Skip to Content

Is Lilly’s Big Alzheimer’s Trial in Trouble?

Eli Lilly has been testing their solanezumab antibody against Alzheimer’s for some years now. The first rounds of clinical trials were. . .inconclusive at best. Perhaps there was a bit of an effect in the mildest-affected patient cohort, perhaps not. The company went for it, though, starting a second Phase III in a more carefully selected population. As I and many other people have noted, the company is very nearly being bet on Alzheimer’s, in a series of very expensive clinical programs on this and other agents, so the progress of this trial is a big issue for Lilly and its shareholders.

That would explain why they’re down about 5% this morning. The company has issued a press release saying that they’re changing the primary endpoint of the trial:

The original study design included co-primary endpoints of cognition and function—measured by ADAS-Cog14 (Alzheimer’s Disease Assessment Scale-Cognitive subscale) and ADCS-iADL (Alzheimer’s Disease Cooperative Study-Instrumental Activities of Daily Living), respectively. Emerging scientific evidence supports the idea that cognitive decline precedes and predicts functional decline in Alzheimer’s disease, particularly in earlier stages of the disease. Thus, Lilly has decided to amend the EXPEDITION3 trial to include a single primary endpoint of cognition (ADAS-Cog14). Functional outcomes will be measured during the trial in the same manner as previously designed, using both the ADCS-iADL and the FAQ (Functional Assessment Questionnaire). These two functional outcomes will now be considered key secondary endpoints for the EXPEDITION3 study.

I’m not enough of a neurologist to say how much of a big switch this is, but I do know, from the drug discovery end, that you don’t change the primary endpoint of a huge clinical trial without a very good reason. Problem is, there are two compelling reasons that come to mind (and they’re not mutually exclusive, either). One is that, as the press release states, that the trial really does need to be looking at cognition, and that it would be irresponsible not to recognize that and act accordingly. The other, though, is that the company fears that the trial as launched is in danger of missing its original primary endpoint, and has decided that since a case can be made for a cognition endpoint and that it has a better chance of yielding something positive.

It all comes down to the idea that cognition is truly a sound marker for functional effects. Is it? Investors clearly have some fears that it isn’t, but I’m going to be interested to hear the reaction from the rest of the Alzheimer’s research community. Is Lilly doing the right thing, or are they gasping for air?

32 comments on “Is Lilly’s Big Alzheimer’s Trial in Trouble?”

  1. watcher says:

    My understanding is that this study started in 2013, and now in 2016 they are changing the primary end-points. I’ve never heard of endpoints being changed during the course of the study. Is the even allowed to make is a valid study? Seems analogous to changing the “success” criteria by using more lax statistical criteria after getting a preview of the outcome.

    1. Isidore says:

      Presumably Lilly was able to make a strong case regarding the “scientific evidence [supporting] the idea that cognitive decline precedes and predicts functional decline in Alzheimer’s disease, particularly in earlier stages of the disease” based on the scientific literature to convince the FDA that it was OK to drop one of the “co-primary” endpoints.

  2. Lane Simonian says:

    Oh, my goodness. I hate to be the first or one of the first to comment–but Eli Lilly is gasping for air. First they had to data mine for a subset of the population that seemed to show some improvement with their drug, then according to some Alzheimer’s experts the small improvement in cognition in that subset was not sufficient enough to produce any recognizable difference, and now they are taking out activities of daily living as a primary endpoint (probably because they know that solanezumab is not going to significantly improve that outcome). I expect Biogen to go down the same dead alley.

  3. tnr says:

    There is really only one question you need to ask the Lilly folks: “Has the FDA approved this change in primary endpoint in writing for the pivotal trial? (yes or no) “. If the answer is not ‘yes’, then the trial is likely dead in the water. It doesn’t matter how many experts say it is scientifically justified.

  4. Anon says:

    I guess you could make an argument that ADL is what you might prescribe against – will this medicine make the patients daily life subjectively easier, whether or not they are better at the trail drawing and 30 min delayed word recall tests? – so hitting that one seems up front useful if a bit vague biologically. I do think that a change in the cognitive domain by ADAS-cog14 would indeed be a more disease relevant endpoint in the sense that it tries to reflect an overall continuum of disease progression. At the same time, the subscales are not equally linear over disease course – so I like the concept of endpoints that try to measure executive memory over the prodromal/MCI/AD transitions.

    I recall Lilly did the same in EXPEDITION 2?

  5. Anon says:

    Cognition is really just a surrogate biomarker for peroxynitrite levels, which is what patients really care about, isn’t it?

  6. Me says:

    FDA probably would have accepted the change already as a protocol amendment. Just means the trial has a valid design that doesn’t put patients at risk.

  7. Diver dude says:

    A better question might be “have the FDA agreed in writing that the results of the study using the modified protocol can be used in any marketing application”? And “do the EMA agree?”. I cannot believe Lilly would do this if the answer to either of these questions was “no”. It’s still cheating tho’.

  8. Toby says:

    Why is it “cheating”? As long as the change is made before the data are unblinded (as in this case, I presume), I can’t see any potential for bias.

  9. bank says:

    Since the results haven’t been unblinded (I hope), my suspicion is that the statistics might look better with a single end-point rather than two. Possibly due to accounting for multiple hypothesis testing. It still doesn’t augur well that they are hoping to “finesse” the outcome at this stage…

  10. anchor says:

    The relentless efforts by Lilly to have this antibody work on “select population” is going to give the caregivers and others nuts. I mean with all those mumbo-jumbo medical terms and scrutiny is almost like selecting people who can walk in between the raindrops! They ought to scrape it and move on instead of giving false hope and empty promises.

  11. Paul says:

    Commenters defending the change: how do you suggest explaining their having a motivation to make this change three years into the trial without access to the data that reveals something about the “promise” of the results from their point of view? And if information of that kind is available, how could the change ever be justified, FDA approval or not?

  12. Anon says:

    In trouble? It never stood a chance in hell. But then we already knew that whan they started dredging the barrel with post-hoc analysis of multiple hypotheses in cherry-picked sub-groups in their earlier trials. Don’t say we didn’t warn them, but this kind of gambling is really bordering on criminal fraud or negligence.

  13. Anon says:

    Presumably Lilly got this change by all the IRBs, but I would have liked to hear the conversations. It seems ethically difficult to recruit for a study in which you hold out the promise of improvement or reduced decline in function like ADLs, and then get almost to the end and say, oops we didn’t really mean that. They did start with a milder population I believe, so scientifically it makes sense. Although it would have made sense at the start of the study as well, so it isn’t clear why the change now, unless it is simply a matter of finally listening to the advise.

  14. Glen says:

    I’m in favor of this sort of large trial. At this point, learning that something does not work, and hopefully *why* it does not work is very valuable.
    We are in dire need of facts. Even learning definitively that solanezumab does not work would be a step forward.

    Plausible as an emotional opinion, not a scientific conclusion. Alzheimer’s research needs more data so it may move from plausible projects to targeted research.

  15. Anon2 says:

    I agree with Glen’s comment.

    Biogen’s aducanumab trials (disclosure: I work there, but obviously am speaking for myself) hope to succeed based on positive image-based diagnosis of prodromal and MCI patients; past anti-amyloid trials have hinted strongly that early intervention is key but their signal has been attenuated by mixed patient populations.

    When I tell friends about these trials, I say they’re either going to be:
    1. a breakthrough, or
    2. a very expensive way to disprove the amyloid hypothesis once and for all.

  16. tnr says:

    One can only speculate why Lilly would do this now. My guess is that they are looking at the data all together (treatment blind not broken) and what they are seeing is that for ADL, everyone is declining by a similar amount and no one is maintaining their baseline score. Under this scenario, it would be very difficult to show any drug difference in ADL.

    If this is the case, unless the drug differences are spectacular large for the ADAS-Cog, I just cannot see a regulatory body approving a product on just the ADAS-Cog.

    1. Mark Thorson says:

      The two tests that show up most often in the clinical studies on AD are ADAS-COG and MMSE. I’m wondering why they didn’t include MMSE too. Maybe they had some earlier work suggesting they wouldn’t do as well on MMSE?

      1. tnr says:

        I am sure Lilly is measuring the MMSE. FDA has consistently mandated that in addition to the ADAS-Cog, a drug has to show a benefit in functional outcomes for approval in Alzheimer’s Disease. I think the MMSE is considered an instrument to measure cognition, not functioning.

      2. Anon says:

        MMSE is too blunt.

        Maybe a little surprised that Lilly didn’t use CDR-SB as per the 2013 FDA draft recommendations on early AD, but there you go…..

    2. tangent says:

      Is the blinding actually strong enough that somebody can’t guess past it based on hints in the data? Itwould seems like if there is any effect of any kind (including side effect), that’s enough to break the blinding and then go hypothesis-shopping.

      1. tnr says:

        Although there probably are some AE’s which unblind individual subjects, I doubt that there is a way to unblind a trial for all 4000+ subjects looking at just side effects. The only people who have seen unblinded data are the independent Data Safety and Monitoring Board. If someone from that group has leaked information to Lilly, the trial is dead on arrival to the FDA.

  17. MTK says:

    This doesn’t sound good in the Lilly press release:

    “Lilly understands that regulators globally will continue to view both cognitive and functional endpoints as necessary for clinical trials in people with mild Alzheimer’s dementia, and regulatory guidance has been to include these as co-primary endpoints. Lilly is submitting the EXPEDITION3 amendment to all appropriate regulatory authorities.”

    So that means the regulatory agencies want cognition and function as primary endpoints and Lilly has decided to go against that guidance? So they’re betting that they can convince the agencies that cognition is enough? Good lord. So now they got to convince the agencies of not only the value of the drug itself based on the clinical trial data, but also the validity of the clinical trial itself?

    Am I reading that right?

    1. tnr says:

      That is how I interpret it also.

      1. MTK says:

        Oh boy.

  18. DTX says:

    Here’s an interesting aside to doing Alzheimer’s clinical trials: One of our clinicians pointed out it’s actually one of the hardest areas in which to recruit study subjects. Think about it: would you want to be tested to see if you have any signs of Alzheimer’s disease?

    Whereas finding out about high blood pressure or diabetes allows you to treat the disease, knowing you have early signs of Alzheimer’s could be quite a burden.

    On top of this, it can be difficult to manage patients who actually have Alzheimer’s because they may not remember why they are in the trial. One colleague whose mother is in a trial says she needs to re-explain the trial every time they go in for a test. In addition, whereas patients with other conditions can often go to a clinical trial site on their own, Alzheimer’s patients typically need someone to take them there, which makes running the trial even harder.

    Hence, running Alzheimer’s trials is difficult both regarding study subjects with and without the disease. (this doesn’t change the Lilly story. It just puts the challenge of these clinical trials into perspective).

  19. Hoosier Daddy says:

    I believe this is a political play, pure and simple. The science isn’t there. Even Lilly backed away from that ridiculous peripheral sink tripe ten years ago. That means the goal is no longer finding a cure or treatment, rather the effort has morphed into a campaign to generate a groundswell of demand for approval via legislative action. This is a HUGE market, and treating even a small fraction of AD patients would be a very lucrative endeavor

    Lilly has moved congressional mountains before (e.g., the insertion of a “Lilly clause” in the homeland security authorization legislation, so it’s not a bad bet to take another shot from a lobbying angle. In a nutshell, Lilly’s survival hinges on whether they can convince desperate families and outright rubes to pick up the phone and call their congresscritter. Given the current level of discourse on the political Right, that seems like a pretty reasonable strategy.

    After this goal post-moving announcement, I’m bumping the odds up to a 35% chance of success.

  20. Mandrake says:

    How is such a massive antibody going to reach the brain?

    1. Mark Thorson says:

      The same way the jellyfish protein does, presumably.

  21. Bob B says:

    Isn’t the use of cognition and function as thresholds for INDs to be swinging for the fence? Seems like after so many failures this field could use a series of singles or doubles strung together. Why can’t slowing the progression of AB accumulation or mopping up be a valid end point? After all isn’t that what the monoclonal antibody is designed to do?

    I have no problem with the “cherry picking” of a cohort for subsequent trials when in fact there appears to have been terrible screening of many prior AD trials as noted by Selkoe and Hardy, 2016 talking about Amyloid PET screening:

    >>>this had not been done in the completed trials reviewed above, where up to 30% of subjects were later found to lack amyloid.<<<<

    So it would seem that selecting a cohort that actually has amyloid present, when testing antibodies targeting amyloid might be a decent approach? One has to wonder how appropriate is one approach juxtapositioned with the heterogeneity of this disease?

    Maybe the solution will ultimately be an antibody that mops up Amyloid, some correction of or stimulation of TREM2, CR1 or CD33 effects, a beta secretase modulator, and the Holy Grail, some form of neurogenisis?

    But until we understand the eitology of the Amyloid Cascade aren't we jousting at wind mills?

    Why Lilly's solanezumab may or may not fail in clearing brain Amyloid may relate to its targeting the mid-region of AB and binding principally to soluble monomers and perhaps low-n oligomers?

Comments are closed.