Abstract
Several sources of evidence indicate that exercise during and after breast cancer could positively modulate the tumor microenvironment. This meta-analysis aimed to determine the effects of exercise training on mediators of inflammation in breast cancer survivors. We searched for randomized controlled trials published from January 1990 to March 2014. An inverse variance method of meta-analysis was performed using a random effects model in the presence of statistical heterogeneity. Eight high-quality trials (n = 478) were included. Exercise improved the serum concentrations of IL6 [weighted mean difference (WMD) = −0.55 pg/mL; 95% confidence interval (CI), −1.02 to −0.09], TNFα (WMD = −0.64 pg/mL; 95% CI, −1.21 to −0.06), IL8 (MD = −0.49 pg/mL; 95% CI, −0.89 to −0.09), and IL2 (WMD = 1.03 pg/mL; 95% CI, 0.40 to 1.67). No significant differences were found in the serum concentrations of C-reactive protein (WMD = −0.15; 95% CI, −0.56 to 0.25) or IL10 (WMD = 0.41; 95% CI, −0.18 to 1.02). Exercise training positively modulates chronic low-grade inflammation in women with breast cancer, which may impact upon carcinogenic mechanisms and the tumor microenvironment. These findings align with the other positive effects of exercise for breast cancer survivors, reinforcing the appropriateness of exercise prescription in this population. Cancer Epidemiol Biomarkers Prev; 25(7); 1009–17. ©2016 AACR.
Introduction
Breast cancer is the most common cancer among women, with nearly 1.4 million cases worldwide annually (1). Several mechanisms have been postulated regarding the etiology and progression of breast cancer (2). Among these mechanisms, chronic inflammation is widely recognized to play a crucial role in cancer development, progression, and risk of recurrence due to its effects on carcinogenesis and the tumor microenvironment (3). Cytokine signaling and oxidative stress result in DNA damage and genomic changes, enhancing tumor progression, angiogenesis, cell proliferation, invasiveness, metastasis, and tumor-cell resistance against several anticancer treatments (4, 5). In addition, mediators of inflammation are associated with reduced overall survival in women with breast cancer, even after adjustments for age, tumor stage, race, and body mass index (6).
A strong body of evidence supports exercise training as a therapy for cancer patients during and after anticancer treatment (7, 8) because exercise training reverses some of the detriments that cancer causes in quality of life, fatigue, depression, muscular strength, and body composition (7–10), without adverse side effects (11). Several of these signs and symptoms that occur commonly in cancer have been associated statistically and linked aetiologically with proinflammatory cytokines (12, 13). Therefore, one crucial mechanism by which physical exercise exerts favorable health effects may be its capacity to reduce chronic low-grade inflammation (Fig. 1).
In 2012, Löf and colleagues published a thorough systematic review of randomized trials attempting to establish the effect of exercise on inflammatory mediators in survivors of breast cancer (14). The review found no significant effects on interleukins (IL) among four trials, and some evidence that exercise may decrease C-reactive protein (CRP) levels in one trial. The authors of that review concluded that further data were needed.
Although the systematic review by Löf and colleagues was published relatively recently, further data have already become available. For example, simple citation tracking from the systematic review by Löf and colleagues via Google Scholar identifies two additional trials with further data about the effect of exercise on numerous inflammatory mediators (15, 16). Further trials may be identified by rigorous searching. Furthermore, the review by Löf and colleagues did not undertake any meta-analysis, but this is possible with the currently available data. Therefore, the aim of the following systematic review was to determine the effect of exercise training on mediators of inflammation in breast cancer survivors, including pooling of the data with meta-analysis where possible.
Materials and Methods
Protocol
The protocol for this systematic review was registered in the PROSPERO database (CRD42014009402) and the PRISMA statement was used to guide the reporting of the review (17). No funding was received.
Identification and selection of trials
Three reviewers (J.F. Meneses-Echávez, J.S. Río-Valle, and E. González-Jiménez) independently screened the search results. The reviewers were blinded to both the name of the authors and to the results of the studies. Searches were conducted between January and May 2014. We searched the PubMed, Embase, Scopus, and Cochrane Central Register of Controlled Trials (CENTRAL) databases using Boolean operators to identify records with terms for the disease (breast cancer, tumor, or carcinoma), the intervention (exercise, physical exercise, or physical training), and the outcomes (inflammation, mediators, anti-inflammatory, cytokines, interleukin*, IL2, IL6, IL8, IL10, CRP, tumor necrosis factor or TNFα). See Supplementary Material S1 for a detailed description of the search strategy. Studies published between 1990 and 2014 were considered for selection. In addition, the reviewers examined the reference lists of the included studies and the conference abstracts of the American Society of Clinical Oncology Annual Meeting on its website from 2004 to 2013, as well as 6 relevant journals: The Lancet Oncology, Journal of Clinical Oncology, Journal of the National Cancer Institute, Journal of Breast Cancer, The Breast Journal, and The Breast. Moreover, the authors contacted high-profile researchers in this area to ask for other possibly relevant trials, published or unpublished. No language restrictions were applied.
Selection criteria
The studies were included if they met the following criteria: (i) a randomized controlled trial involving breast cancer survivors; (ii) included an experimental group performing an exercise training program (categorized as aerobic, resistance, combined aerobic/resistance, yoga or Tai Chi); (iii) included a control group that undertook conventional care only, education only, or no intervention; and (iv) measured serum concentrations of at least one of the following inflammatory mediators: cytokines (IL2, IL6, IL8, IL10), TNFα, and CRP.
Studies were not excluded on the basis of the gender of the participants with breast cancer. Exercise training was defined as any body movement causing an increase in energy expenditure involving a planned or structured movement of the body performed in a systematic manner in terms of frequency, intensity, and duration that was designed to maintain or enhance health-related outcomes (18). Studies were excluded if the exercise intervention included dietary intervention, manual therapy, or psychologic therapeutic approaches. Attempts were made to contact the authors of the trial reports if clarification was necessary. Three reviewers (F. Lobelo, J.E. Correa-Bautista, and M. Elkins) independently screened the studies for eligibility. Disagreements were resolved by discussion and, where necessary, arbitration by a fourth reviewer (E. González-Jiménez).
Outcome measures
The outcome measures evaluated in this systematic review were serum levels of inflammatory mediators (IL2, IL6, IL8, IL10, CRP, and TNFα) after the exercise interventions. The procedures used to measure the serum concentrations of these inflammatory mediators, such as cytokine immunoassay and ELISA kits, were evaluated by one reviewer (R. Ramírez-Vélez) when each study was considered for inclusion.
Ethics declarations
Two investigators (J.F. Meneses-Echávez and R. Ramírez-Vélez) confirmed that the included studies had ethics committee approval and that the participants signed consent forms.
Data extraction
After selecting the studies, the relevant data were extracted by three reviewers (J.F. Meneses-Echávez, M. Elkins, and E. González-Jiménez) blinded to the name of the authors. The following information was extracted: (i) study design: publication year, randomization methods, selection criteria, and intervention groups; (ii) participants: sample size, age, menopausal status, current treatment (yes/no), treatment regimen (chemotherapy, radiotherapy, surgery), stage of disease, and baseline values for outcome measures; (iii) intervention: exercise modality, length (weeks), frequency (sessions/week), duration of training (minutes/session), and intensity of training (maximal heart rate %); (iv) outcome data for each group regarding inflammatory mediators and adverse events.
After data extraction, the data were examined for completeness and accuracy by a third reviewer (J.S. Río-Valle). Disagreements were resolved via review of the trial report and discussion.
Assessment of the risk of bias and completeness of reporting
We used the PEDro scale (19) to assess the risk of bias and the completeness of reporting of the included studies. The PEDro scale is based on the Delphi list (20) and evaluates external validity (criterion 1), internal validity (criteria 2–9), and whether sufficient statistical information is provided to interpret the effect of the intervention (criteria 10–11). Two reviewers (E. González-Jiménez and J.S. Río-Valle) independently performed these assessments, with disagreements resolved by discussion.
Statistical analysis
For continuous outcomes, we recorded the group size, the mean values, and the SDs for each group compared in the included studies. If SDs were not reported, they were calculated from SEs, CIs, or t values (21). Pooled effects were calculated using an inverse of variance model, and the data were pooled to generate a weighted mean difference (WMD) in the original units with corresponding 95% confidence intervals (CI). All the studies for each outcome reported data in the same units, so we were able to pool all studies regardless of whether they reported change data or final data. Significance was set at P < 0.05. Statistical heterogeneity was evaluated using the I2 statistic, and classified according to the Cochrane Handbook (22): negligible heterogeneity, 0%–40%; moderate heterogeneity, 30%–60%; substantial heterogeneity, 50%–90%; and considerable heterogeneity, 75%–100%. Other possible sources of heterogeneity were evaluated via subgroup analysis and a cumulative meta-analysis model if necessary. Throughout the results, the ± symbol represents SD.
A fixed-effect model was used if heterogeneity was low (I2 < 50%); otherwise, a random effects model was used. Subject to data availability, we planned to conduct subgroup analyses according to the modality of exercise investigated (resistance, aerobic, mixed, yoga, Tai Chi), the type of cancer treatment (active or not), and the stage of disease. Meta-regression analysis was performed to examine the association between publication year, length of the intervention program (weeks), duration (minutes/session), and frequency (sessions/week) of exercise training with changes in effect size for each inflammatory mediator. Finally, publication bias was examined via Egger linear regression test for funnel plot asymmetry (P < 0.05), for each outcome with ≥ 10 trials. If no outcomes reached this threshold, publication bias was assessed for the outcome with the greatest number of trials. All analyses were performed by one reviewer (J.F. Meneses-Echávez) using Comprehensive Meta-Analysis (Version 2.0) and checked against the extracted data by one author (M. Elkins). A sensitivity analysis for quality was conducted by excluding those trials with a quality score less than 5 from the meta-analyses to see whether this affected the overall results of the meta-analyses.
Results
Flow of studies and participants into the review
After the removal of duplicates, 367 studies were screened, with 95 studies being retrieved in full text for detailed evaluation of eligibility. Eight trials (n = 478), reported in nine papers, were included in the review (15, 16, 23–29). The results of the search and the reasons for exclusions are presented in Fig. 2. The pooled cohort included 253 women randomized to an exercise training group and 225 women randomized to a control group.
Risk of bias and completeness of reporting
Most of the criteria on the PEDro scale were met by all or most of the included trials. The criteria on the PEDro scale that were met by a minority of the trials were intention-to-treat analysis (38%), concealed allocation (25%), and blinding of participants and therapists (0%). The specific criteria met by each of the trials are presented in Table 1.
Study . | Random allocation . | Concealed allocation . | Groups similar at baseline . | Participant blinding . | Therapist blinding . | Assessor blinding . | < 15% dropouts . | Intention-to-treat analysis . | Between-group difference reported . | Point estimate and variability reported . | Total (0–10) . |
---|---|---|---|---|---|---|---|---|---|---|---|
Bower et al. 2014 (23) | Y | N | Y | N | N | N | Y | N | Y | Y | 5 |
Ergun et al. 2013 (24) | Y | N | Y | N | N | Y | Y | N | Y | Y | 6 |
Gómez et al. 2011 (16) | Y | N | Y | N | N | N | N | N | Y | Y | 4 |
Hutnick et al. 2005 (25) | N | N | Y | N | N | N | Y | N | Y | Y | 4 |
Janelsins et al. 2011 (26) Sprod et al. 2012 (29) | Y | Y | Y | N | N | N | N | Y | Y | Y | 6 |
Jones et al. 2013 (15) | Y | N | Y | N | N | Y | Y | Y | Y | Y | 7 |
Kiecolt-Glaser et al. 2014 (27) | Y | Y | Y | N | N | Y | Y | N | Y | Y | 7 |
Rogers et al. 2013 (28) | Y | N | Y | N | N | Y | N | Y | Y | Y | 6 |
88% | 25% | 100% | 0% | 0% | 50% | 63% | 38% | 100% | 100% |
Study . | Random allocation . | Concealed allocation . | Groups similar at baseline . | Participant blinding . | Therapist blinding . | Assessor blinding . | < 15% dropouts . | Intention-to-treat analysis . | Between-group difference reported . | Point estimate and variability reported . | Total (0–10) . |
---|---|---|---|---|---|---|---|---|---|---|---|
Bower et al. 2014 (23) | Y | N | Y | N | N | N | Y | N | Y | Y | 5 |
Ergun et al. 2013 (24) | Y | N | Y | N | N | Y | Y | N | Y | Y | 6 |
Gómez et al. 2011 (16) | Y | N | Y | N | N | N | N | N | Y | Y | 4 |
Hutnick et al. 2005 (25) | N | N | Y | N | N | N | Y | N | Y | Y | 4 |
Janelsins et al. 2011 (26) Sprod et al. 2012 (29) | Y | Y | Y | N | N | N | N | Y | Y | Y | 6 |
Jones et al. 2013 (15) | Y | N | Y | N | N | Y | Y | Y | Y | Y | 7 |
Kiecolt-Glaser et al. 2014 (27) | Y | Y | Y | N | N | Y | Y | N | Y | Y | 7 |
Rogers et al. 2013 (28) | Y | N | Y | N | N | Y | N | Y | Y | Y | 6 |
88% | 25% | 100% | 0% | 0% | 50% | 63% | 38% | 100% | 100% |
Abbreviations: N, No; Y, Yes.
Characteristics of the included trials
Table 2 summarizes the characteristics of the participants, interventions, and outcome measures in the eight included trials. All eight trials included in the systematic review provided statistical estimates appropriate for meta-analysis.
Study . | Participants . | Intervention . | Outcome measures . |
---|---|---|---|
Bower et al. 2014 (23) | 31 female breast cancer patients (stage 0–II) with fatigue after local and/or adjuvant therapy | Exp = Yoga (90 min × 2/wk × 12 wk)Con = Education (120 min × 1/wk × 12 wk) | IL6, CRP, TNFα |
Exp: n = 16, age (yr) = 54 (SD 6) | |||
Con: n = 15, age (yr) = 53 (SD 5) | |||
Ergun et al. 2013 (24) | 60 female breast cancer patients (stage I–IIIa) after surgery, radiotherapy, and chemotherapy | Exp1 = Aerobic/resistance exercise (45 min × 3/wk × 12 wk) + aerobic exercise (30 min × 3/wk × 12 wk) + education (30 min) | IL6, IL8, TNFα |
Exp1: n = 20, age (yr) = 50 (SD 8) | Exp2 = Aerobic exercise (30 min × 3/wk × 12 wk) + education (30 min) | ||
Exp2: n = 20, age (yr) = 55 (SD 7) Con: n = 20, age (yr) = 55 (SD 10) | Con = Education (30 min) | ||
Gómez et al. 2011 (16) | 16 female breast cancer patients (stage I–II) after surgery, radiotherapy and chemotherapy | Exp = Aerobic/resistance exercise (90 min × 3/wk × 8 wk) | IL2, IL6, IL8, IL10, TNFα |
Exp: n = 8, age (yr) = 50 (SD 6) | Con = Usual care | ||
Con: n = 8, age (yr) = 49 (SD 6) | |||
Hutnick et al. 2005 (25) | 49 female breast cancer patients (stage I–III) during or after chemotherapy and after radiotherapy and surgery | Exp = Aerobic/resistance exercise (40–90 min × 3/wk × 24 wk) | IL6 |
Exp: n = 28, age (yr) = 49 (SD 11) | Con = Usual care | ||
Con: n = 21, age (yr) = 52 (SD 9) | |||
Janelsins | 31 female breast cancer patients (stage 0–IIIb) | Exp = Tai Chi (60 min × 3/wk × 12 wk) | IL2, IL6, IL8 |
et al. 2011 | after surgery, radiotherapy, and chemotherapy | ||
(26) and | |||
Sprod et | Exp: n = 9 completers, age (yr) = 54 (SD 11) | Con = Education and psychosocial support | |
al. 2012(29) | Con: n = 10 completers, age (yr) = 53 (SD 7) | ||
Jones et al. 2013 (15) | 75 female breast cancer patients (stage 0–IIIa) after adjuvant treatment (except endocrine therapy) | Exp = Aerobic exercise (150 min × 3/wk × 24 wk) | IL6, CRP, TNFα |
Exp: n = 37, age (yr) = 56 (SD 10) | Con = Usual care | ||
Con: n = 38, age (yr) = 55 (SD 8) | |||
Kiecolt-Glaser et al. 2014 (27) | 200 female breast cancer patients (stage 0–IIIa) after surgery, radiotherapy, and chemotherapy (except tamoxifen / aromatase inhibitors) | Exp = Yoga (90 min × 2/wk × 12 wk) | IL6, TNFα |
Exp: n = 100, age (yr) = 52 (SD 10) | Con = Usual care | ||
Con: n = 100, age (yr) = 51 (SD 9) | |||
Rogers et al. 2013 (28) | 28 female breast cancer patients (stage I–IIIa) after surgery, radiotherapy, and chemotherapy | Exp = Aerobic exercise (150 min/wk × 12 wk) + resistance exercise (2/wk × 12 wk) | IL6, IL8, IL10, TNFα |
Exp: n = 15, age (yr) = 58 (SD 6) | Con = Educational materials | ||
Con: n = 13, age (yr) = 54 (SD 14) |
Study . | Participants . | Intervention . | Outcome measures . |
---|---|---|---|
Bower et al. 2014 (23) | 31 female breast cancer patients (stage 0–II) with fatigue after local and/or adjuvant therapy | Exp = Yoga (90 min × 2/wk × 12 wk)Con = Education (120 min × 1/wk × 12 wk) | IL6, CRP, TNFα |
Exp: n = 16, age (yr) = 54 (SD 6) | |||
Con: n = 15, age (yr) = 53 (SD 5) | |||
Ergun et al. 2013 (24) | 60 female breast cancer patients (stage I–IIIa) after surgery, radiotherapy, and chemotherapy | Exp1 = Aerobic/resistance exercise (45 min × 3/wk × 12 wk) + aerobic exercise (30 min × 3/wk × 12 wk) + education (30 min) | IL6, IL8, TNFα |
Exp1: n = 20, age (yr) = 50 (SD 8) | Exp2 = Aerobic exercise (30 min × 3/wk × 12 wk) + education (30 min) | ||
Exp2: n = 20, age (yr) = 55 (SD 7) Con: n = 20, age (yr) = 55 (SD 10) | Con = Education (30 min) | ||
Gómez et al. 2011 (16) | 16 female breast cancer patients (stage I–II) after surgery, radiotherapy and chemotherapy | Exp = Aerobic/resistance exercise (90 min × 3/wk × 8 wk) | IL2, IL6, IL8, IL10, TNFα |
Exp: n = 8, age (yr) = 50 (SD 6) | Con = Usual care | ||
Con: n = 8, age (yr) = 49 (SD 6) | |||
Hutnick et al. 2005 (25) | 49 female breast cancer patients (stage I–III) during or after chemotherapy and after radiotherapy and surgery | Exp = Aerobic/resistance exercise (40–90 min × 3/wk × 24 wk) | IL6 |
Exp: n = 28, age (yr) = 49 (SD 11) | Con = Usual care | ||
Con: n = 21, age (yr) = 52 (SD 9) | |||
Janelsins | 31 female breast cancer patients (stage 0–IIIb) | Exp = Tai Chi (60 min × 3/wk × 12 wk) | IL2, IL6, IL8 |
et al. 2011 | after surgery, radiotherapy, and chemotherapy | ||
(26) and | |||
Sprod et | Exp: n = 9 completers, age (yr) = 54 (SD 11) | Con = Education and psychosocial support | |
al. 2012(29) | Con: n = 10 completers, age (yr) = 53 (SD 7) | ||
Jones et al. 2013 (15) | 75 female breast cancer patients (stage 0–IIIa) after adjuvant treatment (except endocrine therapy) | Exp = Aerobic exercise (150 min × 3/wk × 24 wk) | IL6, CRP, TNFα |
Exp: n = 37, age (yr) = 56 (SD 10) | Con = Usual care | ||
Con: n = 38, age (yr) = 55 (SD 8) | |||
Kiecolt-Glaser et al. 2014 (27) | 200 female breast cancer patients (stage 0–IIIa) after surgery, radiotherapy, and chemotherapy (except tamoxifen / aromatase inhibitors) | Exp = Yoga (90 min × 2/wk × 12 wk) | IL6, TNFα |
Exp: n = 100, age (yr) = 52 (SD 10) | Con = Usual care | ||
Con: n = 100, age (yr) = 51 (SD 9) | |||
Rogers et al. 2013 (28) | 28 female breast cancer patients (stage I–IIIa) after surgery, radiotherapy, and chemotherapy | Exp = Aerobic exercise (150 min/wk × 12 wk) + resistance exercise (2/wk × 12 wk) | IL6, IL8, IL10, TNFα |
Exp: n = 15, age (yr) = 58 (SD 6) | Con = Educational materials | ||
Con: n = 13, age (yr) = 54 (SD 14) |
Participants
The mean age of the participants in the included trials ranged from 49 to 56 years, with a mean of 54 ± 4. The majority of these trials involved postmenopausal women. Participants exhibiting different stages of disease were recruited (breast cancer type 0–IIIb). The included trials rarely reported time since diagnosis.
Interventions
Four trials tested a combination of aerobic and resistance training (16, 24, 25, 28). Two trials tested aerobic exercise alone (15, 24). Two trials tested yoga (23, 27). One trial, reported in two articles, tested Tai-chi (26, 29). The exercise interventions were performed for a mean length of 19 ± 13 weeks at a frequency of 3 ± 1 sessions per week for 69 ± 34 minutes per exercise session. The majority of interventions were supervised by health care providers.
Effect estimates of exercise on the inflammatory mediators
With respect to the effects of exercise training on the serum levels of cytokines in breast cancer survivors, the results of all the meta-analyses and subgroup analyses are summarized in Supplementary Table S1. The meta-analyses for each individual cytokine are discussed in detail below.
IL6
The most data were obtained for IL6, with all eight trials contributing data. Exercise improved the concentration of IL6, with a WMD of −0.55 pg/mL, which was statistically significant (95% CI, −1.02 to −0.09). The description of subgroup analysis according to the mode of training and the overall estimate are shown in Fig. 3.
TNFα
Six trials provided data about TNFα (15, 16, 23, 24, 27, 28). Again a significant beneficial effect was observed, with a WMD of −0.64 pg/mL (95% CI, −1.21 to −0.06), as shown in Supplementary Fig. S1.
IL8
The interleukin, IL8, also showed a very similar response. The WMD was −0.49 pg/mL, which was statistically significant (95% CI, −0.89 to −0.09), as shown in Supplementary Fig. S2. This was based on the pooled data from four trials (16, 24, 26, 28, 29), one of which was reported in two publications.
IL2
Two trials, one of which was reported in two publications, reported the effect of exercise on IL2 (16, 26, 29). A significant benefit was observed for IL2 with a mean difference of 1.03 pg/mL (95% CI, 0.40 to 1.67), as shown in Supplementary Fig. S3.
CRP
No significant effect was observed for CRP (WMD −0.15; 95% CI, −0.56 to 0.25) based on data from two trials (15, 23), as shown in Supplementary Fig. S4.
IL10
No significant effect was observed for IL10 (WMD 0.41; 95% CI, −0.18 to 1.02) based on data from two trials (16, 28), as shown in Supplementary Fig. S5.
Adverse events
Ergun and colleagues (24) reported an adverse event: one participant was diagnosed with metastases in the exercise group.
Publication bias
A funnel plot was constructed for IL6. Egger linear regression test did not reveal any significant evidence of publication bias (P = 0.06). See Supplementary Fig. S6 for the funnel plot.
Changes in inflammatory mediators according to exercise mode
Subgroup analysis by exercise mode was conducted if two or more trials were available. Yoga interventions provided significant benefits in the modulation of IL6 and TNFα (P < 0.05). Furthermore, Tai-Chi was effective in reducing IL6. When combined, aerobic and resistance exercise tended to improve IL6, IL8, and TNFα, but these effects did not reach statistical significance. Further details about subgroup analyses are shown in Supplementary Table S1.
Meta-regression analysis
Our meta-regression analysis revealed significant linear interactions between intervention length (> 11 weeks) and duration (> 45 minutes/session) with changes in IL6 levels (P < 0.05). No statistically significant dose–response relationships were observed for year of publication, training intensity, or frequency of exercise. Figure 4 shows the dose–response relationship between exercise intervention length and changes in the effect estimate for reductions in the serum levels of IL6 in breast cancer survivors.
Sensitivity analysis
The overall results of the meta-analyses were not substantially affected by the removal of the two trials with low-quality scores (IL-6 WMD = −0.42 pg/mL; 95% CI, 1.10 to 0.17).
Discussion
Within the last decade, an increasing number of studies have demonstrated that exercise training programs are beneficial for breast cancer patients. This systematic review generated novel evidence that regular exercise reduces the serum concentrations of some proinflammatory mediators, such as IL6, in breast cancer survivors. Similar conclusions were reported in 2012 by Löf and colleagues (14) in a previous systematic review conducted of this topic. In that review, the authors observed weak to moderate evidence that physical activity interventions affect the levels of serum biomarkers (i.e., inflammatory mediators and insulin growth factors) in breast cancer survivors. A key difference between the previous systematic review (14) and our meta-analysis is that we observed significant differences in the levels of IL2, IL8, IL6, and TNFα.
The most data were obtained for the effect of exercise on IL6. Importantly, in breast cancer survivors, IL6 has been associated with symptoms of fatigue, the most common and devastating complaint among cancer survivors (30, 31), and a strong body of evidence has demonstrated that exercise improves fatigue in people with breast cancer specifically (32, 33) and in people with cancer generally (34, 35). Therefore, the results of our meta-analysis lead us to hypothesize that exercise improves fatigue by counteracting key mediators of low-grade inflammation in women with breast cancer. However, acute exposure to exercise training and its effect on the inflammatory profile are short-lived, and it is unlikely that a single bout of exercise causes any adaptive changes; the repetition of exercise appears to be required for its long-term health benefits (36).
In addition to being associated with fatigue, IL6 is also predictive of survival in people with metastatic breast cancer (37). This finding may therefore help in understanding the favorable trend in survival due to exercise in various cancer populations (38, 39). Indeed, the finding of reductions in a range of cytokines (specifically IL2, IL8, IL6, and TNFα) may have similar implications because chronic inflammation is widely recognized to play a crucial role in cancer development, progression, risk, and survival (3–6).
A novel finding in our meta-analysis was the positive effect of exercise training on the levels of IL2, which is broadly involved in the differentiation and proliferation of natural killer cells, suggesting that exercise impacts the proliferation of T and B cells and immunologic function and ultimately enhances natural killer cell activity (40). In 2008, Kintscher and colleagues (41) reported that exercise reduces body fat and increases the expression of certain inflammatory cytokines, including IL2; the authors concluded that these effects reduce the likelihood of tumor reactivation and progression. In contrast to our results, Janelsins and colleagues (26) reported nonsignificant differences in the IL2 levels after a moderately intense 12-week exercise intervention that included Tai Chi in 9 breast cancer survivors compared with nonexercise controls. These discrepancies can likely be explained by the wide range of characteristics of the treatments and disease stages of the breast cancer patients across these studies.
It is well recognized that muscular contractions during exercise induce the release of IL6, which increases the levels of IL10, thereby strengthening systemic inflammatory responses after exercise training (42). Experimental evidence has demonstrated that the circulating levels of IL10, which is released by tumor-associated macrophages, are associated with the regulation of antitumor responses and tumor growth via several pathways, such as angiogenic factors (43, 44). Li and colleagues (45) reported that improvements in the IL10 levels are associated with improved prognosis and life expectancy in breast cancer survivors. Our analysis showed that exercise can improve the serum IL10 levels, although no statistically significant changes were detected, probably due to the fact that only two studies (16, 28) evaluated this cytokine, restricting the strength of this result. Positive changes in the IL10 concentrations highlight the anti-inflammatory and immunoregulatory effects of exercise on the chronic inflammatory status of breast cancer survivors.
We observed significant reductions in the serum levels of IL8 and TNFα after exercise in women with breast cancer. Rotter and colleagues (46) concluded that, by reducing adipose tissue, exercise training reduces the expression of certain proinflammatory cytokines, such as TNFα and IL1β.
We did not observe any significant differences in the CRP levels due to exercise. This is consistent with the nonsignificant effects of exercise on CRP levels in healthy and obese people (47, 48).
Overall, the results of this review suggest that the effect of exercise training on tumor-competitive immune cells and tumor host-relevant mediators, such as cytokines, is an important mechanism that could be exploited to improve prognosis after cancer. However, further investigation is required to fully characterize the roles of cytokines, including the IL system, CRP, and TNFα, as effectors of cancer patient survival (Fig. 1).
Limitations
Although some differences in the effects of various exercise modalities were observed, these may be confounded by differences in the length, frequency, and duration of training in these studies. These discrepancies presented considerable barriers to particular subgroup analyses, such as those for disease progression, the modality of exercise (such as examining aerobic and resistance training separately), and menopausal status. Therefore, further trials that include clear documentation of these variables are warranted to strengthen the conclusions about exercise modality. The studies included in this meta-analysis recruited women of different social and clinical characteristics, including age, menopausal status, stage of breast cancer progression, and therapeutic regimen (i.e., chemotherapy, radiotherapy, or both).
Conclusion
In summary, this review demonstrated that exercise is an effective intervention for controlling low-grade inflammation, which is closely associated with carcinogenesis and the tumor microenvironment in people with breast cancer. The positive effects generated by the meta-analyses for a range of inflammatory mediators justify investigation into the mechanisms underlying these effects so that exercise training exercise can be more precisely prescribed to optimize the prognosis of people with breast cancer. In the interim, exercise training can be encouraged in people during or after breast cancer treatment: because of its known benefits on the problems that cancer induces in physical fitness, function, fatigue, depression, and quality of life (23, 24, 32, 34, 35, 48–51); because of the favorable trends observed in survival with exercise training (38, 39); and now also, given the results of this review, because of its positive effects on inflammatory mediators in the tumor microenvironment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Acknowledgments
The authors thank the Norwegian Knowledge Centre for the Health Services in Oslo, Norway, and the Department of Research at Universidad Santo Tomás in Bogotá, Colombia.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.