How Useful Is EEG in Determining Prognosis After Global Brain Hypoxia/Ischemia?

Abstract & Commentary

By Elayna Rubens, MD, Assistant Professor of Neurology, Weill Cornell Medical College. Dr. Rubens reports no financial relationships relevant to this field of study.

Synopsis: In comatose survivors of cardiac arrest, early malignant EEG findings during therapeutic hypothermia correlated with higher serum neuron-specific enolase levels and poor neurologic outcome.

Source: Rossetti AO, et al. Early EEG correlates of neuronal injury after brain anoxia. Neurology 2012;78:796-802.

Therapeutic hypothermia (TH) improves outcomes in comatose survivors of cardiac arrest, but its use may affect previously validated prognostic parameters in anoxic coma. Re-evaluating the reliability of such parameters in TH is currently the subject of great interest and debate. In this study, Rossetti et al examine the relationship between early EEG findings during hypothermia, serum neuron-specific enolase (NSE) level, and neurologic outcome.

This is a prospective, observational study of patients undergoing mild TH after cardiac arrest who were evaluated with either continuous EEG (for 24-48 hours) or routine EEG (20-30 minutes) recordings during hypothermia. NSE levels were measured 24-48 hours after cardiac arrest. All EEGs included an assessment of background reactivity performed more than 6 hours after initial cardiac arrest. EEGs were evaluated (blinded to NSE results) for the presence of the following "malignant" features: unreactive EEG background, discontinuous background, and epileptiform activity. Functional neurologic outcome was assessed at 3 months according to the Cerebral Performance Categories (CPC) scale.

Of the 61 patients enrolled, 34 survived. Of these, 28 (85%) had a good neurologic outcome at 3 months while six (18%) were severely disabled. All patients with an unreactive EEG background during hypothermia died. Patients with early epileptiform activity also uniformly had unreactive EEG backgrounds and died. Of those with discontinuous patterns (spontaneous burst suppression), four (14%) awoke and two had a good neurologic outcome. Patients with unreactive and/or discontinuous EEG backgrounds had higher NSE levels. A trend toward higher NSE levels was seen in patients with early epileptiform activity, but this did not reach significance. Of note, five survivors had NSE levels greater than 33 µg/L (the AAN guideline threshold for poor prognosis), three of whom had a good neurologic recovery.

The authors conclude that the correlation between early malignant EEG features and higher NSE levels supports the notion that the observed EEG abnormalities reflect permanent anoxic injury rather than transient post-anoxic cerebral dysfunction. The results of the study confirm the prognostic validity of early EEG findings obtained during the hypothermic period, with absent reactivity and epileptiform discharges portending a universally poor outcome. Discontinuous EEG patterns and elevated NSE level > 33 µg/L were highly associated with poor outcome, though neither precluded functional neurologic recovery.


This study is an important contribution to the expanding literature regarding neurologic prognostication after cardiac arrest in the posthypothermia era. It is the first study to look specifically at EEG findings obtained during hypothermia and correlate them with both outcome and a surrogate marker of neuronal injury: NSE. The findings suggest that malignant EEG patterns more reliably predict poor neurologic outcome than do NSE levels alone, though the two assessments are complementary.

The correlation between EEG findings and NSE levels implicates neuronal injury as the likely basis for the background EEG abnormalities. Though this study was not designed to establish pathophysiologic causality, the implication of the findings is clear: an unreactive or discontinuous EEG background during hypothermia suggests structural brain injury that is unlikely to be reversible. EEG pattern during hypothermia appears to be a reliable marker of widespread brain injury and poor prognosis, but the findings need to be replicated in a multicenter study and correlated with additional indicators of neuronal injury (imaging findings and neuropathology) before clinical implementation.

The lack of statistically significant correlation between early epileptiform activity and NSE levels may be due to the small number of patients with epileptiform EEG features in this study, but is nevertheless thought provoking. Is the pathophysiologic correlate of epileptiform activity during TH different from the underlying EEG background abnormalities? Should this early epileptiform activity be treated? Although this study suggests that epileptiform EEG and elevated NSE are predictive of a poor prognosis not amenable to treatment, a larger study is needed to define the role of early antiepileptic therapy. There may be a subset of patients with early epileptiform activity (such as those with lower NSE levels) who might benefit from more aggressive antiepileptic treatment.

Although the study highlights the value of early EEG findings in TH, it does not alter the current recommended practice of prognostication in the setting of TH, which should continue to rely on multiple clinical and physiologic measurements.