An eight-lncRNA signature predicts survival of breast cancer patients: a comprehensive study based on weighted gene co-expression network analysis and competing endogenous RNA network

• DOI: 10.1007/s10549-019-05147-6
• Publication Date: 2019-02-04T07:10:21Z
• Authors: Min Sun; Di Wu; Ke Zhou; Heng Li; Xingrui Gong; Qiong Wei; Mengyu Du; Peijie Lei; Jin Zha; Hongrui Zhu; Xinsheng Gu; Dong Huang
• Keywords: 0301 basic medicine; Gene Expression Profiling; Reproducibility of Results; Breast Neoplasms; Kaplan-Meier Estimate; Middle Aged; Prognosis; 3. Good health; Gene Expression Regulation, Neoplastic; MicroRNAs; 03 medical and health sciences; ROC Curve; Protein Interaction Mapping; Biomarkers, Tumor; Humans; Female; Gene Regulatory Networks; RNA Interference; RNA, Long Noncoding; Protein Interaction Maps; RNA, Messenger; Aged; Proportional Hazards Models

To identify a lncRNA signature to predict survival of breast cancer (BRCA) patients.A total of 1222 BRCA case and control datasets were downloaded from the TCGA database. The weighted gene co-expression network analysis of differentially expressed mRNAs was performed to generate the modules associated with BRCA overall survival status and further construct a hub on competing endogenous RNA (ceRNA) network. LncRNA signatures for predicting survival of BRCA patients were generated using univariate survival analyses and a multivariate Cox hazard model analysis and validated and characterized for prognostic performance measured using receiver operating characteristic (ROC) curves.A prognostic score model of eight lncRNAs signature was identified as Prognostic score = (0.121 × EXPAC007731.1) + (0.108 × EXPAL513123.1) + (0.105 × EXPC10orf126) + (0.065 × EXPWT1-AS) + (- 0.126 × EXPADAMTS9-AS1) + (- 0.130 × EXPSRGAP3-AS2) + (0.116 × EXPTLR8-AS1) + (0.060 × EXPHOTAIR) with median score 1.088. Higher scores predicted higher risk. The lncRNAs signature was an independent prognostic factor associated with overall survival. The area under the ROC curves (AUC) of the signature was 0.979, 0.844, 0.99 and 0.997 by logistic regression, support vector machine, decision tree and random forest models, respectively, and the AUCs in predicting 1- to 10-year survival were between 0.656 and 0.748 in the test dataset from TCGA database.The eight-lncRNA signature could serve as an independent biomarker for prediction of overall survival of BRCA. The lncRNA-miRNA-mRNA ceRNA network is a good tool to identify lncRNAs that is correlated with overall survival of BRCA.