Development of prediction models for distinguishable cognitive trajectories in patients with amyloid positive mild cognitive impairment

Abstract

The clinical outcomes of patients with amyloid beta-positive (Aβ+) mild cognitive impairment (MCI) are heterogeneous. We therefore developed prediction models for distinguishable cognitive trajectories in Aβ+ participants with MCI. We included 238 Aβ+ participants with MCI from the Alzheimer's Disease Neuroimaging Initiative to develop a group-based trajectory model and 63 Aβ+ participants with MCI from the Samsung Medical Center for external validation. Three distinguishable classes, slow decliners (18.5%), intermediate decliners (42.9%), and fast decliners (38.7%), were identified. Intermediate decliners were associated with older age, higher AV45 standardized uptake value ratios (SUVR) and lower fluorodeoxyglucose (FDG) SUVR than slow decliners. Fast decliners were associated with older age, presence of APOE ε4, higher AV45 SUVR and lower FDG SUVR than slow decliners. Prediction models of cognitive decline showed good discrimination and calibration capabilities in the development and validation data sets. Our analysis yields novel insights into the cognitive trajectories of Aβ+ patients with MCI, which will facilitate their effective stratification in Aβ-targeted clinical trials.