Structural covariance network alterations across the spectrum of cognitive status in Parkinson's disease

Parkinson's disease (PD) is a neurodegenerative disorder in which cognitive impairment frequently emerges as the disease progresses. Structural covariance network (SCN) analysis provides a framework for characterizing large-scale brain network organization beyond regional atrophy. This study aimed to assess alterations in SCN topology across the cognitive spectrum of PD.

We retrospectively reviewed 16 healthy controls (HC), 20 patients with PD and normal cognition (PD-CN), and 22 patients with PD and mild cognitive impairment (PD-MCI). Cortical thickness-based SCNs were constructed, and graph-theoretical analyses were performed to assess global network metrics, nodal centrality, and modular organization.

Using a bootstrap-based area-under-the-curve framework (N = 1000 iterations per group; sparsity 0.15-0.40), a progressive increase in modularity was observed across HC, PD-CN, and PD-MCI (all Bonferroni-corrected p < 0.001), and global efficiency was selectively reduced in PD-MCI. Nodal analyses revealed widespread alterations in betweenness, closeness, and eigenvector centrality, with comparable spatial extents across all three metrics. Community detection using the Louvain algorithm showed progressive reorganization of modular structure, with less distinct intermodular boundaries in PD-CN and more spatially extended modules in PD-MCI.

SCN analysis revealed stage-dependent network reorganization in PD, suggesting the utility of network-based approaches for characterizing large-scale brain alterations related to cognitive impairment.