Where do we stand with radioimmunotherapy for acute myeloid leukemia?

Abstract

Introduction:

Despite the approval of several new drugs, deaths from acute myeloid leukemia (AML) remain common. Because of well-defined cell surface antigens, easy accessibility, and radiosensitivity of leukemia cells, there is long-standing interest in radiolabeled antibodies (radioimmunotherapy [RIT]) to complement or replace existing treatments and improve outcomes in AML.

Areas covered:

Targeting primarily CD33, CD45, or CD66, early RIT efforts have focused on β-emitters, including iodine-131 (¹³¹I) and yttrium-90, mostly to intensify conditioning therapy before allogeneic hematopoietic cell transplantation (HCT). An ¹³¹I-labeled CD45 antibody (Iomab-B [apamistamab-I131]) is currently studied in the registration-type phase 3 SIERRA trial (NCT02665065) for this purpose. Of growing interest as therapeutic payloads are α-particle emitting radionuclides such as actinium-225 (²²⁵Ac) or astatine-211 (²¹¹At) since they deliver substantially higher decay energies over a much shorter distance than β-emitters, rendering them more suitable for precise, potent, and efficient target cell killing while minimizing toxicity to surrounding bystander cells, possibly allowing use outside of HCT. Clinical efforts with ²¹¹At-labeled CD45 antibodies and ²²⁵Ac-labeled CD33 antibodies (e.g. ²²⁵Ac-lintuzumab [Actimab-A]) are ongoing.

Expert opinion:

A first anti-AML RIT may soon become available. This might propel further work to develop RIT-based treatments for AML, with many such efforts already ongoing.

1. Introduction

The recent regulatory approval of several new drugs has substantially expanded the therapeutic options for patients with acute myeloid leukemia (AML) [1–3]. Nonetheless, despite their increasing use and integration into evolving treatment algorithms with multiagent chemotherapy and allogeneic hematopoietic cell transplantation (HCT), deaths from leukemia remain all too frequent. Hence, there remains the need for additional, effective agents that could complement or replace existing ones and improve outcomes. Because of well-defined cell surface antigens and easy tumor cell accessibility, antibody-based therapeutics have long been pursued for this purpose in AML. As the success of the first unconjugated antibodies was limited in this malignancy [4], investigators have turned quickly to antibodies linked to toxic payloads. To date most extensively used is gemtuzumab ozogamicin (GO), a CD33 antibody-drug conjugate delivering a DNA damaging derivative of calicheamicin-γ₁^I that is now approved for children and adults with CD33+ AML [5–8]. From early on, however, there has been interest in arming antibodies with radionuclides. This review briefly summarizes past and present efforts with radioimmunoconjugates in AML.

2. Rationale for radioimmunotherapy (RIT) in AML

Conceptually, radionuclides are excellent payloads for antibodies against AML considering acute leukemia cells can be exquisitely sensitive to ionizing radiation [9]. This effect has some dose dependency [10], which is exploitable as demonstrated, for example, by a randomized trial showing significantly fewer relapses with higher compared to lower doses of total body irradiation (TBI) when given as conditioning before allogeneic HCT in adults with AML in first remission [11, 12]. The observation that this benefit did not translate into better survival because of higher non-relapse mortality related to toxicities to lung, liver, and mucous membranes provided a strong rationale for radiolabeled antibodies (radioimmunotherapy [RIT]) to direct radiation toward AML cells. Because normal tissues are relatively spared, it is anticipated RIT creates a therapeutic window relative to conventional radiotherapy [13–16]. As an advantage over some of the other antibody-based therapies, RIT does not require a functional immune system to exert anti-tumor effects.

3. Radionuclides as payloads against AML

Several radionuclides have been exploited for RIT in AML [17–19], each offering advantages and disadvantages. Early studies have primarily involved β-emitters such as iodine-131 (¹³¹I), yttrium-90 (⁹⁰Y), rhenium-188 (¹⁸⁸Re), or lutetium-177 (¹⁷⁷Lu). Because they deliver radiation over a relatively long distance, they can provide a crossfire effect, i.e. they can deliver radiation to bystander cells. This may afford a therapeutic advantage when tumor access of the antibody is limited, when some of the tumor cells express limited amounts of the antibody target or lack it altogether, or when there is interest in delivering radiation to the entire marrow space, e.g. when RIT is used before HCT. However, since normal cells are also subject to this bystander effect, β-emitters are not ideal for selective eradication of leukemia cells. Particularly with the latter in mind, there is increasing interest in α-emitters such as astatine-211 (²¹¹At), actinium-225 (²²⁵Ac), and bismuth-213 (²¹³Bi) [17, 19, 20]. Compared to β-emitters, α-emitters deposit a higher decay energy (5–8 MeV vs. 0.66–2.3 MeV) over much shorter distances (55–70 μm vs. 0.3–2.3 mm) for potent, more precise kill of targeted cells while minimizing toxicities to normal, non-targeted surrounding cells [13, 21, 22], rendering them potentially suitable for non-HCT settings. For several α-emitters, experimental studies have documented that as few as 10 hits per cell or less kill hematopoietic neoplasms [23–26]. This allows their effective use in combination with an antibody even for low-density targets and RIT products where only a small proportion of the antibody molecules are carrying a radionuclide.

4. Suitable targets for RIT of AML

As for any other antigen-directed therapy, the selection of the target is critical for the success of radioimmunoconjugates. Important characteristics include homogenous display at high levels in most or all malignant cells including underlying cancer stem and progenitor cells and minimal or no expression in (vital) normal tissues. Ideally, the target is not shed into the circulation and is not overly rapidly internalized, as metabolism of the radioimmunoconjugate could lead to freely circulating radionuclides if not retained effectively in target cells. Undeniably, the attractiveness of the antigen under consideration is increased if it plays a defined role in disease pathogenesis or prognosis. Although AML is immunophenotypically highly heterogeneous and there is no shortage of antigens to choose from, clinical RIT efforts have so far been limited to a relatively small number of targets, most notably CD33, CD45, and – even though not commonly expressed on AML blasts – CD66. CD33 is displayed on normal maturing and mature myeloid cells, including multipotent myeloid precursors, with a debate ongoing whether some normal hematopoietic stem cells also express CD33 [5, 27, 28]. As a myeloid differentiation antigen, CD33 is displayed on at least a subset of the leukemic blasts in almost all patients and possibly leukemia stem cells in some [27, 29]. Antigen density is relatively low, averaging ~10,000 CD33 molecules per AML blast, and varies greater than 2-log-fold across individual leukemias. CD33 is internalized after antibody binding. While this endocytic property enabled the success of GO, it proved problematic in early efforts with ¹³¹I-labeled CD33 antibodies, in which limited retention of the radioisotope in sites of leukemia were noted.

CD45 (leukocyte common antigen) is a transmembrane protein tyrosine phosphatase that is expressed on almost all hematopoietic cells. Exceptions are mature thrombocytes, mature erythrocytes, and some of their progenitors. Consistent with this broad expression on cells of the blood system, most hematologic malignancies, including most cases of AML, express CD45 [30]. Ideal for RIT applications, CD45 is displayed at a relatively high copy number (200,000/cell). It is relatively stable on the cell surface (although CD45-directed immunotoxins [31] and antibody-drug conjugates [32–34] are currently explored), and there is no appreciable shed after ligand binding [35]. While several isoforms of CD45 exist, the antibodies used so far as basis for CD45-directed therapeutics are pan-specific [30].

Like CD45, CD66 antigens are relatively stably displayed on cells [36]. These glycoproteins are expressed on members of the carcinoembryonic-antigen-related cell-adhesion molecule (CEACAM) family of proteins [37]. CD66 antigens are not only found on epithelial and endothelial cells but also on some hematopoietic cells, in particular myeloid cells from the late myeloblast or early promyelocyte stage. On the other hand, they are only occasionally found on AML cells. Thus, CD66-directed RIT for AML, so far using antibodies recognizing several CD66 antigens, primarily relies on bystander effects for anti-tumor efficacy.

Preclinical studies have suggested other AML-cell associated antigens may have value for RIT. These include the interleukin-3 (IL-3) receptor α-chain (CD123) [38–40]. Compared to CD45 and CD33, CD123 is displayed on a much smaller subset of normal cells. Expressed on some hematopoietic progenitors, CD123 is rapidly lost during erythroid and megakaryocytic differentiation and decreased during monocytic differentiation. On mature cells, CD123 is primarily found on basophils and plasmacytoid dendritic cells. Primitive hematopoietic stem/progenitor cells express little or no CD123, and xenotransplantation studies have shown most normal blood stem cells are unaffected by therapeutics targeting CD123. On the other hand, CD123 is widely displayed on AML blasts (45–95% of cases) [38, 40, 41]. What makes CD123 particularly attractive is its overexpression on leukemic stem/progenitor cells relative to normal hematopoietic stem/progenitor cells. This expression pattern renders CD123 an ideal target for cancer (stem) cell-specific treatment of acute leukemia, expected to cause less on-target toxicities to normal tissues than CD45 or CD33. Like CD33, CD123 is expressed on leukemia cells at several thousand copies per cell, making it amenable to therapeutic targeting with radioisotopes, as indicated by some (but not all) experiments with CD123 antibodies labeled with γ-ray emitting indium-111 (¹¹¹In) [42–44] and, more recently, early findings from human AML xenograft studies in mice with ²¹¹At-labeled CD123 antibodies [45]. Pharmacokinetic and dosimetry studies in mice xenografted with human AML also suggest that ²¹¹At-labeled CXCR4 antibodies could be of therapeutic interest [46], but efficacy assessments for this approach are currently not available.

5. RIT to augment conditioning therapy before HCT

RIT efforts to date in AML have largely aimed at strategies to augment transplant conditioning regimens [47]. Especially with β-emitters, this focus is not surprising considering their path length and expression of target antigens on normal cells. A series of early phase clinical trials have demonstrated that CD33, CD45, and CD66 antibodies carrying β-emitters can be combined with myeloablative or non-myeloablative conditioning regimen before allogeneic HCT for advanced AML or other high-grade myeloid neoplasm and provide radiation to marrow, spleen, and lymph nodes in patients with high or low tumor burden [47]. Most advanced in the testing is an ¹³¹I-labeled CD45 antibody (clone BC8), which can be safely integrated with high-dose chemotherapy/TBI or reduced-intensity conditioning in patients with acute leukemias and deliver 2-to-3-fold higher radiation doses to spleen and bone marrow than any critical normal organ together [48, 49].

Building on these data, ¹³¹I-labeled BC8 (Iomab-B [apamistamab-I131]) was very recently studied in the phase 3 SIERRA trial (NCT02665065). As the first ever conducted randomized trial testing a radiolabeled antibody in AML, the significance of the SIERRA trial cannot be overstated. While it does not address the question whether addition of radiolabeled antibodies to HCT conditioning provides an advantage over conventional HCT conditioning alone (the premise upon which many early phase clinical trials were designed), this trial tested a RIT/HCT platform in adults 55 years of age or older with active relapsed or refractory AML, a disease burden often considered too high for standard allogeneic HCT, facilitating their access to allogeneic HCT. Patients with an 8/8 HLA-matched related or unrelated donor available were randomized 1:1 between immediate HCT with Iomab-B (with a therapeutic dose individualized based on upper limit of 24 Gy liver exposure), fludarabine (30 mg/m²/day for 3 days) and 2 Gy TBI, or conventional care including recently approved drugs followed by standard of care HCT if remission is achieved, with cross-over allowed for people who failed therapy. The study, conducted at 24 academic sites in the U.S. and Canada, completed accrual of 151 patients in September 2021. So far, data on the primary endpoint (durable remission rate) or secondary endpoints (overall and event-free survival) have not been released – topline data are expected for the third quarter of 2022 – but early data on the rates of HCT and engraftment/tolerability have been reported [50]. Of 75 patients randomized to the experimental arm, only 10 did not undergo HCT (all because of decline in performance status). Thirteen (17%) of the 76 patients on the conventional care arm achieved remission and proceeded to standard of care HCT, while 38 of the non-responders crossed over to receive Iomab-B/HCT. Iomab-B delivered a median of 16 Gy (range: 5–45 Gy) to the bone marrow. 100% of all evaluable patients receiving Iomab-B successfully engrafted, including those crossing over to Iomab-B/HCT, with a median time to neutrophil engraftment of 15 days and 18–19 days to platelet engraftment. Among evaluable patients, non-relapse mortality at 100 days after HCT was 10% in the experimental arm (6/59) and 15% in the cross-over patients (2/13) [50].

In parallel to the late-phase testing of β-emitter RIT targeting CD45, early clinical efforts with a ²¹¹At -labeled CD45 antibody (clone BC8) are underway. While other α-emitters were explored [51–53], ²¹¹At was felt particularly useful because of its half-life of 7.2 hours (enabling high-yield radiolabeling and practical drug delivery) and because it does not release α-emitting daughter radionuclides that could cause organ toxicity [22]. In animal models, ²¹¹At-labeled antibodies were highly efficacious against acute leukemia, B-cell lymphoma, and multiple myeloma in vivo, including measurable residual disease (MRD) burdens [54–56]. As translation, 2 first-in-human trials testing ²¹¹At-labeled BC8 combined with fludarabine/2–3 Gy TBI before HCT were initiated for adults with acute leukemia or MDS with HLA-matched related or unrelated donors (NCT3128034) and, more recently, HLA-haploidentical donors (NCT03670966). Early data from the first 20 patients treated on the former indicate that doses of ²¹¹At up to approximately 500 μCi/kg can be safely delivered with the CD45 antibody in combination with nonmyeloablative conditioning [57].

6. RIT as non-HCT AML therapy

Efforts to use radiolabeled antibodies in the non-HCT setting are less far advanced – perhaps unsurprising considering CD33, CD45, and CD66 are widely expressed on normal cells. Therefore, “on-target, off-leukemia cell” toxicities must be expected when targeting these antigens, or other antigens that are AML-associated rather than AML-specific, even with α-emitters. Currently most explored as target antigen for RIT as non-transplant AML therapy is CD33. A phase 1 trial with ²¹³Bi-labeled lintuzumab (HuM195; SGN-33) showing blast reductions in 14/18 treated patients [58] provided initial evidence supporting this concept. Because ²¹³Bi requires high activities to achieve clinical effects and, given the short half-life of ²¹³Bi, an onsite generator for drug preparation, ²²⁵Ac was then explored. ²²⁵Ac is 1,000–10,000 times more potent than ²¹³Bi when bound to antibodies [17]. Because these particles do not penetrate the skin, administration in the clinic is less complex than administration of a typical β-emitter. A phase 1 trial of ²²⁵Ac-labeled lintuzumab (now also known as Actimab-A) showed elimination of peripheral blood blasts in 10/16 adults with relapsed/refractory AML; three patients treated with doses ≥1 μCi/kg achieved marrow blasts of ≤5% [17, 59]. Further testing then focused on the upfront setting in older adults with AML deemed unfit for standard induction chemotherapy. In this patient population, ²²⁵Ac-lintuzumab with low-dose cytarabine with yielded a response rate of 69% among 13 patients who received a 2.0 μCi/kg/dose in a phase 1/2 study. However, severe, prolonged thrombocytopenia and neutropenia occurred in many patients. Ultimately, this led to deaths from infections in some of the trial participants, and the RIT dose had to be reduced to 1.5 μCi/kg for further evaluation [60, 61]. Because objective responses were much less common (4/18 treated patients) at this lower dose, the study was closed early [61].

The above observation not only indicates a step dose/efficacy relationship with Actimab-A but also highlights the importance of “on-target, off-leukemia cell” toxicities as key limitation of potent CD33-directed therapeutics. Thus, in the absence of hematopoietic stem cell support, this drug likely has a very narrow therapeutic window. Still, Actimab-A continues to be tested as non-HCT AML therapy together with other agents. One effort combines it with multiagent chemotherapy (CLAG-M [cladribine, high-dose cytarabine, G-CSF, and mitoxantrone]) for patients with relapsed/refractory disease. In a phase 1 trial, in which 18 adults with relapsed/refractory AML were given a single dose of Actimab-A (0.25, 0.5, 0.75, or 1.0 μCi/kg) between day 7–9 after CLAG-M (NCT03441048). At least 25% of the AML blasts were required for study eligibility. In this trial, the maximum tolerated dose of Actimab-A was identified to be 1.0 μCi/kg, with prolonged neutropenia and mucositis noted as the 2 dose-limiting toxicities among 3 treated patients. With a median time to best response was 40 days, across all patients, 5 complete remissions (CR), 5 CR with incomplete platelet recovery (CRp), and 2 morphologic leukemia-free states (MLFS) were achieved, for an overall CR/CRp and CR/CRp/MLFS rate of 10/18 (56%) and 12/18 (67%); among responders, 72% tested negative for MRD by multiparameter flow cytometry [62]. A phase 2 study building on this experience using Actimab-A at 0.75 μCi/kg is under planning. While this study indicates that the addition of Actimab-A to CLAG-M appears safe at this dose, in what way (if any) the radiolabeled antibody improves outcomes with CLAG-M remains to be determined with larger numbers of patients treated. The response rate reported with CLAG-M/Actimab-A seems quite similar to what has been observed in relapsed/refractory AML with CLAG-M alone [63, 64]. At least in our own experience, the MRD-negative remission rates can be high with CLAG-M. For example, among 40 patients treated in a phase 1/2 trial with CLAG-M using mitoxantrone at a dose of 16 mg/m² (rather than 10 mg/m² as done in the Actimab-A study), a CR/CR with incomplete hematologic recovery (CRi) rate of 60% was observed, with 19 (79%) of these remissions being MRD-negative by multiparameter flow cytometry [63]. Building on preclinical data demonstrating synergistic anti-AML efficacy when ²²⁵Ac-lintuzumab was combined with the BCL-2 inhibitor venetoclax [65], a study testing escalating doses of Actimab-A plus venetoclax is ongoing for adults with relapsed/refractory CD33+ AML (NCT03867682), with early data reported on the first 12 patients treated with doses up to 1.0 μCi/kg indicating that this combination therapy is tolerated with a manageable adverse event profile and has some anti-AML efficacy [66]. Further dose escalation to 1.5 μCi/kg of Actimab-A when used together with venetoclax is ongoing.

7. Conclusion

RIT has been explored for over 3 decades as therapeutic strategy to improve outcomes in patients with AML. Early clinical RIT explorations – mostly done in the form of single-institution, “proof-of-concept” trials – have focused on β-emitters attached to CD33, CD45, or CD66 antibodies, primarily to intensify conditioning therapy before HCT. Building on such efforts, an ¹³¹I-labeled CD45 antibody (Iomab-B [apamistamab-I131]) is currently studied in a multicenter, registration-type phase 3 SIERRA trial (NCT02665065) for this purpose, with emerging early data from SIERRA suggesting usefulness. While work with β-emitters continues, there is growing interest in α-particle emitting radionuclides as therapeutic payloads since they deliver substantially higher decay energies over a much shorter distance, rendering them more suitable for precise, potent, and efficient target cell killing while minimizing toxicity to surrounding bystander cells, possibly allowing use outside of HCT. Several trials testing α-emitters in the context of anti-AML RIT are currently ongoing, including studies using ²¹¹At-labeled CD45 antibodies in the HCT setting and ²²⁵Ac-labeled CD33 antibodies (e.g. ²²⁵Ac-lintuzumab [Actimab-A]) in the non-HCT setting.

8. Expert opinion

Pursued for over 3 decades, progress in the preclinical development and patient testing of RIT for AML has been slow. This is at least partly related to the logistical complexities and infrastructure requirements surrounding the use of radioisotopes, particularly for clinical application. Historically, only a small number of institutions have contributed to early-phase explorations with anti-AML RIT, and there are remaining concerns about how RIT-based therapeutics would be used in the real world if available for routine clinical practice. Nonetheless, an increasing number of uncontrolled, early phase trials demonstrate that radiolabeled antibodies can be given relatively safely and with acceptable toxicities to AML patients, even those of older age. While radiolabeled antibodies have sometimes been used as single agent – and anti-AML effects been observed – the clinical investigations led to date have mainly pursued the strategy of integrating RIT agents into myeloablative or non-myeloablative conditioning regimens to augment the delivery of radiation to hematopoietic sites to improve disease control without increasing treatment-related mortality. As radiolabeled antibodies were therefore used together with other therapeutics in these efforts, their contribution to observed anti-AML effects has typically remained unclear. In fact, a well-controlled study to define the benefit of a RIT/HCT platform in a rigorous fashion for patients with AML has so far been lacking. The registration-type phase 3 SIERRA trial testing Iomab-B with fludarabine/low-dose TBI is a true milestone in the field – not only because it was the first randomized RIT trial in patients with AML but also because it was conducted at over 20 academic sites in the U.S. and Canada, thus providing some assurance that RIT for AML could indeed be implemented in many centers with the expertise to treat AML. The SIERRA trial has recently completed accrual. With positive early data available for engraftment and non-relapse mortality [50], and topline information on efficacy expected for later in 2022, there is the real prospect of a first RIT product becoming available relatively soon for the treatment of AML.

Firmly determining to what degree radiolabeled antibodies provide benefit for patients with AML will be critically important for the field. The SIERRA trial is a first opportunity to accomplish this goal. Demonstration of efficacy would provide much-needed validation for the concept of RIT. Undoubtedly, this would propel further work to develop RIT-based HCT and non-HCT treatment strategies for AML and, for some antibodies, also for other malignancies with transplant indications. To this end, diverse preclinical efforts are already ongoing. These include a variety of α- and β-emitters and exploit target antigens beyond CD33, CD45, and CD66. Especially when envisioning RIT approaches that could be used without the need for stem cell rescue via HCT, antigen that are more selectively expressed on AML cells (e.g. CD123) may provide better tolerability in the clinic. Optimizing the delivery of radioisotopes to target cells (e.g. via pre-targeting approaches rather than use of directly radiolabeled antibodies [67–72]) may provide an alternative, possibly complementary approach to the goal of minimizing toxicities to non-targeted tissues and increasing the safety of RIT. At the same time, efficacy limitations of RIT have been noted, even in hematologic malignancies, highlighting the need for efforts aimed at increasing the anti-cancer cell toxicity properties of RIT, as is for example explored with the combination of radiolabeled antibodies with venetoclax [65]. Put together, it is hoped that over the next 3–5 years significant progress can be made in our understanding of the value of current RIT for patients with AML and how the efficacy and safety of RIT-based therapies could be further improved.

Article highlights.

• Because of well-defined cell surface antigens, easy accessibility, and radiosensitivity of leukemia cells, there is long-standing interest in radioimmunotherapy (RIT) for acute myeloid leukemia (AML).

• Targeting primarily CD33, CD45, or CD66, early RIT efforts have focused on β-emitters, including ¹³¹I and ⁹⁰Y, mostly to intensify conditioning therapy before allogeneic hematopoietic cell transplantation (HCT).

• Building on early single-institution data, an ¹³¹I-labeled CD45 antibody (Iomab-B [apamistamab-I131]) is currently studied as component of an HCT conditioning regimen in the registration-type, multicenter phase 3 SIERRA trial (NCT02665065).

• While efforts with β-emitters continue, there is growing interest in α-particle emitting radionuclides such ²²⁵Ac or ²¹¹At as therapeutic payloads.

• α-emitters deliver substantially higher decay energies over a much shorter distance than β-emitters, rendering them more suitable for precise, potent, and efficient target cell killing while minimizing toxicity to surrounding bystander cells, possibly allowing use outside of HCT.

• Clinical efforts with ²¹¹At-labeled CD45 antibodies and ²²⁵Ac-labeled CD33 antibodies (e.g. ²²⁵Ac-lintuzumab [Actimab-A]) are ongoing.

• The prospect of a first anti-AML RIT becoming available soon may propel further work to develop RIT-based treatments for AML, with many such efforts already ongoing.