EPCO-48. STEMNESS SIGNATURE STRATIFIES GLIOMAS WITH CLINICAL IMPORTANCE AND REVEALS A PROLIFERATIVE PHENOTYPE IN SINGLE CELL DATA

Abstract Gliomas are the most common central nervous system neoplasm and despite past significant progress, its diagnostics faces suboptimal classification which impacts patient management. The stem cell-like phenotype of various cancers is correlated with worst overall prognosis. We propose a Stemness prediction model based on gene expression signatures neural progenitors that can be used to measure dedifferentiation state (or Stemness) glioma samples. To built model, publicly available single-cell RNA sequencing data was identify from fetal astrocyte (AST) population. Subpopulations interest were identified through marker genes. applied one-class logistic regression using AST bulk transcriptomic generate an stemness index (ASTsi). ASTsi able stratify gliomas grade, histology, molecular subtypes. Grade 4, glioblastoma, IDHwt, mitochondrial proliferative functional subtypes had highest stemness. When longitudinal samples we observed increase in IDHmut recurrent decrease IDHwt tumors, compared primary Additionally, RNAseq adult glioblastomas found clusters high-stemness cells (ASTsi > 0.8). A differential combined pathway analysis between high- low-stemness revealed cell cycle, DNA repair mechanisms histone modifications upregulated balance methylation demethylation may directly related these cells. More in-depth genes pathways being carried out provide important information about oncogenesis phenotypic characterization Our stratified by pathological features tumor subpopulations distinct degree IDHwt.