The use of veno - arterial extracorporeal membrane oxygenation (VA - ECMO) is recommended in the current resuscitation guidelines of the European Resuscitation Council as a rescue attempt for selected patients for whom extended resuscitation measures do not lead to the establishment of a spontaneous circulation (ROSC). We aim to characterize and investigate the prognostic value of neuron specific enolase (NSE) for assessing the neurological prognosis after cardiopulmonary resuscitation (CPR) as these have so far only been evaluated in resuscitated patients without extracorporeal circulatory support.

 

We retrospectively analyzed the records of patients, who received a VA - ECMO under ongoing CPR between 2004 and 2021 and were treated in the cardiac intensive care unit (ICU) of the University Hospital Jena. Outcome will be measured clinically by using the Glasgow Outcome Score (GOS) four weeks after ICU discharge. Neurological prognosis will be estimated by serum concentration of NSE (baseline until 96h). Serum free hemoglobin (fHb, baseline until 96h) serves as a marker for identifying a potential confounding effect of relevant parallel hemolysis. 

 

190 patients were included in our study (baseline characteristics, table 1). 154 patients (84.5%) died within 4 weeks after ICU admission or remained unconscious (GOS 1 + 2) and 29 patients (15.5%) survived with a residual slight to severe neurological deficit (GOS 5; 4; 3). Mean duration of VA - ECMO treatment was 103.4 ± 112.1h. In a wake - up attempt 48 - 96h after CPR, patients with GOS 3 - 5 showed significantly more frequent adequate responses to speech (86.7% vs. 30.5%; p <0.001). Starting 48h after CPR, NSE was significantly lower and continued to decrease in patients with GOS 3 - 5 compared to the group with an unfavorable outcome of GOS 1 + 2 (48h: 30.6 ± 39 ng/ml vs. 74.4 ± 97.1 ng/ml // 72h: 26.5 ± 10.8 ng/ml vs. 96.9 ± 149.1 ng/ml // 96h: 25.8 ± 20.2 ng/ml vs. 103.1 ± 170.3 ng/ml; p < 0.01). In addition, when evaluating on the basis of receiver operating characteristic curves (ROC) relevant and stable area under the curve (AUC) values for NSE could be calculated (48h: 0.82 // 72h: 0.81 // 96h: 0.86; p < 0.001, Figure 1) highlighting the level of discrimination between both outcome groups. Serum fHb measurements remained uneventful (24h: 16.9 ± 18.1 ng/ml; 48h: 16.1 ± 18.3 ng/ml; 72h: 15.9 ± 18.4 ng/ml; 96h: 16.5 ± 18.8 pg/ml) and no correlation between NSE and fHb could be found on the respective days (48h: r= 0.13; p = 0.28; 72h: r= 0.12; p= 0.25; 96h: r = 0.097; p = 0.47), so that the extent of known hemolysis induced during the VA - ECMO was not decisive for the level of NSE concentration. On the basis of a binary logistic regression model, relevant odds ratios for the NSE values were found even after adjusting for fHb and the respective adjusted AUCs of the combined predictive probabilities were significant (48h: 0.78 // 72h: 0.76 // 96h: 0.76; p = 0.001, Figure 1).

 

The neuroprotein NSE and clinical examinations make an early, neurological, prognostic assessment of resuscitated patients possible even during VA – ECMO treatment