Convergent Task-State Brain Functional Network Profile Linking Behavioral Symptoms

579 

Abstract Submission 

Yunman Xia¹, Gunter Schumann²

¹Fudan University, Shanghai, China, ²Centre for Population Neuroscience and Stratified Medicine, Berlin, Germany

Yunman Xia  
Fudan University
Shanghai, China

Gunter Schumann  
Centre for Population Neuroscience and Stratified Medicine
Berlin, Germany

Mental illness accounts for nearly 30% of the disease burden among non-communicable diseases worldwide1. Extensive research efforts have been directed toward understanding the pathological mechanisms underlying psychiatric symptoms and developing effective clinical treatments2,3. However, due in part to the complexity of behavioral symptoms and the limitation of sample size, establishing robust and generalizable neuroimaging biomarkers remains challenging. Previous work has identified a shared brain functional network that exhibits a positive relationship with behavioral symptoms during early adolescence4. Here, we aim to leverage follow-up data of this longitudinal neuroimaging cohort to identify cross-disorder associated brain functional network profiles in young adults.

First, we utilized task-fMRI data and clinical measurements obtained from approximately 1000 healthy young adults (from IMAGEN cohort, aged 19 years) to explore the associations between task-state functional brain connectivity and various behavioral symptoms. Specifically, we employed the CONN toolbox to estimate the condition-specific functional connectome derived from the Monetary Incentive Delay (MID)5 and Stop signal task (SST)6. The behavioral symptoms were assessed by the Development and Well-Being Assessment7 and Strengths and Difficulties Questionnaire8. Then we applied the connectome-based predictive model (CPM)9 to predict behavioral symptoms using whole-brain functional connectome. The CPM was iterated 1,000 times, and the edges that were present in over 95% of predicted models and both related to the externalizing and internalizing symptoms were selected. Subsequently, we validated the behavioral implications of these cross-disorder associated edges in the clinical populations (STRATIFY and ESTRA dataset, N=513, aged 18–26, case/control=288/225). The case-control comparisons involving these edges were conducted, while controlling for site effects, sex, and head motion.

We found that task-based functional connectivity (FC) could significantly predict the majority of externalizing and internalizing symptoms at 19 years of age (Fig.1a and b). We delineated two distinct network profiles comprising FCs displaying either positive or negative relationships with both externalizing and internalizing symptoms, respectively (Fig.1c). The FCs exhibiting positive relationships with behavioral symptoms were primarily localized within brain regions such as the dorsal posterior cingulate cortex (dorsal PCC), inferior frontal gyrus (IFG), dorsolateral prefrontal cortex (DLPFC), and cerebellum. However, the FCs demonstrating negative relationships were localized within regions including dorsal PCC, DLPFC, IFG, angular gyrus (AG), and cerebellum. Moreover, we observed notable group differences in functional network profiles that exhibited negative associations with symptoms between healthy and clinical samples. Specifically, the whole clinical samples and subsets with alcohol use disorder (AUD), eating disorder (ED), and major depression disorder (MDD) all exhibited significant higher behavioral symptoms and weaker connections than healthy people (Fig.2a and b). In addition, we identified different patterns of individual FC profiles across specific disorders (Fig.2c), such as the weaker FC between fusiform and visuomotor regions in the AUD group, the weaker FC between AG and DLPFC in the ED group, and the weaker FC between AG and cerebellum in the MDD group.

Our findings identified a reliable neuroimaging biomarker underlying the symptoms across multiple mental disorders, which holds implications for early prevention and therapeutics in psychiatry. We aim to further simulate this brain biomarker in computational brain models and then manipulate brain models to alter the brain biomarker associated with symptoms.

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ¹

Reward and Punishment

Executive Function, Cognitive Control and Decision Making

Connectivity (eg. functional, effective, structural)

fMRI Connectivity and Network Modeling ²

Anxiety

Eating Disorders

Psychiatric Disorders

Other - reward processing