Existing Alzheimer's drugs cause brain bleeds in up to 40% of patients. A Roche antibody candidate may avoid that.

Trontinemab, a Roche antibody presented at the Alzheimer's Association International Conference in London on 14 July, doesn't bind beta-amyloid any better than lecanemab or donanemab. It clears amyloid at a fraction of the dose by borrowing the transferrin receptor the brain already uses to ferry iron across the blood-brain barrier. Roche says that delivery bet is what cuts the rate of the dangerous brain-bleed side effect that hits up to 40% of patients on the approved drugs.

Today's amyloid antibodies clear plaque. They also injure the vasculature that drains it. Amyloid-related imaging abnormalities, or ARIA, are brain swelling, microbleeds, and, in rare cases, fatal hemorrhages. They showed up in roughly 20% of patients on lecanemab in the CLARITY-AD trial and around 40% on donanemab in TRAILBLAZER-ALZ, with deaths attributed to ARIA in both programmes. The U.K.'s National Institute for Health and Care Excellence has declined to fund either drug on the NHS, citing both the modest cognitive effect and a price tag the service refused to pay. The ceiling on this drug class is, in effect, a dose problem: clear amyloid fast and the vessel walls pay the price; clear it slowly and the cognitive benefit shrinks further.

Trontinemab's mechanism is the lever. Antibodies are large molecules, and only a small fraction of an injected dose typically reaches the brain by passive diffusion; the rest circulates through the body. Trontinemab is fused to a fragment that binds the transferrin receptor, the same receptor that pulls iron-bearing transferrin across the blood-brain barrier. That receptor-mediated transcytosis lets Roche administer about a twentieth of the antibody mass and still clear amyloid aggressively, according to the Phase I/II read-out presented by Luka Kulic of Roche in Basel. Lower systemic exposure means less amyloid being mobilised inside the vessel walls on the way out, which is the working explanation for the cleaner safety profile.

The data covers one year of follow-up on patients in the early-stage trial. Roche reports significantly fewer ARIA events than the comparators, though the company did not break out a precise figure in the conference presentation, alongside sustained amyloid reduction. The drug is now moving into a Phase III programme that will test whether the safer profile holds at scale, and whether the cognitive benefit, modest in lecanemab and donanemab, is preserved at the lower dose. Alzforum's coverage of the conference lays out the Phase III design, registered at ClinicalTrials.gov as NCT04639050, and the U.K. Dementia Research Institute's release frames the data as the basis for that pivotal study.

Trontinemab's lower ARIA rate so far is a safety result, not a cognitive one. The Phase II dataset is small, presented at a conference rather than peer-reviewed, and one year of follow-up is a window where late events can still surface. None of this changes the fact that the safety ceiling on amyloid antibodies has been the practical constraint on who actually receives them: lecanemab and donanemab are used in the U.S. and other major markets but have not been adopted in the U.K. A drug that gets in at a lower dose is the most direct way to address that constraint, and MedPage Today's AAIC coverage places the result in the same frame.

What to watch next: Roche's Phase III readout, the first chance to see whether the transferrin-receptor shuttle approach holds at scale and against an active comparator; any ARIA events that surface in the longer tail of the Phase II cohort; and the question every amyloid programme now has to answer, of whether removing plaque at any dose actually moves cognition enough to justify the price tag the U.K. and other single-payer systems have so far refused to pay.