Reduced γ-Catenin Expression and Poor Survival in Oral Squamous Cell Carcinoma

Objective  To investigate whether reduced expression of α-, β-, or γ-catenin predicts poor survival in oral squamous cell carcinoma (OSCC).

Design  Immunohistochemical analyses of a retrospective cohort.

Setting  University-affiliated hospital.

Patients  One hundred twenty-four patients with OSCC.

Main Outcome Measure  The prognostic value of γ-catenin expression on disease-specific survival in different T and N category groups in patients with OSCC.

Results  Reduced expression of γ-catenin correlated with poor tumor differentiation of OSCC (P = .04). Patients with reduced γ-catenin expression in the primary tumor had significantly more frequent lymph node metastasis than did patients with normal γ-catenin expression (P = . 03). Reduced expression of γ-catenin (004) but not of α-catenin (P = .25) or β-catenin (P = .48) correlated with poor clinical outcome. Reduced γ-catenin expression predicted poor disease-specific survival also in the 92 patients with T1 or T2 tumors (P = . 02). In multivariate analysis, advanced T category (P = . 04), neck lymph node metastases (P = . 01), and reduced γ-catenin expression (P = . 05) were independently related to poor survival.

Conclusions  Reduced expression of γ-catenin was associated with poor differentiation of OSCC, with neck lymph node metastases, and, more importantly, with poor disease-specific survival. Loss of γ-catenin expression seems to contribute to metastatic properties of OSCC. Evaluation of the expression pattern of γ-catenin may be useful for predicting outcome in patients with OSCC.

Squamous cell carcinoma (SCC) represents more than 90% of all cancers in the oral cavity and pharynx. The National Cancer Institute estimates that approximately 35 000 Americans were diagnosed as having oral or pharyngeal SCC in 2008, and more than 7500 died of it.¹ Oral SCC (OSCC) is the most common subgroup of all head and neck SSCs, representing approximately 75%. In general, the prognosis of OSCC declines with higher stage and grade of cancer.² However, in certain patients, prognosis can be poor even in stage I primary tumors. Independent tumor-related prognostic markers for poor prognosis in early-stage OSCC are scarce.³ New prognostic markers could help physicians choose appropriate treatment for patients at higher risk for local recurrence and metastasis.

Several factors that regulate cell growth, adhesion, and invasion play a significant role in cancer development. Elevated expression of stromal versican, which is an extracellular matrix proteoglycan, has been shown to correlate with an increased risk of disease recurrence and shortened survival in OSCC.⁴ Reduced expression of hyaluronan, a matrix component suggested to promote cancer cell growth and migration, has been found to correlate with poor survival in OSCC.⁵ Moreover, decreased E-cadherin expression has been noted to be associated with decreased survival in head and neck SCC.⁶

Cadherins and catenins are important factors in intercellular adhesion. Cadherins establish molecular links between adjacent cells. They form zipperlike structures at adherens junctions between different cells. Cadherins are linked to cytoskeleton through catenins. These cytoplasmic molecules are essential for normal cadherin function and formation of adherens junctions. Cadherins are linked to β-catenin, whereas α-catenin is associated with actin microfilaments.⁷ β- and γ-catenin play a crucial role in the cadherin-based adhesion system. β- or γ-catenin is required for stable expression and cell surface localization of E-cadherin and cell adhesion.⁸ γ-Catenin is a structural and regulatory constituent of cell-cell junctions. Whereas β-catenin normally binds exclusively classic cadherins and is restricted to adherens junctions, γ-catenin is associated with classic and desmosomal cadherins. It can be found in desmosomes and in all other types of adherens junctions. γ-Catenin is essential for maintaining and regulating adhesive strength in, for example, keratinocytes by contributing to desmosome assembly and structure.⁹ In addition, it has been demonstrated that γ-catenin suppresses the tumorigenicity of cells in humans.¹⁰

During the metastatic process, loss of tumor cell adhesion is a crucial step because it has been observed that tumor cells migrate from tumor mass as single cells.¹¹ Reduced expression of α-, β-, and γ-catenins has been found to predict an unfavorable prognosis in various malignant neoplasms.^12-14 Reduced γ-catenin expression has also been associated with poor survival in oropharyngeal SCC and OSCC^15-17 and in SCC of the floor of the mouth¹⁸ (Table 1).

The aim of this study is to investigate whether reduced expression of catenins predicts poor survival in OSCC. We also investigated the relation of catenin expression to clinicopathologic variables. For further evaluation, patients were divided into 2 groups according to neck nodal status. Thereafter, we also analyzed the predictive value of catenin expression on patient outcome.

Clinicopathologic data were obtained from the previous study²³ of the same original patient group of 239 patients with OSCC treated at Kuopio University Hospital and Central Hospital of Central Finland, Jyväskylä, between January 1, 1979, and December 31, 1998. This study included a retrospective cohort of 124 patients with OSCC and sufficient tumor material available for immunohistochemical analyses. Median patient follow-up was 53 months (range, 0.5-272 months).

Clinical data from 239 patients were reviewed retrospectively from patient files by a senior otorhinolaryngologist (A.K.). The original sections of the tumors were reviewed by 2 experienced histopathologists (R.P. and K.H.), and the histopathologic grade of the tumors was confirmed according to the World Health Organization classification.^23,24 The reviewers were masked to the clinical data. After primary evaluation, 124 patients were included in the immunohistochemical analyses.

Staging of the tumors was performed according to the International Union Against Cancer staging system²⁵ based on hospital records of tumor localization, size, and possible invasion together with neck nodal status. Performance status was coded using the Karnofsky scale²⁶ at the time of diagnosis.

Deparaffinized and rehydrated sections were heated in a microwave oven (800 W) for 6 × 5 minutes (α- and β-catenin) or 3 × 5 minutes (γ-catenin) in a 0.01M citrate buffer (pH 6.0) and were washed twice in a phosphate-buffered saline (PBS) solution for 5 minutes. Endogenous peroxidase activity was blocked by 5% hydrogen peroxidase for 5 minutes, and, thereafter, the sections were rinsed with distilled water for 3 × 5 minutes and twice with PBS for 5 minutes. Normal horse serum (1.5%) (Vectastain Elite ABC Kit; Vector Laboratories, Burlingame, California) in PBS for 25 minutes at room temperature was used to block unspecific binding. Thereafter, the sections were incubated for 20 hours at 4°C with mouse monoclonal antibodies against α-, β-, and γ-catenins (Transduction Laboratories, Lexington, Kentucky) diluted 1:200, 1:1000, and 1:200, respectively, in PBS and 1.5% normal horse serum.

The sections were rinsed again and were incubated with avidin-biotin peroxidase and biotinylated secondary antibody (Vectastain Elite ABC Kit). Peroxidase activity was detected by using 3,3′-diaminobenzidine tetrahydrochloride (Sigma-Aldrich Corp, St Louis, Missouri). The sections were counterstained with Mayer hematoxylin, washed, dehydrated, cleared, and mounted using DePex (BDH, Poole, England). Normal epithelium and glands beside the tumor tissue served as internal controls. In addition, positive controls from the gut were used in all of the series. In negative controls, primary antibody was omitted. Negative controls showed no positivity.

Staining signals associated with cell membranes were considered positive. The fraction of positive tumor cells in the samples for α-, β-, and γ-catenins was primarily analyzed using a continuous scale by 2 of us (E.L. and V.-M.K.). For statistical analysis, staining of α- and β-catenin was divided into 2 subgroups using a median percentage as a cutoff point: tumors that showed positive staining in 90% or more of the tumor cells were considered to be normal, whereas those with less than 90% staining were considered to be reduced. The same cutoff point has been used successfully in previous studies as well.^27,28 Staining of γ-catenin was also divided into 2 subgroups using the median percentage as a cutoff value at which staining of less than 80% was considered to be reduced and of 80% or more to be normal.

The association between catenin staining and clinicopathologic data was analyzed using the χ² test. Disease-specific survival (DSS) was defined as the interval from diagnosis to death due to OSCC or the end of follow-up. The nonparametric χ² test was used to test the representativeness of the present cohort (n = 124) to the original patient group (n = 239).²³ With continuous variables, the 1-sample t test was used. The nonparametric χ² test was used to compare the proportions of normal and reduced catenin expression in well-differentiated and moderately or poorly differentiated tumors and to evaluate the correlation between histologic grade and catenin expression and mutual expression of different catenins. To evaluate the prognostic value of reduced expression of catenins in the whole cohort (n = 124) and γ-catenin expression in T1 and T2 tumors (n = 92), Kaplan-Meier survival analyses were used. Multivariate analyses were performed using Cox regression. To evaluate the prognostic value of primary OSCC γ-catenin expression in patients with neck metastasis, the patient group was stratified into 2 groups according to nodal status (N category). P ≤ .05 was considered statistically significant. A software program (SPSS for Windows Release 14.0; SPSS Inc, Chicago, Illinois) was used for statistical analysis.

This study was approved by the research ethics committee of Kuopio University and Kuopio University Hospital. Permission for obtaining data from the Finnish Cancer Registry and from hospital records was granted by the Finnish Ministry of Social Affairs and Health.

The present cohort of 124 patients represented well the original group of 239 patients²³ matched by age, sex, Karnofsky performance status, tumor differentiation, and stage (P > .45 for all). Demographic and treatment data are described in Table 2. The median age of patients at the time of diagnosis was 63.8 years (range, 10-94 years). One hundred ten patients underwent surgery, and 50 of them received postoperative adjuvant radiotherapy. Seven patients had definitive radiotherapy only, and 1 received chemotherapy only. Six patients received no cancer-specific treatment for various reasons.

In normal squamous cell epithelium and glands serving as positive controls, α-, β-, and γ-catenins were mainly expressed homogenously on cell membranes. In tumor cells, catenins were also detected on cell membranes. Figure 1 illustrates the different staining patterns of tumor samples stained with mouse monoclonal antibody against γ-catenin. Reduced expressions of catenins correlated with each other: in tumor cells in which γ-catenin expression was reduced more often, α- and β-catenin expression also was likely to be reduced (P < .001). α-Catenin (P = . 75) and β-catenin (P = .95) expression was not associated with tumor differentiation. However, moderately or poorly differentiated tumors more often showed reduced staining for γ-catenin (P = . 04) (Table 3). There were 92 patients with T1 (n = 43) or T2 (n = 49) primary OSCC. No correlation was noticed between T category and expression of α-, β-, or γ-catenin (P > .05 for all) (for γ-catenin, see also Table 3).

Table 4 summarizes the survival analysis data. Briefly, in univariate survival analysis of the whole patient group (n = 124), reduced expression of γ-catenin (P = . 004) but not of α-catenin (P = . 25) or β-catenin (P = . 48) correlated with poor DSS. In a subgroup of patients (n = 43) having T1 tumors, no difference was found in the outcome related to the expression of γ-catenin (P = . 13). The combined analysis of patients with T1 and T2 tumors showed that 16 of 92 patients (17%) died of OSCC during follow-up if γ-catenin expression was reduced in the primary tumor, whereas in the normal-expression group, only 7 (8%) died of OSCC (P = . 02). In patients with T3 and T4 tumors (n = 32), the expression of γ-catenin did not affect survival (P = .27). Also, expression of α- and β-catenin was not associated with survival in patients with T1 or T2 tumors (P = . 08 for α-catenin and P = . 48 for β-catenin).

Twenty-nine of 124 patients with OSCC (23%) had neck lymph node metastasis at the time of diagnosis. Patients with reduced γ-catenin expression in the primary tumor had significantly more frequent lymph node metastasis (20 of 29 [69%]) than did patients with normal γ-catenin expression (9 of 29 [31%]) (Pearson χ² test, P = . 03) (Table 3).

To determine whether γ-catenin expression in the primary tumor can be used to predict clinical outcome in node-negative or node-positive patients, we compared DSS in different groups. Patients were stratified into 4 groups according to reduced or normal expression of γ-catenin and nodal status (any kind of nodal metastasis [N+] or no nodal metastasis [N0]). In the N0 and N+ groups, patients had shorter DSS if γ-catenin expression was reduced in the primary tumor (P < .001) (Figure 2).

In multivariate analysis performed using T category, nodal status, age, Karnofsky performance status, histologic grade, and sex, together with α-, β-, and γ-catenin expression, the strongest predictors of poor survival were advanced T category (P = . 04), positive nodal status (P = . 01), and reduced γ-catenin expression (P = . \05). The other factors did not predict survival (P > .10 for all).

Oral SCC is the most frequent head and neck cancer, with high mortality and morbidity rates. Several prognostic markers have been under evaluation, but only a few studies have investigated the prognostic value of catenins in OSCC (Table 1). The present study was performed to evaluate the prognostic significance of catenins in OSCC because their reduced expression has been shown to be related to poor prognosis in other carcinomas.^13,28,29 In this study, we showed that reduced γ-catenin expression in the primary tumor correlated with poor DSS. Moreover, reduced γ-catenin expression retained its prognostic value in multivariate analysis, predicting poor outcome, together with traditional prognostic factors, such as T and N category.

Catenins are intercellular adhesion molecules located mainly on cell surfaces. In addition to maintaining cell-cell junctions, catenins can function as tumor suppressor–like γ-catenin and have important roles in cell signaling, similar to β-catenin.¹⁰ Reduced catenin expression has been introduced in various malignant neoplasms.^12-16,30 It has been supposed that immunoexpression of the cadherin-catenin complex could be used as a marker to predict neoplastic behavior. Indeed, Ueda et al¹⁷ showed that aberrant expression of α- and γ-catenin was more common in poorly differentiated OSCC. The same has been noted with β-catenin.²² In this study, reduced γ-catenin expression but not reduced α- or β-catenin expression correlated with poor tumor differentiation.

Discovery of prognostic markers is necessary to determine patients with T1N0 or T2N0 tumors at risk for tumor recurrence and neck metastasis. Therefore, we wanted to evaluate catenin expression and its possible relation to survival in T1 and T2 tumors. The number of events in the T1 group alone was low; perhaps, therefore, we could not find any relationship between reduced γ-catenin expression and patient outcome. However, when patients with T1 or T2 tumors were combined and analyzed as a group, reduced γ-catenin expression correlated with poor survival. This finding could be clinically important because it is possible that γ-catenin together with other markers in a predictive model may, in the future, be useful for selecting treatment modalities. The role of γ-catenin as a prognostic marker in T1 tumors has been investigated by Clairotte et al.³¹ They demonstrated that reduced expression of γ-catenin but not of α- or β-catenin functioned as an independent predictor of progression-free survival in patients with stage T1 bladder urothelial tumors, supporting the findings in the present study. They determined that characterization of T1 tumors that will progress could lead to identification of patients who might benefit from surgery to avoid muscle invasion and metastasis.

In the present study, the possible importance of γ-catenin was further confirmed by analyzing patients with N0 separately for γ-catenin expression. Only 15% of patients with N0 and normal expression of γ-catenin died of OSCC during follow-up. However, the mortality of patients with N0 and reduced expression of γ-catenin was as high as 39%. This finding is in line with that of a study¹⁸ demonstrating that in T1N0 and T2N0 SCC of the floor of the mouth, loss of γ-catenin expression correlated with poor disease-free survival. In the present study, lack of γ-catenin expression was related to poor outcome also in node-positive patients. This finding suggests that γ-catenin might have prognostic value in patients with node-negative and node-positive OSCC. The prognostic value of γ-catenin but not β-catenin expression has also been demonstrated in, for example, medulloblastomas.³⁰ In the study by Misaki et al,³⁰ normal γ-catenin expression correlated with a low number of metastases and a good prognosis.

The proportion of α-, β-, or γ-catenin expression did not depend on tumor size. This could suggest that, for example, γ-catenin expression that was demonstrated to be an independent prognostic marker is not only related to tumor growth itself in OSCC but also more likely to its metastatic properties. Defective cell-cell adhesion enables cells to detach from the tumor mass and acquire the capacity to migrate and form metastases in regional lymph nodes. The effect of aberrant expression of different catenins that contribute metastatic properties has been demonstrated on OSCC in various other studies as well (Table 1). Why only γ-catenin expression and not α- or β-catenin expression correlated with survival and neck metastases in this study, and, for example, bladder cancer,³¹ needs to be evaluated.

The TNM staging system has been used to evaluate the prognosis of patients with OSCC. It is well known that patients with advanced (stage III and IV) OSCC have a poor clinical outcome.³ However, clinical prognosis could be poor even in early-stage tumors. In this study, reduced expression of γ-catenin was shown to correlate with positive nodal status and, more important, with poor DSS. Reduced expression of γ-catenin was also related to poor differentiation of the OSCC. These findings suggest that γ-catenin contributes to the metastatic process. Also in T1 and T2 tumors, reduced γ-catenin expression was shown to correlate with poor clinical outcome. Thus, γ-catenin might have prognostic value in patients with OSCC. Additional studies are warranted.

Correspondence: Mervi Närkiö-Mäkelä, MD, PhD, Kuopio University Hospital and University of Kuopio, PO Box 1777, 70211 Kuopio, Finland (mervi.narkio-makela@kuh.fi).

Submitted for Publication: February 3, 2009; final revision received April 26, 2009; accepted April 28, 2009.

Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Pukkila, Virtaniemi, Pirinen, Johansson, Hämäläinen, and Kosma. Acquisition of data: Närkiö-Mäkelä, Lagerstedt, Kosunen, and Lappalainen. Analysis and interpretation of data: Närkiö-Mäkelä, Pukkila, Lagerstedt, Johansson, and Kosma. Drafting of the manuscript: Närkiö-Mäkelä, Pukkila, Kosunen, and Kosma. Critical revision of the manuscript for important intellectual content: Närkiö-Mäkelä, Pukkila, Lagerstedt, Virtaniemi, Pirinen, Johansson, Lappalainen, Hämäläinen, and Kosma. Statistical analysis: Närkiö-Mäkelä and Pukkila. Obtained funding: Närkiö-Mäkelä, Pukkila, Johansson, and Kosma. Administrative, technical, and material support: Närkiö-Mäkelä, Pukkila, Lagerstedt, Virtaniemi, Johansson, Kosunen, Lappalainen, Hämäläinen, and Kosma. Study supervision: Pirinen, Johansson, and Kosma.

Financial Disclosure: None reported.

Funding/Support: This work was supported by grant 229.26.263.08 from the University of Kuopio, Kuopio University Hospital, and by grants from Finnish Cancer Organizations.

Previous Presentation: This study was presented at the Seventh International Conference on Head and Neck Cancer; July 20, 2008; San Francisco, California.

Additional Contributions: Aija Parkkinen provided technical assistance.