Management of severe asthma: a European Respiratory Society/American Thoracic Society guideline

Summary of the evidence

Benefits

Anti-IL-5 and anti-IL-5Rα therapies reduce exacerbations and hospitalisations in patients with severe eosinophilic asthma. Mepolizumab and benralizumab are effective in reducing maintenance OCS dose in patients with corticosteroid-dependent severe asthma.

Harms

All three anti-IL-5 strategy drugs were well tolerated. The frequency of adverse effects was similar when compared with placebo.

Conclusions

An anti-IL-5 strategy reduces exacerbations in patients with severe eosinophilic asthma. Mepolizumab and benralizumab are effective in reducing OCS dose in corticosteroid-dependent asthma. The effects on asthma control, quality of life and FEV₁ are modest for all drugs and did not meet the MCID threshold.

Research needs and additional considerations

Direct comparisons will be needed to further guide selection of the appropriate anti-IL-5 drug. Uncertainty exists around the best biomarker and blood eosinophil threshold that would predict response to anti-IL-5 therapy. In addition to blood eosinophils, the efficacy of anti-IL-5 therapy depends on the degree of pre-existing asthma exacerbations. This should be taken into consideration when considering the clinical and cost-effectiveness of this form of therapy. Data from adolescents are unavailable for mepolizumab and reslizumab, whereas for benralizumab there are data on a limited number of adolescents with severe asthma. There are no data on younger children (age <5 years). Therefore, more evidence is needed to provide higher quality recommendations in the paediatric age group.

What others are saying

The 2019 Global Initiative for Asthma (GINA) guidelines [20] and the National Institute for Health and Care Excellence (NICE) technology appraisal guidance [21–23] include mepolizumab, reslizumab and benralizumab as add-on therapeutic options for severe eosinophilic asthma (at GINA step 5).

ERS/ATS recommendation

We suggest an anti-IL-5 strategy as add-on therapy for adult patients with severe uncontrolled asthma with an eosinophilic phenotype and for those with severe corticosteroid-dependent asthma (the Task Force gave this a conditional recommendation because inclusion criteria across studies did not consistently align with the ERS/ATS severe asthma definition).

Summary of the evidence

We identified 12 randomised controlled trials of anti-IL-5 therapies in children and adults 12–75 years of age that evaluated differential response to therapy among subgroups of individuals with higher or lower levels of eosinophils in blood or sputum in post hoc analyses [10–17, 19, 25, 26]. One study was a meta-analysis of two randomised controlled trials of mepolizumab's therapeutic responsiveness combining the 100 mg s.c. and 75 mg intravenous doses for the analysis by blood eosinophil level [26]. Notably, four of the studies recruited only subjects with evidence of eosinophilic asthma, defined as sputum eosinophils ≥3% or blood eosinophil level ≥300 µL⁻¹ [11–13, 25]. Six of the studies included children age ≥12 years [10, 11, 13, 15, 17, 26]. The most commonly measured biomarker was blood eosinophil count. Only one study evaluated sputum eosinophil level [12]. One additional study evaluated whether the presence of persistently elevated sputum or blood eosinophils was an indicator of therapeutic failure and justified the addition of an alternate anti-IL-5 strategy [27].

Cut-offs assessed for baseline blood eosinophil levels, and hence the definition of what constitutes eosinophilia, varied across anti-IL-5 strategies. Studies of mepolizumab specifically assessed a cut-off of blood eosinophils ≥150 µL⁻¹. For mepolizumab, there was a 73% (95% CI −82– −59%) reduction in exacerbations among those with a blood eosinophil level ≥500 µL⁻¹ compared with a 36–39% reduction in all other groups with eosinophil levels ≥150 µL⁻¹. Notably, subjects with eosinophil levels ≥150 µL⁻¹ constituted nearly three-quarters of the severe asthma population in those studies. Patients treated with reslizumab with a baseline eosinophil level ≥400 µL⁻¹ had a 54% reduction in exacerbations; higher cut-offs were not associated with a greater reduction in exacerbations. For benralizumab, a cut-off ≥300 µL⁻¹ was associated with a significant reduction in exacerbations; however, it is not clear what the optimal cut-off should be since even subjects with an eosinophil level <300 µL⁻¹ experienced a reduction in exacerbations.

For effects on asthma control and quality of life, the data again varied by anti-IL-5 strategy; among those with a baseline eosinophil level ≥150 µL⁻¹, 63% treated with mepolizumab versus 41% treated with placebo achieved a ≥0.5-point reduction from baseline in ACQ-5 (risk ratio 1.53, 95% CI 1.27–1.84). The improvement in asthma control was similar among those with higher baseline levels of eosinophils (≥300 or ≥500 µL⁻¹). For benralizumab, only subjects with a baseline eosinophil level ≥300 µL⁻¹ experienced a significant improvement in asthma control, assessed as change in ACQ-6 score from baseline (mean difference −0.28 points, 95% CI −0.41– −0.15 points), whereas those with an eosinophil level <300 µL⁻¹ did not (mean difference −0.20 points, 95% CI −0.44–0.3 points). Similarly for reslizumab, a cut-off eosinophil level ≥400 µL⁻¹ was associated with improved asthma control (mean difference in ACQ-7) from baseline −0.27 points (95% CI −0.36– −0.19 points), whereas those with an eosinophil level <400 µL⁻¹ did not have a significant benefit (−0.12 points, 95% CI −0.33–0.09 points). Sputum eosinophil level was only considered in one study of reslizumab [12] and sputum levels were categorised as ≥10% or <10%. There were no statistical differences found between groups in level of asthma control. There was a trend for higher blood eosinophil levels to be associated with a greater improvement in asthma control.

One additional study, which was not included in the meta-analysis, assessed treatment response of weight-adjusted i.v. reslizumab in patients previously treated with 100 mg s.c. mepolizumab [27]. It reported that persistently high levels of eosinophils (blood >300 µL⁻¹ and sputum >3%) after treatment with mepolizumab characterised responders. In those subjects a weight-adjusted dose of reslizumab was administered. It was found that further improvements in symptoms and reductions in eosinophilia were possible with addition of reslizumab. These data suggest that evidence of uncontrolled eosinophilic inflammation, as manifest by a high sputum or blood eosinophil level, may be useful in determining which subjects may benefit from additional anti-IL-5 strategies; however, this requires further confirmation.

Benefits

The specific cut-off blood eosinophil count to predict improved asthma control and reduction in exacerbations varies across anti-IL-5 strategies. However, there is very low quality evidence that mepolizumab may provide further benefit in reducing exacerbations in patients with baseline blood eosinophilia ≥500 µL⁻¹ compared with those with eosinophil levels <150, 150– <300 and 300– <500 µL⁻¹.

Harms

There were five studies that assessed adverse events in benralizumab or reslizumab [11, 13–17]. The data for mepolizumab did not assess differences in adverse event rates based on blood eosinophil level. There was no difference in adverse events among those with higher versus lower eosinophil counts for benralizumab. For reslizumab, only subjects with baseline eosinophilia ≥400 µL⁻¹ during screening were recruited; the fewest adverse events occurred in the group who had no data on eosinophil count at the time of recruitment compared with patients with baseline eosinophilia ≥400 µL⁻¹. There was a 5% reduction in the number of adverse events among those with an eosinophil count ≥400 µL⁻¹ which, although statistically relevant, may not be clinically meaningful. More recent studies have now shown that both benralizumab and mepolizumab maintain an adequate safety profile during long-term use for up to 2 and 4.5 years, respectively [28, 29].

Other considerations

Most of the studies focused on blood eosinophils as a biomarker, and there were limited data on sputum eosinophils and no data on F_ENO or serum periostin. Blood eosinophils can be measured in any standard laboratory, increasing its feasibility as a biomarker; however, additional testing beyond the point of care may be required to ascertain baseline levels, particularly among patients on or recently taking systemic corticosteroids. It is more acceptable than sputum eosinophil levels, which are currently only performed in specialised centres. It should be noted that there may be causes other than atopy (e.g. parasitic infections) for peripheral blood eosinophilia, especially in low- and middle-income settings.

Cut-offs to assess response varied across studies of anti-IL-5 medications and there were no data comparing therapeutic regimens using different cut-off levels. Finally, most of the anti-IL-5 strategies use a fixed-dose regimen based on randomised controlled trial data suggesting a plateau in the dose response; however, one study suggested that persistent eosinophilia, despite anti-IL-5 strategies, should be considered as an opportunity to add-on reslizumab using a weight-adjusted dose regimen [27].

Conclusions and research needs

Although the data suggest that subjects with higher levels of blood eosinophil counts benefit more from anti-IL-5 strategies, the evidence we reviewed does not show that a specific level of blood eosinophils ≥150 µL⁻¹ for mepolizumab, ≥300 µL⁻¹ for benralizumab and ≥400 µL⁻¹ for reslizumab is an absolute response threshold, as clinical benefit can still be observed in some patients below these values. Based on currently available evidence (which is very limited), sputum eosinophils may not add to the prediction of response greater than blood eosinophil level.

Determining a patient's baseline eosinophil count may require more than one measurement, as this biomarker is highly variable and significantly reduced by systemic corticosteroid and ICS. It is not known if eosinophil levels obtained during periods of asthma exacerbation are better predictors of treatment response when compared with those measured during periods of clinical stability. Future studies should focus on developing additional noninvasive biomarkers for adults and children that can be used at point of care to predict responsiveness to different anti-IL-5 strategies.

What others are saying

The 2019 GINA guidelines [20] recommend the use of an anti-IL-5 and anti-IL-5Rα strategy for patients who who remain uncontrolled despite step 4 or 5 therapy who have blood eosinophils ≥300 µL⁻¹.

ERS/ATS recommendation

We suggest that a blood eosinophil count cut-off point ≥150 µL⁻¹ can be used to guide anti-IL-5 initiation in adult patients with severe asthma and a history of prior asthma exacerbations (conditional recommendation, low quality evidence).

Summary of the evidence

We identified three randomised, double-blind placebo-controlled trials [31–33] which recruited participants age 12–75 years. Of these, two studies [32, 33] involving 1014 eligible participants formed the evidence for the Task Force recommendations. These two trials included individuals with uncontrolled asthma. In one study [33], patients had uncontrolled symptoms while taking an ICS with or without a controller. In the other study [32], only participants with severe persistent asthma were recruited, whose asthma remained uncontrolled despite ICS and a long-acting β₂-agonist (LABA).

In both trials, eligible participants were randomised 1:1 to receive omalizumab or placebo. Omalizumab dose was determined on the basis of pre-treatment serum total IgE level (IU·mL⁻¹) and body weight (kg) according to the European [33] or ATS [32] omalizumab dosing table, which ensured a minimum omalizumab dose of 0.008 mg·kg⁻¹·IgE⁻¹ (IU·mL⁻¹) every 2 weeks or a minimum of 0.016 mg·kg⁻¹·IgE⁻¹ (IU·mL⁻¹) every 4 weeks.

Busse et al. [33] pre-planned an analysis that divided participants into two subgroups according to eosinophil counts at screening: low (<300 µL⁻¹) and high (≥300 µL⁻¹). A subgroup analysis was performed by Hanania et al. [32], which divided participants into high and low subgroups as: F_ENO low<19.5 ppb, high ≥19.5 ppb; peripheral blood eosinophils low <260 µL⁻¹, high ≥260 µL⁻¹; and serum periostin levels low <50 ng·mL⁻¹, high ≥50 ng·mL⁻¹.

Pooling of the data from the two studies was not possible. In Busse et al. [33], there were significant improvements in exacerbation rates (hazard ratio 0.41, 95% CI 0.20–0.84) and a clinically trivial but statistically significantly greater change in FEV₁ % pred at 24 weeks (mean difference 7.35, 95% CI 1.38–13.31) with omalizumab compared with placebo in patients with a high eosinophil count, whereas there were no differences in patients with low eosinophils (<300 µL⁻¹). In the study by Hanania et al. [32], there was a significantly longer time to first asthma exacerbation with omalizumab compared with placebo in patients with high (≥260 µL⁻¹) eosinophil count at 48 weeks follow-up (hazard ratio 0.64, 95% CI 0.48. 0.85), whereas there were no differences in patients with low (<260 µL⁻¹) eosinophil count (hazard ratio 0.95, 95% CI 0.68–1.33). However, there were no statistically significant differences between these subgroups. There were no differences in AQLQ at 48 weeks when omalizumab was compared with placebo in patients with high (≥260 µL⁻¹) eosinophils (mean difference 0.14 points, 95% CI −0.11–0.30 points), while there was a small statistically, but not clinically significant, difference in the low (<260 µL⁻¹) eosinophil subgroup (mean difference 0.26 points, 95% CI 0.06–0.46 points).

In the subgroup analysis by F_ENO [32], there was a significant relative reduction of exacerbation rates with omalizumab compared with placebo in patients with high (≥19.5 ppb) F_ENO at 48 weeks follow-up (53%, 95% CI 37–70%), whereas there were no differences for those patients with low (<19.5 ppb) F_ENO (16%, 95% CI −32–46%). The time to first asthma exacerbation with omalizumab, compared with placebo, was significantly longer in patients with high (≥19.5 ppb) F_ENO at 48 weeks follow-up (hazard ratio 0.38, 95% CI 0.24–0.60), whereas there were no differences in patients with low (<19.5 ppb) F_ENO (hazard ratio 1.00, 95% CI 0.62–1.61). There were also larger changes of mean AQLQ with omalizumab compared with placebo in high (≥19.5 ppb) F_ENO patients at 48 weeks of follow-up (mean difference 0.39 points, 95% CI 0.06–0.72 points), whereas there were no differences in low (<19.5 ppb) F_ENO patients (mean difference 0.24 points, 95% CI −0.09–0.57 points).

There were no differences in the relative reduction of exacerbation rates at 48 weeks or FEV₁ when omalizumab was compared with placebo in periostin high (≥50 ng·mL⁻¹) or low (<50 ng·mL⁻¹) patients [32]. However, compared with placebo, omalizumab improved AQLQ in patients with low (<50 mg·mL⁻¹) periostin levels at 48 weeks follow-up (mean difference 0.50 points, 95% CI 0.22–0.78 points), whereas there were no differences patients with high (≥50 ng·mL⁻¹) serum periostin levels (mean difference 0.10 points, 95% CI −0.19–0.39 points).

Benefits

In patients treated with omalizumab compared with placebo, the presence of a baseline blood eosinophil count ≥260 µL⁻¹ is associated with greater improvements in FEV₁, and a decreased rate of exacerbations as well as longer time to first exacerbation, compared with those with a blood eosinophil count <260 µL⁻¹.

In patients treated with omalizumab compared with placebo, the presence of F_ENO ≥19.5 ppb is associated with improvements in AQLQ, reduced exacerbation rate and longer time to first exacerbation compared with those with F_ENO <19.5 ppb. In patients treated with omalizumab compared with placebo, the presence of periostin <50 ng·mL⁻¹ was associated with improvements in AQLQ compared with those with periostin ≥50 ng·mL⁻¹. Periostin levels, however, did not predict response in exacerbations or lung function. There is no evidence that periostin is a suitable biomarker to guide asthma treatment in children or adolescents. Levels are influenced by age, skeletal growth and puberty [34].

Harms

There were no differences in the adverse effects in patients treated with omalizumab versus placebo according to high or low F_ENO, blood eosinophils or periostin.

Other considerations

The estimates of effect included one single study (meta-analysis of the two randomised controlled trial was not possible), which introduced some uncertainty due to the limited number of patients included in each subgroup according to the biomarker threshold. Furthermore, the risk of bias was high for completeness of data, due to a considerable number of patients that were not evaluated at baseline for the biomarkers.

Conclusions and research needs

Blood eosinophil counts and F_ENO levels may be useful in choosing patients most likely to achieve a more positive effect on exacerbations and lung function when treated with omalizumab compared with placebo. There were no differences in adverse effects based on the high and low biomarker subgroups, suggesting that the high blood eosinophil and high F_ENO patients achieve clinical benefit without additional adverse effects, whereas low biomarker patients are at risk of adverse effects while potentially having less clinical benefit.

Other excluded studies also make important observations regarding the use of blood eosinophils to select patients most likely to respond to omalizumab. Of particular note is the study by Casale et al. [35], who reported an analysis that pooled the results of two randomised controlled trials. The studies by Busse et al. [36] and Soler et al. [37] were both phase 3, double-blind placebo-controlled trials, comprising a total of 1071 participants comparing omalizumab with placebo in participants with moderate to severe asthma. The pooled analysis published in 2018 investigated the annualised exacerbation rates in the omalizumab group versus placebo according to the subgroups of blood eosinophils: high (≥300 µL⁻¹) and low (<300 µL⁻¹) [35]. The results support the recommendations of the Task Force. There was a more pronounced reduction in exacerbations rates in the omalizumab versus placebo group for the high biomarker subgroup, i.e. for those with an eosinophil count ≥300 µL⁻¹ there was a 67% reduction in exacerbations compared with a 45% reduction in the <300 µL⁻¹ group.

In contrast to the previous studies, one publication found that omalizumab's effectiveness did not vary across biomarker levels. This retrospective study of 872 patients with severe allergic asthma showed that omalizumab reduced exacerbations by 58.4% (95% CI 52.7–63.4%) in the high biomarker (eosinophil count ≥300 µL⁻¹) group versus 58.1% (95% CI 52.7–63.4%) in the low biomarker group (eosinophil count <300 µL⁻¹) [38].

Future randomised controlled trials should evaluate baseline blood eosinophils and F_ENO as individual and combined biomarkers to further determine their ability to predict response to treatment for multiple outcomes including exacerbations, lung function as well as patient-reported outcomes such as AQLQ and asthma control. Furthermore, there is a need to identify biomarkers that support clinical decision making regarding the continuation versus discontinuation of a monoclonal anti-IgE strategy in adults and children with severe asthma.

What others are saying

The 2019 GINA guidelines [20] state that blood eosinophil level ≥260 µL⁻¹ and F_ENO ≥20 ppb are factors that may predict a good response to treatment.

Neither the 2019 British Thoracic Society/Scottish Intercollegiate Guidelines Network (BTS/SIGN) [39] nor the NICE asthma guidelines [30] make comment about predictor biomarkers for anti-IgE treatment response.

ERS/ATS recommendations

In adult and adolescent patients with severe asthma being considered for omalizumab we suggest:

• Using a blood eosinophil cut-off ≥260 µL⁻¹ to identify adolescents (>12 years) and adults with severe allergic asthma more likely to benefit from anti-IgE treatment (conditional recommendation, low quality of evidence).

• Using a F_ENO cut-off ≥19.5 ppb to identify adolescents (>12 years) and adults with severe allergic asthma more likely to benefit from anti-IgE treatment (conditional recommendation, low quality of evidence).

Summary of the evidence

We identified three randomised, placebo-controlled trials in adults 18–75 years of age (one crossover and two parallel designs), one trial in adolescents (age 12–17 years) and one trial in children (age 6–11 years) [40–42]. These trials included individuals with severe uncontrolled asthma on GINA step 4–5 or NAEPP step 5 therapies. Adults were treated with at least high-dose ICS in combination with a LABA, while adolescents and children were treated with medium-dose ICS and a LABA with a third controller.

In the adolescent and paediatric studies, eligible patients were randomised in a 1:1:1 ratio to receive tiotropium 5 µg (2 puffs of 2.5 µg) or 2.5 µg (2 puffs of 1.25 µg) or placebo (2 puffs), each delivered for 12 weeks via the Respimat Soft Mist inhaler as add-on to pre-enrolment background therapy with ICS plus one or more controller therapies. Whereas two adult studies [41] compared 5 µg tiotropium (2 puffs of 2.5 µg) delivered by Respimat over 48 weeks to placebo, one adult study [40] involved an 8-week, three-way crossover design with 5 µg tiotropium (2 puffs of 2.5 µg), 10 µg tiotropium (2 puffs of 5 µg) and placebo, and was excluded from further analyses and the primary meta-analyses. The remaining four trials enrolled a total of 1433 participants (2.5 µg dose, n=528) and were pooled for meta-analyses to inform the Task Force's judgements.

Across the four parallel-arm trials including children, adolescents and adults, the addition of tiotropium 5 µg resulted in improvements in mean peak FEV₁ response compared with placebo (123 mL, 95% CI 88.2–158.7 mL), which was statistically significant but a clinically trivial difference. Serious imprecision in the certainty estimates was also noted for each age group. The addition of tiotropium 5 µg also marginally improved ACQ-7 (−0.11 points, 95% CI −0.2–0.01 points) and prevented asthma worsening (based on exacerbations or symptoms, risk ratio 0.79, 95% CI 0.7–0.89; absolute risk 133, 95% CI 54–122, fewer worsening episodes per 1000) compared with placebo, but again, serious imprecision in the certainty estimates was noted for children and adolescents. In children and adolescents, addition of tiotropium 2.5 µg did not improve asthma control scores but did improve FEV₁ % pred (mean difference 4.99, 95% CI 2.84–7.15) and reduced asthma worsening (risk ratio 0.66, 95% CI 0.45–0.97). Post hoc analyses of adjusted mean trough FEV₁/forced vital capacity responses in children also demonstrated statistically significant improvements at all time-points versus placebo with both tiotropium doses, with the exception of tiotropium 2.5 mg at week 8.

In the two adult trials, treatment with tiotropium 5 µg did not result in significant differences in AQLQ (mean difference 0.10 points, 95% CI −0.04–0.23 points), but did increase the time to first exacerbation requiring OCS (hazard ratio for placebo 0.79, 95% CI 0.62–1.01). Asthma exacerbations requiring hospitalisation were too infrequent in both the tiotropium (16 out of 453 subjects) and placebo (20 out of 454 subjects) arms to draw conclusions [41]. The crossover study in adults [40] that was excluded from the primary analysis similarly noted beneficial effects of tiotropium 5 µg (mean difference 139 mL, 95% CI 96–181 mL) and 10 µg (mean difference 170 mL, 95% CI 128–213 mL) on peak FEV₁ response in adults.

Adverse events were less frequent in the tiotropium arms compared with placebo in these four trials (risk ratio 0.92, 95% CI 0.86–0.98). Severe adverse events were equally infrequent across treatment arms.

Benefits

Long-acting muscarinic antagonist (LAMA) treatment in children, adolescents and adults with severe asthma may improve FEV₁ and may reduce loss of asthma control. In adults, treatment with tiotropium 5 µg also improves asthma control and increases time to first exacerbation.

Harms

There was a lower frequency of adverse events in children, adolescents and adults treated with tiotropium 5 µg compared with placebo. The frequency of severe adverse events was also low and nearly equal to placebo.

Conclusions and research needs

The addition of tiotropium improves FEV₁ and provides beneficial effects on symptom control in children, adolescents and adults with severe asthma not controlled with GINA step 4–5 and NAEPP step 5 combination therapies. There were too few severe exacerbations requiring OCS to draw definitive conclusions as to benefit. Based on the estimated beneficial effects observed for tiotropium, the Task Force judged that these benefits outweigh the adverse effects, burdens and costs associated with this treatment for the management of severe asthma.

In the combined age groups, tiotropium was effective in preventing the composite outcome for asthma worsening inclusive of symptom control and exacerbations. However, the effect of treatment was not significant in adolescents and children, likely due to the smaller sample sizes and shorter study duration of these trials. There is insufficient evidence for the beneficial effects of tiotropium on severe exacerbations in children and adolescents with severe asthma, which should be investigated in longer-term trial cohorts of sufficient size. There are additional LAMAs (umeclidinium and glycopyrronium) currently available which could be alternative long-term bronchodilator therapies for severe asthma. Treatments with umeclidinium and glycopyrronium have beneficial effects on lung function and symptom control in individuals with mild to moderate, persistent asthma [43–45], but have not been evaluated as an adjunct therapy for severe asthma.

Future studies should also focus on the identification of severe asthma subgroups preferentially responsive to LAMAs that might benefit from the step-wide addition of muscarinic antagonists compared with alternative step-up options such as LABAs or increased ICS dosing. Subgroup analyses of trial cohorts with mild to moderate persistent asthma subjects have suggested that subgroups with fixed or baseline airflow obstruction might preferentially respond to LAMAs [45, 46]. Three randomised controlled trials only included subjects with FEV₁ <80% predicted. Kerstjens et al. [47] showed beneficial effects in both those with screening FEV₁ <60% or 60–80% predicted. Two trials in children and adolescents enrolled asthma patients with FEV₁ 60–90% predicted [42, 48]. Hence, it is not clear whether individuals, particularly adults, with severe asthma and higher lung function on combination therapy with high-dose ICS and a LABA will benefit from the addition of a LAMA.

A responder analysis of a severe asthma trial cohort showed equally beneficial effects when comparing subgroups based on baseline lung function, age, sex, ethnicity, body mass index and racial groups. Differential inter-racial effects are difficult to ascertain since minority racial groups (African-Americans and Asians) and Hispanic ethnic groups represented the vast minority of subjects in these trials [47]. Future trials in increasingly ethnically diverse severe asthma cohorts should provide insight into the beneficial effects of LAMAs in these groups, which experience a substantial proportion of asthma-related morbidity. Studies to evaluate responder subgroups based on genetic variation (pharmacogenetic studies) should also be performed using DNA samples from prior and future clinical trials.

What others are saying

The 2019 GINA guidelines [20] recommend the use of tiotropium as an add-on therapeutic option at step 4 or 5 for patients with exacerbations despite treatment with ICS and LABA.

The NAEPP guidelines [49] do not outline any role for the muscarinic antagonists.

ERS/ATS recommendation

For children, adolescents and adults with severe asthma uncontrolled despite GINA step 4–5 or NAEPP step 5 therapies, we recommend the addition of tiotropium (strong recommendation, moderate quality of evidence).

Summary of the evidence

The previous ERS/ATS guidelines made a conditional recommendation that long-term macrolide antibiotics should not be used in the treatment of adults or children with severe asthma, based on available evidence [1]. Since then, six randomised controlled trials have been conducted [50–55], of which five included only adults and one included only children 6– <18 years of age. There were varying definitions of persistent symptomatic or uncontrolled asthma and none met ERS/ATS criteria for severity. Three studies used azithromycin; of these, two (totalling 529 participants) used doses ranging from 250 to 500 mg three times per week for a treatment period of 26–48 weeks [50, 51]. The other (n=97) used a dose of 600 mg·day⁻¹ for 3 days and 600 mg·week⁻¹ thereafter for 11 weeks [53]. The clarithromycin randomised controlled trials (totalling 171 participants) used 600 mg twice daily, ranging from 8 to 16 weeks in treatment duration [54, 55]. In children (n=55), azithromycin nightly doses were given according to body weight, ranging from 250 mg for 25–40 kg and 500 mg for >40 kg for a total of 12 months (the study was prematurely terminated at 30 weeks due to lack of clinical efficacy) [52].

Compared with placebo, during 48 weeks of follow-up, azithromycin reduced the number of combined moderate and severe exacerbations (1.07 versus 1.86 events·patient⁻¹·year⁻¹; risk ratio 0.59, 95% CI 0.47–0.74) [51]. Additionally, macrolides reduced the number of patients with at least one moderate or severe asthma exacerbation and the time to first exacerbation. It did not, however, reduce the rate of severe exacerbations (25.3% versus 34.6%; risk ratio 0.77, 95% CI 0.44–1.34) in children or adults, during a follow-up period ranging from 24 to 48 weeks [50–52]. Neither azithromycin nor clarithromycin treatment improved ACQ-7 (mean difference 0.11 points, 95% CI −0.34–0.12 points) or AQLQ (mean difference 0.16 points, 95% CI −0.06–0.37 points) in adults beyond the MCID.

Relative to placebo, treatment with azithromycin or clarithromycin in adults or children was not associated with changes in post-bronchodilator FEV₁ % pred (mean difference 1.95, 95% CI −2.42–6.32) or pre-bronchodilator FEV₁ L (mean difference 0.37, 95% CI −2.17–2.91) that reached the MCID [50, 53, 54].

The effects of clarithromycin on airway inflammation were inconsistent, with only one of two studies showing significant reductions in airway neutrophilia [55]. Compared with placebo, macrolide therapy in adults was associated with a lower number of lower respiratory tract infections requiring antibiotics (20.9% versus 35.6%; risk ratio 0.60, 95% CI 0.45–0.79) [50, 51].

The number of study participants with at least one adverse event (67.3% versus 72.2%; risk ratio 0.93, 95% CI 0.73–1.19) and the number of serious adverse events (9.1% versus 11.4%; risk ratio 0.81, 95% CI 0.52–1.24) in adults or children were not different from placebo [50, 51, 53, 54].

Benefits

Macrolides reduce the number of asthma exacerbations and at least one study suggests that this effect is similar for participants with or without eosinophilia [51]. The effect on asthma control and quality of life does not reach the MCID.

Harms

Chronic macrolide therapy has been associated with increased incidence of diarrhoea; however, the number of serious adverse events or number of participants with at least one adverse event is not different to placebo. Although macrolides have a potential risk for QT prolongation or hearing loss, the frequency of these events is not reported to be higher than in the placebo arm in patients whom at baseline had no hearing deficits or abnormally prolonged corrected QT [51]. Relative to placebo, the prevalence of nasal and oropharyngeal macrolide-resistant streptococci increased in one study [50], but not in another [51]. Those treated with azithromycin for 48 weeks had reduced airway Haemophilus influenzae load, with no changes to total or pathogenic bacterial loads. Although sputum macrolide resistance genes increased in this group, there was a lower rate of antibiotic use and of adverse events due to clinically diagnosed infections [51, 56].

Conclusions and research needs

Relative to placebo, chronic macrolide therapy reduces the risk of having an asthma exacerbation. However, there is no conclusive evidence that treatment shows any effect in reducing severe exacerbations or hospitalisations. The effects of macrolides on asthma have been limited to participants with uncontrolled or persistently symptomatic disease that may or may not be exacerbation prone; therefore, it is unknown whether this therapy will improve outcomes among those meeting ERS/ATS criteria for severe asthma. The emergence of antimicrobial resistance associated with prolonged antibiotic use such as macrolide therapy is a critical public health issue. Potential benefits in severe asthma need to be carefully considered against this background risk from both the perspective of an individual patient and the wider community.

What others are saying

The 2019 GINA guidelines [20] recommend prescribing add-on low-dose macrolide in patients who do not respond to standard treatment, but classify its use off-label and suggest weighing the benefits against the potential for antibiotic resistance.

In the 2016 BTS/SIGN guidelines [57], the use of macrolide antibiotics in asthma was not recommended; the 2019 BTS/SIGN revised guidelines [39] do not mention the use of chronic macrolide therapy for severe asthma.

The FDA has not approved the use of chronic macrolide therapy for asthma.

ERS/ATS recommendations

We suggest a trial of macrolide treatment to reduce asthma exacerbations in adult asthma subjects on GINA/NAEPP step 5 therapy that remain persistently symptomatic or uncontrolled (conditional recommendation, low quality of evidence)

We suggest against the use of chronic macrolide treatment in children and adolescents with severe uncontrolled asthma (conditional recommendation, low quality of evidence).

Summary of the evidence

Dupilumab is a fully human monoclonal antibody directed against the α subunit of IL-4 receptor. It blocks signalling of two key type 2 cytokines: IL-4 and IL-13. We identified three randomised, placebo-controlled trials evaluating dupilumab as add-on therapy in patients with moderate to severe asthma [58–60]. Two randomised controlled trials included adolescent (age 12–17 years) and adult (age ≥18 years) participants [59, 60], and one trial included only adult participants [58].

In the phase 2b dose-ranging clinical trial [58], four dosing regimens of dupilumab were studied: 200 or 300 mg of the drug administered s.c. every 2 or 4 weeks for 24 weeks. 769 adult patients with uncontrolled asthma, despite use of medium- to high-dose ICS and LABA, were randomised 1:1:1:1:1 into four treatment arms or placebo. The primary end-point was change in FEV₁ (L) at 12 weeks in patients with blood eosinophil counts ≥300 µL⁻¹. Pre-specified secondary end-points at weeks 12 and 24 included asthma exacerbation rate, time to severe exacerbation, asthma symptom score, asthma quality of life and change in FEV₁ (% pred).

One phase 3 efficacy and safety randomised controlled trial [59] was in adolescents and adults with moderate to severe uncontrolled asthma and it evaluated dupilumab add-on therapy at doses 200 mg (after a loading dose of 400 mg) or 300 mg (after a loading dose of 600 mg) every 2 weeks for 52 weeks. A total of 1902 participants were randomised 2:2:1:1 with matched volume placebo. Primary end-points were annualised exacerbation rates (week 52) and absolute change in FEV₁ (week 12). Secondary end-points included change in FEV₁ % pred, ACQ and AQLQ, as well as subgroup analysis by blood eosinophil count.

The second phase 3 randomised controlled trial [60] evaluated dupilumab (300 mg every 2 weeks for 24 weeks) in 210 adolescents and adults with severe OCS-dependent asthma. After a steroid dose-optimisation period, patients were randomised 1:1 to receive dupilumab or placebo. OCS dose was adjusted down during weeks 4–20. The primary end-point was percentage reduction in OCS dose required to maintain asthma control. Secondary end-points included proportion of patients with at least 50% reduction in OCS dose and proportion of patients with reduction in OCS dose to <5 mg·day⁻¹.

These three trials were pooled for meta-analysis (see evidence profiles in the supplementary material). Effects of dupilumab on exacerbation rate, asthma control, asthma quality of life, lung function and side-effects were assessed for 200 and 300 mg doses at 24 and 52 weeks. Differences in effect size by blood eosinophils were also assessed.

Relative to participants assigned to placebo, those assigned to dupilumab (200 or 300 mg every 2 weeks; 24 and 52 weeks) experienced substantial (46–70.5%) reductions in annualised rates of asthma exacerbations. Dupilumab therapy resulted in a greater proportion of participants with OCS-dependent severe asthma experiencing >50% reduction in OCS dose (relative risk 1.49, 95% CI 1.22–1.83; absolute risk 26, 95% CI 12–44, more achieved 50% reduction per 100), reduction in OCS dose to <5 mg·day⁻¹ (relative risk 1.92, 95% CI 1.46–2.53; absolute risk 344, 95% CI 172–572, more per 1000) and discontinuation of maintenance OCS (relative risk 1.81, 95% CI 1.28–2.57). Improvements in FEV₁, ACQ-5 and AQLQ were statistically significant, but did not reach the MCID.

The effect size for all above outcomes was larger in patients with blood eosinophil counts ≥300 µL⁻¹ when compared with eosinophils <300 µL⁻¹ (see evidence profiles in the supplementary material). One study further stratified the study cohort by blood eosinophils <150, 150–300 and ≥300 µL⁻¹ [59]. Rate ratio for annualised severe exacerbation event rate at 52 weeks, pooled for doses 200 and 300 mg every 2 weeks, was 0.33 (95% CI 0.26–0.42; 0.386 versus 1.158 events·patient⁻¹·year⁻¹) for the subgroup with blood eosinophils ≥300 µL⁻¹, 0.60 (96% CI 0.43–0.83; 0.515 versus 0.855 events·patient⁻¹·year⁻¹) for the subgroup with blood eosinophils 150–300 µL⁻¹ and 1.04 (95% CI 0.76–1.43; 0.604 versus 0.576 events·patient⁻¹·year⁻¹) for the subgroup with blood eosinophils <150 µL⁻¹. The same study reported similar results for exacerbations and lung function when stratified by F_ENO ≥50, ≥25– <50 and <25 ppb. A post hoc biomarker interaction analysis found the greatest treatment response in patients with F_ENO ≥25 ppb and blood eosinophils ≥150 µL⁻¹.

Benefits

Dupilumab, as add-on therapy in patients with asthma that is uncontrolled on medium- to high-dose ICS and LABA, may reduce exacerbations and improve asthma symptoms and lung function. The efficacy is greater in patients with type 2 biomarkers (blood eosinophils ≥150 µL⁻¹ or F_ENO ≥25 ppb) Dupilumab may reduce OCS dose in patients with severe corticosteroid-dependent asthma.

Harms

The risk of dupilumab therapy appears to be small, with injection site reaction as the most common treatment-related adverse effect. Frequencies of serious and any side-effects were similar with dupilumab when compared with placebo. However, the mechanisms and potential clinical significance of treatment-related transient blood eosinophilia is not fully understood and needs further elucidation. As dupilumab-mediated eosinophilia has not been associated with adverse events, there are no specific monitoring recommendations.

Conclusions and research needs

Dupilumab add-on therapy substantially decreases exacerbations in moderate to severe uncontrolled asthma [58–60]. It is effective in reducing OCS dose in patients with severe OCS-dependent asthma. Dupilumab therapy is also associated with improvements in lung function, asthma control and quality of life. More robust improvements were observed in patients with higher eosinophil levels.

Ongoing and future studies should provide additional information on long-term safety and durability of response to dupilumab therapy. More data on efficacy and safety are also needed in children and adolescents. Future studies should also focus on identifying specific disease and population characteristics that can predict response to this therapy.

What others are saying

The 2019 GINA guidelines [20] recommend dupilumab as add-on option for patients with severe eosinophilic or type 2 asthma uncontrolled on high-dose ICS and LABA, or requiring maintenance OCS.

The NICE guidelines [30] do not currently include dupilumab as an add-on therapeutic option for asthma.

ERS/ATS recommendation

We suggest dupilumab as add-on therapy for adult patients with severe eosinophilic asthma and for those with severe corticosteroid-dependent asthma regardless of eosinophil levels (conditional recommendation).