Single-Cell RNA Sequencing of Pediatric Ependymoma uncovers Subclonal Heterogeneity associated with Patient Survival

Ependymoma is a malignant glial tumor occurring throughout central nervous system, with one third of relapse rate. The increasing accessibility of single-cell transcriptome has accelerated our understanding of ependymoma. In this study, we generated a high-resolution single-cell dataset with a particular focus on the comparison of subclonal differences within tumor populations. As a proxy to traditional pseudotime analysis, we developed a novel trajectory scoring method to predict survival outcomes and specific molecular characteristics. Moreover, we identified putative cell-cell communication features between relapsed and primary samples and show upregulation of pathways associated with immune cell crosstalk. Our results reveal both inter- and intratumoral gene expression profiles and tumor differentiation and provide a framework for future studies on transcriptomic signatures of individual subclones in brain tumors at single-cell resolution, by which could complement existing published datasets and provide valuable insights into cell-type-specific properties and lay the foundation for therapeutic treatments of diseases. In this study, we report the generation of curated single-cell transcriptomes from EPN patients across PF-A and ST regions. To address intratumoral subclone heterogeneity, we used a deconvolution approach using inferCNV to compare molecular features between subclones in a single PF-A sample as a proof-of-concept. We then identified a subclone-specific cilia-associated program within an individual PF-A EPN sample which was correlated with a more highly differentiated subpopulation. As a proxy to traditional pseudotime analysis commonly used, we also developed and incorporated a trajectory score analysis to our datasets to predict correlations between survival outcomes in EPN and specific molecular characteristics, as well as primary and recurrent tumor populations. Finally, we identified putative cell-cell communication features between relapsed and primary EPN samples and show upregulation of pathways associated with immune cell crosstalk. Our results reveal both inter- and intratumoral gene expression profiles and cell composition differences associated with subclonal variability and provide a framework for future studies on transcriptomic signatures of individual subclones in brain tumors at single-cell resolution.