Breaking down the genetics of depression using brain endophenotypes

Abstract

The Queensland Brain Institute 2 Thesis AbstractThe objective of the work presented in this thesis was to extend our understanding of depression aetiology and to identify brain MRI endophenotypes of depression. Our work builds on previous research that highlighted the genetic relationships between stress, sex and depression. In addition, MRI measurements are heritable and do not relying on selfreport. As such, they represent promising endophenotypes that could help break down further the genetics of depression.Preliminary work involved reporting depression prevalence in a young adult Australian twin sample (N=2,773) as well as comorbidity with affective and substance use disorders (Chapter 1). In addition, we demonstrated the good psychometric properties of the SPHERE questionnaire, which provides self-reported measures of anxiety-depression and chronic fatigue. Properties of the scores include: i) stochastic ordering on the sum score, ii) limited sex and age bias through adolescence, iii) high internal consistency, iv) good 3 months test-retest, v) moderate heritability, vi) significant association with lifetime MDD, social anxiety and alcohol dependence diagnoses (Chapter 2).In an older sample of Australian twins (N=5,221, Chapter 3), we performed a direct test of the diathesis-stress hypothesis in depression, which states that stressors can mediate the effect of genetic predisposition (diathesis) on the disease risk. We used Polygenic Risk Scores as a measure of diathesis and report a significant interaction with personal stressful life events, supporting the idea that stress and predisposition have a multiplicative effect on the depression liability scale. This interaction may be different between sexes as suggested by the stratified analysis.We then performed 3 analyses of resting-state fMRI from which we showed that: 1) Head motion (HM) is reliable and heritable in addition to being a major confound in fMRI analyses. However, this is unlikely to greatly confound twin studies of resting-state functional connectivity (FC). As a by-product of the analysis we showed that FC in the Broca's areas networks exhibited low to moderate test-retest and heritability (Chapter 4).2) HM is phenotypically and genetically associated with ADHD scores of Inattention and Hyperactivity/Impulsivity. This leads to most head motion image processing options to reduce power and/or to induce a sampling bias in RS-fMRI studies of ADHD. We discussed ways of circumventing this problem, via volume pruning or multivariate analysis (Chapter 5). 33) Nuisance regression (global signal regression, CompCor and HM regression) options in RS-fMRI processing have a moderate impact on the test-retest of regional homogeneity. Independently of the processing options, the low to moderate intra-class coefficients estimated in our sample may be due to insufficient scanning time (5mins 20s) and would impair our statistical power to detect RS-fMRI endophenotypes of depression (Chapter 6).Rather, we focused on structural brain images and contributed sample...