Normalisation of airflow limitation in asthma: Post‐hoc analyses of TRIMARAN and TRIGGER

Abstract Background In asthma, persistent airflow limitation (PAL) is associated with poorer control, lung function decline and exacerbations. Using post‐hoc analyses we evaluated: the relationship between post‐salbutamol PAL at screening, airflow limitation (AL) during 52 weeks treatment with extrafine beclometasone dipropionate/formoterol fumarate/glycopyrronium (BDP/FF/G) versus BDP/FF and the risk of moderate/severe asthma exacerbations. Methods TRIMARAN and TRIGGER were double‐blind studies comparing BDP/FF/G with BDP/FF (TRIMARAN medium‐dose ICS; TRIGGER high‐dose) in adults with uncontrolled asthma. Patients were subgrouped according to post‐salbutamol PAL status at screening, and AL over the 52‐week treatment period. Results Most patients with post‐salbutamol PAL at screening had AL at all on‐treatment visits (TRIMARAN 62.8%; TRIGGER 66.8%). A significantly higher proportion of patients had normalised airflow on ≥1 follow‐up visit when receiving BDP/FF/G than BDP/FF (TRIMARAN 44.1 vs. 33.1% [p = 0.003]; TRIGGER 40.1 vs. 26.0% [p < 0.001]). In patients with post‐salbutamol PAL at screening and normalised AL at ≥1 follow‐up visit, exacerbation rates were 15% (p = 0.105) and 19% (p = 0.039) lower in TRIMARAN and TRIGGER versus those with AL on all visits. There was a trend to lower exacerbation rates in patients receiving BDP/FF/G than BDP/FF, particularly in patients in whom AL was normalised. Conclusion In these analyses, AL in asthma was associated with an increased exacerbation incidence. Inhaled triple therapy with extrafine BDP/FF/G was more likely to normalise airflow, and was associated with a trend to a lower exacerbation rate than BDP/FF, particularly in the subgroup of patients in whom treatment was associated with airflow normalisation. ClinicalTrials.gov: TRIMARAN, NCT02676076; TRIGGER, NCT02676089.


| INTRODUCTION
One of the aims of asthma management is to minimise the development of persistent airflow limitation (PAL), 1 usually defined as a forced expiratory volume in 1 s (FEV 1 ) to forced vital capacity (FVC) ratio that does not normalise after administration of a short-acting β 2 -agonist. Although potentially a distinct phenotype, 2 the risk factors associated with PAL are unclear. Indeed, one study has shown that a higher exacerbation rate could subsequently lead to the development of PAL, 3 whereas a second study suggested that the risk factors included age, asthma duration and male sex, but not exacerbation history. 4 Importantly, PAL could be a marker for corticosteroid resistance. 5 The presence of PAL in patients with asthma is associated with increased bronchial inflammation and airway remodelling, [5][6][7] poorer subsequent asthma control, 8 and increased lung function decline. 9 Furthermore, patients with asthma who have PAL are more likely to also have small airways dysfunction. 10 However, whether PAL in asthma is a permanent condition, as it is in chronic obstructive pulmonary disease, or a potentially reversible state has never been properly tested. The ordered setting of a randomised controlled trial seems appropriate for such an assessment.
TRIMARAN and TRIGGER were two 52-week clinical studies that recruited patients with asthma that was uncontrolled despite treatment with a fixed combination of a long-acting β 2 -agonist (LABA) and medium-(TRIMARAN) or high-dose inhaled corticosteroid (ICS; TRIGGER), and pre-bronchodilator FEV 1 <80% of predicted normal, but no limitation on post-bronchodilator FEV 1 or FEV 1 /FVC ratio. 11 As each study recruited more than 1000 patients, post-hoc analyses provide an opportunity to evaluate the relationship between post-salbutamol PAL at screening and the occurrence of asthma exacerbations over the subsequent 52 weeks. Furthermore, patients in both studies were randomised to either triple therapy with extrafine beclometasone dipropionate/formoterol fumarate/glycopyrronium (BDP/FF/G) or to the extrafine ICS/LABA combination BDP/FF. Overall in these studies, the addition of the long-acting muscarinic antagonist (LAMA) component improved lung function and reduced the risk of asthma exacerbations. 11 In previous post-hoc analyses, we demonstrated that the efficacy of BDP/FF/G was more pronounced in the subgroup of patients with post-salbutamol PAL at screening compared to the overall population in both studies, consistently providing statistically superior bronchodilator efficacy to BDP/FF. 12 In addition, the effect of BDP/FF/G on moderate/severe exacerbations appeared greater in the PAL subset in TRIGGER. We therefore used data from TRIMARAN and TRIGGER to better understand the relationship between normalisation of airflow limitation (following maintenance therapy with triple therapy or ICS/LABA) and the risk of asthma exacerbations in the subgroup of patients who exhibited post-salbutamol PAL at screening.

| Trial design and participants
The full design and inclusion/exclusion criteria of TRIMARAN

| Outcomes
These post-hoc analyses focus on the rates of moderate-to-severe exacerbations in each study. Severe exacerbations were defined as asthma worsening requiring treatment with systemic corticosteroids for at least 3 days, whereas moderate exacerbations were episodes of asthma worsening that were self-managed, defined in accordance with an American Thoracic Society/European Respiratory Society joint statement (one or more of: nocturnal awakenings due to asthma or increased daily symptoms; increased short-acting β 2 -agonist use; decline in FEV 1 or peak expiratory flow; and/or emergency room/ study site visit for asthma treatment not requiring systemic corticosteroids). 11,13

| Statistical methods
The analyses include patients with post-salbutamol (400 µg) FEV 1 / FVC ratio data available at the screening visit (corresponding to the peak bronchodilator effect) and with FEV 1 /FVC ratio data available  Figure S1).
The baseline characteristics of the patients subgrouped by postsalbutamol PAL status at screening using the fixed-ratio within each study are shown in Table 1. Compared to those without PAL, patients with post-salbutamol PAL on entry to the studies were more likely to be male, ≥65 years of age, and prior smokers (current smokers were excluded from the studies). Importantly, the exacerbation history prior to enrolment in each study was similar in the PAL and no PAL subgroups (although, overall, patients in TRIGGER were more likely to have a history of >1 exacerbation than those in TRIMARAN).
Results were similar using the LLN-derived PAL status, with the exception of the age category: patients with post-salbutamol PAL on entry were more likely to be <65 years of age (Table S1).

| Relationship between post-salbutamol PAL at screening and on-treatment airflow limitation
The majority of patients with post-salbutamol PAL at screening also  Figure 1) and the LLN cut-off (57.2% and 61.8%; Figure S1). Patients who had post-salbutamol PAL at screening that was then normalised by study treatment (i.e., 3 h post-dose FEV 1 / FVC ratio ≥0.7 on at least one study visit) were more likely to be female and <65 years of age in both studies and using both cut-offs (i.e., 0.7 or LLN; Table 2 and Table S2). In these patients, the number of visits at which airflow was normalised (i.e., 3 h post-dose FEV 1 / FVC ≥0.7) ranged from one to six, with 20.6% of patients in TRIMARAN and 16.6% in TRIGGER having normalisation at all six visits using the fixed-ratio cut-off (Figure 1). Using the LLN cut-off, 22.8% and 21.2% of patients had airflow normalisation at all six visits ( Figure S1). PAPI ET AL.

| Influence of study treatment on normalisation of airflow limitation
There was no imbalance between treatment arms in the proportion of patients with post-salbutamol PAL at screening using either       Although such airflow limitation could be targeted with biologic therapy or systemic corticosteroids, [16][17][18][19] given concerns over cost or side effects, respectively, our results also suggest that a trial with inhaled triple therapy first would be appropriate. That a proportion of patients still had airflow limitation despite inhaled triple therapy could potentially be the result of lingering inflammation, or irreversible airway narrowing due to airway wall remodelling. 20 Evidence suggests that such remodelling may result from repeated bronchoconstriction, rather from inflammation. 21 An appropriate treatment strategy for a patient with PAL despite ICS/LABA could therefore be to escalate promptly to triple therapy, although additional research is required to determine whether this prevents the development of irreversible airflow limitation.

| Relationship between study treatment and the occurrence of moderate/severe asthma exacerbations
Post-hoc, subgroup analyses such as those we present here are not formally statistically powered (due to the smaller number of patients analysed compared to the overall TRIMARAN and TRIGGER populations), and so are also prone to type II errors (or false negatives). This is especially possible when the endpoint of interest has a low incidence (as can be the case for exacerbations). Although we present the rate ratios and associated p values in many of these analyses, we would argue that a lack of statistical significance does not necessarily indicate a lack of relevant difference. In addition, given post-hoc analyses are typically not adjusted for multiplicity, findings in such analyses that have p values lower than the usual threshold (i.e., <0.05) could theoretically also be linked to type I error (or false positives). We have therefore focused on the consistency of findings between TRIMARAN and TRIGGER when drawing conclusions (one of the reasons that we did not pool the data). In a second publication, the authors used data from one of these studies to evaluate the relationship between on-treatment airflow limitation and treatment effect, 22 although since the study was relatively small (386 patients) and only 12 weeks in duration, the effect on exacerbations was not analysed. Our analyses build on these various prior analyses, given we were able to use data from a full 1-year follow-up of a large number of patients who demonstrated reversibility at screening, who attended regular visits and were highly adherent to therapy, and with pre-and post-treatment lung function and exacerbation occurrence collected throughout the follow-up period.
In the initial analyses, we defined airflow limitation on the basis of a fixed FEV 1 /FVC ratio of 0.7. This can introduce an age bias, with older individuals more likely to meet the criterion. 23 The use of a LLN can avoid such a limitation, although obviously adds complexity to the practical implementation of the findings. We therefore replicated the analyses using an LLN-derived airflow limitation definition (i.e., <LLN). Although this impacted some of the data (most notably the proportion of patients aged ≥65 years who met the screening PAL definition, which was reduced from 19.7% to 22.8% in TRIMARAN and TRIGGER, respectively, using the fixed-ratio cut-off to 16.3% and 19.7%, respectively, using the LLN cut-off), the overall results were consistent with the fixed-ratio definition results. This is consistent with one of the prior studies that used both definitions to identify airflow obstruction at baseline, in which the treatment outcomes were generally similar for the two definitions. 8 We therefore chose to focus the manuscript on the fixed-ratio data, since at an individual patient level (when the focus is on changes over time), these results are easier to interpret than is the case for more intangible (and Factors such as differences in time of the day, time to peak or plateau bronchodilation, and regression to the mean (since recruited patients had pre-bronchodilator FEV 1 <80% predicted) should also be considered. However, this scenario mimics treatment administration in the real world making this analysis even more relevant.
In conclusion, in patients with asthma who were receiving medium-or high-dose ICS/LABA or ICS/LABA/LAMA, airflow limitation was associated with a general trend to an increased incidence of moderate-to-severe exacerbations. Treatment with extrafine BDP/ FF/G was more likely to normalise airflow limitation and tended to be associated with a lower exacerbation rate than treatment with BDP/ FF, with the most relevant results in the subgroup of patients in whom treatment was associated with normalisation of airflow limitation.