EEG Abnormalities and Their Radiographic Correlates in a COVID-19 Inpatient Cohort

Discussion

This study represents the largest COVID-19 EEG database in the literature, with a high proportion of patients demonstrating epileptiform findings and nonreassuring periodic patterns.^15,-,18 The in-hospital mortality of this cohort of COVID-19 inpatients requiring EEG was relatively high (37.5%), although consistent with the mortality rate of 44% found in another New York-based hospital system around this period.¹⁹ A considerable number (18.2%) of patients in our study had documented events of clinical concern for seizure before the EEGs being ordered, and a very high proportion of patients presented with encephalopathy or coma. Over a third of patients in our cohort demonstrated overt epileptiform spiking or periodic patterns of concern, although seizures were relatively infrequent, observed in 8 patients (4.3%). For comparison, 1 previous study noted a smaller percentage with epileptiform abnormalities (27%) and a higher number of patients with recorded seizures (7%).¹⁹ However, within our cohort, 3 patients were found to be in NCSE, a particularly serious and treatable condition. Older age, and not surprisingly, ASM administration and epilepsy history were significantly associated with epileptiform abnormalities on EEG. However, only approximately a quarter of the patients with epileptiform abnormalities or electrographic seizures on EEG had known epilepsy, suggesting that most findings were de novo.

In many reports of COVID-19, EEG data are conspicuously lacking. This multifaceted problem may reflect limited availability of EEG and subspecialists in nontertiary hospitals or a purposeful reduction in EEG procedures to decrease exposure risk. Most patients in this cohort had at least 1 continuous EEG recording, with extensive electrographic data and greater chance of recording pertinent abnormalities. The relatively novel approach of RM-EEG was also used, technology potentially suited to the critical care setting for rapid evaluation for seizures. The potential therapeutic implications of EEG studies must be weighed carefully against increased SARS-CoV-2 transmission risk to both patients and healthcare staff. Recent recommendations include utilization of personal protective equipment, disposable electrodes, RM-EEG, and avoiding hyperventilation as an activating procedure.²⁰ In our study, precautions included the above protocols and clinical prescreening, and reduced staffing models. In a previous study of 26 critically ill patients with COVID-19 with altered mentation, most showed diffuse slowing on 30 minute EEG recordings; however, 2 patients had isoelectric EEG consistent with brain death and 5 cases demonstrated frontal maximal GPDs similar to those in this report.¹⁵ Three of the 5 cases of GPDs were associated with myoclonus, and 3 of those patients expired. At least 1 patient had this pattern in the setting of acute encephalopathy without intubation or sedation and diffuse white matter hyperintensity on MRI with normal lumbar puncture results.

A study from France of 36 patients with COVID-19, half critically ill, also showed GPDs or multifocal periodic discharges in 32.5% of routine EEGs, 3 performed in patients with unexplained encephalopathy.¹⁶ EEG was normal or only mildly abnormal in about half of the cases.

Galanopoulou et al.¹⁷ studied 22 COVID-19–positive patients with 2 routine EEGs and 20 RM-EEGs. Sporadic epileptiform discharges were recorded in 40.9%, with frontal maximal sharp waves present in 8 of 9 patients. Periodic or rhythmic discharges occurred in 18.2% of cases. EEG characterization was limited by the predominant use of RM-EEG used to a greater proportion compared with this study.

In an Italian cohort of 15 patients with COVID-19–related encephalopathy, all but 2 patients with anoxia had either severe attenuation or status epilepticus, all with varying degrees of slowing, although without periodic activity noted.¹⁸ One patient demonstrated frontal intermittent generalized rhythmic delta activity. Elevated CSF protein was demonstrated in 1 patient, and MRI showed T2 hyperintensities in 2 patients.

Researchers have pondered whether GPDs may be an EEG pattern specific to COVID-19 encephalopathy.²¹ It is more probable that GPDs reflect severity of illness as opposed to a pathognomonic finding because GPDs are common, in up to 4% of patients in critical care settings undergoing EEG monitoring. Associated etiologies typically involving diffuse cortical or thalamocortical dysfunction.²² GPDs may increase the risk for associated electrographic seizures in at least 16% of patients.²³ Features such as motor manifestations, faster discharge rate, and response to ASMs may be associated with a higher risk of GPDs evolving to seizures or representing NCSE. Follow-up EEG monitoring, trials of nonsedating ASMs, or short-acting benzodiazepines may be warranted to assess clinical response.

The significance of GPDs with triphasic morphology is more controversial and classically described, associated with metabolic encephalopathy. Recent literature supports their occurrence with multiple underlying etiologies—only modest interrater EEG reader agreement regarding triphasic classification, but increased risk of seizures similar to GPDs without triphasic morphology.²²

Focal IEDs, LPDs, or LRDA were observed in a quarter of patients in our study, indicating an elevated risk of focal onset seizures. A recent Critical Care EEG Monitoring Research Consortium study showed LRDA and LPDs, in particular, associated with seizures in 28% and 44% of EEG studies, respectively.²³ Seizure risk approached nearly two-thirds of patients in highest risk subgroups, with a high frequency of discharges and intermixed fast frequency activity or spiking.

In this study, we correlated epileptiform EEG findings with radiographic abnormalities to understand what proportion of patients with these EEG findings may be secondary to new acute lesions related to COVID-19 or because of exacerbation of preexistent lesions. As stated, focal IEDs were found in a quarter of patients, LPDs in 6.3%, and focal slowing in 15.6% of cases, hence raising concern for focal cerebral pathology. The presence of focal epileptiform or periodic EEG abnormalities was associated with a high rate of neuroimaging findings (72.9%), with acute lesions identified among two-thirds of those. Acute imaging abnormalities included ischemic and hemorrhagic stroke, watershed lesions, hypoxic leukoencephalopathy, subdural and subarachnoid hemorrhages, focal inflammatory changes, leptomeningeal enhancement, and PRES. Likewise, GPDs were found in a relatively high proportion of cases (19.3%), likely reflecting severe patient condition, depth of encephalopathy, and more widespread or global cerebral dysfunction because of metabolic factors, anoxia, infection, or inflammatory factors in critically ill patients with COVID-19. Indeed, in this study, patients with coma were 2.6 times more likely to expire during the hospital stay after adjusting for mechanical ventilation use.

The data presented in this study reflect a range of imaging neurologic abnormalities potentially associated with COVID-19. Although it is difficult to determine causality in our cohort, some findings may reflect a prothrombotic state because of immune activation, endothelial cell injury secondary to ACE2-mediated viral tropism, hypoxic ischemia, or cytokine storming. These pathophysiologic mechanisms may lead to acute infarcts with heavy clot burdens, intracranial hemorrhages, leukoencephalopathy, global hypoxic injury, cytotoxic lesions of the corpus callosum, olfactory bulb involvement, and cranial nerve enhancement.²⁴

In a large retrospective study from China, new-onset serious neurologic complications were noted in 32 of 917 patients, mostly stroke or altered mentation.⁹ The authors concluded that evidence for CNS involvement because of SARS-CoV-2 was lacking; however, in attempting to reduce virus transmission, no patients were evaluated by MRI or EEG. Head CT was performed in only 9 cases. However, several subsequent case reports demonstrated CNS-related complications including acute necrotizing encephalopathy, demyelination, stroke, cortical venous sinus thrombosis, PRES, and seizures, indicating serious potential neurologic effects of SARS-CoV-2, possibly from direct CNS invasion or secondary effects from autoimmunity or hypercoagulability.^7,25,26

In a recent review of 61 studies published between 2019 and 2020, radiographic abnormalities were differentiated on mild or severe clinical presentation secondary to COVID-19. Radiographic abnormalities most often included white matter abnormalities, ischemic stroke, and intracerebral hemorrhage.²⁴ Less often, abnormalities included leptomeningeal enhancement, venous thrombosis, PRES, olfactory bulb lesions, and encephalitis.²⁴ The diverse neuroradiographic findings and overlap of severity indicates the utility of radiographic imaging for the management and treatment patients with COVID-19.

Limitations of this study include the retrospective nature of the analysis because causal relationships between the variables presented cannot be drawn from the data. There was inherent variation in clinical decision-making between providers during COVID-19 redeployment, including when an EEG was ordered and, at times, in the absence of neurologic consultation. The high mortality rate may have reflected triaging of EEG resources and a selection bias toward recording more neurologically ill patients, where EEGs less likely to affect clinical decision-making were not performed because of concerns for contagion. Because our data were derived from a large healthcare system with 23 regional hospitals, patients transferred for higher levels of care to the tertiary hospitals may have created a referral bias toward more severe cases. This cohort represents a fraction of the estimated >15,000 patients with COVID-19 admitted to the health system during this time, specific to inpatient COVID-19 cases with neurologic complaints and not necessarily relevant to lesser acute cases. The duration of this study was limited and thus did not provide long-term outlook for persistent electrographic abnormalities or neurologic outcome. Additional variables such as the presence of superinfections, CSF, and cytokine laboratory values may be subject to further study.

This study represents the largest series of EEGs in patients with COVID-19 and highlights the importance of fully assessing neurologic complications that may occur, especially in the inpatient or critical care setting. Epileptiform and periodic patterns on EEG were frequently observed and often associated with the presence of acute neuroimaging abnormalities. Mortality rates in this cohort were high and most elevated in patients with periodic EEG patterns such as GPDs and LPDs, with coma and mechanical ventilation requirements as major risk factors. EEG in patients with COVID-19 may play a critical role in detection of seizure-related patterns, identification of patients at greater risk for neuropathology, and prognostication. Patients with focal or potentially epileptiform patterns raise concern for acute brain lesions and may warrant neuroimaging evaluation. Specific patterns may signal need for further interventions such as ASMs, stroke therapies, or immunomodulatory therapy. Severe EEG abnormalities such as diffuse attenuation, burst suppression, and GPDs may be concerning for anoxia and the possibility of poorer prognosis. Although the data support seizures as a rare occurrence, their presence is concerning for neurologic complications associated with SARS-CoV-2. Most patients with electrographic and observed clinical seizures before EEG in this study had no known history of epilepsy.