Neutrophil and platelet increases with luspatercept in lower-risk MDS: secondary endpoints from the MEDALIST trial

Conflict of interest: COI declared see note COI notes: G.G-M.: Acceleron Pharma, Astex Pharmaceuticals, Bristol Myers Squibb, Helsinn Therapeutics, and Jazz Pharmaceuticals consulting or advisory role; AbbVie, Acceleron Pharma, Astex Pharmaceuticals, Bristol Myers Squibb, and Helsinn Therapeuticshonoraria; AbbVie, Amphivena Therapeutics, Astex Pharmaceuticals, Bristol Myers Squibb, H3 Biomedicine, Helsinn Therapeutics, Merck, Novartis, and Onconova Therapeutics research funding. G.J.M.: AbbVie and Novartis consulting or advisory role; Bristol Myers Squibb and Novartisresearch funding. P.F.: AbbVie, Bristol Myers Squibb, Janssen, and Jazz Pharmaceuticals honoraria, consulting or advisory role; Jazz Pharmaceuticals accommodations, expenses, travel. R.B.: Bristol Myers Squibb honoraria, consulting or advisory role, research funding; TAIHO honoraria, consulting of advisory role, research funding; Takeda research funding. V.S.: Bristol Myers Squibb, and Novartis honoraria; Astex, Bristol Myers Squibb, Geron, Gilead, Menarini, and Novartis consulting or advisory role. M.D.-C.: Bristol Myers Squibb and Novartis consulting or advisory role, honoraria, membership on an entity's board of directors or advisory committee, research funding. C.F.: Bristol Myers Squibb, Janssen, Novartis, and Takeda consulting or advisory role, speakers' bureau; Bristol Myers Squibb research funding. M.A.S.: Bristol Myers Squibb, Millennium, and Syros Pharmaceuticals consulting or advisory role; Pfizer and Takedaresearch funding. A.M.Z.: Abbvie, Acceleron Pharma, Agios, Astellas, Beyond Spring, Boehringer-Ingelheim, Bristol Myers Squibb, Cardiff Oncology, Cardinal Health, Daiichi Sankyo, Epizyme, Incyte, Ionis, Jazz Pharmaceuticals, Novartis, Pfizer, Otsuka, Seattle Genetics, Taiho, Takeda, and Trovagene consulting or advisory role, honoraria; Abbvie, ADC Therapeutics, Aprea, Astex, Boehringer-Ingelheim, Bristol Myers Squibb, Incyte, MedImmune/AstraZeneca, Novartis, Pfizer, Takeda, and Trovagene; Cardiff Oncology, CCITLA, and Leukemia and Lymphoma Society other. R.I.: Bristol Myers Squibb ended employment in the past 24 months, stock and other ownership interests; Eli Lilly and Company employment, stock and other ownership interests. J.Z.: Bristol Myers Squibb employment, stock and other ownership interests. D.S.: Bristol Myers Squibb ended employment in the past 24 months, stock and other ownership interests. A.R.: Bristol Myers Squibb employment, travel, accommodations, expenses, and stock and other ownership interests. J.T.B.: Acceleron Pharma employment, stock and other ownership interests; Bristol Myers Squibb stock and other ownership interests. U.P.: AbbVie, Bristol Myers Squibb, and Novartis consultancy, honoraria. R.S.K.: Agios, Bristol Myers Squibb, Daiichi Sankyo, Incyte, Janssen, Novartis, and Pfizer consulting or advisory role; Alexion Pharmaceuticals, Jazz Pharmaceuticals, and Novartis speakers bureau, AbbVie stock and other ownership interests. O.I.: no conflicts of interest to disclose. Preprint server: No; Author contributions and disclosures: G.G-M. V.S., U.P., R.S.K., designed the study. G.G-M., G.J.M., P.F., R.B., V.S., M.D-C., C.F., O.I., M.A.S., A.M.Z., U.P., R.S.K., collected data. G.G-M., G.J.M., P.F., R.B., V.S., M.D-C., C.F., O.I., M.A.S., A.M.Z., U.P., R.S.K., R.I., J.Z., A.R., D.S., J.T.B., analysed and interpreted the data. R.I., A.R., supervised the clinical study. J.Z., performed statistical analysis. D o w n l o a d e d

Myelodysplastic syndromes (MDS) result in abnormal blood cell development, cytopenias, and risk of progression to acute myeloid leukemia (AML). 1 Most patients with lower-risk MDS (LR-MDS) have anemia, but patients can also have neutropenia and/ or thrombocytopenia with significant clinical implications. [2][3][4] Treatments for anemia include red blood cell (RBC) transfusions, erythropoiesis-stimulating agents (ESAs), hypomethylating agents (HMAs), 5 or lenalidomide. 5 However, RBC transfusions can result in iron overload 6,7 ; patients can become resistant to ESAs, 4,8 and HMAs and lenalidomide have been associated with grade 3 or 4 neutropenia and thrombocytopenia. 9,10 Luspatercept is a first-in-class erythroid maturation agent that binds several transforming growth factor-b (TGF-b) superfamily ligands to diminish Smad2/3 signaling and enhance late-stage erythropoiesis. 11 Its efficacy and safety were demonstrated in the phase 3, placebo-controlled MEDALIST trial in RBC transfusiondependent patients with LR-MDS with ring sideroblasts (RS). 12 In this study, significantly more luspatercept-treated patients achieved RBC transfusion independence for $8 weeks during weeks 1 to 24 (37.9% vs 13.2%; P , .001). 12 Significantly more patients in the luspatercept arm achieved hematologic improvement-erythroid (HI-E), as per 2006 International Working Group (IWG) criteria, 13 during weeks 1 to 24 (52.9% vs 11.8%; P , .001) and weeks 1 to 48 (58.8% vs 17.1%; P , .001). 12 Here, we report the effect of luspatercept on lineages outside the erythroid compartment, including platelets and neutrophils, and the HI for these lineages in MEDALIST patients cytopenic at baseline.
Full details of the MEDALIST trial (NCT02631070) have been published. 12 14 ) with RS (either $15% or $5% if SF3B1 mutation was present), who were refractory to, intolerant of, or unlikely to respond to ESAs (serum erythropoietin .200 U/L) and required RBC transfusions, were randomized 2:1 to receive luspatercept (n 5 153) or placebo (n 5 76) subcutaneously every 3 weeks for 24 weeks.
Of the 25 patients evaluable for HI-N, more of those randomized to luspatercept vs placebo achieved HI-N during weeks 1 to 24 (13.3% vs 0.0%) and weeks 1 to 48 (20.0% vs 10.0%). Similarly, of the 14 patients evaluable for HI-P, more luspatercept-vs placebo-treated patients achieved HI-P during weeks 1 to 24 (50.0% vs 33.3%) and weeks 1 to 48 (62.5% vs 33.3%). These findings potentially support the use of luspatercept to treat patients with LR-MDS with RS who are often neutropenic and/or thrombocytopenic and anemic. However, the HI-N and HI-P responses in the placebo arm might highlight the normal oscillations seen in blood counts of patients with LR-MDS. Coupled with the low numbers of patients evaluable for HI-N and HI-P, these results should be interpreted with caution.
Treatment-emergent grade 3 or 4 neutropenia was infrequently reported, with lower incidence in the luspatercept vs the placebo group (7/153 [4.6%] vs 6/76 [7.9%]) and may have represented normal fluctuations in patients' blood counts. No grade 3 or 4 treatment-emergent thrombocytopenia was reported in either treatment arm. These rates of grade 3 or 4 cytopenias are much lower than those observed with other therapies for MDS, including decitabine, 15 azacytidine, 16 and lenalidomide, which in a phase 3, randomized, placebo-controlled trial in patients with lower-risk non-del(5q) MDS showed high rates of grade 3 or 4 neutropenia (61.9% vs 12.7%) and thrombocytopenia (35.6% vs 3.8%). 10 Despite the increase in neutrophil counts, there was a slight increase in infection rate with luspatercept compared with placebo. Infection was reported in 4 of 9 (44.4%) and 3 of 7 (42.9%) luspatercept-and placebo-treated patients, respectively, who experienced neutropenia (any grade) during the study. Overall infection rates for luspatercept and placebo patients were 53.6% and 40.8%, respectively. The infections were not opportunistic and were mostly grade 1 to 2 in severity. The differences in infection rates were not assessed, as this study was not designed or powered for this purpose. Bleeding was not reported in any luspatercept-or placebo-treated patients who experienced thrombocytopenia (any grade) on study. Among patients who achieved HI-N or HI-P, 1 patient in the luspatercept arm progressed to higher-risk MDS, but none progressed to AML.
Although only a minority of patients were evaluable for HI-P/ HI-N, luspatercept treatment resulted in a mean increase from baseline in platelet and neutrophil counts in most patients overall vs placebo. Mean neutrophil and platelet count increases were observed early on luspatercept treatment and persisted to week 25. This could be associated with the positive effect of luspatercept on hematopoietic stem and progenitor cell expansion by modulating the structure of extracellular matrix 17 or by direct inhibition of transforming growth factor-b signaling. 18 In the 25 patients with baseline neutropenia and 14 patients with baseline thrombocytopenia, higher proportions of patients in the luspatercept vs placebo arms achieved HI-N and HI-P during weeks 1 to 24 and weeks 1 to 48. As meaningful statistical analyses were not possible because of small sample sizes, these results should be treated with caution.

R E F E R E N C E S
Recent studies have revealed the presence of clonally expanded cells with somatically acquired cancer-associated mutations within normal human tissues, 1-5 including in blood from healthy elderly individuals, where these identify individuals with clonal hematopoiesis (CH). [3][4][5] Among the most prevalent CH mutations are those seen in DNMT3A, TET2, ASXL1, and TP53, 3-5 implicated as initiating mutations in myeloid malignancies. 6,7 CH confers increased risk for later development of myeloid malignancies. [3][4][5] However, most CH cases never develop any malignancy, and mechanisms enhancing transformation risk and clonal advantage of CH mutations remain unclear. 5 Unraveling these mechanistic aspects of CH could greatly benefit from studies in genetically modified mice. Such studies have already provided some insights, but with conflicting results. [8][9][10] Because the relevance of mice for modeling of CH mutations, myeloid malignancies, and cancer in general has been questioned, 5,[9][10][11] it would be important to establish to what degree mutations seen in human CH also occur spontaneously and promote clonal expansion in normal-aged mice. CH mutations have yet to be described in mice screened for spontaneous oncogeneic mutations, 12 potentially because of the few mice investigated and sequencing strategies with insufficient sensitivity to detect small clones 12 as human CH mutations, often occur early in life, but are often first detected in aged individuals (.70 years of age) when the clones have become large enough for detection with existing methodology. 5,13 The much lower number of mouse hematopoietic stem cells (HSCs) 14,15 and their shorter lifespan (2 to 3 years) suggest that CH mutations would occur at a much lower rate in mice and prove more difficult to detect than in human. However, the much smaller size of the mouse and fewer HSCs could potentially enable detection of CH clones in aged mice. We screened (supplemental methods, available on the Blood Web site) for the most common CH mutations in up to 24-month-old wild-type C57BL/6j mice, the most extensively used mouse strain for studies of normal and malignant hematopoiesis, including genetically modified mice with CH mutations. 8-10 DNA isolated from single aged human (70 to 75 years; n 5 6) or mouse (24 months; n 5 6) HSC-derived cells was subjected to whole-genome sequencing, and bulk bone marrow (BM) of aged (n 5 97) and transplanted (n 5 88) mice to digital droplet polymerase chain reaction (ddPCR) analysis and errorcorrected targeted DNA sequencing (ECTS). Similar to cultured fibroblasts, 16 but not previously investigated for HSCs, we observed a significantly higher mutation rate (8.5-fold) in mouse compared with human HSCs ( Figure 1A; supplemental Figure 1A). In line with previous reports, 14,17 aged human HSCs contained 1000 mutations ( Figure 1B). Although a 2-year old laboratory mouse approximates a 70-year-old human individual in relative lifespan, the increased mutation rate in mouse HSCs did not result in comparable mutational accumulation, as aged mouse HSC mutations were fivefold lower than in aged human HSCs ( Figure 1B). Despite their differences in lifespan, mutations in aged mouse HSCs were distributed among similar genomic regions, dominated by the agingassociated COSMIC signature 1 featured by enrichment of C.T transitions at CpG dinucleotides 18  Together with the estimated HSC pool size in mice vs humans, 14,15 this suggests that although mutations targeted to CH-associated genes would be much more frequent in aged