Development of a Prognostic Model Based on the Identification of EMT-Related lncRNAs in Triple-Negative Breast Cancer

BACKGROUND: Triple-negative breast cancer (TNBC) remains the most incurable subtype of breast cancer owing to high heterogeneity, aggressive nature, and lack of treatment options. It is generally acknowledged that epithelial-mesenchymal transition (EMT) is the key step in tumor metastasis. METHODS: With the application of TCGA and GEO databases, we identified EMT-related lncRNAs by the Cox univariate regression analysis. Optimum risk scores were calculated and used to divide TNBC patients into high-/low-risk subgroups by the median value using the Lasso regression analysis. The Kaplan–Meier and ROC curve analyses were applied for model validation. Then, we assessed the risk model from multi-omic aspects including immune infiltration, drug sensitivity, mutability spectrum, signaling pathways, and clinical indicators. We also analyzed the expression pattern of lncRNAs involved in the model using qRT-PCR in TNBC cell lines and constructed the ceRNA network. RESULTS: The risk model was composed of EMT-related long noncoding RNAs (lncRNAs), which seemed to be valuable in the prognostic prediction of TNBC patients. The model could act as an independent prognostic factor of TNBC and showed a robust prognostic ability in the stratification analysis. Further investigation demonstrated that the expression of lncRNAs was different between high aggressive and low aggressive TNBC cell lines, as well as TNBC patients. CONCLUSIONS: Together, our study successfully established a risk model with great accuracy and efficacy in the prognostic prediction of TNBC patients.