(P 159) USING THE TRANSCRIPTIONAL SPECTRUM OF UTERINE SMOOTH MUSCLE TUMORS TO IDENTIFY PROGNOSTIC SIGNATURES FOR ABNORMAL LEIOMYOMAS AND SMOOTH MUSCLE TUMOURS OF UNCERTAIN MALIGNANT POTENTIAL

Objective: Clinicians continue to be challenged by the limited clinical, pathologic, and radiologic distinctions preoperatively between uterine leiomyosarcoma (uLMS) and uterine leiomyoma (uLM) variants or smooth muscle tumors of uncertain malignant potential (STUMPs). Given the poor prognosis of uLMS, with an overall 5-year survival of 15-65%, the risk of misdiagnosis prior to resection is significant. While the distinction between typical uLM and uLMS is more straightforward, overlapping histological features of STUMPs and uLM variants can make diagnosis and management challenging. The lack of high quality RNA-seq data for STUMPs has prevented the ability to better understand how these tumors relate to uLM and uLMS transcriptomically. Thus, we profiled uLM, STUMPs, and uLMS tumours to develop an IHC or transcriptomic signature that can be used to classify these intermediate uterine smooth muscle tumours, and aid in identifying more clinically aggressive uLM and STUMPs

Methods: uLM (n=37), STUMP (n=6), and uLMS (n=10) frozen samples were collected from the Toronto Sarcoma patient repository (n=36) and from Stanford University (n=17). They were prepared as FFPE and H&E slides for pathology review to exclude samples which were necrotic (n=4) or non-representative (n=3). A finalized cohort of uLM (n=33), STUMP (n=5), and uLMS (n=8) samples were prepared for bulk RNA sequencing (n=46) (Figure 1). Normal uterine smooth muscle (uSM), uLM, and uLMS RNA-seq data were also included from GTEX (n=52), TCGA (n=81), and other available datasets (n=97). Read count data was inputted into RACCOON, an unsupervised hierarchical clustering algorithm, to create a transcriptomic taxonomy per previously established methods. Clinical data points were retrospectively collected from the Toronto Sarcoma clinical database

Results: uLM, STUMP, and uSM form a distinct cluster away from LMS, both uterine and non-uterine, at the first hierarchical level (Figure 2A). At the second level, the uSM forms a separate cluster away from the uLM and STUMP. However, 2/5 STUMP samples cluster with the uSM, indicating potential sampling of adjacent uterine smooth muscle. Within the uLM and STUMP class, a distinct uLMS subclass emerges at the next hierarchical level, represented as low grade uLMS (ULMS LG) (Figure 2B). Of the 3 STUMP samples within the uLM class, one clustered with the ULMS LG group and was re-reviewed as uLMS at recurrence, while the other 2 clustered with the uLM specimens (Figure 2B). A reduced sample size in the ULMS LG cluster (n=8) may have precluded overall survival analysis from reaching statistical significance (p=0.61) between this group and the ULMS HG cluster (Figure 3). This suggests that transcriptional analysis is able to identify clinically appropriate clusters for STUMPs and can be used to provide insight into differential gene expression differences that could be utilized as biomarkers.

Conclusion: The integrated molecular profiling of STUMP samples holds significant clinical value, providing insight into characterizing these rare uterine mesenchymal neoplasms and elucidating their relationship to other uterine smooth muscle tumours. Despite the limitations of the small STUMP sample size of fresh frozen RNA, this data provides information regarding whether STUMP is truly a distinct tumor entity or if it can be profiled into either uLM or ULMS LG groups based on its expression profile. To our knowledge, this is the first instance of transcriptomic profiling of rare STUMP tumours. Future work aims to identify differentially expressed genes between identified clusters, of which protein expression will be validated in FFPE specimens by IHC. The use of an IHC biomarkers identified based on transcriptomic analysis for abnormal uLMs and STUMPs will enable more accurate diagnosis and more effective management of patients with uterine tumours, reducing the risk of a more aggressive recurrence.