Soluble Urokinase Plasminogen Activator Receptor Levels and Outcomes in Patients with Heart Failure

Abstract

Background:

Soluble urokinase-type plasminogen activator receptor (suPAR) is a marker of immune activation and pathogenic factor for kidney disease shown to predict cardiovascular outcomes including heart failure (HF) in various populations. We characterized suPAR levels in patients with HF and compared its ability to discriminate risk to that of B-type natriuretic peptide (BNP).

Methods and Results:

We measured plasma suPAR and BNP levels in 3,437 patients undergoing coronary angiogram and followed for a median of 6.2 years. We performed survival analyses for the following outcomes: all-cause death, cardiovascular death, and hospitalization for HF. We then assessed suPAR’s ability to discriminate risk for the aforementioned outcomes. We identified 1116 patients with HF (age 65±12, 67.2% male, 20.0% Black, 67% with reduced ejection fraction). The median suPAR level was higher in HF compared to those without HF (3370 [IQR 2610–4371] vs. 2880 [IQR 2270–3670] pg/mL, respectively, P<0.001). In patients with HF, suPAR levels (log-base 2) were associated with outcomes including all-cause death (adjusted hazard ratio aHR 2.30, 95%CI[1.90–2.77]), cardiovascular death (aHR 2.33 95%CI[1.81–2.99]) and HF hospitalization (aHR 1.96, 95%CI[1.06–1.25]) independently of clinical characteristics and BNP levels. The association persisted across subgroups and did not differ between patients with reduced or preserved ejection fraction, or those with ischemic or non-ischemic cardiomyopathy. Addition of suPAR to amodel including BNP levels significantly improved the C-statistic for death (Δ0.027), cardiovascular death (Δ0.017) and hospitalization for HF (Δ0.017).

Conclusions:

SuPAR levels are higher in HF compared to non-HF, are strongly predictive of outcomes, and combined with BNP, significantly improved risk prediction.

Lay Summary:

•Soluble urokinase plasminogen activator receptor (suPAR) is a circulating protein of immune origin notorious for its involvement in kidney disease, and which levels have been found to predict the onset of heart failure.

•Given the pathophysiologic link between heart failure and kidney disease, we sought to examine suPAR levels in patients with heart failure.

•We measured suPAR in 1116 patients with heart failure, and found that levels were strongly predictive of outcomes independently of risk factors and above and beyond the myocardial marker BNP. SuPAR level may be useful as an adjunctive measure for risk stratification of patients with heart failure.

Introduction

Heart failure (HF) affects nearly 6 million Americans, and despite significant advances in its management, major challenges remain such as its rising prevalence and resource utilization.¹ In recent years, a wide array of biomarkers has been explored as means to provide insight into the pathophysiology of HF, identify potential therapeutic targets, detect early signs of myocardial dysfunction to aid in early diagnosis or guide therapy, and lastly refine risk stratification.² Most candidate biomarkers such as ST-2, galectin-3, B-type natriuretic peptides (BNP), and high-sensitivity troponin (hs-TnI) reflect downstream aspects of the pathophysiology of HF such as increased myocardial stretch, cardiomyocyte injury and fibrosis.^3,4

SuPAR is the circulating form of uPAR, a glycosyl-phosphatidylinositol-anchored (GPI) three-domain (DI, DII, and DIII) receptor protein encoded by the Plaur gene, and expressed on a variety of immune cells.(9, 10) SuPAR is produced by cleavage of membrane-bound uPAR at the GPI anchor in response to inflammatory stimuli, and levels are thought to reflect the aggregate activity of the uPAR system: inflammatory pathways, proteolysis, and extracellular matrix remodeling.(12) SuPAR has been implicated in the pathogenesis of kidney disease – which occurs in over half of patients with HF and portends a worse prognosis.^5–13 Most importantly, suPAR levels have consistently been shown to be associated with incident cardiovascular disease (CVD) and poor outcomes in various groups including the general population, critically ill patients, and those with CVD, HIV, cancer, and kidney disease.^8,14–23 SuPAR levels correlated with NT-proBNP levels and were associated with incident HF in a population-based study.²⁴ In experimental models, uPAR was involved in the modulation of TGFβ-1 related post-infarction fibrosis and scar healing,^25,26 and regulation of apoptosis, suggesting a possible direct role for suPAR in the pathogenesis of HF.²⁷

Given its involvement in both inflammation and kidney dysfunction, suPAR levels reflect non-myocardial pathophysiologic processes in HF and may provide prognostic information beyond traditional cardiovascular biomarkers. We sought to (1) characterize the relationship between suPAR levels and HF characteristics, (2) determine whether high suPAR levels are associated with adverse outcomes in patients with HF and (3) incident HF in patients without HF, independently of kidney function and BNP levels.²⁸ Lastly (4), we assess the ability of suPAR levels in improving risk discrimination in patients with HF.

Methods

Study Design and Population

We measured plasma suPAR and BNP levels in 3437 adult participants enrolled in the Emory Cardiovascular Biobank, a prospective cohort of patients undergoing cardiac catheterization for suspected or confirmed coronary artery disease (CAD) at Emory Healthcare sites in Atlanta, GA, between 2004 and 2010.^29,30 A total of 1116 (32.5%) of enrollees carried a diagnosis of HF. Patients without suPAR measurements due to depleted blood samples and those with less than 30-days follow-up were excluded (n=57). Other exclusion criteria were history of cardiac transplant, congenital heart disease, severe anemia, recent blood transfusion, acute myocarditis, history of active inflammatory disease or cancer receiving therapy. Participants were interviewed to collect demographic characteristics, medical history, medication use, and behavioral habits. Medical records were reviewed to confirm self-reported medical history.

Heart failure was defined by the presence of self-reported history of HF, physician diagnosis of HF, or a left ventricular ejection fraction of <50% noted in the medical record at the time of enrollment. We classified patients according left ventricular ejection fraction: (1) HF with reduced ejection fraction (HFrEF) was as a EF <50%, and (2) HF with preserved EF (HFpEF) defined as EF≥50%. Ischemic cardiomyopathy was defined as the presence of HFrEF in addition to at least one of the following: history of myocardial infarction, coronary revascularization, or presence of obstructive coronary artery disease (≥70% stenosis in at least one epicardial artery on angiogram). Non-ischemic cardiomyopathy was defined as HFrEF in the absence of the obstructive coronary artery disease. The study was approved by the institutional review board at Emory University (Atlanta, GA). All subjects provided written informed consent at the time of enrollment.

Sample Collection and Measurement of suPAR, BNP

Fasting arterial blood samples were collected at the time of catheterization and stored at −80°C for a mean duration of 4.9 years. Plasma levels of suPAR were measured by ELISA (suPARnostic kit; ViroGates, Copenhagen, Denmark) with a lower detection limit of 100 pg/mL and intra- and inter-assay variation of 2.75% and 9.17%, respectively. Plasma B-type natriuretic peptide was measured using ARCHITECT BNP assay (Abbott Diagnostics, Chicago, IL) with measurement range between 10 to 5000 pg/mL.

Follow-up and Outcomes

We conducted follow-up as previously described to identify pre-specified incident adverse cardiovascular outcomes including all-cause death, death from cardiovascular causes and lastly, new diagnosis or hospitalization for HF.^30,31 In brief, follow-up was conducted by phone, electronic medical record review, social security death index and state records, and adjudication was conducted by personnel blinded to the biomarker data. The median follow-up of 6.2 years (IQR 3.5–8.0 years).

Statistical Analysis

We reported subject characteristics as descriptive statistics with means, medians, standard deviations and ranges. Differences between groups were assessed using the t-test for continuous variables, and chi-square or Fisher exact tests for categorical variables where appropriate. Two-tailed P-value ≤0.05 were considered statistically significant. For non-normally distributed variables such as suPAR and BNP levels, the Mann-Whitney U test was used to compare groups in unadjusted analyses. We used Spearman-Rank to assess the correlation between suPAR and BNP levels. We used linear regression to determine whether suPAR levels were associated with HF independently of clinical characteristics and BNP levels; including in a single model log-transformed suPAR (base 2), age, gender, race (Black compared to non-Black), body mass index, smoking history, hypertension, diabetes mellitus, estimated glomerular filtration rate (eGFR calculated using the CKD-EPI equation),³² history of coronary artery disease, angiotensin-converting enzyme inhibitors or angiotensin-receptor blocker use, beta-blocker therapy and ejection fraction. We have selected these variables based on biological relevance and their known associations with suPAR and HF.

Survival Analyses.

For multivariable analyses, suPAR levels were examined both as a categorical variable stratified by quartiles defined in separate subgroups (suPAR quartiles in patients with HF: <2610, 2610–3369, 3370–4371, and >4371 pg/mL; suPAR quartiles in patients without HF: <2270, 2270–2879, 2880–3669, >3669 pg/mL), and as a continuous variable after log-transforming (base 2) to a normal distribution and reported as “per 100% increase in suPAR”. We plotted Kaplan-Meier curves and used Cox proportional hazards modeling to examine the association between suPAR and all-cause death, cardiovascular death, and new diagnosis or hospitalization for HF in separate models: Model 0 included suPAR alone. Model 1 included suPAR and age, gender, race (Black compared to non-Black), body mass index, smoking history, hypertension, diabetes mellitus, estimated glomerular filtration rate (eGFR calculated using the CKD-EPI equation),³² history of coronary artery disease, angiotensin-converting enzyme inhibitors or angiotensin-receptor blocker use, beta-blocker therapy and ejection fraction. Lastly, model 2 incorporated suPAR, the aforementioned clinical characteristics, and BNP levels. We used Fine and Gray Cox regression modeling to account for the competing risk of non-cardiovascular death when examining cardiovascular death as an outcome and all-cause death when examining incident HF as an outcome.³³ Sensitivity analyses was performed to explore whether the association between suPAR and outcomes differed according to HF subtype (HFrEF vs. HFpEF and ischemic versus non-ischemic cardiomyopathy), the presence of chronic kidney disease, and BNP levels stratified by a cut-off of 100 pg/ml.

We examined the incremental value of adding suPAR and BNP levels to clinical models for predicting all-cause death and hospitalization for HF. C-statistics, integrated discrimination improvement and net reclassification improvement were calculated based on Cox proportional hazards modeling to evaluate the predictive abilities of models with and without biomarkers using the survC1 and survIDINRI R packages by Uno et al.^34–36 Analyses were performed using IBM SPSS Statistics Version 22, (Armonk, NY, USA), SAS 9.4 (Cary, NC) and R 3.2.2 (R Core Team, Vienna, Austria).

Results

Characteristics of the 3,437 patients stratified by diagnosis of HF are shown in Table 1. Patients with HF were more likely to be Black and have a higher burden of cardiovascular risk factors, coronary artery disease and chronic kidney disease (Table 1). Plasma suPAR levels were 17% higher (3370 vs 2880 pg/mL, P<0.001) in patients with HF compared to those without HF. In multivariable analyses, suPAR remained associated with HF after adjusting for clinical characteristics including medications, eGFR and BNP levels (Table S1). We found a modest positive correlation between suPAR and BNP levels (r=0.352, P<0.001), with a wide range in suPAR levels even in patients with BNP levels <100 pg/ml (Figure S1). Higher BNP levels remained independently associated with higher suPAR levels independently of demographics and clinical risk factors (Table S2). Of note, beta-blocker or ACEi/ARB use were not associated with suPAR levels (Table S2)

Table 1.

Demographics and Clinical Characteristics of the Emory Biobank Cohort Stratified by Diagnosis of Heart Failure

				Heart Failure Subtype
Variables	No HF (n=2321)	HF (n=1116)	P-valuea	HFrEF (n=750)	HFpEF (n=366)	P-valueb
Age, years, mean (SD)	63 (11)	65 (12)	<0.001	65(11)	65(12)	0.31
Male, n (%)	1468(63.2%)	750 (67.2%)	0.023	557 (74.3%)	193 (52.7%)	<0.001
Black race, n (%)	365(15.7%)	223 (20.0%)	0.002	142 (18.9%)	81 (22.1%)	0.21
Body mass index, kg/m2, mean (SD)	30(6)	29 (6)	0.09	29 (6)	30(7)	0.002
Smoking history, n (%)	1504 (64.8%)	756(67.7%)	0.040	531 (70.8%)	225 (61.5%)	0.002
Hypertension, n (%)	1634 (70.4%)	846 (75.8%)	<0.001	549 (73.2%)	297 (81.1%)	0.004
Diabetes mellitus, n (%)	651 (28.0%)	420(37.6%)	<0.001	271 (36.1%)	149 (40.7%)	0.14
Estimated glomerular filtration rate, mL/min/1.73 m2	76(21)	69 (23)	<0.001	69 (23)	68 (25)	0.36
History of coronary artery disease, n (%)	1548(66.7%)	906(81.2%)	<0.001	632 (84.3%)	274 (74.9%)	<0.001
On renin-angiotensin system antagonists, n (%)	1271 (54.8%)	777 (69.6%)	<0.001	544 (72.5%)	233 (63.7%)	0.002
On beta-blockers, n (%)	1373 (59.2%)	858 (76.9%)	<0.001	618(82.4%)	240 (65.6%)	<0.001
Ejection fraction, %, mean (SD)	59 (6)	42 (13)	<0.001	36(10)	56(6)	<0.001
SuPAR, pg/mL, median (IQR)	2880 (2270–3670)	3370 (2610–4371)	<0.001	3340 (2600–4350)	3440 (2640–4390)	0.70
BNP, pg/dL, median (IQR)	50 (20–115)	120(43–341)	<0.001	142 (52–425)	85 (29–192)	<0.001

Values are mean (standard deviation (SD), median (interquartile range IQR), or n (%) as noted. BNP: B-type natriuretic peptide; suPAR: soluble urokinase plasminogen activator receptor. History of coronary artery disease is defined as prior revascularization, myocardial infarction or obstructive disease on coronary angiogram (at least one ≥50% stenotic epicardial artery).

^aP-value is for the comparison between patients with and without heart failure.

^bP-value is for comparison between patients with heart failure with reduced ejection fraction (HFrEF) and those with preserved ejection fraction (HFpEF).

When stratified by HF subtype, patients with HFrEF were more likely to be male and have a history of coronary artery disease compared to those with HFpEF (Table 1). There were no significant differences in suPAR levels between patients with HFrEF (n=750) and HFpEF (n=366) (3340 vs. 3440 pg/ml, P=0.70), or between patients with ischemic (n=587) and non-ischemic cardiomyopathy (n=529) (3320 vs. 3450 pg/ml, P=0.25).

SuPAR and Outcomes in Heart Failure

During a median follow-up time of 6.2 years (IQR 3.5–8.0 years), there were 473 (42.3%) deaths from all causes, 265 (23.7%) deaths from cardiovascular causes, and 195 (17.5%) readmissions for HF. In unadjusted analyses, HF patients in increasingly higher suPAR quartiles had poorer survival free of all cause death, cardiovascular death, and HF readmissions (Fig. 1). At 5 years follow-up, 48% of patients in the highest suPAR quartile (>4371 pg/ml) died, compared to 10% of those within the lowest quartile (<2610 pg/ml). Cardiovascular death and hospitalization for HF occurred in 33% and 35% of patients in the highest suPAR quartile compared to 6% and 10% of patients in the lowest suPAR quartile (P<0.001).

Fig. 1.

Death and Hospitalization for Heart Failure Stratified by SuPAR Quartiles. Kaplan Meier survival curves for (A) all-cause death, (B) cardiovascular death, and (C) hospitalization for heart failure (HF) in patients with HF (n=843), and incident HF (D) in patients without heart failure at enrollment (n=2372), stratified by suPAR quartiles.

SuPAR levels (per 100% increase) were associated with a 2.60-fold (95%CI [2.29–2.95]) increase in the risk of all-cause death, 2.79-fold (95%CI [2.38–3.29]) increase in the risk of cardiovascular death, and 2.34-fold (95%CI [1.94–2.83]) increase in the risk of hospitalization for HF. The associations between suPAR and the aforementioned outcomes were only minimally attenuated after adjusting for demographics, clinical characteristics including kidney function, EF, and medication regimen (Fig. 2, Table S3). After adjusting for BNP, suPAR levels (per 100% increase) remained a significant predictor all-cause death (aHR 2.30, 95%CI[1.90–2.77]), cardiovascular death (aHR 2.33 95%CI[1.81–2.99]) and HF hospitalization (aHR 1.96, 95%CI[1.48–2.61]) (Fig. 2, Table S3). Associations remained consistent when accounting for the competing risk of non-cardiovascular death (aHR 2.12, 95%CI[1.67–2.69] per 100% increase in suPAR levels for cardiovascular death) or all cause-death (aHR 1.97, 95%CI[1.47–2.64] per 100% increase in suPAR levels for HF hospitalization. When examined as a categorical variable, the highest suPAR quartile (>4,371 pg/ml) associated with a 3.65-fold (95%CI [2.27–5.88) increase in the risk of death, 3.54-fold (95%CI [2.51–4.99]) increase in the risk of cardiovascular death, and 2.57-fold (95%CI [1.53–4.32]) increase in the risk of hospitalization for HF compared to the first quartile (<2610 pg/ml) (Table S2).

Fig. 2.

Multivariable Modeling of SuPAR Levels and Outcomes in Patients with Heart Failure. Bar graphs depicting hazard ratios and 95% confidence interval for the association between suPAR levels (log-base 2) and all-cause death, cardiovascular death, hospitalization for heart failure (HF) and incident HF, stratified by model. Model 0 included suPAR alone. Model 1 included suPAR and age, gender, race (Black compared to non-Black), body mass index, smoking history, hypertension, diabetes mellitus, estimated glomerular filtration ratem history of coronary artery disease, angiotensin-converting enzyme inhibitors or angiotensin-receptor blocker use, beta-blocker therapy and ejection fraction. Lastly, model 2 incorporated suPAR, the aforementioned clinical characteristics, and BNP levels.

In sensitivity analyses, the association between suPAR and outcomes was consistent across relevant subgroups including HFrEF and HFpEF, ischemic and non-ischemic cardiomyopathy, and patients with low (<100 pg/ml) or high (≥100 pg/ml) BNP levels (Fig. 3). We did find the association between suPAR and outcomes to be stronger in men compared to women, in patients without hypertension compared to those with hypertension, in patients with diabetes compared to those without diabetes, and in patients with chronic kidney disease compared to those without kidney disease (Fig. 3).

Fig. 3.

SuPAR Levels and Outcomes across Subgroups of Patients with Heart Failure. Forest plot depicting the hazard ratios for the association between suPAR levels (log-base 2) and all-cause death, cardiovascular death, hospitalization for heart failure (HF), stratified by patient subgroup. ^†P-value reported is for the interaction term for the specified variable*suPAR.

Lastly, patients without known HF and high suPAR levels (>3669 pg/ml) had a 3.59-fold (95%CI [1.42–9.10]) increase in the risk of incident HF after adjusting for clinical characteristics and BNP levels. When accounting for the competing risk of all-cause death, suPAR levels (per 100% increase) remained associated with increased risk of incident HF (aHR 1.61, [95%CI[1.06–2.43]).

Risk Discrimination

We tested the incremental value in risk prediction of adding suPAR and BNP levels to a risk factor model which includes age, gender, race, body mass index, smoking history, hypertension, diabetes, estimated glomerular filtration rate at enrollment, history of coronary artery disease, ejection fraction, use of renin-angiotensin system inhibitors, and use of beta-blockers. The model including the aforementioned factors, suPAR and BNP exhibited good risk discrimination, with a C-statistics of 0.732 (95%CI [0.702–0.761]) for all-cause death, 0.752 (95%CI[0.720–0.785]) for cardiovascular death, and 0.731 (95%CI[0.680–0.782]) for HF hospitalization. Both BNP and suPAR levels significantly improved risk discrimination and reclassification indices (C-statistic, net reclassification improvement (NRI) and integrated discrimination improvement (IDI)) for all three outcomes (Table 2). The magnitude of improvement in risk discrimination of death (ΔC-statistic 0.046 vs 0.026) and cardiovascular death (ΔC-statistic 0.028 vs. 0.021) was larger with the addition of suPAR compared to BNP, and similar between both biomarkers for HF hospitalization (ΔC-statistic 0.41 vs 0.40). Combining both suPAR and BNP yielded the strongest improvement in risk discrimination for all three outcomes (Table 2). Adding suPAR to a model with BNP significantly improved the C-statistic for death (ΔC-statistic 0.027, 95%CI [0.009–0.045]), cardiovascular death (ΔC-statistic 0.017, 95%CI[0.001–0.034]) and hospitalization for HF (ΔC-statistic 0.017, 95%CI[0.001–0.032]).

Table 2.

Risk Discrimination Metrics for All-cause Death and Hospitalization for HF in Patients with HF

Models	C-statistic (95% CI)	ΔC-statistic (95% CI)a	IDI (95%CI)	NRI (95%CI)
All-cause Death (n=473)
Model 1: Risk factors only	0.678 (0.646–0.710)	-	-	-
Model 2: Model 1 + BNP	0.705 (0.671–0.738)	0.026 (0.010–0.043)	0.041 (0.017–0.067)	0.192 (0.112–0.281)
Model 3: Model 1 + suPAR	0.724 (0.697–0.752)	0.046 (0.025–0.067)	0.069 (0.039–0.098)	0.259 (0.154–0.329)
Model 4: Model 1 + BNP + suPAR	0.732 (0.702–0.761)	0.053 (0.032–0.075)	0.081 (0.050–0.086)	0.278 (0.170–0.338)
Cardiovascular Death (n=265)
Model 1: Risk factors only	0.714 (0.676–0.753)	-	-	-
Model 2: Model 1 + BNP	0.735 (0.699–0.771)	0.021 (0.003–0.038)	0.040 (0.017–0.066)	0.232 (0.104–0.305)
Model 3: Model 1 + suPAR	0.742 (0.705–0.779)	0.028 (0.008–0.048)	0.059 (0.027–0.087)	0.266 (0.121–0.334)
Model 4: Model 1 + BNP + suPAR	0.752 (0.720–0.785)	0.038 (0.016–0.060)	0.072 (0.043–0.099)	0.275 (0.180–0.359)
Hospitalization for Heart Failure (n=195)
Model 1: Risk factors only	0.674 (0.624–0.725)	-	-
Model 2: Model 1 + BNP	0.714 (0.669–0.759)	0.040 (0.011–0.068)	0.051 (0.009–0.095)	0.345 (0.099–0.470)
Model 3: Model 1 + suPAR	0.715 (0.668–0.762)	0.041 (0.011–0.071)	0.039 (0.005–0.086)	0.139(−0.090–0.357)
Model 4: Model 1 + BNP + suPAR	0.731 (0.680–0.782)	0.057 (0.023–0.090)	0.069 (0.027–0.130)	0.241 (0.042–0.460)

Models 1 through 4 includes age, gender, race, body mass index, smoking history, hypertension, diabetes, estimated glomerular filtration rate at enrollment, history of coronary artery disease, ejection fraction, use of renin-angiotensin system inhibitors, and use of beta-blockers. BNP: b-type natriuretic peptide; CI: confidence interval. IDI: integrated discrimination improvement; NRI: net reclassification improvement.

^aChange in C-statistic is relative to Model 1.

Addition of suPAR to a risk factor model significantly led to significantly improved reclassification of patients as assessed using the IDI and NRI in the prediction of all-cause death and cardiovascular death (Table 2). The magnitude of reclassification with the addition of suPAR was higher than that of BNP in the prediction of all-cause and cardiovascular mortality but not hospitalization for HF (Table 2). Lastly, addition of suPAR to a risk factor model that includes BNP led to additional improvement in reclassification indices for the predication of all-cause death (IDI 0.041 95%CI[0.013–0.071]; NRI 0.193 95%CI[0.081–0.279]), cardiovascular death (IDI 0.032 95CI[0.009–0.057]; NRI 0.195 95%CI[0.072–0.270]), but not hospitalization for HF (IDI 0.019 95%CI [−0.002–0.049]; NRI 0.025 (−0.169–0.220]).

Discussion

In this large prospective cohort of patients undergoing cardiac catheterization for suspected or confirmed coronary artery disease, we found that patients with HF had significantly higher suPAR levels compared to those without HF, independently of kidney function and the subtype of HF. SuPAR levels correlated modestly with BNP, with a substantial proportion of patients with low BNP (<100 pg/ml) still having high suPAR levels. Most importantly, suPAR levels were strongly predictive of incident outcomes including all-cause death, cardiovascular death, and hospitalization for HF, even after adjusting for clinical characteristics including eGFR, ejection fraction, and BNP levels. We found suPAR levels to also be predictive of incident HF independently of BNP levels. Lastly, the addition of suPAR to a risk factor model which includes BNP significantly improved risk discrimination and reclassification metrics – notably in the prediction of all-cause and cardiovascular death. A model combining suPAR with BNP yielded the strongest improvement in risk discrimination.

This is a large study reporting on the association between suPAR and HF characteristics as well as outcomes in this patient population, and is in line with findings from a smaller study of 319 patients with HF.³⁷ We find that suPAR is associated with HF independently of BNP levels and clinical characteristics and provide data on the association between suPAR levels, mortality as well as hospitalization for HF. Another report examined the association between suPAR, NT-proBNP and incident HF in subjects enrolled from the general population enrolled in the Malmo Diet and Cancer Study.²⁴ Notable findings include a positive correlation between suPAR and NT-proBNP, and a 23% increased risk of incident HF in subjects in the highest suPAR level tertile, even after adjusting for NT-proBNP.²⁴ In addition to confirming these findings in a high risk patient group, we find that the association between suPAR and incident HF is independent of underlying coronary artery disease or kidney function, suggesting that it is unlikely to be mediated by coronary atherosclerosis, subclinical myocardial injury or kidney dysfunction. While the association was present across patient subgroups, we did find it to be stronger in men and in patients without hypertension, without diabetes mellitus or chronic kidney disease. The exact reason for these differences is unclear; hypertension, diabetes mellitus and chronic kidney disease are risk factors known to be independently associated with higher suPAR levels that account at least partially for its impact on outcomes. The higher suPAR levels in patients without co-morbidities likely reflect a higher burden of chronic inflammation and potentially undiagnosed associated conditions such as rheumatologic or inflammatory bowel diseases. In-depth characterization of individuals with high suPAR levels but no known CV risk factors will help shed light on the matter.

The association between suPAR and outcomes was strongest for all-cause and cardiovascular mortality; with suPAR levels outperforming and being additive to BNP in the prediction of the aforementioned outcomes, but not hospitalization for HF as measured by the change in C-statistic, IDI and NRI metrics. These findings can be explained by suPAR’s nature as a stable biomarker of chronic extra-cardiac processes that ultimately lead to death, while BNP levels are strongly linked to severity of HF and are variable according to volume status and interventions. We found that combining both suPAR and BNP leads to the strongest improvement in risk discrimination, suggesting a multi-marker approach incorporating cardiac and non-cardiac biomarkers such as suPAR and BNP could help optimize risk prediction in patients with HF.³⁸

A large number of biomarkers have been proposed in recent years as potential candidates for diagnostic and prognostic use in HF.^3,4 Notably, ST2, galectin-3 and hs-TnI are cardiac-specific markers that reflect various aspects of myocardial injury, and their levels correlate well with severity of HF as well as outcomes.³ The incorporation of a biomarker that potentially reflects both cardiac and extra-cardiac pathophysiologic processes may improve risk stratification compared to the use of cardiac-specific markers alone. SuPAR may be an example of such a marker for several reasons. It has consistently been associated with poor outcomes in various groups including the general population, critically ill patients, and those with cardiovascular disease, HIV, cancer, and kidney disease.^8,14–23 SuPAR levels correlate with cardiovascular risk factors and subclinical vascular dysfunction, and can be modified by smoking cessation.^14,20,39 While reflective of immune activation, suPAR levels remain stable during episodes of acute stress such as acute myocardial infarction or surgery, and thus suPAR differs from other inflammatory biomarkers such as C-reactive protein and interleukin-6, which are acute phase reactants.^40,41 We have recently shown that it is associated with coronary artery disease severity, and predicts renal dysfunction; both major contributors to the pathogenesis of HF.^8,17 While we found a modest correlation between suPAR and BNP levels, suPAR levels varied widely, even in patients with a BNP<100 pg/ml and remained associated with outcomes in this patient subgroup. This observation confirms that suPAR levels are not dependent on underlying volume status. SuPAR levels are also not associated with beta-blocker or renin-angiotensin system inhibitors, and more recently have been found not to be altered by dapagliflozin use.⁴²

In summary, the stability of suPAR levels, its strong association with relevant outcomes independently of clinical characteristics, BNP levels and medical therapy for HF make it a potentially valuable biomarker of risk in patients with HF. SuPAR levels are likely a reflection of upstream pathologic processes leading to chronic organ damage such as atherosclerotic disease and renal dysfunction, and subsequently poor outcomes. We can envision a potential use in measuring suPAR and other biomarkers as a strategy to personalize the care. For example, differentiating between a low risk and a higher risk group of patients admitted with HF using biomarkers could guide the allocation of post-discharge care resources to the population at risk, thus lessening the overall cost burden in managing the disease.³⁸ Prospective studies examining the clinical utility of measuring suPAR and other novel biomarkers in HF are needed to justify their routine measurement in this patient population.³⁸

Strengths and Limitations

Our study has several strengths; the large sample size, long follow-up and large number of events allowing for analyses of individual rather than combined outcomes. The cohort included women (36%), Blacks (24%) and patients with a range of LV function and CAD severity. Moreover, the detailed clinical characterization of subjects allows for adjustment for a variety of potential confounders including renal function and medications. The study is limited by its observational nature, and the lack of repeated measures of suPAR and BNP that would have provided valuable information on the relationship between changes in levels and outcomes. The wide confidence interval observed for the association between suPAR and incident HF is likely a reflection of the small number of incident HF events (n=80), but may also be related to existence of differences in the association across subpopulations (for example, patients with and without kidney disease) that could be explored in subsequent studies. Lastly, patients included in this cohort were referred for a coronary angiogram for suspected coronary artery disease and may not be representative of the overall HF population. Our study population is likely enriched for patients with a higher burden of atherosclerotic disease – known to be associated with higher suPAR levels. Nevertheless, we found no significant differences in suPAR levels between patient with and without ischemic cardiomyopathy, and the association between suPAR and outcomes did not differ according to these subgroups.

Conclusions and Clinical Implications

SuPAR levels are higher in HF and are significantly predictive of adverse incident outcomes. In those without HF, higher suPAR levels are independently associated with greater incidence of HF. As a marker of immune activation, suPAR likely reflects upstream pathologic processes such as oxidative stress and inflammation common to various disease states including HF. The predictive value of suPAR levels is additive to data provided by the myocardial-specific marker BNP, and thus may be valuable in identifying a population at higher risk of poor outcomes. Whether suPAR measurements will alter management and resource allocation in the care of patients with HF warrants further study.³⁸