Brain Morphometry of Dimensional Autism: A Twin Study

Background:  

Structural morphology related to Autism Spectrum Disorder (ASD) might give valuable insights into its etiology and contribute to the identification   of biomarkers for diagnostic and treatment purposes. However, though many studies do report abnormal brain morphology, the structural imaging literature on ASD has shown inconsistent results. The latter is probably due in part to the small sample sizes but also to the complex interaction between brain structure and age, gender, IQ and phenotypic heterogeneity  in ASD (Haar et al., 2014; Lefebvre et al., 2015). Moreover, most previous research used a volume-based approach, which only partly characterizes brain morphology, and being less sensitive to subtle alterations in anatomy (Raznahan et al., 2013). Twin design is one way to overcome some of these issues, but to date there have been  few twin-based imaging studies, of which most have used limited  sample sizes (Mevel et al., 2014).

Objectives:  

Our aim was to investigate the relation between both volume- and surface-based brain morphology and a continuous estimate of autistic traits in a population-based sample enriched of broader phenotypic autism and clinical ASD.

Methods:  

T1 weighted anatomical images of 116 subjects aged 8-23 years, both mono- and dizygotic twins, were acquired within the Roots of Autism and ADHD Twin Study Sweden (RATSS) (Bölte et al., 2014). Within twin-pair differences in estimates of lobar volume (FSL, N=116) and gyral volume, surface area and cortical thickness (Freesurfer, N=48) of regions previously implicated in ASD were correlated to within twin-pair differences in total scores on the parent-report Social Responsiveness Scale (SRS). The results from the within twin-pair conditional regression model were compared to a regular linear regression model across all subjects, which is more sensitive to differences in age and gender.

Results:  

Higher SRS scores were associated with an increase in white matter volume in the Right Occipital lobe (z=2.1, p=0.04), and Bilateral Parietal lobes (left: z=-2.3, p=0.02 and right: z=-3.4, p=<0,001), using within twin pair differences on volume-based morphometry. Surface-based analysis confirmed these findings: higher SRS scores were related to 1) an increase in gray matter volume and surface areas in the right Lateral Occipital gyrus (respectively z=2.3, p=0.02 and z=1.8, p=0.066),  and 2) a decrease of gray matter volume and surface areas in the left Inferior Parietal gyrus (respectively z=-2.4, p=0.02 and z=-3.8, p<0.001). No differences in cortical thickness were related to SRS score in these regions. Only the associations with the Right Occipital lobe white matter volume were repeated using the linear regression model.

Conclusions:  

A higher score on the SRS, indicating increasing autism traits, seems to be related to altered volume and surface area in sensory regions namely Occipital and Parietal, when using a within twin-pair design, but not when using a conventional linear regression model. Crucially, twin models allowed us to detect brain structure alterations related to ASD that are sensitive to differences in age, gender and IQ. Our results further add evidence that alterations in cortical thickness and surface area may independently contribute to changes in brain volume and ASD pathology.