Immunotherapy with Long-Lived Anti-CD123 × Anti-CD3 Bispecific Antibodies Stimulates Potent T Cell-Mediated Killing of Human AML Cell Lines and of CD123+ Cells in Monkeys: A Potential Therapy for Acute Myelogenous Leukemia

CD123 (IL-3 receptor alpha) is highly expressed on acute myeloid leukemia stem cells and blasts, and represents a promising target of antibody therapies for AML. Anti-CD123 antibodies such as CSL-362 and KHK2823 are currently in clinical development; however, a limitation of these molecules is that they are unable to stimulate T cell-mediated killing of CD123+ AML cells. To exploit the potent activity inherent to T cell immunotherapy while maintaining the favorable dosing regimen of a therapeutic antibody, we have designed a novel bispecific antibody that recruits T cells to attack CD123+ AML stem and blast cells. Such antibodies act via a mechanism known as "redirected T cell-cytotoxicity" (RTCC), because they stimulate targeted T cell-mediated killing regardless of T cell antigen specificity. Unlike other bispecific formats, these antibodies possess a full Fc domain and spontaneously form stable heterodimers that are readily manufactured. Their Fc domain was also engineered to abolish binding to Fcγ receptors (to reduce the potential for nonselective T cell activation), yet preserve binding to human FcRn (to maintain long serum half-life).

We first generated a library of humanized and affinity-optimized anti-CD123 × anti-CD3 bispecific antibodies and assessed their potency using RTCC assays, in which bispecifics stimulated killing by human T cells of the CD123+ AML cell lines KG-1a and TF-1. From this cell-based screen, we selected the bispecific antibody XmAb14045 for testing in animal models. This antibody has 0.1 nM affinity for human CD123, and a T cell-engaging domain with 8 nM affinity for human CD3. XmAb14045 stimulated T cell-mediated killing of KG-1a and TF-1 cells with an EC50 < 1 ng/ml (8 pM). In contrast, XmAb14045 had no cytotoxic activity against the CD123− Raji B cell line, demonstrating target specificity of the T cells.

XmAb14045 had a prolonged serum half-life in mice of 6.2 days, in marked contrast to non-Fc domain-containing bispecific formats. Because this antibody was optimized for human CD123 and CD3 targets and does not crossreact with mouse antigens, we evaluated efficacy in cynomolgus monkeys. We treated 3 monkeys per group with a single dose of XmAb14045 at 0.01, 0.1, or 1 mg/kg. We quantified CD123+ cell numbers, including basophils and plasmacytoid dendritic cells (pDC) as CD123+ surrogate populations for AML stem and blast cells. Within 4 hours of dosing, XmAb14045 strongly activated T cells and stimulated depletion of over 99% of circulating CD123+ cells within 1 hr, particularly at the 0.1 and 1 mg/kg doses. Basophil and pDC counts fell to baseline within 4 hr and remained low for several weeks. Circulating CD4+ and CD8+ T cells were activated immediately after dosing and this was sustained for 48 hr, as measured by markedly increased levels of the activation markers CD25 and CD69. Notably, XmAb14045 induced rapid margination of CD4+ and CD8+ T cells from the circulation, with blood T cell populations returning to baseline within several days. Bone marrow CD123+ cells were depleted by over 95% at all doses, and these cell populations had not recovered by 8 days after treatment.

Our results demonstrate that bispecific antibodies can recruit and activate T cells to efficiently kill CD123+ cells not only in the circulation but also in the bone marrow. Results in monkeys also suggest that changes in basophil and/or plasmacytoid dendritic cell numbers are readily quantifiable in peripheral blood, and thus these populations may serve as biomarkers for clinical efficacy. Our preclinical data provide a rationale for clinical assessment of anti-CD123 × anti-CD3 bispecific antibodies in patients with acute myeloid leukemia.