Decreased expression of eukaryotic initiation factor 3f is an adverse prognostic factor for stage I–III gastric cancer

A review of surgical specimens at the The Second Hospital of Shandong University between November 2007 and March 2009 yielded 195 cases of GC patients who had undergone curative surgery. The mean age of the patients was 62 years (range, 31–90 years). The median follow-up period was 38 months (range: 3–54 months). The TNM stage of tumors was assessed according to the 7th ed. of the TNM Classification of GC [21]. The Her-2/neu overexpression was detected by postoperative fluorescence in situ hybridization examination. No patient underwent radiotherapy or chemotherapy before operation.

For immunohistochemical staining, the paraffin-embedded tissue sections were obtained from 195 cases of radical resection of GC. The fragments were fixed in neutral formol. All the cases were reviewed by a pathologist to confirm the malignancy. The investigation was approved by the ethics committee of the medical faculty and written informed consents were obtained from all patients.

The polyclonal antibody (ab64177) to eIF3f was bought from Abcom (Cambridge, UK). The paraffin-embedded tissue blocks were sectioned in 3 to 4 mm slices and placed on Anti slides. After de-waxing and hydration, the slides were rinsed in phosphate-buffered saline (PBS) and blocked for 10 min with 3% hydrogen peroxide to deprive the endogenous peroxidase activity. After antigen retrieval with the use of a microwave, the specimens were incubated with the anti eIF3f MAb (diluted 1:100 in PBS) at 37°C for 1.5 hours. The sections were then washed 3 × 3 min in PBS and incubated with 1 and 2 Reagent of PV9000 Mouse/Rabbit hypersensitivity two-step immunohistochemical Kit (Beijing fir Jinqiao, Beijing, China) for a total of 60 min at 37°C in a humid chamber. The sections were washed 3 × 3 min with PBS, followed by the addition of diaminobenzidine as a chromogen for 3 to 5 min, which was strictly controlled under a microscope. Antibodies were optimized using a positive control tissue according to the manufacturer’s instructions. In negative controls, the primary antibody was replaced with PBS. The remaining procedures were performed in parallel with other specimens. Each slide was scored in a blinded fashion by two pathologists according to the manufacturer’s recommended criteria at × 100 and × 200 magnification. The overall percentage of positive cells on an immunostained section was determined according to the pattern of intracellular localization. The extent and pattern of the eIF3f-specific immunostaining within a tissue section were determined by the percentage of cells with cytoplasm staining. The immunostaining was read in a semiquantitative manner. Three visual fields were examined randomly and the rate of positive cells was divided into less than 5% (score 0), 6% to 25% (score 1), 26% to 50% (score 2), 51% to 75% (score 3), and more than 75% (score 4). The staining intensity can be divided into three grades: no staining (score 0), slightly yellowish (score 1), brownish yellow (score 2), and dark brown (score 3). The multiplication of the two were graded as follows: 0 (score 0), 1+ (score 1–4), 2+ (score 5–8), and 3+ (score 9–12). Intensity scores of 0 or 1+ were designated as low expression and 2+ or 3+ were designated as high expression.

Differences between groups were analyzed using a Student’s t-test for continuous variables and a χ² test or Fisher’s exact test for proportions. The overall survival rate was estimated by the Kaplan-Meier method. Univariate analyses (Cox proportional hazard regression models) were also performed to assess the prognostic value of nucleolin expression and other clinicopathological features. We utilized SAS 9.2 software system for statistical analysis.

From the 195 cases, 142 men and 53 women all accepted radical operation of GC, including 116 primary and 79 recurrent GC patients. Pathologically, all of the 195 cases were adenocarcinoma, which covered 12 well differentiated, 27 moderate differentiated, 156 low differentiated, and 4 signet-ring cell carcinoma. For all specimens, 31 cases were demonstrated as Her-2/neu-positive.

Immunohistochemical staining for eIF3f showed a cytoplasm staining pattern in both GC (Figure 1a, b) and ANCT (Figure 1c, d) tissues. Immunoscores were calculated when the presence of eIF3f deposits in tumor glands and cytoplasmic staining were positive. The rate of high eIF3f expression in GC and ANCT tissues was 33.8% (66/195) and 59.5% (116/195), respectively.

The association of eIF3f levels with clinicopathological features in GC was also assessed in Table 1. Low eIF3f expression in GC was significantly associated with more advanced tumor stages (P = 0.02) and likelihood of recurrence (P = 0.04). However, eIF3f expression was not significantly correlated with sex (P = 0.87), age (P = 0.17), tumor differentiation (P = 0.46), and tumor size (P = 0.16). Out of 136 stage I–II cases, 53 (40.0%) had high eIF3f staining densitometry values, and these were higher than cases of stage III (P = 0.02). Furthermore, the overall densitometric analysis of immunostaining revealed that the downregulated eIF3f protein expression varied significantly between primary and recurrent tumors (P = 0.04).

The prognostic effect of eIF3f on the survival rate of GC patients was investigated by comparing the survival rate of patients with high or low levels of eIF3f protein expression in tumors using Kaplan-Meier survival curves and log-rank test. Univariate Cox regression analysis showed that clinical variables, including tumor size (log-rank, P = 0.03; Figure 2a), tumor recurrence (log-rank, P <0.01; Figure 2b), and eIF3f expression (log-rank, P = 0.04; Figure 2c) were all significantly associated with overall survival (Table 2). Data indicates that high expression of eIF3f protein was a significant prognostic factor for better survival of GC patients. Furthermore, multivariate Cox regression analyses showed that eIF3f expression was an independent predictor for the overall survival of patients with GC.