Exercise and mind-body interventions for cancer-related fatigue: A network meta-analysis with exploratory moderator...

To compare multiple exercise and mind-body interventions for cancer-related fatigue (CRF), identify moderators of treatment effects, and examine dose-response patterns to inform integrative symptom management.

Four databases were searched for randomized controlled trials evaluating structured exercise and mind-body interventions for CRF. We performed a multilevel network meta-analysis accounting for correlations in multi-arm trials. Risk of bias and certainty of evidence were assessed using RoB 2 and CINeMA. Exploratory meta-analytic random forest analyses with Shapley additive explanations were used to prioritize candidate moderators, and meta-regression assessed dose-response associations.

We included 205 trials involving 12,629 participants. Low-to-moderate certainty evidence suggested that exercise and mind-body interventions were associated with greater reductions in CRF than control conditions. Several interventions showed favorable estimates, including aquatic exercise (g = -0.59; 95% CrI: -0.72 to -0.46), relaxation training (g = -0.49; 95% CrI: -0.59 to -0.39), dance (g = -0.47; 95% CrI: -0.65 to -0.29), Pilates (g = -0.46; 95% CrI: -0.58 to -0.34), cycling (g = -0.45; 95% CrI: -0.54 to -0.36), and Qigong (g = -0.45; 95% CrI: -0.50 to -0.39), with broadly overlapping credible intervals across modalities. Aquatic exercise and dance were supported by very few arms, and their estimates should not be interpreted as evidence of superiority over better-studied interventions. Dose-response analyses suggested potential benefits across a range of doses, but no clear optimal dose or threshold was identified. Exploratory MARF/SHAP analyses had limited predictive performance and identified reported behavior change techniques as the strongest, but weak, candidate moderator.

Low- to moderate-certainty evidence suggests that multiple exercise and mind-body interventions may help reduce CRF, but current evidence does not support a single clearly superior modality. These interventions may be considered as part of patient-centered supportive care, with attention to patient preference, feasibility, behavioral support, safety, and clinical context.