Simvastatin Suppresses Airway IL-17 and Upregulates IL-10 in Patients With Stable COPD

BACKGROUND: Statins have immunomodulatory properties that may provide beneficial effects in the treatment of COPD. We investigated whether a statin improves the IL-17/IL-10 imbalance in patients with COPD, as has previously been demonstrated in patients with asthma. METHODS: Thirty patients with stable COPD were recruited to a double-blind, randomized, controlled, crossover trial comparing the effect of simvastatin, 20 mg po daily, with that of a matched placebo on sputum inflammatory markers and airway inflammation. Each treatment was administered for 4 weeks separated by a 4-week washout period. The primary outcome was the presence of T-helper 17 cytokines and indoleamine 2,3-dioxygenase (IDO) in induced sputum. Secondary outcomes included sputum inflammatory cells, FEV1, and symptoms using the COPD Assessment Test (CAT). RESULTS: At 4 weeks, there was a significant reduction in sputum IL-17A, IL-22, IL-6, and CXCL8 concentrations (mean difference, −16.4 pg/mL, P = .01; −48.6 pg/mL, P < .001; −45.3 pg/mL, P = .002; and −190.9 pg/mL, P = .007, respectively), whereas IL-10 concentrations, IDO messenger RNA expression (fold change), and IDO activity (kynurenine to tryptophan ratio) were markedly increased during simvastatin treatment compared with placebo treatment periods (mean difference, 24.7 pg/mL, P < .001; 1.02, P < .001; and 0.47, P < .001, respectively). The absolute sputum macrophage count, proportion of macrophages, and CAT score were reduced after simvastatin compared with placebo (mean difference, −0.16 × 106, P = .004; −14.1%, P < .001; and −3.2, P = .02, respectively). Values for other clinical outcomes were similar between the simvastatin and placebo treatments. CONCLUSIONS: Simvastatin reversed the IL-17A/IL-10 imbalance in the airways and reduced sputum macrophage but not neutrophil counts in patients with COPD. TRIAL REGISTRY: ClinicalTrials.gov; No.: NCT01944176; www.clinicaltrials.gov

Cytokines play a critical role in the pathobiologic processes of COPD, including altered innate immune response, chronic infl ammation, emphysema, and small airway fi brosis. 1 Proinfl ammatory cytokines of potential importance include tumor necrosis factor-a , interferon-g , IL-1 b , IL-17, and IL-6. COPD has been associated with an increased IL-17 response directed against innocuous antigens. 2 IL-17 promotes chronic airway infl ammation by primarily acting on the lung epithelium through the upregulation of proinfl ammatory cytokines and chemokines. 3,4 Genetic deletion of IL-17A attenuated cigarette smoke-induced infl ammation and alveolar type 2 apoptosis in mice. 5 Th e expression of IL-17A in human bronchial submucosa was signifi cantly increased in patients with COPD compared with healthy control subjects and normal smokers. 6,7 Accumulating evidence suggests that CD4 1 T cells, including T regulatory (Treg) cells and T-helper (Th )17 cells, possess a greater degree of plasticity in their diff erentiation options than previously appreciated. 8 It appears that expression of Foxp3 by Treg cells or ROR g t by Th 17 cells may not be stable. 8 Th 17 cells produce IL-17 and IL-22, thereby boosting infl ammation, while Treg cells express IL-10 and tumor growth factor-b , suppressing infl ammation at least in part through the immunosuppressive eff ects of indoleamine 2,3-dioxygenase (IDO). Th e induction of IDO-enhanced tryptophan (Tryp) catabolism into kynurenine (Kyn), which can inhibit the accumulation of Th 1 and Th 17 cells at the site of infl ammation and, therefore, attenuate the degree of infl ammation. 9 Th is may explain the imbalance between Th 17 and Treg cytokines in COPD that we have previously documented. 10 We have shown that sputum IL-17A concentrations are associated with COPD severity and inversely correlated with IL-10 concentrations, with reduced expression of the immunosuppressive enzyme IDO and its bioactivity. 11 Th is may contribute, in part, to a further enhancement of airway infl ammation in COPD.
Statins are 3-hydroxy-3-methyl-glutarylcoenzyme A reductase inhibitors that have been clinically used as lipid-lowering agents. Statins, however, have additional pleiotropic pharmacologic eff ects, including antiinfl ammatory, antioxidant, and immunomodulatory activities in vitro and in vivo. [12][13][14][15] Th e immunomodulatory eff ects of statins on Th 17 cell-and IL-17-mediated infl ammatory responses have been well established in autoimmune diseases, including multiple sclerosis in humans and experimental autoimmune encephalomyelitis in mice. Statins mediate their action via suppression of Th 17 diff erentiation with the concurrent induction of Treg diff erentiation. 16 Emerging evidence suggests that statins inhibit the release of airway neutrophilic mediators (CXCL8, IL-6, and granulocyte-macrophage colony-stimulating factor) from bronchial epithelial cells and suppress their upregulation by IL-17. 17 Interestingly, all of these additional actions may counteract the neutrophilic infl ammationpromoting eff ects of IL-17 in COPD. To date, there has been a paucity of studies exploring the antiinfl ammatory eff ects of statins in COPD and particularly on the reversal of IL-17A/IL-10 or Th 17/Treg imbalance.
We conducted a double-blind, placebo-controlled crossover study to ascertain the eff ect of simvastatin on Th 17 cytokines and Th 17-polarizing cytokine expression and chronic airway infl ammation in COPD. To our knowledge, this study is the fi rst to show that simvastatin inhibits IL-17, IL-22, CXCL8, and IL-6, but enhances IL-10, IDO messenger RNA expression, and IDO biologic activity. However, these eff ects of simvastatin were not associated with the attenuation of airway neutrophilia but unexpectedly resulted in a marked decrease in macrophage numbers.

Study Design and Assignment
The study was a 4-week, randomized, double-blind, crossover study comparing the eff ect of oral simvastatin treatment (20 mg daily) with that of a matched placebo on sputum cytokine biomarkers and airway infl ammation in COPD. Aft er a 2-week run-in period, each treatment was administered to randomized patients for 4 weeks, separated by a 4-week washout period. Researchers and participants were blinded to allocation and had no access to the randomization code held by the data center until completion of the study. During the 2-week run-in period and throughout the study, subjects continued their usual COPD medication and withdrew statin therapy for 4 weeks prior to the study entry if they were taking regular treatment. Further visits were undertaken at randomization and after 4 weeks (phase 1). Aft er a washout period of 4 weeks, phase 2 of the crossover was started with a visit after 4 weeks. At each visit, spirometry was performed. Before and aft er each treatment period, sputum induction was performed and blood samples were taken to measure lipid levels and liver function. Tablets were counted at the end of each treatment period as a measure of treatment adherence.

Patients
Patients with COPD aged 45 to 80 years were recruited from the COPD Outpatients Clinic at Siriraj Hospital. Patients with COPD met the following inclusion criteria: current or ex-smokers with a Ն 10 pack-year history; postbronchodilator FEV 1 to FVC ratio , 70% and FEV 1 , 80% and Ն 50% of predicted normal values (GOLD [Global Initiative for Chronic Obstructive Lung Disease] grade 2) at visit 1 (screening); and clinically stable as defi ned by no exacerbations in the previous 6 weeks.
Key exclusion criteria were a primary diagnosis of asthma; history of signifi cant diseases other than COPD, including TB, lung cancer, and HIV; acute worsening of COPD that required treatment within 6 weeks prior to screening or between the screening and baseline visits; cognitive impairment; recent cardiovascular and cerebrovascular diseases within 6 months prior to study entry; current or previous use of immunosuppressive agents; and current administration of macrolides, azole antifungal drugs, and amlodipine.
Recruited study participants who were taking certain COPD medications at screening (ie, tiotropium, long-acting b 2 -agonists [LABAs], combination corticosteroid and LABA products, theophylline, oral b 2 -agonists) were permitted to continue these medications throughout the trial to maintain their clinical stability, with the exception of the study visit. Study participants were required to withhold all COPD medications (including inhaled corticosteroids [ICSs] and short-acting bronchodilators) for at least 8 h before the baseline visit and all subsequent study treatment visits.
At visit 2, study participants were randomized to one of two possible treatment sequences. Each sequence included either 4-week administration of simvastatin (20 mg/d) or placebo in a double-blind manner. Patients returned to the clinic for washout at visit 3, for treatment at visit 4, and for a final visit at visit 5. There was a washout interval between treatments of 4 weeks.

Measurements
Demographic measurements were recorded on the fi rst clinical visit (visit 1 screening). Induced sputum and blood samples were collected before and aft er treatment periods for analysis of sputum cytokines and diff erential leukocyte count and assessment of lipid profi les and highsensitivity C-reactive protein (hsCRP). Spirometry, COPD Assessment Test (CAT) score, and impulse oscillometry were measured at all study visits. Th e primary end point was IL-17A, IL-10, and IL-6 levels, and IDO activity. Secondary end points included CAT scores, sputum neutrophil count, and postbronchodilator FEV 1 .

Sputum Induction and Processing
Sputum induction was performed as previously described. 11 Th e supernatants were kept frozen at 2 70°C until further analysis. For immunocytochemistry, cytospins were fi xed with 4% paraformaldehyde (BDH Ltd) and stored at 2 20°C. Total cell counts were recorded with a hemocytometer, using Kimura staining. Cell viability was determined by Trypan blue exclusion before cytospins were undertaken. Th e slides were stained with May-Grunwald-Giemsa stain and differential cell counts were made by a blinded observer. A total of 400 infl ammatory cells were counted on two slides for each sample in a blinded manner. Diff erential cell counts are expressed as the percentages of total infl ammatory cells. Samples with cell viability . 70% and , 30% squamous cell contamination were considered adequate for analysis.
It should be noted that the standard range of the ELISA assay used in this study was lower than that of the ELISA assays used by previous studies, possibly resulting in the baseline levels of some cytokines reported in this study being different from those reported in other studies. In addition, the differences of cytokine measurements in our study from others might be explained by the nature of our recruits. Th e majority of our subjects were treated with ICSs, whereas other studies recruited steroid-naive patients with COPD, those with mild COPD without treatment, or minority of patients with steroid-treated severe COPD. 18,19 Th e diff erence in the nature of steroid use might contribute to the low basal levels of several cytokines.

RNA Isolation and Real-Time Reverse Transcription-Polymerase Chain Reaction Analysis
Total RNA was extracted from whole sputum cell pellets by using an RNeasy Mini kit (Qiagen NV) according to the manufacturer's instructions. Total RNA was reverse transcribed into cDNA by using Improm-II Reverse Transcription system (Promega Corporation).

Liquid Chromatography-Mass Spectrometry Analysis of L-Tryp and L-Kyn
Analysis of L-Tryp and L-Kyn was performed using a validated highperformance liquid chromatography with tandem mass spectrometry method as previously described. 20 Briefl y, 3-nitro-l -tyrosine was added to each sample as an internal standard, and protein precipitation was performed using 10% trichloroacetic acid. Chromatographic separation of the subsequent organic layer was carried out on liquid chromatographytandem mass spectrometry with C18, 2.5 mm (50 3 3.0 mm internal diameter). A mobile phase consisting of acetonitrile and 0.1% formic acid (gradient condition) was delivered at a fl ow rate of 0.2 mL/min. Mass spectra were obtained using a Quattro Premier XE mass spectrometer (Micromass; Waters Corporation) operated in multiple reaction monitoring mode. Sample introduction and ionization were performed by electrospray ionization in the positive ion. Th e mass transition ion pair for L-Tryp protonated species ([M 1 H] 1 ) and L-Kyn L-Tryp [M 1 H] 1 ions was selected as mass to charge ratio 205.08 . 188.00 and 205.08 . 146.08, respectively. The mass transition ion-pair for 3-nitro-l -tyrosine [M 1 H] 1 ions was selected as mass to charge ratio 227.02 . 181.03. The data acquisition was ascertained by Masslynx 4.1 soft ware (Waters Corporation). Th e lower limit of detection for L-Tryp and L-Kyn in sputum supernatants were 0.05 m g/mL and 0.15 m g/mL, respectively. Th e best linear fi t was achieved with a 1/x weighting factor, showing a mean correlation coeffi cient Ն 0.998.

Statistical Analyses
Baseline characteristics were described by number and percentage of patients for categorical variables and mean ( Ϯ SD) for continuous variables. Response to simvastatin on lung function, induced sputum, and mediator levels vs placebo was assessed by general linear model for the standard 2 3 2 crossover design. When variables were unsuitable for this, the within-patient treatment diff erences were calculated and then analyzed by a paired t test for parametric and Wilcoxon signed rank test for nonparametric data. Th e eff ects of ICS treatment and smoking status on the response to simvastatin were analyzed by an unpaired t test. Signifi cance at a level of 5% was accepted for the primary end point.
A sample size of 21 was calculated to have 80% power to detect a difference in means of 0.23 in IDO activity (the ratio of Kyn to Tryp) and 15 pg/mL in sputum IL-17A (primary end point), assuming an SD of difference of 0.35 and 20 pg/mL using a paired t test with a 5% twosided signifi cance level. A total of 30 patients were recruited to ensure that 21 patients completed the study. All data were analyzed using PASW statistics 18 (SPSS; IBM Corporation).

Recruitment and Baseline Characteristics
Screening visits were arranged for 30 patients; 26 patients were randomized to therapy because four were excluded prior to randomization ( Fig 1 ). Of the 26 patients, 21 completed the treatment phase and fi ve discontinued the study before completion, as indicated in Figure 1 . Baseline demographic and clinical characteristics of the patients are shown in Table 1 and baseline infl ammatory marker levels in Table 2 . Almost all patients had symptomatic GOLD B COPD, according to GOLD classifi cation. Distributions of baseline characteristics were similar for patients starting with placebo and those starting with simvastatin. concentrations was signifi cant in patients with COPD treated with simvastatin compared with control subjects receiving placebo (mean diff erence, 2 16.4 pg/mL; 95% CI, 2 28.3 to 2 4.4; P 5 .01) ( Fig 2A , Table 3 ). Th is suppression was not dependent upon the sequence of treatments either with simvastatin fi rst and placebo second or vice versa ( P 5 .74) ( Table 3 ). Th ere was no significant diff erences in the IL-17A response to simvastatin in patients taking ICSs compared with those not taking ICSs and between ex-smokers and current smokers ( P 5 .9 and P 5 .5, respectively). Similarly, the magnitude of the reduction in sputum IL-22 levels was signifi cant in patients with COPD treated with simvastatin compared with control subjects receiving placebo (mean diff erence, 2 48.6 pg/mL; 95% CI, 2 58.4 to 2 38.9; P , .001) ( Fig 2B , Table 3 ). IL-22 response to simvastatin treatment was not aff ected by ICS treatment and smoking status ( P 5 .09 and P 5 .4, respectively). However, IL-23 was undetectable in sputum (data not shown).

IL-10 and IDO:
Patients with COPD treated with simvastatin had higher concentrations of IL-10 in sputum supernatants compared with the baseline levels, whereas there was no signifi cant increase of IL-10 in patients with COPD who received placebo (median [IQR], 28.8 pg/mL [17.5-41.7 pg/mL] vs 7.2 pg/mL [5.9-9.0 pg/mL], P , .001; 8.3 pg/mL [6.5-10.0 pg/mL] vs 8.1 pg/mL [6.6-9.5 pg/mL], P 5 .78, respectively). In addition, the magnitude of the increase in IL-10 levels was signifi cant in patients with COPD treated with simvastatin compared with subjects treated with placebo (mean diff erence, 24.7 pg/mL;  Data are presented as mean Ϯ SD unless otherwise indicated. BD 5 bronchodilator; D 5 change in; D LCO 5 diff using capacity of the lung for carbon monoxide; ICS 5 inhaled corticosteroid; LABA 5 long-acting b 2 agonist; LAMA 5 long-acting muscarinic antagonist; V A 5 alveolar volume. 95% CI, 15.9-33.6; P , .001) ( Fig 3A , Table 3 ). Th is eff ect was not dependent upon the sequence of treatments between simvastatin and placebo ( P 5 .65). Patients without ICS treatment had a signifi cantly greater response of IL-10 to simvastatin than those with ICS treatment, whereas smoking status did not aff ect IL-10 response ( P 5 .037 and P 5 .9, respectively  Table 3 ). Th is suppressive eff ect was not dependent upon the sequence of treatments in a crossover-designed manner ( P 5 .77 for IL-6 and P 5 .87 for CXCL8). Similarly, ICS treatment and smoking status did not infl uence IL-6 and CXCL8 responses to simvastatin ( P 5 .05 and P 5 .09, respectively, for IL-6; P 5 .06 and P 5 .3, respectively, for CXCL8).

Induced Sputum Cytology
Th e total cell counts recovered from sputum were similar aft er simvastatin and aft er placebo treatment (  Table 3 ). Th ere was a reciprocal increase in the relative proportion of sputum airway epithelial cells (mean proportion diff erence, 7.6%; 95% CI, 2.3-12.9; P 5 .007) ( Table 3 ), but there was no signifi cant changes in the absolute count of these cells or the counts and proportions of the other sputum cell phenotypes under simvastatin treatment.

Changes in Clinical Outcomes
Changes in clinical outcomes aft er simvastatin treatment are listed in Table 4 . At 4 weeks, the change in mean CAT score aft er simvastatin compared with placebo treatment periods was statistically signifi cant (mean difference, 2 3.2; 95% CI, 2 6.0 to 2 0.4; P 5 .02) ( Table 4 ).
No statistically signifi cant eff ect of simvastatin was seen in post-salbutamol FEV 1 , post-salbutamol FVC, or lung resistance at 5 Hz to 20 Hz compared with placebo treatment ( Table 4 ).
No signifi cant adverse events occurred in patients taking simvastatin or in those receiving placebo.

Discussion
Th is study has demonstrated the ability of simvastatin to suppress Th 17 cytokines (IL-17 and IL-22), CXCL8, and IL-6, with concomitant enhancement of IL-10 production and IDO expression and activity in patients with COPD. Surprisingly, simvastatin had no signifi cant eff ect on neutrophilic airway infl ammation, although it did suppress macrophage numbers and improved CAT symptom score.
To our knowledge, no previous study has investigated the antiinfl ammatory eff ects of statins in COPD. In particular, the eff ects on Th 17 cytokines, including IL-17 and Il-22, have not been investigated and are unknown in COPD.
Several studies have suggested that COPD may have an autoimmune component that is related to Th 17 cells, a subset of CD4 1 T cells, present in COPD. [21][22][23] Peripheralblood Th 17 cell count was increased in patients with COPD, which was predictive of the severity of airfl ow limitation. 24    IL-17A is a proinfl ammatory cytokine that regulates airway infl ammation and modulates lung and airway structural cells in COPD through the recruitment of infl ammatory cells, including neutrophils and lymphocytes. IL-17A acts on airway epithelial cells, smooth muscle cells, and airway fi broblasts to release neutrophil chemoattractants, including CXCL8. 25 In addition, IL-17 exerts its eff ect on most parenchymal cells, including macrophages and dendritic cells that express IL-17 receptors, and IL-17-mediated signaling induces the target cells to produce various infl ammatory mediators such as tumor necrosis factor-a and IL-6. Studies have shown that the expression of IL-17A and IL-22 was increased in the airways of patients with COPD and that IL-22 can promote airway infl ammation by acting in synergy with IL-17A. 26,27 Th ese data suggest that inhibition of Th 17-related cytokines might be a useful strategy for COPD treatment but  release. Th is eff ect is associated with geranylgeranylationdependent inhibition of small GTPases, including RhoA, Rac1, and Rab1, that control IL-23 production. 30 In addition, simvastatin can directly inhibit IL-17 gene expression in CD4 1 T cells in a mevalonate-dependent manner. 28 Treg cells exert their immune-suppressive ability, in part, through secretion of IL-10. A study has shown that the Th 17/Treg imbalance, refl ected by an increased IL-17A/IL-10 ratio, exists in mice aft er chronic cigarettesmoke exposure and in patients with stable COPD, and this could play a role in the breakdown of immune selftolerance in COPD. 11,31,32 IL-10 transcription and its levels in the airways and lung parenchyma correlated inversely with the severity of COPD and emphysema extent, respectively. 11,33 In contrast, other studies showed no signifi cant diff erence in IL-10 levels in BAL from patients with COPD compared with smoker control subjects. 34 Nevertheless, boosting IL-10 production by simvastatin, as shown in the present study, might restore an impaired immune self-tolerance in patients with COPD. In addition, simvastatin enhanced the expression of the airway immunomodulating enzyme IDO by upregulating its transcription and biologic activity, refl ected by the generation of the Tryp metabolite Kyn. Increased IDO activity may lead to Treg cell expansion through the activation of aryl hydrocarbon receptors by Kyn. 35 Furthermore, accumulation of Kyn resulted in subsequent tolerogenic eff ects, including increased Treg cell activity. 36 In a mouse model, the neutrophilic alveolitis associated with acute lung injury is markedly reduced with lovastatin treatment. 37 We found no signifi cant reduction in the proportion of neutrophils in induced sputum with simvastatin treatment, despite a marked suppression in CXCL8 and IL-17A levels. Th e absolute neutrophil counts were not signifi cantly diff erent between groups, suggesting that the increased percentage of neutrophils aft er simvastatin treatment was because of a decrease in percentage of macrophages. In contrast, absolute macrophage counts and the proportion of macrophages were markedly reduced in simvastatin treatment compared with placebo treatment, possibly resulting from the reduction in IL-17A that contributes to macrophage recruitment by activating IL-17A receptors on macrophages. 38 In keeping with the fi nding of a reduced sputum macrophage count with simvastatin in COPD, a clinical trial in asthma reported atorvastatin treatment, compared with placebo, was associated with a reduction in the absolute sputum macrophage count. 39 Th ere are several possible explanations for the apparent lack of clinical eff ects of simvastatin on airway neutrophilia. The first possibility is that the calculation of sample size in this study had been based on signifi cant reduction in IDO activity but not the proportion of neutrophils and the absolute neutrophil counts. Th e second possibility is that there was another chemokine driving airway neutrophilia, which could not be suppressed by Figure 5 -A, Absolute macrophage counts were lower in response to simvastatin treatment to a greater extent than placebo. B, A similar pattern was seen in the proportion of macrophages (% of total cells). Th e data are expressed as mean (SD). Data are presented as mean Ϯ SD. CAT 5 COPD Assessment Test; R5-R20 5 lung resistance at 5-20 Hz. See Table 1 for expansion of other abbreviations.
simvastatin, such as the tripeptide collagen-derived neutrophil chemoattractant N-acetyl proline-glycineproline (N-ac-PGP), derived from the breakdown of extracellular matrix in the process of alveolar destruction emerging in emphysema. N-ac-PGP plays a role in the airway and parenchymal neutrophilic infl ammation that drives COPD progression and exacerbations. N-ac-PGP induced neutrophil chemotaxis and superoxide production through CXCR2 receptors. [40][41][42] In an animal model, the administration of N-ac-PGP caused recruitment of neutrophils into lungs of control mice, but not CXCR2defi cient mice. 40 In contrast, the blockade of N-ac-PGP with monoclonal antibody suppressed neutrophil responses. N-ac-PGP treatment also caused alveolar enlargement and, therefore, N-ac-PGP activity links degradation of extracellular matrix with neutrophil recruitment in airway infl ammation. Th e considerable concentrations of N-ac-PGP were detectable in the majority of BAL and sputum samples from patients with stable COPD. 40,43 In addition, cigarette-smoke extractactivated neutrophils can break down collagen to generate N-ac-PGP that can activate neutrophils, leading to the induction of a self-perpetuating cycle of neutrophil infi ltration, chronic infl ammation, and lung emphysema. 44 Sputum N-ac-PGP levels were dramatically increased at the time of an exacerbation of COPD, and azithromycin treatment signifi cantly reduced sputum N-ac-PGP levels in patients with COPD. 45 Suppression of N-ac-PGP may be one of the mechanisms whereby macrolides reduce COPD exacerbations in a clinical trial. 46 All these data raise the possibility that N-ac-PGP contributes to airway neutrophilia in patients with COPD. In addition, the majority of the patients with COPD in this study were current smokers. It is possible, therefore, that persistent extracellular N-ac-PGP generation induces relentless neutrophilic airway inflammation that is resistant to statin therapy. Other mechanisms driving neutrophilic infl ammation in COPD, such as leukotriene B4 or complements, may not be suppressed by a statin, but this requires further studies.
Statins reduce the recruitment of neutrophils and macrophages into the lung. 47,48 Our results extend the previous fi ndings that the mechanisms by which statins suppress the recruitment of neutrophils and macrophages into the lung are possibly mediated through the inhibition of Th 17 cytokines and CXCL8 that could mediate the recruitment. Th e baseline levels of CXCL8 and IL-6 are associated with the frequency of COPD exacerbations. 49,50 During COPD exacerbations, CXCL8 and IL-6 levels are significantly elevated in the airways, compared with the stable phase, 51 suggesting that the development of COPD exacerbations is driven by airway infl ammation driven, in turn, by these cytokines. Th ese biologic eff ects of statins appear to have clinical relevance, as other observational studies reported reduced frequency of COPD exacerbations, [52][53][54][55] and impeded decline of lung function 56 in patients taking statins. Th is was supported by the mechanistic study showing that IL-17 acted as a mediator of IL-1 b -induced neutrophilia in infl uenza-induced COPD exacerbations in mice, and inhibition of IL-17 reduced neutrophil recruitment to the airways both in the initial phase of infection and at the peak of viral replication. 57 Th erefore, biologic eff ects of statins in COPD demonstrated in the current study may explain the fi ndings from the observational studies. Data are presented as mean Ϯ SD. ALT 5 alanine aminotransferase; AST 5 aspartate aminotransferase; BUN 5 blood urea nitrogen; CPK 5 creatine phosphokinase; HDL 5 high-density lipoprotein; hsCRP 5 high-sensitivity C-reactive protein; LDL 5 low-density lipoprotein. See Table 1 and 2 legends for expansion of other abbreviations.
Our results were in agreement with other observational studies, but diff ered from the results of a placebocontrolled trial of a statin on COPD exacerbations (STATCOPE [Simvastatin in the Prevention of COPD Exacerbations]). Previous studies showed the benefi cial eff ects of statins on reduced all-cause mortality, 52,[58][59][60] including that from acute exacerbation, 52,61 reduced frequency of COPD exacerbations, especially in patients with COPD who had coexisting cardiovascular disease (CVD), [52][53][54][55]62 and reduced the decline in lung function. 56 In contrast, in patients with moderate to severe COPD, the STATCOPE study found no reduction in exacerbations. 63 A plausible explanation for the discrepancy between the fi ndings of the STATCOPE study and that of the observational studies and our study is the inclusion of a large percentage of patients with COPD with coexisting, overt CVD who were "previous statin users" and "previous statin nonusers, " but would benefi t from statin therapy. 64 Th ese patients are likely to carry a poor prognosis because of one or a combination of the following: undertreated pulmonary infl ammation, unrecognized systemic infl ammation, or subclinical CVD. [65][66][67] Th ese comorbid phenotypes of COPD are strongly associated with an increased risk of hospitalization with acute exacerbations and greater mortality. [65][66][67] Hence, it is possible that many patients with COPD who have not been prescribed statins in observational studies are undertreated, a hypothesis suggested by STATCOPE investigators to explain this discordance. 63 Another explanation for the STATCOPE fi ndings is that statin therapy has no eff ect on reducing acute exacerbations in patients with COPD where coexisting clinical and subclinical CVD has been all but excluded. 64 Th e presence of cardiovascular comorbidity underlying acute exacerbations of COPD, in particular heart failure, 68,69 may intensify the beneficial eff ect of statins. In addition, the selection of patients with COPD recruited to the STATCOPE study might include a group with much lower risk and who are potentially unresponsive to statin treatment. Any benefi t in such a group might be masked . 64 Th is was supported by the 23% reduction in low-density lipoprotein levels in the STATCOPE study, with 40 mg of simvastatin, which was less than the expected 36% to 40% reduction normally observed. 64 Mortality in the placebo arm of the STATCOPE study over 3 years was 6%, only one-half that reported in the Towards a Revolution in COPD Health (TORCH) study over a similar time. 70 Moreover, the prevalence of comorbid cardiovascular-related diseases in the STATCOPE participants was not reported but is presumably very low or nonexistent. 64 Th e exclusion of cardiovascular comorbidity raises questions about the generalization of STATCOPE fi ndings to COPD populations in general, where overt or subclinical CVD may be associated with approximately 75% of all subjects. 64 Since participants in the STATCOPE study were prescreened through medical record data, it is impossible to estimate what proportion of patients with COPD were excluded because of their cardiovascular profi le alone. Th e exclusion would leave only 20% to 30% "lower risk" patients with COPD eligible for the STATCOPE study. 64 An alternative explanation for the discrepancy is that the diff erences among COPD subpopulations under consideration may have masked any benefi cial eff ects of statins in the STATCOPE study. According to the STATCOPE protocol, any patients who met exclusion criteria based on the risk-based eligibility for statin therapy during follow-up could be prescribed simvastatin and continue in the study on an intent-to-treat basis. 63 Th e proportion of these patients was not stated in the published STATCOPE study fi ndings . If it was large, a dilutional eff ect on outcome may have resulted. In addition, patients with COPD in the STATCOPE study were representative of moderate to severe COPD, albeit the use of supplementary oxygen (signifi cantly greater at . 40%) and duration of statin therapy (signifi cantly shorter, with 56% , 2 years) were also substantially diff erent than those in other observational studies.
Despite its novel fi ndings, our study still has potential limitations. Because the present study included only patients with GOLD stage II (moderate) COPD, the results could not necessarily be extended to patients with severe COPD. In addition, because the study had a randomized, cross-over design, prolonged carry-over eff ects cannot be excluded. However, we attempted to minimize this as much as possible by extending the washout period for 4 weeks. Finally, the number of patients in this study was relatively small and, therefore, a larger number of patients with COPD is required for the future studies to confi rm our observations and to study long-term consequences of statin therapy on exacerbations and disease progression. 63