Abstract
1570
Objectives: The question whether dementia with Lewy bodies (DLB) and Parkinson’s Disease(PD)with cognitive impairment are the same disorder has been debated for decades. Previous neuroimaging studies with FDG PET and multivariate network analysis have demonstrated that PD is associated with a specific disease-related spatial covariance pattern (PDRP) and its expression in individual subjects could be used in differential diagnosis with atypical parkinsonism (i.e., multiple system atrophy, progressive supranuclear palsy), as well as for correlations with clinical disability and disease duration. In this study, we aimed to investigate an analogous spatial covariance pattern exist in patients with DLB and how it characterize the boundary issues between DLB and PD.
Methods: In this study, we prospectively recruited 20 patients with DLB and 20 age/gender matched healthy control subjects to perform FDG PET imaging. 33 patients with PD and 33 age/gender matched healthy control subjects were also enrolled (Scans from these subjects were originally used to identify the PDRP biomarker in Chinese population). DLB-related pattern(DLBRP)was identified in the DLB patients and corresponding control subjects by an automated network computation on a whole brain basis. This applied SSM/PCA toolbox (http://www.fil.ion.ucl.ac.uk/spm/ext/) to the two groups of brain FDG PET images transformed into a common anatomic space. The DLBRP was determined from a linear combination of select principal components (PCs) whose expression in individual cases gave the maximum separation between the patients and control groups. For each pattern (DLBRP&PDRP), the network expression of individual scan was computed and compared between the groups.
Results: For the DLB patients and controls, the first 3 PCs accounted for 57% of the subject × voxel variance. A logistic regression model including PC1 and PC3 yielded the lowest Akaike information criterion (AIC) value and their linear combination was considered to be disease-related, i.e., subject scores for this pattern best discriminated the two subject groups (P<0.001). This DLBRP was characterized by metabolic increases in the bilateral putamen, premotor cortex and cerebellum, covarying with metabolic decreases in the bilateral caudate, thalamus, posterior cingulate and bilateral parietal-occipital regions. The topographic characteristics of DLBRP was similar as that of PDRP, but with an overall wider range. For each pattern (DLBRP/PDRP), subject scores measured in all subjects showed an effect of group (ANOVA; DLBRP: F[3,102]=108.115, p<.0001; PDRP: F[3,102]=74.071, p<.0001) with elevation in the DLB/PD patients (p<.0001, post-hoc tests) compared to the healthy controls. Moreover, network scores were similarly greater in the DLB patients than in the PD patients in both DLBRP (p<.0001) and PDRP (p<.0001).
Conclusions: This is the first study to investigate the specific metabolic brain networks corresponding to DLB using FDG PET imaging. We demonstrated that DLB patients presented a similar topography in cerebral metabolic abnormalities with PD patients, but with wider range as well as higher quantitative expression. These results indicate that PD/DLB maybe under the spectrum of the same disease, while the latter is likely to be associated with greater lesions underlying neuropathological substrates.