Although results of short-term clinical trials showing that ICSs improved FEV₁ have been reported, there have been many reports of long-term large clinical trials showing that ICSs had no preventive effect in relation to FEV₁ or the FEV₁ decline rate (Table 13.1). For example, ICSs had no effect on the FEV₁ decline rate in such long-term, large-scale clinical trials as the European Respiratory Society study on COPD (EUROSCOP), Copenhagen City Lung Study (CCLS), ISOLDE, or Lung Health Study (LHS) [2, 34, 36, 39]. By contrast, in the Groningen Leiden Universities Corticosteroids in Obstructive Lung Disease (GLUCOLD) study, a significant ameliorating effect on the percentage change in FEV₁ was seen in the ICS group (+7.3 mL/year) in comparison with the placebo group (−79 mL/year), and correlations were also seen between slowing of the FEV₁ decline rate and decreases in the numbers of inflammatory cells in the airway [61]. A preventive effect of ICSs on the FEV₁ decline rate was seen in the TORCH trial as well [3], but the possibility of the FEV₁ decline rate in the placebo group having been overestimated because of a dropout case bias was pointed out [66].

In a meta-analysis by van Grunsven et al., it was concluded that an improvement in FEV₁ in response to ICSs had been observed in a 2-year observation period [67]. However, Alsaeedi et al. have stated that the number of RCTs that were suitable for assessment was small and an adequate meta-analysis could not be performed [68]. According to the Cochrane Review, which also analyzed the effectiveness of BDP in improving FEV₁, the number of RCTs that could be assessed was small, and it was impossible to draw a clear conclusion [69]. However, Yang et al. later conducted a meta-analysis of 55 RCTs (n = 16,154 subjects) and concluded that ICSs had been found to have no preventive effect on the rate of decline in FEV₁ [4].

The QOL of COPD patients has been assessed by means of a variety of questionnaires (Table 13.1), and St George’s Respiratory Questionnaire (SGRQ) has recently come into widespread use. The results of the SGRQ in large-scale surveys, including the ISOLDE, Trial of Inhaled Steroids and Long-Acting β2 Agonists (TRISTAN), and TORCH surveys, have shown an ameliorating effect of ICSs on QOL [2, 3, 47, 48, 59, 65]. A similar conclusion was also drawn in a meta-analysis in recent years [4]; however, caution is necessary, because many studies have concluded that ICSs were effective based on statistically significant differences, even though the SGRQ results did not show an improvement of 4 points or more, which is considered significant clinically [70]. Moreover, since the comparisons were not made with the values before using ICSs and the results did not show that ICSs had improved the SGRQ score 4 points or more in comparison with the placebo, the clinical evidence that ICSs improve the QOL of COPD patients can only be said to be scant.

Conflicting results showing that ICSs are effective and ineffective in preventing exacerbations have been published (Table 13.1). A COPD exacerbation-preventing effect of ICSs was shown in ISOLDE, TRISTAN, and TORCH trials [2.3.47]. In the results of meta-analyses, no exacerbation-preventing effect of ICSs was found by van Grunsven et al. [67], whereas Alsaeedi et al. reported a 30 % decrease in exacerbation frequency [68]. In subsequent meta-analyses, Yang et al. concluded that exacerbations had been reduced by ICSs at a rate of one exacerbation in 4 years [4], whereas Agarwal et al. claimed that the exacerbation-preventing effect of ICSs was slight and that the preventive effect of ICSs in previous reports had been exaggerated [5]. Moreover, Ernst et al. pointed out problems in the studies showing that ICSs were effective, i.e., that there were differences between the methods used to diagnose an exacerbation, that intention to treat (ITT) had not been analyzed, that the calculations of the number needed to treat (NNT) were incorrect, etc [71].

A retrospective study by Soriano et al. using the UK General Practice Research Database showed that the hospital admission rate and mortality rate of patients being treated with FP were lower than among patients being treated with just a bronchodilator other than LABA, i.e., with LAMA or short-acting β2 agonist (SABA) [72]. However, no significant differences in mortality rate between the ICS group and placebo group were found in the TORCH trial [3], and a meta-analysis in recent years also reported that ICSs did not affect the COPD mortality rate [4]. Thus, there is little evidence to support ICS effectiveness in reducing the mortality rate.

Clinically, ICSs are being used to treat COPD in combination with bronchodilators, including LABA and LAMA. Szafranski et al. reported that BUD monotherapy did not reduce COPD exacerbations, but that exacerbation frequency declined significantly when BUD was used in combination with formoterol [50], and in the TORCH trial, effectiveness in improving FEV₁ and SGRQ scores in the salmeterol (SAL)/FP combination therapy group was better than in the monotherapy group [3]. Nevertheless, results showing no differences in exacerbation frequency between an ICS/LABA group and a LABA-alone group have also been reported [73], and in the LANTERN trial, which compared ICS/LABA and LABA/LAMA, there were fewer exacerbations in the indacaterol + glycopyrronium (QVA149) group than in the SAL/FP group [74]. Moreover, results of more than one RCT showing greater effectiveness of LABA/LAMA than of ICS/LABA in ameliorating FEV₁ as well have been reported [75, 76]. Furthermore, the results of a meta-analysis also showed that LABA/LAMA was superior to ICS/LABA in relation to ameliorating FEV₁ and its exacerbation-preventing effect, and data showing a lower risk of pneumonia have also been published [77]. According to another meta-analysis, exacerbation-preventing effectiveness was reported to be greatest in a three-drug (tiotropium + BUD/formoterol) combination therapy group [78].

Various studies have also been conducted on the effects of discontinuing ICSs on COPD. In the COPE study, which compared a group in which FP had been discontinued after 4 months and a group in which it had been continued, the interval before exacerbations was shorter in the group in which FP had been discontinued, and health-related QOL had also decreased [79], and in a study by Choudhury et al., the risk of COPD exacerbations was shown to have increased as a result of discontinuing ICSs [80]. Moreover, in the COPD and Seretide: a Multi-Center Intervention and Characterization (COSMIC) study, which compared a group in which FP had been discontinued after 3 months of treatment with SAL/FP and a group in which it had been continued, the magnitude of the decline in FEV₁ was greater in the group in which it had been discontinued. The difference in moderate-to-severe exacerbation frequency in that study was not significant, but mild exacerbations were more frequent in the group in which FP had been discontinued [81]. O’Brien et al. reported finding that when BDP was discontinued in elderly COPD patients, their FEV₁ decreased, and that shortness of breath evaluated on the Borg scale became severer [82]. Furthermore, in the 5-year observation period in the recent GLUCOLD study as well, when the ICS was discontinued after 30 months of treating patients with moderate-to-severe COPD, worsening of their FEV₁, QOL, and airway hypersensitivity was seen [83].

In contrast to the above results, there have also been results showing that it was possible to safely discontinue ICSs in COPD patients. In the INSTEAD trial, which distributed patients with moderate COPD and no history of exacerbations into a group treated with indacaterol (150 μg once daily) after treatment with SAL/FP (50/500 μg twice daily) for 3 months and an SAL/FP (50/500 μg twice daily) group, no significant differences between the groups in FEV₁, SGRQ scores, or exacerbation frequency were seen 12 weeks later [84]. Moreover, in the OPTIMO trial, which compared an ICS continuation group and ICS discontinuation group of 914 patients (FEV₁ predicted >50 %, exacerbation frequency <2 times/year) being treated with a combination of an ICS and a bronchodilator, no differences were seen between the groups in FEV₁, COPD Assessment Test (CAT) scores, or exacerbation rates 6 months later [85]. Thus, it was possible to discontinue the ICS safely in patients with up to moderate COPD, in which the risk of exacerbation is low. On the other hand, in the 52-week WISDOM trial in which FP was continued or gradually discontinued in FEV₁% predicted <50 %, exacerbation frequency >1/year COPD patients (n = 2815) who were receiving tiotropium (TIO) + SAL + FP triple therapy, no significant difference was seen in exacerbation frequency, but FEV₁ decreased in the group in which FP had been discontinued [86]. Consequently, it appeared to be possible to gradually discontinue ICSs even in patients with severe COPD, who have a high risk of exacerbation, but that their respiratory function might decrease.