Clinical evaluation of cellular immunotherapy in acute myeloid leukaemia

Abstract

Immunotherapy is currently under active investigation as an adjuvant therapy to improve the overall survival of patients with acute myeloid leukaemia (AML) by eliminating residual leukaemic cells following standard therapy. The graft-versus-leukaemia effect observed following allogeneic haematopoietic stem cell transplantation has already demonstrated the significant role of immune cells in controlling AML, paving the way to further exploitation of this effect in optimized immunotherapy protocols. In this review, we discuss the current state of cellular immunotherapy as adjuvant therapy for AML, with a particular focus on new strategies and recently published results of preclinical and clinical studies. Therapeutic vaccines that are being tested in AML include whole tumour cells as an autologous source of multiple leukaemia-associated antigens (LAA) and autologous dendritic cells loaded with LAA as effective antigen-presenting cells. Furthermore, adoptive transfer of cytotoxic T cells or natural killer cells is under active investigation. Results from phase I and II trials are promising and support further investigation into the potential of cellular immunotherapeutic strategies to prevent or fight relapse in AML patients.

Introduction

The prognosis for patients with acute myeloid leukaemia (AML) has been steadily improving, in particular for younger patients with an overall survival rate of 60% [1]. However, relapses still occur, and for older patients (>60 years), the outlook remains dismal with less than 10% attaining long-term survival (estimated survival is only 20% at 2 years) [2, 3].

The advent of haematopoietic stem cell transplantation (HSCT) and donor lymphocyte infusions (DLI) has led to a shift in the paradigm of AML therapy [4, 5]. HSCT offers the advantage of dose intensification, administering high-dose chemotherapy with or without total body irradiation, in an attempt to further reduce tumour burden [6]. Furthermore, non-autologous HSCT can result in an immunologic graft-versus-leukaemia (GVL) effect, where leukaemic cells are destroyed by allogeneic transplant cells [7]. However, HSCT is associated with a significant treatment-related mortality and morbidity, including graft-versus-host-disease (GVHD) and complications such as veno-occlusive disease and infections. HSCT is also associated with an elevated risk of developing secondary malignancies such as solid cancers [6, 7].

More recently, treatment has shifted to the use of non-myeloablative and reduced-intensity conditioning regimens, which utilize administration of chemotherapy and/or radiation at reduced doses, sufficient to allow donor stem cell engraftment and to enable eradication of residual leukaemia cells by allogeneic effector immune cells within the graft [8]. Patients are frequently given DLI after treatment to boost this GVL effect. Non-myeloablative transplants have allowed extension of the upper age limit for allogeneic transplantation and reduced toxicity with the use of less intensive conditioning regimens. Unfortunately, this approach is associated with a significant relapse rate and therefore mostly restricted in use to clinical trials for selected patients [8].

The majority of AML patients undergoing HSCT receive donor stem cells from siblings or unrelated donors matched for MHC molecules [6]. When transplantation is performed with a mismatched family member or unrelated donors, detrimental alloreactive responses towards the mismatched MHC molecules can occur, resulting in GVHD. Even with complete matching for MHC molecules, T cell reactivity to minor histocompatibility antigens (mHAgs) due to polymorphisms in common cellular proteins can occur, which also leads to GVHD. In order to reduce the severity of GVHD and its mortality and morbidity, grafts can be manipulated to be depleted of T cells prior to transplantation [9]. This approach suffers from increased relapse rates due to the loss of the beneficial associated GVL effect, and patients continue to die from their disease. Therefore, the need for alternative therapies remains acute to fight relapse. Targeted anti-leukaemia immunotherapy would offer the desired combination of GVL effects with minimal toxicity due to the preferential targeting of AML cells. Recent findings of the group of Fuchs open the gate to a broader use of immunotherapy for patients with advanced haematological malignancies. The authors observed that short-course high-dose cyclophosphamide administered post-transplantation was effective in preventing acute and chronic GVHD in patients after HLA-matched or partially mismatched allo-HSCT [10–12]. In this way, immunosuppressive treatment could potentially be limited after transplantation, paving the way to use immunotherapy strategies early after transplantation, a train of thought that is also followed by others [13]. In this review, we describe preclinical and clinical studies testing cellular immunotherapies for AML, with special focus on recent new developments in this area. Other clinical trials comprising interleukin (IL)-2 administration, monoclonal antibody therapy and peptide-based vaccination have already extensively been reviewed elsewhere [14–18].

Cellular therapy to treat AML

In the last decade, research efforts have been broadening the array of cell therapy approaches which can be used as adjuvant therapy in AML. Cellular therapy comprises the administration of therapeutic cellular vaccines (tumour or DC vaccines) to induce specific immune responses in vivo, as well as adoptive transfer of effector cells to exert anti-tumour immune reactivity (T cells or NK cells) in patients.

NK cells can sense the altered expression of MHC molecules and stress markers on malignant cells through the expression of different NK cell receptors [19]. Furthermore, abnormal expression of normal cellular proteins or expression of mutated proteins in malignant cells give rise to ‘tumour proteins’, fragments of which can be presented on the cell surface, called tumour-associated antigens (TAA) [20]. By binding of TAA to its T cell receptor (TCR), an activated effector T cell is able to recognize and destroy malignant cells expressing these TAA.

It is generally believed that the presence of a large tumour load would overrule the capacity of the immune system to eradicate the malignant cells. Therefore, cellular immunotherapy is particularly being examined in AML patients as an adjuvant therapy following standard therapy to destroy residual leukaemic cells [14, 21]. Ideally, AML immunotherapy would induce an immune response that clears minimal residual disease (MRD), while sparing normal tissue, and would generate immunological memory that protects against disease recurrence. After chemotherapy, T cells are reduced in AML patients [22]. Lymphocyte counts in AML patients during and within 1 month after induction chemotherapy are inversely correlated with relapse rate and overall survival [23, 24], pointing to the importance of immune surveillance, and more specifically lymphocytes, in preventing relapse [16]. Results of active immunization strategies (described below) support the hypothesis that the immune system of AML patients in remission can be boosted to attack AML cells. Examples of high-potential leukaemia-associated antigens (LAAs) are Wilms’ tumour protein (WT1), proteinase 3 (PR3) and the receptor for hyaluronic acid-mediated motility (RHAMM/CD168). The immunogenicity of WT1, PR3 and RHAMM has been shown by the presence of TAA-specific cytotoxic T lymphocytes (CTL) in AML patients [25–27]. Furthermore, in vitro generated or ex vivo isolated T cells specific for WT1, PR3 or RHAMM were able to kill primary myeloid leukaemic cells [26–30]. An in-depth description of LAAs relevant to AML has been published elsewhere [31–37].

Whole tumour cells

Autologous modified whole tumour cells are attractive accessory cells for the in vivo or in vitro stimulation of AML-specific immune responses directed against multiple antigens, thereby decreasing the risk of generating escape mutants. Furthermore, random mutations in tumour cells can generate unique antigens in each individual, providing a rationale for customized immunotherapy approaches. A disadvantage of whole tumour cell approaches is that along with TAA expression, they also contain a wide range of normal self-proteins not linked to the malignant phenotype. Generally, tolerance to self-antigens is tightly controlled, but the aim of cancer immunotherapy is to break tolerance to TAAs and therefore, the danger of inducing autoimmunity to normal tissues should not be discounted [38]. It is hoped that as tumour cells express self-antigens considered as TAA to a higher extent compared to normal tissues, they will be more susceptible to TAA-specific CTL recognition. Focusing on WT1 as a TAA, WT1-specific CTL could possibly also recognize normal hematopoietic progenitor cells expressing WT1 in the bone marrow, causing leucocytopaenia. Until now, leucocytopaenia was only detected in patients with myelodysplastic syndrome (MDS) following peptide vaccination and not in AML [39, 40]. In MDS, the hematopoietic stem cells are transformed and so it is believed that the observed leucocytopaenia is an indicator of the effectiveness of the vaccine [39].

In AML, expression of MHC molecules on primary tumour cells allows them to officiate as accessory cells [41], presenting a range of patient-specific TAA to T cells. However, their antigen-presenting potential may be hampered by insufficient expression of costimulatory molecules, high expression of apoptosis-inducing molecules and/or constitutive secretion of immunosuppressive cytokines [42]. Not only the antigen-presenting capacity, but also the susceptibility of AML cells for immune-mediated cell death is affected by these factors, as well as by heterogenous constitutive chemokine secretion and chemokine responsiveness of primary AML cells [43, 44]. In order to make therapy outcome more predictable, heterogeneity of the expression of immunomodulatory and apoptosis-related molecules by primary AML cells should be reduced (e.g. by genetic modification, vide infra) or should be taken into account to define subsets of patients that are likely to benefit from immunotherapy. Another important parameter to reduce heterogeneity is the choice of culture medium, because it influences the constitutive cytokine secretion of in vitro cultured primary AML cells [45, 46]. It was shown that these cells can be cultured in standardized serum-free conditions [45, 46], thereby excluding the risk of adverse immune reactions against serum-associated molecules.

In order to avoid transfusion of viable, proliferating AML cells when using them as accessory cells, primary AML cells can be γ-irradiated. Bruserud and Ulvestad [47] investigated whether γ-irradiated AML cells could function as accessory cells during anti-CD3-stimulated T cell activation in vitro. The authors showed an irradiation dose-dependent reduction in the secretion of the proinflammatory cytokines IL-1β, IL-6 and TNF-α by primary AML cells. Although highly proliferative T cell responses were not always associated with high levels of cytokine secretion, an irradiation dose-dependent reduction in anti-CD3-stimulated T cell proliferation was observed. A dose of 50 Gy was considered by the authors to be both safe and to conserve the accessory cell function of primary AML cells.

In one of the earliest vaccination trials using autologous whole AML cells, primary cells were frozen in the presence of autologous serum and gamma-irradiated with 100 Gy after thawing [48]. Four AML patients were included in the clinical trial following achievement of chemotherapy-induced remission, receiving intradermal and subcutaneous injections in all four limbs. In vitro analysis of peripheral blood effector cells collected from the patients pre- and post-vaccination showed that immunotherapy increased the activity of patient-derived lymphocytes against autologous leukaemic cells [48]. Later, the protocol was adjusted by co-injecting irradiated autologous AML cells with Bacillus Calmette-Guerin (BCG), resulting in prolonged remission and longer survival after first relapse, compared to non-vaccinated AML patients [49], indicating a clinical effect of the vaccine. No data are available to allow correlation of clinical effects to specific characteristics of the AML cells. A positive outcome was also reported in the study of Zhang et al. [50], where autologous whole cell vaccines comprising Mitomycin C-treated autologous AML cells mixed with IL-2, GM-CSF and IL-6 were administered subcutaneously in the limb of 25 AML patients with relapsed or refractory disease. Partial responses were observed in five patients and CRs seen in four patients were associated with a reduction in serum IL-10 post-vaccination. Efficacy of the vaccine was associated with expression of CD80, which was exclusively expressed on AML cells of some patients with AML FAB M4 and M5.

Genetic modification can be used to reduce heterogeneity and to increase immunogenicity of tumour cells from a wide range of patients by conferring expression of immunostimulatory molecules, including cytokines and costimulatory molecules [51]. Early mouse in vivo studies on the transfer of cytokine or costimulatory genes into murine AML cells or human leukaemia cell lines demonstrated reduced tumourigenicity of the transfected cells [52–55]. Furthermore, vaccinated mice developed protective immunity against rechallenge with wild-type cells and MRD could be eliminated [53–56].

Building on these promising results, several gene transfer methods were tested to modify human primary AML cells, trying to overcome variable levels and/or low persistence of transgene expression. These methods include plasmid transfection [57], nucleofection [58] and transduction with vectors from retroviruses [59, 60], adenoviruses [60–63], herpes simplex viruses [64] and/or adeno-associated viruses [65, 66]. Despite technical obstacles, expression of immunostimulatory molecules by primary AML cells consistently resulted in enhanced T cell stimulation in vitro, mostly shown in allogeneic assays [14]. An efficient method for the genetic modification of human primary AML cells is the use of lentiviral vectors [67–69]. In contrast to classical retroviral vectors, HIV-1-based lentiviral vectors readily transduce mitotic as well as a number of post-mitotic targets [70]. As lentiviral vectors do not encode viral proteins [71], they are suitable for immune gene therapy strategies. A report on the modification of human primary AML cells using lentiviral vectors to express B7.1 and GM-CSF in primary AML cells showed promising levels of transgene expression, but rather low proliferation of autologous T cells [67]. Studies on a fusagene lentiviral vector [72] encoding CD80 and IL-2 resulted in high levels of autologous T cell stimulation, NK cell activation and lysis of targeted AML cells [73–75]. A phase I clinical trial has now been started in poor prognosis AML patients using autologous whole cell vaccines modified with the fusagene lentiviral vector to express CD80 and IL-2 [74]. Furthermore, a phase II clinical study was recently conducted using whole cell vaccines consisting of gamma-irradiated autologous AML cells mixed with K562 cells secreting GM-CSF in pre- and post-transplantation immunotherapy protocols [76]. Tumour-specific immune responses were detected, as well as reductions in MRD as measured by WT1 transcript levels [76]. These results are promising, despite the controversy on the use of whole tumour cell vaccines and GM-CSF as an immune-adjuvant since the halted phase III trial on GM-CSF-secreting allogeneic cells in prostate cancer [77].

In addition to gene transfer, Smits et al. [78] showed that the immunogenicity of AML cells can also be increased by transfecting AML cells with the Toll-like receptor (TLR)3 ligand polyinosinic polycytidylic acid [poly(I:C)]. Toll-like receptor ligands are pathogen-associated molecular patterns, small motifs that are conserved within a group of micro-organisms and can activate immune cells [79]. By loading cells with the non-coding synthetic dsRNA poly(I:C), viral infection was mimicked. The pleiotropic response of primary AML cells and AML cell lines to poly(I:C) transfection included apoptosis, increased expression of costimulatory and MHC molecules, and production of proinflammatory cytokines. Strikingly, in response to electroporation with poly(I:C) all AML cell samples produced type I interferons, known for their anti-tumoural effects [80, 81]. Focusing on dendritic cells (DCs), coculture with apoptotic poly(I:C)-loaded AML cells induced maturation and enhanced their Th1-inducing capacity, which is important for tumour immunotherapy [78]. Furthermore, poly(I:C)-loaded AML cells could activate both the helper and the cytotoxic function of NK cells, resulting in NK cell-derived IFN-γ and killing of AML cells, respectively [82, 83]. Also, the TLR7/8 ligand resiquimod increased the immunogenicity of AML cells and activated IFN-γ secretion by NK cells [84], making TLR ligands attractive molecules to be incorporated in future AML immunotherapy strategies.

If an autologous AML vaccine is used to activate T cells, an important aspect to consider is the in vivo attraction of T cells by the vaccine, as well as recruitment of T cells to the AML compartment. Bruserud et al. [44, 85] showed that only a minority of patients had diminished in vitro chemotaxis of normal lymphocytes towards primary AML cells due to absence of chemokine release by the AML cells. In the broad constitutive chemokine release profile of the other primary AML cells, three chemokine clusters could be identified. Although samples from all clusters had a similar ability to attract normal T cells in vitro (predominantly CD4+ T cells), it is expected that different chemokine release profiles will contribute to the heterogeneity in the outcome of antileukaemic immunotherapy, since there was a significant variation between patients in intravascular and extravascular levels of the T cell chemotactic chemokines CXCL10, CCL5 and CCL17 [43].

The influence of irradiation and genetic modification on chemokine release profiles requires further investigation. Depending on which protocol was used, cytokine-mediated in vitro induction of a DC phenotype in primary human AML cells was associated with a significant increase in CCL17 and CCL22 release, pivotal for chemotaxis of normal T cells [86]. These data provide evidence that modification of primary AML cells influences the secretion of T cell attracting chemokines.

Failure of immune responses to eliminate leukaemic cells can be due to T cell dysfunction and to increased suppressor activity mediated by regulatory T (Treg) cells. Le Dieu et al. [87] showed that T cells from newly diagnosed AML patients had an abnormal genotype and a decreased ability to form functional immune synapses with AML cells. Further research is required to investigate if this is also the case for AML patients in remission as primary target group for immunotherapy. Elevated levels of Treg cells with high levels of suppressor activity were detected both in AML patients that were newly diagnosed and in patients in complete remission [88]. These factors can potentially block positive outcome of vaccination, either using modified tumour cells or DCs.

Dendritic cells

DCs are immune cells located in many normal tissues. Being professional antigen-presenting cells, they are specially adapted to direct the immune response to induce immune activation or tolerance [89]. They are known to be the most potent stimulators of naïve and resting T cells and therefore are promising vehicles for the in vivo delivery of tumour antigens (in the form of a ‘DC vaccine’) or through their in vitro generation of TAA-specific T cells [90, 91]. In order to obtain sufficient DCs for immunotherapeutic purposes, DCs are often cultured in vitro from a starting population of positively selected CD14 monocytes isolated from peripheral blood. The cells are differentiated under the influence of cytokines into ‘monocyte-derived DC’ [92–94]. These immature DC are then amenable to loading with TAA. In AML, another attractive strategy is to differentiate the myeloid leukaemia cells themselves into AML-derived DC to circumvent the need for loading with TAAs [95, 96]. Only a few phase I clinical trials have been performed to date using AML-DC and have reported that the DC vaccine is well tolerated and able to induce immunological responses albeit with limited feasibility in some AML patients in terms of obtaining sufficient AML-DC vaccines and so far without evidence of clinical benefit [97–100]. Currently, protocols are being optimized in order to increase the success rate of generating AML-DC [101, 102] and to increase the cytolytic capacity of T cells primed with AML-DC [103, 104].

For monocyte- or bone marrow-derived DC, two key issues are choice of TAA and method of TAA loading. Until now, preclinical and clinical studies have described loading of human DC with PML-RARα, survivin or WT1 using peptide pulsing or mRNA electroporation [105–107]. Prior identification of LAA is not required when DC are fed with AML cell lysates and/or apoptotic AML cells, or when AML cell-DC fusion hybrids are generated [108–114]. Irrespective of the antigen loading method, these studies have reported DC to be potent activators of autologous or allogeneic HLA-matched CTL in vitro, resulting in cytotoxic responses directed against primary AML cells and providing a rationale for clinical evaluation of DC vaccines in AML.

The primary focus of AML immunotherapy remains the prevention of relapse, and an attractive strategy is to take monocytes from AML patients in remission to generate monocyte-derived DC. Clinical-grade DC with the desired phenotype could be generated in sufficient numbers when starting with monocytes from AML patients in remission [115]. In a recent phase I/II study, DC were loaded with the LAA WT1 by mRNA electroporation. The DC vaccine was well tolerated, and five out of ten AML patients obtained molecular remission following administration of autologous WT1-loaded DC [107]. Furthermore, two of the patients in partial remission achieved CR instead of the expected relapse [107], supporting further clinical evaluation of this DC vaccine. In a phase I clinical study by Lee et al. [112], autologous monocyte-derived DC vaccines were well tolerated in two AML patients in relapse. The DC were loaded with LAA using AML lysates. Despite the detection of immunological responses, no clinical responses were observed, in terms of a lowered percentage of AML cells in the bone marrow [112], pointing to the importance of the choice of antigen and/or time of administration.

Adoptive transfer of T cells

Methods of immunotherapy described so far can be classed as active immunization or vaccination, where the introduction of immunogenic material aims to elicit activation and expansion of effector cells in the endogenous immune repertoire. A disadvantage of this approach is the requirement of the patient for some degree of immune competence. An alternative is passive immunization or adoptive transfer, where transfer of functional immune effector cells confers immunity and could possibly overcome lymphocyte deficiencies in AML patients [116–121].

DLIs comprise the transfer of donor lymphocytes to leukaemia patients in relapse following HSCT [5]. Although DLIs have been successful in the treatment of relapsed CML [122], the induced remission in AML or accelerated phase CML patients has not been as long-lasting, probably due to tumour growth exceeding the capabilities of the immune response [123–125]. Adoptively transferred T cells in DLI can exert anti-leukaemia responses, resulting in a so-called GVL effect. However, GVHD can also be initiated if the allogeneic T cells react against non-leukaemic cells, thereby causing damage to normal tissues [125, 126]. In order to reduce GVHD and to increase the GVL effect, donor lymphocytes can first be manipulated ex vivo to switch off alloreactive effector T cells and to enrich leukaemia-specific T cells. Anergy can be induced in donor alloreactive T cells by costimulatory molecule blockade during coculture with host peripheral blood mononuclear cells [127–130]. In order to avoid possible bystander effects or reversion of anergy in vivo, selective allodepletion is being examined. Donor T cells are cocultured with host antigen-presenting cells, followed by elimination of activated T cells using antibodies targeting activation-induced markers such as CD25, CD69, CD71, CD95, CD134, CD137 and/or HLA-DR [131–142]. Results of clinical trials using allodepletion point out that this promising approach needs optimization in order to further reduce transplant-related morbidities [136, 138].

In a phase I/II trial, enrichment of leukaemia-specific donor T cells was performed ex vivo using leukaemia-derived host antigen-presenting cells. These T cells were administered to fight AML relapse after allo-HSCT as alternative treatment for DLI [143]. In one AML patient, stable disease was detected 8 weeks after T cell infusion, while two other patients had no clinical response. The authors concluded that further optimization was required, because the procedure to generate leukaemia-reactive T cells took so long to move from bench to bedside that leukaemia patients with rapidly expanding disease often succumbed before therapy administration was possible [143]. In a study by the group of Quesenberry [144], a more rapid protocol was developed for refractory AML patients, where haploidentical T cells were administered following a minimal preparative treatment of 100 cGy (nonengrafting haploidentical transplantation with minimal myeloablation). In the group of 13 AML patients, 5 durable complete responses and 4 partial responses were seen, showing that long-term response can be found in transient chimerism. Before infusion, cells were primed with granulocyte colony-stimulating factor (G-CSF) to minimize GVHD by polarizing T cells to a Th2 phenotype.

The generation of CD4+ or CD8+ T cells with anti-leukaemia reactivity for adoptive transfer in AML is also scrutinized by others [28, 29, 145–156]. Until now, clinical efficacy of adoptively transferred T cells has been shown mainly in the allograft setting [157]. Nevertheless, transfer of autologous T cells is an interesting concept to separate GVL reactions from GVHD. Currently, genetic engineering of T cells (described in depth by Vera et al. [158]) is being investigated in an attempt to overcome the difficulty of generating high numbers of leukaemia-reactive autologous T cells in vitro. A promising strategy is the isolation of genes encoding α and β T cell receptor (TCR) chains from LAA-specific T cell clones with high avidity, using them to redirect the specificity of primary T cells. Both WT1-specific and HA-1/2-specific TCR genes have been actively investigated for their anti-leukaemia reactivity in preclinical experiments following transduction of T cells. Explorative clinical trials are now being planned [157, 159]. Toxicity of this approach will depend on distribution of the target antigen in normal tissues. For WT1, specific attention should therefore be paid to the hematopoietic system.

Also, chimeric antigen receptors (CARs) can redirect T cell specificity. By transducing T cells with CARs, antigen specificity of an antibody (single-chain variable fragment) is coupled to the lytic capacity of T cells (transmembrane and intracytoplasmatic structure of CD3ζ) [158]. Advantages are the MHC independency and the extended range of tumour surface antigens that can be recognized by T cells. If CARs are linked to endodomains of the costimulatory molecules 4-1BB and CD28, T cells can obtain a higher activation status. For AML, CAR targeting the Lewis Y antigen had promising effects in preclinical experiments, paving the way for a phase I clinical trial [160]. In vivo toxicity of this approach, related to the potential of CAR-modified T cells to damage normal tissues, is a specific concern that was recently fed by reports on cholestase development (renal cell carcinoma patient [161]) and renal and respiratory failure (B-cell chronic lymphocytic leukaemia patient [162]).

Important characteristics of T cells that determine the success of adoptive transfer are the capacities to survive in vivo and to home to the sites with residual leukaemic cells. Although not all signals that attract T cells to cancer cell sites have been characterized yet, it is clear that expression of adhesion molecules and chemokine receptors on T cells will determine their capacity to migrate. Important chemokines for T cell attraction that can be secreted by AML cells are CXCL10, CCL5, CCL17 and CCL22 [43, 44, 85, 86]. Also, marrow stromal cells can contribute by secreting SDF1 to attract T cells that express the chemokine receptor CXCR4 [163]. In order to attract T cells to the bone marrow, the chemotactic gradient that is determined by the difference between chemokine concentrations in the bone marrow and in the serum must be in favour of the bone marrow [85]. In this regard, Olsnes et al. [43] showed that serum levels of CCL5, CCL17 and CXCL10 vary between AML patients and that interaction between AML cells and the stromal cells can raise the release of CCL17 and CXCL10, supporting the attraction of T cells to the AML sites. Homing capacities could be augmented by genetic transfer of chemokine receptors into T cells before adoptive transfer [158, 164].

Genetic engineering could also be used to increase the survival and functionality of adoptively transferred T cells by codelivery of genes encoding cytokines, costimulatory molecules and/or telomerase reverse transcriptase [158]. On the other hand, suicide genes could be introduced if transferred T cells can potentially harm normal tissues. Examples of suicide genes include thymidine kinase from herpes simplex virus I and Fas [158].

Adoptive transfer of NK cells

NK cells have the ability to kill autologous and allogeneic tumour cells [165, 166]. NK cell-mediated alloreactivity is mediated by killer immunoglobulin-like receptors (KIR), expressed by NK cells and recognizing subgroups of HLA class I molecules. Mismatches in binding of donor KIR with recipient HLA class I molecules triggers donor NK cells to kill recipient mismatched cells [167]. This mechanism has been shown to contribute substantially to GVL effects of HSCT with HLA-haplo-identical transplants from siblings. In this way, NK cell-mediated alloreactive responses can reduce relapse risk, while enhancing engraftment and protecting against GVHD, resulting in improved outcome for AML patients [165, 168–170].

Aiming at increasing the clinical efficacy of allo-reactive NK cells, research is now focusing on the expansion of NK cells in vivo and ex vivo [165]. In a phase I study, enriched NK cell fractions from haplo-identical donors were transferred after activation with IL-2 and expanded in vivo following an intensive preparative regimen, associated with high levels of IL-15. Three out of four AML patients achieved CR following KIR ligand-mismatched NK cell transplantation [171]. In a recent phase I study, haplo-identical KIR-mismatched NK cells were transplanted in ten children with AML in combination with IL-2 following low-dose immunosuppression [172]. The purified cell product was injected without ex vivo expansion or activation. Significant in vivo expansion of NK cells could be detected, and all ten patients remained in remission during follow-up which ranged from 18 to 39 months.

By expanding NK cells ex vivo, AML patients can receive transplantations with high numbers of purified NK cell fractions. Currently, protocols are being optimized for ex vivo expansion of NK cells able to kill leukaemic cells [173, 174] and phase I clinical trials are ongoing [173, 175]. An initial trial involving one child with AML showed feasibility of NK cell isolation and expansion for clinical applications [175].

It has not been specified yet which chemokines are involved in the attraction of adoptively transferred NK cells to the bone marrow of AML patients. Nevertheless, NK cells express several chemokine receptors, so different chemokines might be involved [176–178]. Immunocompetent cells, such as T cells and NK cells, might secrete factors that contribute to cancer growth and/or angiogenesis [177]. Furthermore, investigators should be aware of possible toxic effects of adoptive transfer, e.g. on hematopoiesis. Mesenchyal stromal cells (MSC) are important in the support of hematopoiesis and infusion of MSC is currently being investigated to aid hematopoietic recovery following allo-HSCT [179, 180]. Götherström et al. [181] recently showed that human IL-2-activated NK cells could kill foetal and adult MSC, as well as autologous MSC, so caution will be needed when transferring activated NK cells. On the other hand, the importance of NK cells in controlling AML is illustrated by the fact that relapse in AML patients following allo-HSCT is associated with decreased numbers of NK cells 1 year post-transplant [182]. Therefore, adoptive transfer of NK cells has the potential to become an important future cell therapy strategy to prevent AML relapse.

Conclusions

Knowledge has been gained in recent years on the importance of disease status when applying cancer immunotherapy. Early studies focused on the treatment of advanced disease with a significant tumour bulk. By further understanding the mechanisms of tumour-induced immunosuppression, it is now clear that the immune system may not be able to effectively induce regression of large tumour burdens [183]. Reduced disease loads or immunotherapy treatments in combination with conventional tumour burden-reducing induction therapies may provide more favourable settings to apply therapeutic immunotherapy. In AML, immunotherapy is predominantly investigated to prevent relapse following standard therapy. Ideally, multiple antigenic epitopes and potent adjuvants would be incorporated to induce both innate and adaptive immune responses in order to break AML-mediated immunosuppressive mechanisms and immunological tolerance. Recently, promising progress has been made in the development of cellular immunotherapeutic protocols to prevent or to fight AML relapse.

Therapeutic vaccination using whole tumour cell or DC vaccines and adoptive transfer of NK cells with anti-leukaemic activity have been investigated in phase I or II clinical trials. Positive results associated with the application of these methods include decreased levels of the biomarker WT1, induction of remission (molecular and haematological) and/or increased relapse-free survival. These approaches are now being further evaluated in larger cohorts. Because of the observed GVL effect following DLI and the positive results with adoptively transferred T cells in metastatic melanoma [184], the adoptive transfer of leukaemia-specific T cells is now considered to be worthy of investigation. New and ongoing clinical trials will reveal whether this technique could provide a benefit in AML patients, with genetic engineering of T cells giving a boost to this research area.

Until now, not much attention has been paid to combination strategies. It is important to investigate whether the anti-leukaemic effects of cellular immunotherapeutic approaches could be further increased through the combined administration with immunomodulatory molecules, such as TLR ligands or ipilimumab. Toll-like receptor ligands can activate multiple cells involved in anti-tumour responses simultaneously [185]. Ipilimumab is an antibody currently under active investigation in clinical trials to overcome CTLA-4-mediated immunosuppression [186]. Recent insights have underscored the role of chemokines in the immunoregulatory network formed by immunocompetent cells, bone marrow stromal cells and AML cells [177], making these mediators and their receptors also attractive targets for immunomodulation. Due to the complexity of immunity and the need to overcome anergy to self-antigen-expressing tumour cells, likely personalized therapies and combination strategies are required to further improve the curative potential of cellular therapies, ultimately resulting in improved survival of AML patients.

Abbreviations

AML

Acute myeloid leukaemia

CR

Complete remission

CTL

Cytotoxic T lymphocyte

DC

Dendritic cell

DLI

Donor leucocyte infusion

FAB

French-American-British

GVHD

Graft-versus-host-disease

GVL

Graft-versus-leukaemia

HLA

Human leucocyte antigen

HSCT

Haematopoietic stem cell transplantation

LAA

Leukaemia-associated antigen

LSA

Leukaemia-specific antigen

mHAg

Minor histocompatibility antigen

MHC

Major histocompatibility complex

MRD

Minimal residual disease

NK

Natural killer

poly(I:C)

Polyinosinic polycytidylic acid

PR3

Proteinase 3

RHAMM

Receptor for hyaluronic acid-mediated motility

SSX2IP

Synovial sarcoma X breakpoint 2-interacting protein

TAA

Tumour-associated antigen

TLR

Toll-like receptor

WHO

World Health Organization

WT1

Wilms’ tumour protein