Abstract
Background: Prognosis of renal cell carcinoma (RCC) differs within the same stage and grade. Our aim was to investigate the incidence of COX-2 in primary RCC tumors at different stages according to the occurrence of metastasis, and the impact of this biomarker on the survival of RCC patients. Patients and Methods: The cytoplasmic/membranous COX-2 protein expression was examined by immunohistochemistry in RCC tumors from 102 patients. The patients were divided into those with: no metastasis during 7.5 years' follow-up (nm), no metastasis at the time of nephrectomy but who later developed metastases (lm), and those with metastasis at presentation (pm). The immunoreactivity of COX-2 was classified as none (absent/weak intensity in fewer than 10% of the cancer cells), low (weak intensity in over 10% of the cancer cells) or high immunostaining (strong intensity in the majority of the cancer cells). In addition p53 and Ki-67 immunostaining was also assessed in tumors. Results: Percentages of COX-2 reaction were (no/low/high): 78/16/7 in the nm, 53/28/19 in the lm, 92/8/0 in the pm groups (p=0.014). Median metastasis-free survival was shorter in lm patients with COX-2-negative tumors when compared to those with COX-2-positive ones (15 vs. 46 months; p=0.020). Median overall survival was shorter in pm/lm patients with COX-2-negative tumors when compared to those with COX-2-positive ones (28 vs. 94 months; p=0.027), and with COX-2-negative/Ki-67-positive tumors when compared to COX-2-positive/Ki-67-negative ones (19 vs. 97 months; p=0.004). Findings for patients with COX-2-negative/p53-positive tumors were similar, with shorter survival compared to those with COX-2-positive/p53-negative ones (19 vs. 97; p=0.006). Conclusion: COX-2 protein expression is associated with slow development of metastases, and favourable prognosis in metastatic RCC.
Renal cell carcinoma (RCC) causes approximately 100,000 deaths per year worldwide (1). At diagnosis, 20-30% of patients with RCC have metastases (2). Half of the patients diagnosed with local RCC will suffer from metastases later, and the majority of them within 5 years (2). The expected 5-year survival rate with current treatments for all RCC stages is slightly more than 60% (1), but for metastatic RCC it is less than 15-30% (3-6). The options of systemic therapy for metastatic RCC consist of sunitinib and sorafenib (tyrosine kinase inhibitors) (6), and temsirolimus (mammalian target of rapamycin (mTOR) inhibitor) (6), or interferon-α and interleukin-2 (4, 5). The efficacy of an antiangiogenetic monoclonal antibody, bevacizumab, has also been recently shown (6).
Tumor size and grade are the most important prognostic factors for the survival of RCC patients with a locally confined disease (7, 8). In metastatic RCC, the following independent predictors have been reported: performance status, time to metastasis, number of metastatic sites, prior nephrectomy, involved kidney, and baseline hemoglobin, corrected serum calcium, and serum lactate dehydrogenase levels (3, 4, 9, 10). RCC is a heterogenous disease, and prognosis differs in patients within the same stage and grade (3, 7). p53 and Ki-67 protein expressions have been shown to predict poor overall survival in metastatic RCC (8, 11).
Cyclooxygenase-2 (COX-2), (an isoform of COX enzyme) is an inducible form of an enzyme involved in the first steps of prostanoid, prostaglandins and thromboxane, synthesis (12). COX-2 converts arachidonic acid first into prostaglandin G2, and then by peroxidase activity into prostaglandin H2, a precursor of the prostaglandins (12). COX-2 is undetectable in most normal tissues (13), but it increases in inflammation, immune surveillance, and neoplasia (12-14). COX-2 is highly induced by cytokines, growth factors, oncogenes, and tumor promoters (12). The conversion of procarcinogenes to proximate carcinogens is catalyzed by the peroxidase activity of COX-2 (15). The associations between COX-2 expression and tumor growth, antiapoptotic ability, angiogenesis, and tumor invasiveness, as well as multidrug resistance have been reported (16-18). COX-2 protein has been detected in several human neoplasias including RCC (19, 20), but the significance of COX-2 in RCC biology is still unclear.
In this study, the association of COX-2 protein expression with clinicopathological and histopathological parameters, including p53 and Ki-67 protein expressions, was investigated with emphasis on the prognostic value of COX-2 expression and the development of metastases in RCC patients.
Patients and Methods
Patients, staging, and histology. The study included consecutive tumor samples from 102 nephrectomized patients examined at the Turku University Hospital. Consecutive samples were collected from patients with local RCC treated during 1986-1996, and from patients with metastatic RCC during 1995-2001. As the patients with local disease were expected to be free of metastasis for at least 7.5 years, their samples were from an earlier period than the samples from patients with metastatic disease. The patients of the study were divided into three categories according to the occurrence of metastasis. The patients with M0 staging (pTNM classification system) at primary diagnosis were divided into two subcategories: the first group, with no metastasis (nm), were those patients whose RCC had no metastasis within the follow-up of 7.5 years, and the second group, late metastasis (lm), were those patients whose RCC developed metastasis later (i.e. with metachronous metastasis) after the primary diagnosis during a median follow-up of 24 months (range 2-132 months). The third group, primary metastasis (pm), was formed by those RCC patients who were M1 (pTNM classification system) patients at primary diagnosis (i.e. with synchronous metastasis). There were 45 patients (44%) in the nm group, 32 (31%) in the lm group, and 25 (25%) in the pm group. The patient characteristics are presented in Table I.
Patients with metastatic disease were treated with interferon-α with the schedule described in detail elsewhere (4). During follow-up, radiological evaluations were performed by chest X-ray and abdominal ultrasound at regular 6-month intervals after nephrectomy. During interferon-α treatment, radiological evaluations of the tumor spread included chest and abdominal computed tomography (CT) scan or chest X-ray and abdominal ultrasound at regular 3-4 month intervals. The evaluation of the response followed the criteria of the World Health Organization (WHO) (21).
Histopathological samples were re-evaluated by an experienced pathologist (KOS): the tumors were categorized according to the Heidelberg classification (22) and re-graded according to the WHO classification (23). For T-staging categorization, the 2002 updated UICC pTNM classification system of renal carcinomas was used (24).
Immunohistochemical staining and scoring of COX-2. From archival paraffin-embedded blocks, containing well-preserved cancer tissue, 5-μm-thick sections were cut, deparaffinized with xylene, and rehydrated through a graded series of alcohol. For antigen retrieval, the samples were boiled for 10 min in a microwave oven in 10 mM sodium citrate buffer (pH 6.0). An automated processor (TechMate 500; DAKO Glostrup, Denmark) was used for immunohistochemical staining. Steps were performed in the immunostainer using the avidin-biotin-peroxidase staining methods.
The samples were incubated with commercial mouse monoclonal COX-2 antibody (DAKO) diluted to 1:100 for 27 min. The intensity of COX-2 staining was defined for each tumor slide. The immunoreactivity was divided into three classes: absent or weak intensity in fewer than 10% of the cancer cells was classified as no staining; weak intensity in over 10% of the cancer cells was classified as low staining; and strong intensity in the majority of the cancer cells was classified as high staining (19, 25, 26). Additionally, peritumoral inflammation as an immunological effect was analyzed: the extent of peritumoral lymphocyte infiltration was counted as 0, none; 1, mild; 2, moderate; or 3, severe (27). The samples were also incubated with commercial monoclonal p53 antibody clone DO-7 (DAKO) diluted to 1:300, and commercial monoclonal Ki-67 antibody (MIB-1; DAKO) to 1:100 for 27 min. The intensity of p53 and Ki-67 staining was defined for each tumor slide. Immunoreactivity of p53 and Ki-67 was evaluated by counting the percentage of carcinoma cells exhibiting p53 and Ki-67 nuclear staining. Immunoreactivity was classified as continuous data from undetectable levels (0%) to homogeneous (100%) for both markers. To avoid false positivity, the reaction was considered positive when 10% or more of the carcinoma cells showed staining. The 10% cut-off value was selected according to a previous study on the subject (8). Staining without the primary antibody served as negative control. No significant background staining was detectable. The reliability of staining was measured by standard positive controls used as weekly standard controls in the routine pathology laboratory.
Immunohistochemistry of COX-2, p53 and Ki-67 was scored as a consensus of two investigators (MKT, KOS). The pathologist (KOS) also analyzed the reaction of peritumoral inflammation.
Statistical analysis. The associations of patient characteristics and COX-2 expression were evaluated using cross-tabulation, χ2 or Fisher's exact test, and one-way analysis of variance (age). The Cox proportional hazards model was used to analyze the association of the clinicopathological variables, p53, Ki-67 and COX-2 expression with metastasis-free and overall survival. Variables significantly associated with survival in univariate Cox models were included in the multivariate Cox model. The results were quantified by calculating hazard ratios (HR) with 95% confidence intervals (95% CI). Survival curves were calculated using the Kaplan-Meier method. The differences between the curves were tested with log-rank test.
The association of COX-2, p53 and Ki-67 for the group with metastasis was analyzed using univariate and multivariate multinomial logistic regression. The results were quantified by calculating odds ratios (OR) with 95% CI (28). In all tests, p-values less than 0.05 were considered statistically significant. Statistical calculations were performed using SAS System for Windows, release 8.02/2001 (SAS Institute, Cary, NC, USA).
Results
Immunostaining of COX-2 protein in primary RCC tumors was analyzed in 102 RCC specimens. The staining was positive in 27 specimens (low in 18 and high in 9, respectively). When the patients were divided into three groups according to the occurrence of metastasis, the distribution of COX-2 immunoreactivity (no/low/high) (%) was as follows: 78/16/7 in the nm group, 53/28/19 in the lm group and 92/8/0 in the pm group (p=0.014). Therefore, COX-2 expression was positive in 22% of nm patients, 47% of lm patients, and in only 8% of pm patients. The peritumoral lymphocyte infiltration was not associated with COX-2 expression or metastasis category.
The associations between COX-2 expression and p53 or Ki-67 expression were not significant (Table I), neither were the associations between COX-2 expression and age, gender, performance status, T-stage (p=0.09, a trend), tumor grade, Heidelberg classification, nor response to interferon-α (Table I).
Comparison of COX-2, p53 and Ki-67 for the appearance of metastasis. The associations of COX-2, Ki-67 and p53 with the appearance of metastasis were compared by multinomial logistic regression analysis. COX-2, Ki-67 and p53 protein expressions were significantly different between the metastasis groups (p=0.006, p<0.001, p=0.043, respectively). p53 and Ki-67 were more common in the pm group than in the nm group (OR=4.5, 95% CI=1.3-15.5, p=0.017 and OR=11.1, 95% CI=3.1-40.4, p<0.001, respectively). Ki-67 was also more common in the pm group than in the lm group (OR=3.9, 95% CI=1.2-12.2, p=0.021). COX-2 was more common in the lm group than in both the pm group (OR=10.1, 95% CI 2.0-50.4, p=0.005), and the nm group (OR=3.1, 95% CI=1.2-8.3, p=0.025). In multivariate analysis, only COX-2, and Ki-67 protein expressions were associated with the appearance of metastases (p=0.007, and p=0.006, respectively), indicating that COX-2 and Ki-67 had stronger associations than p53 with the development of metastasis.
Prognostic value of COX-2 for metastasis-free survival. Metastasis-free survival of the lm patients is shown as a function of COX-2 expression (in Figure 1). The median metastasis-free survival was shorter in patients with COX-2-negative tumors when compared to those with COX-2-positive tumors (15 vs. 46 months) (HR=2.5, 95% CI=1.1-5.3, p=0.020, log-rank test). COX-2 was the only variable to have prognostic value for metastasis-free survival (Table II).
Prognostic value of COX-2 and other variables for overall survival in metastatic RCC. The prognostic value of variables for overall survival in the pm/lm group of patients is presented in Table III. COX-2 negativity was associated with shorter overall survival, when compared to COX-2 positivity (median overall survival time 28 vs. 94 months; p=0.027). The higher the T-stage (T3, T4), the shorter was the overall survival (p=0.012); patients with high T-stage (T3, T4) had a greater than two times higher risk of death than patients with low T-stage (T1, T2). Patients with grade 3 tumors had a 2.5 times higher risk of death when compared to patients with grade 1 (p=0.039). Patients with p53 and Ki-67 negativity showed a trend for longer overall survival (p=0.063 and p=0.068, respectively). Neither Heidelberg classification, sex nor age at nephrectomy were associated with overall survival. In multivariate analysis, only T-stage was an independent variable for overall survival.
COX-2 negativity/Ki-67 positivity was associated with shorter overall survival when compared to COX-2 positivity/Ki-67 negativity (median overall survival time 19 vs. 97 months) (HR=3.5, 95% CI=1.5-8.1, p=0.004) (Figure 2). Additionally, double positivity (HR=0.4, 95% CI=0.1-1.1, p=0.083) and double negativity (HR=0.5, 95% CI=0.3-1.0, p=0.061) of COX-2/Ki-67 was associated with longer overall survival when compared with COX-2 negativity/Ki-67 positivity (Figure 2).
COX-2 negativity/p53 positivity was associated with shorter overall survival when compared to COX-2 positivity/p53 negativity (median overall survival time from nephrectomy 19 vs. 97 months) (HR=3.4, 95% CI=1.4-8.2, p=0.006). However, there was no difference in overall survival with double positivity (HR=0.5, 95% CI=0.1-2.0, p=0.345) or double negativity (HR=0.6, 95% CI=0.3-1.2, p=0.160) of COX-2/p53 when compared to COX-2 negativity/p53 positivity.
Discussion
Elevated COX-2 expressions have been observed in neoplastic cells of canine (29) and human RCC (13, 20). COX-2 enhances cell proliferation, apoptosis resistance and angiogenesis in vivo (30). However, the role of COX-2 in RCC carcinogenesis is not fully known. In this study COX-2 expression was highest in the patients who later developed metastatic disease.
Previously published reports indicate that the proportion of COX-2-positive cells varies in human RCCs (20, 31). In our study, weak intensity of COX-2 staining was considered as COX-2 negativity, which resulted in a lower proportion of positive COX-2 cells than in some other RCC studies (27, 31). For comparison, in the study of Miyata et al. the criterion for positive COX-2 expression was 5% (20) whereas in the present study it was considered as 10%. Different antibodies have also been used in other studies (25, 27, 31, 32). This fact and the criteria for immunohistochemical classification may contribute to the difference in the results.
We determined that metastasis-free survival was longer in patients with COX-2-positive tumors; the median metastasis-free survival was 46 months in RCC with COX-2 positivity compared to 15 months in RCC with COX-2 negativity. Our results indicate that COX-2 positivity seems to be associated with the delay of metastasis in those RCC patients who do not have disseminated disease at presentation.
The proportion of COX-2 positive tumors was highest in the lm RCC group, when compared to both the other groups. Previously, Miyata et al. observed that positive COX-2 expression was associated with primary metastases in univariate analysis, when M0 patients were compared to M1 patients (20). Cho et al. found no association between positive COX-2 expression and metastasis, when M0 patients were compared to M1 patients, or the appearance of metastatic disease was compared to non-metastatic disease (31). In these studies, the method of analysis differs from that of our study, where patients were divided into three categories according to the appearance of metastases.
This study indicates that COX-2 negativity is associated with a more aggressive phenotype in metastatic RCC disease. The majority of patients with metastatic RCC and with COX-2 negativity (23/40) had metastases at presentation while 17 out of 40 patients later developed metastases and had a poor median metastasis-free survival of 15 months (Figure 1). Conversely, of RCC patients with COX-2 positivity and metastasis, only 2/17 had metastasis at presentation, the majority developed metastatic disease later, with a median metastasis-free survival of 46 months. This phenomenom is in line with observation that COX-2 negativity is an adverse indicator of overall survival for patients with metastatic RCC.
Few studies have reported the results of an association between COX-2 expression and survival in RCC patients (Table IV). Miyata et al. found that the 5-year survival of patients with COX-2-positive tumors from nephrectomy was 66%, and that of COX-2-negative patients 91% (20). In their study, 86% of patients were M0 and 14% were M1 at nephrectomy. Previously, no results regarding COX-2 and overall survival in patients with metastatic RCC have been published to our knowledge. In this study, COX-2 positivity predicted better overall survival in patients with metastatic RCC treated with interferon-α. This is in line with a previous study of Rini et al. in which COX-2 positivity was associated with longer time to progression in the patients treated with celecoxib plus interferon-α (25). In the present study, no association was found between COX-2 staining and response to interferon-α alone, while in the small study of Rini et al., samples from all the RCC patients with objective responses for celecoxib plus interferon-α exhibited COX-2 staining (25). Additionally, in the present study, COX-2 was not associated with the Heidelberg classification, which is in line with previous results (14).
Tuna and co-workers found an association between COX-2 and peritumoral inflammation in RCC, and thus suggested that COX-2 may be one of the mechanisms that promote RCC carcinogenesis by inflammation (26). Inflammatory activation has been previously associated with favourable prognosis in RCC and melanoma patients (5, 27, 33). In the present study, it may be that the inflammation activation was minimal in RCC patients with initial metastasis, and therefore the proportion of COX-2-positive patients was low in that group. The extent of peritumoral lymphocyte infiltration was not associated with COX-2 expression, or metastatic category. Inflammatory activation in cancer patients arises in various ways (5, 27, 33, 34), and, currently, tumor immunology is a target of active research.
COX-2 expression seems to vary in different tumor types. The mechanisms that predispose to COX-2 up-regulation in different cancer types are not well understood. In many cancer types, COX-2 up-regulation is associated with progressive disease and poor survival (35, 36). Maaser et al. showed increasing expression of COX-2 from normal squamous epithelium to dysplasia and finally to invasive esophageal squamous cell cancer, suggesting that COX-2 up-regulation is associated with esophageal squamous cell carcinogenesis (36). In addition, contrasting results on the role of COX-2 in the disease progression and overall survival in other types of cancer have been reported (38). The involvement of COX-2 in the development of muscle-invasive transitional cell carcinoma of urinary bladder via carcinoma in situ has been suggested: stronger COX-2 expression in carcinoma in situ cases than in muscle-invasive cases (pT2-4) was observed (39). Covariation of low COX-2 and high epidermal growth factor receptor expression predicted poor relapse-free and overall survival in patients with laryngeal squamous cell carcinoma (40).
COX-2 and Ki-67 alone are stronger biomarkers than p53 for the development of metastasis in RCC. COX-2 negativity in Ki-67-positive tumors seems to predict very poor survival in the patients who developed metastasis. In this study, median overall survival of patients with RCC with COX-2 negativity/Ki-67 positivity was 19 months, which was almost five times shorter than of those with COX-2 positivity/Ki-67 negativity, whereas the median overall survival time of those with COX-2 negativity alone was 28 months, which was three times shorter than of those with COX-2 positivity.
The present finding of no association between COX-2 and p53 is in accordance with previous observations (31). The finding of no association between COX-2 and Ki-67, however differs from previous observations (20). In the present study, p53 and Ki-67 were also compared to COX-2 in terms of appearance of metastases, and in multivariate analysis, COX-2 and Ki-67 were independent variables. Published associations between COX-2 and T-stage or tumor grade in RCC have been contradictory (Table IV). Yoshimura et al. demonstrated that COX-2 expressed highest in G1, as well as in pT1 RCC tumors compared to other grades and stages (14), while in Hashimoto et al.'s study, the results were the opposite, with increased COX-2 expression with higher tumor grade and stage (32). In the present study, no association between COX-2 and tumor grade or T-stage was found.
In conclusion, COX-2 positivity in primary RCC tumor appears to be associated with a delay in metastatic development of RCC. COX-2 positivity points to a favourable prognosis in metastatic RCC and COX-2 positivity in Ki-67-negative tumors is strongly associated with improved survival of patients with metastatic RCC.
Acknowledgements
We express appreciation to Tuula Manninen and Sinikka Kollanus for technical assistance. This work was financially supported by Turku University Hospital (EVO) and The Finnish Medical Society Duodecim.
- Received October 18, 2009.
- Revision received May 21, 2010.
- Accepted May 27, 2010.
- Copyright© 2010 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved