Re: the Statnews piece about stopping holding oversight board meetings --
One of the things that is extremely odd, but true, about living in DC is that you realize there are shockingly few "close the loop" mechanisms. A surprisingly large number of Congressionally-mandated reports just aren't done because the person responsible retired, and if they're done, no one reads them (see, e.g., https://www.washingtonpost.com/sf/national/2014/05/03/unrequired-reading/). And there's no central reporting, as far as I'm aware, of board meetings occurring.
Perhaps this is because these reports and board meetings aren't actually (unfortunately) about the content, but just about the gesture, so that's on net good. But then we're just saying that we're incentivizing the production of empty gestures.
Alternate explanation for the fact that knowing more about Covid correlates with being more worried about Covid: the modal person who knows something about Covid knows it from watching/reading the MSM (rather than following data directly), which has been (sometimes deliberately and maliciously) promoting maximalist Covid fearmongering for years now. If Covid is constantly framed by your major news source as an "end of the world" level pandemic, you're more likely to believe that.
As for the Alzheimer's fraud: my trust in science was already pretty low in January 2020 due to the replication crisis, and over the course of the past two and half years it has reached ~zero for anything from the past 30 years or so, unless it has been turned into engineering. The rot is too deep.
What I find fascinating is that people who obsess over COVID never seem to know the probability of a bad outcome for themselves. It's fairly easy to find the odds of dying from COVID, and long-term illness seems uncommon among otherwise healthy people who are vaccinated. For whatever reason, the people who freak out the most understand probability the least.
I think it's a larger problem with the idea that trade-offs exist. If you want to "solve" COVID many other things must suffer. For some reason, many people cannot understand that anything but COVID matters at all. It's a single axis, either less COVID or more COVID with no consideration of cost. Also, they tend to favor solutions that require coercion of other people. Better vaccines and drugs would be the best solution, but they wouldn't require using power to control other people. This is why Paxlovid is ignored or criticized instead of being praised.
I still get crap about taking Paxlovid and having a two-day COVID illness. I was supposed to have a rebound, and didn't. I was supposed to have less immunity as a result of using an antiviral, but my last screening test showed more antibodies than I had from vaccination. The drug won't work in a year. So what? I care about now. It's all total crap. We use drugs all the time. There's resistance to the idea that COVID is another problem conquered by technological innovation. That's how we solve everything. Why would this be different?
TNG is an odd choice for this given that they have the ability to cure basically everything including aging whenever they feel like it, and to use various techniques to make humans far more capable and they then... don't, declaring it immoral, while frequently risking the ship or galaxy for an individual. So there's that.
At least they meet around the conference table to argue about it without anyone leaking the discussion to Twitter and ending the career of anyone who disagrees.
The Alzheimer's fraud seems very bad, but it's impact/importance in the field/research seems to have been dramatically overblown by the media, see here:
One interesting tidbit of the alzheimers story which you did not touch on here. Schrag's discovery was a direct consquence of suspicious truth-seeking short sellers investigating Simufilam. See the quote from the Science article below:
While Schrag is rightly being praised for his role in the discovery, the real heros here are the short sellers here who led the process of uncovering this fraud (and who seemingly used Schrag as an independent expert to verify their own findings/suspicions). You can find their 42 page research report submitted to the FDA in Aug 2021 (under the cover of an attorney) here: https://www.regulations.gov/document/FDA-2021-P-0930-0004
It's worth mentioning that for items like this the FDA has a public commenting process whereby interested parties can submit their thoughts on such matters for consideration. From a cursory review, the comments are unsuprisingly filled with fda lawyers and medical professionals ridiculing the fraud of the short sellers and their motivated research. Go figure. (posted a couple below for fun, don't mean to be too hard on the specific commenters, since their assumptions here may be usually correct, but I also have to imagine that some of them were themselves making use of motivated reasoning...
Last thing I'll note is that in addition to the obvious importance of this discovery, the RAT/EA community should be thrilled that this discovery is in effect maybe the most important real world discovery from 'prediction markets'. For those unfamiliar with the financial markets for early stage biotechs, they are really the closest thing to scaled prediction markets out there. For single-drug companies that are in clinical trials, investors are literally making a asymmetric probabilistic bets on the outcome of trial success, and so as one might suspect the best investors in the space tend to be teams of PhDs conducting their own analysis to derive a P (success) and then betting if their expectations are different than the market. Happy to elaborate further on specifics here when I have time if people are interested.
So who are the short sellers that deserve our praise? As noted in the article, the two short sellers are neuroscientists who submitted the complaint anonymously through an attorney.
Using regulatory databases and some google-fu, it seems rather likely that the initial short seller (or one of them) was Granite Point Capital.
If we using an SEC filing aggregator we can see that three funds had large net put short positions in Cassava in 2H 2021, when the complaint was first filed. These are LMR Partners, Perceptive (one of the very best biotech focused funds), and Granite Point.
Granite Point is smaller, with only ~$500M in AUM, is seemingly biotech focused and active in aggresive short selling (just look up their litigation). A look on linkedin at their partner group shows multiple PhDs/MDs.
Most notably, I stumbled upon this 2019 class action where Labaton, the niche law firm firm filing the cassava, is representing Granite Point, which seems to make it pretty likely that Granite were again working with Labaton in their 2021 complaint against Cassava. Its very possible that they were part of a syndicate or that there was an independent short seller pitching this to multiple funds, but the labaton tie seems to make it very likely they were involved.
"In August 2021, Matthew Schrag, a neuroscientist and physician at Vanderbilt University, got a call that would plunge him into a maelstrom of possible scientific misconduct. A colleague wanted to connect him with an attorney investigating an experimental drug for Alzheimer’s disease called Simufilam. The drug’s developer, Cassava Sciences, claimed it improved cognition, partly by repairing a protein that can block sticky brain deposits of the protein amyloid beta (Aβ), a hallmark of Alzheimer’s. The attorney’s clients—two prominent neuroscientists who are also short sellers who profit if the company’s stock falls—believed some research related to Simufilam may have been “fraudulent,” according to a petition later filed on their behalf with the U.S. Food and Drug Administration (FDA)."
Sample Comments from FDA Complaint
"This is a self serving, deceptive and destructive petition that is not filed on behalf of the human beings this drug is intended to benefit, it is filed on the behalf of one or few short sellers who have gained financially at the expense of investors and future patients.
The petition had made reference to a “whistleblower” which is typically used to refer to someone inside the company. This “whistleblower” was later revealed to be someone outside the company that held a short position in the stock. The scientific queries within this do not seek the truth, they are poorly researched and a scapegoat for this unethical practice of stock manipulation."
"FDA please act expeditiously and firmly deny this abusive and collusive CP against Cassava Sciences by Labaton Sucharow. Already the Journal of Neurosciences have confirmed Cassava’s foundational research paper and found no evidence of manipulation and only 1 human error that doesn’t impact the scientific conclusion. Also, Yale researchers have recently confirmed PTI-125’s moa and affinity binding in mice studies for hard to treat seizures. No patients have been harmed by PTI-125 and you are already aware of the breakthrough improvements AD patients are experiencing. You must not allow any further delay of your decision bc there is an obvious short and distort campaign brought on by bad actors in collusion against Cassava. I believe the SEC is in the process of looking into the matter as many investor complaints (including mine) have been brought to the office of Enforcement. Cassava’s phase 3 SPA trials must not be further impacted."
Turns out a couple of short sellers beat the Journal of Neurosciences. Shows where the real truth-seeking analysis is being done.
I never thought about short-selling companies in clinical trials for amyloid-based cures for AD but in retrospect I should have been doing that since 2002, when I got first involved in neurodegeneration research and even then strongly doubted the amyloid story. Silly me.
I would have made a lot of money in these twenty years....
The whole field of amyloid research in the so-called sporadic Alzheimer's disease was an idiotic boondoggle right from the start. The fraud is just the icing on the cake.
All of the mutations that cause precocious Alzheimers are on the amyloid system. Amyloid plaques are highly correlated with the disease in animal models and in humans.
The amyloid hypothesis seems to be missing something, because just reducing amyloids doesn't fix the problem. But there's very little doubt that amyloids are important to the etiology of Alzheimer's somehow, and we only suspect they're not causitive because of the extensive set of drug failures on them.
It's likely that the subfield has kept going a few years too long on inertia, but the reasons for suspecting amyloids were and are quite strong. (The faked data doesn't change the picture that much; it was at most supplemental evidence that amyloid was important.)
Well, FAD, the familial dementia that often starts when the patient is in his thirties and is often associated with cerebral hemorrhage and is very rare is indeed caused by mutations in the amyloid gene or genes related to the processing of amyloid - but this is a very weak reason to assume that amyloid is the cause of the common senile dementia that develops at age 70 in a non-familial fashion and without hemorrhages, even if it sometimes has some amyloid. These are completely different diseases, with different progression, onset and genetics (see ApoE polymorphisms), so you cannot generalize from one to the other.
Also, there is a very poor correlation between amyloid and disease in animals. What about hibernating animals that accumulate amyloid during hibernation and dissolve it when they emerge in the spring? No, there is a poor correlation between amyloid and dementia in the elderly - a lot of the elderly will have no amyloid but will have dementia and vice versa. And no, dementia is not highly correlated with amyloid in experimental animals. Just read about the triple-transgenic mouse model of AD - three genes from three different dementias were put together in one mouse, causing neurodegeneration because one mutant gene was not enough. That this abomination was actually hailed as a great achievement in AD research is just idiotic.
This is what happens when you take a proxy for the real thing. If you define "Alzheimer's disease" as "the dementia with amyloid plaques" then yes, you will have a perfect correlation between your dementia and amyloid. If you *know* amyloid is the cause you will believe results from models where amyloid processing is sufficiently abnormal to cause cell death, even if your patients do not have the same derangements as seen in your model.
As I outlined above, there were very good reasons to doubt the amyloid hypothesis right from the start. The extensive failures of diverse approaches to amyloid clearance in the clinic, of which Aducanumab is just the latest one, should have ended amyloid research two decades ago. How can you say "the reasons for suspecting amyloids [...] are quite strong" now, after thirty years of failure?
So, I just went back and checked my memory, and I got one thing flat wrong; nothing clearly causes Alzheimer's in animals. The closest model we have is that triple-mut one. (And it's probably better than nothing, but it might not be.)
However, it seems like beta-amyloid uncontraversially causes cell toxicity whenever you engineer cell lines to produce it, and I'm not as willing to write off FAD as a different disease as you are. It still seems to me that, without clinical evidence to the contrary, I'd be betting that malformed amyloids were important to the disease process in some way, and that removing them would at least slow progression.
And that's what I meant by strong evidence in favor; that the evidence in favor largely survives the fraud, not that it survives the wretched clinical results. It clearly doesn't.
Now I agree with you that the amyloid hypothesis was a good idea in the beginning - definitely interesting enough to be worth pursuing. It's true that amyloid does have toxic effects on cells in culture and it was a reasonable guess that it could have the same effect in vivo. However, the problem was that some researchers, like John Hardy, went way beyond the evidence in proclaiming the amyloid hypothesis as *the* Alzheimer's hypothesis and they stuck to it way beyond its past due date.
I remember some time in 2002 or 2003 one of my colleagues was peer-reviewing a submission that had conclusions mildly disproving amyloid. One of the other reviewers was a prominent amyloid proponent who basically said "no way it could be true, are you sure about your data?" with an implied "are sure you want to be rejected?". The authors massaged a few numbers, came up with a conclusion mildly supporting amyloid, and this was published.
Once enough people jump on your bandwagon this is how you can generate tens of thousands of wonky research publications out of a hypothesis and a couple billion dollars.
I skimmed part of your blog post btw, and I'm quite skeptical about a core claim there. You said:
> One of the loose ends is finding mutations in the APP gene in sporadic cases of senile dementia. Yet, here a setback occurs - not only such mutations are not found, it is definitely proven that they are absent. For anybody with a background in genetics (like me), this is a major red flag - you have one condition, early-onset AD, *with* a mutation, and a similar but clearly clinically different condition, late-onset sporadic "AD", *without* the mutation. The geneticist will automatically conclude that these are different diseases with some superficial similarities, rather than slightly differing manifestations of the same disease.
Now, I'm no geneticist by training, but I've done a lot of research into canine genetics specifically and done some consulting for breeders. I don't share your intuition here, and I doubt it's correct.
Take a parallel example: the 'Cholesterol Hypothesis' for heart disease. We know that many people in the general population develop cholesterol plaques in their arteries as they age into 60s+. This is somewhat correlated with diet and genetics, but not strongly. But some people have Familial hypercholesterolemia; specific genetic mutations which raise cholesterol levels in the blood. They have early-onset heart disease with earlier plaque formation and heart attacks in their 20s and 30s. Other people have a knockout to a gene (PCSK9) which causes them to have lower cholesterol levels. They have fantasic cardiac health with less cholesterol plaque and fewer heart attacks than the general population.
I do not believe that the general population not having the gene for early cholesterol is in any way evidence against the Cholesterol Hypothesis. Rather, it's evidence in favor. People with more X develop Y degenerative disease sooner is EXACTLY what we would expect if X caused Y.
Most traits are massively multigenic. In a multigenic trait, you should expect that most people with the disease don't have any given mutation, and that some will lead to earlier and more reliable onset of the disease than others.
In any pathological process in a complex organism you will find hundreds if not thousands of distinct molecular species whose concentrations in affected tissues differ from unaffected controls. Simply noting that there is more of something in the sick people is not enough to conclude that this something is causing the sickness, although such an observation may prompt additional research. In senile dementia there are multiple other abnormalities observed - NFTs, other tau aggregates, mitochondrial dysfunction. Unfortunately the whole research field got stuck in a local maximum for 30 fruitless years when a vocal few concluded they *knew* that amyloid was the answer and seemingly everybody followed suit.
Importantly, in senile dementia there are many patients with low accumulation of amyloid plaque, many healthy controls with high accumulation of plaque and essentially no patients with mutations or even polymorphisms in the amyloid gene or related familial AD (FAD) genes, such as presenilins. This means that amyloid fails multiple Bradford Hill criteria for causality and it should have been rejected as the main cause of AD 20 years ago.
So yes, the absence of amyloid mutations in the senile dementia of the Alzheimer type is not sufficient to reject the amyloid hypothesis but it does put a large question mark over it - certainly enough to send the amyloid proponents back to the lab for more mechanistic research and certainly enough to discourage rushing into the clinic with amyloid-targeted cures.
"Big booster energy. The official stats say ~50% of the vaccinated are boosted (and 10% of them were originally J&J which muddies things) and this has it at 59%, reflecting the bias of people who respond to polls I suppose. "
This discrepancy is mostly due to the official stats being bad, not survey bias. The CDC also believes that 90% of adults have had at least one dose of the vaccine, which is very hard to believe, compared to the poll's answer of 30%, which is plausible. The CDC can't reliably identify which shots are second/third/fourth doses, so they report lots of extra first doses, making the data more or less worthless.
You should be following Elizabeth Bik for research fraud stuff if you aren't already. She's very good at catching this stuff and these fraud-detection organizations already exist.
Though a worthwhile question is "why is peer review failing at this?"
Possibly because when you read a bunch of papers critically, finding no issues, Bayes perniciously dulls your senses. Maybe the panel for obvious frauds should require forced retirement after a year of service.
Or maybe just have a red team that assumes fraud, and speculates on most likely methods for data manipulation in every piece.
Papers commonly include a hand-wavy errors and limitations section. A para on areas of possible data manipulation, or ways the author voluntarily tied their hands to prevent that would be nice. (Did disinterested parties consult on any sudden outlier exclusions? Did you lock your data after collection or register it with a third party? Did you overcollect in a way that would allow shopping for p-values?)
Some risk this would become boilerplate too, but who knows, that might quickly force preregistration and data escrow as a norm.
For the same reason why journalism is so bad: no one wants to make enemies. Academics are not known for integrity or for standing up for the truth. Science works when conflict is accepted.
I think this is a really good point. Further, those academics that are willing to be disagreeable are not consistently disagreeable in the direction of the truth, so much as toward pet rants or self promotion.
Peer review seems to fail at most things. It mostly is acting as a gate keeper. Note the comment above about how peer review caused some slightly negative paper on amyloids to be tweaked to follow the 'company line.' No new ideas allowed.
Re: the Statnews piece about stopping holding oversight board meetings --
One of the things that is extremely odd, but true, about living in DC is that you realize there are shockingly few "close the loop" mechanisms. A surprisingly large number of Congressionally-mandated reports just aren't done because the person responsible retired, and if they're done, no one reads them (see, e.g., https://www.washingtonpost.com/sf/national/2014/05/03/unrequired-reading/). And there's no central reporting, as far as I'm aware, of board meetings occurring.
Perhaps this is because these reports and board meetings aren't actually (unfortunately) about the content, but just about the gesture, so that's on net good. But then we're just saying that we're incentivizing the production of empty gestures.
Alternate explanation for the fact that knowing more about Covid correlates with being more worried about Covid: the modal person who knows something about Covid knows it from watching/reading the MSM (rather than following data directly), which has been (sometimes deliberately and maliciously) promoting maximalist Covid fearmongering for years now. If Covid is constantly framed by your major news source as an "end of the world" level pandemic, you're more likely to believe that.
As for the Alzheimer's fraud: my trust in science was already pretty low in January 2020 due to the replication crisis, and over the course of the past two and half years it has reached ~zero for anything from the past 30 years or so, unless it has been turned into engineering. The rot is too deep.
What I find fascinating is that people who obsess over COVID never seem to know the probability of a bad outcome for themselves. It's fairly easy to find the odds of dying from COVID, and long-term illness seems uncommon among otherwise healthy people who are vaccinated. For whatever reason, the people who freak out the most understand probability the least.
I think it's a larger problem with the idea that trade-offs exist. If you want to "solve" COVID many other things must suffer. For some reason, many people cannot understand that anything but COVID matters at all. It's a single axis, either less COVID or more COVID with no consideration of cost. Also, they tend to favor solutions that require coercion of other people. Better vaccines and drugs would be the best solution, but they wouldn't require using power to control other people. This is why Paxlovid is ignored or criticized instead of being praised.
I still get crap about taking Paxlovid and having a two-day COVID illness. I was supposed to have a rebound, and didn't. I was supposed to have less immunity as a result of using an antiviral, but my last screening test showed more antibodies than I had from vaccination. The drug won't work in a year. So what? I care about now. It's all total crap. We use drugs all the time. There's resistance to the idea that COVID is another problem conquered by technological innovation. That's how we solve everything. Why would this be different?
Watch TNG, people. Get with the program.
TNG is an odd choice for this given that they have the ability to cure basically everything including aging whenever they feel like it, and to use various techniques to make humans far more capable and they then... don't, declaring it immoral, while frequently risking the ship or galaxy for an individual. So there's that.
At least they meet around the conference table to argue about it without anyone leaking the discussion to Twitter and ending the career of anyone who disagrees.
The Alzheimer's fraud seems very bad, but it's impact/importance in the field/research seems to have been dramatically overblown by the media, see here:
https://twitter.com/samuel_marsh/status/1550883407441416193?t=UpvL61XeS30RAq2DC6Kxvg&s=19
One interesting tidbit of the alzheimers story which you did not touch on here. Schrag's discovery was a direct consquence of suspicious truth-seeking short sellers investigating Simufilam. See the quote from the Science article below:
While Schrag is rightly being praised for his role in the discovery, the real heros here are the short sellers here who led the process of uncovering this fraud (and who seemingly used Schrag as an independent expert to verify their own findings/suspicions). You can find their 42 page research report submitted to the FDA in Aug 2021 (under the cover of an attorney) here: https://www.regulations.gov/document/FDA-2021-P-0930-0004
It's worth mentioning that for items like this the FDA has a public commenting process whereby interested parties can submit their thoughts on such matters for consideration. From a cursory review, the comments are unsuprisingly filled with fda lawyers and medical professionals ridiculing the fraud of the short sellers and their motivated research. Go figure. (posted a couple below for fun, don't mean to be too hard on the specific commenters, since their assumptions here may be usually correct, but I also have to imagine that some of them were themselves making use of motivated reasoning...
Last thing I'll note is that in addition to the obvious importance of this discovery, the RAT/EA community should be thrilled that this discovery is in effect maybe the most important real world discovery from 'prediction markets'. For those unfamiliar with the financial markets for early stage biotechs, they are really the closest thing to scaled prediction markets out there. For single-drug companies that are in clinical trials, investors are literally making a asymmetric probabilistic bets on the outcome of trial success, and so as one might suspect the best investors in the space tend to be teams of PhDs conducting their own analysis to derive a P (success) and then betting if their expectations are different than the market. Happy to elaborate further on specifics here when I have time if people are interested.
So who are the short sellers that deserve our praise? As noted in the article, the two short sellers are neuroscientists who submitted the complaint anonymously through an attorney.
Using regulatory databases and some google-fu, it seems rather likely that the initial short seller (or one of them) was Granite Point Capital.
If we using an SEC filing aggregator we can see that three funds had large net put short positions in Cassava in 2H 2021, when the complaint was first filed. These are LMR Partners, Perceptive (one of the very best biotech focused funds), and Granite Point.
Granite Point is smaller, with only ~$500M in AUM, is seemingly biotech focused and active in aggresive short selling (just look up their litigation). A look on linkedin at their partner group shows multiple PhDs/MDs.
Most notably, I stumbled upon this 2019 class action where Labaton, the niche law firm firm filing the cassava, is representing Granite Point, which seems to make it pretty likely that Granite were again working with Labaton in their 2021 complaint against Cassava. Its very possible that they were part of a syndicate or that there was an independent short seller pitching this to multiple funds, but the labaton tie seems to make it very likely they were involved.
https://www.labaton.com/hubfs/Joint%20Declaration%20of%20Carol%20C.%20Villegas%20and%20Adam%20M.%20Apton%20ISO%20Final%20Appro....pdf
Quote from Science Article:
"In August 2021, Matthew Schrag, a neuroscientist and physician at Vanderbilt University, got a call that would plunge him into a maelstrom of possible scientific misconduct. A colleague wanted to connect him with an attorney investigating an experimental drug for Alzheimer’s disease called Simufilam. The drug’s developer, Cassava Sciences, claimed it improved cognition, partly by repairing a protein that can block sticky brain deposits of the protein amyloid beta (Aβ), a hallmark of Alzheimer’s. The attorney’s clients—two prominent neuroscientists who are also short sellers who profit if the company’s stock falls—believed some research related to Simufilam may have been “fraudulent,” according to a petition later filed on their behalf with the U.S. Food and Drug Administration (FDA)."
Sample Comments from FDA Complaint
"This is a self serving, deceptive and destructive petition that is not filed on behalf of the human beings this drug is intended to benefit, it is filed on the behalf of one or few short sellers who have gained financially at the expense of investors and future patients.
The petition had made reference to a “whistleblower” which is typically used to refer to someone inside the company. This “whistleblower” was later revealed to be someone outside the company that held a short position in the stock. The scientific queries within this do not seek the truth, they are poorly researched and a scapegoat for this unethical practice of stock manipulation."
"FDA please act expeditiously and firmly deny this abusive and collusive CP against Cassava Sciences by Labaton Sucharow. Already the Journal of Neurosciences have confirmed Cassava’s foundational research paper and found no evidence of manipulation and only 1 human error that doesn’t impact the scientific conclusion. Also, Yale researchers have recently confirmed PTI-125’s moa and affinity binding in mice studies for hard to treat seizures. No patients have been harmed by PTI-125 and you are already aware of the breakthrough improvements AD patients are experiencing. You must not allow any further delay of your decision bc there is an obvious short and distort campaign brought on by bad actors in collusion against Cassava. I believe the SEC is in the process of looking into the matter as many investor complaints (including mine) have been brought to the office of Enforcement. Cassava’s phase 3 SPA trials must not be further impacted."
Turns out a couple of short sellers beat the Journal of Neurosciences. Shows where the real truth-seeking analysis is being done.
Wow, this is really interesting!
I never thought about short-selling companies in clinical trials for amyloid-based cures for AD but in retrospect I should have been doing that since 2002, when I got first involved in neurodegeneration research and even then strongly doubted the amyloid story. Silly me.
I would have made a lot of money in these twenty years....
Here is something I wrote about AD and amyloid a long time ago:
http://triviallyso.blogspot.com/2009/07/malleus-amyloidarum.html
The whole field of amyloid research in the so-called sporadic Alzheimer's disease was an idiotic boondoggle right from the start. The fraud is just the icing on the cake.
All of the mutations that cause precocious Alzheimers are on the amyloid system. Amyloid plaques are highly correlated with the disease in animal models and in humans.
The amyloid hypothesis seems to be missing something, because just reducing amyloids doesn't fix the problem. But there's very little doubt that amyloids are important to the etiology of Alzheimer's somehow, and we only suspect they're not causitive because of the extensive set of drug failures on them.
It's likely that the subfield has kept going a few years too long on inertia, but the reasons for suspecting amyloids were and are quite strong. (The faked data doesn't change the picture that much; it was at most supplemental evidence that amyloid was important.)
Well, FAD, the familial dementia that often starts when the patient is in his thirties and is often associated with cerebral hemorrhage and is very rare is indeed caused by mutations in the amyloid gene or genes related to the processing of amyloid - but this is a very weak reason to assume that amyloid is the cause of the common senile dementia that develops at age 70 in a non-familial fashion and without hemorrhages, even if it sometimes has some amyloid. These are completely different diseases, with different progression, onset and genetics (see ApoE polymorphisms), so you cannot generalize from one to the other.
Also, there is a very poor correlation between amyloid and disease in animals. What about hibernating animals that accumulate amyloid during hibernation and dissolve it when they emerge in the spring? No, there is a poor correlation between amyloid and dementia in the elderly - a lot of the elderly will have no amyloid but will have dementia and vice versa. And no, dementia is not highly correlated with amyloid in experimental animals. Just read about the triple-transgenic mouse model of AD - three genes from three different dementias were put together in one mouse, causing neurodegeneration because one mutant gene was not enough. That this abomination was actually hailed as a great achievement in AD research is just idiotic.
This is what happens when you take a proxy for the real thing. If you define "Alzheimer's disease" as "the dementia with amyloid plaques" then yes, you will have a perfect correlation between your dementia and amyloid. If you *know* amyloid is the cause you will believe results from models where amyloid processing is sufficiently abnormal to cause cell death, even if your patients do not have the same derangements as seen in your model.
As I outlined above, there were very good reasons to doubt the amyloid hypothesis right from the start. The extensive failures of diverse approaches to amyloid clearance in the clinic, of which Aducanumab is just the latest one, should have ended amyloid research two decades ago. How can you say "the reasons for suspecting amyloids [...] are quite strong" now, after thirty years of failure?
So, I just went back and checked my memory, and I got one thing flat wrong; nothing clearly causes Alzheimer's in animals. The closest model we have is that triple-mut one. (And it's probably better than nothing, but it might not be.)
However, it seems like beta-amyloid uncontraversially causes cell toxicity whenever you engineer cell lines to produce it, and I'm not as willing to write off FAD as a different disease as you are. It still seems to me that, without clinical evidence to the contrary, I'd be betting that malformed amyloids were important to the disease process in some way, and that removing them would at least slow progression.
And that's what I meant by strong evidence in favor; that the evidence in favor largely survives the fraud, not that it survives the wretched clinical results. It clearly doesn't.
Now I agree with you that the amyloid hypothesis was a good idea in the beginning - definitely interesting enough to be worth pursuing. It's true that amyloid does have toxic effects on cells in culture and it was a reasonable guess that it could have the same effect in vivo. However, the problem was that some researchers, like John Hardy, went way beyond the evidence in proclaiming the amyloid hypothesis as *the* Alzheimer's hypothesis and they stuck to it way beyond its past due date.
I remember some time in 2002 or 2003 one of my colleagues was peer-reviewing a submission that had conclusions mildly disproving amyloid. One of the other reviewers was a prominent amyloid proponent who basically said "no way it could be true, are you sure about your data?" with an implied "are sure you want to be rejected?". The authors massaged a few numbers, came up with a conclusion mildly supporting amyloid, and this was published.
Once enough people jump on your bandwagon this is how you can generate tens of thousands of wonky research publications out of a hypothesis and a couple billion dollars.
I skimmed part of your blog post btw, and I'm quite skeptical about a core claim there. You said:
> One of the loose ends is finding mutations in the APP gene in sporadic cases of senile dementia. Yet, here a setback occurs - not only such mutations are not found, it is definitely proven that they are absent. For anybody with a background in genetics (like me), this is a major red flag - you have one condition, early-onset AD, *with* a mutation, and a similar but clearly clinically different condition, late-onset sporadic "AD", *without* the mutation. The geneticist will automatically conclude that these are different diseases with some superficial similarities, rather than slightly differing manifestations of the same disease.
Now, I'm no geneticist by training, but I've done a lot of research into canine genetics specifically and done some consulting for breeders. I don't share your intuition here, and I doubt it's correct.
Take a parallel example: the 'Cholesterol Hypothesis' for heart disease. We know that many people in the general population develop cholesterol plaques in their arteries as they age into 60s+. This is somewhat correlated with diet and genetics, but not strongly. But some people have Familial hypercholesterolemia; specific genetic mutations which raise cholesterol levels in the blood. They have early-onset heart disease with earlier plaque formation and heart attacks in their 20s and 30s. Other people have a knockout to a gene (PCSK9) which causes them to have lower cholesterol levels. They have fantasic cardiac health with less cholesterol plaque and fewer heart attacks than the general population.
I do not believe that the general population not having the gene for early cholesterol is in any way evidence against the Cholesterol Hypothesis. Rather, it's evidence in favor. People with more X develop Y degenerative disease sooner is EXACTLY what we would expect if X caused Y.
Most traits are massively multigenic. In a multigenic trait, you should expect that most people with the disease don't have any given mutation, and that some will lead to earlier and more reliable onset of the disease than others.
In any pathological process in a complex organism you will find hundreds if not thousands of distinct molecular species whose concentrations in affected tissues differ from unaffected controls. Simply noting that there is more of something in the sick people is not enough to conclude that this something is causing the sickness, although such an observation may prompt additional research. In senile dementia there are multiple other abnormalities observed - NFTs, other tau aggregates, mitochondrial dysfunction. Unfortunately the whole research field got stuck in a local maximum for 30 fruitless years when a vocal few concluded they *knew* that amyloid was the answer and seemingly everybody followed suit.
Importantly, in senile dementia there are many patients with low accumulation of amyloid plaque, many healthy controls with high accumulation of plaque and essentially no patients with mutations or even polymorphisms in the amyloid gene or related familial AD (FAD) genes, such as presenilins. This means that amyloid fails multiple Bradford Hill criteria for causality and it should have been rejected as the main cause of AD 20 years ago.
So yes, the absence of amyloid mutations in the senile dementia of the Alzheimer type is not sufficient to reject the amyloid hypothesis but it does put a large question mark over it - certainly enough to send the amyloid proponents back to the lab for more mechanistic research and certainly enough to discourage rushing into the clinic with amyloid-targeted cures.
"Big booster energy. The official stats say ~50% of the vaccinated are boosted (and 10% of them were originally J&J which muddies things) and this has it at 59%, reflecting the bias of people who respond to polls I suppose. "
This discrepancy is mostly due to the official stats being bad, not survey bias. The CDC also believes that 90% of adults have had at least one dose of the vaccine, which is very hard to believe, compared to the poll's answer of 30%, which is plausible. The CDC can't reliably identify which shots are second/third/fourth doses, so they report lots of extra first doses, making the data more or less worthless.
You should be following Elizabeth Bik for research fraud stuff if you aren't already. She's very good at catching this stuff and these fraud-detection organizations already exist.
Someone needs to scale those organizations up!
patio11 (Patrick McKenzie) is one of my favorite people to follow!
I was really sad that his last startup – Stockfighter – folded before I got a chance to play with it!
If you're not following it already, his new blog is great: https://bam.kalzumeus.com/
"Contrast this with Norway’s approach." – LOL; those lucky Norwegians!
I'd second a submission of "checking for obvious frauds" to Open Philanthropy.
https://www.causeexplorationprizes.com/
Though a worthwhile question is "why is peer review failing at this?"
Possibly because when you read a bunch of papers critically, finding no issues, Bayes perniciously dulls your senses. Maybe the panel for obvious frauds should require forced retirement after a year of service.
Or maybe just have a red team that assumes fraud, and speculates on most likely methods for data manipulation in every piece.
Papers commonly include a hand-wavy errors and limitations section. A para on areas of possible data manipulation, or ways the author voluntarily tied their hands to prevent that would be nice. (Did disinterested parties consult on any sudden outlier exclusions? Did you lock your data after collection or register it with a third party? Did you overcollect in a way that would allow shopping for p-values?)
Some risk this would become boilerplate too, but who knows, that might quickly force preregistration and data escrow as a norm.
For the same reason why journalism is so bad: no one wants to make enemies. Academics are not known for integrity or for standing up for the truth. Science works when conflict is accepted.
I think this is a really good point. Further, those academics that are willing to be disagreeable are not consistently disagreeable in the direction of the truth, so much as toward pet rants or self promotion.
Peer review seems to fail at most things. It mostly is acting as a gate keeper. Note the comment above about how peer review caused some slightly negative paper on amyloids to be tweaked to follow the 'company line.' No new ideas allowed.