Chronic obstructive pulmonary disease (COPD) is arguably the most common chronic disease of the lungs at present and, by 2020, it will be the third leading cause of death worldwide.¹ COPD is associated with a relentless decline in forced expiratory volume in 1 second (FEV₁) and, in the later stages, the condition progresses to pulmonary hypertension and hypoxic respiratory failure.

Few interventions have been shown to affect the outcome of COPD. The Lung Health Study-1 (LHS-1) showed that smoking cessation decreases the accelerated decline in FEV₁ characteristic of this disease² and, more recently, that smoking cessation is associated with decreases in cardiovascular and lung cancer mortality in patients with COPD.³ Furthermore, two randomised controlled studies of long term oxygen therapy (LTOT) showed that LTOT improves mortality in patients with COPD complicated by hypoxic respiratory failure.^4,5

Because so few interventions have been shown to affect mortality and FEV₁ decline in COPD, two further outcomes have been studied arising from the observed close association between exacerbation frequency and health related quality of life.⁶ In the ISOLDE study (Inhaled corticosteroids in Obstructive Lung Disease in Europe), inhaled corticosteroids were shown to slow the decline in health status over time and also to decrease exacerbation frequency.⁷ Further statistical modelling has shown that the effect of inhaled corticosteroids on quality of life is largely due to their effect on exacerbation frequency.⁸ Recent studies have closed the loop between exacerbation frequency, mortality, and lung function decline by showing that patients with a history of frequent exacerbations have an accelerated decline in FEV₁^9,10 and increased mortality from COPD.¹¹ From these data one would therefore expect that interventions that reduce COPD exacerbations would also reduce COPD mortality.

In this issue of Thorax we publish a study by Sin and colleagues who show, for the first time, that inhaled corticosteroids are associated with a reduction in all-cause mortality in patients with COPD.¹² This study is important for several reasons. Firstly, it represents a major collaboration between academia and industry that allowed a pooled analysis of the original data from all seven longer term randomised controlled studies of the effects of inhaled corticosteroids in COPD.^7,13–18 This is a major strength of this study. Secondly, it is one of the few studies to report an intervention that affects all-cause mortality in COPD. Thirdly, this study shows that COPD mortality may be affected before developing end stage disease in COPD. A previous systematic review found no significant effect of inhaled corticosteroids on mortality but confirmed the beneficial effects of inhaled corticosteroids on exacerbation frequency.¹⁹ A recent meta-analysis of eight studies of inhaled corticosteroids (four of which are included in the current report) showed that inhaled corticosteroids decreased the decline in FEV₁,²⁰ perhaps through their effect on exacerbations. Thus, before the report of Sin and colleagues, the major effect of inhaled corticosteroids in COPD was considered to be on reduction in exacerbations with no clear effect on mortality.

The authors obtained the original anonymised data from each of the seven randomised placebo controlled trials of inhaled corticosteroids of at least 12 months duration.^7,13–18 The data were then pooled to form a dataset of 5085 patients with allowance for source effect. The study design is unique, and the authors should be commended for their contribution and effort in this regard. Sin and colleagues showed that inhaled corticosteroids reduced all-cause mortality by 27%, and the effects were more pronounced in women, former smokers, and patients with moderate or severe disease. Sex differences in the natural history of COPD have been recognised for some time, with women showing a faster decline in FEV₁.²¹ However, as Sin and colleagues point out in the discussion to the paper, women with COPD have greater bronchial hyperreactivity than men²² and this factor may be associated with the observation that the effects of inhaled corticosteroids on mortality are more pronounced in women. Further studies are required to confirm these sex differences and to investigate the mechanisms associated with these observations.

As with all post hoc analyses of this type, there are limitations. Firstly, only two of the studies (ISOLDE and LHS-2) included mortality on an intention to treat basis, while in the other five studies mortality data were only available for those who completed the study.^7,13 These two studies contributed 37% of the patients to the analysis and were also the two longest studies (34 and 42 months, respectively). The effect of ISOLDE and LHS-2 are therefore likely to influence the results more than the other included studies. The contribution of each study to the overall death rate is not shown in the report and the authors refer to this possible source of bias in their discussion. There are also differences in the pre-intervention treatment of patients between the different studies. For example, in the EUROSCOP study¹⁵ there was a 6 month run-in with no treatment but the ISOLDE study and the study by Calverley and colleagues^7,17 included a run-in period during which patients were treated with oral corticosteroids. Furthermore, the size of the included studies varied because they were each powered to detect different parameters. For example, four of the seven studies were powered to detect a difference in the decline in FEV₁ of 12–20 ml/year between inhaled corticosteroids and placebo controlled groups^7,13–15 and all failed to detect a difference, but the meta-analysis by Sutherland et al²⁰ showed that this difference was in fact much smaller—of the order of 7.7 ml/year. Finally, in all the included studies the difficulty in finding an effect on mortality hitherto has been because of the effect of dropouts. In a further analysis from the ISOLDE study, Calverley and colleagues showed that dropouts were greater in the placebo arm and were associated with a faster decline in FEV₁, thus implying that mortality was probably greater in the placebo arm,²³ which has now been supported by Sin et al.¹²

Currently, for patients with COPD, it may no longer be relevant to ask whether inhaled corticosteroids have a role, but the question still remains—which COPD patients should be given inhaled corticosteroids? The ISOLDE study showed that the effects of inhaled corticosteroids on exacerbation frequency were greatest for patients with an FEV₁ of <50% predicted.⁷ But the LHS-2 found a 50% decrease in unscheduled physician visits for respiratory illnesses (interpreted as exacerbations) in patients with much milder COPD.¹³ A further analysis from the ISOLDE study showed that, when patients with an exacerbation frequency of at least one per year are studied, the effects of inhaled corticosteroids on the reduction of exacerbations are also present in patients with milder COPD.⁸ The study by Sin et al found that the effect of inhaled corticosteroids on COPD mortality was statistically significant in patients with FEV₁ <60% but not in those with a higher FEV₁, although we must be careful about subset analysis in a pooled dataset such as this—and the authors have cautioned us appropriately in the paper.¹² However, one of the further important questions that arises from this study and others is the actual dose of inhaled corticosteroids that should be prescribed in COPD and the long term side effects of this dosage.

Inhaled corticosteroids have similar effect sizes on FEV₁ decline in COPD and on exacerbations (7.7 ml/year is about 25% of the expected FEV₁ decline in normal patients, ISOLDE found a 25% effect on reduction in exacerbation frequency). The effect of inhaled corticosteroids on mortality in patients with COPD may therefore be through decreased exacerbation frequency.

Intriguingly, Sin and colleagues in their meta-analysis found a trend towards a lower incidence of cancer mortality in the treatment arm; however, cancer deaths in this study were low and the study was not powered to look at specific causes of death. A further larger study is therefore required before clear conclusions can be drawn about the effects of inhaled corticosteroids on specific causes of mortality.

Twelve years ago, one of us writing an editorial in this journal made the point that, despite much work on the role of inhaled corticosteroids in COPD, the results of larger controlled trials of the effects of inhaled corticosteroids on FEV₁ decline and disease progression were keenly awaited.²⁴ We now recognise that one of the major effects of inhaled corticosteroids in COPD may be through the reduction in the frequency and severity of exacerbations and thus decreasing mortality. In addition to leading to respiratory failure and mortality in more severe disease, COPD exacerbations are associated with increased systemic inflammatory markers such as plasma fibrinogen and C-reactive protein (CRP), so patients with a history of frequent exacerbations may have an increased risk of cardiovascular disease.^25,26 A reduction in exacerbation frequency in patients with COPD may therefore lead not only to improvements in health status, disease progression and respiratory mortality, but also to a reduction in cardiovascular morbidity and mortality. We now await the results of the TORCH (TOwards a Revolution in COPD Health) study of the long term effects of inhaled corticosteroids and long acting bronchodilators (alone and in combination) on mortality in COPD in one of the largest prospective studies ever undertaken on the treatment of COPD—the TORCH light that will now perhaps guide us to the end of the tunnel?²⁷