Comparative effects of exercise and nutritional interventions on muscle mass and body composition in adults with cancer: a systematic review and network meta-analysis of randomized controlled trials

Introduction Loss of muscle mass and changes in body composition are common in adults with cancer. We compared exercise, nutritional, and combined strategies and assessed the certainty and clinical interpretability of the evidence. Methods We searched PubMed, Embase, the Cochrane Library, and Web of Science through April 2026 for randomized controlled trials reporting appendicular skeletal muscle mass, skeletal muscle index, lean mass, body mass index, body weight, or fat mass. Eight nodes were evaluated: standard of care (SOC), active low-intensity control (ALC), resistance exercise (RE), multicomponent exercise (ME), nutrition-only intervention (NOI), exercise-nutrition intervention without resistance exercise as the primary component (ENI), resistance exercise plus nutrition-only intervention (RE + NOI), and comprehensive exercise-nutrition intervention incorporating resistance exercise (ENI + RE). Random-effects network meta-analysis produced mean differences (MDs) with 95% confidence intervals (CIs); certainty was assessed with GRADE/CINeMA. Results Twenty-one trials and eight intervention nodes were included. Most network estimates were of low certainty. ME was associated with lower BMI than SOC (MD −3.53, 95% CI −7.01 to −0.04), but not with statistically significant differences in body weight or fat mass. ENI + RE was associated with lower lean mass than SOC (MD −6.72, 95% CI −12.43 to −1.02); this sparse, low-certainty estimate should not be interpreted as evidence that combined interventions are harmful or ineffective. No intervention showed a consistent advantage across muscle mass outcomes. Conclusion Relative rankings did not establish clinical superiority. Sparse direct evidence, heterogeneous cancer and treatment contexts, varied intervention protocols and measurement methods, and low certainty limit inference. More standardized head-to-head trials are needed.