Donanemab slows dementia: a modest but real gain

For two decades, a drug that clears amyloid plaques from the brain failed to slow Alzheimer’s disease every single time it was tried. The hypothesis was simple, the target was abundant, and the clinical results were uniformly null. A phase 3 trial published in JAMA on 17 July 2023 is the study that broke that streak, if only narrowly. ‘Donanemab in Early Symptomatic Alzheimer Disease: The TRAILBLAZER-ALZ 2 Randomized Clinical Trial’, led by John Sims and colleagues for Eli Lilly, enrolled 1,736 participants with early symptomatic disease and both amyloid and tau pathology confirmed by positron emission tomography.

The design was a 76-week, double-blind, placebo-controlled trial across 277 centres in eight countries. Participants were randomized one to one to infusions of donanemab or placebo every four weeks, with a distinctive twist: once a patient’s amyloid plaque fell below a threshold, treatment could be stopped and the patient switched to placebo in a blinded fashion. That stop-trial design is the drug’s commercial argument as much as its medical one, because it promises a finite course rather than an indefinite infusion.

The efficacy numbers are real but modest. In the low-to-medium tau subgroup, the least-squares mean decline on the integrated Alzheimer Disease Rating Scale was -6.02 with donanemab against -9.27 with placebo, a difference that translates into 35.1 percent slowing of clinical progression. On the dementia-rating scale the slowing was 36.0 percent in the same subgroup. Across the combined low-and-high tau population the figures fell to 22.3 percent and 28.9 percent, still statistically significant but visibly smaller. Twenty-three of twenty-four gated outcomes were positive. Read plainly, the drug delayed decline by months, not years, and its effect was strongest in the patients whose tau pathology was least advanced.

The safety profile is where the honest reading lives. Amyloid-related imaging abnormalities, the brain swelling or micro-bleeds tied to this whole class of antibodies, appeared in 24.0 percent of the donanemab group against 2.1 percent of placebo, with 52 symptomatic cases. Infusion reactions struck 8.7 percent of treated patients. Three deaths in the donanemab arm and one in placebo were judged treatment-related. These are not exotic side effects; they are the expected price of aggressively removing amyloid, and they concentrate in carriers of the APOE4 risk gene, which is why the trial’s open-label guidance moved toward genotyping before treatment.

The paper’s quietest contribution is methodological. By stratifying patients on tau burden, the authors argue that earlier negative trials may have enrolled people whose disease was already too advanced for clearance to matter. That is a plausible reconciliation, but it is also a convenient one, and the 35 percent figure comes with a wide confidence interval that a sceptic can fairly describe as anything from roughly 20 to 50 percent. The biological argument, that removing amyloid should help, is finally supported by data, yet the size of the benefit raises the older question of whether amyloid is cause or merely passenger.

What the record now supports is a narrow but genuine conclusion. An anti-amyloid antibody can measurably slow early Alzheimer’s disease, in a defined subgroup, over 18 months, at the cost of monitoring for brain swelling and accepting a finite course of infusions. It is a treatment, not a cure, and it prolongs independence by a matter of months. After decades of failure that is a real result, and it is reasonable to call it modest. Both adjectives belong in the same sentence.