Longitudinal associations between depression symptoms and brain structure in youth

641 

Abstract Submission 

Eira Aksnes¹, Dani Beck², Niamh MacSweeney³, Marieke Bos⁴, Lia Ferschmann¹, Linn Norbom¹, Valerie Karl¹, Lars Westlye⁵, Christian Tamnes¹

¹University of Oslo, Oslo, Norway, ²Diakonhjemmet Hospital, Oslo, Oslo, ³University of Oslo/Diakonhjemmet Hospital, Oslo, Oslo, ⁴Leiden University, Leiden, Netherlands, ⁵Norwegian Centre for Mental Disorders Research (NORMENT), Oslo University Hospital, Oslo, Norway

Eira Aksnes, Phd  
University of Oslo
Oslo, Norway

Dani Beck  
Diakonhjemmet Hospital
Oslo, Oslo

Niamh MacSweeney  
University of Oslo/Diakonhjemmet Hospital
Oslo, Oslo

Marieke Bos  
Leiden University
Leiden, Netherlands

Lia Ferschmann  
University of Oslo
Oslo, Norway

Linn Norbom  
University of Oslo
Oslo, Norway

Valerie Karl  
University of Oslo
Oslo, Norway

Lars Westlye  
Norwegian Centre for Mental Disorders Research (NORMENT), Oslo University Hospital
Oslo, Norway

Christian Tamnes  
University of Oslo
Oslo, Norway

Emotional disorders have become the largest disease burden in adolescence during the last 10 years, especially in girls1. Adolescent-onset Major Depressive Disorder (MDD) is associated with a host of psychiatric and somatic comorbidities. Structural magnetic resonance imaging (sMRI) studies have sought to understand the neurobiological underpinnings of MDD, but findings have been inconsistent2. Compared to case-control studies of MDD, symptom-based investigations can capture the heterogenous nature of MDD3. Previous studies have reported associations between specific MDD symptoms and distinct brain structural features in adults4,5. However, this research is limited by cross-sectional designs, small sample sizes and few studies to date have focused on adolescent samples5. To this end, we investigate the association between dimensional aspects of depressive symptomatology and regional cortical thickness (CT) and hippocampal (HC) volume in youth using a large longitudinal sample. Based on previous work, we hypothesized that lower CT in the insula, medial orbitofrontal cortex (mOFC), cingulate (CI), fusiform gyrus (FG), and HC would be associated with specific symptoms of depression, and that these effects would exacerbate over time.

We used longitudinal symptom and sMRI data from the Adolescent Brain Cognitive Development (ABCD) Study. Data was extracted from three timepoints. Due to limited data availability, we used a subset of the ABCD youths who had data at the third wave, in addition to the first and/or second wave. Our final sample included 2892 children and adolescents (53 % male, mean age=12.11, SD=1.82). To match core MDD symptoms in DSM-5, we chose four items (sad, interest loss, worthless and low energy) derived from the parent-reported Child Behavioral Checklist (CBCL). Building on previous studies on MDD case-control differences4,5, we based our a priori MRI metrics and regions of interest on brain associations showing the largest bilateral effects6,7: CT in the CI, insula, mOFC, and FG, as well as HC volume. We harmonized imaging data across scanners using the long.combat R package, averaged left and right hemispheres, and residualized HC volume by intracranial volume.

To test for associations between specific depressive symptoms and select brain measures, we estimated a panel graphical vector-autoregression network model using the psychonetrics package. This model estimates both between-person and within-person (contemporaneous and temporal) dynamics over time. Bootstrapping was applied to check for network stability. An Individual Network Invariance Test (INIT) was estimated to investigate potential differences between male and female networks.

Both the saturated and pruned models showed good fit (TLI=.94/.96, CFI=.96, RMSEA=.042/.036). In the temporal network, we found positive associations between HC and CI in the brain domain, with auto-regressions for the HC, mOFC and insula. For symptoms, worthless was associated with sad, sad was associated with interest loss, and interest loss and low energy predicted each other over time. Auto-regressions were present for all symptoms except feeling worthless. The contemporaneous network showed associations between all symptoms, and all brain regions were associated apart from the HC. The between-persons network showed positive associations between low energy and sad, sad and worthless, and positive associations between all cortical areas, as well as the insula and HC. There was a negative association between HC and mOFC. Lastly there were no associations between brain-symptom domains in any of the three networks (Figure 1). The INIT did not identify significant sex differences.

In the first three wave large-scale longitudinal study investigating brain-symptom relations in youth, we found no significant associations between select brain regions and depression symptoms. Further analysis will use self-report measures, which may be a preferred measure for depressive symptoms.

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ¹

Emotion and Motivation Other

Early life, Adolescence, Aging ²

Anatomical MRI

Development

Emotions

MRI

Other - Depression