with overall HF and HF subtypes of HFpEF and HFrEF in the pooled sample.12 These models accounted for the competing risks of death, other HF subtype, and unclassified HF. This technique was chosen to account for informative censoring when evaluating the association of biomarkers and incident HF subtypes. Models were adjusted for age and sex, and then additionally for race/ethnicity, previous myocardial infarction, hypertension treatment, systolic blood pressure, smoking status, presence of LV hypertrophy or left bundle branch block, and diabetes.5 The “strata” statement was included to account for study cohort as well as for stratified recruitment in PREVEND (24-hour urine albumin excretion >10 mg/L vs <10 mg/L at recruitment [to convert to grams per liter, multiply by 10.0]). Secondary analyses accounted for both cohort and recruitment center among CHS and MESA participants, without substantial differences in findings. Results for primary analyses were considered significant using a Bonferroni-corrected, 2-sided P value threshold corrected for the 12 biomarkers tested, P = .05/12 = .004. Results with .004 < P < .05 were considered suggestive. The β coefficients associating each biomarker with HFpEF vs HFrEF were formally tested for equality using the method of Lunn and McNeil.13

In secondary analyses, we additionally adjusted for estimated glomerular filtration rate, interim myocardial infarction, and natriuretic peptides. We examined a multimarker model containing biomarkers available in all 4 cohorts (hs-Tn, natriuretic peptides, C-reactive protein, and cystatin C) along with clinical factors. We also conducted cohort-specific analyses (age- and sex-adjusted and multivariable-adjusted) and tested for age × biomarker associations.

We assessed the incremental effect of each biomarker on its association with HFpEF and HFrEF. C statistics were compared between models with clinical covariates alone (based on a previously validated prediction model)5 and models including both clinical covariates and the given biomarker. We calculated the category-free net reclassification index using previously described methods.14 We created a biomarker risk score from significant biomarkers that were available in 3 or more cohorts.15 The biomarker risk score for HFpEF was composed of UACR, natriuretic peptides, and hs-Tn, and the risk score for HFrEF included UACR, natriuretic peptides, hs-Tn, cystatin C, and C-reactive protein. We then examined the risk of HF subtype by quartiles of biomarker risk score. All statistical analyses were conducted with SAS, version 9.4 software (SAS Institute).

Results

There were 22 756 participants for analysis: 3431 (15.1%) from FHS, 5277 (23.2%) from CHS, 7369 (32.4%) from PREVEND, and 6679 (29.4%) from MESA. Mean (SD) baseline ages of participants ranged from 49 (12) years in PREVEND to 73 (6) years in CHS, with a mean (SD) age of 60 (13) years for the total cohort; 12 087 of the participants (53.1%) were women. Baseline clinical characteristics by cohort are detailed in Table 1. Over a mean (SD) follow-up of 12 (3) years, there were a total of 2095 incident HF events. Within this group, 1474 (70.4%) had cardiac imaging available at or around the time of HF presentation, allowing for classification of HF subtypes. Among the participants whose HF subtype was classified, there were 633 HFpEF events (42.9%) and 841 HFrEF events (57%.1). Over the same follow-up period, there were 5187 deaths (22.8%).

Biomarker Distribution in the Cohort

Mean biomarker concentrations at baseline are displayed in eTable 2 in the Supplement and generally were within the normal range, as expected in community-based samples. There were modest intrabiomarker correlations, with the 5 largest age- and sex-adjusted partial correlation coefficients for hs-Tn and natriuretic peptides (r = 0.29; P < .001), galectin-3 and cystatin-C (r = 0.30; P < .001), C-reactive protein and fibrinogen (r = 0.49; P < .001), interleukin 6 and fibrinogen (r = 0.40; P < .001), and interleukin 6 and C-reactive protein (r = 0.53; P < .001) (eTable 3 in the Supplement).

Biomarker Associations With HF Subtypes

Cohort-specific associations of biomarkers with HFpEF and HFrEF are presented in Table 2 and show some differences in effect sizes between cohorts. In pooled analyses, after adjusting for age, sex, race/ethnicity, systolic blood pressure, hypertension treatment, body mass index, diabetes, smoking status, presence of LV hypertrophy, and left bundle branch block, only UACR (hazard ratio [HR], 1.33; 95% CI, 1.20-1.48; P < .001) and natriuretic peptides (HR, 1.27; 95% CI, 1.16-1.40; P < .001) were associated with risk for HFpEF (Table 3 and Figure). Suggestive associations were observed with hs-Tn (HR, 1.11; 95% CI, 1.03-1.19; P = .008), plasminogen activator inhibitor 1 (HR, 1.22; 95% CI, 1.03-1.45; P = .02), and fibrinogen (HR, 1.12; 95% CI, 1.04-1.22; P = .01) (Table 3 and Figure). By contrast, 6 individual biomarkers were positively associated with HFrEF: natriuretic peptides (HR, 1.54; 95% CI, 1.41-1.68; P < .001), UACR (HR, 1.21; 95% CI, 1.11-1.32; P < .001), hs-Tn (HR, 1.37; 95% CI, 1.29-1.46; P < .001), cystatin C (HR, 1.19; 95% CI, 1.11-1.27; P < .001), D-dimer (HR, 1.22; 95% CI, 1.11-1.35; P < .001), and C-reactive protein (HR, 1.19; 95% CI, 1.11-1.28; P < .001). All significant biomarkers appeared to have linear associations with incident HF outcomes when examined using restricted cubic splines.

When formally tested, natriuretic peptides, C-reactive protein, and hs-Tn had differential associations with HFrEF vs HFpEF; these biomarkers were more strongly associated with HFrEF compared with HFpEF. Specifically, a 1-SD increase in log-transformed natriuretic peptides was associated with a 1.5-fold increased risk of future HFrEF (95% CI, 1.41-1.68) compared with a 1.3-fold increased risk of HFpEF (95% CI, 1.16-1.40; P = .005 for difference). Similarly, a 1-SD increase in hs-Tn was associated with a 1.4-fold increased risk of HFrEF (95%, CI, 1.29-1.46), whereas hs-Tn was more weakly associated with HFpEF (HR, 1.11; 95% CI, 1.03-1.19; P = .008; P < .001 for difference).

Incremental Performance Metrics of Biomarkers in Clinical Models

To further investigate if the addition of biomarkers to our base clinical model would improve estimation of risk, we calculated the increment in C statistics from the addition of individual biomarkers to clinical models estimating HFpEF and HFrEF (Table 4). Several biomarkers increased the C statistics, especially for HFrEF, such as natriuretic peptides (+0.022) and hs-Tn (+0.021). For HFpEF, the increment was smaller, with the largest increment for UACR (+0.010). Similarly, the category-free net reclassification index was modest for single biomarker models—the largest net reclassification index for HFpEF was with the addition of natriuretic peptides (0.16; 95% CI, 0.06-0.26). High-sensitivity troponin also had the largest net reclassification index for association with HFrEF (0.25; 95% CI, 0.17-0.33).

Multimarker Models

In a multimarker model including biomarkers available in all 4 cohorts (hs-Tn, natriuretic peptides, C-reactive protein, and cystatin C), only natriuretic peptides remained associated with HFpEF after accounting for clinical HF risk factors (HR, 1.27; 95% CI, 1.14-1.41; P < .001), whereas other biomarkers were not significant. By contrast, hs-Tn (HR, 1.24; 95% CI, 1.15-1.34; P < .001) and natriuretic peptides were associated with HFrEF (HR, 1.44; 95% CI, 1.30-1.58; P < .001), with a suggestive association for C-reactive protein (HR, 1.10; 95% CI, 1.01-1.20; P = .03). For HFpEF, the C statistic increment with addition of multimarkers was limited (0.787- 0.795); for HFrEF, addition of the 4 biomarkers to the model was associated with a modest increment (0.804-0.839).

We created a biomarker risk score of significant biomarkers available across at least 3 cohorts and found an increase in HR across risk score quartiles (eTable 4 in the Supplement). For HFpEF, HR among the fourth quartile was 4.96 (95% CI, 2.54-9.70), with the first quartile serving as the reference. For HFrEF, the HR among the fourth quartile was 10.95 (95% CI, 5.67-21.17) compared with the first quartile.

Secondary Analyses and Effect Modification

We calculated cohort-specific associations of the biomarkers with HF subtypes to query if outcomes could be driven by the results in 1 cohort, with age- and sex-adjusted analyses presented in eTable 5 in the Supplement and multivariable models in Table 2. The direction of associations between the biomarkers and incident HF subtypes was generally consistent across cohorts despite modest differences in individual β coefficients. In further analyses, we additionally adjusted multivariable models for estimated glomerular filtration rate, interim myocardial infarction, and natriuretic peptides (eTable 6A and 6B in the Supplement). Overall, these adjustments resulted in minor changes in effect estimates.

We tested for effect modification by age on the associations of biomarkers and HF subtypes using cohort-specific models (eTable 7 in the Supplement). In models adjusted for clinical covariates, statistically significant interactions between biomarkers and age were observed for the association of HFpEF with UACR, natriuretic peptides, hs-Tn, and aldosterone to renin ratio in at least 1 cohort. For the HFrEF analyses, statistically significant age × biomarker associations were observed for UACR, natriuretic peptides, and C-reactive protein.

Discussion

In the current analysis, we evaluated the associations of 12 cardiovascular biomarkers with incident HFpEF and HFrEF among 4 longitudinal community-based cohorts with more than 2000 incident HF events.5 Our principal findings were as follows: natriuretic peptides and hs-Tn were more strongly associated with HFrEF compared with HFpEF, UACR was associated with both HFpEF and HFrEF, fewer biomarkers were associated with HFpEF compared with HFrEF, and the incremental value of individual biomarkers when added to a robust clinical model was modest. Whereas HFpEF and HFrEF have a shared clinical presentation and common clinical risk factors, these findings suggest that some antecedent factors preceding HFpEF and HFrEF may be distinct. For example, myocyte necrosis (hs-Tn) appears to play a larger role in the development of HFrEF, whereas low-grade albuminuria, a marker of endothelial dysfunction, seems to precede both HFpEF and HFrEF. Overall, our findings also suggest that traditional cardiovascular biomarkers are associated with HFrEF more so than HFpEF, highlighting the need for further studies focused on elucidating drivers of HFpEF.

Heart failure is expected to become even more prevalent in the decades to come,1,2 a trend that will undoubtedly result in growing morbidity, mortality, and associated costs.2 This trend underlines the need for HF prevention strategies in light of major challenges of treating clinically overt HF.1,2 An integral component of disease prevention is a refined understanding of risk estimation. Numerous publications of individual population-based cohorts have examined the association of biomarkers with cardiovascular end points, including new-onset HF.15-30 By far, the strongest evidence exists for natriuretic peptides: individual studies have consistently shown that natriuretic peptides are among the strongest associations with new-onset HF, which was recently confirmed for the estimation of new-onset HF in an individual patient data meta-analysis with more than 95 000 participants and more than 2000 HF events.30 Therefore, in recent years, several trials were launched that reported the feasibility of acting on elevated levels of natriuretic peptides, which appeared to lower the risk of new onset-HF.3,4 A downside of this potential strategy is that a sizeable number of participants will need to be treated; therefore, strategies to further enrich risk estimation to target biomarker-based prevention strategies have been put forward.31 Furthermore, the usefulness of other biomarkers for the estimation of new-onset HF remains largely unclear,32 with previous studies being limited by modest power and by a limited set of biomarkers.

Although previous community-based studies have focused largely on biomarker associations with overall HF,15-18,20,21,27,28,30 our unique multicohort collaboration enabled an initial look at specific biomarkers preceding HFpEF vs HFrEF. Interestingly, the association of biomarkers with risk of HFpEF was clearly less convincing than it was for risk of HFrEF. Likely, the pathophysiology of HFrEF is more easily captured by biomarkers of stretch (NT-proBNP), ischemia (hs-Tn), and inflammation (C-reactive protein). Markers of subclinical renal disease and endothelial dysfunction (indicated by cystatin C and UACR) and subclinical vascular disease (indicated by fibrinogen and D-dimer) were associated with new-onset HFrEF. Whether strategies aimed at targeting these specific pathways may prevent HFrEF remains unknown.

By contrast, only natriuretic peptides and UACR were associated with HFpEF, with suggestive associations for plasminogen activator inhibitor 1, hs-Tn, and fibrinogen. The fact that single biomarker associations overall appeared less powerful at estimating HFpEF highlights the complexity of HFpEF, which is thought to consist of many subphenotypes.33,34 Natriuretic peptides were associated with HFpEF, albeit with a smaller effect size compared with the association with HFrEF. This finding is recapitulated in clinical HF studies, in which natriuretic peptide levels are observed to be lower in HFpEF vs HFrEF.35 We observed a strong association of UACR with both HF subtypes, consistent with previous studies, which have demonstrated the association of UACR as a marker of endothelial dysfunction with adverse cardiac mechanics, including diastolic function.36 The association between cystatin C and HF subtypes was less clear, underscoring that estimated glomerular filtration rate and albuminuria assess distinct components of kidney disease. In a treatment trial of PREVEND, the PREVEND IT (Prevention of Renal and Vascular End-stage Disease Intervention Trial) study, treatment of participants with microalbuminuria but without other signs of overt cardiovascular disease resulted in a reduction of cardiovascular endpoints over the course of 4 to 8 years.9,37,38 The association of UACR and new-onset HF in individuals with impaired glucose tolerance has been reported,39 and we extend this association to community-based participants. The interplay between UACR, associated vascular disease, and endothelial dysfunction has been studied in HFpEF.40 However, the exact association between renal disease, albuminuria, estimated glomerular filtration rate, and HF is complex.41

The cohort-specific analyses revealed several interesting findings. First, when we considered the direction of associations between the markers and new-onset HF, we observed that the direction of the association was consistent among cohorts, adding robustness to our analyses. However, there were clear numerical differences in the effect sizes between the cohorts. The largest differences were observed when comparing FHS and MESA on one hand and PREVEND and CHS on the other hand. In CHS, participants were much older, with a higher prevalence of comorbidities, which possibly resulted in more complex multifactorial HF phenotypes. Indeed, the associations with biomarkers and HF tended to be weaker in CHS, although the direction of the associations again was consistent with the other cohorts. This finding might imply that, in elderly participants with comorbidities, biomarkers are less strongly associated with HF subtypes. In PREVEND, on the other hand, the youngest participants were enrolled with the lowest event rates. A previous study reported that biomarkers are more strongly associated with HF subtypes in participants who have higher (clinical) risk, whereas the association with HF subtypes in a very low-risk population appears less strong.21 We conclude that the value of biomarkers, as with many diagnostic tests, appears best in the population with intermediate risk.

Limitations

Several limitations deserve mention. We were limited by the available biomarkers in each cohort, and all markers were not available in all cohorts. However, the consistency among cohorts in associations leads us to believe that the biomarkers were measured reliably. Although the availability of 12 biomarkers in a large sample of community-dwelling individuals is a notable strength of our study, several biomarkers previously associated with incident HF were not available for the present analyses. Future studies incorporating additional biomarkers are warranted. The duration between enrollment in the study and the incident HF event was variable; it is well known that the association of biomarkers with HF subtypes may be altered over prolonged periods. In our pooled data set, we defined HFpEF and HFrEF only in individuals who underwent LV function assessment at or around the time of HF presentation, leaving 30% of cases as unclassified HF. We have only a one-time biomarker measurement, while it has been acknowledged that serial sampling over time may be superior in estimation of HF.42 Biomarker measurements should always be considered in light of their natural variability (assays and individual-dependent); for most biomarkers, such data are available.43 Last, the LVEF cutpoint that we used to distinguish between HFrEF and HFpEF was 50% and may be debated, although previous sensitivity analyses demonstrate only minor differences when using an LVEF cutpoint of more than 45%,5 and those with mid-range LVEF between 40% and 50% are largely of intermediate phenotype between HFpEF and HFrEF.

Conclusions

We demonstrate that markers of neurohormonal activation and myocyte necrosis appear to be more strongly associated with HFrEF compared with HFpEF in a unique international collaboration of 4 community-based longitudinal cohorts. Furthermore, biomarkers of renal dysfunction, endothelial dysfunction, and inflammation were associated with incident HFrEF. By contrast, only natriuretic peptides and UACR were associated with HFpEF. In general, biomarkers modestly improved risk estimation, and discrimination metrics overall were lower for HFpEF models, highlighting current limitations in our understanding of factors underlying the development of HFpEF. Although some studies demonstrate the potential utility of biomarker-guided prevention strategies,3,4 nuances in antecedent factors differentiating HFpEF and HFrEF highlight the need for future studies to examine the role of biomarkers in HF subtype-specific risk estimation.

Back to top
Article Information

Accepted for Publication: November 15, 2017.

Corresponding Author: Jennifer E. Ho, MD, Cardiovascular Research Center, Department of Medicine, Massachusetts General Hospital, 185 Cambridge St, Simches Research Bldg 3192, Boston, MA 02114 (jho1@mgh.harvard.edu).

Published Online: January 10, 2018. doi:10.1001/jamacardio.2017.4987

Author Contributions: Drs de Boer, Nayor, deFilippi, Enserro, van Gilst, Gottdiener, Bertoni, and Ho contributed equally to this work. Dr Ho had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: de Boer, Nayor, Enserro, Blaha, Brouwers, Lima, Bahrami, Psaty, Januzzi, Shah, Levy, Larson, van Gilst, Ho.

Acquisition, analysis, or interpretation of data: de Boer, Nayor, deFilippi, Enserro, Bhambhani, Kizer, Brouwers, Cushman, Lima, Bahrami, van der Harst, Wang, Gansevoort, Gaggin, Kop, Liu, Vasan, Lee, Hillege, Bartz, Benjamin, Chan, Allison, Gardin, Shah, Levy, Herrington, van Gilst, Bertoni, Ho.

Drafting of the manuscript: de Boer, Nayor, Enserro, Januzzi, Ho.

Critical revision of the manuscript for important intellectual content: Nayor, deFilippi, Enserro, Bhambhani, Kizer, Blaha, Brouwers, Cushman, Lima, Bahrami, van der Harst, Wang, Gansevoort, Fox, Gaggin, Kop, Liu, Vasan, Psaty, Lee, Hillege, Bartz, Benjamin, Chan, Allison, Gardin, Januzzi, Shah, Levy, Herrington, Larson, van Gilst, Bertoni, Ho.

Statistical analysis: de Boer, Nayor, Enserro, Bhambhani, Bahrami, Ho.

Obtained funding: de Boer, van der Harst, Gansevoort, Psaty, Levy, van Gilst, Ho.

Administrative, technical, or material support: Brouwers, Cushman, Lima, van der Harst, Gansevoort, Lee, Hillege, Bartz, Chan, Ho.

Study supervision: de Boer, Blaha, Brouwers, Cushman, van der Harst, Benjamin, Allison, Januzzi, Shah, Larson, van Gilst, Ho.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr de Boer reported receiving grants from AstraZeneca, Bristol-Meyers Squibb, and Trevena; serving as a consultant for and receiving research and/or personal honoraria from Roche; and serving as a consultant for and receiving research and/or personal honoraria from Novartis. Dr DeFilippi reported receiving research support from Roche Diagnostics; receiving consulting fees from Roche, Siemens Healthcare Diagnostics, Alere, Metanomics, and Ortho Diagnostics; serving on the end-point committee for Radiometer and Quintiles; and receiving royalties from UpToDate. Dr Kizer reported owning stock in Pfizer, Gilead Sciences, and Amgen. Dr Blaha reported receiving grants and personal fees from Amgen and the US Food and Drug Administration, grants from Aetna, the American Heart Association, the National Institutes of Health; and personal fees from Sanofi, Regeneron, Novartis, and MedImmune. Dr Gaggin reported receiving grants from Roche and Portola and personal fees from Roche Diagnostics, Amgen, Ortho Clinical, EchoSense, and Radiometer. Dr Psaty reported serving on a data and safety monitoring board for a clinical trial funded by the manufacturer (Zoll LifeCor) and serving on the steering committee of the Yale Open Data Access Project funded by Johnson & Johnson. Ms. Bartz reported receiving grants from the National Institutes of Health. Dr Benjamin reported receiving grants from the National Institutes of Health and from the American Heart Association/National Institutes of Health. Dr Januzzi reported receiving grants and personal fees from Roche, Abbott, Singulex, and Novartis and personal fees from Critical Diagnostics, Janssen, Boehringer-Ingelheim, Abbvie, Pfizer, GE, and Bayer. Dr Herrington reported receiving grants from the National Institutes of Health. Dr Bertoni reported receiving grants from the National Institutes of Health/National Heart, Lung, and Blood Institute. Dr Ho reported receiving grants from the National Institutes of Health and Massachusetts General Hospital. No other disclosures were reported.

Funding/Support: This work was partially supported by the National Heart, Lung, and Blood Institute (Framingham Heart Study, contract N01-HC25195 and HHSN268201500001I; Cardiovascular Health Study, contracts HHSN268201200036C, HHSN268200800007C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and U01HL130114 and grant U01HL080295). The Multi-Ethnic Study of Atherosclerosis and the Multi-Ethnic Study of Atherosclerosis National Institutes of Health SNP Health Association Resource project are conducted and supported by the National Heart, Lung, and Blood Institute in collaboration with the Multi-Ethnic Study of Atherosclerosis investigators. Support for the Multi-Ethnic Study of Atherosclerosis is provided by contracts HHSN268201500003I, N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, N01-HC-95169, UL1-TR-000040, UL1-TR-001079, UL1-TR-001420, UL1-TR-001881, and DK063491. Funding support for the Multi-Ethnic Study of Atherosclerosis Renal Function data set was provided by grant DK083538-01. The Cardiovascular Health Study received additional contributions from the National Institute of Neurological Disorders and Stroke and grant R01AG023629 from the National Institute on Aging. A full list of principal Cardiovascular Health Study investigators and institutions can be found at https://chs-nhlbi.org/. A full list of participating Multi-Ethnic Study of Atherosclerosis Renal Function investigators and institutions can be found at https://www.mesa-nhlbi.org. The Prevention of Renal and Vascular End-Stage Disease study has been made possible by grants from the Dutch Kidney Foundation. Dr de Boer is supported by the Netherlands Heart Foundation (CVON-DOSIS, grant 2014-40) and the Innovational Research Incentives Scheme program of the Netherlands Organization for Scientific Research (NWO VIDI, grant 917.13.350). Dr Nayor received support from the Clinical Skills Development Core Training grant U10HL110337 from the National Heart, Lung, and Blood Institute and by grant K23-HL138260. Dr Ho is supported by grants K23-HL116780 and R01 HL134893 and a Hassenfeld Research Scholar Award (Massachusetts General Hospital, Boston). Dr Lee is supported by a clinician-scientist award from the Canadian Institutes of Health Research. In Framingham Heart Study samples, measurement of soluble suppressor of tumorigenicity was performed by Critical Diagnostics, Inc, and measurement of high-sensitivity troponin I was performed by Singulex, Inc.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Additional Contributions: Dr Shah is an Associate Editor of JAMA Cardiology but did not participate in any discussions or decisions regarding publication of this manuscript.

References
1.
Lloyd-Jones  DM, Larson  MG, Leip  EP,  et al; Framingham Heart Study.  Lifetime risk for developing congestive heart failure: the Framingham Heart Study.  Circulation. 2002;106(24):3068-3072.PubMedGoogle ScholarCrossref
2.
Ponikowski  P, Anker  SD, AlHabib  KF,  et al.  Heart failure: preventing disease and death worldwide.  ESC Heart Fail. 2014;1(1):4-25.PubMedGoogle ScholarCrossref
3.
Huelsmann  M, Neuhold  S, Resl  M,  et al.  PONTIAC (NT-proBNP selected prevention of cardiac events in a population of diabetic patients without a history of cardiac disease): a prospective randomized controlled trial.  J Am Coll Cardiol. 2013;62(15):1365-1372.PubMedGoogle ScholarCrossref
4.
Ledwidge  M, Gallagher  J, Conlon  C,  et al.  Natriuretic peptide–based screening and collaborative care for heart failure: the STOP-HF randomized trial.  JAMA. 2013;310(1):66-74.PubMedGoogle ScholarCrossref
5.
Ho  JE, Enserro  D, Brouwers  FP,  et al.  Predicting heart failure with preserved and reduced ejection fraction: the international collaboration on heart failure subtypes.  Circ Heart Fail. 2016;9(6):e003116.PubMedGoogle ScholarCrossref
6.
Dawber  TR, Kannel  WB, Lyell  LP.  An approach to longitudinal studies in a community: the Framingham Study.  Ann N Y Acad Sci. 1963;107:539-556.PubMedGoogle ScholarCrossref
7.
Kannel  WB, Feinleib  M, McNamara  PM, Garrison  RJ, Castelli  WP.  An investigation of coronary heart disease in families: the Framingham Offspring Study.  Am J Epidemiol. 1979;110(3):281-290.PubMedGoogle ScholarCrossref
8.
Psaty  BM, Kuller  LH, Bild  D,  et al.  Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study.  Ann Epidemiol. 1995;5(4):270-277.PubMedGoogle ScholarCrossref
9.
Hillege  HL, Fidler  V, Diercks  GF,  et al; Prevention of Renal and Vascular End Stage Disease (PREVEND) Study Group.  Urinary albumin excretion predicts cardiovascular and noncardiovascular mortality in general population.  Circulation. 2002;106(14):1777-1782.PubMedGoogle ScholarCrossref
10.
Bild  DE, Bluemke  DA, Burke  GL,  et al.  Multi-Ethnic Study of Atherosclerosis: objectives and design.  Am J Epidemiol. 2002;156(9):871-881.PubMedGoogle ScholarCrossref
11.
Paulus  WJ, Tschöpe  C.  A novel paradigm for heart failure with preserved ejection fraction: comorbidities drive myocardial dysfunction and remodeling through coronary microvascular endothelial inflammation.  J Am Coll Cardiol. 2013;62(4):263-271.PubMedGoogle ScholarCrossref
12.
Fine  JP, Gray  RJ.  A proportional hazards model for the subdistribution of a competing risk.  J Am Stat Assoc. 1999;94(446):496-509. doi:10.2307/2670170Google ScholarCrossref
13.
Lunn  M, McNeil  D.  Applying Cox regression to competing risks.  Biometrics. 1995;51(2):524-532.PubMedGoogle ScholarCrossref
14.
Pencina  MJ, D’Agostino  RB  Sr, Steyerberg  EW.  Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers.  Stat Med. 2011;30(1):11-21.PubMedGoogle ScholarCrossref
15.
Wang  TJ, Wollert  KC, Larson  MG,  et al.  Prognostic utility of novel biomarkers of cardiovascular stress: the Framingham Heart Study.  Circulation. 2012;126(13):1596-1604.PubMedGoogle ScholarCrossref
16.
Vasan  RS, Sullivan  LM, Roubenoff  R,  et al; Framingham Heart Study.  Inflammatory markers and risk of heart failure in elderly subjects without prior myocardial infarction: the Framingham Heart Study.  Circulation. 2003;107(11):1486-1491.PubMedGoogle ScholarCrossref
17.
Frankel  DS, Vasan  RS, D’Agostino  RB  Sr,  et al.  Resistin, adiponectin, and risk of heart failure: the Framingham Offspring Study.  J Am Coll Cardiol. 2009;53(9):754-762.PubMedGoogle ScholarCrossref
18.
Ho  JE, Liu  C, Lyass  A,  et al.  Galectin-3, a marker of cardiac fibrosis, predicts incident heart failure in the community.  J Am Coll Cardiol. 2012;60(14):1249-1256.PubMedGoogle ScholarCrossref
19.
Ho  JE, Lyass  A, Lee  DS,  et al.  Predictors of new-onset heart failure: differences in preserved versus reduced ejection fraction.  Circ Heart Fail. 2013;6(2):279-286.PubMedGoogle ScholarCrossref
20.
Xanthakis  V, Larson  MG, Wollert  KC,  et al.  Association of novel biomarkers of cardiovascular stress with left ventricular hypertrophy and dysfunction: implications for screening.  J Am Heart Assoc. 2013;2(6):e000399.PubMedGoogle ScholarCrossref
21.
Brouwers  FP, van Gilst  WH, Damman  K,  et al.  Clinical risk stratification optimizes value of biomarkers to predict new-onset heart failure in a community-based cohort.  Circ Heart Fail. 2014;7(5):723-731.PubMedGoogle ScholarCrossref
22.
Brouwers  FP, de Boer  RA, van der Harst  P,  et al.  Incidence and epidemiology of new onset heart failure with preserved vs reduced ejection fraction in a community-based cohort: 11-year follow-up of PREVEND.  Eur Heart J. 2013;34(19):1424-1431.PubMedGoogle ScholarCrossref
23.
Agarwal  SK, Chambless  LE, Ballantyne  CM,  et al.  Prediction of incident heart failure in general practice: the Atherosclerosis Risk in Communities (ARIC) Study.  Circ Heart Fail. 2012;5(4):422-429.PubMedGoogle ScholarCrossref
24.
Seliger  SL, de Lemos  J, Neeland  IJ,  et al.  Older adults, “malignant” left ventricular hypertrophy, and associated cardiac-specific biomarker phenotypes to identify the differential risk of new-onset reduced versus preserved ejection fraction heart failure: CHS (Cardiovascular Health Study).  JACC Heart Fail. 2015;3(6):445-455.PubMedGoogle ScholarCrossref
25.
Kalogeropoulos  A, Psaty  BM, Vasan  RS,  et al; Cardiovascular Health Study.  Validation of the Health ABC heart failure model for incident heart failure risk prediction: the Cardiovascular Health Study.  Circ Heart Fail. 2010;3(4):495-502.PubMedGoogle ScholarCrossref
26.
Bahrami  H, Kronmal  R, Bluemke  DA,  et al.  Differences in the incidence of congestive heart failure by ethnicity: the Multi-Ethnic Study of Atherosclerosis.  Arch Intern Med. 2008;168(19):2138-2145.PubMedGoogle ScholarCrossref
27.
Bahrami  H, Bluemke  DA, Kronmal  R,  et al.  Novel metabolic risk factors for incident heart failure and their relationship with obesity: the MESA (Multi-Ethnic Study of Atherosclerosis) Study.  J Am Coll Cardiol. 2008;51(18):1775-1783.PubMedGoogle ScholarCrossref
28.
AbouEzzeddine  OF, McKie  PM, Scott  CG,  et al.  Biomarker-based risk prediction in the community.  Eur J Heart Fail. 2016;18(11):1342-1350.PubMedGoogle ScholarCrossref
29.
Echouffo-Tcheugui  JB, Greene  SJ, Papadimitriou  L,  et al.  Population risk prediction models for incident heart failure: a systematic review.  Circ Heart Fail. 2015;8(3):438-447.PubMedGoogle ScholarCrossref
30.
Willeit  P, Kaptoge  S, Welsh  P,  et al; Natriuretic Peptides Studies Collaboration.  Natriuretic peptides and integrated risk assessment for cardiovascular disease: an individual-participant-data meta-analysis.  Lancet Diabetes Endocrinol. 2016;4(10):840-849.PubMedGoogle ScholarCrossref
31.
Meijers  WC, de Boer  RA, Ho  JE.  Biomarkers to identify and prevent new-onset heart failure in the community.  Eur J Heart Fail. 2016;18(11):1351-1352.PubMedGoogle ScholarCrossref
32.
de Boer  RA, Daniels  LB, Maisel  AS, Januzzi  JL  Jr.  State of the art: newer biomarkers in heart failure.  Eur J Heart Fail. 2015;17(6):559-569.PubMedGoogle ScholarCrossref
33.
Shah  SJ, Katz  DH, Selvaraj  S,  et al.  Phenomapping for novel classification of heart failure with preserved ejection fraction.  Circulation. 2015;131(3):269-279.PubMedGoogle ScholarCrossref
34.
Senni  M, Paulus  WJ, Gavazzi  A,  et al.  New strategies for heart failure with preserved ejection fraction: the importance of targeted therapies for heart failure phenotypes.  Eur Heart J. 2014;35(40):2797-2815.PubMedGoogle ScholarCrossref
35.
Kitzman  DW, Little  WC, Brubaker  PH,  et al.  Pathophysiological characterization of isolated diastolic heart failure in comparison to systolic heart failure.  JAMA. 2002;288(17):2144-2150.PubMedGoogle ScholarCrossref
36.
Katz  DH, Selvaraj  S, Aguilar  FG,  et al.  Association of low-grade albuminuria with adverse cardiac mechanics: findings from the Hypertension Genetic Epidemiology Network (HyperGEN) Study.  Circulation. 2014;129(1):42-50.PubMedGoogle ScholarCrossref
37.
Brouwers  FP, Asselbergs  FW, Hillege  HL,  et al.  Long-term effects of fosinopril and pravastatin on cardiovascular events in subjects with microalbuminuria: ten years of follow-up of Prevention of Renal and Vascular End-stage Disease Intervention Trial (PREVEND IT).  Am Heart J. 2011;161(6):1171-1178.PubMedGoogle ScholarCrossref
38.
Asselbergs  FW, Diercks  GF, Hillege  HL,  et al; Prevention of Renal and Vascular Endstage Disease Intervention Trial (PREVEND IT) Investigators.  Effects of fosinopril and pravastatin on cardiovascular events in subjects with microalbuminuria.  Circulation. 2004;110(18):2809-2816.PubMedGoogle ScholarCrossref
39.
Wong  YW, Thomas  L, Sun  JL,  et al.  Predictors of incident heart failure hospitalizations among patients with impaired glucose tolerance: insight from the Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research Study.  Circ Heart Fail. 2013;6(2):203-210.PubMedGoogle ScholarCrossref
40.
Katz  DH, Burns  JA, Aguilar  FG, Beussink  L, Shah  SJ.  Albuminuria is independently associated with cardiac remodeling, abnormal right and left ventricular function, and worse outcomes in heart failure with preserved ejection fraction.  JACC Heart Fail. 2014;2(6):586-596.PubMedGoogle ScholarCrossref
41.
Ter Maaten  JM, Damman  K, Verhaar  MC,  et al.  Connecting heart failure with preserved ejection fraction and renal dysfunction: the role of endothelial dysfunction and inflammation.  Eur J Heart Fail. 2016;18(6):588-598.PubMedGoogle ScholarCrossref
42.
van der Velde  AR, Meijers  WC, Ho  JE,  et al.  Serial galectin-3 and future cardiovascular disease in the general population.  Heart. 2016;102(14):1134-1141.PubMedGoogle ScholarCrossref
43.
Meijers  WC, van der Velde  AR, Muller Kobold  AC,  et al.  Variability of biomarkers in patients with chronic heart failure and healthy controls.  Eur J Heart Fail. 2017;19(3):357-365.PubMedGoogle ScholarCrossref
×