Integrated miRNA-cytokine profiling reveals a molecular signature of cachexia in patients with chronic heart failure

Abstract

Introduction

Cardiac cachexia (CC) complicates chronic heart failure (CHF). Circulating microRNAs (miRNAs) and cytokines, such as growth differentiation factor-15 (GDF-15) and fibroblast growth factor-21 (FGF-21), may reflect the catabolic-anabolic imbalance and structural remodelling associated with this syndrome. To define the circulating signature of CHF by profiling muscle and remodelling-related miRNAs alongside GDF-15 and FGF-21, and to assess their associations with cardiac function and clinical phenotypes.

Methods

Twenty-five CHF patients (median age 78.0 years, 40% male) and 10 controls (median age 56.8 years, 60% male) were studied. CC was defined as unintentional, non-oedematous weight loss ≥6% within 6 months. Circulating miRNAs were quantified by RT-qPCR, GDF-15 and FGF-21 by ELISA. CHF patients were classified as having reduced (HFrEF) or preserved ejection fraction (HFpEF). Associations with CC and clinical CHF phenotypes were evaluated.

Results

Compared with controls, CHF patients showed lower levels of miR-15b, miR-29b, miR-133a, and miR-486 (all P < .01) and elevated GDF-15 levels (P < .001). FGF-21 was higher in patients with CC (P = .012), in HFrEF vs HFpEF (P = .039), and in NYHA III-IV vs I-II (P = .05). miR-29b levels were lower in CC (P = .019) and in HFrEF + CC vs HFpEF + nCC (P = .019). In CC, FGF-21 inversely correlated with miR-486 (ρ = −.69, P = .035), whereas GDF-15 inversely correlated with miR-21 (ρ = −.81, P = .007).

Conclusions

CHF is associated with miRNA suppression and elevated stress cytokine levels, amplified in CC. miR-29b and FGF-21 showed associations with CC, functional impairment, and structural remodelling, warranting validation in larger longitudinal cohorts.

Graphical Abstract

Graphical Abstract.

Schematic representation of the circulating molecular profile associated with chronic heart failure and cardiac cachexia, characterized by dysregulation of muscle- and remodelling-related miRNAs and increased stress-responsive cytokines. CHF patients showed reduced miR-15b, miR-29b, miR-133a, and miR-486 and increased GDF-15, while cardiac cachexia was further associated with lower miR-29b and higher FGF-21 levels.

Introduction

Chronic heart failure (CHF) represents a global health concern,¹ affecting over 64 million individuals worldwide and contributing significantly to morbidity, mortality, and healthcare burden. CHF causes metabolic dysfunction that extends from the heart to the skeletal muscle (SM).² Among the CHF most severe complications is cardiac cachexia (CC), a complex metabolic syndrome defined by non-oedematous weight loss, SM wasting, and systemic inflammation.³ CC affects ∼10%–40% of patients with CHF and is associated with a markedly poor prognosis, with mortality approaching 50% within 18 months.⁴ According to the 2016 European Society of Cardiology (ESC) heart failure (HF) guidelines, patients are classified according to left ventricular ejection fraction (LVEF) as HF with reduced ejection fraction (HFrEF, LVEF <40%), mildly reduced ejection fraction (HFmrEF; LVEF 41%–49%), and with preserved ejection fraction (HFpEF, LVEF ≥50%).⁵ Although occurring across the entire spectrum of CHF, CC is particularly frequent in advanced disease and in patients with HFrEF.⁴

The underlying pathophysiological mechanisms driving CC reflect a sustained disruption of the equilibrium between anabolic and catabolic pathways, driven by chronic systemic inflammation, autonomic imbalance, activation of the renin-angiotensin-aldosterone system, immune dysregulation, and profound metabolic alterations that can lead to myocardial fibrosis and accelerated cardiac muscle depletion, further exacerbated by systemic catabolic conditions. Despite its substantial clinical impact, CC remains frequently under recognized, with adverse effects on clinical outcomes.⁶ The need for early identification and tailored intervention of CC has prompted increasing interest in circulating mediators that mirror underlying molecular alterations associated with CHF and CC. Among these, Growth Differentiation Factor-15 (GDF-15) has emerged as a robust biomarker of cellular stress and HF severity.⁷ Experimental studies have shown that pharmacological inhibition of GDF-15 can attenuate cardiac dysfunction and prevent the onset of cachexia in HF models.⁸ Similarly, Fibroblast Growth Factor-21 (FGF-21) is elevated in several CC-related conditions, including CHF, systemic inflammation, and myopathy, where it predicts poorer outcomes.⁹ Although both cytokines are mechanistically linked to pathways relevant to CC, including appetite suppression, mitochondrial stress responses, and lipid mobilization,¹⁰ their specific roles in CHF patients with CC remain incompletely defined. In parallel, circulating microRNAs (miRNAs), short non-coding RNAs that fine-tune gene expression, have emerged as attractive biomarkers and potential mediators of SM wasting and cardiac remodelling in CHF and CC.¹¹ However, comparative studies jointly evaluating GDF-15, FGF-21, and miRNA signatures in CC are scarce, and their diagnostic or prognostic value, particularly in combination, remains to be clarified.

In this study, we profiled selected circulating miRNAs, along with GDF-15 and FGF-21, in a well-characterized CHF cohort to evaluate whether their combined patterns associate with clinical phenotypes, functional class, left-ventricular structural burden, and cachexia status. By integrating cytokine and miRNA profiling with clinical characterization, we aimed to identify molecular signatures that could refine diagnosis, enhance prognostic assessment, and deepen understanding of the mechanisms underpinning CC in CHF.

Methods

Study design and patients

This single-centre, observational study included 25 consecutive patients with a confirmed diagnosis of CHF who were referred to the Department of Translational and Precision Medicine, Internal Medicine and Clinical Nutrition Unit, Policlinico Umberto I, Sapienza University of Rome.

Patients were classified according to the NYHA functional classes as I-II and III-IV¹² and met the current ESC diagnostic criteria for CHF.

Inclusion criteria were: age ≥50 years; documented diagnosis of CHF for at least 3 months; presence of clinical signs and symptoms of chronic HF with either reduced LVEF (<40%, HFrEF) or preserved LVEF (≥50%, HFpEF); negative history of active or previous malignancy; and ability to provide written informed consent.

Exclusion criteria were: age <18 years, refusal to participate in the study, active malignancy, chronic inflammatory or autoimmune diseases unrelated to CHF, end-stage renal disease, chronic pulmonary disease, liver cirrhosis, cognitive impairment, dysphagia, recent (<3 months) major surgery, acute infection, congenital heart disease, acute decompensated heart failure, and any other condition known to cause systemic inflammation or malnutrition. We also excluded patients with HFmrEF due to the small overall prevalence (10%–25%) of this condition among CHF patients, and since the prevalence of CC is higher in patients with advanced disease and HFrEF.⁵

The control group (n = 10) consisted of sex-matched healthy volunteers without a history of cardiovascular disease, chronic illness, or inflammatory conditions.

The study was approved by the Local Ethics Committee (approval number 27/19), and all participants provided written informed consent in accordance with the Declaration of Helsinki.

Clinical assessment and sample collection

At enrolment, a comprehensive clinical, anthropometric, and biochemical evaluation was performed for all participants, including NYHA functional class, major comorbidities (hypertension, type 2 diabetes, dyslipidemia, atrial fibrillation).

Height (m) and weight (kg) were measured, and body mass index (BMI, kg/m²) was calculated.

Patients reported their usual body weight and any involuntary, non-oedematous weight loss. CC was diagnosed as unintentional weight loss ≥6% within 6 months according to criteria proposed for cardiac cachexia assessment in CHF.¹³ The absence of oedema/congestion was clinically assessed.

Fasting peripheral blood was collected and processed within 2 hours. EDTA tubes were used for plasma isolation and circulating miRNA analysis, whereas serum samples were used for GDF-15 and FGF-21 measurement. Plasma and serum samples were obtained by centrifugation at 3000 rpm for 15 minutes at 4°C, then aliquoted and stored at −80°C until further analysis.

Echocardiographic evaluation and left ventricular mass

All patients with CHF underwent standard transthoracic echocardiography using a commercially available Sonoscape X3 portable ultrasound system (Sonoscape Medical Corp, China) equipped with a phased-array 7P-B cardiac probe, operating at ∼2.0–9.0 MHz, performed by experienced sonographers blinded to biomarker data.

Left ventricular mass (LVM) was calculated according to the Devereux-modified American Society of Echocardiography (ASE) cube formula, which provides a highly accurate non-invasive estimate of LV mass. References values were defined as ≤115 g for men and ≤95 g for women.¹⁴

RNA isolation and real-time PCR

Total RNA was extracted from plasma samples using RNeasy Serum/Plasma Kit (Qiagen, Germantown, MD, USA) according to the manufacturer’s instructions. Plasma samples were visually inspected for haemolysis and spectrophotometrically assessed by measuring absorbance at 414 nm. Samples showing evidence of haemolysis were excluded from the analysis. RNA concentration and purity were assessed using a Multiskan Sky spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Reverse transcription was carried out from 10 ng of total RNA using the TaqMan Advanced miRNA cDNA Synthesis kit (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer’s instructions. Quantitative real-time PCR was performed with TaqMan Fast Advanced Master Mix (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA), using the Step-OnePlus Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA). The following TaqMan miRNA assays (A25576, Applied Biosystems) were used: hsa-miR-15b-3p, hsa-miR-21-5p, hsa-miR-29a-3p, hsa-miR-29b-3p, hsa-miR-133a-3p, and hsa-miR-486-5p. The expression of hsa-miR-16-5p served as the endogenous reference for normalization after having verified the stability of its expression. Data were processed using SDS 2.4 Software (Applied Biosystems), and miRNA relative expression levels were calculated by the 2^−ΔΔCt method. All reactions were performed in duplicate.

Enzyme-linked immunosorbent assay

Serum concentrations of GDF-15 (#ELH-GDF15-1) and FGF-21 (#ELH-FGF21-1) were measured using ELISA (RayBiotech, Norcross, GA, USA) according to manufacturers’ instructions.

The amount of cytokine present in the samples was detected at a wavelength (λ) of 450 nm, using the Multiskan Sky spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), by calculating the interpolation of the standard curve and the arithmetic average of the duplicates.

Standard curves were generated for each plate, and cytokine concentrations were expressed as pg/mL. The lower detection limits were 0.947 pg/mL for GDF-15 and 8 pg/mL for FGF-21. All measurements were performed in duplicate.

Statistical analyses

Given the exploratory nature of the present study, no sample size calculation was performed, and all consecutive eligible patients were referred to the Department of Translational and Precision Medicine, Internal Medicine and Clinical Nutrition Unit, Policlinico Umberto I, Sapienza University of Rome. Data are presented as the mean ± standard deviation (SD) or median with interquartile range (IQR) for continuous normally or non-normally distributed variables, as appropriate. Student’s t-test or one-way ANOVA were used for comparisons of normally distributed data, while Mann–Whitney U test or Kruskal–Wallis test were applied for non-normally distributed data. Categorical variables were compared using the chi-square (χ²) test or Fisher’s exact test when appropriate. Spearman’s correlation coefficient (ρ) was used to assess associations between circulating miRNAs and clinical/echocardiographic parameters, including LVEF, left atrial diameter, NYHA functional class, and comorbidities. Due to the limited sample size and subgroup numerosity, multivariable adjustment was not performed to avoid unstable estimates and model overfitting. Therefore, all analyses should be considered unadjusted and exploratory. A two-tailed P-value <.05 was considered statistically significant. No formal correction for multiple testing was applied, and all P-values should be interpreted as exploratory. Analyses were performed using IBM^® SPSS Statistics version 26.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 5.0 (San Diego, California, USA).

Results

Patients’ characteristics

The study included 25 patients with CHF (40% males), with a median age of 78.0 years, and 10 healthy controls (60% males), with a median age of 56.8 years. Demographic, anthropometric, clinical, and echocardiographic characteristics are summarized in Table 1.

Table 1.

Clinical and demographic characteristics of the study population

	Controls n = 10	HF patients n = 25	P-value
Male, n (%)	6 (60)	10 (40)	P = .27
Age (years), median (IQR)	56.8 (53.25; 62.25)	78.0 (74.2; 83.5)	P  <  .001
BMI (kg/m2), mean ± SD	25.2 ± 1.6	25.0 ± 3.9	P = .93
Weight loss in previous 6 months (%), mean ± SD	0	4.1 ± 5.7
Cardiac cachexia, n (%)	0	10 (40)
NYHA functional class:	0
I-II, n (%)		9 (36)
III-IV, n (%)		13 (52)
LVM (g), mean ± SD		260.6 ± 74.7
LVEF phenotype:
HFrEF, n (%)		12 (48)
HFpEF, n (%)		13 (52)
Aetiology:
Hypertensive heart disease, n (%)		15 (60)
Coronary artery disease, n (%)		9 (36)
Device therapy:
PMK/CRT implantation, n (%)		13 (52)
Blood parameters:
Hemoglobin, g/dL	13.3 ± 0.7	11.3 ± 1.6	P = .01
C-Reactive Protein, mg/dL	3.4 ± 5.7	30.7 ± 47.0	P = .41
Creatinine, mg/dL	0.7 ± 0.1	1.4 ± 1.4	P = .33
Albumin, g/dL	2.3 ± 0.9	2.8 ± 1.8	P = .50
Hypertension, n (%)	4 (40%)	18 (72%)	P = .17
Diabetes mellitus, n (%)	2 (20%)	11 (44%)	P = .35
Dyslipidemia, n (%)	4 (40%)	10 (40%)	P = 1.0
Atrial fibrillation, n (%)	0	7 (28%)	P = .08

P values < 0.05 are shown in italics.

Abbreviations: BMI, body mass index; CRT, Cardiac resynchronization therapy; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; IQR, interquartile range; LVM, left ventricular mass; NYHA, New York Heart Association; PMK, pacemaker; SD, standard deviation.

Mean unintentional weight loss in the previous 6 months among CHF patients was 4.1 ± 5.6%. CC was diagnosed in 10 patients (40%). Within our cohort, 60% of patients with CHF was primarily attributed to LV hypertrophy secondary to arterial hypertension (n = 15). Coronary artery disease represented the second most prevalent aetiology, accounting for 36.0% of cases (n = 9). Additionally, 52% of the CHF patients had previous undergone pacemaker (PMK)/Cardiac resynchronization therapy (CRT) implantation (n = 13).

CHF patients were distributed across NYHA classes I-II (36%) and III-IV (52%). Both HFrEF (48%) and HFpEF (52%) phenotypes were represented with a balanced distribution. Echocardiographic evaluation revealed a mean LVM of 281.5 g. In the CHF cohort, comorbidities were frequent, with hypertension (72%), type 2 diabetes mellitus (44%), dyslipidemia (40%), and atrial fibrillation (28%) among the most frequent.

No significant structural or functional cardiac abnormalities were detected in the control group.

Circulating miRNAs and cytokines as molecular signatures of heart failure

To investigate molecular alterations associated with CHF, we quantified plasma levels of circulating selected miRNAs and serum concentration of stress-responsive cytokines (GDF-15 and FGF-21) in patients with CHF (n = 25) and healthy controls (n = 10).

We first analyzed, by quantitative Real-Time PCR, plasma miRNA involved in body composition changes, SM homeostasis, cardiac remodelling, and systemic metabolic stress responses (Figure 1). We selected these muscle-specific miRNAs based on our recent work where we observed their association with levels of muscularity in a cohort of patients cancer.¹⁵ Our results showed a significant downregulation of miR-15b, miR-29b, miR-133a, and miR-486 in CHF patients compared with controls (P < .01) (Figure 1A, D–F). In contrast, miR-21 and miR-29a did not differ significantly between the two groups (Figure 1B, C). Absolute miRNA values for CHF patients and controls are reported in Table 2.

Figure 1.

Circulating miRNAs and cytokines as molecular signatures of heart failure. Differences in circulating miRNA levels between CHF patients and controls. Circulating miRNA levels were analyzed by quantitative Real-Time PCR from patients with heart failure (HF) (n = 25) and control group (n = 10). Data shows significant downregulation of miR-15b (P < .001) (A), −29b (P < .001) (D), −133a (P = .005) (E) and −486 (P < .001) (F) in CHF vs controls. No statistical significance was observed for miR-21 (B) and −29a (C) levels. Data were normalized to miR-16 from two biological replicates, with a calibrator used as the internal control. Data are expressed as fold change and are presented as mean ± SEM. Student’s t-test or Mann–Whitney test was used for normally distributed data or non-normally distributed data, respectively. ** P ≤ .01, *** P ≤ .001.

Table 2.

Circulating biomarkers levels in chronic heart failure patients and controls

	Controls n = 10	HF patients n = 25	P-value
miR-15b	4.3 (2.2; 5.1)	1.0 (0.7; 1.8)	P  <  .001 *
miR-21	1.9 ± 0.9	3.3 ± 2.1	P = .054
miR-29a	1.5 (1.3; 2.1)	1.1 (0.9; 1.9)	P = .076
miR-29b	2.0 ± 0.7	1.0 ± 0.7	P  <  .001 *
miR-133a	26.2 ± 10.2	14.6 ± 14.1	P = .005 *
miR-486	1.1 (0.9; 1.5)	0.3 (0.2; 0.4)	P  <  .001 *
GDF-15	395.0 (328.1; 485.5)	1106.5 (835.5; 1281.1)	P  <  .001 *
FGF-21	265.6 ± 239.2	527.4 ± 668.9	P = .120

P values < 0.05 are shown in italics.

* P < .05.

Data are reported as mean ± SD or median (IQR), according to data distribution.

Abbreviations: FGF-21, fibroblast growth factor 21; GDF-15, growth differentiation factor 15; IQR, interquartile range; miR-, microRNA; SD, standard deviation.

Next, to evaluate systemic stress signalling, we measured serum levels of GDF-15 and FGF-21 by ELISA (Figure 2).

Figure 2.

Circulating miRNAs and cytokines as molecular signatures of heart failure. Differences in serum levels of GDF-15 and FGF-21 in CHF patients and controls. Serum levels of GDF-15 by ELISA assay in patients with heart failure (HF) (n = 25) and the control group (n = 10). (A) Data showed significant upregulation of GDF-15 in CHF vs controls (P < .001). (B) FGF-21 levels did not differ significantly between groups. Correlations between GDF-15 and selected miRNAs in CHF patients. (C, D) In CHF patients, GDF-15 levels were positively associated with miR-29b (ρ = 0.42, P = .037) and miR-486 (ρ = 0.42, P = .039). Data are expressed as fold change and are presented as mean ± SEM. Student's t-test or Mann–Whitney test was used for normally distributed data or non-normally distributed data, respectively. *** P ≤ .001.

Our data showed that GDF-15 levels were significantly elevated in CHF patients compared with controls (P < .001) (Figure 2A). Spearman’s correlation analysis revealed that in CHF patients, GDF-15 positively correlated with miR-29b (ρ = 0.42, P = .037) and miR-486 (ρ = 0.42, P = .039) (Figure 2C, D). Although median FGF-21 values were higher in CHF patients, differences between groups did not reach statistical significance (Figure 2B). No other significant correlations were observed in our cohort (data not shown). Absolute cytokines values for CHF patients and controls are reported in Table 2.

None of samples used showed signs of haemolysis.

Molecular profiles in CHF patients

We next analyzed circulating miRNA in our cohort of CHF patients to determine whether specific molecular alterations characterize the clinical phenotype of CC.³

miR-29b was significantly downregulated in CHF patients with CC compared with those without CC (P = .019) and with respect to controls (P = .001) (Figure 3D). In addition, both cachectic and non-cachectic CHF patients showed significantly lower circulating levels of miR-15b, miR-133a, and miR-486 compared with controls (all P < .05) (Figure 3A, E, F). No significant differences were observed for miR-21 and miR-29a (Figure 3B, C).

Figure 3.

Molecular profiles in CHF patients. Differences in circulating miRNA levels among heart failure patients with and without cardiac cachexia. Circulating miRNA levels were analyzed in CHF patients with cardiac cachexia (CC) (n = 10) and without cardiac cachexia (nCC) (n = 15), and a control group (n = 10). (D) Data show significant downregulation of miR-29b (P = .019) in HF with CC vs nCC and vs controls (P = .001). Significant modulations were observed between CHF patients with CC or nCC vs controls for miR-15b (A), −133a (E), and −486 (F) (all P < .05). No other statistical significances were observed for miR-21 (B), and −29a (C). Data are expressed as fold change and are presented as mean ± SEM. One-way ANOVA test or Kruskal–Wallis test was used for normally distributed data or non-normally distributed data, respectively. * P < .05, ** P ≤ .01, *** P ≤ .001.

We next examined whether GDF-15 and FGF-21 could differentiate cachectic and non-cachectic patients within the CHF population (Figure 4). Levels of FGF-21 were significantly higher in cachectic than in non-cachectic CHF patients (P = .012) and compared with controls (P = .010) (Figure 4B). Interestingly, FGF-21 negatively correlated with miR-486 (ρ = −0.69, P = .035) in CHF patients with CC (Figure 4D). GDF-15 levels were significantly higher in both cachectic and non-cachectic CHF patients compared with controls (all P < .001) (Figure 4A). Additionally, in CHF patients with CC, GDF-15 negatively correlated with miR-21 (ρ = −0.81, P = .007) (Figure 4C). No other significant correlations were observed (data not shown).

Figure 4.

Molecular profiles in CHF patients. Differences in serum levels of GDF-15 and FGF-21 among heart failure patients with and without cardiac cachexia. Circulating levels of GDF-15 and FGF-21 were analyzed in CHF patients with cardiac cachexia (CC) (n = 10) and without cardiac cachexia (nCC) (n = 15), and a control group (n = 10). (A) Significant difference in GDF-15 levels was observed between patients with CC or nCC vs controls (all P < .001). (B) Levels of FGF-21 were significantly higher in CHF with CC compared with nCC (P = .012) and vs controls (P = .010) (C). Spearman’s correlation analysis revealed that in CHF patients with CC, GDF-15 negatively correlated with miR-21 (ρ = −0.81, P = .007). (D) In the same group, FGF-21 negatively correlated with miR-486 (ρ = −0.69, P = .035). Data are expressed as fold change and are presented as mean ± SEM. One-way ANOVA test or Kruskal–Wallis test was used for normally distributed data or non-normally distributed data, respectively. * P < .05, *** P ≤ .001.

Molecular profiles in cachectic severe CHF patients

We further explored whether molecular alterations were associated with cardiac functional phenotypes by stratifying CHF patients according to left ventricular ejection fraction (LVEF). As described in the Materials and Methods section, patients with HFmrEF were not included in the present study.

FGF-21 concentrations were significantly higher in patients with reduced LVEF (HFrEF) compared with those with preserved LVEF (HFpEF) (P = .039) and vs the control group (P = .050) (Supplementary Figure S1A). Moreover, within the HFrEF group, FGF-21 positively correlated with miR-133a (ρ = 0.68, P = .014) (Supplementary Figure S1B).

We then assessed whether these molecular changes were further modulated at the by incorporating cachexia status into the stratification. This allowed us to exploratorily compare phenotypes combining reduced LVEF and cachexia (HFrEF + CC) with those showing preserved cardiac function and no cachexia (HFpEF + nCC), as a descriptive analysis of clinically distinct CHF subgroups. Our results showed that miR-29b was significantly reduced in HFrEF + CC compared with HFpEF + nCC (P = .019) and compared with controls (P = .002) (Figure 5A). While no significant differences between HFrEF + CC and HFpEF + nCC were observed for miR-15b, miR-21, miR-29a, or miR-133a, miR-486 was significantly reduced in both HFrEF + CC (P = .002) and HFpEF + nCC (P = .005) compared with controls (Supplementary Figure S2).

Figure 5.

Molecular profiles in cachectic patients with severe CHF. Circulating miRNA levels and GDF-15 correlations according to ejection fraction and cachexia status in CHF patients. Circulating miRNA levels were analyzed in CHF patients with preserved ejection fraction without cachexia (HFpEF + nCC) (n = 7) and with reduced ejection fraction with cachexia (HFrEF + CC) (n = 5) and a control group (n = 10). (A) Data shows significant downregulation of miR-29b (P = .019) in HFrEF + CC vs HFpEF + nCC and vs controls (P = .002). (B) In HFpEF + nCC, GDF-15 showed a strong positive correlation with miR-29b (r = 0.82, P = .034). Data are expressed as fold change and are presented as mean ± SEM. One-way ANOVA test or Kruskal–Wallis test was used for normally distributed data or non-normally distributed data, respectively. * P < .05, ** P ≤ .01.

GDF-15 showed a strong positive correlation with miR-29b in the group with HFpEF + nCC (ρ = 0.82, P = .034) (Figure 5B).

To gain a more comprehensive view of the clinical context in which these molecular changes occur, we next explored their associations with distinct clinical phenotypes and quantitative structural measures of cardiac remodelling and functional status.

GDF-15 levels were strongly and positively correlated with miR-133a (ρ = 0.93, P = .007) in CHF patients with atrial fibrillation, whereas in those in sinus rhythm, GDF-15 positively correlated with miR-29b (ρ = 0.50, P = .033) (Supplementary Figure S3).

When we considered functional status, FGF-21 concentrations were higher in NYHA class III-IV than in class I-II (P = .050) (Figure 6A). Within NYHA III-IV, FGF-21 concentrations were positively correlated with miR-15b (ρ = 0.68, P = .014) (Figure 6B).

Figure 6.

Molecular profiles in cachectic patients with severe CHF. FGF-21 concentrations and their associations with NYHA functional class in patients with CHF. (A) Data show higher FGF-21 levels in NYHA class III-IV (n = 13) compared with class I-II (n = 9) (P = .050) and the control group (n = 10) (P = .050). (B) In NYHA class III-IV, FGF-21 positively correlated with miR-15b (r = 0.68, P = .014). Data are expressed as fold change and are presented as mean ± SEM. One-way ANOVA test or Kruskal–Wallis test was used for normally distributed data or non-normally distributed data, respectively. * P < .05.

Discussion

CHF is frequently accompanied by progressive alterations in body composition that may culminate in cardiac cachexia (CC), a severe metabolic condition associated with poor clinical outcomes and increased mortality.¹⁶ CC arises from chronic inflammation, neurohormonal activation, and metabolic stress, leading to a persistent imbalance between anabolic and catabolic pathways and ultimately driving SM and adipose tissue wasting.¹⁷ However, early detection remains challenging, and reliable biomarkers for the early stages of CC are still lacking.^4,6

In this context, circulating miRNAs and stress-responsive cytokines may provide mechanistic insight and potential early biomarkers by integrating signals from cardiac remodelling, SM catabolism, and systemic metabolic stress. Consistently, our results identify a molecular signature in CHF characterized by coordinated alterations in selected miRNAs and cytokines associated with impaired SM homeostasis, structural remodelling, and metabolic stress responses.

Among the analyzed miRNAs, miR-15b was significantly reduced in CHF patients compared with controls, suggesting an association with disease-related molecular remodelling. This reduction was observed in both CC and nCC patients, indicating that lower miR-15b expression may be related to CHF itself rather than to cachexia status, at least in our cohort. This observation is consistent with previous reports describing dynamic regulation of miR-15b during cardiac stress and remodelling.¹⁸ Functionally, miR-15b has been implicated in regulating cardiomyocyte stress responses¹⁹ and SM cell homeostasis,²⁰ including processes related to proliferation, apoptosis, and mitochondrial regulation.²¹ In this context, the reduced circulating levels observed in our cohort may reflect impaired adaptive regulation of cardiac and SM stress-response pathways during CHF progression.

In contrast, miR-21, a well-established player in cardiac pathophysiology,²² did not differ significantly between CHF patients and controls, nor between cachectic and non-cachectic subgroups, and showed no association with the CHF phenotypes examined in this study. Although miR-21 is widely described as a pro-fibrotic²³ and anti-apoptotic²⁴ regulator, its circulating levels remained stable in our cohort and did not reflect these processes. This discrepancy likely reflects the complex and often contradictory functions of miR-21,²⁵ where tissue-specific expression and disease-stage variability²⁶ may limit the ability of circulating levels to capture local pro-fibrotic or remodelling processes in CHF.

The miR-29 family is recognized as a central antifibrotic regulator, directly repressing the expression of collagens and extracellular matrix proteins, and their downregulation in heart failure has been associated with pathological fibrosis and adverse remodelling.²⁷ In our cohort, miR-29b was the most consistently downregulated isoform, with a further reduction in patients with CC and in the subgroup with CC and HFrEF, suggesting an association with more advanced disease states. These results are consistent with experimental and clinical evidence,²⁸ suggesting that reduced miR-29b may reflect the convergence of metabolic and fibrotic signalling that drives catabolic remodelling. Downregulation of miR-29b has been shown to correlate with lower LVEF and higher inflammatory markers in acute myocardial infarction,²⁹ while its therapeutic overexpression attenuates fibrosis and improves ventricular function.³⁰ In this context, the reduced circulating levels observed in our cohort may be consistent with ongoing fibrotic remodelling and the systemic catabolic environment characterizing CHF and cachexia. In contrast, miR-29a showed no significant changes across any groups or phenotypes in our cohort. This stability may suggest isoform-specific regulation, with miR-29a potentially less involved in the systemic alterations captured by circulating biomarkers in CHF.

As for miR-133a, a myomiR that regulates myoblast differentiation, cardiomyocyte proliferation, and structural remodelling,³¹ our findings demonstrated clear downregulation in CHF patients compared with controls, consistent with previous studies.¹⁸ Early mechanistic studies demonstrated that miR-133a is a negative regulator of cardiac hypertrophy by repressing key sarcomeric and signalling genes,³¹ and synthetic or overexpressed miR-133a has been shown to prevent heart failure, attenuate fibrosis, and improve cardiac function.³² We did not observe any additional modulation between the cachectic and non-cachectic subgroups, suggesting that, in our dataset, miR-133a downregulation may primarily reflect HF-related remodelling rather than cachexia-specific signalling.

Similarly, miR-486, a muscle-enriched miRNA involved in maintaining both cardiac and SM homeostasis, was markedly reduced in CHF patients compared with controls, aligning with previous clinical reports of reduced miR-486 in HF and SM wasting disorders.³³ While previous data have reported miR-486’s involvement in SM atrophy and cardiac remodelling, our findings did not show cachexia-specific changes in the circulation. This pattern may suggest that circulating miR-486 in our cohort primarily reflects HF-related alterations rather than cachexia-specific signalling detectable in plasma.

Our analysis of the stress-responsive cytokines GDF-15 and FGF-21 revealed distinct patterns and functional interplay, reflecting their roles in metabolic stress, cardiac function, and remodelling.³⁴

Consistent with previous evidence, we found markedly elevated circulating GDF-15 levels in patients with CHF compared with controls.³⁵ As a stress-inducible member of the TGF-β superfamily, GDF-15 integrates inflammatory and oxidative cues³⁶ and is widely considered a biomarker of myocardial stress and adverse prognosis in CHF.^8,37 Mechanistically, GDF-15 acts on GDNF-family alpha-like (GFRAL) receptors in the brainstem,³⁶ leading to reduced appetite and increased energy expenditure,³⁸ hallmarks of catabolic states. Accordingly, GDF-15 blockade has emerged as a therapeutic concept in catabolic conditions.³⁹ Ponsegromab (PF-06946860), a humanized monoclonal antibody designed to neutralize GDF-15 selectively, has shown encouraging results in early trials for cancer cachexia, improving anorexia, preventing weight loss, and enhancing physical function.⁴⁰ These data support the rationale for exploring GDF-15–targeted strategies also in CC, and are consistent with benefits reported in preclinical CHF models.

However, recent clinical evidence from the phase 2 GARDEN-TIMI 74 trial reported that neutralization of GDF-15 with ponsegromab unexpectedly increased the risk of worsening heart failure despite substantial suppression of circulating GDF-15. These findings suggest that excessive inhibition of GDF-15 may interfere with its endogenous compensatory functions, supporting the concept that this cytokine may act not only as a stress biomarker but also as a tissue-protective mediator.⁴¹

In our cohort, GDF-15 inversely correlated with miR-21 specifically in cachectic CHF patients, suggesting that elevated GDF-15 may reflect intensified metabolic and inflammatory stress in this subgroup, while reduced miR-21 could be associated with pro-inflammatory signalling pathways. This interpretation is consistent with evidence linking miR-21 dysregulation to inflammatory signalling and adverse cardiac remodelling.⁴² Conversely, the positive correlation between GDF-15 and miR-29b in HFpEF without cachexia may suggest an interaction with fibrotic remodelling processes, which are particularly relevant in HFpEF, where reactive fibrosis and concentric hypertrophy predominate.³⁰ This association, observed specifically in patients in sinus rhythm, may indicate that preserved electrophysiological organization is associated with a remodelling signature characterized by coupled GDF-15/miR-29b regulation. These phenotype-specific interactions may support the convergence of GDF-15 and fibrosis-related miRNAs on shared remodelling pathways in CHF.⁴³ Similarly, the positive correlation between GDF-15 and miR-486 may indicate a coordinated adaptive response to cardiac stress, potentially reflecting mechanisms aimed at preserving myocardial integrity. This interpretation is consistent with evidence showing that both GDF-15 and miR-486 participate in adaptive myocardial stress responses.^44,45

FGF-21, a stress-responsive myokine, is induced by mitochondrial and metabolic stress, and recognized as a biomarker of impaired bioenergetics, lipid mobilization, and disease severity in CHF.⁴⁶ While our analysis of the overall cohort of CHF patients revealed a trend toward higher circulating FGF-21 levels than in healthy controls, this difference did not reach statistical significance. This lack of significance may reflect the limited sample size and clinical heterogeneity of our cohort, consistent with a recent meta-analysis reporting substantial heterogeneity across studies.⁴⁷ Interestingly, we observed significantly elevated FGF-21 concentrations in those with CC. This finding is consistent with prior studies documenting elevated FGF-21 levels in patients with CC.⁴⁸ FGF-21 also inversely correlated with miR-486 in cachectic CHF patients, suggesting an interaction between metabolic stress signalling and pathways involved in SM homeostasis.⁴⁵ FGF-21 levels were also higher in patients with CHF with HFrEF compared with those with HFpEF, supporting its role as a biomarker of more advanced myocardial dysfunction and metabolic stress.^46,49 Within the HFrEF subgroup, FGF-21 positively correlated with miR-133a, possibly reflecting a coordinated activation of stress- and remodelling-related pathways in more severely impaired ventricles,⁵⁰ where both markers may respond to increased haemodynamic and metabolic load. Additionally, FGF-21 concentrations were significantly higher in patients classified with more advanced functional impairment, specifically in NYHA class III-IV, compared with class I-II. In NYHA III-IV, the positive correlation between FGF-21 and miR-15b further points to a potential synergistic or interconnected pathway involving metabolic dysregulation and cellular stress, which could exacerbate cardiac pathology and contribute to the poor prognosis typically associated with advanced NYHA classes.

This study has several limitations that should be acknowledged. First, the sample size was relatively small, for both CHF patients and controls, particularly in the stratified subgroups, which may have limited the ability to detect subtler differences. Additionally, an age imbalance existed between groups. This has limited the strength of CHF vs control comparisons, as circulating biomarkers may be influenced by age. In addition, the limited number of CHF patients, particularly after stratification by cachexia status, LVEF phenotype, NYHA functional class, and rhythm status, may have reduced the ability to detect subtler differences. Therefore, both CHF vs control comparisons and within-CHF subgroup analyses should be interpreted with caution and considered hypothesis-generating rather than confirmatory. Additionally, analyses were not adjusted for potential confounders, including age, sex, BMI, comorbidities, medication use, renal function, nutritional status, and HF phenotype, which may influence circulating miRNAs, GDF-15, and FGF-21 levels. Furthermore, because numerous comparisons and correlation analyses were performed across multiple biomarkers and clinical subgroups without formal correction for multiple testing, some statistically significant findings may reflect exploratory associations. Moreover, patients with HFmrEF were excluded, potentially limiting the cohort's representativeness across the full spectrum of CHF phenotypes. Second, the cross-sectional design and single-centre setting preclude causal inferences and may introduce selection bias. At the same time, the use of circulating biomarkers alone does not capture potential tissue-specific regulation of miRNAs and cytokines. Additionally, potential confounders such as comorbidities, medication use, and nutritional status were not controlled through multivariate analyses, and HF-specific variables were not systematically available for all patients, limiting a more detailed assessment of disease burden. Moreover, HF-specific variables such as NT-proBNP, iron status, detailed diuretic burden, and objective exercise capacity were not consistently available. Finally, because circulating biomarkers were assessed cross-sectionally, the present study cannot establish whether the observed miRNA and cytokine alterations represent causal mechanisms, compensatory responses, or markers of disease severity. These findings, therefore, should be interpreted as exploratory and require validation in larger, longitudinal cohorts.

Conclusions

Our findings suggest a coordinated molecular signature that links fibrotic remodelling, metabolic stress, and muscle wasting across the spectrum of CHF. Among all circulating markers, miR-29b showed the most consistent association with disease severity, supporting its potential relevance in fibrotic and catabolic pathways. GDF-15 was markedly elevated in CHF patients and showed phenotype-specific associations with several miRNAs, supporting its association with different clinical and molecular CHF phenotypes. FGF-21 further contributed a metabolic dimension, with higher levels in cachectic patients and in advanced HF phenotypes, and associations with miR-133a and miR-15b, supporting a link between metabolic stress and structural deterioration in CHF. In parallel, the reduced circulating levels of miR-133a and miR-486 across CHF patients may reflect alterations in muscle-related regulatory pathways associated with cardiac and SM remodelling. Overall, our findings support a complementary biomarker framework associated with, systemic stress activation, and metabolic-catabolic alterations in CHF. Given the exploratory nature and cross-sectional nature of this work, larger longitudinal studies are warranted to validate these biomarkers, and to further clarify their biological and clinical significance in patients with CHF and CC.

Author contributions

FT: Methodology, Investigation, Data curation, Formal Analysis, Writing—original draft, Writing—review and editing. GI: Performed Clinical Studies, Formal Analysis. CG: Contributed to Patient Enrolment and Provided Study Materials. PN: Contributed to Patient Enrolment and Provided Study Materials. SO: Methodology, Provided Administrative Support. RB: Methodology, Provided Administrative Support. CR: Methodology, Provided Administrative Support. SDA: Conceptualization, Data curation, Supervision, Formal Analysis, Writing—original draft, Writing—review and editing. AM: Conceptualization, Data curation, Supervision, Formal Analysis, Writing—original draft, Writing-review and editing. MM: Conceptualization, Data curation, Supervision, Formal Analysis, Funding acquisition, Methodology, Writing—original draft, Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Supplementary data

Supplementary data are available at ESC Heart Failure online.

Declarations

Disclosure of Interest

SDA declares grants from CSL-Vifor and Abbott Laboratories, and personal fees for consultancies, trial committee work and/or lectures from Actimed, Alleviant, Amgen, Astra Zeneca, Bayer, Berlin Heals, Bimyo, Boehringer Ingelheim, Brahms, Cardiac Dimensions, Cardior, Cordio, Corvia, CSL-Vifor, CVRx, Cytokinetics, Edwards, Impulse Dynamics, Lilly, Mankind Pharma, Novo Nordisk, Occlutech, Pfizer, Pulnovo, Regeneron, Relaxera, Repairon, Scirent, Sensible Medical, Vectorious, Veru, Viscardia, Vivus, and V-Wave. He is a named co-inventor of two patent applications regarding MR-proANP (DE 102007010834 and DE 102007022367), but he does not personally benefit from the related issued patents. The other authors declare that they have no relevant financial or non-financial interests to disclose.

Data Availability

No data were generated or analysed for this manuscript.

Funding

The author(s) declare that financial support was received for the research, authorship, and/or publication of this article. This study was supported by funding from the European Union received by MM (SEED PNR 2021, CUP B89J21032850001 and National Recovery and Resilience Plan, Project PE_00000019 ‘HEAL ITALIA’, CUP B53C22004000006).

Ethical Approval

Ethical Approval was not required.

Pre-registered Clinical Trial Number

None supplied.