TMIC-04. SPATIAL SEQUENCING OF MEDULLOBLASTOMA REVEALS BIOLOGICAL NICHES CORRELATED WITH HIGH-RISK FEATURES AND RELAPSE

Abstract Medulloblastoma (MB) defines a heterogeneous set of neuroepithelial cancers of the posterior fossa. The MB tumor microenvironment (MB-TME) influences tumor progression and therapy response and has emerged as a growing target for novel therapeutic approaches. Prior single-cell analysis has characterized the MB-TME composition but does not reveal the spatial orientation of components. To address this gap, we performed 10X Visium spatial sequencing on sixteen pediatric MB samples obtained at surgical resection with samples representing the four molecular subtypes (WNT, N=1; SHH, N=6; Group 3, N=2; Group 4, N=5) and two matching samples at diagnosis and relapse. Spatial voxel clustering yielded clusters corresponding to malignant, immune, and stromal components of the TME, including tumor-associated astrocytes (TAAs), macrophages (TAMs), and vascular endothelium. Ten distinct clusters emerged representing malignant cell states, defined primarily by programs of cell cycle progression and neuronal differentiation. Notably, these MB cell clusters were differentially abundant between the MB subgroups, highlighting heterogeneity between disease subtypes. Moreover, examining the patterns of cellular spatial distribution in high-risk (HR) patients revealed dense regions of quiescent MB progenitors, while standard-risk patients showed greater heterogeneity within their TME. Proximity-dependent intercellular signaling further highlighted increased signaling by TAAs and vascular endothelium with decreased heat shock response signaling in HR patients. Comparing samples from relapse to diagnosis, TAMs, TAAs, and vascular endothelium were found to constitute a greater proportion of the TME. Subsequent niche-dependent gene expression and pathway analysis for TAAs localized near tumor vasculature in relapsed samples revealed increased expression of metastasis-associated pathways. Collectively, the results of our study enable new insight into the heterogeneity of the MB-TME, with close spatial association of quiescent MB progenitors correlating with high-risk clinical features. Quiescent MB progenitors have been documented to confer resistance through cellular dormancy, and identifying its molecular drivers may reveal potential targets for anticancer therapy.