Dissection of spatial hypoxic and inflammatory ecosystem in glioblastoma

The spatial organization of the glioblastoma (GBM) microenvironment is a critical determinant of tumor progression and therapeutic resistance. Although cellular heterogeneity and microenvironmental stressors such as hypoxia and inflammation are well recognized, their coordinated spatial organization and functional interplay remain incompletely understood. Here, we developed Spatial PHenotype Ecosystem RElationship (SPHERE), a computational framework that quantitatively decodes the spatial traits of the GBM ecosystem by integrating single-cell-resolution spatial transcriptomic data from a large patient cohort. We identify hypoxic and inflammatory niches as spatially distinct microenvironment states that are frequently peri-necrotic but exhibit a degree of partial overlap or spatial adjacency. Mesenchymal-like tumor cells showed strong spatial coupling to both hypoxic and inflammatory regions, whereas glial-lineage cells preferentially localize to peripheral areas. Immune checkpoint molecules, including LGALS3, HMGB1, CD47, etc., are significantly enriched within hypoxic zones. In contrast, inflammatory regions are characterized by elevated chemokines and cytokines, highlighted by a spatially resolved LIF-LIFR signaling axis. Malignant cells residing in inflammatory niches displayed marked upregulation of LIF, a response reinforced by crosstalk with macrophages and associated with poor patient prognosis. Collectively, our findings establish a quantitative model of the spatially structured GBM microenvironment, linking stress-driven niche formation to malignant cellular states and immune regulation, and reveals spatially defined vulnerabilities that may be exploited for precision therapeutic intervention.