Reduced brain cortical folding in schizophrenia revealed in two independent samples

Abstract

The cerebral cortex is highly convoluted, and principal folding patterns are determined early in life. Degree of cortical folding in adult life may index aberrations in brain development. Results from previous studies of cortical folding in schizophrenia are inconsistent. Here we investigated cortical folding patterns in the hitherto largest sample of patients with schizophrenia drawn from two independent cohorts. Magnetic resonance imaging scans were acquired from 207 patients and 206 healthy subjects recruited to two separate research projects in Sweden and Norway. Local gyrification index (lGI) was estimated continuously across the cortex using automated methods. Group differences in lGI were analyzed using general linear models. Patients had lower lGI in three large clusters of the cortex with peak differences found in the left precentral gyrus, right middle temporal gyrus, and right precuneus. Similar, although not completely overlapping results were found when the two cohorts were analyzed separately. There were no significant interaction effects between age and diagnosis and gender and diagnosis. The finding of reduced degree of folding in large regions of the cerebral cortex across two independent samples indicates that reduced gyrification is an inherent feature of the brain pathology in schizophrenia.

Introduction

The first clinical manifestation of schizophrenia usually occurs in late adolescence or early adulthood, but genetic and epidemiological findings suggest that the disease process involves alterations in early brain development to the developing brain (Weinberger, 1987, Rapoport et al., 2012). Neurodevelopment occurs in a programmed and gradual fashion with cortical neurons migrating to their destination before birth (Bystron et al., 2008). Major cortical folding patterns are mainly determined before birth and undergo only minor changes in childhood and adolescence. Importantly, the degree of folding relative to brain size remains relatively stable from early childhood (Armstrong et al., 1995, Zilles et al., 2013), and is thus a suitable subject for investigation of early brain development. A range of methods for measuring cortical folding has been developed (Mangin et al., 2010, White et al., 2010). The most widely used method is the gyrification index (GI), i.e. the ratio between the folded cortical surface and an outer cerebral surface tightly warping the brain without entering the sulci (Zilles et al., 1988). One of the authors of the present paper (MS) has developed an automated method for measuring vertex-wise gyrification in three-dimensional (3D) space across the entire cortex based on magnetic resonance imaging (MRI) data (Schaer et al., 2008).

Findings from MRI studies of gyrification in schizophrenia have been mixed, as reviewed by White and Hilgetag (2011). Both reduced (e.g. Sallet et al., 2003) and increased (e.g. Falkai et al., 2007) GI have been found using manual or automated methods on coronal sections of MR images, and one study did not find significant group differences (Highley et al., 2003). Higher GI in prefrontal cortex has been found among high-risk patients who later developed schizophrenia compared to those who did not (Harris et al., 2004, Harris et al., 2007). Studies using the automated lGI method have shown reduced folding in the right prefrontal cortex among patients with adolescent onset (Janssen et al., 2009) and adult onset schizophrenia (Palaniyappan et al., 2011), and reduced folding in the left insula and medial parieto-occipital cortex in adult onset schizophrenia (Palaniyappan and Liddle, 2012).

Given the discrepant findings in the literature, it is still unclear if, where, and to what extent the cortex is abnormally folded in schizophrenia. In the present study, the automated lGI method was applied to a large group of patients with schizophrenia and healthy subjects drawn from a Swedish sample with predominantly long-term treated patients, and a Norwegian sample with a high proportion of patients with recent onset schizophrenia. Our aim was to test if patients and controls differed in degree of cortical folding across two large independent samples.