Die Geheimnisse des Gliosarkoms: Räumliche Einblicke in eine seltene Tumorvariante

Gliosarcoma (GS) is a rare, aggressive variant of glioblastoma (GB), comprising about 2% of GB cases. While GS and GB share genetic and epigenetic traits, GS is distinguished from GB by its biphasic histological growth pattern — glial and sarcomatous differentiation — and a greater tendency for skull invasion and distant metastases. The mechanisms driving these differences remain unclear. We propose that GS's unique architecture and clinical behavior stem from distinct interactions between tumor cells and the tumor microenvironment (TME).

To investigate this hypothesis, we employed sequencing-based spatial transcriptomics (ST) to analyze 15 GS samples from 13 patients, propensity matched with GB samples based on age, sex and methylation profiles. Using unsupervised clustering and non-negative matrix factorization, we identified transcriptional metaprograms in GS and GB. Cell-type deconvolution mapped tumor cell subtypes to TME components. Cell-to-cell communication analysis further elucidated the interactions between tumor phenotypes and TME cells.

In the GS samples, we identified four spatial metaprograms that were abundant across multiple samples and characterized by shared gene expression patterns. Notably, one of these was a hypoxia-related program, with upregulated genes such as MT1X and VEGFA. Cell-type deconvolution revealed a strong spatial correlation between this program and tumor cells exhibiting a mesenchymal phenotype. The proportion of these mesenchymal-like tumor cells was significantly higher in GS tumors compared to GB tumor cells (p < 0.01).

This study provides the first spatially resolved transcriptional insights into gliosarcoma, highlighting the tumor's unique architecture and interactions with the TME, and potentially offering new therapeutic targets for both GS and GB.