Comparative effectiveness of biologic therapies for preventing asthma exacerbations in severe asthma: a systematic review and network meta-analysis

The comparative effectiveness of biologic therapies for preventing exacerbations in severe asthma remains uncertain because no eligible direct head-to-head randomized comparisons were identified. We compared six biologic therapies using a network meta-analysis (NMA). This systematic review followed PRISMA 2020 and PRISMA-NMA guidance and a prospectively registered PROSPERO protocol (CRD420261428438). PubMed/MEDLINE, Embase, Scopus, Web of Science, and CENTRAL were searched from inception through 30 June 2026. Randomized controlled trials evaluating omalizumab, mepolizumab, reslizumab, benralizumab, dupilumab, or tezepelumab in severe or persistently uncontrolled asthma were included. The primary outcome was annualized asthma exacerbation rate (AAER). Risk of bias was assessed using the Cochrane Risk of Bias 2 tool. A frequentist random-effects NMA estimated rate ratios (RRs) with 95% confidence intervals (CIs). Treatments were ranked using P-scores, and confidence was assessed using Confidence in Network Meta-Analysis (CINeMA). Thirty-two trials involving 17,384 participants were included; 19 studies contributing 20 independent contrasts and 10,892 unique participants entered the primary NMA. All six biologics significantly reduced AAER versus placebo. Tezepelumab had the most favourable estimate and highest P-score (RR 0.34, 95% CI 0.24–0.49; P-score 0.934). RRs were 0.46 (95% CI 0.27–0.79) for reslizumab, 0.47 (0.35–0.62) for benralizumab, 0.47 (0.35–0.64) for dupilumab, 0.49 (0.38–0.64) for mepolizumab, and 0.68 (0.50–0.93) for omalizumab. The entirely indirect comparison favoured tezepelumab over omalizumab (RR 0.50, 95% CI 0.31–0.80); no other active-treatment comparison was statistically significant. Heterogeneity was substantial (τ² = 0.062; I² = 70.9%), and exploratory analyses indicated possible small-study effects. Confidence in biologic-versus-placebo estimates was moderate except for omalizumab, for which it was low; confidence in most biologic-versus-biologic estimates was low or very low. All six biologics reduced AAER versus placebo, but entirely indirect active-treatment evidence and substantial heterogeneity precluded a universal treatment hierarchy. Treatment selection should remain individualized according to eligibility, inflammatory phenotype, biomarkers, comorbidities, treatment burden, patient preferences, access, and cost.