# Differentiating Late Awakeners from Non-Awakeners in Comatose Cardiac Arrest Survivors: Diagnostic Value of Multimodal Monitoring in Patients with Indeterminate Prognosis

**Authors:** Hyo Joon Kim, Sang Hoon Oh, Jee Yong Lim

PMC · DOI: 10.3390/diagnostics16040558 · 2026-02-13

## TL;DR

This study identifies markers to distinguish patients who later awaken after cardiac arrest from those with poor outcomes, to avoid premature withdrawal of life support.

## Contribution

A multimodal model combining NSE levels and brainstem reflexes improves identification of late awakeners with good outcomes.

## Key findings

- 9.4% of comatose patients at 72 hours became late awakeners with favorable outcomes.
- Preserved corneal reflexes and NSE < 90 ng/mL predicted late awakening with high specificity.
- Multimodal model achieved an optimism-corrected AUC of 0.896 for predicting recovery.

## Abstract

Background: Current guidelines recommend prognostication at 72 h after cardiac arrest, yet a subset of patients (Late Awakeners) recover consciousness after this window. This study investigated diagnostic markers to distinguish Late Awakeners from those with permanent poor outcomes (Non-Awakeners) to prevent premature withdrawal of life-sustaining therapy. Methods: We analyzed adult OHCA patients treated with TTM from 2009 to 2019 who remained comatose (Glasgow Coma Scale Motor score < 6) at 72 h. Patients were categorized as Late Awakeners (obeyed commands > 72 h) or Non-Awakeners. The diagnostic performance of maximal Neuron-Specific Enolase (NSE) levels within 72 h and brainstem reflexes was assessed using receiver operating characteristic (ROC) analysis. Model calibration was evaluated using the Hosmer–Lemeshow test, and internal validation was performed using bootstrap resampling. Results: Of 213 patients comatose at 72 h, 20 (9.4%) were identified as Late Awakeners. The median time to awakening was 4.4 days (IQR 3.4–8.3) from ROSC. Late Awakeners exhibited significantly preserved corneal reflexes (85.0% vs. 20.2%) compared to Non-Awakeners. The optimal NSE cut-off value to predict late awakening was <89.5 ng/mL (Sensitivity 95.0%, Specificity 50.3%, AUC 0.801). A multimodal approach combining NSE < 90 ng/mL and preserved corneal reflexes achieved a high specificity of 93.2% and an AUC of 0.899 (optimism-corrected: 0.896) for predicting late recovery. At six-month follow-up, 74.3% of Late Awakeners achieved good neurological outcome (CPC 1–2). Conclusions: Approximately 9% of patients comatose at 72 h eventually regain consciousness with favorable long-term outcomes. A multimodal diagnostic model combining intermediate NSE thresholds and preserved brainstem reflexes can effectively identify these Late Awakeners, suggesting that observation should be extended for patients fitting this profile.

## Linked entities

- **Diseases:** cardiac arrest (MONDO:0000745)

## Full-text entities

- **Genes:** ENO2 (enolase 2) [NCBI Gene 2026] {aka HEL-S-279, NSE}
- **Diseases:** Neuromuscular blockade (MESH:D020879), Comatose (MESH:D003128), disability (MESH:D009069), HIBI (MESH:D020925), neuronal destruction (MESH:D008105), dysfunction (MESH:D006331), TTM (MESH:D000377), neuronal injury (MESH:D009410), PCAS (MESH:D000080942), chest compressions (MESH:D013898), COVID-19 (MESH:D000086382), CPC 5 (OMIM:614066), hypertension (MESH:D006973), brain injury (MESH:D001930), death (MESH:D003643), reperfusion injury (MESH:D015427), hypothermia (MESH:D007035), hemolysis (MESH:D006461), fever (MESH:D005334), seizure (MESH:D012640), AKI (MESH:D058186), organ dysfunction (MESH:D009102), Anoxic (MESH:D002534), respiratory failure (MESH:D012131), cerebral edema (MESH:D001929), chronic kidney disease (MESH:D051436), edema (MESH:D004487), Cardiac Arrest (MESH:D006323), ischemic (MESH:D002545), diabetes (MESH:D003920), CPC (MESH:D002547), injury to (MESH:D014947), OHCA (MESH:D058687), hepatic impairment (MESH:D008107)
- **Chemicals:** creatinine (MESH:D003404), TTM (MESH:C048192), rocuronium (MESH:D000077123), vecuronium (MESH:D014673), remifentanil (MESH:D000077208), Propofol (MESH:D015742), fentanyl (MESH:D005283), midazolam (MESH:D008874)
- **Species:** Homo sapiens (human, species) [taxon 9606]

## Figures

3 figures with captions in the complete paper: https://tomesphere.com/paper/PMC12939699/full.md

---
Source: https://tomesphere.com/paper/PMC12939699