Imagine a security guard who is supposed to throw intruders out of a building. Now imagine the intruders give him a gift that changes his behavior, and he starts protecting them instead. This is what we found in the bone marrow of children with acute myeloid leukemia: macrophages, the immune system’s guards, receive a gift of mitochondria from leukemia cells and stop attacking them.
Acute myeloid leukemia (AML) is an aggressive blood cancer and the second most common leukemia in children. Although chemotherapy has improved outcomes, around 20-30% of children with AML still do not survive, because of treatment toxicity or relapse. Better and less toxic therapies are urgently needed.
Immunotherapies, which harness the patient’s own immune system to attack cancer, have transformed the treatment of several other tumors, but progress for children with AML has been slow. One reason is the bone marrow microenvironment, the niche inside our bones where blood cells are made. Many cell types live there, and leukemia cells can exploit them to grow and to resist treatment.
Among the cells in this niche are macrophages, immune cells whose name means “big eaters.” Their job is to engulf pathogens and abnormal cells, including cancer cells. In other cancers, macrophages are known to switch to a state in which they support the tumor and weaken the immune response, yet in children with AML they have received little attention.
In our recent research, published in Leukemia, we investigated how macrophages contribute to leukemia progression instead of fighting the tumor.
The Role of Macrophages in the children with AML
Macrophages are often described as two broad types. M1-like macrophages are pro-inflammatory and fight tumors. M2-like macrophages are anti-inflammatory and can protect and support cancer cells. This is a simplification, but it is a useful starting point.
By analyzing bone marrow data from more than 300 children with AML, we found that M2-like macrophages dominate the bone marrow of AML patients, while children without leukemia have relatively more M1-like macrophages. In the lab, we also showed that M2-like macrophages support the growth of AML cells.
We then asked whether the leukemia cells themselves are responsible for the dominance of M2-like macrophages. When we grew AML cells together with M1 macrophages, the macrophages initially held back leukemia growth, but this effect was lost within days. After six days of contact, the macrophages had lost their inflammatory features, switched off key inflammatory pathways such as NF-κB, and taken on an M2-like profile. They also became much worse at “eating”: their ability to engulf leukemia cells dropped sharply.
How does it happen?
How do leukemia cells achieve this? When we separated the two cell types with a thin membrane that lets molecules through but prevents contact, macrophages still lost some of their pro-inflammatory features, but they largely did not acquire anti-inflammatory ones. They also kept their ability to hold back leukemia growth. Direct contact was needed for the full switch.
Under the microscope we saw the formation of tunneling nanotubes, thin membrane bridges that connect AML cells to macrophages. Through these bridges, the leukemia cells pass over their mitochondria, the small structures that produce most of a cell’s energy. The mitochondria were healthy and functional, and the transfer was largely one-way, from leukemia to macrophage.
When we blocked nanotube formation, by drug or by gentle mechanical disruption, or interfered with mitochondrial transport, the macrophages largely failed to acquire their M2-like features. So, the picture seems to have two parts. Soluble signals from leukemia cells appear to switch off the macrophages’ inflammatory features, and mitochondrial transfer through nanotubes is needed to complete the switch to a tumor-supporting state.
Why this matters for patients
Reprogrammed macrophages did more than change their features: they also changed how leukemia responds to treatment.
- Chemotherapy: Macrophages reprogrammed by AML cells can protect the AML cells from the effects of daunorubicin and cytarabine, two standard chemotherapy agents;
- T cells: M1 macrophages help activate T cells and boost their killing of leukemia cells, but macrophages reprogrammed by AML cells lost this ability.
- Bispecific antibodies: Blinatumomab is a drug that links T cells to cancer cells. M1 macrophages made it work better in our experiments, but macrophages that had been exposed to AML cells completely lost this supportive effect, and leukemia cells survived more.
We also looked at adult patients with relapsed or refractory AML treated with flotetuzumab, a similar bispecific antibody. Patients who responded had significantly fewer M2-like macrophages in their bone marrow before treatment than those who did not. This is a small, retrospective analysis, but it suggests macrophage status could one day help predict who benefits from T cell-based therapies.
Looking ahead
Our findings suggest that immunotherapies might work better if we stop leukemia cells from reprogramming macrophages, for example by blocking nanotube formation or mitochondrial transport. Drugs that disrupt nanotubes broadly would likely be too toxic, so more selective strategies are needed.
Important questions remain. Do the transferred mitochondria carry features that actively instruct macrophages? How much do other communication routes contribute? And how well do lab models capture a real bone marrow, where macrophages are highly plastic and dynamically shaped by signals from the tumor microenvironment?
For now, our work shows that macrophages are not bystanders in childhood AML. They are active participants that leukemia can recruit, and understanding this relationship may help us design better therapies for children.
Read the full paper: Perzolli et al., Leukemia (2026). https://www.nature.com/articles/s41375-026-03164-1