The role of CD33 as therapeutic target in acute myeloid leukemia

Abstract

CD33 is a myeloid differentiation antigen that is displayed on acute myeloid leukemia (AML) blasts in most patients and, possibly, leukemic stem cells in some, and has thus served as target for antibody-based therapies for many years. Validation for this approach comes from the antibody–drug conjugate, gemtuzumab ozogamicin, which improves survival of some patients with AML when added to induction chemotherapy. Still, CD33 is a challenging target because of its low expression and slow internalization; these characteristics limit antibody-dependent cell-mediated cytotoxicity and intracellular drug accumulation and, consequently, the activity of unlabeled and toxin-carrying antibodies. Very promising preclinical data are now available from an improved antibody–drug conjugate and CD33-targeted strategies that redirect immune effector cells to eradicate the leukemia, most notably bispecific antibodies and chimeric antigen receptor T-cell immunotherapy. In parallel to their clinical testing, efforts will be needed to identify the patients that most likely benefit from such agents and the disease stage in which they are most efficacious. With enhanced activity of CD33-directed therapies, toxic effects on normal hematopoiesis will increase and require excellent supportive care measures, or even rescue with donor cells, to minimize morbidity and mortality from expected cytopenias and to optimize treatment outcomes with these therapeutics.

Keywords::

• acute myeloid leukemia
• antibody
• antibody–drug conjugate
• bispecific antibody
• bispecific T-cell engager
• CD33
• chimeric antigen receptor
• gemtuzumab ozogamicin
• immunotherapy

1. Introduction

CD33 is a myeloid differentiation antigen that can be displayed on some normal B-cells and activated T- and natural killer cells but is not expressed on pluripotent hematopoietic stem cells or outside the hematopoietic system [1]. It is found on at least a subset of blasts in nearly all acute myeloid leukemias (AMLs). With an average of ∼ 10⁴ molecules/leukemic cell, CD33 is not highly abundant but levels vary considerably across individual patients. Besides its broad expression on AML blasts, a main impetus to pursue CD33 as therapeutic target emanated from the notion that some AMLs may predominantly or entirely involve committed CD33⁺ myeloid precursors, suggesting that this antigen could serve to eradicate malignant stem cells in such leukemias, with acute promyelocytic leukemia (APL) being one possible example [1].

Although crosslinking of CD33 can induce AML cell apoptosis in vitro, the results with unconjugated CD33 antibodies have so far been disappointing. For instance, the best-studied molecule (lintuzumab) has very modest activity as single agent even at supra-saturating doses and failed to improve survival when added to intensive or non-intensive chemotherapy in two randomized trials [1]. Because of the endocytic property of CD33, it was suspected early that adding a toxic payload might improve the efficacy of CD33 antibodies. Indeed, gemtuzumab ozogamicin (GO), a CD33 immunoconjugate carrying a calicheamicin-γ₁ derivative, induces remissions in a subset of AML patients [1]. More importantly, several randomized studies have demonstrated that addition of GO to conventional chemotherapy reduces the risk of relapse and improves survival of previously untreated patients with favorable-risk and, to a lesser degree, intermediate-risk AML [2], justifying its continued availability for the treatment of these patients and validating CD33 as therapeutic target.

2. Emerging strategies to improve CD33-directed therapy

Regardless, CD33 is a challenging target even for armed antibodies: the development of a CD33 antibody conjugated to a thiol-containing maytansinoid derivative (AVE9633) was terminated because of modest single agent activity [1], and similarly modest activity was observed with an immunoconjugate carrying recombinant gelonin (HuM-195/rGel) [3]. Two potential reasons for the poor clinical activity of CD33 antibodies are the oftentimes relatively low cell surface density of CD33 and the slow internalization kinetic of CD33/antibody complexes, which limit antibody-dependent cell-mediated cytotoxicity (ADCC) and intracellular accumulation of antibody-delivered payloads. The latter may then be extruded from the AML cell before exerting any cytotoxic effects. Such limitations may explain why GO is ineffective in many patients even though a highly potent toxin is delivered [1]. Nonetheless, the survival improvement seen in some patients with GO, together with the fact that GO is no longer available in most countries, provides a strong impetus to develop novel CD33-directed drugs that might be more efficacious than this first-generation antibody–drug conjugate.

2.1 Unconjugated antibodies

Efforts are ongoing to generate engineered CD33 antibodies with better ADCC than lintuzumab, and a fully human antibody with these properties (mAb 33.1) is currently tested clinically (NCT01690624).

2.2 Antibody–drug conjugates

Besides cellular extrusion of the calicheamicin-₁ derivative, another factor limiting GO's activity was heterogeneous and incomplete drug loading [1]. Obvious technological advancements would therefore include improvements of conjugation and linker technology as well as use of highly potent toxins that are poor substrates for drug transporters. One such candidate is SGN-CD33A, a humanized CD33 antibody with engineered cysteines carrying a synthetic DNA cross-linking pyrrolobenzodiazepine dimer [4]. After preclinical studies have demonstrated that SGN-CD33A is more potent than GO against AML cell lines and primary AML cells and maintains activity in models of multidrug-resistant disease, this drug has recently entered clinical testing (NCT01902329).

2.3 Radiolabeled antibodies

Because of the radiosensitivity of AML, radionuclides are an attractive means to augment the efficacy of therapeutic antibodies. However, the low CD33 abundance limits how much radiation can be delivered via antibodies, and immunoconjugate metabolization shortens the intracellular retention of radio-isotopes. Nevertheless, studies with ¹³¹I demonstrated that radiolabeled CD33 antibodies could be used as single agent or part of pretransplant conditioning regimens [5]. Later studies with α-emitters similarly found these to be safe and to have anti-leukemia activity [5], and trials with an ²²⁵Ac-labeled antibody are ongoing (NCT00672165, NCT01756677).

2.4 Bispecific antibodies

A long-pursued strategy to improve antitumor antibodies encompasses bispecific constructs that harness the immune system in the elimination of cancer cells. Many types have been explored over the years, but their success was limited by suboptimal effector cell recruitment and challenges with antibody production [6]. As evidenced by recent data on T-cell-directed constructs, such shortcomings may not pertain to small single-chain constructs that only contain the minimal binding domains of the two antibodies. By binding CD3, they bring polyclonal T-cells in close proximity of targeted tumor cells and force formation of a synapse that triggers lymphocyte activation and serial destruction of attached tumor cells through perforin/granzyme-mediated apoptosis at low effector-to-target (E:T) ratio. Therapeutic utility of these molecules for acute leukemias is suggested by emerging results from small studies with the CD19/CD3 bispecific T-cell engager (BiTE) antibody, blinatumomab, showing a 70 – 80% response rate and high relapse-free survival rates among adults with CD19⁺ acute lymphoblastic leukemia (ALL) that persisted or relapsed after chemotherapy [7].

Developing small bispecific antibodies targeting CD33 is a logical consequence of the experience with GO and the encouraging experience with blinatumomab. We and others have recently demonstrated that a CD33-directed candidate built on the BiTE platform (AMG 330) causes highly potent cytolysis of CD33⁺ AML cell lines and primary AML cells in the presence of healthy donor or patient T-cells at low E:T ratio in vitro and in immunodeficient mice [8-10]. Target antigen density, antibody dose and E:T ratio are critical determinants for the drug's activity, whereas cytolysis is not affected by drug transporter activity [10]. Other CD33/CD3-targeting single-chain antibodies confirm the high anti-AML activity of these constructs [11]. None of these bispecific antibodies have been brought to the clinic yet. Despite their high preclinical potency, however, further improvements seem possible. For example, a modular targeting system that not only brings CD33⁺ AML cells together with CD3⁺ T-cells but also provides a co-stimulatory signal yielded more efficient cytolysis of AML cells expressing low levels of CD33 [12]. Moreover, several groups explore whether targeting a second tumor antigen simultaneously with CD33, or targeting additional/alternative immune effector cell antigens, could provide superior anti-AML efficacy.

2.5 Chimeric antigen receptor T-cell immunotherapy

An alternative strategy to redirect the immune system in the eradication of leukemia cells is adoptive immunotherapy using T-cells genetically engineered to express chimeric antigen receptors (CARs). These hybrid single-chain receptor constructs contain an extracellular tumor antigen-recognizing domain linked to an intracellular component composed of the CD3 zeta chain as primary domain with or without additional co-stimulatory endodomains that, together, activate immune effector cells upon binding to the tumor antigen [13]. Complete responses were observed in small series of ALL patients treated with CAR T-cells recognizing CD19 [13]. Various types of T-cells modified to express CD33-directed CARs have now been generated and shown to very efficiently reduce the burden of AML cells in preclinical models [14-16]. A first clinical trial testing CD33-directed CAR T-cells is currently ongoing (NCT01864902).

3. Conclusion

Considering that survival improvements were observed with a CD33 immunoconjugate that had significant limitations with regard to drug labeling and susceptibility to drug extrusion, the true potential of CD33-targeted therapies may not have been elucidated yet. The emerging preclinical data from several novel CD33 antibody-based therapeutics indeed suggest that they might be active against a much broader subset of leukemias than GO. It is thus conceivable that these agents could benefit a high proportion of AML patients. Nevertheless, to avoid the mistake made with GO where little attention was paid to the heterogeneity of AML, efforts should be made early to identify the patient subsets that most likely respond to such therapies to optimize the clinical outcome of these targeted treatments.

With the almost universal expression of CD33 on AML blasts of individual patients, CD33 antibody-based therapeutics may be useful for tumor debulking. On the other hand, it currently remains debated whether this antigen is displayed on AML stem and progenitor cells, and CD33-directed therapies may eradicate such cells only in a subset of leukemias (e.g., APL). Resolving this controversy will be instrumental for the understanding in which disease stages CD33-targeted agents may be beneficial.

As CD33 is widely expressed on hematopoietic progenitor cells, effects on these cells are likely substantial if the CD33-targeted drug is highly efficacious even at low CD33 expression levels or in the presence of drug transporter activity. Patients may thus experience very prolonged cytopenias and require excellent supportive care or even rescue with donor hematopoietic cells to minimize morbidity and mortality. Anticipated cytopenias may prove a particular hurdle for CD33-directed CAR T-cells as forming myeloid cells could be continuously destroyed, which could lead to permanent cytopenias without the use of suicide genes or myeloablative conditioning to limit their long-term persistence in patients.

Undoubtedly, with the various new CD33 antibody-based therapeutics on the horizon, we are entering a new treatment era. Many questions remain open, but these agents offer exciting new prospects for a disease for which the outcomes remain unsatisfactory for many patients.

Declaration of interest

The author was supported by 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.