Cancer cells reprogram themselves under therapy pressure, switching survival strategies.
For decades, acute myeloid leukemia has resisted medicine's best efforts not because treatments were weak, but because the disease itself was misread — assumed to be one thing when it was quietly four. Researchers at the German Cancer Research Center have now mapped the distinct subtypes of leukemia stem cells that drive relapse, revealing how each one exploits different molecular survival strategies to outlast therapy. The discovery reframes AML not as a single adversary but as a coalition of related threats, each requiring its own answer. In doing so, it opens a path from uniform treatment toward something more honest: medicine shaped to the precise architecture of each patient's disease.
- Nearly every AML patient treated with venetoclax eventually relapses — a pattern that has haunted oncologists for years without a clear biological explanation.
- The culprit turns out to be not one rogue cell population but four distinct leukemia stem cell subtypes, each occupying a different developmental stage and carrying different vulnerabilities.
- Under therapeutic pressure, resistant cells abandon their dependence on BCL-2 and quietly switch to a backup survival protein, BCL-xL, rendering venetoclax effectively invisible to them.
- In mouse models using actual patient leukemia cells, combining venetoclax with a BCL-xL inhibitor dismantled the resistance — the cancer's escape route was blocked from both directions.
- Biomarkers capable of identifying each subtype at the moment of diagnosis could soon allow oncologists to design personalized combination therapies before resistance even has a chance to form.
Acute myeloid leukemia moves fast and kills with regularity. Venetoclax offered real hope — it outperformed older chemotherapy and gave patients something to hold onto. Then, almost universally, the cancer returned. The drug stopped working, and the reasons remained stubbornly unclear.
Researchers at the German Cancer Research Center and the HI-STEM Stem Cell Institute spent years pursuing that question through tissue samples from more than 150 AML patients. Their focus was the leukemia stem cells — rare, self-renewing cells responsible for both the disease's persistence and its relapse. What they found dismantled a foundational assumption: there wasn't one type of leukemia stem cell. There were at least four, each resembling blood cells at different stages of normal development, and each responding to treatment in fundamentally different ways.
Venetoclax works by blocking BCL-2, a protein that keeps leukemia cells alive. But some stem cell subtypes had already shifted away from BCL-2 dependence before treatment began — and others reprogrammed themselves under therapeutic pressure, switching to a related survival protein called BCL-xL. The drug became useless. The cancer adapted.
The finding, though sobering, pointed toward a solution. If each subtype used different survival machinery, each could be targeted differently. In mouse models transplanted with patient leukemia cells, combining venetoclax with a BCL-xL inhibitor proved far more effective than standard treatment — the resistant cells had no remaining refuge.
The more practical breakthrough was the identification of molecular biomarkers capable of distinguishing one subtype from another at the time of diagnosis. In principle, a patient's leukemia could be profiled before treatment begins, revealing which subtypes are present and which drug combinations would be most effective against them. Study leader Andreas Trumpp has called for clinical trials to test this approach in real patients. The goal is a future where an AML diagnosis triggers not a uniform protocol, but a tailored plan — one built around the specific biological architecture of each person's disease.
Acute myeloid leukemia is a brutal disease. It strikes mostly older people, moves fast, and kills with regularity despite decades of medical progress. In recent years, a drug called venetoclax changed the equation somewhat—it worked better than the chemotherapy it was meant to replace, giving patients and doctors something that felt like real hope. Then the cancer came back. Nearly every patient relapsed. The drug stopped working. And no one fully understood why.
Researchers at the German Cancer Research Center and the HI-STEM Stem Cell Institute spent years chasing that question. They examined tissue samples from more than 150 AML patients, focusing on the cells that matter most: the leukemia stem cells, the rare ones that can renew themselves indefinitely and are responsible for both the disease's persistence and its return. What they found upended a simple assumption. There wasn't one type of leukemia stem cell. There were at least four distinct subtypes, each one different in ways that proved crucial to survival and resistance.
The differences lay in developmental stage. Each subtype resembled blood cells at different points in their normal maturation process, and that resemblance determined everything about how the cell would respond to treatment. Venetoclax works by blocking a protein called BCL-2, which keeps leukemia cells alive. When the drug shuts down BCL-2, the cells should die. But the researchers discovered that some leukemia stem cells don't rely heavily on BCL-2 at all. They've already switched to backup survival systems. Under the pressure of therapy, cancer cells can reprogram themselves entirely, abandoning their dependence on BCL-2 and shifting to a related protein called BCL-xL instead. Venetoclax becomes useless against them. The cancer adapts, and the patient relapses.
The finding was grim, but it opened a door. If different stem cell subtypes used different survival mechanisms, then different drugs could target each one. The researchers tested this hypothesis in mice transplanted with leukemia cells from actual patients. They combined venetoclax with a BCL-xL inhibitor—a drug that blocks the backup survival system the resistant cells were using. The combination worked far better than standard treatment alone. The resistant cells had nowhere to hide.
The real breakthrough, though, was simpler and more practical. The researchers identified specific biomarkers—molecular signatures—that could distinguish one stem cell subtype from another. This meant that at the moment of diagnosis, before treatment even began, doctors could theoretically look at a patient's leukemia cells and know which subtypes were present. They could predict which drugs would work and which ones the cancer would evade. Treatment would no longer be one-size-fits-all. It would be tailored to the biology of each individual patient's disease.
Andreas Trumpp, who led the study, framed the next step clearly: clinical trials. The science was solid. The mouse models showed promise. But the real test would come in actual patients, in real time, with real stakes. The hope is that within a few years, an AML diagnosis won't automatically mean the same protocol for everyone. It will mean a blood test, a biomarker panel, and a treatment plan designed specifically for the four subtypes lurking in that particular patient's bone marrow. It's a small shift in language—from treating a disease to treating a disease's specific architecture. But in oncology, that shift can mean the difference between relapse and remission.
Citas Notables
Treatment would thus become more individualized, targeted, and potentially successful.— Alexander Waclawiczek, first author of the study
Testing this new treatment strategy in a clinical trial with AML patients would now be the next step.— Andreas Trumpp, study leader
La Conversación del Hearth Otra perspectiva de la historia
So the problem isn't that venetoclax doesn't work. It's that it doesn't work on all the cells.
Exactly. It's like having a lock that opens four different doors, but the drug only has one key. Some cells use that lock. Others don't.
And the cells that don't use it—they just switch to something else?
Under therapy pressure, yes. They reprogram. It's almost like they're learning. They sense the drug is there, and they find an escape route.
That sounds like evolution happening in real time.
It is, in a way. Cancer cells are incredibly plastic. They can change their dependencies faster than we can kill them.
But you found that if you block both pathways at once—BCL-2 and BCL-xL—the cells can't escape.
Right. In the mouse models, combination therapy was significantly more effective. The cells ran out of options.
The biomarker piece—that's the part that actually changes how doctors practice, isn't it?
That's the part that makes it real. Right now, every AML patient gets roughly the same treatment. With biomarkers, you could know on day one which subtypes you're dealing with and which drugs will actually work for that specific patient.
When does that happen? When do we see this in the clinic?
That's the question everyone's asking. The science is there. Now it needs a clinical trial to prove it works in actual patients.