Targeting CD47/SIRPα in Acute Myeloid Leukemia and Myelodysplastic Syndrome: Preclinical and Clinical Developments of Magrolimab

CD47 is a widely expressed transmembrane protein and represents the ligand for the signal regulatory protein alpha (SIRPα), which is identified on macrophages and dendritic cells.^[1] The activation of SIRPα triggers a signal transduction cascade, leading to the inhibition of phagocytosis.^[2–4] In a preclinical model, the expression of mouse CD47 in a human acute myeloid leukemia (AML) cell line resulted in the inhibition of phagocytosis.^[5] Moreover, CD47 was overexpressed on myeloid leukemia cells, mediating cancer cell evasion of phagocytosis by the innate immune system.^[6] Thus, through its interaction with SIRPα, CD47 on the surface of tumor cells (or leukemic cells) appears to be an essential molecule in allowing tumor and potentially cancer stem cells to overcome intrinsic expression of their prophagocytic “eat me” signals, and thereby escape phagocytosis.

As a macrophage inhibitory checkpoint, CD47-SiRPa interaction may be amenable to targeted strategies. In mouse xenograft models, a monoclonal antibody (mAb) targeting CD47 was able to enhance phagocytosis, leading to the inhibition of tumor growth and metastasis. This allowed the elimination of cancer cells from various hematologic malignancies and solid cancers.^[7–12] Although macrophage phagocytosis of the malignant cells represented the major mechanism of antitumor activity, different studies demonstrated that phagocytic cells further triggered a downstream antitumor and cytotoxic, CD8⁺ T-cell immune response. These findings suggest that anti-CD47 antibody-mediated phagocytosis also may mediate downstream T-cell elimination of tumor cells.^[13,14]

A key principle to the success of anti-CD47 therapies is the potential specificity of the CD47/SIRPα signaling pathway blockade, leading to the elimination of cancer cells, while leaving most normal cell counterparts unaffected. The selective targeting of tumor cells by anti-CD47 mAb is attributed to the significantly higher expression potent prophagocytic signal on the surface of cancer cells. These prophagocytic signals, which serve as a second hit that is required to induce phagocytosis in addition to interruption of the CD47-SiRPa axis, are minimally expressed on normal healthy cells. Calreticulin was identified as prophagocytic molecule present on the surface of multiple hematologic and solid malignancies.^[9]

This robust rationale led to the development of an anti-CD47 antibody that showed promising efficacy both in vitro and in vivo, with significant phagocytosis and elimination of leukemic cells.^[15] A first-in-class humanized IgG4 anti-CD47 antibody, magrolimab (previously known as 5F9), was developed, with potent efficacy and favorable pharmacokinetics properties and toxicity profile. The use of this drug in vivo allowed the complete eradication of human AML cells, leading to long-term, disease-free survival of patient-derived xenografts.^[16]

Acute myeloid leukemia is a heterogeneous disease, diagnosed at a median age of 67 years, with around one-third of patients aged 75 years or older.^[17] Cure-rates are relatively low and decrease with age, with a 5-year overall survival (OS) rate of around 35% in patients younger than 60 years and 11% in patients 60 years or older.^[18] Treatment has traditionally been built around intensive cytotoxic chemotherapy, including high-intensity, cytarabine-based regimens, and/or allogeneic stem cell transplantation for patients at high risk of relapse.^[19] This approach is associated with a high risk of induction mortality in older patients and individuals with comorbidities or poor performance status.^[20,21] Over the past 15 to 20 years, frail patients were frequently treated with lower-intensity therapies with hypomethylating agents (HMAs). These regimens were associated with lower rates of remission, less early mortality, and a median OS of around 7 to 8 months.^[22] Recently, the combination of azacitidine with venetoclax showed improvement over azacitidine alone with higher remission rates (complete remission [CR]/ incomplete hematologic recovery [CRi] rate approximately 70%) and improved median OS. There still remains significant room for further improvement.

Over the past few years, a better characterization of biomarkers and pathways underlying the evolution of AML led to the approval of novel targeted therapies and allowed a personalized treatment approach.^[23–25] This resulted in an increased emphasis on molecular profiling that allows the selection of more potent, targeted drugs, in association with a less-intensive backbone therapy, namely HMAs, particularly in patients 60 years and older.^[26,27]

The increased understanding of the role of CD47 overexpression on leukemic cells led to its exploitation as a therapeutic target. However, while preclinical trials demonstrated robust antileukemic effects of anti-CD47 mAb monotherapy, the single-agent activity was modest in humans.^[15] In a Phase 1 trial, magrolimab was evaluated as a single agent in patients with relapsed and/or refractory AML. Anemia was the most common drug-related adverse event (AE). Of patients, 58% had blast count reduction, but no objective responses were observed; insufficient efficacy to allow further development as monotherapy.^[28]

Therefore, combination strategies aiming to enhance the efficacy of CD47 blockade were tested in clinic as follows: relying on the synergy between blocking antiphagocytic and enhancing prophagocytic signals. Prophagocytic signals on cancer cells can be induced by cell damage after treatment with cytotoxic chemotherapies or by epigenetic therapies, such as hypomethylating agents.^[29] This strategy was investigated preclinically with the combination of the hypomethylating agent, azacitidine, and the anti-CD47 antibody, magrolimab.^[30] Azacitidine exerts its antitumoral effect by inducing DNA demethylation as well as its antimetabolite activity. In addition, azacitidine has been shown to significantly increase the expression of the prophagocytic signals, specifically calreticulin, in AML cell lines. The combination of increased prophagocytic calreticulin exposure on leukemia cell surface and interruption of CD47-SiRPa axis was more effective than either single agent alone; it significantly enhanced the phagocytic elimination of AML cells, with dramatically improved survival in AML xenograft models.^[31]

Magrolimab was then investigated in a Phase 1b trial in combination with azacitidine in 52 untreated AML patients, ineligible for induction chemotherapy. The median age of the cohort was 73 years, with 64% of participants having poor-risk cytogenetics, and 65% having TP53 mutations. The safety profile of the combination was similar to azacitidine monotherapy, with the most frequent AEs being anemia (31%), fatigue (19%), increase in bilirubin (19%), neutropenia (19%), thrombocytopenia (17%), and nausea (15%). Only 2 patients discontinued treatment due to an AE. Among the 34 evaluable patients, 65% had an objective response; 44% CR, 12% CRi, 3% partial response, 6% morphologic leukemia-free state, 32% stable disease, and 3% progressive disease. The rate of minimal residual disease negativity by flow cytometry in patients with CR/CRi was 37%. Responses occurred earlier in patients treated with the combination compared with azacitidine alone, with a median time to response of 2.04 months. Of patients, 56% became transfusion-independent. Among the highest-risk group harboring TP53 mutation, 15 of 21 patients (71%) achieved an objective response; 48% CR, 19% CRi, 5% morphologic leukemia-free state, 24% stable disease, and 5% progressive disease, suggesting the efficacy of magrolimab in poor-risk disease. The median duration of response was 9.9 months, with 89% of patients maintaining response at 6 months. After a median follow-up of 12 months, median OS for TP53 wild-type patients was 18.9 months. The median OS for TP53-mutant patients was 12.9 months after a median follow-up of 4 months.^[32]

The superiority of doublet regimens compared with monotherapy was also seen with the combination of HMAs with venetoclax, a BCL-2 inhibitor, tested in a Phase 1b trial among untreated AML patients 65 years and older. The median age was 74 years, and 49% of patients had poor-risk cytogenetics. The CR/CRi rate of the entire cohort was 67% with a median duration of CR/CRi of 11.3 months. The median OS for the entire cohort was 17.5 months.^[33] These results compared favorably with historic outcomes of older patents with newly diagnosed AML with CR/CRi rates of 15% to 28% and median OS of 6 to10 months with HMA.^[34,35]

The encouraging efficacy of these two doublet regimens warrants further investigations of triplet combination strategies with potential improvement in outcomes, particularly with azacitidine plus venetoclax and magrolimab, among older AML patients ineligible for intensive chemotherapy. This triplet combination is expected to increase the duration of CR/CRi, median OS and minimal residual disease–negativity rate, with a focus especially among patients with less than stellar outcomes with the azacitidine and venetoclax doublet, such as patients with secondary AML, or those with adverse cytogenetics or high variant allele frequency TP53 mutations. An additional benefit of this regimen is the potentially favorable safety profile with anticipated lower cumulative myelosuppression compared with other venetoclax-based doublet and triplet combinations. This strategy is being evaluated in an ongoing Phase Ib/II trial (NCT04435691).^[36]

Other investigational approaches include the combinations of magrolimab with mAbs, which were shown to have a synergistic effect.^[8] For AML or MDS, combinations of magrolimab with mAbs targeting CD33 or CD123 may be efficacious. The association of magrolimab with anti–PD-1 or PD-L1 agents may lead to improved responses and further enhancement of T-cell responses and efficacy.^[13,37–39] A Phase 1b trial (NCT03922477)^[40] was recently initiated to evaluate the combination of magrolimab with the anti–PD-L1 agent, atezolizumab, in patients with relapsed and/or refractory AML. Similarly, combining magrolimab or other CD47 antibodies with cytotoxic therapy may be synergistic as cytotoxic therapy will likely cause significant upregulation of stress response prophagocytic signals, and such strategies warrant evaluation in clinic for AML.

Myelodysplastic syndrome is a premalignant clonal hematopoietic disorder affecting patients around the age of 70 years, characterized by ineffective hematopoiesis with varying degrees of dysplasia and cytopenias, and risk of leukemic transformation.^[41] The therapeutic approach and goals of therapy are based on the Revised International Prognostic Scoring System.^[42] Patients with very low or low-risk MDS carry a low risk of leukemic progression and are usually treated with erythroid and myeloid growth factor support.^[43] Conversely, patients with intermediate, high, or very high-risk disease carry a high risk of leukemic transformation (≈ 25%) and treatment with HMAs is considered standard of care.^[44] HMAs therapy resulted in ORR of 17% to 31%, median OS of around 24.5 months, and a longer median time to acute AML progression or death of around 1 year, compared with conventional care regimens or supportive care.^[45,46] However, outcomes can be improved by the incorporation of novel therapies to the armamentarium of higher-risk MDS therapies, the primary focus of MDS therapeutic development.

Similarly to what has been observed in AML, preclinical studies in MDS patients revealed that CD47 expression increased throughout the evolution of the disease. In fact, while CD47 is normally expressed on blast cells in low-risk patients, it becomes overexpressed in high-risk patients.^[47] Thus, CD47 blockade was tested in high-risk MDS patients and showed preclinical efficacy by inducing significant phagocytosis of MDS progenitor cells.^[15] Moreover, treatment with azacitidine induced a 4- to 6-fold upregulation of CD47 expression in MDS cell models, as well a 7- to 10-fold increase in calreticulin expression, suggesting a synergy for this combination by potentiation of the prophagocytic signal.^[48]

In the Phase 1b trial of patients with intermediate to very high-risk MDS, 39 patients with a median age of 70 years were treated with magrolimab plus azacitidine. Sixty-four percent had poor-risk cytogenetics and 13% had TP53 mutation. Myelosuppression was the most frequent AE (anemia 44%, neutropenia 8%, thrombocytopenia 5%) along with infusion reactions and fatigue, in 18% each. No patients discontinued the drug due to AEs. Among the 33 evaluable patients, 91% had an objective response; 42% with CR, 24% with marrow CR, 3% with partial response, 21% with hematologic improvement alone, and 9% with stable disease. Over time, patients experienced deepening responses with CR rate increasing to 56% after 6 months of follow-up. Twenty-two patients with CR/CRi/marrow CR were minimal residual disease negative by flow cytometry. Faster time to initial response was observed with the combination, with a median time to response of 1.9 months. Of patients, 58% became transfusion independent. With a median follow-up of 5.8 months, neither median duration of response nor median OS have been reached, with 6-month OS estimate of 100%.^[49]

Of all somatic mutations identified in MDS, TP53 mutations have been associated with the most inferior outcomes, with a median OS of 6 to 12 months.^[50] Of note, in a previous report of patients with MDS treated in a Phase 1b trial of magrolimab plus azacitidine, four patients had TP53-mutant disease and three of them achieved an objective response (CR/marrow CR). Despite these small numbers, this strategy shows promising efficacy in a poor-prognosis and therapy-refractory population.^[51] The Phase 3 ENHANCE trial (NCT04313881)^[52] is currently evaluating the combination of azacitidine + magrolimab compared with azacitidine alone in untreated patients with MDS.

In conclusion, the novel anti-CD47 antibody magrolimab, demonstrated significant antileukemic activity when combined with azacitidine for the frontline treatment of AML and MDS patients, with high-response rates, prolonged survival and duration of response, both in TP53-mutant and wild-type populations. With a favorable safety profile and activity against patients with traditionally refractory mutations, such as TP53 disease, magrolimab-based combinations are promising to address areas of unmet need, particularly among elderly patients and patients with TP53 mutations.