Despite several systematic reviews, meta-analysis of studies on NSE after cardiac arrest has been challenging due to, a) variable cutoffs reported in each study and b) multiple cut-offs within the same study for each time point.

Data from a 2014 systematic review (ERC 2014) were updated with literature searches from Pubmed, Cochrane and Scopus until May 2020. Cohort studies on comatose (GCS < 8) survivors aged > 16 y, which defined poor outcome as CPC 3-5 (or equivalent) were included. After grading study quality based on the AAN guideline process manual, contingency tables were extracted and combined with the ERC 2014 data. Using the multiple thresholds meta-analysis model, summary receiver-operating curves (SROC) were generated for NSE levels at time points <24, 48 and 72 h post resuscitation. Area-under-curves (AUCs) [95% CI] were calculated as well as cut-off values [95% CI] associated with false positive rate (FPR) < 5%.

Evidence from a total of 52 studies was reviewed; 11 studies (n = 2153) at 24 h, 22 studies (n = 3012) at 48 h, and 13 studies (n = 2605) at 72 h. Quality of evidence was low or very low for most studies due to the risk of incorporation bias. AUCs for prediction of poor outcomes were 0.64 [0.52-0.74]; 0.82 [0.76-0.87] and 0.83 [0.77-0.88] respectively. For prediction of poor outcome with FPR < 5%, NSE cut-offs (μg/L) were 50.1 [30.5-70.4] at 48 h, and 47.1 [35.8-58.6] at 72 h.

NSE had the best prognostic performance at 48-72 h. Levels above 70.4 μg/L at 48 h and 58.6 μg/L at 72 h are highly predictive of poor outcome with FPR < 5%.