TMIC-57. CHARTING THE SPATIAL ARCHITECTURE OF GLIOBLASTOMA USING GBMAP: INSIGHTS FROM IN SITU SEQUENCING

Abstract

Abstract Glioblastoma (GB), an aggressive brain malignancy, is associated with poor prognosis and quality of life. Single-cell RNA sequencing has helped unravel the complexity of GB cell states. However, to comprehensively unravel the pathobiology of GB, it is essential to consider the spatial context in which these cells reside within the tumor microenvironment (TME). In this study, we aimed to understand the spatial architecture of GB tumors and its implications on cellular and molecular heterogeneity. To achieve this, we developed GBmap, a comprehensive resource that harmonizes 26 datasets comprising 239 patients and over 1.1 million cells. Leveraging computational advancements, we utilized GBmap to create a GB-specific panel for charting the spatial architecture of tumors using an in situ sequencing (ISS)-based platform (HybISS) in an independent cohort of 13 tumors. We curated a panel of 209 genes based on the GBmap, capturing cell phenotypic features, intra- and inter-tissue heterogeneity, and in silico L-R analysis predictions. The panel includes markers for specific cellular processes such as inflammation, extracellular matrix remodeling, hypoxia, and angiogenesis, providing an additional layer of functional information. Our optimized pipeline incorporates a new method for ISS-based transcriptomics data (direct RNA targeting), which enhances sensitivity and tolerance for tissue specimens of poorer RNA quality. This new approach enables improved cell type/state resolution and classification confidence. Using this advanced approach, we observed heterogeneous cell-cell colocalization patterns that vary among patients, revealing spatiotemporal distributions of specific cancer phenotypes and distinct co-localization patterns with neighboring cells in the TME. Our findings showcase the unprecedented spatial and cellular resolution achieved, surpassing previous GB studies. Our study sheds light on the spatial architecture of GB tumors, unveiling previously unexplored cellular and molecular interactions. GBmap and our optimized pipeline offer a powerful resource for researchers to investigate GB heterogeneity, generate hypotheses, and contribute to advancing GB research.