TMIC-68. Deciphering the spatial architecture of transcriptional heterogeneity in glioblastoma at single cell resolution

Abstract The diffusely infiltrative growth pattern of glioblastoma (GBM) is a major obstacle to effective therapy. A complex network of tumor intrinsic and microenvironmental features interact to shape the landscape of this growth pattern, but deciphering the molecular substrates of these interactions remains challenging. While single cell RNA sequencing has reshaped our understanding of GBM heterogeneity and uncovered cellular phenotypes of many cellular populations in these tumors, the spatial context of this heterogeneity could not be assessed with earlier technologies. By contrast, spatial transcriptomics platforms have the potential to map this transcriptome-wide information in situ. We recently reported the use of this technology to develop a layered model of spatial architecture in GBM and showed that hypoxia represents a key organizing feature of GBM heterogeneity. However, the limited resolution of this platform precluded assessment of whole-transcriptome features at truly cellular resolution. In this work, we now present our findings utilizing the next generation of this technology and showcase the ability to map whole transcriptome data in spatial context at true single cell resolution in primary patient GBM samples. We explore heterogeneous GBM phenotypes in the context of GBM spatial microarchitecture. Using copy number inference, we demonstrate the ability to assess clonal heterogeneity in situ and map the pattern of invasion of unique GBM clones. Finally, we discuss our progress to discover organizing principles underlying distinct patterns of GBM cell invasion. Taken together, we demonstrate the potential of single cell spatial transcriptomics to decipher the molecular heterogeneity of GBM microenvironment in situ and uncover the biology driving the aggressive infiltrative growth of this disease.