- Email: [email protected]
BCG-refractory vs. BCG-relapsing non–muscle-invasive bladder cancer: A prospective cohort outcomes study
Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
Original article
BCG-refractory vs. BCG-relapsing non–muscle-invasive bladder cancer: A prospective cohort outcomes study Harry W. Herr, M.D.*, Tanya N. Milan, B.S., M.P.H, Guido Dalbagni, M.D. Department of Urology, Memorial Sloan-Kettering Cancer Center, New York, NY Received 4 February 2014; received in revised form 25 February 2014; accepted 25 February 2014
Abstract Purpose: Patients with recurrent or persistent high-grade non–muscle-invasive bladder cancer after bacille Calmette-Guérin (BCG) therapy are termed “BCG failures.” We hypothesize that BCG-refractory patients who fail to respond to BCG have worse outcomes after bladder-sparing treatments compared with BCG-relapsing patients whose tumors recur after at least a 6-month disease-free interval. Materials and methods: We screened 32 patients who had failed BCG therapy for eligibility in a multicenter investigational trial. BCGrefractory patients received instillations of a mycobacterial cell wall–DNA complex extract. BCG-relapsing patients were treated with additional BCG or other intravesical agents. Both groups of patients were followed prospectively with transurethral biopsy after 6 months, and cystoscopy every 3 to 6 months for more than 2 years. Median follow-up time for all patients was 53 months (range: 24–72 mo). Results: Seventeen patients were classified as BCG refractory and 15 patients defined BCG relapsing. Recurrence-free median survival time was 10 months for BCG-refractory patients receiving mycobacterial cell wall–DNA complex vs. 23 months for BCG-relapsing patients who received another induction course of BCG therapy (P ¼ 0.002). Progression-free survival time was 18 months for BCG-refractory vs. 52 months for BCG-relapsing patients (P ¼ 0.001). Of the 17 BCG-refractory patients, 8 (47%) have died vs. 3 (20%) of the 15 BCGrelapsing patients. Conclusions: BCG-refractory and BCG-relapsing categories differentiate BCG-failed patients into high-and lower-risk prognostic groups that may be useful in guiding treatment strategies. r 2015 Elsevier Inc. All rights reserved.
Keywords: BCG failure; Bladder cancer
1. Introduction Intravesical bacille Calmette-Guérin (BCG) is the standard adjuvant therapy after transurethral resection of highgrade non–muscle-invasive bladder cancer. Unfortunately, not every patient is cured with BCG therapy. Some patients fail to respond to induction BCG and are never rendered free of disease. They are defined as BCG refractory [1]. Others who initially respond to BCG may have a recurrence of tumor after variable disease-free intervals, usually less than 2 years from the start of induction BCG therapy. They are labeled BCG relapsing [2]. Both groups of patients— BCG refractory and BCG relapsing—are often lumped together as “BCG failures” [3]. Such patients are believed Corresponding author. Tel.: þ1-646-422-4411; fax: þ1-212-988-0768. E-mail address: [email protected] (H.W. Herr). *
http://dx.doi.org/10.1016/j.urolonc.2014.02.020 1078-1439/ r 2015 Elsevier Inc. All rights reserved.
to have a greater risk of tumor progression and they are considered for cystectomy or alternative intravesical treatments. On the contrary, if prognosis differs significantly among “BCG failures,” some patients may undergo an unnecessary cystectomy when they might still respond to conservative measures, and other patients may receive additional futile intravesical treatments when they would be better served by cystectomy. Additionally, if all patients whose tumor recur after BCG are lumped together simply as “BCG failures,” it obscures the real value intravesical alternatives to BCG may have in defined patients. We hypothesize that BCG-refractory and BCG-relapsing patients have different outcomes, especially in response to salvage bladder-sparing treatments; however, this hypothesis has never been validated in a prospective study. Recently, we had the unique opportunity to test our hypothesis while participating in a multicenter prospective
108.e2
H.W. Herr et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
trial to investigate the effect of mycobacterial cell wall– DNA complex (MCC) in patients who had failed BCG therapy [4].
Table Patient characteristics Variable
BCG refractory
BCG relapsing
2. Patients and methods
No. of patients
17
15
Age, y: median (range)
65 (51–81)
62 (37–83)
Sex: males
13 (76%)
11 (73%)
T-stage Ta T1
7 (41%) 10 (59%)
5 (33%) 10 (67%)
High grade
17 (100%)
15 (100%)
Carcinoma in situ
17 (100%)
15 (100%)
From August 2007 to January 2009, we screened 32 patients who were deemed to be BCG failures. They had multiple, recurrent, high-grade Ta or T1 cancers associated with carcinoma in situ. Although cystectomy was advised, they elected to pursue bladder-sparing strategies. Seventeen patients had BCG-refractory disease, defined as evidence of high-grade bladder cancer at least 6 months from the start of a full induction course of BCG, with or without maintenance BCG or retreatment at 3 months. They were accepted on the protocol to receive 6 weekly intravesical instillations of 8-mg MCC, followed by 3 once weekly planned instillations at 3, 6, 12, 18, and 24 months. Fifteen patients were classified as BCG relapsing. They became disease free by 6 months in response to induction BCG courses, but they developed recurrent cancers less than 2 years following the last dose of BCG. BCG-relapsing patients were not eligible to receive MCC because they did not fit the strict definition of BCG-refractory status. Some patients were also excluded because they had prior upper tract disease (2 patients), prostatic urethra involvement (1 patient) or remote history of muscle invasion (1 patient) cured by transurethral resection, and radiation therapy for localized prostate cancer (1 patient). BCGrelapsing patients opted to receive other intravesical therapies, consisting of BCG plus interferon or gemcitabine [5]. Both patient groups were followed up concurrently with cystoscopy, urine cytology, and biopsy as necessary every 3 months for 2 years, then every 6 months up to 5 years. Patients provided written informed consent for this institutional review board–approved study. End points were recurrence-free, progression-free and overall survival. Tumor recurrence included any evidence of disease after 3 months on follow-up evaluations. Tumor progression was defined as new or persistent lamina propria (pT1) invasion, muscle invasion, or metastases. Overall survival included deaths from any cause. Time to event was measured from the date of first MCC or other intravesical treatment. Patients who underwent cystectomy were censored for recurrence or progression, but they were included in survival analysis. Correlation between groups was performed using chi-square test. Survival times were calculated by Kaplan-Meier method and compared using the log-rank test. All tests were 2 sided and P o 0.05 were considered significant. 3. Results The Table shows patient characteristics of the 2 groups. All had high-volume recurrent non–muscle-invasive bladder
No. of 6-week BCG cycles: mean (range)
2
3.3 (2–5)
Diagnosis to BCG failure, mo: median (range)
6 (3–6)
18 (13–26)
11 (65%)
5 (33%)
Radical cystectomy
cancer and had failed BCG therapy. BCG-refractory patients had failed 2 induction courses of BCG within 6 months before starting MCC therapy. BCG-relapsing patients received an average of 3 or more 6-week courses of BCG and had a median time from diagnosis to BCG failure of 18 months (range: 13–26 mo). BCG-refractory patients received MCC therapy. BCG-relapsing patients received additional intravesical BCG, with or without interferon, or intravesical gemcitabine. Radical cystectomy was reserved for rapid tumor recurrence or progression to muscle invasion. Eleven BCG-refractory (65%) and 5 BCGrelapsing patients (33%) underwent salvage cystectomy. Pathologic stage in the BCG-refractory group was muscle invasion in 5 cases and 6 had recurrent pT1 tumors. In BCG-relapsing patients, 3 had carcinoma in situ only and 2 had pT1 disease. Of 11 BCG-refractory patients who had cystectomy, 7 were salvaged and 4 died of disease; of the 5 BCG-relapsing patients, 2 died and 3 survived after cystectomy. The mean follow-up time for all patients was 53 months (95% CI: 48–57 mo), 48 months (95% CI: 41–55 mo) for BCG-refractory patients, and 58 months (95% CI: 55–61 mo) for BCG-relapsing patients. All but 1 patient in both groups recurred. Fig. 1 shows median recurrence-free survival time was 10 months (95% CI: 6–17 mo) in BCG-refractory patients compared with 23 months (95% CI: 15–27 mo) in the BCG-relapsing patients (P ¼ 0.002). Tumors progressed in 14 of the 17 BCGrefractory patients (82%) compared with 5 of the 15 BCGrelapsing (33%) patients (P ¼ 0.04). Fig. 2 shows a median progression-free survival time of 18 months (95% CI: 7–28 mo) among BCG-refractory patients vs. a mean of 52 months (95% CI: 47–59 mo) in BCG-relapsing patients (P ¼ 0.001). Eleven patients have died (10 from bladder cancer and 1 from acute leukemia). Of the 17 BCG-refractory patients, 8 (47%) died vs. 3 of the 15
H.W. Herr et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
108.e3
Fig. 1. Recurrence-free survival. (Color version of figure is available online.)
Fig. 3. Overall survival. (Color version of figure is available online.)
BCG-relapsing patients (20%). Fig. 3 shows 54 months (95% CI: 24–72 mo) of median overall survival in BCGrefractory patients vs. median survival time not yet reached (mean ¼ 58 mo, 95% CI: 48–72 mo) among BCGrelapsing patients (P ¼ 0.04).
The major weakness of this study is the small numbers of patients. However, we believe that a few patients followed up prospectively according to prescribed protocol and recording real-time events as they unfold yields more reliable and robust information than retrospective analysis of databases confounded by selection bias, missing data, and unquantifiable variables. Segregating BCG failures into prognostic groups is not a new concept. Shirakawa et al. [6] defined 4 prognostic groups among BCG failures, with BCG-refractory patients having worse outcomes. The 10-year disease-specific survival rate in BCG-refractory and BCG-relapsing patients was 62% and 91%, respectively, similar to the 53% and 80% respective survival results from our study. Prognostic risk groups identified in retrospective analyses require prospective validation and that, albeit the small numbers of patients in our study, the information derived is persuasive. BCG refractory and BCG relapsing are clinically useful categories to define high-risk and lower-risk prognostic groups among patients who have failed BCG therapy. BCG refractory identifies patients who fail to respond to adequate BCG exposure, including maintenance BCG, and are never rendered disease free. BCG relapsing denotes later recurrences after a short-term (o2 y) disease-free interval. BCGrefractory patients are at risk for early tumor progression and cancer death and should be considered for immediate cystectomy. BCG-relapsing patients, although still at risk of dying of urothelial cancer, tend to have more indolent clinical courses favoring additional attempts at protocolbased bladder-sparing regimens. However, progression of disease and outcomes in both groups are poor, suggesting cystectomy should be discussed in all heavily pretreated BCG patients.
4. Discussion The major finding of this study is that BCG-refractory patients have worse outcomes in response to conservative salvage treatments than BCG-relapsing patients. Overall, 80% of BCG-relapsing patients survive compared with 53% of BCG-refractory patients with mature follow-up. Muscleinvasive tumor progression was more common in BCG-refractory than in BCG-relapsing patients, whose recurrences remained confined to bladder mucosa or submucosa. This is the first prospective study to discriminate different outcomes among high-risk and lower-risk cohorts of BCG failures.
References Fig. 2. Progression-free survival. (Color version of figure is available online.)
[1] Herr HW, Dalbagni G. Defining BCG refractory superficial bladder tumors. J Urol 2003;169:1706.
108.e4
H.W. Herr et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
[2] Saint F, Salomon L, Quintela R, et al. Do prognostic parameters of remission versus relapse after BCG immunotherapy exist? Analysis of a quarter century of literature. Eur Urol 2003;43:351. [3] Witjes JA. Management of BCG failures in superficial bladder cancer: a review. Eur Urol 2006;49:790. [4] Morales A, Herr H, Steinberg G, et al. Efficacy and safety of mycobacterial cell wall DNA complex in the treatment of patients with non-muscle invasive bladder cancer at high risk of progression
who are refractory to BCG. J Urol 2011;185:[No. 4 S (Abst 1650)]. [5] Sternberg IA, Dalbagni G, Chen LY, et al. Intravesical gemcitabine for high risk, nonmuscle invasive bladder cancer after Bacillus CalmetteGuérin treatment failure. J Urol 2013;190:1686–91. [6] Shirakawa H, Kikuchi E, Tanaka N, et al. Prognostic significance of BCG failure classification in non-muscle-invasive bladder cancer. BJU Int 2012;110:16–21.
Original article
BCG-refractory vs. BCG-relapsing non–muscle-invasive bladder cancer: A prospective cohort outcomes study Harry W. Herr, M.D.*, Tanya N. Milan, B.S., M.P.H, Guido Dalbagni, M.D. Department of Urology, Memorial Sloan-Kettering Cancer Center, New York, NY Received 4 February 2014; received in revised form 25 February 2014; accepted 25 February 2014
Abstract Purpose: Patients with recurrent or persistent high-grade non–muscle-invasive bladder cancer after bacille Calmette-Guérin (BCG) therapy are termed “BCG failures.” We hypothesize that BCG-refractory patients who fail to respond to BCG have worse outcomes after bladder-sparing treatments compared with BCG-relapsing patients whose tumors recur after at least a 6-month disease-free interval. Materials and methods: We screened 32 patients who had failed BCG therapy for eligibility in a multicenter investigational trial. BCGrefractory patients received instillations of a mycobacterial cell wall–DNA complex extract. BCG-relapsing patients were treated with additional BCG or other intravesical agents. Both groups of patients were followed prospectively with transurethral biopsy after 6 months, and cystoscopy every 3 to 6 months for more than 2 years. Median follow-up time for all patients was 53 months (range: 24–72 mo). Results: Seventeen patients were classified as BCG refractory and 15 patients defined BCG relapsing. Recurrence-free median survival time was 10 months for BCG-refractory patients receiving mycobacterial cell wall–DNA complex vs. 23 months for BCG-relapsing patients who received another induction course of BCG therapy (P ¼ 0.002). Progression-free survival time was 18 months for BCG-refractory vs. 52 months for BCG-relapsing patients (P ¼ 0.001). Of the 17 BCG-refractory patients, 8 (47%) have died vs. 3 (20%) of the 15 BCGrelapsing patients. Conclusions: BCG-refractory and BCG-relapsing categories differentiate BCG-failed patients into high-and lower-risk prognostic groups that may be useful in guiding treatment strategies. r 2015 Elsevier Inc. All rights reserved.
Keywords: BCG failure; Bladder cancer
1. Introduction Intravesical bacille Calmette-Guérin (BCG) is the standard adjuvant therapy after transurethral resection of highgrade non–muscle-invasive bladder cancer. Unfortunately, not every patient is cured with BCG therapy. Some patients fail to respond to induction BCG and are never rendered free of disease. They are defined as BCG refractory [1]. Others who initially respond to BCG may have a recurrence of tumor after variable disease-free intervals, usually less than 2 years from the start of induction BCG therapy. They are labeled BCG relapsing [2]. Both groups of patients— BCG refractory and BCG relapsing—are often lumped together as “BCG failures” [3]. Such patients are believed Corresponding author. Tel.: þ1-646-422-4411; fax: þ1-212-988-0768. E-mail address: [email protected] (H.W. Herr). *
http://dx.doi.org/10.1016/j.urolonc.2014.02.020 1078-1439/ r 2015 Elsevier Inc. All rights reserved.
to have a greater risk of tumor progression and they are considered for cystectomy or alternative intravesical treatments. On the contrary, if prognosis differs significantly among “BCG failures,” some patients may undergo an unnecessary cystectomy when they might still respond to conservative measures, and other patients may receive additional futile intravesical treatments when they would be better served by cystectomy. Additionally, if all patients whose tumor recur after BCG are lumped together simply as “BCG failures,” it obscures the real value intravesical alternatives to BCG may have in defined patients. We hypothesize that BCG-refractory and BCG-relapsing patients have different outcomes, especially in response to salvage bladder-sparing treatments; however, this hypothesis has never been validated in a prospective study. Recently, we had the unique opportunity to test our hypothesis while participating in a multicenter prospective
108.e2
H.W. Herr et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
trial to investigate the effect of mycobacterial cell wall– DNA complex (MCC) in patients who had failed BCG therapy [4].
Table Patient characteristics Variable
BCG refractory
BCG relapsing
2. Patients and methods
No. of patients
17
15
Age, y: median (range)
65 (51–81)
62 (37–83)
Sex: males
13 (76%)
11 (73%)
T-stage Ta T1
7 (41%) 10 (59%)
5 (33%) 10 (67%)
High grade
17 (100%)
15 (100%)
Carcinoma in situ
17 (100%)
15 (100%)
From August 2007 to January 2009, we screened 32 patients who were deemed to be BCG failures. They had multiple, recurrent, high-grade Ta or T1 cancers associated with carcinoma in situ. Although cystectomy was advised, they elected to pursue bladder-sparing strategies. Seventeen patients had BCG-refractory disease, defined as evidence of high-grade bladder cancer at least 6 months from the start of a full induction course of BCG, with or without maintenance BCG or retreatment at 3 months. They were accepted on the protocol to receive 6 weekly intravesical instillations of 8-mg MCC, followed by 3 once weekly planned instillations at 3, 6, 12, 18, and 24 months. Fifteen patients were classified as BCG relapsing. They became disease free by 6 months in response to induction BCG courses, but they developed recurrent cancers less than 2 years following the last dose of BCG. BCG-relapsing patients were not eligible to receive MCC because they did not fit the strict definition of BCG-refractory status. Some patients were also excluded because they had prior upper tract disease (2 patients), prostatic urethra involvement (1 patient) or remote history of muscle invasion (1 patient) cured by transurethral resection, and radiation therapy for localized prostate cancer (1 patient). BCGrelapsing patients opted to receive other intravesical therapies, consisting of BCG plus interferon or gemcitabine [5]. Both patient groups were followed up concurrently with cystoscopy, urine cytology, and biopsy as necessary every 3 months for 2 years, then every 6 months up to 5 years. Patients provided written informed consent for this institutional review board–approved study. End points were recurrence-free, progression-free and overall survival. Tumor recurrence included any evidence of disease after 3 months on follow-up evaluations. Tumor progression was defined as new or persistent lamina propria (pT1) invasion, muscle invasion, or metastases. Overall survival included deaths from any cause. Time to event was measured from the date of first MCC or other intravesical treatment. Patients who underwent cystectomy were censored for recurrence or progression, but they were included in survival analysis. Correlation between groups was performed using chi-square test. Survival times were calculated by Kaplan-Meier method and compared using the log-rank test. All tests were 2 sided and P o 0.05 were considered significant. 3. Results The Table shows patient characteristics of the 2 groups. All had high-volume recurrent non–muscle-invasive bladder
No. of 6-week BCG cycles: mean (range)
2
3.3 (2–5)
Diagnosis to BCG failure, mo: median (range)
6 (3–6)
18 (13–26)
11 (65%)
5 (33%)
Radical cystectomy
cancer and had failed BCG therapy. BCG-refractory patients had failed 2 induction courses of BCG within 6 months before starting MCC therapy. BCG-relapsing patients received an average of 3 or more 6-week courses of BCG and had a median time from diagnosis to BCG failure of 18 months (range: 13–26 mo). BCG-refractory patients received MCC therapy. BCG-relapsing patients received additional intravesical BCG, with or without interferon, or intravesical gemcitabine. Radical cystectomy was reserved for rapid tumor recurrence or progression to muscle invasion. Eleven BCG-refractory (65%) and 5 BCGrelapsing patients (33%) underwent salvage cystectomy. Pathologic stage in the BCG-refractory group was muscle invasion in 5 cases and 6 had recurrent pT1 tumors. In BCG-relapsing patients, 3 had carcinoma in situ only and 2 had pT1 disease. Of 11 BCG-refractory patients who had cystectomy, 7 were salvaged and 4 died of disease; of the 5 BCG-relapsing patients, 2 died and 3 survived after cystectomy. The mean follow-up time for all patients was 53 months (95% CI: 48–57 mo), 48 months (95% CI: 41–55 mo) for BCG-refractory patients, and 58 months (95% CI: 55–61 mo) for BCG-relapsing patients. All but 1 patient in both groups recurred. Fig. 1 shows median recurrence-free survival time was 10 months (95% CI: 6–17 mo) in BCG-refractory patients compared with 23 months (95% CI: 15–27 mo) in the BCG-relapsing patients (P ¼ 0.002). Tumors progressed in 14 of the 17 BCGrefractory patients (82%) compared with 5 of the 15 BCGrelapsing (33%) patients (P ¼ 0.04). Fig. 2 shows a median progression-free survival time of 18 months (95% CI: 7–28 mo) among BCG-refractory patients vs. a mean of 52 months (95% CI: 47–59 mo) in BCG-relapsing patients (P ¼ 0.001). Eleven patients have died (10 from bladder cancer and 1 from acute leukemia). Of the 17 BCG-refractory patients, 8 (47%) died vs. 3 of the 15
H.W. Herr et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
108.e3
Fig. 1. Recurrence-free survival. (Color version of figure is available online.)
Fig. 3. Overall survival. (Color version of figure is available online.)
BCG-relapsing patients (20%). Fig. 3 shows 54 months (95% CI: 24–72 mo) of median overall survival in BCGrefractory patients vs. median survival time not yet reached (mean ¼ 58 mo, 95% CI: 48–72 mo) among BCGrelapsing patients (P ¼ 0.04).
The major weakness of this study is the small numbers of patients. However, we believe that a few patients followed up prospectively according to prescribed protocol and recording real-time events as they unfold yields more reliable and robust information than retrospective analysis of databases confounded by selection bias, missing data, and unquantifiable variables. Segregating BCG failures into prognostic groups is not a new concept. Shirakawa et al. [6] defined 4 prognostic groups among BCG failures, with BCG-refractory patients having worse outcomes. The 10-year disease-specific survival rate in BCG-refractory and BCG-relapsing patients was 62% and 91%, respectively, similar to the 53% and 80% respective survival results from our study. Prognostic risk groups identified in retrospective analyses require prospective validation and that, albeit the small numbers of patients in our study, the information derived is persuasive. BCG refractory and BCG relapsing are clinically useful categories to define high-risk and lower-risk prognostic groups among patients who have failed BCG therapy. BCG refractory identifies patients who fail to respond to adequate BCG exposure, including maintenance BCG, and are never rendered disease free. BCG relapsing denotes later recurrences after a short-term (o2 y) disease-free interval. BCGrefractory patients are at risk for early tumor progression and cancer death and should be considered for immediate cystectomy. BCG-relapsing patients, although still at risk of dying of urothelial cancer, tend to have more indolent clinical courses favoring additional attempts at protocolbased bladder-sparing regimens. However, progression of disease and outcomes in both groups are poor, suggesting cystectomy should be discussed in all heavily pretreated BCG patients.
4. Discussion The major finding of this study is that BCG-refractory patients have worse outcomes in response to conservative salvage treatments than BCG-relapsing patients. Overall, 80% of BCG-relapsing patients survive compared with 53% of BCG-refractory patients with mature follow-up. Muscleinvasive tumor progression was more common in BCG-refractory than in BCG-relapsing patients, whose recurrences remained confined to bladder mucosa or submucosa. This is the first prospective study to discriminate different outcomes among high-risk and lower-risk cohorts of BCG failures.
References Fig. 2. Progression-free survival. (Color version of figure is available online.)
[1] Herr HW, Dalbagni G. Defining BCG refractory superficial bladder tumors. J Urol 2003;169:1706.
108.e4
H.W. Herr et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 108.e1–108.e4
[2] Saint F, Salomon L, Quintela R, et al. Do prognostic parameters of remission versus relapse after BCG immunotherapy exist? Analysis of a quarter century of literature. Eur Urol 2003;43:351. [3] Witjes JA. Management of BCG failures in superficial bladder cancer: a review. Eur Urol 2006;49:790. [4] Morales A, Herr H, Steinberg G, et al. Efficacy and safety of mycobacterial cell wall DNA complex in the treatment of patients with non-muscle invasive bladder cancer at high risk of progression
who are refractory to BCG. J Urol 2011;185:[No. 4 S (Abst 1650)]. [5] Sternberg IA, Dalbagni G, Chen LY, et al. Intravesical gemcitabine for high risk, nonmuscle invasive bladder cancer after Bacillus CalmetteGuérin treatment failure. J Urol 2013;190:1686–91. [6] Shirakawa H, Kikuchi E, Tanaka N, et al. Prognostic significance of BCG failure classification in non-muscle-invasive bladder cancer. BJU Int 2012;110:16–21.