Thanks. If you had a scan showing a high amyloid load but no symptoms, would you start using these drugs? Why aren’t insurers re imbursing in such cases? If you can avoid or delay pheno conversion to proper AD that would result in cost savings for them?
The first GLP-1RAs (exenatide twice-daily) was approved in 2005 but the class only became very popular with semaglutide in 2017. So aducanumab was approved in 2021 and we might have to wait 10 years as well?
It would depend on my age and other risk factors, but if I had a high amyloid load and had the opportunity, I expect I would take them.
None of these antibodies are approved for this scenario yet: they’ve only been tried in and shown effective in people with “mild” dementia or in some cases “mild” cognitive impairment (MCI), which is actually a substantial level of CI. There are trials underway to test this question and presumably get label expansions for both L and K.
Beyond the evidentiary and regulatory issues, the cost savings question is a problem of misaligned incentives: unlike in places like the UK and Canada (and to a lesser extent Germany, Switzerland, and Australia), Americans frequently change insurers in response to job changes or changes in insurance options. So the odds that Insurer A who pays for Lequembi for a sixty-year-old will still be the patient’s insurer when they avoid dementia at age 75 (and therefore reap the cost savings) are very slim — and, as noted, the antibodies are at present very expensive.
My first post in this thread was in response to this:
That’s not an apples-for-apples comparison. The semaglutide bet is to merely be FDA-approved for Alzheimer’s, which Lequembi and Kinsula have already done. The amyloid bet is for a therapy that not only works and gets approved, but slows cognitive decline by 75%.
The markets are giving 58% that Ozempic will be approved for AD. If they asked that question for amyloid therapies, the answer would be 100%, because there are already three amyloid therapies approved for AD.
They’re not going to get from 30% slowing to 75% by doing better at clearing amyloid: these therapies already clear nearly all of it out. The thing limiting their effectiveness is that by the time people were enrolled in these trials, the patients’ brains were already a mess because of all the downstream damage already inflicted by the time they were diagnosed.
To do better, they’re going to have to start a lot earlier, before that downstream damage has occurred. Trials in people who are positive for amyloid beta but are cognitively normal are underway now, but it’s likely going to take more than 12 years to show a cognitive difference that large when everyone starts the trial cognitively normal to begin with. It’s similar to the problem of showing a reduction in ACM in statin trials.
But can you see anything if you start “a lot earlier”? And how long do you need to follow people to get a statistically significant result? Big pharma won’t do trials that last more than 4y (that’s already very long!). FYI: Donanemab NCT05026866 will last 182 weeks = 3.5y.
Said otherwise: even if these drugs work, could it just be super hard and expensive to prove it? To some extent, it’s a problem similar to rapamycin and all longevity trials.
Which is why there’s a search for biomarkers which could be proxies for the process. If we establish firmly that, say, ApoB levels at given values “guarantee” no atherosclerosis, then all we have to show is that a given drug brings ApoB levels to that given value and we don’t have to wait for 20 years to show atherosclerosis prevention. Obviously, that has all the disadvantages of biomarkers as proxies, questions of drug tolerability and long term side effects etc., but you release the drug into the wild and wait for the clinical data to gather over the years (statins).
If amyloid and tau tangles prevention can definitively guarantee no ND disease (say, AD), then perhaps some biomarker(s) can be a proxy. Then you show that the drug candidate affects the biomarker etc.
What biomarkers or measurements can act as such proxies for any given NDD?
How can we measure aging? If a given drug (say, rapamycin) affects those key markers then we don’t have to wait a whole human lifespan to assess effectiveness, because we can show impact on those proxies. We first validate those proxies in animal models, and then see if there’s a similar effect in humans. That’s what Fontana et al tried to do in the CALERIE study for the effects of CR in humans. We have a bunch of biomarkers associated with CR in animal models. If we can now show that initiating CR in humans gives us a similar biomarker profile as in animal model CR, then we have a leap of faith that there will be a similar longevity outcome. It’s the best we can do, because nobody is doing a lifelong controlled study in humans. On the other hand, we have the Framington multigenerational CVD data etc., so observational approaches have been used with decent confidence - based on this, recommendations have been generated. Maybe something similar could be tried.
You start with animal models. First validate it there, then do trials of the candidates in humans. We don’t have good animal models for AD to begin with, the mice models we are using are shit. Even then, when it comes to amyloid it’s chicken and egg.
Chinese, but Beijing:
Animal models of Alzheimer’s disease: Applications, evaluation, and perspectives
Your last statement is exactly why I assume there’s skepticism of getting a drug proven to improve cognitive decline by 75%: not the science, but the logistics, incentives, and timelines. " it’s a problem similar to rapamycin and all longevity trials" — and ACM for statins, as I pointed out.
However, I do think we will get this answer eventually, and in the positive. In 3.5 years, I expect that they will be able to prove that they can clear amyloid beta out of the brains of cognitively normal people with minimal side-effects. Then they can observe them without further intervention (or, at their option, invite them for further treatment on an open-label basis, possibly on a less frequent dosing schedule since they’ve already roto-rootered their brains) for 5-15 years more and see if it substantially reduces the rate at which people transition into MCI or dementia, which (a) would likely allow them a license extension, and (b) might already qualify as a 75% reduction in cognitive decline and certainly would if sustained for another few years.
As CronosTempi notes, they might try for a conditional license extension earlier on the basis of the biomarker change plus safety, analogous to what was previously done with Aduhelm but on more solid footing since their Phase 3s unambiguously succeeded. (CT: Aduhelm was approved in substantial part on the basis of having reduced brain amyloid beta. We now have additional, well-validated serum markers, such as ptau-217).
Aducanumab was abandoned. We still have donanemab and lecanemab. They’re doing what you’re suggesting:
NCT04437511: “Following the double-blind 76-week main study period, a double-blind 78-week long-term extension period is added to further evaluate donanemab efficacy and safety over time.”
NCT05026866: Donanemab: “Approximately 800 additional participants will be enrolled in a 12-month addendum to assess safety of a different titration regimen.” (that’s the one cited by the author when he writes: “On the other hand, if a clinical trial completes earlier than 12 years from now (perhaps [73], reading out in 2027), sustains extremely good amyloid clearance at the preclinical stage, and has a good safety profile, but doesn’t make substantial progress towards this 75% goal, then I would consider this prediction refuted in advance.”)
NCT05508789: “The study duration including screening and follow-up is up to 93 weeks.” in “prodromal AD and mild dementia due to AD”
NCT06566170: “The study will last about 273 weeks and may include up to 28 visits.” (5 years!)
NCT06889818: 10y follow up
NCT03887455: Lecanemab, 69 months (almost 6y)
NCT04468659: Lecanemab, 216 weeks (4y)
These drugs clear amyloid perfectly, yet prediction markets are skeptical about their effectiveness: why?
My other concern is that even if we manage to prove that taking these drugs early slows down cognition decline by >75%, what about other effects? If they increase ACM, then who would take them? If they decrease ACM then they would be amazing longevity drugs but we would see lifespan extension in mice if that was the case. That’s why targeting tau might be more promising, even if more downstream than amyloid: you can intervene later and act when people are already somewhat impaired so that you improve their quality of life and ACM matters less. (And it costs less to prove the efficiency with shorter trials…)
From the summaries you gave, none of these studies are doing what I’m suggesting (though I expect some of them will if they’re successful as I defined above). Most of of these studies last three years or less. NCT06566170 is in “Participants With Early Symptomatic Alzheimer’s Disease”. NCT06889818 is an observational study, not a trial, and is in people with AD. NCT03887455 is the extension study of the already-completed Phase 3 for Lequembi, whose subjects had AD. NCT04468659 is the right kind of study, but only 4 y (though I expect they will ultimately do what I suggest, if successful as I defined above).
To avoid AD, I actually expect a lot of people would take them. But in any case, there’s no reason to think they would do this.
… if anyone had ever done a lifespan study with them, which no one has …
Actually, you wouldn’t expect a LS increase from these drugs in wild-type mice, because mice don’t naturally accumulate amyloid beta in their brains. You could do it in AD-transgenic mice, but I don’t think many people would take a “life extension” in such mice seriously (and they shouldn’t).
I do think it’s causal — newborn infants have apoB, but it’s still causal for ASCVD (yes, I know their levels are low) — but it doesn’t matter whether it’s causal — just that it’s predictive as a biomarker, which it is.
But often on top of previous studies of ~2y so aligned with the 5y goal you mentioned.
Aren’t there long-term carcinogenicity studies in mice before approval? I remember those for dapagliflozin and empagliflozin that we discussed in the canagliflozin thread. Not a lifespan study but gives some hint.
None of the relevant studies as defined are on top of anything. The studies in cognitively-normal people positive for beta-amyloid are only now starting from scratch. And I suggested we would get a positive result if a 3.5 year trial were followed by an additional 5-15 years of open label, so that’s 8.5-18.5 years, not 5, 3, or 2.
“Gives some hint” in a 2-y carcinogenicity study is not “are amazing longevity drugs” as you proposed in the negative, arguing from silence. And an antibody designed to fix a problem that wild-type mice don’t have could conceivably even be slightly harmful to them but positive for ACM in humans.
As I said, that kind of concern “is exactly why I assume there’s skepticism of getting a drug proven to improve cognitive decline by 75%: not the science, but the logistics, incentives, and timelines.” But I do anticipate we’ll get the data.
I will note that Tanzi is one of the most prominent advocates of an infectious hypothesis for AD — and in his view, it goes through amyloid beta.
No new worrisome safety signals have appeared; rates of the treatments’ most serious adverse event have not exceeded those seen in clinical trials. “If there was a really bad safety signal, we would have seen it,” noted clinical trialist Robert Howard, MD, a professor of old age psychiatry at University College London. “Give the drugs credit where it’s due. That’s good news.”
Eisai presented data from its 48-month follow-up at this year’s AAIC. However, the company this year compared changes in cognition and function over time, again as determined by the CDR-SB, only between trial participants who received lecanemab from the start and 2 historical control groups. The delayed-start trial participants were not included in the comparison. Howard said he can’t help but wonder whether Eisai this year decided to omit the delayed-start patients from the comparison because they had caught up with the early-start patients, which would suggest that lecanemab wasn’t a disease-modifying therapy after all.
In an email, Eisai spokeswoman Libby Holman said that the company had not planned to compare the early-start vs delayed-start trial participants past 36 months.
Their manufacturers did not provide uptake data on the therapies requested by JAMA Medical News. University College London’s Howard says his sources tell him “the numbers are really, really small,” with only around 13 500 patients worldwide having been treated with lecanemab.
Whether the treatments actually extend the time patients can run errands or play with their grandchildren—the kinds of benefits that Pike of the Alzheimer’s Association referred to when the FDA granted traditional approval to lecanemab—can’t be determined in clinical practice because there isn’t a placebo comparison group, University College London’s Howard pointed out. “There is a massive placebo effect.”
Just as the safety profiles of the antiamyloid therapies in the real world appear to be about the same as in their clinical trials, Howard said, “the drugs are probably only as effective as they were in the trials, and that isn’t very effective.”
Colleagues in Japan and South Korea who have treated patients with the drugs have told him that “all we’re seeing is people deteriorating the way they do without treatment,” Howard said.
