Identification and validation of a gene-based signature reveals SLC25A10 as a novel prognostic indicator for patients with ovarian cancer

Abstract

Background: Ovarian cancer is a common gynecological cancer with poor prognosis and poses a serious threat to woman life and health. In this study, we aimed to establish a prognostic signature for the risk assessment of ovarian cancer.

Methods: The Cancer Genome Atlas (TCGA) dataset was used as the training set and the International Cancer Genome Consortium (ICGC) dataset was set as an independent external validation. A multi-stage screening strategy was used to determine the prognostic features of ovarian cancer with R software. The relationship between the prognosis of ovarian cancer and the expression level of SLC25A10 was selected for further analysis.

Results: A total of 16 prognosis-associated genes were screened to construct the risk score signature. Survival analysis showed that patients in the high-risk score group had a poor prognosis compared to the low-risk group. Accuracy of this prognostic signature was confirmed by the receiver operating characteristic (ROC) curve and decision curve analysis (DCA), and validated with ICGC cohort. This signature was identified as an independent factor for predicting overall survival (OS). Nomogram constructed by multiple clinical parameters showed excellent performance for OS prediction. Finally, it's found that patients with low expression of SLC25A10 generally had poor survival and higher resistance to most chemotherapeutic drugs.

Conclusions: In sum, we developed a 16-gene prognostic signature, which could serve as a promising tool for the prognostic prediction of ovarian cancer, and the expression level of SLC25A10 was tightly associated with OS of the patients.

Keywords: Ovarian Cancer; Prognosis; Risk Score; SLC25A10.

Fig. 1

The workflow chart of this study. A flow chart of the systematic identification and validation of the 16-gene signature for the prognostic prediction of ovarian cancer

Fig. 2

Identification of risk score based on gene signature of patients with ovarain cancer in TCGA. A Risk plot of each point sorted based on risk score, representing one patient. Green and red points represent patients with low- and high-risk scores, respectively. B Distribution of risk score and significant genes of ovarian cancer in TCGA. C Kaplan–Meier analysis of ovarian cancer patients was stratified by median risk in TCGA. High risk scores are associated with poor survival. D ROC curves of risk score in prognosis prediction of ovarian cancer in TCGA. E DCA curves of risk score in prognosis prediction of ovarian cancer in TCGA

Fig. 3

Characterization of prognostic signature in ovarian cancer of TCGA. (A, B) Univariate (A) and multivariate (B) Cox proportion hazard regression for OS of ovarian cancer in training group of TCGA. (C) Multi-index ROC curve of risk score and other indicators for predictions of survival time in ovarian cancer patients. (D-F) Boxplots showing the distribution of risk score in ovarian cancer samples stratified by different factors, including age (D), stage (E) and grade (F)

Fig. 4

Validation of risk score based on gene signature of patients with ovarain cancer in ICGC. A Risk plot of each point sorted based on risk score, representing one patient. Green and red points represent patients with low- and high-risk, respectively. B Distribution of risk score and patient survival time of ovarian cancer in ICGC. C Kaplan–Meier analysis of ovarian cancer patients was stratified by median risk in ICGC. D ROC curves of risk score in prognosis prediction of ovarian cancer in ICGC

Fig. 5

Contruction of a nomogram to predict the patient survival in ovarian cancer. A The nomogram using age, stage, grade and risk score to predict the OS of patients with ovarian cancer in TCGA. B-D The calibration plot to evaluate the nomogram. Y-axis, actual survival. X-axis, predicted survival of 1-year (B), 3- year (C), and 5-year (D), respectively. The solid line represents the predicted nomogram

Fig. 6

The correlation of SLC25A10 with the overall survival and molecular functions in patients with ovarian cancer. A Kaplan–Meier survial analysis of SLC25A10 with ovarian cancer using the information from TCGA dataset. Patients are divided into low and high SLC25A10 groups by median expression level. B-D Results of GO analysis showing the consistently altered gene profiles with SLC25A10 in TCGA dataset, including biological process (B), cellular component (C) and molecular function (D)

Fig. 7

The sensitivity analysis of SLC25A10 and multiple chemotherapeutic drugs. The top 16 drugs with high correlation with SLC25A10 expression were demonstrated