An Alzheimer's antibody lowered brain plaque but did not slow decline in a trial of 542 people

ALZHEIMER'S DISEASE slowly takes away memory and then the skills of daily life. One of its clearest physical signs is amyloid, a sticky protein that builds up in the brain. It gathers into large deposits called plaques and also into small loose clumps, known as oligomers, that drift between nerve cells. The newest drugs for the disease are antibodies, lab-made immune proteins designed to grab amyloid so the body can remove it.

Acumen Pharmaceuticals built its antibody, sabirnetug, on the idea that the small clumps rather than the plaques do most of the damage to nerve cells. In a phase 1 study of 65 people, published in 2025, the antibody bound amyloid oligomers more than 650 times more strongly than single amyloid molecules and showed little binding to plaque. Even so, plaque fell by about 25% and 20% over three months in the two groups given the most intensive doses. That study was built to test safety and dosing, not benefit to patients.

Testing benefit needs a fairer kind of test, in which some volunteers receive a dummy infusion that looks like the drug but contains none of it. That group is called the placebo arm, and comparing the others with it separates the effect of the drug from the ordinary course of the illness. The test is also stronger when chance alone decides who gets which infusion and when nobody involved knows who got what until the end, a design known as randomized and double-blind. Acumen's trial, ALTITUDE-AD, used it, and its record on ClinicalTrials.gov says the volunteers, their care providers, the investigators and the people scoring the results were all kept unaware of the assignments.

The trial enrolled 542 people with mild memory problems or mild dementia due to Alzheimer's with amyloid confirmed in the brain. They were split among three arms: an infusion every four weeks of 35mg of the drug per kilogram of body weight, an infusion every four weeks of 50mg per kilogram after two starting doses of 35mg, or the dummy, with each arm planned to hold up to 180 people.

The one result fixed before a trial starts to decide its success is called the primary endpoint, and for ALTITUDE-AD it was the change after 18 months on a scale called iADRS, which runs from 0 to 144 and combines a test of memory and thinking with a rating of how well the person manages everyday tasks.

A difference between groups counts as real only if it is larger than chance would be likely to produce in a trial of that size. Such a difference is said to reach statistical significance. Falling short does not prove the drug has no effect, but it means the trial could not tell any effect apart from random variation between people.

On October 30th Acumen reported that neither dose met the primary endpoint. Other planned measures of decline, including a common dementia rating scale, also fell short, and the company said there was no "meaningful difference" between either treated arm and the placebo arm. The company did not release effect sizes, p-values, the amount by which plaque fell or the number of people who left the trial.

Inside the brain, the drug did what it was designed to do. A biomarker is a sign of disease that can be measured directly, such as plaque seen on a PET scan, a type of imaging that shows amyloid deposits. Scans showed a statistically significant fall in plaque at both doses, at 36 weeks and again at 76 weeks. Yet by the company's account, treated people did not lose memory and daily skills more slowly than those given the dummy. About 10% of treated participants showed brain swelling on MRI scans, a known side effect of amyloid antibodies that doctors call ARIA-E.

In earlier work, a secondary analysis of 1,582 people in a trial of donanemab, a plaque-targeting antibody, published in JAMA Neurology in 2025, found that lower amyloid after treatment went with slower decline. Without detailed numbers it is not yet possible to compare the two drugs or to say whether the idea of targeting small clumps is wrong or whether this antibody did not test it well enough.

Acumen expects more blood and imaging data in the coming weeks but says it does not expect them to change the trial's conclusions. The company plans to stop developing sabirnetug, including an extension of the trial in which all participants would have received it.