Serial amyloid PET in 32 patients treated outside a clinical trial showed a median reduction of 59.11 Centiloids after treatment with lecanemab or donanemab, and the size of that reduction was not significantly associated with baseline amyloid burden, APOE-e4 carrier status or amyloid related imaging abnormalities (ARIA), according to a retrospective single-centre study by Jana Ivanidze and colleagues at Weill Cornell Medicine, published online in the American Journal of Neuroradiology (doi 10.3174/ajnr.A9387). The reduction was statistically significant across the cohort (p less than 0.001) and within each treatment group.

The study reviewed records of patients with mild cognitive impairment or mild dementia due to Alzheimer's disease who had amyloid PET before and after treatment: 15 received lecanemab and 17 received donanemab. Scans were read visually and quantified using the Centiloid scale and regional Z-scores, and clearance was graded by the magnitude of Centiloid change. Among those assessable, 26 of 29 (89.7 per cent) met criteria for full or partial treatment-related clearance, and regional Z-scores fell across all regions examined. Baseline amyloid burden, and cognitive outcomes, did not differ significantly across clearance categories (p = 0.25). In adjusted analyses, greater global Centiloid reduction was associated with greater improvement on the Mini-Mental State Examination, particularly at lower baseline burden; the authors themselves frame this as showing the complexity of linking amyloid clearance to short-term cognition, not as proof of benefit.

The question is no longer academic for clinics. The US prescribing information for donanemab tells prescribers to consider stopping dosing based on reduction of amyloid plaques to minimal levels on amyloid PET imaging, a rule built into the label since the drug was approved by the Food and Drug Administration on 2 July 2024. That makes the follow-up scan a decision point, and this report is one of the first descriptions of how such scans behave in a single centre's ordinary patients rather than in a trial population.

The design sets hard limits on what can be read into it. There was no control group, no randomisation, and clinical follow-up was short, so the study can document amyloid removal but cannot establish that removal changed the course of anyone's illness; observational associations, including the one with MMSE change, are not evidence of cause. A separate prospective real-world cohort in Milan, reported in Alzheimer's Research and Therapy on 20 March 2026, covered 29 treatment courses and found a mean fall of 52.4 Centiloids among the 11 patients with six-month follow-up, with no significant change in cognitive measures over that period.