The role of neuron-specific enolase as a predictor of neurological recovery and functional status in a selected cardiac intensive care unit population
M Urpina Matias, R Barbosa, M I Soares, R Montalvao, J Machado, M Presume, A R Bello, M Sousa Paiva, J Presume, J Ferreira, C BrizidoAbstract
Background
Neuron-specific enolase (NSE) is a biomarker of neuronal injury with strong prognostic value for neurologic outcome after sudden cardiac arrest (CA). Recommended cutoffs in the 2025 European Resuscitation Council (ERC) Guidelines derive from several studies, but a universal threshold is lacking.
Purpose
To evaluate NSE performance as a prognostic predictor in CA intensive care unit (CICU) cohort; to define optimal NSE cutoffs, comparing with guideline-recommended values.
Methods
Retrospective analysis of consecutive CICU patients admitted after in-hospital or out-of-hospital CA (IHCA or OHCA), between 2015-2025. Comatose patients (GCS<8) with multimodal neuroprognostication and NSE at 24, 48, and 72 hours after ROSC were included. Neuroprognosis followed ERC criteria. Neurological status at 30 days was classified according to the Cerebral Performance Category (CPC) scale, as good (CPC 1-2) or poor (CPC 3–5), and disability by modified Rankin Scale (mRS). ROC curves identified optimal NSE cutoffs for CPC 3-5 and mRS≥4 and estimated positive predictive value (PPV).
Results
94 patients (mean age 62 ± 17 y, 66% male) were included, 32% had ICHA and 68% OHCA. Mean no-flow time was 1.8, median low-flow time was 15 [8.5-18.8], and time to ROSC 25 [16-40] min. A shockable rhythm was present in 65%, 70% underwent coronary angiography, and 6% had a non-cardiac cause for CA. Among 90 patients completing neuroprognostication, a third had poor prognosis, 37% were indeterminate, and 7% were brain dead. At 30 days, 71% had CPC 3-5, and 68% significant functional dependence (mRS≥4). Mortality at 30 days was 62%. Median NSE values at 24, 48 and 72 hours were 49 [32-91], 45 [25-122] and 41 [23-101] ng/mL, respectively. NSE values were significantly higher in patients with very likely poor neuroprognosis; and in CPC 3-5 (122 vs 39, p=0.0005, 162 vs 32, p <0.0001 and 145 vs 28, p = 0.0005; 58 vs 36, p=0.007, 46 vs, p =0.0001 and 45 vs 22, p = 0.005; respectively). NSE at 48h showed highest accuracy for predicting poor CPC and mRS≥4 (AUC 0.89, p<0.001). The optimal 48h cutoff of 25 ng/mL (Youden index) yielded 72–75% specificity, 89% sensitivity and 91–92% PPV. Higher 48h NSE values further increased PPV, with 48.5 ng/mL providing 100% PPV and specificity but 54% sensitivity. NSE at 24 and 72 h also performed well (AUC>0.7, p<0.05). A 48 h level >60 ng/mL, as currently recommended, had 100% PPV for CPC 3–5 and mRS≥4.
Conclusion(s)
In this CICU CA cohort, NSE showed good prognostic performance at all recommended time points, with maximal accuracy at 48 h. A cutoff of 48.5ng/mL at 48 hours was the most accurate for predicting poor CPC and mRS ≥4 at 30 days, slightly lower than guideline-recommended values, suggesting population-dependent cutoffs and reinforcing the importance of multimodal neuroprognostication.For image description, please refer to the figure legend and surrounding text.