Schizophrenia is a devastating neurodevelopmental disorder that assaults the afflicted with visual and auditory hallucinations, confusion and cognitive problems, emotional withdrawal, a lack of motivation, a misunderstanding of what constitutes appropriate social behavior, and a lower life expectancy. While the majority of schizophrenia cases are of unknown etiology, recent work has highlighted the impact that rare genetic mutations can have on an individual’s predisposition to developing a psychotic disorder. 22q11.2 Deletion Syndrome (22q11DS, aka velocardiofacial syndrome or DiGeorge syndrome;), resulting from the loss of a 1.5-3 Megabase portion of chromosome 22, is one such disorder; it imparts a greatly increased risk of psychosis (up to 30%), amongst other psychiatric disorders, for those with the deletion. Concurrently, the use of non-invasive neuroimaging techniques such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) to investigate the brain-behavior relationship has rapidly grown to the method of choice in the quest to characterize the neural correlates of normal and abnormal cognition; a large body of literature now exists that reports alterations to functional brain network (both task-oriented and resting-state) across a range of psychiatric disorders. However, despite the clear utility in studying a disorder with a known genetic etiology and the neuroscience community’s growing interest in probing the intrinsic connectivity of the human brain and how it relates to behavior and cognition, very little research to date has attempted to characterize the dynamics and extent of resting state networks (RSNs) in 22q11DS. Accordingly, we sought to explore the 22q11DS-related alterations to RSN (if any) via a range of analysis methods, and to uncover the cognitive and behavioral correlates of aberrant connectivity within these RSNs.
An initial region-of-interest-based exploration of the Default Mode Network (DMN), in a cross-sectional analysis of 22q11DS subjects and controls (N=77), showed evidence of weakened long-range connectivity in this network for 22q11DS subjects relative to controls. The strength of long-range connectivity was inversely correlated with scores on the social responsiveness scale, such that individuals with more robust DMN connectivity exhibited improved social behavior. Subsequently, we commenced a model-free investigation of resting state network connectivity in youth with 22q11DS and matched control subjects (N=66), utilizing spatial Independent Components Analysis (ICA) to parse the observed variance in the subjects’ data into multiple RSNs at once. Upon identifying networks of interest, rigorous statistical testing yielded group differences of significant within-network hypoconnectivity in 5 RSNs: ACC/Precuneus network, Executive network, DMN, Posterior DMN and Salience network. No cortical RSN tested showed any evidence of within-network hyperconnectivity in 22q11DS. Concurrently, each of the identified RSNs was vectorized and used to train and cross-validate a sparse diagnostic classifier to differentiate between 22q11DS and controls on the basis of their within-network connectivity alone. Of all RSNs, the within-network connectivity differences between subjects were most robust in the DMN, allowing the associated diagnostic classifier to partition the groups with 100% accuracy during cross-validation. Additionally, the DMN-derived classifier could identify the presence of psychotic symptoms dimensionally and further partition the 22q11DS subjects into low-risk and high-risk cohorts on the basis of their DMN hypoconnectivity alone. The findings of RSN hypoconnectivity and DMN utility as a classifier were replicated in an independent cohort of 22q11DS subjects and controls (N=56), showing that cortical RSN hypoconnectivity, particularly within the DMN, is a stable distinguishing feature of 22q11DS. Finally, a targeted analysis of thalamocortical connectivity, in a cross-sectional cohort (N=79) as well as a subsample of individuals with longitudinal data (N=26), was performed to characterize the subcortical-cortical connectivity differences between 22q11DS and controls. Results were largely consistent with current literature in idiopathic schizophrenia and mouse models of 22q11DS, showing hyperconnectivity between the thalamus and bilateral sensory cortices, and hypoconnectivity between the thalamus, striatum, occipital cortex and cerebellum. The observed hyperconnectivity between thalamus and sensory cortex became significantly exacerbated with time in the patient group, and could be used to predict psychotic symptom severity and prodromal risk status in 22q11DS subjects.
Overall, this collected work provides evidence of aberrant within-RSNs connectivity in 22q11DS and lays the groundwork for future investigations into how changes in developing
brain networks can predispose an individual to psychosis.