25357
A Multi-Site Investigation of Functional MRI Responses in the Longitudinal European Autism Project (LEAP) Cohort
Objectives: As part of the Longitudinal European Autism Project (LEAP), we aim at identifying markers of aberrant neural processing in four major cognitive systems (emotion processing, Theory of Mind [ToM], executive control, reward processing) using task-based functional magnetic resonance imaging (fMRI) in a large cohort of ASD and typically developing (TD) subjects.
Methods: A task battery was designed which taps into core ASD deficits, allows for the inclusion of children and less able subjects and meets test-retest reliability criteria for the use in an accelerated longitudinal design including two measurement time points. FMRI was performed at six European centers following standard operations procedures. A total of 672 subjects (6-30 years, 58% ASD, 69% male) participated in the fMRI assessment and tasks were selected according to the individual’s ability level. The task battery included a social and monetary reward task, a spontaneous mentalizing task involving animated shapes, an emotional face-matching task, and a flanker/GoNogo task. Baseline assessment was performed between April 2014 and September 2016, and a 12-24 month follow-up is currently under way. A quality control pipeline has been implemented and applied to the data in order to assess effects of site, diagnosis and age on raw data quality. First pass analyses have been performed using standard processing routines implemented in SPM12 (http://www.fil.ion.ucl.ac.uk/spm/).
Results: Quality assessment identified prominent effects of site on raw data QC metrics, with no or only modest interaction with diagnosis and age. Motion was elevated in ASD subjects and in younger children across tasks. Preliminary statistical analyses performed within the framework of the General Linear Model revealed robust activation of each of the networks of interest, while simple case-control differences did not pass the significance threshold (p<0.05, family-wise error corrected across the whole brain).
Conclusions: The LEAP cohort represents the largest European task-based fMRI data set on autism. The employed tasks successfully engaged functional activation within four major cognitive systems. In addition, first-pass analyses suggest that a thorough investigation of different sources of variance is warranted. This not only includes the control of various sources of noise (e.g. between-site differences, motion), but also the application of sample stratification procedures in order to address the hypothesized heterogeneity of the sample (e.g. symptom profiles, interaction with sex and age). These analyses will set the stage for an in-depth investigation of the autism phenotype in various functional imaging measures, ranging from regional activation to network-based connectivity metrics.