COVID-19 EEG Studies: The Other Coronavirus Spikes We Need to Worry About.

Commentary

By now it is well-known that neurological manifestations occur in ∼35% of COVID-19 patients, and potentially in a majority of patients with more severe infections.[1-3] Results of electroencephalogram (EEG) in these patients are of particular interest: (1) How frequently is epileptiform activity seen in this patient population, and are the underlying risks for epileptiform activity different from other critically ill patients? (2) Are there novel EEG features specific to COVID-19? (3) Is the EEG informative in guiding the understanding of the pathophysiology of the frequently observed altered mental status, or prognosis, in COVID-19?

We are now starting to see the early COVID-19 EEG studies in the literature. Obtaining EEG in a safe manner to limit exposure will necessarily limit the number of patients in these studies.[4] Thus, early studies likely represent highly select groups of patients in whom suspicion of seizures was likely substantial. The range of prevalence of electrographic seizures in the intensive care unit (ICU) ranges between ∼10% to 45%,[5,6] with the largest series estimating ∼15%.[7] The prevalence of electrographic seizures in sepsis is ∼15%,[8] although it is lower in patients with sepsis without evidence of acute central nervous system disease.[9] These statistics can serve as a useful baseline.

The following 3 papers are peer-reviewed studies that systematically assessed EEGs in patients with COVID-19, as of July 26, 2020. That the study authors, clinicians engaged in the momentous task of treating these patients in high prevalence hospitals, were able to quickly disseminate this information is a noteworthy and commendable achievement.

The first, published in the Annals of Neurology (Vespignani et al[10]) is an observational study performed across multiple ICU departments in Paris during a 2 week period in March 2020. Electroencephalograms were done with a limited 9-electrode (plus reference) device for 30 minutes. They noted that 19 of 26 had “diffuse and nonspecific theta and alpha wave activity” which likely refers to nonspecific slowing. Two patients had isoelectric EEGs. A total of 5 patients were reported to have periodic discharges. Examining the snapshots of the EEGs, 4 of these were labeled as generalized rhythmic delta activity by the authors (which would technically not be periodic) and one of them as lateralized periodic discharges. In my review, 4 of those 5 did have periodic activity. All of them were critically ill, and 3 of them died before completion of the study. The authors suggest that the findings in these 5 patients could be indicative of brain injury, either related to COVID-19 directly or concurrent illness. The presented examples in the manuscript do not necessarily appear to me to differ significantly from patterns seen in other critically ill patients. In fact, what seems unusual and reassuring is that only 1 of 26 patients had focal discharges, and no patient had electrographic seizures.

The second study, published in Epilepsia Open, originates from Albert Einstein College of Medicine, one of the epicenters of the COVID-19 pandemic in New York City during a 6-week period in March to April 2020 (Galanopoulou et al[11]). Electroencephalograms, utilizing either a 10-electrode device or full montage EEGs, were obtained from acutely ill patients who were under investigation for COVID-19. A total of 28 patients were included, most of whom would eventually test positive for COVID-19 (22 patients). The EEG readers were aware of the results of at least some of the COVID testing a priori, potentially introducing both selection and confirmation bias. As expected, all EEGs had abnormal background slowing. Although the high prevalence of COVID-19 did not allow for a more statistically favorable split in this study (thus resulting in mostly nonsignificant results), there were some provocative findings: COVID-positive patients had more seizure-like events (motor events or confusion) prompting the EEG request (63% vs 33%), new encephalopathy (68% vs 33%), sporadic epileptiform discharges (41% vs 17%), and rhythmic/periodic discharges (18% vs 0%). A significantly greater proportion of COVID-negative patients (83% vs 24%) were found to have some other infection other than COVID. This study suggests that perhaps COVID-positive patients differ from a similar COVID-negative cohort. Again, as in the previous study, no electrographic seizures were recorded.

The third study, a single institution retrospective review from France (Petrescu et al[12]), reports 40 consecutive routine EEGs from 36 patients who tested positive for COVID, after excluding brain death studies and asymptomatic patients, half of the patients from the ICU. Electroencephalograms were normal or nearly normal in a little more than half of the recordings (57.5%). Patients with more severe abnormalities on EEG were associated with concurrent neurological disorders, including posterior reversible encephalopathy syndrome, malignancy, or dementia. Periodic discharges were seen in 8 patients, but it was felt that none of the patients had epileptiform activity.

Collectively, these studies appear to suggest that encephalitis and cortical irritability may not be major features of central nervous system manifestations of COVID-19, supporting findings that the virus or its specific encephalitic manifestations are not typically found in brain tissue.[13] Initial reports from China that there was no evidence of increased risk of symptomatic seizures from COVID-19 are reassuring.[14] This is, after all, still a primarily non-neurological disease. There are notable limitations to these early studies. The sample sizes are tiny, and studies obtained are by and large short routine EEGs, which are unlikely to record an electrographic seizure. Studies with sample sizes an order of magnitude larger, with continuously recorded EEG studies, and with recorded longer-term clinical outcomes are required for more reasonable assessments of the full spectrum of EEG abnormalities and their consequences; a multicenter consortium may be a reasonable way to address this.[15] There are already several case reports of seizures in the setting of COVID-19 as well as preprints of continuous EEG studies detailing potentially greater risk of epileptiform activity. Thus, full “damage reports” are still pending, and many questions remain unanswered.