Few prognostic indicators with differential expression have been reported among the differing ER statuses. We aimed to screen important breast cancer prognostic genes related to ER status and to construct an efficient prognostic prediction system. mRNA expression profiles were downloaded from TCGA and GSE70947 dataset. Two hundred seventy-one overlapping differentially expressed genes (DEGs) between the ER- and ER+ breast cancer samples were identified. Among the 271 DEGs, 109 prognostically relevant mRNAs were screened. mRNAs such as RASEF, ITM2C, CPEB2, ESR1, ANXA9, and VASN correlated strongly with breast cancer prognosis. Three modules, which contained 28, 9 and 8 enriched DEGs, were obtained from the network, and the DEGs in these modules were enriched in response to hormone stimulus, epithelial cell development, and host cell entry. Using bayes discriminant analysis, 48 signature genes were screened. We constructed a prognostic prediction system using the 48 signature genes and validated this system as relatively accurate and reliable. The DEGs might be closely associated with the prognosis in patients with breast cancer. We validated the effectiveness of our prognostic prediction system by GEO database. Therefore, this system might be a useful tool for preliminary screening and validation of potential prognosis indicators for ER+ breast cancer derived from mechanistic research.
Keywords: Breast cancer; Differentially expressed genes; Estrogen receptor; Prognosis; Prognostic prediction system.