Luspatercept response predictors in lower-risk myelodysplastic syndrome

Abstract Background: Luspatercept is a recombinant fusion protein that targets TGF-β superfamily ligands to enhance late-stage erythropoiesis. It has shown superior efficacy over erythropoietin stimulating agents (ESAs) and placebo in phase 2 and 3 trials. It is FDA-approved for the treatment of anemia in patients (pts) with lower-risk myelodysplastic syndrome (LR-MDS), particularly in pts with ring sideroblasts (RS) or SF3B1 mutation. However, predictors of response beyond RS/SF3B1 remain unclear. Methods: We retrospectively analyzed adult pts with LR-MDS treated with luspatercept between 2016–2025 at The University of Texas MD Anderson Cancer Center. Pts were classified into 3 transfusion burden (TB) groups per IWG 2018 criteria: non–transfusion dependent (NTD; 0 RBC units in 16 weeks), low TB (LTB; 3–7 RBC units in 16 weeks), and high TB (HTB; ≥8 RBC units in 16 weeks or ≥4 in 8 weeks). Response was defined as sustained hemoglobin (Hb) increase (≥1.5 g/dL) in NTD, transfusion independence (TI) (≥8 weeks) in LTB, and either TI or ≥50% transfusion reduction in HTB, lasting ≥16 weeks. We collected baseline clinical characteristics, prior therapies, cytogenetics, and mutational data. Logistic regression identified predictors of response; overall survival (OS) was assessed using Kaplan-Meier and Cox proportional hazards models. Results: Among 104 response-evaluable pts, median age was 71.2 (range, 34–90) yrs, and 63% were male. Median baseline Hb was 8.5 g/dL (7.8–9.2), platelets (PLT) 193 × 10⁹/L (97–293), ANC 2.3 × 10³/µL (1.4–3.3), and median IPSS-R score 2.5 (2.0–3.5). Common mutations were SF3B1 (39%), TET2 (25%), ASXL1 (17%), DNMT3A (12%), U2AF1 (8%), and SRSF2 (6%). Response defined by the IWG 2018 criteria was observed in 52 pts (50%). Among these, 22.1% achieved a Hb increase of ≥1.5 g/dL in the NTD category, 25.9% attained TI in the LTB category, and 1.9% exhibited a >50% reduction in the HTB category. Responders had significantly higher baseline Hb (8.7 vs 8.2 g/dL, P = 0.04) and PLT (237 vs 135 ×10⁹/L, P = 0.03) with a trend towards lower median RS% (30% vs 51%, P = 0.06) and EPO level (52 vs 83 mIU/mL, P = 0.15). There was no difference in baseline ANC, marrow blasts, fibrosis, IPSS-R/IPSS-M scores. Cytogenetics, including del5q, -7/del(7q), +8, -20/del(20q), and complex karyotype, were not associated with response. Among 84 pts with NGS data, no mutation was significantly predictive. Trends were noted for U2AF1 (OR 4.23, 95% CI 0.96–29.41; P = 0.08), TET2 (OR 2.07, 95% CI 0.85–5.28; P = 0.12), and SF3B1 (OR 1.70, 95% CI 0.77–3.81; P = 0.19). Response rates differed significantly by TB category (NTD, LTB, HTB; P < 0.01). Prior hypomethylating agents (HMAs) exposure was significantly associated with lower response rates (25% vs 75% in HMA-naïve, P = 0.01). There was no significant association with prior ESA or lenalidomide use. First-line luspatercept showed a trend towards better response (p=0.09). In multivariate analysis, adjusting for key clinical covariates, including age, IPSS-R, baseline Hb, PLT, and line of therapy, HTB (OR 0.08; 95% CI, 0.008–0.59; P = 0.02) and higher RS% (OR 0.98; 95% CI, 0.96–0.99; P = 0.04) independently predicted reduced odds of response. In a subgroup analysis of SF3B1 wild-type pts (N = 37), RS% (OR 0.97; 95% CI 0.93–0.99; P = 0.04), and earlier line of therapy was significantly associated with better response (OR 0.15; 95% CI 0.02–0.63; P = 0.02). Prior HMA exposure trended toward reduced response (OR 0.19; 95% CI 0.03–0.92; P = 0.06), while higher Hb showed a positive trend (OR 2.07; 95% CI 1.03–4.76; P = 0.06). Other variables, including IPSS-R, EPO levels, and co-mutations, were not predictive. In the SF3B1-mutated subgroup (N = 48), TI at baseline was the only significant predictor of response to luspatercept (OR 7.20, 95% CI: 1.20–62.51; P = 0.04). Responders had longer OS (median not reached vs 89 weeks; P < 0.01). One- and two-year OS rates were 85% and 82% in responders versus 74% and 34% in non-responders, respectively. In univariate analysis, attaining response was associated with reduced risk of death (HR 0.44, 95% CI 0.19–1.01; P = 0.05). Conclusions: Patients who respond to luspatercept tend to have lower baseline transfusion burden, lower RS%, and no prior HMA therapy. Attainment of response was associated with significantly improved overall survival. Prospective studies are needed to validate these predictors and inform optimal sequencing of luspatercept in LR-MDS.