LAUSR

Abstract METHODS Through the Advancing Treatment for Pediatric Craniopharyngioma (ATPC) consortium we accumulated preoperative MRIs and tumor RNA for 50 unique ACP patients. MRIs were assessed quantitatively for 28 different features and analyzed using Multiple Factor Analysis (MFA) and optimal clustering was determined via maximization of Bayesian Information Criterion (BIC). Following bulk RNAseq, differential expression and pathway enrichment were performed using standard methodologies (i.e., DESeq2 and GSEA). RESULTS MRI features were well represented in the first 3 dimensions of MFA (variance explained=67.32%); specifically tumor/cyst size, ventricular size, and cyst fluid diffusivity. Using this three-way axis, we identified 3 patient subgroups. Transcriptional differences between these subgroups indicated one group was enriched for DNA damage response and MYC related pathways, one group enriched for SHH, and one group enriched for WNT/β-catenin and EMT-related pathways. CONCLUSION This preliminary work suggests that there may be unique gene expression variants within ACP, which may be identified preoperatively using easily quantifiable MRI parameters. These radiogenomic signatures could provide prognostic information and/or guidance in the selection of antitumor therapies for children with ACP.