The First-in-Class Anti-CD47 Antibody Magrolimab (5F9) in Combination with Azacitidine Is Effective in MDS and AML Patients: Ongoing Phase 1b Results

Background

Magrolimab (previously named 5F9) is a first-in-class antibody targeting CD47, a macrophage immune checkpoint and "don't eat me" signal on cancers. CD47 blockade induces tumor phagocytosis and eliminates leukemia stem cells (LSC) in AML models. Azacitidine (AZA) synergizes with magrolimab by inducing "eat me" signals on AML, to enhance phagocytosis. A Phase 1b trial of magrolimab+AZA was initiated in MDS/AML patients with preliminary reported results mainly from the safety cohort demonstrating high response rates in both diseases. Here we report data from the expansion cohort of this ongoing trial.

Methods

Results from this Phase 1b reported here focus on treatment of magrolimab+AZA in untreated intermediate to very high risk MDS patients by IPSS-R and untreated AML (induction chemotherapy ineligible) patients. A magrolimab priming/intrapatient dose escalation regimen (1-30 mg/kg weekly) was utilized to mitigate on target anemia. AZA dosing was 75mg/m² days 1-7 on a 28 day cycle. Responses were assessed by IWG 2006 and ELN 2017 criteria for MDS and AML patients, respectively.

Results

43 patients (18 MDS and 25 AML) with a median of 73 years of age were treated with magrolimab+AZA. 19% were intermediate cytogenetic risk with 63% poor risk (19% unknown). 28% of patients harbored a TP53 mutation. Magrolimab+AZA was well-tolerated with a safety profile similar to AZA monotherapy. Treatment-related AEs (>15% of patients) for magrolimab+AZA were anemia (37%), neutropenia (26%), and thrombocytopenia (26%). Treatment-related febrile neutropenia occurred in only 1 (2%) patient. Only 1 patient discontinued due to an AE. 29 patients were evaluable for efficacy at time of data cut. 13/13 (100%) untreated MDS patients had an objective response with 7 patients (54%) achieving a CR, 5 (39%) with marrow CR (3/5 also had hematologic improvement (HI)), and 1 (7%) with HI alone. In AML, 11/16 (69%) had an objective response; 8/16 (50%) with CR or CRi, 2 (13%) with PR, 1 (6%) with MLFS, and 5 (31%) with stable disease. Time to response was more rapid (median 1.9 mos) than expected for AZA alone. For those with abnormal cytogenetics at baseline, 40% and 44% of MDS and AML patients achieved a cytogenetic CR, respectively. 4/8 (50%) AML patients with CR/CRi and 2/12 (17%) MDS patients with CR or marrow CR were MRD negative by flow cytometry. 11/16 (69%) AML patients became RBC transfusion independent and 11/13 (85%) MDS patients had hematologic improvement.

Given that CD47 is an LSC marker on leukemic cells, CD34+CD38- putative LSC frequency was measured by flow cytometry in the bone marrow in 5F9+AZA treated AML/MDS patients. In data available for analysis, LSCs were completely eliminated in 10/16 (63%) of AML/MDS patients who had a clinical response. Lastly, mutational analyses are ongoing to correlate subgroups with response. Interestingly, 7/8 (88%) evaluable TP53 mutant patients (5/6 AML patients [5 CR/CRi], 2/2 MDS [1 CR, 1 marrow CR]) achieved an objective response, highlighting efficacy in a poor prognosis and therapy-refractory population. No median duration response or overall survival has been reached for either MDS or AML patients with a median follow-up of 4.9 months (range 3.1 - 8.8 months) for MDS and 5.8 months (range 1.9 - 9.5 months) for AML.

Conclusions

Magrolimab+AZA is a novel immunotherapy regimen that blocks a key macrophage checkpoint. The combination therapy continues to be well tolerated with robust activity in MDS and AML patients with an ORR of 100% and 69%, respectively. High rates of putative LSC eradication suggest potential durable responses, with no median duration of response yet reached. Initial data indicate that 5F9+AZA may be particularly effective in TP53 mutant patients, a treatment-refractory subgroup. Expansion cohorts are ongoing (NCT03248479) with registrational studies in MDS being initiated. Additional patients, follow-up, and mutational characterization will be reported at time of presentation. Funded by Forty Seven and the California Institute for Regenerative Medicine.