CADTH Reimbursement Review

Budesonide-Glycopyrronium-Formoterol Fumarate Dihydrate (Breztri Aerosphere)

Sponsor: AstraZeneca Canada Inc.

Therapeutic area: Chronic obstructive pulmonary disease (COPD)

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Clinical Review

Executive Summary

An overview of the submission details for the drug under review is provided in Table 1.

Introduction

Chronic obstructive pulmonary disease (COPD) is a chronic inflammatory lung disease, often associated with chronic bronchitis and emphysema, that causes obstructed airflow from the lungs, lung hyperinflation, systemic manifestations, and increasing frequency and severity of exacerbations.^1,2 COPD is a preventable and treatable disease with approximately 80% to 90% of cases being caused by smoking.¹ COPD is an under-diagnosed illness; thus, prevalence statistics are likely to underestimate the number of people currently living with COPD. In Canada, COPD is the fifth-leading cause of death.¹ Patients with COPD often experience negative consequences that impact their day-to-day life, including their ability to breathe, talk, sleep, work, and socialize. Overall, these patient experiences describe a physically and mentally exhausting disorder that can result in anxiety, depression, and decreased quality of life. The goals of COPD management are to reduce the frequency and severity of exacerbations, alleviate symptoms, improve exercise tolerance and daily activity, prevent and treat exacerbations and complications, improve health status, and reduce mortality.⁴ Management decisions are guided by disease severity (i.e., symptoms or disability and spirometry) and the frequency of acute exacerbations. Smoking cessation is the single most effective intervention to reduce the risk of developing COPD and the only intervention shown to slow the rate of lung function decline.¹ Bronchodilators form the mainstay of pharmacotherapy for COPD¹ and include long-acting beta-agonists (LABAs) and antimuscarinic drugs (long-acting muscarinic antagonists [LAMAs]). Antimuscarinic and beta-agonist drugs used in combination as a step-up therapy are recommended for patients with stable COPD with exacerbations despite the use of LAMA or LABA therapy.⁵ According to the Canadian Thoracic Society, based on consensus, a step-up to triple therapy, where a LAMA plus a LABA plus an inhaled corticosteroid (ICS) (LAMA-LABA-ICS) is used, may be considered in COPD with high symptom burden and poor health status despite the use of an inhaled LAMA plus LABA (LAMA-LABA) dual therapy.⁵

Breztri Aerosphere is a triple-combination therapy composed of an ICS (budesonide), a LAMA (glycopyrronium), and a LABA (formoterol fumarate dihydrate). Budesonide, the ICS component, is a well-established corticosteroid and can be administered through oral inhalation mechanisms.^3,4 Glycopyrronium, the LAMA component, uses muscarinic receptors to create bronchodilatory effects on smooth muscle in the trachea and bronchi. Formoterol fumarate, the LABA component, is a well-established LABA that is often used as a monotherapy for the treatment of COPD and other indications. This fixed-dose triple therapy is administered twice daily(total dose 320 mcg budesonide plus 14.4 mcg glycopyrronium plus 9.6 mcg formoterol fumarate) through oral inhalation using a metered-dose inhalation (MDI) device, which enables delivery of an exact volume of micronized drugs at each valve actuation.⁵ Aerosphere offers advantages over traditional MDIs, such as consistent delivery of drugs across patient conditions, optimal particle size, and effective delivery of the particles to peripheral and central airways.⁶ The components of Breztri Aerosphere are available in other formulations that also have Health Canada–approved indications for the management of COPD.^3,4

The objective of this review was to perform a systematic review of the beneficial and harmful effects of budesonide plus glycopyrronium plus formoterol fumarate (BGF) dihydrate inhalation aerosol (Breztri Aerosphere 182 mcg, 8.2 mcg, 5.8 mcg per metered actuation) for long-term maintenance treatment of patients with COPD, to reduce exacerbations of COPD, and to relieve symptoms, including chronic bronchitis and/or emphysema.

Stakeholder Perspectives

The information in this section is a summary of input provided by the patient and clinician groups that responded to CADTH’s call for input and from a clinical expert consulted by CADTH for the purpose of this review.

Patient Input

Three patient groups, COPD Canada, the Lung Health Foundation (formerly Ontario Lung Association), and the British Columbia Lung Association, provided input for this submission. The patient groups noted that Breztri Aerosphere would be a welcome addition to provincial formularies across the country. They also expressed the belief that compliance would increase due to the ease of use of this triple therapy, and the effectiveness of the Aerosphere pressurized MDI would ensure this triple-drug combination reached the lungs without relying on a COPD patient’s ability or strength to inhale properly. However, there are continued accessibility issues throughout Canada and provincial drug coverage varies considerably between drug plans.

Clinician Input

Input From Clinical Experts Consulted by CADTH

Input was provided by 1 clinical expert consulted by CADTH for this review. The expert noted that triple-drug combination therapy with an MDI and a spacer or holding chamber, such as an Aerochamber, would be beneficial to patients, as they are currently encouraged to use an Aerochamber for rescue medications. Having their maintenance medication administered using an MDI would also be beneficial. The expert pointed out that the pharmacological components of BGF MDI is familiar to physicians. Lastly, the clinical expert added that although small, the population that might benefit from twice-daily BGF MDI has been better identified as Canadian recommendations evolve.

Clinician Group Input

Input was provided by 3 clinicians from the COPD clinic in the Division of Respirology and Sleep Medicine at Queen’s University. The group noted that a stepwise add-on approach was used in COPD management in clinical settings. They pointed out that COPD patients who experience frequent exacerbations despite being on dual therapies would be the group that might benefit the most from BGF MDI. The clinicians added that fixed-dose triple-therapy combinations, such as BGF MDI, would likely improve the deposition of the bronchodilators and ICS in the target location, leading to better ventilated alveolar units and better outcomes for the patient.

Drug Program Input

The drug program asked about the appropriate comparators for assessing the efficacy of BGF MDI. The clinical expert noted that, although other triple-therapy combinations (fixed dose or other) would be the most appropriate comparators, dual therapies (i.e., ICS plus a LAMA [ICS-LAMA] or a LAMA plus a LABA [LAMA-LABA]) can also be considered to establish the efficacy of BGF.

Clinical Evidence

Pivotal Studies and Protocol Selected Studies

Description of Studies

Two randomized controlled trials (RCTs), KRONOS and ETHOS, were included in the main report of the study. ETHOS was a 52-week, multi-site, double-blind, parallel-group study comparing BGF MDI with glycopyrronium plus formoterol fumarate (GFF) MDI and budesonide plus formoterol fumarate (BFF). Outcomes of the ETHOS trial included the rate of moderate or severe (primary end point) and severe COPD exacerbations, symptoms, health-related quality of life (HRQoL), pulmonary function, and safety. A total of 8,588 patients were randomized to 1 of 4 treatment groups. KRONOS was a 24-week, multi-site, double-blind parallel-group study comparing BGF MDI with GFF MDI and BFF MDI and comparing budesonide plus formoterol fumarate (BUD-FOR) inhalation powder with Symbicort Turbuhaler (TBH) inhalation powder (active control). The KRONOS trial had outcomes similar to the ETHOS trial, including rate of moderate-to-severe COPD exacerbations, symptoms, HRQoL, and safety; change in pulmonary function based on the forced expiratory volume in 1 second (FEV₁) was the primary outcome. Overall, 1,902 patients were randomized among the treatment groups in the KRONOS study.

Efficacy Results

Table 2 presents a summary of key end point results from the included studies.

Exacerbations

In ETHOS, the adjusted rates of moderate or severe exacerbations per year were 1.08, 1.42, and 1.24 for the BGF MDI 320, GFF MDI, and BFF MDI arms, respectively. The rate difference between BGF MDI 320 and GFF MDI was −0.35 (95% confidence interval [CI], −0.46 to −0.23), and between BGF MDI 320 and BFF MDI it was −0.17 (95% CI, −0.27 to −0.06). In KRONOS, the adjusted annualized rates of moderate or severe exacerbations per year were numerically lower for BGF MDI 320 (0.46) compared with GFF MDI (0.95), BFF MDI (0.56), and BUD-FOR dry powder inhaler (DPI) (0.56); however, rate differences were not reported.

BGF MDI 320 was associated with significantly lower rates of moderate or severe COPD exacerbations compared with GFF MDI (rate ratio = 0.76; 95% CI, 0.69 to 0.83 at 52 weeks [ETHOS] and rate ratio = 0.48; 95% CI, 0.37 to 0.64 at 24 weeks [KRONOS]), and BFF MDI at 52 weeks (rate ratio = 0.87; 95% CI, 0.79 to 0.95 [ETHOS]). In the KRONOS trial, there was no statistically significant difference found in the rate of moderate-to-severe exacerbations between BGF MDI and BUD-FOR DPI (rate ratio = 0.83; 95% CI, 0.59 to 1.18) nor between BGF MDI and BFF MDI (rate ratio = 0.82; 95% CI, 0.58 to 1.17) at 24 weeks.

Lung Function

Lung function, measured as FEV₁ area under the curve from 0 to 4 hours post dose (AUC_0-4) over 24 weeks, was the primary outcome for the comparisons of BGF MDI 320 versus BFF MDI (ETHOS and KRONOS) and BGF MDI 320 versus BUD-FOR DPI (KRONOS). In ETHOS, this outcome was assessed in a pulmonary function test (PFT) substudy population. BGF MDI 320 showed statistically significant improvement in lung function compared with both BFF MDI (least squares mean = 104 mL; 95% CI, 77 to 131) and BUD-FOR DPI (least squares mean = 91 mL; 95% CI, 64 to 117). Based on a minimal clinically important difference (MCID) of 0.10 L to 0.14 L, these differences were likely clinically significant. Lung function measured as morning pre-dose trough FEV₁ over 24 weeks was the primary outcome for the comparisons of BGF MDI 320 versus GFF MDI (ETHOS and KRONOS). In ETHOS, this outcome was assessed in a PFT substudy population. The change from baseline in morning pre-dose trough FEV₁ at 24 weeks for BGF MDI 320 compared with GFF MDI was not clinically significant (22 mL: 95% CI, 4 to 39).

Use of Rescue Medication

In both trials, the evaluation of the average daily number of puffs of rescue medication over 24 weeks was restricted to the rescue Ventolin use population. In ETHOS, BGF MDI 320 was associated with a statistically significant reduction in the use of rescue medication compared with GFF MDI (difference = −0.51 puffs per day; 95% CI, −0.68 to −0.34) and BFF MDI (difference = −0.37 puffs per day; 95% CI, −0.54 to −0.20). No statistically significant differences were found between the groups in KRONOS.

Symptoms

The change from baseline in the Transition Dyspnea Index (TDI) focal score was used by both trials to assess improvement in dyspnea symptoms with treatment. Although both trials found that BGF MDI improved symptoms compared with GFF MDI, BFF MDI, and BUD-FOR DPI, these were not clinically significant improvements. In ETHOS, the difference in the least squares mean of the TDI focal score in BGF MDI 320 compared with GFF MDI was 0.40 units (95% CI, 0.24 to 0.55) and, compared with BFF MDI, was 0.31 units (95% CI, 0.15 to 0.46). In KRONOS, the difference in the least squares mean of the TDI focal score in BGF MDI 320 versus GFF MDI was 0.177 units (95% CI, −0.071 to 0.426), 0.237 units (95% CI, −0.068 to 0.542) compared with BFF MDI, and 0.461 units (95% CI, 0.156 to 0.766) compared with BUD-FOR DPI. Compared with other groups, there were no clinically meaningful improvements in symptoms for BGF MDI as measured using the Exacerbations of Chronic Pulmonary Disease Tool (EXACT) or the Evaluating Respiratory Symptoms in COPD (E-RS) instruments.

Health-Related Quality of Life

St. George's Respiratory Questionnaire (SGRQ) was used to measure HRQoL in both trials. Patients in all treatment groups showed clinically significant improvement in SGRQ total score before and after treatment. However, between groups, these improvements were not clinically significant (MCID = 4 units). In ETHOS, the mean difference in SGRQ total score was −1.62 units (95% CI, −2.27 to −0.97) for BGF MDI versus GFF MDI, and −1.38 (95% CI, −2.02 to −0.73) for BGF MDI 320 versus BFF MDI. These differences were statistically significant, but not clinically meaningful. In KRONOS, the mean differences were −1.22 units (95% CI, −2.30 to −0.15) and −0.45 units (95% CI, −1.78 to 0.87) for BGF MDI versus GFF MDI, and BGF MDI versus BFF MDI, respectively.

Mortality

All-cause mortality was evaluated in ETHOS (as a secondary outcome) but not in KRONOS. The risk of death (all causes) was lower during treatment with BGF MDI 320 relative to GFF MDI (hazard ratio [HR] = 0.51; 95% CI, 0.330 to 0.80), but not different relative to BFF MDI (HR = 0.72; 95% CI, 0.44 to 0.1.16) as assessed by the Cox proportional hazards model.

Harms Results

Within each trial, adverse events (AEs) were similar across treatment arms. The most common treatment-associated AEs were COPD (9.5% to 11.3% in ETHOS, 2.5% to 5.1% in KRONOS), nasopharyngitis ((9.4% to 11% in ETHOS, 7.7% to 9.4% in KRONOS), and upper respiratory tract infections (4.8% to 5.7% in ETHOS, 5.7% to 10.2% in KRONOS). Around 20% of patients in ETHOS and 9% of those in KRONOS reported 1 or more serious adverse events (SAEs). Incidence of pneumonia was 1.6% to 2.8% in ETHOS and 0 to 1.3% in KRONOS. In ETHOS, there were 112 deaths on treatment (1.3%) across all treatment groups, whereas in KRONOS there were 12 deaths (0.6%).

Approximately 6% of patients in ETHOS and 4% of patients in KRONOS withdrew due to AEs. The number of patients who withdrew was relatively similar across treatment arms. The most common reason for withdrawal was lack of efficacy, reported as COPD. Notable harms such as cardiovascular events, anticholinergic events, and corticosteroid-related events were reported in a small number of patients in each trial. The incidences were relatively similar across treatment arms. Details on these notable harms are shown in Table 2.

Critical Appraisal

One of the limitations of the included study was the amount of missing data, likely due to early discontinuation. In ETHOS, in particular |||||||||||||||||||||||||||||||||||||||||||||||||| for the primary end point. The relatively high rates of treatment discontinuation (22% in ETHOS and 14% KRONOS) could have biased the results in favour of BGF MDI. The BUD-FOR TBH arm in KRONOS was open label, which could have introduced a subjective bias in patient-reported outcomes. The chances of inadvertent unblinding due to AEs were low, given the similarities in the events across the treatment groups. The degree and type of training provided for the inhaler device was not described in the trials. However, treatment adherence was high across the treatment arms in both trials, which, measured by the ratio of daily puffs taken and the expected number of daily puffs, was 93.2 and 95.2 in ETHOS and KRONOS, respectively.

There were several limitations related to generalizability. The percentage of female participants was lower than what would be expected in real-world settings. Participants’ treatment history with LAMA-LABA (14%) combinations was low compared with the Canadian COPD population. Approximately 3/4 of patients in both trials used an ICS at baseline, which was higher than in clinical settings for respirologists in Canada. This, coupled with the low threshold of baseline eosinophil levels considered for subgroup analysis, was of some concern to the clinical expert consulted by CADTH. However, the threshold and proportion of patients with a baseline eosinophil count of 150 cells/mm³ or greater in ETHOS was similar to that in trials of other triple-therapy combinations.⁷ Although the ideal comparator for BGF MDI would have been other triple-therapy combinations, the comparators used in the study were considered appropriate for establishing efficacy. Some outcomes that were pointed out as important by the patient groups were not considered in the trials, such as exercise tolerance and patient satisfaction, except for items within the SGRQ.

Indirect Comparisons

One network meta-analysis (NMA), submitted by the sponsor, was identified to provide indirect evidence. The NMA compared BGF MDI 320 with other open and fixed triple-therapy combinations of LAMA-LABA-ICS for the treatment of moderate-to-very-severe COPD. A systematic literature search and the study selection and quality assessments were conducted appropriately. For the NMA analysis, all LAMA-LABA combinations were grouped together into a single node to create networks and an assumption of similar efficacy was made. Analyses were conducted using a 3-level hierarchical Bayesian NMA model.

The population, intervention, comparators, and outcomes considered by the NMA were relevant. Fifteen double-blinded RCTs were included in the NMA. The baseline characteristics of the study participants and results of the included studies were not reported, making the interpretation across trials regarding potential effect modifiers and homogeneity challenging. However, an assessment of heterogeneity and planned sensitivity analyses were conducted for most of the outcomes. The results of the NMA found comparable levels of efficacy and safety between BGF MDI 320, fluticasone furoate plus umeclidinium plus vilanterol (FF-UMEC-VI) (Trelegy Ellipta) and other open triple-therapy combinations.

Other Relevant Evidence

A 52-week extension study of the KRONOS trial, the results of which were included in the appendix of the NMA, evaluated the safety of the triple BGF MDI fixed-dose combination and its effects on bone mineral density (BMD) and ocular safety in patients with moderate-to-severe COPD. Overall, 456 patients were included in the safety population and randomized to 1 of the treatment groups (BGF MDI, BFF MDI, or GFF MDI). The changes from baseline in all 3 groups were small and not clinically meaningful, and there were no new or unexpected safety findings. The main limitations of the extension safety study were the study duration and the relatively small sample size.

Conclusions

Two multinational double-blind RCTs sponsored by AstraZeneca were included in this review, ETHOS (52 weeks) and KRONOS (24 weeks), comparing the efficacy of BGF MDI 320 with that of dual therapy for patients with COPD. Overall, BGF MDI 320 reduced the rate of moderate or severe exacerbations compared with GFF MDI and BFF MDI at 52 weeks. Although numerical improvements in HRQoL and symptoms were reported, the between-group differences were not clinically significant. BGF MDI 320 improved FEV₁ over 24 weeks compared with all active comparators. However, this improvement in FEV₁ was not clinically significant. The average daily use of rescue medications was decreased by 1 unit after treatment with BGF MDI for 24 weeks. The between-groups difference in the use of rescue medications was inconclusive. The lack of head-to-head trials between triple-therapy combinations is an important gap in the evidence. Indirect evidence from a sponsor-submitted NMA compared BGF MDI 320 with FF-UMEC-VI (Trelegy) and other open triple-therapy combinations. BGF MDI likely showed similar or comparable efficacy and safety compared with other triple therapies, with notable limitations inherent with indirect comparison. AEs associated with BGF MDI 320 were consistent with those expected for each of the individual components.

Introduction

Disease Background

COPD is a chronic inflammatory lung disease, often associated with chronic bronchitis and emphysema, that causes obstructed airflow from the lungs, lung hyperinflation, systemic manifestations, and increasing frequency and severity of exacerbations.^1,2 COPD is a preventable and treatable disease with approximately 80% to 90% of cases being caused by smoking.¹ A number of factors may cause COPD and contribute to its complexity, including long-term cumulative exposure to occupational dusts and chemicals, second-hand smoke or wood smoke and other biomass fuels used for cooking; frequent lung infections as a child; or genetic reasons (alpha₁-antitrypsin deficiency).^2,8 This disease is associated with several comorbidities, including ischemic heart disease, osteopenia and osteoporosis, glaucoma and cataracts, cachexia and malnutrition, anemia, peripheral muscle dysfunction, cancer, and metabolic syndrome.¹

COPD is an under-diagnosed illness; thus, prevalence statistics likely underestimate the number of people currently living with COPD. In 2015, the global prevalence of COPD in adults aged 40 and over was reported to be 6.4%, marking a slight decline since 2008 (7.2%).^9,10 However, the prevalence, morbidity, and mortality of COPD varies across the world due to a variety of factors, including a country’s income level (high or low), air quality, and health insurance coverage.¹⁰ According to a recent Statistics Canada report, COPD is the fifth-leading cause of death in Canada and was previously ranked as the fourth-leading cause of death up until 2015.^11,12 Historically, the prevalence and incidence of COPD has been higher in males compared with females; however, this has changed over the years, as the rates of smoking have increased among females.⁹ Moreover, new evidence suggests that women may be more susceptible to the effects of tobacco than men, possibly due to a difference in lung physiology, leading to more severe disease in women.⁸

Patients with COPD are often limited in their day-to-day life, including their ability to breathe, talk, sleep, work, and socialize. Overall, patients describe COPD as a physically and mentally exhausting disorder that can result in anxiety, depression, and a decrease in quality of life. In addition, COPD has a profound effect on caregivers, who cite a number of challenges, including limited time for managing their own health and well-being, feelings of depression and isolation, anxiety, stress, fatigue, feeling of unending days, and increased requirements for social support.

The goals of COPD management are to reduce the frequency and severity of exacerbations, alleviate symptoms, improve exercise tolerance and daily activity, prevent and treat exacerbations and complications, improve health status, and reduce mortality.¹

Standards of Therapy

Management decisions are guided by disease severity (i.e., symptoms, disability, and spirometry) and the frequency of acute exacerbations. Smoking cessation is the single most effective intervention to reduce the risk of developing COPD and the only intervention shown to slow the rate of lung function decline; regular exercise with cardiorespiratory conditioning can improve functional status and sensation of dyspnea in COPD patients more than medications alone. Education and self-management skills are also integral to the non-pharmacological management of COPD. Pulmonary rehabilitation is recommended for all COPD patients who are symptomatic.

Bronchodilators form the mainstay of pharmacotherapy for COPD¹³ and include LABAs such as salmeterol, formoterol, indacaterol, and vilanterol, and LAMAs such as tiotropium, glycopyrronium, aclidinium, and umeclidinium. LAMAs and LABAs used in combination as a step-up therapy are recommended for patients with stable COPD who are still experiencing exacerbations despite the use of LAMA or LABA monotherapy.¹³ Combinations of a fixed-dose LABA-ICS, such as fluticasone plus salmeterol, may be considered for certain patients with COPD. ICS is not recommended as monotherapy in COPD and, should only be combined with an inhaled LABA when used.. An ICS may not be useful for mild disease; but may have more of a role in the management of patients with moderate-to-severe COPD who are experiencing 2 or more exacerbations (or 1 or more exacerbations leading to hospital admission) per year, or in those with persistent symptoms.^1,8,14,15 There may also be a subpopulation of COPD patients who have concomitant asthma or airway eosinophilia where ICS use may be beneficial.^8,13,16-18 Patients with persistent symptoms and poor health status who continue to experience exacerbations despite inhaled LAMA-LABA dual therapy may be recommended to step up to LAMA-LABA-ICS triple therapy.^8,13 Methylxanthines (such as theophylline) and phosphodiesterase inhibitors (roflumilast) are adjunctive therapies for COPD management that have a limited place in the treatment of COPD in Canada. Oxygen therapy is used in patients with very severe COPD with persistent hypoxemia.

Drug

Breztri is a fixed-dose combination of BGF in a pressurized inhalation aerosol for oral inhalation, marketed as the Aerosphere inhaler device. BGF MDI is currently under review by Health Canada for long-term maintenance treatment to reduce exacerbations of COPD and to relieve symptoms in patients with COPD, including chronic bronchitis and/or emphysema. BGF MDI delivers budesonide 182 mcg, glycopyrronium 8.2 mcg, and formoterol fumarate dihydrate 5.8 mcg per metered actuation. Table 3 provides an overview of BGF MDI and the other currently available LAMA-LABA-ICS fixed-dose combination inhaler for COPD, FF-UMEC-VI (Trelegy Ellipta).

Stakeholder Perspectives

Patient Group Input

This section was prepared by CADTH staff based on the input provided by patient groups.

1. Brief Description of Patient Group(s) Supplying Input

Three patient groups, COPD Canada, the Lung Health Foundation (Ontario Lung Association), and the British Columbia Lung Association, provided input for this submission. These not-for-profit organizations help inform and support Canadians living with COPD and other lung diseases. Furthermore, these organizations act as educational resources and patient advocacy groups, providing patient education materials and services and producing quality-of-life seminars for patients and their caregivers.

COPD Canada reviews and interprets scientific literature related to emphysema and chronic bronchitis so that it can be easily interpreted by the community. The British Columbia Lung Association works closely with the Canadian Thoracic Society, Canadian Lung Association, the Lung Health Association, and other partners while providing funding to various British Columbia research initiatives pertaining to lung disease. The Lung Health Foundation invests in lung research and advocates for improved treatment and care for lung health.

Declarations of financial support for each patient group are available on the CADTH website.

2. Condition Related Information

COPD Canada collected Canadian-applicable patient input from the personal experiences of the organization’s members and from published scientific articles. Members of COPD Canada provided their experiences during group pulmonary rehabilitation sessions, lung issue support groups, and in direct one-on-one consultations. In addition, COPD Canada distributed an email survey in January 2021 for which they received 64 written responses.

The British Columbia Lung Association conducted telephone interviews with some of its members (5 patients and 5 caregivers). The interviewees were either on or caring for patients on triple therapy prescribed by a respirologist.

The Lung Health Association collected online survey responses (received in December 2020) from people living with COPD (n = 39), emphysema (n = 12), and chronic bronchitis and bronchiectasis (n = 4) as well as from 18 caregivers caring for a family member living with COPD. Additionally, 2 phone interviews were completed in January 2021 and input was received from a certified respiratory educator who reviewed disease experience and experiences with available treatments and outcomes.

COPD has a profound effect on the lives of both patients and caregivers. COPD is associated with a considerable burden of disease, affecting many things that are fundamental to everyday life, such as the ability to breathe, talk, sleep, work, and socialize. As the disease progresses and worsens, patients become less physically active and more socially isolated. Many patients with COPD are of working age, so even in the early stages of the disease, the breathlessness and fatigue caused by COPD reduces the ability of the patient to go to work or carry out their normal work activities. Some patients are forced to go into early retirement as a consequence of the progressive disease’s severity.

Even many of the day-to-day activities most take for granted are virtually impossible or extremely difficult for people with severe COPD. Changing bed sheets, bathing and dressing, shopping and carrying bags (e.g., groceries), climbing stairs, and walking and talking at the same time are all examples of such day-to-day activities.

Patients with COPD can often develop anxiety, depression, and a sense of reduced self-worth because of their inability to perform common tasks due to their lung condition. Additionally, patients with COPD often have difficulty fighting infections and ongoing exacerbations may lead to a worsening of lung function.

Caregivers face considerable challenges that commonly include limited time for managing their own health and well-being, feelings of depression and isolation, anxiety, stress, fatigue, feeling of unending days, and increased requirements for social support. In the case of grown children who become their parent’s caregivers, they are often torn between the needs of their young families and the needs of their elderly parent with COPD.

One respondent described their experience living with COPD:

• It has seriously slowed activity including ruling out some of my favourites such as, hiking, walking in the woods, climbing hills, keeping up with anyone even on flat ground and I am totally unable to walk up hills. Sexual activity is very impacted. Can't do even simple jobs around the home such as mowing the grass, lifting things, or even carrying groceries and other items. The deterioration is so fast that now I could never do the renovation work I did on my house only 8 years ago.

3. Current Therapy Related Information

Typical maintenance therapy usually includes the use of tiotropium bromide (Spiriva) with fluticasone plus salmeterol (Advair), or budesonide plus formoterol (Symbicort) twice per day. Rescue medications vary from patient to patient, although salbutamol (Ventolin) is used quite extensively. These products are to control the condition, but they do not improve long-term lung function. When a patient experiences an exacerbation, prednisone and antibiotics are often prescribed. Prednisone works quickly but has very dangerous side effects. The overuse of antibiotics has become a national and international concern due to increased resistance, particularly in long-term care facilities.

The British Columbia Lung Association noted that when patients were on triple-therapy medications, symptoms subsided, and patients were able to resume or had improvements in certain activities like walking and household chores.

A respondent had the following to say about their experience with treatments they have used:

• Have used Spiriva and Advair. Got thrush in my mouth a couple of times. Advair leaves too much powder behind inside my mouth and I wonder if any meds even get to my lungs. Trelegy does the same. They both seem to waste a lot of the medicine. Spiriva dispenser is much easier to use and its design seems to target the back of my mouth and throat much better.

4. Expectations About the Drug Being Reviewed

Although none of the patients with COPD who responded to the survey had direct experience with the drug under review, their input clearly identified that they want additional therapies that work to improve breathing and lung function, are easy to use, and do not just offer symptomatic or emergency relief. Because COPD is treated in a stepwise manner, where treatments are layered on as the disease progresses, additional treatment options are often needed to address continual disease progression and severity. As well, the input noted that long-term use of some of these compounds results in a perceived diminishing of the drug’s effectiveness. Thus, it would be desirable to have alternative but equivalent drugs available beyond what there is currently.

The patient groups noted that BGF MDI would be a welcome addition to provincial formularies across the country. Their input indicates they expect that adherence to treatment would increase with the ease of use of this triple therapy. They also expressed hope the Aerosphere inhaler would be easy to use and would ensure the drug is reaching the lungs without relying on a patient’s ability or strength to inhale properly to administer a dose. The British Columbia Lung Association noted that the choice of inhaler device was important and significant for COPD patients, as it has been shown that some devices do not dispense medications appropriately to large and small airways.

COPD Canada is aware of accessibility issues throughout Canada and notes that provincial drug coverage varies considerably among the plans. COPD Canada points out that most COPD patients are over 65 years of age and rely on provincial drug coverage.

Clinician Input

Input From Clinical Experts Consulted by CADTH

All CADTH review teams include at least 1 clinical specialist with expertise in diagnosing and managing the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., providing guidance on the development of the review protocol, assisting in the critical appraisal of clinical evidence, interpreting the clinical relevance of the results, and providing guidance on the potential place in therapy). The following input was provided by a clinical specialist with expertise in the diagnosis and management of COPD.

Since the recent publications of the WISDOM, FLAME, and Trelegy studies, there has been great debate as to what the place of ICS in the management of COPD and the prevention of exacerbations should be. WISDOM and FLAME highlighted the role of dual bronchodilators in the prevention of COPD exacerbations in patients known to have had an exacerbation in the last 12 months. WISDOM provided evidence that withdrawing ICS did not lead to an increased risk of exacerbation, while FLAME showed superiority of dual bronchodilation over an ICS-LABA in preventing all exacerbations (although there was no statistically significant difference in severe exacerbations). The Global Initiative for Obstructive Lung Disease (GOLD) group published revised recommendations taking into consideration the new literature and reviewed the role of ICS in the management of COPD. To illustrate the changes proposed in GOLD, if we look at group C, a group with patients who had an exacerbation requiring hospitalization or who had more than 2 exacerbations requiring treatment with antibiotics and prednisone, dual bronchodilation is recommended and the ICS-LABA combination is an alternative treatment. Similarly, in group B, which regroups the patients with 1 exacerbation or less, there is no recommendation to use an ICS-LABA. Trelegy noted an increase in mortality rates in COPD patients with a history of exacerbations if ICS treatment was withdrawn suddenly; however, after that first month, the mortality rate stabilized between the groups of patients who were either on or off an ICS.

When reviewing AEs, especially when considering pneumonia as an AE in treating COPD patients, the risk seems less with budesonide than with other ICSs.

Something that has been missing in the treatment options for COPD is a triple-therapy combination using an MDI device. Patients are encouraged to use salbutamol as an emergency treatment using an Aerochamber, but this implies that treating teams need to teach patients how to use both their regular everyday inhaler as well as another device for emergency situations; thus, having a combination triple-therapy device in MDI format could represent an advantage.

Physicians are familiar with the individual pharmacological components of this product, as they are all available on the Canadian market, although they are in somewhat different form and delivered using different devices.

The place that a twice-daily MDI triple-therapy combination product would take is probably a small one. As the positive effects of dual-combination therapy on exacerbation rates and symptom management have been shown, and as international and Canadian recommendations have evolved, the population that would benefit from triple therapy has been better identified.

Clinician Group Input

This section was prepared by CADTH staff based on the input provided by clinician groups.

A joint submission providing input from 3 clinicians from Ontario was received on the reimbursement review of BGF MDI. All 3 clinicians work at the COPD outpatient clinic in the Division of Respirology and Sleep Medicine at Queen’s University. The clinicians noted their work is at a university-based academic centre where they are involved in research, teaching, and long-term care (including rehabilitation) of patients with COPD.

Information for this input was collected through a peer-reviewed paper, recent reviews, and updates on the topic up to February 9, 2021. The clinicians also noted that they see patients with mild to end-stage COPD and are active researchers with extensive lecturing experience in clinical issues involving patients with COPD.

Unmet Needs

The clinicians noted that, as per Canadian Thoracic Society recommendations, management of COPD includes smoking cessation, anti-influenza and anti-pneumococcal vaccination, regular physical activity (or pulmonary rehabilitation in more dyspneic patients), and short-acting bronchodilators, regardless of disease severity. The clinicians also added that addressing comorbidities (e.g., chronic rhinosinusitis, gastroesophageal reflux disease, sleep disorder breathing, ischemic heart disease) is important.¹⁹

For more symptomatic patients (at least grade 1 according to the Modified Medical Research Council [mMRC] Dyspnea Scale), the clinicians noted that a stepwise add-on approach to long-acting bronchodilators is applied. The clinicians noted that patients usually start with either a LAMA or LABA followed by a LAMA and LABA combination if monotherapy is deemed insufficient for adequate control of symptoms (that are at least grade 2 according to the mMRC scale). ICSs are added to a LABA or LAMA-LABA for patients experiencing frequent exacerbations (≥ 2 moderate exacerbations that prompted a prescription for antibiotics and/or oral steroids or at least 1 severe exacerbation that required hospitalization or an emergency room visit in the last year) and the clinicians noted that some physicians use eosinophil counts to guide adding an ICS (e.g., ≥ 300 cells/mm³).²⁰

The group of clinicians added that oral theophylline is used in some patients with advanced disease as well as low-dose opiates and anxiolytics. Additionally, they noted that macrolide prophylaxis,²¹ oral N-acetylcysteine, and roflumilast might be used to further decrease the frequency and severity of exacerbations. The clinicians added that action plans, usually containing a respiratory fluoroquinolone and oral steroids, are made available for patients who may be more educated on their use.²²

The clinicians added that long-term oxygen therapy is considered for hypoxemic patients (either rest or exercise), whereas non-invasive positive-pressure ventilation may be helpful in hypercapnic patients.²³ Furthermore, the clinicians noted that bullectomy and lung volume reduction surgeries (including endoscopic) are restricted to carefully selected patients, and eligible patients who have end-stage disease may be referred for lung transplantation. The clinicians noted that no treatment modifies the underlying disease mechanism; treatments are fundamentally focused on reducing the burden of moderate-to-severe exacerbations and dyspnea control.

The current goals of therapy for this patient population include reducing dyspnea (particularly activity-related), reducing the burden of moderate-to-severe acute exacerbations, decreasing the rate of lung function decline, and improving HRQoL. Additionally, the clinicians noted that, more recently, improving survival has been considered a potential target for treatments.²⁴

For those patients whose needs are not being met by currently available treatments, the clinicians noted that highly variable responses are seen for dyspnea control, and there is reduced efficacy in controlling exacerbations for those who need such controls. The clinicians added there are important limitations in delivering drugs to smaller airways that need to be addressed, as well as shortcomings in the activation of DPIs due to insufficient peak inspiratory flows in patients who are hyperinflated. They also noted there is poor round-the-clock bronchodilatation in once-daily medications and there is a lack in the Canadian market of twice-daily triple therapy (which offers more consistent bronchodilation through the day and night) delivered by an MDI (which has better airway deposition). Clinicians also commented that convenience (e.g., once daily versus twice daily) is usually an issue in chronically dyspneic patients who derive sensory benefit from long-acting bronchodilators. No treatments are available to reverse the course of the disease for hypoxemic patients, apart from smoking cessation and long-term oxygen therapy.

The clinicians noted that COPD patients who experience frequent exacerbations despite being on dual therapies like LABA and ICS or LAMA and LABA and, in particular, patients who report at least moderate dyspnea in daily life (mMRC ≥ 2), are most in need of therapy.²⁵ The clinicians added that the drug combination under review would address this unmet patient need. They added there is recent evidence that the drug combination under review may also decrease the exacerbation burden, even in patients who do not present with a recent (within the past year) exacerbation.²⁶

Place in Therapy

The clinicians noted that BGF MDI can be used either after a trial of LABA and ICS or LAMA and LABA, or as a first-line treatment in patients with a particularly high burden of moderate-to-severe exacerbations.

In addition, the clinicians added that triple therapy in a single inhaler complements the foundations of COPD treatment. They added there is another triple therapy in the Canadian market (FF-UMEC-VI, Trelegy), but this product is administered once daily through a DPI. The clinicians expressed concerns about both drug delivery and insufficient bronchodilation over the 24 hours with FF-UMEC-VI, and that fluticasone furoate may be associated with a higher rate of bacterial pneumonia compared with budesonide.²⁷ The clinicians noted that formoterol is a dual short- and long-acting bronchodilator with a faster onset of action compared with vilanterol, which leads to quicker relief of dyspnea.²⁸ The clinicians noted this may result in FF-UMEC-VI being insufficient for dyspnea control in more symptomatic patients.

When asked whether it would be appropriate to recommend that patients try other treatments before initiating treatment with the drug combination under review, the clinicians noted that a standard recommendation on this is not appropriate, given the large heterogeneity of the disease. The clinician group noted that the available evidence to date clearly indicates that triple therapy (LAMA-LABA-ICS) is superior to dual therapies (LAMA and LABA and ICS and LABA) in reducing the burden of moderate-to-severe exacerbations while improving dyspnea to the same extent as LAMA and LABA.^26,29 Additionally, with respect to sequencing, the clinicians note that triple therapy may be the more appropriate first choice for patients with an unordinary burden of moderate-to-severe exacerbations and who are at high risk for a negative outcome if a major exacerbation occurs. The clinicians added that the likelihood of a negative, life-threatening outcome is usually decided on a patient-by-patient basis based on the severity of functional impairment, previous history of life-threatening exacerbations (including intensive care unit [ICU] admission), and the severity of comorbidities (particularly cardiovascular disease).

Patient Population

When asked which patients would be best suited for treatment with the drug combination under review, the clinicians noted this would be COPD patients presenting with a high burden of moderate-to-severe exacerbations, as these patients are more in need of an intervention.^25,30 The clinicians said they would also consider using it to treat patients without a recent (within the last year) exacerbation. In addition, the clinicians added that the presence of at least moderate dyspnea in daily life (at least grade 2 according to the mMRC scale) would strengthen the indication for use, but it should not be a requirement.

According to the clinicians, patients would be identified through the mandatory diagnosis of chronic, poorly reversible airflow limitation on spirometry. In addition, they noted that spirometry alone might underestimate the severity of functional impairment and therefore measurements of lung volumes and lung-diffusing capacity might be warranted for patients with out-of-proportion dyspnea. The clinicians also added that under-diagnosis due to the lack of PFTs is common; however, once a PFT is available, the diagnosis is usually straightforward in a patient with a high pretest likelihood of disease, i.e., smoker or ex-smoker, aged 40 or older. In addition, they added that clinical history is crucial to identify those patients experiencing frequent exacerbations and those reporting a higher dyspnea burden.²⁴ Blood eosinophils counts were noted by the clinicians as possibly providing auxiliary information. They commented that there is not sufficient evidence to indicate that treatment with triple therapy should be considered in pre-symptomatic COPD patients.

The patients least suited for treatment with the drug combination under review would be those with a low burden of exacerbations (≤ 1 moderate exacerbation and no severe exacerbation in the preceding year).³¹ Patients with mild dyspnea (mMRC dyspnea grade ≤ 1) being treated with a LAMA and/or LABA and/or short-acting bronchodilators would also be least suited.

Clinical history (dyspnea burden and dyspnea in daily life) as well as, occasionally, circulating eosinophilia, may help to identify those patients who are most likely to exhibit a response to the drug combination under review. According to the clinicians, the severity of functional impairment as measured with PFTs may also influence some treatment choices.

Assessing Response to Treatment

When asked which outcomes are used to determine whether a patient is responding to treatment in clinical practice, the clinicians noted that frequency and severity of COPD exacerbations and dyspnea in daily life would be the most appropriate outcome assessed. Similarly, the clinicians noted that the absence of severe exacerbations and 1 or fewer moderate exacerbations in the year subsequent to treatment initiation would demonstrate a clinically meaningful response. This, along with improvement in at least 1 grade in mMRC dyspnea score and lower as-needed use of short-acting bronchodilators, would demonstrate a clinically meaningful response. The clinicians noted that the magnitude of the treatment effect can be assessed by any physician and response should be assessed 1 to 2 months after treatment initiation to ensure compliance and lack of side effects. Thereafter, a response assessment every 6 months would be appropriate for most patients. However, this time period may be shortened in the presence of repetitive exacerbations and the use of an action plan despite adherence to treatment.

Discontinuing Treatment

When asked about considerations for treatment discontinuation for the drug combination under review, the clinicians noted that once triple therapy is initiated in a patient for whom treatment with the drug combination is appropriate, there is conflicting evidence on whether or not it is safe to de-escalate to LABA and LAMA or LABA and ICS after exacerbation control. The clinicians noted that this situation is likely to be safely accomplished by a specialist (respirologist) on a patient-by-patient basis.³²

Prescribing Conditions

Community settings such as outpatient clinics or specialty clinics were identified by the clinicians as the most appropriate treatment setting for the drug combination under review.

In addition, the clinicians noted that for most patients, a specialist is not necessary to diagnose, treat, and monitor patients who may receive the drug combination under review, provided the family physician sees the patient regularly. However, the clinicians added that a respirologist or a COPD nurse practitioner might be better at monitoring severe patients with several hospital admissions and frequent emergency room visits, and those patients who require multiple interventions such as long-term oxygen therapy, non-invasive ventilation, opiates for dyspnea control, and so forth.²⁴

Additional Information

The clinicians noted the recent data, observed in 2 large RCTs,^30,33 showing that a LAMA-LABA-ICS (including the drug combination under review) administered using a single device has a positive effect on all-cause and respiratory mortality in patients with COPD presenting with a history of moderate-to-severe exacerbations. They noted that a strong signal in relation to lower cardiovascular-related mortality, a key cause of early death in Canadians with COPD, was observed in both studies.

The clinicians added that a LAMA-LABA-ICS may be prescribed on different inhalers and there is good evidence that administration on a single inhaler is associated with decreased health care resource utilization and improved cost-effectiveness compared with multiple inhalers. The clinicians commented that the use of a single device likely improves the deposition of the bronchodilators and steroids at the same location (i.e., the better ventilated alveolar units).³⁴

Drug Program Input

The drug programs provide input on each drug being reviewed through CADTH’s reimbursement review processes by identifying issues that may impact their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical expert consulted by CADTH are summarized in Table 4.

Clinical Evidence

The clinical evidence included in the review of BGF (BGF MDI 320, Breztri Aerosphere 182 mcg per 8.2 mcg per 5.8 mcg) is presented in 3 sections. The first section, the systematic review, includes pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those studies that were selected according to an a priori protocol. The second section includes indirect evidence from the sponsor and indirect evidence selected from the literature that met the selection criteria specified in the review. The third section includes sponsor-submitted long-term extension studies and additional relevant studies that were considered to address important gaps in the evidence included in the systematic review.

Systematic Review (Pivotal and Protocol Selected Studies)

Objectives

To perform a systematic review of the beneficial and harmful effects of BGF (BGF MDI 320, Breztri Aerosphere 182 mcg per 8.2 mcg per 5.8 mcg) for the long-term maintenance treatment to reduce exacerbations of COPD and to relieve symptoms in patients with COPD, including chronic bronchitis and/or emphysema.

Methods

The studies selected for inclusion in the systematic review include pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those meeting the selection criteria presented in Table 5. Outcomes included in the CADTH review protocol reflect outcomes considered to be important to patients, clinicians, and drug plans.

Of note, the systematic review protocol presented in Table 5 was established before the granting of a Notice of Compliance from Health Canada.

The literature search for clinical studies was performed by an information specialist using a peer-reviewed search strategy according to the PRESS Peer Review of Electronic Search Strategies checklist (https://www.cadth.ca/resources/finding-evidence/press).³⁵

Published literature was identified by searching the following bibliographic databases: MEDLINE All (1946‒) through Ovid and Embase (1974‒) through Ovid. The search strategy comprised both controlled vocabulary, such as the National Library of Medicine’s MeSH (Medical Subject Headings), and keywords. The main search concept was a combination of 3 drugs (budesonide- glycopyrrolate-formoterol). Clinical trials registries were searched: the US National Institutes of Health’s clinicaltrials.gov, WHO’s International Clinical Trials Registry Platform (ICTRP) search portal, Health Canada’s Clinical Trials Database, and the European Union Clinical Trials Register.

No filters were applied to limit the retrieval by study type. Retrieval was not limited by publication date or by language. Conference abstracts were excluded from the search results. See Appendix 1 for the detailed search strategies.

The initial search was completed on February 26, 2021. Regular alerts updated the search until the meeting of the CADTH Canadian Drug Expert Committee (CDEC) on June 16, 2021.

Grey literature (literature that is not commercially published) was identified by searching relevant websites from the Grey Matters: A Practical Tool For Searching Health-Related Grey Literature checklist (https://www.cadth.ca/grey-matters).³⁶ Included in this search were the websites of regulatory agencies (US FDA and European Medicines Agency). Google was used to search for additional internet-based materials. See Appendix 1 for more information on the grey literature search strategy.

These searches were supplemented by reviewing bibliographies of key papers and through contact with appropriate experts. In addition, the sponsor of the drug was contacted for information regarding unpublished studies.

Two CADTH clinical reviewers independently selected studies for inclusion in the review based on titles and abstracts, according to the predetermined protocol. Full-text articles of all citations considered potentially relevant by at least 1 reviewer were acquired. Reviewers independently made the final selection of studies to be included in the review, and differences were resolved through discussion.

Findings From the Literature

A total of 2 studies were identified from the literature for inclusion in the systematic review (Figure 1). The included studies are summarized in Table 6. A list of excluded studies is presented in Appendix 2.

Figure 1: Flow Diagram for Inclusion and Exclusion of Studies

Description of Studies

Two RCTs (ETHOS and KRONOS) were submitted by the sponsor and were included in the systematic review. The focus of this section is on these 2 pivotal trials. Both studies are described in Table 6.

ETHOS

ETHOS was a multinational, 52-week, phase III, parallel-group, double-blind RCT sponsored by AstraZeneca. Patients were enrolled from Europe, Asia, North America, South America, Australia, New Zealand, and South Africa from 784 study centres (24 in Canada). This study took place from July 15, 2015 (first patient randomized) to July 26, 2019 (last patient’s last visit). Database lock was on August 9, 2019. The primary objective of ETHOS was to assess the effect of BGF MDI on the rate of moderate or severe COPD exacerbations compared with GFF MDI and BFF MDI in patients with moderate-to-severe COPD who had at least 1 exacerbation in the past year. For those patients who met the eligibility criteria, current COPD medications were reviewed and adjusted, as per the study protocol. Patients received training with an e-diary (for recording device use, symptoms, and so forth) at screening (day 1). Patients were required to demonstrate acceptable e-diary adherence during the screening period to be eligible for randomization. All patients who successfully completed the screening period of 4 weeks were randomized 1:1:1:1 to 4 treatment arms and were stratified based on exacerbation history, post-bronchodilator FEV₁, blood eosinophil count, and country. Randomization was followed by a treatment period of 52 weeks that included 5 additional clinic visits and 6 telephone calls. Overall, 8,588 patients were enrolled to receive BGF MDI 320 mcg-14.4 mcg-9.6 mcg twice daily, BGF MDI 160 mcg-14.4 mcg-9.6 mcg twice daily, BFF MDI 320 mcg-9.6 mcg twice daily, or GFF MDI 14.4 mcg-9.6 mcg twice daily. Among them, BGF MDI 160 mcg-14.4 mcg-9.6 mcg is not aligned with the Health Canada review and is not considered for the current review. Figure 2 shows a visual representation of the study design for ETHOS.

KRONOS

KRONOS was a multinational, 24-week, phase III parallel-group, double-blind RCT sponsored by AstraZeneca. Patients were enrolled from the US, Canada, Japan, and China from 208 study centres (9 in Canada). The trial took place from August 20, 2015 (first patient randomized) to January 5, 2018 (last patient’s last visit). Database lock was on May 8, 2018. The primary objective of KRONOS was to assess the effect of BGF MDI and Symbicort TBH on lung function in patients with moderate-to-very-severe COPD. For those patients who met the eligibility criteria, all prohibited medications were discontinued with predetermined washout periods, and the allowed medications were adjusted as per the study protocol. Patients received training with an e-diary (for recording device use, symptoms, and so forth) at screening (day 1). All patients who successfully completed the screening period of 4 weeks were randomized in a 2:2:1:1 scheme to 4 treatment arms and were stratified based on reversibility to salbutamol (Ventolin) hydrofluoroalkane, country, and disease severity. Randomization was followed by a treatment period of 24 weeks, and each participant attended 10 study visits and a follow-up call 14 days after the last visit. Overall, 1,902 patients were randomized to receive BGF MDI 320 mcg-14.4 mcg-9.6 mcg twice daily, GFF MDI 14.4 mcg-9.6 mcg twice daily, BFF MDI 320 mcg-9.6 mcg twice daily, or BUD-FOR DPI through Symbicort TBH 400 mcg-12 mcg. Among them, the budesonide plus formoterol (Symbicort) TBH arm was designed as open label.

The KRONOS trial was designed based on local regulatory agency requirements; therefore, 3 different registration approaches were used, based on region. The approaches were based on requirements in Japan and China, the EU and Canada, and the US. The treatment comparison of interest, end points, and analysis time frames were different for each region. This review focused on the EU-Canada approach.

Figure 3 shows a visual representation of the study design for KRONOS.

Figure 2: ETHOS Study Design

BFF = budesonide-formoterol fumarate; BGF = budesonide-glycopyrronium-formoterol fumarate; BD = bronchodilator; GFF = glycopyrronium-formoterol fumarate; MDI = metered-dose inhaler; V = visit.

Source: ETHOS Clinical Study Report.

Figure 3: KRONOS Study Design

BFF = budesonide-formoterol fumarate; BGF = budesonide-glycopyrronium-formoterol fumarate; BID = twice daily; GFF = glycopyrronium-formoterol fumarate; HFA = hydrofluoroalkane; MDI = metered-dose inhaler; Symb = Symbicort; TBH = Turbuhaler; V = visit.

Source: KRONOS Clinical Study Report.

Populations

Inclusion and Exclusion Criteria

The study population for ETHOS and KRONOS consisted of current or former smokers with COPD (diagnosed based on American Thoracic Society and European Respiratory Society definitions) who were 40 to 80 years old and were on COPD maintenance therapy with 2 or more inhaled medications for at least 6 weeks. Other inclusion criteria were an FEV₁ to FVC ratio of less than 0.70 and a score of 10 or more on the COPD Assessment Test (CAT). The ETHOS trial enrolled only patients who had a documented history of at least 1 exacerbation within the previous year. On visit 1, ETHOS required the FEV₁ to be less than 65% of the predicted normal value for the patient to be eligible, whereas the required FEV₁ for KRONOS was less than 80% of the predicted normal. Exclusion criteria across the 2 trials were patients with: a current diagnosis of asthma, poorly controlled COPD (defined as corticosteroids or antibiotics use within the previous 6 weeks or hospitalization within the previous 3 months for KRONOS and 6 weeks for ETHOS), COPD associated with alpha₁-antitrypsin deficiency, other respiratory disorders, lung volume reduction within 6 months of screening, other clinically significant diseases, or pregnancy. Patients with cardiovascular diseases such as unstable ischemic heart disease, left ventricular failure, congestive heart failure, clinically significant arrhythmias, and uncontrolled hypertension were also excluded from the trials.

Baseline Characteristics

The baseline characteristics were relatively balanced between arms for each study. The mean (standard deviation [SD]) ages of the patients were 64.7 (7.6) and 65.2 (7.7) for ETHOS and KRONOS, respectively. In ETHOS, 40.3% of patients were female compared with 28.8% female in KRONOS. In KRONOS, there was a higher proportion of Asian patients (44.9%), presumably due to the number of study centres in Japan and China. In the ETHOS trial, 60% of the patients had severe COPD and 28.5% had moderate COPD based on GOLD grade. In KRONOS, the proportion of patients with moderate and severe COPD were 49.1% and 42.9%, respectively.

Most of the patients (74.4%) in KRONOS did not have a history of COPD exacerbation within the previous 12 months, whereas more than 1-half (56.5%) of ETHOS patients had 2 or more COPD exacerbations within the year before screening. Additionally, approximately 80% of ETHOS patients and 71% of KRONOS patients were taking an ICS at baseline. Table 7 summarizes the baseline characteristics for each trial.

Interventions

In ETHOS, the study treatments were BGF 320 mcg-14.4 mcg-9.6 mcg, BFF 320 mcg-9.6 mcg, or GFF 14.4 mcg-9.6 mcg. It was administered as 2 oral inhalations twice daily using an MDI that contained 120 inhalations. Treatment duration was 52 weeks. Patients were provided with and trained to use an e-diary in which time of dosing was to be recorded along with symptoms and use of rescue medications. It was unclear whether the study patients were given training on how to use the inhaler.

In KRONOS, the study treatments were BGF 320 mcg-14.4 mcg-9.6 mcg, GFF 14.4 mcg-9.6 mcg, BFF 320 mcg-9.6 mcg, or BUD-FOR 400 mcg-12 mcg inhalation powder. In the double-blinded study arms, BGF 320 mcg-14.4 mcg-9.6 mcg, GFF 14.4 mcg-9.6 mcg, or BFF 320 mcg-9.6 mcg were administered as 2 oral inhalations twice daily using an MDI that contained 120 inhalations. In the open-label study arm, BUD-FOR inhalation powder was administered as 2 inhalations twice daily using a DPI using EU-sourced Symbicort TBH. Treatment duration was 24 weeks. Patients were provided with and trained to use an e-diary in which time of dosing was to be recorded along with symptoms and use of rescue medications.

In both trials, salbutamol (provided as Ventolin hydrofluoroalkane 90 mcg MDI) was used as rescue medication. No other COPD rescue medications were allowed. COPD medications used by the patients at screening were discontinued and a minimum washout period was observed before starting study treatments, as shown in Table 8. These medications were also not allowed to be used during the study. Medications such as systemic corticosteroids were allowed for the treatment of COPD exacerbations for a maximum of 14 days. Other prohibited medications during the study period included drugs that could prolong the QT interval, non-selective beta-blocking drugs, cardiac arrhythmic class Ia and III drugs, anticonvulsants, tricyclic antidepressants, monoclonal antibodies, mono-amine oxidase inhibitors, anti–tumour necrosis factor-alpha antibodies, and Chinese herbal therapies.

Outcomes

A list of efficacy end points identified in the CADTH review protocol that were assessed in the clinical trials included in this review is provided in Table 9. These end points are further summarized in Table 9. A detailed discussion and critical appraisal of the outcome measures is provided in Appendix 4.

The primary outcome in ETHOS was the rate of moderate or severe exacerbations. COPD exacerbation was defined as a change from usual COPD symptoms that lasted for 2 or more days beyond normal variation, was acute in onset, and required a change in medication. An exacerbation was considered moderate if it required the use of systemic corticosteroids and-or antibiotics for at least 3 days. An exacerbation was considered severe if it resulted in COPD-related hospitalization for 24 hours or more or in a COPD-related death. Rate of exacerbations was considered a secondary outcome in KRONOS. Time to first moderate or severe COPD exacerbation and time to death were secondary outcomes of interest in ETHOS. In KRONOS, the rate of moderate or severe exacerbations was considered a secondary outcome.

As described previously, outcomes in KRONOS were different based on the registration requirement for that region. This review reports the outcomes per the EU-Canada approach. The primary outcome in KRONOS, lung function measured using spirometry, varied depending on the registration approach. For the EU-Canada registration approach, the primary outcome was the AUC_0-4 for the FEV₁ (for BGF MDI 320 versus BFF MDI, and BGF MDI 320 versus BUD-FOR DPI) and the change from baseline in the morning pre-dose trough FEV₁ over 24 weeks (for BGF MDI 320 versus GFF MDI). Peak change from baseline in FEV₁ within 4 hours post dosing over 24 weeks was considered a secondary outcome. In ETHOS, lung function measured as FEV₁ AUC_0-4 over 24 weeks was assessed as 1 of the primary outcomes in the 4-hour PFT substudy. FEV₁ is the volume of air that, after a full inspiration, can be forcibly expired in 1 second. Trough FEV₁ is recognized as a component of the GOLD classification of airflow limitation severity in COPD.³⁹ The generally accepted clinically important change in FEV₁ is between 0.10 L and 0.14 L.⁴⁰

The change from baseline in the SGRQ total score was evaluated as a secondary outcome in ETHOS (at 52 weeks) and KRONOS (over 24 weeks). SGRQ measures health status (HRQoL), with scores ranging from 0 to 100. A higher score on the SGRQ indicates a poorer level of HRQoL and decreases in score are indicative of improvement in HRQoL. A decrease of 4 points from baseline is considered a clinically meaningful improvement.^40,41 The percentage of patients who achieved this MCID were considered to be responders, which was a secondary outcome in both trials.

HRQoL was also measured using change from baseline in the EuroQol 5-Dimensions (EQ-5D) 5-Levels questionnaire (EQ-5D-5L) and was evaluated as “other” outcome in ETHOS and KRONOS. EQ-5D-5L measures the patient’s immediate situation using the EuroQol Visual Analogue Scale (EQ VAS) and 5 dimensions. The EQ VAS records the respondent’s self-rated health on a vertical VAS, where the end points are 100 (“best imaginable health state”) and 0 (“worst imaginable health state”). The MCID of the EQ VAS is estimated to be from 6.5 (anchored to the Chronic Respiratory Questionnaire [CRQ] total using receiver operating characteristic [ROC] methods) to 10.1 (distribution SD = 0.5).⁴²

Symptoms in patients were measured using the change from baseline EXACT score in ETHOS, and the TDI focal score and E-RS score in both trials. EXACT is a valid patient-reported outcome consisting of 14 items (MCID not yet established).⁴³ It was evaluated as a secondary outcome in ETHOS. The TDI, considered a secondary outcome in ETHOS and KRONOS, measures the change in dyspnea severity from the baseline as established by the Baseline Dyspnea Index (BDI). The TDI consists of 24 items that are graded; lower scores indicate more deterioration related to an increase in severity of dyspnea from baseline. The TDI focal score is composed of 3 different scales for functional impairment, magnitude of task, and magnitude of effort. The TDI focal score is calculated as the sum of the 3 individual scores and then divided by 2. The range of the TDI focal score is −9 to 9.0.⁴⁴ A change in 1 point is considered clinically meaningful.^40,45 The E-RS is an 11-item subset of the EXACT questionnaire to evaluate respiratory symptoms in COPD, with higher scores indicating greater severity.⁴⁶ The E-RS was evaluated as a secondary outcome in KRONOS (over 24 weeks) and as an “other” outcome in ETHOS.

Use of rescue medications was calculated as the change from baseline in the number of average daily puffs of Ventolin hydrofluoroalkane taken by the participant. Use of rescue medications was considered as an “other” outcome in ETHOS (over 52 weeks) and was a secondary outcome in KRONOS (over 24 weeks).

Several harms outcomes were reported, including AEs, SAEs, withdrawals due to AEs, and mortality. Notable harms were reported for anticholinergic AEs, corticosteroid-associated AEs, cardiovascular effects, and pneumonia.

Lastly, health care resource utilization, such as the number of hospitalizations, ICU visits, emergency room visits, and the number of workdays missed, were measured and reported descriptively by both trials, as well.

Statistical Analysis

Summaries of the statistical analysis of the efficacy end points in ETHOS and KRONOS are presented in Table 10 and Table 11 respectively.

ETHOS

Sample size was estimated based on the primary outcome, i.e., rate of moderate-to-severe COPD exacerbations. It was assumed the average exposure would be 0.83 years, and the number of moderate or severe COPD exacerbations per year would be 1.142 in the BGF MDI 320 group, 1.344 in the GFF MDI group, and 1.344 in the BFF MDI group. A relative reduction of 15% was assumed between BGF MDI 320 and the comparator groups. With type I error controlled at a 1-sided alpha level of 0.025, enrolling 8,400 patients in a 1:1:1:1 ratio would result in a probability of demonstrating differences between the intervention and comparators of approximately 93% (96% for each comparison). A blinded sample-size reassessment was conducted before the planned interim analysis, which increased the required sample size from 8,000 to 8,400. All comparisons relevant to this review were done for superiority. The order of hypothesis testing for control of type I error is shown in Figure 4. The primary estimand of interest was the efficacy estimand, which was defined as the effect of the randomized treatments in all patients, assuming they continued the treatments for the duration of the study regardless of actual compliance. For the secondary end points, the main analysis used the efficacy estimand except for the end point time to death, which used the treatment policy estimand. The treatment policy estimand is the effect of randomized treatment over the study period, regardless of whether randomized treatment is continued.

The primary end point was analyzed using negative binomial regression, with the logarithm of time at risk as an offset term. All missed data were assumed to be missing completely at random or missing at random. The comparisons were adjusted for baseline post-bronchodilator percent predicted FEV₁, log baseline blood eosinophil count, baseline COPD exacerbation history, region, and ICS use at screening. For the efficacy estimand, the time at risk was the duration of exposure to treatment, which was calculated as the days from first dose to date of treatment completion (both days inclusive), or the days from first dose to the day after premature treatment discontinuation. A sensitivity analysis for missing data was conducted using a tipping-point analysis that used a maximum delta value of 1.5 exacerbations per year. Multiple imputation techniques were used to impute the mission data, up to 10 imputations per the tipping-point analysis set. For the primary outcome, a pre-specified subgroup analysis based on several demographic categories was performed. The subgroups relevant for the current review were baseline blood eosinophil count and COPD exacerbation history.

Time to first moderate or severe COPD exacerbation and time to death were analyzed using a Cox regression model. The model for time to first moderate or severe COPD exacerbation included treatment, baseline post-bronchodilator percent predicted FEV₁, log baseline blood eosinophil count, baseline COPD exacerbation history, region, and ICS use at screening. Rates of severe outcomes and the primary outcome were analyzed similarly (negative binomial regression). SGRQ total score, TDI focal score, EXACT score, E-RS score, and use of rescue medications were analyzed using a linear repeated measures analysis of covariance (ANCOVA) model. An unstructured correlation matrix was used to model additional autocorrelation within a patient. The SGRQ responder rate (percentage of patients with an improvement of ≥ 4 points) was analyzed using logistic regression with PROC GENMOD. Robustness of results to missing data was assessed using a tipping-point analysis. All covariates used in the analysis of each of the secondary outcomes are listed in Table 10.

Other end points were analyzed using log-rank tests or ANCOVA, without adjustments for multiplicity.

Figure 4: Order of Hypothesis Testing for Type I Error Control (ETHOS)

AUC_0-4 = area under the curve from 0 to 4 hours post dose; BFF = budesonide-formoterol fumarate; BGF = budesonide-glycopyrronium-formoterol fumarate; COPD = chronic obstructive pulmonary disease; COPDX = chronic obstructive pulmonary disease exacerbations; COPDHX = history of chronic obstructive pulmonary disease exacerbations; FEV₁ = forced expiratory volume in 1 second; GFF = glycopyrronium-formoterol fumarate; MDI = metered-dose inhaler; M-S = moderate or severe; vs. = versus .

*All comparisons of BGF MDI 160-14.4-9.6 mcg vs BFF MDI were for non-inferiority using the PP Estimand followed by superiority. Superiority was not required to advance to the next comparison. All other comparisons were for superiority and used the Efficacy Estimand unless otherwise stated.

**Hochberg-controlled secondary end points: Time to first moderate or severe COPD exacerbation, rescue use, rate of severe COPD exacerbation, SGRQ (ex-US), SGRQ responders (US), EXACT Total (ex-US), and TDI (ex-US).

Source: ETHOS Clinical Study Report. .

KRONOS

Sample size was estimated based on the primary outcomes. Based on these assumptions, the sample size had:

• 99% power to detect a difference of 75 mL between BGF MDI and BFF MDI in FEV₁ AUC_0-4 over 24 weeks

• 96% power to detect a difference of 35 mL between BGF MDI and GFF MDI in morning pre-dose trough FEV₁ over 24 weeks (92% power over weeks 12 to 24)

• 97% power to detect a difference of 50 mL between BGF MDI and BFF MDI in morning pre-dose trough FEV₁ over weeks 12 to 24

• 96% power to demonstrate noninferiority of BFF MDI to Symbicort TBH in morning pre-dose trough FEV₁ over 24 weeks (92% power over weeks 12 to 24), based on a margin of 50 mL (1-sided alpha = 0.025); a sample size of 1,800 patients (in a 2:2:1:1 ratio) was estimated with a type I error control at a 2-sided alpha level of 0.05.

All comparisons relevant to this review were done for superiority. The type I error control approach varied based on the region and registration. For the EU-Canada approach, the order of hypothesis testing for type I error control is shown in Figure 5. The primary estimand of interest was the efficacy estimand, which is the effect of the randomized treatments in all patients, assuming they continued the treatments for the duration of the study regardless of actual compliance.

One of the primary outcomes, FEV₁ AUC_0-4, was analyzed using a repeated measures linear mixed model. The covariates in the analysis were baseline FEV₁, percent reversibility to Ventolin hydrofluoroalkane, baseline eosinophil count, treatment, visit, treatment by visit interaction, and ICS use at screening. The other primary outcome, change from baseline in morning pre-dose trough FEV₁, was also analyzed using a repeated measures linear mixed model. The covariates in the analysis were treatment, visit, treatment by visit interaction, ICS use at screening, baseline FEV₁, baseline eosinophil count, and percent reversibility to Ventolin hydrofluoroalkane. Correlation within a patient was modelled using an unstructured variance-covariance matrix. Sensitivity analyses for missing data in both outcomes were conducted using tipping-point analyses. Multiple imputation techniques were used to impute the missing data under the pattern mixture model framework.

The secondary end points related to FEV₁ and the primary outcomes were analyzed similarly. Peak FEV₁, SGRQ total score, TDI focal score, E-RS score, and use of rescue medications were analyzed using a linear repeated measures model. The rate of moderate-to-severe exacerbations was analyzed using a negative binomial model and the SGRQ responder rate was analyzed using a logistic regression model. Other end points were analyzed using a repeated measures linear mixed model or Cox regression models, as appropriate, without adjustments for multiplicity. Other details about the analysis of each of the outcomes are listed in Table 11.

Lastly, to compare the primary and secondary continued end points (in the modified intention-to-treat [ITT] population) in each approach, a correlation analysis was conducted.

Figure 5: Order of Hypothesis Testing for Type I Error Control (KRONOS)

AUC_0-4 = area under the curve from 0 to 4 hours post dose; BFF = budesonide-formoterol fumarate; BGF = budesonide-glycopyrronium-formoterol fumarate; FEV₁ = forced expiratory volume in 1 second; GFF = glycopyrronium-formoterol fumarate; MDI = metered-dose inhaler; TBH = Turbuhaler; tFEV₁ = trough forced expiratory volume in 1 second; vs. = versus .

Source: KRONOS Clinical Study Report.

Analysis Populations

ETHOS and KRONOS included the following analysis populations:

• The ITT population included all patients who were randomized to treatment and received any amount of the study drug.

• The modified ITT population included all patients who were randomized with post-randomization data obtained before discontinuation from the study drug.

• The per-protocol population included all patients with post-randomization data obtained before any major protocol deviation.

• The safety population included all patients who were randomized to treatment and received any amount of the study drugs (similar to ITT). Patients were analyzed according to treatment received rather than randomized.

• The rescue Ventolin use population included all patients in the ITT population who reported an average baseline daily use of 1 or more puffs of rescue Ventolin.

Results

Patient Disposition

Patient dispositions in ETHOS and KRONOS are summarized in Table 12. Over 8,500 patients were randomized to the 4 treatment arms in the ETHOS trial, whereas 1,902 patients underwent randomization in KRONOS. The proportion of patients who discontinued each trial was balanced between trial arms. In ETHOS, 20.6% to 25.9% of patients discontinued compared with 11.4% to 16.4% in KRONOS. This discrepancy could be due to the longer trial duration of ETHOS (52 weeks). The most common reasons for study discontinuation were AEs (5.5% to 6.9%) in ETHOS, lack of efficacy (4.8% to 8%) in ETHOS, and patient discretion (2.2% to 6.1%) in KRONOS.

Among the 16,033 and 3,047 patients screened for ETHOS and KRONOS, respectively, 7,455 (46.5%) and 1,139 (27.4%) patients were not randomized due to various reasons. The most common reason for screening failure was severity of disease in both ETHOS and KRONOS.

Exposure to Study Treatments

In ETHOS, the mean duration of treatment was similar across the treatment groups, ranging from 304.2 days to 322.9 days. Overall, 78.9% of patients were exposed to the study drugs for 48 weeks or more. In KRONOS, the mean duration of treatment was similar across the treatment groups, ranging from 153.8 days to 158.5 days, with 76.5% of patients overall exposed to the study drugs for 24 weeks or more.

In both trials, treatment adherence was high and balanced across treatment groups. Compliance to treatment was measured as the number of puffs of study drug taken per day divided by the total expected number of puffs taken per day. Between trials, KRONOS had a slightly higher treatment compliance (overall mean [SD] = 95.2 [9.0]) compared with ETHOS ||||||||||||||||||||||||||||||||||||||||||, which could be due to the longer study duration of the ETHOS trial.

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Efficacy

Only those efficacy outcomes and analyses of subgroups identified in the review protocol are reported subsequently. Efficacy results from the included studies are summarized in Table 13 and Table 14. See Appendix 3 for detailed efficacy data.

Mortality

The outcome mortality rates (all causes or COPD specific) were evaluated by ETHOS but not KRONOS.

Time to death (all causes) was a secondary efficacy outcome in ETHOS. The time to death was defined as the number of days from randomization to the date of death. Numerically, there were fewer deaths in the BGF MDI 320 group compared with GFF MDI and BFF MDI. The risk of death (all causes) was lower during treatment with BGF MDI 320 relative to GFF MDI (HR = 0.544; 95% CI, 0.340 to 0.870; P = 0.0111), but not different relative to BFF MDI (HR = 0.782; 95% CI, 0.472, to 1.296; P = 0.3401) as assessed by the Cox proportional hazards model.

Analysis of the final retrieved dataset (99.6% of the ITT population), which included missing week 52 vital status for 354 patients obtained after the database lock, showed similar results as the original analysis.³⁰ The risk of death (all causes) with the supplemental vital status data (on and off treatment) was nominally lower during treatment with BGF MDI 320 relative to GFF MDI (HR = 0.51; 95% CI, 0.33 to 0.80) and was numerically lower relative to BFF MDI (HR = 0.72; 95% CI, 0.44 to 1.16). The Kaplan–Meier curve for the time of death (all causes) is shown in Figure 6.

Figure 6: Time to Death (All-cause), ETHOS

BFF = budesonide-formoterol fumarate; BGF = budesonide-glycopyrronium-formoterol fumarate; GFF = glycopyrronium-formoterol fumarate.

Source: Martinez et al. (2021)³⁰. Copyright 2021 American Thoracic Society. Reprinted in accordance with Creative Licence Attribution License CC BY 4.0.³⁰

Health Care Resource Utilization

In ETHOS and KRONOS, health care resource utilization outcomes were reported descriptively. The results were relatively balanced within trial arms. Emergency room visits ranged from |||||||||||||||||| in ETHOS and ||||||||||||||||||||||||||||. The proportion of patients who required hospitalization with time spent in ICU ranged from |||||||||||||| in ETHOS to |||||||||||| in KRONOS. Additional results for health care resource utilization outcomes are summarized in Appendix 3.

Exacerbations

In ETHOS, the rate of moderate or severe COPD exacerbations during the 52-week study was lower for BGF MDI 320 (1.02 per year) than GFF MDI (1.24 per year) and BFF MDI (1.15 per year). The rate ratio of moderate or severe exacerbations for BGF MDI 320 versus GFF MDI was 0.76 (95% CI, 0.69 to 0.83; P < 0.0001); for BGF MDI 320 versus BFF MDI, the rate ratio was 0.87 (95% CI, 0.79 to 0.95; P = 0.0027). The HRs for the time to first moderate or severe COPD exacerbation was 0.880 (95% CI, 0.807 to 0.959) for BGF MDI 320 versus GFF MDI, and 0.887 (95% CI, 0.814 to 0.966) for BGF MDI 320 versus BFF MDI (Table 13). The Kaplan-Meir curve for the time to first moderate or severe COPD exacerbation is shown in Figure 7.

The rate of severe exacerbations during the 52-week study was lower for BGF MDI 320 (0.14 per year) than for GFF MDI (0.16 per year) and BFF MDI (0.18 per year). Compared with BFF MDI, BGF MDI 320 was associated with a reduction in the rate of severe exacerbation (rate ratio = 0.80; 95% CI, 0.66 to 0.97) and an HR for the time to the first severe exacerbation of 0.805 (95% CI, 0.673 to 0.964). However, no similar statistically significant reductions were found in the BGF MDI 320 group compared with the GFF MDI group (Figure 13). The Kaplan-Meir curve for the time to first severe COPD exacerbation is shown in Figure 8.

In KRONOS, the annualized rate of moderate or severe COPD exacerbations during the 24-week study was lower for BGF MDI 320 (0.46 per year) than for GFF MDI (0.95 per year), BFF MDI (0.56 per year), and BUD-FOR DPI (0.55 per year). The rate ratio of moderate-to-severe COPD exacerbations of BGF MDI versus GFF MDI was statistically significant (rate ratio = 0.48; 95% CI, 0.37 to 0.64; P < 0.0001). . The HR for the time to first moderate or severe exacerbation also favoured BGF MDI 320 compared with GFF MDI (HR = 0.593; 95% CI |||||||||||||||||), which was statistically significant. However, the HRs were not statistically significantly different for BGF MDI 320 versus BFF MDI, or BGF MDI 320 versus BUD-FOR DPI (Table 14).

The annualized rates of severe exacerbations during the 24-week study were 0.05 per year for BGF MDI 320, 0.13 per year for GFF MDI, 0.05 per year for BFF MDI, and 0.07 per year for BUD-FOR DPI. Compared with GFF MDI, BGF MDI 320 was associated with a reduction in the rate of severe exacerbation (rate ratio = 0.36; 95% CI = 0.18 to 0.70) and in the HR for the time to first severe exacerbation (HR = 0.473; 95% CI, 0.263 to 0.850). However, no similar statistically significant reductions were found in the BGF MDI 320 group compared with the BFF MDI group or the open-label BUD-FOR DPI group (Table 14).

Figure 7: Time to First Moderate or Severe COPD Exacerbation (ETHOS)

BFF = budesonide-formoterol fumarate; BGF = budesonide-glycopyrronium-formoterol fumarate; COPD = chronic obstructive pulmonary disease; GFF = glycopyrronium-formoterol fumarate; MDI = metered-dose inhaler.

Source: ETHOS Clinical Study Report.

Figure 8: Time to First Severe COPD Exacerbation (ETHOS)

Figure 8 was removed at the request of the sponsor because it contained confidential information.

Health-Related Quality of Life

HRQoL was assessed using the change from baseline SGRQ score as a secondary outcome in ETHOS and KRONOS. HRQoL was also measured using the change from baseline in EQ-5D-5L (EQ VAS) score as an “other” outcome in both trials. The results of the EQ-5D-5L are summarized in Appendix 3.

In ETHOS, all treatment arms showed within-group improvement in SGRQ at 24 weeks from baseline. Comparing the change from baseline in SGRQ score between groups, the BGF MDI 320 arm had more improvement than the GFF MDI arm, with a mean difference of −1.62 units (95% CI, −2.27 to −0.97). Similarly, the BGF MDI 320 arm also had more improvement than the BFF MDI arm, with a mean difference of −1.38 (95% CI, −2.02 to −0.73). These differences were statistically significant (P < 0.001), but not clinically significant (MCID = 4 units). The proportion of patients achieving an improvement of at least 4 points on the SGRQ, termed as responders, was higher for BGF MDI 320 compared with GFF MDI (difference = 7.60%; 95% CI, 4.52 to 10.68) and BFF MDI (difference = 5.47%; 95% CI, 2.39 to 8.55) (Table 13).

In KRONOS, all treatment arms showed within-group improvement in SGRQ over 24 weeks from baseline. Comparing the change from baseline in SGRQ score between groups, the BGF MDI 320 arm had more improvement than the GFF MDI arm, with a mean difference of −1.22 units (95% CI, −2.30 to −0.15). This difference was statistically significant (P < 0.001), but not clinically significant (MCID = 4 units). The BGF MDI 320 arm also had more improvement than the BFF MDI arm, with a mean difference of −0.45 (95% CI, −1.78 to 0.87) as well as the BUD-FOR DPI arm, with a mean difference of −1.26 (95% CI, −2.58 to 0.06). These differences were not statistically or clinically significant. The percentage of responders was higher for BGF MDI compared with GFF MDI |||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| (Table 14).

Lung Function

In ETHOS, lung function measured as FEV₁ AUC_0-4 over 24 weeks was assessed as 1 of the primary outcomes in the 4-hour PFT substudy (N = 3,088) comparing BGF MDI 320 versus BFF MDI. BGF MDI 320 was associated with an improvement in the mean FEV₁ AUC_0-4 at week 24 compared with BFF MDI, with the mean difference being 99 units ||||||||||||||||||||||||. These improvements were statistically significant. The other primary outcome of the PFT substudy was morning pre-dose trough FEV₁ for the comparison of BGF MDI 320 versus GFF MDI. The change from baseline in the morning pre-dose trough FEV₁ over 24 weeks for BGF MDI 320 was significantly higher than for GFF MDI (mean difference = 35 mL: 95% CI, 12 to 57). This improvement in FEV₁ was statistically significant but of uncertain clinical significance (MCID = 0.10 L to 0.14 L).⁴⁰

In KRONOS, lung function measured as FEV₁ AUC_0-4 over 24 weeks was the primary outcome, for the comparison of BGF MDI 320 versus BFF MDI, and BGF MDI 320 versus BUD-FOR DPI (in the EU-Canada approach). Change from baseline in the morning pre-dose FEV₁ over 24 weeks was the primary outcome for the comparison of BGF MDI 320 versus GFF MDI (in the EU-Canada approach). In the trial, BGF MDI 320 showed improvement in the FEV₁ AUC_0-4 over 24 weeks compared with BFF MDI (least squares mean = 104 mL; 95% CI, 77 to 131) and compared with BUD-FOR DPI (least squares mean = 91; 95% CI, 64 to 117) (Table 14). These improvements were statistically significant. The change from baseline in morning pre-dose trough FEV₁ at 24 weeks for BGF MDI 320 compared with GFF MDI was 22 mL (95% CI, 4 to 39). This improvement in FEV₁ was statistically significant but not clinically significant (MCID = 0.10 L to 0.14 L)⁴⁰ (Table 14).

Symptoms

Symptoms in study participants were measured using the change from baseline in EXACT in ETHOS, and change in TDI focal score and E-RS in both trials.

In ETHOS, the difference in least squares mean for the TDI focal score showed improvement in BGF MDI 320 compared with both GFF MDI (difference = 0.40 units; 95% CI, 0.24 to 0.55) and BFF MDI (difference = 0.31 units; 95% CI, 0.15 to 0.46). The improvements were not clinically significant (MCID = 1 unit). There were statistically significant improvements in the change from baseline in the EXACT total score in BGF MDI 320 compared with GFF MDI (difference = −1.14 units; 95% CI, −1.64 to −0.65) and BFF MDI (difference = −1.04; 95% CI, −1.53 to −0.55). As for the change from baseline in E-RS score over 52 weeks, BGF MDI 320 was associated with a statistically significant improvement compared with GFF MDI |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| and BFF MDI ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||; however, these improvements were not clinically meaningful (MCID = 3.35 units) (Table 13).

In KRONOS, the difference in the least squares mean of the TDI focal score showed improvement in BGF MDI 320 compared with the open-label BUD-FOR DPI (difference = 0.46 units; 95% CI, 0.156 to 0.766). The improvement was not clinically meaningful (MCID = 1 unit). Compared with GFF MDI and with BFF MDI, BGF MDI 320 did not demonstrate a clinically or statistically significant improvement in the TDI focal score over 24 weeks (Table 13). As for the change from baseline in E-RS score over 24 weeks, BGF MDI 320 was associated with a slight improvement in scores compared with GFF MDI (difference = −0.38), BFF MDI (difference = −0.16), and the open-label BUD-FOR DPI (difference = −0.16); however, these improvements were not clinically or statistically significant (MCID = 3.35 units) (Table 14).

Use of Rescue Medication

In ETHOS and KRONOS, the evaluation of the average daily number of puffs of rescue medication over 24 weeks was restricted to the rescue Ventolin use population. In ETHOS, BGF MDI 320 was associated with a significant reduction in daily rescue medication use compared with GFF MDI (difference = −0.51 puffs per day; 95% CI, −0.68 to −0.34) and BFF MDI (difference = −0.37 puffs-day; 95% CI, −0.54 to −0.20) (Table 13).

In KRONOS, BGF MDI 320 was associated with small reductions in daily rescue medication use compared with GFF MDI (difference = −0.25 puffs-day; 95% CI, −0.60 to 0.09) and BFF MDI (difference = −0.24 puffs-day; 95% CI, −0.65 to 0.18). However, compared with BUD-FOR DPI, BGF MDI 320 was found to be associated with a slightly higher average daily use of rescue medication (difference = 0.23 puffs per day; 95% CI, −0.17 to 0.63). These differences were not statistically significant (Table 14).

Productivity

This outcome was not evaluated by any of the included trials; therefore, a summary could not be provided.

Harms

Only those harms identified in the review protocol are reported subsequently. See Table 15 for detailed harms data.

Adverse Events

Within each trial, AEs were similar across treatment arms. In ETHOS, the proportions of patients who reported at least 1 AE at 52 weeks were 63.8% in the BGF MDI 320 arm, 61.7% in the GFF MDI arm, and 64.5% in the BFF MDI arm. The most common AEs were nasopharyngitis (9.4% to 11.3% across arms) and COPD (9.5% to 11.3%). In KRONOS at 24 weeks, AEs were reported in 60.7% of patients in the BGF MDI 320 arm, 61.4% of those in the GFF MDI arm, 55.7% of those in the BFF MDI arm, and 57.5% of those in the BUD-FOR DPI arm. The most common AEs were upper respiratory tract infections (5.7% to 10.2% across arms) and nasopharyngitis (6.6% to 9.4%). AEs that occurred in 2% or more of the population are presented in Table 15.

Serious Adverse Events

Within each trial, SAEs were similar across treatment arms. In ETHOS, the proportion of patients who reported 1 or more SAE were 19.9% in the BGF MDI 320 arm, 20.4% in the GFF MDI arm, and 20.6% in the BFF MDI arm at 52 weeks. The most common SAEs were COPD (9.4% to 11.3%) and pneumonia (1.6% to 2.8%). In KRONOS at 24 weeks, SAEs were reported in 8.6% of patients in the BGF MDI 320 arm, 10.9% of those in the GFF MDI arm, 6.7% of those in the BFF MDI arm, and 9.1% of those in the BUD-FOR DPI arm. The most common SAEs were COPD (2.5% to 5.1% across arms) and pneumonia (0 to 1.3%). Other SAEs that were reported in the trials are presented in Table 15.

Withdrawals Due to Adverse Events

Within each trial, the number of patients who stopped treatment due to AEs was relatively similar across the treatment arms. In ETHOS, 5.6% of patients in the BGF MDI 320 arm, 6.9% of patients in the GFF MDI arm, and 6.6% of patients in the BFF MDI were reported as discontinuing treatment due to AEs, most commonly due to COPD (1% to 2.1%). In KRONOS, the percentage of patients who discontinued treatment due to AEs was 4.7% in the BGF MDI arm, 4.8% in the GFF MDI arm, 3.5% in the BFF MDI arm, and 3.5% in the open-label BUD-FOR DPI arm. The most common reason for discontinuing treatment was COPD (0.3% to 1.3%). Other reasons for withdrawal from the study treatments reported in the trials are presented in Table 15.

Mortality

In ETHOS, 1.3% of the study population (n = 112) died on treatment. The most common causes of death related to cardiovascular causes (n = 33, 0.4%) and respiratory causes (n = 23, 0.3%). In KRONOS, there were 12 deaths (0.6%) reported on treatment. Among them, 3 were due to cardiovascular causes and 3 were due to respiratory causes.

Notable Harms

Cardiovascular events affected each arm similarly across trials. Among them, 1.1% to 2.1% of patients within each arm of ETHOS and 0.3% to 0.6% of patients within each arm of KRONOS reported major adverse cardiovascular events. Other notable cardiovascular AEs are presented in Table 15.

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Short-term corticosteroid-related AEs were reported similarly across trials. AEs such as weight gain and insomnia were reported in less than 2% of patients in ETHOS and |||||||||||||||||||||||||||||||||||||||||. Oral candidiasis was reported by 2.3% of ETHOS patients and ||||||||||||||||||||||||||||||||. Long-term corticosteroid AEs such as osteoporosis and osteopenia were reported in a small number of patients in similar numbers across the arms of each trial. Across trials, around 2% of patients in each arm of ETHOS and |||||||||||| of patients in each arm of KRONOS reported bone fractures (overall). Other notable harms associated with corticosteroid use is presented in Table 15.

Critical Appraisal

Internal Validity

ETHOS and KRONOS were randomized, double-blind, parallel-group trials. In both trials, randomization was centralized using an interactive web response system. In ETHOS, randomization was stratified by exacerbation history, post-bronchodilator FEV₁, blood eosinophil count, and country. In KRONOS, randomization was stratified by reversibility to Ventolin hydrofluoroalkane, country, and disease severity. Patients in the different intervention groups were recruited over the same period of time and from the same centres. Measures were taken to ensure blinding throughout the studies. All study drugs were administered by oral inhalation devices that were provided by the sponsor as MDIs. However, in the KRONOS trial, the BUD-FOR (Symbicort TBH) arm could not be blinded, as it used the TBH inhaler. Overall, the occurrence of AEs was similar across the treatment arms in both trials. The chances of inadvertent unblinding due to AEs were low, given the similarities in the events across the treatment groups.

The baseline demographics and disease characteristics were similar between treatment arms in each trial, suggesting successful randomization. The patient disposition in both trials was described clearly.

The duration of the KRONOS trial was 24 weeks, which was a reasonable duration to evaluate the pulmonary function outcome measure (e.g., trough FEV₁) and symptoms. However, for exacerbations, HRQoL, and safety outcomes, the shorter duration of the trial could have affected the results. The 52-week trial period of ETHOS was more appropriate for evaluating these outcomes, especially given the variation within the 1-year period for exacerbations.

Both trials had a 4-week screening period before randomization and the start of study treatments. There were pre-specified washout periods for the COPD medications used by the patients at screening, as well. Background medications such as rescue inhalers were standardized. However, the degree and type of training for the use of the inhalers was not described. This is particularly noteworthy for the KRONOS study, which included open-label budesonide-formoterol delivered via a different inhaler compared with the other treatment groups. Overall, treatment adherence was high and balanced across treatment arms in ETHOS and KRONOS. Despite some uncertainty as to how well patients used the inhaler devices, there did not appear to be meaningful differences between groups regarding the treatment received, other than the study treatment each patient was assigned.

The outcomes considered in the trials were validated. The primary outcomes assessed in the trials, related to trough FEV₁ (KRONOS) and the annual rate of on-treatment moderate or severe exacerbations (ETHOS), are established outcomes in evaluating drug interventions for COPD. The FEV₁ is used in clinical settings and recommended by GOLD to grade the severity of airflow limitation in patients with COPD. The generally accepted clinically important change in FEV₁ is between 0.10 L and 0.14 L.⁴⁰ The assessment of exacerbations was based on moderate exacerbations (defined as requiring treatment with oral [systemic] corticosteroids and-or antibiotics [not involving hospitalization]) and severe exacerbations (defined as requiring inpatient hospitalization); the definitions were considered acceptable by the clinical expert consulted by CADTH. In KRONOS, the primary and secondary outcomes and time points of analysis were different based on the registration approach (EU and Canada, Asia, US). The open-label study arm in KRONOS may have biased the subjective outcome results in favour of BGF MDI 320.

In ETHOS and KRONOS, all comparisons (relevant to this review) were for testing the superiority of triple therapy versus dual therapy. Appropriate statistical tests were used to analyze the study outcomes in both trials and were adequately adjusted for covariates. The primary end point in ETHOS (rate of moderate or severe COPD exacerbations) was analyzed using negative binomial regression, with the logarithm of time at risk as an offset term. Both primary outcomes in KRONOS were analyzed using a repeated measures linear mixed model. It is unknown whether the data distribution was examined for final model selection. In the presence of Poisson overdispersion, an alternative of negative binomial distribution may yield a better model fit of the count data than traditional Poisson regression. However, this needed to be tested and confirmed.

A hierarchical structure of hypothesis testing was employed for controlling type I error. The primary and secondary outcomes were analyzed, adjusting for multiplicity. Outcomes designated as “other” were analyzed without adjustments for multiplicity and should be interpreted with caution.

Approximately 22% of patients in ETHOS and 14% in KRONOS discontinued the study drug. The major reasons for early discontinuation from the trials were either AEs or lack of efficacy. In ETHOS, there were slightly more patients in the GFF and BFF arms than in the BGF arm who discontinued the trial due to lack of efficacy (8.0% or 6.4% versus 4.8%) or AEs (6.9% or 6.5% versus 5.5%). This could bias the result in favour of BGF, particularly when compared with GFF, as the percentage of patients withdrawing due to lack of efficacy was 8.0% in the GFF group compared with 4.8% in the BGF group. These percentages were relatively lower in KRONOS (lack of efficacy: 2.6% or 1.9% versus 1.6%; AEs: 4.8% or 3.5% versus 4.4%), likely because of the shorter duration of the study. The difference in efficacy is perhaps not surprising, given the stepwise treatment approach (i.e., adding medications to existing ones as the disease and symptoms progress). It has been established that stepping up to LAMA-LABA-ICS improves lung function and patient-reported outcomes and reduces exacerbations, as noted by GOLD. Most patients (> 60%) in ETHOS had severe COPD at baseline (GOLD grade 3) and had experienced 2 or more exacerbations in the previous year (> 55%). As well, one-third had been receiving an ICS-LABA and 40% had been receiving a combination LAMA-LABA-ICS at screening; only approximately 14% had been receiving dual LAMA-LABA treatment. Patients randomized to BGF MDI would have been stepping up to or staying at triple therapy, while those in the other groups mostly would have stayed on dual therapy or been stepped down to dual therapy from triple. Although testing BGF MDI against dual therapies may have been practical for establishing efficacy, it likely biased results in favour of BGF MDI, considering the baseline severity and COPD treatment history.

Generally, the robustness of the results to missing data was assessed using a tipping-point analysis under the assumption of missing not at random for the primary outcomes and most of the secondary outcomes in both trials. In ETHOS, |||||||||||||||||||||||||||||||||| missing data for the primary end point between the treatment arms. The percentage of missing data was comparable between treatment arms. In KRONOS, an attributable estimand (imputed for missing data) also showed results consistent with the efficacy estimand for the outcome change from baseline in morning pre-dose trough FEV_1. As for the other primary outcome (FEV₁ AUC_0-4), the tipping-point analysis demonstrated that the conclusions were robust to missing data. Sensitivity analyses for missing data were not conducted for SGRQ total score, TDI focal score, EXACT score, or E-RS score in ETHOS, or for TDI focal score or E-RS score in KRONOS.

Subgroup analyses were planned a priori. Based on the randomization stratification, the subgroups relevant to this report were able to maintain randomization in ETHOS. In KRONOS, additional subgroups that were not considered during randomization stratification were analyzed. It is possible that randomization was not maintained for the baseline eosinophil subgroup or exacerbation history subgroup. Other potentially important subgroups such as prior bronchodilator therapy and baseline bronchodilator reversibility were not considered in the trials.

External Validity

The ETHOS and KRONOS trials were multi-centre multinational studies that included Canadian participation, |||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||. Other inclusion criteria for patient eligibility were restrictive, based on FEV₁, reversibility, and past exacerbation (ETHOS). This could lower the generalizability of results to real-world settings. The clinical expert believed the demographics of patients in the included studies were generally consistent with that of the population that would be expected to use the drug, although the percentage of women was likely lower than expected (though a higher percentage were enrolled in ETHOS and KRONOS than in previous trials of COPD drugs). There was a higher proportion of Asian participants in the KRONOS trial compared with Canadian settings. The clinical expert noted that most of the patients (71% to 80%) used ICS-containing treatments at screening, which was higher than what is used in clinical settings for respirologists in Canada. As well, only 14% of patients had been receiving LAMA-LABA dual therapy at screening, which is not consistent with the current approach to treatment. Patient adherence was high for both trials which, measured by the ratio of daily puffs taken and the expected number of daily puffs, were 93.2 and 95.2 in ETHOS and KRONOS, respectively. However, it is uncertain how much of that could be translated to real-world clinical settings. The comparators and doses used in the trials were deemed by the clinical expert to be clinically relevant and appropriate; however, it was noted that comparisons with fixed-dose triple-therapy combinations would have been of interest. The concomitant medication prohibitions were found to be reasonable by the expert. The study-mandated treatment duration for exacerbations (maximum of 14 days) was also considered adequate by the expert.

For subgroup analysis, the cut-off for baseline eosinophil was 150 cells per mm³ in the studies. The clinical expert expressed concern that this was low, especially considering the majority of patients used ICS at screening.

The outcomes assessed in the included trials were appropriate and clinically important. Outcomes such as productivity, exercise tolerance, and patient satisfaction were considered important by the patient groups that provided stakeholder input for this review. Among them, exercise tolerance and patient satisfaction were included within HRQoL, as measured with the SGRQ, but were not compared separately. Productivity was not considered by any of the trials. Since the follow-up duration of ETHOS was 52 weeks, the efficacy and safety of treatments that are administered longer than that needs to be assessed, as BGF MDI 320 is indicated for long-term maintenance therapy. Lastly, the expert also noted that the percentage of patients who discontinued the study drug (22.4% in ETHOS and 14% in KRONOS) is higher than the rate of discontinuation from an inhaler device in the Canadian COPD population.

Indirect Evidence

Objectives and Methods for the Summary of Indirect Evidence

There was limited evidence from head-to-head RCTs that compared BGF MDI 320 with other fixed-combination triple therapies in patients with COPD. The purpose of this section is to summarize and appraise the literature for the comparative efficacy and safety of BGF MDI 320 and other triple therapies for COPD through indirect comparisons or NMAs.

An NMA submitted by the sponsor was reviewed. Additionally, a focused literature search for indirect comparisons with COPD was run in MEDLINE All (1946–) on February 26, 2021. Through the search, 121 citations were identified, among which 16 potentially relevant articles were screened. After the full-text review, 1 sponsor-submitted NMA was included in this review. Sixteen NMAs were excluded due to irrelevant interventions, either because triple-therapy LAMA-LABA-ICS combinations were considered as a class, or because the specific triple therapy of BGF was not considered.

Description of NMA

One systematic review and NMA of triple combinations in moderate-to-severe COPD submitted by the sponsor was included in this review for critical appraisal. The included NMA was sponsored by the sponsor of BGF MDI (Breztri). Study selection criteria, outcomes of interest, and the methods of the NMA are described in Table 16. Of note, the NMA considered comparators such as Trimbow and Breztri 160 (BGF MDI 160), which are not available in Canada. Only the comparative results of the triple-therapy combinations relevant to the current review are summarized and presented subsequently.

Methods of the Sponsor-Submitted NMA

Objectives

The objective of the included NMA was to evaluate the comparative efficacy and safety of Breztri Aerosphere and other triple-therapy combinations in the treatment of moderate-to-very-severe COPD.

Study Selection Methods

A systematic literature search was conducted to search and identify studies for inclusion in the NMA. Briefly, a systematic literature search was conducted in multiple electronic databases (MEDLINE, MEDLINE In-Process, Cochrane Central Register of Controlled Trials [CENTRAL], conference proceedings, and other sources to identify relevant studies. The searches were performed in July 2019 and updated in June 2020.

The eligible population comprised patients 40 years or older with moderate-to-severe COPD. Treatment using triple therapy (LAMA-LABA-ICS) in fixed-dose and open combinations was considered a relevant intervention. The relevant comparators of interest were triple, dual, or monotherapies of ICS, LAMA, and LABA along with placebo and best supportive care. The efficacy outcomes evaluated in the NMA were rate of exacerbations, lung function, symptom relief, use of rescue medications, HRQoL |||||||||||| ||||||||. Safety outcomes such as AEs, SAEs, pneumonia, and upper respiratory tract infection, along with withdrawals (due to any reason and due to AEs) were also analyzed.

The study selection was conducted by 2 reviewers independently with discrepancies resolved by a third reviewer. Data extraction from the included studies was conducted by 2 reviewers using a predetermined grid and was checked and reconciled by a third reviewer. The risk of bias in the included trials was assessed using the single technology appraisal guide from the National Institute for Health and Care Excellence (NICE). Quality assessment criteria included randomization, allocation concealment, blinding, baseline comparability, follow-up, selective reporting, and analysis.

NMA Analysis Methods

The NMA was conducted using a 3-level hierarchical Bayesian NMA model, which considers both treatments and classes of medications. This model assumed exchangeability between treatments within the same class. Primary base-case analyses were performed using this class-effect model. Class effects were considered for ICS-LABA and open triple therapies, in addition to including separate model nodes by individual therapy. In addition, independent treatment-effect models, in which treatments were pooled into classes defined by open triple therapies, ICS-LABAs, and LABA-LAMAs, and no differentiation was made between treatments within each class, were conducted for key outcomes relevant to economic modelling (exacerbations, use of rescue medication, |||||||| AEs, and treatment withdrawals). Multiple pairwise meta-analyses were conducted along with the NMA, following the recommended practice of the NICE decision support unit for evidence synthesis. The synthesis was conducted using WinBUGS. A total of 20,000 simulations were run for the burn-in and 180,000 simulations for inferences. Convergence was assumed if the Gelman-Rubin statistic was equal to 1 and was assessed by running 3 chains. Deviance information criterion (DIC) was used to determine the best fit, in which the model with the lowest DIC was considered to have the best fit. Markov chain Monte Carlo simulation for inference was increased to 200,000 and a thin of 30 was applied when autocorrelation was present. Models were run using non-informative priors.

A list of outcomes, the types of distribution, and output statistics is presented in Table 17. The primary outcome was exacerbations (moderate to severe and severe only). The comparative effects sizes of the NMA were reported using rate ratio (for log-normal distributions), mean difference (for continuous outcomes), odds ratio (for dichotomous outcomes), and HR (for time to event distributions) with 95% credible intervals (CrIs).

Due to the lack of a common comparator between the treatments, all LAMA-LABA combinations were grouped together to create an interlinked network under the assumption of similar efficacy for all LAMA-LABA therapies. The outcomes were analyzed at 24 and 52 weeks. All studies that reported outcome data between 20 and 28 weeks were included in the 24-weeks analysis. All studies that reported data between 48 and 56 weeks were included in the 52-weeks analysis.

Potential inconsistencies in the network were investigated with a feasibility assessment before conducting the NMA. A list of potential effect modifiers to assess clinical heterogeneity was not reported. Statistical heterogeneity in the NMA was assessed using the I² statistic.

The base case for efficacy and safety outcomes included double-blinded studies. Some sensitivity analyses were conducted by removing outlier studies from the base-case model. For the efficacy outcomes, the planned sensitivity analyses included studies with a symptomatic population, exacerbation history, and trial duration. Other planned sensitivity analyses included analyses of open-label studies along with double-blind RCTs. Lastly, a univariate meta regression (using BMI, smoking status, COPD severity, and exacerbation history) was conducted for the rate of exacerbations (moderate and severe exacerbations as well as severe exacerbations). For safety outcomes, the planned sensitivity analyses that were performed included studies with a duration of 24 weeks and those with confirmed pneumonia cases (for the pneumonia outcome).

Results of the NMA

The systematic review identified 15,542 publications from the original database search. After excluding irrelevant and duplicate studies, the full text of 1,589 publications was screened along with 33 publications from other sources (conferences, bibliographies, Clinical Study Reports). Nineteen trials (from 151 publications) were found to be eligible and included in the NMA. The search update did not find any additional eligible trials. Of the included studies, 7 studies were 52 weeks in duration, 5 studies were 24 weeks in duration, 1 study was 26 weeks in duration, and 6 studies were 12 weeks in duration. The FULFIL study lasted 24 weeks; however, a subgroup of patients continued in an extension study up to 52 weeks. Two of the included studies were open label (Jung, 2012 and Lee, 2016), and 17 were double-blinded RCTs.

The characteristics of the included studies, such as the inclusion criteria, number of randomized patients, baseline demographics, and clinical characteristics of patients, were not reported. The quality assessment conducted using the NICE checklist showed that all of the included studies had a low risk of bias across domains such as randomization, baseline characteristics, withdrawals, and statistical analysis. Most of the included studies (except the 2 open-label trials) had a low risk of bias in terms of blinding. All studies except 1 (Welte, 2009) had a low risk of bias in outcome selection and reporting.

The feasibility assessment conducted before the NMA showed that 2 studies (Wheeler, 2016 and Siler, 2016) were not connected in the base-case network and were therefore excluded from the NMA. Open-label studies were not included in the base-case analysis. Thus, 15 RCTs were included in the NMA.

Results

Table 18 and Table 19 present the summary of the results of the sponsor-submitted NMA for efficacy outcomes. The results of the safety outcomes are presented in Table 20.

Rate of Exacerbations

For the outcome of moderate-to-severe exacerbations without restriction on time point, 14 studies provided data for the NMA. In a random-effects model, BGF MDI showed comparable efficacy against FF-UMEC-VI (Trelegy) and other open triple-therapy combinations (Table 18). The rate ratio of BGF MDI 320 versus Trelegy was 1.0 (95% CrI, 0.9 to 1.1). The network diagram is shown in Figure 9 and a summary plot is shown in Figure 10.

A sensitivity analysis was conducted excusing studies that did not require a prior exacerbation history. Results of the sensitivity analysis and meta regression were consistent with the base-case model.

For the outcome of severe exacerbations without restriction on time point, 13 studies provided data for the NMA. BGF MDI showed comparable efficacy against FF-UMEC-VI (Trelegy) and other open triple-therapy combinations. The rate ratio of BGF MDI 320 versus FF-UMEC-VI (Trelegy) was 1.03 (95% CrI, 0.84 to 1.49). The network diagram is shown in Figure 11 and a summary plot is shown in Figure 12. A sensitivity analysis was conducted excusing studies that did not require a prior exacerbation history. Results of the sensitivity analysis and meta regression were consistent with the base-case model.

Figure 9: Network Diagram for a Composite of Moderate-to-Severe Exacerbations (n = 14; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 10: Summary Plot of Moderate-to-Severe Exacerbations for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CrI = credible interval; FLU = fluticasone; FOR = formoterol; MDI = metered-dose inhaler; O = open triple combination; od = once daily; SAL = salmeterol; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

Figure 11: Network Diagram for a Composite of Severe Exacerbations (n = 13; Double-Blind Studies)

bid = twice daily; BDP = beclomethasone; BUD = budesonide; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 12: Summary Plot of Severe Exacerbation for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CrI = credible interval; FLU = fluticasone; FOR = formoterol; O = open triple combination; od = once daily; MDI = metered-dose inhaler; red = reducing dose of tiotropium; SAL = salmeterol; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

Health-Related Quality of Life

HRQoL measured using the change from baseline in SGRQ || |||||||||||||||||||| were considered outcomes of interest in the NMA.

For the NMA of change from baseline in SGRQ score at 24 weeks, 9 studies provided data to the network. Compared with FF-UMEC-VI (Trelegy) and other open triple-therapy combinations, BGF MDI showed comparable efficacy in improving HRQoL as measured using the SGRQ (Table 19). The network diagram is shown in Figure 13 and a summary plot is shown in Figure 14. For the NMA of change from baseline in SGRQ score at 52 weeks, 8 studies provided data to the network. Compared with FF-UMEC-VI (Trelegy), there was no difference in change from baseline in SGRQ score for BGF MDI at 52 weeks (mean difference = 0.0; 95% CrI, −0.93 to 0.93); the results were similar for triple-combination tiotropium 18 mcg once daily plus beclomethasone 200 mcg twice daily plus formoterol 12 mcg twice daily (Table 18). The network diagram is shown in Figure 15 and a summary plot is shown in Figure 16. A heterogeneity assessment was not performed, as the number of included studies was low. For analysis at both time points, the results of the planned sensitivity analyses were consistent with the base-case analysis.

Figure 13: Network Diagram for SGRQ CFB at 24 Weeks (n = 9; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; CFB = change from baseline; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; SGRQ = St George’s Respiratory Questionnaire; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 14: Summary Plot of SGRQ CFB at 24 Weeks for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CFB = change from baseline; CrI = credible interval; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; MDI = metered-dose inhaler; O = open triple combination; od = once daily; SAL = salmeterol; SGRQ = St George’s Respiratory Questionnaire; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

Figure 15: Network Diagram for SGRQ CFB at 52 Weeks (n = 8; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; CFB = change from baseline; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; SGRQ = St. George’s Respiratory Questionnaire; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 16: Summary Plot of SGRQ CFB at 52 Weeks for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CFB = change from baseline; CrI = credible interval; FLU = fluticasone; FOR = formoterol; MDI = metered-dose inhaler; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; SGRQ = St. George’s Respiratory Questionnaire; TIO = tiotropium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

The data for the NMA of SGRQ responders at 24 and 52 weeks were contributed by 8 and 6 studies, respectively (network diagram not shown). The results of the analysis at 24 weeks showed that BGF MDI was associated with an improvement in the responder rate of SGRQ comparable to FF-UMEC-VI (Trelegy) and open triple-therapy combinations (Table 19). The results of the planned sensitivity analyses were consistent with the base-case analysis. At 52 weeks, compared with FF-UMEC-VI (Trelegy) and the open combination of tiotropium 18 mcg once daily plus beclomethasone 200 mcg twice daily plus formoterol 12 mcg twice daily, BGF MDI 320 showed comparable improvement in the SGRQ responder rate (Table 18).

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Lung Function

Measures of lung function such as change from baseline in the trough FEV₁ (at 24 and 52 weeks) and |||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.

The data for the NMA of change from baseline in trough FEV₁ at 24 and 52 weeks were contributed by 10 and 8 studies, respectively. At 24 weeks, BGF MDI showed improvements in trough FEV₁ comparable to FF-UMEC-VI (Trelegy) and other open triple-therapy combinations (Figure 19). The network diagram for analysis at 24 weeks is shown in Figure 17 and a summary plot is shown in Figure 18. Overall, moderate-to-low statistical heterogeneity (I² = 0% to 45%) was observed. The results of the planned sensitivity analyses were consistent with the base-case analysis.

Figure 17: Network Diagram for Trough FEV₁ CFB at 24 Weeks (n = 8; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; CFB = change from baseline; F = fixed-dose combination; FEV₁ = forced expiratory volume in 1 second; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; SAL = salmeterol; SGRQ = St. George’s Respiratory Questionnaire; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 18: Summary Plot of Trough FEV₁ CFB at 24 Weeks for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CFB = change from baseline; CrI = credible interval; F = fixed-dose combination; FEV₁ = forced expiratory volume in 1 second; FLU = fluticasone; FOR = formoterol; MDI = metered-dose inhaler; O = open triple combination; od = once daily; SAL = salmeterol; SGRQ = St. George’s Respiratory Questionnaire; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

At 52 weeks, BGF MDI showed improvements in trough FEV₁ comparable to FF-UMEC-VI (Trelegy) and other open triple-therapy combinations (Table 18). The network diagram for analysis at 52 weeks is shown in Figure 19 and a summary plot is shown in Figure 20. A heterogeneity assessment was not performed, as the number of included studies was limited. The results of the planned sensitivity analyses were consistent with the base-case analysis (Figure 18).

Figure 19: Network Diagram for Trough FEV₁ CFB at 52 Weeks (n = 8; Double-Blind Studies)

bid = twice daily; CFB = change from baseline; F = fixed-dose combination; FEV₁ = forced expiratory volume in 1 second; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 20: Summary Plot of Trough FEV₁ CFB at 52 Weeks for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CFB = change from baseline; CrI = credible interval; F = fixed-dose combination; FEV₁ = forced expiratory volume in 1 second; FLU = fluticasone; FOR = formoterol; MDI = metered-dose inhaler; O = open triple combination; od = once daily; red = reducing dose of tiotropium; SAL = salmeterol; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Symptoms

The burden of symptoms measured using TDI focal score at 24 and 52 weeks was considered an outcome of interest in the NMA. The data for the NMA of TDI focal score at 24 and 52 weeks were contributed by 6 and 4 studies, respectively.

At 24 weeks, BGF MDI showed improvements in TDI focal score comparable to FF-UMEC-VI (Trelegy) and other open triple-therapy combinations (Table 18). The network diagram is shown in Figure 21 and Figure 19, and a summary plot is shown in Figure 22. The results of the planned sensitivity analyses were consistent with the base-case analysis.

Figure 21: Network Diagram for TDI Focal Score at 24 Weeks (n = 6; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; SAL = salmeterol; TDI = Transition Dyspnea Index; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 22: Summary Plot of TDI Focal Score at 24 Weeks for BGF MDI 320 Versus Comparators

BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CrI = credible interval; FLU = fluticasone; MDI = metered-dose inhaler; O = open triple combination; od = once daily; SAL = salmeterol; TDI = Transition Dyspnea Index; TIO = tiotropium; UMEC = umeclidinium; VIL = vilanterol trifenatate; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

At 52 weeks, the TDI focal score for BGF MDI was comparable to FF-UMEC-VI (Trelegy) and tiotropium 18 mcg once daily plus fluticasone 500 mcg twice daily plus salmeterol 50 mcg twice daily (Table 18). The network diagram is shown in Figure 23 and Figure 19, and a summary plot is shown in Figure 24. As the network was sparse and the DIC did not favour a single model, a random-effects model with informative priors was chosen as the base case. A heterogeneity assessment was not performed, as the number of included studies was limited. The results of the planned sensitivity analyses were consistent with the base-case analysis.

Figure 23: Network Diagram for TDI Focal Score at 52 Weeks (n = 4; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; CFB = change from baseline; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; SAL = salmeterol; TDI = Transition Dyspnea Index; TIO = tiotropium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 24: Summary Plot of TDI Focal Score at 52 Weeks for BGF MDI 320 Versus Comparators

BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CrI = credible interval; FLU = fluticasone; MDI = metered-dose inhaler; O = open triple combination; od = once daily; SAL = salmeterol; TDI = Transition Dyspnea Index; TIO = tiotropium; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

Use of Rescue Medication

The mean puffs per day of rescue medication over 52 weeks was considered for an NMA, with 6 studies providing data. The results showed that BGF MDI was associated with a reduction in the use of rescue medication, which was comparable to FF-UMEC-VI (Trelegy) (mean difference = −0.05; 95% CrI, −0.34 to 0.18) and tiotropium 18 mcg once daily plus beclomethasone 200 mcg twice daily plus formoterol 12 mcg twice daily (mean difference = −0.09; 95% CrI, −0.48 to 0.16) (Table 18). The network diagram is shown in Figure 25 and Figure 19 and a summary plot is shown in Figure 26. The results of the planned sensitivity analyses were consistent with the base-case analysis.

Figure 25: Network Diagram for Use of Rescue Medication Over 52 Weeks (n = 6; Double-Blind Studies)

BDP = beclomethasone; bid = twice daily; F = fixed-dose combination; FLU = fluticasone; FOR = formoterol; LABA = long-acting beta2-agonist; LAMA = long-acting muscarinic antagonist; O = open triple combination; od = once daily; TIO = tiotropium; VIL = vilanterol trifenatate.

Source: Sponsor-submitted network meta-analysis.

Figure 26: Summary Plot of Use of Rescue Medication Over 52 Weeks for BGF MDI 320 Versus Comparators

BDP = beclomethasone; BGF = budesonide-glycopyrronium-formoterol fumarate; bid = twice daily; CrI = credible interval; FOR = formoterol; MDI = metered-dose inhaler; O = open triple combination; od = once daily; TIO = tiotropium; vs. = versus .

Source: Sponsor-submitted network meta-analysis.

||||||||||||||||||||

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||

Figure 27: ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Figure 27 was removed at the request of the sponsor because it contained confidential information.

||||||||||||||||||||||||||||||

Figure 28: ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Figure 28 was removed at the request of the sponsor because it contained confidential information.

||||||||||||||||||||||||||||||

Safety

The safety outcomes considered in the NMA were any AEs, any SAEs, upper respiratory tract infection, and pneumonia. The time point for analysis for all safety outcomes was 52 weeks. The overall AEs and SAEs were comparable between BGF MDI 320 and FF-UMEC-VI (Trelegy) and other triple-therapy combinations. For upper respiratory tract infections, Trelegy was the relevant comparator. Compared with Trelegy, BGF MDI showed no difference (odd ratio = 1.0; 95% CrI, 0.73 to 1.3) (Table 20). There were no differences in the occurrence of pneumonia (any grade) in BGF MDI 320 compared with FF-UMEC-VI (Trelegy) or other triple-therapy combinations. The results of the NMA of safety outcomes are presented in Table 20. The results of the planned sensitivity analyses were consistent with the base-case analysis.

Outcomes in the NMA related to tolerability were assessed as all withdrawals and withdrawals due to AEs at 52 weeks. Eight studies contributed to the NMA for each of these outcomes. The results showed that the tolerability of BGF MDI 320 is comparable to FF-UMEC-VI (Trelegy) and other triple-therapy combinations (Table 20). The results of the planned sensitivity analyses were consistent with the base-case analysis.

Critical Appraisal of NMA

The sponsor-submitted NMA was critically appraised, in part, by using recommendations from the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Task Force on Indirect Treatment Comparisons⁴⁷ as a guide.

Relevance

The objectives and rationale for the NMA were clearly reported. The population, interventions, and comparators were appropriate and clear. Relevant to the current report, BGF MDI (triple therapy) was compared with other triple therapies (open and fixed dose) as well as dual and monotherapies of ICS, LAMA, and LABA. Comparisons with placebo and best supportive care were also considered. The outcomes considered in the NMA were relevant and described well. Across the included studies, the definitions of the exacerbation outcomes were similar to those used in Canadian settings, thereby increasing the generalizability of the results.

Credibility

The systematic review was based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement, and the inclusion and exclusion criteria allowed for the identification of studies with relevant outcomes. Multiple electronic databases were searched to identify eligible RCTs. In addition, conference abstracts, trial registries, and bibliographies of systematic reviews were searched. Study selection and data extraction were conducted by 2 independent reviewers with a third reviewer employed for the resolution of discrepancies and data checking. However, the potential for publication bias was not reported and the literature search used was limited to those with an English abstract.

Quality assessments of the included RCTs were conducted using the NICE Single Technology Appraisal process. The risk of bias of the individual studies was shown. Overall, the studies were of low risk, with 2 studies having issues related to blinding (2 with unclear risk), and 1 study having a high risk associated with outcome selection (2 with unclear risk). All outcomes of interest for the NMA were not considered by all of the included studies and, therefore, not included in some of the NMAs. Additionally, a list of excluded studies (along with reason for exclusion) was not reported. As publication bias was also not assessed, it is likely the selective reporting of outcomes in the study introduced bias; however, not enough information was available to know for certain.

The baseline characteristics of the included studies were not described, except for the outcomes and follow-up time. Therefore, the number of patients in each of the included studies and their demographic and clinical characteristics were not clear. This made interpretation across trials regarding potential effect modifiers challenging. It was unclear whether all potential effect modifiers were considered, as such a list was not reported. A feasibility analysis was conducted by the reviewers for BGF MDI versus other triple-therapy combinations before conducting the NMA. Since a common comparator was lacking, all LAMA and LABA combinations were grouped together as 1 class to develop networks under the assumption of similar efficacy.

Analysis

The NMA was conducted using a 3-level hierarchical Bayesian NMA model. Convergence was assessed by running 3 chains and was assumed if the Gelman-Rubin statistic was 1. Both fixed-effect and random-effect models were generated. The goodness of fit was assessed using the DIC, and the best model was chosen based on the lowest DIC value. In the presence of autocorrelation, the Markov chain Monte Carlo simulation for inference was increased to 200,000 and a thin of 30 was applied. A class effect and independent treatment model were conducted, the base-case analysis being a class-effect model. The NMA was conducted based on relative treatment effects and no naive comparisons were made. Multiple pairwise comparisons were performed to allow indirect and direct comparisons to be synthesized. Heterogeneity in random-effect models was assessed using the I² statistic. Consistency between direct and indirect comparisons was assessed. For 2 outcomes (moderate-to-severe exacerbation and severe exacerbation), a pre-specified univariate meta regression was conducted using BMI, smoking status, COPD severity, and exacerbation history. However, for most outcomes, meta regression was not possible due to the small number of studies reporting those outcomes.

Three different sensitivity analyses were conducted for the efficacy outcomes, as described in the earlier section (adding open-label studies to the base, symptomatic studies only, studies with a follow-up time greater than 24 weeks, and studies with a requirement of exacerbation history). Other potential effect modifiers, such as disease severity, baseline eosinophilia, and baseline reversibility of bronchodilators, were not considered.

Reporting Quality

Network diagrams and summary plots for all outcomes considered in the NMA were reported. Study results of the individual studies were not provided in the report. The results of the direct comparison and indirect comparison within the NMA were not reported separately. However, due to the lack of head-to-head trials between triple-therapy combinations, all results relevant to this review were from indirect comparisons. The results of the NMA were reported using appropriate effect estimates and measures of random variability (95% CrI). Relative treatment effects, based on certain patient characteristics such as BMI, smoking history, COPD severity, and exacerbation history were reported, as univariate meta regression, for the exacerbation outcomes. However, other potential effect modifiers, such as baseline eosinophilia and baseline reversibility to bronchodilators, were not considered.

The included NMA was funded by the sponsor of BGF MDI 320. Potential bias due to conflict of interest, and the steps taken to mitigate such bias, were not addressed.

Lastly, the authors of the sponsor-submitted NMA concluded that the efficacy and safety of BGF MDI 320 is comparable to other triple therapies for COPD. Based on the effect sizes and uncertainty levels (CrIs) of all outcomes, the relative efficacy and safety of BGF MDI 320 were found to be comparable between the treatments.

Other Relevant Evidence

Objective

The objective was to summarize the efficacy and safety results of the KRONOS extension study.³⁸ This study was identified in the systematic review search but was not included in the main report, as the outcomes were not identified in the protocol.

Findings

Study Design

This 52-week, double-blind, randomized, parallel-group study was an extension of the phase III, 24-week, KRONOS study. The objective of the study was to assess the BMD and ocular safety of BGF MDI compared with BFF MDI and GFF MDI in patients with moderate-to-very-severe COPD. The primary end points were percentage change from baseline in lumbar spine BMD and change from baseline in the Lens Opacities Classification System version III (LOCS III) posterior subcapsular cataract score at week 52. AEs were also assessed in this study. A summary of study characteristics is presented in Table 21.