Distinct brain functional networks are associated with state- and trait-depression
Poster No:
531
Submission Type:
Abstract Submission
Authors:
Wei Zhang1, Rosie Dutt2, Daphne Lew1, Deanna Barch1, Janine Bijsterbosch1
Institutions:
1Washington University in St. Louis, Saint Louis, MO, 2Washington University in St. Louis; University of Chicago, Saint Louis, MO; Chicago, IL
First Author:
Co-Author(s):
Introduction:
Depression is a significant contributor to global disability, affecting over 300 million people worldwide1. Despite its broad recognition as a disorder involving large-scale brain networks, the foundational brain mechanisms remain elusive. Although some evidence points to altered connectivity patterns within major brain networks, prior findings often display inconsistency2,3 or lack replication4, offering, at best, a partial explanation for a limited proportion of the observed variance5. This inconsistency presents an enduring challenge in comprehensively understanding the etiology of depression. One potential contributing factor to this challenge may be the conflation of state (current symptoms; variable) and trait (general propensity; stable) depression. Addressing this issue, this study aims to disentangle state and trait depression, seeking to identify potentially dissociable resting-state correlates for these interconnected yet distinct constructs.
Methods:
We analyzed longitudinal data from the UK Biobank, using the resting-state fMRI data (scan1 and follow-up scan2), and depression assessments with Recent Depressive Scale (RDS)6 at baseline, scan1 and scan2. We differentiated trait from state depression, defining "trait" as individuals in remission (i.e., RDSscan1≤5) with a prior history (i.e., RDSbaseline≥7), and "state" as significant longitudinal fluctuation between two scans (i.e., RDS|scan2-scan1|≥3). We identified N=311 and N=265 participants for state and trait groups, respectively, and selected the same number of control participants with minimal RDS across all time points (RDSbaseline≤5 & RDSscan1≤5 & RDSscan2≤5) and matched on covariates (i.e., sex, age, head motion, scanning site and alcohol intake frequency), using optimal pair matching.
The resting-state data were decomposed into a set of 15 modes (i.e., resting-state networks), using Probabilistic Functional Modes (PROFUMO)7. Each mode is described by a spatial map (from which the spatial overlap was estimated as the correlation between each pair of mode spatial maps8), network matrix (i.e., connectivity matrix - estimated as the partial correlation between mode time-courses), and mode amplitude (indicative of overall BOLD signal fluctuations). We obtained these PROFUMO outputs separately per scan per participant.
After removing 2 spurious modes, we conducted two separate linear regression models to estimate group differences (i.e., state vs. control, trait vs. control; Fig 1A) in all PROFUMO outputs while accounting for covariates. We also repeated statistical analyses for two subgroups with stricter definitions of "state" (i.e., adding RDSbaseline≤5; N=148) and "trait" (adding PHQonline≥5; N=193) to replicate our findings. False Discover Rate (FDR) was applied to account for multiple testing within each class of PROFUMO output.
The resting-state data were decomposed into a set of 15 modes (i.e., resting-state networks), using Probabilistic Functional Modes (PROFUMO)7. Each mode is described by a spatial map (from which the spatial overlap was estimated as the correlation between each pair of mode spatial maps8), network matrix (i.e., connectivity matrix - estimated as the partial correlation between mode time-courses), and mode amplitude (indicative of overall BOLD signal fluctuations). We obtained these PROFUMO outputs separately per scan per participant.
After removing 2 spurious modes, we conducted two separate linear regression models to estimate group differences (i.e., state vs. control, trait vs. control; Fig 1A) in all PROFUMO outputs while accounting for covariates. We also repeated statistical analyses for two subgroups with stricter definitions of "state" (i.e., adding RDSbaseline≤5; N=148) and "trait" (adding PHQonline≥5; N=193) to replicate our findings. False Discover Rate (FDR) was applied to account for multiple testing within each class of PROFUMO output.
·Figure 1. Model equations and PROFUMO modes
Results:
After FDR corrections, we found no significant differences in network matrix (i.e., temporal coupling between modes) or spatial overlap matrix (i.e., similarity of spatial maps between modes) between state and control groups, nor between trait and control groups. However, we observed significantly smaller magnitude of changes in amplitude of the primary and sensorimotor cortices in individuals with state depression in contrast to controls (β=-0.02, pFDR=0.036), and significantly higher amplitude in the primary visual cortex in individuals with trait depression in contrast to controls (β=0.04, pFDR=0.02). These results replicated in subgroups with stricter definitions of state/trait (p's<0.03).
·Figure 2. Group differences in PROFUMO ampiltude
Conclusions:
Our findings show that both state and trait depression are associated with abnormal amplitudes of large-scale resting-state networks. Although the motor areas and visual cortex have been consistently implicated in depression symptoms9,10, our data suggest that these brain circuits may play different roles in differentiating state vs. trait depression experiences.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Modeling and Analysis Methods:
Bayesian Modeling
fMRI Connectivity and Network Modeling 2
Task-Independent and Resting-State Analysis
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
Affective Disorders
FUNCTIONAL MRI
MRI
Other - trait depression, state depression, PROFUMO
1|2Indicates the priority used for review
Provide references using author date format
2. Greene AS, Shen X, Noble S, et al. Brain–phenotype models fail for individuals who defy sample stereotypes. Nature. 2022;609(7925):109-118. doi:10.1038/s41586-022-05118-w
3. Tozzi L, Zhang X, Chesnut M, Holt-Gosselin B, Ramirez CA, Williams LM. Reduced functional connectivity of default mode network subsystems in depression: Meta-analytic evidence and relationship with trait rumination. Neuroimage Clin. 2021;30:102570. doi:10.1016/j.nicl.2021.102570
4. Kennis M, Gerritsen L, van Dalen M, Williams A, Cuijpers P, Bockting C. Prospective biomarkers of major depressive disorder: a systematic review and meta-analysis. Mol Psychiatry. 2020;25(2):321-338. doi:10.1038/s41380-019-0585-z
5. Schmaal L. The Search for Clinically Useful Neuroimaging Markers of Depression—A Worthwhile Pursuit or a Futile Quest? JAMA Psychiatry. 2022;79(9):845-846. doi:10.1001/jamapsychiatry.2022.1606
6. Dutt RK, Hannon K, Easley TO, Griffis JC, Zhang W, Bijsterbosch JD. Mental health in the UK Biobank: A roadmap to self-report measures and neuroimaging correlates. Hum Brain Mapp. 2022;43(2):816-832. doi:10.1002/hbm.25690
7. Harrison SJ, Bijsterbosch JD, Segerdahl AR, et al. Modelling subject variability in the spatial and temporal characteristics of functional modes. NeuroImage. 2020;222:117226. doi:10.1016/j.neuroimage.2020.117226
8. Bijsterbosch JD, Beckmann CF, Woolrich MW, Smith SM, Harrison SJ. The relationship between spatial configuration and functional connectivity of brain regions revisited. Ivry RB, Honey C, Margulies DS, Seidlitz J, eds. eLife. 2019;8:e44890. doi:10.7554/eLife.44890
9. Ray D, Bezmaternykh D, Mel’nikov M, Friston KJ, Das M. Altered effective connectivity in sensorimotor cortices is a signature of severity and clinical course in depression. Proceedings of the National Academy of Sciences. 2021;118(40):e2105730118. doi:10.1073/pnas.2105730118
10. Wu F, Lu Q, Kong Y, Zhang Z. A Comprehensive Overview of the Role of Visual Cortex Malfunction in Depressive Disorders: Opportunities and Challenges. Neuroscience Bulletin. 2023;39(9):1426. doi:10.1007/s12264-023-01052-7