Abstract
BACKGROUND: Ovarian cancer (OV) is characterized by the highest mortality rate among gynecological malignancies. Suboptimal early diagnosis and ineffective prognostic prediction of OV contribute to poor survival outcomes for most patients. This study aimed to identify immune-related programmed cell death (IPCD) signatures and find the valuable biomarker for predicting OV prognosis. METHODS: All OV datasets were downloaded from public databases of TCGA, GEO and ICGC. Prognostic genes from IPCD-related differential genes were screened by univariate cox regression analysis. The construction of the IPCDS model was performed via 101 algorithm combinations. The prognostic performance of IPCDS model were examined by Kaplan-Meier analysis and timeROC curves. TIDE algorithm was used to predict the immune response of TCGA data. RESULTS: 88 IPCD-related prognostic genes were screened for modeling IPCDS. A significantly higher OS in low-IPCDS group was observed among most OV datasets than in high-IPCDS group. 2-, 3-, and 5-year timeROC results of each OV dataset revealed the excellent predictive value of IPCDS for OV, showing a relatively higher AUC after treated 2 years. The low IPCDS group had a higher TIDE value, while the no response group had a higher IPCDS value. In the pan-cancer immunotherapy dataset, patients in the low IPCDS group had a longer overall survival period and a significant immunotherapy effect. Besides, model gene PDGFRA were found to have predictive performance for OV. We validated the upregulation of PDGFRA and C-MYC in ovarian cancer tissues through Western blot and confirmed their co-localization and immunofluorescence analyses. CONCLUSION: Our study develops a novel prognostic model IPCDS for OV. IPCD-related gene PDGFRA serves as a prognosis factor play in predicting survival outcomes of OVs.