Role of Androgen Receptor (AR) as a Prognostic Factor in Primary Breast Cancer (BC)
Sharon Nofech-Mozes, Carlos Parra-Herran, Rebecca Dent, Cynthia Villarreal, Mary Anne Quintayo, John MS Bartlett, Jane Starczynski, Ping Sun, Steven Narod, Wedad Hanna. University of Toronto, Toronto, Canada; Sunnybrook Health Sciences Centre, Toronto, Canada; Ontario Institute for Cancer Research, Toronto, Canada; Women's College Hospital, Toronto, Canada
Background: There is evidence that androgen signaling pathway may play a role in breast carcinogenesis through the activation of estrogen-responsive genes. In BC cell lines, when lacking estrogens, androgens stimulate tumor growth by binding to estrogen receptor (ERα). When estrogens are present, however, androgens act as their antagonists and inhibit the growth of the tumor; this is probably mediated by AR and can be blocked by antiandrogens. Our aim is to determine the frequency of AR expression in BC and correlate it with established clinicopathological prognostic factors and outcome.
Design: We studied 1008 primary BC from the Henrietta Banting database, with a mean follow up of 9.1 yrs, treated in a single institution. Immunohistochemistry for AR, ER, progesterone receptor (PR) and HER2 was applied to 1 mm core tissue microarrays in triplicates. Interpretation of AR was performed using the Ariol platform (Leica Biosystems 2012) which produced a histoscore [(% cells staining 1+)x1 +(%staining 2+)x2 + (%staining 3+)x3 for a total ranging from 0 to 300]. 152 cases were also scored by two pathologists and manual and automated readings were compared. For analysis, cases were divided in AR-rich (histoscore ≥200) and AR-poor (histoscore <200). The association between AR expression and standard prognostic factors (age, tumor size, nodal status, ER, PR, and HER2 status) was examined. Relative risk of death was calculated for age, tumor size and AR expression in univariate and multivariate analyses. Estimates were considered statistically significant for two-tailed values of P<0.05.
Results: The interclass correlation coefficient for the manual vs the automated reading was 0.9195. Among 1008 cases, 256 (25.4%) were AR-poor and 752 were (74.6%) AR-rich. AR-rich status was significantly associated with older age (57 vs 52 years, p<0.0001), smaller tumor size (2.54 vs 2.96 cm, p=0.001), ER+ (90.4 vs 45.5%, p<0.0001) and PR+ (72.4 vs 26.8%, p<0.0001), but not with nodal status (p=0.67) or HER2+ (p=0.53). AR rich status was associated with better BC specific survival (75% vs 61.3%, p<0.0001). Multivariate Cox-regression revealed a relative risk of 0.67 (95%CI 0.52-0.85, p=0.001) for AR expression.
Conclusions: AR is frequently expressed in our population. A histoscore cut-off value of >200 demonstrated significant association with multiple prognostic clinicopathologic factors and BC specific survival.Our study indicates that high AR expression is a prognostic factor in BC.
Tuesday, March 5, 2013 9:30 AM
Poster Session III # 4, Tuesday Morning