Prognostic biomarkers for enhanced risk stratification in extraskeletal myxoid chondrosarcoma: a retrospective cohort...

Extraskeletal myxoid chondrosarcoma (EMC) is a rare soft-tissue sarcoma driven by the EWSR1::NR4A3 fusion. Its unpredictable clinical course and lack of validated molecular prognostic tools complicate patient treatment.

We performed whole-transcriptome targeted RNA sequencing (TempO-Seq) on archival formalin-fixed paraffin-embedded (FFPE) tumor specimens from 12 molecularly confirmed EMC cases. To explore potential biological markers, the cohort was stratified into good-prognosis (n = 6; overall survival (OS) >8 years) and poor-prognosis (n = 6; OS < 8 years) groups. We conducted differential expression analysis, gene set enrichment analysis (GSEA), and applied least absolute shrinkage and selection operator (LASSO)-Cox regression with bootstrap validation (1,000 iterations). Candidate genes were cross-validated using 31 external EMC cases across public microarray and NGS datasets. Immune infiltration was estimated computationally and spatially validated via multiplex immunofluorescence (mIF) on two representative specimens.

We identified 550 differentially expressed genes and derived a three-gene prognostic candidate (PXN, TYMS, H1FX) that stratified patients by overall survival in the training cohort (log-rank P = 0.0099), despite internal validation indicating model overfitting. H1FX was significantly upregulated in EMC versus other cartilaginous tumors across independent datasets (P = 0.001). Immune profiling revealed higher B-cell infiltration in low-risk tumors (P = 0.005). Proof-of-concept mIF in two specimens suggested a spatially localized immunosuppressive microenvironment in the high-risk tumor, with increased exhausted CD3⁺CD8⁺PD1⁺ T cells (P = 6.4 × 10^-5), FOXP3⁺ regulatory T cells (P = 0.006), and elevated PXN protein expression (P = 0.008). All mIF comparisons are within-specimen region of interest (ROI) analyses and should be interpreted as hypothesis-generating.

This integrated transcriptomic and spatial immune characterization identifies PXN, TYMS, and H1FX as exploratory EMC biomarker candidates, with PXN showing the most consistent cross-platform support. Immune microenvironment heterogeneity in EMC warrants further investigation. Given the small cohort and model overfitting, these findings are exploratory. Prospective validation in larger, multi-center cohorts is required.