Imaging genetics evidence for the neurodevelopmental model of schizophrenia

Abstract

This thesis applies a wide range of imaging genetics methods to disentangle the complex genetic relationship between brain structural phenotypes and schizophrenia. Chapter 1 reviews current imaging and genetic evidence for the neurodevelopmental model of schizophrenia. The neurodevelopmental model of schizophrenia posits that genetic variation and interactions with early environmental risk factors impact long-lasting brain developmental processes, which ultimately predispose individuals to the development of the disorder in early adulthood. In line with the neurodevelopmental model, schizophrenia is highly heritable, and genetic risk factors suggest that pathogenesis of the disease begins during early neurodevelopment. Concurrently, a large body of neurobiological research has consistently reported brain structural abnormalities measured using magnetic resonance imaging (MRI) in schizophrenia patients. However, links between the genetic risk for schizophrenia and schizophrenia-associated brain structural abnormalities have thus far been lacking. Understanding the genetic relationship between schizophrenia and brain structural abnormalities is thus a crucial aspect of the neurodevelopmental model of schizophrenia, and for unraveling the biological mechanisms underlying schizophrenia in general. Chapter 2 investigates whether the genetic risk for schizophrenia measured using poly- genic risk scores (PRSs) is associated with multiple micro- and macrostructural MRI metrics measured at each of 180 cortical areas, seven subcortical structures and 15 major white matter tracts, in a large sample of the UK Biobank (N = 29,878). Polygenic risk was significantly associated with reduced neurite density index (NDI) at global brain scale, at 149 cortical regions, five subcortical structures and 14 white matter tracts. Other micro-structural parame- ters that were correlated with NDI, e.g., fractional anisotropy (FA), were also significantly associated with PRS. Genetic effects on multiple MRI phenotypes were co-located in the temporal, cingulate and prefrontal cortical areas, as well as the insula and hippocampus. Risk-related reduction in NDI is plausibly indicative of reduced density of myelinated axons and dendritic arborization in large-scale cortico-subcortical networks. Cortical, subcortical and white matter micro-structure may therefore be linked to the genetic mechanisms of schizophrenia. Chapter 3 examines the genetic architecture of normative cortical MRI phenotypes, two macro-structural and one micro-structural, as well as the genetic relationship between individual brain regions. We accessed genome-wide association studies (GWASs) for surface area (SA), cortical thickness (CT) and NDI measured at 180 cortical areas (N = 36,843,UK Biobank). Using Hi-C coupled MAGMA (H-MAGMA), we identified 318 genes that were significantly associated with SA, 157 genes with CT, and 86 genes with NDI. Genes associated with each MRI metric shared transcriptional trajectories that peak during mid-late periods of fetal life and were enriched for neurodevelopmental processes. Additionally, genetic similarity networks of cortical regions were strongly coupled to their structural covariance networks, suggesting that structural covariance between regions represents close equivalence in the genetic determinants of their development. Chapter 4 investigates whether there is evidence for pleiotropic associations between genes and both normative brain regional phenotypes and schizophrenia. We accessed GWASs of schizophrenia (N = 69,369 cases; 236,642 controls) and, using Hi-C-coupled MAGMA, showed that 61 genes were significantly associated with both schizophrenia and one or more normative MRI metrics analysed in Chapter 3. Whole-genome analysis with partial least squares demonstrated significant genetic covariation between schizophrenia and SA, CT or NDI of most cortical regions. Additionally, genetic covariation between schizophrenia and brain phenotypes was strongest in the hubs of structural covariance networks. Pleiotrop- ically associated genes were enriched for neurodevelopmental processes and positionally concentrated in chromosomes 3p21, 17q21 and 11p11. Parallel analyses of GWAS on bipolar disorder and Alzheimer’s disease showed that pleiotropic association with MRI metrics was stronger for schizophrenia compared to other disorders. In Chapter 5 we investigate whether the associations between schizophrenia and brain structural MRI metrics identified in Chapter 2 and Chapter 4 represent causal pathways. We conducted bi-directional Mendelian randomization (MR) on a total subset of 103 regional MRI phenotypes measured using NDI, SA or CT. We report significant causal effects between lower thalamic NDI and increased risk of schizophrenia, and between higher SA of the posterior cingulate cortex and increased risk of schizophrenia. In line with the neurodevelopmental model, these findings are preliminary evidence for the traditional causal pathway in which genetic risk for schizophrenia is mechanistically and/or proximally mediated by its impacts on brain structure over development. Finally, Chapter 6 contextualises these findings with the neurodevelopmental model, and discusses current challenges as well as future directions in the field.