The impact of childhood trauma on structural brain connectivity in schizophrenia: A diffusion MRI and network based study
Costello, Laura
Costello, Laura
Loading...
Publication Date
2021-12-29
Type
Thesis
Downloads
Citation
Abstract
Schizophrenia is a severe and heterogeneous neuropsychiatric disorder affecting approximately 1% of the world’s population. In recent years, exposure to childhood trauma has been established as a prominent environmental risk factor in the aetiology of schizophrenia, with studies demonstrating a higher prevalence of trauma-reported reported among individuals with schizophrenia and more severe clinical symptoms, poorer treatment response and functional outcomes. However, the biological mechanism underpinning this association has not yet been established. Advancements in in-vivo MRI techniques have helped to highlight the involvement of white matter microstructural and network-level structural connectivity impairments among individuals with a diagnosis of schizophrenia across all stages of the disorder. Structural brain differences have also been documented among non-clinical populations with a history of maltreatment, relative to unexposed individuals, leading some researchers to speculate about the potential effects of trauma-exposure on the brain and whether this may confer an increased neurobiological vulnerability for the development of schizophrenia. Although limited, some studies have started to examine the relationship between trauma-exposure and white matter alterations in schizophrenia. However, heterogeneous methodological and clinical factors in addition to technical limitations limit the interpretability of the findings. Further, the majority of findings to date use measures of white matter organisation that are not sensitive to changes at the tissue level or fail account for the brains integrative network structure encompassing both grey and white matter regions. Building on existing neuroimaging research, this thesis aims to address several existing gaps in the literature by examining the relationship between reported experiences of childhood trauma, white matter microstructure, and network-level connectivity impairments in adults with a diagnosis of schizophrenia using traditional measures of anisotropy (Manuscript 1), novel applications of a free water elimination model (Manuscript 2) and a graph-theory based network analyses approach (Manuscript 3). After confirming existing microstructural and connectivity alterations in individuals with a diagnosis of schizophrenia, I hypothesised that the relationship between reported experiences of childhood trauma, white matter microstructure and impairments in structural brain network organisation would be most pronounced in the group with a diagnosis of schizophrenia and a history of childhood trauma relative to trauma-exposed controls and unexposed individuals with a diagnosis of schizophrenia or healthy controls. Self-reported experiences of childhood trauma that may have occurred between birth and eighteen years of age were assessed retrospectively using the 28-item Childhood Trauma Questionnaire Short-Form (CTQ-SF) for all participants. The CTQ manual cut-off scores were first used to categorise the severity of trauma-exposure as none/low/moderate, or severe across each of the five CTQ subtypes (emotional abuse, physical abuse, sexual abuse, emotional and physical neglect). Following this categorisation, high levels of trauma-exposure was defined as the presence of moderate-severe abuse and/or neglect. In contrast, ‘none or low levels of trauma-exposure’ was defined as meeting the criteria for none or low across each of the five CTQ subscales in addition to a single case of moderate or severe physical/emotional neglect. All participants underwent T1-weighted structural and 32-gradient direction diffusion-weighted magnetic resonance imaging (MRI, 3T, b = 1000s/mm 2, Philips Medical Systems, Best, The Netherlands). In manuscripts 1 and 2, estimation of eigenvalues from the diffusion-weighted MR image was performed using RESTORE (ExploreDTIv4.8.6) and sample-specific registration. In manuscript 1 (Chapter 4), a general linear model was designed as a Multivariate Analysis of Covariance (MANCOVA), covarying for age and sex to examine the relationship between trauma-exposure and group-differences in anisotropy within deep white matter voxels (Tract-Based Spatial Statistics; TBSS, FSL v 6.0.1) between healthy controls and individuals’ schizophrenia. In manuscript 2 (Chapter 5), a novel free water elimination script (Matlab v 2017b) was applied to participants diffusion-weighted MRI data to separate the diffusion signal into two components reflecting 1) fractional volume of extracellular free water (FW) and 2) tissue-specific measures of anisotropy following the elimination of extracellular free water (FAT). Covarying for age and sex (MANCOVA GLM), I examined the relationship between trauma-exposure, and group-differences in measures of FW and FAT within deep white matter voxels (TBSS, FSL v 6.0.1) between healthy controls and individuals with schizophrenia. In the third and final manuscript (Chapter 6), participants structural connectivity matrices were constructed from ‘nodes’ (AAL-90 template atlas) derived from T1-weighted structural MR images and ‘edges’ were derived from whole-brain and non-tensor-based diffusion-MRI tractography (constrained spherical deconvolution; CSD). Covarying for age and sex, a MANCOVA was designed to examine group differences across both weighted (fractional anisotropy; FA, number-of-streamlines; NOS) and unweighted whole-brain measures of structural topology. A permutation-based statistical analyses (network-based statistics, NBS) was also used to examine the effects of trauma-exposure across connected anatomical subnetworks in the schizophrenia group relative to controls. In all three empirical studies, I confirmed that having a diagnosis of schizophrenia (n=37) relative to controls (n=129) was associated with white matter microstructural reductions across both conventional measures of anisotropy (p < 0.05, TFCE, Chapter 4), localised reductions in tissue-specific measures of anisotropy (p < 0.05, TFCE, Chapter 5), as well as structural connectivity impairments across whole-brain measures of network topology (efficiency, strength, clustering, density and increases in path length and betweenness), and dysconnectivity within anatomical subnetwork involving frontoparietal, frontotemporal and occipital connection involved in the default mode network (p < 0.05, Chapter 6). Despite these detected differences, a history of high levels of trauma-exposure (n=13) did not relate to any significant neuroanatomical differences in schizophrenia across either i) conventional measures of anisotropy (p > 0.05, TFCE, Chapter 4), ii) tissue-specific measures of anisotropy (Chapter 5), iii) whole-brain measures of structural network topology or a differentially connected anatomical subnetwork (p > 0.05, TFCE, Chapter 6). Irrespective of diagnosis, however, the findings from this thesis demonstrated that high levels of trauma-exposure, irrespective of diagnosis (n=48) was associated with reductions in FA (p < 0.05, TFCE, Chapter 4), localised reductions in FAT within fronto-thalamic regions and widespread increases in FW throughout white matter tissue (p < 0.05, TFCE, Chapter 5). However, these differences in white matter tissue did not extend to encompass any significant differences across any whole-brain measures of structural network organisation or connectivity when graph theory was used to map individuals' structural brain networks (p > 0.05, Chapter 6). Despite confirmation of white matter microstructure and network-level connectivity differences in the group with a diagnosis of schizophrenia relative to controls, the results from all three empirical studies included in this thesis suggest that reported experiences of trauma-exposure are not related to microstructural or network-level impairments in connectivity among individuals with schizophrenia. While it is not possible to draw any firm conclusions based on these cross-sectional findings, we propose that given that schizophrenia is a highly complex and multifaceted disorder, additive rather than interactive effects of trauma-exposure is more likely to explain the possible role of trauma-exposure as a driver of increased risk in the aetiology of schizophrenia. Furthermore, the results from this thesis extend previous findings highlighting the involvement of both cellular white matter microstructure in participants with early-life adversity, irrespective of diagnosis. In particular, the novel observation of an association between trauma-exposure and widespread increases in free water is indicative of a possible inflammatory process that alters the extracellular space. These findings also raise the possibility that neuroanatomical alterations present among controls may underpin compensatory or adaptive mechanisms relating to early life adversity. Considering the brains network structure as a dynamic system that is continually refined by genetic and environmental factors, future studies should aim to develop a more comprehensive and multi-scale understanding of how various biological and psychosocial risk factors may interact over time to underpin mechanisms of increased risk or resilience to the development of schizophrenia.
Publisher
NUI Galway