Neuro-anatomical and Functional Substrates in Autism Spectrum Disorder

Autism Spectrum Disorders (ASD) are highly heritable, lifelong, neurological disorders, typically presenting within the first three years of life. Each of these disorders is behaviourally characterized by developmental impairments in social interaction and communication and by a limited range of interests and activities. Despite these theoretically well-defined core symptom categories, the expression of the disorder in individual patients varies greatly. The term spectrum disorder refers to the broad spectrum of clinical presentations. Symptoms may range from mild to severe and may fluctuate, abate, change, and even disappear over time. There has been a clear and substantial rise in reported cases of ASD over the past few decades. However, it remains uncertain whether this rise is the result of changes in diagnostic methodologyor reflects a true increase in the frequency of the disorder. To answer this question adequately, an understanding of the biological substrate of ASD is required. The existing literature describes a wide range of structural and functional alterations in the brain of autistic patients. The above described inherent heterogeneity of ASD is probably one of the most important confoundingfactors. In an attempt to deal with this phenotypical heterogeneity a categorical distinction into subclasses was proposed by the American Psychiatric Association (DSM-IV-TR): autistic disorder, pervasive developmental disorder not otherwise specified (PDD-NOS), Asperger syndrome, Rett syndrome and childhood disintegrative disorder (CDD). The last two disorders are very rare, and are typically not considered in the autism spectrum.In the search for anatomical and functional defects in the brain, however, the subclasses proposed in the DSM-IV-TR have proven of limited help. Recent findings suggest that the search for a biological explanation of single (endo)phenotypical markers, may lead to more consistent insights in the complex neurobiological basis of autism. As a consequence, investigators have moved away from the former categorical classifications and have started to use quantifiable andobjective continuous variables or (endo)phenotypes as the focus of their research. Language impairments embody a fundamental component of the autism phenotype and represent a core deficit in ASD individuals. Although the classical description highlights pragmatic language aspects (i.e. language usage, prosody, understanding and production of gestures), recent linguistic studies clearly demonstrate that language deficits in ASD often also encompass the semantic and syntactic fields and sometimes even the phonological domain. Language impairments in the latter areas are often designated as structural language deficits. These structural language problems can objectively be measured using standardized and validated language tests. In this thesis we used structural language as a behavioural endophenotype. To this end we selected a subgroup of ASD patients with a comorbid structural language impairment (ASD-LI). This selection resulted in a more homogeneous subgroup, well defined by objective, well quantifiable and continuous language parameters. By the use of this endophenotype we aimed to unravel a more consistentbrain-behaviour relationship. Complementary findings in neurocognitive, genetic and brain imaging research have recently suggested that ASD may be considered a disorder of brain connectivity. It is hypothesizedthat, as a result of a miswiring of connections in the developing brain, integration and synchronization of distinct brain regions may be defective, ultimately leading to an improper flow of information.Diverse studies have provided indirect evidence for abnormal brain connectivity in autism. The work in this thesis is an attempt to demonstrate in a more direct way that autism is a disorder of brain connectivity and to visualize and quantify the underlying defects, using non-invasive, advanced magnetic resonance imaging (MRI) techniques. To do so, we focused on defects in the language system in a preselected (more homogeneous) subgroup of ASD patients (ASD-LI). Diffusion tensor imaging (DTI) and functional magnetic resonance imaging (fMRI) techniques were used to assess structural and functional brain connectivity, respectively. The DTI technique is a relatively new MRI technique that allows the evaluation of the orientation and integrity of white matter fibres in vivo by measuring the normal diffusion of water molecules in the brain. In this thesis DTI was used to investigate the white matter microstructural properties of major white matter pathways, with particular attention for the language pathways. Quantitative metrics such asthe fractional anisotropy and mean diffusivity were used to evaluate white matter tract integrity providing a measure for structural connectivity. Functional connectivity was assessed using resting state fMRI. This technique examines the temporal coherence of low frequency BOLD fluctuations in functional brain networks. A loss in the synchrony of activation within a particular network denotes a functional disconnection between the cortical regions involved. To identify the cortical regions involved in the language network in a reliable way we performed a task-based fMRI experiment and analysis. Former studies using DTI in ASD patients have demonstrated some evidence of white matter integrity loss in several white matter regions. The alterations described, however, display a considerable degree of variability. Here we searched for white matter alterations in a more homogeneous subgroup of ASD-LI patients. Three different methods were used to evaluate the white matter microstructural properties of the autistic brain: tract-based segmentations inatlas space, trackings in individual space and voxel-based analyses. Tract-based analyses in individual space and in atlas space, revealed no significant alterations in the mean white matter characteristics of major white matter bundles. At the voxel level only subtle differences could be detected. The clusters were generally quite small and did not survive corrections for multiple comparisons. Considering the limitations of voxel-based analyses we concluded that, at least in the subgroup of ASD-LI patients, DTI did not provide a clinically relevant biomarker. While in the ASD-LI group structural language impairments are part of a global developmental disorder, similar language problems can also be encountered in an otherwise normal developmental trajectory (Specific Language Impairment, SLI). The debate about a possible overlap of the neurobiogical origin of the language deficits in ASD-LI and SLI patients is ongoing. To examine whether a similar neuroanatomical substrate underlies this phenotypical linguistic overlap, we compared in a second part of this thesis the white matter microstructural properties of the superior longitudinal fascicle in SLI and ASD-LI patients. The superior longitudinal fascicle can be considered as a major tract in language processing, as it serves the expressive - receptive language areas connection. DTI revealed a clearly affected superior longitudinal fascicle in the children with an elementary language problem. In contrast, no microstructural differences could be demonstrated when comparing the quantitative diffusion metrics of the superior longitudinal fascicle of the ASD-LI patients with those of the age-matched controls, indicating that language impairments common to ASD-LI and SLI involve, at least in part, different biological mechanisms. As the DTI study did not provide adequate answers on the neurobiological origin of the language problems found in ASD-LI, we assessed in a last part of this thesis the functional connectivity properties of the language network using resting state fMRI. To confidently identify the cortical areas involved in language processing, we first performed a short task-based fMRI sequence, using a verb-generation task. A conjunction analysis of the group-activation maps of ASD-LI patients and controls identified language components commonly active in both groups. Eight joint language components were identified and subsequently used as seeds in a resting state analysis. Interestingly, both the interregional and the seed-based whole brain connectivity analysis confirmed a preserved connectivity between the classical language centres, the Wernicke s and Broca s area, an observation that is in agreement with the findings at the microstructural levelobtained by DTI. Conversely, a marked loss of connectivity was observed between the right cerebellar region and the supratentorial regulatory language area. This finding suggested a critical role for the cerebellum in the impaired language functioning of ASD-LI patients. Moreover a negative correlation was observed between the severity of the autism and the strength of theconnectivity between the cerebellar seed region and one of the supratentorial modulatory language regions suggesting that this connectivity defect is directly linked to the autistic phenotype. In conclusion: we have tried to unravel the neurobiological basis of ASD by applying advancedneuroimaging techniques to a well-defined, more homogeneous subgroup of ASD patients with language impairment. The final goal of this approach was to identify structural or functional abnormalities that can be used as biomarkers to improve the diagnosis of ASD and to provide better care and therapy for individual patients. Our work provided a number of new insights into the neurobiology of ASD. Firstly, the use a more homogeneous subgroup resulted in the disappearance of most of the variable and inconsistent white matter findings described in the literature. A lot of these findings may probably be attributed to heterogeneity of the included study population. Therefore, we consider the DTI technique as a less suitable approach to find biologicalmarkers for ASD(-LI). Secondly, the DTI technique does appear to possess the appropriate sensitivity to differentiate, based on WM alteration, two phenotypically overlapping groups. The biological basis ofthe language problems in the SLI group is clearly different from that of the comparable language problems in the ADS-LI group.</&gt;Thirdly, both our DTI data and our fMRI data support the contention that the language impairment in ASD-LI is not related to disturbances at the level of the cortical connections between the frontal and the temporal language areas. Interestingly, however, our fMRI data link the language impairment in ASD-LI to a relative dissociation of the cortical language system from normal cerebellar control, suggesting that the disturbed development and function of the language system in ASD-LI may be due to a loss of the normal modulatory control and automation function ofthe cerebellum. Further and prospective studies are needed to evaluate the potential of resting state cerebellar dysconnectivity as a clinically relevant biomarker for ASD-LI. The microstructural characteristics of the cortico-cerebellar connections in ASD patiens also warrant more detailedinvestigation.status: publishe