Evaluation of the PPAR-γ Agonist Pioglitazone in Mild Asthma: A Double-Blind Randomized Controlled Trial

Background Peroxisome proliferator-activated receptor gamma (PPAR-γ) is a nuclear receptor that modulates inflammation in models of asthma. To determine whether pioglitazone improves measures of asthma control and airway inflammation, we performed a single-center randomized, double-blind, placebo-controlled, parallel-group trial. Methods Sixty-eight participants with mild asthma were randomized to 12 weeks pioglitazone (30 mg for 4 weeks, then 45 mg for 8 weeks) or placebo. The primary outcome was the adjusted mean forced expiratory volume in one second (FEV1) at 12 weeks. The secondary outcomes were mean peak expiratory flow (PEF), scores on the Juniper Asthma Control Questionnaire (ACQ) and Asthma Quality of Life Questionnaire (AQLQ), fractional exhaled nitric oxide (FeNO), bronchial hyperresponsiveness (PD20), induced sputum counts, and sputum supernatant interferon gamma-inducible protein-10 (IP-10), vascular endothelial growth factor (VEGF), monocyte chemotactic protein-1 (MCP-1), and eosinophil cationic protein (ECP) levels. Study recruitment was closed early after considering the European Medicines Agency’s reports of a potential increased risk of bladder cancer with pioglitazone treatment. Fifty-five cases were included in the full analysis (FA) and 52 in the per-protocol (PP) analysis. Results There was no difference in the adjusted FEV1 at 12 weeks (-0.014 L, 95% confidence interval [CI] -0.15 to 0.12, p = 0.84) or in any of the secondary outcomes in the FA. The PP analysis replicated the FA, with the exception of a lower evening PEF in the pioglitazone group (-21 L/min, 95% CI -39 to -4, p = 0.02). Conclusions We found no evidence that treatment with 12 weeks of pioglitazone improved asthma control or airway inflammation in mild asthma. Trial Registration ClinicalTrials.gov NCT01134835


Introduction
Asthma is a common and chronic condition that affects 235 million people worldwide and is responsible for substantial healthcare expenditures [1]. Inhaled corticosteroids and bronchodilators are the mainstay of asthma pharmacotherapy. Although these treatments are highly effective, many patients fail to achieve optimal asthma control and are at risk of impaired quality of life, recurrent exacerbations, and even death. In some patients, these outcomes are due to poor concordance with medication, while others dislike or exhibit reduced responses to inhaled medications, creating a need to identify novel therapeutic targets if the burden of asthma is to be reduced.
PPAR-γ is thought to exert its immune modulating effects by reducing the downstream gene transcription of targets of the pro-inflammatory transcription factors activator protein-1 (AP-1) and nuclear factor-kappaB (NF-kB) [11][12][13]. In cultured human airway smooth muscle cells (HASMs), PPAR-γ activation suppresses the release of the chemokines MCP-1 (monocyte chemotactic protein-1/CCL2) and eotaxin (CCL11) in response to the potent inflammatory stimulant tumor necrosis factor-alpha (TNF-α) [3]. This suppression of eotaxin release is secondary to reduced histone acetylation and reduced binding of NF-kB to the eotaxin promoter, reducing gene transcription [3].
The PPAR-γ agonists pioglitazone and rosiglitazone were developed to treat the hyperglycemia and insulin resistance of type 2 diabetes mellitus, and were used widely until rosiglitazone was withdrawn due to an adverse cardiovascular safety profile [14][15][16]. Prior to its withdrawal, rosiglitazone had been studied in two short-term (up to 28 days) randomized controlled trials (RCTs) in asthma that found modest effects on the late asthmatic reaction and airway obstruction in smokers with asthma [17,18]. A longer term study of pioglitazone in a subset of obese subjects with poorly controlled asthma resulted in no significant improvement in measures of asthma control [19]. The aim of the present study was to test the hypothesis that the activation of PPAR-γ with pioglitazone (which has a more favorable safety profile than that of rosiglitazone) would improve measures of asthma control and markers of airway inflammation over a 12-week period in mild asthma. Study recruitment was terminated after 68 participants had been recruited after reports emerged that pioglitazone treatment may be associated with an increased risk of developing bladder cancer [20].

Study design and participants
A single-center randomized, double-blind, placebo-controlled, parallel-group trial comparing pioglitazone and placebo over 12 weeks in adults with mild asthma was performed at City Hospital, Nottingham, UK. The study design was approved by Nottingham Research Ethics Committee 2 (08/H0408/120). All participants gave written informed consent. The full trial protocol is given in S1 Text.
Potential participants aged 18-75 years with a clinical diagnosis of asthma who were taking 0-800 μg of inhaled beclomethasone dipropionate (or equivalent) and short-acting β 2 -agonists (SABAs) as required were identified from primary care practices and our research database between January 2010 and February 2012. At screening, participants were required to have a post-bronchodilator forced expiratory volume in one second (FEV 1 ) 60% of predicted with 12% reversibility of the pre-bronchodilator FEV 1 with 400 μg of salbutamol or 12% peak flow variability during the run-in period (maximum peak expiratory flow [PEF] minus minimum PEF divided by average PEF) to be eligible.
The exclusion criteria included an exacerbation of asthma within six weeks, current smoking, or a history of more than 10 pack years of smoking, or treatment with leukotriene antagonists, long-acting beta-agonists, theophylline, or oral steroids. Patients with a history of diabetes or liver or cardiovascular disease; patients who were pregnant, lactating, or not using adequate contraception; or patients with contraindications to pioglitazone were excluded for safety reasons.
Screening visit. FEV 1 was measured as the greater of two values within 100 ml using a dry bellows spirometer (Vitalograph, Buckingham, UK), prior to and 15 minutes after the inhalation of 400 μg of salbutamol via a spacer (Volumatic TM , Allen & Hanburys Ltd, Middlesex, UK). Sputum induction with hypertonic saline was performed as described previously [21,22] using the ultrasonic nebulizer OMRON NE-U17 (OMRON Health Care Ltd, Milton Keynes, UK). A clinician reviewed and examined participants at each visit.
Run-in period. Participants entered a two-week run-in period, during which morning and evening PEF (Mini-Wright, Clement Clarke International Ltd, Harlow, UK) and asthma medication use were recorded.
Randomization. Eligible participants were randomized to 12 weeks pioglitazone (30 mg once a day for 4 weeks, then 45 mg once a day for 8 weeks) or matched placebo using an internet-based randomization schedule stratified by inhaled corticosteroid use. The allocation sequence was generated using random permuted blocks of two, four, and six and maintained on a secure server by the Nottingham Clinical Trials Unit (CTU).
Study intervention. Pioglitazone tablets were over encapsulated with DBcaps1 capsules (Capsugel Belgium NV, Bornem, Belgium), and a matched placebo backfilled with lactulose was manufactured by Catalent Pharma Solutions, Bolton, UK. An independent pharmacist issued 30 capsules of the study medication to each participant every four weeks and compliance was assessed by counting the returned capsules. All participants and clinical investigators performing the clinical assessments were blinded to allocation until all trial procedures were performed and the analysis was complete.
Outcome measures and statistical analysis. The primary outcome measure was the adjusted mean FEV 1 at 12 weeks compared between treatment groups by analysis of covariance (ANCOVA) incorporating terms for baseline value, treatment arm, inhaled corticosteroid use, age, gender, and height. The secondary outcome measures at 12 weeks included mean morning and evening PEF for 14 days prior to the study visit, ACQ and AQLQ scores, FeNO, PD 20 , SABA use, and body mass index (BMI). We considered a PD 20 of 800 μg to be equivalent to a PC 20 of 8 mg/mL [26]. Final follow up visits were in December 2011.
In subsets of participants, induced sputum cell counts, sputum supernatant interferon gamma-inducible protein-10 (CXCL10/IP-10), vascular endothelial growth factor (VEGF), MCP-1/CCL2, and eosinophilic cationic protein (ECP) were measured. Activation of PPAR-γ was assessed by comparing levels of cytoplasmic and nuclear PPAR-γ from induced sputum cells between groups. Adverse effects (reported spontaneously at clinic visits and on diary cards) were recorded, and random glucose, full blood count (FBC), and liver function tests (LFTs) were measured pre-and post-treatment.
Where appropriate, secondary endpoints were log-transformed to approximate a normal distribution (PD 20 , percentage eosinophil count, and FeNO) and compared between groups using ANCOVA, incorporating terms for baseline value treatment arm, inhaled corticosteroid use, and, where appropriate, age, gender, and height. Values that could not be transformed were compared using the Mann-Whitney U test. In participants with values below the lower limit of detection in laboratory assays (five participants for ECP), a value of 0.5 of the lower detection limit was used. Sputum mediators were expressed per 1 x 10 6 cells and were reported as geometric means with 95% confidence intervals (CIs). Stata v11.0 (StataCorp LP, College Station, TX, USA) and GraphPad Prism Version 6 (GraphPad Software, La Jolla, CA, USA) were used to perform the statistical analyses.
The a priori calculation suggested that 88 participants would allow us to detect a difference of 150 mL between groups in the mean FEV 1 at 12 weeks, with 80% power at a 5% two-sided significance level, assuming a standard deviation (SD) of FEV 1 over eight weeks of 250 mL [27].
A Trial Steering Committee (TSC) and independent Data Monitoring and Ethics Committee (DMEC) monitored the progress and safety of the trial.
Sputum processing. Induced sputum samples were processed within two hours of induction as described previously [21,22]. Sputum supernatant IP-10, MCP-1, and VEGF were assayed using the Bio-Plex1 200 system (Bio-Rad, Hemel Hempstead, UK) and supernatant ECP was assayed using ImmunoCAP 100 (Phadia AB, Uppsala, Sweden) according to the manufacturers' instructions. PPAR-γ was assayed in the nuclear and cytoplasmic fractions of sputum cells using a TransAM™ PPAR-γ Kit (Active Motif, Rixensart, Belgium) and a competitive enzyme-linked immunosorbent assay (ELISA) PPAR-γ kit (Biosources, Inc., San Diego, CA, USA) according to the manufacturers' instructions. The optical density at 450 nm (OD 450 ) was used for quantification of nuclear PPAR-γ and the reciprocal of the OD 450 was used for quantification of cytoplasmic PPAR-γ.

Results
Of the 119 individuals screened, 68 participants were randomized, 34 to pioglitazone and 34 to placebo between January 2010 and February 2012 (Fig 1). The full trial dataset is given in S1 Data and Consort checklist in S2 Text.
The full analysis was performed on 55 completed cases, and after excluding those with protocol deviations, the per-protocol analysis included 52 participants. The baseline characteristics of the study population are shown in Table 1. Subjects were well matched between treatment  Short-acting beta 2 -agonist use during run-in (actuations/day) 0.7 ± 1 1.5 ± 1.8 0.7 ± 1 Sputum differential eosinophil count % Geometric mean (95% CI) arms, 75% were taking inhaled corticosteroids with a SABA, and 25% were taking a SABA alone. Forty-one participants produced viable induced sputum samples pre-and post-treatment (Tables 1 and 2). Medication adherence assessed by tablet counting was greater than 95% in both groups over the course of the study.

Closure of recruitment
The European Medicines Agency (EMA) reported that pioglitazone may be associated with an increased risk of developing bladder cancer in July 2011 (20). Because there were no proven benefits associated with taking pioglitazone in asthma, the TSC concluded that the risk to benefit ratio had become unacceptable for participants with mild asthma and recruitment was stopped on 14 October, 2011. At that point, 68 of the minimum requirement of 88 participants had been randomized, with 52 participants having completed the study and 10 still in the treatment phase of the study. Among those still participating in the study, 70% withdrew after receiving notification of the EMA report, while three participants elected to complete the study. All participants and investigators remained blinded to allocation until the analysis was completed.

Efficacy
There was no significant difference observed in the primary outcome measure, the adjusted mean FEV 1 at 12 weeks, between the pioglitazone and placebo groups (adjusted mean difference -0.014 L, 95% CI -0.15 to 0.12, p = 0.84). In the FA there were no differences in any of the secondary outcomes between the groups at 12 weeks, although there was a trend toward a lower adjusted mean morning PEF (-16 L/min 95% CI -33 to 2, p = 0.075) and evening PEF (-17 L/min, 95% CI -35 to 1, p = 0.068) in the pioglitazone group. There was also a trend toward a higher use of as required SABA use in the pioglitazone group (0.6 actuations/day, 95% CI -0.05 to 1.2, p = 0.07) (Fig 2 and Table 3).
The per protocol analysis replicated the full analysis, except that the mean evening PEF was significantly lower in the pioglitazone group after 12 weeks (-21 L/min, 95% CI -39 to -4, p = 0.02) ( Table 4). In subsets of participants that provided induced sputum samples pre-and post-treatment, there were no significant differences in the induced sputum differential cell counts or the induced sputum supernatant ECP, MCP-1, IP-10, or VEGF, or the nuclear and cytoplasmic levels of PPAR-γ between groups (Figs 2 and 3).
There was an increase in the reporting of expected side effects of pioglitazone in the active group, particularly ankle swelling, weight gain, insomnia, and lethargy (Table 5). There were no asthma exacerbations requiring oral corticosteroids or hospital admission in the study; however, five participants in the pioglitazone group and two in the placebo group reported mild/moderate asthma exacerbations that required an increase in SABAs.

Discussion
In this randomized controlled trial, the administration of pioglitazone for 12 weeks in mild asthma did not result in any improvements in a range of markers of asthma control or measures of airway inflammation. In fact, we observed a lower post-treatment mean evening PEF and a trend toward higher SABA reliever use in the pioglitazone group, which was not accompanied by significant deterioration in asthma control or scores on quality of life questionnaires. Although these findings most likely reflect the multiplicity of analyses, they could have been caused indirectly by side effects of pioglitazone such as insomnia and fatigue. If the reduction in PEF is a true effect, and assuming a minimum important difference (MID) for PEF of 15 to 20 L/minute, this is of a magnitude that could have been perceived by the participants [28]. Values are reported as arithmetic mean unless specified ± Standard deviation ¶ Geometric means are reported with a 95% confidence interval (CI) The adjusted mean of the secondary clinical outcomes at 12 weeks was compared between groups using ANCOVA including terms for baseline value, treatment arm, inhaled corticosteroid use, age, gender, and height and expressed as the difference in the adjusted means with a 95% confidence interval Despite evidence that PPAR-γ activation has anti-inflammatory effects in experimental models of asthma, there are a number of potential reasons why pioglitazone may not have exerted an effect in the present study. Firstly, the early closure of recruitment limited the number of participants completing the study, leaving the study underpowered to detect a difference of 150 ml between groups. However, considering the narrow 95% CIs for the difference in FEV 1 , the lack of change in airway hyper-responsiveness, lower PEF, and trend toward higher SABA use in the pioglitazone group, it is unlikely that we missed a clinically important effect on airway dysfunction in mild asthma.
It is possible that the observed lack of efficacy was because pioglitazone was not biologically active. Although we did not measure plasma levels of pioglitazone, more than 95% of the study  Values are reported as arithmetic mean unless specified ± Standard deviation ¶ Geometric means are reported with a 95% confidence interval (CI) The adjusted mean of the secondary clinical outcomes at 12 weeks was compared between groups using ANCOVA including terms for baseline value, treatment arm, inhaled corticosteroid use, age, gender, and height and expressed as the difference in the adjusted means with a 95% confidence interval (CI) and p-value.   Table 5. Reported adverse events occurring in more than 1 participant by system.

Investigations
Weight increased (> 5% of baseline) 3 1 capsules were taken by participants and we observed a decrease in hemoglobin, a recognized systemic side effect of pioglitazone [29][30][31]. However, despite evidence of systemic biological activity, we were unable to confirm that pioglitazone successfully activated PPAR-γ in the lungs to induce nuclear translocation of PPAR-γ within the induced sputum cells. As pioglitazone is highly bound to plasma proteins (>99%) and has a low volume of distribution across tissues, it may not have entered the airways in sufficient concentration to exert an effect. An alternative explanation for the lack of efficacy is that the study participants had adequate asthma control with low levels of eosinophilic airway inflammation and preserved lung function at baseline, leaving little potential for improvement. Recruiting patients with evidence of eosinophilic inflammation would have been better but would have rendered recruitment into the study even more difficult. Therefore, we have not excluded a possible anti-inflammatory effect in patients with more severe asthma and more marked airway inflammation, and a study with longer duration in this cohort would be required to assess the effect of pioglitazone on the risk of asthma exacerbations. Values reported as arithmetic mean unless specified ± Standard deviation The adjusted mean of the secondary outcomes at 12 weeks was compared between groups using ANCOVA including terms for baseline value, treatment arm, inhaled corticosteroid use, age, gender, and height and expressed as the difference in the adjusted means with a 95% CI and p-value. ALT-alanine aminotransferase. GGT-gamma glutamyl transferase. ANCOVA-analysis of covariance doi:10.1371/journal.pone.0160257.t006 To our knowledge, this is the largest and longest study of a PPAR-γ agonist in asthma performed to date [16][17][18][19]. A previously reported RCT compared rosiglitazone to inhaled corticosteroids in 46 current smokers with asthma (a group associated with poor response to inhaled corticosteroids), and reported a trend toward a higher pre-bronchodilator FEV 1 in the group treated with rosiglitazone at 28 days (183 mL, 95% CI -1.0 to 367, p = 0.051) (18). There was also an improvement in the forced expiratory flow between 25 and 75% of the forced vital capacity after 28 days of treatment with rosiglitazone (FEF 25-75 0.24 L/second, 95% CI 0.03 to 0.46, p = 0.03), but no statistically significant improvements in ACQ scores or sputum cell counts [18]. A second RCT compared 28 days of treatment with rosiglitazone with placebo in 32 adults with mild asthma and reported a 15% attenuation of the late asthmatic reaction, with no change in the early asthmatic reaction compared to placebo. The authors also reported no effect on airway hyper-responsiveness after allergen challenge or on a panel of inflammatory markers [17]. Dixon and colleagues reported a 12 week RCT comparing pioglitazone (n = 23) to placebo (n = 19) in obese subjects with poorly controlled moderate to severe asthma [19]. Using 30mg pioglitazone for 2 weeks increased to 45mg for 10 weeks there were no significant differences in the airway hyper-responsiveness to methacholine, FeNO or lung function between treatment and placebo. Recruitment was halted early (as in our study) due to the impact of emerging safety concerns surrounding the potential risk of bladder cancer affecting recruitment, lack of efficacy signal and significant weight gain in pioglitazone treated patients [19]. The thiazolidinediones are associated with significant adverse effects when administered systemically and these studies support our conclusion that systemic treatment with pioglitazone is not associated with beneficial effects on asthma control.

Conclusion
We found no evidence to support the hypothesis that 12 weeks of treatment with the PPAR-γ agonist pioglitazone improves asthma control or airway inflammation in individuals with mild asthma. Opportunities to further investigate the potential anti-inflammatory effects of pioglitazone in more severe asthma are currently limited by safety considerations.