Abstract Spinal ependymomas comprise glial neoplasms that are classified by histological and molecular features. Two of the respective types, myxopapillary ependymomas (MPE) and spinal ependymomas (SP-EPN) are largely understudied. For each of these, our group identified clinically relevant subtypes with poorer (MPE-A; SP-EPN-A) and more favorable progression-free survival (MPE-B; SP-EPN-B) employing global methylation and transcriptomic profiling. As cellular origin and pathogenesis of these tumors are largely unknown, therapeutic options for patients with unfavorable subtypes are unavailable. We performed single cell transcriptomic sequencing of 12 formalin-fixed and paraffin-embedded spinal ependymomas including all MPE and SP-EPN subtypes and obtained a median of 10,091 cells per tumor after quality control. Uniform Manifold Approximation and Projection (UMAP) analysis revealed intertumoral heterogeneity as most samples formed separate clusters with tumors of the same subtype located in proximity. The expression of literature-defined neoplastic consensus cell states was examined for each cell of the data set and revealed intratumoral heterogeneity. Tumors expressed predominantly two programs, being a ciliated (ependymal-like) cell state (61% of all neoplastic cells) and the astrocytic state (31 % of all neoplastic cells). Interestingly, third and fourth most abundant states (mesenchymal and hypoxia) were almost exclusively expressed in MPE-A and SP-EPN-A. Furthermore, distinct clones of chromosomal copy number variations were identified across all tumors. These clones can be distinguished in UMAP representation of MPE single cell transcriptomes and to a lesser extend in SP-EPN. Next, we sought to characterize the resemblance of integrated tumor subtypes to normal spinal cord cell populations and observed a high similarity to human spinal cord ependymal cells. All spinal ependymoma subtypes display intratumoral heterogeneity with respect to the transcriptomic cell state resemblance of the neoplastic cells. Together with the transcriptomic similarity to ependymal and astrocytic cells, our data suggest the tumor origin within the astro-ependymal lineage.
Read full abstract