|Home | About | Journals | Submit | Contact Us | Français|
The notion that immunocompetent cells, contained within adult bone marrow or peripheral blood, are capable of mediating an antitumor effect was first validated experimentally in 1957. T-cell immunotherapy for malignant disease is now routinely used in the context of allogeneic bone marrow transplantation. After 50 years of investigations into the use of T-cells for cancer therapy, adoptive cellular immunotherapy for cancer has progressed from the delivery of unspecific cellular products to the transfer of engineered tumor-specific T-cells. Adoptive cellular immunotherapy for cancer has now reached a stage of increasing feasibility and efficacy.
The idea that immunocompetent cells are capable of mediating an antitumor effect was first validated experimentally in 1957 by Barnes and Loutit  who showed that leukemic animals that were lethally irradiated and reconstituted with allogeneic bone marrow had a lower tumor burden following transplantation than animals that were reconstituted with syngeneic marrow. In 1973, Bortin and colleagues [2,3] attempted to quantify the immunologic antitumor effect, which they called "Graft versus Leukemia, GvL-effect," of donor lymphocytes. These observations in animal models led Mathé and colleagues  to speculate that leukocyte transfusions could mediate antitumor effects in cancer-bearing recipients. To test this hypothesis, pooled white cell products were transfused into non-transplanted patients with end-stage acute leukemia, which resulted in responses . Meanwhile, a series of clinical observations provided evidence for GvL activity in humans. These included a higher relapse rate in recipients of syngeneic compared with allogeneic transplants , a reduction in relapse rates in patients with graft-vs-host disease (GvHD) compared with those free of GvHD , induction of remission in patients after withdrawal of immunosuppression , and higher relapse rates in recipients receiving T-cell-depleted grafts compared with unmanipulated grafts .
The first patient to receive donor lymphocyte infusions (DLI) for a hematologic malignancy in relapse after Bone Marrow Transplantation (BMT) was a boy with acute lymphocytic leukemia (ALL) that was resistant to chemotherapy and cytokines . He ultimately obtained a sustained complete remission of his disease by receiving multiple transfusions of lymphocytes from his sister, the original bone marrow donor. Then, Kolb and colleagues  reported on 3 patients with relapsed chronic myeloid leukemia (CML) who failed to respond to treatment with interferon-alpha but who obtained complete remissions with the combination of IFN-alpha plus DLI. Thus, the era of adoptive immunotherapy to treat post-transplant relapse of hematologic malignancies was born.
Several pathways by which the GvL effect may eliminate tumor cells have been suggested . Extensive analysis in both human and animal studies has shown that GvL activity is mediated primarily by T-cells and NK cells, although other cells can also contribute through direct or indirect mechanisms. Cytotoxic T-cells execute their function by direct killing of leukemia cells via the perforin/granzyme system or by triggering the target cell’s own apoptotic pathways via cytokines, such as tumor necrosis factor-alpha and IFN-gamma. In some cases, donor T-cells can recognize a tumor-specific antigen (eg, Bcr/Abl in CML) or a minor histocompatibility antigen, such as a polymorphic antigen with restricted expression on hematopoietic cells (eg, HA-1/2); however, the major contributors to the GvL activity of donor T-cells are likely alloreactive T-cells recognizing alloantigens on tumor cells and normal tissue cells in the recipient. Clinical studies support this notion and have demonstrated an inverse correlation between GvHD (especially chronic GvHD) and the risk of post-transplantation relapse. However, DLI-induced leukemia responses can occur in the absence of GvHD and it is still difficult to separate GvL from GvH mechanistically.
The best responses to DLI occur in patients with CML. There are important issues that we have learned from the DLI treatment strategy in relapsed CML. First, the most powerful predictor of the response to DLI is disease stage. Thus, DLI induces cytogenetic complete remissions in 81% of patients whose relapse was detectable only by cytogenetic analysis, but in only 33% of patients whose disease had transformed to either accelerated phase or blast crisis [13,14]. Second, the T-cell dose, expressed as the number of CD3+ cells per kilogram of recipient body weight, seems to influence response rates, at least in relapsed CML. In two elegant studies Mackinnon and colleagues [13,14] found that after HLA-identical sibling BMT for CML a total CD3 dose >107 cells/kg given in an escalation manner (and not as a single bulk dose regimen) provide the best response rates while GvHD rates remain low. Third, GvL after DLI in CML is prolonged. The time to complete cytogenetic response and molecular response following the initial infusion of DLI is 8-16 weeks and >6 months, respectively. Patients with other malignancies respond less frequently to DLI. Response rates of 25-50% have been reported in relatively indolent diseases like multiple myeloma, chronic lymphocytic leukemia and other low grade lymphomas . In acute myeloid (AML) leukemia, remissions have been documented even less frequently (about 21%) . Remissions after DLI are generally of short duration, and long-term survival is <20% .
In two large retrospective series of DLI published in the late 1990s, response rates in AML were only 15-29% [18,19]. DLI from unrelated donors maybe associated with a higher response rate than that seen from related donors, with one series reporting 42% of patients achieving a complete response after unrelated DLI . When DLI has been used for recurrent or persistent AML or MDS after nonmyeloablative transplantation (Seattle protocol), the response rates have been low and the 1-year survival rate was only 11% for AML and 8% for MDS . The most recent comprehensive assessment of DLI in AML was conducted by the European Group for Blood and Marrow Transplantation and published in 2007 . In this retrospective analysis, estimated survival at 2 years was superior in the 171 patients receiving DLI compared with the 228 patients who did not receive DLI (21% vs 9%), despite the fact that a smaller number of patients went on to second allogeneic haematopoietic stem cell transplantation in the DLI group compared with the non-DLI group (13 vs 33). After adjustment for all the clinical variables in the two groups, DLI administration (P=0.04) appeared to be associated with improved outcome, together with younger patient age (P=0.008) and relapse occurring more than 5 months following transplantation (P<0.0001). Among AML patients who received DLI, marrow blast count at relapse, female gender, favourable cytogenetics and remission at the time of DLI were predictive of survival in multivariable analysis. For patients who received DLI in remission and had favourable cytogenetics, 2-year overall survival was estimated at 56%. In contrast, patients who received DLI during active disease had an OS of 9% to 20% (overall 15%), depending on other risk factors.
Limited efficacy of DLI for relapsed AML may be probably related to the rapid leukemia cell growth which may outpace the cytotoxicity of donor leukocytes. After donor cell infusion, the T -cells must recognize antigen, become activated, proliferate to a critical number of effectors and target the malignant cells before cell killing can occur. That interval may be too long and patients with relapsed acute leukemia may progress rapidly and die before a GvL effect is evident. Alternatively, lack of response to DLI in AML may be due to intrinsic differences in susceptibility of different tumor types to adoptive immunotherapy (lack of expression of recognizable tumor-specific antigens, lack of costimulatory molecules on malignant cells, inhibition of T cell activation or function or weak tumor cell killing). Other immune escape mechanisms of leukemic cells from GvL include defects in presentation of antigenic peptides, presence of areas of immunological sanctuary in the recipient (e.g. a significant fraction of patients relapsing after an initial response to DLI have done so in extramedullary sites) and downregulation of HLA molecules. In a very elegant paper published in New England Journal of Medicine, Vago et al.  reported that after haploidentical transplantation, leukemic cells can escape from the donor’s antileukemic T-cells through the loss of the mismatched HLA haplotype.
Most commonly, the DLI product is obtained by leukapheresis of unstimulated peripheral blood, containing approximately 2x108 mononuclear cells, of which about 50% are CD3+ cells, but also includes B-cells, dendritic cells and others. The composition of DLI is altered when G-CSF is given to the donors prior to apheresis or DLI are isolated from the G-CSF primed stem cell graft. G-CSF priming in normal donors increases the yield of CD3+ cells by almost threefold and changes the cytokine profile resulting in Th2-type differentiation, increase of Tregs and expansion of immature antigen presenting cells (APCs). Whether G-SCF mobilized DLI will affect GVHD or GvL remains to be determined.
Regarding dosing of DLI, no obvious dose–response relationship has been identified for diseases other than CML [18-20]. Similarly, there appears to be no relationship between DLI dose and GVHD. Although many investigators use a lower dose for unrelated DLI than for matched sibling DLI, limited retrospective data have identified neither a higher risk of GVHD for unrelated DLI nor a relationship between unrelated DLI dose and either response or toxicity.
Several unique methods explore the dual goals of improving the efficacy of DLI without increasing GvHD/Transplant Related Mortality (TRM). Small numbers of patients have been treated with each of these novel strategies and no solid conclusions can be made. These methods, however, appear promising and deserve further exploration. These include: (1) Ex vivo activation and expansion of donor T-cells through costimulation (activated DLI) . (2)Generation and infusion of minor histocompatibility antigen-specific T-cells . (3) Infusion of selected T-cell subsets (CD8+ cell depletion or CD4+ cell selection) . (4) Promote the in vivo clonal expansion of leukemia-reactive T-cells (e.g. via IL-2 administration) . (5) Enhancement of AML antigen presentation to T-cells (e.g. via administration of GM-CSF) . (6) Inactivation of alloreactive T-cells (that is, through transduction of suicide genes into donor T-cells, photochemical inactivation, chemotherapy inactivation, irradiation) [29,30]. (7) Infusion of regulatory cells as an attempt to minimize post DLI-GvHD. (8) Generation and infusion of Th2-type T-cells. (9) Lymphodepletion followed by DLI (eg promote in vivo expansion of the infused lymphocytes by providing lymphoid space and eliminating suppressive regulatory T-cells) . (10) Epigenetic modulation with low-dose 5-azacytidine followed by DLI (e.g. enhance immune effects through upregulation of HLA or cancer testis antigens on leukemic blasts) [32,33]. (11) Biological markers predicting the safety and efficacy of DLI. (e.g. CTLA-4 Single Nucleotide Polymorphism may be used as a surrogate marker for DLI outcome after allogeneic hematopoietic cell transplantation for acute leukemia) . (12) Since immunotherapy seems to be more effective in stages of impending, rather than overt relapse, perhaps the most appropriate platform for DLIs is in a prophylactic setting [35-37].
In order to make T-cell products more specific and potent, researchers seeked to generate leukemia specific T-cells by in vitro stimulation of donor cells with patients’ leukemia. Falkenburg and colleagues  were the first to show proof of principle that clonally selected antileukemic T-cells could induce long remissions in patients with CML relapsing after bone marrow transplantation. Accumulating clinical data provide evidence in support of the effectiveness of adoptive T-cell therapy to control cancer [39-42]. The group of Rosenberg pioneered the use of tumor infiltrating lymphocytes (TIL) to patients with melanoma  and Heslop et al. used virus-specific T-cells from peripheral blood to control and eliminate post-transplant EBV-specific lymphoproliferative disease . These early clinical trials have helped to understand the hallmarks of T-cell therapies and provided critical insights to the manufacturing of a therapeutically robust product, such as selection of specific tumor antigens (e.g. PR1, WT1)  and generation of T-cells with high avidity for specific tumor antigens and with specific phenotypes (e.g. both CD4 and CD8 T-cells, central memory subsets ). Next generation antigen-specific T-cells which show resistance to TGF-beta  or T-cells that were introduced to express a tumor-specific T-cell receptor (TCR)  or a chimeric antigen receptor (CAR, composed of a tumor targeting antibody fused to an activation molecule)  have been shown to overcome tolerance and are tested in trials presently under way at a number of academic institutions.
Although many hurdles remain unresolved, adoptive cellular immunotherapy for cancer has now reached a stage of increasing feasibility and efficacy . Development of facilities and experienced staffs for Good Manifacturing Practice (GMP) production of cellular immunotherapeutic products is currently extremely expensive. As T-cell manufacturing processes become more robust, closed and automated , it is anticipated that smaller facilities will be able to consistently manufacture a GMP cell product, leading to the wider dissemination of T-cell based therapies to patients (Figure 1).
This work has been supported by the Research Committee of Patras University within the framework of K. Karatheodoris scholarships (to AS). The authors declare no conflict of interest.