EPEN-13. Clinically relevant molecular hallmarks of PFA ependymomas display intratumoral heterogeneity and correlate with tumor morphology

Abstract

Abstract PFA ependymomas are aggressive posterior fossa tumors that mainly effect children and have a poor prognosis despite intensive therapy. Histopathology reveals glial neoplasms that usually show perivascular pseudorosettes throughout the tumor. Tumor cell density may vary within a tumor and sometimes shows well demarcated nodules of extremely densely packed tumor cells within surrounding loose glial tissue harboring diffusely distributed tumor cells. To determine potential molecular differences in such areas and to understand the clinical meaning, we analysed 70 clinically well-annotated PFA samples and 9 corresponding relapse samples. Areas of low and high cell density were digitally identified with a cut-off of 8,500 cells/mm². Global DNA methylation and copy-number profiling was performed in 52/79 cases for the entire tumor area of the sample and in 15/79 cases for the cell-dense and less cell-dense regions, separately. Fluorescence-in-situ-hybridization was performed for 20 cases, and scRNA-Seq data were analysed for 12 cases. The proportion of cell-dense areas proved to be highly variable on a continuum ranging from 0 % to 100 % of the tumor area. It was significantly higher in relapses in comparison to the respective primary tumors (p=0.036). Also, cell-dense areas displayed a significantly higher proportion of proliferating tumor cells (p<0.01). Global DNA methylation was similar, but not identical in different tumor areas. In 6/15 cases, the PFA subtype changed as determined by the current version of the brain tumor classfier (v12, www.molecularneuropathology.org). Chromosomal changes at 1q and 6q were only detectable in cell-dense areas in 5/12 and in 3/12 cases, respectively. Finally, scRNA-Seq data confirmed a widespread heterogeneity regarding 1q and 6q changes in PFA ependymoma cells. These data suggest that PFA ependymoma habor a previously unrecognized heterogeneity that has to be accounted for when selecting material for the analysis of clinically relevant DNA methylation profiles or copy number changes.