Metabolic subtypes of Posterior Cortical Atrophy: a data-driven approach controlling for pathophysiology-related hypometabolism

Objectives: The International Working Party on Posterior Cortical Atrophy (PCA) has suggested the existence of distinct subsyndromes within the PCA spectrum. However, this hypothesis has been challenging to validate in neuroimaging studies, which reveal substantial neuroanatomical overlap across PCA phenotypes. Our FDG-PET study aims at addressing one potential source of such overlap: the temporo-parietal hypometabolism related to amyloid deposits. Amyloid is indeed the most frequent pathology underlying PCA. Materials and Methods: We enrolled 55 patients with PCA, and 40 patients with biomarker-confirmed typical Alzheimer’s Disease (tAD) matched to the PCA group for age and MMSE score. Using a software for the automated analysis of PET scans, CortexID Suite, we extracted FDG uptake values from five bilateral posterior regions of interest (ROI) in each of the 95 participants. Values extracted from PCA patients were entered into a hierarchical cluster analysis to identify distinct metabolic patterns, and each ROI’s mean uptake values were then calculated for each cluster, and standardized relative to the mean uptake values extracted from the tAD group. Regions with z-scores at least two standard deviations below the tAD group’s values were considered specific of PCA, and used to define PCA metabolic variants. Clusters were also compared for socio-demographic and neuropsychological features. Results: Cluster analysis identified five metabolic subgroups within the PCA cohort, which were partially overlapping. However, this overlap decreased after uptake values standardization relative to tAD. This procedure identified the following discrete PCA metabolic variants: Cluster 1 (n. 11 cases, 20%) showed right parietal and lateral occipital hypometabolism, and was associated with a higher prevalence of cases with simultanagnosia and Gerstmann syndrome; Cluster 2 (n. 14, 25.4%) and Cluster 3 (n. 13, 23.6%) showed left and right occipital hypometabolism, respectively, and both included the majority of cases with visual agnosia; Cluster 4 (n. 11, 20%) showed left inferior parietal hypometabolism, and was characterized by a high prevalence of patients with Gerstmann syndrome and limb apraxia. Cluster 5 (n. 6, 11%) did not show areas that were hypometabolic over and beyond the areas of hypometabolism in the tAD group. Discussion: Our findings support the idea that PCA comprises distinct neurocognitive subtypes, which correspond to three of the four variants suggested by consensus criteria: (i) a dorsal, parieto-occipital, simultanagnosic variant; (ii) a ventral, occipito-temporal, visual agnosic variant; (iii) a dominant inferior parietal variant characterized by Gerstmann syndrome and apraxia. Metabolic subtyping highlights PCA complexity and could aids tailored interventions.