01.18 The negative impact of androgen receptor expression on overall survival in ER positive breast cancer

T. Takeshita1, M. Okano1, E. Katsuta1, X. Peng2, L. Yan2, K. Takabe1  1Roswell Park Cancer Institute,Surgical Oncology,Buffalo, NY, USA 2Roswell Park Cancer Institute,Biostatistics & Bioinformatics,Buffalo, NY, USA

Introduction: The androgen receptor (AR) is one of the members of the steroid nuclear receptor family, which includes estrogen receptor (ER) and progesterone receptor (PR). AR is expressed in 50–90 % of breast cancers. The role of AR in breast cancer is mechanistically complex and remains controversial. Some in vitro data have shown that androgen and AR have a role in proliferation of normal and malignant breast tissues. High AR expression also demonstrated resistance to tamoxifen and aromatase inhibitors in both in vitro and in vivo systems. The possible mechanism of this resistance was that breast cancer tumor cells could be changed from ER-dependent to AR-dependent. Further, it has been demonstrated that AR supports estradiol-mediated ER activity in ER/AR both positive breast cancer cells. In this study, we investigated the association of AR mRNA and protein expression and patient survival using gene and protein expression data of the publically available large cohort.

Methods:  Clinical, gene and protein expression data were obtained from The Cancer Genome Atlas (TCGA) and METABRIC through cBioPortal. Disease free survival (DFS), overall survival (OS) , gene set enrichment analysis (GSEA) and CIBERSORT analysis were conducted comparing high and low AR expression groups, which were defined as lower quartile based upon previous publications.

Results: AR high and low expression group were 817 and 272 patients in TCGA whole cohort and 1068 and 356 patients in METABRIC whole cohort, respectively. AR expression was significantly higher in ER positive tumors compared to ER negative tumors (p<0.001) in both cohorts. The high expression AR group showed significantly worse OS in ER positive patients in TCGA cohort (p=0.007). In METABRIC cohort, AR high group showed significantly worse OS in Luminal B patients (p=0.007). No significant difference in survival was observed by AR protein expression in TCGA cohort. To explore the mechanism of these results, GSEA was conducted. As expected, it was demonstrated that androgen response related gene set was significantly enriched with AR mRNA high expression (Normalized enrichment score; NES=1.75, p=0.003). Protein secretion related gene set (NES=1.76, p=0.01) and estrogen response related gene set (NES=1.67, p=0.02) were also significantly enriched with high AR. On the other hand, DNA repair related gene sets was significantly enriched in AR low expressed tumors in ER positive tumors (NES=-1.75, p=0.01). In CIBERSORT analysis, AR high tumors were negatively associated with immune-eliminating cells, such as CD8 T-cells, Gamma-Delta T-cells and memory B-cells (p>0.01).

Conclusion: High expression of AR showed worse progress in ER positive breast cancer. High AR expression tumor was enriched estrogen response related gene expression that might associate with worse OS in ER positive patients.