Semaglutide Improved Alzheimer's Biomarkers and Did Nothing for Patients

The evoke and evoke+ trials asked whether semaglutide could slow Alzheimer's disease. They were serious trials: 3,808 people aged 55 to 85 with mild cognitive impairment or mild dementia, all with amyloid buildup confirmed by testing, given oral semaglutide or placebo for two years.

The answer was no.

On the main measure, a standard scale of cognition and daily function called the CDR-SB, there was no difference between semaglutide and placebo at two years. The secondary measure of daily living activities showed no difference either. The company discontinued the planned extension phase. The full results have since been published in The Lancet.

The part worth sitting with

The trials were not a blank. Semaglutide did improve Alzheimer's-related biomarkers, including reductions of up to about 10 percent in markers linked to neuroinflammation. Those changes were statistically real.

They just did not help anyone.

Patients on the drug did not think more clearly, remember more, or manage daily life better than patients on placebo. The biology moved in the direction everyone hoped, and the people did not.

Why this is the most useful result on this site

Almost every claim made for longevity and anti-aging peptides is a biomarker claim, or one step further back, a mechanism claim.

Epitalon activates telomerase and telomeres get longer. MOTS-c and humanin signal mitochondrial health, and levels are higher in the children of people who lived to 100. SS-31 stabilizes cardiolipin and protects mitochondrial function. Every one of those statements is, as far as anyone can tell, true. We say as much on the epitalon, MOTS-c, and SS-31 pages, and we went through the longevity versions in detail in our look at telomeres and hype.

What the evoke trials demonstrate is that a true mechanism and a moving biomarker can coexist with zero benefit to the person. This was not a small or sloppy study. It was large, long, well controlled, properly blinded, and run by a company with every incentive to find an effect. It still came back empty on the outcomes that matter.

If that can happen to a drug taken by millions, in a trial of this size, it can certainly happen to a peptide whose case rests on cell cultures and mice.

Intermediate measures and real ones

There is a useful contrast with something we wrote about last week. SS-31 received accelerated approval based on knee extensor muscle strength, an intermediate measure, precisely because regulators accepted it as reasonably likely to predict real benefit. That approval remains conditional on a confirmatory trial proving the benefit is real.

That is the system working as designed: an intermediate measure can open a door, but it does not close the question. The evoke trials are what it looks like when an intermediate measure moves and the door turns out to lead nowhere.

What this does not say

It does not say semaglutide is bad for the brain, or that the biomarker findings were noise. Researchers have taken them as a signal about which pathways are worth pursuing, possibly in combination with other treatments.

It also does not mean biomarkers are useless. They are how research makes progress. They are simply not evidence of benefit on their own, and the distinction gets erased in most of what is written about peptides.

The practical version

When you read that a compound raised, lowered, or restored some measurable thing, the useful question is not whether the measurement is real. It usually is. The question is whether anyone has shown that changing it makes a person better off.

For most peptides sold for aging, energy, and cognition, nobody has tested that. For semaglutide in Alzheimer's disease, somebody did, in 3,808 people over two years, and the answer was no.

Sources: Alzheimer Europe on the evoke and evoke+ topline results and the full trial report in The Lancet.