HETEROGENEITY OF HYPOMETABOLIC BRAIN DYSFUNCTION IN AMNESTIC MCI

Abstract

Amnestic mild cognitive impairment (aMCI) is a clinical diagnosis with considerable etiologic heterogeneity. While the condition often represents prodromal Alzheimer's disease (AD), patients may also progress to other dementia types, stay stable, or even reverse to normal cognition. Detection of glucose hypometabolism by means of FDG-PET is a very sensitive in vivo marker of neuronal dysfunction, and represents one of the best established methods for early detection of an AD-typical neurodegeneration pattern. However, little is known about the heterogeneity of hypometabolic patterns among subjects with aMCI. We applied Ward's hierarchical clustering to voxel-wise FDG-PET data of 403 aMCI patients from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort to explore subgroups of distinct brain-wide metabolic patterns in this population. Identified subgroups were compared to 179 cognitively normal (CN) controls and characterized with respect to cognitive characteristics, AV45-PET and CSF-based biomarker abnormalities, and clinical progression over 2 year follow-up. Three aMCI subgroups with markedly differing metabolic profiles were identified (see Figure), including (i) individuals with an “AD-typical” hypometabolic pattern (N=199; 49%), who also showed AD characteristic biomarker abnormalities and highest clinical progression rates, (ii) individuals with a “diffuse cortical” hypometabolism (N=85, 21%), who were characterized by relatively more non-amnestic features, less amyloid pathology, and slower clinical decline, and (iii) patients with largely “normal” metabolism (N=119, 30%), who did not show increased biomarker levels or cognitive decline compared to CN controls. “AD-typical” and “diffuse cortical” subgroups could be further stratified into relatively mild and advanced hypometabolic stages with consistent differences in severity of cognitive impairment and clinical progression rates.