Identification of a novel signature with prognostic value in triple-negative breast cancer through clinico-transcriptomic analysis

Although perceived as a highly aggressive disease, triple-negative breast cancer (TNBC) constitutes heterogeneous features with various outcomes. In this study, we aimed to establish a prognostic signature for patients with TNBC to improve risk stratification. Gene expression data were obtained from The Cancer Genome Atlas (TCGA). Differentially expressed genes (DEGs) were detected pairwise between TNBC and other subtypes of samples. Then, TNBC-correlated modules were determined using coexpression network analysis. A gene signature was established based on the prognostic genes in the intersection between DEGs and selected gene modules using least absolute shrinkage and selection operator (LASSO) Cox regression. Finally, a clinico-transcriptomic signature was developed to predict overall survival (OS). Model performance was quantified, and the bootstrap resampling method was used for validation. The gene signature included 6 messenger RNAs (mRNAs) and a clinical score indicating an increased likelihood of death when used as continuous or categorical predictors. A nomogram was built by integrating the pathological stage and gene signature to predict 2-, 3-, and 5-year OS. The addition of pathological stage increased the concordance index (C-index) compared with pathological stage alone and the gene signature alone. Bootstrap resampling revealed a stable performance of the nomogram. A 6-mRNA signature was established to inform prognosis for patients with TNBC. Its combination with pathological stage can contribute to improving performance and provide additional supporting evidence for clinical decision-making.