What's known on the subject? and What does the study add?
Non-muscle-invasive bladder cancer has a significant recurrence and progression rate despite transurethral resection. The current standard of care to lower the risk of recurrence and progression is adjuvant BCG followed by maintenance BCG. Despite this, a significant number of patients experience recurrence and progress to invasive cancer. Several randomized trials have studied combination therapy (BCG with chemotherapy) to try to reduce the recurrence and progression rate.
We performed a systematic review with meta-analysis and found that adjuvant BCG followed by maintenance therapy is the appropriate standard of care when compared with combination therapy. We conclude that further trials are warranted to test the effects of adding chemotherapy to BCG in patients with Ta or T1 disease, but not in those with Tis alone.
To determine if the combination of intravesical chemotherapy and maintenance bacille Calmette-Guérin (BCG), used in sequence, is superior to maintenance BCG alone in the treatment of non-muscle-invasive bladder cancer (NMIBC).
We searched biomedical literature databases for randomized controlled trials that compared sequential, intravesical chemotherapy added to maintenance BCG with maintenance BCG alone.
Studies that did not use maintenance BCG were excluded.
The meta-analysis was performed using the fixed effects model.
Four trials were identified, including 801 patients.
Adding chemotherapy to maintenance BCG did not result in a significant reduction in recurrence (relative risk [RR] 0.92; 95% confidence interval [CI] 0.79–1.09; P = 0.32) or progression (RR 0.88; 95% CI 0.61–1.27; P = 0.5).
The risk of recurrence (RR 0.75; 95% CI 0.61–0.92; P = 0.006) and progression (RR 0.45; 95% CI 0.25–0.81; P = 0.007) were reduced when the single trial that included isolated Tis was excluded.
Toxicity was similar for both groups.
Adjuvant therapy with induction BCG followed by maintenance BCG is the appropriate standard of care for patients with resected NMIBC at high risk of recurrence.
Further trials are warranted to test the effects of adding chemotherapy to BCG in patients with Ta or T1 disease, but not in those with Tis alone.
Bladder cancer has a significant morbidity and mortality from a global perspective. It is the seventh most common cancer in males and its mortality ranks ninth in recently published global cancer statistics . In developed countries, bladder cancer is ranked fourth for incidence and seventh for mortality among males . In northern Africa, it represents the second highest cause of cancer mortality in the male population .
Approximately 92% of all new cases of urothelial (transitional cell) bladder cancer were classified as non-muscle-invasive bladder cancer (NMIBC) in a recent population-based study , a higher figure than that cited in earlier reports [4, 5]. NMIBC includes Ta (papillary) and T1 (lamina-propria-invasive) tumours, and Tis (carcinoma in situ) which account for ∼70, 20 and 10% of non-muscle-invasive cancers, respectively . An estimated 15–70% of NMIBCs will recur within 12 months of a transurethral resection of the bladder tumour (TURBT) , and 7–40% will progress to muscle-invasive disease [8, 9] and, for this reason, more aggressive therapy is often required, even after initial, apparently complete, TURBT.
Intravesical therapy delivers high concentrations of a therapeutic agent locally within the bladder, potentially destroying viable tumour cells that remain after TURBT and preventing tumour implantation. The intravesical route also substantially reduces the potential for the systemic side effects of toxic therapies. BCG has been the mainstay of adjuvant therapy for high-grade Ta, T1 and Tis NMIBC since its anti-tumour effects were first described by Morales et al.  in 1976. Intravesical BCG yields a complete response in 55–65% of papillary tumours and in 70–75% of carcinomas in situ [11, 12]; however, a significant number of patients experience recurrence and progress to invasive cancer despite BCG therapy . Recently, the combined therapy of BCG plus intravesical chemotherapy has been used in an attempt to reduce the rates of treatment failure ; however, it remains uncertain if this new approach offers any significant benefit over existing therapy in patients with Ta, T1 and Tis. We performed a systematic review and meta-analysis of randomized controlled trials to determine the effects of intravesical chemotherapy with BCG compared with BCG alone on recurrence and progression in patients with NMIBC.
We included randomized controlled trials on patients who had undergone complete resection of NMIBC via transurethral resection. Patients with untreated or recurrent NMIBC were included. The intervention examined was intravesical BCG plus sequential intravesical chemotherapy (given on separate occasions), compared with the current standard of intravesical BCG alone. Both arms had to have a minimum of 6 months of maintenance BCG treatment. Different preparations of chemotherapy and treatment time courses were included and examined as possible sources of heterogeneity in the results. Trials that included i.v. anti-neoplastic or immunomodulatory agents, neoadjuvant treatment, or radiotherapy were excluded.
Trials were identified by searching MEDLINE, EMBASE and the Cochrane Central Register of Controlled Trials (1976 to August 2010). An attempt to identify unpublished trials was undertaken by hand searching the annual meeting proceedings of the American Society of Clinical Oncology, the European Association of Urology and the AUA. In addition, the references of selected articles were scanned for any other relevant trials. The search was not limited by language. The search strategy included, but was not limited to, the following medical subject headings (MeSH) and non-MeSH terms: urinary bladder neoplasms; bladder cancer; carcinoma, transitional cell; TCC; urologic neoplasms; chemotherapy, adjuvant; antineoplastic protocol; antineoplastic combined chemotherapy protocols; antineoplastic agents; gemcitabine; docetaxel; epirubicin; mitomycin C; doxorubicin; Adriamycin; BCG vaccine; BCG; Bacillus Calmette-Guerin; adjuvants, immunologic; administration, intravesical; intravesical; endovesical; randomized controlled trial; controlled clinical trial; randomized; placebo; clinical trials; randomly; and trial. Search strategy modifications were made as required for each database.
Selection of Trials and Data Collection
Two reviewers (B.H. and V.C.) independently assessed the eligibility of abstracts identified by the search. The full-text article of any trial that appeared to meet the inclusion criteria was retrieved for closer examination. There were no disagreements. Data were independently extracted by each reviewer using a standardized form. Data extracted from the reports included details about the trial methods, characteristics, interventions, duration of follow-up and outcome measures. The data extracted from the trials were entered into the Cochrane Collaboration software (RevMan version 5.0.2; http://www.cochrane.org).
Methodological Quality Assessment
Quality assessment was based on the reporting of the study methods and results. A ‘risk of bias’ assessment was made for each study as per the Cochrane Handbook  based on the following features: adequate sequence generation; allocation concealment; blinding of patient reported outcome and mortality; incomplete short- and long-term data; freedom from selective reporting; and freedom from other bias.
The main outcome measures for this meta-analysis were rates of recurrence and progression. Recurrence was defined by the detection of a subsequent bladder cancer during follow-up. Progression was defined by the detection of a recurrence with histological evidence of muscle invasion. Secondary outcomes were time to recurrence, time to progression and toxicity. Time to recurrence was time from randomization to first appearance of tumour. Time to progression was time from randomization to first histological evidence of muscle invasion.
The primary outcomes of interest were tumour recurrence and disease progression after treatment within the follow-up period of each individual trial. For dichotomous data, a summary relative risk (RR) and its 95% CI were calculated. A RR of <1 indicated an advantage of sequential chemotherapy and BCG over BCG alone.
The statistical heterogeneity of trial results was assessed by visual inspection of forest plots, chi-squared tests, and the I-squared statistic. A P value > 0.10 for the chi-squared test and an I2 value of <25% was considered to signify a low level of heterogeneity. Where heterogeneity was not significant, a summary estimate of the effect of the intervention on the outcome was reported, along with the 95% CI. There are many variations for intravesical treatment regimens and no standard is currently accepted practice for superficial bladder cancer. As a result it was assumed a priori that there would be substantial heterogeneity with the results. Thus, data were pooled using a random effects model. Publication bias was assessed with funnel plots.
Subgroup analyses were planned a priori to assess the effects of treatment duration (<1 years, ≥1year) and pathological risk group (Ta vs. T1 vs. Tis).
Four trials, including 801 patients, were selected using the strategy shown in Fig. 1 [14, 16-18]. The characteristics of the four included trials are shown in Table 1 [14, 16-18]. All trials included patients with T1 disease, three included Ta disease and one included Tis. The duration of therapy in three studies was 12 months, and was 36 months in the remaining study. Each trial used different doses and schedules to administer the BCG and intravesical chemotherapy. The trial from Mansoura University used an alternating regimen of Pasteur BCG 150 mg and epirubicin 50 mg given as an induction course (weekly for 6 weeks), followed by monthly dosing for 10 months . Cai et al.  gave a single early dose of epirubicin 80 mg (within 6 h of TURBT), followed by an induction course of BCG (5 × 108 colony-forming units of OncoTice) for 6 weeks, and then boosters of BCG according to the protocol described by Lamm et al.  (3-weekly instillations of BCG given at 3, 6, 12, 18, 24, 30 and 36 months). Di Stasi et al.  gave an induction course of BCG 81 mg (Connaught substrain [ImmuCyst]) and electromotive mitomycin C 40 mg, which was scheduled as three cycles given weekly for 9 weeks (each 3-week cycle was BCG, BCG, mitomycin C). The maintenance regime consisted of monthly infusions of either electromotive mitomycin C or BCG given according the following recurring schedule: mitomycin C, mitomycin C and BCG. This schedule was repeated continuously for 9 months (i.e. patients received six doses of mitomycin C and three doses of BCG during the maintenance period) . Kaasinen et al.  used only mitomycin C 40 mg in the induction course (six weekly instillations), followed by an alternating schedule of BCG 120 mg (Connaught) and mitomycin C monthly for 12 months. The control arm in all studies received an induction course of BCG, followed by maintenance BCG alone.
Table 1. Characteristics of included trials
Study and year
Mean age, years
Tumour stage, grade
*Alternating dosing. †Single dose within 6 h of surgery. BCG was given as maintenance in control and intervention arms for all studies.
There were 351 recurrences in the four trials, 173 in the combination arms and 178 in the BCG-alone arms. Meta-analysis showed no significant overall effect (RR 0.92, 95% CI 0.79–1.08, P = 0.32), but substantial heterogeneity (P = 0.01; I2 = 72%; Fig. 2).
The pre-planned subgroup analysis showed that the heterogeneity in the overall results was explained by tumour stage (Fig. 3): analysis according to tumour stage showed a substantial benefit for adding chemotherapy in patients with Ta or T1 disease (RR = 0.75; 95% CI 0.61 to 0.92; P = 0.006, heterogeneity P = 0.48, I2 = 0%), with a non-significant trend favouring BCG alone in patients with Tis (RR 1.22, 95% CI 0.97–1.54, P = 0.09). Subgroup analysis did not suggest any interaction between treatment group and treatment duration on recurrence.
The meta-analysis of all four trials showed no significant effect on progression of adding chemotherapy (RR = 0.88; 95% CI 0.61–1.27; P = 0.50), but again with substantial heterogeneity (P = 0.02, I2 = 70%; Fig. 4). Subgroup analysis by tumour stage again showed a substantial effect in patients with Ta or T1 disease (RR = 0.45; 95% CI 0.25–0.81; P = 0.007) without significant heterogeneity (P = 0.96, I2 = 0%; Fig. 5), with a nonsignificant trend favouring BCG alone in patients with Tis (RR 1.55, 95% CI 0.94–2.57, P = 0.09). Subgroup analysis did not suggest any interactionbetween treatment group and treatment duration on progression.
Toxicity data were available from two of the studies [14, 16]. There were no demonstrable differences in toxicities (cystitis, haematuria and fever) between combined chemotherapy plus BCG and BCG alone.
The present meta-analysis showed that adding chemotherapy to BCG had no effect overall, but suggested the possibility that its effects might depend on tumour stage with substantial observed benefit in patients with Ta or T1 disease, but no benefit and a trend towards harm in patients with Tis alone. This observation may be attributable to chance, but it raises the possibility that the addition of chemotherapy might be beneficial and deserves further study, particularly in patients with Ta or T1 disease.
The Kaasinen et al. trial  was restricted to patients with Tis, and included patients with recurrent disease and previous intravesical therapy. It was also the only study not to use BCG during induction therapy in the combination arm, using weekly mitomycin C instead. It is possible that mitomycin C is inferior to BCG as induction therapy for Tis. It is also possible that the previous use of BCG increases the effectiveness of subsequent chemotherapy. The Di Stasi et al. trial  used electromotive delivery, which is not commonly used or widely available. These authors also used a different sequencing of therapy from the other trials: they used BCG for induction, and then the electromotive mitomycin C. It remains unclear whether the benefit they found was attributable to the sequencing (i.e. giving the BCG before the chemotherapy) or the electromotive therapy.
There was no increase in toxicity with the addition of chemotherapy to intravesical BCG.
As expected, the trials identified were a heterogeneous mix of intervention and duration of therapy reflecting the lack of evidence to support one chemotherapeutic agent over another. Two studies used mitomycin C and two used epirubicin. The doses of BCG also varied. The currently accepted standard dose of BCG is 81 mg, but recent trials have suggested that a much lower dose of 27 mg may be equally effective with fewer side effects .
A limitation of this meta-analysis is that it is based on published results and not individual patient data. This may overestimate the effect of adding chemotherapy to BCG. The other limitations are the heterogeneity of treatment regimens, with all trials using different doses of BCG, and different chemotherapy agents or dosages for the experimental arm. Toxicity data were limited by the small number of studies that reported detailed toxicity outcomes; however, excluding trials with unconventional treatments (duration and type) did not alter the overall outcome as the trials were small.
The available evidence suggests that adjuvant therapy with induction BCG followed by maintenance BCG is the appropriate standard of care for routine practice and for trials in patients with resected high grade Ta, T1 or Tis NMIBC. Further trials are warranted to test the effects of adding chemotherapy to BCG in patients with Ta or T1 disease, but not in those with Tis alone.
Ian Davis is an NHMRC Practitioner Fellow. ANZUP Cancer Trials Group is supported by the Australian Government through Cancer Australia.