Highlights
This comprehensive plasma peptidome analysis reveals critical insights into heart failure pathophysiology and patient stratification.
First, the study quantified 21,694 unique peptides in plasma from 486 heart failure patients compared to 98 age-matched controls, identifying 1,924 differentially expressed peptides that distinguish diseased from healthy states.
Second, machine learning analysis identified 141 high-scoring bioactive peptide candidates, with 65 peptides independently associated with clinical outcomes, predominantly involving the renin-angiotensin system, natriuretic peptides, and cardiometabolic pathways.
Third, hierarchical clustering revealed three major patient clusters with distinct peptide signatures, where the cluster showing the lowest survival probability exhibited heightened acute phase response and increased inflammatory markers.
Background
Heart failure represents a global pandemic affecting over 64 million individuals worldwide, with staggering morbidity, mortality, and healthcare costs. Despite advances in pharmacotherapy and device-based interventions, prognosis remains poor, particularly for patients with reduced ejection fraction. The identification of reliable biomarkers for early diagnosis, risk stratification, and therapeutic targeting remains an urgent clinical unmet need.
Peptides serve as critical signaling molecules in cardiovascular homeostasis, with established roles for natriuretic peptides in diagnosis and angiotensin peptides in treatment. However, previous biomarker discovery efforts have predominantly focused on intact proteins, potentially missing crucial information encoded in the low molecular weight peptidome. The plasma peptidome—comprising endogenous peptides generated through proteolytic processing—reflects real-time physiological and pathological processes with potential for capturing disease-specific signatures.
The absence of systematic, unbiased studies examining the complete peptidome in heart failure patients has limited our understanding of disease-specific peptide networks. This gap represents both a scientific opportunity and a clinical imperative, as deciphering these signals could revolutionize diagnostic approaches, prognostic models, and therapeutic targeting strategies.
Study Design
This cross-sectional investigation employed mass spectrometry-based proteomics to comprehensively characterize the plasma peptidome in heart failure patients. The study enrolled 486 patients meeting established heart failure criteria alongside 98 age-matched non-heart failure controls from comparable demographic backgrounds.
Peptide quantification followed a rigorous three-tier ranking methodology. First, relative upregulation in heart failure versus control populations was assessed. Second, pattern similarity to known bioactive peptides was evaluated using an adapted machine learning approach. Third, association with clinical outcomes—including mortality, hospitalization, and composite endpoints—was examined through multivariate regression modeling.
Differential expression analysis utilized stringent statistical thresholds to identify peptides with significant abundance changes. Hierarchical clustering was performed to subgroup patients based on peptide signature similarity, enabling identification of distinct phenotypic clusters with potential clinical implications.
Key Findings
Peptidome Landscape in Heart Failure
The mass spectrometry analysis revealed an unprecedented view of the plasma peptidome, quantifying 21,694 unique peptides across the study population. Among these, 1,924 peptides demonstrated statistically significant differential expression between heart failure patients and healthy controls, representing approximately 9% of the total peptide repertoire analyzed.
This substantial number of regulated peptides underscores the profound remodeling of circulating peptide networks in heart failure, extending far beyond traditionally studied biomarkers such as B-type natriuretic peptide (BNP) and N-terminal pro-BNP (NT-proBNP).
High-Ranking Peptide Candidates
Among peptides demonstrating elevated abundance in heart failure patients, several categories emerged as particularly prominent. Angiotensin-related peptides featured prominently, with angiotensin 1-9 showing marked upregulation. This finding is noteworthy given the traditional focus on angiotensin II in heart failure pathophysiology, suggesting additional roles for alternative angiotensin metabolites in disease progression.
Propeptides derived from metabolic hormones—including gastric inhibitory polypeptide (GIP), osteocalcin, and cholecystokinin—were substantially regulated, indicating cross-talk between cardiovascular dysfunction and metabolic regulation. These findings support the emerging concept of cardiometabolic interplay in heart failure pathogenesis.
Peptides mapping to the extracellular domain of the natriuretic peptide clearance receptor (NPR-C) demonstrated altered abundance, potentially reflecting compensatory mechanisms for managing natriuretic peptide signaling. Similarly, peptides derived from integrin alpha-7, a component of the cardiomyocyte extracellular matrix, suggest involvement of structural and adhesion pathways in disease manifestation.
Machine Learning-Derived Bioactivity Scoring
Applying an adapted machine learning algorithm to assess pattern similarity between identified peptides and established bioactive peptide sequences yielded 141 candidates scoring within the top 5% of all analyzed peptides. This computational approach prioritized peptides most likely to possess biological activity rather than merely serving as disease markers.
Among these high-scoring candidates, 65 peptides demonstrated independent association with clinical outcomes after adjustment for conventional risk factors and established biomarkers. This subset represents potential therapeutic targets or biomarker candidates warranting further investigation.
Patient Stratification and Cluster Analysis
Hierarchical clustering of heart failure patients based on peptide signatures revealed three major patient clusters with distinct biological and clinical characteristics. This unsupervised approach identified natural groupings independent of traditional classification systems such as ejection fraction category or New York Heart Association (NYHA) functional class.
Critically, the cluster exhibiting the lowest survival probability demonstrated a specific peptide degradation pattern characterized by higher proportions of peptides linked to acute phase response and inflammatory pathways. This observation suggests that systemic inflammation and catabolic processes may drive adverse outcomes in this subgroup, potentially identifying patients who might benefit from targeted anti-inflammatory or immunomodulatory therapies.
The identification of peptide-defined clusters with prognostic significance represents a paradigm shift from traditional phenotype-based stratification toward molecularly-defined patient subgroups.
Expert Commentary
The study by Madsen and colleagues represents a methodological tour de force in cardiovascular biomarker discovery, introducing several innovations that advance the field beyond conventional approaches. The integration of mass spectrometry-based peptidomics with machine learning and clinical outcome modeling creates a comprehensive framework for translating molecular observations into clinically actionable insights.
Several aspects merit particular attention. First, the study’s scale—nearly 500 heart failure patients with comprehensive peptide profiling—provides statistical power rarely achieved in proteomic investigations. Second, the three-tier ranking system addresses a fundamental challenge in biomarker discovery: distinguishing between markers that merely reflect disease presence versus those with biological relevance and prognostic significance.
However, certain limitations warrant consideration. The cross-sectional design precludes causal inference regarding peptide abundance changes and disease progression. The single-center nature of the study raises questions about generalizability across diverse populations and healthcare settings. Additionally, while the machine learning approach enhances bioactivity prediction, functional validation through in vitro and in vivo studies remains essential.
The observation linking inflammatory peptide patterns to adverse outcomes aligns with emerging evidence supporting a role for immune activation in heart failure progression. Whether these peptide signatures represent passive markers of systemic inflammation or active contributors to disease pathogenesis remains to be elucidated. Nonetheless, this finding opens avenues for exploring anti-inflammatory interventions in selected patient populations.
Conclusion
This groundbreaking peptidome analysis establishes a new frontier in heart failure biomarker research, demonstrating that the circulating peptide landscape harbors rich information beyond traditional protein-based biomarkers. The identification of 1,924 differentially expressed peptides, including 65 independently outcome-associated candidates, provides an extensive resource for future investigation.
The study’s most compelling contributions include the characterization of angiotensin-related, natriuretic peptide, and cardiometabolic peptides as key players in heart failure pathophysiology, and the demonstration that peptide-defined patient clusters can identify individuals at differential mortality risk. The inflammatory signature associated with worst outcomes particularly warrants therapeutic investigation.
For clinicians, these findings suggest that comprehensive peptide profiling could eventually inform personalized risk stratification and therapeutic decision-making. For researchers, the identified peptide candidates provide starting points for mechanistic studies and targeted therapy development. For the broader scientific community, this work establishes methodology applicable to other cardiovascular conditions where peptidome characterization might yield diagnostic or prognostic value.
Future directions should include longitudinal validation studies, functional characterization of top peptide candidates, and investigation of therapeutic modulation of identified pathways. The convergence of mass spectrometry, computational biology, and clinical investigation exemplified in this study points toward a future where molecular phenotyping guides precision cardiology.
Funding and Study Registration
This study received support from institutional research grants. The investigation was conducted according to Good Clinical Practice guidelines and applicable regulatory requirements. Full funding disclosure and conflict of interest statements are available in the original publication.
References
Madsen CT, Refsgaard JC, Voordes GHD, van Essen BJ, Ouwerkerk W, Hoegl A, Grønborg M, Tromp J, Lang CC, Barascuk-Michaelsen N, Voors AA. Decoding the Heart Failure Peptidome. Circulation. Heart failure. 2026-03-24:e013290. PMID: 41874184.
Ponikowski P, Voors AA, Anker SD, et al. 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J. 2016;37(27):2129-2200.
Chung MK, Szymanski MK, Greenberg RA, et al. Emerging biomarkers in heart failure: from bench to bedside. JACC Heart Fail. 2024.
