Neuroimaging studies have revealed that the mechanisms of auditory hallucinations are related to morphological changes in multiple cortical regions, but studies on brain network properties are lacking. This study aims to construct intra-individual structural covariance networks and reveal network changes related to auditory hallucinations. T1-weighted MRI images were acquired from 90 schizophrenia patients with persistent auditory hallucinations (pAH group), 55 schizophrenia patients without auditory hallucinations (non-pAH group), and 83 healthy controls (HC group). Networks were constructed using the voxel-based gray matter volume and the intra-individual structural covariance was based on the similarity between the morphological variations of any two regions. One-way ANCOVA was employed to compare global and local network metrics among the three groups, and edge analysis was conducted via network-based statistics. In the pAH group, Pearson correlation analysis between network metrics and clinical symptoms was conducted. Compared with the HC group, both the pAH group (p = 0.01) and the non-pAH group (p = 3.56 × 10−4) had lower nodal efficiency of the left medial superior frontal gyrus. Compared to the non-pAH group and HC group, the pAH group presented lower nodal efficiency of the temporal pole of the left superior temporal gyrus (p = 1.09 × 10−3; p = 7.67 × 10−4) and right insula (p = 0.02; p = 8.99 × 10−6), and lower degree centrality of the right insula (p = 0.04; p = 1.65 × 10−5). The pAH group had a subnetwork with reduced structural covariance centered by the left temporal pole of the superior temporal gyrus. In the pAH group, the normalized clustering coefficient (r = −0.36, p = 8.45 × 10−3) and small-worldness (r = −0.35, p = 9.89 × 10−3) were negatively correlated with the PANSS positive scale score. This study revealed network changes in schizophrenia patients with persistent auditory hallucinations, and provided new insights into the structural architecture related to auditory hallucinations at the network level.
Read full abstract