Androgen receptors frequently are expressed in breast carcinomas

Potential relevance to new therapeutic strategies

Abstract

BACKGROUND

Several studies have demonstrated the biologic and therapeutic significance of estrogen and progesterone receptors (ER and PR) in breast carcinomas. The aim of the current study was to examine the presence of androgen receptors (AR) in breast carcinomas.

METHODS

Two hundred cases of breast carcinoma, consisting of 145 invasive and 55 noninvasive (ductal carcinoma in situ [DCIS]) lesions, were examined using a monoclonal antibody against AR on formalin-fixed, paraffin-embedded archival material. The results were analyzed for correlations with immunohistochemically determined ER, PR, and HER-2/neu expression.

RESULTS

Eighty-seven of the 145 cases (60%) of invasive carcinoma and 45 of the 55 cases (82%) of DCIS were AR-positive according to internationally standardized guidelines. The vast majority of Grade 1 carcinomas were positive for AR (90% of invasive Grade 1 carcinomas and 95% of Grade 1 DCIS), whereas in Grade 3 invasive carcinomas and DCIS, positive immunoreactions for AR were observed in 46% and 76% of cases, respectively. Among the cases of Grade 3 carcinoma, 33 invasive carcinomas (39%) and 17 DCIS lesions (68%) were ER-negative but AR-positive. Among Grade 1 carcinomas (invasive and DCIS), not a single case was positive for HER-2/neu, but most cases were intensely positive for AR. In contrast, many invasive Grade 3 carcinomas exhibited agreement between AR status and HER-2/neu status (AR-positive and HER-2/neu-positive, 30.5%; AR-negative and HER-2/neu-negative, 42.5%).

CONCLUSIONS

Androgen receptors are commonly expressed in DCIS and in invasive breast carcinoma. A significant number of poorly differentiated carcinomas are ER-negative and PR-negative but AR-positive. Immunohistochemical examination of AR would be desirable because it would provide additional information about steroid receptors in breast carcinomas. Cancer 2003;98:703–11. © 2003 American Cancer Society.

DOI 10.1002/cncr.11532

The role of determining estrogen and progesterone receptor status in the management of breast carcinoma, particularly as a guide to identifying patients who are likely to respond to hormonal manipulations, is well established.1, 2 Estrogen and progesterone receptors (ER and PR) also have gained widespread acceptance as independent prognostic parameters in breast carcinoma.3–5 Several studies have shown that primary invasive breast carcinomas contain ER and PR in approximately 55–65% and 45–55% of cases, respectively.6, 7 The presence of both ER and PR in a breast tumor increases its likelihood of responding to hormonal manipulations from 55%, as observed in patients with ER-positive tumors, to 75–80%.7, 8 Studies have shown that PR status is at least as valuable in predicting the behavior of breast carcinoma as is ER status, and the loss of PR by tumor cells is associated with a less favorable prognosis.9

HER-2/neu, also known as c-erb-B2, is another prognostic and predictive parameter in breast carcinoma.10–12 It is a member of the erb-B oncogene family that is closely related to epidermal growth factor.13 HER-2/neu protein overexpression in breast carcinoma cells has been detected in 20–30% of invasive and 43–48% of noninvasive breast carcinomas.14–16 Recognition of HER-2/neu oncoprotein overexpression plays an increasingly important role in the selection of patients with breast carcinoma who may benefit from adjuvant therapy (i.e., treatment with a monoclonal antibody against HER-2/neu).17

Although numerous studies have examined ER and PR and their correlations with other prognostic indicators, surprisingly little is known about the role of androgen receptor (AR) and its prognostic value in breast carcinoma.18–20 Previous studies have demonstrated AR expression in other malignancies, including endometrial carcinoma.21 Although determining ER and PR status on biopsy specimens before performing hormonal manipulations has become standard practice in the management of breast carcinoma, assessment of AR currently is not practiced.

The aim of the current study was to investigate the expression of AR in a large series of breast carcinomas using immunohistochemical techniques. The results were analyzed for correlations with ER, PR, and HER-2/neu expression, as determined immunohistochemically in tissue sections from paraffin-embedded archival material.

MATERIALS AND METHODS

Two hundred cases of breast carcinoma, consisting of 145 invasive carcinomas and 55 ductal carcinomas in situ (DCIS), were retrieved from the files of the Department of Pathology at the University of Graz (Graz, Austria). Determination of tumor type and histopathologic grade was performed according to standardized guidelines.22, 23 A recently proposed translational classification system for ductal intraepithelial neoplasia also was retrospectively applied.24

The invasive breast carcinoma series consisted of 131 ductal carcinomas (IDCs) and 14 lobular carcinomas (ILCs). Of the 131 IDCs, 17 were well differentiated (Grade 1 [G1]), 31 were moderately differentiated (Grade 2 [G2]), and 83 were poorly differentiated (Grade 3 [G3]). Among the ILCs, there were three G1, nine G2, and two G3 tumors. The series of noninvasive tumors consisted of 25 cases of high-grade DCIS (DIN 3), 11 cases of intermediate-grade DCIS (DIN 2), and 19 cases of low-grade DCIS (DIN 1c).

Formalin-fixed, paraffin-embedded tissue blocks were cut into 4 μm thick serial sections that were mounted on precoated slides. The sections were deparaffinized, rehydrated, and rinsed in distilled water. Immunohistochemical assays for AR, ER, PR, and HER-2/neu were performed on consecutive paraffin sections using standardized automated procedures (Ventana Medical Systems, Tucson, AZ; and Dako, Glostrup, Denmark) (Table 1). Monoclonal mouse antihuman antibody clones 6F11 and 1A6 (Ventana Medical Systems) were used as primary antibodies for ER and PR, respectively. For determination of AR expression, the monoclonal mouse antihuman androgen receptor antibody (clone AR441; Dako) was used. In brief, antigen retrieval was achieved with microwave treatment (ER and PR) or heating in a water bath (AR) (Table 1). An ES (Ventana Medical Systems) or ChemMate (Dako) autostainer was used in conjunction with an indirect streptavidin-biotin method. After incubation with the primary antibody, incubation with the secondary (link), biotinylated antibody was performed for 30 minutes. After washing, sections were incubated with streptavidin-peroxidase for 30 minutes. Finally, the enzyme was visualized after a 15-minute incubation with diaminobenzidine. Counterstaining was performed with hematoxylin.

Table 1. Summary of Antibodies and Methods Used in the Current Study

Antigen	Antibody (manufacturer)	Method (autostainer)	Dilution	Incubation time (min)	Antigen retrieval method
DAB: 3,3′-diaminobenzidine; W: watts.
Estrogen receptor	6F11 (Ventana Medical Systems, Tucson, AZ)	Biotin-DAB (Ventana ES)	Ready to use (no-dilution)	32	Microwave at 160 W in sodium citrate, pH 6.0, for 30 min
Progesterone receptor	1A6 (Ventana Medical Systems)	Biotin-DAB (Ventana ES)	Ready to use (no dilution)	32	Microwave at 160 W in sodium citrate, pH 6.0, for 30 min
Androgen receptor	AR441 (Dako, Glostrup, Denmark)	ChemMate-DAB (Dako ChemMate)	1:100	20	Water bath (100°C) for 40 min, followed by cooling in Epitope Retrieval Solution (Dako)
HER-2/neu	K5205 (Dako) Hercep Test	Dako EnVision Polymer–DAB (Dako ChemMate)	Ready to use (no dilution)	30	Water bath (100°C) for 40 min, followed by cooling in Epitope Retrieval Solution (Dako)

Samples were scored as positive when at least 10% of nuclei were immunoreactive. HER-2/neu expression was determined using the HercepTest (Dako) in accordance with the manufacturer's protocol and scoring guidelines (possible score range, 0–3+). Only a score of 3+ was regarded as a positive result for HER-2/neu overexpression. Positive controls included normal breast tissue surrounding the tumors; negative controls included substitution of the primary antibody with normal sera or phosphate-buffered saline, omission of the secondary antibody, and incubation of the primary antibody solution with lymphoid tissue. All immunoslides were evaluated independently by at least two investigators (F. M. and W. W.). The rare cases in which disagreement occurred were reevaluated using a multiheaded microscope; a final agreement was reached in all cases that were examined.

Initial, exploratory statistical analysis (categorical principal component analysis [CPCA] and hierarchical cluster analysis [HCA]) were performed using SPSS software (Version 10.0; SPSS, Chicago, IL).25, 26 CPCA allowed us to determine the directions and magnitudes of correlations between numerous pairs of differently scaled variables in a simple (i.e., two-dimensional) space. HCA was used to compute pairwise ‘distances’ between variables; these distances are inversely related to the correlations between pairs of variables. Next, a stepwise agglomeration of variables into clusters was performed. In the first step of this agglomeration process, variables that are closest to each other (i.e., those that are most closely correlated) were joined. Subsequently, variables at greater distances from each other were joined, until all variables had been merged. The steps of the analysis are represented as a horizontal dendrogram, which should be read from left to right. Correlations that were detected with CPCA and/or HCA were tested for significance using the chi-square or Fisher exact test as appropriate.

RESULTS

Distribution of Steroid Receptors in Normal Breast Tissue

All 200 cases contained normal breast tissue (ducts and lobules) adjacent to or set apart from the tumors. In all cases, ERs were identified in normal epithelial cells (NE). (The proportion of ER-positive NE ranged from 10–80%, with an average of 45%.) Myoepithelial cells (ME), however, were completely negative for ER in the vast majority of cases (180 of 200 [90%]). In rare cases, a small proportion of ME nuclei (1–2%) were positive for ER. Positive reactions for PR in NE were observed in all cases (average proportion of stained nuclei, 20%; range, 10–50%). In contrast, no positive reaction could be identified in ME in most cases (176 of 200 [88%]), although some cases (24 of 200 [12%]) exhibited a level (1–2%) of PR-positivity. Positive reactions for AR in NE were observed in all cases (average proportion of stained nuclei, 40%; range, 10–70%). ME were completely negative for AR in most cases (180 of 200 [90%]). In some cases (10%), a small proportion of ME nuclei (1%) exhibited AR-positivity. All stromal cells examined were completely negative for ER, PR, and AR.

Distribution of Steroid Receptors and HER-2/neu in DCIS

AR was expressed in nearly all G1 tumors (95%) but was expressed somewhat less frequently in G2 and G3 DCIS (73% and 76% of cases, respectively). ER and PR were expressed frequently in G1 DCIS (100% and 95% of cases, respectively) and in G2 DCIS (91% and 54% of cases), but most G3 tumors did not express ER and PR (expression in 8% and 0% of cases, respectively). In contrast with ER and PR, HER-2/neu overexpression was rare in G1 and G2 tumors (0% and 11% of cases, respectively) and frequent in G3 tumors (84% of cases) (Table 2).

Table 2. Positive Immunoreactivity for Steroid Receptors and HER-2/neu in Invasive Ductal Carcinoma and Ductal Carcinoma In Situ

	G1 DCIS	G2 DCIS	G3 DCIS	G1 IDC	G2 IDC	G3 IDC
G1: Grade 1 (well-differentiated); DCIS: ductal carcinoma in situ; G2: Grade 2 (moderately differentiated); G3: Grade 3 (poorly differentiated); IDC: invasive ductal carcinoma; ER-estrogen receptor; PR: progesterone receptor; AR: androgen receptor.
ER-positive (%)	19/19 (100)	10/11 (91)	2/25 (8)	17/17 (100)	19/31 (61)	8/83 (9.5)
PR-positive (%)	18/19 (95)	6/11 (54.5)	0/25 (0)	14/17 (82.5)	8/31 (26)	3/83 (3.5)
AR-positive (%)	18/19 (95)	8/11 (73)	19/25 (76)	15/17 (88)	22/31 (71)	39/83 (47)
HER-2/neu-positive (%)	0/19 (0)	1/11 (9)	21/25 (84)	0/17 (0)	8/31 (26)	35/83 (42)

Comparison of ER and AR expression (Table 3) revealed that the vast majority of G1 and G2 DCIS lesions (95% and 64%, respectively) showed concordant expression of both ER and AR, whereas the majority of G3 DCIS lesions (68%) were ER-negative but AR-positive. Tumors that were both ER-negative and AR-negative were rare and were observed in only a small number of G3 DCIS lesions and not in any G1 or G2 DCIS lesions (Fig. 1).

Table 3. Immunoreaction Results for Androgen Receptor and Estrogen Receptor in Invasive Carcinoma and Ductal Carcinoma In Situ

	G1 DCIS	G2 DCIS	G3 DCIS	G1 ICA	G2 ICA	G3 ICA
G1: Grade 1 (well-differentiated); DCIS: ductal carcinoma in situ; G2: Grade 2 (moderately differentiated); G3: Grade 3 (poorly differentiated); ICA: invasive carcinoma; AR: androgen receptor; ER: estrogen receptor.
AR-positive/ER-positive (%)	18/19 (95)	7/11 (63.5)	2/25 (8)	18/20 (90)	22/40 (55)	6/85 (7)
AR-positive/ER-negative (%)	0/19 (0)	1/11 (9)	17/25 (68)	0/20 (0)	8/40 (20)	33/85 (39)
AR-negative/ER-positive (%)	1/19 (5)	3/11 (27.5)	0/25 (0)	2/20 (10)	6/40 (15)	2/85 (2)
AR-negative/ER-negative (%)	0/19 (0)	0/11 (0)	6/25 (24)	0/20 (0)	4/40 (10)	44/85 (52)

Figure 1. Examples of immunoreactions in ductal carcinoma in situ. (A-D) Ductal carcinoma in situ, high-grade (ductal intraepithelial neoplasia Grade 3 [DIN 3]), with comedo-type central necrosis. (B, C) Positive immunoreactivity for androgen receptor (AR) in different areas of the tumor. The lesion, however, was negative overall for estrogen receptor (ER) and progesterone receptor. (D) Lack of immunoreaction for ER. (E-G) Ductal carcinoma in situ, intermediate grade (DIN 2). (E) Hematoxylin and eosin stain, (F) negative immunoreaction for ER, and (G) intense and almost diffuse positive immunoreaction for AR.

The results of the immunoreactions for AR and HER-2/neu (Table 4) indicate that most G1 and G2 DCIS lesions (95% and 64%, respectively) were AR-positive but HER-2/neu-negative. In contrast, the majority of G3 DCIS lesions (64%) were positive for both AR and HER-2/neu.

Table 4. Immunoreaction Results for Androgen Receptor and HER-2/neu in Invasive Carcinoma and Ductal Carcinoma In Situ

	G1 DCIS	G2 DCIS	G3 DCIS	G1 ICA	G2 ICA	G3 ICA
G1: Grade 1 (well-differentiated); DCIS: ductal carcinoma in situ; G2: Grade 2 (moderately differentiated); G3: Grade 3 (poorly differentiated); ICA: invasive carcinoma; AR: androgen receptor.
AR-positive/HER-2/neu-positive (%)	0/19 (0)	1/11 (9)	16/25 (64)	0/20 (0)	5/40 (12.5)	26/85 (30.5)
AR-positive/HER-2/neu-negative (%)	18/19 (95)	7/11 (63.5)	3/25 (12)	18/20 (90)	25/40 (62.5)	13/85 (15)
AR-negative/HER-2/neu-positive (%)	0/19 (0)	0/11 (0)	5/25 (20)	0/20 (0)	3/40 (7.5)	10/85 (12)
AR-negative/HER-2/neu-negative (%)	1/19 (5)	3/11 (27.5)	1/25 (4)	2/20 (10)	7/40 (17.5)	36/85 (42.5)

Distribution of Steroid Receptors and HER-2/neu in Invasive Carcinomas

ER, PR, AR, and HER-2/neu demonstrated comparable expression profiles in DCIS cases. Among all invasive ductal and lobular breast carcinomas (ICAs), ER and PR were expressed more frequently in G1 (100% and 82% of cases, respectively) and G2 (61% and 26% of cases) ICAs compared with G3 ICAs (10% and 4% of cases). AR was expressed most frequently in G1 ICAs (88% of cases) and less frequently in G2 ICAs (71% of cases) and G3 ICAs (47% of cases). HER-2/neu showed an increase in overexpression frequency with increasing tumor grade, from 0% in G1 ICAs to 26% in G2 ICAs and 42% in G3 ICAs (Table 2).

Comparison of ER and AR (Table 3) revealed that the majority of G1 and G2 ICAs (90% and 55%, respectively) expressed both ER and AR; in contrast, the majority of G3 ICAs (52%) were both ER-negative and AR-negative. AR-positivity with ER-negativity was observed in 39% of G3 ICAs (Fig. 2).

Figure 2. Examples of immunoreactions in well-differentiated (G1) to poorly differentiated (G3) invasive breast carcinomas. (A-C) Well-differentiated mucinous carcinoma (A), with intense and diffuse positive reactions for (B) estrogen receptor (ER) and (C) androgen receptor (AR). (D-F) Moderately differentiated invasive lobular carcinoma, with intense and diffuse positive reactions for (E) ER and (F) AR. (G, H) Moderately differentiated invasive ductal carcinoma, (H) with diffuse positivity for AR. Tumor cells, however, were completely negative for ER and progesterone receptor (PR) in this case (not shown). (I-J) Poorly differentiated invasive ductal carcinoma, with (I) intense but (J) heterogeneous immunoreaction for AR. Tumor cells were completely negative for ER and PR (not shown).

The results of the immunoreactions for AR and HER-2/neu (Table 4) indicate that most G1 and G2 ICAs (90% and 63%, respectively) were AR-positive but HER-2/neu-negative. In contrast, 42% of G3 ICAs possessed a common AR and HER-2/neu positivity.

The results of IDC and ILC immunostaining were not significantly different from each other in all differentiation grades (Table 5). In summary, G1 and G2 ILCs exhibited a higher proportion of ER-, PR-, and AR-positive cases compared with G1 and G2 IDCs.

Table 5. Positive Immunoreactivity for Steroid Receptors and HER-2/neu in Invasive Ductal Carcinoma and Invasive Lobular Carcinoma, Grade 1 to Grade 3

	G1 IDC	G1 ILC	G2 IDC	G2 ILC	G3 IDC	G3 ILC
G1: Grade 1 (well-differentiated); IDC: invasive ductal carcinoma; ILC: invasive lobular carcinoma; G2: Grade 2 (moderately differentiated); G3: Grade 3 (poorly differentiated); ER: estrogen receptor; PR: progesterone receptor; AR: androgen receptor.
ER-positive (%)	17/17 (100)	3/3 (100)	19/31 (61)	9/9 (100)	8/83 (9.5)	0/2 (0)
PR-positive (%)	14/17 (82.5)	3/3 (100)	8/31 (26)	5/9 (55.5)	3/83 (3.5)	0/2 (0)
AR-positive (%)	15/17 (88)	3/3 (100)	22/31 (71)	8/9 (89)	39/83 (47)	0/2 (0)
HER-2/neu-positive (%)	0/17 (0)	0/3 (0)	8/31 (26)	1/9 (11)	35/83 (42)	0/2 (0)

Distribution of Steroid Receptors and HER-2/neu in Special Types of Invasive Carcinoma

Among the poorly differentiated invasive carcinomas, we found 12 cases (14%) that showed extensive areas of apocrine differentiation. Seven (58%) of these tumors were negative for ER and PR but positive for AR. One tumor (8.5%) was ER-positive but AR-negative. Three carcinomas with apocrine differentiation (25%) were negative for AR. Six carcinomas (50%) were HER-2/neu-positive, and 5 of these 6 (85%) also were AR-positive.

Two lobular G3 carcinomas (pleomorphic variant of invasive lobular carcinoma) were negative for ER, PR, AR, and HER-2/neu.

Statistical Analysis

The results of CPCA for noninvasive tumors (DCIS) are shown in Figure 3A. HER-2/neu expression and tumor grade were closely correlated with each other in DCIS and were inversely correlated with ER and PR expression. AR expression was found to be independent of other variables. Similar results were obtained with HCA (Fig. 3B). Tumor grade and ER, PR, and HER-2/neu expression all were merged into one group of variables at the beginning of clustering, whereas AR remained independent until the end of the analysis. According to the Fisher exact test and the chi-square test, all pairwise correlations within the set of variables containing tumor grade, ER, PR, and HER-2/neu were highly significant (P < 0.0005).

Figure 3. (A) Location of variables in a two-dimensional space as computed by categorical principal component analysis of in situ breast carcinoma data. Black points denote variables, and dotted lines denote projections of variables onto the origin. Parallel projections indicate a direct correlation, projections with opposite slopes indicate an inverse correlation, and perpendicular projections indicate no correlation. (B) Horizontal dendrogram representing steps in the hierarchic clustering of variables for ductal carcinoma in situ data. Variables merged in early steps of the analysis (i.e., variables with shorter distances between them) represent clusters and are thought to be closely correlated. AR: androgen receptor; PR: progesterone receptor; ER: estrogen receptor.

For invasive carcinomas, CPCA revealed a strong positive correlation between ER and PR expression and a negative correlation between tumor grade and receptor expression (Fig. 4A). In contrast, HER-2/neu expression and AR expression were not found to be correlated with any other variables.

Figure 4. (A) Location of variables in a two-dimensional space as computed by categorical principal component analysis of invasive breast carcinoma data. Black points denote variables, and dotted lines denote projections of variables onto the origin. Parallel projections indicate a direct correlation, projections with opposite slopes indicate an inverse correlation, and perpendicular projections indicate no correlation. (B) Horizontal dendrogram representing steps in the hierarchic clustering of variables for invasive breast carcinoma data. Variables merged in early steps of the analysis (i.e., variables with shorter distances between them) represent clusters and are thought to be closely correlated. AR: androgen receptor; PR: progesterone receptor; ER: estrogen receptor.

HCA of the invasive carcinoma results yielded similar findings (Fig. 4B). Tumor grade, ER expression, and PR expression formed a cluster in the first step of the analysis, whereas AR and HER-2/neu expression remained independent until the last step. Cross-tabulation confirmed these findings (P < 0.0005 [Fisher exact test] for all pairwise correlations within the set of variables containing tumor grade, ER expression, and PR expression).

DISCUSSION

Although several previous studies have shown the biologic and therapeutic significance of ER and PR in breast carcinoma,1–5 few, to our knowledge, have dealt with the role of AR in breast carcinoma. Studies that did focus on AR in breast carcinoma examined cultures of breast carcinoma cells27–29 or frozen material using biochemical techniques.20 To our knowledge, immunohistochemical determination of AR expression in breast carcinoma rarely has been performed using formalin-fixed, paraffin-embedded material.30 Previous studies of AR in breast carcinoma dealt primarily with a small number of invasive carcinomas or examined the presence of AR in certain subtypes of breast carcinoma, such as apocrine carcinoma.31–34 Furthermore, most of these studies focused on invasive carcinomas and did not include patients with intraepithelial neoplasia (i.e., in situ carcinoma).

The results of the current study clearly show that AR expression is a common feature of invasive and noninvasive breast carcinomas. The vast majority of G1 carcinomas were AR-positive, and AR-positivity was observed in 46% and 76% of G3 invasive and noninvasive carcinomas, respectively. As was expected, all well-differentiated (invasive and noninvasive) carcinomas were ER-positive, and only 9.5% of all poorly differentiated invasive carcinomas and 8% of all poorly differentiated noninvasive carcinomas were ER-positive.

One notable finding in the current study was yielded by the comparison of AR with ER in poorly differentiated carcinomas. Thirty-nine percent of invasive G3 carcinomas were ER-negative but AR-positive. Among noninvasive G3 carcinomas, a greater percentage of cases (68%) were ER-negative but AR-positive.

Another noteworthy finding resulted from the comparison of AR and HER-2/neu expression. Depending on the grade of differentiation, concordant or discordant immunochemistry results for AR and HER-2/neu were observed. Seventy-three percent of invasive G3 carcinomas exhibited concordant AR and HER-2/neu expression as determined by immunochemistry, and similar results were observed for G3 DCIS. In contrast, 90% of invasive G1 carcinomas were AR-positive but HER-2/neu-negative, and results were similar for G1 DCIS.

Regarding apocrine differentiation of carcinoma cells, some studies have reported a characteristic constellation of AR-positive and ER- and PR-negative immunoreactions.31, 32 It is well known that apocrine metaplastic epithelial cells within cystic areas of nonneoplastic breast (i.e., fibrocystic breast changes) characteristically are positive for AR but negative for both ER and PR.31, 34 In the current study, although the majority of examined invasive ductal carcinomas with apocrine differentiation were negative for ER and PR, 2 carcinomas (17%) showed intense and diffuse staining for ER (but not for PR). In contrast, most apocrine invasive ductal carcinomas were AR-positive. Three carcinomas, however, did not show any AR-positivity despite having typical apocrine differentiation. The two lobular carcinomas (pleomorphic variant) with prominent apocrine features were negative for ER, PR, and AR. These results suggest that although tumor cells with apocrine differentiation usually are ER-negative and AR-positive, in some cases these cells can be either intensely positive for ER or negative for AR.

Most examined invasive lobular carcinomas were AR-positive. The small number of cases of invasive lobular carcinoma, however, does not allow any statistically meaningful conclusions to be made in the current study. Future studies focusing on a larger number of cases of lobular intraepithelial neoplasia (i.e., lobular carcinoma in situ) and invasive lobular carcinoma are necessary for analysis of the occurrence of AR in this type of breast neoplasm.

It must be noted that some of the previous studies of AR in breast carcinoma were based on biochemical assays that measure androgen binding in cytosolic fractions of tumor homogenates.20 Several methodologic shortcomings of the cytosolic AR assay have been described,30, 35 including the thermolability of the receptor proteins and the binding of androgen to albumin and sex hormone–binding globulin. Furthermore, steroid receptor assays that use tissue homogenates are not capable of distinguishing between receptor-containing malignant and nonmalignant cells.

Using a monoclonal AR antibody on frozen sections of 76 primary breast carcinomas, Isola35 reported positive immunostaining in 79% of all tumors. (The cutoff point for ER-, PR-, and AR-positivity was 20% positively stained cells.) Nonetheless, the use of formalin-fixed, paraffin-embedded archival tissue by Isola yielded negative results, despite the application of several antigen-unmasking techniques. Among the breast carcinomas examined by Isola, 7 cases (9%) were negative for ER and PR but positive for AR. In the current study, using a monoclonal antibody against AR on formalin-fixed, paraffin-embedded archival material, we observed the presence of AR in 132 of 200 (66%) invasive and noninvasive breast carcinomas. More cases that were negative for both ER and PR but positive for AR were observed in the current study compared with the study performed by Isola.35

Statistical analysis of the results of the current study showed that in both DCIS and invasive carcinomas, AR was expressed independently of tumor grade as well as ER, PR, and HER-2/neu status. Furthermore, HER-2/neu expression was independent of tumor grade and ER, PR, and AR expression in invasive carcinomas. In DCIS, however, HER-2/neu overexpression did not appear to be independent of tumor grade, ER expression, or PR expression. As indicated by CPCA and HCA, the correlation between HER-2/neu expression and tumor grade was much more pronounced in DCIS than in invasive carcinomas; the biologic explanation for this phenomenon is unclear.

The frequent expression of AR in breast carcinoma cells, as observed in the current study, raises the important question of the interaction between androgens and human breast carcinoma.36, 37 Studies have shown that androgens may affect the growth of breast carcinoma in animals.36 For example, pharmacologic administration of androgens to rats bearing dimethylbenzanthracene-induced breast carcinoma leads to tumor regression.37 Tumor proliferation in human mammary carcinoma also is significantly altered by androgens.27 Approximately 20% of patients with metastatic breast carcinoma may experience tumor regression after treatment with androgens.38 In contrast, studies analyzing the effects of androgens and antiandrogens on breast carcinomas in long-term tissue cultures indicated that some human breast carcinomas, at least in vitro, may be stimulated by androgens.27, 36, 39 Furthermore, the results of adjuvant treatment with aromatase inhibitors, which block the conversion of adrenal steroids (mainly androgens) into estrogens, have been reported.40–44 These studies also underscore the important role of androgens (albeit in an indirect way, through estrogens) in the stimulation of human mammary carcinoma growth.42–44 Thus, androgens can have either stimulatory or inhibitory effects on tumor growth. These seemingly paradoxical effects may depend on carcinoma cell type and/or may be related to the presence or absence of other steroid receptors, such as ER and PR. In addition, the heterogeneity of carcinoma cells in terms of steroid receptor positivity and the proportional distribution of each steroid receptor among carcinoma cells may influence the activity of androgens in either a proliferative or inhibitory direction.

In summary, the current study demonstrates that androgen receptors frequently are expressed in breast carcinomas. Most notable was the finding by immunohistochemical analysis of poorly differentiated breast carcinomas that 39% of invasive carcinomas and 68% of DCIS were AR-positive but ER-negative.

Aside from ER and PR expression, the immunohistochemical assessment of AR expression may lead to new adjuvant hormonal treatment strategies for patients with breast carcinoma.

Acknowledgements

The authors thank Mrs. Margit Gogg-Kamerer and Mrs. Andrea Sommersacher for their excellent technical assistance.