This review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (20) and is registered in PROSPERO (CRD42023456724). The review followed the registered protocol without any deviations.

A comprehensive search of English literature using the database of PubMed, EMBASE, Web of Science, Cochrane Library, CHINAHL, China National Knowledge Infrastructure(CNKI) Databases, Wanfang Database. Screen the reference of the included articles to identify any other eligible studies. The following Mesh terms and keywords were include: ‘bladder’, ‘urothelial carcinoma’, ‘muscle-invasive bladder cancer’, ‘non-muscle-invasive bladder cancer’ ‘ sarcopenia’, ‘skeletal muscle index’, ‘muscle strength’, ‘Psoas muscle index’, the detailed search strategy is shown in Supplementary Table S1 .

We use the Boolean operators “OR” and “AND” between the groups. The publication year of these articles was limited to January 1, 2008 to September 2023, and only full-text original research articles are considered. All search results are downloaded and imported directly into Zotero, version 6.0.

We enrolled studies according to the following inclusion criteria: (1) study population: patients with any type of bladder cancer; (2)) indicator: sarcopenia (each study definition was applied, because no unique definition exists); (3) evaluated the prognostic value of preoperative sarcopenia; (4) outcomes: overall survival (OS), cancer-specific survival(CSS), the prevalence of sarcopenia, or other available data of survival; (5) study type: prospective or retrospective studies. The exclusion criteria are: (1) the sample size was less than 50; (2) no available data for analysis; (3) studies with incomplete data, such as prevalence of sarcopenia, incomplete baseline characteristics, or other critical survival data.

The study selection process was conducted by two independent reviewers (Zhang and Liu), who initially screened titles and abstracts to identify potentially eligible articles. Subsequently, a separate pair of reviewers (Li and Cao) independently assessed the full texts to determine study inclusion or exclusion. Reasons for exclusion were documented, and any discrepancies were resolved through discussion. If consensus could not be achieved, a third reviewer was consulted to make the final decision.

Two authors (Zhang and Yang) independently extracted relevant data from all eligible studies, including author names, study design, sample size, disease types, treatment modalities, patient age, sarcopenia definitions, and follow-up durations. The extracted datasets were cross-checked, and any discrepancies were resolved through consultation of the original articles. The study selection process is illustrated in the PRISMA flow diagram ( Figure 1 ).

Two reviewers independently assessed the risk of bias and overall study quality using the Newcastle-Ottawa Quality Assessment Scale (NOS) (21). The NOS scores range from 0 to 9, with studies categorized as low (0–3), moderate (4–6), or high quality (7–9). Any disagreements were resolved through discussion until consensus was achieved.

To ensure consistency in the analysis, we have adopted standardized operational definitions for OS and CSS based on widely accepted criteria in the oncology field. OS is defined as the time from the date of diagnosis or treatment initiation to the date of death from any cause, or the last known follow-up date if the patient is still alive. CSS is defined as the time from the date of diagnosis or treatment initiation to the date of death specifically due to cancer, with patients who die from other causes being censored.

Given the variability in definitions across studies, we have taken the following approaches:1. For studies that did not explicitly define OS or CSS or used a non-standard definition, we referred to the most commonly accepted definitions in the literature, as described above. When possible, we consulted with the authors of these studies for clarification to ensure the consistency of the outcomes. 2. For studies where the definition of OS or CSS significantly differed from the standardized definition, we excluded them from the pooled analysis to prevent excessive heterogeneity.

For analysis, the results extracted from the included studies were input into Stata 15.0 software package. The endpoints of OS and CSS were characterized by HR and 95%CI, and the pooled prevalence were estimated by proportion. The degree of heterogeneity was tested by Cochrane’s Q test (p value) and the I² statistic (22). We set the I² values as 25%, 50%, and 75%, indicating low, medium, and high heterogeneity, respectively. When no significant heterogeneity was detected (I² ≤ 50), the fixed-effects model was used for pooled prevalence and risk estimates, otherwise the random-effects model was used (23). We also employed funnel plot asymmetry to detect the potential publication bias. An Egger’s regression was applied to test the funnel plot symmetry (24).

Finally, the sensitivity analyses was performed to examine the influence of each study on the pooled estimates of the primary outcome. The data of included studies were divided into subgroups according to regions, measurement and median follow-up time. Due to the limited number of studies in certain subgroup analyses, sensitivity analyses were performed only for meta-analyses comprising more than two studies. All statistical tests were two-sided, with a significance threshold set at P < 0.05.

In total, 950 related citations (PubMed: 259, EMBASE: 246, Web of science:160, Cochrane Library:160, CINAHL: 55, CNKI:15, Wanfang:12) were identified and qualified through electronic database search, of which 235 were duplicates. After screening titles and abstracts and removing duplicate references, 715 articles were selected on the basis of inclusion criteria. Of these studies, 45 were excluded because of irrelevant outcome and population, not evaluate the sarcopenia, not treated with RC. A total of 18 studies (17–19, 23, 25–38) were utilized in this study. The selection process for the study is shown in the PRISMA flow chart ( Figure 1 ).

A total of 18 studies were included, all of which were retrospective and included 3,110 patients who were treated with RC. Eligible patients were relatively old, with a median age ranging from 44 to 92 years, and most were from the United States and Japan. In terms of the definition of sarcopenia, most of these studies identified sarcopenia by measuring SMI (17–19, 23, 25–28, 30–32, 34–36, 38) and PMI (29, 33, 37) at the level of the L3 using computed tomography (CT) images, only one study diagnosed based on PMV. The definition of sarcopenia is detailed in Table 1 . The characteristics of the 18 studies are summarized in Table 2 .

Quality assessment

The quality assessment and risk of bias were conducted in accordance with the NOS. The NOS score of each included study is listed in Table 1 ; All studies evaluated were of high quality (NOS score ≥ 6).The detailed quality assessment results are displayed in Supplementary Table S2 .

Prevalence of sarcopenia

In the 18 studies available for the meta-analysis, the pooled prevalence of sarcopenia was estimated to be 49% (95% CI: 41% to 57%, I² = 95.3%, P < 0.001), which was based on a random-effects model. Among them, the prevalence of sarcopenia defined by SMI (International Consensus) was estimated to be 56% (95% CI: 48% to 64%, I² = 91.6%, P < 0.001), the sarcopenia defined as SMI(Martin) was estimated to be 49% (95% CI: 32% to 66%, I² = 94%, P < 0.001)( Figure 2 ). In addition, a stratified analysis was conducted according to regions, and the results are shown in Figure 3 .

Data from 15 studies, including 2,784 participants, were available to meta-analyze overall survival. Of the included studies, OS was defined in 12 from treatment initiation to death, or the last follow-up (15, 17–19, 26, 27, 29–33, 37); and from the time of diagnosis to death or the last follow-up, in the other 3 studies (25, 35, 38), OS was not clearly defined in the remaining two studies. We observed that patients with sarcopenia had a worse OS compared with those without sarcopenia, the pooled HR was 1.64 (95% CI: 1.30 to 1.97, I² = 76.5%, P < 0.001; Figure 4 ). Because of high heterogeneity was revealed, so we used the random-effect model.

Data from 9 studies, including 1514 participants, were available to meta-analyze cancer-specific survival. Of the included studies, CSS in 7 studies was defined as the interval from RC to death attributable to bladder cancer progression or metastasis. CSS was not clearly defined in the remaining two studies. The results of meta-analysis showed that sarcopenia was associated with poor CSS, the pooled HR is 1.86 (95% CI: 1.45 to 2.27, I² = 0.0%, P < 0.001; Figure 4 ).

We conducted subgroup analyses based on geographical region, measurement methods, and follow-up time separately for the outcomes of OS and CSS, as shown in Table 3 . The studies were divided into two groups: Asian and non-Asian regions, with 7 and 8 studies included, respectively. The meta-analysis results revealed that sarcopenia was a predictive factor associated with a decrease in both OS (HR: 1.55, 95% CI: 1.45 to 2.27, P < 0.001; HR: 1.45, 95% CI: 1.06 to 1.84, P < 0.002) and CSS (HR: 2.34, 95% CI: 1.50 to 3.17, P < 0.001; HR: 1.75, 95% CI: 1.30 to 2.21, P < 0.001), irrespective of whether the cases originated from Asian or non-Asian populations.

In the subgroup analysis based on measurement methods, due to limited study numbers, we only conducted meta-analyses for SMI (International Consensus) and SMI(Martin). The results showed that sarcopenia measured using SMI (International Consensus) was correlated with a decrease in OS (HR: 1.53, 95% CI: 1.35 to 1.70, P < 0.001), while SMI(Martin) did not reach statistical significance (HR: 1.75, 95% CI: 0.66 to 2.83, P < 0.080). Both measurement methods, SMI (International Consensus) and SMI(Martin), had an impact on the decrease in CSS (HR: 1.76, 95% CI: 1.32 to 2.21, P < 0.001; HR: 2.34, 95% CI: 1.28 to 3.40, P < 0.001).

Regarding median follow-up time, we used a cut-off of 30 months for subgroup analysis. The results showed that sarcopenia was associated with a decrease in OS regardless of whether the median follow-up time exceeded 30 months (HR: 1.53, 95% CI: 1.36 to 1.71, P < 0.001; HR: 1.89, 95% CI: 1.12 to 2.67, P < 0.013). For CSS, subgroup analysis with a median follow-up time exceeding 30 months revealed an association with sarcopenia (HR: 1.88, 95% CI: 1.44 to 2.29), but no statistically significant association was observed in the subgroup with a median follow-up time of less than 30 months (HR: 2.08, 95% CI: 0.81 to 3.36, P < 0.187). Forest plots for all outcomes are provided in Supplementary Figures 5 - 10 .

To evaluate the robustness and reliability of the primary analysis, sensitivity analyses were conducted by sequentially excluding individual studies. The results indicated that the overall survival outcomes remained consistent regardless of the removal of any single study, including those of relatively lower methodological quality. The results are shown in Supplementary Figure 1 .

Publication bias was assessed using Begg’s test, and funnel plots were examined for symmetry. The P-values for Begg’s test were 0.223 for OS and 0.978 for CSS, suggesting no significant evidence of publication bias in the meta-analysis. The funnel plots are provided in the Supplementary Materials .

While this meta-analysis consolidates current evidence and highlights sarcopenia as an important prognostic indicator in BC, several limitations should be acknowledged. First, substantial clinical heterogeneity among the included studies limited the feasibility of performing meta-analyses for several outcome measures. Second, the diverse definitions of sarcopenia employed across the studies, despite primarily relying on CT scans, exhibited variations in scanning levels and thresholds. Such differences between studies may have biased our results. Likewise, inconsistencies in the definitions of OS and CSS may introduce bias in the pooled results. Third, all included studies were retrospective in nature, which may have introduced a higher risk of bias and contributed to the observed inter-study heterogeneity. Furthermore, it is important to note that all of the included studies were published in English or Chinese, which may have led to language bias. Consequently, the findings should be interpreted with caution, and their applicability to clinical practice remains limited. Prospective, well-designed studies are warranted to further validate the prognostic significance of sarcopenia.