Grading the evidence on the effects of exercise interventions in children and adolescents during and beyond cancer treatment: an umbrella review of systematic reviews with meta-analyses Authors: Samuel Ruiz-Campos, Joaquin Sanchez-Gomez, David Ruiz-González, Ana León-Rodríguez, Juan Luis Sánchez Sánchez, Andrea Rodriguez-Solana, Alberto Soriano-Maldonado, Borja Martínez-Tellez, Pedro L Valenzuela, Javier S Morales Published: 2026 ABSTRACT AbstractObjectiveTo synthesise the evidence on the effects of exercise interventions in children and adolescents during and beyond cancer treatment, and to evaluate the methodological quality and certainty of this evidence.DesignUmbrella review of systematic reviews with meta-analyses.Data sourcePubMed, Web of Science, Scopus, SPORTDiscus and CINAHL were searched from inception to January 2026.Eligibility criteria for selecting studiesSystematic reviews with meta-analysis of controlled trials (randomised or non-randomised) evaluating exercise interventions in children and adolescents (≤19 years) during and beyond cancer treatment. Reviews without quantitative synthesis, those including only observational studies, or interventions where the isolated effect of exercise could not be determined were excluded.Results19 systematic reviews with meta-analyses, including 53 primary studies (n=2361 participants), were identified, yielding 80 effect estimates. Attending to the most comprehensive meta-analyses, exercise induced significant benefits on cardiorespiratory fitness (mean difference (MD) 6.92% (95% CI 1.01% to 12.82%)), cognitive function (standardised mean difference (SMD) 0.26 (95% CI 0.08 to 0.44)) and cognitive performance (SMD 0.41 (95% CI 0.17 to 0.65)), with moderate certainty of evidence. Exercise also induced significant benefits on muscle strength (SMD 0.81 (95% CI 0.05 to 1.57)), functional mobility (MD −1.17 s (95% CI −1.86 to −0.49)) and cancer-related fatigue (SMD −0.62 (95% CI −1.21 to −0.03)), although the certainty was low or very low. These effects were observed across meta-analyses focused exclusively on studies conducted during treatment (n=26), after treatment (n=3) or both (n=51). In contrast, no significant effects of exercise were observed for depression, bone mineral density, health-related quality of life, anthropometry or other clinical outcomes (such as mortality, relapse risk and withdrawal rates).ConclusionExercise appears to be a beneficial intervention in childhood cancer for improving some physical, cognitive and psychosocial outcomes. However, more high-quality research is warranted to strengthen clinical recommendations and inform personalised exercise prescriptions, given that the certainty of evidence remains variable and overall low across outcomes.PROSPERO registration numberCRD420251031412.Keywords: Physical activity, Exercise training, Child Health, Medical Oncology WHAT IS ALREADY KNOWN ON THIS TOPIC Children and adolescents often show impaired physical fitness, cognitive difficulties, fatigue and reduced quality of life during and beyond cancer treatment. Children and adolescents often show impaired physical fitness, cognitive difficulties, fatigue and reduced quality of life during and beyond cancer treatment. Exercise interventions have been proposed as a supportive therapy to mitigate these adverse effects. Exercise interventions have been proposed as a supportive therapy to mitigate these adverse effects. Existing systematic reviews and meta-analyses report potential benefits, but results are heterogeneous and inconsistent, and the certainty of evidence remains unknown. Existing systematic reviews and meta-analyses report potential benefits, but results are heterogeneous and inconsistent, and the certainty of evidence remains unknown. WHAT THIS STUDY ADDS This umbrella review shows that exercise interventions improve cardiorespiratory fitness, muscle strength, functional mobility, cancer-related fatigue and cognitive-related outcomes, although the certainty of evidence was low for some outcomes. This umbrella review shows that exercise interventions improve cardiorespiratory fitness, muscle strength, functional mobility, cancer-related fatigue and cognitive-related outcomes, although the certainty of evidence was low for some outcomes. No significant effects were found for depression, bone health, quality of life, anthropometry or other clinical outcomes. No significant effects were found for depression, bone health, quality of life, anthropometry or other clinical outcomes. Only 10% of effect estimates were supported by moderate-certainty evidence, underscoring the need for well-designed, adequately powered trials. Only 10% of effect estimates were supported by moderate-certainty evidence, underscoring the need for well-designed, adequately powered trials. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Exercise interventions should be considered a supportive strategy in childhood cancer care. Exercise interventions should be considered a supportive strategy in childhood cancer care. Larger, well-designed trials are needed to strengthen the certainty of evidence and optimise exercise prescriptions in this population. Larger, well-designed trials are needed to strengthen the certainty of evidence and optimise exercise prescriptions in this population. Introduction Survival rates for childhood cancer have markedly improved over recent decades, rising from approximately 58% in the mid-1970s to over 85% in high-income countries in recent years.1 2 However, children and adolescents during and beyond cancer treatment (herein referred to as childhood cancer survivors (CCS) for the sake of conciseness) experience persistent adverse effects from the disease and its therapy,3 which compromise long-term health and quality of life. Indeed, it has been estimated that their life expectancy is on average 9–16 years shorter than that of individuals without a history of cancer.2 Compared with their peers without cancer, CCS show significantly lower cardiorespiratory fitness, mobility and muscle strength, as well as performance limitations in daily life activities.4 5 CCS also show an increased prevalence of cardiovascular risk factors, including impaired cardiac function, notably due to anthracycline-induced cardiotoxicity, reduced high-density lipoprotein-cholesterol levels and excess abdominal adiposity.3 Fatigue, cognitive impairment and adverse psychosocial outcomes such as depression and reduced health-related quality of life (HRQoL) are also prevalent in CCS.6,10 Therefore, unravelling adjuvant strategies to improve health-related outcomes in CCS is of major clinical and public health interest. Physical exercise is a promising intervention as part of a strategy to mitigate a broad range of impairments in CCS.11 However, most primary studies in childhood cancer are limited by small sample sizes, heterogeneity across interventions and outcomes, and varying methodological quality,11,13 which hinder robust conclusions. Although several systematic reviews and meta-analyses have synthesised the effects of exercise interventions in CCS, their findings remain heterogeneous and inconsistent, resulting in mixed results for several health outcomes (eg, cardiorespiratory fitness (CRF), bone health, body mass index (BMI) and HRQoL) across reviews.14,17 This might be partly due to the inclusion of studies with heterogeneous intervention designs (eg, supervised vs unsupervised, aerobic vs multicomponent) and treatment timing (during vs post-treatment). To synthesise evidence in this field, Rapti et al conducted an umbrella review that included systematic reviews both with and without meta-analyses and concluded that exercise interventions may confer beneficial effects across several health outcomes in paediatric cancer populations.18 While this work provided an overview of the literature, the absence of quantitative synthesis across all included reviews and the lack of formal assessment of certainty of evidence limit its ability to inform the magnitude, robustness and clinical relevance of observed effects. As highlighted in recent methodological guidance for umbrella reviews,19 evaluating the certainty of evidence is essential to support evidence-based recommendations and to identify meaningful research gaps. In addition, an expanding body of meta-analytic evidence has since become available,1214 15 17 20,31 enabling a reassessment of the effects of exercise training in CCS across a broader range of health-related outcomes. A comprehensive and methodologically rigorous synthesis is therefore warranted to consolidate and critically appraise the available meta-analytic evidence and to evaluate the strength and certainty of findings across key health domains. Accordingly, the present umbrella review aims to synthesise current meta-analytic evidence on the effects of exercise interventions on physical, cognitive, psychosocial and clinical outcomes in CCS, and to assess the overall certainty of evidence using established appraisal frameworks. Methods Protocol and registration This umbrella review was conducted in accordance with the Preferred Reporting Items for Overviews of Reviews statement32 and the PERSiST guidance for implementing Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) in Exercise, Rehabilitation, Sport medicine and SporTs science.33 The protocol was prospectively registered in PROSPERO (registration number: CRD420251031412). Information sources and search strategy A comprehensive literature search was conducted in PubMed, Web of Science, Scopus, SPORTDiscus and CINAHL, covering all records from inception to 17 January 2026. Two authors (JSG and SRC) independently conducted a systematic search using terms related to “cancer”, “children/adolescents”, “exercise”, “physical activity”, “systematic review”, and “meta-analysis” (full search strategy in online supplemental material 1). Reference lists of included reviews were screened for additional publications. Eligibility criteria Two authors (JSG and SRC) independently screened all records. The selection process was conducted in two phases: first, titles and abstracts were reviewed to exclude clearly irrelevant studies; second, the full texts of potentially eligible studies were assessed against the inclusion criteria. Disagreements at any stage were resolved through discussion or, if necessary, in consultation with a third reviewer (JSM). Inclusion and exclusion criteria were defined according to the Population, Intervention, Comparison, Outcome and Study design framework.34 The umbrella review was limited to studies that met the criteria shown in table 1. Table 1Eligibility criteria based on the Population, Intervention, Comparison, Outcome and Study design approachCategoryInclusion criteriaExclusion criteriaPopulationStudies conducted in children and adolescents (individually or with a mean age of ≤19 years, in accordance with the WHO definition of adolescence68 during and beyond cancer treatment.Studies including only adults or with a mean age of >19 years.InterventionStructured exercise interventions in childhood cancer survivors.Interventions that did not involve structured physical exercise. Interventions combined with other therapies in which the effect of exercise could not be isolated.ComparatorA group that did not undertake any exercise intervention and did not receive any other intervention differing from the exercise group (eg, psychological or nutritional support).Comparisons involving additional interventions beyond exercise (other than usual care).OutcomeReported at least one physical, cognitive, psychosocial or clinical outcome.Studies not reporting outcomes within these categories.Study designSystematic reviews with meta-analysis of controlled trials (randomised or non-randomised).Narrative reviews or systematic reviews without meta-analysis. Meta-analyses including only observational studies. Eligibility criteria based on the Population, Intervention, Comparison, Outcome and Study design approach Data collection process From each included systematic review, two authors (JSG and SRC) independently extracted the following data: (1) first author, publication year and population characteristics; (2) type, frequency, duration and intensity of the exercise intervention; (3) outcome categories and measurement tools and (4) main results, pooled effect sizes and corresponding 95% CIs. Any disagreements were resolved in discussion with a third author (JSM). Quality assessment The methodological quality of the included reviews was assessed independently by two authors (JLSS and JSM) using the AMSTAR-2 (A MeaSurement Tool to Assess systematic Reviews 2).35 Any disagreements were resolved through discussion with a third author (PLV). AMSTAR-2 includes 16 items and provides an overall rating of confidence in the results of the review (high, moderate, low or critically low) based on the presence of critical and non-critical weaknesses in methodological design and reporting. Statistical analysis To quantify the overlap among the included meta-analyses, we calculated the Corrected Covered Area (CCA), which reflects the degree to which primary studies are shared across reviews.36 A CCA of 0% indicates no overlap, while 100% indicates complete redundancy. Following established thresholds, overlap was classified as slight (0–5%), moderate (6–10%), high (11–15%) or very high (>15%).37 Additionally, a visual network analysis of shared primary studies was performed using VOSviewer (V.1.6.20) to illustrate overlap patterns among the meta-analyses. For each meta-analysis, we extracted the reported effect sizes and corresponding 95% CI of the individual primary studies. To ensure consistency across analyses, we applied a uniform random-effects model using the DerSimonian and Laird method.38 Pooled effect sizes were reported with their corresponding 95% CIs. Statistical significance was set at p<0.05 (two-sided tests). Between-study heterogeneity was assessed using the I² statistic,39 with values of ≥25%, ≥50% and ≥75% interpreted as low, moderate and high heterogeneity, respectively. Potential small-study effects were evaluated using Egger’s regression asymmetry test,40 with a p<0.10 on this test taken as statistical evidence of the presence of small-study effects. All analyses were conducted using the metafor package in R (V.4.5.0). Assessment of the certainty of the evidence The certainty for each effect estimate was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. This assessment considered the following five domains: risk of bias, inconsistency, imprecision, indirectness and publication bias.41 Each outcome was rated as high, moderate, low or very low certainty, and justifications for downgrading were recorded.41 Two reviewers (JSM and PLV) independently applied the GRADE framework, and any disagreements were resolved by consensus or through consultation with a third reviewer (JLSS). Equity, diversity and inclusion statement This study was conducted by a multidisciplinary team of researchers from different institutions across Spain and France, including early-career and senior scientists, with diverse expertise in exercise science, oncology and paediatric research. The team is committed to promoting equity, diversity and inclusion in research practices and collaborations. All contributions were valued equally, and decisions were made collectively to ensure transparency and fairness. Results The literature search retrieved 1727 records, of which 19 systematic reviews with meta-analysis (including 53 unique primary studies, n=2361 participants) met the inclusion criteria (figure 1).12,1720 The temporal trend of the included reviews and their primary studies is shown in figure 2. A network map of the included reviews and their overlapping primary studies is shown in figure 3. Flow diagram of the systematic search. Publication trends of primary studies and meta-analyses over time. The figure illustrates the number of primary studies published per year (grey bars) and the cumulative number of meta-analyses published during the same period (green line). Network map of included meta-analyses and their overlapping primary studies. Each rectangular node represents a meta-analysis, while circles correspond to primary studies included in one or more meta-analyses. Meta-analyses are grouped into four colour-coded clusters (red, green, blue and yellow) based on shared primary studies. Meta-analyses within the same colour cluster are more closely connected. Proximity between nodes reflects the degree of overlap: meta-analyses positioned closer together share more primary studies. The colour intensity of each rectangle reflects connection density, that is, the number of shared primary studies with other meta-analyses. Circles are colour-coded by sample size, with yellow representing smaller studies and purple representing larger ones. Grey lines indicate primary studies included in multiple meta-analyses. The characteristics of the included systematic reviews with meta-analysis are summarised in online supplemental material 2). Of note, two Cochrane reviews addressed identical outcomes.13 27 To avoid duplication, only the most recent version was retained for the analyses.13 The number of primary studies included per meta-analysis ranged from 1 to 13 (median=3), with total sample sizes per meta-analysis ranging from 14 to 987 participants. Most of the included primary studies were randomised controlled trials (n=43), although some reviews also included non-randomised controlled trials (n=10), and Bernal et al23 included data from two studies (ISRCTN85913225 and NCT04070131) that were unpublished at the time of inclusion. Primary studies included participants undergoing active treatment (n=28), those in post-treatment phases (n=19), or a combination of both (n=4); two studies did not report treatment timing. The exercise interventions comprised aerobic training (n=12), resistance training (n=1), combined aerobic and resistance training (n=12), or other modalities (n=18) such as flexibility, stretching, inspiratory muscle training or plyometric exercise, either alone or combined with aerobic and/or resistance training. Three primary studies integrated exercise with additional interventions (eg, psychosocial or nutritional), and seven did not report sufficient detail on the intervention type. Intervention duration and intensity varied, ranging from inpatient to home-based or community-based programmes, lasting from 4 weeks to 2.5 years (most often 6 weeks to 6 months). The methodological quality of the 19 included reviews was assessed using the AMSTAR-2 tool (online supplemental material 3). Of these, four reviews were rated as having high confidence,13 15 23 27 four as moderate,16 17 25 28 one as low29 and 10 as critically low.1214 20,22 24 26 30 31 42 The main methodological shortcomings observed across the reviews were the lack of justification for the selection of study designs (53%), failure to report funding sources of primary studies (100%), absence of risk of bias impact assessment in the synthesis (53%), inadequate consideration of risk of bias in the interpretation of results (47%) and insufficient assessment of publication bias (47%). A total of 80 effect estimates of exercise interventions were reported. A detailed summary of the methodological characteristics of each meta-analysis, including the timing of exercise relative to cancer treatment (during, after or both), supervision level, type of exercise and study design (randomised controlled trials, non-randomised controlled trials or mixed), is provided in online supplemental material 2). Although a minority of these effect estimates were derived from meta-analyses focused exclusively on studies conducted during active treatment (n=26, 32.5%) or after treatment (n=3, 3.8%), most of them mixed primary studies conducted both during and after treatment (n=51, 63.7%). Of these 80 effect estimates, 34 (42.5%) revealed statistically significant exercise benefits. None of the effect estimates was graded as high certainty. Of the 80, only eight effect estimates (10%) were graded as moderate certainty, whereas 33 effect estimates (41.3%) were rated as low certainty and 39 (48.7%) as very low certainty. Full statistical details for each effect estimate are presented in online supplemental material 2. Additionally, a detailed summary of all outcomes together with their corresponding GRADE assessments and the review overlap analysis is provided in online supplemental materials 4 and 5. Physical performance 11 effect estimates focused on CRF measures (eg, peak oxygen uptake, 6 min walking test or a mix of different tests), with eight (72.7%) showing significant improvements. However, all effect estimates had low to very low certainty, except for two that were graded as moderate certainty. The largest meta-analysis, derived from studies conducted during and/or after treatment, reported a significant benefit (mean difference (MD) 6.92% (95% CI 1.01% to 12.82%), p=0.022; table 2), supported by moderate certainty of the evidence.29 Table 2Summary of the most comprehensive meta-analysis per outcomeStudyOutcomeStudiesTreatment timingParticipantsEffect size (95% CI)P valueGRADEBourdon et al29CRF9During and/or after treatment287MD 6.92% (1.01 to 12.82)0.022ModerateDong et al24Muscle strength8During and/or after treatment470SMD 0.81 (0.05 to 1.57)0.036Very lowLiu et al20Functional mobility4During and/or after treatment123MD −1.17 s (−1.86 to −0.49)<0.001LowDong et al24Physical activity levels8During and/or after treatment439SMD 0.58 (−0.11 to 1.27)0.102Very lowWang et al21Cancer-related fatigue8During treatment465SMD −0.62 (−1.21 to −0.03)0.038Very lowBernal et al23Cognitive performance5During and/or after treatment245SMD 0.41 (0.17 to 0.65)0.001ModerateBernal et al23Cognitive function13During and/or after treatment987SMD 0.26 (0.08 to 0.44)0.004ModerateZang et al14Depression2During and/or after treatment80SMD 0.06 (−0.38 to 0.50)0.799Very lowMarmol-Perez et al15BMD8During and/or after treatment341SMD 0.05 (−0.11 to 0.22)0.538Very lowDong et al24HRQoL8During and/or after treatment504SMD 0.53 (−0.04 to 1.10)0.067Very lowDong et al24Anthropometry and body composition5During treatment238SMD 0.02 (−0.21 to 0.25)0.881Very lowMorales et al12Mortality risk3During treatment105OR 0.96 (0.36 to 2.59)0.939LowMorales et al12Relapse risk3During treatment105OR 0.83 (0.25 to 2.77)0.758LowZang et al14Withdrawal rates4During and/or after treatment113RR 0.54 (0.12 to 2.56)0.443Very lowSignificant results are in bold.Colours indicate certainty of evidence according to the GRADE approach (yellow = moderate; orange = low; red = very low).BMD, bone mineral density; CRF, cardiorespiratory fitness; HbA1c, glycated haemoglobin; HRQoL, health-related quality of life; MD, mean difference; RR, risk ratio; SMD, standardised mean difference. Summary of the most comprehensive meta-analysis per outcome Significant results are in bold. Colours indicate certainty of evidence according to the GRADE approach (yellow = moderate; orange = low; red = very low). BMD, bone mineral density; CRF, cardiorespiratory fitness; HbA1c, glycated haemoglobin; HRQoL, health-related quality of life; MD, mean difference; RR, risk ratio; SMD, standardised mean difference. 10 effect estimates focused on muscle strength. When grouped by type, only one effect estimate focused specifically on isometric strength, reporting a significant improvement.31 One effect estimate targeted upper-limb dynamic strength, and another one targeted lower-limb dynamic strength14; only the latter showed a statistically significant benefit. The remaining seven effect (70%) estimates revealed overall or region-specific strength (eg, knee, ankle and back), and four (57.1%) reported significant positive effects. The largest meta-analysis, which focused on overall strength and was derived from studies conducted during and/or after treatment, found a significant effect (standardised mean difference (SMD) 0.81 (95% CI 0.05 to 1.57), p=0.036; table 2), although with very low certainty.14 Additionally, three effect estimates focused on respiratory muscle strength. None reported significant effects. Six effect estimates focused on functional mobility outcomes, of which four (66.7%) showed significant improvements following exercise interventions. These were based on performance tests such as the Timed Up and Go and Timed Up and Down Stairs (TUDS). The largest meta-analysis, drawing on data from interventions implemented during and/or after treatment, reported a significant improvement in functional mobility (MD −1.17 s (95% CI −1.86 to −0.49), p<0.001; table 2), although the certainty of the evidence was rated as low.20 Two effect estimates focused on flexibility outcomes (ankle dorsiflexion).13 One focused on passive ankle dorsiflexion and showed a significant effect supported by moderate certainty, while the other focused on active range of motion and found no effect, with low certainty. One effect estimate focused on a composite outcome combining several performance domains (CRF, muscle strength, functional mobility and flexibility) and found an overall beneficial effect (SMD 0.66 (95% CI 0.17 to 1.11), p=0.008), based on studies conducted during and/or after treatment, although the certainty was very low.42 Physical activity levels Four estimates focused on the effects of exercise interventions on physical activity levels, of which two (50%) showed significant benefits. The largest meta-analysis, derived from studies conducted during and/or after treatment, found no significant effect (SMD 0.58 (95% CI −0.11 to 1.27), p=0.102; table 2), with very low certainty of the evidence.24 Additionally, Dong et al24 conducted two subanalyses based on intervention duration. While the analysis for interventions shorter than 6 months showed no significant effect, the subanalysis restricted to interventions lasting 6 months or longer did report a significant benefit, although certainty remained critically low. Cancer-related fatigue Six effect estimates focused on cancer-related fatigue in general, of which four (66.7%) showed significant beneficial effects. Two additional effect estimates focused on specific fatigue-related items (ie, sleep/rest, cognitive fatigue).13 The largest meta-analysis, based on studies conducted during treatment, revealed a significantly beneficial effect on cancer-related fatigue (SMD −0.62 (95% CI −1.21 to −0.03), p=0.038; table 2), although with very low certainty of the evidence.21 Cognitive-related outcomes Two effect estimates focused on cognitive-related outcomes. Both (100%) reported significant beneficial effects, supported by moderate-certainty evidence. Specifically, exercise interventions were reported to improve cognitive performance—objectively assessed through standardised neuropsychological or academic tests measuring domains such as attention, memory and executive function—(Hedges’ g 0.41 (95% CI 0.17 to 0.65), p=0.001; table 2), derived from studies conducted during and/or after treatment.23 Similarly, cognitive function—assessed through validated patient- or proxy-reported questionnaires reflecting perceived cognitive abilities in daily life—also improved (Hedges’ g 0.26 (95% CI 0.08 to 0.44), p=0.004; table 2), drawing on data from interventions implemented during and/or after treatment.23 Mental health Two effect estimates focused on mental health outcomes, of which only one (50%) reported significant benefits. The largest meta-analysis, which focused on depressive symptoms, found no significant exercise benefits (SMD 0.06 (95% CI –0.38 to 0.50), p=0.799; table 2), based on studies conducted during and/or after treatment.14 The certainty of the evidence was rated as very low. The other effect estimate focused on a composite outcome combining several psychological outcomes (negative mood, interpersonal problems, self-esteem, ineffectiveness, anhedonia and fatigue) and found a significant beneficial effect (SMD 0.61 (95% CI 0.17 to 1.05), p=0.006), derived from studies conducted during treatment. However, the certainty was critically low.42 Bone health Eight effect estimates focused on bone mineral density (BMD), with only one (12.5%) reporting statistically significant benefits. The most comprehensive meta-analysis, which pooled data from interventions delivered both during and/or after treatment, found no significant effect of exercise on BMD (SMD 0.05 (95% CI –0.11 to 0.22), p=0.538; table 2).15 The certainty of the evidence was graded as very low. Health-related quality of life A total of 12 effect estimates focused on HRQoL. Seven of these estimates focused on general HRQoL, of which two (28.5%) reported statistically significant improvements. One effect estimate assessed cancer-specific HRQoL,13 two additional effect estimates assessed cancer-specific HRQoL from the perspective of parents13 and two targeted specific HRQoL domains, such as social or school functioning30—none of which reported significant effects. One additional effect estimate evaluated self-efficacy,24 showing a large effect size, though results were not statistically significant and the certainty of the evidence was critically low. The largest meta-analysis, which examined general HRQoL, found no significant exercise effect (SMD 0.53 (95% CI −0.04 to 1.10), p=0.067; table 2), drawing on data from interventions implemented during and/or after treatment, with very low certainty.24 Cardiometabolic outcomes Five effect estimates focused on cardiometabolic outcomes, including anthropometry and body composition measures, haemoglobin and glycated haemoglobin. Among these, only one effect estimate (20%) showed a significant improvement. The most comprehensive meta-analysis, which focused on anthropometry and body composition and derived from studies conducted during treatment, reported no significant exercise effect (SMD 0.02 (95% CI −0.21 to 0.25), p=0.881; table 2), with very low certainty of the evidence.24 One effect estimate focused on a composite outcome combining several cardiometabolic markers (BMI, haemoglobin and glycated haemoglobin) and found no significant exercise effect (SMD 0.36 (95% CI −0.06 to 0.73), p=0.093), which pooled data from interventions delivered during treatment, with very low certainty.42 Clinical outcomes Three effect estimates focused on clinical outcomes, including mortality risk, relapse risk and treatment withdrawal rates. None (0%) showed statistically significant effects of exercise interventions. Specifically, mortality risk (OR 0.96 (95% CI 0.36 to 2.59), p=0.939; low certainty of the evidence; table 2), based on studies conducted during treatment,12 and relapse risk (OR 0.83 (95% CI 0.25 to 2.77), p=0.758; low certainty; table 2), also derived from interventions implemented during treatment,12 showed no significant effects. Likewise, withdrawal rates (risk ratio 0.54 (95% CI 0.12 to 2.56), p=0.443; very low certainty; table 2), from studies conducted during and/or after treatment, were not significantly affected by exercise.14 Discussion This umbrella review provides an overview and synthesis of the available evidence on the effects of exercise interventions on physical, cognitive, psychosocial and clinical outcomes in CCS. We observed significant effects for CRF, muscle strength, functional mobility, cancer-related fatigue and cognitive-related outcomes, but no significant effects were found for physical activity levels, mental health, BMD, HRQoL, anthropometry and body composition, or other clinical outcomes. These effects were observed across meta-analyses focused exclusively on studies conducted during treatment (n=26), after treatment (n=3) or both (n=51). However, while 42.5% of the 80 analysed effect estimates reported significant exercise benefits, only eight (10%) were supported by moderate-certainty evidence and none by high certainty. This gap between positive trends and low confidence reflects methodological limitations that weaken the robustness of clinical recommendations. The present umbrella review supports the beneficial effects of exercise training across multiple outcomes in CCS, in line with the conclusions of the umbrella review by Rapti et al.18 However, the present work provides a methodologically distinct and substantively updated appraisal of the evidence base. Unlike the review by Rapti et al,18 which included systematic reviews both with and without meta-analyses, we deliberately restricted inclusion to reviews conducting meta-analyses of controlled trials, thereby prioritising quantitative synthesis and internal validity. In addition, our umbrella review synthesises a substantially expanded body of evidence, incorporating 16 meta-analyses not included in the umbrella review by Rapti et al and offering broader coverage of primary studies (25 new primary studies, 22 of which were randomised controlled trials), participants and clinically relevant outcomes (such as mortality, relapse risk and withdrawal rates). Importantly, this is the first umbrella review in this field to formally evaluate the certainty of evidence using the GRADE framework, alongside methodological appraisal with AMSTAR-2, an essential step for informing clinically meaningful recommendations and identifying research gaps, as emphasised in recent methodological guidance.19 Collectively, this approach moves beyond summarising effects to provide a rigorous and clinically informative synthesis of both the magnitude and robustness of evidence supporting exercise interventions in CCS. Some of the most consistent benefits of exercise were observed for cognitive-related outcomes, supported by moderate-certainty evidence. These findings are consistent with previous research suggesting that exercise might enhance brain health by reducing systemic inflammation in CCS,43 and in other populations, such apparently healthy young adults, exercise has been reported to increase cerebral blood flow and neurotrophic factor release (eg, brain-derived neurotrophic factor).44 45 These biological pathways are considered relevant in CCS, particularly in those treated for brain tumours, among whom neurocognitive deficits have been reported in 40–100% of cases.46 Emerging neuroimaging data also suggest that exercise may induce favourable structural brain changes in this population,47 48 further supporting its neuroprotective role. Exercise was also associated with improvements in CRF and reductions in cancer-related fatigue. In the largest pooled analysis, exercise improved VO₂peak (oxygen consumption) by ∼7%, supported by moderate-certainty evidence.29 This may be clinically meaningful given the impairment in CRF typically seen in CCS, even several years after treatment.49 For instance, a recent meta-analysis found that CCS have a VO₂peak approximately 7.1 mL/kg/min lower than age-matched peers without cancer,49 a deficit that may increase their risk of cardiovascular disease and all-cause mortality.50 Similarly, cancer-related fatigue improved in most meta-analyses.14 16 20 21 Although the exact mechanisms underlying these benefits remain unknown, increased proinflammatory cytokines trigger cancer-related fatigue in CCS,51 which can be alleviated by exercise.43 Improvements in CRF also align with well-established physiological adaptations to exercise in healthy and clinical populations, such as increased stroke volume and cardiac output, better oxygen utilisation in muscles and improved overall endurance.52 Briefly, exercise interventions with a frequency of ≥3 sessions/week, duration <45 min per session and total programme length <12 weeks appear to be more effective than lower-frequency or longer-duration protocols in alleviating cancer-related fatigue.21 Consistent with the observed improvements in CRF and cancer-related fatigue, we found that exercise led to gains in muscle strength and functional mobility in daily activities. This is relevant because other studies have reported muscle weakness occurring as early as during treatment and persisting for years after its completion.4 53 For example, Deisenroth et al reported strength deficits of up to 57% in the lower limbs within the first month after diagnosis in children and adolescents with different types of cancer (including leukaemia, bone and brain tumours), with links to poorer quality of life.54 Similarly, Valenzuela et al found comparable reductions in muscle strength (∼50%) and functional mobility (∼23%, assessed through the TUDS) among CCS of both haematological and solid tumours compared with cancer-free controls assessed more than 4 years after treatment completion.4 This indicates that muscle weakness and mobility limitations are consistent consequences of cancer and its treatment, regardless of diagnosis. In this regard, our umbrella review shows that exercise seems to improve performance in functional tests such as the TUDS in most meta-analyses, indicating gains in agility, balance and coordination. These findings support the role of exercise in mitigating early declines in strength and preserving daily functioning. In contrast to the above outcomes, our review found no significant effects of exercise on certain endpoints (notably depression, HRQoL, BMD and anthropometry). Several clinical and methodological factors are likely to account for these null findings. For instance, interventions were not specifically designed to target depression or HRQoL, as their primary objective was to improve physical fitness.55 56 Moreover, depressive symptoms in CCS are often transient and tend to decline naturally over time as children and families adapt to the diagnosis and treatment. A recent longitudinal study reported that although nearly half of children with cancer met criteria for anxiety or depressive disorders within the first year after diagnosis, the prevalence of depression decreased to about 9% at 1 year follow-up.9 In addition, although previous systematic reviews and meta-analyses have reported exercise benefits on HRQoL in CCS,14 20 some studies have observed only domain-specific improvements (eg, better emotional functioning after a 6-week supervised in-hospital exercise intervention) without a substantial overall change in total HRQoL.57 On the other hand, the lack of effect on body composition and anthropometry can be due to the catabolic effects of cancer therapies (eg, cranial and pelvic irradiation, glucocorticoids) and subsequent endocrinopathies,58 which a typical exercise regimen alone (eg, with no concomitant nutritional intervention) may struggle to counteract. Consistent with this interpretation, attempts to prevent treatment-related bone loss with exercise have largely been unsuccessful,15 20 suggesting that the prescribed exercise type might not be appropriate or the intensity required to modify bone parameters was not achievable.59,61 A finding warranting careful interpretation is the effect of exercise on clinical outcomes. While recent evidence from adult CCS has shown that higher physical activity levels are associated with a lower risk of subsequent neoplasms,62 and exercise interventions have also been linked to improved survival and reduced recurrence rates in adults with colon cancer63; our findings were not able to confirm such associations. Nevertheless, none of the effect estimates related to mortality, relapse risk or treatment withdrawal showed statistically significant effects; the absence of adverse trends provides indirect support for the safety of exercise interventions. This is particularly relevant in childhood cancer, where safety concerns may limit the implementation of supportive therapies. However, the rate of other adverse events during or after exercise sessions could not be analysed, which limits drawing conclusions on the safety of exercise interventions in CCS. From a clinical perspective, the consistently positive direction of exercise effects observed for some outcomes provides reassurance that exercise is likely beneficial for CCS. However, given the low certainty of the evidence in many domains, more research is still warranted. Moreover, our review highlights important research gaps. The small sample size and single-centre nature of most trials can limit the generalisability of the findings. In this regard, ongoing large-scale studies such as the FORTEe trial (n=450) are expected to substantially advance our understanding of the efficacy and safety of exercise interventions in children and adolescents with cancer.64 It is also worth noting that most interventions included in the available meta-analyses were relatively short in duration (typically between 6 and 24 weeks) and did not include follow-up assessments after intervention completion, limiting our understanding of the long-term sustainability of exercise-induced benefits. In this regard, a prospective cohort study with up to 5-year follow-up showed that the benefits induced by an exercise intervention applied during treatment (∼22 weeks) were not maintained at 1 year of follow-up or thereafter,65 suggesting that exercise may need to be ideally maintained after treatment ends. This underscores the importance of designing longer or continued interventions beyond treatment cessation. Future studies should therefore assess whether early exercise interventions can promote persistent behavioural changes that translate into healthier lifestyles and improved long-term outcomes among CCS. Additionally, most of the included meta-analyses mixed primary studies with exercise interventions applied during and after treatment, which limits drawing strong conclusions on the role of exercise timing (ie, whether interventions during active treatment, after treatment or across the full continuum of care provide the greatest and most sustainable benefits). Establishing this temporal dimension is essential to inform precise clinical recommendations. Evidence from adult oncology shows that patients who initiate exercise during chemotherapy experience smaller declines in both CRF and muscle strength compared with those who start exercising only after treatment completion.66 Translating this to childhood cancer, exercise interventions should ideally be initiated during the inpatient phase to mitigate early sequelae associated with cancer treatment and support healthier transitions into adulthood. Likewise, studies should seek to optimise exercise prescriptions by comparing different training modalities and delivery formats (eg, aerobic vs resistance, supervised vs unsupervised and in-hospital vs home-based). Understanding how these variables influence health outcomes will help refine evidence-based guidelines, as the present results do not yet allow us to specify an optimal training protocol (eg, a precise frequency or intensity that works best). Notwithstanding, current evidence does support the principle that some activity is better than none. This is in line with current guidelines,67 which encourage regular physical activity for CCS as tolerated, emphasising flexibility and personalisation. Strengths and limitations of the study Several limitations should be noted. We did not evaluate potential adverse effects of exercise interventions, as these outcomes were inconsistently reported, and some important health-related outcomes known to be affected in CCS, such as cardiac function, lipid profile or adiposity, were not analysed in the present umbrella review due to the absence of meta-analyses addressing these specific outcomes in paediatric populations. We were also unable to compare the relative effectiveness of different types of exercise (eg, aerobic vs resistance) or timing (eg, during vs post-treatment), since most reviews pooled heterogeneous interventions without stratified analyses. In addition, although we reanalysed pooled effects and applied standardised quality assessments (GRADE, AMSTAR-2), our conclusions remain constrained by the scope and reporting of the underlying reviews. Despite these limitations, our umbrella review has several strengths, including a strict critical appraisal at both the review level (AMSTAR-2) and outcome level (GRADE), and an overlap analysis (with CCA and visual mapping of primary study overlap). Together, these methodological strengths enhance the robustness of our conclusions. We believe this approach delivers a more balanced and evidence-informed interpretation of exercise effects than was previously available. Conclusion In conclusion, the present umbrella review suggests that exercise is a potentially beneficial intervention for CCS, associated with improvements in CRF, muscle strength, functional mobility, cancer-related fatigue and cognitive-related outcomes. However, the overall certainty of the evidence was generally low and varied across outcomes. These findings underscore the need for high-quality research to strengthen clinical recommendations and inform personalised exercise prescriptions in childhood cancer. In this regard, future studies should also ensure consistent reporting of key intervention features and treatment timing to improve comparability and reproducibility. Supplementary material Footnotes Funding: BMT, JSM and PLV are supported by the grant RYC2022-036473-I, RYC2023-045783-I and RYC2024-048275-I by MCIN/AEI/10.13039/501100011033 and by ESF. JSM is also supported by grant PID2024-159960OA-I00 funded by MCIN/AEI/10.13039/501100011033. SRC was supported by the University of Almeria’s programme for research and knowledge transfer (PPIT-UAL: CPRE2025-044). DRG (grant number: FPU21/04573) was supported by the Ministry of Science, Innovation and Universities of the government of Spain. JSG was supported by the grant PID2022-141442OA-I00 (FPI) funded by MICIU/AEI/10.13039/501100011033 and by 'ESF+'. ALR was supported by a bridge contract (Plan Propio de Investigación, University of Málaga) and by MICIU/AEI/10.13039/501100011033. ARS was supported by the Spanish Ministry of Science, Innovation and Universities (JDC2024-053847-I), funded by MCIU/AEI/10.13039/501100011033 and by the European Social Fund Plus (FSE+). Funders had no role in the study design, data collection, analysis or the decision to publish. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: Not applicable. Data availability free text: All data analysed in this umbrella review were extracted from previously published systematic reviews with meta-analyses. No new primary data were collected. Data supporting the findings of this study are available from the corresponding author on reasonable request. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.