IMPACT OF PLASMA GLUCOSE ON THE PATTERN OF BRAIN FDG UPTAKE IN COGNITIVELY NORMAL ELDERLY SUBJECTS

Abstract

Several studies reported increased blood glucose levels (BGL) to be associated with alterations of regional uptake of F-18-fluorodeoxyglucose (FDG) in the brain that mimic the characteristic pattern of Alzheimer's disease (AD)-related hypometabolism. Aim of the present study was to test this relationship in well characterized normal controls of the Alzheimer's Disease Neuroimaging Initiative (ADNI). The study included all ADNI control subjects with baseline FDG-PET that were cognitively stable for at least 36 months (n=87, 74.2±5.3y, 40 females). Pre-processed FDG-PET data was downloaded and stereotactically normalized to MNI space using SPM8. The mean signal in brain parenchyma was used for intensity scaling. Voxelwise multiple regression was used to test the effect of BGL on scaled FDG uptake, using a cluster-defining voxel-level threshold of p=0.05, uncorrected. Resulting clusters were considered statistically significant if the family-wise error corrected p-value of the cluster was 0.05 or less. BGL ranged between 59 and 149 mg/dl (98±16 mg/dl). There was a significant cluster of negative correlation between scaled FDG uptake and BGL comprising precuneus and bilateral parieto-occipital cortex (cluster volume 123 ml, p=0.020). The cluster did not include the posterior cingulate. Positive correlation was detected in a large cluster of white matter in the centrum semiovale (245 ml, p<0.0005). A piecewise linear broken stick model used to fit mean FDG uptake in the cluster of negative correlation as a function of BGL identified a breaking point at 94 mg/dl. Unexpectedly, the decline of scaled FDG uptake in this cluster was steeper below the breaking point than above. The broken stick provided a better fit of the data than a straight line (Akaike Information Criterion).