Individual differences in effective connectivity across resting state networks implicated in anxiety in a youth cohort: a preliminary study

Pediatric anxiety, or anxiety that arises starting in childhood through adolescence has high prevalence (Guyer et al., 2013). Specifically, with a median age of onset at age 11 years, population prevalence rates among children and adolescents have been estimated to be between 5.7% and 12.8% (Ramsawh et al., 2010). Pediatric anxiety is a potential marker for development of even greater severity anxiety symptomatology after onset from childhood and late adolescence into adulthood (Guyer et al., 2013). For example, Generalized anxiety disorder (GAD), a subset of pediatric anxiety, affects 5.7% of individuals throughout life (Sievertsen et al., 2025). GAD arises typically in late childhood or adolescence (Sievertsen et al., 2025). These ages also cover a period of considerable brain development, during which plasticity is high (Fandakova & Hartley, 2020; Sievertsen et al., 2025). The nature of this temporal association needs to be explored, as it is not known whether it is causal, due to the effects of background factors, or coincidence. This study will investigate pediatric anxiety, or the collective group (per (Sylvester et al., 2012) and (Guyer et al., 2013)) of high trait anxiety and anxiety disorders (including namely generalized anxiety disorder, separation anxiety disorder, social phobia) as expressed in an adolescent cohort, where onset of the disorders is expected relative to childhood. Specifically, pediatric anxiety will be assessed on a continuum based on the Child Behavior Checklist (CBCL) DSM (Diagnostic and Statistical Manual of Mental Disorders) oriented anxiety problems subscale, which will serve as a proxy to assess anxiety in this cohort. Functional Magnetic Resonance Imaging (fMRI) is a tool which measures neuronal activity by proxy of blood oxygenation level dependent (BOLD) content in the brain. Using resting-state fMRI, resting-state functional connectivity (rsFC) between different brain regions can be inferred. Resting state functional connectivity has been used to characterize normal and abnormal states in the brain (Li et al., 2022; Ma et al., 2024; Mwansisya et al., 2017). By itself, rsFC is a useful metric to characterize any normal or aberrant connections between brain regions during resting state (Li et al., 2022; Ma et al., 2024; Mwansisya et al., 2017). However, rsFC is limited in that it assesses statistical dependencies between brain regions (K. J. Friston et al., 2003, 2014). Understanding causal interactions among regions of interest requires methods that go beyond statistical dependence. Dynamic causal modeling (DCM) (K. J. Friston et al., 2003) provides a framework for inferring effective connectivity based on known or hypothesized interregional interactions. The DCM framework has been validated through empirical research (Cook et al., 2019; K. J. Friston et al., 2016; Stephan et al., 2007). In order to understand the causal interactions between regions, and the directed network architecture at the neural level between given regions of interest, dynamic causal modeling (DCM) provides a framework given known statistically dependent regions or hypothesized interactions (Cook et al., 2019; K. J. Friston et al., 2016; Stephan et al., 2007). Specifically, spectral DCM (spDCM) provides a framework for modeling the effective connectivity of the brain using resting state functional MRI (rsfMRI) data (K. J. Friston et al., 2014). It is thought that high trait anxiety, and anxiety disorders have aberrations of functional networks in the brain (Sylvester et al., 2012). Specifically, it is thought that there is overactive functioning of the cingulo-opercular (CON) and ventral attention network (VAN) (Sylvester et al., 2012). In contrast it is thought that there is underactive functioning of the frontoparietal network (FPN) and default mode network (DMN) (Sylvester et al., 2012). The cingulo-opercular network (CON) is implicated in detection of errors or conflict (Sylvester et al., 2012). The fronto-parietal network has been implicated in being responsible for implementing adjustments in cognitive control (Sylvester et al., 2012). The ventral attention network is implicated in directing an individual’s attention to new onset stimuli (Sylvester et al., 2012). The default mode network (DMN), a task negative network (active when not performing a specific task, for example) is involved in emotion regulation, self-inspection, rumination and introspection (Sylvester et al., 2012; Xiong et al., 2020). (Xiong et al., 2020) noted that the default mode network has aberrations in GAD (Xiong et al., 2020). One meta-analytic study reported that with regard to GAD, specific hypo-activation of the prefrontal cortex and anterior cingulate cortex as it interacts with the amygdala, relative to participants without anxiety occurs in the brain (Mochcovitch et al., 2014). For youth with generalized anxiety disorder (GAD) (pediatric GAD), hyperconnectivity within the VAN, hyperconnectivity between CON and the amygdala and across striatal regions with the CON, DMN, and salience networks (implicated in detection behaviorally relevant stimuli) was found (Sievertsen et al., 2025). In conjunction with (Sylvester et al., 2012), specific regions in each of the four networks are implicated in anxiety, and separately the amygdala of the subcortex (which is involved in fear/threat detection) (Sievertsen et al., 2025). Another study found connectivity involving the dorsal attention network pertaining to angular gyrus/inferior parietal connectivity in trait anxiety in adulthood (De la Peña-Arteaga et al., 2024). An rsFC exclusive study in pediatric anxiety hypothesized that anxiety was related to altered connectivity of the FPN, DMN, CON, and VAN (Perino et al., 2021). The study found correlations between anxiety and altered connectivity primarily in within-CON, between CON and VAN, as well as CON-DMN, and CON and visual networks (Perino et al., 2021). In support of DMN underreactivity (Sylvester et al., 2012), a recent study found a relationship of higher trait anxiety to lower DMN connectivity (Baggio et al., 2025). Investigating effective connectivity in the context of pediatric anxiety, therefore, has the potential to provide improved evidence in assessing dual-processing theories of emotion control, where initial activity of bottom-up nodes is tonically controlled by more dorsal and lateral frontocortical regions functioning. In addition, there is potential to extend adirectional functional connectivity hypotheses with directional flow of information. Furthermore, the study aims to investigate the generalizability of hypothesized adirectional abnormalities found in either pediatric and adult populations to pediatric anxiety, and whether there is a graded, or dimensional, effect. In this study, the four networks, the VAN, CON, FPN, DMN, and the amygdala were selected for dynamic causal modelling analyses. Specifically, 1) the dorsal anterior cingulate cortex (dACC) and anterior insula in the cingulo-opercular network, 2) the dorsolateral prefrontal cortex (dlPFC) and inferior parietal lobule in the fronto-parietal network, 3) the ventrolateral prefrontal cortex (vlPFC) and temporoparietaljunction (TPJ) of the ventral attention network, 4) the subgenual anterior cingulate cortex (sACC) and posterior cingulate cortex (PCC) of the default mode network, and 5) the amygdala of the subcortical regions. Additionally, in order to investigate pediatric anxiety, and generalizability of how individual differences in anxiety occurs in individual differences in the brain in youth, the Adolescent Brain Cognitive Development (ABCD) study will be utilized (Hagler et al., 2019; Maes et al., 2023). Specifically, the ABCD study enrolled over 10,000 participants ages 9-10 over multiple time points until ages 19-20, where both neuroimaging measures and diagnostic and other measures are collected (Hagler et al., 2019; Maes et al., 2023). Therefore, for this study, the goal is to use spectral DCM in a youth cohort to establish individual differences in effective connectivity patterns involving the VAN, CON, FPN, DMN and amygdala in the youth population. and compare specific individual differences in effective connectivity features with individual differences in anxiety subscale scores. Specifically, this study will investigate the causal effective connectivity interactions involved in pediatric anxiety in an adolescent cohort across the age of 16. The study will determine 1) the expected causal interaction network model across the population of individuals and 2) individual differences of prominent directional connections (effective connectivity edge weights) and how they correlate with individual differences in anxiety subscale scores of interest.