Mental disorders are the major cause of disease burden and disability in the world. There are many factors that play a role in the development of these diseases and we have a great need for new knowledge. Therefore, identification of underlying pathophysiology is crucial.
In a number of clinical and experimental studies, based on our closely integrated and thematically oriented research organization with complementary research groups, established infrastructure, large clinical cohorts with extensive clinical phenotype characterization and state-of-the-art experimental methods, we have identified new genetic variants and brain phenotypes that associate gene variants with clinical symptoms.
Our partners of the Center for Multimodal Imaging and Genetics at the University of California San Diego (UCSD) have developed the first genetics-based atlas of the human cortex, using that twin data and fuzzy clustering (Chen CH et al., Science 2012). Not only did this confirm that human brain phenotypes are hereditary, it also demonstrated a distinct region-specific genetic pattern in which the cortex can be divided into various pleiotropic regions. In this project, we have been working on a new statistical method to investigate aggregated contributions from all SNPs to phenotypic variation, and which capture a much larger portion of heredity than conventional GWAS (genome-wide association studies). We also looked for SNPs associated with each genetic division to establish a database tool that has information on the contributions of specific genetic variants to phenotypic variation in different parts of the brain. We have also studied SNPs and CNVs associated with each genetic division to establish a database tool that has information on the contributions of specific genetic variants to phenotypic variation in different parts of the brain.
The ongoing challenge has been to integrate different types of data (eg. image processing, genotypes, cognitive and clinical data) and analyze them. For this purpose, we implemented multi-block multivariate methods, ie. Consensus Principal Component Analysis (CPCA) and Multi-block Partial Least Squares Regression (MBPLSR), on MR and SNPs data. Morphometric modes of schizophrenia (SCZ) and bipolar disorders (BD) have been investigated by MBPLSR discriminant analysis (MBPLSR-DA). In this study, MBPLSR-DA was implemented with respect to cortical thickness, white matter area and gray substance density. The results were then mapped back to MR data for further interpretation. We have discovered groupings of SCZ and BD patients. The results are presented in "Classifying schizophrenia and bipolar disorder with multi-block Partial Least Squares Regression analysis of brain MRI measures" manuscript that is "under submission".
We also have integrated genetics with brain based studies by running a pilot study aimed at identifying relationships between various previously reported Autism Spectrum Disorder (ASD) -related modules. We have used CPCA to examine the patterns of genes shared between time of gene expression, cellular / biochemical functions and disorders that are comorbid with ASD. We have generated new groups of ASD genes together with specific comorbidity categories. The manuscript of this study is "under submission" in the Biological Psychiatry journal.
Relevance and benefit to society.
Identifying the underlying disease mechanisms of psychiatric disorders can contribute to
substantial health benefits through the development of new treatment and prevention regimes. A potential breakthrough will have value to other research disciplines, from various brain diseases to studies of normal brain development and function, as well as through and MRI-based diagnostic tools.
It is well documented that multiple biological and genetic causes are involved in ASD . Moreover, ASD lacks clearly defined diagnostic boundaries with other disorders. Therefore, a more comprehensive model that can relate ASD to other disorders and to specific biological or biochemical components would benefit both clinicians and patients.
The overall aim of the current project has been to identify common and rare genetic factors associated to altered brain development, and explore their role in neurodevelopmental disorders. The ongoing challenge was to integrate different types of data, analyze them and interpret the outcomes. We have adopted a novel statistical framework to examine the contribution of all SNPs to phenotypic variation in aggregate. We have identified SNPs that provided enriched genetic association and showed increased replication rates across independent samples. Finally, we have studied genetic variants and mechanisms that are involved in disease or symptom pathogenesis. Together these aims will fill a knowledge gap by uncovering SNPs associated with individual brain regions defined by genetically-based atlases. The information of SNPs associated with brain phenotypes will be further applied to enhance our ability to identify disease-related genetic variants in individuals with psychiatric disorders.
Severe mental disorders usually starts in adolescence and are major public health problems with unknown pathophysiology. Recently we produced the first genetically-based atlas of the human cortex derived from magnetic resonance imaging (MRI) data of twins using fuzzy clustering. This finding not only confirms that human brain phenotypes are heritable traits but also demonstrates a very clear region-specific genetic pattern in which the cortex can be subdivided into various pleiotropic regions. In this pro ject, we will extend this approach further and apply it to a uniquely large in-house sample with both MRI and single nucleotide polymorphism (SNP) data (n=5,000), in order to generate SNP-based atlases. We will adopt a novel statistical framework develope d by Visscher to examine the contribution of all SNPs to phenotypic variation in aggregate. This approach is analogous to twin modeling which examines aggregate genetic influences on a trait, and captures a much larger portion of heritability than the con ventional genome-wide association study approach. However, neither the twin nor the Visscher methods is informative about specific genetic variants underlying each genetic division. Next, we will take the genetic maps and find SNPs associated with each ge netic division to establish a database tool that has information about contributions of specific genetic variants to phenotypic variation in different parts of the brain. Finally, it is well-recognized that psychiatric and substance use disorders affect b oth brain and behavior. We will study genetic variants and mechanisms that are involved in disease or symptom pathogenesis with benefit from incorporating the knowledge of polygenic basis of the brain. Partners include UC San Diego and Univ of Oslo, in ad dition to Oslo Univ. Hosp and deCODE Genetics. The results will provide new understanding of the pathophysiology of psychiatric disorders and form the basis of future treatment studies.