Luspatercept in β-thalassemia: Meta-analysis of efficacy and safety across transfusion-dependent and non-transfusion-dependent populations

Abstract

Abstract Background: Luspatercept is a first-in-class erythroid maturation agent approved for adult transfusion-dependent β-thalassemia. Phase III trials demonstrated significant reductions in transfusion requirements via TGF-β superfamily ligand binding and SMAD2/3 attenuation. We conducted a meta-analysis to quantify efficacy and safety across transfusion-dependent (TDT) and non-transfusion-dependent (NTDT) populations. Methods: PubMed, Embase, and Google Scholar were searched through July 2025 for prospective luspatercept trials in β-thalassemia, excluding duplicate extensions. Four studies (n=393) met inclusion: two randomized trials (BELIEVE in TDT and BEYOND in NTDT) and two phase II single-arm trials. Pooled proportions for ≥33% and ≥50% transfusion reduction (TDT) and ≥1.0 and ≥1.5 g/dL hemoglobin increase (NTDT), along with safety data on any grade ≥3 treatment-emergent adverse events and 95% confidence intervals, were estimated using a random-effects model (DerSimonian-Laird). Heterogeneity was assessed by I², publication bias by funnel plot symmetry, and sensitivity analyses excluded single-arm trials. Risk of bias (Cochrane) and evidence quality (GRADE) were evaluated. Results: In transfusion-dependent β-thalassemia (TDT), the pooled proportion achieving ≥33% reduction in transfusion burden was 74% (95% CI 65–82%; I²=48%), and 27% (95% CI 20–35%; I²=35%) achieved ≥50% reduction. These results align with the BELIEVE trial, in which 70.5% vs 29.5% of patients achieved ≥33% reduction with luspatercept vs placebo. In BELIEVE, the odds of ≥33% reduction with luspatercept were 5.8 (95% CI 4.1-7.8), nearly sixfold that with placebo. In non–transfusion-dependent β-thalassemia (NTDT), pooled response rates were 78% (95% CI 69–85%; I²=22%) for achieving ≥1.0 g/dL hemoglobin increase and 54% (95% CI 44–64%; I²=30%) for achieving ≥1.5 g/dL increase. This is consistent with BEYOND trial findings: 77.1% of luspatercept-treated NTDT patients achieved a ≥1.0 g/dL Hb increase (vs. 0% with a placebo), and >50% achieved a ≥1.5 g/dL increase. Overall, any TEAE occurred in 98% of patients (95% CI 96–99%; I²=12%), and grade ≥3 AEs occurred in 24% (95% CI 18–31%; I²=18%). Luspatercept was generally well tolerated, with predominantly musculoskeletal pain, headache, and arthralgia. No new safety signals were identified. Conclusions: Luspatercept demonstrates substantial and consistent benefit in β-thalassemia: in TDT it markedly reduces transfusion burden, and in NTDT it significantly increases hemoglobin, with an acceptable safety profile. Despite limited trial numbers and heterogeneity, these findings support luspatercept as a transformative therapy in β-thalassemia, though evidence certainty is limited by the available data. The key limitations included a small number of studies, which introduced a single data theory' and limited generalizability, as well as heterogeneity in study design, endpoints, and follow-up durations. Only BELIEVE was placebo-controlled; the others were single-arm, and several endpoints lacked uniform reporting. Risk-of-bias was moderate overall, and the GRADE assessment noted imprecision and inconsistency affecting evidence certainty. Further large trials with standardized endpoints are warranted. Keywords: β-thalassemia; luspatercept; meta-analysis; transfusion reduction; hemoglobin increase