Intravesical bacillus Calmette-Guérin is superior to mitomycin C in reducing tumour recurrence in high-risk superficial bladder cancer: a meta-analysis of randomized trials
M.D. Shelley, Cochrane Prostatic Diseases and Urologic Cancers Group, Velindre NHS Trust, Cardiff UK.
The first paper in this section is a meta-analysis of randomized trials into the effect of two intravesical agents used to reduce tumour recurrence in high-risk superficial bladder cancer. This is the highest level of evidence-based medicine and the results presented here suggest that intravesical BCG is superior to mitomycin C. There was no difference in the effect of the two agents on disease progression or survival. In a study from Amsterdam and Rome, the authors describe a phase I trial into the toxicity and safety of intravesical gemcitabine, and found that it was well tolerated with no systemic toxicity.
In a paper from Taiwan the link between viral infections and tumour foundation is evaluated, in this case specifically infections with the Epstein-Barr virus and its correlation with urothelial cancer. The authors found that there was indeed a strong association between infection with this virus and a considerable proportion of primary urothelial TCCs.
To assess, in a systematic review and meta-analysis, the relative effectiveness of intravesical mitomycin C and bacillus Calmette-Guérin (BCG) for tumour recurrence, disease progression and overall survival in patients with medium- to high-risk Ta and T1 bladder cancer.
The major medical databases were searched comprehensively up to June 2003, and relevant journals hand-searched for randomized controlled trials, in any language, that compared intravesical mitomycin C with BCG in medium- to high-risk patients with Ta or T1 bladder cancer.
Twenty-five articles were identified but only seven were considered eligible for the analysis. This represented 1901 evaluable patients in all, 820 randomized to mitomycin C and 1081 to BCG. Six trials had sufficient data for meta-analysis and included 1527 patients, 693 in the mitomycin and 834 in the BCG arm. There was no significant difference between mitomycin C and BCG for tumour recurrence in the six trials, with a weighted mean log hazard ratio, LHR, (variance) of −0.022 (0.005). However, there was significant heterogeneity between trials (P = 0.001). A subgroup analysis of three trials that included only high-risk Ta and T1 patients indicated no heterogeneity (P = 0.25) and a LHR for recurrence of −0.371 (0.012). With mitomycin C used as the control in the meta-analysis, a negative ratio is in favour of BCG and, in this case, was highly significant (P < 0.001). The seventh trial (in abstract form only) used BCG in low doses for two arms of the trial (27 mg and 13.5 mg) compared with a standard dose of mitomycin C (30 mg), and reported a significantly lower recurrence rate with BCG (27 mg) than for mitomycin C (P = 0.001). Only two trials included sufficient data to analyse disease progression and survival, representing 681 patients (338 randomized to BCG and 343 to mitomycin C). There was no significant difference between mitomycin C and BCG for disease progression, with a LHR of 0.044 (0.04) (P = 0.16), or survival, at −0.112 (0.03) (P= 0.50). Adverse events were slightly more frequent with BCG. Local toxicity (dysuria, cystitis, frequency and haematuria) were associated with both mitomycin C (30%) and BCG (44%). Systemic toxicity, e.g. chills, fever and malaise, occurred with both agents (12% and 19%, respectively) although skin rash was more common with mitomycin C.
Tumour recurrence was significantly lower with intravesical BCG than with mitomycin C only in those patients at high risk of tumour recurrence. However, there was no difference in disease progression or survival, and the decision to use either agent might be based on adverse events and cost.
log hazard ratio
European Organization for Research and Treatment of Cancer.
The incidence of bladder cancer is increasing and represents the fifth most common cancer in European men and the fourth most common in the USA. Each year ≈ 12 000 new cases are diagnosed in the UK and > 50 000 in the USA [1,2]. Bladder cancer is more common in men than women, with a male/female ratio of ≈ 4 : 1. Most patients present with early-stage disease confined either to the urothelium (Ta) or the lamina propria (T1), and the standard treatment for these patients is transurethral resection (TUR) of all visible tumour. However, despite complete surgical resection, tumour recurs in about two-thirds of patients . High-risk patients are typically treated with adjuvant intravesical therapy to minimize the risk of tumour recurrence.
Many intravesical cytotoxic agents have some activity in superficial bladder cancer, including mitomycin C, adriamycin, thiotepa, epirubicin, bleomycin and cytosine-arabinoside. The antitumour antibiotic mitomycin C is widely used for treating superficial bladder cancer and yields response rates for tumour recurrence of 7–81%. In randomized clinical trials mitomycin C has been reported to be better than adriamycin  and thiotepa , and equivalent to epirubicin . Side-effects commonly associated with intravesical mitomycin C are chemical cystitis and contact dermatitis. Other forms of intravesical therapy include nonspecific immunotherapy with BCG, which has been used to treat superficial bladder cancer since 1976 . Many subsequent reports have advocated the efficacy of adjuvant BCG in reducing tumour recurrence in patients with bladder cancer [9,10], and in delaying disease progression .
Intravesical BCG is significantly more effective than TUR alone in the prophylaxis of Ta and T1 bladder cancer , but the role of BCG compared with other intravesical agents is not clear. The AUA Guidelines on early-stage bladder cancer  advocate the use of either BCG or mitomycin C for Ta and T1 bladder cancer, but do not define the criteria to decide on the choice of agents. Studies by the European Organization for Research and Treatment of Cancer (EORTC), comparing the efficacy of intravesical BCG with mitomycin C in superficial bladder cancer, have been inconclusive . Intravesical BCG is associated with serious side-effects in a few patients, e.g. sepsis, and consequently there is a clear need to establish the most appropriate agent for treating early-stage bladder cancer.
The aim of the present systematic review was to assess the relative efficacy and side-effects of intravesical BCG compared with mitomycin C in Ta and T1 bladder cancer from published randomized trials, and to undertake a meta-analysis if sufficient data were available.
Only randomized controlled trials comparing intravesical BCG with intravesical mitomycin C were considered relevant to this review. Included studies were those on adults with histologically confirmed Ta and T1 bladder cancer. Eligible patients were those who had a medium or high risk of tumour recurrence and/or progression; the former was defined as a solitary tumour at presentation and tumour recurrence at 3 months or multiple tumours at presentation and no tumours at 3 months, and the latter as multiple tumours at presentation and recurrence at 3 months . Medium/high risk for disease progression was as defined by Kurth et al., i.e. high tumour grade, tumour size > 3 cm and previous recurrence. Intravesical BCG of any schedule or strain was considered appropriate for inclusion, as was any dose or schedule of mitomycin C. Studies that included other intravesical agents and reported BCG and mitomycin C groups were considered for inclusion. The main outcome measures were tumour recurrence, disease progression, overall survival and toxicity.
Databases of publications were comprehensively searched to identify all relevant randomized clinical trials, in any language, comparing intravesical BCG with mitomycin C in superficial bladder cancer, and published before June 2003. The 51-step search strategy included an electronic search of Medline (available on request) which was then modified to search Excerpta Medica Database, the Cochrane Library, ISI Science Citation (EMBASE) Index, Database of Abstracts of Reviews of Effectiveness. Cancerlit and Healthstar databases were searched to their final date limit as both have ceased publication. Recent Proceedings of the American Society for Clinical Oncology were also hand-searched (1996–2003). In addition, the reference list contained within each primary reference was scrutinized for additional randomized trials.
The literature search was screened and by consensus the relevant articles obtained. Data were extracted from each identified paper independently by two reviewers and included information on the trial design, participants, the type of intervention and outcome measures. Trials were assessed for quality according to the concealment of treatment allocation and adequate descriptions of numbers and reasons for patient withdrawal. Studies were categorized according to whether the included patients were at high risk for recurrence and/or disease progression. In studies where data from low- and high-risk patients were not analysed separately, or where there was doubt whether low-risk patients had been included, the patients were classified as ‘all risk’.
For time-to-event data the log hazard ratio (LHR) and its variance (in parentheses) were determined. Trials were combined by calculating the mean LHR, weighted by the reciprocal of the variance . To assess comparability of outcome data, chi-square tests for heterogeneity were used and sensitivity analyses carried out where considered appropriate.
In all, 25 published reports of randomized trials comparing BCG with mitomycin C were identified. Where there was more than one publication of the same trial, the most recent one was used for assessment. Fourteen articles were excluded as repeat publications, and four further articles comparing mitomycin C with an alternating regimen of BCG and mitomycin C and were also excluded. The remaining reports of seven different randomized trials comparing intravesical BCG with mitomycin C were considered for assessment [18–24].
In all, 2442 patients were entered into these seven trials and of these 2288 (94%) were eligible for inclusion. For some patients data were missing or unavailable, so that altogether 1901 patients were evaluable for the present review. One trial  was in abstract form only and had insufficient data to accurately determine the LHR. Therefore only six of the seven trials were used for the pooled meta-analysis, representing 1527 patients; 693 randomized to mitomycin C and 834 to BCG. The male to female ratio for all seven trials was ≈ 4–1, with a mean age of 65–68 years and median follow-up 2–7.2 years.
A German study enrolled 337 patients with stage pTa or pT1 grades 1–3 bladder cancer from 15 participating clinics . This was a three-armed study with 122 patients undergoing TUR alone, 112 receiving intravesical mitomycin C after TUR and 102 receiving intravesical BCG. Data from the mitomycin C and BCG arms only were analysed in this review. Intravesical mitomycin C was given at a dose of 20 mg every 2 weeks for 1 year, then monthly for 1 year, and BCG Connaught 120 mg was given weekly for 6 weeks, then monthly for 4 months. At a mean follow-up of 20.2 months, 30 (27%) patients given mitomycin C and 26 (25%) receiving BCG developed at least one tumour recurrence. In a log-rank test there were no significant differences among the three arms in those cases stratified for primary tumours (P = 0.255). In the case of recurrent tumours, while the three arms differed significantly (P = 0.007), there was no significant difference between BCG and mitomycin C (P = 0.191). To obtain a LHR between BCG and mitomycin C directly, this study used a Cox regression model. Both high- and low-risk patients were included in the analysis and therefore, for the purposes of this review, these patients were classified as ‘all risk’. Comparing mitomycin C with BCG for the time to first recurrence, the calculated LHR was 0.1948 (0.0729), and not significant (P = 0.47). In that study tumour progression was defined as an increased level of tumour stage and/or grade. The observed progression rate for mitomycin C was 2.0% per year, compared with 4.5% per year for BCG, and the difference among the three treatment arms was not statistically significant (chi-square, P= 0.207).
A joint Norwegian and Swedish study  reported on the 5-year follow-up of a randomized prospective trial comparing mitomycin C with BCG in 250 evaluable patients. Patients with stage Ta grades 1–3 or stage T1 grades 1 and 2 were included if they had had at least three tumour recurrences during the previous 18 months. Those with T1 grade 3, or primary or concomitant carcinoma in situ were included even if no previous tumour events were recorded. These patients were classified as ‘high risk’ for tumour recurrence. Patients were randomized to receive either 20 mg mitomycin C or 120 mg BCG Danish strain 1331 weekly for 6 weeks then monthly for 1 year and every 3 months for 2 years. At a median follow-up of 64 months, 42 of the 125 in the mitomycin C group were disease-free (34%) compared with 59 of the 125 in the BCG group (47%, P = 0.04). The LHR for recurrence, determined from the log-rank statistic quoted in the report , yielded −0.242 (0.0389) in favour of BCG, which was not significant (P = 0.22). Tumour progression, defined as increasing tumour stage, occurred in 27 patients (22%) in the mitomycin C arm and 20 (16%) in the BCG arm (P = 0.39). A total of 74 patients died (30%), giving an overall 5-year survival of 80% and 75% for mitomycin C and BCG, respectively (P = 0.9).
Intravesical BCG was compared with mitomycin C in a randomized intergroup study conducted by the South-west Oncology Group, USA . This study included 469 patients with completely resected stage Ta or T1 bladder tumours and all were considered to be at ‘high risk’ of recurrence, based on having had two recurrences within 56 weeks, a stage T1 tumour, two or three recurrences within 16 weeks, or concurrent carcinoma in situ. In all, 377 eligible patients received intravesically either 50 mg of Tice BCG or 20 mg mitomycin C weekly for 6 weeks then monthly for a year. These data were analysed using proportional hazard regression, giving direct LHRs. The recurrence-free interval was significantly longer with BCG than with mitomycin C and the LHR for recurrence (or death) was −0.3436 (0.0214) in favour of BCG (P = 0.017). Of the 178 patients with disease recurrence (47%), 39 had evidence of worsening disease, with 15 in the BCG arm and 24 in the mitomycin arm (proportional hazard regression, P = 0.482). Of the 53 patients who died 25 were in the BCG arm and 28 in the mitomycin C arm (proportional hazard regression, P = 0.572).
In a Finnish prospective multicentre randomized trial comparing mitomycin C with BCG, 109 patients with Tis grades 1–3 (18) or Ta-T1 grades 1–3 (91) were enrolled from 1984 to 1987 . An inclusion criterion was a minimum of two episodes of tumour recurrence before the study, and consequently these patients were considered as ‘high risk’ for recurrence. After TUR patients were randomly allocated to receive intravesically 2 weeks later either mitomycin C (20–40 mg) or BCG (Pasteur F, 6 × 108 per 50 mL). The instillations were repeated weekly during the first month, and then once a month for a 2-year period. In Ta and T1 patients there was a complete response in 67% with mitomycin C and 90% with BCG after 12 months of follow-up. The LHR for tumour recurrence and its variance were calculated directly from the presented Kaplan-Meier plots using the intervals method of Parmar et al. and were −0.868 (0.1074), significantly in favour of BCG (P = 0.001).
The Dutch Cooperative Group evaluated the efficacy of intravesical mitomycin C vs either BCG-Tice or BCG-RIVM in 469 patients with primary or recurrent pTa/T1 carcinoma and Tis of the urinary bladder . In all, 437 (93%) patients were eligible and of these 43 (10%) had solitary pTa grade 1 tumours, which are generally considered to be at low risk for recurrence; therefore patients in this study were classified as ‘all risk’. Mitomycin C 30 mg was instilled once weekly for 1 month and thereafter once a month for 6 months. Both BCG strains (5 × 108 per 50 mL saline) were given once a week for 6 weeks and, if recurrence was detected at 6 weeks, a second course of six weekly doses was given. For patients with pTa and T1 tumours the estimated percentages rendered disease-free at 5 years were 57% for mitomycin C, 54% with BCG-RIVM and 36% for BCG-Tice. Mitomycin C and BCG-RIVM treatment were equally effective (log rank P = 0.53), mitomycin C was significantly more effective than BCG-Tice (log rank P = 0.01), whereas there was no significant difference between the BCG strains (log rank P = 0.07). The LHR for recurrence was calculated from the quoted P values and the number of events in each arm , and was 0.1147 (0.0334) for mitomycin C vs. BCG-RIVM (P = 0.53) and 0.4492 (0.0304) for mitomycin C vs BCG Tice (P = 0.01), favouring mitomycin C. Minimal data on progression were presented and survival was not reported.
The long-term follow-up of an EORTC randomized prospective trial comparing intravesical BCG with mitomycin C in superficial bladder cancer was reported . From January 1985 to October 1986, 361 patients from 24 institutions were entered and of these 344 (95%) were eligible; 171 in the BCG and 173 in the mitomycin C arm. Patients with primary or recurrent pTa and pT1 tumours were included in this study and were classified in the present review as ‘all risk’, as there was no information on the multiplicity of primary tumours. Mitomycin C was instilled (30 mg) once a week for a month and thereafter once a month for a total of 6 months. BCG-RIVM (5 × 108 per 50 mL saline) was administered once a week for 6 weeks. Hazard ratios were quoted in the text. After a median follow-up of 7.2 years the time to first recurrence was similar for both therapeutic regimens, with a small and insignificant trend towards an improved outcome after mitomycin C, with an LHR of 0.1393 (0.0270) (P = 0.396). There was progression to muscle-invasive disease in 33 (9.6%) of patients, again with an insignificant benefit after mitomycin C, with a LHR of 0.4943 (0.1207) (P = 0.155). In 14 patients there was progression during treatment (11 in the BCG and three in the mitomycin C group), and in 19 patients after treatment (10 for BCG and nine for mitomycin C). Overall 33 (10%) patients died from malignant disease, 15 (46%) in the BCG and 18 (54%) in the mitomycin C arm. The difference was not statistically different, with a LHR of 0.0862 (0.0420) (P= 0.672).
Intravesical mitomycin C was compared to two low-dose schedules of BCG in the Spanish Cueto study 95011 , which was reported only as an abstract. Data were presented on 374 patients with stage Ta (37) or T1 (33) tumours with grades G1 (96) and G2 (278). Patients were randomized to receive mitomycin C (30 mg, 127), low-dose BCG (27 mg, 118) or very low dose BCG (13.5 mg, 129). Instillations were given once a week for 6 weeks followed by another six instillations every 2 weeks. The statistical methods used were not stated. At a median follow-up of 24.5 months, 40% of patients on mitomycin C had tumour recurrence, 11% on the low-dose BCG and 22% on the very low dose BCG (P = 0.001), with respective times to recurrence of 33.8, 47.2 and 42.9 months (P < 0.001). Progression, defined as stage progression to T2, occurred in 6% with mitomycin C, 3% with low-dose BCG and 5% with very low dose BCG, and was not statistically significant among the groups (P = 0.08). There were insufficient data in the abstract to calculate a LHR for tumour recurrence or progression and therefore the study was not included in the meta-analysis.
The meta-analysis was based on the method of Parmar et al. and calculates an overall weighted mean LHR and variance. When both strains of BCG from the study by Witjes et al. were included and considered as two separate analyses, the overall LHR was −0.0221 (0.00512). A positive LHR indicates a greater benefit from mitomycin C and a negative value a greater benefit with BCG. The overall LHR translates into < 3% difference in the probability of tumour recurrence per unit time between the intravesical regimens, which was not significantly different (P = 0.76), although there was significant heterogeneity among studies (P = 0.0011). In a sensitivity analysis, a similar result was found if the BCG-RIVM from Witjes et al. was omitted (overall LHR −0.0469, variance 0.006; P = 0.55) or by omitting only the BCG-Tice from the same study (overall LHR −0.1175, variance 0.006, P= 0.13).
In a further sensitivity analysis including those trials classified as high-risk [19–21] the overall LHR was −0.3714 (0.0122); this was highly significant in favour of BCG (P < 0.001) with no evidence of heterogeneity (P = 0.25). This indicates that a 31% reduction in the probability of tumour recurrence per unit time was associated with BCG rather than mitomycin C in patients with a high risk of recurrence.
Only two papers presented sufficient data to enable the calculation of the LHR for disease progression and overall survival [20,23]. This represents a total of 721 patients, 362 randomized to BCG and 359 to mitomycin C, and with a median follow-up of 2.2–7.2 years. The combined LHR for progression  was 0.044 (0.039), and with no evidence of heterogeneity (P = 0.11), which was not significantly different between the treatment regimens (P = 0.82). The LHR for survival was −0.112 (0.027) (heterogeneity P = 0.84) and again was not significantly different (P = 0.45).
Cystitis (3–27% BCG, 2–21% mitomycin C) and allergy (1–34% BCG, 7–16% mitomycin C) were common side-effects of intravesical treatment. Drug-induced cystitis, dysuria, frequency/urgency and systemic side-effects such as chills, fever, malaise and nausea, were significantly more frequent in the BCG than in the mitomycin C groups [18,20–23]. One patient treated with BCG required cystectomy after ulcerating cystitis  and there were two cases of pulmonary pneumonitis after BCG treatment [21,22]. Three of 125 patients treated with mitomycin C developed a contracted bladder, one of 125 receiving initial BCG and two further cases after crossover to BCG . Overall, ≈ 30% of those patients receiving mitomycin C developed local toxicity compared with 44% with BCG, with respective values of 12% and 19% for systemic side-effects.
Tumour recurrence after TUR is a significant problem for both the patient and the clinician. However, it is clear that intravesical treatment with either BCG or mitomycin C is effective for the prophylaxis of Ta and T1 bladder cancer, and is superior to TUR alone [12,25]. The present meta-analysis compared the effectiveness of these two agents using data from six published randomized trials. The overall analysis indicates that both agents were equivalent in preventing tumour recurrence, although there was significant heterogeneity. The patient selection for each study may be one source of heterogeneity, particularly of tumour characteristics. Patients with solitary Ta grade 1 tumours have a low probability of recurrence and may be cured by TUR alone , and the inclusion of such low-risk patients in studies where the outcome is tumour recurrence might underestimate the likelihood of recurrences in patients with medium-high risk tumours. Therefore, this might minimize any potential differences in outcomes between agents. We therefore used a sensitivity analysis on trials that only entered high-risk patients based on the definitions reported by Hall et al.. The results of this sub-analysis indicate that intravesical BCG is significantly better than mitomycin C for reducing the risk of tumour recurrence, but both agents were equivalent for tumour progression and survival.
A recently published meta-analysis compared the recurrence rate and toxicity of intravesical BCG with mitomycin C in stages Ta and T1 bladder cancer, combining data from randomized and observational studies . That study reported an overall statistically significant benefit of BCG in reducing tumour recurrence compared with mitomycin C (odds ratio 0.56, 95% CI 0.38–0.84, P < 0.001). Although the present study was solely confined to high-grade evidence from randomized trials and used the LHR for time-to-event data, the previous report agrees with the findings from the present analysis. A second recent meta-analysis reported on the risk of disease progression with intravesical BCG after TUR . In a sub-analysis, there was no significant difference in the size effect when comparing BCG with mitomycin C (odds ratio 0.86), confirming the present analysis of disease progression.
A greater delay in tumour recurrence with intravesical BCG could represent a net patient benefit as fewer tumour resections would be required, and the patient's risk of recurrence should be carefully assessed before intravesical therapy is given. However, BCG appears to induce greater local and systemic toxicity than mitomycin C and, although most are manageable, this needs to be considered when deciding management policy for patients with Ta and T1 disease. The lack of definite superiority for either agent in terms of disease progression and survival may reflect the limited data available for the present meta-analysis. Only two studies could be included [20,23] and this reflects to some extent the poor reporting of trial data. In addition, both outcomes require a longer follow-up than for disease recurrence; some trials were too short to report on progression and survival. Future trial updates may provide further information about these outcomes. Alternatively, there may be no direct relationship between the mechanisms of tumour recurrence and disease progression, and therefore survival , so that an agent affecting recurrence may not necessarily affect progression and survival.
In the seven randomized trials analysed in the present study there was variation in the treatment schedules used, with BCG varying from 13.5 mg to 120 mg and mitomycin C from 20 mg to 40 mg. Clearly more research is needed to define the optimum dose and schedule for both mitomycin C and BCG. The use of low BCG doses  may lead to a reduction in side-effects and warrants further study.
It is unclear why intravesical BCG should be more effective than mitomycin C in high- risk patients. By definition, one of the characteristics of these patients’ tumours is a greater proliferative capacity than for tumours from low-risk patients, and this may be associated with specific chromosomal abnormalities . However, it is uncertain if any differences in gross tumour morphology, e.g. tumour size, will influence the respective response to either agent. It may be that in high-risk tumours the relatively reduced action of mitomycin C relates to poor drug absorption, as mitomycin C penetration depth is an important determinant of treatment effectiveness . There may also be a higher degree of drug resistance to mitomycin C in high-risk tumours, as the activity of this drug is attenuated in bladder cancer cells over-expressing the MDR-1 gene , an increased expression of intracellular glutathione  or metallothionein . Alternatively, rather than mitomycin C being less active in high-risk patients, it could be that BCG has enhanced activity. This could be through increased expression of fibronectin receptors which are required for the endocytosis of the bacterium into the urothelium , but further studies are required to confirm this. In addition, a greater understanding of the causes of bladder cancer, and the natural history and mechanisms of tumour recurrence at the molecular level, would provide data to better explain why patients at high risk for tumour recurrence should benefit more from intravesical BCG than mitomycin C. This may also help elucidate the relationship between tumour recurrence and disease progression.
This review was prepared under the aegis of the Cochrane Collaboration.