Maybe we're on to something that could help millions
In the long struggle against Alzheimer's disease — a condition that slowly unmakes the self — researchers at UC San Francisco have found an unexpected ally in the medicine cabinet of oncology. By asking not what new drug might fight the disease but which existing ones already speak its molecular language, a team of computational biologists identified two cancer medications that, together, reversed hallmarks of Alzheimer's in mice. It is a reminder that knowledge, like medicine, sometimes heals best when repurposed.
- With over 55 million people living with Alzheimer's globally and that number expected to double within 25 years, the urgency for a breakthrough has never been greater.
- Two cancer drugs — letrozole and irinotecan — disrupted expectations by reducing toxic tau protein clumps and measurably restoring learning and memory in mouse models of the disease.
- The drugs work through different cellular mechanisms, targeting neurons and glial cells respectively, allowing the combination to attack Alzheimer's across multiple biological pathways at once.
- Because both drugs already carry FDA approval, the usual decade-long runway to human trials could be significantly shortened — a rare structural advantage in a field littered with failed attempts.
- Researchers caution that side effects tolerable in cancer patients may require fresh scrutiny for Alzheimer's use, and that mouse results must still survive the test of human clinical trials.
A team at UC San Francisco chose an unconventional starting point: instead of designing a new drug, they asked which existing ones might already counteract Alzheimer's at the molecular level. Using the Connectivity Map — a database that tracks how drugs alter gene expression — they searched for medications capable of reversing the specific genetic changes Alzheimer's inflicts on the brain. Cross-referencing their findings with patient records from cancer treatment added a second, independent layer of evidence. Two drugs rose to the top: letrozole, used for breast cancer, and irinotecan, used against colon and lung cancers.
When the combination was tested in mouse models of Alzheimer's, the results were striking. Harmful tau protein clumps — a defining feature of the disease — were significantly reduced, and the mice showed measurable improvements in learning and memory. The drugs appeared to complement each other, with letrozole acting on neurons and irinotecan on glial cells, together addressing the disease across multiple cell types simultaneously.
This matters because Alzheimer's is not a single-gene problem. It involves cascading changes across many genes and proteins, which has long confounded drug development strategies built around one target at a time. The computational approach allowed researchers to think in terms of pathways rather than single molecules — and to find drugs already shaped to fit them.
Lead researcher Marina Sirota noted the power of convergence: when independent data sources — gene expression profiles and real-world clinical records — point to the same drugs and the same pathways, the signal becomes harder to dismiss. Still, the work remains in its early stages. Both medications carry side effects that were acceptable in cancer treatment but will need careful re-evaluation for long-term use in Alzheimer's patients. Human trials are the essential next step.
The broader vision is ambitious: a computational framework that could one day be tailored to individual patients based on how their own gene expression has shifted — a more precise medicine for a disease that has, so far, resisted every simple answer.
A team of researchers at UC San Francisco took an unconventional path to Alzheimer's treatment: they looked backward at drugs already in use. What they found was that two cancer medications—letrozole, typically prescribed for breast cancer, and irinotecan, used against colon and lung cancers—appeared to reverse some of the brain damage caused by Alzheimer's disease when tested together in mice.
The approach began with a computational question. The scientists examined how Alzheimer's alters gene expression in the brain, then searched the Connectivity Map, a medical database, for drugs that could reverse those changes. They cross-referenced this with patient records from people who had taken these same medications for cancer treatment, looking for signals that the drugs might have lowered their Alzheimer's risk. The computational work narrowed the field to letrozole and irinotecan as the most promising candidates.
When tested in mouse models of Alzheimer's disease, the combination worked. The harmful tau protein clumps that accumulate in Alzheimer's brains were significantly reduced. The mice also showed measurable improvements in learning and memory tasks—cognitive abilities that Alzheimer's typically erodes. The two drugs appeared to work through different mechanisms: letrozole targeted neurons while irinotecan worked in glial cells, allowing the combination to address the disease across multiple cell types.
The significance of this finding lies partly in what it sidesteps. Alzheimer's is not a single-gene disease. It involves numerous alterations across many genes and proteins, all working together to damage the brain. Traditional drug development has struggled with this complexity, typically producing one medication to target one gene or protein. The computational approach allowed researchers to think differently—to identify drugs that could address multiple pathways simultaneously.
Because both medications are already FDA-approved for cancer treatment, the path to human trials could move faster than developing a drug from scratch. Marina Sirota, the computational biologist who led the work, emphasized the convergence of evidence: "If completely independent data sources, such as single-cell expression data and clinical records, guide us to the same pathways and the same drugs, and then resolve Alzheimer's in a genetic model, then maybe we're on to something."
But significant hurdles remain. The drugs have only been tested directly in mice. Both letrozole and irinotecan carry side effects that were acceptable when treating cancer but would need careful evaluation if the drugs are repurposed for Alzheimer's. The researchers acknowledge that clinical trials in people with Alzheimer's represent the necessary next step.
The stakes are substantial. More than 55 million people currently have Alzheimer's disease globally. Projections suggest that number will more than double within the next 25 years as populations age. A treatment that could reverse cognitive decline, or even prevent it, would reshape how the disease is managed. The researchers suggest this computational approach could eventually enable personalized treatments tailored to how gene expression has shifted in individual patients—a more precise medicine for a disease that has resisted simple solutions.
Notable Quotes
Alzheimer's disease comes with complex changes to the brain, which has made it tough to study and treat, but our computational tools opened up the possibility of tackling the complexity directly.— Marina Sirota, computational biologist, UC San Francisco
Alzheimer's is likely the result of numerous alterations in many genes and proteins that, together, disrupt brain health. This makes it very challenging for drug development—which traditionally produces one drug for a single gene or protein that drives disease.— Yadong Huang, neuroscientist, UC San Francisco and Gladstone Institutes
The Hearth Conversation Another angle on the story
Why did researchers think to look at cancer drugs in the first place?
They didn't start with cancer drugs. They started with the question of what happens to genes in an Alzheimer's brain, then searched databases for any existing drugs that could reverse those changes. Cancer medications happened to be the answer.
So this is drug repurposing—taking something already approved and using it for a different disease?
Exactly. And it's smart because these drugs already have safety data. The FDA has already vetted them. That means if the mouse results hold up in humans, you could potentially move to clinical trials much faster than waiting for a brand-new drug to be developed and approved.
The mice improved in memory and learning. How confident should we be that the same thing happens in people?
That's the honest question. Mouse models are useful for understanding mechanisms, but they're not people. The brain is vastly more complex. What works in a controlled lab setting might not translate. That's why the next step—human trials—is so critical.
Both drugs come with side effects from their cancer use. Doesn't that complicate things?
It does. When you're treating cancer, the benefit of stopping a life-threatening disease outweighs significant side effects. With Alzheimer's, the calculus changes. You'd need to prove the benefit is worth whatever toxicity comes with the drug. That's a real barrier.
Why combine two drugs instead of using one?
Because Alzheimer's hits multiple cell types in different ways. One drug worked better on neurons, the other on glial cells. Together, they addressed more of the damage. It's treating the disease as the complex problem it actually is, rather than hunting for a single magic bullet.
What happens next?
Human trials. If those work, then the real work begins—figuring out which patients benefit most, how to minimize side effects, and whether this computational approach can be refined to find even better drug combinations. This is the beginning, not the finish line.