Identification and validation of glycolysis-related signatures for prognosis in patients with breast cancer

Abstract Background: Emerging evidence has demonstrated roles of glycolysis in the tumorigenesis and progression of human tumors. However, their underlying clinical implications have not been well elucidated in breast cancer. In present study, we aimed to generate a risk-score from glycolysis-related signatures to predict prognosis of patients with breast cancer. Methods: We acquired mRNAs expression and clinical datasets in patients with breast cancer from The Cancer Genome Atlas (TCGA), then identifying glycolysis-related mRNAs by Gene Set Enrichment Analysis (GSEA), followed by construction of prognostic risk-score. The altered expression of glycolysis-related mRNAs was identified as candidates for further investigation. We constructed a risk-score from the prognostic glycolysis-related mRNAs by Cox regression. Receiver Operating Characteristic (ROC) and clinical subgroups analysis were performed to evaluate the values of risk-score to predict prognosis of breast cancer. Besides, we also compared the expression patterns of the signatures in breast cancer tissues and cell lines.Results: Total of 1208 cases were obtained, including 112 normal tissues and 1096 tumor tissues. We found 4 glycolysis-related pathways significantly involved in breast cancer. And 298 mRNAs involved in the 4 pathways were defined as glycolysis-related mRNAs; of these, 241 dysregulated mRNAs were candidates for further exploration. Then we constructed a risk-score from the 5 candidates (IL13RA1, PGK1, SDC3, NUP43 and SDC1). The area under the curve (AUC) for the risk-score to predict prognosis was 0.729. Patients with high-risk score had poor prognosis among overall or clinical subgroups ( P <0.05). And IL13RA1, PGK1, NUP43 and SDC1 were up-regulated in tumor tissues and cell lines (MDA-MB-231 and BT474) as compared to normal tissues and cell line (MCF-10A), while SDC3 was down-regulated.Conclusions: We construct a risk-score based on 5 glycolysis-related signatures, which can well predict prognosis in breast cancer. Additionally, our findings further unveil the molecular mechanisms of glycolysis in cancer, providing promising directions for the prognostic and therapeutic biomarkers for breast cancer.