ENHANCE: magrolimab added toxicity but no benefit in higher-risk MDS

Magrolimab was built on an appealing idea. Cancer cells display a protein, CD47, that tells passing immune cells "don't eat me"; an antibody that blocks CD47 should lift that shield and let macrophages clear the cancer. In myelodysplastic syndromes — bone-marrow disorders that often progress to leukaemia — early results looked promising enough to launch a large phase 3 trial. ENHANCE is that trial, and it did not work [s1].

What the trial tested

ENHANCE enrolled treatment-naïve patients with higher-risk myelodysplastic syndromes, classified as intermediate- to very-high-risk on the Revised International Prognostic Scoring System [s1]. Patients were randomly assigned to magrolimab plus azacitidine or to matched placebo plus azacitidine [s1].

Azacitidine is the standard backbone here, so both arms received active treatment; the question was whether adding the CD47 antibody improved on it [s1]. The magrolimab schedule was a graded ramp — 1 mg/kg on days 1 and 4, 15 mg/kg on day 8, then 30 mg/kg on days 11 and 15 and weekly for five doses, followed by 30 mg/kg maintenance every two weeks — while azacitidine was given at 75 mg/m2 on a standard 28-day cycle [s1]. The graded start is a deliberate feature of CD47 blockade: because red blood cells also carry CD47, an abrupt full dose can trigger anaemia, so the schedule builds up slowly.

The trial had dual primary endpoints: complete remission rate and overall survival [s1]. Requiring both a short-term response measure and a hard survival outcome is a demanding design — a drug can shrink disease without helping people live longer, and dual endpoints are meant to catch that gap. Higher-risk myelodysplastic syndromes, the population enrolled here, are the ones most likely to progress to acute leukaemia and the ones where a genuinely effective new agent would matter most.

What happened

At the final analysis, 539 patients had been randomised — 268 to magrolimab plus azacitidine and 271 to placebo plus azacitidine [s1]. Baseline characteristics were generally well balanced [s1].

Neither primary endpoint was met. The complete remission rate was 21.3% with magrolimab plus azacitidine versus 23.6% with placebo plus azacitidine — an odds ratio of 0.876 (95% confidence interval, 0.585 to 1.312), P = .5218 [s1]. Median overall survival was 15.9 months with magrolimab plus azacitidine versus 18.6 months with placebo plus azacitidine, a hazard ratio of 1.203 (95% CI, 0.947 to 1.528), P = .1299 [s1].

Both numbers point the wrong way for the experimental drug. Remissions were slightly less common, and survival was numerically shorter — not by a statistically significant margin, but with no hint of the benefit the mechanism predicted.

The safety data are where the result turns from disappointing to harmful. The magrolimab arm had a higher incidence of grade 3 or higher adverse events (92.8% versus 79.2%), more adverse-event-related study-drug discontinuations (24.0% versus 12.1%), more serious adverse events (71.9% versus 51.5%) and more fatal adverse events (15.2% versus 9.8%) than placebo plus azacitidine [s1]. Adding the antibody made treatment more toxic across every category the trial measured, including death.

The authors' conclusion is blunt: ENHANCE did not meet its primary endpoints of complete remission rate and overall survival, and showed more frequent severe adverse events in the magrolimab arm [s1].

The funding and the ending

ENHANCE was run for magrolimab's developer and registered as NCT04313881, where its status is recorded as terminated [s1][s2]. That ending is the substance of the story. A drug that looked promising in early, uncontrolled data was tested against the proper comparator in a randomised trial, failed on efficacy, harmed patients on safety, and the programme was stopped. The value of the controlled trial is precisely that it caught what the early signal missed.

It is also a caution about how much weight to put on mechanism and single-arm results. CD47 blockade is biologically elegant, and the phase 1 data were encouraging; neither survived contact with a randomised comparison.

Limits

ENHANCE tested one CD47 antibody in one disease — higher-risk myelodysplastic syndromes — on an azacitidine backbone, so it does not by itself close the door on CD47 as a target in other cancers or combinations [s1]. But within its own question the result is decisive on both efficacy and safety, and the trial's termination reflects that [s2].

What to watch

Whether other CD47-directed agents and combinations report, and whether any separates benefit from the toxicity ENHANCE recorded. For magrolimab in higher-risk myelodysplastic syndromes, the trial's answer is final: it added risk without adding benefit [s1].

This article describes trial results. It is not medical advice, and nothing here should be used to start, stop or change any treatment.

Sources

• Magrolimab Plus Azacitidine Versus Placebo Plus Azacitidine in Untreated Higher-Risk MDS: The Phase III ENHANCE Study, Journal of Clinical Oncology, 13 July 2026
• Magrolimab + Azacitidine Versus Azacitidine + Placebo in Untreated MDS (NCT04313881), ClinicalTrials.gov