Inhaled Triple Therapy at Two Glucocorticoid Doses in Moderate-to-Very Severe COPD

Klaus F. Rabe, M.D., Ph.D.1, Fernando J. Martinez, M.D.2, Gary T. Ferguson, M.D.3, Chen Wang, M.D., Ph.D.4, Dave Singh, M.D.5, Jadwiga A. Wedzicha, M.D.6, Roopa Trivedi, M.S.7, Earl St. Rose, M.S.8, Shaila Ballal, M.S.8, Julie McLaren, M.D.9, Patrick Darken, Ph.D.8, Magnus Aurivillius, M.D., Ph.D.10, Colin Reisner, M.D.8, and Paul Dorinsky, M.D.7, for the ETHOS Investigators*

1. LungenClinic Grosshansdorf and Christian-Albrechts University Kiel, Airway Research Center North, Member of the German Center for Lung Research (DZL), Grosshansdorf, Germany

2. Joan and Sanford I. Weill Department of Medicine, Weill Cornell Medicine, New York, NY, USA

3. Pulmonary Research Institute of Southeast Michigan, Farmington Hills, MI, USA

4. National Clinical Research Centre for Respiratory Diseases, China-Japan Friendship Hospital, Beijing, China

5. Medicines Evaluation Unit, University of Manchester, Manchester University NHS Foundation Hospitals Trust, Manchester, UK

6. National Heart and Lung Institute, Imperial College London, London, UK

7. AstraZeneca, Durham, NC, USA

8. AstraZeneca, Morristown, NJ, USA

9. AstraZeneca, Gaithersburg, MD, USA

10. AstraZeneca, Gothenburg, Sweden

*A complete list of investigators in the ETHOS trial is provided in the Supplementary Appendix, available at NEJM.org.

Corresponding author: K.F. Rabe, LungenClinic Grosshansdorf, Wöhrendamm 80, 22927 Großhansdorf, Germany; k.f.rabe@lungenclinic.de

Word count: 2,763

Tables/figures: 3/2 (5 maximum)

ABSTRACT (260/250 words)

BACKGROUND

Triple fixed-dose combinations (inhaled glucocorticoid [ICS]/long-acting muscarinic antagonist [LAMA]/long-acting β2-agonist [LABA]) for chronic obstructive pulmonary disease (COPD) have only been studied at a single ICS dose level.

METHODS

This 52-week, phase 3, randomized trial assessed the efficacy and safety of budesonide/glycopyrrolate/formoterol (ICS/LAMA/LABA) triple therapy at two ICS dose levels (320/18/9.6 µg and 160/18/9.6 µg), versus glycopyrrolate/formoterol 18/9.6 µg (LAMA/LABA) and budesonide/formoterol 320/9.6 µg (ICS/LABA) twice-daily, in patients with moderate-to-very severe COPD and ≥1 exacerbation in the previous year. All treatments were administered via an otherwise identical Aerosphere inhaler. The primary endpoint was the rate of moderate/severe COPD exacerbations over 52 weeks.

RESULTS

The primary analysis population (a modified intent-to-treat population using on-treatment data only) included 8509 patients. Rates of moderate/severe exacerbations were 1.08 with 320 µg-budesonide triple therapy (N=2137), 1.07 with 160 µg-budesonide triple therapy (N=2121), 1.42 with glycopyrrolate/formoterol (N=2120), and 1.24 with budesonide/formoterol (N=2131). Rate ratios for 320 µg-budesonide triple therapy were 0.76 versus glycopyrrolate/formoterol (95% confidence interval [CI], 0.69–0.83; 24% reduction; P<0.001) and 0.87 versus budesonide/formoterol (95% CI, 0.79–0.95; 13% reduction; P=0.003). Rate ratios for 160 µg-budesonide triple therapy were 0.75 versus glycopyrrolate/formoterol (95% CI, 0.69–0.83; 25% reduction; P<0.001) and 0.86 versus budesonide/formoterol (95% CI, 0.79–0.95; 14% reduction; P=0.002). The overall incidence of adverse events was similar across treatment groups (61.7%–64.5%); the incidence of confirmed pneumonia was 3.5–4.5% in the ICS groups and 2.3% with glycopyrrolate/formoterol.

CONCLUSIONS

Triple combination therapy with budesonide/glycopyrrolate/formoterol at 160 µg and 320 µg budesonide dose levels twice-daily reduced moderate/severe COPD exacerbations compared with glycopyrrolate/formoterol and budesonide/formoterol.

(Funded by AstraZeneca, ETHOS ClinicalTrials.gov number, NCT02465567.)

INTRODUCTION

Treatment recommendations for chronic obstructive pulmonary disease (COPD) utilize a step-wise approach whereby treatments are added as necessary to control symptoms and reduce or eliminate exacerbations,1 with the additional goal of reducing mortality from the disease. Triple therapy with an inhaled glucocorticoid (ICS)/long-acting muscarinic antagonist (LAMA)/long-acting β2-agonist (LABA) combination reduces COPD exacerbation risk and improves lung function, symptoms, and health-related quality of life compared with dual therapies,2-6 and is recommended for patients with continued symptoms or exacerbations while receiving dual therapy with a LAMA/LABA or ICS/LABA.1

It is known that ICS-associated adverse events include pneumonia, bone fractures, and cataracts, for which the magnitude of risk may depend on the duration, dose, and type of ICS treatment.7-10 Therefore we evaluated the efficacy and safety of two single-inhaler triple fixed-dose combinations, i.e., budesonide (at two dose levels) with a LAMA and LABA, over 52 weeks compared with the corresponding LAMA/LABA and ICS/LABA therapies in symptomatic patients with moderate-to-very severe COPD with a history of exacerbation(s) in the previous year.

METHODSTrial Design and Oversight

ETHOS (NCT02465567) was a phase 3, randomized, double-blind, parallel-group trial conducted in 26 countries. The trial design has been published,11 and the protocol and statistical analysis plan are available at NEJM.org.

Patients were 40–80 years of age with symptomatic COPD (COPD Assessment Test score ≥10; range, 0–40, with higher scores indicating more symptoms; minimum clinically important difference, 2 units), receiving ≥2 inhaled maintenance therapies at screening, and had a post-bronchodilator forced expiratory volume in 1 second [FEV1]/forced vital capacity ratio of <0.7 with a post-bronchodilator FEV1 25–65% of predicted normal. Patients had to have a smoking history of ≥10 pack-years, and a documented history of ≥1 moderate/severe COPD exacerbation (if their FEV1 was <50% of predicted normal), or ≥2 moderate or ≥1 severe COPD exacerbations (if their FEV1 was ≥50% of predicted normal), in the year before screening. Patients with a current diagnosis of asthma were excluded, but those with a previous history of asthma (e.g. as a child or adolescent) were eligible. Investigators were advised not to enroll patients with an active asthma diagnosis within the past 5–10 years.

Patients were randomized 1:1:1:1 to 52 weeks’ treatment with ICS/LAMA/LABA (budesonide/glycopyrrolate/formoterol fumarate with 160/9/4.8 µg or 80/9/4.8 µg per actuation), LAMA/LABA (glycopyrrolate/formoterol fumarate 9/4.8 µg per actuation), or ICS/LABA (budesonide/formoterol fumarate 160/4.8 µg per actuation), all administered from identical metered dose inhalers (Aerosphere, AstraZeneca). Patients received two doses per day, and each dose consisted of two actuations (i.e., each dose of triple therapy delivered 320 or 160 µg budesonide). Randomization was stratified by exacerbation history (1 or ≥2 moderate/severe exacerbations), post-bronchodilator FEV1 (25% to <50% or 50 to <65% predicted), blood eosinophil count (<150 or ≥150 cells/mm3), and country. Patients discontinued COPD maintenance medications after Visit 1 (except ICS, if used before screening), and received scheduled ipratropium and as-needed albuterol during screening (1–4 weeks). Ipratropium and ICS were discontinued at randomization.

The protocol and informed consent form were approved by appropriate institutional review boards, independent ethics committees, or health authorities; all patients signed written informed consent before screening. Independent data monitoring and clinical endpoint committees reviewed safety data throughout the trial, including cardiovascular, cerebrovascular, and pneumonia events, and cause-specific mortality.

The trial was designed by AstraZeneca and the principal academic investigators. Data were collected by clinical investigators and analyzed by employees of Everest Clinical Research Services and AstraZeneca. The first draft of the manuscript was written by a medical writer funded by the sponsor, under the direction of the authors. All the authors critically reviewed and provided feedback on all subsequent versions of the manuscript and, along with the sponsor, made the decision to submit the manuscript for publication. All authors had access to the data, contributed to data interpretation, and vouch for the accuracy and completeness of the data and the fidelity of the trial to the protocol.

Endpoints

The primary efficacy outcome was the rate of moderate/severe COPD exacerbations over 52 weeks. Moderate exacerbations were defined as those leading to treatment with systemic glucocorticoids and/or antibiotics for at least 3 days; severe exacerbations were defined as those resulting in hospitalization or death. Secondary endpoints were defined according to region;11 for the US statistical approach these included the time to first moderate/severe COPD exacerbation over 52 weeks, change from baseline in average daily rescue medication use over 24 weeks, the percentage of patients achieving a ≥4 unit decrease in the St. George’s Respiratory Questionnaire (SGRQ) total score at Week 24 (total scores range from 0–100, with lower scores indicating better health-related quality of life; the minimum clinically important difference is 4 units), time to death (all-cause), and rate of severe COPD exacerbations over 52 weeks. Details of the ex-US approach and additional pre-defined endpoints are listed in Table S1 in the Supplementary Appendix.

Other analyses of the primary endpoint included pre-specified subgroup analyses by exacerbation history (≥2 moderate/severe exacerbations in the previous year), by ICS use at screening, and by blood eosinophil count (<150 or ≥150 cells/mm3; supplemented by a generalized additive model predicting exacerbation rates based on eosinophil count as a continuous variable), and a post-hoc subgroup analysis by reversibility at screening.

A comprehensive framework, described in recent regulatory guidelines,12 was used to provide clarity on the description of estimates of treatment effect, including the handling of potentially confounding events such as treatment discontinuations. Each approach is denoted by a different “estimand”, which includes the study population, endpoint/variable(s), and pre-specifies how these confounding events will be handled in the analysis. Most efficacy analyses were conducted using an efficacy estimand in the modified intent-to-treat population, which includes on-treatment data from all randomized and treated patients. There were two exceptions: the first secondary analysis of the rate of moderate/severe exacerbations was conducted using an attributable estimand, where data following treatment discontinuation due to lack of efficacy or adverse events were imputed with poor responses,13 and time to death was assessed using a treatment policy estimand, with on- and off-treatment data from all randomized and treated patients. Although this terminology is relatively new, analyses using the efficacy estimand are comparable to those traditionally used in previous trials evaluating COPD exacerbations, as is the approach of using treatment policy for time to death.

For the primary endpoint, the order of treatment comparisons was: 320 µg-budesonide triple therapy versus glycopyrrolate/formoterol, 320 µg-budesonide triple therapy versus budesonide/formoterol, 160 µg-budesonide triple therapy versus glycopyrrolate/formoterol (all for superiority), and 160 µg-budesonide triple therapy versus budesonide/formoterol (for non-inferiority followed by superiority).

Safety assessments included physical examinations, vital signs, electrocardiograms, clinical laboratory tests, and adverse event monitoring.

Subsets of patients also participated in 4-hour pulmonary function test and 24-hour Holter monitoring sub-studies (not reported in this manuscript).

Statistical Analysis

The sample size of 8400 was selected to provide 93% power to detect a 15% reduction in the rate of moderate/severe exacerbations for 320 µg-budesonide triple therapy versus both glycopyrrolate/formoterol and budesonide/formoterol (96% power for each comparison), with Type I error controlled at an equivalent of a two-sided alpha level of 0.05 (further details are provided in the Supplementary Appendix).

Exacerbation rates were analyzed using negative binomial regression. Time to first event analyses were performed using Cox regression. Rescue medication use was analyzed using a linear mixed model with repeated measures, and SGRQ responder analysis was performed using logistic regression.

RESULTS Patient Population

Overall, 8588 patients were randomized, and 8573 were treated (Fig. 1). The safety population included 8529 patients; the ITT and mITT populations included 8509 patients. In total, 7187 (83.8%) patients completed the trial, with 6654 (77.6%) completing 52 weeks of treatment (79.4% and 80.4% with 320 µg- and 160 µg-budesonide triple therapy, respectively, 74.1% with glycopyrrolate/formoterol and 76.6% with budesonide/formoterol).

Baseline demographics in the mITT population were similar across treatment groups (Table 1). Overall, 80.5% of patients were using ICS at screening; other prior COPD medications are detailed in Table S3 of the Supplementary Appendix.

Primary Efficacy Analyses

The model-adjusted rates of moderate/severe exacerbations over 52 weeks were 1.08 with 320 µg-budesonide triple therapy, 1.07 with 160 µg-budesonide triple therapy, 1.42 with glycopyrrolate/formoterol, and 1.24 with budesonide/formoterol. The rate of moderate/severe exacerbations was significantly lower with 320 µg-budesonide triple therapy versus both glycopyrrolate/formoterol (rate ratio 0.76, 95% CI, 0.69 to 0.83; 24% reduction; P<0.001) and budesonide/formoterol (rate ratio 0.87, 95% CI, 0.79 to 0.95; 13% reduction; P=0.003). Similarly, the moderate/severe exacerbation rate was significantly lower with 160 µg-budesonide triple therapy versus glycopyrrolate/formoterol (rate ratio 0.75, 95% CI, 0.69 to 0.83; 25% reduction; P<0.001), and budesonide/formoterol (rate ratio 0.86, 95% CI, 0.79 to 0.95; 14% reduction; P=0.002) (Table 2). No difference between the 320 µg- and 160 µg-budesonide triple therapies was observed (rate ratio 1.00, 95% CI 0.91 to 1.10).

Secondary and Other Efficacy Analyses

Results for moderate/severe exacerbation rate using the attributable estimand were similar to the primary analysis (Table 2). Both doses of triple therapy significantly prolonged the time to first moderate/severe exacerbation versus both dual therapies (Fig. 2A; Table 2).

Rates of severe exacerbations over 52 weeks were 0.13 for 320 µg-budesonide triple therapy, 0.14 for 160 µg-budesonide triple therapy, 0.15 for glycopyrrolate/formoterol, and 0.16 for budesonide/formoterol. The rate of severe exacerbations was significantly lower with 320 µg-budesonide triple therapy versus budesonide/formoterol (rate ratio 0.80, 95% CI 0.66 to 0.97; 20% reduction; P=0.02). The rate ratio for 320 µg-budesonide triple therapy versus glycopyrrolate/formoterol was 0.84 (95% CI 0.69 to 1.03; 16% reduction). For 160 µg-budesonide triple therapy, non-inferiority to budesonide/formoterol was demonstrated (rate ratio 0.82, 95% CI 0.68 to 1.00; per-protocol population); however, treatment differences were not significant versus either dual therapy (Table 2).

Both triple therapies significantly reduced rescue medication use over 24 weeks, and significantly increased the odds of being an SGRQ responder at Week 24 versus glycopyrrolate/formoterol and budesonide/formoterol (P≤0.01 for all) (Table S4 in the Supplementary Appendix).

Triple therapy with 320 µg budesonide reduced the risk of death (all-cause) versus glycopyrrolate/formoterol (28 vs 49 deaths; hazard ratio 0.54; 95% CI 0.34 to 0.87; 46% reduction). The hazard ratio for time to death with 320 µg-budesonide triple therapy versus budesonide/formoterol was 0.78 (28 vs 34 deaths; 95% CI, 0.47 to 1.30); for 160 µg-budesonide triple therapy, hazard ratios were 0.79 versus glycopyrrolate/formoterol (39 vs 49 deaths; 95% CI, 0.52 to 1.20) and 1.13 versus budesonide/formoterol (39 vs 34 deaths; 95% CI 0.72 to 1.80) (Fig. 2B; Table 2). Adjudicated causes of death are provided in Table S5 of the Supplementary Appendix.

Data on pre-specified analyses of moderate/severe exacerbation rates by eosinophil counts, prior ICS use, and reversibility at screening are provided in the supplemental appendix (Fig. S4-S5 and Table S6). Benefits of triple therapy on exacerbation rates versus dual therapies were observed in both eosinophil subgroups (<150 and ≥150 cells/mm³), and regardless of whether patients had prior ICS use or reversibility to albuterol.

Safety and Adverse Event Profile

Overall, the percentage of patients experiencing at least one adverse event ranged from 61.7%–64.5% across groups. Rates of serious adverse events were 19.9–21.0% across treatment groups. The most frequently reported adverse events overall were nasopharyngitis (10.5%), COPD (10.4%), and upper respiratory tract infection (5.6%) (Table 3).

The incidence of confirmed pneumonia was 3.5–4.5% in the ICS groups (41.3–57.8 events per 1000 patient-years) and 2.3% with glycopyrrolate/formoterol (28.8 events per 1000 patient-years). The time to first confirmed pneumonia event was longer with glycopyrrolate/formoterol than in the ICS-containing groups (P<0.05 for all comparisons; Table S7 in the Supplementary Appendix). Serious confirmed pneumonia events occurred more often in the ICS-containing groups (2.4–3.0%) than with glycopyrrolate/formoterol (1.3%) (P<0.05 for all comparisons; Table 3 and Table S8 in the Supplementary Appendix). A similar pattern was observed for pharmacologically expected local steroid effects (dysphonia and candidiasis), with a lower incidence in the glycopyrrolate/formoterol group than with ICS-containing treatments (P<0.05 for all comparisons), while the incidences of other systemic steroid effects including diabetes mellitus, bone fracture, and ocular effects were similar across groups (P>0.05 for all comparisons; Table S10 in the Supplementary Appendix).

There were no clinically meaningful findings across treatment groups in vital signs, electrocardiograms, or clinical laboratory assessments.

DISCUSSION

In this randomized trial of more than 8,500 patients with moderate-to-very severe COPD, triple combination therapy with 320 µg or 160 µg budesonide twice-daily resulted in significantly lower rates of moderate/severe exacerbations versus LAMA/LABA and ICS/LABA. In addition, both triple therapy doses significantly improved patient-reported outcomes versus both dual therapies.

ETHOS assessed two different ICS dose levels in a fixed-dose triple therapy for COPD. While the evaluation of a statistical dose-response was not part of our testing hierarchy and the trial was not powered to detect a significant difference between the triple therapy doses, unadjusted comparisons between doses showed similar efficacy for most exacerbation endpoints. These comparisons also showed trends in favor of 320 µg-budesonide triple therapy for severe exacerbations, SGRQ responders, and rescue medication use. Furthermore, despite a mortality rate overall of 1.8%, a risk reduction in all-cause mortality was seen versus LAMA/LABA only with the 320 µg-budesonide triple therapy. The hazard ratio for 320 µg- versus 160 µg-budesonide triple therapy for time to death was 0.69, but the 95% CI was 0.42 to 1.13, precluding any definitive conclusions regarding dose response. This is the second trial to show a benefit on mortality for triple therapy versus LAMA/LABA in patients with COPD, with risk reductions versus LAMA/LABA in the ITT population (including on- and off-treatment data) of 46% in ETHOS and 29% in IMPACT.4 The difference observed between the two budesonide doses in mortality, but not other outcomes, is unexplained but may reflect a beneficial effect on cardiovascular outcomes in this high-risk population. This has been previously suggested by findings from the TORCH study,14,15 but was not proven in a subsequent interventional study in a patient population with cardiovascular risk factors but less severe respiratory disease (SUMMIT).16

The reductions in moderate/severe exacerbation rates versus dual LAMA/LABA and ICS/LABA therapies in ETHOS were comparable with findings of previous 52-week trials, including IMPACT, TRILOGY, and TRIBUTE.4-6 In addition, the results of ETHOS further build on the findings from KRONOS, a 24-week trial which showed the benefits of 320 µg-budesonide/glycopyrrolate/formoterol on lung function, symptoms, and exacerbations versus dual therapies in COPD in a population in which most patients (74%) had not had an exacerbation in the previous year.2 It has been proposed that the benefits of triple therapy versus LAMA/LABA therapy in previous studies may have resulted from a short-term increase in exacerbation rates in the LAMA/LABA groups due to ICS withdrawal in patients who were using ICS prior to trial entry.17 However, in ETHOS, benefits of both triple therapy doses versus LAMA/LABA were similar in patients who were using ICS at screening and those who were not, indicating that results were not driven by acute ICS withdrawal.

Current recommendations from the Global Initiative for Chronic Obstructive Lung Disease (GOLD) suggest that patients with elevated eosinophil levels who experience continued exacerbations on a single bronchodilator should initially step-up to ICS/LABA therapy.1 Here, triple therapy with 160 µg budesonide showed significant benefits versus ICS/LABA therapy with 320 µg budesonide on exacerbations and symptoms, calling into question the role for ICS/LABA in patients with moderate-to-very severe COPD who are symptomatic and have a history of exacerbations. Both triple therapy doses showed benefits on exacerbations versus LAMA/LABA that were greater in patients with higher blood eosinophil counts, consistent with previous studies of ICS response,18,19 as well as current GOLD recommendations.1

No unexpected safety signals were identified in ETHOS. As previously shown in other 52-week COPD studies,4,20 the pneumonia incidence was higher in ICS-containing treatment groups than with LAMA/LABA. The incidence of confirmed pneumonia with 160 µg-budesonide triple therapy (3.5%), 320 µg-budesonide triple therapy (4.2%), and 320-µg budesonide/formoterol (4.5%) were all higher relative to glycopyrrolate/formoterol (2.3%). Across treatment groups, the pneumonia incidence in ETHOS was higher than in KRONOS (1.3–1.9%),2 which may reflect the longer trial duration as well as the differences in the patient population, as more severe airflow limitation and frequent exacerbations have been associated with a greater risk of pneumonia.21

In conclusion, our findings demonstrate the benefits of budesonide/glycopyrrolate/formoterol in reducing the rate of moderate or severe COPD exacerbations and improving symptoms and health-related quality of life compared with LAMA/LABA and ICS/LABA dual therapies in patients with moderate-to-very severe COPD at risk of exacerbations. Triple therapy with 320 µg-budesonide also reduced all-cause mortality versus LAMA/LABA therapy. We also showed that triple therapy with 160 µg-budesonide was an effective treatment option for COPD, demonstrating greater efficacy than higher-dose ICS/LABA therapy in reducing exacerbations and improving symptoms and health-related quality of life.

Supported by AstraZeneca. Dave Singh is supported by the National Institute for Health Research (NIHR) Manchester Biomedical Research Centre (BRC). The Aerosphere brand of inhalers is owned by AstraZeneca.

Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.

We thank all the patients, their families, and the team of investigators, research nurses, and operations staff involved in ETHOS. Medical writing support, under the direction of the authors, was provided by Julia King, Ph.D., of CMC Connect, McCann Health Medical Communications, which was funded by AstraZeneca, Gaithersburg, USA in accordance with Good Publication Practice (GPP3) guidelines.22

REFERENCES

1.Global Initiative for Chronic Obstructive Lung Disease. 2020 Report: Global Strategy for the Diagnosis, Management and Prevention of COPD. 2020. (Accessed November 05, 2019 at https://goldcopd.org.)

2.Ferguson GT, Rabe KF, Martinez FJ, et al. Triple therapy with budesonide/glycopyrrolate/formoterol fumarate with co-suspension delivery technology versus dual therapies in chronic obstructive pulmonary disease (KRONOS): a double-blind, parallel-group, multicentre, phase 3 randomised controlled trial. Lancet Respir Med 2018;6:747-58.

3.Lipson DA, Barnacle H, Birk R, et al. FULFIL trial: once-daily triple therapy for patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2017;196:438-46.

4.Lipson DA, Barnhart F, Brealey N, et al. Once-daily single-inhaler triple versus dual therapy in patients with COPD. N Engl J Med 2018;378:1671-80.

5.Papi A, Vestbo J, Fabbri L, et al. Extrafine inhaled triple therapy versus dual bronchodilator therapy in chronic obstructive pulmonary disease (TRIBUTE): a double-blind, parallel group, randomised controlled trial. Lancet 2018;391:1076-84.

6.Singh D, Papi A, Corradi M, et al. Single inhaler triple therapy versus inhaled corticosteroid plus long-acting β2-agonist therapy for chronic obstructive pulmonary disease (TRILOGY): a double-blind, parallel group, randomised controlled trial. Lancet 2016;388:963-73.

7.Kew KM, Seniukovich A. Inhaled steroids and risk of pneumonia for chronic obstructive pulmonary disease. Cochrane Database Syst Rev 2014:CD010115.

8.Cumming RG, Mitchell P, Leeder SR. Use of inhaled corticosteroids and the risk of cataracts. N Engl J Med 1997;337:8-14.

9.Suissa S, Patenaude V, Lapi F, Ernst P. Inhaled corticosteroids in COPD and the risk of serious pneumonia. Thorax 2013;68:1029-36.

10.Loke YK, Cavallazzi R, Singh S. Risk of fractures with inhaled corticosteroids in COPD: systematic review and meta-analysis of randomised controlled trials and observational studies. Thorax 2011;66:699-708.

11.Rabe KF, Martinez FJ, Ferguson GT, et al. A Phase III study of triple therapy with budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler 320/18/9.6 μg and 160/18/9.6 μg using co-suspension delivery technology in moderate-to-very severe COPD: the ETHOS study protocol. Respir Med 2019;158:59-66.

12.International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH). E9(R1). Guideline on statistical principles for clinical trials: Addendum on estimands and sensitivity analysis in clinical trials. 2019. at https://database.ich.org/sites/default/files/E9-R1_Step4_Guideline_2019_1203.pdf.

13.Darken P, Nyberg J, Ballal S, Wright D (in press). The attributable estimand: A new approach to account for intercurrent events. Pharm Stat 2020. DOI: 10.1002/pst.2019

14.Calverley PMA, Anderson JA, Celli B, et al. Salmeterol and fluticasone propionate and survival in chronic obstructive pulmonary disease. N Engl J Med 2007;356:775-89.

15.Calverley PMA, Anderson JA, Celli B, et al. Cardiovascular events in patients with COPD: TORCH study results. Thorax 2010;65:719-25.

16.Vestbo J, Anderson JA, Brook RD, et al. Fluticasone furoate and vilanterol and survival in chronic obstructive pulmonary disease with heightened cardiovascular risk (SUMMIT): a double-blind randomised controlled trial. Lancet 2016;387:1817-26.

17.Suissa S, Ariel A. Triple therapy trials in COPD: a precision medicine opportunity. Eur Respir J 2018;52.

18.Bafadhel M, Peterson S, De Blas MA, et al. Predictors of exacerbation risk and response to budesonide in patients with chronic obstructive pulmonary disease: a post-hoc analysis of three randomised trials. Lancet Respir Med 2018;6:117-26.

19.Pascoe S, Locantore N, Dransfield MT, Barnes NC, Pavord ID. Blood eosinophil counts, exacerbations, and response to the addition of inhaled fluticasone furoate to vilanterol in patients with chronic obstructive pulmonary disease: a secondary analysis of data from two parallel randomised controlled trials. Lancet Respir Med 2015;3:435-42.

20.Chapman KR, Hurst JR, Frent SM, et al. Long-term triple therapy de-escalation to indacaterol/glycopyrronium in patients with chronic obstructive pulmonary disease (SUNSET): a randomized, double-blind, triple-dummy clinical trial. Am J Respir Crit Care Med 2018;198:329-39.

21.Williams NP, Coombs NA, Johnson MJ, et al. Seasonality, risk factors and burden of community-acquired pneumonia in COPD patients: a population database study using linked health care records. Int J Chron Obstruct Pulmon Dis 2017;12:313-22.

22.Battisti WP, Wager E, Baltzer L, et al. Good publication practice for communicating company-sponsored medical research: GPP3. Ann Intern Med 2015;163:461-4.

Table 1. Baseline demographics (modified intent-to-treat population).*

Budesonide/glycopyrrolate/formoterol

320/18/9.6 µg(N = 2137)

Budesonide/glycopyrrolate/formoterol

160/18/9.6 µg(N = 2121)

Glycopyrrolate/formoterol

18/9.6 µg(N = 2120)

Budesonide/formoterol

320/9.6 µg(N = 2131)

Age — yr

64.6±7.6

64.6±7.6

64.8±7.6

64.6±7.6

Male sex — no. (%)

1260 (59.0)

1298 (61.2)

1244 (58.7)

1279 (60.0)

Current smoker — no. (%)

910 (42.6)

865 (40.8)

856 (40.4)

864 (40.5)

Number of pack-years smoked

47.0±25.1

47.9±25.8

48.4±26.5

47.1±26.3

COPD exacerbations in the past 12 months

0 moderate or severe — no. (%)

1 moderate or severe — no. (%)

≥ 2 moderate or severe — no. (%)

≥ 1 severe — no. (%)

1.7±0.8

2 (0.1)

940 (44.0)

1195 (55.9)

451 (21.1)

1.7±0.9

2 (0.1)

932 (43.9)

1187 (56.0)

463 (21.8)

1.7±0.8

2 (0.1)

907 (42.8)

1211 (57.1)

429 (20.2)

1.7±0.9

2 (0.1)

912 (42.8)

1217 (57.1)

458 (21.5)

Blood eosinophil count

Median (range), cells/mm3

≥ 150 cells/mm3 — no. (%)≥ 300 cells/mm3 — no. (%)

165.0 (0–2510)

1277 (59.8)

310 (14.5)

166.7 (5–1590)

1258 (59.3)

318 (15.0)

170.0 (5–2305)

1272 (60.0)

293 (13.8)

166.7 (0–2430)

1294 (60.7)

333 (15.6)

Post-albuterol FEV1 — % of predicted normal value

50 – <80% (moderate COPD)

30 – <50% (severe COPD)

<30% (very severe COPD)

43.6±10.3

613 (28.7)

1305 (61.1)

217 (10.2)

43.1±10.4

604 (28.5)

1270 (59.9)

245 (11.6)

43.5±10.2

596 (28.1)

1293 (61.0)

229 (10.8)

43.4±10.4

614 (28.8)

1283 (60.2)

233 (10.9)

Reversibility to albuterol

Change in FEV1 from pre- to post-albuterol — mL

Reversibility ≥ 12% and ≥ 200

mL, n (%)

146.3±158.0

657 (30.7)

144.4±151.7

631 (29.8)

148.7±151.1

669 (31.6)

142.3±144.8

654 (30.7)

Use of ICS at screening — no. (%)

1706 (79.8)

1729 (81.5)

1707 (80.5)

1704 (80.0)

COPD Assessment Test score

19.7±6.5

19.6±6.6

19.5±6.6

19.5±6.5

*Plus–minus values are means ±SD. The doses of glycopyrrolate and formoterol fumarate are equivalent to 14.4 µg of glycopyrronium and 10 µg of formoterol fumarate dihydrate, respectively. COPD denotes chronic obstructive pulmonary disease, FEV1 forced expiratory volume in 1 s, ICS inhaled glucocorticoid.

1

Table 2. Efficacy endpoints (efficacy estimand, modified intent-to-treat population unless stated).

Budesonide/glycopyrrolate/formoterol320/18/9.6 µg(N = 2137)

Budesonide/glycopyrrolate/formoterol160/18/9.6 µg(N = 2121)

Glycopyrrolate/formoterol 18/9.6 µg(N = 2120)

Budesonide/formoterol320/9.6 µg(N = 2131)

Primary endpoint

Rate of moderate or severe COPD exacerbations over 52 weeks

Model-estimated rate

1.08

1.07

1.42

1.24

320 µg-budesonide triple therapy versus comparators*

Rate ratio (95% CI)

P-value

—

1.00 (0.91 to 1.10)

—

0.76 (0.69 to 0.83)

<0.001

0.87 (0.79 to 0.95)

0.003

160 µg-budesonide triple therapy versus comparators

Rate ratio (95% CI)

P-value

—

—

0.75 (0.69 to 0.83)

<0.001

0.86 (0.79 to 0.95)

0.002

Secondary exacerbation and mortality endpoints

Rate of moderate or severe COPD exacerbations over 52 weeks (attributable estimand)

Model-estimated rate

1.25

1.23

1.63

1.47

320 µg-budesonide triple therapy versus comparators*

Rate ratio (95% CI)

P-value

—

1.01 (0.94 to 1.10)

—

0.76 (0.71 to 0.83)

<0.001

0.85 (0.78 to 0.92)

<0.001

160 µg-budesonide triple therapy versus comparators

Rate ratio (95% CI)

P-value

—

—

0.75 (0.70 to 0.82)

<0.001

0.84 (0.77 to 0.90)

<0.001

Time to first moderate or severe COPD exacerbation over 52 weeks

No. of patients with exacerbations

1026 (48.0)

1013 (47.8)

1056 (49.8)

1085 (50.9)

320 µg-budesonide triple therapy versus comparators*

Hazard ratio (95% CI)

P-value

—

1.02 (0.93 to 1.11)

—

0.88 (0.81 to 0.96)

0.004

0.89 (0.81 to 0.97)

0.006

160 µg-budesonide triple therapy versus comparators

Hazard ratio (95% CI)

P-value

—

—

0.87 (0.79 to 0.94)

0.001

0.87 (0.80 to 0.95)

0.002

Rate of severe COPD exacerbations over 52 weeks

Model-estimated rate

0.13

0.14

0.15

0.16

320 µg-budesonide triple therapy versus comparators*

Rate ratio (95% CI)

P-value

—

0.96 (0.78 to 1.17)

—

0.84 (0.69 to 1.03)

0.09

0.80 (0.66 to 0.97)

0.02

160 µg-budesonide triple therapy versus comparators

Rate ratio (95% CI)

P-value

—

—

0.88 (0.72 to 1.08)

0.22

0.83 (0.69 to 1.01)

0.06

Time to death over 52 weeks (all-cause; treatment policy estimand)†,‡

No. of patient deaths

28 (1.3)

39 (1.8)

49 (2.3)

34 (1.6)

320 µg-budesonide triple therapy versus comparators*

Hazard ratio (95% CI)

P-value

—

0.69 (0.42 to 1.13)

—

0.54 (0.34 to 0.87)

0.01‡

0.78 (0.47 to 1.30)

0.34

160 µg-budesonide triple therapy versus comparators

Hazard ratio (95% CI)

P-value

—

—

0.79 (0.52 to 1.20)

0.27

1.13 (0.72 to 1.80)

0.59

Additional secondary endpoints are reported in Table S4 in the Supplementary Appendix. *Comparisons between 320 µg-budesonide triple therapy and 160 µg-budesonide triple therapy were not included in the testing hierarchy and, therefore, no p-values are provided. †Intent-to-treat population. ‡P-values for this endpoint are considered unadjusted due to results higher in the testing hierarchy for the rate of severe exacerbations.

The means presented are least-squares means. COPD denotes chronic obstructive pulmonary disease, ICS inhaled glucocorticoid.

Table 3. Summary of adverse events (safety population).

Budesonide/glycopyrrolate/formoterol 320/18/9.6 µg(N = 2144)

Budesonide/glycopyrrolate/formoterol 160/18/9.6 µg(N = 2124)

Glycopyrrolate/formoterol 18/9.6 µg(N = 2125)

Budesonide/formoterol320/9.6 µg(N = 2136)

No. of patients, n (%)

Events (rate per 1000 patient-yr)

No. of patients, n (%)

Events (rate per 1000 patient-yr)

No. of patients, n (%)

Events (rate per 1000 patient-yr)

No. of patients, n (%)

Events (rate per 1000 patient-yr)

Any adverse event

1368 (63.8)

4527 (2388.4)

1356 (63.8)

4382 (2318.7)

1312 (61.7)

4074 (2301.7)

1377 (64.5)

4746 (2587.0)

Serious adverse events

426 (19.9)

664 (350.3)

445 (21.0)

681 (360.3)

433 (20.4)

639 (361.0)

440 (20.6)

653 (355.9)

Adverse events that led to early discontinuation

119 (5.6)

150 (79.1)

112 (5.3)

146 (77.3)

146 (6.9)

187 (105.7)

140 (6.6)

160 (87.2)

Confirmed* major adverse cardiovascular event

31 (1.4)

32 (16.9)

30 (1.4)

31 (16.4)

44 (2.1)

47 (26.6)

23 (1.1)

24 (13.1)

Confirmed* pneumonia

90 (4.2)

93 (49.1)

75 (3.5)

78 (41.3)

48 (2.3)

51 (28.8)

96 (4.5)

106 (57.8)

Deaths (all causes; on treatment)

20 (0.9)

19 (10.0)†

28 (1.3)

28 (14.8)

35 (1.6)

35 (19.8)

29 (1.4)

29 (15.8)

Adverse events that occurred in ≥3% of patients overall

Nasopharyngitis

227 (10.6)

290 (153.0)

239 (11.3)

315 (166.7)

199 (9.4)

255 (144.1)

234 (11.0)

331 (180.4)

COPD

203 (9.5)

256 (135.1)

221 (10.4)

263 (139.2)

219 (10.3)

268 (151.4)

242 (11.3)

300 (163.5)

Upper respiratory tract infection

123 (5.7)

149 (78.6)

137 (6.5)

176 (93.1)

102 (4.8)

129 (72.9)

115 (5.4)

154 (83.9)

Pneumonia

98 (4.6)

101 (53.3)

85 (4.0)

93 (49.2)

61 (2.9)

66 (37.3)

107 (5.0)

117 (63.8)

Bronchitis

66 (3.1)

74 (39.0)

68 (3.2)

76 (40.2)

76 (3.6)

81 (45.8)

69 (3.2)

83 (45.2)

*Confirmed by the clinical endpoint committee. †Twenty patients in the budesonide/glycopyrrolate/formoterol 320/18/9.6 µg group had on-treatment adverse events linked to an outcome of death; however, one patient’s death was adjudicated based on a post-treatment adverse event, and therefore was not included in the calculation of the on-treatment event rate. COPD denotes chronic obstructive pulmonary disease.

Figure 1. Patient flow diagram.Forty-four patients with overlapping treatment exposure, from participating multiple times in the same study or in another study concurrently, were excluded from all analysis populations. An additional twenty patients were included in the safety population (N=8529) but not the modified intent-to-treat population (N=8509). Nineteen of these patients were excluded based on multiple enrollments with non-overlapping treatment exposure, and one was excluded due to administrative reasons.

24

Figure 2. Time-to-first-event analyses.

Panel A shows the time to first moderate or severe COPD exacerbation in the modified intent-to-treat population (efficacy estimand), and Panel B the time to death (all-cause) in the intent-to-treat population (treatment policy estimand). BFF MDI denotes budesonide/formoterol fumarate metered dose inhaler, BGF MDI budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler, COPD chronic obstructive pulmonary disease, GFF MDI glycopyrrolate/formoterol fumarate metered dose inhaler.

35

Supplementary Appendix

1.Trial investigators2

2.METHODS9

Figure S1. Order of hypothesis testing for Type I error control: US and ex-US approaches.15

Figure S2. Forest plots for time-to-first event analyses.16

Figure S3. Change from baseline in SGRQ total score over 52 weeks.17

Figure S4. Rate of moderate or severe COPD exacerbations by eosinophil subgroups.18

Figure S5. Rate of moderate or severe COPD exacerbations by blood eosinophil count.19

Table S1. Pre-specified endpoints in the ETHOS trial protocol.20

Table S2. Additional baseline demographics24

Table S3. Prior COPD-related inhaled therapies.25

Table S4. Secondary and other pre-specified symptom and health-related quality of life endpoints26

Table S5. Summary of adjudicated causes of death.28

Table S6. Subgroup analyses of the rate of moderate or severe COPD exacerbations by exacerbation history, ICS use at screening, and reversibility29

Table S7. Time to first MACE and time to first confirmed pneumonia31

Table S8. Incidence of MACE and pneumonia adverse events.32

Table S9. Adverse events occurring in ≥2% of patients in any treatment group (by preferred term)33

Table S10. Adverse events of special interest occurring in ≥1% of patients in any treatment group (by category).34

References35

1. Trial investigators

Argentina

Hector Hugo Altieri, Norma Aramayo, Germán Arce, Miguel Bergna, Xavier Bocca, Horacio Budani, Victor Hugo Cambursano, Cristian Carminio, Fernando Antonio de la Riestra, Maria De Salvo, Héctor Defranchi, Ricardo del Olmo, Andrés Luis Echazarreta, Pedro Elías, Cristián Fazio, Marcelo Fernández, Gabriel García, Adriana Gosn, Virginia Larivey, Luis Larrateguy, Veronica Patricia Lawriwskyj, Martin Maillo, Fernando Massola, Walter Mattarucco, Andrea Medina, Alicia Beatriz Molina Bonetto, Laura Nardone, Juan Facundo Nogueira, Angel Esteban Piacenza, Federico Promencio, Miguel Quiroga, Luisa Beatriz Rey, Mariana Rivera, Alicia Rodriguez, Juan José Rodriguez Moncalvo, Ramón Rojas, Gonzalo Saenz, Pablo Sáez, Damián Silva, Adriana Sosso, Jorge Taborda, Alberto Tolcachier, Nora del Valle Vega, Fernando Verra, Luis Wehbe

Australia

Simon Bowler, Peter Bremner, Patrick Carroll, Michael Chia, Huw Davies, Mark Holmes, Jeff Karrasch, Matthew Peters, Frederick Graham Simpson, Anna Tai, Francis Thien, Peter Wark, John Wheatley

Austria

Othmar Ablinger, Bernhard Forstner, Michael Studnicka, Mahmud Sweilem, Norbert Vetter, Robert Voves, Josef Würtz, Peter Würtz

Belgium

Luc Capiau, Benoit Colinet, Ulrike Himpe, Eduard Janssens, Jean Benoit Martinot, Jaak Mortelmans, Rudi Peche, Thierry Pieters, Luc van Zandweghe, Guy Vereecken, Geert Vileyn

Brazil

Marina Andrade Lima, Alexandre Cardoso, Claudia Costa, Alberto Cukier, Elie Fiss, Jussara Fiterman, Guilherme Garcia, Irma Godoy, José Roberto Jardim, Marli Knorst, Fernando Lundgren, Waldo Mattos, Danuza Ávila de Mello, Maria Eunice Moraes de Oliveira, Marcio Neis, Karina Oliveira, Andreia Pez, Emilio Pizzichini, Marcelo Rabahi, Adalberto Rubin, Roberto Stirbulov

Canada

Syed Anees, François Blouin, Guy Chouinard, Guy Deslauriers, Anthony Dowell, Anthony D'Urzo, Francis Ervin, Tharwat Fera, Murdo Ferguson, Sam Henein, Lawrence Homik, Allan Kelly, Patrick Killorn, François Maltais, Darcy Marciniuk, Giuseppe Mazza, Andrew McIvor, Lyle Melenka, Denis O'Donnell, Bonavuth Pek, Sean Peterson, George Philteos, Claude Poirier, Dennis Reich, Don Sin, Eric St-Amour, John Taliano, Guy Tellier, Arthur Vasquez, Brian Zidel

Chile

Manuel Barros, Claudia Cartagena, Juan Cristóbal Celis Carrasco, Germán Cruz, Patricia Fernández, Roxana Fuentes, José Fuentes, Paula Galleguillos, Victor Martinez, Roxana Maturana, Eduardo Mendez, Laura Mendoza, Juan Carlos Palma Carvajal, Juana Pavie, Carlos Quilodran, Andrés Rosenblut, Patricia Schonffeldt, Rafael Silva, Sergio Vargas

China

Xiaoyue Chang, Gang Chen, Hong Chen, Rongchang Chen, Juan Du, Ganzhu Feng, Jiaxi Feng, Yingyun Fu, Xiuhua Fu, Xiwen Gao, Zhancheng Gao, Yuhai Gu, Wei Gu, Yubiao Guo, Zhongliang Guo, Xiaowen Han, Huijie He, Zhiyi He, Jian-an Huang, Jin Huang, Mao Huang, Shujuan Jiang, Mingyan Jiang, Jian Kang, Jiulong Kuang, Yali Li, Yingxiang Lin, Zeying Liu, Xiaoju Liu, Chuntao Liu, Dongming Liu, Shuying Liu, Qiaofa Lu, Bailing Luo, Zhuang Luo, Xiaodong Mei, Zongxing Ou, Wenjun Pei, Dejun Sun, Yuling Tang, Haoyan Wang, Changhui Wang, Dexi Wang, Limin Wang, Guangfa Wang, Xuefen Wang, Liping Wei, Bin Wu, Xixin Yan, Ting Yang, Lan Yang, Yang, Hongzhong Yang, Kejing Ying, Changhe Yu, Xiangyan Zhang, Min Zhang, Guohou Zhao, Xiangdong Zhou, Huili Zhu, Lei Zhu, Shuyang Zhu

Czech Republic

Masroor Ali, Michal Bláha, Jana Bursová, Martina Čmakalová, Josef Frátrik, Elena Güttlerová, Stanislav Holub, Daniela Kopecká, Jaromír Musil, Ilona Pavlišová, Alexandra Popelková, Lenka Povýšilová, Jolana Presperinova, Josef Veverka, Jiří Votruba, Jiří Vytiska

France

Alain Boye, Ari Chaouat, Gilles Devouassoux, Dominique Lejay, Mathieu Larrousse, Charles Le Merre, Hervé Pegliasco

Germany

Sabine Ballenberger, Robert Bals, Ekkehard Beck, Peter Berg, Burkhard Bewig, Hans Christian Blum, Lutz Bollmann, Hans Brüggen, Falk Brunner, Stephan Budweiser, Klaus Dalhoff, Andres de Roux, Regina Deckelmann, Andreas Deimling, Rolf Dichmann, Frank Eberhardt, Tamara Eckermann, Martin Ehlers, Andreas Eich, Moritz Erlinger, Guido Ern, Jan Feimer, Frank Feldmeyer, Karin Forster, Andreas Forster, Karl Heinz Franz, Andreas Fritzsche, Christian Gessner, Thomas Ginko, Peter-Uwe Haase, Peter Hammerl, Hans Peter Hauber, Stefan Heindl, Gerd-Ulrich Heinz, Rudolf Hennig, Peter Hofbauer, Martin Hoffmann, Gerhard Hoheisel, Gabriele Illies, Margret Jandl, Matthias John, Thomas Jung, Josef Junggeburth, Frank Käßner, Frank Kanniess, Claus Keller, Joachim Kirschner, Anne-Marie Kirsten, Uwe Kleinecke-Pohl, Andrea Koch, Hans-Joachim König, Stephanie Korn, Claus Kroegel, Matthias Krull, Petra Kühne, Evelin Liefring, Anneliese Linnhoff, Andrea Ludwig-Sengpiel, Matthias Luttermann, Silke Mronga, Ingomar F. K. Naudts, Ruth Nischik, Axel Overlack, Ronald Redlich, Peter Ruckert, Heiner Saueressig, Axel Schaefer, Christian Schäfer, Lennart Schaper, Isabelle Schenkenberger, Volker Schlegel, Christian Schlenska, Olaf Schmidt, Gerhard Scholz, Michael Sebert, Helena Sigal, Heiner Steffen, Inga Steinebach, Christoph Stolpe, Liebhild Stratmann, Harald Sudhoff, Hilke Temme, Lutz Volgmann, Volker von Behren, Christian von Mallinckrodt, Jürgen Wachter, Matthias Waltert, Sabina Wehgartner-Winkler, Joachim Weimer, Tobias Welte, Jörg Winkler, Stefan Zielen

Hungary

Anna Bartha, Ildikó Breining, Valéria Csajbók, Katalin Gömöri, Zsuzsanna Mark, Imre Mészáros, Lajos Molnár, János Mucsi, Eva Radeczky, Marianna Rakvács, Judit Schlezák, Mihály Tímár, Erika Unger, István Várkonyi, Hilda Zibotics

Italy

Elena Bacci, Bianca Beghé, Francesco Blasi, Damiano Capaccio, Mauro Carone, Stefano Centanni, Isa Cerveri, Alfredo Chetta, Giuseppe Fiorentino, Antonio Foresi, Maria Pia Foschino Barbaro, Francesco Mazza, Stefano Nava, Paolo Palange, Giovanni Passalacqua, Claudio Pedone, Stefano Picciolo, Massimo Pistolesi, Renato Prediletto, Paola Rogliani, Alessandro Sanduzzi Zamparelli, Nicola Scichilone, Antonio Spanevello

Japan

Tomoki Kimura, Masaharu Kinoshita, Hiroyuki Nakamura, Hiroyuki Ohbayashi, Hironori Sagara, Yuji Tohda, Takashi Yamada, Eiji Yamagata

Mexico

Moisés Acuña, Sylvia Colmenero, Dante Hernández, Luis Natera, Alejandra Ramírez Venegas, Alicia Ramírez, Jorge Rosas, Héctor Glenn Valdez López

Netherlands

Mazin Alhakim, Paul Bresser, Robert Costongs, Frank Custers, Remco Djamin, Michiel Eijsvogel, Steven Gans, Martijn Goosens, Christian Melissant, Sunil Ramlal, M.A. Tiemessen, Hans Timmer, Vivienne van de Walle, Jan Willem van den Berg, Maarten van den Berge, Pascal Wielders

New Zealand

Benedict Brockway, Simon Carson, Catherina Chang, Paul Dawkins, Michael Epton, John Kolbe, Paul Noonan, Dean Quinn, John Richmond, Andrew Veale

Peru

José Cabrera, Oscar Carbajal, Socorro Castro, William Chávez, Octavio Cubas, Silvia Giovana del Pilar Cubas Durango, Javier Diaz, Rolando Estrella, Maria Faverio, Ronal Gamarra, Alfredo Guerreros, Carlos Iberico, Alberto Matsuno, Ramon Mendoza, Fernando Roberto Rodriguez Chariarse, Danilo Salazar, César Villarán, José Antonio Zaga Ortega

Poland

Adam Śmiałowski, Anna Biełous-Wilk, Anna Bogusz, Kornelia Ciekalska, Krystyna Folcik, Hanna Gęsińska, Tomasz Kaziród, Violetta Łabij, Danuta Mądra-Rogacka, Bernadetta Majorek-Olechowska, Janusz Milanowski, Robert Mróz, Władysław Pierzchała, Robert Staniszewski, Marzenna Tarnowska-Matusiak, Sławomir Tokarski

Russia

Elena Alexeeva, Gregory Arutyunov, Alexander Averyanov, Alexander Bezlepko, Natalia Galvas, Mikhail Ilkovich, Sergey Mikhailov, Elena Ovchinnikova, Valery Podzolkov, Elena Polkanova, Irina A. Semenova, Natalia Shaporova, Vasilii Trofimov, Elena Vishneva, Anna Zateyshchikova

Serbia

Ivan Cekerevac, Vesna Dopudja Pantic, Aleksandra Ilic, Ivan Kopitovic, Zorica Lazic, Ivana Mikavica, Branislava Milenkovic, Dragica Mirkovic, Natasa Petrovic-Stanojevic, Gorana Sovljanski, Ivana Stankovic, Ana Stojanovic, Miodrag Vukcevic, Vladimir Zugic, Dejan Zujovic, Biljana Zvezdin

South Africa

Ismail Aboobaker Abdullah, Ismail Abdullah, Paul Abrahams, Luthando Adams, David Bernhardi, Johannes Breedt, Nazira Carrim-Ganey, Douwe De Jong, Nyda Fourie, Elvis Irusen, Dirkie Janse van Rensburg, Umesh Lalloo, Johan Lombaard, Hannelie Lottering, Ismael Mitha, Essack Mitha, Mohamed Salim Mookadam, Murimisi Mukansi, Chantal Muller, Haylene Nell, Trevenesan Padayachee, Gerhard Ras, Danelle Richter, Susanna Roux, Saadiya Seedat, Mohammed Tayob, Michael van der Linden, Eugene Van der Walt, Albie Van Zyl, Richard van Zyl-Smit, Marianne Elizabeth Viljoen, Agatha Wilhase

South Korea

Joong Hyun Ahn, Min Kwang Byun, Jung Hyun Chang, Jaehwa Cho, Hee Soon Chung, Seung Joon Kim, Yee Hyung Kim, Kwan Ho Lee, Sang Yeub Lee, Yong Chul Lee, Sang Haak Lee, Sung Soon Lee, Sung-Chul Lim, Seong Yong Lim, Choon Sik Park, Seungsoo Sheen, Jae Jeong Shim, Soojung Um, Sukjoong Yong, Chul-Gyu Yoo, Kwang Ha Yoo, Hyoungkyu Yoon, Hoil Yoon, Ho Kee Yum

Spain

Macarena Arroyo, Ferran Barbé, Miguel Barrueco, Albert Boada Valmaseda, José Luis de la Cruz Ríos, Luis de Teresa, José Echave-Sustaeta, Xavier Farrès, Antonio Ferrer, Juan Luis García Rivero, Javier Hueto Pérez de Heredia, José María Ignacio García, Jordi Juanola Pla, Luis Lores, José María Marin, Pyrene Martínez, Carlos Martínez Rivera, Luis Mateos Caballero, Fernando Molina Nieto, Eduard Monsó, Elsa Naval Sendra, Jacinto Ortiz, Juan Ortiz de Saracho Bobo, Mercè Pérez Vera, Germán Peces-Barba, Luis Puente-Maestu, David Ramos Barbón, Ana Rañó, José Miguel Rodriguez González-Moro, Juan Roldán, Fernando Sánchez-Toril López, Antonio Santa Cruz, José Gregorio Soto Campos, Zoran Stojanovic, Pere Torán-Monserrat, José Luis Velasco Garrido

Sweden

Johan Berglund, Ulla-Britt Ericsson, Pekka Koskinen, Anders Luts, Åke Olsson, Stefan Rustscheff, Ines Vinge

Taiwan

Shih-Lung Cheng, Wen-Feng Fang, Liang-Wen Hang, Jeng-Yuan Hsu, Ming-Shyan Huang, Yu Chih Liu, Yu-Feng Wei, Cheng Ta Yang, Chong-Jen Yu

United Kingdom

Joshua Asubiaro, Ronnie Beboso, Mark Blagden, John Calvert, Gourab Choudhury, Rebecca Clark, David Collier, Anthony Gunstone, John Hurst, Paul Ivan, Venkata Kondagunta, Brian Leaker, Ravi Mahadeva, Gerry McKaig, Damien McNally, Monica Nordstrom, Babatunde Oyesile, Janice Patrick, Elizabeth Sapey, Dinesh Saralaya, Rex Sarmiento, Dave Singh, Johnston Stewart, Usha Sukumaran, Amrit Takhar, Alice Turner, Dennis Wat, Jadwiga Wedzicha, Mark Wilkinson, Andrew Wilson

United States

Chandar Abboy, Ladly Abraham, Roger Abrahams, Amable R. Aguiluz, Bassil Aish, Mohamed Ali, Karen Allen, Gregory Allen, Jr., William F. Alleyne, II., Jose Alvarez, Rajasekaran Annamalai, Rami Arfoosh, Ahmed Arif, Samir Arora, Robby Ayoub, Francisco Badar, Anil Badhwar, Peggy Barnhill, Glenn Beard, Richard Beasley, George Bensch, Peter Bercz, Larry Berman, David Bernard, David I. Bernstein, Saligrama Bhat, Maria Blahey, Joseph A. Boscia, III., Susan Bottone, Cynthia Bowman-Stroud, Shari Brazinsky, Gregory Brooks, Barry Buffman, John R. Burk, Robert Buynak, William Byars, James Cain III, Ronald Caldwell, Robert Call, Jose F. Cardona, James Carswell, Ivan Castano, Ravi Chandran, Christopher Chappel, Eric Chenworth, Kenneth Chinsky, Jim Christensen, Bertrand Cole, Jeremy Cole, Clinton Corder, Bruce Corser, Maria C. Cubillas, Richard Cutchin, Nizar Daboul, Jerome Daniel, Timothy Dao, Enrique Davis, Gilda M. De La Calle, Luis De La Cruz, Ronald DeGarmo, Samuel DeLeon, Michael B. Denenberg, Sandip Desai, Ernesto Diaz, Jose Diaz, Robert Dillon, Calvin Dixon, Leonard Dunn, Hugh Durrence, Timothy Elder, Kirk Elliott, David Erb, Neil Ettinger, Emeka Eziri, Faisal Fakih, Richard Fei, Gregory Feldman, Jeremy Feldman, Gary Ferguson, Maria Fernandez, Herbon Fleming, Charles Fogarty, Rodney Folz, Miguel Franco, George Freeman, Stephen Fritz, David Fuentes, Gregory Funk, Nashwa Gabra, Bernard Garcia, Francisco Garcia, Robert Garver, Padmaja Reddy Gayam, Arthur Gelb, Marcy Goisse, William Gonte, Learned Gonzales, Robert Gordon, Brian Gotkin, Gregory Gottschlich, Kanakadurga Govindaraju, Richard Gower, Joseph Graif, Benny Green, Gary Greenwald, Darin Gregory, Kevin Grullon, Sridhar Guduri, Giancarlo Guido, Tarsem Gupta, James Haaksma, Ghassam Hadi, Kenneth S. Haft, Michael Hagan, Athir Hajjar, Gregory Hammond, Hoadley Harris, Aaron Hartman, Ernest Hendrix, Carlos Herrera, Mitzie Hewitt, Albrecht Heyder, Jeffrey Hirschfield, James Hitchcock, John Holcomb, Donald Howard, Cathy Hurley, Iftikhar Hussain, Thomas Hyers, Abraham Ishaaya, Younus Ismail, Richard Jackson, Joshua Jacobs, Ajay Jain, Ahmad Jalloul, Mikell Jarratt, Frank Johnson, Stephen Jones, Rebecca Jordan, John Joseph, Kishor Joshi, Thomas Kaelin, Monroe Karetzky, Najmuddin Karimjee, Robert Kaufmann, Mitchell Kaye, Edward Kerwin, Ahtaram Khan, Sohail Khan, Yasuko Kidokoro, Jeffrey Kingsley, Gaylon Kipp, Ryan Klein, Andras Koser, Raymond Kovalski, James Krainson, Camil Kreit, Kannappan Krishnaswamy, Shana Krstevska, Ted Kubicki, Steven Kulback, Amrendra Kumar, Sanjeev Kumar, Ritsu Kuno, Mitchell Kuppinger, Shahrukh Kureishy, Joseph Labuda, Albert Lai, Mitchell Douglas Lee, Lawrence Levinson, Vadim Leyenson, Joseph Lillo, Govinda Lohani, Aslam Loya, Peter Lutz, Lon Lynn, Sharan Mahal, Naveed Mahfooz, Sashi Makam, Rickey Manning, Lyndon Mansfield, Abbas Mansour, Hipolito Mariano, David Marks, Jennifer Martin, Mark Martin, Rafael Martinez, Cindy Martinez, Maria Mascolo, Samuel Mast, Jonathan Matz, David Maybee, James McDonnell, Isaac Melamed, Curtis Mello, Eric Melvin, Kevin Merkes, Peter Meyers, Bernard Michlin, Magdy Mikhail, Thomas Minor, Gowdhami Mohan, Jason Mohr, Elizabeth Mones, Anthony Montanaro, Eliot Moon, Emily Morawski, Nabil Morcos, Courtney Morgan, Timothy Moriarty, Scott Morin, Nidal Morrar, Felix Morris, Alvaro Murcia, Alexander Murray, Murtaza Mussaji, Salil Nadkarni, Pedro Nam, Cheta Nand, Ikeadi Maurice Ndukwu, Brooke Nevins, Rachel Nisbet, Timothy L. Norcross, Daniel Norman, Thomas Nugent, Lazaro Nunez, David Nyanjom, Dany Obeid, Onwura Obiekwe, Thomas O’Brien, Joseph Ojile, Ikechi Okwara, Robert Onder, Robert Orr, David Ostransky, Raidel Oviedo, Ward Paine, Mikhail Palatnik, Michael Palumbo, Jorge Paoli-Bruno, Kaushik Patel, Amit Patel, Nancy L.S. Patel, Christopher Perry, Shirin Peters, Michael Pfeffer, David Pham, Jean Philippe, Regina Pillai, Jennifer M. Piwowarski, Joe Pouzar, Bruce Prenner, Kevin Pritchett, Raman Purighalla, George Pyrgos, Syed Rahman, William Randall, Orlando Rangel, Syed Rehman, Robert Remler, A Lynn Ridgeway, Ernie Riffer, Eustace Riley, Clifford Risk, Demetrius Rizos, Emory Robinette, Lilia Rodriguez, Hugo Romeu, Bridgett Ronan, Shari Rozen, Timothy Rummel, Gary Ruoff, Eugene Ryan, Stephen Ryan, Boris Sagalovich, Satinder Saini, Robert Salazar, Victor Salcedo, D. Andrew Sams, Mercedes Samson, Jay Sandberg, William Sargeant, Alan Schecter, Eric Schenkel, Jay Schmidt, C. Andrew Schroeder, James Schultz, Jeffrey Scott, Anita Scribner, Michael Seep, Sudhir Sehgal, Sudhir Sekhsaria, Jean-Louis Selam, John Sensenbrenner, Marvin Sexton, Amit Shah, Suresh Shah, Ajit Shah, Vipul Shah, Ranjan Shah, Heena Shah-Patel, Paul Shapero, Timothy Shepherd, Lawrence Sher, James R Shoemaker, John Scott Sibille, Ather Siddiqi, Barry Sigal, Vinay Sikand, Thomas Siler, William Sims, Naresh Singh, David Smith, Stephen Keith Smith, Clark Soderlund, Royce Solano, Guillermo Somodevilla, Weily Soong, John Southard, Clyde Southwell, Selwyn Spangenthal, Roy St. John, Thomas Stern, Asha Stern, Gary Steven, Vijay Subramaniam, Sever Surdulescu, Ricardo Tan, Tonny Tanus, Horia Tatu, Gilbert Teixeira, Alan Thomas, Charles Thompson, Ernest Thompson, Jr, Letitia Thompson-Hargrave, Craig Thurm, Raymond Tidman, Louis A. Torres Jr., Rodolfo Trejo, Miguel Trevino, Barry Troyan, John Updegrove, Sanjay Vadgama, Sunil Verma, Manuel Villareal, Srinivas Vodnala, Andrew Wachtel, George Walker, Luke Webb, Dave E Webster, Paul Weinberg, Terry Wells, Howard Wenocur, Jan Westerman, Bram Wieskopf, Hugh Windom, David Winslow, Robert Wise, Bryan Wolf, Patrick Wright, Daniel Yang, Douglas Young, Thomas Yunger, Irfanullah Yusufzai, Zahid Zafar, Joseph Zaky

2. METHODSTrial Procedures

For budesonide/glycopyrrolate/formoterol 320/18/9.6 µg, budesonide/glycopyrrolate/formoterol 160/18/9.6 µg, glycopyrrolate/formoterol 18/9.6 µg, and budesonide/formoterol 320/18/9.6 µg, the doses of 18 µg glycopyrrolate and 9.6 µg formoterol fumarate are equivalent to 14.4 µg of glycopyrronium and 10 µg of formoterol fumarate dihydrate, respectively. Patients were instructed to take two inhalations in the morning and two in the evening (approximately 12 hours apart). Doses represent the sum of two inhalations (i.e., the doses per actuation were 160 or 80 µg budesonide, 9 µg glycopyrrolate, and 4.8 µg formoterol fumarate, resulting in a total daily dose of 640 or 320 µg budesonide, 36 µg glycopyrrolate, and 19.2 µg formoterol fumarate).

Exacerbations

A COPD exacerbation was defined as a change in the patient’s usual COPD symptoms that lasted 2 or more days, was beyond normal day-to-day variation, was acute in onset, and may have warranted a change in regular medication. The change in symptoms must have included at least one major COPD symptom and at least one other major or minor symptom from the list below:

· Major COPD symptoms: dyspnea, sputum volume, and sputum color

· Minor COPD symptoms: cough, wheeze, sore throat, cold symptoms (rhinorrhea or nasal congestion), and fever without other cause.

All moderate or severe COPD exacerbations were captured using the COPD Exacerbation electronic case report form. Mild COPD exacerbations were captured based on symptoms as recorded by the patient in an electronic diary. COPD exacerbations of any severity were considered to be expected trial endpoints and were not reported as adverse events unless considered a serious adverse event.

The investigator was required to justify the decision for defining the event as an exacerbation, and record it in the electronic case report form if symptoms were acute or progressed rapidly and required treatment less than 2 days from onset of symptoms, or if a patient’s symptoms and the overall clinical findings supported the diagnosis of a COPD exacerbation, but the patient did not experience a worsening of at least one major COPD symptom and at least one other major or minor symptom.

Pulmonary Function Tests

All pulmonary function tests were performed in accordance with American Thoracic Society criteria.1 All trial staff responsible for performing pulmonary function testing received standardized training, and all technicians were required to demonstrate proficiency in the use of the equipment and the ability to perform technically acceptable pulmonary function tests prior to performing testing on trial patients. Spirometry assessments were required to meet protocol-specified acceptability and repeatability criteria.

At Visit 1, a single spirometry assessment was conducted. At Visits 2 and 3, spirometry was conducted 60 and 30 minutes prior to bronchodilator administration and at 30 minutes post-bronchodilator. For patients participating in the 4-hour pulmonary function sub-study, additional spirometry assessments were conducted at Visits 4, 5, 7, 10, 12, and 14 (Day 1 and Weeks 4, 12, 24, 36, and 52). At these visits, spirometry was conducted at 60 and 30 minutes prior to trial drug administration and at 15 and 30 minutes, and 1, 2, and 4 hours after trial drug administration. An additional spirometry assessment was conducted at 5 minutes after trial drug administration at Visit 4 (Day 1) only.

Pneumonia

To adequately assess and characterize the risk of pneumonia in patients in a nonbiased manner, an external clinical endpoint committee (CEC) reviewed all adverse events reported as pneumonia to ensure appropriate pre-defined and clinically consistent pneumonia criteria were met.

A clinically consistent definition of pneumonia was implemented, requiring all of the following criteria:

1. Clinical diagnosis of pneumonia by the investigator

2. Documentation of chest imaging obtained within 14 days of the diagnosis of pneumonia that was compatible with the diagnosis of pneumonia

3. Treatment with antibiotics (and/or appropriate antiviral and/or antifungal agents)

4. At least two of the following clinical signs, symptoms, or laboratory findings: increased cough, increased sputum purulence or production, adventitious breath sounds on auscultation, dyspnea or tachypnea, fever, elevated white blood cell counts, and hypoxemia.

The CEC could request any additional information, including copies of chest X-rays or CT scans, if needed, to confirm the pneumonia diagnosis. Radiographs were evaluated locally, and the results (infiltrate compatible with pneumonia) were documented within the source document at the sites. If the investigator became aware that a diagnosis of pneumonia was made without a chest image having been performed, he or she was to obtain a chest X-ray (frontal and lateral) up to 10 to 14 days after the date of pneumonia diagnosis.

Statistical Analysis

Sample Size

The probability of demonstrating differences for 320 µg-budesonide triple therapy versus both dual therapies with a P-value of <0.005 was approximately 78% (87% for each comparison). These calculations assumed yearly rates of moderate or severe exacerbations of 1.14, 1.21, 1.34, and 1.34 with 320 µg-budesonide triple therapy, 160 µg-budesonide triple therapy, glycopyrrolate/formoterol, and budesonide/formoterol, respectively. The sample size was increased from 8000 to 8400 following a blinded sample size re-estimation.2

Populations and Estimands

The intent-to-treat (ITT) population included all patients who were randomized and received any amount of trial treatment. Patients were analyzed according to randomized treatment group, and efficacy data obtained after discontinuation of treatment but prior to withdrawal from the trial were included. The modified ITT (mITT) population was a subset of the ITT population that included all patients with post-randomization data obtained prior to discontinuation from treatment. Any data collected after completion of, or discontinuation from, randomized trial medication was excluded from the mITT analysis but included in the ITT analysis. The safety population was similar to the mITT population, but patients were analyzed according to treatment received rather than treatment randomized, and patients with no post-dose safety assessments were excluded. The per-protocol population was a subset of the mITT population defined as all patients with post-randomization data obtained prior to any major protocol deviations and was used for non-inferiority analyses. The primary estimand of interest was the efficacy estimand, which assumed continuation of randomized treatments for the trial duration, regardless of actual compliance. A secondary analysis of the primary endpoint was performed using the attributable estimand, which accounted for patients who discontinued treatment because of lack of efficacy or tolerability by imputing missing data. The treatment policy estimand, i.e. the effect of randomized treatment over the study period regardless of whether randomized treatment was continued, was assessed using the ITT population, in which all observed data were utilized regardless of whether patients remained on randomized treatment.

Type I Error Control

All treatment comparisons were made for superiority, except for 160 µg-budesonide triple therapy to budesonide/formoterol, which was for non-inferiority followed by superiority. However, attaining statistical significance in the superiority comparison was not a pre-requisite to proceeding down the testing hierarchy (Figure S1).

This trial had a planned interim efficacy analysis overseen by an external data monitoring committee. Because the criteria for unequivocal efficacy were not met for the interim analysis, and the trial continued until all participants completed, the one-sided α criteria for the final analysis was adjusted based on the percentage of the information available at the interim analysis such that the overall Type I error was controlled at a one-sided α of 0.025. The final analysis one-sided critical level of significance was 0.023, thereby making the two-sided level 0.046.

For both registration approaches, US and ex-US, and for each dose of triple therapy, if the primary and first secondary measures were statistically significant for all comparisons, then Type I error for the remainder of the secondary measures with the exception of time to death (all-cause) was to be controlled by using Hochberg within each comparison.3 If all of the secondary measures were significant, then time to death (all-cause) was to be tested within each comparison.

Non-Inferiority Margins

For the comparison of 160 µg-budesonide triple therapy to budesonide/formoterol, the non-inferiority margins were as follows (all for two-sided 95% CIs): for exacerbation endpoints, a rate ratio of 1.1 for the upper bound; for Transition Dyspnea Index (TDI) focal score, –0.75 units for the lower bound; for rescue medication use, 0.75 puffs/day for the upper bound; for St. George’s Respiratory Questionnaire (SGRQ), 3.0 units for the upper bound; for SGRQ responders, 10 percentage points for the upper bound; for the EXAcerbations of Chronic pulmonary disease Tool (EXACT) total score, 1.5 units for the upper bound.

Analysis Methods

The rate of moderate or severe COPD exacerbations and the rate of severe exacerbations were analyzed using negative binomial regression, adjusting for baseline post-bronchodilator percent predicted FEV1 and log baseline blood eosinophil count as continuous covariates and baseline COPD exacerbation history, ICS use at screening, and region (US and Canada; Asia; Western Europe; Eastern Europe; Mexico, Central America, and South America; Australia, New Zealand, and South Africa) as categorical covariates. Time at risk was used as an offset variable. The Cox regression model for time to first moderate or severe COPD exacerbation included the same covariates.

TDI focal score and SGRQ total score were analyzed using linear repeated measures analysis of covariance (ANCOVA) models including treatment, visit, and the treatment by visit interaction, and ICS use at screening as categorical covariates and log baseline blood eosinophil count, baseline score, baseline post-bronchodilator percent predicted FEV1, and percent reversibility to bronchodilator as continuous covariates. The mean change from baseline in EXACT total score was analyzed using a similar repeated measures model as for TDI, but using the corresponding baseline mean score instead of the Baseline Dyspnea Index as a covariate.

Change from baseline in rescue medication use was analyzed using a linear repeated measures ANCOVA model including treatment, 4-week time interval, the treatment by time interval interaction, and screening ICS use as categorical covariates and baseline post-bronchodilator percent predicted FEV1, baseline rescue medication use, log baseline blood eosinophil count, and percent reversibility to bronchodilator as continuous covariates.

For the SGRQ responder analysis, logistic regression was used to compare the treatment groups with baseline SGRQ score, log baseline blood eosinophil count, and baseline post-bronchodilator percent predicted FEV1 and percent reversibility to bronchodilator as continuous covariates and treatment, and ICS use at screening as categorical covariates.

Time to death was analyzed using a Cox regression model, adjusting for baseline post-bronchodilator percent-predicted FEV1 and baseline age as covariates.

Analyses of the rate of moderate or severe exacerbations according to eosinophil count were performed using generalized additive modeling, combining nonparametric regression for the relationship of eosinophil levels to response with a negative binomial model.

Figure S1. Order of hypothesis testing for Type I error control: US and ex-US approaches.

All comparisons of BGF MDI 160 versus BFF MDI are for non-inferiority using the per-protocol estimand followed by superiority. Superiority is not required to advance to the next comparison. All other comparisons are for superiority and use the efficacy estimand unless otherwise stated. Endpoints are at Week 24 for the US approach and over 24 weeks for the ex-US approach wherever applicable (lung function, SGRQ score, and symptoms). Hochberg-controlled secondary endpoints were the time to first moderate or severe COPD exacerbation, rescue medication use, rate of severe COPD exacerbation, SGRQ total score (ex-US), SGRQ responders (US), EXACT total score (ex-US), and TDI focal score (ex-US). BFF MDI denotes budesonide/formoterol fumarate metered dose inhaler, BGF MDI 160 budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler 160/18/9.6 µg, BGF MDI 320 budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler 320/18/9.6 µg, COPD chronic obstructive pulmonary disease, exac. exacerbations, EXACT EXAcerbations of Chronic pulmonary disease Tool, GFF MDI glycopyrrolate/formoterol fumarate metered dose inhaler, SGRQ St. George’s Respiratory Questionnaire, TDI transition dyspnea index.

Figure S2. Forest plots for time-to-first event analyses.

Panel A shows the time to first moderate or severe COPD exacerbation in the modified intent-to-treat population (efficacy estimand), and Panel B the time to death (all-cause) in the intent-to-treat population (treatment policy estimand). Error bars represent 95% confidence intervals. Hazard ratios are plotted on a logarithmic scale. BFF MDI denotes budesonide/formoterol fumarate metered dose inhaler, BGF MDI budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler, COPD chronic obstructive pulmonary disease, GFF MDI glycopyrrolate/formoterol fumarate metered dose inhaler.

Figure S3. Change from baseline in SGRQ total score over 52 weeks (efficacy estimand; modified intent-to-treat population).

The I bars represent standard errors. Total scores on the SGRQ range from 0 to 100, with lower scores indicating better health-related quality of life; the minimum clinically important difference is 4 units4. BFF MDI denotes budesonide/formoterol fumarate metered dose inhaler, BGF MDI budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler, GFF MDI glycopyrrolate/formoterol fumarate metered dose inhaler, SGRQ St George’s Respiratory Questionnaire.

Figure S4. Rate of moderate or severe COPD exacerbations by eosinophil subgroups (efficacy estimand; modified intent-to-treat population).

Error bars represent 95% confidence intervals. Hazard ratios are plotted on a logarithmic scale. BFF MDI denotes budesonide/formoterol fumarate metered dose inhaler, BGF MDI budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler, COPD chronic obstructive pulmonary disease, GFF MDI glycopyrrolate/formoterol fumarate metered dose inhaler.

Figure S5. Rate of moderate or severe COPD exacerbations by blood eosinophil count (modified intent-to-treat population).

Banded areas indicate 95% confidence intervals. BFF MDI denotes budesonide/formoterol fumarate metered dose inhaler; BGF MDI budesonide/glycopyrrolate/formoterol fumarate metered dose inhaler; COPD chronic obstructive pulmonary disease; GFF MDI glycopyrrolate/formoterol fumarate metered dose inhaler.

Table S1. Pre-specified endpoints in the ETHOS trial protocol.

Endpoints

Reported in this manuscript or supplement

Not reported in this manuscript or supplement

Primary endpoint

Rate of moderate or severe COPD exacerbations

X

Secondary endpoints

Time to first moderate or severe COPD exacerbation

X

Change from baseline in average daily rescue medication use over 24 weeks

X

TDI focal score over 24 weeks (ex-US)

X

Change from baseline in EXACT total score over 52 weeks (ex-US)

X

Change from baseline in SGRQ total score over 24 weeks (ex-US)

X

Percentage of participants achieving an MCID of 4 units or more in SGRQ total score at Week 24 (US)

X

Time to death (all-cause)

X

Rate of severe COPD exacerbations

X

Other endpoints

Rate of COPD exacerbations of any severity

X

Time to first COPD exacerbation of any severity

X

Time to first severe COPD exacerbation

X

Change from baseline in average daily rescue medication use over 52 weeks

X

Time to death (respiratory)

X

Time to treatment failure (treatment discontinuation for any cause, moderate or severe exacerbation, or death)

X

Change from baseline in the EXACT total score over 24 weeks and over each 4-week interval of the 52-week treatment period

X

Change from baseline in the E-RS total score over 24 weeks, over 52 weeks, and over each 4-week interval of the 52-week treatment period

X

Percentage of days with “no rescue medication use”

X

TDI focal score over 24 weeks (US)

X

TDI focal score over 52 weeks, and at each post-randomization visit

X

Percentage of participants achieving an MCID threshold of 1 unit or more on average over 24 weeks in TDI focal score

X

Percentage of participants achieving an MCID threshold of 1 unit or more on average over 52 weeks in TDI focal score

X

Change from baseline in SGRQ total score over 52 weeks, and at each post-randomization visit

X

Percentage of participants achieving an MCID of 4 units or more in SGRQ total score at Week 52

X

Percentage of participants achieving an MCID of 4 units or more in SGRQ total score over 52 weeks and over 24 weeks

X

EQ-5D-5L variables, including the EQ-5D index score, the EQ-5D VAS, and each of the five-dimension single item 5-level responses at each post-randomization visit.

X

Pulmonary function tests sub-study endpoints

Primary endpoints

Change from baseline in morning pre-dose trough FEV1 at Week 24 (US) for the comparisons of triple therapies to glycopyrrolate/formoterol

X

Change from baseline in morning pre-dose trough FEV1 over 24 weeks (ex-US) for the comparisons of triple therapies to glycopyrrolate/formoterol

X

FEV1 AUC0-4 at Week 24 (US) for the comparisons of triple therapies to budesonide/formoterol

X

FEV1 AUC0-4 over 24 weeks (ex-US) for the comparisons of triple therapies to budesonide/formoterol

X

Other endpoints

Change from baseline in morning pre-dose trough FEV1 over 24 weeks, over Weeks 12 to 24, over 52 weeks and at each post-randomization visit

X

FEV1 AUC0-4 over 24 weeks, over Weeks 12 to 24, over 52 weeks and at each post-randomization visit where measured

X

Peak change from baseline in FEV1 over 24 weeks, over Weeks 12 to 24, over 52 weeks and at each post-randomization visit where measured

X

Rate of decline in pre-dose FEV1 over 52 weeks

X

Rate of decline in FEV1 AUC0-4 over 52 weeks

X

Time to onset of action on Day 1

X

24-h Holter monitoring sub-study endpoints

Primary endpoint

Change from baseline in mean heart rate averaged over 24 hours

X

Secondary endpoints

Change from baseline in mean nighttime (22:00 to 06:00) and daytime (06:00 to 22:00) heart rate

X

Change from baseline in the maximum 24-hour heart rate

X

Change from baseline in the minimum 24-hour heart rate

X

Change from baseline in the frequency of isolated ventricular ectopic events (including a single PVC)

X

Change from baseline in the frequency of ventricular couplets (defined as two PVCs preceded or followed by regular beats)

X

Change from baseline in the frequency of ventricular runs (defined as three or more PVCs preceded or followed by regular beats)

X

Change from baseline in the frequency of supraventricular ectopic beats

X

Change from baseline in the frequency of isolated supraventricular ectopic events

X

Change from baseline in the frequency of supraventricular couplets

X

Change from baseline in the frequency of supraventricular runs

X

Incidence of withdrawal criteria being met during 24-hour Holter monitoring

X

Incidence of atrial fibrillation with a rapid ventricular response (>100 bpm)

X

Other endpoints

Proportion of participants with maximum heart rate >180, >160−180, >140−160, >120−140, >100−120, and 100 bpm or less

X

The proportion of participants with a minimum heart rate >60, >50−60, >40−50, and ≤40 bpm

X

The proportion of participants in each category of change from baseline in the number of PVCs per hour (no change, increase of >0−<60, 60−<120, and ≥120, and decrease of >0−<60, 60−<120, and ≥120)

X

Healthcare resource utilization assessments

X

AUC0–4 denotes area under the curve from 0 to 4 hours, bpm beats per minute, COPD chronic obstructive pulmonary disease, EQ-5D-5L EuroQol 5 Dimensions 5 Levels health questionnaire, ERS EXACT-Respiratory Symptoms, EXACT EXAcerbations of Chronic pulmonary disease Tool, FEV1 forced expiratory volume in 1 second, MCID minimal clinically important difference, SGRQ St. George’s Respiratory Questionnaire, PVC premature ventricular contraction, TDI Transition Dyspnea Index, VAS Visual Analog Score.

Table S2. Additional baseline demographics (modified intent-to-treat population).*

Budesonide/glycopyrrolate/formoterol

320/18/9.6 µg(N = 2137)

Budesonide/glycopyrrolate/formoterol

160/18/9.6 µg(N = 2121)

Glycopyrrolate/formoterol

18/9.6 µg(N = 2120)

Budesonide/formoterol

320/9.6 µg(N = 2131)

Race — no. (%)

White

Asian

Black

American Indian/Alaska Native

Other

1819 (85.1)

162 (7.6)

78 (3.6)

33 (1.5)

45 (2.1)

1783 (84.1)

166 (7.8)

88 (4.1)

40 (1.9)

44 (2.1)

1808 (85.3)

157 (7.4)

75 (3.5)

30 (1.4)

50 (2.4)

1816 (85.2)

166 (7.8)

64 (3.0)

39 (1.8)

46 (2.2)

Body mass index — kg/m2

27.6±6.2

27.5±6.3

27.6±6.2

27.1±6.2

COPD duration — yr

8.4±6.5

8.2±6.1

8.2±6.1

8.4±6.1

Reversibility to ipratropium

Change in FEV1 from pre-to post-ipratropium — mL

Reversibility ≥ 12% and ≥ 200

mL, n (%)

144.0±150.6

633 (29.6)

148.7±149.5

634 (29.9)

141.1±144.5

599 (28.3)

136.5±141.8

598 (28.0)

Cardiovascular medical history†

No. of patients evaluated

≥1 cardiovascular risk factor

Hypertension

High total cholesterol

Diabetes

2144

1511 (70.5)

1272 (59.3)

769 (35.9)

398 (18.6)

2124

1505 (70.9)

1252 (58.9)

769 (36.2)

400 (18.8)

2125

1492 (70.2)

1250 (58.8)

744 (35.0)

339 (16.0)

2136

1513 (70.8)

1251 (58.6)

761 (35.6)

358 (16.8)

*Plus–minus values are means ±SD. †Safety population. The doses of glycopyrrolate and formoterol fumarate are equivalent to 14.4 µg of glycopyrronium and 10 µg of formoterol fumarate dihydrate, respectively. COPD denotes chronic obstructive pulmonary disease, FEV1 forced expiratory volume in 1 second.

Table S3. Prior COPD-related inhaled therapies (safety population).

Budesonide/glycopyrrolate/formoterol320/18/9.6 µg(N = 2144)

Budesonide/glycopyrrolate/formoterol160/18/9.6 µg(N = 2124)

Glycopyrrolate/formoterol 18/9.6 µg(N = 2125)

Budesonide/formoterol320/9.6 µg(N = 2136)

All patients

(N = 8529)

PRN SABA and/or SAMA only

1 (<0.1)

1 (<0.1)

1 (<0.1)

1 (<0.1)

4 (<0.1)

Used MA only

LAMA

24 (1.1)

23 (1.1)

19 (0.9)

14 (0.7)

23 (1.1)

22 (1.0)

21 (1.0)

21 (1.0)

87 (1.0)

80 (0.9)

Used BA only

LABA

18 (0.8)

15 (0.7)

12 (0.6)

10 (0.5)

14 (0.7)

10 (0.5)

12 (0.6)

11 (0.5)

56 (0.7)

46 (0.5)

Used ICS only

9 (0.4)

6 (0.3)

4 (0.2)

7 (0.3)

26 (0.3)

Used MA and BA only

LAMA/LABA

393 (18.3)

315 (14.7)

366 (17.2)

278 (13.1)

383 (18.0)

286 (13.5)

401 (18.8)

305 (14.3)

1543 (18.1)

1184 (13.9)

Used ICS and BA only

ICS/LABA

681 (31.8)

663 (30.9)

695 (32.7)

672 (31.6)

677 (31.9)

660 (31.1)

706 (33.1)

672 (31.5)

2759 (32.3)

2667 (31.3)

Used ICS and MA only

ICS/LAMA

27 (1.3)

14 (0.7)

31 (1.5)

13 (0.6)

39 (1.8)

24 (1.1)

37 (1.7)

21 (1.0)

134 (1.6)

72 (0.8)

Used ICS, MA, and BA

ICS/LAMA/LABA

986 (46.0)

839 (39.1)

991 (46.7)

840 (39.5)

981 (46.2)

846 (39.8)

949 (44.4)

833 (39.0)

3907 (45.8)

3358 (39.4)

Did not use any ICS, MA, BA, or PRN SABA or SAMA

5 (0.2)

3 (0.1)

3 (0.1)

2 (<0.1)

13 (0.2)

Data are n (%). The components (e.g. ICS, MA, and/or BA) must have each been used for at least 30 days prior to screening. Scheduled (non-PRN) use of SAMA or SABA was included in MA and BA categories, respectively. BA denotes β2-agonist, COPD chronic obstructive pulmonary disease, ICS inhaled corticosteroid, LABA long-acting β2-agonist, LAMA long-acting muscarinic antagonist, MA muscarinic antagonist, PRN as needed, SABA short-acting β2-agonist, SAMA short-acting muscarinic antagonist.

Table S4. Secondary and other pre-specified symptom and health-related quality of life endpoints (efficacy estimand; modified intent-to-treat population).*

Budesonide/glycopyrrolate/formoterol320/18/9.6 µg(N = 2137)

Budesonide/glycopyrrolate/formoterol160/18/9.6 µg(N = 2121)

Glycopyrrolate/formoterol 18/9.6 µg(N = 2120)

Budesonide/formoterol320/9.6 µg(N = 2131)

Secondary endpoints

Change from baseline in average daily rescue medication use over 24 weeks†

No. of patients evaluated

Mean (SE)

1425

–1.2 (0.06)

1389

–1.0 (0.07)

1387

–0.7 (0.07)

1426

–0.8 (0.06)

320 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

–0.15 (–0.32 to 0.01)

—

–0.51 (–0.68 to –0.34)

<0.001

–0.37 (–0.54 to –0.20)

<0.001

160 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

—

–0.35 (–0.53 to –0.18)

<0.001

–0.22 (–0.39 to –0.05)

0.01

Percentage of patients achieving an increase of ≥ 4 units in SGRQ total score at Week 24

No. of patients evaluated

Responders, n (%)

2119

1068 (50.4)

2102

1024 (48.7)

2096

893 (42.6)

2122

949 (44.7)

320 µg-budesonide triple therapy versus comparators

Odds ratio (95% CI)

P-value

—

1.1 (0.9 to 1.2)

—

1.4 (1.2 to 1.5)

<0.001

1.2 (1.1 to 1.4)

<0.001

160 µg-budesonide triple therapy versus comparators

Odds ratio (95% CI)

P-value

—

—

1.3 (1.1 to 1.5)

<0.001

1.2 (1.0 to 1.3)

0.01

TDI focal score over 24 weeks

No. of patients evaluated

Mean (SE)

2044

1.3 (0.06)

2023

1.3 (0.06)

1983

0.9 (0.06)

2021

1.0 (0.06)

320 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

0.03 (–0.12 to 0.19)

—

0.40 (0.24 to 0.55)

<0.001

0.31 (0.15 to 0.46)

<0.001

160 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

—

0.37 (0.21 to 0.52)

<0.001

0.27 (0.12 to 0.43)

<0.001

Change from baseline in EXACT total score over 52 weeks

No. of patients evaluated

Mean (SE)

2126

–1.8 (0.18)

2109

–1.6 (0.18)

2105

–0.7 (0.19)

2120

–0.8 (0.18)

320 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

–0.21 (–0.70 to 0.28)

—

–1.14 (–1.64 to –0.65)

<0.001

–1.04 (–1.53 to –0.55)

<0.001

160 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

—

–0.93 (–1.43 to –0.44)

<0.001

–0.83 (–1.32 to –0.34)

0.001

Change from baseline in SGRQ total score over 24 weeks

No. of patients evaluated

Mean (SE)

2076

–6.5 (0.25)

2056

–6.2 (0.25)

2017

–4.9 (0.25)

2056

–5.1 (0.25)

320 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

–0.34 (–0.99 to 0.30)

—

–1.62 (–2.27 to –0.97)

<0.001

–1.38 (–2.02 to –0.73)

<0.001

160 µg-budesonide triple therapy versus comparators

Mean (95% CI)

P-value

—

—

–1.28 (–1.93 to –0.63)

<0.001

–1.04 (–1.68 to –0.39)

0.002

Other endpoints

Change from baseline in SGRQ total score at Week 52

No. of patients evaluated

Mean (SE)

1681

–6.4 (0.35)

1680

–6.0 (0.36)

1562

–4.5 (0.36)

1631

–4.9 (0.36)

320 µg-budesonide triple therapy versus comparators

Mean (95% CI)

—

–0.37 (–1.32 to 0.59)

–1.88 (–2.84 to –0.91)

–1.47 (–2.43 to –0.51)

160 µg-budesonide triple therapy versus comparators

Mean (95% CI)

—

—

–1.51 (–2.48 to –0.54)

–1.10 (–2.06 to –0.14)

Percentage of patients achieving an increase of ≥ 4 units in SGRQ total score at Week 52

No. of patients evaluated

Responders, n (%)

2119

937 (44.2)

2102

907 (43.1)

2096

766 (36.5)

2122

832 (39.2)

320 µg-budesonide triple therapy versus comparators

Odds ratio (95% CI)

—

1.0 (0.9 to 1.2)

1.4 (1.2 to 1.6)

1.2 (1.1 to 1.4)

160 µg-budesonide triple therapy versus comparators

Odds ratio (95% CI)

—

—

1.3 (1.2 to 1.5)

1.2 (1.0 to 1.3)

The means presented are least-squares means. *P-values are provided only for comparisons included in the testing hierarchy. †Rescue medication user population.

CI denotes confidence interval, EXACT EXAcerbations of Chronic pulmonary disease Tool, ICS inhaled corticosteroid,