Abstract: Objective To construct a novel immune-related prognostic model of ovarian cancer and to preliminarily screen the prognostic biomarkers. Methods The ovarian cancer samples from the TCGA database were randomly divided into a training cohort and a testing cohort according to the ratio of 7∶3, with GSE26712 being external validate cohort. The immune-related differential expressed genes (IRDEGs) were analyzed by “limma” package and the prognostic IRDEGs were screened by univariable Cox regression. The signatures were constructed by robust LASSO and multivariable Cox regression. Additionally, the receiver operating characteristic (ROC) curves and C-index were employed to evaluate the signatures. A nomogram was established, and the prediction performance of the nomogram was evaluated by calibration curves and decision curves. Results The 11-gene signature (C5AR1, CX3CR1, CXCL11, CXCL13, IGF1, IL27RA, NFKBIB, PENK, PI3, PSMC1 and PSME3) was successfully constructed in the training cohort with C-index of 0.69, the area under curves (AUC) at 1-, 3- and 5-year were 0.67, 0.71 and 0.75, respectively. Multivariable Cox regression showed that the risk signature was an independent prognostic factor for ovarian cancer patients (HR=2.58, 95%CI=2.15-3.25). Based on the risk score, the nomogram signature was successfully constructed, agreeing well with calibration curves, and the decision curve showed that it had a positive net benefits in guiding clinical decision-making. Conclusions The novel immune-related prognosis signature constructed in this study has good efficacy in predicting the prognosis of ovarian cancer, and the related genes are the potential biomarkers for immunotherapy for ovarian cancer patients.

Key words: Ovarian cancer, Bioinformatics analysis, Prognostic signature, Prognostic biomarker, TCGA