T cell therapy in allogeneic stem cell transplantation

Introduction

T cells derived from healthy stem cell donors can mediate curative immune responses after allogeneic stem cell transplantation (SCT) by eliminating (residual) hematopoietic tumor cells life-threatening viral infections. The first evidence for the crucial role of T cells came from the observations that depletion of T cells from a stem cell graft to prevent the development of acute graft versus host disease (GVHD) was associated with an increased incidence of relapse of the malignancy and life threatening viral infections including CMV disease and EBV mediated post-transplant lymphoproliferative disease (PTLD). Apparently, after allogeneic SCT T cells are not only responsible for the detrimental GVHD, but also can mediate a graft versus tumor (GVT) effect and are essential for long term of control of persistent endogenous viral infections. Unmanipulated T cell populations derived from the stem cell donor administered to the patient may result in concurrent development of GVHD, GVT and anti-virus responses. Immune suppression to control GVHD may lead to reduction or elimination of GVT and pathogen specific T cell responses. Obviously, if T cell responses leading to GVHD could be separated from GVL or pathogen-specific reactivity, morbidity and survival after allogeneic SCT can be significantly improved. One approach to manipulate the effect of donor T cells after transplantation has been the depletion of donor T cells from the graft with postponed administration of donor T cells ¹. The rationale for this approach came from several mouse and human studies illustrating that the recipient microenvironment plays a significant role in the development of GVHD². Tissue damage during the conditioning regimen leading to the initiation of a ‘cytokine storm’, the strong lymphodepletion after transplant leading to homeostatic proliferation of donor T cells shortly after transplant, and the presence of patient derived activated dendritic cells (DC) which are required for the induction of allo-immune responses all contribute to these strong allo-response immediately after transplantation. Postponed T cell administration at the time of replacement of the majority of recipient DC by donor DCs, after repair of the tissue damage, and after partial reconstitution of T cells in peripheral blood of patients is associated with lower risk of developing severe acute GVHD. Administration of T cells from 6 months after transplantation can be performed with relatively low risk of inducing severe GVHD, although chronic GVHD remains a significant complication. However, recurrence of the malignancy and the development of severe opportunistic infections frequently take place within the first half year after transplantation. Thus, although postponed administration of T cells can result in less severe GVHD, overall survival did not significantly improve.

Delivering stem cell transplants that are free from GVHD while conferring enhanced protection against malignant disease recurrence and viral reactivation is considered by many transplanters to be a key goal (the “Holy Grail”) for improving transplant success. We will discuss several approaches for the selective administration of non-GVHD reactive donor T cells: selective allodepletion, adoptive transfer of virus-specific T cells, allo-reactive anti-tumor T cells recognizing minor histocompatibility antigens (mHag) or MHC molecules, or T cells recognizing tumor-associated self antigens.

Selective allodepletion

It is possible to separate GvHD from GvL effects by co-incubating donor lymphocytes with allogeneic stimulator cells. Under these conditions, alloreactive donor cells can be selectively identified by their surface phenotype (eg CD25, CD69), proliferative potential, or preferential retention of photoactive dyes, and can be subsequently targeted for elimination using an immunotoxin, immunomagnetic bead separation, fluorescence-activated cell sorting, or photodynamic purging ³. Using standard proliferation assays, such as the MLR and the helper T-lymphocyte precursor (HTLp) frequency assay, alloreactivity can be substantially depleted in both the HLA-mismatched and HLA-matched setting, while maintaining third party responses. Responder cells obtained after allodepletion also maintain anti-tumor and anti-viral activity. This promising approach has been validated in a number of centers in animal models, and three clinical trials using a CD25 immunotoxin to selectively deplete alloreactivity (two in haploidentical PBSCT in pediatric patients and one in adults with hematological malignancies in HLA identical sibling transplants) indicate the concept is feasible, reducing GVHD while promoting rapid immune reconstitution ^4–7. To overcome fluctuations in activation-based surface marker expression and achieve a more consistent and profound allodepletion, we developed a photodepletion process targeting activation-based changes in p-glycoprotein that result in an altered efflux of the photosensitizer TH9402 ⁸. Third-party responses were maintained while the response to the original stimulator was reduced 2–3 logs. Photodepletion is currently being investigated in several clinical trials in Europe and North America.

Adoptive transfer of virus-specific T cells

An alternative approach to allodepletion that has been evaluated for the prevention and treatment of viral infection is to generate cytotoxic T cells ex vivo for subsequent transfer to the recipient by repeated stimulation with antigen presenting cells expressing viral antigens. One prerequisite for this strategy is knowledge of which viral antigens are expressed and are the most crucial for viral persistence and thus suitable targets for immunotherapy strategies. The viral antigens expressed at different stages of viral infection have been well characterized for some viruses such as CMV and EBV but are less well defined for other viruses that cause morbidity. It is also necessary to have a source of the identified viral antigen suitable for use in GMP manufacture and an antigen presenting cell that will present peptides derived from this antigen and express costimulatory molecules to produce T-cell activation. Initial studies evaluating virus specific CTLs have focused on CMV and EBV as the immune response is well understood and viral antigens are defined ⁹.

CMV specific CTLs

CMV matrix protein pp65 contains immunodominant epitopes against a range of HLA backgrounds, and several studies have shown that CTLs specific for pp65 protect against CMV disease in humans ^10–14. These studies have used a variety of sources of CMV antigen including CMV antigen and lysate, peptides and an adenoviral vector encoding pp65 antigen. Antigen presenting cells have included dendritic cells, fibroblasts and peripheral blood mononuclear cells. Prophylaxis with adoptively transferred donor-derived, CMV specific CTL was first explored by Riddell and colleagues ¹⁰. When CD8+ T cell clones specific for the immunodominant viral matrix proteins, pp65 and pp150 were transferred to recipients of matched sibling grafts, there were no adverse effects, CMV specific immune responses were reconstituted and no patients developed CMV disease or late recurrences. Other studies have confirmed the ability of adoptively transferred CTLs to prevent and treat CMV reactivation ^11,13,14.

EBV specific T cells

In EBV PTLD in SCT recipients, the transformed cells are of donor origin and express all latent cycle virus-associated antigens, providing excellent targets for virus-specific T cells. EBV transformed B lymphoblastoid cell lines (LCL) that express the same array of viral proteins can be readily prepared from any donor and provide a source of antigen-presenting cells that endogenously express the appropriate antigens for presentation of HLA class I-restricted epitopes. Our group generated EBV-specific T cell lines from donor lymphocytes and used them as prophylaxis and treatment for EBV-induced lymphoma in patients post SCT ¹⁵. The resultant EBV-specific CTL are polyclonal and contain both CD4− and CD8− positive EBV-specific T cells which is considered advantageous since the presence of antigen-specific CD4-helper T cells is important for in vivo survival of cytotoxic CD8-positive T cell populations. These CTLs have been administered as prophylaxis or therapy for EBV lymphoma in high-risk SCT recipients and have survived for up to 86 months after infusion and were able to reduce the high virus load that is observed in about 20% of patients. EBV-CTL also appeared to prevent progression to EBV-lymphoma, since none of 60 patients who received prophylactic CTL developed this malignancy, compared with 11.5% of controls. Further, 5 of 6 patients who received CTL as treatment for overt lymphoma achieved complete remissions. In the patient who failed to respond, the tumor was transformed with a virus that had deleted the two CTL epitopes for which the donor CTL line was specific. Other studies have confirmed the activity of EBV specific CTLs post transplant ¹⁶.

Multivirus specific CTLs

Although effective these strategies only target one of the many viruses that cause morbidity post transplant. We have therefore developed an approach to generate CTL that can restore cellular immunity to CMV, EBV and adenovirus simultaneously ¹⁴ by using mononuclear cells transduced with a recombinant adenoviral vector encoding the CMV antigen pp65 for the initial stimulation followed by stimulations with EBV-lymphoblastoid cell lines transduced with the same vector. 14 of 15 CTL lines generated showed specific activity against all three viruses and one line recognized CMV and EBV but not adenovirus targets. We have treated 14 patients in a phase I prophylaxis study, and observed up to an 8-fold increase in CMV-and EBV- specific T cells and these patients were able to control viral reactivations. We observed an increase in adenovirus-specific T cells in 5 patients who had evidence of adenovirus infection pre infusion and all were able to clear their infections. Of note one of these patients who received CTLs when he had progressive adenoviral pneumonia had a rise in adenovirus-specific T cells post-infusion coincident with a complete recovery. Multivirus CTLs can therefore expand in response to viral challenge and produce clinically relevant effects against all three viruses. We have also evaluated the use of bispecific lines targeting adenovirus and EBV in 12 recipients of CMV seronegative products ¹⁷. Again increased EBV-specific T cell frequency was detected in all patients, but rises in adenovirus-specific T cell frequency was only seen in patients with active infection. Expansion of virus specific CTLs may therefore require the presence of antigen to stimulate the infused cells. We are currently modifying the construct to express additional viral antigens to provide broader coverage.

Extending the Applicability of Virus Specific CTLs

While these studies effectively demonstrate proof of principle the wider application of this approach is limited by the time required for CTL generation using current good manufacturing processing techniques. Rapid selection techniques such are tetramer selection and gamma-interferon capture are therefore being evaluated. CMV-pp65-specific T cells from stem cell donors were selected from peripheral blood using tetramers and after direct infusion into patients expanded by several logs and reconstituted immunity to CMV ¹². A concern with this approach is that the product has limited specificity for one epitope and is only available for some HLA types. Gamma interferon capture has been used to isolate adenovirus specific donor T cells which were infused into nine children with systemic adenovirus infection with responses in five of six evaluable patients ¹⁸. An alternative approach is to use closely matched banked allogeneic lines which could be available as an “off the shelf” product. A concern with this approach is that persistence of a mismatched product may be suboptimal. However a recent Phase II study using EBV-specific CTLs to treat PTLD after solid organ transplant or SCT has shown an encouraging response rate of 64% with better responses with more closely matching lines ¹⁹.

Treatment of hematological malignancies with alloreactive T cells

The observation that autologous stem cell transplantation and allogeneic stem cell transplantation using homozygous twins as donors did not result in a similar control of the hematological malignancy after transplantation as compared to allogeneic SCT has indicated that the mere presence of T cells in the graft is not sufficient to elicit a GVT reactivity ²⁰. Apparently, genetic differences between donor and patients strongly contribute to not only GVHD but also a GVT response. After partially HLA matched allogeneic SCT, the immune response may likely to be directed against the mismatched HLA allele, since the number of alloreactive T cells in peripheral blood of normal individuals is relatively high with frequencies of up to 1% of circulating T cells. After HLA-matched allogeneic transplantation, minor histocompatibility antigens (mHag) are the most likely targets for the immune-reactivity after allogeneic SCT. Although alloreactive T cell responses against HLA or mHag can elicit both GVHD and GVT, characterization of the fine specificities of the T cells has indicates that GVHD responses can be separated from alloreactive GVT.

Minor histocompatibility antigens

MHag are defined as target antigens that are capable of eliciting an allogeneic immune response mediated by T cells from a fully HLA-identical donor. The human genome contains many single nucleotide polymorphisms (SNP). If a SNP is present within the coding region of a gene, this may lead to amino acid substitutions in the protein. If a peptide derived from a polymorphic gene is presented by HLA molecules, differences in the peptide content between donor and recipient may lead to an allo-immune response. The polymorphism in the gene can lead to differences in amino acid sequence of the peptides presented in the groove of the HLA molecules, to differential binding of peptides to the MHC molecule, or to differential processing of the peptide leading to the presence of absence of the potentially immunogenic peptide in HLA molecules. Donor T cells that are capable of recognizing immunogenic peptides on cells from the patient may lead to destruction of the cells expressing these polymorphic genes. Obviously, a prerequisite for the destruction of the target tissues by the immune response is the presence of the presenting HLA molecules on the target cells as well as the expression and processing of the protein encoding the mHag in these tissues. Thus, both the tissue distribution of the protein encoding the mHag as well as the expression of the MHC molecules will determine the likelihood of destruction of the patient-derived tissues.

MHag as targets for GVT-reactivity

After transplantation, normal hematopoiesis in the patient is replaced by donor hematopoiesis. Residual tumor cells are of recipient origin, and some residual normal hematopoietic cells may also be derived from the patient. An allogeneic T cell response recognizing polymorphic antigens expressed on hematopoietic cells from the recipient that are co-expressed on the hematopoietic tumor cells will lead to eradication of the tumor and elimination of residual recipient hematopoietic cells, without impairment of donor hematopoiesis in the patient. Whether or not this GVT response will be accompanied by GVHD may depend on the tissue distribution of the MHC molecules and the polymorphic protein encoding the mHag in non-hematopoietic tissues. Several mHag have been found to be restricted to hematopoietic tissues. For instance, the genes encoding the mHag HA-1, HA-2, HB-1, BCL2A1, LRH-1, PANE1 and LB-ECGF-1H have been found to be relatively restricted to cells of hematopoietic origin. Analyses of immune responses from patients successfully treated with donor lymphocyte infusion (DLI) have illustrated that the development of a T cell response against the hematopoiesis associated mHag can result in a potent anti-tumor effect, conversion to full donor chimerism, with no or only temporary GVHD ²¹. Since many hematopoietic cells including dendritic cells are present in normal tissues, an immune response against polymorphic antigens expressed on hematopoietic cells may lead to an inflammatory response in these tissues, but the severity of this immune response is likely to be low. Analyses of patients successfully treated with DLI have demonstrated that the kinetics of the immune response against mHag specific antigens associated with an anti-tumor effect, resembles the development of an immune response against viral infections. After initial rise of antigen specific T cells to high frequencies exceeding 1% of all circulating T cells, the immune response declines after eradication of the tumor and a memory response may develop. This memory response may be relevant for sustained and prolonged suppression of the tumor. Similarly, we have found strong T cell responses against class II associated antigens in patients successfully responding to DLI in the absence of GVHD. The limited distribution of MHC class II, mainly restricted to hematopoietic cells may allow the development of immune responses against class II antigens as relatively tumor-specific.

In vitro generation of mHag specific T cells for treatment of leukemia

Since we demonstrated a potential clinical effect of treatment of patients with refractory hematological malignancies with in vitro generated leukemia-reactive T cells ²², several approaches are being explored to isolate mHag specific T cells for adoptive transfer, including the isolation of antigen specific T cells using multimeric peptide/MHC complexes coupled to immunomagnetic beads, or based on specific production of INF-γ by activated T cells after antigen specific triggering ²³. Alternatively, large numbers of antigen specific T cells can be generated using T cell receptor (TCR) gene transfer. High affinity mHag specific TCRs have been characterized and isolated from high avidity T cells. Following transfer of the TCR α and β genes to primary donor T cells, the specificity of the TCR can be transferred to these T cells. Since pathogens specific T cells like CMV specific T cells are responsible for an effective immune response after transplantation with control of CMV, the possibility to transfer mHag specific TCRs to these virus-specific T cells has been explored ²⁴. Transfer of the TCRs resulted in combined reactivity of these re-engineered T cells against mHag and the pathogens. Since the persistent presence of CMV antigen or other relevant pathogens after transplantation in the recipient may lead to persistent activation of these specific transduced T cells, a memory T cell response may be obtained allowing persistent immune surveillance against both pathogens and residual tumor cells.

Leukemia specific antigens as targets for GVL

In addition to mHag, tumor cells can overexpress proteins that are often intrinsic to the maintenance of the malignant phenotype. Peptide sequences from these proteins can be antigenic by virtue of their overexpression (eg proteinase 3 (PR3) and human neutrophil elastase (HNE)), aberrant expression (eg onco-fetal antigens such as Wilms tumor 1), or uniqueness (eg fusion proteins such as the BCR-ABL breakpoint sequence of the Ph chromosome). Because they are not usually alleleic, peptides derived from these proteins behave as self antigens rather than mHag ²⁵. Nevertheless it is clear that low frequencies of CTL recognizing PR3, HNE and WT1 (but not BCR-ABL) circulate in normal individuals and are increased in patients with leukemia and more so after allogeneic SCT ^26,27. These studies suggest that stimulation by leukemia, especially after SCT, can induce expansion of antigen-specific effector-memory CTL from a pool of antigen-experienced central memory cells. The occurrence of low frequencies of leukemia specific CTL in normal donors has encouraged clinical trials with peptide vaccines against HLA A0201 and HLA A 2401 restricted peptides of PR3 and WT1 ²⁸. Notable responses including complete remissions have been reported using PR1 peptide vaccine to treat patients with relapsed myeloid malignancies after allogeneic SCT ²⁵. The demonstration that leukemia-antigen specific T cell clones from the donor can transfer and expand in the recipient supports the development of new strategies for the adoptive transfer of GVL immunity. Firstly, vaccination of a donor not yet tolerized to the leukemia could increase the frequency of leukemia-specific T cells conveyed in the graft. Second the cytokine storm and the lymphopenia that follows the conditioning regime may be favorable to the expansion of these leukemia specific CTL which could be further boosted by early vaccination of the transplant recipient.

Future prospects - combining strategies for the “perfect transplant”

Although the approaches we have described to shape immune reconstitution after SCT are currently being developed separately at different centers, the successful introduction of these individual strategies could be only a prelude to the design of future transplants where the techniques are combined to optimize GVT effects and antimicrobial immunity. For example vaccination of the donor with leukemia-antigen specific peptides could be used to induce antigen-specific T cells in donor lymphocytes for further expansion in vitro for adoptive transfer to the recipient. Selectively depleted T cell transplants (which would not require post-transplant immunosuppressive GVHD prophylaxis) would be an ideal platform for the adoptive transfer of antigen-specific T cells in the first few weeks after SCT when the lymphopenic milieu best favors antigen-specific T cell expansion. Finally adoptively transferred T cells could be further boosted by the relevant vaccine. Although we are aware that the careful construction of post-transplant immunity is likely to remain technically challenging and expensive, the effort taken to create the “perfect transplant” will be justified economically if it is demonstrated that most patients have a complication-free post-transplant course, without disease recurrence.