Materials and methods

PATIENTS AND FOLLOW UP

The study population comprised 261 patients with resected NSCLC treated in eastern Finland from 1978 to 1996 (table 1). There were 239 men and 22 women in the study cohort. The mean age of the patients was 63 years. Most patients (222) underwent either radical lobectomy (50% of cases) or pneumonectomy (35% of cases). Palliative resection was performed in 10% of the patients and explorative thoracotomy only in 5% of the patients. None of the patients received radiotherapy or chemotherapy before surgery. Postoperative radiotherapy was given to 61 patients and chemotherapy to 11. TNM classification and stage of the disease²⁰ were determined by reviewing radiological examinations, findings during surgery, and the histopathological evaluation of the tumour. Follow up was performed regularly by a senior physician according to routine protocol. All clinical data were collected retrospectively by reviewing the patients' files.

HISTOPATHOLOGICAL EVALUATION

The histological diagnosis and grade of differentiation were re-evaluated (V-MK and SH) using haematoxylin and eosin stained sections according to the WHO classification.²¹ There were 163 squamous cell carcinomas, 68 adenocarcinomas, and 30 anaplastic/large cell carcinomas in our study. The most representative slide was used in the immunohistochemical analyses.

IMMUNOHISTOCHEMISTRY

Catenins

Briefly, the paraffin wax embedded samples were cut into 5 μm thick sections and dewaxed in graded alcohols. For the antigen retrieval, the sections were heated in a microwave oven for three periods of five minutes in Tris/HCl buffer, pH 9.7 (α-catenin), or citrate buffer, pH 6.0 (β-catenin and γ-catenin). The slides were washed with phosphate buffered saline (PBS) for two times five minutes and endogeneous peroxidase activity was blocked by 5% H₂O₂ for five minutes. The sections were washed twice with PBS and treated with 1.5% normal serum (ABC Elite kit; Vector, Burlingame, California, USA). The primary antibody (Transduction Laboratories, Lexington, Kentucky, USA) was used at a dilution of 1/200 (in 1% bovine serum albumin in PBS (α-catenin), 1/1000 (β-catenin), or 1/100 (γ-catenin). Slides were incubated overnight at 4°C, washed with PBS, and treated with the biotinylated secondary antibody and avidin–biotin peroxidase reagent (Vector). Diaminobenzidine was used as a chromogen. The slides were washed with PBS, lightly counterstained with haematoxylin, dehydrated, cleared, and mounted in Depex (PDH, Poole, Dorset, UK). A sample of normal colonic mucosa, known to be positive for all catenins, was used as a positive control. In the negative controls, the same samples were stained without adding the primary antibody.

Ki-67

To determine the proliferative activity of the tumour, anti-Ki-67 (MIB1; Immunotech, Marseille, France) was used as a primary antibody, diluted 1/200. The indirect ABC technique was used as described above.

EVALUATION OF STAINING

All samples were evaluated (RP, PH, and V-MK) unaware of the clinical data. In the case of disagreement (less than 10% of the cases) the slides were reviewed again and a consensus was reached. The staining reaction localised on the cell membranes was graded into three groups according to the percentage of the positively stained carcinoma cells as follows: < 10%, 10–90%, and ≥ 90% of the tumour cells showing a positive staining pattern. Tumours showing 90% or more positively stained cells were categorised as normally stained, others were classified as reduced.²² Nuclear staining of β-catenin and γ-catenin was recorded as positive or negative. In positive samples nuclear staining was seen in more than 20% of cancer cells.

For the Ki-67 staining, the percentage of the stained nuclei (proliferation index; PI) was determined by a continuous scale using the CAS 200 (Becton Dickinson, Elmhurst, Illinois, USA) image analyser as described previously.²³ The data were available for analysis from datasheets of our previous study.²⁴

STATISTICAL ANALYSIS

SPSS version 8.0 was used in the analyses. The Mantel-Haenzel or Pearson χ² test, when appropriate, was used to clarify the relation between staining results and other variables studied. The cell proliferation associated with catenins was analysed with Kruskall-Wallis and Mann-Whitney U tests. Overall (OS) and disease free survival (DFS) were estimated by means of the Kaplan-Meier test and Cox's multivariate hazards model. In DFS analyses only the patients who were treated with radical resection of tumour were included. Deaths by any other cause than lung cancer were considered as censored events—corrected survival rates were used.

ETHICS

Our study plan was approved by the ethical committee of the Kuopio University Hospital and the Finnish Ministry of Social Affairs and Health.

Results

Two hundred and thirty one samples were immunostained successfully for α-catenin, 229 for β-catenin, and 231 for γ-catenin. Some samples were discarded because of technical problems, such as partial folding of the tissue section or lack of invasive carcinoma in the section. The distribution of clinicopathological characteristics (stage, sex, grade, Karnofsky scale, histological type of tumour) among the patients with tissue samples for immunohistochemistry did not differ from the original patient group (n = 261). Benign bronchial glands showed normal staining patterns for α-catenin, β-catenin, and γ-catenin. In addition, normal staining reactions for all catenins were found in the bronchial pseudostratified columnar epithelium and basal cells.

STAINING OF CATENINS AND CLINICOPATHOLOGICAL PARAMETERS

Table 2 summarises the results. Normal staining for α-catenin, β-catenin, and γ-catenin was found significantly more often in adenocarcinomas (62%, 66%, and 58% of cases, respectively) than in squamous cell or anaplastic/large cell carcinomas (p < 0.0001 for all; fig 1A and B; table 2). Poorly differentiated tumours more often showed reduced staining for all catenins (p = 0.005, p = 0.007, and p = 0.005 for α-catenin, β-catenin, and γ-catenin, respectively; fig 1C). Nuclear staining of β-catenin and γ-catenin was found in 16 (7%) and 29 (13%) cases, respectively, and was seen more often in squamous cell carcinomas (11 cases with β-catenin and 24 cases with γ-catenin) than in adenocarcinomas or anaplastic/large cell carcinomas. Nuclear staining correlated inversely with the membranous staining pattern (p = 0.001 and p = 0.0003, respectively; fig 1D).

CELL PROLIFERATION

Tumours with reduced β-catenin or γ-catenin staining had higher PI values (p = 0.0001 for β-catenin; p = 0.006 for γ-catenin) than did those with a normal staining pattern. Furthermore, nuclear positivity for β-catenin and γ-catenin was associated with a higher PI (p = 0.01 and p = 0.002, respectively). α-Catenin staining was not related to cell proliferation. Moreover, PI was associated with tumour type (p = 0.001) and grade (p = 0.004).

SURVIVAL ANALYSIS

At the end of follow up 136 (52.1%) patients had died of lung cancer and 56 (21.5%) were alive. A recurrent tumour was noted in 124 cases (47.5%) during follow up. In univariate survival analysis, more advanced stage (II–IV) was a sign of poor OS (p = 0.0006). Patients with adenocarcinoma or large cell/anaplastic carcinoma had poorer OS than did patients with squamous cell carcinoma (p = 0.0003). Similarly, the stage (p = 0.006) and histological type of the tumour (p = 0.0001) significantly predicted DFS (fig 2). The expression of catenins did not predict survival. In Cox's multivariate analysis including stage and histological type of tumour, both retained their independent prognostic value to predict OS and DFS (p < 0.0005 for both).

Discussion

Alterations in E-cadherin–catenin complexes act as an important link in the progression of different cancers.¹³ Lung cancer is the most frequent cancer worldwide, and despite some downward trends in mortality prognosis remains poor.¹⁵ Few studies have investigated the prognostic value of catenins in lung cancer.^{18, 19} In our present study, we investigated the distribution of different catenins in a large population of patients with resected NSCLC. Altered expression of catenins was found to correlate with the tumour type, with anaplastic/large cell and squamous cell carcinomas more often showing reduced staining of all catenins compared with adenocarcinomas. In addition, reduced staining of catenins was often seen in poorly differentiated tumours. Furthermore, nuclear staining of β-catenin and γ-catenin was related to increased proliferative activity of the tumour.

The distribution of catenins seems to be related to tumour type. Previous results indicate that lung adenocarcinomas often show normal staining for catenins, which supports our observations.¹⁹ Squamous cell carcinomas at different localisations show variable staining of catenins, although staining is mostly reduced.^{22, 25, 26} In many human cancers, reduced expression of α-catenin and β-catenin has been associated with poor tumour differentiation,^27–29 which is in line with our present study. In addition, lack of expression of α-catenin and β-catenin has been shown to correlate with a more advanced stage in laryngeal, ovarian, and bladder cancers.^{5, 27, 29} In NSCLC, reduced expression of E-cadherin has been associated with increased lymph node metastases,³⁰ but the expression of α-catenin and γ-catenins has not been related to TNM classification of the tumour,^{17, 19} a finding that is supported by our study.

In our study, increased proliferative activity of the tumour was associated with reduced expression of β-catenin and γ-catenin. This supports the theory that a reduction in the expression of catenins is a sign of more aggressive tumour behaviour.³ However, a similar relation between cell proliferation and reduced expression of catenins was not found in some previous studies.^{5, 31} Furthermore, the tumours in our series that showed nuclear β-catenin staining also had significantly higher PI values. According to recent studies, the nuclear accumulation of β-catenin is a consequence of the disturbance of the system that normally degrades cytosolic β-catenin.¹⁰ Nuclear β-catenin interacts with the Wnt/lef transcription pathway, resulting in changes in gene expression and cell differentiation during the malignant process.¹⁰ Recently, Nhieu and colleagues³² found an association between cell proliferation and nuclear accumulation of β-catenin in hepatocellular carcinoma, which is in agreement with our findings. Interestingly, we found that nuclear staining of γ-catenin was also related to increased cell proliferation. However, the role of γ-catenin in these regulatory events remains unclear.

Nuclear staining of β-catenin has been linked with clinicopathological factors in some previous studies.¹¹ In particular, increased nuclear staining of β-catenin has been demonstrated in colorectal carcinoma, being predominantly located at the invasion front.¹¹ This suggests a role for nuclear β-catenin in tumour cell migration and invasion. In addition, the nuclear accumulation of β-catenin was related to unfavourable prognosis in pharyngeal cancer.³³ In our study, we found no correlation between nuclear staining and clinicopathological parameters or survival. One explanation for these results might be the relatively low number of cases included in the analysis. However, other studies on carcinomas have found no evidence of nuclear accumulation of β-catenin or have questioned the clinical prognostic value of nuclear staining.^{1, 11, 34} It seems that the expression of nuclear β-catenin might be tumour type dependent, and studies with a larger number of positive cases are needed to clarify its clinicopathological value.

The loss of expression of catenins has been related to an unfavourable outcome in many human cancers,^{4, 5, 22} although conflicting data also exist.³⁵ In NSCLC, reduced expression of β-catenin and γ-catenin has been related to poor prognosis,^{17, 18} which contradicts our data. One explanation for our divergent results could be the relatively high number of recurrences noted in the adenocarcinomas. However, the analyses within the histological subgroups did not change the results.

Taken together, we have shown that the expression of different catenins (α, β, and γ) in NSCLC is related to the histological type and differentiation of the tumour. Furthermore, our study supports the important role of β-catenin and γ-catenin in regulating the proliferative activity of the tumour. However, we found no association between the catenins and the other clinicopathological parameters and survival. We conclude that catenins play an important role in regulating the behaviour of NSCLC, although they have no independent clinical prognostic value beyond the well known factors affecting the survival of patients.