Effects of Magrolimab Therapy in Conjunction with Conventional Chemotherapeutics in Pediatric Acute Myeloid Leukemia Patient Derived Xenograft Models

Introduction: Magrolimab (Magro) is a humanized naked anti-CD47 monoclonal antibody that blocks the SIRPα CD47 interaction, allowing macrophages to target and destroy tumor cells. The drug has undergone extensive testing and has been evaluated in phase III studies (NCT05079230, NCT04778397). In this project, Magro was tested as a single agent and in combination with backbone chemotherapy drugs, cytarabine (AraC) or azacitidine (Aza), in three different pediatric AML (pAML) patient derived xenograft (PDX) models – AML006 (KMT2A::MLLT1), AML010 (+10, WT1), AML013 (KMT2A::MLLT4) - in order to evaluate its pre-clinical efficacy in vivo.

Methods: For each PDX model tested, 60-65 NSGS mice were tail vein injected with 2x10⁵ isolated mononuclear cells and the cells’ engraftment was tracked via surface flow of peripheral blood (PB) using hCD45-APC-Cy7, hCD33-AF700, and hCD47-BV421. Once 50% of the mice met the engraftment threshold of 1% human cells by hCD45+ in PB, we began treatment with six treatment groups: hulgG4 (VC, vehicle control for Magro), Magro, AraC + VC, Aza + VC, AraC + Magro, and Aza + Magro. Magro and VC were given in a dose escalation manner: 100 ug/kg/dose IP on days 1, 2, 3; 0.5 mg/kg/dose on day 4; 1mg/kg/dose on days 5, 6; and 10mg/kg/dose on days 7, 10, 15, 22. AraC was given 50mg/kg/day IP on days 1-5. Aza was given 2.5 mg/kg/day IP on days 1-7. After treatment was complete, all mice were followed for survival and PB disease burden every two weeks. Mice that met humane euthanasia endpoints had tissues harvested - PB, spleen, bone marrow (BM) - to measure disease burden via surface flow.

All AML010 mice were euthanized for clinical illness by day 43. AML006 mice that had not yet met clinical endpoints were euthanized at 70-72 days (7/10 Magro, 10/10 AraC+Magro, 4/10 Aza + Magro). For AML013, surviving mice were euthanized at 152-153 days (5/10 Magro, 6/10 AraC+Magro, 1/10 Aza, 3/11 Aza + Magro).

Results: All survival data are given in medians and disease burden data (hCD45+) are given in means. Two of 3 models had improved overall survival (OS) for VC vs. Magro: AML006, 23 vs undefined days (p <0.0001); AML013, 41 vs 129 days (p=0.003); and AML010, 14 vs 15 days (p=ns). Two of 3 models had decreased disease burden in the BM at the time of euthanasia: AML006, 78.61% vs 31.11% (p =0.009); AML013, 87.27% vs 31.84% (p=0.002); and AML010, 95.11% vs 90.09% (p =ns). Therefore, Magro alone significantly improved survival and decreased BM disease burden in AML006 and AML013, but not for AML010.

For AraC only and AraC + Magro groups, all 3 models had statistically improved OS: AML006, 37 vs undefined days (p <0.0001); AML013, 43 vs undefined days (p=0.001); AML010, 25 vs 28 days (p=0.002). Two of 3 models had decreased disease burden: AML006, 97.49% vs 2.56% (p <0.0001); AML013, 77.29% vs 39.63% (p=0.048); and AML010, 49.21% vs 39.86% (p=ns). Therefore, Magro + AraC significantly improved survival and decreased disease burden in AML006 and AML013, and less so in AML010, in comparison to AraC alone.

For Aza only and Aza + Magro groups, two of 3 models had improved OS: AML006, 12 vs 14 days (p=ns); AML013, 58 vs 104 (p=0.047); AML010, 37 vs 39 days (p=0.017). Only one model had decreased disease burden: AML010, 88.90% vs 24.41% (p=0.001); AML006, 85.04% vs 50.27% (p=ns); and AML013, 78.39% vs 60.18% (p=ns). Therefore, Magro + Aza resulted in decreased disease burden in AML010 and improved survival in AML013 in comparison to Aza alone.

Conclusion: Magrolimab, when given in a dose escalation manner, is well tolerated in PDX models, with evidence of efficacy in 2 of 3 models tested - AML013 and AML006 - as a single agent and in combination with chemotherapeutics. While Magro in combination with Aza or AraC resulted in statistically significant improved median survival for AML010, the difference was not clinically meaningful. Interestingly, the two models that demonstrated improvement in survival with magrolimab both harbored KMT2A rearrangements, suggesting a subset of patients that may be more responsive to the effects of CD47 blockade. As this drug is still being evaluated for use in other malignancies, future studies may focus on investigating the possible importance of biomarker-based selection of patients.