Research ArticleBladder Cancer

Preexisting BCG-Specific T Cells Improve Intravesical Immunotherapy for Bladder Cancer

See allHide authors and affiliations

Science Translational Medicine  06 Jun 2012:
Vol. 4, Issue 137, pp. 137ra72
DOI: 10.1126/scitranslmed.3003586


Therapeutic intravesical instillation of bacillus Calmette-Guérin (BCG) is effective at triggering inflammation and eliciting successful tumor immunity in patients with non–muscle invasive bladder cancer, with 50 to 70% clinical response. Therapeutic success relies on repeated instillations of live BCG administered as adjuvant therapy shortly after tumor resection; however, the precise mechanisms remain unclear. Using an experimental model, we demonstrate that after a single instillation, BCG could disseminate to bladder draining lymph nodes and prime interferon-γ–producing T cells. Nonetheless, repeated instillations with live BCG were necessary for a robust T cell infiltration into the bladder. Parenteral exposure to BCG before instillation overcame this requirement; after the first intravesical instillation, BCG triggered a more robust acute inflammatory process and accelerated T cell entry into the bladder, as compared to the standard protocol. Moreover, parenteral exposure to BCG before intravesical treatment of an orthotopic tumor markedly improved response to therapy. Indeed, patients with sustained preexisting immunity to BCG showed a significant improvement in recurrence-free survival. Together, these data suggest that monitoring patients’ response to purified protein derivative, and, in their absence, boosting BCG responses by parenteral exposure before intravesical treatment initiation, may be a safe and effective means of improving intravesical BCG-induced clinical responses.


Bacillus Calmette-Guérin (BCG) therapy of superficial bladder cancer is one of the few examples of successful immunotherapy in the clinic and therefore offers a unique opportunity to define mechanisms by which the immune system may be used to target tumor cells. Carcinoma of the bladder is the most common cancer of the urinary tract and the fourth most common malignant disease in males in the developed world (1). Most tumors are diagnosed at a superficial stage and are surgically removed by transurethral resection. Depending on the stage and grade of the non–muscle invasive tumors, adjuvant therapy is recommended as a strategy for reducing recurrence and diminishing risk of progression (2). Since the work of Morales in 1976 (3), BCG therapy, which consists of six weekly intravesical instillations shortly after resection, has been the standard of care for high-risk urothelial carcinoma: carcinoma in situ and high-grade Ta/T1 bladder lesions (2, 4).

BCG is an attenuated strain of Mycobacterium bovis, initially developed as a vaccine against Mycobacterium tuberculosis (Mtb) infection (5). It is a slow-growing bacterium with a doubling time of ~24 hours. Macrophages are the principal host cells where mycobacteria either multiply or remain latent (6). Several effector lymphocyte subsets are induced upon mycobacterial infection, including multifunctional CD4+ and cytolytic CD8+ T cells (7). Interferon-γ (IFN-γ)–producing CD4+ T cells are responsible for phagocyte activation, an essential process for controlling mycobacterial infection (7). Although robust, it has been demonstrated that T cell recruitment to the lung mucosa of mice infected with aerosolized mycobacteria takes several weeks to establish (7, 8).

In the context of bladder cancer, BCG therapy is known to trigger a strong innate immune response, followed by the influx of type 1 polarized lymphocyte subsets (9, 10). Using orthotopically transplanted urothelial tumors in mice, several groups have reported that BCG-mediated antitumor activity relies on a functional immune system of the tumor-bearing host (1114). In particular, CD4+ and CD8+ T lymphocytes seem to be essential effector cells for eliminating the tumor in a mouse model (12), and correlates have been established between T cell infiltration and clinical response in patients (15).

We recently reported that repeated instillations with BCG were required to trigger a robust inflammatory response in patients (16). On the basis of these findings, our goal was to establish an experimental mouse model to study the dynamics of the immune response after intravesical BCG regimen and to specifically address several important unknowns, including the link between bacterial persistence/dissemination and T cell priming, the role of multiple instillations in triggering T cell influx in the bladder, and the temporal relationship between T cell priming and T cell entry into the inflamed bladder microenvironment. We demonstrate herein that T cell priming was necessary but not sufficient for achieving T cell trafficking to the bladder. Parenteral exposure to BCG before intravesical regimen accelerated the kinetics of bladder inflammation, resulting in an improved antitumor response after orthotopic tumor challenge. Moreover, patients with preexisting immunity to BCG had a greater likelihood of achieving recurrence-free survival. Together, these data provide new insight into a clinically effective immunotherapeutic regimen and predict strategies that may improve patient management.


Repeated intravesical instillations of live BCG result in a robust infiltration of activated αβ T cells into the bladder

To determine the dynamics of T cell infiltration into the bladder, we intravesically instilled age-matched female C57BL/6 mice with either phosphate-buffered saline (PBS; control) or clinical-grade BCG (ImmuCyst, Sanofi Pasteur) once a week for a total of three instillations (Fig. 1A, instillations indicated by black arrow). At defined time points, bladders were resected, digested as detailed in Materials and Methods, and stained for cytometric analysis. Infiltrating T cells were defined as CD45.2+CD3ε+NK1.1 cells (Fig. 1A). Twenty-nine days after the start of the treatment, there was a robust increase in both the percentage of T cells among total leukocytes infiltrating the bladder (Fig. 1B; instillation with BCG versus PBS, P < 0.01) and their absolute number (Fig. 1C; instillation with BCG versus PBS, P < 0.01). Once established, this infiltration was sustained in the absence of additional treatments for greater than 10 days (Fig. 1B). Administration of a fourth weekly instillation did not alter the kinetics of T cell influx into the bladder (fig. S1). Bladder T cells were predominantly found within the submucosa in the vicinity of blood vessels, with some having infiltrated the urothelium (Fig. 1D). All bladder T cells had an antigen-experienced phenotype based on the expression of CD44 and the absence of CD45RA (Fig. 1E). Although fewer in number, resident T cells were also CD45RACD44hi, suggesting that entry into the submucosa was restricted to previously activated T cells. Phenotypic assessment demonstrated that greater than 70% of the T cells were αβ CD4+ and CD8+ T cells (Fig. 1, F and G).

Fig. 1

Repeated instillations of BCG result in a robust infiltration of activated αβ T cells into the bladder. (A) Female mice received three weekly intravesical instillations of PBS (control) or clinical-grade BCG (ImmuCyst) at days 0, 7, and 14 (indicated by black arrows). At day 29, bladders were resected, digested with collagenase, and stained for flow cytometry. T cells were gated as live CD45.2+CD3ε+NK1.1 cells. Representative FACS plots are shown. (B) Mice were treated as above and the kinetics of T cell infiltration was evaluated. The dashed line indicates basal level T cells in naïve controls; seven independent experiments were combined with n = 3 to 7 mice analyzed per time point; means and SEM are shown. A general linear mixed model was used to compare the percentage of T cells among total leukocytes (***P < 0.0001). (C) Data from (B) was reanalyzed and absolute T cell numbers are shown for individual mice during the time window of maximal infiltration (days 29 to 35). Black bars indicate median. A Mann-Whitney test was performed (**P < 0.01). (D) Immunofluorescence staining at day 33 is shown. Nuclei were stained with DAPI (gray), leukocytes with CD45.2 (green), and T cells with CD3ε (red), whereas α-SMA (blue) staining indicates the smooth muscle layer of blood vessels (V) and the bladder muscle layer (M). A dotted white line demarcates the bladder lumen (L); the thin urothelial layer (U) and the submucosa (S) are indicated. Scale bar = 100 μm. (E) Bladder infiltrating T cells were assessed by flow cytometry for an activated phenotype on the basis of CD44 expression and absence of CD45RA. A representative histogram is shown (green, naïve; blue, PBS; red, BCG). Shaded histograms indicate fluorescence minus one. (F and G) T cells infiltrating the bladder were further gated as γδ TCR positive or negative, and the latter population was assessed for CD4 or CD8α expression. Representative FACS plots and gating strategy are shown (F). Average numbers of T cell subpopulations are displayed; n = 4 mice per group (G).

Priming of T cells is necessary but not sufficient for their entry into the bladder after intravesical BCG instillation

Early clinical investigation in humans suggested that live BCG was required to achieve tumor immunity (17, 18). To establish that T cell recruitment was indeed dependent on live bacilli, we compared repeated intravesical instillations of clinical-grade BCG (containing live BCG) with heat-killed BCG. As expected, the latter did not result in T cell recruitment to the bladder (fig. S2).

To assess the requirement for live BCG to activate adaptive immunity components, we evaluated BCG dissemination with the hypothesis that its entry into the bladder draining lymph nodes (LNs) is a prerequisite for priming and subsequent T cell infiltration of the bladder. Such a requirement has indeed been well documented in the context of low-dose Mtb lung infection (1921). We first assessed the persistence of BCG after intravesical injection, assaying colony-forming units (CFUs) that remain in the bladder after first voiding (at removal of catheter) at 2 hours and on day 1 after instillation. Consistent with what has been suggested for human treatments (22, 23), the BCG load in the bladder rapidly decreased to 1% of the instilled dose and was barely detectable by 24 hours after instillation (Fig. 2A). Next, we investigated the presence of live BCG in the periaortic draining LNs. As an additional parameter, we compared single and weekly repeated instillations, because the rapid decrease of BCG load in the bladder led us to hypothesize that repeated doses of live BCG might be required to result in efficient BCG dissemination to the draining LNs. Mice were intravesically instilled with live BCG and the periaortic LNs were homogenized and plated at defined time points. Analysis of early time points (hours) after a single instillation demonstrated no bacterial growth (fig. S3), thus indicating that bacilli were not tracking to the LNs owing to passive processes (such as from potential trauma and/or antegrade pressure during the instillations). Mice were tested for up to 1 month after single or repeated BCG instillations and when bacterial growth was observed: The total CFUs per LN ranged from 8 to 1700 colonies (median CFU = 50). Because of the high variance, we scored animals as positive or negative for the presence of live BCG in the LNs. Overall, we observed that 40 to 60% mice harbored BCG in their periaortic LNs after a single intravesical instillation, which was consistent across a time course of 15 to 36 days (Fig. 2B). In comparison, mice receiving multiple instillations also showed a mixed response at 15 days (Fig. 2B, P = not significant), but by days 30 to 36, BCG could be cultivated from the periaortic LNs of all mice (Fig. 2B, P < 0.01).

Fig. 2

Priming of T cells is necessary but not sufficient for their entry into the bladder. (A) At 2 and 27 hours after instillation, bladders were homogenized in PBS and total CFUs per organ were enumerated. Limit of detection was 20 CFUs and is demarcated by a dotted line. (B) Bladder draining LNs were resected at indicated time points, after either a single or repeated instillations of live BCG, homogenized in PBS, and plated. Mice were stratified as either CFU positive (black; limit of detection = 4 CFUs) or CFU negative (white); several independent experiments were combined (n = 13 to 24 mice per group). Two-sided Fisher’s exact tests were performed (ns, not significant; **P < 0.01). (C and D) Mice were treated and stratified as above and the BCG-specific response was analyzed on splenocytes using H2-Db–Mtb32309–318 tetramers on days 30 to 36. CD8+ T cells were gated as live, dump negative (dump channel including CD45RB (B220), NK1.1, CD11b, F4/80, and CD4), CD3ε+CD8α+, and the percentage of tetramer-positive (Tet+) cells among this population was analyzed. A representative FACS plot for tetramer assays is shown for an animal receiving PBS or weekly intravesical instillations of BCG (C). The percentage of Tet+ cells among CD8+ T splenocytes is shown for individual mice across the different treatment conditions; black bars represent medians. Mann-Whitney tests were performed (ns, not significant; *P < 0.05) (D). (E) Mice were treated as above, and at day 29, purified CD8+ T cells from spleen and draining LNs from mice that were CFU+ were restimulated ex vivo for 20 hours using splenocytes pulsed with Mtb32309–318 peptide. Unpulsed splenocytes served as a negative control. The number of SFCs per 106 CD8+ T cells for individual mice is shown. (F) Mice were treated and stratified as above and absolute numbers of T cells infiltrating the bladder were enumerated after either a single or repeated instillations on days 30 to 36. Individual mice are shown; black bars represent medians. Mann-Whitney tests were performed (**P < 0.01).

We next evaluated the priming of BCG peptide–specific T cells, assessed using H2-Db–Mtb32309–318 tetramers (also known as PepA or GAP) (24) (Fig. 2C). When mice were stratified on the basis of the presence of live BCG in their periaortic LNs, most of the CFU+ animals had a high frequency of BCG-specific CD8+ T cells among total splenocytes. In contrast, there was no expansion of Db-Mtb32309–318–reactive T cells in mice for which live BCG was undetectable (Fig. 2D, P < 0.05). When comparing mice that had received single or repeated instillations, the critical parameter was the presence of live BCG (Fig. 2D, CFU+ single versus repeated, P = not significant). We next assessed the capacity of CD8+ T cells purified from spleen and periaortic LNs to produce IFN-γ upon restimulation with Mtb32309–318 peptide in an enzyme-linked immunosorbent spot (ELISPOT) assay. In mice harboring live BCG within their LNs, we found similar numbers of spot forming cells (SFCs) irrespective of the number of instillations (Fig. 2E). These data demonstrate that the priming of IFN-γ–producing BCG-specific T cells can occur after a single instillation and correlates with BCG dissemination to the bladder draining LNs.

To investigate whether dissemination of BCG also correlated with local adaptive immunity, we examined lymphocyte populations in the bladder. Although we observed low levels of T cell infiltration in CFU+ animals, the level of infiltration was significantly lower in mice that had received single versus repeated instillations (Fig. 2F, P < 0.01). Together, these data suggest that priming of T cells is necessary but not sufficient for achieving T cell accumulation in the bladder.

Preexisting adaptive immunity supports a robust intravesical immune response

To further test the dissociation of priming from T cell trafficking, we evaluated whether the activation of BCG-specific T cells before bladder instillations would affect T cell recruitment during the first intravesical instillation. Mice were injected subcutaneously with BCG, and after 21 days, intravesical instillations were initiated, comparing single and repeated BCG challenge. In mice primed by subcutaneous BCG, we observed a robust T cell infiltration as early as 12 days after a single instillation (Fig. 3; subcutaneous BCG W4), which lasted up to 35 days after instillation (Fig. 3; subcutaneous BCG W1). The level of T cell accumulation in the bladder was similar to that achieved by multiple intravesical treatments (Fig. 3; BCG W1 to W4). Repeated instillations in the subcutaneously primed group (Fig. 3; subcutaneous BCG W1 to W4) did not result in an enhanced accumulation of T cells compared to other treatment conditions, suggesting that maximal intravesical responses can be achieved by a subcutaneous injection of BCG followed by a single instillation of BCG. Notably, live BCG could be detected in the draining LNs of mice 20 days after subcutaneous immunization but could not be detected in the bladder (fig. S4).

Fig. 3

Subcutaneous immunization with BCG before intravesical instillations results in accelerated T cell entry into the bladder. Twenty-one days before intravesical instillation, mice were subcutaneously immunized with BCG, compared to nonimmunized (Ø) controls (subcutaneous injection is represented by a star). Mice subsequently received either a single or repeated intravesical instillations with PBS or BCG (instillations represented by a black arrow). Bladder T cell infiltration was analyzed by flow cytometry on days 33 to 35. A Kruskal-Wallis test was performed among all groups that received intravesical BCG (ns, not significant).

To further characterize the differential bladder T cell trafficking after different BCG regimens, we evaluated the local inflammation of the bladder mucosa. Shortly after the first and the third instillation, we observed a rapid but short-lived (less than 42 hours after instillation) influx of neutrophils (characterized as Ly-6G+ leukocytes; Fig. 4, A and B) and inflammatory monocytes (characterized as Ly-6ChighCD11b+Ly-6G leukocytes; Fig. 4, A and B). Notably, accumulation of inflammatory monocytes was significantly more pronounced after the third instillation (Fig. 4B, P < 0.01). In animals that had received previous subcutaneous BCG, the infiltration of neutrophils and inflammatory monocytes after a single dose of intravesical BCG was more pronounced than in nonvaccinated animals (Fig. 4C; isotype control). The inflammatory response was even stronger than that observed after repeated instillations with no previous subcutaneous exposure to BCG (Fig. 4, B and C).

Fig. 4

Preexisting adaptive immunity supports a robust, albeit short-lived, innate immune response. (A) Neutrophils were defined as live CD45.2+Ly-6G+ cells; inflammatory monocytes were defined as live CD45.2+Ly-6GLy-6ChighCD11b+ cells. For each cell population, a representative FACS plot is shown (16 hours after third BCG instillation). (B) Sixteen and 42 hours after either the first or the third BCG instillation, bladder-infiltrating neutrophils (upper graph) and inflammatory monocytes (lower graph) were quantified by flow cytometry (n = 3 to 9 mice per group). Mean values and SEM are shown. A Mann-Whitney analysis was performed to compare infiltration at 16 hours after the first and the third instillation with BCG (ns, not significant; **P < 0.01). (C) Mice were subcutaneously immunized with BCG 21 days before instillation, compared to nonimmunized controls, followed by a single intravesical instillation with PBS or BCG. Forty-eight hours before instillation, mice were treated with depleting monoclonal antibodies specific for CD4+ and CD8+ T cells or isotype control antibodies. Sixteen hours after intravesical instillation, infiltration of neutrophils (upper graph) and inflammatory monocytes (lower graph) was assessed by flow cytometry. Individual mice are shown and medians are indicated by black bars. Mann-Whitney analyses were performed (ns, not significant; *P < 0.05).

Considering the robust inflammatory process observed in mice immunized subcutaneously with BCG, we hypothesized that the existence of BCG-specific T cells at the time of instillation has an impact on the acute inflammatory process. To test this possibility, we subjected mice previously immunized subcutaneously with BCG to anti-CD4 and anti-CD8 depleting antibodies 48 hours before intravesical instillation. After T cell depletion, we demonstrated a decrease in the number of neutrophils and inflammatory monocytes infiltrating the bladder (Fig. 4C, P < 0.05). The level of the inflammatory response in the group of mice that underwent transient depletion was in the range of what is observed after the first instillation with no previous subcutaneous BCG exposure (Fig. 4C). Similar results were obtained using an alternative strategy for T cell depletion (fig. S5). Together, these data suggest that T cells, which are primed by subcutaneous BCG, mediate the “boosted” inflammatory response after intravesical BCG.

Preexisting BCG-specific immunity improves antitumor response

On the basis of the ability to achieve stronger inflammation and earlier T cell recruitment to the bladder microenvironment, we reasoned that subcutaneous exposure to BCG before intravesical BCG therapy might improve the antitumor response. To test that hypothesis, we used an orthotopic tumor model—implantation of syngeneic MB49 tumor cells into the bladders of C57/BL6 mice. Although the derived epithelial MB49 tumors grow in an aggressive manner (25), this model remains, to our knowledge, the only mouse model in which intravesical BCG treatment has been shown to induce antitumor responses when initiated 1 to 2 days after tumor implantation (26). To evaluate the impact of preexisting BCG-specific T cells, we subcutaneously immunized mice with BCG, and after 3 weeks, we implanted 80,000 MB49 cells into the bladder mucosa, as described in Materials and Methods. Two days later, intravesical BCG therapy was initiated, and mice were monitored twice daily for survival. One hundred percent of mice that received BCG subcutaneously before intravesical therapy survived as late as 70 days after tumor challenge; in comparison, 80% of mice with no previous BCG immunization succumbed within 50 days, despite intravesical BCG therapy (Fig. 5). As a control, mice that received BCG subcutaneously were challenged with tumors, and received intravesical PBS, showing no evidence of delayed tumor growth (Fig. 5).

Fig. 5

Preexisting BCG-specific immunity improves antitumor response in a mouse model for bladder cancer. Three weeks before orthotopic MB49 tumor challenge, mice were subcutaneously immunized with BCG (solid lines) or left untreated (dashed lines). Starting 2 days after tumor implantation, mice received five weekly intravesical instillations of either PBS (blue lines) or BCG (red lines) and were monitored twice daily for survival until termination of the experiment on day 70. A log-rank test was performed to compare groups that received intravesical BCG, either immunized subcutaneously or not (**P < 0.01).

These results prompted us to investigate the relevance of preexisting BCG-specific immunity in patients with high-risk non–muscle invasive bladder cancer undergoing BCG therapy. Analysis of available clinical data was performed, accessing data from an observational study in which patients underwent a purified protein derivative (PPD) skin test before intravesical therapy (Table 1). A positive skin test is the signature of previous exposure and active immune response to BCG, Mtb, or other mycobacteria. We therefore stratified patient outcome data according to their PPD status before treatment and observed that patients with a positive PPD had a significantly better recurrence-free survival than patients with a negative PPD skin test (Fig. 6, P < 0.02). Together, these data suggest that boosting BCG-specific immunity before intravesical therapy might improve clinical response and tumor immunity.

Table 1

Patient and tumor characteristics.

View this table:
Fig. 6

Preexisting BCG-specific immunity improves the antitumor response in patients with high-risk non–muscle invasive bladder cancer undergoing intravesical BCG therapy. Patients were stratified according to their pretherapy PPD status (+, positive; −, negative), and analysis of their recurrence-free survival was performed over 60 months. The median recurrence-free survival was 25 months in the PPD-negative group and was not reached in the PPD-positive group. A log-rank test was performed (*P < 0.05); hash marks along the lines indicate censored events (for example, death from causes other than bladder cancer).


BCG was generated by the repetitive passage of a virulent strain of M. bovis. This live attenuated strain was developed as a vaccine for tuberculosis, and in line with the work of William Coley, BCG was evaluated for use as an anticancer therapeutic vaccine. Over the past century, BCG has been injected into many solid tumors, and although there were reports of some success, controlled clinical trials did not provide statistical significance (10, 27, 28). Nonetheless, animal studies with BCG continued and it was recognized that long-lasting direct contact with the live bacteria resulted in optimal tumor immunity (29). These results prompted Morales and colleagues to evaluate BCG as an adjuvant intravesical treatment for carcinoma of the bladder (3, 30). Since these initial observations, several large prospective studies have been conducted and BCG remains the standard of care for non–muscle invasive disease. Although now in use for over 35 years, many questions remain about the mechanism of action by which BCG mediates the observed clinical response (10); additionally, there is interest in identifying strategies for optimizing therapy (31, 32).

Although previous efforts have evaluated immunologic response during therapy in human observational studies or in experimental mouse models, our study provides the first systematic evaluation of BCG-induced T cell infiltration of the bladder mucosa. Using histological and cytometric analyses, and paying careful attention to mycobacterial persistence and antigen-specific T cell priming, we defined the parameters required for achieving effective adaptive immune responses in the bladder. We identify a requirement for live bacteria that disseminate to local draining LNs to achieve T cell priming, and repeated instillations are needed to trigger recruitment of T cells to the bladder microenvironment. Careful analysis of these parameters has not been previously documented, in part due to the inability to access patient material (such as bladder mucosa and LNs) during a clinically approved therapeutic intervention.

On the basis of experimental work in humans, we focused our attention on the activation and recruitment of T lymphocytes. Immunohistochemical analysis of patient bladder biopsies has indeed demonstrated T cell infiltration during BCG therapy and up to 3 months after therapy (15, 33), and the degree of T cell infiltration correlated with treatment response (15). These findings were consistent with our study of BCG-induced inflammation after intravesical instillations in mice. On the basis of our observations of a delayed influx of T cells, we hypothesized that parenteral exposure to BCG before standard of care might accelerate the kinetics of bladder inflammation. We demonstrate that such an approach provides an optimized strategy for T cell recruitment and that this treatment protocol improves the host antitumor response.

From the perspective of the host response to infection, one marked observation is that despite the presence of proinflammatory mycobacterial components [including Toll-like receptor agonists and inflammatory lipids (34, 35)] and the induction of cell death (36), the bladder T cell infiltration after repeated BCG instillations occurs only after day 29 (Fig. 1C). There is an important precedent for such an observation—the cellular immune response to aerosolized Mtb takes several weeks to initiate, with T cells reaching the lungs only 20 days after infection (7, 8). Several hypotheses have been proposed to explain such a delayed cellular immune response. One possibility is that Mycobacteria sp. regulate T cell activation by inhibiting antigen-presenting cell migration and/or function (7). Alternatively, it has been speculated that slow induction could be a consequence of the low number of bacteria used for infection, which could result in insufficient inflammation and antigen load. Arguing against this point, however, is the observation that exponentially increasing the dose of aerosolized bacteria in lung studies does not significantly accelerate T cell recruitment to the mucosa (20). In our bladder instillation model, the number of BCG CFUs decreases quickly; however, testing the response to a higher dose of BCG remains technically challenging because of the size of the freeze-dried BCG cake as well as the dead volume of the catheter used for instillations. Once established, the response is sustained, lasting at least 21 days after the third instillation.

We also discovered that BCG dissemination to the regional LNs is critical for achieving efficient T cell priming, again showing similarity to what has been shown in the lung Mtb infection model (1921). T cell priming, however, was not sufficient to achieve T cell recruitment to the bladder, as shown by the relatively low level of T cell infiltration after a single instillation, even in the presence of measurable BCG-specific T cell responses. To further assess the relationship between priming and trafficking to the bladder, we performed studies in mice that were previously primed through the subcutaneous route. These data demonstrated that trafficking of T cells to the bladder could be dissociated from the route of priming. Bladder T cell recruitment correlated with a robust but short-lived innate immune response, which is suggestive of a delayed-type hypersensitivity (DTH) response (also known as type 4 hypersensitivity reaction). Although not typically attributed to inflammation in the bladder mucosa, DTH reactions are mediated by antigen-specific effector T cells (for example, induration induced by PPD challenge in the skin of a primed individual). This reaction is defined by antigen-specific T cells mediating the rapid recruitment of inflammatory cells to the site of injection (37, 38). We report here that subcutaneous immunization 21 days before BCG intravesical instillation results in a more robust inflammatory response after intravesical BCG, which is dependent on T cells, thereby suggesting that bladder inflammation in response to BCG should be considered a DTH reaction. Although we demonstrate a critical role for primed T cells in the BCG-mediated influx of inflammatory innate cells, we were unable to define the cellular mechanisms governing T cell entry into the bladder. Notably, depletion of neutrophils, monocytes, and natural killer cells did not result in impaired T cell trafficking to the bladder (fig. S6).

To apply our insights into the dynamics of bladder inflammation, we tested our modified treatment regimen using an orthotopic bladder tumor model. We demonstrate the ability to achieve up to 100% survival compared to 80% lethality at day 70 (median survival time being ~45 days). These data are remarkable considering the aggressive nature of MB49, but even more so for the ease of translating our results for testing in human clinical trials. Supporting the important role for pretreatment BCG-specific responses, we analyzed the available clinical data from a recent observational trial and identified the absence of a PPD response to be a risk factor for treatment failure. Although these data emerged from the assessment of parameters dictating T cell recruitment, we acknowledge the historical interest in PPD as a clinical marker for response to therapy. More than 25 years ago, Badalament and colleagues reported a related finding in a cohort of U.S. patients with recurrent superficial bladder cancer: Pretherapy PPD-positive patients—likely primed as a result of their previous intravesical treatment—showed greater benefit from their subsequent course of treatment (39). In line with this report, in patients who fail intravesical therapy, an additional 50% of patients respond to the second cycle of BCG (40, 41). In light of our data, we suggest that the first cycle of BCG might serve to prime patients, thus enhancing bladder inflammation and the chance to achieve tumor clearance during subsequent rounds of intravesical treatment. Other studies have focused on PPD conversion achieved by intravesical therapy, demonstrating correlation with clinical response (4244). These results, however, have been somewhat controversial, because others only observed a trend (45) or did not observe any correlation at all (46, 47).

Clinical studies have also investigated the combined use of intravesical and intradermal (or scarification) routes for treating patients with BCG (44, 45). These trials showed no evidence of enhanced clinical response. Arguably, our findings indicate that treatment protocols combining both routes of injection were well conceived but misguided with respect to the timing required for achieving systemic BCG-specific immunity. We favor a treatment strategy where patients are exposed parenterally to BCG before initiation of intravesical therapy. To our knowledge, this has not yet been evaluated and may represent a straightforward approach to improving treatment response.

In summary, we have demonstrated that although BCG dissemination to regional LNs and priming of IFN-γ–producing T cells can occur after a single instillation, repeated instillations of live BCG are necessary to achieve robust bladder T cell infiltration. Parenteral exposure to BCG before instillation overcomes the requirement for repeated instillations, triggering a more robust acute inflammatory process at the first instillation and accelerating the recruitment of T cells to the bladder. Moreover, parenteral exposure to BCG before orthotopic tumor challenge markedly improves response to BCG therapy. Patients with preexisting immunity to BCG respond significantly better to therapy. Together, these data suggest that checking patients’ immunity to BCG before intravesical therapy, and boosting it if necessary, might improve BCG-induced clinical responses.

Materials and Methods

Mouse intravesical instillations and subcutaneous immunization

For intravesical instillations, 7- to 12-week-old C57BL/6 female mice (Charles River) were water-starved for 7 to 8 hours, reflecting the clinical practice of patients being asked not to drink before treatment. Mice were anesthetized [ketamine (125 mg/kg) and xylazine (12.5 mg/kg) intraperitoneally) and drained of any urine present by application of slight digital pressure to the lower abdomen. The urethral orifice was disinfected with povidone-iodine, and a 24-gauge catheter (BD Insyte Autoguard, Becton Dickinson) adapted to a 1-ml tuberculin syringe (Braun) containing 50 μl of either PBS (Invitrogen) or BCG (~3 × 106 CFUs) was carefully inserted through the urethra. The injection was made at a low rate to avoid trauma and vesico-ureteral reflux, and there was no dead volume in the catheter. Mice were kept under anesthesia for 2 hours, with catheter and syringe maintained in place to retain the intravesical solution. For tumor implantation, mouse bladders were pretreated with poly-l-lysine (0.1 mg/ml) (Sigma-Aldrich) for 20 min, before instillation of 80,000 MB49 cells in 50 μl of PBS, which were retained for 1 hour into the bladder. For subcutaneous immunization, mice received a single injection of 2 × 106 to 5 × 106 CFUs of BCG. Mice were housed under specific pathogen-free conditions and used under approved protocols.

BCG and determination of bacterial load

For instillations, ImmuCyst (Sanofi Pasteur) was reconstituted in 3 ml of PBS following the manufacturer’s instructions. For subcutaneous administration, either ImmuCyst (once) or frozen aliquots of BCG Pasteur (1137P2) were used with similar results. BCG Pasteur was grown at 37°C in Middlebrook 7H9 medium supplemented with bovine albumin, dextrose, and catalase (ADC, Difco), harvested in exponential growth phase, washed, dispersed with 3-mm glass beads, resuspended in PBS, aliquoted, and then frozen at −80°C. A defrosted aliquot was used to determine the lot titer on 7H11 medium supplemented with oleic acid and ADC (OADC, Difco). In addition, all preparations used for intravesical or subcutaneous injections were titrated. For organ bacterial load, bladders were resected in sterile PBS and homogenized for 2 min at 25 Hz in a TissueLyser II (Qiagen), whereas draining LNs were mashed with the back of a syringe in sterile PBS. Fivefold serial dilutions of the homogenates were plated on 7H11 supplemented with OADC, and CFUs were assessed after 17 to 28 days of growth at 37°C.

Antibodies and reagents

For fluorescence-activated cell sorting (FACS), CD16/CD32 (clone 2.4G2, Fc block), CD45.2 (clone 104), CD3ε (clone 145-2C11), NK1.1 (clone PK136), CD8α (clone 53-6.7), CD44 (clone IM7), CD45RA (clone 14.8), CD45R/B220 (clone RA3-6B2), CD11c (clone HL3), CD86 (clone GL1), Ly-6C (clone AL-21), and Ly-6G (clone 1A8) antibodies (Abs) were purchased from BD Pharmingen; CD4 (clone GK1.5), CD11b (clone MAC-1), pan-γδ TCR (T cell receptor) (clone GL3), IAb-IEb (clone M5), and F4/80 (clone BM8) Abs were from eBioscience; and CD45.2 (clone 104-2) was from SouthernBiotech. Dead cells were stained either with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) or with LIVE/DEAD Fixable Aqua Dead Cell Staining Kit (Invitrogen). Cells were enumerated with AccuCheck counting beads (Invitrogen). For histology, CD3ε (clone 500A2) and CD45.2 (clone 104) Abs were obtained from BD Pharmingen, α-smooth muscle actin (α-SMA, clone 1A4) from Sigma-Aldrich, and Syrian Hamster secondary Ab from Jackson ImmunoResearch Laboratories. Abs used in the IFN-γ ELISPOT assays were purchased from Mabtech. H2-Db–restricted Mtb32309–318 peptide (GAPINSATAM) was obtained from PolyPeptide Laboratories. Depleting anti-CD4 (clone GK1.5) and anti-CD8 (clone YTS169.4) as well as rat immunoglobulin G 2b (IgG2b) isotype control monoclonal Abs were purchased from Bio X Cell. MB49 cells were received from the Brandau Group and cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen), complemented with 10% fetal calf serum (FCS, Eurobio) and 1% penicillin/streptomycin (Invitrogen). Poly-l-lysine was purchased from Sigma-Aldrich.

Tissue processing and flow cytometry

Bladders were resected and incubated in DMEM (Invitrogen) containing collagenase D (1 mg/ml) (Roche), Liberase TM (0.17 U/ml) (Roche), and deoxyribonuclease 1 (1 U/ml) (Invitrogen) at 37°C for two successive cycles of 30 min. Tissue suspensions were washed in DMEM plus 10% FCS, pressed through a 70-μm mesh, washed in PBS plus 2% FCS, pressed through a 40-μm mesh, and pelleted for FACS staining. Spleens were mashed, incubated at 37°C in 1.66% ammonium chloride (VWR International) in water for 5 min for red blood cell lysis, and filtered through a 70-μm mesh. All cells were preincubated with Fc block (and Aqua, if used), washed, and incubated with appropriate Abs for 20 min in PBS plus 0.5% FCS. Samples were run on a BD FACSCanto II flow cytometer (BD Biosciences) and analyzed with FlowJo (Treestar) software.

Monitoring BCG-specific T cell response

For IFN-γ secretion assays, at indicated time points, spleens and bladder draining LNs were harvested and combined; CD8+ T cells were purified with microbeads and MS columns (Miltenyi Biotec); and ELISPOT assays for IFN-γ–producing cells were performed as previously described (48). The ELISPOT plate evaluation was performed in a blinded fashion by an independent evaluation service (Zellnet Consulting). For tetramer staining, soluble Db-Mtb32309–318 monomers were produced with a modified version of that described (49) and conjugated with premium-grade streptavidin-phycoerythrin (Invitrogen), added for 1 hour at room temperature.

Immunofluorescence histology

Tissues were processed as previously described (50). Briefly, samples were fixed overnight at 4°C in a fresh solution of 4% paraformaldehyde (Sigma-Aldrich) in PBS, embedded in OCT compound (Sakura Finetek), and frozen at −80°C. Frozen blocks were cut at 8-μm thickness and sections were collected onto SuperFrost Plus slides (VWR International). Slides were dried for 1 hour and processed for staining or stored at −80°C. For staining, slides were first hydrated in PBS-XG [PBS containing 0.1% Triton X-100 (Sigma-Aldrich) and 1% FCS] for 5 min and blocked with 10% FCS in PBS-XG for 1 hour at room temperature. Slides were then incubated with primary antibodies in PBS-XG overnight at 4°C, washed, incubated with secondary antibodies for 1 hour at room temperature, incubated with DAPI for 5 min at room temperature, washed, and mounted with Fluoromount-G (SouthernBiotech). Slides were examined under an AxioImager M1 fluorescence microscope (Zeiss) equipped with a charge-coupled device camera, and images were processed with AxioVision software (Zeiss).

T cell depletion

Mice were injected intraperitoneally with a mixture of 100 μg of anti-CD4 and 100 μg of anti-CD8 antibodies, or with 200 μg of isotype control, 48 hours before instillation. Depletion efficiency was controlled on blood and splenocytes, and confirmatory studies indicated that splenic dendritic cell populations were unaffected by depletion protocols.


We performed analysis of recurrence-free survival stratified to available pre-BCG PPD status in 55 patients with high-risk non–muscle invasive bladder cancer, which were defined as any high-grade tumor or as any low-grade tumor with more than two recurrences within 2 years. Patients underwent transurethral resection of all visible bladder tumors and random bladder biopsies in case of a positive bladder wash cytology. High-grade tumors underwent a second resection 2 to 4 weeks after first resection, and then, within 2 weeks, BCG therapy was initiated at the Department of Urology, University Hospital of Bern, Switzerland. Patient and tumor characteristics were equally distributed over the patient cohorts (PPD positive versus negative; Table 1). Patients were followed as part of an observational clinical trial, in accordance with clinical guidelines. Tumor recurrence was defined on the basis of biopsy and urine cytology. The local ethical commission of Bern, Switzerland, approved the study and all patients provided informed consent.


Unless otherwise indicated, two-tailed Mann-Whitney nonparametric tests were used for statistical analyses with Prism software (GraphPad). Differences with a P value of 0.05 or less were considered statistically significant. For the mouse tumor challenge (Fig. 5), a log-rank test was performed. For patient data analysis (Fig. 6), a log-rank test was performed with SPSS 18.0 (SPSS Inc.). For kinetic studies (Fig. 1B), we used a general linear mixed model with Stata 11.0 (Stata Corporation). For comparison of CFUs in the draining LN (Fig. 2B), we used two-sided Fisher’s exact tests.

Supplementary Materials

Materials and Methods

Fig. S1. Administration of three or four weekly repeated instillations of BCG results in a similar kinetics of T cell infiltration into the bladder.

Fig. S2. Instillation of live but not heat-killed BCG results in T cell accumulation in the bladder.

Fig. S3. No bacterial growth is detected in the bladder draining LNs shortly after instillation.

Fig. S4. CFU measurements after subcutaneous BCG immunization.

Fig. S5. Preexisting adaptive immunity supports a robust, albeit short-lived, innate immune response.

Fig. S6. Inflammatory cells are dispensable for T cell recruitment to the bladder in preimmunized animals.

References and Notes

  1. Acknowledgments: We thank all members of the Albert Laboratory, G. Milon, L. Majlessi, R. Simeone, R. Brosch, R. Breban, and H. Law for their support and advice. We also thank S. Leroy and A. Fontanet for assistance with statistical analyses. The Center for Human Immunology and the Animal Facilty at Institut Pasteur were instrumental for the studies. We thank C. Leclerc for providing the ascitic fluid clones GK1.5 and H35.17.2. Funding: Work was funded by La Ligue Contre le Cancer and L’Institut National du Cancer and through the generosity of the Caisse de Retraite et de Prévoyance des Clercs et Employés de Notaires (M.L.A.) and the Swiss National Foundation (C.A.R.). Author contributions: C.B. conceived the project, performed all experiments, analyzed the data, and wrote the manuscript; C.A.R. contributed to tumor model development, generated and analyzed clinical data, and reviewed the manuscript; J.R.G. performed tumor treatment studies; F.D.B. generated and analyzed clinical data and reviewed the manuscript; H.J.-S. contributed to developing key protocols; F.L. contributed critical tools for T cell analysis; C.A. contributed technical assistance for mouse instillations; A.B. provided support for the study and reviewed the manuscript; P.B. provided technical expertise, experimental advice, and critical reagents; C.D. provided support for studies on BCG; L.P. contributed key protocols for assaying bladder leukocytes; G.N.T. generated and analyzed clinical data and reviewed the manuscript; M.L.A. conceived the project, analyzed the data, and wrote the manuscript. Competing interests: C.B., J.R.G., C.A.R., and M.L.A. are the inventors listed on patent application #EP12305086.6, “Improved cancer treatment by immunotherapy with BCG or antigenically related nonpathogenic mycobacteria.” The authors declare that they have no competing interests.

Stay Connected to Science Translational Medicine

Navigate This Article