IMG-07. MOLECULAR PROCESSES GIVE RISE TO EMERGENCE OF RADIO-PHENOTYPES ON MRI SCANS OF PEDIATRIC LOW-GRADE GLIOMA

Abstract

Abstract There is mounting evidence that tumor microenvironmental pressure selects for somatic genetic alterations that contribute to the formation of distinct morphological characteristics, captured on radiology scans as radio-phenotypes. Such phenotypic variations are a source of heterogeneity in clinical manifestation of tumors of the same histology across the patients, and in part, their heterogeneous responses to therapies. Deeper understanding of the associations between genotype and radio-phenotype in pediatric low-grade glioma (pLGG), the most histologically diverse childhood brain tumor, may facilitate precision diagnostic and therapeutic approaches. Here, we categorize pLGGs into distinct and relatively homogeneous imaging subtypes based on radiomic features and further explore the associations of these imaging subtypes with genotype. From multiparametric MRI scans of 167 pLGGs from the Children’s Brain Tumor Network (CBTN), 881 radiomic features and clinical variables (tumor location, age, and gender) were extracted. After dimensionality reduction using principal component analysis (PCA), K-Means clustering was applied on 19 principal components to group the patients into three imaging subtypes. Using whole transcriptomic data from OpenTargets, differential expression and co-expression of network- and pathway-level and immune-related signaling were compared among these three imaging subtypes. Gene Set Enrichment Analysis (GSEA) revealed differentially higher expression of cell cycle regulatory, extracellular matrix (ECM) remodeling, and cell migratory pathways in imaging subtype1 than subtypes 2 and 3, and upregulation of ECM and immune-related pathways in subtypes 1 and 2 compared to subtype3. Based on Gene Sets Net Correlations Analysis (GSNCA), subtype1 exhibited differential co-regulation of TNF/TNFR1 signaling compared to subtype2, and differential co-regulation of RHOG GTPase and TGFB1pathways when compared against subtype3. Subtype2 showed differential co-regulation in NOTCH1 signaling and transcriptional regulatory pathways. Our proposed multi-disciplinary radiomic-genomic analysis approach elucidates the molecular and biological processes in the genotype of the tumors that are associated with emergence of distinct imaging subtypes in pLGG.