Autologous Acute Myeloid Leukemia Research Articles

Identification and Characterization of T Cell Epitopes Deduced from RGS5, a Novel Broadly Expressed Tumor Antigen.

Identification of tumor-associated antigens (TAA) has resulted in the development of therapeutic vaccines for the treatment of cancer. We applied an integrated functional genomics approach to identify TAA in malignant tissues of patients with renal cell carcinoma (RCC). A comparative DNA chip analysis of tumor and the corresponding non-malignant tissue from patients with RCC followed by sequencing of peptides bound to the HLA-class I molecules by mass spectrometry was applied to identify novel TAA in RCC. To confirm the immunogenicity of identified epitopes cytotoxic T lymphocytes (CTL) were generated using dendritic cells (DC). Utilizing this approach, two peptides derived from RGS5 binding to either HLA-A*02 or 03 were identified. The key function of regulators of G protein-signalling (RGS) is to bind to G protein α subunits and to stimulate their intrinsic GTPase activity. The hydrolysis of guanosine triphosphate (GTP) to guanosine diphosphate (GDP) thereby is accelerated and the inactive heterotrimer more rapidly restored. Thus, RGS proteins inhibit the biological activity of G proteins. Interestingly, it was recently shown that RGS5 is overexpressed in pericytes of newly developing tumor vessels, indicating that RGS5 plays an important role during tumor angiogenesis.Using RT-PCR analysis we found that RGS5 is expressed on a broad variety of tumor cells including RCC, colorectal, breast and ovarian cancer, malignant melanoma and multiple myeloma as well as in acute and chronic leukemias making this protein an interesting candidate for the development of vaccination strategies to target the tumor cells and the tumor vessels. CTL that were induced using the RGS5 peptides lysed autologous DC pulsed with the cognate peptide or transfected with in vitro transcribed RGS5 RNA as well as HLA-matched tumor cell lines. The specificity and HLA restriction was confirmed using blocking monoclonal antibodies and in cold-target inhibition assays. We next utilized DC transfected with RGS5 RNA to generate specific CTL. Using this approach we confirmed the processing and presentation of the identified peptides by malignant cells. These CTL lysed tumor cells in an antigen specific manner while sparing non-malignant cells. To analyze the induction of RGS5 specific CTL in an autologous setting in patients with malignant diseases, we used blood samples from a patient with acute myeloid leukemia (AML) in complete remission after chemotherapy and were able to generate specific CTL capable of recognizing autologous AML blasts while sparing non-malignant cells.Our results demonstrate that RGS5 is a novel tumor rejection antigen expressed in a wide range of malignancies that can be applied to target malignant cells and tumor angiogenesis.

Randomized Trial of GM-CSF and G-CSF Following High-Dose Cytarabine and Mitoxantrone Chemotherapy for Relapsed and Refractory Acute Leukemia.

Patients with acute myeloid leukemia (AML) or acute lymphoblastic leukemia (ALL) refractory to or in relapse following induction chemotherapy have a poor prognosis, and inducing an immune response to autologous AML or ALL cells following chemotherapy is an attractive approach to improving outcome. Immune responses to autologous leukemia cells may be stimulated by dendritic cell presentation of leukemia cell antigens, dendritic cells may be deficient in acute leukemia, and administration of the recombinant hematopoietic growth factors granulocyte-monocyte colony-stimulating factor (GM-CSF) or granulocyte colony-stimulating factor (G-CSF) following chemotherapy may increase dendritic cell numbers. We compared the effects of GM-CSF and G-CSF administered following high-dose chemotherapy. Adult relapsed and refractory AML and ALL patients received salvage chemotherapy consisting of high-dose cytarabine 3 g/m2 (1.5 g/m2 for age ≥50 years) over one hour every 12 hours × 12 doses and mitoxantrone 12 mg/m2 daily × 3 (HiDAC/Mx), and at completion of chemotherapy were randomized to receive GM-CSF 250 mcg/m2 or G-CSF 5 mcg/kg daily beginning 12 hours after the last chemotherapy dose, until absolute neutrophil count ≥5 × 109〈1 year) and late (≥1 year) first and subsequent relapse. Peripheral blood was collected when ANC reached 5 × 109/L for measurement of myeloid dendritic cell (lineage-negative, HLADr+, CD11c+) percentages by flow cytometry. Sixty patients were enrolled, ages 18 to 82 (median 66) years, 41 male and 19 female, 47 with AML and 13 with ALL, and 15 with primary refractory disease, 27 in early and 17 in late first relapse and one in subsequent relapse; 6 had relapsed following allogeneic transplantation. Overall, 22 of 60 patients (37%) achieved CR and 4 (7%) CR with incomplete count recovery (CRi), while 23 (38%) had resistant disease and 11 (18%) died. The regimen was generally well tolerated, the most frequent grade ≥4 toxicities pulmonary, infectious and cardiac, in 8, 7 and 6 patients, respectively, and 13 patients subsequently received transplant-based therapies (9 allogeneic, 4 autologous). 56 patients were randomized, as 4 died or stopped therapy before randomization, and randomization was to GM-CSF in 29 patients and G-CSF in 27. CR and CRi were achieved by 13 and 1 patients of 29 patients receiving GM-CSF and 9 and 3 of 27 receiving G-CSF (p NS, Fisher's Exact Text). ANC ≥0.5 was achieved at 22 to 98 days (median 27) from start of chemotherapy in 25 GM-CSF patients and at 18 to 65 days (median 25) in 20 G-CSF patients (p=0.08, Wilcoxon Rank Sum Test). Toxicities did not differ significantly on the two arms. Only 17 patients (G-CSF: 7 and GM-CSF: 10) had blood samples submitted and successfully studied for myeloid dendritic cell percentages. Myeloid dendritic cell percentages were 0 to 40 (median 22), and the comparison by treatment group showed no evidence of a difference. In summary, HiDAC/Mx is an effective salvage regimen in this high-risk population and may serve as a bridge to transplant-based therapies or, possibly, a backbone for targeted therapies, myeloid dendritic cells are present at count recovery in patients receiving GM-CSF or G-CSF following HiDAC/Mx, and treatment outcome, toxicities, count recovery and myeloid dendritic cell percentages did not differ in patients receiving GM-CSF or G-CSF following HiDAC/Mx.

RHAMM/CD168-R3 Peptide Vaccination of Patients with Acute Myeloid Leukemia (AML), Myelodysplastic Syndrome (MDS) and Multiple Myeloma (MM) Elicits Immunological and Clinical Responses.

Most patients with AML, MDS, MM and chronic lymphatic leukemia (CLL) express the receptor for hyaluronic acid mediated motility (RHAMM/CD168) on their tumor cells. We characterized RHAMM/CD168 as a leukemia-associated antigen (LAA) eliciting both humoral and cellular immune responses in patients with different hematological malignancies. CD8 positive T cells primed with the RHAMM/CD168-derived peptide R3 (ILSLELMKL) were able to lyse autologous AML blasts expressing this LAA. Therefore, we initiated a phase I/II R3 peptide vaccination to induce immunological and hematological responses for patients with AML, MDS or MM overexpressing RHAMM/CD168. Patients were included with positive RHAMM/CD168 expression but with a limited tumor load. At a biweekly interval, RHAMM R3 peptide (300 mcg for the first 12 patients and 1000 mcg for patients 13–24) emulsified with the incomplete Freund's adjuvant (day 3) and GM-CSF (100 mcg, days 1–5) was administrated four times subcutaneously. The primary aim of the study is safety and feasibility of this peptide vaccination, secondary aims the evaluation of a specific T cell immune response to RHAMM/CD168 R3 peptide and the assessment of the influence of the R3 peptide vaccination on the remission status. Since January 2005, 14 patients have been enrolled in the study. The first ten patients (2 AML, 4 MDS, 4 MM) have completed the course of four vaccinations and have been completely evaluated. Therapy related adverse events observed under R3-peptide vaccination were erythema and induration of the skin at the site of injection (CTC I°). In 5/10 patients, we detected in the peripheral blood a significant increase of specific CD8+ T cells recognizing the R3 peptide in ELISPOT analysis and seven-color flow cytometry including tetramer staining. As expected, after vaccination with the HLA class I peptide R3 no significant increase of IgG titers against the antigen RHAMM could be detected. Clinical responses have been assessed by the examination of peripheral blood and bone-marrow samples before and after vaccination. Patients showed a reduction of the tumor-specific expressed antigen RHAMM/CD168 in real-time RT-PCR analysis after vaccination. 3/6 patients with myeloid disorders (1 AML, 2 MDS/RAEB1) showed a reduction of CD33+ cells in FACS analysis of the bone-marrow after four vaccinations from 10 and 7 % to 1–2 and <1%, respectively. One patient with MDS did not need further erythrocyte substitution. Two patients with MM showed a reduction of plasma cells in the bone-marrow as by FACS analysis and of free light chains. One patient with AML and one patient with MM showed a progressive disease. Taken together, RHAMM/CD168 induced both immunological and clinical results and therefore constitutes a promising target antigen for immunotherapies in patients with hematological malignancies. Further entities with RHAMM/CD168 expression such as CLL might also be eligible for immunotherapeutic approaches using RHAMM/CD168.

Generation of Acute Myeloid Leukemia (AML)-Specific Autologous CD4 and CD8 T Cell Lines by Limiting Dilution AML Dendritic Cell Culture.

Naturally occurring cytotoxic T cells directed against various leukemia associated antigens (LAA) expressed by acute myeloid leukemia (AML) cells have been described. However, these LAA-specific T cells are rare and obviously unable to initiate effective anti-leukemia responses. The challenge is how to investigate, select, activate and expand the rare LAA-specific T cells from the vast population of blood cells in patients with AML for immunotherapy. Based on our studies of inducing AML dendritic cell (AMLDC) differentiation and priming in situ AML-reactive T cells, we have developed a novel method of generating multiple autologous AML reactive T cell lines by limiting dilution AMLDC (LD-AMLDC) culture. The principle of LD-AMLDC is based on the assumption that autologous AML-reactive T cells or precursors are randomly distributed in the AML PBMC suspension, and that each one has an equal opportunity to respond to AML cells in the 96-well plates under optimized culture condition. By culturing AML PBMC (&gt;90% blasts) in culture medium supplemented with GM-CSF/IL4/IL2/IL7/IL12 to induce AML DC differentiation and activate in situ autologous T cells, highly reactive anti-AML T cell lines (both CD4+ and CD8+ lines) were selected and expanded from LD-AMLDC culture using the appropriate numbers of AML PBMC in each culture well by the criterion of release of IFN-gamma in response to autologous AML blasts. By maximum likelihood solution, the estimated average frequency of AML reactive T cells or precursors is 6±3/1,000,000 AML PBMC (n=8). Strong intracellular IFN-gamma release of T cell lines obtained in LD-AMLDC was demonstrated by flow cytometry analysis after stimulation by autologous AML cells but not autologous B-lymphoblastoid cell line (LCL) (Figure). Effective specific lysis (up to 70% at E:T=20:1) of autologous AML cells but not autologous LCL or allogeneic AML cells by these T cell lines was observed. Two PR1 specific T cell lines were obtained by screening 39 AML reactive HLA-A2+ CD8+ T cell lines generated from 5 LD-AMLDC cultures, suggesting that other unidentified CD4 or CD8 lines with strong autologous AML responses may be reactive to known or unknown LAAs. These results encourage continued efforts to induce, activate and select T cells lines with high autologous AML reactivity using LD-AMLDC culture and to expand multi-LAA reactive T cell lines acquired from limiting dilution AML-DC culture for AML immunotherapy. Figure Figure

IL-2/B7.1 (CD80) Fusagene Transduction of AML Blasts by a Self-Inactivating Lentiviral Vector Stimulates T Cell Responses in Vitro: a Strategy to Generate Whole Cell Vaccines for AML

Combined expression of costimulatory factors and proinflammatory cytokines stimulate effective immune-mediated tumor rejection in a variety of murine tumor models. Specifically, syngeneic tumor cells genetically modified to express B7.1 (CD80) have been shown to induce rejection of previously established murine solid tumors, and transduction with IL-2 can further increase survival. However, poor rates of gene transfer and inefficient expression of multiple transgenes encoded by single vectors have hampered the development of such autologous tumor cell vaccines for clinical trials in acute myeloid leukemia (AML) patients. Here we describe the development of a self-inactivating lentiviral vector encoding B7.1 and IL-2 as a single fusion protein postsynthetically cleaved to generate biologically active membrane-anchored B7.1 and secreted IL-2. This enables the efficient transduction of both established and primary AML blasts, resulting in expression of the transgenes in up to 98% of the cells following a single round of infection at an m.o.i. of 10. The combined expression of IL-2 and B7.1 in AML blasts enables increased stimulation of both allogeneic and autologous T cells. The stimulated lymphocytes secrete greater levels of Th1 cytokines and show evidence of specificity, as indicated by their increased proliferation in the presence of autologous AML compared to remission bone marrow cells.

Naturally Processed Peptides Eluted from Acute Myeloid Leukemia Cells : A Potent Antigen Source for Immunotherapy.

Acute myeloid leukemia (AML) is a heterogenous group of diseases characterized by a clonal proliferation of myeloid progenitors. Its poor prognosis with conventional chemotherapy justifies seeking for adjuvant immunotherapeutic approaches to eliminate minimal residual disease. We evaluated an immunotherapeutic strategy that bypass the need for epitope identification and the limitation due to HLA restriction. Naturally processed peptides were extracted by acid elution from AML cells at diagnosis, and loaded on mature dendritic cells (mDCs) derived from autologous monocytes obtained when the patients were in complete remission (CR). We evaluated i) the feasibility to elute naturally processed peptides from AML cells at diagnosis, ii) the capacity of mDCs loaded with eluted peptides (mDC/EP) to stimulate specific T cell lines in vitro. We showed that stimulation by mDC/EP was able to generate anti-leukemic T cells lines from PBMC of 6 AML patients in CR. CD4+ and CD8+ T cells were isolated from T cell lines of 5 patients and analyzed for their proliferation, INF-γ production and cytotoxicity in response to autologous or allogeneic AML targets, or to normal autologous PBMC. We showed that both CD4+ and CD8+ leukemia-specific T cells were generated in vitro by mDC/EP stimulations since proliferation of CD4+ T cells, IFN-γ secretion by CD4+ and CD8+ T cells and cytotoxicity mediated by CD8+ T cells were induced in response to stimulation with autologous AML cells. Furthermore, we could not detect auto-immune recognition of autologous normal PBMC, consistent with the specificity of the T cell response induced by mDC/EP. These results provide the proof of concept for using mDC/EP to vaccinate patients with poor-risk AML, and will soon be evaluated in a phse I/II clinical trial.

Cytokine-induced Killer Cells: NK-like T Cells with Cytotolytic Specificity against Leukemia

Cytokine-induced killer (CIK) cells are a unique population of cytotoxic T lymphocytes (CTL) with the characteristic CD3+CD56+ phenotype. These cells have demonstrated higher proliferative and cytolytic activities in comparison to the reported CD3−CD56+ lymphokine activated killer (LAK) cells that are essentially activated natural killer (NK) cells. CIK cells are non-MHC-restricted in target cell recognition and killing. We have shown the feasibility of generating CIK cells from a series of marrow samples of patients with acute myeloid leukemia (AML) collected at diagnosis. At maturity, the CIK cells exhibit potent cytotoxicity against autologous AML targets as well as allogeneic myeloid leukemia cells, regardless of the HLA types of these targets. This observed cytotoxicity is not entirely due to NK cells as prior pre-absorption of the NK cells cytolytic activities does not abolish the subsequent cytotolytic activities against leukemic targets. It has also been reported by others that CIK cells are cytolytic against chronic myeloid leukemia (CML) cells, both in vitro and in the SCID mouse tumor model. In a mouse transplant model across MHC barrier, the CIK cells generated from the donor do not induce graft vs. host disease as observed for unfractionated donor splenocytes. In comparison to untreated control mice, the infusion of CIK cells results in the prolonged survival of murine leukemia-bearing mice. CIK cells also express CD94, part of the NK receptor comprising of CD94-NKG2 heterodimer. However, only low level of the killer immunoglobulin-like receptors are expressed by the CIK cells. In addition, as reported for the classical CTL, CIK cells could interact with dendritic cells (DC) to result in the enhancement of cytotolytic activities against tumor cells. The characteristic biological properties of the CIK cells would, therefore, enable them to be exploited for anti-leukemic therapy.

Sequential modulation of growth factors: a novel strategy for adoptive immunotherapy of acute myeloid leukemia.

Evidence from allogeneic hematopoietic stem cell transplantation indicates a possible immune response against leukemia-associated antigens in patients with either acute myeloid leukemia (AML) or chronic myeloid leukemia (CML). However, autologous immune responses are less evident. We have developed a method using sequential modulation of growth factors (SMGF) to generate specific anti-AML T-cells from primary cultures of mononuclear cells (MNCs) from patients with AML. This culture method induces greater degrees of antigen presentation by inducing dendritic cell (DC) differentiation of AML in the presence of autologous lymphocytes, which are then expanded by interleukin (IL)-2 and costimulatory molecule ligation. MNCs consisting of 92.3% +/- 5.1% AML blasts and 3.4% +/- 3.2% CD3+ T-cells were obtained from AML patients (n = 12) and cultured in AIM-V medium with IL-4 and recombinant granulocyte-monocyte colony-stimulating factor. Recombinant IL-2 was added on day 8. On day 21, culture conditions were changed to anti-CD3/anti-CD28 monoclonal antibodies and IL-2. By day 42, 354 +/- 182-fold CD3+ T-cell expansion had occurred. Cytotoxic T-lymphocyte assays demonstrated that these T-cells caused significant lysis of autologous leukemia cells and AML cell lines, but not of cells of other lineages, in an HLA class I-dependent manner. Specific Vbeta subgroups (Vbeta3, -7, and -12a), possibly representing T-cell clones specific to AML-specific antigens, were expanded in the cultures of cells from 3 AML patients. SMGF can be used to induce and expand autologous T-cells with HLA class I-dependent antileukemia potential from the peripheral blood of AML patients. Adoptive transfer of these expanded T-cells to patients is a possible therapeutic approach for further study.

Ex vivo expansion of normal progenitor cells from acute myeloid leukemia cell-contaminated CD34+ peripheral blood progenitor cells after mafosfamide purging.

The rationale for purging of autologous acute myeloid leukemia (AML) grafts is to eradicate contaminating leukemic cells that might contribute to relapse. However, in vitro purging generally delays post-transplant hematopoietic recovery, thus increasing treatment-related complication rates. Theoretically, this prolonged aplasia might be shortened by the additional transplantation of ex vivo-generated progenitor cells. Therefore, we investigated whether nonleukemic progenitors could be expanded ex vivo from AML cell-contaminated CD34(+) peripheral blood progenitor cell (PBPC) preparations. Nonleukemic CD34(+)-selected PBPC and AML cells (Kasumi-1, KG-1, primary AML blasts) were cultured in cytokine-supplemented liquid culture for up to 19 days. Cells were used either unmanipulated or following in vitro purging with mafosfamide (30, 50, 75 microg/ml). Ex vivo-generated cells were assessed by flow cytometry, progenitor cell assays, and polymerase chain reaction. Without prior purging, ex vivo culture markedly amplified AML cells as well as nonleukemic CD34(+) PBPC (day 12: Kasumi-1, 18.5 +/- 0.6-fold; KG-1, 52.2 +/- 2.6-fold; CD34(+), 74.1 +/- 5.6-fold). Co-culture with leukemic cells did not affect CD34(+) cell growth and vice versa. Following in vitro purging, CD34(+) PBPC were expanded even at the highest mafosfamide dose (day 19: 25 +/- 15-fold), whereas leukemic cells were markedly depleted (approx. 1.5 log). Furthermore, normal colony-forming units (CFU) could be effectively recovered (day 19: 10 +/- 3.1% of prepurging input CFU), whereas CFU-L were depleted to undetectable levels in six of seven experiments. Finally, leukemic cells were undetectable following ex vivo co-culture of purged cells (CD34(+) PBPC plus 10% Kasumi-1 cells or primary blasts), but were clearly detectable without purging. Taken together, these data demonstrated that ex vivo expansion of normal progenitors from mafosfamide-purged AML cell-contaminated grafts might be feasible.

Targeting of a B7-1 (CD80) immunoglobulin G fusion protein to acute myeloid leukemia blasts increases their costimulatory activity for autologous remission T cells

AbstractTransfection of tumor cells with the gene encoding the costimulatory molecule B7-1 (CD80), the ligand for CD28 and cytotoxic T lymphocye antigen-4 on T cells, has been shown to result in potent T-cell–mediated antitumor immunity. As an alternative approach, this study analyzed the costimulatory capacity of a human B7-1 immunoglobulin G (IgG) fusion protein targeted to the cell membrane of human acute myeloid leukemia (AML) blasts. Flow cytometric analysis revealed a low constitutive expression of B7-1 on human AML blasts (on average, 3.0 ± 4.3%; n = 50). In contrast, the expression of B7-2 (CD86) was highly heterogeneous and higher in AML blasts of French-American-British classification types M4 and M5 (P < .0001). The B7-1 IgG fusion protein used in this study efficiently costimulated the proliferation of resting and preactivated T cells when immobilized on plastic. After preincubation with B7-1 IgG, specific binding of the fusion protein to the high-affinity Fcγreceptor I (CD64) on leukemic cells was demonstrated and was found to increase the proliferation of both allogeneic and autologous T cells in costimulation experiments. Furthermore, targeting of B7-1 IgG to the tumor membrane resulted in increased proliferation of autologous remission T cells and had the potential to generate an enhanced redirected cytotoxic T-cell response against autologous AML blasts. In summary, the targeting of B7-1 IgG fusion protein described in this study represents a strategy alternative to gene therapy to restore the expression of the costimulatory molecule B7-1 on human AML blasts, thereby enhancing their immunogenicity for autologous T cells. This new approach may have implications for T-cell–mediated immunotherapy in AML.

A Bispecific Antibody Enhances Cytokine-Induced Killer-Mediated Cytolysis of Autologous Acute Myeloid Leukemia Cells

An anti-CD3 Fab' x anti-CD13 Fab' bispecific antibody (BsAb) was generated. This BsAb reacted with both CD3+ T cells and CD13+ acute myeloid leukemia (AML) cells. We investigated whether cytokine-stimulated peripheral blood mononuclear cells (PBMC) could lyse patient AML cells after addition of the BsAb. When interleukin-2 (IL-2)-stimulated PBMC were assayed for their cytotoxicity against 51Cr-labeled allogeneic and autologous CD13+ AML cells, their activity was markedly enhanced by the addition of the BsAb. PBMC stimulated with IL-2 plus anti-CD3 monoclonal antibody (MoAb) showed higher proliferative ability and higher cytotoxicity if this was expressed as lytic units per culture. IL-7-stimulated PBMC also exhibited enhanced cytotoxicity against CD13+ AML cells after addition of the BsAb. Ultrastructurally, CD13+ AML cells incubated with IL-2 plus anti-CD3 MoAb-stimulated PBMC and the BsAb showed apoptotic morphologic changes. A colony assay for AML blast progenitors showed that the colony formation of CD13+ AML cells was inhibited by the addition of autologous IL-2 plus anti-CD3 MoAb-stimulated PBMC, and that this inhibition was further enhanced by the addition of the BsAb. A colony assay for normal bone marrow progenitor cells showed that the addition of autologous IL-2 plus anti-CD3 MoAb-stimulated PBMC and the BsAb inhibited the formation of granulocyte-macrophage colonies and mixed-cell colonies. However, the degree of inhibition was smaller than that for the AML blast colonies. Taken together, these findings suggest that this BsAb may be useful for ex vivo purging of CD13+ AML cells in autologous bone marrow transplantation.

A bispecific antibody enhances cytokine-induced killer-mediated cytolysis of autologous acute myeloid leukemia cells

An anti-CD3 Fab' x anti-CD13 Fab' bispecific antibody (BsAb) was generated. This BsAb reacted with both CD3+ T cells and CD13+ acute myeloid leukemia (AML) cells. We investigated whether cytokine- stimulated peripheral blood mononuclear cells (PBMC) could lyse patient AML cells after addition of the BsAb. When interleukin-2 (IL-2)- stimulated PBMC were assayed for their cytotoxicity against 51Cr- labeled allogeneic and autologous CD13+ AML cells, their activity was markedly enhanced by the addition of the BsAb. PBMC stimulated with IL- 2 plus anti-CD3 monoclonal antibody (MoAb) showed higher proliferative ability and higher cytotoxicity if this was expressed as lytic units per culture. IL-7-stimulated PBMC also exhibited enhanced cytotoxicity against CD13+ AML cells after addition of the BsAb. Ultrastructurally, CD13+ AML cells incubated with IL-2 plus anti-CD3 MoAb-stimulated PBMC and the BsAb showed apoptotic morphologic changes. A colony assay for AML blast progenitors showed that the colony formation of CD13+ AML cells was inhibited by the addition of autologous IL-2 plus anti-CD3 MoAb-stimulated PBMC, and that this inhibition was further enhanced by the addition of the BsAb. A colony assay for normal bone marrow progenitor cells showed that the addition of autologous IL-2 plus anti- CD3 MoAb-stimulated PBMC and the BsAb inhibited the formation of granulocyte-macrophage colonies and mixed-cell colonies. However, the degree of inhibition was smaller than that for the AML blast colonies. Taken together, these findings suggest that this BsAb may be useful for ex vivo purging of CD13+ AML cells in autologous bone marrow transplantation.