The hype about the new Alzheimer’s drug Donanemab has gone too far
A “historic moment” in September 2022. A “momentous breakthrough” a couple of months later. And today, a “turning point”.
From all these headlines, which are about a new set of drugs for Alzheimer’s disease, you’d be forgiven for thinking we were well on our way towards a working treatment. But despite the media hype, a lot of major questions remain: and that’s no less true of the newest drug, donanemab.
The drugs rest on what’s known as the “amyloid theory” of Alzheimer’s. The idea is that Alzheimer’s is due to a build-up of plaques made of the protein called amyloid-beta in the brain. The drugs are targeted at clearing out those proteins.
At that, donanemab does a tremendous job: today’s study shows that after eighteen months of treatment, nearly 80% of the participants on the drug had seen a substantial reduction in amyloid.
If amyloid causes the cognitive impairments that are characteristic of Alzheimer’s—and it should be noted that many scientists have questioned the amyloid theory and suggested alternatives—removing the amyloid plaques should help those cognitive problems. We might not expect the worsening to stop or reverse. But the drug might make the decline slower, and mean sufferers can remain independent for longer.
The researchers found that donanemab had a statistically significant effect on cognitive decline: compared to the placebo group. That is, the effect compared to placebo was very unlikely to appear if the drug had no effect.
But statistically significant isn’t the same as clinically significant: it’s perfectly possible for a drug to work, but to have an effect so small that the average person wouldn’t notice it: they won’t see changes in the things they can do in their everyday life, or their level of independence.
And this leads to some questions about the new results. The researchers state that a difference greater than 20% between the treatment and the placebo groups is meaningful, and cite three references that apparently back this up. But none of those provide a rationale for this number: one of them doesn’t discuss numbers at all, and two give the seemingly arbitrary figure of 25%.
Their headline result—that decline was slowed by a third—only applies to patients who had low-to-medium brain levels of another Alzheimer’s-related protein called tau. If you include patients who had high levels (fully two thirds of the sample), the effect is 22% – depending on where you put the threshold, it’s either below or on the very borderline of clinical significance.
For two of the cognitive tests they use, a 5-9 point difference and a 1-2 point difference are respectively said to be clinically meaningful for an individual person’s level of decline. In the study, the effect of the drug was a 2-3 point difference and a 0.7 point difference on the respective tests. It’s argued that these within-person changes aren’t the same as group differences in a trial. But if a person wouldn’t notice a change in their own life, it’s odd to argue that it would be meaningful in the trial of a drug.
Another strange thing: one of the “secondary” results of the trial is that 47% of patients taking the drug showed no change in their score on one of the cognitive tests after year, compared to only 29% of the patients in the placebo group. This was the first thing mentioned at the very top of the pharmaceutical company’s press release for the drug back in May. But now that we see the published study, this seemingly blockbuster result has been relegated to an online-only appendix. It’s not clear why.
It’s even possible that the effect could have been inadvertently overestimated in this trial – a lot of people with additional illnesses were excluded from participating, and that kind of selection effect can mean that trials find effects that are larger than the true effect of a drug. If that’s the case, we should revise down the effect size found here. It’s also important to note that we don’t know the extent to which the drug works beyond 18 months.
In a perfect world with no side effects, you’d recommend that every patient took these drugs, even if the average effect was very modest. But there are side effects, and some are quite concerning: many more of the treatment participants had brain swellings than those on the placebo; three donanemab patients died in ways that were considered relevant to the treatment compared to one in the placebo group.
Then there’s the question of cost-effectiveness. That’s something for government medical regulators to work out. The previous two drugs, aducanemab and lecanemab, were approved for use in the US despite even more borderline data on their effectiveness, and to be consistent they’ll have to approve this one.
But they—and our own UK regulators—should be pressing the pharmaceutical company to make a much more reasoned, detailed argument about the clinical significance of their drug’s effect before they make any decisions.
