Immunomodulatory cytokines and clonal dynamics in low-risk myelodysplastic syndromes patients treated with luspatercept.

Myelodysplastic neoplasms (MDS) with ring sideroblasts represent about 30% of all MDS patients and are characterized by severe anemia, transfusion dependence, and limited response to erythropoiesis stimulating agents (ESAs). Most patients harbor a somatic mutation of Splicing Factor 3B1 (SF3B1),1 whose presence identifies a distinct entity in the fifth edition of the WHO classification of myeloid neoplasms (MDS-SF3B1).2 For MDS-SF3B1 patients not candidate to, refractory or relapsing after ESAs, luspatercept has recently become an effective therapeutic option.3 The drug acts as a TGF-β fusion trap protein and neutralizes TGF-β ligands, which negatively regulate terminal erythroid maturation, thus improving ineffective erythropoiesis.3 In the phase 3 trial, 37.9% of patients obtained transfusion independence (TI) for at least 8 weeks during the first 24 weeks of study, versus 13.2% in the placebo group.3 However, no clear predictors of response emerged, and the contribution of the pro-inflammatory/immunologic microenvironment to response was not established. The relationship between MDS and immune system activation has been largely supported by the epidemiologic association with autoimmune diseases/phenomena, by the observation of a T-cell infiltrate and of a pro-apoptotic cytokine milieu in the bone marrow, and by the clinical response to several immunomodulating agents (i.e. steroids, androgens, cyclosporine, and anti-thymocyte globulin).4 Here we systematically studied a cohort of MDS-SF3B1 patients treated with luspatercept focusing on the dynamics of cytokine levels during treatment, the distribution of polymorphisms of cytokine genes, and the mutational status and allele burden of several genes related to myeloid neoplasms. Patients with MDS-SF3B1 treated with luspatercept from January 2021 until the time of writing at two tertiary hematologic centers in Milan, Italy, were included in the study. They were classified according to IPSS-R and IPSS-M, and transfusion need (Appendix S1). Luspatercept was administered subcutaneously according to current data sheet (1–1.75 mg/kg every 3 weeks) and response to treatment was classified as attainment of TI or reduction of transfusion burden (<50%) over an 8-week period as compared with baseline. Cytokine serum levels were evaluated in peripheral blood at baseline and at 9 and 33 weeks of treatment using high-sensitivity ELISA kits. Cytokine genotyping was done by the Cytokine Genotyping Tray (One Lambda, Canoga Park, CA), based on PCR-sequence specific primer (PCR-SSP) amplification. Molecular analysis of MDS-related genes was performed by targeted gene panel including coding exons and splice sites of 73 genes involved in myeloid malignancies (Appendix S1). A total of 23 patients were included in the analysis (Table S1). Patients were mainly elderly (median age 72 years, range 46–86) with a male to female ratio of 4.75, and belonged to either very low, low, or intermediate IPSS-R risk category, while IPSS-M was mostly low or moderate-low (91%). Median time from diagnosis to luspatercept was 52 months (2.8–235), and all subjects had been previously treated with ESAs. Overall, patients required a median of 7 (3–13) red blood cell units in an 8-week period before luspatercept, and 74% were classified as having a “high transfusion burden”. After a median follow up of 18 months (13–24 months), a total of 15 patients (65%) obtained a hematologic improvement: 11 (48%) became transfusion independent, 4 (17%) displayed a reduction ≥50% of transfusion need. The only baseline parameter associated with response was low transfusion burden, present in 40% of responders versus 0% of nonresponders (p = .02). Median time on luspatercept was 11.4 months (3.7–24). Twenty-one of 23 patients required dose titration at 1.75 mg/kg every 3 weeks, 1 at 1.33 mg/kg and 1 continued at 1 mg/kg with persistent TI. Nine patients discontinued treatment after at least seven doses, mainly due to nonresponse. Two patients (one responder and one non-responder) eventually evolved to acute leukemia and died. Table S2 shows cytokine levels in MDS patients versus controls at baseline: IL-6, IL-10, and IL-17 levels were significantly higher in patients versus controls (p ≤ .005). Likewise, TGF-β levels were higher in MDS patients versus controls, although not significantly. On the contrary, TNF-α levels were lower in patients versus controls (p < .001). Considering response to luspatercept, responders showed markedly lower IL-17 levels as compared with nonresponders, both at baseline (p < .001) and along treatment (p < .001 at week 9 and p = .04 at week 33), being more similar to healthy controls. Likewise, IL-10 levels were lower in responders versus nonresponders, reaching statistical significance after 33 weeks of therapy (p = .01). The dynamics of cytokine levels at week 9 and 33 of therapy in responders and nonresponders are depicted in Figure 1A. We next analyzed the prevalence of genotypes of cytokine gene polymorphisms and found that the frequency of the TGF-β gene codon 10 T/T genotype was significantly lower in MDS versus controls, including those responding to luspatercept (p < .05) (Table S3), while no differences were noted in the allelic frequency nor in the distribution of low/intermediate/high producer profiles (Tables S4 and S5). NGS-based mutational analysis was performed in 18 patients (11 responders and 7 nonresponders) at two time points, with a 6- to 18-week interval after treatment start. At baseline, all patients had at least one driver mutation (median 2, range 1–5) (Figure 1B). All but one displayed SF3B1 mutation (median VAF 34%, 9–43). Co-mutations occurred in 12 of 17 SF3B1-mutated patients, involving mainly TET2 (n = 6). Patients showed variable trends of SF3B1 mutation's VAF during treatment, with increasing, decreasing, and stable levels: six patients experienced a significant increase, seven stable, and five displayed reduced VAFs. A trend emerged associating nonresponders to stable driver mutations' VAF (Figure S1). VAF delta changes were similar among responders and nonresponders to luspatercept. Notably, five of six patients whose SF3B1 VAF increased during treatment were responders to luspatercept (Figure S2). Here we report for the first time the association of the immune-regulatory cytokine pattern with luspatercept efficacy in a prospective cohort of MDS-SF3B1 patients and describe a possible immunomodulatory effect of the drug. We confirm that luspatercept is beneficial in two of three patients with MDS-SF3B1, inducing TI in nearly half of them, particularly in patients with low transfusion burden.3, 5 Mutational data did not predict response, as SF3B1 VAFs as well as type and number of co-mutated genes were similar between responders and nonresponders.3, 5 However, five of the six patients who showed increased SF3B1 VAF were responding to the drug, in line with the mechanism of action of luspatercept that is not aimed at “killing the clone” but rather at improving ineffective erythropoiesis.3 Although MDS is a clonal disease, the surrounding inflammatory and immunological microenvironment may have a role in the pathophysiology and severity of disease.4 In general, TNF-α, IL-6, and IL-8 were found significantly higher in MDS patients, and IFN-γ and IL-17 moderately increased.6 This applies to all MDS, which are highly heterogeneous, while we studied SF3B1 patients. We observed that TNF-α levels were lower in patients versus controls, possibly indicating a greater utilization. Additionally, we detected markedly increased levels of IL-6 and IL-10 as compared with controls. The former is a well-known inflammatory cytokine while the latter is mainly an anti-inflammatory mediator, and both are involved in T helper 2 responses.6 This scenario may reflect an attempt to downregulate an inflammatory T helper 1 and cytotoxic microenvironment. More interestingly, IL-17 has been found increased in patients versus controls in line with previous findings.6 IL-17 is a pleiotropic cytokine produced by T helper 17 and involved in the differentiation of T regulatory cells (Tregs) into cytotoxic effector cells.6 This cytokine has a pivotal role in autoimmune diseases, including immune-mediated cytopenias,6 and may be involved in the generation of unfavorable cytotoxic bone marrow microenvironment. More interestingly, a different cytokine pattern emerged in patients responding to luspatercept versus nonresponders, with lower IL-10 and IL-17 levels in the former at baseline and along treatment. This cytokine signature, more similar to healthy controls, may in turn have favored a less cytotoxic bone marrow milieu. Considering TGF-β, which was expected to be modulated by luspatercept, we found slightly higher levels in MDS patients versus controls. In MDS, TGF-β signaling has been identified as key mediator of apoptosis and ineffective erythropoiesis.6 Additionally, TGF-β is a crucial inducer of T helper 17 and Tregs, suggesting that its slight increase may be responsible for the greater IL-17 levels observed in our series. As regards the distribution of polymorphisms of cytokine genes, we found no significant differences in patients versus healthy controls, nor in responders versus nonresponders, indicating that the genetic predisposition to high or low cytokine production is not determinant in this context. Our study carries the limitation of including a limited number of patients; however, the systematic and prospective clinical evaluation, and the centralization of laboratory and data analysis increase the value of the results. In conclusion, these results suggest that the immunological microenvironment may dynamically influence response to luspatercept in MDS-SF3B1 patients and may pave the way to combination therapy with immunomodulatory agents. BF, LR, MR, LP, NG, FM, and WB followed patients, designed the study, collected data, and wrote the article. AM, ML, AkM, AM, MCDV, and NB performed molecular studies and wrote the manuscript. AZ and EF designed the study, performed cytokine essays and cytokine polymorphisms analysis, and wrote the manuscript. FP wrote the article. All authors revised the manuscript for important intellectual content. Open access funding provided by BIBLIOSAN. The study was partially funded by Italian Ministry of Health, Current Research Grant. All authors declare that they have no conflict of interest to disclose. All authors approved present submission. All data are available within the manuscript and further may be available upon reasonable request to the corresponding author. Appendix S1. Supporting Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.