Search
Health

Alzheimer's blood test matched spinal fluid accuracy with real-world cutoffs, study finds

Lund University researchers judged the test by the approved thresholds doctors actually use and found it detected Alzheimer's changes at least 90% of the time.

Lund University's white Main Building behind a fountain and flowerbeds of tulips and pansies under a partly cloudy sky.
File photo. Photo: Einaz80 / Wikimedia Commons, CC BY-SA 4.0

A blood test for Alzheimer's disease identified the brain changes of the disease as accurately as a spinal fluid test, long a diagnostic standard, when both were judged by the thresholds doctors use in clinics. The study, from Lund University in Sweden, was published in Nature Aging.

The test measures p-tau217, a protein the researchers describe as specific to Alzheimer's. The team checked results against PET brain scans, which can confirm whether Alzheimer's-related changes are present. The blood test detected those changes with at least 90% accuracy and performed as well as cerebrospinal fluid testing in three separate groups: a U.S. research cohort, a Swedish memory clinic cohort and a Swedish primary care cohort.

"It is the first time this has been demonstrated," said Noëlle Warmenhoven, a doctoral student in Lund's Clinical Memory Research Unit and the paper's first author. Versions of the test already have regulatory approval for clinical use in Europe, the United States and Japan, among other places.

The finding matters most for patients facing a lumbar puncture. Collecting cerebrospinal fluid means inserting a needle into the lower back, while a blood draw can happen at any routine visit. If the two tests are equally reliable, many people with memory complaints could skip the more invasive procedure.

Why the cutoffs make this study different

Earlier studies of Alzheimer's blood tests have reported strong accuracy, so the headline number alone is not the news. The more useful contribution is how the Lund team reached it.

Every test of this kind sorts results into positive, negative or inconclusive based on cutoff values. Before the tests were approved, researchers had to set their own cutoffs, typically choosing whatever values worked best for the group of patients in front of them. That can make a test look better on paper than it will in a clinic. Warmenhoven gave an example: a threshold that perfectly separates people with and without Alzheimer's in one study could produce many false positives in another study where far more participants are disease-free.

Now that the tests are approved, manufacturers supply predefined cutoffs, and those are the values physicians will rely on. The Lund researchers used those regulator-approved thresholds instead of tuning their own. "These new findings are based on cutoff values that directly reflect clinical practice," Warmenhoven said.

That design choice is what makes the result credible to working clinicians. Testing a tool under the same rules doctors will apply is a fairer test than an idealized one, and the blood test held up across a research cohort, a specialist clinic and a primary care setting. That spread matters, because a test that performs well only in carefully selected research volunteers would be of limited use.

Sebastian Palmqvist, an associate professor of neurology at Lund and a senior consultant at the memory clinic at Skåne University Hospital, said earlier studies had compared blood tests against spinal fluid tests as the reference. As a result, he said, many clinicians came to view the blood tests as good but probably a step below the fluid tests already in use. Measured against PET imaging instead, the new results suggest that "a blood test is just as accurate as cerebrospinal fluid testing," he said.

The study also carries a pointed finding about test design. The blood test cleared for use in the U.S. measures p-tau217 together with amyloid-beta 42, a protein that builds up into plaques in the brains of people with Alzheimer's. The Lund group reported, as it has in earlier work, that this combination does not necessarily outperform p-tau217 measured on its own. The researchers did not say the combined test performs worse, but their results suggest the simpler single-protein measurement gives up little or nothing.

Who can use the test, and its limits

For now, the blood test is recommended only in memory clinics and other specialist settings, Palmqvist said, and only for people who already have cognitive symptoms. It is not a screening tool for people with no memory problems.

It also does not always deliver a clear answer. Some results come back inconclusive, and those patients may still need a PET scan or a spinal fluid test to reach a reliable diagnosis. The study's summary does not say how often that happens, which is a practical question for any clinic weighing how many invasive tests the blood draw will actually replace.

Other details readers might want are also missing from the published summary, including how many people were in each cohort and what the test costs. The findings come from the same research group that has studied these markers for years, and no independent assessment of the paper was reported.

The biggest unmet need is in primary care. Specialized memory clinics have access to PET scans and spinal fluid testing, but family doctors, who see most people with early memory complaints, have few quick tools. The researchers note that one in five women and one in 10 men will develop Alzheimer's during their lifetime, a burden specialist clinics alone cannot absorb.

Work is underway to figure out how the test could be rolled out in primary care, Palmqvist said. The study's inclusion of a Swedish primary care cohort gives that effort a useful starting point, though a recommendation for general practice has not been made.

The pressure for faster diagnosis has grown since the anti-amyloid drugs lecanemab and donanemab reached the market in 2023 and 2024. Both are meant for people in early stages of the disease, and both require confirmation of amyloid in the brain before treatment. A reliable blood test could shorten the path from a first appointment to that confirmation.

Palmqvist also pointed further ahead. Screening people with no memory problems at all could become relevant, he said, if drug treatments prove effective at that stage. That is a conditional and distant prospect. It would also reopen the problem Warmenhoven described: in a population where most people do not have the disease, even a highly accurate test can generate a meaningful number of false positives. Getting the cutoffs right for symptomatic patients, as this study did, is the necessary first step before anyone tries to answer that harder question.

Sources

Amara Okafor

Amara Okafor covers community health, access to services, and developments affecting everyday wellbeing. Her work follows research and public information about the conditions that influence health across different communities. She focuses on explaining how services are organized, whom new initiatives may reach, and what the available evidence says about their effects.

View all stories