*3.10. Synthesis of Results*

#### 3.10.1. Main Results

The main results of each included meta-analysis are summarised in Table 3. Different methods were used to present aggregate effects, including Cohen's d, Hedges' g statistic, and the standardised mean difference (SMD) using a random-effects model. In four out of the six included meta-analyses, the authors observed a significant reduction in depressive symptoms favouring the exercise group. In studies from Patsou et al. [20] and Vashistha et al. [21], no statistically significant decrease in depressive symptoms for the exercise group was observed. Three meta-analyses, Brown et al. [17], Craft et al. [18] and Patsou et al. [20] observed small effect sizes, whereas two meta-analyses, Gonzalez et al. [19] and Yi et al. [22], reported moderate effect size. All the included meta-analyses presented a large or very large heterogeneity (I2 from 55% to 84%).

**Table 3.** Results of the meta-analyses included in the study.


Abbreviations: d, mean change scores (Cohen's d); g, Hedges' g statistic to estimated effect size; I2, I-squared statistic for heterogeneity; SMD, standardised mean difference.

#### 3.10.2. Sensitivity and Subgroup Analyses

Four of the six included meta-analyses presented sensitivity or subgroup analyses. Regarding the type of cancer, the Brown et al. [17] subgroup analysis revealed significant reductions in depressive symptoms among breast cancer patients (d= −0.17; 95% CI: −0.32, −0.02), but non-significant differences for prostate, leukaemia, lymphoma and colorectal cancer patients. Gonzalez et al. [19] proceeded with a subgroup analysis of only breast cancer patients and presented a significant moderate reduction in depressive symptoms favouring exercise (g = −0.41; 95% CI: −0.59, −0.23).

A subgroup analysis compared supervised vs. non-supervised exercise. Brown et al. [17] showed that supervised exercise was the most effective in reducing depressive symptoms (ß = −0.26, *p* = 0.01). Moreover Craft et al. [18] founded that supervised exercise presented a greater reduction in depressive symptoms (ES = −0.67; 95% CI: −1.11, −0.23) than non-supervised exercise (ES = 0.25; 95% CI: −0.01, 0.50).

Another aspect observed in the subgroup analyses was the exercise dose. Craft et al. [18] founded that exercise bout durations >30 min had larger effects (ES = −0.57; 95% CI: −0.91, −0.23) on depression than exercise bouts ≤30 min (ES = 0.01; 95% CI: −0.20, 0.22). Lastly, Patsou et al. [20] demonstrated that exercising ≤135 min/week yielded a moderate to a large effect (g = −0.82; 95% CI: −1.54, −0.10; I2 = 35%) and exercising ≥135 min/week presented no significant effect. Moreover, exercise up to 12 weeks yielded a moderate to

a large effect (g = −1.69; 95% CI: −2.66, −0.73; I2 = 32%), while exercise duration over 12 weeks presented no significant effect.

Regarding participants' age, only Brown et al. [17] explored this subgroup analysis and showed that exercise was the most effective when cancer patients were between 47 and 62 years (ß = −0.27, *p* = 0.01). Among older adults, the effect of exercise on depressive symptoms was not significant.

Craft et al. [18] also analysed potential moderators of effect, including exercise location, observing that home-based exercise was associated with increased depressive symptoms (ES = 0.16; 95% CI: −0.15, 0.47), while other exercise locations presented a reduction in depressive symptoms (ES = −0.45; 95% CI: −0.77, −0.14).

The Gonzalez et al. [19] sensitivity analysis showed that removing one study greatly reduced heterogeneity (I2 = 36.9%) but also reduced the effect size to the small-medium range (g = −0.41; 95% CI: −0.55, −0.28). Additionally, comparing studies that used active and inactive control interventions did not find differences; both had a significant, medium to small effect size post-intervention.

Lastly, Patsou et al. [20] performed several subgroup analyses showing that: aerobic exercise interventions yield a large and significant effect on depression (g = −1.23; 95% CI: −1.97, −0.49; I2 = 0%), no significant effect was found regarding resistance exercise interventions, combined aerobic and resistance exercise and Yoga interventions. Exercise during treatment yielded a moderate effect (g = −0.54; 95% CI: −1.16, 0.08; I2 = 25%), while exercise post-treatment yielded no significant effect.

#### **4. Discussion**

This umbrella review included six meta-analyses that comprised 100 individual studies with little overlap that investigated the effect of exercise on depressive symptoms among cancer survivors. Overall, a small significant reduction in depressive symptoms in this vulnerable population was observed in the studies. However, high-quality evidence for the efficacy of exercise on depressive symptoms is limited. For a more detailed analysis, some points need to be considered, such as the type of cancer, the specificity of exercise prescription, the time of interventions, and during or after cancer.

In our umbrella review, participants had mainly breast cancer in the included metaanalysis and were mostly women. Only one study did not include breast cancer [27] and was with prostate cancer patients, and it was the one that did not observe a significant effect of exercise on depressive symptoms. In a subgroup analysis, Brown et al. found a significant reduction in depressive symptoms among breast cancer survivors but did not find the same in prostate, leukaemia, lymphoma and colorectal cancer [17]. The prevalence of depression among breast cancer survivors is higher than in other cancers and can achieve 32.8% [35]. Moreover, depression is more prevalent in women than men [36], and breast cancer is prevalent in women. Evidence suggests that depression in breast cancer patients decreases over time and is more common throughout the disease and in the recurrent phase of breast cancer [37]. The occurrence of depression among patients with breast cancer is due to several factors, such as treatment-related distress, worries regarding fear of death and disease recurrence, and altered body image, sexuality and attractiveness [38–40]. In addition, a study exposes the association between depression and tumour levels of estrogen receptors and progesterone receptors [41]. A study found that fatigue and pain are significant risk factors for developing depression among breast cancer survivors [37]. Fatigue is also a recognised barrier to exercise [42]; however, exercise can reduce fatigue among women with breast cancer [43]. The benefits of exercise can be extended to improve physical functioning and multiple aspects of quality of life among cancer patients [44]. Moreover, exercise is a feasible alternative to control symptoms burden and improve well-being among breast cancer patients [39].

Another sample characteristic that must be highlighted is that most patients were older adults (>50 years old). In the general population, the prevalence of depression symptoms rises with increasing age, 10% to 15% of older adults have clinically significant depressive symptoms [45]. Older patients with cancer often experience depression, fatigue, pain, and sleep disturbance [46]. Only one included meta-analysis directly explored the role of age in the effectiveness of exercise on depression symptoms and found that the efficacy seems to disappear among old age patients [24]. However, an RCT with older cancer patients receiving chemotherapy found that after the six-week structure exercise program, participants' anxiety and mood improved [47]. Besides the effects of exercise on mental health, physically active old age patients improve general health, such as physical fitness outcomes, quality of life and increased life expectancy [48].

When considering the effects of exercise on depressive symptoms, it is necessary to consider the characteristics of the exercise we are referring to. Many dimensions of exercise exist, which are captured in part by the principle (frequency, intensity, time and type of exercise), as well as the way of practising, whether accompanied or not and if exercise occurs indoors or outdoors. However, the included systematic meta-analyses showed great variance concerning exercise. Except for Gonzalez et al. and Yi et al. [26,29], which analysed the effects of yoga intervention, the others included meta-analyses that examined a variety of exercises, such as aerobic (e.g., walking, cycling), resistance (e.g., weight machine, resistance bands) and qigong. Only the Patsou et al. study explored the difference between the types of exercise and found that aerobic intervention yields a large significant effect on depressive symptoms. At the same time, resistance training presents a small significant effect, and combined aerobic and resistance training yielded a moderate effect [27]. This statement is in accordance with the American College of Sports Medicine (ACSM), which describes that resistance training alone does not seem effective for depression [3]. Aerobic activities are cost-effective and should be popularised in clinical practice.

Regarding yoga, both included meta-analyses that analysed only intervention found significant and medium effects on depressive symptoms [26,29]. However, in the Patsou et al. study, which included aerobic exercise, resistance exercise and yoga intervention, when a subgroup analysis proceeded and considered only yoga intervention, no significant difference in depression symptoms was observed [27]. The contradictory results found in the three studies can be explained by the fact that yoga combines breathing (pranayama) and meditative techniques during a series of postures (asanas), but different types of yoga were being practised, which made it difficult to understand the effects of this practice [49].

Two included meta-analyses found that supervised exercise is more efficient than non-supervised exercise [32,38], which also appears in the ACSM recommendation [3]. Craft et al.'s study [25] explored the effects of exercise session durations and found that more than 30 min had larger effects compared with less than 30 min of the exercise session. In Patsou et al. [27] ≤135 min/week yielded a moderate to large effect and no effect with ≥135 min/week of exercise. The ACSM describes that aerobic training performed three times per week and for at least 12 weeks or twice weekly with combined aerobic plus resistance training lasting 6 to 12 weeks, can significantly reduce depressive symptoms in cancer survivors during and after treatment. However, the exact exercise duration per week has not yet been established by the ACSM. Gonzalez et al.'s study [26] explored the frequency and found no differences between one class per week and two or more classes per week. In contrast, Patsou et al. [27] explored the exercise intervention program duration and found that exercise for up to 12 weeks yielded a moderate to large effect compared with a small effect of over 12 weeks. Aside from the efficacy of depressive symptoms, exercise has other benefits in health outcomes among cancer survivors that must be considered such as improving cardiorespiratory fitness [50], and muscle strength [51,52].

Another important aspect of the efficacy of exercise on depressive symptoms among cancer survivors is the time of intervention, before the diagnosis, during treatment or in a recovering phase. This aspect was explored, and no difference was found between patients receiving cancer treatment, following treatment or mixed treatment status [26]. On the other hand, patients under treatment yielded a moderate effect, and patients post-treatment yielded a small effect [27]. Exercise increases the chemotherapy completion rate during treatment without causing lymphedema or significant adverse events [53]. In addition, exercise appears to reduce chemotherapy-induced peripheral neuropathy symptoms in patients receiving taxane-, platinum-, or vinca alkaloid-based chemotherapy [54].

Concerning the methodological quality, the AMSTAR 2 scores show that the majority were of moderate methodological quality. Nevertheless, as for item #10, "Did the review authors report on the sources of funding for the studies included in the review", no study reported the source of funding which can entail a risk of bias.

#### *Strengths and Limitation*

The strength of our study is that we included two recent meta-analyses, one from 2020 [19] and one from 2021 [22], and compiled current data regarding the effectiveness of exercise in depressive symptoms among cancer patients. However, some limitations should be exposed. The prevalence of breast cancer was substantial, which prevented us from generalising the findings to other types of cancer. Two studies did not perform subgroup analysis, and those who performed subgroup analysis did not analyse the same constructs, specifically in relation to the FITT principles of exercise. Future research is required to clarify the effectiveness of different intervention modalities (frequency, intensity, time and type) for depressive symptoms among cancer survivors to allow determination of exercise dose-response. Moreover, additional studies are needed to evaluate the cost-analysis and adverse events of exercise, and there is an urgent need for innovative methods to generate high-quality evidence.

#### **5. Conclusions**

Our critical review contributed to the evidence of the effects of exercise on depressive symptoms among cancer patients. Four in six studies found a significant effect of exercise. Some aspects should be highlighted and should be used for future interventions. Supervised exercise is better than non-supervised. The dose of exercise seems to be important, however, the finding did not present a specific dose-response relation. Exercise outside of the home is better than home-based exercise. Aerobic exercise is the most effective type of exercise. The effect of exercise on depression seems to be more effective among breast cancer patients. Future studies should explore other types of cancer.

**Author Contributions:** P.M.: conceptualisation, formal analysis, writing—original draft, writing review and editing; A.M.: supervision, formal analysis, writing review and editing; G.F.: writing review and editing; É.R.G.; writing review and editing; M.P.: formal analysis, writing review and editing; A.I.: supervision, writing review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Swiss National Centre of Competence in Research LIVES—Overcoming vulnerability: life course perspectives, granted by the Swiss National Science Foundation (grant number: 51NF40-185901). AI acknowledges support from the Swiss National Science Foundation (grant number: 10001C\_189407).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

