Expanding use of CD33-directed immunotherapy

KEYWORDS:

• Acute myeloid leukemia (AML)
• CD33
• drug target
• immunotherapy
• myeloid differentiation antigen

The sialic acid-binding immunoglobulin (Ig)-type lectin (Siglec) CD33 is a cell surface glycoprotein with elusive function(s) characterized by a membrane distal V-set and a membrane proximal C2-set Ig-like domain [1]. CD33 (Siglec-3) is primarily found on maturing and mature myeloid cells (including monocytes, macrophages, and microglial cells) [1,2] and their neoplastic cell counterparts. Because of this expression pattern, CD33 has long been an attractive therapeutic target, particularly for acute myeloid leukemia (AML) because CD33 is displayed on at least a subset of AML blasts in almost all cases and, possibly, underlying leukemia stem cells in some [3,4]. As unconjugated CD33 antibodies proved ineffective, attention turned to antibody-based delivery of toxic effector molecules, exploiting the endocytic properties of CD33 [2–5]. Most work has focused on the antibody–drug conjugate gemtuzumab ozogamicin (GO), which induces remissions in up to 25–35% of patients with newly diagnosed or relapsed/refractory AML when given alone. Demonstration of improved survival of some patients with GO add-on to intensive induction chemotherapy [5,6] led to GO’s approval in the U.S. and Europe for CD33⁺ AML and validates CD33 as target for AML immunotherapy.

In parallel to the development of new, more effective CD33-directed therapeutics (e.g. antibody-drug conjugates, radioimmunoconjugates, bispecific antibodies, chimeric antigen receptor [CAR]-modified T cells) to overcome the limitations noted with GO [5,7], interest has grown in CD33 as drug target for other malignant and nonmalignant disorders. Herein, I will briefly summarize these emerging efforts which include the targeting of CD33 splice variants not recognized by most current CD33 antibodies as well as the targeting of CD33⁺ tumor cells in other hematologic malignancies, CD33⁺ myeloid-derived suppressor cells (MDSCs) in a variety of diseases, and normal CD33⁺ microglial cells in Alzheimer disease.

1. Targeting CD33 splice variants

Alternative splicing results in several shorter isoforms of CD33 [5]. Currently of greatest interest in AML is one isoform missing exon 2 (CD33^∆E2) which contains the C2-set but not V-set domain. A single nucleotide polymorphism (SNP), rs12459419, present in exon 2 close to the intron/exon junction, modulates exon 2 splicing efficiency. The minor (T) allele results in preferential transcription of CD33^∆E2 and reduced expression of full-length CD33 (CD33^FL). CD33 isoforms without the V-set domain are important because this domain contains the immune-dominant epitope(s) to which most CD33 antibodies, including GO, bind [5]. Data from the COG-AAML0531 trial, showing improved outcome with GO add-on to intensive chemotherapy was limited to patients homozygous for the major C allele in rs12459419, suggest CD33 splicing events may be clinically important [5]. This observation has prompted interest in targeting CD33^∆E2 with antibodies recognizing the C2-set domain, perhaps particularly for the 50% of patients with rs12459419 CT or TT genotypes. However, whether endogenous CD33^∆E2 mRNA is translated, post-translationally modified, and transported to the cell surface of AML cells (or, by extension, other cells) is unclear. In our recent studies conducted with newly developed CD33^∆E2-specific antibodies, we were unable to detect CD33^∆E2 on human AML cell lines or primary blast cells from a smaller cohort of AML patients, which contained cells with CC, CT, and TT rs12459419 genotypes [8]. Still, the development of a novel C2-set domain-binding CD33/CD3 bispecific antibody (JNJ-67571244) has recently been reported [9], and this drug has entered early testing in patients with relapsed/refractory AML and high-risk myelodysplastic syndrome (MDS; NCT03915379).

2. Targeting CD33 in MDS and other myeloid hematologic malignancies

In addition to AML and acute promyelocytic leukemia – the latter being the leukemia with the greatest sensitivity to GO [5] – CD33 is frequently expressed on neoplastic cells in other myeloid malignancies such as MDS and myeloproliferative neoplasms (e.g. chronic myeloid leukemia). Patients with these malignancies have been included in some of the trials with GO [5] but have not been a primary focus of CD33-targeted immunotherapy. This may change, particularly for MDS. Part of this interest is based on the increasing recognition that CD33⁺ MDSCs are expanded in MDS and may contribute to the immune suppressive tumor microenvironment, disease progression, and ineffective hematopoiesis [10]. Paralleling similar observations with AML-derived MDSCs [11,12], a fully human Fc-engineered CD33 antibody (BI 836858) and CD33-directed T cell and natural killer (NK) cell-engaging bispecific antibodies have been shown to kill putative MDSCs from patients with MDS and reverse MDSC immune suppression [12–14]. It is possible CD33 offers more than just a target for MDSC therapy solely because of its cell surface expression. Studies in MDS-derived MDSCs showed knock-down of CD33 via CD33-specific shRNA resulted in improved hematopoiesis, consistent with the idea CD33 may play a direct role in the suppressive functions of MDSCs [15]. Indeed, these studies identified S100A9 as a CD33-interacting protein and suggested S100A9/CD33 forms a functional ligand/receptor pair that recruits components to the immunoreceptor tyrosine-based inhibitory motifs of CD33, leading to the secretion of suppressive cytokines [15]. These observations suggest that inhibiting CD33 signaling, e.g. via antibodies that block the interaction between CD33 and ligands such as S100A9 (as suggested in studies with BI 836858 [14]) or via downmodulation of cell surface expression of CD33, may partly underlie how CD33-directed therapy could inhibit MDSCs and improve outcomes in MDS. A trial using BI 836858 in lower-risk MDS has been completed (NCT02240706). Several trials with CD33-targeted bispecific or trispecific antibodies (e.g. GTB-3550 [NCT03214666], JNJ-67571244 [NCT03915379], AMV564 [NCT03516591, no longer recruiting], or CD33-targeting CAR-modified T cells have been initiated for patients with MDS and other non-AML myeloid neoplasms.

3. Targeting CD33 in non-myeloid hematologic malignancies

CD33 is expressed on a subset of cases of non-myeloid hematologic malignancies, including acute lymphoblastic leukemia of B and T cell lineage, some mature T cell malignancies, anaplastic large cell lymphoma, mast cell neoplasms, blastic plasmacytoid dendritic cell neoplasms, and plasma cell neoplasms [16,17]. The latter are an increasing focus of interest for CD33-targeted immunotherapies. In a large series of multiple myeloma cases, 20% of 915 cases were found to be CD33⁺ with either a homogenous (12%) or heterogenous (8%) expression pattern [18]. In smaller series, the proportion of CD33⁺ multiple myeloma cases ranged between 6.5% and 35% [19–26]. Limited data indicate CD33 is also displayed on abnormal plasma cells in monoclonal gammopathies of undetermined significance although perhaps only in a small minority of cases [19]. Adding to the attractiveness as drug target, CD33 expression has been associated with adverse disease features (e.g. immature or plasmablastic cell type, higher-risk cytogenetic abnormalities, higher LDH), lower rates of very good partial remissions with non-transplant therapies, reduced effectiveness of autologous transplantation, increased 1-year mortality, and shorter survival [21,22,25,26], with some multivariable models indicating independent prognostic significance of CD33 expression for shorter progression-free and overall survival [25,26]. Moreover, CD33 expression levels may be increased after a variety of myeloma therapies or at relapse [22,26]. One trial is currently testing the CD33 antibody lintuzumab (HuM195, SGN-33) conjugated to the alpha-particle emitting radionuclide actinium-225 in adults with CD33⁺ relapsed/refractory multiple myeloma (NCT02998047).

A second rationale to pursue CD33-directed immunotherapy for multiple myeloma, potentially expanding its use beyond patients whose neoplastic plasma cells express CD33, comes from recent studies implicating MDSCs in myeloma cell growth and disease progression as well as immune suppressive properties of the surrounding microenvironment [27,28].

4. Targeting CD33⁺ MDSCs in disorders other than CD33⁺ hematologic malignancies

Data from an increasing number of studies support a pivotal role of MDSCs in the pathogenesis of many disorders. These include cancers where MDSCs are considered important contributors to immune suppression and evasion, tumor angiogenesis, drug resistance (including resistance to checkpoint inhibitor therapy, bispecific antibodies, and CAR-modified T cells), and tumor metastasis, but also infections and chronic inflammatory diseases [29–33]. Particularly for patients with cancer, this has raised interest in therapeutically targeting MDSCs, with a variety of approaches being currently explored [32,34,35]. The data summarized above with MDSCs from patients with AML and MDS suggest that CD33 might be a suitable target for MDSC therapy even beyond the treatment of CD33⁺ hematologic malignancies. In fact, based on the notion that the CD33/CD3 tandem diabody AMV564 targets AML- and MDS-associated MDSCs [12], a trial with this agent for adults with advanced solid tumors has been initiated (NCT04128423).

5. Targeting CD33 in Alzheimer disease

A SNP in the proximal promoter of CD33 (rs3865444) has been associated with susceptibility for Alzheimer disease [1,36]. rs3865444 is in strong linkage disequilibrium with rs12459419, with the minor allele being associated with reduced risk of Alzheimer disease. While underlying mechanisms remain incompletely understood, a direct role for CD33 in the pathogenesis of Alzheimer disease is hypothesized based on the observation that CD33^FL but not CD33^∆E2 reduces uptake and clearance of toxic amyloid β species by microglial cells, possibly by inhibiting activating receptors involved in microglial phagocytosis such as TREM2 [1]. Monocytes from people homozygous for the major C allele in rs3865444 have decreased phagocytic activity compared to cells from people homozygous for the protective minor A allele. Based on these data, it has been proposed that blocking the function of CD33^FL could be exploited as a means to enhance microglial phagocytosis of amyloid β plaques. Early-phase clinical testing of AL003, a CD33 antibody claimed to block the function of CD33 and increase the activity of microglia cells, has begun in healthy volunteers and patients with mild to moderate Alzheimer disease (NCT03822208). An alternative approach to alter the function of CD33 might be the use of selective small molecules occupying the carbohydrate-binding site of CD33^FL, as indicated by recent studies in which a sialic acid mimetic (P22) presented on microparticles increased uptake of toxic amyloid β species into microglial cells [37].

6. Expert opinion

CD33-targeted immunotherapy as come of age. With improved outcome of some patients treated with GO, it now has an established role in the treatment of AML [5]. Still, CD33-targeted therapy for AML is evolving, with major current efforts dedicated to the development of new, more potent therapeutics that can overcome GO’s shortcomings. Using more and more potent CD33-targeted agents comes at a price, however. Because of physiologic expression of CD33 on several normal cell types, unwanted CD33-specific (‘on target’) toxicities have to be expected with the use of CD33-targeted therapies. AML has served as a paradigm for the demonstration of the clinical challenges with potent CD33-directed therapeutics with regard to toxicities to normal CD33⁺ cells. These have manifested primarily as prolonged cytopenias and associated risks of infections and/or bleeding, but also as hepatic toxicities that may be, at least partly, due to the targeting of CD33-expressing cells (Kupffer cells, sinusoidal endothelial cells, stellate cells) in the liver [5,38]. A potential breakthrough in widening the therapeutic window of such therapeutics has been the observation that normal hematopoietic stem and progenitor cell populations can be engineered to lack CD33 (e.g. via CRISPR/Cas9), seemingly without resulting impairment of engraftment potential or other functionality [39–41]. This might enable the use of engineered hematopoietic stem cell products in people considered for CD33-targeted therapies to mitigate toxic effects on normal cells – a strategy that will soon be tested in the clinic.

Partly fueled by an increasing understanding where CD33-expressing cells play a pivotal pathophysiologic role, increasing efforts are taken to explore CD33 as drug target for other malignant and nonmalignant disorders. Particularly for the targeting of MDSCs in MDS, preclinical studies provide strong rationale to pursue such strategies in patients. Early clinical result can be expected for many of the targeted disorders in the next 2–5 years and will hopefully clarify whether CD33 will provide a validated drug target for diseases other than AML.

Declaration of interest

RB Walter received laboratory research grants and/or clinical trial support from Amgen, Aptevo, Celgene, Immunogen, Macrogenics, Jazz, Pfizer, and Selvita; has ownership interests in Amphivena; and is (or has been) a consultant for Agios, Amgen, Amphivena, Astellas, Aptevo, Bristo Myers Squibb, Janssen, Jazz, Kite, Macrogenics, Pfizer, and Race Oncology. 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.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial relationships or otherwise to disclose.