Impact of triple therapy on mortality in COPD

Context

COPD is an airway disease with persistent respiratory symptoms that causes globally ∼3 million deaths every year [1]. Tobacco exposure has been proven to be the main risk factor for the development of COPD [2]. With the increasing prevalence of smoking in developing countries, and ageing populations in high-income countries, prevalence of COPD is expected to rise over the next 40 years. By 2030, there may be over 4.5 million deaths annually from COPD and related conditions [1]. Only smoking cessation in mild and moderate COPD patients [3], vaccination, long-term oxygen therapy in severely hypoxaemic COPD patients, and lung volume reduction surgery in selected COPD patients with emphysema were found to have a consistent benefit on mortality [4–6]. Pharmacological treatment choices such as inhaled bronchodilators are central to symptom management and are commonly given on a regular basis to prevent or reduce symptoms. However, none of the individual studies that were powered to evaluate the effects of pharmacological treatments on COPD mortality have detected a significant survival benefit [7–10].

In recent years, two large randomised controlled trials provided new evidence on reduction of exacerbations with combination therapy including an inhaled corticosteroid (ICS), a long-acting muscarinic antagonist (LAMA) and a long-acting β₂-agonist (LABA) given in a single inhaler (triple therapy). These trials are called the IMPACT (“Informing the Pathway of COPD Treatment”) and ETHOS (“Efficacy and Safety of Triple Therapy in Obstructive Lung Disease”) trials [11, 12]. These trials also suggested a reduction in mortality with triple therapy compared with dual bronchodilation [11–14].

Here we describe the methods and results of the IMPACT and ETHOS trials with a specific focus on the impact of triple therapy on COPD mortality and discuss the strengths and limitations of these trials and how their design could have affected their findings.

Commentary

COPD is a leading cause of mortality worldwide and several large trials have evaluated whether COPD-related treatments reduce mortality. Although rarely powered by design to investigate mortality, various previous randomised controlled trials have assessed the effect of different bronchodilators on mortality in COPD as well. No statistically significant difference in mortality was observed with LABA/ICS, LABA alone or ICS alone compared with placebo in previous trials [7, 10]. Even the SUMMIT trial (Study to Understand Mortality and Morbidity in COPD) involving patients with moderate COPD with heightened cardiovascular risk which was powered to study all-cause mortality, did not show a significant effect for FF/VI compared with FF (hazard ratio 0.91 (95% CI 0.77–1.08); p=0.284) or VI alone (hazard ratio 0.96 (95% CI 0.81–1.14); p=0.655) [7].

Besides smoking cessation, only few therapies have been shown to prolong survival in specific patients with COPD including long-term oxygen therapy, vaccination and lung volume reduction surgery in selected patients [1, 4, 5]. The IMPACT and ETHOS trials suggested for the first time that bronchodilators could also improve mortality [11–14]. Both trials reported a markedly large (40–50%) reduction of mortality with LAMA/LABA/ICS over LAMA/LABA combinations in patients with COPD [11–14]. These results need, however, to be interpreted with caution.

First, in both trials, mortality was a secondary or an exploratory outcome. Therefore, these trials were not powered by design to evaluate the impact of triple therapy on mortality. Furthermore, mortality reduction has to be put in perspective with the low mortality rate in both studies. Indeed, mortality findings were based on 164 deaths out of 8588 participants (mortality rate of 1.9%) in ETHOS and 138 deaths out of 10 355 participants (mortality rate of 1.3%) in IMPACT [11–14].

Second, both the ETHOS and IMPACT trials have shown a significant reduction in mortality with triple therapy in the first year of treatment. However, by stratification over follow-up time, some discrepancies have to be pointed out [13–15]. Indeed, during the first 90 days after treatment initiation, there was a significantly lower proportion of deaths in the triple therapy group than in the dual therapy group in both the IMPACT and ETHOS studies (rate ratio of mortality of 0.24 (95% CI 0.12–0.48) and 0.37 (0.15–0.95), respectively). By contrast, during the subsequent 91–365 days of follow-up, the death rate was more similar or no longer statistically significant for triple and dual therapy (rate ratio 1.03 (0.71–1.50) and 0.86 (0.57–1.31) for the IMPACT and ETHOS trials, respectively) [13–15]. This would suggest that triple therapy only has a protective effect during the first 3 months of use, with no sustained benefit thereafter when the large majority of deaths occurred, or that withdrawal of ICS has an acute harmful effect compared to maintenance of ICS. Therefore, several methodological issues have been raised to explain this controversial benefit of triple therapy [16]. A key methodological issue is the management of treatment of patients at enrolment. Treatment had to be discontinued at randomisation for all patients. However, up to 70–80% of patients had ICS withdrawal at the enrolment of both the ETHOS and IMPACT studies (table 1) in the LABA/LAMA arms but none of the patients in the ICS containing treatment arms. By comparison, only 33% of patients had ICS before inclusion in SUMMIT [7]. It is possible that ICS discontinuation may have contributed to some early death events. This effect of ICS discontinuation is especially prominent in IMPACT where, in the first month, the incidence rate of first exacerbation under LAMA/LABA was higher than under LAMA/LABA/ICS. This may be due to the fact that in both the IMPACT and ETHOS trials, patients had raised eosinophil counts at baseline (table 1). In these patients, especially those with blood eosinophils over 300 cells·µL⁻¹, ICS discontinuation increases exacerbations and might therefore contribute to a higher mortality [17].

Third, the inclusion and exclusion criteria of both ETHOS and IMPACT were designed to select patients likely to respond to ICS. Patients with a previous history of asthma or atopy were not excluded. In addition, up to one-fifth of patients in the IMPACT study and almost one-third in ETHOS had significant bronchodilator reversibility (table 1). This is in contrast to other trials, such as FLAME [18], that included patients that were less likely to respond to ICS, as they excluded patients with a current or previous diagnosis of asthma or atopy or allergic rhinitis, or patients with respiratory symptoms before the age of 40 years, or patients with eosinophils ≥600 cells per mm³. Furthermore, the patients included in IMPACT and ETHOS were more symptomatic with more frequent moderate or severe exacerbations compared with previous trials [11, 12]. In both trials, more than half of patients had two or more moderate or severe exacerbations during the year before inclusion. By contrast, a recent pooled analysis from six randomised controlled trials including COPD patients with a lower exacerbation risk (e.g. <20% of patients reported ≥2 exacerbations in the prior year) showed contrasting results [19]. No statistically significant difference in time to death was observed between LAMA/LABA (3133 patients) and LAMA/LABA/ICS (3133 patients) after propensity score matching including exacerbation history in the previous year (hazard ratio for LAMA/LABA 1.06 (95% CI 0.68–1.64); p=0.806). This suggests triple therapy may predominantly benefit selected patients with a high risk of exacerbation (>2 exacerbations per year).

Therefore, the difference of inclusion criteria and baseline characteristics of patients in the ETHOS and IMPACT trials may have affected all outcomes and also mortality compared with previous trials.

Implications for research and practice

The management of COPD should be based on a systematic approach to identify the phenotype and comorbidities that can influence the prognosis. The current COPD treatment guidelines recommend a stepwise approach and personalisation of treatments [1]. In patients with symptomatic severe COPD (FEV₁ <50%) and a history of exacerbations, triple therapy has shown a reduction of exacerbations over a LAMA–LABA association. However, as discussed above, there are no conclusive data yet that triple therapy reduces mortality. Even if there is mortality benefit, this may be driven by maintaining corticosteroid therapy in ICS responsive groups. Therefore, future studies need to consider ICS responsiveness to evaluate the effect of ICS withdrawal on mortality. Furthermore, following an adaptive selection method that aims at avoiding the previous medications' withdrawal effect will be important for future studies to evaluate the effect of triple therapies [20].

Conclusion

There are no conclusive data yet that triple therapy causally reduces mortality in COPD. In both the ETHOS and IMPACT trials, even if a potential reduction in mortality was observed (secondary outcome), there are several methodological issues that prompt to interpret these results with caution. Therefore, reduction of exacerbations in symptomatic patients with severe COPD and a history of exacerbations should remain the main objective of triple therapy. Future, well designed and powered trials are needed to validate the findings on mortality.