The review, published on 27 October 2026, distinguishes carefully between three concepts that are often conflated in the biomarker literature. Prognostic biomarkers are associated with outcomes regardless of treatment. Treatment-response-associated biomarkers correlate with response within a treated cohort but do not by themselves prove treatment-specific benefit. Predictive biomarkers, in the strict sense, require comparative evidence demonstrating a biomarker-by-treatment interaction. Because most available studies are single-arm or treatment-specific cohorts, the authors conclude that the vast majority of reported immune features in bladder cancer remain treatment-response-associated or prognostic patterns rather than validated predictive tools.

A recurring theme is that apparently similar levels of immune infiltration can carry opposite clinical meanings. CD8+ T cells are the principal cytotoxic effectors of the antitumor response, but their density alone does not establish cytotoxic activity. Studies of muscle-invasive and compartment-specific tumor-infiltrating lymphocytes have found CD8+ cells in both the tumor stroma and tumor nests, with non-uniform distribution across compartments in some cohorts. A tumor may harbor abundant CD8+ T cells yet remain clinically resistant when those cells are excluded from tumor nests or accompanied by exhaustion-associated marker patterns. Conversely, PD-1 or PD-L1 expression may reflect exhaustion-related or adaptive immune-resistance phenotypes, but marker expression alone does not demonstrate functional exhaustion.

Compartmentalization extends across the full immune repertoire. CD4+ T cells include Th1, Th2, memory, cytotoxic, regulatory, and CXCL13-expressing subsets with divergent effects. One study found CXCL13-expressing and cytotoxic CD4+ subsets enriched within tumor tissue, while memory and activated CD4+ cells were more common in peritumoral regions; another reported a Th2-predominant intratumoral phenotype with a high GATA3-to-T-bet ratio and compartment-specific enrichment at the invasive front. Regulatory T cells, which suppress immunity through IL-10, TGF-beta, CTLA-4, and PD-L1-related mechanisms, preferentially localize to the tumor stroma or peritumoral region, potentially forming a physical and functional barrier between effector lymphocytes and tumor cells. Natural killer cells show heterogeneous evidence, with CD56bright NK cells associated with improved survival in some tissues and single-cell data suggesting higher NK infiltration in muscle-invasive disease than in non-muscle-invasive tumors or adjacent normal tissue.

Less abundant populations add further layers. Innate lymphoid cells with ILC1-like features and mixed inflammatory and Th17-like programs have been detected within bladder tumors. Gamma-delta T cells, including Vdelta2 cells, are present in tissue and blood and can recognize bladder tumor cells, with reactivity potentially enhanced by BCG or zoledronate in experimental settings. Th17 cells appear at increased frequencies in tumor tissue, T follicular helper cells have been characterized in draining lymph nodes, and single-cell profiling has identified NKT cells with exhaustion-associated transcriptional states. The B-cell compartment is particularly context dependent: atypical B cells and regulatory B cells, including IL-10-producing and TGF-beta-producing phenotypes, have been linked to progression and poor BCG response, whereas other B-cell populations participate in antigen presentation and organized tertiary lymphoid structures containing B-cell follicles surrounded by T cells. Early and secondary follicle-like tertiary lymphoid structures show different distributions across tissue compartments, underscoring that B-cell aggregates cannot be classified as uniformly beneficial or suppressive.

The myeloid compartment offers the clearest example of compartmentalized phenotypes. Tumor-associated macrophages are often described with an M1-like versus M2-like framework, but the review stresses that this terminology is descriptive shorthand for a plastic continuum rather than fixed states. M1-like phenotypes are associated with inflammatory antitumor programs, while M2-like phenotypes are associated with immune suppression, angiogenesis, matrix remodeling, and invasion. Markers such as CD163 and CD204 are frequently used, but their expression alone does not demonstrate immunosuppressive activity. Single-cell transcriptomic profiling has identified heterogeneous macrophage populations, including SPP1+, FOLR2+, MKI67+, C1QC+, and LY6D+ states, with SPP1+, FOLR2+, and C1QC+ populations showing relatively high M2-signature scores. Notably, SPP1+ macrophages shifted from a phagocytic phenotype in non-muscle-invasive disease toward an angiogenesis-associated phenotype in muscle-invasive tumors. Histologically, macrophages are enriched in the tumor stroma and invasive fronts rather than uniformly within tumor nests, a pattern that may support immune exclusion and invasion even when lymphocytes are detectable elsewhere. Myeloid-derived suppressor cells add another suppressive layer by inhibiting CD8+, CD4+, and NK-cell function, though their identification in human tissue remains methodologically challenging because of phenotypic overlap with conventional neutrophils and monocytes.

Disease-state associations are uneven across cell types. The most consistent findings link macrophage enrichment, particularly M2-like or CD204- and CD163-positive patterns, with higher stage, higher grade, or invasive behavior, though these observations come mainly from mixed-stage cohorts. Regulatory B cells were increased in high-grade tumors in one study, and FATP2-high, RIPK3-low myeloid-derived suppressor cell patterns were reported in stage III/IV versus stage I/II disease. CD8+ and Treg findings, by contrast, are context dependent: some studies found no clear association between CD8+ density and stage, and Treg data vary between cohorts. The authors emphasize that these cross-sectional associations do not establish that immune changes causally drive progression, and that differences in antibody markers, scoring methods, and cohort composition limit comparisons.

On treatment-associated remodeling, the strength of evidence varies by modality. BCG immunotherapy has relatively stronger human support. In one profiling study, BCG responders showed increased CD4+FOXP3-negative and CD8+PD-1-negative T cells, whereas non-responders showed increased CD8+PD-1-positive cells characterized as exhausted, along with increased CD8+ numbers without corresponding CD69 activation. Paired specimens from BCG-resistant tumors showed increased PD-L1 expression after treatment, and sequential tumor analyses found BCG exposure associated with PD-L1 changes in tumor and inflammatory cells. For chemotherapy, paired studies provide direct evidence: analysis of 54 paired muscle-invasive specimens before and after neoadjuvant chemotherapy found significant decreases in CD68-positive and CD68-positive PD-L1-negative macrophages, while CD68-positive PD-L1-positive macrophages and evaluated T-cell populations showed no significant paired changes. In a broader platinum-treated cohort of 95 paired samples, chemotherapy was associated with increased intratumoral CD8+ proportions and PD-L1-positive immune cells, alongside increased fibroblast TGF-beta signaling and greater spatial separation between immune and cancer cells.

Evidence for immune checkpoint inhibitors and radiotherapy is more heterogeneous. A comparison of untreated tumors with tumors exposed to preoperative anti-PD-1 and anti-CTLA-4 therapy found increased CD8+PD-1-positive T cells and altered tertiary lymphoid structure architecture, but the unpaired design cannot exclude between-patient differences. A preliminary conference report from the neoadjuvant pembrolizumab PURE-01 setting suggested that even non-responders may show increased CD8+ and CD20+ densities after treatment, indicating that post-treatment immune-cell increases should not be read as evidence of effective antitumor response. For radiotherapy, paired human samples are limited; preclinical models suggest irradiation can enhance CD8+ activity while inducing compensatory PD-L1 upregulation and expansion of suppressive myeloid cells and Tregs, and clinical evidence derives largely from concurrent chemoradiotherapy rather than radiotherapy alone.

For baseline biomarkers, the review finds that unfavorable associations across BCG-treated cohorts more often involve regulatory, checkpoint-associated, or exhaustion-related patterns in specific compartments rather than simply low infiltration. A higher CD8+ density gradient across the tumor-stromal interface was associated with recurrence-free survival after BCG; high pretreatment CD68-positive macrophage and CD83-positive dendritic-cell infiltration was associated with poorer response; and higher peritumoral Treg density was associated with non-response in a carcinoma in situ cohort. In the neoadjuvant chemotherapy setting, a higher pretreatment CD8-to-Treg ratio was associated with response, while non-proliferative Ki-67-negative CD8+ cells combined with CD204-positive macrophage accumulation indicated a suppressive state associated with resistance. For checkpoint inhibitors, baseline PD-L1 positivity was associated with response in single-arm studies including KEYNOTE-052 and PURE-01, but randomized evidence from IMvigor130, in which atezolizumab plus platinum chemotherapy did not achieve a significant overall-survival benefit, illustrates that such associations do not establish treatment-specific prediction. The authors conclude that composite, compartment-resolved immune assessment is more informative than any single marker, that most TIME-based biomarkers remain investigational, and that future progress requires prospective longitudinal studies with matched pre- and post-treatment specimens and formal biomarker-by-treatment interaction testing.

Subject of Research: The tumor immune microenvironment of bladder cancer, including immune-cell localization, disease-state associations, treatment-induced remodeling, and baseline biomarkers of response and prognosis

Article Title: Bladder cancer immune microenvironment: immune-cell distribution, disease associations, treatment remodeling, and baseline immune features associated with treatment response and prognosis

Article References: Qin, S., Yu, X., Zuo, S., He, X., & Liu, J. (2026). Bladder cancer immune microenvironment: immune-cell distribution, disease associations, treatment remodeling, and baseline immune features associated with treatment response and prognosis. Frontiers in Oncology, 16, 1985263. https://doi.org/10.3389/fonc.2026.1985263

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1985263

Keywords: bladder cancer, tumor immune microenvironment, BCG immunotherapy, tumor-associated macrophages, CD8+ T cells, regulatory T cells, immune checkpoint inhibitors, neoadjuvant chemotherapy, biomarkers, tertiary lymphoid structures, radiotherapy, myeloid-derived suppressor cells