Purpose: To evaluate whether tumor androgen receptor (AR) expression was prognostic and/or predictive for endocrine treatment alone or in combination with estrogen receptor (ER). The AR has been hypothesized to have differential prognostic roles in breast cancer depending on tumor ER status, and to influence endocrine treatment response.
Experimental design: A population-based prospective cohort of 1,026 patients diagnosed with primary invasive breast cancer in Lund, Sweden, between 2002 and 2012 was followed until June 2014. Associations between immunohistochemical AR expression in tumor tissue microarrays, patient and tumor characteristics, and AR genotypes were analyzed. Disease-free survival (DFS) by AR status, and combined ER/AR status was assessed in various treatment groups.
Results: AR expression was assessable in 913 tumors. AR(+) tumors (85.0%) were associated with higher age (P = 0.036) and favorable tumor characteristics. The AR(+) status was a prognostic marker for DFS (LogRank P = 0.025). There was an interaction between AR and ER expression with respect to prognosis (adjusted P(interaction) ≤ 0.024). Tumors with discordant hormone receptor expressions (ER(+)AR(-) or ER(-)AR(+)) demonstrated worse prognosis compared with concordant tumor expressions (ER(+)AR(+) or ER(-)AR(-)) in multivariable models [adjusted HRs (95% confidence intervals); ≥ 1.99 (1.28-3.10), P ≤ 0.002]. ER(+)AR(-) indicated early treatment failure with aromatase inhibitors (AI) among chemonaïve patients aged 50 or older.
Conclusions: Prediction of breast cancer prognosis and treatment response was improved by combining AR and ER status. AR negativity predicted early treatment failure with AI but not tamoxifen, a finding that warrants confirmation in a randomized setting. Patients may benefit from anti-androgens or selective AR modulators.
©2015 American Association for Cancer Research.