Sudden cardiac arrest (CA) is one of the most common causes of death in the developed world, affecting nearly 556 000 people annually in the United States.1 Although a large proportion of CA deaths occur before or during resuscitation maneuvers, a substantial percentage of deaths occur in patients who have been successfully resuscitated and subsequently admitted to an intensive care unit. In these patients, most of initial mortality is secondary to hemodynamic failure, while brain injury accounts for most of the later deaths. In addition, withdrawal of life sustaining therapies (WLST) based on poor neurologic prognosis represents the most frequent cause of death in this subgroup.2,3 Therefore, neurological prognostication remains a significant clinical challenge in postresuscitation care and current guidelines recommend a multimodal approach combining clinical neurological examination, electrophysiological studies (somatosensory evoked potentials and electroencephalography), neuroimaging, and biomarkers, including neuron-specific enolase (NSE).4

NSE is a glycolytic enzyme involved in glucose metabolism and is mainly found in neuronal and neuroendocrine cells. Serum NSE concentrations are not affected by hypothermia and positively correlate with both the extent of anoxic-ischemic neurological injury and malignant electroencephalography changes after CA.5,6 A prospective study before the advent of targeted temperature management (TTM) reported that a serum NSE concentration higher than 33 μg/L on days 1 to 3 after CA was strongly associated with poor neurological outcome (false-positive rate [FPR] 0%)7 and the American Academy of Neurology subsequently introduced this cutoff value into the 2006 prognostication guidelines8. However, results were more variable in subsequent studies with unacceptably high FPRs being repeatedly reported when this serum NSE concentration threshold was used in patients whether they underwent TTM or not.9–11 For these reasons, and given the difficulty of finding a consistent threshold to identify patients at risk of a poor neurological outcome with a high degree of certainty, current resuscitation guidelines do not recommend any specific NSE threshold for prediction of poor neurological outcome, but instead suggest consecutive sampling at multiple time-points to establish trends in NSE levels and to reduce the risk of false-positive results due to hemolysis.4 Although the trend in serum NSE values could be more accurate than isolated values, the optimal approach, timelines and interpretation remain unclear.5

Since 2009, we have included assessment of NSE values at hospital admission and daily during the first 72hours of hospitalization as part of our acute cardiac care unit (ACCU) postcardiac arrest management protocol. The aim of this study was to analyze NSE kinetics and their correlation with neurological outcome in CA survivors treated with TTM in our ACCU.

We performed a retrospective analysis of a prospective database that included consecutive CA survivors who were treated with TTM at the ACCU of a single tertiary care center from September 2006 to May 2018. The study population included in- and out-of-hospital CA survivors of suspected cardiac origin, irrespective of initial rhythm and cooled to 32°C to 34°C for 24hours, as previously described.12 Patients were considered eligible if there was confirmation of both persistent comatose state after ROSC and absence of contraindications for TTM. We included only patients who had at least 1 serum NSE determination during ACCU hospitalization and did not die within first 72hours after admission. The study protocol was approved by the institutional ethics committee (PI-3328).

The TTM protocol of our center consisted of implantation of either intravascular or surface cooling automatic feedback devices and cooling patients to a target temperature of 32°C to 34°C for 24hours. Rewarming was controlled at a set rate of 0.1°C to 0.2°C/h until achievement of normothermia (36.5°C-37°C) and body temperature was continuously monitored through a bladder probe. All patients underwent mechanical ventilation and were sedated with intravenous infusion of midazolam adjusted upon continuous bispectral index (BIS) monitoring. Muscular relaxation was achieved by continuous perfusion of cisatracurium until completion of rewarming. If acute coronary syndrome was suspected as the underlying cause of CA, coronary angiography was performed in a 24/7 cath-lab.

Blood samples for NSE measurement were obtained at hospital admission and daily during the first 72hours of hospitalization. Blood samples were routinely processed by the hospital's Laboratory Medicine Department and serum NSE measurement was performed using a Cobas e411 electro-chemiluminescence assay (Roche Diagnostic). The measurement range extended from 0.05 to 370 ng/mL, with a between-day precision coefficient of variation <4.4% and biological reference interval <16.3 ng/mL. All blood samples with visible hemolysis were rejected from the analysis to avoid falsely elevated values.

Once patients reached normothermia, and sedation and muscular relaxation were withdrawn, they underwent clinical neurological examination, followed by an electroencephalogram (EEG) if there was no documented response to simple commands and also taking into account BIS monitoring. Additionally, head computed tomography was performed when required by treating physicians. One week after hospitalization, if patients remained in a coma, a second EEG was performed. At this time, based on neurological examinations, biochemical markers as well as EEGs and neuroimaging findings, a multidisciplinary team established WLST, always in agreement with the patient's representatives.

The primary outcome of the study was the best neurological functional status achieved 3 months after CA determined according to the Pittsburgh Cerebral Performance Category (CPC) score and characterized for the purpose of the present analysis as good (CPC 1-2) or poor (CPC 3-5). A CPC score of 3 to 5 is equivalent to severe disability, coma, or death. Neurological status information was assessed retrospectively based on routine postcardiac arrest medical follow-up or by telephone interviews with patients and/or close family members.

The present study describes the kinetics of serum NSE in the first few days after CA in patients treated with TTM and analyzes their correlation with neurological outcome. Despite the simplicity and accessibility of this biomarker as an additional tool for neurological prognostication and the absence of an effect of sedative agents and TTM on its values or its relative low cost, NSE is still underused in routine clinical practice.13 This situation probably reflects both technical limitations and the inconclusive results of previous studies, explained by their small sample sizes and the lack of standardization in measuring techniques.14,15 It follows that the NSE threshold for prediction of poor neurological outcome with 0% FPR in TTM-treated patients varies in the literature between 25 ng/mL and 151.5 ng/mL at 48hours, and between 57.2 ng/mL and 78.9 ng/mL at 72hours after ROSC4. In our cohort, although serum NSE values after hospital admission were significantly higher in the CPC 3-5 group than in the CPC 1-2 group, there was still wide variability, especially in the poor neurological outcome group showing, at least in part, the challenge of attempting to generalize a cutoff value of this biomarker with high specificity and narrow confidence intervals (Figure 2). In line with findings of other studies,6,16 the best strategy for the use of isolated serum NSE values in our study was at 48 and 72hours after ROSC with a ROC-AUC of 0.85 and 0.88. In contrast, the predictive ability of NSE at admission and 24hours after ROSC was significantly lower (ROC-AUC of 0.62 and 0.73, respectively).

Figure 2 also shows the ascending temporal trend of NSE values in the CPC 3-5 group compared with the descending trend in the CPC 1-2 group. Most recent postcardiac arrest care guidelines4,17 state that the limited information available on NSE kinetics in the first few days after CA remains a knowledge gap, and therefore a barrier to establishing its usefulness as a prognostic biomarker. Since NSE determination has been part of our postcardiac arrest protocol for 9 years and has been quantified daily from admission to 72hours after ROSC, the strength of our study lies in the description of NSE kinetics in a large cohort of CA survivors treated with TTM and the assessment of their correlation with the best CPC score at 3 months. In addition, all blood samples from the cohort were analyzed at the same laboratory to reduce variation between different analytical methods. Because the estimated half-life of NSE in serum is approximately 24hours18 an increasing trend in serum NSE values could be explained by ongoing hypoxic brain damage and continuous release from dying neuronal cells in patients with poor neurological outcome.19

In our cohort, serum NSE changes between 72hours after ROSC and hospital admission showed the highest ROC-AUC and the cutoff value of+55 ng/mL represented 100% specificity for poor neurological outcome. In addition, delta serum NSE values were also integrated in a multivariable model in which every 10 ng/mL NSE increase between day 3 and hospital admission showed an OR of 1.97 (95%CI, 1.43-2.72) for poor neurological outcome. Although we did not include variables such as somatosensory evoked potentials, clinical examination or electroencephalogram recordings in our predictive model, the ROC-AUC of the model (0.94; 95%CI, 0.89-0.97) was higher than that of other models proposed in the literature using the recommended multimodal approach. For example, a proposed predictive model by Oddo and Rossetti20 based on a prospective cohort of 134 patients that included clinical examination, EEG, isolated NSE values, and somatosensory evoked potentials showed a ROC-AUC of 0.88 (95%CI, 0.82-0.93) for poor neurological outcome (CPC 3-5) at 3 months of follow-up. This similar prognostic accuracy compared with the isolated NSE values at 72hours after ROSC and delta NSE values between 72hours after ROSC and hospital admission in our study strengthens the position of NSE as a prognostic biomarker of neurological outcome in CA survivors. However, NSE is not specific to neuronal damage and can be produced by noncentral nervous system sources (hemolysis, neuroendocrine tumors, myenteric plexus, muscle, and adipose tissue breakdown)17 and its levels could be influenced by age, sex, and body mass index.21 For these reasons, in line with current guidelines and recent studies,6,19,22 we recommend including serum NSE kinetics as part of a multimodal approach based on the integration of different methods to increase its predictive value to identify patients with unfavorable neurological outcome.

The present study has several limitations due to its retrospective nature. First, a substantial number of patients were excluded from the analysis either because of a lack of NSE samples during ACCU admission or because their blood samples showed hemolysis. The number of patients excluded was greater due to the absence of NSE determination at hospital admission or at 72hours after ROSC, probably given the difficulty of the requirement of this biomarker in the context of an admission for CA and due to a selection bias in view of their worse CA characteristics and neurological outcome. Secondly, serum NSE values were mostly available at the time of neurological prognostication, probably influencing decision-making on WLST. Thirdly, we were not able to include in our predictive model variables such as electroencephalogram recordings, somatosensory evoked potentials, clinical examination, or neuroimaging findings. However, we feel our results represent the current daily practice of a tertiary referral center in postcardiac arrest care that included a substantial cohort of CA survivors in comparison with other studies. A cutoff value of delta NSE+55 ng/mL between 72hours after ROSC and hospital admission for poor neurological outcome should be evaluated and validated in an external prospective cohort before its implementation in current practice.

The present study shows that delta NSE value between 72hours after ROSC and hospital admission has a high predictive value for poor neurological outcome. Our results support the idea that serial NSE measurements rather than isolated values should be included in a multimodal postcardiac arrest neurological prognostication strategy. In our cohort, a delta NSE cutoff value of+55 ng/mL between 72hours after ROSC and hospital admission showed 100% specificity for poor neurological outcome at 3 months of follow-up after CA.

Fellowship for Training and Research in Acute Cardiovascular Care granted by the Working Group for Ischemic Heart Disease and Acute Cardiovascular Care of the Spanish Society Of Cardiology.

None declared.

.