Study Type – Therapy (RCT)
Level of Evidence 1b
Study Type – Therapy (RCT)
Level of Evidence 1b
non-muscle-invasive bladder cancer
Southwest Oncology Group
transurethral resection of bladder tumour
BCG intravesical instillation is the most effective immunotherapy for intermediate and high-risk non-muscle-invasive bladder cancer (NMIBC) [1–7]. Lamm et al.  reported in the Southwest Oncology Group (SWOG) 8507 study that 3-year BCG maintenance therapy markedly prolonged the recurrence-free survival (RFS) and time to disease progression in comparison with the conventional 6-week induction therapy. Since that report, other randomized, comparative clinical studies have been carried out, but they used different treatment regimens and were unable to verify the efficacy of BCG maintenance therapy [8–10]. To date, the SWOG 8507 study has been the only report to clearly demonstrate the efficacy of BCG maintenance therapy by means of a randomized comparative clinical trial. In consideration of the present situation, Herr  recently argued that we have not yet generated sufficient evidence to support the efficacy of BCG maintenance therapy for bladder cancer.
In the SWOG 8507 study, stratified analysis of the efficacy of BCG maintenance therapy in preventing disease recurrence in the study, which enrolled patients with recurrent or multiple stage Ta or T1 NMIBC and patients with carcinoma in situ, showed that the efficacy was most striking in patients with stage Ta and T1 disease . However, that study was not controlled by inclusion of either a transurethral resection of bladder tumour (TURBT) monotherapy group or an anti-cancer drug instillation group. In addition, it was reported that the maintenance therapy had to be discontinued in many patients because of exacerbation of drug-related adverse events (AEs) and that only 16% of the patients were able to complete the full treatment schedule covering 36 months . With regard to clinical research performed in Japan, the Japanese Urological Cancer Research Group carried out a combined analysis and reported that recurrence manifested within 500 days of surgery in most patients with stage Ta or T1 bladder cancer .
The present clinical trial was designed to compare and confirm the recurrence-preventing efficacy of BCG Connaught strain maintenance therapy in patients with stage Ta or T1 bladder cancer. A randomized, comparative clinical study was designed that included three treatment groups: a BCG maintenance group, a non-maintenance group and an epirubicin (EPI) group as an active control. The duration of the BCG maintenance therapy was set at 18 months because that period was thought to present the greatest risk of disease recurrence  and would also make it possible to achieve a certain level of treatment compliance.
The protocol for this clinical trial was approved by the Institutional Review Board of each participating institution. In addition, the trial was initiated only after having obtained written informed consent from each of the enrolled patients, and it was conducted in accordance with good clinical practice.
Patients who had been diagnosed with recurrent or multiple tumours of bladder cancer, had undergone excision of all of their tumours by TURBT, and were confirmed to have stage Ta or T1 transitional cell carcinoma of the bladder by histopathological analyses were included in this trial if they fell into any of the following three categories: (a) the presence of at least three bladder tumours at the time of TURBT (b) the present recurrence is at least the third such event or (c) the present recurrence was diagnosed within 12 months from the date of previous TURBT for the NMIBC.
The patients satisfied all of the following inclusion criteria: age between 20 and 79 years; no history of intravesical instillation of BCG or an anthracycline anti-tumour drug within the 12-month period following the day on which TURBT was performed; one course of BCG treatment that had been performed more than 12 months earlier was permitted; for patients who had been intravesically administered an anti-cancer drug such as mitomycin C or cytarabine or had received an oral antimetabolite, a washout period of at least 4 weeks was required, but administration of other anti-cancer treatments after the present TURBT was not permitted; white blood cell count 3300/mm3 to 10 000/mm3; platelet count ≥10 × 104/mm3; haemoglobin: ≥9.5 g/dL; aspartate and alanine aminotransferases no more than double the upper limit of normal for the performing institution; serum total bilirubin ≤2 mg/dL; serum creatinine ≤1.5 mg/dL; and Eastern Cooperative Oncology Group performance status 0–2.
In addition, patients were excluded from this trial if they met any of the following exclusion criteria: presence of carcinoma in situ lesion in the bladder; stage T2 or higher muscle-invasive bladder cancer, or a history thereof; presence of a malignant tumour of the upper urinary tract or the urethra as a complication, or a history thereof; a history of intravenous administration of an anti-cancer agent or intra-arterial chemotherapy for bladder cancer, or a history of radiation therapy; currently being administered such drug therapies as an immunosuppressive dose of a steroid; congenital or acquired immunodeficiency; active double cancers; showing a strongly positive tuberculin reaction; active tuberculosis or currently undergoing treatment with anti-tubercular drugs; serious bacterial infections of the urinary tract or, in the 4 weeks before registration for this trial, a serious viral or bacterial infection; a history of serious hypersensitivity or having some other serious complication; presence of a contracted bladder; female patients who are lactating or may be pregnant; or other patients whom the investigator considers to be inappropriate for inclusion in the study.
Figure 1 shows the study design. Patients who had undergone excision of all of their bladder tumours by TURBT were registered with the Central Registration Centre. The centre then randomly allocated them to three treatment groups: a BCG maintenance group (hereinafter, ‘maintenance group’), a BCG induction monotherapy group (hereinafter, ‘non-maintenance group’) and an EPI group.
The allocation was performed by a minimization method  that employed the main background factors (primary and multiple tumours, recurrent and solitary tumours, and recurrent and multiple tumours; whether or not BCG intravesical instillation had been performed previously; whether or not anthracycline anti-tumour drug intravesical instillation had been performed previously; the institution) as allocation factors.
Treatment of each patient was to be started no later than within 1 month following TURBT. For the patients in the non-maintenance group, 81 mg of the BCG Connaught strain (ImmuCyst®; manufactured by Sanofi Pasteur Ltd., Toronto, Canada) was first suspended in 3 mL diluent, followed by dilution with 40 mL physiological saline. That suspension was then instilled into the bladder via a urethral catheter, and the patient was instructed to endeavour to retain it for 2 h, and only then to urinate. This intravesical instillation procedure was to be performed once weekly, for a total of six treatments. For the maintenance group, the 6-week induction therapy (i.e. the above BCG instillation protocol) was performed first, and then maintenance therapy was carried out by performing additional BCG intravesical instillations according to a schedule of once weekly for 3 weeks, at each of 3, 6, 12 and 18 months from the date when the induction therapy had been started. EPI (Farmorubicin®; Pfizer Japan, Tokyo, Japan), 40 mg, was dissolved in 40 mL physiological saline and instilled into the bladder. The patient was to retain that solution for 1–2 h and then urinate. This EPI intravesical instillation was performed twice at a 1-week interval and then seven times at 2-week intervals. This administration regimen was adopted from that reported by the Japanese Urological Cancer Research Group for a series of clinical studies [15–17] and which had yielded good results .
Efficacy of the treatment regimens was assessed on the basis of cystoscopy and urinary cytology findings. Patient follow up was performed at 3-month intervals for the first 3 years, and then at 6-month intervals beginning from the 4th year. When tumour recurrence was suspected on the basis of the findings of cystoscopy and urinary cytology, confirmation was carried out by performing TURBT. The date of confirmation of recurrence or progression was defined as the earliest date on which the diagnosis had been made based on the findings of cystoscopy, urinary cytology or some other diagnostic imaging technique.
Progression at the time of recurrence was defined as progression in the disease stage (i.e. progression to stage T2 or higher, metastasis, or appearance of tumours in the upper urinary tract or the urethra) or as exacerbation of the tumour grade.
For assessment of the safety of each of the treatment regimens, the severity of AEs was judged in accordance with the criteria stipulated in NCINational Cancer Institute-Common Toxicity Criteria, Version-2.0 .
The primary endpoint was the RFS from randomization and the secondary assessment endpoints were the progression rate at the time of first recurrence and the safety. The RFS was estimated by the Kaplan-Meier method. As the primary analyses, the study protocol called for two comparisons regarding the RFS. The first-step comparison was of the combined BCG group (the maintenance group plus the non-maintenance group) with the EPI group. Only if the result for the RFS was statistically superior for the combined BCG group compared with the EPI group was the second-step comparison performed, comprising a comparison of the maintenance group with the non-maintenance group. The treatment groups were compared using the generalized Wilcoxon test. The necessary number of patients in each treatment group was calculated for each comparison on the basis of assuming a registration period of 2.5 years, a study period of 4.5 years, an 80% power of the test and a 5% type one error. The expected 2-year recurrence-free rates for each group were estimated to be 85% for the maintenance group, 60% for the non-maintenance group and 45% for the EPI group, based on the earlier reported results [1,15–17]. Accordingly, the necessary numbers of patients were estimated to be 52 patients in the combined BCG group and 26 patients in the EPI group for the first-step comparison, and 36 patients in each of the treatment groups for the second-step comparison. When consideration was given to the possibility of dropouts, the total target number of patients was set at 110, consisting of 40 patients in each of the maintenance and non-maintenance groups and 30 patients in the EPI group. For unequal treatment allocation, the biased-coin minimization method  was used in the allocation programme. Smoothed hazards analysis [7,12,20] was applied to estimate the hazard of recurrence in each of the treatment groups.
There had initially been no plan to perform an interim test for significant differences during the course of this trial. However, as the trial progressed, it became evident that there were more recurrent cases in the EPI group than had been expected at the start. This raised the question of whether, from an ethical standpoint, the trial should be terminated earlier than scheduled. To address this issue, the Independent Efficacy and Safety Assessment Committee for this trial recommended revising the analytical plans of the study. As a result, new conditions for termination of the trial were established, and it was decided to perform interim data analyses to determine the advisability of early termination. The date of cut-off for interim data analyses was set as the last day of August 2008, which was 4 years from the date of registration of the first patient in the trial instead of the initially planned 4.5 years. The main statistical test was the generalized Wilcoxon test, with a P value of 0.035 indicating statistical significance, which was calculated using an α consumption function that permitted adjustment for the type one error for the study as a whole.
Figure 2 explains the grouping of the patients in this clinical trial. A total of 120 patients were enrolled in the trial at 37 participating institutions between August 2004 and the end of December 2006. Following registration and before the start of treatment, four patients were excluded. The enrolled patients were randomly allocated to the three treatment groups after registration, and a total of 116 patients were started on their respective treatments. One of the patients in the maintenance group retracted consent and was excluded from the efficacy analysis. Hence, the ‘full analysis set’ analytical population numbered 115 patients in total.
Analysis of the per protocol set analytical population was carried out as a secondary endpoint of analysis. The study protocol stipulated that, for a maintenance group patient to be included in the per protocol set analysis, it was necessary for the patient to have been administered at least one BCG intravesical instillation as maintenance therapy.
Table 1 presents the data on the background characteristics of the FAS population.
|Characteristic||BCG* 83||EPI32||P value||M41||Non-M42||P value†|
|EORTC Recurrence score|
|Intermediate risk (1–9)||72||25||0.2637||36||36||1.0000|
|High risk (≥10)||11||7||5||6|
Figure 3 shows, the respective RFS plots for the combined BCG group and the EPI group, and for the maintenance group and the non-maintenance group.
In the first step analysis, the generalized Wilcoxon test showed that the RFS was significantly prolonged in the combined BCG group compared with the EPI group (P < 0.0001). Accordingly, comparison of the RFS between the maintenance group and the non-maintenance group was carried out as the second step analysis. That analysis showed the RFS to be significantly prolonged in the maintenance group compared with the non-maintenance group (generalized Wilcoxon test: P= 0.0190). This result for the main analyses satisfies the criterion for statistical significance that was stipulated for the interim analyses. For that reason, this result was considered to represent the final analysis, and the clinical trial was terminated early.
In the comparison of the per protocol set analytical population, only two of the 36 patients in the maintenance group experienced recurrence. That analysis also showed the RFS to be significantly prolonged in the maintenance group compared with the non-maintenance group (generalized Wilcoxon test: P= 0.0007).
Figure 4 presents the results of smoothed hazards analysis of the recurrence in each of the treatment groups. In the EPI group, a high recurrence hazard continued for 2 years after TURBT, whereas in the non-maintenance group the recurrence hazard decreased from the half-year point onward. On the other hand, in the maintenance group, in addition to suppressing the initial hazard to about half of that in the non-maintenance group, a low hazard was maintained for up to approximately 2 years after TURBT.
Table 2 presents the data on the numbers of patients in each of the treatment groups who showed progression at the time of first recurrence. The progression rates were compared between the combined BCG group and the EPI group and between the maintenance group and the EPI group. Fisher’s exact test showed statistically significant differences in those comparisons (P= 0.0047 and P= 0.0021, respectively). When the maintenance group and the non-maintenance group were then compared, no statistically significant difference was found (Fisher’s exact test: P= 0.2410). However, there were no cases of progression in the maintenance arm.
|No. of patients||Patients with recurrence, n (%)||Patients with progression, n (%)||Nature of progression|
|Stage progression to ≥T2||Extracystic progression*||Exacerbation of tumour grade|
|Maintenance group||41||5 (12.2)||0 (0)||0||0||0|
|Non-maintenance group||42||14 (33.3)||3 (7.1)||0||2†||2†|
|Epirubicin group||32||22 (68.8)||7 (21.9)||1||2||4|
The incidence of treatment-related AEs was higher in the maintenance group than in the non-maintenance group (Table 3). However, all of these symptoms could be controlled by suspending the treatment or administering therapy such as an anti-inflammatory or analgesic. Even for the maintenance treatment cycle, at the 18-month point approximately 70% of the patients were able to be given at least one of the three planned instillations (Table 4).
|Event||Maintenance arm (N= 42)*||Non-maintenance arm (N= 42)*|
|Pts n||Incidence (%)||≥Grade 3 (%)||Pts n||Incidence (%)||≥Grade 3 (%)|
|Pain on urination||39||92.9||9.5||29||69.0||2.4|
|Difficulty in urination||21||50.0||4.8||12||28.6|
|Lower abdominal pain||4||9.5||2||4.8||2.4|
|Pyrexia (≥38 °C)||18||42.9||11||26.2|
|Urinary protein positive||20||47.6||10||23.8|
|Urinary red blood cell increase†||29||69.0||24||57.1|
|Urinary white blood cell increase†||36||85.7||31||73.8|
|Cycle||N||Performance rate||Completed rate|
|At 3 months||36||36||100.0||32||88.9|
|At 6 months||36||35||97.2||23||63.9|
|At 12 months||36||30||83.3||19||52.8|
|At 18 months||36||25||69.4||15||41.7|
The incidence of treatment-related AEs was clearly lower in the EPI group patients compared with in the BCG groups. Grade 3 or more severe treatment-related AEs consisted of gross haematuria (3.1%) and malaise (3.1%).
The SWOG 8507 study did not include either a TURBT-only group or a chemotherapeutic drug instillation group. Our present study included an EPI group as an active control group. The design made it possible to clearly show the difference in efficacy among the three treatment groups. The 2-year RFS rate for the BCG maintenance group was 92.7%, which was markedly better than the rates of 65.4% for the non-maintenance group and 33.2% for the EPI group. Two-year RFS rates of approximately 80% and 60% were estimated for the maintenance group and the non-maintenance group in the SWOG 8507 study, respectively . In addition, Saint et al.  carried out a clinical study that administered the BCG Connaught strain according to a 3-year maintenance schedule, and they reported a 2-year RFS rate of 84.9%. Furthermore, van der Meijden et al.  reported an estimated 2-year RFS rate of 70% with maintenance therapy that employed the BCG Tice strain. It is clear that the 92.7% 2-year RFS rate was reported in our study is significant and similar to the results in those earlier studies.
The SWOG 8507 study found that maintenance therapy significantly prolonged the period until disease progression (log rank test, P= 0.04) . In addition, Bohle and Bock  performed a meta-analysis of 2410 patients to compare the efficacy of BCG and mitomycin C in suppressing disease progression. They reported that administration of BCG maintenance therapy for at least 1 year resulted in significantly superior suppression of the risk of disease progression that was 34% greater than that with mitomycin C (OR = 0.66; 95% CI 0.47–0.94; P= 0.02). Moreover, Sylvester et al.  carried out a meta-analysis of 4863 patients in which they compared the therapeutic results in a BCG group, a TURBT-only group and an anti-cancer drug group. They reported that the BCG group showed statistically significant suppression of disease progression (OR = 0.73; 95% CI 0.60–0.89; P= 0.001). Also, when the analysis was performed for patients who received some form of BCG maintenance therapy for at least 1 year, striking efficacy was demonstrated (OR = 0.63; 95% CI 0.51 P= 0.79; P= 0.00004). In addition, a recent systematic review  by the Cochrane group concluded that BCG maintenance significantly reduced recurrence in patients with non-muscle-invasive bladder cancer. On the basis of these and other reports, the guidelines for treatment of bladder cancer all recommend that BCG maintenance therapy be continued for at least 1 year [26–31].
On the other hand, Herr  recently argued that there is insufficient evidence indicating that BCG maintenance therapy shows significant efficacy in suppressing disease progression. However, most events of progression are detected during the course of repeated recurrence in the bladder, and it can therefore be thought that achieving RFS over a long period of time by performing maintenance therapy will also reduce the risk of progression. Actually, in our study no progression was observed in the maintenance group. The efficacy results for BCG maintenance therapy that we have demonstrated in our present randomized trial not only support the findings reported by Lamm et al.  but are also able to partially counter the recent discussion regarding BCG maintenance therapy [11,32–34].
For the tolerability evaluation of 81 mg of BCG dose, our present protocol stipulated that the dose of BCG could not be reduced, because one objective of the study was to confirm the tolerability of BCG when administered in a dose of 81 mg. Treatment-related AEs were clearly more severe during the maintenance therapy. It was reported that treatment-related AEs could be reduced, without sacrificing efficacy, by lowering the BCG dose [35,36] or by administering a quinolone antibiotic 8 h after BCG instillation . We think that future studies are warranted to investigate the possibility of further increasing the duration of long-term prevention of bladder cancer recurrence by improving treatment compliance through better management of treatment-related AEs during the course of BCG maintenance therapy.
In conclusion, this randomized, comparative clinical trial demonstrated that BCG intravesical instillation maintenance therapy is able to significantly prolong post-TURBT RFS in patients with recurrent or multiple, stage Ta or T1, bladder cancer that is at moderate-to-high risk of recurrence. The results of this clinical trial represent valuable new evidence in support of the efficacy of BCG maintenance therapy in preventing bladder cancer recurrence.
We appreciate the cooperation afforded by all of the patients and institutions involved in this study. In Japan, neither the usage of BCG adjuvant treatment after TURBT nor a schedule of maintenance therapy for NMIBC has yet been approved. Accordingly, the present clinical study was carried out with the objective of supporting an approval application for BCG maintenance therapy for preventing the recurrence of NMIBC in Japan. The trial is supported by Nippon Kayaku Co. Ltd (Tokyo, Japan), which is the current Japanese licence holder for the BCG Connaught strain.
The authors wish to thank the investigators for performing this study. Investigators and institutions: T. Tsukamoto, Sapporo Medical University Hospital; T. Ohyama, Hirosaki University School of Medicine and Hospital; A. Ito, Tohoku University Hospital; N. Tsuchiya, Akita University Hospital; N. Miyanaga, Tsukuba University Hospital; K. Tanabe, Tokyo Women’s Medical University Hospital; E. Kikuchi, Keio University Hospital; I. Fukui, The Cancer Institute Hospital of JFCR; K. Tomita, The University of Tokyo Hospital; S. Egawa, The Jikei University Hospital; S. Komatsubara, Niigata Cancer Centre Hospital; S. Mugiya, Hamamatsu University School of Medicine, University Hospital; Y. Hirano, Fujieda Municipal General Hospital; H. Kobayashi, Nagoya Daini Red Cross Hospital; K. Kohri, Nagoya City University Hospital; K. Miki, University Hospital, Kyoto Prefectural University of Medicine; H. Nishiyama, Kyoto University Hospital; M. Usami, Osaka Medical Centre for Cancer and Cardiovascular Disease; N. Nonomura, Osaka University Hospital; K. Fujimoto, Nara Medical University Hospital; T. Inagaki, Wakayama Medical University Hospital; M. Takahashi, Tokushima University Hospital; M. Inui, Faculty of Medicine, Kagawa University; A. Yokomizo, Kyushu University Hospital; M. Nakagawa, Kagoshima University Medical and Dental Hospital. The Protocol and Evaluation Committee: Y. Kubota, Yokohama City University Hospital; T. Habuchi, Akita University Hospital; H. Kanetake, Nagasaki University Hospital. Coordinating Committee: T. Tsukamoto, Sapporo Medical University Hospital; Y. Hirao, Nara Medical University Hospital.