Cognitive Phenotypes in Epilepsy are Associated with Unique Functional Connectivity Profiles: Findings from the Epilepsy Connectome Project
Abstract number :
1.441
Submission category :
11. Behavior/Neuropsychology/Language / 11A. Adult
Year :
2019
Submission ID :
2421434
Source :
www.aesnet.org
Presentation date :
12/7/2019 6:00:00 PM
Published date :
Nov 25, 2019, 12:14 PM
Authors :
Bruce P. Hermann, University of Wisconsin; Lisa L. Conant, Medical College of Wisconsin; Cole J. Cook, Medical Physics; Gyujoon Hwang, Department of Radiology; Kevin Dabbs, University of Wisconsin; Veena A. Nair, University of Wisconsin; Jedidiah Mathis,
Rationale: Individuals with temporal lobe epilepsy (TLE) are known to have impaired cognitive function at the group level compared to healthy individuals. However, there is variability in the cognitive presentation among individual patients, suggesting the presence of cognitive subgroups. The goal of this study was to identify the underlying cognitive phenotypes of TLE and characterize their associated neurobiology using resting-state functional connectivity (RSFC). Methods: Participants included 55 healthy controls and 111 TLE patients aged 18 to 60, all of whom underwent assessment of intelligence, language, memory, executive function, speeded performance and motor dexterity using traditional cognitive tests and NIH Toolbox measures. All test scores were converted to age-adjusted z-scores. Measures were assigned to cognitive domains supported by Cronbach alpha and independent factor analysis, with mean cognitive domain scores then subjected to k-means clustering. Resting state fMRI images (TR=0.8s, 2.0mm isotropic) were acquired for 20 minutes with 3T GE scanners using simultaneous multi-slice imaging, and processed using the Human Connectome Project pipelines. Images were bandpass filtered (0.01-0.1Hz) and white matter, cerebrospinal fluid, global signal, and motion vectors were regressed out. Glasser parcellation and FreeSurfer subcortical regions were extracted. Pairwise Pearson correlations and Fisher’s z transformations were computed to generate connectivity matrices (a total of 71,631 connections). Nine patients did not complete MRI scans and were excluded from the following analyses. For each cluster of TLE patients, the association of group (TLE cluster vs. healthy) with the RSFC was tested using a generalized linear model, with chronological age as a covariate. Benjamini-Hochberg false discovery rate correction was made on all p-values. Results: Consistent with published literature, all mean cognitive domain scores for the TLE group were significantly lower than controls. However, k-means clustering identified three subgroups (Figure 1): Cluster 1 (18% of TLE group) exhibited significant generalized cognitive impairment affecting all domains, Cluster 2 (31%) exhibited particularly depressed language, memory and executive function/processing speed, and Cluster 3 (51%) was intact and comparable to controls across all domains. fMRI results (Figure 2) for Cluster 1 showed 167 significantly (corrected p<0.05) increased connectivity (hyperconnectivity) and 503 decreased connectivity (hypoconnectivity) compared to healthy controls. Most changes were seen from the bilateral temporal and medial frontal lobes. Cluster 2 showed 5 significant hyerperconnectivity and 13 hypoconnectivity differences compared to controls. Cluster 3 showed no significant differences from controls. Years of education (
Behavior/Neuropsychology/Language