A four-gene signature predicts the efficacy of paclitaxel-based neoadjuvant therapy in human epidermal growth factor receptor 2-negative breast cancer.

Neoadjuvant chemotherapy (NAC) is the major preoperative treatment of breast cancer (BC) with negative human epidermal growth factor receptor 2 (HER2), and the efficacy of NAC and the optimization of regimen are under intensive research. The current study aimed to define the predictive biomarkers for paclitaxel (PTX) response in NAC of HER2-negative BC. Data from GSE25065, GSE26065, GSE41998, as well as drug sensitivity data of breast and ovarian cancer cell line from NCI60, were used. Through logistic regression, COX regression, and correlation analysis with bootstrapping, we found that four genes (CDK8, FAM64A, MARC2, and OCEL1) were associated with drug sensitivity of PTX. The four gene "≥3" model had the best classification accuracy. Subgroup analysis found that the model performed well in the hormone receptor positive, HER2-negative subgroup and did not perform well in the triple-negative subgroup. Decision curve analysis showed that the model could enhance the predictive effect of clinical features. Subsequent gene set enrichment analysis, network analysis showed that these genes may be related to the cell cycle, mitosis and other pathways. The current study demonstrated the promising potential of the novel four-gene signature as a predictive biomarker for pathological complete response of HER2-negative BC patients and indicated the drug sensitivity of PTX.