Skip to main content
NCBI home page
The Journal of Nutrition, Health & Aging logoLink to The Journal of Nutrition, Health & Aging
. 2026 Feb 8;30(4):100805. doi: 10.1016/j.jnha.2026.100805

Sarcopenia and recurrence risk in high-risk non–muscle-invasive bladder cancer: a nonlinear modeling approach

Gerald Klinglmair a,1, Maximilian Lutz b,1, Tatia Telia b,1, Nils CH van Creij c, Hamed Wafa a, Piotr Tymoszuk d, Beatrice Heim e, Dora Niedersuess-Beke f, Angelika Terbuch g, José Daniel Subiela h, Francesco del Giudice i, Roman Mayr j, Anna Luger b, Renate Pichler a,*
PMCID: PMC12907225  PMID: 41655456

Graphical abstract

graphic file with name fx1.jpg

Open in a new tab

Keywords: Bladder cancer, BCG, Sarcopenia, Recurrence, SMI, PMI, BMI

Abstract

Background

Sarcopenia and altered body composition are established prognostic markers in several malignancies, but their relevance in non–muscle-invasive bladder cancer (NMIBC) remains unclear. This study evaluated associations between body composition indices—body mass index (BMI), skeletal muscle index (SMI), and psoas muscle index (PMI)—and recurrence risk in high-risk NMIBC patients treated with intravesical Bacillus Calmette–Guérin (BCG).

Methods

Body composition was assessed on computed tomography at primary diagnosis. Sarcopenia was defined using published sex-specific cut-offs. Recurrence-free survival (RFS) was analyzed using Kaplan–Meier estimates, Cox regression with linear and spline terms, and a gradient boosted machine (GBM) model.

Results

A total of 118 patients (median age 71 years; 78% male) were included. Tumor stages were pTa (40%), pT1 (54%), and primary carcinoma in situ (CIS) in 6%. During a mean follow-up of 32 months, 30 patients (25.4%) developed recurrence. Median BMI was 26 kg/m². SMI and PMI were higher in men and declined with age, whereas BMI remained stable. Sarcopenia prevalence varied widely across definitions (6–66%) with low-to-moderate concordance (Cohen’s κ: 0–0.58). No sarcopenia definition independently predicted recurrence. In the GBM model, BCG maintenance, tumor stage, CIS, age, BMI, SMI, and PMI were important predictors. Spline Cox models demonstrated nonlinear U-shaped associations between BMI, SMI, PMI, and recurrence risk, with increased risk at both low and high values.

Conclusions

Body composition shows nonlinear associations with recurrence in high-risk NMIBC. Both low and excessive body and muscle mass may adversely affect disease control. The heterogeneity and limited prognostic value of current sarcopenia definitions highlight the need for standardized, multiparametric, and nonlinear risk modeling approaches.

1. Introduction

Bladder cancer (BC) is the tenth most common malignancy worldwide [1]. With an estimated 573,000 new cases and 213,000 deaths worldwide in 2020, it represents a significant public health and economic burden [1]. Approximately 70% of newly diagnosed BCs are non–muscle-invasive bladder cancer (NMIBC), which is typically managed with transurethral resection of the bladder tumour followed by intravesical therapy, most commonly Bacillus Calmette-Guérin (BCG) in intermediate- and high-risk disease [2]. Still, recurrence and progression rates in NMIBC are high, necessitating long-term surveillance and significant healthcare resources [2].

Traditional general NMIBC prognostic schemes rely on selected pathological variables such as stage, grade, tumor multiplicity, or concomitant carcinoma in situ (CIS) [3]. For BCG-treated NMIBC patients, the Club Urológico Español de Tratamiento Oncológico (CUETO) risk model is based on seven prognostic factors: gender, age, prior recurrence status, number of tumours, T category, associated CIS, and WHO 1973 tumour grade [4]. However, neither general nor therapy-tailored established risk models integrate systemic, patient-specific factors like body mass and body composition parameters that may influence disease course or treatment response.

Sarcopenia, a progressive and systemic loss of skeletal muscle mass and function, has emerged as a significant prognostic biomarker in oncology [5]. It often results from a complex interplay between cancer-related inflammation, metabolic dysregulation, malnutrition, and physical inactivity [6]. Sarcopenia is commonly associated with poor treatment tolerance, postoperative complications, impaired immune function, and worse oncologic outcomes [7]. While the adverse impact of sarcopenia on survival in muscle-invasive and metastatic bladder cancer is increasingly recognized, its prognostic role in NMIBC remains largely unexplored [8]. Muscle mass metrics such as the skeletal muscle index (SMI) and psoas muscle index (PMI) can be derived from routine cross-sectional imaging in NMIBC patients and may offer cost-effective, noninvasive prognostic insights [9].

Given the aging population and the increasing use of immunomodulatory therapies such as BCG and immune checkpoint inhibitors, understanding the interplay between sarcopenia, immunotherapy response, and oncologic outcomes is both timely and clinically relevant [10]. This study aims to assess body composition parameters at the time of cancer diagnosis—specifically body mass index (BMI), SMI, and PMI—as predictors of relapse risk in NMIBC patients undergoing intravesical BCG therapy. Secondary objectives include the characterization of associations among biometric, demographic, pathological, and therapy-related variables, as well as the evaluation of concordance and prognostic performance of previously published definitions of sarcopenia.

2. Materials and methods

2.1. Patients

Following approval by the local ethics committee (study number 1349/2020), medical records of patients with available baseline CT at diagnosis of a high-risk NMIBC who underwent intravesical BCG therapy were retrospectively reviewed. Intravesical BCG therapy was administered in accordance with the European Association of Urology (EAU) guidelines [2]. Induction therapy consisted of weekly instillations for six weeks, followed by maintenance therapy over 1–3 years (BCG strain RIVM seed 1173-P2; BCG Medac, Wedel, Germany). Patients were monitored by cystoscopy and urinary cytology every three months for the first two years, every six months for years 3–5, and annually thereafter. CT urography was performed at primary diagnosis and repeated annually or in cases of tumor recurrence. Recurrence was defined as high-grade or low-grade NMIBC relapse, progression of disease, or appearance of upper tract urothelial carcinoma.

2.2. CT image analysis and body composition measurements

Baseline CT was performed with intravenous contrast using a Somatom Definition Flash or a Somatom Drive scanner (both Siemens Healthineers, Erlangen, Germany) at primary diagnosis of NMIBC. Acquisition parameters were tailored to patient size, with tube voltages ranging from 90 to 120 kV. Axial images were reconstructed with slice thicknesses of 1 mm and 3 mm using soft tissue kernels. Anonymized image analysis was conducted independently by two radiologists who were blinded to all clinical data. Skeletal Muscle Index (SMI) and Psoas Muscle Index (PMI) were assessed as described by Gómez-Pérez et al. [11]. Axial CT slices at the level of the third lumbar vertebra (L3), with slice thicknesses between 1 and 5 mm, were evaluated using ImageJ (National Institutes of Health, Bethesda, MD, USA). The Hounsfield unit (HU) range was set from −29 to +150 HU, which is standard for skeletal muscle segmentation [[11], [12], [13]]. First, the outer contours of all skeletal muscles within the L3 cross-section were manually delineated to determine the total muscle area, including only pixels within the specified HU range. Next, the inner contour of the skeletal muscles and the vertebral body were segmented. The inner area was subtracted from the outer muscle contour to obtain the skeletal muscle area (SMA). The left and right psoas major muscles were manually segmented and measured individually using the same HU range. Their areas were summed to obtain the psoas muscle area (PMA). Finally, SMI and PMI were calculated by dividing SMA and PMA by the patient’s height squared (cm2/m2).

2.3. Statistical analysis

All statistical analyses were performed using R version ≥4.2.3. Patients were classified as underweight, normal weight, overweight, or obese, as proposed by Martin et al. [14]. Sarcopenia was defined according to several published schemes based on sex-specific cutoffs of SMI and PMI at diagnosis [[14], [15], [16], [17]]. Differences in numeric variables were evaluated using Mann–Whitney U tests with the biserial correlation coefficient (r) as an effect size measure. Distributions of categorical variables were compared using the χ2 test with Cramer’s V as an effect size metric. Spearman’s rank correlation tests were used to assess associations between continuous variables, with ρ as the correlation coefficient. Differences in relapse-free survival were analyzed by Kaplan–Meier estimation and Peto–Peto tests. P values obtained from statistical hypothesis testing and model inference were adjusted for multiple comparisons within each analytical task using the false discovery rate (FDR) method.

A correlation network was constructed for candidate predictors of relapse-free survival (Suppl. Table S1/Suppl. Figure S1). Inter-rater reliability, i.e., concordance between sarcopenia definitions, was assessed using Cohen’s κ. Relapse risk was modeled using univariable and sex- and age-adjusted Cox proportional hazards models, as well as two multi-parameter survival machine learning algorithms. Model performance was evaluated on the full dataset and using leave-one-out cross-validation (LOOCV), based on the concordance index (C-index), integrated Brier score (IBS), and R2 metrics. Risk predictions from the machine learning models were interpreted with Shapley additive explanations (SHAP). Finally, model-free survival analyses were conducted using the Kaplan–Meier method, with Peto–Peto tests for differences in relapse-free survival between patient strata. Details of the statistical analyses, algorithms, and software are provided in the Supplementary Methods.

3. Results

3.1. Patient characteristics

A total of 118 patients were included in this retrospective analysis, of whom 26 (22%) were female. The median age at diagnosis was 71 years. At diagnosis, the median BMI was 26 kg/m2 (IQR 23–29), median SMI was 49 cm2/m2 (IQR 42–56), and median PMI was 6.4 cm2/m2 (IQR 5.2–7.8).

At diagnosis, 47 patients (40%) presented with pTa tumors, 64 (54%) with pT1, and 7 (5.9%) with primary CIS. The majority (94%) had high-grade NMIBC (n = 111). Concurrent CIS was observed in 53 patients (45%). BCG maintenance therapy was initiated in 101 patients (86%) with a median of 4 (IQR 2–6) maintenance cycles. BCG-associated adverse events (AEs) occurred in 28 patients (24%), and in 19 individuals therapy was discontinued due to AEs. Treatment was discontinued for any reason in 46 patients (39%).

During a median follow-up of 32 (IQR 19–43) months, 30 (25%) patients experienced recurrence. Among these, 12 (40%) recurred as Ta, 10 (33%) as T1, 2 (7%) as T2, and 6 (20%) as primary CIS. Twenty-three patients (79%) had high-grade recurrence. Radical cystectomy was performed in 7 patients (5.9%), and metastatic progression occurred in 2 (1.7%). Overall mortality during follow-up was 6.8% (n = 8), Suppl. Table S2.

3.2. Significant impact of age and sex on muscle mass indices

There was a difference in age between males and females. BMI was higher in females (median 29 kg/m2, IQR 25–32) than in males (26 kg/m2, IQR 23–28), yet this difference was not significant after multiple testing correction (raw p = 0.021, corrected p = 0.2). Both the SMI (males: median 52 cm2/m2, IQR 45–58; females: 43 cm2/m2, IQR 39–46) and PMI (males: median 6.9 cm2/m2, IQR 5.7–8.3; females: 4.7 cm2/m2, IQR 4.2–5.8) were significantly higher in males than females (p < 0.001). BMI did not change substantially with age (correlation analysis, ρ = −0.16, p = 0.09), whereas both SMI and PMI decreased significantly with age (ρ = −0.34, p < 0.001 and ρ = −0.31, p = 0.0012, respectively). Moreover, BMI correlated with both SMI (ρ = 0.39, p < 0.001) and PMI (ρ = 0.29, p = 0.0012). A particularly strong positive correlation was observed between SMI and PMI (ρ = 0.8, p < 0.001, Suppl. Figure S2).

Collectively, these results indicate that while BMI remains largely unaffected by age and sex, muscle-related indices are significantly influenced by both factors—declining with age and being consistently higher in males. The correlation of SMI and PMI suggests a close relationship between whole-body and single-muscle indices of muscle mass.

3.3. Low concordance between established definitions of sarcopenia

Based on the sex-specific SMI cut-offs [[16], [17], [18]], 66 (56%), 47 (40%), and 15 (13%) patients were sarcopenic, respectively. Using the PMI-based classification by Hamaguchi et al. [17], 78 (66%) patients were sarcopenic. When applying the Bahat et al. [16] PMI thresholds, only 18 patients (15%) and 8 (6.8%) were sarcopenic according to the 5th percentile and 2 × SD cut-offs, Suppl. Table S2. The strongest agreement was observed between SMI- and PMI-based definitions of sarcopenia proposed for the healthy population by Bahat et al. [16] (κ = 0.58–0.67). For cancer-specific sarcopenia definitions, the highest agreement was observed between the schemes by Fearon et al. [15] and Martin et al. [14] (κ = 0.55). Interestingly, concordance between sarcopenia schemes for healthy populations [17,17,18] and cancer-specific definitions [14,15] was absent to weak (κ = 0.058–0.17), Suppl. Figure S3.

3.4. No prognostic significance of established sarcopenia definitions for relapse-free survival in NMIBC

BCG maintenance was identified as the strongest favorable prognostic factor (HR = 0.076, 95% CI: 0.035–0.17; 25% survival quantile: 72 months with BCG maintenance vs. 4.4 months without BCG maintenance). In contrast, primary CIS at diagnosis was strongly associated with relapse (HR = 5.9, 95% CI: 1.7–20; 25% survival quantiles: CIS 3.7 months, pT1 24 months, pTa 48 months). Other potential unfavorable prognostic factors were concomitant CIS (HR = 2.1, 95% CI: 0.99–4.3), pT1 stage (HR = 2.3, 95% CI: 0.95–5.4), and age (HR = 1.4 per decade, 95% CI: 0.98–2.1), yet these effects missed the statistical significance cutoff.

Neither BMI, SMI, nor PMI were associated with RFS in univariable analysis, assuming a simple linear relationship between the relapse risk and parameter value. Interestingly, none of the investigated sarcopenia definitions was associated with RFS in NMIBC, with HR values ranging from 0.56 (95% CI: 0.27–1.2) for Hamaguchi et al. [17] to HR 1.5 (95% CI: 0.34–6.1) for sarcopenic individuals identified with the 2 × SD SMI definition by Bahat [16], Suppl. Figure S4. These findings indicate no simple linear relationship between BMI, SMI, and PMI as standalone biomarkers and RFS.

Machine learning reveals non-linear relationship between relapse risk and body composition metrics in NMIBC

To better understand the complex interplay between clinical, pathological, and body composition parameters and RFS, we resorted to multi-parameter modeling of RFS with two machine learning algorithms, Suppl. Figure S5. Notably, the Elastic Net Cox proportional regression algorithm does not implement interactions between survival predictors, and a linear relationship between continuous variables such as BMI, PMI, or SMI is assumed. The GBM algorithm inherently handles both interactions between predictors as well as their non-linear relationships with RFS. The explanatory variables included age and sex, pathological variables, BCG maintenance therapy and its AEs, and body composition features (BMI, SMI, PMI).

The GBM (training data: concordance index [C] = 0.94, integrated Brier score [IBS] = 0.08, R² = 0.9; leave-one-out cross-validation [LOOCV]: C = 0.67, IBS = 0.15, R² = 0.45) outperformed the Elastic Net algorithm (training data: C = 0.69, IBS = 0.12, R² = 0.69; LOOCV: C = 0.66, IBS = 0.16, R² = 0.31) in predicting RFS in our NMIBC cohort (Suppl. Figure S6).

As investigated using Shapley additive explanation (SHAP) techniques for the GBM models, the most influential factors associated with elevated relapse risk were pT stage and age at diagnosis, while the most influential favorable predictor was BCG maintenance therapy. Interestingly, the importance of BMI, SMI, and PMI for prediction of relapse risk by the GBM model was comparable with tumor stage or age as measured by mean absolute SHAP values, Fig. 1A. An analysis of local variable importance, i.e., of SHAP values for particular observations, revealed clearly non-linear relationships of age, BMI, PMI, and SMI with predicted relapse risk quantified by SHAP values, Fig. 1B. Relapse risk steeply increased at older age, particularly in patients aged >70 years. Analogously, increased predicted relapse risk was observed for high PMI values >7.5 cm2/m2. BMI and SMI demonstrated S-shaped patterns, with both low and high values associated with elevated predicted relapse risk.

Fig. 1.

Fig. 1

Open in a new tab

Interpretation of the GBM model of relapse risk with Shapley additive explanations. Predictions made by the best performing Gradient Boosted Machines (GBM) Cox proportional hazard regression model were interpreted by the Shapley additive explanations (SHAP) method. Note that rising SHAP values correspond to increasing relapse risk, while decreasing SHAP values suggest reduced relapse risk. (A) SHAP values and minimum/maximum scaled values of the explanatory variables are presented in a swarm plot (left): each point represents a single value of an explanatory value, point color corresponds to the minimum/maximum-scaled explanatory variable value, SHAP values distributions are visualized with violins. Mean absolute values of SHAP serving as measures of global variable importance are shown in a bar plot (right) with bar colors coding for the association sign between SHAP and the feature values. (B) Detailed representation of SHAP values for sex, age, and key biometry parameters (body mass index [BMI], skeletal muscle index [SMI], psoas muscle index [PMI]). SHAP values are represented by points. For the sex variable, the SHAP value distributions in females and males are visualized by violins. For the remaining variables, the LOESS (locally estimated scatterplot smoothing) trends of associations between the SHAP and variable values are depicted as red lines. Gray ribbons represent standard error regions of the LOESS trends.

To corroborate the non-linear associations of anthropometric parameters with recurrence risk, we constructed age- and sex-adjusted Cox proportional hazards models of RFS with linear and spline (i.e., non-linear) terms of BMI, SMI, and PMI. As shown in Fig. 2, models with spline terms of BMI (LOOCV linear: C = 0.56, IBS = 0.19, R2 = 0.022; spline: C = 0.63, IBS = 0.19, R2 = 0.06), SMI (LOOCV linear: C = 0.56, IBS = 0.19, R2 = 0.033; spline: C = 0.6, IBS = 0.18, R2 = 0.11), and PMI (LOOCV linear: C = 0.55, IBS = 0.19, R2 = 0.034; spline: C = 0.59, IBS = 0.19, R2 = 0.06) outperformed the Cox proportional hazards models with linear terms of body composition parameters and the age- and sex-only model (LOOCV: C = 0.58, IBS = 0.19, R² = 0.054) in the training data and cross-validation.

Fig. 2.

Fig. 2

Open in a new tab

Modeling of relapse risk with age, sex, and splines of BMI, SMI, and PMI. Association of body mass index (BMI), skeletal muscle index (SMI), and psoas muscle index (PMI) at diagnosis with relapse risk was investigated by a series of Cox proportional hazard models adjusted for sex and age at diagnosis. In the models, the BMI, SMI, and PMI variables were included either as linear terms (i.e. linear relationship between the feature and relapse risk) or spline term (i.e. non-linear relationship between the feature and relapse risk). The modeling data set consisted of n = 118 patients with n = 30 relapses. Performance of the models was evaluated in the entire collective and leave-out-one cross-validation (LOOCV). (A) Metrics of model accuracy (Harrell’s concordance index [C-index]: high values reflect good model accuracy), confidence and calibration (integrated Brier score [IBS]: low values are characteristic for a well calibrated and confident model), and relapse risk variance explained by the model (R2). The statistics for the sex/age reference model, and sex/age-adjusted models with linear and spline terms of BMI, SMI, and PMI are presented in dot plots. Point sizes correspond to R2, point shapes code for data set type, points representing the same algorithm are connected with lines. (B) Brier scores (squared differences between the predicted relapse risk and 0/1-coded relapse) for the unique time points for the null model, and the sex/age-adjusted models with the spline terms of BMI, SMI, and PMI terms in the entire data set and LOOCV. The Brier score values are presented in line plots; line color codes for model and data type. IBS values are indicated in the plot legends. (C) Log hazard ratios (log HR) of relapse as functions of BMI, SMI, and PMI values predicted by the sex/age-adjusted models with the spline terms of BMI, SMI, and PMI. The log HR values are visualized as blue lines, and gray ribbons represent 95% estimation intervals obtained by bootstrap.

The estimated HR for particular values of BMI, SMI, and PMI in the age- and sex-corrected spline Cox proportional hazards models displayed U- and S-shaped patterns. In particular, for SMI (HR local maxima at 41 and 65 cm2/m2, local minimum at 52 cm2/m2) and PMI (HR local maxima at 5.6 and 10 cm2/m2, local minimum at 7.2 cm2/m2), increased recurrence risk was estimated for both low and high indices of muscle mass.

4. Discussion

Sarcopenia, defined as a progressive loss of skeletal muscle mass and function, has emerged as an independent prognostic factor in various malignancies, including muscle-invasive bladder cancer (MIBC). Multiple studies have demonstrated that sarcopenia is associated with inferior oncologic outcomes and increased perioperative morbidity following radical cystectomy (RC), [[18], [19], [20], [21]]. In a meta-analysis of over 3,000 patients, Psutka et al. reported that preoperative sarcopenia was associated with significantly reduced overall survival (HR = 1.64, 95% CI: 1.25–2.16) and cancer-specific survival (HR = 1.86, 95% CI: 1.35–2.56) in patients undergoing RC [18]. Similarly, Fukushima et al. demonstrated that sarcopenic patients had higher rates of postoperative complications and shorter RFS compared with non-sarcopenic counterparts [19]. These findings have been corroborated by later analyses showing that reduced muscle mass is predictive not only of survival but also of impaired tolerance to neoadjuvant chemotherapy [20,[19], [20], [21]]. Mechanistically, sarcopenia reflects both cancer-related catabolism and systemic inflammation, which may contribute to reduced immune competence, metabolic dysregulation, and poor recovery capacity [22].

Focusing on NMIBC, the prognostic role of sarcopenia remains uncertain, showing inconsistent results. Two trials reported no significant association between sarcopenia and survival after transurethral resection and intravesical therapy [23,24]. In contrast, Fukushima et al. observed that sarcopenic patients were more likely to experience BCG intolerance, leading to premature discontinuation and potentially inferior disease control [25]. These findings suggest that in NMIBC, sarcopenia may not directly influence tumor biology but affects treatment adherence and the immunological efficacy of BCG. Emerging evidence indicates that body composition is a clinically relevant modifier of immune responsiveness to BCG in NMIBC. Increased adiposity has been associated with higher recurrence risk and inferior oncologic outcomes following BCG therapy [26], potentially reflecting obesity-related chronic low-grade inflammation and impaired antitumor immunity. In parallel, sarcopenia has been linked to BCG non-responsiveness and unfavorable outcomes [27], consistent with reduced immune reserve and increased inflammatory burden in patients with low skeletal muscle mass. In our study, we comprehensively evaluated the role of body composition metrics in relation to recurrence in high-risk NMIBC patients undergoing BCG therapy. While spline-based and machine learning models revealed that anthropometric parameters contribute to relapse prediction in a nonlinear manner, traditional survival analyses demonstrated that only tumor stage, age and BCG maintenance therapy remain the dominant prognostic determinants in NMIBC. Notably, our data highlight both the methodological and biological complexity underlying the relationship between body composition and oncologic outcomes.

Despite consistent evidence linking low skeletal muscle mass to adverse outcomes in MIBC [[18], [19], [20], [21]], definitions of sarcopenia remain heterogeneous, with variable cut-off values across populations and imaging techniques [14]. In line with these findings, we found no independent prognostic value of SMI or PMI for RFS after correction for multiple testing. This discrepancy may reflect the heterogeneity of sarcopenia definitions, as demonstrated by the low concordance among existing classification schemes. Depending on the chosen threshold [[14], [15], [16], [17]], sarcopenia prevalence ranged in our study from 6% to 66%, highlighting the urgent need for standardized, cancer- and population-specific reference values. The inconsistency among current cut-offs likely contributes to the mixed evidence regarding the prognostic role of sarcopenia in NMIBC [[23], [24], [25]]. Thus, standardization of diagnostic criteria and integration of functional measures such as muscle strength and performance are therefore essential to improve the clinical applicability of sarcopenia assessment in bladder cancer management.

The observed sex-specific differences in muscle mass, with higher SMI and PMI in men, align with established physiological differences and confirm previous findings in oncologic and geriatric cohorts [14,16]. The inverse correlations of SMI and PMI with age might reflect the expected age-related decline in skeletal muscle mass, a process often accelerated by systemic inflammation and catabolic stress in cancer patients. In contrast, BMI remained largely stable across age and sex, suggesting that weight-based indices may obscure underlying changes in muscle and fat composition—a limitation increasingly recognized in clinical nutrition research [28]. These findings underscore the importance of muscle-specific indices over BMI when assessing nutritional and functional status in oncology.

In our analysis, advanced modeling approaches revealed that body composition parameters exhibit nonlinear associations with relapse risk. Particularly, BMI followed a U-shaped curve, suggesting increased risk at both low and high extremes—a pattern consistent with prior studies linking both undernutrition and obesity to impaired immune response and altered treatment tolerance [28,29]. Similarly, nonlinear trends in SMI and PMI suggest that both muscle depletion and excess may negatively affect clinical outcomes, possibly through immune modulation, altered pharmacokinetics of intravesical therapy, or systemic metabolic imbalance [30]. These results emphasize that conventional dichotomous cut-offs for sarcopenia or obesity may inadequately capture the biological continuum of nutritional risk [8,14].

Among clinical variables, the presence of CIS and the absence of BCG maintenance emerged as the strongest predictors of recurrence, consistent with established evidence on tumor aggressiveness and the protective effect of adequate BCG exposure [4]. The robust inverse association between BCG maintenance and recurrence risk underscores the importance of treatment adherence and supports the continued implementation of maintenance protocols despite potential adverse effects [4].

Our study has several limitations. The retrospective design and single-center setting may limit generalizability. The overall sample size is small relative to the heterogeneous high-risk NMIBC population, limiting statistical power, particularly for subgroup analyses and machine learning approaches. The relatively short median follow-up of approximately three years likely results in a low number of overall survival events, leading to immature survival estimates. Sarcopenia was assessed solely based on muscle quantity, without functional parameters, which are now central to current consensus definitions. In addition, the absence of inflammatory, nutritional, and immunological biomarkers (e.g., C-reactive protein, albumin) restricts a more comprehensive interpretation of the observed associations.

5. Conclusion

In conclusion, although BMI, SMI, and PMI show nonlinear associations with recurrence risk, they do not independently predict recurrence in NMIBC. Tumor stage, age, and BCG maintenance remain the main determinants of disease control. Muscle mass should therefore be interpreted within the broader clinical context. Future studies should standardize sarcopenia definitions and integrate functional, biochemical, and mechanistic assessments to clarify interactions between nutrition, immunity, and treatment response. Integrating clinical and anthropometric variables into prognostic models is recommended.

CRediT authorship contribution statement

All authors had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Conceptualization: GK, RP, RM, BH, AL; Data curation: HW, GK, RP, ML, AL, TT; Formal analysis: ML, AL, TT, PT; Investigation: HW, NVC, GK, ML; Methodology: BH, DNB, RM; FDG; Project administration: RP; Resources: GK, RP, AL, NVC; Software: PT, AL, ML, TT; Supervision: RP, RM, FDG, JDS, AT, DNS; Validation: RP, RM, FDG, JDS; AT, DNB; Visualization: GK, HW, NVC; Writing - original draft: GK, RP, ML, AL; Writing - review & editing: All authors.

Declaration of Generative AI and AI-assisted technologies in the writing process

The authors used ChatGPT (GPT-5, OpenAI, San Francisco, CA, USA) to assist with language editing (e.g., improving grammar, readability, and phrasing). The study design, data analysis, interpretation of results, and all scientific conclusions were entirely developed by the authors. All AI-assisted text was carefully reviewed and revised by the authors to ensure accuracy and appropriateness.

Funding

Research reported in this publication received no funding.

Data availability

The study data will be made available on request to the corresponding author (renate.pichler@i-med.ac.at). The R code is available from the GitHub repository (https://github.com/PiotrTymoszuk/sarcoBLCA).

Declaration of competing interest

The authors have no conflict of interest to declare.

Acknowledgments

None.

Footnotes

A modified abstract of this article has been submitted to the European Congress of Radiology 2026.

Appendix A

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jnha.2026.100805.

Appendix A. Supplementary data

The following are Supplementary data to this article:

mmc1.pdf (2.8MB, pdf)
mmc2.xlsx (11.5KB, xlsx)

References

  • 1.Sung H., Ferlay J., Siegel R.L., Laversanne M., Soerjomataram I., Jemal A., et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209–249. doi: 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
  • 2.Babjuk M., Burger M., Compérat E.M., Gontero P., Mostafid A.H., Palou J., et al. EAU guidelines on non–muscle-invasive bladder cancer (TaT1 and CIS) – 2022 update. Eur Urol. 2022;81(1):75–94. doi: 10.1016/j.eururo.2021.08.010. [DOI] [PubMed] [Google Scholar]
  • 3.Sylvester R.J., Rodríguez O., Hernández V., Turturica D., Bauerová L., Bruins H.M., et al. European Association of Urology (EAU) Prognostic Factor Risk Groups for Non–muscle-invasive Bladder Cancer (NMIBC) Incorporating the WHO 2004/2016 And WHO 1973 Classification Systems for Grade: An Update from the EAU NMIBC Guidelines Panel. Eur Urol. 2021;79(4):480–488. doi: 10.1016/j.eururo.2020.12.033. [DOI] [PubMed] [Google Scholar]
  • 4.Fernández-Gómez J., Madero R., Solsona E., Unda M., Martínez-Piñeiro L., Ojea A., et al. Predicting non–muscle-invasive bladder cancer recurrence and progression in patients treated with Bacillus Calmette-Guérin: the CUETO scoring model. J Urol. 2009;182(5):2195–2203. doi: 10.1016/j.juro.2009.07.016. [DOI] [PubMed] [Google Scholar]
  • 5.Cruz-Jentoft A.J., Bahat G., Bauer J., Boirie Y., Bruyère O., Cederholm T., et al. Sarcopenia: revised european consensus on definition and diagnosis. Age Ageing. 2019;48(1):16–31. doi: 10.1093/ageing/afy169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Argilés J.M., Busquets S., Stemmler B., López-Soriano F.J. Cancer cachexia: understanding the molecular basis. Nat Rev Cancer. 2014;14(11):754–762. doi: 10.1038/nrc3829. [DOI] [PubMed] [Google Scholar]
  • 7.Prado C.M., Purcell S.A., Alish C., Pereira S.L., Deutz N.E., Heyland D.K., et al. Implications of low muscle mass across the continuum of care: a narrative review. Ann Med. 2018;50(8):675–693. doi: 10.1080/07853890.2018.1511918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Fukushima H., Nakanishi Y., Kataoka M., Tobisu K., Koga F. Prognostic significance of sarcopenia in muscle-invasive and metastatic bladder cancer: a systematic review and meta-analysis. Int J Urol. 2021;28(8):831–845. doi: 10.1111/iju.14547. [DOI] [Google Scholar]
  • 9.Antoun S., Borget I., Lanoy E. Impact of sarcopenia on the prognosis of advanced cancer. Ann Oncol. 2013;24(10):2518–2521. doi: 10.1093/annonc/mdt199. [DOI] [PubMed] [Google Scholar]
  • 10.Kamat A.M., Hahn N.M., Efstathiou J.A., Lerner S.P., Malmström P.U., Choi W., et al. Bladder cancer. Lancet. 2016;388(10061):2796–2810. doi: 10.1016/S0140-6736(16)30512-8. [DOI] [PubMed] [Google Scholar]
  • 11.Gomez-Perez S.L., Haus J.M., Sheean P., Patel B., Mar W., Chaudhry V., et al. Measuring abdominal circumference and skeletal muscle from a single cross-sectional computed tomography image: a step-by-step guide for clinicians using national institutes of health ImageJ. JPEN J Parenter Enteral Nutr. 2016;40(3):308–318. doi: 10.1177/0148607115604149. Epub 2015 Sep 21. Erratum in: JPEN J Parenter Enteral Nutr. 2016 Jul;40(5):742-3. doi: 10.1177/0148607116648918. PMID: 26392166; PMCID: PMC4767633. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Mourtzakis M., Prado C.M., Lieffers J.R., Reiman T., McCargar L.J., Baracos V.E. A practical and precise approach to quantification of body composition in cancer patients using computed tomography images acquired during routine care. Appl Physiol Nutr Metab. 2008;33(5):997–1006. doi: 10.1139/H08-075. [DOI] [PubMed] [Google Scholar]
  • 13.Molwitz I., Leiderer M., McDonough R., Fischer R., Ozga A.K., Ozden C., et al. Skeletal muscle fat quantification by dual-energy computed tomography in comparison with 3T MR imaging. Eur Radiol. 2021;31(10):7529–7539. doi: 10.1007/s00330-021-07820-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Martin L., Birdsell L., MacDonald N., Reiman T., Clandinin M.T., McCargar L.J., et al. Cancer cachexia in the age of obesity: skeletal muscle depletion is a powerful prognostic factor, independent of body mass index. J Clin Oncol. 2013;31:1539–1547. doi: 10.1200/JCO.2012.45.2722. [DOI] [PubMed] [Google Scholar]
  • 15.Fearon K., Strasser F., Anker S.D., Bosaeus I., Bruera E., Fainsinger R.L., et al. Definition and classification of cancer cachexia: an international consensus. The Lancet Oncology. 2011;12:489–495. doi: 10.1016/S1470-2045(10)70218-7. [DOI] [PubMed] [Google Scholar]
  • 16.Bahat G., Turkmen B.O., Aliyev S., Catikkas N.M., Bakir B., Karan M.A. Cut-off values of skeletal muscle index and psoas muscle index at L3 vertebra level by computerized tomography to assess low muscle mass. Clin Nutr. 2021;40:4360–4365. doi: 10.1016/J.CLNU.2021.01.010. [DOI] [PubMed] [Google Scholar]
  • 17.Hamaguchi Y., Kaido T., Okumura S., Kobayashi A., Hammad A., Tamai Y., et al. Proposal for new diagnostic criteria for low skeletal muscle mass based on computed tomography imaging in Asian adults. Nutrition (Burbank, Los Angeles County, Calif) 2016;32:1200–1205. doi: 10.1016/J.NUT.2016.04.003. [DOI] [PubMed] [Google Scholar]
  • 18.Psutka S.P., Carrasco A., Schmit G.D., Moynagh M.R., Boorjian S.A., Frank I., et al. Sarcopenia in patients with bladder cancer undergoing radical cystectomy: impact on cancer-specific and all-cause mortality. Eur Urol Oncol. 2019;2(5):458–466. doi: 10.1002/cncr.28798. [DOI] [PubMed] [Google Scholar]
  • 19.Fukushima H., Yokoyama M., Koga F. Sarcopenia as a prognostic biomarker of muscle-invasive bladder cancer treated with bladder-preserving therapy. Transl Androl Urol. 2023;12(4):542–553. [Google Scholar]
  • 20.Smith A.B., Deal A.M., Yu H., Boyd B., Wallen E.M., Pruthi R.S., et al. Sarcopenia as a predictor of complications and survival following radical cystectomy. J Urol. 2014;191(5):1714–1720. doi: 10.1016/j.juro.2013.12.047. [DOI] [PubMed] [Google Scholar]
  • 21.Miyake M., Morizawa Y., Hori S., Tatsumi Y., Nakai Y., Anai S., et al. Clinical impact of sarcopenia on oncologic outcomes after radical cystectomy for muscle-invasive bladder cancer. Urol Oncol. 2017;35(12):671.e1–671.e9. [Google Scholar]
  • 22.Cruz-Jentoft A.J., Bahat G., Bauer J., Boirie Y., Bruyère O., Cederholm T., et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(1):16–31. doi: 10.1093/ageing/afy169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Kato T., Hatakeyama S., Iwamoto H., Narita T., Fujita T., Okamoto T., et al. Impact of preoperative sarcopenia on oncologic outcomes of patients with non–muscle-invasive bladder cancer treated with transurethral resection and intravesical therapy. Int J Urol. 2020;27(10):873–881. [Google Scholar]
  • 24.Ito H., Takemura K., Nakano T., Watanabe H., Kinoshita H., Tanaka M., et al. Association between skeletal muscle index and recurrence in patients with non–muscle-invasive bladder cancer. BMC Urol. 2021;21(1):118. [Google Scholar]
  • 25.Ferro M., Vartolomei M.D., Russo G.I., Cantiello F., Farhan A.R.A., Terracciano D., et al. An increased body mass index is associated with a worse prognosis in patients administered BCG immunotherapy for T1 bladder cancer. World J Urol. 2019;37(3):507–514. doi: 10.1007/s00345-018-2397-1. [DOI] [PubMed] [Google Scholar]
  • 26.Liu P., Chen S., Gao X., Liang H., Sun D., Shi B., et al. Preoperative sarcopenia and systemic immune-inflammation index can predict response to intravesical Bacillus Calmette-Guérin instillation in patients with non-muscle invasive bladder cancer. Front Immunol. 2022;13:1032907. doi: 10.3389/fimmu.2022.1032907. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Fukushima H., Nakanishi Y., Koga F. Sarcopenia and bacillus Calmette–Guérin therapy outcomes in non–muscle-invasive bladder cancer: impact on tolerance and recurrence. Urol Oncol. 2022;40(8):342.e1–342.e8. [Google Scholar]
  • 28.Prado C.M., Wells J.C.K., Smith S.R., Stephan B.C.M., Siervo M. Sarcopenic obesity: a critical appraisal of the current evidence. Clin Nutr. 2022;41(5):1112–1123. doi: 10.1016/j.clnu.2012.06.010. [DOI] [PubMed] [Google Scholar]
  • 29.Caan B.J., Cespedes Feliciano E.M., Kroenke C.H. The obesity paradox in cancer: how important is muscle? Annu Rev Nutr. 2018;38:357–379. doi: 10.1146/annurev-nutr-082117-051723. [DOI] [PubMed] [Google Scholar]
  • 30.Gonzalez M.C., Pastore C.A., Orlandi S.P., Heymsfield S.B. Obesity paradox in cancer: new insights provided by body composition. Am J Clin Nutr. 2014;99(5):999–1005. doi: 10.3945/ajcn.113.071399. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mmc1.pdf (2.8MB, pdf)
mmc2.xlsx (11.5KB, xlsx)

Data Availability Statement

The study data will be made available on request to the corresponding author (renate.pichler@i-med.ac.at). The R code is available from the GitHub repository (https://github.com/PiotrTymoszuk/sarcoBLCA).

Articles from The Journal of Nutrition, Health & Aging are provided here courtesy of Elsevier

RESOURCES