Fax: (011) 41 223724920
Expression of cell cycle proteins in T1a and T1b urothelial bladder carcinoma and their value in predicting tumor progression
Article first published online: 3 MAY 2004
Copyright © 2004 American Cancer Society
Volume 100, Issue 11, pages 2367–2375, 1 June 2004
How to Cite
Mhawech, P., Greloz, V., Oppikofer, C., Szalay-Quinodoz, I. and Herrmann, F. (2004), Expression of cell cycle proteins in T1a and T1b urothelial bladder carcinoma and their value in predicting tumor progression. Cancer, 100: 2367–2375. doi: 10.1002/cncr.20306
- Issue published online: 18 MAY 2004
- Article first published online: 3 MAY 2004
- Manuscript Accepted: 23 MAR 2004
- Manuscript Revised: 20 FEB 2004
- Manuscript Received: 26 NOV 2003
- cell cycle protein;
- prediction of progression
Cell cycle proteins are important markers in predicting tumor behavior in urothelial carcinoma of the bladder. The objectives of this study were 1) to determine the expression levels of some of those markers in a series of patients with bladder carcinoma, 2) to define their value in distinguishing T1a (minimally invasive) from T1b (invasive) tumors, and 3) to evaluate their use as predictive factors in the progression of T1a and T1b tumors.
Tumor specimens from 101 patients were included (22 Ta specimens, 34 T1a specimens, 15 T1b specimens, and 30 T2 specimens). A tissue microarray from the 101 paraffin embedded tissue blocks was constructed. Immunohistochemistry for p16, p27, p21, p53, cyclin D1, and Ki-67 were performed. To evaluate T1a and T1b tumor progression, clinical and follow-up data were available for all 49 patients.
Cyclin D1 and p27 were the only markers that showed a significant association with tumor stage and tumor grade (cyclin D1: P = 0.002 and P > 0.00, respectively; p27: P = 0.024 and P = 0.031, respectively). The results indicated that a combination of p21 (odds ratio, 5.7; 95% confidence interval [95% CI], 1.3–24.8 [P = 0.022]) and p16 (odds ratio, 3.7; 95% CI, 0.8–16.5 [P = 0.081]) may have potential use in distinguishing T1b tumors from T1a tumors. Finally, none of the markers examined were found to have predictive value for T1a and T1b tumor progression.
The expression of cyclin D1 and p27 was associated with the most important prognostic factors (tumor stage and grade). The combination of p21 and p16 may have value in distinguishing T1b tumors from T1a tumors, although this finding must be evaluated in much larger series. Finally, none of the markers studied appeared to have predictive value for disease progression in patients with T1a and T1b urothelial bladder tumors. Cancer 2004. © 2004 American Cancer Society.
Urothelial bladder carcinoma is the fourth leading malignancy in men and the seventh cause of cancer-related deaths in the U.S.1 The majority of newly diagnosed patients present with superficial tumors, including 70% noninvasive tumors (pathologic Ta [pTa] tumors) 30% superficially invasive tumors (pT1). The tumor classification still is considered one of the most significant variables for predicting the outcome of patients with urothelial bladder carcinoma and, consequently, for patients management. Although Ta tumors have high likelihood of recurrence and display less propensity to progress, T1 tumors have a greater rate of progression, reaching 30% and even 50% in some instances.2 The identification of the muscularis mucosae (MM) by Dixon and Gosling in 1983 had a great impact on patient prognosis and management.3 The MM, which is comprised of scattered, muscular fibers running along large blood vessels, has changed T1 bladder staging and led to its classification into three groups and, subsequently, into two groups: T1a (minimally invasive) tumors (i.e., tumors that extend into the lamina propria but are located above the level of the MM), and T1b (invasive) tumors (i.e., tumors that invade beyond the MM). Furthermore, it has been proven that this subclassification has prognostic value. Numerous studies have shown that T1b tumors behave more aggressively and have a greater probability of progressing to muscle invasion and metastasis compared with T1a tumors.4–8 Based on this behavior, some clinicians recommended a conservative attitude for patients with T1a tumors and more aggressive treatment, such as a radical cystectomy, for patients with T1b tumors.9, 10
Despite the advances made in understanding the molecular biology of urothelial bladder carcinoma, T1 tumors still present a great challenge for both pathologists and clinicians. First, regarding histologic diagnosis, the interpretation of the depth of tumor invasion within the T1 classification often is very difficult and has some limitations, mainly due to the nature of the resection material itself. Some of these limitations are misorientation, cautery artifact, or necrosis, which are the leading causes of interpretation subjectivity and, thus, interpathologist variation and disagreement.11, 12 Second, with regard to patient management, prognostic factors that predict the behavior of T1a or T1b tumors are not yet available. Thus, criteria to identify patients who are at greater risk of invasion and thus to manage their disease more aggressively are not feasible.
The cell cycle is controlled by a complex of inhibitor and stimulator proteins. Deregulation of these proteins is a common finding in malignancies, leading to uncontrolled tumor cell proliferation. Most of those cell cycle proteins have been studied in numerous carcinoma types using immunohistochemistry. The expression of some cell cycle proteins appears to have significant value in predicting recurrence and disease progression in patients with urothelial bladder carcinoma.13 Despite the large amount of literature using cell cycle proteins to compare T1 tumors with, for instance, Ta and T2 tumors, to our knowledge only a few reports are available that discuss subgroups of patients with T1a and T1b tumors.14–17
The objective of the current study was re to evaluate the expression of cell cycle proteins such as p16, p27, p21, p53, cyclin D1, and Ki-67 in T1a and T1b tumors using immunohistochemistry in an attempt to answer some of the following questions: 1) Is there any association of these proteins with the two most important prognostic factors, tumor grade and stage? 2) Can we explain the aggressive behavior of T1b tumors by the pattern of expression of these six markers? Thus, if there is any value of those markers in distinguishing T1a tumors from T1b tumors, then we could overcome one of the most difficult tasks for pathologists. 3) Finally, and most important, do these proteins have value in predicting progression of T1a and T1b tumors, so that patients who have a greater risk of disease progression and metastasis may be identified?
MATERIALS AND METHODS
Patient and Tissue Selection
The study was comprised of a selection of patients with T1 tumors who were seen in the study institution over 3 years (1997–2000). Patients with a first time diagnosis and adequate follow-up were considered for inclusion. The specimens were comprised of transurethral resection (TUR) samples when abundant material was available. Thus, patients with small tumor samples were excluded from the study. Only specimens with very good histologic orientation, including the identification of MM in the lamina propria and the presence of muscularis propria, were considered for analysis. Most specimens were chosen from our previous study.18 Tumor grade and stage were reviewed by one pathologist (P.M.). Tumor grading was based on the 1999 World Health Organization classification system, and specimens were stratified in two groups: low-grade lesions and high-grade lesions.19 Follow-up data were retrieved from the patient files. Progression was defined as disease progression to higher tumor stage or metastasis, with the progression confirmed by histology. Treatment was comprised of repeated TUR, TUR plus bacillus Calmette–Guerin intravesical therapy, and cystectomy. Cystectomy was performed in patients in whom conventional treatment had failed or the disease had progressed to invade the muscularis propria. Patients with pTa and pT2 tumors at first diagnosis also were included in the current study to explore the correlation of cell cycle proteins and Ki-67 with tumor grade and stage.
Paraffin embedded tissues were used. The tissue microarray was constructed as described previously by Kononen et al.20 After carefully choosing the morphologically representative region on the individual paraffin embedded urothelial bladder carcinoma blocks (donor blocks), a core tissue biopsy measuring 1.6 mm was punched and transferred to the donor paraffin embedded block (receiver block). To overcome tumor heterogeneity, five punch biopsies were performed from different areas of each tumor. Sections measuring 4-μm thick were cut from the tumor microarray block and transferred to glass slides using the paraffin sectioning aid system (adhesive-coated slides, adhesive tape, and an ultraviolet lamp; Instrumedics Inc., Hackensack, NJ). One section was stained by with hematoxylin and eosin to evaluate the presence of the tumor (see Fig. 1).
Tissue sections were deparaffinized with xylene and washed with ethanol. The characteristics of the antibodies used are summarized in Table 1. After the pretreatment of the tissue sections, they were incubated with the appropriate antibody. Endogenous peroxidase was blocked with 0.3% hydrogen peroxidase for 5 minutes. Then, the sections were incubated with mouse envision horseradish peroxidase for 30 minutes. Diaminobenzidine complex was used as the chromogen. In negative controls, a normal goat serum was used instead of the primary antibody, resulting in a lack of detectable staining. All five punch biopsies were scanned. The percentage of positive stained nuclei were evaluated semiquantitatively by two pathologists (P.M. and I.S.-Q.) who were not aware of the original histologic diagnosis or the clinical data. For p53 expression, a cut-off value of > 10% positive nuclei was considered positive, based on well established data.14, 21, 22 For all of the markers, samples were stratified into 3 groups, depending on the percentage of positive cells, as follows: < 10%, > 10–50%, and > 50% positive nuclei. In a second phase, after the first statistical analysis showed no difference in the results between these 3 predefined cut-off values, we defined < 10% as negative and ≥ 10% as positive.
|p27, Kip 1 (monoclonal)||G173-524||Microwave (10)||1:80||Pharmingen (Hamburg, Germany)|
|p53 (monoclonal)||DO-7||Citrate (3)||1:400||Dako (Glostrup, Denmark)|
|p21, Cip1 (monoclonal)||4D10||Citrate (3)||1:20||Novocastra (Newcastle upon-Tyne, United Kingdom)|
|Cyclin D1 (monoclonal)||P2D 11F11||Citrate (5)||1:10||Novocastra|
|Ki-67 (monoclonal)||MIB-1||Microwave (10)||1:50||Novocastra|
For statistical analysis, the following baseline variables were considered for their prognostic value: patient age at presentation; gender; tumor stage and grade; tumor multifocality; treatment modality; presence of carcinoma in situ; and p16, p27, p21, p53, cyclin D1, and Ki-67 expression levels (considered as binary variables). Associations between categorical variables were studied using either the chi-square test or the Fisher exact test, as appropriate. The date of diagnosis was considered the time of origin. For progression-free intervals, the first occurrence of disease progression was considered an event. Univariate and multivariate logistic regression models were defined to predict differences between T1a tumors and T1b tumors and to predict the first occurrence of disease progression while adjusting for all baseline variables. Pseudo-R2 statistics were used to estimate the amount of variance of the outcome variable explained by the different models. Multivariate analyses based on Cox proportional hazards model were used to identify the most significant factors related to disease progression. The results of the multivariate analyses are expressed in terms of either odds ratios (ORs) or hazard ratios, which were derived from the estimated regression coefficients along with their 95% confidence intervals (95% CIs). The results are presented as the mean ± standard deviation (SD) or the median ± interquartile range. P values ≤ 0.05 were considered statistically significant. All analyses were performed with the Stata statistical software package (version 8.1; Stata Corporation, College Station, TX).
In all, 101 patients who were included in the study had 30 T2 tumors, 22 Ta tumors, and 49 T1 tumors. The patients ranged in age from 46–92 years (mean ± SD, 70.3 ± 11.4 years; median ± interquartile range, 70.0 ± 18.0 years). There were 84 males and 17 females, with a male:female ratio of 5.0. Sixty-two tumors were low-grade lesions, and 39 tumors were high-grade lesions. To determine whether p16, p27, p21, p53, cyclin D1, and Ki-67 were associated with tumor progression, we focused on the 49 patients with T1 tumors, for whom the clinical findings and follow-up data are summarized in Table 2. Among 49 T1 tumors, 34 were T1a tumors and 15 were T1b tumors. Nineteen tumors were multifocal (13 T1a tumors and 6 T1b tumors), and 30 tumors were unifocal (21 T1a tumors and 9 T1b tumors). Follow-up ranged from 3–77 months (mean ± SD,19.0 ± 16.5 months; median ± interquartile range, 12.0 ± 15.0 months). Progression to muscular invasion was observed in 11 of 15 T1b tumors and in 2 of 32 T1a tumors. The treatment was as follows. Thirty-six patients underwent transurethral resection (26 with T1a tumors and 10 with T1b tumors), 7 patients underwent transurethral resection followed by intravesical therapy (6 with T1a tumors and 1 with T1b tumors), and 6 patients underwent radical cystectomy for disease progression or for disease that was nonresponsive to therapy (2 with T1a tumors and 4 with T1b tumors). One patient with a T1b tumor died from unrelated causes.
|Classification (no. of patients)|
|Grade (no. of patients)|
|Gender ratio (male:female)||44/5|
|Mean ± SD||70.3 ± 11.4|
|Mean follow-up (mos)||19.0|
|Disease progression (no. of patients)|
|Therapy (no. of patients)|
|pT1a (n = 34 patients)||26|
|pT1b (n = 15 patients)||10|
|TUR and BCG|
|pT1a (n = 34 patients)||6|
|pT1b (n = 15 patients)||1|
|pT1a (n = 34 patients)||2|
|pT1b (n = 15 patients)||4|
There was a significant association between tumor stage and grade (P < 0.0001), as expected. The immunohistochemistry results are summarized in Table 3. Only cyclin D1 and p27 were associated significantly with lower grade (P = 0.002 and P = 0.024, respectively) and lower stage (P ≤ 0.001 and P = 0.031, respectively). The other markers (p16, p21, p53 and KI-67) showed no association with tumor grade or stage.
|Marker||Tumor grade||Tumor stage|
|Low||High||Fisher P value||Ta||T1a||T1b||T2||Fisher P value|
|No. of patients||62||—||39||—||—||22||—||34||—||15||—||30||—||—|
To identify the value of the markers used to distinguish T1a tumors from T1b tumors, we focused only on the 49 patients with T1 tumors (34 with T1a tumors and 15 with T1b tumors). The results are summarized in Tables 3 and 4. Applying logistic regression to each marker, only p21 was associated significantly with an increased risk having a T1b tumor instead of a T1a tumor (OR, 3.60; 95% CI, 1.01–12.81 [P = 0.048]). However, none of the other markers (p53, p16, p27, cyclin D1, and Ki-67) were found to be significant for distinguishing T1a tumors from T1b tumors. When all the markers were included together in a logistic regression model, only p16 emerged as significant (OR, at 6.26; 95% CI, 1.02–38.49; [P = 0.048]). Then, by performing both forward and backward stepwise logistic regressions, we found that only 2 markers (p21 and p16) were able to distinguish these 2 substages (p21: OR 5.7; 95% CI, 1.3–24.8 [P = 0.022] and p16: OR, 3.7; 95% CI, 0.8–16.5 [P = 0.081]) (see Table 4). It is interesting to note that, even if p16 (pseudo R2 = 2.3%) is not statistically significant, it does double the predictive ability of p21, from a pseudo-R2 of 6.7% when considered alone to 12.3% when both markers are included in the models.
|Marker||Crude||Adjusted full model||Adjusted with p21 and p16|
|OR||95% CI||P value||OR||95% CI||P value||OR||95% CI||P value|
With regard to predicting tumor progression, in our study, all six markers failed to predict disease progression in patients with T1a tumors and T1b tumors. In addition, to investigate the value of the clinical data in predicting progression, gender had to be removed from the prediction of disease progression, because none of the five women developed progressive disease. Among the clinical parameters investigated, such as age, multifocality, presence of carcinoma in situ, treatment modality, tumor grade, and depth of invasion in the lamina propria, the only parameter that was found to be predictive of tumor progression in patients with T1a and T1b lesions was the depth of tumor invasion (P < 0.0001) (Table 5).
|Characteristic||Model 1: Prediction of disease progression||Model 2: Prediction of time to disease progression|
|OR||95% CI||P value||HR||95% CI||P value|
|Carcinoma in situ||4.98||0.14–171.67||0.374||19.96||0.75–534.71||0.074|
|Pathologic Stage 1b||62.21||55.1–701.92||0.001||224.18||6.49–7740.33||0.003|
To our knowledge, the current study is one of the very few available in the literature discussing tumor markers in T1a and T1b lesions. This substaging for patients with urothelial carcinoma of the bladder is one of the most challenging problems facing the pathologist, who must make a precise diagnosis, and the clinician, who must make decisions regarding the patient's management. In the literature, we found four scientific articles that discussed different aspects of this subject. Three reports have determined the value of p53 expression using immunohistochemistry staining, and all three reached the same conclusion: The overexpression of p53 does not have any significant value in predicting tumor behavior in T1a and T1b lesions.14–16 The fourth study by Santos et al. included 56 superficial, low-grade urothelial carcinomas (14 Ta tumors, 42 T1 tumors, 35 T1a tumors, and 7 T1b tumors), and assessed expression levels of p53, p27, p16, Rb, and Ki-67.17 Their study showed that only Ki-67 was useful in predicting recurrences and disease progression in patients with superficial, low-grade bladder tumors. To our knowledge, the current study is the first to evaluate five cell-cycle proteins plus Ki-67 in a series of 49 patients with well documented staging and follow-up data. In addition to the predictive value of these cell markers in tumor progression, the study presented herein is to our knowledge the first to raise a different issue (i.e., finding the value of these markers in distinguishing T1a tumors from T1b tumors).
We found a strong association between cyclin D1 expression and two powerful prognostic variables: tumor grade and stage. That is, we observed that low cyclin D1 expression was more frequent in low-grade and low-stage lesions, which is in agreement with other studies.23, 24 In addition, we also demonstrated a significant association between p27 expression and tumor grade and stage. p27 expression was observed more frequently in low-grade and low-stage tumors, which also is in agreement with the literature.25
With regard to distinguishing T1a lesions from T1b lesions, the results of the current study showed that p21 was the only marker that was useful: In various combinations of these 6 markers, the p16-positive/p21-positive pattern of expression increased the value of p21 even more in distinguishing T1b tumors. Unfortunately, this was true only in the context of our model and was not valid if it was used in a routine setting as a discriminator marker in an individual patient. However, the combination of these two proteins may have potential use in the future if these results are confirmed in a much larger series.
Although tumor recurrences are an important issue, disease progression is a more serious problem and requires more aggressive treatment. Cell cycle proteins are important in the regulation of cell proliferation. In addition, they appear to be significant markers for predicting tumor behavior in urothelial bladder carcinomas,26, 27 which raises the third question: whether these proteins have any value in predicting tumor progression. Loss of parts of chromosome 9q and/or 9p is observed frequently in different types of carcinoma. p16 is located on the 9p21 locus and, by inhibiting cyclin-dependent kinase (CDK), plays an important role in the G1 cell cycle checkpoint. Numerous investigators have studied p16 expression in urothelial bladder carcinoma. The value of the loss of p16 in predicting tumor recurrences in superficial urothelial carcinoma remains controversial.28–30 Some investigators have demonstrated that p16 loss is a predictive factor for recurrence, whereas others failed to do so. However, to our knowledge its value in tumor progression has been evaluated only rarely. The results of the current study demonstrated that p16 has no prognostic value in predicting tumor progression, a finding that is in agreement with the findings of Santos et al.17
The p53 gene is a tumor-suppressor gene located at 17 p13.1 that regulates DNA replication, cell proliferation, and cell death. When DNA is damaged, p53 switches off replication, allowing cells the time to repair their damage. With the loss of p53 by mutation or deletion, as documented in various types of carcinoma, the function of p53 is compromised, leading to genomic instability and contributing to clonal evolution and tumor progression. The predictive value of p53 for tumor recurrence and progression in superficial bladder carcinoma remains controversial: Some studies reported that p53 showed good predictive value, and others did not.21, 22, 31 We did not find any value for p53 accumulation in predicting T1 tumor progression, which is in agreement with all four studies mentioned earlier.14–16
Cyclin D1 has a central role in the control of the cell cycle. It is essential for the G1/S-phase transition and works as a positive regulator of the G1 checkpoint of the cell cycle. It is an independent prognostic factor for survival in patients with muscle-invasive carcinoma.32 In superficial bladder carcinoma, the value of cyclin D1 has been somewhat controversial: Some have reported that overexpression of cyclin D1 was associated with early tumor recurrences and tumor progression, and others found no such association.33–35 The results of the current study showed no value for cyclin D1 expression in predicting tumor progression in T1a and T1b tumors. Unfortunately, we could not compare our results with others because of the lack of similar studies in the literature. p27 belongs to the CIP/KIP family, which comprises cyclin-CDK complex inhibitors. Reduced expression of p27 has been reported in numerous types of carcinoma and in urothelial bladder carcinoma. Loss of p27, combined with other cell cycle protein abnormalities, has been associated with aggressive tumor and an unfavorable prognosis in patients with bladder carcinoma.36, 37 In the current study, we failed to demonstrate a value for p27 in predicting tumor progression in T1a and T1b tumors, which is in agreement with the findings of Santos et al.17
Simlar to p27, p21 belongs to the CIP/KIP family; and, as discussed earlier, p21 is a universal inhibitor of the cyclin-CDK complex and is capable of arresting the cell cycle at the G1 phase. Loss of p21 has been observed in 33–64% of invasive bladder tumors and in 50–60% of superficial bladder tumors. Loss of p21 has been associated with aggressive tumor behavior and is a predictor of tumor progression.38, 39 In patients with superficial bladder carcinoma, loss of p21 has not been associated with tumor recurrence.31 However, its value in predicting tumor progression in T1 tumors has not been explored fully. To our knowledge, the current study is the first to show that p21 has no clear value in predicting tumor progression in T1a or T1b tumors. With an increase in the sample size, p21 may be associated with the time to tumor progression (P = 0.054); however, this marker will not be as good a predictor as tumor depth.
Ki-67 is a nuclear protein that is expressed in all stages of the cell cycle except in G0 phase. Ki-67 antigen expression, as detected by the monoclonal MIB-1 antibody on paraffin fixed tissue, showed that cell proliferative activity correlates with tumor growth in different types of carcinoma. Ki-67 expression has been correlated with tumor grade; however, the value of Ki-67 in predicting tumor recurrences in superficial tumors remains controversial. Although Vorreuther et al. found that Ki-67, in combination with other cell cycle proteins, may provide information regarding tumor recurrence in superficial bladder tumors,39 Pfister et al. showed no value of Ki-67 expression alone or in combination with other cell cycle proteins in predicting tumor recurrence in superficial bladder tumors.31 In the current series, we did not find that Ki-67 had any value in predicting tumor progression. This is in disagreement with the results of Santos et al., who found that Ki-67 was a reliable marker for predicting tumor recurrence and progression in patients with superficial tumors (Ta, T1a, and T1b). However, Santos et al. analyzed a selection of low-grade lesions, and approximately one-third of the tumors in their report were Ta tumors; consequently, the combination of Ta and low-grade tumors make tumor progression less frequent and may bias the value of Ki-67.
Among all of the clinical parameters investigated, such as age, gender, tumor multifocality, therapy modality, presence of carcinoma in situ, and depth of tumor invasion in the lamina propria, only the depth of invasion had predictive value for tumor progression. This finding has been well documented in previous studies, thus, indicating the importance of accurate tumor staging by the pathologists.15, 16, 40 However, the value of the other parameters in predicting tumor progression may be hampered by the administration of intravesical therapy, which may change the natural course of the tumor.
To our knowledge, the current study is the first to evaluate the simultaneous expression of all six markers (cyclin D1, p27, p21, p16, p53, and Ki-67) in T1a and T1b tumors in a well documented series with good follow-up. Our conclusions are as follows: 1) There is a significant association of cyclin D1 and p27 expressions with tumor stage and grade. 2) Expression of p21 can distinguish T1b tumors from T1a tumors, and the p21-positieve/p16-positive expression pattern renders the value of p21 more significant in predicting T1b tumors. However, this finding must be evaluated in larger series before it can be recommended for use in the future. 3) There was no value found for any of the six markers alone or in combination with the others in predicting disease progression in patients with T1a and T1b tumors. 4) Finally, among all of the clinical parameters evaluated, only the depth of invasion had independent predictive value for tumor progression.
The author thank Mr. Ludovic Metral for his help in searching the archives and the histology laboratory staff, especially Ms. Danielle Fontana for her technical assistance. They also thank Dr. Sophia Taylor for her review of the article.
- 3Histology and fine structure of the muscularis mucosae of the human urinary baldder. J Anat. 1983; 13: 265–271., .
- 14Prognostic value of p53 nuclear accumulation and histopathologic features in T1 transitional cell carcinoma of the urinary bladder. Urology. 200;56: 735–740., , , et al.
- 30Deletions of the INK4A gene in superficial bladder tumors: association with recurrence. Am J Pathol. 1999; 55: 105–113., , , et al.