Catestatin as a predictor for cardiac death in heart failure with mildly reduced and preserved ejection fraction

Abstract

Aims

Endogenous catecholamine release‐inhibitory peptide catestatin has been associated with heart failure (HF). This subgroup analysis of our cohort of HF compared the different effects of catestatin as a predictor for cardiac outcomes in patients with HF with reduced (HFrEF), mildly reduced (HFmrEF) or preserved (HFpEF) ejection fraction.

Methods

Plasma catestatin was measured in the HF patient cohort of 228 cases with a whole spectrum of ejection fraction. The cardiac deaths were analysed according to prespecified subgroups.

Results

Over a median follow‐up of 52.5 months, the association between plasma catestatin and cardiac death was different in patients with HFrEF, HFmrEF or HFpEF [hazard ratio (HR) 1.53, 95% confidence interval (CI) 0.99–2.37 and HR 2.73, 95% CI 1.56–4.75, respectively; interaction P = 0.022]. Patients with HFmrEF/HFpEF were older and more likely to be female, with non‐ischaemic cardiomyopathy and atrial fibrillation but lower levels of plasma B‐type natriuretic peptide (BNP). Similar adverse cardiac events occurred in patients with HFmrEF/HFpEF as in HFrEF. Plasma catestatin was a better predictor for cardiovascular death in the HFmrEF/HFpEF patients [area under the receiver operating characteristic curve (AUC) = 0.72, 95% CI 0.45–0.74] than in the HFrEF patients (AUC = 0.59, 95% CI 0.587–0.849). The optimal cut point of plasma catestatin level of 0.86 ng/mL predicted a 2.80‐fold elevated risk for cardiac death in HFmrEF/HFpEF.

Conclusions

Elevated plasma catestatin might be a more sensitive predictor for cardiac outcome in patients with HFmrEF/HFpEF than in HFrEF.

Introduction

Heart failure (HF) with mildly reduced (HFmrEF) and preserved (HFpEF) ejection fraction (EF) has been defined as HF with left ventricular EF (LVEF) of more than 40%. Although these patients may exhibit better systolic function than patients with HF with reduced EF (HFrEF), they still experience considerable clinical events, including HF readmission and cardiovascular (CV) death.

The sympathetic adrenergic system is one of the most essential neurohumoral mechanisms involved in HF. Increased local myocardial release of catecholamine has been demonstrated in HFpEF, ¹ yet beta‐blockers show no clinical benefits as in HFrEF. ² Moreover, beta‐blockers have been reported to impair patients' exercise tolerance further in HFpEF. ³ Non‐selective beta‐blockers also have an unfavourable influence on glucose and lipid metabolism. ⁴ Therefore, sympathetic adrenergic regulation involved in HF with higher EF might differ from that in HFrEF. ⁵ , ⁶ Whether initiation of other anti‐sympathetic therapies might improve clinical outcomes in these patients is unknown.

Catestatin (CST) is a proteolytic fragment of chromogranin A (CgA). It has been shown to act as an endogenous catecholamine release‐inhibitory peptide. ⁷ Many researchers, including us, have revealed an association between the plasma level of CST and multiple cardiac diseases. ⁸ , ⁹ , ¹⁰ To be noted, we have observed elevated plasma CST in patients with acute coronary syndrome and HF exacerbation in contrast to other authors' results of low circulatory CST levels in asymptomatic coronary heart disease and HF population. ¹¹ , ¹² We proposed that the elevated CST could play a compensatory cardioprotective role at the activated adrenal sympathetic condition. ¹⁰

Recently, HF, especially HFpEF, has been suggested to be a metabolic disease. ¹³ Meanwhile, CST has been reported to be involved in glucose and lipid profile metabolic regulation. ¹⁴ Decreased plasma CST in obese adolescents was associated with ventricular hypertrophy and future HF risks. ¹⁵ Therefore, we suspect that CST might take part in the development of HFpEF at an early stage. The relative insufficiency of CST contributes to HFpEF, while the adaptive increase of CST might suggest the progression of cardiac dysfunction.

Our prior observation demonstrated that plasma CST acted as a predictor for elevated risks for cardiac and all‐cause death in patients with HF. Our study population permitted enrolment of patients with the whole spectrum of LVEF and thus provided an opportunity to examine the role of CST in this previously understudied group of HFmrEF/HFpEF patients. The objective of this analysis was to determine the efficacy of CST in predicting cardiac outcomes in patients with HFmrEF and HFpEF. We hypothesized that plasma CST in this population would be a better predictor than in HFrEF patients.

Methods

Study population and design

Consecutive 228 adult HF patients consistent with the modified Framingham criteria who had been hospitalized or presented to cardiac clinics were enrolled between July 2010 and November 2011 at the Department of Cardiology of Peking University First Hospital. For the diagnosis of HF, modified Framingham criteria were used. Two major criteria or one major and two minor criteria are needed. ¹⁶ Patients with concurrent acute coronary syndrome, pheochromocytoma, systemic infection, autoimmune disease, end‐stage renal disease, and confirmed or suspected malignant diseases were excluded. ¹⁷

At the time of enrolment, all patients had echocardiography performed by the same team of trained physicians. Blood samples were drawn at the time of enrolment, and plasma was frozen at −80°C for further analyses. Plasma CST was measured by using a human CST enzyme immunoassay kit (EK‐053‐27, Phoenix Pharmaceuticals, Inc.). The minimum detectable concentration of the kit is 0.07 ng/mL. Plasma B‐type natriuretic peptide (BNP) was tested by the chemiluminescent immunoassay method (UniCel DxI 800, Beckman Coulter, Inc., Fullerton, CA, USA).

The follow‐up of all included patients was conducted every 3 months with time‐to‐event analysis until the occurrence of fatal or nonfatal CV events. Twenty‐six patients were excluded because of incomplete baseline information or loss of follow‐up. The protocol was approved by the Ethics Committee of Peking University First Hospital and complied with the Declaration of Helsinki. Each patient provided written informed consent in accordance with established guidelines.

Demographic characteristics, underlying heart disease, comorbidities and medication usage were identified from the electronic medical record system. The primary outcome was CV death. Secondary outcomes include readmission due to HF, sudden cardiac death and all‐cause death. All the outcome events were retrieved and confirmed from the cohort follow‐up database. Death causes on the certificate were documented from the electoral health records system. CV death was defined as death attributed to a CV cause [International Classification of Disease 10 (ICD‐10) Codes I00–I99]. Clinical outcomes were adjudicated by an independent, blinded adjudication committee. The statistical analysis plan was prespecified and registered prior to analyses. The analysis of the HFmrEF/HFpEF and HFrEF subgroups was performed per the statistical analysis plan.

Statistical analysis

Data from all enrolled patients were included in this analysis. Primary and secondary outcomes from the cohort were compared between LVEF ≤ 40% and LVEF > 40% subgroups. Baseline characteristics in participants were summarized as means and SD, medians and inter‐quartile ranges or percentages and compared by chi‐squared test for categorical variables and Wilcoxon test and two‐sample Student's t test for non‐normal and normally distributed continuous variables, respectively. Specifically, the plasma CST levels were non‐normally distributed and analysed by non‐parametric tests. Time‐to‐event data for the primary outcome and secondary clinical outcomes were evaluated using the Kaplan–Meier estimator and Cox proportional hazard models, stratified by EF statuses. Interactions between EF status and other parameters were evaluated. In patients with different EF statuses, the prediction accuracy of plasma CST for the primary outcome was evaluated in the area under the receiver operating characteristic (ROC) curve (C statistic). A Hosmer–Lemeshow goodness‐of‐fit test was used to assess the model fit (Hosmer–Lemeshow statistic ≥0.05). Sensitivity and specificity values were reported. All analyses were performed in SPSS 25.0 software for Windows (SPSS Inc., Chicago, IL, USA).

Results

Of these 228 HF patients, complete baseline and follow‐up data were available for the primary endpoint in 202 patients.

Plasma CST as a predictor for CV death for prespecified subgroups

Over a median follow‐up of 52.5 months, the effect of plasma CST as a predictor for cardiac death was different in patients with HFrEF and HFmrEF/HFpEF [hazard ratio (HR) 1.53, 95% confidence interval (CI) 0.99–2.37 and HR 2.73, 95% CI 1.56–4.75, respectively; interaction P = 0.022]. Moreover, CST was a more robust predictor for cardiac death in patients with atrial fibrillation (AF), normal range pulmonary artery systolic pressure (PASP) and lower plasma BNP (Figure 1).

Figure 1.

Plasma catestatin for prediction of cardiac death in prespecified subgroups. AF, atrial fibrillation; BNP, B‐type natriuretic peptide; CI, confidence interval; HF, heart failure; LVEF, left ventricular ejection fraction; PASP, pulmonary artery systolic pressure.

Baseline characteristics by LVEF status

Compared with those with LVEF ≤ 40%, patients with LVEF > 40% were older, were more likely to be female and had a higher LVEF at baseline [56.5 (47.0–66.3) % vs. 31.0 (26.3–36.0) %, P < 0.001]. A higher percentage of patients with HFmrEF/HFpEF compared with those with HFrEF had a history of AF but were less likely to have coronary artery disease. Patients with HFmrEF/HFpEF, compared with those with HFrEF, were less likely to be treated with digoxin and angiotensin‐converting enzyme inhibitors or angiotensin receptor blockers. The baseline BNP levels were lower in HFmrEF and HFpEF than in HFrEF patients (Table 1).

Table 1.

Baseline characteristics and treatment by left ventricular ejection fraction (LVEF) statuses.

Characteristic	LVEF ≤ 40%	LVEF > 40%	P value
	N = 100	N = 102
Age (years), mean (SD)	64 (14)	70 (15)	0.003
Male sex, n (%)	82 (82.0)	60 (58.8)	<0.001
NYHA class, n (%)			0.574
I–II	42 (42.0)	47 (46.1)
III–IV	58 (58.0)	55 (53.9)
Ischaemic cardiomyopathy, n (%)	59 (59.0)	41 (40.2)	0.012
Atrial fibrillation, n (%)	23 (23.0)	53 (52.0)	<0.001
Hypertension, n (%)	9 (9.0)	16 (15.7)	0.557
Diabetes, n (%)	40 (40.0)	38 (37.3)	0.773
eGFR (mL/min), n (%)			0.462
<60	32 (32.0)	38 (37.3)
≥60	68 (68.0)	64 (62.7)
LVEF (%), median (IQR)	31.0 (26.3–36.0)	56.5 (47.0–66.3)	<0.001
PASP (mmHg), median (IQR)	40.5 (29.3–55.0)	36.0 (31.0–47.0)	0.121
Medications, n (%)
Diuretics	62 (62.0)	60 (58.8)	0.668
Beta‐blockers	81 (81.0)	72 (70.6)	0.101
ACEI/ARB	60 (60.0)	39 (38.2)	0.003
Spironolactone	39 (39.0)	34 (33.3)	0.464
Digoxin	43 (43.0)	20 (19.6)	<0.001
Plasma biomarker
Catestatin (ng/mL), median (IQR)	0.91 (0.58–1.42)	0.73 (0.59–1.12)	0.184
BNP (pg/mL), median (IQR)	968.7 (530.5–2238.5)	559.0 (261.5–1202.8)	<0.001

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitor; ARB, angiotensin receptor blocker; BNP, B‐type natriuretic peptide; IQR, inter‐quartile range; PASP, pulmonary artery systolic pressure; SD, standard deviation.

Outcomes by LVEF status

The rates of primary outcomes—CV death—were similar among those with LVEF ≤ 40% and those with LVEF over 40% [25 (25.0%) vs. 24 (23.5%), P = 0.870]. Rates of secondary endpoints, including readmission for HF events, sudden cardiac death and all‐cause death, were also similar between those with LVEF ≤ 40% and those with LVEF over 40% (Table 2).

Table 2.

Outcomes according to LVEF status.

Outcome, n (%)	LVEF ≤ 40%	LVEF > 40%	P value
	N = 100	N = 102
Primary endpoints
Cardiovascular death	25 (25.0)	24 (23.5)	0.870
Secondary endpoints a	63 (63.0)	62 (60.8)	0.773
Readmission due to HF	50 (50.0)	53 (52.0)	0.888
Sudden cardiac death	7 (7.0)	6 (5.9)	0.782
All‐cause death	32 (32.0)	27 (26.5)	0.440

Abbreviation: LVEF, left ventricular ejection fraction.

^a Combined endpoints.

CST as a predictor for CV death in patients with different LVEF

For the primary endpoint of the CV deaths, the Hosmer–Lemeshow test for goodness‐of‐fit of the models indicated good calibration for the plasma CST level as a continuous variable. Plasma CST level showed a moderate predictive ability with a C statistic of 0.59 (95% CI 0.45–0.74) in the HFrEF population for CV death. Meanwhile, the predictor efficacy of CST showed a better ability with a C statistic of 0.72 (95% CI 0.59–0.85) and higher sensitivity in the HFmrEF/HFpEF population (Table 3 and Figure 2).

Table 3.

Efficacy of plasma CST in the prediction of cardiovascular death in patients with LVEF ≤ 40% and LVEF > 40%.

	C statistic (95% CI)	Sensitivity	Specificity	χ 2 (P value) a
HFrEF	0.59 (0.45–0.74)	51.8%	71.8%	9.48 (0.303)
HFmrEF/HFpEF	0.72 (0.59–0.85)	70.8%	71.8%	7.66 (0.467)

Abbreviations: CI, confidence interval; CST, catestatin; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction.

^a Hosmer–Lemeshow χ ² statistic.

Figure 2.

The receiver operating characteristic curves for plasma catestatin in predicting cardiovascular death in patients with heart failure with reduced ejection fraction (HFrEF) and heart failure with mildly reduced ejection fraction (HFmrEF)/heart failure with preserved ejection fraction (HFpEF).

Optimal cut‐off values were determined by the Youden index extracted from ROC curve analyses of plasma CST in HFmrEF/HFpEF. The optimal cut‐off of 0.86 ng/mL of plasma CST with sensitivity of 70.8% and specificity of 71.8% suggested an elevated risk for CV death in this population.

HFmrEF/HFpEF patients were then grouped according to the CST cut‐off of 0.86 ng/mL. In multivariate Cox regression, patients with plasma CST higher than 0.86 ng/mL had significantly higher risks for CV death than those with lower plasma CST levels (adjusted HR = 2.80, 95% CI 1.08–7.26, P = 0.035) (Figure 3).

Figure 3.

The incidence of cardiovascular deaths assessed in a time‐to‐event analysis stratified by catestatin (CST) levels. Covariates adjusted were age, gender, New York Heart Association functional class, B‐type natriuretic peptide, pulmonary artery systolic pressure and estimated glomerular filtration rate. CI, confidence interval; HR, hazard ratio.

Discussion

In this secondary analysis of the HF cohort with plasma CST measurement, we found that CST was a better predictor for CV death in patients with HFmrEF and HFpEF than in those HFrEF patients with higher sensitivity and accuracy.

CST is one of the proteolytic fragments of CgA. CST binds to nicotinic acetylcholine receptors, non‐competitively inhibits catecholamine (CA) release from chromaffin cells and central and peripheral neurons, and acts as an important endogenous inhibitor for catecholamine release. ¹⁸ We, as other researchers, have noted the possible involvement of CST in the pathophysiological process of the CV system. Circulatory CST concentrations have been associated with hypertension, coronary heart disease, cardiac remodelling and HF, ⁸ , ⁹ , ¹⁰ in which sympathetic excitation has taken part in the disease progress. For HF, sympathetic adrenergic activation has been reported in both HFrEF and HFmrEF/HFpEF but has been diversely regulated. ⁵ , ⁶ Meanwhile, beta‐blockers are well‐established and valuable for managing HFrEF but not for HFmrEF and HFpEF. As a significant part of the endogenous sympathetic regulatory mechanism, the possible difference between CST plasma concentrations and the implications in HF patients with stratified EF spectrum warrants further assessment. However, there have been few explorations in this field.

Multiple metabolic disorders such as obesity, hypertension, diabetes and hyperlipidaemia have been associated with the pathophysiological process of HFpEF. As a pleiotropic peptide, CST has been found to be critical in maintaining metabolic homeostasis other than direct haemodynamic regulation. CST improves insulin sensitivity by attenuating endoplasmic reticulum stress, reducing inflammation and enhancing fatty acid oxidation. ¹⁴ CST‐KO mice are not only hypertensive but also have elevated levels of pro‐inflammatory cytokines and catecholamines and display left ventricular hypertrophy. ¹⁹ Serum CST concentrations are decreased in obese children and adolescents and normotensive offspring of patients with hypertension, suggesting that diminished CST might predispose to later development of metabolic syndrome. ¹⁵ , ²⁰ In addition, hyperglycosylated CgA has impaired conversion to CST, which is associated with adverse outcomes in acute HF. ²¹ , ²² We suspect that the relative CST deficiency might contribute to the development of HFpEF.

The CST level has been shown to be lower than healthy controls in stable coronary disease and asymptomatic HF patients in Chen et al. ¹¹ and Zhu et al. ¹² In contrast, our prior study showed elevated plasma CST in acute coronary syndrome and HF patients. ⁹ , ¹⁷ The distinct phenomena suggest the possible compensatory mobilization of CST in reaction to sympathoadrenal stress. Indeed, we have observed that the circulatory CST increased simultaneously with norepinephrine levels in HF patients, and the elevated CST concentrations independently predict cardiac and all‐cause deaths. ¹⁰ Recently, the results were supported by Borovac et al. in decompensatory HF patients, suggesting a similar prognostic indicator role of CST. ²³

The present study further conducted subgroup analysis in our HF cohort according to LVEF. Individuals with HFmrEF and HFpEF comprised more than half of the total patients in our HF cohort population, and the constituent ratio was in line with the overall epidemiological data of HF. Current guidelines have largely been limited in recommendations for this population because of the lack of evidence. These patients had baseline characteristics that were different from those of the HFrEF population, including older age, female sex, more concomitant AF and less frequent treatment with renin–angiotensin–aldosterone system (RAAS) inhibitors. Despite the higher EF, these patients are at substantial risk for poor quality of life and adverse outcomes. We noted that patients with HFmrEF and HFpEF enrolled in our study cohort had a similar symptom burden [as indicated by New York Heart Association (NYHA) classes] to those with HFrEF. There was no significant difference in the incidence of endpoint events between HFmrEF/HFpEF and HFrEF as well.

The elevated CST level appears to be a stronger predictor for CV death in HFmrEF/HFpEF than in HFrEF based on the ROC curve analysis. In contrast to the moderate predictive ability in the HFrEF population, the predictor efficacy of CST showed a better ability with a C statistic of 0.72 and much higher sensitivity in the HFmrEF/HFpEF population. The same tendency for CST as a better predictor existed in patients with AF and lower PASP and BNP concentrations than in their counterparts. In our study population, more concurrent AF and lower BNP levels in HFmrEF/HFpEF were noted. Meanwhile, AF is also a symptomatic stimulator, while elevated CST levels in AF patients have been reported. ²⁴ Despite this, we have suggested a more sensitive biomarker for CST in HFmrEF/HFpEF irrespective of AF history in our multivariate regression analysis. The results further strengthen the significance of CST in suggesting adverse outcomes in this population.

Additionally, the optimal cut‐off value of CST level identified from the ROC curve in the HFmrEF/HFpEF population was lower than the median value in the general HF population and suggested an ~3‐fold elevated risk for CV death. The data indicate the high sensitivity of the CST concentration in forecasting future adverse events in the HFmrEF/HFpEF population.

Although other cardiac biomarkers, such as BNP, have shown a correlation with cardiac dysfunction, CST as a predictor is still quite attractive. BNP acted as a more consistent biomarker across different LVEF spectra in our analysis with moderate accuracy in predicting cardiac death (Data S1). CST, in contrast, has shown the different association with cardiac death in the two subgroups with different LVEF ranges. The improved sensitivity of CST as a predictor in HFpEF might suggest that CST represents an important early detection of cardiac performance alteration. On the other hand, the CST C statistic in patients with HFrEF is relatively low, which might suggest a less pronounced contributor to the outcomes in this subgroup. In our prior work, a multi‐marker strategy approach has been developed and proven to be more informative than a single biomarker in HF prognosis. ¹⁷ Also, we have observed the BNP level returned to baseline after decongestion treatment as the chamber strain reduced while CST would not change soon. ¹⁰ Therefore, the elevated CST might suggest a continuous compensatory mechanism to balance the relative insufficiency. Considering that extra‐cardiac burden (such as obesity and diabetes mellitus) contributed more than cardiac burden to death and HF hospitalization in patients with higher LVEF and the possible involvement of CST in metabolic regulation, treatment targeting morbidities other than cardiac function per se should be highlighted. This suggestion is consistent with a recent update of European Society of Cardiology (ESC) guidelines stating the importance of treatment for CV and non‐CV comorbidities. Also, CST might possess more therapeutic potential than beta‐adrenergic receptor blockers in this regard, considering the unfavourable impact of the latter on glucose and lipid metabolism.

Limitation

The patients' inclusion and exclusion criteria for this cohort were identical to those of our prior serial studies. Specific disease conditions were excluded based on past or current history extracted from electronic health records. No specific diagnostic measurements were conducted to further confirm. However, considering the relative rareness of these disease conditions, such as pheochromocytoma and autoimmune disease, and the low probability of these conditions concurrently existing in chronic HF patients, we speculated that this exclusion method would have little impact on the data analysis and conclusion. Measurement of the biomarkers was performed on blood samples that were stored deep‐frozen until analysis, which cannot exclude the possibility of target protein degradation. As all the samples have been handled identically, this approach should not affect the validity of our results. In analysis of endogenous catecholamine regulation, the CST/epinephrine ratio might be a better indicator suggesting the relative balance. In our study, plasma epinephrine level was not determined. Therefore, we could not exclude the possible altered CST/epinephrine ratio that might temper the strength of the conclusions. However, the elevated CST level in the acute phase of HF was consistent with prior reports. Some medications, such as digoxin and RAAS inhibitors, were not applied evenly in subgroups. Meanwhile, the medications involved were not significant confounders in the multifactorial analysis. Finally, our study is an observational exploration conducted in an HF cohort from a single tertiary hospital. As expected, the study sample size is limited. And a relatively small number of events have been registered during follow‐up in this HF population, which might lead to unstable results, especially in the multivariate survival analysis. Results must be confirmed in further prospective large‐scale studies directly addressing CST as a CV predictor in HFmrEF/HFpEF.

Conclusions

Elevated plasma CST might be a more sensitive predictor for cardiac outcome in patients with HFmrEF/HFpEF than in HFrEF.

Conflict of interest statement

The authors have no conflicts of interest to declare.

Supporting information

Data S1. Supporting information.

Chu, S.‐Y. , Peng, F. , Wang, J. , Liu, L. , Zhao, J. , Han, X.‐N. , and Ding, W.‐H. (2025) Catestatin as a predictor for cardiac death in heart failure with mildly reduced and preserved ejection fraction. ESC Heart Failure, 12: 517–524. 10.1002/ehf2.15107.