Since the first report by Morales et al.,1 the efficacy of BCG in the prophylaxis of high-risk superficial bladder tumors and treatment of carcinoma in situ has been demonstrated2 Although BCG therapy is the most significant advance in conservative management of these patients, its efficacy is limited by the frequency and severity of adverse events and by the empirical nature of instillation schedules and doses. Lamm et al.3 suggested that an initial induction cycle of 6 weekly intravesical BCG instillations is suboptimal, unless maintenance therapy (3 consecutive weekly instillations) is given 3, 6, 12, 18, 24 and 30 months later. However, maintenance therapy is also hindered by troublesome adverse events.3, 4
The antitumor response is closely related to the immune response induced by BCG. Skin test reactivity, bladder-wall granulomas, urinary excretion of cytokines (e.g., IL-8 and IL-2) and leukocyturia have been studied as immune markers in this setting. Some have prognostic value for tumor recurrence during the first BCG induction course, but their clinical relevance is controversial.5, 6, 7, 8, 9, 10 Nevertheless, BCG nonresponders must be identified rapidly as early failure is associated with a grim prognosis.11, 12, 13 Immune response are often categorized as either type 1 (Th1) or type 2 (Th2), based on the profile of cytokines produced by both CD4+ and CD8+ T cells.14 A Th1 (IFN-γ/IL-2) cytokine profile is most often associated with cell-mediated immunity and the most suitable for eradication of malignant cells, whereas Th2 (IL-4/IL-10) cytokines favor the development of antibody responses.14, 15
BCG involves the generation of an enhanced Th1 cytokine immune environment in the bladder wall, which may contribute to the generation of productive tumor-specific immunity and may generate bladder tumor–killing cells comparable to lymphokine-activated killer cells.5, 6 Enhanced urinary Th1 cytokine levels (IL-2 and IFN-γ) have been observed during intravesical BCG therapy and may be associated with the immune reaction against bladder cancer.7, 8 However, even if a large body of evidence suggests that a Th1 urinary response after a first course of intravesical BCG is necessary for a favorable antitumor response and that early urinary IL-2 excretion after a second course might identify patients who qualify for additional BCG instillations, tumor-specific immunity, either clinically or in experimental animal models, has not been convincingly demonstrated.7, 8, 10 Moreover, in an animal model, the Th1-type immune response appears to be markedly increased in IL-10 KO mice, suggesting the potential role of IL-10 in controlling the Th1 response.16
We therefore prospectively investigated urinary Th1(IL-2)/Th2(IL-10) immune responses in patients receiving intravesical BCG induction and maintenance therapy as a marker for response to BCG.
BCG, bacillus Calmette-Guérin; iAES, adverse events score during induction course; KO, knockout; NK, natural killer; NPV, negative predictive value; PPD, purified protein derivative; PPV, positive predictive value; SITA, sophisticated individualized therapeutic approach; tAES, adverse events score during maintenance cycle.
MATERIAL AND METHODS
Patients and treatment
From April 1996 to January 2000, 39 patients (5 women and 34 men) with a history of multifocal or recurrent papillary transitional cell carcinoma (primary/secondary carcinoma in situ,n = 1; stage Ta/T1 carcinoma associated with grade 3 carcinoma in situ,n = 1; stage Ta/T1 grade 3, n = 18; stage Ta/T1 grades 1 and 2, n = 19) were monitored during intravesical BCG induction and maintenance therapy.
The instillation schedule was that described by Lamm et al.3 The first induction cycle of 6 weekly instillations was given 3–6 weeks after the last transurethral resection of the papillary tumor(s) [81 mg of BCG Connaught (Immucyst, Aventis Pasteur SA, Lyon, France) dissolved in 50 ml of saline and instilled via a 14 F catheter]. This was followed by maintenance therapy consisting of 3 weekly instillations given at 3, 6, 12, 18, 24, 30 and 36 months. Intradermal BCG inoculation was not given. Urine was cultured before each instillation. Patients were instructed to retain the solution for approximately 2 hr, to drink large amounts of fluids and to avoid physical activity on the day of instillation. They were seen at 3-month intervals, notably for voided urinary cytology and flexible cystoscopy. Transurethral resection and biopsy were used for all recurrent papillary tumors and for suspected recurrences. Maintenance therapy was stopped if severe adverse events or recurrences occurred. Historic mycobacterial contact status was assessed before instillation using an intermediate-strength PPD skin test, standard chest radiography and a questionnaire on vaccination against tuberculosis.
Complete responses were defined on the basis of histologic and cytologic results and classified as complete when histology and cytology were negative. Tumor recurrence was defined as positive transurethral resection and/or biopsy, without stage progression. Tumor progression was defined according to Herr et al.17 as tumor stage progression, muscle infiltration, metastasis or tumor recurrence requiring additional therapy or cystectomy.
Urinary cytokine assays
Spot urine samples were collected from each patient prior to and 4, 6 and 8 hr after each instillation. Samples were immediately centrifuged for 10 min at 800 rpm to remove cells and debris. They were then dialyzed, with m.w. cut-off of 6–8 kDa, for 24 hr in PBS (pH 7.4). Samples were stored in aliquots at −70°C until analysis. IL-2 and IL-10 were determined using a highly specific and reproducible commercial oligoclonal ELISA 8 hr after each instillation.18 The ELISA kits used for IL-2 and IL-10 were D2050 (R & D Systems, Minneapolis, MN), with a detection limit of 6 pg/ml, and 80-3747-00 (Genzyme, Cambridge, MA), with a detection limit of 5 pg/ml. All samples were tested in duplicate. Cytobacteriologic examinations and leukocyturia evaluation were done 3 days after each instillation, on freshly voided urine.19
Adverse events scale
A 4-class scale for the severity and duration of local and systemic adverse events was used prospectively, as previously described, and an adverse events score calculated for each patient during the induction course (iAES) and throughout the maintenance cycle (tAES).4
Data were analyzed with StatView software (SAS, Cary, NC). The following tumor and host variables were analyzed: age, sex, stage, grade, multifocality, tumor size, blood group, prior intravesical BCG therapy, history of primary infection by Mycobacterium tuberculosis, PPD skin test status before BCG instillations, tuberculosis vaccination status, cumulative doses of BCG, leukocyturia 3 days after instillation, highest leukocyturia observed during induction course, urinary IL-2 and IL-10 levels, adverse events after each instillation and iAES and tAES. The log-rank test was used to identify correlations between cytokine levels and the risk of tumor recurrence and progression. Times to recurrence and progression were estimated using the Kaplan-Meier method. Correlations between urinary cytokine levels and tumor and host variables were analyzed with the Mann-Whitney U-test. Correlations between cytokine levels were determined using Spearman's rank correlation coefficient (r). All reported p values are 2-sided, and significance was assumed at p < 0.05.
Median follow-up was 32 months (range 7–72 months), and average age was 64 (range 49–85) years. Complete responses to BCG were obtained in 30/39 patients (77%); the other 9 patients (23%) relapsed, and 4 (10.2%) of these progressed.
Mean times to recurrence and progression were 25.2 (range 7.2–48.5) months and 34.2 (range 16.9–72) months, respectively. Table I lists the relevant clinical and pathologic parameters of tumor aggressiveness. Two patients with progression were treated by cystectomy and 2 with systemic chemotherapy. Table II lists relevant tumor and host variables and their correlation with urinary IL-2 and IL-10 levels during treatment. Responses were assessed on the basis of histologic and cytologic parameters, and patients were classified as responders if urinary cytology and bladder-wall biopsy were negative at the last follow-up.
Table I. Clinical and Pathologic Characteristics of Tumor Aggressiveness in BCG Responders and Nonresponders
Number of patients
Carcinoma in situ, grade 3, primary/secondary
Stage pTa/pT1 carcinoma associated with grade 3 carcinoma in situ
Stages pTa/pT1, grade 3 carcinoma
Stages pTa/pT1, grades 1 and 2 carcinoma
Previous treatment with BCG
Table II. Tumor and Host Variables Correlating with Mean Urinary IL-2 and IL-10 Values During BCG Induction and Maintenance Therapy
IL-2 mean value (range) during treatment (pg/ml)
IL-10 mean value (range) during treatment (pg/ml)
Mann-Whitney U-test; all p values are 2-sided, with significance assumed at p < 0.05. NS, nonsignificant.
≥65 (n = 23)
<65 (n = 16)
Male (n = 34)
Female (n = 5)
Recurrent disease before BCG
Yes (n = 22)
No (n = 17)
High-risk tumor (pT1G3, carcinoma in situ)
Yes (n = 20)
No (n = 19)
Yes (n = 19)
No (n = 20)
Tumor size (> 11 mm)
Yes (n = 17)
No (n = 17)
O (n = 12)
Non-O (n = 13)
Recurrence after BCG
Yes (n = 9)
No (n = 30)
Progression after BCG
Yes (n = 4)
No (n = 35)
Vaccination against tuberculosis
Yes (n = 29)
No (n = 10)
PPD skin test
≥8 mm (n = 10)
<8 mm (n = 8)
Prior BCG therapy
Yes (n = 10)
No (n = 29)
≥1 (n = 15)
<1 (n = 24)
≥2 (n = 23)
<2 (n = 16)
Greatest leukocyturia during induction course
≥100,000/ml (n = 27)
<100,000/ml (n = 12)
Depending on the detection of urinary IL-2 during any of the 6 weekly instillations of the induction course, patients were classified as IL-2 producers or nonproducers. Urinary IL-2 was never detected in 53.8% patients (21/39), while 46.2% of patients (18/39) had detectable levels during at least one instillation. IL-2 was detected in 56.6% (17/30) of responders and 11.1% (1/9) of nonresponders. IL-2 nonproducers during the induction course had a shorter time to recurrence after BCG than IL-2 producers (p = 0.01, Mantel-Cox proportional hazards regression model) (Fig. 1a). IL-2 was detected from the first instillation onward in responders, peaking at the fifth or sixth instillation. In nonresponders, IL-2 was detected from the second instillation onward, peaking at the sixth instillation. Nonresponders with detectable urinary IL-2 had a late recurrence (21.3 months after the beginning of BCG treatment).
During the first extended induction cycle by BCG (6+3 schedule), urinary IL-2 was not detected in 35.9% of patients (14/39), while 64.1% of patients (25/39) had detectable urinary IL-2 during at least one instillation. IL-2 was detected in the urine of 73.3% (22/30) of responders and 33.3% (3/9) of nonresponders. Patients with persistent absence of IL-2 release during the first extended induction cycle (6+3 schedule) had shorter times to recurrence and progression than IL-2 producers (both p = 0.01, Mantel-Cox proportional hazards regression model) (Fig. 1b,c).
During maintenance therapy, IL-2 was never detected in 25.6% of patients (10/39), while 74.4% of patients (29/39) had detectable urinary IL-2 during at least one instillation. IL-2 was detected in the urine of 80% (24/30) of responders and 55% (5/9) of nonresponders (Fig. 2Aa,b). Five nonresponders who had detectable IL-2 during maintenance therapy had relatively late recurrences (mean 30.6 months, range 17.7–48 months after starting treatment); in contrast, 4 nonresponders in whom IL-2 was never detected during maintenance therapy had relatively early recurrences (mean 15.2 months, range 7.2–29.5 months). Urinary IL-2 was detected after the first instillation of each extended induction cycle in most cases. Levels fell after the second and third instillations in both responders and nonresponders. Urinary IL-2 levels were higher in responders than in nonresponders after the fifth instillation (p = 0.04) and the seventh instillation (p = 0.02). After the seventh instillation, 22/39 patients (56.4%) had produced IL-2. Of these, 21 (95.5%) remained free of tumor during follow-up. One patient, with pTaG2 and urinary IL-2 level close to 200 pg/ml, recurred 36.7 months after the beginning of treatment. Sensibility, specificity, PPV and NPV to detect patients free of tumor recurrence, using ELISA IL-2 urinary assay after the seventh instillation, were, respectively, 72.4%, 83%, 95% and 38%.
During maintenance therapy, urinary IL-2 levels increased with age (p = 0.02), adverse events (tAES, p = 0.0009) and tuberculosis nonvaccination status (p = 0.0002). IL-2 levels were not related to cumulative dose of BCG.
Depending on the detection of urinary IL-10 during any instillation of the induction course or the reinduction courses, patients were classified as IL-10 producers or nonproducers. During the induction course, urinary IL-10 was never detected in 59% of patients (23/39), while 41% of patients (16/39) had detectable IL-10 during at least one instillation. IL-10 was detected in 30% (10/30) of responders and 66.6% (6/9) of nonresponders (Fig. 2c,d). In responders, IL-10 was detected from the third instillation onward, peaking at the sixth instillation. In nonresponders, IL-10 was first detected after the second instillation, also peaking at the sixth instillation. Detection of urinary IL-10 during initial or subsequent induction courses did not correlate with time to recurrence or progression (p > 0.05, Mantel-Cox proportional hazards regression model).
During maintenance therapy, IL-10 was detected in the urine of 80% (27/30) of responders and 66.6% (6/9) of nonresponders. IL-10 was detected after the first instillation of each extended induction cycle. Levels peaked after the second or third instillation in both responders and nonresponders. IL-10 levels were higher in responders than in nonresponders after the seventh, ninth, eleventh and twelfth instillations (p = 0.02, p = 0.05, p = 0.03 and p = 0.03, respectively). IL-10 levels increased with cumulative dose of BCG (p = 0.01) and tAES (p = 0.003) and were higher in blood group O patients (p = 0.003) and those with a baseline PPD skin test wheel larger than 8 mm (p = 0.0007). Urinary IL-10 levels were lower in women (p < 0.001). Patients previously failing BCG had lower urinary IL-10 level compared to patients never previously treated with intravesical BCG (p = 0.0006).
Mean urinary IL-2 and IL-10 values and kinetics in responders and nonresponders are shown in Figure 2e,f. IL-2 and IL-10 levels correlated with each other (p = 0.009). During maintenance therapy, IL-2 and IL-10 peaked after the seventh and ninth instillations, respectively, in responders. Subsequent instillations were associated with a fall in IL-2 and an increase in IL-10. In nonresponders, IL-10 peaked after the eleventh instillation and IL-2 after the thirteenth instillation.
During subsequent extended induction cycles, the first instillation was associated with higher IL-2 levels than IL-10 levels, whereas the contrary was observed after the second and third instillations. This switch occurred in 75% of patients and was associated with leukocyturia above 100,000/ml during induction therapy (p = 0.0001) and with an iAES >1 (p = 0.03). The switch was not related to subsequent recurrence or progression, though it occurred earlier in nonresponders than in responders (fifth and eleventh instillations, respectively).
During maintenance therapy, 43.6% of patients stopped treatment after developing severe adverse events. The average iAES was 1.1 (range 0–2), and the average tAES was 1.5 (range 0.5–2.4). The iAES was the only predictive factor of the tAES (p = 0.001). The severity of adverse events after each instillation (class 0–IV) increased with leukocyturia (p < 0.0001) and cumulative dose of BCG (p < 0.0001) (data not shown).
Although BCG is now recommended as an adjunctive treatment for superficial bladder tumors, the optimal treatment schedule remains to be defined. Lamm et al.3 suggested that an initial induction cycle of 6 weekly intravesical BCG instillations is suboptimal unless maintenance therapy (3 consecutive weekly instillations) is given 3, 6, 12, 18, 24, 30 and 36 months later. However some questions remain concerning the need to apply the 3 weekly maintenance therapy systematically or not. A better SITA, based on immune parameters, might be valuable in this setting.
Several parameters of the immune response, such as the presence of granulomas in the bladder wall and conversion of the PPD skin test, have been linked to the clinical response to BCG instillation.20, 21 However, their individual predictive value is low, and they may lack immunologic specificity.6 Urinary IL-2 may be more relevant for routine monitoring as its level reflects the local immunologic reaction in the entire bladder.10, 22
We studied the kinetics of urinary IL-2 (Th1) and IL-10 (Th2) in responders and nonresponders during a first 6 weekly BCG induction course after transurethral resection and during a series of 3 weekly instillations given 3, 6, 12, 18, 24, 30 and 36 months later. We confirm that urinary IL-2 levels are predictive of time to recurrence and show prognostic value for time to tumor progression, pointing to the relevance of routine IL-2 monitoring.7, 10
IL-2 production is predominantly restricted to T lymphocytes and can be considered a sign of immune responsiveness to intravesical BCG. Nevertheless, while individuals unable to generate a Th1 response are considered immunologically unresponsive to the tumor,23 the absence of effective antitumor immunity (Th1) does not necessarily correlate with the lack of an antitumor response but may be associated with an ineffective or unproductive response. However, some patients have shown apparent long-term durable complete response without any detectable urinary IL-2 and IL-10. This might be explained by the fact that BCG response is probably related to multiple factors, such as immunologic response and tumor characteristics (biologic aggressiveness).
During the BCG induction course, IL-10 peaked before IL-2 in nonresponders, whereas IL-2 peaked before IL-10 in responders. Although urinary IL-10 excretion was not associated with clinical outcome, these results confirm that an unproductive response (absence of IL-2) or an ineffective one (low level of IL-2, high and/or early levels of IL-10) is associated with BCG failure. Moreover, the occurrence of late tumor relapse or progression, despite low levels of IL-2 during the induction or extended induction cycle, suggests that a threshold level of local IL-2 production in the bladder may be needed for an optimal antitumor response.
During the extended induction cycle, urinary IL-2 and IL-10 levels peaked earlier than during the induction courses in both responders and nonresponders, pointing to accelerated cytokine induction on repeat BCG exposure.18, 24 This supports the use of weekly instillations to boost the BCG immune response as IL-2 induces CD45RA+/CD45RO− cells (a naive T-cell subset) to express the memory phenotype.25 Interestingly, 80% of responders to maintenance therapy (compared to only 56.6% of responders to an initial induction course) had detectable urinary IL-2. Most (73.3%) produced urinary IL-2 during the first extended induction cycle. This may lead to a corresponding increase in delayed-type hypersensibility reactions and subsequent tumor elimination, reinforcing the usefulness of the first extended induction cycle and maintenance instillations.3, 23, 26
Unfortunately, maintenance instillations gradually became less effective over time, IL-2 levels falling and IL-10 levels increasing. Several explanations might be associated with this phenomenon. During the initial instillations of BCG treatment, minimal inflammation would permit substantial mycobacterial pathogen expansion, resulting in the generation of high foreign peptide levels sufficient to prime T cells and to generate IL-2. Gradual recruitment and activation of inflammatory cells (macrophages and NK cells) would subsequently restrict pathogen expansion and thereby inhibit peptide generation, even if some authors have shown that BCG retention after BCG instillation in an animal model increased with the number of instillations.27, 28 Moreover, prior exposure to BCG could partly explain the lack of strong Th1 responses during subsequent courses, in the same way that environmental mycobacterial infection reduces the efficacy of BCG vaccination.29, 30 A further possible explanation for the failure of BCG to induce strong Th1 responses during the extended induction cycle is that Th1 reactions can be downregulated by Th2 cytokines.26, 31 By contrast to Th1, Th2 responses elicit production of functionally antagonistic cytokines (IL-4 and IL-10) that suppress delayed type hypersensitivity reactions.26 A mutually Th1/Th2 antagonistic response coexists via effects on the antigen-presenting cells: IFN-γ inhibits Th2 T-cell proliferation, while IL-4 and IL-10 inhibit cytokine production by Th1 T cells.31 IL-10 can suppress local macrophage activation in vivo.32 Macrophage effector functions are tightly regulated, presumably to limit bystander damage.33 In the context of cell-mediated reactivity to ongoing antigenic stimulation, as in chronic infections, macrophages are usually subjected to both positive and negative stimuli. IL-10 could limit IFN-γ production at the same time as it activates macrophages, limiting the antigen-presentation function, reactive oxygen species generation and membrane B7 and MHC class II antigen expression.34 During maintenance therapy, we found that IL-2 and IL-10 levels correlated with each other, pointing to an antagonistic feedback loop between Th1 and Th2 cytokines.
Th1/Th2 cytokine imbalances, with IL-10 overproduction, may be associated with a worse prognosis of some cancers.35 We found that IL-10 peaked before IL-2 in nonresponders to induction courses, whereas IL-10 was associated with bladder inflammation (leukocyturia) and mainly local and regional adverse events during maintenance therapy.4 Whether these IL-2/IL-10 cytokine imbalances have a deleterious effect on future antitumor activity remains to be determined in longer follow-up studies. Nevertheless, IL-10 appears to suppress T-cell activation and function in an antigen-specific manner, resulting in an induced state of antigen-specific anergy in CD4+ and CD8+ cells.36 These anergic T cells show impaired proliferation, reduced IL-2 and IFN-γ production and reduced capacity to lyse malignant cells.37 Early IL-10 production, associated with little or no IL-2 production during the BCG induction course, could reflect such anergy in nonresponders. Likewise, IL-10 production and the switch from a Th1 to a Th2 urinary cytokine pattern during BCG maintenance therapy could be an indirect sign of adverse events, as suggested by Miller et al.37 in chronic prostatitis.
We observed a correlation between the cumulative dose of BCG and urinary IL-10 levels. Low and high antigen doses are thought to promote a Th2 response, while moderate antigen doses predispose naive cells to become Th1 cells.38 Low and high doses of high-affinity antigens yield Th1 cells, while moderate doses yield Th2 cells.39 At almost any dose, antigens favor a Th0 phenotype; in this case, the key to subsequent differentiation is the level of available IL-2.40 Further studies of relevant BCG strain antigens and the dose inducing the optimal Th1 response are needed. The level of signaling achieved through the T-cell receptor and its association with induction of Th1/Th2 cytokines remain to be explored. Moreover, use of recombinant BCG inducing a strong Th1-like response is needed if the relationship between vaccination against tuberculosis, poor IL-2 production and recurrence is to be demonstrated.41, 43
In conclusion, during maintenance BCG therapy of bladder cancer, high urinary levels of IL-2 were associated with protection from recurrence and progression. IL-2 management during maintenance therapy might be proposed. Patients in whom urinary IL-2 is undetectable after the 6+3 schedule might benefit more from careful medical supervision and/or alternative therapeutic regimens than from the extended induction cycle. After the first BCG reinduction course, the Th1 profile switched gradually to a Th2 profile. This switch was associated with an increase in adverse events and IL-10 levels.
During BCG maintenance therapy, monitoring extended induction instillations based on IL-2/IL-10 levels might allow the physician to delay and/or to decrease adverse events, the IL-10 retro-control loop, and to promote IL-2 overproduction and subsequent tumor elimination.