SPATIAL TUMOR ECOSYSTEMS:

Abstract

Diffuse midline glioma (DMG) is a devastating pediatric brain cancer with an almost universally fatal prognosis. Despite detailed knowledge of its genetic and epigenetic drivers, durable therapeutic responses remain rare. This thesis investigates whether the spatial organization of tumor and immune cells helps explain the gap between molecular understanding and clinical progress. We propose that DMG functions as a spatially organized, adaptive ecosystem comprising distinct but interdependent cellular niches. Embedded within the developing brain, these communities coexist and may compensate for one another under therapeutic pressure. Understanding their organization and interactions could reveal vulnerabilities overlooked by approaches targeting individual pathways or cell types. In Chapter 2, we introduce mLSR-3D, a protocol for large-scale three-dimensional imaging of intact tissue. By preserving spatial relationships lost during tissue dissociation, this approach reveals patterns of immune infiltration and tumor organization that are not readily captured by conventional two-dimensional imaging or dissociative analyses. In Chapter 3, we identify the IGSF11-VISTA checkpoint axis as a mechanism through which DMG maintains immune quiescence. Tumor cells engage this pathway to preserve microglia, the brain’s resident immune cells, in a homeostatic, low-activation state, limiting spontaneous antitumor immunity. Disrupting this axis in preclinical models reactivates microglial effector functions and induces tumor regression. These findings demonstrate that microglia retain antitumor potential and suggest that DMG exploits regulatory mechanisms of developmental immune homeostasis to evade immune attack. Chapter 4 provides a framework for multi-omics profiling of sequential tissue sections, enabling integrated analysis of transcriptomic, proteomic, and spatial features within the same tumors. In Chapter 5, we develop pHGGmap, a spatially informed single-cell atlas of pediatric high-grade gliomas integrating datasets from more than 100 patients. This resource identifies recurrent tumor and immune cell states and supports a model of DMG organized around three spatial niches: an astro-microglial niche containing more differentiated tumor cells and homeostatic microglia; a tumor-reactive niche shaped by hypoxia and inflammation and containing immunosuppressive macrophages; and a persistent, stem-enriched niche associated with immunosuppressive myeloid cells and tumor invasion. Although their relative abundance changes following therapy, the overall ecosystem architecture remains remarkably stable, potentially contributing to treatment resistance. In Chapter 6, we reconsider immune checkpoint inhibition, oncolytic virotherapy, and CAR T-cell therapy through this ecosystem framework. Beyond molecular resistance and antigen escape, incomplete disruption of tumor niches and compensatory interactions may allow the disease to persist. Durable control may therefore require combinations that target multiple niches simultaneously and adapt to changes in the tumor over time. Together, this thesis defines architectural and regulatory principles governing tumor-immune interactions in DMG. It highlights both the resilience of spatially organized tumor communities and the therapeutic potential of reactivating brain-resident immunity, providing a foundation for ecosystem-informed strategies aimed at lasting disease control.