Multivariate analysis of survival, recurrence, progression and development of mestastasis in T1 and T2a transitional cell bladder carcinoma

Authors


Abstract

BACKGROUND

Determination of prognosis factors associated with survival, recurrence, progression, and development of metastasis in T1 and T2a transitional cell carcinoma (TCC) of the bladder is discussed.

METHODS

A study was conducted of a group of 210 patients with primary bladder TCC at classification T1 (n = 175) and T2aN0M0 (n = 35). A total of 177 variables were studied in each patient. The monoclonal antibodies used were the following: DO7 (p53) and MIB-1 (Ki-67). Prognosis was obtained using Kaplan–Meier methodology and Cox proportional hazards model.

RESULTS

The average follow-up period was 6.7 years. Cancer-related survival rates at 5 and 10 years were 82.96% and 74.78%, respectively. The independent survival variables were the following: age and expression of p53. Recurrence free survival at 5 and 10 years stood at 51.80% and 42.71%, respectively. The independent recurrence variables were T2a classification, tumor multifocality, tumor size of greater than 3 cm, carcinoma in situ in random biopsy, and expression of Ki-67. Progression free survival rates at 5 and 10 years were 75.31% and 69.16%, respectively. The independent progression variables were age, T2a classification, and expression of p53. Metastasis free survival rates at 5 and 10 years stood at 87.23% and 84.55%, respectively. The expression of p53 was the sole variable to provide an independent prediction of metastasis.

CONCLUSIONS

The expression of p53 clearly has an independent effect on the prediction of survival, progression and development of metastasis, showing a dose–response effect. Tumor multifocality and T2a classification are the variables that best predict recurrence. Cancer 2002;94:1677–84. © 2002 American Cancer Society.

DOI 10.1002/cncr.10376

Numerous factors have been associated with the prognosis of transitional cell carcinoma (TCC) of the bladder. Traditional prognosis factors have been based mainly on histopathologic characteristics of the tumor. Patients then are classified into various risk groups to provide them with optimum treatment at the most appropriate time.

More recently, however, new prognosis factors have come into the limelight, related to cytometric and molecular tumor characteristics, which would appear to provide further information, but which so far have not enabled urologists to abandon the use of the more traditional factors. The new prognosis factors include DNA ploidy, the cell proliferation activity markers, growth factors and growth factor receptors, various cell adhesion and cytoskeleton molecules, angiogenesis stimulators and inhibitors, oncogenes, and tumor suppression genes.1

T1 and T2a classification tumors often create a therapeutic dilemma that ideally would require a definitive prognosis factor, which so far remains undiscovered.

The objective of our study is to determine the prognosis factors related to survival, recurrence, and progression of bladder TCC at classifications T1 and T2a, focusing especially on the prognosis value of the immunohistochemical expression of p53 and Ki-67.

MATERIALS AND METHODS

An observational retrospective and prospective study was conducted on a group of 210 patients, all diagnosed with primary bladder TCC, at classification T1 and T2aN0M0, during the period 1987–1992, at the Xeral-Cíes Hospital in Vigo, Spain.

A total of 177 variables were studied in each case, including the following: gender, age, smoking habits, initial clinical symptoms, time periods from the onset symptoms to diagnosis, urine cytology, the existence of hydronephrosis, carcinoma in situ (CIS) in random bladder biopsy, presence of CIS in prostate urethra, T classification, histologic grade, number of tumors, size of the largest tumor, the endoscopic shape of the tumor, microscopic growth pattern, cell nucleus size, percentage of p53 expression in quartiles, and the percentage of Ki-67 expression in quartiles. Further clinical, and follow-up-related variables also were studied.

Treatment for Primary Tumors

All the primary tumors were treated with transurethral resection (TUR). In all cases, the resection was performed to pericyst and included a biopsy of the tumor bed. All tumors at classification T1 received intravesical installation of chemical or immunotherapy agents, after the resection of the primary tumor. None of the patients was subjected to systemic adjuvants or neoadjuvant chemotherapy.

Patient Follow-Up after the Initial TUR

Follow-up included the following: 1) quarterly cystoscopy during the first year, to be followed by a cystoscopy every 6 months during the second year and an annual cystoscopy from the third year onward. Follow-up in those patients with zero cystoscopy tolerance was by means of a bladder echography, although the same revision timetable was maintained; 2) urine cytology and study of renal function by means of blood urea and creatinine counts, every 6 months during the first 2 years and once a year from the third year onward; 3) urine sediment check every six months during the first 2 years and once a year from the third year onward; 4) yearly intravenous urography during the first 2 years and once every 2 years after that; 5) an abdominopelvic computerized tomography scan conducted in all T2a tumors 1 month after TUR of the primary tumor.

Cancer-related death was defined as that caused by bladder carcinoma. Recurrence was defined as the appearance of a new urothelial tumor in any part of the urinary tract. Progression was defined as an increase in T, N, or M classification in any of the recurrences. Within the context of progression, special attention was paid to the development of metastasis.

Pathologic Assessment

All the tumors were graded in accordance with the World Health Organization scale.2 Clinical and pathologic classification were determined in accordance with the American Joint Committee on Cancer/International Union Against Cancer TNM 1997 classification.3 A section was taken from each tumor, based exclusively on the amount of tumor present, and the apparent morphologic feasibility of the tumor cells.

Immunohistochemical Study

A microtome was used to obtain sections 4 μm in thickness from the selected paraffin block. These sections then were deparaffinized, beginning with their incubation in a heater at 60 °C for 12 hours, followed by their immersion in increasingly weaker concentrations of xylol and ethanol until they were completely rehydrated. The next step consisted of a systematic recovery of the antigens, by using a pressure cooker to heat the preparation submerged in buffer Citrat 10 mM (pH 6.0) or buffer EDTA 1 mM (pH 8.0) for 2–3 minutes at full pressure. For the immunohistochemical staining, an automated system known as Ventana NexES (Ventana Medical Systems, Inc., Tucson, AZ) was used. After an initial stage consisting of deactivating the endogenous peroxidase, this system provides doses for the primary antibody: p53 protein, clone DO7; Novocastra Laboratories, Newcastle, UK[reference NCL-p53-DO7]; dilution, 1:25) and antigen Ki-67, clone MIB-1 (Immunotech, Marseilles, France (reference PN IM0505); dilution, 1:100).

Incubation time for these primary antibodies was 30 minutes at a temperature of 40 °C. The system later revealed an immunologic reaction in the tissue, thanks to the avidin-biotin-peroxidase technique, which uses diaminobenzine as the chromogen. The preparations then were counterstained using hematoxiline and finally prepared for the standard procedure.

The quantitative and qualitative assessment of the immunohistochemical study was conducted using an Olympus BX 40 optical microscope. The percentage of p53 and Ki-67 expression was determined by calculating the average of the four microscopic fields (×40 magnification) with the most stained nuclei. More than 3000 nuclei were counted in all cases.

Statistical Analysis

A descriptive study of all the variables included in the study was conducted, with their 95% confidence interval. The quantitative variables were expressed using their centralization measurements, and in some cases they were categorized according to their median or quartile value. The mean showed the mean ± standard deviation. To determine the effect of the different variables on prognosis, we completed a survival analysis using Kaplan–Meier methodology. Dependent variables of interest were cancer-related survival, recurrence, and progression. To verify the probability of accumulated survival, in accordance with several strata of variables, the log-rank test was used. A P value of up to less than 0.05 was considered to be of statistical significance. To determine the way in which survival, recurrence, and progression were affected by other covariables, and to control their confounding effect, a Cox proportional hazards analysis was conducted. The hazards were estimated with their 95% confidence interval.

RESULTS

The mean age of the patients participating in the study was 66.05 ± 11.64; 83.33% were men. The average follow-up period was 79.95 months (6.7 years). One hundred seventy-five tumors (83.33%) belonged to classification T1, and 35 (16.66%) to classification T2a. The most relevant clinical and pathologic characteristics of the patients involved are shown in Table 1.

Table 1. Patient Clinical and Pathologic Characteristics
VariableFrequency (%)95% CI
  1. CI: confidence interval; CIS: carcinoma in situ.

Gender
 Male175 (83.33)77.83–87.92
 Female35 (16.66)12.08–22.17
Smoking
 Smoker150 (71.43)65.04–77.23
 Nonsmoker60 (28.57)22.77–34.96
Onset symptoms
 Macrohematuria160 (76.19)70.08–81.58
 Others50 (23.81)18.41–29.92
Duration of symptoms (mo)
 ≤ 6125 (59.52)52.77–66.01
 > 638 (18.09)13.32–23.74
Tumor multifocality
 Unknown47 (22.38)17.13–28.39
 Unifocal141 (67.14)60.57–73.24
 Multifocal69 (32.86)26.75–39.42
Tumor size (cm)
 ≤ 3106 (50.47)43.73–57.21
 > 3104 (49.52)42.79–56.27
Endoscopic shape
 Papillary174 (82.85)77.30–87.51
 Solid36 (17.15)12.49–22.69
Microscopic growth pattern
 Papillary180 (85.71)80.48–89.97
 Solid30 (14.28)10.03–19.52
Hydronephrosis
 No hydronephrosis192 (91.42)87.04–94.68
 Unilateral8 (3.81)1.78–7.10
 Bilateral1 (0.47)0.02–2.32
Random bladder biopsy
 Positive (CIS)10 (4.76)2.44–8.32
 Negative (no CIS)185 (88.09)83.17–91.97
 Not performed15 (7.14)4.21–11.25
Prostatic urethra
 Presence of CIS4 (2.28)0.73–5.42
 Absence of CIS97 (55.43)48.00–62.67
 Unknown74 (42.28)35.12–49.70
T classification
 T1175 (83.33)77.83–87.92
 T2a35 (16.66)12.08–22.17
Histologic grade
 120 (9.52)6.08–14.07
 2129 (61.43)54.71–67.83
 361 (29.04)23.21–35.46
Table 2. Cox Regression in the Prediction of Cancer-Related Survival
VariableRR95% CIP value
  1. RR: relative risk; CI: confidence interval.

Gender (female/male)1.840.72–4.700.2037
Age1.051.01–1.100.0130
T classification (T2a/T1)1.820.52–6.360.3446
Histologic grade0.9545
 11
 20.710.08–6.180.7613
 30.730.07–7.140.7869
Tumor multifocality (yes/no)2.490.97–6.410.0585
Tumor size (> 3/≤ 3 cm)1.810.74–4.400.1898
Endoscopic shape (solid/papillary)1.340.47–3.850.5833
Microscopic growth pattern (solid/papillary)3.800.79–18.140.0941
p53 expression (%)0.0946
 0–0.21
 0.2–23.890.72–20.950.1131
 2–146.661.36–32.720.0194
 > 147.711.38–42.960.0197
Ki-67 expression (%)0.8686
 0–51
 5–120.790.20–3.010.8686
 12–271.170.33–4.100.8034
 > 271.170.31–4.420.8115
Table 3. Cox Regression in the Prediction of Recurrence
VariableRR95% CIP value
  1. RR: relative risk; CI: confidence interval; CIS: carcinoma in situ.

Gender (female/male)1.020.57–1.840.9360
Age1.010.99–1.040.2135
T classification (T2a/T1)2.661.35–5.250.0045
Histologic grade0.2206
 11
 20.590.28–1.210.1534
 30.440.17–1.110.0837
Tumor multifocality (yes/no)2.831.70–4.710.0001
Tumor size (> 3/≤ 3 cm)1.821.12–2.950.0150
Random bladder biopsy (CIS/no CIS)1.391.05–2.520.0044
p53 expression (%)0.5459
 0–0.21
 0.2–21.030.53–2.010.9281
 2–141.490.79–2.820.2171
 < 141.100.50–2.410.8042
Ki-67 expression (%)0.0461
 0–51
 5–121.860.99–3.520.0544
 12–270.940.46–1.950.8775
 > 271.880.88–4.040.1045
Table 4. Cox Regression in the Prediction of Progression
VariableRR95% CIP value
  1. RR: relative risk; CI: confidence interval.

Gender (female/male)1.250.44–3.580.6742
Age1.051.01–1.090.0177
T classification (T2a/T1)2.711.22–6.010.0139
Histologic grade0.2334
 11
 20.550.25–1.190.1312
 30.430.16–1.150.0931
Tumor multifocality (yes/no)2.090.94–4.650.0702
Tumor size (> 3/≤ 3 cm)1.120.52–2.410.7629
Endoscopic shape (solid/papillary)2.520.88–7.220.0850
Microscopic growth pattern (solid/papillary)2.190.42–11.390.3508
p53 expression (%)0.0197
 0–0.21
 0.2–23.190.77–13.150.1075
 2–144.081.07–15.600.0397
 > 148.982.12–38.060.0029
Ki-67 expression0.2629
 0–51
 5–121.820.51–6.470.3521
 12–272.690.78–9.260.1161
 > 273.450.95–12.570.0600
Table 5. Cox Regression in the Prediction of Metastasis Development
VariableRR95% CIP value
  1. RR: relative risk; CI: confidence interval.

Gender (female/male)0.470.11–1.930.3000
Age1.030.98–1.090.1826
T classification (T2a/T1)2.620.54–12.640.2284
Histologic grade0.9233
 11
 20.790.09–6.860.8305
 30.660.06–6.890.7313
Tumor multifocality (yes/no)2.460.83–7.290.1035
Microscopic growth pattern (solid/papillary)2.020.37–10.980.4163
p53 expression (> 2%/≤ 2%)6.901.89–25.200.0035
Ki-67 expression (> 12%/≤ 12%)0.910.32–2.530.8567

Survival

Cancer-related survival after 5 and 10 years was 82.96% and 74.78%, respectively (Fig. 1). The variables in the univariate analysis that significantly altered survival were the following: T2a classification, histologic grade, tumor size of greater than 3 cm, urine irritation syndrome at the onset of the disease, a solid endoscopic shaped tumor, solid microscopic growth pattern, large cell nucleus, p53 expression, and Ki-67 expression.

Figure 1.

(A) Cancer-related survival. (B) Recurrence free survival.

To predict cancer-related survival, we adjusted Cox multivariate recurrence analysis for the following variables: gender, age, T classification, histologic grade, tumor size, endoscopic shape, microscopic growth pattern, p53 expression in quartiles, and Ki-67 in quartiles. The variables that independently predicted survival were age and p53 expression in quartiles, with a dose–response in p53, to the extent that the larger the expression of p53 the lower the probability of survival (Table 2).

Recurrence

Recurrence free survival rates after 5 and 10 years were 51.80% and 42.71%, respectively (Fig. 1).

The variables that were closely related to recurrence in the univariate analysis were the following: tumor size of greater than 3 cm, tumor multifocality, solid endoscopic shaped tumor, solid microscopic morphology, classification T2a, histologic grade, location of the tumor on anterior wall, urine irritation syndrome at onset, presence of CIS in the random bladder biopsy, expression of p53, and expression of Ki-67.

Cox multivariate recurrence analysis was adjusted for the following variables: gender, age, T classification, histologic grade, tumor multifocality, tumor size, presence of CIS in the random bladder biopsy, p53 expression in quartiles, and Ki-67 expression in quartiles. The variables capable of the independent prediction of recurrence were T2a classification, tumor multifocality, tumor size of greater than 3 cm, presence of CIS in bladder biopsy, and expression of Ki-67 (Table 3).

Progression

Progression free survival at 5 and 10 years stood at 75.31% and 69.16%, respectively (Fig. 2). Metastasis free survival at 5 and 10 years was 87.02% and 84.35%, respectively (Fig. 2).

Figure 2.

(A) Progression free survival. (B) Metastasis free survival.

The variables showing a significant association with progression in the univariate analysis were the following: T2a classification, histologic grade, tumor size of greater than 3 cm, solid endoscopic shaped tumor, solid microscopic morphology, location of the tumor on anterior wall, urine irritation syndrome at onset, presence of CIS in the random bladder biopsy, expression of p53, and expression of Ki-67.

The multivariate analysis was adjusted for the following variables: age, gender, T classification, histologic grade, tumor multifocality, number of tumors, endoscopic tumor shape, microscopic tumor growth pattern, p53 expression in quartiles, and Ki-67 expression in quartiles. The variables with the capacity to predict progression independently were the following: age, T2a classification, and p53 expression. A dose–response effect is observed in p53, as the greater the expression of p53 the higher the probability of progression (Table 4).

Metastasis free survival rates at 5 and 10 years were 87.23% and 84.55%, respectively. The variables linked significantly to the development of metastasis in the univariate analysis were T2a classification, histologic grade, expression of p53, and expression of Ki-67.

The multivariate analysis used to predict the development of metastasis was adjusted by the following: age, gender T classification, histologic grade, tumor size, tumor multifocality, microscopic growth pattern, expression of p53 (> 2% vs. ≤ 2%), and expression of Ki-67 (> 12% vs. ≤ 12%). The only variable capable of an independent prediction of the development of metastasis was the expression of p53 (Table 5).

DISCUSSION

In our study, patient age is shown as an independent variable of survival and tumor progression, to the extent that the older the patient the greater the probability of progression. These results are consistent with several multivariate studies of survival in bladder carcinoma, such as those by Takashi et al.4 and Narayana et al.5 However, in the study by Jahnson et al.6 age did not behave as an independent variable of survival.

It is hard to draw a comparison between the results obtained from the above mentioned studies, including our own, mainly because of the various stages of the tumors studied in each one.

The existence of infiltration in the muscle layer is a crucial factor in the early prognosis of bladder carcinoma. As a result, and as Martínez Piñeiro et al.7 point out, survival after 5 years in patients at classification T1 is between 60% and 80%, but this decreases to 35–50% in patients with T2 tumors. In our study, cancer-related survival rates in classification T1 after 5 and 10 years is 89.51% and 80.68%, respectively, whereas survival in classification T2a stands at 44.59% at both the 5 and 10-year period. There are statistically significant differences between the two tumor classifications (log-rank test, 37.13; P = 0.0000).

The T classification is an independent variable of recurrence and progression among the patients included in our study, yet it does not provide an independent form of predicting cancer-related survival or the development of distant metastasis. Despite this, we have observed that the risk of death from classification T2a bladder carcinoma is 1.82 times higher than that of death from classification T1 bladder carcinoma, and the risk of developing metastasis is 2.62 times higher in T2a tumors than in T1 tumors. It would be logical to think that a variable that independently predicts recurrence and especially tumor progression therefore should be capable of predicting survival, because all the deaths from bladder carcinoma in classifications T1 and T2a are caused by tumor progression. For this reason, we believe that if we were to increase the number of T2a tumors in the sample, the differences in the multivariate survival analysis would be significant for the T classification, as occurs in the univariate analysis.

There are no existing studies to our knowledge that specifically analyze both T1 and T2a tumors. The multivariate studies of Pfister et al.8 and Allard et al.9 highlight the importance of the T classification in tumor recurrence, but in both the object of the study was classification Ta and T1 tumors. Fujii et al.10 and Kiemeney et al.11 observe an independent association between the T classification and tumor progression, and the study by Kiemeney et al. observes a similar association in recurrence.

Lipponen et al.12 analyze a sample of Ta–T4 tumors and reach the conclusion that the T classification is an independent variable of cancer-related survival in the Ta–T2 tumor subgroup. They also conclude that the T classification is the most important factor of prognosis in predicting survival. However, these authors do not observe a significant link between the T classification and progression in the multivariate analysis.

Patients suffering from multiple tumors right from the onset, probably show a generalized genetic instability of the transitional epithelium,13 which likewise conditions the synchronized appearance of several tumors and suggests a particularly marked tendency to develop metachronic tumors. This would explain the overwhelming tendency of patients with multifocal tumors to develop new bladder tumors, most of which are new occurrences (70%) and not true recurrences, as noted by Nogueira March.14

Our study shows that the risk of recurrence in multiple tumors is almost three times greater than that of single tumors, making tumor multifocality the main factor determining the risk of recurrence. Research by Allard et al.,9 Witjes et al.,15 Parmar et al.,16 and Mulders et al.,17 has shown the independence of tumor multifocality and highlight the vital role this variable plays in tumor recurrence.

The existence of CIS in random bladder biopsy has been the subject of numerous studies,9, 11, 18 in an attempt to define its importance in prognosis. However, few have viewed CIS as a variable possessing independent prognosis capacity. With regards to this variable, there is a considerable lack of consistency among the various studies, even in such basic issues as the concept of CIS.

In the study by Millán-Rodríguez et al.,18 the presence of CIS in the bladder biopsy is shown as an independent variable of survival, recurrence, and progression. In our study, we observed that the presence of CIS is an independent prediction recurrence factor, yet it has no significant link with survival and progression in the multivariate analysis. This variation in results probably is caused by the population of tumors considered in both studies, because whereas the study by Millán-Rodríguez et al.,18 was conducted using Ta and T1 tumors, ours was completed with T1 and T2a classification tumors.

The expression of p53 has been viewed as an independent factor in several studies looking into the prognosis of bladder carcinoma, in the case of both superficial and muscle-infiltrating tumors. The patients in our study with an expression p53 greater than 14% are 8 times more likely to die as a consequence of the tumor than those with an expression of less than or equal to 0.2%. Similarly, it was observed that the risk of progression in tumors with an expression of p53 greater than 14% is 9 times higher than in tumors whose expression of p53 is less than or equal to 0.2%. Tumors with an expression of p53 between 2% and 14% also show a significantly high risk of progression and cancer-related death.

With regards to metastasis, it was observed that tumors with an expression of p53 greater than 2% have a long-term risk of developing metastasis, which is 7 times higher than in those tumors with an expression of less than or equal to 2%. The expression of p53 is the sole variable to provide an independent prediction for the development of metastasis.

The results of studies analyzing the relation between the expression of p53 and survival vary enormously. This is probably because of the diversity existing among samples: differing tumor classifications, a range of different treatments, different cutoff points for p53, different monoclonal antibodies, etc.

Esrig et al.19 and Tsuji et al.20 point to a significant link between the expression of p53 and survival, although the latter study included only a univariate analysis. However, in the studies by Jahnson et al.6 and Liukkonen et al.21 the expression of p53 does not appear as an independent variable of survival.

The relation between the expression of p53 and tumor progression creates fewer discrepancies, as in most cases p53 appears as an independent variable of progression, as occurs in the studies by Schmitz-Dräger et al.22 and Serth et al.23

The studies analyzing the expression of p53 in bladder carcinoma show that the threshold of positivity adopted to obtain two populations with significantly different risks is considerably higher in the case of muscle-infiltrating tumors than in superficial ones. In the case of muscle-infiltrating tumors it can reach the 40% used by Kuczyk et al.,24 with the corresponding loss in the specificity of the immunohistochemical technique.

In our study, we observed that the expression of Ki-67 independently predicts tumor recurrence, yet it does not objectify a dose–response effect. If we consider our results and those of other authors such as Zlotta et al.25 and Pfister et al.,8 in which the expression of Ki-67 does not appear as an independent variable of recurrence, it would seem that the expression of Ki-67 does not play an important role in predicting tumor recurrence.

The role played by the histologic grade in the prognosis of T1 and T2a bladder tumors is a controversial one. In our study, the grade does not appear as an independent prognosis variable. In this aspect, our results are consistent with those included in other multivariate studies, in terms of survival,6, 12, 19, 21 recurrence,8, 11, 15–19, 25 and progression.21–23, 25

Our study shows that there are several significant variables in the prognosis of bladder TCC at T1 and T2a classifications. The expression of p53 shows its independence in predicting survival, progression, and metastasis development, whereas tumor multifocality, the T2a classification, and the presence of CIS in random bladder biopsy are the variables associated with a greater risk of recurrence.

Ancillary