Abstract
Purpose: RON and MET belong to a subfamily of tyrosine kinase receptors. They both can induce invasive growth, including migration, cell dissociation, and matrix invasion. Cross-linking experiments show that RON and MET form a noncovalent complex on the cell surface and cooperate in intracellular signaling. We wanted to examine the clinical significance of RON and MET expression patterns in node-negative breast cancer.
Experimental Design: We studied the protein expressions of RON and MET in five breast cancer cell lines and a homogeneous cohort of 103 T1-2N0M0 breast carcinoma patients, including 52 patients with distant metastases and 51 patients with no evidence of disease after at least a 10-year follow-up.
Results: Both HCC1937 and MDA-MB-231 cancer cell lines co-overexpressed RON and MET. The MCF-7 cell line did not express RON or MET. In multiple logistic regression analysis, RON expression (odds ratio, 2.6; P = 0.05) and MET expression (odds ratio, 4.7; P = 0.009) were independent predictors of distant relapse. RON+/MET+ and RON−/MET+ tumors resulted in a large risk increase for 10-year disease-free survival after adjusting for tumor size, histologic grade, estrogen receptor, bcl-2, HER-2/neu, and p53 status by multivariate Cox analysis (risk ratio, 5.3; P = 0.001 and risk ratio, 3.76; P = 0.005). The 10-year disease-free survival was 79.3% in patients with RON−/MET− tumors, was only 11.8% in patients with RON+/MET+ tumors, and was 43.9% and 55.6% in patients with RON−/MET+ and RON+/MET− tumors.
Conclusions: Co-expression of RON and MET seems to signify an aggressive phenotype in node-negative breast cancer patients.
- RON
- MET
- node-negative breast cancer
- Tumor Progression, Invasion, and Metastasis
- Breast cancer
INTRODUCTION
Compared to breast cancer patients whose lymph nodes show metastasis at the time of surgery, node-negative breast cancer patients have a low risk of disease recurrence. Nevertheless, up to 30% of node-negative breast cancer patients will relapse within 10 years after surgery and eventually die of distant metastasis (1). Traditional factors, such as tumor size, histologic grade, hormone receptor status, age, and menopausal status, have been used to identify the high-risk node-negative patients who might benefit from adjuvant systemic therapy (2). Following these guidelines, >75% of patients with node-negative breast cancer will receive adjuvant systemic therapy, although only about one third will develop distant metastasis (2). Hence, adjuvant systemic therapy is not always necessary and may induce unwanted negative side effects in subgroups not at risk of developing distant metastasis. The success of adjuvant systemic therapy is based on identifying patients at risk of developing distant metastasis. Although there is abundant literature on new prognostic factors and technologies, including gene profiling, the exact application of these technologies remains to be properly defined (3).
Metastasis development is a complex process involving cell dissociation, cell migration, matrix invasion, and transport to remote sites with adhesion (4). Many genes have been implicated in the metastasis of breast cancer (5). Recently, members of the MET proto-oncogene family, a subfamily of receptor tyrosine kinases, have drawn special attention to the association between invasion and metastasis (6, 7). The MET family, including MET and RON receptors, can function as oncogenes, like most tyrosine kinases (8). MET and RON receptors binding with their ligands can induce a complex genetic program that results in invasive growth, including cell dissociation, migration, and extracellular matrix invasion (9–11). MET overexpression is associated with aggressive cases in various tumors (12–14) and with death due to metastatic disease in patients with node-negative breast carcinoma (15).
RON shares a similar structure, biochemical features, and biological properties with MET. RON receptor is a heterodimeric protein similar to MET. It has a single-chain precursor of 170 kDa and a mature heterodimeric form of 185 kDa, which is composed of a 35 kDa extracellular α chain and a 150 kDa transmembrane β chain with tyrosine kinase activity (7, 16). RON and its ligand have diverse biological effects on numerous cell types, including the inhibition of inducible nitric oxide synthase in macrophages, a role in wound healing and liver regeneration, and the proliferation and migration of keratinocytes (17–19). Recent studies have shown RON overexpression in a significant fraction of breast carcinomas (20) and colorectal adenocarcinomas (21), but not in normal breast epithelia or benign lesions (20). In vitro, RON mutations that alter catalytic properties of the RON kinase induce oncogenic and metastatic potential (22). The overexpression of human wild-type RON can cause the formation of lung tumors in transgenic mice (23). Although the RON gene is associated with the pathogenesis of cancer, the link between RON overexpression and its prognostic role in breast cancer is not clear.
Cross-linking experiments have shown that RON and MET form a noncovalent complex on the cell surface and cooperate in intracellular signaling (24). Co-expression of a kinase-inactive RON with mutant MET suppresses the transforming phenotype. A recent study (25) reported neither mutation in RON or MET genes nor co-expression of their receptor ligands in ovarian carcinoma samples or cell lines. RON and MET genes are significantly co-expressed in ovarian carcinoma, and RON and MET receptor ligands trigger ovarian cancer cell motility and invasiveness. This suggests that co-expression of these two related receptors might confer a selective advantage to ovarian carcinoma cells during either tumor onset or progression.
The aims of this study were to search the predictors for distant metastasis to provide reliable information for patient-tailored therapy strategies in node-negative breast cancer patients. We did this study to examine RON and MET expression patterns in five breast cancer cell lines and in specimens of primary breast carcinoma from T1-2N0M0 patients with at least a 10-year follow-up.
MATERIALS AND METHODS
Cell Lines and Culture. Five human breast cancer cell lines (HCC1395, HCC1937, MCF-7, ZR-75-1, and MDA-MB-231) were obtained from the American Type Culture Collection (Manassas, VA). HCC1395, HCC1937, and ZR-75-1 cells were grown in RPMI 1640 (Life Technologies, Grand Island, NY), whereas MCF-7 and MDA-MB231 cells were maintained in DMEM (HyClone, Logan, UT). All media were supplemented with 10% FCS, l-glutamine 2 mmol/L, penicillin (100 U/mL), and streptomycin (100 μg/mL) at 37°C in a 5% CO2-humidified atmosphere. The HCC1395, MCF-7, and ZR-75-1 cells were positive for estrogen receptor. The MCF-7 cell line expressed the wild-type p53 gene; the HCC1937 and MDA-MB-231 cell lines expressed the mutant p53 gene.
Western Blot Analysis of RON and MET. Protein extracts were prepared as previously described (26). Equal amounts of cellular protein (50 μg) were separated by SDS-PAGE and transferred to polyvinylidene difluoride membrane (Amersham Pharmacia Biotech, Piscataway, NJ), and probed with antibodies against RON and MET (Santa Cruz Biotechnology, Santa Cruz, CA). The antibody for RON is a rabbit polyclonal antibody raised against a peptide mapping to the COOH terminus of the human RON, and the anti-MET antibody is a rabbit polyclonal antibody raised against the peptide mapping at the COOH terminus of human c-MET p140. The targeted protein was visualized by using enhanced chemiluminescence reagents (Santa Cruz Biotechnology) and the amount was quantitated by densitometry.
Patient Population. From 1988 to 2002, we recruited, at National Cheng Kung University Hospital, a homogeneous cohort of 103 T1-2N0M0 breast cancer patients who had undergone primary surgery, and we used available paraffin blocks of their primary tumors. According to their divergent clinical behavior, these patients were divided into the metastasis group (unfavorable group) and the disease-free group (favorable group). The metastasis group (52 patients) showed distant relapse within 10 years, and had a median disease-free survival of 1.7 years (range, 0.2-8.8 years) and a median follow-up of 3.8 years (range, 0.8-13.5 years). Metastatic diseases were established by characteristic features revealed by radiography, sonography, bone scans, or pathologic evidence. The disease-free group (51 patients) showed no evidence of disease for at least 10 years and had a median disease-free survival of 12 years (range, 10.1-15.6 years). All patients had received either modified radical mastectomy (98 patients) or breast-conserving surgery with complete axillary node dissection and subsequent breast irradiation (5 patients). The median number of dissected lymph nodes was 18 (range, 12-40).
Adjuvant chemotherapy was given based on the physician's discretion and the patient's preferences, and often on risk factors, such as the size of the primary tumor, age, or menopausal status, and the patient's performance status. The adjuvant treatment for these two groups did not vary significantly. Briefly, patients with larger or higher-grade tumors received chemotherapy with either cyclophosphamide, 5-fluorouracil, and methotrexate, or doxorubicin-containing agents. Tamoxifen was given to most patients with hormone receptor–positive tumors. Since 1995, adjuvant systemic therapy has been based on the St. Gallen consensus (27). The principles of our adjuvant treatment followed the then-current consensus. Histologic slides were reviewed by a breast pathologist (W.Y. Lee) to ensure that they were adequate and representative of the actual tumor. Histologic typing and grading were according to the WHO and modified Scarff-Bloom-Richardson classifications. Patient characteristics are shown in Table 1.
Comparison of clinicopathologic factors and biomarkers between metastasis group and disease-free group
Primary Antibodies. Polyclonal antibody to human RON (diluted 1:100) was used, as was polyclonal antibody to MET, which is specific to human c-Met. Polyclonal anti-human c-erbB-2 oncoprotein (HER-2/neu; DAKO Corporation, Carpinteria, CA; diluted 1:200) was applied. Monoclonal antibodies to p53 (clone PAb1801; Oncogene Research Products, EMD Biosciences, Inc., San Diego, CA; diluted 1:50) and bcl-2 (clone 100; BioGenex Laboratories Inc., San Ramon, CA; diluted 1:200) were also used.
Immunohistochemistry. Tissue sections were obtained from a representative formalin-fixed paraffin-embedded tissue block of each patient's tumor. Immunohistochemical staining by the avidin-biotin-peroxidase complex method was done with an LSAB kit (DAKO). Briefly, 4-μm-thick sections were prepared and endogenous peroxidase activity was blocked with H2O2. Microwave antigen retrieval was done in 10 mmol/L citrate buffer (pH 6.0) at 750 W. The sections were incubated with primary antibodies at 4°C overnight. Antigens were detected using an LSAB kit (DAKO) and visualized using an aminoethyl carbazole substrate kit (AEC kit, Zymed Laboratories, San Francisco, CA). Finally, the sections were lightly counterstained with hematoxylin. Positive and negative controls were included in all runs. For negative controls, we omitted the primary antibodies. Positive controls consisted of breast cancer tissue known to express bcl-2, p53, and HER-2/neu. Normal breast ducts were used as an internal positive control for MET and normal renal tubules as a positive control for RON.
All slides were interpreted by two independent observers blinded to the clinical outcomes. For each case, at least 1,000 tumor cells were analyzed and the percentage of positively stained tumor cells was recorded. Cutoff points with clear precedents in the literature (e.g., for estrogen receptor, bcl-2, and p53) were used (28, 29). For HER-2/neu, a score of 2+ or 3+, as illustrated by the HercepTest kit (DAKO) scoring guidelines, was considered overexpression. To qualify for 2+ and 3+ scoring, >10% of the tumor cells had to show complete membrane staining with moderate and strong intensity. In less well-established markers (e.g., RON and MET), breakpoints that were most consistent with the literature were used (20, 30). Tumors were considered positive for RON (RON+) if >5% of the tumor cells had membranous staining and positive for MET (MET+) if >5% of the tumor cells had intense cytoplasmic or membranous reactivity.
Statistics. Associations between variables and clinical behavior were analyzed with χ2 tests or t tests. All potential predictors for metastasis were entered in multiple logistic regression analyses, both stepwise forward and stepwise backward, until all of the remaining factors were significant at P = 0.05. Disease-free survival was determined as the time from the date of surgery until the date of detection of distant metastasis or the last follow-up. Disease-free survivals were constructed using the Kaplan-Meier method. Log-rank tests were used to test for differences in time to survival between subgroups. The Cox proportional hazards model was used to assess the simultaneous effects of several possible prognostic factors, using univariate analyses followed by multivariate analyses. All statistical tests were done using STATVIEW software (v. 5.0, 1998, Abacus Concepts, Inc., Berkeley, CA) at the two-tailed significance level of P < 0.05.
RESULTS
Western Blot Analysis of RON and MET in Breast Cancer Cell Lines. RON protein was overexpressed in HCC1937, ZR-75-1, and MDA-MB-231 cell lines, and MET was overexpressed in HCC1395, HCC1937, and MDA-MB-231 cell lines (Fig. 1). Both HCC1937 and MDA-MB-231 cancer cell lines, which had p53 mutant and negative estrogen receptor, showed co-expression of RON and MET. HCC1395 and ZR-75-1 cell lines, which had positive estrogen receptor, showed no expression of either receptor. MCF-7 cell line, which had wild-type p53 and positive estrogen receptor, revealed no expression of either receptor.
Expression of RON (top) and MET (bottom) in breast cancer cell lines by Western blotting. RON overexpression in HCC1937, ZR-75-1, and MDA-MB-231 cell lines. MET overexpression in HCC1395, HCC1937, and MDA-MB-231 cell lines.
Comparisons Between Metastasis Group and Disease-Free Group. No significant differences between the two groups of patients were found in age, tumor size, histologic type, histologic grade, hormone status, chemotherapy, hormone therapy, or bcl-2 or HER-2/neu expression (Table 1). Positive immunostainings of p53, RON (Fig. 2A), and MET (Fig. 2B) were present more frequently in the tumors of the metastasis group (P = 0.04, P = 0.007, and P = 0.0002, respectively).
Immunohistochemical stainings for RON and MET on serial tissue sections of an invasive ductal carcinoma from metastasis group. A, carcinoma cells showing strong RON immunostaining in the cell membranes. B, MET expression in the cell membranes. Original magnification, ×200.
Multivariate Logistic Analysis. Multivariate logistic regression showed that RON expression [odds ratio (OR), 2.6; P = 0.05] and MET expression (OR, 4.7; P = 0.009) were independent predictors of distant metastasis. Other potential markers, such as tumor size, histologic grade, estrogen status, p53, bcl-2, and HER-2/neu expression, were not significant predictors.
Cox's Proportional Hazards Regression Analysis. Univariate analyses showed that p53, RON, and MET were significant prognostic factors for disease-free survival and associated with risk ratios (RR) of 1.8 (P = 0.03), 2.1 (P = 0.01), and 3.3 (P = 0.0008), respectively. In the multivariate Cox model, we included tumor size, histologic grade, estrogen receptor status, p53, bcl-2, HER-2/neu, and RON/MET expression patterns (Table 2). Patients with RON+/MET+ and RON−/MET+ tumors showed a large increase in risk for 10-year disease-free survival (RR, 5.3; P = 0.001 and RR, 3.76; P = 0.005, respectively).
Multivariate Cox model for potential prognostic factors of disease-free survival
RON and MET Immunostainings and Patient Survival. The 10-year disease-free survival of all patients was 53.4% (95% confidence interval, 43.8-63.6%). The 10-year disease-free survival as a function of RON and MET expression is shown in Fig. 3. Patients with RON+ tumors had significantly worse 10-year disease-free survival than those with RON− tumors (30.3% versus 58.6%, P = 0.009). Similarly, the 10-year disease-free survival in patients with MET− tumors was 73.7% compared with 35.4% if MET expression was positive (P = 0.0004).
Disease-free survival (DFS) curves in T1-2N0M0 patients with RON+ tumors and RON− tumors (A), and MET+ tumors and MET− tumors (B).
To further assess the relative importance of RON and MET expression, we combined the two biomarkers to evaluate their association with the 10-year disease-free survival (Fig. 4). The highest 10-year disease-free survival was seen in patients with RON−/MET− tumors (79.3%), and patients with RON+/MET+ tumors had the lowest 10-year disease-free survival (11.8%, P = 0.0008). The 10-year disease-free survival in patients with RON−/MET+ tumors and RON+/MET− tumors was 43.9% and 55.6%, respectively.
RON/MET expression pattern and its impact on disease-free survival.
DISCUSSION
In this study, we showed that expression of RON or MET was an independent predictor for distant metastasis, and that co-expression of RON and MET correlated with the lowest 10-year disease-free survival in patients with T1-2N0M0 breast cancer. Taken together, the cooperation of RON and MET signaling events seems to confer an aggressive phenotype on breast cancer cells. Therefore, evaluation of the RON and MET expression status may be of great help in identifying a subgroup of node-negative breast cancer patients for intensive therapy.
A current hypothesis proposes that ligand-induced activation of tyrosine kinase receptors results in autophosphorylation of tyrosine residues, followed by initiation of downstream signaling molecules (31). Cross-linking experiments have also shown that both RON and MET receptors can form homodimers and heterodimers even in the absence of the respective ligand (24). Hence, it is plausible to speculate that overexpressed receptors able to induce dimerization bypass the need for the ligand. Members of the same receptor family can heterodimerize (32), and the heterodimers produce more efficient signaling, as has been described for the ErbB2/ErbB3 dimer (33). In vitro experiments have shown that concomitant activation of RON and MET might cooperate to transduce the same signals, activating the Ras and phosphatidylinositol-3 kinase pathways to mediate invasive growth (34, 35). A recent study showed that MET or RON with the Metp+1loop → Thr point mutation (2B mutation) constitutively phosphorylated signal transducers and activators of transcription 3, leading to the formation of tumors with high metastatic potential (36). Thus, it is possible that overexpression of RON and MET might be due to mutations leading to constitutive activation of RON/RON, MET/MET, or RON/MET dimers.
The conclusion of this study regarding the prognostic importance of RON and/or MET expression in a breast cancer cohort is supported by a number of in vitro experiments. In the present study, the RR of MET expression on 10-year disease-free survival was higher than that of RON expression, which is in agreement with a previous report showing that RON is a less efficient kinase than MET (9). Patients with a RON+/MET+ tumor had significantly poorer outcome than patients with either a RON+/MET− or RON−/MET+ tumor. These results are supported by findings that the heterodimers of RON and MET receptors result in a more efficient signal than that induced by the RON homodimer (9). In addition, kinase-inactive RON behaves as a dominant-negative MET effector, suppressing the transforming phenotype induced by MET receptor (24). Taken together, concomitant activation of RON and MET synergistically increases invasive growth, and inactive RON decreases the signal by the MET receptor. MET can regulate invasive growth and transformation (37), whereas RON promotes invasive growth but not transformation (9). Thus, we suggest that activated RON is involved in tumor progression rather than in the early steps of tumorigenesis.
Further support for our hypothesis comes from the finding that RON and MET were co-expressed in the HCC1937 and MDA-MB-231 cell lines, both of which had p53 mutant and negative estrogen receptor. In contrast, no expression of both receptors was shown in MCF-7 cells that carried wild-type p53 and positive estrogen receptor. Given that p53 mutation and loss of estrogen receptor are poor prognostic indicators (38), co-expression of RON and MET seems to signify an invasive phenotype for breast cancer cells.
In conclusion, our data emphasize the significance of the co-expression of RON and MET in promoting breast cancer metastasis. Co-expression of RON and MET can be used to identify a subgroup of node-negative breast cancer patients at high risk for distant metastasis and who, therefore, may benefit from intensive adjuvant chemotherapy. Furthermore, this observation has clinical significance for patient-treatment options and for the therapeutic target of drug development.
Footnotes
-
Grant support: National Science Council grants NSC91-2314-B-006-130 and NSC92-2320-B-006-072, National Cheng Kung University Hospital grant NCKUH-92-45, and Ministry of Education Program for Promoting Academic Excellence in Universities grant 91-B-FA09-1-4 Taiwan.
-
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.
- Accepted December 23, 2004.
- Received August 30, 2004.
- Revision received December 2, 2004.