If not beta-amyloid, then what?

f you or a loved one has Alzheimer’s disease (AD), you already know it is a dark time. Researchers know it as well. Despite 251,069 scientific papers on the subject, no one still has a clue what causes it or how to treat it.

Now, Cochrane Reviews, which does respected meta-reviews that institutionalize what everybody in the field already knows, has pronounced the beta-amyloid (Abeta or Aβ) theory dead.^[1] It’s not paywalled, but the abstract alone tells you the story:

The 17 studies assessed seven different amyloid-beta-targeting monoclonal antibodies: aducanumab (n = 3), bapineuzumab (n = 4), crenezumab (n = 2), donanemab (n = 1), gantenerumab (n = 4), lecanemab (n = 1), and solanezumab (n = 2). All used placebo as a comparison. Eleven studies lasted 18 months, four lasted 24 months, and two lasted more than 24 months. All studies were funded by the pharmaceutical industry.

Compared to placebo, amyloid-beta-targeting monoclonal antibodies probably result in little to no difference in cognitive function.

It’s been known for some time that antibodies have no clinical benefit for severe AD, so companies focused on early, mild AD or cognitive impairment. Now that hope is gone, and the Aβ theory is to blame.

But there’s a catch: we need something to replace it with. Global warming, PFASs, air pollution, and microplastics are not going to do the job. Without an alternative, the failed Aβ theory could last forever.

Alas, finding an alternative is not Big Pharma’s job, as I was reminded when I once accident­ally used the word ‘research’ to a recruiter. I was reminded that their task is not to do research but to take sure things from the academic literature and develop a product on it. If it works, they then do process research to improve it. If it doesn’t, the manager who made the decision has somebody—academic researchers—to blame.

Evidence for and against Aβ

1. Aβ is a peptide that comes in several forms. The larger form, Aβ42, is increased in the brain of AD patients. It is decreased elsewhere in the body. This Aβ40/42 ratio was once proposed as a diagnostic test for AD: if it increased in the blood, the test would be positive. Unfortunately, the test was unreliable. Aβ42 is hard to measure chemically, and MRI studies show wide variation among patients: some have it and some don’t. Nonetheless, the Aβ theory became so dominant that many people concluded that those without it did not have AD.

2. The normal function, if any, of Aβ is still unclear. One theory was that Aβ does something beneficial, so depleting it might be harmful. The fact that Aβ immunotherapy often causes ARIA (brain swelling) doesn’t prove that, because no antibody is perfectly specific. The massive doses of it needed to have any effect could well be causing side effects by undetected binding to additional targets, or simply from a reaction to the antibody itself.

3. No other species gets AD. Not mice, rats, rabbits, dogs, or any other primates. This means it’s impossible to test any theories directly. All we have are transgenic mice that are engin­eered to produce large amounts of Aβ in their brain. Unsurprisingly, that makes them sick and shortens their lifespan. But they don’t get AD.

4. The reason Aβ would cause AD is also unknown. There are only hypotheses: maybe it’s toxic, maybe it chemically produces oxidative stress, maybe it chelates essential metals. The most that can be said is that AD might a form of amyloidosis. This is reasoning by analogy: ‘amyloid’ is a functional definition, and there are many other amyloids, all different. Most are considered immune disorders.

5. Big pharma used its immense clout in enormous trials of thousands of patients. The results were not statistically significant, but anyone could make a 30 second calculation to see that doubling the population would have made it significant. So that’s exactly what they did.

Fraud?

As often happens, there are also accusations of fraud in AD research. I’ve examined many cases of purported fraud, including one lab that had five papers forcibly retracted by journals and several more in the process before they contacted me. It turned out that their problem was not fraud, but lack of basic knowledge of how to do scientific imaging and data preservation, coupled with inept attempts to defend their own results and a rush by the university to protect its reputation by sacrificing the accused laboratory and firing as many people as possible.

Indeed, in most cases I’ve seen, the accusers were even more incompetent than the researchers. In one case, the accuser looked at a re-stained blot (where the investigator washes off the first antibody and probes it with a second one) and claimed the two images didn’t line up, so it must be fraudulent. In fact, the investigators had simply rotated their blot by two degrees when they put the blot back on the imager. This doesn’t even qualify as an error: it’s nearly impossible to line up a little PVDF membrane under a camera exactly.

Many cases of ‘fraud’ are actually a combination of four factors: incompetent scientists, incompetent accusers, competitors seeking to advance their own careers, and institutional cowardice. Papers are rapidly turning into legal documents. This means scientists will have to use GLP, which is the form of protective science used in industry. It is inflexible, slower, and more expensive than today’s research.

This doesn’t mean there are no fraudulent results. But misinterpreted results and miscalc­ulated stats far outnumber fake results. Indeed, one could say that calling a transgenic mouse a case of AD is fraudulent. But everyone recognizes that it’s just a verbal shorthand. And maybe that’s the problem. Reasoning by analogy and using proxies always gets you in trouble.

If not Aβ, then what?

All those papers advising dietary or lifestyle changes to prevent AD are meaningless because we still have no idea what causes it. Most of them invoke brain inflammation, which happens in AD. So there’s one theory that a virus, maybe HSV1, becomes reactivated when a person reaches old age. Another theory is that since AD starts in the entorhinal cortex, which is near the nose, that perhaps a virus could travel from the nose to the brain and trigger AD. Another hypothesis is that air pollution could cause inflammation that could somehow affect the brain.

Another theory says that the blood-brain barrier gets clogged up, preventing the brain from eliminating toxic metabolites (which might, the researchers always say, include Aβ). Or maybe phospho-tau, which is now used as a biomarker, might be toxic.

These ideas are all worth considering, but they’re also all dead ends because there is no animal model. Many researchers simply dismiss any study that uses a transgenic mouse model. We obviously can’t test humans, so there are only two viable strategies:

1. Try treatments with random drugs based on a wild guess, preferably repurposed drugs that have been shown not to kill too many people.

2. Look for correlations. For that, people always go with the most fashionable new technology. For a time that was GWAS, which correlates the expression of genes with the diagnosis. GWAS failed because the only consistent gene was APOE, which we already knew was involved. Nowadays, they use single-cell GWAS and more sophisticated DNA techniques, but there have been few interesting results.

Inflammation

Even inflammation isn’t certain. Cell death, as occurs in AD, itself can cause neuro­inflam­mation by releasing cell contents such as DNA and RNA. And it remains to be explained why most people have these viruses, but only a few get AD. Epidemiological studies don’t see a correlation with urban vs rural, or even meaningful differences among states, which seems to rule out pollution or some other environmental cause.

Without a viable alternative, the so-called Aβ Mafia will keep publishing paper after paper insisting that Aβ is still the culprit. Even papers that have nothing to do with Aβ are forced to add a paragraph speculating how Aβ could be involved. If they don’t, they don’t get published. Since Aβ is a ‘safe’ topic, many labs just continue to grind out papers with Aβ mice, hoping to discover something worthwhile.

Funding

Foundations like the Alzheimer’s Association (AA) long ago stopped accepting applications that assumed the Aβ hypothesis. But AA, ADDF, and all the other foundations are dwarfed by NIH. And that’s where the roadblock lies.

NIH still uses the top-down funding model: one bureaucrat rises to the top at CSR (Center for Scientific Review) (though not necessarily by killing off the competition as some people think). He or she then decides what NIH will and will not fund. This drives people with novel ideas out of the field.

In fact, NIH has essentially given up. At the moment, there are only 16 funding programs for AD, mostly for tangential or trivial topics: bioinformatics, discovery of new pain targets, and ‘health disparities’ (for which there are four programs), and a few looking for neuropsychiatric symptoms. Remarkably, NINDS (National Institute for Neurological Disorders and Stroke) does not have a single program for AD in which they are the issuing organization and only 2 R01s (one on sleep and one on ‘dissemination’) where they even participate. (R01 is the only NIH grant big enough to be worth spending the 4–6 months needed to write it.)

NIH’s top-down model tells us that the government simply doesn’t trust scientific researchers to discover things. Until a superior theory comes along, the Aβ theory cannot die. Until funding agencies once again start funding basic research, a new theory will not happen. And for current patients, that will be a death sentence.

apr 17 2026, 8:44 am