A novel twelve-gene signature to evaluate neoadjuvant chemotherapy response and predict prognosis in breast cancer

Abstract Background Accurate evaluation of breast cancer response to neoadjuvant chemotherapy (NAC) provides important information about systemic therapies, which elucidates tumor biology and prognosis and guides further therapies. Gene profiles overcome the limitations of some indicators of the classical pathological evaluation criteria and the subjectivity of observation, but are complicated and expensive. Therefore, it is essential to develop a more accurate, repeatable, and economical evaluation approach for NAC responses. Methods We analyzed the transcriptional profiles of epirubicin-resistant breast cancer cell lines and the tumors derived from NAC-resistant patients from the GEO25066 dataset. We initially screened common significantly differentially expressed genes and constructed a NAC response risk model using LASSO regression and univariate and multivariate analyses. Bioinformatics Analysis were performed to find the differences in bioinformatic features of tumor cells, immune characteristics, and prognosis between the high- and low-risk groups. We screened candidate drugs that could reverse chemotherapy resistance in breast cancer by the Connectivity Map database. Cytotoxicity assay and TUNEL assay were performed to verify the killing effect of drug candidates on drug-resistant breast cancer cells. Results Thirty-six genes were commonly up/down-regulated in both the chemo-resistant tumors and cell lines, compared to the sensitive tumors and wild-type cell lines. We obtained a risk model composed of 12 genes trough LASSO regression. The risk model divided patients into high- and low-risk groups. Univariate and multivariate Cox regression analyses suggested that the risk score is an independent prognostic factor for the evaluation of NAC response to breast cancer. Tumors in these risk groups exhibited significant differences in molecular biological characteristics, tumor-infiltrating lymphocytes, and immunosuppressive molecule expression. Our results suggested that the risk score was also a good prognostic factor for breast cancer. Finally, our results suggest that pravastatin, isocarboxazid, imexon, axitinib, and crizotinib could inhibit or kill chemoresistant cells. Conclusion A novel 12 gene-signature could be used to evaluate NAC response and predict prognosis in breast cancer.