Alzheimer's symptoms such as memory loss could be prevented by targeting a chemical that dismantles brain connections, research suggests.
Scientists have already started work searching for a drug that will block the mechanism, discovered in mice.
If successful, a treatment that effectively protects against the effects of Alzheimer's could be available in the next 10 years.
It has long been known that the disease is linked to a build-up of toxic amyloid-beta protein in the brain.
Researchers at University College London have now found that amyloid-beta stimulates production of another protein, Dkk1, which is largely responsible for Alzheimer's symptoms.
Dkk1 destroys synapses, connections between neurons, in the hippocampus area of the brain vital to learning and memory.
Studying samples of mouse brain in the laboratory, the scientists found they could neutralise Dkk1 with a specific antibody.
Neurons exposed to the antibody remained healthy with no synaptic disintegration.
In practical terms, it is unlikely that a vaccine-type treatment could tackle Alzheimer's the same way.
But study leader Dr Patricia Salinas said now that Dkk1's role was known, there was a chance of developing drugs to target it.
"These novel findings raise the possibility that targeting this secreted Dkk1 protein could offer an effective treatment to protect synapses against the toxic effect of amyloid-beta," she said.
"Importantly, these results raise the hope for a treatment and perhaps the prevention of cognitive decline early in Alzheimer's disease."
Her team is now working with a biotech company to develop molecules that can block Dkk1.
A major obstacle is overcoming the "blood-brain barrier" - a natural "firewall" that prevents damaging substances entering the brain.
"It's a long shot but we're talking about a period of about 10 years," Dr Salinas added.
Since Dkk1 is normally present in such small quantities in a healthy human brain, targeting it is unlikely to produce serious side effects, she said.
The protein is thought to play an important role in early development, helping the brain to construct its "wiring". In adult life, Dkk1 ceases to have any known useful function in the brain and its levels are drastically reduced.
Amyloid-beta leads to increased production of Dkk1 at a time when it is not wanted.
"The key thing is that this factor gets developed very, very quickly by amyloid-beta," said Dr Salinas.
"If we could target Dkk1 early in Alzheimer's disease we may be able to ameliorate or delay loss of memory."
Another possibility is that the protein could be used as a biomarker for early Alzheimer's diagnosis, she said.
It could be detected in cerebrospinal fluid, although such a test would be invasive.
The study, published in the Journal of Neuroscience, was funded by the charity Alzheimer's Research UK and the Biotechnology and Biological Sciences Research Council (BBSRC).
Dr Simon Ridley, head of research at Alzheimer's Research UK, said: "By understanding what happens in the brain during Alzheimer's, we stand a better chance of developing new treatments that could make a real difference to people with the disease.
"Studies like this are an essential part of that process but more work is needed if we are to take these results from the lab bench to the clinic."
Alzheimer's is the most common form of dementia, which affects around 750,000 people in the UK and costs the economy an estimated £23 billion per year.