Biting back: BiTE antibodies as a promising therapy for acute myeloid leukemia

Abstract

The experience with gemtuzumab ozogamicin has highlighted both the potential value and limitations of antibodies in acute myeloid leukemia (AML). Recently, bispecific T-cell engager (BiTE) antibodies have emerged as a means to harness polyclonal cytotoxic T-cells and cause highly efficient lysis of targeted tumor cells. Promising early results have been obtained with the CD19-directed BiTE antibody, blinatumomab, in patients with acute lymphoblastic leukemia. A first candidate for AML is the CD33/CD3 molecule, AMG 330, for which several recent preclinical studies demonstrated high potency and efficacy in destroying CD33⁺ human AML cells. Many questions remain to be addressed, but BiTE antibodies may offer an exciting new tool in a disease for which the outcomes in many patients remain unsatisfactory.

Keywords::

• acute myeloid leukemia
• AMG 330
• antibody
• bispecific
• bite
• cd3
• cd33
• gemtuzumab ozogamicin
• immunotherapy

The demonstration that the CD33 antibody-drug conjugate, gemtuzumab ozogamicin, improves survival of some patients with acute myeloid leukemia (AML) has highlighted the potential of antibodies in this disease [1]. While this observation validates CD33 as therapeutic target, this antigen is challenging for toxin-loaded immunoconjugates because of its relatively low abundance and slow internalization. Together with drug transporter activity, such limitations may explain why gemtuzumab ozogamicin is ineffective in many patients – a major reason for its market withdrawal in most countries [1].

What comes next? Further development of antibody-drug conjugates is ongoing, and one second-generation molecule, SGN-CD33A [2], has recently entered clinical testing. Another line of research pursues the use of bispecific antibodies recognizing a tumor-associated antigen and an immune effector cell antigen – most commonly CD3 on T cells – to harness the immune system in the elimination of cancer cells [3–6]. Many types of such antibodies have been explored over the last 3 decades, but their clinical success was limited by suboptimal effector cell recruitment with need for costimulation or preactivation, requirement for high effector-to-target ratios and antibody concentrations, challenges with construct production and serious side effects from widespread cytokine release [3–6].

Many of these shortcomings may not pertain to bispecific T cell engager (BiTE) antibodies [4–8]. These stable single-chain variable fragment antibodies combine the minimal binding domains of two different antibodies fused in tandem on one polypeptide chain of approximately 55 kDa by a short flexible linker. They bind the invariant epsilon subunit of CD3 to recruit polyclonal cytotoxic T cells without dependency on T cell receptor specificity or MHC class I presentation. Thereby, BiTE antibodies bring CD3⁺ T cells in close proximity of target tumor cell membranes and force formation of an immunological lytic synapse that triggers lymphocyte activation, cytotoxic granule fusion, transient cytokine release and proliferation. Destruction of the attached tumor cell then occurs through membrane perforation by perforin and induction of programmed cell death with activation of caspases via granzymes [4–8]. In various preclinical cancer models, BiTE antibodies were highly potent and effective in a strictly T cell as well as target antigen-dependent fashion and lysed tumor cells serially at low effector-to-target ratios without the need for T cell costimulation or preactivation [4].

The therapeutic utility of BiTE antibodies for acute leukemia is suggested by emerging results with the CD19/CD3 molecule, blinatumomab. Among 20 adults with CD19⁺ acute lymphoblastic leukemia (ALL) who had persistent or relapsed minimal residual disease after chemotherapy, blinatumomab yielded an 80% minimal residual disease response rate and relapse-free survival of 61% after a median follow-up of 33 months [9]. Blinatumomab appears highly efficacious even in situations with larger tumor loads, as evidenced by preliminary data showing that 26 of 36 adults with relapsed/refractory ALL achieved either a hematological complete remission or complete remission with partial hematological recovery [10]. Together, these limited data indicate that CD19-directed BiTE antibodies are noncross-resistant to commonly used chemotherapeutics (and possibly allotransplantation) and can be effective in otherwise treatment-refractory patients.

Developing BiTE antibodies for AML is a logical consequence of these encouraging observations. A first candidate built on this platform is the CD33-directed molecule, AMG 330. Several recent preclinical studies have demonstrated that AMG 330 is highly potent in causing cytolysis of CD33⁺ AML cell lines or primary human AML cells in the presence of healthy donor T cells or autologous T cells from AML patients at low effector-to-target ratio ex vivo [11–13] and in immunodeficient mice in vivo [11]. Target antigen density, antibody dose and effector-to-target cell ratio were identified as critical determinants for the activity of AMG 330. In contrast, cytolysis was not affected by drug transporter activity and unlike bivalent antibodies, AMG 330 did not modulate CD33 expression after continued exposure [13].

With these findings, BiTE antibodies appear promising for AML although several important questions remain open at this early stage. First, and perhaps foremost: which cell surface antigen is best suited as target? To date, most efforts with antibodies in AML have focused on CD33. Depending on how antigen positivity is defined, some amount of CD33 is found on AML blasts in nearly 100% of patients [12,14]. While expression levels vary considerably in human AML [12,14], preclinical data with AMG 330 suggest activity at very low CD33 density [13]. Thus, CD33-directed BiTE antibodies may be active against a wide range of CD33⁺ leukemias. Besides CD33, however, a staggering number of AML-associated antigens have been identified, including some thought to be present on putative AML stem or progenitor cell populations [15]. Since BiTE antibodies do not require internalization of target antigens, but may still be effective if they internalize [8], a vast selection of targets is available for exploitation. So far, only limited data are available on the efficacy of CD3-directed bispecific antibodies directed at antigens other than CD33 [16,17], and their relative therapeutic values are currently unknown. Given the biological heterogeneity of AML, it is conceivable that there is no single best antigen but that, rather, individualized targeting strategies need to be developed and tailored to distinct subsets of patients.

Second: what is the main goal of therapy? With the almost universal expression of CD33 on AML blasts of individual patients [12,14], molecules targeting CD33 may be useful for AML cell debulking. On the other hand, it remains controversial whether, and to what degree, this myeloid differentiation antigen is displayed on AML stem and progenitor cells [1], and CD33-directed BiTE antibodies may eradicate such cells only in a subset of leukemias (e.g., acute promyelocytic leukemia). Other antigens, for example, CD123, may be selected because of their presumed broad expression on AML stem cells. However, since true AML-specific antigens have yet to be identified, targets will need to be chosen prudently, with careful assessment of their expression in normal essential and regenerative tissue and determination of whether overexpression on AML cells provides a suitable therapeutic window.

Third: can the administration be simplified? Because of its short half-live [4], blinatumomab was given by continuous intravenous infusion for 28 days in the ALL trials [9,10], providing a logistic challenge and risk for complications. Patients may accept arduous treatments if they are effective, but future modifications of BiTE antibodies, which enable intermittent dosing would certainly facilitate their clinical use tremendously.

Fourth: what side effects will we see? The most frequent early side effects with blinatumomab are flu-like symptoms (e.g., pyrexia, headache, chills and fatigue) coinciding with the maximal release of cytokines from T cells [18], with some notion that the severity increases with higher tumor burdens [7] and, in some cases, may involve abnormal macrophage activation [19]. As a consequence of T cell activation, this toxicity is likely to occur also in AML patients. Additional toxicities may come from the destruction of AML cells, with anticipated differences based on the amount of efficiently lysed cells and, perhaps, content of cells; the latter may differ across the subtypes and differentiation stages of individual leukemias. Treatment limiting, fully reversible central nervous symptoms (e.g., confusion, disorientation or seizure) that occurred with blinatumomab were suspected to be due to collateral damage from B cell destruction in the brain [7] and pointed to the possibility that toxicities from lysed target cells could be substantial. Finally, perhaps most important are toxicities that may arise from affected normal tissues and cells expressing the tumor-associated antigen, in particular mature and immature hematopoietic cells including underlying stem cells. With CD33, for example, the most primitive stem cells may be spared, but this antigen is widely expressed on maturing progenitor cell populations [1], and significant cytopenias are expected with CD33-directed BiTE antibodies. As BiTE antibodies can be highly effective even at very low target antigen levels, the possibility that target antigens may be expressed on normal hematopoietic stem or progenitor cells is of great concern and may require excellent supportive care strategies or rescue with donor hematopoietic cells to minimize morbidity and mortality.

Fifth: how will AML cells avoid death by BiTE antibodies? The early results with blinatumomab indicate that not all patients will derive long-term benefit from BiTE antibodies. The selection of target antigen-negative tumor cells may play a major role [7], but the relevance of additional mechanisms employed by tumor cells to resist BiTE antibodies is unknown. Target antigen density will likely matter. Other strategies that have allowed tumor cells to evade cytotoxic T cells in other situations and that may be relevant for BiTE antibodies include the secretion of molecules that interfere with T cell adhesion, the expression of proteins that block perforin/granzyme- and caspase-induced apoptosis or the expression of signaling molecules that lead to apoptosis or anergy of T cells or impede their differentiation [4].

And finally: assuming suboptimal responses in some patients, how could the efficacy of BiTE antibodies be enhanced? So far, very little is known about strategies to improve their activity in AML [13], and identifying rational treatment combinations will be an important part in optimizing the therapeutic value of these antibodies. Clearly, we are only at the beginning of the journey with BiTE antibodies; however, they may represent an exciting new tool in the armamentarium against AML for which the outcomes in many patients remain unsatisfactory.

Financial & competing interests disclosure

The author has received research funding from Amgen, Inc., and Seattle Genetics, Inc. and has served as a consultant for Seattle Genetics, Inc. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.