Out-of-hospital cardiac arrest (OHCA) is defined as the cessation of cardiac mechanical activity as confirmed by the absence of signs of circulation occurring outside the hospital (1). Coronary artery disease with initial ventricular fibrillation and subsequent degeneration into asystole is a leading cause of non-traumatic OHCA (2, 3). OHCA remains a major health problem, as it frequently results in patient death or severe neurological disability, even with early resuscitation efforts and high-quality intensive care (4–7). Prolonged care for unconscious patients with severe brain injury and little hope of a meaningful life is costly and burdensome both for relatives and medical staff. Early discontinuation of intensive treatment should, therefore, be considered in selected cases with a grave neurological prognosis. However, an early determination of the patient's prognosis is difficult and often biased by unavoidable patient sedation (8).

To improve early predictions of neurological outcomes after an OHCA, studies have used various serum biomarkers of neuronal injury (9). Advantages of these biomarkers over other prognostic modalities and clinical examination include quantitative results and independence from the effects of sedatives. Their disadvantage is the lack of a consistent threshold for identifying patients destined to a poor outcome with a high degree of certainty (10). Astroglial S100 calcium-binding protein has been shown to correlate with the extent of nerve tissue damage and neurological outcomes (10). The prognostic value of S100 has been investigated in several studies; however, most of them suffered from significant limitations: (1) they included mixed populations of OHCA and in-hospital cardiac arrest patients (11–16); (2) targeted temperature management (TTM), which is the currently recommended therapy with demonstrated neuroprotective effects, was not routinely used (16–26); (3) they included all-cause mortality as a part of a combined clinical endpoint (12, 26, 27), or the causes of death were not analyzed (13–15, 21–25, 28–30), which imposed a significant bias since S100 levels reflect brain injury/death but not death from other causes.

To avoid the above-mentioned limitations, we conducted this prospective study to verify the validity of baseline and 48 h serum S100 levels as a predictor of the best neurological outcome during the 30 day period after a non-traumatic OHCA.

Of the 48 patients, a total of 30 (63%) had a favorable neurological outcome (CPC 1–2), while 18 (38%) had an unfavorable neurological outcome (CPC 3–4) 30 days after the OHCA. Eleven patients (23%) died, of those, eight died from irreversible brain damage. A comparison of patient clinical characteristics, based on the best achieved 30 day neurological outcome, is shown in Table 1.

In the univariate analysis, an unfavorable neurological outcome was associated with higher 48 h S100 levels, higher initial left ventricular ejection fraction, absence of a shockable rhythm at first medical contact, the presence of ST-segment depressions on the first electrocardiogram, and the need for vasopressors during the 24 h following the OHCA. On the other hand, there was no significant association between neurological outcomes and baseline S100 levels or between the absolute decrease in S100 between baseline and 48 h (Figure 2). Furthermore, there was no relationship between neurological outcomes and the time delay from OHCA to the start of TTM or to the achievement of the target temperature.

In the multivariate analysis, increased S100 levels at 48 h were the only independent predictors of an unfavorable neurological outcome up to 30 days. The area under the receiver operating characteristic curve for S100 levels at 48 h, used for prediction of neurological outcomes, was 0.85 (CI 0.74–0.96) (Figure 3). At 48 h, S100 levels above the cut-off value of ≥0.37 μg/L had 100% specificity and 39% sensitivity for predicting an unfavorable 30 day neurological outcome (Table 2), while a cut-off value of 0.12 μg/L, determined as the Youden index, provided an unacceptably high false-positive rate for unfavorable outcomes (17%).

Table 3 provides univariate analysis of clinical characteristics relative to 48 h S100 cut-off value of 0.37 μg/L. Current tobacco smoking was found to be the only significant predictor of lower S100 levels. Importantly, the S100 values were influenced by neither the resuscitation-related characteristics nor the cardiac function.

In a homogeneous population of OHCA victims undergoing TTM, this study found a significant association between 30 day neurological outcomes and S100 levels measured 48 h after the OHCA. However, the baseline values of S100 had no relationship to outcome. In fact, the 48 h S100 was the only independent predictor of neurological outcome. Using a cut-off value of ≥0.37 μg/L, S100 could predict an unfavorable outcome in 39% of patients with a 100% specificity. These findings suggest that the biomarker could be helpful for neurological prognostication, especially in patients who are still unconscious 48 h after their OHCA.

Dealing with comatose patients after resuscitation from an OHCA is notoriously difficult. Importantly, it is essential to minimize the risk of a falsely pessimistic prediction. When predicting poor outcomes, the false positive rate should be zero. Advantages of biomarkers over other prognostic modalities include quantitative results and likely independence from the effects of sedatives (10). In the central nervous system, the S100 family of proteins is present in glial and Schwann cells, where they stimulate cellular processes such as proliferation, differentiation, and regulation of intracellular Ca²⁺ homeostasis (38, 39). Extra-cerebrally, S100 is found in muscle, chondrocytes (40), and adipocytes (41). The latter sources may be confounding in OHCA patients who received chest compressions since they can lead to high baseline S100 levels. This increase was independent of the neurological outcomes in our patients.

The association between 48 h S100 serum levels and neurological outcomes in our cohort corresponds to previous observational trials, in which S100 measured at various time points after an OHCA were predictive of the neurological outcome (11–17, 19, 20, 24, 25, 27–30); nonetheless, only a few studies showed a significant association when the multivariate analysis was applied (23, 24, 26, 30). All the published evidence, to date, has significant shortcomings, including heterogeneous populations (11–16), or lack of routine TTM after OHCA, which is the current standard of care (16–26).

Previously, a biobank sub-study of the “TTM trial” analyzed serum S100 values of 687 patients after an OHCA (27, 35). After adjusting for relevant clinical factors, the authors found no independent predictive value of S100 measured at 24, 48, or 72 h. The sub-study compared S100 levels with a definitive 6 month CPC score, including all-cause mortality. However, this is a weak point of the observation, since by default, S100 levels cannot be used to predict all-cause mortality. To overcome this bias in analysis, our study considered the best CPC score, both during hospitalization and post-hospitalization, starting on day 3 and continuing through day 30. For instance, two of our patients, who had neurologically recovered during the first days (both were CPC 1), subsequently died from cardiac-related death; however, both were treated in the analysis as being CPC 1 (Figure 1). On the other hand, none of our patients who achieved a CPC score of 1 or 2 died during the 30 day study due to brain injury. We believe that our clinically more relevant and straightforward definition of the neurological endpoint was the main reason why our results differ from the “TTM trial” sub-study.

Our study has limitations. The required study population size was calculated based on previously published data. Although the study showed a significant association between high S100 levels and unfavorable neurological outcomes, the relatively modest sample size translates into wider CIs in the statistical analyses. The patient population was biased toward cardiac causes of OHCA since patients with obvious extra-cardiac causes of OHCA were predominantly transported to other intensive care units. On the other hand, this referral bias could be considered as an advantage since our study population was very homogeneous. Therefore, the study results are better applicable to CICU clinical practice. Finally, our results may be influenced by the automated method of electrochemiluminescence immunoassay using Elecsys S100® kits, which is more sensitive and has a faster turnaround time compared to kits of some other manufacturers (42).

This study showed that S100 values examined 48 h after OHCA can independently estimate 30 day neurological outcomes in a population of patients treated with TTM. Most importantly, this predictive value can be achieved without interference with TTM-associated sedation. Thus, the biomarker S100 may be particularly helpful in patients who remain comatose after completion of TTM. However, before making any critical clinical decisions, a neurological prognosis in these patients should be assessed by a multimodal approach, which should be based on repeated neurological assessments, and may include imaging, somatosensory evoked potentials, electroencephalography, and quantitative pupillometry (10, 43). In this regard, evaluation of S100 levels at 48 h would be a useful tool, though it is only one piece in the clinical mosaic.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of the Institute for Clinical and Experimental Medicine and Thomayer Hospital, Prague. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

MK: formal analysis, data curation, writing—original draft, and visualization. MS: data curation, writing—review and editing, and visualization. JKo: investigation. JKa: writing—review and editing. JKe: conceptualization, methodology, resources, supervision, and project administration. All authors contributed to the article and approved the submitted version.

JKa has received speaker's honoraria from Boehringer Ingelheim, Biosense Webster, Biotronik, Daiichi Sankyo, Medtronic, Merck Sharp & Dohme, Pfizer, and Abbott - St. Jude Medical; and has served as a consultant for Bayer, Boehringer Ingelheim, Biosense Webster, Daiichi Sankyo, Medtronic, Merit Medical, and Abbott - St. Jude Medical. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.