What's new in Alzheimer's disease research

he news media are constantly telling us about wonderful new treatments and clever ways of diagnosing Alzheimer's disease (AD). Is any of it really true?

The anti-Abeta antibody Donanemab from Eli Lily just got through the FDA the same way the infamous Aducanumab did: with a gigantic clinical trial showing a significant reduction of Abeta in the brain after suitable post-hoc population slicing and dicing ^[1], this time in patients with early symptomatic AD. And just in time, too: those persistent reports of ARIA caused by Aduhelm keep clinicians wary that they could be trading one neurological disorder for another.

Each of these drug trials has hit the same ceiling effect: it's now clear that the best to be expected from immunotherapy is a 30% decrease in the rate of cognitive decline.

A big challenge in clinical drug trials for AD is that almost anything you give the patients seems to work for a day or so. Everyone starts to think the drug is working, but after a few days the patient goes back to their previous state. It's as if the brain is mainly reacting to some novel stimulus. Maybe there's a clue there.

We're trapped! Like rats!

That's a famous Three Stooges gag. But not even Moe ever commented on how research was being strangled at CSR in favor of the beta-amyloid theory. The biggest risk factor is still ApoE4, but NIH hasn't had a program announcement on that since 2018.

A quick look on grants.gov shows the problem. Most of the AD grants are either for clinical studies and drug trials, or for “diversity.” For basic research, there are only the old general-purpose ones, which have a low funding rate, and grants for researchers to “validate” drugs found by other screening projects.

‘Validation’ is a polite way of saying “going through somebody else's dregs.” You can't validate an AD drug unless you know the mechanism of action and a clear link to the patho­genesis. And you can't validate any AD drug in cells or even in animals because even after all these years and all those poor mice squeaking their hearts out, there is still no good model for AD in any species.

Nothing happens in science unless somebody funds it. NIH is by far the biggest funder of neurology research, and something weird is going on at CSR, the division of NIH that decides grant funding. Morale is in the pits. The officials who contact researchers to ask them to review grants have strict rules about what they're allowed to say. Reviewers also have strict rules: you can't say, for instance, that somebody's drug is likely to cause cancer, because that would kill their grant. Yet, considering the inverse relationship of those two diseases, cancer promoters might be just what we need. If I were an 80-year-old with Alzheimer's I'd trade it for the possibility of future cancer in a second. But no drug company would bother: even if it just flunks the Ames test, it gets dropped.

Unless we know what causes AD, a clinical trial is just a way of picking a chemical on a hunch, giving it to people, and seeing if it cures them. It's essentially a jobs program for doctors. That's why I ended up going to the DoD, which wants to keep its veterans alive. The best way to do that, aside from by convincing them to stay away from VA hospitals, is to discover what causes the disease. If only NIH cared more about that.

Viruses

Even those claims that AD was caused by gingivitis / periodontitis or by herpesvirus that migrated into the brain have run out of steam. When an AD patient has bacteremia, does it mean the bacteria caused the disease, that the inflammatory response to the bacteria (which goes through TLR4, a receptor in the innate immune system that's “implicated” in AD) caused it, or—more likely—that a patient who loses his or her memory also forgot to brush their teeth?

Maybe all of the above. All you can do is look for correlations. These theories can't be tested in humans unless you want to go all Mengele, or unless there's an animal model. And for all you mice out there in the audience, that means you, baby.

Diagnostics

On the bright side, diagnostics are now working. Plasma phospho-tau levels are now a generally accepted and affordable way of testing for neuronal loss, and the phosphorylation pattern is reasonably specific for AD. MRI and PET scanning are too expensive and not reliable enough, but they can confirm that the person has an advanced case. The main value is not for patients, but to make clinical drug trials cheaper. And that's a good cause.

Simufilam fraud

There's also bad news. Hoau-Yan Wang, who has been promoting an anti-Alzheimer drug called simufilam^[2], has been indicted for allegedly falsifying data to fraudulently obtain $16 million from the NIH.

According to Science magazine ^[3], Wang was charged with manipulating images of Western blots. Many of Wang's papers on simufilam have been retracted. Cassava's stock dropped 47% on June 28, the day Wang got indicted. Science mag says the company released a statement saying the CUNY lab where Wang worked was not required to be compliant with GLP standards (which is true if it's not used as evidence for safety or efficacy). The Science reporter, C. Piller, wrote “the CUNY panel concluded that 20 of his papers—some co-authored by Lindsay Burns, a leading Cassava scientist—contained evidence highly suggestive of deliberate scientific misconduct by Dr. Wang.”

That sounds bad—there are so many false accusations out there it's hard to know for sure—but if true, some of the blame lies with industry, which has consistently failed to provide biologists with a reliable way of measuring and discovering proteins.

It's almost impossible to discover a new protein using electrophoresis. You'd have to laboriously cut the band out of a gel and have it sequenced. But every band in a 1D gel may contain 100 proteins (not counting keratin and BSA, which always manage to get into the sample), so you learn nothing. Two-dimensional gels are so unreliable, expensive, and technically tedious that almost no one does them anymore.

In Western blots, you can only measure proteins that are already known and somebody has created an antibody for. So all you can do is test a hypothesis about some existing protein—one at a time. Staining the gel, quantifying it, and ensuring that exactly the same amounts of protein got loaded are all error-prone steps. And thanks to all the hysteria about image falsification, where people fix up their images by jacking up the contrast, erasing their thumbprints, and losing their originals in misguided attempts to get past peer review, the number of controls we have to do makes them impractical. ELISAs are no solution: reviewers won't accept them, and demand that you confirm the result with a Western.

So everyone is waiting for somebody else to invent a way of identifying and quantifying the proteins in their samples. Mass spec might work if it was affordable, but imaging mass spec bombed out. One company that tried commercializing it didn't have enough customers, mainly because the technology wasn't advanced enough. So instead of doing the research, which I had hoped to help them with, all they wanted to do was standard protein LCMS. Shortly after I interviewed with them, they went out of business.

The system isn't set up to cure diseases. It's set up for people to extract money from each other. Most of the time, despite the rhetoric, a cure is not even on the table. I can't remember the last time we ever had a discussion about science in academia; their meetings always turn into a question of whether NIH would like it. And increasingly NIH is like Mikey, who hates everything.

jul 07 2024, 8:54 am