Identification of hub genes and important pathways in determination of breast cancer survival using bioinformatics approaches

Abstract Backgrounds Breast cancer is a highly heterogeneous disorder characterized by dysregulation of expression of numerous genes and cascades. The conventional pathologic classification of breast cancer is not sufficient for the prediction of breast cancer behavior and response to therapy.Methods We have retrieved data of two microarray datasets (GSE65194 and GSE45827) from the NCBI Gene Expression Omnibus database (GEO). R package was used for identification of differentially expressed genes (DEGs), assessment of gene ontology (GO) and pathway enrichment evaluation. The DEGs were integrated to construct a protein-protein interaction (PPI) network. Next, hub genes were recognized using the Cytoscape software and lncRNA-mRNA co-expression analysis was performed to evaluate the potential roles of lncRNAs. The interactive information among DEGs and the PPI network was obtained using the STRING online database. Finally, the clinical importance of the obtained genes was assessed using Kaplan-Meier survival analysis.Results After excluding the outliers from the GSE65194 and GSE45827 datasets and data normalization, 866 DEGs including 712 upregulated and 154 downregulated DEGs were detected between breast cancer and normal samples. Up-regulated DEGs were enriched in six pathways including ‘Cell cycle’, ‘Oocyte meiosis’ and ‘Focal adhesion’. Down-regulated DEGs were enriched in five pathways including ‘Peroxisome-proliferator-activated receptors (PPAR) signaling pathway’, ‘Metabolism of xenobiotics by cytochrome P450’, ‘Adipocytokine signaling pathway’ and ‘Cytokine-cytokine receptor interaction’ pathways. CCNA2, CDK1, MAD2L1, and CCNB2 were significantly enriched in several biological pathways. These four genes showed strong expression in breast cancer samples as compared to normal breast tissue. We also identified 12 lncRNAs with a significant correlation with MAD2L1 and CCNB2 genes. MAD2L1, CCNA2, RAD51-AS1, and LINC01089 have the most prediction potential among all candidate hub genes.Conclusion Our study offers a framework for recognition of the mRNA-lncRNA network in breast cancer and the detection of important pathways that could be used as therapeutic targets in this kind of cancer.