Only 53 of 1,230 people screened, 4.3 percent, met trial-like eligibility criteria for anti-amyloid therapy in a UK brain-donor cohort where the diagnosis was confirmed at autopsy rather than inferred in life. The retrospective analysis, led by Haque and colleagues and published online on 29 June 2026 in Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring, linked in-life cognitive assessments from the Brains for Dementia Research cohort to standardised neuropathology, then applied eligibility rules in sequence. The largest single step of attrition was co-existing vascular neuropathology, which removed 70.7 percent of those who had otherwise qualified.
Of the 1,230 donors assessed, 945 had both cognitive and neuropathology data. Of those, 788 (83.4 percent) met harmonised thresholds for early clinical stage, defined as a Clinical Dementia Rating of 0.5 to 1 with an MMSE of 20 or above or a MoCA of 18 or above. Within that group, 232 (29.4 percent) had Alzheimer's pathology confirmed as Thal amyloid phase 3 or higher and Braak stage IV or higher. Exclusions were then applied hierarchically: vascular neuropathology (70.7 percent), anticoagulant use (19.1 percent of those remaining) and APOE e4/e4 homozygosity (3.6 percent). Eligibility fell with age, and the authors attribute the decline primarily to vascular comorbidity.
The direction of the finding matches other autopsy and clinic work. In a study of 428 autopsied older people published in Neurology on 18 August 2025, 94 percent had Alzheimer's disease neuropathologic change but only 26 percent had it without infarcts, Lewy bodies or LATE-NC. At the Alzheimer Center Amsterdam, among 1,309 new memory clinic patients, 79 (6 percent) remained eligible for lecanemab after label-based exclusions. In the UK the question is currently theoretical for NHS patients: NICE's final draft guidance of 19 June 2025 did not recommend lecanemab or donanemab for routine NHS use.
The caveats are substantial and the authors describe their framework as conservative. Brain-donor cohorts are older and enriched for dementia, so they are not a memory clinic population, and post-mortem vascular staging is not the test a clinician can run in a living person, where eligibility turns on MRI findings such as microbleeds and siderosis. The analysis is observational and retrospective, and no one in it received an anti-amyloid drug.