Abstract
Background/Aim: To evaluate immunohistochemical expression of nuclear factor-kappa B (NFκB), cyclo-oxygenase (COX)-2, and vascular endothelial growth factor (VEGF) and the impacts thereof on clinicopathological tumor features and survival in patients with colorectal cancer. Materials and Methods: Sixty-six patients with colorectal cancer (stage II or III) were enrolled. Results: The positive expression rates of NF-κB, COX2, and VEGF were 62.1%, 51.5%, and 63.6%, respectively. Sixteen tumor samples (24.2%) coexpressed all three markers. Coexpression of all three markers correlated with pTNM III, poor histological grade, larger tumor diameter, and elevated carcinoembryonic antigen level. pTNM III and coexpression of all three markers were independent prognostic factors for cancer-specific and disease-free survival. Conclusion: The combination of NFκB, COX2, and VEGF expression correlated with advanced pathological features and had a prognostic impact on cancer-specific and disease-free survival. These findings suggest that coexpression of three markers may have a synergistic effect on aggressive tumor biology.
Inflammation is closely associated with the development and progression of colorectal cancer. The exact mechanisms by which molecular markers mediate inflammatory pathways and cancer development are still being investigated. Nuclear factor-kappa B (NFκB) is a well-known transcription factor and mediates inflammation-related carcinogenesis by increasing tumor cell proliferation and angiogenesis, inhibiting apoptosis, and promoting tumor invasion and metastasis (1). Cyclo-oxygenase (COX)-2, an inducible enzyme, converts arachidonic acid to prostaglandin and promotes tumor invasion and metastasis in colorectal cancer (2). Vascular endothelial growth factor (VEGF) is involved in angiogenesis and plays a key role in promoting local tumor growth and metastasis (3).
NFκB, COX2, and VEGF may be closely related at the gene level, in that the COX2 promoter has transcriptional regulatory elements for the NFκB and activator protein-1 (AP1) transcription factors. AP1 regulates the downstream target gene VEGF (4). In colorectal cancer tissue, increased NFκB expression correlates with enhanced COX2 expression (5), and constitutive NFκB activation was found to be associated with angiogenesis (6). In addition, it has been observed that COX2 and VEGF were overexpressed at both the protein and mRNA levels in colorectal cancer. COX2 may influence tumor angiogenesis of colorectal cancer via the VEGF pathway (7).
Although altered expression of NFκB, COX2, and VEGF in colorectal cancer has been observed, the exact mechanisms thereof and the relationships between these markers are not clearly understood. To date, there have been few studies investigating coexpression of NFκB, COX2, and VEGF, and their impact on prognosis in colorectal cancer. Thus, this study was designed to evaluate NFκB, COX2, and VEGF expression, and their impact on clinicopathological tumor features and survival in patients with colorectal cancer.
Materials and Methods
Patients. Sixty-six patients who underwent curative resection and adjuvant chemo(radiation) therapy for stage II or III colorectal cancer were enrolled from 2005 to 2006. Patients were excluded if they had preoperative radiation therapy or a history of pelvic irradiation. Of the 66 patients, 39 had stage II disease and the remaining 27 patients had stage III colorectal cancer. All patients were registered in a dedicated prospective database and underwent close follow-up. Patient follow-up lasted until either death or the cut-off date of April 30, 2013. The median follow-up interval was 78 months (range=6-89 months). This retrospective study was approved by the Institutional Review Board of Severance Hospital (No.: 420100151).
Representative expression of nuclear factor-kappa B (A) and cyclooxygenase-2 (B) in the tumor cell cytoplasm, as determined by immunohistochemistry. The intensity score was graded as negative (0) (i), weak (1) (ii), moderate (2) (iii), or strong (3) (iv). Magnification: ×200.
Immunohistochemical (IHC) staining. Tumor samples stored in formalin-fixed and paraffin-embedded tissue blocks were selected. Prepared paraffin-embedded tissue blocks were sectioned to 4 μm, pretreated with xylene, and rehydrated with alcohol. Antigens were retrieved three times using antigen-retrieval buffer (LabVision, Fremont, CA, USA) at 99°C for 5 minutes. The sections were then incubated at 65°C for 20 minutes. Endogenous peroxidase activity was blocked by treating sections with Tris-buffered saline with Triton X-100 (TBST buffer) and 0.3% hydrogen peroxide for 10 minutes. The following primary antibodies were then applied for 2 hours at room temperature: rabbit monoclonal antibody to NFκB (Epitomics, Burlingame, CA, USA), rabbit monoclonal antibody COX2 to (Cell Marque, Rocklin, CA, USA), and rabbit monoclonal antibody to VEGF (Epitomics). The sections were then washed in TBST. Secondary antibodies were subsequently applied at room temperature for 30 minutes, followed by a wash in TBST. Three-amino-9-ethyl carbazole was applied for 5-10 minutes for the peroxidase reaction, and the sections were counterstained with hematoxylin.
Representative expression of vascular endothelial growth factor in the intratumoral stroma, as determined by immunohistochemistry. The intensity score was graded as negative (0) (A), weak (1) (B), moderate (2) (C), or strong (3). Strong (3) expression was not detected. Magnification: ×200.
Assessment of IHC expression. The expression of NFκB and COX2 was determined by examining the tumor cell cytoplasm (Figure 1). VEGF expression was determined by evaluating cancer cells in the intratumoral stroma (Figure 2).
Two pathologists (JKP and ES) evaluated all IHC slides using the weighted-histoscore method. The slides were evaluated for the intensity and percentage of expression by microscopy at ×40, ×100, and ×200 magnifications. Tumor cell intensities were recorded (negative, 0; weak, 1; moderate, 2; and strong, 3). Histoscores were calculated by summing the percentage of cells exhibiting weak intensity plus 2× the percentage of cells exhibiting moderate intensity plus 3× the percentage of cells exhibiting strong intensity. Calculated histoscores were categorized using a 4-point grading system: negative (histoscore=0), weak (histoscore, 1-100), moderate (histoscore, 101-200), and strong (histoscore, 201-300). For statistical analysis, the range of histoscore was classified into two groups: negative (0 to 100) or positive (101 to 300) expression.
Inter-class correlation coefficients (ICCCs) were determined to identify the differences between the two observers (8). An ICCC >0.9 is regarded as an excellent correlation and an ICCC >0.8 as a good correlation (9, 10). The ICCCs comparing the histoscores between the two pathologists were 0.97 for NFκB, 0.999 for COX2, and 0.97 for VEGF (p<0.001).
Statistical analysis. All statistical analyses were performed using IBM SPSS Statistics for Windows, version 20.0 (IBM, Armonk, NY, USA). Categorical variables were analyzed using Chi-squared tests or Fisher's exact tests. A Kaplan–Meier curve with a log-rank test and Cox proportional hazards model were used for survival analysis. A p-value of less than 0.05 was considered to be statistically significant.
Results
Clinicopathological characteristics according to NFκB, COX2, and VEGF expression. Among the 66 samples, the positive expression rates of NFκB, COX2, and VEGF were 62.1% (n=41), 51.5% (n=34), and 63.6% (n=42) respectively. Positive NFκB expression correlated with poor histological grade/mucinous histology (p=0.045). Positive COX2 expression correlated with pTNM III (p=0.002), poor histological grade/mucinous histology (p=0.01), and elevated carcinoembryonic antigen (CEA) level (≥5 ng/ml, p=0.04). Positive VEGF expression showed no correlation with patient age, gender, tumor location, pTNM stage, histological grade, lymphovascular invasion, or microsatellite instability (Table I).
Correlations of immunohistochemical expression. There was a significant correlation between NFκB and VEGF expression (p=0.04). Among the 41 tumors expressing NFκB, 30 samples (73%) had VEGF expression.
Tumors expressing COX2 had increased expression of NFκB (59%) and VEGF (74%) compared to tumors not expressing COX2. However, correlation of COX2 expression with either NFκB (p=0.1) or VEGF expression (p=0.09) did not reach statistical significance (Table II).
Correlation of clinicopathological characteristics with NFκB, COX2, and VEGF coexpression. Sixteen tumor samples (24.2%) coexpressed all three markers (NFκB, COX2, and VEGF). Thirty-four samples (51.5%) coexpressed two markers, and 14 samples (21.2%) expressed only one marker. Two cases showed no positive expression by IHC.
Clinicopathological characteristics according to NFκB, COX2, and VEGF expression.
Coexpression of all three markers correlated with pTNM III (p=0.04), poor histological grade/mucinous histology (p=0.01), larger tumor diameter (p=0.03), and elevated CEA level (p=0.04).
Elevated CEA levels were more commonly observed concurrently with positive NFκB expression (22% vs. 16%), positive COX2 expression (29% vs. 9%) and positive VEGF expression (24% vs. 13%). However, only COX2 expression was statistically significant. Coexpression of all three markers correlated with elevated CEA levels (p=0.04) in the combined analysis. However, significance was not observed in tumors coexpressing only one or two markers (Table III).
Survival analysis. Univariate analysis of cancer-specific survival showed that pTNM III (p=0.03), CEA (p=0.02), COX2 expression (p=0.01), and coexpression of all three markers (p<0.001) were significant risk factors. Multivariate analysis of cancer-specific survival showed that pTNM III [hazard ratio (HR)=2.8, p=0.045] and coexpression of all three markers (HR=4.8, p=0.01) were independent prognostic factors for poorer survival.
Univariate analysis of disease-free survival showed that pTNM III (p=0.004), high CEA level (p<0.001), COX2 expression (p=0.02), and coexpression of all three markers (p<0.001) were significant risk factors of poorer survival. Multi-variate analysis of disease-free survival showed that pTNM III (HR=3.4, p=0.01), elevated CEA level (HR=13, p<0.001), and coexpression of all three markers (HR=3, p=0.03) were independent risk factors of poorer survival (Table IV).
Correlations of immunohistochemical expression of nuclear factor kappa B (NFκB), cyclo-oxygenase-2 (COX2) and vascular endothelial growth factor (VEGF).
Discussion
The major finding of the present study was that coexpression of NFκB, COX2, and VEGF correlated with advanced tumor features, such as pTNM III, poor histological grade or mucinous histology, larger tumor diameter, and elevated CEA level. In addition, coexpression status was an independent prognostic factor in cancer-specific and disease-free survival.
NFκB is an inducible transcription factor and regulates the expression of diverse genes associated with inflammation, apoptosis, and tumor progression (11). In the unstimulated state, the NFκB protein is located in the cytoplasm, bound to a specific inhibitory protein called inhibitor of kappa B (IκB) (12). Phosphorylation of IκB causes release of NFκB proteins from this complex. Freed NFκB moves to the nucleus and induces target gene expression. Overexpression of NFκB has been observed in 47.3% to 86.7% of colorectal carcinomas (13, 14), which is similar to the rate observed in this study (62.1%).
COX is a key enzyme in the synthesis of prostaglandin from arachidonic acid, and anti-inflammatory agents usually inhibit the action of COX. Two isoforms of COX, COX1 and COX2, have been well-characterized. Unlike COX1, which is expressed constitutively, COX2 is an inducible form and plays a role in the pathogenesis of colorectal cancer (15). Enhanced COX2 expression has been found in 42.4% to 72.7% of colorectal cancer tissue (16, 17). In this study, overexpression of COX2 was observed in 51.5% of colorectal cancer samples.
VEGF is a growth factor produced by cells that stimulates angiogenesis. Angiogenesis refers to new blood vessel formation and is a critical step for tumor growth. A small tumor deposit of 1–2 mm in size requires an adequate blood supply for further growth and progression (3). Increased VEGF expression has been reported in 61.5% to 72.4% of colorectal cancer tissue (16, 18). We found overexpression of VEGF in 63.6% of colorectal cancer tissue samples.
Correlation of clinicopathological characteristics with coexpression of nuclear factor-kappa B (NFκB), cyclo-oxygenase-2 (COX2), and vascular endothelial growth factor (VEGF).
Although overexpression of NFκB, COX2, and VEGF has been reported, the mechanisms that regulate their expression in colorectal cancer are not completely understood. NFκB is a major contributing factor for COX2 expression as it regulates the COX2 gene. NFκB-binding sites exist within the COX2 promoter, and NFκB is a positive effector of COX2 induction in response to various cytokine-mediated stimuli (4, 5). Vandoros et al. observed that activation of the NFκB pathway is related to COX2 up-regulation in stromal myofibroblasts in colon cancer, in contrast to levels in normal colon tissue (19). Expression of NFκB correlates strongly with COX2 expression in colorectal cancer by IHC (5, 19). In this study, positive NFκB expression was more commonly observed when COX2 expression was positive, but this correlation did not reach statistical significance. The association between NFκB and VEGF has been previously studied. Sakamoto et al. observed constitutive NFκB activation in cell lines and colorectal cancer tissues (6). Those authors also showed that NFκB activation is involved in angiogenesis using human umbilical vein endothelial cells. Kwon et al. reported that NFκB expression in tumor tissue is associated with VEGF expression and inferior overall survival in patients with stage III colorectal cancer (14). In this study, positive NFκB expression significantly correlated with positive VEGF expression. COX2 and VEGF are involved in neovascularization, invasiveness, and metastatic activity in colorectal cancer (18). It has been shown that COX2 is associated with tumor angiogenesis in colorectal cancer. Wu et al. found that COX2 expression correlated with VEGF expression based on mRNA and protein analysis (7). In this study, although not statistically significant, positive COX2 expression was more commonly found when VEGF expression was also positive.
Multivariate survival analysis using Cox proportional hazard model.
The association between expression of NFκB, COX2, and VEGF, and clinicopathological variables and survival remains controversial. NFκB expression has been reported as related to vascular invasion (14), T stage (20), histological grade (this study), or unrelated to clinicopathological parameters (5). NFκB expression has been reported to be a prognostic factor for overall survival (14), but this was not the case in our survival analysis. COX2 expression has been reported to be associated with TNM stage and survival in some studies [(16, 17, 21) and this study] but not in others (5, 19). VEGF expression has been reported to be correlated with TNM stage and survival (21, 22), whereas others have reported no such correlation [(14, 23) and this study]. An inverse correlation between COX2 or VEGF expression and microsatellite instability (MSI)-high tumor has been observed (24, 25). Although statistically insignificant, we obtained similar findings, namely that an MSI-high phenotype is more frequent when VEGF expression is negative (17%) than when it is positive (7%). In this study, overexpression of one marker alone showed little or no correlation with clinicopathological parameters or prognoses. However, the combination of NFκB, COX2, and VEGF expression is indicative of advanced tumor features and a grave prognosis. Based on these results, it may be postulated that coexpression of NFκB, COX2, and VEGF has a synergistic effect on tumor aggressiveness. Peng et al. investigated the expression of COX2, matrix metalloproteinase-2, and VEGF in patients with stage II and III colorectal cancer (21). Coexpression was found in 32.9% of patients and was significantly associated with TNM stage, CEA level, recurrence, and death. Those authors suggested that coexpression of molecular markers could be used as a surrogate marker for malignant biological behavior of tumors.
CEA is a well-known prognostic factor. In this study, CEA level was a significant risk factor for disease-specific survival in multivariate survival analysis. Based on the combined analysis of three markers, coexpression of three markers correlated well with elevated CEA level. These findings suggest that coexpression of markers represents prognostic significance and, indeed, coexpression status was a significant prognostic factor for both cancer-specific and disease-specific survival.
The main limitation of this study is that it was a single-Center study with a small number of patients. We observed that coexpression of the three markers correlated with advanced disease features, however, we did not evaluate the exact underlying mechanism of these findings. It would be helpful to evaluate mRNA and protein expression in the same cancer tissue. Additionally, the correlation between mRNA and protein expression may reveal the underlying relationship among the three markers. Different methodologies for assessing protein expression, such as quantitative immunofluorescence with different fluorescent dyes, may provide a better correlation among the three markers. However, to the best of our knowledge, this is the first report to evaluate simultaneous expression of NFκB, COX2, and VEGF in cancer tissue.
In summary, the combination of NFκB, COX2, and VEGF expression correlated with advanced tumor features and had a negative prognostic impact on cancer-specific and disease-free survival. In the future, elucidating the role of this coexpression in tumor biology and it role in survival would be useful in the development of novel chemotherapeutic agents or optimal adjuvant treatment strategies for patients with colorectal cancer.
Acknowledgements
The Authors express their special thanks to Professor Ik Yong Kim, who provided statistical support and critically reviewed this manuscript.
Footnotes
-
Conflicts of Interest
Young Wan Kim, Jin Kyu Park, Eunah Shin, and Nam Kyu Kim have no conflicts of interest or financial ties to disclose.
- Received July 11, 2014.
- Revision received September 2, 2014.
- Accepted September 5, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
