We are interested in developing new therapeutics to treat glioblastoma, but a deep understanding of glioblastoma cell complexity hinders progress. We thus wanted to understand just how heterogenous these cancers really are, and if we could find some organization to this heterogeneity. We also thought that such knowledge might provide insight into glioblastoma origin cells and reveal therapeutic vulnerabilities. 

We harvested tumor samples directly from the operating room and immediately processed cancer cells for single-cell RNA sequencing and functional experiments. Tens of thousands of cancer cells from 16 glioblastoma patients were analysed. What we found was striking. Every cancer cell from every patient expressed a transcriptomic signature similar to one of the cell types found in the developing brain: astrocytes; oligodendrocytes; neurons; and dividing progenitor cells. The presence of genomic alterations in the cancer cells such as chromosomal alterations, however, made these comparisons difficult. To solve this problem, we performed single-cell RNA sequencing of the developing human brain to build a normal brain roadmap for comparison. We found that all tumours shared a conserved hierarchical organization. Progenitor cancer cells give rise to differentiated cancer cells reminiscent of all 3 neural lineages, with varying proportions between patients or clones of cells. These data provide insight into the nature of glioblastoma stem cells and reveal a common pathway for glioblastoma development.

Our study suggests glioblastoma is derived from normal neural stem cells that have acquired cancer causing alterations but have maintained multipotency. That this developmental pattern exists in each patient informs inter- and intra-tumoral heterogeneity and exposes new therapeutic opportunities.