Prognostic value of different biomarkers in bladder cancer: protocol for an umbrella review of systematic reviews and meta-analyses

Abstract

Introduction

Bladder cancer (BC), the second most common urinary cancer, is classified as either non-muscle-invasive BC or muscle-invasive BC. Although numerous studies have identified potential prognostic biomarkers, none have yet been implemented in clinical practice. A comprehensive umbrella review comparing all prognostic biomarkers in BC remains absent. Thus, we aim to synthesise evidence from systematic reviews and meta-analyses on prognostic factors in patients with BC.

Methods and analysis

We will search PubMed/MEDLINE, Scopus, Web of Science (WOS), EMBASE, Cochrane Database of Systematic Reviews (CDSR), Google Scholar and EPISTEMONIKOS for systematic reviews of BC prognostic biomarkers from the inception date to 1 November 2025. Title and abstract will be screened for potentially relevant studies. Then two reviewers will independently select eligible reviews based on eligibility criteria. A data extraction sheet will be used to extract relevant information. The risk of bias in the included reviews will be assessed with the Risk of Bias in systematic reviews tool. Using the random-effects model for each prognostic biomarker, we will estimate the pooled HR and its 95% CI as well as the prediction interval and its 95% CI. To quantify heterogeneity between studies, we will use the I² metric. We will use contour funnel plots, doi plots, the Egger test and the Copas method to evaluate evidence of small-study effects. The credibility assessment will be assessed using the following items: level of statistical significance, number of cases, degree of statistical heterogeneity, significance of the PI, evidence of small-study effects and excess significance bias. We will perform sensitivity analyses with multiple approaches.

Ethics and dissemination

Formal ethical approval is not required, as primary data will not be collected in this study. The findings of this study will be disseminated through peer-reviewed publications.

PROSPERO registration number

CRD42024602432.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Re-meta-analysis of primary studies combined with calculation of the corrected covered area will improve the accuracy of pooled estimates.

• Excess statistical significance evaluation and multiple approaches to assess small-study effects will strengthen the robustness of bias detection.

• Residual confounding and variability in adjustment models across primary studies may affect the validity of pooled effect sizes.

Background

Bladder cancer (BC) is the second most prevalent kind of cancer in the urinary system.¹ BC is generally classified into two categories: non-muscle-invasive BC (NMIBC) and muscle-invasive BC (MIBC).² Tumours that are localised to the epithelium (urothelium) are classified as NMIBC (stages Ta, T1 and Tis), while tumours that infiltrate the muscle layer of the bladder (stage T2), perivesical fat (stage T3) or other nearby tissues (stage T4) are categorised as MIBC. About 75% of cases are classified as NMIBC, characterised by a significant rate of recurrence and progression, whereas 25% present with MIBC or metastatic illness.^{3 4}

Patients with BC frequently ask their urologists about the course of their disease and factors improving their prognosis. Patients with NMIBC are at risk of recurrence or progression to muscle-invasive disease, which necessitates frequent cystoscopic and histopathological monitoring.⁵ Cystoscopy, however, is an invasive procedure for the patient and costly for public health organisations. Therefore, it is critical to gain more knowledge about the association of biomarkers with BC prognosis and identify patients who require close monitoring.

With the advancement of research, a wide range of biomarkers has been examined for their ability to refine prognoses and guide personalised treatment strategies. They are derived from diverse biological samples, including serum, urine, histopathology and genetic data, providing unique insights into disease progression. These biomarkers encompass a broad spectrum, from genetic mutations and inflammatory markers to long non-coding RNAs (lncRNA), micro RNA (miRNA) and circulating tumour DNA.^6–9

Despite promising findings, none of these biomarkers have yet been incorporated into clinical decision-making. One study, for instance, divided patients with NMIBC at high risk of progression into groups with good, moderate and poor risk of progression based on mutations in FGFR3 and GATA2 methylation.¹⁰ Studies have also demonstrated a correlation between poor outcomes of tumour mutational burden and APOBEC-associated mutations and increased invasion and progression of NMIBC.¹¹ lncRNAs and miRNAs are also emerging prognostic biomarkers for BC patients. A recent systematic review and meta-analysis indicated that high expression of the miR-200 family may correlate with better prognoses in BC patients.¹²

To our knowledge, no umbrella reviews have comprehensively evaluated and compared all prognostic factors in BC patients. Nonetheless, the only published umbrella review related to our objective focused on the role of negative likelihood ratio (NLR) as a prognostic biomarker across all urologic cancers and examined its association with survival rates. Based on 10 systematic and meta-analyses, the results of this study on BC revealed that the HR in MIBC for OS was 1.32 (95% CI 1.20 to 1.45), the progression-free survival (PFS) was 1.57 (95% CI 1.23 to 2.00), the recurrence-free survival (RFS) was 1.56 (95% CI 1.28 to 1.90) and the CSS was 1.42 (95% CI 1.28 to 1.57). In NMIBC, HR for PFS was 1.87 (95% CI 1.55 to 2.25), for RFS was 1.85 (95% CI 1.43 to 2.40). Nonetheless, most of the survival outcomes showed significant heterogeneity.¹³

Here, we aim to comprehensively evaluate all prognostic factors to have a usable and informative document to clearly address the factors associated with survival outcomes of BC patients. We will also discuss the challenges and prospects of these biomarkers for clinical implementation in BC.

Methods

The umbrella review is registered in the PROSPERO database (CRD42024602432). In the final article, any amendments to the protocol will be described. This study is reported in accordance with the PRISMA-P guidelines, and the corresponding checklist is provided in online supplemental table 1.¹⁴ The study process will be illustrated through a detailed flow chart (online supplemental figure 1). This study is planned to commence following the acceptance of the study protocol article, and it is expected to be completed approximately 5 months after protocol article approval.

Search strategy

In order to identify systematic reviews and meta-analyses of prognostic biomarkers in patients with BC, we will search PubMed/MEDLINE, Scopus, Web of Sciences (WOS), EMBASE, Cochrane Database of Systematic Reviews (CDSR), Google Scholar, PROSPERO and EPISTEMONIKOS databases from the inception date to 1 November 2025. There will be no language restrictions. Moreover, we will manually check the references of included systematic reviews and meta-analyses to find any relevant records that were not found through an electronic search. As an example, table 1 displays a search strategy for PubMed/MEDLINE database.

Table 1. Example of search strategy in Pubmed/Medline database.

#1	[mesh]	“Urinary Bladder Neoplasms”
#2	[tiab]*	(((Neoplasm) AND (“Urinary Bladder”)) OR “Urinary Bladder Neoplasm” OR “Bladder Neoplasms” OR “Bladder Neoplasm” OR ((Neoplasm) AND (“Bladder”)) OR “Bladder Tumors” OR “Bladder Tumor” OR ((Tumor) AND (“Bladder”)) OR ((Tumors) AND (“Bladder”)) OR ((Neoplasms) AND (“Bladder”)) OR “Urinary Bladder Cancer” OR ((Cancer) AND (“Urinary Bladder”)) OR “Bladder Cancer” OR “Bladder Cancers” OR ((Cancer) AND (“Bladder”)) OR “Cancer of Bladder” OR “Cancer of the Bladder” OR “Malignant Tumor of Urinary Bladder”)
#3	[tiab]*	“urothelial carcinoma” OR “urothelial cancer” OR “urothelial neoplasm”
#4		#1 OR #2 OR #3
#5	[tiab]	Prognos*
#6		#4 AND #5
#7	[pt]	systematic review
#8	[pt]	meta-analysis
#9	[sb]	systematic
#10	[ti]	(“systematic review*” OR meta-analysis OR metaanalysis OR “systematic literature review” OR “systematic scoping review” OR “systematic narrative review” OR “systematic qualitative review” OR “systematic evidence review” OR “systematic quantitative review” OR “systematic meta-review” OR “systematic critical review” OR “systematic mixed studies review” OR “systematic mapping review” OR “systematic cochrane review” OR “systematic search and review” OR “systematic integrative review”)
#11		#7 OR #8 OR #9 OR #10
#12		#6 AND #11

^*Search tags may be changed based on the proper number needed to read.

Eligibility criteria

Systematic reviews and meta-analyses involving participants at any stage and with any type of BC or urothelial carcinoma regardless of age, gender or ethnicity will be considered in our study. Systematic reviews and meta-analyses that have assessed the role of any type of biomarkers (genomic, epigenetic, protein and metabolite, immunological, inflammatory, pathological and tumour-specific biomarkers) from any source (including serum, urine, tissue and etc) will be included. Systematic reviews and meta-analyses that compare the prognosis of patients with BC with the prognosis of any control group are eligible. The outcomes of systematic reviews and meta-analyses can include any of the following: overall survival (OS), PFS, disease-free survival (DFS), disease-specific survival (DSS) and RFS. Original studies of these systematic reviews and meta-analyses must be retrospective or prospective cohorts, prospective case–control studies.

In our umbrella review, only systematic reviews will be included. There will be no exclusion based on language. The following categories of articles will not be accepted: animal experimental systematic reviews and meta-analyses, systematic reviews without meta-analysis, narrative reviews, scoping reviews, letters, editorials and protocols.

Screening, selection and data extraction

First, we will screen the title and abstract of retrieved systematic reviews and meta-analyses from databases. Then, two independent reviewers will examine the full text of screened articles for inclusion criteria and eligible studies will be selected. Discrepancies will be resolved through discussion and, if necessary, consultation with a third expert author (SMKA).

Two independent authors will accomplish data extraction from studies.

We will retrieve the following data:

• General information: author, title, source, study dates, publication date, mean or median of follow-up time, study design (eg, randomised clinical trials, prospective or retrospective cohort studies (CS), nested case–control and case–cohort studies),

• Participants’ information: type of cancer (urothelial carcinoma vs non-urothelial), number of patients with BC and control group in order to develop a 2 by 2 table, number of participants is true positive, true negative, false positive and false negative group, treatments received, ethnicity, age and sex distributions

• Tumour characteristics: tumour grade, tumour stage, tumour number, tumour site

• Biomarker characteristics: type of biomarkers, methods of quantification of biomarkers, biosamples (plasma, urine, tissue), biomarker cut-points (in Kaplan-Meier curve)

• Effect size: The HR and its 95% CIs for survival outcomes such as OS, RFS, MFS, PFS or DSS (CSS), diagnostic accuracy characteristics such as the area under the ROC curve (AUC), diagnostic odds ratio (DOR), sensitivity, specificity, positive likelihood ratio and NLR.

The data-extraction sheet will be pilot tested for consistency on a small sample of included studies (5–10 studies). Following the pilot, the extraction form will be refined as necessary to ensure clarity and inter-reviewer consistency before proceeding with full data extraction.

Reviewers will discuss any discrepancies first by discussion and if needed through consulting with a third expert author (SMKA). Excluded studies will be reported with reasons in online supplemental table 2.

Risk of bias (quality) assessment

Risk of bias of systematic reviews and meta-analyses will be evaluated by two independent reviewers using the Risk of Bias in systematic reviews tool (online supplemental figure 2 and online supplemental table 3).¹⁵

The tool consists of three phases: (1) assessing relevance (optional), (2) identifying concerns about the review process and (3) judging bias risk. The second phase covers four domains where bias can be introduced into a systematic review: study eligibility criteria, study identification and selection, data collection and study appraisal. As part of phase 3, the risk of bias in interpretation of review findings is assessed, as well as consideration of any limitations identified in phase 2. To assist in judging concerns about the review process (phase 2) and the overall bias risk in the review (phase 3), signalling questions are included; these questions highlight aspects of the review design related to the potential for bias and provide an assessment of the risk of bias in the review process, results and conclusions.

The methodological quality of each included review will be rated high, moderate, low or critically low. In case of discrepancies, we will use discussion and if needed a third expert reviewer (SMKA) will be consulted.

Data synthesis

For each prognostic biomarker, we will estimate the pooled HR and its 95% CI using the random-effects model. The DerSimonian-Laird (DL) method will be used in meta-analyses of more than 20 studies, the Maximum Likelihood Ratio (REML) method is used when there are 10 or more studies, and the REML method with Hartung-Knapp-Sidik-Jonkman correction is applied when there are less than 10 studies. The interpretation areas of HR for the prognostic efficacy biomarkers are as follows: 1 to 1.21: trivial (inconsiderable), 1.22 to 1.85: small, 1.86 to 2.99: moderate, 3 or more: large.¹⁶

In addition, we will estimate the prediction interval (PI) and its 95% CI, which measures the certainty of the association in the event of a new study addressing the same association. We will estimate the SE of the effect size and examine whether the SE or the difference between the effect estimate and the upper or lower 95% CI is 0.20 (ie, this uncertainty is less than what is considered a small effect size) so as to identify the largest study in each meta-analysis.

To address the issue of overlapping primary studies across the included systematic reviews and meta-analyses, we adopted a two-step strategy. First, we extracted the original primary studies from each meta-analysis and conducted a re-meta-analysis by pooling these studies. This approach ensured that each primary study was included only once. Second, we calculated the corrected covered area to quantify the degree of overlap and was not used as an exclusion criterion.

We will use the I² metric to quantify the heterogeneity between studies. I² values exceeding 50% or 75% will be considered to represent large or very large heterogeneity, respectively. We will evaluate the evidence of small-study effects (publication bias) using contour funnel plots and doi plots, the Egger test and the Copas method. If any signs of a small-study effect are seen, we will adjust our results using the Copas method. P value of less than 0.10 is considered to be significant for all the small-study effect assessment tests. In case of sufficient data, the following subgroup analyses on methodological quality, stage of BC, patients receiving radical cystectomy and BC subtypes (eg, urothelial carcinoma) will be performed.¹⁷

We will apply the excess statistical significance test to evaluate whether there is a relative excess of formally significant findings in the published literature due to any reason (eg, publication bias, selective reporting of outcomes or analyses). Excess statistical significance will be claimed if a two-sided p value of less than 0.10 is observed. Credibility evaluation of our results will be conducted based on the criteria presented in table 2 and each finding will be categorised into one of five credibility levels: Convincing, Highly suggestive, Suggestive, Weak or Non-significant evidence.

Table 2. Criteria for the assessment of the credibility of evidence.

Convincing evidence	Statistical significance according to random-effects model at p<10−6 based on more than 1000 cases without large between-study heterogeneity (I2, 50%) 95% PI excluding the null value No evidence of small-study effects and excess significance
Highly suggestive evidence	Statistical significance according to random-effects model at p<10−6 associations with at least 1000 cases largest study presenting a statistically significant effect
Suggestive evidence	Statistical significance according to random-effects model at p<10−3 associations supported by at least 1000 cases
Weak evidence	Statistical significance according to random-effects model at p<0.05
Non-significant	Non-significant association p>0.05

Studies involving all histological subtypes of BC will be included to ensure a comprehensive assessment of the evidence. Given the differences in biological behaviour across tumour subtypes, subgroup analyses by histology and/or tumour behaviour will be conducted, where feasible, to explore heterogeneity and reduce potential misinterpretation.

As sensitivity analysis approaches we will perform the one-out remove method, meta-analysis only on trials, meta-analysis on high quality studies. All statistical analyses will be conducted in Stata, V.17.0 (StataCorp) and R, V.4.3.3 (R Foundation for Statistical Computing).

Discussion

Patients with BC frequently ask their urologists about the course of their disease and factors improving their prognosis. As a result, more attention has been paid to research on BC prognosis and a large amount of evidence is available regarding the course of this health condition. Multiple systematic reviews and meta-analyses have studied the prognostic factors associated with this cancer.^18–20

We aim to present a systematic evidence synthesis to describe the factors that affect the risk of mortality, recurrence or progression of BC and to provide the supporting evidence.

We expect that the results of this umbrella review will help numerous stakeholders, such as clinicians and healthcare managers, to evaluate the prognosis of these patients. This will enable them to focus their time and resources on those who need them the most. The results of this overview could also give insight to decision-makers and policymakers about the importance of early identification of prognostic factors to prevent life-threatening outcomes. Highlighting all the potential prognostic factors studied in adults with bladder neoplasm could benefit researchers who wish to further explore this topic. Moreover, our results will lay a foundation for others to develop clinical predictive models for patients suffering from this health problem.