Beyond blood-brain barrier disruption and molecular weight: compartmental kinetics of S100B and NSE for neurological prognostication after cardiac arrest

Critical Care volume 29, Article number: 342, Published: 04 August 2025

 Background

The prognostic value of serum biomarkers after out-of-hospital cardiac arrest (OHCA) depends on timing, but the physiological basis remains unclear. We investigated whether blood–brain barrier (BBB) integrity and biomarker-specific properties explain the time-dependent differences in prognostic performance.

Methods

This retrospective study included comatose adult OHCA survivors who underwent paired serum and cerebrospinal fluid (CSF) measurements of neuron-specific enolase (NSE; 47 kDa) and S100 calcium-binding protein B (S100B; 21 kDa) at 0 (H0), 24 (H24), 48 (H48), and 72 (H72) h after return of spontaneous circulation. BBB disruption was assessed using the CSF/serum albumin quotient (Q_A). Prognostic performance was assessed using AUC analysis for 6-month poor neurological outcome (Cerebral Performance Category 3–5).

Results

Among 111 patients (59% poor outcome), 646 serum and 620 CSF samples were analyzed. BBB disruption was more severe in the poor outcome group at all timepoints (all P < 0.001), peaking at H24 (Q_A 0.0282 [IQR 0.0150–0.120]) and remaining elevated at H72 (0.0228 [IQR 0.0147–0.0598]). In the poor outcome group, serum S100B levels peaked at H0 (0.80 ng/mL [IQR 0.39–2.81]) and declined despite a persistent elevation in CSF levels at or above the upper detection limit (≥ 30 ng/mL). Conversely, NSE levels progressively increased in both compartments, with serum and CSF levels increasing in parallel over time. Serum NSE concentrations showed a time-dependent improvement in prognostic accuracy, peaking at H72 (AUC 0.88), whereas S100B concentrations maintained stable performance across all timepoints (AUCs 0.79–0.85, all P > 0.4). Notably, the prognostic performance of S100B remained relatively consistent regardless of BBB disruption severity, whereas NSE showed progressively improved predictive accuracy with increased BBB disruption. Across all timepoints, CSF biomarkers—particularly S100B and NSE—showed consistently higher AUCs than serum, suggesting superior prognostic utility.

Conclusions

Serum NSE levels closely reflect the degree of BBB disruption and CSF levels, while S100B exhibits a transient early-phase profile, with decreased serum detectability over time, even in the presence of sustained CSF elevation or severe BBB disruption. These findings highlight the importance of interpreting biomarker kinetics across compartments and timepoints rather than relying on molecular weight or BBB status alone.