Multiomics profiling reveals excessive fibroblast activation and collagen deposition are associated with poor prognosis in desmoid tumors

Backgroud Desmoid tumors (DT) is a rare locally aggressive soft tissue tumor of fibroblastic origin, exhibiting unique biological behavior characterized by spontaneous shrinkage or enlargement. It demonstrates high recurrence rates after surgical resection and lacks of prognostic biomarkers. Methods A total of 102 paraffin-embedded tissue samples from 88 surgical DT patients were collected for proteomic sequencing. Validation was conducted using 17 internal pairs of specimens and 128 external specimens collected for transcriptomic sequencing. Single-cell transcriptomics sequencing was used to analyze cellular composition and interactions in 3 fresh DT tissue samples. Primary cells were isolated and cocultured to evaluate the antitumor effect of TGF-β inhibitors in vitro. Results Tumor location was the only clinical feature associated with tumor relapse-free survival. 65 differential signatures were screened out and clustered into three groups, largely corresponding to clinically recurrent and nonrecurrent patients. Excessive fibroblast activation and collagen deposition are associated with DT recurrence. Using tumor location and three key signatures (POSTN, COL1A1, and COL1A2), a clinical predictive model was successfully constructed with a prediction accuracy of more than 80%. Tumor cells promoted the activation of myofibroblastic cancer-associated fibroblasts primarily through POSTN-ITGAV/ITGB5 signaling. In vitro, a TGF-β inhibitor had limited effects on tumor cell proliferation but significantly suppressed cell invasion. Conclusion A new clinical prediction model integrating tumor location with POSTN, COL1A1, and COL1A2 expression was constructed for DT prognosis. POSTN-ITGAV/ITGB5-driven excessive fibroblast activation and collagen deposition are associated with poor prognosis. TGF-β inhibitors can inhibit cell invasion and represent a novel therapeutic option for DT.