Mutations associated with neuropsychiatric conditions delineate functional brain connectivity dimensions contributing to autism and schizophrenia

16p11.2 and 22q11.2 Copy Number Variants (CNVs) confer high risk for Autism Spectrum Disorder (ASD), schizophrenia (SZ), and Attention-Deficit-Hyperactivity-Disorder (ADHD), but their impact on functional connectivity (FC) remains unclear. Here we report an analysis of resting-state FC using magnetic resonance imaging data from 101 CNV carriers, 755 individuals with idiopathic ASD, SZ, or ADHD and 1,072 controls. We characterize CNV FC-signatures and use them to identify dimensions contributing to complex idiopathic conditions. CNVs have large mirror effects on FC at the global and regional level. Thalamus, somatomotor, and posterior insula regions play a critical role in dysconnectivity shared across deletions, duplications, idiopathic ASD, SZ but not ADHD. Individuals with higher similarity to deletion FC-signatures exhibit worse cognitive and behavioral symptoms. Deletion similarities identified at the connectivity level could be related to the redundant associations observed genome-wide between gene expression spatial patterns and FC-signatures. Results may explain why many CNVs affect a similar range of neuropsychiatric symptoms.


Idiopathic schizophrenia
We used fMRI data retrospectively aggregated from 10 distinct sites and studies. Brain imaging multistate data were obtained through either the SchizConnect and OpenfMRI data sharing platforms (http://schizconnect.org 4 ;https://openfmri.org 5 ) or local scanning at the University of Montréal. All patients were diagnosed with SZ according to DSM-IV or DSM-V criteria, as a function of the time of study. Sites samples were obtained after subjects were selected in order to ensure even proportions of SZ patients and controls within each site (from N = 9 to N = 42 per group) and to reduce betweengroup differences with regards to gender ratio (74% vs. 75% males in patients and controls, respectively), age distribution (34 vs. 32 years old on average) and motion levels (averaged frame displacement: 0.16 vs. 0.14 mm). Such matching of SZ and controls subjects was achieved based on propensity scores. In total, we retained 242 SZ patients and 242 healthy controls in statistical analyses.
Depending on the study, positive and negative symptoms were assessed with either with the Positive and negative syndrome scale (PANSS, 6 ) or the Scales for the assessment of positive/negative symptoms (SAPS/SANS, 7 ). In order to allow for group analyses, SAPS/SANS scores were converted into PANSS scores using published regression-based equations 8 .

Idiopathic ADHD
We used data provided by the ADHD-200 Consortium and The Neuro Bureau ADHD-200 Preprocessed repository (8 cohorts http://fcon_1000.projects.nitrc.org/indi/adhd200/ 9 ). Data from seven sites were retained after exclusion of 184 individuals. We included in our study a total of 763 subjects, 289 patients diagnosed with ADHD and 474 healthy controls.
This database provided child and adolescent data. Scores related to ADHD symptoms were measured using Conner's Parent Rating Scale-Revised, Long Version (CPRS-LV 10 ).

Preprocessing
All datasets were preprocessed using the same parameters with the same Neuroimaging Analysis Kit (NIAK) version 0.12.4, an Octave-based open-source processing and analysis pipeline 11 . The first four volumes of each rs-fMRI time series were discarded to allow for magnetization to reach a steady state. Each data set was corrected for differences in slice acquisition time. Head motion parameters were estimated by spatially re-aligning individual timepoints with the median volume in the time series. This reference median volume was then aligned with the individual anatomical T1 image, which in turn was co-registered onto the MNI152 template space using an initial affine transformation, followed by a nonlinear transformation. Finally, each individual timepoint was mapped to the MNI space 12 using the combined spatial transformations. Slow frequency drifts were modelled on the entire time series as discrete cosine basis functions with a 0.01 Hz high-pass cut-off.
Timepoints with excessive in-scanner motion (greater than 0.5 mm framewise displacement) were then censored from the time series by removing the affected timepoint as well as the preceding and following two timepoints 13 . Nuisance covariates were regressed from the remaining time series: the previously estimated slow time drifts, the average signals in conservative masks of the white matter and lateral ventricles, and the first principal components (95% energy) of the estimated six rigid-body motion parameters and their squares. Data were then spatially smoothed with a 3D Gaussian kernel (FWHM = 6mm).

Quality Control
Preprocessed data were visually controlled for quality of the co-registration, head motion, and related artefacts by one rater. Not all six datasets were examined by the same raters, yet all raters followed the same standardized quality-control procedure 14  Additional information on motion for each cohort after preprocessing Supplementary

Sensitivity analyses on age distribution in 16p11.2 deletion carriers
We performed a sensitivity analysis after excluding older controls to obtain identical age distributions (mean 12.7 and 13.0 years respectively) in the 16p11.2 deletions and control groups. The CWAS performed before and after excluding adults provides the same results albeit with decreased power: The 2 beta maps are highly correlated (r=0.967) and their distribution (b) overlaps perfectly.
Supplementary The alterations of each region and network relative to one another remain similar before and after adjustment and both FC-signatures (beta maps) are highly correlated (r=0.98). This also demonstrates that ASD and SZ are associated with a mean shift in connectivity as well as a reorganisation of networks relative to one another.