Dean's Research Fellow Department of Neurological Surgery, Columbia University Vagelos College of Physicians and Surgeons
Introduction: The poor prognosis for Glioblastoma (GBM) is related to invasion of the brain and complex interactions with surrounding non-neoplastic cells in the tumor microenvironment which contribute to treatment failure. To identify potential new treatment strategies, we investigated a novel comprehensive single nucleus RNA sequencing (snRNAseq) and spatial transcriptomics approach to human GBM samples. This was used to elucidate the cellular composition of the tumor microenvironment in GBM and reveal interactions between GBM and Astrocytes that represent novel therapeutic targets.
Methods: Nine IDH-wildtype GBM samples were analyzed using snRNAseq and spatial transcriptomics. Deconvolution was used to identify the distribution and abundance of predominant tumoral and non-tumoral cell types within our samples and spatial cross-correlation was used to quantify the patterns of colocalization between the different cell types. Colocalization patterns were aggregated across samples to determine which patterns persisted across samples. The derived transcriptional signature associated with these patterns was correlated with survival using The Cancer Genome Atlas (TCGA) database.
Results: Spatial cross-correlation analysis showed that several non-neoplastic cell types are found in proximity to specific subtypes of GBM cells. Notably, reactive astrocytes and mesenchymal GBM cells showed significant colocalization across the dataset. In the TCGA database, the transcriptional signature associated with the colocalization of these cell types showed a significant correlation with poor survival that is not accounted for by either cell type alone, suggesting that interactions between these cell types plays a role in glioma progression.
Conclusion : Our analysis of snRNAseq and spatial transcriptomics revealed prognostically significant patterns of colocalization, and identified potential therapeutic targets to interrupt crosstalk signals and metabolic interactions between GBM cells and reactive astrocytes in the brain tumor microenvironment.
