Comprehensive Plasma Oxylipin Profiling Reveals a Pro-Inflammatory Eicosanoid Signature and Diagnostic Biomarker Panel in Dilated Cardiomyopathy

Introduction

Dilated cardiomyopathy (DCM) is a primary myocardial disorder characterized by ventricular enlargement and impaired systolic function, which frequently progresses to advanced heart failure with high morbidity and mortality [1]. Despite the widespread use of echocardiography, cardiac magnetic resonance imaging, and conventional circulating biomarkers – such as N-terminal pro-B-type natriuretic peptide (NT-proBNP) and troponins – current tools remain limited in their ability to detect DCM at an early stage or to accurately predict disease progression. Although omics-based studies, including genomics and transcriptomics, have provided valuable mechanistic insights, translation into clinically actionable diagnostics has been slow, partly due to phenotypic heterogeneity, modest discriminatory power, and the lack of validated molecular targets [2]. These limitations underscore the urgent need for novel, biologically informative biomarkers to improve early detection, refine risk stratification, and guide individualized management strategies.

Emerging evidence has highlighted the pivotal role of lipid metabolism in myocardial structure and function, and there is increasing recognition that dysregulated lipid signaling contributes to adverse remodeling in cardiomyopathy [3,4]. Oxidized lipid mediators, collectively termed oxylipins, are bioactive molecules generated through enzymatic (cyclooxygenase [COX], lipoxygenase [LOX], cytochrome P450) and non-enzymatic oxidation of polyunsaturated fatty acids. These mediators regulate vascular tone, inflammatory cascades, and immune cell function; their imbalance has been implicated in a spectrum of cardiovascular disorders, including ischemic heart disease, myocarditis, and heart failure [5–9]. In DCM specifically, altered profiles of oxylipins derived from arachidonic acid (AA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) have been reported, yet findings remain inconsistent and mechanistic links to myocardial dysfunction are poorly defined [10]. Conflicting results regarding the direction and magnitude of oxylipin alterations – potentially reflecting differences in patient populations, disease stage, or analytical platforms – highlight the need for systematic, high-resolution lipidomic profiling in well-characterized DCM cohorts.

Against this background, the present study was designed to evaluate the diagnostic potential of a 6-oxylipin plasma panel in distinguishing patients with DCM from age- and sex-matched healthy controls, and to investigate the relationship between oxylipin patterns and established clinical phenotypes. Using a high-resolution liquid chromatography-tandem mass spectrometry (LC-MS/MS) lipidomic platform, we quantified targeted oxylipin species derived from AA, EPA, and DHA. Multivariate modeling was conducted to identify a lipid signature with maximal discriminatory capacity, assess its correlations with functional and biochemical indices, and develop a support vector machine (SVM)–based diagnostic model. Beyond improving diagnostic precision, such lipidomic insights may inform the development of precision therapies – for example, by identifying patients who could benefit from interventions targeting inflammatory lipid pathways, pharmacologic modulation of COX/LOX/P450 activity, or dietary strategies designed to optimize polyunsaturated fatty acid substrates.

Material and Methods

STUDY POPULATION AND ETHICAL APPROVAL:

Between July 2024 and January 2025, 30 patients with a confirmed diagnosis of DCM and 30 age- and sex-matched healthy controls undergoing routine health examinations were prospectively enrolled. Inclusion and exclusion criteria were applied to minimize confounding influences on lipidomic analysis. Inclusion criteria: A diagnosis of DCM was established in accordance with international clinical guidelines [11]. Specifically, DCM was defined by the presence of left ventricular (LV) dilation – indexed LV end-diastolic diameter exceeding the upper limit of normal for age and body surface area – together with reduced systolic function, typically a left ventricular ejection fraction (LVEF) below 45%, as determined by echocardiography or cardiac magnetic resonance imaging. These findings were required to occur in the absence of abnormal loading conditions (e.g., uncontrolled hypertension, clinically significant valvular disease) or coronary artery disease sufficient to explain the degree of LV dysfunction. Secondary forms of cardiomyopathy, including ischemic, hypertensive, valvular, or congenital heart disease, were systematically excluded by clinical evaluation, imaging, and, when indicated, coronary angiography. Additional criteria included age 40 to 65 years to reduce variability related to age-associated changes in lipid metabolism and stable cardiac function without acute decompensation during the preceding 3 months. Exclusion criteria: Patients with metabolic disorders affecting lipid profiles (e.g., diabetes mellitus, obesity, dyslipidemia), use of lipid-lowering drugs (statins or fibrates) within the past 4 weeks, renal or hepatic dysfunction (abnormal serum creatinine or liver enzymes), or recent cardiovascular procedures (e.g., pacemaker or defibrillator implantation) were excluded. Healthy controls were matched to cases by age (±2 years) and sex; they were screened to exclude cardiovascular disease, metabolic disorders, and recent medication use affecting lipid metabolism. The selection of age- and sex-matched controls was intended to reduce demographic confounding while providing a baseline lipidomic profile representative of healthy middle-aged individuals.

The study protocol was approved by the Institutional Review Board of Tianjin Chest Hospital (No. 2024LW-015), and written informed consent was obtained from all participants. Confidentiality was maintained via coded identifiers; de-identified data may be shared in accordance with institutional guidelines.

SAMPLE SIZE DETERMINATION:

The sample size (n=60) was estimated based on unpublished pilot observations from our group, used solely to guide power calculations, which indicated large effect sizes (Cohen’s d >1.0) for multiple oxylipins between DCM and control groups, providing greater than 90% power to detect significant differences at α=0.05 with 25 participants per arm. An enrollment target of 30 participants per group was selected to enhance model stability for multivariate classification techniques.

BLOOD SAMPLE COLLECTION AND PROCESSING:

Fasting peripheral venous blood was collected from patients with DCM at hospital admission and from controls during routine health examinations, using sodium ethylenediaminetetraacetic acid tubes to prevent coagulation and minimize enzymatic degradation. Samples were placed on ice immediately after collection, transported to the laboratory on the same day, and centrifuged at 1,500×g for 10 min at 4°C. Plasma was aliquoted into pre-labeled tubes and stored at −80°C until analysis. Environmental conditions during sample handling (ambient temperature 20–22°C; relative humidity 45–55%) were monitored to ensure consistency. Demographic and clinical data were obtained for subsequent correlation analyses.

REAGENTS:

AA, EPA, docosapentaenoic acid, DHA, corresponding metabolite standards, and isotopically labeled internal standards (deuterated 6-keto prostaglandin F1α [6-keto-PGF1α-d4], deuterated prostaglandin E2 [PGE2-d4], deuterated leukotriene B4 [LTB4-d4], deuterated 11,12-dihydroxyeicosatrienoic acid [11,12-DHET-d11], deuterated 20-hydroxyeicosatetraenoic acid [20-HETE-d6], deuterated 5-hydroxyeicosatetraenoic acid [5-HETE-d8], deuterated 8,9-epoxyeicosatrienoic acid [8,9-EET-d11], deuterated arachidonic acid [ARA-d8], deuterated eicosapentaenoic acid [EPA-d5], deuterated docosahexaenoic acid [DHA-d5]) were obtained from Cayman Chemical Co. (Ann Arbor, MI, USA). Prostaglandin F2α (PGF2α) was also purchased from Cayman Chemical. Butylated hydroxytoluene was purchased from Sigma-Aldrich (St. Louis, MO, USA). LC-MS-grade acetonitrile and high-performance liquid chromatography (HPLC)-grade methanol were obtained from Merck (Darmstadt, Germany). Ethyl acetate and formic acid (HPLC-grade or higher) were obtained from Fisher Scientific (Pittsburgh, PA, USA). Oasis hydrophilic-lipophilic balance 10 mg solid-phase extraction cartridges were purchased from Waters Co. (Milford, MA, USA). Centrifuge tube filters were obtained from Corning Co. (Corning, NY, USA). Unless otherwise specified, additional reagents were purchased from Sigma-Aldrich.

LIPID EXTRACTION AND TARGETED LIPIDOMIC ANALYSIS:

Metabolite profiling was conducted using a rigorously validated LC-MS/MS method [12,13]. Briefly, plasma lipids were extracted by solid-phase extraction. Cartridges were preconditioned with 1 mL methanol and 1 mL Milli-Q water. Samples were spiked with the internal standards mixture (5 ng each) prior to loading. Cartridges were washed with 1 mL 5% methanol, evacuated under high vacuum, dried for 20 min, and eluted with 1 mL methanol. Eluates were dried under nitrogen, reconstituted in 100 μL of 30% acetonitrile, vortexed, and filtered through a 0.22-μm nylon membrane.

Chromatography was performed on a UPLC BEH C18 column (1.7 μm, 100×2.1 mm) at 25°C using water (A) and acetonitrile (B) at 0.6 mL/min with the following gradient: 0 to 1.5 min, 30% to 40% B; 1.5 to 6.5 min, 40% to 60% B; 6.5 to 7.6 min, 60% to 80% B; 7.6 to 8.6 min, hold; 8.6 to 8.8 min, 80% to 30% B; followed by 0.2 min re-equilibration. The same program was utilized for 32 AA-derived and 37 ω-3 PUFA-derived metabolites. Injection volume was 10 μL.

Detection was carried out on a 5500 QTRAP mass spectrometer (AB Sciex, Foster City, CA, USA) in negative multiple reaction monitoring mode (curtain gas=40 psi; ion source gas 1, gas 2=30 psi; ion spray voltage=−4500 V; collision-activated dissociation=medium; source temperature=500°C). Peaks were integrated and quantified using MultiQuant 3.0. Features detected in at least 33% of samples were retained, and missing values were imputed using the group mean.

QUALITY CONTROL:

Technical replicates and pooled QC samples were analyzed every 10 injections. Internal standards were used to correct for variability in extraction efficiency and instrument response. Calibration curves (R² >0.99) were generated for all targeted analytes. Batch effects were assessed by principal component analysis (PCA) and corrected when detected. Analysts were blinded to group labels during peak integration.

ANNOTATION CONVENTION:

Oxylipins were annotated as FattyAcid_Enzyme_CompoundName (e.g., DHA_LOX_8-HDoHE, AA_COX_PGJ2) to indicate precursor, biosynthetic route, and identity.

UNIVARIATE SCREENING:

Group differences were evaluated with 2-sample Wilcoxon rank-sum tests in MetaboAnalyst 6.0 (https://www.metaboanalyst.ca/) using Benjamini-Hochberg false discovery rate (FDR) correction (significant if FDR <0.05).

COMPLEMENTARY FEATURE SELECTION:

Two complementary approaches were applied. In significance analysis of microarrays (SAM), Δ was 1.3 and estimated FDR was 3.6%; metabolites with |d-scores| above 1.3 were retained. SAM controls FDR by permutation-based estimation of significance. In empirical Bayes analysis of microarrays (EBAM), posterior probability was at least 0.90 (local FDR approximately 18%); positive z-scores indicated upregulation, and negative scores indicated downregulation. EBAM uses empirical Bayes shrinkage to improve stability in small-n, large-P datasets. The intersection of features identified by Wilcoxon, SAM, and EBAM was regarded as the high-confidence candidate set.

MULTIVARIATE PATTERN RECOGNITION:

PCA and orthogonal partial least squares-discriminant analysis (OPLS-DA) were performed in MetaboAnalyst 6.0. Seven-fold cross-validation and 200 permutations were used to evaluate model performance.

PATHWAY ENRICHMENT:

Significant oxylipins (FDR <0.05) with corresponding fold changes and p values were uploaded to Ingenuity Pathway Analysis (IPA, Qiagen Inc., Content v73620684). Filters were set to species=human, tissue=cardiovascular system, and confidence=experimentally observed or high predicted. Canonical pathways and “Diseases & Bio Functions” with Fisher’s P<0.05 were reported. Activation z-scores indicated directionality, and network scores greater than 5 were considered robust.

BIOMARKER MODELING:

Receiver operating characteristic (ROC) curves (area under the curve [AUC], optimal cutoff, sensitivity, specificity) were generated in MetaboAnalyst 6.0 and R (pROC package). SVM models were trained on the top 20 ranked metabolites, iteratively refined, and validated on a one-third hold-out set. Features identified by Least Absolute Shrinkage and Selection Operator (LASSO, conducted in R via glmnet package) with greater than 50% selection frequency were retained for the final 6-marker model. Sensitivity, specificity, and overall accuracy were calculated from the confusion matrix of the validation set.

CORRELATIONS WITH CLINICAL VARIABLES:

Spearman’s rank correlations were calculated to assess interactions between oxylipin concentrations and clinical indices, including B-type natriuretic peptide (BNP), New York Heart Association (NYHA) functional class, and LVEF. To control for multiple testing, P-values were adjusted using the Benjamini-Hochberg FDR procedure, with a significance threshold of FDR <0.05.

CROSS-PLATFORM INTEGRATION AND COMPUTATIONAL ENVIRONMENT:

Data processing and statistical analyses were conducted using the combined functionalities of R (v4.0.5) and MetaboAnalyst 6.0. Integration was performed by exporting intermediate data tables between platforms, enabling the use of R packages for advanced modeling and MetaboAnalyst’s curated workflows for exploratory and confirmatory analyses. Multivariate modeling was conducted in SIMCA (v15.0; Umetrics, Sartorius Stedim Data Analytics AB, Umea, Sweden). Additional statistical procedures were carried out in SPSS (v24.0; SPSS Inc., Chicago, IL, USA). All software versions and parameter settings are available upon request to ensure reproducibility.

CONTROL OF POTENTIAL CONFOUNDING FACTORS:

In addition to the stated inclusion and exclusion criteria, participants were instructed to maintain a standardized diet (avoiding fish oil supplements, excessive alcohol consumption, and unusual fat intake) for 72 h prior to blood collection. Lifestyle factors, including smoking and alcohol use, were recorded and considered during interpretation. Medication histories were reviewed to exclude agents with potential effects on lipid metabolism. Environmental conditions during sample handling (temperature, humidity) were monitored to minimize variability.

Results

TARGETED LC-MS/MS QUANTIFICATION AND GLOBAL OXYLIPIN PROFILING:

A validated targeted LC-MS/MS assay (annotating each metabolite with a predefined lipid code, e.g., AA_COX_PGE2, AA_LOX_9-HETE) quantified 73 oxylipins in plasma from 30 patients with DCM and 30 age- and sex-matched healthy controls. Baseline demographic and clinical features are presented in Table 1. A heatmap of normalized intensities (Figure 1) demonstrated broad shifts in oxylipin abundance in DCM, with substantial increases in AA-derived LOX products (e.g., 5-hydroxyeicosatetraenoic acid [HETE], 9-HETE, and 12-HETE) and COX products (e.g., prostaglandin E2 [PGE2] and PGF2α).

MULTIVARIATE SEPARATION OF STUDY GROUPS:

PCA (Figure 2A, left) of all 73 quantified metabolites revealed clear separation between DCM and control samples, primarily along principal component 1 (PC1; 58.5% variance explained); PC2 and PC3 explained an additional 8.6% and 3.6%, respectively. OPLS-DA (Figure 2A, right) further improved group discrimination (predictive T score 6.6%, orthogonal T score 55.5%), consistent with a distinct plasma oxylipin signature in DCM.

UNIVARIATE SCREENING AND PATHWAY ENRICHMENT:

Two-sample Wilcoxon rank-sum tests with FDR <0.05 (as prespecified in the statistical plan) identified 16 differentially abundant oxylipins (Table 2). The largest case-control differences were observed for AA_LOX_12-HETE, AA_LOX_5-HETE, AA_LOX_9-HETE, AA_COX_PGE2, and AA_LOX_11-HETE (Figure 2B). These metabolites, along with associated fold changes and P-values, were submitted to IPA using human/cardiovascular system/high-confidence content filters. Enriched canonical pathways (Figure 2C) included eicosanoid signaling, prostanoid biosynthesis, triggering receptor expressed on myeloid cells 1 (TREM1) signaling, and macrophage migration inhibitory factor-mediated glucocorticoid regulation (all -log10 P>1.5). Eicosanoid and TREM1 pathways were predicted to be activated (positive z-scores), whereas T helper cell 17 (Th17)/interleukin (IL)-17A-related pathways were predicted to be suppressed (negative z-scores).

Top “Diseases and Biological Functions” categories (Figure 2D) encompassed Infectious Diseases/Inflammatory Disease/Organismal Injury (-log10 P=16.8), Reproductive System Disease/Cell-to-Cell Signaling/Inflammatory Response (-log10 P=8.4), and Cellular Movement/Hematological System Development/Hypersensitivity Response/Immune Cell Trafficking (-log10 P=4.2). Associated network analysis identified 3 high-scoring interaction modules. Network 1 (Score 43) comprised Infectious Diseases, Inflammatory Disease, and Organismal Injury and Abnormalities. Network 2 (Score 24) consisted of Cell-to-Cell Signaling and Interaction, Cell Signaling, and Molecular Transport. Network 3 (Score 6) constituted Cellular Development, Growth and Proliferation, and Connective Tissue Development and Function.

CORRELATIONS WITH CLINICAL SEVERITY:

Spearman’s rank correlation was used to evaluate relationships between significantly altered oxylipins and clinical severity indicators (BNP, NYHA class, and LVEF). Among the 6 most altered oxylipins, only PGE2 demonstrated significant positive correlations with BNP (r=0.413, P=0.01) and NYHA class (r=0.401, P=0.02) (Figure 3). No other metabolites exhibited significant associations with these clinical parameters after FDR correction, highlighting the specificity of PGE2 as a potential clinical correlate in this cohort.

SAM- AND EBAM-DERIVED SIGNATURES:

To identify robust metabolite patterns beyond univariate analysis, SAM (Δ=1.3, estimated FDR=3.6%) and EBAM (posterior probability ≥0.90, local FDR=18%) were performed as described in the Methods. SAM identified 24 significant oxylipins (Figure 4A), with top scores for AA_COX_PGE2 (d=3.51, q=0.0012), EPA_P450_18-HEPE (d=3.47, q=0.0012), AA_LOX_9-HETE (d=3.47, q=0.0012), and AA_P450_16-HETE (d=3.32, q=0.0012).

EBAM identified 62 features (Figure 4B), including strong upregulation of AA_LOX_12-HETE (z=4.66, local FDR=1.0×10−6), AA_LOX_5-HETE (z=4.20, local FDR=7.8×10−6), and AA_LOX_9-HETE (z=3.93, local FDR=2.5×10−5), as well as downregulation of specialized pro-resolving mediators such as DHA_LOX_RvD2 (z=−2.10, local FDR=0.016).

Fourteen oxylipins were common to the Wilcoxon, SAM, and EBAM outputs (Figure 4C), corresponding to the subset listed in Table 2 after exclusion of AA_LOX_15-HETE and AA_LOX_11-HETE from the original 16. Pathway analysis of this intersected set identified glycolysis/gluconeogenesis as the top enriched metabolic pathway (P<0.05; Figure 4D).

DIAGNOSTIC MODEL DEVELOPMENT AND PERFORMANCE:

To evaluate diagnostic potential, we performed ROC analyses. Univariate ROC curves identified 6 metabolites with an AUC greater than 0.75; 5 were among the 14-metabolite core set (Figure 5). In multivariate modeling, SVM classification using the top 20 ranked features achieved a cross-validated AUC of 0.875 (Figure 6A, 6B).

Feature reduction with LASSO retained 6 metabolites – AA_LOX_15-oxo-ETE, AA_LOX_9-HETE, AA_LOX_6R-LXA4, AA_COX_PGE2, AA_P450_16-HETE, and EPA_P450_18-HEPE – for the final model. This 6-marker SVM panel yielded an AUC of 0.876, sensitivity of 74.2%, and specificity of 75.9% in the hold-out validation set (Figure 6C, 6D). For clinical context, an AUC of 0.876 indicates an 87.6% probability that the model will correctly predict a randomly chosen patient with DCM to have a higher probability of disease than a randomly chosen control, representing good discriminative ability by accepted clinical biomarker standards.

Discussion

SUMMARY OF KEY FINDINGS AND NOVELTY:

This study constructed, to our knowledge, the first comprehensive plasma oxylipin atlas in DCM, integrating targeted ultra-high performance LC-MS/MS quantification of 73 oxidized lipids with univariate (Wilcoxon rank-sum), SAM, and EBAM analyses. By intersecting the outputs of these complementary approaches, we identified a robust 14-marker signature – predominantly LOX-derived HETEs (12-, 5-, 9-, and 11-HETE), COX-derived PGE2, and P450-derived hydroxyeicosapentaenoic acids (HEPEs). We propose a novel SVM-based 6-oxylipin panel (AA_LOX_15-oxo-ETE, AA_LOX_9-HETE, AA_LOX_6R-LXA4, AA_COX_PGE2, AA_P450_16-HETE, EPA_P450_18-HEPE) that achieved an AUC of 0.876, with sensitivity of 74.2% and specificity of 75.9% – comparable to or exceeding the performance of BNP/NT-proBNP (AUC=0.80–0.85 [14]) and approaching that of advanced imaging modalities [15,16]. This combination of broad lipidomic profiling with machine learning-driven biomarker selection represents a translational advance in cardiovascular research.

INTEGRATION WITH PREVIOUS RESEARCH:

Our results extend earlier observations of elevated HETEs and PGE2 in failing myocardium, which have been linked to NF-κB and mitogen-activated protein kinase (MAPK) activation and downstream pro-inflammatory remodeling [17–19]. Whereas prior studies in heart failure largely emphasized LOX- and COX-derived products, the present analysis also highlights elevated levels of EPA_P450_18-HEPE and AA_P450_16-HETE, revealing a previously underappreciated role for P450-derived oxylipins in DCM-related vascular dysfunction and leukocyte trafficking [20,21]. The observed downregulation of resolvin D2 (RvD2) aligns with the “eicosanoid storm” characteristic of non-resolving inflammation [22,23].

Compared with earlier cardiovascular oxylipin studies, our dataset provides broader metabolite coverage, integrative statistical validation, and a direct link to diagnostic modeling [24].

MECHANISTIC IMPLICATIONS:

The predominance of LOX-derived HETEs in our signature supports a mechanistic role for 12/15-LOX upregulation in amplifying oxidative stress and inflammatory cascades, thereby promoting cardiomyocyte apoptosis and maladaptive remodeling [25]. Experimental models have shown that overactivation of 12/15-LOX triggers NF-κB and MAPK signaling, which increases cytokine release, myofibroblast differentiation, and excessive extracellular matrix deposition – changes reflected in our DCM cohort [25]. Pharmacologic or genetic suppression of 12/15-LOX reduces HETE production, limits fibroblast activation, and shifts immune responses toward resolution [26,27].

In parallel, PGE2-prostaglandin E2 receptor subtype 4 (EP4) signaling facilitates fibroblast-cardiomyocyte crosstalk. Under stress, EP4 activation in cardiomyocytes promotes transforming growth factor-β1 release, driving fibroblast activation and extracellular matrix deposition [28]. EP4 deletion mitigates fibrosis and preserves ventricular compliance, underscoring its therapeutic relevance [29].

Deficiencies in specialized pro-resolving mediators, particularly RvD2, suggest impaired efferocytosis, prolonged immune activation, and sustained fibrotic drive [30–33]. RvD2 enhances apoptotic cell clearance, supports non-inflammatory macrophage proliferation, and promotes a reparative phenotype; its depletion likely prolongs inflammation and tissue injury [33,34].

Alterations in EPA- and DHA-derived HEPEs and hydroxydocosahexaenoic acids (HDoHEs) reflect disruption of lipid networks regulating vascular tone, endothelial reactivity, and leukocyte chemotaxis [9,35]. The imbalance between omega-6 (AA-derived) and omega-3 (EPA/DHA-derived) substrates may skew LOX/P450 activity toward pro-inflammatory output [35,36]. Moreover, polyunsaturated fatty acid precursors themselves – AA, EPA, and DHA – exert dual immunomodulatory effects. AA-derived eicosanoids (e.g., leukotrienes and prostaglandins) generally promote inflammation and fibrosis, whereas EPA- and DHA-derived metabolites are associated with pro-resolving functions, including the generation of specialized mediators such as resolvins and protectins [36,37].

Finally, enrichment of differential oxylipins in glycolysis/gluconeogenesis pathways suggests that oxylipin signaling is intertwined with metabolic reprogramming in DCM [38,39]. Several oxylipins, including PGE2 and selected HETEs, regulate glycolytic enzymes such as phosphofructokinase and pyruvate kinase M2, thereby altering energy substrate utilization and cellular redox balance. This crosstalk between lipid mediators and central energy pathways may exacerbate maladaptive remodeling and could represent a therapeutic nexus for intervention.

CLINICAL RELEVANCE AND DIAGNOSTIC POTENTIAL:

The 6-oxylipin SVM panel demonstrated strong discrimination (AUC=0.876), indicating an 87.6% probability that a randomly selected patient with DCM will be correctly ranked above a control. This performance compares favorably with BNP, troponin, and echocardiography in similar contexts. However, only PGE2 was significantly correlated with BNP levels and NYHA class, suggesting that most oxylipins in the panel capture pathophysiologic variation beyond conventional severity metrics. This observation likely reflects the complex and heterogeneous lipidomic responses in DCM; it should guide future validation studies.

The multiplexed ultra-high performance LC-MS/MS platform is increasingly available in clinical laboratories, supporting potential implementation of oxylipin panels for early detection, longitudinal monitoring, and therapy personalization. Such panels may inform individualized interventions, including selective LOX/COX inhibition or supplementation with specialized pro-resolving mediators, tailored to patient-specific lipid signatures.

THERAPEUTIC AND TRANSLATIONAL OUTLOOK:

Our findings support further exploration of LOX, COX, and P450 modulation: 12/15-LOX inhibitors attenuate remodeling [40], COX-2-biased agents reduce fibrosis [41], and synthetic RvD2 analogs improve outcomes after myocardial infarction [42]. The cyclic guanosine monophosphate-protein kinase G (cGMP-PKG) signaling axis may also be co-targeted for synergistic benefit [43].

LIMITATIONS:

The cross-sectional, single-center design restricts causal inference and generalizability. The SVM model was trained and tested on the same dataset, raising the risk of overfitting; external validation is therefore essential. Despite FDR control, residual confounding from diet, medications, and comorbidities may persist. The 73-oxylipin panel excluded other bioactive lipids detectable by untargeted approaches. Mechanistic and interventional validation in both animal models and randomized trials is required to confirm causality and therapeutic potential. Finally, the limited clinical correlation observed (significant only for PGE2) should be considered when interpreting the translational applicability of the biomarker panel.

Conclusions

This study profiled oxylipin metabolites derived from essential polyunsaturated fatty acids in DCM, revealing a shift toward pro-inflammatory mediators (HETEs, PGE2) and depletion of pro-resolving mediators (resolvins). Integrative modeling identified an SVM-based 6-oxylipin panel with diagnostic performance comparable to that of established biomarkers and highlighted mechanistic pathways as potential therapeutic targets.

These findings underscore the clinical relevance of lipid signaling in myocardial remodeling. In addition to diagnostic applications, the panel may guide personalized interventions, such as selective LOX/COX inhibition or specialized pro-resolving mediator supplementation, tailored to individual lipidomic profiles. Future studies should include validation in independent cohorts and interventional trials to determine whether modulation of lipid mediator pathways improves clinical outcomes.