Multimodal Analysis of Brain Alterations in Bipolar Disorder and Relationship to Clinical Covariates

2159 

Abstract Submission 

Melody J.Y. Kang¹, Leila Nabulsi^1,2, Genevieve McPhilemy², Fiona Martyn², Brian Hallahan², Ole Andreassen³, Colm McDonald², Paul Thompson¹, Christopher Ching¹, Dara Cannon²

¹Imaging Genetics Center, Mark and Mary Stevens Neuroimaging & Informatics Institute, USC, Los Angeles, CA, ²University of Galway, Galway, Galway, ³NORMENT, Oslo, Norway

Melody J.Y. Kang  
Imaging Genetics Center, Mark and Mary Stevens Neuroimaging & Informatics Institute, USC
Los Angeles, CA

Leila Nabulsi, PhD  
Imaging Genetics Center, Mark and Mary Stevens Neuroimaging & Informatics Institute, USC|University of Galway
Los Angeles, CA|Galway, Galway

Genevieve McPhilemy, PhD  
University of Galway
Galway, Galway

Fiona Martyn, PhD  
University of Galway
Galway, Galway

Brian Hallahan, MD  
University of Galway
Galway, Galway

Ole Andreassen  
NORMENT
Oslo, Norway

Colm McDonald, PhD  
University of Galway
Galway, Galway

Paul Thompson, PhD  
Imaging Genetics Center, Mark and Mary Stevens Neuroimaging & Informatics Institute, USC
Los Angeles, CA

Christopher Ching  
Imaging Genetics Center, Mark and Mary Stevens Neuroimaging & Informatics Institute, USC
Los Angeles, CA

Dara Cannon, PhD  
University of Galway
Galway, Galway

Bipolar disorder (BD) is a chronic and severe mental illness with no validated biomarkers to guide diagnosis or treatment. Altered cortical and subcortical volumes (Hibar et al., 2016, Ching et al., 2022), as well as functional dysconnectivity of the default mode network (DMN), have been reported in BD (Nabulsi et al., 2020, Philemy et al., 2020). However, differences in processing and analysis methods challenge the consistency of study findings, calling for large-scale, standardized analyses to improve replication and generalizability. In addition, the impact of illness severity and common treatments on brain measures in BD remains unclear. We aimed to evaluate both structural and functional brain alterations in BD using standard protocols from the ENIGMA Consortium and map alterations associated with both symptom severity and medication use.

Structural T1-weighted and resting-state functional MRI data were acquired from one pilot ENIGMA-BD site, with 100 participants (44 BD, 56 controls (CN); age range 19-65, 54% female). Subcortical volumes of lateral ventricles, thalamus, caudate, putamen, pallidum, hippocampus, amygdala, and nucleus accumbens were derived using an ENIGMA-standardized protocol (based on FreeSurfer v5.3). ENIGMA-HALFpipe (v1.2.1) (Waller et al., 2022), a standardized pipeline based on fMRIPrep (Esteban et al., 2019), was used to derive overall within-network connectivity (N=94) of the DMN as well as 3 subnetworks ('core', dorsal medial, and medial temporal), using individual Pearson's correlation matrices (Schaefer el al., 2018). Multiple linear regression was used to test subcortical volumes for the effect of diagnosis, treatments (lithium, antipsychotics, antidepressants) and symptom severity (number of mood episodes). Any significant associations were also evaluated in DMN network connectivity. Models were covaried for age, sex, intracranial volume (structural), motion parameters (functional) and adjusted for multiple comparisons (FDR q<0.05).

No subcortical volume differences were detected between BD and CN. In those with BD, antidepressant treatment (primarily targeting serotonin; SSRIs) was associated with smaller pallidum volumes (pFDR=0.035, R2= 0.35). Greater number of (hypo)manic episodes was associated with larger putamen (pFDR=0.04, R2=0.46) and pallidum volumes (pFDR=0.001, R2=0.56) even when adjusting for medication use and duration of illness. Number of depressive episodes was not associated with subcortical volume. Neither antidepressant treatment or (hypo)manic episodes was associated with DMN functional connectivity measures derived from the current implementation of the ENIGMA-HALFPipe protocol.

In this pilot sample, we found no subcortical volume differences between BD and CN, highlighting the challenge of detecting subtle BD-related structural brain alterations without large-scale samples (Hibar et al., 2016). Whereas serotonergic antidepressant treatment was related to smaller basal ganglia volumes, greater number of hypo- and manic episodes was related to larger volumes. The differential effects between medication and illness severity are intriguing and merits further evaluation in the larger ENIGMA-BD sample. The ENIGMA Schizophrenia working group found larger pallidum volumes associated in patients (van Erp et al., 2016), and the ENIGMA obsessive-compulsive disorder working group has also shown smaller striatal volumes associated with antidepressant treatment (Ivanov et al, 2022). Neither antidepressant use or number of (hypo)manic episodes was related to alterations in DMN connectivity in this current implementation of the ENIGMA-HALFpipe protocol. However, prior findings from the team have highlighted BD-related functional alterations in this sample (Quirke et al., 2023). Ongoing work is focused on assessing the downstream effects of variable rsfMRI processing streams in large, multisite samples to improve detection of the subtle functional brain alterations in BD.

Connectivity (eg. functional, effective, structural)

Segmentation and Parcellation

Task-Independent and Resting-State Analysis ²

Subcortical Structures ¹

Anatomical MRI

Data analysis

FUNCTIONAL MRI

MRI

Psychiatric Disorders

STRUCTURAL MRI

Sub-Cortical

Other - Bipolar Disorder