IC‐P‐120: Multivariate data analysis of FDG‐PET and PiB‐PET in three conditions with underlying Alzheimer's pathology

Abstract

The goal of this study was to apply parallel ICA to FDG-PET and PIB-PET data from a clinically heterogeneous population of patients with underlying AD in order to: (1) identify specific components from each modality that correlate with clinical presentation; (2) identify relationships between spatial patterns of PIB and FDG across the population. Our sample included patients with amnestic AD (ADmem, n = 27), visuospatial AD (posterior cortical atrophy (PCA), n = 9) and language AD (logopenic variant primary progressive aphasia (lvPPA), n = 10). All patients were PIB-positive, 59% were female, mean age was 63.0 (7.7) and MMSE 22.0 (4.8). All data were Z transformed and FDG data inverted. Parallel ICA was applied using Fusion ICA Toolbox (Calhoun et al., 2006). Relationship to clinical diagnosis was examined using a Receiver Operator Characteristic approach, including age, gender, education and ApoE4 as covariates. Four significant components were identified for FDG (see Figure). These included a left inferior frontal and temporal component correlated with lvPPA (AUC 0.90, P = 0.009), a bilateral inferior frontal (low metabolism) and PCC/precuneus (high metabolism) component correlated with ADmem (0.83, P = 0.039) and two others correlated with PCA (bilateral occipito-temporal (0.88, P = 0.012) and right PCC/lateral parietal (0.78, P = 0.05)). None of the 4 significant PIB components (left PCC/lateral parietal, right parieto-temporal, bilateral inferior frontal, bilateral PCC/precuneus) correlated with diagnosis. A significant mixed FDG-PIB coefficient was found in which decreased frontal and increased PCC/precuneus FDG uptake was correlated with the inverse pattern (increased frontal, decreased PCC/precuneus) of PIB uptake (partial r = 0.76, P <0.0001).