In this study, we aimed to detect promising prognostic factors of breast cancer and interpreted the relevant mechanisms using an integrated bioinformatics analysis. RNA sequencing profile of breast cancer was downloaded from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) databases, which were combined as a group (TCGA_GTEx). GSE70947 dataset was from Gene Expression Omnibus. Blue and turquoise modules, respectively identified in TCGA_GTEx database and GSE70947 dataset using weighted co-expression network analysis (WGCNA), were both notably associated with breast cancer. By comparing genes in the two significant modules with differentially expressed genes (DEGs), we obtained a set of 40 shared genes, which were mainly enriched in chromosome segregation and mismatch repair pathway. After protein-protein interaction (PPI) network and overall survival analysis, two hub genes EXO1 and KIF4A were extracted from the set of 40 shared genes, which were up-regulated and associated with the dismal outcome of breast cancer patients. There was a notable negative correlation between EXO1 and KIF4A expression and age of breast cancer patients, whereas a positive relationship with two another clinical traits stage and tumor category was detected. Univariate and multivariate Cox regression analysis revealed that the two hub genes could be independent prognostic factors of breast cancer. Mechanistically, gene correlation analysis suggested that EXO1 and KIF4A exerted their oncogenic role via promoting breast cancer cell proliferation. Overall, our findings identify two promising individual prognostic predictors of breast cancer and pave the new way for diagnosis and therapy strategy of breast cancer.
Keywords: Bioinformatics; Biomarker; Breast cancer; Prognosis.
Copyright © 2020. Published by Elsevier Inc.