Molecular profiling is promising, but BCG selection still rests on clinical risk alone
Intravesical BCG remains the backbone of adjuvant therapy for intermediate- and high-risk non-muscle-invasive bladder cancer, yet response is far from uniform. A substantial fraction of patients either fail to respond or recur within two years, and current clinical and pathologic models do not reliably identify who will benefit most from treatment. The review pulls together a growing body of molecular data and makes one point plainly: the field has several biologically plausible candidate markers, but none has earned a validated place in treatment selection.
That gap matters because BCG is not a marginal intervention. It is the standard bladder-preserving immunotherapy for a disease in which recurrence and progression can alter both surveillance burden and the chance of later invasive spread. When a therapy is both effective and imperfect, precision selection becomes attractive; the challenge is separating signal from cohort-specific noise.
Transcriptomic classes define biologic behavior, not yet treatment decisions
The strongest molecular signals come from transcriptomic classifications, including UROMOL2021, BCG Response Subtypes (BRS), MSP888, and subtype systems developed in stage T1 disease. These frameworks sort tumors into biologic phenotypes that correlate with recurrence, progression, and outcomes after BCG exposure. In practical terms, they suggest that some tumors are intrinsically more immunologically receptive, while others carry programs that may blunt response to intravesical immune activation.
Among these systems, BRS3 stands out because it aligns with basal-squamous and immunosuppressive biology and is linked to shorter recurrence-free and progression-free survival. The mechanism is intuitive. BCG depends on an inflamed local immune response to clear residual disease, so a tumor microenvironment already biased toward immune suppression may be less likely to mount the effective antitumor cascade that therapy is trying to provoke.
A molecular subtype can describe risk without yet qualifying as a companion diagnostic.
Genomic and epigenetic markers have repeatedly appeared, but validation remains the barrier
Several genomic alterations recur across studies. FGFR3 and TERT promoter mutations, along with APOBEC signatures, have been associated with outcomes in individual cohorts. DNA methylation markers such as GATA2, TBX3, and GPR158 have also emerged as candidate predictors. These findings are biologically credible because they reflect pathways that shape tumor growth, differentiation, genomic instability, and transcriptional control, all of which may influence how a bladder tumor interacts with an intravesical immune stimulus.
Even so, association is not the same as treatment specificity. A marker can correlate with recurrence and still fail to predict whether the benefit of BCG itself differs from the baseline prognosis of the disease. That distinction is central. Without prospective biomarker-stratified studies, it remains impossible to know whether these alterations identify BCG sensitivity, BCG resistance, or simply a more aggressive tumor class that does poorly regardless of therapy.
The tumor microenvironment may be the most mechanistically relevant layer
The review places particular weight on the tumor microenvironment, where regulatory T cells, exhausted CD8+ T cells, and PD-L1 expression have all been associated with outcomes in selected cohorts. This is the part of the biology that most directly connects to BCG’s mechanism of action. BCG is intended to recruit and amplify local immunity inside the bladder wall, and the presence of suppressive immune populations could plausibly dampen that effect before tumor eradication is achieved.
Exhausted cytotoxic T cells and PD-L1 expression point to immune evasion, a state in which the host response is present but functionally blunted. In that setting, more immune stimulation is not automatically better. If the inhibitory circuitry dominates, the bladder may accumulate inflammatory cells without generating the tumoricidal activity that translates into durable remission.
BCG-unresponsive disease has expanded bladder-preserving options, but not biomarker-guided selection
For patients with BCG-unresponsive disease, bladder-preserving options now include pembrolizumab, nadofaragene firadenovec, nogapendekin alfa inbakicept-pmln with BCG, and the gemcitabine intravesical system. These therapies widen the menu for patients who are no longer good candidates for continued BCG alone, but they do not solve the core biomarker problem. None is being selected today on the basis of a validated molecular classifier that can predict who will respond to one strategy over another.
Randomized trials that have added checkpoint inhibition to BCG in BCG-naive disease have produced inconsistent results. That inconsistency is clinically important because it suggests the immune contexture of the tumor and host may govern benefit, yet the field still lacks the standardized biomarker framework needed to identify who should receive an intensified immunotherapy approach up front.
Clinical use remains premature without standardized assays and prospective trials
The review’s final message is conservative for good reason. No molecular biomarker is currently validated for treatment selection in non-muscle-invasive bladder cancer. Molecular profiling may eventually complement established clinical risk assessment, but it cannot replace it now. The available studies differ in definitions, assay methods, endpoints, and patient populations, which makes cross-trial comparison difficult and weakens the case for routine adoption.
Prospective biomarker-stratified trials will be necessary if these candidates are to move beyond hypothesis-generating work. Those studies will need standardized definitions of BCG response, uniform assay platforms, and clinically meaningful endpoints such as recurrence, progression, and bladder preservation. Until then, the most defensible use of molecular data is as a research tool and a refinement of risk discussion, not as a basis for withholding or assigning BCG in daily practice.
For now, the biology is moving faster than the clinic. The field has identified plausible markers, but it has not yet proven that any of them can steer treatment better than established risk models at the bedside.