Soluble urokinase plasminogen activator receptor and outcomes in HFpEF: A TOPCAT ancillary study

Abstract

Aims

Inflammation is postulated to be a key pathogenic mechanism in heart failure with preserved ejection fraction (HFpEF). Soluble urokinase plasminogen activator receptor (suPAR), a regulator of innate immune activity, is associated with incident heart failure; however, its role in HFpEF remains unclear. We aimed to elucidate the role of suPAR in HFpEF outcomes.

Methods

In this secondary analysis of the TOPCAT trial's North American cohort, suPAR was measured at baseline and 1 year in a subset of patients with HFpEF (n = 406) treated with either spironolactone or placebo. We assessed the association between suPAR levels and adverse outcomes, whether spironolactone influenced suPAR levels and whether the association between spironolactone and outcomes is dependent on suPAR levels. The primary outcome was a composite of cardiovascular death, cardiac arrest or hospitalization for heart failure management.

Results

The mean age of participants was 69.5 years, and 46.6% were female. After a median follow‐up of 2.9 years, 19.9% experienced the primary outcome event. The 5‐year cumulative incidence of the primary outcome in the highest tertile of suPAR (>3.93 ng/mL) was 44%, compared with 14% in the lowest tertile (≤2.94 ng/mL) (P = 0.001). Spironolactone did not significantly change suPAR levels at 1 year, nor was its effect on outcomes modified by baseline suPAR (P for interaction = 0.6). In multivariable analysis, each doubling of baseline suPAR levels was associated with nearly twofold increased risk of the primary outcome, independent of traditional risk factors and natriuretic peptide (NP) levels (HR 1.94 [95% CI 1.33–2.83]). suPAR's risk discrimination ability was superior and additive to that of NP.

Conclusions

While suPAR levels independently predict poor outcomes in HFpEF patients, spironolactone does not modulate this inflammatory pathway. The findings suggest that suPAR represents a stable inflammatory biomarker in HFpEF, highlighting the need for further evaluation of targeted anti‐inflammatory strategies in this population.

Introduction

Heart failure with preserved ejection fraction (HFpEF) is a major cause of cardiovascular morbidity and mortality worldwide. ¹ Evidence suggests inflammation plays an important role in the pathophysiology of HFpEF. ² , ³ , ⁴ , ⁵ Risk factors for HFpEF, such as obesity, diabetes, hypertension and chronic kidney disease (CKD), lead to a chronic inflammatory state characterized by the release of pro‐inflammatory cytokines. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ Inflammation is associated with oxidative stress, endothelial dysfunction, myocardial fibrosis and diastolic dysfunction. ⁶ , ⁷ While inflammation plays a significant role in HFpEF development, the degree of inflammatory burden varies substantially among patients, which may explain the inconsistent results observed in clinical trials targeting inflammatory pathways. ⁹ , ¹⁰ , ¹¹ Recent evidence suggests that HFpEF patients exhibit heterogeneous inflammatory profiles, highlighting the critical need for better patient stratification. ¹² By using robust inflammatory biomarkers to identify patients with a predominantly inflammatory phenotype, clinicians could better direct anti‐inflammatory therapies to those most likely to benefit, thereby optimizing treatment outcomes and advancing the field towards more personalized heart failure care.

The Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist (TOPCAT) trial specifically examined the efficacy of spironolactone—a mineralocorticoid receptor antagonist (MRA) with anti‐inflammatory effects—on the prognosis of patients with HFpEF. The trial did not find a significant reduction in the incidence of the primary outcome, a composite of cardiovascular death, cardiac arrest or hospitalization for heart failure management in patients treated with spironolactone compared to placebo. ¹⁰ , ¹¹ However, a post hoc analysis suggested that higher risk patients may derive benefit from the therapy. ¹⁰ , ¹¹ , ¹³ Reliable biomarkers for identifying patients who might benefit from spironolactone remain elusive. ¹⁴

Soluble urokinase plasminogen activator receptor (suPAR) is an immune‐derived signalling glycoprotein, the levels of which reflect the activity of chronic inflammation. ¹⁵ suPAR is strongly associated with incident kidney disease, atherosclerotic disease and type II diabetes mellitus and is predictive of poor outcomes, including mortality, across patient populations. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ Elevated suPAR levels have been associated with incident heart failure and are strongly predictive of heart failure‐related outcomes, independent of clinical risk factors or natriuretic peptides (NPs). ¹⁶ , ²⁵ , ²⁶ , ²⁷ , ²⁸ Whether spironolactone improves outcomes through the suPAR pathway in patients with HFpEF is unknown. We leveraged HFpEF patients enrolled in the North American arm of the TOPCAT trial to (1) characterize the association between suPAR and outcomes in patients with HFpEF, (2) determine whether spironolactone reduces suPAR levels, (3) assess whether the association between spironolactone outcomes is dependent on suPAR levels and (4) compare suPAR's risk discrimination ability to that of NPs in patients with HFpEF.

Methods

Study design and population

We performed a secondary analysis of data from participants in the North American cohort of the TOPCAT trial (NCT00094302), a randomized placebo‐controlled trial conducted from 2006 to 2012 that treated HFpEF patients with either spironolactone (escalating dose from 15 to 45 mg daily) or placebo. ¹⁰ The TOPCAT trial conducted a sub‐study to collect biologic samples from a subset of patients that had provided informed consent for collection and submission to the National Institute of Health's National Heart, Lung and Blood Institute Biological Specimen and Data Repositories Information Coordinating Center (BioLINCC) biorespository. ¹⁰ In 2020, we gained approval to receive deidentified samples from BioLINCC to measure suPAR in all available serum samples (n = 406), collected at baseline and 1‐year follow‐up. Detailed inclusion and exclusion criteria of TOPCAT have been previously published. ¹⁰ , ²⁹ Briefly, included patients were 50 years or older, with at least one symptom of HF, a left ventricular ejection fraction of 45% or greater, a controlled systolic blood pressure, and a serum potassium level of less than 5.0 mmol/L. Eligible patients also had a history of hospitalization for HF management within the past 12 months as a major part of their care, or elevated N‐terminal pro‐B‐type (NT‐proBNP) (≥360 pg/mL) levels or B‐Type NP (BNP) levels (≥100 pg/mL) within 60 days before randomization. Patients with a life expectancy of less than 3 years due to severe systemic illness or renal dysfunction were excluded. ¹⁰ Informed consent was obtained from participants as per the TOPCAT trial protocol. This study complies with the Declaration of Helsinki and was approved by the institutional review board at the University of Michigan Medicine (Ann Arbor, MI).

Data variables and outcome measures

The dataset obtained from the TOPCAT trial included demographics, comorbidities, baseline laboratory measures and treatment assignment (spironolactone or placebo). Baseline measurements of NPs, BNP (n = 99) or NT‐proBNP levels (n = 174) were available in a subset of patients (n = 273). Outcomes were assessed by an independent, blinded committee from the original TOPCAT trial. ¹⁰ The primary outcome was a composite of cardiovascular death, aborted cardiac arrest or hospitalization for HF.

Sample collection and measurement of suPAR

All serum samples from the TOPCAT BioLINCC repository were obtained and stored at −80°C until analysis. Plasma suPAR levels were measured using a commercially available enzyme‐linked immunosorbent assay with a double monoclonal antibody sandwich (ViroGates, Copenhagen, Denmark). The assay has a lower detection limit of 100 pg/mL and intra‐ and inter‐assay variations of 2.75% and 9.17%, respectively. suPAR measurements were obtained for all 406 participants at baseline, and a subset of 331 participants at baseline and 1‐year follow‐up. Laboratory technicians performing the suPAR assays were blinded to outcome status related to each sample.

Statistical analysis

Descriptive statistics for demographic data, comorbidities, laboratory measures and outcomes were stratified by suPAR tertiles: first tertile ≤2.94 ng/mL, second tertile >2.94 and ≤3.91 ng/mL and third tertile >3.91 ng/mL. Assessments of normality were conducted by visual inspection of histograms and Q‐Q plots. Skewed continuous variables were presented using medians with interquartile ranges (IQRs). Normally distributed continuous variables were presented using means with standard deviations (SDs), while categorical variables were presented using counts and percentages (n [%]).

Correlation assessments were performed using Pearson correlation for normally distributed variables while Spearman's rank was used for non‐normally distributed variables. We compared the suPAR tertiles using the Chi‐squared test, Fisher's exact or pairwise Wilcoxon rank‐sum tests for categorical variables, as appropriate. For continuous variables, we used the Kruskal–Wallis test for nonparametric data and one‐way ANOVA for parametric data based on assessment of normality. In the subset of patients with suPAR measured at both time points (n = 331), the change in suPAR levels from baseline to Year 1 were analysed for significance using the Wilcoxon signed‐rank paired t‐test for within‐group differences and Wilcoxon rank sum for between‐group differences.

We plotted Kaplan–Meier curves and used Cox proportional hazards modelling to assess the association between baseline suPAR levels and outcomes, as well as the interaction between suPAR levels and treatment arm (spironolactone vs. placebo). In multivariable linear regression models, continuous suPAR was log‐base 2‐transformed. In Cox proportional hazards models, we assessed suPAR levels as both tertiles, with the first tertile serving as a reference group, and as a continuous variable (log‐base 2‐transformed). Hazard ratios (HRs) and 95% confidence intervals (CIs) were reported. Relevant variables were included in the models a priori or if they changed the association of suPAR with the outcomes by greater than 10%. Sequential models were examined: Model 0 included suPAR alone; Model 1 incorporated age, sex and race; Model 2 further adjusted for clinical risk factors including continuous estimated glomerular filtration rate (eGFR), hypertension, diabetes, smoking history, body mass index (BMI) and history of cardiovascular disease (CVD); Model 3 included all aforementioned variables in addition to NP tertiles. NP tertiles were created by transforming BNP and NT‐proBNP into tertiles and combining tertiles of each into a single categorical variable. ³⁰ In sensitivity analyses, we examined whether the association between suPAR and outcomes differed according to age (<65 vs. >65 years), sex (male vs. female), race (White vs. non‐White) and treatment assignment (spironolactone vs. placebo). In the sample with available NP values (n = 273), the association of suPAR (dichotomized at 3 ng/mL) with outcomes was assessed for differences by levels of NP tertiles.

Lastly, we computed risk discrimination metrics (C‐statistics, integrated discrimination improvement [IDI] and net reclassification index [NRI]), using the 'survC1' and 'survIDINRI' packages in R, to examine the incremental prediction ability of adding suPAR and NP, separately and together, for the primary outcome at 1 year. Statistical significance was determined by a nonparametric permutation test with 300 resamples.

Statistical analyses were performed using R version 4.3.2. A two‐tailed P‐value <0.05 was used to determine statistical significance.

Results

SuPAR and patient characteristics

Baseline characteristics of 406 patients included in this study are shown in Table 1 , stratified by suPAR tertiles: first tertile (≤2.94 ng/mL), second tertile (>2.94 and ≤3.91 ng/mL) and third tertile (>3.91 ng/mL). The median suPAR level in the cohort at baseline was 3.37 (IQR 2.68–4.46) ng/mL. Patients were 46.6% female (n = 189/406), 91.6% White (n = 372/406) with a mean age of 69.5 years (SD 9.5) (Table  1 ). Approximately 10% were current smokers, 33.0% had diabetes mellitus and 39.7% had CKD Stage III or higher (eGFR <60). Age, BMI, BNP, NT‐proBNP and the proportion of patients with diabetes mellitus, chronic obstructive pulmonary disease (COPD) and peripheral arterial disease increased with higher suPAR tertiles. The correlation between suPAR and NT‐proBNP was weak (r = 0.04, P = 0.6, n = 174) while the correlation between suPAR and BNP was moderate (r = 0.27, P = 0.006, n = 99). In multivariable analysis, suPAR levels were independently associated with age, sex, race, eGFR, smoking, BMI and creatinine levels (Table  2 ).

Table 1.

Baseline characteristics of TOPCAT study sample by tertiles of suPAR.

Characteristics	All	Tertile 1	Tertile 2	Tertile 3	P‐value
	(N = 406)	(N = 136)	(N = 135)	(N = 135)
SuPAR (ng/mL) (range)	(1.19–60.97)	(1.19–2.94)	(2.95–3.91)	(3.93–60.97)
Demographics
Age, years (mean ± SD)	69.5 ± 9.5	67.1 ± 8.2	69.9 ± 9.6	71.5 ± 10.1	0.001
Female, n (%)	189 (46.6)	60 (44.1)	60 (44.4)	69 (51.1)	0.4
Race, n (%)
White (non‐Hispanic)	372 (91.6)	126 (92.6)	126 (93.3)	120 (88.9)	0.7
Black (non‐Hispanic)	28 (6.9)	8 (5.9)	7 (5.2)	13 (9.6)
Other	6 (1.5)	2 (1.5)	2 (1.5)	2 (1.5)
Body mass index (mean ± SD)	32.8 ± 6.7	31.5 ± 5.2	32.9 ± 6.6	33.9 ± 8.0	0.013
Comorbidities, n (%)
Current smoker	39 (9.6)	10 (7.4)	15 (11.1)	14. (10.4)	0.6
Hypertension	383 (94.3)	128 (94.1)	125 (92.6)	130 (96.3)	0.4
Diabetes mellitus	134 (33.0)	25 (18.4)	47 (34.8)	62 (45.9)	<0.001
Asthma	38 (9.4)	9 (6.6)	13 (9.6)	16 (11.9)	0.3
Chronic obstructive pulmonary disease	49 (12.1)	8 (5.9)	20 (14.8)	21 (15.6)	0.025
Previous myocardial infarction	127 (31.3)	47 (34.6)	41 (30.4)	39 (28.9)	0.6
Atrial fibrillation	174 (42.9)	50 (36.8)	63 (46.7)	61 (45.2)	0.2
Peripheral arterial disease	42 (10.3)	7 (5.1)	12 (8.9)	23 (17.0)	0.005
Dyslipidaemia	295 (72.7)	96 (70.6)	102 (75.6)	97 (71.9)	0.6
Thyroid disease	65 (16.0)	17 (12.5)	21 (15.6)	27 (20.0)	0.2
History of cardiovascular disease	150 (36.9)	55 (40.4)	51 (37.8)	44 (32.6)	0.4
History of stroke	32 (7.9)	7 (5.1)	10 (7.4)	15 (11.1)	0.2
Laboratory values upon admission (mean ± SD)
Estimated glomerular filtration rate	66.5 ± 19.5	73.7 ± 18.6	65.6 ± 16.9	60.2 ± 20.6	<0.001
Creatinine (mg/dL)	1.11 ± 0.30	1.01 ± 0.21	1.11 ± 0.26	1.22 ± 0.37	<0.001
Aspartate aminotransferase (U/L)	25.7 (12.0)	24.6 (10.1)	26.6 (12.4)	26.0 (13.4)	0.4
Alanine aminotransferase (U/L)	27.3 (14.9)	27.3 (14.5)	27.9 (14.5)	26.6 (15.9)	0.8
Ejection fraction full sample	57.9 ± 7.6	57.5 ± 7.5	58.9 ± 7.25	57.4 ± 7.9	0.2
BNP (pg/mL), median	234.0	174.0	191.0	331.0	0.02
(IQR)	(150.5–418.5)	(145.2–284.0)	(135.0–323.0)	(170.5–568.0)
NT‐proBNP (pg/mL), median	713.5	560.0	611.0	1017.0	0.001
(IQR)	(422–1460)	(390–1094)	(392–1374)	(610–2237)
Natriuretic peptides, n (%)					0.1
Tertile 1	91 (22.4)	30 (22.1)	40 (29.6)	21 (15.6)
Tertile 2	91 (22.4)	31 (22.8)	28 (20.7)	32 (23.7)
Tertile 3	91 (22.4)	28 (20.6)	23 (17.0)	40 (29.6)
Missing (n)	133 (32.8)	47 (34.6)	44 (32.6)	42 (31.1)
Outcomes, n (%)
Mortality	64 (15.7)	12 (8.8)	14 (10.4)	38 (28.1)	<0.001
Cardiovascular disease (CVD) mortality	43 (10.6)	9 (6.6)	9 (6.7)	25 (18.5)	<0.001
Myocardial infarction	18 (4.4)	3 (2.2)	7 (5.2)	8 (5.9)	0.3
Stroke	10 (2.5)	3 (2.2)	5 (3.7)	2 (1.5)	0.5
Any hospitalization	193 (47.5)	45 (33.1)	66 (48.9)	82 (60.7)	<0.001
Heart failure hospitalization	52 (12.8)	6 (4.4)	15 (11.1)	31 (23.0)	<0.001
Aborted cardiac arrest	1 (0.25)	1 (0.7)	0 (0.0)	0 (0.0)	‐‐
Primary outcomea	81 (19.9)	13 (9.6)	21 (15.6)	47 (34.8)	<0.001

Note: Values are mean (standard deviation [SD], median [interquartile range, IQR], or n (%) as noted. Significance is highlighted in bold.

^aComposite of death from a cardiovascular cause, aborted cardiac arrest or hospitalization for heart failure management.

Table 2.

Multivariable linear regression analysis of association between covariates and (log‐base 2) suPAR.

	Model 1	Model 2	Model 3
	Beta (95% CI)	Beta (95% CI)	Beta (95% CI)
Age	0.007 (0.001, 0.013)	0.015 (0.009, 0.021)	0.014 (0.007, 0.022)
Sex (females vs. males)	0.028 (−0.082, 0.139)	0.451 (0.285, 0.616)	0.558 (0.352, 0.764)
Race
Black (vs. White)	0.185 (−0.033, 0.403)	−0.293 (−0.526, −0.060)	−0.315 (−0.595, −0.035)
Other (vs. White)	−0.015 (−0.464, 0.433)	−0.027 (−0.386, 0.440)	−0.052 (−0.506, 0.402)
BMI	‐	0.011 (0.003, 0.020)	0.015 (0.004, 0.027)
Current smoker (yes vs. no)	‐	0.212 (0.033, 0.390)	0.128 (−0.096, 0.353)
Hypertension	‐	0.092 (−0.127, 0.310)	−0.036 (−0.298, 0.227)
History of cardiovascular disease	‐	−0.072 (−0.175, 0.032)	−0.084 (−0.216, 0.048)
Diabetes mellitus	‐	0.093 (−0.029, 0.215)	0.064 (−0.085, 0.213)
Chronic obstructive pulmonary disease	‐	0.161 (0.005, 0.318)	0.110 (−0.083, 0.302)
Peripheral artery disease	‐	0.159 (−0.009, 0.327)	0.199 (−0.003, 0.400)
eGFR	‐	0.020 (0.013, 0.026)	0.024 (0.016, 0.033)
Creatinine	‐	1.483 (1.037, 1.929)	1.802 (1.213, 2.391)
Natriuretic peptides (tertiles)	‐	‐	0.042 (−0.035, 0.118)

Note: Bolded effect estimates are significant at P < 0.05.

Impact of spironolactone on suPAR levels

The median suPAR level at baseline (n = 192/406) and 1‐year follow‐up (n = 155/331) was 3.46 and 3.70 ng/mL in the spironolactone arm (6.9% change, P = 0.22) and 3.30 ng/mL (baseline, n = 214/406) and 3.27 ng/mL (1‐year, n = 176/331) in the placebo arm (0.9% change, P = 0.79). There was no significant difference in median change of suPAR levels from baseline to 1‐year follow‐up comparing placebo to spironolactone (P = 0.60, Figure 1 ). Of note, there were no significant differences in baseline clinical characteristics between the spironolactone arm and placebo arm (Table  S1 ).

Figure 1.

Median change in suPAR levels by treatment arm (placebo versus spironolactone). The boxplot displays the median (IQR) change in suPAR (ng/mL) for placebo (blue) and spironolactone (orange). The P‐value indicates the statistical difference between groups using the Wilcoxon rank sum test.

SuPAR and outcomes

A total of 81 (19.9%) participants experienced the primary composite outcome (34.8% in the highest suPAR tertile [>3.93 ng/mL], compared to 9.6% in the lowest suPAR tertile [≤2.94 ng/mL], P < 0.001) (Table  1 ). The cumulative incidence of the primary outcome increased in a stepwise fashion according to suPAR tertile (log‐rank P < 0.001) (Figure  2 ). In multivariable analysis, the highest suPAR tertile was associated with more than a twofold higher risk of primary outcome (adjusted HR [aHR] 2.38, 95% CI [1.19–4.76]) compared to the lowest tertile, after adjusting for demographics (age, sex and race) and clinical risk factors (eGFR, hypertension, diabetes, history of smoking, history of CVD and BMI). When examined as a continuous variable, suPAR was associated with a twofold increase in the risk of the primary outcome (aHR 1.94, 95% CI [1.39–3.21]) per 100% increase in levels (Table  3 ).

Figure 2.

Kaplan–Meier plots illustrating the cumulative incidence of the primary outcome based on baseline tertiles of suPAR across the 5‐year period. The log‐rank test was used to test for significance between tertiles, with significance indicated by a P‐value of <0.05. suPAR, soluble urokinase plasminogen activator receptor. Primary outcome, composite of death from a cardiovascular cause, aborted cardiac arrest or heart failure hospitalization.

Table 3.

Association of suPAR with adverse HFpEF outcomes over time.

	Primary outcome, n = 81	All‐cause mortality, n = 64	CVD‐related mortality, n = 43	Any hospitalization, n = 193	HF hospitalization, n = 52
	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)
suPAR continuous (log‐base 2)
Model 0: unadjusted	1.82 (1.43, 2.31)	1.80 (1.36, 2.38)	1.71 (1.20, 2.45)	1.43 (1.18, 1.74)	1.90 (1.43, 2.51)
Model 1: + demographics	1.79 (1.37, 2.34)	1.84 (1.36, 2.49)	1.72 (1.19, 2.50)	1.37 (1.10, 1.69)	1.90 (1.35, 2.66)
Model 2: + clinical risk factors	1.94 (1.33, 2.83)	2.02 (1.35, 3.03)	1.78 (1.11, 2.87)	1.28 (1.01, 1.62)	2.13 (1.32, 3.46)
Model 3: + natriuretic peptides	2.05 (1.34, 3.14)	2.74 (1.72, 4.37)	2.57 (1.50, 4.40)	1.21 (0.91, 1.62)	1.98 (1.09, 3.59)
suPAR tertiles (ranges, ng/mL)
Model 0: unadjusted
Tertile 1 (1.19–2.94)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Tertile 2 (2.95–3.91)	1.61 (0.81, 3.22)	1.09 (0.50, 2.36)	0.94 (0.37, 2.37)	1.56 (1.07, 2.28)	2.50 (0.97, 6.44)
Tertile 3 (3.93–60.97)	4.13 (2.23, 7.64)	3.23 (1.69, 6.19)	2.85 (1.33, 6.11)	2.21 (1.53, 3.18)	5.84 (2.44, 14.01)
Model 1:
Tertile 1 (1.19–2.94)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Tertile 2 (2.95–3.91)	1.59 (0.79, 3.20)	0.97 (0.45, 2.12)	0.90 (0.35, 2.28)	1.50 (1.02, 2.20)	2.39 (0.92, 6.25)
Tertile 3 (3.93–60.97)	3.40 (1.80, 6.42)	2.89 (1.48, 5.65)	2.74 (1.25, 6.01)	1.92 (1.02, 2.79)	4.30 (1.75, 10.57)
Model 2:
Tertile 1 (1.19–2.94)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Tertile 2 (2.95–3.91)	1.29 (0.63, 2.66)	0.84 (0.38, 1.86)	0.78 (0.30, 2.05)	1.42 (0.95, 2.12)	1.88 (0.70, 5.08)
Tertile 3 (3.93–60.97)	2.38 (1.19, 4.76)	2.24 (1.07, 4.69)	2.23 (0.93, 5.36)	1.76 (1.18, 2.63)	2.82 (1.06, 7.48)
Model 3:
Tertile 1 (1.19–2.94)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Tertile 2 (2.95–3.91)	2.54 (0.93, 6.93)	2.08 (0.63, 6.88)	2.01 (0.48, 8.42)	1.62 (0.99, 2.65)	2.30 (0.66, 8.03)
Tertile 3 (3.93–60.97)	4.33 (1.64, 11.44)	6.00 (1.94, 18.59)	6.54 (1.74, 24.63)	1.81 (1.09, 3.01)	3.21 (0.93, 11.06)

Note: Model 1: Model 0 + demographics (age, sex, race). Model 2: Model 1 + clinical risk factors (eGFR, hypertension, diabetes, smoking, cardiovascular disease, BMI). Model 3: Sample size = 273 due to availability of natriuretic peptide (NP); primary outcome (n = 55), all‐cause mortality (n = 42), CVD‐related mortality (n = 27), any hospitalization (n = 134), HF hospitalization (n = 38).

Components of the primary outcome (CVD‐related mortality and hospitalization for HF management) along with all‐cause mortality and any hospitalization each increased in frequency with increasing suPAR tertiles when assessed individually (P < 0.001 for each) (Table  1 ). In multivariable analysis, levels of suPAR in the highest tertile remained significantly associated with increased risk for all‐cause mortality (aHR 2.24, 95% CI [1.07, 4.69]), any hospitalization (aHR 1.76 95% CI [1.18, 2.63]) and hospitalization for HF management (aHR 2.82, 95% CI [1.06, 7.48]) (Table  3 ) after adjustment for demographic and clinical risk factors.

We found no significant interaction between suPAR and treatment arm in the association with the primary outcome or individual outcomes (P interaction = 0.8, Table S2 ).

In the subset of participants with NP data, inclusion of NP in models with demographic and clinical risk factors revealed the highest tertile of suPAR was associated with a fourfold higher risk of the primary outcome compared to the lowest tertile (aHR 4.33, 95% CI [1.64, 11.44]) When assessed continuously, the association of each doubling of suPAR and the primary outcome persisted with the addition of NP to the model adjusted for demographic and clinical risk factors (aHR 2.05, 95% CI [1.34–3.14]).

In sensitivity analyses, the association between suPAR and outcomes did not differ by relevant subgroups, including by treatment arm (Figure  3 ). In the subset of participants with NP values available (n = 273), there was a significant interaction between suPAR and NP (interaction P = 0.008) in the fully adjusted model (Figure  4 ). Specifically, NP was more strongly associated with outcomes in patients with high suPAR (≥3 ng/mL) levels compared to those with suPAR <3 ng/mL.

Figure 3.

Sensitivity analysis among subgroups for the association of baseline suPAR (log‐base 2) and the primary outcome. Models are adjusted for age, race, sex, hypertension, diabetes, BMI, smoking history and history of cardiovascular disease The interaction P‐value was significant at a level of <0.05.

Figure 4.

Effect of natriuretic peptides on the primary outcome, stratified by high and low suPAR levels (cutoff 3 ng/mL). The model is adjusted for age, sex, race, eGFR, hypertension, diabetes, smoking, history of cardiovascular disease and BMI.

Risk discrimination

In a subset of 273 participants with available NP measurements, the baseline model (age, sex and race) yielded a C‐statistic of 0.616 (95% CI 0.478–0.755). Adding log2‐suPAR alone improved the C‐statistic by 0.140 (95% CI 0.026–0.254) to 0.756 (95% CI 0.642–0.871). Incorporation of NP alone led to a smaller increase in the C‐statistic (0.065, 95% CI [−0.044 to 0.173) from 0.616 to 0.681 (95% CI 0.550–0.812). When both log2‐suPAR and NP were added, the C‐statistic increased by 0.130 (95% CI −0.003 to 0.263) to 0.746 (95% CI 0.679–0.818). Further details, including changes in the integrated discrimination improvement and net reclassification improvement, are presented in Table 4 .

Table 4.

Risk discrimination metrics for the primary outcome in patients with HFpEF through 1‐year follow‐up. ^a

Models (NP subset: n = 273)	C‐statistic (95% CI)	Δ C‐statistic (95% CI)	IDI (95% CI)	NRI (95% CI)
Model 1	0.616 (0.478, 0.755)	‐‐	‐‐	‐‐
Model 1 + Log2‐suPAR	0.756 (0.642, 0.871)	0.140 (0.026, 0.254)	0.018 (−0.001, 0.055)	0.248 (0.000, 0.423)
Model 1 + NP	0.681 (0.550, 0.812)	0.065 (−0.044, 0.173)	0.010 (−0.003, 0.038)	0.130 (−0.069, 0.372)
Model 1 + Log2‐suPAR + NP	0.746 (0.679, 0.818)	0.130 (−0.003, 0.263)	0.032 (0.003, 0.093)	0.301 (0.071, 0.498)

Note: Model 1: age, sex, race. IDI and NRI statistical significance was determined via a nonparametric permutation test with 300 resamples, with a significance threshold of 0.05. NP subset includes the sample with available measures for NT‐proBNP or BNP.

^a Primary outcome = composite of death from a cardiovascular cause, aborted cardiac arrest or HF hospitalization at 1‐year follow‐up.

Discussion

In this ancillary analysis of TOPCAT's North American cohort, we examined whether suPAR—a regulator of innate immune activity—could enhance risk stratification and predict spironolactone response in HFpEF. Our main findings are that (1) elevated suPAR was independently associated with worse cardiovascular outcomes, (2) spironolactone did not reduce suPAR levels over 1 year and (3) the therapeutic effect of spironolactone was not modified by suPAR levels. These observations underscore that while suPAR captures a distinct inflammatory pathway in HFpEF, this pathway remains unaffected by mineralocorticoid receptor blockade.

The heterogeneous nature of HFpEF presents a significant challenge in developing effective treatments, particularly when targeting inflammation. While inflammation is recognized as a key pathogenic mechanism in HFpEF, clinical trials of anti‐inflammatory therapies have shown mixed results, likely because not all HFpEF patients have the same degree of inflammatory burden.

SuPAR as a stable inflammatory marker in HFpEF

Consistent with prior evidence in broader HF populations—including heart failure with reduced ejection fraction (HFrEF)—suPAR strongly associates with incident heart failure and adverse outcomes. ¹⁷ , ¹⁸ , ¹⁹ , ²⁶ , ²⁸ , ³¹ Our data extend these findings to HFpEF, demonstrating suPAR's prognostic value above NP. ²⁶ , ²⁸ The relatively weak correlation between suPAR and NPs suggests that inflammation (reflected by suPAR) and volume/pressure overload (captured by NPs) represent complementary drivers of HFpEF progression. Participants with elevated levels of both markers appear particularly vulnerable to adverse outcomes. Combining suPAR with NPs may offer a more nuanced approach to identifying high‐risk individuals. ²⁸ This concept aligns with the multifactorial nature of HFpEF, which involves myocardial hypertrophy and fibrosis, altered diastolic compliance and renal dysfunction. ³² , ³³ , ³⁴ , ³⁵

A notable property of suPAR is its stability over time and across acute‐phase inflammatory states, such as myocardial infarction or cardiac surgery. ²¹ , ³⁶ , ³⁷ In our cohort, suPAR levels remained stable in patients receiving spironolactone, suggesting that suPAR‐driven inflammatory processes may be upstream of—or parallel to—the mineralocorticoid‐receptor‐mediated pathways targeted by spironolactone. Moreover, suPAR's connections to CKD, atherosclerosis and diabetes ¹⁶ , ²⁵ , ²⁶ further underscore its role in reflecting systemic immune activation common in the comorbidity profile of HFpEF.

Spironolactone does not target the suPAR pathway

MRAs possess well‐documented anti‐inflammatory and antifibrotic properties. Their potential benefits in HFpEF—such as reducing congestion, lowering blood pressure and limiting fibrosis—were key justifications for the TOPCAT trial. ¹⁰ However, our findings indicate that spironolactone does not reduce suPAR levels, despite suPAR's involvement in inflammation and organ dysfunction. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²⁵ , ²⁶ This lack of effect was consistent across suPAR tertiles, suggesting that a high inflammatory risk, as determined by suPAR, does not necessarily equate to a greater therapeutic response to spironolactone. Therefore, any protective effects observed with spironolactone in HFpEF are likely mediated through mechanisms outside of the suPAR pathway.

Interaction with NPs

We noted an interesting interplay between suPAR and NPs (BNP or NT‐proBNP). Specifically, participants with both elevated suPAR and high levels of NPs appeared to face a particularly high risk of cardiovascular events. ²⁶ , ³⁸ This interplay may indicate a convergence of inflammatory and haemodynamic stress, where increased immune activation amplifies the deleterious consequences of elevated cardiac filling pressures. ³³ , ³⁴ , ³⁵ Notably, in our cohort, NPs became more prognostically significant in the presence of higher suPAR levels. These results warrant further exploration of combined biomarker strategies to identify high‐risk subpopulations in HFpEF, especially those for whom concurrent inflammation and elevated wall stress drive disease progression.

Clinical implications

Although suPAR did not identify patients who would derive greater or lesser benefit from spironolactone, it may serve as a valuable tool for risk stratification. suPAR, a stable marker of chronic inflammation, might help pinpoint a distinct inflammatory phenotype within the broader HFpEF population. In the TOPCAT trial, elevated suPAR levels were independently associated with worse cardiovascular outcomes, suggesting that suPAR could act as a significant biomarker for identifying HFpEF patients at increased inflammatory risk. This subgroup stratification is crucial, as these patients may benefit the most from targeted anti‐inflammatory therapies. In clinical practice, suPAR could complement NP assessment to refine prognostic estimates and guide closer monitoring or more aggressive management. ²⁶ The ability to identify specific inflammatory phenotypes within HFpEF could lead to more personalized therapeutic approaches, potentially enhancing the effectiveness of anti‐inflammatory interventions by directing them to the most suitable patient subgroups. Notably, suPAR has already demonstrated clinical utility as an entry criterion for patient selection in COVID‐19 trials with anakinra, representing the first example of suPAR being used clinically for therapeutic decision‐making. ³⁹ A recent meta‐analysis shows promising results for clinical trials targeting inflammation in HFpEF to reduce risk of death, HF hospitalization and worsening HF. ⁴⁰ Currently, the HERMES trial is underway to assess the efficacy and safety of interleukin‐6, another pro‐inflammatory cytokine, to reduce death, hospitalization and urgent visits for events related to HFpEF and mild HFrEF. ⁴¹ This meta‐analysis and ongoing clinical trial justify future investigations into whether suPAR can be directly lowered by novel anti‐inflammatory or immunomodulatory agents—and, if so, whether reducing suPAR levels can improve clinical outcomes in HFpEF.

Strengths and limitations

The strengths of this analysis include the utilization of well‐characterized, prospectively collected data from a randomized, double‐blind trial; repeated suPAR measurements at baseline and 1 year; and adjustments for comprehensive clinical and demographic variables. However, we acknowledge several limitations. First, our analysis was limited to a small subset of the North American TOPCAT cohort, which may restrict generalizability. ¹⁰ External validation is required in future studies. Second, missing NP data for some participants could diminish power to detect certain interactions; nevertheless, we still identified a significant interaction between suPAR and NPs. Finally, while our data reinforce suPAR's prognostic role, the observational nature of this ancillary study prevents definitive causal conclusions regarding suPAR's role in HFpEF pathophysiology.

Conclusions

In summary, we confirm suPAR as a stable and independent predictor of adverse outcomes in HFpEF. We also demonstrate that suPAR levels remain unaltered by spironolactone and do not dictate MRA response. These findings highlight the need for alternative strategies to target the persistent inflammatory processes reflected by suPAR. Future investigations into treatments aimed at lowering suPAR—and thereby potentially improving clinical outcomes—are warranted in HFpEF, a complex syndrome that continues to challenge existing therapies.

Funding

This work was supported by the National Center for Advancing Translational Sciences (NCATS) (UL1 TR002240‐05). SSH is supported by NHLBI R01HL153384. CGH is supported by a National Heart, Lung, and Blood Institute‐funded postdoctoral fellowship (T32‐HL007853). SNG is supported by a VA MERIT grant (1I01CX002560, NIH/NHLBI R01HL150392) and Taubman Medical Research Institute (Wolfe Scholarship). AAL is supported by VA Merit (1I01CX002684), the University of Michigan Research Scholars Program and the Michigan Biology of Cardiac Aging Center.

Conflict of interest

None declared.

Supporting information

Table S1. Comparison of baseline clinical characteristics between treatment arms.

Table S2. Interaction between suPAR and treatment arm in the association with each adverse HF outcome.

Hutten, C. G. , Tekumulla, A. , Ismail, A. , Vasbinder, A. , Farhat, T. , Kunkle, P. , Goonewardena, S. N. , Abdel‐Latif, A. , Pitt, B. , and Hayek, S. S. (2025) Soluble urokinase plasminogen activator receptor and outcomes in HFpEF: A TOPCAT ancillary study. ESC Heart Failure, 12: 4208–4218. 10.1002/ehf2.15423.