Antioxidant supplementation for lung disease in cystic fibrosis
Abstract
Background
Airway infection leads to progressive damage of the lungs in cystic fibrosis and oxidative stress has been implicated in the etiology. Supplementation of antioxidant micronutrients (vitamin E, vitamin C, ß‐carotene and selenium) or glutathione may therefore potentially help maintain an oxidant‐antioxidant balance. Current literature suggests a relationship between oxidative status and lung function.
Objectives
To synthesize existing knowledge of the effect of antioxidants such as vitamin C, vitamin E, ß‐carotene, selenium and glutathione in cystic fibrosis lung disease.
Search methods
The Cochrane Cystic Fibrosis and Genetic Disorders Group's Cystic Fibrosis Trials Register and PubMed were searched using detailed search strategies. We contacted authors of included studies and checked reference lists of these studies for additional, potentially relevant studies.
Last search of Cystic Fibrosis Trials Register: 29 August 2013.
Selection criteria
Randomized controlled studies and quasi‐randomized controlled studies of people with cystic fibrosis comparing antioxidants as listed above (individually or in combination) in more than a single administration to placebo or standard care.
Data collection and analysis
Two authors independently selected studies, extracted data and assessed the risk of bias in the included studies. We contacted trial investigators to obtain missing information. Primary outcomes are lung function and quality of life; secondary outcomes are oxidative stress, inflammation, nutritional status, days on antibiotics and adverse events during supplementation. If meta‐analysed, studies were subgrouped according to method of administration and the duration of supplementation.
Main results
One quasi‐randomized and nine randomized controlled studies were included, with a total of 436 participants. Eight studies analyzed oral supplementation with antioxidants and two inhaled supplements.
One study (n = 46) of an oral combined supplement demonstrated a significant difference in forced expiratory volume at one second expressed as per cent predicted after two weeks in favour of the control group, mean difference ‐4.30 (95% confidence interval ‐5.64 to ‐2.96); however a further study (n = 41) of oral supplementation with glutathione showed a significant improvement in this outcome and in forced vital capacity after six months from the treatment start, mean difference 17.40 (95% confidence interval 13.69 to 21.11) and 14.80 (95% confidence interval 9.66 to 19.94) respectively. The combined supplement study also indicated a significant improvement in quality of life favouring control, mean difference ‐0.06 points on the quality of well‐being scale (95% confidence interval ‐0.12 to ‐0.01). Based on one study (n = 41) of oral glutathione supplementation in children, the supplements had a positive effect on the nutritional status (body mass index %) of the patients, mean difference 17.20 (95% confidence interval 12.17 to 22.23). In two studies (n = 83) that supplemented vitamin E, there was an improvement after two months in the blood levels of vitamin E, mean difference 11.78 μM/L (95% confidence interval 10.14 to 13.42).
Based on one of the two studies of inhaled glutathione supplementation, there was an improvement in the forced expiratory volume at one second expressed as per cent predicted after three and six months (n = 153), mean difference 2.57 (95% confidence interval 2.24 to 2.90) and 0.97 (95% confidence interval 0.65 to 1.29) respectively. Only one of the studies reported quality of life data that could be analysed, but data showed no significant differences between treatment and control.
None of the 10 included studies was judged to be free of bias.
Authors' conclusions
There appears to be conflicting evidence regarding the clinical effectiveness of antioxidant supplementation in cystic fibrosis. Based on the available evidence, glutathione (administered either orally or by inhalation) appears to improve lung function in some cases and decrease oxidative stress; however, due to the very intensive antibiotic treatment and other treatments that cystic fibrosis patients receive, the beneficial effect of antioxidants is very difficult to assess in patients with chronic infection without a very large population sample and a long‐term (at least six months) study period. Further studies, especially in very young patients, examining clinically relevant outcomes, dose levels, timing and the elucidation of clear biological pathways by which oxidative stress is involved in cystic fibrosis, are necessary before a firm conclusion regarding effects of antioxidants supplementation can be drawn.
PICOs
Plain language summary
What are the effects of vitamins E and C, beta‐carotene, selenium and glutathione on lung disease in people with cystic fibrosis?
Background
Frequent chest infections in people with cystic fibrosis cause long‐lasting inflammation in their lungs. Cells causing inflammation produce a kind of oxygen molecule (reactive oxygen species (ROS)) which can easily harm proteins and DNA (oxidative damage). To fight these effects, the body may produce antioxidants. The genetic defect in cystic fibrosis leads to an imbalance favouring the high production of harmful ROS over the low level of protective antioxidants. Antioxidant supplements might help reduce the oxidative damage in the lungs from constant infection and build up low levels of antioxidants.
There are many different antioxidants. Vitamin E and beta‐carotene are fat‐soluble and levels are low in people with cystic fibrosis because of the problems they have absorbing fat. Glutathione is the most abundant antioxidant in cells, but in people with cystic fibrosis it is not released into the lungs properly. Some enzymes which help antioxidants work are dependent on a mineral called selenium, so selenium supplements aim to stimulate antioxidant action. Vitamin C is water‐soluble and decreases with age in people with cystic fibrosis, so vitamin C supplements aim to rebuild these levels.
Most supplements are taken by mouth, but glutathione can also be breathed directly into the lungs.
Search date
We last looked for evidence on 29th August 2013.
Study characteristics
We included 10 studies with 436 people with cystic fibrosis (almost equal numbers of males and females) aged from six months to 32.9 years. Eight studies compared oral supplements to placebo (a supplement appearing the same as the treatment but containing no medicine) and two compared inhaled supplements to placebo.
Key results
The main outcomes were lung function and quality of life; other outcomes were markers of oxidative stress, inflammation, body mass index, days on antibiotics and adverse events.
In one study (46 volunteers), a combined supplement showed a better score for forced expiratory volume at one second (% predicted) in the placebo group compared to the supplement group at two weeks, but a study of oral glutathione (41 children) showed that after six months people taking antioxidants had on average a 17.40% better score for this measure of lung function compared to the placebo group. The same study also showed a 14.80% better score for forced vital capacity for the supplemented group. The combined supplement study showed a improvement in quality of life in the placebo group. The oral glutathione study showed that those receiving glutathione had a body mass index on average 17.20% higher than the placebo group by the end of the study. All studies (regardless of the supplement) showed improvements in the blood levels of vitamin E. Those studies supplementing beta carotene and selenium also showed improvements in blood levels of those antioxidants.
The largest of the two inhaled glutathione studies (153 patients) showed that, compared to the placebo group, patients inhaling glutathione had on average a 2.57% better score for forced expiratory volume at one second (% predicted) after three months and a 0.96% better score after six months. We analysed quality of life data from this study, but found no difference between supplement and placebo groups.
Conclusions
It is too early to judge the effects of antioxidant supplements as the results of this review are conflicting and it is difficult to tell which changes are due to antioxidants and which are due to other treatments (e.g. antibiotics). Glutathione (either oral or inhaled) appears to improve lung function in some cases and lower oxidative stress. Larger studies, especially in very young patients, should look at important clinical outcomes for at least six months before firm conclusions regarding the effects of antioxidant supplements can be drawn.
Quality of the evidence
None of the studies was free of possible bias. Most problems were because data were not fully reported and we thought this likely to affect the results. We were mostly unsure if volunteers knew in advance which group they were going to be in and if they knew once the trials started whether they received the supplements or placebo. We are not sure how this might affect the results.
Authors' conclusions
Background
Description of the condition
Cystic fibrosis (CF) is the most prevalent inherited, life‐limiting disorder in Caucasian populations affecting approximately one in 2000 births. It is estimated that the present number of CF cases is 35,000 in Europe, 30,000 in North America and 3000 in Canada (CCFF 2002; CFF 2005). Approximately 1000 new cases of CF are diagnosed in the USA each year. The median predicted lifespan for people with CF has risen steadily over the last 25 years. Since 2002, the median predicted survival age has increased by almost 10 years; from 31.3 years in 2002 to 41.1 years in 2012 (CFF 2012). Most people with CF are diagnosed before the age of two years. Today, in several countries, CF is typically diagnosed shortly after birth through newborn screening programs and this early diagnosis may have played an important role in improving survival. Since 2010, all newborns have been screened for CF in the USA. Research shows that people with CF who are diagnosed through the newborn screening programs have a higher weight and healthier lungs later in life than those diagnosed at a later time point because of CF symptoms (CFF 2012).
There are more than 1500 mutations in the cystic fibrosis transmembrane conductance regulator gene (CFTR) on chromosome 7, which lead to a malfunction of the chloride channel in people with CF. This malfunctioning of the chloride channel in people with CF leads to a decreased volume of the paraciliary fluid in the lower respiratory tract, which in turn leads to impaired mucociliary clearance of inhaled microbes. This impairment of the non‐inflammatory defence mechanism of the respiratory tract leads to early recruitment of inflammatory defence mechanisms such as polymorphonuclear leukocytes (PMN) and antibodies. However, in spite of an inflammatory response and intensive antibiotic therapy, infections caused by particularly Pseudomonas aeruginosa (P. aeruginosa) persist and lead to respiratory failure or death. This pathogen is able to survive by switching to the biofilm mode of growth, which provides tolerance to the inflammatory defence mechanisms and antibiotic treatment.
Therefore, from early childhood, people with CF have recurrent and chronic respiratory tract infections characterised by PMN inflammation. Counts of PMNs in CF airway fluid have been found to be thousands of times higher than normal. A consequence of the PMN‐dominated inflammation is the release of proteases and reactive oxygen species (ROS), which are believed to be the main modulators of tissue damage in CF. Besides the increased production of ROS, people with CF have an impaired absorption of dietary antioxidants in the gut and the inability of cells bearing mutant CFTR to efflux glutathione (GSH) ‐ the most abundant intracellular antioxidant ‐ into the extracellular milieu of the lung. As a primary water‐soluble antioxidant, GSH performs several important functions in the epithelial lining fluid (ELF) by directly scavenging hydrogen peroxide and other free radicals (Kelly 1999). In this process, GSH is oxidized to glutathione disulfide (GSSG). Furthermore, GSH also plays multiple, pivotal roles in the immune system as normal intracellular levels of GSH are essential for chemotaxis, phagocytosis, oxidative burst etc. In CF, reduced levels of total GSH (GSH + GSSG) and an increased GSSG to GSH redox ratio may partly explain the chronic and excessive inflammation in the respiratory system.
Thus, in CF, the source of oxidative stress is due to the imbalance between increased ROS production and impaired antioxidant systems. It is thought that ROS, which are the key players in oxidative stress, cause tissue damage in the lungs by attacking e.g. polyunsaturated fatty acids (PUFAs) in cell membranes. These PUFAs are one of the main components of dietary fats and are converted to arachidonic acid, a component of phospholipids in cell membranes. It is thought that ROS attack phospholipids (peroxidation) and produce a free radical, which in turn initiates attacks on adjacent arachidonic acid chains, thus compromising cell‐membrane structure. Free radical damage is propagated until host defence systems counteracts and terminates these actions. The peroxidation products of arachidonic acid are F2‐isoprostanes and these have become the gold‐standard indicator of oxidative stress in vivo (Mayne 2003). The mechanism of peroxide generation, propagation and termination is shown in the figures (Figure 1).
Peroxide chain reaction characterized by initiation, propagation and termination. (RH: PUFA; R·: free radical; ROO·: peroxide; ROOH: hydroxyl peroxide; AH: vitamin E; A·: oxidized Vitamin E. Adapted from: Tappel AL. Vitamin E and free radical peroxidation of lipids. Annals of the New York Academy of Sciences. 1972; 203(1):12‐28.
Description of the intervention
Unusually high levels of oxidative stress in CF deplete the host‐defense system, which includes exogenous antioxidant micronutrients vitamin E, vitamin C, ß‐carotene and selenium and the major endogenous antioxidant, GSH. Supplementation of these micronutrients or of GSH, alternatively referred to as free‐radical scavengers, may help in preventing the unfavorable shift towards redox imbalance observed in people with CF.
Since the CFTR channel is the major mechanism of GSH efflux into the extracellular milieu of the lung from lung epithelial cells, this efflux is severely compromised in CF resulting in GSH system dysfunction. People with CF experience GSH deficiency both locally in the ELF of the lung but also as a systemic GSH deficiency in blood (Roum 1993).
Pilot studies have shown that it is possible to replete alveolar GSH after GSH inhalation therapy and several clinical trials employing GSH or a GSH precursor such as N‐acetyl‐cysteine as an intervention have resulted in improved clinically‐relevant markers in CF (Tirouvanziam 2006). Therefore, both local treatment (inhalations) and systemic administration (orally) of GSH or GSH precursor have been proposed to be beneficial for people with CF.
Although many other antioxidants exist, vitamin E, vitamin C, ß‐carotene, selenium and glutathione have been chosen in this review due to their well‐defined antioxidant properties, mechanisms of action and long history of study in the body (Rock 1996). Other, more recently proposed antioxidants include, for example, other carotenoids (lycopene, zeaxanthin, lutein), melatonin and retinol (Pryor 2000). People with CF are largely affected by malfunctioning pancreatic enzymes that, despite enzyme supplements and high‐fat diets, prevent the absorption of fat from the digestive tract, and consequently, the fat‐soluble vitamins E and β‐carotene. Lowered plasma antioxidant status of vitamin C and decreased activity of erythrocyte glutathione peroxidase (GSHPx), an antioxidant enzyme dependent on the mineral selenium, have also been reported in people with CF (Benabdeslam 1999; Wood 2001). As such, vitamins E and C, ß‐carotene and selenium as well as glutathione comprise the antioxidant interventions that will be assessed in this review; as their mechanisms of action are sufficiently different, they are subgrouped accordingly.
How the intervention might work
Literature suggests that a relationship exists between oxidative stress status and lung function. Specifically, elevated levels of oxidative stress and inflammatory stress indicators with corresponding reduced lung function have previously been found in individuals with CF (Brown 1994; Brown 1996; Mayer‐Hamblett 2007; Wood 2001). Such indicators (oxidative and inflammatory markers) are often used as surrogate outcomes of lung function in respiratory research (Montuschi 1998; Repine 1997; Schunemann 1997). Lung function status or improvements, or both, are also routinely reported in the literature. Due to the chronic and progressive nature of CF, clinical benefits of antioxidant therapy may be difficult to determine.
Why it is important to do this review
A synthesis of all available clinical trials on the effects of antioxidants on lung disease will indicate the relevance of antioxidants to health status in people with CF and will guide future therapeutic decisions. Currently, fat‐soluble vitamins (vitamins A, D, E and K) are routinely supplemented in CF to prevent deficiencies associated with fat malabsorption; however, the therapeutic use of antioxidants, such as vitamins C and E, β‐carotene, selenium and glutathione is limited. Vitamin A supplementation is the subject of a recent Cochrane Review (Bonifant 2012), which aimed to establish whether supplementation reduced the frequency of vitamin A deficiency disorders, improved general and respiratory health or increased the frequency of vitamin A toxicity; the review did not identify any eligible studies. Reviews of vitamin D and vitamin K supplementation have also been published (Ferguson 2012; Jagannath 2013). The present review aims to establish whether antioxidant supplementation with micronutrients such as vitamins C and E, β‐carotene, selenium or with glutathione are promising adjunct therapies in CF.
Objectives
The central objective of the review is to synthesize existing knowledge on the effect of antioxidants on lung function through inflammatory and oxidative stress markers in people with CF.
Methods
Criteria for considering studies for this review
Types of studies
Included studies were controlled clinical trials (randomized (RCTs) and quasi‐randomized (CCTs)).
Types of participants
Studies of all people of either gender reporting a CF diagnosis and all degrees of severity (Pellegrino 2005), including those who have undergone lung transplant, were considered eligible for inclusion.
Types of interventions
The interventions considered were antioxidants including vitamin E, vitamin C, ß‐carotene, selenium and glutathione in more than a single administration, with any route of administration and solubility taken individually or in combination compared to placebo or standard medication or care.
Types of outcome measures
Primary outcomes
-
Lung function tests (e.g. forced expiratory volume at one second (FEV1) (% predicted or litres), forced vital capacity (FVC) (% predicted or litres))
-
Quality of life (QOL) (using validated measurement tools only)
Secondary outcomes
-
Oxidative stress markers in serum
-
hydrogen peroxide (H2O2 exhalation)
-
lipid peroxidation (F2 isoprostanes)
-
antioxidant enzyme function (post hoc change)
-
potency (post hoc change)
-
plasma antioxidant status
-
plasma fatty acids
-
-
Inflammation
-
inflammatory markers (i.e. IL‐6, IL‐8, TNF‐α, IL‐1β)
-
hyperinflation of chest
-
-
Nutritional status (e.g. body mass index (BMI) or BMI percentile for children)
-
Pulmonary exacerbations requiring intravenous antibiotic therapy or hospitalization
-
Adverse events
Since measures of oxidative stress reported were not confined to those anticipated, a post hoc decision was made to include all reported markers of oxidative stress encountered. We categorized oxidative stress outcomes using the classification scheme defined by Dotan (Dotan 2004). Since multiple oxidative stress outcomes exist and within each outcome multiple measures have been identified to quantify the same outcome, oxidative stress was collected as follows.
-
Lipid peroxidation products (F2‐isoprostanes, malondialdehyde (MDA) or thiobarbutic acid reactive substances (TBARS, an unspecific measure of lipid peroxidation), hydroperoxides (H2O2))
-
Promoters (Luminol)
-
Inhibitors (i.e. antioxidant micronutrients and enzymes)
-
Potency (i.e. trolox‐equivalent antioxidant capacity (TEAC))
-
Oxidizability (i.e. lag time, propagation)
We also decided to collect data for antioxidant enzymes as measured by erythrocyte glutathione peroxidase (GPX), which is a selenium‐dependent enzyme, and superoxide dismutase (SOD).
"Pulmonary exacerbations requiring intravenous antibiotic therapy or hospitalization" was revised to "days of antibiotic therapy" after data extraction began and data were found to be presented in the latter manner rather than the former.
Search methods for identification of studies
No language restrictions were imposed in the process of identifying studies.
Electronic searches
Relevant trials were sought from the CF Trials Register using the terms: antioxidants. The terms 'vitamin C' and 'glutathione' were not indexed keywords within the register and therefore could not be searched for.
The Cystic Fibrosis Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of The Cochrane Library), quarterly searches of MEDLINE, a search of Embase to 1995 and the prospective handsearching of two journals ‐ Pediatric Pulmonology and the Journal of Cystic Fibrosis. Unpublished work is identified by searching the abstract books of three major cystic fibrosis conferences: the International Cystic Fibrosis Conference; the European Cystic Fibrosis Conference and the North American Cystic Fibrosis Conference. For full details of all searching activities for the register, please see the relevant sections of the Cystic Fibrosis and Genetic Disorders Group Module.
Date of the latest search of the CF Trials Register: 29 August 2013.
PubMed (1950 to August 2013) has been searched. Previously other databases were searched such as CINAHL (1937 to 21 Dec 2007) and AMED (1985 to 21 Dec 2007), details of which can be found in the following appendices (Appendix 1; Appendix 2; Appendix 3). The new team who have taken on this review in 2013 do not have access to these databases and are of the opinion that the searches of the Cystic Fibrosis and Genetic Disorders Group's Cystic Fibrosis Trials Register and the search of PubMed will capture the literature of interest.
A search of the Register of international standard randomised controlled trial numbers (ISRCTN) and clinicaltrials.gov for relevant ongoing studies was run on 11 July 2014 with the search terms: 'cystic fibrosis and antioxidants' and 'cystic fibrosis and oxidative stress'.
Searching other resources
We checked the bibliographies and contacted investigators of included studies for possible references to previously unidentified RCTs (published or unpublished) for inclusion that may have been missed.
We have received additional data from the authors of three studies (Griese 2013; Keljo 2000; Visca 2013); these data have been used in the analysis.
Data collection and analysis
Selection of studies
The two authors (OC and JL) assessed studies independently for inclusion into the review. In the case of conflict of opinion between the two authors, they resolved this by discussions until they reached a common agreement. The first stage of screening included systematically screening electronic titles or abstracts (or both) of all studies according to the pre‐specified criteria. The authors then reviewed the full‐text hard copies, again applying selection criteria.
Data extraction and management
The two authors (OC and JL) extracted data independently for all outcomes of interest using pre‐developed extraction forms. In the case of conflict of opinion between the two authors, they resolved this by discussions until they reached a common agreement.
The authors presented different routes of administration as separate comparisons. If one study compared two arms of an antioxidant intervention to control, the authors combined the intervention arms using appropriate statistical methods (seeUnit of analysis issues).
The concentrations of vitamin E in the older studies (Harries 1971; Levin 1961) were expressed as mg/100 ml and have been recalculated in this review to μmol/L by using the formula 1 mg/100 ml = 23.22 μmol/L in order to be compared to the values presented in the newer studies (Visca 2013).
Assessment of risk of bias in included studies
Two authors independently assessed the risk of bias of each study following the domain‐based evaluation as described in the Cochrane Handbook for Systematic Reviews of Interventions 5.1 (Higgins 2011a). The tool for assessing risk of bias in each included study comprises a judgement and support for the judgement for each entry in a 'Risk of bias' table, where entry addresses a specific feature of the study. The judgement for each entry assesses the risk of bias as low, high or unclear risk, with the last category indicating either lack of information or uncertainty over the potential for bias. In the case of conflict of opinion between the two authors, they resolved this by discussions to lead to a common agreement.
We assessed the domains listed below.
-
Randomisation
-
Concealment of allocation
-
Blinding (of participants, personnel and outcome assessors)
-
Incomplete outcome data (whether investigators used an intention‐to‐treat analysis)
-
Selective outcome reporting
-
Other potential threats to validity
Measures of treatment effect
For binary outcomes, we planned to report relative risks (RR) and 95% confidence intervals (CIs). When possible, we reported the proportion of participants reporting adverse events for each treatment arm. As we expected adverse events to be rare, we analysed these outcomes using the Peto odds ratio (OR) statistic and 95% CIs.
We recorded continuous outcomes as either mean relative changes from baseline or mean end‐point values and standard deviations (SD). Where standard errors (SE) were reported, these were converted to SDs. We calculated the mean difference (MD) and 95% CI for most outcome measures except for outcomes of oxidative stress for which we used standardized mean differences (SMDs) and 95% CI, since we identified multiple measures which quantify the same process.
Unit of analysis issues
Cross‐over trials
If we had been able to include cross‐over studies with sufficient data, we planned to analyse these by paired t‐test for continuous data, as long as there was no evidence of carry‐over or period effect (Elbourne 2002). Where papers reported cross‐over study data insufficiently, i.e. so that only first‐period data were available, we treated data from the first period as a parallel trial (Elbourne 2002).
Studies with multiple treatment arms
For studies reporting multiple intervention and placebo groups, we combined all relevant intervention groups and placebo groups, each to be analysed as a single group as recommended in the Cochrane Handbook for Systematic Reviews of Interventions to avoid a unit of analysis error (Higgins 2011b).
Dealing with missing data
We made up to two attempts to contact each of the authors of studies for which information was missing. If authors did not respond, we left out incomplete data.
We have received additional data from the authors of three studies (Griese 2013; Keljo 2000; Visca 2013); these data have been used in the analysis.
Assessment of heterogeneity
We planned to measure the inconsistency of study results using the Chi2 test and the I2 heterogeneity statistic to determine if variation in outcomes across trials was due study heterogeneity rather than chance (Higgins 2003). This Chi2 test assesses whether observed differences in results are compatible with chance alone. A low P value (or a large Chi2 statistic relative to its degree of freedom) provides evidence of heterogeneity of intervention effects (variation in effect estimates beyond chance). A P value of 0.10, rather than the conventional level of 0.05, is used to determine statistical significance.
The I2 statistic, as defined by Higgins (Higgins 2011a), measures heterogeneity as a percentage (%) where a value:
-
0% to 40%: might not be important;
-
30% to 60%: may represent moderate heterogeneity;
-
50% to 90%: may represent substantial heterogeneity;
-
75% to 100%: considerable heterogeneity.
The importance of the observed value of I2 depends on (i) magnitude and direction of effects and (ii) strength of evidence for heterogeneity (e.g. P value from the Chi2 test, or a confidence interval for I2).
Assessment of reporting biases
Using the method by Light, if we had included a sufficient number of studies (n > 10, by convention), we planned to assess publication bias using a funnel plot (Light 1994). A funnel plot is a graph that plots treatment effect for each study against a measure of precision (i.e. 1/standard error (SE)).
We present information regarding selective reporting of outcomes within individual studies in the risk of bias assessment (Risk of bias in included studies).
Data synthesis
The main comparisons were between antioxidant supplementation and control (standard of care, other therapy, no treatment). We have presented a forest plot for each outcome for which data are available. Where we have included more than one study for a single subgroup, we have pooled data into a single effect estimate. Since each antioxidant works by a different mechanism of action, we analysed each micronutrient or unique combination of micronutrients as a separate subgroup, as per the first originally planned subgroup analysis to explore methodological heterogeneity.
We intended to use a fixed‐effect model for all analyses with a low degree of heterogeneity (I2 less than 40%). We later decided to employ a random‐effects model for all analyses, since there were known differences (i.e. doses, duration and solubility of supplement) and unknown differences between trials that may potentially influence the size of the treatment effect.
All studies were analysed using the Review Manager software (RevMan 2011).
Subgroup analysis and investigation of heterogeneity
Where at least 10 studies per outcome were included (Deeks 2011), the following a priori subgroup analyses were planned to investigate both clinical and methodological heterogeneity.
Clinical heterogeneity
Planned clinical subgroups were:
-
age: pediatric (up to 18 years) versus adult (over 18 years);
-
disease severity as measured by FEV1 (70% to 80% will be considered mild; 60% to 70% moderate; 50% to 60% moderately severe; 34% to 50% severe; and less than 34% very severe as defined by American Thoracic Society guidelines (Pellegrino 2005)).
Methodological heterogeneity
Planned methodological subgroups were:
-
combined antioxidant supplementation and single antioxidant supplementation (i.e. each single micronutrient or combination thereof are listed separately);
-
antioxidant(s) alone versus antioxidant(s) alongside concurrent treatment;
-
timing of intervention: antioxidant(s) as prophylactic or therapeutic treatment.
Post hoc, it was decided that, regardless of the number of studies per outcome, individual supplements or unique combinations thereof should not be combined in a single meta‐analysis as it would not be appropriate due to the aforementioned differences between micronutrients. Therefore, sub‐grouping by supplement was employed in the meta‐analyses.
Sensitivity analysis
While the protocol for this review indicated that we would base sensitivity analysis on only randomization, allocation concealment, blinding, and intention‐to‐treat versus per‐protocol analysis, we later decided to evaluate risk of bias using the newly introduced risk of bias tool, therefore altering planned sensitivity analyses.
We planned sensitivity analyses to evaluate treatment effect by excluding trials with an overall high risk of bias.
Results
Description of studies
Results of the search
Out of 323 unique studies yielded from the search strategy, 64 remained after title and abstract screening. After full text screening, 10 studies met the inclusion criteria (Bishop 2005; Griese 2013; Harries 1971; Homnick 1995b; Keljo 2000; Levin 1961; Portal 1995a; Renner 2001; Visca 2013; Wood 2003). Three of the five studies that were listed as 'Awaiting classification' in the initial version of the review have been included in the analysis at the 2014 update (Harries 1971; Keljo 2000; Levin 1961). A further study which was also listed as 'Awaiting classification' in the initial version of the review was a single dose bioavailability study and did not meet the inclusion criteria and so was excluded (Jacquemin 2009). One study which was included in the first version of the review, has now (update 2014) been excluded as it was presenting bioavailability data after a single administration (Homnick 1995a). One study remains listed under 'Characteristics of studies awaiting classification' as it is only currently available in abstract format and, if included, may compromise the validity of results due to unavailability of a complete set of data (Wong 1988). There are two studies identified from the searches of trials registries in July 2014 which are still ongoing (Characteristics of ongoing studies).
The flow of studies through the screening process of the review is shown in the figures (Figure 2); this process uses the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) flow diagram (Moher 2009). During full‐text screening, three study reports were translated but did not meet final inclusion criteria.
Study flow diagram.
Included studies
Several of the 10 included studies had multiple papers linked to them. Two reports represented the Portal study (Portal 1995a), two reports (an article and a conference abstract) represented the Griese study (Griese 2013) and the same for the Harries study (Harries 1971); there were also two reports (an article and a letter) for the Levin study (Levin 1961). There were three reports and four abstracts representing the Renner study (Renner 2001) and one report and three abstracts representing the Bishop study (Bishop 2005). One report represented two studies, one of which was included and one excluded (Homnick 1995a; Homnick 1995b).
Trial Characteristics
Four studies were conducted in the USA (Bishop 2005; Homnick 1995b; Keljo 2000; Levin 1961), one in Australia (Wood 2003) and the remaining five studies were conducted in Europe (one in France (Portal 1995a), one in Austria (Renner 2001), one in Germany (Griese 2013), one in Italy (Visca 2013) and one in Great Britain (Harries 1971)).
Ten studies were RCTs; one did not contain any information regarding sequence generation or allocation concealment has been interpreted as a controlled clinical trial (Homnick 1995b). Eight studies were of parallel design (Bishop 2005; Griese 2013; Harries 1971; Keljo 2000; Levin 1961; Renner 2001; Visca 2013; Wood 2003). One of the included studies was of cross‐over design with each arm lasting five months and a washout period in between treatment arms lasting two months (Portal 1995a). Baseline data for the second period were not reported. Since no mean difference could be calculated using available period data, data from this period were omitted from the analysis. As such, the study was treated as a parallel‐group RCT rather than cross‐over RCT.
Time points for reporting data in the included studies ranged from one month to 50 weeks. Harries reported outcomes at one month (Harries 1971). Bishop and Wood reported outcome data at eight weeks (Bishop 2005; Wood 2003). Griese reported at one, three and six months (Griese 2013); Keljo reported outcomes at three months only (end of the study) (Keljo 2000). In the 2001 study, Renner reported at three and six months (Renner 2001), whereas Levin reported outcomes at two and six months (Levin 1961). Portal reported data at five months (Portal 1995a) and Visca at six months (Visca 2013). The longest study was by Homnick, who reported outcome data at 50 weeks (Homnick 1995b).
Eight of the 10 studies reported the source of trial funding; of these, one author (DPRM) of one study (Harries 1971) was supported by Roche Products Ltd. None of the other authors received funding from industry (Bishop 2005; Griese 2013; Homnick 1995b; Keljo 2000; Portal 1995a; Visca 2013; Wood 2003).
Participants
The 10 studies included in this review represent 436 participants. Sample sizes ranged from 20 (Homnick 1995b) to 153 participants (Griese 2013). One study reports on power considerations when choosing the number of patients included and uses cytokine levels as primary outcome (Keljo 2000). A further study reports on sample size calculation (Visca 2013) and a third study reports on calculations of the total sample size to detect changes in the primary outcome, FEV1 (Griese 2013).
The age of participants was not consistently reported in all studies, but the minimum reported age for inclusion was six months (Harries 1971) and maximum was 32.9 years (Griese 2013). Of the eight included studies of oral antioxidant supplementation, one did not report the age of participants (Homnick 1995b); four included just children, but with a large range of ages from 18 months to 14.5 years (Harries 1971; Levin 1961; Visca 2013; Wood 2003); and three included a mixture of children and adults (Keljo 2000; Portal 1995a; Renner 2001). For the two studies of nebulised supplements, one included people with CF with a mean age of 23 years (Griese 2013), while the remaining study included people with CF aged 6 to 19 years, mean age 13 years (Bishop 2005).
The gender of the participants was reported by seven studies (Bishop 2005; Griese 2013; Keljo 2000; Levin 1961; Portal 1995a; Renner 2001; Wood 2003). There were approximately equal numbers of males and females in three studies (Griese 2013; Keljo 2000; Portal 1995a). There were slightly more males in the Bishop study, 67% in the treatment group and 60% in the placebo group (Bishop 2005). This was also true for the placebo group in the Levin study (68%), but not for the treatment group (45%) (Levin 1961). Wood reported 59% males in the treatment group, but only 33% in the placebo group (Wood 2003); there were also significantly fewer males (25%) than females reported in the Renner study (Renner 2001). Details of gender split were not reported by the three remaining studies (Harries 1971; Homnick 1995b, Visca 2013).
Given the small number of studies included in the review, we were not able to split data by clinical subgroups.
Interventions
Antioxidant supplements were either oral (Harries 1971; Homnick 1995b; Keljo 2000; Levin 1961; Portal 1995a; Renner 2001; Visca 2013; Wood 2003) or nebulised (Bishop 2005; Griese 2013).
There were not enough data to examine other planned methodological subgroups. Data were grouped according to combined and single supplementation such that each unique micronutrient or combination thereof were included in separate subgroups. Since at least 10 studies are thought to be necessary for meaningful subgroup analysis (Deeks 2011), the subgroup analyses presented are meant to be exploratory.
Oral supplementation
Participants in all studies received standard pancreatic enzymes and vitamin supplements in addition to the study interventions.
One study evaluated a combination of 200 mg vitamin E, 300 mg vitamin C, 25 mg β‐carotene, 90 μg selenium and 500 μg vitamin A compared to routine vitamin treatment (10 mg vitamin E and 500 μg of vitamin A) (Wood 2003). The vitamin E supplement was administered as RRR‐α‐tocopherol.
Three studies evaluated supplementation with vitamin E (α‐tocopherol) (Harries 1971; Keljo 2000; Levin 1961). Harries compared supplementation with vitamin E (10 mg/kg/day D,L‐α‐tocopheryl acetate) in a single daily dose to control group without vitamin E supplement; both a fat‐soluble and a water‐miscible preparation were assessed (Harries 1971). The second study compared supplementation with vitamin E (naturally occurring RRR‐α‐tocopherol) in tablet form to placebo; doses of the supplement were determined according to weight ‐ 600 UI/day for patients under 20 kg and 1200 UI/day for patients who weighed over 20 kg (1 IU is the biological equivalent of 0.45 mg of D,L‐α‐tocopheryl acetate) (Keljo 2000). Levin compared supplementation with vitamin E in a dose of 10 mg/kg/day of D,L‐α‐tocopheryl acetate in a water‐miscible dispersion divided in two or three doses to placebo (Levin 1961).
Two studies examined β‐carotene supplementation (Homnick 1995b; Renner 2001). Homnick reports on a comparison of patients in the β‐carotene group who received 30 mg β‐carotene twice a day (60 mg/day) to a control group (Homnick 1995b). The β‐carotene dose was increased individually and periodically during the study in an attempt to obtain plasma concentrations of 0.37 to 0.74 μM/L believed to be consistent with baseline concentrations in normal people. Eight participants in the control and five in the β‐carotene group finished the study, but there are no data reported from the control group (Homnick 1995b). In the Renner study, investigators compared a weight‐dependent dose of β‐carotene (1 mg/kg of body weight/day up to a maximum of 50 mg/day) to placebo for three months, after which point the β‐carotene was supplemented in a standard, non‐weight‐dependent dose (10 mg/day) for all participants for another three months (Renner 2001). Since the average weight‐dependent dose during the first part of the study was not reported, measurements at this time point were not meaningful and only endpoint data (i.e. change from baseline to six months) were included for meta‐analysis.
A further study examined selenium supplementation in a cross‐over study (Portal 1995a). The investigators examined a 2.8 mg/kg of body weight/day dose of selenium compared to placebo (Portal 1995a).
One study evaluated oral supplementation with L‐glutathione (65 mg/kg) divided into three doses per day compared to placebo (Visca 2013).
Inhaled supplementation
Two studies reported on supplementation with nebulised glutathione (GSH) compared to placebo (Bishop 2005; Griese 2013). In the earlier of these studies, patients inhaled buffered GSH 66 mg/kg distributed across four inhalation sessions per day, spaced three to four hours apart (Bishop 2005). The later study treated patients with GSH inhalations of 646 mg every 12 hours via an eFlow nebulizer (Griese 2013).
Outcomes
Outcomes for the different interventions are reported and analyzed separately.
Oral supplementation
Three studies reported the primary outcomes of this review (Renner 2001; Visca 2013 ; Wood 2003). All three studies reported FEV1%; and two studies additionally reported FVC (Visca 2013; Wood 2003).
Only one study reported QoL using a validated measure ‐ quality of well‐being (QOWB) (Wood 2003); no other trials reported any measure of QoL, validated or not.
For markers of oxidative stress, three studies reported lipid peroxidation measures: one reported F2‐isoprostanes (Wood 2003); one reported both H2O2 and thiobarbituric acid reactive substances (TBARS) (Portal 1995a); and one reported malondialdehyde (MDA) levels (Renner 2001). Two studies reported glutathione peroxidase (GPX) function (Portal 1995a; Wood 2003) and one reported superoxide dismutase (SOD) activity (Wood 2003). One study reported oxidative stress potential by trolox‐equivalent antioxidant capacity (TEAC) (Renner 2001).
All studies measured the plasma status of at least the antioxidant being supplemented. Four studies reported changes in the plasma levels of α‐tocopherol (vitamin E) (Harries 1971; Keljo 2000; Levin 1961; Visca 2013). As stated above, the concentrations of vitamin E in the older studies were expressed as mg/100ml and in the newer studies as μmol/L (Harries 1971; Levin 1961; Visca 2013). One study measured the plasma fatty acid status of 17 plasma fatty acids; since we did not pre‐specify which to analyze, only data for total plasma fatty acid status were included in the analysis (Wood 2003).
Three studies measured β‐carotene antioxidant status ( Homnick 1995b; Portal 1995a; Renner 2001). However, one of these did not completely report endpoints for the control group; as such, we did not have complete data to enter into a meta‐analysis (Homnick 1995b).
One study measured BMI % (Visca 2013) and one trial measured weight (Levin 1961). One study reported assessing BMI but did not provide complete outcome data; no additional data was provided by the study authors (Renner 2001).
Two studies reported the number of days of antibiotic therapy (Renner 2001; Wood 2003).
Data on death during the studies or adverse events were reported in five studies (Keljo 2000; Levin 1961; Portal 1995a; Renner 2001; Visca 2013).
Inhaled supplementation
Both studies of inhalation with glutathione reported the primary outcome of this review, lung function (Bishop 2005; Griese 2013). Each study used a range of measures for this. Bishop reported mean differences (post‐baseline) between GSH and placebo groups for FVC, FEV1, FEF25‐75 and peak flow measures (Bishop 2005). The Griese paper presents data on changes in the absolute FEV1; after contacting the author, data for FEV1 in % predicted have been also obtained (Griese 2013).
Both studies reported on QOL, but used different methods; Griese used a validated measurement tool (Wenninger 2003), and Bishop used self‐reported parameters (Bishop 2005; Griese 2013).
Only one study reported markers of oxidative stress, measurements of glutathione and its metabolites in both sputum and blood (Griese 2013). Both reduced glutathione and reduced forms of its metabolites (named free glutathione or free forms) and the sum of reduced and oxidized glutathione and that of its metabolites (named total glutathione or total forms) were measured. The intracellular levels of glutathione in neutrophils from sputum and blood were also measured in a small subgroup of patients. Lipid mediators (isoprostanes) and protein carbonyls were reported in sputum.
Griese also reported cytokines (IL‐10, IL‐8) in the sputum in a subgroup of patients from the intervention group compared to placebo (Griese 2013).
Both studies reported changes in nutritional status (Bishop 2005; Griese 2013). Griese measured weight (Griese 2013); and Bishop reported the average difference in BMI between baseline and after two months (Bishop 2005).
Only one study reported the number of exacerbations per patient (Griese 2013).
Both studies reported adverse events (Bishop 2005; Griese 2013).
Excluded studies
Upon title and abstract screening 259 trials were excluded and a further 51 were excluded after full‐text screening and one is awaiting classification (seeCharacteristics of excluded studies). Twelve studies described as controlled trials were excluded from this review (Cobanoglu 2002; Congden 1981; Farrell 1977; Knopfle 1975; Lancellotti 1996; Lepage 1996; Madarasi 2000; Portal 1995b; Underwood 1972b; Winklhofer‐Roob 1995; Winklhofer‐Roob 1996c; Winklhofer‐Roob 1997a). In four studies, the antioxidant intervention was compared to an active control arm, therefore not meeting the pre‐specified selection criteria for the review (Nasr 1993; Papas 2007; Peters 1996; Winklhofer‐Roob 1996b); in one, a micronutrient mix was compared to placebo; however, the intervention contained a mixture of micronutrients in addition to those being studied and the sole effects of those of interest could not be obtained (Oudshoorn 2007); five trials did not include any of the interventions under study (Abdulhamid 2008; Best 2004; Mischler 1991; Powell 2010; Sokol 1989). Twelve were prospective cohort studies (Bines 2005; Ekvall 1978; Kauf 1995; Kawchak 1999; Kelleher 1987; Munck 2010; Rawal 1974; Rettammel 1995; Richard 1990; Sokol 1989; Sung 1980; Wood 2002), seven were review articles (Anonymous 1975; Beddoes 1981; Goodchild 1986; Oermann 2001; van der Vliet 1997; Winklhofer‐Roob 2003; Zoirova 1983), three concerned letters (Winklhofer‐Roob 1996a; Winklhofer‐Roob 1997b; Winklhofer‐Roob 1997c), three reported on single dose administration for tolerance investigations (Casale 2012; Homnick 1995a; Jacquemin 2009), two were case‐reports (Hoogenraad 1989; Hubbard 1980), one was a retrospective cohort study (Underwood 1972a) and one included patients with chronic pancreatitis (Uden 1990).
Out of the excluded studies, one was represented by three separate reports (Winklhofer‐Roob 1996c) and two were represented by a report and an abstract (Abdulhamid 2008; Winklhofer‐Roob 1996b). The remaining studies were each represented by a single report.
Risk of bias in included studies
As can be seen from the risk of bias summaries none of the 10 included studies was free of bias and when each of the domains are considered across studies, none of these were apparently free of bias (Figure 3; Figure 4). Of those studies that had assessable (i.e. not unclear) domains (green and red dots), there were 12 instances of studies being judged to have a high risk of bias and 20 instances of a low risk of bias assessment. Most studies failed to adequately describe allocation concealment and blinding, resulting in an unclear risk of bias with respect to these domains (yellow dots). Each domain is individually described below.
Risk of bias graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each methodological domain for each included study.
Allocation
Sequence generation
Four studies adequately described sequence generation and were judged to have a low risk of bias (Griese 2013; Levin 1961; Visca 2013; Wood 2003). Griese randomized participants by central telephone block randomization at 1:1 ratio within each age group to receive study medication or placebo (Griese 2013). In the Levin study, cards labelled 1 or 2 were individually placed in sealed envelopes in groups of four, two for each mixture number. Envelopes were divided into three groups, according to the age of the patients: less than 5, between 5 and 10 and 10 years or older (Levin 1961). In a third study, once patients were enrolled, they were randomly assigned to treatment or placebo by use of a random number generator (Visca 2013). Finally, Wood states that the sequence was derived using a random‐numbers computer program (Wood 2003). We judge there to be an unclear risk of bias for the remaining six studies (Bishop 2005; Harries 1971; Homnick 1995b; Keljo 2000; Portal 1995a; Renner 2001). In the Bishop study, it is stated that the patients were first paired by age and sex, and then each member of the pair was randomly assigned to the treatment or placebo group, but the actual method of randomisation is not described (Bishop 2005).
Allocation concealment
We judged two studies to have a low risk of selection bias (Bishop 2005; Griese 2013;). In the Bishop study, no member of the clinical team was involved in the coding or assignment to treatment or placebo; non‐clinical researchers involved in the study were only provided with patient identification numbers, not patients names (Bishop 2005). Griese used a central telephone system to inform clinicians of participant allocation (Griese 2013). The risk of bias with respect to allocation concealment is unclear for the remaining eight studies (Harries 1971; Homnick 1995b; Keljo 2000; Levin 1961; Portal 1995a; Renner 2001; Visca 2013; Wood 2003). Levin does state that they concealed the allocation schedules in sealed envelopes, but does not state if these envelopes were opaque, thus the risk of bias is unclear.
Blinding
We judged there to be a low risk of bias with respect to blinding for four studies (Bishop 2005; Keljo 2000; Levin 1961; Renner 2001). The Bishop study matched the taste and smell of treatment and placebo medications to ensure that both patients and the clinical team were blinded (Bishop 2005). In the study by Keljo, treatment (naturally occurring RRR‐α‐tocopherol) and placebo were both provided in vegetable oil (Keljo 2000). Levin also matched the taste of treatment and placebo and the preparations were labelled "1" and "2" so neither investigator nor patients knew which was being taken (Levin 1961). In the Renner study, all patients received capsules of identical appearance (Renner 2001).
Although the study by Griese was a double‐blind study with respect to the packaging of the vials and visual appearance of the medication, those participants treated with verum could recognize glutathione by its smell, which could not be masked; therefore, the blinding bias was judged as high (Griese 2013). The authors provide this bias as explanation for the significantly higher dropout rate due to early termination by patient request in the placebo group compared to the treatment group.
One study was judged as having a high risk of bias as the control group did not receive a placebo but rather no treatment. Moreover, the two interventions used were physically different (tablet versus liquid preparations) (Harries 1971).
The remainder of the studies did not describe the blinding process in enough detail in order to allow a proper assessment of this domain; therefore, the risk of bias with respect to blinding is unclear for these studies (Homnick 1995b; Portal 1995a; Visca 2013; Wood 2003).
Incomplete outcome data
Three out of 10 studies are judged to have a low risk of attrition bias as the number of withdrawals from the studies as well as the reasons for these are described in detail for each group (Bishop 2005; Griese 2013; Visca 2013). A further three studies did not provide a description of withdrawals or dropouts and are judged to have an unclear risk of bias (Harries 1971; Renner 2001; Wood 2003). The remaining four studies reported incomplete data for the outcomes of interest and have a high risk of bias (Homnick 1995b; Keljo 2000; Levin 1961; Portal 1995a). One study did not described which study arm participants withdrew from; furthermore, the authors of this study did not provide control group data, thereby preventing a comparison between groups in a meta‐analysis (Homnick 1995b). Authors were contacted but unable to provide further information because the original data were on a computer they no longer had access to (Homnick 2008). In the paper by Keljo, there are inconsistencies in the number of patients included in each group between the tables of data reporting and the table describing patient inclusion criteria (Keljo 2000). Furthermore, data from one subgroup of patients are not reported at all due to the very limited number of patients (Keljo 2000). The third study, reports that there were 45 patients in the final analysis who had been followed for at least two months and 37 patients who completed the six‐month study period (18 in the tocopherol group and 19 in the placebo group); serum tocopherol was reported at two and six months in 18 and 15, respectively out of 20 patients initially included in the study (Levin 1961). While the remaining study did not explicitly state the number of participants originally randomized to each group, it is the only study which states the reasons for participant withdrawal (Portal 1995a).
Selective reporting
Six studies reported data for all outcomes measured and are judged to have a low risk of bias (Bishop 2005; Harries 1971; Levin 1961; Portal 1995a; Visca 2013; Wood 2003). One study is judged to have an unclear risk of bias as the full paper does not state in the 'Methods' section what the authors planned to report on and the protocol is not available (Keljo 2000). Three studies appeared to have a high risk of bias in this domain (Griese 2013; Homnick 1995b; Renner 2001). Griese reported data on cellular and biochemical markers such as glutathione and its metabolites in an exploratory manner in a very limited number of patients (Griese 2013). In one study, authors claimed to take measurements at least monthly for 56 weeks, but only report data for baseline and week 50 (Homnick 1995b). In the remaining study, actual data for BMI were not reported and the difference between groups only described as non‐significant; when contacted, the author was unable to provide further data due to relocation of the involved statistician (Renner 2001).
Other potential sources of bias
Two studies included in this review appear to be subject to duplicate publication (Portal 1995a; Renner 2001). In the case of Portal, authors describe the same study in full‐length manuscripts, published two years apart (Portal 1995a). The journals in which they are published appear related, but are independent – Clinical Chemistry andClinica Chimica Acta (International Journal of Clinical Chemistry). Although the two reports appear to describe different outcomes of the same study based on their titles (the 1993 paper reports on biological indices of selenium status and the 1995 paper reports on lipid peroxidation markers), the later paper does not reference the methods already reported in the earlier report. Although the earlier report assesses two outcomes not later described and the latter report describes two not previously described, there is an overlap of two outcomes; neither of which is referred to as having already been reported. As such, the two studies were taken as one here since the outcomes of interest were contained in both studies and the authors of this review did not want to ‘double count’ participants (Portal 1995a). Another study appeared in the literature in seven different instances ‐ three full‐text reports and four abstracts (Renner 2001). At the screening stage of this review, one full‐text report was included and the other two were excluded on the basis of unstated diagnostic criteria. Eventually, data from all reports were included for meta‐analysis according to Cochrane policy. None of the full‐text reports referenced the others and all are reported as 'original' publications.
There is an unclear risk of bias for the Keljo study which has not been published in a peer‐reviewed journal; an abstract was presented at the North American Cystic Fibrosis Conference and additional data presented in this review were supplied directly by the authors (Keljo 2000).
With the exception of one study which includes 153 participants (Griese 2013), most studies in this review suffer from relatively small sample sizes, ranging from 20 to 49 participants.
Another source of potential bias is the one cross‐over study included in this review (Portal 1995a). While the authors describe a proper cross‐over regimen, they failed to measure and report baseline measurements for all outcomes after the washout period and before the start of the second period. This prevented the authors of this review from assessing whether a ‘carry‐over’ effect occurred; data from the second period were incomplete and hence could not be included for analysis in this review.
In the Griese study which compares the effect of inhaled glutathione to placebo, a number of patients were allowed to continue the oral administration of N‐acetylcysteine, which is a precursor of glutathione and these patients could not be identified from the report. We are therefore unable to assess the possible influence of this treatment on the results (Griese 2013).
In the study by Visca, there were more patients homozygous for delta F508 (known to have a more severe disease manifestation than heterozygotes) in the placebo group (27.7%) compared to the GSH group (13.6%) (Visca 2013).
We judge the Homnick study Homnick 1995b to be at a high risk of bias since the authors do not describe baseline demographics and do not state a sample size calculation. Furthermore, investigators did not systematically control dose levels throughout the study (Homnick 1995b).
Effects of interventions
Only significant summary statistics are described in the text below; statistics showing non‐significant effects are available in the meta‐graphs (Data and analyses).
Oral antioxidant supplementation versus control
Primary outcomes
1. Lung function tests
a. FEV1
Three studies reported FEV1 (% predicted) (Renner 2001; Visca 2013; Wood 2003). After two months of a combined supplement, Wood reported a significant difference in favour of control, MD ‐4.30% (95% CI ‐5.64 to ‐2.96). At six months the meta‐analysis showed a significant overall effect of oral GSH (Visca 2013) compared to the control group in favour of the antioxidants, MD 16.90 (95%CI 13.24 to 20.55) (Analysis 1.1), but a non‐significant difference for the and β‐carotene supplement (Renner 2001).
b. FVC
Two studies reported on FVC (% predicted), one of combined supplement after two months (Wood 2003) and one of oral GSH after six months (Visca 2013). The difference between treatment and control groups at two months was not significant, but a difference was found after six months of oral supplementation with GSH in favour of the antioxidant group, MD 14.80 (95% CI 9.66 to 19.94) (Analysis 1.2).
2. QoL
This outcome was assessed using the Quality of Wellbeing scale (QOWB) in one study (Wood 2003) and was found to significantly favour control over antioxidant supplementation, MD ‐0.06 points (95% CI ‐0.12 to ‐0.01) (Analysis 1.3).
Secondary outcomes
1. Oxidative stress
a. Lipid peroxidation
Three measures of lipid peroxidation were reported by three studies: H2O2 (Portal 1995a); TBARS (Portal 1995a; Renner 2001); and 8‐iso‐prostoglandin F2α (Wood 2003). There was no significant difference between groups in the meta‐analysis containing H2O2 (Analysis 1.4), TBARS (Analysis 1.5) or F2‐isoprostanes (Analysis 1.6).
b. Antioxidant enzyme function
Two studies contributed data for this outcome; one of combined supplementation with data reported at two months (Wood 2003) and one of selenium supplementation reported at five months (Portal 1995a). There was a significant improvement in GPX for both combined supplementation, MD 1.60 units per gram of haemoglobin (U/g Hb) (95% CI 0.30 to 2.90) and for selenium supplementation 10.20 U/g Hb (95% CI 2.22 to 18.18) (Analysis 1.7).
Only the study of combined supplements reported on SOD (at two months); there was no significant difference between groups (Analysis 1.8).
c. Potency
One study of β‐carotene supplementation reported on antioxidant potency using TEAC as an outcome measure (Renner 2001). At six months, there was no significant different found between supplement and placebo groups (Analysis 1.9).
d. Plasma antioxidant status
i. Vitamin E
Four studies provided data for this outcome (Harries 1971; Levin 1961; Visca 2013; Wood 2003). Two studies supplemented antioxidants in form of vitamin E as D,L‐α‐tocopheryl acetate (Harries 1971; Levin 1961), Harries supplemented with both fat‐soluble and water‐miscible forms of vitamin E; and one study supplemented vitamin E as RRR‐α‐tocopherol as part of a combined antioxidant supplement (Wood 2003). A further study supplemented with oral GSH (Visca 2013). The supplementation led to significantly increased plasma levels of vitamin E in favour of the supplements as follows: at one month fat‐soluble vitamin E supplements, MD 11.37 (95% CI 9.69 to 13.05) and water‐miscible vitamin E supplements, MD 27.86 (95% CI 23.95 to 31.77); at two months water‐miscible vitamin E, MD 11.60 (95% CI 9.73 to 13.47) and combined supplement, MD 12.40 (95% CI 8.99 to 15.81); and at six months water‐miscible vitamin E, MD 19.73 (95% CI 17.73 to 21.73) and oral GSH, MD 1.70 (95% CI 1.39 to 2.01) (Analysis 1.10). The difference between oral GSH and control groups was less than the supplements containing vitamin E; however, the effect of oral GSH supplementation on the serum levels of vitamin E is indirect (increase regeneration of the oxidized vitamin E) compared to the direct supplementation with vitamin E.
One study included in the review reported changes in the serum vitamin E levels without SDs which make them unsuitable for the meta‐analysis (Keljo 2000). The study reported that levels increased from 28.2 to 35 μM/L with vitamin E treatment and from 25.4 to 28.6 μM/L in the placebo group. Baseline serum vitamin E levels were not reported and the paper states "the baseline serum α‐tocopherol level did not differ between the placebo and α‐tocopherol groups, and there was no difference in the baseline α‐tocopherol levels between subgroups (data not shown)".
ii. β‐carotene
One study included β‐carotene as part of a combined antioxidant supplement (Wood 2003) and two studies included it as a single supplement (Homnick 1995b; Renner 2001); however only two studies presented data suitable for analysis (Renner 2001; Wood 2003). There was a significant improvement in β‐carotene levels in favour of both combined supplementation at two months, MD 0.10 μmol/L (95% CI 0.02 to 0.18) and single β‐carotene supplementation at six months, MD 0.24 μmol/L (95% CI 0.02 to 0.46) (Analysis 1.11).
Homnick reported that the mean (SD) serum levels of β‐carotene increased significantly from baseline 0.09 (0.02) μmol/L to 0.62 (0.19) μmol/L during supplementation with β‐carotene; however, no data are available on the β‐carotene serum levels in the control group and therefore these results could not be included in the analysis (Homnick 1995b). It is stated in the paper that "no control patient had a significant increase in β‐carotene levels throughout the duration of the study" and a mean (SD) baseline of 0.12 (0.05) μmol/L is given, but it is not clear which patients are included.
iii. Selenium
Two studies supplemented selenium (Portal 1995a; Wood 2003). Both combined supplementation (Wood 2003) and single supplementation (Portal 1995a) showed a significant improvement in plasma selenium status in favour of antioxidant supplementation at two months, MD 0.60 μmol/L (95% CI 0.39 to 0.81) and at six months, MD 0.39 μmol/L (95% CI 0.27 to 0.51) respectively (Analysis 1.12).
iv. Vitamin C
One study supplemented vitamin C as part of a combined antioxidant supplementation in 46 participants (Wood 2003); there was no significant difference in improvement between antioxidant and control (Analysis 1.13).
e. Plasma fatty‐acid status
One study of a combined antioxidant supplementation examined this outcome (Wood 2003). Data showed no significant difference between groups (Analysis 1.14).
2. Inflammation
a. Inflammatory markers (i.e. IL‐6, IL‐8, TNF‐α, IL‐1β)
Keljo measured IL‐6 and TNF‐α in three subgroups of patients defined according to lung function and treatment with DNase (Keljo 2000). After three months supplementation with RRR‐α‐tocopherol, a non‐significant difference was found in the levels of these cytokines in all but one of the groups of patients treated with antioxidants compared to placebo, with a slight decrease in the levels of these cytokines in the treated group (Analysis 1.15; Analysis 1.16). In those patients whose FEV1 measurements lay between 70% and 85% and who were taking DNase (n = 11), there was a significant difference in favor of RRR‐α‐tocopherol, MD ‐0.94 (95% CI ‐1.61 to ‐0.26) (Analysis 1.16).
b. Hyperinflation of chest
No studies examined this outcome.
3. Nutritional status
One study measured the effects of supplementation on BMI, but only reported baseline values and stated that there was a non‐significant effect of supplementation on this outcome (Renner 2001). We were unable to obtain full data for this outcome from the study investigators.
One study reported on the % change in BMI after six months of oral supplementation with GSH (Visca 2013). Data showed a significant difference in favour of the supplementation with antioxidants, MD 17.20 (95% CI 12.17 to 22.23) (Analysis 1.17).
One study reported on the actual weight measured in kg (Levin 1961). Analysis of the data found no significant difference between groups (Analysis 1.18).
4. Antibiotic days
The number of antibiotic days per patient in both treatment groups was reported in two studies (Renner 2001; Wood 2003). No significant difference between groups was found (Analysis 1.19).
5. Adverse events
While it was possible to identify specific adverse events, the rates of specific events were not calculable due to inadequate reporting. Five studies reported adverse events or deaths during the study (Keljo 2000; Levin 1961; Portal 1995a; Renner 2001; Visca 2013). Keljo reported a number of mild adverse events with no deaths (Keljo 2000); none of these were significant (Analysis 1.20). The cross‐over study stated that one death occurred in the group which received selenium first followed by placebo; however, investigators did not state a time point or period during which the death occurred, other than to say that only baseline data were used in the analysis (Portal 1995a). Another study reported three deaths, all of which were in the control group (Levin 1961). The remaining trials reported that no adverse events occurred (Renner 2001; Visca 2013).
Inhaled antioxidant supplementation versus control
Primary outcomes
1. Lung function tests
a. FEV1
Both studies of inhaled GSH reported FEV1 (Bishop 2005; Griese 2013). One of the studies (n = 153) reported absolute values at one, three and six months; values for % predicted were obtained from the first author after contact (Griese 2013). All of these results from this study were statistically significant in the favour of the intervention group: at one month, MD 0.93 (95% CI 0.64 to 1.22); at three months, MD 2.57 (95% CI 2.24 to 2.90); and at six months, MD 0.97 (95% CI 0.65 to 1.29) (Analysis 2.1). The second study (n = 16) reported % predicted at a single time point after two months (eight weeks) (Bishop 2005). Data analysis showed a non‐significant difference between the group receiving GSH as an inhalation and placebo (Analysis 2.1).
b. FVC
Both studies reported FVC (Bishop 2005; Griese 2013). As for FEV1, Griese reported values at one, three and six months (in a graphical form), but values for FVC % predicted were obtained from the first author after contact (Griese 2013). Again, all of these results were statistically significant: at one month, MD 1.31 (95% CI 1.07 to 1.55); at three months, MD 3.10 (95% CI 2.80 to 3.40); and at six months, MD 0.65 (95% CI 0.35 to 0.95) in favour of the intervention group (Analysis 2.2). Bishop reported FVC % predicted after two months (eight weeks) (Bishop 2005). Data analysis showed a non‐significant difference between the group receiving GSH as inhalation and placebo (Analysis 2.2).
2. QoL
Both studies reported on QoL (Bishop 2005; Griese 2013). However, only one study used a validated measure (CF Questionnaire for QOL) (Griese 2013); Bishop used a self‐reported scale which we do not present in the analysis (Bishop 2005). Data analysis showed a non‐significant difference between antioxidant group and placebo at all three time points (one, three and six months) (Analysis 2.3).
Secondary outcomes
1. Oxidative stress in sputum and blood
a. hydrogen peroxide (H2O2) exhalation
Neither of the studies reported this parameter (Bishop 2005; Griese 2013).
b. lipid peroxidation (8‐isoprostanes) in sputum
Griese measured 8‐isoprostane in the sputum of a small number of patients and reported data at three and six months (Griese 2013). Neither result was significant, but the level of lipid peroxidation was lower at the three‐month and six‐month time points in the group treated with antioxidants (Analysis 2.4).
c. antioxidant enzyme function (post hoc change)
Neither of the studies reported this parameter (Bishop 2005; Griese 2013).
d. potency (post hoc change)
Neither of the studies reported this parameter (Bishop 2005; Griese 2013).
e. plasma and sputum antioxidant status
Griese measured levels of free and total GSH and its metabolites in plasma at six months (Griese 2013). No statistically significant differences between the two groups were found for any of these levels (Analysis 2.5; Analysis 2.6).
Griese also measured levels of free and total GSH and its metabolites in sputum. For free GSH, at one month and three months there was no statistically significant difference between groups; however, data did show a significant difference in free GSH at six months in favour of the GSH‐treated group, MD 59.10 (95% CI 3.68 to 114.52) (Analysis 2.7). For total GSH results at one and three months were significant in favour of the GSH‐treated group, MD 405.30 (95% CI 105.27 to 705.33) and MD 329.20 (95% CI 167.04 to 491.36) respectively (Analysis 2.8). The results at six months were not statistically significant.
The intracellular levels of GSH in neutrophils in the blood were reported at the six‐month time point for a subset of patients (four patients in the GSH‐treated group and nine in the placebo group) (Griese 2013). Results were not statistically significant (Analysis 2.9).
Griese also measured intracellular levels of GSH in neutrophils in the sputum in a small subgroup of patients (eight in the GSH‐treated group and eight in the placebo group) at one, three and six months (Griese 2013). Statistically significant differences between the two groups were found for GSH levels at three and six months, MD 3.70 (95% CI 0.27 to 7.13) and MD 4.40 (95% CI 1.52 to 7.28) respectively, but not at one month (Analysis 2.10).
f. plasma fatty acid status
Neither of the studies reported this outcome (Bishop 2005; Griese 2013).
g. Carbonylated proteins
One study measured carbonylated proteins in the sputum of a subgroup of patients as a marker of oxidative stress (Griese 2013). No statistical significant difference was found for the change from baseline in levels of protein carbonyls in the sputum at one, three or six months (Analysis 2.11).
2. Inflammation
a. inflammatory markers (i.e. IL‐6, IL‐8, IL‐10,TNF‐α, IL‐1β)
One study analysed levels of these cytokines and chemokines in the sputum of a subgroup of patients (24 in GSH‐treated group and 29 in the placebo group) at six months (Griese 2013). No statistically significant differences between the two groups in change in levels from baseline were found (Analysis 2.12; Analysis 2.13; Analysis 2.14).
b. hyperinflation of chest
Neither of the studies reported this outcome (Bishop 2005; Griese 2013).
3. Nutritional status (e.g. BMI or BMI percentile for children)
One study measured BMI at baseline and at the end of the study (two months) and presented the differences between the two groups (Bishop 2005). No statistically significant difference was found (Analysis 2.15).
The second study measured the change in weight from baseline to one, three and six months (Griese 2013). There was a statistically significant gain in weight in the GSH‐treated group after three months, MD 1.00 kg (95% CI 0.39 to 1.61), but results at one and six months were not significant (Analysis 2.16).
4. Pulmonary exacerbations requiring intravenous antibiotic therapy or hospitalization
Only one study reported the time to pulmonary exacerbations as the mean (SD) number of days and the rate of exacerbations (Griese 2013). The rate of exacerbation was reduced by 18% in the treated group compared to placebo, but this was not statistically significant. The time to first pulmonary exacerbation (days) was significantly longer in the GSH treated group, MD 22.00 (95% CI 20.05 to 23.95) (Analysis 2.17).
5. Adverse events
One study reported the number of patients with specific adverse events: hospitalization for non‐acute pulmonary exacerbations; rhinitis or sinusitis; cough; pharyngitis; stomach pain or cramps; headache; chest tightness or bronchospasm; nose bleed; and shortness of breath (Bishop 2005). The second study also reported some specific adverse events after inhalation with antioxidants which are presented the graphs (Analysis 2.18). There were no statistical significant differences between the placebo and intervention groups in the number of cases of rhinitis or sinusitis, or upper respiratory tract infection, cough, pharyngitis and headache, which were reported by both studies (Analysis 2.18).
Bishop reported no serious adverse events (Bishop 2005). Griese reported that the number of serious adverse events were similar between the group treated with GSH inhalations and the placebo group (11% and 10%, respectively) (Griese 2013).
Sensitivity Analysis
Since there were so few studies contributing data to the primary outcomes, a sensitivity analysis with regards to risk of bias was not conducted. However, this may be a useful analysis in the future, especially with respect to the high risk of incomplete data and selective reporting which plagued the current review.
Due to inadequacies of reporting numbers of enrolled participants, completed participants and analysed participants in most studies, an intention‐to‐treat analysis was not possible.
Sensitivity analyses excluding studies with industry funding was planned but not conducted (no studies were funded by industry).
Publication bias
A funnel plot was not generated, since we were not able to include and combine a sufficient number of studies in this review (Light 1994). Also, only limited data were available for analyses from those included studies.
Discussion
Summary of main results
In the revised version of this review, we chose to extend the list of antioxidant micronutrients (vitamin E, vitamin C, ß‐carotene and selenium) with GSH administered orally or by inhalation in CF patients and the effects of the two routes of administration are presented separately.
There appears to be conflicting evidence regarding the clinical effectiveness of oral supplementation with antioxidants in CF; however only a small number of studies contributed data towards analysis in this systematic review. Three studies describing 114 participants reported lung function measured by FEV1 (Renner 2001; Wood 2003; Visca 2013). Data from two studies suggest that oral antioxidant micronutrient supplementation does not improve lung function (Renner 2001; Wood 2003). However, addition of data to the meta‐analysis from the Visca trial which compared oral GSH supplementation of 65 mg/kg three times per day in a pediatric population seem to be beneficial for the patients (Visca 2013). Due to the short half‐life of GSH (Reed 2008), repeated daily dosing for relatively extended time periods (six months) is essential for the effect. Only one study with 46 participants assessed QOL, and showed that QOL improvement actually favoured the control group (Wood 2003). There was a significant difference between antioxidants and control in both improvement of GPX and plasma antioxidant status for all antioxidants except vitamin C. There was an improvement in the blood levels of vitamin E in all trials that used supplementation with this vitamin, although vitamin E was administered in different forms and for different periods of time (Harries 1971; Levin 1961; Wood 2003). Adverse events were not adequately reported; there was only one death in a study of 27 participants reported, but this was not clearly attributable to the supplement (selenium) or placebo (Portal 1995a).
Only two studies contributed to the analysis of supplementation with inhaled GSH (Bishop 2005; Griese 2013); and the size of Griese study with 153 patients is much larger than Bishop study with only 19 patients. Both studies reported the primary outcome of this review (lung function). An effect on lung function in favour of the antioxidant supplementation was observed at one, three months and six months only in the study by Griese (Griese 2013). Both studies reported on QOL (Bishop 2005; Griese 2013), but only Griese reported validated parameters and showed a difference in favour of the antioxidant supplementation (Griese 2013). The levels of GSH, both the oxidized and reduced form, in the sputum, as well as the intracellular levels in sputum neutrophils were higher in the group of patients receiving inhaled GSH compared to controls. No differences between the two groups were observed in levels of GSH in the blood (Griese 2013).
Overall completeness and applicability of evidence
For oral supplements, the primary outcomes had very few data to contribute to the meta‐analysis ‐ only three out of nine studies assessed lung function and one included study assessed QOL. Given the limited number of included studies with data for the meta‐analysis, stating that antioxidant supplementation has either an effect or no effect based on these outcomes appears to be premature. The analysis performed was exploratory in nature and the results should be interpreted with caution. Furthermore, small sample sizes of included studies, incomplete reporting and per protocol analyses demonstrate that there is not yet a single, well‐designed and reported RCT in this area. Small and unachieved sample sizes reduce the power of a study, thereby increasing the chance of a type II error ‐ wrongly accepting the null hypothesis when it is false. The absence of reporting of methods used to determine sample size in all of the included studies yields questions regarding minimum important difference of outcomes, possibly because these data do not exist for many of the biological markers used as primary outcomes.
There was one cross‐over RCT, from which complete data were only reported from the first period, thereby halving the intended sample size and yielding an underpowered study, which makes a significant difference undetectable (Portal 1995a). A completely reported sufficiently‐powered study is necessary before concluding that antioxidant supplementation had no effect on lung function. Specifically, investigators did not present baseline measurements for the second treatment period following the wash‐out period making assessment of carryover effect unfeasible. The authors acknowledge that since only half of the intended population was included in meta‐analysis, issues of reduced power may prevent the study results from revealing true differences between intervention and control. This also contributed to the decision not to pool the treatment effect.
Plasma antioxidant status was the most completely reported outcome in studies included for review. As one might expect, since they are the most direct measure of plasma levels, there was evidence that antioxidant supplementation improved plasma status for the respective micronutrient being supplemented. However, the correlation of plasma antioxidant status to clinically important outcome measures in CF has not been adequately explored. Only two out of five studies examined clinically important outcomes: lung function, in which there was no significant difference in improvement between groups; and QoL, in which there was a significant improvement in favour of the control group. These studies were the two most recent ones. It is possible that investigators of studies more than 10 years old may not have perceived today's clinically important outcomes as relevant at the time. The study of antioxidants has increased in recent years and the mechanisms of action of many oxidative stress processes were largely unknown 10 years ago.
For inhaled supplements, only two studies contribute to the analysis and although small, an effect on lung function was observed after three months of supplementation from the larger of these studies.
A common draw‐back of these studies is that the patients are intensively treated with inhaled antibiotics and other treatments that lead to a significant improvement of their lung function making further improvements by addition of antioxidants difficult to assess without very large number of patients. Griese included 153 participants in his study and calculated that 276 participants would have been necessary to show a 2.2% predicted change in lung function (FEV1) to be statistically significant (Griese 2013).
Quality of the evidence
An overall risk of bias in this review was largely unclear, due to inadequate reporting of methods and results of included studies. There is an unclear and potentially large amount of bias in the results of this review and we judge that further studies are necessary before firm conclusions can be made. Three out of the ten included studies had a low risk of bias in all domains which had sufficient information to make a clear judgement (Bishop 2005; Visca 2013; Wood 2003), while none of the domains were free of bias. The risk of bias relative to sequence generation was unclear in all studies except four (Griese 2013; Levin 1961; Visca 2013; Wood 2003). With the exception of two studies, where allocation concealment was judged as low (Bishop 2005; Griese 2013), the remaining eight studies had an unclear risk of bias in this domain. The blinding procedure was described in four studies, which were judged as having a low risk of detection and performance bias (Bishop 2005; Keljo 2000; Levin 1961; Renner 2001). Two studies were judged as having a high risk of bias for this domain; one as it was not possible to blind the 'no treatment' arm (Harries 1971) and in the second, it was possible for the patients to find out if they were inhaling the verum due to the smell and taste of the active substance (Griese 2013). Low attrition bias was ascertained for three studies (Bishop 2005; Griese 2013; Visca 2013) and high attrition bias for four studies (Homnick 1995b; Keljo 2000; Levin 1961; Portal 1995a); the remaining studies had an unclear risk of bias. One out of 10 studies did not contribute data to any of the outcomes measured in this review (Homnick 1995b); this highlights the need for complete selection and reporting of outcomes for studies in this area in order to make treatment decisions. Authors of some studies were contacted for a more complete data set, but were unable to locate the appropriate data (likely due to length of time since study completion). Furthermore, important clinical outcomes listed in this review were omitted in many studies and may be evidence of selective reporting.
One study in which multiple publication was apparent was a single‐centre RCT examining the effects of β‐carotene supplementation on multiple biological markers of CF lung disease (Renner 2001). When redundancy is not made explicit and study reports fail to disclose association with other reports of the same population under study, this can be particularly challenging for systematic reviewers (Huston 1996). If systematic reviewers were unaware of redundant publications, especially when published under different first author names (as is true in this case where papers were published with lead authors Engl, Renner and Rust (Renner 2001)), data may be counted twice and further overestimate true treatment effect (Huston 1996).
Potential biases in the review process
No articles on the Cochrane Cystic Fibrosis and Genetic Disorders Group's CF Trials Register have been recorded as containing the terms 'vitamin C' or 'glutathione', hence these terms were not searchable keywords in that register. Previously, additional searches of other databases were conducted using these terms (seeAppendices).
Two studies reported data for the number of antibiotic days (Renner 2001; Wood 2003). Of those, one reported a range rather than an SD (Wood 2003). As such, the SD was imputed using the range yielding an inaccurate estimate, since ranges are distorted by outliers in the data. If one were to exclude data from this study, the MD between groups would be ‐23.00 days (95% CI ‐34.71 to ‐11.29) (or 23 less days) in favour of antioxidants based on the remaining study (Renner 2001) and may better represent antioxidant effect on this outcome.
Agreements and disagreements with other studies or reviews
The data presented within this systematic review have not been previously synthesized. During the screening phase of this review, numerous case‐control and cohort studies on this topic were identified (seeCharacteristics of excluded studies) and such studies have been the basis for clinical trials in this area. Previous studies suggest that antioxidant micronutrients are likely to play a role in the oxidative stress that occurs in CF lung disease and have shown beneficial results (Winklhofer‐Roob 1994; Winklhofer‐Roob 1997a; Winklhofer‐Roob 2003; Wood 2002). However, the aim of this review was to obtain the most rigorous studies on which to base conclusion that have been asserted by multiple cohort and case‐control studies to date.
Peroxide chain reaction characterized by initiation, propagation and termination. (RH: PUFA; R·: free radical; ROO·: peroxide; ROOH: hydroxyl peroxide; AH: vitamin E; A·: oxidized Vitamin E. Adapted from: Tappel AL. Vitamin E and free radical peroxidation of lipids. Annals of the New York Academy of Sciences. 1972; 203(1):12‐28.
Risk of bias graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each methodological domain for each included study.
![Comparison 1 Oral antioxidants versus control, Outcome 1 Lung function FEV1 [% pred].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-01.png)
Comparison 1 Oral antioxidants versus control, Outcome 1 Lung function FEV1 [% pred].
![Comparison 1 Oral antioxidants versus control, Outcome 2 Lung function FVC [% pred].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-02.png)
Comparison 1 Oral antioxidants versus control, Outcome 2 Lung function FVC [% pred].
Comparison 1 Oral antioxidants versus control, Outcome 3 QoL: Quality of Well Being Scale.
![Comparison 1 Oral antioxidants versus control, Outcome 4 Oxidative stress: lipid peroxidation (H2O2) [μmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-04.png)
Comparison 1 Oral antioxidants versus control, Outcome 4 Oxidative stress: lipid peroxidation (H2O2) [μmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 5 Oxidative stress: Lipid peroxidation (TBARS) [μmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-05.png)
Comparison 1 Oral antioxidants versus control, Outcome 5 Oxidative stress: Lipid peroxidation (TBARS) [μmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 6 Oxidative stress: Lipid peroxidation (F2‐isoprostanes) [ng/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-06.png)
Comparison 1 Oral antioxidants versus control, Outcome 6 Oxidative stress: Lipid peroxidation (F2‐isoprostanes) [ng/L].
![Comparison 1 Oral antioxidants versus control, Outcome 7 Oxidative stress: Enzyme function ‐ GPX [U/g Hb].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-07.png)
Comparison 1 Oral antioxidants versus control, Outcome 7 Oxidative stress: Enzyme function ‐ GPX [U/g Hb].
![Comparison 1 Oral antioxidants versus control, Outcome 8 Oxidative stress: Enzyme function ‐ SOD [U/mg Hb].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-08.png)
Comparison 1 Oral antioxidants versus control, Outcome 8 Oxidative stress: Enzyme function ‐ SOD [U/mg Hb].
![Comparison 1 Oral antioxidants versus control, Outcome 9 Oxidative stress: Potency (TEAC) [mmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-09.png)
Comparison 1 Oral antioxidants versus control, Outcome 9 Oxidative stress: Potency (TEAC) [mmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 10 Plasma antioxidant status ‐ vitamin E [μmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-10.png)
Comparison 1 Oral antioxidants versus control, Outcome 10 Plasma antioxidant status ‐ vitamin E [μmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 11 Plasma antioxidant status ‐ β‐carotene [μmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-11.png)
Comparison 1 Oral antioxidants versus control, Outcome 11 Plasma antioxidant status ‐ β‐carotene [μmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 12 Plasma antioxidant status ‐ selenium [μmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-12.png)
Comparison 1 Oral antioxidants versus control, Outcome 12 Plasma antioxidant status ‐ selenium [μmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 13 Plasma antioxidant status ‐ vitamin C [μmol/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-13.png)
Comparison 1 Oral antioxidants versus control, Outcome 13 Plasma antioxidant status ‐ vitamin C [μmol/L].
![Comparison 1 Oral antioxidants versus control, Outcome 14 Inflammation: plasma fatty acid status [mg/L].](/cdsr/doi/10.1002/14651858.CD007020.pub3/media/CDSR/CD007020/rel0003/CD007020/image_n/nCD007020-CMP-001-14.png)
Comparison 1 Oral antioxidants versus control, Outcome 14 Inflammation: plasma fatty acid status [mg/L].
Comparison 1 Oral antioxidants versus control, Outcome 15 Inflammation: IL‐6 (pg/ml) at 3 months (vitamin E).
Comparison 1 Oral antioxidants versus control, Outcome 16 Inflammation: TNF‐α (pg/ml) at 3 months (vitamin E).
Comparison 1 Oral antioxidants versus control, Outcome 17 Nutritional status (BMI %).
Comparison 1 Oral antioxidants versus control, Outcome 18 Nutritional status: weight (kg).
Comparison 1 Oral antioxidants versus control, Outcome 19 Antibiotic days per patient.
Comparison 1 Oral antioxidants versus control, Outcome 20 Adverse effects (vitamin E).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 1 Lung function FEV1 (% predicted).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 2 Lung function FVC (% predicted).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 3 QoL score (CF questionnaire for QoL).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 4 Sputum oxidative stress: lipid peroxidation (8‐isoprostan).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 5 Plasma antioxidant status: free glutathione (pM).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 6 Plasma antioxidant status: total glutathione (pM).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 7 Sputum antioxidant status: free glutathione in sputum (pM).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 8 Sputum antioxidant status: total glutathione in sputum (pM).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 9 Plasma antioxidant status: glutathione in blood neutrophils (MFI).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 10 Sputum antioxidant status: glutathione in sputum neutrophils (MFI).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 11 Sputum oxidative stress: protein carbonyls (U).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 12 Local inflammation: cytokines in sputum IL‐8 (pg/ml).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 13 Local inflammation: cytokines in sputum IL‐10 (pg/ml).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 14 Local inflammation: cytokines in sputum TNF‐α (pg/ml).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 15 Nutritional status: BMI.
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 16 Nutritional status: weight (kg).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 17 Time to first pulmonary exacerbation (days).
Comparison 2 Inhaled antioxidant (glutathione) versus control, Outcome 18 Adverse events.
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Lung function FEV1 [% pred] Show forest plot | 3 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 1.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.2 At 6 months (β‐carotene) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.3 At 6 months (oral GSH) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 Lung function FVC [% pred] Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 2.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.2 at 6 months (oral GSH) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 3 QoL: Quality of Well Being Scale Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 3.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 4 Oxidative stress: lipid peroxidation (H2O2) [μmol/L] Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 4.1 At 5 months (selenium) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 5 Oxidative stress: Lipid peroxidation (TBARS) [μmol/L] Show forest plot | 2 | Std. Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 5.1 At 5 months (selenium) | 1 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 5.2 At 6 months (ß‐carotene) | 1 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 6 Oxidative stress: Lipid peroxidation (F2‐isoprostanes) [ng/L] Show forest plot | 1 | Std. Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 6.1 At 2 months (combined supplement) | 1 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 7 Oxidative stress: Enzyme function ‐ GPX [U/g Hb] Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 7.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 7.2 At 5 months (selenium) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 8 Oxidative stress: Enzyme function ‐ SOD [U/mg Hb] Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 8.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 9 Oxidative stress: Potency (TEAC) [mmol/L] Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 9.1 At 6 months (ß‐carotene) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10 Plasma antioxidant status ‐ vitamin E [μmol/L] Show forest plot | 4 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 10.1 At 1 month (fat‐soluble vitamin E) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.2 At 1 month (water‐miscible vitamin E) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.3 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.4 At 2 months (water‐miscible vitamin E) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.5 At 6 months (water‐miscible vitamin E) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.6 At 6 months (oral GSH) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 11 Plasma antioxidant status ‐ β‐carotene [μmol/L] Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 11.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 11.2 At 6 months (ß‐carotene) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 12 Plasma antioxidant status ‐ selenium [μmol/L] Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 12.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 12.2 At 5 months (selenium) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 13 Plasma antioxidant status ‐ vitamin C [μmol/L] Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 13.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 14 Inflammation: plasma fatty acid status [mg/L] Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 14.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 15 Inflammation: IL‐6 (pg/ml) at 3 months (vitamin E) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 15.1 FEV1 >85% and no DNase | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 15.2 FEV1 >85% and DNase | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 15.3 FEV1 range 70% ‐ 85% and DNase | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 16 Inflammation: TNF‐α (pg/ml) at 3 months (vitamin E) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 16.1 FEV1 >85% and no DNase | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 16.2 FEV1 >85% and DNase | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 16.3 FEV1 range 70% ‐ 85% and DNase | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 17 Nutritional status (BMI %) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 17.1 At 6 months (oral GSH) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 18 Nutritional status: weight (kg) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 18.1 At 6 months (water‐miscible vitamin E) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 19 Antibiotic days per patient Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 19.1 At 2 months (combined supplement) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 19.2 At 6 months (ß‐carotene) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 20 Adverse effects (vitamin E) Show forest plot | 1 | Odds Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 20.1 sinusitis | 1 | Odds Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 20.2 pulmonary exacerbations | 1 | Odds Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 20.3 elevated liver enzymes(ALT) | 1 | Odds Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 20.4 distal intestinal obstruction syndrome | 1 | Odds Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 20.5 diarrhea | 1 | Odds Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Lung function FEV1 (% predicted) Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 1.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.2 At 2 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.3 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.4 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 Lung function FVC (% predicted) Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 2.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.2 At 2 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.3 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.4 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 3 QoL score (CF questionnaire for QoL) Show forest plot | 1 | Std. Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 3.1 At 1 month | 1 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 3.2 At 3 months | 1 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 3.3 At 6 months | 1 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 4 Sputum oxidative stress: lipid peroxidation (8‐isoprostan) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 4.1 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 4.2 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 5 Plasma antioxidant status: free glutathione (pM) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 5.1 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 6 Plasma antioxidant status: total glutathione (pM) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 6.1 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 7 Sputum antioxidant status: free glutathione in sputum (pM) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 7.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 7.2 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 7.3 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 8 Sputum antioxidant status: total glutathione in sputum (pM) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 8.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 8.2 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 8.3 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 9 Plasma antioxidant status: glutathione in blood neutrophils (MFI) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 9.1 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10 Sputum antioxidant status: glutathione in sputum neutrophils (MFI) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 10.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.2 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 10.3 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 11 Sputum oxidative stress: protein carbonyls (U) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 11.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 11.2 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 11.3 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 12 Local inflammation: cytokines in sputum IL‐8 (pg/ml) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 12.1 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 13 Local inflammation: cytokines in sputum IL‐10 (pg/ml) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 13.1 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 14 Local inflammation: cytokines in sputum TNF‐α (pg/ml) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 14.1 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 15 Nutritional status: BMI Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 15.1 At 2 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 16 Nutritional status: weight (kg) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 16.1 At 1 month | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 16.2 At 3 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 16.3 At 6 months | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 17 Time to first pulmonary exacerbation (days) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 17.1 By end of study (6 months) | 1 | Mean Difference (IV, Random, 95% CI) | 0.0 [0.0, 0.0] | |
| 18 Adverse events Show forest plot | 4 | Odds Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 18.1 hospitalization for non‐acute pulmonary exacerbations | 1 | 19 | Odds Ratio (M‐H, Random, 95% CI) | 0.39 [0.03, 5.21] |
| 18.2 rhinitis/sinusitis or upper respiratory tract infection | 2 | 172 | Odds Ratio (M‐H, Random, 95% CI) | 1.18 [0.46, 3.00] |
| 18.3 cough | 2 | 172 | Odds Ratio (M‐H, Random, 95% CI) | 1.14 [0.62, 2.09] |
| 18.4 pharyngitis | 2 | 172 | Odds Ratio (M‐H, Random, 95% CI) | 1.12 [0.61, 2.07] |
| 18.5 stomach pain/cramps | 1 | 19 | Odds Ratio (M‐H, Random, 95% CI) | 0.14 [0.01, 1.61] |
| 18.6 headache | 2 | 172 | Odds Ratio (M‐H, Random, 95% CI) | 1.02 [0.51, 2.04] |
| 18.7 chest tightness/bronchospasm | 1 | 19 | Odds Ratio (M‐H, Random, 95% CI) | 0.22 [0.02, 2.67] |
| 18.8 nose bleed | 1 | 19 | Odds Ratio (M‐H, Random, 95% CI) | 0.5 [0.06, 4.00] |
| 18.9 shortness of breath | 1 | 19 | Odds Ratio (M‐H, Random, 95% CI) | 0.39 [0.03, 5.21] |
| 18.10 sputum increase | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 1.13 [0.55, 2.33] |
| 18.11 pyrexia | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 1.77 [0.68, 4.61] |
| 18.12 haemoptysis | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 0.91 [0.43, 1.91] |
| 18.13 lung disorder | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 0.98 [0.38, 2.58] |
| 18.14 sputum abnormal | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 1.60 [0.60, 4.22] |
| 18.15 infection | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 1.43 [0.53, 3.84] |
| 18.16 rales | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 1.11 [0.39, 3.12] |
| 18.17 oropharyngeal pain | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 0.46 [0.15, 1.40] |
| 18.18 condition aggravated | 1 | 153 | Odds Ratio (M‐H, Random, 95% CI) | 1.06 [0.52, 2.17] |
| 18.19 no adverse events | 2 | 68 | Odds Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] |
