Key functional circuitry altered in schizophrenia involves parietal regions associated with sense of self

Abstract

There is still no clear consensus as to which of the many functional and structural changes in the brain in schizophrenia are of most importance, although the main focus to date has been on those in the frontal and cingulate cortices. In the present study, we have used a novel holistic approach to identify brain-wide functional connectivity changes in medicated schizophrenia patients, and functional connectivity changes were analyzed using resting-state fMRI data from 69 medicated schizophrenia patients and 62 healthy controls. As far as we are aware, this is the largest population reported in the literature for a resting-state study. Voxel-based morphometry was also used to investigate gray and white matter volume changes. Changes were correlated with illness duration/symptom severity and a support vector machine analysis assessed predictive validity. A network involving the inferior parietal lobule, superior parietal gyrus, precuneus, superior marginal, and angular gyri was by far the most affected (68% predictive validity compared with 82% using all connections) and different components correlated with illness duration and positive and negative symptom severity. Smaller changes occurred in emotional memory and sensory and motor processing networks along with weakened interhemispheric connections. Our findings identify the key functional circuitry altered in schizophrenia involving the default network midline cortical system and the cortical mirror neuron system, both playing important roles in sensory and cognitive processing and particularly self-processing, all of which are affected in this disorder. Interestingly, the functional connectivity changes with the strongest links to schizophrenia involved parietal rather than frontal regions.

Keywords: VBM; functional connectivity; network; partial correlation; schizophrenia.

Figure 1

Changes in individual links and community scores in schizophrenia patients. (a) Bar plot of the largest changes in connectivity scores between the networks for schizophrenia patients and healthy controls. (b) Summary of the score changes for links within each RSN. [Color figure can be viewed in the online issue, which is available at http://wileyonlinelibrary.com.]

Figure 2

Altered connections in schizophrenia patients displayed in a single hemisphere. (a) Functional network structure of the 6 RSN showing the connections which are increased in schizophrenia patients (red) and in normal healthy subjects (blue). (b) Brain images showing the location of the structures and circuits where changes have occurred. [Color figure can be viewed in the online issue, which is available at http://wileyonlinelibrary.com.]

Figure 3

Significant ROIs with altered interhemispheric partial correlation coefficients between schizophrenia patients and control subjects. (a) Shows a pseudocolor plot of comparisons between all individual healthy subjects (1–62) and schizophrenia patients (63–131). (b) Histograms show specific brain regions with significantly altered interhemispheric connectivity—P value is given above each set. (c) Location of these significant ROIs. (d) Correlation of the mean correlation coefficients (MCC) of these significant ROIs and the illness duration/negative PANSS score. [Color figure can be viewed in the online issue, which is available at http://wileyonlinelibrary.com.]

Figure 4

Predictive value of connectivity changes for identifying schizophrenia patients. (a) Histogram shows (mean ± SD)% accuracy in the SVM classifier analysis using all connection changes in the group of schizophrenia patients compared with healthy controls across a range of training samples and when one connection is randomly left out. The table gives mean percentage values for specificity and sensitivity as well as accuracy. (b) Same as (a) but just for the circuit most changed in schizophrenia patients (IPG‐SPGPCUN‐SMG‐ANG). [Color figure can be viewed in the online issue, which is available at http://wileyonlinelibrary.com.]

Figure 5

GM and WM changes in schizophrenia patients. (a) The detected ROIs with significant GM and WM volume difference between patients and healthy controls. There are 58 ROIs with significant GM difference in which the ratios of GM volume of healthy controls over patients are all larger than 1, the sequence and names of the ROIs are as listed in Table 2. There are four ROIs with significant WM volume difference in which the volume of healthy controls are all smaller than the patients. (b) ROIs which have significant correlation between GM and illness duration. [Color figure can be viewed in the online issue, which is available at http://wileyonlinelibrary.com.]