Postdoctoral Researcher University of Pennsylvania, Pennsylvania
This abstract is a recipient of the Grass Foundation Young Investigator Award This abstract will be presented during the Neuro Imaging Platform poster session
Rationale: Working memory (WM) is a building block for all cognitive tasks, and is impaired in focal epilepsies, including temporal and frontal lobe epilepsy (TLE, FLE), as well as in genetic generalized syndromes, such as juvenile myoclonic epilepsy (JME). Prior functional MRI (fMRI) work used conventional, static metrics of activity and connectivity, that cannot capture the time-varying, dynamic reconfigurations of brain networks elicited by task demands. Thus, network mechanisms accounting for epilepsy-related WM dysfunction remain poorly characterized, and knowledge of whether different syndromes present with unique neural signatures, which can inform targeted treatment approaches, is also lacking. Here, we apply the emerging framework of dynamic network analysis to a large (n=274) fMRI sample, decode network mechanisms of WM dysfunction in epilepsy, and identify syndrome-specific and shared patterns in three common syndromes, capturing trait distributions across the disease spectrum. Methods: We analyzed 120 TLE patients (59 left), 62 FLE patients (36 left), 37 JME patients, and 55 controls comparable in age, sex, and handedness, who completed IQ, WM and mental flexibility tests, and a 3T-fMRI visuo-spatial WM task. After preprocessing and regional parcellation, we applied open source MATLAB code to obtain sliding windows of functional connectivity matrices, coupled these into a multi-layer network, derived network modules, and tracked their evolution during the task (Figure 1). To quantify dynamic reconfigurations on the backbone of canonical subcortical and cortical (Yeo) networks, we estimated (a) recruitment, the stability of a given cognitive network, and (b) integration, the frequency of transient interactions among different networks. We compared groups via multivariate analysis of variance, after adjusting for age and sex (pFDR< 0.05). Results: TLE and FLE had lower scores than controls in all tests; only mental flexibility was impaired in JME. Compared to controls (Fig.2), all patient groups had reduced stability of salience (SAL) and default-mode networks (DMN). Enhanced subcortical and reduced frontoparietal stability were common to JME and FLE, while impaired stability of the dorsal attention network (DAN) was specific to TLE. The DMN had abnormally high integration to “task-positive” networks (DAN, SAL) in both focal epilepsies. In JME, DMN changes were less marked, whereas we found enhanced integration of the somatomotor with both DAN and frontoparietal control network, that discriminated patients from controls (AUC= 0.73/0.69, p< 0.001/0.004). Increased integration between somatomotor and frontoparietal control networks, however, was also detected in FLE and TLE. Task performance significantly correlated with dynamic metrics (partial Spearman rho= 0.17/-0.19, p< 0.008 for SAL stability and integration with DMN). Conclusions: We show widespread alterations of cognitive network dynamics in epilepsy. Impaired stability of the DMN and salience network is a shared dysfunctional marker, and points to impaired architecture of high-order systems. In focal syndromes, abnormal interactions mostly involve DMN and task-positive networks, while altered communication between somatomotor and task-positive systems is more prominent in JME. Enhanced interactions affecting somatomotor and frontoparietal control networks, however, also represent a trans-syndromic feature. Collectively, these patterns point to abnormal segregation of large-scale networks in common epilepsies. Our findings imply a conceptual change in our understanding of cognitive dysfunction across the epilepsy spectrum, and pave the way for future application of this framework to other cognitive domains. Funding: Please list any funding that was received in support of this abstract.: NINDS (R01NS099348).
