A Positive Amyloid Trial, Finally?

The big news last night was the announcement of a positive readout in Eisai and Biogen’s trial of lecanemab, another attempt an an anti-amyloid antibody for Alzheimer’s therapy. It was big news because positive results have been very rare indeed in this area, and there is a barrage of headlines this morning. As I write, Biogen stock (BIIB) is up about $70/share, a sudden 35% jump that adds billions of dollars to their market captitalization, so there’s definitely a lot of optimism out there. And as a longtime vocal skeptic of such anti-amyloid therapies, I’ve had plenty of messages asking what I think about all this!

Well, first off, let me say that I hope that these results are real and that they hold up. There are bigger things than being right (or wrong) about biological hypotheses, and the amount of suffering caused by Alzheimer’s disease worldwide – a problem that will likely increase in many countries as populations age – is beyond calculating. So anything that really does help slow the progression of this disease would indeed be good news (and not just for Biogen and Eisai!)

At the same time, though, news like this needs to be examined carefully. As the world knows, the anti-amyloid clinical landscape for Alzheimer’s is absolutely littered with failures in every direction: anti-amyloid antibodies of various types, attempts to inhibit beta-secretase and gamma-secretase enzymes, attempts to prevent aggregation, you name it. Nothing has worked. The presumption at this point is that such therapies will not succeed, so if lecanemab has indeed worked, the question is what makes it different. There’s a ready answer (up to a point) because antibodies can indeed be quite different (that’s their point!) and if you do manage to hit exactly what needs to be hit, you could expect efficacy when apparently similar attempts have led to nothing.

Lecanemab came out of work on the so-called “Arctic” mutation in human amyloid (E693G), which was first described over twenty years ago in a Swedish population. This led to lower plasma levels of the putatively toxic forms of beta-amyloid (Abeta40 and Abeta42), but to higher formation of “protofibrils” in cells. The hypothesis was that a rapid buildup of these led to accelerated amyloid deposition, sort of skipping the excess 40 and 42 stage of things, and perhaps represented an alternative form of Alzheimer’s pathology (one, interestingly, that apparently leads to lower numbers of amyloid plaques). That led to the thought that clearing these protofibrils could be of benefit to patients without this mutation as well, that this might be a mechanism acting to some extent in “normal” Alzheimer’s. Lecanemab was developed by a small Swedish company called BioArctic, who later did a development deal with Eisai. It’s certainly true that other companies with other antibodies have been trying to target amyloid in this “upstream” phase – the protofibril/soluble amyloid/oligomer hypothesis has been the subject of a lot of work (and a lot of failures), and it’s important to realize that this is not the first general whack at this piñata. But it’s a confusing area, and the details of how amyloid assembles (and what the toxicity of the various forms are) have not truly been worked out. And to that point, lecanemab does indeed seem to bind to protofibrils, but it also shows activity against oligomers of various kinds and even insoluble fibrils. And that’s another reason I have been skeptical of its prospects, because other antibodies have shown such spectra of binding activity that (to me) are difficult to differentiate.

The results from last night were from a Phase III trial, so lecanemab has been in the works for a while. In 2017, a Phase 2b study of the drug looked like a definite failure. Those results of that work (with more data from extended dosing) were published last year, and it was the typical half-empty/half-full readout in these things. The antibody was dosed at 10mg/kg twice a week, and the Bayesian trial was set to look for an 80% chance of a 25% or greater decline in disease progression, as measured by ADCOMS, the Alzheimer’s Disease Composite Score. It missed that endpoint at the specified 12-month evaluation point, and at the 18-month point it was actually showing weaker effects than it did at 12 months. So you could go with that viewpoint (missed, and missed further the longer it was dosed), or you could go with the “positive effects seen all the way out to 18 months” view, depending on your personality. Now, it has to be said those positive effects included reduction of amyloid levels, which is also arguable as a positive indicator, since several other antibodies have shown such reductions without showing any clinical benefit. But anyway. There was also about a 10% incidence rate of cerebrovascular edema, which has been a constant worry with such antibody treatments. Readers may recall that the evidence for the earlier Biogen/Eisai anti-amyloid antibody (Aduhelm) causing such edema was much stronger and clearer than any evidence that it improved Alzheimer’s pathology.

The companies went ahead into Phase III, and that’s what we’re seeing now. But we’re not seeing much – a more complete look at the data is expected in November, but for now, here’s what we have: this trial’s primary endpoint was the Clinical Dementia Rating – Sum of Boxes evaluation, CDR-SB, with ADCOMS and other ratings scales as secondary endpoints. Lecanemab reduced the decline in the treated patients by 27% as compared to placebo, which corresponds to an absolute score difference of 0.45 points. It’s not in the press release, but Eisai apparently mentioned on a conference call last night that the plateau-and-decline effect that showed up in the Phase 2b trial was not seen here – they say that the effect increased, with the largest values at the 18-month time point.

Now, here’s where the arguing starts: all the talk of “highly significant” statistical results from this trials comes from this rating score, but the natural question is whether a significant difference on the score means anything in the real world. This article, for example, includes the phrase “That difference is near what some analysts had previously said would represent a clinically meaningful benefit“.  As you would imagine, Biogen and Eisai are taking the position that such a question is laughable, even borderline offensive – of course this is going to be meaningful for Alzheimer’s patients! But there are a lot of comments this morning from clinicians in the field with significant doubts. It is no small thing, scientifically, to show such a difference in a patient rating score – but it might be too small a thing to notice for Alzheimer’s patients, their families, and their caregivers. The similarity to some cancer therapies is worth keeping in mind: in oncology, if you die at the about the same time you would have otherwise, but with smaller tumors (and that only confirmed by imaging, not by quality of life), no one cares. It’s a scientifically valid point and might lead to something greater down the line, but the real-world impact is another matter. So I hope that this is not the case with lecanemab.

The incidence of cerebral edema and related symptoms is going to be a big area of argument as well. It’s a complex subject – you can look at rates of imaging abnormalities (and those break down into two categories, effusion and microhemorrhages), and you can look at symptomatic/clinically noticeable ones compared across both of those. All of these were definitely higher in the treatment group than in the placebo controls, although they don’t seem to be as high as in the earlier Aduhelm trial. This good article at Stat quotes Michael Greicius at Stanford, who points out that the imaging results may well have essentially unblinded the trial, and would like to know what the effect of lecanemab was in the patients who did not show imaging abnormalities. One hopes we’ll see such data in November. It’s important to keep in mind that we’re going to be arguing about relative rates of decline: everyone who didn’t get the antibody got worse, and everybody who did get the antibody got worse, too. They just got worse at a statistically different rate than the placebo group.

And speaking of Aduhelm, its effects as measured on the CDR-SB score came out to 0.39 points as opposed to 0.45, but that was in the trial that gave a positive readout. Another trial, as many will recall, showed no such effect, and that was a big source of controversy in the eventual FDA approval. And here we come to another personality quiz: do you look at this comparison and think “Good, this one’s even better”, or do you look at the two trials and wonder what would have happened if two Phase III trials had been run for lecanemab as well, since you can apparently see a 0.39-point score change appear or evaporate? It’s an annoying question to ask, but it’s sadly a valid one when you’re right at the edge of showing benefit.

We’ll revisit this in November, when I hope that we have more data to talk about. For now, I would assume that we are already seeing all the positive news there is to see, and that later disclosures will just make everything more complicated! But that’s my personality talking. . .