Immune subtypes and diagnostic genes revealed by neutrophil trap-associated transcriptomic signatures in ischemic stroke

Immune subtypes and diagnostic genes revealed by neutrophil trap-associated transcriptomic signatures in ischemic stroke | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Immune subtypes and diagnostic genes revealed by neutrophil trap-associated transcriptomic signatures in ischemic stroke Xiaoqian Lan, Qing Li, Yuting Shi, Guangli Feng, Yingrui Hu, Shiyi Qin, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7515929/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Objective To investigate the involvement of neutrophil extracellular trap (NET)-associated genes in ischemic stroke (IS) and create a diagnostic model for precision stroke treatment. Methods Transcriptomic datasets from GEO were integrated to identify NETs-related gene signatures in ischemic stroke. We conducted differential expression analysis and Weighted Gene Co-expression Network Analysis (WGCNA), followed by biomarker identification using support vector machine-recursive feature elimination (SVM-RFE), least absolute shrinkage and selection operator (LASSO), and Random Forest (RF) algorithms. A multigene diagnostic model was developed and validated. Immune subtypes were defined via consensus clustering based on hub genes. Immune infiltration, functional enrichment, drug-gene interaction analysis, and molecular docking were performed to identify therapeutic targets specific to subtypes. Results Six hub genes ( KCNJ15 , ARG1 , CLEC4E , ABCA1 , ANXA1 , and MMP9 ) were recognized as promising biomarkers for diagnosis with excellent performance (AUC = 0.990). Two immune subtypes of IS were revealed, characterized by distinct metabolic activity, immune cell infiltration, and proinflammatory signaling. Functional analysis confirmed significant immunometabolic divergence between the subtypes. Rosuvastatin was identified as a potential therapeutic agent targeting ABCA1, MMP9, and KCNJ15, suggesting subtype-specific therapeutic effects. Conclusion Our study provides a novel framework for immune subtyping and targeted therapy in IS, demonstrating the diagnostic and therapeutic potential of NETs-related biomarkers. These findings offer promising implications for precision stroke management. ischemic stroke neutrophil extracellular traps immune subtypes machine learning biomarker discovery drug repositioning molecular stratification Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 1 Introduction Ischemic stroke (IS) continues to be a major contributor to long-term disability and death globally(1, 2). The global incidence of ischemic stroke (IS) has been steadily increasing, driven by an aging population and a growing prevalence of vascular risk factors(3). IS is primarily triggered by the sudden blockage of cerebral arteries, which sets off a series of pathological events such as neuronal damage, oxidative damage, immune system disruptions, and impairment of vascular integrity(4, 5). Despite advancements in reperfusion therapies, such as mechanical thrombectomy and intravenous thrombolysis, the clinical benefits of these treatments are limited by factors such as narrow therapeutic windows, strict patient eligibility, and risks of hemorrhagic transformation or reperfusion injury(6, 7). Therefore, discovering new molecular biomarkers and potential therapeutic targets is crucial for is essential for enabling precision subtyping and individualized treatment strategies in IS. Over recent years, inflammation and immune activation have emerged as critical contributors to stroke pathogenesis(8). Neutrophils, as key effector cells of the innate immune system, have recently garnered increasing attention for their ability to release neutrophil extracellular traps (NETs)(9). NETs are extracellular networks primarily made of decondensed chromatin, including DNA and histones, and contain granular enzymes released by neutrophils(10). Originally identified as antimicrobial agents, NETs are now recognized for their proinflammatory, prothrombotic, and tissue-damaging effects in non-infectious settings(11, 12). Following cerebral ischemia, neutrophils are rapidly recruited to the ischemic region and release NETs both in the brain and peripheral circulation(13). These NETs disrupt the blood-brain barrier (BBB), exacerbate cerebral edema and neuronal necrosis(14), promote thrombus formation(15, 16), and activate the coagulation cascade(17), thereby amplifying neurovascular injury and delaying tissue repair(18). High levels of NETs have been linked to greater stroke severity and worse clinical outcomes, highlighting their potential as both prognostic biomarkers and targets for therapeutic intervention(19, 20). Although early research has emphasized the role of specific NETs-related molecules in ischemic stroke (IS), a comprehensive understanding of their broader regulatory context remains lacking. For instance, platelet-derived high mobility group box 1 has been demonstrated to trigger NETs release, thereby exacerbating ischemic damage(21). Similarly, brain border-associated CXCL2⁺ neutrophils have been shown to mediate reperfusion injury through NETs formation mechanisms(22). Several circulating NETs biomarkers, such as citrullinated histone H3 and myeloperoxidase-DNA complexes, have been found to be markedly elevated in IS patients and show positive correlations with neurological impairment scores(23, 24). Therapeutic interventions aimed at inhibiting NETs, including the neonatal NET-inhibitory factor, have shown promise in preclinical models by reducing infarct volume and improving neurological outcomes(25). Moreover, with the advancement of high-throughput sequencing technologies, emerging studies have identified NETs-associated genes (including CEACAM3 , TNF , and SELP ) enriched in neutrophils from IS animal models. These genes are implicated in leukocyte adhesion, ferroptosis, and IL-17 signaling pathways, and exhibit potential diagnostic relevance(26). Nonetheless, most existing studies have focused on individual molecules or discrete pathways, with limited efforts to systematically map the molecular network that underpins NETs biology in IS. In addition, NETs-based immune subtyping remains an underexplored area, significantly constraining its translational value for patient stratification and targeted intervention. To address these limitations, the present study integrates transcriptomic analysis from multiple public datasets, leveraging differential gene expression analysis, weighted gene co-expression network analysis (WGCNA), and three complementary machine learning algorithms (support vector machine-recursive feature elimination (SVM-RFE), least absolute shrinkage and selection operator (LASSO), and Random Forest (RF)) to identify NETs-associated hub genes. Subsequently, we developed a reliable diagnostic model utilizing these genes. Additionally, functional enrichment and drug-gene interaction mapping are performed to explore the therapeutic implications of these key hub genes, while molecular docking is employed to predict drug interactions. This comprehensive analytical framework provides a novel approach for immune subtyping and targeted therapy in IS, demonstrating the clinical applicability and potential of NETs-related biomarkers in precision stroke management. 2 Materials and Methods Data Acquisition and Preparation Gene expression data were sourced from the Gene Expression Omnibus (GEO) database ( https://www.ncbi.nlm.nih.gov/geo/ ). Two datasets (GSE16561(27) (platform: GPL6883) and GSE58294(28) (platform: GPL570)) were selected as the training cohort, including 108 IS samples and 47 healthy controls. The external validation cohort comprised GSE195442(29) (platform: GPL31275) and GSE66724(30) (platform: GPL570), resulting in a combined dataset of 18 IS and 18 control samples (Supplementary Table 1). All data processing was conducted in R (v4.4.1). Repeated probes were merged using the avereps() function in the limma package, followed by log2 transformation and normalization via the normalizeBetweenArrays() function. To adjust for batch effects across datasets, the ComBat function from the sva package was applied. Principal component analysis (PCA) was then used to assess and visualize variations before and after the correction, utilizing the ggplot2 and ggpubr packages. To identify NETs-related genes, 770 protein-coding genes with a relevance score > 8 were retrieved from the GeneCards Human Gene Database ( https://www.genecards.org/ ). An additional 241 genes were integrated from published literature (Supplementary Table 2)(31–33). After deduplication, a total of 855 unique NETs-related genes were included for downstream analysis. The overall analytical workflow is illustrated in Fig. 1 . Differential Expression Analysis After constructing the design matrix, differential gene expression was analyzed using the limma package. DEGs were identified using the criteria |log2 fold change (FC)| >0.5 and a false discovery rate (FDR) < 0.05. The 30 most upregulated and downregulated DEGs were visualized using the pheatmap package, while overall gene distribution was presented via volcano plots using ggplot2. Functional Enrichment Analysis Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the ClusterProfiler package to explore the biological functions of DEGs. The GO terms analyzed included biological process (BP), cellular component (CC), and molecular function (MF). Terms with P < 0.05 and adjusted P -values ( p .adjust) < 0.05 were considered significant. The results were visualized through bubble and bar plots. Weighted Gene Co-expression Network Analysis (WGCNA) A co-expression network was generated from the training datasets using the WGCNA package (v1.73). Genes with a standard deviation ≤ 0.5 were excluded. Sample clustering was performed to detect outliers, using a cut-off height of 20,000. The optimal soft-thresholding power (β) was selected based on the pickSoftThreshold function, ensuring an R² value > 0.9. A topological overlap matrix (TOM) was calculated and converted into a dissimilarity matrix (1 − TOM). Modules were detected using dynamic tree cutting with a minimum module size of 50, and modules with an eigengene correlation distance < 0.3 were merged. Finally, the association between module eigengenes and sample traits was analyzed to identify key modules. Overlapping Genes Identification and Integration To identify overlapping genes, DEGs, NETs-related genes, and genes from significant WGCNA modules were intersected. Venn diagrams were generated using the ggvenn package. The expression levels of the overlapping genes were extracted, and intergroup differences were assessed using the Wilcoxon rank-sum test (FDR < 0.05). Boxplots were created using ggpubr. Spearman correlation coefficients were calculated among overlapping genes within the IS group using the corrplot package, and hierarchical clustering was used to identify gene clusters. Genomic locations of overlapping genes were mapped using the hg38 reference genome, and Circos plots (circlize) along with Manhattan plots were constructed to visualize chromosomal distribution and gene-level significance. Immune Cell Infiltration Profiling Single-sample gene set enrichment analysis (ssGSEA) was conducted using the gene set variation analysis (GSVA) package to quantify immune cell infiltration based on 28 immune-related gene sets from the immune.gmt file. Enrichment scores were normalized (min-max) across samples. Heatmaps of immune infiltration were plotted using pheatmap and clustered by group (IS vs. control). Intergroup differences were evaluated using the Wilcoxon test. Spearman correlations between overlapping genes expression and immune cell scores were computed and visualized as correlation heatmaps using ggplot2, with a blue-white-red gradient indicating correlation strength. Machine Learning-Based Hub Gene Selection To identify reliable diagnostic biomarkers from the overlapping gene set, three distinct machine learning algorithms were applied: LASSO: Implemented via the glmnet package with 10-fold cross-validation to select the optimal penalty parameter (lambda). Non-zero coefficient genes at minimum lambda were retained. SVM-RFE: Recursive feature elimination was performed using the e1071 package and a custom script (NETs21.msvmRFE.R), selecting the gene subset with the lowest cross-validation error. RF: The caret package was used to upsample minority classes and build the RF model. Feature importance was ranked by MeanDecreaseGini, and genes with importance > 5 were selected. Venn diagrams were employed to determine the common genes identified by all three methods, which were then retained as final robust hub genes. Development and Validation of the Diagnostic Model A multivariate logistic regression model was constructed using glmnet based on the selected hub genes. In the training set, area under the curve (AUC) and receiver operating characteristic (ROC) curves values were calculated using the pROC package to evaluate discriminatory ability. Gene expression levels were visualized using boxplots (ggpubr and reshape2), and differences were tested with Wilcoxon rank-sum tests. To further assess model performance, calibration curves and nomograms (rms package) were used, while clinical utility was evaluated through decision curve analysis (DCA) with the rmda package. The same procedure was repeated for external validation. Molecular Subtyping and Functional Profiling Based on the expression levels of hub genes identified by machine learning in IS samples, molecular subtyping was performed using ConsensusClusterPlus (k-means algorithm, Euclidean distance, max k = 9, 50 resampling iterations). The ideal number of clusters was selected using consensus cumulative distribution function (CDF) and consensus heatmaps. Subtype-specific expression patterns were visualized with heatmaps and PCA. Boxplots were used to compare gene expression among subtypes. Immune cell infiltration patterns among subtypes were evaluated using ssGSEA scores. Functional enrichment of KEGG pathways across subtypes was assessed using GSVA, and pathways with P < 0.05 were visualized in bar plots (top 10 upregulated and downregulated). Drug Enrichment and Regulatory Network Construction Potential drug-gene interactions were explored using DGIdb ( https://www.dgidb.org/ ) and DSigDB ( https://maayanlab.cloud/DSigDB/ ). Drug enrichment was conducted using the enricher function in ClusterProfiler ( P < 0.05 and FDR < 0.05). Up to 10 drugs per gene were selected, and drug-gene interaction networks were built using Cytoscape software, with visualization facilitated by ggplot2 and enrichplot. Molecular Docking Analysis To further elucidate ligand-target interactions and identify candidate therapeutic agents, structure-based molecular docking was performed. Rosuvastatin, the top-ranked drug from enrichment analysis, was selected for docking with its key targets: MMP9, ABCA1, and KCNJ15. The 3D structure of Rosuvastatin was retrieved from the PubChem database. Protein structures for MMP9 (PDB ID: 1ITV) and ABCA1 (PDB ID: 5XJY) were obtained from the Protein Data Bank ( https://www.rcsb.org/ ), while KCNJ15 structure was predicted using AlphaFold ( https://alphafold.com/entry/Q99712 ). Docking was performed using the CB-Dock platform ( https://cadd.labshare.cn/cb-dock2/php/index.php ), which identifies potential binding cavities and executes blind docking via AutoDock Vina. Binding affinity scores (in kcal/mol) were used to evaluate binding strength, with thresholds of ≤ -5.0 kcal/mol indicating strong interaction. Binding poses and energy rankings were visualized and used to select optimal compound–target pairs. 3 Results Identification of Differentially Expressed Genes and Batch Effect Correction To reduce technical variation and improve cross-cohort comparability, PCA was performed on the merged expression matrix from GSE16561 and GSE58294. Before correction, samples from the two datasets showed clear separation along the PC1 and PC2 axes, indicating significant batch effects (Fig. 2 A). Following batch correction and normalization using the sva package, sample distributions became more homogeneous in PCA space, demonstrating effective mitigation of batch-related variability (Fig. 2 B). Subsequently, differential expression analysis was performed between ischemic stroke (IS) and control samples, revealing 368 differentially expressed genes (DEGs), including 212 upregulated and 156 downregulated genes (Fig. 2 C). Heatmap visualization further highlighted distinct expression patterns between the two groups (Fig. 2 D). Functional Enrichment of Differentially Expressed Genes To investigate the biological significance of the DEGs, GO and KEGG pathway enrichment analyses were performed. GO results revealed that the DEGs were significantly enriched in immune-related biological processes, including activation of immune responses, regulation of cytokine production, T-cell receptor signaling, and acute inflammatory responses (Fig. 3 A). At the cellular component level, DEGs were primarily associated with secretory granule membranes, membrane microdomains, and the external side of the plasma membrane. Molecular function enrichment pointed to roles in immune receptor activity, immunoglobulin binding, and pattern recognition receptor signaling. KEGG pathway analysis further revealed prominent enrichment in immune and infection-associated pathways such as hematopoietic cell lineage, NF-κB signaling, T and B cell receptor signaling, and cytokine–cytokine receptor interaction (Fig. 3 B). Additionally, metabolic and pathogen-related pathways, including Staphylococcus aureus infection, Tuberculosis, and Pantothenate and CoA biosynthesis, were significantly enriched, suggesting that these DEGs may regulate immune activation and inflammation in the pathogenesis of IS. Identification of Key Co-expression Modules via WGCNA WGCNA was performed to identify gene modules linked to IS. A soft-thresholding power of β = 6 was chosen based on the scale-free topology criterion (R² >0.9) and average connectivity (Fig. 4 A). Several gene modules were identified via hierarchical clustering and color-coded (Fig. 4 B). Correlation analysis revealed a strong negative association between the blue module and IS status (r = -0.61, P = 5e − 17 ), while the yellow module showed a positive correlation with IS (r = 0.65, P = 4e − 20 ); green and red modules also showed statistically significant associations (Fig. 4 C). Sample clustering heatmaps confirmed distinct expression patterns between IS and control samples across these modules, underscoring their biological relevance (Fig. 4 D). Identification and Characterization of NETs-Associated Overlapping Genes To identify overlapping genes implicated in both IS and NETs biology, we intersected DEGs, genes from the significant WGCNA module (yellow), and a curated NETs-related gene set. Twenty-one candidate hub genes were identified at the intersection (Fig. 5 A). All were significantly differentially expressed between IS and control samples (Wilcoxon test, *** P < 0.001), with the majority being upregulated in the IS group (Fig. 5 B). Pearson correlation analysis demonstrated strong co-expression among these genes (Fig. 5 C), suggesting they may operate within coordinated regulatory networks. Chromosomal mapping revealed non-random genomic distribution. Some genes were clustered on specific chromosomes. For example, FCGR1A and F5 on chromosome 1; DYSF and IL18R1 on chromosome 2; ARG1 and VNN3P on chromosome 6; ABCA1 and ANXA1 on chromosome 9; SIGLEC5 , PGLYRP1 , and CD177 on chromosome 19; and MMP9 and KCNJ15 on chromosome 21 (Fig. 5 D). Manhattan plot analysis confirmed significant genomic enrichment for these genes, with SLC22A4 , KCNJ15 , F5 , ARG1 , and ABCA1 among the most prominent (Fig. 5 E), indicating potential diagnostic or mechanistic relevance in IS. Immune Infiltration Profiling in Ischemic Stroke To assess immune alterations in IS, ssGSEA was employed to estimate the infiltration of 28 immune cell types. The immune heatmap illustrated clear differences in immune cell abundance between IS and control groups (Fig. 6 A). IS samples exhibited elevated levels of activated dendritic cells (DCs), macrophages, neutrophils, and Th17 cells. Violin plots confirmed that these differences were statistically significant (Fig. 6 B, P < 0.001). Spearman correlation analysis between overlapping genes and immune cells revealed that ABCA1 and CLEC4E were positively correlated with neutrophils and plasmacytoid DCs; ARG1 was significantly associated with neutrophils, macrophages, and activated DCs; ANXA1 correlated positively with mast cells ( P < 0.01) but negatively with CD56 bright NK cells and Th17 cells ( P < 0.05) (Fig. 6 C). These findings suggest that the identified genes may play regulatory roles in the immune landscape of IS. Machine Learning Identifies Robust Hub Genes Three complementary machine learning algorithms (SVM-RFE, LASSO, and RF) were applied to identify hub genes. LASSO regression selected 16 genes (Figs. 7 A, B), while SVM-RFE achieved optimal accuracy (95.4%) with 13 genes (Fig. 7 C), corresponding to the lowest cross-validation error (Fig. 7 D). The RF model stabilized at 100 trees and identified 8 important genes (importance score > 5), with ANXA1 , ABCA1 , SLC22A4 , and ARG1 scoring highest (Figs. 7 E, F). Robust diagnostic biomarkers identified through machine learning are shown in Supplementary Table 3. A Venn diagram highlighted six hub genes ( KCNJ15 , ARG1 , ABCA1 , CLEC4E , ANXA1 , and MMP9 ) shared across all three methods (Fig. 7 G), all of which were significantly upregulated in IS samples (Fig. 7 H). Construction and Validation of a Multi-Gene Diagnostic Model ROC curve analysis revealed high discriminatory power for each of the six hub genes, with AUCs ranging from 0.835 ( MMP9 , ANXA1 ) to 0.914 ( ARG1 ) (Fig. 8 A). A logistic regression model incorporating all six genes yielded an AUC of 0.990 (95% Confidence Interval: 0.977–0.998) in the training set (Fig. 8 B). Nomogram visualization illustrated the relative contribution of each gene (Fig. 8 C), with calibration and decision curve analyses confirming the model's accuracy and clinical benefit (Figs. 8 D, E). In the external validation set, none of the six genes showed significant differential expression individually, and their AUCs were all below 0.70 (Supplementary Fig. 1). However, the combined model retained moderate diagnostic power, with an AUC of 0.713 (95% Confidence Interval: 0.522–0.873) (Fig. 8 F). Molecular Subtype Identification and Immune Landscape Characterization Consensus clustering based on the expression profiles of the six hub genes identified two stable molecular subtypes (Fig. 9 A), with optimal cluster number determined as k = 2 (Figs. 9 B, C and Supplementary Table 4). Five genes ( KCNJ15 , ARG1 , ABCA1 , CLEC4E , MMP9 ) showed significantly higher expression in subtype C2 compared to C1 (Fig. 9 D). Heatmap analysis further validated these differences (Fig. 9 E). Immune infiltration profiling revealed that subtype C2 exhibited higher abundance of activated DCs, macrophages, monocytes, NK cells, and plasmacytoid DCs, whereas subtype C1 showed enrichment of multiple B and T cell subpopulations, including memory B cells and effector CD4 + T cells (Fig. 9 F). PCA confirmed clear transcriptomic separation between C1 and C2 (Fig. 9 G). GSVA-based pathway enrichment revealed functional divergence between subtypes. Subtype C2 was enriched in inflammatory and signaling pathways, such as MAPK signaling, Fc gamma R-mediated phagocytosis, Toll-like receptor signaling, and complement cascades. In contrast, subtype C1 was enriched in metabolic and mitochondrial pathways including oxidative phosphorylation, citrate cycle, and steroid biosynthesis (Fig. 9 H). Drug Enrichment and Molecular Docking Drug enrichment analysis using the DSigDB database identified several candidates, including Rosuvastatin, L-proline, and Diosgenin (Fig. 10 A and Supplementary Table 5). Rosuvastatin was selected for further validation. Molecular docking was conducted between Rosuvastatin and three target proteins: ABCA1 (Fig. 10 B), MMP9, and KCNJ15 (Supplementary Fig. 2). The binding affinities (vina scores) were − 8.4, -6.6, and − 6.8 kcal/mol, respectively, with the highest affinity observed for ABCA1 . Structural analysis revealed that Rosuvastatin binds stably within the ABCA1 active pocket, interacting with residues F819, E792, Y793, K1524, and N1523 via hydrogen bonding and hydrophobic contacts. π–π stacking and polar interactions further stabilized the complex (Fig. 10 B). A compound-gene interaction network constructed in Cytoscape illustrated potential multitarget effects of Rosuvastatin and other agents (Fig. 10 C). 4 Discussion IS is a highly heterogeneous neurovascular disorder involving complex pathophysiological processes such as inflammation, immune activation, apoptosis, and vascular remodeling(34). Recent studies have highlighted NETs as crucial factors in exacerbating post-stroke inflammation and facilitating secondary thrombosis, significantly influencing the extent of neuronal damage and clinical outcomes(35). However, existing studies on NETs in IS have primarily focused on single-gene functional validation, lacking a comprehensive understanding of their molecular networks and immunophenotypic characteristics. In this study, by integrating multiple GEO datasets and applying differential gene expression analysis, WGCNA, and three complementary machine learning algorithms, we identified six NETs-related hub genes: KCNJ15 , ARG1 , ABCA1 , CLEC4E , ANXA1 , and MMP9 . These genes were significantly upregulated in IS patients and exhibited strong associations with the immune microenvironment, suggesting pivotal regulatory roles in inflammation-coagulation interplay during stroke progression. A diagnostic model constructed from these six genes showed excellent discriminative performance in the training cohort (AUC = 0.990). Although none of the individual genes reached statistical significance in the independent validation cohort (n = 36, P > 0.05, AUC < 0.7), the multi-gene model still demonstrated moderate discriminatory power (AUC = 0.713), indicating favorable robustness and potential for clinical translation. It is worth noting that the limited sample size of the validation cohort may have reduced the statistical power for assessing individual biomarkers. Larger clinical cohorts are thus warranted to further evaluate the diagnostic utility and generalizability of the identified gene signature. These findings also highlight that for highly immune-heterogeneous diseases like IS, multi-gene models may offer greater clinical utility than single biomarkers. Functionally, the six hub genes contribute to NETs formation, inflammation regulation, or neurovascular protection via diverse mechanisms. ARG1, an arginase enzyme, modulates immune responses during the acute phase of stroke by competitively inhibiting iNOS-mediated NO synthesis, whereas its co-expression with M2 macrophage markers and IL-10 in the recovery phase suggests dual immunomodulatory roles(36). Although elevated ARG1 levels correlate with stroke severity and immune suppression(37), direct evidence for its involvement in NETs formation remains lacking. ANXA1, a known pro-resolving mediator(38), has been shown to suppress NETs release and protect the BBB via the FPR2 pathway during post-stroke resolution(39). ABCA1 facilitates cholesterol efflux and maintains BBB integrity(40), and its deficiency exacerbates white matter injury and increases IS risk(41). KCNJ15, a potassium channel protein(42), has been proposed as a biomarker for atherosclerotic cerebral infarction(43). It may contribute indirectly to neutrophil activation by modulating membrane potential; however, its direct involvement in the formation of NETs remains to be experimentally confirmed. CLEC4E, a pattern recognition receptor, is functionally linked to innate immunity and NETs release(44). MMP9 is a critical protease that facilitates neutrophil migration and extracellular matrix degradation, contributing to BBB disruption and hemorrhagic transformation in stroke. It may act synergistically with NETs to amplify inflammation and thrombosis(45). These results enhance our understanding of the inflammation-coagulation axis in IS and underscore the potential of NETs-related molecular markers. Using the expression patterns of the six hub genes, we conducted NETs-related immunophenotypic subtyping for the first time in IS. Two molecular subtypes were identified: C1 and C2. Subtype C2 exhibited higher expression of KCNJ15, ARG1, ABCA1, CLEC4E, and MMP9, with enrichment in pathways such as MAPK signaling, complement cascade, Toll-like receptor signaling and coagulation pathways. This subtype also showed prominent infiltration of innate immune cells (neutrophils, macrophages, monocytes), reflecting a “NETs-high, immune-activated” phenotype, which is closely linked to heightened inflammation and thrombotic risk after stroke(46, 47). In contrast, subtype C1 was characterized by enrichment of metabolic pathways (e.g., oxidative phosphorylation, lipid metabolism) and higher abundance of adaptive immune cells (memory B/T cells), representing a “metabolically regulated, immune-modulated” phenotype. Notably, these subtypes do not completely align with the traditional N1/N2 neutrophil classification(48), suggesting that neutrophils in stroke may exhibit more complex and plastic immunophenotypes, shaped by disease stage and the local immune-metabolic environment. For instance, metabolic comorbidities such as diabetes can skew neutrophils toward a pro-inflammatory N1 phenotype, enhance NETs formation, and exacerbate neuroinflammation and tissue injury after stroke(49). This immunophenotypic classification may provide a framework for risk stratification and tailored therapy. From a clinical standpoint, subtype C2 patients may have higher risk for hemorrhagic transformation and re-occlusion, particularly following thrombolytic therapy. For these patients, early NET-targeted interventions such as PAD4 inhibitors(50), FPR2 agonists(51), or nNIF peptides(52) may help mitigate excessive inflammation and thrombotic complications. Meanwhile, C1 subtype patients may benefit more from antioxidant therapy, BBB stabilizers, or neuroprotective agents. These findings support the rationale for immune subtype-guided therapy in IS. To explore therapeutic implications further, we performed drug enrichment analysis and molecular docking simulations, identifying rosuvastatin as a promising candidate for clinical repurposing. Rosuvastatin, a selective HMG-CoA reductase inhibitor, is commonly prescribed for preventing both primary and secondary atherosclerotic cardiovascular events, and it exerts a range of pleiotropic effects, including lipid-lowering, anti-inflammatory, plaque stabilization, and endothelial protection(53). Our docking analysis revealed that rosuvastatin binds stably to several NETs-related proteins (ABCA1, MMP9, and KCNJ15), with a binding affinity of -8.4 kcal/mol for ABCA1. Structural modeling suggested that rosuvastatin occupies the transporter’s active pocket and may enhance ABCA1 function by promoting cholesterol efflux and modulating miR-33b-5p expression(54). While its effect on NETs formation remains to be clarified, previous studies indicate that rosuvastatin attenuates BBB damage and hemorrhagic transformation in IS by inhibiting MMP9 via PDGFR-α/LRP1-MAPK signaling(55). While its direct impact on NETs formation requires further investigation, these mechanistic insights and binding data provide a foundation for future therapeutic exploration. In conclusion, we established a comprehensive framework encompassing NETs-related gene identification, diagnostic modeling, immune subtype classification, and therapeutic exploration in IS. This study is the first to delineate immunologically distinct molecular subtypes of stroke from the perspective of NETs, highlighting the central role of neutrophils in the inflammation-thrombosis axis. Despite the enhanced robustness from multi-omics integration and cross-validation, limitations such as small sample size, lack of single-cell/protein-level verification, and the need for experimental validation of drug-target interactions remain. Future studies should integrate dynamic time-course analyses, animal models, and functional assays to elucidate the mechanistic roles of key regulators such as KCNJ15 and ARG1, and develop immune-subtype-based precision therapies to support individualized stroke management and improve clinical outcomes. These findings may inform immune subtype-guided treatment decisions in ischemic stroke patients, facilitating early intervention and risk stratification in clinical settings. This study has a few limitations. Firstly, the relatively small sample size of the external validation cohort may have restricted the statistical power to identify significant differences at the individual gene level. Second, the study relied primarily on transcriptomic data, without confirmation from single-cell sequencing, proteomics, or in vivo functional validation. Third, although we utilized cross-validation and testing with external datasets to validate the model's robustness, prospective clinical trials and mechanistic studies are crucial to verify the diagnostic and therapeutic potential of the identified gene signature. 5 Conclusion Our findings offer a novel framework for NETs-related biomarker identification, molecular subtyping, and precision therapeutic exploration in ischemic stroke. The proposed immune subtypes, based on neutrophil-driven transcriptomic features, may facilitate individualized treatment planning and early risk assessment. Moreover, the identification of actionable targets such as MMP9 and ABCA1 opens avenues for drug repurposing strategies in stroke management. Future studies should prioritize validation in larger and more diverse clinical cohorts, integration with longitudinal immune profiling, and experimental elucidation of the roles of key regulators like KCNJ15 and ARG1 . These efforts could ultimately inform immune-based personalized interventions for improving stroke outcomes. Abbreviations AUC area under the curve BBB blood-brain barrier BP biological processes CC cellular components CDF cumulative distribution function DCA decision curve analysis DCs dendritic cells DEGs differentially expressed genes FDR false discovery rate FC fold change GEO Gene Expression Omnibus GO Gene Ontology GSVA gene set variation analysis IS ischemic stroke KEGG Kyoto Encyclopedia of Genes and Genomes LASSO least absolute shrinkage and selection operator MF molecular functions NETs neutrophil extracellular traps PCA Principal component analysis RF Random Forest ssGSEA single-sample gene set enrichment analysis SVM-RFE support vector machine-recursive feature elimination ROC receiver operating characteristic TOM topological overlap matrix WGCNA Weighted Gene Co-expression Network Analysis. Declarations Ethics approval and consent to participate Not applicable. The study was based on publicly available datasets, and no human or animal experiments were conducted. Consent for publication Not applicable. No individual patient data are included in this manuscript. Availability of data and material All data and findings generated during this study are available within the article and its supplementary materials. For any additional information, readers are encouraged to contact the corresponding author. Competing interests The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Funding This study was supported by the Yunnan Science and Technology Program (Grant No. 202401AT070176). Authors' contributions XL: Conceptualization, Methodology, Formal analysis, Visualization, Writing-original draft. QL: Data curation, Software, Validation, Funding acquisition. YS: Formal analysis, Software, Visualization. GF: Investigation, Resources, Data curation. YH: Resources, Supervision. SQ: Formal analysis, Software. LZ: Conceptualization, Supervision, Writing-review & editing, Project administration. All authors (XL, QL, YS, GF, YH, SQ, and LZ) reviewed and approved the final manuscript. Acknowledgements The authors express their sincere gratitude to all individuals who took part in this study. References Qin C, Yang S, Chu YH, Zhang H, Pang XW, Chen L, et al. Signaling pathways involved in ischemic stroke: molecular mechanisms and therapeutic interventions. Signal Transduct Target Ther. 2022;7(1):215. 10.1038/s41392-022-01064-1. Feigin VL, Brainin M, Norrving B, Martins SO, Pandian J, Lindsay P, et al. World Stroke Organization: Global Stroke Fact Sheet 2025. Int J Stroke. 2025;20(2):132-44. 10.1177/17474930241308142. Zhu W, He X, Huang D, Jiang Y, Hong W, Ke S, et al. Global and Regional Burden of Ischemic Stroke Disease from 1990 to 2021: An Age-Period-Cohort Analysis. Transl Stroke Res. 2024. 10.1007/s12975-024-01319-9. Majumder D. Ischemic Stroke: Pathophysiology and Evolving Treatment Approaches. Neurosci Insights. 2024;19:26331055241292600. 10.1177/26331055241292600. Wang H, Zhang S, Xie L, Zhong Z, Yan F. Neuroinflammation and peripheral immunity: Focus on ischemic stroke. International Immunopharmacology. 2023;120:110332. 10.1016/j.intimp.2023.110332. Hurd MD, Goel I, Sakai Y, Teramura Y. Current status of ischemic stroke treatment: From thrombolysis to potential regenerative medicine. Regenerative Therapy. 2021;18:408-17. 10.1016/j.reth.2021.09.009. Mosconi MG, Paciaroni M. Treatments in Ischemic Stroke: Current and Future. Eur Neurol. 2022;85(5):349-66. 10.1159/000525822. Gong Z, Guo J, Liu B, Guo Y, Cheng C, Jiang Y, et al. Mechanisms of immune response and cell death in ischemic stroke and their regulation by natural compounds. Front Immunol. 2023;14:1287857. 10.3389/fimmu.2023.1287857. Luo H, Guo H, Zhou Y, Fang R, Zhang W, Mei Z. Neutrophil Extracellular Traps in Cerebral Ischemia/Reperfusion Injury: Friend and Foe. Curr Neuropharmacol. 2023;21(10):2079-96. 10.2174/1570159x21666230308090351. Ji YM, Li T, Qin YH, Xiao SY, Lv YH, Dong Y, et al. Neutrophil Extracellular Traps (NETs) in Sterile Inflammatory Diseases. J Inflamm Res. 2025;18:7989-8004. 10.2147/jir.S526936. Liaptsi E, Merkouris E, Polatidou E, Tsiptsios D, Gkantzios A, Kokkotis C, et al. Targeting Neutrophil Extracellular Traps for Stroke Prognosis: A Promising Path. Neurol Int. 2023;15(4):1212-26. 10.3390/neurolint15040076. Gu X, Dong M, Xia S, Li H, Bao X, Cao X, et al. γ-Glutamylcysteine ameliorates blood-brain barrier permeability and neutrophil extracellular traps formation after ischemic stroke by modulating Wnt/β-catenin signalling in mice. Eur J Pharmacol. 2024;969:176409. 10.1016/j.ejphar.2024.176409. Denorme F, Portier I, Rustad JL, Cody MJ, de Araujo CV, Hoki C, et al. Neutrophil extracellular traps regulate ischemic stroke brain injury. J Clin Invest. 2022;132(10). 10.1172/jci154225. Wang R, Zhu Y, Liu Z, Chang L, Bai X, Kang L, et al. Neutrophil extracellular traps promote tPA-induced brain hemorrhage via cGAS in mice with stroke. Blood. 2021;138(1):91-103. 10.1182/blood.2020008913. Gao X, Zhao X, Li J, Liu C, Li W, Zhao J, et al. Neutrophil extracellular traps mediated by platelet microvesicles promote thrombosis and brain injury in acute ischemic stroke. Cell Commun Signal. 2024;22(1):50. 10.1186/s12964-023-01379-8. Dhanesha N, Ansari J, Pandey N, Kaur H, Virk C, Stokes KY. Poststroke venous thromboembolism and neutrophil activation: an illustrated review. Res Pract Thromb Haemost. 2023;7(4):100170. 10.1016/j.rpth.2023.100170. Zhao Z, Pan Z, Zhang S, Ma G, Zhang W, Song J, et al. Neutrophil extracellular traps: A novel target for the treatment of stroke. Pharmacol Ther. 2023;241:108328. 10.1016/j.pharmthera.2022.108328. Li J, Liu L, Zhang R, Pan L, Tan J, Ou M, et al. Associations of NETs with inflammatory risk and clinical predictive value in large artery atherosclerosis stroke: a prospective cohort study. Front Immunol. 2024;15:1488317. 10.3389/fimmu.2024.1488317. Wu ZR, Zhou TQ, Ai SC. Neutrophil extracellular traps correlate with severity and prognosis in patients with ischemic stroke: a systematic review and meta-analysis. Acta Neurol Belg. 2024;124(2):513-22. 10.1007/s13760-023-02409-5. Novotny J, Oberdieck P, Titova A, Pelisek J, Chandraratne S, Nicol P, et al. Thrombus NET content is associated with clinical outcome in stroke and myocardial infarction. Neurology. 2020;94(22):e2346-e60. 10.1212/wnl.0000000000009532. Oh SA, Seol SI, Davaanyam D, Kim SW, Lee JK. Platelet-derived HMGB1 induces NETosis, exacerbating brain damage in the photothrombotic stroke model. Mol Med. 2025;31(1):46. 10.1186/s10020-025-01107-7. Huang T, Guo Y, Xie W, Yin J, Zhang Y, Chen W, et al. Brain border-derived CXCL2(+) neutrophils drive NET formation and impair vascular reperfusion following ischemic stroke. CNS Neurosci Ther. 2024;30(8):e14916. 10.1111/cns.14916. Fang H, Bo Y, Hao Z, Mang G, Jin J, Wang H. A promising frontier: targeting NETs for stroke treatment breakthroughs. Cell Commun Signal. 2024;22(1):238. 10.1186/s12964-024-01563-4. Vogelgesang A, Lange C, Blümke L, Laage G, Rümpel S, Langner S, et al. Ischaemic stroke and the recanalization drug tissue plasminogen activator interfere with antibacterial phagocyte function. J Neuroinflammation. 2017;14(1):140. 10.1186/s12974-017-0914-6. Denorme F, Portier I, Cody M, Grandhi R, Neal MD, Majersik JJ, et al. Platelet-Mediated NET Formation Exacerbates Ischemic Stroke Brain Injury. Blood. 2021;138:437. 10.1182/blood-2021-151423. Qin R, Xu W, Qin Q, Liang X, Lai X, Xie M, et al. Identification of NETs-related genes as diagnostic biomarkers in ischemic stroke using RNA sequencing and single-cell analysis. Mamm Genome. 2025;36(2):651-64. 10.1007/s00335-025-10117-z. Liu Y, Wang W, Cui X, Lyu J, Xie Y. Exploring Genetic Associations of 3 Types of Risk Factors With Ischemic Stroke: An Integrated Bioinformatics Study. Stroke. 2024;55(6):1619-28. 10.1161/strokeaha.123.044424. Li T, Kang X, Zhang S, Wang Y, He J, Li H, et al. Integrating machine learning and multi-omics analysis to reveal nucleotide metabolism-related immune genes and their functional validation in ischemic stroke. Front Immunol. 2025;16:1561544. 10.3389/fimmu.2025.1561544. Yang WX, Wang FF, Pan YY, Xie JQ, Lu MH, You CG. Comparison of ischemic stroke diagnosis models based on machine learning. Front Neurol. 2022;13:1014346. 10.3389/fneur.2022.1014346. Li RB, Yang XH, Zhang JD, Cui W. GAS6-AS1, a long noncoding RNA, functions as a key candidate gene in atrial fibrillation related stroke determined by ceRNA network analysis and WGCNA. BMC Med Genomics. 2023;16(1):51. 10.1186/s12920-023-01478-y. Wang Y, Liang S, Hong Q, Mu J, Wu Y, Li K, et al. Construction of a neutrophil extracellular trap formation-related gene model for predicting the survival of lung adenocarcinoma patients and their response to immunotherapy. Transl Lung Cancer Res. 2024;13(12):3407-25. 10.21037/tlcr-24-463. Shi H, Pan Y, Xiang G, Wang M, Huang Y, He L, et al. A novel NET-related gene signature for predicting DLBCL prognosis. J Transl Med. 2023;21(1):630. 10.1186/s12967-023-04494-9. Zhang Y, Guo L, Dai Q, Shang B, Xiao T, Di X, et al. A signature for pan-cancer prognosis based on neutrophil extracellular traps. J Immunother Cancer. 2022;10(6). 10.1136/jitc-2021-004210. Salaudeen MA, Bello N, Danraka RN, Ammani ML. Understanding the Pathophysiology of Ischemic Stroke: The Basis of Current Therapies and Opportunity for New Ones. Biomolecules. 2024;14(3). 10.3390/biom14030305. Lapostolle A, Loyer C, Elhorany M, Chaigneau T, Bielle F, Alamowitch S, et al. Neutrophil Extracellular Traps in Ischemic Stroke Thrombi Are Associated Wth Poor Clinical Outcome. Stroke: Vascular and Interventional Neurology. 2023;3(3):e000639. doi:10.1161/SVIN.122.000639. Wanrooy BJ, Wen SW, Wong CH. Dynamic roles of neutrophils in post-stroke neuroinflammation. Immunol Cell Biol. 2021;99(9):924-35. 10.1111/imcb.12463. Petrone AB, O'Connell GC, Regier MD, Chantler PD, Simpkins JW, Barr TL. The Role of Arginase 1 in Post-Stroke Immunosuppression and Ischemic Stroke Severity. Transl Stroke Res. 2016;7(2):103-10. 10.1007/s12975-015-0431-9. Tang C, Lan R, Ma DR, Zhao M, Zhang Y, Li HY, et al. Annexin A1: The dawn of ischemic stroke (Review). Mol Med Rep. 2025;31(3). 10.3892/mmr.2024.13427. Ansari J, Gavins FNE. Neutrophils and Platelets: Immune Soldiers Fighting Together in Stroke Pathophysiology. Biomedicines. 2021;9(12). 10.3390/biomedicines9121945. Cui X, Chopp M, Zacharek A, Karasinska JM, Cui Y, Ning R, et al. Deficiency of brain ATP-binding cassette transporter A-1 exacerbates blood-brain barrier and white matter damage after stroke. Stroke. 2015;46(3):827-34. 10.1161/strokeaha.114.007145. Au A, Griffiths LR, Irene L, Kooi CW, Wei LK. The impact of APOA5, APOB, APOC3 and ABCA1 gene polymorphisms on ischemic stroke: Evidence from a meta-analysis. Atherosclerosis. 2017;265:60-70. 10.1016/j.atherosclerosis.2017.08.003. Wang T, Kim DH, Ding C, Wang D, Zhang W, Silic M, et al. Inwardly rectifying potassium channels regulate membrane potential polarization and direction sensing during neutrophil chemotaxis. bioRxiv. 2025. 10.1101/2025.03.06.641746. Zhang B, Huang W, Yi M, Xing C. Gene Differential Expression and Interaction Networks Illustrate the Biomarkers and Molecular Mechanisms of Atherosclerotic Cerebral Infarction. J Healthc Eng. 2022;2022:3912697. 10.1155/2022/3912697. Wang X, Liu X. Exploration of the shared gene signatures and molecular mechanisms between cardioembolic stroke and ischemic stroke. Front Neurol. 2025;16:1567902. 10.3389/fneur.2025.1567902. Turner RJ, Sharp FR. Implications of MMP9 for Blood Brain Barrier Disruption and Hemorrhagic Transformation Following Ischemic Stroke. Front Cell Neurosci. 2016;10:56. 10.3389/fncel.2016.00056. Chen R, Zhang X, Gu L, Zhu H, Zhong Y, Ye Y, et al. New Insight Into Neutrophils: A Potential Therapeutic Target for Cerebral Ischemia. Front Immunol. 2021;12:692061. 10.3389/fimmu.2021.692061. Ju H, Park KW, Kim ID, Cave JW, Cho S. Phagocytosis converts infiltrated monocytes to microglia-like phenotype in experimental brain ischemia. J Neuroinflammation. 2022;19(1):190. 10.1186/s12974-022-02552-5. Xie M, Hao Y, Feng L, Wang T, Yao M, Li H, et al. Neutrophil Heterogeneity and its Roles in the Inflammatory Network after Ischemic Stroke. Curr Neuropharmacol. 2023;21(3):621-50. 10.2174/1570159x20666220706115957. Liu S, Cai W, Hu M, Lu Z. Abstract WP317: Neutrophil Subtypes and Nets Changes in Diabetes With Acute Ischemic Stroke. Stroke.51(Suppl_1):AWP317-AWP. 10.1161/str.51.suppl_1.WP317. Kang L, Yu H, Yang X, Zhu Y, Bai X, Wang R, et al. Neutrophil extracellular traps released by neutrophils impair revascularization and vascular remodeling after stroke. Nat Commun. 2020;11(1):2488. 10.1038/s41467-020-16191-y. Zhu J, Li L, Ding J, Huang J, Shao A, Tang B. The Role of Formyl Peptide Receptors in Neurological Diseases via Regulating Inflammation. Front Cell Neurosci. 2021;15:753832. 10.3389/fncel.2021.753832. Bircher JS, Denorme F, Cody MJ, de Araujo CV, Petrey AC, Middleton EA, et al. Neonatal NET-inhibitory factor inhibits macrophage extracellular trap formation. Blood Advances. 2024;8(14):3686-90. 10.1182/bloodadvances.2024013094. Mehta A, Jain P, Patil R, Sashi Kant T, Indurkar SA, Kota SK, et al. Real-World Clinical Experience of Rosuvastatin as a Lipid-Lowering Therapy for Primary and Secondary Prevention of Cardiovascular Events (REAL ROSE). Cureus. 2022;14(11):e31468. 10.7759/cureus.31468. Santovito D, Marcantonio P, Mastroiacovo D, Natarelli L, Mandolini C, De Nardis V, et al. High dose rosuvastatin increases ABCA1 transporter in human atherosclerotic plaques in a cholesterol-independent fashion. Int J Cardiol. 2020;299:249-53. 10.1016/j.ijcard.2019.07.094. Lu D, Mai HC, Liang YB, Xu BD, Xu AD, Zhang YS. Beneficial Role of Rosuvastatin in Blood-Brain Barrier Damage Following Experimental Ischemic Stroke. Front Pharmacol. 2018;9:926. 10.3389/fphar.2018.00926. Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterial.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 20 Oct, 2025 Reviewers agreed at journal 19 Oct, 2025 Reviewers invited by journal 08 Oct, 2025 Editor invited by journal 11 Sep, 2025 Editor assigned by journal 04 Sep, 2025 Submission checks completed at journal 04 Sep, 2025 First submitted to journal 02 Sep, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7515929","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":532253808,"identity":"69d4928d-1aca-47e8-86e5-e0ebc1ab7b24","order_by":0,"name":"Xiaoqian Lan","email":"","orcid":"","institution":"The First Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaoqian","middleName":"","lastName":"Lan","suffix":""},{"id":532253810,"identity":"f45be0f4-b458-4ac2-bd11-a61becb658ed","order_by":1,"name":"Qing Li","email":"","orcid":"","institution":"The First Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qing","middleName":"","lastName":"Li","suffix":""},{"id":532253813,"identity":"4aef1894-cd88-43ab-b428-97bd742df4b6","order_by":2,"name":"Yuting Shi","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yuting","middleName":"","lastName":"Shi","suffix":""},{"id":532253814,"identity":"74f47cc9-5bf8-495e-a140-5f654c3d311f","order_by":3,"name":"Guangli Feng","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Guangli","middleName":"","lastName":"Feng","suffix":""},{"id":532253816,"identity":"fb7aec30-13d6-4852-b2dc-d7e2579a3e20","order_by":4,"name":"Yingrui Hu","email":"","orcid":"","institution":"Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yingrui","middleName":"","lastName":"Hu","suffix":""},{"id":532253817,"identity":"0927a670-0e91-4ae0-b502-617168aee065","order_by":5,"name":"Shiyi Qin","email":"","orcid":"","institution":"The First Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shiyi","middleName":"","lastName":"Qin","suffix":""},{"id":532253820,"identity":"1ad168df-7d79-42ce-a272-79bb9b9510c5","order_by":6,"name":"Lianmei Zhong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAsklEQVRIiWNgGAWjYNCCAgY5Nvb2A6RoMWAw5uM5k0CalsR5Eg4GRCq+kWMm+cNgW3qbBEMCw4+KbYS1SM7IMZOQMLid2ybdeICx58xtwlr4pYFaDEBaZA4kMDO2EaGFDaQlweB2OhuQJE4L2JYDBrcTiNciOf9ZsWWDwW3DNmAgHyTKLwZnDm+8+aPitrx8e/vBB0AGYS0MDBwmEjDmAWLUAwH74w9EqhwFo2AUjIKRCgDPaDmHAYXJaQAAAABJRU5ErkJggg==","orcid":"","institution":"Capital Medical University","correspondingAuthor":true,"prefix":"","firstName":"Lianmei","middleName":"","lastName":"Zhong","suffix":""}],"badges":[],"createdAt":"2025-09-02 09:23:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7515929/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7515929/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94053312,"identity":"93c0c6fc-d3fa-44f2-b84d-f75d8eead1c8","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10977906,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/ab42ad6c7f752531aba6e6ac.docx"},{"id":94053306,"identity":"75e6115c-5ab2-46eb-90b1-00bcbc9f3b25","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8487,"visible":true,"origin":"","legend":"","description":"","filename":"08c1437def3e43f1b2e9fb6936cb3882.json","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/fa395a4c650e100630582997.json"},{"id":94053411,"identity":"d9f965f3-5323-4db2-ad52-29c8011d4df0","added_by":"auto","created_at":"2025-10-22 01:50:09","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1124905,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/baf22e52abb4cd72f03c6dde.docx"},{"id":94053011,"identity":"2025d922-53cc-4d05-ac1e-b3ec30476e7a","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":90255,"visible":true,"origin":"","legend":"","description":"","filename":"08c1437def3e43f1b2e9fb6936cb38821enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/d1c0e7c4ae8510a42b5741ce.xml"},{"id":94053307,"identity":"1e4e71a4-641f-49d7-8a54-f960fb93e0a6","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":548060,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/bc21b201d232ce5294a8d272.jpeg"},{"id":94053310,"identity":"dc30dde0-6d3a-4826-8bfe-abec94d1cda6","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1094118,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage10.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/bcfda87b7205f0f56061add5.jpeg"},{"id":94053474,"identity":"5cb0a2cc-63ab-40eb-be86-f89e9c0618e8","added_by":"auto","created_at":"2025-10-22 01:58:09","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2634384,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/ae02708cda3930170d39f2cd.jpeg"},{"id":94053473,"identity":"2a85f207-512d-475e-bdfc-5be3f298f1d0","added_by":"auto","created_at":"2025-10-22 01:58:09","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":924539,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/fd38eec963000539562f8f83.jpeg"},{"id":94053432,"identity":"4c2a8e20-e07e-440b-9615-c625b7d0c8d0","added_by":"auto","created_at":"2025-10-22 01:50:15","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1045797,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/8e855e6502d71236f4195f36.jpeg"},{"id":94053320,"identity":"66e83536-96ac-47ca-ba82-2a59c4998a7d","added_by":"auto","created_at":"2025-10-22 01:42:10","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1869439,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/18ae56d577f9f157c8a4fda7.jpeg"},{"id":94053040,"identity":"ab98a4e3-4706-4b0c-92bb-eeadd6af7b25","added_by":"auto","created_at":"2025-10-22 01:34:10","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1060536,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/481158aa8683c49a7e58d7d6.jpeg"},{"id":94053037,"identity":"1ffeddd2-2bd0-443d-9d86-a8c5ee3afa7a","added_by":"auto","created_at":"2025-10-22 01:34:10","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":507177,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/8441b3602db7bb335e9ceef2.jpeg"},{"id":94053413,"identity":"da6143fd-ef8b-496c-847b-da2c4e01a894","added_by":"auto","created_at":"2025-10-22 01:50:09","extension":"jpeg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":438745,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/83b96b6409346ef1bc307121.jpeg"},{"id":94053031,"identity":"e2cffc73-d735-4eba-930c-cfe32844ea59","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"jpeg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":709043,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/884807369c3467416cae93bb.jpeg"},{"id":94053023,"identity":"cc0edc64-b9b9-4f48-ac45-05c7e3cc20a6","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":135834,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/27c7d0cca5f658c590e7775d.png"},{"id":94053021,"identity":"31dc38eb-0d3b-4b91-a30c-4a5fff2c20f1","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":203212,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/2f47da7d4a88c3345ce24791.png"},{"id":94053029,"identity":"9c02457c-710b-497c-80dd-d1c04e537cc7","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":171583,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/9e363cd64e5d05293b1cb570.png"},{"id":94053414,"identity":"850e2c73-9c10-43fb-b1be-d267ab627024","added_by":"auto","created_at":"2025-10-22 01:50:10","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":156333,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/acbfb9c949eaa0d6a7f8ca4f.png"},{"id":94053042,"identity":"b9b420aa-24a1-40b9-aba4-68257f1126bd","added_by":"auto","created_at":"2025-10-22 01:34:10","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":217261,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/d819089edf51bd9b66cfb9ae.png"},{"id":94053315,"identity":"298a37df-6112-4252-8732-b433ae7b102e","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":322650,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/bbc429ac556f3d036b896da4.png"},{"id":94053033,"identity":"7eebf06b-e4f5-48db-aa21-a2094e94301f","added_by":"auto","created_at":"2025-10-22 01:34:10","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":178330,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/bc7c3c08558834c364104d16.png"},{"id":94053321,"identity":"d4edd3b4-39fc-4df2-8dac-9e9c6e752940","added_by":"auto","created_at":"2025-10-22 01:42:10","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":94892,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/cd076128f925510870eeda76.png"},{"id":94053318,"identity":"ce1661b8-62e1-443a-a436-9d0017943150","added_by":"auto","created_at":"2025-10-22 01:42:10","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":95844,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/abd129bc4d5bc5ee1c0f5d7e.png"},{"id":94053041,"identity":"89ca15c3-6b7c-4ba9-bf9d-9c1272485bdd","added_by":"auto","created_at":"2025-10-22 01:34:10","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":166580,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/187905a9947c03f7015e3c74.png"},{"id":94053043,"identity":"704d409f-f83b-4027-95f2-12bbfcee2fcf","added_by":"auto","created_at":"2025-10-22 01:34:10","extension":"xml","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89019,"visible":true,"origin":"","legend":"","description":"","filename":"08c1437def3e43f1b2e9fb6936cb38821structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/2fbdac5d78d9bb220a21d46c.xml"},{"id":94053322,"identity":"02d490da-a84b-4989-8247-9545c85257c1","added_by":"auto","created_at":"2025-10-22 01:42:10","extension":"html","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":106100,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/46752bdebe4b23287723cc65.html"},{"id":94053007,"identity":"ab9dd1c5-4e4c-421e-ac55-d29ab775cd6e","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2333536,"visible":true,"origin":"","legend":"\u003cp\u003eThe design and procedural framework of the study.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/b93aebd9269e30fe13174e83.png"},{"id":94053015,"identity":"e3f958dc-f4cb-48d3-a138-1e3c7cb7e018","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":6191866,"visible":true,"origin":"","legend":"\u003cp\u003eBatch effect correction and identification of DEGs. \u003cstrong\u003e(A-B)\u003c/strong\u003e PCA plots illustrating sample distribution before \u003cstrong\u003e(A)\u003c/strong\u003e and after \u003cstrong\u003e(B)\u003c/strong\u003ebatch correction. \u003cstrong\u003e(C)\u003c/strong\u003e Volcano plot showing DEGs between IS and control groups. \u003cstrong\u003e(D)\u003c/strong\u003e Heatmap showing the top 30 up- and downregulated DEGs.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/54f208278197e0939a2c48b3.png"},{"id":94053008,"identity":"7985dc05-30cb-4b71-adde-bbb49e3682f3","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":3318078,"visible":true,"origin":"","legend":"\u003cp\u003eFunctional enrichment analysis of DEGs.\u003cstrong\u003e (A)\u003c/strong\u003e GO enrichment analysis of DEGs, highlighting significantly overrepresented terms in BP, CC, and MF categories. \u003cstrong\u003e(B)\u003c/strong\u003e KEGG pathways associated with immune and metabolic responses in ischemic stroke.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/78c21d9c1fb9fbfb7ceb76e4.png"},{"id":94053010,"identity":"ff1c0502-2acc-4c18-a213-5cab8204ee9e","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":3969707,"visible":true,"origin":"","legend":"\u003cp\u003eWGCNA reveals gene modules associated with IS.\u003cstrong\u003e (A)\u003c/strong\u003e Determination of soft-thresholding power β based on scale-free topology. \u003cstrong\u003e(B)\u003c/strong\u003e Gene clustering and module assignment. \u003cstrong\u003e(C)\u003c/strong\u003e Heatmap showing correlation between disease status and module eigengenes. \u003cstrong\u003e(D)\u003c/strong\u003e Sample clustering with phenotype annotation.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/41099615120bca5490d12726.png"},{"id":94053019,"identity":"932556c3-a22f-42c9-8aa0-15690ed4e5ea","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":8391213,"visible":true,"origin":"","legend":"\u003cp\u003eIdentification and genomic distribution of overlapping genes. \u003cstrong\u003e(A)\u003c/strong\u003eVenn diagram showing intersecting genes from DEGs, WGCNA module, and NETs database. \u003cstrong\u003e(B)\u003c/strong\u003e Expression levels of shared genes in IS vs. control. \u003cstrong\u003e(C)\u003c/strong\u003eCorrelation matrix among overlapping genes. \u003cstrong\u003e(D)\u003c/strong\u003e Circos plot of chromosomal locations. \u003cstrong\u003e(E)\u003c/strong\u003e Manhattan plot of gene-level significance.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/88d3c54776fbb74874e2fac0.png"},{"id":94053018,"identity":"65487040-70a0-4712-ac9e-8db43ecb5e4e","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":4795007,"visible":true,"origin":"","legend":"\u003cp\u003eImmune infiltration patterns and their association with overlapping genes. \u003cstrong\u003e(A)\u003c/strong\u003e ssGSEA-based heatmap showing immune cell abundance. \u003cstrong\u003e(B)\u003c/strong\u003e Violin plots comparing infiltration scores between groups. \u003cstrong\u003e(C)\u003c/strong\u003e Spearman correlation heatmap between overlapping genes and immune cells.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/50ccd3e499d61399715cd9cd.png"},{"id":94053017,"identity":"2a918c36-a689-4546-a334-c3048dae9f5d","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":2166076,"visible":true,"origin":"","legend":"\u003cp\u003eMachine learning selection of diagnostic hub genes.\u003cstrong\u003e (A-B)\u003c/strong\u003eLASSO model selection and coefficient paths. \u003cstrong\u003e(C-D)\u003c/strong\u003e SVM-RFE classification accuracy and error. \u003cstrong\u003e(E-F)\u003c/strong\u003e RF tree optimization and gene importance.\u003cstrong\u003e (G) \u003c/strong\u003eHub genes identified by three algorithms. \u003cstrong\u003e(H)\u003c/strong\u003eExpression levels of selected hub genes.\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/b9c88010fc891456eb0ec494.png"},{"id":94053308,"identity":"3f1ec857-dad3-453b-bb0b-9dbe7b6ce33a","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1727458,"visible":true,"origin":"","legend":"\u003cp\u003eEvaluation of diagnostic performance. \u003cstrong\u003e(A) \u003c/strong\u003eROC curves for individual feature genes. \u003cstrong\u003e(B)\u003c/strong\u003eROC curve for combined gene model. \u003cstrong\u003e(C)\u003c/strong\u003e Nomogram based on selected genes. \u003cstrong\u003e(D)\u003c/strong\u003e Calibration curve of the logistic regression model. \u003cstrong\u003e(E)\u003c/strong\u003e DCA. \u003cstrong\u003e(F)\u003c/strong\u003eExternal validation ROC curve.\u003c/p\u003e","description":"","filename":"image8.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/098184579a781c8e3dd87181.png"},{"id":94053027,"identity":"4082fff9-c5db-43de-a39f-0ccd28968784","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":3203606,"visible":true,"origin":"","legend":"\u003cp\u003eConsensus clustering identifies molecular subtypes of ischemic stroke.\u003cstrong\u003e (A)\u003c/strong\u003eConsensus matrix (k = 2). \u003cstrong\u003e(B-C)\u003c/strong\u003e CDF and cluster score plots confirm cluster stability. \u003cstrong\u003e(D)\u003c/strong\u003e Differential gene expression between subtypes. \u003cstrong\u003e(E)\u003c/strong\u003eHeatmap of hub gene profiles. \u003cstrong\u003e(F)\u003c/strong\u003e Immune infiltration comparison across subtypes. \u003cstrong\u003e(G) \u003c/strong\u003ePCA shows transcriptomic separation. \u003cstrong\u003e(H)\u003c/strong\u003e GSVA-based enrichment of subtype-specific pathways.\u003c/p\u003e","description":"","filename":"image9.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/8668355c08a6cac4896a866c.png"},{"id":94053317,"identity":"ae2a56fa-d9c2-4b59-8c44-2e57309ea75b","added_by":"auto","created_at":"2025-10-22 01:42:09","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":4679472,"visible":true,"origin":"","legend":"\u003cp\u003eDrug enrichment and docking analysis. \u003cstrong\u003e(A)\u003c/strong\u003e Top enriched drugs targeting hub genes from DSigDB. \u003cstrong\u003e(B)\u003c/strong\u003e Predicted binding of rosuvastatin to ABCA1 protein.\u003cstrong\u003e (C)\u003c/strong\u003e Drug-gene interaction network illustrating potential multitarget effects.\u003c/p\u003e","description":"","filename":"image10.png","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/8f11ccbd688d7f2a22009d35.png"},{"id":94053479,"identity":"55c2e3a1-a243-4ddf-8823-747650855200","added_by":"auto","created_at":"2025-10-22 01:58:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":41289420,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/b716527c-163b-4e4d-986a-4d163bbfe10c.pdf"},{"id":94053013,"identity":"8f834730-8f25-4135-814d-19046cbc326f","added_by":"auto","created_at":"2025-10-22 01:34:09","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1124905,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-7515929/v1/cb9cd9349d1057d85d375535.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eImmune subtypes and diagnostic genes revealed by neutrophil trap-associated transcriptomic signatures in ischemic stroke\u003c/p\u003e","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eIschemic stroke (IS) continues to be a major contributor to long-term disability and death globally(1, 2). The global incidence of ischemic stroke (IS) has been steadily increasing, driven by an aging population and a growing prevalence of vascular risk factors(3). IS is primarily triggered by the sudden blockage of cerebral arteries, which sets off a series of pathological events such as neuronal damage, oxidative damage, immune system disruptions, and impairment of vascular integrity(4, 5). Despite advancements in reperfusion therapies, such as mechanical thrombectomy and intravenous thrombolysis, the clinical benefits of these treatments are limited by factors such as narrow therapeutic windows, strict patient eligibility, and risks of hemorrhagic transformation or reperfusion injury(6, 7). Therefore, discovering new molecular biomarkers and potential therapeutic targets is crucial for is essential for enabling precision subtyping and individualized treatment strategies in IS.\u003c/p\u003e\u003cp\u003eOver recent years, inflammation and immune activation have emerged as critical contributors to stroke pathogenesis(8). Neutrophils, as key effector cells of the innate immune system, have recently garnered increasing attention for their ability to release neutrophil extracellular traps (NETs)(9). NETs are extracellular networks primarily made of decondensed chromatin, including DNA and histones, and contain granular enzymes released by neutrophils(10). Originally identified as antimicrobial agents, NETs are now recognized for their proinflammatory, prothrombotic, and tissue-damaging effects in non-infectious settings(11, 12). Following cerebral ischemia, neutrophils are rapidly recruited to the ischemic region and release NETs both in the brain and peripheral circulation(13). These NETs disrupt the blood-brain barrier (BBB), exacerbate cerebral edema and neuronal necrosis(14), promote thrombus formation(15, 16), and activate the coagulation cascade(17), thereby amplifying neurovascular injury and delaying tissue repair(18). High levels of NETs have been linked to greater stroke severity and worse clinical outcomes, highlighting their potential as both prognostic biomarkers and targets for therapeutic intervention(19, 20).\u003c/p\u003e\u003cp\u003eAlthough early research has emphasized the role of specific NETs-related molecules in ischemic stroke (IS), a comprehensive understanding of their broader regulatory context remains lacking. For instance, platelet-derived high mobility group box 1 has been demonstrated to trigger NETs release, thereby exacerbating ischemic damage(21). Similarly, brain border-associated CXCL2⁺ neutrophils have been shown to mediate reperfusion injury through NETs formation mechanisms(22). Several circulating NETs biomarkers, such as citrullinated histone H3 and myeloperoxidase-DNA complexes, have been found to be markedly elevated in IS patients and show positive correlations with neurological impairment scores(23, 24). Therapeutic interventions aimed at inhibiting NETs, including the neonatal NET-inhibitory factor, have shown promise in preclinical models by reducing infarct volume and improving neurological outcomes(25). Moreover, with the advancement of high-throughput sequencing technologies, emerging studies have identified NETs-associated genes (including \u003cem\u003eCEACAM3\u003c/em\u003e, \u003cem\u003eTNF\u003c/em\u003e, and \u003cem\u003eSELP\u003c/em\u003e) enriched in neutrophils from IS animal models. These genes are implicated in leukocyte adhesion, ferroptosis, and IL-17 signaling pathways, and exhibit potential diagnostic relevance(26). Nonetheless, most existing studies have focused on individual molecules or discrete pathways, with limited efforts to systematically map the molecular network that underpins NETs biology in IS. In addition, NETs-based immune subtyping remains an underexplored area, significantly constraining its translational value for patient stratification and targeted intervention.\u003c/p\u003e\u003cp\u003eTo address these limitations, the present study integrates transcriptomic analysis from multiple public datasets, leveraging differential gene expression analysis, weighted gene co-expression network analysis (WGCNA), and three complementary machine learning algorithms (support vector machine-recursive feature elimination (SVM-RFE), least absolute shrinkage and selection operator (LASSO), and Random Forest (RF)) to identify NETs-associated hub genes. Subsequently, we developed a reliable diagnostic model utilizing these genes. Additionally, functional enrichment and drug-gene interaction mapping are performed to explore the therapeutic implications of these key hub genes, while molecular docking is employed to predict drug interactions. This comprehensive analytical framework provides a novel approach for immune subtyping and targeted therapy in IS, demonstrating the clinical applicability and potential of NETs-related biomarkers in precision stroke management.\u003c/p\u003e"},{"header":"2 Materials and Methods","content":"\u003cp\u003eData Acquisition and Preparation\u003c/p\u003e\u003cp\u003eGene expression data were sourced from the Gene Expression Omnibus (GEO) database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Two datasets (GSE16561(27) (platform: GPL6883) and GSE58294(28) (platform: GPL570)) were selected as the training cohort, including 108 IS samples and 47 healthy controls. The external validation cohort comprised GSE195442(29) (platform: GPL31275) and GSE66724(30) (platform: GPL570), resulting in a combined dataset of 18 IS and 18 control samples (Supplementary Table\u0026nbsp;1).\u003c/p\u003e\u003cp\u003eAll data processing was conducted in R (v4.4.1). Repeated probes were merged using the avereps() function in the limma package, followed by log2 transformation and normalization via the normalizeBetweenArrays() function. To adjust for batch effects across datasets, the ComBat function from the sva package was applied. Principal component analysis (PCA) was then used to assess and visualize variations before and after the correction, utilizing the ggplot2 and ggpubr packages.\u003c/p\u003e\u003cp\u003eTo identify NETs-related genes, 770 protein-coding genes with a relevance score\u0026thinsp;\u0026gt;\u0026thinsp;8 were retrieved from the GeneCards Human Gene Database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.genecards.org/\u003c/span\u003e\u003cspan address=\"https://www.genecards.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). An additional 241 genes were integrated from published literature (Supplementary Table\u0026nbsp;2)(31\u0026ndash;33). After deduplication, a total of 855 unique NETs-related genes were included for downstream analysis. The overall analytical workflow is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eDifferential Expression Analysis\u003c/p\u003e\u003cp\u003eAfter constructing the design matrix, differential gene expression was analyzed using the limma package. DEGs were identified using the criteria |log2 fold change (FC)| \u0026gt;0.5 and a false discovery rate (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.05. The 30 most upregulated and downregulated DEGs were visualized using the pheatmap package, while overall gene distribution was presented via volcano plots using ggplot2.\u003c/p\u003e\u003cp\u003eFunctional Enrichment Analysis\u003c/p\u003e\u003cp\u003eGene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the ClusterProfiler package to explore the biological functions of DEGs. The GO terms analyzed included biological process (BP), cellular component (CC), and molecular function (MF). Terms with \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and adjusted \u003cem\u003eP\u003c/em\u003e-values (\u003cem\u003ep\u003c/em\u003e.adjust)\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered significant. The results were visualized through bubble and bar plots.\u003c/p\u003e\u003cp\u003eWeighted Gene Co-expression Network Analysis (WGCNA)\u003c/p\u003e\u003cp\u003eA co-expression network was generated from the training datasets using the WGCNA package (v1.73). Genes with a standard deviation\u0026thinsp;\u0026le;\u0026thinsp;0.5 were excluded. Sample clustering was performed to detect outliers, using a cut-off height of 20,000. The optimal soft-thresholding power (β) was selected based on the pickSoftThreshold function, ensuring an R\u0026sup2; value\u0026thinsp;\u0026gt;\u0026thinsp;0.9. A topological overlap matrix (TOM) was calculated and converted into a dissimilarity matrix (1\u0026thinsp;\u0026minus;\u0026thinsp;TOM). Modules were detected using dynamic tree cutting with a minimum module size of 50, and modules with an eigengene correlation distance\u0026thinsp;\u0026lt;\u0026thinsp;0.3 were merged. Finally, the association between module eigengenes and sample traits was analyzed to identify key modules.\u003c/p\u003e\u003cp\u003eOverlapping Genes Identification and Integration\u003c/p\u003e\u003cp\u003eTo identify overlapping genes, DEGs, NETs-related genes, and genes from significant WGCNA modules were intersected. Venn diagrams were generated using the ggvenn package. The expression levels of the overlapping genes were extracted, and intergroup differences were assessed using the Wilcoxon rank-sum test (FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Boxplots were created using ggpubr. Spearman correlation coefficients were calculated among overlapping genes within the IS group using the corrplot package, and hierarchical clustering was used to identify gene clusters. Genomic locations of overlapping genes were mapped using the hg38 reference genome, and Circos plots (circlize) along with Manhattan plots were constructed to visualize chromosomal distribution and gene-level significance.\u003c/p\u003e\u003cp\u003eImmune Cell Infiltration Profiling\u003c/p\u003e\u003cp\u003eSingle-sample gene set enrichment analysis (ssGSEA) was conducted using the gene set variation analysis (GSVA) package to quantify immune cell infiltration based on 28 immune-related gene sets from the immune.gmt file. Enrichment scores were normalized (min-max) across samples. Heatmaps of immune infiltration were plotted using pheatmap and clustered by group (IS vs. control). Intergroup differences were evaluated using the Wilcoxon test. Spearman correlations between overlapping genes expression and immune cell scores were computed and visualized as correlation heatmaps using ggplot2, with a blue-white-red gradient indicating correlation strength.\u003c/p\u003e\u003cp\u003eMachine Learning-Based Hub Gene Selection\u003c/p\u003e\u003cp\u003eTo identify reliable diagnostic biomarkers from the overlapping gene set, three distinct machine learning algorithms were applied:\u003c/p\u003e\u003cp\u003eLASSO: Implemented via the glmnet package with 10-fold cross-validation to select the optimal penalty parameter (lambda). Non-zero coefficient genes at minimum lambda were retained.\u003c/p\u003e\u003cp\u003eSVM-RFE: Recursive feature elimination was performed using the e1071 package and a custom script (NETs21.msvmRFE.R), selecting the gene subset with the lowest cross-validation error.\u003c/p\u003e\u003cp\u003eRF: The caret package was used to upsample minority classes and build the RF model. Feature importance was ranked by MeanDecreaseGini, and genes with importance\u0026thinsp;\u0026gt;\u0026thinsp;5 were selected.\u003c/p\u003e\u003cp\u003eVenn diagrams were employed to determine the common genes identified by all three methods, which were then retained as final robust hub genes.\u003c/p\u003e\u003cp\u003eDevelopment and Validation of the Diagnostic Model\u003c/p\u003e\u003cp\u003eA multivariate logistic regression model was constructed using glmnet based on the selected hub genes. In the training set, area under the curve (AUC) and receiver operating characteristic (ROC) curves values were calculated using the pROC package to evaluate discriminatory ability. Gene expression levels were visualized using boxplots (ggpubr and reshape2), and differences were tested with Wilcoxon rank-sum tests. To further assess model performance, calibration curves and nomograms (rms package) were used, while clinical utility was evaluated through decision curve analysis (DCA) with the rmda package. The same procedure was repeated for external validation.\u003c/p\u003e\u003cp\u003eMolecular Subtyping and Functional Profiling\u003c/p\u003e\u003cp\u003eBased on the expression levels of hub genes identified by machine learning in IS samples, molecular subtyping was performed using ConsensusClusterPlus (k-means algorithm, Euclidean distance, max k\u0026thinsp;=\u0026thinsp;9, 50 resampling iterations). The ideal number of clusters was selected using consensus cumulative distribution function (CDF) and consensus heatmaps. Subtype-specific expression patterns were visualized with heatmaps and PCA. Boxplots were used to compare gene expression among subtypes. Immune cell infiltration patterns among subtypes were evaluated using ssGSEA scores. Functional enrichment of KEGG pathways across subtypes was assessed using GSVA, and pathways with \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were visualized in bar plots (top 10 upregulated and downregulated).\u003c/p\u003e\u003cp\u003eDrug Enrichment and Regulatory Network Construction\u003c/p\u003e\u003cp\u003ePotential drug-gene interactions were explored using DGIdb (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.dgidb.org/\u003c/span\u003e\u003cspan address=\"https://www.dgidb.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and DSigDB (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://maayanlab.cloud/DSigDB/\u003c/span\u003e\u003cspan address=\"https://maayanlab.cloud/DSigDB/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Drug enrichment was conducted using the enricher function in ClusterProfiler (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Up to 10 drugs per gene were selected, and drug-gene interaction networks were built using Cytoscape software, with visualization facilitated by ggplot2 and enrichplot.\u003c/p\u003e\u003cp\u003eMolecular Docking Analysis\u003c/p\u003e\u003cp\u003eTo further elucidate ligand-target interactions and identify candidate therapeutic agents, structure-based molecular docking was performed. Rosuvastatin, the top-ranked drug from enrichment analysis, was selected for docking with its key targets: MMP9, ABCA1, and KCNJ15. The 3D structure of Rosuvastatin was retrieved from the PubChem database. Protein structures for MMP9 (PDB ID: 1ITV) and ABCA1 (PDB ID: 5XJY) were obtained from the Protein Data Bank (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.rcsb.org/\u003c/span\u003e\u003cspan address=\"https://www.rcsb.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), while KCNJ15 structure was predicted using AlphaFold (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://alphafold.com/entry/Q99712\u003c/span\u003e\u003cspan address=\"https://alphafold.com/entry/Q99712\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Docking was performed using the CB-Dock platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cadd.labshare.cn/cb-dock2/php/index.php\u003c/span\u003e\u003cspan address=\"https://cadd.labshare.cn/cb-dock2/php/index.php\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), which identifies potential binding cavities and executes blind docking via AutoDock Vina. Binding affinity scores (in kcal/mol) were used to evaluate binding strength, with thresholds of \u0026le; -5.0 kcal/mol indicating strong interaction. Binding poses and energy rankings were visualized and used to select optimal compound\u0026ndash;target pairs.\u003c/p\u003e"},{"header":"3 Results","content":"\u003cp\u003eIdentification of Differentially Expressed Genes and Batch Effect Correction\u003c/p\u003e\u003cp\u003eTo reduce technical variation and improve cross-cohort comparability, PCA was performed on the merged expression matrix from GSE16561 and GSE58294. Before correction, samples from the two datasets showed clear separation along the PC1 and PC2 axes, indicating significant batch effects (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Following batch correction and normalization using the sva package, sample distributions became more homogeneous in PCA space, demonstrating effective mitigation of batch-related variability (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Subsequently, differential expression analysis was performed between ischemic stroke (IS) and control samples, revealing 368 differentially expressed genes (DEGs), including 212 upregulated and 156 downregulated genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Heatmap visualization further highlighted distinct expression patterns between the two groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFunctional Enrichment of Differentially Expressed Genes\u003c/p\u003e\u003cp\u003eTo investigate the biological significance of the DEGs, GO and KEGG pathway enrichment analyses were performed. GO results revealed that the DEGs were significantly enriched in immune-related biological processes, including activation of immune responses, regulation of cytokine production, T-cell receptor signaling, and acute inflammatory responses (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). At the cellular component level, DEGs were primarily associated with secretory granule membranes, membrane microdomains, and the external side of the plasma membrane. Molecular function enrichment pointed to roles in immune receptor activity, immunoglobulin binding, and pattern recognition receptor signaling. KEGG pathway analysis further revealed prominent enrichment in immune and infection-associated pathways such as hematopoietic cell lineage, NF-κB signaling, T and B cell receptor signaling, and cytokine\u0026ndash;cytokine receptor interaction (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Additionally, metabolic and pathogen-related pathways, including Staphylococcus aureus infection, Tuberculosis, and Pantothenate and CoA biosynthesis, were significantly enriched, suggesting that these DEGs may regulate immune activation and inflammation in the pathogenesis of IS.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIdentification of Key Co-expression Modules via WGCNA\u003c/p\u003e\u003cp\u003eWGCNA was performed to identify gene modules linked to IS. A soft-thresholding power of β\u0026thinsp;=\u0026thinsp;6 was chosen based on the scale-free topology criterion (R\u0026sup2; \u0026gt;0.9) and average connectivity (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Several gene modules were identified via hierarchical clustering and color-coded (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Correlation analysis revealed a strong negative association between the blue module and IS status (r = -0.61, P\u0026thinsp;=\u0026thinsp;5e\u003csup\u003e\u0026minus;\u0026thinsp;17\u003c/sup\u003e), while the yellow module showed a positive correlation with IS (r\u0026thinsp;=\u0026thinsp;0.65, P\u0026thinsp;=\u0026thinsp;4e\u003csup\u003e\u0026minus;\u0026thinsp;20\u003c/sup\u003e); green and red modules also showed statistically significant associations (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). Sample clustering heatmaps confirmed distinct expression patterns between IS and control samples across these modules, underscoring their biological relevance (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIdentification and Characterization of NETs-Associated Overlapping Genes\u003c/p\u003e\u003cp\u003eTo identify overlapping genes implicated in both IS and NETs biology, we intersected DEGs, genes from the significant WGCNA module (yellow), and a curated NETs-related gene set. Twenty-one candidate hub genes were identified at the intersection (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). All were significantly differentially expressed between IS and control samples (Wilcoxon test, ***\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with the majority being upregulated in the IS group (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). Pearson correlation analysis demonstrated strong co-expression among these genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC), suggesting they may operate within coordinated regulatory networks. Chromosomal mapping revealed non-random genomic distribution. Some genes were clustered on specific chromosomes. For example, \u003cem\u003eFCGR1A\u003c/em\u003e and \u003cem\u003eF5\u003c/em\u003e on chromosome 1; \u003cem\u003eDYSF\u003c/em\u003e and \u003cem\u003eIL18R1\u003c/em\u003e on chromosome 2; \u003cem\u003eARG1\u003c/em\u003e and \u003cem\u003eVNN3P\u003c/em\u003e on chromosome 6; \u003cem\u003eABCA1\u003c/em\u003e and \u003cem\u003eANXA1\u003c/em\u003e on chromosome 9; \u003cem\u003eSIGLEC5\u003c/em\u003e, \u003cem\u003ePGLYRP1\u003c/em\u003e, and \u003cem\u003eCD177\u003c/em\u003e on chromosome 19; and \u003cem\u003eMMP9\u003c/em\u003e and \u003cem\u003eKCNJ15\u003c/em\u003e on chromosome 21 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). Manhattan plot analysis confirmed significant genomic enrichment for these genes, with \u003cem\u003eSLC22A4\u003c/em\u003e, \u003cem\u003eKCNJ15\u003c/em\u003e, \u003cem\u003eF5\u003c/em\u003e, \u003cem\u003eARG1\u003c/em\u003e, and \u003cem\u003eABCA1\u003c/em\u003e among the most prominent (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE), indicating potential diagnostic or mechanistic relevance in IS.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eImmune Infiltration Profiling in Ischemic Stroke\u003c/p\u003e\u003cp\u003eTo assess immune alterations in IS, ssGSEA was employed to estimate the infiltration of 28 immune cell types. The immune heatmap illustrated clear differences in immune cell abundance between IS and control groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). IS samples exhibited elevated levels of activated dendritic cells (DCs), macrophages, neutrophils, and Th17 cells. Violin plots confirmed that these differences were statistically significant (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Spearman correlation analysis between overlapping genes and immune cells revealed that \u003cem\u003eABCA1\u003c/em\u003e and \u003cem\u003eCLEC4E\u003c/em\u003e were positively correlated with neutrophils and plasmacytoid DCs; \u003cem\u003eARG1\u003c/em\u003e was significantly associated with neutrophils, macrophages, and activated DCs; \u003cem\u003eANXA1\u003c/em\u003e correlated positively with mast cells (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01) but negatively with CD56 bright NK cells and Th17 cells (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC). These findings suggest that the identified genes may play regulatory roles in the immune landscape of IS.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eMachine Learning Identifies Robust Hub Genes\u003c/p\u003e\u003cp\u003eThree complementary machine learning algorithms (SVM-RFE, LASSO, and RF) were applied to identify hub genes. LASSO regression selected 16 genes (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA, B), while SVM-RFE achieved optimal accuracy (95.4%) with 13 genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC), corresponding to the lowest cross-validation error (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD). The RF model stabilized at 100 trees and identified 8 important genes (importance score\u0026thinsp;\u0026gt;\u0026thinsp;5), with \u003cem\u003eANXA1\u003c/em\u003e, \u003cem\u003eABCA1\u003c/em\u003e, \u003cem\u003eSLC22A4\u003c/em\u003e, and \u003cem\u003eARG1\u003c/em\u003e scoring highest (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eE, F). Robust diagnostic biomarkers identified through machine learning are shown in Supplementary Table\u0026nbsp;3. A Venn diagram highlighted six hub genes (\u003cem\u003eKCNJ15\u003c/em\u003e, \u003cem\u003eARG1\u003c/em\u003e, \u003cem\u003eABCA1\u003c/em\u003e, \u003cem\u003eCLEC4E\u003c/em\u003e, \u003cem\u003eANXA1\u003c/em\u003e, and \u003cem\u003eMMP9\u003c/em\u003e) shared across all three methods (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eG), all of which were significantly upregulated in IS samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eH).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eConstruction and Validation of a Multi-Gene Diagnostic Model\u003c/p\u003e\u003cp\u003eROC curve analysis revealed high discriminatory power for each of the six hub genes, with AUCs ranging from 0.835 (\u003cem\u003eMMP9\u003c/em\u003e, \u003cem\u003eANXA1\u003c/em\u003e) to 0.914 (\u003cem\u003eARG1\u003c/em\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA). A logistic regression model incorporating all six genes yielded an AUC of 0.990 (95% Confidence Interval: 0.977\u0026ndash;0.998) in the training set (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB). Nomogram visualization illustrated the relative contribution of each gene (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eC), with calibration and decision curve analyses confirming the model's accuracy and clinical benefit (Figs.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eD, E). In the external validation set, none of the six genes showed significant differential expression individually, and their AUCs were all below 0.70 (Supplementary Fig.\u0026nbsp;1). However, the combined model retained moderate diagnostic power, with an AUC of 0.713 (95% Confidence Interval: 0.522\u0026ndash;0.873) (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eF).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eMolecular Subtype Identification and Immune Landscape Characterization\u003c/p\u003e\u003cp\u003eConsensus clustering based on the expression profiles of the six hub genes identified two stable molecular subtypes (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eA), with optimal cluster number determined as k\u0026thinsp;=\u0026thinsp;2 (Figs.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eB, C and Supplementary Table\u0026nbsp;4). Five genes (\u003cem\u003eKCNJ15\u003c/em\u003e, \u003cem\u003eARG1\u003c/em\u003e, \u003cem\u003eABCA1\u003c/em\u003e, \u003cem\u003eCLEC4E\u003c/em\u003e, \u003cem\u003eMMP9\u003c/em\u003e) showed significantly higher expression in subtype C2 compared to C1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eD). Heatmap analysis further validated these differences (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eE). Immune infiltration profiling revealed that subtype C2 exhibited higher abundance of activated DCs, macrophages, monocytes, NK cells, and plasmacytoid DCs, whereas subtype C1 showed enrichment of multiple B and T cell subpopulations, including memory B cells and effector CD4\u0026thinsp;+\u0026thinsp;T cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eF). PCA confirmed clear transcriptomic separation between C1 and C2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eG). GSVA-based pathway enrichment revealed functional divergence between subtypes. Subtype C2 was enriched in inflammatory and signaling pathways, such as MAPK signaling, Fc gamma R-mediated phagocytosis, Toll-like receptor signaling, and complement cascades. In contrast, subtype C1 was enriched in metabolic and mitochondrial pathways including oxidative phosphorylation, citrate cycle, and steroid biosynthesis (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eH).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eDrug Enrichment and Molecular Docking\u003c/p\u003e\u003cp\u003eDrug enrichment analysis using the DSigDB database identified several candidates, including Rosuvastatin, L-proline, and Diosgenin (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eA and Supplementary Table\u0026nbsp;5). Rosuvastatin was selected for further validation. Molecular docking was conducted between Rosuvastatin and three target proteins: ABCA1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eB), MMP9, and KCNJ15 (Supplementary Fig.\u0026nbsp;2). The binding affinities (vina scores) were \u0026minus;\u0026thinsp;8.4, -6.6, and \u0026minus;\u0026thinsp;6.8 kcal/mol, respectively, with the highest affinity observed for \u003cem\u003eABCA1\u003c/em\u003e. Structural analysis revealed that Rosuvastatin binds stably within the \u003cem\u003eABCA1\u003c/em\u003e active pocket, interacting with residues F819, E792, Y793, K1524, and N1523 via hydrogen bonding and hydrophobic contacts. π\u0026ndash;π stacking and polar interactions further stabilized the complex (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eB). A compound-gene interaction network constructed in Cytoscape illustrated potential multitarget effects of Rosuvastatin and other agents (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eC).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eIS is a highly heterogeneous neurovascular disorder involving complex pathophysiological processes such as inflammation, immune activation, apoptosis, and vascular remodeling(34). Recent studies have highlighted NETs as crucial factors in exacerbating post-stroke inflammation and facilitating secondary thrombosis, significantly influencing the extent of neuronal damage and clinical outcomes(35). However, existing studies on NETs in IS have primarily focused on single-gene functional validation, lacking a comprehensive understanding of their molecular networks and immunophenotypic characteristics.\u003c/p\u003e\u003cp\u003eIn this study, by integrating multiple GEO datasets and applying differential gene expression analysis, WGCNA, and three complementary machine learning algorithms, we identified six NETs-related hub genes: \u003cem\u003eKCNJ15\u003c/em\u003e, \u003cem\u003eARG1\u003c/em\u003e, \u003cem\u003eABCA1\u003c/em\u003e, \u003cem\u003eCLEC4E\u003c/em\u003e, \u003cem\u003eANXA1\u003c/em\u003e, and \u003cem\u003eMMP9\u003c/em\u003e. These genes were significantly upregulated in IS patients and exhibited strong associations with the immune microenvironment, suggesting pivotal regulatory roles in inflammation-coagulation interplay during stroke progression. A diagnostic model constructed from these six genes showed excellent discriminative performance in the training cohort (AUC\u0026thinsp;=\u0026thinsp;0.990). Although none of the individual genes reached statistical significance in the independent validation cohort (n\u0026thinsp;=\u0026thinsp;36, P\u0026thinsp;\u0026gt;\u0026thinsp;0.05, AUC\u0026thinsp;\u0026lt;\u0026thinsp;0.7), the multi-gene model still demonstrated moderate discriminatory power (AUC\u0026thinsp;=\u0026thinsp;0.713), indicating favorable robustness and potential for clinical translation. It is worth noting that the limited sample size of the validation cohort may have reduced the statistical power for assessing individual biomarkers. Larger clinical cohorts are thus warranted to further evaluate the diagnostic utility and generalizability of the identified gene signature. These findings also highlight that for highly immune-heterogeneous diseases like IS, multi-gene models may offer greater clinical utility than single biomarkers.\u003c/p\u003e\u003cp\u003eFunctionally, the six hub genes contribute to NETs formation, inflammation regulation, or neurovascular protection via diverse mechanisms. ARG1, an arginase enzyme, modulates immune responses during the acute phase of stroke by competitively inhibiting iNOS-mediated NO synthesis, whereas its co-expression with M2 macrophage markers and IL-10 in the recovery phase suggests dual immunomodulatory roles(36). Although elevated ARG1 levels correlate with stroke severity and immune suppression(37), direct evidence for its involvement in NETs formation remains lacking. ANXA1, a known pro-resolving mediator(38), has been shown to suppress NETs release and protect the BBB via the FPR2 pathway during post-stroke resolution(39). ABCA1 facilitates cholesterol efflux and maintains BBB integrity(40), and its deficiency exacerbates white matter injury and increases IS risk(41). KCNJ15, a potassium channel protein(42), has been proposed as a biomarker for atherosclerotic cerebral infarction(43). It may contribute indirectly to neutrophil activation by modulating membrane potential; however, its direct involvement in the formation of NETs remains to be experimentally confirmed. CLEC4E, a pattern recognition receptor, is functionally linked to innate immunity and NETs release(44). MMP9 is a critical protease that facilitates neutrophil migration and extracellular matrix degradation, contributing to BBB disruption and hemorrhagic transformation in stroke. It may act synergistically with NETs to amplify inflammation and thrombosis(45). These results enhance our understanding of the inflammation-coagulation axis in IS and underscore the potential of NETs-related molecular markers.\u003c/p\u003e\u003cp\u003eUsing the expression patterns of the six hub genes, we conducted NETs-related immunophenotypic subtyping for the first time in IS. Two molecular subtypes were identified: C1 and C2. Subtype C2 exhibited higher expression of KCNJ15, ARG1, ABCA1, CLEC4E, and MMP9, with enrichment in pathways such as MAPK signaling, complement cascade, Toll-like receptor signaling and coagulation pathways. This subtype also showed prominent infiltration of innate immune cells (neutrophils, macrophages, monocytes), reflecting a \u0026ldquo;NETs-high, immune-activated\u0026rdquo; phenotype, which is closely linked to heightened inflammation and thrombotic risk after stroke(46, 47). In contrast, subtype C1 was characterized by enrichment of metabolic pathways (e.g., oxidative phosphorylation, lipid metabolism) and higher abundance of adaptive immune cells (memory B/T cells), representing a \u0026ldquo;metabolically regulated, immune-modulated\u0026rdquo; phenotype. Notably, these subtypes do not completely align with the traditional N1/N2 neutrophil classification(48), suggesting that neutrophils in stroke may exhibit more complex and plastic immunophenotypes, shaped by disease stage and the local immune-metabolic environment. For instance, metabolic comorbidities such as diabetes can skew neutrophils toward a pro-inflammatory N1 phenotype, enhance NETs formation, and exacerbate neuroinflammation and tissue injury after stroke(49). This immunophenotypic classification may provide a framework for risk stratification and tailored therapy. From a clinical standpoint, subtype C2 patients may have higher risk for hemorrhagic transformation and re-occlusion, particularly following thrombolytic therapy. For these patients, early NET-targeted interventions such as PAD4 inhibitors(50), FPR2 agonists(51), or nNIF peptides(52) may help mitigate excessive inflammation and thrombotic complications. Meanwhile, C1 subtype patients may benefit more from antioxidant therapy, BBB stabilizers, or neuroprotective agents. These findings support the rationale for immune subtype-guided therapy in IS.\u003c/p\u003e\u003cp\u003eTo explore therapeutic implications further, we performed drug enrichment analysis and molecular docking simulations, identifying rosuvastatin as a promising candidate for clinical repurposing. Rosuvastatin, a selective HMG-CoA reductase inhibitor, is commonly prescribed for preventing both primary and secondary atherosclerotic cardiovascular events, and it exerts a range of pleiotropic effects, including lipid-lowering, anti-inflammatory, plaque stabilization, and endothelial protection(53). Our docking analysis revealed that rosuvastatin binds stably to several NETs-related proteins (ABCA1, MMP9, and KCNJ15), with a binding affinity of -8.4 kcal/mol for ABCA1. Structural modeling suggested that rosuvastatin occupies the transporter\u0026rsquo;s active pocket and may enhance ABCA1 function by promoting cholesterol efflux and modulating miR-33b-5p expression(54). While its effect on NETs formation remains to be clarified, previous studies indicate that rosuvastatin attenuates BBB damage and hemorrhagic transformation in IS by inhibiting MMP9 via PDGFR-α/LRP1-MAPK signaling(55). While its direct impact on NETs formation requires further investigation, these mechanistic insights and binding data provide a foundation for future therapeutic exploration.\u003c/p\u003e\u003cp\u003eIn conclusion, we established a comprehensive framework encompassing NETs-related gene identification, diagnostic modeling, immune subtype classification, and therapeutic exploration in IS. This study is the first to delineate immunologically distinct molecular subtypes of stroke from the perspective of NETs, highlighting the central role of neutrophils in the inflammation-thrombosis axis. Despite the enhanced robustness from multi-omics integration and cross-validation, limitations such as small sample size, lack of single-cell/protein-level verification, and the need for experimental validation of drug-target interactions remain. Future studies should integrate dynamic time-course analyses, animal models, and functional assays to elucidate the mechanistic roles of key regulators such as KCNJ15 and ARG1, and develop immune-subtype-based precision therapies to support individualized stroke management and improve clinical outcomes. These findings may inform immune subtype-guided treatment decisions in ischemic stroke patients, facilitating early intervention and risk stratification in clinical settings.\u003c/p\u003e\u003cp\u003eThis study has a few limitations. Firstly, the relatively small sample size of the external validation cohort may have restricted the statistical power to identify significant differences at the individual gene level. Second, the study relied primarily on transcriptomic data, without confirmation from single-cell sequencing, proteomics, or in vivo functional validation. Third, although we utilized cross-validation and testing with external datasets to validate the model's robustness, prospective clinical trials and mechanistic studies are crucial to verify the diagnostic and therapeutic potential of the identified gene signature.\u003c/p\u003e"},{"header":"5 Conclusion","content":"\u003cp\u003eOur findings offer a novel framework for NETs-related biomarker identification, molecular subtyping, and precision therapeutic exploration in ischemic stroke. The proposed immune subtypes, based on neutrophil-driven transcriptomic features, may facilitate individualized treatment planning and early risk assessment. Moreover, the identification of actionable targets such as MMP9 and ABCA1 opens avenues for drug repurposing strategies in stroke management. Future studies should prioritize validation in larger and more diverse clinical cohorts, integration with longitudinal immune profiling, and experimental elucidation of the roles of key regulators like \u003cem\u003eKCNJ15\u003c/em\u003e and \u003cem\u003eARG1\u003c/em\u003e. These efforts could ultimately inform immune-based personalized interventions for improving stroke outcomes.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003earea under the curve\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eBBB\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eblood-brain barrier\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eBP\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ebiological processes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ecellular components\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCDF\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ecumulative distribution function\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDCA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003edecision curve analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDCs\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003edendritic cells\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDEGs\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003edifferentially expressed genes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eFDR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003efalse discovery rate\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eFC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003efold change\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGEO\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGene Expression Omnibus\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGO\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGene Ontology\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGSVA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003egene set variation analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eIS\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eischemic stroke\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eKEGG\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eKyoto Encyclopedia of Genes and Genomes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eLASSO\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eleast absolute shrinkage and selection operator\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eMF\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003emolecular functions\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eNETs\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eneutrophil extracellular traps\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003ePCA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePrincipal component analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eRF\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRandom Forest\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003essGSEA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003esingle-sample gene set enrichment analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSVM-RFE\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003esupport vector machine-recursive feature elimination\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eROC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ereceiver operating characteristic\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eTOM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003etopological overlap matrix\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eWGCNA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eWeighted Gene Co-expression Network Analysis.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eNot applicable. The study was based on publicly available datasets, and no human or animal experiments were conducted.\u003c/p\u003e\n\u003cp\u003eConsent for publication\u003c/p\u003e\n\u003cp\u003eNot applicable. No individual patient data are included in this manuscript.\u003c/p\u003e\n\u003cp\u003eAvailability of data and material\u003c/p\u003e\n\u003cp\u003eAll data and findings generated during this study are available within the article and its supplementary materials. For any additional information, readers are encouraged to contact the corresponding author.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.\u003c/p\u003e\n\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThis study was supported by the Yunnan Science and Technology Program (Grant No. 202401AT070176).\u003c/p\u003e\n\u003cp\u003eAuthors\u0026apos; contributions\u003c/p\u003e\n\u003cp\u003eXL: Conceptualization, Methodology, Formal analysis, Visualization, Writing-original draft.\u003c/p\u003e\n\u003cp\u003eQL: Data curation, Software, Validation, Funding acquisition.\u003c/p\u003e\n\u003cp\u003eYS: Formal analysis, Software, Visualization.\u003c/p\u003e\n\u003cp\u003eGF: Investigation, Resources, Data curation.\u003c/p\u003e\n\u003cp\u003eYH: Resources, Supervision. SQ: Formal analysis, Software.\u003c/p\u003e\n\u003cp\u003eLZ: Conceptualization, Supervision, Writing-review \u0026amp; editing, Project administration.\u003c/p\u003e\n\u003cp\u003eAll authors (XL, QL, YS, GF, YH, SQ, and LZ) reviewed and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eThe authors express their sincere gratitude to all individuals who took part in this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eQin C, Yang S, Chu YH, Zhang H, Pang XW, Chen L, et al. Signaling pathways involved in ischemic stroke: molecular mechanisms and therapeutic interventions. Signal Transduct Target Ther. 2022;7(1):215. 10.1038/s41392-022-01064-1.\u003c/li\u003e\n\u003cli\u003eFeigin VL, Brainin M, Norrving B, Martins SO, Pandian J, Lindsay P, et al. World Stroke Organization: Global Stroke Fact Sheet 2025. Int J Stroke. 2025;20(2):132-44. 10.1177/17474930241308142.\u003c/li\u003e\n\u003cli\u003eZhu W, He X, Huang D, Jiang Y, Hong W, Ke S, et al. Global and Regional Burden of Ischemic Stroke Disease from 1990 to 2021: An Age-Period-Cohort Analysis. Transl Stroke Res. 2024. 10.1007/s12975-024-01319-9.\u003c/li\u003e\n\u003cli\u003eMajumder D. Ischemic Stroke: Pathophysiology and Evolving Treatment Approaches. Neurosci Insights. 2024;19:26331055241292600. 10.1177/26331055241292600.\u003c/li\u003e\n\u003cli\u003eWang H, Zhang S, Xie L, Zhong Z, Yan F. Neuroinflammation and peripheral immunity: Focus on ischemic stroke. International Immunopharmacology. 2023;120:110332. 10.1016/j.intimp.2023.110332.\u003c/li\u003e\n\u003cli\u003eHurd MD, Goel I, Sakai Y, Teramura Y. Current status of ischemic stroke treatment: From thrombolysis to potential regenerative medicine. Regenerative Therapy. 2021;18:408-17. 10.1016/j.reth.2021.09.009.\u003c/li\u003e\n\u003cli\u003eMosconi MG, Paciaroni M. Treatments in Ischemic Stroke: Current and Future. Eur Neurol. 2022;85(5):349-66. 10.1159/000525822.\u003c/li\u003e\n\u003cli\u003eGong Z, Guo J, Liu B, Guo Y, Cheng C, Jiang Y, et al. Mechanisms of immune response and cell death in ischemic stroke and their regulation by natural compounds. Front Immunol. 2023;14:1287857. 10.3389/fimmu.2023.1287857.\u003c/li\u003e\n\u003cli\u003eLuo H, Guo H, Zhou Y, Fang R, Zhang W, Mei Z. Neutrophil Extracellular Traps in Cerebral Ischemia/Reperfusion Injury: Friend and Foe. Curr Neuropharmacol. 2023;21(10):2079-96. 10.2174/1570159x21666230308090351.\u003c/li\u003e\n\u003cli\u003eJi YM, Li T, Qin YH, Xiao SY, Lv YH, Dong Y, et al. Neutrophil Extracellular Traps (NETs) in Sterile Inflammatory Diseases. J Inflamm Res. 2025;18:7989-8004. 10.2147/jir.S526936.\u003c/li\u003e\n\u003cli\u003eLiaptsi E, Merkouris E, Polatidou E, Tsiptsios D, Gkantzios A, Kokkotis C, et al. Targeting Neutrophil Extracellular Traps for Stroke Prognosis: A Promising Path. Neurol Int. 2023;15(4):1212-26. 10.3390/neurolint15040076.\u003c/li\u003e\n\u003cli\u003eGu X, Dong M, Xia S, Li H, Bao X, Cao X, et al. \u0026gamma;-Glutamylcysteine ameliorates blood-brain barrier permeability and neutrophil extracellular traps formation after ischemic stroke by modulating Wnt/\u0026beta;-catenin signalling in mice. Eur J Pharmacol. 2024;969:176409. 10.1016/j.ejphar.2024.176409.\u003c/li\u003e\n\u003cli\u003eDenorme F, Portier I, Rustad JL, Cody MJ, de Araujo CV, Hoki C, et al. Neutrophil extracellular traps regulate ischemic stroke brain injury. J Clin Invest. 2022;132(10). 10.1172/jci154225.\u003c/li\u003e\n\u003cli\u003eWang R, Zhu Y, Liu Z, Chang L, Bai X, Kang L, et al. Neutrophil extracellular traps promote tPA-induced brain hemorrhage via cGAS in mice with stroke. Blood. 2021;138(1):91-103. 10.1182/blood.2020008913.\u003c/li\u003e\n\u003cli\u003eGao X, Zhao X, Li J, Liu C, Li W, Zhao J, et al. Neutrophil extracellular traps mediated by platelet microvesicles promote thrombosis and brain injury in acute ischemic stroke. Cell Commun Signal. 2024;22(1):50. 10.1186/s12964-023-01379-8.\u003c/li\u003e\n\u003cli\u003eDhanesha N, Ansari J, Pandey N, Kaur H, Virk C, Stokes KY. Poststroke venous thromboembolism and neutrophil activation: an illustrated review. Res Pract Thromb Haemost. 2023;7(4):100170. 10.1016/j.rpth.2023.100170.\u003c/li\u003e\n\u003cli\u003eZhao Z, Pan Z, Zhang S, Ma G, Zhang W, Song J, et al. Neutrophil extracellular traps: A novel target for the treatment of stroke. Pharmacol Ther. 2023;241:108328. 10.1016/j.pharmthera.2022.108328.\u003c/li\u003e\n\u003cli\u003eLi J, Liu L, Zhang R, Pan L, Tan J, Ou M, et al. Associations of NETs with inflammatory risk and clinical predictive value in large artery atherosclerosis stroke: a prospective cohort study. Front Immunol. 2024;15:1488317. 10.3389/fimmu.2024.1488317.\u003c/li\u003e\n\u003cli\u003eWu ZR, Zhou TQ, Ai SC. Neutrophil extracellular traps correlate with severity and prognosis in patients with ischemic stroke: a systematic review and meta-analysis. Acta Neurol Belg. 2024;124(2):513-22. 10.1007/s13760-023-02409-5.\u003c/li\u003e\n\u003cli\u003eNovotny J, Oberdieck P, Titova A, Pelisek J, Chandraratne S, Nicol P, et al. Thrombus NET content is associated with clinical outcome in stroke and myocardial infarction. Neurology. 2020;94(22):e2346-e60. 10.1212/wnl.0000000000009532.\u003c/li\u003e\n\u003cli\u003eOh SA, Seol SI, Davaanyam D, Kim SW, Lee JK. Platelet-derived HMGB1 induces NETosis, exacerbating brain damage in the photothrombotic stroke model. Mol Med. 2025;31(1):46. 10.1186/s10020-025-01107-7.\u003c/li\u003e\n\u003cli\u003eHuang T, Guo Y, Xie W, Yin J, Zhang Y, Chen W, et al. Brain border-derived CXCL2(+) neutrophils drive NET formation and impair vascular reperfusion following ischemic stroke. CNS Neurosci Ther. 2024;30(8):e14916. 10.1111/cns.14916.\u003c/li\u003e\n\u003cli\u003eFang H, Bo Y, Hao Z, Mang G, Jin J, Wang H. A promising frontier: targeting NETs for stroke treatment breakthroughs. Cell Commun Signal. 2024;22(1):238. 10.1186/s12964-024-01563-4.\u003c/li\u003e\n\u003cli\u003eVogelgesang A, Lange C, Bl\u0026uuml;mke L, Laage G, R\u0026uuml;mpel S, Langner S, et al. Ischaemic stroke and the recanalization drug tissue plasminogen activator interfere with antibacterial phagocyte function. J Neuroinflammation. 2017;14(1):140. 10.1186/s12974-017-0914-6.\u003c/li\u003e\n\u003cli\u003eDenorme F, Portier I, Cody M, Grandhi R, Neal MD, Majersik JJ, et al. Platelet-Mediated NET Formation Exacerbates Ischemic Stroke Brain Injury. Blood. 2021;138:437. 10.1182/blood-2021-151423.\u003c/li\u003e\n\u003cli\u003eQin R, Xu W, Qin Q, Liang X, Lai X, Xie M, et al. Identification of NETs-related genes as diagnostic biomarkers in ischemic stroke using RNA sequencing and single-cell analysis. Mamm Genome. 2025;36(2):651-64. 10.1007/s00335-025-10117-z.\u003c/li\u003e\n\u003cli\u003eLiu Y, Wang W, Cui X, Lyu J, Xie Y. Exploring Genetic Associations of 3 Types of Risk Factors With Ischemic Stroke: An Integrated Bioinformatics Study. Stroke. 2024;55(6):1619-28. 10.1161/strokeaha.123.044424.\u003c/li\u003e\n\u003cli\u003eLi T, Kang X, Zhang S, Wang Y, He J, Li H, et al. Integrating machine learning and multi-omics analysis to reveal nucleotide metabolism-related immune genes and their functional validation in ischemic stroke. Front Immunol. 2025;16:1561544. 10.3389/fimmu.2025.1561544.\u003c/li\u003e\n\u003cli\u003eYang WX, Wang FF, Pan YY, Xie JQ, Lu MH, You CG. Comparison of ischemic stroke diagnosis models based on machine learning. Front Neurol. 2022;13:1014346. 10.3389/fneur.2022.1014346.\u003c/li\u003e\n\u003cli\u003eLi RB, Yang XH, Zhang JD, Cui W. GAS6-AS1, a long noncoding RNA, functions as a key candidate gene in atrial fibrillation related stroke determined by ceRNA network analysis and WGCNA. BMC Med Genomics. 2023;16(1):51. 10.1186/s12920-023-01478-y.\u003c/li\u003e\n\u003cli\u003eWang Y, Liang S, Hong Q, Mu J, Wu Y, Li K, et al. Construction of a neutrophil extracellular trap formation-related gene model for predicting the survival of lung adenocarcinoma patients and their response to immunotherapy. Transl Lung Cancer Res. 2024;13(12):3407-25. 10.21037/tlcr-24-463.\u003c/li\u003e\n\u003cli\u003eShi H, Pan Y, Xiang G, Wang M, Huang Y, He L, et al. A novel NET-related gene signature for predicting DLBCL prognosis. J Transl Med. 2023;21(1):630. 10.1186/s12967-023-04494-9.\u003c/li\u003e\n\u003cli\u003eZhang Y, Guo L, Dai Q, Shang B, Xiao T, Di X, et al. A signature for pan-cancer prognosis based on neutrophil extracellular traps. J Immunother Cancer. 2022;10(6). 10.1136/jitc-2021-004210.\u003c/li\u003e\n\u003cli\u003eSalaudeen MA, Bello N, Danraka RN, Ammani ML. Understanding the Pathophysiology of Ischemic Stroke: The Basis of Current Therapies and Opportunity for New Ones. Biomolecules. 2024;14(3). 10.3390/biom14030305.\u003c/li\u003e\n\u003cli\u003eLapostolle A, Loyer C, Elhorany M, Chaigneau T, Bielle F, Alamowitch S, et al. Neutrophil Extracellular Traps in Ischemic Stroke Thrombi Are Associated Wth Poor Clinical Outcome. Stroke: Vascular and Interventional Neurology. 2023;3(3):e000639. doi:10.1161/SVIN.122.000639.\u003c/li\u003e\n\u003cli\u003eWanrooy BJ, Wen SW, Wong CH. Dynamic roles of neutrophils in post-stroke neuroinflammation. Immunol Cell Biol. 2021;99(9):924-35. 10.1111/imcb.12463.\u003c/li\u003e\n\u003cli\u003ePetrone AB, O\u0026apos;Connell GC, Regier MD, Chantler PD, Simpkins JW, Barr TL. The Role of Arginase 1 in Post-Stroke Immunosuppression and Ischemic Stroke Severity. Transl Stroke Res. 2016;7(2):103-10. 10.1007/s12975-015-0431-9.\u003c/li\u003e\n\u003cli\u003eTang C, Lan R, Ma DR, Zhao M, Zhang Y, Li HY, et al. Annexin A1: The dawn of ischemic stroke (Review). Mol Med Rep. 2025;31(3). 10.3892/mmr.2024.13427.\u003c/li\u003e\n\u003cli\u003eAnsari J, Gavins FNE. Neutrophils and Platelets: Immune Soldiers Fighting Together in Stroke Pathophysiology. Biomedicines. 2021;9(12). 10.3390/biomedicines9121945.\u003c/li\u003e\n\u003cli\u003eCui X, Chopp M, Zacharek A, Karasinska JM, Cui Y, Ning R, et al. Deficiency of brain ATP-binding cassette transporter A-1 exacerbates blood-brain barrier and white matter damage after stroke. Stroke. 2015;46(3):827-34. 10.1161/strokeaha.114.007145.\u003c/li\u003e\n\u003cli\u003eAu A, Griffiths LR, Irene L, Kooi CW, Wei LK. The impact of APOA5, APOB, APOC3 and ABCA1 gene polymorphisms on ischemic stroke: Evidence from a meta-analysis. Atherosclerosis. 2017;265:60-70. 10.1016/j.atherosclerosis.2017.08.003.\u003c/li\u003e\n\u003cli\u003eWang T, Kim DH, Ding C, Wang D, Zhang W, Silic M, et al. Inwardly rectifying potassium channels regulate membrane potential polarization and direction sensing during neutrophil chemotaxis. bioRxiv. 2025. 10.1101/2025.03.06.641746.\u003c/li\u003e\n\u003cli\u003eZhang B, Huang W, Yi M, Xing C. Gene Differential Expression and Interaction Networks Illustrate the Biomarkers and Molecular Mechanisms of Atherosclerotic Cerebral Infarction. J Healthc Eng. 2022;2022:3912697. 10.1155/2022/3912697.\u003c/li\u003e\n\u003cli\u003eWang X, Liu X. Exploration of the shared gene signatures and molecular mechanisms between cardioembolic stroke and ischemic stroke. Front Neurol. 2025;16:1567902. 10.3389/fneur.2025.1567902.\u003c/li\u003e\n\u003cli\u003eTurner RJ, Sharp FR. Implications of MMP9 for Blood Brain Barrier Disruption and Hemorrhagic Transformation Following Ischemic Stroke. Front Cell Neurosci. 2016;10:56. 10.3389/fncel.2016.00056.\u003c/li\u003e\n\u003cli\u003eChen R, Zhang X, Gu L, Zhu H, Zhong Y, Ye Y, et al. New Insight Into Neutrophils: A Potential Therapeutic Target for Cerebral Ischemia. Front Immunol. 2021;12:692061. 10.3389/fimmu.2021.692061.\u003c/li\u003e\n\u003cli\u003eJu H, Park KW, Kim ID, Cave JW, Cho S. Phagocytosis converts infiltrated monocytes to microglia-like phenotype in experimental brain ischemia. J Neuroinflammation. 2022;19(1):190. 10.1186/s12974-022-02552-5.\u003c/li\u003e\n\u003cli\u003eXie M, Hao Y, Feng L, Wang T, Yao M, Li H, et al. Neutrophil Heterogeneity and its Roles in the Inflammatory Network after Ischemic Stroke. Curr Neuropharmacol. 2023;21(3):621-50. 10.2174/1570159x20666220706115957.\u003c/li\u003e\n\u003cli\u003eLiu S, Cai W, Hu M, Lu Z. Abstract WP317: Neutrophil Subtypes and Nets Changes in Diabetes With Acute Ischemic Stroke. Stroke.51(Suppl_1):AWP317-AWP. 10.1161/str.51.suppl_1.WP317.\u003c/li\u003e\n\u003cli\u003eKang L, Yu H, Yang X, Zhu Y, Bai X, Wang R, et al. Neutrophil extracellular traps released by neutrophils impair revascularization and vascular remodeling after stroke. Nat Commun. 2020;11(1):2488. 10.1038/s41467-020-16191-y.\u003c/li\u003e\n\u003cli\u003eZhu J, Li L, Ding J, Huang J, Shao A, Tang B. The Role of Formyl Peptide Receptors in Neurological Diseases via Regulating Inflammation. Front Cell Neurosci. 2021;15:753832. 10.3389/fncel.2021.753832.\u003c/li\u003e\n\u003cli\u003eBircher JS, Denorme F, Cody MJ, de Araujo CV, Petrey AC, Middleton EA, et al. Neonatal NET-inhibitory factor inhibits macrophage extracellular trap formation. Blood Advances. 2024;8(14):3686-90. 10.1182/bloodadvances.2024013094.\u003c/li\u003e\n\u003cli\u003eMehta A, Jain P, Patil R, Sashi Kant T, Indurkar SA, Kota SK, et al. Real-World Clinical Experience of Rosuvastatin as a Lipid-Lowering Therapy for Primary and Secondary Prevention of Cardiovascular Events (REAL ROSE). Cureus. 2022;14(11):e31468. 10.7759/cureus.31468.\u003c/li\u003e\n\u003cli\u003eSantovito D, Marcantonio P, Mastroiacovo D, Natarelli L, Mandolini C, De Nardis V, et al. High dose rosuvastatin increases ABCA1 transporter in human atherosclerotic plaques in a cholesterol-independent fashion. Int J Cardiol. 2020;299:249-53. 10.1016/j.ijcard.2019.07.094.\u003c/li\u003e\n\u003cli\u003eLu D, Mai HC, Liang YB, Xu BD, Xu AD, Zhang YS. Beneficial Role of Rosuvastatin in Blood-Brain Barrier Damage Following Experimental Ischemic Stroke. Front Pharmacol. 2018;9:926. 10.3389/fphar.2018.00926.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-neurology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"nurl","sideBox":"Learn more about [BMC Neurology](http://bmcneurol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/nurl","title":"BMC Neurology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"ischemic stroke, neutrophil extracellular traps, immune subtypes, machine learning, biomarker discovery, drug repositioning, molecular stratification","lastPublishedDoi":"10.21203/rs.3.rs-7515929/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7515929/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjective\u003c/h2\u003e\u003cp\u003eTo investigate the involvement of neutrophil extracellular trap (NET)-associated genes in ischemic stroke (IS) and create a diagnostic model for precision stroke treatment.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eTranscriptomic datasets from GEO were integrated to identify NETs-related gene signatures in ischemic stroke. We conducted differential expression analysis and Weighted Gene Co-expression Network Analysis (WGCNA), followed by biomarker identification using support vector machine-recursive feature elimination (SVM-RFE), least absolute shrinkage and selection operator (LASSO), and Random Forest (RF) algorithms. A multigene diagnostic model was developed and validated. Immune subtypes were defined via consensus clustering based on hub genes. Immune infiltration, functional enrichment, drug-gene interaction analysis, and molecular docking were performed to identify therapeutic targets specific to subtypes.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eSix hub genes (\u003cem\u003eKCNJ15\u003c/em\u003e, \u003cem\u003eARG1\u003c/em\u003e, \u003cem\u003eCLEC4E\u003c/em\u003e, \u003cem\u003eABCA1\u003c/em\u003e, \u003cem\u003eANXA1\u003c/em\u003e, and \u003cem\u003eMMP9\u003c/em\u003e) were recognized as promising biomarkers for diagnosis with excellent performance (AUC\u0026thinsp;=\u0026thinsp;0.990). Two immune subtypes of IS were revealed, characterized by distinct metabolic activity, immune cell infiltration, and proinflammatory signaling. Functional analysis confirmed significant immunometabolic divergence between the subtypes. Rosuvastatin was identified as a potential therapeutic agent targeting ABCA1, MMP9, and KCNJ15, suggesting subtype-specific therapeutic effects.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eOur study provides a novel framework for immune subtyping and targeted therapy in IS, demonstrating the diagnostic and therapeutic potential of NETs-related biomarkers. These findings offer promising implications for precision stroke management.\u003c/p\u003e","manuscriptTitle":"Immune subtypes and diagnostic genes revealed by neutrophil trap-associated transcriptomic signatures in ischemic stroke","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-22 01:34:04","doi":"10.21203/rs.3.rs-7515929/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2025-10-20T10:27:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"266432315908145123612503357403206624105","date":"2025-10-19T16:54:56+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-10-08T17:03:57+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-11T17:08:39+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-04T07:36:24+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-04T07:35:06+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Neurology","date":"2025-09-02T09:11:58+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-neurology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"nurl","sideBox":"Learn more about [BMC Neurology](http://bmcneurol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/nurl","title":"BMC Neurology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6c2c1e04-4d53-41a8-a11e-60e26c6c41f1","owner":[],"postedDate":"October 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-10-22T01:34:05+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-22 01:34:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7515929","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7515929","identity":"rs-7515929","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}