Identification and validation of glycosaminoglycan metabolism-associated immune biomarkers in patients with ischemic stroke based on weighted gene co-expression network analysis and machine learning

Abstract

Background Acute ischemic stroke (IS) represents a cerebrovascular emergency characterized by abrupt cessation of cerebral perfusion and subsequent systemic immune activation. Accumulating evidence suggests that glycosaminoglycan (GAG) metabolic pathways critically modulate neuroinflammatory cascades and vascular dysfunction. However, the role of GAG metabolism in generating peripheral molecular signatures and translatable diagnostic indicators remains poorly characterized. This investigation sought to identify GAG metabolism–associated immune biomarkers and decode their immunometabolic mechanisms in IS. Methods We integrated two transcriptomic repositories (GSE58294 and GSE16561) and compiled 127 GAG metabolism–related genes from published literature. Weighted gene co-expression network analysis was used to extract GAG-associated modules. Machine-learning algorithms combined with differential expression validation prioritized candidate biomarkers, and diagnostic performance was assessed via nomogram construction. Mechanistic insights were obtained through gene set enrichment analysis and computational immune deconvolution, while pharmacological databases identified potential therapeutic compounds. Biomarker expression was experimentally validated using RT-qPCR in peripheral blood samples obtained from 5 IS patients (3 males and 2 females; mean age 67.6 years, range 55–77 years) and 5 healthy controls (2 males and 3 females; mean age 68.6 years, range 40–88 years). Blood samples from IS patients were collected 3–7 days after stroke onset, whereas control samples were obtained on the second day after hospital admission. Results Three biomarkers, MCEMP1, PRRG4, and VSIG4, demonstrated concordant transcriptional elevation in IS samples. The constructed nomogram exhibited robust discriminatory accuracy, confirming its clinical utility. Pathway enrichment revealed associations with ribosomal biogenesis, neutrophil extracellular trap formation, and neuroactive ligand–receptor signaling, indicating roles in translational regulation and innate immunity. Immune profiling disclosed substantial alterations in neutrophils, M0 macrophages, CD8+ T cells, and eosinophils, with strong correlations between biomarkers and neutrophil-driven inflammation. Drug–gene interaction analysis identified trichostatin A as a shared molecular target. RT-qPCR confirmed a significant upregulation of MCEMP1 and PRRG4. Conclusion MCEMP1, PRRG4, and VSIG4 constitute reliable molecular indicators that link systemic GAG metabolic dysregulation to IS-associated immune reorganization. These findings advance our mechanistic understanding of GAG-mediated inflammatory networks in stroke and establish actionable targets for both clinical risk stratification and therapeutic development.