Androgen receptor expresion in breast cancer: Relationship with clinicopathological characteristics of the tumors, prognosis, and expression of metalloproteases and their inhibitors

In the last two decades, the molecular mechanisms related to the hormone dependence of breast tumors have been extensively investigated and the role of the estrogen and progesterone receptors (ER and PgR) in promoting breast cancer has been well documented. However, the role of androgens and their receptors (AR) in breast cancer etiology and progression has been less profoundly studied and remains an unanswered question [1, 2]. There is evidence showing that androgens can directly stimulate the growth of human breast cancer cell lines [3]. In addition, both retrospective and prospective studies have reported statistically significant associations between increased levels of testosterone and higher breast cancer risk in both pre- and postmenopausal women [4‐6]. Likewise, AR is expressed in approximately 70% to 90% of invasive breast cancers, a frequency comparable with or higher than the one reported for ER (70–80%) and PgR (50–70%) [1, 7‐10]. Although a relationship between AR and both ER and PgR status has been demonstrated [10‐14], a significant percentage of tumors are positive for AR and negative for ER and PgR [13]. This finding reveals the independent expression of AR in human breast cancer. However, there are apparently divergent data on the biological and clinical signification of AR in breast cancer. AR have also been detected in a significantly higher percentage of AR-positive ductal carcinomas "in situ" (DCIS) adjacent to invasive carcinomas of the breast than in pure DCIS lesions [15], suggesting that AR correlates with tumor invasiveness, at least in the early phases of tumor progression. In invasive breast carcinomas, AR-positive tumors have been associated with a low or intermediate histological grade (G1, G2) [10, 13, 14, 16, 17]. In addition, certain types of breast carcinoma, even high grade ones, are typically ER- and PR-negative, but AR-positive; a typical example of such tumors is the apocrine breast carcinoma [18, 19]. However, the expression level of both the AR gene and the AR protein in breast cancer was found to be positively correlated with axillary lymph node involvement [20]. In addition, it is remarkable that among the steroid hormone receptors, the androgen receptor is the best preserved one during metastases development and is expressed in the majority of metastatic tumors [8, 21]. There is evidence as well indicating that AR/steroids are able to up-regulate matrix metalloproteases (MMPs), contributing to invasiveness via destruction of basement membrane and extracellular matrix [22, 23]. Nevertheless, only a few studies have examined the impact of AR expression on patient prognosis in early breast cancer. Patients with AR-positive tumors were shown to have a significant trend toward longer relapse-free and/or overall survival in the univariate analysis than those patients with AR-negative tumors [1, 16, 24], but none in the multivariate analysis [1, 16]. In addition, other studies have not found any significance of AR expression in predicting prognosis in breast cancer [20, 25].

Since AR are expressed by an important percentage of breast carcinomas and there are evidences pointing their role in tumor progression, in the present study we analyze the tumor expression of AR, its relationship with clinicopathological characteristics, with several MMPs and their inhibitors (TIMPs) and with prognosis, in patients with breast cancer.

More than 2,800 determinations were performed on TMAs of cancer specimens from 111 patients with primary invasive ductal carcinoma of the breast and controls. Minimal internal variance of score data between duplicate tissue cores from the same patients was detected in the tissue arrays, showing a high agreement for each protein (r > 0.95 and p < 0.0001, for each protein). In the validation study there was a total concordance in the global expression, as well as in the intensity of immunostaining for AR, between TMA cases and the corresponding whole-tissue sections. In addition, there were highly significant correlations in the immunostaining scores between these two paired sets (r > 0.90 and p < 0.0001, for each protein).

Figure 1 shows examples of tissue immunostained for the proteins evaluated. As it was expected, AR immunostaining was of nuclear localization. A total of 83 cases (74.8%) showed a positive immunostaining for AR, although with a wide variation in the immunostaining score values (Figure 2). There were only five cases with a score value of less than ten points (three with a score of 5, and two with a score of 8). Immunostaining for MMPs and TIMPs was localized predominantly in the cytoplasm of the malignant cells, but also in stromal cells in a considerable percentage of cases.

Bild vergrößern

Bild vergrößern

There were not any significant associations between the total immunostaining scores for AR and clinicopathological parameters such as tumor size, lymph node involvement, stage, histological grade, high Nottingham prognostic index, infiltrating edge, vascular invasion, desmoplastic reaction or peritumoral inflammation, ER or PgR status (Table 1).

In the present study we also investigated the possible relationship between AR expression and both MMPs and TIMPs expression in tumors, which have been associated with an aggressive behaviour and a poor prognosis in breast cancer patients. Our data demonstrated some significant associations. Thus, score values for MMP-1, -7 and -13, were significantly higher in AR-positive tumors than in AR-negative tumors (Table 2). Likewise, when we considered the cellular type expressing each factor, we found that the scores for AR were significantly higher in malignant cells from tumors positive for MMP-1, -7, -11, and TIMP-2, as well as for MMP-1 in intratumor fibroblasts (Table 3). However, there was not any significant association between MMPs or TIMPs expression by mononuclear inflammatory cells and the AR status of the tumors (Table 3).

We initially investigated the possible association between each immunostaining score value for AR, as cut-off points, and relapse-free survival. We found that none of these cut-off points were significantly associated with relapse-free survival in our patient population (data not shown). However, our results demonstrated that when patients were dichotomized in two different groups with regard to the more optimal cut-off point of score values for AR (score = 0 v.s. score > 0), patients with AR-positive tumors had a significantly longer survival than patients with AR-negative tumors (p = 0.01) (Figure 3), but there was no difference regarding the occurrence of distant metastases (data no shown). In addition, and in accordance with previous studies indicating the prognostic value of AR expression in ER-negative tumors, we investigated that value in the subset of 59 ER-negative tumors included in the present study, but we could not find any significant value of AR expression able to predict either relapse-free or overall survival in the corresponding patients (data not shown). Multivariate analysis according to Cox model demonstrated that tumor stage (stage II: relative risk (RR) (confidence interval): 3.46(1.11–10.78); stage III: 7.29(2.37–22.41); p < 0.001) and PgR status (positive: 0.19(0.08–0.49), p < 0.001) were significantly associated with overall survival. Multivariate analysis also confirmed that patients with AR-positive carcinomas had a significant longer overall survival than those with AR-negative breast neoplasms (AR-positive: 0.46(0.23–0.93), p = 0.03).

Bild vergrößern

Our results confirmed previous biochemical and immunohistochemical studies indicating that AR are expressed in a considerable proportion of breast carcinomas [1, 7‐10]. We also found that AR expression in breast carcinomas is highly variable because of tumor heterogeneity. We determined that there are some correlations between AR and MMPs expression. However, AR expression was not related with the occurrence of distant metastases, but instead associated with a longer overall survival in breast cancer patients.

Contrary to other studies, we found no significant differences between AR expression and the clinicopathological characteristics of the tumors, such as histological grade [10, 13, 14, 16, 17, 29], axillary lymph node involvement [20], ER or PgR status [10‐14]. Nevertheless, we found a positive and significant relationship between AR expression and the expression of parameters potentially indicative of invasiveness, such as MMP-1, -7 and -13. This association could be relevant because it is now known that MMPs promote metastases not only by modulating the remodelling of extracellular matrix, but because MMPs are able to impact in vivo on tumor cell behaviour as a consequence of their ability to cleave growth factors, cell surface receptors, cell adhesion molecules, and chemokines/cytoquines [30‐33]. Furthermore, by cleaving pro-apoptotic factors, MMPs are able to produce a more aggressive phenotype via generation of apoptotic resistant cells [34]. MMPs may also regulate cancer-related angiogenesis, both positively through their ability to mobilize or activate pro-angiogenic factors [35], and negatively via generation of angiogenesis inhibitors, such as angiostatin and endostatin, cleaved from large protein precursors [36]. In addition, it is now understood that TIMPs are multifactorial proteins also involved in the induction of proliferation and the inhibition of apoptosis [37, 38]. In a prior report we found an increase in AR expression in the transition from pure DCIS to DCIS adjacent to the invasive component of breast carcinoma, which led us to consider that there could be a relationship between both AR and MMPs/TIMPs expression in breast carcinoma [39]. There are evidences indicating the existence of a steroid regulation of the gelatinases (MMP -2 and MMP -9) in both breast [40, 41] and prostate cancer [42, 43]. Likewise, it is of note that the expression of MMP -13 (collagenase-3), which has been associated with the microinvasive component of "in situ" carcinomas [44], has been found to be up-regulated by androgens in prostate cancer derived the cell line LNCaP [23]. Nevertheless, the regulation of MMPs production by androgens seems to be a quite complex process. Thus, experimental studies showed that androgens, via AR-Ets, negatively regulate the expression of interstitial collagenase (MMP -1), stromelisin-1 (MMP -3), and matrilysin -1 (MMP -7) [45]. Even so, in despite of assuming an association between AR expression and some MMPs/TIMPs production in the context of breast cancer, we could not determine any significant relationship between AR status and the occurrence of distant metastases. This may be due to the consideration of the cellular type expressing each factor in the tumor scene. In accordance with other authors, we found that AR immunoreactivity is localized in the nuclei of tumor cells and no stromal staining was observed [1, 13]. However, there is a biological variability with regard to the cellular type expressing MMPs or TIMPs (cancerous cells and/or stromal cells -fibroblasts or mononuclear inflammatory cells-). When we considered this morphological aspect, we found that AR-positive tumors had a higher percentage of cases positive for MMP-1, -7, -11, and TIMP-2 in their malignant cells, when compared to AR-negative tumors. The only association with AR-positive status in stromal cells was for MMP-1 in intratumor fibroblasts. We believe that these findings could explain our results pointing the lack of any significant association between AR status and the occurrence of distant metastases because, such as it was recently reported from our group, the expression of these MMPs and the TIMP-2 correlate with distant metastases mainly when those are expressed by stromal cells [15, 28]. Thus, our results led us to consider the existence of a regulation of MMPs/TIMPs expression via AR in the same tumor cells, but without a significant influence in the development of distant metastases. We considered that both AR and MMPs/TIMPs expression could be more important in the early phases of tumor progression, but less in primary invasive breast carcinomas.

On the other hand, our data demonstrated that the AR status correlates significantly and independently with overall patient survival. Other authors have also found that breast cancer patients with AR-negative tumors show a trend toward a shorter overall survival than those patients with AR-positive tumors [24]. It has been proposed that this trend may be secondary to the AR-positive tumors' capability to retain a hormone-sensibility that confers a low biological aggressiveness. In fact, among the steroid hormone receptors, AR is the best preserved one during metastases development and is expressed in the majority of metastatic tumors [21]. Furthermore, the effects of tamoxifen [46] and medroxyprogesterone acetate are mediated by AR [47]. A recent study showed that reduced levels of AR or impaired AR function contribute to the failure of medroxyprogesterone acetate therapy, potentially due to the abrogation of the inhibitory effect of AR on ER signaling [48]. In addition, Aggof et al. have reported a significant association in the univariate analysis (p = 0.049) between AR expression and relapse-free survival in patients with ER-negative tumors (n = 57), but none with overall survival [16]. In the present study we did not find this association. Nevertheless, there are possible explanations for that discrepancy with our results due to differences in the studied patient populations. Thus, although we have a similar number of patients with ER-negative tumors (n = 59), it is of note that our study included a higher number of events (tumor relapses) (61%) than in the study of Aggof et al, (33%), because our population was selected stratifying on the basis of the occurrence of distant metastases. Likewise, in our study we applied different criteria for patient selection, such as ductal being the chosen histological type, considering distant metastases as the only type of tumor recurrence, and including only T1 and T2 tumors. On the other hand it is remarkable that Schippinger et al., did not find in their multivariate analysis any independent prognostic value for AR-expression in patients with metastatic breast cancer [25]. Nevertheless, this patient population differs clinically of that of non-metastatic breast cancer included in our study. Even so, the latter finding seems to indicate that the prognostic significance of AR status may be lost once distant metastatic disease occurs.

Our results confirm that AR are commonly expressed in breast cancer, and correlate with the expression of some MMPs and TIMP-2. Although we found a value of AR expression to be a prognostic indicator in breast cancer, the functional role of AR in these neoplasms is still unclear and further data are needed to determine their biological signification in breast cancer, i.e. if AR could be used as a marker for efficiency of endocrine therapy and for new hormonal therapeutic strategies in women with ER-negative carcinomas, as well as on the prognostic signification of AR microsatellites polymorphism in breast cancer, which could affect the transactivation capacity of the receptor [49], and whether or not patients with ER-negative, PR-negative, but AR-positive cancers behave differently from those with triple negative breast carcinomas.