(18)F-FDG PET database of longitudinally confirmed healthy elderly individuals improves detection of mild cognitive impairment and Alzheimer's disease.

Abstract

The normative reference sample is crucial for the diagnosis of Alzheimer's disease (AD) with automated (18)F-FDG PET analysis. We tested whether an (18)F-FDG PET database of longitudinally confirmed healthy elderly individuals ("normals," or NLs) would improve diagnosis of AD and mild cognitive impairment (MCI). Two (18)F-FDG PET databases of 55 NLs with 4-y clinical follow-up examinations were created: one of NLs who remained NL, and the other including a fraction of NLs who declined to MCI at follow-up. Each (18)F-FDG PET scan of 19 NLs, 37 MCI patients, and 33 AD patients was z scored using automated voxel-based comparison to both databases and examined for AD-related abnormalities. Our database of longitudinally confirmed NLs yielded 1.4- to 2-fold higher z scores than did the mixed database in detecting (18)F-FDG PET abnormalities in both the MCI and the AD groups. (18)F-FDG PET diagnosis using the longitudinal NL database identified 100% NLs, 100% MCI patients, and 100% AD patients, which was significantly more accurate for MCI patients than with the mixed database (100% NLs, 68% MCI patients, and 94% AD patients identified). Our longitudinally confirmed NL database constitutes reliable (18)F-FDG PET normative values for MCI and AD.