Clinical and molecular features of primary gliosarcoma with digital spatial whole-transcriptome analysis of glial and mesenchymal components

Abstract

Gliosarcoma is a rare subtype of IDH-wildtype glioblastoma defined by mixed malignant glial and high-grade sarcomatous histological elements. Gliosarcoma is clinically managed similarly to glioblastoma and has a poor outcome. The sarcoma-like regions of gliosarcoma are thought to represent extreme mesenchymal metaplasia of neoplastic glial cells. Factors contributing to this phenomenon are not completely understood. Here we report a single-institution series of 37 gliosarcomas including next-generation sequencing data on 25 cases and digital spatial whole-transcriptome analysis on four cases to characterize differential gene expression between glial and mesenchymal components. Gliosarcoma demographic and genetic features were compared to a cohort of 75 primary adult hemispheric IDH-wildtype non-sarcomatous glioblastomas. Patient age, tumor location, sex, and overall survival in gliosarcoma were similar to glioblastoma. Gliosarcomas showed a significantly lower rate of EGFR amplification and a higher rate of NF1 mutation compared to glioblastomas in next-generation sequencing analysis. Digital spatial whole-transcriptome analysis showed a distinct transcriptomic profile in sarcomatous regions with over-expression of genes involved in extracellular matrix development and remodeling. Selected differentially expressed transcripts were examined further by immunohistochemistry. The glial elements of gliosarcomas showed higher immunoreactivity for Chitinase-3-like protein 1 (CHI3L1) than glioblastomas, and low to absent immunoreactivity within the sarcomatous elements. Lymphoid Enhancer-Binding Factor 1 (LEF1) immunoreactivity was identified within sarcomatous regions of gliosarcoma without detectable nuclear β-catenin, suggesting a role for β-catenin independent wingless (WNT) effector signaling in sarcomatous transformation. This study adds to the growing literature demonstrating differences in the genetic underpinning of gliosarcoma and glioblastoma, establishes feasibility of spatial transcriptomic approaches in gliosarcoma, and builds on digital spatial profiling-based results as a discovery platform to identify pathways and immunohistochemical markers for further study.