Abstract Background Androgen receptor (AR) is expressed in 70-90% of estrogen receptor (ER) positive breast cancer but its expression in triple negative (TN) tumors is less frequent and extremely variable across studies, probably due to their heterogeneity. FOXA1 is essential for expression of 50% of ER-related genes and is a good prognostic factor in luminal tumors. Microarray studies identified the subgroup of molecular apocrine tumors (ER-, AR+) expressing luminal genes including FOXA1, for which pre-clinical data suggested that FOXA1 may direct AR to sites normally occupied by ER in luminal tumors, inducing an estrogen-like gene program stimulating proliferation. We aimed at precising AR expression-associated profiles among a prospective collection of HER2-negative, progesterone receptor (PR)-negative breast tumors. Patients and methods Expression of AR and FOXA1 were evaluated by immunohistochemistry on tissue microarrays in 762 patients enrolled in the PACS08 adjuvant randomized trial evaluating ixabepilone in patients with TN or ER+, PR-, HER2- poor prognosis breast cancer. Two observers evaluated independently AR and FOXA1 expressions. A tumor was considered as positive when at least 10% of the tumor cells were stained. Kappa coefficients were used to assess the agreement between the 2 observers. Categorical clinico-pathological characteristics according to AR and/or FOXA1 expression were compared using Chi-square test or Fisher's exact test when appropriate. Student's t-test and the One way Anova were used to compare continuous variable across patient groups. Results Evaluation of AR and FOXA1 expressions were highly reproducible with inter-observer Kappa coefficients of 0.96 and 0.90 respectively. Both AR and FOXA1 expressions were available in 592 cases, 38% of which were AR+ and 30% FOXA1+. The majority (59.6%) of the tumors was AR-/FOXA1-, 27.9% were AR+/FOXA1+. AR+/FOXA1+ tumors were more frequently: found in older patients (p<0.0001), multifocal (p = 0.02), lobular (p<0.0001), ER+ (p<0.0001), of lower nuclear grade (p<0.0001), and exhibited more in situ component (p = 0.0002), lympho-vascular invasion (p = 0.02) and node involvement (p<0.0001) than other subgroups. Among 487 TN and 136 ER+ tumors, 25.9% and 80.9% respectively were AR+. Among 468 TN and 134 ER+ tumors, 16.4% and 77.6% respectively were FOXA1+. In TN patients, both AR and FOXA1 expressions were available in 460 cases, 72.8% of which were AR-/FOXA1- and 15.2% AR+/FOXA1+. FOXA1+ tumors were more frequently of lobular histology (p = 0.02). AR+/FOXA1+ tumors were more frequently found in older patients (p = 0.0002), had lower nuclear grade (p<0.0001), showed more lympho-vascular invasion (p = 0.001) and more node involvement (p<0.0001). The follow-up is currently too short to report the prognostic value of these markers. Conclusions ER+ tumors were mainly AR+/FOXA1+. In this large series, 25.9% of TN tumors were AR+. AR+/FOXA1+ tumors represent 15.2% of TN tumors and seem to behave like luminal tumors (older patients, lobular histology, lower nuclear grade, node involvement). Anti-androgen therapies are currently under investigation in TN breast cancer; these results stimulate also consideration of FOXA1 as potential therapeutic target. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P5-05-02.