NEW YORK – Pediatric ependymomas that include more undifferentiated cells are more likely to be aggressive, a single-cell RNA-sequencing analysis has found.
Ependymomas are heterogeneous tumors of the central nervous system, and while some are relatively benign, others are much more aggressive tumors. DNA methylation pattern analyses have previously uncovered nine molecular subgroups of ependymomas. Researchers from Dana-Farber Cancer Center and elsewhere aimed to further characterize these subgroups and trace the cellular programs using single-cell transcriptomes.
As they reported on Monday in Cancer Cell, they found ependymomas to be composed of cell types typically found in brain cell development but stalled at various differentiation states, with three potential trajectories. Tumors associated with poor prognosis contain more undifferentiated cell populations, they noted, and the tumors' transcriptomic signatures highlighted possible treatment strategies.
"Taken together, our analyses reveal a developmental hierarchy underlying ependymomas relevant to biological and clinical behavior," senior author Mariella Filbin from the Dana-Farber Boston Children's Cancer and Blood Disorders Center and her colleagues wrote in their paper.
The researchers generated single-cell transcriptomic data on 20 fresh surgical tumor samples from 18 patients, as well as on eight patient-derived cell models and two patient-derived xenografts. At the same time, they performed single-nucleus RNA-seq on 14 snap-frozen ependymoma samples. In all, they analysed 74,927 single tumor cells or nuclei, and, based on the samples' DNA methylation patterns, determined the tumors' molecular subgroups.
The transcriptomes of cells found within ependymomas, the researchers found, are similar to those of normal brain cells. However, they noted that the tumor cells appeared to have stopped at various points along the differentiation process. Within ependymomas, they uncovered three differentiation trajectories: ependymal-like, glial progenitor-like, and neuronal-like cells.
For instance, when the researchers focused on posterior fossa ependymoma samples — posterior fossa group A is the most aggressive of the subgroups and posterior fossa group B is linked to better patient prognosis — they uncovered nine recurrent transcriptional metaprograms. Two programs were linked with cell-cycle genes and were particularly found among PF-A samples. Other metaprograms were associated with mature cell types, astrocytes, or immature stem-like cells and neuronal or glial lineage precursors.
Similarly, the researchers identified 10 transcriptional metaprograms, including two associated with cell-cycle genes and others linked to radial glial-like or neuronal precursor-like cell types, among eight supratentorial ependymoma samples.
Undifferentiated cell states were also more common among tumors with poorer prognosis, such as the typically more aggressive PF-A tumors, the researchers found. The more benign groups like PF-B or PF-subependymoma had less proliferative and more differentiated cell populations.
Within bulk RNA expression data, tumors' transcriptional signatures correlated with survival. For instance, the PF-ependymal-like signature stratified overall patient survival and progression-free survival, as well as survival within a cohort of only PF-A tumors. PF-A cases with low PF-ependymal-like signature had a 7.3-fold increased risk of death.
These gene signatures also suggested potential drug targets. For example, when the researchers integrated the cell-population-specific genes with the Drug Gene Interaction database, they noted that the Wnt-signalling regulator LGR5 and the anti-apoptotic gene MCL1 were linked to the PF-NSC-like program and could be potentially druggable. Small-interfering RNA-knockdown of LGR5 indeed inhibited sphere formation within a patient-derived PF-A model, the researchers noted.
Similarly, a radial glial cell-like program is marked by FGFR expression, and treatment with dovitinib, a FGFR inhibitor, affected cell viability in culture.
This "[d]econvolution of heterogeneous [ependymoma] subpopulations pinpoints key malignant transcriptomic signatures," the researchers wrote, adding that it can further "identify high-risk tumors and subsequently inform the development of more effective anti-[ependymoma] treatments."