Aberrant dynamics of cognitive control and motor circuits predict distinct restricted and repetitive behaviors in children with autism

Introduction
Restricted and repetitive behaviors (RRBs) have long been recognized as a core symptom of autism spectrum disorders (ASD). RRBs are the earliest detectable behavioral predictors of ASD and have adverse long-term consequences for acquisition of crucial life skills in individuals with the disorder. Critically, recent changes to the Diagnostic and Statistical Manual of Mental Disorders have identified RRBs as central to understanding heterogeneity of clinical presentations in ASD.
However, RRBs remain a grossly understudied aspect of ASD research and the underlying brain circuits are unknown.
RRBs include behaviors such as preoccupation with objects, ritualized patterns of behavior, highly restricted/fixated interests and stereotyped/repetitive motor (RM) movements. Although RRBs were traditionally defined as a unitary construct, there is growing evidence that RRBs are a heterogeneous construct that can be factored into three distinct phenotypic characteristics: circumscribed interests (CI), insistence on sameness (IS), and RM actions.
...

Here we address critical gaps in our knowledge regarding heterogeneous expression of RRBs in childhood ASD using dynamic brain circuit analysis. We test the hypothesis that brain circuit dynamics underlying RRB symptoms can be dissociated, and specifically that, aberrant cognitive control circuit dynamics would underlie CI and IS symptoms whereas aberrant motor circuit dynamics would underlie RM symptoms. We characterize the dynamic properties of two distinct brain circuits: (i) a cognitive control circuit consisting of salience (SN), central executive (CEN), and default-mode (DMN) network nodes that play a key role in salience detection, allocation of attentional resources, and flexible behavior and (ii) a motor circuit, consisting of cortical (cMN) and subcortical (sMN) motor network nodes important for implementing motor planning, control, and execution (Fig. 1a). We predicted that compared to TD children, children with ASD would show less flexible, aberrant time-varying brain circuit dynamics. In addition, we predicted that aberrant dynamics of the cognitive control circuit would be associated with CI and IS symptoms of RRB, but not RM symptoms and that aberrant dynamics of the motor circuit would be associated with RM symptoms of RRB, but not CI and IS symptoms. Finally, we predicted that, compared to static functional circuits, dynamic functional circuits would better distinguish and predict distinct RRB clinical symptoms.
..
Results
Our dynamic connectivity analysis revealed that children with ASD have less flexible cognitive control circuit dynamics, characterized by brain states with impaired coupling of the SN with CEN and DMN, consistent with findings from a recent study that reported that adults with ASD show dominant neural states with aberrant functional interactions between SN and CEN and between SN and DMN. Notably, we found that CI and IS symptoms of RRB were associated with the degree of inflexible interactions between the three key cognitive control networks: SN, CEN, and DMN. Specifically, severity of CI and IS symptoms was associated with aberrant temporal engagement of the SN with the CEN and DMN. Notably, no such relation was found with RM symptoms pointing to the specificity of our findings with respect to cognitive inflexibility. Critically, no static time-averaged functional connectivity measures predicted CI or IS symptoms. These findings demonstrate that aberrant circuits dynamics associated with SN, CEN, and DMN carry clinically relevant neurobiological signatures of cognitive, but not motoric, components of RRB.

Cross-network interactions between the SN, CEN, and DMN play a key role in effectively responding to dynamic demands of changing environment. In particular, interactions of the SN with the CEN and the DMN are thought to facilitate switching between externally-oriented attention and internally-oriented mental processes in response to salient events to guide flexible behavior. Our dynamic network analysis revealed that this switching is impaired in children with ASD, and that the degree of impairments predicts cognitive inflexibility. These results are consistent with and extend previous studies based on static time-averaged measures demonstrating hyper-connectivity within the SN, CEN, and DMN in children with ASD. Aberrant functioning of the anterior insula node of the SN in ASD may be a key mechanism contributing to inflexible circuits and behaviors given its key role as causal hub for switching between these networks. Together, results suggest that reduced cross-network interactions in the cognitive control circuit contribute to core phenotypic features and inflexible behaviors such as intense focus, unusual attachment to objects of interest, and difficulty with changes in the environment, and provide support for a neurocognitive model of RRB in ASD based on dynamic circuit properties.

Analysis of motor circuit dynamics revealed a different pattern of association with specific RRB phenotypic features. Children with ASD had less flexible motor circuits characterized by stronger intermittent coupling between sMN and cMN. Moreover, we found a strong association between aberrant motor circuit dynamics and RM, but not CI and IS, pointing to the specificity of our findings with respect to motor symptoms. Our results highlight a tight link between sMN-cMN dynamics and RM symptoms observed in children with ASD.

The sMN and cMN nodes including the cerebellum, motor, and premotor regions are critical for motor control and execution, and have been shown to have structural abnormalities in individuals with ASD. Our results suggest reduced differentiation of these motor networks can lead to more rigidity in motor behaviors. We previously suggested that intrinsically hyper-connected circuits may make it more difficult to modulate connectivity in response to task demands, thereby resulting in task-related under-connectivity compared to the baseline state. Consistent with this proposal, a previous study reported reduced static connectivity between the subcortical and cortical motor regions during a finger sequencing task in individuals with autism. Thus, we hypothesize that the propensity of children with ASD to remain in brain states in which sMN and cMN nodes are intrinsically hyper-connected, potentially due to structural deficits in the fronto-thalamo-cerebellum pathway, could lead to inflexible motor control and RM behaviors that are characteristic of the disorder1.

To address growing concerns about reproducibility of neuroscientific findings, we leveraged our sample and conducted cross-validation analyses following procedures typically used in machine learning. Cross-validation is a powerful approach for validating research findings, and its use for demonstrating generalization and reproducibility has been advocated in psychiatry, psychology, and many other disciplines. The results of these analyses were consistent with our original results, demonstrating the robustness of our findings. Finally, findings were replicated with RRB measures derived from a previously published factor structure.
..

Link zur Studie bei Nature Communications