P1-338: the Consistency of Hypometabolic Brain Voxels in Probable Alzheimer's Disease and Amnestic Mild Cognitive Impairment Patients from the Alzheimer's Disease Neuroimaging Initiative

Kewei Chen,Wendy Lee,Xiaofen Liu,Gene Alexander,Dan Bandy,Cole Reschke,Norman Foster,Michael Weiner,Robert Koeppe,William Jagust,Eric Reiman

DOI: https://doi.org/10.1016/j.jalz.2008.05.919

2008-01-01

Abstract:Fluorodeoxyglucose positron emission tomography (FDG-PET) studies find characteristic patterns of cerebral hypometabolism in probable Alzheimer's disease (AD) and amnestic mild cognitive impairment (MCI) patients. To define regions-of-interest (ROIs) or voxel-based search regions for future studies use, it would help to characterize the consistency of hypometabolic brain voxels in these patient groups. The objective of this study is to characterize the consistency of hypometabolic brain voxels in AD and MCI patients from the AD Neuroimaging Initiative (ADNI) using statistical parametric mapping (SPM) and bootstrap resampling. Batched SPM5 was used for each of 1000 bootstrap iterations to compare FDG images from 74 probable AD patients and 142 amnestic MCI patients to those from 82 normal controls (NC). Hypometabolic voxels were defined using uncorrected P≤0.005 for each SPM5 run. We examined the map formed with hypometabolic voxels detected at a specific percentage of times or higher over the 1000 runs. Compared to NC, AD was associated with hypometabolic voxels in extensive posterior cingulate/precuneus, parietotemporal and frontal regions ≥1% of the 1000 runs, many fewer voxels from frontal regions and only slightly fewer voxels from the posterior regions in at least 50% of runs, and no voxels from the frontal and only slightly fewer voxels from the posterior brain regions in at least 90% and 100% of runs. In comparison with NC, MCI was associated with hypometabolic voxels in extensive posterior cingulate/precuneus, parietotemporal and frontal regions in at least 1% of the 1000 runs, many fewer voxels from these regions in at least 50% of runs, restricted to only regions of posterior cingulate/precuneus and medial frontal cortex in at least 90% of runs, and only a few voxels from the posterior cingulate/precuneus in 100% of runs, providing empirical support for the more heterogeneous nature of the amnestic MCI group. Bootstrap analysis provides an alternative to the parametric family-wise error approach to examine the consistency of hypometabolic brain voxels in patients with AD and amnestic MCI. It also provides a foundation for identifying posterior cingulate/precuneus and parietotemporal ROIs or search regions for future FDG PET studies of AD and MCI.