Heart failure (HF) is a clinical syndrome recognized as a global epidemic, representing the end stage of most cardiovascular diseases, and is one of the leading causes of death and disability. Over the past few decades, the incidence of HF has gradually reached a stable and decreasing trend in developed countries. For instance, the incidence rates are 6.5 per 1,000 in Germany, 2.9–3.9 per 1,000 in Spain, 2.2–3.2 per 1,000 in America, and 0.7 per 1,000 in Hong Kong (Conrad et al., 2018; Savarese et al., 2023). Despite this slight decline in incidence, the prevalence of HF is gradually increasing and varies considerably across countries and regions. The highest prevalence in 2017 was noted in Central Europe, North Africa, and the Middle East, while lower rates were observed in Eastern Europe and Southeast Asia. The prevalence rates range from 0.9% to 6.8% in Spain, 3.9% in Germany, 2.4%–3.0% in America, and 0.4% in Thailand (Savarese et al., 2023). Furthermore, HF imposes a significant economic burden on healthcare systems worldwide, with the total cost of treating HF in the United States projected to rise from $31 billion to $70 billion between 2012 and 2030 (Savarese and Lund, 2017; Heidenreich et al., 2013). This escalation undoubtedly places a substantial strain on healthcare expenditures.

The diagnosis of HF primarily relies on electrocardiograms, imaging, laboratory tests, and biomarker assessments. B-type Natriuretic Peptide (BNP) and N-terminal Pro-B-Type Natriuretic Peptide (NT-proBNP) are widely regarded as the most effective biomarkers for diagnosing HF due to their significant roles in diagnosis and prognostic evaluation (Castiglione et al., 2022). The American College of Cardiology/American Heart Association (ACC/AHA) endorses the use of BNP and NT-proBNP to assist in diagnosing HF (Yancy et al., 2017). However, the European Society of Cardiology (ESC) guidelines advocate for the use of these biomarkers to rule out HF, considering the impact of gender, age, and comorbidities (Ponikowski et al., 2016). Consequently, many guidelines suggest that thresholds for biomarkers should be determined with consideration for age, gender, and ethnic region stratification (Kavsak et al., 2019). In terms of treatment, HF therapeutic strategies have evolved from the traditional “Golden Triangle”- comprising angiotensin-converting enzyme inhibitors or angiotensin II receptor antagonists, beta-blockers, and mineralocorticoid receptor antagonists-to the “New Quadruple Combination,” which adds sodium-glucose cotransporter 2 inhibitors (McDonagh et al., 2021). The approach has further advanced to the current “Five Golden Flowers,” with the addition of soluble guanylate cyclase stimulators, such as vericiguat (Metra et al., 2023). This progression is due to the residual risk of HF exacerbation and death that persists even with the “New Quadruple Combination” therapy (Docherty et al., 2020). It highlights that, despite ongoing advancements in HF drug development, the diagnosis and treatment of HF continue to face numerous challenges. These challenges stem from the complexity and refractory nature of HF, as well as its poor prognosis, and include individual differences, comorbidity management, medication side effects, and economic burdens.

Significant gender differences exist in HF, encompassing symptoms, susceptibility, risk factors, pathophysiology, and response to treatment. Studies have shown that women are more likely to present with severe symptoms, with dyspnea being more predominant in women and peripheral edema in men (Maidana et al., 2023). Risk factors such as hypertension, diabetes, smoking, and obesity make women more susceptible to HF than men. Additionally, women face unique risk factors, including those related to menopause, breast cancer treatments, and pregnancy (Maidana et al., 2023; Lala et al., 2022). Biomarkers, including NT-proBNP, CA125, high-sensitivity troponin, galectin-3, and osteopontin, also exhibit gender-specific differences (Maidana et al., 2023). Racial differences are equally important factors influencing the diagnosis and prognosis of HF. Research published in the Journal of the American Medical Association indicates that African American individuals have nearly twice the incidence of HF, experience an earlier onset of the disease, present with higher severity at diagnosis, and have a higher mortality rate among younger individuals (45–64 years) (Yancy, 2024). Age is also a key risk factor for HF, with significant variations in incidence, progression, and biomarker expression across different age groups. Global Burden of Disease database analysis indicates that the number of HF cases increases significantly with age, particularly among individuals over 65 years old (Ran et al., 2025; Kang et al., 2025). Multiple studies have demonstrated that NT-proBNP levels, a marker for HF, exhibit significant variations across different age groups and genders, with elevated levels being more common in the general middle-aged population (Welsh et al., 2022; Mu et al., 2023). Therefore, studies focusing on gender, age, and racial differences may offer insights to improve the diagnosis of HF and potentially contribute to the development of new drugs targeting HF.

Due to database limitations, specifically the lack of basic age information, this study focused on four distinct populations: African American, Caucasian, German, and Spain. Differential expression analysis was employed to identify gender-specific differentially expressed genes (DEGs) within these populations. Enrichment analysis was utilized to explore the potential pathogenesis of HF, while immune infiltration analysis was applied to investigate the microenvironment of immune infiltration across populations. Additionally, Weighted Gene Co-expression Network Analysis (WGCNA) was applied to identify the most relevant gene modules associated with HF in different populations. Finally, three machine learning models were applied to screen for gender-specific Hub genes in different populations: the least absolute shrinkage and selection operator (LASSO) regression, support vector machine-recursive feature elimination (SVM-REF), and random forest (RF). These models were used to construct nomograms to predict the risk of HF.

Symptoms, pathogenesis, and biomarkers of HF are influenced by multiple factors, particularly gender, age, and race. Several scholars have emphasized the importance of linking HF biomarkers to gender (Maidana et al., 2023; Blumer et al., 2023). Studies have shown that baseline NT-proBNP levels are higher in women than in men, especially in premenopausal women, and that other HF biomarkers, such as the soluble isoform of suppression of tumorigenesis-2 (sST2), are higher in men than in women (Maidana et al., 2023). Racial disparities also play a significant role in the development of HF. Despite improvements in HF treatments and overall survival, the mortality disparity for African American patients continues to widen over time (Lewsey and Breathett, 2021). Hale suggests that future HF research should be grounded in understanding these racial disparities (Hale and Yancy, 2023). Therefore, this study focused on race and gender to identify HF-specific biomarkers.

Enrichment analyses in this study highlight the importance of the wnt signaling pathway in the pathogenesis of HF. The Wnt signaling pathway is a fundamental cellular communication system comprising the β-linker classical pathway and the nonclassical pathways, namely, the planar cell polarity and the calcium-dependent pathways. It has been shown that the classical Wnt signaling promotes inflammation and fibrosis in the context of cardiac hypertrophy and HF (Horitani and Shiojima, 2024); non-classical WNT signaling produces contractile dysfunction by affecting myocardial oxidative stress, inflammation, reparative capacity, energetics, and remodeling, including fibrosis or fatty infiltration of the myocardium (Akoumianakis et al., 2022).

Several studies have investigated biomarkers for HF. For instance, Fan’s research identified core genes such as EIF1AY, RPS4Y1, USP9Y, KDM5D, DDX3Y, NPPA, HBB, TSIX, LOC28556, and XIST through protein-protein interaction networks (Fan and Hu, 2022). Zhu’s study identified NPPA, OMD, and PRELP as biomarkers for dilated cardiomyopathy and HF using random forests (Zhu et al., 2022). Chen’s research identified 16 differentially expressed genes (DEGs) for HF using random forests, which included ECM2, LUM, ISLR, ASPN, PTN, SFRP4, GLT8D2, FRZB, FCN3, TEAD4, NPTX2, LAD1, ALOX5AP, RNASE2, IL1RL1, and CD163 (Chen et al., 2023). Additionally, Chen identified NSG1, NPPB, PHLDA1, and SERPINE2 using LASSO and SVM-REF (Chen et al., 2022). These genes were also addressed in this study. For example, five of the ten genes identified by Fan (EIF1AY, RPS4Y1, USP9Y, KDM5D, DDX3Y) were found to be upregulated exclusively in males, in both control and disease groups; NPPA identified by Zhu did not differ in this study in the Spanish population (P > 0.05); Chen similarly identified ASPN as DEGs for HF; and ASERPINE2 identified by Chen was considered as male-specific DEGs in this study. In conclusion, this study encompassed datasets from multiple regions, stratified by gender, and employed three machine learning models (LASSO, SVM-REF, and RF) to identify Hub genes. First, analysis of 650 samples through differential expression analysis, DEGs, WGCNA, gender stratification, external validation across four datasets, and machine learning identified three Hub genes: MYH6, ASPN, and COL14A1.

ASPN is a member of the small leucine-rich proteoglycan family, specifically class I. Multiple bioinformatics analyses have identified ASPN as a potential biomarker for HF (Boyang et al., 2022; Guo et al., 2022; Wang et al., 2019; Huang et al., 2024). ASPN primarily encodes the asporin protein, which acts as an inhibitor of transforming growth factor-β1 (TGF-β1) and is considered a beneficial regulator of cardiac remodeling (Huang et al., 2022). In the ASPN knockout (Aspn^−/−) mouse model, increased fibrosis and reduced cardiac function were observed following pressure overload (Huang et al., 2022). The TGFβ superfamily is one of the most important families of profibrotic cytokines in the regulation of myocardial fibrosis. Studies have shown that while inhibiting TGF-β1 may exacerbate early cardiac dysfunction, it can prevent late remodeling post-infarction, and inhibiting TGF-β1 is a significant factor in protecting the myocardium from fibrosis (Ikeuchi et al., 2004). Huang’s research found that asporin, released by cardiac fibroblasts, was able to attenuate TGFβ signaling, thereby inhibiting the progression of myocardial fibrosis (Huang et al., 2022). However, other studies have indicated that asporin plays a key role in glycated low-density lipoprotein (gly-LDL)-induced apoptosis of cardiomyocytes, increasing H9C2 cardiomyocyte apoptosis by downregulating Bcl-2, upregulating TGF-β1, Bax, type III collagen, fibronectin, and the phosphorylation of smad2 and smad3 (Li et al., 2020). On the other hand, Medzikovic’s research found that miR-129-5p expression was reduced and ASPN expression was enhanced in cardiac fibrosis and calcified human heart fibroblasts. Overexpression of miR-129-5p was able to downregulate ASPN expression, and targeting the miR-129-5p/ASPN signaling axis in cardiac fibroblasts attenuated myocardial fibrosis and calcification and restored cardiac function in mice (Medzikovic et al., 2023). In conclusion, the role of ASPN in HF is complex, but it is considered a promising potential biomarker for HF, and its specific role in HF still needs to be further investigated to clarify (Zhang et al., 2021).

COL14A1 is a major fibrillar collagen produced by fibroblasts and plays a crucial role in regulating the extracellular matrix component of the cardiac remodeling process in HF (Frangogiannis, 2019). Portokallidou has identified COL14A1 as a key gene in both dilated and ischemic cardiomyopathy through transcriptomic and proteomic analyses (Portokallidou et al., 2023). COL14A1-deficient mouse ventricles exhibit morphological defects and disorganization of collagen fibers (Tao et al., 2012). COL14A1 functions as a regulator of tissue differences, particularly during the early stages of collagen fiber formation, which is crucial for myocardial growth and structural integrity (Tao et al., 2012; Ansorge et al., 2009).

Myosin is a hexamer composed of two heavy chain subunits, two light chain subunits, and two regulatory subunits, possessing ATPase activity and the ability to bind actin. Among these components, the myosin heavy chain (MyHC) is an essential part of myocardial structure and function, playing a vital role in cardiac contractile function (Toepfer et al., 2020). The MYH6 and MYH7 genes encode the α-MyHC subunit and the β-MyHC subunit, respectively. In the human heart, α-MyHC is predominant in the atria, while β-MyHC is predominant in the ventricles (Walklate et al., 2021). The content of α-MyHC varies among different mammalian hearts: it is 100%/100% in mouse ventricles/atria, 90%/99% in rats, and 5%/75% in humans (Walklate et al., 2021). In pathological states such as HF, myosin genes may undergo “return to the fetal gene program,” where α-MyHC expression decreases and β-MyHC expression increases, leading to a reduced α-MyHC/β-MyHC ratio. This change may help maintain myocardial contractility and compensate for cardiac function in the short term, but if it persists, it can adversely affect energy metabolism (Rajabi et al., 2007; Taegtmeyer et al., 2010; Papait et al., 2013; Herron and McDonald, 2002). Several studies have shown that mRNA and protein expression levels of α-MyHC are significantly downregulated in patients with HF or cardiac hypertrophy, as well as in various animal models of HF (Reiser et al., 2001; Miyata et al., 2000; Lowes et al., 1997; Nakao et al., 1997). Furthermore, the expression of the MYH6 gene changes with the improvement of clinical symptoms during the treatment of cardiomyopathy or HF. Before treatment, MYH6 gene expression is downregulated, but it is upregulated during many therapeutic measures aimed at improving cardiac function. For instance, β-blockers can lead to upregulation of MYH6 gene expression and downregulation of MYH7 gene expression while improving ejection fraction and cardiac function. Among patients treated with β-blockers, those with improved ejection fraction exhibit an increase in α-MyHC mRNA and a decrease in β-MyHC mRNA compared to non-responders (Lowes et al., 2002).

In addition, this study identified Omecamtiv Mecarbil, Danicamtiv, and Mavacamten as potential drugs for the treatment of HF through database screening. Omecamtiv Mecarbil and Danicamtiv act as cardiac myosin activators (Nanasi et al., 2018; Kooiker et al., 2023), while Mavacamten is a cardiac-specific myosin inhibitor (Braunwald et al., 2023). Omecamtiv Mecarbil enhances myocardial contractility by specifically binding to the catalytic S1 structural domain of cardiac myosin, thereby improving cardiac function, reducing ventricular wall stress, reversing ventricular remodeling, and promoting sympathetic regression for the treatment of HF (Teerlink et al., 2020; Teerlink et al., 2021). Danicamtiv potentially enhances myocardial force and calcium sensitivity by increasing myosin recruitment and slowing cross-bridge turnover (Kooiker et al., 2023). However, Danicamtiv is still in clinical trials, and although it improves cardiac systolic function, it may limit diastolic function at high concentrations (Voors et al., 2020; Ráduly et al., 2022). Mavacamten reduces contractility by normalizing cross-bridging between myosin and actin and is commonly used to treat obstructive hypertrophic cardiomyopathy (Schenk and Fields, 2023). A meta-analysis showed that Mavacamten reduced New York Heart Association (NYHA) class and post-exercise left ventricular outflow tract gradient, and increased mixed venous oxygen pressure in patients with hypertrophic cardiomyopathy, but it may also cause adverse effects such as atrial fibrillation and reduced left ventricular ejection fraction (Bishev et al., 2023). The DGIdb database indicates that Omecamtiv Mecarbil, Danicamtiv, and Mavacamten may act on MYH6. However, there have been an absence of studies confirming the targeted effects of these drugs on MYH6, and these findings must be validated through subsequent in-depth in vitro and in vivo experiments.

This study also identified gender-specific DEGs across different regions. After stratifying by gender, differential expression analysis, WGCNA, and machine learning were performed to identify race- and gender-specific DEGs. Subsequently, a nomogram was constructed based on Hub genes (MYH6, ASPN, and COL14A1) and combined with specific DEGs to predict the risk of HF onset. Exercise stimulation induced cardiac-specific expression of the C-terminal domain of CCDC80, which prevented angiotensin II-induced cardiac hypertrophy and fibrosis in mice (Yin et al., 2022). CX3CR1 has been suggested to be a prerequisite for the development of cardiac hypertrophy and left ventricular dysfunction in mice upon transverse aortic constriction (Weisheit et al., 2021). Furthermore, the activation of the cardiac CX3CL1/CX3CR1 signaling axis delays β-adrenergic-induced HF (Flamant et al., 2021). C1QTNF3 has been found to be decreased in patients with HF with reduced ejection fraction in the Xi’an population of China and is strongly associated with increased morbidity and mortality (Gao et al., 2019). However, this study found that C1QTNF3 was upregulated in Caucasian males, which contrasts with another study that found C1QTNF3 upregulated in mouse hypertrophied hearts and in human hearts with HF, derived from cardiomyocytes and induced by the production of reactive oxygen species during the hypertrophic response. Additionally, CTRP3 facilitates pressure-overload-induced cardiac hypertrophy through activation of the TAK1-JNK axis (Ma et al., 2019). Yu identified m6A-modified C14orf132 as a potential diagnostic gene for idiopathic cardiomyopathy (Yu et al., 2024). The relationship between other genes and HF remains unclear and requires further investigation.

In conclusion, this study utilized bioinformatics to identify the pathogenesis and potential biomarkers of HF associated with dilated cardiomyopathy. First, three Hub genes (MYH6, ASPN, and COL14A1) associated with dilated cardiomyopathy-related HF were identified using a large-scale training dataset. External validation confirmed the differential expression of these genes. The research underscores that the pathogenesis of HF is closely related to inflammatory responses, immune responses, vascular regulation, the Wnt signaling pathways, glutathione metabolism, and apoptosis. The myocardial immune infiltrate microenvironment in HF patients is dysregulated and exhibits significant differences across distinct patient populations. The myocardial immune infiltrate microenvironment is dysregulated in patients with HF, characterized by a high abundance of naïve B cells and CD8 T cells, and a lower abundance of resting memory CD4 T cells, M2 macrophages, and eosinophils. Differential analyses were conducted to pinpoint population-specific DEGs, and gender-specific Hub genes were identified using three machine learning models: LASSO, SVM-REF, and RF. This study is innovative in its focus on race and gender, identifying HF-specific DEGs across different races and genders, thereby reflecting the principles of precision and individuality in medicine. Furthermore, the study identified Hub genes using multiple machine learning models, which were then validated in four external validation sets. Additionally, the study identified five genes that were upregulated only in male and not associated with HF.

However, there are limitations to this study. Although efforts were made to include HF datasets from various regions, database limitations precluded the inclusion of more HF-related datasets from other countries. The Japanese-related dataset GSE8331 was also not considered for inclusion due to its small sample size. While many specific DEGs were identified in this study, the relationship between many of these genes and HF remains unclear, necessitating further experimental validation.

However, this study has several limitations. First, the large number of datasets included in this study were sourced from various platforms and countries. Despite batch effect correction, differences in platforms, processing pipelines, and sample demographics may still introduce confounding effects that are difficult to completely eliminate. Second, although efforts were made to include HF datasets from different regions, limitations of the databases restricted the inclusion of more HF-related datasets from other countries. Moreover, there was significant variation in sample sizes across different datasets, with the Spanish cohort having a particularly small sample size, which may have reduced the reliability of the results. Additionally, while the US datasets provided detailed racial information, the population information for the other datasets was inferred based on the country of origin. However, the included populations may not have had strictly defined racial information sources, which could have affected the reliability of the results. Furthermore, although CIBERSORT has been widely applied across various tissue types, it is important to acknowledge that its reference matrix was primarily developed using peripheral blood mononuclear cells. Potential differences in gene expression profiles between blood and heart cells may have affected the accuracy of immune cell estimation. Future studies incorporating heart-specific signature matrices will help to validate and refine these findings. Age is also an important factor associated with HF. However, many of the included datasets did not provide patient age information, and thus, this study did not perform subgroup analyses based on age. This is a key aspect that needs to be addressed in future research. In addition, this study predicted drug interactions for Hub genes based on databases. However, such drug repositioning predictions are merely preliminary hypothesis-generating tools, and the reliability of their results must be verified through subsequent in-depth in vitro and in vivo experiments. Finally, although this study identified many specific DEGs, these computational results lack experimental validation. The relationships between many genes and HF remain unclear, and they have not been independently verified in patient samples or experimental models (such as qRT-PCR, Western blot, and immunohistochemistry). The biological reliability and translational applicability of the proposed biomarkers remain uncertain and require further experimental validation.

The biomarkers of HF vary significantly across different populations and genders. MYH6, ASPN, and COL14A1 may be potential biomarkers for HF in dilated cardiomyopathy.