Intra-tumoral heterogeneity in glioma

Glioblastoma is an incurable malignancy that remains poorly understood. Heterogeneous genetic, epigenetic and developmental programs are thought to drive glioblastoma, but their precise characterization remains challenging. Here we propose an integrative approach to model glioblastoma, combining single-cell RNA-sequencing of 28 tumors with bulk genetic and expression analysis of 401 specimens from the TCGA. We find that malignant cells in glioblastoma exist in four main cellular states that recapitulate distinct steps of neural development, are influenced by the tumor micro-environment, and exhibit plasticity. Importantly, we show that while each glioblastoma contains similar cellular states, their relative frequency varies between tumors and is influenced by chromosomal aberrations that each influence a particular state. By providing a roadmap of the cellular programs of malignant cells in glioblastoma and their modulation by genetic drivers, our work proposes a unifying model for glioblastoma.