Transdiagnostic neural correlates of executive dysfunction

1795 

Abstract Submission 

Hajer Karoui¹, Danielle Kurtin¹

¹Imperial College London, London, London

Hajer Karoui  
Imperial College London
London, London

Danielle Kurtin, PhD  
Imperial College London
London, London

The aetiology of Parkinson's Disease (PD), Obsessive-Compulsive Disorder (OCD), and Huntington's Disease (HD) are distinct, yet all share symptoms of executive dysfunction^1-4, such as challenges in planning, heightened impulsivity, disinhibition, and obsessional tendencies. Functional Magnetic Resonance Imaging (fMRI) studies have revealed comparable neural patterns of executive dysfunction in PD^1-2,HD³, and OCD⁴ patient (PT) groups, including diminished functional connectivity compared to healthy controls (HC). Here, we assess whether the brain state dynamics from each PT group are different from HC and other PT groups. The aim of this work is to evaluate if the cumulative effects of cellular-level pathology converge into similar neural markers of executive dysfunction.

fMRI data were collected during an executive planning task (Tower of London) ⁵ from 3 cohorts (PD=68, HD=30, OCD=24) of PTs and their matched HCs. Data preprocessing with fMRIPrep included brain extraction, BBR of BOLD data to T1w subject space using 6 DOF, slice-time correction, and estimation of head motion through 6 rotation and translation parameters. 14 subcortical regions were identified using Freesurfer, and the 400-region Schaefer cortical atlas was transformed to subject space. Regional timeseries were extracted, z-scored, filtered between 0.02-0.1 Hz, and Hilbert-transformed, enabling the pairwise computation of the cosine of the difference in phase angle, and generating a symmetric coherence matrix for all timepoints. The connectivity matrix for each timepoint was masked with a binarised matrix for the 8 Yeo functional networks ⁶ and 1 subcortical network. After calculating the mean of the coherence values within each network's mask, the timepoint was labelled with the network with the highest mean coherence, generating a network-based state timeseries (ST) for each subject.
State lifetime, Lempel Ziv Complexity (LZC)⁷, and Block Decomposition Methods of Complexity (BDMC) of ST were computed. Kruskal-Wallis tests evaluated group effects on state dynamic metrics, with Wilcoxon sign-rank post-hoc pairwise comparisons between PTs, and between PTs and HCs within groups. All p-values were FDR-corrected.

Across all subjects, the limbic state had the highest average percent occupancy, followed by the visual, dorsal attention, subcortical, temporal parietal, somatomotor, control, salience, and DMN states (Fig 1). While the low prevalence of the DMN was expected, the low prevalence of the control state was surprising, given its role in the planning processes engaged in TOL ^1-4. The high prevalence of the limbic state was driven by a minority of subjects displaying a total occupancy of their ST by the limbic state. While there was a significant main group effect on LZC and BDMC of ST, post-hoc pairwise differences were not significant (Fig 2). A significant main group effect was also seen for the lifetime of all states (Fig 2). Post-hoc pairwise tests showed PD PTs had significantly higher visual, somatomotor, and DMN lifetime than HCs. Inappropriate DMN engagement during tasks have been associated with poor cognitive task performance ⁸; thus, we suggest the higher lifetime of the DMN state in PD PTs may contribute to the previously reported worse performance of our PD vs HC subjects ¹.
PD PTs showed significantly longer lifetimes across all states than OCD PTs, and all states but subcortical ones compared to HD PTs. This is likely due to older age of PD PTs compared to other patient groups, as age disrupts the metastable, flexible neural dynamics associated with healthy cognition ⁸.

Our results show that metrics of state dynamics capture differences in how neural resources are organised over time to subserve the cognitive demands. Relationships among age, task performance, and measures of state dynamics are currently being investigated.

Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s)

Psychiatric (eg. Depression, Anxiety, Schizophrenia)

Executive Function, Cognitive Control and Decision Making ²

fMRI Connectivity and Network Modeling ¹

Cognition

Computational Neuroscience

FUNCTIONAL MRI

Modeling