Vascular and microvascular dysfunction in the human heart and brain: investigating the underlying pathways using an AI-based approach

Vascular and microvascular dysfunction contribute to both cardiovascular and cerebrovascular diseases, but the shared molecular pathways linking these conditions remain partially understood. Here we applied an artificial-intelligence (AI)–assisted literature-mining pipeline to identify shared genes and pathways underlying vascular and microvascular dysfunction across cardiovascular and cerebrovascular disorders. PubMed records from January 2000 to November 2024 were systematically queried for genomic studies of vascular and microvascular disease using the Entrez API, with stringent MeSH and text-based filtering. A large-language-model assisted extraction pipeline extracted and standardized gene associations, which were then analysed using Reactome pathway enrichment, gene-set intersection, and network analysis. To reduce candidate-gene bias, the analysis was restricted to large-scale genomic study designs, including GWAS, WES, and WGS. Among 690 genes associated with vascular disorders (398 cardiovascular, 292 cerebrovascular), 24 were shared across both categories, suggesting a limited shared molecular component alongside substantial system-specific heterogeneity. Enriched pathways included inflammatory and metabolic processes, NOTCH, VEGF, and TGF-β signaling, immune modulation, lipid transport, and cellular senescence). Network analysis revealed a tightly interconnected core, with APOE, NOTCH3, EPHB4, MYLK, ZFHX3, and HDAC9 emerging as high-degree hubs. As many included studies relate to macrovascular phenotypes, the microvascular relevance of these shared genes is inferred and requires further validation. Our results are consistent with the presence of a restricted cross-organ vascular gene set and provide an AI-assisted framework for large-scale integration and prioritization of genomic evidence across organ systems. Experimental validation and further orthogonal validation is warranted.