The brain metabolic signature of visual hallucinations in dementia with Lewy bodies

Visual hallucinations (VH) are a core clinical feature of dementia with Lewy bodies (DLB), but their specific neural substrate remains elusive. We used F-18-FDG-PET to study the neural dysfunctional signature of VH in a group of 38 DLB patients (mean age +/- SD 72.9 +/- 7.5) with available anamnestic records, cognitive and neurological examination and Neuro-Psychiatric Inventory assessing VH. We tested the voxel-wise correlation between F-18-FDG-PET hypometabolism and VH NPI scores at the whole-group level, then adopting inter-regional correlation analysis to explore the resting-state networks (RSNs) metabolic connectivity in DLB patients with and without visual hallucinations, as compared to N = 38 age-matched healthy controls (HCs) (mean age +/- SD 71.5 +/- 6.9). At the whole-group level, we found a negative correlation between VH NPI scores and F-18-FDG-PET hypometabolism in the right occipito-temporal cortex (p < .001 uncorrected, p < .05 Family-Wise Error cluster corrected). Then, splitting the group according to VH presence, we found that DLB non hallucinators presented a pattern of connectivity seeding from this occipito-temporal cluster and extending to the ventral visual stream. At difference, the DLB hallucinators showed a metabolic connectivity pattern limited to the occipital-dorsal parietal regions. As for RSNs, both the DLB subgroups showed a markedly reduced extent of attention and visual networks compared to HCs, with a variable alteration in the topography. DLB-VH patients showed a more pronounced shrinkage of the primary visual network, which was disconnected from the higher visual hubs, at difference with both HC and DLB non hallucinators. These findings suggest that an altered brain metabolic connectivity withinand beyond visual systems may promote VH in DLB. These results support the most recent neurocognitive models interpreting VH as the result of an inefficient recruitment of the ventral visual stream and of a large-scale multi-network derangement. (C) 2018 Elsevier Ltd. All rights reserved.