Mutational Landscape and Depth of Response in Patients with Acute Myeloid Leukemia (AML) Treated with Magrolimab in Combination with Venetoclax and Azacitidine Compared to Placebo

Background: AML is a genetically heterogeneous disease. A key determinant of therapy response and overall survival (OS) is the underlying molecular architecture of a patient’s disease. ENHANCE-3 tested the innate checkpoint inhibitor magrolimab (anti-CD47) in combination with azacitidine and venetoclax compared to placebo control for safety and efficacy in newly diagnosed AML patients ineligible for intensive therapy (NCT5079230). The prespecified futility boundary for OS was crossed and the trial was terminated early. We report the baseline mutational landscape, flow cytometry (FCM) and next generation sequencing (NGS) measurable residual disease (MRD) data and its correlation with clinical outcome.

Methods: Clinical data was obtained from the trial database. DNA from pre-treatment bone marrow or peripheral blood mononuclear cells (PBMNC or BMMNC) from patients (magrolimab N=120, placebo N=124) was subject to NGS using a harmonized 37 gene panel at Oxford or HOVON (sensitivity: variant allele frequency >1%). FLT3 and NPM1 fragment analysis was performed using PCR followed by capillary gel electrophoresis. Central NGS MRD (Oxford) was performed on BMMNC or PBMNC using a 97 gene NGS panel with a sensitivity cut off of >0.1%. Central FCM MRD analysis on BMMNC had a cut off of 0.1% (LabCorp).

Results: The ITT and biomarker analysis cohorts had similar demographics and median OS (mOS). The median follow up time for the biomarker cohort was 7.4 months. The mOS for the biomarker cohort placebo arm was not reached and 10.7 months in the magrolimab arm compared to 14.1 and 10.7 months in the ITT set. TET2 (28.2%, 28.3%), DNMT3A (21.8%, 25.0%), and TP53 (30.6%, 22.5%) were the most frequent mutations in the placebo and magrolimab arms, respectively.

We stratified patients using the molecular prognostic risk score (mPRS) for venetoclax and azacitidine treated patients (lower benefit, (TP53 mutated); intermediate benefit, (FLT3-ITD and/or N/KRas mutated) and higher benefit (all other mutations)) to determine if the placebo arm behaved as expected and if magrolimab conferred benefit in any of the risk groups. In the placebo and magrolimab treated arms, mPRS stratified patients into lower (N=38, 30.6%, N=27, 22.5%), intermediate (N=26, 21.0%, N=22, 18.3%) and higher (N=60, 48.4%, N=71, 59.2%) benefit. Patients with lower benefit had a mOS of 6.9 months in the placebo and 7.4 months in the magrolimab arm, consistent with the prognosis of patients with mutated TP53. For intermediate benefit patients, the magrolimab arm had a mOS of 9.8 months and was not reached in the placebo arm. In patients with higher benefit, mOS was not reached in either arm.

Given the prevalence of TET2 mutations in the biomarker cohort and prior reports that patients with TET2 mutations may have a better response to azacitidine, we assessed mOS in TET2 mutated patients in the placebo (N= 35) and magrolimab (N=34) arms. The mOS for the patients in the placebo arm was not reached, in marked contrast to 7.0 months in the magrolimab arm (HR=2.02 [0.94-4.34]). There was a difference in the proportion of deaths in TET2 mutant patients in the magrolimab (N=17, 50.0%) compared to placebo arm (N=11, 31.4%). Deaths related to pneumonia or lung disease (N=3 vs N=1) and disease progression (N=10 vs N=6) in were increased in the magrolimab arm compared to placebo.

In contrast, TET2 wild type patients did not reach mOS in the placebo arm and was 10.7 months in the magrolimab arm. The death rate from any cause (N = 30, 33.7% vs N=29, 33.7%) in the placebo and magrolimab treated arms was balanced in TET2 wild type patients. Further analysis of the biomarker dataset is ongoing to identify any additional sub-groups that show differential benefit. The correlation with FCM and NGS MRD, mutational state and clinical outcome will be presented.

Conclusions: The biomarker cohort is clinically representative of the ENHANCE-3 ITT population. TET2, DNMT3A, and TP53 were enriched and ASXL1, NPM1 and FLT3 mutations were underrepresented compared to expected distributions in an AML population ineligible for intensive treatment. Patients with a TET2 mutation likely fare worse with magrolimab treatment, including an increased rate of respiratory tract infection. It is noteworthy that TET2m mature myeloid cells have pro-inflammatory features raising the hypothesis that magrolimab and TET2 mutant myeloid cells may synergise to exacerbate a maladapted innate response to infection.