Impact of E2F-1 Expression on Clinical Outcome of Gastric Adenocarcinoma Patients with Adjuvant Chemoradiation Therapy Free

Gastric cancer is the second leading cause of cancer death worldwide, with the incidence of 18.9/100,000 per year (1, 2). The incidence of gastric cancer was estimated to be 934,000 cases, with 56% of the new cases occurring in East Asia (3). Gastric cancer accounts for 20.8% of all cancers in Korea according to the Central Tumor Registry data for 2002 (4). Although gastrectomy is the only curative treatment in gastric cancer patients, a high recurrence rate ranging from 40% to 60% following curative surgery still accounts for poor overall survival. One of the milestone studies in adjuvant trials was the U.S. Intergroup Study INT-0116, which reported a significantly better survival from chemoradiation therapy with 5-fluorouracil (5-FU)/leucovorin and 45-Gy radiation mostly in D₀-D₁ nodal dissected gastric cancer patients (5). Moreover, we recently showed a potential survival benefit from adjuvant chemoradiation therapy (INT-0116) in D₂-dissected patients (5-year survival rate, 57.1% versus 51.0%, adjuvant chemoradiation group versus control group, respectively; P = 0.0198; ref. 6). To further improve survival, there is an urgent need to identify reliable molecular prognostic markers for survival or recurrence following adjuvant chemoradiation therapy, which will eventually evolve to the development of patient-tailored treatment strategies.

E2F-1 is a transcription factor that acts either as an oncogene or tumor suppressor depending on the tumor type and predominant signal (7, 8). Wikonkal et al. (9) reported increased levels of keratinocyte apoptosis in response to UV radiation in the E2F-1 knock-out mice and the subsequent suppression of apoptosis upon targeting of ectopic E2F-1 to the epidermis. In addition, these E2F-1 knock-out mice had predilection to develop spontaneous solid tumors (9, 10). More recently, several lines of evidence showed positive correlation between E2F-1 overexpression and radiosensitivity and/or chemosensitivity in certain solid tumor types. The overexpression of E2F-1 transcription factor by transduction with adenoviral E2F-1 considerably enhanced radiosensitivities in p53^wild-type and p53^null prostate cancer cells (11). Additionally, in osteosarcoma cells (U20S), E2F-1 overexpression resulted in a marked increase in sensitivity to paclitaxel and vinblastine (12). However, another study showed increased sensitivity to etoposide and doxorubicin in HT-1080 cells with E2F-1 overexpression (13).

E2F-1 controls the transcription of several genes encoding proteins involved in DNA synthesis such as thymidylate synthase (TS), dihydrofolate reductase, thymidine kinase, and ribonucleotide reductase (14–16). Recently, a meta-analysis incorporating 20 eligible studies with 887 cases of advanced colorectal cancer and 2,610 patients with localized colorectal cancer concluded that high levels of TS correlate with poorer survival in advanced tumors (17). However, the association between TS levels and progression-free survival and overall survival was less clear in adjuvantly treated colorectal cancer patients (17). For gastric cancer, the relationship between TS expression and clinical outcome has not been systematically studied as well in colorectal cancer. Several small studies reported an inverse correlation between TS expression levels with clinical outcome to 5-FU–based chemotherapy in gastric cancer (18–21). On the contrary, Choi et al. (22) observed no survival difference according to different TS expressions in 103 gastric cancer patients treated with adjuvant chemotherapy. Hence, a more reliable molecular marker, which consistently predicts clinical outcome after adjuvant therapy, needs to be identified in gastric cancer.

The aim of this study was to determine expression levels of E2F-1 and TS and to correlate these with disease-free and overall survival using tissue microarrays (TMA) of 465 primary tumors of stages IB to IV (M₀) of gastric cancer patients, who participated in an adjuvant chemoradiation study using the INT-0116 regimen.

Patients and tissues. We have previously reported the outcome of 544 stage II to IV (M₀) gastric patients who received adjuvant chemoradiation therapy after curative surgery (6). Of these patients and additional 23 stage IB patients who were included in our previous study, formalin-fixed, paraffin-embedded primary tumor tissues were available from 467 patients. The postoperative adjuvant treatment adopted was the same as that used for the INT-0116 (SWOG-9008) trial, and the results were previously reported (5). All patients provided written informed consent according to the institutional guideline, and the study was approved by the Institutional Review Board. The expressions of E2F-1 and TS were analyzed in all patients included. The clinical and pathologic features of the patients are shown in Table 1. The median age was 54, with a range of 23 to 70. By Lauren classification, 29.7% of patients had intestinal type. All patients received D₂ or greater lymph node dissection, and 33.8% of patients had stage IB or II. After a median follow-up duration of 91.1 months (65.6-134.5 months), a 5-year overall survival rate was 58.1%, and a 5-year disease-free survival rate was 57.0%. At the time of analysis, 211 (45.4%) patients were dead.

Tumor cell line and culture. Human gastric carcinoma cell, MKN74, was purchased from the American Type Culture Collection, and SNU638 and SNU668 cells were purchased from the Korean Cell Line Bank. All of the cell lines were maintained in RPMI 1640 (Life Technologies BRL) supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin and streptomycin. All cells were incubated in a humidified atmosphere containing 5% CO₂ at 37°C.

DNA transfection. The full-length E2F-1 open reading frame was cloned from normal fibroblast mRNA by reverse transcription-PCR for cloning into pCMVTaq4C (Invitrogen). Cells were transfected using Effectene transfection reagent (Qiagen) according to the manufacturer's recommendations.

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. The cell proliferation assay was used as a quantitative colorimetric method for measurement of cellular cytotoxicity. E2F-1–transfected cells were seeded at a density of 1 × 10⁴ cells per well in a 96-well plate. After culturing for 24 h, the media was changed to DMEM without phenol red and serum containing anticancer drugs at different concentrations. Four replicates were tested for each concentration. After 72-h treatment with 5-FU, the media supplemented with 0.5 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was added and incubated for 4 h. Formazan crystals were dissolved in DMSO for 5 min, and the absorbance was read using spectrophotometer at 540 to 650 nm.

Radiation therapy in vitro. The cells were transfected using Effectene (Qiagene) as recommended by the manufacturer, with E2F-1 plasmid cloned into the pcDNA3.1 expression vector (Invitrogen). After 24 h, cells were exposed to 4 Gy irradiation. Immediately after irradiation, cells were trypsinized, and 500 cells were plated into 60-mm dishes. After 14 days, colonies were stained with methylene blue and counted. Five experiments were done, and the results were averaged.

Western blot analysis. Total cell extracts were obtained using lysis buffer containing 20 mmol/L HEPES (pH, 7.4), 150 mmol/L NaCl, 1 mmol/L MgCl₂, 1 mmol/L EDTA, 2 mmol/L EGTA, 10% glycerol, 1% Triton X-100, 1 μg/mL leupeptin, and 1 μg/mL aprotinin. Equal amounts (30 μg) of cell lysates were dissolved in 12% SDS-PAGE and subjected to Western blot analysis using the enhanced chemiluminescence system (Amersham Corp.). Antibodies used were E2F-1 (KH95, Santa Cruz Biotechnology Inc.; 1:1,000) and β-actin (Sigma-Aldrich; 1:5,000).

TMAs. All H&E-stained slides were reviewed, and representative tumor tissue samples were selected for each case. The corresponding formalin-fixed, paraffin-embedded tissue blocks were retrieved. The selected area was circled on the block with a marker pen for TMA construction. Each 2.0-mm tissue core was taken from a representative region of each tissue block using the AccuMax (Isu Abxis Co., Ltd.). Eleven TMAs were constructed, and each TMA contained 45 sample cores and three nontumor controls.

Immunohistochemistry. Tissue microarray blocks were sectioned with 4 μm thickness. Immunohistochemical study was done using the streptavidin-biotin complex method and TechMate 1000 automated staining system (DakoChemmate). Primary antibodies used and working dilutions employed were as follows; E2F-1 mouse monoclonal antibody (mAb; KH95, Zymed; 1:100) and TS mouse mAb (TS106, Chemicon; 1:50). Deparaffinized sections were treated with 3% hydrogen peroxide in methanol for 10 min to inhibit endogenous peroxidase. Sections were immersed in 0.01 mol/L citrate buffer (pH, 6.0) and heated in a pressure cooker for 30 min. Sections were then incubated with primary antibody for 50 min at room temperature. Each section was treated sequentially with biotinylated secondary antibody (anti-mouse immunoglobulin) and streptavidin peroxidase complex (DakoChemmate). 3,3′-Diaminobenzidine tetrahydrochloride was used as a chromogen, and then Mayer's hematoxylin counterstain was applied. Breast cancer section and tonsil section were processed in parallel as a positive control for E2F-1 and TS, respectively. Negative controls (isotype-matched irrelevant antibody) were run simultaneously.

The results of staining were evaluated by two independent pathologists (C.K.P. and K.M.K.) who were blinded to the clinical data. There was close agreement (>90%) between both investigators. In cases of disagreement, final grading was determined by consensus. For assessment of the positivity of immunostaining for each section, only nuclear staining was regarded as positive. Positive cells were counted by monitoring at least 1,000 cancer cells from more than five high-power fields where positive cells were present at a relatively uniform density. We considered the E2F-1 and TS as positive when >5% and >25% of cancer cell nuclei showed positive immunostaining, respectively.

Statistical analyses. Disease-free survival was defined as the time from surgery to the first relapse of cancer, occurrence of a second primary tumor, or death of any cause. Overall survival was measured from the date of surgery to the date of death. Overall survival and disease-free survival were calculated using the Kaplan-Meier method. The time of analyses for survival and disease-free survival was as of September 2006. Correlation analyses of the expression of E2F-1 or TS with clinical and pathologic variables were done using the two-sided χ² test or Fisher's exact test. Differences in disease-free and overall survival according to expressions of E2F-1 and TS were compared using log-rank tests and Cox proportional hazard analysis. P value <0.05 was considered statistically significant.

Overexpression of E2F-1 increased the sensitivity to 5-FU and radiation in vitro. To investigate the effect of overexpression of E2F-1 on 5-FU sensitivity in vitro, gastric cancer cells (MKN74, SNU638, SNU668) were transfected with the full-length E2F-1 open reading frame, and a cell proliferation assay using MTT was done (Fig. 1). The E2F-1 protein overexpression was confirmed by Western blot analysis (Fig. 1D and E). The IC₅₀ of 5-FU was significantly lower in E2F-1–overexpressed gastric cancer cells (0.639-0.660 μg/mL) as compared with those transfected with mock or vector only (1.035-1.396 and 1.045-1.383 μg/mL, respectively; Fig. 1A). Furthermore, all three of the E2F-1–overexpressed gastric cancer cells (MKN74, SNU638, SNU668) showed markedly increased radiosensitivity as compared with those transfected with mock or vector only (Fig. 1B). The percentage of colony formation decreased as the pE2F-1 vector concentration increased (Fig. 1C).

Expression of E2F-1 and TS. Immunohistochemistry for E2F-1 and TS was done on all 467 cases. Four hundred and sixty-five cases were interpretable for E2F-1, and 463 cases were analyzable for TS. The E2F-1 immunopositivity was observed in 103 (22.2%) of the 465 cases (Fig. 2A). The TS immunoreactivity was found in 88 (19.0%) of the 463 cases (Fig. 2B). Correlative analyses between E2F-1 and clinical parameters showed no significant relationship with age, lymphovascular invasion, histology, lymph node metastasis, or tumor infiltration status (Table 2). However, the proportion of E2F-1 negativity was significantly higher in the diffuse type (77.3%) when compared with intestinal type (22.7%, P < 0.001).

Disease-free and overall survival in relation to E2F-1 and TS expression at univariate analyses. The variables tested at the univariate level were as follows: sex, age, grade, histology, stage, lymphovascular invasion, Lauren classification, TS expression, and E2F-1 expression. Among these variables, E2F-1 immunopositivity was significantly associated with better disease-free survival [5-year DFS, 54.7% versus 65.0%, E2F-1 (−) versus E2F-1 (+), respectively; P = 0.023] and overall survival [5-year OS, 55.5% versus 67.0%, E2F-1 (−) versus E2F-1 (+), respectively; P = 0.007] as compared with those for E2F-1 immunonegativity patients by log-rank test (Table 3, Fig. 3). On the contrary, TS expression was not predictive of disease-free [5-year DFS, 56.3% versus 60.2%, TS (−) versus TS (+), respectively; P = 0.361] or overall survival [5-year OS, 57.3% versus 61.4%, TS (−) versus TS (+), respectively; P = 0.312].

Correlation between E2F-1 expression and outcome of adjuvant chemoradiation therapy in multivariate analyses. Using backward stepwise Cox proportional hazards regression modeling, the following variables were tested for their influence on disease-free and overall survival: age (≤60 versus >60), sex, Lauren classification, histology (tubular adenocarcinoma versus others, namely, signet ring cell, mucinous, papillary, adenosquamous, and hepatoid), stage (IB/II versus III/IV M₀), TS expression, and E2F-1 expression. For overall survival in all patients, the E2F-1 immunopositivity predicted more favorable survival as compared with the E2F-1 immunonegativity with borderline statistical significance [P = 0.050, hazard ratio [HR] = 0.702, 95% confidence interval (95% CI), 0.487, 1.013]. However, the E2F-1 immunopositivity did not retain its statistical significance at multivariate analysis for predicting disease-free survival (data not shown, P = 0.270), but stage was the only influential factor for disease-free survival in stage IB to IV (M₀) patients (P < 0.001). TS immunopositivity did not influence survival (P = 0.459) or disease-free survival (P = 0.447).

Herein, we report for the first time that the E2F-1 immunoreactivity independently predicts favorable overall survival in gastric cancer patients who received adjuvant chemoradiation therapy with 5-FU and leucovorin after D₂ or greater resection. In a correlative analysis, E2F-1 reactivity was associated with localized tumor, intestinal type, and TS expression with statistical significances. In contrast, TS expression did not predict the clinical outcome of gastric cancer patients who were treated with adjuvant chemoradiation therapy.

E2F-1 is a transcriptional factor that has dual roles as tumor suppressor and oncogene depending on the cell type and cell milieu. Several lines of evidence may account for the increased sensitivity to 5-FU/leucovorin in gastric cancer patients, subsequently leading to survival prolongations. First, the E2F-1 protein expression was shown to be induced in human medulloblastoma, glioma, lung, colon, and bladder cancer cell lines after treatment with Adriamycin or etoposide (23). Tumor cells sensitized by overexpression of a proapoptotic protein, E2F-1, may further enhance chemosensitivity to chemotherapeutic drugs by producing a synergistic effect. This was verified by infecting melanoma cells with adenoviral E2F-1, which sensitized the cells to apoptosis induced by etoposide or Adriamycin (24). In support of this report, another group illustrated that osteosarcoma cells with E2F-1 overexpression were sensitized to paclitaxel and vinblastine (12). Yet another study also revealed an additive effect of adenoviral E2F-1 and etoposide in pancreatic carcinoma cells (25). Although several in vitro studies reported 5-FU resistance in E2F-1–overexpressed cells (12, 13, 23–25), the drug sensitivity was cell type specific. We tested in vitro chemosensitivity test to 5-FU in gastric cancer and found a significantly increased sensitivity in E2F-1–overexpressed cells in all three cell lines (MKN74, SNU638, SNU668; Fig. 1). Moreover, E2F-1–overexpressed cells showed a marked decrease in survival following radiation therapy. These findings support our clinical data, which suggested a favorable outcome in patients with E2F-1 overexpression.

E2F-1 overexpression has been shown to augment cellular radiosensitivity and, hence, cell deaths in several cell types in vitro, such as fibrosarcoma cells (26), myeloid cells (27), and prostate cancer cells (11). Interestingly, adenoviral E2F-1 transduction of LNCaP and PC3 cells showed significantly enhanced apoptosis upon the addition of radiation therapy, suggesting a possible application of adenoviral E2F-1 as an adjunct to radiation therapy in prostate cancer therapy (11). Given the fact that all tumor samples were from gastric cancer patients treated with adjuvant chemotherapy and radiation therapy and that the E2F-1–positive group showed improved survival, it can be postulated that an E2F-1 apoptotic effect may have been augmented by the addition of radiotherapy and a chemotherapeutic agent, 5-FU.

E2F-1 expression has not been extensively studied in human gastric cancer specimens. One small study showed that gene amplification of E2F-1 was detected in 4% of gastric carcinoma (n = 23), increased expression of E2F-1 mRNA was observed in 40% (n = 30), and E2F-1 protein was overexpressed in 63% (n = 30) by Western blotting (28). Another study on 38 gastric cancer tissues from patients postoperatively treated with 5-FU/methotrexate showed 37% positivity for E2F-1, with a cutoff value of 25% nuclear reactivity, and reported no direct relationship between E2F-1 and clinical outcome (29). We observed the E2F-1 immunopositivity rate of 22% (103 of 465 samples) with a cutoff value of 5% immunoreactivity. The discrepancies in percentage of E2F-1 positivity may be due to differences in antibody used and tissue fixation and variations in staining methods among studies. One of the possible limitations of our study would be the statistical impact of E2F-1 immunopositivity for survival at multivariate analysis (Table 3). Although the E2F-1 expression retained its statistical significance at a multivariate level (P = 0.05), the stage was still the most powerful prognostic factor in our series (P < 0.001). Moreover, the upper limit of the 95% CI for HR of E2F-1 expression slightly crossed the midline (1.013). A long-term follow-up of the cohort of patients may further clarify the aspect in future studies.

E2F-1 transcription factor is known to induce genes encoding S-phase proteins, including TS (30). Thus, we investigated the correlation between E2F-1 and TS expressions and observed the E2F-1 (−)/TS (−) concordance in 66.0% (n = 307/465) and E2F-1 (+)/TS (+) concordance in 7.5% (n = 35/465; Table 3). There are conflicting reports on the correlations of E2F-1 and TS expressions in the literature. One study reported that the level of TS mRNA expression correlated closely with the level of E2F-1 mRNA expression in 23 colon cancer tissues (31). However, a recent study showed no correlation between E2F-1 and TS expression by immunohistochemistry in 25 colorectal tissues treated with palliative 5-FU–based chemotherapy (16). It is difficult to draw any conclusion from the previous reports due to a small cohort of patients.

The role of TS in predicting clinical outcome to 5-FU–based chemotherapy is also controversial in gastric cancer, both in adjuvant and palliative settings. Several small studies reported that gastric cancer patients with high TS expression levels showed poorer clinical outcome to 5-FU–based chemotherapy as compared with those with low TS expression in palliative setting (18–21). There are few reports on TS expression levels in adjuvant setting for gastric cancer, however. Choi et al. (22) observed no survival difference according to different TS expressions in 103 gastric cancer patients treated with adjuvant chemotherapy. On the contrary, recent correlative study on 94 gastric cancer tissues from patients treated with epirubicin/mitomycin C followed by doxifluridine showed that a high TS expression was associated with longer survival (P = 0.0392; ref. 32). In our study, TS expression did not predict disease-free or overall survival with statistical significance.

We recommend E2F-1 as a potential biological marker that is predictive of clinical outcome following adjuvant chemoradiation therapy in gastric cancer patients. Based on our data, TS is not a predictive marker in this particular setting. A partial explanation for this would be that E2F-1 regulates TS expression at upstream, and further analyses on upstream molecules are warranted. One of the caveats of the study may be that in the United States, only about 10% of the patients undergo D₂ dissection according to the intergroup study (5), and that it may or may not be applicable to the U.S. population; however, there is no difference in survival between patients who undergo D₁ versus D₂ resection in randomized trials. Although it should be verified in future studies, it can be speculated that gastric cancer patients, who show a high level of E2F-1 expression, may benefit most from the adjuvant chemoradiation therapy with the INT-0116 regimen.