Restlessness or agitation is one of the core symptoms of CH. However, the neurological substrate underlying this phenomenon has not been thoroughly analyzed. Whether they are attributed to the core aggression circuit or other CH-related structures remains unclear.

We prospectively enrolled CH patients who consistently reported either the presence or absence of restlessness during attacks, and categorized them into restlessness and non-restlessness groups. All participants underwent fMRI; those in the restlessness group exhibited restlessness during scanning, while those in the non-restlessness group showed no restlessness during fMRI. In this study, thirty-two regions of interest (ROIs) relevant to CH pathophysiology and the core aggression circuit were selected. To identify restlessness-related networks, ROI-to-ROI functional connectivity was compared between two groups. ROIs showing significant differences were then used as seeds for ROI-to-voxel analyses to explore downstream networks. Multiple comparisons were corrected using FDR and FWE methods.

A total of 24 CH patients were recruited and categorized into two groups: restlessness (N=14) and non-restlessness (N=10). The ROI-to-ROI functional connectivity analysis of CH patients with restlessness revealed increased functional connectivity between the non-pain side locus coeruleus (LC) and the pain-side substantia nigra pars compacta (SNpc), which survived FDR correction (p-FDR = 0.016). Seed-based general linear model analysis further revealed decreased connectivity between the pain-side SNpc and pain-side superior frontal gyrus, which survived FWE correction (p = 0.037). However, no significant cortical connectivity from the LC survived FDR correction.

In CH, fMRI findings suggest that the neurological substrates of restlessness involve the LC and SNpc rather than the core aggression network. Weakened connectivity from the SNpc to the superior frontal cortex may represent the downstream pathway contributing to restlessness in CH.