Personalized neuron-specific enolase level based on EEG pattern for prediction of poor outcome after cardiac arrest

Background After cardiac arrest (CA), the European recommendations suggest to use a neuron-specific enolase (NSE) level > 60 mu g/L at 48-72 h to predict poor outcome. However, the prognostic performance of NSE can vary depending on electroencephalogram (EEG). The objective was to determine whether the NSE threshold which predicts poor outcome varies according to EEG patterns and the effect of electrographic seizures on NSE level. Methods A retrospective study was conducted in a tertiary CA center, using a prospective registry of 155 adult patients comatose 72 h after CA. EEG patterns were classified according to the Westhall classification (benign, malignant or highly malignant). Neurological outcome was evaluated using the CPC scale at 3 months (CPC 3-5 defining a poor outcome). Results Participants were 64 years old (IQR [53; 72,5]), and 74% were male. 83% were out-of-hospital CA and 48% were initial shockable rhythm. Electrographic seizures were observed in 5% and 8% of good and poor outcome patients, respectively (p = 0.50). NSE blood levels were significantly lower in the good outcome (median 20 mu g/L IQR [15; 30]) compared to poor outcome group (median 110 mu g/l IQR [49;308], p < 0,001). Benign EEG was associated with lower level of NSE compared to malignant and highly malignant patterns (p < 0.001). The NSE level was not significantly increased in patients with seizures as compared with malignant patterns (p = 0.15). In patients with a malignant EEG, a NSE > 45.2 mu g/L was predictive of unfavorable outcome with 100% specificity and a higher sensitivity (70.8%) compared to the recommended NSE cut-off of 60 mu g/l (Se = 66%). Combined to electrographic seizures, a NSE > 53.5 mu g/L predicts poor outcome with 100% specificity and a higher sensitivity (77.7%) compared to the recommended cut-off (Se = 66.6%). Combined to a benign EEG, a NSE level > 78.2 mu g/L was highly predictive of a poor outcome with a higher specificity (Sp = 100%) compared to the recommended cut-off (Sp = 94%). Conclusion In comatose patients after AC, a personalized approach of NSE according to EEG pattern could improve the specificity and sensitivity of this biomarker for poor outcome prediction. Compared to others malignant EEG, no significant difference of NSE level was observed in case of electrographic seizures.