Skip to main content

Alzheimer's Disease

Biogen and Eisai: Let the Alzheimer’s Arguing Commence

Well, the tease is over: Biogen and Eisai have released data on their Alzheimer’s antibody, BAN2401. And the situation is messy, as many had feared. The top-line results are mostly positive, but there are several confounding factors that make them a matter of argument (see below).

The companies were scheduled to go public with the information at 4:30 EST yesterday afternoon at a meeting in Chicago, and trading in Biogen shares was pretty lively during the day (of course, it had been pretty lively ever since the announcement a few weeks ago). In mid-morning the shares suddenly leapt like a trout, ramping up steeply over the next hour or so, and then falling back just as rapidly. Some folks were moving an awful lot of a rather expensive stock on both sides of that trade, that’s for sure. For the rest of the day, the stock generally moved up in anticipation of the results, closing at a three-year high.

And that’s how we know that the news didn’t leak, actually. Because while the initial impression was favorable (as the cognitive assay percentage changes were released), further inspection as the slides went on raised several questions. One of the cognitive measures, for example, did not reach significance. The 10mg dose of the antibody did look favorable – but lower doses were, if anything, worse than the placebo group, which is a bit odd. The actual scores on these cognitive scales were not available, just the percentage changes, leading some to think that if those scores had been clear winners that someone might have been willing to recall them. Most worryingly, it turns out that there was a clear difference between the placebo group and the treatment group in APOE4 status.

APOE4, a lipoprotein, has been recognized as a risk factor for Alzheimer’s since back in the 1990s, although the mechanism behind that has been the subject of debate. But there was a regulatory-mandated change in enrollment of APOE4 patients during the (Bayesian) trial, for fear of a brain-swelling side effect that these patients are also prone to, and there ended up being far more APOE4-positive patients in the placebo group compared to the high-dose group (70% versus 30%). Which might well make that group show faster deterioration during the period of the study. Which might well blow your statistical proof that your drug substance is an improvement.

This issue, more than any other, seems to be the cloud hanging over these results today. But there are other thing to think about – for example, Lilly’s failed antibody solenuzemab was able to put up similar cognitive score changes in Phase II but ultimately failed (and more than once) in Phase III. And in general, drugs just don’t get better in Phase III – it can happen, but it’s not the way to bet. Moving to a larger, more diverse population (which is one of the things that Phase III is supposed to represent) tends to flatten things out as compared to often-more-carefully-chosen Phase II population. So the optimistic case would be that BAN2401 could continue to be mildly positive later in the clinic, and “mildly positive” would likely be enough for approval, given the state of Alzheimer’s therapy. But how likely is that compared to the other possibility, that the antibody has just managed to show the best possible numbers that it could manage to produce, and that they might be artifactual.

Well, one thing seems clear: the stock-market-bull case for this drug is in ruins. You had people running around thinking “Big win! Fast track! Early approval!”, which is why the stock (below $290/share in late June) had run up to almost $384 at the close yesterday. You can forget about all that early approval stuff (although Biogen and Eisai aren’t saying what they’ll do yet). My own opinion is that there will have to be a large, lengthy, expensive Phase III effort on this one, just like there’s a large, lengthy, expensive Phase III effort on every Alzheimer’s drug. And both investors and patients will have to hope that this one doesn’t end up like the rest of those Phase III trials did. It’s anyone’s guess. But the historical track record for Alzheimer’s drugs, anti-amyloid therapies, and anti-amyloid antibodies in particular is there for anyone to see. While I’m offering opinions, I think that any sort of early approval for BAN2401 would be a mistake. I completely understand the huge medical need for such a therapy, but what we equally don’t need (I exclude Biogen and Eisai shareholders from that “we”) is a rush by millions of patients, and their insurance companies, and the US government’s reimbursement system, into a drug that doesn’t actually do anything.

That would be both wasteful and cruel. I wish that I had better news for people looking for some therapeutic option for Alzheimer’s, but all I can see is “Maybe, and maybe not, and not for some time”.

Full disclosure: I went short Biogen stock during the morning yesterday, just in time to watch up immediately go up ten dollars a share. I covered my position this morning at the open, and I’ve timed this blog post to appear afterwards, not that think that I can move Biogen’s stock price. Other things being equal, I will tend to be short pretty much anyone who’s about to announce a big, important Alzheimer’s result, at least so long as the field is in the state it’s in today.

32 comments on “Biogen and Eisai: Let the Alzheimer’s Arguing Commence”

  1. Paul Brookes says:

    Once again the mainstream media gives a balanced and nuanced look at the data and doesn’t just buy into company hype 😉

  2. Anon says:

    Yea it was all over the network news yesterday! Biogen got that “name and fame” label and nothing could stop them, now. So you got to pick the patent very early for this drug at highest dose to work, marginally if that news has to be believed. Does that mean others can go back and do the clinical studies of their drug candidate, again knowing well that they did not pick up their patients “very early?”

  3. BASparks says:

    People should turn their attention to Anavex. A2-73 is not just an “Alzheimer’s drug”, but a cellular rejuvenation therapy. Their data continues to impress, as they have now identified genomic precision to their approach. Strong responders in the P2a trial have stabilized and some have improved from baseline. Three trials about to start: Alz P2b/3 – AU; PDD P2 – Spain; Rett P2 – US.

    1. Derek Lowe says:

      “Cellular rejuvenation therapy”? Looks like a muscarinic ligand to me. . .

    2. Big Richard says:

      Are you one of the people Anavex pays to promote its stock? AVXL was exposed years ago, only people buying it are novices.

    3. Belgian PhD student says:

      You put Alzheimer Drug in quotation marks but cellular rejuvenation therapy you leave unquoted?

  4. Lambchops says:

    All looks fairly underwhelming (I work on the standard assumption that a percentage change and a p-value with no values attached are on the wrong side of ‘clinically meaningful’).

    So as a side note from the data what are the thoughts of anyone here with experience of AD on ADCOMS as an endpoint in early/mild AD?

    I noted from the paper that it has some heavyweight clinical names behind it (eg Jeffrey Cummings, who authored the NPI scale) –

    Is this likely to become a standard for early AD? Are there any possible issues?

  5. gwern says:

    I guess the next question is, what happens if one stratifies by APOE status and compares slopes to control for that baseline difference? Unbalanced covariates are bad for a RCT, but not so bad if you’ve measured them at baseline and can control for them.

    1. Derek Lowe says:

      Yep, good question. Biogen says that Eisai is doing all these sorts of subgroup analyses.

      1. C_B says:

        “Our results didn’t turn out the way we wanted, but what if we did some subgroup analyses?”

        That’s certainly never gone wrong before

        1. Diver Dude says:

          Time for the obligatory XKCD link?

  6. luysii says:

    Forget it. We may well be treating Alzheimer’s disease with acyclovir (Zovirax) or something in this drug class in the future.

    Two recent papers in Neuron (vol. 99 pp. 56 – 63 and 64 – 82 ’18) show that (1) herpes simplex virus 1 triggers aBeta peptide formation in the brain of the 5FX mouse, and (2) there is evidence that other herpes viruses (HHV6 and HH7) are alive and well in our brains.

    I wrote an author of one of the studies and got this back — “Thank you for your interest in the studies. The results of the studies suggest a number of strategies to explore the use of existing drugs such as acyclovir and others in Alzheimers. Our team and no doubt Rudy, Rob and team are engaged in follow-up studies along these lines. There are some limited trials of antiretrovirals in Alzheimer’s that were recently published, but unfortunately I don’t have the citations on hand. I’m sure we’ll see many folks investigating antivirals as a result of the two papers. ”

    For more details — see

    1. cynical1 says:

      I have not seen those papers but HSV-1 and HSV-2 are not ubiquitous. So not everyone carries either of those viruses. For instance, for people under 50, ~67% carry HSV-1. It should be the easiest epidemiology study in the world to correlate either with Alzheimer’s. Has such a study been done? Of course, HHV-6 and HHV-7 are pretty much ubiquitous which makes the epidemiology very difficult. And finding herpetic viral residue in the CNS has been shown in more diseases than you can shake a stick at.

      If memory serves (and it may not), I do not believe that 1. acyclovir. has much intrinsic activity against HHV-6 and HHV-7 and 2. reaches a sufficient level in the CNS to inhibit replication. A 3g per day dose of valacyclovir gives a level of about 2.5 uM in the CSF. That would cover you for HSV-1 and HSV-2 but not HHV-6 and HHV-7.

      1. luysii says:

        Cynical1 — Pretty much everyone has HSV1 in their Gasserian ganglion. If a low grade herpetic infection is inciting Abeta production, then unless totally insoluble, stopping production of Abeta should at least diminish the size of the plaques by le Chatelier’s principle. Reducing plaque size should allow the glymphatic system to function better.

        I agree, this is far out, but anti-Abeta therapy hasn’t worked in the clinic. Time to look at all the antivirals for herpes viruses (man is infected by 8 of them) and perhaps develop new ones.

        1. cynical1 says:

          luysii said “Pretty much everyone has HSV1 in their Gasserian ganglion”.

          For clarification, is that pretty much everyone who has been infected with HSV-1 or does it magically just establish a latent infection in people who have not been exposed to the virus? Because based on my limited reading, that expression correlates with being seropositive to HSV-1 and that’s not pretty much everyone. Inquiring minds want to know.

          As I said above, HSV-1 is not understood or believed to be ubiquitous in the human population. Depending on the study, latent infection of HSV-1 in the human population is estimated to be in the 52-84% range for HSV-1 and if HSV-1 (or HSV-2) is being implicated in Alzheimers, I’d like to see a simple epidemiology study that shows that 100% of Alzheimer’s patients show evidence of being seropositive and similarly only 52-84% of their age and sex matched controls show evidence of being seropositive. Actually, that study is literally just about the most dead easy study I can imagine doing. (Come to think of it, I can’t imagine why Ascherio at Harvard isn’t doing it already.). But if it doesn’t come back 100%, I’m going to ask whoever pipes up to explain how a virus that a patient doesn’t appear to harbor is responsible for their disease. And I think that would be a reasonable question to ask.

          But I am going to make an assertion here: “Pretty much everyone” is not 52-84%. Pretty much everyone is 99+%.

          There’s a lot more to Koch’s postulates than you simply stating without reference that “Pretty much everyone has HSV1 in their Gasserian ganglion”. I’d like a little more than that before I believe that HSV-1 or any other herpes virus is positively correlated to Alzheimer’s. I can easily believe that pretty much every HSV-1 seropositive person has HSV1 in their Gasserian ganglion which is quite different than what you stated.

          Guess what? 100% of MS and SLE patients carry EBV. And that’s a big difference between the rest of the population where the number is 90% especially when you start studying a thousand patients and not quoting a study with a handful of corpses you studied who had Alzheimer’s. And in the case of childhood lupus, it’s not even 90% in the controls. It’s more like 50% in the controls versus 100% in the patient population.

  7. Chrispy says:

    The New York Times had in interesting article a few days ago about how hard it was to recruit sufficient patients to support all the trials in Alzheimer’s disease.
    It is really to the point where we can’t just fling things at it and hope something works; we have to choose the most promising candidates.

    1. Hap says:

      1) But how am I going to get investors if I admit that our drug isn’t likely to cure anything other than my financial difficulties?

      2) Less cynically, people have had a difficult time recognizing good drug candidates, sometimes – Lipitor? (why another statin?) and making that the gateway to trials could be a problem. It seems to be a fine line between persistence and insanity.

      1. John Wayne says:

        I think both your points 1 and 2 have an interesting interplay when people are trying to guess the winners. If folks are honest, it is really hard to tell what is going to work both scientifically and commercially. At the same time, there are some things that the consensus would agree a extreme long shots.

        So, it is hard to tell if new therapies are being put forward because we don’t know what we’re doing (your point 2,) or because we do (your point 1.) The answer is probably a bit of each.

        1. Hap says:

          There’s too much of a market to leave untouched (and the first one to find something useful is probably going to make a lot of money), so even if no one knows enough biology to know what to do, people are going to try anyway. If there’s so much uncertainty and so much demand, people with ill-intentions (or less-than-good intentions) can use the uncertainty to their advantage.

          Developing an Alzheimer’s drug looks like (not being there) all of the uncertainties of regular drug design with added uncertainty and long development and detection timelines. It seems like a very large lottery ticket paid long in the future. I don’t know how anyone does it.

  8. Sunyilo says:

    Digesting the reactions to the Biogen presentation it seems obvious that a statistical model should have been presented incorporating Apoe4 status as co-variate. One wonders how dishonest was the company in disclosing preliminary data without correct statistical treatment (a month ago)?
    In all likelihood this is the end of BAN2401 as it’s hard to see how another multibillion$ phase III in AD would be launched in addition to the already iffy aducanumab trial.

    1. tnr says:

      Agreed. I guarantee you that the stratified analysis has been done since it would be obvious to any statistician that it was needed. If they didn’t present the results of those analyses, then well, you know what the results were…..

  9. Ron Louie says:

    To Derek Lowe’s point about early approval, FDA Feb Draft Guidance may well pave the way to approve agents like BAN2401, or solanezumab, etc with their new biomarker criteria and staging system. I posted my comments in the Fed Register 05/16/18, and here:

  10. Peter C says:

    BIOGEN / EISAI: Was it worth it?
    Biogen shares soared 30% since partner Eisai made their claim to have seen a slowing in Alzheimer disease caused by their shared beta-amyloid candidate BAN-2401. Even after yesterday’s shareprice falls when the cheat and the mislead came to light, Biogen are still +15.4%, and Eisai are still +29.6%. Was it worth it? Go compute. Heard Alexion say at their Q2 telco they had ‘evidence’ of a role played by Complement and were throwing Soliris at AD. Can’t get this one out of my head, and like the idea. Anyone with experience at hand?

  11. Dolph Dümpling says:

    I wouldn’t even be surprised if the sole purpose of this story was some nice oldfashioned insider trading…

  12. Oligodendroctye Expert says:

    Peter C: there is (and has been for almost ten years) an evolving story concerning the role of the complement system in (supposedly) aberrant synaptic pruning in neurodegenerative diseases such as Alzheimer’s and Huntington’s.

    The late Ben Barres (Stanford), a prolific scientist and revered mentor to many in the glial cell biology field, co-founded Annexon Biosciences in 2011, in part to pursue the classical complement pathway as a target in Alzheimer’s disease.

    Among others, the work of Ben’s onetime graduate student Beth Stevens—now at Harvard, and the winner of a prestigious MacArthur Foundation grant in 2015—continues to elucidate the potential role of the complement system in neurodegeneration.

  13. RB Woodward says:

    Geeeez!…This is a blog for drug development chemists and nobody mentions the lack of brain exposure of these antibodies. If I proposed a CNS drug candidate that had an excess of one H-bond donor the entire med chem community would be criticizing me saying “It’ll never get through the blood-brain barrier”. Yet, these Alzheimer’s antibodies seem to be buoyed by some sort of magical thinking. Fact is, antibodies, cytokines, growth factors…These large molecules do not get into the brain. Antibodies that have failed in Alzheiemer’s disease clinical trials are NOT an apt test of the amyloid hypothesis, or any hypothesis for that matter. They never crossed the BBB. CNS diseases need be addressed using small molecule drugs.

    1. Biotonic says:

      but how would you explain the strong reduction of amyloid-beta, if the antibodies cannot cross the BBB.

    2. Ian Malone says:

      As Biotonic mentions, previous antibody trials (e. g. bapineuzumab) have shown reduction of plaques, plus the ARIA-E changes that these studies now check for with safety scans, the stuff is getting in. As for small molecules, it’s no good them getting in if they can’t affect the disease.

  14. Amitabh says:

    Recent draft guidelines did mention that FDA May consider drugs for approval based on biomarker changes and some clinical benefits as secondary outcomes. Looking at the difficulty of conducting AD trials and recruitment and most big MNCs moving out of AD clinical development it’s imperative for FDA to consider approvals. It’s very tough to get a clinical benefit in a long course disease progression where it might require 3-5 years to show any meaningful clinical benefit. The AMyloid Beta clearance is huge and scores on ADASCog and ADCOMS are significant.

  15. Biotonic says:

    APOE4 carriers have an earlier onset, but is their expected rate of decline different from patients with other APOE genotypes?

    1. Biotonic says:

      I am quite confused now:
      Memory was more impaired among APOE Ε4 carriers than among noncarriers. By contrast, naming, executive functions and mental speed were more impaired among APOE Ε4 noncarriers. This suggests that the APOE genotype modifies the clinical phenotype in terms of cognitive impairment in AD.

      Patients with early onset AD show more rapid cognitive decline than patients with late onset, suggesting that early onset AD follows a more aggressive course. Furthermore, this effect seems to be most prominent in patients with early onset who do not carry the genetic APOE ε4 risk factor for AD.

Comments are closed.