This was a Phase IV, multinational, multicentre, double-blind, randomised, active- and placebo-controlled, complete block crossover study. After the screening visit, patients completed an incremental exercise test on a computer-driven cycle ergometer to evaluate their peak exercise response (Fig. 1). On a subsequent day they completed a training constant work-rate cycle ergometry test (at 80% of the maximum workload [Wmax] achieved in the incremental exercise test; see the supplement for additional detail). A dedicated cardiopulmonary exercise test manual was provided to each site to standardise the test and minimise inter-operator variability [24, 25]. At the end of a 7–10-day run-in period, eligible patients were randomised to one of six treatment sequences using a balanced-block randomisation scheme generated by the interactive response technology provider. Each sequence comprised three, 3-week treatment periods, with a 7–10-day washout between treatment periods. During the treatment periods, patients received BDP/FF/G 100/6/10 µg per actuation (Trimbow, Chiesi Farmaceutici SpA, Parma, Italy), BDP/FF 100/6 µg per actuation (Foster, Chiesi Farmaceutici SpA, Parma, Italy), or matching placebo, all administered as two inhalations twice daily via identical pressurised metered-dose inhalers. Patients, investigators, site staff and sponsor personnel were blinded to treatment for the duration of the study.

Fig. 1

Pre- and 2-h post-dose at the start and end (i.e., after 3 weeks) of each treatment period, patients undertook plethysmography and spirometry assessments. Plethysmography parameters included residual volume (RV), total lung capacity (TLC), RV/TLC ratio, and functional residual capacity (FRC). Spirometry, assessed with standardised spirometry equipment and a central reading service, comprised slow inspiratory manoeuvres to assess resting inspiratory capacity (IC), followed by forced manoeuvres to assess FEV1 and forced vital capacity (FVC). Pre-dose on Day 1 of each treatment period and post-dose after 3 weeks (in both cases after forced spirometry), patients completed a constant work-rate cycle ergometry test (at 80% of the Wmax of the incremental exercise test), during which dyspnoea and muscle fatigue were assessed using a modified Borg scale [26], and IC was measured (prior to initiation, every 2 min during loaded pedalling, and at the end of exercise). Inhaled salbutamol was permitted as rescue medication throughout the study (including during the run-in and washout periods) but not within 6 h prior to any spirometry or cycle ergometry assessment; patients recorded this rescue medication use daily.

All patients provided written informed consent prior to any study-related procedure. The study was approved by the independent ethics committees at each institution (listed in the supplement), and was performed in accordance with the Declaration of Helsinki, and Good Clinical Practice. The study was registered at ClinicalTrials.gov (NCT05097014, registered 27th October 2021). The protocol was amended three times; none of the amendments were substantial or impacted recruitment.

Participants

Adults ≥ 40 years of age, diagnosed with COPD ≥ 12 months prior to screening, with post-bronchodilator FEV1/FVC < 0.7 and FEV1 40–80% predicted were eligible for the study. Participants were hyperinflated (FRC ≥ 120% predicted [9, 11, 14, 17, 18, 20, 22, 23]), symptomatic (modified Medical Research Council dyspnoea scale ≥ 2), and were receiving mono- or dual inhaled maintenance COPD therapy at a stable dose for ≥ 3 months (a regular, scheduled short-acting β2-agonist or muscarinic antagonist, alone or in combination, was acceptable), which were to be suspended prior to the screening visit and for the overall study period. Exclusion criteria included known respiratory disorders other than COPD, an abnormal, clinically significant 12-lead electrocardiogram reading that may impact patient safety, unstable concurrent disease or any other disease/condition that may impact the efficacy or safety assessments, and a moderate or severe COPD exacerbation in the previous 3 or 12 months, respectively. The full list of inclusion and exclusion criteria is in the supplement, together with the required wash-out periods prior to the screening visit for maintenance COPD therapy.

Outcomes

The primary objective was to evaluate the effects of BDP/FF/G and BDP/FF vs. placebo in terms of change from baseline in 2-h post-dose IC, assessed using slow spirometry prior to constant work-rate cycle ergometry (i.e., resting IC) at Week 3 of treatment. The key secondary objectives were to evaluate the effect of BDP/FF/G and BDP/FF vs. placebo in terms of change from baseline at Week 3 in IC at isotime (i.e., dynamic IC) and in exercise endurance time during constant work rate cycle ergometry. Since patients completed two constant work-rate cycle ergometry tests in each treatment period (at the start [baseline], and after 3 weeks), isotime was defined as the shortest exercise endurance time achieved by a patient in either the baseline or Week 3 exercise test, and was derived separately for each treatment period.

Exploratory endpoints included:

BDP/FF/G vs. BDP/FF comparisons of the primary and key secondary endpoints.

Change from baseline at Week 3 in:

Pre-dose resting IC.

Pre-dose FEV1.

Pre-dose FVC.

Pre-dose and 2-h post-dose FRC.

Pre-dose and 2-h post-dose RV.

Pre-dose and 2-h post-dose RV/TLC ratio.

Dyspnoea intensity at isotime (using the modified Borg scale).

Percentage of rescue medication-free days over the 3-week treatment period.

In addition, post-hoc analyses were performed on pre-dose and 2-h post-dose TLC. Safety and tolerability were assessed throughout the study in terms of the occurrence of adverse events, and vital signs, haematology and blood chemistry evaluations.

Sample size and statistical methods

Assuming a within-patient standard deviation (SD) of 318 mL, using a complete crossover design, 78 evaluable patients would be required to detect a treatment difference of 170 mL in the change from baseline of 2-h post-dose resting IC at Week 3, with 91% power at a two-sided significance level of 0.05. Jointly considering the two comparisons (BDP/FF/G vs. placebo and BDP/FF vs. placebo), the overall power for the primary endpoint would be at least 83%. With a non-evaluable rate of 20%, 102 patients would need to be randomised. This sample size would provide 89% power to detect a difference of 155 mL in dynamic IC at isotime, assuming a within-patient SD of 298 mL, and 84% power to detect a treatment difference of 90 s in exercise endurance time, assuming a within-patient SD of 187 s, both at a two-sided significance of 0.05. The assumptions were based on studies included in a meta-analysis by Di Marco et al. [27].

The primary endpoint was analysed using a linear mixed model assuming an unstructured covariance matrix, including treatment and period as fixed effects, with baseline values for the current period and averaged across all treatment periods as covariates, and patient included as a random effect. Baseline IC was collected pre-dose on Day 1 of each treatment period. The key secondary endpoints were analysed using similar models as the primary endpoint, with baseline dynamic IC and exercise endurance time values taken from the constant workload test conducted pre-dose on Day 1 of each treatment period. Similar models were used to analyse the exploratory and post-hoc (TLC) endpoints. Missing data were not imputed.

Type 1 error was controlled for the primary and key secondary endpoints using a hierarchical strategy. Step 1 was the comparison of BDP/FF/G vs. placebo for the primary endpoint; Step 2 was the comparison of BDP/FF vs. placebo for the primary endpoint; Steps 3 and 4 were the comparisons of BDP/FF/G vs. placebo and BDP/FF vs. placebo, respectively, for IC at isotime at Week 3; Steps 5 and 6 were the comparisons of BDP/FF/G vs. placebo and BDP/FF vs. placebo, respectively, for exercise endurance time. Multiplicity was not controlled for the exploratory endpoints.

The efficacy analyses were evaluated in the intention-to-treat (ITT) set, which was all patients who received at least one dose of study medication and who had at least one post-baseline efficacy evaluation. The per-protocol set, which was all patients in the intention-to-treat set without any important protocol deviations, was used for supportive analyses of the primary and key secondary endpoints. The safety set was all patients who received at least one dose of study medication, and was used for all safety analyses.