The Neuroanatomy of ASD in a Large and Clinically Heterogeneous Sample – Preliminary Results of the EU-AIMS Longitudinal European Autism Project (LEAP)

Background: Evidence suggests that Autism Spectrum Disorder (ASD) is accompanied by neurodevelopmental differences in brain anatomy and connectivity. Across studies, however, the size of reported effects is often inversely correlated with sample size (i.e. larger effects are typically reported by studies examining a small number of participants). It is thus currently under debate whether findings replicate in large and clinically heterogeneous samples of individuals, and which particular aspects of the cortical architecture underlying ASD are most representative of the broader ASD phenotype in the general population.

Objectives: Here, we therefore examined differences in various aspects of brain anatomy in a large and heterogeneous sample of ASD individuals, acquired across multiple acquisition sites within Europe, in order to (1) determine which neuroanatomical characteristics best represent the ASD phenotype in the general population, (2) to characterize differences in developmental trajectories, and (3) to examine the extent to which demographic variables such as biological sex may affects the neuroanatomy of ASD.

Methods: Across six European sites, we collected structural Magnetic Resonance Imaging (MRI) scans on 350 well-characterized individuals with ASD (n=254 males, n=96 females, mean age = 17.49+5.59 years, FSIQ = 99.28+19.47), and 255 typically developing (TD) controls (n=159 males, n=96 females, mean age = 17.34+5.87 years, FSIQ = 105.13+17.41). All individuals with ASD were diagnosed using gold-standard assessment tools for ASD (i.e. ADI-R, and ADOS). The FreeSurfer software suite (v.5.3) was utilized for data pre-processing, and to derive a set of neuroanatomical features that included total grey and white matter volumes, subcortical volumes, and surface-based morphometric features (cortical thickness (CT) and surface area (SA)). Multivariate general linear models were used to examine (1) between-group differences, (2) differences in neurodevelopmental trajectories on the global and local (i.e. vertex-wise) level of brain anatomy, and (3) diagnosis-by- sex interactions.

Results: Overall, we found that individuals with ASD did not differ significantly from TD controls in total grey and white matter volumes, total intracranial volume, or cerebrospinal fluid (p>0.05, two-tailed). There were also no significant between-group differences in the volumes of subcortical structure following correction for multiple statistical comparisons. On the vertex level, we confirmed that individuals with ASD have significant - and mostly non-overlapping - differences in CT and SA (Figure 1), which mediated differences in regional brain volume. As expected, within our observed age range (6.81–30.78 years), both CT and SA displayed complex (i.e. non-linear) neurodevelopmental trajectories, where ASD individuals significantly differed from controls. Last, we report significant sex-by-diagnosis interactions in surface anatomy, thus confirming that the neuroanatomy of ASD is significantly modulated by biological sex.

Conclusions: Taken together, when examining the neurobiological underpinnings of ASD in a large and clinical heterogeneous sample of individuals, our findings suggest that ASD is best characterized by regional differences in CT and SA. This is of importance as CT and SA have different neurodevelopmental origins, and may therefore be used to stratify individuals into genetically and/or phenotypically distinct subgroups. Moreover, these differences vary across the human life span, and also differ between men and women.