Graft-versus-leukemia Reactivity Research Articles

AML-142 Patients at High Risk of Relapse Post-Transplant: A Phase 1 Study Design With a Novel Treatment Strategy Using the Estimand Framework

The risk of disease relapse in patients allografted for acute myeloid leukemia (AML) is high and remains the main cause of transplant failure. A potent graft-versus-leukemia (GvL) effect is exerted post-stem cell transplantation (SCT); strategies that may augment a GvL effect have the potential to reduce relapse risk post SCT. Siremadlin/HDM201 is a novel investigational MDM2 inhibitor with single-agent (SA) anti-AML activity. MDM2 inhibitors possess potent immunomodulatory effects (IME) in murine solid tumor and AML models. The post-allogeneic SCT setting is ideal for investigating IME of MDM2 inhibition on the GvL effect. To present a study design assessing the safety and preliminary efficacy of treatment with monotherapy siremadlin and in combination with donor lymphocyte infusions (DLIs) for AML post SCT using the estimand framework (ICH E9 [R1] addendum). Phase Ib/II proof-of-concept. AML patients in CR post SCT at high risk for relapse based on pre-SCT risk factors. The estimand framework mandates definition of experimental intervention according to the clinical question. Due to the risk of GvHD with DLI early post SCT, siremadlin will first be given to eligible patients as priming monotherapy to prevent early relapse; patients who tolerate siremadlin with no evidence of GvHD are eligible to start siremadlin/DLI. Post combination, patients may continue maintenance SA siremadlin, which intends to prolong/enhance the GvL reaction to eradicate residual leukemic blasts. According to their treatment journey, patients may receive only part of the intervention's sequence (e.g., siremadlin priming monotherapy patients not eligible for DLI combination will continue priming). To determine 1) the siremadlin recommended dose (RD) across all parts of the treatment strategy and 2) preliminary efficacy regardless of the treatment parts received. Siremadlin RD will be determined using a Bayesian logistic regression model to guide dose finding of siremadlin during priming/Cycle 1 and a Bayesian time-to-DLT model to assess the siremadlin combination dose. Efficacy will be assessed by the proportion of patients remaining in CR ≥6 months. The estimand framework guided the design of an AML clinical study post SCT to conceptualize the assessment of safety and efficacy of the treatment strategy, accounting for different patient journeys.

P583: PATIENTS AT HIGH RISK OF RELAPSE POST-TRANSPLANT: A PHASE 1 STUDY DESIGN WITH A NOVEL TREATMENT STRATEGY USING THE ESTIMAND FRAMEWORK

Background: The risk of disease relapse in patients (pts) allografted for acute myeloid leukemia (AML) is high and remains the main cause of transplant failure. The great majority of pts destined to relapse post-allogeneic stem cell transplant (SCT) will do so within the first 12 months and survival after early relapse is dismal. Therefore, strategies to prevent relapse are urgently required. A potent graft-versus-leukemia (GvL) effect is exerted after SCT, and strategies that may augment a GVL effect have the potential to reduce relapse risk post-SCT. Siremadlin (HDM201) is a novel investigational MDM2 inhibitor with single-agent (SA) anti-AML activity. MDM2 inhibitors, including siremadlin, possess potent immunomodulatory effects (IME) in murine solid tumor and AML models. The post-allogeneic SCT setting is ideal for investigating the IME of MDM2 inhibition on enhancing GvL effect. Here we present a phase Ib/II proof-of-concept study designed to assess safety and preliminary efficacy of a treatment strategy with siremadlin as monotherapy and, subsequently, in combination with donor lymphocyte infusions (DLIs) for AML in post-SCT setting using the estimand framework (ICH E9 (R1) addendum). Aims: The objective of the planned treatment strategy is to enhance GvL post-SCT by siremadlin as monotherapy and in combination with DLIs. Due to the risk of Graft-vs-Host Disease (GvHD) with DLI early after SCT, siremadlin will be given as priming monotherapy to prevent early relapse prior to starting siremadlin/DLI combination therapy. SA siremadlin maintenance after the combination therapy is intended to prolong and enhance the GvL reaction to eradicate residual leukemic blasts. Methods: The estimand framework mandates a careful definition of the experimental intervention according to the clinical question of interest. In this trial, the experimental intervention follows a treatment strategy: Patients (pts) with AML in complete remission (CR) post-SCT but at high risk for relapse based on pre-SCT risk factors will start with siremadlin priming monotherapy. Pts who tolerate the siremadlin dose with no evidence of GvHD are eligible to start siremadlin/DLI combination therapy. After combination therapy, pts may continue with SA siremadlin maintenance. Depending on the treatment journey, pts may receive only a part of the intervention’s sequence (e.g., pts in siremadlin priming monotherapy who are not eligible for DLI combination will continue the priming part). The clinical questions addressed in this trial are to determine the recommended dose (RD) of siremadlin across all parts of the treatment strategy and to assess preliminary efficacy regardless of the parts of the treatment strategy pts received. The doses of interest were preselected based on SA studies and in pts who received siremadlin after SCT. Results: The proposed approach is tailored to investigate safety and efficacy of the treatment strategy. The RD of siremadlin will be determined over 2 parts: A Bayesian logistic regression model will guide dose finding of siremadlin during priming (Cycle 1) and a Bayesian time-to-DLT model will assess the siremadlin dose for combination. Efficacy will be assessed by proportion of pts remaining in CR for at least 6 months. Image:Summary/Conclusion: In clinical practice, treatment decisions follow a strategy. The estimand framework guided the design of a clinical study in AML post SCT to conceptualize the assessment of a treatment strategy. Accordingly, safety and efficacy of the treatment strategy as a whole will be assessed taking into account different patient journeys.

In silico analysis of T-cell receptors specifi c to the minor histocompatibility antigen HA-2

Hematopoietic stem cells transplantation of (HSCT) from the related or unrelated donor is used as a treatment for hematopoietic system malignancies. However, transplantation triggers an immune response of the donor cells to the recipient’s antigens. The response to the healthy tissues is called a graft versus host reaction (GVHD), and the response to the hematopoietic tissue in the context of malignant disease is called a graft versus leukemia (GVL) eff ect. The development of GVL reactivity is a favorable consequence of transplantation, since it eliminates residual tumor cells and prevents the relapse. It was demonstrated that immune response arises towards polymorphic peptides, presented in the molecules of the major histocompatibility complex (HLA). Such peptides are derived from the proteasomal degradation of proteins expressed from the genes with non-synonymous single nucleotide polymorphisms and are referred to as minor histocompatibility antigens (MiHA). Studying the structure of T- cell receptor (TCR) repertoires that recognize MiHAgs can help identify the mechanisms for the formation of the alloreactive response and is important for predicting the antigen of alloreactive clones with unknown specificity. In this article the genetic sequences encoding T-cell receptors specific to the HA-2 minor antigen were determined and analyzed in silico. We found the predominant use of the V21 and J42 segments in the formation of the CDR3 region of the α-chain and the presence of the V7-8 segment in most CDR3 β-chain regions, which indicates the existence of a conservative motif responsible for recognizing the HA-2 antigen.

Selective graft-versus-leukemia depends on magnitude and diversity of the alloreactive T cell response.

Patients with leukemia who receive a T cell-depleted allogeneic stem cell graft followed by postponed donor lymphocyte infusion (DLI) can experience graft-versus-leukemia (GVL) reactivity, with a lower risk of graft-versus-host disease (GVHD). Here, we have investigated the magnitude, diversity, and specificity of alloreactive CD8 T cells in patients who developed GVL reactivity after DLI in the absence or presence of GVHD. We observed a lower magnitude and diversity of CD8 T cells for minor histocompatibility antigens (MiHAs) in patients with selective GVL reactivity without GVHD. Furthermore, we demonstrated that MiHA-specific T cell clones from patients with selective GVL reactivity showed lower reactivity against nonhematopoietic cells, even when pretreated with inflammatory cytokines. Expression analysis of MiHA-encoding genes showed that similar types of antigens were recognized in both patient groups, but in patients who developed GVHD, T cell reactivity was skewed to target broadly expressed MiHAs. As an inflammatory environment can render nonhematopoietic cells susceptible to T cell recognition, prevention of such circumstances favors induction of selective GVL reactivity without development of GVHD.

Angiogenic cytokines are antibody targets during graft-versus-leukemia reactions.

The graft-versus-leukemia (GVL) reaction is an important example of immune-mediated tumor destruction. A coordinated humoral and cellular response accomplishes leukemia cell killing, but the specific targets remain largely uncharacterized. To learn more about the antigens that elicit antibodies during GVL reactions, we analyzed patients with advanced myelodysplasia (MDS) and acute myelogenous leukemia (AML) who received an autologous, granulocyte-macrophage colony-stimulating factor (GM-CSF)-secreting tumor cell vaccine early after allogeneic hematopoietic stem cell transplantation (HSCT). A combination of tumor-derived cDNA expression library screening, protein microarrays, and antigen-specific ELISAs were used to characterize sera obtained longitudinally from 15 patients with AML/MDS who were vaccinated early after allogeneic HSCT. A broad, therapy-induced antibody response was uncovered, which primarily targeted intracellular proteins that function in growth, transcription/translation, metabolism, and homeostasis. Unexpectedly, antibodies were also elicited against eight secreted angiogenic cytokines that play critical roles in leukemogenesis. Antibodies to the angiogenic cytokines were evident early after therapy, and in some patients manifested a diversification in reactivity over time. Patients that developed antibodies to multiple angiogenic cytokines showed prolonged remission and survival. These results reveal a potent humoral response during GVL reactions induced with vaccination early after allogeneic HSCT and raise the possibility that antibodies, in conjunction with natural killer cells and T lymphocytes, may contribute to immune-mediated control of myeloid leukemias.

Collateral Damage of Nonhematopoietic Tissue by Hematopoiesis-Specific T Cells Results in Graft-versus-Host Disease During an Ongoing Profound Graft-versus-Leukemia Reaction

After allogeneic stem cell transplantation (allo-SCT), donor T cells may recognize minor histocompatibility antigens (MiHA) specifically expressed on cells of the recipient. It has been hypothesized that T cells recognizing hematopoiesis-restricted MiHA mediate specific graft-versus-leukemia (GVL) activity without inducing graft-versus-host disease (GVHD), whereas T cells recognizing ubiquitously expressed MiHA induce both GVL and GVHD reactivity. It also has been hypothesized that alloreactive CD4 T cells are capable of mediating specific GVL reactivity due to the hematopoiesis-restricted expression of HLA class II. However, clinical observations suggest that an overt GVL response, associated with expansion of T cells specific for hematopoiesis-restricted antigens, is often associated with GVHD reactivity. Therefore, we developed in vitro models to investigate whether alloreactive T cells recognizing hematopoiesis-restricted antigens induce collateral damage to surrounding nonhematopoietic tissues. We found that collateral damage to MiHA-negative fibroblasts was induced by misdirection of cytotoxic granules released from MiHA-specific T cells activated by MiHA-positive hematopoietic cells, resulting in granzyme-B–mediated activation of apoptosis in the surrounding fibroblasts. We demonstrated that direct contact between the activated T cell and the fibroblast is a prerequisite for this collateral damage to occur. Our data suggest that hematopoiesis-restricted T cells actively participate in an overt GVL response and may contribute to GVHD via induction of collateral damage to nonhematopoietic targets.

Graft Versus Leukemia Separates From Graft Versus Host Disease By Magnitude and Avidity Of The Allo-Reactive T Cell Response After Allogeneic Stem Cell Transplantation and Donor Lymphocyte Infusion

Donor lymphocyte infusion (DLI) after allogeneic stem cell transplantation (alloSCT) can be a curative treatment for patients with hematological malignancies due to the capacity of allo-reactive donor derived T cells to mediate a curative potent graft versus leukemia (GVL) effect. However, associated acute graft versus host disease (GVHD) remains a major risk. To study the role of CD8+ T cells in GVL reactivity and GVHD, we selected patients who responded to DLI (without preceding cytoreductive treatment) for recurrent disease or incomplete donor chimerism after alloSCT. The patients were grouped according to absence (7 patients) or presence (6 patients) of GVHD. To quantify the number of circulating activated CD8+ T cells before DLI and at the time of disease regression or conversion to full donor chimerism we measured the frequencies of CD8+ HLA-DR+ T cells in peripheral blood samples by flowcytometry. Before DLI, highly variable numbers of CD8+ HLA-DR+ T cells were found (37.8 ± 42.9 x106/L), that significantly increased after DLI (309±473 x106/L, p<0.005), demonstrating involvement of CD8+ HLA-DR+ T cells in immune responses after DLI. To determine the specificity and functional avidity of the CD8+ HLA-DR+ T cells, these cells were isolated using flowcytometric cell sorting and clonally expanded. From a total of 30 samples, on average 225 T cell clones per sample were obtained and tested for recognition of patient and donor derived EBV-LCL, CD40L stimulated B cells (CD40L-B cells) and monocyte derived dendritic cells (monoDC). Surprisingly, in many samples from both patient cohorts high percentages of clones recognizing EBV-LCL derived from both patient and donor but not recognizing CD40L-B cells and monoDC were found. These T cells may be involved in anti-EBV responses irrespective of the presence of a GVL effect or GVHD. To investigate whether the magnitude of the allo-immune response was different in patients with or without GVHD coinciding the GVL effect, we compared the frequencies of allo-reactive T cell clones in samples from both patient groups. Significantly lower percentages of allo-reactive T cell clones were found in patients without GVHD as compared to patients with GVHD (5.1 ± 7.0% versus 32.5 ± 20.0% respectively, p<0.01), showing that coinciding GVHD is associated with an increased magnitude of the allo-reactive T cell response. Per patient, we determined the number of unique antigens targeted by the isolated T cell clones by characterizing the targeted MiHA using whole genome association scanning. In line with the lower total number of allo-reactive T cells, a lower number of unique MiHA was targeted in patients without GVHD (2.7±3.5) as compared to patients with GVHD (10.2±5.8, p=0.015). To determine whether occurrence of GVHD could be explained by the tissue specificity and functional avidity of the allo-reactive T cell response after DLI, we tested the T cell clones obtained from both patient cohorts for recognition of fibroblasts (FB) derived from skin biopsies of the patient. To mimic pro-inflammatory conditions, FB were pretreated for 4 days with 100 IU/ml IFN-γ. Recognition of untreated FB was exclusively mediated by T cell clones obtained from patients with GVHD, whereas recognition of IFN-γ pretreated FB was found for clones isolated from patients with or without coinciding GVHD. In addition, several T cell clones isolated from patients without GVHD were found to be directed against MiHA encoded by genes with a broad expression profile in non-hematopoietic cells comprising FB, despite absence of FB recognition under non-inflammatory conditions. This suggests that in addition to the tissue expression profile of the MiHA other factors, comprising the local inflammatory milieu, play a role in the risk of developing GVHD. In conclusion, our data show a strong correlation between the magnitude and the functional avidity of the allo-reactive CD8+ T cell response and the occurrence of GVHD after DLI. We hypothesize that the limited production of pro-inflammatory cytokines due to the moderate magnitude of the immune response in patients mounting a GVL response without coinciding GVHD reactivity may have prevented the induction of GVHD by the lower avidity allo-reactive T cells, that under pro inflammatory conditions can mediate GVHD by recognition of normal non-hematopoietic cells of the patient. Disclosures:No relevant conflicts of interest to declare.

LB-ARHGDIB-1R As Novel Minor Histocompatibility Antigen For Therapeutic Application

Patients with hematological malignancies can be successfully treated with allogeneic hematopoietic stem cell transplantation (alloSCT). Beneficial Graft-versus-Leukemia (GvL) reactivity, however, is often accompanied with undesired Graft-versus-Host Disease (GvHD). In HLA-matched alloSCT, donor T-cells can mediate GvL reactivity by recognition of minor histocompatibility antigens (MiHA) on malignant cells, but also GvHD when MiHA are recognized on non-hematopoietic tissues. To decrease the incidence and severity of GvHD, T-cells can be (partially) depleted from the graft and re-administered later as donor lymphocyte infusion (DLI). MiHA are polymorphic peptides that are presented in HLA-molecules and can be recognized as non-self by donor-derived T-cells. Only a minority of MiHA have therapeutic relevance based on hematopoietic-restricted expression and only 25% and 40% of recipients transplanted with sibling and unrelated donors, respectively, are eligible for T-cell therapies in which one of the known hematopoietic-restricted MiHA is targeted. Therefore, to increase the efficacy and limit the toxicity of T-cell therapy, more therapeutic MiHA are needed. Recently, we identified a MiHA encoded by ARHGDIB, a gene that has been described to be highly expressed on hematopoietic cells, and we here studied the therapeutic relevance of LB-ARHGDIB-1R in detail. First, we confirmed hematopoietic-restricted expression of ARHGDIB by microarray gene expression analysis and demonstrated &gt;10-fold overexpression in the majority of malignant and healthy hematopoietic versus non-hematopoietic cells. In line with its hematopoietic-restricted gene expression profile, LB-ARHGDIB-1R in the context of HLA-B* 07:02 was specifically recognized on different hematological malignancies, but not on non-hematopoietic fibroblasts and keratinocytes cultured in the absence or presence of IFN-γ, which was added to mimick the pro-inflammatory cytokine milieu of the early post-transplantation period. Next, we investigated the cytolytic capacity of LB-ARHGDIB-1R specific T-cells, and demonstrated specific lysis of patient, but not donor, EBV-B cells and specific lysis of an ALL sample in a 10 hrs 51Cr-release assay. Specific lysis of additional ALL and AML samples could be measured after 48 hrs of co-incubation in a flowcytometry-based cytotoxicity assay. To determine the in vivo immunogenicity of LB-ARHGDIB-1R, 11 MiHA-disparate HLA-B* 07:02 positive patient-donor pairs were screened for specific CD8+ T-cells by dual color tetramer analysis. All patients received partial T-cell depleted alloSCT and sampling was done at different time points after alloSCT (and DLI). In 4 out of 11 patients, LB-ARHGDIB-1R-specific T-cells (&gt;0.01%) could be detected directly ex vivo and in 4 additional patients after 7 days of in vitro peptide stimulation, indicating that LB-ARHGDIB-1R is highly immunogenic. High frequencies of LB-ARHGDIB-1R specific T-cells were measured ex vivo in a patient whose hematological relapse was successfully treated with DLI, without development of GvHD, further supporting the therapeutic relevance of LB-ARHGDIB-1R. In summary, we confirmed hematopoietic-restricted expression of LB-ARHGDIB-1R and demonstrated T-cell mediated lysis of primary leukemic cells in long-term co-incubation assays. Furthermore, we showed that LB-ARHGDIB-1R is highly immunogenic and that specific T-cells could be detected in a patient who responded to DLI in the absence of GvHD. Altogether, our data support the clinical relevance of LB-ARHGDIB-1R as therapeutic MiHA with the potential to shift the delicate balance between GvL and GvHD in favor of a desired anti-tumor effect. Disclosures: No relevant conflicts of interest to declare.

Influence Of Killer Immunoglobulin-Like Receptor (KIR) and HLA Genotypes On Outcomes After Reduced-Intensity Conditioning Allogeneic Hematopoietic Stem Cell Transplantation For Patients With AML and MDS: A Report From The Center For International Blood and Marrow Transplant Research Immunobiology Working Committee

Disease relapse is a significant cause of treatment failure after allogeneic hematopoietic stem cell transplantation (HCT). In the setting of reduced-intensity conditioning (RIC), a graft-versus-leukemia (GVL) effect is critical for successful outcomes in patients with advanced myeloid malignancies. A GVL effect has been attributed in part to donor-derived alloreactive natural killer (NK) cells, which are regulated by interaction of KIRs with their HLA-class I ligands. Several models of NK reactivity (missing KIR ligand, centromeric haplotype-B content, non-tolerized KIR2DS1) have been associated with improved outcomes following HLA-matched, HLA-mismatched, related, and unrelated donor HCT, particularly for AML patients given myeloablative conditioning. The effect of KIR-HLA combinations on outcomes after RIC HCT, however, is not known. We retrospectively analyzed donor-recipient KIR/HLA genotypes from 929 URD HCTs facilitated by the National Marrow Donor Program for patients with AML (n=624) or MDS (n=305) treated with RIC between 1990 and 2007. 664 donor-recipient pairs were 10/10 HLA-matched and 265 were 9/10 HLA-matched. 332 (37%) patients received ATG and 73 (8%) received alemtuzumab. P-values less than 0.01 were considered significant. We hypothesized that donor-recipient KIR-HLA interactions may be associated with improved post-transplant outcomes following RIC HCT for AML and MDS. Patients lacking the HLA-Bw4 ligand for donor inhibitory KIR3DL1 experienced lower relapse following HCT at 1 year (27 vs 36%, p=0.002) and 5 years (31 vs 42%, p=0.003) compared to patients with the Bw4 ligand (Figure). The lower risk for relapse was confirmed in multivariate analysis (HR 0.71, p=0.005). However, there was no significant association of Bw4 ligand with disease-free survival (HR 0.84, p=0.05) and overall survival (HR=0.97, p=0.70). Risk for acute GVHD was higher among patients lacking KIR ligands after adjusting for other clinical factors. In particular, patients lacking HLA-C2 for donor KIR2DL1 experienced higher grade 2-4 (HR 1.3, p=0.005) and 3-4 acute GVHD (HR 1.5, p=0.002), and patients lacking multiple KIR ligands experienced higher grade 3-4 acute GVHD (HR 1.5, p=0.007). The analysis was then restricted to AML patients, the patient population with greatest reported KIR-HLA effects. Patients whose donors were KIR2DS1+ and HLA-C2C2 (n=33) had higher transplant-related mortality (TRM) (HR, 2.4, p=0.002) compared to all other patients. There was no significant effect of KIR2DS1 with HLA on relapse. In a multivariate analysis, lack of HLA-C2 in AML patients was associated with higher grade 2-4 (HR 1.4, p=0.002) and 3-4 acute GVHD (HR 1.5, p=0.01), and risk for grade 3-4 acute GVHD was higher in patients lacking multiple KIR ligands (HR 1.6, p=0.005). There were no significant associations between donor homozygosity for the centromeric B-haplotype (cenBB) or overall B-haplotype KIR content and RIC HCT outcomes.Overall, these results suggest that in the RIC HCT setting, lack of Bw4 ligand for KIR3DL1 is associated with a lower risk of relapse for AML/MDS. This observation corroborates previous findings in myeloablative HCT transplants. Furthermore, tolerance of donor KIR2DS1 by HLA-C2C2 was associated with worse outcome, as manifested by higher TRM, in AML patients. In contrast, it appears that in RIC HCT homozygosity for the centromeric B-haplotype does not have a significant role in leukemia relapse. The associations of KIR ligands with acute GVHD were not previously observed suggesting that NK cell alloreactivity depends on multiple variables, including RIC. Further elucidation of the biology of NK cell alloreactivity in the RIC setting may provide guidance for future approaches to help optimize conditions for generating GVL reactions without GVHD and less TRM in this transplant population. [Display omitted] Disclosures:No relevant conflicts of interest to declare.

A new approach for eradication of residual lymphoma cells by host nonreactive anti–third-party central memory CD8 T cells

AbstractGeneration of T cells endowed with graft-versus-leukemia (GVL) and depleted of graft-versus-host (GVH) activity represents a highly desirable goal in bone marrow transplantation (BMT). Here, we demonstrate that donor anti–third-party CD8 T cells with central memory phenotype (Tcm) exhibit marked GVL reactivity through a unique T-cell receptor–independent mechanism. Thus, in a residual disease mouse model, Tcm therapy following autologous BMT led to significant survival prolongation, with 30% to 40% of the treated mice displaying long-term tumor-free survival. A more impressive finding was that infusion of donor Tcm in an allogeneic model rapidly eliminated residual lymphoma cells and led to long-term survival of 100% in the absence of GVH disease. Collectively, the strong GVL reactivity of anti–third-party Tcm, coupled with their demonstrated enhancement of bone marrow allografting, suggests that the use of Tcm therapy in conjunction with allogeneic T-cell–depleted BMT could be of particular benefit in patients with B-cell malignancies who cannot tolerate intensive myeloablative conditioning.

Donor T Cells Administered Over HLA Class II Barriers Mediate Antitumor Immunity without Broad Off-Target Toxicity in a NOD/Scid Mouse Model of Acute Leukemia

Alloreactive (allo)-HLA–directed T cell responses after HLA-mismatched allogeneic hematopoietic stem cell transplantation and donor lymphocyte infusion are typically considered detrimental responses mediating graft-versus-host disease (GVHD). Allo-HLA-reactive T cells with beneficial and selective graft-versus-leukemia (GVL) reactivity, however, can also be identified within an HLA-mismatched context. We investigated whether allo-HLA class II–directed T cells with beneficial GVL reactivity induced in NOD/scid mice engrafted with human chronic myelogenous leukemia in lymphoid blast crisis after treatment with donor lymphocyte infusion – mediated detrimental xenogeneic GVHD as a result of broad off-target cross-reactivity. The results demonstrate that beneficial GVL reactivity and xenogeneic GVHD are mediated by separate T cells. GVL reactivity was mediated by human T cells recognizing allo-HLA class II molecules, whereas xenoreactivity was exerted by human T cells recognizing H-2 molecules. Taken together, our data indicate a limited risk for detrimental off-target effects by allo-HLA class II–directed T cells and thereby provide a basis for the development of strategies for selecting allo-HLA–restricted T cells with selective GVL reactivity for adoptive transfer after HLA-mismatched allogeneic hematopoietic stem cell transplantation.

HLA Class II Disparity Is Necessary and Sufficient for Induction of Effective Anti-Tumor Immunity by Donor Lymphocyte Infusion in a NOD/Scid Mouse Model for Human Acute Lymphoblastic Leukemia

Abstract 648Donor lymphocyte infusion (DLI) can be a curative treatment for patients with relapsed hematological malignancies after HLA matched allogeneic stem cell transplantation (alloSCT). However, curative responses in patients with acute lymphoblastic leukemia (ALL) and chronic myeloid leukemia in lymphoid blastic phase (CML-BP) are infrequent after HLA matched DLI. This may be partly explained by the poor immunogenicity of these malignancies, since we previously demonstrated efficient induction of Graft-versus-Leukemia (GvL) immune responses in vitro and in vivo upon modification of ALL and CML-BP cells into leukemic antigen presenting cells (APC). Leukemic-APC may be particularly relevant for efficient generation of GvL immune responses after HLA matched DLI, since T cells recognizing allo-antigens in matched HLA molecules are known to reside in the naïve T cell compartment. In contrast, T cells recognizing allo-antigens in mismatched HLA molecules reside in the memory T cell compartment as well. Since memory T cells can also be activated by non-professional APC, HLA mismatched alloSCT and DLI may particularly be considered as a treatment modality for induction of GvL reactivity against poorly immunogenic malignancies. However, T cell responses across HLA barriers can induce severe Graft-versus-Host Disease (GvHD). Mismatched HLA class I molecules, which are broadly expressed on all nucleated cells, are frequent targets of alloreactive T cells. Since HLA class II molecules are predominantly expressed on hematopoietic cells, HLA class II mismatched alloSCT and DLI may more selectively induce GvL reactivity without inducing severe GvHD. In this study, we investigated the in vivo immunogenicity of established B-ALL or CML-BP by comparing the anti-tumor responses after fully HLA matched versus HLA class II mismatched DLI in a NOD/scid mouse model.Mice engrafted with primary B-ALL and CML-BP were treated with DLI from HLA matched (12/12 match) or HLA class II mismatched, but HLA class I matched donors. In mice engrafted with B-ALL or CML-BP, treatment with HLA matched DLI induced expansion of human CD4+ and CD8+ T cells in peripheral blood, but leukemic cells were only delayed in growth, and not eliminated. In contrast, after HLA class II mismatched DLI, leukemic cells rapidly disappeared upon emergence of human CD4+ and CD8+ T cells in peripheral blood. To analyze the specificity of the T cells, we clonally isolated CD4+ and CD8+ T cells from bone marrow and spleens of mice after treatment with DLI. All T cell clones were tested for recognition of patient leukemic cells, donor EBV transformed B cells (EBV-LCL) and murine bone marrow derived dendritic cells in IFNg ELISA. Isolated CD8+ and CD4+ T cell clones recognized either patient leukemic cells or murine cells, indicating that the T cell clones were either leukemia-reactive or xeno-reactive. After HLA matched DLI, only 2 of the 106 CD4+ T cell clones, and none of the 183 CD8+ T cell clones, recognized patient leukemic cells. The majority of isolated CD4+ and CD8+ T cell clones were xeno-reactive, as demonstrated by specific recognition of murine bone marrow derived dendritic cells, or non-reactive against any of the tested target cells. In contrast, after HLA class II mismatched DLI, 95 of the 322 CD4+ T cell clones specifically recognized patient leukemic cells. These leukemia-reactive CD4+ T cell clones were shown to be restricted by the mismatched HLA-DRB3, -DQB1 and –DPB1 alleles of the patient. None of the 49 CD8+ T cell clones were leukemia-reactive, but a significant number of CD8+ T cell clones and remaining CD4+ were xeno-reactive.In conclusion, our data show that HLA class II mismatched, but HLA class I matched, DLI is far more effective in inducing anti-tumor reactivity as compared to HLA matched DLI, whereas the in vivo capacity of both DLI's to induce allo-immune reactivity based on the induction of xeno-reactive T cells was similar. Our study emphasizes the necessity of HLA class II disparity for efficient in vivo induction of HLA class II mediated anti-tumor immunity against poorly immunogenic B-ALL and CML-BP in NOD/scid mice. We therefore hypothesize that use of HLA class II mismatched as compared to HLA matched alloSCT and DLI, despite an increased risk for GvHD, may improve the outcome for patients with HLA class II positive high risk acute lymphoblastic leukemia. Disclosures:No relevant conflicts of interest to declare.

Human allo-reactive CD4+ T cells as strong mediators of anti-tumor immunity in NOD/scid mice engrafted with human acute lymphoblastic leukemia

Adoptive immunotherapy with donor lymphocyte infusion (DLI) after allogeneic stem cell transplantation (alloSCT) may not only mediate Graft-versus-Leukemia (GvL) reactivity, but also induce Graft-versus-Host Disease (GvHD). As HLA-class II molecules are predominantly expressed on hematopoietic cells, CD4+ T cells may selectively mediate GvL reactivity without GvHD. Here, we assessed the capacity of human CD4+ T cells to act as sole mediators of GvL reactivity in a NOD/scid mouse model for human acute lymphoblastic leukemia and chronic myeloid leukemia in lymphoid blast crisis. Highly purified CD4+ DLI eradicated the leukemic cells. The anti-tumor immunity was mediated by a polyclonal CD4+ T cell response comprising cytokine-producing T-helper and cytolytic T-effector cells directed against the mismatched HLA-class II molecules of the patients. Furthermore, primary leukemic cells acquired an antigen-presenting cell (APC) phenotype in vivo after DLI, as well as in vitro after co-culture with leukemia-specific CD4+ T cells. In conclusion, our results show that CD4+ T cells can be strong mediators of anti-tumor immunity, and provide evidence that cross-talk between CD4+ T cells and leukemic cells is the basis for induction of leukemic cells with an APC phenotype. These data emphasize the clinical relevance of CD4+ T cell based immunotherapy as treatment modality for hematological malignancies after alloSCT.

Ex Vivo Generated Donor Central Memory CD8 T Cells, Previously Shown to Enhance Engraftment of Allogeneic Bone Marrow, Also Exhibit Significant GVL Activity without Causing Gvhd In An In Vivo b Cell Lymphoma Model

AbstractAbstract 424Generation of T cells endowed with graft versus leukemia (GVL) and depleted of graph versus host (GVH) activity represents a highly desirable goal in BM transplantation (BMT). Donor CTLs directed against 3rd party stimulators were shown to effectively enhance engraftment of BM allografts without causing GVHD. Recently, we demonstrated that activated anti 3rd party CD8 cells with central memory phenotype (Tcm), can further support and improve BM engraftment. This is likely due to the improved lymph node homing of the Tcms, their proliferative capacity and prolonged persistence in BMT recipients. Surprisingly, when tested in vitro, anti 3rd party human CTLs were found to eliminate B CLL and other lymphoma cell types. Investigating the mechanism of the cytotoxicity revealed a unique TCR independent effect mediated against both autologous and allogeneic targets. Inhibition studies using specific blocking Ab revealed initial binding of CTLs to their targets via a rapid ICAM1-LFA1 adhesion, followed by slow induction of apoptosis following interaction of the CD8 on the CTL with MHCI on the tumor cell. The killing did not involve classical CTL death molecules. To establish an appropriate mouse model, we initially verified that anti 3rd party Tcm derived from (B6xBALB/c)F1 exhibit TCR independent killing of A20 lymphoma cells of BALB/c origin (34.8±12.1% in 4 experiments, 5:1 Tcm/lymphoma cell ratio, in comparison to A20 cells incubated without Tcm, p<0.05). Moreover, after 16 hours of incubation with Tcm, AnnexinV staining of A20 cells was enhanced compared to basal staining (14.8±4.5% and 5.2±2.2% respectively, in three experiments, p<0.05), suggesting an apoptosis based mechanism as previously described for the killing of human lymphoma. Using luciferase expressing A20 cells, we were able to follow the fate of the malignant cells in vivo, and study the effect of added donor anti 3rd party Tcm to either syngeneic or allogeneic BMT in a model simulating minimal residual disease. In the syngeneic model, 3×106 Nude BALB/c BM cells were transplanted into lethally irradiated BALB/c mice together with 5000 A20 cells. On the next day, syngeneic Tcm were infused. As can be seen in Fig. 1, none of the untreated mice survived (0/7) 100 days post BMT (23 days median survival). Administration of 1×107 or 2×107 syngeneic Tcms led to significantly diminished tumor burden (Fig. 1A) and improved overall survival of 28% (2/7, P<0.0001) and 40% (2/5, P<0.002) with median survival of 49 and 80 days, respectively (Fig. 1B). As expected, further improvement of tumor eradication was found in the allogeneic setting, in which an additional GVL effect, mediated by residual alloreactivity in the absence of GVHD, occurred in addition to the TCR independent killing. In the allogeneic model, 3×106 allogeneic Nude B6 BM cells were transplanted together with 5000 A20 cells into lethally irradiated BALB/c mice, and mice were treated with donor type Tcm. Similar to the results in the syngeneic model, none of the untreated mice survived 100 days (0/8) (23 days median survival), while administration of 5×106 donor Tcms, led to remarkable overall survival of 100% (7/7) at 100 days post BMT (Fig. 2). Although the allogeneic Tcm displayed enhanced GVL activity compared to syngeneic Tcm, this effect was not associated with any manifestation of GVHD. Thus, as previously described, the weight and overall appearance of mice receiving allogeneic anti 3rd party Tcms were the same as that of mice in the control group, radioprotected with a transplant of Nude BM alone. Collectively, we demonstrate for the first time, by in vivo imaging the GVL reactivity of murine anti 3rd party Tcm. Considering our previous finding that anti 3rd party Tcm markedly enhance BM allografting, these results suggest that anti 3rd party Tcm can provide a ‘double supportive effect’ by promoting BM engraftment, and concurrently inducing GVL reactivity without causing GVHD. Such cell therapy is highly attractive, in particular for elderly patients with B CLL and other B cell malignancies who might not tolerate aggressive conditioning. [Display omitted] [Display omitted] Disclosures:No relevant conflicts of interest to declare.

Blockade of interleukin-23 signaling results in targeted protection of the colon and allows for separation of graft-versus-host and graft-versus-leukemia responses

AbstractAllogeneic stem cell transplantation is the most potent form of effective adoptive immunotherapy. The graft-versus-leukemia (GVL) effect mediated by the allogeneic graft, however, is typically coexpressed with graft-versus-host disease (GVHD), which is the major complication of allogeneic stem cell transplantation. In this study, we used genetic and antibody-based strategies to examine the effect that blockade of interleukin 23 (IL-23) signaling had on GVH and GVL reactivity in murine transplantation recipients. These studies demonstrate that the selective protection of the colon that occurs as a consequence of inhibition of IL-23 signaling reduces GVHD without loss of the GVL effect. The separation of GVH and GVL reactivity was noted in both acute and chronic hematologic malignancy models, indicating that this approach was not restricted by the kinetic profile of the underlying leukemia. Furthermore, a potent GVL response could be mounted in the colon under conditions where tumor cells migrated to this site, indicating that this organ did not serve as a sanctuary site for subsequent systemic relapse in GVHD-protected animals. These studies demonstrate that blockade of IL-23 signaling is an effective strategy for separating GVH and GVL responses and identify IL-23 as a therapeutic target for the regulation of alloresponses in humans.

Human T Cells with Distinct Specificities Mediate Graft-Versus-Leukemia Reactivity and Xenogeneic Graft-Versus-Host Disease in a NOD/Scid Mouse Model for Human Acute Leukemia.

Abstract 1330Poster Board I-352Immunodeficient mice inoculated with human leukemia can be used as a model to investigate Graft-versus-Leukemia (GvL) effects of donor lymphocyte infusions (DLIs). In addition to GvL reactivity, treatment with DLI induces xenogeneic Graft-versus-Host Disease (GvHD) in mice, characterized by pancytopenia and weight loss. In patients treated with DLI for relapsed or residual leukemia after allogeneic stem cell transplantation, immune responses against non-leukemic cells may also cause GvHD. It has been suggested that GvL reactivity and GvHD, which co-develop in vivo, can be separated and that distinct T cells exist with the specific capacity to mediate GvL reactivity or GvHD. Since adoptive T cell transfer models that allow analysis of separation of GvL and GvHD are rare, we aimed to establish whether GvL reactivity and xenogeneic GvHD could be separated using our model of human leukemia-engrafted NOD/scid mouse after treatment with human donor T cells. In this study, non-conditioned NOD/scid mice engrafted with primary human acute lymphoblastic leukemic cells were treated with CD3+ DLI. Established tumors were effectively eliminated by emerging human T cells, but also induced xenogeneic GvHD. Flowcytometric analysis demonstrated that the majority of emerging CD8+ and CD4+ T cells were activated (HLA-DR+) and expressed an effector memory phenotype (CD45RA-CD45RO+CCR7-). To investigate whether GvL reactivity and xenogeneic GvHD were mediated by the same T cells showing reactivity against both human leukemic and murine cells, or displaying distinct reactivity against human leukemic and murine cells, we clonally isolated and characterized the T cells during the GvL response and xenogeneic GvHD. T cell clones were analyzed for reactivity against primary human leukemic cells and primary NOD/scid hematopoietic (BM and spleen cells) and non-hematopoietic (skin fibroblasts) cells in IFN-g ELISA. Isolated CD8+ and CD4+ T cell clones were shown to recognize either human leukemic or murine cells, indicating that GvL response and xenogeneic GvHD were mediated by different human T cells. Flowcytometric analysis demonstrated that all BM and spleen cells expressed MHC class I, whereas only 1-3 % of the cells were MHC class II +. Primary skin fibroblasts displayed low MHC class I and completely lacked MHC class II expression. Xeno-reactive CD8+ T cell clones were shown to recognize all MHC class I + target cells and xeno-reactive CD4+ T cells clones displayed reactivity only against MHC class II + target cells. To determine the MHC restriction of xeno-reactive T cell clones, NOD/scid bone marrow (BM) derived dendritic cells (DC) expressing high levels of murine MHC class I and class II were tested for T cell recognition in the presence or absence of murine MHC class I and class II monoclonal antibodies in IFN-g ELISA. Xeno-reactive CD8+ T cell clones were shown to be MHC class I (H-2Kd or H-2Db) restricted, whereas xeno-reactive CD4+ T cell clones were MHC class II (I-Ag7) restricted, indicating that xeno-reactivity reflects genuine human T cell response directed against allo-antigens present on murine cells. Despite production of high levels of IFN-gamma, xeno-reactive CD8+ and CD4+ T cell clones failed to exert cytolytic activity against murine DC, as determined in a 51Cr-release cytotoxicity assay. Absence of cytolysis by CD8+ T cell clones, which are generally considered as potent effector cells, may be explained by low avidity interaction between human T cells and murine DC, since flowcytometric analysis revealed sub-optimal activation of T cells as measured by CD137 expression and T cell receptor downregulation upon co-culture with murine DC, and therefore these results indicate that xenogeneic GvHD in this model is likely to be mediated by cytokines. In conclusion, in leukemia-engrafted NOD/scid mice treated with CD3+ DLI, we show that GvL reactivity and xenogeneic GvHD are mediated by separate human T cells with distinct specificities. All xeno-reactive T cell clones showed genuine recognition of MHC class I or class II associated allo-antigens on murine cells similar as GvHD-inducing human T cells. These data suggest that our NOD/scid mouse model of human acute leukemia may be valuable for studying the effectiveness and specificity of selectively enriched or depleted T cells for adoptive immunotherapy. Disclosures:No relevant conflicts of interest to declare.

Diversity of HLA Class I and Class II Restricted Minor Histocompatibility Antigens in Graft-Versus-Leukemia Reactivity.

Abstract 4084Poster Board III-1019Donor lymphocyte infusion (DLI) can be an effective cellular immunotherapy for patients with hematological malignancies after HLA-matched allogeneic stem cell transplantation (alloSCT). The effect of DLI is mediated by donor derived T-cells recognizing minor histocompatibility antigens (mHags) encoded by single nucleotide polymorphisms (SNPs) on malignant cells of the recipient. Donor T-cells may also induce Graft-versus-Host Disease (GvHD) when directed against mHags with broad expression on non-malignant tissues. The aim of this study was to investigate the specificity and diversity of mHags recognized by T-cells in Graft-versus-Leukemia (GvL) reactivity.Activated (HLA-DR+) CD8+ and CD4+ T-cell clones were isolated from a patient successfully treated with DLI for relapsed chronic myeloid leukemia (CML) more than one year after HLA-matched alloSCT. GvL reactivity in this patient was accompanied with mild GvHD of the skin. Isolated T-cell clones were shown to recognize 13 different mHags. CD8+ T-cell clones were specific for HA-1 and HA-2 in HLA-A*0201, one unknown mHag in B*0801 and 4 unknown mHags in B*4001. CD4+ T-cell clones were specific for one unknown mHag in HLA-DQ and 5 unknown mHags in DR. By screening plasmid (class I) and bacteria (class II) cDNA libraries, we identified a mHag in HLA-DQ encoded by the PI4K2B gene (Griffioen et al., PNAS 2008), 4 mHags in HLA-DR encoded by the PTK2B, MR-1, LY75 and MTHFD1 genes (Stumpf et al., Blood 2009) and a mHag in B*4001 encoded by the TRIP10 gene. For the 3 T cell clones recognizing unknown mHags in B*4001, we performed Whole Genome Assocation scanning (WGAs). A panel of 60 EBV-LCL was retrovirally-transduced with B*4001 and tested for T-cell recognition. In parallel, genomic DNA was isolated and more than one million single nucleotide polymorphisms (SNPs) were determined by the Illumina beadchip array. Statistical analysis revealed significant association between T-cell recognition of EBV-LCL and the presence of coding SNPs in the SON DNA-binding protein and SWAP-70 genes. To get more insight into the role and potential use of the mHags in GvL reactivity and GvHD, all T-cell clones were analyzed in detail for reactivity against hematopoietic and non-hematopoietic cells. Hematopoietic cells included peripheral blood cells (monocytes, B-cells and T-cells), professional antigen presenting cells (APC) and leukemic cells (CML, ALL and AML). All CD8+ T-cell clones recognized (subsets of) peripheral blood cells as well as CML cells, except for the T-cell clone for TRIP10. Recognition of (subsets of) peripheral blood cells was also observed for all CD4+ T-cell clones, but CML cells were differentially recognized. CML cells were strongly recognized by the T-cell clones for MTHFD1 and the unknown mHag in HLA-DR, whereas no or low reactivity was observed for all other CD4+ T-cell clones. All CD8+ and CD4+ T-cell clones strongly recognized professional APC, including monocyte-derived dendritic cells and in vitro differentiated CML cells with APC phenotype. All T-cell clones were also capable of recognizing AML and ALL, except for the T-cell clone for TRIP10, which showed restricted recognition of AML-M4 and -M5 of monocytic origin. As non-hematopoietic cells, patient-derived fibroblasts were cultured with and without IFN-γ and tested for T-cell recognition. In the absence of IFN-γ, all T-cell clones failed to recognize fibroblasts, except for the T-cell clone for the unknown mHag in B*0801. After treatment with IFN-γ, additional reactivity was observed for the T-cell clones for SON DNA-binding protein and the unknown mHag in B*4001.Our data showed the specificity and diversity of mHags recognized by T-cells induced in a patient successfully treated with DLI for relapsed CML. The T-cell response was directed against 13 different mHags, of which 10 mHags in HLA class I and class II have now been identified by different techniques. Detailed analysis of T-cell recognition of hematopoietic and non-hematopoietic cells provides evidence that the mHags played different roles in the onset and execution of GvL and GvHD. Moreover, only one of the 10 identified mHags was expressed on fibroblasts after treatment with IFN-γ, indicating the characterization of mHags with potential relevance for T-cell based immunotherapy. Disclosures:No relevant conflicts of interest to declare.

Collateral Damage: a Mechanism for Hematopoiesis-Restricted Mhag-Specific T Cells to Play a Role in the Effector Phase of GvHD.

Abstract 3552Poster Board III-489Graft versus host disease (GvHD) after allogeneic stem cell transplantation (SCT) may occur when donor T cells respond to antigens specifically expressed on cells of the recipient, such as minor histocompatibility antigens (mHags). It has been hypothesized that ubiquitously expressed mHags, like HY and ADIR-1F, may be targets for both GvHD and graft versus leukemia (GvL), whereas hematopoiesis-restricted mHags, such as HA-1, may give rise to a more specific GvL effect without GvHD. However, it has been reported that mismatching for HA-1 is correlated with the development of GvHD. In addition, there is clinical evidence that the presence of a high tumor load leading to a profound GvL reaction may be associated with the induction of GvHD. To explain these observations, we analyzed whether T cells directed against hematopoiesis-restricted antigens, like HA-1, are able to contribute to destruction of the surrounding non-hematopoietic tissues in the effector phase of the immune response. First, we investigated whether HA-1 specific T cells could directly attack non-hematopoietic tissues, such as human primary skin fibroblasts from HLA-A2+, HA-1+ individuals using a 51Cr-release assay. In accordance with the absent gene expression of HA-1 in fibroblasts, no direct antigen-specific kill of the fibroblasts was seen. In contrast, exposure of ADIR-1F mHag positive fibroblasts to ADIR-specific CTLs resulted in 70% lysis after 20 hours. Even in a more inflammatory environment, induced by pretreatment with interferon gamma (IFNg), primary human fibroblasts from HLA-A2+, HA-1+ individuals were not directly killed by HA-1 specific CTLs. However, based on the clinical observation that the induction of a profound GvL effect in patients with a high tumor load correlated with the induction of GvHD, we hypothesized that these hematopoiesis-restricted T cells may have the capacity to induce collateral damage to surrounding non-hematopoietic tissues during an ongoing profound immune response against the hematopoietic target cells. To test this hypothesis, we developed an in vitro system in which monolayers of 51Cr-labeled primary human fibroblasts from HLA-A2+, HA-1+ individuals were co-cultured with HA-1-specific CTLs in the presence or absence of HA-1-expressing hematopoietic targets (EBV-LCL) for 4, 8 and 20hrs. Activation of HA-1-specific T cells by the EBV-LCL was shown to be dose dependent and was found to be optimal at stimulator/responder (S/R) ratios higher than 1/1 as measured by their specific IFNg production. Despite optimal activation of the HA-1-specific CTLs by the EBV-LCL (S/R ratio 10/1), initially no collateral damage to surrounding fibroblast was observed. However, under pro-inflammatory conditions, mimicked by the treatment of the fibroblasts with IFNg, HA-1-specific CTLs optimally activated by the HA-1+ EBV-LCL which were seeded between the fibroblasts (S/R ratio 10/1), exerted 40% (25-60%) bystander cytotoxicity to the surrounding fibroblasts at a 10/1 T cell/fibroblast-ratio after 20 hrs. Interestingly, there was a strict correlation between the amount of collateral damage to the surrounding fibroblasts and the strength of the T cell activation by the hematopoietic cells expressing the relevant target antigen, being optimal at S/R ratios higher than 1/1, possibly reflecting the clinical situation of a GVL response in the presence of a high tumor load. Similarly, the magnitude of the immune response, reflected by the number of T cells, was associated with the amount of collateral damage, since no collateral damage could be observed at CTL/fibroblast ratios below 3/1. In conclusion, these data indicate that a profound immune response against hematopoietic targets, mediated by T cells directed against hematopoiesis-restricted antigens, may result in the induction of GvHD via collateral damage to the surrounding non-hematopoietic tissues without the necessity for these tissues to express the specific target antigen. Disclosures:No relevant conflicts of interest to declare.

Human Alloreactive CD4+ T Cells as Potent Effector Cells and Sole Mediators of Anti-Tumor Responses in a NOD/SCID Mouse Model for Human Acute Leukemia.

Donor lymphocyte infusion (DLI) following allogeneic stem cell transplantation (alloSCT) can be a curative treatment for patients with hematological malignancies. The therapeutic benefit of DLI is attributed to a graft versus leukemia (GvL) reactivity mediated by donor T cells recognizing allo-antigens on malignant cells of the patient. Donor T cells, however, often recognize allo-antigens which are broadly expressed in non-malignant tissues of the patient, thereby causing severe graft versus host disease (GvHD). In contrast to HLA class I molecules which are ubiquitously expressed on all nucleated cells, HLA class II molecules are predominantly expressed on cells of the hematopoietic system, and therefore CD4+ T cells may selectively mediate GvL reactivity without GvHD. Several clinical studies have indeed demonstrated that CD8-depleted DLI after alloSCT can lead to clinical remissions with reduced incidence of GvHD. Since in most of these studies DLI was contaminated with CD8+ T cells, it remained unclear whether CD4+ T cells alone are capable of mediating GvL reactivity. To assess the capacity of purified CD4+ T cells to solely exert GvL reactivity we compared the anti-tumor effects of CD4+ DLI and CD3+ DLI in a NOD/SCID mouse model of human acute leukemia. Iv injection of primary human leukemic cells from three different patients reproducibly resulted in engraftment of leukemia in mice, as monitored by peripheral blood analysis. Three weeks after inoculation of leukemic cells, established tumors were treated by infusion of human donor T cells. In mice treated with CD4+ DLI (5*106 CD4+ T cells), the emergence of activated (HLA-DR+) T cells coincided with rapid disappearance of leukemic cells, showing similar kinetics as for CD3+ DLI (consisting of 5*106 CD4+ T cells and 3*106 CD8+ T cells). To analyze the specific reactivity of T cells responsible for the anti-leukemic effect, we clonally isolated human CD45+ T cells during the anti-tumor response following CD4+ DLI in which the donor was matched for HLA class I and mismatched for the HLA-DR (DRB1*1301), -DQ (DQB1*0603) and –DP (DPB1*0301/0401) alleles of the patient. A total number of 134 CD4+ T cell clones were isolated expressing various different TCR Vbeta chains. Most of the isolated CD4+ T cell clones (84%) were shown to be alloreactive, as determined by differential recognition of patient and donor EBV-transformed B cells (EBV-LCL) in IFN-g ELISA. A substantial number of these CD4+ T cell clones also exerted cytolytic activity (17%), as demonstrated by specific reactivity with patient EBV-LCL but not donor EBV-LCL in a 10 hr 51Cr-release cytotoxicity assay. Further characterization of the specificity of 20 CD4+ T cell clones using blocking studies with HLA class II specific monoclonal antibodies illustrated HLA class II restricted recognition directed against HLA-DR (n=3), HLA-DQ (n=16) and HLA-DP (n=1) molecules of the patient. Of the 127 alloreactive CD4+ T cell clones, only 36 clones directly recognized primary leukemic cells of the patient. Flowcytometric analysis demonstrated that HLA class II, and in particular HLA-DQ, molecules were expressed at relatively low levels on patient leukemic cells as compared to patient EBV-LCL. Upregulation of HLA class II and costimulatory molecules on patient leukemic cells upon differentiation in vitro into leukemic antigen presenting cells (APC) resulted in recognition of patient leukemic cells by all alloreactive CD4+ T cell clones. Therefore, we hypothesize that the alloreactive CD4+ T cells have been induced in vivo by patient leukemic cells, which, upon interaction with T cells or other environmental factors, acquired an APC phenotype. In conclusion, our data show that alloreactive CD4+ T cells can be potent effector cells and sole mediators of strong antitumor responses in a NOD/SCID mouse model for human acute leukemia.

Detailed Analysis of CD8+ T Cell Immunity and Identification of a Novel Minor Histocompatibility Antigen Contributing to Graft-Versus- Leukemia Reactivity.

Donor lymphocyte infusion (DLI) can be an effective cellular immunotherapy for patients with hematological malignancies after HLA-matched allogeneic stem cell transplantation (alloSCT). The effect of DLI is mediated by donor derived T cells recognizing minor histocompatibility antigens (mHags) on malignant cells of the recipient. These donor originated T cells may also induce Graft-versus-Host Disease (GvHD) when directed against mHags with broad expression on non-malignant tissues of the patient. In this study, we performed a detailed analysis and characterization of mHags recognized by CD8+ T cells contributing to Graft-versus-Leukemia (GvL) reactivity in a patient treated with DLI for relapsed chronic myeloid leukemia (CML) more than one year after HLA-matched alloSCT. The GvL effect in this patient was accompanied with only mild GvHD of the skin.To investigate the specificity of the CD8+ T cell response induced in this patient, activated (HLA-DR+) CD8+ T cells were single cell sorted from a bone marrow sample obtained five weeks after DLI by flowcytometry. A number of isolated CD8+ T cell clones were shown to be specific for mHags, as determined by differential recognition of patient and donor EBV-transformed B cells (EBV-LCL) in IFN-g ELISA and 51Crrelease assays. By screening a panel of third party EBV-LCL sharing one or more HLA class I restriction molecules with the patient, CD8+ T cell clones directed against 7 different mHags were identified, including the known hematopoiesis restricted mHags HA-1 and HA-2, which are presented in HLA-A*0201. Of the 5 remaining specificities, 4 mHags were presented in HLA-B*4001 (B60) and one mHag in HLA-B*0801. To determine the tissue distribution patterns in more detail, we tested recognition of selected non-malignant hematopoietic cells (monocytes, B cells, T cells), malignant CD34+ CML precursor cells, and skin-derived fibroblasts. One HLA-B*4001-restricted T cell clone (clone ZRZ16) failed to recognize all primary hematopoietic cells and skin fibroblasts. The four remaining T cell clones were all capable of recognizing and lysing (specific subsets of) non-malignant hematopoietic cells and malignant CD34+ CML precursor cells. Fibroblast recognition could be demonstrated for two of these four T cell clones. Since clone ZRZ16 failed to recognize all hematopoietic and non-hematopoietic cells, except for EBV-LCL, we tested whether this clone selectively recognizes antigen presenting cells (APC), which are known to be required for efficient induction of immune responses in vivo. Clone ZRZ16 showed indeed strong recognition of monocyte-derived dendritic cells as well as in vitro differentiated CD34+ CML cells with APC phenotype. To identify the mHag recognized by the CD8+ T cell clone, we screened a cDNA expression library constructed from EBV-LCL from the patient. One single cDNA was isolated as the target for B*4001 restricted CD8+ T cell clone ZRZ16. The epitope recognized by this clone was derived from the 3' untranslated region (UTR) of a cDNA encoding thyroid hormone receptor interactor 10 (TRIP10). The peptide epitope was translated in a reading frame different from the TRIP10 protein and comprises three single nucleotide polymorphisms, which were all different between patient and donor. Two of the three SNPs were shown to be important for recognition by clone ZRZ16. Despite ubiquitous tissue expression of the TRIP10 gene as determined by public microarray analysis, CD8+ T cells specific for the newly-identified LB-TRIP10-1EPC mHag selectively recognized APC and failed to recognize CD34+ CML precursor cells, suggesting a predominant role in the initiation, but not effector, phase of the anti-tumor response. In conclusion, our data show a detailed analysis of mHag specific CD8+ T cell immunity induced in a patient successfully treated with DLI for relapsed CML and provide evidence for differential involvement of HLA class I restricted mHags in the onset and execution of GvL reactivity.