This phase 2, multicenter, single-arm trial was conducted at 16 clinical sites in Japan. As shown in supplemental Fig. 1, the trial comprised a screening period, treatment period, and post-treatment follow-up period, including 42-day safety and long-term follow-up.

Patient disposition. OS, overall survival

The study was conducted in accordance with the Good Clinical Practice Guideline E6 from the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use, the general ethical principles outlined in the Declaration of Helsinki, and applicable national, prefectural, and local laws. All patients provided written informed consent. The study protocol was approved by the institutional review board or independent ethics committee at each clinical site. The study was registered at www.ClinicalTrials.gov on 3 April 2019 (NCT03900715).

Eligible patients were aged ≥ 20 years and had MDS diagnosed according to the World Health Organization (WHO) 2016 criteria [23], with < 5% bone marrow blasts. They had LR-MDS, defined according to the revised International Prognostic Scoring System (IPSS-R) as disease of very low, low, or intermediate risk [24]. Patients had symptomatic anemia with a mean hemoglobin level < 10.0 g/dL (from two measurements, one performed 7–35 days before the first dose of luspatercept and one within 1 day prior to the first dose of luspatercept), and no requirement for RBC transfusions (no transfusions ≤ 16 weeks before the first dose of luspatercept). Previous ESA use, administered ≥ 8 weeks before the first dose of luspatercept, was permitted. Patients were required to have an Eastern Cooperative Oncology Group performance status ≤ 2.

Patients were not eligible to participate in the study if they had previously received disease-modifying agents for MDS (e.g., lenalidomide, hypomethylating agents), luspatercept, or an allogeneic or autologous hematopoietic cell transplant. Other exclusion criteria included a diagnosis of myelodysplastic/myeloproliferative neoplasms according to the WHO 2016 classification (i.e., chronic myelomonocytic leukemia, atypical chronic myeloid leukemia, BCR-ABL1 negative, juvenile myelomonocytic leukemia) [23].

Luspatercept was administered to patients as a subcutaneous injection once every 3 weeks for ≥ 24 weeks unless the patient met any of the treatment discontinuation criteria. The luspatercept starting dose was 1.0 mg/kg. To maintain the hemoglobin level within a target range of 10–12 g/dL (6.2–7.5 mmol/L) without RBC transfusions, starting from week 7 (i.e., dose 3), the luspatercept dose could be increased in a stepwise manner, first to 1.33 mg/kg, up to a maximum of 1.75 mg/kg.

In the event of a grade ≥ 3 treatment-related treatment-emergent adverse event (TEAE), luspatercept dosing was delayed until the TEAE resolved to grade ≤ 1, at which point a reduced dose was administered. A reduced dose was also required if, compared with the pre-dose level of the previous cycle, the hemoglobin level increased by > 2.0 g/dL. If a patient’s pre-dose hemoglobin level was ≥ 12 g/dL, a dose delay was required until the hemoglobin level was < 11.0 g/dL.

All patients underwent an MDS disease assessment on day 169 and every 24 weeks thereafter. Patients could continue to receive luspatercept after day 169 if they had evidence of clinical benefit, which was noted by either remaining NTD or having an International Working Group (IWG)-defined hematological improvement-erythroid (HI-E) response (≥ 1.5 g/dL increase in hemoglobin level for ≥ 8 weeks) within the 12 weeks before day 169 and every 24 weeks thereafter, and if they did not have disease progression according to IWG criteria [25]. Patients who did not have evidence of clinical benefit or had disease progression discontinued luspatercept and entered post-treatment follow-up. Patients could receive best supportive care, including RBC transfusions, antibiotics, antivirals, and antifungals, as needed.

In the first 10 patients to be enrolled in the study, blood samples for pharmacokinetic analyses were collected on days 1 and 3 of weeks 1 and 2; day 1 of weeks 3, 4, 10, 13, 16, 17, 18, 19, and 22; at the 24-week MDS disease assessment visit; and every 12 weeks thereafter for up to 1 year from the first dose (dense sample collection schedule). For the subsequent 10 patients, blood samples were collected on day 1 of weeks 1, 2, 3, 4, 10, 16, and 22; at the 24-week MDS disease assessment visit; and every 12 weeks thereafter for up to 1 year from the first dose (sparse sample collection schedule). Serum concentrations of luspatercept were analyzed using a validated enzyme-linked immunosorbent assay.

All patients who discontinued treatment with luspatercept entered long-term follow-up, during which information was collected on subsequent MDS therapies, progression to AML and other malignancies/pre-malignancies, and overall survival. Patients were followed for 5 years from the date of the first dose of luspatercept or for 3 years from the last dose, whichever occurred later, or until they died, were lost to follow-up, or withdrew consent.

The primary endpoint of the study was the proportion of patients who achieved HI-E response per IWG criteria (≥ 1.5 g/dL increase in hemoglobin level for a consecutive 8-week period [25]) in the absence of RBC transfusions within the first 24 weeks of luspatercept treatment. In the event of dose delays because of a pre-dose hemoglobin level ≥ 12 g/dL, the time from dose delay to achieving an increase in hemoglobin of ≥ 1.5 g/dL after treatment resumption was included in the period of sustained hemoglobin increase even if there was a temporary decrease in hemoglobin to < 1.5 g/dL.

Secondary endpoints included the proportion of patients who achieved HI-E response within 48 weeks in the absence of RBC transfusions; the proportion of patients who achieved modified HI-E (mHI-E) response (≥ 1.5 g/dL mean increase in hemoglobin level for ≥ 8 weeks) in the absence of RBC transfusions within the first 24 and 48 weeks of treatment; time to and duration of HI-E and mHI-E response; the proportion of patients who remained NTD (no RBC transfusions during the treatment period) at weeks 24, 48, and 72; and the safety and pharmacokinetic profiles of luspatercept.

Safety was evaluated by the incidence of TEAEs until 42 days after the last dose and their causality to luspatercept (treatment-related TEAEs), as well as by the frequency of TEAEs leading to death, or to discontinuation or dose modification of luspatercept. Selected TEAEs of special interest included asthenia, hypertension, pre-malignant disorders, renal toxicity, immunogenicity type reactions, thromboembolic events, malignancies, extramedullary hematopoiesis masses, and liver toxicity. TEAEs of special interest were defined on the basis of non-clinical findings or the known safety profile [20, 22, 26,27,28,29] of the study drug. TEAEs were encoded using the Medical Dictionary for Regulatory Activities version 25.0 and presented as the worst grades based on the Common Terminology Criteria for Adverse Events version 4.03.

For the primary efficacy analysis, the null hypothesis was that the proportion of patients who would achieve HI-E response by week 24 would be ≤ 10%. This proportion was based on the results of ARCADE, a phase 3 randomized placebo-controlled trial conducted in European countries, in which the HI-E response rate for patients with IPSS-defined low- or intermediate-1 risk MDS treated with darbepoetin-α was 14.7% (95% confidence interval [CI], 7.6–24.7) by week 24 [13]. In the present trial, 45% of patients were expected to achieve HI-E response based on the results of the phase 2 PACE-MDS dose-finding trial, in which the preliminary HI-E response rate was 63% (95% CI, 48–76) for patients with low- or intermediate-1 risk MDS treated with luspatercept 0.75–1.75 mg/kg [29]. With an expected HI-E response rate of 45%, a sample size of 20 patients would provide 95% power with a one-sided significance α of 0.025, based on a one-sample binomial exact test for the null hypothesis, which would be rejected if the lower limit of the 95% CI for the HI-E response rate was > 10%.

Exploratory analyses were performed for the primary endpoint across predefined patient subgroups: age (≤ 64 years, 65–74 years, ≥ 75 years); sex; RS status; baseline serum erythropoietin (≤ 200 U/L, > 200 U/L, ≤ 500 U/L, > 500 U/L); prior ESAs treatment; time since original MDS diagnosis (≤ 2 years, > 2 to ≤ 5 years, > 5 years); IPSS-R classification; splicing factor 3B subunit 1 (SF3B1) mutation status; and platelet count (< 100 × 109/L, 100–450 × 109/L, > 450 × 109/L).

Secondary endpoints were summarized descriptively. Durations of HI-E and mHI-E response were estimated using the Kaplan–Meier method. Pharmacokinetic parameters within the first treatment cycle were determined using non-compartmental analyses of serum luspatercept concentrations from samples collected according to the dense schedule after the first dose. In addition, pharmacokinetic analyses were based on a one-compartment model with first-order absorption and elimination and were conducted using all available samples.

All efficacy analyses were performed on the intention-to-treat (ITT) population, defined as all enrolled patients. The pharmacokinetic population included all patients who received at least one dose of luspatercept and had sufficient samples collected and assayed for analysis. The safety population included all patients who received at least one dose of luspatercept.

Safety and efficacy analyses were conducted using SAS software, version 9.4 (SAS Institute, Cary, North Carolina, USA). Phoenix WinNonlin software, version 8.2 (Certara, Princeton, New Jersey, USA) was used for pharmacokinetic analyses.