Abstract
High-grade serous ovarian carcinoma (HGSOC) is a gynaecological malignancy that is often fatal. Poor prognosis of HGSOC patients is primarily attributed to concealed initial symptoms, diagnostic challenges, postsurgical recurrence , and chemoresistance. Distinct programmed cell death (PCD) patterns play a pivotal role in tumour progression, serving as valuable predictors for postoperative intervention outcomes in HGSOC. Additionally, they provide insights into HGSOC’s pathogenesis and the exploration of immunomodulatory therapeutic mechanisms. Transcriptome and clinical data were collected from TCGA-OV and the GSE26193 databases. We constructed an ovarian carcinoma death score intervention model using eight genes and machine learning algorithms based on 13 PCD modes (apoptosis, necroptosis, pyroptosis, cuproptosis, ferroptosis,entotic cell death, netotic cell death, parthanatos, lysosome-dependent cell death, autophagy, alkaliptosis, oxeiptosis, and disulfidptosis). Three molecular subtypes of HGSOC with different biological processes were identified using unsupervised clustering models. A nomogram was constructed by combining the cell death index (CDI) with clinical features, which exhibited high predictive performance. The correlation between CDI and immune checkpoint genes, components within the tumour microenvironment, and drug therapy sensitivity was analysed. After multiple dataset validation, the prognosis of HGSOC patients with high CDI was relatively poor. CDI and immune checkpoint genes were related to components of the tumour microenvironment. Patients with HGSOC and high CDI may have resistance to standard adjuvant therapy; therefore, targeting these genes could be a potential therapeutic strategy. Finally, we found that our model had better predictive ability than published models. We conducted a comprehensive analysis of 13 PCD patterns and established a novel CDI model, which can evaluate the prognosis of HGSOC and provide a theoretical basis for its clinical treatment. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1038/s41598-026-42628-3.