Helmholtz Gemeinschaft


State-dependent signatures of anti-NMDA-receptor encephalitis: a dynamic functional connectivity study

Item Type:Preprint
Title:State-dependent signatures of anti-NMDA-receptor encephalitis: a dynamic functional connectivity study
Creators Name:von Schwanenflug, N. and Krohn, S. and Heine, J. and Paul, F. and Pruess, H. and Finke, C.
Abstract:INTRODUCTION: Anti-N-methyl-d-aspartate receptor encephalitis (NMDARE) is an autoimmune disorder associated with severe neuropsychiatric symptoms. While patients with NMDARE exhibit disrupted functional connectivity (FC), these findings have been limited to static connectivity analyses. This study applies time-resolved FC analysis to explore the temporal variability of large-scale brain activity in NMDARE and to assess the discriminatory power of functional brain states in a supervised classification approach. METHODS: Resting-state fMRI data from 57 patients with NMDARE and 61 controls was included. To capture brain dynamics, four discrete connectivity states were extracted and state-wise group differences in FC, occurrence, dwell time and transition frequency were assessed. Furthermore, logistic regression models with embedded feature selection were trained for each state to predict group status in a leave-one-out cross validation scheme. RESULTS: Patients showed FC alterations in three out of four states. Besides a reduction in hippocampal-frontal connectivity, we observed connectivity decreases within the default mode network and between frontal areas and subcortical as well as visual regions, which remained undetected in static FC. Furthermore, patients displayed a shift in dwell time from the weakly connected dominant state to a higher connected, but less frequent state, accompanied by increased transition frequencies. Discriminatory network features and predictive power varied dynamically over states, reaching up to 78.6% classification accuracy.CONCLUSION: Patients showed state-specific alterations in FC along with a shift in dwell time and increased volatility of state transitions. These measures were associated with disease severity and duration, highlighting the potential of spatiotemporal dynamics in FC as prognostic biomarkers in NMDARE.
Keywords:Autoimmune Encephalitis, Functional Connectivity, Dynamic Functional Connectivity, Machine Learning
Publisher:Cold Spring Harbor Laboratory Press
Article Number:bioRxiv:2020.06.12.141945
Date:12 June 2020
Official Publication:https://doi.org/10.1101/2020.06.12.141945
Related to:
https://edoc.mdc-berlin.de/21357/Final version

Repository Staff Only: item control page

Open Access
MDC Library