All four patients who received the optimal dose of the drug, Gammagard, had no decline in several measures of cognition and daily function for three years, researchers said.

Dr. Norman Relkin of Weill Cornell Medical College said the results were "remarkable" because patients with Alzheimer's disease typically worsen within 12 months.

But some experts not involved in the trial said it was premature to conclude that the drug was responsible for the stabilization.

"There will be some patients who at three years don't have a decline" even without an experimental therapy, said Dr. Rachelle Doody, director of the Alzheimer's Disease and Memory Disorders Center at Baylor College of Medicine. She said some patients go six or eight years without worsening.

Dr. Samuel Gandy, director of the Center for Cognitive Health at the Mount Sinai School of Medicine, said the results left him "optimistic to a minor degree, not really gushing."

Whether the results are a fluke or not could be known by the first half of next year, when the results of a Phase 3 trial are expected. The evidence "really suggests this is a drug effect and not just an accident," said Bill Thies of the Alzheimer's Association.

Go to full story: http://www.nytimes.com