NCHR Statement at FDA Alzheimer’s Advisory Committee Meeting On Aducanumab

November 6, 2020

I’m Dr. Diana Zuckerman, president of the National Center for Health Research.  Our center is a nonprofit think tank that scrutinizes the safety and effectiveness of medical products, and we don’t accept funding from companies that make those products.  My expertise is based on post-doc training in epidemiology and public health, and as a faculty member and researcher at Vassar, Yale, and Harvard.  I’ve also worked at HHS, the U.S. Congress, and the White House, and I’m on the Board of the Alliance for a Stronger FDA, which lobbies for more appropriations for FDA.  

We’ve all seen the devastating impact of Alzheimer’s Disease. Those of us whose loved ones have suffered desperately want to help them and to help ourselves to avoid that fate. Current treatments are widely used but not particularly effective. 

I will talk about what we’ve learned from the 2 studies of Aducanumab, based on quotes from the statistical slides from the FDA’s Dr. Massie.  Unfortunately, it seems those slides were not presented today, even though they were prepared for this meeting.  

  • “Study 302 was positive with high-dose treatment effect changing from -18% (futility dataset) to -22% (final dataset) on CDR-SB.”
  • “Study 301 was negative with treatment difference favoring placebo changing from 15% to 2%.”  Dr. Fountain called it a “failed sister study.”

Blinding is very important in a randomized double blind study.  A large number of patients experienced ARIA, which can be painful and have other symptoms.

  • ARIA was responsive to dose.
  • ARIA results in different management of patients, so patients will know they are getting additional tests. 
  • Mitigation: Raters were independent of patient care and were to remain blinded to dosing and adverse events  — BUT many patients weren’t blind because they are being treated for a side effect – brain edema.

Most of you know that post-hoc analyses are not considered scientifically appropriate.  But given the desperate need to find a treatment for Alzheimer’s, I have no argument with analyzing the hell out of the data in an effort to find out if the drug is effective for some patients under some conditions.  Nevertheless, I completely agree with FDA’s statistical concerns expressed in Dr. Massie’s slides, again using direct quotes:

  • “INCONSISTENCY: We have an equally-sized and identically designed study 301 that directly contradicts 302”
  • “301 high dose is numerically worse than placebo, p=.8252”
  • “Under null, .0975 chance of falsely rejecting the null across 2 studies” 
  • “If you have 2 studies and you take the best and pretend like it’s the only one, your estimate is likely biased.” 
  • “At best, evidence is from 302 only and there exists conflicting adequate well controlled evidence.”

And perhaps most important, we need to be concerned about the impact of approving this drug in terms of interfering with ongoing trials and future trials for any other promising Alzheimer’s drugs.  As Dr. Massie noted: 

  • “Noninferiority would be questionable with mixed divergent results” 
  • Approval “creates recruitment challenges for ongoing trials” 

You have a very tough decision today. 

  • There is a terribly urgent need for an effective treatment. 
  • Families’ Dilemma: They want you to approve this drug because they want to have hope, but this would be a very expensive treatment that might not work and might not be safe.  As a result, we would expect many desperate families would lose their life savings for a treatment that might not be at all beneficial for most patients. 
  • Inability to gather long-term data on this drug once it is approved.  This is especially important because the study was terminated early due to the pandemic. We need better long-term data on efficacy, and especially on safety.

Thank you for the opportunity to share our perspectives today.


The Advisory Committee members agreed with many of the issues we raised, and 10 of them voted that it did not meet the standard of evidence of effectiveness, with 1 abstention, and 0 voting yes.