A connectome-based model of delusion in schizophrenia using functional connectivity under working memory task

Abstract

Delusion is an important feature of schizophrenia, which may stem from cognitive biases. Working memory (WM) is the core foundation of cognition, closely related to delusion. However, the knowledge of neural mechanisms underlying the relationship between WM and delusion in schizophrenia is poorly investigated. Two hundred and thirty patients with schizophrenia (dataset 1: n=130; dataset 2: n=100) were enrolled and scanned for an N-back WM task. We constructed the WM-related whole-brain functional connectome and conducted Connectome-based Predictive Modelling (CPM) to detect the delusion-related networks and built the correlation model in dataset 1. The correlation between identified networks and delusion severity was tested in a separate, heterogeneous sample of dataset 2 that mainly includes early-onset schizophrenia. The identified delusion-related network has a strong correlation with delusion severity measured by the NO.20 item of SAPS in dataset 1 (r=0.433, p=2.7×10-7, permutation-p=0.035), and can be validated in the same dataset by using another delusion measurement, that is, the P1 item of PANSS (r=0.362, p=0.0005). It can be validated in another independent dataset 2 (NO.20 item of SAPS for r= 0.31, p=0.0024, P1 item of PANSS for r=0.27, p=0.0074). The delusion-related network comprises the connections between the default mode network (DMN), cingulo-opercular network (CON), salience network (SN), subcortical, sensory-somatomotor network (SMN), and visual networks. We successfully established correlation models of individualized delusion based on the WM-related functional connectome and showed a strong correlation between delusion severity and connections within the DMN, CON, SMN, and subcortical network.