Key points

Introduction

For decades, intravesical instillation therapy with the attenuated mycobacterial strain Bacillus Calmette-Guérin (BCG) has been a mainstay of treatment for patients with intermediate- and high-risk nonmuscle-invasive bladder cancer (NMIBC). However, disease progression and recurrence are common among patients on appropriate BCG therapy, and options for bladder-preserving subsequent therapy remain limited. Ongoing efforts to develop better second-line bladder-sparing therapies rely on clinical trials of patients deemed to have failed management with BCG. Both the design of these studies and critical interpretation of their results require the use of consistent definitions of BCG failure. Furthermore, these definitions serve a critical role in clarifying the clinical threshold for switching therapeutic strategies for a given patient. Proper definitions of BCG failure must therefore also be rational, systematic, and data-driven, to ensure that patients across all points along the BCG treatment and response spectrum receive optimal management. ^, Further complicating this challenge is the fact that accurate risk stratification in such patients must be dynamic, taking into account not only the parameters of a patient’s disease at the time of initial diagnosis such as stage and grade, but also his or her individual treatment history, recurrence timeline, and the parameters of any recurrent tumors. This article describes historical definitions of BCG failure, as well as recent efforts to better delineate and refine the clinical criteria for identifying individual patients who will not benefit from further intravesical BCG therapy. The authors also review guidance from the most recent expert consensus panels and professional association guidelines regarding which patients should not receive additional BCG therapy.

Bacillus Calmette-Guérin in bladder cancer management: a brief history

The antitumor potential of mycobacteria and the body’s inflammatory reaction to them grew out of early observations from an autopsy series published by Pearl in 1929, in which patients infected with tuberculosis were found to have lower rates of cancers. The Mycobacterium bovis strain Bacillus Calmette-Guérin (BCG), isolated from a heifer in 1913 and attenuated by serial passaging on glycerin-ox-bile, was already in widespread use in Europe as a vaccine against tuberculosis, and offered a more tenable mycobacterial model for capturing these desirable properties. Several groups showed promising evidence of antitumor activity conferred by BCG injection therapy in murine models using transplanted tumors or tumor cell lines, and demonstrated the importance of close contact between BCG and the tumor in achieving these effects.

The modern standard of adjuvant intravesical BCG following transurethral resection of bladder tumors (TURBT) was first described by Morales and colleagues in 1976, in a retrospective series of 10 patients who received paired intravesical instillations and intradermal injections with the mycobacteria. The authors describe weekly dosing with a combination of intradermal and intravesical therapy for a total course of 6 weeks, and reported no tumor recurrences in limited follow-up. A short time later in 1982, a larger randomized controlled trial of 57 patients receiving either TURBT alone or TURBT with intravesical and intradermal BCG instillations showed a significant reduction in tumor recurrence (21% vs 50%, P =.027) with the addition of BCG adjuvant therapy. Additional randomized trials demonstrated superior disease control with BCG when compared against existing adjuvant intravesical therapies: thiotepa and doxorubicin. ^, In 1990, BCG was approved by the US Food and Drug Administration (FDA) for the treatment of superficial bladder cancer. However, with this tremendous advance in therapy for NMIBC came a new set of problems in determining how best to manage patients who did not respond. Indeed, BCG nonresponders were described in the literature as early as 1986.

Historical definitions of Bacillus Calmette-Guérin failure

With published rates of tumor recurrence and/or progression in the range of 30% to 50%, there are clearly many patients who do not exhibit the desired degree of antitumor response to intravesical BCG therapy. However, it has proven difficult for the field to agree upon a rigorous evidence-based and uniform set of criteria to identify those patients who will not benefit from further BCG therapy, and should therefore be advanced to alternative management strategies ( Table 1 ). It is generally agreed that radical cystectomy is the current standard of care for such patients, but with its invasiveness and high rates of postoperative morbidity, this can be a difficult choice to justify for many patients and providers. As highlighted throughout this issue, efforts to develop novel salvage therapies for this difficult-to-manage patient population are ongoing.

The first modern attempt to rigorously define the patient population who would not benefit from further intravesical BCG therapy was published by Herr and Dalbagni in 2003. The authors retrospectively reanalyzed data from a 1987 study randomizing patients to receive either induction BCG alone (6 weeks) or induction BCG plus an early iteration of maintenance BCG (monthly for 2 years). All patients had repeat TURBT at 3 months and 6 months. This early iteration of maintenance BCG showed no beneficial effect in its population. The authors found that tumor status at the 6-month evaluation, but not at the 3-month evaluation, was highly predictive of the overall 2-year recurrence-free rate (3-month tumor status hazard ratio [HR] 1.51, P =.24; 6-month tumor status HR 9.18, P =.001). Their work helped to establish the timing of tumor recurrence as an important indicator of BCG responsiveness, and hinted at the existence of an early (3-month) time window in which persistent tumors may still respond favorably to further BCG therapy, a finding further supported by several groups reporting durable response rates of 50% to 60% to a second induction course of BCG for persistent tumors seen after a single induction course. ^{, ,} In contrast, further attempts at BCG reinduction after failing 2 induction courses have been repeatedly shown to have limited or no added benefit. ^, Taken together, these data provide the foundation for one of the most widely accepted criteria for defining patients unlikely to benefit from further BCG: those who have recurrent or persistent tumor present upon completion of 2 courses of BCG (typically at the 6-month timepoint from initial diagnosis). It is also worth noting that, depending on their tumor status after the first induction course, patients may receive either a second induction course (6 weekly doses) or a first maintenance course (3 weekly doses). These 2 populations have typically been grouped together in defining BCG responsiveness categories, although it remains possible that they may exhibit somewhat different underlying pathobiology.

Two years later, in 2005, an international group of authors issued a consensus guidance statement regarding the management of stage T1 bladder tumors, which included the first detailed description of a nomenclature system for classifying BCG nonresponders into several subcategories. The authors postulate some of these subcategories that were once grouped together as BCG failure may in fact be optimally managed with further BCG, including some patients with late relapsing disease. However, the authors did not provide specific clinical decision-making criteria. Their subcategory definitions were echoed in a 2006 review, and include

• •
  

  BCG refractory: failure to achieve a disease-free state by 6 months after initial BCG therapy with either maintenance or retreatment at 3 months because of either persistent or rapidly recurrent disease, also includes any progression in stage, grade, or disease extent by 3 months after the first cycle of BCG that is nonimproving or worsening disease despite BCG

  
  
• •
  

  BCG resistant: recurrence or persistence of disease at 3 months after induction cycle but of lesser degree, stage, or grade that subsequently is no longer present at 6 months from BCG retreatment with or without  TUR (disease improves then resolves with further BCG)

  
  
• •
  

  BCG relapsing: recurrence of disease after achieving a disease-free status by 6 months (ie, disease resolves after BCG then returns); relapse is further defined by time of recurrence: early (within 12 months), intermediate (12–24 months), late (>24 months); relapsing disease while on active maintenance (within 3 months) may qualify as BCG refractory

  
  
• •
  

  BCG intolerant: disease recurrence after a less than adequate course of therapy is applied because of a serious adverse event or symptomatic intolerance that mandates discontinuation of further BCG (ie, recurrent disease in setting of inadequate BCG treatment from drug toxicity)

This nomenclature system has largely persisted, with subtle alterations, and remains commonly utilized in the current literature. Notably, however, the original descriptions did not provide a comprehensive definition of a population of patients who should not receive further BCG. The first attempt to do so did not come until 2015, when an expert panel was convened at the Genitourinary Cancers Symposium (GU-ASCO) in Florida to help guide clinical trial design for patients with BCG-unresponsive NMIBC. This group of authors provided a new definition for a category of patients to be known as “BCG unresponsive,” “who have been treated with adequate BCG…and are unlikely to benefit from and should not receive further intravesical BCG.” Specifically, the GU-ASCO group included the following patients in its BCG unresponsive definition: (1) new or persistent high-grade tumor present at or around 6 months after initiation of BCG therapy, (2) relapse with high-grade tumor within 6 months of last intravesical exposure to BCG despite an initial complete response, or (3) high-grade T1 tumor present at first evaluation following induction BCG alone (at least 5 of 6 doses). They further stipulate that patients qualifying under either (1) or (2) must have completed “adequate BCG therapy,” defined as receiving at least 5 of 6 doses of an initial induction course, and at least 2 of 3 doses of a maintenance course. This was later broadened by others to also include patients who have received 5 of 6 doses of a first induction course followed by 2 of 6 doses of a second induction course, in order to accommodate the increasing practice of administering a second induction course of BCG to those with a partial response after their first induction course.

Beginning in 2011, the International Bladder Cancer Group (IBCG) expert panel also published several consensus statements regarding management of NMIBC and clinical trial design in this patient population, and specifically addressing the issue of defining and managing BCG failure. ^, Panel members propose that an important distinction must be made between treatment failure (any recurrence or progression during therapy) and recurrence (reappearance of disease after completion of therapy). Echoing the position of the international consensus statement from 2005, they posit that certain patients with recurrent disease appearing after completion of BCG therapy can often be successfully managed with additional intravesical therapy.

The major focus of the IBCG, however, has been on providing recommendations for standardization of clinical trials in NMIBC. They specify that the subcategories of BCG refractory, BCG relapsing, and BCG intolerant as defined by Nieder and colleagues should be used. Furthermore, they reaffirmed the GU-ASCO 2015 consensus panel definition of a group of BCG unresponsive patients, including all patients with BCG refractory disease, as well as those with BCG relapsing disease within 6 months of their last BCG exposure. ^, For this group of BCG unresponsive patients, they specify that additional BCG intravesical therapy is not a feasible option, and radical cystectomy is considered the only standard of care management option. Because placebo controls are considered unethical in this population, and radical cystectomy is a drastically more invasive option that cannot practicably be randomized, the IBCG proposes that single-arm studies or comparison against investigator-choice alternative therapies are appropriate for testing novel salvage therapies in the BCG unresponsive patient population.

In aggregate, these studies and consensus statements present a relatively consistent, if not completely unified, set of definitions for patients who should not receive further BCG therapy. Yet further refinements to the nomenclature and clinical criteria for BCG failure are ongoing. In 2016, Steinberg and colleagues performed retrospective cohort analysis on a group of patients enrolled in a phase 2 trial of intravesical BCG (at one-third the standard dose) mixed with interferon (IFN)-α. They specifically examined those patients in the trial who had previously failed intravesical BCG, and evaluated potential risk factors for developing recurrence of either pure papillary disease or CIS. Although those patients who had previously received 2 or more courses of BCG were more likely to fail the BCG plus IFN treatment regimen regardless of whether they had papillary disease or CIS, the time to BCG failure appeared to confer different levels of risk for the different disease subtypes. Hazard ratios for failing the BCG plus IFN regimen were calculated, using as a reference those patients with a history of BCG treatment whose recurrent disease developed greater than 12 months after therapy initiation. For those with papillary disease, a prior BCG failure within 6 months conferred significantly increased risk of failure of BCG plus IFN (HR 1.82, P =.02), while failure at 6 to 12 months carried identical risk compared to those with a prior failure interval greater than 12 months (HR 1.00, P =1.00). By contrast, for patients with a history of CIS and BCG failure, a failure interval of either less than 6 months (HR 2.59, P <.01) or 6 to 12 months (HR 2.29, P =.04) conferred a significantly greater risk of failing the BCG plus IFN treatment regimen, suggesting that the recurrence interval has dramatically different implications for patients with pure papillary disease versus CIS. Taken together, these data indicate that a patient with a pure papillary recurrence at greater than 6 months may have disease characteristics similar to a BCG-naïve patient, whereas recurrent CIS at up to 12 months carries an elevated risk of failing any attempted further BCG therapy.

The most recent consensus statement addressing BCG failure was published in early 2019 by the International Consultation on Urologic Diseases (ICUD) and the Société Internationale d’Urologie (SIU) and reflects an international expert panel convened in 2017. They utilize the GU-ASCO framework nomenclature, reaffirming that patients with BCG unresponsive disease (BCG refractory disease, plus those with early relapse within 6–9 months of last BCG exposure) should not receive further BCG, and should be offered radical cystectomy as the standard of care.

Association guidelines

Professional society guidelines for the management of NMIBC devote significant attention to the definition and management of BCG failure (see Table 1 ). The AUA/SUO guideline on diagnosis and treatment of NMIBC was most recently updated in 2016, and recommends that no further BCG therapy be administered to patients who either (1) are intolerant of BCG or (2) have a documented recurrence on TURBT of high-grade nonmuscle invasive disease and/or CIS within 6 months of 2 induction courses of BCG or induction BCG plus maintenance. The European Association of Urology (EAU) guidelines, most recently updated in 2017, are highly concordant but with a slightly more wide-ranging timeline for high-grade recurrence. In the European guideline, unsuccessful BCG treatment is defined by the persistence of high-grade disease at 3 months or the appearance of new high-grade disease (in a patient who previously had only low-grade disease) at any time during therapy, or by the presence of CIS at both 3 and 6 months after the initiation of BCG therapy (noting that a second course of BCG for those with CIS present at 3 months can achieve a complete response in up to 50% of patients by 6 months). The NCCN guidelines for bladder cancer, most recently updated in 2019, do not provide specific criteria for defining BCG failure or subcategories. They do not make specific management recommendations, but stipulate that for post-treatment recurrent NMIBC, providers should take into account individual risks for disease progression including tumor stage, grade, size, and number, and viable management strategies may include further BCG, alternative adjuvant therapies, and radical cystectomy.

The extended 12-month recurrence window for CIS as described by Steinberg and colleagues was incorporated into a guidance statement issued by the US Food and Drug Administration (FDA) in 2018, targeted at standardizing the design of clinical trials for BCG-unresponsive NMIBC. Among other specifications, they reaffirm that radical cystectomy is the standard of care for such patients, and that single-arm studies are acceptable in the BCG unresponsive disease state given the known likelihood of disease progression without therapy, impracticality of randomization to cystectomy, and lack of effective comparator therapies. They provide a strict definition of BCG unresponsive disease largely based on the framework of the GU-ASCO panel, which includes

• 
  

  Patients with persistent or recurrent CIS alone or with recurrent Ta/T1 disease within 12 months of completion of adequate BCG therapy (defined as at least 5 of 6 doses of initial induction, plus at least 2 of 3 doses of a maintenance course or 2 of 6 doses of a second induction course)

  
  
• 
  

  Recurrent high-grade Ta/T1 disease within 6 months of completion of adequate BCG therapy

  
  
• 
  

  T1 high-grade disease at the first evaluation following an induction BCG course

Although the FDA guidance statement does not explicitly describe how to categorize recurrent disease arising in the context of additional ongoing BCG therapy (ie, maintenance therapy), it is generally accepted by the field that the 6- and 12-month timeframes used in the definitions are to be measured from the time of most recent BCG exposure.

Future directions

As described throughout this issue, much remains to be determined in the optimal management of NMIBC and BCG failure. In particular, the authors recommend that future efforts should focus on improving the prognostic stratification and management of intermediate-risk NMIBC, as defined by AUA and EAU guidelines. This common disease category is a particular challenge for patients and providers alike, because of the large degree of heterogeneity and uncertainty in determining its optimal management. Novel biomarkers and prognostic tools will be particularly important in this group to better predict disease behavior and guide management. This is also an important patient population for the development of novel therapeutic strategies, although the high degree of heterogeneity within this group should be carefully considered and controlled for when designing studies. Another area of interest is to better elucidate the cellular and molecular mechanisms for the BCG antitumor response, including the concept of “priming,” whereby early antigenic exposure may help train the body’s immune system to mount a powerful antitumor response. Finally, even as development of new therapies continues, disruptions in the usable supply of BCG impose logistical constraints onto its use, further emphasizing the need for an expanded armamentarium against this common disease entity.

Summary

Intravesical BCG therapy is the adjuvant treatment of choice and standard of care following thorough TURBT for intermediate- and high-risk NMIBC. However, rates of BCG failure remain high. Efforts to develop and improve alternative management strategies for NMIBC patients who fail BCG are ongoing, as highlighted throughout this issue. However, radical cystectomy, with high rates of perioperative morbidity, remains the accepted gold standard of care. Shortages of the BCG supply chain, such as the current shortage in the United States, further emphasize the need for viable alternative management strategies.

Accurate assessment of the risks of disease recurrence and progression in NMIBC patients requires careful consideration of multiple dynamic factors, including patient and tumor biology, as well as prior therapies received, with the timeline of therapeutic interventions and disease states playing a pivotal and complex role. Despite limited high-quality primary data, there is general consensus among expert panels and association guidelines in identifying those patients and tumors unlikely to respond to further BCG therapy. Future efforts must focus on further refinement and unification of precise, clinical evidence-based thresholds for abandoning further BCG therapy, with the goal of minimizing the number of patients whose disease and treatment history place them in the uncertain gray area between receiving further BCG or progressing to alternative management strategies. Current evidence suggests that these thresholds may be markedly different for different stages and subtypes of NMIBC. If one is to provide optimal care for each patient with this challenging disease entity, further study is needed.