Integrated molecular and histopathological analysis of uterine smooth muscle tumors of uncertain malignant potential (STUMP) elucidates features associated with progression towards leiomyosarcoma

Uterine smooth muscle tumors of uncertain malignant potential (STUMPs) represent a major diagnostic challenge in gynecological pathology, occupying a gray zone between benign leiomyomas (LMs) and malignant leiomyosarcomas (LMSs). A subset of STUMPs progresses to LMS, an event associated with poor clinical outcomes, underscoring the need for improved risk stratification. This study aimed to characterize the molecular and immune landscape of uterine smooth muscle tumors (USMTs), with particular emphasis on identifying features associated with STUMP progression. Targeted transcriptomic profiling was performed using the nCounter PanCancer IO360 panel on 55 USMTs, including 16 non-progressed STUMPs (npSTUMPs), 3 progressed STUMPs (pSTUMPs), 18 LMSs, and 18 LMs. Transcriptomic analysis revealed an aggressive molecular profile in pSTUMPs, with significant upregulation of cell proliferation pathways (p = 0.0015), PI3K-AKT signaling (p = 0.0015), angiogenesis (p = 0.0015), and DNA damage repair pathways (p = 0.0015), alongside other pathways associated to tumor aggressiveness. Moreover, direct comparison between pSTUMPs and LMSs did not highlight any differences, strengthening the hypothesis that pSTUMPs closely resemble LMSs. In contrast, npSTUMPs exhibited intermediate pathway activation, with significantly lower activation of the cell proliferation pathway compared with LMSs (p = 0.0010) but higher than LMs (p = 0.0009). Direct comparison between pSTUMPs and npSTUMPs did not identify a progression-specific gene signature, although PTEN expression was reduced in pSTUMPs. To further refine molecular classification, array-based DNA methylation profiling was performed on 16 npSTUMPs, 3 pSTUMPs, 11 LMs, and 11 LMSs and analyzed using the Epignostix Sarcoma Classifier v13.1. pSTUMPs displayed marked molecular heterogeneity, being variably classified as uterine-type LMS, non-uterine-type LMS, or remaining unclassified. In contrast, npSTUMPs were more consistently recognized as uterine smooth muscle tumors, although subclass assignment between LM and uterine-type LMS lacked robustness. Methylation analysis also allowed the evaluation of genomic complexity, revealing that LMSs have the highest percentage (mean 19.6%), followed by pSTUMPs (mean 13%). Cytogenetic events such as loss of chromosome 13p were significantly 3 associated with tumor class (p = 0.021) and accompanied by frequent RB1 loss, detected in 63.6% of LMSs and 66.7% of pSTUMPs, compared with 9.1% of LMs and 18.8% of npSTUMPs. Immunohistochemical analysis of the tumor immune microenvironment revealed increased CD8⁺ T-cell infiltration and PD-L1 expression in LMSs but did not distinguish pSTUMPs from npSTUMPs, leading to the hypothesis that the immune compartment is not associated to the progression of these lesions. Collectively, these data support the concept that STUMPs are biologically intermediate tumors along a molecular continuum between leiomyomas and leiomyosarcomas. Moreover, methylation analysis showed promising results; however, while sarcoma classifier shows excellent performance in benign uterine smooth muscle tumors, its limited accuracy in malignant cases highlights current constraints in molecular classification of uterine tumors. Finally, the analysis of genomic complexity highlighted how RB1 loss might be used as a promising marker that might improve progression risk stratification when integrated with histopathological evaluation.