Dynamic change in soluble suppressor of tumorigenicity 2 (sST₂) during acute decompensated heart failure with a reduced ejection fraction, and its mortality implications

Abstract

Aims

To evaluate the association of soluble suppression of tumorigenicity 2 (sST₂) concentrations measured at clinician-determined decompensation and recompensation, and their change between these states, with one-year mortality in acute decompensated heart failure (ADHF).

Methods

Paired sST₂ samples were obtained at decompensation and recompensation from 168 patients hospitalized with ADHF and a left ventricular ejection fraction of <50%. Associations with one-year all-cause mortality were assessed using Cox regression and Kaplan–Meier analysis. Discrimination was evaluated using receiver operating characteristic curves, with exploratory cut-offs derived using Youden's index.

Results

The median age was 71.5 years (IQR 62–80.8), 39 (23.2%) were women, median LVEF was 26% (IQR 17.5–32.5), and 158 (94.1%) had NYHA class III or IV symptoms. During one-year follow-up, 34 patients (20.2%) died. Each doubling of decompensation sST₂ was associated with higher mortality after adjustment for creatinine (HR 1.67, 95% CI 1.23–2.25). Recompensation sST₂ remained associated with mortality after adjustment for creatinine and NT-proBNP (HR per doubling 1.71, 95% CI 1.18–2.48). The respective areas under the curve were 0.686 and 0.688, with exploratory cut-offs of 66.5 and 57 ng/mL. Unchanged or increased sST₂ was associated with higher mortality than decreased sST₂ in unadjusted analysis (HR 2.76, 95% CI 1.37–5.59, p = 0.003).

Conclusion

Elevated sST₂ at decompensation and unchanged or increased concentrations during recompensation were associated with one-year mortality in patients with ADHF. These findings support further evaluation of sST₂ as an adjunctive prognostic biomarker in this setting; the cohort-derived cut-offs require validation.

Graphical abstract

1. Introduction

Suppressor of tumorigenicity 2 (ST₂), first described in 1989, is a member of the interleukin-1 receptor superfamily [1]. The ST₂ gene encodes and produces two major isoforms, ST₂L (transmembrane) and sST₂ (soluble) [2]. Binding of interleukin-33 (IL-33) to ST₂L activates signalling pathways involved in inflammation and immune regulation [3]. Experimental studies suggest that IL-33/ST₂L signalling exerts cardioprotective effects through limitation of cardiomyocyte apoptosis, hypertrophy and fibrosis [4].

In contrast, sST₂ acts as a decoy receptor [5], binding IL-33 and attenuating these protective effects (Fig. 1). Circulating sST₂ has therefore attracted interest as a cardiovascular biomarker, with early studies in acute myocardial infarction (MI) associating high sST₂ with elevated creatine kinase, reduced left ventricular ejection fraction and poorer outcomes [6], [7].

Fig. 1.

Illustration of the interaction between interleukin-33 and transmembrane ST₂L (left), and the action of sST₂ as decoy receptor in blocking the interleukin-33/ST₂L effects (right).

With emerging data showcasing its diagnostic and prognostic capabilities, the American Heart Association (AHA) / American College of Cardiology (ACC) gave a class IIb recommendation for the additive utility of sST₂ in risk stratification of HF in both acute and ambulatory settings [8], [9]. In 2014, the international ST₂ consensus panel proposed an sST₂ concentration of 35 ng/mL as a prognostic threshold in patients with chronic heart failure [10], [11], [12]. However, the applicability of this threshold to acute decompensated heart failure (ADHF) requires further evaluation.

Evidence also supports the prognostic value of sST₂ in acute HF. In the Serum Catestatin Expression and Cardiometabolic Parameters in Patients with Congestive Heart Failure (CATSTAT-HF) study, sST₂ was as an independent mortality predictor during an admission for acute worsening of heart failure, with levels significantly higher in patients who died in-hospital compared to survivors [13]. In the N-terminal Pro-BNP Investigation of Dyspnea in the Emergency Department (PRIDE) study, elevated sST₂ was independently associated with one-year mortality, including among patients with acute HF [14]. A meta-analysis of ten studies further supported the association between elevated sST₂ concentrations and mortality in acute HF, with an average sST₂ concentration of 68 ng/mL at acute heart failure, reported to be an important prognostic threshold [15], [16].

Serial measurement of sST₂ may provide additional prognostic information, but the optimal timing and sampling remain uncertain. Previous studies have evaluated repeat sST₂ measurements at predefined intervals or at discharge, which may not coincide with clinical recompensation [17], [18]. Because the pace of recovery varies between patients, sampling at clinically defined decompensation and recompensation may help contextualize sST₂ changes in relation to clinical recovery. The prognostic value of this approach remains insufficiently characterized.

The purpose of this study therefore was to assess the association of sST₂ concentrations at clinically defined decompensation and recompensation, and their change between these time points, with one-year mortality in patients hospitalized with ADHF and an LVEF <50%.

2. Methods

2.1. Study design

INST₂ANT-HF (Investigating the Soluble Suppressor of Tumorigenicity 2 Biomarker in addition to Natriuretic Peptides in Managing Patients with Heart Failure and a Reduced Ejection Fraction) was a prospective, observational cohort study evaluating the relationship between changes in sST₂ levels and adverse outcomes at clinician-determined decompensation and recompensation of acute heart failure. The primary endpoint for this study was all-cause mortality within one year after discharge following hospitalization with acute decompensated heart failure (ADHF). The study was sponsored by Mid and South Essex NHS Foundation Trust and received ethical approval from the East of England, Cambridgeshire and Hertfordshire Research Ethics Committee and Health Research Authority and Health and Care Research Wales (REC reference 19/EE/0269). The study was conducted in accordance with the principles of Good Clinical Practice and the World Medical Association Declaration of Helsinki.

2.2. Patient population

We enrolled consecutive adult patients aged ≥18 years, who were admitted to Basildon University Hospital with an acute decompensation of chronic or newly diagnosed HF with a reduced left ventricular ejection fraction <50%, based on conventional imaging. Patients were excluded if LVEF was ≥50%, they required renal replacement therapy, were on a palliative care pathway or were participants in other HF trials involving new drugs or procedural interventions. All participants provided oral and written informed consent.

2.3. Study procedure

Before enrolment, eligible participants were reviewed by HF specialists, specifically, a heart failure cardiologist or specialist nurse, to confirm HF decompensation. This ensured the enrolled cohort included only patients with the correct clinical diagnosis. All patients received contemporary inpatient HF treatment in line with published guidance [19]. Enrolled patients had daily clinical reviews, including assessments of congestion, throughout hospitalization. Recompensation was determined by the treating HF specialist, through collective assessment of symptoms, clinical signs, laboratory biomarkers and fluid status. Findings considered included symptom resolution; reductions in jugular venous pressure and pulmonary and/or peripheral oedema; net negative fluid balance; a downward trend in body weight towards pre-admission levels; resolution of any additional oxygen requirement; and haemodynamic stability.

As part of usual care, blood samples were collected within 24 h of clinician-determined HF decompensation and furthermore at recompensation. sST₂ was measured in addition to usual analytes from both samples. Biochemical analyses of paired sST₂ samples were performed blinded to baseline characteristics and clinical parameters of the study participants, and results of sST₂ measurements were only unmasked to INST₂ANT-HF investigators at the end of the study period. These results could therefore not be utilized in clinical decision-making. The primary outcome was all-cause mortality within one year after discharge from the index hospitalization.

2.4. Effect size

The estimated change in sST₂ levels during a hospitalization for heart failure, reported in literature and thought to be clinically relevant, is between 11 and 33% [17], [20]. For this study, we hypothesized that an effect size of 20% would be a reasonable estimate that would represent an anticipated relative change in sST₂, given the fact that our patient cohort was in a clinically validated state of congestion.

2.5. Biochemical analyses

Non-fasting ethylenediaminetetraacetic acid (EDTA) blood samples were obtained via venepuncture for analysis by the essential service laboratory at Basildon University Hospital. Samples intended for sST₂ measurements were centrifuged, with plasma aliquots frozen and stored at −80 °C, until batch analyses, which took place at the end of the study period. All samples went through no more than two freeze-thaw cycles. sST₂ concentrations were determined using the SEQUENT-IA™ sST₂ assay (Critical Diagnostics, San Diego, CA, USA) on the Beckman Coulter® (Brea, California) AU680 automated chemistry analyser. The SEQUENT-IA™ sST₂ assay has a linear dynamic range from 15 ng/mL to 300 ng/mL, and a minimum detectable value of 10.8 ng/mL. The intra-assay and inter-assay coefficients of variation are 6.5% and 8.5%, respectively, at an sST₂ level of 25 ng/mL.

2.6. Statistical analyses

Continuous variables are reported as means ± standard deviation (SD) or as medians with interquartile ranges (IQR), according to their distribution. Categorical variables are reported as frequency and percentages. Between-group comparisons were made using Student's t-test or Mann-Whitney U test, and categorical variables with Chi-Square or Fisher's exact test, as appropriate.

Associations between sST₂ concentrations and continuous clinical parameters were assessed using Spearman's rank correlation, because of the skewed distribution. Paired sST₂ concentrations at HF decompensation and recompensation were compared using the Wilcoxon signed-rank test. Concentrations in patients who died and those who survived to one year were compared using the Kruskal-Wallis test. Receiver operating characteristic (ROC) analysis was used to assess the discrimination of sST₂ concentrations at decompensation and recompensation, and their percentage change, for one-year all-cause mortality. Discrimination was expressed as the area under the ROC curve (AUC) with 95% confidence intervals.

To evaluate the prognostic association of sST₂, selected continuous biomarkers were log₂-transformed, with hazard ratios interpreted per doubling of concentration. Associations between sST₂ at decompensation and recompensation and one-year all-cause mortality were examined separately using univariable and multivariable Cox proportional hazards regression. Proportional hazards were assumed. Hazard ratios (HRs) are reported with 95% confidence intervals (CIs).

Covariates selected for adjustment included creatinine, to account for renal function, which influences natriuretic peptide concentrations and prognosis in ADHF, and NT-proBNP, to assess whether sST₂ remained associated with mortality after accounting for an established prognostic biomarker [21], [22]. Covariates were added sequentially, and changes in model fit were assessed using likelihood-ratio tests. The final decompensation model included creatinine; NT-proBNP was not retained because its addition did not significantly improve model fit. The final recompensation model included both creatinine and NT-proBNP. Model complexity was limited in relation to the number of deaths, using approximately ten events per estimated predictor parameter as a pragmatic guide rather than a guarantee of model stability [23].

In secondary exploratory analyses, sST₂ concentrations were dichotomised using cohort-derived cut-offs that maximised Youden's index in ROC analysis. Survival was estimated using the Kaplan–Meier method and compared using the log-rank test. Associations with mortality were further evaluated using univariable and multivariable Cox regression. Because the cut-offs were derived and evaluated in the same cohort, these analyses were considered exploratory.

Patients with unchanged or increased sST₂ between decompensation and recompensation were also compared with those whose concentrations decreased, using Kaplan–Meier analysis, the log-rank test and univariable Cox regression. All analyses were conducted using IBM SPSS Statistics software package (version 27.0.1.0). A p-value <0.05 was considered statistically significant.

3. Results

Between December 2019 and January 2022, 199 patients were enrolled into the INST₂ANT-HF study. The COVID-19 pandemic disrupted recruitment and access to assay supplies. Paired assays were unavailable for 31 patients; samples were not collected from 10 patients because of deterioration or end-of-life care, and samples from 21 patients were lost following delays in assay access. 168 patients who had reconcilable paired sST₂ at decompensation and recompensation were included in the final analysis.

Baseline characteristics are displayed in Table 1. The median age was 71.5 (IQR 62–80.8) years; 39 (23.2%) were women, and 160 (95.2%) were ethnically white. 59 (35.1%) had heart failure secondary to an ischaemic aetiology. 89 (53), 52 (31%) and 61 (36.5%) were co-morbid with hypertension, ischaemic heart disease and atrial fibrillation, respectively. 158 (94.1%) presented with NYHA class III or IV symptoms, and the median LVEF was 26 (IQR 17.5–32.5)%. At admission, 65 (38%) were naïve to any of the foundational heart failure medicines.

Table 1.

Demographics, clinical characteristics, and biomarker measurements of participants of INST₂ANT-HF.

Demographics and baseline characteristics, median (IQR) or n (%)	N = 168
Age (years)	71.5 (62–80.8)
Female gender	39 (23.2)
Ethnicity
White	160 (95.2)
Black African	4 (2.4)
Asian	3 (1.8)
Body mass index (kg/m2)	29.2 (25.9–33.5)
Hypertension	89 (53)
Pre-existing heart failure	64 (38.1)
Ischaemic heart disease	52 (31)
Atrial fibrillation	61 (36.5)
Diabetes
Type 1	3 (1.5)
Type 2	51 (30.4)
Prior cerebrovascular accident	15 (8.9)
Obstructive airways diseases	31 (18.5)
Chronic kidney disease	41 (24.4)
Heart failure aetiology
Ischaemic	59 (35.1)
Non-ischaemic	97 (57.7)
Dual pathology	7(4.2)
Unknown	5 (3)
NYHA
II	10(6)
III	67 (39.9)
IV	91 (54.2)
Left ventricular ejection fraction (%)	26 (17.5–32.5)
Foundational medicines on admission (%)
ACEi/AIIRB/ARNI	63 (37.5)
Beta-blocker	73 (43.4)
Mineralocorticoid receptor antagonist	38 (22.7)
Cardiac devices
CRT-D	9 (5.4)
CRT-P	2 (1.2)
ICD	5 (3)
Permanent pacemakers	3 (1.8)
Smoker
Current	24 (14.4)
Ex	51 (30.6)

ACEi = angiotensin-converting enzyme inhibitor, AIIRB = angiotensinogen II receptor blocker, ARNI = angiotensin receptor–neprilysin inhibitor, NYHA = New York Heart Association Classification of Heart Failure, CRT-D = Cardiac Resynchronisation Therapy with a Defibrillator, CRT-P = Cardiac Resynchronisation Therapy with a Pacemaker, ICD = Implantable Cardioverter Defibrillator.

Median NT-proBNP and sST₂ at decompensation were 5893 ng/L (IQR 3537–11,426) and 62 ng/mL (IQR 40–97), respectively. The median time between clinician-determined HF decompensation and recompensation in the entire cohort was 8 (IQR 5–13) days. Median sST2 at recompensation was 39.5 ng/mL (IQR 24–64), and the median within-patient reduction was 32.3% (IQR 9.6–53.6).

3.1. Clinical correlates of sST₂ in ADHF

Bivariate associations are presented in Table 2. At both decompensation and recompensation, sST₂ concentrations correlated positively with C-reactive protein and NT-proBNP. No statistically significant correlations were observed with QRS duration, haemoglobin, ferritin or creatinine.

Table 2.

Correlation between sST₂ with other clinical measurements.

Clinical measurements	sST2 at decompensation (ng/l)		sST2 at recompensation (ng/l)
	ρ	Sig.	ρ	Sig.
LVEF (%)		0.126	0.176	0.023
QRS duration (ms)		0.862	–	0.739
Haemoglobin (g/dl)		0.234	–	0.086
Ferritin (ng/ml)		0.590	–	0.519
Creatinine (micromol/l)		0.498	–	0.190
C-reactive protein (mg/l)	0.360	<0.001	0.241	0.002
sST2 at decompensation (ng/l)			0.642	<0.001
sST2 at recompensation (ng/l)	0.642	<0.001
NT-proBNP at decompensation (ng/l)	0.304	<0.001	0.186	0.018
NT-proBNP at recompensation (ng/l)	0.279	<0.001	0.423	<0.001

LVEF = left ventricular ejection fraction, sST₂ = soluble suppressor of tumorigenicity 2; NT-proBNP = N-terminal pro-B-type natriuretic peptide; ρ = Spearman correlation coefficient.

3.2. One-year mortality and continuous sST₂ concentrations

At one-year follow-up, 34 (20.2%) patients had died. Compared to survivors (s), decedents (d) were typically older [d 78.5 (IQR 72–84.3) vs s 69.5 (IQR 60–79)], and comorbid with obstructive airways disease (d 32.4% vs s 14.9%; p = 0.019) or chronic kidney disease (d 44.1% vs s 19.4%; p = 0.003). They were also more likely to have an ischaemic aetiology for HF (d 50% vs s 31.3%; p = 0.042) and higher LVEF [d 32.5 (IQR 21.3–40)% vs s 25 (IQR 16–32.5)%; p = 0.043]. Significantly higher median sST₂ levels were observed in decedents compared to survivors at both decompensation [d 85 (IQR 52–122) ng/mL vs s 55 (IQR 37–84) ng/mL, p = 0.029] and recompensation [d 58 (IQR 36–100) ng/mL vs s 35 (IQR 22–61) ng/mL, p = 0.013].

The AUCs for one-year mortality were 0.686 (95% CI 0.581–0.791; p = 0.001) for decompensation sST₂ and 0.688 (95% CI 0.588–0.788; p < 0.001) for recompensation sST₂. In univariable Cox analyses, each doubling of sST₂ was associated with a higher mortality hazard at decompensation (HR 1.62, 95% CI 1.22–2.16, p < 0.001) and recompensation (HR 1.85, 95% CI 1.32–2.61, p < 0.001; Table 3a and b). Decompensation sST₂ remained associated with mortality after adjustment for creatinine (HR per doubling 1.67, 95% CI 1.23–2.25; p < 0.001). Adding NT-proBNP did not significantly improve model fit (likelihood ratio χ²(1) = 1.11, p = 0.292). Recompensation sST₂ remained associated with mortality after adjustment for creatinine and NT-proBNP (HR per doubling 1.71, 95% CI 1.18–2.48, p = 0.004).

Table 3.

Cox proportional hazards models for one-year mortality — sST₂ modelled continuously (N = 168, 34 events).

Model	Variable	HR	95% CI	P	Δχ2 (df)	P (Δχ2)
(a) sST2 at decompensation, log2-transformed
Model 1 (unadjusted)	sST2, per doubling	1.62	1.22–2.16	<0.001	9.50 (1)	0.002
Model 2 (+creatinine) Primary model	sST2, per doubling	1.67	1.23–2.25	<0.001	17.45 (1)	<0.001
	Creatinine, per doubling	6.50	2.79–15.14	<0.001
Model 3 (+NT-proBNP) Sensitivity analysis	sST2, per doubling	1.60	1.18–2.18	0.003	1.11 (1)	0.292
	Creatinine, per doubling	5.39	2.16–13.47	<0.001
	NT-proBNP, per doubling	1.18	0.86–1.62	0.299
(b) sST2 at recompensation, log2-transformed
Model 1 (unadjusted)	sST2 at recompensation, per doubling	1.85	1.32–2.61	<0.001	11.77 (1)	<0.001
Model 2 (+creatinine)	sST2 at recompensation, per doubling	1.90	1.32–2.73	<0.001	11.37 (1)	<0.001
	Creatinine, per doubling	4.50	1.89–10.70	<0.001
Model 3 (+NT-proBNP) Primary model	sST2 at recompensation, per doubling	1.71	1.18–2.48	0.004	4.61 (1)	0.032
	Creatinine, per doubling	3.13	1.20–8.15	0.020
	NT-proBNP, per doubling	1.40	1.02–1.91	0.034

3.3. Exploratory sST₂ thresholds

ROC analysis identified cohort-derived cut-offs of 66.5 ng/mL (sensitivity 75%, specificity 61%) at decompensation and 57 ng/mL at recompensation (sensitivity 57.1%, specificity 73.2%). Patients with decompensation sST₂ > 66.5 ng/mL had lower one-year survival than those with concentrations ≤66.5 ng/mL (log-rank χ²(1) = 11.94, p < 0.001). This cut-off was associated with mortality in unadjusted analysis (HR 5.04, 95% CI 2.15–11.79, p < 0.001) and after adjustment for creatinine (HR 5.30, 95% CI 2.26–12.44, p < 0.001; Table 4a). The mean survival time for patients with decompensation sST2 > 66.5 ng/mL was 289.5 days (95% CI 261.6–317.3) compared to 344.8 days (95% CI 332.6–356.9) in those <66.5 ng/mL. At baseline, patients with sST₂ > 65 ng/mL were more likely to be older, male and comorbid with atrial fibrillation, but there were no significant differences in LVEF, NYHA, and other comorbidities (Table 5).

Table 4.

Cox proportional hazards models for one-year mortality — sST₂ dichotomised (N = 168, 34 events).

Model	Variable	HR	95% CI	P	Δχ2 (df)	P (Δχ2)
(a) sST2 > 66.5 ng/mL at decompensation
Model 1 (unadjusted)	sST2 > 66.5 ng/mL	5.04	2.15–11.79	<0.001	16.78 (1)	<0.001
Model 2 (+creatinine) Primary model	sST2 > 66.5 ng/mL	5.30	2.26–12.44	<0.001	18.34 (1)	<0.001
	Creatinine, per doubling	7.05	3.01–16.54	<0.001
Model 3 (+NT-proBNP) Sensitivity analysis	sST2 > 66.5 ng/mL	4.98	2.11–11.76	<0.001	1.08 (1)	0.300
	Creatinine, per doubling	5.73	2.24–14.63	<0.001
	NT-proBNP, per doubling	1.19	0.85–1.65	0.307
(b) sST2 > 57 ng/mL at recompensation
Model 1 (unadjusted)	sST2 > 57 ng/mL	3.58	1.68–7.65	<0.001	11.12 (1)	<0.001
Model 2 (+creatinine)	sST2 > 57 ng/mL	3.25	1.51–6.98	0.003	10.13 (1)	0.001
	Creatinine, per doubling	4.24	1.76–10.20	0.001
Model 3 (+NT-proBNP) Primary model	sST2 > 57 ng/mL	2.91	1.36–6.24	0.006	6.48 (1)	0.011
	Creatinine, per doubling	2.98	1.14–7.76	0.026
	NT-proBNP, per doubling	1.46	1.08–1.98	0.014

Table 5.

Demographics and baseline characteristics in sST₂ at decompensation ≤66.5 vs sST₂ at decompensation >66.5.

Demographics and baseline characteristics, median (IQR) or n (%) N = 168
	sST2 at decompensation ≤66.5 ng/mL (N = 96)	sST2 at decompensation >66.5 ng/mL (N = 72)	P
Age (years)	70.00 (61.00–81.00)	74.00 (62.25–80.00)	0.043
Female gender	31 (32.3%)	8 (11.1%)	0.002
Ethnicity
White	90 (93.8%)	70 (97.2%)	0.497
Black African	2 (2.1%)	2 (2.8%)	1.000
Asian	3 (3.1%)	0 (0.0%)	0.355
BMI (kg/m2)	29.48 (24.71–33.86)	29.10 (26.23–32.72)	0.799
Hypertension	54 (56.3%)	35 (48.6%)	0.409
Pre-existing heart failure	33 (34.4%)	31 (43.1%)	0.324
Ischaemic heart disease	31 (32.3%)	21 (29.2%)	0.791
Atrial fibrillation	28 (29.2%)	33 (45.8%)	0.044
Diabetes
Type 1	1 (1.0%)	2 (2.8%)	0.801
Type 2	29 (30.2%)	21 (29.2%)	1.000
Prior cerebrovascular accident	9 (9.4%)	6 (8.3%)	1.000
Obstructive airways diseases	16 (16.7%)	15 (20.8%)	0.626
Chronic kidney disease	26 (27.1%)	15 (20.8%)	0.452
HF aetiology			0.520
Ischaemic	36 (37.5%)	23 (31.9%)
Non-ischaemic	53 (55.2%)	44 (61.1%)
Dual pathology	3 (3.1%)	4 (5.6%)
Unknown	4 (4.2%)	1 (1.4%)
NYHA			0.228
II	4 (4.2%)	6 (8.3%)
III	43 (44.8%)	24 (33.3%)
IV	49 (51.0%)	42 (58.3%)
LVEF (%)	25.00 (15.50–32.50)	27.50 (17.50–35.63)	0.546
Cardiac devices
CRT-D	7 (7.3%)	2 (2.8%)	0.347
CRT-P	1 (1.0%)	0 (0.0%)	1.000
ICD	3 (3.1%)	2 (2.8%)	1.000
Permanent pacemakers	3 (3.1%)	3 (4.2%)	1.000
Smoker
Current	22 (22.9%)	11 (15.3%)	0.300
Ex	33 (34.4%)	17 (23.6%)	0.180

NYHA = New York Heart Association Classification of Heart Failure, CRT-D = Cardiac Resynchronisation Therapy with a Defibrillator, CRT-P = Cardiac Resynchronisation Therapy with a Pacemaker, ICD = Implantable Cardioverter Defibrillator.

Similarly, patients with recompensation sST₂ > 57 ng/mL had lower survival than those with concentrations ≤57 ng/mL (log-rank χ²(1) = 12.16; p < 0.001). The association persisted in unadjusted analysis (HR 3.58, 95% CI 1.68–7.65) and after adjustment for creatinine and NT-proBNP (HR 2.91, 95% CI 1.36–6.24; Table 4b). The mean survival time for sST₂ > 57 ng/mL was 279.5 days (95% CI 246.5–312.5), compared to sST₂ ≤ 57 ng/mL of 341.3 days (95% CI 328.7–353.9).

3.4. Percentage change in sST₂ and mortality

Between decompensation and recompensation, the median percentage reduction of sST₂ was 32.3% (IQR 9.6–53.6) across the entire cohort. The median percentage change in sST₂ was −25% (IQR −41 to 15) among patients who died and − 35% among survivors (IQR −55 to −13). This difference was not statistically significant (p = 0.084). The AUC for percentage change was 0.575 (95% CI 0.455–0.694; p = 0.221; Fig. 2).

Fig. 2.

Receiver operating characteristic curves for sST₂ at decompensation, at recompensation, and percentage change, for predicting one-year mortality.

3.5. Trajectory change in sST₂ and mortality

Between HF decompensation and recompensation, sST₂ concentrations were unchanged or increased in 32 patients (19.0%) and decreased in 136 (81.0%). Patients with a decrease were younger [71 (IQR 61–80) vs 77.5 (IQR 68.3–82.5); p = 0.031] and less frequently had obstructive airways disease (14% vs 37.5%; p < 0.002). One-year mortality was 37.5% (12/32) among patients with unchanged or increased sST₂ and 16.2% (22/136) among those with a decrease (Pearson χ² = 7.30; p = 0.007; Table 6). Survival was lower in the unchanged/increased group (log-rank p = 0.003; Fig. 3). In unadjusted Cox regression, unchanged or increased sST₂ was associated with a higher mortality hazard (HR 2.76, 95% CI 1.37–5.59, p = 0.003).

Table 6.

Contingency table for sST₂ trajectory vs mortality at one year.

Contingency table for sST2 trajectory vs mortality at one year
	Deceased at 1 year	Survived at 1 year
sST2 (<-> or ↑)	12	20
sST2 (↓)	22	114

Fig. 3.

Kaplan-Meier curve for the survival probability of the two groups (Increased or unchanged sST₂ vs reduced sST₂ during ADHF episode) during one-year follow up.

4. Discussion

In this prospective cohort of patients hospitalized with ADHF and an LVEF <50%, higher sST₂ concentrations at clinician-determined decompensation and recompensation were associated with one-year mortality after adjustment for selected covariates. Although sST₂ concentrations generally declined during clinical recovery, percentage change showed limited discrimination for mortality. Unchanged or increased concentrations were nevertheless associated with higher mortality in an unadjusted analysis.

Relief of congestion is a central goal of ADHF treatment. Biomarkers that complement clinical assessment and identify patients at increased risk after apparent recovery may therefore be useful.

Natriuretic peptides (NPs) remain the recommended gold standard biomarkers to support diagnosis, risk stratification, prognostication, and to guide treatment of acute and chronic heart failure (HF) [24]. Despite their well-established utility, NPs are limited by relatively high biological variability, and their clinical interpretation can be affected by comorbidities such as age, renal impairment and body mass index [25], [26]. Whereas NPs primarily reflect myocardial stretch, sST₂ is associated with inflammatory and remodelling pathways and may provide complementary information [27]. Lower biological variability than NT-proBNP has also been reported in stable chronic HF, although this does not establish its performance for monitoring acute decompensation [28].

Prior studies have investigated sST₂ at predefined time points during and after an ADHF episode and established temporal trends of sST₂ levels and its prognostic associations [17], [18], [20], [29]. To our knowledge, our study is the first to extend this work by linking sampling to clinician-determined decompensation and recompensation. Because the pace of recovery varies between patients, this approach provides clinical context for interpreting biomarker changes. Compared to other studies [17], [18], [20], [29], most participants presented with NYHA class III or IV symptoms, and the median LVEF was 26%, indicating substantial symptomatic and systolic impairment. This likely represents a considerably sicker cohort of patients with ADHF. Two-thirds of our cohort presented with incident HF with a reduced LVEF <50%, which is reflective of the current landscape of HF diagnosis in the UK [30].

The median sST₂ level at decompensation was 62 ng/mL, exceeding the 35 ng/mL threshold proposed for prognostic assessment in chronic HF [11]. Exploratory analyses identified cut-offs of 66.5 ng/mL at decompensation and 57 ng/mL at recompensation, above which mortality was higher. However, these cut-offs were derived and evaluated in the same cohort and should not be interpreted as established clinical thresholds. The continuous analyses provide stronger support for the prognostic association: each doubling of sST₂ was associated with a higher mortality hazard at both time points. Discrimination was modest, with AUCs of approximately 0.69.

Recompensation sST₂ remained associated with mortality after adjustment for creatinine and NT-proBNP, suggesting that it may capture prognostic information not fully reflected by these measurements. In addition, decompensation sST₂ remained associated with mortality after adjustment for creatinine. These findings do not establish superiority over natriuretic peptides or establish that sST₂ is unaffected by renal dysfunction; however, the lack of significant interaction with renal function, which can typically confound the utility of NPs, suggests a potential advantage.

There was noted a strong association of sST₂ with CRP indicating a possible inflammatory pathway in its expression. The correlation with CRP and higher prevalence of obstructive airways disease/CKD among decedents suggest that sST₂ may reflect systemic inflammatory burden, not only heart failure-specific remodelling. Ongoing studies targeting inflammation in HF may provide additional clarity [31].

Failure of NT-proBNP to decline during ADHF has previously been associated with adverse outcomes [32]. In our cohort, the magnitude of percentage change in sST₂ did not differ significantly between patients who died or survived, and its AUC did not demonstrate discrimination beyond chance. By contrast, mortality was higher among patients whose sST₂ remained unchanged or increased than among those whose concentrations decreased. These analyses address different aspects of biomarker change and should be interpreted together.

The trajectory finding remains exploratory. Its association with mortality was unadjusted, and patients with unchanged or increased sST₂ were older and more frequently had obstructive airways disease. Small changes may also reflect analytical or biological variability. Persistent sST₂ elevation despite clinical recompensation may identify patients with residual risk. Further studies are required to establish whether sST₂ should guide treatment intensification or other care decisions.

4.1. Strength and limitations

Strengths include prospective recruitment, paired sampling at clinically defined states, specialist assessment throughout hospitalization, and blinding of sST₂ results until study completion. These features allowed examination of biomarker concentrations in relation to clinical recovery without sST₂ measurements influencing treatment decisions.

Several limitations warrant consideration. This was a single-centre study with 168 analysable participants and 34 deaths, limiting the precision of estimates and the number of covariates that could be included in adjusted models. Residual confounding remains possible, particularly for the unadjusted trajectory analysis. Multiple exploratory comparisons and the cohort-derived cut-off may also have produced optimistic estimates of prognostic performance. External validation is required.

The analysis was restricted to patients who survived to discharge and had paired measurements. Missing samples, including those not collected because of patients' clinical condition, may have introduced selection bias. Consequently, the findings cannot be extrapolated directly to all patients presenting with ADHF, particularly those who died before recompensation. Pandemic-related disruption also affected recruitment, sample availability and completion of quality-of-life follow-up.

Recompensation was determined by the treating HF specialist without a standardized congestion score, introducing potential variability in assessment. Sampling intervals differed between patients, and two measurements could not characterize the full course of sST₂ during hospitalization. Further serial measurements might clarify these patterns, although their additional prognostic value remains uncertain.

Participants were predominantly White, and all had an LVEF <50%, limiting generalisability to more diverse populations and patients with preserved LVEF. Finally, the observational design establishes associations but cannot determine whether sST₂-guided management improves outcomes.

5. Conclusion

Higher sST₂ concentrations at clinician-determined decompensation and recompensation were associated with one-year mortality in patients hospitalized with ADHF and an LVEF <50%. Unchanged or increased sST₂ also identified patients with higher mortality in unadjusted analysis, whereas percentage change showed limited discrimination. These findings support further evaluation of sST₂ as an adjunctive prognostic biomarker. The cohort-derived cut-off of 66.5 ng/mL at decompensation requires external validation, and studies should establish whether sST₂ adds useful information to clinical assessment and natriuretic peptides or improves outcomes when used to guide care.

CRediT authorship contribution statement

Brian Li: Conceptualization, Evidence synthesis, Formal analysis, Writing – Original Draft; Michaela Dowley: Validation, Investigation; Samantha Langley: Project administration; Paraic Cliffe: Project administration; Stephen Whiting: Resources, Validation; Archana Ganapathy: Investigation; Anthony D. Dimarco: Writing – Review & Editing; Rossella Barbagallo: Writing – Review & Editing; Jason N. Dungu: Supervision, Review & Editing; Henry Oluwasefunmi Savage: Supervision, Critical Revision to structure and intellectual content of manuscript, Guarantor. All authors approved the final manuscript.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study included organ or tissue donors. This study includes human biological material and consent was obtained by donors, or their next of kin or legal representatives, for use in this study and for publication of the article. The samples used in this research were not sourced from executed prisoners or prisoners of conscience.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the East of England - Cambridgeshire and Hertfordshire Research Ethics Committee. (Approval No. 19/EE/0269).

The results of this clinical trial and any associated work have been posted in a registry. This clinical trial was registered with number 268762 (Integrated research application system (IRAS)).

Sources of funding

None.

Declaration of competing interest

B. Li received an unrestricted educational grant from Abbott Medical paid to institution to support his salary as a research fellow. All other authors have no conflicts of interest.

Data availability

Individual Patient Data from this study cannot be shared publicly due to the restrictions of ethical approval obtained, internal regulations, patient consent, and data regulations. Researchers are invited to contact the corresponding author to discuss collaboration.