Between 70 and 85 percent of 184 people with biomarker-confirmed atypical Alzheimer's disease would not meet the eligibility criteria for lecanemab or donanemab, according to a retrospective study published on August 5, 2026 in Neurology, the journal of the American Academy of Neurology. The patients had posterior cortical atrophy, logopenic variant primary progressive aphasia, dysexecutive Alzheimer's disease or corticobasal syndrome, that is, Alzheimer's pathology presenting first through vision, language, planning or movement rather than memory. Researchers led by Dror Shir at Mayo Clinic in Jacksonville, Florida applied the criteria used in the pivotal trials and the appropriate use recommendations written for the two drugs. Performance on cognitive screening tests accounted for 50 to 67 percent of the exclusions, in patients the authors describe as early in their disease. "These results suggest a mismatch between how we measure disease severity and how atypical Alzheimer's disease presents," Shir said in a statement issued by the academy. The work was supported by the National Institutes of Health and the Alzheimer's Association. No one in the study was treated: eligibility was estimated from existing clinical records.
The structural complaint is that the gate is built from short memory-weighted screens and dementia staging, so a person whose deficit is language can score in the excluded range while still managing much of daily life. That is the question a smaller Japanese series set out to answer directly. Writing in Psychogeriatrics (March 2026 issue), Shoya Inagawa and colleagues at Tokyo Medical University, with Niigata University and the Tokyo Metropolitan Institute for Geriatrics and Gerontology, reported 13 patients with primary progressive aphasia assessed with the Standard Language Test of Aphasia, MRI, perfusion SPECT, cerebrospinal fluid biomarkers and amyloid PET, then classified as logopenic, semantic or nonfluent/agrammatic variant.
Their numbers point the other way. Of the seven logopenic patients, six (86 percent) were amyloid positive, and five (71 percent) met Japan's optimal use guidelines for lecanemab or donanemab. One logopenic patient was negative on Alzheimer's biomarkers, showed a perfusion pattern atypical for the disease, and was rediagnosed with frontotemporal lobar degeneration. The authors' stated aim was diagnostic rather than political: they conclude that SPECT may help separate logopenic Alzheimer's from the other aphasia variants, and they call for larger studies.
The two results are not directly comparable and should not be averaged. Thirteen patients at a small number of Japanese centres, screened against Japanese guidance, cannot settle what happens in a large American referral clinic applying trial criteria and appropriate use recommendations, and seven logopenic cases is a very small denominator. Both studies are retrospective, both estimate eligibility from records rather than assessing patients against the rules in real time, and neither examined whether the drugs help or harm people with atypical presentations. Atypical phenotypes were largely absent from the trials that won the drugs their approvals, so the underlying question, what these antibodies do in a person whose Alzheimer's began in language, remains unanswered by either paper.