Fragile minds

New Scientist, 01 February 2003

THE only way to know for sure if someone has Alzheimer's is to wait until they die and then autopsy their brain. On the outside, the disease can look just like other forms of dementia. But inside the brain there's something special a pathologist will be looking for: telltale tangles and plaques in and around the brain cells, made from a protein called amyloid beta.

The protein connection was first discovered in the mid-1980s, and ever since, A has been considered the main culprit. The amyloid hypothesis is still by far the most widely accepted explanation for the disease. It proposes that A, which is usually soluble, for some unknown reason spontaneously clumps together, killing off nearby neurons and causing dementia. The tantalising implication was that if you could get rid of the peptide clumps, you could cure Alzheimer's.

Millions of research dollars have already been gambled on the idea. Scientists around the world are either trying to persuade the body to produce less A, suck out what has already been deposited, or crank up the body's own systems for removing it. Some approaches are now being tested in people. But even though it has proved relatively straightforward to cleanse a rodent brain of the rogue protein, a cure for Alzheimer's seems as elusive as ever. The path from lab mouse to cure is never straightforward, but maybe there's a more fundamental problem. Despite being such a characteristic marker of the disease, there is no hard evidence that A is causing Alzheimer's. Are efforts to find the real cause blinkered by the protein's presence? Could we be going after the wrong thing?

No one doubts the peptide is there. But is it responsible for the slow, undignified decline of the brain's function? Even though he believes the hypothesis, Bart De Strooper from the Catholic University of Leuven in Belgium acknowledges that there's a disconnect. "What's lacking is a clear link between amyloid and neuronal toxicity," he says.

If A isn't to blame, then the proposed treatments could well be worthless (see "In the works"). Or worse. What if A is actually shielding the brain as it fights the real killer of neurons? Getting rid of the peptide could actually accelerate the disease.

This worrying debate has been fuelled by speculation about why several patients developed dangerous side effects during one A-busting trial last year, forcing an abrupt halt to the research. Researchers are still waiting for the full explanation. It might prove an innocent one, and the strategy of ridding the brain of A will be vindicated. On the other hand, proposed treatments for Alzheimer's may need a radical rethink.

To be sure, there's plenty of evidence that A plays an important part in the disease process. First of all, clumps of protein in the body tend to be associated with other kinds of poor health. People with diabetes, prion diseases and amyotrophic lateral sclerosis - a type of motor neuron disease - are all proof of that. More specifically, a small number of Alzheimer's patients have mutations on chromosome 21. This is where the gene for the amyloid precursor protein or APP, from which the amyloid peptide is made, is found. The mutation seems to increase the amount of amyloid their bodies produce.

Down's syndrome is caused by an extra chromosome 21 and, interestingly, all people with Down's eventually develop amyloid plaques in their brains. And mice that have been genetically engineered to overproduce APP invariably develop brain plaques, and experience some learning and memory problems that seem to go away when the amyloid is removed.

Mutations in two other genes, presenilin-1 and presenilin-2, are also associated with Alzheimer's. Both have a hand in overproducing the longer, most clump-prone form of amyloid beta, A1-42. Another Alzheimer's gene produces a protein called ApoE4. The Apo family of proteins appears to play a role in clearing out amyloid plaques, and the E4 form isn't very good at it.

But after almost two decades in the limelight, the theory still has some serious shortcomings. For one thing, lots of people get dementia without plaques, and lots of others develop plaques without dementia. Even in patients who clearly have Alzheimer's, the number of plaques does not correlate well either with neuron loss or the severity of their mental decline.

In fact, an intriguing 1991 study by Dennis Dickson, a neuropathologist now at the Mayo Clinic in Jacksonville, Florida, left Alzheimer's researchers scratching their heads. He and his team followed 14 elderly people for several years until their deaths. None showed signs of Alzheimer's, but about half were considered "high functioning" - minds as sharp as tacks - while the others suffered a slight decline in their mental faculties. During autopsy, Dickson observed a clear divide. Some had numerous plaques in their brains, the others had few, if any. Surprisingly, however, the ones with the plaques had all been in the high-functioning group.

Did the plaques indicate early-stage Alzheimer's, caught before the patients had suffered any ill effects, or could the plaques have been protecting the brains of the high-functioning individuals? "I think the jury's still out," says Dickson.

Another weak point in the hypothesis is that mice with plaques don't suffer the same fate as people with plaques. Rodents never develop Alzheimer's naturally. And while mice genetically modified to overproduce amyloid, used to test potential drugs, do develop the characteristic plaques and perform a little worse on tests of memory and learning, even when their brains are packed with plaques they don't cause anything like the severity of the human disease. "There was so much amyloid and so little neurodegeneration," says De Strooper. "Even me, being a believer...I had a big problem with that."

When drug giant Eli Lilly used the mice to test the effects of an antibody against A, they too discovered something odd. Steven Paul of Eli Lilly in Indianapolis says that they found, to their surprise, that they could reverse the memory and learning deficits very quickly without any reduction in A deposits or blood levels of amyloid. Longer-term administration of the antibody did dissolve the plaques, but the researchers admit that the immediate results are extremely puzzling.

The theory also leaves a few basic questions unanswered. For instance, APP is found in virtually all our tissues and cells. What does it do in a healthy individual? No one knows. If amyloid is inherently bad, why do we produce it? Again, no answers. Also, what exactly is killing neurons in Alzheimer's? If it's the plaques, how do they wreak their havoc? That's anyone's guess.

Some researchers think all these problems add up to a duff theory. Even pro-amyloid researchers admit that they are eagerly awaiting some clinching evidence to show that taking the peptide out of a human brain is beneficial.

The most promising test of the hypothesis so far should have been a clinical trial conducted by pharmaceutical companies Elan and Wyeth. They tested a vaccine, known as AN-1792, which consisted of a synthetic form of the full peptide. The idea was to provoke an immune response that would chase natural A out of the brain. In earlier animal studies, the vaccine not only reduced pre-existing deposits in the brains of older animals, but also prevented A deposition in the younger ones. But 15 of the 360 human volunteers who were given the vaccine developed serious brain inflammation and the trial was called off in March last year. To the irritation of many academics, the companies have revealed very little about what went wrong.

Elan has confirmed that at least one person who participated in the trial has since died, of apparently unrelated causes. Rumours are circulating that upon autopsy, no amyloid whatsoever was found in the brain.

This could be strong support for the amyloid camp. "The suggestion would be that the general premise may have worked," says Larry Sparks, an Alzheimer's expert at the Sun Health Research Institute in Sun City, Arizona. "But they would be crazy not to report the data if there was clearance of A," he adds.

Elan may not have released that information, but it certainly hasn't lost faith in the overall approach. It has other vaccination strategies in the pipeline, such as inoculating sufferers with fragments of the peptide rather than the whole molecule, or just injecting the ready-made antibodies. There is also talk of adding anti-inflammatory drugs to the mix. "The story is far from over," says Dale Schenk of Elan.

But Mark Smith, a neurobiologist at Case Western Reserve University in Cleveland, Ohio, thinks there could be another explanation for the Elan vaccine's adverse effects: far from being the cause of the disease, A could be fighting it. "The one thing that nobody is raising is that removing amyloid is bad," he says. "No one is even considering that."

He invokes his favourite analogy. A doctor is called to the top of Everest to look into why so many mountaineers get sick up there. Blood tests reveal that they are producing more red blood cells than normal. Back in the lab, doctors genetically alter mice to overproduce red blood cells and, sure enough, they get sick. They bleed them, and the animals get better. But when sick mountaineers are bled on Everest, they get even sicker. Needless to say, the extra blood cells were an adaptation to low oxygen levels. In other words, care must be taken not to confuse the symptoms with the cause. "I think that's the situation with Alzheimer's disease," says Smith. "If you remove amyloid you will make the patients worse. The 'better' the treatment, the worse it will make the patient."

He suggests that A may actually be protective. This idea fits well with his own findings of an inverse relationship between the amount of amyloid in a brain and signs of "oxidative stress" - the more amyloid, the less damage seems to have been caused by highly reactive molecules called free radicals. He thinks that oxidative stress happens early on in the disease process, years if not decades before dementia sets in, and that amyloid could be playing a role in managing it. For instance, following head trauma or stroke, deposits of A rapidly form. Smith thinks amyloid might be a kind of brain scab, and that the most important factor in Alzheimer's is the way ageing increases the brain's sensitivity to harm, including that from oxidative stress. But the amyloid camp point out that he has little direct evidence.

Like Smith, however, Glenda Bishop and Stephen Robinson at Monash University in Clayton, Australia, argue that A is harmless and actually contributes to normal brain function and aids in recovery from brain injury. It does this, they suggest, by binding to toxic agents in the brain - such as metal ions and excessive amounts of certain neurotransmitters - and clumping them into plaques for easier removal from the brain. They think A clumps are a consequence of Alzheimer's, not the cause. Alzheimer's, they speculate, is a disorder in which the brain's defences - including A - are overwhelmed.

Ashley Bush at Harvard University agrees that A appears to be mopping up metal ions. Several years ago, Bush's team found that in the test tube, adding metals such as copper and iron made otherwise soluble A molecules clump together. They also found that these metals were present in human A plaques at levels much higher than would normally be found in a healthy brain.

Bush claims that amyloid plaques are not themselves responsible for killing neurons. Instead, he proposes that when soluble A binds with iron or copper in the brain, it clumps together and begins to act as an enzyme that churns out the powerful oxidising agent hydrogen peroxide. And it's this that kills off neurons (New Scientist, 26 August 2000, p 36).

In collaboration with Colin Masters at the University of Melbourne, Bush is now testing a disused antibiotic called clioquinol on Alzheimer's patients. The drug should work as a "chelator", removing excess metals from their brains. Results from his phase II clinical trials are due out any day now, but he has already confirmed that the treatment looks promising, slowing the progression of the disease (New Scientist, 3 August 2002, p 14). Bush's theory doesn't totally discredit the amyloid hypothesis. In fact, the more success he enjoys, the more the amyloid folk are eager to consider him one of their own.

But some amyloid aficionados are beginning to question whether plaques are the main problem after all. Dennis Selkoe at Harvard University recently proposed that short pieces of soluble amyloid may pose the more serious danger. Correlations between levels of soluble A and neuron loss or dementia are stronger, he notes. In his lab, he was able to show that the short strings of A, at levels found in humans, were themselves able to interfere with learning and memory in rats (Nature, vol 416, p 535). If he's right, many of the proposed anti-Alzheimer's treatments, including both Bush's and the Elan vaccine, may be treading on dangerous ground by breaking down plaques and raising levels of the soluble protein.

Mark Pepys of the Royal Free and University College Medical School in London is not convinced. He has identified a compound, dubbed CPHPC, that removes a substance called serum amyloid P (SAP) from the blood. SAP is thought to stabilise amyloid deposits by protecting them from digestion. CPHPC has not so far been tested in Alzheimer's patients, but is currently being tested in 30 patients with a disease called systemic amyloidosis, in which amyloid clogs up various organs (New Scientist, 18 May 2002, p 16).

The hope is that, when the drug is eventually tested on people with Alzheimer's, it will break down the plaques, flushing soluble amyloid into the bloodstream from where it will be removed by the liver. Pepys believes this will halt or even reverse patients' decline. The idea that dissolving amyloid will make things worse is wrong-headed, says Pepys. "There's no evidence that amyloid is toxic, or that getting rid of amyloid is bad for you." The body does it naturally all the time. He stresses that the patients with systemic amyloidosis in the trial, who have been treated with CPHPC for over a year now, have experienced no ill effects. "It's very, very reassuring."

So is amyloid guilty or not guilty? The jury will soon be in. "We will know in five years," predicts De Strooper. Many of the proposed therapies for treating Alzheimer's assume that amyloid plaques are behind the disease. If this is so, patients will surely get better. If they don't, it will be well past time to start listening to the naysayers. Says Pepys: "The beauty of science is that what's true always comes out in the end."

In the works

Cholesterol-lowering drugs

High cholesterol levels increase the risk of developing Alzheimer's. Cholesterol may encourage your body to produce A from amyloid precursor protein (APP), whereas low cholesterol favours a different APP-processing pathway. Cholesterol-lowering drugs appear to provide some protection. Clinical trials are under way.


Vaccination with the peptide seems to prompt an immune response that not only prevented A from depositing in the brains of young mice but cleared it from the brains of older mice. But there were side effects in the first human trial. New strategies may now focus on vaccinating with smaller peptide fragments or with antibodies.

Anti-inflammatory drugs

Long-term use of nonsteroidal anti-inflammatory drugs significantly reduces the risk of Alzheimer's, perhaps independently of their effect on inflammation. NSAIDs such as ibuprofen and indomethacin seem to inhibit production of the most plaque-prone kind of amyloid, A1-42.


A tends to clump together when it binds to metals, so "chelating" agents, which mop up metal ions, may dissolve A plaques from Alzheimer's brains. The antibiotic and chelating agent clioquinol has been shown to decrease A deposition in animals, and is now in clinical trials.

Secretase inhibitors

Beta secretase enzymes cleave APP to produce A. One, known as BACE, is considered a drug target. Mice that have been genetically engineered to lack BACE appear to be anatomically and physiologically normal and don't produce A. Of course, just because mice born without BACE are normal doesn't mean everything will be fine in adult human brains.