With the EMA's unfortunate reversal on lecanemab here's a condensed (anti)anti-amyloid antibody argument: 1) No correlation between amyloid removal and clinical response 2) Functional unblinding may explain much/all of the "response" 3) No effect on tau. pubmed.ncbi.nlm.nih.gov/38250779/🧪 #alzsky - ThreadSky

mikegreicius.bsky.social • 99 days ago

With the EMA's unfortunate reversal on lecanemab here's a condensed (anti)anti-amyloid antibody argument: 1) No correlation between amyloid removal and clinical response 2) Functional unblinding may explain much/all of the "response" 3) No effect on tau.
https://pubmed.ncbi.nlm.nih.gov/38250779/🧪 #alzsky

Comments

wandawitch.bsky.social•99 days ago

Thanks. Sometimes I feel like the only person who doubts about the amyloid/tau road everyone gone down. As if it's settled. Fact. Show me the mechanism and I'll rethink.
Examples of increased amyloid plaques in the absence of any symptoms just gets pushed aside

dweisman.bsky.social•98 days ago

Does HIV cause AIDS? Because there are examples of HIV+ people in the absence of symptoms or immunologic problems.
Never the fall that gets you, always the landing.

wandawitch.bsky.social•98 days ago

But the mechanism by which HIV causes AIDS is understood and has been explained.
There may be a correlation between AD and amyloid plaque build up but the mechanism hasn't been explained, unless I missed something really big.
Explain the mechanism and I'll re-evaluate

dweisman.bsky.social•98 days ago

Totally serious question: have you played a role in these trials as PI or sub-I to witness possible unblinding firsthand and how it plays out?

There was an effect on tau:

mikegreicius.bsky.social•98 days ago

dweisman.bsky.social•96 days ago

What do you make of t-tau in CSF? Less tau seems to be a good thing.

Regardless, spend some time in clinical trial, raters are blinded to safety MRIs. And you see IRRs and ARIA in placebo.

mikegreicius.bsky.social•95 days ago

Promising but CSF data is from 200 people, study enrolled 1700. Donanemab had largest clinical "effect", tau PET readout included 800+ people and showed nothing. Yes, IRR and ARIA in placebo but the rates are ~2-10x more with drug. Patient/study partner unblinded; CDR-SB depends heavily on them.

mikegreicius.bsky.social•98 days ago

Plasma p-tau is not a measure of brain tau pathology, but rather, counterintuitively, reflects brain amyloid pathology. Donanemab has the largest reported effect size in this class yet at 18 months there was no difference in frontal tau despite having 400+ per arm (eFigure 5 in Sims et al.).

dweisman.bsky.social•98 days ago

Jury might be out on what these represent, but anti-amyloid mabs do affect plasma and CSF tau at least.

Again, do you serve as a PI or sub-I on site?

mikegreicius.bsky.social•98 days ago

Jury is, in fact, back, after a short deliberation, with a guilty verdict: Tau aggregation in brain is bad, failure to slow its accrual suggests no effect on neurodegeneration.

Don't do trial work. If I did I'd be EVEN MORE concerned that 37% of people on donanemab had ARIA and a protocol change.

dweisman.bsky.social•98 days ago

As a therapeutic biomarker, seems to me that tau is a measurement of tau.

If you did trials, you would see there’s a ton of thought about how to maintain the blind, you’d also see placebo giving rise to IRRs and ARIA. Post hoc data analyses bears this out.

andrewmstern.bsky.social•98 days ago

Your main claim against efficacy (rather than pharmacodynamics) seems to be the unblinding argument. Do you see a relationship between rate of infusion reactions and efficacy? These could also unblind, but should have no biological relation to amyloid removal.

andrewmstern.bsky.social•98 days ago

I would be skeptical that ARIA, which in theory could unblind due to needing MRI but occurs in 8-10% of placebo recipients, would unblind more often than infusion reactions, which occur far less frequently in placebo groups.

mikegreicius.bsky.social•98 days ago

Good ?, but hard to answer without subject-level data which none of the companies have been willing to provide us. I think ARIA, triggering a dosing pause and more frequent MRIs, is the bigger concern for unblinding. Infusion reaction/unblinding will be a big issue for trontinemab.

andrewmstern.bsky.social•98 days ago

I don’t quite understand why you think dose pausing and additional monitoring would be more likely to unblind than an infusion reaction. Could you repeat your Fig 2 analysis but with infusion reaction instead of ARIA-E?

mikegreicius.bsky.social•98 days ago

Bigger adverse event, results in patient/partner being contacted by a study coordinator. These "across trial" correlation plots are weak tea. We want to look at "within trial" correlations but Eisai & Lilly declined our requests. Recently got gantenerumab data and will look at both aria and IRR.

andrewmstern.bsky.social•98 days ago

Is it really a bigger adverse event to the patient? Most ARIA is asymptomatic. Infusion reactions, in my experience, can make patients feel quite sick. If testing the unblinding hypothesis, it would make sense to test it against all possible causes of unblinding.

andrewmstern.bsky.social•98 days ago

And if you present some weak tea in Figure 2 for ARIA as a basis of your argument, wouldn’t it make sense to present the weak tea of the other potential causes of unblinding?

athelroy.bsky.social•98 days ago

Well said. Effective antibody for an irrelevant mechanism, as demonstrated by the NEJM paper. Lecanemab will cause a lot of brain bleeds and break the bank, to no purpose.

Comments

Posting Rules

Reply