Keita Sakurai, Daita Kaneda, Yuto Uchida, Haruto Shibata, Satoru Morimoto, Yoshio Hashizume
BackgroundAlthough neuroimaging differences from Alzheimer's disease (AD) in argyrophilic grain disease (AGD), a limbic-predominant 4-repeat tauopathy, have been evaluated, edge-wise alterations in structural similarity remain unexamined.ObjectiveAs an exploratory pilot study, to apply network-based statistics (NBS) to individualized structural covariance (ISC) matrices to identify morphological-similarity alterations in AGD and AD.MethodsWe analyzed 3D T1-weighted MRI from 13 pathologically confirmed AGD, 17 AD, and 18 healthy controls (HC). Gray-matter volumes from a 170-region atlas were used to compute ISC matrices. Between-group differences were assessed using NBS (threshold t = 3.5; 5000 permutations). Component strength was the sum of ISC edge weights within the significant component. Stability was assessed via leave-one-out analyses and bootstrap confidence intervals. Significant edges were mapped to Yeo 7 resting-state networks.ResultsAll three comparisons yielded a single significant component (HC versus AD: 340 edges; HC versus AGD: 80 edges; AGD versus AD: 39 edges). Component strength was higher in the group with less widespread morphological disruption (HC > AD; HC > AGD; AGD > AD), with non-overlapping bootstrap confidence intervals. Mapping to resting-state networks revealed widespread posterior cortico-limbic involvement in AD and a limbic-predominant pattern in AGD. In the AGD versus AD comparison, AD showed greater disruption within frontoparietal, somatomotor, and ventral attention systems. Component strength correlated positively with Mini-Mental State Examination scores (r = 0.48, p = 0.01).ConclusionsNBS applied to ISC matrices provides preliminary evidence of altered morphological similarity in AGD and AD, with AGD showing a limbic-predominant subnetwork and AD demonstrating broader posterior cortico-limbic involvement.