Contextualized network dysfunction in schizophrenia: Granger Causality to Graph Theory
Poster No:
545
Submission Type:
Abstract Submission
Authors:
Kalyyanee Nanaaware1, John Kopchick2, Asadur Chowdury2, Patricia Thomas2, Usha Rajan2, Dalal Khatib2, Luay Haddad2, Alireza Amirsadri2, Jeffrey Stanley2, Vaibhav Diwadkar2
Institutions:
1Wayne State University School of Medicine, Detroit, MI, 2Wayne State University, Department of Psychiatry, Detroit, MI
First Author:
Co-Author(s):
Introduction:
In-degree and out-degree centrality are graph theoretic metrics that approximate the flow of information in a directed graph (Rubinov and Sporns, 2010). Such metrics are useful in understanding disordered task-driven connectomics in schizophrenia (Meram et al., 2023). Here, we used a learning paradigm (Stanley et al., 2017) with distinct task conditions (encoding, retrieval) to evoke brain network interactions in a group of controls (HC) and schizophrenia patients (SCZ). From a reduced network of co-activated nodes (dACC, dlPFC, Hippocampus, Superior Parietal, Inferior Temporal, Fusiform), we derived a directed graph where the edges between these six vertices were estimated using Wiener-Granger Causality (Baajour et al., 2020; Bressler and Seth, 2011). From these, the in- and out-degree centrality of each node was estimated before conducting parametric analyses.
Methods:
fMRI data were collected in 55 participants (31 patients, Siemens Verio 3T). During the task, participants encoded object-location associations (ENCoding) and were tested in separate blocks (RETrieval) using location cues. fMRI data were preprocessed using typical methods (SPM 12). A co-activated network (HC ∩ SCZ) was identified using a conjunction analyses based on the minimum inference statistic (Nichols et al., 2005). Time series from these (dACC, dlPFC, HPC, ITG, SPC, FG) were extracted from each participant. Next, in each participant inter-node causality was estimated from time series using multi-variate autoregressive models (MVAR, consistent with Weiner Granger Causality). From the resultant directed graphs in each participant (6 vertices, 30 unique edges), we estimated the in- and out-degree centrality of each vertex for each of encoding and retrieval, and in each of the early and late phases of the paradigm. As the paradigm evoked negatively accelerated learning, we studied in-group (HC vs SCZ) and inter-phase (early learning vs. late learning) differences on degree centrality in an analyses of variance framework (separately for ENC and RET).
Results:
Figure 1 provides a comprehensive overview of our results. During ENC, a) the in-degree centrality of the dlPFC increased over the course of learning, and was higher in SCZ while b) the out-degree centrality of the SPC increased over the course of learning. During RET, we observe a) an interaction on the in-degree centrality in the dlPFC, wherein in controls in decreased over time, but in patients it increased; b) the out-degree centrality of the FG decreased over time.
Conclusions:
Effects during ENC and RET were somewhat distinct in terms of node specificity and patterns. During ENC, the dlPFC showed higher in-degree centrality in SCZ, suggestive of increased "flow" of information into this region. This increase may be seen as a class of "compensatory" response, often seen in the illness (Zovetti et al., 2022). In SCZ, we also observed reduced in-degree centrality of the ITG during RET. This effect is notable in that the retrieval cue (in part originating in the SPC and the dlPFC) must engage the retrieval of the object identity (an ITG related function). The ancillary effects of phase and the interaction between group and phase (dlPFC during RET) highlight a) the complex set of directed functional interactions that accrue in the brain during learning and b) the highly contextualized nature of connectivity deficits evoked in schizophrenia.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Higher Cognitive Functions:
Executive Function, Cognitive Control and Decision Making
Learning and Memory:
Long-Term Memory (Episodic and Semantic)
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 2
fMRI Connectivity and Network Modeling
Keywords:
FUNCTIONAL MRI
Learning
Memory
Multivariate
Schizophrenia
1|2Indicates the priority used for review
Provide references using author date format
Bressler, S.L. (2011), Wiener-Granger causality: a well established methodology. NeuroImage 58(2), 323-329.
Meram, E.D. (2023), The topology, stability, and instability of learning-induced brain network repertoires in schizophrenia. Network Neuroscience 7(1), 184-212.
Nichols, T. (2005), Valid conjunction inference with the minimum statistic. NeuroImage 25(3), 653-660.
Rubinov, M. (2010), Complex network measures of brain connectivity: uses and interpretations. NeuroImage 52(3), 1059-1069.
Stanley, J.A. (2017), Functional dynamics of hippocampal glutamate during associative learning assessed with in vivo 1H functional magnetic resonance spectroscopy. NeuroImage 153, 189-197.
Zovetti, N. (2022), Inefficient white matter activity in Schizophrenia evoked during intra and inter-hemispheric communication. Translational psychiatry 12(1), 449.