Natural Killer Immunotherapy for Minimal Residual Disease Eradication Following Allogeneic Hematopoietic Stem Cell Transplantation in Acute Myeloid Leukemia

The most common cause of death in patients with acute myeloid leukemia (AML) who receive allogeneic hematopoietic stem cell transplantation (allo-HSCT) is AML relapse. Therefore, additive therapies post allo-HSCT have significant potential to prevent relapse. Natural killer (NK)-cell-based immunotherapies can be incorporated into the therapeutic armamentarium for the eradication of AML cells post allo-HSCT. In recent studies, NK cell-based immunotherapies, the use of adoptive NK cells, NK cells in combination with cytokines, immune checkpoint inhibitors, bispecific and trispecific killer cell engagers, and chimeric antigen receptor-engineered NK cells have all shown antitumor activity in AML patients. In this review, we will discuss the current strategies with these NK cell-based immunotherapies as possible therapies to cure AML patients post allo-HSCT. Additionally, we will discuss various means of immune escape in order to further understand the mechanism of NK cell-based immunotherapies against AML.


Introduction
Allogeneic hematopoietic stem cell transplantation (allo-HSCT) has been recognized as the only curative therapy for patients with acute myeloid leukemia (AML). Allo-HSCT's mode of action is primarily attributed to the graft-versus-leukemia (GVL) effect mediated by donor T-cells and natural killer (NK) cells. However, approximately 40% of the AML patients who undergo allo-HSCT will relapse, and the two-year post relapse survival among these patients is less than 20% [1][2][3][4][5][6]. With the use of targeted sequencing or flow cytometry, the persistent detection of minimal residual disease (MRD) is associated with post-transplantation relapse [7][8][9]. It is therefore important to provide additional therapies to eliminate MRD after allo-HSCT, particularly in high-risk AML. Donor lymphocyte infusion (DLI) or repeat allo-HSCT as a donor cell-based therapy has been associated with improved survival in patients who relapse after allo-HSCT [1][2][3]5,10]. Although the efficacy of therapeutic DLI in relapsed AML may be suboptimal, pre-emptive, or prophylactic, DLI may have an important role [5,[11][12][13]. Use of the hypomethylating agent azacitidine appears to be effective in AML following allo-HSCT [14,15]. Additionally, pre-emptive treatment with azacitidine may prevent a relapse while monitoring for MRD (NCT01462578) [16].
Previously, it had been presumed that most additional therapies would not be able to suppress the proliferation of leukemia cells in the long term in relapsed AML after allo-HSCT. However, early use of these therapies might prevent relapse in AML, and could be an important step toward improving prognosis. Recently, immunotherapies, including NK cells administration and immune checkpoint inhibitors (ICIs), have been reported as new treatment modalities after allo-HSCT in hematologic malignancies [17][18][19][20][21][22][23]. Previous studies had demonstrated that ICIs had antitumor immune responses for several solid tumors and hematologic malignancies [24][25][26]. However, their responses remained limited because of the lack of MHC classes I and II, which leads to less T-cell activation and proliferation, and is observed in ICI-resistant tumors [27][28][29]. In contrast, while NK cells express limited MHC (e.g., human leukocyte antigen (HLA)-Bw4, C1, and C2)-dependent receptors, they express non-MHC-dependent receptors including NKG2D, natural cytotoxicity receptors, CD96, T-cell immunoreceptor with Ig, and immunoreceptor tyrosine-based inhibition motif domains (TIGIT), DNAM-1, SLAMF6 (also known as NTB-A), NKRP1-B, and 2B4 [21,30,31]. Additionally, consistent with donor T-cell mediated GVL, donor T-cells' contribution to graft versus host disease (GVHD) is dependent upon recognition of HLA disparities following allo-HSCT. While administration of some ICIs post allo-HSCT may lead to severe GVHD [17,24], donor NK cells confer alloreactivity against tumors without GVHD [32,33]. Recently, we have noted that high NK cell levels in the bone marrow microenvironment immediately following allo-HSCT were associated with better overall survival (OS) and progression-free survival [34]. Moreover, AML patients with lower TIGIT expression following allo-HSCT had superior OS and progression-free survival [35]. Therefore, strategies to activate NK cells in order to reinforce GVL effect as a pre-emptive or prophylactic immunotherapy may improve MRD clearance in high-risk AML after allo-HSCT ( Figure 1). In this review, we focus on NK cell-based immunotherapies following allo-HSCT and explore emerging therapies to eradicate MRD.

Adoptive NK Cell Therapy and Cytokine-Based NK Cell Therapy
Previous studies have reported an association between clinical outcomes and NK cell recovery after allo-HSCT. This likely occurs because NK cells play an essential role in GVL effects and also in preventing infection following allo-HSCT [34,36,37]. To date, adoptive transfer of NK cells from Some patients with AML after conventional allo-HSCT will relapse. For the prevention of relapse, immunotherapies may play an important role in the elimination of MRD.

Adoptive NK Cell Therapy and Cytokine-Based NK Cell Therapy
Previous studies have reported an association between clinical outcomes and NK cell recovery after allo-HSCT. This likely occurs because NK cells play an essential role in GVL effects and also in preventing infection following allo-HSCT [34,36,37]. To date, adoptive transfer of NK cells from allogeneic donors to patients with AML has been performed following allo-HSCT [38][39][40][41][42][43][44]. Additionally, NK cell infusion has been combined with the administration of IL-2 to boost in vivo expansion ( Figure 2) [45][46][47][48]. T-regulatory cells (Tregs) are significantly increased in number following NK cell infusion and IL-2 administration, which may inhibit NK cell functionality and hinder the efficacy of adoptively transferred NK cells (Figure 3). In cases with prior IL-2-diphtheria toxin fusion protein treatment for the depletion of host Tregs, increased in vivo expansion of NK cells was noted, and relapsed/refractory AML patients were able to achieve complete remission (CR) (NCT00274846 and NCT01106950) [47]. Besides IL-2 administration, NK cells activated by IL-12, IL-15, IL-18, and IL-21 have enhanced antitumor functionality [49][50][51][52]. These cytokines also lead to an increase in varying degrees of host and/or donor CD8 + T-cells. Therefore, these therapies may result in adverse events, including severe GVHD. However, previous studies have demonstrated that adoptively transferred NK cells activated by these cytokines had GVL effect without life-threatening GVHD [49][50][51][52]. IL-15/IL-15Ra-Fc (ALT-803) therapy (NCT01885897), for instance, promoted an increase in CD8 + T-cells of the effector or effector memory phenotype without increasing Tregs, and no patient developed severe GVHD despite the induction of CD8 + T-cell activation [51]. One possible reason may be that the preferential expansion of NK cells mediates a reduction of GVHD by inhibiting CD8 + donor T-cell proliferation [53]. Although adoptive transfer of NK cells during allo-HSCT may be a promising therapy for AML, further studies are required in order to design protocols that balance the persistence of donor NK cells and host/donor T-cell activation. These studies must include the timing of transferred NK cells, NK cell dosage, combination with cytokines, the conditioning regimen, donor selection, and GVHD prophylaxis. allogeneic donors to patients with AML has been performed following allo-HSCT [38][39][40][41][42][43][44]. Additionally, NK cell infusion has been combined with the administration of IL-2 to boost in vivo expansion ( Figure 2) [45][46][47][48]. T-regulatory cells (Tregs) are significantly increased in number following NK cell infusion and IL-2 administration, which may inhibit NK cell functionality and hinder the efficacy of adoptively transferred NK cells ( Figure 3). In cases with prior IL-2-diphtheria toxin fusion protein treatment for the depletion of host Tregs, increased in vivo expansion of NK cells was noted, and relapsed/refractory AML patients were able to achieve complete remission (CR) (NCT00274846 and NCT01106950) [47]. Besides IL-2 administration, NK cells activated by IL-12, IL-15, IL-18, and IL-21 have enhanced antitumor functionality [49][50][51][52]. These cytokines also lead to an increase in varying degrees of host and/or donor CD8 + T-cells. Therefore, these therapies may result in adverse events, including severe GVHD. However, previous studies have demonstrated that adoptively transferred NK cells activated by these cytokines had GVL effect without life-threatening GVHD [49][50][51][52]. IL-15/IL-15Ra-Fc (ALT-803) therapy (NCT01885897), for instance, promoted an increase in CD8 + T-cells of the effector or effector memory phenotype without increasing Tregs, and no patient developed severe GVHD despite the induction of CD8 + T-cell activation [51]. One possible reason may be that the preferential expansion of NK cells mediates a reduction of GVHD by inhibiting CD8 + donor T-cell proliferation [53]. Although adoptive transfer of NK cells during allo-HSCT may be a promising therapy for AML, further studies are required in order to design protocols that balance the persistence of donor NK cells and host/donor T-cell activation. These studies must include the timing of transferred NK cells, NK cell dosage, combination with cytokines, the conditioning regimen, donor selection, and GVHD prophylaxis.

ICIs for Intensifying the Activation of NK Cells
NK cells express various co-inhibitory receptors, including killer immunoglobulin-like receptors (KIRs), NKG2A, programmed cell death protein 1 (PD-1), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), T-cell immunoglobulin, and mucin domain-containing protein 3 (TIM-3), TIGIT, and lymphocyte activation gene 3 (LAG-3). These receptors have recently been recognized as immune checkpoints (Figures 2 and 4). T-cells also express these immune checkpoints in which PD-1, CTLA-4, LAG-3, and B-and T-lymphocyte attenuator (BTLA) had greater expression than in NK cells ( Figure 5) [54,55]. In order to block a receptor's inhibitory signal, NK cell activation and leukemia cell killing are induced by a cognate ligand by ICIs. Therefore, administration of ICIs in the first few months post allo-HSCT might be a powerful tool for eradicating MRD following AML. Several clinical trials for ICIs as monotherapy or as part of combination treatment for AML after allo-HSCT have been reported recently [17,20,24,[56][57][58]. CTLA-4 (e.g., ipilimumab) and PD-1 (e.g., nivolumab) blockade have been administered following allo-HSCT in several hematologic malignancies, and beneficial GVL responses have been achieved. However, serious immune-related adverse events and/or severe GVHD were accompanied by exposure to these two ICIs [59]. This might explain how increased donor-derived alloreactive T-cells might cause life-threatening GVHD [57,58]. It would be interesting to test whether NK cell-specific ICIs might enhance anti-leukemia activity without aggravating GVHD more than CTLA-4 or PD-1 blockade.

ICIs for Intensifying the Activation of NK Cells
NK cells express various co-inhibitory receptors, including killer immunoglobulin-like receptors (KIRs), NKG2A, programmed cell death protein 1 (PD-1), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), T-cell immunoglobulin, and mucin domain-containing protein 3 (TIM-3), TIGIT, and lymphocyte activation gene 3 (LAG-3). These receptors have recently been recognized as immune checkpoints (Figures 2 and 4). T-cells also express these immune checkpoints in which PD-1, CTLA-4, LAG-3, and B-and T-lymphocyte attenuator (BTLA) had greater expression than in NK cells ( Figure 5) [54,55]. In order to block a receptor's inhibitory signal, NK cell activation and leukemia cell killing are induced by a cognate ligand by ICIs. Therefore, administration of ICIs in the first few months post allo-HSCT might be a powerful tool for eradicating MRD following AML. Several clinical trials for ICIs as monotherapy or as part of combination treatment for AML after allo-HSCT have been reported recently [17,20,24,[56][57][58]. CTLA-4 (e.g., ipilimumab) and PD-1 (e.g., nivolumab) blockade have been administered following allo-HSCT in several hematologic malignancies, and beneficial GVL responses have been achieved. However, serious immune-related adverse events and/or severe GVHD were accompanied by exposure to these two ICIs [59]. This might explain how increased donor-derived alloreactive T-cells might cause life-threatening GVHD [57,58]. It would be interesting to test whether NK cell-specific ICIs might enhance anti-leukemia activity without aggravating GVHD more than CTLA-4 or PD-1 blockade.      IPH2101, also known as anti-KIR1-7F9 monoclonal antibody (mAb; lirilumab is a recombinant version of this mAb), blocks common inhibitory KIRs (KIR2DL/DS-1, -2, and -3), which bind to HLA-C alleles and augment NK cell-mediated killing in HLA-C-expressing leukemia cells [60]. Although this KIR2D blockade showed no clinical effectiveness [61] in a phase 2 trial in smoldering multiple myeloma (NCT01248455), adoptive transfer of NK cells combined with IPH2101 after allo-HSCT may have a therapeutic benefit. With lirilumab as with ICIs, CD94-NKG2A receptors on NK cells primarily recognize HLA-E, which is expressed by leukemia cells. Anti-NKG2A mAb (monalizumab) administration showed anti-leukemia effects in hematologic malignancies [62][63][64]. Previous studies have demonstrated that anti-NKG2A mAb can induce NKG2A + NK cell killing activity against HLA-E-expressing leukemia cells in vitro and in vivo [62][63][64]. Additionally, reduced numbers of NKG2A + NK cells after allo-HSCT are associated with the occurrence of severe GVHD [65,66]. Moreover, NKG2A + NK cells inhibited T-cell proliferation and activation and might prevent GVHD [66]. Therefore, NKG2A + NK cells may play a crucial role in GVHD and GVL effect following allo-HSCT, and monalizumab administration may have a promising clinical role following allo-HSCT.
TIGIT is expressed by both T and NK cells, and its ligands are the poliovirus receptors PVR (also known as CD155) and PVRL2 (also known as CD112) [30]. PVR is overexpressed in several tumors including AML, and its overexpression has been linked to a poor prognosis in AML [35,67,68]. Meanwhile, AML patients with higher TIGIT expression after allo-HSCT had an inferior prognosis [35]. Previous studies have shown that blockade of TIGIT could prevent the exhaustion of NK cells and enhance NK cell-dependent antitumor effect [20,69]. Additionally, CMV-induced adaptive NK cells had less TIGIT expression compared to conventional NK cells, and overcame myeloid-derived suppressor cell (MDSC)-mediated immune suppression. The cytotoxic function of NK cells co-cultured with MDSCs against tumor cells could be restored by blockade of TIGIT [70]. Unlike PD-1 and CTLA-4 inhibitors, NK cells may play a critical role in TIGIT-based immunotherapy, and blockade of TIGIT may have therapeutic effects in GVL by controlling NK-cell activity after allo-HSCT. However, because blockade of TIGIT can also promote T-cell activity [71], it is imperative to optimize the clinical setting to prevent severe GVHD and intensify the GVL effect in AML patients undergoing allo-HSCT.
TIM-3 is expressed on all mature CD56 dim CD16 + NK cells and activated immature CD56 bright CD16 − NK cells. Its ligand is galectin-9 [72], which induces interferon-gamma (IFN-γ) production by NK cells [73]. TIM-3 blockade restores NK cell exhaustion and leads to an increased NK cell cytotoxicity in several cancers [74,75], whereas the TIM-3 antibody agonist leads to a decrease in the cytotoxicity of NK cells [72]. TIM-3 blockade reduces NK cell-mediated killing of pancreatic cancer cell lines [76]. Although a phase 1 study evaluating the blockade of TIM-3 (TSR-022) in advanced solid tumors is in progress (NCT02817633), further studies will be needed to determine the precise role of TIM-3 in AML after allo-HSCT.
LAG-3 (also known as CD223) is a ligand which has been identified as MHC class II. It is widely expressed not only on activated T and NK cells but also on dendritic and B-cells ( Figure 5) [75,77,78]. LAG-3 is involved in inhibiting T-cell effector function, and blockade of LAG-3 promotes T-cell proliferation in vitro [79]. However, the function of LAG-3 on NK cells remains unclear. Blockade of LAG-3 had no effect on NK-cell-mediated cytotoxicity [80]. Further investigation on the role of LAG-3 on NK cells is necessary.
BTLA (also known as CD272), which belongs to the immunoglobulin superfamily, is expressed by most lymphocytes (Figure 5). BTLA acts as a negative modulator of immune responses regulating T-cell activation and proliferation [77,81,82]. Its ligand, herpesvirus entry mediator (HVEM, also known as TNFSF14), is expressed in several tumor cells [82,83]. Blocking BTLA-HVEM interaction leads to a decrease in suppressor T-cells in the tumor microenvironment and enhances antitumor immunity [84]. Additionally, BTLA blockade promotes an increase in NKT-cells and expression of cytotoxic marker genes [85]. However, the functional role of BTLA on NK cells is controversial and requires further investigation.

Bi/Trispecific Engagers and Chimeric Antigen Receptors (CAR) NK Cells
Bispecific and trispecific killer cell engagers (BiKEs and TriKEs), which are composed of a single-chain variable fragment (scFv) containing a variable heavy and variable light chain of an antibody, can specifically target both CD16 expressed on NK cells and tumor antigens (Figure 2). Previous studies have shown that NK-cell-mediated cytotoxicity could occur by CD16 × CD33 (1633) BiKEs that ligated CD16 on NK cells and CD33 on tumor cells, including myelodysplastic syndromes (MDS) and AML [86,87]. Recently, 161533 TriKE, which is an NK-cell stimulatory cytokine with IL-15 added onto BiKE, has been found to restore NK cell proliferation and function through a low expression of TIGIT in NK cells. It is also able to enhance NK-cell-mediated cytotoxicity against MDS cells more than 1633 BiKE [88]. Moreover, NK cells treated with 161533 TriKE can overcome immune suppression mediated by MDSCs. Although IL-15 also stimulates cytotoxic T-cells, 161533 TriKE induces the proliferation of NK cells with minimal effect on T-cells [87]. Therefore, the administration of this agent after allo-HSCT may be a potentially promising treatment to decrease relapse of AML after allo-HSCT with less T-cell-mediated GVHD.
Chimeric antigen receptors (CARs) consist of scFv (extracellular domains) combined with CD3ζ, DAP10, or DAP12 as intracellular signal domains, and CD28, 4-1BB (also known as CD137), and 2B4 (also known as CD244) as costimulatory domains (Figure 2) [89][90][91][92][93]. In a murine allogeneic transplant model using donor-derived CD19-CAR T-cells, allogeneic CAR T-cells eliminated acute lymphoblastic leukemia [94]. However, its administration caused lethal GVHD. Additionally, CD123-redirected T-cells (CART123) eliminated AML and also eradicated normal hematopoietic stem cells (HSCs) in a mouse model because CD123 is highly expressed in HSC [95]. In a phase 1 clinical trial of CD33-CAR NK cells for relapsed and refractory AML patients (NCT02944162), the administration of CD33-CAR NK cells was not clinically efficacious [92]. Recently, cord blood-derived NK cells with CAR-CD19, IL-15, and inducible caspase-9-based suicide gene (iC9) (iC9/CAR.19/IL15-transduced CB-NK cells) enhanced their cytotoxicity against CD19-expressing tumors in a murine model [96]. CAR NK cells may provide a cost-effective treatment with a reduced risk of GVHD compared to CAR T-cells, but further clinical studies will be needed to demonstrate the safety and efficacy of CAR NK cells against AML following allo-HSCT.
The incidence of HLA loss following allo-HCT is one of the major immune escape mechanisms that lead to relapse in AML, and may account for approximately one third of all relapses [107]. Because loss of mismatched HLA through copy-neutral loss of heterozygosity results in the elimination of the incompatible HLA alleles while keeping the expression of HLA class I molecules, cytotoxic killing by NK cells does not occur. Also, recent studies have shown that the downregulation of HLA class II molecules (HLA-DPA1, HLA-DPB1, HLA-DQB1, and HLA-DRB1) and their related molecules (CIITA, IFI30, HLA-DMA, HLA-DMB, and CD74) could allow leukemia relapse after allo-HSCT [108].
Patients who exhibited a high expression of CD200, CD47, PD-L1, PVR, or PVRL2, which is associated with an immune response or immune checkpoints, had a poor prognosis [35,67,[109][110][111]. In addition to PD-1, exhausted T-cells, including exhausted CD8 + T-cells, express inhibitory receptors such as CTLA-4, LAG-3, and TIM-3 [21,112,113]. T-cell exhaustion contributes to AML relapse after allo-HSCT [113]. In contrast to T-cell exhaustion, expression of activating NKG2D ligands such as MHC class I-related chain A (MICA) and UL16-binding protein 1 (ULBP1) on AML cells at diagnosis is associated with an improved OS and a reduced incidence of relapse [114]. Activated NKG2D on NK cells recognizes NKG2D ligands (MICA/B and ULBPs) and enables the induction of NK-cell-mediated cytotoxicity on AML cells [115,116]. AML cells which express low levels of NKG2D ligands are able to evade immune surveillance by NK cells [115].
Janus kinase (JAK) mutations affect the interferon (IFN) signaling pathway by inducing an increase in STAT1 expression, the loss of beta-2-microglobulin, which can detect HLA class I antigen processing, and the loss of PTEN, which increases the production of immunosuppressive cytokines such as VEGF and can increase STAT3 expression. These mutations, which represent various mechanisms of resistance to ICIs, have been reported in several cancers [117][118][119][120], but it remains unknown in the case of AML. These studies demonstrated the association between T-cell activity and resistance to ICIs. However, the mechanisms of resistance to ICIs on NK cells are less well explored and require further elucidation.

Conclusions
Currently, there are numerous NK cell-based immunotherapies for AML post allo-HSCT that have been incorporated into pre-clinical and clinical trials. We have described some clinical trials associated with NK cell-based immunotherapies (Table 1). NK cell immunotherapies such as adoptive NK cells, cytokine-based therapies, ICIs, and bi/trispecific engagers have the potential to significantly enhance conventional therapies for the elimination of AML after allo-HSCT. In the future, combinations of these approaches require to be optimized to further enforce donor NK-cell mediated GVL in AML patients who received allo-HSCT. Moreover, for the next generation of NK-cell immunotherapies, therapeutic approaches based on CAR-engineered NK cells, memory-like NK cells, NKT-cells, and induced pluripotent stem cell-derived NK cells may be considered in a future study [49,89,[121][122][123][124]. However, particularly with the use of ICIs, especially PD-1 or CTLA-4 blockade, after allo-HSCT, there is a distinct need for caution due to the risk of GVHD-related mortality. These ICIs appear to promote T-cell activity more than NK-cell activity against leukemia cells. Well-designed clinical trials should be required to demonstrate the safety and efficacy of these therapies. In addition, because NK cells have a short lifespan compared to T-cells, further improvements in manufacturing and expansion techniques are needed. Previously, immunotherapies had primarily focused on T-cell-mediated cytotoxicity. Thus, some mechanisms for NK-cell immunotherapies, including immune escape of AML, remain unclear. Further studies will be needed to predict which type of AML after allo-HSCT will be affected by NK-cell immunotherapies. The collection of large patient series and datasets will allow the investigation of the various factors which may potentially influence NK-cell immune responses. These factors will include the expression genes, mutations, alterations of resistance to immunotherapies in leukemia cells, tumor microenvironment consisting of Tregs, tumor-associated macrophages, MDSCs, and the production of cytokines.