Serum Levels of the Interleukin-1 Receptor Family Member ST2, Cardiac Structure and Function, and Long-Term Mortality in Patients With Acute DyspneaCLINICAL PERSPECTIVE
Background— ST2, a biomarker of cardiomyocyte stretch, powerfully predicts poor outcomes in patients with acute dyspnea, but nothing is known about associations between soluble ST2 (sST2) and cardiac structure and function, or whether sST2 retains prognostic meaning in the context of such measures.
Methods and Results— One hundred thirty-four dyspneic patients with and without decompensated heart failure had echocardiography during index admission and vital status was ascertained at 4 years. Echocardiographic and clinical correlates of sST2 as well as independent predictors of death at 4 years were identified. sST2 correlated with left ventricular end-systolic dimensions/volumes and left ventricular ejection fraction. sST2 was inversely associated with right ventricular fractional area change (ρ=−0.18; P=0.046), higher right ventricular systolic pressure (ρ=0.26; P=0.005), and right ventricular hypokinesis (P<0.001) and was correlated with tissue Doppler Ea wave peak velocity, but not to other indices of diastolic function. In multivariate regression, independent predictors of sST2 included right ventricular systolic pressure (t=2.29; P=0.002), left ventricular ejection fraction (t=−2.15; P=0.05) and dimensions (end systolic, t=2.57; end diastolic, t=2.98; both P<0.05), amino-terminal pro-B-type natriuretic peptide (t=3.31; P=0.009), heart rate (t=2.59; P=0.01), and presence of jugular venous distension (t=2.00; P=0.05). In a Cox proportional hazards model that included echocardiographic results and other biomarkers, sST2 independently predicted death at 4 years (hazard ratio=2.70; P=0.003).
Conclusions— Among dyspneic patients with and without acute heart failure, sST2 concentrations are associated with prevalent cardiac abnormalities on echocardiography, a more decompensated hemodynamic profile and are associated with long-term mortality, independent of echocardiographic, clinical, or other biochemical markers of risk.
Received November 4, 2008; accepted April 23, 2009.
In patients with acute heart failure (HF), elevated filling pressures and ventricular dysfunction contribute to increased wall stress and promote the synthesis and release of neurohormones, including the natriuretic peptides (eg, B-type natriuretic peptide [BNP] and its amino-terminal cleavage fragment [NT-proBNP]). These biomarkers, induced by pressure or volume overload,1,2 may be associated with the presence and severity of abnormalities in cardiovascular structure and function, such as poorer ventricular systolic function, abnormalities in ventricular nonsystolic function, and dilatation of cardiac chambers, abnormalities in heart valves, and pulmonary hypertension.3,4 Furthermore, natriuretic peptides have been shown to predict long-term mortality and readmission for HF independent of echocardiographic parameters.4–9
Clinical Perspective on p 311
A member of the interleukin (IL)-1 receptor family, ST2 is a novel biomarker of mechanical stress that is measurable in serum. ST2 expression is upregulated in isolated cardiomyocytes exposed to mechanical strain,10 and in patients with acute myocardial infarction, elevated serum level of soluble ST2 (sST2) is associated with an increased risk of mortality or HF, independent of BNP.11,12 sST2 levels are strongly correlated with BNP in patients with chronic advanced HF, and serial alterations in sST2 level over time predict outcomes independent of natriuretic peptides.13 Furthermore, in acute decompensated HF, our group and others have demonstrated that sST2 levels strongly associate with presence and severity of HF, while forecasting 1-year mortality in an additive manner to NT-proBNP.14–16
Although the clinical and prognostic correlates of sST2 concentrations have been defined, nothing is known regarding correlates of cardiac structure and function and sST2 values, and it remains yet unclear if sST2 remains prognostically relevant in those with echocardiographic data. We now investigate the associations among sST2 levels, cardiac structure and function (as assessed with detailed echocardiography studies), and outcomes among patients with acute dyspnea enrolled in the Pro-BNP Investigation of Dyspnea in the Emergency Department study.17
The design and main results of the Pro-BNP Investigation of Dyspnea in the Emergency Department study are described elsewhere.17 In brief, the Pro-BNP Investigation of Dyspnea in the Emergency Department study was a prospective, blinded study of 599 dyspneic patients presenting to the emergency department of the Massachusetts General Hospital examining the use of NT-proBNP testing compared with clinical judgment for the diagnosis of acute HF. This study population involved 139 subjects from the overall cohort who underwent detailed echocardiographic examinations during index admission, as described in Ref. 4. As described in our previous work,4 adjudicated diagnosis of the cause of dyspnea was determined by study physicians blinded to natriuretic peptide results, using hospital records from 60 days after presentation and a 60-day follow-up call. Echocardiography was determined to be indicated by the treating physicians. Patients were followed up for vital status at 4 years; follow-up for outcomes was available in 100% of the present cohort. Patient outcomes follow-up with respect to mortality was assessed via Social Security Death Index database search.
Transthoracic echocardiography was performed during the index hospitalization with standard techniques as described.4 The median time between hospital contact and echocardiography in these patients was 45 hours (interquartile range, 23 to 73 hours). Off-line analysis was performed by observers blinded to the patients’ clinical and biomarker data. Structural indices assessed on echocardiography included left ventricular (LV) end-diastolic and end-systolic volumes, LV wall thickness, LV mass, left atrial volume index, and right ventricular (RV) end-diastolic and end-systolic areas. LV ejection fraction was determined using biplane modified Simpson’s measurements. Markers of diastolic function included early and late transmitral diastolic velocities (E and A); early deceleration time, and early and late diastolic tissue Doppler velocities at the lateral mitral annulus (Ea and Aa). Indices of RV function included RV fractional area change, the presence or absence of RV hypokinesis or dilation by visual assessment, and tricuspid regurgitation velocity. Mitral regurgitation and tricuspid regurgitation severity were graded from 0 (none) to 4 (severe) on the basis of visual assessment of structural and Doppler parameters as described.4 Measurements were averaged over 3 cycles (5 if atrial fibrillation was present). As all patients did not have all echocardiographic measurements collected (eg, only 101 patients had full pulmonary Doppler information), analyses were based on echocardiographic data available.
Blood collected at the time of presentation was analyzed for concentrations of sST2 with an enzyme-linked immunosorbent assay (Medical and Biological Laboratories Company, Woburn, Mass) with characteristics described previously.18
sST2 results were log-transformed to establish normality, and correlations between sST2 and continuous echocardiographic variables as well as NT-proBNP were evaluated with Spearman regression. Multivariate linear regression analyses were performed with log-transformed sST2 concentrations as the dependent variable, including all echocardiographic, clinical and serological biomarkers of HF, and final diagnosis (HF versus other cause) with a P value of <0.10. Comparison of clinical characteristics between patients who were alive and deceased at 4 years after index admission were performed with χ2 tests for categorical data and via Wilcoxon rank-sum tests for nonnormally distributed continuous variables. A Cox proportional hazards model was used to evaluate independent predictors of 4-year mortality, including log-transformed NT-pro-BNP and log-transformed sST2 level as covariates. Variables entered into the mortality model included those associated with death at 4 years in univariable screening with a P value of <0.10 for retention; for all statistical analyses, a P value of 0.05 was considered significant. To assess the additive importance of sST2 relative to other risk predictors, the C-statistic for mortality was assessed using the method of Harrell. All statistics were performed with SPSS (Chicago, Ill) or STATA (College Station, Tex) software.
Baseline Echocardiographic and Clinical Characteristics
Baseline echocardiographic and clinical characteristics have been described previously4,15 and are detailed in Tables 1 and 2⇓. In brief, the study population was an elderly (69±14 years), overweight (body mass index 28±7 kg/m2) cohort evenly divided by gender with a history of hypertension (61%), coronary artery disease (31%), HF (37%), obstructive pulmonary disease (27%), and preserved renal function. Acute HF was the cause of dyspnea in 66%.
Overall, LV function was at the lower limit of normal (median LV ejection fraction, 53%; with interquartile range, 39% to 61%), with RV hypokinesis in 21%, mild pulmonary hypertension (median RV systolic pressure, 43 mm Hg), and preserved LV diastolic filling parameters (median tissue Doppler E and A wave peak velocities each 8.4 cm/s).
Characteristics of patients as a function of diagnosis (non-HF versus HF) as well as the presence or absence of impaired LV function are detailed in supplemental Tables I and II.
The median concentration of sST2 for the group as a whole was 0.45 ng/mL (interquartile range=0.19 to 1.03 ng/mL); those with acute HF had higher sST2 values compared with those without (0.53 [0.29 to 1.38] versus 0.20 [0.12 to 0.48]; P<0.001). Of note, the median sST2 value of patients who had echocardiography was higher than those who did not (0.45 [0.19 to 1.03] versus 0.23 [0.12 to 0.76] ng/mL; P=0.001).
Echocardiographic Correlates of sST2 Concentration
Spearman correlation analysis of echocardiographic indices and sST2 concentration is shown in Table 3. The sST2 concentration was significantly correlated with higher LV end-systolic area and volume (in the 4-chamber view) and end-systolic dimension, but not with left atrial dimension or volume. The sST2 was inversely related to LV ejection fraction (Figure 1) and RV fractional area change; sST2 was also associated with a higher RV systolic pressure (Figure 2A), more severe tricuspid regurgitation, and a higher frequency of RV hypokinesis (Figure 2B and 2C). With respect to LV diastolic function, sST2 was negatively correlated with tissue Doppler E wave peak velocity but not with other traditional markers of diastolic dysfunction.
In multivariate linear regression (Table 4), independent predictors of sST2 concentrations included RV systolic pressure (t=2.29; P=0.002), LV ejection fraction (t=2.15; P=0.05) and dimensions (end systolic, t=2.57; P=0.01; end diastolic, t=2.98; P=0.005), log-transformed NT-proBNP (t=3.31; P=0.009), heart rate (t=2.59; P=0.01), and presence of jugular venous distension (t=2.00; P=0.05).
Predictors of Long-Term Mortality in Patients With Acute Dyspnea
Clinical and echocardiographic characteristics stratified by vital status at 4 years after index admission are shown in Tables 5 and 6⇓. Compared with survivors, patients deceased at 4 years’ follow-up were older, more likely to have a history of HF, had a lower body-mass index, and were more likely to use loop diuretics or an angiotensin-converting enzyme inhibitor on presentation. In addition, decedents were more likely to have evidence of volume overload on admission chest radiography (pulmonary edema or pleural effusion), worse renal function, a lower hemoglobin concentration, and higher concentrations of sST2 and NT-proBNP. Furthermore, patients with higher sST2 levels (as stratified by quartile) had incrementally higher risk of death at 4 years (Figure 3).
With respect to echocardiographic findings, those dead at 4 years from enrollment had a higher RV systolic pressure. Differences in several other echocardiographic indices between deceased and living patients approached statistical significance, including wall thickness, nonsystolic function, and measures of RV function.
In a Cox proportional hazards analysis of 4-year mortality in patients with acute dyspnea (Table 7), log-transformed sST2 concentration remained independently predictive of mortality (hazard ratio=2.70; 95% CI, 2.00 to 2.98; P=0.003) even in the presence of echocardiographic results and other relevant covariates; in this context, addition of log-transformed sST2 concentration as a continuous variable to other markers of clinical risk into the model resulted in a significant increase in the C-statistic from 0.76 to 0.80.
As a marker of cardiomyocyte stretch, sST2 concentration has emerged as a potent prognostic marker in patients with acute dyspnea with and without HF, and seems to be prognostically meaningful even in the context of other markers of myocardial stretch, including NT-proBNP.15 A previous work by our group and others has identified elevated sST2 concentration in patients with acute dyspnea as predictive of mortality at a follow-up of 1 year.15,16 Although increases in ST2 gene expression during conditions of myocardial stretch has been demonstrated in vitro,10 there had been no clinical evidence linking sST2 to parameters of cardiac structure or function. In addition, the implications of sST2 on long-term mortality, as well as the impact of sST2 on prognosis independent of echocardiographic, clinical, and other serological biomarkers of risk remained unclear.
Our results demonstrate numerous cardiac structure and function predictors of sST2 concentrations, with independent associations between sST2 concentration and biventricular size and function (both systolic and nonsystolic), as well as RV systolic pressure in a population of patients with and without acute HF. We further demonstrate that an elevated sST2 concentration predicts mortality at 4 years, independent of clinical or echocardiographic indices or NT-proBNP concentration. Furthermore, the relationship of sST2 concentration with mortality is continuous, such that patients with incrementally higher sST2 levels among our cohort experience a correspondingly higher relative risk of mortality when compared with patients with lower values. Addition of sST2 concentration to a multivariate model of risk prediction significantly improved the C-statistic in a receiver-operator analysis. These data are novel, as they represent the first clinical association between sST2 and real-time assessment of cardiac structure and function and further establish the value of this interesting biomarker for prediction of hazard independent of other relevant clinical and biochemical predictors of risk.
From a hemodynamic perspective, patients with a poorer ventricular function or compliance typically have higher ventricular filling pressures and transmural wall stress, thereby increasing myocardial stretch and providing a stimulus for both dyspnea and sST2 release. In this setting, the strong associations between sST2 and LV and RV functions provide a clinical link for the proposed role of ST2 in the pathogenesis of HF; the biological role of ST2 in this context is speculative, but may relate to its role in cardiac fibrosis in response to pressure and volume overload.
Along these lines, it is known that ST2 is expressed both in a soluble isoform (measured in serum) and in a transmembrane receptor isoform for which IL-33 serves as a ligand.19 IL-33 signaling via the transmembrane ST2 receptor has been shown to limit LV hypertrophy in response to pressure overload in mice, and sST2 competitively inhibits the antihypertrophic action of IL-33.19 In an aortic constriction model of cardiac pressure overload, ST2-deficient mice experience a greater increase in LV hypertrophy and decrement in LV function, effects that could not be rescued by treatment with IL-33.19 Our association of a higher sST2 concentration with more advanced adverse biventricular remodeling (as reflected by poorer LV and RV functions, more ventricular dilatation, and pulmonary hypertension) is the first clinical evidence that supports a role for sST2 in cardiac fibrosis, adverse ventricular remodeling—the hallmark of HF in humans.
We and others have shown that elevation of sST2 is very strongly predictive of mortality in the setting of HF. Why excessive levels of sST2 (which would hypothetically have a beneficial role to respond to pressure or volume overload in concert with IL-33) are associated with such a powerfully negative prognosis remains speculative. It is theorized that excessive amounts of sST2 may act as a “decoy receptor,” binding IL-33 and preventing its binding to membrane bound ST2, leading to unchecked fibrosis, hypertrophy, and promoting a biological spiral promoting worse remodeling and a deleterious prognostic state.
In this analysis, sST2 concentration during index admission was significantly higher in patients who died at 4 years and was independently prognostic, even with the presence of detailed cardiac phenotyping and NT-proBNP in the model. Given that poor LV and RV functions and higher pulmonary arterial pressures are well-known markers of poor prognosis in patients with chronic advanced HF, it is possible that the prognostic power of sST2 may be partly explained via association with these variables. However, our data indicate that the relationship of elevated sST2 to mortality in patients with acute dyspnea remains after echocardiographic parameters of systolic function and pulmonary hypertension are taken into account, implying that the prognostic import of sST2 may extend beyond that imparted by abnormalities in cardiac structure and function. In addition, given that IL-1β, a proinflammatory cytokine, is expressed post-MI and induces ST2 mRNA expression in vitro,10 sST2 elevations in patients with poor ventricular function and dyspnea may represent a link between inflammation and ventricular injury and remodeling.
Our study has several important limitations. Although these results were derived from a small patient cohort, our study population was carefully and extensively phenotyped with respect to clinical, echocardiographic, and biochemical parameters. Given the limited number of patients in this study, relationships between sST2 and echocardiographic variables not necessarily definitive, should be viewed as hypothesis-generating only for study in larger patient cohorts, and need to be confirmed in such larger studies before definitive conclusions can be drawn. Echocardiography was obtained for clinical indications during the index admission for HF such that there may have been delays between collection of biomarkers and echocardiographic parameters for correlation. Because the biological half-life of sST2 in patients with acute dyspnea is not well understood, it is possible that correlations between echocardiography and sST2 would have been different if both measures were taken simultaneously.
In addition, we studied a subset of the initial Pro-BNP Investigation of Dyspnea in the Emergency Department cohort who received echocardiography for clinical indications, introducing referral bias into the study. However, our cohort had significantly higher sST2 level than patients who did not receive echocardiography, suggesting that our patients may have been sicker. Furthermore, sST2 still emerged as an independent predictor of mortality beyond echocardiographic, clinical, and other serological (eg, NT-proBNP) markers of risk. This suggests that even if patients were being selected for echocardiography based on an a priori estimate of higher clinical risk, measuring sST2 can add further to prognosis and risk stratification in these patients.
The small number of patients in this study limited our evaluation of associations between sST2 level and echocardiographic indices in patients with preserved versus depressed LV ejection fraction difficult. However, given sST2 release may be stimulated by mechanisms distinct from cardiomyocyte stretch (eg, inflammation and noncardiac comorbidities, such as pulmonary disease), variation in sST2 (and hence its relationship with other echocardiographic indices) is likely only partly determined by LV ejection fraction.
In conclusion, among patients with acute dyspnea, sST2 levels are associated with prevalent structural and functional abnormalities on echocardiographic analysis, as well as a more clinically decompensated hemodynamic profile. Higher levels of sST2 are associated with long-term mortality, independent of traditional echocardiographic, serological, or clinical biomarkers of risk. Our data—establishing cardiac structural and functional associations with sST2—provide valuable data extending the link between sST2 and cardiovascular disease.
We thank the echocardiography laboratory at the Massachusetts General Hospital.
Sources of Funding
This study was supported by grant from the National Heart, Lung, and Blood Institute Heart Failure Research Network, Boston, Mass (to R.V.S.); in part by the Balson Scholar Fund, Boston, Mass (to J.L.J.); and by a research fellowship of the Interuniversity Cardiology Institute of the Netherlands (ICIN) (to R.R.J.v.K.).
J.L.J. reports grant support, speaker’s fees, and/or consulting income from Roche Diagnostics, Siemens, Ortho Clinical Diagnostics, Critical Diagnostics, BG Medicine, and Inverness. The remaining authors report no conflicts.
Liang F, Gardner D. Mechanical strain activates BNP gene transcription through a p38/NF-κB-dependent mechanism. J Clin Invest. 1999; 101: 1603–1612.
Chen A, Wood M, Krauser D, Baggish A, Tung R, Anwaruddin S, Picard M, Januzzi JL. NT-proBNP levels, echocardiographic findings, and outcomes in breathless patients: results from the ProBNP Investigation of Dyspnea in the Emergency Department (PRIDE) echocardiographic substudy. Eur Heart J. 2006; 27: 839–845.
Januzzi J, van Kimmenade R, Lainchbury J, Bayes-Fenis A, Ordonez-Llanos J, Santalo-Bel M, Pinto Y, Richards M. NT-proBNP testing for diagnosis and short-term prognosis in acute destabilized heart failure: an international pooled analysis of 1256 patients. Eur Heart J. 2006; 27: 330–337.
Anand I, Fisher L, Chiang Y, Latini R, Masson S, Maggioni A, Glazer R, Tognoni G, Cohn J; Val-HeFT Investigators. Changes in brain natriuretic peptide and norepinephrine over time and mortality and morbidity in the Valsartan Heart Failure trial (Val-HeFT). Circulation. 2003; 107: 1278–1283.
Weinberg E, Shimpo M, De Keulenaer G, MacGillivray C, Tominaga S, Solomon S, Rouleau J, Lee R. Expression and regulation of ST2, an interleukin-1 receptor family member, in cardiomyocytes and myocardial infarction. Circulation. 2002; 106: 2961–2966.
Sabatine M, Morrow D, Higgins L, MacGillivray C, Guo W, Bode C, Rifai N, Cannon C, Gerzten R, Lee R. Complementary roles for biomarkers of biomechanical strain ST2 and N-terminal prohormone B-type natriuretic peptide in patients with ST-elevation myocardial infarction. Circulation. 2008; 117: 1936–1944.
Shimpo N, Morrow D, Weinberg E, Sabatine M, Murphy S, Antman E, Lee R. Serum levels of the interleukin-1 receptor family member ST2 predict mortality and clinical outcome in acute myocardial infarction. Circulation. 2004; 109: 2186–2190.
Weinberg E, Shimpo M, Hurwitz S, Tominaga S, Rouleau J, Lee R. Identification of serum soluble ST2 receptor as a novel heart failure biomarker. Circulation. 2003; 107: 721–726.
Januzzi JL, Peacock W, Maisel A, Chae C, Jesse R, Baggish A, O'Donoghue M, Sakuja R, Chen A, van Kimmenade R, Lewandrowski K, Lloyd-Jones D, Wu A. Measurement of the interleukin family member ST2 in patients with acute dyspnea: results from the PRIDE study. J Am Coll Cardiol. 2007; 50: 607–613.
Myeller T, Dieplinger B, Gegenhuber A, Poelz W, Pacher R, Haltmayer M. Increased plasma concentrations of soluble ST2 are predictive for 1-year mortality in patients with acute destabilized heart failure. Clin Chem. 2008; 54: 752–756.
Januzzi J, Camargo C, Anwaruddin S, Baggish A, Chen A, Krauser D, Tung R, Cameron R, Nagurney J, Chae C, Lloyd-Jones D, Brown D, Doran-Melanson S, Sluss P, Lee-Lewandrowski E, Lewandrowski K. The N-terminal Pro-BNP Investigation of Dyspnea in the Emergency Department (PRIDE) study. Am J Cardiol. 2005; 95: 948–954.
ST2, a member of the interleukin receptor family, is a novel biomarker of cardiomyocyte stretch. In animal models of heart failure, abnormalities of ST2 signaling are associated with left ventricular dysfunction, myocardial fibrosis, and chamber remodeling, as well as reduced survival in this setting. Moreover, concentrations of soluble ST2 (sST2) are strongly associated with reduced survival in acute heart failure in humans; however, mechanistic associations between sST2 concentrations and cardiac structure and function are unknown. Among a cohort of patients with acute dyspnea (with and without heart failure), comprehensive echocardiography was performed, and correlated with sST2 concentrations drawn on admission to the hospital. Higher sST2 levels were associated with larger left ventricular end-systolic dimensions/volumes, poorer left and right ventricular functions, and higher pulmonary arterial pressures as well as more evidence for congestion on examination. In a Cox proportional hazards model that adjusted for echocardiography, clinical presentation, and other biomarkers, sST2 concentrations independently predicted death at 4 years from enrollment. Therefore, among dyspneic patients with and without acute heart failure, sST2 concentrations are independently associated with prevalent cardiac abnormalities on echocardiography, a more decompensated hemodynamic profile and are associated with long-term mortality, independent of other predictors of risk.
The online-only Data Supplement is available at http://circheartfailure.ahajournals.org/cgi/content/full/CIRCHEARTFAILURE.108.833707.