In silico analysis of differentially expressed genesets in metastatic breast cancer identifies potential prognostic biomarkers

Abstract

Abstract Background Identification of specific biological functions, pathways, and appropriate prognostic biomarkers is essential to accurately predict the clinical outcomes of and apply efficient treatment for breast cancer patients. Methods To search for metastatic breast cancer-specific biological functions, pathways, and novel biomarkers in breast cancer, gene expression datasets of metastatic breast cancer were obtained from Oncomine, an online data mining platform. Over- and under-expressed genesets were collected and the differentially expressed genes were screened from four datasets with large sample sizes (N > 200). They were analyzed for gene ontology (GO), KEGG pathway, protein-protein interaction, and hub gene analyses using online bioinformatic tools (Enrichr, STRING, and Cytoscape) to find enriched functions and pathways in metastatic breast cancer. To identify novel prognostic biomarkers in breast cancer, differentially expressed genes were screened from the entire twelve datasets with any sample sizes and tested for expression correlation and survival analyses using online tools such as KM plotter and bc-GenExMiner. Results Compared to non-metastatic breast cancer, 193 and 144 genes were differentially over- and under-expressed in metastatic breast cancer, respectively, and they were significantly enriched in regulating cell death, epidermal growth factor receptor signaling, and membrane and cytoskeletal structures according to the GO analyses. In addition, genes involved in progesterone- and estrogen-related signalings were enriched according to KEGG pathway analyses. Hub genes were identified via protein-protein interaction network analysis. Moreover, four differentially over-expressed (CCNA2, CENPN, DEPDC1, and TTK) and three differentially under-expressed genes (ABAT, LRIG1, and PGR) were further identified as novel biomarker candidate genes from the entire twelve datasets. Over- and under-expressed biomarker candidate genes were positively and negatively correlated with the aggressive and metastatic nature of breast cancer and were associated with poor and good prognosis of breast cancer patients, respectively. Conclusions Transcriptome datasets of metastatic breast cancer obtained from Oncomine allow the identification of metastatic breast cancer-specific biological functions, pathways, and novel biomarkers to predict clinical outcomes of breast cancer patients. Further functional studies are needed to warrant validation of their roles as functional tumor-promoting or tumor-suppressing genes.