Overexpression of Phosphorylated Mammalian Target of Rapamycin Predicts Lymph Node Metastasis and Prognosis of Chinese Patients with Gastric Cancer Free

Gastric cancer is the second leading cause of cancer-related deaths worldwide. More than one-third of all gastric cancer cases occur in China (1, 2). Researchers have described many biological markers associated with gastric cancer progression and outcome and observed differences in survival and clinicopathologic prognostic factors in patients with this disease in different countries and racial and ethnic groups within given populations and areas (3–8). Molecules involved in several signaling pathways were potential prognostic and therapeutic markers (9, 10). Determining the expression profiles of key molecules in the survival pathways in gastric cancer progression may aid in diagnosing and predicting disease progression. Because of their potential roles in gastric cancer progression, the phosphatidylinositol 3-kinase-dependent pathway and angiogenesis pathway have become the foci for development of novel anticancer drugs (9), including mammalian target of rapamycin (mTOR).

The mTOR is a serine/threonine downstream mediator in phosphatidylinositol 3-kinase signaling pathway. Physiologically, mTOR is a central controller of eukaryotic cell growth and proliferation and plays a critical role in regulating important cellular functions, including proliferation, growth, survival, mobility, and angiogenesis (10, 11). Authors have reported activation of the mTOR pathway and overexpression of mTOR in several types of tumors, including hepatocellular (12), renal cell (13), prostate (14), and breast (15) cancers. Moreover, patients with breast cancer who had mTOR overexpression had a risk of recurrence three times greater than that in patients without mTOR overexpression (15). Lang et al. (16) recently reported overexpression of p-mTOR in 60% of intestinal-type and 64% of diffuse-type human gastric cancer cases and that inhibition of expression of mTOR significantly impaired gastric cancer cell migration and proliferation. However, the role of mTOR activation in gastric cancer carcinogenesis and progression is poorly understood. The clinical significance of the expression mTOR protein and particularly the activated form of p-mTOR in patients with gastric cancer also remains unclear.

Angiogenesis is a complex process and a crucial step in tumor formation and progression. This process is regulated by a balance between proangiogenic and antiangiogenic molecules. Among numerous proangiogenic factor, vascular endothelial growth factor (VEGF) has been established as the major angiogenic factor (17). Frequently observed in gastric cancer (18–20), overexpression of VEGF closely correlates with lymph node metastasis (21), liver metastasis (22), age (23), and tumor size (24) and often indicates a poor prognosis (25, 26) and therapy resistance (27). Although VEGF is constitutively expressed in many tumor cells, its expression is subject to several regulatory mechanisms (17). The potential role of mTOR pathway in VEGF regulation is supported by several recent studies, showing that inhibition of mTOR associated with a decreased VEGF expression (28–30). Given the critical role of VEGF in cancer angiogenesis in general, those recent findings suggest a critical link between mTOR activation and VEGF-mediated angiogenesis in gastric cancer.

The present study sought to provide both clinical and experimental evidence for the significance of mTOR activation in gastric cancer and the potential underlying mechanisms of its effect on gastric cancer pathogenesis, particularly angiogenesis.

Patient specimens and tissue microarray construction. A total of 1,072 patients with gastric adenocarcinoma who underwent curative surgery without prior treatments at Changhai Hospital from 2001 to 2005 were enrolled in this study. Patients with other gastric tumors, such as neuroendocrine tumors, lymphoma, and sarcoma, were excluded from this study. The patients' medical records were reviewed to obtain data including age at diagnosis, sex, tumor location, tumor size (diameter), nerve invasion, and American Joint Committee on Cancer stage. The mean age of patients at tumor resection was 59 years; 757 (71%) were male and 315 (29%) were female. Clinical follow-up results were available only for 669 patients from the Shanghai area (mean follow-up duration, 37 months; range, 21-73 months). Twenty-one paraffin-embedded tissue microarray blocks of gastric tumor, matched normal gastric, and lymph node metastasis tissue specimens obtained from those patients were created using a manual arrayer (Beecher Instruments). Each block had two 1.5 mm cores of primary tumor tissue and at least one 1.5 mm core of matched nonneoplastic mucosal tissue. For patients with lymph node and/or liver metastasis, one or two 1.5 mm cores of metastatic tissue were included. Of the primary gastric cancer specimens, 57.6% (618 of 1,072 cases) were intestinal-type and 42.4% (454 of 1,072 cases) were diffuse-type according to the Lauren classification. Other patient characteristics are listed in Table 1. All of the tissue specimens were obtained for the present study with patient informed consent, and the use of the human specimens was approved by the Changhai Hospital Institutional Review Board.

Immunostaining and evaluation. Consecutive sections (4 μm) of paraffin-embedded tissue microarrays blocks were prepared and processed for immunohistochemical analysis as described previously (7). Total mTOR, p-mTOR, and VEGF protein expression in the sections was detected with appropriate antibodies against total mTOR (dilution, 1:50; clone Y391; Abcam), p-mTOR (Ser²⁴⁴⁸; dilution, 1:100; clone 49F9; CST), and VEGF (dilution, 1:200; clone SP28; Lab Vision). Total mTOR, p-mTOR, and VEGF protein expression in those 1,072 cases was evaluated by two individuals using an Olympus CX31 microscope (Olympus Optical). A semiquantitative scoring system was used as described previously (7). Specifically, a underexpression was defined as no staining or staining positivity in tumor tissue being less than matched normal tissue, a normal expression as staining positivity being similar to matched normal tissue, and an overexpression as staining positivity being higher than normal tissue. Staining was scored independently by two individuals who were blinded to each other's findings. All conflicting calls on scoring were adjudicated by a third individual.

Quantification of tumor microvessel density. For CD34 staining, formalin-fixed, paraffin-embedded tumor tissue sections were processed and stained with a monoclonal goat anti-CD34 (PECAM1-M20; Santa Cruz Biotechnology; 1:100 dilution). For CD31 staining, frozen sections (5 μm thick) of human tumor xenograft specimens were fixed with acetone. For quantification of tumor microvessel density, highly vascular areas were initially identified by scanning tumor sections using light microscopy at low power. Vessels in five high-power fields (×200 magnification: ×20 objective and ×10 ocular) were counted by two independent investigators without knowledge of the patient outcome (double-blinded) as described previously (7).

Western blot analysis. Whole-cell lysates were prepared from human gastric cancer and normal gastric tissue specimens. Standard Western blotting was done using rabbit polyclonal antibodies against human mTOR, p-mTOR, p-AKT (Ser⁴⁷³), and pS6 kinase (Lab Vision) and an anti-rabbit IgG antibody, which was a horseradish peroxidase-linked F(ab′)₂ fragment obtained from a donkey (Amersham). Equal protein sample loading was monitored by probing the same membrane filter with an anti-β-actin antibody. The probe proteins were detected using the Amersham enhanced chemiluminescence system according to the manufacturer's instructions.

VEGF protein measurement. The VEGF protein levels in the culture supernatants were determined using the Quantikine VEGF ELISA kit (R&D Systems), which is a quantitative immunometric sandwich enzyme immunoassay. A curve of the absorbance of VEGF versus its concentration in the standard wells was plotted. By comparing the absorbance of the samples with the standard curve, we determined the VEGF concentration in the unknown samples (7).

Cell lines and xenograft models. The human gastric cancer cell line NCI-N87 was purchased from the American Type Culture Collection, and the SK-GT5 cell line was obtained from Gary K. Schwartz (Memorial Sloan-Kettering Cancer Center). The growths of those tumor cell lines in nude mice were determined as described previously (7).

Endothelial cell tube formation assay. The tube formation assay was done as described previously (7, 15) with a modification of using conditioned medium from either GT5 and N87 cells or rapamycin-treated GT5 and N87 cells.

Statistical analysis. Categorical data were analyzed using χ² statistics tests. Within-group correlations of continuous and ordinal variables were assessed using Pearson's correlation coefficient or Spearman's rank correlation coefficient when appropriate. The Kaplan-Meier method was used to estimate survival rates, and the log-rank test was used to assess survival differences between groups. The Cox proportional hazards model for multivariate survival analysis was used to assess predictors related to survival. In general, statistical modeling with a detection of a hazard ratio of 1.5 with 93% power was used to determine the sufficiency of the sample sizes for various statistical calculations. Analyses were done using the SPSS statistical software program for Microsoft Windows (SPSS). Each experiment was done independently at least twice with similar results; one representative experiment was presented. The significance of the in vitro data was determined using Student's t test (two-tailed), whereas that of the in vivo data was determined using the two-tailed Mann-Whitney U test. In all of the tests, a two-sided P < 0.05 was defined as statistically significant.

Overexpression of mTOR and p-mTOR in patients with gastric cancer. Immunostaining revealed overexpression of total mTOR and p-mTOR in primary tumor tissue compared with that in normal gastric tissue (Supplementary Fig. S1). The results were confirmed by Western blotting using both tumor tissues and cell line cultures (Fig. 1). Increased expression of total mTOR and p-mTOR protein was primarily in the membrane and/or cytoplasm of tumor cells (Fig. 1). The overall rates of overexpression of total mTOR and p-mTOR were 50.8% (545 of 1,072) and 46.5% (499 of 1,072), respectively. We also detected VEGF [59.4% (637 of 1,072)] and proliferating cell nuclear antigen [PCNA; 89.3% (957 of 1,072)] overexpression (Supplementary Fig. S1).

Association of mTOR and p-mTOR overexpression with clinicopathologic factors in patients with gastric cancer. To delineate the clinical significance of mTOR and p-mTOR overexpression, we analyzed the correlations between the overexpression of total mTOR and p-mTOR and clinicopathologic factors. The overexpression of total mTOR protein was significantly associated with tumor location (cardia; P = 0.014), tubular/papillary histology (P < 0.001), well/moderate differentiation (P < 0.001), T₁/T₂ tumors (P = 0.022), stage I/II disease (P = 0.046), and PCNA expression (P = 0.038). However, p-mTOR overexpression was associated with age (>60 years; P = 0.001), lymph node metastasis (P < 0.001), stage III/IV disease (P = 0.001), and PCNA expression (P = 0.002; Table 1). Our data showed that p-mTOR, not total mTOR, significantly correlated with lymph node metastasis.

Differential influence of overexpression of mTOR and p-mTOR on survival in patients with gastric cancer. The cohort consisted of 474 male (70.9%) and 195 female (29.1%) patients with a median age of 59 years (range, 20-86 years). The median cumulative survival duration in patients with resected gastric carcinoma was 30 months. Patients with total mTOR overexpression tumors had a significantly shorter median survival duration than patients without total mTOR overexpression tumors (P = 0.0001). Similarly, patients with p-mTOR overexpression tumors had a significantly shorter median survival duration than patients without p-mTOR overexpression tumors (P < 0.0001; Fig. 2). Our data showed that overexpression of both total mTOR and p-mTOR was associated with decreased survival durations. Additionally, we noticed that all other factors, except patient genders, significantly influenced patient survival in univariate analyses (Table 2), including age (P < 0.0001), tumor size (P < 0.0001), anemia (P < 0.001), differentiation (P < 0.001), gastric wall invasion (P < 0.0001), perineural invasion (P < 0.001), tumor-node-metastasis (TNM) stage (P < 0.0001; Supplementary Fig. S2), lymph node metastasis (P < 0.0001; Supplementary Fig. S3), and VEGF overexpression (P = 0.0067; Supplementary Fig. S4).

Expression of p-mTOR, not total mTOR, as an independent prognostic factor in patients with gastric cancer. In contrast to the univariate analyses, multivariate analysis using the Cox proportional hazards model showed that p-mTOR overexpression, VEGF overexpression, age at diagnosis, lymph node metastasis, differentiation, and TNM stage were independent prognostic factors (Table 2). Furthermore, subgroup analyses of total mTOR, p-mTOR, and VEGF according to TNM revealed that the outcomes of patients with total mTOR overexpression were worse in stages I to III, not in stage IV, than that without mTOR overexpression (Fig. 3). However, patients with p-mTOR overexpression tumors had a worse median survival than patients without p-mTOR overexpression tumors in every stage (Fig. 4). Therefore, p-mTOR appears to be a more sensitive biomarker than total mTOR in predicting patient survival.

Subgroup analyses also revealed that the survival durations were significantly worse in patients of stages III and IV with VEGF overexpression than those of stages III and IV without VEGF overexpression. However, there was no significant difference in survival durations in patients of stages I and II with or without VEGF overexpression (Supplementary Fig. S5). Therefore, the VEGF overexpression status in advanced-stage disease significantly influence patient survival. Additionally, overexpression of VEGF was associated with age (>60 years; P < 0.001), male sex (P = 0.004), tumor location (cardia; P = 0.002), adenocarcinoma (P < 0.001), high/moderate differentiation (P < 0.001), gastric wall invasion (T3-T4; P = 0.03), lymph node metastasis (P = 0.017), PCNA expression (P < 0.001), and p53 expression (P = 0.003; Supplementary Table S1).

Direct correlation of total mTOR/p-mTOR overexpression with VEGF overexpression. To explore the underlying mechanisms of the effects of mTOR activation on gastric cancer pathogenesis, we analyzed the correlation among mTOR, p-mTOR, and VEGF protein overexpression and observed significant correlations among the biomarkers: r = 0.442 for total mTOR and VEGF (P < 0.001; Supplementary Table S2), r = 0.282 for p-mTOR and VEGF (P < 0.001; Supplementary Table S3), and r = 0.242 for total mTOR and p-mTOR (P < 0.001; Supplementary Table S4). The expression of mTOR, p-mTOR, and VEGF also significantly correlated with tumor microvessel density assessment according to CD34 staining (Supplementary Tables S5-S7). These data provided strong clinical evidence for a critical role of mTOR pathway in gastric cancer angiogenesis.

Regulation of gastric cancer angiogenesis by mTOR pathway. To provide potential direct evidence for the critical role of mTOR pathway in gastric cancer angiogenesis, we determined the effect of inhibition of mTOR pathway on angiogenic phenotype of gastric cancer cells in vitro and in animal models. First, treatment of N87 and GT5 cells with rapamycin suppressed mTOR activities (inhibitions of both pS6 and pS6 kinase) and reduced VEGF expression (Supplementary Fig. S6A and B), which was consistent with reduced angiogenic potential as determined using tubulogenesis assay (Supplementary Fig. S6C). Treatment of rapamycin also suppressed tumor growth, metastasis, which was consistent with reduced tumor vessel formation in animal models (Supplementary Fig. S6D-G). These experimental data were consistent with our clinical evidence, supporting the critical role of mTOR activation in gastric cancer angiogenesis and growth and metastasis.

In the present study, we reported a construction of gastric cancer tissue microarray containing 1,072 tumor tissues and matched noncancerous tissues and used both microarray and standard molecular biology and animal models to evaluate the activation and function of mTOR pathway in gastric cancer. We showed that mTOR was frequently activated in gastric cancer. Although both mTOR and p-mTOR overexpression was associated with tumor progression, only p-mTOR overexpression was an independent predictor of survival after resection of primary gastric cancer. Moreover, p-mTOR directly correlated with nodal metastasis and VEGF expression and microvessel density, suggesting a novel molecular basis for the critical role of mTOR activation in gastric cancer development and progression.

Many previous studies from various countries have examined clinicopathologic prognostic factors for gastric cancer after radical gastrectomy. Researchers have regarded controversial factors, including age at diagnosis, tumor size, gastric wall invasion, surgical margins, differentiation, lymph node metastasis, and TNM stage as independent prognostic factors in different reports (25, 31–34). However, they did not reach a consensus concerning which clinicopathologic factors are associated with the optimum prognosis (35–37). In our study, we found that age at diagnosis, lymph node metastasis, differentiation, and TNM stage were independent predictors of survival in the Chinese patients with gastric cancer. However, two factors appeared to have the most prognostic importance not only in our study but also in almost all published studies: TNM stage and lymph node metastasis (33, 34). Particularly, lymph node metastasis directly correlated with overexpression of p-mTOR, not mTOR, strongly suggesting the potential role of mTOR activation in gastric cancer metastasis. This motion was further supported by our in vitro and animal experiments, showing that inhibition of mTOR suppressed metastatic potential of human gastric cancer cells.

Furthermore, researchers introduced biological markers (VEGF, epidermal growth factor receptor, HER-2, etc.) into the Cox regression model, which may have helped improve prediction of the median survival duration in patients after resection of primary gastric carcinoma. Previous studies showed that expression of biological markers such as p53 (38), nm23 (39), CD44 (40), and ErbB2 (41) is predictive of survival. Other studies, however, failed to show a correlation between many of these markers and overall prognosis (8, 37). Molecules in signaling pathways, including mTOR and VEGF (9), are now recognized as potential therapeutic targets. To that end, it is crucial to understand the clinical effects and their underlying mechanisms of mTOR activation and signaling in human gastric cancer development and progression. In our current study, we chose to focus on mTOR and VEGF as potential prognostic factors in patients undergoing curative resection of primary gastric cancer. Evidently, mTOR was expressed in normal gastric, primary gastric tumor, and metastatic tissues; however, this expression was higher in primary gastric tumor and metastatic cells than in normal cells. Our findings were consistent with previous findings, showing overexpression of mTOR, especially p-mTOR, in various types of primary cancers, including gastric cancer (12–16). However, the rates of mTOR and p-mTOR overexpression in these studies varied greatly, ranging from 5% to 64%. The predominant expression pattern observed was an increase in cytoplasmic mTOR or p-mTOR overexpression in cancer cells compared with that in normal cells. Our study using a large cohort of patients has identified overexpression of mTOR as a common event in the development of gastric cancer. Moreover, we found that a different significance of total mTOR overexpression and p-mTOR overexpression. Specifically, total mTOR overexpression was associated with high/moderate differentiation, T₁/T₂ tumors, and stage I/II disease, suggesting that total mTOR mainly contributes to the earlier development of gastric carcinogenesis, whereas p-mTOR overexpression was correlated with lymph node metastasis and advanced-stage disease, suggesting that p-mTOR is particularly important to gastric cancer progression. More importantly, our multivariate analysis showed that overexpression of p-mTOR, but not mTOR, was an independent predictor of survival of gastric cancer. Interestingly, patients with p-mTOR overexpression had a poor outcome at every stage than those patients without p-mTOR overexpression. This finding suggested that mTOR overexpression, especially p-mTOR, is an important event in gastric tumorigenesis and is a useful biomarker for predicting outcome.

Researchers have widely studied VEGF expression in gastric cancer cases. For example, some studies found that VEGF expression was positively correlated with tumor size, depth of invasion, lymphatic and venous invasion, lymph node metastasis, International Union Against Cancer stage, and microvessel density in patients with gastric carcinoma (42, 43) and was a good prognostic factor for this disease (22). In our present study with a large cohort of patients, we found that overexpression of VEGF was significantly correlated with differentiation (high/moderate), lymph node metastasis, gastric wall invasion, age (>60 years), sex (male), tumor location (upper gastric), histology (adenocarcinoma), p53 expression, and PCNA expression. Moreover, VEGF overexpression is an independent prognostic factor for gastric cancer at every stage, which is consistent with previous studies (21, 43). Therefore, our results revealed that VEGF overexpression is associated not only with malignant phenotypes of gastric cancer (such as tumor growth and gastric wall invasion) but also with more general characteristics (such as age, sex, and tumor type), suggesting an important role of VEGF overexpression in gastric cancer development and progression.

Mechanistically, a significant relationship among total mTOR, p-mTOR, and VEGF protein expression further supports the critical role of mTOR signaling in VEGF expression, particularly cap-dependent regulation of VEGF translation (44–48). This notion also was supported by several other studies. For example, anti-mTOR and rapamycin analogues potentially inhibit angiogenesis directly and/or indirectly in human rhabdomyosarcoma (28) and breast cancer models in vivo and in vitro (44). Moreover, Stephan et al. (29) showed that rapamycin, alone and in combination with the anti-VEGF antibody 2C3, strongly inhibited primary and metastatic pancreatic tumor growth in vivo. Better understanding and targeting mTOR/VEGF signaling for gastric cancer prevention and treatment is very important and warrants further investigation.

In summary, mTOR and VEGF were frequently activated in human gastric cancer and their overactivation/overexpression were independent prognostic factors in Chinese patients with gastric cancer, suggesting that mechanisms other than clinicopathologic factors such as tumor stage and lymph node metastasis also influence outcome in patients with gastric cancer, including mTOR activation and/or VEGF overexpression. Our findings also suggest that deregulated mTOR/VEGF signaling could be a promising new molecular target for designing novel preventive/therapeutic strategies to control this malignancy.

No potential conflicts of interest were disclosed.

We thank Don Norwood for editorial comments.