Inhaled corticosteroids in chronic obstructive pulmonary disease: a pro–con perspective

Abstract

Current guidelines limit regular use of inhaled corticosteroids (ICS) to a specific subgroup of patients with chronic obstructive pulmonary disease (COPD) in whom the forced expiratory volume in 1 s is <60% of predicted and who have frequent exacerbations. In these patients, there is evidence that ICS reduce the frequency of exacerbations and improve lung function and quality of life. However, a review of the literature suggests that the evidence available may be interpreted to favour or contradict these observations. It becomes apparent that COPD is a heterogeneous condition. Clinicians therefore need to be aware of the heterogeneity as well as having an understanding of how ICS may be used in the context of the specific subgroups of patients with COPD. This review argues for and against the use of ICS in COPD by providing an in-depth analysis of the currently available evidence.

Introduction

Chronic obstructive pulmonary disease (COPD) is the fourth leading cause of death worldwide; its mortality is rising and it is expected to move to the third leading cause of death in the next 20 years. Globally, ∼10% of people ≥40 years old have airflow limitation of GOLD (Global initiative of chronic Obstructive Lung Disease) stage 2 or worse [forced expiratory volume in 1 s (FEV1) <80% predicted] and up to 25% may have undiagnosed COPD, i.e. ‘the missing millions’ [1]. The major goals in the management of COPD include assessment, monitoring, reducing risk factors, preventing and treating acute exacerbations and managing associated comorbidities. To date, the only measures that reduce mortality are smoking cessation and supplemental oxygen [2,3]. Debate looms over what the ideal treatment for COPD should be. The role of inhaled corticosteroids (ICS) has been viewed favourably by some and sceptically by others. Recent trials on COPD, instead of shining light on the role of ICS, have only added to the controversy [4–6].

In this review, we look at the evidence for the role of ICS in COPD with an aim of providing readers with a balanced view of the effects of ICS on various aspects of COPD, namely airway inflammation, lung function, health-related quality of life (HRQoL), exacerbations, mortality and adverse effects. Full-text papers and abstracts from international respiratory meetings were identified from 1968 to date from Medline and Embase databases. The terms applied for the electronic searches were as follows: COPD/or/inhaled corticosteroids/or/beclomethasone diproprionate (BDP)/or/budesonide/or/budesonide/formoterol (BF)/or/fluticasone proprionate (FP)/or/FP/salmeterol (FPS)/or/inflammation/or/quality of life/or/exacerbations/or/mortality/or/pneumonia/or/diabetes mellitus (DM)/or/osteoporosis/or/glaucoma/or/cataracts/or/tuberculosis (TB)/or/non-tuberculous mycobacterial (NTM) infections. We applied no language or publication status restrictions and reviewed the reference lists of all selected papers, primary studies and review papers for relevant studies. We contacted drug companies for resources as well. All the large relevant randomized, controlled trials (RCTs) using ICS, with their primary and secondary outcomes, are summarized in Table 1. This review analyses and deliberates some important questions and weigh ups the evidence for and against ICS in the management of COPD.

Table 1.

Important randomized controlled trials in COPD assessing a variety of parameters

				Outcomes assessed
Study [citation]	Number of patients	Study intervention/specifics	Duration of study	Primary outcome(s)	Secondary outcomes
Calverley et al. (2007) [TORCH] [4]	6112	Salmeterol, 100 μg day−1 FP, 1000 μg day−1 FPS, 1000/100 μg day−1 Placebo	3 years	Death from any cause for the comparison between the combination regimen and placebo in COPD patients	Frequency of exacerbations, health status, adverse events and spirometric values between groups
Pauwels et al. (1999) [EUROSCOP] [5]	1277	Budesonide, 800 μg day−1 Placebo	3 years	Lung function decline in patients with mild COPD who continued to smoke	Adverse events
Burge et al. (2000) [ISOLDE] [6]	751	SFP, 1000/100 μg day−1 Placebo	3 years	Rate of postbronchodilator lung function decline in moderate-to-severe COPD patients	Exacerbation rates, health status, adverse events and respiratory withdrawals
Sin et al. (2008) [17]	289	FP, 1000 μg day−1 FPS, 1000/100 μg day−1 Placebo	4 weeks	Change in systemic inflammation as measured by C-reactive protein	Change in circulating IL-6 and SP-D, health status and FEV1
Vestbo et al. (1999) [50]	290	Budesonide, 1200 μg day−1 for 6 months, then 800 μg day−1 for 30 months Placebo	3 years	Rate of FEV1 decline in COPD patients	Exacerbations and respiratory symptoms
Calverley et al. (2010) [Lung Health Study] [46]	1116	Triamcinalone acetate, 1200 μg day−1 Placebo	40 months	Rate of postbronchodilator lung function decline in mild-to-severe COPD patients	Respiratory symptoms, use of healthcare services, airway reactivity and bone density measurements
Szafranski et al. (2003) [52]	812	BF, 320/9 μg day−1 Budesonide, 400 μg day−1 Formoterol, 9 μg day−1 Placebo	12 months	Severe exacerbation rates and FEV1 in moderate-to-severe COPD patients	Peak expiratory flow, COPD symptoms, HRQoL, mild exacerbations, use of reliever β2-agonist therapy and safety variables
Calverley et al. (2003) [53]	1022	BF, 320/18 μg day−1 Budesonide, 400 μg day−1 Formoterol, 9 μg day−1 Placebo	12 months	Time to first exacerbation and FEV1 in moderate-to-severe COPD	SGRQ, withdrawals, peak expiratory flow, symptoms, use of reliever medication and adverse events
Calverley et al. (2003) [TRISTAN] [54]	1465	Salmeterol, 100 μg day−1 FP, 1000 μg day−1 FPS, 1000/100 μg day−1 Placebo	12 months	Pretreatment (12 h) and prebronchodilator (6 h) FEV1 in COPD patients	Lung function, symptoms, rescue medication use, number of exacerbations, patient withdrawals and disease-specific health status
Mahler et al. (2002) [55]	691	Salmeterol, 100 μg day−1 FP, 1000 μg day−1 FPS, 1000/100 μg day−1 Placebo	24 weeks	Change in predose and 2 h postdose FEV1 in moderate-to-severe COPD patients	Morning peak expiratory flow, supplementary reliever use, transitional dyspnoea index score, chronic bronchitis symptom questionnaire, exacerbations, chronic respiratory disease questionnaire score and safety variables
Dransfield et al. (2013) [56]	Study 1, 1622; Study 2, 1633	Vilanterol, 25 μg day−1 FF/vilanterol, 50/25 μg day−1 FF/vilanterol, 1000/25 μg day−1 FF/vilanterol, 200/25 μg day−1	1 year replicate studies	Yearly rate of moderate and severe exacerbations in moderate-to-severe COPD patients	Adverse events, time to first moderate or severe exacerbations, yearly rate of exacerbations needing OCS, change in baseline trough FEV1, yearly rate of severe exacerbations, number of night awakenings due to symptoms, rescue reliever usage and dyspnoea scores
Aaron et al. (2007) [OPTIMAL] [60]	449	Tiotropium, 18 μg day−1 and placebo Tiotropium, 18 μg day−1 and salmeterol, 100 μg day−1 Tiotropium, 18 μg day−1 and FPS, 1000/100 μg day−1	1 year	Proportion of patients with moderate or severe COPD who experienced an exacerbation of COPD requiring treatment with OCS or antibiotics	Lung function, disease-specific quality of life and number of hospitalizations for COPD exacerbations and all-cause hospitalizations
Wouters et al. (2005) [COSMIC] [61]	373	FPS, 1000/100 μg day−1 for 3 months, followed by salmeterol, 100 μg day−1 FPS, 1000/100 μg day−1	15 months	Investigation of the duration of the benefits of FP following 1 year withdrawal	Lung function, time to first exacerbation, incidence and severity of exacerbations, symptoms and health status
Sharafkhaneh et al. (2012) [76]	1219	BF, 640/18 μg day−1 BF, 320/18 μg day−1 Formoterol, 9 μg day−1	1 year	Exacerbation rates in patients with severe/very severe COPD	Time to first exacerbation, exacerbations treated with antibiotics, adverse events
Choudhury et al. (2007) [77]	260	FPS, 1000 μg day−1 Placebo	1 year	COPD exacerbation frequency in moderate-to-severe COPD patients	Time to first COPD exacerbation, reported symptoms, peak expiratory flow rate, reliever usage, lung function and HRQoL
Spencer et al. (2001) [78]	751	FPS, 1000 μg day−1 Placebo	3 years	Change in health status as measured by SGRQ and SF-36 scores in severe COPD patients	FEV1 and smoking status
Ferguson et al. (2008) [79]	782	Salmeterol, 100 μg day−1 FPS, 1000/100 μg day−1	1 year	Moderate to severe exacerbations reduction in patients with severe/very severe COPD	Risk of time to first severe exacerbation, annual rate of exacerbations needing OCS and adverse events
Anzueto et al. (2009) [80]	797	Salmeterol, 100 μg day−1 FPS, 1000/100 μg day−1	1 year	Effect on moderate to severe exacerbation in severe/very severe COPD patients	Annual rate of exacerbations needing OCS and hospitalization, reliever use, dyspnoea scores, night-time awakenings and HRQoL scores
Kardos et al. (2007) [88]	994	Salmeterol, 100 μg day−1 FPS, 1000/100 μg day−1	44 weeks	Number of moderate and severe exacerbations in each group of patients with severe/very severe COPD	Time to first exacerbation, prebronchodilator peak expiratory flow, postbronchodilator FEV1 and SGRQ scores
Paggiaro et al. (1998) [94]	281	FP, 1000 μg day−1 Placebo	6 months	Number of patients with at least one exacerbation in mild-to-severe COPD patients	Number and severity of exacerbations, lung function, diary card symptoms, peak expiratory flow and 6 min walk distance
Hanania et al. (2003) [96]	723	Salmeterol, 100 μg day−1 FP, 1000 μg day−1 FPS, 1000/100 μg day−1 Placebo	6 months	Morning trough and 2 h postdose FEV1 in moderate-to-severe COPD patients	Morning peak expiratory flow, transitional dyspnoea index, chronic respiratory questionnaire, chronic bronchitis questionnaire, exacerbations and symptom assessments
Wedzicha et al. (2008) [INSPIRE] [97]	1323	FPS, 1000/100 μg day−1 Tiotropium, 18 μg day−1	2 years	Healthcare utilization exacerbation rate in severe/very severe COPD patients	SGRQ scores, mortality, adverse events and withdrawal from study

Abbreviations are as follows: BF, budesonide/formoterol; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 s; FF, fluticasone furoate; FP, fluticasone proprionate; FPS, fluticasone proprionate/salmeterol; HRQoL, health-related quality of life; IL, interleukin; OCS, oral corticosteroids; SGRQ, St George's Respiratory Questionnaire; SP-D, surfactant protein D.

Inflammation

Inflammation in the lungs has been implicated in the pathogenesis of COPD, with the involvement of numerous cell types [1,7–9]. Parenchymal infiltration by inflammatory cells and mediators is associated with destruction of the normal alveolar architecture and submucosal gland thickening within an intact epithelium [10,11]. These findings have been mirrored by results from induced sputum [9,12,13]. Systemic inflammation is also a feature of COPD, with elevated levels of circulatory inflammatory markers [14–17].

Pro – ICS in inflammation

The key inflammatory cells that mediate inflammation in COPD are CD68⁺ macrophages, neutrophils and CD8⁺ cytotoxic lymphocytes [11,18]. Studies have shown that ICS can reduce the airway and the systemic inflammatory responses in COPD. In a double-blind, placebo-controlled (DBPC) trial in 30 patients with COPD, treatment with FP over 3 months resulted in a reduction of CD8:CD4 ratio, with no reduction in CD8⁺, CD68⁺ cells or neutrophils, suggesting that ICS works on specific aspects of airway inflammation [19]. In another biopsy study, use of FPS and placebo for 13 weeks resulted in reduction in CD8⁺, CD45⁺ and CD4⁺ cells, but no change in CD68⁺ cells [20]. A large Dutch study analysed sputum inflammatory cells and bronchial biopsies from 114 steroid-naive COPD patients receiving 500 μg FP for 6 months. Inhaled corticosteroid therapy decreased counts of mucosal CD3⁺, CD4⁺, CD8⁺ and mast cells in comparison to placebo, with effects maintained after 30 months [21]. Moreover, treatment with ICS resulted in a reduction in bronchoalveolar lavage fluid (BALF) cellularity, lactoferrin, lyzozyme and albumin levels, all of which are markers of inflammation [22]. On a similar note, Ozol et al. demonstrated that ICS treatment resulted in a reduction in interleukin (IL)-8 levels in BALF mean percentage of neutrophils in patients with COPD [23]. Budesonide/formoterol has been shown to attenuate inflammation by significantly reducing sputum eosinophilia in comparison to placebo by a Dutch group [24].

In a meta-analysis assessing the effects of ICS on inflammatory sputum indices, six studies met the inclusion criteria [25]. The authors used cumulative doses, which were calculated on the basis of mean daily dose and duration of therapy. In studies where patients were exposed to ≥60 mg of beclomethasone diproprionate (BDP) or its equivalent, ICS were effective in reducing the sputum total cell, neutrophil and lymphocyte counts. In contrast, trials with a cumulative dose of <60 mg failed to demonstrate such a favourable effect. Of note, trials with the higher cumulative dose had ICS exposure for at least 6 weeks, whereas those with the lower cumulative dose were <6 weeks in duration, suggesting that prolonged therapy with ICS is effective in attenuating airway inflammation in stable COPD. A recent systematic review and meta-analysis looked at the effects of ICS on bronchial biopsies and BALF in COPD patients; ICS reduced CD4⁺ and CD8⁺ cells in the four bronchial biopsy studies, while from five studies using BALF, ICS reduced the neutrophil and lymphocyte counts [26].

Evidence also exists for the role of ICS in attenuating systemic inflammation, a feature of COPD. Withdrawal of ICS (FP) increased baseline C-reactive protein (CRP) levels by 71%, and 2 weeks with inhaled FP reduced CRP levels by 50%. These results echoed prednisone, which reduced CRP by 63% compared with placebo [27]. The effects of FP, with or without a long-acting β₂-agonist (LABA), were examined in a larger population; although there was no reduction in CRP in the FP and FPS groups, there was a significant reduction in SP-D (a lung-specific systemic inflammatory biomarker) [17].

Con – ICS in inflammation

The postulated mechanism of ICS resistance in COPD is the reduction in histone deacetylase-2 activity that is caused by cigarette smoking and oxidative stress [28,29]. A number of studies have assessed the effects of ICS on pulmonary and systemic inflammation, but have failed to demonstrate any anti-inflammatory efficacy. Keatings et al. studied the effects of a 2 week course of ICS (budesonide) and later oral prednisolone on sputum inflammatory indices of differential inflammatory cell counts, IL-8, tumour necrosis factor α, eosinophilic cationic protein, eosinophil peroxidise, myeloperoxidase and human neutrophil lipocalin in severe COPD patients [30]. Neither ICS nor oral prednisolone resulted in any significant changes in inflammatory cells or markers, suggesting a lack of efficacy in the short term and the presence of corticosteroid resistance. In a double-blind crossover study by the same group using FP for 4 weeks, they showed no improvements in lung function, subjective symptom scores or sputum indices of differential cell counts, IL-8, neutrophil elastase activity, matrix metalloproteinase (MMP)-1, MMP-9, secretory leukoprotease inhibitor and tissue inhibitor of metalloproteinase (TIMP)-1 levels [31]. A DBPC crossover trial using mometasone furoate or placebo in COPD subjects on bronchodilator monotherapy reported that there were no treatment-related improvements in lung function, subjective assessments or various sputum characteristics [32]. Recently, a DBPC crossover study in COPD subjects treated for 6 months with budesonide or placebo noted no change in neutrophilic or eosinophilic inflammation [33]. Moreover, a Dutch DBPC bronchoscopic study assessing the effects on reactive oxygen species (ROS) reported no improvements in ROS and differential cell counts at 6 months in BALF [34]. The same group, in a double-blind study of 23 COPD subjects with histamine-induced bronchial hyper-responsiveness receiving FP or placebo for 6 months, showed no effect on bronchial hyper-responsiveness and various inflammatory cells on bronchial biopsies [35]. To evaluate the effect of ICS on COPD inflammation, a Canadian group assessed 60 COPD patients in a 6 month DBPC study using FPS, FP alone or placebo in bronchoscopic biopsies [36]. Although FPS-treated subjects had significant reductions in CD8⁺ cells and macrophages in comparison to placebo, there was no effect on these inflammatory cells with FP monotherapy. Moreover, there were no differences in neutrophil and eosinophil numbers in either treatment group. This would imply and reconfirm that ICS lack effect in pulmonary inflammation.

Systemically, a 12 week DBPC crossover study using BDP in severe COPD subjects conducted by John et al. failed to demonstrate any effect on the release of inflammatory mediators from peripheral blood monocytes [16]. Moreover, in a multicentre DBPC study of 289 COPD subjects, FPS or FP compared with placebo had no effect on serum CRP or IL-6 levels [17].

Summary

In summary, inflammation is a key pathological process in COPD. The evidence can be argued for and against the effects of ICS on the inflammatory processes in COPD [32,36]. The subgroup of COPD patients who have exhibited most benefits from ICS are those with the presence of eosinophilic airway inflammation and frequent exacerbators [37,38]. More recently, the presence of blood eosinophilia during exacerbations has been reported to improve recovery more rapidly with the use of systemic steroids compared those who did not have systemic eosinophilia [39]. There are also limitations of the studies presented relating to the small sample size and possible interactions between sputum processing and assays used for assessment of inflammatory markers. Moreover, the majority of the studies investigated the role of ICS in phenotypically unselected populations. Although larger studies are needed to show the efficacy of improved response in subjects with an eosinophilic phenotype, one may extrapolate from the data available that this phenotype may have an improved response to ICS and/or oral corticosteroids in routine states and during exacerbation. The above data suggest that ICS are not only useful in reducing the airway inflammatory profile in COPD, but also, more importantly, that they have an effect on systemic inflammation. Overall, there is more evidence to support the role of ICS in reducing the inflammatory process of COPD; however, this is not compelling, suggesting that further studies are required before finite conclusions can be drawn regarding the effect of ICS on inflammation in COPD.

Lung function

In healthy subjects, lung function declines with age at a rate of ∼29 ml year⁻¹ [40]. Subjects who smoke and have COPD, as well as their exacerbations, air pollutants and bronchial hyper-responsiveness, have been shown to have a rapid decline in lung function [41]. Spirometry, in particular FEV1, is one of the pillars of establishing a diagnosis of COPD [1]. Over 3 decades ago, Fletcher and Peto described the phenomenon of rapid lung function decline in patients with COPD [42]. Moreover, reports suggest that a brisk lung function decline is independently associated with an enhanced risk of hospitalization and mortality; hence, maintaining or halting further decline would be beneficial [43,44]. Although recently, there is a suggestion that pharmacotherapy improves lung function [41], to date only smoking cessation genuinely reduces the progression of COPD [2].

Pro – ICS and lung function

In the European Respiratory Society study on COPD (EUROSCOP), patients with mild disease showed improvement in FEV1 of 17 ml year⁻¹ in the ICS group, in contrast to a decline of 81 ml year⁻¹ in the placebo group. This was true for the first 6 months; however, from 9 months to the end of the study the decline in FEV1 was similar in both groups [5,45]. Although true for patients with milder disease, the ISOLDE trial investigated moderate-to-severe COPD patients and reported that 500 μg FP twice daily did not affect the rate of FEV1 decline, but produced a small, sustained increase in FEV1 throughout the study period [6]. Recently, BDP/formoterol has been assessed in severe stable COPD, demonstrating similar benefits to BF in predose morning FEV1, but superior to formoterol monotherapy [46].

An earlier meta-analysis found that the estimated 2 year difference in postbronchodilator FEV1 was +0.039 l year⁻¹; the authors concluded that in patients with clearly defined moderately severe COPD, there were improvements in FEV1 with treatment at relatively high daily ICS doses [47]. Other studies have also shown similar findings. In a meta-analysis of eight studies assessing the annual change of FEV1 in subjects with mild-to-moderate COPD, ICS reduced the rate of FEV1 decline by 7.7 ml year⁻¹, whereas analysis of studies with high-dose regimens revealed a greater effect of 9.9 ml year⁻¹. The authors inferred that ≥2 years of ICS treatment attenuates the rate of lung function decline and that high-dose ICS regimens have greater effect than that with all regimens combined [48]. A post hoc analysis of the TORCH study reported that the adjusted rate in FEV1 decline in moderate-to-severe COPD patients was 55 ml year⁻¹ for placebo, 42 ml year⁻¹ for salmeterol, 42 ml year⁻¹ for FP and 39 ml year⁻¹ for FPS [49], confirming that LABAs, ICS and ICS/LABAs can attenuate lung function decline.

Con – ICS and lung function

Unlike asthma, in which the role of ICS is established, in COPD there is a relentless decline in lung function in the presence of inflammation. A number of trials at the turn of the millennium set out to assess the hypothesis to combat inflammation, which may result indirectly in an attenuated decline in lung function [5,6,50,51]. The study by Vestbo and colleagues showed that following 3 years of treatment with budesonide or placebo in mild COPD subjects, there were no significant differences in lung function decline between the treatments [50]. Similar results were reported in the EUROSCOP study with 3 years of use of budesonide compared with placebo [5]. The use of other ICS, i.e. FP [6] and triamcinolone acetonide [51], in moderate-to-severe COPD demonstrated no benefits in lung function deterioration in comparison to placebo, although the use of FP resulted in a small but significant increase in FEV1 that was sustained throughout the study period.

In the last decade, several trials using combination inhalers (FPS and BF) have reported improvements in lung function [4,52–56], although interpretation of the results needs to be done with caution because they have various limitations that need consideration. Firstly, it is challenging to ascertain whether the benefits are due to the addition of ICS to the LABA or the components alone. This conundrum was answered by the post hoc analysis of the TORCH study, which assessed the rate of decline in FEV1 in groups treated with FPS, FP and salmeterol vs. placebo [49]. All three treatment arms resulted in a significantly slower FEV1 decline compared with placebo; however, there was no difference between the treatment arms, raising a question regarding the role of ICS in FEV1 decline. Moreover, a meta-analysis of more than 16 000 COPD subjects concluded that continued ICS use had no benefit in FEV1 decline [57]. Other meta-analyses of ICS vs. LABA [58] and ICS/LABA vs. LABA monotherapy [59] have reported in favour of LABA compared with ICS (20 ml) for the former and trivial, nonclinically significant improvement of 4–6 ml in favour of ICS/LABA compared with LABA monotherapy. Secondly, there is the issue of ICS withdrawal at randomization in subjects in the design of these combination inhaler studies, i.e. the withdrawal of the effect of ICS, as in the OPTIMAL study [60]. This needs careful consideration in data interpretation. The COSMIC study confirmed this, because ICS withdrawal resulted in worsening lung function decline, despite patients being on LABA therapy [61]. Thirdly, the methodology of statistical analyses in these studies is inadequate, due to the absence of a pure intent-to-treat analysis [62–64]. Using the TORCH study [4], for example, there was an incomplete follow-up of FEV1 measurements as opposed to mortality. In fact, under one-fifth of patients in the placebo arm of the study did not contribute a single FEV1 measurement, compared with >10% in the FPS arm. Moreover, patients withdrawing prior to the end of follow-up had a sharper decline than those completing the follow-up. This does not take into consideration the fact that subjects who were excluded even before any follow-up measurement probably had lower FEV1 values at their initial visit. The slope of decline in the remaining subjects with better FEV1 values at the first visit may thus have been exaggerated as a consequence of regression to the mean. This would result in an overestimation of FEV1 decline in the placebo arm and hence a bias that ICS attenuates FEV1 deterioration [63]. Lastly, in a recently published study of two pooled dose-ranging RCTs of a novel combination therapy, fluticasone furoate (FF)/vilanterol, vs. vilanterol, it was reported that there was an overall significant improvement in lung function with the combination therapy [56]. However, when FEV1 was considered individually in each of the two studies there were no statistical differences between the treatment arms.

Summary

In summary, there is evidence to indicate that treatment with ICS in patients with COPD, be it mild or moderate to severe, leads either to an improvement in FEV1 or to a slowing down in the progression of disease by reducing the decline in FEV1 over time; however, the effects are relatively small and clinically irrelevant. Moreover, none of the studies was powered for decline in lung function, and many were powered for effects against placebo. Lung function is an important marker of the disease, and any improvement or slowing of the progressive decline would be beneficial for the patient, not only for disease control but also from the perspective of HRQoL and disease progression. The available literature suggests that more evidence is required before definite conclusions can be drawn.

Health-related quality of life

Patients with COPD suffer a diminished HRQoL [65], which is often related to the severity [66] of the condition and progresses over time [67]. It has been observed that there is a relationship between FEV1 and HRQoL [66,68]; however, they do not necessarily follow a similar pattern of improvement or decline [69]. In fact, HRQoL is more strongly associated with patients' symptoms than is lung function [70]. The HRQoL can be assessed using generic [71] or disease-specific questionnaires [72,73]. The St George's Respiratory Questionnaire (SGRQ) is the most frequently used subjective HRQoL-validated assessment tool.

Pro – ICS and HRQoL

Improved health status, as assessed by SGRQ and other questionnaires, favours the use of ICS in COPD. This has been shown in the ISOLDE and TORCH studies, with more improvement in severe disease [4,6]. The COSMIC study showed that withdrawal of fluticasone in patients with COPD resulted in persistent dyspnoea and an increase in the percentage of disturbed nights, as well as in a decrease in the percentage of rescue medication-free days and worsening of the Clinical COPD Questionnaire score despite the continued use of salmeterol, suggesting a key role for ICS [61]. More recently, BDP/formoterol has been assessed in severe stable COPD and demonstrated improvement in COPD symptoms and HRQoL measures [46].

Recently Ai-Kassimi and Alhamad [74] reviewed 11 studies that reported on changes in HRQoL, 10 of which were ≥12 months in duration. While LABAs were associated with deterioration of HRQoL in two studies and improvement in six other studies, ICS were associated with deterioration in one study and improvement in five others. A recent Cochrane review showed that ICS slowed the rate of decline in quality of life, as measured by the SGRQ [mean difference (MD) −1.22 units year⁻¹, 95% confidence −1.83 to −0.60, 2507 participants] [57].

Con – ICS and HRQoL

A number of trials in COPD patients have reported trivial improvements in HRQoL [4,6,61,74]; however, sustained use of ICS or ICS/LABA has not demonstrated clinically relevant results in COPD patients. Although a reliable and responsive instrument, changes in SGRQ scores need to be interpreted on the basis of clinical relevance, not statistical changes alone. Importantly, despite the minimal clinical improvement in SGRQ score being 4 units, this suggests only minor efficacy, with 8 and 12 units implying moderate and marked efficacy, respectively [75]. Hence, any changes of <4 units should be rejected even if statistically different. With this in mind, only one study in the last decade has demonstrated a clinically relevant SGRQ score improvement in comparison to placebo [53]. This may have been related to ICS withdrawal resulting in marked deterioration in the placebo group and the fact that the scores were not a change from baseline. Additionally, there were no clinical significant improvements in the SGRQ scores with ICS monotherapy compared with placebo, suggesting that the observed HRQoL benefit may be due to formoterol. More recently, similar findings have been reported by Sharafkhaneh and colleagues [76]. Another consideration is that improvements in the SGRQ scores were compared with baseline and were not clinically relevant. This was noted in the WISP [77] and TRISTAN studies [54], although in the former there was no difference between FP and placebo, whilst in the latter FP alone resulted in no significant change in the SGRQ score. In the ISOLDE study, despite the reduction in the rate of SGRQ score decline in the treatment arms, at 36 months the scores were worse than at baseline [78]. Lastly, to date no COPD trials have had HRQoL as a primary outcome assessment.

Recently, two clinical trials comparing FPS vs. salmeterol have reported statistically significant worsening in SGRQ scores with the use of combination therapy compared with LABA [79,80]. Moreover, a meta-analysis of over 8000 patients reported that combination therapy was associated with a statistical improvement in the SGRQ and dyspnoea scores; however, these were not clinically relevant [59].

Summary

In summary, COPD is a chronic and debilitating disorder that has a negative effect on patients' HRQoL; therefore, many therapeutic interventions have been assessed from the perspective of their effects on HRQoL. In this context, ICS have been shown to have little effect on HRQoL and in slowing down the decline in quality of life. These improvements in HRQoL may be greater when ICS are combined with LABA, especially in those patients with frequent exacerbations.

Exacerbations

Exacerbations of COPD are defined as acute episodes of worsening respiratory symptoms that require a change in medication, or hospitalization if severe [1]. The severity and frequency of exacerbations determine the course and burden of the condition to a patient. Exacerbations may be precipitated by respiratory infections (bacterial or viral), pollution, non-adherence to regular therapy or, in one-third of cases, may be due to an unknown trigger [1,37]. Approximately one-half of all exacerbations remain untreated or unreported [81]. Moreover, the frequency of these exacerbations has been associated with a number of aspects, including eosinophilic airway inflammation, frequency of viral infections, hypersecretion of mucus and lower airway bacterial colonization [37]. The consequences of COPD exacerbations affect not only subjective and objective measures, but also HRQoL, all of which take time to recover [82,83]. It is well known that exacerbations lead to a rapid decline in lung function and increased mortality, especially if severe [84–87]; hence, avoidance of exacerbations or decreasing their severity is a vital measure in preventing this vicious cycle. Current guidelines suggest the use of ICS/LABA for patients with severe/very severe COPD and recurrent exacerbations [1].

Pro – ICS in exacerbations

The ISOLDE study reported that FP reduced exacerbation rates by 25%, and a subgroup analysis revealed that this was predominantly in patients with more severe disease (FEV1 <50%) [6]. In the TRISTAN study, FPS, FP and salmeterol reduced exacerbation rates in comparison to placebo [54]. In the TORCH study (n = 6112 subjects), the annual exacerbation rate was 0.85 for FPS compared to 1.13 for placebo [4]. There was a 25% reduction in the exacerbation rate ratio favouring the active treatment arm. A key question is whether the addition of LABA reduces COPD exacerbations. In a study comparing FPS and salmeterol monotherapy in moderate-to-severe exacerbations, there were 334 exacerbations in the FPS arm compared with 464 in the salmeterol arm (P < 0.0001). Hence, the annualized rate of moderate and severe exacerbations per patient was 0.92 and 1.4, respectively, corresponding to a 35% decrease in favour of FPS. Additionally, the mean time to first exacerbation with FPS was significantly longer than with salmeterol (128 vs. 93 days, respectively; P < 0.0001) [88].

In a meta-analysis of nine randomized trials (3976 COPD patients), the use of ICS therapy reduced exacerbation rates (relative risk = 0.70), providing further evidence for the benefits ICS in reduction of exacerbation [89]. In a Cochrane meta-analysis involving 7598 severe COPD subjects from 10 trials (FPS and BF), the exacerbation rates with combination inhalers were reduced in comparison to LABAs alone; however, there was no difference in mortality [90]. Lastly, the most convincing evidence from a recent meta-analysis by Yang and colleagues reported that long-term use of ICS significantly reduced the mean rate of exacerbations in studies where pooling of data was possible (seven studies), irrespective of the means of analysis [57]. Pooling of data from the long-term studies showed a statistically significant benefit of ICS in reducing the mean rate of exacerbations (generic inverse variance analysis: MD −0.26 exacerbations per patient per year; P < 0.0001).

Con – ICS in exacerbations

Previous studies (five budesonide and one FPS) have concluded that ICS, at various doses in a variety of COPD severities over a range of 6–36 months, compared with placebo resulted in no significant reduction in exacerbations [50,91–94]. This was reconfirmed by a meta-analysis of the largest three studies [47]. In another meta-analysis of the earlier negative studies and two positive studies (LHS [51] and ISOLDE [6]), the authors unexpectedly reported that ICS reduced exacerbations by 30% [89]; this was highly criticised for the lack of unweighted statistical approach and marked dropouts from the ISOLDE study that contributed to the bias [95].

In the last decade, most of the studies conducted have used combination inhalers, with comparison to placebo and individual components, and were generally larger. Mahler et al. were the first to report that there were no statistical differences in time to exacerbations between FPS, FP, salmeterol and placebo [55]. Subsequently, in a 12 month DBPC four-arm study of BF, budesonide, formoterol and placebo that was appropriately weighted to avoid bias, no statistically significant difference was found in the severe exacerbation rate ratios between budesonide and placebo [52]. However, BF was significantly superior to both placebo and formoterol, with an annual severe exacerbation rate reduction of 24 and 23%, respectively. Moreover, the TRISTAN study [54] and another by Hanania et al. [96] using FPS, FP, salmeterol and placebo showed that there were no significant reductions in exacerbation rates, numbers or time to first exacerbation between the treatment groups. The COSMIC study likewise found no benefit of ICS treatment on moderate or severe exacerbations [61]. In the largest COPD study of its time, the TORCH study, whilst FPS and salmeterol significantly decreased the annual exacerbation rate in comparison to placebo, there were no differences between them, suggesting similar efficacy of FPS and salmeterol [4]. The OPTIMAL study, in which the primary end-point was exacerbations requiring oral corticosteroids (OCS) or antibiotics, assessed FPS/tiotropium, salmeterol/tiotropium and placebo/tiotropium in moderate-to-severe COPD patients [60]. It was observed that the addition of FPS to tiotropium therapy did not significantly influence the rates of COPD exacerbations. In a head-to-head study of tiotropium vs. FPS, the INSPIRE study, a similar lack of superiority of FPS over tiotropium was noted [97]. More recently, BDP/formoterol has been assessed in severe stable COPD and demonstrated similar benefits to BF and formoterol monotherapy [46]. To improve compliance, there is a drive to develop therapies that need to be administered once daily. Recently, a once daily combination inhaler containing FF/vilanterol has been assessed in comparison to vilanterol. Although the times to first moderate and severe exacerbations favoured FF/vilanterol, there were no differences in severe exacerbation rates between vilanterol monotherapy and the various doses of the combination therapy [56]. Contentiously, two studies using ICS reported marginally higher exacerbation rates in comparison to placebo; however, this was not significant [98,99].

Recently, two systematic reviews have been published. The first reviewed 18 RCTs with over 12 000 subjects and concluded that combination therapy had no effect on severe COPD exacerbations and was not superior to LABA monotherapy [59]; and the second review of 11 studies with more than 8000 subjects suggested that there was no benefit of ICS in preventing exacerbations, which was independent of the severity of COPD as assessed by FEV1 [100].

Summary

In summary, reduction of exacerbation forms an important strategy in the management of COPD, and there is good evidence that patients who experience frequent exacerbations have a rapid decline in lung function [84]. This is the situation in which ICS have been shown to be of particular benefit, with studies suggesting an average reduction in exacerbations by one-quarter [6,47,100]. The results from the meta-analysis by Yang and colleagues suggest a reduction of one exacerbation every 4 years for a patient with COPD. Although the magnitude of the effect is relatively small, this could be important in the context that exacerbations worsen lung function. The most significant reduction in exacerbations has been reported when ICS were combined with LABA [4,52,53]. The studies showing that ICS reduce exacerbations in COPD have some limitations that need to be reflected upon. Firstly, as mentioned above, there is the issue of ICS withdrawal at randomization in subjects. This can be noted in a number of studies. On re-analysis of exacerbation data in the OPTIMAL study [60], Suissa et al. established that the effect of ICS on the likelihood of the first severe exacerbation was markedly protective in subjects previously on ICS who had to discontinue ICS treatment [62]; however, this was not the case between treatment arms in ICS-naive subjects. Secondly, in combination treatment trials, there have been no significant differences between the ICS/LABA and LABA or LAMA arms, suggesting that the beneficial effect over placebo may be related to each of these medications. Lastly, there is no consensus in the multitude of studies on the definition or assessment of an exacerbation. This not only makes studies challenging to compare but may also lead to bias in the estimation of treatment effects [101]. Hence, despite the positive role of ICS in exacerbation reduction, they need to be used in the appropriate patients.

Mortality

Chronic obstructive pulmonary disease accounts for 5% of all-cause mortality worldwide and is expected to increase by >30% in the next decade. Although two COPD mega-trials reported a mortality rate of around 15% in moderate-to-severe subjects [4,102], longer follow-up observations of >13 years in subjects of similar severity have shown a staggeringly high rate of 56% [103]. Moreover, patients with GOLD stages I and II had nearly double the mortality of patients without COPD [104]. Thus far, improved survival has been reported only with smoking cessation [2] and supplementary oxygen therapy in severe hypoxaemia [3,105]. There are emerging considerations that pharmacotherapy may improve some aspects of COPD [41]; however, to date there is lack of evidence for this. Hence, more trials with a primary end-point of mortality are required to establish the efficacy of pharmacotherapy in COPD.

Pro – ICS and mortality

Chronic obstructive pulmonary disease studies using ICS have reported significant improvements in exacerbation, inflammation and HRQoL, but none has shown an effect on mortality, although some have suggested trends in mortality reduction. Soriano et al. examined the 3 year survival of 1045 FPS-treated COPD patients compared with 3620 COPD patients on bronchodilator monotherapy from a UK general practice (GP) database. They observed a reduction in mortality with increased prescriptions of FP and/or salmeterol, supporting a role for ICS in attenuating COPD mortality [106]. The same investigators, in another GP database study, showed that death within a year occurred in 24.3% of the reference COPD patients, in 17.3% with LABAs monotherapy, 17.1% with ICS monotherapy and 10.5% with use of ICS/LABA. In multivariate analyses, the risk of rehospitalization or death was reduced by 10% in LABA only (not significant), by 16% in ICS only (P < 0.05) and by 41% in ICS/LABA users (P < 0.05) [107]. Although the TORCH study did not show a significant reduction in mortality with FPS in comparison to placebo, there a marked reduction in mortality of 17.5% in favour of combination therapy. In post hoc analysis of TORCH and in the INSPIRE study, FPS has been shown to decrease mortality in comparison to placebo in GOLD stage II patients and in those treated with tiotropium [97,108].

In a pooled analysis of seven studies, Sin and colleagues examined mortality as the primary end-point, with an overall mortality of 4% at a mean follow-up of 26 months (adjusted hazard ratio of 0.73). This systematic review had the methodological strength of access to individual patient data, in order to adjust for age, sex, baseline lung function, smoking status and body mass index. Despite considerable overlap between subgroups in effect sizes, the beneficial effect was pronounced in women and former smokers. The authors concluded that ICS reduced all-cause mortality in COPD patients [109].

Con – ICS and mortality

In a time-dependent analysis of mortality in over 8000 patients, of whom about one-third were on ICS, irrespective of the ICS dose there were no significant mortality benefits in favour of ICS therapy [110]. Recently, a meta-analysis of >14 000 COPD subjects comparing usage of ICS and non-ICS therapies reported a lack of difference in all-cause mortality between the two groups at 1 year [111]. Similar absence of efficacy of ICS vs. placebo and LABAs on mortality has been reported in Cochrane reviews [57,58]. Observational and meta-analyses have suggested survival benefits with ICS use [107,109,112]; however, the former are highly predisposed to immortal-time bias [113–115]. This is noted by the discrepancy in all-cause mortality reduction in one meta-analysis of 27% [109], >30% reduction in an observational study [116] but a lack of any benefit in the TORCH trial [4]. Besides immortal-time bias, such disagreement, as aforementioned, may be due to poor follow-up data as a result of improper study design and intention-to-treat analyses [62–64]. In the TORCH study, when analysed using factorial methodology to estimate the effect of each drug and adjusting for the other, it was observed that the mortality effect of FPS was entirely due to the salmeterol component and was highly significant [117]. This was not the case with the fluticasone component. Furthermore, in order to establish a synergy between ICS and LABA in the TORCH study, it was reported that the use of an interaction term was insignificant, implying that the combination was no more efficacious than its components [62].

Besides COPD-related and all-cause mortality, ICS have been implicated in beneficial effects on cardiovascular morbidity and mortality [118–120]. However, in a well-designed meta-analysis of 23 RCTs, it was noted that ICS use had no added effect on myocardial infarction, cardiovascular deaths or all-cause mortality [121]. They also observed positive effects in favour of ICS in the observational studies assessed, causing a dilemma regarding the true influence of ICS on cardiovascular morbidity and mortality as well as all-cause mortality. Rodrigo et al. in their meta-analysis of 18 RCTs also reported that the ICS/LABA combination compared with LABA monotherapy had no advantage in all-cause, respiratory and cardiovascular mortality [59].

Summary

In summary, the issues related to the effects of ICS on mortality in COPD are not clear. The studies with ICS on COPD mortality show a positive trend, but this is not significant. Furthermore, the observations are confounded by the coexistence of comorbidities, such as ischaemic heart diseases, cardiac failure, diabetes mellitus, stroke or bronchiectasis [122–125]. Chronic obstructive pulmonary disease is also associated with an increased prevalence of anxiety and depression, which may affect patients' symptoms and ability to cope with their disease [126,127]. Therefore, further evidence would be required to assess whether ICS have any effects on mortality in patients with COPD.

Inhaled corticosteroids and adverse events

On prescribing any drug, the physician needs to be aware of the risk–benefit balance of taking the drug to improve the condition vs. its adverse effects. Hypothetically, inhaled drugs can manifest adverse events locally (orophyrangeal area) or systemically via the pulmonary vasculature or gut absorption. The use of ICS as monotherapy or in combination with LABAs is increasing, and commonly reported side-effects of ICS include hoarseness, candidiasis, upper respiratory tract infection, sore throat and cough [128]. The incidence of these side-effects ranges from 1 to 10%, and they occur in a minority of patients without major sequelae [129]. Of note, the licensed doses of ICS are in the upper limits of the dosage scale [4,52,53,97] and a large proportion of patients initiated on treatment are elderly; thus they may be more prone to ICS therapy-related adverse events. Hence, it is pivotal that we consider using the lowest possible dose to achieve the optimal symptom management.

Pro – ICS and adverse events

There are reports of a significant increase in the incidence of pneumonia in patients treated with ICS. Unfortunately, most of the trials were not designed to assess this risk, and it is challenging to ascertain whether radiographic confirmation was available in those patients believed to have had pneumonia. The diagnosis of pneumonia was based on clinical symptoms and signs, which are unreliable, and as the clinical presentation of an exacerbation of COPD and pneumonia may overlap, in the absence of radiological abnormality an adverse event could potentially be misclassified as pneumonia and vice versa. Another confounding factor would be the use of antibiotics in clinical trials in the treatment and control arms. Furthermore, there is also the question of whether this is a class effect. While numerous studies found an increased association between FP use and pneumonia, the safety data of inhaled budesonide is heterogeneous and limited to smaller numbers of RCTs [130]. An earlier study by Vestbo and colleagues showed that pneumonia was a feature in 16 budesonide-treated patients (n = 145) compared with 24 patients on placebo (n = 145); 34 patients in each group had at least one reported viral infection [50]. Sin et al. assessed pooled patient data from seven trials and found that budesonide treatment for 12 months failed to increase the risk of pneumonia during the study period [131]. Recently, a retrospective study reported that compared with taking BF, the rates of pneumonia and COPD-related hospitalizations were higher in FPS-treated patients (1.73 and 1.74, respectively). Although the mean durations of pneumonia-related admissions were similar, mortality related to pneumonia was higher in FPS-treated (97 deaths) than in BF-treated subjects (52 deaths) [132]. Recently, a nested-control insurance database Canadian study has also reported higher rates of pneumonia with FPS compared with BF [133]. This raises the possibility that molecular characteristics may influence the manifestation of pneumonia. Despite this increased risk of pneumonia, pneumonia-related mortality in trials is not increased, although observational studies do report an increase in pneumonia-related fatalities [134,135].

Oral corticosteroids are associated with an increased incidence of diabetese mellitus; however, the data from ICS are fragmented. A large case–control study in patients with asthma and COPD has reported that ICS use was associated with a 34% increased risk of both new-onset diabetes mellitus and progression of diabetes mellitus [136]. A retrospective analysis from Canada evaluated all DBPC trials in patients ≥4 years of age involving budesonide or BF in asthma (26 trials; budesonide, n = 9067; placebo, n = 5926) and in COPD (eight trials; budesonide, n = 4616; non-ICS, n = 3643). In the asthma data set, the occurrence of diabetes mellitus/hyperglycaemia was 0.13% for both budesonide and placebo, and serious adverse events were 0% for budesonide and 0.05% for placebo; whilst in the COPD data set, the occurrence of diabetese mellitus/hyperglycaemia adverse events was 1.3% for budesonide and 1.2% for non-ICS, and the occurrence of serious adverse events was 0.1% for budesonide and 0.03% for non-ICS. The authors concluded that treatment with ICS in asthma or COPD patients was not associated with increased risk of new-onset diabetes mellitus or hyperglycaemia [137].

The evidence for osteoporosis as an adverse event of ICS is mixed. A Cochrane database review concluded that in patients with asthma or mild COPD, there is no evidence for an effect of ICS at conventional doses given for 2–3 years on bone mineral density or vertebral fracture. Higher doses were associated with biochemical markers of increased bone turnover, but data on bone mineral density and fractures at these doses were not available [138]. Similar results were observed in other clinical trials that were not included in the Cochrane review [5,139].

The risk of mycobacterial disease is relevant in countries with a high incidence. There is evidence for a propensity to develop tuberculosis in patients who are immunosupressed with OCS. However, a nested case–control Canadian database study showed that exposure to ICS is not associated with risk of tuberculosis in the presence of OCS [140].

A systematic review and meta-analysis observed that the adverse effects of ICS in patients with COPD were generally tolerable, and pooled discontinuation rates were not different between the ICS and placebo groups [141].

Con – ICS and adverse events

Irrespective of brand, ICS can cause dose-dependent undesirable systemic adverse events. In fact, it has been proposed that the systemic activity of 1000 μg of FP is equivalent to 10 mg of prednisolone [142], prompting us to be more cautious in the use of ICS in older COPD patients and those who have comorbidities.

A 5 year population-based case–control observational study eluded to the finding that ICS use may be associated with a 70% increased risk of hospitalization for pneumonia and a 53% subsequent 30 day mortality [134]. Furthermore, the risk of pneumonia was dose dependent. This was reinforced by adverse event reports from large RCTs [4,88,97]. A post hoc analysis of the TORCH study not only reported a significantly higher rate of pneumonia with FP (monotherapy or combination) compared with placebo and salmeterol, but also highlighted risk factors associated with the development of pneumonia [135]. Despite there being no increase in the number of deaths observed with FPS use, this could not be concluded for FP. Furthermore, the INSPIRE study reported a near doubling of pneumonia incidence with FPS use over tiotropium [97]. More recently, a UK-based prospective observational COPD study compared patients on ICS and non-ICS therapies [143]. They observed that prior ICS use did not influence the outcome in patients admitted with community-acquired pneumonia at 30 day and 6 month mortality or the need for inotropic support and/or mechanical ventilation. Furthermore, once-daily administration of FF/vilanterol was reported in a pooled RCT to result in an increased frequency of pneumonia and, at the higher dose, more cases of fatal pneumonia [56]. There have been two published meta-analyses [111,144] and Cochrane reviews [57,145] concluding that ICS therapy in COPD is associated with an increased risk of pneumonia. In fact, the two Cochrane reviews comparing ICS vs. placebo and ICS/LABA vs. LABA reported similarly increased odds ratios in the ICS- and ICS/LABA-treated groups [57,145]. More recently, a retrospective Swedish analysis has suggested differences in rates of pneumonia, pneumonia-related hospitalizations and deaths between BF and FPS use [132].

The use of corticosteroids is understandably associated with much trepidation in both patients and physicians. This is especially the case when subjects have underlying osteoporosis or in COPD patients in whom osteoporosis is associated [146]. Although there are no RCTs with evidence of ICS causing osteoporosis and related fractures, there is support of this from observational and case–control studies. In a UK GP database case–control study of >16 000 hip fractures and just <30 000 controls, it was noted that there was an odds ratio of 1.26 of having hip fractures in ICS-exposed patients compared with those who were not ICS exposed [147]. Additionally, they observed that there was a dose-related association between ICS and hip fractures despite correcting for the OCS doses. Similar results were noted between ICS use and nonvertebral fractures in a US cohort of COPD patients [148]. In a recent meta-analysis of 16 RCTs and seven observational studies, it was reported that ICS use was associated with an increased risk of fractures [149]. Dransfield et al. conducted a pooled analysis of two RCTs, which reported an increase in nontraumatic fractures in all FF/vilanterol arms in a dose-dependent manner compared with vilanterol alone [56]. Based on this limited information, there seems to be a potential risk of osteoporosis and fractures with long-term ICS use.

Akin to osteoporosis and fractures, there is no RCT-based evidence of increased incidence of cataracts and glaucoma. However, considering the ocular side-effects of OCS and the older age group of COPD subjects there may be a causal relationship with ICS use. Currently available data are from observational studies and a meta-analysis, which we discuss here. A UK GP database case–control study noted that there was an increased association of ICS use and cataracts, which increased dramatically with high-dose ICS therapy [150]. Similar results have been reported in earlier studies [151–153]. A recent meta-analysis of four case–control studies not only confirmed the findings of the observational studies of an increased association between cataracts and ICS use, but also found that there was a 25% increased risk for every 1000 μg increase in BDP-equivalent dose [154]. Moreover, two well-designed observational population-based case–control studies, although not specifically in COPD subjects, have observed an increased risk of intraocular hypertension or open-angle glaucoma with the use of ICS [155,156].

Despite RCTs reporting the lack of association between ICS use and the onset and progression of diabetes mellitus in COPD subjects [5,51], observational studies, although not in COPD-specific subjects, suggest the contrary [136,157]. In a large population-based observational study with 5.5 years follow-up, ICS use was associated with a 34% increase in the risk of developing diabetes mellitus and rate of diabetes progression. Moreover, the chances of acquiring diabetes mellitus (64%) and progressing to insulin use from oral hypoglycaemics (54%) were greatest with high daily ICS doses of 1000 μg FPS or more [136]. Also, ICS use in diabetic patients is associated with a dose-dependent increase in serum glucose concentration [157].

Theoretically, akin to prolonged OCS use, ICS may decrease localized immunity of the lung, raising concerns regarding mycobacterial infections. Two nested-control database studies, although not specifically in COPD patients, reported that the use of ICS and the risk of tuberculosis was ≥20% [140,158]. Moreover, the risk of developing tuberculosis was dose dependent. Controversially, in patients using both ICS and OCS the risk the tuberculosis was nonsignificant compared with controls. Likewise, the risk of nontuberculous mycobacterial disease is also increased in a dose-dependent manner in COPD patients who use ICS compared with those who do not [159]. Analogous to the pneumonia outcomes, there seems to be disparate difference in the signal of increased nontuberculous mycobacterial disease depending on the ICS used; hence, keeping a high index of suspicion and monitoring for the development of mycobacterial disease with ICS use is essential.

Summary

In summary, although there have not been any specific studies conducted to elucidate the rate of adverse events in patients using ICS, it seems likely from the evidence of secondary outcomes measures, adverse events reporting and observational studies that the use of ICS in COPD is associated with adverse events. This is especially the case with pneumonia, which could be a class effect, but there are more reports, albeit small numbers, of the other above-mentioned adverse events.

Conclusion

Current guidelines on management of asthma recommend the use of ICS in all but patients with very mild disease. In contrast, the regular use of ICS should be limited to a specific subgroup of COPD patients, in whom FEV1 is <60% of predicted and who have frequent exacerbations [1]. In these patients, there is evidence that ICS reduce the frequency of exacerbations, improve lung function and reduce the decline in HRQoL.

There is a significant burden of COPD, with a global prevalence of 9–10%, with an estimated 210 million people affected [1], and the annual costs per COPD patient on prescription medications are estimated at approximately $2359 and that for medical spending at $21 488 [160–162]. The highest costs to health systems come from patients with frequent exacerbations. Recent studies have shown that various strategies of using ICS/LABA or LAMAs and ICS/LABA were cost effective, with significant improvement in quality-adjusted life years, especially in patients with COPD and frequent exacerbations, mainly through exacerbation reduction [163–165]. In addition, alternative modalities of treating COPD patients may be of relevance in a selected few, including lung volume reduction surgery and bronchoscopic lung volume reduction procedures, such as endobronchial valves and coils and bronchial thermal vapour ablation [166].

The present pros and cons discussion with regard to the use of ICS in COPD has raised a number of important issues. The effects of ICS on inflammation and mortality would require further evidence. In contrast, there is now a large body of evidence to suggest that ICS reduce exacerbation frequency, and this is particularly relevant in patients with moderate-to-severe COPD. Moreover, there is evidence to suggest that ICS may reduce the rate of decline in lung function, although a critical review of the available data would suggest that the combination of ICS/LABA may have greater effects in comparison to ICS monotherapy. Likewise, the effects of ICS on improvement of HRQoL are more noticeable when ICS are used in combination with LABA or LAMA. However, it is clear that ICS may be associated with potential adverse events, both localized and systemic. Possible adverse events such as increased risk of diabetes mellitus, ocular issues or osteoporosis may have weaker evidence and require further evaluation in future studies. The evidence does point to a higher incidence of pneumonia in patients with COPD on ICS, and there are some data to suggest that this is likely to be a class effect, with a higher incidence being noted with fluticasone.

The main role of physicians managing patients with COPD is to provide the most appropriate combination of treatments for each individual patient. This would require a careful assessment of potential benefits against adverse events. In the context of ICS prescribing, patients require specific characteristics related to their FEV1 and frequency of exacerbations. In addition, in the majority of patients ICS would be better used in combination with a LABA and/or LAMA. Therefore, the issues related to the use of ICS in COPD are complex. Moreover, the role of ICS in COPD should be analysed in the context of other available therapeutic modalities and the phenotypic characteristics of individual patients.

Competing Interests

KSB has received speaker fees and sponsorship to attend meetings from Cheisi, AstraZeneca and GSK. JAK has no conflict of interest to declare. JBM has received speaker fees and sponsorship to attend the ERS, ATS and BTS Conferences, educational grants and is on advisory boards from Novartis, Boehringer Ingelheim, GSK, Chiesi, MSD, Pfizer, Almirall, Teva and Napp. There are no other relationships or activities that could appear to have influenced the submitted work.