Functional Connectivity during Frustration: Predictive Modeling of Irritability in Children and Adolescents

• In Event: Using Connectomics and Machine Learning to Examine Brain Development

Abstract

Introduction: Irritability, defined as an increased proneness to anger and frustration relative to peers (Brotman et al., 2017; Leibenluft, 2017), is among the most common presenting problems in childhood psychopathology (Stringaris et al., 2018). Clinically, irritability manifests as developmentally inappropriate, frequent, and severe temper outbursts and irritable mood (Brotman et al., 2017; Leibenluft, 2017) and cuts across many pediatric disorders including disruptive mood dysregulation disorder (DMDD), oppositional defiant disorder, attention-deficit/hyperactivity disorder (ADHD), and anxiety and depressive disorders. It has been well-documented that childhood irritability predicts long-term impairments including high suicidality and decreased education and income attainment (Vidal-Ribas et al., 2016). Despite the impairing nature of childhood irritability, the underlying neural mechanisms remain unclear. Given that many irritable youths have a low threshold for frustration, paradigms evoking frustration provide an effective means to study irritability in a laboratory setting (Tseng et al., 2019). Here, we used two frustrating fMRI tasks to examine functional connectivity during frustration and its predictability of irritability symptoms in two transdiagnostic samples of children and adolescents with varying levels of irritability.
Methods: Sample 1 included 69 youths: 20 DMDD, 14 ADHD, 12 anxiety disorder, and 23 healthy controls (mean age=14.6 years; 55% males). Sample 2 included 195 youths: 52 DMDD, 40 ADHD, 42 anxiety disorder, and healthy 61 controls (mean age=12.9 years; 50% males). Sample 1 completed a modified Change-Signal task (Brown and Braver, 2005), and Sample 2 completed a modified Affective Posner cued-attention task (Tseng et al., 2019) while their fMRI data were acquired on a 3T scanner. We evoked frustration by presenting participants with rigged feedback (both tasks) and by manipulating task difficulty (modified Change-Signal). Irritability was measured using the parent-report and child-report of the Affective Reactivity Index scale (Stringaris et al., 2012). Standard preprocessing procedures were applied to the fMRI data (e.g., despiking, slice timing correction, co-registration). Several covariates of no interest were regressed out from the data, including 12 motion parameters, white matter signal, cerebrospinal fluid signal, global signal, and the linear, quadratic, and cubic drifts. Only group results from Sample 1 were presented here; group analyses for Sample 2 are underway and will be included in the presentation (should the submission be accepted).
Results: Using the connectome-based predictive modeling with the Shen 268-node functional brain atlas (CPM; Shen et al., 2013, 2017) and a 10-fold cross-validation, we found that functional connectivity during frustration (not non-frustration) blocks was predictive of child-reported irritability (r=.24, root mean square error=2.02, p=.03, permutation testing, 1000 iteration, one-tailed; Figure 1). Results were adjusted for age, motion, ADHD, and anxiety symptoms. The most predictive networks of irritability were within the motor-sensory networks;
among motor-sensory, subcortical, and salience networks; and between these networks and frontoparietal and medial frontal networks (Figure 2).
Conclusions: We found that functional connectivity during a “frustrated” state was more predictive of youths’ trait-level irritability than functional connectivity during a “non-frustrated” state. This highlights the importance of examining functional connectivity during a phenotype-relevant state to better study individual differences in brain-behavior relationships and thus facilitate biomarker discovery.



 

Authors

• Wan-Ling Tseng, Yale School of Medicine
  Presenting Author

• Eva McAdam Freud, Yale School of Medicine
  Non-Presenting Author

• Dustin Scheinost, Yale School of Medicine
  Non-Presenting Author