Correlation of Cyclooxygenase-2 Expression With Molecular Markers, Pathological Features and Clinical Outcome of Transitional Cell Carcinoma of the Bladder

0022-5347/03/1703-0985/0 THE JOURNAL OF UROLOGY® Copyright © 2003 by AMERICAN UROLOGICAL ASSOCIATION

Vol. 170, 985–989, September 2003 Printed in U.S.A.

DOI: 10.1097/01.ju.0000080401.85145.ee

CORRELATION OF CYCLOOXYGENASE-2 EXPRESSION WITH MOLECULAR MARKERS, PATHOLOGICAL FEATURES AND CLINICAL OUTCOME OF TRANSITIONAL CELL CARCINOMA OF THE BLADDER SHAHROKH F. SHARIAT,* KAZUMASA MATSUMOTO, JAHONG KIM,† GUSTAVO E. AYALA,‡ JAIN-HUA ZHOU, WEIGUO JIAN, WILLIAM F. BENEDICT AND SETH P. LERNER§ , 㛳 From the Scott Department of Urology (SFS, KM, JK, GEA, WJ, SPL) and Department of Pathology (GEA), Baylor College of Medicine and Methodist Hospital, and Department of Genitourinary Medical Oncology (J-HZ, WFB), University of Texas M. D. Anderson Cancer Center, Houston, Texas

ABSTRACT

Purpose: We investigated the relationship between cyclooxygenase-2 (COX-2) expression and molecular alterations commonly found in transitional cell carcinoma (TCC) of the bladder and determined whether COX-2 immunoreactivity is associated with cancer stage, progression and survival in patients undergoing radical cystectomy. Materials and Methods: Immunohistochemical staining for COX-2 was done in archival tumor specimens from 80 patients who underwent radical cystectomy. Immunoreactivity was categorized as positive (reactivity in greater than 10% tumor cells) or negative. Microvessel density, E-cadherin, pRB, p16, p21, p53 and transforming growth factor (TGF)-␤1 and its receptors (types I and II) were also studied because evidence suggests a biological association between COX-2 and alteration of these molecules. Results: COX-2 was over expressed in 62 patients (78%). COX-2 over expression was associated with muscle invasive pathological stage (p ⫽ 0.022), TGF-␤1 over expression (p ⫽ 0.004), decreased E-cadherin expression (p ⬍ 0.001), and altered expression of pRB (p ⫽ 0.003) and p16 (p ⫽ 0.006). At a median followup of 101 months COX-2 over expression was associated with disease progression (p ⫽ 0.038) and bladder cancer specific survival (p ⫽ 0.042). However, when adjusted for the effects of standard pathological features, only lymph node metastasis was associated with bladder cancer progression (p ⫽ 0.027) and mortality (p ⫽ 0.042). Conclusions: COX-2 is commonly expressed in patients with bladder TCC. Using the cutoff of 10% abnormal COX-2 expression is associated with the degree of invasiveness, alterations in TGF-␤1 and pRB/p16 pathways, and loss of cell adhesion. While COX-2 expression has limited prognostic value in patients with bladder TCC, it may serve as a target for therapy with selective COX-2 inhibitors. KEY WORDS: bladder; bladder neoplasms; carcinoma, transitional cell; cyclooxygenase

Cyclooxygenase-2 (COX-2) over expression has been reported recently in various tumors, including bladder cancer.1– 4 COX-2 is known to contribute to tumorigenesis through multiple and complex mechanisms, including apoptosis inhibition, increased angiogenesis, increased invasiveness, modulation of inflammation and immunity, and conversion of procarcinogens to carcinogens. Recent evidence suggests that COX-2 promotes tumor angiogenesis by releasing and activating proangiogenic proteins, interfering with endothelial adhesion molecules and producing eicosanoid products that directly simulate endothelial cell migration and vasculature dependent solid tumor growth.5 A mechanism for elevated tumor associated COX-2 expression could be related to marked repression of transcription of the COX-2

gene by WT p53.6 A recent study suggested that loss of response to the inhibitory effect of WT p53 leads to increased COX-2 mRNA and protein expression.6 Furthermore, there appears to be an important regulatory interaction between COX-2 and transforming growth factor-␤1 (TGF-␤1) pathways.7 TGF-␤1 has been shown to induce or augment COX-2 expression in various cell lines.7 We have previously reported that altered tissue expression of p53, pRB, p16, E-cadherin, TGF-␤1 and its receptors I (TGF-␤-RI) and II (TGF-␤-RII) are associated with established markers of biologically aggressive transitional cell carcinoma (TCC) and it predicts the clinical outcome in patients treated with radical cystectomy.8 –10 An important target of p53 is p21WAF1, a cyclindependent kinase inhibitor that can arrest the cell at the G1 phase by inhibiting DNA replication. To our knowledge the possible association between COX-2 expression and the immunohistochemical status of these molecules has not yet been investigated in bladder cancer tissue. We determined if COX-2 expression is associated with the expression of other molecular markers, including cell cycle regulators such as p53, pRB, p16 and p21 that have been shown to have a role in bladder cancer, with the adhesion molecule E-cadherin, TGF-␤1 and its receptors I and II, and with the angiogenesis related protein CD34. In addition, we evaluated the association of COX-2 expression with the pathological characteristics and clinical outcome of bladder

Accepted for publication March 7, 2003. Supported by grants from the Frost Foundation, Inc. and Austrian Science Fund, and Grant CA54672 from the National Cancer Institute. * Current address: Department of Urology, University of Texas Southwestern Medical Center, Dallas, Texas 75390-9110. † Current address: Urologic Institute, Cleveland Clinic Foundation, Cleveland, Ohio 44195-5041. ‡ Financial interest and/or other relationship with Pintex. § Corresponding author: Scott Department of Urology, 6560 Fannin, Suite 2100, Houston, Texas 77030 (telephone: 713-798-6841; FAX: 713-798-5553; e-mail: [email protected]). 㛳 Financial interest and/or other relationship with Chevron Phillips Chemical, National Cancer Institute and PhotoCure ASA. 985

986

CYCLOOXYGENASE-2 AND BLADDER TRANSITIONAL CELL CANCER

cancer in a well defined set of 80 patients who underwent radical cystectomy and had long-term followup. MATERIALS AND METHODS

Patient population. We identified 80 patients with TCC of the bladder who had undergone radical cystectomy and pelvic lymphadenectomy, and for whom we had adequate archival material. Histology, grade, stage and the presence of carcinoma in situ were confirmed by blinded review of the original pathology slides by one of us. The 1997 TNM classification was used for pathological staging and the WHO classification was used for pathological grading. No patient was treated preoperatively with radiation or systemic chemotherapy, and none had distant metastatic disease at cystectomy. There were 65 males (81%) and 15 females (19%) with a median age of 65 years (range 41 to 80). We collected formalin fixed, paraffin embedded blocks representing the most invasive areas of each tumor. Median followup was 101 months (range 62.4 to 177.3) for patients alive at the time of analysis. Western blot analysis. Electrophoresis was done in a 10% sodium dodecyl sulfate-polyacrylamide gel using 100 ␮g total protein per lane. After electrophoresis proteins were transferred to a 0.2 mm nitrocellulose membrane (BioRad Laboratories, Hercules, California). After transfer the membranes were incubated overnight in blocking buffer. Subsequently the membranes were incubated with SC-1746 (SCB, Inc., Santa Cruz, California) primary polyclonal antibody against COX-2 at a dilution of 1:1,000 for 2 hours at room temperature and goat antirabbit secondary antibody (BioRad Laboratories) at a dilution of 1:3,000 for 1 hour at room temperature. Immunoreactive bands were visualized by enhanced chemiluminescence (APB, Inc., Piscataway, New Jersey). To confirm antibody activity antiCOX-2 antibody was incubated with the blocking peptide SC-1746p before use. Immunohistochemistry. Antigen retrieval was performed by immersing tissue sections in 0.1 M citrate buffer and microwaving at 800 W for 5 minutes. The tissue was then incubated with SC-1746 primary polyclonal antibody against COX-2 at a 1:100 concentration in a humidified chamber at 4C overnight. Secondary antibody (Vector Laboratories, Inc., Burlingame, California) was applied at a dilution of 1:400. Known strong COX-2 staining colon cancer specimens served as a positive control. Consecutive tumor sections with the primary antibody replaced with a normal rabbit IgG served as negative controls. All slides were reviewed independently by 2 investigators blinded to clinical and pathological data. Multiple sections were evaluated to eliminate staining technique as the reason for grouping. Bladder tumors were classified as normal (no reactivity or a few focally positive cells, that is 1% to 10% of tumor cells) or abnormal (11% or more positive cells). At least 10% immunoreactivity was considered positive since COX-2 is an inducible enzyme. In a preliminary study we assessed the discriminative value of COX-2 as a continuous and as a categorical variable with

serial increments of cutoffs ranging from 5% to 90% COX-2 positive cells with regard to bladder cancer characteristics and prognosis (data not shown). Kaplan-Meier analysis revealed that the 10% cutoff for positive prognosis and negative was the best discriminator for bladder cancer progression and survival. Likewise, evaluation of the association of the continuous variable and each cutoff for COX-2 with clinical, pathological and molecular characteristics showed that 10% was the only cutoff that yielded statistically significant findings (data not shown). We have previously performed p53, pRB, p16, p21, E-cadherin, CD-34, TGF-␤1, TGF-␤-RI and TGF-␤-RII immunohistochemical staining. Staining and scoring protocols for p53 with DO-7 (Dako Corp., Carpinteria, California) antip53 monoclonal antibody,11 for pRB with polyclonal antiRB antibody RB-WL-1, for p16 with clone DCS-50 (Vector Laboratories) monoclonal antip16,10 for p21 with clone Waf1(Ab1), (Oncogene Research, Cambridge, Massachusetts),12 for CD-34 with clone QBEnd10 (Dako Corp.) monoclonal antibody, and for E-cadherin with antiE-CD (Transduction Laboratories, Lexington, Kentucky) monoclonal mouse antibody.13 Protocols for TGF-␤1, TGF-␤-RI and TGF-␤-RII (SBC, Inc.) have been previously described.8 All markers were placed in 1 of 2 categories (altered or normal) by at least 2 investigators blinded to clinical outcome. Statistical analyses. For purposes of analysis tumor pathological stage (P1 or less vs P2 or greater), grade (1 and 2 vs 3) and lymph node status (N0 vs N1 and N2) were evaluated as dichotomized variables. Fisher’s exact test was used to evaluate associations among COX-2, p53, pRB, p16, p21, E-cadherin, TGF-␤1, TGF-␤-RI, TGF-␤-RII, patient gender, pathological stage, pathological grade, presence of carcinoma in situ, lymph node status and lymphovascular invasion. The difference in microvessel density (MVD) between COX-2 positive and negative groups was tested by the Mann-Whitney U test. The Kaplan-Meier method was used to calculate survival functions and differences were assessed with the log rank statistic. Multivariable survival analyses were performed with the Cox proportional hazard regression model. Statistical significance in this study was considered as p ⬍0.05. All reported p values are 2-sided. All analyses were performed with SPSS version 11.0 for Windows (SPSS, Chicago, Illinois). RESULTS

COX-2 expression in normal bladder and bladder TCC. To confirm the specificity of the COX-2 antibody a limited set of tissue samples (2 matched nontumor and carcinoma tissues, and 5 carcinoma tissues) was subjected to Western blot analysis (fig. 1). COX-2 antibody pretreated with the blocking peptide did not react with COX-2 protein from tumor samples (data not shown), demonstrating the specificity of the antibody used. We confirmed COX-2 up-regulation in carcinoma tissues compared with normal paired mucosa. The 4 invasive tumors (1 pT1, 2 pT2 and 1 pT3) showed intense immunore-

FIG. 1. Immunoblot assay of COX-2 in matched samples of bladder transitional cell carcinoma (pTa to pT2) and normal (n) epithelial bladder tissue from 2 patients with muscle invasive disease (pT3 in 1 and pT2 in 1) and in tumor tissue from 3 with pTa, 1 with pT1 and 1 with pT2.

987

CYCLOOXYGENASE-2 AND BLADDER TRANSITIONAL CELL CANCER

active bands of the expected COX-2 molecular weight (approximately 70 kDa), whereas the 3 pTa tumors had undetectable COX-2 protein expression. Figure 2 shows representative results of the immunohistochemistry for COX-2 in cancerous bladder tissues. Within tumor cells most COX-2 immunoreactivity was observed in the cytoplasm. We stained tumor sections from 11 patients twice to confirm the reproducibility of staining results between batches. When scored in blinded fashion to the first batch, the concordance rate for COX-2 status between the 2 runs using the 10% cutoff for positive and negative was 100% (p ⬍0.001). Between the 2 runs the median difference in the percent of cells staining positive for COX-2 was only 12.5%.

Association of COX-2 expression with clinicopathological characteristics and molecular markers. The table lists clinicopathological characteristics of the 80 patients treated with cystectomy and associations with COX-2. The expression of COX-2 protein was associated with muscle invasive disease (p ⫽ 0.022) when 10% COX-2 expression was used as the cutoff. In addition, COX-2 expression was associated with TGF-␤1 over expression (p ⫽ 0.004), decreased E-cadherin expression (p ⬍0.001), and abnormal expression of p16 and pRB (p ⫽ 0.003 and 0.006, respectively). Median CD-34 MVD was 82 (range 22 to 263) at 200⫻ magnification and 24 (range 3 to 81) at 400⫻ magnification. Neither patient age at cystectomy, MVD at 200⫻ or MVD at 400⫻ was statistically different between COX-2 positive and negative cases (p ⫽ 0.432, 0.923 and 0.834, respectively). Association of COX-2 expression with clinical outcomes. Disease progressed in 25 patients and 34 were dead at the time of analysis. Of these 34 patients 24 died of metastatic bladder cancer and 10 died of other causes without evidence of disease progression. At a median followup of 101 months Kaplan-Meier analyses showed that altered COX-2 expression was associated with an increased risk of disease progression and cancer specific mortality (p ⫽ 0.0376 and 0.0421, respectively, fig. 3). On multivariable Cox proportional hazards regression analyses that included pathological stage,

Association of COX-2 expression using 10% cutoff between positive and negative with clinical, pathological and molecular characteristics of 80 patients who underwent radical cystectomy for bladder TCC No. COX-2 Expression (%)

FIG. 2. COX-2 immunohistochemistry in high grade invasive bladder cancer. A, negative stain. B, intense positive cytoplasmic staining in papillary morphology. C, intense positive cytoplasmic staining in solid morphology, respectively. Reduced from ⫻200.

Totals Gender: Female Male Pathological tumor stage: Pa, Pis, P1 P2–P4 Pathological grade: 1⫹2 3 Lymph node status: N0 N1, N2 Lymphovascular invasion: Neg Pos Carcinoma in situ: Neg Pos p53 Expression: Normal Altered p21 Expression: Normal Altered pRB expression: Normal Altered p16 Expression: Normal Altered E-cadherin expression: Normal Altered TGF-␤1 expression: Normal Altered TGF-␤ RI expression: Normal Altered TGF-␤ RII expression: Normal Altered

No. Pts (%)

Normal Abnormal

80 (100)

18 (23)

62 (78)

15 (19) 65 (81)

5 (33) 13 (20)

10 (67) 52 (80)

0.308

18 (23) 62 (77)

8 (44) 10 (16)

10 (56) 52 (84)

0.022

23 (29) 57 (71)

4 (17) 14 (25)

19 (83) 43 (75)

0.567

54 (68) 26 (32)

13 (24) 5 (19)

41 (76) 21 (81)

0.778

56 (70) 24 (30)

15 (27) 3 (13)

41 (73) 21 (88)

0.244

49 (61) 31 (39)

8 (16) 10 (32)

41 (84) 21 (68)

0.109

35 (44) 45 (56)

11 (31) 7 (16)

24 (69) 38 (84)

0.111

41 (51) 39 (49)

13 (32) 5 (13)

28 (68) 34 (87)

0.061

37 (46) 43 (54)

14 (38) 4 (9)

23 (62) 39 (91)

0.003

38 (47) 42 (53)

14 (37) 4 (10)

24 (63) 38 (91)

0.006

23 57

12 (52) 6 (11)

11 (48) 51 (90)

⬍0.001

29 (36) 51 (64)

12 (41) 6 (12)

17 (59) 45 (88)

0.004

46 (57) 34 (43)

11 (24) 7 (21)

35 (76) 27 (79)

0.729

42 (52) 38 (48)

11 (26) 7 (18)

31 (74) 31 (82)

0.436

p Value (2-sided Fisher’s exact test)

988

CYCLOOXYGENASE-2 AND BLADDER TRANSITIONAL CELL CANCER

FIG. 3. Probabilities after radical cystectomy according to COX-2 expression with 10% cutoff between positive and negative. A, progression. B, mortality.

tumor grade, lymphovascular invasion, lymph node status and COX-2 status only lymph node status was associated with bladder cancer progression (OR 3.026, 95% CI 1.033 to 10.559, p ⫽ 0.027) and survival (OR 2.785, 95% CI 1.012 to 9.776, p ⫽ 0.042). DISCUSSION

We found that COX-2 is commonly expressed in bladder TCC. COX-2 over expression was associated with the degree of tumor invasiveness and with altered expression of TGF␤1, E-cadherin, p16 and pRB. While univariable analysis showed that COX-2 over expression was associated with a higher risk of bladder cancer progression and mortality following radical cystectomy, multivariable analysis that adjusted for the effects of standard pathological features showed that it did not retain its predictive value. In accordance with previous studies we found that a high number of bladder TCCs expressed COX-2.1, 2, 4 Of the patients 78% had COX-2 immunoreactivity in at least 10% of tumor cells and 51% had COX-2 expression in at least 60%. We found that using the 10% cutoff between positive and negative COX-2 expression was associated with advanced pathological stage. While 2 studies have shown that COX-2 immunoreactivity is associated with tumor stage,2, 4 1 failed to confirm this finding.3 Compounding this lack of clarity is the fact that groups at these laboratories relied on different antibodies, used widely varying interpretation and stratifi-

cation criteria, and differed in specimen handling and technical procedures. Previous studies have used randomly chosen cutoffs ranging from 5% to 33% to categorize COX-2 immunoreactivity into altered and normal or classified the tumors into ranked groups without attempting to quantify the number of COX-2 expressing tumor cells.1, 3 In the current study 2 investigators blinded to clinical and pathological data determined the percent of cells expressing COX-2 in all specimens, providing internal consistency to the application of the subjective interpretation criteria. In a preliminary evaluation we assessed the discriminative value of COX-2 expression as a continuous variable and as a categorical variable using each cutoff ranging from 5% to 90% COX-2 positive cells with regard to the clinical, pathological and molecular characteristics of TCC and prognosis. These analyses consistently revealed that the categorical variable with 10% as the cutoff was the optimal classification. The inducible character of the COX-2 enzyme, which is normally absent from most cells but can be transiently up-regulated in response to growth factors, tumor promoters or cytokines, supports the choice of this cutoff.14 Altered COX-2 expression was associated with an increased risk of bladder cancer progression and bladder cancer related mortality on univariable but not multivariable analysis. This finding is consistent with that of Shirahama et al, who found no association between COX-2 expression and overall survival on multivariable analysis.4 Indeed, to our

CYCLOOXYGENASE-2 AND BLADDER TRANSITIONAL CELL CANCER

knowledge the prognostic role of COX-2 remains unproven. Most studies that showed an association between COX-2 over expression and poor prognosis on univariable models could not confirm these results on multivariable Cox proportional hazard regression models. However, further studies of the mechanistic implications of COX-2 in bladder cancer progression and the possible application of COX-2 inhibitor to prevent a promotional role of COX-2 in disease progression are warranted. We found that COX-2 over expression was associated with TGF-␤1 over expression but not with altered expression of its receptors TGF-␤-RI and TGF-␤-RII. In various cancer tissues TGF-␤1 expression co-localizes with COX-2.14 It has been proposed that COX-2 acts as a landscaping tumor promoter that regulates the expression of pro-angiogenic factors, including TGF-␤1.5 Reciprocally TGF-␤1 has been shown to induce COX-2 at the cellular level.7 We have previously reported that TGF-␤1 over expression is associated with advanced pathological stage and risk of disease progression in patients undergoing radical cystectomy for bladder TCC.8 However, we also found that the loss of expression of its membrane bound receptors TGF-␤-RI and/or TGF-␤-RII is clinically more important because it is associated with bladder cancer stage, grade, progression and survival. While the biological and clinical significance of the interaction of these molecules in bladder cancer is unclear, we postulate that COX-2 induction is involved in the loss of cell growth inhibition and other procarcinogenic events caused by TGF-␤1 via mechanisms other than TGF-␤-RI and II down-regulation. We also found an association between COX-2 expression and the cell-cell adhesion molecule E-cadherin as well as the cell cycle regulators p16 and pRB but not p53 or p21. In concordance with these findings Tsujii and DuBois observed that COX-2 over expression induces the loss of E-cadherin mediated cell-cell communication in epithelial cells.15 An association of p53 with COX-2 on immunohistochemistry has been reported in various cancers.16 However, our study and studies performed in other cancers failed to detect an association between COX-2 and p53 immunoreactivity.17 In agreement with 1 previous study of gliomas17 COX-2 expression was associated with pRB and p16 tissue expression. COX-2 has been shown to be highly expressed in retinoblastoma.18 In addition, COX-2 over expression in intestinal epithelial cells has been demonstrated to result in the loss of retinoblastoma kinase activity, which is associated with cyclin-dependent kinase 4, conferring resistance to apoptosis.19 We found no association between COX-2 expression and CD34 stained MVD. COX-2 is known to partly mediate its effects through angiogenesis related events, such as increased expression of the pro-angiogenic growth factors, stimulation of endothelial cell migration and inhibition of endothelial cell apoptosis. While some groups have found an association between COX-2 and CD34 expression in epithelial cancers,14 we and others have not.20 Our choice of the angiogenesis related marker and our methodology may have obscured the association between COX-2 and angiogenesis. CONCLUSIONS

We found that COX-2 is commonly expressed in bladder cancer. In addition, COX-2 over expression was associated with pathological and molecular features of biologically aggressive disease, suggesting a role for COX-2 in bladder cancer development and invasion. However, COX-2 expression seems not to add independent prognostic value to standard pathological features in patients who undergo radical cystectomy for bladder TCC. These findings provide a rationale for further evaluation of COX-2 and its downstream signaling pathways, and raise the potential for targeted therapy for bladder cancer.

989

REFERENCES

1. Mohammed, S. I., Knapp, D. W., Bostwick, D. G., Foster, R. S., Khan, K. N., Masferrer, J. L. et al: Expression of cyclooxygenase-2 (COX-2) in human invasive transitional cell carcinoma (TCC) of the urinary bladder. Cancer Res, 59: 5647, 1999 2. Komhoff, M., Guan, Y., Shappell, H. W., Davis, L., Jack, G., Shyr, Y. et al: Enhanced expression of cyclooxygenase-2 in high grade human transitional cell bladder carcinomas. Am J Pathol, 157: 29, 2000 3. Ristimaki, A., Nieminen, O., Saukkonen, K., Hotakainen, K., Nordling, S. and Haglund, C.: Expression of cyclooxygenase-2 in human transitional cell carcinoma of the urinary bladder. Am J Pathol, 158: 849, 2001 4. Shirahama, T., Arima, J., Akiba, S. and Sakakura, C.: Relation between cyclooxygenase-2 expression and tumor invasiveness and patient survival in transitional cell carcinoma of the urinary bladder. Cancer, 92: 188, 2001 5. Tsujii, M., Kawano, S., Tsuji, S., Sawaoka, H., Hori, M. and DuBois, R. N.: Cyclooxygenase regulates angiogenesis induced by colon cancer cells. Cell, 93: 705, 1998 6. Subbaramaiah, K., Altorki, N., Chung, W. J., Mestre, J. R., Sampat, A. and Dannenberg, A. J.: Inhibition of cyclooxygenase-2 gene expression by p53. J Biol Chem, 274: 10911, 1999 7. Sheng, H., Shao, J., Hooton, E. B., Tsujii, M., DuBois, R. N. and Beauchamp, R. D.: Cyclooxygenase-2 induction and transforming growth factor beta growth inhibition in rat intestinal epithelial cells. Cell Growth Differ, 8: 463, 1997 8. Kim, J. H., Shariat, S. F., Kim, I. Y., Menesses-Diaz, A., Tokunaga, H., Wheeler, T. M. et al: Predictive value of expression of transforming growth factor-beta(1) and its receptors in transitional cell carcinoma of the urinary bladder. Cancer, 92: 1475, 2001 9. Byrne, R. R., Shariat, S. F., Brown, R., Kattan, M. W., Morton, R. A., Jr., Wheeler, T. M. et al: E-cadherin immunostaining of bladder transitional cell carcinoma, carcinoma in situ and lymph node metastases with long-term followup. J Urol, 165: 1473, 2001 10. Tokunaga, H., Shariat, S. F., Green, A. E., Brown, R. M., Zhou, J. H., Benedict, W. F. et al: Correlation of immunohistochemical molecular staging of bladder biopsies and radical cystectomy specimens. Int J Radiat Oncol Biol Phys, 51: 16, 2001 11. Shariat, S. F., Weizer, A. Z., Green, A., Laucirica, R., Frolov, A., Wheeler, T. M. et al: Prognostic value of P53 nuclear accumulation and histopathologic features in T1 transitional cell carcinoma of the urinary bladder. Urology, 56: 735, 2000 12. Stein, J. P., Ginsberg, D. A., Grossfeld, G. D., Chatterjee, S. J., Esrig, D., Dickinson, M. G. et al: Effect of p21WAF1/CIP1 expression on tumor progression in bladder cancer. J Natl Cancer Inst, 90: 1072, 1998 13. Shariat, S. F., Pahlavan, S., Baseman, A. G., Brown, R. M., Green, A. E., Wheeler, T. M. et al: E-cadherin expression predicts clinical outcome in carcinoma in situ of the urinary bladder. Urology, 57: 60, 2001 14. Fosslien, E.: Review: molecular pathology of cyclooxygenase-2 in cancer-induced angiogenesis. Ann Clin Lab Sci, 31: 325, 2001 15. Tsujii, M. and DuBois, R. N.: Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase 2. Cell, 83: 493, 1995 16. Rajnakova, A., Moochhala, S., Goh, P. M. and Ngoi, S.: Expression of nitric oxide synthase, cyclooxygenase, and p53 in different stages of human gastric cancer. Cancer Lett, 172: 177, 2001 17. Shono, T., Tofilon, P. J., Bruner, J. M., Owolabi, O. and Lang, F. F.: Cyclooxygenase-2 expression in human gliomas: prognostic significance and molecular correlations. Cancer Res, 61: 4375, 2001 18. Karim, M. M., Hayashi, Y., Inoue, M., Imai, Y., Ito, H. and Yamamoto, M.: Cox-2 expression in retinoblastoma. Am J Ophthalmol, 129: 398, 2000 19. DuBois, R. N., Shao, J., Tsujii, M., Sheng, H. and Beauchamp, R. D.: G1 delay in cells overexpressing prostaglandin endoperoxide synthase-2. Cancer Res, 56: 733, 1996 20. Sumi, T., Ishiko, O., Yoshida, H., Hyun, Y., Nakagawa, E., Hirai, K. et al: Expression of cyclooxygenase-2 in ovarian mature cystic teratomas with malignant transformation. Int J Mol Med, 8: 495, 2001