Decoding the Molecular Landscape of 262 Uterine Sarcomas: RNA-Seq Clustering of ESS, UTROSCT, and UUS with Prognostic Insights.

Low-grade endometrial stromal sarcomas (LG-ESS), high-grade endometrial stromal sarcomas (HG-ESS), undifferentiated uterine sarcomas (UUS), and uterine tumors resembling ovarian sex cord tumors (UTROSCT) are distinct non-smooth muscle cell neoplasms with varying clinical outcomes often exhibiting overlapping characteristics. Diagnosis can be supported by identifying characteristic recurrent translocations, which may be absent in some cases, complicating the distinction of equivocal cases. Additionally, cases with overlapping features of LG and HG characteristics are recognized. To address these challenges, we analyzed RNA-Seq profiles of 262 cases. Our results revealed that LG-ESS, with and without recurrent fusions, clustered into two partially overlapping expression profiles associated with distinct overall and relapse-free survival outcomes, with the cluster containing a majority of fusion-negative tumors demonstrating better prognoses. UTROSCT expression profiles closely resembled those of both LG-ESS subgroups, with NCOA3 fusion-positive cases clustering in groups with better survival outcomes. Furthermore, a distinct cluster for HG-ESS with BCOR and YWHAE fusions was identified, differentiating these tumors from HG-ESS without fusions. ONECUT3 emerged as a potential specific marker for this HG-ESS-fusion entity. A significant expression overlap was observed between monomorphic HG-ESS without fusions and pleomorphic UUS. These samples separated further into two mixed clusters distinguished by differences in immune activity, which significantly influenced overall survival and relapse-free survival outcomes. Unsupervised clustering of UUS revealed subgroups resembling either HG-ESS or muscle-cell differentiated tumors, suggesting that UUS may include poorly differentiated distinct entities, such as leiomyosarcoma, and that the distinction from HG-ESS may, in some cases, be arbitrary. Our transcriptome analysis highlights several entities with distinct survival characteristics, providing a foundation for further characterization of these rare, often difficult-to-classify, tumors.