Amyloid status imputed from a multimodal classifier including structural MRI distinguishes progressors from nonprogressors in a mild Alzheimer's disease clinical trial cohort

Abstract Introduction Mild‐Alzheimer's disease (AD) subjects without significant Aβ pathology represent a confounding finding for clinical trials because they may not progress clinically on the expected trajectory, adding variance into analyses where slowing of progression is being measured. Methods A prediction model based on structural magnetic resonance imaging (MRI) in combination with baseline demographics and clinical measurements was used to impute Aβ status of a placebo‐treated mild‐AD sub‐cohort (N = 385) of patients participating in global phase 3 trials. The clinical trajectories of this cohort were evaluated over 18 months duration of the trial, stratified by imputed Aβ status within a mixed‐model repeated measures statistical framework. Results In the imputed Aβ‐positive cohort, both cognitive (ADAS‐Cog 14 and MMSE) and functional (ADCS‐iADL) measures declined more rapidly than in the undifferentiated population. Discussion Our results demonstrate imputing Aβ status from MRI scans in mild‐AD subjects may be a useful screening tool in global clinical trials if amyloid measurement is not available.