For decades, electroencephalogram (EEG) has been a cornerstone for prognostication of comatose patients after cardiac arrest. Along with clinical examination, somatosensory evoked potentials (SSEP), serum biomarkers, and brain imaging, 鈥渉ighly malignant patterns鈥� (suppressed or burst-suppressed background with or without periodic discharges) are routinely used to identify patients with poor functional prognosis.¹ About one-third of patients may show epileptiform abnormalities鈥攕pikes, sharp waves, or spike waves鈥攊n the first days, often taking a periodic or rhythmic pattern (RPP, along the ictal-interictal continuum) or even fulfilling criteria for electrographic seizures. These patterns have traditionally been considered to herald poor prognosis and may be coupled with myoclonic jerks in two-thirds of patients.^2,3

However, several reports have described patients who awoke from coma despite epileptiform abnormalities on EEG,^4,5 challenging the historical pessimism and suggesting that pharmacological interventions treating epileptiform abnormalities under video EEG monitoring may be beneficial. While advocates of this strategy underscore the necessity of attempting to save each patient, others argue that systematically allocating limited resources to patients with a low likelihood of recovery is inefficient, generates unrealistic expectations among caregivers and relatives, and may increase the proportion of patients with severe neurological deficits, including unresponsive wakefulness, which is observed relatively frequently in environments that culturally do not foresee limitation of maximal care. The clinical community has struggled to reconcile these differing perspectives.

The Treatment of Electroencephalographic Status Epilepticus after Cardiopulmonary Resuscitation (TELSTAR) trial² was established to build, for the first time, a stronger evidence base to guide clinicians in these challenging scenarios. Comatose patients with RPP on continuous EEG within 3 days of cardiac arrest were randomized to standard care vs aggressive antiseizure medication treatment with levetiracetam, phenytoin, or valproate, combined with propofol or midazolam, and escalation to high-dose barbiturates if needed to suppress RPP for 48 hours. Cerebral performance categories (CPCs) at 3 months represented the primary outcome, dichotomized into favorable (CPC 1-2, representing no to moderate disability) vs unfavorable (CPC 3-5, representing several disabilities, unresponsive wakefulness, or death).² Among 172 patients, 79 of 88 patients in the treatment group (90%) and 77 of 84 control group patients (92%) had an unfavorable outcome (95% CI of the difference, 鈭�7% to 11%; P鈥�=鈥�.68), demonstrating that prognosis at 3 months in that cohort was unaffected by EEG-guided treatment, thus reinforcing the view that treating epileptiform EEG patterns is futile in this setting.

However, a prespecified subgroup analysis showed that among 36 patients with electrographic seizures or evolving, rhythmic EEG patterns, 6 of 20 patients randomized to the pharmacologic treatment group (30%) had a favorable outcome compared to 0 of 16 patients randomized to standard care. While the study was underpowered to examine this subgroup and this difference was nonsignificant, for some observers, this finding (in addition to other issues within the trial design) challenges the simple, pessimistic conclusions from the primary analysis.

While the TELSTAR investigators used multicenter recruitment to enroll a clinically diverse population, about 80% of participants displayed generalized periodic discharges, which have been shown to be poorly responsive to treatment,⁵ likely diluting the potential treatment effect (ie, inclusion bias). Additionally, the trial included patients with certain types of EEG patterns not belonging to the ictal-interictal continuum (such as generalized rhythmic delta) that most specialists would not consider justifying antiseizure treatment. Roughly one-third of patients displayed suppressed or burst-suppressed EEG backgrounds, which are widely regarded as 鈥渉ighly malignant鈥� patterns strongly related to unfavorable outcomes, especially if lacking background reactivity.⁶ Furthermore, including patients solely according to EEG results, without considering other routinely recommended ancillary prognostic tests,¹ may have reduced the sensitivity to detect favorable outcomes. Finally, the use of phenytoin could aggravate myoclonus in this population at high risk, thus possibly reducing favorable outcomes.

While awaiting a larger trial on a selected cohort with electrographic seizures, how should clinicians manage comatose patients after cardiac arrest with epileptiform abnormalities or RPP on EEG (Figure)? As clinical situations are virtually never absolute, a pragmatic approach should avoid generalized assumptions and instead be informed by multimodal assessments including, but not limited to, EEG. In assessing EEG results, diffuse, high-voltage poly-spikes that are time-locked with myoclonus, superimposed on a suppressed background, strongly correlate with poor prognosis, while lower-voltage, midline spikes on a continuous background may represent early forms of Lance-Adams posthypoxic myoclonus, which is treatable.⁷ Additional EEG characteristics, such as the timing of the appearance of epileptiform abnormalities (the earlier the worse, despite routine administration of 纬-aminobutyric acid [GABA]-mediated agents that are also anticonvulsants), background continuity (the earlier the better, reflecting a timely recovery of brain function),⁸ and the presence of a reactive background (ie, better prognosis) should also be considered. These aspects have been summarized in a validated EEG score.³ After assessing other ancillary tests, it seems reasonable to initiate antiseizure treatment in patients with results compatible with favorable prognosis rather than poor outcome, such as patients with preserved brainstem reflexes, intact cortical SSEP, and low serum NSE, and those without widespread alterations in diffusion-weighted magnetic resonance imaging.^9,10 In the absence of high-level evidence, we recommend antimyoclonic compounds (eg, intravenous levetiracetam or valproate or perampanel, topiramate, or zonisamide through the nasogastric tube), escalating to general anesthetics if needed.⁹ Treating these patients for at least a few days and up to 2 to 3 weeks seems reasonable, following the clinical evolution (awakening to follow command on sedation weaning) and the EEG (limitation or resolution of epileptiform features).^5,9 For patients with results not compatible with favorable prognosis, antiseizure treatment may be warranted for symptomatic relief and comfort to mask the clinical manifestations of seizures and myoclonus, without aiming at prolonging life. Above all, a patient-tailored approach should consider family preferences and beliefs, premorbid conditions, and directives.

Further research is clearly needed in this challenging field. An ideal randomized clinical trial should at least initially enroll an adequate cohort of comatose cardiac arrest survivors showing epileptiform EEG patterns with or without clinical counterparts, excluding those fulfilling criteria of 鈥渉ighly malignant鈥� features (ie, unreactive, suppressed, or burst-suppressed background) and/or those with ancillary tests that herald poor prognosis (eg, absent brainstem reflexes, lack of cortical SSEP, markedly elevated serum markers, or severe alterations on brain imaging) to focus on the population that may be amenable to meaningful clinical improvement.

Corresponding Author: Andrea O. Rossetti, MD, FAES, Department of Neurology, Lausanne University Hospital and University of Lausanne, CHUV-BH07, CH-1011 Lausanne, Switzerland (andrea.rossetti@chuv.ch).

Published Online: November 11, 2024. doi:10.1001/jamaneurol.2024.3831

Conflict of Interest Disclosures: None reported.