# CD47

Source: https://onco.cc/targets/cd47/  
OnCo record `cd47` (Target). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

CD47 is the 'don't eat me' signal: it binds SIRP-alpha on macrophages to stop them engulfing the cell, and over 90% of AML blasts and large B-cell lymphoma cells display it. Blocking it should let macrophages eat tumour cells, but red cells carry CD47 too, so anaemia is built in, and the lead antibody magrolimab was dropped after failed trials.

## Summary

CD47 is the 'don't eat me' signal: it binds SIRPα on macrophages to inhibit phagocytosis, and tumour cells display it broadly, with more than 90 percent of AML blasts and DLBCL cells carrying it. Blocking CD47 should let macrophages engulf tumour cells, particularly when combined with an opsonising antibody such as rituximab or with azacitidine. Because CD47 is also ubiquitous on red cells, anaemia is the built-in on-target toxicity. Magrolimab, the lead antibody, was discontinued after failed trials in MDS and AML (ENHANCE), with excess deaths. The macrophage checkpoint concept persists through SIRPα-targeted agents and bispecifics designed to spare red cells and lower haematological toxicity. For a newcomer: it looked like a promising immune brake on macrophages, but the first drug against it failed.

## Fields

- Kind: Target
- Last checked: 2026-09-04
- Tags: checkpoint; innate; failed-so-far
- Symbol: CD47
- Class: checkpoint
- Biology: Binds SIRPα on macrophages to inhibit phagocytosis; ubiquitous on red cells, causing anaemia.
- Where found: Broadly on tumour cells

## Notes

- Lymphoma: Lymphoma cells, as on most tumour cells; it is a signal that says do not eat me rather than a lineage marker. Also on healthy cells: Nearly everything, and red cells most of all. CD47 density on an erythrocyte is how a macrophage decides not to clear it, which is the whole problem with the class. What the medicine does: Blocking the CD47-SIRPa interaction releases the macrophage brake so that an opsonising antibody such as rituximab can be acted on. In a phase 1b study of 22 patients with relapsed or refractory non-Hodgkin lymphoma, 95% of whom had rituximab-refractory disease, Hu5F9-G4 with rituximab gave an objective response in 50% (Advani 2018). How tumours lose it: Not applicable: there is no approved CD47 agent in lymphoma and no established acquired-resistance literature. What that costs the patient: On-target anaemia, because the drug removes the same signal from red cells. The dosing strategy that made the class tolerable is a low priming dose that clears the oldest red cells first, followed by higher maintenance doses.

## Sources

- Wikipedia: https://en.wikipedia.org/wiki/CD47
- Wikipedia: https://en.wikipedia.org/wiki/CD47
- Advani et al., N Engl J Med 2018: CD47 blockade with Hu5F9-G4 and rituximab in 22 patients with non-Hodgkin lymphoma: https://doi.org/10.1056/NEJMoa1807315

## Connected records

- cancers: [Acute myeloid leukaemia](https://onco.cc/cancers/aml/), [Diffuse large B-cell lymphoma](https://onco.cc/cancers/dlbcl/), [Higher-risk myelodysplastic syndromes](https://onco.cc/cancers/mds-higher-risk/), [Myelodysplastic syndromes / neoplasms (MDS)](https://onco.cc/cancers/mds/), [Non-Hodgkin lymphoma (all types)](https://onco.cc/cancers/non-hodgkin-lymphoma/), [Secondary and therapy-related acute myeloid leukaemia](https://onco.cc/cancers/aml-secondary/)
- technologies: [CD47 and SIRP-alpha blockade](https://onco.cc/technologies/cd47-blockade/)
- drugs: [AK117](https://onco.cc/drugs/ak117/), [Magrolimab](https://onco.cc/drugs/magrolimab/), [Peluntamig](https://onco.cc/drugs/peluntamig/), [Spevatamig](https://onco.cc/drugs/spevatamig/)
- pathways: [CD47 / SIRPα (the 'don't eat me' signal)](https://onco.cc/pathways/cd47-sirpa/), [MYC](https://onco.cc/pathways/myc/), [Myeloid suppression: TAMs, MDSCs & don't-eat-me signals](https://onco.cc/pathways/myeloid-suppression-axis/), [Tumour microenvironment (TME)](https://onco.cc/pathways/tumor-microenvironment/)
- terms: [Checkpoint (two meanings)](https://onco.cc/terms/checkpoint/), [Macrophage](https://onco.cc/terms/macrophage/), [Tumour-associated macrophages (TAMs)](https://onco.cc/terms/tumor-associated-macrophages/), [What it costs to aim at a lineage antigen](https://onco.cc/terms/lymphoma-bio-lineage-antigen-cost/)
- institutions: [Stanford Health Care / Stanford Cancer Institute](https://onco.cc/institutions/stanford/)
- people: [Irving L. Weissman](https://onco.cc/people/irving-weissman/)
- bottlenecks: [Cold tumours and the immunosuppressive microenvironment](https://onco.cc/bottlenecks/b-tme-immunosuppression/)
- ideas: [Bispecific antibodies that engage macrophages instead of T cells](https://onco.cc/ideas/idea-bio2-myeloid-engager-bispecific/), [Can MYC be drugged directly, and will patients tolerate it?](https://onco.cc/ideas/idea-drugging-myc/), [Engineered bacteria that live in tumours and manufacture drugs there](https://onco.cc/ideas/idea-bio2-engineered-bacteria-payloads/), [Inhaled immune therapy to make the lung hostile to arriving tumour cells](https://onco.cc/ideas/idea-bio2-inhaled-lung-niche-immunotherapy/), [Make every cold tumour hot: a coordinated programme to reprogramme immune-excluded tumours](https://onco.cc/ideas/idea-moon-cold-to-hot-programme/)
- companies: [ALX Oncology](https://onco.cc/companies/alx-oncology/)
- targets: [CD24](https://onco.cc/targets/cd24/)

---
JSON: https://onco.cc/api/v1/entities/cd47.json