Abstract
Drug resistance in breast cancer (BC) is a clinical challenge. Exploring the mechanism and identifying a precise predictive biomarker for the drug resistance in BC is critical. Three first-line drug (paclitaxel, doxorubicin and tamoxifen) resistance datasets in BC from GEO were merged to obtain 1,461 differentially expressed genes for weighted correlation network analysis, resulting in identifying ATRX as the hub gene. ATRX is a chromatin remodelling protein, therefore, ATRX-associated transcription factors were explored, thereby identifying the network of AR, GLI3 and GATA2. GO and KEGG analyses revealed immunity, transcriptional regulation and endocrinotherapy/chemotherapy resistance were enriched. Moreover, CIBERSORT revealed immunity regulation was inhibited in the resistance group. ssGSEA showed a significantly lower immune status in the ATRX-Low group compared to the ATRX-High group. Furthermore, the peaks of H3K9me3 ChIP-seq on the four genes were higher in normal tissues than in BC tissues. Notably, the frequency of ATRX mutation was higher than BRCA in BC. Moreover, depressed ATRX revealed worse overall survival and disease-free survival in the human epidermal growth factor receptor 2 (HER2)-/hormone receptor (HR)+ BC. Additionally, depressed ATRX predicted poor results for patients who underwent endocrinotherapy or chemotherapy in the HER2-/HR+ BC subgroup. A nomogram based on ATRX, TILs and ER exhibited a significantly accurate survival prediction ability. Importantly, overexpression of ATRX significantly inhibited the IC50 of the three first-line drugs on MCF-7 cell. Thus, ATRX is an efficient predictive biomarker for endocrinotherapy and chemotherapy resistance in HER2-/HR+ BC and acts by suppressing the AR, GLI3 and GATA2 transcriptional network.