30602
Heterogeneity and Reproducibility of Functional Connectivity in Autism
Objectives: 1) Which functional connectivity features are reproducible across a large multisite sample of participants with autism? 2) Can consensus features of autism that are heterogenous across sites be identified within a single sample using high temporal resolution, long-duration modern acquisition techniques? 3) To what extent do distinct functional connectivity features track together in the same participants vs. representing different aspects or endophenotypes of autism?
Methods: We used a large multi-site dataset (ABIDE 1+2) comprised of a heterogenous population of individuals with autism and typically-developing individuals to compare a number of resting-state functional connectivity features of autism. ABIDE data were aggressively screened for image quality and head motion leaving 1402 subjects from 25 sites, then processed using SPM12 software with motion realignment, coregistration to anatomic image, normalization to MNI space, and regression of realignment parameters and derivatives, white matter, CSF, and soft tissues of the face to mitigate physiological artifacts, followed by volume censoring (scrubbing) for mean head motion greater than 0.3 mm. These features were also tested in a single site sample (n=90) that utilized a high temporal resolution (multiband, multiecho), long-duration (30 min per subject) resting-state acquisition technique, analyzed using AFNI software with multiecho ICA approach for artifact rejection.
Results: Of over 1200 features tested that demonstrated uncorrected p<0.05 across the combined ABIDE sample, no single method of analysis provided reproducible results across research sites, ABIDE 1 and ABIDE 2 samples, and the high-resolution dataset, using a model that included site, age, sex, and mean head motion as covariates. Distinct categories of functional connectivity features that differed in autism such as homotopic, default network, salience network, long-range connections, and corticostriatal connectivity, did not align with differences in clinical and behavioral traits in individuals with autism. Features that emphasized temporal relationships such as lag-based functional connectivity were not correlated to other methods in describing patterns of resting-state functional connectivity and their relationship to autism traits. High temporal resolution single-site sample showed qualitatively similar results across feature types to combined ABIDE (n=1402 subjects) sample.
Conclusions: Overall, functional connectivity features predictive of autism demonstrated striking heterogeneity across sites, with consistent results only for large samples. Different types of functional connectivity features do not consistently predict different features of autism. Rather, specific features that predict autism severity and social dysfunction are distributed across feature types. Findings suggest a need for longer-duration, higher temporal resolution acquisition strategies to improve single subject reproducibility, emphasis on temporal domain of brain function, investigation of cohorts with lower cognitive and verbal abilities, and models that combine many imaging and genetic features to identify clinical subphenotypes.