New Dementia Trial Results

post by PeterMcCluskey · 2021-07-02T23:04:46.548Z · LW · GW · 5 comments

This is a link post for http://www.bayesianinvestor.com/blog/index.php/2021/07/01/new-dementia-trial-results/

Contents

  1.
  2.
  3.
  4.
  5.
  6.
None
5 comments

There's a new clinical trial result showing that Bredesen's approach is able to at least partially cure common forms of Alzheimer-like dementia. (Press release here). It has not received as much attention as it deserves.

The 9 month study seemed a bit less impressive than what I'd hoped for, but the outcomes still support the claim that common forms of dementia are partly curable.

Out of 25 patients, 21 or 19 improved their cognition compared to the start of the trial, depending on which measure I look at, and 2 or 3 declined.

Side effects included occasional improvements in hypertension and diabetes, enough to allow patients to stop taking medications for those conditions.

1.

Compared to my prediction in 2017, the trial happened a bit later than I expected, but achieved much higher patient compliance than I expected. If I had predicted the study size, I'd have likely said something way too optimistic (a thousand?).

The trial registration reports an actual enrollment of 30 patients, whereas the paper reports 28 were recruited, 3 of whom dropped out within 3 months. I don't see any explanation for that difference. For comparison, it looks like Biogen's phase 3 trials had completion rates of less than 60%.

The pandemic started in the middle of the trial, impairing compliance. Exercise seems likely to have been reduced. Two patients showed reversal of their improvement that was attributed to sheltering in moldy homes. The pandemic reportedly hurt patients' diets. I'm unsure what to make of that. Restaurant closures would have reduced access to food that tempts people to cheat - my diet better resembles Bredesen's when I'm cooking at home. But the effect would likely be the opposite for those who don't cook, as I'm guessing they'd eat more junk food.

2.

The study does not report ADAS-Cog results. ADAS-Cog seems to be the usual measure for clinical trials. It may be inappropriate for milder impairment due to ceiling effects - I'm unclear whether that's a reason for them to have avoided it.

They used several measures that seem to be fairly common tests of cognition, and they pre-registered those, so it doesn't look like they ran a lot of tests and selected those that favored the protocol.

3.

Their criteria for excluding people includes statin use, unless eligible to discontinue. Yet they report one patient discontinuing statin use after the treatment apparently improved the patient's lipids (I'm not 100% clear on whether that patient was in the trial).

Did they exclude many statin users, and if so did that cause them to have a more treatable population than most other dementia trials? Bredesen is pretty concerned that statins cause dementia. The evidence on this topic seems pretty weak. I'll guess that if statins were a big enough cause of dementia to impact this trial in an important way, someone would have managed to report better evidence than I've found. But I'm still avoiding my doctor's advice to use a statin, partly due to Bredesen's opinion.

4.

The trial didn't directly verify that patients had Alzheimer's. They claim that if the treatment was only working on non-Alzheimer's forms of dementia, then the 12 patients with an Apoe4 allele would likely have done poorly, which was not the case. That's not a very rigorous argument, but it would be fairly hard to accidentally produce the indicated results without improving the health of some Alzheimer's patients.

I've minimized my use of the term Alzheimer's here, due to uncertainty about whether it means much. Some authorities treat amyloid beta as a defining feature of Alzheimer's, yet amyloid beta has been largely demonstrated to have little causal influence on dementia.

The FDA has panicked recently over the failure of the amyloid beta industry, and is trying to provide hope for that industry by allowing it to market a drug (Biogen's Aduhelm) that reduces amyloid beta, as if that treated a disease. Maybe Aduhelm is better than nothing, but the FDA announcement seems to include a number of false statements that appear designed to suppress awareness of better alternatives.

5.

This trial makes me slightly more optimistic about Bredesen's protocol, mainly due to the evidence that they were able to get better compliance than I expected. That evidence is somewhat tempered by the small size, which likely means that only the most eager patients ended up in the trial.

We ought to have a higher prior for Bredesen's lifestyle protocol than for some novel drug, partly due to the evidence from other cultures which suggests that most dementia is due to modern lifestyles. That evidence has been neglected due to not fitting the High Modernist notion of what qualifies as evidence.

In addition, Bredesen's protocol is composed of parts that look reasonable to adopt even in the absence of effects on dementia. It looks like the reluctance to adopt most of those parts comes from issues such as price (much less than Aduhelm), hassle, and/or the fact that the evidence for effectiveness is confusing. How does that compare to, say, the CDC's reluctance to advocate mask use in March 2020?

6.

Two other recent trials showed drug-based promise for dementia.

Cassava Science's trial results: cognition improved 10% over 6 months. I bought stock in Cassava due to that announcement.

Annovis Bio has reported a 30% improvement in cognition in 25 days, in 14 patients. I'm still wondering what to make of that announcement. Note that their comparison to Biogen in their press release is confusingly favorable to Biogen, which reported an improvement relative to placebo, but those patients still declined. Whereas Annovis, Cassava, and the Bredesen trials show improvement relative to baseline.

I've seen three promising clinical trial results for dementia in the last 6 months. Can anyone show me prior dementia trial results that had similar promise but failed at a later stage?

5 comments

Comments sorted by top scores.