A Siriraj-Yale team profiled RNA from 12 molecularly confirmed extraskeletal myxoid chondrosarcomas, split into six good- and six poor-prognosis cases. A PXN-TYMS-H1FX signature separated survival in training data, but internal validation explicitly indicated overfitting.
Key findings
- The study found 550 DEGs and a three-gene signature associated with training-set OS at log-rank p=0.0099, but overfit. H1FX was higher than in other cartilaginous tumors (p=0.001), B-cell infiltration was higher in low-risk tumors (p=0.005), and high-risk ROIs showed more exhausted CD8 PD1+, FOXP3+ Tregs, and PXN.
Why this matters globally
EMC is a rare sarcoma without validated molecular prognostic tools. Multicenter data and tissue networks may clarify biological and immune heterogeneity, but treatment or surveillance decisions require larger prospective evidence.
Thai researcher contribution
Siriraj teams in precision medicine, pharmacology, pathology, oncology, and orthopedics built a Thai rare-disease cohort and linked it with international public data and Yale expertise.
Limitations to consider
Only 12 training cases and an eight-year dichotomy make the model unstable, and overfitting was observed. The 31 external cases span platforms, while two mIF specimens cannot represent between-patient variation. Numerous ROI-level comparisons may risk pseudoreplication if regions are treated as independent.
Verify the original sources
PeerJRead the original article↗DOI: 10.7717/peerj.21497