Impact of Mutational Landscape and CD47 Levels in the Magrolimab + Azacitidine Vs Placebo + Azacitidine Trial in Patients with Untreated Higher Risk Myelodysplastic Syndrome

Background

Magrolimab (Hu5F9-G4) is a monoclonal antibody that blocks CD47, an antiphagocytic signal largely overexpressed on tumor cells, but also ubiquitously expressed on non-neoplastic cells of myeloid origin. Following promising Ph 1 clinical data, the safety and efficacy of magrolimab in combination with azacitidine (aza) compared to aza+ placebo in patients with untreated higher risk MDS (HR-MDS) was investigated in a phase 3 study (ENHANCE, NCT04313881). The combination was expected to block the antiphagocytic signal generated by the interaction of CD47 on tumor cells and SIRP alpha on macrophages, while increasing pro-phagocytic signals, notably calreticulin, with the use of aza¹. The study was discontinued for futility at a prespecified interim analysis with a median overall survival (mOS) of 18.6 months in the placebo + aza arm and 15.9 months in the magrolimab + aza treated arm.

We report the impact of CD47 levels and the baseline mutational landscape of HR-MDS patients enrolled in ENHANCE.

Methods

Newly diagnosed adults with intermediate, high, or very high risk MDS per the revised International Prognostic Scoring System (IPSS-R) were enrolled to receive magrolimab (1 mg/kg on days 1 and 4, 15 mg on day 8, and 30 mg/kg on day 11 and 15, and then weekly for 5 weeks, followed by bi-weekly maintenance dosing). Aza was given at 75 mg/m² for days 1-7 or day 1-5, 8-9 in 28-day cycles.

To address CD47 expression, flow cytometry was performed on fresh, pre-treatment bone marrow aspirate (BMA) from placebo (N=71) or magrolimab treated (N=69) patients. CD47 expression was normalized to molecules of equivalent fluorescence and the median used to stratify by CD47 low or high expression. Multivariate cox regression analysis was performed to determine if CD47 was an independent prognostic factor, using data obtained from the clinical database.

Next generation sequencing (NGS) of 523 genes was performed centrally on DNA isolated from available pre-treatment BMA (N=465) using a custom pipeline to identify tumor mutations (TSO500HT, Illumina).

Results

In the biomarker analysis subset for CD47 expression, the mOS was 20.7 and 19.0 months for placebo + aza and magrolimab + aza, respectively. For placebo treated patients, CD47 expression had no influence on outcomes with a mOS of 20.7 months in the CD47 low group (N=30) and 21.1 months in the CD47 high group (N=41). The worst outcomes were observed in patients treated with magrolimab + aza that were CD47 low (N=40) with a mOS of 14.9 months. Magrolimab+aza treated patients who had high CD47 expression (N=29) had the best outcomes and mOS was not reached.

Magrolimab + aza treated patients with high CD47 expression (N=29) were predominantly classified as having very good/good/intermediate cytogenetic risk (69%, N=20) compared to those who were CD47 low (N=40) and where the majority of patients had poor/very poor cytogenetic risk (52.5%, N=21). In the placebo arm, cytogenetic risk was balanced across CD47 low and CD47 high patients with 50% (N=15) and 53.7% (N=22) having very good/good/intermediate and 43.3% (N=13) and 43.9% (N=18) with poor/very poor risk, respectively. Multivariate Cox regression analysis demonstrated that cytogenetic status was the key prognostic factor and not CD47 expression.

We will additionally report the NGS results to determine if there was a balance of poor prognostic and frequently mutated genes between the study arms and address if any patient sub-groups were differentially impacted by magrolimab treatment. Patients were initially enrolled on IPSS-R criteria and we will provide the molecular IPSS score.

Conclusions

One of the key questions for CD47 targeting agents has been whether expression levels impact outcomes. The data from this study suggests that higher CD47 expression in the bone marrow of HR-MDS patients may associate with OS benefit following magrolimab treatment. Given the imbalanced cytogenetic risk between CD47 high and CD47 low in the magrolimab+ aza arm, the observation warrants further investigation into the connection between cytogenetic risk, CD47 levels, and CD47 blockade.

Additional data with respect to the mutational landscape will be forthcoming, including central evaluation of TP53 status.

References

1.Chao MP, Takimoto CH, Feng DD, McKenna K, Gip P, Liu J, Volkmer JP, Weissman IL, Majeti R. Therapeutic Targeting of the Macrophage Immune Checkpoint CD47 in Myeloid Malignancies. Front Oncol. 2020 Jan 22;9:1380.

Johnson:Gilead Sciences Inc.: Current Employment. Zhang:Gilead Sciences Inc.: Current Employment.