Anatomic Pathology / ANDROGEN RECEPTOR IN BREAST CANCER Androgen Receptor Expression in Estrogen Receptor Negative Breast Cancer Immunohistochemical, Clinical, and Prognostic Associations S. Nicholas Agoff, MD, 1 Paul E. Swanson, MD, 1 Hannah Linden, MD, 2 Stephen E. Hawes, PhD, 3 and Thomas J. Lawton, MD 1 Key Words: Androgen receptor; Breast cancer; Estrogen receptor negative Abstract We sought to determine the prevalence of androgen receptor (AR) expression in a predominantly estrogen receptor (ER)-negative subset of breast cancers and delineate the immunohistochemical and clinical associations, including whether AR expression has prognostic significance in ER-negative tumors. We identified 69 ER-negative and 19 ER-positive breast cancer cases with concurrent immunohistochemical prognostic panels (ER, PR, HER-2/neu, Ki-67, and p53); immunohistochemical analysis was performed for AR using standard techniques. Clinical data were extracted from medical records. χ 2 tests were used to assess associations between variables. AR was found in 49% (34/69) of ER-negative and 89% (17/19) of ERpositive cases. In ER-negative tumors, AR was associated with increased age (P =.02), postmenopausal status (P <.001), tumor grade (P =.03), tumor size (P =.03), and HER-2/neu overexpression (P =.003). In ER-positive tumors, AR was associated with progesterone receptor expression (P <.03). In univariate analysis of ER-negative tumors, patients with AR-positive tumors had significantly better disease-free survival (P =.049). AR is expressed in a significant number of ER-negative cases and shows significant associations with important clinical and pathologic prognostic factors. The role of hormone receptors as prognostic and therapeutic tools has widespread acceptance in the management of breast cancer. The expression of estrogen receptor (ER), in particular, is thought to be of great importance, predicting an approximately 50% to 75% response rate to hormonal therapy, while ER-negative tumors have a less than 10% chance of response. 1,2 ER positivity also is prognostic of delayed recurrence in primary breast cancer. 3 The data concerning progesterone receptor (PR) are not as clear, but ER-negative tumors with PR are thought to have an intermediate response rate compared with ER- and PR-positive tumors. 2 In addition, estrogen is thought to play a major role in the development and progression of breast cancer. 4 Androgens also are thought to have an important role in breast cancer. The risk of breast cancer is increased in postmenopausal women with high estrogen levels as well as in women with high androgen levels. 5,6 The mechanism by which androgens contribute to breast cancer is not well understood, but studies have shown that androgens can induce proliferative changes in breast tissue, and animal models have shown that administration of both estrogen and androgens can induce tumor formation. 7,8 Also, recent studies have shown that the breast cancer susceptibility gene 1 (BRCA1) is a coactivator of the androgen receptor (AR). 9 Studies of AR expression in breast cancer have shown that AR is expressed in the majority of ER-positive tumors; however, these studies have not emphasized the expression of AR in ER-negative tumors; most had a majority of ER-positive tumors. 10-14 In addition, while some of these studies found that expression of AR predicted response to hormonal therapy and overall survival, they also showed that AR status was not predictive of disease-free survival in patients with ER-positive Am J Clin Pathol 2003;120:725-731 725 725 725
Agoff et al / ANDROGEN RECEPTOR IN BREAST CANCER tumors. In these studies, ER status, not AR status, was an independent prognostic factor for disease-free survival. 11-13 The purposes of the present study were to analyze expression of AR in paraffin-fixed tissues in a subset of patients from a university hospital with predominantly ERnegative tumors and to correlate AR expression with other prognostic variables as well as clinicopathologic data, including disease-free survival. Materials and Methods During the period July 1990 to July 2002, ER-negative (n = 69) and ER-positive (n = 19) breast cancer cases with concurrent immunohistochemical prognostic panels (ER, PR, HER-2/neu, Ki-67, and p53) were randomly selected from the files of the Department of Pathology, University of Washington Medical Center, Seattle. Approval was obtained for this study from the University of Washington Medical Center Human Subjects division (application 02-2811-E01). The following information was obtained from all patient s medical records (when available): age, menopausal status, grade and type of tumor, tumor size, lymph node status, clinical stage, family history of breast and ovarian cancer, BRCA mutation, pregnancy, and clinical follow-up data. Tumors were graded according to the modified Scarff- Bloom-Richardson system. 15 Staging followed the current American Joint Committee on Cancer staging criteria. 16 All patients with primary breast carcinoma received current standard therapy: lumpectomy and radiation or modified radical mastectomy; sentinel node biopsy (and axillary dissection if the sentinel node was positive); or axillary dissection, with radiation therapy for tumors with extensive axillary involvement (>5 cm primary tumor) or with close margins. Chemotherapy and hormone therapy treatments were based on characteristics of the tumor and patient preferences. In general, patients with ER-negative, Her2-neu positive or multiply node-positive tumors received anthracycline-based therapy, whereas patients with more favorable prognostic characteristics were treated with a modified Cooper-type regimen of cyclophosphamide (Cytoxan), methotrexate, 5-fluorouracil (CMF) weekly therapy. Only patients with hormone receptor positive tumors received hormonal therapy. Immunohistochemical Analysis ER, PR, AR, Ki-67, and p53 Antibodies, vendor sources, working dilutions, and tissue pretreatments are listed in Table 1. Antigen localization was performed with a modified avidin-biotin (ABC) immunoperoxidase method according to standard protocols (Vector Laboratories, Burlingame, CA). 3,3'-diaminobenzidine was used as the chromogen. The immunostaining results for ER, PR, and AR were assessed semiquantitatively and reported as positive if more than 5% of cells immunostained in a tumor. Ki-67 expression was graded semiquantitatively as low (<10%), intermediate (10%-20%), or high (>20%). 17 p53 overexpression was defined as more than 50% of the cells with strong nuclear staining. 18 Her-2/neu Immunohistochemical analysis for Her-2/neu was performed using a polyclonal antibody to the Her-2/neu protein (Table 1). Four-micrometer sections were deparaffinized and rehydrated in graded alcohols. For each case, 1 slide was subjected to pretreatment (Table 1) and the other was not. Slides then were incubated with the anti HER- 2/neu antibody, which is the same currently available in the HercepTest kit (DAKO, Carpinteria, CA), except it is not in the prediluted form. For positive controls, breast carcinomas known to be immunoreactive for HER-2/neu with fluorescent in situ hybridization (FISH)-proven gene amplification were used. Benign breast tissue in the background provided the internal negative control. FISH for HER-2/neu Gene Amplification For an intermediate HER-2/neu immunohistochemical result (9 cases), FISH was performed as specified by the Table 1 Antibodies and Dilutions Used Antibody to Clone Vendor * Pretreatment Working Dilution Estrogen receptor 1D5 AMAC 18 min microwave in 10-mmol/L citrate concentration, ph 6.0 1:1,000 Progesterone receptor PR88 BioGenex 18 min microwave in 10-mmol/L citrate concentration, ph 6.0 1:500 Androgen receptor F39.4.1 BioGenex 15 min microwave in 0.5-mol/L tris(hydroxymethyl)aminomethane 1:500 concentration, ph 10 Her-2/neu Polyclonal DakoCytomation A, no pretreatment; B, 15 min microwave in 10-mmol/L citrate A, 1:1,000; concentration, ph 6.0 B, 1:2,000 Ki-67 MIB-1 DakoCytomation 18 min microwave in 10-mmol/L citrate concentration, ph 6.0 1:500 p53 DO7 DakoCytomation 15 min microwave in 10-mmol/L citrate concentration, ph 6.0 1:1,000 * AMAC (Immunotech International), Westbrook, ME; BioGenex, San Ramon, CA; DakoCytomation, Carpinteria, CA. 726 Am J Clin Pathol 2003;120:725-731 Downloaded 726 from https://academic.oup.com/ajcp/article-abstract/120/5/725/1759235
Anatomic Pathology / ORIGINAL ARTICLE manufacturer using the PathVysion HER-2 DNA Probe Kit (Vysis, Downers Grove, IL). Briefly, 4- to 5-µm tissue sections were baked for 20 minutes at 60 C, deparaffinized in two 10-minute changes of xylene, transferred through two 5-minute changes of 100% ethanol, and air dried. The slides then were pretreated, digested in protease solution, fixed, denatured, and hybridized as specified by the manufacturer. FISH slides were scored according to the manufacturer s guidelines. Statistics Univariate χ 2 tests were used to assess associations between patient and tumor characteristics and AR positivity. Mantel-Haenszel tests for trend were used to measure associations in ordered categorical factors, such as tumor grade and size and clinical stage. Cox proportional hazards modeling was applied for analyses regarding time to disease recurrence, using likelihood ratio tests to assess differences by AR positivity. Kaplan-Meier actuarial methods were used to calculate disease-free survival probabilities. Results Patient and Tumor Characteristics Patient and tumor characteristics are summarized in Table 2. The patients ranged in age from 26 to 91 years (mean, 54.9 years); 31 (35%) of 88 patients were younger than 50 years of age, and 37 (42%) were premenopausal. Of the cases, 84 were biopsy specimens of primary breast carcinomas, and 4 were biopsy specimens of metastatic disease. The majority of tumors were ductal carcinomas (n = 75), with a few lobular (n = 7) and pleomorphic lobular carcinomas (n = 4). Most tumors (58/86 [67%]) were high grade, and slightly more than half of the primary tumors had lymph node metastasis at diagnosis (43/76 [57%]). In 2 metastatic tumors, the type and grade were not specified. In ER-negative tumors, AR positivity Image 1 was associated with increasing age (P =.02), postmenopausal status (P <.001), lower tumor grade (P =.03), smaller tumor (P =.03), and Her-2/neu overexpression (P =.003). Lack of AR expression showed a trend toward association with high Ki-67 expression and with pregnancy (only 1 of 6 pregnant women had tumors that were AR-positive), but these trends were not statistically significant (P =.06 and.10, respectively). In ER-positive tumors, AR positivity was associated with PR positivity (P =.03) and showed a trend toward association with lack of Her-2/neu overexpression, although this was not statistically significant (P =.06). There were no other associations with AR positivity in ER-positive tumors; however the number of ER-positive cases was small, and nearly all were positive for AR. The presence of AR strongly correlated with estrogen positivity: 17 (89%) of 19 ER-positive breast cancers were positive for AR compared with only 34 (49%) of 69 ERnegative breast tumors (P =.002). Samples from only 2 patients with proven BRCA1 mutations were included in the study, and both of those tumors were negative for ER and AR. There was no significant correlation between AR and family history of breast or ovarian cancer. Follow-up for Disease Recurrence Follow-up for disease recurrence was complete for 66 women (75%), including 57 of 69 women with ER-negative tumors (83%). Follow-up after treatment ranged from 4 to 120 months, with a mean follow-up of 25 months. Followup data are summarized in Table 3. During this time, 21 (84%) of 25 patients with ER-negative, AR-positive tumors remained disease free, while only 17 (53%) of 32 women with ER-negative, AR-negative tumors were disease free. Of 57 women with ER-negative tumors, 19 (33%) experienced disease relapse: 15 (47%) of 32 with AR-negative tumors but only 4 (16%) of 25 with AR-positive tumors (Table 3). None of the women with ER-negative, AR-positive tumors died of disease, but 4 with ER-negative, ARnegative tumors died of disease. The follow-up data for women with ER-positive tumors were too scant (9 of 19 patients) to draw meaningful conclusions. In univariate Cox regression analysis of disease-free survival in patients with ER-negative tumors, patients with tumors that were AR-positive had only 33% the risk of recurrence compared with that for patients with AR-negative tumors (P =.049; hazard ratio, 0.33; 95% confidence interval, 0.1-1.0) Table 4 and Figure 1. Other factors that showed a significant association with poor prognosis in ER-negative tumors in univariate analyses included age younger than 50 years, 4 or more positive lymph nodes, tumor larger than 5 cm, high level of Ki-67 expression, and stage III tumors. The other factors (age 60 years or older, Her-2/neu status, grade, PR status [data not shown], pregnancy, family history, p53, ductal histologic features [data not shown], and menopausal status) were not significantly associated (P.2) with a poor prognosis in ERnegative tumors. In multivariate analysis, adjusted hazard ratios for AR expression were not significant (P =.5) when adjusting for age, tumor size, positive nodes, clinical stage, and Ki-67 expression. However, those factors also lost significance in multivariate analysis (P.5), except for Ki-67 expression (hazard ratio, 8.7; P =.02). Am J Clin Pathol 2003;120:725-731 727 727 727
Agoff et al / ANDROGEN RECEPTOR IN BREAST CANCER Table 2 Patient and Tumor Characteristics Associated With ER and AR Status * ER Tumors (n = 69) ER+ tumors (n=19) Characteristic AR (n = 35) AR+ (n = 34 [49%]) P AR (n = 2) AR+ (n = 17 [89%]) P Age (y).02.4 <40 5 2 (29) 0 2 (100) 40-49 11 5 (31) 0 6 (100) 50-59 11 16 (59) 1 1 (50) 60-69 8 4 (33) 0 4 (100) 70+ 0 7 (100) 1 4 (80) Menopausal status <.001.2 Premenopausal 22 7 (24) 0 8 (100) Postmenopausal 12 27 (69) 2 9 (82) Tumor grade.03.9 I 1 1 (50) 1 8 (89) II 3 11 (79) 0 3 (100) III 31 20 (39) 1 6 (86) Tumor type.5.2 Ductal 34 30 (47) 2 9 (82) Lobular 1 2 (67) 0 4 (100) Pleomorphic lobular 0 0 0 4 (100) Tumor size (cm).03.9 2 8 16 (67) 1 9 (90) >2-5 22 12 (35) 1 6 (86) >5 4 2 (33) 0 0 Lymph nodes.9.3 Negative 12 10 (45) 2 9 (82) Positive 21 17 (45) 0 5 (100) Clinical stage.4.9 I 4 6 (60) 0 5 (100) II 15 17 (53) 2 8 (80) III 11 4 (27) 0 3 (100) IV 5 6 (55) 0 1 (100) Progesterone receptor.3.03 Negative 34 34 (50) 2 4 (67) Positive 1 0 0 13 (100) Her-2/neu.003.06 Negative 32 20 (38) 1 16 (94) Positive 3 13 (81) 1 1 (50) Ki-67 expression.06.8 Low 2 6 (75) 1 4 (80) Intermediate 3 6 (67) 0 5 (100) High 26 20 (43) 1 7 (88) p53.3.6 Negative 10 14 (58) 2 14 (88) Positive 20 16 (44) 0 2 (100) Pregnant.10 No 29 32 (52) 2 16 (89) Yes 5 1 (17) 0 0 Family history No 26 26 (50).4 2 13 (87).8 Yes 9 5 (36) 0 4 (100) AR, androgen receptor; ER, estrogen receptor. * Data for AR+ tumors are given as number (percentage of the row total for ER tumors or ER+ tumors). Data were not available for all cases for some characteristics. Modified Scarff-Bloom-Richardson grading system. 15 At least 1 first-degree relative with breast and/or ovarian cancer. Discussion The prognostic significance of ER and PR expression in breast cancer and their roles in therapy are well recognized, but the significance of AR expression is less well characterized. Previous studies have focused on ER-positive tumors and have shown some association between AR status and disease-free survival, as well as response to endocrine therapy, but these findings are obscured by the presence of ER. 12-14 Our study focused on ER-negative tumors, since most ER-positive tumors are AR-positive, and ER-negative tumors characteristically have a poorer prognosis and are thought to respond poorly to endocrine therapy. 1 We found that AR expression correlated with longer survival in a cohort of women with ER-negative tumors. In one of the largest studies of AR status in breast cancer (1,371 patients), Bryan and colleagues 11 found a highly significant association between AR status and survival (P <.001) that remained significant even when taking ER status into account. In addition, they found that AR positivity influenced the response of the primary tumor and metastasis to tamoxifen therapy. Unfortunately, they did not assess the response to hormonal therapy in any ER-negative, AR-positive tumors. In a more recent study of 153 728 Am J Clin Pathol 2003;120:725-731 Downloaded 728 from https://academic.oup.com/ajcp/article-abstract/120/5/725/1759235
Anatomic Pathology / ORIGINAL ARTICLE A B C Image 1 Androgen receptor (AR) expression in an estrogen receptor (ER)-negative tumor (A, H&E; B, ER; C, AR; 40). patients, Kuenen-Boumeester and colleagues 12 showed that while AR status was a significant prognostic factor for disease-free survival in univariate analysis, in multivariate analysis, only lymph node status, tumor size, and ER status were independent prognostic variables. The findings in our study are in accord with previous studies in that we found that AR expression was associated significantly with disease-free survival in univariate analysis. However, unlike previous analyses, our study focused on ERnegative tumors and found an association between AR expression and survival in this subset of patients, who characteristically have a worse prognosis than patients with ERpositive tumors. Whether this difference in disease-free survival is indicative of a more indolent nature of AR-positive tumors or whether it reflects sensitivity to treatment is unclear, but none of these patients received antiestrogen or antiandrogen therapy. In multivariate analysis, the association between AR status and survival did not maintain significance; however, neither did other factors such as tumor size, lymph node status, and clinical stage. This implies that owing to the small data set, the statistical power was lacking to detect significant differences after adjusting for other important factors. Other authors have found significant associations between AR status and prognostic factors as well. Isola 14 found a significant correlation between AR status and tumor grade, similar to our findings, but did not find a positive correlation between AR status and HER-2/neu status, which was present in our study. Also, unlike the findings of Kuenen-Boumeester et al, 12 we did not find a significant correlation between AR status and Ki-67 expression, although there was a trend toward an association. Bryan and colleagues 11 did not find a significant correlation between Am J Clin Pathol 2003;120:725-731 729 729 729
Agoff et al / ANDROGEN RECEPTOR IN BREAST CANCER Table 3 Follow-up Information on Patients According to Receptor Status * Receptor Status No Evidence of Disease Alive With Disease Died of Disease ER AR+ (n = 25) 21 (84) 4 (16) 0 (0) ER AR (n = 32) 17 (53) 11 (34) 4 (13) ER+ AR+ (n = 8) 6 (75) 1 (13) 1 (13) ER+ AR (n = 1) 0 (0) 1 (100) 0 (0) Total (n = 66) 44 (67) 17 (26) 5 (8) AR, androgen receptor; ER, estrogen receptor. * Data are given as number (percentage). AR and menopausal status in their patients; however, we found that AR expression not only correlated with increasing age, but also was highly significantly correlated with menopausal status. Some of these differences may be Table 4 Univariate Analysis of Factors Associated With Poor Prognosis in Patients With ER-Negative Tumors Hazard Ratio Characteristic (95% Confidence Interval) * P Age (y) <50 3.5 (1.1-11.5).04 50-59 1.0 (ref) 60+ 3.2 (0.8-12.0).09 Androgen receptor Positive 0.33 (0.1-1.0).05 Tumor size (cm) 2 1.0 (ref) >2-5 3.7 (0.8-16.7).08 >5 8.4 (1.4-50.4).02 Lymph nodes 1-3 1.4 (0.4-5.1).7 4+ 4.2 (1.4-12.7).01 Clinical stage I or II 1.0 (ref) III 3.6 (1.3-10.1).02 IV 1.9 (0.5-7.3).3 Ki-67 expression Low or intermediate 1.0 (ref) High 5.4 (1.2-24.8).03 Tumor grade 15 I or II 1.0 (ref) III 2.3 (0.7-8.0).2 Her-2/neu Positive 0.55 (0.2-1.6).3 p53 Positive 1.8 (0.5-6.9).4 Pregnant No 1.0 (ref) Yes 1.9 (0.4-8.5).4 Menopausal status Premenopausal 1.0 (ref) Postmenopausal 0.64 (0.2-1.7).8 Family history No 1.0 (ref) Yes 2.0 (0.7-5.6).2 ref, reference value. * From Cox proportional hazards regression analysis. explained by the fact that our study emphasized ER-negative tumors, which tend to be high-grade ductal carcinomas with more advanced stage at diagnosis. Although the functional role of androgens in breast cancer is uncertain, epidemiologic associations and animal models implicate androgens in breast carcinogenesis. The incidence of breast cancer is high in postmenopausal women when androgenic levels are high, and the risk of breast cancer is increased in women with high estrogen levels and in those with high androgen levels. 5,6,19 Wong and Xie 7 and Xie and colleagues 8 have shown in animal models that testosterone in combination with estrogen can induce a high incidence of mammary carcinoma and that administration of flutamide can block changes induced by androgens. Similarly, Boccuzzi et al 20 showed that flutamide inhibited growth of rat mammary tumors. A major strength and distinguishing feature of our study is that it concentrates on ER-negative breast cancers, a group of tumors that have a poorer prognosis than ER-positive tumors; our data show that AR status can separate ER-negative tumors into more and less favorable prognostic groups. Because we chose to focus our study on this group of tumors, it included a preponderance of ER-negative, high-grade ductal carcinomas, as these features tend to be associated. Survival Distribution Function 1.00 0.75 0.50 0.25 0.00 0 20 40 60 80 100 120 Disease-Free Survival (months) Figure 1 Disease-free survival curve for patients with estrogen receptor negative tumors stratified according to androgen receptor (AR) positivity (P =.049). Solid line, ARnegative; hatched line, AR-positive. 730 Am J Clin Pathol 2003;120:725-731 Downloaded 730 from https://academic.oup.com/ajcp/article-abstract/120/5/725/1759235
Anatomic Pathology / ORIGINAL ARTICLE Thus, we have less information regarding low-grade ERnegative tumors and about ER-negative, PR-positive tumors. This study was retrospective, so treatment was not influenced by AR expression. Because all patients received current standard therapies, treatment differed depending on the tumor, and our survival curves may reflect treatment sensitivity and the innate characteristics of the tumor. AR is expressed in almost 50% of ER-negative tumors, has significant associations with clinical, pathologic, and immunohistochemical prognostic factors, and is a predictor of disease-free survival in univariate analysis. Although statistical significance was lacking in our multivariate analysis, a larger sample and additional research may help clarify whether AR status is a significant independent prognostic factor. While the functional role of AR in breast cancer remains unclear, further exploration of this area could expand the repertoire of potential treatments for patients with ER-negative breast cancers. From the Departments of 1 Pathology, 2 Oncology, and 3 Epidemiology, University of Washington, Seattle. Supported in part by a generous gift from the Avon Foundation, New York, NY (Dr Linden). Address reprint requests to Dr Agoff: Assistant Professor, Pathology and Cytopathology, UWMC/Harborview Medical Center, Box 356100, 1959 NE Pacific St, Seattle, WA 98195. References 1. Wittliff JL. Steroid-hormone receptors in breast cancer. Cancer. 1984;53:630-643. 2. Osborne CK, Yochmowitz MG, Knight WA, et al. The value of estrogen and progesterone receptors in the treatment of breast cancer. Cancer. 1980;46:2884-2888. 3. Knight WA, Livingston RB, Gregory EJ, et al. Estrogen receptor as an independent prognostic factor for early recurrence in breast cancer. Cancer Res. 1977;37:4669-4671. 4. Vessey MP. Effect of endogenous and exogenous hormones on breast cancer: epidemiology. Verh Dtsch Ges Pathol. 1997;81:493-501. 5. Cauley JA, Lucas FL, Kuller LH, et al. Elevated serum estradiol and testosterone concentrations are associated with a high risk for breast cancer: study of Osteoporotic Fractures Research Group. Ann Intern Med. 1999;130:270-277. 6. Hankinson SE, Willett WC, Manson JE, et al. Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst. 1998;90:1292-1299. 7. Wong YC, Xie B. The role of androgens in mammary carcinogenesis. Ital J Anat Embryol. 2001;106:111-125. 8. Xie B, Tsao SW, Wong YC. Sex hormone induced mammary carcinogenesis in female noble rats: the role of androgens. Carcinogenesis. 1999;20:1597-1606. 9. Park JJ, Irvine RA, Buchanan G, et al. Breast cancer susceptibility gene 1 (BRCAI) is a coactivator of the androgen receptor. Cancer Res. 2000;60:5946-5949. 10. Brys M, Wojcik M, Romanowicz-Makowska H, et al. Androgen receptor status in female breast cancer: RT-PCR and Western blot studies. J Cancer Res Clin Oncol. 2002;128:85-90. 11. Bryan RM, Mercer RJ, Bennett RC, et al. Androgen receptors in breast cancer. Cancer. 1984;54:2436-2440. 12. Kuenen-Boumeester V, Van der Kwast TH, Claassen CC, et al. The clinical significance of androgen receptors in breast cancer and their relation to histological and cell biological parameters. Eur J Cancer. 1996;32A:1560-1565. 13. Soreide JA, Lea OA, Varhaug JE, et al. Androgen receptors in operable breast cancer: relation to other steroid hormone receptors, correlations to prognostic factors and predictive value for effect of adjuvant tamoxifen treatment. Eur J Surg Oncol. 1992;18:112-118. 14. Isola JJ. Immunohistochemical demonstration of androgen receptor in breast cancer and its relationship to other prognostic factors. J Pathol. 1993;170:31-35. 15. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer, I: the value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology. 1991;19:403-410. 16. Greene FL, Page DL, Fritz A, et al, eds. The AJCC Cancer Staging Manual. 6th ed. New York, NY: Springer-Verlag; 2002:223. 17. Brown RW, Allred CD, Clark GM, et al. Prognostic value of Ki-67 compared to S-phase fraction in axillary node negative breast cancer. Clin Cancer Res. 1996;2:585-592. 18. Barnes DM, Dublin EA, Fisher CJ, et al. Immunohistochemical detection of p53 protein in mammary carcinoma: an important new independent indicator of prognosis? Hum Pathol. 1993;24:469-476. 19. Secreto G, Toniolo P, Berrino F, et al. Increased androgenic activity and breast cancer risk in premenopausal women. Cancer Res. 1984;44:5902-5905. 20. Boccuzzi G, Tamagno E, Brignardello E, et al. Growth inhibition of DMBA-induced rat mammary carcinomas by the antiandrogen flutamide. J Cancer Res Clin Oncol. 1995;121:150-154. Am J Clin Pathol 2003;120:725-731 731 731 731