Abstract
INTRODUCTION: Leiomyosarcoma (LMS) is a rare and aggressive soft tissue sarcoma with limited therapeutic options and poor prognosis. Identifying reliable prognostic markers and therapeutic targets is critical for improving personalized treatment strategies. METHODS: We integrated single-cell RNA sequencing (scRNA-seq) data of LMS from the Gene Expression Omnibus (GEO) with bulk RNA-seq and clinical data from The Cancer Genome Atlas (TCGA). Malignant cells were identified using machine learning approaches, and their stemness index was calculated. Cells were stratified into high and low stemness index groups, and differential gene expression analysis was performed. Prognostic markers were identified through a sequential pipeline: univariate Cox regression to screen candidate genes, Lasso regression for feature selection, and multivariate Cox regression for model construction and survival analysis. RESULTS: Cells with a high stemness index exhibited a more complex tumor immune microenvironment (TIME) and enhanced intercellular interactions compared to those with a low stemness index. Differential expression analysis identified genes distinguishing high versus low stemness cells. Through the regression pipeline, six prognostic markers were identified: BOP1, CTBP1, DSE, PMSD10, SRPK1, and HACD4. These markers were significantly associated with tumor cell proliferation and patient survival outcomes. DISCUSSION: Our findings suggest that stemness-related heterogeneity in LMS shapes the tumor immune microenvironment and contributes to disease progression. The six identified prognostic markers not only provide insights into the molecular mechanisms of LMS but also represent potential therapeutic targets for personalized treatment.