Abstract

Background: Six biologic agents are now approved for patients with severe asthma. This meta-analysis aimed to assess the efficacy and safety of licensed biologic agents in patients with severe asthma, including the recently approved tezepelumab.

Methods: We searched MEDLINE, Embase and CENTRAL to identify randomised controlled trials involving licensed biologics until 31 January 2023. We used random-effects meta-analysis models for efficacy, including subgroup analyses by individual agents and markers of T2-high inflammation (blood eosinophils and fractional exhaled nitric oxide), and assessed safety.

Results: 48 studies with 16 350 patients were included in the meta-analysis. Biologics were associated with a 44% reduction in the annualised rate of asthma exacerbations (rate ratio 0.56, 95% CI 0.51–0.62) and 60% reduction of hospitalisations (rate ratio 0.40, 95% CI 0.27–0.60), a mean increase in the forced expiratory volume in 1 s of 0.11 L (95% CI 0.09–0.14), a reduction in asthma control questionnaire by 0.34 points (95% CI −0.46–−0.23) and an increase in asthma quality of life questionnaire by 0.38 points (95% CI 0.26–0.49). There was heterogeneity between different classes of biologics in certain outcomes, with overall greater efficacy in patients with T2 inflammation. Overall, biologics exhibited a favourable safety profile.

Conclusions: This comprehensive meta-analysis demonstrated that licensed asthma biologics reduce exacerbations and hospitalisations, improve lung function, asthma control and quality of life, and limit the use of systemic corticosteroids, with a favourable safety profile. These effects are more prominent in patients with evidence of T2 inflammation.

Shareable abstract

Asthma biologics reduce exacerbations and hospitalisations, improve lung function, asthma control and quality of life, and limit the use of systemic corticosteroids, with a favourable safety profile, notably in patients with T2 inflammation. https://bit.ly/3Ijs1OQ

Introduction

Asthma is a heterogeneous disease affecting an estimated 339 million people worldwide [1]. Severe asthma requires treatment with high-dose inhaled corticosteroids (ICS) plus an additional controller, usually a long-acting β-agonist (LABA) and/or systemic corticosteroids, to achieve asthma control or when the condition remains uncontrolled despite maximal optimised therapy [2]. The Global Initiative for Asthma (GINA) reports an estimated prevalence of severe asthma in around 3.7% of asthma patients [1, 3]. In different studies, approximately 6.9% of children and 10% of adult patients have severe asthma, although a large prevalence range (1.8–38%) has been reported [4, 5]. Despite the use of inhaled asthma controller medicine, currently available biologic therapies and oral corticosteroids (OCS), severe asthma in many patients remains uncontrolled [2, 6]. Patients with severe asthma are at an increased risk of exacerbations, hospitalisations and mortality [7], leading to a poor quality of life [8] and imposing a significant socioeconomic burden [9].

The United States Food and Drug Administration and European Medicines Agency has approved monoclonal antibodies against immunoglobulin E (anti-IgE, omalizumab), interleukin 5 (anti-IL5, mepolizumab and reslizumab) and its receptor-α (anti-IL5Rα, benralizumab), interleukin 4 receptor-α (anti-IL4α, dupilumab), and, more recently, against thymic stromal lymphopoietin (anti-TSLP, tezepelumab). Tezepelumab is the first biologic to receive an indication for severe asthma irrespective of the presence of type 2 inflammation. The mechanism of action and effects of approved biologics for asthma are summarised in supplementary appendix 1.

In the present systematic review and meta-analysis, we evaluate the efficacy and safety of currently licensed biologics through the results of randomised controlled trials (RCTs), with a particular focus on studies of the recently approved tezepelumab.

MethodsLiterature search and inclusion criteria

We conducted a systematic literature search of MEDLINE (PubMed), Embase and CENTRAL, from inception until 31 January 2023, to identify RCTs that assess the efficacy of approved biological therapies in severe or uncontrolled asthma. Searches were carried out independently by two reviewers (C. Kyriakopoulos and A. Gogali) using a standardised approach, followed by screening through relevant titles and abstracts, and then a full-text review. Disagreements were resolved by consensus, with unresolved conflicts decided by a third reviewer (K. Kostikas). In studies that contained parallel arms of different biologic agent doses, only the arm with a dose close to the approved dose was used in the analysis (supplementary appendix 1). The search strategy algorithm and study selection are shown in supplementary appendix 4. We submitted the systematic review protocol for registration on PROSPERO (CRD42022361075) (supplementary appendix 8) and followed PRISMA reporting guidelines (supplementary appendix 2) [10].

Eligible studies met the following criteria based on the participants–intervention–comparator–outcome–time-frame (PICOT) format.

Participants

Patients aged >12 years old with moderate-to-severe, severe, uncontrolled/inadequately controlled asthma.

Intervention

Any approved biologic therapy for asthma (at the approved dose).

Comparator group

Standard of care treatment.

Outcomes

Primary end-point:

Secondary end-points:

Hospitalisations for asthma exacerbation (annual rate)

Lung function (forced expiratory volume in 1 s (FEV1) in L)

OSC use and sparing effect (reduction to <5 mg·day−1 prednisolone, reduction >50%, dose reduction %, discontinuation)

Asthma control questionnaire (ACQ) and asthma control test (ACT)

Asthma symptom score (ASS)

Asthma quality of life questionnaire (AQLQ)

Safety (adverse events, serious adverse events, pneumonia, death)

Time-frame

Without restriction in duration of studies.

Data extraction

Two authors (C. Kyriakopoulos and A. Gogali) concurrently reviewed all eligible studies to perform data extraction. The reviewers worked independently during study data extraction; disagreements, if any, were resolved by discussion to obtain consensus, with unresolved conflicts decided by a third reviewer (K. Kostikas). Obtained data were validated by a third independent author (G. Markozannes).

From each eligible study, we recorded information about the first author, publication year, journal, study design, follow-up time, population characteristics, total and biological-treated sample size, treatment indication, biological agent and comparator dose, and use of corticosteroids. Rate ratios along with their confidence intervals were extracted for exacerbations and hospitalisations, assessed as binary outcomes, while risk ratios were extracted for OCS reduction to <5 mg prednisolone equivalent, OCS discontinuation, OCS reduction >50%, adverse effects, serious adverse effects, pneumonias and deaths. Mean change differences from baseline along with their confidence intervals were extracted for FEV1 in litres, ACQ, AQLQ, ACT, ASS and OCS dose reduction.

Data analysis

We calculated rate ratios, risk ratios and summary mean differences, along with the corresponding 95% confidence intervals, using DerSimonian and Laird [11] random-effects meta-analysis models. If only subgroup-specific effects were reported in a study, we estimated the overall study effect using a fixed-effect model before including it in the meta-analysis, considering that participants were sampled from the same target population within the same trial. The presence and the degree of heterogeneity were assessed using the I2 test (ranging from 0 to 100%) [12]. Heterogeneity between subgroups was assessed by a Q-test [13] based on an analysis of variance.

We assessed the efficacy of the biologic agents in a priori defined subgroups of patients with high and low blood eosinophils (≥300 cells·μL−1 and <300 cells·μL−1, respectively) and exhaled nitric oxide fraction (FENO) (≥25 ppb and <25 ppb, respectively) on exacerbations, FEV1 (litres), ACQ (points) and AQLQ (points). We also evaluated other cut-points used in trials, e.g. eosinophils (<150 and ≥150 cells·μL−1; <150, 150–299 and ≥300 cells·μL−1) and FENO (<25, 25–49 and ≥50 ppb).

We assessed possible small-study effects (an indication of publication bias) by visual inspection of funnel plots and Egger test [14] when at least 10 studies were available in the meta-analyses, with a p-value <0.1 indicating the presence of small-study effects. All analyses were performed using Stata (version 16.1; StataCorp, College Station, TX, USA).

ResultsStudy identification and selection

The search identified a total of 3666 studies; following removal of duplicates, screening and full-text review, 51 RCTs with 16 400 participants with published data were shortlisted for inclusion in the systematic review. Figure 1 shows a flow chart of the study selection process. Three studies that included 20 or fewer participants were subsequently excluded from quantitative synthesis [15–17].

FIGURE 1

Systematic review and meta-analysis flow-diagram illustrating systematic search and screening strategy, including number of studies meeting eligibility criteria and number of excluded studies.

The 48 studies included in the meta-analysis involved 16 350 participants; 7582 (46.4%) in the control arm and 8768 (56.6%) received a biologic agent, out of whom 4211 (48%) received omalizumab, 681 (7.8%) mepolizumab, 393 (4.5%) reslizumab, 1791 (20.4%) benralizumab, 945 (10.8%) dupilumab and 747 (8.5%) tezepelumab. Table 1 shows the characteristics of the included studies.

TABLE 1

Study characteristics of 51 studies that met eligibility criteria for assessment of biologic agents licensed for severe asthma

Study outcomesPrimary outcome: effect of biologics on exacerbations

In 25 studies with 10 867 participants (5836 on biologics), the use of biologics reduced exacerbations by 44% (rate ratio 0.56, 95% CI 0.51–0.62; I2=58.4%) (figure 2a). Exacerbations were reduced in all biologic categories, with a nonsignificant greater improvement in anti-IL4α (rate ratio 0.44, 95% CI 0.32–0.61, I2=49.4%) and anti-TSLP (rate ratio 0.45, 95% CI 0.30–0.66, I2=70%) (psubgroup_heterogeneity=0.15).

FIGURE 2

a) Forest plot of asthma exacerbations rate ratios comparing biologic agents and control treatments. b) Forest plot of hospitalisations due to asthma exacerbation rate ratios comparing biologic agents and control treatments. Sample sizes are given for participants receiving biologic agent and participants receiving standard of care treatment, included in the study, when data were available. Summary estimates presented separately for each biologic agent category. DL: DerSimonian and Laird method IL: interleukin; TSLP: thymic stromal lymphopoietin.

Exacerbation reduction by type 2 inflammation subgroups

12 studies with 6015 participants assessed the efficacy of biologics in subgroups based on blood eosinophil counts ≥300 and <300 cells·μL−1, demonstrating greater reductions in patients with higher eosinophil counts (rate ratio 0.38 versus 0.67, psubgroup_heterogeneity=0.001) (supplementary figure S1).

Four studies with 2372 participants assessed the efficacy of dupilumab and tezepelumab in subgroups based on blood eosinophil counts ≥150 and <150 cells·μL−1, again demonstrating greater exacerbation reduction in patients with higher eosinophil counts (rate ratio 0.40 versus 0.80, psubgroup_heterogeneity=0.007) (supplementary figure S2).

Two studies with 2012 participants assessed the efficacy of dupilumab and tezepelumab in terms of asthma exacerbations in three subgroups of patients based on blood eosinophil counts ≥300, 150–299 and <150 cells·μL−1, which consistently showed a greater exacerbation reduction in patients with higher eosinophil counts (ratio ratio 0.31 versus 0.57 versus 0.83, psubgroup_heterogeneity<0.001) (supplementary figure S3).

Four studies with 2485 participants assessed the efficacy of dupilumab and tezepelumab in subgroups based on FENO levels ≥25 and <25 ppb, showing greater exacerbation reduction in patients with higher FENO (rate ratio 0.33 versus 0.70, psubgroup_heterogeneity<0.001) (supplementary figure S4).

Three studies with 2196 participants assessed the efficacy of dupilumab and tezepelumab in subgroups based on FENO levels >50, 25–49 and <25 ppb, with greater exacerbation reduction in patients with higher FENO (rate ratio 0.30 versus 0.39 versus 0.72, psubgroup_heterogeneity<0.001) (supplementary figure S5).

Effect of biologics on hospitalisations due to asthma exacerbation

11 studies with 4182 participants (2232 on biologics) reported a 60% reduction in hospitalisations (rate ratio 0.40, 95% CI 0.27–0.60, I2=32%) (figure 2b). The greatest numerical reduction was shown with anti-TSLP (rate ratio 0.28).

Effect of biologics on lung function

31 studies with 10 323 participants (5551 on biologics) reported results on FEV1 (in L). Biologics improved FEV1 by 0.11 L (95% CI 0.09–0.14; I2=50.1%) (figure 3). There was an indication of heterogeneity between the biologics subgroups (psubgroup_heterogeneity=0.08) that could mainly be attributed to the differences in effect sizes between anti-IL4α (0.17, 95% CI 0.11–0.22 L, I2=27.3%) and anti-IgE (0.08, 95% CI 0.04–0.13 L, I2=50.6%).

FIGURE 3

Forest plot of the mean change difference in forced expiratory volume in 1 s comparing biologic agents and control treatments. Sample sizes are given for the total number of participants and the participants receiving intervention included in the study. Summary estimates presented separately for each biologic agent category. DL: DerSimonian and Laird method; IL: interleukin; TSLP: thymic stromal lymphopoietin.

Improvement in lung function by type 2 inflammation subgroups

12 studies with 5405 participants assessed the efficacy of biologics in subgroups based on blood eosinophil counts ≥300 and <300 cells·μL−1, showing a greater FEV1 improvement in patients with higher eosinophil counts (0.18 versus 0.07 L, psubgroup_heterogeneity<0.001) (supplementary figure S6).

Four studies with 2185 participants assessed the efficacy of dupilumab and tezepelumab in subgroups based on blood eosinophil counts ≥150 and <150 cells·μL−1, with a greater FEV1 improvement in patients with higher eosinophil counts (0.18 versus 0.10 L, psubgroup_heterogeneity=0.074) (supplementary figure S7).

One study with 922 participants assessed the efficacy of dupilumab on lung function in three subgroups of patients categorised by blood eosinophil counts ≥300, 150–299 and <150 cells·μL−1, with a greater FEV1 improvement in patients with higher eosinophil counts (0.24 versus 0.00 versus 0.09 L, psubgroup_heterogeneity=0.001) (supplementary figure S8).

Three studies with 1372 participants assessed the efficacy of dupilumab and tezepelumab in subgroups of patients categorised by FENO ≥25 versus <25 ppb, showing a greater improvement in FEV1 in patients with higher FENO (0.21 L versus 0.08, psubgroup_heterogeneity=0.023) (supplementary figure S9).

Two studies with 1107 participants assessed the efficacy of dupilumab in three subgroups categorised by FENO levels ≥50, 25–49 and <25 ppb, with a greater improvement in FEV1 in patients with higher FENO (0.37 versus 0.13 versus 0.11 L, psubgroup_heterogeneity=0.004) (supplementary figure S10).

Effect of biologics on asthma control, symptoms and quality of lifeImprovement in ACQ score

25 studies with 7999 participants (4315 on biologics) reported results on the change in ACQ score. Biologics led to improved asthma control, reducing the ACQ score by 0.34 points (95% CI −0.46–−0.23, I2=89.5%) with no indication of heterogeneity in the effect sizes between biologics (psubgroup_heterogeneity=0.962); however, the results did not reach the minimal clinically important difference (MCID) for the ACQ (−0.50 points) (figure 4a) [66].

FIGURE 4

a) Forest plot of the mean change difference in asthma control questionnaire comparing biologic agents and control treatment. b) Forest plot of the mean change difference in asthma quality of life questionnaire (AQLQ) comparing biologic agents and control treatment. Sample sizes are given for the total number of participants and the participants receiving intervention included in the study. Summary estimates presented separately for each biologic agent category. DL: DerSimonian and Laird method; IL: interleukin; TSLP: thymic stromal lymphopoietin.

Improvement in ACQ score by type 2 inflammation subgroups

10 studies with 4239 participants assessed the efficacy of biologics in subgroups of patients based on blood eosinophil counts ≥300 and <300 cells·μL−1, demonstrating a greater improvement of ACQ score in patients with higher eosinophil counts (−0.39 versus −0.20 points, psubgroup_heterogeneity=0.008) (supplementary figure S11).

Two studies with 1091 participants assessed the efficacy of tezepelumab in subgroups of patients categorised by blood eosinophil counts ≥150 and <150 cells·μL−1, demonstrating a greater improvement in patients with higher eosinophil counts (−0.42 versus −0.07 points, psubgroup_heterogeneity=0.011) (supplementary figure S12).

One study with 220 participants assessed the efficacy of tezepelumab in subgroups of patients with high and low FENO, demonstrating a greater reduction of ACQ score in patients with FENO ≥25 versus <25 ppb (−0.67 versus −0.10 points, psubgroup_heterogeneity=0.014) (supplementary figure S13).

Change in ACT score

Three studies with 688 participants reported the efficacy of omalizumab on the change in ACT, with a trend for improvement by 0.58 points (95% CI −0.12–1.29); however, they did not reach statistical significance or the MCID (three points for adults and two points for patients aged 12–18 years) (supplementary figure S14) [67, 68].

Change in ASS

Eight studies with 3380 participants reported an improvement in asthma symptoms by reducing the ASS by 0.30 points (95% CI −0.48–−0.12, I2=73.4%). Significant heterogeneity between biologics was observed (psubgroup_heterogeneity<0.001) (supplementary figure S15).

Effect of biologics on AQLQ score

20 studies with 7202 participants (3810 on biologics) reported AQLQ results. Biologics improved quality of life, increasing the AQLQ score by 0.38 points (95% CI 0.26–0.49, I2=75.5%); however, not reaching the MCID for the AQLQ (0.50 points) [69]. No significant heterogeneity was observed between biologics (psubgroup_heterogeneity=0.179) (figure 4b).

Improvement in AQLQ by type 2 inflammation subgroups

Six studies with 2453 participants assessed the efficacy of biologics in subgroups of patients categorised by blood eosinophil counts ≥300 and <300 cells·μL−1, demonstrating a greater numerical increase of AQLQ score in patients with higher eosinophil counts (0.34 versus 0.21 points, psubgroup_heterogeneity=0.263) (supplementary figure S16).

One study with 947 participants assessed the efficacy of tezepelumab in subgroups of patients based on blood eosinophil counts ≥150 and <150 cells·μL−1, demonstrating a greater improvement in AQLQ score in patients with higher eosinophil counts (0.41 versus 0.11 points, psubgroup_heterogeneity=0.055) (supplementary figure S17).

One study with 200 participants assessed the efficacy of tezepelumab in subgroups of patients with FENO levels ≥25 and <25 ppb, demonstrating a greater increase of AQLQ score with higher FENO (0.65 versus 0.02 points, psubgroup_heterogeneity=0.012) (supplementary figure S18).

Effect of biologic agents on OCS useOCS reduction to doses <5 mg·day−1 prednisolone

Five studies with 820 participants (407 on biologics) reported results on OCS reduction to doses <5 mg·day−1 prednisolone. Biologics increased the probability of reduction by 74% (risk ratio 1.74, 95% CI 1.23–2.46; I2=44.1%) (figure 5a).

FIGURE 5

a) Forest plot of oral corticosteroids reduction to doses <5 mg·day−1 prednisolone risk ratios comparing biologic agents and control treatment. Sample sizes are given for participants receiving biologic agent and participants receiving standard of care treatment, included in the study, when data were available. Summary estimates presented separately for each biologic agent category. b) Forest plot of oral corticosteroids dose reduction >50% risk ratios comparing biologic agents and control treatment. Sample sizes are given for the total number of participants and the participants receiving intervention included in the study. Summary estimates presented separately for each biologic agent category. c) Forest plot of oral corticosteroids discontinuation risk ratios comparing biologic agents and control treatment. Sample sizes are given for the total number of participants and the participants receiving intervention included in the study. Summary estimates presented separately for each biologic agent category. d) Forest plot of the mean change difference of oral corticosteroids reduction (%) comparing biologic agents and control treatment. Sample sizes are given for the total number of participants and the participants receiving intervention included in the study. Summary estimates presented separately for each biologic agent category. DL: DerSimonian and Laird method; IL: interleukin; TSLP: thymic stromal lymphopoietin.

OCS dose reduction >50%

Five studies with 820 participants (407 on biologics) reported results on OCS dose reduction by >50%. Biologics increased the probability of >50% reduction by 68% (risk ratio 1.68, 95% CI 1.29–2.19; I2=27.2%) (figure 5b).

OCS discontinuation

Four studies with 620 participants (307 on biologics) reported results on OCS discontinuation. The use of biologics showed an indication of an increase in the probability of OCS discontinuation (risk ratio 1.63, 95% CI 0.99–2.70; I2=57.2%) (figure 5c).

OCS dose reduction (%)

Four studies with 668 participants (331 on biologics) reported results on OCS dose reduction (%) demonstrating a mean reduction of OCS dose by 34.67% (95% CI 18.46–50.89%; I2=65.7%) (figure 5d).

Safety profile of biologicsAdverse events

39 studies with 14 879 participants (8165 on biologics) reported adverse events, demonstrating that treatment administration was not associated with increased risk; in fact, there was a 5% reduction of adverse events (risk ratio 0.95, 95% CI 0.92–0.99, I2=72.5%) (supplementary figure S19).

Serious adverse events

38 studies with 14 905 participants (8043 on biologics) reported serious adverse events, demonstrating that treatment administration was associated with a significant reduction of adverse events (risk ratio 0.76, 95% CI 0.65–0.89, I2=24.3%) (supplementary figure S20). There was an indication of heterogeneity between agents (psubgroup_heterogeneity=0.012) that could be driven by the difference in the effect estimates of anti-IL5/5Rα (risk ratio 0.71, 95% CI 0.56–0.89) and anti-IL4α (risk ratio 1.13, 95% CI 0.78–1.62).

Deaths

28 studies with 12 951 participants (7177 on biologics) reported deaths during the study period, with no difference between the biologics and control groups (risk ratio 0.91, 95% CI 0.39–2.09, I2=0%) (supplementary figure S21).

Pneumonias

10 studies with 5550 participants (3239 on biologics) reported results on pneumonias during the study period, demonstrating that the use of biologics was associated with an overall trend of a 30% reduction of pneumonias compared to the control group (risk ratio 0.70, 95% CI 0.36–1.37, I2=0%) (supplementary figure S22).

Risk of bias, small-study effects and influence analysis

Risk of bias in eligible trials was assessed by using the RoB2 tool [70, 71]. Most studies included in the meta-analysis were judged to be at moderate risk of bias (23 (48%)) [21, 22, 25, 28, 30, 31, 33, 35–38, 40, 41, 45, 47, 48, 50, 53, 57–59, 64 64]. Four (8%) studies included in the analyses were judged to be at a serious risk of bias [18, 24, 42, 44]. The main sources of bias were bias arising from the randomisation process, bias due to deviation from intended intervention and bias due to missing data (supplementary appendix 7).

No indication of small-study biases was observed by visual inspection of the funnel plots and Egger's test for asthma exacerbations (p=0.323), hospitalisations due to asthma exacerbation (p=0.343), change in FEV1 (p=0.350), and AQLQ score (p=0.123). However, Egger's test was suggestive of publication bias for ACQ score (p<0.001). No indication of small-study effects bias was observed for adverse events, serious adverse events, deaths and pneumonias (p=0.143, 0.553, 0.851 and 0.925, respectively) (supplementary figures S23–S31). The trim and fill sensitivity analysis indicated that the number of possible missing studies was 0 to 4, with minimal effects on the observed summary estimates (supplementary figures S32–S39). The single exception was again the meta-analysis of the ACQ score, where the trim and fill method indicated a very large number of potentially missing studies (n=36) (supplementary figure S40). These unlikely results were driven by one study with a very small confidence interval (Haldar et al. [37]). When this study was excluded, there was no longer an indication of small-study bias (p=0.421) (supplementary figures S41–S43).

The leave-one-out influence analysis identified influential studies in only four instances. Excluding Rubin et al. [57], the summary effect size for AQLQ score changed from 0.53 (95% CI 0.28–0.78) to 0.32 (95% CI 0.23–0.41) in the anti-IgE subgroup. Excluding Bernstein et al. [21], the summary effect size for percentage of OCS reduction changed from 37.84 (95% CI 13.09–62.52) to 50.00 (95% CI 36.95–63.05) in the anti-IL5/5Rα subgroup. Excluding Corren et al. [32], the summary effect size for adverse events changed from 0.66 (95% CI 0.39– 1.10) to 0.92 (95% CI 0.81–1.03) and for serious adverse events changed from 0.53 (95% CI 0.28–1.02) to 0.73 (95% CI 0.54–0.98), both in the anti-TSLP subgroup (supplementary figures S44–S58).

Discussion

This meta-analysis that included 48 RCTs of approved biologic agents for severe asthma with 16 350 participants demonstrated that their use led to a significant reduction in exacerbations and hospitalisations, along with modest improvements in lung function, asthma control and quality of life. Moreover, biologics significantly reduced the use of systemic corticosteroids, exhibiting concurrently an acceptable safety profile. We also observed significant heterogeneity of the effect estimates between subgroups, which could be driven by the more profound effect estimates of anti-IL4α and anti-TSLP in reducing exacerbations, anti-TSLP in reducing hospitalisations, anti-IL4α and anti-TSLP in improving FEV1, and anti-IL5/5Rα in reducing OCS use compared to the other agent in those analyses. The effect of the biologics was greater overall in the subgroups of participants with features of T2-high inflammation. This is, to the best of our knowledge, the largest meta-analysis on the efficacy and safety of the approved doses of biologic agents, including the newly approved tezepelumab.

Exacerbations are critical events in the natural history of asthma, increasing hospitalisations, morbidity and mortality, deteriorating lung function and quality of life, and posing a significant socioeconomic burden [72]. In our study, monoclonal antibodies reduced exacerbations overall by 44%. The reduction was more apparent in the subgroups of patients with high blood eosinophil counts (62%) and FENO (67%); however, the efficacy of the medications was also evident in patients with features of T2-low inflammation, with approximately 33% efficacy observed in patients with eosinophil counts <300 cells·μL−1 and 30% for FENO <25 ppb. The signal of exacerbation reduction is mainly driven by studies of tezepelumab and, to a lesser extent, by dupilumab. Moreover, biologic agents reduced the risk of hospitalisation for asthma exacerbation by 60%.

Lung function is important in patients with asthma, with healthcare initiatives aiming to decelerate, preserve or optimally improve lung function parameters [73]. In this meta-analysis, biologics improved FEV1 by a mean of 0.11 L. Although this does not achieve the MCID of 0.2 L in asthma [74, 75], it is comparable and potentially greater than the improvement of 0.08 L that is achieved by the addition of a long-acting muscarinic antagonist (LAMA) to the standard of care treatment of ICS/LABA [76]. The greater numerical improvements were detected by the use of anti-IL4α (0.17 L) and anti-TSLP (0.15 L), and in patients with evidence of T2-high inflammation that showed FEV1 increases of 0.18, 0.18 and 0.21 L in patients with eosinophil counts ≥300 cells·μL−1, ≥150 cells·μL−1 and FENO ≥25 ppb, respectively.

The use of biologics was also associated with improvements in asthma control and quality of life. Specifically, they reduced the ACQ score by 0.34 and augmented the AQLQ score by 0.38 points, with comparable efficacy among the different categories. The improvements in ACQ and AQLQ scores were more prominent in patients with eosinophil counts ≥300 cells·μL−1 and FENO ≥25 ppb. The mean improvements did not reach the MCID of 0.5 points for both tools; however, this was achieved for the AQLQ among patients receiving anti-IgE and for both tools among patients with FENO ≥25 ppb treated with anti-TSLP [32]. Despite the modest overall improvements in asthma control and quality of life with biologics on top of standard of care treatment, it is important to note that these improvements were much greater than those achieved with the addition of an LAMA to the standard of care treatment of ICS/LABA (−0.06 and 0.05 points for the ACQ and AQLQ, respectively) [76].

Monoclonal antibodies exhibited a significant OCS-sparing effect, increasing the probability of reduction to <5 mg prednisolone·day−1 by 74%, the probability of reduction >50% by 68% and the probability of OCS discontinuation by 63%, reducing the OCS dose by 35%. Importantly, the steroid-sparing effect was more evident in studies of anti-IL5/5Rα in patients with severe eosinophilic asthma, which confirms data from pragmatic and real-life studies of these agents [77].

Biologic agent administration for severe asthma exhibited an excellent safety profile. Not only did they not increase, but, on the contrary, they were associated with fewer reported adverse events, serious adverse events, deaths and pneumonias compared to standard of care treatment. Taking into consideration that the RCTs regarded patients with severe asthma who received medium- or high-dose ICS and the majority of whom received OCS, it is interesting that only 36 pneumonias were reported in total, 16 in the treatment arm and 20 in the control arm. These results are consistent with data reported from phase 4 [19, 44, 78] and real-world studies [77, 79–81], enhancing the widespread perception that monoclonal antibodies are safe for the treatment of severe asthma in adolescents and adults.

Our findings are consistent with the network meta-analysis by Pitre et al. [82] where omalizumab, mepolizumab, reslizumab, benralizumab, dupilumab and tezepelumab reduced exacerbations, omalizumab, mepolizumab and tezepelumab reduced hospital admissions, and dupilumab improved FEV1 compared with a placebo. Moreover, reslizumab and benralizumab produced a modest reduction in ACQ score and omalizumab, mepolizumab, benralizumab and dupilumab reduced OCS use compared with a placebo. They also evaluated patients with high (≥300 cells·μL−1) and low blood eosinophil counts (<300 cells·μL−1) [82]. Although the results are in a similar direction, there are significant differences between our analysis and that by Pitre et al. [82]. In this analysis, we have chosen to include only RCTs at the licensed doses of approved biologics, whereas Pitre et al. [82] also evaluated different doses (e.g. in early studies of mepolizumab) and agents that have not been approved (including astegolimab, fevipiprant, itepekimab and tralokinumab), thus making the current analysis more relevant for practicing clinicians. We additionally chose not to perform a network meta-analysis, as all these studies compared active treatment with placebo (usually on top of standard care), due to the potential bias of the magnitude of the placebo effect [83]. Finally, we have performed analyses based on different cut-points of T2 biomarkers (blood eosinophils and FENO) in order to evaluate the efficacy of biologics in clinically relevant subgroups. Therefore, we believe that our analysis is likely more useful to practicing clinicians.

Our study has the limitation of including RCTs with heterogeneous populations regarding disease severity and T2 inflammation. However, we assessed heterogeneity in the different biologic categories and in the different subgroups of participants, simultaneously performing influence analysis in order to detect studies that could affect the overall size effect.

In conclusion, in this meta-analysis that included all the RCTs on all approved biologic agents for severe asthma we demonstrated that their administration reduced exacerbations and hospitalisations, improved lung function, asthma control and quality of life, and reduced the use of systemic corticosteroids, with a favourable safety profile. These effects were more prominent in patients with evidence of T2 inflammation, with some evidence for the efficacy of tezepelumab also in patients with T2-low asthma.

Points for clinical practice

This meta-analysis, which included 48 RCTs regarding all the approved biologic agents for severe asthma, with 16 350 participants in total, demonstrated that their administration reduced exacerbations and hospitalisations, improved lung function, asthma control and quality of life, and reduced the use of systemic corticosteroids, with a favourable safety profile.

Anti-IL4α and anti-TSLP were the most effective agents in reducing exacerbations, anti-TSLP in reducing hospitalisations, anti-IL4α and anti-TSLP in improving FEV1, anti-IgE in reducing ACQ score and improving AQLQ score, and anti-IL5/5Rα in reducing OCS use and discontinuation.

The effect of the biologics was greater overall in the subgroups of participants with features of T2-high inflammation.

Questions for future research

Biologic agents have been approved and included in GINA guidelines in step 5 for the treatment of asthma. Our study contributes to the growing body of evidence on the efficacy and safety of biologic agents in the different phenotypes of severe, “uncontrolled” asthma and their utility in the effort to “control” the disease. The currently available evidence does not warrant a change in policies; however, there is an unmet need for the development of biologic agents or other medications for the optimal treatment of patients with features of T2-low inflammation.

Supplementary materialSupplementary Material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material ERR-0238-2023.SUPPLEMENT

Acknowledgements

We thank George Ntritsos (Department of Hygiene and Epidemiology, University of Ioannina Faculty of Medicine, Ioannina, Greece) who helped in the literature search and the data extraction.

Footnotes

Provenance: Submitted article, peer reviewed.

Data sharing: Extracted data are available from the corresponding author on request.

Author contributions: C. Kyriakopoulos and K. Kostikas developed the protocol. G. Markozannes designed and C. Kyriakopoulos and A. Gogali ran the literature search. C. Kyriakopoulos, A. Gogali, G. Markozannes and K. Kostikas screened records, extracted data, and assessed risk of bias. C. Kyriakopoulos and A. Gogali independently extracted the estimates used in meta-analyses (the underlying data), and all authors had access to the extracted data and the scripts used for statistical analyses. G. Markozannes did statistical analyses with input from C. Kyriakopou