Purpose: The peripheral benzodiazepine receptor (PBR) expression has been shown dramatically increased in neoplastic tissues and tumor cell lines originated from ovary, liver, colon, breast, or brain relative to untransformed tissues. Its expression has been also associated with tumor progression and aggressiveness. To explore whether PBR expression level could be of prognostic value in invasive breast cancer, we studied a series of 117 patients who underwent surgery for primary breast carcinomas and were followed-up for 8 years.
Experimental Design: Using an immunohistochemical approach, we first compared PBR expression in normal and tumoral tissues, then we studied PBR expression together with clinicopathological variables (histological type, histological grade, lymph node, estrogen and progesterone receptor status), and biological markers such as BclII, Ki-67, and HER2/Neu.
Results: Our results revealed a significant increase of PBR expression in tumoral versus normal breast cells. We found a negative correlation between PBR expression and estrogen receptor status (P = 0.03) as well as a positive correlation between PBR and Ki-67 (P = 0.044). Although the disease-free survival was not affected by PBR in the whole population, high PBR expression level was significantly correlated with a shorter disease-free survival in the lymph node-negative patients, P = 0.038.
Conclusions: As the axillary lymph node-negative status is generally considered as a good prognosis factor, the high expression of PBR in this patient subgroup may be used to identify a new high risk population, for which a more specific therapy would be beneficial.
The peripheral benzodiazepine receptor (PBR) is a mitochondrial Mr 18,000 D protein, which was shown to modulate a variety of cell processes, including steroidogenesis, immune responses, apoptosis, mitochondrial oxidative phosphorylation, and cell proliferation (for recent review, see Ref. 1 ). PBR is present in peripheral tissues as well as in the central nervous system and exhibits various expression levels. Glandular and steroid producing tissues (adrenal glands and gonads) are particularly rich in PBR, whereas other tissues show relatively low to intermediate levels (2, 3, 4) . PBR expression has been shown to modulate in different pathological conditions, including brain lesions or trauma (5) , stress (6) , inflammation (7) , and cancer (8) . Specially, some of the highest densities of PBR are observed in neoplastic tissues and cell lines. For instance, ovarian, hepatic and colonic carcinomas, adenocarcinoma, and glioma (8, 9, 10) all show increased PBR densities relative to the untransformed tissues. Higher levels of PBR density were also observed in more rapidly proliferating breast cancer cells (11 , 12) , and the PBR gene was recently found to be amplified in an aggressive breast cancer cell line relative to a nonaggressive cell line (13) . In addition, Hardwick et al. (14) showed that PBR levels in breast cancer cell line-derived subclones correlated with the ability of cells to grow in vivo when implanted into scid mice. Taken together with the increased PBR expression and in relation to tumor aggressiveness, a nuclear and perinuclear localization of the protein was described in breast cancer cell lines, in aggressive metastatic human breast tumor biopsies (12) , and in glioblastoma biopsies (15) . The nuclear localization of PBR is thought to participate in the development and progression of the disease as it would regulate cell proliferation by facilitating cholesterol transport into the nucleus (12) . These observations lead to the hypotheses that (a) the presence of PBR may be a determinant factor for the aggressive phenotype of the tumor, and thereby, (b) the expression of PBR may be monitored in tumor samples for diagnostic and/or prognostic purposes in the clinic.
Supporting this, PBR expression was shown to be positively correlated with tumor malignancy grade and negatively with patient survival in human astrocytoma (16) . In addition, one study recently demonstrated the prognostic relevance of PBR in colorectal cancer. Precisely, using an immunohistochemical approach, Maaser et al. (17) showed that the mean survival of patients with stage III colorectal cancer was reduced by 35% when patients presented with high PBR expression.
In the present study, to address the relevance of PBR expression as a prognostic marker in human breast cancer, we investigated PBR expression in a series of 117 patients with invasive breast carcinomas, which were followed-up for a median 8-year period. We used a monoclonal antibody, specific for the human PBR and examined its relationship with well-established prognostic factors, including standard histological criteria [tumor grading, nodal status, estrogen (ER) and progesterone (PR) receptor status] immunohistochemical markers of cell proliferation (Ki-67), cell death (using the Bcl-2 proto-oncogene), and with the disease-free survival (DFS).
PATIENTS AND METHODS
From 1992 to 1993, 117 patients with primary invasive breast carcinoma underwent primary surgical tumor resection completed with axillary dissection at the Val d’Aurelle Cancer Center in Montpellier, France. One patient presented with two tumors, but only data on one specimen were included in the analysis. The median age was 61 years (range, 26–83 years). Patients with conservative breast surgery underwent postoperative radiotherapy. After surgery, 18% patients received adjuvant chemotherapy and 70% patients received adjuvant hormonal therapy. According to the Val d’Aurelle Cancer Center routinely assessed clinical management of the disease, patients with lymph node involvement underwent adjuvant chemotherapy, whereas patients without lymph node involvement but with positive receptors status underwent hormonotherapy. Clinicopathological parameters, which include histological type, Scarff-Bloom-Richardson (SBR) grade lymph nodes status, Tumor-Node-Metastasis classification (Union International Contre le Cancer; Ref. 18 ), estrogen and progesterone status, menopausal status, were documented. Patients were observed for disease recurrence and death, with a median follow-up of 98 months (range, 30–119 months).
At surgery, all patients had two small portions of the tumor removed. One portion was formalin-alcohol-fixed, paraffin-embedded and subsequently processed with routine techniques followed by immunohistochemical analysis. The other portion was snap-frozen in liquid nitrogen and stored at −80°C for ER and PR analysis as described previously (19) .
Five-μm-thick tumor slides were stained with H&E for the histopathological study. Tumor grading was performed according to the methodology described in Ref. 20 . Mitosis counts were performed in 10 consecutive high power fields (×400) using a Leica microscope (Leitz DMRB). Axillary lymph node status was assessed for each case by histopathological examination for a minimum of seven lymph nodes.
The expression of PBR, BclII, Her2/Neu, and the proliferative marker Ki67 was analyzed using an immunohistochemical procedure. The antibodies used were a mouse monoclonal antihuman PBR [8D7, dilution 1:300 (21)] , a mouse monoclonal anti-Bcl-2 antibody (clone 124, dilution 1:50; DAKO), a mouse monoclonal anti-Her2/Neu (CB11), and the anti-Ki67 MIB-1 antibody (dilution 1:100; Immunotech). Two-μm-thick paraffin-embedded sections of tumor samples were analyzed and mounted on DAKO-silanized slides. All procedures were carried at room temperature. Immunohistochemical detection of the different markers was done using the streptavidin-biotin (LSAB) method (LSAB kit; DAKO). The sections, which had been preincubated with 3% H2O2 solution for 10 min to block endogenous peroxidase, were incubated for 20 min with blocking agent for 2 h with the different primary antibodies; they were then rinsed and incubated with the secondary antibody for 10 min. They were then incubated with streptavidin conjugated to horseradish peroxidase: a positive reaction was visualized with 3-amino-9-ethylcarbazol. Before mounting, the sections were counterstained with Mayer’s hematoxylin. For the negative control, the primary antibody was omitted and replaced by an irrelevant antibody (monoclonal mouse antihuman IgG; DAKO). For the positive control, sections from normal breast structures were used. Double immunostaining for PBR and Ki-67 was performed on 41 samples using a sequential immunoenzymatic double staining method as previously described (22) , using the mouse monoclonal antihuman PBR and the rabbit anti-Ki-67 primary antibodies, revealed with the Envision reagent (DAKO).
The different marker’s immunoreactivity was then evaluated by two observers using a high-power lens (×400).
Semiquantitative Evaluation of PBR, Bcl-2, Ki-67, and HER2/Neu Staining.
The different antibody labeling was evaluated using a semiquantitative method taking into account the staining intensity and the number of stained cells in different random fields. For PBR, the staining intensity of tumor tissue was compared with that of the corresponding normal tissue for each patient (0, no increase; 1, weak increase; 2, moderate increase; and 3, strong increase). Values were then calculated as the product of the increase in staining intensity and the frequency of stained cancer cells (the latter was 0, <10%; 1, 10–25%; 2, 25–50%; 3, >50%). Finally, the staining for PBR was scored as 0, 1 (values 1 and 2), 2 (values 3 and 4), and 3 (values 6 and 9). Scores of 2 and 3 referred to the overexpression of PBR. A similar score calculation was used for Bcl-2. For Ki-67, staining 0 means no expression of the marker; 1, weak expression of the marker; 2, moderate expression of the marker; and 3, strong expression of the marker. For the determination of HER2 protein expression, only the membrane staining intensity and pattern in the infiltrating component of cancers were evaluated according to the DAKO Herceptest scale. Score 0: no staining at all, or membrane staining in <10% of the tumor cells. Score 1+: although there is a faint membrane staining in >10% of the tumor cells, with just some membrane portions stained, the labeling is considered as negative. Score 2+: a weak to moderate staining of the whole membrane in >10% of the tumor cells, the labeling is considered as weakly positive. Score 3+: a strong staining of the entire membrane in >10% of the tumor cells. No staining of the normal breast structures indicated a valid assay.
Correlations between the clinicopathological data and the expression of the four immunohistochemical markers analyzed were assessed using the χ2 tests. Locoregional disease relapse, distant metastasis, second primary, and death due to cancer were considered as end points for DFS. DFS curves starting from the date of surgery until the first event were estimated using the Kaplan-Meier method. Patients alive without ever failing from disease were censored at the last follow-up. The statistical significance of each variable was evaluated for prognosis using the log-rank test for univariate analyses and the Cox proportional hazards model for multivariate analyses. For all statistical analyses, P < 0.05 was considered statistically significant. A subgroup analysis for the node-negative patient population was also investigated.
Patient Clinicopathological Characteristics.
Patients were characterized according to their age, their menopausal status, the tumor grade, the axillary nodal status, according to the Elston and Ellis modification of the Scarff-Bloom-Richardson grading system (SBR) and the Tumor-Node-Metastasis staging (based on the Union International Contre le Cancer atlas criteria; Ref. 18 ). All of the data are included in Table 1⇓ .
PBR Expression in Invasive Breast Carcinoma.
PBR expression was addressed in human breast carcinoma and breast normal tissues using a monoclonal antihuman PBR antibody (8D7) targeting the COOH-terminal end of PBR (21) . Normal breast components (ductal and acinar epithelial cells) showed a homogeneously distributed staining regarding cells or structures. They showed a weak granular cytoplasmic immunostaining, with a typical mitochondrial localization. In some rare cases, the staining was strong. In many ducts, there was a basal or/and a luminal increase of the staining. A very similar staining was observed in dystrophic structures (Fig. 1A)⇓ .
Considering tumor tissues, 23 patients (20%) exhibited no or little PBR expression (scores 0, 1), whereas 94 patients (80%) exhibited a strong staining (scores 2, 3; Table 2⇓ ). In most malignant breast sections, PBR immunostaining was heterogeneously distributed in cells and structures (Fig. 1B–D)⇓ . When the staining in a structure was weak, it dramatically increased at the periphery of the structure, in the infiltrating cancer lobules. Intraductal components of infiltrating carcinomas generally showed only a weak staining. At the cellular level, PBR immunostaining was cytoplasmic and granular (varying from ten to hundreds of labeled granules). A perinuclear localization was often observed. The perinuclear labeling was more intense than the cytoplasmic one in some tumor cells. No labeling was obtained on cytoplasmic membranes and nuclei.
Bcl-2, Ki-67, and HER2 Expression.
Bcl-2 immunoreactivity was studied on 114 patients because it could not be detected in 3 patients due to a small tumor sample available for analysis (Table 2)⇓ . The staining was always cytoplasmic (data not shown). Forty-six tumors patients (40%) demonstrated no or little Bcl-2 staining (scores 0–1), and 68 (60%) exhibited a strong staining (scores 2–3). MIB-1 anti-Ki-67 antibody nuclear staining was studied in 111 patients. The staining was weak for 47 patients (42%), intermediate for 23 patients (21%), and strong for 41 patients (37%). HER2 protein expression was determined in 76 patients and only 10 patients (13%) showed high expression levels according to the DAKO Herceptest scale (scores 2+ and 3+).
Association between PBR Overexpression and Clinicopathological Variables.
No correlation was obtained between PBR overexpression (scores 2, 3) and the age, menopausal status at diagnosis, clinical stage, SBR grade, or tumor size. More patients without an axillary lymph node involvement (N−) were found to overexpress PBR (84% in N− patients versus 75% in N+ patients), but the difference was not statistically significant (P = 0.23, Table 3⇓ ). Also, all 6 T3 patients overexpressed PBR (P = 0.12). In addition, considering the receptor status, all but 1 ER-negative patient overexpressed PBR (97% in ER- patients versus 80% in ER+ patients), and the difference was statistically significant (P = 0.03). However, PBR overexpression could not discriminate patients according to their PR status (P = 0.31).
Associations between PBR Expression and Bcl-2, Ki-67, and HER2/Neu Immunostaining.
No significant relationship was found between PBR and Bcl-2 expression (P = 0.36; Table 3⇓ ). All 10 patients who overexpressed HER2 showed high PBR expression level concomitantly (100%), but the difference with low expression HER2 population where 86% patients expressed high PBR levels was not statistically significant (P = 0.21). By contrast, a significant positive relationship was found between PBR and Ki-67. The difference considering PBR overexpression between low and high expressing Ki-67 patients was statistically significant, with P = 0.044 (72 and 88%, respectively). Consistent with this, a double labeling showed that mitotic cells were often strongly stained with PBR antibody (Fig. 2)⇓ .
The median follow-up was 98 months (range, 30–119 years). Only 1 patient was lost to follow-up. At the time of the analysis, 36 patients failed, among them 11 died. The survival rates were 97 and 87%, and the DFS rates were 76 and 67% at 5 and 8 years, respectively (Fig. 3A)⇓ . The impact of the different variables on DFS was determined in the overall population using an univariate analysis (Table 4)⇓ . Patients with a SBR grade III had a lower DFS rate (50% at 5 years) than patients with SBR I or II (84% at 5 years, P = 0.001; Table 4⇓ ). Patients with a T3 size tumor (>50 mm) had a lower DFS rate (33% at 5 years) than patients with T1 or T2 (DFS = 84 and 71% at 5 years, P = 0.006). On the contrary, lymph node involvement was not a relevant prognostic factor for DFS (P = 0.71). PBR expression did not impact the DFS rate in the overall population; they were 77 and 69% for patients scored 2 and 3, respectively, which was not significantly different from that of low PBR scores (100 and 84% for scores 0 and 1, respectively, P = 0.10). Among the three other markers studied, only Bcl-2 had a significant impact on DFS with 5-year survival rates of 64 and 83% for low and high expressions, respectively (P = 0.006; Table 4⇓ ).
The patients were then dichotomized according to their nodal status (Table 5)⇓ . Interestingly, although no specific impact of PBR expression was obtained in the n + population (data not shown), we observed that PBR expression has a negative impact on DFS in the N− population. Among the 54 N− patients with high PBR expression level, 18 patients (33%) relapsed (DFS = 69%), whereas no relapses (DFS = 100%) were observed among the 10 N− patients with low PBR expression level. Here, the difference between PBR high and low expression levels was statistically significant with P = 0.038 (Fig. 3B)⇓ .
This article described for the first time the assessment of the expression of PBR as a prognostic factor in patients with breast cancer. To this aim, using the 8D7 monoclonal anti-PBR antibody that specifically and exclusively recognized the human PBR, we analyzed the expression of PBR in breast tumor biopsies from a group of 117 patients with operable primary breast carcinoma, followed-up for a median 8 year-period. PBR expression was investigated together with common clinicopathological variables (histological type, histological grade, lymph node, and ER and PR status) and biological markers (BclII, Ki-67, and HER2/Neu). In this study, although no correlation was observed in the overall population, we demonstrated that high PBR expression may be of prognostic value considering the lymph node-negative patients. We evidenced that high PBR expression level correlated with a poor prognostic in lymph node-negative invasive breast carcinoma, where it is associated with a shorter DFS, P = 0.038.
The comparison of PBR protein expression level in normal versus breast tumor biopsies studied here mirrors that seen in other human cancers. In normal breast tissues, PBR staining was homogeneous, either strong or weak, and of granular type, which indicates a mitochondrial localization. By contrast, PBR immunoreactivity was rather heterogeneous in breast carcinoma; very often PBR exhibited high expression level, as shown by the intense staining of many areas in a tumor and the weak staining of surrounding normal structures. Mitotic cells were positive for PBR expression, and we observed that high PBR expression was related to high expression of the proliferative marker, Ki-67 (P = 0.044). Importantly, the cellular distribution of PBR did not differ in normal and tumoral tissues; both clearly exhibit a cytoplasmic and mostly mitochondrial pattern. In a previous study (12) , PBR has been described primarily localized in and around the nucleus in aggressive metastatic human breast tumor biopsies. Contrastingly, here and as in the recently published study performed in colorectal carcinoma (17) , we did not evidence a nuclear localization of the protein in breast cancer cells. We observed an increase of the labeling around the nucleus, which was always of a granular type, consistent with the mitochondrial localization of the protein.
To a functional point of view, PBR has been implicated in cell proliferation and apoptotic process. On the one hand, PBR expression is associated with the regulation of the proliferation rate of cancer cells (11 , 23) , on the other hand, the protein per se exhibits antiapoptotic properties (24) . Collectively, these PBR-mediated effects would favor tumor cell proliferation and survival and thereby contribute to the poor prognostic in patients with elevated PBR expression.
Regarding breast cancer, studies in breast cancer cell lines, animals models, and human biopsies have suggested a close correlation between the expression of PBR and the progression of cancer. However, until this study, no direct data were available regarding survival. In breast cancer, the lymph node status is currently one of the best prognostic factors, but alone it is not sufficiently accurate to predict the clinical course of the disease (25) . The repertoire of the predictive factors contains many different markers characterized thus far, but early prognostic markers, which are significantly sooner at the beginning of tumor growth, are scarce. In that context, given the impact of PBR high expression on DFS in lymph node-negative patients, the determination of PBR status may help to identify a high-risk population early in the tumor process. As the axillary lymph node-negative status is generally associated with a good prognostic, high PBR expression in those patients should be taken into account to establish a more adequate and likely more aggressive therapy.
We thank Nadine Lequeux and Sylvie Roques who provided expert technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Pierre Casellas, Immunology-Oncology Department, Sanofi-Synthelabo Recherche, 371 rue du Professeur Joseph Blayac, 34184 Montpellier cedex 04, France. Phone: 33-4-67-10-62-90; Fax: 33-4-67-10-60-00; E-mail:
- Received June 29, 2003.
- Revision received December 22, 2003.
- Accepted December 29, 2003.