Tenascin Expression in Cancer Cells and Stroma of Human Breast Cancer and Its Prognostic Significance1

Sections of formalin-fixed, paraffin-embedded tissues from 210 human breast cancers were immunohistochemically examined using the mAb against human tenascin (TN) RCB1. Immunoreactive TN was detected in the breast cancer stroma in 77 (36.7%) cases, whereas the remaining 133 (63.3%) were negative. Of the 77, 12 (5.7%) cases also showed positive staining in the carcinoma cell cytoplasm. The positive cells were often observed in the margin of the cancer nests at the site adjacent to the stroma. According to the staining pattern of TN, the breast cancer cases were classified into the three groups of cancer cell TN(+)/stromal TN(+), cancer cehl(-)/stromal TN(+), and cancer celh(-)/ stromal TN(-). Analysis of the relationship of these TN patterns with various clinicopathological characteristics of the tumors and the patient outcome revealed that, in cornparison to the cancer cell(-)/strornal TN(-) group, the cancer cell TN(+)/stromal TN(+) group exhibited increased frequency of lymph node metastasis and exceptionally poor outcome, and the cancer cehl(-)/stromal TN(+) group also showed more frequent metastasis and poorer outcome. Most of the cancer cell TN(+)/stromal TN(+) cases were c-erbB-2 positive and estrogen receptor negative. Furthermore, in situ hybridization of freshhy obtained breast cancer tissues demonstrated that both cancer cells and stromal cells express TN mRNA. These results indicate that the TN in breast cancer is produced by cancer epithehial cells as well as by stromal mesenchymal cells, and that cancer cell TN might be involved in cancer spreading, resulting in unfavorable patient prognosis. INTRODUCTION TN3 is an extracelluhar matrix glycoprotein with a unique six-armed macromolecular structure, which is known to bc an essential factor for modulation of reciprocal interactions between the epithehium and mesenchyme during embryogenesis (1-3). A number of studies have demonstrated prominent TN Received 2/28/95; revised 4/14/95; accepted 5/1/95. I This research was supported in part by Grants-In-Aid from the Mmistry of Education, Science and Culture, Japan. 2 To whom requests for reprints should be addressed. 3 The abbreviations used are: TN, tenascin; ER, estrogen receptor; TBS, Tris-buffered saline: DIG, digoxygenin. expression in human cancers including those of brain (4), colon (5, 6), liver (7, 8), lung (9, 10), uterus (1 1), skin (12), prostate (13), and mammary gland (14-17). Because it was initially proposed as a stromal marker for epithelial malignancy (14), TN expression in cancers has usually been observed in the stroma. Examining human breast cancers by immunohistochemistry, investigators have found that TN is consistently present in the stroma of malignant tumors (14-16). Our previous examination also indicated intense staining in the connective tissue of invasive ductal carcinomas (17). Accordingly, it has been generally accepted that TN is produced by the mesenchyme and has an active function in cancer development, probably by promoting cancer cell proliferation and invasion. Shoji et a!. (18) proposed TN to be a marker useful in predicting the survival of breast cancer patients. They investigated immunoreactive TN in 82 patients with primary invasive breast carcinomas, and found a significantly superior outcome during the 5-year period after surgery in the TN-positive patients compared with the negative patients. Immunohistochemistry of primary colon carcinomas demonstrated no lymphogenous metastasis in patients in whom the cancer stroma showed strongly positive staining for TN (19). Taking these results into consideration, it is conceivable that TN may prevent rather than enhance the cancer cell outgrowth, probably by creating a barricade surrounding the cancer nests and inhibiting the movement of these cells. In fact, TN has been proposed as a boundary molecule in the somatosensory cortical barrel field during the development of mouse cerebral cortex (20). Thus, there are two different lines of speculation as to the role of TN in cancer development. Ligthner et a!. (21) have recently demonstrated that normal mammary epithelial cells express TN in culture and incorporate the protein into the underlying matrix. They have also examined human breast tissues by in situ hybridization and found that both normal and malignant mammary epithehial cells can express TN mRNA (21). Our previous findings have shown that A431 epidermal cancer cells which express no TN in culture can produce TN, with accumulation in the surrounding mesenchyme, when injected into the nude mouse subcutis (22). Thus, the TN in the breast cancer stroma is considered to be originated from both the epithelium and mesenchyme. We have proposed that TN in the tissue is heterogeneous in its structure and function (23). In breast cancer, the structure and function of epithelial TN are possibly different from those of the mesenchymal TN. Therefore, it is important to determine by which cells the TN in the cancer stroma was produced. In this study, we used archival formalin-fixed and paraffinembedded surgical materials of human breast cancers. In immunohistochemical studies of cancers using anti-TN antibody, we have found positive staining in the cytoplasm of cancer cells. Therefore, this study was designed to investigate the presence of both the cancer cell TN and stromal TN in breast cancer, and the implications regarding the relationship between the TN staining Research. on April 13, 2017. © 1995 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 1036 Tenascin Expression in Human Breast Cancers pattern and clinicopathological characteristics. TN staining was detected in cancer cells in the patients whose outcome was poor. The results were also compared with the c-erbB-2 and ER status. Furthermore, the cellular source of TN was identified in freshly obtained cancer tissues by in situ hybridization with human TN mRNA. MATERIALS AND METHODS Tissues. Primary breast cancers and axillary lymph nodes from 210 female patients who had undergone surgical resection at Matsusaka Chuo General Hospital between 1979 and 1987 were used in this study. The average age at the time of diagnosis was 54.4 (range, 28-89) years. The tumor diameter varied from 0.6 to 12 cm (mean diameter, 2.8 cm). Five years after the initial operation, 167 patients were still alive, 32 had died from relapsed breast cancer, and 1 1 had died of other causes. The tumors were classified according to the WHO histological typing for breast cancer (24). The histological malignancy was categorized into three grades (I-Ill), based on the grades of architectural atypia and nuclear atypia, along with the number of mitotic figures as described initially by Bloom and Richardson (25). All cases were classified into the clinical TNM stages I, II, lIla and b, and IV according to the size of the tumor, degree of lymph node metastasis, and distant metastasis status (26). The ER bevels were determined at a commercial laboratory (Otsuka Assay Company, Tokushima, Japan). Survival times were measured from the date of diagnosis using the multiple regression model developed by Cox (27) for censored survival data. The curves of the probability of survival were obtained using the product limit method of Kaplan and Meier (28). Differences among the groups for other prognostic parameters were analyzed using Student’s t test with the Cochran-Cox method and x2 test. Immunohistochemistry. The surgical specimens were fixed routinely in 10% phosphate-buffered formalin, embedded in paraffin, and cut at 4-p.m thickness. Sections were examined by immunostaining for TN, c-erbB-2, and vimentin. After deparaffinization with xylene and hydration with downgraded ethanol, the sections were incubated in 0.3% H2O2 in methanol for 15 mm to block endogenous peroxidase activity. For the staining of TN, the sections were treated in 10 mt i TBS solution supplemented with 1% normal rabbit serum for 10 mm to block nonspecific binding of rat immunogbobulins. Development and characterization of a rat mAb, RCB1, against human TN punfied from conditioned medium of umbilical fibnobbasts were previously described (29). Antibody RCB1 is commercially available from Cosmo Bio Japan, Ltd., as a clone name of 8C9. Sections were then incubated with RCB1 at a concentration of 1 p.g/ml for 2 h, washed three times with TBS, incubated with biotinylated rabbit antinat IgG for 30 mm, washed three times with TBS, and incubated with a complex of avidin and biotinylated horseradish peroxidase (Vector Laboratories, Bunlingame, CA) for 30 mm. After the sections were washed with TBS, the color reaction was developed with a freshly prepared solution of 0. 1% diaminobenzidine tetrahydrochbonide (Sigma, St. Louis, MO)/0.02% H2O2 in 50 mt t Tris-HC1 (jH 7.6). They were washed with TBS and counterstained with hematoxybin. All Table 1 Expression of immunoreactive TN in human br east cancers Histological No. of classification cases TN