Transcriptome profiling of uterine leiomyosarcomas identifies a leiomyoma-like expression pattern that indicates better survival.

Abstract

Background: Uterine leiomyosarcomas are rare and aggressive cancers with poor survival. Their non-cancerous counterparts, uterine leiomyomas, are common tumors affecting many women during reproductive years. Distinguishing leiomyosarcomas from leiomyomas remains a diagnostic challenge. This study aimed to identify molecular biomarkers that differentiate leiomyosarcomas from leiomyomas and to assess their prognostic value.

Methods: We analyzed 3'RNA sequencing data from 51 leiomyosarcomas and 44 leiomyomas with differential gene expression analysis and machine learning to identify diagnostic biomarkers. We utilized immunohistochemistry to validate the findings. We used Kaplan-Meier and Cox regression models to assess disease-specific survival in leiomyosarcoma patients.

Results: Leiomyosarcomas exhibited significant dysregulation of the retinoblastoma and cell cycle pathways. Clustering based on retinoblastoma pathway genes identified a subset of leiomyosarcomas with a leiomyoma-like expression profile that associated with better survival. Machine learning-based classification identified genes with high predictive accuracy, including genes from the retinoblastoma pathway. Immunohistochemistry validated TOP2A and CDK1 as potential diagnostic biomarkers, and their higher expression was associated with worse survival. Leiomyoma driver alterations were found in 27% of leiomyosarcomas, but these showed no association with survival.

Conclusion: Our results reveal molecular heterogeneity in uterine leiomyosarcomas and provide potential diagnostic and prognostic biomarkers for improved patient management.