Among patients with chronic heart failure (HF), it is known that soluble concentrations of the interleukin receptor family member ST2 (sST2) are prognostic for adverse outcome, including risk for progression of HF and death. Considerably less was known about the merits of serial testing of sST2 and whether such testing provides incremental information beyond single measurements; additionally, the influence of HF therapies on sST2 concentrations and whether sST2 values predicted benefit of therapy changes remained unclear. Recently, several studies indicate that serial testing for sST2 increases the prognostic information gained compared with a single measurement. When measured in patients with chronic HF, concentrations of sST2 appear to predict worsening left ventricular remodeling, and serial measurements of sST2 appear particularly useful for predicting HF events, such as worsening HF and risk for either hospitalization or death from HF. Values for sST2 are lower in those treated with several HF therapies; in turn, sST2 concentrations may predict benefit from specific HF therapy changes. Using an upper reference limit of 35 ng/ml, serial testing for sST2 in chronic HF thus appears useful. The promising role of sST2 for serial chronic HF therapy monitoring will be discussed.
During the course of chronic heart failure (HF) therapy, a wide range of risk may be encountered. Some patients have a relatively lower risk course, whereas others have a substantially worse prognosis, complicated by frequent episodes of instability. Understanding the likelihood for adverse outcome is a prerequisite for tailoring therapeutic strategies to improve prognosis for HF, and indeed, judging risk in those with chronic HF is recommended by current clinical practice guidelines. However, assessment of risk for complications in this setting may be challenging; signs or symptoms of congestion are often difficult to recognize, risk for arrhythmia is often impossible to recognize, and the sine qua non of progression in HF, myocardial remodeling, is not recognized until after it has occurred. This last process—left ventricular (LV) myocardial remodeling—is particularly insidious, as it may occur in the absence of symptoms or signs, and when significant, LV remodeling leads to ventricular dilation, reduced function, and a higher risk for HF events, including death.
Patients with chronic HF are typically assessed with history and physical examination as well as echocardiography to monitor their risk over time. However, such standard means have limitations. Accordingly, the development of tools to supplement clinical judgment and imaging to better recognize risk in chronic HF would be welcome. One option that has gained significant attention in recent years is the use of biomarkers to achieve this goal.
Biomarker Testing to Monitor Risk in Chronic HF: Promises and Pitfalls
On a theoretical level, measurement of circulating biomarkers indicative of HF risk is an attractive option. Considerable caveats exist before a biomarker may be adopted for clinical use, however. We recently articulated that clinically useful biomarkers of HF (1) should be easily measured with high analytical precision, (2) should reflect important pathophysiologic processes involved in HF presence and progression, (3) should not recapitulate clinical information already available at the bedside (including other biomarkers), and (4) must provide clinically useful information for caregivers to more swiftly and reliably establish and/or reject a diagnosis, to more accurately estimate prognosis, or to inform more successful therapeutic strategies.
Beyond these points, it is critical that the biologic aspects of a circulating marker be well understood before it reaches widespread clinical testing; among the factors to be ascertained include the biologic variation of the biomarker, knowledge of the clinical correlates that lead to biomarker elevation and/or reduction, and whether serial measurement adds to single/spot assessment for risk prediction. Thus, substantial work is necessary before a biomarker makes the translation from bench to bedside.
While only the natriuretic peptides have presently met the requirements listed previously, select novel biomarkers appear to have promise in HF. Among the most promising of the novel biomarkers is ST2.
ST2 in Heart Failure: Biology and Promise as a Clinical Biomarker
The role played by the ST2 system in HF is reviewed elsewhere in this consensus document (consensus report on ST2 biology). Briefly, ST2 is expressed in a transmembrane form (ST2 ligand, or ST2L) as well as in a soluble circulating form (sST2). Clarifying the understanding of the role played by ST2 in cardiovascular (CV) disease, interleukin-33 (IL-33) was identified as the ligand for ST2. IL-33 has antiremodeling effects on the myocardium, by preventing hypertrophy, fibrosis, and apoptosis; these effects are transduced by ST2L. Administration of sST2 to rat myocytes blocks the antihypertrophic influence of IL-33 in a dose-dependent fashion, suggesting that sST2 may serve as a “decoy receptor” for circulating IL-33. Seki and coworkers identified that sST2 promotes myocardial apoptosis, fibrosis, and hypertrophy, compelling results given the importance of these processes in LV remodeling. Therefore, in a clinical translation, we could expect that elevated concentrations of sST2 maintained over time, on repeated measurements, identify a phenotype of adverse remodeling, which might profit from tailored antiremodeling therapies. In fact, regarding the pathologic correlates of elevated concentrations of sST2 in HF populations, it has been established that concentrations of the marker are significantly related to the presence of a more adversely remodeled LV, significant diastolic dysfunction, and worse right ventricular performance.
Measurement of sST2 in circulation is possible, and significant efforts have been made to elucidate the various factors needed to understand its measurement in humans. A description of the variables that affect sST2 in healthy subjects is reviewed elsewhere in this consensus document (report on normal populations); beyond the understanding of factors predicting values of sST2 in health, it is important to understand the biologic variation of the biomarker if it is to be measured serially and the variables predictive of its concentrations in disease states.
Biologic variation is the expected degree of increase and/or decrease of any biologic marker when measured serially; it is the sum total of normative biologic processes that lead to changes in the secretion of a biomarker, and a critical piece of information to understand if the marker is to be measured serially. In short, understanding biologic variation answers the question of “How much of an increase or decrease in the biomarker is needed to identify a pathologic situation?”
The biologic variation of sST2 was recently examined by Wu et al. In this study, concentrations of sST2 were measured in 17 healthy subjects over a period of 8 weeks, and the amount of change in the marker assessed in the absence of significant clinical instability was assessed. Wu et al noted that the reference change value for sST2 was 30% (compared with 61% for galectin-3 and 92% for aminoterminal pro–B-type natriuretic peptide [NT-proBNP]). The index of individuality (a measure of whether serial measurements added significantly to a single assessment) for sST2 was 0.25, suggesting substantial value from serial measurement; in contrast, the index of individuality for galectin-3 was 1.0, suggesting serial galectin-3 measurement added nothing to a baseline value. The investigators of this study suggested based on these results that sST2 may be incrementally useful for monitoring long-term HF risk.
With this background established, it became possible to examine the potential value of sST2 as a monitoring tool for patients with chronic HF.
ST2 in Heart Failure: Biology and Promise as a Clinical Biomarker
The role played by the ST2 system in HF is reviewed elsewhere in this consensus document (consensus report on ST2 biology). Briefly, ST2 is expressed in a transmembrane form (ST2 ligand, or ST2L) as well as in a soluble circulating form (sST2). Clarifying the understanding of the role played by ST2 in cardiovascular (CV) disease, interleukin-33 (IL-33) was identified as the ligand for ST2. IL-33 has antiremodeling effects on the myocardium, by preventing hypertrophy, fibrosis, and apoptosis; these effects are transduced by ST2L. Administration of sST2 to rat myocytes blocks the antihypertrophic influence of IL-33 in a dose-dependent fashion, suggesting that sST2 may serve as a “decoy receptor” for circulating IL-33. Seki and coworkers identified that sST2 promotes myocardial apoptosis, fibrosis, and hypertrophy, compelling results given the importance of these processes in LV remodeling. Therefore, in a clinical translation, we could expect that elevated concentrations of sST2 maintained over time, on repeated measurements, identify a phenotype of adverse remodeling, which might profit from tailored antiremodeling therapies. In fact, regarding the pathologic correlates of elevated concentrations of sST2 in HF populations, it has been established that concentrations of the marker are significantly related to the presence of a more adversely remodeled LV, significant diastolic dysfunction, and worse right ventricular performance.
Measurement of sST2 in circulation is possible, and significant efforts have been made to elucidate the various factors needed to understand its measurement in humans. A description of the variables that affect sST2 in healthy subjects is reviewed elsewhere in this consensus document (report on normal populations); beyond the understanding of factors predicting values of sST2 in health, it is important to understand the biologic variation of the biomarker if it is to be measured serially and the variables predictive of its concentrations in disease states.
Biologic variation is the expected degree of increase and/or decrease of any biologic marker when measured serially; it is the sum total of normative biologic processes that lead to changes in the secretion of a biomarker, and a critical piece of information to understand if the marker is to be measured serially. In short, understanding biologic variation answers the question of “How much of an increase or decrease in the biomarker is needed to identify a pathologic situation?”
The biologic variation of sST2 was recently examined by Wu et al. In this study, concentrations of sST2 were measured in 17 healthy subjects over a period of 8 weeks, and the amount of change in the marker assessed in the absence of significant clinical instability was assessed. Wu et al noted that the reference change value for sST2 was 30% (compared with 61% for galectin-3 and 92% for aminoterminal pro–B-type natriuretic peptide [NT-proBNP]). The index of individuality (a measure of whether serial measurements added significantly to a single assessment) for sST2 was 0.25, suggesting substantial value from serial measurement; in contrast, the index of individuality for galectin-3 was 1.0, suggesting serial galectin-3 measurement added nothing to a baseline value. The investigators of this study suggested based on these results that sST2 may be incrementally useful for monitoring long-term HF risk.
With this background established, it became possible to examine the potential value of sST2 as a monitoring tool for patients with chronic HF.
ST2 Monitoring in Chronic HF
The general prognostic value of sST2 measurement in chronic HF is reviewed elsewhere in this consensus document (reference for chronic HF prognosis report); the great majority of studies evaluating the prognostic merit of sST2 in chronic HF have focused on single measurements. The take-home from most studies is that single concentrations of sST2 are among the strongest predictors of HF complications, including hospitalization, arrhythmia, and death. The goal of the present review was to consider the use of sST2 as a monitoring tool for chronic HF and to discuss potential effects of therapies for HF.
Of the published data regarding serial measurement, 3 provide the greatest information: 1 from the Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA) study, 1 from the Pro-BNP Outpatient Tailored Chronic HF Therapy (PROTECT) study, and 1 from the Valsartan Heart Failure Trial (Val-HeFT). It is reassuring to note that all 3 analyses were performed using the Presage ST2 assay (Critical Diagnostics, San Diego, CA), so clinicians may interpret the data with the understanding that results from these 3 trials are applicable to those obtainable in clinical practice.
In the CORONA analysis, Broch et al measured concentrations of sST2 using the Presage ST2 assay in 1,449 subjects with HF due to LV systolic dysfunction; in 1,309 subjects, a second sample 3 months after randomization was available for sST2 measurement. The median follow-up time for this study was 2.6 years; during this period, 28.2% met the primary end point of CV death, nonfatal myocardial infarction (MI), or stroke, whereas 29.1% died. The median concentration of sST2 at baseline for this study was 17.8 ng/ml (interquartile range 13.0 to 25.0). As expected, those with sST2 values in the highest tertile (in this case >28.8 ng/ml) were older and more likely to be male, to have lower LV ejection fraction, to have more prevalent atrial fibrillation and greater NT-proBNP and C-reactive protein concentrations. Notably, patients with highest values of sST2 in CORONA were least likely to be taking angiotensin-converting enzyme inhibitors (ACEi) or angiotensin II receptor blockers (ARB), compared with those with lower sST2 values.
In the CORONA study, after initial adjustment for established clinical and biochemical variables, baseline sST2 was a significant predictor of all end points examined, including the primary end point of CV death, nonfatal MI, or stroke, as well as death, and particularly worsening HF and hospitalization for HF. With addition of NT-proBNP and C-reactive protein to the model, sST2 was no longer a predictor of the primary end point (reflecting coronary ischemic complications). However, sST2 remained significantly predictive of death from worsening HF (hazard ratio [HR] 1.57, 95% confidence interval [CI] 1.05 to 2.34, p = 0.03), hospitalization for CV causes (HR 1.28, 95% CI 1.07 to 1.52, p = 0.006), and hospitalization for worsening HF (HR 1.30, 95% CI 1.04 to 1.62, p = 0.02). Treatment allocation to rosuvastatin did not interact with the predictive value of sST2, and there was no interaction between age and the predictive value of sST2.
When analyzed dichotomously using the median concentration of 17.8 ng/ml, sST2 demonstrated a predictive ability additional to that of NT-proBNP. For most end points, patients with NT-proBNP below the median level had a significantly worse outcome if sST2 was above its median level. In patients with NT-proBNP above the median level, the risk of all-cause hospitalization, as well as hospitalization for worsening HF, significantly increased if sST2 was also above median.
In the 1,309 patients with sST2 re-measurement 3 months after randomization, the change in sST2 was negligible overall (median 0, interquartile range −3 to 3 ng/mL); however, although the vast majority of patients had no change in their sST2 serum concentration over the first 3 months, a few patients experienced large change in the biomarker. In fully adjusted Cox regression analyses, although most of the coronary end points were not predicted by change in sST2 values, a decrease in sST2 by 3 months was associated with a reduced risk of hospitalization for worsening HF (HR 0.87, p = 0.02) and hospitalization for CV causes (HR 0.88, p = 0.006). No treatment interaction was observed (data not shown). Notably, an increase in sST2 of ≥15.5% was associated with hospitalization for CV causes (p = 0.04), but not with any other end point on univariate analyses. After full adjustment, an increase in sST2 was a statistically significantly predictor of both the primary end point and hospitalization for CV causes.
In the PROTECT study analysis, Gaggin et al measured concentrations of sST2 in a cohort of 151 subjects with HF due to LV systolic dysfunction (LV ejection fraction <40%). Of these, 145 subjects had >1 sample available for serial assessment. In this analysis, concentrations of sST2, highly sensitive troponin T (hsTnT), and growth differentiation factor 15 (GDF15) were added to a model that included clinical variables and NT-proBNP; at baseline, all 3 biomarkers significantly improved risk prediction compared with the clinical model alone, displacing NT-proBNP for prognosis.
Examined serially, the investigators noted significant change in sST2 concentration over a median of 10 months of follow-up; in contrast, neither hsTnT nor GDF15 showed substantial change from baseline. Correlating baseline to final values, a correlation coefficient of 0.67 was seen for sST2 (similar to NT-proBNP), compared with 0.87 and 0.86 for hsTnT and GDF15, respectively. These data suggest significant increase and/or decrease in sST2 over time, which is greater than most other novel biomarkers. Using a conventional threshold of 35 ng/ml (representing the 97.5th percentile of a normal population and commonly used value for HF risk prediction) and again focused on baseline and final measurements, 40% of subjects changed from either below or above this threshold ( Table 1 ).
| Biomarker | Cut-off | Always above | Above to below | Below to above | Always below | Change in category |
|---|---|---|---|---|---|---|
| sST2 | <35 ng/mL | 31.7% | 22.1% | 17.9% | 28.3% | 40% |
| NT-proBNP | <1000 pg/mL | 55.6% | 29.7% | 4.9% | 19.7% | 25% |
| GDF15 | <2000 ng/L | 65.8% | 8.2% | 6.9% | 19.2% | 21% |
| hsTnT | <14 pg/mL | 54.8% | 9.6% | 11.6% | 24.0% | 15% |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
