Brain networks involved in generalized periodic discharges (GPD) in post-anoxic-ischemic encephalopathy

Pia De Stefano,Margherita Carboni,Deborah Pugin,Margitta Seeck,Serge Vulliémoz

doi:10.1016/j.resuscitation.2020.07.030

Pia De Stefano, Margherita Carboni + Show 3 more

Open Access

https://doi.org/10.1016/j.resuscitation.2020.07.030

Copy DOI

Journal: Resuscitation	Publication Date: Aug 11, 2020
Citations: 5	License type: cc-by

Affiliation: University of Geneva

Abstract

AimGeneralized periodic discharge (GPD) is an EEG pattern of poor neurological outcome, frequently observed in comatose patients after cardiac arrest. The aim of our study was to identify the neuronal network generating ≤2.5 Hz GPD using EEG source localization and connectivity analysis. MethodsWe analyzed 40 comatose adult patients with anoxic-ischemic encephalopathy, who had 19 channel-EEG recording. We computed electric source analysis based on distributed inverse solution (LAURA) and we estimated cortical activity in 82 atlas-based cortical brain regions. We applied directed connectivity analysis (Partial Directed Coherence) on these sources to estimate the main drivers. ResultsSource analysis suggested that the GPD are generated in the cortex of the limbic system in the majority of patients (87.5%). Connectivity analysis revealed main drivers located in thalamus and hippocampus for the large majority of patients (80%), together with important activation also in amygdala (70%). ConclusionsWe hypothesize that the anoxic-ischemic dysfunction, leading to hyperactivity of the thalamo-cortical (limbic presumably) circuit, can result in an oscillatory thalamic activity capable of inducing periodic cortical (limbic, mostly medial-temporal and orbitofrontal) discharges, similarly to the case of generalized rhythmic spike-wave discharge in convulsive or non-convulsive status epilepticus.

Highlights

The vast majority of patients resuscitated from cardiac arrest (CA) present in coma or with an altered level of consciousness caused by the widespread nature of brain injury,[1] and they show pathological patterns at electroencephalogram (EEG)
We used an atlas-based head model parcellation of the brain into 82 regions of interest (ROIs) (Fig. 2C) and applied a linear distributed inverse solution, as well as functional connectivity estimation between cortical sources as described in our previous work[13] (Supplementary S2, S3). These steps allowed us to estimate the cortical activity of each region of interest (ROI) and further calculate their time varying relation by means of Partial Directed Coherence (PDC), based on Granger Causality (Fig. 2D)
One patient died of a respiratory tract infection (RTI) 166 days after CA and two patients were still alive with GOS 4 at discharge in 2009, but lost to follow-up

Summary

Introduction

The vast majority of patients resuscitated from cardiac arrest (CA) present in coma or with an altered level of consciousness caused by the widespread nature of brain injury,[1] and they show pathological patterns at electroencephalogram (EEG).Certain EEG patterns À namely generalized periodic discharges (GPD), post-anoxic non-convulsive status epilepticus (NCSE), alpha coma and burst suppression or generalized suppression À are Abbreviations: CA, cardiac arrest; GPD, generalized periodic discharge; NCSE, non convulsive status epilepticus; ESI, Electrical Source Imaging; iPDC, Information Partial Directed Coherence; LSMAC, Locally Spherical Model with Anatomical Constraints; LAURA, Local AUtoRegressive Average; ROI, region of interest; ROIs, regions of interest.associated with a poor prognosis in comatose patients.[2]. The vast majority of patients resuscitated from cardiac arrest (CA) present in coma or with an altered level of consciousness caused by the widespread nature of brain injury,[1] and they show pathological patterns at electroencephalogram (EEG). Certain EEG patterns À namely generalized periodic discharges (GPD), post-anoxic non-convulsive status epilepticus (NCSE), alpha coma and burst suppression or generalized suppression À are Abbreviations: CA, cardiac arrest; GPD, generalized periodic discharge; NCSE, non convulsive status epilepticus; ESI, Electrical Source Imaging; iPDC, Information Partial Directed Coherence; LSMAC, Locally Spherical Model with Anatomical Constraints; LAURA, Local AUtoRegressive Average; ROI, region of interest; ROIs, regions of interest. Recent criteria support a diagnosis of NCSE in the presence of GPD with frequency !2.5 Hz or, in case of slower frequencies, either an electro-clinical improvement after intravenous antiepileptic drugs (within 10 min after antiepileptic drugs administration), or subtle clinical ictal phenomena or spatio-temporal evolution.[6]

Objectives

Methods

Results