The predictive role of tertiary lymphoid structures in the prognosis and response to immunotherapy of bladder cancer patients: a systematic review and meta-analysis

Abstract

Background

The association between tertiary lymphoid structures (TLS) and clinical outcomes in bladder cancer (BCa) patients, particularly their value in predicting immunotherapy response, remains inconsistent and has not been systematically evaluated.

Objective

This systematic review and meta-analysis aimed to: (1) evaluate the association between TLS and survival outcomes in BCa; (2) examine the relationship between TLS and clinicopathological characteristics; and (3) summarize evidence on TLS and immunotherapy response.

Methods

We systematically searched PubMed, Embase, and Web of Science from inception to February 1, 2026. Hazard ratios (HRs) with 95% confidence intervals (CIs) for overall survival (OS), progression-free survival (PFS), and disease-free survival (DFS) were extracted and pooled. The study protocol was registered with PROSPERO (CRD420261282595).

Results

Eight studies involving 1706 patients were included in the quantitative synthesis. Pooled analysis revealed that high TLS levels were significantly associated with improved OS (HR = 0.49, 95% CI: 0.33–0.74), PFS (HR = 0.51, 95% CI: 0.45–0.58), and DFS (HR = 0.42, 95% CI: 0.22–0.79). TLS showed a significant correlation only with tumor multiplicity among various clinicopathological features. Qualitative synthesis of six studies suggested that while baseline TLS had limited predictive value for immunotherapy response, therapy-induced increases in TLS number, density, and maturation were consistently associated with favorable treatment outcomes.

Conclusion

High TLS levels are associated with favorable survival outcomes in bladder cancer patients, supporting their role as a prognostic biomarker. Therapy-induced TLS dynamics, rather than baseline status alone, may be a more relevant predictive biomarker for immunotherapy efficacy. These findings warrant validation in larger, prospective studies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16326-2.

Introduction

Bladder cancer (BCa) represents a significant global public health concern. According to the GLOBOCAN 2020 estimates, BCa accounted for approximately 573,000 new cases and 213,000 deaths worldwide in 2020. The disease exhibits a pronounced male predominance; incidence and mortality rates are approximately four times higher in men than in women globally. Consequently, bladder cancer ranks as the 6th most commonly diagnosed cancer and the 9th leading cause of cancer death among men [1]. Compared to non–muscle-invasive bladder cancer (NMIBC), muscle-invasive bladder cancer (MIBC) is associated with poor prognosis and a higher risk of progression to metastatic disease [2]. Currently, radical cystectomy combined with neoadjuvant or adjuvant therapy remains the standard treatment for MIBC [2].

With advances in the understanding of the tumor microenvironment (TME) and immune checkpoint blockade (ICB), immunotherapy has emerged as a promising treatment for advanced BCa [3]. However, clinical benefit is limited to a subset of patients [4–7]. The IMvigor010 trial showed no survival benefit of adjuvant atezolizumab in the overall population [8], and in the ABACUS trial, a considerable proportion of patients did not respond to neoadjuvant atezolizumab. These findings highlight the urgent need for reliable biomarkers to identify patients who may benefit from immunotherapy.

Increasing evidence suggests that tertiary lymphoid structures (TLS) presence correlates with improved clinical outcomes and enhanced responses to immunotherapy across various cancer types [9–13]. Primarily consisting of dendritic cells, B cells, T cells, and specialized blood vessels [14], TLS are considered as ectopic lymphoid aggregates that develop in areas of chronic inflammation, including tumors. It’s reported that TLS can facilitate immune cell recruitment and local immune activation within the TME [15]. However, studies on TLS in BCa remain limited and inconsistent, particularly regarding their prognostic value and relationship with immunotherapy response [16, 17].

Therefore, to consolidate the existing evidence, we performed a systematic review and meta-analysis with the following objectives: (1) evaluate the association between TLS and survival outcomes in BCa; (2) examine the relationship between TLS and clinicopathological characteristics; and (3) summarize current evidence on the interaction between TLS and immunotherapy in BCa. (Fig. 1).

Fig. 1.

Visualization of TLS by HE staining (Left) and spatial transcriptomics (Right) (18). The spatial transcriptomics map illustrates the calculated TLS signature score across the tissue region, highlighting areas with high enrichment of TLS-associated genes (shown in red/orange)

Method

Protocol and registration

This systematic review and meta-analysis was conducted in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. The study protocol was prospectively registered on the PROSPERO database (registration number CRD420261282595). Since the analysis exclusively utilized data from previously published studies, neither ethical approval nor patient consent was required.

Search strategy

A systematic search of three electronic databases (PubMed, Embase, and Web of Science) was performed, encompassing literature from their inception until February 1, 2026. According to the core research themes, the keywords used for searching were closely related to “tertiary lymphoid structure” (including “tertiary lymphoid organ”, “ectopic lymphoid structure”), “bladder cancer” (including “urothelial carcinoma”, “urothelial cancer”, “transitional cell carcinoma”, “urinary bladder neoplasm”), immune-related terms (e.g., “lymphocyte”, “tumor-infiltrating lymphocyte”, “T cell”, “B cell”, “tumor immune microenvironment”), and “prognosis” (including “survival”, “overall survival”, “progression-free survival”, and “recurrence”). The initial identification of records was based on screening titles and abstracts. Potentially relevant articles subsequently underwent full-text review. To supplement the database search, the reference lists of all included studies were manually reviewed to identify any additional relevant publications that might have been overlooked.

Inclusion and exclusion criteria

Studies were considered eligible if they met all the following criteria: [1] Prospective or retrospective cohort studies and clinical trials that reported original research data; [2] Studies focused on human patients with histologically confirmed bladder cancer, including but not limited to NMIBC, MIBC, and upper tract urothelial carcinoma (UTUC), irrespective of treatment received (e.g., surgery, chemotherapy, radiotherapy, or immunotherapy); [3] Studies that assessed TLS using standard pathological methods, such as hematoxylin and eosin (HE) staining, multiplex immunofluorescence (mIF), immunohistochemistry (IHC), or spatial transcriptomics, with a clear description of the assessment criteria (e.g., presence, count, density, or maturity); [4] Studies that reported at least one of the following with extractable data: survival outcomes including overall survival (OS), progression-free survival (PFS), or disease-free survival (DFS), provided with hazard ratios (HRs) and 95% confidence intervals (CIs).

Studies were excluded based on the following criteria: [1] Case reports, reviews, meta-analyses, conference abstracts, editorials, and studies based on animal or in vitro models; [2] Studies that lacked sufficient quantitative data for meta-analysis or qualitative data for synthesis regarding clinical or immunotherapeutic relevance; [3] Studies with suspected or confirmed overlapping patient populations with another included study; where overlap was suspected, the study with the larger sample size or more comprehensive data was retained; [4] Studies that did not primarily investigate TLS as an organized structure (e.g., those focusing solely on isolated immune cell subsets like specific T-cells or high endothelial venules without assessing the TLS architecture) or that examined TLS exclusively in distal peritumoral areas, excluding the primary tumor site; [5] Studies not published in English.

Three reviewers independently screened the titles and abstracts of all records. The full texts of studies deemed eligible were retrieved for detailed assessment. Any discrepancies were resolved through consensus discussion between the reviewers, with a fourth reviewer consulted to arbitrate if disagreements persisted.

Data extraction

A standardized form was created for data extraction. Two independent reviewers collected the following data from each eligible study: first author’s name, publication year, country, cancer type, sample size, timing of sample collection, treatment received, laboratory methods, criteria or cut-offs used for TLS assessment, outcome measures, HR and 95% CI. In studies presenting both univariate and multivariate analyses, data from multivariate analyses were prioritized for extraction. Univariate and multivariate results were combined separately. Any discrepancies were resolved through consensus discussion with a third researcher.

Regarding the consistency of response definitions across the included immunotherapy studies, it should be clarified that the terms PR (partial response), CR (complete response), and pCR (pathological complete response) were based on the standard clinical or pathological assessment systems reported in the individual original studies. While these categories are universally recognized, subtle variations in the specific assessment criteria and reporting methodologies might exist among the different studies.

Study quality and bias assessment

The methodological quality and potential bias of the included studies were assessed using the Quality in Prognosis Studies (QUIPS) tool [20]. This tool evaluates six domains of potential bias: study participation, study attrition, prognostic factor measurement, outcome measurement, study confounding, and statistical analysis and reporting. Each domain contains 3 to 7 specific items. The risk of bias for a single study can be rated as low, medium, or high. To minimize individual bias, three researchers performed this assessment independently. Any scoring discrepancies were resolved by a joint re-evaluation and discussion until a consensus was reached among the authors.

Statistical analysis

The prognostic value of TLS in BCa was assessed by pooling HRs and their 95% CIs. A random-effects model was applied in the presence of significant heterogeneity; otherwise, a fixed-effects model was used. Heterogeneity was quantified using the statistic, with > 50% and a  p-value < 0.05 considered indicative of substantial heterogeneity. Sensitivity analyses were also conducted. Subgroup analyses were performed to explore potential sources of heterogeneity, including differences in study origin, sample size, and detection methods. Publication bias was examined by Egger’s [21] regression asymmetry test, and Begg’s [22] rank correlation test. Article screening and management were performed using EndNote X9. All statistical analyses were conducted in R software (version 4.5.2), primarily employing the “meta” and “forestploter” packages.

Result

Literature search and selection

The initial search resulted in 477 records: 252 from Embase, 79 from PubMed, 146 from Web of Science. After removing duplicates, 358 studies remained for further evaluation. Following title and abstract screening, 318 studies were excluded, leaving 40 studies for full-text review. Ultimately, 8 studies involving 1706 patients met the eligibility criteria and were included in the meta-analysis [16, 18, 23–28], 6 studies were included in the qualitative synthesis [23, 25, 29–32]. The process of study selection is detailed in the PRISMA flowchart (Fig. 2).

Fig. 2.

PRISMA flowchart for literature search and study selection

Study characteristics

The final selection consisted of twelve cohort studies and four clinical trials conducted across six countries, with sample sizes ranging from 8 to 313 participants. The cancer types involved in the eligible studies included NMIBC, MIBC, UTUC. The evaluated immunotherapies associated with positive treatment response included anti-PD-1/PD-L1, anti-CTLA-4 agents, and oncolytic immunotherapy. TLS detection methodologies primarily included HE staining, mIF, IHC, spatial transcriptomics, and the HALO image analysis platform. Six studies investigated the correlation between TLS levels and response to immunotherapy [23, 25, 29–32], and we included them in qualitative synthesis. Eight studies explored the correlation between TLS levels and survival outcomes [16, 18, 23–28], and we included them in quantitative synthesis (meta-analysis). The baseline characteristics and methodological quality of the 8 studies included in meta-analysis are summarized in Table 1.

Table 1.

The main characteristics of the studies included in the meta-analysis

Researcher	Country	Study interval	Patients	Sample time	Treatment	Method	Detection method	Cut-off criterion	Outcome
Chen et al. 2026 [23]	China	Cohort Study (SYSMH)	102 MIBC	After NAC + ICB	neoadjuvant GC+anti PD-(L)1 ICB	mIF(PNAd CD79a CD3 CD21 CD23)	TLS count	median expression of TLS size	OS, PFS
		Cohort Study (Multi-center)	327 MIBC	After NAC + ICB	neoadjuvant GC+anti PD-(L)1 ICB	mIF(PNAd CD79a CD4 CD8 CD23)	TLS count	median expression of TLS size	OS, PFS
Teng et al. 2025 [24]	China	Cohort Study	119 MIBC	Before RC	RC	IHC(CD3 CD20)	TLS count	Presence	OS
Nour Hassan et al. 2026 [25]	China	Cohort Study	118 MIBC	Before RT	RT	IHC(CD21 CD23)	TLS density	0.57%	OS, CSS
							TLS mature	GC size (22,800 μm²)	OS
		Cohort Study	88 MIBC	Before NAC	NAC (cisplatin)	IHC(CD21 CD23)	TLS density	0.38%	OS
							TLS mature	GC size (15500 μm²)	OS
Lin J et al. 2025 [18]	China	Cohort Study (RMYY)	51MIBC	After RC	RC	IHC (CD20 CD3 CD21 CD68)	TLS count	≥ 4	OS
							TLS density	1.05	OS
Yilmaz F et al. 2024 [26]	Turkey	Cohort Study	135 BC	After TURBT	TURBT	HE	TLS count	≥ 3	OS, PFS
Ma G et al. 2024 [27]	China	Cohort Study	313 NMIBC	After surgery	NA	IHC (CD20 CD3 Bcl-6 CD21)	TLS count	Presence	DFS
			267 MIBC	After surgery	NA		TLS count	GC TLS Presence	DFS
Masuda T et al. 2022 [28]	Japan	Cohort Study	51 MIBC	After RC	RC	IHC (CD20 CD3 Bcl6 CD10 CD21)	TLS count	Presence	DFS
Pfannstiel C et al. 2019 [16]	Germany	Cohort Study (CCC-EMN)	135 MIBC	After RC	RC	IHC (PD-L1 CD3 CD8 CD79a PD-1 CD68 CD56)	TLS count	TLS Amount (n) > = 4 VS Absent	DFS
							TLS distance to tumor	< 129.5 micrometer VS TLS absent	DFS

DFS Disease-free survival, OS Overall survival, PFS Progression-free survival, HE Hematoxylin–eosin, IHC Immunohistochemical staining, mIF multiplex immunofluorescence, NAC Neoadjuvant chemotherapy, GC Gemcitabine and cisplatin, RC Radical cystectomy, RT Radiation therapy, ICB Immune checkpoint blockade, TURBT Transurethral resection of bladder tumor

Clinicopathological relevance of TLS in bladder cancer

Table 2 summarizes the relationships between TLS and clinicopathological features, including age, sex, tumor multiplicity, perineural invasion, pathological T stage, and vascular invasion. Our analysis revealed a significant association between TLS and tumor multiplicity (OR = 1.74, 95% CI: 1.09–2.79, p < 0.05). In contrast, TLS status did not show a statistically significant correlation with the patient’s age (OR = 0.89, 95% CI: 0.26–3.07, p = 0.71), gender (OR = 1.03, 95% CI: 0.68–1.56, p = 0.84), nerve invasion status (OR = 0.66, 95% CI: 0.27–1.62, p = 0.18), pT stage (OR = 1.30, 95% CI: 0.11–14.90, p = 0.75), or vascular invasion status (OR = 1.11, 95% CI: 0.40–3.09, p = 0.77) (Fig. 3).

Table 2.

Correlation analysis between TLS and various clinicopathological features

Characteristics	No.of studies	No.of patients	OR and 95% CI	Heterogeneity
					p value
Age (elder vs. young)	3	305	0.89 (0.26–3.07)	25.0	0.263
Gender (male vs. female)	5	885	1.03 (0.68–1.56)	0.0	0.645
Tumor Multiplicity (multiple vs. single)	3	710	1.74 (1.09–2.79)	0.0	0.770
Nerve invasion (present vs. absent)	3	699	0.66 (0.27–1.62)	0.0	0.460
pT stage (T3-4 vs. Ta-2)	4	572	1.30 (0.11–14.90)	87.2	< 0.0001
Vascular invasion (present vs. absent)	4	750	1.11 (0.40–3.09)	46.1	0.135

Fig. 3.

Forest plot of clinicopathological relevance of TLS in bladder cancer. A age; (B) gender; (C) tumor multiplicity; (D) nerve invasion; (E) pT stage; (F) vascular invasion. OR, odds ratio; CI, confidence interval. Data were pooled using a random-effects model, with an odds ratio (OR) > 1 indicates that the clinicopathological feature is associated with an increased likelihood of the presence of TLS

TLS and prognosis

Eight studies involving 1706 patients investigated the association between TLS levels and survival outcomes, including OS, PFS and DFS [16, 18, 23–28]. The pooled analysis demonstrated that higher TLS levels were significantly associated with better OS (HR = 0.49, 95% CI: 0.33–0.74), PFS (HR = 0.51, 95% CI: 0.45–0.58), and DFS (HR = 0.42, 95% CI: 0.22–0.79) in bladder cancer patients (Fig. 4). No publication bias was found using both Begg’s test (p = 0.189 for OS, p = 0.602 for PFS, p = 0.497 for DFS) and Egger’s test (p = 0.293 for OS, p = 0.936 for PFS, p = 0.245 for DFS).

Fig. 4.

Forest plot for the associations between TLS and OS (A), DFS (B), PFS C. Data were pooled using a random-effects model, with HR < 1 indicating a favorable prognostic effect associated with the presence of TLS

To assess the robustness of the meta-analysis, a leave-one-out sensitivity analysis was performed. The pooled analysis of the remaining studies did not go beyond the expected range, as shown in Supplementary Fig. 1. There is no significant difference between the revised overall estimate and the primary combined estimate, confirming the stability and reliability of the overall results.

The analysis revealed that the overall pooled Hazard Ratios (HRs) for OS, PFS, DFS remained statistically significant and within a consistent range upon the sequential omission of each study. This confirms the robustness of our primary findings. However, the point estimate and the confidence intervals showed some variability when certain studies were excluded, providing insight into the influence of specific study characteristics.

For instance, in the OS analysis, the exclusion of the study by Chen et al. [23], which had the largest sample size among the included studies and reported a pronounced protective effect (HR = 0.39, 95% CI: 0.22–0.70), resulted in a widening of the confidence interval for the pooled HR. While the association remained significant, the point estimate shifted slightly, indicating that this large, multicenter study contributed considerable weight and precision to the meta-analysis. Its exclusion underscores that the strength of the association is not dependent on a single large study but is reinforced by concordant findings from other cohorts.

Conversely, the omission of the study by Yilmaz F et al. [26], which utilized HE staining for TLS detection and reported a more conservative effect estimate, led to a marginally more favorable pooled HR. This suggests that methodological differences in TLS assessment, as discussed in our subgroup analysis, contribute to the observed statistical heterogeneity.

Subgroup analyses

Subgroup analysis identified sample size and detection method as significant sources of heterogeneity. In Chinese cohorts, a higher abundance of TLS was associated with significantly improved overall survival compared with low or TLS-negative levels (HR = 0.44, 95% CI: 0.31–0.62). In European populations, high TLS/TLS + was associated with improved DFS (HR = 0.32, 95% CI: 0.18–0.55). However, no significant differences were observed in the Turkey subgroup (HR = 0.87, 95% CI:0.44–1.70). Regarding sample size, a significant association between high TLS/TLS + and favorable OS was found in studies with fewer than 120 participants (HR = 0.46, 95% CI: 0.27–0.78), whereas studies with ≥ 120 participants showed no such correlation (HR = 0.57, 95% CI: 0.00–90.73). The detection method also influenced the results. When IHC was used, elevated TLS levels were significantly associated with improved prognosis. In contrast, studies based on HE staining (HR = 0.87, 95% CI: 0.44–1.70) or mIF (HR = 0.36, 95% CI: 0.06–2.23) did not reach statistical significance, although a beneficial trend was observed. Detailed results are presented in Tables 3 and 4.

Table 3.

Subgroup analysis of TLS for OS

Subgroup	No.of studies	Sample size	HR and 95% CI	Heterogeneity
					p value
OS	6	852	0.49 (0.33–0.74)	12.8	0.333
Country
China	5	717	0.44 (0.31–0.62)	0	0.644
Turkey	1	135	0.87 (0.44–1.70)	-	-
Sample Size
< 120	4	390	0.46 (0.27–0.78)	0	0.516
≥ 120	2	462	0.57 (0.00-90.73)	67.8	0.078
Detection Method
mIF	2	429	0.36 (0.06–2.23)	0	0.574
IHC	3	288	0.50 (0.25–0.98)	0	0.554
HE	1	135	0.87 (0.44–1.70)	-	-
Treatment
ICB	2	429	0.36 (0.06–2.23)	0	0.574
Surgery	3	305	0.61 (0.24–1.55)	21.4	0.280
Radiation	1	118	0.43 (0.14–0.77)	-	-
TLS Level
TLS count	5	734	0.49 (0.29–0.84)	28.8	0.229
TLS density	1	118	0.43 (0.14–0.77)	-	-

Table 4.

Subgroup analysis of TLS for DFS

Subgroup	No.of studies	Sample size	HR and 95% CI	Heterogeneity
					p value
DFS	4	766	0.42 (0.22–0.79)	42.8	0.155
Country
Asia	3	631	0.46 (0.15–1.41)	38.3	0.198
Europe	1	135	0.32 (0.18–0.55)	-	-
Sample Size
< 200	2	186	0.30 (0.08–1.12)	0	0.680
≥ 200	2	580	0.56 (0.04–7.98)	9.0	0.294
Treatment
Surgery	2	186	0.30 (0.08–1.12)	0	0.680

Considerable heterogeneity in some subgroup analyses likely stem from clinical and methodological diversity across the included studies. Key sources include: (1) Clinical heterogeneity: Variations in patient cohorts, pathological staging, and received treatments which can differentially modulate TLS. (2) Methodological heterogeneity: The use of disparate TLS detection methods (HE, IHC, mIF) and inconsistent quantification criteria (count, density, maturity with varying cut-offs) directly impacts classification and pooled estimates. Our subgroup analysis indicated that significant prognostic value was primarily observed with IHC-based assessments. (3) Demographic and statistical factors: Observed differences in effect sizes between geographical subgroups (e.g., China, Europe, Turkey) may reflect ethnic or regional variations. Furthermore, the inconsistent reporting of TLS location (intratumoral vs. invasive margin) also precluded stratified analysis, representing an unmeasured source of variability. These factors collectively underscore the need for cautious interpretation.

Certain subgroup comparisons, such as those for the Turkey and Europe subgroups, were based on a limited number of studies. Conclusions drawn from these analyses, therefore, should be interpreted with considerable caution. To reliably determine whether the prognostic significance of TLS varies across different populations or geographic regions, findings from this meta-analysis necessitate validation in larger-scale, multi-center, prospective studies designed with adequate power for such subgroup assessments.

TLS and response to immunotherapy

Six studies systematically evaluated TLS as biomarkers of response to immune-based therapies in MIBC (Table 5). The predictive value of baseline TLS remains inconsistent. Most clinical studies of immune checkpoint blockade reported that pre-treatment TLS abundance was not significantly associated with clinical response, suggesting limited predictive utility [25, 30, 32]. In contrast, a small spatial analysis indicated that higher baseline TLS quantity, density, and maturity might correlate with pathological complete response following combined chemo-immunotherapy, although these results require validation in larger cohorts [29].

Table 5.

The main characteristics of the studies included in the qualitative synthesis

Researcher	Study interval	Patients	Sample time	Treatment	Method	Prognostic value
Chen et al. 2026 [23]	multiple center clinical trial (ChiCTR 2000037670)	47 MIBC	After NAC + ICB	neoadjuvant Tislelizumab combined with GC	HALO/3D mIF (CD3 + T CD79a + T)	Post-treatment TLS density and count were significantly higher in patients achieving PR/CR.
	Cohort Study (SYSMH)	102 MIBC	After NAC + ICB	neoadjuvant GC+anti PD-(L)1 ICB	mIF(PNAd CD79a CD3 CD21 CD23)	Post-treatment PR/CR patients displayed a greater number of E-TLS, PFL-TLS, and SFL-TLS, along with an overall higher level of maturity.
Nour Hassan et al. 2026 [25]	Cohort Study	118 MIBC	Before RT	RT	IHC(CD21 CD23)	Baseline TLS density and maturity predicted favorable response to RT.
	Cohort Study	88 MIBC	Before NAC	neoadjuvant chemotherapy (cisplatin)		Baseline TLS density and maturity showed no significant association with response to NAC.
Wahafu W et al. 2025 [29]	Cohort Study	8 MIBC	Before NAC + ICB	neoadjuvant camrelizumab combined with GC	IHC(PNAd CD3 CD20); Spatial transcriptomics	Baseline TLS quantity, density, and maturity correlated with pCR.
Li et al. 2025 [30]	Clinical trial (NCT04610671)	21 MIBC	Before and after treatment	Oncolytic immunotherapy with nivolumab	mIF (CD3 CD4 CD8 CD20 CD138 PCK) mIF (PIGR CD20 PCK IgA IgM IgG)	Baseline TLS levels did not predict response; however, responders demonstrated increased post-treatment TLS quantity/density and markers of maturation (CD138 + plasma cells, IgG/IgA).
Zhang L et al. 2024 [31]	Cohort Study	92 MIBC	Before and after treatment	neoadjuvant gemcitabine plus cisplatin	HE and mIF (CD20 CD21 CD23 CD3 Ki67 PanCK)	NAC significantly increased TLS quantity and density
	multiple center clinical trial (RJBLC-I2N003)	20 MIBC		neoadjuvant toripalimab immunotherapy		ICB significantly improved TLS positivity rate and maturity.
	Cohort Study	41 MIBC		NAC and immunotherapy		Combination therapy (NAC + ICB) induced greater increases in TLS positivity and maturity compared to monotherapies.
van Dijk N et al. 2020 [32]	Clinical trial (NCT03387761) NABUCCO	24 stage III BCa	Before and after treatment	Preoperative ipilimumab plus nivolumab+surgery	mIF (CD21 CD23 PNAd CD20 CD3 CD27)	Baseline TLS levels were not predictive; CR patients showed post-treatment increase in TLS numbers across all maturation stages. Corticosteroid use inhibited TLS formation.

NAC Neoadjuvant chemotherapy,ICB Immune checkpoint blockade,GC Gemcitabine and cisplatin,RC Radical cystectomy,RT Radiation therapy,TURBT Transurethral resection of bladder tumor,HE Hematoxylin–eosin,IHC Immunohistochemical staining,PR Partial response,CR Complete response,pCR Pathological complete response,E-TLS Early tertiary lymphoid structures,PFL-TLS Primary follicle-like tertiary lymphoid structures,SFL-TLS Secondary follicle-like tertiary lymphoid structures

By contrast, therapy-induced TLS changes show more consistent associations with treatment outcomes [23, 30–32]. Across cohorts receiving immune checkpoint inhibitors, including Nivolumab or combined Ipilimumab + Nivolumab, responders exhibited significant increases in TLS number, density, and maturation after treatment, with greater expansion across developmental stages compared with non-responders. This expansion often displays considerable heterogeneity, suggesting that certain TLS subpopulations with stronger antitumor functionality may be preferentially induced, potentially explaining the superior benefit in some complete responders.

This aligns with the known effects of different treatments on the tumor immune microenvironment. Neoadjuvant chemotherapy (NAC) can remodel the immune landscape and has been shown to promote TLS neogenesis and maturation [31]. Immune checkpoint inhibitors (ICIs) are potent inducers of TLS [33]. Notably, emerging evidence suggests a synergistic effect, where combination chemoimmunotherapy may be superior to either modality alone in enhancing TLS positivity and maturity, correlating with improved pathological response rates [34–37].

The interaction between TLS and treatment response may be further contextualized by molecular subtypes [33, 38, 39]. Basal-subtype tumors, which frequently exhibit an “inflamed” phenotype, may harbor TLS and demonstrate greater sensitivity to immunotherapy. In contrast, non-infiltrated luminal subtypes (e.g., luminal-papillary), often characterized as “immune-desert,” typically show poorer responses to ICIs. Thus, the presence and inducibility of TLS may represent a morphological correlate of the active, immune-responsive tumor microenvironment that defines treatment-sensitive disease, particularly in the context of modern immunotherapy and chemoimmunotherapy regimens [16, 29, 38].

Mechanistically, TLS expansion is accompanied by enhanced B-cell differentiation and humoral immune activation, including increased infiltration of CD138⁺ plasma cells and elevated IgG and IgA levels, indicating that immune checkpoint blockade may promote TLS maturation and local antibody-mediated antitumor immunity [30]. Conversely, corticosteroid treatment inhibited TLS formation and maturation, highlighting the sensitivity of TLS dynamics to immunosuppressive interventions [32]. Together, these findings underscore the importance of characterizing functional TLS subsets to better understand and predict immunotherapy responses.

Discussion

Growing evidence in recent years has highlighted the significant prognostic value of TLS across multiple cancer types [40, 41]. TLS are considered a promising biomarker reflecting an active anti-tumor immune microenvironment. However, the prognostic relevance of TLS in BCa, particularly its predictive role for immunotherapy, has not been systematically evaluated and remains inconsistent across individual studies. This systematic review and meta-analysis aimed to integrate current evidence to evaluate the prognostic value of TLS in BCa patients and to preliminarily examine its association with immunotherapy response. The pooled analysis revealed that elevated TLS levels were significantly linked to better OS, PFS, and DFS in BCa patients. Importantly, we observed that while baseline TLS showed limited predictive value for immunotherapy response, therapy-induced dynamic changes in TLS were more consistently linked to favorable treatment outcomes.

TLS are ectopic immune cell aggregates that play a crucial role in orchestrating local adaptive anti-tumor immunity. Their structural components facilitate the recruitment, activation, and clonal expansion of lymphocytes within the tumor microenvironment [42, 43]. The presence of TLS is believed to create a favorable immune contexture, supporting the generation of effective and sustained anti-tumor responses [44]. However, the clinical significance of TLS exhibits heterogeneity across malignancies—a heterogeneity rooted in variations in maturation state, spatial distribution, cellular composition, and the tumor microenvironment [14, 45–48]. Contextualizing the findings of our meta-analysis within this broader landscape reveals conserved biological principles while illuminating the specificity of BCa.

Our finding that high TLS density is associated with favorable survival outcomes in BCa aligns with the dominant paradigm observed in most solid tumors, including non-small cell lung cancer (NSCLC), colorectal cancer (CRC), melanoma, and breast cancer, where TLS presence correlates with improved prognosis. This is mechanistically supported by the role of TLS as ectopic hubs for local antigen presentation and lymphocyte activation, fostering a sustained anti-tumor immune response. However, divergence occurs in clear cell renal cell carcinoma (ccRCC), where TLS presence is paradoxically associated with poor prognosis [28]. Masuda et al. demonstrated that TLS in ccRCC exhibit a fundamentally different immunophenotype compared to BCa: significantly lower proportions of CD8⁺, FOXP3⁺ cells within the TLS, but significantly higher PD-1 positivity. Similarly, in hepatocellular carcinoma, the spatial location of TLS dictates clinical significance. TLS resident within the tumor core predicts a reduced risk of early recurrence, whereas TLS in the adjacent non-tumoral liver parenchyma are associated with increased late recurrence risk, potentially serving as “microniches” for tumor progenitor cells [49, 50]. Our data places BCa in the “favorable prognosis” camp alongside NSCLC, CRC, and melanoma, but this position requires rigorous spatial standardization in future studies to avoid conflating functionally disparate TLS subtypes.

The development of ICB has revolutionized the treatment landscape for advanced bladder cancer. However, the clinical benefit is confined to a subset of patients [4, 36], highlighting the urgent need for reliable predictive biomarkers. Our qualitative synthesis of six studies reveals that the predictive value of baseline TLS levels appears limited, as most studies reported no significant association between pre-treatment TLS abundance and clinical response to ICB. This finding suggests that the mere presence of TLS before treatment may not adequately capture the dynamic and inducible nature of an effective immune response.

In contrast, a more consistent and compelling narrative emerges when focusing on therapy-induced TLS dynamics [23, 30–32]. Across several cohorts involving patients treated with ICB (e.g., Nivolumab, Ipilimumab + Nivolumab) or combined chemo-immunotherapy, responders exhibited significant increases in TLS number, density, and maturation after treatment compared to non-responders. This expansion encompassed various TLS developmental stages, from early aggregates to more organized structures containing germinal centers. Mechanistically, this expansion was accompanied by enhanced B-cell differentiation and humoral immunity, evidenced by increased infiltration of CD138 + plasma cells and elevated IgG/IgA levels [30]. Conversely, corticosteroid treatment was found to inhibit TLS formation and maturation [32]. These observations collectively suggest that the ability to induce or expand functional TLS upon therapeutic intervention, rather than their static pre-existing level, may be a key determinant of clinical benefit. This aligns with the concept that effective immunotherapy can remodel the tumor immune microenvironment, fostering the development of tertiary lymphoid organization which in turn sustains the anti-tumor immune response [51].

This “dynamic evolution” paradigm is increasingly recognized across multiple tumor types. In HCC, neoadjuvant anti-PD-1 therapy actively induces de novo intratumoral TLS formation and maturation, with high post-treatment TLS density significantly associated with pathologic response and improved relapse-free survival; responding tumors exhibited a conversion from a pre-existing immune landscape to one rich in organized TLS [52]. In NSCLC, responding patients to neoadjuvant PD-1 blockade display enriched TLS in post-treatment specimens, and the presence of tumor-infiltrating PD-1hi CD8⁺ T cells within TLS before treatment was predictive of response [53]. The negative corollary further validates this paradigm: corticosteroids, which impair GC formation in TLS, abrogate the prognostic benefit of TLS in lung squamous cell carcinoma [54].

Subgroup analyses in our meta-analysis identified the detection method and sample size as potential sources of heterogeneity among studies. The method used to identify and quantify TLS is critical. Our analysis showed that when IHC was employed, high TLS levels were significantly associated with better survival outcomes [16, 18, 24, 25, 27, 28]. In contrast, when HE staining or mIF was used alone, the association, while trending positive, did not reach statistical significance. HE staining, though cost-effective, cannot reliably assess TLS organization or maturity, while mIF, though powerful, may have varied analytical pipelines. This underscores the lack of a standardized, universally accepted method for TLS assessment in BCa. For consistent evaluation of TLS prognostic value, IHC, either alone or in combination, could provide a more reliable approach, though future studies are needed to establish and validate a standardized detection protocol.

In our included studies, the definition and reporting of TLS location were inconsistent: some focused on the intratumoral region, some on the invasive margin with varying definitions, and others did not specify the location. This variability precluded a stratified meta-analysis based on TLS location. Given that TLS in different anatomical niches may encounter distinct immune pressures and cellular compositions, leading to potentially divergent impacts on prognosis, future research should explicitly define and stratify TLS based on their precise spatial arrangement within the tumor-immune ecosystem.

Our analysis could not perform a stratified assessment based on histologic subtypes (e.g., Sarcomatoid, Micropapillary) or divergent differentiation of urothelial carcinoma [55, 56]. Only a minority of included studies reported such data, and definitions were inconsistent, preventing quantitative synthesis. Qualitative insights from the literature suggest subtype-specific differences in TLS density and immune contexture (e.g., TLS in sarcomatoid tumors showed a significant decrease in CD8 + T cells, suggesting that divergent differentiation may be associated with a more immunosuppressive microenvironment), which may influence prognosis and therapy response [28]. Future studies with standardized pathological reporting are needed to clarify the role of TLS across different bladder cancer subtypes.

Conclusion

This meta-analysis demonstrated that high levels of TLS are significantly associated with improved overall survival, progression-free survival, and disease-free survival in bladder cancer patients, supporting their role as a robust prognostic biomarker. While baseline TLS status shows limited predictive value for immunotherapy response, therapy-induced increases in TLS number, density, and maturation are consistently correlated with favorable treatment outcomes. These findings suggest that dynamic changes in TLS, rather than static baseline levels, may serve as a more relevant predictive biomarker for immunotherapy efficacy. Validation in larger prospective studies and standardization of TLS assessment methods (encompassing maturation state, spatial distribution, cellular composition) are needed to translate these insights into clinical practice.

Limitations

Several limitations should be acknowledged. First, the available data lacked sufficient granularity for subgroup analyses based on histological subtypes or treatment type (immunotherapy alone, chemotherapy alone, or combined). Second, substantial methodological heterogeneity existed in TLS assessment across studies. Detection methods (IHC, HE, mIF), positivity cut-offs, maturation criteria, and spatial localization reporting were inconsistent, hindering cross-study comparability. This variability contributed to observed heterogeneity and underscores the need for standardized TLS evaluation protocols. Third, therapy-induced dynamic changes appear more predictive of treatment benefit, particularly for immunotherapy. However, the limited number of studies and small sample sizes for immunotherapy response analyses warrant cautious interpretation. Fourth, TLS biology is highly tumor-specific, as evidenced by divergent prognostic associations in other malignancies (e.g., renal cell carcinoma, hepatocellular carcinoma). This limits direct extrapolation of pan-cancer findings and reinforces the need for tumor-specific validation. Finally, all included studies were retrospective in design, with heterogeneous patient treatments that may confound TLS–prognosis associations.

Authors’ contributions

TX conceived and designed the study. PH, CW, XZ, JH, QW, XL, SJ, YD, YS were responsible for the collection and assembly of data, data analysis, and interpretation. PH, CW, XZ, SJ were involved in writing the manuscript. TX revised the manuscript. All authors read and approved the final manuscript. PH, CW, XZ, SJ contributed equally to this work.

Funding

Noncommunicable Chronic Diseases-National Science and Technology Major Project (2024ZD0525700), The National Key Research and Development Program of China (2023YFC2507000), Innovation Fund for Outstanding Doctoral Candidates of Peking University Health Science Center (BMU2024BSS001), National Natural Science Foundation of China (82471866, 82271877, 82472912), Natural Science Foundation of Beijing, China (7242150, 7264343, QY25150, 7264340), Beijing Municipal Science & Technology Commission (Z221100007422097), Capital’s Funds for Health Improvement and Research of China (2022-4-4087) and Peking University People’s Hospital Scientific Research Development Funds (RDEB2025-02, RDGS2022-02, RDX2024-01), the China Postdoctoral Science Foundation (2025M781883).

Data availability

All datasets generated for this study are included in the article/ Supplementary Material.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.