Home-based, supervised, and mixed exercise intervention on functional capacity and quality of life of colorectal cancer patients: a meta-analysis

This systematic review and meta-analysis of randomized controlled trials tested the effects of home-based, supervised, or mixed exercise interventions on the functional capacity (FC) and quality of life (QoL) in colorectal cancer patients. A literature search was performed using the PubMed, Embase, Cochrane, and Medline databases. Two reviewers screened the literature through March 10, 2021 for studies related to exercise and colorectal cancer. Of the 1161 screened studies in the initial search, 13 studies met the eligibility criteria (home-based = 6 studies; supervised or mixed = 7 studies). Overall, 706 patients were enrolled in the trials, and 372 patients were submitted to home-based, supervised, or mixed exercise intervention. The overall results from the main meta-analysis showed a significant effect regarding supervised or mixed intervention (6 studies; p = 0.002; I2 = 43%; PI 0.41–1.39); however, no significant effect was observed for home-based intervention (5 studies; p = 0.05; I2 = 25%; PI − 0.34–0.76). A sensitivity analysis based on studies with intervention adherence ≥ 80% (home-based = 3 studies; supervised or mixed = 4 studies) revealed that home-based intervention or intervention entirely supervised or with some level of supervision (mixed) are effective in improving the QoL and FC of CRC patients. In summary, this meta-analysis verified that supervised and home-based exercise can modify QoL and FC when intervention adherence ≥ 80%. Regardless of the supervision characteristics, future RCTs are strongly encouraged to provide a detailed description of the exercise variables in physical interventions for CRC prescription. This perspective will allow a refined exercise prescription for patients with CRC, mainly according to their clinical status.

Eligibility criteria. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 25 was used to screen the studies involving CRC patients. Additionally, the PICOS criteria 26 (Table 1) were adopted as the eligibility criteria. Meticulous inclusion consisted of: (a) RCT, preliminary, or feasibility RCTs involving the exercise intervention in CRC patients and published in English; (b) studies involving interventions rather than an only association between the outcomes; (c) studies including other cancer types were considered as long as they provided results of the CRC subgroup; (d) the only intervention consisted of exercise, which was structured and provided detailed information on the intensity or duration; (e) the length of the intervention (e.g. number of weeks) and type (i.e. home-based, supervised, or mixed) was clearly presented; (f) studies that described the exercises performed by patients (e.g. running, walking, cycling); and (g) the outcomes are related to the QoL (e.g. global scores from validated QoL questionnaires, or scores from other questionnaires focused on the parameters that directly affect the QoL of CRC patients, such as fatigue, depression, anxiety, sleep), and/ or FC (e.g. measures of aerobic power-peak oxygen uptake-or capacity-anaerobic threshold, cardiovascular responses, such as the heart rate, total distance covered, and time to exhaustion of physical evaluations). The meticulous exclusion consisted of: (a) studies with animals; (b) grey literature without detailed information, data, or full text; (c) abstracts without full-text; (d) lack of detailed information on the exercises, duration, intensity, or the length of intervention; and (e) studies that did not include pre-post statistical analysis.
Data extraction. Two reviewers (LM and AG) independently screened the studies based on the PICOS and meticulous inclusion/exclusion. The discrepancies were checked by a third reviewer (IR), and a consensus was reached through discussion. Further data on the strength or resistance outcomes were not shown and demonstrated since we were particularly interested in the effects of home-based, supervised, or mixed exercise interventions on the QoL and FC. The follow-up results were not presented or discussed according to the goals of this systematic review.  27 . PEDro is composed of 11 items that yield one point (except for the first item) to the final score and its reliability of items. Additionally, the final score varied from fair to substantial and fair to good, respectively 28 .

Meta-analysis of QoL and FC.
The quantitative data related to the QoL and FC from home-based or supervised exercise interventions were extracted for the meta-analysis and considered as continuous variables. Given that mixed approaches have some level of supervision, we considered these results as supervised interventions for the meta-analysis. A sensitivity analysis was conducted with high intervention adherence and retention rate (≥ 80%). The ≥ 80% was considered as high adherence based on previous suggestions with cancer patients [29][30][31] . Adherence was defined as the number of sessions attended out of total offered. Two studies did not report exercise attendance 32,33 , but informed that the data from patients with low adherence to exercise was not included in the analysis. In these cases, the retention rate (i.e. number of patients that completed the exercise intervention out of total patients allocated to the exercise group) was considered instead of the attendance. The post-intervention means and standard deviations (SDs) were used for comparisons. The effect size index was considered as the standardized mean difference (SMD) and was used for the meta-analysis with 95% confidence intervals (CI). A random effects model was used for the analysis. The Z-value, I 2 , and prediction interval (PI) were adopted for significance analysis, proportion estimation of variance, and variation in the treatment effects, respectively. SMD was plotted against the standard error, and funnel plots were generated to evaluate the asymmetries around the metanalytic summary effect 34 . Two studies were excluded from the meta-analysis because they did not present the standard deviation 31,35 . In three studies 33,36,37 the inferences of the results were the opposite. This means that as the parameter decreases, the outcome is observed to improve. In such cases, the minus sign was inserted into the means of both the intervention and control values. All statistical analyses, except PI, were performed using the Review Manager 5 software (version 5.3, The Nordic Cochrane Centre, Copenhagen, Denmark). The PI was calculated using a comprehensive meta-analysis software (CMA) 38 . The statistical significance was set at p < 0.05.

Results
Literature search and quality assessment. The initial literature search identified 1161 studies (Fig. 1).
Based on the first search, 25 duplicates were identified. After screening for the title, abstract, and full text, 13 studies 17,31-33,35-37,39-44 attained the eligibility criteria ( Table 2). The included studies were published between the years 2003 and 2020. The mean and standard deviation of the PEDro score among the studies was 6.46 ± 0.92. Three studies were classified with a score of 8, one with a score of 7, eight studies with a score of 6, and one study with a score of 5.
Additional records identified through other sources (n = 7)

Included
Records screened at title and abstract (n = 141)

Eligibility
Full-text articles assessed for eligibility (n = 71) Studies included in qualitative synthesis (n = 13) Records identified through database searching (n = 1154) Studies included in quantitative synthesis (meta-analysis) (n = 11) Figure 1. Flow diagram of the preferred reporting items for systematic reviews and meta-analyses (PRISMA).      41 and Pinto et al. 35 respectively. Lee et al. 43 assessed the FC using both 6-MWT and Tecumseh protocols. Table 3 shows the RCTs that conducted home-based exercise interventions for CRC patients. The duration of the intervention ranged from 6 to 16 weeks. Two studies 35,41 reported the prescribed exercise session frequency (i.e., days per week). The remaining studies used the metabolic equivalent tasks (MET) per hour 32,42,43 and the amount of minutes per week 40 to classify the frequency, and details on the exercise distribution over a week were not provided. Christensen et al. 40 instructed the patients to perform exercises for 150 min/week according to their preferences. Likewise, Lee et al. 43 submitted CRC patients to home-based exercises of more than 18 METs hours per week, which should count for approximately 360 min/week. While two studies applied an upper limit of 30-min daily exercise session duration 35,41 , others used the number of steps as a goal (i.e., > 10.000) 32,42,43 . With one exception 40 , all the studies prescribed the intensity based on a percentage (range 64-76%) of the maximum heart rate (HR max ). Two studies applied only aerobic exercises 40,41 , and the remaining conducted both aerobic and resistance exercises 32,35,42,43 .

Effects of home-based exercise intervention on quality of life and functional capacity.
A meta-analysis of the QoL and FC is shown in Fig. 2. No significant effect of home-based exercise on QoL and a considerable proportion of variance were observed. Further, the PI of QoL was − 4.76-5.24. A low proportion of variance was observed for FC and the significance was almost reached; besides, CMS returned that all studies share a common effect size. Overall, the significance was observed at p = 0.05, with a low proportion of variance and a PI of − 0.34-0.76. The mean adherence to home-based interventions was observed at 80.2 ± 6.1%. The sensitivity analysis included three studies 32,40,42 and revealed a significant effect of home-based intervention regardless of the outcome (overall, p = 0.01; I 2 = 0%) (Fig. 3).

Effects of supervised or mixed exercise intervention on the quality of life and functional capacity.
The characteristics and outcomes of the three studies that conducted supervised exercise interventions in CRC patients are shown in Table 4. The length of the intervention ranged from 8 to 24 weeks, with 2 17,44 , 4 39 , or 5 33 weekly exercise session frequency. The upper limit of the exercise session duration of supervised intervention ranged from 40 to 90 min. The intensity of exercise was not specified in two studies 33,39 . The two remaining studies used the HR max or rating of perceived exertion (RPE) scores for prescribing endurance activities. Additionally, Zimmer et al. 17 prescribed resistance exercises based on the percentage of one maximum repetition (1RM) or Borg's category ratio-scale (Borg CR10). Lin et al. 44 and Zimmer et al. 17 applied both endurance and strength exercises, while the intervention of Lu et al. 33 was solely based on Baduanjin qigong. Cantarero-Villanueva et al. 39 conducted aerobic and stabilization exercises.
Only one study fixed the length of mixed exercise intervention in CRC patients 37 . This approach was not possible in studies by Ahn et al. 36 and Karlsson et al. 31 since the main goals of these RCTs are directly associated with the intervention length. The exercise session duration ranged from 30 to 60 min in these studies. Additionally, only the RCT by Karlsson et al. 31 prescribed an exercise intensity based on an inspiratory test (maximal inspiratory pressure [MIP]) or ratings from the Borg CR10. Two studies adopted both strength and aerobic exercises 31,36 and Hwang et al. 37 mostly prescribed resistance exercises. The QoL and FC did not improve in these studies (Table 5).
A meta-analysis revealed a significant effect regarding exercise intervention entirely supervised or with some level of supervision (mixed) for CRC patients (PI − 0.41-1.39) (Fig. 4). The subgroup analysis for the QoL showed a high proportion of variance, near p-value of significance, and a PI of − 6.06-7.16. A considerable I 2 was obtained  17,33,36,39 and reinforced the significant effect of intervention with any level of supervision for CRC patients (overall, p = 0.00001, I 2 = 0%) (Fig. 5).

Discussion
A meta-analysis revealed that exercise intervention entirely supervised or with any level of supervision is effective in improving the FC in CRC patients. The same outcome was not observed for the QoL. Similarly, no significant effects were observed for home-based interventions in terms of the QoL and FC. However, regardless of exercise intervention supervision, achieving ≥ 80% intervention adherence was associated with improved QoL and FC in CRC patients.

Modulation of functional capacity and quality of life by home-based exercise intervention.
Four home-based interventions evaluated FC and two observed improvements in this outcome 35,42 . The latter interventions adopted aerobic and resistance exercises, but used different instruments for FC measurement. The studies that observed improvements in FC and 83% of the eligible home-based RCTs 32,35,41-43 provided details in terms of the exercise intensity, which ranged from 64 to 75% of the HR max . This meta-analysis supports the relevance of exercise adherence to improve QoL and FC of CRC patients. The sensitivity analysis revealed that when the adherence was ≥ 80%, the benefits of home-based programs were significant. While Christensen et al. 40 applied only walking exercises, Lee et al. 42 and Kim et al. 32 also allowed hiking, cycling, and resistance exercises. The survivors of ovarian cancer 48 and CRC 49 prefer walking as an exercise intervention, and this may have contributed to the improvements in the QoL 32,40 or FC 42 . Additionally, these studies suggest that gains are observed regardless of the exercises adopted when ≥ 80% of adherence is attained.  www.nature.com/scientificreports/ Moreover, these interventions lasted 12 weeks, and two 32,42 detailed the exercise load progression during the training. Although the load progression magnitude was not specified by Christensen et al. 40 , the participants were instructed to perform new tests with the InterWalk application to ensure the exercise program progression. These factors are relevant since the report by Lee et al. 43 did not observe modifications in the 6-MWT results in a short intervention (6 weeks) and without load progression; however, the physical activity levels and the heart rate after the Tecumseh test were modified in this report. Together, these data highlight the importance of considering training factors such as intervention duration and intensity prescription along with adherence ≥ 80%. The contamination of the control group may explain the non-modification of QoL in the study by Courneya et al. 41 . Pinto et al. 35 suggested that their small sample size and a possible ceiling effect on the QoL limited the detection of home-based exercise intervention in this outcome.
The eligible studies-except one-evaluated sedentary 35 patients or patients with low levels of physical activity 32,40,42,43 . Not less important, all eligible home-based RCTs evaluated the CRC patients who underwent surgery or therapy and were mostly non-metastatic. Therefore, future RCTs are required to investigate whether home-based exercise interventions can positively affect the FC and QoL of patients with CRC undergoing ongoing therapy, mainly for those with advanced cancer. This discussion was highlighted in a systematic review by Dittus et al. 50 in which some eligible studies observed improvements in the QoL, mobility, fatigue, and the sleep Table 4. Studies design and Quality of Life (QoL) and Functional Capacity (FC) outcomes of supervised exercise interventions in colorectal patients. The EORTC QLQ-C30 provides several outcomes related to the quality of life. Therefore, we opted to expose those who were modified by the intervention; Pos 1-reevaluation at the 12th week; Pos 2-reevaluation at the 24th. EORTC QLQ-C30 European Organization for Research and Treatment of Cancer Quality of Life Core Questionnaire, 6-MWT 6 Minute Walk Test, TOI Trial Outcome Index, RPE Rating of perceived exertion, HR max Maximum Heart Rate, 1RM one maximum repetition, Borg CR10 Borg's category ratio-scale, BFI Brief Fatigue Inventory. a Retention rate.

Modulation of functional capacity and quality of life by exercise interventions entirely supervised or with some level of supervision. Both the controls and CRC patients who exercised in the RCT
by Lin et al. 44 improved the FC and QoL. Since the controls were not restricted to physical activity, these authors suggest possible exercise contamination in this group. However, the within-group effect size for the supervised group was moderate to large in some EORTC QLQ-C30 subscales. The FC of CRC patients from the study of Zimmer et al. 17 did not improve after the intervention. However, among the interventions included in the sensitivity analysis, this study provided the largest SMD on QoL. Other important outcomes, such as the side effects of chemotherapy-induced peripheral neuropathy, balance, and strength were positively modulated. These improvements can be explained by the cautious prescription of the exercise length and intensity (e.g. intensity individualization by heart rate and 1RM). The dose-response of exercise is directly affected by the length of intervention and the precise prescription of the intensity 54 . Therefore, these factors should not be overlooked. Cantarero-Villanueva et al. 39 observed improvements of 79.7 m in the FC of CRC patients, while their controls only improved by 4.9 m. These authors emphasized that their exercise program was not designed to improve FC, but health-related parameters including muscle strength. The FC improvements were attributed to gains in abdominal strength, which may have positively influenced the total distance covered in the 6-MWT. Several reports demonstrating the association between muscle strength and FC were brought by the review of Maestroni et al. 55 , strengthening the suggestion of Cantarero-Villanueva et al. 39 . Lu et al. 33 demonstrated that supervised Baduanjin qigong can decrease the BFI score after 24 weeks of intervention. On the other hand, a recent Table 5. Studies design and Quality of Life (QoL) and Functional Capacity (FC) outcomes of exercise interventions with some level of supervision (i.e. mixed) in colorectal patients. The results of Karlsson et al. 31 were expressed as median. HR heart rate, Borg CR10 Borg's category ratio-scale, Pos 1 pre-surgery, Pos 2 possurgery. a The intervention and its length are directly related.  56 . Further RCTs involving this practice would confirm its benefits for patients with CRC. Two mixed interventions were not included in the meta-analysis for those who did not report SD 31 or those who did not present with exercise adherence 37 . The SMD of the remaining study was favorable to the intervention. Since only one mixed RCT was included in the meta-analysis, further studies with this approach are recommended to verify its effectiveness in improving the QoL and FC in this population.

Exercise session duration (D) Intensity (I) Exercises
Potential advantages and disadvantages of home-based and supervised exercise interventions for patients with CRC . Authors have debated the strengths and limitations of supervised and nonsupervised exercise interventions [18][19][20][21][22] . Home-based intervention is a valid strategy to overcome common barriers reported by cancer patients, such as access, time, and cost 57 . However, the home-based definition requires caution interpretation 21 , and proper planning must be conducted even during independent exercise. In line with this, the transition from a supervised environment to a non-supervised was reported as the most significant barrier to exercise by cancer patients 58 . Moreover, a systematic review by Ormel et al. 59 emphasizes that supervision in a home-based setting can increase exercise adherence by enhancing family support and improving the knowledge and exercise skills of the cancer patient. The latter perspective is aligned with one of the results reported by Hardcastle et al. 60 in which most of the interviewed non-metropolitan cancer patients were unaware of physical activity recommendations or had incorrect knowledge regarding the intensity required in these guidelines.  www.nature.com/scientificreports/ One potential advantage of supervised exercise is associated with the control of exercise intensity. The systematic review and meta-analysis of Sweegers et al. 61 (74 eligible exercise arms) found significant beneficial effects on the QoL and physical function of supervised exercise RCTs in cancer patients. In agreement, non-eligible studies according to our inclusion criteria demonstrated that moderate 62 and high 15 intensity supervised exercise promote benefits to cancer patients, including those diagnosed with CRC. Exclusively for CRC patients, the studies suggest high-intensity exercise as a feasible and efficacious intervention for improving the FC 63,64 .
Mixed interventions are feasible 31 and can reduce the hospital stay after colectomy procedure 36 and are associated with inhibited global hypermethylation 37 . This approach has emerged as a notable strategy to overcome barriers in home-based or entirely supervised exercise interventions. However, without proper exercise planning, the engagement of the patient, and some experience with exercise, the same limitations previously mentioned for supervised or unsupervised intervention can occur here. Overall, opinion articles have provided an important discussion on the necessity of considering the principles of training and optimizing the safety and efficacy of exercise through cancer treatment 65,66 . Such a perspective must be considered regardless of the supervision level of the exercise intervention.
Limitations, strengths, and insights for future research. This systematic review and meta-analysis should be interpreted in light of these limitations. Few eligible studies have measured both FC and QoL. None of the mixed interventions have measured the QoL. Moreover, the primary goals of some included RCTs were heterogeneous and the length, frequency, and intensity of the exercise intervention varied substantially. Only 15% of the studies included metastatic patients, and a small sample size was observed in 61% of the RCTs. Furthermore, we cannot discuss the effects of exercise interventions on specific CRC sites (e.g., colon or rectum), which have distinct genotypes and phenotypes, since the included RCTs did not separate the intervention groups according to the tumor location. Future studies should consider this perspective, as CRC types require personalized and individualized treatment 64 . Further research on exercise intervention, regardless of the supervision, for CRC patients must clearly report: (a) the intensity prescription (individualized vs. non-individualized) and the basis for it (i.e., which parameter was adopted to prescribe the intensity); (b) the weekly as well as the total duration; (c) the calculation of load performed during the intervention and how the load progression was performed. If future studies provide efforts to describe these factors, future meta-analyses will be capable of creating subgroups and refining the exercise prescriptions for this population. The strengths of this study are completing the meta-analysis, comparing home-based versus supervised exercise, and performing a novel sensitivity analysis to determine the influence of exercise adherence.

Conclusion
This systematic review and meta-analysis verified that improvements in the QoL and FC of CRC patients were observed when the adherence to the intervention was ≥ 80%. Such results do not necessarily imply that low physical intervention adherence for this population does not provide any benefits. However, the higher the adherence, the higher are the chances of acquiring benefits in the QoL and FC. Regardless of the supervision characteristics, future RCTs are strongly encouraged to provide a detailed description of exercise variables in physical interventions for CRC prescription. This perspective will allow a refined exercise prescription for patients with CRC, mainly according to their clinical status.