Mild Amyloid Burden in the Brains of Cognitively Unimpaired Individuals is Associated with Long‐Term Cognitive Impairment

OBJECTIVE: We aimed to identify the minimal amount of amyloid burden measured by positron emission tomography (PET) in amyloid-negative cognitively unimpaired individuals that predicts the development of cognitive impairment. METHODS: A total of 1,896 cognitively unimpaired participants underwent longitudinal clinical/cognitive assessments for up to 20.3 years after amyloid PET imaging. Cox proportional hazards models were used to assess how the risk of cognitive impairment was associated with subthresholds of amyloid PET Centiloid. Linear mixed effects models were used to associate the longitudinal cognitive decline and amyloid accumulation with baseline amyloid burden. RESULTS: Amyloid-positive participants (Centiloid >20; n = 421) were 154% (hazard ratio 2.54, p < 0.0001) more likely to develop cognitive impairment than amyloid-negative participants. Approximately 48.7% of the amyloid-negative participants had mild amyloid burden (defined by Centiloid from 0.93 to 20) and were 43% more likely to develop cognitive impairment (hazard ratio 1.43, p = 0.008) than the lowest amyloid group (Centiloid <0.93) approximately 10 years post baseline. Longitudinal changes in amyloid PET Centiloid and cognition followed a similar pattern: amyloid-positive participants were the fastest, followed by the mild amyloid burden group and finally the lowest amyloid group, and the cognitive trajectories for the latter 2 started to separate only approximately 10 years post baseline (p = 0.003). INTERPRETATION: Almost half of amyloid-negative cognitively unimpaired individuals were at an elevated long-term risk of cognitive impairment, when compared with those with the lowest amyloid levels. These findings suggest that the current threshold for amyloid positivity may be overly restrictive, potentially excluding a large portion of high-risk participants from prevention trials. ANN NEUROL 2026.