Uncovering the Tumour Ecosystem of Glioblastoma at Single Cell Resolution

Glioblastoma is the most aggressive and prevalent form of malignant brain cancer. It is largely incurable and particularly heterogeneous in nature. Although the four subtypes of GBM -- established by TCGA on the basis of genomic and clinical features -- have provided a framework for scientists and clinicians to tackle the disease, a comprehensive understanding of GBM is still missing. Moreover, GBM affects both children and adults, but despite marked differences in response to therapy the two have traditionally been combined in studies on high-grade gliomas. As such paediatric form remains under-investigated.­

Our work addresses these outstanding problems. Using fresh tumour samples from paediatric and adult GBMs, we perform single-cell computational analyses on the transcriptomes of thousands of cells and dissect the intra-tumour composition of this disease. We find overarching features at the level of individual cells that at once offer a generalised understanding of GBM across its various subgroups whilst providing explanations as to their subtle but important differences.

We shed light on a small number of gene expression programs that are conserved across GBM cohorts and that can describe glioblastoma in a manner that is consistent with the basic biology. These findings match the TCGA subtype classifications with one important difference: the subtypes share the same sets of gene expression programs. What sets them apart is a matter of proportions. Similarly, distinctions between the adult and paediatric forms of GBM are attributed to varying capacities of cell types to undertake roles in the tumour ecosystem. These observations set us on a path that can improve the therapeutic outcome of this fatal disease.