Seventy-two patients diagnosed with GC and identified from the pathology archives of the General Hospital of PLA (from 2004 through 2006) were randomly enrolled in this study. Patients that underwent curative surgery had no prior treatment such as chemotherapy or radiation therapy. Of the 72 patients, 40 were men and 32 women with a mean age of 52 years (range 41–78). In each case, GC tissues and adjacent non-tumorous tissues (non-tumor group) were obtained for pathological examination. The detailed pathological results were obtained from the Department of Pathology of our hospital. The study protocol was approved by the Clinical Research Ethics Committee of the Chinese PLA General Hospital. Informed consent was obtained from each patient.

The histological tumor type of GC was defined according to Lauren’s classification. In this study, the 72 patients with GC included 40 intestinal- and 32 diffuse-type. Tumor node metastasis (TNM) staging was performed at the pathologic diagnosis of GC according to the American Joint Committee on Cancer staging system (Table I). Patients were followed up for 8–30 months.

Tumor and normal cells from the same patients were selectively procured from hematoxylin and eosin-stained slides, using a 30G1/2 hypodermic needle (Becton Dickinson, Franklin Lakes, NJ, USA). The needle was affixed to a micromanipulator by microdissection. DNA extraction was performed by a modified single-step DNA extraction method and then overnight digestion with SDS and proteinase K at 37°C. This procedure was followed by standard phenol-chloroform (1:1) extraction and ethanol precipitation.

Sections (5 μm) of formalin-fixed, paraffin-embedded tissues were dewaxed in xylene and rehydrated through a series of ethanol. Antigen retrieval was carried out at this stage in a microwave oven. Sections were then blocked with 3% hydrogen peroxidase followed by incubation with a 50% protein blocking agent. Fetal bovine serum (10%), with or without the HER2/neu antibody (1/100), was applied to each slide and incubated for 30 min. Slides were counterstained with hematoxylin and mounted. Omission of the specific antibody was used as the negative control. The anti-HER2 monoclonal antibody (Dako Herceptin Test kit) and Dako EnVision™ kit (DakoCytomation Co., Denmark) were used for the IHC staining.

After the IHC staining, slides were microscopically interpreted in a blinded fashion by two pathologists. Membrane staining for HER2/neu was taken into account. IHC expression of HER2/neu was scored as: no staining or in <10% of the tumor cells (score 0); faint/barely perceptible partial staining in >10% of tumor cells (score 1+); weak to moderate staining of the entire membrane and/or cytoplasma in >10% of tumor cells (score 2+) and strong staining in >10% of tumor cells (score 3+). Scores 0 and 1+ were considered to be negative, and scores 2+ and 3+ positive for HER2/neu overexpression.

HER2 mutations in exons 18–23 encoding the kinase domain were detected. Genomic DNA from each of the tumor and normal cells was amplified with three primer pairs covering exon 18–23. The primer sequences were (forward and reverse, respectively): exons 18–19 (5-GACACCTAGCGGAGCGATGC-3 and 5-ATGGGGTC CTTCCTGTCCTC-3), exons 20–21 (5-GTGATGGTTGGGA GGCTGTG-3 and 5-CTGCTCCTTGGTCCTTCAC-3), exons 22–23 (5-GGCCACCTCCCCACAACACA-3 and 5-GCTCAG CCACGCACATTTGAC-3). Calculations of cDNA of HER2 were carried out with respect to the ATG start codon (NM_004448).

The PCR reaction mixture was denatured for 1 min at 94°C and incubated for 35 cycles (denaturing for 60 sec at 94°C, annealing for 50 sec at 58–59°C and extension for 50 sec at 72°C). Nest-PCR was performed when the amplification was not adequate for direct sequencing. The primers designed for the nest-PCR are not listed.

When the qualified PCR products were obtained, their direct sequencing was carried out using a cyclic sequencing kit (Perkin-Elmer, Foster City, CA, USA) according to the manufacturer’s recommendation. When mutations were identified, further confirmation using sequencing in both directions was carried out.

The SPSS software package (SPSS, Inc., Chicago, IL, USA) was used in the statistical analysis. Comparison of the frequencies between the HER2/neu expression status and clinicopathological variables was performed with Fisher’s exact test (two-sided) by SPSS version 12.0. The log-rank test was used for survival analysis. The Kaplan-Meier method was used to calculate survival curves. Multivariate analysis was carried out using Cox’s proportional hazards model in order to identify the primary prognostic indicators that were independently associated with survival. A probability of 0.05 was considered to be a statistically significant difference.

Overexpression for HER2/neu using IHC staining was detected in 13 of the 72 GC patients and in 4 of the 72 tissue samples in the non-tumor group. A significant difference for HER2/neu overexpression between the GC and non-tumor group was observed (18.1 vs. 5.6%, P<0.05) (Table I). The location of HER2/neu protein was limited to the membrane (Fig. 1). No detectable immunoreactivity was noted in the negative control (data not shown), indicating the specificity of IHC staining.

As shown in Table I, HER2/neu overexpression was detected in 11 of the 37 intestinal-type and in 2 of the 35 diffuse-type GC cases. The rate of HER2/neu overexpression was higher in the intestinal-type GCs than that of the diffuse-type (29.7 vs. 5.7%, P<0.05).

No significant difference was noted in gender, age, tumor location and tumor size with respect to the overexpression of HER2/neu (data not shown). Table I shows that 11 of the 39 stage III–IV cases were HER2-positive, with a higher rate than that of stage I–II (28.2 vs. 6.6%, P<0.05).

Univariate analysis (log-rank test) showed that age (≥60 years), advanced TNM stage (III–IV) and the overexpression of HER2/neu correlated with a less favorable patient survival (Table II). Survival curves computed according to the Kaplan-Meier method (Fig. 2) showed that overexpression of HER2/neu correlated significantly with a decreased patient survival (P=0.046). Although the TNM stage was the most dominant prognostic factor, Cox’s proportional hazard model also identified HER2/neu overexpression as an independent prognostic factor (Table III).

Each case in our study exhibited qualified DNA bands comprising the kinase domain of HER2 for direct DNA sequencing (Fig. 3). In the 72 cases examined, no mutations were identified in the tumor or corresponding tissues.