From January 1996 to December 2001, 2,104 pathologically confirmed gastric cancer patients were registered in our hospital. Of these, there were 1,158 curatively resected patients, and immunohistochemical analyses of EGFR and c-erbB-2 were performed in 739 of these patients. We reviewed the clinicopathologic parameters of TNM stage, the World Health Organization classification, histological grade, Lauren classification, Ming classification, vascular invasion and neuronal invasion of these patients. Staging evaluation was done by the guidelines of the American Joint Committee of the Cancer, 5th edition (8).

Curative resection was defined as the removal of all gross tumors and the demonstration of tumor-negative proximal and distal surgical margins by microscopic examination. A total gastrectomy was performed in 114 patients (15.4%), subtotal gastrectomy in 625 patients (84.6%), D1 and D1+α resection was performed in 232 patients (31.4%) and D2 and D 2+α resection was performed in 507 patients (68.6%). Four hundred ninety-five (67.0%) of these patients were male, and 244 (33.0%) of these patients were female. The median age of the subjects was 59.0 years (the range was 19~80 years). The staging was as follows; IA in 340 (46.0%) patients, IB in 104 patients (14.1%), II in 116 patients (15.7%), IIIA in 106 patients (14.3%), and IIIB in 73 patients (9.9%) (Table 1).

Immunohistochemical staining was performed using the avidin-biotin-peroxidase complex with monoclonal antibodies raised against EGFR (Sigma, diluteed 1:500, St Louis, MO and C-erbB-2 (NCL-L-CB11, diluted 1:40, Novocastra, UK). Representative paraffin blocks containing tumor from each case were sectioned in 5µm slices and affixed to slides and then they were dried for 1 hour at 60℃. The sections were de-paraffined in xylene and rehydrated with a descending series of alcohol concentrations. Endogenous peroxidase activity was blocked by 3% hydrogen peroxidase for 15 min, and this was followed by washing with phosphate buffered saline (PBS), pH 7.2. The sections were then subjected to a heated antigen retrieval process by autoclaving with 1% zinc sulfate solution for 5 min. After cooling for 20 min at room temperature, the sections were incubated with 10% normal horse serum (Vectastain Elite kit) for 30 min. After decanting away the excess serum, the sections were incubated with primary antibody for 2 hour at 37℃. For the EGFR study, Sigma E2760 monoclonal antibody was used at a 1:500 dilution, and for the c-erbB-2 study, the monoclonal antibody NCL-L-CB11 was used at a 1:40 dilution (Novocastra, UK). The sections were subsequently incubated with prediluted biotinylated anti-mouse immunoglobulins (Vectastain Elite kit) for 30 min at 37℃. After washing with PBS, the sections were reacted with peroxidase-conjugated streptoavidin (Dako, Denmark) at a dilution of 1:500 for 30 min at 37℃. After washing with PBS, the peroxidase activity was detected using 3,3'-diaminobenzidine tetrahydrochloride (DAB), and the sections were counterstained with Mayer's hematoxylin.

Two pathologists, who were blind to the clinical outcome and features of the patients, independently evaluated all the immunohistochemical slides. The tumor cells with a positive cytoplasmic staining for EGFR and C-erbB-2 in more than 10% of tumor cells were interpreted as having a positive expression.

Statistical analysis was performed using χ-2 tests to compare percentages in the cross tabulations, and independent sample T tests were used to compare the means. Survival curves were generated by the Kaplan-Meier method (9), and they were compared by using the log-rank test (10). To determine the significant prognostic factors related to survival and relapse-free survival, multivariate analysis was performed with the Cox proportional hazards regression model (11). All the significance levels refer to two-sided tests. P values less than 0.05 were considered significant.

Statistical analyses were performed using SPSS for Windows 11.0 (Microsoft, Chicago, IL).

According to the World Health Organization classification, overexpression of C-erbB-2 was inversely associated with vascular invasion; vascular invasion was detected in 34.8% of patients with positive c-erbB-2 expressions, and in 51.2% of patients with negative c-erbB-2 expressions (p=0.001). Positive EGFR expressions were associated with more poorly differentiated tumor; there was EGFR expression in 32.7% of poorly differentiated tumor, 20.4% in moderately differentiated tumor and 12.5% in well differentiated tumor, respectively (p=0.000), and in neuronal invasions (p=0.03) (Table 2).

No significant correlation was observed for the overexpression of EGFR and C-erbB-2, and the Lauren classification, Ming classification, depth of invasion (pT category), lymph node involvement (pN category), and pathologic stage (Table 3).

The 7-year overall survival rate (Fig. 1), and the relapse-free survival rate (Fig. 2) were 87.2%, and 79.8%, respectively. There were no significant differences in the overall survival rate and relapse-free survival rates between patients with positive and negative EGFR expression (p=0.954, p=0.181, respectively) (Fig. 3, 4), and positive and negative C-erbB-2 expression (p=0.134, p=0.982, respectively) (Fig. 5, 6). According to multivariate Cox regression analysis, the tumor stage, size and age were important prognostic factors for overall survival, and tumor stage was important for the relapse-free survival. Positive expression of EGFR and C-erbB-2 were not independent prognostic factors for survival rates (Table 4).