BACKGROUND: Our goal was to evaluate the impact of level of androgen receptor (AR) expression on outcomes in women with estrogen receptor α (ER) positive breast cancer. We sought to corroborate our preclinical findings that AR-agonists were efficacious in patients with ER-positive tumors that also expressed high levels of AR. METHODS: Tissue microarrays (TMAs) were prepared from primary tumor blocks from patients entered on a prospective randomized adjuvant trial of tamoxifen (Tam) alone or combined with fluoxymesterone (Flu), an AR-agonist, (NCCTG 89-30-52). TMAs were stained for ER and AR and expression examined in decile increments (0-100%) of positive invasive tumor nuclei. The primary endpoint was relapse-free survival (RFS). RESULTS: 301 (59%) of the 514 patients had sufficient tissue to determine ER and AR expression, where nuclear staining of >
70% was considered "enriched" and nuclear staining of ≤ 70% was considered "poor/moderate". Eleven (4%) of these patients had poor/moderate ER staining and were excluded from these analyses. The proportion of the ER-enriched tumors that also had AR-enriched expression levels was 56.3% in the Tam arm and 51.8% in the Tam + Flu arm. Within the AR-enriched patients, the cumulative incidence of RFS events showed an advantage for Tam + Flu over Tam alone that reached significance (Gray's test p = 0.0472). CONCLUSIONS: Our findings suggest that an AR-agonist may be of value in AR-enriched, ER-enriched breast cancers and should be studied in future trials because of the availability of new, more tolerable AR-agonists.