Multi-omics integration reveals hypoxia-driven mechanisms in vascular dementia: a machine learning and single-cell sequencing approach.

Background: This study aimed to identify and analyze hub hypoxia-related genes in vascular dementia (VaD) and explore their roles in metabolism, immune response, and cell differentiation, thereby offering potential biomarkers and therapeutic targets. Methods: Using VaD datasets (GSE122063, GSE282111) from Gene Expression Omnibus (including high-throughput and single-cell sequencing data), analyses were performed via R preprocessing, WGCNA, and machine learning. A chronic cerebral hypoperfusion model was established by two-vessel occlusion (2VO), with verification through immunohistochemistry. Results: WGCNA identified 7451 module genes and 36 overlapping hypoxia-related genes; machine learning pinpointed DUSP1, MAFF, and TGFBI as hub genes. ssGSEA linked these genes to metabolic pathways (e.g., cysteine-methionine metabolism, glycolysis) and cell death pathways (apoptosis, pyroptosis). They associated with immune cells like M2 macrophages and neutrophils. Single-cell analysis showed their expression in astrocytes, endothelial cells, and microglia, with endothelial cells exhibiting a hypoxic phenotype via pathways like PI3K-Akt. Immunohistochemistry revealed increased DUSP1, MAFF and TGFBI in models. Conclusions: This study identified DUSP1, MAFF, and TGFBI as key players in hypoxia-related mechanisms in VaD, highlighting their pivotal roles in metabolic regulation, cell death pathways, immune microenvironment modulation, and neural differentiation. These insights enhance our understanding of VaD pathogenesis and suggest that these genes may be potential therapeutic targets for cognitive impairment.