The expanding horizon of immunotherapy in the treatment of malignant disorders: Allogeneic hematopoietic stem cell transplantation and beyond

Abstract

Allogeneic hematopoietic stem cell transplantation (allo-SCT) is a very effective therapeutic modality with curative potential in patients with hematological malignancies. The therapeutic efficacy is mainly based on the alloreactive reaction of donor lymphocytes against malignant cells of the recipient named as ‘graft-versus-leukemia’ or ‘graft-versus-tumor’ (GVL, GVT) effect. However, besides the beneficial GVL effect, alloreactive reaction attacks normal cells and provokes the deleterious ‘graft-versus-host disease’ (GVHD) which represents the major limitation of allo-SCT. Current trials have focused on a dual goal: augmentation of GVL and complete abolishment of GVHD. From a theoretical point of view complete dissociation of GVL from GVHD can occur by selecting antigenic targets present on malignant and absent from normal cells. Hematopoietic tissue-restricted minor histocompatibility antigens and leukemia or tumor-associated antigens are ideal candidates for tumor-targeted immunotherapy. Other options for inducing anti-tumor immunity in the absence of GVHD are natural killer (NK) cell immunotherapy, amplification of immune responses by using monoclonal antibodies, and bispecific T and NK-cell engagers. Genetically modified immune effectors such as T-cells armed with chimeric antigen receptors (CAR) or transduced with T-cell receptors with anti-tumor specificity are another exciting field of immunotherapy against malignancies.

Key words::

• Cellular adoptive immunotherapy
• graft versus host disease
• graft versus leukemia effect
• targeted therapy

Key messages

• The therapeutic potential of allogeneic stem cell transplantation (allo-SCT) is mainly due to an immunological process named graft-versus-leukemia effect (GVL). GVL and graft-versus-host disease (GVHD) represent the two sides of the same coin because donor lymphocytes provoke an immune reaction against host antigens expressed on both normal and malignant tissues.

• Complete dissociation of GVL from GVHD can occur by selecting antigenic targets present on malignant and absent from normal cells. Ideal antigenic targets are: 1) minor histocompatibility antigens (mHAs) with restricted tissue expression and 2) leukemia or tumor-associated antigens (LAAs, TAAs). Other methods aiming at dissociation of GVL from GVHD are: 1) donor lymphocytes transduced with a gene encoding T-cell receptors (TCR) specific for mHA or for LAA/TAA, 2) vaccination against LAAs or TAAs in patients with various malignant disorders, 3) selective allodepletion of host alloreactive cells, 4) infusion of T-cells transduced with suicide genes, and 5) donor-purified NK cell infusion.

• Promising immunotherapeutic approaches outside the setting of allo-SCT are: 1) autologous lymphocytes transduced with a gene encoding TCR specific for LAA or TAA, 2) amplifying anti-tumor immunity by using monoclonal antibodies against inhibitory immunoreceptors such as anti-PD-1 and bispecific T and NK-cell engagers, and 3) T and NK-cells armed with chimeric antigen receptors.

Introduction

Allogeneic hematopoietic stem cell transplantation (allo-SCT) is the most powerful immunotherapeutic approach employed so far in clinical practice. Therapeutic efficacy of allo-SCT is mostly mediated through donor lymphocytes with reactivity against host antigens. The beneficial alloreactivity of donor against malignant cells of the host has been given the name ‘graft-versus-leukemia’ or ‘graft-versus-tumor’ (GVL, GVT) effect. However, the alloreactive effect of the donor against the host is not limited to leukemic cells but can also affect normal epithelial tissues, thus giving rise to the unwanted graft-versus-host disease (GVHD). From this point of view, GVL and GVHD are the two sides of the same coin. Widespread use of allo-SCT is limited due to the serious morbidity and mortality mainly associated with GVHD. Moreover, highly aggressive tumors like chemo-resistant leukemia are not sensitive enough to GVL effect, and the vast majority of these cases usually relapse after allo-SCT. Augmentation of the GVL effect and complete abolishment of GVHD is the goal for future improvements in the field of allo-SCT. Therefore research efforts have been focused on the dissociation of the beneficial GVL from the deleterious GVHD.

Although the GVL effect (observed after allo-SCT) remains the most effective immunotherapy in use, various other immunotherapeutic approaches outside the setting of allo-SCT are in the early phase of clinical development, or have been already introduced in everyday practice. Passive immunotherapy by using monoclonal antibodies (mabs) with specificity against antigenic targets expressed on the surface of malignant cells has shown excellent results as in the case of the anti-CD20 antibody rituximab in the treatment of B-cell lymphoid malignancies. Advances in the field of genetic engineering resulted in the production of genetically modified T-cells armed with chimeric antigen receptors or with T-cell receptors with predefined specificity. A brief overview of these approaches is presented in this review. Although development occurred in the autologous setting we consider that most if not all of these immunotherapeutic strategies will be employed in the near future in the field of allo-SCT.

Donor lymphocyte infusions

Barnes et al. first suggested the existence of GVL effect in 1956 when they noted eradication of leukemia in irradiated mice receiving allo-SCT but not syngeneic transplants (1). In humans evidence for GVL came from studies that indicated lower relapse rates post allo-SCT in patients that developed GVHD. Further indications for GVL were the observation of higher relapse rates: 1) in allogeneic transplants with T-cell-depleted grafts, 2) in syngeneic transplants, 3) with higher doses of cyclosporine (CSA) and longer immunosuppression post allo-SCT (2–4).

These observations set up the rationale for the development of donor lymphocyte infusion (DLI) as an immunotherapeutic approach for treatment and prevention of post allo-SCT relapse in high-risk malignancies. The safety and efficacy of therapeutic DLI (t-DLI) in different diseases is briefly summarized below.

Chronic myeloid leukemia

Kolb et al. first reported the significant efficacy of DLI administered as treatment for post allo-SCT relapse in patients with chronic myeloid leukemia (CML) (5). A retrospective analysis conducted by the European Bone Marrow Transplantation Team (EBMT) examined the efficacy and safety of DLI infusions in a cohort of 298 patients with CML treated in different centers. Patients were categorized in three groups depending on the initial total mononuclear cell (MNC) dose they received. Briefly, group 1 consisted of patients who received an initial TNC dose of less than 2 × 10⁷/kg. Group 2 and 3 consisted of patients who received an initial dose of TNC between 2 × 10⁷/kg and 2 × 10⁸/kg, and above 2 × 10⁸/kg respectively. Additional DLIs were given to 62%, 20%, and 5% of patients in groups 1, 2, and 3, respectively. The incidence of GVHD, of non-relapse mortality (NRM), and of DLI-related mortality was significantly reduced, while overall survival (OS) and disease-free survival (DFS) were significantly prolonged in group 1 patients. Authors concluded that the initial dose of DLI should not exceed 2 × 10⁷/kg in patients with relapsed CML after allo-SCT (6).

Besides the significance of cell dose, other parameters predictive of response to DLI are the tumor load and the phase of disease (7,8): 1) The less the magnitude of disease, the higher the probability of response. Patients in molecular or cytogenetic relapse almost always achieve remission after DLI, while the remission rate is approximately 75% in patients with chronic-phase hematological relapse. 2) The earlier the phase of disease, the higher the probability of response. The percentage of patients in accelerated and blastic phase who achieve remission post DLI is significantly reduced and is estimated at 33% and 12%, respectively.

Acute myeloid leukemia

The efficacy of DLI in patients with acute myeloid leukemia (AML) who relapse post allo-SCT is not so impressive as is the case in patients with CML. Complete remission rates range between 15% and 40%, while 2-year OS is approximately 15%–20% (9,10).

In a recent retrospective study by Schmid et al. on behalf of the Acute Leukemia Working Party (ALWP) of the EBMT, the authors analyze the outcome of 399 patients with AML who relapsed (first relapse) after allo-SCT. In 171 out of 399 patients DLI was administered as part of treatment of post allo-SCT relapse. Patients who received DLI for treatment of relapse had a better OS as compared with patients who did not receive DLI, and this difference remained statistically significant in multivariate analysis. Factors predictive for better OS after treatment with DLI were: 1) low tumor burden at relapse (< 35% blasts), 2) favorable cytogenetic abnormalities, and 3) achievement of hematological remission before DLI (10).

Acute lymphoblastic leukemia

Acute lymphoblastic leukemia (ALL), responds even more poorly to DLI than AML. Although some patients may achieve remission after DLI with or without chemotherapy, most remissions are short in duration. However, DLI should be part of relapse treatment because occasionally some patients may experience long-term disease control. Slavin et al. have reported a child with ALL who entered complete remission after repeated DLI infusions and remains in remission 15 years after the DLI (11). In general, chemotherapy is administered before DLI in most cases of relapsed ALL in order to achieve rapid disease control. The median initial dose of DLI administered for ALL is approximately 2 × 10⁸/kg T-cells and is significantly higher in comparison with relapsed CML patients. The overall response rate after DLI ranges from 10% to 20%, while 2-year OS rates range from 5% to 20% (12,13).

Other hematological malignancies

DLI has been tested as treatment of post allo-SCT relapse in almost all cases of hematological malignancies. DLI is effective in cases of low-grade lymphomas, especially follicular lymphoma. The efficacy is less pronounced in cases of multiple myeloma, Hodgkin disease, and high-grade lymphomas (14,15).

Methods to improve the efficacy while minimizing toxicity of standard DLI

In an effort to decrease the incidence of GVHD various methods have been developed during past years. Previous studies have shown that depletion of CD8-positive T-cells from DLIs or from the hematopoietic stem cell graft reduced the incidence and severity of GVHD without completely abrogating the GVL effect (16). The effectiveness of CD8-depleted DLI has been shown in a small randomized trial (17).

Prophylactic or pre-emptive DLI

Many groups explored the safety and efficacy of DLI administration either as prophylaxis of relapse in high-risk cases, or pre-emptively in cases with mixed chimerism or minimal residual disease (MRD). DLI administering, prophylactically or pre-emptively, is usually given in lower and gradually escalating doses in order to reduce the rates of severe GVHD (18,19). A research team from China is testing the efficacy of ‘modified’ DLI as a means to prevent relapse in high-risk acute leukemia after allo-SCT. Donors received granulocyte colony stimulating factor (G-CSF) prior to lymphocyte collection, and recipients of modified DLIs received a short course of immunosuppression (usually a calcineurin inhibitor) in order to reduce the incidence of severe GVHD (20). The efficacy, safety, and the most appropriate method for prophylactic DLI administration should be further evaluated in large prospective randomized trials.

In conclusion, DLI represents the first immunotherapeutic approach for treatment or prevention of relapse post allo-SCT. However, the efficacy of DLI is limited in cases of highly aggressive tumors as well as in cases with overt relapse. Moreover, DLI can be complicated by severe GVHD because it provokes a generalized immune reaction against host antigens without any discrimination between leukemic and normal tissues. Even in cases where GVL effect can be observed in the absence of clinical GVHD, this is due to the greater sensitivity of malignant cells to graft-versus-host (GVH) effect in comparison with normal tissues, and these cases should not be considered as examples of GVL dissociation from GVHD.

GVL without GVHD: myth or reality?

Complete dissociation of GVL from GVHD remains the holy grail of transplant immunotherapy. The alloreactive process after human leukocyte antigen (HLA)-matched allo-SCT is directed against mismatched minor histocompatibility antigens (mHA) present in the recipient and absent from the donor. Alloreactive donor T-cells display T-cell receptors (TCRs) with specificity against the complex mismatched mHA/self HLA (21). Since most of mHAs are broadly expressed, donor T-cells attack not only hematopoietic tissue but also normal epithelial cells, thus causing not only depletion of host hematopoietic and lymphoid system but also severe damage in skin, liver, and intestine, manifesting clinically as acute GVHD. From a theoretical point of view these two alloreactive processes can be separated only if powerful immune reactions can be generated against leukemia-associated antigens (LAA) not expressed on normal tissues (22). Another opportunity for GVL in the absence of GVHD is offered by mHAs expressed only on cells of hematopoietic origin. In this situation alloreactive T-cells specific for hematopoietic tissue-restricted mHAs will cause depletion of host hematopoietic cells and thus a strong GVL effect (only in case of hematological malignancies) without damage of normal epithelial tissues and thus with no GVHD. However, because identification of LAA is not possible in every case, and moreover due to the fact that immune responses against LAA are not so strong as is the case with immune responses against allo-antigens, a few other options will be discussed. A general overview of immune reactivity against antigens expressed on tumor cells is shown schematically in Figure 1.

Figure 1. Antigens expressed on tumor cells: targets for immunotherapy. 1) Tumor-associated antigens (TAAs): antigens expressed on tumor cells, while expression on normal cells is minimal or absent. Immune reactivity against TAAs will provoke GVL in the absence of GVHD. Allogeneic as well as autologous immune effectors can be used against TAAs. 2) Hematopoietic tissue-restricted mismatched minor HAs: allogeneic antigens expressed on normal as well as on malignant cells of hematopoietic origin. Immunotherapy against minor HAs can be performed only in the setting of allo-SCT and in cases when donor and recipient are mismatched in the GVH direction. Immune reactivity against minor HAs will provoke GVL in the absence of GVHD. 3) Mismatched HLA, and broadly expressed mismatched minor HAs: allogeneic antigens expressed on malignant cells as well as on normal hematopoietic and epithelial tissues. Immunotherapy can be performed only in the setting of allo-SCT. Immune reactivity will provoke GVL and GVHD. 4) Antigens expressed on malignant cells as well as on normal counterparts, e.g. CD19. Immune reactivity will provoke GVL in the absence of GVHD, but depletion of normal cells expressing the same antigen with all the relevant consequences will occur (e.g. depletion of normal B-cells, hypogammaglobulinemia, infections in case of CD19 targeting). Allogeneic as well as autologous immune effectors can be used.

Minor histocompatibility antigens

Minor histocompatibility antigens (mHAs) are peptides generated after proteolytic cleavage of polymorphic proteins; mHAs are presented as complexes with HLA molecules after binding to the groove of HLA class I and II (23). In the setting of allo-SCT from HLA-identical donors, mHAs mismatches between donor and recipient are responsible for the alloreactive process. Previous experiments and analysis of clinical data have shown that mHAs mismatches in the host-versus-graft (HVG) direction might provoke graft rejection, while mHAs mismatches in graft-versus-host (GVH) direction are responsible for both the GVHD and GVL effect.

Regarding tissue distribution, mHAs can be separated in two major subcategories. Broadly distributed mHAs are expressed on almost every tissue and cell type, including liver, intestine, skin, and hematopoietic cells. Donor T-cells with specificity against broadly expressed mHAs are the mediators of both GVHD and GVL effect (24). For a subset of mHAs expression is restricted on hematopoietic cells. From a theoretical point of view, hematopoietic tissue-restricted mHAs are ideal targets for immunotherapy because alloreactive reactions will provoke the GVL effect in the absence of GVHD (25). An attractive target for treating hematological malignancies with high risk of relapse after allo-SCT is antigen HA-1. HA-1 is a hematopoietic tissue-restricted mHA presented in the context of HLA-A*02016. However, at the present time only 30% of patients are expected to be mHAs (hematopoietic tissue-restricted) mismatched in the GVH direction with their respective donors, and therefore only a small percentage of patients in need would be eligible for mHAs- directed immunotherapy (26). Molecular identification of mHAs expressed on hematopoietic tissue will help in the expansion of patient–donor pairs with mHAs mismatched in GVH direction.

Data from animal models have shown that cytotoxic T-lymphocytes (CTLs) specific for a single but broadly expressed mHA do not cause GVHD when infused in mice after allo-SCT (27,28). From these data it seems that a co-ordinated alloreaction against several mHAs is required for significant damage and thus for development of the clinical manifestations of GVHD. On the contrary, CTLs against a single broadly expressed mHA can cause significant anti-tumor activity when infused in mice after allo-SCT (29). If these data can be extrapolated to humans then single mHAs can be used as targets for immunotherapy without GVHD development. An absolute prerequisite in order to prevent GVHD is to avoid completely any concomitant infusion of T-cells with specificity against other mHAs.

In healthy donors not previously immunized, the frequency of T-cells with specificity against mHAs is in the order of 1 out of 10⁵ cells in peripheral blood (30). Therefore massive in vitro expansion of allo-specific T-cells is required before their administration to recipients. The methods employed so far for ex vivo expansion of mHAs-specific T-cells are extremely time and labor-consuming. More importantly, ex vivo expanded cells do not retain the ability of self-renewal; they are thus short-lived, resulting in only a transient GVL effect. Warren et al. explored the safety and efficacy of ex vivo expanded donor CTLs with specificity against hematopoietic tissue-restricted mHAs. Seven patients with relapsed leukemia after HLA-matched allo-SCT were included in the study. Pulmonary toxicity was observed in three patients after CTLs infusion and was attributed to the presence of mHAs in lung tissue. GVHD was not observed in any of the treated patients. Treatment was effective with five patients achieving complete but transient remission. Infused cells were detected only for a very limited period of time, consistent with the transient nature of the remissions (31). Culture conditions result in massive expansion of CTLs by promoting extensive differentiation of central memory to effector memory cells resulting in the exhaustion of central memory cells. Future trials should focus on exploring culture conditions that will result in the preservation of significant numbers of cells with central memory phenotype.

Instead of cloning T-cells from the peripheral blood of donors, another method is to transduce T-cells with a gene encoding TCR specific for a mHA expressed on the recipient's cells (32). Indeed generation of T-cells genetically modified to express specific anti-HA-1 TCR receptors is feasible, and it is an attractive target for leukemia immunotherapy in the absence of GVHD. Many other mHAs with expression limited to hematopoietic tissue have been identified and are promising targets for immunotherapy. UTA2-1 is a mHA presented in the context of HLA-A*02:01 molecule. This antigenic peptide was identified after cloning a CTL from a multiple myeloma patient who achieved long-lasting complete remission after receiving DLI from his HLA-identical sibling. UTA2-1 as well as other recently identified mHAs are promising targets for minor mHAs-directed immunotherapy (33,34).

Another option is the use of donors previously immunized against mHAs expressed by the recipient. In donors previously immunized against the relevant mHAs the frequency of Ag- specific T-cells is much higher as compared with unimmunized individuals. Moreover, a significant percentage of Ag-specific T-cells are central memory cells with self-renewal capacity. Selection of Ag-specific T-cells from peripheral blood of donor can be performed using various methods as in the case of viral-specific cells. It is possible that infusing Ag-specific cells from pre-immunized donors might provoke a permanent GVL effect since central memory cells are long-lived. However, such a methodology has not been explored so far, mainly due to ethical concerns regarding safety of the donor.

Leukemia or tumor-associated antigens

Many previous experiments have shown that leukemia or tumor-associated antigens (LAA, TAA) do really exist, although the immunogenicity of these antigens, as well as the capacity to mount an effective immune response, is questionable. TAA are antigens expressed on malignant cells, while expression on normal cells is either absent or limited to very low levels. Subtypes of potentially immunogenic TAAs are briefly summarized. 1) Antigens expressed in immune-privileged sites and aberrantly expressed on malignant cells: A typical example is the MAGE family of proteins expressed in testis and therefore ignored by the immune system and considered as foreign. In certain malignant diseases such as melanomas, MAGE molecules are aberrantly expressed on neoplastic cells (35). 2) Antigens expressed at very low levels on normal cells but overexpressed in tumor cells: There are many proteins such as proteinase-3 (PR3) and Wilms's tumor antigen-1 (WT1) physiologically expressed at very low levels on normal tissues but overexpressed on leukemic cells (36). 3) Novel antigenic epitopes derived from mutated or chimeric proteins present in the vast majority of neoplastic cells: A few examples are peptides derived from mutated p53, and bcr-abl (37). Potentially immunogenic TAAs are shown in Table I.

Table I. Potentially immunogenic tumor-associated antigens.

Type of antigen	Expression on normal cells	Expression on malignant cells	Immunogenicity	Anti-tumor activity	Autoreactivity
Broadly expressed antigens	Yes	Yes	Very low or absent	Low	Yes?^a
Antigens expressed in immune-privileged sites	No	Yes in some cases	Yes	Yes	No
Low-level-expressed antigens	Very low levels	Overexpression in some cases	Yes	Yes	No
Novel antigens	No	Novel proteins encoded by mutated or translocated genes	Yes	Yes	No
Viral antigens	No	Yes in virus-infected malignant cases	Yes	Yes	No

^aAutoreactivity in cases of successful generation of immune reaction.

However, a fundamental question concerning the immunogenicity of TAAs remains to be answered. If TAAs really exist, then how and why do tumors escape immune surveillance? It is possible that only tumors that manage to evade the immune system finally develop and produce a clinically overt malignant disease. It is conceivable that due to selection pressure malignant cells expressing highly immunogenic TAAs are rejected by the immune system and finally only tumors expressing TAAs with weak or absent immunogenicity manage to develop. Consistent with this hypothesis is the observation that the GVL effect is practically absent after transplantation using a syngeneic twin as a donor.

A few examples of TAAs currently tested as antigenic targets are given below:

PR3 is serine protease present in the cytoplasm of granulocytes. Autoantibodies against PR3 are present in patients with Wegener granulomatosis (36). Myeloid cells express PR3 at very low levels under normal conditions. However, overexpression of PR3 has been observed in many AML, CML, and myelodysplastic syndrome (MDS) patients. PR3 is one of the well- characterized LAAs that are capable of eliciting specific immune responses. The peptide PR1 derives after proteolytic cleavage of PR3 protein and binds to the HLA-A-0201 molecule. Indeed, T-cells that are specific for the complex HLA-A-2/PR1 can be identified in the peripheral blood of patients with myeloid malignancies in remission. With this theoretic background and following observations from in vitro studies, Qazilbash et al. performed a clinical trial to explore the efficacy of vaccination against PR1 in patients with myeloid malignancies. Thirty-five patients were included in the trial and followed for 1–4 years. Response rates including complete remissions after vaccination were observed in 25% of patients, while toxicity was minimal. This is the first study showing that vaccination can produce complete remission (CR) in patients with active leukemia (38).

An additional target is the protein WT1. The WT1 gene encodes for a transcription factor involved in cell proliferation, differentiation, and apoptosis. WT1 is expressed at very low levels on various tissues such as ovary, kidney, and testis. Similarly to PR3, WT1 is aberrantly overexpressed in a variety of human malignancies including acute leukemias. A large number of HLA-A2 and HLA-A24-restricted WT1 peptides have been identified in the peripheral blood of patients with leukemia in remission. Keilholz et al. explored the safety and efficacy of WT1 peptide vaccination in 19 patients with AML and MDS. Impressively, more than 50% of evaluable patients had an objective clinical response, while toxicity was again minimal. Of note is the observation that one patient with active AML achieved CR. Interestingly, peripheral blood WT1 transcripts decreased in all clinical responders (39). A new and very interesting approach is the generation of mabs with a ‘TCR-like’ motif specific against tumor antigen-derived peptide/HLA complexes expressed on the surface of cancer cells. A WT1-derived peptide named RMF presented in the context of HLA-A0201 has been recently discovered, and a fully human ‘TCR-like’ mab (known as ESK1) specific against the complex RMF/HLA-A0201 has been constructed. Binding experiments showed that ESK1 binds with high avidity to cell lines from solid tumors and acute leukemias only in the presence of WT1 and HLA-A0201 expression. More importantly, low doses of ESK1 showed significant therapeutic activity when infused in non-obese diabetic with severe combined immunodeficiency (NOD/SCID) mice transplanted with human acute leukemia. In contrast, no toxicity was observed in healthy HLA-A0201 transgenic mice (40).

The third important TAA in hematological malignancies is the PHAMM protein. Similarly to PR1 and WT1, the PHAMM protein under normal circumstances is expressed at very low levels in various tissues like thymus, placenta, etc. PHAMM is overexpressed in various hematological malignancies including AML, CML, MDS, and multiple myeloma (MM). PHAMM-R3 peptide is derived from PHAMM and is presented in the context of HLA-A2 molecule. T-cells with specificity against PHAMM-R3/HLA-A2 have been identified in the peripheral blood of patients with myeloid malignancies in remission, as well as in normal volunteers. Similar to the case of PR1 and WT1, in a recent study Schmitt et al. showed that PHAMM-R3 peptide vaccination produces objective clinical responses in patients with myeloid malignancies (41).

A large body of data supporting the existence of TAAs and the feasibility of inducing an autologous immune response against them comes from studies in patients with metastatic melanoma. A dense lymphocyte infiltrate has been detected in tumors taken from patients with melanoma. These lymphocytes have been defined as tumor-infiltrating lymphocytes (TIL) and display significant cytotoxic activity against melanoma cells. Isolation of TIL followed by ex vivo expansion and reinfusion to the patients has been associated with clinical benefit (42). A lympho-depleting conditioning regimen is administered before infusion of TILs in order to allow homeostatic expansion of infused cells (43).

Based on these data, further studies focused on TCR engineering and generation of T-cells expressing TCRs with pre-defined specificities. The first clinical trial of TCR gene therapy was performed in patients with advanced melanoma. Autologous cells were collected from the peripheral blood of patients and were transduced by using a retroviral vector with a gene encoded for a TCR specific for melanoma-associated antigen-1 recognized by T-cells (MART-1) presented in the context of HLA-A2. TCR-transduced cells were detected in patients for as long as 1 year post infusion. Four out of 31 (13%) treated patients had an objective clinical response (44). In this trial autoimmune manifestations occurred and were mainly attributed to destruction of normal melanocytes in the skin, eye, and ear resulting in vitiligo, uveitis, and hearing loss. In a more recent trial a TCR with specificity against the NY-ESO-1 peptide presented in the context of the HLA-A*0201 was constructed. NY-ESO-1 is tumor antigen-expressed in many cases of melanoma as well as in various other tumors such as breast, prostate, thyroid, ovarian cancer, and synovial cell sarcomas. Autologous T-cells from patients with melanoma and synovial cell sarcoma were transfected with the gene encoding for the relevant TCR and reinfused to patients. Objective clinical responses were observed in four out of six and in five out of 11 patients with synovial cell sarcoma and melanoma, respectively. Interestingly, two out of 11 melanoma patients achieved complete remission that persisted for more than 1 year (45).

Selectively allodepleted T-cells

GVHD can be completely abrogated after in vitro T-cell depletion (TCD) of the graft. Although GVHD-related mortality is minimized, an excess risk of infectious deaths as well as increased relapse rates and somewhat higher rates of non-engraftment are the major drawbacks after extensive in vitro TCD, resulting eventually in similar overall survival as compared with the use of unmanipulated grafts (46).

Stem cell grafts contain a significant number of donor lymphocytes. Host alloreactive donor lymphocytes are responsible for GVHD. However, alloreactive lymphocytes represent only a minor fraction of the total lymphocyte population. The vast majority of lymphocytes contained in the graft are innocent passengers with significant activity against infectious agents, and against LAAs or TAAs. Theoretically selective elimination of host alloreactive cells from the graft will result in significant reduction in GVHD rates, without the limitations associated with unselected in vitro TCD. Indeed, many different methods have been explored by different centers during the last years specifically to eliminate the alloreactive lymphocytes from the graft. These methods have been named ‘selective allodepletion’ (SD).

SD methods are based on the same concept (47): Irradiated host peripheral blood mononuclear cells (PBMCs) or different subpopulations are co-cultured with donor T-cells for 3–5 days. Under culture conditions donor T-cells, with host alloreactivity, are activated and express various surface-specific markers that can be used as targets for T-cell elimination. Elimination of activated cells can be performed with the use of monoclonal antibodies coupled to magnetic beads or to toxins. A large number of surface markers, such as CD25, CD69, CD137, CD134, etc., have been used for targeting of allostimulated cells (47–49).

Another SD approach is the photodepletion procedure, and a brief summary is given below. Allostimulated T-cells start to proliferate and deactivate the protein pump multidrug resistance p-glycoprotein (MDR1) which is responsible for the cell efflux of various toxic substances. Following co-culture in the presence of stimulators the phototoxic rhodamide-dye (TH9402) is added to culture medium, and cells are exposed to light energy (5 J/cm²) (50). TH9402 is basically an inert molecule, but it becomes extremely toxic after photosensitization. High intracellular concentration of photosensitized TH9402 provokes apoptosis of allostimulated cells. In contrast, the vast majority of resting cells will survive the SD procedure because active MDR1 pump will keep intracellular TH9402 at very low levels.

Roy et al. explored the safety and efficacy of SD-treated DLI using a photodepletion methodology, named Add-Back of T Cell for Immune Reconstitution (ATIR) (Kiadis Pharma, Amsterdam, Netherlands) (51). Preliminary results were promising, and a multicenter international trial evaluating the efficacy of ATIR-treated DLI is currently underway.

T-cells transduced with suicide genes

The goal of suicide gene therapy is the transfection of T-cells with a gene that will set up the apoptotic machinery of the cells in case of unwanted effects. Among the different suicide genes, the herpes simplex virus-derived thymidine kinase (TK) gene is the most extensively explored in clinical trials. TK is an enzyme that phosphorylates ganciclovir (GCV) into the active drug. The active drug acts as a GTP analog and causes inhibition of DNA replication. Therefore GCV when administered is toxic only to proliferating TK-expressing lymphocytes (52). After haplo-SCT TK-transduced lymphocytes are infused in order to harness immune recovery and prevent relapse. In case the patient develops GVHD, GCV is administered to the patient. GCV is toxic to proliferating alloreactive transduced cells leading to apoptosis and to GVHD termination. More importantly the rest of TK-transduced cells with activity against infectious agents and leukemic cells are not affected by GCV since they remain in a resting state.

The seminal study performed so far has been published by Ciceri et al. (53). Fifty patients who underwent T-cell- depleted haplo-SCT with purified CD34 + cells were included in this phase II study. Twenty-eight out of 50 patients received TK-transduced lymphocyte infusions. Immune recovery was achieved in all patients at a median of 75 days. No adverse effects related to infusion or to retroviral transfection were recorded. Ten patients developed acute GVHD (mainly cutaneous, Grade I–II), and one patient chronic GVHD. Interestingly, in all cases GVHD resolved completely after GCV administration.

TK protein is of viral origin, and immune reactions against TK-expressing cells had been observed in initial experiments. Indeed, when TK-transduced cells were infused in patients after HLA-identical SCT, rejection of TK cells was observed in a few cases (54). Moreover, authors noticed that the longer the interval between allo-SCT and infusion, the higher the possibility of immune rejection, concluding that generation of immune reactivity against TK protein is dependent on the degree of immune competence of the patient at the time of infusion. However, initial concerns regarding the immunogenicity of TK protein were not confirmed in the setting of haplo-SCT, and no case of immune reactivity against TK cells has been reported so far (53).

Furthermore, the anti-leukemic efficacy of TK-transduced cells has been tested in previous experiments in 23 patients with various hematological malignancies who relapsed after HLA-matched SCT. Six and five out of 17 evaluable patients achieved complete and partial remission post DLI, respectively, for an overall response rate of 65%. Responses correlated with expansion of TK-expressing cells in the peripheral blood of the patients, and the anti-tumoral effect was durable. GVHD developed in seven patients and completely resolved after GCV administration (55). A similar GVL effect of TK cells infusion has been observed after haplo-SCT. Interestingly, the administration of the TK-transduced cells boosts the immune recovery and conventional T-cell recovery, leading to a reduction of transplant related mortality (TRM) from 60% in the historical controls to less than 20%, mostly due to significant reduction in infection complications and in infection-related mortality (53).

To overcome the problem of TK-induced immunogenicity, other teams explored the efficacy of transduction with other suicide genes with less immunogenic potential, such as a gene that encodes for the death domain of caspase 9. However, this approach is still in an early stage, and the available clinical data are limited so far (56).

NK cell immunotherapy

Natural killer (NK) cells are lymphocytes that belong to the arm of innate immunity and express significant cytotoxic activity against neoplastic and virally infected cells. Detailed information regarding the physiology of NK cells is given in other reviews (57). There is a large body of experimental data supporting the role of NK cells in immune surveillance against tumors. It is known that NK cells from the peripheral blood of AML patients in remission display potent cytotoxic activity against leukemic blasts collected at the time of initial presentation (58). Moreover, data from in vitro experiments showed that IL-2-activated NK cells have significant cytotoxic activity against a variety of solid tumors (59,60). Rosenberg explored the clinical efficacy of lymphokine-activated killer cells (LAK) in patients with metastatic melanoma. LAK represent a mixture of activated NK and T-cells respectively. In his seminal trial objective responses were observed in a small percentage of patients (61).

NK cell immunobiology became again the focus of scientific interest when a Perugia team reported on the beneficial role of NK cells on the outcome after T-cell-depleted haplo-SCT. In this study, killer immunoglobulin-like receptor (KIR) alloreactivity in the GVH direction was associated with decreases of both relapse and GVHD rates, resulting in significant survival advantage especially in the group of patients with AML transplanted in remission (62).

Following the results from the Perugia team, other teams explored the efficacy of mismatched NK cells adoptive immunotherapy (outside the setting of transplantation) in patients with advanced malignancies that did not undergo stem cell transplantation. A lympho-depleting conditioning (including fludarabine and cyclophosphamide) was administered to patients before NK infusion with the aim to achieve transient engraftment and expansion of infused cells. Miller et al. tested the efficacy of mismatched NK cells infusions in 43 patients with various refractory malignancies including solid tumors and AML (63). Peripheral blood mononuclear cells (PBMCs) were collected from haplo-identical family donors. Patients received CD3-depleted NK infusions by using immunomagnetic selection in order to prevent severe GVHD. Interestingly in severely lympho-depleted patients infused NK cells underwent remarkable in vivo expansion, most probably due to increased blood levels of IL-15, and were detectable in peripheral blood for a median of 30 days. More importantly objective clinical benefit was observed in this group of patients with five out of 19 AML patients achieving CR. No clinical or laboratory sign of GVHD was observed after infusion of the mismatched NK cells. The Curti et al. explored the same procedure in 13 patients with high-risk AML. Purified NK cells were collected from haplo-identical family donors with KIR alloreactivity in GVH direction. Fludarabine plus cyclophosphamide was administered before infusion as a lympho- depleting regimen, and patients received IL-2 post infusion. All NK infusions were contaminated with less than 10⁵/kg T-cells. GVHD did not occur in any of the patients. Results regarding anti-leukemic efficacy were strikingly promising. One out of five patients was treated at a state of active disease, and two out of two patients with molecular relapse achieved CR which lasted for several months. In addition, three out of six patients with morphological remission achieved long-lasting disease-free survival (64).

In another study, authors explored the safety and efficacy of intentionally mismatched LAK cells in 40 patients with metastatic solid tumors and refractory hematological malignancies (65). Despite all patients receiving unmanipulated cells, only one patient developed self-limited skin GVHD. Objective clinical benefit was observed in five out of 21 and in four out of five patients with metastatic solid tumors and hematological malignancies, respectively.

The major limitation of infusing NK or LAK cells in patients with malignancies that did not undergo stem cell transplantation prior to the cellular therapy is the transient nature of the anti-tumor effect due to the inevitable rejection of allogeneic cells. From a theoretical point of view infusion of allogeneic NK cells after allo-SCT might be more effective, because it is conceivable that the NK cells will not be rejected and therefore the anti-tumor effect will last longer. Indeed, several teams explored the efficacy and safety of post allo-SCT infusion of purified NK cells (66–69).

In one such study, 19 patients with high-risk hematological malignancies were included. Among them, 15 patients underwent a previous T-cell depleted haplo-SCT, while four patients underwent allo-SCT from matched donors either related or unrelated. The donors of NK cells were the original donors in all cases. PBMCs were cultured for 2–3 days in the presence of IL-2. Following initial activation and expansion, NK cells were selected using antiCD56 antiCD3 monoclonal antibodies coupled to magnetic beads (66). The procedure was safe, with only four patients suffering from acute GVHD (in all four cases NK infusions were contaminated with high numbers of T-cells, and authors speculated that GVHD could be prevented in future trials by more extensive TCD). The above data suggest that purified NK infusions can be administered safely without the risk of adverse reactions such as severe GVHD (66–69). Future trials need to be performed in order to define the timing, dose, and number of required administrations in order to achieve maximal clinical benefit.

Another method to increase cytotoxic activity of NK cells is by using antibodies against KIRs. Monoclonal antibodies against KIR receptors might result in activation of resting autologous NK cells and augment anti-tumor activity. IPH2101 is a fully humanized monoclonal antibody with specificity against inhibitory KIRs such as KIR2DL-1, KIR2DL-2, and KIR2DL-3 which collectively interact with all HLA-C alleles. In in vitro experiments IPH2101 enhances NK cytotoxic activity against fresh AML blasts. The safety and efficacy of IPH2101 was tested in a phase I study including 23 elderly patients with AML in first CR. Adverse events were manageable, and OS as well as DFS compared favorably to a historical well-matched control group (70). IPH2101 has been tested also in 32 patients with relapsed/ refractory MM in a phase I trial. Although in vitro NK cytotoxic activity against MM was increased, no objective response was observed in any of the patients (71).

Amplifying anti-tumor immunological attack by mabs with or without DLI-targeted immunotherapy

Monoclonal antibodies targeting antigens expressed on tumor cells e.g. CD20 have shown excellent results in the treatment of various malignant disorders including non-Hodgkin lymphoma and breast cancer, resulting in prolongation of DFS and OS. Progress in biotechnology engineering resulted in generation of numerous monoclonal antibodies blocking inhibitory receptors present on the surface of immune effectors as well as of novel molecules like bispecific T-cell engagers (BiTEs). Blocking antibodies augment autologous anti-tumor immunity, while BiTEs bind to autologous T-cells with various specificities, and engaged T-cells exert cytotoxic activity against tumor cells in a major histocompatibility complex (MHC)-independent manner. Blocking antibodies and BiTEs can be used after allo-SCT either as single agents or in combination with DLI. In both cases, augmentation of GVL with minimal or absent GVHD is theoretically expected to occur.

T-cell and NK-cell engagers—BiTEs and BiKEs

Bispecific T-cell engagers (BiTEs) are novel molecules consisting of two single-chain antibody variable fragments (scFv). A linker is used in order to combine the two Fv of the two antibodies with different specificities. One chain is specific for CD3, and the other is specific for an antigen expressed on the surface of a target tumor. BiTE antibodies activate T-cells by binding to their surface CD3 and bring these T-cells in close proximity with the target. In this situation there is no need for previous priming of cytotoxic T-cells, and cytotoxicity is not MHC-restricted. The anti-tumor response is polyclonal since every T-cell engaged by BiTE exerts cytotoxic activity against the target (72). The mechanism of action is shown schematically in Figure 2.

Figure 2. T-cell and natural killer cell engagers (BiTEs and BiKEs). A: Bispecific T and natural killer cell engagers (BiTEs and BiKEs) consist of two single-chain antibody variable fragments (scFv). In the case of BiTE antibody scFv is specific for CD3 antigen, while in the case of BiKE scFv is specific for CD16. The other scFv is specific for an antigen expressed on the surface of tumor cells, like CD19, CD33, etc. A linker is used in order to combine the two variable regions of the two antibodies. B: BiTE antibodies activate T-cells by binding to their surface CD3 and bring these T-cells in close proximity with the target. In an analogous manner BiKEs activate NK cells by binding to CD16. Activation of effector cell results in perforin and granzyme degranulation and tumor cell apoptosis.

In case of B-cell malignancies one of the antigenic targets is CD19, and BiTE antibody is a two single-chain CD3/CD19 molecule. The anti-CD19 BiTE antibody has been given the name blinatumomab. CD19 was chosen because it is expressed in all B-cells, from immature pre-B-cells to mature antigen-specific B-cells, while it is lost during terminal differentiation to plasma cells. Moreover, expression of CD19 is stable and minimal, and no shedding has been observed in various experiments. In preclinical testing, blinatumomab has shown significant cytotoxic activity in B-ALL and B-lymphoma cell lines (73). Following exposure to blinatumomab all CD19 expressing cells including normal B-cells undergo apoptosis, resulting in long-lasting severe lymphopenia and hypogammaglobulinemia.

The first clinical trial with blinatumomab was a dose-escalating study in B-cell-indolent lymphomas. In this initial study, 38 patients were treated with escalated doses of the antibody. Response rate was 30% including four CR (74). Blinatumomab was also tested in 21 patients with B-ALL with persistent or relapsed molecular disease. Response rate was impressive with 80% of patients achieving complete molecular remission (75). These data are promising, taking into consideration the poor prognosis of ALL patients with persistent MRD after induction. In a phase II trial performed by the ALL German study group, 36 patients with B-ALL in hematological relapse were treated with blinatumomab. Twenty-six patients achieved hematological CR for an overall response rate of 72%. Thirteen patients underwent allo-SCT while in remission, and 12 out of 13 remain disease-free for a median follow-up period of 10 months (76).

In analogy to BiTEs, BiKEs and TRiKEs are bispecific and trispecific killer cell engagers. BiKEs consist of two different single Fv regions of two monoclonal antibodies. One chain is specific for CD16, and the other is specific for an antigen expressed on the surface of the target tumor. BiKE antibodies engage NK cells and bring them in close proximity to the target tumor. NK cells exert their cytotoxic activity by using degranulation of granzymes and perforin. A recently constructed BiKE is the anti-CD16/anti-CD19 antibody designed for use against B-cell malignancies. TriKEs are similar to BiKEs, but instead of one they target two antigens expressed on the surface of the tumor with the aim to potentiate NK cell cytotoxicity even further. A TriKE designed for use against B-lymphoid diseases is the anti-CD16/anti-CD19/anti-CD22 antibody. In in vitro experiments BiKEs and TriKEs showed significant cytotoxic activity against B-lymphoma cell lines (77).

Ipilimumab

The failure of immunotherapeutic approaches including the GVL effect provoked by DLI can be attributed to insufficient co-stimulation of effector cells or to the negative action of T-regulatory (Tregs) cells that are usually recruited in the local tumor microenviromnment of many malignancies. Cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) is a key regulator of T-cell activation and helps in the homeostatic balance of the immune reactions. CTLA-4 antigen is expressed on the surface of activated T-cells and interacts with the activating ligands CD80 and CD86 expressed on the surface of antigen-presenting cells (APCs). CTLA-4 and the co-stimulatory receptor CD28 are both present on activated T-cells and compete for binding to the same activating ligands. When CTLA-4 interacts with CD80 and CD86 instead of CD28 then termination of the immune reaction occurs, preventing tissue damage from an uncontrolled immune system (78). CTLA-4 is also constitutively expressed on the Treg cell surface. Tregs are the key players in maintaining peripheral tolerance. In the setting of allo-SCT Tregs are mediators of tolerance and therefore help in the prevention of GVHD. However, inhibition of the alloreactive process results also in down-regulation of the beneficial GVL effect. The mechanism of Tregs-mediated immune modulation requires cell-to-cell contact of the Tregs with the effector cells, as well as secretion of anti-inflammatory cytokines such as TGFb and IL-10. Immune effectors when in close contact with Tregs become anergic, and this action is mediated through interaction of CTLA-4 expressed on Tregs with ligands expressed on the surface of immune effectors (79). Tregs are recruited in the local microenvironment of many tumors, cause a state of local immune suppression, and help in tumor evasion and growth. Indeed, numerous clinical studies have shown the negative impact of increased Tregs in the tumor microenvironment on response to treatment as well as on overall survival (80).

In experimental animal models administration of a monoclonal antibody against CTLA-4 resulted in significant anti-tumor responses, especially in metastatic melanoma. Ipilimumab is the first fully human monoclonal antibody with specificity against CTLA-4 (Medarex, Bloomsbury, NJ, USA, and Bristol-Myers Squibb, Wallingford, CT, USA) (81).

Animal models exploring the effect of ipilimumab after allo-SCT showed that administration of ipilimumab is associated with increased incidence of GVHD if administered early after allo-SCT. In contrast, delayed administration of ipilimumab resulted in augmentation of the GVL effect without causing more GVHD (82). The safety and efficacy of ipilimumab administered after allo-SCT have been tested in a group of 29 patients with relapsed or progressive hematological malignancies. This was a dose escalation study, and no dose-limiting toxicity was identified. GVHD or graft rejection was not observed in any of the treated patients. Autoimmune manifestations occurred in four patients including hyperthyroidism, arthritis, and pneumonitis. Despite the poor prognosis, two patients with Hodgkin lymphoma and one with mantle cell lymphoma achieved CR and PR, respectively (83).

A clinical trial exploring the efficacy of ipilimumab in MDS and AML patients is currently underway (NCT01757639).

Programmed death-1 protein

Analogous to CTLA-4, programmed death 1 (PD-1) protein is another inhibitory receptor expressed on the surface of T-cells. The PD-1 molecule has significant homology with CTLA-4 but has different ligand specificity. The ligands for PD-1 are PD-L1 and PD-L2. PD-L1 is expressed on many tumor cells and mediates tumor escape from immune surveillance. After binding to the PD-1 receptor expressed on T-cells, PD-L1 inhibits both cytokine secretion and cytotoxic activity of activated CD4 and CD8 cells, respectively. PD-1 is also present on the surface of NK cells, and interaction of PD-1 with its ligands results also in inhibition of NK cells (84). Based on these data it seems that blockage of PD-1 might result in significant anti-tumor responses.

The first anti-PD-1 blocking antibody tested in a clinical trial was CT-011 (humanized IgG1 monoclonal antibody). CT-011 (pidilizumab) interacts with PD-1 and leads to NK stimulation and enhancement of NK anti-tumor responses. Efficacy and safety of CT-011 was tested in 17 patients with advanced malignancies in a dose escalating study. CT-011 was safe and well tolerated in this patient population, and no maximum tolerated dose was identified in this study. Clinical benefit was observed in a significant percentage (33%) of patients, with one complete remission (85).

BMS-936559 is a high-affinity, fully human monoclonal antibody specific for PD-L1. BMS-936559 blocks the interaction of PD-L1 expressed on the tumor surface with both PD-1 and CD80 receptors present on immune effectors. The safety and efficacy of BMS-936559 antibody has been tested in more than 200 patients with various advanced solid tumors. Objective clinical responses were observed in approximately 20%, 10%, and 10% of patients with melanoma, renal, and non-small lung cancer, respectively (86).

In a recent trial, pidilizumab was administered post autologous SCT (auto-SCT) in patients with diffuse large B-cell (DLBCL) and primary mediastinal B-cell (PML) lymphoma. Sixty-six patients with relapsed DLBCL or PML entered the study. DFS for the whole cohort of patients was 72%, and there was no difference between patients with PET-positive and PET-negative response to salvage chemotherapy. Interestingly, the overall response rate for patients treated for residual disease after auto-SCT was as high as 52%. Pidilizumab was well tolerated, and no serious adverse reaction was observed in this trial (87).

Regarding PD-L2, interesting data have been observed in animal studies. Animal experiments using mouse models have shown that PD-L2 expression was limited to hematopoietic cells, while PD-L1 was broadly expressed in all tissues (88). Consistent with this finding was the observation that blockage of PD-1/PD-L1 interaction resulted in increasing GVHD-related mortality as opposed to blockage of PD-1/PD-L2 interaction. Taken together, these data provide new insight into the differential roles of PD-L1 and PD-L2 in the setting of allo-SCT. Theoretically selective blockage of PD-1/PD-L2 interaction might result in GVL augmentation without increasing the severity of GVHD. Trials evaluating the efficacy of blocking PD-1/PD-L1/PD-L2 interaction in patients with various malignant disorders are currently underway (NCT00729664, NCT00730639, NCT01096602).

Theoretically it is assumed that the combination of blocking antibodies might display synergistic activity against neoplastic cells. Experimental data showed that blockade of PD-1 is counterbalanced by an increased expression of CTLA-4 on human T-cells. Previous animal models suggest that the concurrent blockade of more than one immune-inhibitory molecule results in synergistic anti-tumor activity (89). These data set up the rationale for an ongoing clinical trial, testing the efficacy of combining ipilimumab (anti-CTLA-4) with nivolumab (anti-PD-1) (NCT01024231).

T-cells with genetically modified receptors

A novel and most promising technique is genetic modification of lymphocytes. Techniques of gene transferring offer the unique possibility to transduce polyclonal T-cells with TCR specific for the relevant mHAs or LAAs in the context of the appropriate HLA molecule. Another method is generation of T-cells armed with chimeric antigen receptors (CAR) with a predefined specificity, such as anti-CD19. Genetic modification is associated with a number of advantages: 1) different T-cell subpopulations can be transduced, and their efficacy can be tested in in vivo experiments. From this point of view, central memory cells might be proved as the ideal T-cell population for TCR transfection. 2) Autologous as well as allogeneic T-cells can be used as the targets for TCR transfection. Use of autologous T-cells would obviate the need for previous allo-SCT as the necessary platform for tumor immunotherapy.

Chimeric antigen receptors

First-generation CARs consist of an extracellular portion which is a single-chain F variable antibody fragment with specificity against an antigen expressed on a tumor target. The extracellular portion is linked via a transmembrane portion to the signaling motif CD3z chain. First-generation CARs exert cytotoxic activity but do not allow the cell to proliferate after engaging the ligand. To overcome these problems second-generation CARs were constructed in a similar manner, but the intracellular portion consisted of the CD3z chain coupled to the signaling motif of a co-stimulatory receptor like CD28. Second-generation CARs contain activation and co-stimulating domains and after engaging the cognate ligand have the ability to secrete cytokines, exert cytotoxicity, proliferate, and expand in the host. Third-generation CARs have an intracellular portion that combines a CD3z chain with two co-stimulating domains acting in parallel. Third-generation CARs are even more effective. Different co-stimulatory molecules have been used for the construction of CARs like CD28, DAP10, 4-1BB, etc. The efficacy of different CARs has not been tested comparatively (90). Both CD4 and CD8 genetically modified to express CAR display cytotoxic activity against target tumors. Cytotoxic activity displayed by CAR-engineered cells is not-MHC restricted and does not require previous activation by an antigen-presenting cell. Moreover, autologous T-cells can be used for CAR engineering because these cells will not be rejected by the patient (91).

CARs have been studied in phase I clinical trials mainly against B-cell malignancies. The most widely constructed CAR is against CD19 antigen expressed on almost all normal and malignant B-cells. Approximately 30 patients with chronic lymphocytic leukemia (CLL), low indolent lymphomas, and B-cell ALL have been treated so far with autologous (mainly second-generation) anti-CD19 CAR cells. Results were promising, with six responders (including two CRs) out of 14 patients with refractory CLL. Responses were also observed in patients with B-ALL. Second-generation CARs resulted in a significantly better in vivo expansion and persistence as compared with first-generation CARs. Response rates were inversely correlated with tumor burden, at the time of infusion: the higher the tumor burden, the lower the probability of response. Current trials explore the role of administering a lympho-depleting conditioning regimen before CAR infusion. It seems that pre-infusion lympho-depleting conditioning results in creating ‘space’ for homeostatic expansion of infused cells to take place. Indeed most clinical responses occurred in patients previously conditioned with chemotherapy. Another important issue that needs to be explored in future clinical trials is the role of post-infusion administration of cytokines like IL-2 (92).

Another interesting target for the CAR technology is the hyaluronate receptor CD44 that is expressed on various malignant tumors. CD44 is overexpressed in hematologic as well as in epithelial tumors. CD44 is also expressed on the surface of normal and leukemic stem cells. A monoclonal antibody with specificity against CD44 has been tested in xenograft models using NOD/SCID mice. Anti-CD44 showed excellent activity in eradication of human AML after infusion in NOD/SCID mice. An interesting observation is that different CD44 isoforms have different expression patterns. In more detail, the isoform variant 6 (CD44v6) is expressed in many cases of AML and multiple myeloma, as well as in pancreatic, breast, and head and neck cancers. In contrast, CD44v6 isoform is not expressed on normal hematopoietic stem cells, and low-level expression has been found on monocytes and keratinocytes. CAR T-cells with specificity against CD44v6 have been constructed by using lentiviral vectors. In vitro experiments with CAR anti-CD44v6 cells showed significant anti-tumor activity against AML blasts, while no effect was observed on normal hematopoietic stem cells or keratinocytes (93). Using the same methodology like in the case of T-cells, chimeric antigen receptors (CAR) can be expressed on the surface of NK-cells. NK armed with CARs recognize tumor antigens on the surface of tumor cells and exert cytotoxic activity (94).

Conclusions and future perspectives

The therapeutic potential of allo-SCT is mainly due to the alloreactive activity of lymphocytes contained in the graft. However, the vast majority of donor lymphocytes provoke a generalized immune reaction against host antigens without discriminating between normal and malignant tissues. Therefore the beneficial GVL effect is usually manifested in association with the deleterious GVHD. From a theoretical point of view dissociation of GVL from GVHD can occur by selecting antigenic targets present on malignant and absent from normal cells. Significant advances in the field of molecular biology and genetic engineering resulted in the development of innovative technologies such as transfection of lymphocytes with TCRs and CARs with predefined specificities. Moreover generation of mabs with specificity against receptors such as CTLA-4, PD-1, and KIR expressed on immune effectors might augment anti-tumor immunity by blocking the inhibitors. Although allo-SCT still remains the ideal platform for immunotherapy, the new developments offer the possibility of effective immunotherapy outside the setting of transplantation.

Declaration of interest: The authors have no relevant conflict of interest to declare.