Bacillus Calmette-Guérin (BCG) treatment of bladder cancer: a systematic review and commentary on recent publications

Abstract

Bacillus Calmette-Guérin (BCG) is the standard immune therapy for non-muscle invasive bladder cancer. A systematic review of published articles regarding BCG treatment of bladder cancer was conducted and a commentary of these major developments is provided. Emerging strategies are aimed to improve the clinical response to BCG through vaccination with live organisms or BCG proteins. Several BCG strains are utilized worldwide. As the understanding of genetic and phenotypic differences in these strains is elucidated, inquiries into the potential clinical effects of these various strains have been studied. Data suggest that some strains could be more effective than others but further study is needed. Although response to BCG is heterogenous, current clinical practice does not incorporate use of biomarkers to delegate treatment selection. Thus, biomarker prediction is an important area of research in this area. Novel urine and tissue markers show promise in this endeavor. Significant developments have also occurred in understanding BCG’s mechanism of action, which remains incompletely understood. Notable publications include mechanistic studies showing a role for T cells, NK cells, mast cells and granulocytes in BCG’s antitumor efficacy. Future work includes efforts to create recombinant BCG strains to decrease side effects, repeated instillations, and increase overall efficacy.

Introduction: Discovery and Treatment of Bladder Cancer

Bacillus Calmette-Guerin (BCG) was developed by Albert Calmette and Calmille Guerin at the Pasteur Institute over a decade of culturing mycobacterium bovis in an effort to develop a vaccine against tuberculosis which was rampant in Paris in the early 21^st century. The idea that BCG could be utilized to treat cancer was demonstrated by Berton Zbar and colleagues at the National Cancer Institute in 1970¹ and subsequently adopted to treat patients with melanoma and bladder cancer (BC)². Based on its clinical activity, including the favorable results from a landmark study comparing intravesical BCG against intravesical doxorubicin³, BCG was approved for the treatment of carcinoma in-situ (CIS) of the bladder and for the prevention of relapse of papillary Ta and T1 BC. Subsequent randomized controlled trial determined that maintenance therapy provided an improved long-term benefit compared to treatment without maintenance⁴, leading to the current standard of care utilized in 2018. BCG is given weekly, retained in the bladder for 2 hours as an induction course (6 weekly treatments) followed by maintenance course (3 weekly treatments) given at 3, 6, 12, 18, 24, 30, and 36 months following tumor removal (total maintenance courses = 7, total BCG instillations = 27).

Methods

We conducted a search of published literature using pubmed using the following search approach for BCG: (“BCG Vaccine”[Mesh] OR “antigen 85, Mycobacterium bovis” [Supplementary Concept] OR “Mycobacterium bovis”[Mesh] OR “BCG fibronectin attachment protein, Mycobacterium bovis” [Supplementary Concept] OR “BCG lipopolysaccharide” [Supplementary Concept] OR “BCG Connaught” [Supplementary Concept]) NOT Urinary bladder neoplasms) and the following search term BC “Urinary Bladder Neoplasms”[Mesh]. Examination of the trends in publication regarding BCG and BC reveals noteworthy findings. In the 1980’s, the number of BCG (non-BC) publications steadily decreased whereas the number of publications of BCG (BC) increased. This is a result of the successful implementation of BCG use for BC treatment. However, this also reflects the unfortunate waning amount of research in BCG tuberculosis. Fortunately, an increase in both categories was subsequently observed in the 1990’s and now, publications have plateaued but at a steady average/year (Figure 1). These trends reflect promisingly on a robust scientific interest in BCG and also indicate that much remains to be discovered regarding BCG’s therapeutic consequences, response prediction, and mode of action. Here, we review novel discoveries made in BCG/BC literature over the last 2 years.

Figure 1. Trend of publications from 1970–2018.

The average number of articles per year involving research with BCG and BCG in bladder cancer (BC) and published through the years 1970–2018 are plotted.

BCG strains

Currently used BCG has undergone two separate phases of attenuation. First, in Dr. Guerin’s laboratory, the Mycobacterium bovis organism that had been isolated from a cow infected with bovine tuberculosis was repeatedly cultured through 230 in vitro passages from 1908–1921 until it was attenuated enough to be used in humans. Then, in 1924 BCG was widely disseminated for use worldwide including other parts of Europe, Russia, and Japan. In these various locations, BCG was grown in culture and evolved into different strains until 1966 when a seed lot system is implemented to curtail further evolution of the strains. Thus, BCG strains currently existing today share many mutations attributable to the first phase of attenuation that took place in 1908–1921. Further mutations reflect unique changes that occurred in various locations throughout the world.

Characterization of BCG strains have identified differences in genetics and laboratory measures of activity such as immunogenicity and virulence. However, it is currently unknown if different BCG strains exhibit different clinical activity either for use in prevention of tuberculosis or in treatment or prevention of BC. Recent evidence suggests that BCG strains could influence clinical response for patients with BC, but studies fall short of validating this concept. No adequately powered head-to-head trials have been completed to determine the effect of BCG strains on clinical outcomes for patients with BC⁵. A prospective randomized trial with 142 patients found a significantly greater 5-year recurrence-free survival in patients given Connaught compared to TICE BCG (P=0.01)⁶. However, patients were not given maintenance BCG and the superiority of Connaught over TICE in patients treated with induction only may be mitigated with maintenance BCG⁷. A systemic review and network meta-analysis examined 65 clinical trials and over 12 thousand patients⁵. This analysis found that BCG strains exhibit significant differences in efficacy compared to chemotherapy, with Tokyo-172 out-performing other strains, but was unable to conclude superiority of any particular BCG strain due to lack of adequate comparative studies. A randomized trial⁸ of 129 patients showed similar recurrence-free survival rates between Connaught and Tokyo-172 strains. This trial was underpowered and did not meet accrual because it was interrupted due to manufacturing shortage and lack of Connaught product. Since then, Sanofi-Pasteur has ceased all production of Connaught BCG and patients in the U.S. are currently dependent on one manufacturer (Merck) for BCG (TICE). S1602 trial⁹ is a subsequent phase III study to examine Tokyo-172 in 969 patients during a 3 year accrual period.

BCG priming

Priming describes the immunologic phenomenon of activation and clonal expansion of a naïve T or B cell upon its initial encounter with an antigen. Vaccination is administration of the antigenic material (vaccine) also termed immunization. To be successful, some vaccines require repeated immunizations after priming and these are termed boosting. Recently, a major development occurred in BCG treatment of BC when it was discovered that priming with subcutaneous BCG could significantly improve the response to intravesical BCG in an animal model of BC¹⁰. In this model, repeated instillations of intravesical BCG were necessary to generate a robust T cell infiltration into the bladder. However, parenteral BCG overcame this requirement. In addition, the authors provided clinical data that response to BCG was improved in patients who exhibited BCG immunity prior to receiving BCG, scored by a positive purified protein derivative (PPD) test¹⁰. This work led to the development of PRIME¹¹, a prospective clinical trial to test the value of priming prior to intravesical BCG administration.

If BCG priming could enhance response to intravesical BCG, could administration of PPD, which has similar antigens as BCG, also improve response to BCG? PPD is cell-free purified protein fraction obtained from Mycobacterium tuberculosis. The mycobacterium was initially isolated from an infected patient but subsequently inactivated and grown on a protein-free synthetic medium. Administration of PPD is not usually adequate to elicit an immunity to mycobacterium since an effective immune response requires recognition of an infection, termed a “danger” signal. Nevertheless, PPD shares many antigenic proteins with BCG and induction of BCG-specific immunity has been described following PPD testing.

In Japan where people are routinely immunized with BCG at birth, persistence of PPD positivity is variable among patients presenting with BC. This is, in part, because immunization status wanes with time. To determine if pretreatment BCG-specificity predicted response to BCG therapy, Niwa and colleauges¹² at the Keio University School of Medicine in Tokyo examined patients subjected to PPD skin testing prior to intravesical BCG therapy. The strength of the PPD skin reactions were categorized as positive, slightly positive, or negative. The recurrence-free survival rate was significantly higher for patients who were positive (89%) compared to those patients who were slightly positive (66%), who were also higher than patients with a negative PPD test (56%). The authors concluded that PPD skin testing could be used before BCG therapy to predict clinical outcomes. This same group also reported that PPD placement prior to BCG administration could be therapuetic¹³. They identified n=320 patients who underwent PPD testing 1–2 weeks prior to BCG instillation for BC and compared outcomes to n=178 patients who did not have PPD testing prior. Those patients that received PPD testing experienced a higher 5-year recurrence-free survival rate (67%) compared to those patients that did not receive a PPD test (59%). Interestingly, the product insert for TICE BCG recommends PPD testing prior to instillation of intravesical BCG¹⁴, but this has not been standard practice for patients in the U.S. Given that PPD is inexpensive and safe, this approach needs to be considered.

BCG in combination with tumor vaccine

Tumor vaccines utilizes tumor-specific proteins to improve antitumor immune responses. As monotherapy, tumor vaccine strategies, generally, have poor efficacy. However, combination strategies could be beneficial by providing complementary facets of the immune response. Recently, a tumor vaccine strategy was tested in a phase I exploratory study¹⁵. Patients (n=24) with NMIBC. The vaccine consisted of recombinant MAGE-A3 protein formulated in the AS15 adjuvant (also referred to as MAGE-A3 immunotherapeutic) which includes a TLR4 (monophosphoryl lipid A) and a TLR9 (CpG) agonists as well as saponin (QS-21).. This MAGE-A3 cancer/testis antigen is expressed in ~ 40% of NMIBC and up to 60% of MIBC. The study demonstrated safety of combining a tumor vaccine with BCG and reported induction of circulating vaccine-specific T cells in half of the patients.

BCG related adverse events

BCG is commonly prescribed and each patient received multiple instillations. Thus, the number of serious adverse events per BCG instillation is low. Nevertheless, serious adverse events do occur. A contemporary list of BCG complications and treatments reported over the last 24 months are found in Table 1. A comparative review of AEs associated BCG used for preventative tuberculosis versus therapeutic use for BC is noteworthy¹⁶. While generally, AEs were different between percutaneous/intradermal versus intravesical BCG, they identified AEs common to both BCG uses including chills, pneumonia, and an increase in C-reactive protein. Potential explanations for the differential AEs include routes of administration, age of patients and administration of BCG to healthy, in the case of Tb, versus diseased, in the case of BC, tissues.

Table 1:

Complications of BCG therapy and their treatments.

Complication	Signs/symptoms	Time	Lab testing	Treatment
Reiter’s syndrome34	Urethral discharge, conjunctivis, low back pain	4 weeks	systematic clinical examination, no single confirmatory laboratory or radiological test	Withhold BCG NSAIDs, eye ointment
Parotid gland infection35	Bilateral swelling of the parotid glands, parotid gland ulcer with continuous pus discharge	10 months	GenoType Mycobacterium, scintigraphy	Isoniazid + Vitamin B6, ethambutol, levofloxacin
Ateriocutaneous fistula36	Rapidly expanding mass	2 years	Culture, PCR, HAIN test	isoniazid, ethambutol, rifampin
Pleural effusion	Dyspnea, pleuritic chest pain	1 year	Culture	RIE
Disseminated BCG37	Fever, night sweats, weakness	2 years	Zn stain, PCR,	RIE
Psoas abscess, iliac artery rupture38	Fever, night sweats, abdominal pain	10 months	PCR, culture	Endovascular stent, RI
Poncet’s disease39	Diffuse arthritis	8 months	polymerase chain reaction revealed the presence of A24/AX, B44, B27, BW4/BW4, DQ7 and DQ5	etoricoxib and isoniazid for 4 months

Predictors of response to BCG

Identifying patients that will respond to BCG treatment is critical as use of this therapy in non-responders can delay receipt of other therapeutic treatments, potentiating progression of disease. Thus, the identification of measurable predictors of response to BCG is necessary to select for treatment of appropriate patients and optimize clinical responses. To identify novel markers, a study screening for surveillance biomarkers from Memorial Sloan Kettering¹⁷ used next generation DNA sequencing of 105 index tumors from patients with NMIBC. Despite the extensive amount of data generated and multiple mutations unique to NMIBC over MIBC, only mutations in ARID1A (a chromatin modifying gene) were associated with tumor recurrence in the cohort treated with BCG (n=62; HR = 3.14, 1.51–6.51; p=0.002). The authors note that several mutations, including ARID1A, identified in their cohort may be exploited as therapeutic targets.

In a retrospective study, Husek et.al. used primary TURB tissue samples from 66 patients who subsequently received BCG therapy, of which 45 (68%) were responders and 21 (32%) were non-responders¹⁸. DNA isolated from formalin-fixed tissue was then queried for aberrant methylation of tumor suppressor genes. The authors found two unique genes, CDKN2B (cyclin-dependent kinase inhibitor 2A) and MUS81a (MUS81 endonuclease homolog) whose unmethylated status correlated with BCG failure patients. While the cohort is small, this paper highlights that detection of other epigenetic changes may help to predict response to BCG.

In a similar retrospective cohort analysis, tissue from the Nordic T1 and BCG-MMC trials run in Sweden were used to validate three previously identified predictors of BCG and mitomycin-C responsiveness (MMC): ezrin, CK20, and Ki-67¹⁹. Tissue from patients treated with BCG and/or MMC were used to generate tissue microarrays and immunohistochemistry used for analysis of protein expression. Of the potential markers assessed, ezrin, a scaffold protein that helps organize cell polarity, correlated positively with progression free survival and treatment failure free survival in the BCG treatment arm. On multivariable analysis, however, ezrin significantly predicted treatment failure only in the entire cohort (HR 2.287, 1.125–4.648; p=0.022) and was not significant in the BCG treatment arm (HR 0.91, 0.369–2.245; p=0.838). Multifocality of tumor was the only predictor of progression after BCG in the multivariate model, consistent with AUA risk-stratification¹⁷. The authors note that the negative results of the study show the difficulty in finding biomarkers that predict clinical effectiveness.

A prospective study to assess the accuracy of fluorescence in situ hybridization (FISH) using UroVysion® (Abbot Molecular) technology in predicting response to BCG was performed and funded by the makers of UroVysion® ²⁰. Five centers enrolled patients (n=114) that underwent TURB and subsequent BCG instillation. Bladder washings were taken prior to BCG therapy (n=114), 6 weeks following the start of BCG induction therapy (n=106), and at the first surveillance cystoscopy (3 months; n=66). A positive FISH analysis at 3 months was predictive of tumor recurrence on corrected Cox regression analysis (HR 4.0, 1.45–11.10) and had sensitivity and specificity of 59% and 84% respectively. Although helpful in assessing response post-BCG induction, this test does not help predict response or lack thereof prior to any BCG therapy.

Together these studies show the effort to find markers of clinical response to BCG therapy. Markers may be urine-based or tissue-based with gene mutations, epigenetic changes, and whole protein expression as potential targets.

Mechanism of action of BCG therapy

Despite the use of BCG in the clinic for a significantly long time, the mechanism behind its therapeutic efficacy still lacks clarity. Urothelial cells and cells of the immune system including CD4⁺ and CD8⁺ T cells, natural killer cells, granulocytes, macrophages, and dendritic cells have been previously shown to be critically involved in the therapeutic effect of BCG. Evidence show that attachment and internalization of BCG results in the secretion of cytokines and presentation of BCG and/or cancer cell antigens to the immune cells for the activation of the tumor protective immune response. In 2017 Kates et al. characterized the T-cell subpopulations post BCG therapy²¹. They found that even though BCG therapy increased CD4⁺ T-cell population in the urothelium, no differences were found in gene expression in the intravesical CD4⁺ and CD8⁺ T cells²¹. These suggest that BCG recruits T cells but does not necessarily affect the T-cell phenotype opening the avenue for using T-cell-activating agents with BCG therapy in BC. Further, it was also shown that even through Th1-type immune responses are mostly involved in BCG efficacy, Th2-promoting factors such as GATA3 may induce the Th1-type immune responses resulting in an effective BCG response²². The study showed that there is a Th2 predisposition of tumor-infiltrating immune cells in patients who respond to BCG and an increase in the number of GATA3⁺ immune cells was observed in these patients relative to patients who fail BCG therapy. Further, increased pretherapeutic GATA3/T-bet ratio was observed in responders relative to non-responders and smokers with low GATA3 expression did not respond favorably to BCG. It is well established that changing a Th2 immune landscape towards a Th1 immune response during BCG therapy is required for an efficient BCG response²³. However, this current study focused on an alternative mechanism that an adequate Th2 capacity is required for initiating an effective Th1 response during BCG therapy²². Derré et al. also showed the increase in vaccine-specific T cells in the bladder when BCG is combined with subunit cancer vaccine (recMAGE-A3 protein+AS15)¹⁵. These suggested that cancer vaccines with local immunostimulation can be useful in BC therapy. This is especially relevant because despite the long clinical experience with BCG, it is still plagued with issues of non-response and progression of cancer²⁴. Generation of tumor protective T cells by cancer vaccines seems to hold combinatorial therapeutic potential with BCG.

In addition to T cells, David et al. recently hypothesized that granulocytes can also be one of the underrated critical effector cells for the BCG therapy²⁵ which may have been overlooked due to the huge emphasis on delineating the role of T cells in BCG response and lower number of available assays. He showed that granulocytes can distinguish cancer cells from normal cells by detecting the difference in surface charges and target these tumor cells by the process of degranulation. This idea is supported by evidence showing that there is an infiltration of granulocytes into the bladder wall and bladder cavity immediately after BCG instillation^26,27. BCG treatment also fails in a murine model depleted of granulocytes²⁸. IL-17⁺ mast cells have also been shown to positively affect outcomes from BCG therapy for patients with CIS suggesting their potential involvement in BCG therapy in CIS patients²⁹. This can be useful as a potential novel therapeutic strategy focused on increasing the beneficial effects of IL-17⁺ mast cells during BCG therapy in CIS patients. Lastly, another member of innate immune cells family which have been shown to modulate BCG response is the NK cells. BCG exposure resulted in the activation and expansion of a novel population of CD56^bright NK cells which are traditionally cytokine producers and have low cytotoxicity. However, these BCG activated CD56^bright population became specialized to have high cytotoxicity by maintaining the expression of receptors such as CD16 and KIR³⁰. This novel CD56^bright cell population degranulated against bladder tumor cells. Further, CD16 expression was significantly expressed in peripheral blood CD56^bright cells derived from BCG treated patients compared to patients on Mitomycin C .Proliferation and degranulation of this unconventional population of CD56^bright NK cells post BCG therapy suggested that activation of NK cells can also be involved in an effective BCG response in BC patients³⁰

Recombinant BCG

How could elucidation of BCG’s mechanism of action contribute to improving outcomes for patients? Mechanistic studies demonstrated that BCG requires interleukin (IL)-17A produced by γδ T cells to recruit antitumor neutrophlis to the bladder after BCG instillation³¹. IL-15 is known to play an important role in neutrophil migration during inflammation. Working on this knowledge, researchers developed a recombinant BCG strain expressing the fusion protein of IL-15 and BCG’s immunodominant antigen (Ag85B), termed BCG-IL-15³². Using a mouse model of BC to study this recombinant BCG, they found that BCG-IL-15 treatment significantly increased the survival of mice challenged with tumor compared to standard BCG³².

Interferon alpha 2b is a critical antitumor cytokine and has been utilized for treatment of malignancy, including BC. In theory, a recombinant BCG strain expressing IFNα−2b could be more effective than BCG. Using a recombinant plasmid expressin hIFNα−2b, researchers transformed BCG and studied its activity in vitro³³. They demonstrated that this recombinant BCG could stimulate lymphocytes and enhance cytotoxicity against BC cells. Further studies are planned for the development of this product for human testing.

Conclusion

Despite entering the fifth decade of BCG therapy for management of non-muscle invasive BC, novel insights into the various aspects of this immunotherapeutic strategy continue to emerge. Recent developments include methods to improve BCG’s efficacy either through improving the immune response to BCG, combining BCG with tumor vaccines or transforming BCG to express novel proteins through recombination techniques. Further work in prognostication and prediction of BCG efficacy holds promise to enable clinicians to identify patients most likely to benefit or to steer patients towards alternative modalities based on biomarker profiles. Despite these important developments, gaps in the understanding of biology of BCG and its effect on the bladder remain. Future research to understand BCG’s precise mechanisms of activity are needed to move the field forward and improve patient outcomes.

Key points:

• Strategies to improve the clinical response to BCG through vaccination with live organisms or BCG protein are ongoing.

• Some BCG strains could be more effective than others, but further research is required.

• Important publications came out in 2016–2018 delineating a role for T cells, NK cells, mast cells and granulocytes in BCG’s antitumor efficacy.

• Further, novel urine and tissue markers, including epigenetic proteins show promise in the use of biomarkers for BCG treatment selection.