The results hit like a ton of bricks — and they have critical implications for the recently FDA-approved Alzheimer's blood test.
Researchers were packed wall-to-wall at a conference in San Diego. They were there to hear some of the most highly anticipated results in Alzheimer's research in years. Those results came from the EVOKE and EVOKE+ trials, which had been testing the impact of GLP-1 medications on patients with early-stage Alzheimer's disease.
GLP-1 receptor agonists had already proven transformative. They were a game changer for treating type 2 diabetes. Then they revolutionized weight-loss medicine. And there was good reason to think they might prove a powerful defense against dementia too.
Then the numbers went up on the screen.
It was a total failure. A daily dose of semaglutide did nothing to delay Alzheimer's progression [1].
"All told, semaglutide failed to distinguish itself from placebo on the primary end point of change in CDR-Sum of Boxes (CDR-SB) over a 104-week stretch, or 2-year period… there was an estimated –0.06-point (95% CI, –0.48 to 0.36) difference between oral semaglutide (2.2) and placebo (2.2; P = .7727) in EVOKE during that time… a 0.15-point difference (95% CI, –0.24 to 0.54) between semaglutide (2.1) and placebo (2.0), which was also not significant (P = .4604)."
There has been failure after failure in Alzheimer's treatment trials. And now another one has joined the list.
One possible reason for these repeated disappointments: Alzheimer's disease is being detected too late. By the time a diagnosis is made, the condition has already silently wreaked havoc on brain tissue for years — sometimes decades. More than anything else, what the field has needed is a reliable way to find Alzheimer's early, when there is still time to intervene.
With a fresh FDA approval of a simple blood test, that may now be possible.
Table of Contents
- A Long History of Late Detection
- How Researchers Have Tried to Diagnose Alzheimer's
- A Major Step Forward (and Its Limitations)
- References
A Long History of Late Detection
There is reason to be cautious about any Alzheimer's diagnostic breakthrough. The field has been here before.
For most of the hundred-year history of Alzheimer's disease, the only way to confirm a diagnosis with certainty was to wait until a person died and examine their brain post-mortem.
That began to change in the 1990s. By then, researchers had a better understanding of the disease's progression and had identified specific proteins and protein fragments characteristic of the condition. The working hypothesis was that samples of the cerebrospinal fluid (CSF) — the fluid surrounding the spinal cord and brain — might contain detectable clues about the activity of these key proteins.
Unlike blood, which must cross the blood-brain barrier, cerebrospinal fluid is in direct contact with the brain [2]. Sampling it for Alzheimer's biomarkers is a bit like searching through discarded mail for clues about someone's financial situation — messy and indirect, but informative.
A breakthrough followed. Scientists found markers in cerebrospinal fluid that tracked tightly with Alzheimer's diagnoses. One of the most studied was beta-amyloid 42: levels of this protein fragment were significantly lower in patients with confirmed Alzheimer's disease [3].
Based on these discoveries, CSF tests — using fluid drawn from the lower back via lumbar puncture — were developed and deployed extensively in Alzheimer's research in the United States. However, their use has not spread widely beyond research settings [4].
Imaging tools such as PET scans have also advanced the field considerably. For the first time, PET scanning gave clinicians a way to detect the physical changes in the brain caused by Alzheimer's disease during a patient's lifetime.
But both approaches carry serious drawbacks. PET scans are expensive and inaccessible for many patients. Lumbar punctures are highly invasive [5]. And for much of the period when these tests were being developed, the limitations barely mattered — because there was little clinicians could do even if a diagnosis was made.
That has changed. Now that approved medications exist that can slow disease progression in some patients [6], the stakes around early, accurate diagnosis have risen sharply. Clinicians need a simple, affordable tool that puts reliable diagnosis in their hands — and in the hands of the patients who need it most.
How Researchers Have Tried to Diagnose Alzheimer's
It looked like a simple solution might arrive in 2023.
Midway through that year, Quest Diagnostics announced it would begin offering an Alzheimer's blood test directly to consumers through its clinics.
The test was built on the theory that amyloid plaque formation is a central driver of Alzheimer's disease progression [7]. The logic has an intuitive appeal — amyloid plaques have been associated with Alzheimer's for decades. However, researchers have grown increasingly concerned that amyloid's causal role may not be as central as once assumed.
Adding to the controversy, some scientists have found evidence of potential fabrication in published research supporting the amyloid theory — a finding that has cast a shadow over a significant body of literature in this area [7].
How well did the consumer blood test actually work? Early data presented at a conference drew on a group of 209 participants. The test was reported to have a sensitivity of 89% — meaning that out of 100 people who actually had Alzheimer's, the test would correctly flag 89 as positive. The specificity was 71%, meaning that for every 100 people without the disease who took the test, 29 would receive a false positive result [8].
Critics were swift to raise concerns. A 29% false positive rate was not adequate for a disease diagnosis. At population scale, this would generate mass quantities of incorrect diagnoses — driving a surge in demand for follow-up testing and specialist services at dementia clinics that could overwhelm capacity and divert resources from those who genuinely needed care. The psychological toll of a false Alzheimer's diagnosis on individuals and families is also not trivial.
Because the test was sold directly to consumers — not prescribed by a physician — it did not require FDA approval under the regulatory framework at the time. But the chorus of concern from the medical community was sufficient to limit its uptake among clinicians, and the test never gained meaningful clinical traction.
A Major Step Forward (and Its Limitations)
That context is what makes the recent FDA approval of a new Alzheimer's blood test significant. It appears the field may finally have a simple, accurate diagnostic tool capable of filling the gap — though one important limitation must be understood before interpreting any result.
The newly FDA-approved test measures levels of p-tau181 in the blood. It is intended for use in adults aged 55 and older who are already showing symptoms of cognitive decline [9].
P-tau181 — phosphorylated tau at threonine-181 — is a protein that has been extensively studied as a reliable biomarker for the presence of Alzheimer's pathology. Elevated p-tau181 levels correlate with amyloid burden and neurofibrillary tangle formation in the brain, and multiple independent research groups have validated its diagnostic utility [10].
But how effective is the test in practice? This is where the nuance becomes important.
The p-tau181 blood test is designed to rule out Alzheimer's disease — not to rule it in [11]. In other words, a negative result is highly informative. A positive result requires further investigation.
The test performs the rule-out function exceptionally well. In a clinical trial evaluating the test in a population similar to a general primary care setting — where the disease was relatively rare and early stage when present — a negative result carried a probability of approximately 98% that the person did not have Alzheimer's disease [12]. That is a clinically meaningful negative predictive value.
A positive result is a different matter. In that same trial, a positive test result corresponded to confirmed Alzheimer's pathology only about 22% of the time [13]. That is not a high positive predictive value — meaning most positive results, at least in a lower-prevalence population, will not reflect true disease.
Despite this limitation, the test represents a genuine advance. Rather than applying broad treatments across everyone who might possibly have Alzheimer's, clinicians can now use the blood test to identify patients who almost certainly do not have the disease — and spare them unnecessary, expensive, and invasive follow-up procedures. For those who do test positive, follow-up with PET imaging or cerebrospinal fluid analysis via lumbar puncture remains necessary to confirm the diagnosis [14].
This is also why the test carries a specific clinical indication: adults aged 55 and older who are presenting with symptoms of cognitive decline. It is not a routine screening tool for otherwise healthy, asymptomatic individuals. In a healthy, low-risk population, a positive result would very likely be a false positive — generating significant anxiety and triggering a cascade of costly, invasive follow-up testing with low probability of finding genuine disease.
This same reasoning explains why direct-to-consumer companies marketing blood tests that claim to detect Alzheimer's long before symptoms appear warrant careful scrutiny [15]. One such company, for example, markets its p-tau217 test with the following claim:
"P-tau217 is a key blood biomarker linked to Alzheimer's disease that begins to change long before memory symptoms surface. Measuring p-tau217 through a simple blood test allows for earlier detection and ongoing monitoring of potential cognitive changes."
This is a classic illustration of why the correct test, applied to the right population, at the right time, matters enormously — and why running every available test on healthy individuals, even if cost were no barrier, is not necessarily in the interest of those being tested.
References
2. https://pmc.ncbi.nlm.nih.gov/articles/PMC2915796
3. https://pubmed.ncbi.nlm.nih.gov/7574461/
4. https://pmc.ncbi.nlm.nih.gov/articles/PMC10013957
5. https://jamanetwork.com/journals/jama/article-abstract/2842578
6. https://www.sciencedirect.com/science/article/pii/S1878747925000480
9. https://www.roche.com/investors/updates/inv-update-2025-10-13b
10. https://jamanetwork.com/journals/jama/article-abstract/2842578
11. https://jamanetwork.com/journals/jama/article-abstract/2842578
12. https://jamanetwork.com/journals/jama/article-abstract/2842578
13. https://jamanetwork.com/journals/jama/article-abstract/2842578
14. https://jamanetwork.com/journals/jama/article-abstract/2842578
