ADU: Neither Far Nor Deep

by Dr. David C. Weisman, MD

They cannot look out far.
They cannot look in deep.
But when was that ever a bar
To any watch they keep?

Robert Frost, “Neither Out Far Nor In Deep”

In 2007, Mr. O’Shea1 presented to my clinic, aged 56, after he couldn’t figure out a golf score. The way the numbers lined up no longer made sense. Around the same time he started drifting close to cars on either side of him. He tailgated a truck, telling his distressed wife he had to center himself on the road by following a close target. He went to an eye doctor who told him everything was normal. Then he came to see me. When I showed him a photo of a person in a hammock, he pointed at the edge, “string.” He could only see the small parts, not the whole, not even the person laying in the hammock.

His was a classic case of Posterior Cortical Atrophy (PCA), a variant of Alzheimer’s disease that first affects visual association areas. Like the typical variant that presents with memory loss, it is progressive and fatal. Dr. Oliver Sacks made the condition famous in his case report “The Man Who Mistook His Wife for a Hat,” written before we knew it as a type of Alzheimer’s disease.

Mr. O’Shea got into a clinical trial investigating a drug called bapineuzumab, bapi for short. Bapi was a man-made antibody against β-amyloid, aiming to clear it from the brain. Bapi eventually showed that it did in fact reduce β-amyloid accumulation. But sadly it had no clinical benefit. A few years later, it was a bittersweet honor to see my work, along with hundreds of others, recognized as co-investigators on that paper.2

More bitter than sweet because that drug failed. No matter that it cleared a bit of amyloid. No matter how many papers got published, nor how smart these people were, nor how hard we tried, nor our training, caring, diligence, Mr. O’Shea progressed rapidly.

Mr. O’Shea was the first person randomized and enrolled in a clinical trial at Abington Neurological Associates. Since then, hundreds more enrolled, each one put their lives on the line, all heroes, who have my deepest admiration.

And we’ve made progress. We’ve learned a lot about amyloid, the right targets, right dose, and right subjects. One of these amyloid drugs, aducanumab, ‘adu’ for short, recently got an ‘accelerated’ approval (more on ‘accelerated’ later).

It should have been a happy day, celebrating a new era in Alzheimer treatment. June 7 should have marked the start of human victory against a dread disease. What should have happened did not. The win is hollow, more bitter than sweet. Why more bitter? We need to go back a bit to understand where we find ourselves now: a place of acrimony, uncertainty, and danger.

Biogen conducted two identical trials, twins really, and in early 2019 their board halted both. They found that the combined trials had a low chance of being positive. But when every trial subject’s final visit was tallied and analyzed, something wondrous happened: one trial was positive, one trial was negative. One twin blessed, one cursed. Since the FDA requires two positive trials in usual situations, and the data did not meet this standard, many expected another clinical trial. Instead, the FDA accepted Biogen’s submission for approval.

Like the twin studies themselves, two opposite camps emerged: a skeptical majority, an enthusiastic minority, with very few in the middle. One skeptic even wrote an article saying that he would never prescribe the drug if it was approved. This is perhaps overly rigid, but understandable, as the drug causes brain swelling in 35% of the subjects. Here’s what that looks like on MRI (red circle):3

MRI of swelling in the brain

This event fully resolved, but left the subject in a medical ICU with headache, confusion, and high blood pressure.

The FDA convened a meeting with 11 independent experts with backgrounds in medicine, policy, law, and biostatistics. As the rest of the world watched the pandemic unfold, the Alzheimer world watched as Biogen made their case. As I listened, I found myself hoping for some secret in the data that would make it sensible. But like Kung Fu Panda, there is no secret ingredient. The data remained: one trial positive, one negative. The FDA advisory panel grew frustrated. After Biogen was done, Dr. Scott Emerson, a biostatistician at the University of Washington said this:

This analysis seems to be subject to the Texas sharpshooter fallacy, a name for the joke of someone first firing a shotgun at a barn and then painting a target around the bullet holes.

Dr. Scott Emeron, University of Washington

But that doesn’t capture the how much the panel disliked the data and bickered with Biogen and the FDA. A Biogen doctor told the panel, in what could most charitably be called bulls–t, that one negative study “does not detract” from a positive study. Dr. Emerson said this was incorrect, “with some complicity from the FDA clinical staff.”

"This is easily the most heated FDA committee meeting that I've covered this year. One of the advisory committee members just said, 'I feel, to a certain extent, the clock has been run out and we have not been able to ask questions.'" – Rowan Walrath

The panel voted against approval, without a single yes vote:

Committee votes 10 against, 1 uncertain

The enthusiasts also had their say. Patients, families, and some trialists thought they saw an effect. Advocacy groups pushed for approval. But the most important enthusiast was the FDA, and they had the biggest say.

Dr. Dunn of the FDA, whom Dr. Emerson accused of showing “complicity,” did not exactly approve aducanumab. Instead, he used the FDA’s Accelerated Approval Program. This permitted aducanumab’s removal of amyloid, and not clinical effect, as allowing aducanumab on the market. Because clinical effectiveness is not established, Biogen must complete another trial showing effect with a timeline that extends to 2030. The FDA never gave the Accelerate Approval proposition to their advisors, and as of today three members resigned in protest.

“Accelerated Approval” contains a bit of Orwellian double speak, sounding like we’re moving in the right direction, so let’s instead call it a conditional approval. The drug does get rid of brain amyloid. As to clinical benefit, we really just don’t know. In medicine, with lives on the balance, we must never pretend to know something we do not. A conditional approval asks us to pretend it works.

If you’ve been paying attention, another problem becomes obvious. Bapi also removed amyloid to a more limited extent, but it did Mr. O’Shea no good. Many predict a clinical benefit with adu’s robust removal of amyloid (I’m one of them). But that prediction is not clear and might be wrong. Dr. Dunn is pointing at a hammock and saying “string.” Dr. Dunn focused on the small parts that are favorable; not seeing the whole; and, dare I say it, not even seeing the person.

“Never treat the scan, treat the person,” my mentors told me, a hundred times, then a thousand, tattooed in my mind. And now, at the most delicate tipping point, Dr. Dunn based the approval on the scans, not on the people. It is also possible that adu’s brain swelling knocked out those who would have had the worst progression, tipping a trial positive.

The first airplanes, cars, and rockets were unsafe and ugly just like the first drugs to treat MS, HIV, and cancer were all terrible. But they were a start, the airplane got off the ground. AZT kept HIV patients alive a bit longer. We now have a first drug to treat Alzheimer’s disease, but don’t know if it works. This is not exactly what progress is supposed to look like.

Where to go from here? Where do we follow Dunn’s string? In a letter to the advisory panel, Dr. Dunn himself doesn’t quite know, and abandons both his responsibility and decision rights, “Ultimately, the decision on whether aducanumab will be used for treatment will be made by patients, their families and caregivers, and health care professionals.”

It is appealing to be pure minded and restrict this drug to trials. In clinical medicine, like life, never say never. We humans do not see out far nor in deep. In time, Dr. Dunn’s decision may be vindicated, as heroic as Mr. O’Shea. We must balance the data with the real people suffering from Alzheimer’s disease. Adopting a hard skepticism disallows autonomy, which seems an evil. The skeptics ignore a single positive study, which may be the truth, the drug may work. And yet, an evil twin lurks on the other side: all the risk of brain swelling with a simply unknown benefit. Another mantra, tattooed in my mind, comes into focus: do no harm. How do we balance these opposing mandates outside the safety of a clinical trial? I’m not sure. This uncertainty feels terrible, but better than embracing dogma.

While we might not have clear progress, we do have an advance. A staggering step forward is better than one backwards. Only three years ago it was hard to imagine removing amyloid at all. We now live in a world where we can see an amyloid PET scan turning negative. We will follow this string, to one day see patients progress no further out, no deeper in.

  1. Details changed to de-identify and protect privacy.
  2. Liu E, Schmidt ME, Margolin R, et al. Amyloid-β 11C-PiB-PET imaging results from 2 randomized bapineuzumab phase 3 AD trials. Neurology. 2015;85(8):692-700.
  3. VandeVrede L, Gibbs DM, Koestler M, et al. Symptomatic amyloid-related imaging abnormalities in an APOE ε4/ε4 patient treated with aducanumab  Alzheimers Dement (Amst). 2020;12(1):e12101. Published 2020 Oct 9. doi:10.1002/dad2.12101