Integrated Multi-Omics and Machine Learning Identify PTTG1 as a Prognostic Biomarker and Therapeutic Target in Endometrial Cancer

Abstract Background Endometrial carcinoma (UCEC) exhibits high molecular heterogeneity, and the lack of precise prognostic biomarkers and therapeutic targets severely limits the clinical management of high-risk patients. This study aimed to identify a robust multi-omics biomarker and elucidate its intrinsic oncogenic mechanisms and extrinsic immune-remodeling roles in the UCEC microenvironment. Methods We systematically integrated transcriptomic data from TCGA and GEO cohorts. WGCNA coupled with four machine learning algorithms (LASSO, SVM-RFE, XGBoost, Boruta) were employed for feature selection. Single-cell RNA sequencing and spatial transcriptomics were integrated to map the cellular landscape. In silico virtual knockout, virtual overexpression, and pharmacogenomic analyses were conducted. Computational predictions were experimentally validated using Western blot, qRT-PCR, and immunohistochemistry in paired clinical specimens, and corroborated by CPTAC proteomic analysis. Results PTTG1 was identified as an independent prognostic biomarker, with significant overexpression confirmed at both mRNA and protein levels and predominant nuclear localization in tumor epithelium. Mechanistically, virtual knockout revealed PTTG1 drives malignant progression through hyperactivation of cell cycle, ribosome biogenesis, and glutathione metabolism, whereas virtual overexpression activated immune-related transcriptional networks including MHC class II antigen presentation. Integrated scRNA-seq and spatial transcriptomics demonstrated PTTG1 enrichment in malignant epithelial cells and tumor-associated macrophages. Drug sensitivity predictions identified selective vulnerability to MEK inhibitors, including selumetinib and trametinib. Conclusion PTTG1 serves as a multidimensional oncogenic driver and independent prognostic biomarker in UCEC, intrinsically fueling tumor proliferation while extrinsically orchestrating an immunosuppressive microenvironment. Bidirectional in silico perturbation revealed distinct transcriptional consequences, substantiated by multi-level experimental validation. Drug sensitivity predictions identified selective MEK inhibitor vulnerabilities, providing a rationale for PTTG1-guided prognostic stratification and targeted therapeutic strategies in high-risk UCEC.