Dimensional brain-behavior relationships in children with attention-deficit/hyperactivity disorder

Abstract

Background: Emerging neuroscientific and genetic findings emphasize the dimensional rather than the categorical aspects of psychiatric disorders. However, the integration of dimensional approaches within the current categorical diagnostic framework remains unclear. Here, we used resting state functional magnetic resonance imaging to examine whether dimensional measures of psychiatric symptomatology capture brain-behavior relationships unaccounted for by categorical diagnoses. Additionally, we examined whether dimensional brain-behavior relationships are modified by the presence of a categorically defined illness, attention-deficit/hyperactivity disorder (ADHD).

Methods: Resting state functional magnetic resonance imaging scans were collected from 37 typically developing children (aged 10.2 ± 2; 21 female subjects) and 37 children meeting DSM-IV Text Revision criteria for ADHD (9.7 ± 2; 11 female subjects). Parent-rated Child Behavior Checklist Externalizing and Internalizing scores served as dimensional measures in our analyses of default network (DN) resting state functional connectivity (RSFC).

Results: Regardless of diagnosis, we observed several significant relationships between DN RSFC and both internalizing and externalizing scores. Increased internalizing scores were associated with stronger positive intra-DN RSFC, while increased externalizing scores were associated with reduced negative RSFC between DN and task-positive regions such as dorsal anterior cingulate cortex. Several of these brain-behavior relationships differed depending on the categorical presence of ADHD.

Conclusions: Our findings suggest that while categorical diagnostic boundaries provide an inadequate basis for understanding the pathophysiology of psychiatric disorders, psychiatric illness cannot be viewed simply as an extreme of typical neural or behavioral function. Efforts to understand the neural underpinnings of psychiatric illness should incorporate both categorical and dimensional clinical assessments.

FIGURE 1. Positive relationships between resting state functional connectivity (RSFC) and dimensional measures (Internalizing and Externalizing scores per Child Behavior Checklist [CBCL])

For Externalizing (left panel) and Internalizing CBCL scores (right panel), surface inflated maps display the clusters with significant positive relationships between scores and resting state functional connectivity (RSFC) for each of the default network (DN) seeds examined. Red colored areas represent regions for which a significant brain-behavior relationship was observed for multiple seeds. The scatter plots at the bottom illustrate each participant’s relationship between RSFC and CBCL scores. Left: Higher Externalizing scores were associated with weaker negative RSFC between A) posterior inferior parietal lobule (pIPL) and dorsal anterior cingulate cortex (displayed in khaki). Right: Higher Internalizing scores were associated with increased positive RSFC between B) the hippocampal formation (HF) and left and right temporal poles and C) the core nodes of the default network (DN) (i.e., anterior medial prefrontal cortex (aMPFC) and PCC, displayed in yellow and orange respectively). Similar relationships were found for the RSFC between the pIPL and the PCC (displayed in khaki) and between the lateral temporal cortex (LTC) and the precuneus (displayed in blue). Bonferroni correction accounting for the number of subsystems exceeding an omnibus cluster-level correction of p<0.017 was employed for multiple comparison correction.

FIGURE 2. Negative relationships between resting state functional connectivity (RSFC) and dimensional measures (Internalizing and Externalizing symptoms per Child Behavioral Checklist [CBCL])

For Externalizing (left panel) and Internalizing CBCL scores (right panel), surface inflated maps display the clusters with significant negative relationships between scores and resting state functional connectivity (RSFC) for each of the default network (DN) seeds examined. Red colored areas represent regions for which a significant brain-behavior relationship was observed for multiple seeds. The scatter plots at the bottom illustrate each participant’s relationship between RSFC and CBCL scores. Left: Higher Externalizing symptoms were associated with reduced positive RSFC between A) retrosplenial cortex (Rsp) and posterior parietal/dorsal occipital cortex (represented in bright green) Right: Higher Internalizing scores were associated with stronger negative RSFC between B) Rsp and anterior insula (displayed in bright green) and C) anterior medial prefrontal cortex (aMPFC) and ventrolateral prefrontal and premotor cortex. Bonferroni correction accounting for the number of subsystems exceeding an omnibus cluster-level correction of p<0.017 was employed for multiple comparison correction.

FIGURE 3. Relationships between resting state functional connectivity (RSFC) and categorical X dimensional interactions

For Externalizing scores measured with the Child Behavior Checklist (CBCL), surface maps display regions showing significant differences between typically developing children (TDC) and children with ADHD in the relationship between CBCL scores and RSFC for each of the default network (DN) seeds. The scatter plots at the bottom illustrate each participant’s relationship between RSFC and symptom scores according to diagnostic status. Higher Externalizing scores were associated with decreased positive RSFC between A) posterior inferior parietal lobule (pIPL) and frontal pole (displayed in khaki) and B) TempP and precuneus (displayed in blue) in children with ADHD while in TDC, such relationships were absent. Bonferroni correction accounting for the number of subsystems exceeding an omnibus cluster-level correction of p<0.017 was employed for multiple comparison correction.