Background: Cognitive impairment in Parkinson's disease (PD) is associated with altered connectivity of the resting state networks (RSNs). Longitudinal studies in well cognitively characterized PD subgroups are missing.
Objectives: To assess changes of the whole-brain connectivity and between-network connectivity (BNC) of large-scale functional networks related to cognition in well characterized PD patients using a longitudinal study design and various analytical methods.
Methods: We explored the whole-brain connectivity and BNC of the frontoparietal control network (FPCN) and the default mode, dorsal attention, and visual networks in PD with normal cognition (PD-NC, n = 17) and mild cognitive impairment (PD-MCI, n = 22) as compared to 51 healthy controls (HC). We applied regions of interest-based, partial least squares, and graph theory based network analyses. The differences among groups were analyzed at baseline and at the one-year follow-up visit (37 HC, 23 PD all).
Results: The BNC of the FPCN and other RSNs was reduced, and the whole-brain analysis revealed increased characteristic path length and decreased average node strength, clustering coefficient, and global efficiency in PD-NC compared to HC. Values of all measures in PD-MCI were between that of HC and PD-NC. After one year, the BNC was further increased in the PD-all group; no changes were detected in HC. No cognitive domain z-scores deteriorated in either group.
Conclusion: As compared to HC, PD-NC patients display a less efficient transfer of information globally and reduced BNC of the visual and frontoparietal control network. The BNC increases with time and MCI status, reflecting compensatory efforts.
Keywords: Between-network connectivity; Parkinson’s disease; cognitive resting state brain networks; functional MRI; graph measures; longitudinal; mild cognitive impairment; partial least squares analysis.