News · Brain & Mental Health
An Alzheimer's blood test estimated 5-year risk before symptoms
A team led from Harvard and Mass General Brigham pooled 2,684 older adults with no memory or thinking problems, from six Alzheimer's studies, and reported in JAMA that those with the very highest blood p-tau217 had about a 38 percent chance of decline within five years. The cohorts were selected research volunteers, so the figure may not fit an average older adult.
Based on a peer-reviewed study in JAMA
- A team led from Harvard and Mass General Brigham pooled blood samples and follow-up records from 2,684 older adults with no memory or thinking problems, drawn from six Alzheimer's research cohorts in North America, Japan, and Australia. The results were published in JAMA.
- Adults with very high p-tau217, more than 2.5 standard deviations above the mean, had a 38 percent absolute risk (95% CI, 33 to 43 percent) of progressing to cognitive impairment within 5 years.
- Adults in the high band, 1.1 to 2.4 standard deviations above the mean, had a 24 percent absolute risk (95% CI, 20 to 28 percent) over the same 5 years.
- Each 1 standard-deviation rise in p-tau217 was linked to higher risk of progression (hazard ratio 1.38; 95% CI, 1.30 to 1.46), and the link held after accounting for amyloid PET brain scans (hazard ratio 1.32; 95% CI, 1.24 to 1.41). Higher p-tau217 also tracked with faster cognitive decline.
- The cohorts were people already enrolled in Alzheimer's studies, not a random sample of the public, so the 38 percent figure may not apply to an average older adult. The authors say validation in unselected populations is still needed, and 10-year estimates were limited by sparse data. This is a way to gauge risk, not a diagnosis or a screening test to request.
Blood tests can now catch the earliest, silent signs of Alzheimer’s disease long before anyone notices a lapse. What they could not do, until recently, was tell a healthy person how likely real decline actually was, or when. A large analysis in the journal JAMA has taken a step toward that.
Pooling records from thousands of older adults with no memory or thinking problems, researchers measured a blood marker called p-tau217 and tracked who declined. Those with the very highest levels had close to a 4 in 10 risk of progression to cognitive impairment, the stage where memory and thinking problems start to disrupt daily life, within five years.
There is a large asterisk. Everyone in the study had already enrolled in Alzheimer’s research, which makes them a particular slice of the population; people who volunteer for brain studies tend to differ from those who never do. The authors are explicit that further validation in unselected populations is needed before a number like that should shape any individual’s care.
What p-tau217 actually is
Tau is a protein that, in Alzheimer’s, twists into tangles inside brain cells. A small piece of it, phosphorylated tau 217, leaks into the blood as those changes take hold. Measured in a sample, it becomes p-tau217, one of several blood-based biomarkers for Alzheimer’s disease whose levels climb early, well before memory falters. A biomarker is just something measurable, here a protein you can pick up in a sample, that reflects a process inside the body.
The appeal is easy to see. A blood draw costs less and asks less of a person than a brain scan or a spinal tap. But a high p-tau217 reading only signaled that the underlying biology was in motion. It could not tell someone with no symptoms how likely they were to actually decline, or how soon.
Putting a number on the risk
That gap is what this study closed. By pooling six observational and clinical trial cohorts, groups of older adults tracked over time, the researchers could for the first time estimate how much a single p-tau217 reading shifts a person’s odds.
They ranked people by how high the marker sat. The absolute risk of progression over the next five years ran close to 4 in 10 for those in the highest band, well above average, and about 1 in 4 for the band just below. Those with low levels mostly stayed steady. Even in the highest group, then, most had not declined by that point, though their odds were far worse than a low reading would suggest.
Higher p-tau217 also tracked with faster cognitive decline, the gradual slipping of memory and thinking, as time went on. And even after the team accounted for amyloid, a sticky protein that is another hallmark of Alzheimer’s and normally seen only on a costly brain scan, a high reading still carried an increased risk of progression.
Who the study actually followed
The strength here is scale. The analysis pooled 2,684 cognitively unimpaired older adults, free of memory or thinking problems at the outset, at a median age of about 70. Over a median follow-up of about five years, and more than a decade for some, 478 of them, close to 500, met the study’s threshold for progression to cognitive impairment. Because the estimate drew on several cohorts at once, it held steady from one group to the next, a sign it was not a fluke of any single study.
The weakness is who those people were. Every cohort was a selected sample of cognitively unimpaired older adults, volunteers who had already joined Alzheimer’s research. That is a particular slice of the public, and those who sign up for brain studies often differ from those who do not, so the exact odds may not carry to a random person of the same age. The authors say as much: further validation in unselected populations is needed before the numbers can guide individual care.
Two other limits are worth keeping in view. The five-year figures rest on far more data than the ten-year ones: risk did climb further over a decade, but those longer-term estimates were constrained by limited data. And levels of the marker can be pushed around by other factors that influence biomarker levels and dementia risk, from kidney function to obesity, which is part of why one result is never the whole story.
What it means if you are worried about dementia
For now, this is not a test to go asking your doctor for. It was built and checked in research volunteers, and it estimates risk for a group. It cannot diagnose an individual.
Where it already earns its place is research. A cheap marker that flags who is most likely to decline lets scientists enroll the right volunteers into prevention trials, well before symptoms. Rachel Buckley of Mass General Brigham, who led the work, put the near-term use plainly: “Once verified, these blood tests could be used to recruit patients for clinical trials of treatments to prevent cognitive decline and dementia.”
If those treatments do prove out, a test like this could one day help doctors decide who to treat, and when. That day has not arrived. For a healthy person today, the honest takeaway is modest: even the highest results left most people free of impairment within five years. The everyday things that help an aging brain, steady sleep, exercise, blood pressure, hearing, still matter more than any single test result.
People also ask
What is p-tau217, and what does the blood test show?
P-tau217 is a piece of the tau protein that leaks into the blood as Alzheimer's changes build in the brain. Levels rise early, often before any memory or thinking problems, so researchers use it as a marker of Alzheimer's brain pathology. The test gauges risk; it does not diagnose the disease.
Does a high result mean I will get Alzheimer's?
No. Even in the very high group, about 38 percent progressed to cognitive impairment within five years, which means most did not. A high level raises the odds of decline over time; it is not a diagnosis, and the timing varies from person to person.
Should I ask my doctor for a p-tau217 test?
Not on the strength of this study. The people measured were already enrolled in Alzheimer's research, not a cross-section of the public, so the risk figures may not carry over to an average older adult. The authors say the test needs validation in unselected populations before it guides individual care. For now it is mainly a research and clinical-trial tool.
How solid is the 5-year risk estimate?
It came from pooling six cohorts totaling 2,684 people, which makes the 5-year estimate reasonably stable. The 10-year estimates were weaker, limited by how few people had been followed that long. Risk did rise further over 10 years, but with less certainty.