P1‐325: Predicting Progression to Mild Cognitive Impairment in Cognitively Unimpaired Individuals Using Neuroimaging Biomarkers
Cynthia M. Stonnington,Wendy Lee,Robert Bauer,Sameen Sharieff,Yinghua Chen,Richard J. Caselli,Dona E. Locke,Eric M. Reiman,Kewei Chen
DOI: https://doi.org/10.1016/j.jalz.2016.06.1076
2016-01-01
Abstract:Neuroimaging biomarkers can detect disease pre-clinically but their prognostic value, alone or in combination, is uncertain. Volumetric MRI and FDG-PET data were compared in 18 cognitively unimpaired participants approximately 2 years (1.8±0.8) before the clinical diagnosis of aMCI due to AD and 35 participants matched for age and sex who remained unimpaired for ≥4 years (nonprogressors) after their imaging acquisition. We further compared the 18 progressors to 20 nonprogresors matched for age, sex, and APOE status. Participants were followed in the Arizona APOE Cohort, have a reported first degree family history of possible AD dementia, and were enriched for APOE4 gene dose. Voxel-wise analyses were used to characterize reductions in regional gray matter volume and cerebral glucose metabolism between groups. Binary logistic regression, leave-one-out, and Receiver Operating Curve analyses were performed to characterize the extent to which posterior cingulate glucose metabolism, the AD-related hypometabolic conversion index (HCI), hippocampal volumes (HCV), and entorhinal cortical thickness, alone or in combination, could distinguish between subsequent progressors and non-progressors. The progressor and non-progressor groups did not differ significantly in their age (68±4), sex, education, or general intelligence. The progressor group had significantly more APOEe4 carriers than the group that was not matched for APOE. Compared to both nonprogressor groups, those who developed aMCI in approximately 2 years had significant (FWE corrected p ≤ 0.05) decreases in glucose metabolism, gray matter volume, and cortical thickness in regions known to be preferentially affected by AD. The best predictor of aMCI when matched for APOE was right and left HCV at 79% sensitivity and 78% specificity, with no improvement in performance when adding FDG-PET data. The best predictor for the groups not balanced for APOE was left HCV at 74% sensitivity and 72% specificity; combining MRI and FDG PET data increased the sensitivity and specificity to 77% and 78%. Based on logistic regression models, APOE e4 partially mediated the effect on the left hippocampus. This study supports the possibility that regional MRI and FDG PET measurements, individually or in combination, could help predict the likelihood of progressing to the clinical stages of AD.