Neuroanatomical Differences Between ASD Patients and Controls in the Abide Cohort

Background: The aetiology of Autism spectrum disorders (ASD) appears to be very heterogeneous. In addition to an environmental component, recent studies suggest that hundreds of different genetic mutations may also trigger an ASD phenotype. On the other hand, the central nervous system shows a remarkable plasticity and robustness, and is in many cases capable of compensating important perturbations. The possibility of studying large populations is then critical if we aim at distinguishing the subtle pathological traces that may differentiate the ASD neuroanatomy from the huge neuroanatomical diversity characteristic of humans.

Objectives:  We have analysed various neuroanatomical parametres in a large, multicentric, freely available, cohort of high-functioning ASD patients (N=525) and matched controls (N=560) – the ABIDE project. Our objective was to characterise the neuroanatomical variability in the ASD group within the context of normal neuroanatomical diversity.

Methods:  We used validated automatic segmentation tools (Afni, FSL, FreeSurfer, and our own software) to obtain measurements of intracranial volume, brain volume, total grey and white volume, lobar volumes, volumes of different subcortical structures as well as cortical surface reconstructions of the pial surface and the white/grey matter interface. The original datasets and the results of the automatic analyses were visually quality controlled. All measurements were log-transformed and we used a general-linear model to account for the confounding effects of age, sex and scanning centre. We performed univariate comparisons of the differences in mean values between cases and controls, and used F-tests to compare differences in dispersion. We are currently using multivariate analyses to study the pattern of variability between cases and controls, as well as analyses of allometric scaling for regional volume, cortical surface extension, gyrification and thickness.

Results:  Our first analyses do not reveal significant differences in mean volume of the different structures, however, we observed a statistically significant larger dispersion within the ASD group compared with controls.

Conclusions:  The availability of large open databases is a major step forward in the study of the structure and function of the brain of patients with ASD. Being able to study this large cohort will allow the community to have a common ground to test different methodological approaches and validate previous results obtained in smaller case/control groups. We performed an extensive neuroanatomical analysis of the ABIDE cohort, which we plan to make open to the community. Our results did not show the larger incidence of macrocephaly or total brain volume observed in previous studies, however, we observed a statistically significant larger variability. This larger variability may be the sign of a system failing to compensate a too important perturbation during development. Further analyses, including multivariate analyses of the patterns of neuroanatomical variability, should allow us to better characterise these finding.