Triple-negative breast cancer (TNBC) is a high-risk subtype of breast cancer defined by negative expression of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2. Accumulating evidence indicates that alternative splicing (AS) events are correlated with the prognosis of cancer. RNA sequencing data and AS event data were manually curated from The Cancer Genome Atlas (TCGA) dataset and TCGA Splice Seq, respectively. Univariate and multivariate Cox regression analyses were applied to screen AS events associated with TNBC survival and to establish a prognostic model. A receiver operating characteristic (ROC) curve was used to evaluate the performance of the prognostic model. Differentially expressed gene analysis and functional enrichment analysis were harnessed to reveal the functional role of gene sets and to screen novel biomarkers. By integrated bioinformatics analysis of AS events and gene expression in TNBC, our study is the first to generate specific AS event profiles, prognostic AS event interaction networks, and splice factor-AS interaction networks for TNBC. Surprisingly, we found that the performance of the AS-based prognostic model was encouraging with a mean area under the ROC curve of 0.957 at 2-10 years. We also found that chemokine (C-C motif) ligand 16 (CCL16) expression was correlated with TNBC grade and could be a potential novel biomarker. In conclusion, this study provided a systematic analysis of prognostic AS event profiles and gene expression in TNBC. A novel prognostic model based on AS events may establish a foundation for future research investigating the diagnosis and treatment of TNBC.
Keywords: TNBC; alternative splicing; biomarker; prognostic signature.