Preferential Nuclear and Cytoplasmic NY-BR-1 Protein Expression in Primary Breast Cancer and Lymph Node Metastases

Purpose:NY-BR-1 is a recently isolated differentiation antigen, which is expressed in normal mammary tissue and in breast cancer. However, current data are based on RT-PCR analysis and nothing is known about the presence of NY-BR-1 on a protein level. We previously generated a monoclonal antibody to NY-BR-1 to study the protein expression of NY-BR-1.

Methods: In our immunohistochemical study, NY-BR-1 was analyzed in normal tissues, various tumor types, 124 primary breast cancers, and 37 paired lymph node metastases.

Results: Among normal tissues, NY-BR-1 was present solely in ductal epithelium of the breast. In tumors, carcinoma in situ and invasive carcinoma of the breast were NY-BR-1 positive whereas other tumors and normal tissues were negative. Sixty percent of invasive breast carcinomas were NY-BR-1 positive, displaying cytoplasmic and/or nuclear immunoreactivity. This coexpression was verified by confocal microscopy. Although the monoclonal antibody identified intratumoral heterogeneity, a majority (72%) of NY-BR-1-positive carcinomas revealed immunoreactivity in >50% of the tumor cells. NY-BR-1 expression was more frequent in estrogen receptor–positive and lymph node–negative primary carcinomas (P < 0.05 each) and was more common in grade 1 (77%) than in grade 2 (63%) or grade 3 (50%) carcinomas (P < 0.05). This suggests that NY-BR-1 expression is lost with tumor progression. Forty-nine percent of lymph node metastases were NY-BR-1 positive.

Conclusion: This study supports the notion that NY-BR-1 is a differentiation antigen of the breast, which is present in normal and tumorous mammary epithelium. The organ-specific expression of NY-BR-1 and its high prevalence in metastases indicate that it could be a valuable target for cancer immunotherapy.

Differentiation antigens are expressed in cells of particular lineage or in cells at specific stages of differentiation (1, 2). Expression of differentiation antigens is often preserved in their neoplastic counterparts. Their cell/tissue specificity makes differentiation antigens useful markers for the differential diagnosis in surgical pathology. Consequently, differentiation antigens, such as CD20 for B-lymphocytes and gp100 as detected by monoclonal antibody (mAb) HMB45 for melanomas, have become valuable tools in surgical pathology. Based on their expression pattern and their presence in tumors, differentiation antigens seem to be ideal targets for the immunotherapy of cancer (2–7). Various methods such as autologous T-cell assays, serological analysis of recombinant expression libraries (SEREX), representational difference analysis, and others have been used to identify novel antigens for immunotherapy.

We have recently identified NY-BR-1 by SEREX analysis of a patient with metastatic breast carcinoma (3, 8, 9). The NY-BR-1 gene has been mapped to chromosome 10p11-12 and is composed of 37 exons. It encodes a peptide of M_r 150,000 to 160,000, which is regarded as a putative transcription factor. NY-BR-1 mRNA expression was found in a high percentage of breast cancers (84%) and normal breast tissue, and at a much lower level in normal adult testis, but not in other normal tissues (8). The present knowledge about NY-BR-1 expression is primarily based on RT-PCR analysis. However, knowledge about the actual protein expression of NY-BR-1 is crucial in assessing its therapeutic and diagnostic potential (2, 4, 10, 11). We recently generated several mAbs to NY-BR-1 for the analysis of NY-BR-1 on a protein level.⁵

However, the determination of antigen expression is typically carried out using primary tumor samples although immunotherapy mostly targets metastases, which can manifest years after the resection of the primary. Predicting the response to therapy in metastatic disease on the basis of biomarkers analyzed in the primary tumor is based on the assumption that the status of these markers does not change during metastatic progression. Consequently, the present study analyzes the presence of NY-BR-1 in an extensive panel of various tumor tissues including invasive and in situ carcinomas of the mammary gland as well as in a panel of normal tissues. We will show that the expression of NY-BR-1 is restricted to normal breast epithelium and to tumors thereof, confirming its true nature as a differentiation antigen of the mammary gland. We will also show that NY-BR-1 expression is lost during tumor progression.

Generation of an anti-NY-BR-1 mAb. The generation of NY-BR-1 mAbs NY-BR-1 clone 1 and NY-BR-1 clone 3 has been carried out according to standard procedures⁶

and as previously described for several other serologic reagents to other tumor-associated antigens such as MAGE-A1 (12, 13), CT7 (14), and NY-ESO-1 (15). NY-BR-1-specific hybridomas were generated using standard fusion protocols (8, 16). The two anti-NY-BR-1 specific clones were selected based on ELISA, Western blotting, and immunohistochemical analysis. For the immunohistochemical analysis, culture media of hybridoma clones or antibodies purified from ascetic fluid of mice have been used. For the present analysis, mAb NY-BR-13 was selected.

Patients and tissues. Formalin-fixed paraffin-embedded tissues were retrieved from the archives of the Institute of Surgical Pathology of the University Hospital Zurich. Four-micrometer sections were applied to histology slides for immunohistochemistry (Superfrost Plus, Menzel, Braunschweig, Germany). Paraffin sections were dried overnight at 60°C. Conventional large sections as well as tissue microarrays of normal and neoplastic tissues were tested as indicated in Table 1. Five different samples of normal tissue were used. Tumor-specific tissue microarrays with large numbers of endometrial (n = 352) and ovarian carcinomas (n = 268) have been recently constructed as described and were also used to verify the specificity of NY-BR-1 expression for breast cancer (17, 18). Naturally, breast carcinomas were a focus of the present analysis for which 124 breast cancer cases were randomly selected. These consisted of 107 invasive ductal carcinomas, 7 lobular, 2 tubular, 2 micropapillary, 1 papillary, 1 cribriform, 1 apocrine, 2 metaplastic, and 1 mucinous carcinoma, according to the WHO tumor classification (15). The median age of the patients was 59 years (range, 28-91 years). Follow-up ranged from 6 to 24 months. Patients were operated with segmental resection or mastectomy and had not received any preoperative therapy. All cases were reevaluated by one pathologist (Z.V.) and histopathologic grading was done according to the modified Bloom and Richardson system (19). The paraffin blocks selected for the study contained normal breast tissue and invasive carcinoma. A ductal carcinoma in situ (or lobular neoplasia) component was identified in all cases. Regional lymph node metastases were present in 55 patients. Paraffin blocks with metastatic tissue were available for analysis from 37 patients.

Immunohistochemistry. For initial analysis, immunostaining of mAb NY-BR-13 was adjusted to the Ventana Benchmark (Ventana Medical Systems, Tucson, AZ) system by doing titrations and initial reactivity assessment in a small series of invasive breast cancer samples. Sections were treated with and without a heat-based steamer using various antigen different retrieval solutions (citrate buffer, EDTA buffer, cell conditioning solution). Positive immunostaining was defined as nuclear or cytoplasmic staining. Best staining was achieved at a dilution of 1:400 (concentration, 2 μg/mL), which was used for all analysis.

Eventually, large tissue sections as well as tissue microarrays were analyzed with the Ventana Benchmark automated staining system using Ventana reagents for the entire procedure. For antigen retrieval, slides were heated with cell conditioning solution (CC1) in a standard protocol. Endogenous peroxidase was blocked using the Ventana endogenous peroxidase blocking kit. Primary antibodies were detected using the iVIEW DAB detection kit and the signal enhanced with the amplification kit.

A tumor was considered positive when at least a cluster of carcinoma cells exhibited an assessable reaction product in the cytoplasm or in the nucleus. To obtain an impression of the intratumoral heterogeneity, the extent of immunohistochemical reactivity was graded as follows: focal staining of single cells or small clusters of cells (<10% positive cells); moderate staining (11-50% positive cells); and diffuse staining (51-100% positive cells). This three-tiered scoring system of NY-BR-1 expression reflected the observation that tumors with strong expression were almost always positive in 90% to 100% of the tumor cells. Differences in staining intensity were not considered.

Hormone receptor and Her2 status. In breast cancer tissues, estrogen and progesterone receptor expression was detected using the Ventana Benchmark automated stainer according to the recommendations of the manufacturer and mAbs 6F11 (concentration, 1 μg/mL; Ventana) and 1A6 (concentration, 1 μg/mL; Ventana) for the detection of estrogen receptor and progesterone receptor, respectively. The status of the Her2 proto-oncogene was assessed by fluorescence in situ hybridization using direct fluorescent-labeled DNA probes obtained from Pathvysion (Vysis, Abott AG Diagnostic Division, Baar, Switzerland) and an Olympus microscope. The centromere region of chromosome 17 was visualized by a spectrum green-labeled probe and the Her2 gene with a spectrum-orange labeled probe. The probe mixture was hybridized on paraffin slides containing invasive tumor areas as control on an adjacent H&E-stained section. The interpretation of the fluorescence signals was carried out according to the recommendations of the manufacturer. Presence of five or more Her2 copy number signals was defined as amplification.

Confocal microscopy. For laser confocal microscopy, a breast carcinoma lesion with known NY-BR-1 expression was selected. Ten-micrometer sections were made from paraffin blocks and processed for NY-BR-1 immunolabeling as described above. To avoid problems with autofluorescence, Cy5-conjugated Fab fragments of goat anti-mouse immunoglobulin G (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) were used. Chromatin was labeled with 4′,6-diamidino-2-phenylindole (Sigma, St. Louis, MO). Immunofluorescence was recorded with a Leica confocal laser scanning microscope using the 100× objective (1.4) using Imaris software (Bitplan AG, Zurich, Switzerland). In double fluorescence overlays, effects of pixel shift were excluded. The z-axis resolution of this equipment was at maximum 300 nm per voxel and the x-y settings were between 50 and 250 nm per voxel.

Statistics. For statistical analysis, categorical features were evaluated by a χ² test for linear-by-linear association. A measure of agreement (κ analysis) was done to analyze the congruence of NY-BR-1 expression in primary tumors and metastases. P ≤ 0.05 was considered statistically significant.

NY-BR-1 expression in normal and tumor tissues. In normal tissues, NY-BR-1 staining was solely present in epithelium of the mammary gland whereas all other normal tissues (including adult brain and myocardium from autopsy material) were immunonegative (Table 1). There was a faint unspecific nuclear staining in stromal fibroblasts, smooth muscles, and endothelial cells. A faint cytoplasmic reactivity of plasma cells was also occasionally observed. No staining was detectable in testis. Glandular epithelial cells (pancreas, renal tubules, and lung) exhibited an unspecific cytoplasmic granular reaction corresponding to secretory granula. RNA in situ hybridization showed no NY-BR-1 expression in endothelial cells, smooth muscles, fibroblasts, and secretory glandular cells (data not shown). Therefore, these stains were regarded as unspecific. No NY-BR-1 expression was detectable in various malignant tumors such as carcinomas of the head and neck, ovary, endometrium, urinary bladder, colon, prostate, and the kidney. Other neoplasms, such as melanomas, sarcomas, and lymphomas, were also NY-BR-1 negative. There were no positive ovarian and endometrial carcinomas in the tumor-specific tissue microarrays with 352 endometrial and 268 ovarian carcinomas.

Expression of NY-BR-1 in breast epithelium and in breast cancer. Immunohistochemical staining using the monoclonal NY-BR-1 antibody showed strong cytoplasmic and weak nuclear staining in the epithelial cells of the ductulolobular units of the breast (Fig. 1). The staining of breast carcinoma cells also was cytoplasmic and/or nuclear. Positive immunostaining was confined to the ductal epithelial cells and the carcinoma cells whereas the stromal and inflammatory cells in the tumor stained negative. Confocal microscopy confirmed a nuclear and cytoplasmic NY-BR-1 expression in breast cancer cells as well as in normal terminal ducts (Fig. 2). By confocal microscopy, cytoplasmic labeling of NY-BR-1 was diffusely distributed within the cytoplasm and did not allow any correlation to specific cellular organelles. No positive signal was found in the cell membrane. A nuclear expression was also shown by confocal microscopy. Some nuclei showed a double signal resulting from an overlap of NY-BR-1 and 4′,6-diamidino-2-phenylindole chromatin stain as shown in Fig. 2.

In noninvasive carcinoma lesions, all ductal carcinoma in situ and lobular neoplasia (n = 124) adjacent to invasive breast carcinoma components showed NY-BR-1 immunopositivity (100%). In invasive carcinoma lesions, 74 of 124 (60%) tumors were positive (Table 2); these were 64 of 107 (60%) invasive ductal carcinomas and 10 of 17 (59%) of other histologic subtypes (papillary, tubular, lobular, and mucinous). NY-BR-1 was significantly associated with the differentiation grade, lymph node status, estrogen receptor expression, and Her2-neu gene amplification. Of the 74 positive invasive carcinoma lesions, 20 of 26 grade 1 (77%), 24 of 38 grade 2 (63%), and 30 of 60 grade 3 (50%) carcinoma lesions were positive for NY-BR-1 (P < 0.05). NY-BR-1 expression was also significantly associated with lymph node–negative carcinomas (pN₀ stage category), presence of Her2 amplification, and estrogen receptor expression (P < 0.05). Presence of NY-BR-1 was not associated with tumor size, menopausal status, progesterone receptor status, and histologic subtypes.

NY-BR-1 expression in lymph node metastases. In 37 grade 2 and 3 tumors, regional lymph node metastases were available for analysis (grade 2, n = 12; grade 3, n = 25). In total, 18 of 37 (49%) lymph node metastases were NY-BR-1 immunopositive. In 17 of 18 immunopositive metastases, the corresponding primary tumor was also NY-BR-1 positive. Of 19 NY-BR-1-negative lymph node metastases, the corresponding primary tumor was also immunonegative in 16 cases. This indicated a high concordance of NY-BR-1 expression in the primary tumor and lymph node metastasis and mismatches were more frequently observed in grade 3 than in grade 2 carcinomas. The latter retained their expression in 11 of 12 cases, whereas in one positive primary tumor, the regional lymph node metastasis was negative. In grade 3 carcinomas, 22 of 25 cases had an identical expression pattern (11 cases +/+; 11 cases −/−) whereas two primary positive tumors were NY-BR-1 negative in the lymph node metastasis and one case showed the reverse expression pattern.

The concordance of NY-BR-1 expression in primary tumors and metastasis was 88% for grade 2 cases and 91% for grade 3 tumors. The measure of agreement (κ analysis) was 0.83 for grade 2 carcinomas and 0.76 for grade 3 carcinomas (P < 0.01).

Intratumoral heterogeneity. A homogeneous expression (i.e., presence of NY-BR-1 in >50% of the primary tumor) was seen in 53 of 74 (72%) of invasive breast carcinomas whereas 10 of 74 (14%) tumors exhibited immunoreactivity in 11% to 50% of the tumor cells and the remaining 11 of 74 (15%) positive tumors showed expression in single cells or clusters of neoplastic cells. Interestingly, the degree of NY-BR-1 expression heterogeneity significantly varied between grade 1, grade 2, and grade 3 carcinomas and in lymph node metastases (P < 0.01; Table 3). Expression in >10% of tumor cells (moderate or diffuse expression) was observed in 73% of grade 1, 47% of grade 2, 43% of grade 3 carcinomas, and in only 35% of lymph node metastases.

According to our data, NY-BR-1 protein is frequently expressed in primary breast cancer and metastases but not in any normal tissue, except breast. This finding establishes NY-BR-1 as a potential target for immunotherapy of breast carcinoma patients. Other differentiation antigens such as melanocytic differentiation antigens Melan-A and gp100 as well as cluster of differentiation antigens have already been used in phase I and II clinical trials (20, 21).

A prerequisite to evaluate a potential or actual antigenic target is the knowledge about its presence in a particular tumor. We previously generated several mAbs to NY-BR-1, which can be used on standard archival formalin-fixed paraffin-embedded tissue. Using one of these antibodies, we confirmed previous mRNA analysis, showing NY-BR-1 expression in breast cancer as well as in normal breast epithelium. NY-BR-1 expression was found in ∼60% of all invasive breast carcinomas. This is in concordance with our mRNA expression analysis, showing a higher expression in ∼80% of invasive breast cancer lesions. Although the mRNA expression analysis already suggested a high percentage of positive tumors for this particular antigen, our investigation shows that an immunohistochemical examination renders a different picture for protein expression, which is essential for the evaluation of targets for immunotherapy.

There was no NY-BR-1 protein expression in human testis although NY-BR-1 mRNA was previously found by RT-PCR analysis, indicating that NY-BR-1 is not a member of the family of cancer testis antigens (8). Cancer testis antigens are a group of antigens, which are expressed in a variety of malignant tumors and are restricted in normal adult tissues to germ cells of the testis (2, 22, 23). Cancer testis antigens have been employed as antigenic targets for immunotherapy in numerous clinical trials (1, 24, 25).

We have shown by immunohistochemistry and confocal microscopy that NY-BR-1 antigen expression is present in both cytoplasm and nucleus. A nuclear site for NY-BR-1 is indicated by a bipartite nuclear localization motif (8). The nuclear and cytoplasmic NY-BR-1 expression might reflect a nuclear/cytoplasmic traffic of the protein, similar to the nuclear/cytoplasmic traffic of the von Hippel-Lindau protein. We have recently confirmed such an expression pattern by immunohistochemistry (26). Further studies might be necessary to characterize the subcellular compartment with NY-BR-1 expression.

Although differentiation antigens are used for antigen-targeted and specific antineoplastic immunotherapies, the response rates are difficult to predict. Reasons for therapy failures might be intratumoral heterogeneity and expression heterogeneity between primary tumors and metastases. Unlike other tumor-associated antigens, NY-BR-1 shows a quite homogeneous expression pattern. More than 70% of our positive tumors were immunopositive in >50% of the tumor areas. This is higher than what has been reported for antigens such as MAGE-A1, NY-ESO-1, and CT7 (5, 6, 13, 14, 27). Nevertheless, there are cases with heterogeneous expression pattern, and even in cases with homogeneous presence of NY-BR-1, there are foci of immunonegative cells, indicating some level of heterogeneity. At this point, we can only speculate about the mechanisms leading to the loss of NY-BR-1 expression in a considerable population of breast cancer cells. One cause could be tumor suppressor gene inactivation mechanism with chromosome 10p11-12 loss and consecutive loss of protein expression. Interestingly, NY-BR-1 has been mapped to chromosome 10p11-12 (8) and deletions of chromosome 10p11-12 have been observed in ∼80% of breast cancer by loss of heterozygosity analyses (28). The latter results are in contrast to our recent study of 40 invasive breast carcinomas by comparative genomic hybridization showing no relative DNA copy number losses in this area (29). This may be explained by missing of small interstitial deletions because comparative genomic hybridization enables the detection of losses if the affected region spans >10 Mb.

Importantly, the level of intratumoral heterogeneity of NY-BR-1 was dependent on tumor differentiation. A diffuse expression was seen in normal breast epithelium and in all preinvasive lesions. Interestingly, tumors with lower malignant potential (grade 1 invasive carcinomas, estrogen receptor–positive and lymph node–negative tumors) preferentially expressed NY-BR-1 in a more homogeneous pattern and no NY-BR-1 expression or a high degree of heterogeneity was more often present in poorly differentiated tumors and lymph node metastases. This indicates that NY-BR-1 expression is lost during disease progression. Considering the breast cancer progression model with a stepwise accumulation of genetic changes, our results indicate that loss of NY-BR-1 is a late event in breast cancer progression and may facilitate dissemination of cancer cells via lymphatic or blood circulation. We had therefore expected that the less differentiated tumor cells with a higher tendency for metastases cause low NY-BR-1 expression in lymph node metastases. From this point of view, it was surprising to note that NY-BR-1-positive cells are observed in a considerable number of lymph node metastases, suggesting that NY-BR-1-positive cells are able to disseminate and metastasize. It is unclear, however, if NY-BR-1 loss occurs at the metastatic site and provides a significant selective growth advantage or if positive and negative cells simultaneously metastasize to lymph nodes.

Another crucial point is the congruence of antigen expression in primary tumor and metastatic lesion. Therapeutic decisions in metastatic breast cancer are often based on the presence of biomarkers assessed solely in the primary tumor and metastatic lesions remain unconsidered because they are rarely removed. Consequently, it is often assumed that the biomarker status in the primary reflects the metastatic lesion (30–32). This is shown by Her2, where treatment options are discussed almost always on the basis of the Her2 status in the primary tumors and only few studies have addressed the presence of biomarkers in primary tumors and corresponding axillary lymph node metastases (31). The present data indicate that NY-BR-1 expression in the primary tumors is concordant with its metastatic expression pattern.

In our series, only four cases showed disconcordant NY-BR-1 expression between primary tumors and metastasis. In one case, there was no NY-BR-1 expression in the primary tumor but in the lymph node metastases whereas three cases showed lack of NY-BR-1 expression in the metastases, although the primary tumor had NY-BR-1-positive subpopulations of tumor cells. Clonal selection of individual tumor cells may possibly account for switch of immunoreactivity about NY-BR-1 expression. A selection of clones with different metastatic behavior is potentially more likely in tumors with higher intratumoral NY-BR-1 heterogeneity. According to our results, a higher intratumoral heterogeneity is present in less differentiated tumors (grade 3) and lymph node metastases than in grade 1 and grade 2 carcinomas. This can explain divergent results of NY-BR-1 expression in individual primary tumors and corresponding metastases. On the other hand, fixation artifacts with consecutive loss of epitopes and decreased sensitivity for immunohistochemistry may play a role as well in this phenomenon.

The finding of a concordant NY-BR-1 status in most primary tumors and their lymph node metastases indicates that NY-BR-1 expression analysis of the primary tumor will reliably predict the NY-BR-1 status of the metastasis. The nondisparate expression pattern of primary and metastatic site will be a useful feature for clinical applications, such as NY-BR-1 vaccine–based clinical trials or as a tool for surgical pathology (1, 33, 34).

The relationship of Her2 amplification and NY-BR-1 protein expression requires further analyses. Her2 overexpression and/or amplification is associated with poor clinical outcome. In our study, however, the majority of NY-BR-1-positive carcinomas exhibited Her2 gene amplification as well (35–37). On the other hand, there was no association of NY-BR-1 expression with histologic tumor size or menopausal status. Taken together, these findings suggest that loss of NY-BR-1 might be an indicator of a tumor phenotype with higher metastatic potential.

Another finding with important diagnostic relevance is the breast cancer–specific expression of NY-BR-1. Due to the tissue-specific expression of NY-BR-1, it is of potential value for routine diagnostic purposes in characterizing primary tumors of unknown origin. Our tissue microarray analysis of 352 endometrial and 268 ovarian carcinomas revealed NY-BR-1 expression only in breast cancer. Therefore, NY-BR-1 expression is obviously very rare in gynecologic carcinomas except breast. Currently, there are no breast cancer–specific immunohistochemical markers (34) whereas hormone receptor expression is also seen in endometroid and ovarian carcinomas. Recently, weak mRNA expression was detected in prostate cancer (9). In this study, we have tested more than 10 prostate cancer cases and found no NY-BR-1 expression by immunohistochemistry. Even by mRNA in situ hybridization, all prostate carcinomas were negative, suggesting organ specificity of NY-BR-1. Organ-specific markers are helpful for diagnostic purposes in the differential diagnosis of metastatic diseases if the primary tumor is unknown. As previously indicated, our results indicate that NY-BR-1 is a useful diagnostic marker for cells of mammary gland lineage with potential as a breast cancer marker.

In conclusion, our analysis of normal tissues and a limited number of tumor lesions confirms the tissue-specific expression pattern of NY-BR-1 in normal ductal epithelium and in breast carcinomas and suggests that NY-BR-1 is a potential target for immunotherapy in breast cancer and a useful diagnostic marker in surgical pathology.

We thank S. Bingge, S. Behnke, C. Zuber, and M. Storz for skillful technical assistance; N. Wey for help in preparing the photographical illustrations; and Prof. Dr. Burkhardt Seifert for performing the statistical analysis.