OUP user menu

Decreased Androgen Receptor Expression Is Associated With Distant Metastases in Patients With Androgen Receptor–Expressing Triple-Negative Breast Carcinoma

Lisa M. Sutton MD, Dengfeng Cao MD, PhD, Venetia Sarode MD, Kyle H. Molberg MD, Kwame Torgbe MD, Barbara Haley MD, Yan Peng MD, PhD
DOI: http://dx.doi.org/10.1309/AJCP8AVF8FDPTZLH 511-516 First published online: 1 October 2012


To characterize prognostic values of androgen receptor (AR) in triple-negative (TN) breast cancers, we investigated AR expression status and levels, explored an association of AR expression with metastatic disease, and correlated AR expression with Ki-67 in TN invasive breast carcinomas. AR expression was analyzed with immunohistochemistry in 121 cases of TN tumors. Thirty-nine cases had distant metastatic disease and 82 had locoregional disease only. AR was positive in 38 (31.4%) of the 121 cases. Our results indicate that among the AR-positive TN tumors, distant metastases are significantly associated with lower expression of AR compared with cases with only locoregional disease, and that AR expression negatively correlates with Ki-67 expression. These findings suggest that decreased intratumoral AR expression may be predictive of distant metastatic disease and AR expression levels may have potential prognostic value in AR-expressing TN tumors.

Key Words
  • Androgen receptor
  • Triple-negative breast cancer
  • Distant metastasis

High androgen receptor (AR) expression in breast cancer has been correlated with a low risk of recurrence and death. Significant differences in AR expression have been identified in different molecular subtypes of breast cancer. Triple-negative (TN) breast carcinoma, characterized by estrogen receptor (ER), progesterone receptor (PR), and HER2 negativity, comprises a group of aggressive tumors that characteristically have a low AR expression.1 AR has been shown to have prognostic implications in breast carcinoma, and higher AR expression levels have been associated with older age at diagnosis, higher expression of ER or PR, lower nuclear grades, and smaller tumor size.2,3 Multiple studies have revealed that AR positivity has been detected in approximately half of all breast carcinoma cases.1,2 However, it has been detected much less frequently in TN breast carcinomas (25%-35%).1,2,4-6 Positive AR expression has been associated with favorable outcomes in ER-positive tumors.2 Previous studies looking at AR expression in TN breast carcinoma have demonstrated that AR negativity has been associated with a shorter disease-free interval and overall survival than AR-positive TN cancer.4,7,8 These studies suggest that AR expression could be a useful prognostic marker in TN tumors.

TN carcinomas lack molecular targets commonly used in targeted therapy, making this group of tumors difficult to treat, and chemotherapy remains the main treatment modality.9-11 Conventional neoadjuvant chemotherapy results in complete pathologic response in 13% to 45% of TN tumors, and patients not exhibiting a complete response have a high likelihood of disease relapse.12 Despite having relatively high rates of clinical response to presurgical/neoadjuvant chemotherapy, patients with TN tumors have frequent distant recurrences and a poorer prognosis than patients with non-TN subtypes of breast cancer.13 Fewer than 30% of patients with metastatic TN tumors survive 5 years, and almost all die of their disease despite neoadjuvant chemotherapy.14 The development of novel therapies is crucial to improve the prognosis of TN tumors.

TN breast carcinoma was traditionally considered a distinct, molecular subset of breast carcinoma, but currently it has been proposed that TN breast cancer includes a heterogeneous group of tumors.15 Subtyping TN cancer is necessary to better identify molecular targeted therapies. Lehmann et al15 recently identified 6 TN cancer subtypes using gene expression analysis, including 2 basal-like (BL1 and BL2), an immunomodulatory, a mesenchymal-like, a mesenchymal stem cell–like, and a luminal androgen receptor (LAR) subtype. The LAR subtype was characterized with AR signaling, and expressed numerous downstream AR targets and coactivators. Further characterization of AR expression in TN tumors using immunohistochemistry (IHC) may help better elucidate the molecular basis of TN tumors and provide additional information on prognosis and treatment options for patients with TN tumors that lack specific therapeutic targets.

Recent studies suggest a possible antiproliferative effect of AR stimulation and pathway activation in breast cancer.16,17 High expression of Ki-67, a proliferation index, has been associated with a worse prognosis in TN breast cancer. However, little is known about the correlation status between AR and Ki-67 expression in TN cancers. In this study, we investigated AR expression status and levels in TN breast carcinomas, explored an association of AR expression with metastatic disease, and correlated AR expression with Ki-67.

Materials and Methods

Patient Identification

Patient information was obtained with a search of case files from 2007 to 2010 at the University of Texas Southwestern Medical Center and Parkland Memorial Hospital, Dallas. We included all patients in whom TN breast carcinoma was established with IHC and confirmed with HER2 fluorescence in situ hybridization assay. The criteria for determining triple negativity were based on immunohistochemical staining and image quantitation of ER, PR, and HER2 as previously described.18,19

We identified 163 patients with primary TN invasive breast carcinoma. IHC was performed in 120 of the 163 cases, 1 of which had bilateral primary tumors that were analyzed separately, for a total of 121 cases included in this study. Follow-up data were available for a mean of 30.3 months (range, 1-60 months). The 121 cases of TN breast carcinomas were divided into either distant metastatic disease group (pathologic stage pM1) or locoregional disease group (pM0). Mean follow-up period for the pM1 group was 30.7 months (range, 1-60 months) and for the pM0 group was 30 months (range, 3-60 months).

The locoregional disease group included cases with or without axillary node involvement and with no evidence of distant metastases. Of the 121 cases, 112 (92.56%) were determined to be basal-like TN tumors defined as cytokeratin 5/6 (CK5/6) positivity and/or epidermal growth factor receptor (EGFR) positivity on IHC.

IHC for Tumor Biomarkers

Slides cut from the tumor tissue were immunostained for ER (SP1, prediluted, Ventana Medical Systems, Tucson, AZ), PR (PgR, prediluted, Ventana Medical Systems), HER2 (prediluted, Ventana Medical Systems), CK5/6 (D5/16B4, 1:50, Cell Marque, Rocklin, CA), and EGFR (3C6, 1:2, Ventana Medical Systems). These IHC tests were performed as previously described.18,19

IHC was also used for several additional markers including AR (sc-816, 1:15, Santa Cruz Biotechnology, Santa Cruz, CA), Ki-67 (MIB-1, prediluted, Ventana Medical Systems), and p53 (DO-7, 1:16,000; DAKO, Carpinteria, CA). Immunostaining for AR, Ki-67, and p53 was performed using IHC in conjunction with automated image analysis (ACIS, Clarient, San Juan Capistrano, CA), and all of the results were confirmed on manual review by a pathologist (Y.P.). The ACIS system used for image quantitation consisted of an automated robotic bright-field microscope module, a computer, and a Windows NT (Microsoft, Redmond, WA)–based software interface. The robotic microscope module scanned the IHC-stained slides, and the computer monitor displayed the digitalized tissue images. After viewing the high-magnification images on the ACIS computer, more than 10 subregions of the digitalized tissue images were selected for analysis by the ACIS. To assess the level of AR, Ki-67, or p53 expression, the ACIS provided the percentage of positively stained cells for these markers in the selected subregions. When the percentage of cells staining positive for AR was more than or equal to 1%, the tumor was considered to be positive for AR. Ki-67 and p53 expression was considered to be high when positive cells were more than or equal to 10%.

Statistical Analysis

The frequency of AR expression in the pM1 and pM0 groups was examined using the Fisher exact test. The percentage of intratumoral expression of AR in the pM1 and pM0 groups was compared (data shown as mean ± standard error of mean [SEM]) using the unpaired t test to explore an association of AR expression with metastatic disease in TN tumors. Also, the Pearson correlation analysis was used to correlate AR expression levels with Ki-67 expression, p53 expression, and tumor size. All analyses were performed using GraphPad Prism 5 (GraphPad Software, La Jolla, CA).


Of the 121 cases of TN tumors included in this study, 39 were determined to have distant metastatic disease (pM1) and 82 had evidence of locoregional disease only (pM0). Of the latter 82 cases of local disease, 34 had axillary node metastases. In the pM1 group, 41% of the cases were treated with neoadjuvant chemotherapy before surgery and the most common therapeutic agents used were doxorubicin, cyclophosphamide, and docetaxel; 35% were treated with a partial mastectomy and 65% with a total mastectomy. In the pM0 group, 21% of the cases were treated with presurgical chemotherapy; 62% were treated with a partial mastectomy, and 38% with a total mastectomy. The pM1 cases were more likely than the pM0 cases to receive neoadjuvant chemotherapy.

AR was positive in 38 (31.4%) of the 121 cases. The frequency of AR positivity in the pM1 and pM0 groups was similar, with 10 positive pM1 cases (25.64%) and 28 positive pM0 cases (34.15%, P = .4). There was no significant difference in the rate of neoadjuvant chemotherapy use between the AR-positive and AR-negative cases.

The clinicopathologic characteristics of the pM1 and pM0 groups are summarized in Table 1, and the AR-positive and AR-negative cases are summarized in Table 2. No significant differences were found in the patient age, histologic subtypes, histologic grade, or size of tumor between the pM1 and pM0 groups or the AR-positive and AR-negative groups.

View this table:
Table 1
View this table:
Table 2

Among the AR-positive cases, the intratumoral expression level was significantly higher in the pM0 group (52.2% ± 7.2%) compared with the pM1 group (19.4% ± 9.1%; P = .02) Figure 1. Of note, 8 of the 10 AR-positive cases with distant metastases (pM1 group) had 20% or less AR positivity Figure 2. Image 1 shows tumor AR immunoreactivity in representative cases in the pM1 and pM0 groups.

Figure 1

Percentage of intratumoral androgen receptor (AR) expression in the distant metastasis (pM1) and locoregional (pM0) groups with AR-positive tumors (data shown as mean ± standard error of mean).

Figure 2

Individual intratumoral androgen receptor (AR) expressions in the distant metastasis (pM1) group with a total of 10 AR-positive cases.

Image 1

Triple-negative carcinoma with androgen receptor (AR) immunoreactivity in the distant metastasis (pM1) group (A, H&E, ×10; B, AR immunostain, ×10) and locoregional (pM0) group (C, H&E, ×20; D, AR immunostain, ×20).

Differences in Ki-67 expression levels between AR-positive and AR-negative cases are shown in Table 2. The mean Ki-67 expression among the AR-positive cases (48.4% ± 5.0%) was significantly lower than that of the AR-negative cases (73.1% ± 3.5%; P < .0001) Figure 3. The AR expression levels showed a significant negative correlation with Ki-67 expression in the AR-positive tumors (r = −0.55; P = .0006). No correlation was found between AR expression levels and either p53 expression or tumor size.

Figure 3

Percentage of Ki-67 expression in the androgen receptor (AR)–positive tumors and AR-negative tumors (data shown as mean ± standard error of mean).


Our results reveal that 31.4% of TN breast cancers express AR, which is similar to results seen in previous reports.1,2,5,6,9,20 Among AR-positive TN breast tumors, distant metastases (pM1) are significantly associated with a lower intratumoral expression of AR compared with cases with only locoregional disease (pM0). The results suggest that, among AR-positive TN cancers, the tumors with a higher expression of AR may be less likely to develop metastatic disease. There is no significant difference in the frequency of AR positivity between the distant metastases and the locoregional disease groups in the study.

Androgens have long been thought to play a role in breast cancer, and previous studies have shown that androgens can induce proliferative changes and tumor formation.21 Endogenous androgens inhibit breast development, the effect of which is normally balanced by the stimulatory effect of estrogens.22 Androgens, including testosterone and dihydrotestosterone, function by binding to and activating intracellular ARs. AR expression is abundant in normal breast epithelial cells. However, the functional role of AR expression in breast cancer is not well understood.23

Ki-67 is a proliferative marker with expression peaking during mitosis. The use of Ki-67 as a predictive and prognostic marker in breast cancer has been investigated extensively with higher levels associated with worse prognosis.19,24 Our results demonstrated a negative correlation of AR expression with Ki-67, and a significantly lower Ki-67 expression in the AR-positive tumors than that in the AR-negative tumors, which may be related to the antiproliferative effect of AR stimulation; these findings further support the idea that high AR levels may be associated with a better prognosis in TN tumors.16,17,20

Our results suggest that intratumoral AR expression level rather than the frequency of AR positivity in TN tumors may have potential prognostic value. Decreased intratumoral AR expression is associated with a worse prognosis in AR-expressing TN tumors and may be predictive of distant metastatic disease. In contrast, TN tumors with higher AR expression may have a better prognosis in part due to decreased tumor cell proliferation caused by the increased antiproliferative effect of AR stimulation. Although IHC assessment of AR appears to have significant prognostic value, routine testing for AR in breast cancer specimens clinically is not recommended at this point. Further work is needed to better understand the role of AR expression in tumor metastases.

Our results support the concept that TN breast cancers are a heterogeneous group of tumors with only the minority of cases expressing AR.15 The Lehmann et al15 study revealed that the percentage of TN tumor cells scored with nuclear AR immunostaining and the intensity of the staining were significantly higher in the luminal AR subtype (>10-fold; P < .004) compared with all other TN cancer subtypes. The AR-expressing TN tumors in our study may represent the luminal AR subtype of TN tumors. In addition, their study demonstrated that luminal AR cell lines are uniquely sensitive to bicalutamide, an AR antagonist.15 The AR signaling pathway may be a molecular driver that can be therapeutically targeted in AR-expressing TN tumors, suggesting that AR inhibitor may be a novel therapeutic agent for treating TN breast cancers, especially those tumors not responding to chemotherapy. Knowing the AR expression status of TN tumors is important for the selection of patients with TN breast cancer for antiandrogen therapy.


View Abstract