In this study, we investigated whether intermittent EEG could be used to predict neurological outcomes of patients who underwent ECPR. Major findings of this study were as follows. First, regardless of sedation, malignant EEG patterns were more common in patients with poor neurological outcome than in patients with favorable neurological outcome. Especially, all patients with highly malignant EEG patterns had poor neurological outcome. In addition, patients with moderate malignant EEG patterns had poor neurological outcome except for one patient. Second, benign EEG patterns alone did not necessarily imply a favorable neurological outcome. Third, in multivariable analysis, malignant EEG patterns and CPR duration were significantly associated with poor neurological outcomes in patients who underwent ECPR. Therefore, early intermittent EEG scan and CPR duration could be helpful for predicting neurological outcomes of post-cardiac arrest patients after ECPR.
EEG signals mainly reflect cerebral cortical function and some subcortical function . EEG is very sensitive to ischemia because cortical neurons of the brain need consistent blood supply to maintain signaling and integrity . Therefore, EEG scan is a standard and useful tool to predict neurological outcomes after cardiac arrest [3, 15]. Especially, malignant EEG patterns such as suppressed background, status epilepticus, burst suppression, periodic patterns, and unreactive EEG are associated with poor neurological prognosis after cardiac arrest [8, 15, 18]. In addition, early continuous wave with normal voltage could be a predictor of favorable neurological outcome after cardiac arrest .
Cerebral autoregulation may be changed in survivors after cardiac arrest . Highly oxygenated continuous ECMO flow could affect cerebral autoregulation after ECPR . In addition, neurological outcomes may be affected by functional recovery of native heart and lung, the amount of ECMO support, and changed cerebral autoregulation . Altered cerebral hemodynamics by ECMO support may influence neurological outcome after ECPR. Therefore, it is difficult to predict neurological prognosis by these changed situations after ECPR . Ultimately, the interaction between cerebral autoregulation and ECMO flow may affect neurological recovery and prognosis in ECPR patients through mechanisms of primary ischemic damage and secondary additive injury . Thus, EEG change by this interaction should be studied for neurological prediction after ECPR. However, there has been no report of EEG according to neurological outcomes after ECPR.
Sedation may confuse outcome prediction in survivors of cardiac arrest [1, 8, 20]. Sedatives are commonly used in survivors after cardiac arrest for 72 hours as important confounders [1, 20]. A motor response to noxious stimuli, corneal reflex, caloric testing, and some electrophysiologic studies may also be confounded by sedation [20, 21]. Although mild to moderate hypothermia does not significantly affect EEG in patients with induced hypothermia [8, 22], a confounder accompanied by induced hypothermia such as analgesics, sedatives, or artifacts from shivering, mechanical ventilator, or electrical devices may affect the reliability of EEG interpretation . However, a recent study has reported that the predictive performance of EEG after cardiac arrest is similar between patients with ongoing sedation and those without ongoing sedation [8, 15]. In this study, sedation or targeted temperature management did not significantly affect the prediction of poor neurological outcome after ECPR. Regardless of sedation or targeted temperature management, patients with malignant EEG patterns had poor neurological outcome in this study.
Benign EEG patterns may be associated with a favorable neurological outcome in survivor after cardiac arrest . Especially, early continuous wave with normal voltage could be a predictor of favorable neurological outcome after cardiac arrest [8, 15, 19]. However, in previous studies, benign EEG patterns are not always associated with good neurological outcome [15, 18]. Additive secondary injury is characterized by an imbalance in post-resuscitation cerebral oxygen delivery and use . This injury is associated with reperfusion injury, impaired autoregulation, fluctuations in oxygen support and arterial carbon dioxide, hyperthermia, and concomitant anemia . Early EEG findings may not be shown to be malignant EEG patterns in patients with poor neurological outcome if the secondary cerebral injury is more serious than the primary cerebral injury. In this study, benign EEG patterns were not always associated with a favorable neurological outcome. In addition, intermittent EEG scan may be less sensitive for predicting favorable neurological outcome than continuous EEG monitoring in this study.
This study has several limitations. First, this was a retrospective review. Thus, CPC score was determined based on medical records. By using two independent specialists’ agreement on the score, any bias may be mitigated to some extent. In addition, although the cause of death had to be accurately verified, its identification was insufficient due to the retrospective nature of this study. Second, the nonrandomized nature of registry data might have resulted in selection bias. Particularly, during the study period, EEG scans were not performed in all patients. They were only performed in patients with abnormal consciousness, seizure, abnormal movements, or other symptoms. Although EEG scans were performed within 96 hours following ECPR, a major limitation of the study might be that EEG scans were performed in different time settings. Lastly, our study was conducted in a small cohort at a single institution. Therefore, future studies with larger cohorts are needed to confirm our findings.