So let me say two things immediately about the now-published results for lecanemab, the latest attempt at anti-amyloid antibody therapy for Alzheimer’s. I think that both of these are true: the numbers are the best that I have yet seen for such an antibody, and at the same time I doubt if the drug will actually make a difference in the treatment of Alzheimer’s.
I realize that this is in great contrast to a lot of the headlines and coverage (although not all of it). The British press in particular has been completely useless in covering Alzheimer’s for decades now, a nearly unbroken record that is difficult for me to explain by reason, and sure enough, the BBC Health Twitter account says that lecanemab is being “hailed as a momentous breakthrough”. Most of the US coverage is less delirious, but you would still get the impression that some big step forward has been taken. Why should I disagree about that?
One reason is that I have become increasingly skeptical that amyloid is in fact the cause of Alzheimer’s disease. It is definitely involved in Alzheimer’s disease, and for many decades the idea that it was causative made a great deal of sense. The eventual explanation for Alzheimer’s etiology is going to have an amyloid component, but (as the common analogy goes) smoke is a component of a house fire, but it is not the reason that houses burn down. It’s possible that amyloid’s relationship to Alzheimer’s is similar. After so many clinical failures with agents (of several different kinds) that have taken their best shots at this hypothesis, I can’t at this point hold out much hope for clinically useful effects from an amyloid-directed therapy. Another such antibody, in fact, failed two weeks ago.
Lecanemab is designed to bind to the soluble protofibrils of amyloid. Other antibodies have been tried with somewhat different profiles (here’s a recent review), but this is hitting as early in the amyloid deposition process as any of them ever has. It should be noted that small-molecule drugs have aimed even further upstream at the beta-secretase and gamma-secretase enzymes that cleave beta-amyloid from amyloid precursor protein (APP) in the first place, and none of these have ever shown any beneficial effects, either. Lilly’s gamma secretase inhibitor actually made patients slightly worse.
What are the clinical results, then? Let’s get some depressing disclaimers out of the way first. No Alzheimer’s therapy has ever stopped the progression of the disease, for starters. No one really expected lecanemab to do that, either, and it didn’t. After that, it should go without saying that no Alzheimer’s therapy has ever actually reversed any of the damage of the disease, either, and thus the emphasis on identifying and treating patients as early as possible before too much of that damage has taken place. Lecanemab was most certainly not aimed at producing any result like that, because its developers know like the rest of the medical community how far we are from any such thing. Anyone who tries to sell you something with claims to reverse Alzheimer’s is cynically trying to take your money.
So if our Alzheimer’s candidates don’t reverse the damage and don’t stop the disease where it is, what do they do? The trials are designed to look for a slowing of the progression of the disease, to see if the agent being tested manages to make people decline at a slightly slower rate than they would otherwise. Lecanemab is arguably the first antibody to even show that much of an effect. Patients (50 to 90 years old) received either the antibody at a 10 mg/kg dose every two weeks or an iv placebo, with 900 patients enrolled in each group and over 700 completing the trial in each. The primary endpoint was the change at 18 months of dosing on a ratings scale called CDR-SB (Clinical Dementia Rating – Sum of Boxes). That’s a widely used measure that looks at memory, orientation, problem solving, interactions with the community and home life, and personal care, and these sorts of rating scales are pretty much the best measures we have of the progression of such diseases. They have their shortcomings (subjective ratings and all), but these are still valid measurements. Other broadly similar ratings scales were used as secondary endpoints.
Over 18 months, both the treatment group and the placebo group declined by this measure – the participants were all showing signs of mild cognitive impairment and showed evidence of amyloid deposition by PET scans or in their cerebrospinal fluid, so you would unfortunately expect everyone in that situation to decline. The treatment group did decline less, and the difference between it and the placebo group was indeed statistically significant. That’s what I mean (and all I mean) when I say that these are the best numbers I’ve seen, because previous anti-amyloid antibodies have not even shown that. (Yes, I know that the FDA approved another one (aducanumab), but I thought that was a huge mistake, and the FDA’s own statisticians had come out against approval).
That result is the reason for all the excitement. And now that we have it, there are three questions to answer: (1) how likely is it that this result is true? (2) If it is, how much difference will it make for Alzheimer’s patients and their families? And (3), if that difference is meaningful, are there any other things about lecanemab that might argue against its use? We’ll take those in order.
I think that the slower-decline result is probably real, but I can’t be completely sure. Recall that with aducanumab there were two sets of Phase III data, and one of them was notably more positive than the other (which was not positive at all). That’s enough to show you that you can get rather variable clinical readouts in this area. Lecanemab in fact missed its primary endpoint at 12 months in the Phase IIb trial, but results at 18 months were enough to continue with Phase III.
But let’s stipulate that the result is real, for the sake of argument. That takes us into the very contentious question of real-world utility. As the NEJM paper says, “A definition of clinically meaningful effects in the primary end point of the CDR-SB score has not been established”. Clinicians are already disagreeing over whether the difference between lecanemab and placebo is something that would even be noticeable. That last link features a quote of Mdhav Thambisetty, a neurologist at the National Institute on Aging: “From the perspective of a physician caring for Alzheimer’s patients, the difference between lecanemab and placebo is well below what is considered to be a clinically meaningful treatment effect”. This is not an uncommon take.
And that leads to question 3. A constant problem with these anti-amyloid antibody ideas is the complication of brain edema, an inflammation response that can be serious trouble. The term of the art is “amyloid-related imaging abnormalities with edema or effusions”, ARIA-E. This latest trial kept a constant watch for this, as well it should have, and any such trial also has to keep in mind the possibility of “functional unblinding” as any incidents develop. ARIA-E was noted in 0.8% of the treatment group (and in none of the placebo patients, naturally). Overall, adverse events that were enough to lead to patient discontinuation in the trial occurred in 6.9% of the treatment group and 2.9% of the placebo group. Most seriously, two patients in the treatment group have died from what could well be treatment-related vascular issues. In these cases, the patients involved appear to have had significant amyloid deposits associated with cerebral blood vessels, which then could have been weakened by the amyloid loss brought on by antibody treatment. Lecanemab did indeed show substantial amyloid clearance in the brain, so it is definitely working on target – it’s just that I don’t think that the target is worth very much. The fact that the drug has such effects on amyloid and still just barely slows the course of the disease argues for that point of view as well. The idea that vigorous amyloid clearance might be a hazard of its own is going to be quit
Eisai and Biogen are surely going to ask the FDA for accelerated approval, and I would expect the agency to grant it, contingent on another confirmatory trial starting up. (After all, if they approved aducanumab, they have to approve this one, although the second half of that proposition can unfortunately be filled in by all sorts of stuff). Safety concerns are really the only thing that can derail this, and I do not envy the agency having to work these issues out.
Addendum: there’s an interesting report that Eisai is very much taking the lead on this compound, to the point that they’ve apparently said that they are not taking queries from analysts who cover Biogen. I don’t know quite what to make of that, but there may well be residual frictions from aducanumab experience. . .
