An experimental vaccine for Alzheimer's disease, which was quickly pulled from testing last year after it caused serious side effects, has halted and even reversed the brain disease in some who got the shots, according to the first follow-up study of those patients.

The results, which several experts described as remarkable, represent a bittersweet coda for the now abandoned vaccine, which was designed to rally the immune system against the debilitating ailment.

The vaccine's developers do not plan to resurrect the product, which caused brain inflammation in 18 of 300 patients who got the shots. But the positive findings, published in today's issue of the journal Neuron, offer tantalizing evidence that similar vaccines under development may be able to save people from the syndrome's devastating decline. Alzheimer's afflicts 4 million Americans and is expected to strike twice that number in the next two decades as baby boomers enter their senior years.

"When we learned about the side effects, everyone was disappointed," said Peter St. George-Hyslop, a University of Toronto Alzheimer's expert. "But this now says we shouldn't be so fast to run away from this, and we should consider going back and redesigning or trying another related approach."

Researchers cautioned that the new results should be considered preliminary because they reflect tests conducted on just the first 28 participants evaluated after getting the vaccine. Nonetheless, several experts said they found the results highly encouraging.

"It shows that the concept of vaccination for Alzheimer's is really something that is possible," said Bengt Winblad, chief of geriatric medicine at the Karolinska Institute in Stockholm, who wrote an editorial accompanying the new study. "I'm sure that in five years we'll have some kind of vaccination for Alzheimer's."

Alzheimer's is a disease characterized by a buildup of abnormal proteins in the brain, progressive memory loss and dementia. Scientists have feuded for years over the significance of the protein deposits -- known as amyloid plaques -- with some suspecting that they directly cause the symptoms of Alzheimer's and others believing they are mere byproducts of the disease.

The vaccine -- made by Elan of Dublin, Ireland, and Wyeth of Madison, N.J. -- was developed in the belief that the plaques are indeed the problem. The vaccine triggers the body's immune system to attack the plaques and to aid in their removal by white blood cells.

Animal studies had suggested the approach was safe and would work. Vaccinated mice with an Alzheimer's-like condition developed antibodies that helped clear some of the amyloid from their brains. None suffered brain inflammation as a side effect, and many started scoring better on memory tests.

But early human testing in patients with mild to moderate Alzheimer's was halted 15 months ago when a dozen participants developed meningoencephalitis, an inflammation of the brain and surrounding membranes that can lead to seizures and other serious complications. Another six were diagnosed with the problem soon after. None died as a direct result of the vaccine, but some had difficulty recovering.

The new analysis, the first to compare cognitive test results before vaccination and one year after, was conducted by researchers at the University of Zurich, one of 28 sites where the vaccine was being tested. To prevent bias, the researchers were not told which patients had been vaccinated. They documented steep mental declines in nine patients, marginal declines in 13 and stable scores or improvements in six.

Separate tests later showed that the nine patients who fared worst had no amyloid antibodies in their blood -- evidence that they were in the unvaccinated control group or for some reason did not respond to the vaccine. The 13 who experienced only slight declines had modest concentrations of antibodies, indicating an average response to the vaccine. Those who remained stable or improved had the highest levels of antibodies.

That "dose-response" relationship between antibody levels and clinical effects constitutes strong evidence that amyloid proteins are indeed a primary cause of Alzheimer's symptoms, scientists said.

"That was important to show, because otherwise you might solve the amyloid problem and leave the clinical symptoms intact," said Zaven Khachaturian, senior science adviser to the Alzheimer's Association and a former director of the federal Office of Aging Research. "People don't go to the doctor saying, 'Help me, I have lots of plaques in my head.' They say, 'I can't remember; I can't function.' "

Most impressive, two patients with high antibody levels enjoyed significant increases on the Mini Mental State Examination, a widely used measure of cognitive health that tests orientation, attention, memory and the ability to follow simple spoken and written commands.

Some researchers said they wished the vaccine's developers had not dropped the product so quickly and wondered aloud whether concurrent doses of anti-inflammatory drugs could have prevented the few cases of encephalitis.

"If I knew that I were at risk for Alzheimer's disease . . . I would take the vaccine right away and watch carefully for side effects -- with the cortisone close to my pillow," said neuroscientist Roger M. Nitsch, who led the Swiss study.

Several companies are developing other approaches to attacking amyloid in the brain. And Elan is developing modified versions of its vaccine designed to raise antibodies but not the T lymphocyte cells that appear to have been the cause of the brain inflammation.

"We believe they will optimize and focus on the good qualities and eliminate the negative qualities," said Dale B. Schenk, a vice president at Elan. He said the company hopes to gain Food and Drug Administration permission to start testing those products in patients by the end of this year.