Background
Asthma is a multifactorial disease with genetic, environmental and inflammatory components in its etiology. The principal pathophysiology of asthma is chronic inflammation of the lower respiratory tract [
1]. Pharmacotherapy and avoidance of allergens are the primary therapies emphasized in all asthma guidelines [
2,
3]. Anti-inflammatory agents such as inhaled steroids and leukotriene receptor antagonists along with long acting bronchodilators are the mainstay of asthma pharmacotherapy; however, persistent inflammatory cell infiltration has been demonstrated even after a course of oral steroids (methyl-prednisone 40 mg daily for 14 days) [
4]. In addition, airway remodeling in established asthma responds poorly to current medicinal therapies [
5]. Current asthma therapies have not achieved asthma prevention or asthma cure and there are currently no medications that can alter the natural history of the disease [
6]. Potential long term side-effects, prohibitive costs, and suboptimal adherence to asthma medications are on-going challenges to optimal asthma control. Even for the newer and expensive asthma therapies, such as omalizumab, only a very small subpopulation of refractory or severe asthma appears to respond [
7,
8]. Treatment options are therefore quite limited for asthma and the need to search for other therapies has been recognized by many experts in the field [
4,
5,
9].
Physical exercise training is thought to be beneficial in asthma management, at least in children, but it has not been extensively studied [
10]. There are no specific recommendations on physical training type, intensity, duration, or frequency in any asthma guidelines [
2,
3]. Moreover, low physical activity in asthmatics is a reality because they usually avoid exercise [
11]. Asthma and chronic obstructive pulmonary disease (COPD) are, according to the ‘Dutch hypothesis’ [
12,
13], different manifestations of the same disease entity. In addition, asthma and COPD might share common pathogenetic pathways [
14]. Positive impacts of physical exercise training and rehabilitation in COPD have been extensively studied [
15,
16]; therefore, physical training is recommended in COPD guidelines [
15,
17,
18]. Given that asthma and COPD may share a common pathogenesis, and that both diseases are manifested by chronic airway inflammation, it is imperative that we discern the role of physical training in asthma management. This systematic review represents an endeavour to shed light on this relationship, as we assess the effects of physical training on markers of airway inflammation in asthmatics.
To the best of our knowledge, there exists no systematic review to date focusing on the effects of physical training on airway inflammation in asthmatics. A Cochrane review published in 2012 summarized the effects of physical training on asthma and included only randomized controlled trials (RCTs) [
10]. The objective of this review was to extend the evidence base by using all types of observational epidemiologic research in combination with RCTs [
19] to summarize effects of physical training on airway inflammation in asthmatics.
Methods
A separate protocol for this review was not previously published.
Data sources and search
Studies were identified by searching electronic databases, and scanning reference lists of articles. The search was applied to Medline (1948 – 2012), Embase (1947 – 2012), and adapted for Web of Science (1898 – 2012) including SCI-EX (Science citation index Expanded), SCPCI-S (Conference Proceedings Citation Index – Science), and SPCI-SSH (Conference Proceedings Citation Index). It was also adopted for Cochrane (CENTRAL), and DARE (Database of Abstract of Reviews of Effectiveness), and the Cochrane Database of Systematic Reviews (Cochrane Reviews). These databases were searched for published studies that included physical training program in asthmatics. Variants of key words such as “Physical Education and Training”, “physical activity”, “physical training” were used. The search was run on April 19th, 2012. As we planned to investigate both humans and animal models, the search was not restricted to human subjects. No language or region restrictions were applied. The search was peer-reviewed by a librarian at The Ottawa Hospital, Ottawa, Canada.
Full electronic search strategy
The following search strategy was applied to Medline and Embase (peer-reviewed by a librarian at The Ottawa Hospital, Ottawa, Canada):
(exp Asthma/ OR asthma.tw) AND (exp Exercise/ or exp Exercise Therapy/ or exp Exercise Movement Techniques/ or exp Physical Conditioning, Animal/ or exp "Physical Education and Training"/ or exp Physical Exertion/ or (exercis$ or aerobic train$ or physical activi$).tw) AND (inflammation or inflammatory).tw or Inflammation/ or exp Inflammation Mediators/or bronchial biopsy.mp or Cell Count/ or Sputum/ or Adenosine Monophosphate/ or Methacholine Chloride/ or Histamine/ or tumstatin.mp or Basement Membrane/ or Collagen/ or Collagen Type IV/ or Immunoglobulin E/ or T-Lymphocytes, Regulatory/ or eicosanoids/ or leukotrienes/ or prostaglandins/ or thromboxanes/ or cytokines/ or chemokines/ or interleukin-8/ or platelet factor 4/ or interferons/ or interferon type i/ or interferon-gamma/ or interleukins/ or interleukin-1/ or interleukin-2/ or interleukin-3/ or interleukin-4/ or interleukin-5/ or interleukin-6/ or interleukin-7/ or interleukin-9/ or interleukin-11/ or interleukin-12/ or interleukin-13/ or interleukin-15/ or interleukin-16/ or interleukin-17/ or interleukin-18/ or lymphokines/ or leukocyte migration-inhibitory factors/ or macrophage-activating factors/ or transforming growth factor beta/ or tumor necrosis factors/ or tumor necrosis factor-alpha/ or endothelial growth factors/ or fibroblast growth factors/ or platelet-derived growth factor/ or tolloid-like metalloproteinases/ or transforming growth factors/ or cells/ or antigen-presenting cells/ or granulocytes/ or basophils/ or eosinophils/ or neutrophils/ or leukocytes, mononuclear/ or cytokine-induced killer cells/ or killer cells, lymphokine-activated/ or monocytes, activated killer/ or t-lymphocytes, cytotoxic/ or lymphocytes/ or killer cells, natural/ or lymphocyte subsets/ or t-lymphocyte subsets/ or t-lymphocytes/ or cd4-positive t-lymphocytes/ or cd8-positive t-lymphocytes/ or natural killer t-cells/ or monocytes/ or fibroblasts/ or mast cells/ or epithelial cells/ or goblet cells/ or phagocytes/ or histiocytes/ or Nitric Oxide/ or lung inflammation.mp or Th2 Cells/ or Bronchoalveolar Lavage Fluid/ or airway inflammation.mp. or Bronchial Hyperreactivity/).
Study selection
A scoping search revealed few studies in this area pertaining to either humans or animal models of asthma; therefore, we decided to investigate this question for both animal models and human asthmatic subjects of all age groups (the former published separately [
20]). For this review, the inclusion criteria were 1) subjects with asthma as per the Canadian Thoracic Society Asthma Guidelines [
3], or the Global Initiative for Asthma (GINA) guidelines, or physician diagnosed asthma [
2]); 2) the intervention - a physical training program; and 3) at least one marker of airway inflammation evaluated at the end of the training program. Importantly, we included all observational epidemiological research studies and RCTs. In addition, we also included studies satisfying inclusion criteria published in abstract forms only. We generated a list of candidate markers of airway inflammation based on literature searches, but allowed markers not on this list if there was appropriate justification.
Data extraction and quality assessment
The search was applied to each database and the results were combined. Duplicate articles were removed. For the first search stage, two authors (V.L. and A.B.) independently scanned the titles and abstracts for selecting studies. At the first stage, studies were included if insufficient information was available to reliably exclude them. For the second stage, the full text of each study was obtained, and both authors independently applied the inclusion and exclusion criteria blinded to author and publication data. Studies meeting the inclusion/exclusion criteria were included for data extraction. Finally, the reference lists of all studies from stage 2 were reviewed to ensure that all relevant studies were considered. If there were any disagreements as to whether an article should be included or excluded, a third reviewer (S.P.) was called upon to reach a consensus. Human studies were used for this review, and animal asthma model studies were set aside for a separate review [
20].
Data extraction forms were created and piloted on six articles obtained from the initial scoping search. The data extraction forms were then modified to facilitate extraction of the relevant data. The data was extracted independently by two authors (V.L. and A.B.) blinded to the study authors, and publication date. Any disagreements were resolved by consensus after reviewing the study. The accuracy of the information was not verified with the authors of the original studies. Studies where the data was represented in graphical format only, or if there was insufficient raw numerical data to perform meta-analysis, study authors were contacted [
21‐
25]; we successfully obtained further information only for three of these studies [
21‐
23].
The following data was extracted from each of the included studies if available: study design, characteristics of subjects (including method of diagnosis of asthma, baseline asthma severity, baseline medications, atopy, body mass index, smoking status, baseline exercise routine), physical training intervention/s (type of training, intensity and duration of each session, frequency of sessions, duration of training program), control intervention (if any), list of markers of airway inflammation, and the change in the evaluated markers of airway inflammation with physical training.
To determine the validity of the included trials one author (A.B.) applied the Cochrane risk of bias tool to RCTs [
26] and the Ottawa Newcastle tool for all observational epidemiologic research studies to assess study-level and outcome-level risk of bias [
27].
Data synthesis and analysis
We anticipated heterogeneity in studies in the primary outcome measures, methods of assessing outcome measures, and study designs. It is difficult to interpret pooled or summary measures in the presence of significant heterogeneity [
28]. We decided a priori to avoid pooling of the data in anticipation of heterogeneity of the studies, specifically heterogeneity of airway inflammatory markers studied as outcome measures. Hence, no pooled estimates of effect were calculated.
We grouped the studies measuring similar markers of airway inflammation as outcome measures creating forrest plots for their mean differences to aid visual interpretation of effect estimates. We converted medians to means where the data appeared normally distributed in the graphs. Using appropriate formulae, we converted interquartile ranges, measures of variation, and 95% confidence intervals to standard deviations where the data appeared normally distributed [
28].
Though many studies reported on exercise-induced bronchoconstriction, this outcome was measured in a myriad of different ways hence the results were described only qualitatively.
It is unknown if there was any selective reporting bias, as protocols for individual studies were not widely available. Funnel plots were not performed because, except EIB, there were less than ten studies reporting on any single outcome [
29]. Though there were 13 studies reporting on EIB, data was heterogeneous.
Data analysis was performed using the software Comprehensive Meta-Analysis version 2.0. [
30].
Discussion
Physical exercise has multi-faceted benefits in health and disease [
61,
62] and is therefore part of many guidelines for chronic diseases [
63]. Physical exercise is a suggested strategy for targeting T-cells in diseases like asthma [
64], and has been shown to reduce airway inflammation in animal models of asthma [
20]. It remains unclear, however, whether physical exercise reduces airway inflammation in human asthmatics. Specific recommendations on physical exercise dose, duration and frequency hence are not included in current asthma guidelines [
2,
3].
Physical exercise does seem to have beneficial effects on reducing airway inflammation as shown by reduction in CRP [
51,
57,
60]. CRP is a systemic inflammatory marker, used often for disease monitoring during treatment. In asthmatic subjects, in absence of any other systemic inflammatory disease, reduction in CRP could presumably be due only to a reduction in airway inflammation.
The results from single studies do demonstrate that there is a signal for reduction in airway inflammation in response to physical training; however, we cannot conclude the benefits based upon the data from single studies. Nevertheless, the effects on reactive oxygen species after physical training are important: there is reduction in malondialdehyde (MDA) and total plasma nitric oxide (NO), no further increase in SOD beyond that seen with ICS alone, and increase in GSH-Px with the combination of ICS and exercise [
53]. Similarly, the finding of decreased total sputum cells and eosinophils in asthmatics post-training [
56], decreased serum endothelin-1 [
49], and increased adiponectin and decreased leptin post-training [
60] are significant and could be due to reduction in airway inflammation. Adiponectin and leptin however could represent an increased systemic inflammatory milieu from obesity [
65,
66]. An increase in the prevalence of asthma and obesity has been linked [
67‐
69], however the directionality of this relationship is unclear; it is yet unknown whether the presence of asthma leads to obesity or whether obesity leads to development of asthma.
Multiple studies have evaluated fractional excretion of Nitric Oxide (FeNO) (Figure
2A), bronchial hyperresponsiveness to methacholine or histamine (Figure
2B) and exercise-induced bronchoconstriction (EIB) with mixed results. Noninvasive inflammatory markers in asthma such as FeNO demonstrated heterogeneous results. As such, the literature remains unconvincing and inconclusive on the role of FeNO in asthma management, as FeNO could be affected by non-disease related factors such as atopy, height/age and infection [
70,
71]. Though there were eight studies estimating bronchial hyperresponsiveness, we could not combine the results in a forest plot as bronchial hyperresponsiveness was measured to either methacholine or histamine. Six of these eight studies did not show any significant change which could be due to the non-specific nature of the agent provoking bronchial hyperreactivity. Bonsignore et al. demonstrated that airway response to methacholine and EIB decreased after 12 weeks of training with or without montelukast [
44]. Improvements in EIB, if they were real and not related to a decreased ventilator load post training, suggest that the mechanism of provocation in exercise is different compared to methacholine [
22,
23,
44] and that the former can be positively impacted by training. These results are not conclusive because methods used in these studies varied significantly and were not standardized.
Finally, the method of diagnosis of asthma was not uniform in all the studies which could introduce selection bias (please see the Additional file
1). Establishing a diagnosis of asthma can be challenging even at a tertiary care center because of its variable natural history. Even though systematic, algorithmic diagnosis of asthma is cost-effective, asthma misdiagnosis could be as common as 30% [
72‐
74]. Asthma misdiagnosis in the studies included in this review might drive airway inflammation outcome data towards null if some truly non-asthmatic subjects were misclassified as asthmatics; and hence might nullify effects of physical training.
This systematic review and the Cochrane Review [
10] have demonstrated the main problem: there are very few RCTs and observational studies that have evaluated the effects of physical exercise training on asthmatic subjects and all are plagued by small sample sizes (10;20). One non-randomized controlled study [
75] with 15 adult asthmatics demonstrated a clinically significant improvement in Asthma Control (0.5 increase) as measured by the Asthma Control Questionnaire (ACQ) and improvements in quality of life, as measured by the Asthma Quality of Life Questionnaire (AQLQ) with the structured-physical exercise program. The US Nurses’ Health Study cohort database demonstrated prospectively in 2818 asthmatics that physical exercise (median physical activity = walking at a brisk pace for 20 minutes three times per week) might be associated with lower risk of asthma exacerbation irrespective of asthma severity and other covariates [
76]. This database study demonstrated that there was a dose–response relationship; higher the level of physical activity lower the risk of asthma exacerbation [
76].
This systematic review provides important insight into the role of physical training in the treatment of asthma in children and adults. There is some evidence that physical training might provide beneficial effects by reducing airway inflammation in asthmatics, though this cannot be conclusively proven from the existing literature. Physical training seems to be safe in asthmatics [
10] though we do not know the safety limits of duration, intensity, and frequency.
Strengths and limitations of the review
The current review has several strengths. We have employed comprehensive search strategies using multiple databases of published literature and employed standardized methods of conducting systematic reviews. Despite our comprehensive efforts, it is possible that we have missed some relevant studies in the process of publication or other unpublished sources.
Heterogeneity of the included studies was predicted and confirmed and hence data was not pooled. We described the data, when possible, using forest plots. Many studies did not use standardized methods of testing (e.g. testing protocols for EIB), thereby limiting the conclusions that could be drawn. In addition, reporting was not uniform across studies, especially e.g. asthma diagnosis, asthma medications, asthma severity and control. Finally, different markers of airway inflammation were used in different studies, leading to few studies reporting on any single outcome. Most importantly, physical training intervention(s) (type of training, intensity and duration of each session, frequency of sessions, duration and site of training program) and control intervention(s) were heterogeneous across studies. These issues combined to limit the strength of our conclusions.
Conclusions
Effectiveness of physical exercise on airway inflammation is yet unproven. There is, however, some evidence to suggest that physical exercise may reduce inflammation in an asthmatic airway. Should such evidence emerge in future research, physical exercise may well prove to be an affordable, accessible, healthy, non-invasive, and enjoyable asthma management strategy. We urgently need to design studies to understand if this non-pharmacological adjunctive strategy can effectively decrease airway inflammation, control symptoms of asthma, and be safely administered in the community.
Acknowledgements
We would like to acknowledge the following people for their contributions: Dr. D. Moher (The Ottawa Hospital Research Institute) for his thoughtful input during the systematic review process and manuscript edits; Dr. M. Emtner (Uppsala University, Sweden) for supplying raw data from her studies and Ms. Risa Shorr (The Ottawa Hospital Library) for her help with the search strategies and peer review of searches.
Financial
Dr. S. Pakhale is supported by the Ottawa Hospital Research Institute and the Department of Medicine, The Ottawa Hospital, Ottawa, Canada. Dr. Moher is supported by a University of Ottawa Research Chair. There was no role of the funders in conducting this systematic review. All authors declare no conflict of interest.
Non-financial
None for all authors.
Competing interests
No competing interests for any author.
Authors’ contributions
SP had full access to the data and takes full responsibility for the integrity of the data and the accuracy of the data analysis, contributed to the concept, design, implementation, statistical analysis, interpretation and writing. VL contributed to the data management, statistical analysis, interpretation and writing. AB contributed to the data management, statistical analysis, interpretation and writing. LT contributed to the data analysis, interpretation and writing. All authors read and approved the final manuscript.