Skip to Content

Alzheimer's Disease

Another Alzheimer’s Antibody Fails. There Will Be More.

To what should be absolutely no human being’s surprise, another amyloid-antibody trial has failed. Roche announced today that the Phase III work (two 750-patient trials) on crenezumab after an interim analysis showed a strong chance of futility. They’re still going on with a study in familial early-onset disease in Colombia, and they have yet another amyloid antibody (gantenerumab) in the pipeline, but crenezumab itself for Alzheimer’s in the general population appears to have failed.

You may be wondering about your own memory by this point – it’s hard to keep track, but crenezumab had already failed a few years ago. But Roche kept the faith after subgroup analysis, and went on what is now the traditional route for Alzheimer’s antibodies: move into earlier-stage patients at even higher doses. In fact, that’s exactly the same place that gantenerumab is in, because it too failed an earlier trial only to have development continue. You can imagine what its chances for success are at this point. Not to worry, though – the company’s chairman gave an interview just last weekend saying that he expected breakthroughs in Alzheimer’s treatment in the next few years. To be fair, I don’t think he said that he expected them from his company.

This is even harder on Roche’s partner, AC Immune, who are also working on an amyloid-targeting vaccine. They have programs targeting tau protein as well (such as RG-6100, an antibody that’s also a Roche partnership). Working on tau has the advantage of not having to climb over the huge pile of failures that amyloid-targeting therapies have left, but it might just be the devil that we don’t know so well yet. It’ll be quite interesting to see how the current crop of tau-targeting ideas works out, that’s for sure.

But there are still amyloid diehards out there, and Biogen has found itself leading that list, whether they envisioned themselves in that position or not. As those watching the field know, they have an amyloid antibody late in trials (aducanumab), and another one partnered with Eisai (BAN-2401). The company had its earnings call just the other day, and spent a good part of it fending off questions about these programs, especially since analysts were expecting the Roche analysis to come out very shortly. Biogen, naturally enough, didn’t want to speculate in front of everyone, but I’ll be glad to do so here.

I don’t think that aducanumab is going to work. And I don’t think that BAN-2401 is going to work, either. (Those two links in the last paragraph will take you to longer reasons why I believe that). I would be very happy to be wrong about those predictions, but you’d have to be out of your mind to be optimistic given how anti-amyloid antibodies (and anti-amyloid therapies in general) have gone. I know, I know – earlier patients. Higher doses. Targeting the real form of disease-forming amyloid, the perfect little oligomers/soluble thingies/whatevers that actually like give you Alzheimer’s, instead of what all those crude fist-ax antibodies were doing that came before. I’ve heard all of that. I just don’t believe it. We as an industry have slammed into the amyloid hypothesis over and over, and the results might be trying to tell us something.

Now, I could be wrong about these last anti-amyloid therapies. There is an outside chance that one of them may come through. But “Hey, you never know!” is not a business plan.

As a side point, this field illustrates in fact what I was talking about yesterday on the role of chemistry in drug discovery. In my recent post on a possible infectious disease mechanism for Alzheimer’s, I mentioned Lilly’s gamma-secretase inhibitor in passing as having “done nothing”. A longtime friend in the industry (who knew that program well) wrote to take me to task about that, pointing out that the compound was in fact an excellent gamma-secretase inhibitor and was the best shot at the target ever taken. He’s right, and I’m adding a note to that earlier post. Lilly’s med-chem team and preclinical development group did a great job by getting a compound that good in an area that difficult. But in the end, it was to no avail, because gamma-secretase inhibition does not seem to do anything for Alzheimer’s no matter how good your compound is and no matter how talented and hard-working your medicinal chemists are. So when I say “did nothing”, I mean “for the patients in the trial”, which (sadly) is indeed the bottom line.

26 comments on “Another Alzheimer’s Antibody Fails. There Will Be More.”

  1. sad says:

    This is so sad 🙁

  2. HR says:

    Does that longtime friend happen to be Jim Audia?

  3. Lambchops says:

    Can’t say I’m surprised. Though in a lot of ways I’m more interested in the study in familial early-onset disease. This might tell us a bit more about whether the failures are down to intervening at the wrong time (or measuring the wrong outcomes) and just plain aiming at the wrong target.

    Speaking of familial Alzheimer’s disease , this study identifying a potential biomarker is pretty interesting and perhaps suggests that we shouldn’t be too optimistic about targeting amyloid in the familial Alzheimer’s population either (the biomarker was present when patients became symptomatic but was not strongly associated with amyloid deposition) – https://www.nature.com/articles/s41591-018-0304-3

    As ever the over enthusiastic UK press deemed this a “breakthrough in identifying Alzheimer’s TM” The article I read at least acknowledged that this was for familiar Alzheimer’s and that the biomarker was not predictive in the general population (I’m not sure whether this improves my opinion on the journalist or not, but I’m willing to concede that they didn’t write their own headline). I can imagine that some of the other coverage was overblown without even bothering to qualify it given the state of scientific journalism here.

    1. Lambchops says:

      Mixing my “and”s and “or”s there. It’s a good thing I’m not a programmer!

    2. Ian Malone says:

      NFL is very worth looking at (I would say that, I’m tangentially involved in looking at it). The thing to bear in mind with it will be that it’s not disease specific. Or, not very; it wont tell you what type of neurodegeneration you have, but it is a nice marker that things are not going well. This makes it good for genetic cohorts like the one in the paper, since we’ve got a good idea what the gene carriers are going to develop, but what we don’t know is when. For presymptomatic carriers you have predicted age of onset, essentially the age their carrier parent developed the disease, which is correlated, but not great, and also poses problems when analysing them together with symptomatic carriers for whom you have an actual age of diagnosis.

      If you want to hold out hope for the amyloid hypothesis… as Derek notes, lack of detectable amyloid deposition doesn’t necessarily mean there aren’t toxic species like oligomers floating around beforehand. With techniques like PET we need enough amyloid laid down to separate people from the normal range of tracer uptake for unaffected people. The odds on that are getting longer though, and there’s been some work done on investigating how much we can affect disease progression even if amyloid is part of the process. (Disclosure, our centre recruits for a Biogen study, but my involvement is limited to occasionally making sure MRI get sent to the sponsors, so no idea how it’s going.)

  4. Emjeff says:

    Sub-group analysis – the last act of desperation…

  5. Mauro Fa says:

    “Hey, you never know!”
    -cit.
    Unfortunately, it seems a very popular refrain nowadays.

  6. some guy says:

    “Hey, you never know!” is not a business plan. (c)
    Challenge accepted! – said Vivek Ramaswamy

  7. Anonn says:

    Anyone care to link today’s discussion the recent one about gingipains/leaky blood brain barriers from 2 days ago? What if the appearance of amyloid is actually the cavalry showing up, instead of being the agent saboteur? Or is that totally outlandish?

    1. Chris says:

      Not outlandish at all! The amyloid-as-antimicrobial-peptide idea has been having a good few years. Lots of work from Rudolph Tanzi, Robert Moir and others on that, and from Ruth Itzhaki and others on the possibility that AD involves a response to viral infection. Still hard to see exactly how/when brain immunity leads to disease… Super interesting that they showed the gingipains to be involved specifically, as opposed to just plain host-driven inflammation. Wonder if anyone’s really tried any protease inhibitors for AD in the past?

      1. SAS says:

        It is interesting to note that there have been reports (albeit conflicting) of sub-antimicrobial dosage of doxycycline improving cognitive outcome in AD.

        Background here is that sub-antimicrobial doxycycline doses are functional as MMP inhibitors, and were approved a couple decades ago for gum disease. The MMP inhibition reduces self-inflicted gum tissue degradation during immune response against P. gingivals infection (gum disease). It is further interesting to note that doxycycline does in fact also have low micromolar IC50 against gingipains…

    2. Todd says:

      Not only is that principle not outlandish, but based on publications in the past year, we might be approaching a point where offering anti-amyloid therapy for a trial might be unethical. Considering how targeting amyloid hasn’t shown to improve disease outcomes, and how amyloid is being shown to try and mitigate the damage, I think the amyloid train has hit a dead end.

  8. neil says:

    Do enough trials and you’ll get a type I error…gets you over the regulatory hurdle. Proving lack of efficacy in a real world setting would be very hard….

    1. quantum of solace says:

      Need TWO trials with Type 1 error in the same program. That’s going to need a lot of antibody Phase 3 programs.

  9. azetidine says:

    Frankly, I think it was a poor judgement to ban Lane. I’m not saying that I believe his hypotheses, but I am saying that given the years of articles on Alzheimer’s on this blog, that they are surely no worse than any of the other hypotheses. We continue to see work on amyloid as the target, when the majority of the evidence now points to amyloid being an effect, not a cause. Yet there are still huge efforts being put forth on amyloid, and those people aren’t banned. If I had to start an Alzheimer’s project and I had to choose between amyloid and peroxynitrite, I’d choose the latter.

    1. tangent says:

      I don’t think it was a judgment on the science (that peroxynitrites are an outlier of badness) but on the conversational practices (that bulk-pasting in semi-related abstracts of low-grade papers in every thread was unproductive). If somebody were blatting in minor articles to promote the amyloid hypothesis, I’d disfavor that at least as much.

      You know, if Lane were offered the opportunity to write a thoughtful guest post reviewing what’s actually known and unknown about peroxynitrites in AD, I’d give that a shot (and the comments).

      I know I’ve always read Lane with a strong sense of “there but for the grace of God”, that I could easily, easily be at least as factually wrong about something, and t hat I could be just as attached to my thing as being good science and having pounds of evidence.

      1. MatthewTKK says:

        Agreed – I would be interested in a cogent argument laid out in full, rather than the buttonholing and crank-style repetition he used to advance his argument.
        It’s also pretty much axiomatic that if you have an outside-plausible theory, you can find enough in the sprawling, frequently crappy literature on AD to support almost anything. It’s the disprovable hypothesis and the well designed experiment which are the crucible. I literally have 2-3 related hypotheses about how the disease starts and progresses, based on this stuff and other findings plus my knowledge of cortical anatomy and function, but I keep them to myself because the world doesn’t need another conjecture without supporting evidence.

    2. Anon says:

      I don’t think Lane was banned. In fact I recall seeing a post from him recently. I think he was discouraged from automatically posting after every mention of Alzheimer’s. He is always a polite guy, never gets into flame wars, and was not IMHO trolling with his peroxy nitrite theory.

      1. MatthewTKK says:

        He wasn’t trolling, he just used every mention of AD as a pretext to restate his hypothesis, and derailed discussion.

  10. matt says:

    The really astonishing thing to me is that when you start going through the candidates that are left, it seems like every one of them has ALREADY failed a trial, whether II or III. So not only have all the competitors on that same target crashed and burned around you, but your own candidate has failed to generate good results, and yet STILL hope springs eternal, and let’s have another couple of Phase IIIs, shall we?

    Dr. Lowe talks about how hard it is to know when to give up and move on, versus persist in trying to problem-solve and keep things going…well, this looks like the new benchmark in Don’t Drag It Out THAT Far. It’s like a whole fleet of Titanics trying to ram that iceberg into submission.

    Still, I’m inclined to agree with Roche’s Franz that we may be poised for some breakthroughs. I completely agree that the progress in calibrating diagnostic imaging methods and the a/t/n classification scheme, are crucial helps to testing candidates and getting better clinical readouts. And the industry’s new-found humility, at least at the managerial/investment level, will help fertilize a hundred flowers to bloom, a hundred schools of thought to contend.

    And Lane, if you’re still reading this, I’d say don’t walk but RUN to some investor whose name ends in “swamy” and get $100 million to start Peroxynitrovant, hire some Harvard or MIT professor who can couch your ideas in the proper intellectual hand-waving and appeals to expertise (a Sinclair or Guarente, perhaps), and run a trial (perhaps using lavender oil as an “adjuvant” in the active arm, some other pleasant oil in the control arm). Pay yourself well, of course–those guys will certainly be getting their money early and often.

    1. Nick K says:

      Another tip for our friend Lane: cash out and buy your yacht BEFORE the results of the clinical trials.

      1. Isidore says:

        A suggestion regarding Lane: Since he has been consistently honoring Derek’s request of some months ago not to post on Alzheimer’s threads perhaps one might also refrain from goading him on.

        1. NoLongerMedChem says:

          I agree. He isn’t bothering you, why are you continuing to pick on him.

  11. FoodScientist says:

    I like the microbial hypothesis, it seems a little hopeful. It’s just you’d think some of the anti amyloid antibodies would make the disease outcome worse and give a really good clue as to the mechanism. It seems odd, but a drug that rapidly accelerates ALZ would be very exciting and a major breakthrough(for people not in the trial). Is it possible this has happened and the results just weren’t published? Though it would have to not be a big company, because they would probably stop funding most of the ALZ trials.

  12. loupgarous says:

    This is where (and I know I sound a little like Lane on this topic, but I have the discovery of discrete modes of fluoroquinolone toxicity as evidence) big data analysis of medical records comes in.

    It took the Taiwan and Toronto studies of thousands of patient charts (only possible because those charts already existed in easily data-minable form owing to centralized national medical insurance) to go from “anecdotal evidence” of severe AEs from at least some of the fluoroquinolones (such as Levaquin) to “a significant rate of those severe adverse events”. And the fact of those AEs (especially in older patients) is accepted as fact.

    We’ve got several plausible hypotheses for what causes Alzheimer’s disease. In the US alone there are over three hundred million people who get regular medical care. That “free-market” model of medical care which some US senators don’t like provides us with bills submitted to various payers for that medical care. So we know, at a minimum, who’s being treated for diagnosed Alzheimer’s disease, from the IDC codes for that medical care.

    Placing everyone’s medical charts in reasonable digitized form wouldn’t be cheap, but it would be possible (I’ve taken part in data transformation projects for products like human growth hormone that take patient data from multiple formats to create searchable post-marketing safety and efficacy databases). The only way to find out what causes Alzheimer’s diseases and what seems to reduce its incidence is to look at the patient data, not just from diagnosed AD sufferers, but everyone else, too.

    You’d still need to be vigilant to avoid post hoc. ergo propter hoc analysis, and the amyloid hypothesis appears to have been that sort of thing – but we had no real reason to believe it so until clinical study data showed lack of efficacy against AD of drugs known in animal and human models to suppress amyloid secretion.

    We have masses of actual patient data for retrospective studies, the sort which only really work with masses of actual patient data. Why don’t we use them?

  13. Andre Brandli says:

    I agree with Derek that the approach to target and clear Abeta plaques with antibodies, may ultimately not work out and fail in clinical studies. Failure may be attributed to either (1) lack of sufficient in vivo affinity, (2) poor penetration across the blood-brain barrier, or (3) off-target effects reducing effective concentrations in vivo. In addition, poorly stratified patient populations can hide efficacy in subgroups of patients. Importantly, failure to demonstrate therapeutic efficacy with antibodies targeting Abeta in clinical trials does however not refute the amyloid cascade hypothesis. It just means that your therapeutic agent was not doing what it was supposed to do. Furthermore, drug candidates may fail as the underlying hypothesis was wrong from the start. The notion for developing gamma-secretase and BACE inhibitors was that the inhibition of these proteases may block toxic Abeta production. As pointed out by Derek and others medicinal chemists at Eli Lilly and other companies had developed excellent gamma-secretase inhibitors that did what they were supposed to do in vitro and in vivo. Unfortunately, we go to know from mouse genetics that many of the PSEN1 mutations underlying familial AD are loss-of-function mutations. With this knowledge in mind, I was not surprised to see that the treatment of patients with gamma-secretase inhibitors promoted disease progression rather than slowed it down. Billions of dollars were spent on poor hypothesis.

    With regard to familial AD, the genetics unequivocally demonstrate that APP, PSEN1 and PSEN2 are the key disease causing factors. Hence, any successful treatment for familial AD patients must aim at addressing the miss- or loss-of-function of these gene products, either directly or indirectly. In my opinion, the AD field has been obsessed with either developing anti-Abeta antibodies or secretase inhibitors. There has been little room and resources for alternative approaches to target the key components underlying familial AD. More focus should rest on developing strategies leading to brain penetrating small molecules with therapeutic AD efficacy.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.