Novel biomarkers and prediction model for the pathological complete response to neoadjuvant treatment of triple-negative breast cancer

Objective: To develop and validate a prediction model for the pathological complete response (pCR) to neoadjuvant chemotherapy (NCT) of triple-negative breast cancer (TNBC). Methods: We systematically searched Gene Expression Omnibus, ArrayExpress, PubMed gene expression profiles operable TNBC accessible NCT. Molecular heterogeneity was detected with hierarchical clustering method, biological differentially expressed genes were investigated by Ontology, Kyoto Encyclopedia Genes Genomes analyses, Set Enrichment Analysis (GSEA). Next, machine-learning algorithms including random-forest analysis least absolute shrinkage selection operator (LASSO) synchronously performed and, then, intersected proportion significant undergone binary logistic regression fulfill variables selection. The predictive score (pRS) system built as product coefficient obtained from analysis. Last, cohorts randomly divided in 7:3 ratio into training cohort validation introduction robust model, nomogram constructed independent predictors pCR rate. Results: A total 217 individuals four datasets (GSE32646, GSE25065, GSE25055, GSE21974) clinicopathological information included. Based on microarray data, six-gene panel (ATP4B, FBXO22, FCN2, RRP8, SMERK2, TET3) identified. nomogram, adopting pRS clinical tumor size stage, established performance successively validated calibration curves receiver operating characteristic area under curve 0.704 0.756, respectively. Results GSEA revealed that processes apoptosis, hypoxia, mTORC1 signaling myogenesis, oncogenic features EGFR RAF proactivity attribute an inferior response. Conclusions: This study provided rate potential mechanisms distinct NCT TNBC, which promising warranted further perspective.