An MMSE cutoff is deciding who gets anti-amyloid therapy, and it is the wrong instrument

The edition in brief

In 184 patients with biomarker-confirmed atypical Alzheimer's disease seen at one centre, 70 to 85% would not have met eligibility criteria for lecanemab or donanemab, depending on the framework applied. The dominant reason was not disease severity: bedside cognitive thresholds drove half to two-thirds of exclusions, even though 82% had early symptomatic disease by Clinical Dementia Rating. Posterior cortical atrophy and logopenic aphasia score low on the Mini-Mental State Examination because the test loads on the domains those syndromes attack, so the instrument, not the stage, is doing the excluding. A companion Personal View argues that the barrier to using blood-based Alzheimer's biomarkers in primary care is not assay performance but the absence of a pathway in which the result has a defined role: in low-prevalence, heterogeneous populations pretest probability is unstable and testing drifts towards screening. In Parkinson's disease, loss-of-function variants of MC1R, the pigmentation gene carried by over 60% of people of European descent, were associated with 30% faster motor decline in the Parkinson Progression Markers Initiative and 50% faster in a replication cohort drawn from three trials, with a more than fourfold phenoconversion risk in a small prodromal group. The practice-changer is chronic subdural haematoma, where trials since 2020 have refined operative strategy, argued robustly against routine steroids, and tested middle meningeal artery embolisation as an adjunct — with stratification by risk of poor outcome now the live question rather than which operation to perform.