Brain structure in people at ultra-high risk of psychosis, patients with first-episode schizophrenia, and healthy controls: a VBM study

Abstract

Early intervention research in schizophrenia has suggested that brain structural alterations might be present in subjects at high risk of developing psychosis. The heterogeneity of regional effects of these changes, which is established in schizophrenia, however, has not been explored in prodromal or high-risk populations. We used high-resolution MRI and voxel-based morphometry (VBM8) to analyze grey matter differences in 43 ultra high-risk subjects for psychosis (meeting ARMS criteria, identified through CAARMS interviews), 24 antipsychotic–naïve first-episode schizophrenia patients and 49 healthy controls (groups matched for age and gender). Compared to healthy controls, resp., first-episode schizophrenia patients had reduced regional grey matter in left prefrontal, insula, right parietal and left temporal cortices, while the high-risk group showed reductions in right middle temporal and left anterior frontal cortices. When dividing the ultra-high-risk group in those with a genetic risk vs. those with attenuated psychotic symptoms, the former showed left anterior frontal, right caudate, as well as a smaller right hippocampus, and amygdala reduction, while the latter subgroup showed right middle temporal cortical reductions (each compared to healthy controls). Our findings in a clinical psychosis high-risk cohort demonstrate variability of brain structural changes according to subgroup and background of elevated risk, suggesting frontal and possibly also hippocampal/amygdala changes in individuals with genetic susceptibility. Heterogeneity of structural brain changes (as seen in schizophrenia) appears evident even at high-risk stage, prior to potential onset of psychosis.

Introduction

Early intervention and detection of people at risk of developing psychosis has become a major focus of clinical research on schizophrenia (Yung and Nelson, 2011, Stafford et al., 2013). There are now well-validated clinical instruments (Daneault et al., 2013), which have been used to screen young people at high risk for later onset of psychosis, both with the aim of identifying early intervention strategies, as well as enabling biological research into well-defined high-risk populations. Most of these assessments rely on clinical signs and symptoms, including basic symptoms, the occurrence of brief or attenuated psychotic symptoms, psychometric schizotypy, or biological factors such as familial liability (Addington and Heinssen, 2012, Schultze-Lutter et al., 2012, Daneault et al., 2013). Among the most widely used clinical and research instruments, the CAARMS interview (comprehensive assessment of at-risk mental state (Yung et al., 2002)), for example, considers several clinical factors, including higher genetic load (e.g. first-degree relatives of patients with schizophrenia with drop in functioning), attenuated psychotic symptoms, or brief self-limiting intermittent psychotic symptoms (BLIPS). Any of these factors or “routes” towards an at-risk mental state (ARMS) is considered, and screened subjects might meet criteria for ARMS based on one or more of these aspects (Yung et al., 2002, Daneault et al., 2013). People at ultra-high risk (UHR) for psychosis can therefore be assumed to be a heterogeneous group, independent of whether they eventually convert to schizophrenia or develop another psychiatric condition, since they vary in degrees of genetic liability, symptom profiles, and other phenotypic variables such as cognitive function (Kohler et al., 2014).

Along with the established and validated research criteria for high-risk states for psychosis, there have been several studies investigating neurobiological changes in high-risk populations, including volumetric and voxel-based morphometry (VBM) approaches (for review, see (Jung et al., 2010, Lawrie et al., 2008, Wood et al., 2013)) as well as functional MRI (Fusar-Poli, 2012). Reviews and meta-analyses in this area, however, differ, with regards to the definition and inclusion of high-risk subjects: while some have provided overviews on studies in genetic high-risk relatives (Palaniyappan et al., 2012), others have included studies with a broader spectrum of the high-risk paradigm, including individuals at risk for psychosis not only through affected relatives, but also through either psychometric or subclinical symptom profiles (Chan et al., 2011, Wood et al., 2013).

There has been little research into the biological diversity of subgroups of people within at-risk mental state for the psychosis spectrum, i.e. testing the hypothesis that distinct brain structural changes characterize subgroups of at-risk populations depending on their risk profile. So far, subgroups of UHR subjects have been defined in longitudinal studies according to clinical outcome, i.e. whether brain structural parameters might predict eventual conversion into psychosis, and in particular schizophrenia (Lawrie et al., 2008, Wood et al., 2008, Koutsouleris et al., 2009a, Koutsouleris et al., 2009b, Koutsouleris et al., 2012, Sprooten et al., 2013, Cooper et al., 2014). Although such a distinction is relevant for using brain imaging for prediction or monitoring, it still leaves unanswered the question of heterogeneity within this population of subjects who are at risk of developing a disorder, which in itself is highly heterogeneous with regards to clinical presentation, long-term outcomes, and treatment. Only two studies have divided subgroups according to family history of psychosis, which might be a biologically more plausible discriminant of subgroups: one study, using volumetry and assessment of gross morphological features, found differences with reduced hippocampal volume in those UHR subjects without family history (Wood et al., 2005), another assessed cavum septum pellucidum prevalence and its features, but failed to find differences in UHR subgroups (Takahashi et al., 2008). Hence, to our knowledge, there is no study to assess which regions that are structurally compromised in UHR would be related to or specific for subgroups.

In this study, we aim to test the hypothesis that the biological “route” into high-risk status, i.e. whether through genetic liability or attenuated psychotic symptom profiles, differs with regards to regional grey matter. Using a cross-sectional design, we compared a group of ultra-high risk (UHR) individuals (defined by CAARMS criteria (Yung et al., 2002)) with both healthy controls and people with first-episode schizophrenia; in a second set of analyses, we then divided the UHR group into two subgroups, which were compared to identify changes (each compared to healthy controls and head-on) that would distinguish the two groups. Specifically, we hypothesized diverging effects in the lateral prefrontal, lateral temporal and hippocampal areas identified in the studies mentioned above, while testing voxel-wise across the whole brain to additionally provide an explorative analysis of other brain regions.