The present study was approved by the Ethics Committee of the University of South China. The patients and healthy volunteers provided signed informed consent. The present study was conducted in accordance with the Declaration of Helsinki.

Gastric cancer cell lines MGC-803 and SGC-7901 and normal gastric mucosal epithelial GES-1 cells were cultured in RPMI-1640 medium containing heat-inactivated 15% fetal bovine serum (FBS) on 6-well plate with a cover slide. Cells were cultured for three days at 37°C in 5% CO₂.

Cells were washed three times in phosphate-buffered saline (PBS), fixed for 15 min in 10% formaldehyde, and then again washed three times with PBS. The S-P immunohistochemical kit was purchased from Fujian Maixin Biological Technology Ltd. (Fujian, China). The cells were then incubated 10 min with peroxidase blocking solution and washed three times with PBS. Cells were incubated at 37°C for 10 min with enough non-immunologic animal serum to block non-specific binding sites, and then anti-hnRNP K antibody (diluted 1:1,500; ab-32969; Abcam Trade Company) was added and incubated at 4°C overnight. Next, the cells were incubated at 37°C for 10 min with the secondary antibody marked by biotin, and then incubated for 10 min with Streptomyces avidin-peroxidase. The reaction was visualized using DAB substrate chromogen (Fujian Maixin Biological Technology Ltd.).

Human gastric cancer tissue microarrays (TMA) were obtained from Dr Yongjun Wu (Department of Pathology, The First People's Hospital of Xiangtan, Hunan, China). The specimens were biopsy or gastric cancer resection specimens collected from year 2003 to 2006, including 199 gastric cancer, 31 tumor-adjacent gastric mucosal and 98 normal gastric mucosal specimens. None of the patients received preoperative radiotherapy or chemotherapy.

We placed the paraffin sections at 55°C constant temperature in an incubator for one night, and then they were dewaxed with xylene, and washed three times with PBS. After that an antigen retrieval step was performed. Immunohistochemistry (IHC) was performed according to S-P kit instructions. Immunohistochemical reactions were developed in freshly prepared 3,3′-diamino-benzidine tetrahydrochloride (DAB kit; Fujian Maixin Biological Technology Ltd.) for immune complex visualization. Finally, the TMA IHC was evaluated by light microscopic examination, and the intensity of immunostaining was assessed by two independent professors of pathology. We selected five different high magnifications, counting the total number of cells and hnRNP K-positive cells. The results were evaluated based on both the intensity of immunostaining and the positive cell percentage. The intensity of immunostaining in each core was scored 0–3 (0, negative; 1, weak, 2, moderate; and 3, strong). In addition, the samples were scored into four groups based on the percentage of positively stained cells: 0–25% positivity scored as 1, 26–50% as 2, 51–75% as 3, and 76–100% as 4. The staining intensity score and the percentage immunoreactivity score were then calculated to obtain a composite score (composite score = intensity score × percentage score); 0–5 was defined as low expression, and 6–12 was defined as high expression (16).

Total cell proteins from the MGC-803, SGC-7901 and GES-1 cells were quantified using a BCA protein assay kit. Then, the samples were separated by 10% SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membranes, and then blocked with Tris-buffered saline Tween-20 (TBST) containing 5% non-fat milk for 1 h, and washed with TBST three times. Then, the samples were probed at 4°C overnight with rabbit anti-hnRNP K antibody (diluted 1:2,000; ab-32969), washed with TBST three times 5 min each time, and incubated with the appropriate secondary antibody (goat anti-rabbit IgG; diluted 1:3,000) for 1 h. The samples were then washed with TBST three times again for 5 min each time and enhanced chemiluminescence (ECL) western blotting detection reagents were used to visualize the target proteins (both from KeyGen Biotech, Nanjing, China).

We used a total of 96 serum samples including 37 paired samples from gastric cancer patients (preoperative and postoperative) and 22 samples from healthy volunteers with no evidence of cancer and other disease. All samples were collected from the Affiliated First Hospital of the University of South China according to our previously published protocol (17). The serum samples were analyzed for hnRNP K using a commercially available ELISA (USCN Life Science Inc., Houston, TX, USA).

Data are reported as mean ± SD. The differences between two groups were compared using Chi-square tests, t-test, one-way ANOVA. Kaplan-Meier survival analysis and log-rank test were performed to determine survival differences between the different groups. Factors associated with the outcomes were evaluated by Cox multivariate regression analysis. Statistical analysis was carried out using SPSS version 17.0 software program (SPSS, Inc., Chicago, IL, USA). All analyses were regarded as statistically significant at the P<0.05 level, and all P-values were two-tailed.

To evaluate the expression of hnRNP K between gastric cancer and normal gastric mucosal epithelial cells, we performed cell immunochemical staining analysis in two gastric cancer cell lines (MGC-803 and SGC-7901) and a normal gastric mucosal epithelial cell line (GES-1). Obviously, the intensity of immunostaining in the MGC-803 and SGC-7901 cells was stronger than that found in the GES-1 cells (Fig. 1A). We also performed western blot analysis in the three cell lines. Compared with the normal gastric mucosal epithelial cell line GES-1, hnRNP K protein expression was significantly higher in the gastric cancer cell lines MGC-803 and SGC-7901 (P<0.05; Fig. 1B).

To confirm whether hnRNP K expression is elevated in human gastric cancer tissues, we analyzed the level of hnRNP K protein by IHC in the human gastric carcinoma tissue microarrays (TMAs). The hnRNP K expression in gastric cancer tissues was significantly higher compared with that found in the tumor-adjacent gastric mucosal and normal gastric mucosal specimens. Both in the gastric cancer and non-gastric carcinoma tissues, hnRNP K was localized in the nucleus and showed low staining in the cytoplasm. However, we found that hnRNP K expression was significantly elevated in the gastric glandular neck epithelium when compared to that in the other glandular epithelium (Fig. 2). According to various studies, gastric gland stem cells localize in the gastric neck glands, where they have the ability to proliferate and differentiate. This indicates that elevated expression of hnRNP K promotes the ability of gastric cancer cells to grow and proliferate.

We also evaluated the relationship between the clinicopathological parameters of the gastric cancer patients and the nRNP K expression level. We found that there was no correlation between the protein expression level and gender, age, degree of differentiation or tumor size. However, hnRNP K expression was significantly elevated in the group with lymph node metastasis than that found in the group without lymph node metastasis, and was also higher in stage III and IV than stage I and II samples (Table I; Fig. 3)

One hundred ninety-nine patients with gastric cancer were followed-up for 8 years. There were 34 cases lost to follow-up and 121 patients died during this time; overall survival (OS) was 26.7%. Twenty-eight of the 53 (52.8%) patients with low expression of hnRNP K died; while 93 of 112 (83.0%) patients with high hnRNP K died. The death rate of the high hnRNP K group was 1.57 times higher than that in the low hnRNP K group. We found that patients with elevated hnRNP K expression in cancer cells had poorer survival compared with those patients with low hnRNP K expression (log-rank=62.339, P=0.000; Fig. 4). The mean survival time of patients with low hnRNP K expression was 52.277 months and the median survival time was 61 months, while the mean survival time in the cohort with higher hnRNP K expression was 20.657 months and the median survival time was 18 months.

Cox multivariate analysis indicated that the degree of differentiation and lymph node metastasis were associated with poor survival. Moreover, the risk of death in the poorly differentiated group was 2.203 times (death ratio=1/HR) higher than that of the moderately and well differentiated groups. The mortality risk in the lymph node metastasis group was 1.976 times higher than that in the patients with no lymph node metastasis (Table II).

To evaluate the expression level of serum hnRNP K between gastric cancer patients and normal controls, we measured the levels of hnRNP K in serum samples by ELISA. Most serum samples showed low expression of hnRNP K. There was no significant difference between the gastric cancer patients and the healthy volunteers or between the preoperative and postoperative groups (Table III). We believe that hnRNP K is not a suitable circulating tumor biomarker for detecting gastric cancer.