Every person who lives long enough will develop cancer mutations in their blood. By age 70, between 10% and 30% of us carry detectable clones of mutant white blood cells, each descended from a single stem cell that picked up a mutation in a cancer-driver gene and started outcompeting its neighbors. The condition has a clinical name, clonal hematopoiesis of indeterminate potential (CHIP). For most people, it never causes symptoms. But CHIP raises the risk for leukemia and can accelerate atherosclerosis.
Recently, two research groups asked what happens when these mutant blood cells reach the brain. Despite both groups studying the same mutations in the same genes, finding them in the same cell type in the same organ, and agreeing on the basic mechanism by which they get there, they arrived at starkly different answers.
CHIP Carriers Show Lower Alzheimer’s Risk
One group, led by Siddhartha Jaiswal, MD, PhD, a hematologist at Stanford University School of Medicine, Stanford, California, published a study in 2023 showing people with CHIP were 36% less likely to develop Alzheimer’s dementia. Mendelian randomization supported a causal link, meaning that the genetic variants that make a person more likely to develop CHIP also made them less likely to develop Alzheimer’s disease, independent of other risk factors — the closest thing to a randomized trial that epidemiology can offer.
And when Jaiswal’s team examined the postmortem brains of eight people in their 80s who were known CHIP carriers, most of whom had no dementia at death, they found the mutations had colonized up to 95% of the microglia, the brain’s resident immune cells. “The cells need to get into the brain to exert their protective effect,” Jaiswal said. “It could relate to perhaps better clearance of amyloid plaques or tau tangles.”
- CHIP common by age 70; linked to leukemia + atherosclerosis risk.
- CHIP associated with ↓ Alzheimer dementia risk in one cohort (36%).
- Postmortem AD brains showed CHIP mutations in microglia, up to 95% in some cases.
- Deeper sequencing found more small CHIP clones in AD brains; clone-size threshold changed risk signal.
- TET2 ≠ DNMT3A: TET2 linked to ↓ late-onset AD (47%); DNMT3A showed no benefit.
In every other disease context where CHIP has been studied, such as atherosclerosis, stroke, chronic liver disease, and arthritis, CHIP mutations worsen outcomes. Jaiswal’s finding pointed in the opposite direction. “Alzheimer’s is really kind of the one exception that we found where the result is actually a better outcome,” he said.
Deeper Sequencing Reveals a Different Picture
Meanwhile, in June this year, Christopher A. Walsh, MD, PhD, of Boston Children’s Hospital and the Howard Hughes Medical Institute in Boston and his group published a study that complicated the picture. Where Jaiswal had started from a population-level question — does CHIP protect? — and used brain tissue secondarily to confirm the mutations were present, Walsh started from the diseased tissue itself. His team examined postmortem brain samples from 190 people who died with Alzheimer’s disease and 121 age-matched control individuals, all from a cohort with neuropathologic diagnoses. And they sequenced far deeper than Jaiswal had, using a technique sensitive enough to detect mutations shared by fewer than 1 in 1000 cells.
Alzheimer’s disease brains harbored significantly more CHIP mutations than control individuals, and the mutations were concentrated in microglia.
“The microglial cells are the cells that carry these mutations, and the ones that we studied are shared in the blood, suggesting they probably originate from blood cells,” Walsh said.
The mutations were not arising independently in the brain. As we age, immune cells from the bone marrow that carry CHIP mutations cross the blood-brain barrier and settle into the brain, where they become indistinguishable from the microglia that have been there since birth. They look like microglia, they behave like microglia, but they arrived carrying mutations they acquired in the blood.
Those mutant microglia had shifted into what researchers call the disease-associated microglia (DAM) state, a transcriptional profile characterized by inflammation and proliferation that has been linked to neurodegeneration. “The vast majority of the microglia that are in the DAM state are carrying these CHIP mutations,” Walsh said.
When his team engineered the same mutations into microglia grown from stem cells, the cells shifted into that inflammatory profile on their own. “I was amazed at how clean the results were,” Walsh said.
Small Clones, Big Questions
What struck Walsh most was the nature of the clones involved. Unlike in cancer, where a single large clone transforms and dominates, a brain with Alzheimer’s disease contains many small, competing clones expanding simultaneously. Jaiswal’s methods could only pick up clones that had grown large enough to make up at least 5%-8% of a blood sample. Walsh could see the small ones.
“It’s not big clones, it’s little clones. And the little clones are responding to an inflammatory environment,” said Walsh. “If you include the smallest clones, then actually CHIP becomes a risk factor, not a protective factor, although at a nonsignificant p value. It’s no longer protective at all.”
In Walsh’s paper’s supplemental figures, when his team analyzed the data across different clone-size thresholds, large clones in CHIP genes trended protective, reproducing Jaiswal’s result. The risk signal came from the small ones.
A new study published in Nature in July 2026 by Jaiswal’s group established that the migration of blood cells into the brain isn’t confined to people carrying CHIP mutations. By using the random mutations that accumulate naturally in blood stem cells as a kind of genetic fingerprint, the team traced the origins of immune cells in the brain and showed that bone-marrow-derived cells routinely cross the blood-brain barrier during aging and become indistinguishable from the resident microglia that have been there since birth.
Furthermore, Jaiswal’s analysis in the July paper suggested that the arrival of these immune cells from the bone marrow was associated with protection against Alzheimer’s disease, not harm, fortifying the evidence he presented in his 2023 paper. Walsh’s deeper sequencing showed that these cells cause harm, yet Jaiswal was unconvinced. Though Walsh demonstrated mutant microglia look different, looking different and causing damage are not the same thing. “As far as I know, in the Walsh paper they didn’t try to do any sort of functional experimentation to prove that these microglia were actually actively making Alzheimer’s disease or dementia worse,” Jaiswal said. The inflammatory state Walsh described, Jaiswal argued, is more likely a consequence of the disease than a cause. “The DAM signature is very clearly a response signature,” he said. “When the microglia see a lot of amyloid in the brain, they change their gene expression, and that’s what it looks like.”
He pointed to decades of clinical failure as evidence. “There’s a very long history of testing whether anti-inflammatory drugs would improve dementia outcomes, and the results have all been negative,” Jaiswal said. “There’s never been a proven link that inflammation makes Alzheimer’s worse, or that stopping inflammation would prevent Alzheimer’s.”
Gene, Timing May Determine Direction of Effect
Katherine Y. King, MD, PhD, of Baylor College of Medicine in Houston, who studies how infection and inflammation reshape the behavior of blood stem cells, explained how both groups may have reached different conclusions. “You have to think about where you’re looking in the disease process. Inflammation can be good and it can be bad. It depends on what stage of the disease process you’re in,” King said.
Walsh and Jaiswal had studied the same handful of genes that drive the vast majority of CHIP, most prominently TET2 and DNMT3A, which together account for more than 90% of cases. But in a study of more than 450,000 people in the UK Biobank, on which King was a senior author, her team found that people who carried TET2 mutations were 47% less likely to develop late-onset Alzheimer’s disease. People who carried DNMT3A mutations, the most common form of CHIP, saw no benefit at all.
“Clonal hematopoiesis is not monolithic. It is gene specific, and we’re still at the beginning of understanding the differences,” King said.
The TET2-mutant cells infiltrated the brain more efficiently than wild-type or DNMT3A-mutant cells and showed enhanced phagocytic function, clearing amyloid more aggressively, suggesting a protective benefit. But King emphasized that her mice were young, modeling early disease. “That’s a big difference between both the Jaiswal study and the Walsh study. They’re looking at cadaveric brain specimens, from people who have died from Alzheimer’s and who have presumably been affected by the disease for many years, if not decades,” King said.
Early in the disease process, King argued, inflammation may be beneficial: the immune system clearing damaged cells and pathologic debris. Later in the disease, the same inflammatory response may become maladaptive, either contributing to tissue damage or simply reflecting the degree of destruction that has already occurred.
And she raised the possibility that the causal arrow might point the other way. Walsh found more CHIP mutations in Alzheimer’s disease brains and interpreted them as a driver of disease. But TET2-mutant cells have a natural predisposition to infiltrate the brain.
“You could equally conclude that having Alzheimer’s causes clonal hematopoiesis,” King said. If that is the case, Walsh’s finding that Alzheimer’s brains harbor more mutations may reflect the disease recruiting mutant cells, not mutant cells driving the disease.
The iPSC-derived microglia in Walsh’s paper were produced by the same Mount Sinai laboratory that made the cells for King’s study. Walsh’s team saw the inflammatory profile and interpreted it as pathologic, while King’s team saw cells that were inflammatory and highly phagocytic and interpreted the combination as protective.
“The exact same data was used for both papers to come to opposite conclusions,” Jaiswal said.
Competing Predictions Face Clinical Tests
The therapeutic implications split accordingly. Walsh saw potential in repurposing cancer drugs that already target these pathways, though he noted the approach would need to be customized gene by gene.
“My prediction is that many of these drugs will be successful at preventing heart disease, but probably not neurodegeneration,” Jaiswal said.
King envisioned a third path. If TET2-mutant cells protect the brain by infiltrating it early and clearing damage, could that mechanism be harnessed without the mutation itself, which carries its own risks of leukemia and heart disease?
“Could we get there without starting with a TET2 mutation? Could we engineer that particular chemokine axis in a way that’s transient, that augments the CNS’s ability to address tissue damage?” King asked.
Both predictions are testable. A landmark cardiovascular trial called CANTOS showed in 2017 that canakinumab, a drug that blocks the inflammatory molecule interleukin (IL)-1beta, significantly reduced recurrent heart attacks in more than 10,000 patients. IL-1beta is one of the key molecules that CHIP-mutant cells overproduce.
Last year, a new generation of drugs targeting the NLRP3 inflammasome, the molecular machinery inside immune cells that produces IL-1beta, entered phase 2 trials for cardiovascular risk reduction.
Jaiswal noted other trials are testing the same anti-inflammatory approaches in neurodegeneration. “My prediction is that many of these drugs will be successful at preventing heart disease, but probably not neurodegeneration. There’s trials that will read out in the next few years that will show whether we’re right or not,” he said.
Walsh, Jaiswal, and King reported having no relevant financial disclosures. Disclosure information for study authors is available in the original study publications.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.