Self-Agency Experiment in First-Episode Schizophrenia Patients: A Comparative Study Across Two Sites

Abstract

INTRODUCTION: Schizophrenia is a complex psychiatric disorder that affects perception, emotion, judgment, and self-awareness, leading to profound changes in personality and reality perception and impairing social integration and work capacity. It affects approximately 0.3–0.7 % of the population (Saha, 2005). Disturbance in the sense of self, including thought insertion and delusions of control, is a core marker, linked to abnormalities in cortical midline structures like the medial prefrontal cortex and posterior cingulate (Northoff, 2004). Early-stage schizophrenia (first-episode spectrum disorders) presents an opportunity for effective intervention, as anatomical changes are less pronounced than in chronic cases (Ellison-Wright, 2008), though the precise origins and mechanisms remain uncertain. This study focuses on alterations in the sense of agency, whose neural substrate is believed to be mediated by network structures, such as the default mode network (DMN). METHODS: For the purposes of this study, datasets collected at two different sites—the Institute for Clinical and Experimental Medicine in Prague (IKEM) and the National Institute of Mental Health in Klecany (NIMH)—were used. The IKEM dataset consists of 131 subjects (76 FES, 55 HC), and the NIMH dataset includes 158 subjects (92 FES, 66 HC). Both datasets are comprised of fMRI data obtained during the Self-Agency Experiment (Spaniel, 2016). The data from both sites was preprocessed using the same preprocessing pipeline consisting of a standard approach for fMRI BOLD data. Using the GIFT toolbox (Calhoun, 2001; Du, 2020), a group spatial ICA was performed on each dataset separately. Relevant components were further selected based on their stability, correlation with the time course of the experiment, and a between-groups (FES, HC) two-sample t-test. Next, on each dataset, we conducted a voxel-wise analysis using the SPM12 toolbox (Penny, 2011). To detect group differences during the experiment, the two groups were compared using a two-sample t-test. Additionally, the two groups were statistically compared across the sites. RESULTS: The ICA results, further narrowed using post-hoc selection (p < 0.05, FWE corrected), found 5 and 4 relevant components for IKEM and NUDZ, respectively. Two of these components were overlapping, specifically in the anterior and posterior parts of the default mode network. The voxel-wise analysis also revealed the between-group differences to be in the DMN (p < 0.05, FWE corrected, min. cluster size > 20 voxels). The two-sample t-test conducted between the two sites found statistical differences only between the groups of HC (p < 0.05, FWE corrected, min. cluster size > 20 voxels) in the area of the dorsal anterior cingulate cortex. CONCLUSIONS: Regarding the results of this study, the Self-Agency experiment introduced by Spaniel (2016) provides an efficient means to detect group differences between first-episode schizophrenia patients and controls. Using ICA, we found stable components across two independent sites. These components are parts of the default mode network and were also found in the previous study (Spaniel, 2016). The voxel-wise analysis showed the main differences between the two groups to also be in regions of the DMN. Additionally, the comparison of the two datasets revealed the importance of site effects, which were present mainly in the HC groups.