The dossiers landed in Amsterdam eight months apart, and both got the same answer at first: no. On 25 July 2024 the Committee for Medicinal Products for Human Use recommended against authorising Leqembi (lecanemab), stating that the observed effect on delaying cognitive decline "does not counterbalance the risk of serious side events associated with the medicine", singling out amyloid-related imaging abnormalities, the brain swelling and small bleeds that shadow this whole drug class. On 27 March 2025, after consulting a scientific advisory group of neurologists and other experts, the same committee refused Kisunla (donanemab). Two antibodies, one hypothesis, two rejections. Then both were reversed.

That sequence is the reason a Personal View published in The Lancet Regional Health Europe on 10 March 2026 is worth a family's attention. It is written by Lorenzo Guizzaro and colleagues, including serving CHMP members and European Medicines Agency staff, and it is not new trial data; it is the regulator explaining its own reasoning in public. The paper sets out the standard 210-day evaluation, led by appointed rapporteurs, with the Pharmacovigilance Risk Assessment Committee alongside and an optional scientific advisory group consultation, followed where a company asks for it by a 60-day re-examination and finally a European Commission decision. Both medicines, the authors note, were authorised in the EU in 2025 only after that re-examination route, and only for mild cognitive impairment or mild dementia due to Alzheimer's disease in people who are not homozygous for ApoE4. Both antibodies showed statistically significant effects on cognitive and functional outcomes, the authors write, though "the clinical meaningfulness of these effects remains debated". They are equally blunt that the drugs do not halt the disease and carry substantial risks.

So what actually changed between no and yes? Not the underlying phase 3 results, which stayed where they were. What changed was who would be given the drug and how. Lecanemab returned on 14 November 2024 with a positive opinion confined to ApoE4 non-carriers and heterozygotes, the group at lower risk of ARIA, and the European Commission granted marketing authorisation on 15 April 2025, a decision Alzheimer's Disease International reported had passed through a Commission standing committee and an appeal board that each returned no opinion, meaning member states could not agree among themselves. Donanemab returned on 25 July 2025 with the same genetic restriction plus a different way of getting the dose in: Eli Lilly says the CHMP relied on TRAILBLAZER-ALZ 2 for efficacy and on TRAILBLAZER-ALZ 6, which tested a modified titration schedule that lowered the incidence of ARIA with edema at 24 and 52 weeks compared with the original schedule while achieving similar plaque clearance. Alzheimer Europe recorded the conditions attached: a controlled access programme, regular MRI monitoring, supervision by teams trained to spot and manage ARIA, and an EU-wide registry to track what happens in real practice.

Read that way, the European story is less about two drugs than about one recurring judgement. The benefit was accepted as real and modest; the harm was accepted as real and concentrated in identifiable people; and authorisation became possible only when the population was narrowed and the monitoring was hard-wired into the licence. It is a regulator reasoning by subtraction, and the paper is candid that these were difficult benefit-risk assessments rather than clean ones.

Not everyone reading the explanation was satisfied. In a correspondence published in the same journal on 11 May 2026, clinicians welcomed the transparency but pushed back on the yardsticks, arguing that the outcome scales used in these trials were not designed to measure cognitive decline in early symptomatic Alzheimer's disease. That is the unresolved core of the whole debate: the statistics can be robust while the question of whether a family would notice the difference stays open.

For someone in a memory clinic in Lisbon, Warsaw or Dublin, the practical shape of all this is narrow and specific. Eligibility now turns on confirmed amyloid pathology and on ApoE4 genotype, which means a genetic test before a treatment decision. Access runs through centres that can do repeat MRI and manage ARIA. And national payers, not the EMA, decide what is actually funded, which is why authorisation across 27 member states has not translated into availability across 27 member states.