Background There is a paucity of effective treatments for myelodysplastic neoplasms (MDS) patients who are relapsed or refractory. Luspatercept shows effective results in lower-risk MDS (LR-MDS) patients in a phase 3 study. However, the real-world data of luspatercept in LR-MDS are limited, particularly in large cohorts with long-term follow-up in China. Methods Data from patients diagnosed with LR-MDS at Peking Union Medical College Hospital (PUMCH) in China between June 2022 and March 2024 were retrospectively collected. All the enrolled patients were refractory/relapsed to previous treatment. Patients had been treated with luspatercept (at a dose of 1.0 and up to 1.75 mg per kilogram of body weight) for at least 3 courses. Erythroid response (HI-E), rate of red blood cell transfusion independence (RBC-TI), changes in other parameters, adverse events, clonal evolution, long-term relapse and survival, and factors that could affect efficacy were analyzed. Results Sixty LR-MDS patients were enrolled in the study, with a median age of 67 years (range: 46-85), and 63.4% were males. The median follow-up time was 12 months (range: 3-24), and the median duration of luspatercept was 7 months (range: 2-23). HI-E occurred in 28/60 (46.7%), 25/49 (51.0%), 17/35 (48.6%), and 26/60 (43.3%) patients at the 3rd, 6th, 12th months, and at the end of follow-up, respectively. Among patients who were transfusion-dependent prior to luspatercept, 15/31 (48.3%), 12/31 (38.7%), and 8/31 patients (25.8%) exhibited RBC-TI for 8, 12, and 16 weeks or longer at the 6th month. 13/22 (59.1%), 10/22 (45.5%), and 9/22 patients (40.9%) achieved RBC-TI for 8, 12, and 16 weeks or longer at the 12th month, respectively. Over time, patients treated with luspatercept had a significant increase in hemoglobin level compared with that of the baseline at the 1st 2nd, 3rd, 6th, and 12th month, and at the end of follow-up (P < 0.001, < 0.001, < 0.001, < 0.001, =0.026, and 0.002, respectively). Besides, patients treated with luspatercept had lower serum ferritin level at the end of follow-up compared to baseline (P = 0.039). Five of thirty-two patients (15.6%) who had response relapsed at the 8th, 9th, 10th, 12th and 17th months, respectively. Three patients (5.0%) transformed to acute myeloid leukemia (AML) at the 11th, 17th, and 17th months, and 3 patients (5.0%) died of pulmonary infection after 3, 5 and 6 months of treatment.In univariate regression, non-transfusion dependent (NTD) and low transfusion burden (LTB) before luspatercept (P = 0.025) and serum erythropoietin (EPO) < 500 IU/L (P = 0.021) were associated with a higher HI-E rate. Whereas in a subsequent multivariate regression analysis, serum EPO < 500 IU/L was the only independent predictor for higher HI-E rate at the 3rd month (P = 0.018). It is noteworthy that relapsed or refractory to prior therapy and IPSS-M higher-risk was not a predictive factor of luspatercept treatment, either in univariate or multivariate analysis. Conclusion Luspatercept was effective for refractory LR-MDS patients in the long-term study, with mild side effects. Serum EPO < 500 IU/L was the only independent predictor for higher HI-E rate. Keywords: lower-risk myelodysplastic neoplasms, luspatercept, refractory, effect, follow-up