Purpose: To construct a competing endogenous RNA (ceRNA) topology network of RNA-seq data and micro RNA-seq (miRNA-seq) data to identify key prognostic long non-coding RNA (lncRNAs) in luminal breast cancer, and validate the results by human luminal breast cancer samples.
Materials and Methods: The RNA-seq data and miRNA-seq data of luminal A breast cancer in the The Cancer Genome Atlas (TCGA) database were downloaded and compared with those in the miRcode database to obtain lncRNA–miRNA relationship pairs. Final target genes were predicted by all three databases (miRDB, miRTarBase, and TargetScan), thereby obtaining the miRNA-messenger RNA (miRNA-mRNA) relationship pairs and a ceRNA topology network was constructed, then mRNA enrichment analysis, ceRNA topological and stability analysis, univariate and multivariate Cox regression analysis were performed. Overall survival (OS) was evaluated and the key prognostic RNAs were identified. The expression difference between normal and tumor, as well as the correlation of high expression in tumor with pathological parameters (Ki-67, Grade, tumor diameter) were validated by human breast cancer specimens.
Results: A ceRNA topology network was constructed and six lncRNAs were finally identified (The higher expression of PART1, IGF2.AS, WT1.AS, OIP5.AS1, and SLC25A5.AS1  was associated with poor prognosis while AL035706.1 was adverse) and the poor prognostic ones were higher expressed in tumor tissue and correlated with a higher Ki-67 (> 10%), tumor grades (II, III) and tumor diameters (> 1.5 cm). Using six lncRNAs, we constructed a prognostic model, which performed well for the classification of prognosis in the module.
Conclusion: We identified and verified six biomarkers (OS-predicting) in luminal breast cancer, which significantly enriched the prediction and potential targets of this subtype.
Keywords: competing endogenous RNA, breast cancer, long non-coding RNA, prognosis, clinical sample