The increasing use of intravesical therapies for stage T1 bladder cancer coincides with decreasing survival after cystectomy


Erica H. Lambert, Urology, Columbia Presbyterian Medical Center, NY, NY, USA.



Intravesical therapy (IVT), chemo and immunotherapy, has made conservative, bladder-sparing strategies a viable option for managing patients with high grade T1 bladder cancer. However, many of these patients will have recurrence and occasionally progression, questioning delayed intervention. This study examines the patterns of use of IVT in high-grade T1 bladder cancer and the subsequent impact on survival for patients ultimately proceeding to radical cystectomy (RC).


Between 1990 and 2005, 104 patients were identified with T1 high-grade transitional cell carcinoma (TCC) and who underwent RC. Patients were divided into two groups; those having RC before 1998 (median year of surgery) and those after 1998. Trends in time from diagnosis to RC, courses of IVT, recurrence and pathological stage were analysed using two-sample t-tests with 95% confidence intervals. Kaplan–Meier analysis was used to determine the disease-free and overall survival rates.


Before 1998, 28 of 38 patients (74%) proceeded directly to RC with no IVT, vs 20 of 47 (43%) after 1998 (P = 0.004). The mean number of IVT courses per patient was 0.53 before 1998 and 1.2 afterward (P = 0.016). Patients who had RC before 1998 had a 69.7% disease-free survival at 5 years, vs 39.6% for those after 1998 (P = 0.05).


In the past 15 years, our experience indicates that patients having RC for T1 high-grade TCC after 1998 were more likely to receive IVT. These same patients had a worsening disease-free survival. In very few other cancers has disease-free survival decreased over time. We postulate that the decrease in survival might be related to an increased use of IVT.


intravesical therapy


radical cystectomy


carcinoma in situ


transurethral resection


disease-specific survival


disease-free survival.


Intravesical therapy (IVT), by chemotherapy and immunotherapy, has made conservative bladder-sparing strategies a viable option for managing high-risk superficial bladder cancer. Nevertheless, patients with such disease, including high-grade Ta, high-grade T1 and carcinoma in situ (CIS), will have recurrence and/or progression [1,2]. Transurethral resection (TUR) and adjuvant IVT with BCG is the accepted first-line treatment for these patients. Although BCG has been shown to lower the risk of recurrence and progression, the long-term follow-up shows no positive impact on disease-specific survival (DSS), with progression and mortality as high as 53% and 34%, respectively [3]. Specifically, patients with high-grade T1 disease treated conservatively show a continuous decline in DSS over time, with a 30% risk of progression and death at 10 years [4].

The response to BCG therapy at the first follow-up cystoscopy is the most powerful predictor of tumour recurrence and progression [5]. A lack of response or disease refractory to multiple IVTs warrants consideration of more invasive treatment [6]. At the time of radical cystectomy (RC), pathological staging indicates that 35–50% of patients with T1 high-grade bladder cancer harbour muscle-invasive disease, including micrometastases in 10–15%[7]. The excessive use of IVT could theoretically worsen survival for those understaged patients whose surgery is delayed by these more conservative measures. The question of delay of intervention and its role in the high-risk population is therefore relevant. In this study, we examined the patterns of use of IVT in high-grade T1 bladder cancer over the past 15 years at one institution, and the subsequent impact on survival for patients ultimately proceeding to RC.


The Urologic Oncology Database approved by the Institutional Review Board was retrospectively reviewed. Between 1990 and 2005, 505 patients had RC for TCC of the bladder; 104 were identified with high-grade T1 bladder cancer on biopsy immediately before RC. High-grade was defined by the WHO/International Society of Urological Pathology Consensus Classification [8]. Patients with Ta and Tis were excluded from the sample. All patients had data available on courses of IVTs, number of recurrences and time to RC from the date of diagnosis. Data after RC were complete in all patients. The decision to proceed to RC was at the surgeon’s discretion, evaluating for depth of lamina propria invasion in the TUR specimen, multifocality, and/or concomitant CIS. Patients receiving IVT had RC primarily for persistent disease refractory to IVT.

The patients’ data were examined for trends in treatment, response and survival. Clinical and pathological prognostic factors before RC, including year of RC, interval between tumour diagnosis and RC, number of IVTs, number of recurrences, and presence or absence of CIS, were evaluated. Patients were analysed for pathological stage and disease status after RC. The group was divided into two groups based on the median year of surgery, i.e. those having RC before 1998 (early group), and those during or after 1998 (late group).

Chi-squared analyses with 95% CIs were used to assess differences between patients in each group who had IVT, recurred, had RC within 120 days of diagnosis, had CIS on pathology before and after RC, and who were up-staged. Recurrences were defined as positive visual and pathological findings on subsequent cystoscopy and/or resection. Two-sample t-tests with 95% CIs were used to assess the difference in the mean number of IVTs, recurrences, and time to RC between the early and late groups.

Survival was analysed using the Kaplan–Meier method with validation by log-rank testing. Disease-free survival (DFS), DSS and overall survival were assessed between groups based on the median year of surgery (1998). Results were also analysed between patients who did not have IVT vs those who received any IVT regardless of year of surgery. Using a Cox proportional hazard model, we analysed year of surgery as a continuous variable and its impact on survival after RC. In all tests, statistical significance was assumed when P < 0.05.


Table 1 describes the overall patient characteristics. Of 505 patients who had RC, 104 were included in the final analysis who were diagnosed with high-grade T1 bladder cancer at the time of RC (mean age 67.5 years at RC). There was no statistically significant difference in mean age between patients in the early and late groups (P = 0.23). Gender was evenly distributed amongst each group; 21% of the overall population was female.

Table 1.  Patient information before and after RC
VariableN*Early (before 1998)Late (after 1998)P
  • *

    Complete information on IVT, recurrences and time to RC could not be obtained for all patients.

  • †For some patients, despite knowing that IVT was received, the number of IVTs could not be obtained.

Mean (range) age at RC, years10466.7 (49.8–78.8)68.1 (38.8–87.7)0.23
n/N (%):
Female 22/104 (21.2) 7/44 (15.9)15/60 (25.0)0.26
IVT 37/85 (43.5)10/38 (26.3)27/47 (57.5)0.004
≥2 IVT 19/78 (24.6) 4/36 (11.1)15/42 (35.7)0.01
Mean (range) courses of IVT 78 0.53 (0–5) 1.2 (0–5)0.016
Recurrence, n/N (%) 29/76 (38.2)16/35 (45.7)13/41 (31.7)0.21
Mean (range) no. of recurrences 76 1.4 (0–9) 2.1 (0–13)0.12
Days to RC <120, n/N (%) 40/93 (43.0)23/40 (57.5)17/53 (32.8)0.04
Mean (range) years to cystectomy 93 1.4 (0.01–16.7) 2.0 (0.02–23.3)0.21
Pathological stage, n (%)10444600.63
 pT0  5 (4.8) 2 (4.6) 3 (5.0) 
 pTis 17 (16.3) 8 (18.2) 6 (10.0) 
 pTa 14 (13.5) 9 (20.5) 8 (13.3) 
 pT1 26 (25.0) 9 (20.5)17 (28.3) 
 pT2 13 (12.5) 5 (11.4) 8 (13.3) 
 pT3 24 (23.1)10 (22.7)14 (23.3) 
 pT4  5 (4.8) 1 (2.3) 4 (6.7) 
Up-staged, n/N (%) 42/104 (39.6)16/44 (36.4)26/60 (43.3)0.47
CIS: n/N (%)
 before RC 50/104 (48.1)21/44 (47.7)29/60 (48.3)0.95
 after RC 78/104 (75.0)33/44 (75.0)45/60 (75.0)1.00

Before 1998, 28 of 38 patients (74%) proceeded directly to RC with no IVT, vs 20 of 47 (43%) after 1998 (P = 0.017; Table 2). The mean number of IVTs per patient increased with time (P = 0.016), and significantly more patients in the late group had two or more IVTs than those in the early group (P = 0.01, Table 2). Patients in the late group tended to have more recurrences before RC, although this was not statistically significant (P = 0.1). The median time to RC from the date of first diagnosis of high-grade T1 TCC was not statistically different between the groups. However, 58% of patients in the early group had RC within 120 days of a high-risk diagnosis, vs 33% of those in the late group (P = 0.04, Table 2). Of all patients, ≈ 40% had stage progression in each group, which was statistically equivalent (Table 1).

Table 2.  A comparison of courses of IVT and time to RC between the early and late groups; the mean number of IVTs was 0.53 and 1.2 (P = 0.016), respectively, and the distribution of patients receiving none, one and two or more IVTs was different between groups (P = 0.017); the mean time to RC was 16.4 and 23.6 months, respectively (P = 0.2), and the percentage of patients proceeding to RC within 120 days was different between groups (P = 0.01)
VariableEarly (before 1998), %Late (after 1998), %
IVT courses
 ≥2 1136
Time to RC, days

The Kaplan–Meier survival analysis demonstrated differences in DFS (69.7% vs 39.6%; P = 0.05, Fig. 1), DSS (100% vs 87%; P = 0.03, data not shown) and overall survival (94.1% vs 81.6%; P = 0.10; data not shown) between the early and late groups, respectively, at 5 years. Independent of date of surgery, patients receiving IVT had no survival difference from those who proceeded directly to RC (P = 0.95). Patients who were up-staged had a significant worsening of DFS (P = 0.003).

Figure 1.

DFS between the early and late groups; at 5 years the DFS rates were 69.7% and 39.6% for the early and late groups, respectively (P = 0.05).

The year of surgery was examined as a continuous variable using a Cox proportional hazard model; the hazard ratio was 1.15 (P = 0.035), indicating a 15% increase in the risk of disease recurrence and death from disease as the year of surgery progressed from 1990 to 2005.


The vast majority of patients with high-grade T1 bladder cancer are treated with conservative, bladder-sparing strategies, e.g. TUR followed by IVT (immunotherapy and/or chemotherapy). However, these same high-risk patients often have multiple recurrences and occasional progression to muscle-invasive disease. This has coincided with a progressive increase in the use of intravesical BCG and an increasing interval from the diagnosis of high-grade T1 disease to RC. The possibility that IVT might be delaying curative surgery and adversely affecting survival is an important consideration.

Before 1998, 74% (28/38) of patients with high-grade T1 TCC proceeded directly to RC after initial TUR, with no trial of IVT, compared to 43% (20/47) after 1998. Most of these later patients received at least one 6-week cycle of BCG. For subsequent recurrences, patients had varying cycles of BCG, mitomycin C, thiotepa, adriamycin and other investigational agents. The decision to proceed directly to RC was based on a combination of prognostic clinical and pathological factors, including depth of invasion into the lamina propria, lymphovascular invasion, multifocal disease, and the presence of CIS. Deep invasion into the lamina propria and high histological grade has been shown to most commonly correlate with the likelihood of tumour progression in high-grade T1 bladder cancer [9,10]. In addition, patients with T1 disease and concomitant CIS are more likely to be upstaged to invasive disease after RC [11,12].

The likelihood of receiving more than one IVT was significantly greater among patients treated for T1 disease after 1998. However, the absolute number of IVT courses instilled, regardless of period of surgery, did not significantly affect survival. It is possible that we were unable to detect a difference as there were relatively few patients and a restricted distribution of the number of IVTs in these groups. However, unlike most malignancies at present, the estimated DFS for patients with high-grade T1 disease and having RC at our institution has declined over the last 15 years. This decline in DFS could be explained by a delay in RC; before 1998 patients more frequently had RC within 120 days of diagnosis, as opposed to after 1998. We postulate that the delay is due to the increasing use of IVT, shown by the near doubling in the number of IVTs between the periods. However, the delay from the original diagnosis of T1 disease to RC was not significantly associated with worsening DFS. Again, the study was probably under-powered to detect a significant difference. In addition, ≈ 15% of the present patients were excluded from this survival analysis because we could not determine the date of the original diagnosis of some patients who had been referred to our institution from outside hospitals.

Inaccurate staging at the initial TUR is common, as 27–62% of high-grade T1 tumours are up-staged after RC [13–15]. Patients who have such stage progression have a worse DSS due to a delay in the treatment of muscle-invasive TCC [16]. In the present study, 40% of patients had stage progression; this was consistent between groups and, as expected, these patients had a lower DFS than those accurately staged with T1 disease before RC.

Stage progression has prompted many investigators to use a second or re-staging aggressive TUR before initiating BCG. Repeat TUR at the site of the initial resection has been shown to improve the initial response rate to BCG therapy, by removing residual tumour [17]. Others have shown that muscle-invasive cancer is present in 24–38% of repeat TURs that were initially found to be high-grade T1 [18,19]. Re-staging procedures were not uniformly used in the present patients.

At our institution, patients with tumour characteristics considered as high-risk for progression, or in whom attempts at bladder preservation, failed were offered RC. Herr and Sogani [7] reported a survival benefit for an early vs delayed RC for patients with high-risk superficial disease. For those who progressed to muscle invasion or who had persistent disease on surveillance cystoscopy after BCG, the survival rate was 92% when the RC was within 2 years of initial BCG therapy, compared to 56% in patients who had RC after 2 years. In addition, it was reported that patients with a history of superficial bladder cancer and progression to muscle-invasive disease have worse survival than patients who present with primary invasive tumours [20]. The mechanism is not clearly understood. Our results and these studies emphasize the need to develop more reliable factors to identify patients at risk of progression. Patient outcomes are better with early aggressive surgery.

Notably, the present groups comprised only patients with clinical high-grade T1 disease who had RC. Thus, patients treated with IVT (immunotherapy and/or chemotherapy) without having had RC were not included in the analysis. Although there appears to be a selection bias, our study focuses on trends in the treatment algorithm over time. Importantly, these data show a shift in the treatment of high-grade T1 disease, from early RC to multiple cycles of IVT, followed by delayed RC. Interestingly, when time was analysed as a continuous variable, patients in 2005 had a significantly greater likelihood of disease recurrence and dying from disease, as opposed to those in 1990. Overall, the present study seems to indicate that the trend toward increasing use of IVTs over time has resulted in worsening DFS in patients with clinical high-grade T1 bladder cancer.

In the past 15 years our experience indicates that patients are more likely to have a longer and more courses of IVT before RC for T1 bladder cancer. We also noted a worsening of DFS during the same interval. In very few other cancers has DFS decreased over time. We postulate that the decrease in survival might be related to a delay in definitive treatment.


None declared.