Our group at Moffitt Cancer Center recently published a large retrospective study characterizing the BCG-E phenotype using contemporary definitions.5 Among 470 evaluable BCG-treated patients with high-grade recurrence, 245 (52.1%) met BCG-E criteria and 225 (47.9%) met BCG-UR criteria. In the BCG-E cohort, 50.2% experienced subsequent high-grade recurrence and 15.5% progressed to muscle-invasive or metastatic disease; median recurrence-free survival (RFS) was 27.2 months, while median progression-free, metastasis-free, and overall survival were not reached.
A majority of patients (86.9%) received bladder-sparing therapy following BCG-E diagnosis, with salvage intravesical BCG the most common salvage modality (62%), followed by gemcitabine plus docetaxel (18.4%). No consistent survival differences were observed across salvage treatment regimens for metastasis-free survival (MFS) or overall survival (OS). Radical cystectomy was performed in 45 patients (18.4%), with upstaging to muscle-invasive disease found in 26.1%, lower than the 40–50% previously reported.6–8 Perhaps the most surprising finding was the head-to-head comparison between BCG-E and BCG-UR groups: no statistically significant differences were identified for RFS (p = 0.9), PFS (p = 0.5), MFS (p = 0.9), or OS (p = 0.5).
This last observation deserves careful attention. The BCG-UR definition retains clear therapeutic value: it identifies patients unlikely to benefit from additional BCG, for whom alternative salvage strategies or radical cystectomy should be prioritized. However, the absence of meaningful prognostic separation between BCG-E and BCG-UR populations challenges the assumption that these categories reflect fundamentally different disease biology. A patient who recurs with a high-grade Ta/T1 lesion at 5 months after adequate BCG is labeled BCG-UR; a patient who recurs at 7 months is BCG-E. While the therapeutic implications of that distinction remain relevant, the data from this study suggest that the underlying risk of recurrence, progression, and death is remarkably similar across both groups — raising the question of whether these time-based cutoffs capture meaningful biological differences or simply procedural ones.
The current classification system functions as a crude sieve. It sorts patients into administrative buckets based on timing and BCG dose thresholds, not on any understanding of why a given tumor failed to respond. We know that BCG exerts its antitumor effects through a complex interplay of innate and adaptive immunity—trained immunity, T-cell activation, cytokine cascades—and that failure likely involves a heterogeneous mix of mechanisms, including T-cell exhaustion,9 immunosuppressive microenvironment remodeling, tumor-intrinsic immune evasion, and perhaps simple pharmacokinetic inadequacy of drug exposure. Yet our classification framework captures none of this biology.
What we can confidently state from the cumulative evidence—reinforced by this study—is that BCG remains the clear winner in the intravesical therapy space. It was the most commonly used salvage agent even in the BCG-E setting (62% of patients), and its 1-year and 3-year RFS rates of 65% and 50% in this context provide the benchmark that any novel agent must surpass. The recent debates at EAU 2025 and AUA 2025 have reaffirmed BCG’s central role, and the emerging data from CREST,10 KEYNOTE-676,11 and POTOMAC12 only reinforce this by showing that even the most promising systemic additions are built around a BCG backbone. However, with each subsequent recurrence, the benefit of repeated BCG appears to wane. What is happening behind the scenes biologically—why a tumor that initially responded to immune activation eventually escapes—remains poorly understood, and this represents the most important frontier in NMIBC research.
This is where emerging technologies offer genuine promise. Urinary tumor DNA (utDNA) analysis is rapidly maturing from a research curiosity into a potentially transformative clinical tool. Recent work from the SWOG S1605 trial demonstrated that UroAmp-based utDNA profiling could stratify recurrence risk in BCG-unresponsive patients treated with atezolizumab, with only 13% of utDNA-positive patients at baseline achieving complete response compared to 71% of utDNA-negative patients.13 A landmark study published in Cell in early 2026 introduced a field-effect-informed utDNA approach that distinguished surgical responders from BCG responders from non-responders, revealing that molecular predictors of response to surgery and BCG are fundamentally different.14 Notably, the senior author of the present study (Dr. Roger Li) has been at the forefront of this work, presenting at SUO 2025 on tumor-informed utDNA minimal residual disease detection in NMIBC, demonstrating that utDNA signals frequently turn positive before or at the time of visual recurrence detection on cystoscopy.15
The implications are clear: rather than continuing to refine clinical categories based on timing of recurrence and dose counting, the field needs to move towards a biologically grounded classification. Transcriptomic studies have already identified molecular subtypes (BRS1/2/3) that predict BCG response more accurately.16 The integration of utDNA dynamics—clearance patterns, residual disease burden, emergence of new mutations—with tumor molecular profiling could enable a precision approach where treatment decisions are guided by what the tumor is actually doing immunologically, not by how many calendar months have elapsed since the last BCG instillation.
This study provides the pragmatic benchmarks that the field needs for trial design and patient counseling. But it also, perhaps inadvertently, makes the strongest case yet for moving beyond phenotype-based categorization entirely. The next generation of clinical trials in this space should be designed not just to test new agents against BCG, but to integrate correlative biomarker studies—utDNA, tumor transcriptomics, immune microenvironment profiling—that will allow us to finally understand why BCG works when it does, why it fails when it doesn’t, and how to rationally sequence or combine therapies based on biology rather than bureaucratic definitions.
Written by:
- Renzo G. Di Natale, MD, MSc, Department of Genitourinary Oncology, H. Lee Moffitt Cancer Center, Tampa, FL
- Roger Li, MD, Department of Genitourinary Oncology, H. Lee Moffitt Cancer Center, Tampa, FL; Department of Immunology, H. Lee Moffitt Cancer Center, Tampa, FL
- Matulewicz RS, Steinberg GD. Non-muscle-invasive bladder cancer: overview and contemporary treatment landscape of neoadjuvant chemoablative therapies. Rev Urol. 2020;22(2):43–51.
- Lerner SP, Dinney C, Kamat A, et al. Clarification of bladder cancer disease states following treatment of patients with intravesical BCG. Bladder Cancer. 2015;1(1):29–30.
- US Food and Drug Administration. BCG-unresponsive non-muscle-invasive bladder cancer: developing drugs and biologics for treatment. Guidance for industry. Silver Spring, MD: FDA; 2018.
- Roumiguié M, Kamat AM, Bivalacqua TJ, et al. International Bladder Cancer Group consensus statement on clinical trial design for patients with bacillus Calmette-Guérin-exposed high-risk non–muscle-invasive bladder cancer. Eur Urol. 2022;82(1):34–46.
- Jazayeri SB, DiNatale RG, Guske C, et al. Bacillus Calmette-Guérin–exposed non–muscle-invasive bladder cancer: survival benchmarks, bladder-sparing strategies, and implications for trial design. Eur Urol Focus. 2025. doi:10.1016/j.euf.2025.12.013.
- Soloway MS, Hepps D, Katkoori D, Ayyathurai R, Manoharan M. Radical cystectomy for BCG failure: has the timing improved in recent years? BJU Int. 2011;108(2):182–185.
- Guzzo TJ, Magheli A, Bivalacqua TJ, et al. Pathological upstaging during radical cystectomy is associated with worse recurrence-free survival in patients with bacillus Calmette-Guérin-refractory bladder cancer. Urology. 2009;74(6):1276–1280.
- Nieder AM, Simon MA, Kim SS, Manoharan M, Soloway MS. Radical cystectomy after bacillus Calmette-Guérin for high-risk Ta, T1, and carcinoma in situ: defining the risk of initial bladder preservation. Urology. 2006;67(4):737–741.
- Strandgaard T, Lindskrog SV, Nordentoft I, et al. Elevated T-cell exhaustion and urinary tumor DNA levels are associated with bacillus Calmette-Guérin failure in patients with non-muscle-invasive bladder cancer. Eur Urol. 2022;82(6):646–656.
- Shore ND, Powles TB, Bedke J, et al. Sasanlimab plus BCG in BCG-naive, high-risk non-muscle invasive bladder cancer: the randomized phase 3 CREST trial. Nat Med. 2025;31(8):2806–2814.
- Kamat AM, Shore N, Hahn N, et al. KEYNOTE-676: phase III study of BCG and pembrolizumab for persistent/recurrent high-risk NMIBC. Future Oncol. 2020;16(10):507–516.
- De Santis M, Palou Redorta J, Nishiyama H, et al. Durvalumab in combination with BCG for BCG-naive, high-risk, non-muscle-invasive bladder cancer (POTOMAC): final analysis of a randomised, open-label, phase 3 trial. Lancet. 2025;406(10516):2221–2234.
- St-Laurent MP, Singh P, McConkey DJ, et al. Urine tumor DNA to stratify the risk of recurrence in patients treated with atezolizumab for bacillus Calmette-Guérin–unresponsive non–muscle-invasive bladder cancer. Eur Urol. 2025;88(5):430–434.
- Shi WY, Liu KJ, Esfahani MS, et al. Field-effect-informed urine liquid biopsy for bladder cancer. Cell. 2026. doi:10.1016/j.cell.2025.12.054. [Epub ahead of print].
- Li R. MRD detection in NMIBC: the role of utDNA and clinical implications. Presented at: 2025 Society of Urologic Oncology Annual Meeting; December 2–5, 2025; Phoenix, AZ.
- de Jong FC, Laajala TD, Hoedemaeker RF, et al. Non–muscle-invasive bladder cancer molecular subtypes predict differential response to intravesical bacillus Calmette-Guérin. Sci Transl Med. 2023;15(697):eabn4118.