Are there any drugs that can slow or stop the progress of Alzheimer’s?

The short version: the drugs approved for Alzheimer’s slow the slide modestly for some people, but none have been shown to stop or reverse the disease, and the newest ones carry serious brain-safety risks.

The evidence here covers two main drug classes. The older ones, acetylcholinesterase inhibitors like donepezil and galantamine, boost a brain chemical that fades as Alzheimer’s progresses. A 1998 open-label study found that donepezil slowed cognitive decline by about 5 points per year on a 70-point scale compared to an estimated untreated rate [2]. A 2006 cohort of 224 patients found that people taking these drugs had about half the odds of rapid decline over 12 months [1]. But these are modest, temporary effects—symptom management, not disease modification. The same 2006 study found that antipsychotics and benzodiazepines were linked to faster decline, with odds two to three times higher [1], a reminder that what you add matters as much as what you avoid.

The newer class, anti-amyloid monoclonal antibodies (lecanemab, donanemab, aducanumab), aim to clear the sticky plaques in the brain. A 2025 review of the phase 3 trials reports that lecanemab slowed cognitive decline by about 27% on one scale over 18 months, and donanemab by about 35% [4]. Those sound meaningful, but the absolute difference is small—a few points on a 100-point scale. The same review reports that these drugs cause a side effect called ARIA (brain swelling or small bleeds): lecanemab caused ARIA-E in 12.6% of patients, donanemab in 24%, and aducanumab in up to 41% at the highest dose [4]. APOE4 gene carriers, who already have higher Alzheimer’s risk, had even higher ARIA rates [4]. The benefit is modest; the risk is real and measurable.

The evidence does not include any drug that stops or reverses Alzheimer’s. The older drugs offer modest, temporary symptom relief. The newer antibodies clear plaque but produce small cognitive gains at the cost of a real risk of brain swelling or bleeding. A 2025 review of the field concludes that multi-target approaches are still in development [3]. No trial here compared treated patients to a truly untreated group over many years, so the long-term net effect—whether the small cognitive benefit is worth the brain-safety risk—remains unknown.

My call: for someone already diagnosed, the older drugs may offer modest temporary help with known risks; the newer antibodies carry a real chance of brain injury for a small cognitive gain. Confidence: moderate.