Welcome, this website is intended for all international healthcare professionals in uro-oncology. By clicking the link below you are declaring and confirming that you are a healthcare professional.

You are here

Bacillus Calmette-Guérin Strain Differences Have an Impact on Clinical Outcome in Bladder Cancer Immunotherapy

European Urology, Volume 66 Issue 1, October 2014, Pages 677-688

Comment from Maria Ribal: In a current situation of shortage in BCG supply the information about differences on clinical outcomes of different BCG strains are results of great importance.



Whether the commonly used bacillus Calmette-Guérin (BCG) strains Connaught and Tice confer different treatment responses in non–muscle-invasive bladder cancer (NMIBC) is unknown.


To compare clinical efficacy, immunogenicity, and genetics of BCG Connaught and Tice.

Design, setting, and participants

A prospective randomized single-institution trial with treatment of 142 high-risk NMIBC patients with BCG Connaught or Tice.


Patients were randomized to receive six instillations of BCG Connaught or Tice. For experimental studies, BCG strains were compared in C57Bl/6 mice. Bladders and lymphoid tissues were analyzed by cytometry and the latter cultivated to detect live BCG. BCG genomic DNA was sequenced and compared with reference genomes.

Outcome measurements and statistical analysis

Recurrence-free survival was the primary end point of the clinical study. The Kaplan-Meier estimator was used for estimating survival and time-to-event end points. Nonparametric tests served for the analysis of the in vivo results.

Results and limitations

Treatment with BCG Connaught conferred significantly greater 5-yr recurrence-free survival compared with treatment with BCG Tice (p = 0.0108). Comparable numbers of patients experienced BCG therapy-related side effects in each treatment group (p = 0.09). In mice, BCG Connaught induced stronger T-helper cell 1–biased responses, greater priming of BCG-specific CD8+T cells, and more robust T-cell recruitment to the bladder than BCG Tice. Genome sequencing of the BCG strains revealed candidate genes potentially involved in the differential clinical responses.


BCG strain may have an impact on treatment outcome in NMIBC immunotherapy.

Patient summary

We compared the efficacy of two commonly used bacillus Calmette-Guérin (BCG) strains for the treatment of NMIBC and found that treatment with BCG Connaught prevented recurrences more efficiently than BCG Tice. Comparison of the immunogenicity of the two strains in mice indicated superior immunogenicity of BCG Connaught. We also identified genetic differences that may explain the differential efficacy of the Connaught and Tice BCG strains.

Trial registration

NCT00003779 .

Take Home Message

Genetically different bacillus Calmette-Guérin strains exert different clinical efficacy and immunogenicity. The urologic community is prompted to investigate this issue prospectively in randomized clinical trials.

Keywords: Urinary bladder neoplasms, BCG vaccine, BCG Connaught, BCG Tice, Intravesical drug administration, Immunotherapy, Randomized controlled trial, Models, Animal, Flow cytometry.

1. Introduction

Intravesical instillation of bacillus Calmette-Guérin (BCG) represents one of the most successful immunotherapies and has been the standard of care for patients with non–muscle-invasive bladder cancer (NMIBC) for >35 yr [1] . From its initial release in 1921 as an attenuated live vaccine for tuberculosis until the introduction of lyophilized seed lots in the 1960s, BCG was distributed worldwide and cultivated by continuous serial passage [2] . As a result, cultivated BCG strains accumulated mutations as compared with the original 1921 strain, for example, single nucleotide polymorphisms (SNPs) and genetic loss of regions of difference (RDs)[3] and [4]. More than eight different strains are currently used for the treatment of NMIBC [2] , and although they are considered biosimilar agents, it is debated among vaccination specialists and urologists whether BCG strain differences have an impact on efficacy and/or adverse effects secondary to treatment[4] and [5].

BCG Connaught and BCG Tice are the two most widely used BCG strains in North America and Europe [2] . When the use of BCG Pasteur was discontinued in Switzerland in 1994, the introduction of BCG Connaught and BCG Tice permitted us to examine possible differences between the two strains. We conducted a prospective randomized clinical trial comparing the two BCG strains, established experimental animal models, and performed genome-wide analysis of the two strains. We report (1) that patients treated with BCG Connaught had a significantly better recurrence-free survival compared with patients treated with BCG Tice and (2) that BCG Connaught is superior to BCG Tice in inducing T-helper 1–biased T-cell responses and priming CD8+T cells. These results suggest that a reevaluation of current clinical guidelines may be necessary to reflect the differing efficacy among BCG strains.

2. Patients and methods

2.1. Patients

The local ethics committee of the Bern canton approved the prospective randomized trial (reference 34/98). The multicenter trial was registered under ClinicalTrials.gov identifier NCT00003779 . The initially calculated sample size was defined to reach at least 300 patients based on a reduction of the relapse rate of 20%, a two-sided α of 0.05 and a power of 80%, and a correction for eventual censoring of patients. However, the multicenter trial was closed in 2003 by the Swiss Group for Clinical Cancer Research due to insufficient accrual and was instead conducted as a single-center trial at the Department of Urology, University of Bern, Switzerland. The trial was stopped in 2010 because of the clinically evident discrepancy in the efficacy between the two BCG strains, and recurrence-free, progression-free, and overall survival, along with side effects, were analyzed.

From 1998 to 2010, 142 patients with high-risk NMIBC consented and were randomly assigned to receive either BCG Connaught (ImmuCyst, Sanofi Pasteur, France) or BCG Tice (OncoTICE, MSD, NJ, USA) as the primary BCG therapy using sealed envelopes. From 1998 to 2003 and from 2004 to 2010, 58 and 84 patients, respectively, were recruited to the trial at the Department of Urology, University of Bern, Switzerland. High-risk NMIBC was defined as any high-grade tumor, any low-grade tumor with more than two recurrences within 2 yr, or carcinoma in situ. Because the study was initiated before studies indicated a potential benefit of maintenance therapy, no maintenance therapy was given. Only patients who had no prior intravesical BCG therapy were included.

Overall, 75 and 67 patients were allocated to receive BCG Connaught and BCG Tice, respectively. Four patients were lost to follow-up, six patients discontinued BCG treatment, and one patient was excluded from analysis because of protocol violation due to later BCG maintenance therapy. A total of 131 patients, 71 patients treated with BCG Connaught and 60 patients with BCG Tice, were analyzed ( Fig. 1 and Table 1 ). Pertinent patient and tumor characteristics were recorded. Surgical treatment consisted of a conventional white-light transurethral resection of bladder tumor (TURBT) of all visible tumors and random bladder biopsies in case of a preoperative positive bladder wash cytology. Patients with high-grade tumors underwent a second resection 2 wk after the initial TURBT to confirm no evidence of residual tumor or muscle invasion before treatment initiation.


Fig. 1 Consolidated Standards of Reporting Trials flow diagram of the prospective randomized clinical trial. BCG = bacillus Calmette-Guérin.

Table 1 Patient and tumor characteristics

  BCG Connaught

(n = 71)
BCG Tice

(n = 60)
p value
Sex, n (%)     0.8084
 Male 61 (86) 50 (83)  
 Female 10 (14) 10 (17)  
Age, yr, median (range) 72 (47–92) 72 (46–96) 0.819
Previous bladder tumor history, n (%)     0.1991
 No previous TURBT 49 (69) 43 (72)  
 1 previous TURBT 17 (24) 13 (21.5)  
 ≥2 previous TURBT 5 (7) 4 (6.5)  
Tumor status, n (%)     0.1178
 pTa 27 (38) 19 (32)  
 pT1 41 (58) 32 (53)  
 pTis 3 (4) 9 (15)  
CIS status, n (%)     1
 Any CIS positive 31 (44) 26 (43)  
No. of tumors, n (%)     0.8797
 1 38 (54) 32 (53)  
 2 9 (13) 11 (18)  
 ≥3 13 (18) 10 (17)  
 No visible tumor 1 (1) 1 (2)  
 Unknown 10 (14) 6 (10)  

BCG = bacillus Calmette-Guérin; CIS = carcinoma in situ; TURBT = trasnurethral resection of bladder tumor.

At 2–15 d after surgery, patients started instillation therapy consisting of six weekly intravesical instillations with 6.6–19.2 × 108colony-forming units (CFU) BCG Connaught (ImmuCyst) or 2–8 × 108CFU BCG Tice (OncoTICE) resuspended in 50 ml 0.9% saline that was retained in the bladder for 2 h. No BCG maintenance therapy was administered after the BCG induction courses every 6 wk, and no postoperative single-shot intravesical chemotherapy was performed because these concepts were introduced after the study was started. Patients were followed by cystoscopy and bladder wash cytology at 3-mo intervals for the first 3 yr, then at 6-mo intervals for the following 2 yr. An intravenous urography or computed tomography scan was performed 1 yr and 3 yr after BCG therapy.

A recurrence was defined as a return of tumor of any stage and grade confirmed by TURBT and histologic or cytologic assessment. A progression was defined as recurrence with increased stage and/or grade. Trial reporting was performed according to the Consolidated Standards of Reporting Trials 2010 standards [6] .

2.2. Bacillus Calmette-Guérin culture

Laboratory-grade BCG strains were cultured in Sauton medium and harvested after incubation for 8 d at 37 °C. Thereafter, the concentration of BCG cultures was adjusted to 50 mg/ml wet weight (approximately 10 million CFU/mg wet weight) in Beck Proskauer medium supplemented with 6% glycerol that preserved the viability of the bacilli at −20 °C. Regional medial iliac lymph nodes (LNs) were mashed with the plunger of a syringe in sterile phosphate-buffered saline (PBS) and plated on Middlebrook 7H11 solid medium supplemented with oleic acid, albumin dextrose, and catalase (OADC, Difco) [7] . CFUs were assessed after 17–28 d of growth at 37 °C. In addition, all preparations used for intravesical or subcutaneous injections were titrated on solid 7H11 agar to determine the number of live bacteria in each preparation.

2.3. DNA sequencing of bacillus Calmette-Guérin strains and protein prediction

Genomic DNA of the BCG strains was prepared starting from 200 ml of liquid culture [4] . Then 5 μg DNA were used to generate libraries and analyzed with an Illumina HiSeq 2000 sequence analyzer (Illumina Inc., San Diego, CA, USA). Single-end reads of 70 and 75 base pairs were obtained from cultured strains, grown directly from clinical-grade BCG Connaught and BCG Tice. Sequence coverage obtained corresponded to approximately 300×. Short-read data sets were aligned against the BCG Pasteur reference genome [4] (GenBank accession number NC_008769 ) by using MAQ software [8] . SNPs were called using SNIFER ( https://bitbucket.org/clafooty/tango/wiki/Home ), which is based on a comparison of aligned read sequences to the reference genome from mapping positions.

Mismatches identified were then filtered using five criteria: (1) a coverage sum >10, (2) a variant frequency of at least 0.89, (3) a threshold of quality >20 according to the Sanger format, and both (4) mean coverage and (5) quality >20 for the 10 bases surrounding the variant (−5/+5). A de novo assembly was performed on the read data sets by using the Perl script VelvetOptimiser, provided with the Velvet package [9] . Then 2000 and 1890 contigs were generated, with k-mers of 57 and 61; N50 lengths of 3467 and 3676 base pairs for BCG Connaught and BCG Tice, respectively.

For the search of indels, the contigs generated by Velvet were compared with the reference genome of BCG Pasteur followed by a filtering of gapped hits with a percentage identity >90%. Sequence read data were deposited in the EMBL Sequence Read Archive under accession number ERP001539. All identified polymorphisms in BCG Connaught and BCG Tice were also compared with available BCG sequences in public databases and are listed in Supplemental Table 1. Three-dimensional models of SodC were generated using the SWISS-MODEL server (swissmodel.expasy.org) [10] .

2.3.1. In vivo experiments

All animal studies were approved by the Pasteur Institute animal safety committee in accordance with French and European guidelines (Directive 86/609/CEE and Decree 87–848 of October 19, 1987). Female 7-wk-old C57BL/6 mice were purchased from Charles River (France). Six to eight mice per group were used, and each experiment was repeated two to eight times. Prior to intravesical instillations, animals were water-starved for 6 h. BCG was prepared by resuspending lyophilized clinical-grade BCG in 3 ml sterile PBS.

Mice were anesthetized with a solution of ketamine/xylazine; bladders were transurethrally catheterized using 24G catheters and instilled with 107CFU BCG Tice or Connaught in a volume of 50 μl as previously described [11] . Catheters were left in place for 2 h. Lymphocyte recruitment to the bladder was evaluated 1–2 wk after the final instillation. For clinical- and laboratory-grade BCG comparison, 100 μl of 2 × 106CFUs of the indicated preparations were injected via the subcutaneous route [11] . Spleens were harvested after 14 d, and BCG-specific immunity was assessed as described below.

2.3.2. Flow cytometric analysis

Bladders, draining LNs, and spleens were removed. Bladders were minced using a scalpel followed by digestion in a mixture of 1 U/ml Collagenase D (Roche), 0.17 U/ml Liberase TM (Roche), and 1 U/ml DNase I (Invitrogen) in prewarmed DMEM (Invitrogen). The cell suspension obtained was consecutively filtered through 100-μm and 40-μm cell strainers (Becton Dickinson). Spleens were mashed and incubated at 37 °C in a 1.66% ammonium chloride solution (VWR International) for 5 min to induce red blood cell lysis and filtered through a 70-μm cell strainer. Dead cells were stained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) or with a live/dead fixable Aqua Dead Cell Stain Kit (Invitrogen). Antibodies to CD16/CD32 (clone 2.4G2, FcR block), CD45.2 (clone 104), CD3ɛ (clone 145-2C11), NK1.1 (clone PK136), and CD8α (clone 53-6.7) were purchased from BD Pharmingen; CD4 (clone GK1.5), and T-bet (clone 4B10) antibodies were from eBioscience. For tetramer staining, soluble Db-Mtb32309-318(GAP) monomers [12] were produced using a modified version of that described previously [13] and conjugated to premium-grade streptavidin-PE (Invitrogen). All cells were preincubated with CD16/CD32 to block nonspecific Fc receptor binding, washed, incubated with tetramers when indicated, and subsequently incubated with indicated antibodies for 30 min in PBS and 1% fetal calf serum. Samples were acquired on a BD FACS Canto II cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star Inc., Ashland, OR, USA).

2.3.3. Statistical analysis

The Fisher exact test was used to compare patient and tumor characteristics. The Kaplan-Meier estimator was used for estimating survival function of time-to-event end points. We used a log-rank test to compare the survival functions; the method proposed by Klein et al. [14] was applied for the comparison of 5-yr survival rates. The follow-up time was calculated by using the inverse Kaplan-Meier method. The Mann-Whitney test was used for statistics on cellular distribution differences in the bladders, LNs, and spleens. PBS or untreated controls were not included in the analysis where indicated. We used R software v.3.0.1 (R Foundation for Statistical Computing, Vienna, Austria), GraphPad Prism v.5 (GraphPad Software Inc., La Jolla, CA, USA), and SPSS v.18.0 (IBM Corp., Armonk, NY, USA).

3. Results

3.1. Patients treated with Bacillus Calmette-Guérin (BCG) Connaught experience longer recurrence-free survival compared with patients treated with BCG Tice

Patient characteristics for age, gender, tumor stage, and tumor grade were well balanced between the two treatment arms ( Table 1 ). Median follow-up of patients treated with BCG Connaught (n = 71) and Tice (n = 60) was 47.6 mo (95% confidence interval [CI], 42.5–60.5 mo) and 51.4 mo (95% CI, 39.1–63.6 mo), respectively. Five-year recurrence-free survival of patients treated with BCG Connaught (74.0%; 95% CI, 62.8–87.2) was significantly greater than of those treated with BCG Tice (48.0%; 95% CI, 35.5–65.1;p = 0.0108; Fig. 2 a). Five-year progression-free survival was 94.1% (95% CI, 87.8–100) for patients treated with BCG Connaught and 87.9% (95% CI, 76.5–100) for patients treated with BCG Tice (p = 0.3442; Fig. 2 b). Five-year overall survival and disease-specific survival were comparable between patients treated with BCG Connaught and BCG Tice ( Table 2 ).


Fig. 2 Estimated probability of recurrence-free survival and progression-free survival by treatment. Kaplan-Meier survival curves of patients with high-risk non–muscle-invasive bladder cancer treated with either bacillus Calmette-Guérin (BCG) Connaught (orange lines) or BCG Tice (blue lines). (a) Estimated probability of recurrence-free survival. The calculated hazard ratio (log rank) for treatment with BCG Connaught is 0.4 (95% confidence interval, 0.25–0.71). (b) Estimated probability of progression-free survival.

Table 2 Survival probability at 5 years

  Survival probability at 5 yr (%) 95% CI p value
Recurrence-free survival     0.0108
 BCG Connaught 74.0 62.8–87.2  
 BCG Tice 48.0 35.5–65.1  
Progression-free survival     0.3442
 BCG Connaught 94.1 87.8–100  
 BCG Tice 87.9 76.5–100  
Overall survival     0.2652
 BCG Connaught 84.9 75.5–95.5  
 BCG Tice 93.6 85.2–100  
Disease-specific survival     NA
 BCG Connaught 93.0 86.5–100  
 BCG Tice 100 100–100  

BCG = bacillus Calmette-Guérin; CI = confidence interval; NA = not applicable.

Five-year recurrence-free survival was significantly greater in patients treated with BCG Connaught than in patients treated with BCG Tice. Progression-free survival, overall survival, and disease-specific survival were comparable between both treatment groups.

A comparable number of patients treated with BCG Connaught (n = 20 [28%]) and BCG Tice (n = 25 [42%]) experienced side effects caused by BCG instillation therapy (p = 0.09). A subanalysis of side effects demonstrated that significantly fewer patients treated with BCG Connaught (n = 9 [13%]) experienced dysuria compared with patients treated with BCG Tice (n = 18 [30%];p = 0.0151; Table 3 ). None of the patients experienced BCG sepsis. In a multivariate analysis assessing tumor recurrence rate including BCG strain, gender, dysuria, tumor stage (pTa, pT1, pTis), tumor grade (G1, G2, G3), and multifocality (one, two, or more than three tumors), BCG strain remained the only significant variable (hazard ratio: 2.91; 95% CI, 1.12–7.58;p = 0.0288; Supplemental Table 2).

Table 3 Side effects

  BCG Connaught

(n = 71)
BCG Tice

(n = 60)
p value
Side effects, n (%)
 No 48 (68) 30 (50) 0.09
 Yes 20 (28) 25 (42)  
 Unknown * 3 (4) 5 (8) 0.4685
Dysuria, n (%)
 No 59 (83) 37 (62) 0.0151
 Yes 9 (13) 18 (30)  
 Unknown * 3 (4) 5 (8) 0.4685
Frequency, n (%)
 No 64 (90) 47 (78.5) 0.1327
 Yes 4 (6) 8 (13.5)  
 Unknown * 3 (4) 5 (8) 0.4685
Nocturia, n (%)
 No 66 (93) 52 (87) 0.6556
 Yes 2 (3) 3 (5)  
 Unknown * 3 (4) 5 (8) 0.4685
Hematuria, n (%)
 No 63 (89) 51 (85) 1
 Yes 5 (7) 4 (7)  
 Unknown * , n (%) 3 (4) 5 (8) 0.4685
Temperature >37.5 °C, n (%)
 No 58 (82) 47 (78.5) 1
 Yes 10 (14) 8 (13.5)  
 Unknown * 3 (4) 5 (8) 0.4685

* The p value for “unknown” is calculated by considering that the unknown is a category of the variables.

BCG = bacillus Calmette-Guérin.

3.2. Bacillus Calmette-Guérin (BCG) Connaught induces greater priming of BCG-specific CD8+ T cells and enhanced recruitment of T cells to the bladder than BCG Tice

Concurrent with the clinical trial, we evaluated priming of BCG peptide-specific T cells, assessed using H2-DbMtb32309318tetramers [12] . Female C57BL/6 mice were intravesically instilled with PBS (control), clinical-grade BCG Connaught, or BCG Tice once a week for a total of 4 wk ( Fig. 3 a). Representative flow cytometry plots illustrate that following intravesical instillation with BCG Tice, about 0.1% of the CD8+T cell population was specific for H2-Db-Mtb32309318tetramers (range: <0.01–0.4%), whereas BCG-specific CD8+T cells in PBS-treated animals were undetectable (<0.01% of CD8+T cells were bound by the tetramer) ( Fig. 3 b and 3c). Intravesical instillation with BCG Connaught resulted in more robust CD8+T cell priming, with a median value of 0.45% tetramer-positive CD8+T cells (range: 0.2–0.8%;p < 0.002 compared with BCG Tice) ( Fig. 3 b and 3c). We found a significantly higher number of CFUs in the regional LNs of animals treated with BCG Connaught compared with animals treated with BCG Tice ( Fig. 3 d).


Fig. 3 Comparison of bacillus Calmette-Guérin (BCG) Connaught and BCG Tice in the induction of BCG-specific CD8+T cells and recruitment of CD8+T cells to the bladder. (a) The scheme illustrates the experimental design. Animals were instilled intravesically with either phosphate-buffered saline (PBS) or one of two BCG strains, BCG Connaught or BCG Tice, at indicated times (arrows). After four weekly instillations, spleens and lymph nodes were harvested and digested, and the resulting cell suspensions were analyzed by flow cytometry. (b) Representative flow cytometric plots of tetramer staining for H2-Db-Mtb32309–318positive CD8+T cells in spleens from mice intravesically treated with PBS, BCG Tice, and BCG Connaught (left to right). Plots represent (b) concatenated data from (c) individual mice. (c) The graph depicts the percentage of H2-Db-Mtb32309–318positive CD8+T cells among all CD8+T cells from the spleen. (d) The graph depicts colony-forming units (CFUs) recovered from regional lymph nodes in mice treated with intravesical BCG. Bars indicate medians. **p < 0.01; Mann-Whitney test. DLN = draining lymph node.

Next we evaluated the ability of the two strains to induce recruitment of immune cells into the bladder following repeated intravesical BCG instillation. CD45 expression marked leukocytes that were present in the bladder parenchyma, and staining with CD3ɛ and NK1.1 permitted the assessment of T cells (CD3ɛ+NK1.1), natural killer (NK) cells (CD3ɛNK1.1+), and natural killer T cells (CD3ɛ+and NK1.1+). T cells were further distinguished based on their expression of CD4 and CD8. Representative dot plots are shown for animals treated with BCG Connaught and BCG Tice ( Fig. 4 a). Absolute cell numbers of CD45+leukocytes, NK cells, NKT cells, CD3+T cells, CD4+T cells, and CD8+T cells were compared between mice treated with BCG Connaught and BCG Tice ( Fig. 4 b). BCG Connaught instillation induced a greater overall cellular recruitment to the bladder compared with BCG Tice treatment (p < 0.05) including significantly more CD4+and CD8+T cells.


Fig. 4 Recruitment of T cells to the bladder. Animals were instilled intravesically with either phosphate-buffered saline (PBS) or bacillus Calmette-Guérin (BCG) according to the scheme depicted in Figure 3 a. (a) Representative flow cytometry plots from bladders treated with BCG Tice and Connaught depicting gating strategies to delineate T, natural killer (NK), and natural killer T (NKT) cells. (b) Absolute numbers of indicated events per bladder of animals treated with phosphate-buffered saline (PBS), BCG Connaught, or BCG Tice. (c) Representative histogram of bladder-derived CD4+T cells expressing intracellular T-bet transcription factor. For each treatment group, three bladders were pooled for analysis. Yellow and green lines: respective isotype controls for animals treated with BCG Connaught and BCG Tice, respectively; blue line: Tice; orange line: Connaught. (d) Graphs show the frequency and absolute numbers of T-bet positive CD4+T cells in regional lymph nodes. In graphs, bars indicate medians. *p < 0.05; Mann-Whitney test. PBS-treated animals were not included in the statistical analysis.

To gain insight into the polarization of the T-cell response, we assessed expression of Th cell transcription factors. Consistent with enhanced Th1-cell priming, we observed higher T-bet expression in CD4+Th cells derived from pooled treated bladders ( Fig. 4 c). We also observed a higher percentage and absolute number of T-bet expressing CD4+T cells in the regional LNs in mice treated with BCG Connaught compared with mice treated with BCG Tice ( Fig. 4 d).

3.3. Bacterial titer and formulation do not contribute to strain differences

We considered the possibility that differences in in vivo T-cell priming ( Fig. 3 and 4) may be due to differential bacterial titer or formulation of the two freeze-dried clinical-grade preparations. We measured the number of live bacilli in each preparation for each experiment performed in the course of this study. No significant differences were observed between the two strains in the number of live bacilli present in each clinical preparation ( Fig. 5 a). To address the impact of formulation, we cultured the clinical BCG preparations and reformulated them as a concentrated suspension of live bacilli as described in the Methods section. Our cultivated preparations contained approximately fivefold higher CFUs than the clinical freeze-dried BCG preparations ( Fig. 5 b). The two laboratory-formulated preparations were compared for their ability to prime CD8+T cells as a measure of their in vivo potency. Although the laboratory formulations achieved better overall CD8+T cell priming, as determined by the percentage of H2-Db-Mtb32309318tetramer-positive CD8+T cells, there remained a significant difference between BCG Connaught and BCG Tice in their ability to induce priming of BCG-specific T cells ( Fig. 5 c and 5d).


Fig. 5 Evaluation of bacterial titers and bacillus Calmette-Guérin (BCG) formulation. (a) Titrations of clinical-grade (purchased) BCG freshly reconstituted from vials; (b) laboratory-grade BCG preparations (produced at Institut Pasteur, Paris). (c, d) Mice were immunized subcutaneously with a single dose of BCG. After 14 d, animals were killed and T-cell priming assessed. Graphs depict the frequency of H2-Db-Mtb32309–318tetramer positive CD8+T cells in spleens of animals injected with either (c) clinical or (d) laboratory-grade BCG Tice or BCG Connaught. *p < 0.05; Mann-Whitney test. Untreated animals were not included in the statistical analysis. CFU = colony-forming unit.

3.4. Candidate bacillus Calmette-Guérin single nucleotide polymorphisms may contribute to differential host response

Sequencing of the two BCG strains confirmed the previously reported RD15 (Rv0309–Rv0312) locus to be missing in BCG Connaught[3] and [4]. Comparison of aligned read sequences to the reference genome identified a total of 35 SNPs and 13 small insertion/deletion events (indels) differing among BCG Connaught, Tice, and Pasteur. Twelve SNPs were common to BCG Connaught and BCG Tice, of which seven were intergenic and six were synonymous nucleotide changes. We mapped the 22 genes affected by nonsynonymous SNPs against recently published sequences of multiple BCG reference strains andMycobacterium tuberculosisstrain H37Rv (Supplemental Table 1)[15], [16], [17], and [18], which helped to define SNPs and indels specific for BCG Connaught and BCG Tice. BCG Tice contains 9 nonsynonymous SNPs compared with BCG Pasteur; BCG Connaught contains 22 nonsynonymous SNPs. Interestingly, BCG Tice contained three strain-specific differences (two SNPs and one indel) not present in any other BCG strain orM. tuberculosis. Although one SNP and the indel extend or truncate, respectively, open reading frames of hypothetical proteins of unknown function (BCG_3474c,BCG_2151c; Supplemental Table 1), the second SNP is in the gene encoding the Cu, Zn superoxide dismutase C (sodC) and accounts for a nonsynonymous change from the highly conserved CAC (His) codon to a CAG (Gln) codon (Supplemental Fig. 1a–1d). Given the importance of SodC in defense against exogenous oxidative stress, and the evolutionary conservation of this His residue throughout a wide range of Cu, Zn Sod homologs (mycobacteria to human; Supplemental Fig. 1c and 1d), we evaluated the putative structural implications of this mutation. Using modeling software (SWISS-MODEL [10] ), we compared the predicted tertiary structure of the dimer formed by SodC from BCG Connaught and Tice (Supplemental Fig. 1e and 1f). The model showed close agreement with other bacterial Cu, Zn Sods whose structures have been resolved [19] . Our model predicted a lack of the Cu metal binding site in the active site of the BCG Tice enzyme, as a direct result of the amino acid change at position 118 (His → Gln) (due to the use of different numbering schemata, His118corresponds to His86in the reference) [19] . Loss of this binding site would likely result in a loss of enzyme activity.

4. Discussion

Here we provide evidence that BCG Connaught elicits a stronger immune response resulting in a significant better recurrence-free survival in patients with NMIBC. In a single-institution prospective randomized clinical trial comparing the intravesical use of BCG Connaught with BCG Tice in NMIBC patients with high risk for recurrence, a significantly better 5-yr recurrence-free survival for patients treated with BCG Connaught (74.0%) compared with patients treated with BCG Tice (48.0%;p = 0.0108) ( Fig. 2 a) was observed. These results, to the best of our knowledge, are the first to demonstrate that different BCG strains have a differential impact on clinical response to NMIBC immunotherapy. The low numbers of events for progression-free survival, disease-specific survival, and overall survival did not allow for conclusive statistical analyses. Although one would expect that recurrence-free survival might translate into better progression-free survival, disease-specific survival, and overall survival, such a conclusion would require a larger, prospective trial.

The study population received childhood vaccination for protection against tuberculosis. Because the ability of the body to mount a cellular immune response to BCG is a prerequisite for successful BCG therapy[11] and [20], our results may not be directly applied to populations that are not BCG vaccinated prior to intravesical therapy for NMIBC.

The following caveats limit the conclusions achieved of our study. First, treatment was performed in the absence of maintenance therapy because the benefit of maintenance therapy was only recognized after the start of the trial [21] . As such, patients treated with BCG Tice might have profited from maintenance regimens with improved outcomes and on par with patients treated with BCG Connaught. Second, we did not prospectively assess history of smoking, a known risk factor for high-risk disease [22] .

In contrast to BCG strain efficacy as a vaccine against tuberculosis, the impact of the genetic drift in BCG on treatment efficacy in patients with human NMIBC has not been previously demonstrated. Notably, there have been few head-to-head comparisons of different BCG strains used in the management of bladder cancer because bacterial preparations are considered to be biosimilar. Vegt and colleagues published the results of a randomized trial comparing the efficacy of BCG RIVM, BCG Tice, and mitomycin in 1995 [23] . Although the efficacy of BCG Tice was inferior to that of BCG RIVM and mitomycin in preventing recurrences (36% vs 54% recurrence-free survival after 5 yr for patients treated with BCG Tice and BCG RIVM, respectively), the study was underpowered to allow for the detection of statistically significant differences in outcome between the two BCG strains [24] .

The question of BCG strain differences is of ongoing interest to urologists and specialists in tuberculosis vaccination. However, the field still lacks convincing population-based studies to define a relationship conclusively between the differential efficacy of available BCG strains and protection against disease[4], [5], and [25]. Nonetheless, immunity and efficacy of different BCG strains have been studied using experimental models for tuberculosis. These reports indicate profound differences in the ability of different BCG strains to induce purified protein derivative reactions and to prevent tuberculosis in exposed animals[26] and [27]. Similar to the results reported here, BCG Connaught appears to be a stronger inducer of a Th1 polarized response compared with BCG Tice in a model of tuberculosis [28] . In the clinical trial, BCG Connaught proved to be more efficient in preventing recurrences than BCG Tice. Although a direct comparison between results in mice and humans are difficult to make, it is intriguing that the weaker immune stimulation induced by BCG Tice observed in mice correlated with the treatment outcome in humans.

A reported marker for genetic difference between BCG Connaught and BCG Tice is the loss of region RD15 in BCG Connaught (also referred to as RD08 [3] ). Our genomic analysis confirmed the absence of RD15 from commercial BCG Connaught preparations and also identified SNP and indel differences among BCG Connaught, Tice, and Pasteur (Supplemental Table 1). Among the 3 strains evaluated, 12 mutations were unique to BCG Pasteur, 18 found only in BCG Connaught, and 3 were specific to BCG Tice. Based on the similar clinical efficacy of BCG Pasteur and BCG Connaught [29] , we focused on the three mutations specific to BCG Tice:sodC,BCG_3474c, andBCG_2151c(Supplemental Table 1). Although the latter two encode hypothetical proteins whose functions are not known, considerable information exists regarding the role of SodC in permitting bacteria to resist oxidative stress. As such, reactive oxygen species (ROS) detoxification by bacterial SodC may enhance resistance of the organism to killing within the host environment[30] and [31]. The putative loss in SodC catalytic activity may result in BCG Tice being more sensitive to ROS and thus less fit for in vivo survival. We previously showed in an animal model that the presence of bacilli in LNs after intravesical BCG instillation is a correlate for the induction of BCG-specific T cells in vivo [11] . The near absence of bacilli in LNs of animals treated with BCG Tice supports the supposition of BCG Tice being more sensitive to ROS ( Fig. 3 d), and thus it might be an explanation for its less robust T-cell activation (Fig 3, Fig 4, and Fig 5). A recent prospective clinical trial conducted in Japan compared the BCG strains Tokyo and Connaught without evident differences in outcome or side effects [32] . Of note, both strains used in the Japanese trial express a functional SodC (Supplemental Table 1). Based on these observations, we conclude that BCG strain differences may account for the diminished ability of BCG Tice to stimulate immune responses against NMIBC, and present a set of possible target genes that may serve as the starting point for future experimental approaches to dissect the molecular mechanisms involved in generating adaptive tumor immunity.

5. Conclusions

In this prospective randomized clinical phase 3 trial of BCG Connaught versus BCG Tice, the Connaught strain was significantly more effective in terms of recurrence-free survival. In an experimental animal model, BCG Connaught induced a stronger Th1-biased T-cell response, more efficient priming of BCG-specific CD8+T cells in the spleen, and more robust T-cell recruitment to the bladder than BCG Tice. The genetic differences observed between the two BCG strains will permit further dissection of the key determinants of effective BCG immunotherapy. Based on the worldwide use of BCG Connaught and BCG Tice, we believe our findings should result in further trials assessing the efficacy of different BCG strains to improve the outcome of patients with NMIBC.

Author contributions:George N. Thalmann had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design:Thalmann, Albert, Rentsch, Brosch.

Acquisition of data:Thalmann, Birkhäuser, Rentsch, Biot, Gsponer, Bisiaux, Ingersoll, Wetterauer, Marchal, Lagranderie, Orgeur, Bouchier.

Analysis and interpretation of data:Rentsch, Birkhäuser, Orgeur, Bouchier, Brosch, Biot, Marchal, Lagranderie.

Drafting of the manuscript:Rentsch, Birkhäuser, Albert, Ingersoll, Brosch, Thalmann.

Critical revision of the manuscript for important intellectual content:Rentsch, Brosch, Ingersoll, Birkhäuser, Biot, Bachmann, Lagranderie, Marchal, Albert, Thalmann.

Statistical analysis:Rentsch, Gsponer, Ingersoll.

Obtaining funding:Albert, Thalmann, Rentsch, Brosch.

Administrative, technical, or material support:Lagranderie, Marchal, Orgeur, Bouchier, Brosch.

Supervision:Albert, Thalmann.

Other(specify): None.

Financial disclosures:George N. Thalmann certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor:The Cancéropôle Île de France, Ligue Foundation, and the Association for Research on Cancer financed the experimental animal work. Support for the clinical studies was provided by the Swiss National Science Foundation Grant number PSAMP 129381 and by the “Freiwillige Akademische Gesellschaft Basel.” Aurélie Bisiaux was funded by the French Foundation for Medical Research, and Molly A. Ingersoll by the LabEx Immuno-Oncology (Agence Nationale de la Recherche). Mickael Orgeur and Roland Brosch acknowledge support by the European Community's FP7 program under grant agreement number 241745, and FRM DEQ20130326471. Molly A. Ingersoll and Matthew L. Albert acknowledge support by La Ligue Contre le Cancer and Institut National du Cancer, France. The study was also supported by the Swiss Group for Clinical Cancer Research.

Acknowledgment statement:Cyrill A. Rentsch thanks Stewart T. Cole for his encouragement in initiating this project. Matthew L. Albert thanks the Center for Human Immunology for their support of the work. The clinical study was initiated with the help of the Swiss Group for Clinical Cancer Research (SAKK), and we thank Qiyu Li (SAKK Coordinating Center, Bern, Switzerland) and K. Stucki (Institute of Mathematical Statistics and Actuarial Science, University of Bern, Switzerland) for the support with the statistical analysis of the clinical data. We also thank Wafa Frigui, Romain Veyron-Churlet, and Laurence Ma for their help in genomic DNA and/or library preparation.

Appendix A. Supplementary data



  • [1] S. Brandau, H. Suttmann. Thirty years of BCG immunotherapy for non-muscle invasive bladder cancer: a success story with room for improvement. Biomed Pharmacother. 2007;61:299-305 Crossref
  • [2] H.W. Herr, A. Morales. History of bacillus Calmette-Guerin and bladder cancer: an immunotherapy success story. J Urol. 2008;179:53-56 Crossref
  • [3] M.A. Behr, M.A. Wilson, W.P. Gill, et al. Comparative genomics of BCG vaccines by whole-genome DNA microarray. Science. 1999;284:1520-1523 Crossref
  • [4] R. Brosch, S.V. Gordon, T. Garnier, et al. Genome plasticity of BCG and impact on vaccine efficacy. Proc Natl Acad Sci U S A. 2007;104:5596-5601 Crossref
  • [5] M.A. Behr. BCG—different strains, different vaccines?. Lancet Infect Dis. 2002;2:86-92 Crossref
  • [6] K.F. Schulz, D.G. Altman, D. Moher. CONSORT 2010 statement: updated guidelines for reporting parallel group randomized trials. PLoS Med. 2010;7:e10000251
  • [7] W. Van den Broeck, A. Derore, P. Simoens. Anatomy and nomenclature of murine lymph nodes: descriptive study and nomenclatory standardization in BALB/cAnNCrl mice. J Immunol Methods. 2006;312:12-19 Crossref
  • [8] H. Li, J. Ruan, R. Durbin. Mapping short DNA sequencing reads and calling variants using mapping quality scores. Genome Res. 2008;18:1851-1858
  • [9] D.R. Zerbino, E. Birney. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 2008;18:821-829 Crossref
  • [10] K. Arnold, L. Bordoli, J. Kopp, T. Schwede. The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics. 2006;22:195-201 Crossref
  • [11] C. Biot, C.A. Rentsch, J.R. Gsponer, et al. Preexisting BCG-specific T cells improve intravesical immunotherapy for bladder cancer. Sci Transl Med. 2012;4:137ra72 Crossref
  • [12] S.M. Irwin, A.A. Izzo, S.W. Dow, et al. Tracking antigen-specific CD8 T lymphocytes in the lungs of mice vaccinated with the Mtb72F polyprotein. Infect Immun. 2005;73:5809-5816 Crossref
  • [13] P. Bousso, A. Casrouge, J.D. Altman, et al. Individual variations in the murine T cell response to a specific peptide reflect variability in naive repertoires. Immunity. 1998;9:169-178 Crossref
  • [14] J.P. Klein, B. Logan, M. Harhoff, P.K. Andersen. Analyzing survival curves at a fixed point in time. Stat Med. 2007;26:4505-4519 Crossref
  • [15] Y. Pan, X. Yang, J. Duan, et al. Whole-genome sequences of four Mycobacterium bovis BCG vaccine strains. J Bacteriol. 2009;193:3152-3153
  • [16] L.H. Gomes, T.D. Otto, E.A. Vasconcellos, et al. Genome sequence of Mycobacterium bovis BCG Moreau, the Brazilian vaccine strain against tuberculosis. J Bacteriol. 2011;193:5600-5601 Crossref
  • [17] M. Seki, I. Honda, I. Fujita, et al. Whole genome sequence analysis of Mycobacterium bovis bacillus Calmette-Guerin (BCG) Tokyo 172: a comparative study of BCG vaccine substrains. Vaccine. 2009;27:1710-1716 Crossref
  • [18] P. Orduna, M.A. Cevallos, S.P. de Leon, et al. Genomic and proteomic analyses of Mycobacterium bovis BCG Mexico 1931 reveal a diverse immunogenic repertoire against tuberculosis infection. BMC Genomics. 2011;12:493 Crossref
  • [19] L. Spagnolo, I. Toro, M. D’Orazio, et al. Unique features of the sodC-encoded superoxide dismutase from Mycobacterium tuberculosis, a fully functional copper-containing enzyme lacking zinc in the active site. J Biol Chem. 2004;279:33447-33455 Crossref
  • [20] B. Zbar, H.J. Rapp. Immunotherapy of guinea pig cancer with BCG. Cancer. 1974;34(Suppl):1532-1540 Crossref
  • [21] D.L. Lamm, B.A. Blumenstein, J.D. Crissman, et al. Maintenance bacillus Calmette-Guerin immunotherapy for recurrent TA, T1 and carcinoma in situ transitional cell carcinoma of the bladder: a randomized Southwest Oncology Group Study. J Urol. 2000;163:1124-1129 Crossref
  • [22] M. Burger, J.W. Catto, G. Dalbagni, et al. Epidemiology and risk factors of urothelial bladder cancer. Eur Urol. 2013;63:234-241 Crossref
  • [23] P.D. Vegt, J.A. Witjes, W.P. Witjes, et al. A randomized study of intravesical mitomycin C, bacillus Calmette-Guerin Tice and bacillus Calmette-Guerin RIVM treatment in pTa-pT1 papillary carcinoma and carcinoma in situ of the bladder. J Urol. 1995;153:929-933
  • [24] W.P. Witjes, J.A. Witjes, G.O. Oosterhof, M.J. Debruyne. Update on the Dutch Cooperative Trial: mitomycin versus bacillus Calmette-Guerin-Tice versus bacillus Calmette-Guerin RIVM in the treatment of patients with pTA-pT1 papillary carcinoma and carcinoma in situ of the urinary bladder. Dutch South East Cooperative Urological Group. Semin Urol Oncol. 1996;14:10-16
  • [25] N. Ritz, B. Dutta, S. Donath, et al. The influence of bacille Calmette-Guerin vaccine strain on the immune response against tuberculosis: a randomized trial. Am J Respir Crit Care Med. 2011;185:213-222
  • [26] M.R. Lagranderie, A.M. Balazuc, E. Deriaud, C.D. Leclerc, M. Gheorghiu. Comparison of immune responses of mice immunized with five different Mycobacterium bovis BCG vaccine strains. Infect Immun. 1996;64:1-9
  • [27] S.M. Irwin, A. Goodyear, A. Keyser, et al. Immune response induced by three Mycobacterium bovis BCG substrains with diverse regions of deletion in a C57BL/6 mouse model. Clin Vaccine Immunol. 2008;15:750-756 Crossref
  • [28] A.I. Castillo-Rodal, M. Castanon-Arreola, R. Hernandez-Pando, et al. Mycobacterium bovis BCG substrains confer different levels of protection against Mycobacterium tuberculosis infection in a BALB/c model of progressive pulmonary tuberculosis. Infect Immun. 2006;74:1718-1724 Crossref
  • [29] R.J. Sylvester, M.A. van der, D.L. Lamm. Intravesical bacillus Calmette-Guerin reduces the risk of progression in patients with superficial bladder cancer: a meta-analysis of the published results of randomized clinical trials. J Urol. 2002;168:1964-1970
  • [30] D.L. Piddington, F.C. Fang, T. Laessig, et al. Cu,Zn superoxide dismutase of Mycobacterium tuberculosis contributes to survival in activated macrophages that are generating an oxidative burst. Infect Immun. 2001;69:4980-4987 Crossref
  • [31] O. Dussurget, G. Stewart, O. Neyrolles, et al. Role of Mycobacterium tuberculosis copper-zinc superoxide dismutase. Infect Immun. 2001;69:529-533 Crossref
  • [32] A. Sengiku, M. Ito, Y. Miyazaki, et al. A prospective comparative study of intravesical bacillus Calmette-Guerin therapy with the Tokyo or Connaught strain for nonmuscle invasive bladder cancer. J Urol. 2013;190:50-54 Crossref


a Department of Urology, University Hospital Basel, Basel, Switzerland

b Laboratory of Dendritic Cell Biology, Department of Immunology, Institut Pasteur, Paris, France

c Department of Urology, University of Bern, Inselspital, Bern, Switzerland

d Inserm U818, Paris, France

e Mines ParisTech, Paris, France

f Immunotherapix BioTop, Institut Pasteur, Paris, France

g Institut Pasteur, Unit for Integrated Mycobacterial Pathogenomics, Paris, France

h PF1-Plate-Forme Génomique, Institut Pasteur, Paris, France

i Université Paris Descartes, Paris, France

j Center for Human Immunology, Institut Pasteur, Paris, France

lowast Corresponding author. Department of Urology, Anna Seiler-Haus, University Hospital Inselspital, Bern, Switzerland. Tel. +41 31 632 36 64; Fax: +41 31 632 21 80.

These authors contributed equally.

Please visit www.eu-acme.org/europeanurology to read and answer questions on-line. The EU-ACME credits will then be attributed automatically.