Journal of the American College of Cardiology
Soluble ST2 for Predicting Sudden Cardiac Death in Patients With Chronic Heart Failure and Left Ventricular Systolic Dysfunction
Author + information
- Domingo A. Pascual-Figal, MD, PhD⁎,⁎ (dapascual{at}servicam.com),
- Jordi Ordoñez-Llanos, MD, PhD†,
- Pedro L. Tornel, PhD§,
- Rafael Vázquez, MD, PhD∥,
- Teresa Puig, MD, PhD‡,
- Mariano Valdés, MD, PhD⁎,
- Juan Cinca, MD, PhD‡,
- Antoni Bayes de Luna, MD, PhD¶,
- Antoni Bayes-Genis, MD, PhD‡,
- MUSIC Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Domingo A. Pascual-Figal, Heart Failure Unit, Cardiology Department, University Hospital Virgen de la Arrixaca Ctra Madrid-Cartagena s/n, 30120 Murcia, Spain
Abstract
Objectives We studied whether the measurement of the soluble form of ST2 (sST2), an interleukin-1 receptor family member, could identify heart failure (HF) patients at risk of sudden cardiac death (SCD).
Background The prediction of SCD remains an important challenge in patients with mild-to-moderate chronic HF. Concentrations of sST2 have been found increased and related to worse long-term outcomes in patients with acute HF. Whether sST2 has a prognostic role in SCD is unknown.
Methods A nested case-control study was performed on 36 cases of SCD and 63 control patients (matched for age, sex, and left ventricular ejection fraction) obtained from the MUSIC (MUerte Súbita en Insuficiencia Cardíaca) registry, a 3-year multicenter registry of ambulatory HF patients (New York Heart Association functional class II to III, left ventricular ejection fraction ≤45%). Demographic, clinical, echocardiographic, electrical, and biochemical data were collected at enrollment.
Results Concentrations of sST2 were greater among decedents (0.23 ng/ml [interquartile range 0.16 to 0.43 ng/ml] vs. 0.12 ng/ml [interquartile range 0.06 to 0.23 ng/ml], p = 0.001) and were predictive of experiencing SCD (+0.1 ng/ml, odds ratio: 1.39, 95% confidence interval: 1.09 to 1.78, p = 0.006). On the basis of a combined biomarker status, only 4% of patients experienced SCD for neither sST2 nor N-terminal pro–B-type natriuretic peptide (NT-proBNP) above receiver-operator characteristic-derived cut-off points (0.15 ng/ml and 2,000 ng/l, respectively), 34% for either biomarker above, and 71% for both biomarkers above (p < 0.001 for trend). This combined variable added incremental prognostic value to the multivariable regression model (p < 0.001).
Conclusions Elevated sST2 concentrations are predictive of SCD in patients with chronic HF and provide complementary information to NT-proBNP levels. A combined biomarker approach may have an impact on clinical decision-making.
The prediction and prevention of sudden cardiac death (SCD) remains a major challenge in the management of patients with chronic heart failure (HF). Sudden cardiac death is the cause of 50% of all HF deaths, mainly affecting patients with mild-to-moderate symptoms, and the number of patients with HF at risk is increasing (1). The challenge lies in identifying patients with HF who are at significant risk of SCD. Severe left ventricular (LV) systolic dysfunction has long been recognized as a risk factor for SCD, yet left ventricular ejection fraction (LVEF) lacks enough sensitivity and specificity by itself (2,3). Among electrical factors, there is evidence that prolonged QRS duration and the presence of nonsustained ventricular tachycardia may identify a subset of patients at greater risk of SCD (4,5). Natriuretic peptides also have been associated with a greater risk of SCD (6–9). The mechanism of SCD remains uncertain, and additional diagnostic tests to replace or complement currently existing tools are required for a better prediction of SCD in patients with HF (10,11).
ST2, an interleukin (IL)-1 receptor family member, has been identified as a novel biomarker for cardiac strain (12,13). Under the induction of separate promoters, the ST2gene expresses 2 unique proteins: sST2, the circulating form of ST2, and ST2L, which is the transmembrane receptor form and signals through a complex involving IL-33 (14–16). This IL-33/ST2L signaling plays a previously unrecognized cardioprotective role in mechanical overload and exerts antifibrotic actions (17,18). Sanada et al. (17) demonstrated that the sST2 protein acts as a decoy receptor, reversing the protective effects of IL-33/ST2L in a dose-dependent manner, probably by binding IL-33 and preventing signaling through ST2L. Thus, it is postulated that sST2 could be a potential pathophysiological mediator of myocardial fibrosis.
Concentrations of sST2 have been found to be predictive of the rate of mortality in acute coronary syndromes (19,20) and also in the long-term follow-up of patients after an acute HF episode (21–23). Whether sST2 could have a role in the identification of patients at risk of SCD is not known. On the basis of its fibrosis-modulating effect (17,18) and the interaction between cardiac fibrosis and electrical heterogeneity in SCD (11); we hypothesized that sST2 could identify patients at risk of SCD and add useful prognostic information for stratifying care. The aim of this study was to investigate such prognostic value in a nested case-control study of patients with mild-to-moderate HF and LV systolic dysfunction.
Methods
Study population and protocol
The study was a post-hoc, nested case-control study of participants included in the MUSIC (MUerte Súbita en Insuficiencia Cardíaca) study (24,25), a prospective multicenter longitudinal study designed to assess risk predictors of cardiac mortality and SCD in patients with HF in New York Heart Association functional class II to III. Participants were those patients consecutively enrolled in 3 specialist HF clinics, between April 2003 and December 2004, with both LVEF ≤45% and absence of an implanted cardiac defibrillator (n = 356). All patients received standard management as recommended by contemporary guidelines, including pharmacologic therapy at the maximal tolerated doses (angiotensin-converting enzyme inhibitor or angiotensin receptor blocker 93%, beta-blockers 73%, loop diuretics 86%, spironolactone 56%, digoxin 40%).
All patients were followed for 3 years, and case participants (n = 36) consisted of all individuals who died of SCD. After identifying cases of SCD, controls were those patients who remained alive during follow-up (3 years; n = 247). Patients were sorted according to the time of inclusion, and we attempted to match 2 control patients to each case by using the nearest control subjects that fulfilled the following criteria: same sex, age within ±5 years, and LVEF ± 5%. A total of 63 control subjects were matched to cases. The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) recommendations (26) were used in the development of the study.
Detailed information about symptoms, clinical history, 12-lead electrocardiogram, and medication usage was collected at baseline. An echocardiogram was performed at baseline in all patients, and standardized measures were obtained according to the American Society of Echocardiography recommendations (27). The 24-h ambulatory 3-channel electrocardiographic recordings were obtained at baseline and analyzed by the use of ELA Synescope Holter software (ELA Medical, Sorin Group, Paris, France) to evaluate the presence of ventricular arrhythmias. The study protocol was approved by the respective institutional investigation committees, and all patients signed informed consent. Follow-up visits were conducted on an outpatient basis every 6 months or according to patients' clinical status for 3 years.
Definition of SCD
Cases of death were verified by hospital and autopsy records and by either relatives or the primary physicians who had witnessed the death. Death was defined as “sudden cardiac death” (28) by an “ad-hoc committee” (R.V., J.C., A.B.d.L., and A.B.-G.) if it was: 1) a witnessed death occurring within 60 min from the onset of new symptoms, unless a cause other than cardiac was obvious; 2) an unwitnessed death (<24 h) in the absence of pre-existing progressive circulatory failure or other causes of death; or 3) a death during attempted resuscitation.
Biochemical analysis
Blood samples were obtained by venipuncture at patient enrollment and aliquots of serum stored at –80°C until analyzed. Concentrations of sST2 were determined by the use of an enzyme-linked immunosorbent assay from Medical & Biological Laboratories Co. (no. 7638, Woburn, Massachusetts) (12). This nonautomated assay, which uses monoclonal antibodies to human sST2 for both capture and detection of sST2, requires several manual steps, and its total imprecision at the concentration found as cut-off point (0.150 ng/ml) was of 18.5% (ranging from 5.7% at 4.0 ng/ml to 20% at 0.0625 ng/ml). The total imprecision we found was close to the recommended as desirable (<15%) in international guidelines (29) for the cut-off values of similar HF biomarkers, as B-type natriuretic peptides (BNPs), measurable with fully automated methods. N-terminal pro–B-type natriuretic peptide (NT-proBNP) levels were measured at the time of patient inclusion by electrochemiluminescence immunoassay by the use of an Elecsys 2010 analyzer (Roche Diagnostics GmbH, Mannheim, Germany) with a total imprecision <3% (30). Both sST2 and NT-proBNP were evaluated in a central laboratory (J.O.-L).
Statistical analysis
Continuous variables were tested for normal distribution by use of the Kolmogorov-Smirnov test. Variables not normally distributed are expressed as median (interquartile range [IQR]). Normally distributed continuous variables are expressed as mean ± SD. Frequencies of categorical variables were expressed as numbers (percentage) and compared by the Fisher exact test. The comparison of continuous variables between the 2 groups was carried out by use of the Mann-Whitney Unonparametric test. Receiver-operating characteristic (ROC) curve analysis was constructed to determine the optimal cut-off points for NT-proBNP and sST2 for predicting SCD. A new ordinal value was built on the basis of a combined biomarker status, taking into account the presence of neither, either, or both biomarkers above the ROC-derived cut-off points.
Univariable logistic regression analysis was used for studying the association between SCD and biomarkers and was evaluated as continuous variables and as the new combined ordinal variable. The clinical, electrical, echocardiographic, and biochemical characteristics are described in Table 1were also evaluated in the univariable logistic regression analysis. Variables that showed a significant result (p < 0.05) were included in the multivariable logistic regression model (forward stepwise) to determine those independently related to SCD. The incremental information added by sST2 was studied by taking the optimal regression multivariable model and then adding sST2 to this model. The combined variable sST2/NT-proBNP also was the last variable put into the multivariable model to assess its incremental value added to the other variables. All tests were 2-sided, and a p value <0.05 was considered statistically significant. All data analyses were performed by the use of SPSS version 15.0 software (SPSS Inc., Chicago, Illinois).
Characteristics of SCD (Cases) and Survivors (Control Subjects)
Results
The characteristics of SCD patients (cases) and HF survivors (control subjects) are shown in Table 1. After enrollment, the majority (53%) of SCD cases occurred during the first year; the follow-up in the SCD group had a median of 10.8 months (IQR 6.4 to 20.1 months). Decedents and controls were matched by age, sex, and LVEF; yet among SCD patients, previous myocardial infarction was slightly more prevalent, and differences between groups also were found for New York Heart Association functional class, left atrial diameter, and glomerular filtration rate. No group differences were observed in the other clinical, echocardiographic, electrical, or biochemical parameters. As shown in Figure 1,concentrations at enrollment of sST2 (0.23 ng/ml [IQR 0.16 to 0.43 ng/ml] vs. 0.12 ng/ml [IQR 0.06 to 0.23 ng/ml], p = 0.001) and NT-proBNP (2,873 ng/l [IQR 1,541 to 6,979 ng/ml] vs. 966 ng/l [IQR 440 to 2,405 ng/ml], p < 0.001) were significantly greater in patients who experienced SCD during the 3-year follow-up. In the logistic regression analysis, sST2 levels (+0.1 ng/ml, odds ratio [OR]: 1.39, 95% confidence interval [CI]: 1.09 to 1.78, p = 0.006) and NT-proBNP levels (OR: 1.02, 95% CI: 1.00 to 1.03, p = 0.01) were predictive of SCD.
sST2 and NT-proBNP in Patients With SCD and Survivors
Box plots showing the baseline concentrations of soluble ST2 (sST2) (A)and N-terminal pro–B-type natriuretic peptide (NT-proBNP) (B)in patients experiencing sudden cardiac death (SCD) and the control survivor group.
Figure 2illustrates the ROC curves for both sST2 and NT-proBNP, which identified the following optimal cut-off point values: 0.15 ng/ml for sST2 and 2,000 ng/l for NT-proBNP. When these values were used to predict outcome, sST2 had 83% sensitivity (95% CI: 67% to 94%), 51% specificity (95% CI: 38% to 64%), a positive predictive value of 49%, and a negative predictive value of 84%; and NT-proBNP had 72% sensitivity (95% CI: 53% to 86%), 73% specificity (95% CI: 60% to 83%), a positive predictive value of 58%, and a negative predictive value of 83%.
Receiver-Operator Characteristic Curve Analysis of sST2 and NT-proBNP Concentrations for Predicting SCD
AUC = area under the curve; other abbreviations as in Figure 1.
After adjustment in the multivariable model, introducing variables with a univariate difference between groups and excluding sST2, the independent predictors were left atrial diameter (OR: 1.12, 95% CI: 1.032 to 1.231, p = 0.006), estimated glomerular filtration rate (OR: 0.956, 95% CI: 0.918 to 0.995, p = 0.01), and NT-proBNP >2,000 ng/l (OR: 5.46, 95% CI: 1.85 to 16.1, p = 0.002; chi-square = 42.2). When added to this model, sST2 >0.15 ng/ml was an independent predictor of SCD (OR: 4.56, 95% CI: 1.31 to 15.9, p = 0.017) and added incremental value to the model (chi-square: 48.6, p = 0.012).
In a combined biomarker approach, the presence of none, either (p = 0.015), or both elevated biomarkers (p < 0.001) were predictive of SCD (p < 0.001 for trend). As shown in Table 2,this combined variable added incremental prognostic value to the multivariable regression model (p < 0.001). The presence of none, 1, or 2 elevated biomarkers was associated with increasing rates of SCD (Fig. 3);only 4% of patients experienced SCD for neither sST2 nor NT-proBNP above the ROC-derived cut-off points; 34% for either biomarker above and 71% for both biomarkers above.
Events According to the Presence of None, Either, or Both sST2 >0.15 ng/ml and NT-proBNP >2,000 ng/l
Red bars= sudden cardiac death; open bars= survivors. Abbreviations as in Figure 1.
Multiple Regression Analysis: Adjusted Models for Prediction of SCD
Discussion
This study shows that increased serum levels of sST2, a novel biomarker of cardiac strain, were predictive of SCD in ambulatory patients with mild-to-moderate chronic HF and left ventricular systolic dysfunction. The prognostic value of sST2 was independent of other clinical variables and, importantly, provided complementary information to NT-proBNP.
This study is the first to demonstrate a significant association between this novel biomarker and SCD. The authors of previous reports (22,23) already recognized sST2 as a powerful prognosticator for all-cause mortality during the long-term follow-up (1 year) in patients with acute HF. In our population, sST2 added independent prognostic information to predict SCD over the other studied variables, including NT-proBNP. Therefore, this study provides new data about the prognostic value of sST2 in the ambulatory setting of chronic HF and the complementary roles of sST2 and NT-proBNP. Such a multimarker predictive approach, sST2 plus NT-proBNP, also has been reported in patients with acute coronary syndromes and acute HF (20,22,23). In our HF population, elevation of both biomarkers above the cut-off values was associated with a very high rate (71%) of SCD, and interestingly, a very low rate (4%) of SCD existed when both markers were below the cut-off thresholds. At present, no single test reliably predicts SCD in patients with HF (10), but the combination of sST2 and NT-proBNP markedly improves risk stratification to identify high- and low-risk patients; this fact, may have an important impact on clinical decision-making, particularly for delineating optimal preventive strategies.
It is noticeable that the sST2 cut-off value for SCD in HF patients found in this study (0.150 ng/ml) is very close to the 95 percentile of a blood donor population found by Rehman et al. (22) (0.168 ng/ml) and to the cut-off value found in the PRIDE (Pro-Brain Natriuretic Peptide Investigation of Dyspnea in the Emergency Department) study to predict 1-year mortality in acutely dyspneic and acute HF patients (0.200 ng/ml) (31). Therefore, our results support the evidence of a “solid” cut-off value between 0.150 and 0.200 ng/ml for predicting and ruling-out major cardiac events in HF patients. It is likely that sST2 may fluctuate over time, as occurs with BNP, and it has been reported that changes in sST2 concentrations have also a prognostic value (21,32). In this study, we did not collect serial sST2 measures to explore the effect of changes over time, yet the presented data support the predictive value of a single inclusion measurement.
The relationship between ST2 and SCD is at present incompletely characterized, but some hypothesis could arise based on the physiology of IL-33/ST2 pathway. IL-33/ST2 has emerged as an intercellular signaling system that participates in inflammatory responses, autoimmunity, organ fibrosis, and cardiac injury (18). IL-33/ST2 signaling protects the myocardium under mechanical strain and acts as a biomechanically activated fibroblast-cardiomyocyte paracrine system that prevents cardiac hypertrophy and fibrosis (17,18). Sanada et al. (17) suggest that sST2 protein abrogates this adaptative response in a dose-dependent manner by binding IL-33 and preventing signaling through ST2L. Moreover, IL-33/ST2 system might participate in inflammatory and remodeling processes of the myocardium (15,33). Concentration of sST2 correlate positively with parameters of HF severity, as norepinephrine levels, natriuretic peptides levels, diastolic filling pressures, and C-reactive protein levels (21–23,34). It is thus possible that sST2 is not only a biomarker of outcome but also a true pathophysiological mediator of disease progression and predisposition to SCD.
sST2 modulates extracellular matrix remodeling and regulates inflammatory and electrical signals between cardiac cells types that may facilitate lethal cardiac arrhythmias (18,35) or other cardiac events ultimately leading to SCD. Moreover, its relationship with fibrosis and inflammation could explain its complementary role with NT-proBNP; in fact, procollagen and high-sensitivity C-reactive protein also have been shown to be complementary of BNP in the prediction of ventricular tachycardia, among patients with implanted cardioverter-defibrillator (9).
Sudden cardiac death is a hard end point in patients with HF, yet this mode of death is difficult to obtain, and it has been mainly evaluated by clinical trials of primary prevention. The registry from which current data were obtained was specifically designed to study SCD, and therefore patients were closely followed and the mode of death rigorously verified by an independent “ad-hoc committee,” as prespecified. However, without cardiac monitoring at the time of death, the assumption of an underlying arrhythmic cause is only presumptive. After 3 years of follow-up, 36 sudden deaths were identified. This is a large group of SCD patients, comparable or greater than those of several published clinical trials on sudden death (4–8,28,36). The greatest-risk patients, those who already had a defibrillator implanted, were excluded. Nevertheless, although the highest- and lowest-risk patients obtain less benefit from risk stratification, the patients at an intermediate risk, like those included in this study, are most benefited by SCD prediction.
Conclusions
This is the first study to demonstrate that elevated sST2 concentrations are predictive of SCD and provide complementary information to clinical variables and NT-proBNP. Future trials are needed to elucidate whether this novel biomarker, alone or in combination with NT-proBNP, is useful in clinical routine for preventing SCD.
Footnotes
This study was supported by “Red Temática de Investigación Cooperativa MUSIC” and the national network of investigation on heart failure “REDINSCOR”: Grants G03/078 and RD06/0003, Instituto de Salud Carlos III, Ministry of Health, Madrid, Spain. Dr. Pascual-Figal has received speaking and research grants from Roche Diagnostics. Dr. Ordoñez-Llanos has received speaking, consulting, and research grants from Roche Diagnostics. Dr. Bayes-Genis has received speaking and research grants from Roche Diagnostics. Critical Diagnostics freely provided the reagents used to measure sST2. The first 2 authors contributed equally to this work.
- Abbreviations and Acronyms
- BNP
- B-type natriuretic peptide
- CI
- confidence interval
- HF
- heart failure
- IL
- interleukin
- IQR
- interquartile range
- LV
- left ventricle/ventricular
- LVEF
- left ventricular ejection fraction
- NT-proBNP
- N-terminal pro–B-type natriuretic peptide
- OR
- odds ratio
- ROC
- receiver-operator characteristic
- SCD
- sudden cardiac death
- sST2
- soluble ST2
- Received June 9, 2009.
- Accepted July 8, 2009.
- American College of Cardiology Foundation
