Integrated analysis of single-cell and bulk transcriptomics reveals the prognostic value and underlying mechanisms of crotonylation in ovarian cancer

Background Ovarian cancer remains the deadliest gynecological malignancy with 5-year survival rates below 40% due to frequent recurrence and chemoresistance. Aberrant crotonylation, a type of epigenetic modification, has been implicated in the proliferation, metastasis, and immune evasion of various cancers. However, its role in the ovarian cancer microenvironment and clinical outcomes remains unexplored. The aim of this study was to develop a prognostic model for ovarian cancer on the basis of crotonylation and to investigate the underlying mechanisms and potential of crotonylation for targeted therapy. Methods We systematically analyzed single-cell RNA-seq and bulk transcriptomic datasets from ovarian cancer patients. Cellular crotonylation activity was quantified using AUCell algorithm. Potential prognostic genes were identified through DEG analysis and Weighted gene correlation network analysis (WGCNA), and the associated molecular mechanisms were elucidated via Gene set enrichment analysis (GSEA). An ovarian cancer prognosis model were constructed by integrating machine learning algorithms. Immune microenvironment features were assessed using CIBERSORT, ESTIMATE and TIDE algorithms, with drug sensitivity predicted via genomics of drug sensitivity in cancer. Results The ovarian cancer microenvironment is characterized by abundant immune cell infiltration, with significant differences in crotonylation levels among 7 cell subtypes. We identified 451 key crotonylation-related genes. The crotonylation risk score (RS) model demonstrated robust prognostic performance. High-RS groups showed immunosuppressive characteristics: decreased follicular helper T cells and activated NK cells, concomitant with M2 macrophage enrichment. Elevated RS was associated with increased stromal activation, as indicated by a higher ESTIMATE score, and enhanced immune evasion potential, reflected by an elevated TIDE score. Notably, high-RS patients exhibited upregulated PDL1 and CD40, suggesting increased immunotherapy susceptibility. Pharmacogenomic analysis identified vinblastine with differential sensitivity, providing actionable targets for RS-stratified therapy. Conclusion We elucidated the significant impact of crotonylation on the ovarian cancer microenvironment and prognosis. We developed and validated a novel prognostic model for ovarian cancer that can serve as a tool for predicting patient outcomes and characterizing the immune microenvironment. These findings enhance our understanding of the role of crotonylation in ovarian cancer and establish a robust framework for developing therapeutic strategies targeting crotonylation.