From Immune Dysregulations to Therapeutic Perspectives in Myelodysplastic Syndromes: A Review

The pathophysiology of myelodysplastic syndromes (MDSs) is complex and often includes immune dysregulation of both the innate and adaptive immune systems. Whereas clonal selection mainly involves smoldering inflammation, a cellular immunity dysfunction leads to increased apoptosis and blast proliferation. Addressing immune dysregulations in MDS is a recent concept that has allowed the identification of new therapeutic targets. Several approaches targeting the different actors of the immune system have therefore been developed. However, the results are very heterogeneous, indicating the need to improve our understanding of the disease and interactions between chronic inflammation, adaptive dysfunction, and somatic mutations. This review highlights current knowledge of the role of immune dysregulation in MDS pathophysiology and the field of new drugs.


Introduction
Myelodysplastic syndromes (MDSs) are acquired clonal myeloid malignancies characterized by ineffective hematopoiesis resulting in peripheral cytopenia and a risk of progression to acute myeloid leukemia (AML) [1]. MDSs are a heterogeneous group of diseases with differences in clinical presentation, biological features, prognosis, and treatment. Severity is defined by the Revised International Prognostic Scoring System (IPSS-R) [2] but some studies identified other prognostic factors such as patient characteristics (comorbidities) [3] or somatic mutations [4] and new scores are actually being developed [5].
In Low-risk MDS (LR-MDS), the objectives of treatments are to correct peripheral cytopenia. In contrast, blastic progression and risk of AML transformation are the major concerns of High-Risk MDS (HR-MDS) patients, and treatments target the malignant clone [6]. In both situations, however, available therapeutic options are limited, and a better comprehension of MDS pathophysiology is needed to identify new targets [7].
The central role of the immune system in tumor surveillance is well known and targeted in solid tumors [8] but its implication in myeloid malignancies pathogenesis, especially in MDS, is less well described. However, recent studies tried to explore dysregulations in both innate and adaptive compartments of immunity and their consequences on MDS presentation, prognosis, and risk of AML transformation. Moreover, T-cells from patients with MDS may originate from the malignant clone [9]. Together, these studies showed that MDS is associated with several immune alterations leading to cytopenia

Smoldering Inflammation Acts Early in MDS Pathophysiology
We previously described the role of inflammation in clonal selection and progression, but dysregulations of innate immune system components are also involved in MDS pathophysiology.

NLRP3 Inflammasome
Inflammasomes play a key role in MDS pathophysiology, particularly in LR-MDS [44][45][46]. In an activated NLR family, NOD-like receptor pyrin domain-containing 3 (NLRP3) complexes are implicated in the pyroptosis of MDS cells, which is an inflammatory cell death process [47]. The NLRP3-pyroptosis axis is found to be activated in MDS BM cells [48] and is responsible for LR-MDs features such as macrocytosis and ineffective hematopoiesis resulting in cytopenia. Activation of the NLRP3 inflammasome is initiated by damage-associated molecular pattern signals (DAMP) (e.g., S100A8/9 [49]), and involves signal transducers such as serine/threonine kinases IRAK1/4 [50] and the E3 ubiquitin ligase TRAF6. Once activated, NLRP3 recruits and activates pro-caspase-1, which then activates pro-IL-1β, pro-IL-18, and the pore-forming protein gasdermin D [51], followed by pyroptotic cell death with the release of active IL-1β, IL-18, and other intracellular proteins that contribute to local inflammation [48,52]. Interestingly, S100A8 and S100A9 proteins are increased in the HSCs or blood of MDS patients, especially in LR-MDS [53]. These proteins also expand Myeloid-Derived Suppressor Cells (MDSCs) in the BM of patients with MDS [54].

Toll-like-Receptors (TLR)
The TLR family contains 10 subtypes in humans, and these are expressed on a variety of hematopoietic cell types [55]. TLR signaling regulates hematopoietic stem and progenitor cell function [56]. MDS HSCs are particularly sensitive to pyroptosis because of high levels of TLR (TLR 1-2-6) expression and activation [42, [57][58][59]. Moreover, while the overexpression of TLR-2 or TLR-4 is observed in LR-MDS, increased expression of TLR-6 correlates with HR disease [60]. HSCs of LR-MDS patients are more sensitive to TLR ligands (DAMP) and some of these ligands are more abundant in the BM and/or serum of MDS patients compared to healthy controls (ex HMGB1 [61], S100A8/9 [53]). In LR-MDS, S100A8/A9 is more abundant (also secreted by MDSCs) and binds to TLR-4 and CD33, stimulating inflammasome assembly and pyroptotic cell death by promoting the production of inflammasome components, Reactive Oxygen Species (ROS), and proinflammatory cytokines such as IL-1β and IL-18, and the expansion of MDSCs. The consequence is the suppression of normal hematopoiesis via the production of cytokines such as IL-10 and TGF-β.

Immune Cells
MDSCs [62] are increased in the BM and blood of MDS patients compared to healthy donors, especially in HR-MDS, and are involved in ineffective hematopoiesis. MDSCs also induce T-cell immunosuppression via the secretion of IL-10 and TGF-β, leading to MDS progression. MDSC expansion is driven by the interaction of S100A9 with CD33.
Monocytes: The number of monocytes in the blood or BM of MDS patients is increased, especially in HR-MDS, but their ability to differentiate into macrophages to then use their phagocytic function is decreased [63,64]. Cytokines, chemokines, and TLR genes are also downregulated in BM monocytes from MDS patients. Moreover, as the disease progresses, the number of abnormal monocytes increases in the BM of the patients [65]. In addition, the ratio of M2 macrophages to monocytes is higher in patients with MDS whereas the ratio of M1 to M2 macrophages is lower in MDS patients. Because monocytes play a key role in the defense against microbial agents, these alterations may explain the susceptibility to infections observed in MDS patients.
Dendritic cells (DCs): Some studies described quantitative or qualitative defects of DCs in MDS patients [66] with, for example, an altered cytokine profile [67]. High secretion of IL-10 may contribute to the immunosuppressive phenotype. Interestingly, it was shown that cytogenetic abnormalities in DCs from MDS patients were similar to the malignant clones of MDS, suggesting that they originated from the malignant clones [68]. DCs derived from MDS monocytes also have morphologic and phenotypic abnormalities [64].
Lymphocytes: A reduced number of lymphocytes, or their dysfunction, can also have prognostic implications. For example, the number of Tregs could be a prognostic factor in LR-MDS and predict the severity of anemia, AML transformation, and overall survival [69]. A low number of NK-cells, their reduced cytotoxic capacity, or their Killercell immunoglobulin-like receptors (KIR) gene haplotype are also associated with poor prognosis and AML transformation [70][71][72].

Positive and Negative Immune Regulators
Immune cell signaling includes co-stimulatory and co-inhibitory (named Immune check points or ICPs) molecules that are differentially expressed by a variety of cell types (such as antigen-presenting cells, B-cells, macrophages, and T-cells) and may be regulated by activation [73]. Some of these molecules were studied in MDS and appeared to be potential therapeutic targets. Other positive regulators, such as CD244, CD80, and CD40, have been studied but their implication is not clearly validated and they are not yet interesting therapeutic leads [75,76].
Program cell death-1, and L1 (PD-1 and PD-L1): The PD-1/PD-L1 axis is well studied. Upregulation of PD-1 and PDL-1 was seen in CD34+ BM cells from patients with MDS and in T-cells, with further upregulation seen in those previously treated with HMAs [77,78]. Moreover, a higher rate of PD-L1 expression on BM blasts and HSCs from TP53-mutated AML and MDS patients is observed [74]. The expression of ICPs could be upregulated in the presence of pro-inflammatory cytokines [79] or in response to S100A9 signaling, contributing to MDSC-induced HSC death [80].
T cell immunoreceptor with Ig and ITIM domains (TIGIT): TIGIT was found to be highly expressed in NK-cells and T-cells from the blood of MDS patients with an association with disease progression [85]. The overexpression of TIGIT can be associated with decreased NK and T-cell functions, and lower secretions of activating cytokines such as IFN-γ and TNF-α.
ICPs expression is also associated with resistance to hypomethylating agents [77]. Moreover, the impact of ICP expression on MDS prognosis was assessed in a recent study, which evaluated the mutation burden in genes coding for ICP molecules (LAG-3, CTLA-4, B7H3, PD-1, PD-L1, etc.). The authors observed a high prevalence of mutations in these genes, with an impact on overall survival [86].

Immune Phenotypes Can Be Associated with Somatic Mutations
Clonal hematopoiesis and MDS progression are driven by the acquisition of somatic mutations. However, these mutations not only act on clonal selection but also on the immune microenvironment. Then, mutations affecting DNA methylation (TET-2 and DNMT3A), histone modification (ASXL1), or splicing factors (SF3B1, SRSF2, U2AF1) have been associated with increased inflammation in patients with CHIP or MDS.
TET2 mutation seems to promote clonal HSC dominance by creating an inflammatory environment [32,95]. Interestingly, TET2 mutant CHIP is more highly associated with elevated IL-6, while DNMT3A mutant CHIP is more highly associated with elevated TNF-α or IFN-γ [96,97]. Recently Zhao et al. found that TET2 or IDH 1/2 mutations were more frequent in MDS patients presenting with systemic autoinflammatory or autoimmune diseases [98]. In this study, TET2/IDH mutant patients' phenotyping showed a reduction of Tregs and deep alterations in CD8+ T-cell distribution.
As previously described, Sallman et al. recently described a distinct immune phenotype associated with TP53 mutated MDS/AML patients [74]. These patients overexpressed ICPs PD-L1 (especially in HSCs) and CTLA-4, associated with poor survival. Moreover, TP53 mutant patients displayed reduced numbers of BM OX40+ cytotoxic T-cells and helper T-cells, reduced ICOS+ and 4-1BB+ NK cells, and an increased number of immunosuppressive regulatory T-cells and MDSCs. Finally, they found that a higher proportion of BM infiltrating ICOShigh/PD-1neg Tregs was a highly significant predictor of overall survival.
Finally, driver mutations have been found in lymphoid precursors, raising the question of the potential impact of these mutations on immune responses in the MDS context [108].

Immune Strategies in MDS: Past, Present and Future
Targeting the immune system in MDS is not a novel concept, but modern approaches such as ICP inhibitors (ICPi) or an adoptive T-cell transfer may be promising and are still under evaluation (Table 1).

"Old" Treatments
Immunosuppressive and immunomodulatory therapies have been used in selected MDS patients.
First, allogenic Hematopoietic stem-cell transplantation (HSCT) is actually the only curative treatment for MDS. It is considered an immune approach because of the graft versus leukemia effect but also because of its immune modulatory properties, especially in the context of reduced-intensity conditioning transplant programs [109].
In del(5q) LR-MDS, lenalidomide, an immunomodulatory drug, has shown a high rate of efficacy with sustained responses [116]. It acts on T-cell activity via several mechanisms, including reduced T-cell tolerance and increased effector functions [117]. Other studies have shown its role in the ubiquitination and degradation of specific substrates [118]. Lenalidomide is currently used in LR-MDS with del(5) patients who are still transfusiondependent despite treatment with erythropoietic-stimulating agents [119].
Alemtuzumab is a humanized monoclonal antibody directed against CD52, a glycosyl phosphatidyl inositol-anchored cell surface marker abundantly expressed on B and T-cells and at low levels on effector cells of the innate immune system. It was evaluated in a subset of MDS patients and showed good efficacy and a good safety profile [120].
Targeting pro-inflammatory cytokines is also an option in MDS, especially in LR-MDS patients. Infliximab, an antibody targeting TNF-α, was evaluated in LR-MDS and showed a moderate hematologic response but a good safety profile [121]. Other agents such as Etanercept, were associated with HMA and showed interesting results [122]. Studies using the anti-TNF-α antibody in combination with other drugs are ongoing. Other biologics are actually used in MDS-associated systemic inflammatory or auto-immune disorders (such as anti-IL-1, IL-6 antibodies) but are also evaluated in combination with other drugs [123].

Targeting TGF-β Superfamily in LR-MDS with Anemia
Increased concentrations of TGF-β superfamily ligands, including growth differentiation factor 11 (GDF11) in bone marrow have been linked to ineffective erythropoiesis in MDS [124,125]. Luspatercept (ACE-536) is a soluble fusion protein that binds to GDF11 and other TGF-β superfamily ligands. It acts as an activin receptor type II ligand trap and allows the restoration of terminal maturation of erythroid progenitors by diminishing Smad2/3 signaling. It was recently evaluated in a phase III, randomized, double blind, placebo-controlled study for LR-MDS patients with ring sideroblasts who were refractory/intolerant or ineligible for ESA, and transfusion dependent. Results are encouraging, and luspartecept obtained FDA and EMA approbation in this indication [126,127].

Checkpoint Inhibitors
Following the successes with ICPi in solid tumors, these therapies are being evaluated in hematologic malignancies, including AML and MDS [133]. The first ICPi molecules challenged in clinical trials for HR-MDS patients are ipilimumab [134][135][136] (CTLA-4 inhibitor), nivolumab [137,138] or pembrolizumab [139,140] (PD-1 inhibitors), and atezolizumab or durvalumab [141][142][143] (PD-L1 inhibitors). Targeting TIGIT, CD70, or TIM-3 also seems to be an interesting option [84,85,[144][145][146]. Response rates in monotherapy approaches are actually very low and combined strategies are more promising, especially with HMA. Indeed, treatment with HMA has been shown to upregulate several of these ICPs (such as PD-L1, TIM-3, and CD47) on MDS cells [77]. Moreover, the association of two ICPis (ipilimumab and nivolumab) has been also evaluated, finding the same efficacy as the results for monotherapy [135]. More recently, new ICPs have been identified as CD47, especially expressed by macrophages and acting as a "do not eat me" signal [147][148][149]. MDS cells from HR-MDS patients overexpress CD47 and this is associated with poor survival. Thus, the CD47-SIRPa interaction could be targeted by specific inhibitors [150]. Two anti-CD47 inhibitors have been evaluated in HR-MDS patients. First, CC-90002 was tested in phase 1, but was associated with a major toxicity profile and poor preliminary efficacy [151]. Another CD47 inhibitor, Magrolimab (HuF9-G4 or 5F9), seems to be less toxic than CC-90002 in a phase I trial [152]. Investigators reported high efficacy of magrolimab in association with HMA and a phase III evaluation of this association is currently in the recruitment stage (NCT04313881) [152].
Together, ICPis showed moderate and variable efficacy in early-phase trials with HR-MDS patients. The association with HMA seems to be more interesting than monotherapy, especially in the front line. Responders must be characterized more to identify the predictive factors of response with each ICPi.

Targeting NK-Cells
Similar to T-cells, NK-cells can also be exhausted in HR-MDS with the loss of their cytotoxic capacity [85,153,154]. Restoring their functional activity could help to avoid MDS progression. To address this hypothesis, new molecules are actually being evaluated, such as GTB-3550 TriKE™, a novel CD16/IL-15/CD33 Tri-Specific Killer Engager (TriKE) [155,156]. Pre-clinical data show specific NK cell activation and targeted cytolytic activity in xenogeneic AML mouse models. Another way to improve NK cell activity consists of blocking interactions between KIRs and human leukocyte antigen-C (HLA-C) molecules. Lirilumab is a human IgG4 monoclonal antibody that blocks KIR/HLA-C interaction [157].

Vaccination Strategies
Several tumor-associated antigens (TAA) have been identified and found to be overexpressed in MDS patients, leading to the development of specific vaccines to stimulate tumor-specific T cells [158][159][160]. TAAs targeted by peptide vaccines include Wilm's tumor-1 (WT1), Proteinase-3+ neutrophil elastase (PR-1), or NY-ESO-1. The expression of WT1 is correlated with the marrow blast percentage, poor response to HMA, and poor OS [159,[161][162][163][164]. The first studies of the peptide vaccine of WT1 have demonstrated an acceptable safety profile, good biological response (expansion of WT1 reactive T cells), but with no significant sustained clinical response [159,[165][166][167][168]. PR-3 is another potential target, and a PR-1 vaccine (an HLA-A2-restricted peptide-based PR3 in combination with neutrophil elastase) was recently evaluated with the same profile of response as the WT1 vaccine [160]. NY-ESO1, a cancer testis antigen, has been found expressed on some solid tumors [169]. NY-ESO1 expression is low in MDS due to the silencing of TAA genes by the methylation mechanism [170]. As expected, early studies showed an increase in NY-ESO1 expression after treatment with HMA [171] and the results in MDS patients are encouraging [172]. New multiepitope vaccines are also being evaluated in combination with HMA, but the first results are disappointing (negative results and AML progression) [173].
To conclude, vaccine strategies in MDS have actually shown a good safety profile and interesting biological results but with poor clinical responses. Because dendritic cells play a key role in vaccine response and because these cells present some dysfunction in MDS patients [64,67,174], it could be important to consider this issue. Other vaccine strategies are also evaluated [175].

Adoptive T-Cell Transfer Therapy
Chimeric antigen receptor: MDS cells have been shown to overexpress the ligands for Natural killer group 2 (NKG2D) receptors [176,177], and clinical trials with NKG2D receptors CAR-T cells have been evaluated, but without favorable outcomes [178]. The expansion process could be improved by lymphodepleting chemotherapy prior to infusion, but a major issue seems to be the inability of CAR-T to persist as memory lymphocytes. Other studies evaluating new CAR-T cells are ongoing [179,180]. Because CD123 delineates MDS stem cells in HR-MDS patients, CD123 CARs have been generated and have demonstrated promising results in vitro and in vivo [181].
Donor-directed lymphocytes: While HSCT is the only curative treatment for MDS, relapses can occur and are associated with poor outcomes. In these cases, unmanipulated DLIs may be an option but these are associated with severe GVHD. DLI could be stimulated ex vivo by TAA, which selects for an enriched, polyclonal CD4+ and CD8+ specifically directed against MDS. TTA-DLI appears to be safer, with respect to GVHD, than unselected DLI [182,183].

Conclusions
Recent interest for alterations in the immune system in MDS has led to an improved understanding of its pathogenesis (Figure 1). Both the innate and adaptive immune systems have been shown to participate in MDS pathogenesis and progression, which has allowed the development of different therapeutic approaches in LR-MDS and HR-MDS. However, results are heterogeneous and future investigations should further explore the MDS immune landscape. In parallel, future studies will have to determine which patients will respond to a predefined treatment, in order to achieve a personalized approach.
MDS immune landscape. In parallel, future studies will have to determine which patients will respond to a predefined treatment, in order to achieve a personalized approach.