Brain Glucose Metabolism in Cerebral Amyloid Angiopathy: An FDG-PET Study

The in vivo diagnosis of cerebral amyloid angiopathy (CAA) is currently based on the Boston criteria, which largely rely on hemorrhagic features on brain magnetic resonance imaging. Adding to these criteria ¹⁸F-fluoro-deoxy-D-glucose (FDG) positron emission tomography, a widely available imaging modality, might improve their accuracy. Here we tested the hypothesis that FDG uptake is reduced in posterior cortical areas, particularly the primary occipital cortex, which pathologically bear the brunt of vascular Aβ deposition.

From a large memory clinic database, we retrospectively included all patients in whom both brain magnetic resonance imaging and FDG positron emission tomography had been obtained as part of routine clinical care and who fulfilled the Boston criteria for probable CAA. None had a history of symptomatic intracerebral hemorrhage. FDG data processing involved (1) spatial normalization to the Montreal Neurology Institute/International Consortium for Brain Mapping 152 space and (2) generation of standardized FDG uptake (relative standardized uptake value; relative to the pons). The relative standardized uptake value data obtained in 13 regions of interest sampling key cortical areas and the cerebellum were compared between the CAA and age-matched control groups using 2 separate healthy subject databases and image-processing pipelines. The presence of significant hypometabolism (2-tailed P<0.05) was assessed for the bilaterally averaged regions-of-interest relative standardized uptake values.

Fourteen patients fulfilling the Boston criteria for probable CAA (≥2 exclusively lobar microbleeds) were identified. Significant hypometabolism (P range, 0.047 to <0.0001) consistently affected the posterior cortical areas, including the superior and inferior parietal, primary visual, lateral occipital, lateral temporal, precuneus, and posterior cingulate regions of interest. The anterior cortical areas were marginally or not significantly hypometabolic, and the cerebellum was spared.

Supporting our hypothesis, significant glucose hypometabolism predominantly affected posterior cortical regions, including the visual cortex. These findings from a small sample may have diagnostic implications but require replication in larger prospective studies. In addition, whether they generalize to CAA-related symptomatic intracerebral hemorrhage warrants specific studies.