Identification of cholinergic centro-cingulate topography as main contributor to cognitive functioning in Parkinson’s disease: Results from a data-driven approach

BACKGROUND: Degeneration of the cholinergic system plays an important role in cognitive impairment in Parkinson’s disease (PD). Positron emission tomography (PET) imaging using the presynaptic vesicular acetylcholine transporter (VAChT) tracer [(18)F]Fluoroethoxybenzovesamicol ([(18)F]FEOBV) allows for regional assessment of cholinergic innervation. The purpose of this study was to perform a data-driven analysis to identify co-varying cholinergic regions and to evaluate the relationship of these with cognitive functioning in PD. MATERIALS AND METHODS: A total of 87 non-demented PD patients (77% male, mean age 67.9 ± 7.6 years, disease duration 5.8 ± 4.6 years) and 27 healthy control (HC) subjects underwent [(18)F]FEOBV brain PET imaging and neuropsychological assessment. A volume-of-interest based factor analysis was performed for both groups to identify cholinergic principal components (PCs). RESULTS: Seven main PCs were identified for the PD group: (1) bilateral posterior cortex, (2) bilateral subcortical, (3) bilateral centro-cingulate, (4) bilateral frontal, (5) right-sided fronto-temporal, (6) cerebellum, and (7) predominantly left sided temporal regions. A complementary principal component analysis (PCA) analysis in the control group showed substantially different cholinergic covarying patterns. A multivariate linear regression analyses demonstrated PC3, PC5, and PC7, together with motor impairment score, as significant predictors for cognitive functioning in PD. PC3 showed most robust correlations with cognitive functioning (p < 0.001). CONCLUSION: A data-driven approach identified covarying regions in the bilateral peri-central and cingulum cortex as a key determinant of cognitive impairment in PD. Cholinergic vulnerability of the centro-cingulate network appears to be disease-specific for PD rather than being age-related. The cholinergic system may be an important contributor to regional and large scale neural networks involved in cognitive functioning.