Observational Study Medicine ®

Development of a strategy for assessing blood-

brain barrier disruption using serum S100
calcium-binding protein B and neuron-specific
Downloaded from http://journals.lww.com/md-journal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy

enolase in early stage of neuroemergencies

wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 05/25/2024

A preliminary study
Gi Su Yun, MDa, Yong Nam In, MD, PhDb,c,*, Changshin Kang, MD, PhDa,* , Jung Soo Park, MD, PhDa,b,
Yeonho You, MD, PhDa, Jin Hong Min, MD, PhDb,c, Hong Joon Ahn, MD, PhDa,b, Insool Yoo, MD, PhDa,b,
Seung Whan Kim, MD, PhDa,b, Se Kwang Oh, MD, PhDb,c, In Ho Lee, MD, PhDd, Da Mi Kim, MD, PhDd

Abstract
Background: Rapid disease progression in neuroemergencies is associated with blood-brain barrier (BBB) disruption. We
investigated a less invasive strategy for assessing BBB status by evaluating S100 calcium-binding protein B (S100B) and neuron-
specific enolase (NSE) at early stages of the hypoxic-ischemic brain injury (HIBI) cascade.
Methods: This retrospective study used prospectively collected data from patients with out-of-hospital cardiac arrest (August
2019–July 2021). Albumin specimens obtained from serum and cerebrospinal fluid via arterial catheter and lumbar puncture were
used to measure the albumin quotient (Qa), which is widely accepted as the gold standard method for detecting BBB disruption.
Serum S100B and NSE levels were measured simultaneously following the return of spontaneous circulation. We conducted linear
regression to evaluate the relationship between S100B and Qa and the predictive performance of S100B for abnormal Qa. The
primary study outcome was abnormal Qa (>0.007).
Results: Forty-one patients were enrolled; 30 showed an abnormal Qa suggestive of BBB disruption. S100B levels were
significantly higher than in those with a normal Qa (0.244 μg/L [interquartile range [IQR], 0.146–0.823 vs 0.754 μg/L [IQR, 0.317–
2.228], P = .03). We report a positive correlation between serum S100B and Qa (R2 = 0.110; P = .04). The area under the receiver
operating characteristics curve (AUROC) evaluating the predictive performance of S100B with respect to abnormal Qa was 0.718
(95% confidence interval, 0.556–0.847). The cutoff value for S100B (with respect to BBB disruption) in the total cohort was 0.283
μg/L (sensitivity, 80.0%; specificity, 72.7%). Subgroup analyses in patients with serum neuron-specific enolase (NSE) levels of
<40.8 ng/mL (excluding those with established neuronal cell injury) showed an improved correlation coefficient (R2 = 0.382; P <
.01) and predictive performance (AUROC, 0.836 [95% confidence interval, 0.629–0.954]) compared with the total cohort.
Conclusions: Serum S100B obtained at an early stage of the HIBI cascade is associated with abnormal Qa, suggesting BBB disruption.
The predictive performance of S100B and the correlation between serum S100B and Qa can be improved using a complementary
strategy (i.e., evaluations of S100B and NSE levels) that combines considerations of cell damage in astrocytes and neurons.
Abbreviations: AUROC = area under the receiver operating characteristics curve, BBB = blood-brain barrier, CPR =
cardiopulmonary resuscitation, CSF = cerebrospinal fluid, ECLIA = electrochemiluminescence immunoassay, HIBI = hypoxic-
ischemic brain injury, OHCA = out-of-hospital cardiac arrest, Qa = albumin quotient, ROC = receiver operating characteristic,
S100B = S100 calcium-binding protein B, SPCC = Spearman’s rank correlation coefficient.
Keywords: biomarker, blood-brain barrier, out-of-hospital cardiac arrest

This work was supported by research fund of Chungnam National University *Correspondences: Yong Nam In, MD, PhD, Department of Emergency Medicine,
Hospital (2021) and the Korean Society of Emergency Medicine (2022). College of Medicine, Chungnam National University, 266 Munwha-ro, Jung-gu,
The authors have no conflicts of interest to disclose. Daejeon 35015, Republic of Korea (e-mail: ynsoft@naver.com).

The datasets generated during and/or analyzed during the course of conducting
the current study are not publicly available due to privacy concerns and
institutional policy.
a
Department of Emergency Medicine, Chungnam National University Hospital,
Jung-gu, Daejeon, Republic of Korea, b Department of Emergency Medicine,
College of Medicine, Chungnam National University, Jung-gu, Daejeon, Republic
of Korea, c Department of Emergency Medicine, Chungnam National University
Sejong Hospital, Sejong, Republic of Korea, and d Department of Radiology,
College of Medicine, Chungnam National University, Jung-gu, Daejeon, Republic
of Korea.
Copyright © 2022 the Author(s). Published by Wolters Kluwer Health, Inc.
This is an open-access article distributed under the terms of the Creative Commons
Attribution-Non Commercial License 4.0 (CCBY-NC), where it is permissible to
download, share, remix, transform, and buildup the work provided it is properly
cited. The work cannot be used commercially without permission from the journal.
How to cite this article: Yun GS, In YN, Kang C, Park JS, You Y, Min JH, Ahn HJ,
Yoo I, Kim SW, Oh SK, Lee IH, Kim DM. Development of a strategy for assessing
blood-brain barrier disruption using serum S100 calcium-binding protein B and
neuron-specific enolase in early stage of neuroemergencies: a preliminary study.
Medicine 2022;101:28(e29644).

*Correspondences: Changshin Kang, MD, PhD, Department of Emergency Received: 27 January 2022 / Received in final form: 24 April 2022 / Accepted:
Medicine, Chungnam National University Hospital 282, Munhwa-ro, Jung-gu, 9 May 2022
Daejeon 35015, Republic of Korea (e-mail: changsiny@naver.com). http://dx.doi.org/10.1097/MD.0000000000029644

1
 Yun et al. • Medicine (2022) 101:28Medicine
1. Introduction
The blood-brain barrier (BBB) is a dynamic platform for
exchanging substances between the extracellular or intersti-
tial fluid and the brain parenchyma, allowing for selective
molecular transport across the endothelial cells into the brain
parenchyma.[1,2] Anatomically, the BBB is a cellular assembly
constituted by the close functional association of several cen-
tral nervous system (CNS) cell types, including endothelial cells,
Downloaded from http://journals.lww.com/md-journal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
pericytes, and astrocytes.[3] This structure contributes to the
maintenance of brain homeostasis in consideration of metabolic
wastes and extrinsic environmental factors.[4]
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 05/25/2024
It is widely acknowledged that BBB dysfunction can not
only be a cause of neurogenic disease but can also accelerate
disease progression.[5,6] Accordingly, it is also recognized that
BBB dysfunction is a hallmark of neurovascular pathologies.[4]
Calculating the cerebrospinal fluid (CSF) to serum albumin quo-
tient (Qa) is commonly accepted as the gold standard method
for assessing BBB status.[7–9] However, due to the limitations
involved in the routine application of Qa in the clinical field (i.e.,
invasive spinal tap or external ventricular drainage), alternative
markers reflecting BBB status have been suggested in several
prior studies.[10–13] More specifically, S100 calcium-binding pro-
tein B (S100B) has been suggested as an alternative biomarker
for assessing BBB status as its level statistically significantly cor-
relates with BBB integrity.[14,15] A previous study on the diag-
nostic performance of S100B for assessing BBB status using
timed serial samples (12, 24, and 48 hours after traumatic brain
injury) demonstrated that S100B obtained at 12 hours showed
the best indication of BBB dysfunction.[15] This may be due to
the rapid elimination of S100B from the blood, with a relatively
short half-life (between 0.5 and 2 hours).[14] In addition, more
careful consideration needs to be given to assessing BBB status
using serum biomarkers due to the confounding effect of sever-
ity or duration of cerebral injury.[16]
Hence, we hypothesized that excluding a confounding effect
from already established neuronal injury, which can be assessed
using the biomarker of neuron-specific enolase (NSE), may
improve predictive performance for assessing BBB status using
serum S100B. Therefore, we aimed to investigate the relationship
between serum S100B levels and BBB status within 12 hours of
cerebral injury, which more closely encompasses the proposed
half-life of S100B compared with previous work. Furthermore, we
developed a strategy to assess BBB status using combined consider-
ation of 2 biomarkers reflecting astrocyte and neuronal cell injury.
2. Materials and Methods
2.1. Study design and setting
We conducted a retrospective study using prospectively collected
data from a registry of patients experiencing out-of-hospital
Figure 1. Scheme for our protocol to initiate postcardiac arrest care; obtain serum and cerebrospinal fluid samples. *Brain CT was performed to investigate
whether definite severe cerebral edema or brain hemorrahge or not. CA = cardiac arrest, CSF = cerebrospinal fluid, CT = computed tomography, NSE = neu-
ron-specific enolase, OHCA = out-of-hospital cardiac arrest, ROSC = return of spontaneous circulation, S100B = S100 calcium-binding protein B.
2
cardiac arrest (OHCA) who underwent targeted temperature
management (TTM) at our medical center over the course of
24 months (August 2019–July 2021). All patients included in
our registry were managed according to our previously pub-
lished TTM protocol, which was established based on current
international guidelines (Fig. 1).[17] We enrolled adult patients
(≥18 years of age) who were treated with TTM after OHCA
and whose CSF albumin levels were measured within 12 hours
of the return of spontaneous circulation (ROSC) using lumbar
catheters. Patients lacking data on serum S100B were excluded
from this study.
The study was approved by the Institutional Review Board
(or Ethics Committee) of Chungnam National University
Hospital (No. 2021-07-047) and was conducted according to
the guidelines of the Declaration of Helsinki. The extracted data
included clinical data only; it does not include any personally
identifiable information. Therefore, the need for informed con-
sent was waived.
2.2. Serum S100B and the albumin quotient
Serum samples for evaluating S100B and albumin levels were
collected simultaneously using an arterial catheter (Fig. 1).
Serum S100B and albumin concentrations were determined
using an electrochemiluminescence immunoassay (ECLIA)
(Elecsys S100, COBAS e801, Roche Diagnostics, Rotkreuz,
Switzerland) and an automated biochemical analyzer (TBA-
2000FR; Canon Medical Systems Corporation, Otawara,
Japan), respectively. The measurement range for serum S100B
was 0.005 to 30 μg/L. CSF albumin samples were obtained
via a lumbar catheter by an expert physician using a Hermetic
lumbar catheter accessory kit (Integra Neurosciences,
Plainsboro, NJ) under aseptically guided sonography with the
patient lying in the lateral decubitus position with their hips
and knees flexed. CSF albumin samples were obtained simul-
taneously as the serum samples (within 12 hours of ROSC).
CSF albumin levels were detected using the analyzer used for
serum albumin samples.
2.3. Primary outcome and data collection
The primary outcome evaluated in this study was abnormal
Qa, defined as a Qa of >0.007.[16] The Qa was calculated using
the following formula: Qa = CSF albumin/serum albumin. The
following data were collected from the medical records data-
base: age, sex, presence of a witness at the time of the collapse,
bystander cardiopulmonary resuscitation (CPR), first moni-
tored rhythm, etiology of cardiac arrest, time from collapse to
CPR (no flow time), time from CPR to ROSC (low-flow time),
and time from ROSC to obtaining serum and lumbar catheter
samples.
 Yun et al. • Medicine (2022) 101:28www.md-journal.com
2.4. Subgroup analysis
We assumed that our entire cohort, which comprises comatose
patients with OHCA, would display a confounding effect due
to established neuronal cell injury. This may affect the perfor-
mance of serum S100B as a marker for BBB disruption. Thus,
we performed a subgroup analysis in patients with serum NSE
levels of <40.8 ng/mL measured simultaneously with S100B, as
suggested according to the lower limit of the 95% confidence
Downloaded from http://journals.lww.com/md-journal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
interval (CI) for serum NSE obtained within 24 hours following
cardiac arrest within a recent meta-analysis.[18] We performed
this subanalysis to estimate the relationship between serum
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 05/25/2024
S100B and BBB status after excluding the confounding effects
of established neuronal cell injury.
2.5. Statistical analyses
Categorical variables were presented as frequencies and per-
centages. Continuous variables were presented as medians
and interquartile ranges (IQR); all continuous variables in this
study showed a nonnormal distribution via the Shapiro-Wilk
test. Categorical and continuous variables were analyzed using
Fisher exact tests and Mann-Whitney U tests, respectively.
Simple linear regression was conducted to evaluate the relative
impact of S100B on Qa using the coefficient of determination
(R2). Receiver operating characteristic (ROC) curves were con-
structed to assess the diagnostic performance of serum S100B
with respect to reflecting BBB disruption.
ROC curves plot the sensitivity of a measure on the y-axis
and 1-sensitivity on the x-axis and thus measure a test over-
all accuracy. The most important summary index of the ROC
curve is the area under the ROC curve (AUROC). The opti-
mal cutoff value for serum S100B for predicting abnormal
Qa was the largest value based on calculating the Youden
index. Statistical analyses were performed using SPSS sta-
tistical software (IBM Corp., SPSS Statistics for Windows,
Version 25.0. Armonk, NY). AUROC curves were constructed
using MedCalc software (version 15.2.2, MedCalc Software,
Mariakerke, Belgium). Results were considered statistically
significant at a P < .05.
Figure 2. Diagram showing the patient selection process for the current study. CPC = Glasgow-Pittsburgh cerebral performance category, CT = computed
tomography, ECMO = extracorporeal membrane oxygenation, GCS = Glasgow Coma Scale, LP = lumbar puncture, OHCA = out-of-hospital cardiac arrest, Qa
= albumin quotient, ROSC = return of spontaneous circulation, S100B = S100 calcium-binding protein B, TTM = targeted temperature management.
3
3. Results
3.1. Study population characteristics
We evaluated 68 patients with hypoxic-ischemic brain injury
due to OHCA who underwent postcardiac arrest care at our ter-
tiary medical center, including TTM. Of these patients, 27 were
excluded for the following reasons: 20 patients were ineligible
for lumbar catheter placement, and 7 had no data on serum
S100B (Fig. 2). Thus, 41 patients were enrolled in this study, of
whom 30 (73.2%) showed an abnormal Qa suggestive of BBB
disruption (Fig. 2). Baseline medical and demographic charac-
teristics were not significantly different between the normal and
abnormal Qa groups (Table 1). Serum NSE levels measured at
baseline (before postcardiac arrest care) were statistically sig-
nificantly higher in the abnormal Qa group (20.8 ng/mL [IQR,
17.7–29.8] vs 37.0 ng/mL [IQR, 25.0–64.4], P = .01, Table 1).
3.2. Relationship between albumin quotient and serum
S100B in the total cohort
Serum S100B levels were statistically significantly higher in the
abnormal Qa group than in the normal Qa group (0.244 μg/L
[IQR, 0.146–0.823] vs 0.754 μg/L [IQR, 0.317–2.228], P = .03,
Fig. 3), with a statistically significant correlation between serum
S100B and Qa (R2 = 0.110, P = .04, Fig. 3). The AUROC value
of serum S100B for its predictive performance with respect to
BBB disruption was 0.718 (95% CI, 0.556–0.847, Fig. 3). The
cutoff value of serum S100B with respect to reflecting BBB dis-
ruption in the total cohort was 0.283 μg/L (sensitivity, 80.0%;
specificity, 72.7%; Fig. 3).
3.3. Subgroup analysis
Table 2 shows the demographic and medical characteristics of
the patients included in the subgroup analysis (i.e., those with
serum NSE levels of <40.8 ng/mL). A total of 24 patients were
included in the subgroup analysis; 16 (66.7%) had an abnormal
Qa. Baseline characteristics were not statistically significantly
different between groups, except for low-flow time (9.0 minutes
 Yun et al. • Medicine (2022) 101:28Medicine

Table 1
Baseline demographics and characteristics in total cohort

Total patients, Normal Qa Abnormal Qa

n = 41 n =11 n = 30 P
Age, yrs 60 (47–70) 45 (38–76) 64 (52–70) .15
Male 28 (68.3) 6 (54.5) 22 (73.3) .28
Pre-existing illnesses
Downloaded from http://journals.lww.com/md-journal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy

 CAD 6 (14.6) 0 6 (20) .16

 Hypertension 17 (41.5) 3 (27.3) 14 (46.7) .31
 Diabetes mellitus 13 (31.7) 2 (18.2) 11 (36.7) .45
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 05/25/2024

 Ischemic stroke 3 (7.3) 1 (9.1) 2 (6.7) 1.00

 Pulmonary disease 6 (14.6) 3 (27.3) 3 (10.0) .31
 Renal disease 8 (19.5) 0 8 (26.7) .08
 Neurologic disorders* 1 (2.4) 1 (9.1) 0 .27
Cardiac arrest characteristics
Witnessed 23 (56.1) 7 (63.6) 16 (53.3) .73
 Bystander CPR 28 (68.3) 10 (90.9) 18 (60.0) .13
 Shockable rhythm 13 (31.7) 3 (27.3) 10 (33.3) 1.00
 Cardiac etiology 18 (43.9) 4 (36.4) 14 (46.7) .73
 No flow time, min 1.0 (0.0–12.5) 1.0 (0.0–5.0) 3.0 (0.0–14.0) .62
 Low-flow time, min 21.0 (13.5–30.0) 15.0 (9.0–22.0) 24.5 (16.5–32.3) .07
 Time to obtain samples, hours 4.0 (3.0–6.0) 3.2 (2.5–6.0) 4.6 (3.2–6.1) .11
 Serum NSE, ng/mL 33.0 (20.2–48.8) 20.8 (17.7–29.8) 37.0 (25.0–64.4) .01
Mortality, n (%) 13 (31.7) 1 (9.1) 12 (40.0) .27
 Multiorgan failure 7 (17.1) 1 (9.1) 6 (20.0)
  WLST 5 (12.2) 0 5 (16.7)
 Brain death 1 (2.4) 0 1 (3.3)
Data are presented as n (%) or median (interquartile range).
* Neurologic disorder includes epilepsy, Alzheimer dementia, Parkinson disease, and muscular dystrophy.
CAD = coronary artery disease, CPR = cardiopulmonary resuscitation, NSE = neuron-specific enolase, Qa = albumin quotient, WLST = withdrawal of life-sustaining therapy.

Figure 3. Analysis of the relationship between Qa and serum S100B levels in the total cohort. (A) Serum S100B levels comparing normal and abnormal Qa
groups. (B) A simple linear regression graph comparing serum S100B and Qa values. (C) The predictive performance of serum S100B with respect to abnormal
Qa. AUROC = the area under the receiver operating characteristic curve, CI = confidence interval, Qa = albumin quotient, S100B = S100 calcium-binding
protein B.

[IQR, 6.8–15.0] vs 21.0 minutes [IQR, 15.0–29.0], P = .01,
Table 2). Serum S100B was statistically significantly higher in
the abnormal Qa group than in the normal Qa group (0.214
μg/L [IQR, 0.120–0.268] vs 0.324 μg/L [IQR, 0.238–0.663], P
= .01, Fig. 4), and there was a statistically significant correlation
between serum S100B and Qa (R2 = 0.382, P < .01, Fig. 4). The
AUROC value of serum S100B for its predictive performance in
BBB disruption was 0.836 (95% CI, 0.629–0.954; Fig. 4). The
cutoff value of serum S100B with respect to reflecting BBB dis-
ruption in the total cohort was 0.283 μg/L (sensitivity, 62.5%;
specificity, 100.0%; Fig. 4).
4. Discussion
The present study was designed to assess BBB disruption
according to abnormal Qa and serum S100B values obtained
at an early stage of the HIBI cascade. Our results align with
previous findings that S100B is primarily present in the CSF.[19]
In addition, we found a numerical improvement in predictive
performance for assessing BBB status by considering a combi-
nation of 2 biomarkers suggestive of cell injury in astrocytes
and neurons.
We analyzed the relationship between serum S100B and BBB
status using samples obtained during the early process of brain
injury (within 12 hours) and compared it to the findings of a
previous study evaluating BBB dysfunction (in which serum
S100B was first obtained at 12 hours).[15] Since S100B is rapidly
eliminated from the blood (within 2 hours), we assumed that
earlier sampling might lead to better predictive performance for
S100B with respect to BBB dysfunction. However, the predictive
performance in our study was numerically lower than that of
the prior investigation (AUROC, 0.718 [95% CI, 0.556–0.847]
vs 0.800 [95% CI, 0.511–1.089]).[15] We suggest that this result
can be supported by the discrepancy of cascades between trau-
matic brain injury (TBI) and HIBI after cardiac arrest. The times
of BBB disruption onset from cerebral injury, either hemorrhagic

4
 Yun et al. • Medicine (2022) 101:28www.md-journal.com

Table 2
Baseline demographics and characteristics of subgroup

Total patients, Normal Qa Abnormal Qa

n = 24 n=8 n = 16 P
Baseline demographics and characteristics
Age, yrs 57 (39–65) 42 (24–59) 61 (45–67) .10
Male 17 (70.8) 5 (62.5) 12 (75.0) .65
Downloaded from http://journals.lww.com/md-journal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy

Pre-existing illnesses
 CAD 5 (20.8) 0 5 (31.3) .14
 Hypertension 11 (45.8) 2 (25.0) 9 (56.3) .23
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 05/25/2024

 Diabetes mellitus 9 (37.5) 2 (25.0) 7 (43.8) .67

 Ischemic stroke 3 (12.5) 1 (12.5) 2 (12.5) 1.00
 Pulmonary disease 2 (8.3) 2 (25.0) 0 .10
 Renal disease 4 (16.7) 0 4 (25.0) .27
 Neurologic disorders* 1 (4.2) 1 (12.5) 0 .58
Cardiac arrest characteristics
 Witnessed 15 (62.5) 5 (62.5) 10 (62.5) 1.00
 Bystander CPR 18 (75.0) 7 (87.5) 11 (68.8) .62
 Shockable rhythm 12 (50.0) 3 (37.5) 9 (56.3) .67
 Cardiac etiology 14 (58.3) 4 (50.0) 10 (62.5) .67
 No flow time, min 1.0 (0.0–5.0) 1.0 (0.0–4.8) 1.0 (0.0–6.3) .87
 Low-flow time, min 17.0 (9.0–25.0) 9.0 (6.8–15.0) 21.0 (15.0–29.0) .01
 Time to obtain samples, hours 3.6 (3.0–6.0) 3.3 (3.0–5.6) 4.0 (2.8–6.0) .85
Mortality, n (%) 3 (12.5) 0 3 (18.8) .53
 Multiorgan failure 3 (12.5) 0 3 (18.8)
 WLST 0 0 0
 Brain death 0 0 0
Serum NSE 22.9 (18.1–34.5) 20.0 (16.6–23.3) 29.3 (18.8–36.5) .08
Data are presented as n (%) or median (interquartile range).
* Neurologic disorders include epilepsy, Alzheimer dementia, Parkinson disease, and muscular dystrophy.
CAD = coronary artery disease, CPR = cardiopulmonary resuscitation, NSE = neuron-specific enolase, Qa = albumin quotient, WLST = withdrawal of life-sustaining therapy.

Figure 4. Subgroup analysis of the relationship between serum S100B levels and Qa in patients with serum NSE levels of <40.8 ng/mL. (A) Serum S100B levels
comparing normal and abnormal Qa groups. (B) A simple linear regression graph comparing serum S100B and Qa values. (C) The predictive performance of
serum S100B with respect to abnormal Qa. AUROC = the area under the receiver operating characteristic curve, CI = confidence interval, NSE = neuron-spe-
cific enolase, Qa = albumin quotient, S100B = S100 calcium-binding protein B.

or ischemic stroke, were revealed as minimally 3 and 6 hours
from cerebral injury in TBI and HIBI models, respectively.[20,21]
Moreover, Foerch et al[22] reported that S100B levels are ele-
vated in the systemic circulation at 6 hours following stroke.
We note that S100B samples were collected within 6 hours of
cardiac arrest in 20 patients within our total cohort. In addition,
another study revealed that the variation in time between BBB
dysfunction and obtaining a serum sample for detecting S100B
levels could limit the utility of S100B as a marker for BBB sta-
tus since the rapid elimination of S100B from the blood occurs
within 2 hours.[23] Therefore, we suggest that the time to obtain
S100B measurements following brain injury and an etiological
characteristic of the injury could play a critical role in the utility
of S100B as a BBB status marker.
Our subgroup analysis was completed to exclude confound-
ing effects from established neuronal cell injury. The subgroup
consisted of subjects with serum NSE levels of <40.8 ng/mL.
This reflects findings of a previous study that suggested a certain
cutoff value of serum NSE with respect to evaluating neuro-
logical outcomes in patients with OHCA (based on the lower
limit of the 95% CI for serum NSE obtained within 24 hours of
OHCA).[18] The subgroup analysis in the current study showed
numerically improved AUROC values for serum S100B with
respect to predicting BBB disruption (AUROC, 0.718–0.836)
and a 0% false-positive ratio for predicting abnormal Qa.
Moreover, we found an approximately 3.5-fold increase in the
coefficient of determination between serum S100B and Qa (R2,
0.110–0.382).
The delay between injury/insult and irreversible neuronal cell
death can offer a therapeutic window, which provides a unique
opportunity to administer therapeutic interventions in acutely
progressed disease cascades (such as in acute ischemic stroke or
HIBI).[24–26] For this purpose, we suggest that it is necessary to dis-
tinguish the presence of BBB impairment without considerable

5
 Yun et al. • Medicine (2022) 101:28Medicine
neuronal cell injury from that of progressed neuronal cell injury
(which could suggest irreversible brain injury). Notably, S100B
showed improved diagnostic performance for BBB impairment
and agreement with Qa in the subgroup with relatively low NSE
values (reflective of neuronal cell injury). Hence, we suggest that
a complementary strategy reinforcing the strength of each bio-
marker is essential for administering effective treatment during
the novel therapeutic window according to BBB status.
This study had several limitations that should be considered in
Downloaded from http://journals.lww.com/md-journal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
the clinical setting. First, since this was a single-center retrospec-
tive study with a small number of enrolled patients, the nature
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 05/25/2024
of this study may inherently affect the selected threshold and
various statistical outcomes. Therefore, a multicenter prospective
study with large sample size is required to enhance the general-
izability of our findings in the future. Second, CSF albumin was
obtained through a lumbar catheter. The continuous diffusion of
serum molecules from the blood vessels along the CSF flow paths
and spine and into the CSF creates a rostrocaudal concentration
gradient in the lumbar to ventricular space.[27] Therefore, CSF
albumin levels may have been overestimated in our study, leading
to inaccurate designations of normal or abnormal Qa. However,
a previous study suggested that Qa calculated using CSF albu-
min was a reliable parameter for indicating BBB disruption.[28]
Third, S100B has an extracerebral source to increase serum lev-
els such as diabetes mellitus and cerebral infarction.[22,29] Thus,
its contamination issue can disturb the results. Nonetheless, we
suggest that this potential bias from the extracerebral source of
increased S100B was minimized by showing the nonsignificant
difference in past medical history between normal and abnormal
Qa groups. Finally, our inclusion criteria for subgroup analy-
sis (<40.8 ng/mL of NSE) lacked evidence regarding excluding
neuronal cell injury completely and was also only acceptable in
patients with cardiac arrest. Therefore, this criterion is less clini-
cally relevant to other brain injury cascades.
5. Conclusions
Serum S100B obtained at an early stage of the HIBI cascade is
associated with an abnormal Qa suggestive of BBB disruption.
We conclude that the predictive performance of serum S100B
with respect to accurately reflecting BBB disruption can be
improved using a complementary strategy that combines find-
ings for serum S100B and NSE. Additional multicenter studies
with larger sample sizes are required to generalize our results.
Author contribution
Conceptualization, C. Kang; methodology, J.S. Park and S.K.
Oh; software, Y. You; formal analysis, J.H. Min and H.J. Ahn;
investigation, I. Yoo and D.M. Kim; data curation, S.W. Kim and
I.H. Lee; writing—original draft preparation, G.S. Yun; writ-
ing—review and editing, G.S. Yun and Y.N. In; visualization, C.
Kang; supervision, C. Kang; funding acquisition, C. Kang and
Y.N. In. All authors have read and agreed to the published ver-
sion of the manuscript.
References
[1] Langen UH, Ayloo S, Gu C. Development and cell biology of the
blood–brain barrier. Annu Rev Cell Dev Biol. 2019;35:591–613.
[2] Daneman R, Prat A. The blood–brain barrier. Cold Spring Harb
Perspect Biol. 2015;7:a020412a020412.
[3] Segarra M, Aburto MR, Hefendehl J, et al. Neurovascular interactions
in the nervous system. Annu Rev Cell Dev Biol. 2019;35:615–35.
[4] Segarra M, Aburto MR, Acker-Palmer A. Blood–brain barrier dynamics
to maintain brain homeostasis. Trends Neurosci. 2021;44:393–405.
[5] Yang Y, Rosenberg GA. Blood–brain barrier breakdown in acute and
chronic cerebrovascular disease. Stroke. 2011;42:3323–8.
[6] Zhao M, Jiang XF, Zhang HQ, et al. Interactions between glial cells and
the blood–brain barrier and their role in Alzheimer’s disease. Ageing
Res Rev. 2021;72:101483.
[7] Andersson M, Alvarez-Cermeño J, Bernardi G, et al. Cerebrospinal
fluid in the diagnosis of multiple sclerosis: a consensus report. J Neurol
Neurosurg Psychiatry. 1994;57:897–902.
[8] Stahel PF, Morganti-Kossmann MC, Perez D, et al. Intrathecal levels of
complement-derived soluble membrane attack complex (sc5b-9) cor-
relate with blood–brain barrier dysfunction in patients with traumatic
brain injury. J Neurotrauma. 2001;18:773–81.
[9] Reiber H, Felgenhauer K. Protein transfer at the blood cerebrospinal
fluid barrier and the quantitation of the humoral immune response
within the central nervous system. Clin Chim Acta. 1987;163:319–28.
[10] Roberts DJ, Hall RI, Wang Y, et al. S100B as a biomarker of blood–
brain barrier disruption after thoracoabdominal aortic aneurysm
repair: a secondary analysis from a prospective cohort study. Can J
Anaesth. 2021;68:1756–68.
[11] Koh SX, Lee JK. S100B as a marker for brain damage and blood–
brain barrier disruption following exercise. Sports. Sports Med.
2014;44:369–85.
[12] Reinsfelt B, Ricksten SE, Zetterberg H, et al. Cerebrospinal fluid mark-
ers of brain injury, inflammation, and blood–brain barrier dysfunction
in cardiac surgery. Ann Thorac Surg. 2012;94:549–55.
[13] Kapural M, Krizanac-Bengez L, Barnett G, et al. Serum S-100beta
as a possible marker of blood–brain barrier disruption. Brain Res.
2002;940:102–4.
[14] Marchi N, Fazio V, Cucullo L, et al. Serum transthyretin monomer
as a possible marker of blood-to-csf barrier disruption. J Neurosci.
2003;23:1949–55.
[15] Blyth BJ, Farhavar A, Gee C, et al. Validation of serum mark-
ers for blood–brain barrier disruption in traumatic brain injury. J
Neurotrauma. 2009;26:1497–507.
[16] Prakash R, Carmichael ST. Blood–brain barrier breakdown and neo-
vascularization processes after stroke and traumatic brain injury. Curr
Opin Neurol. 2015;28:556–64.
[17] Kang C, In YN, Park JS, et al. Impact of low and high partial pressure
of carbon dioxide on neuron-specific enolase derived from serum and
cerebrospinal fluid in patients who underwent targeted temperature
management after out-of-hospital cardiac arrest: a retrospective study.
Resuscitation. 2020;153:79–87.
[18] Sharma K, John M, Zhang S, et al. Serum neuron-specific enolase
thresholds for predicting postcardiac arrest outcome: a systematic
review and meta-analysis. Neurology. 2022;98:e62–72.
[19] Marchi N, Rasmussen P, Kapural M, et al. Peripheral markers of brain
damage and blood–brain barrier dysfunction. Restor Neurol Neurosci.
2003;21:109–21.
[20] Doll DN, Hu H, Sun J, Lewis SE, Simpkins JW, Ren X. Mitochondrial
crisis in cerebrovascular endothelial cells opens the blood–brain barrier.
Stroke. 2015;46:1681–9.
[21] Cash A, Theus MH. Mechanisms of blood–brain barrier dysfunction in
traumatic brain injury. Int J Mol Sci 2020;21:3344.
[22] Foerch C, Wunderlich MT, Dvorak F, et al. Elevated serum S100B levels
indicate a higher risk of hemorrhagic transformation after thrombo-
lytic therapy in acute stroke. Stroke. 2007;38:2491–5.
[23] Townend W, Dibble C, Abid K, et al. Rapid elimination of pro-
tein S-100B from serum after minor head trauma. J Neurotrauma.
2006;23:149–55.
[24] Kadry H, Noorani B, Cucullo L. A blood–brain barrier overview on
structure, function, impairment, and biomarkers of integrity. Fluids
Barriers CNS. 2020;17:69.
[25] Mossakowski MJ, Lossinsky AS, Pluta R, et al. Abnormalities of the
blood–brain barrier in global cerebral ischemia in rats due to experi-
mental cardiac arrest. Acta Neurochir Suppl (Wien). 1994;60:274–6.
[26] Pluta R, Lossinsky AS, Wiśniewski HM, et al. Early blood–brain bar-
rier changes in the rat following transient complete cerebral ischemia
induced by cardiac arrest. Brain Res. 1994;633:41–52.
[27] Reiber H. Flow rate of cerebrospinal fluid (CSF)—A concept common
to normal blood-CSF barrier function and to dysfunction in neurologi-
cal diseases. J Neurol Sci. 1994;122:189–203.
[28] Tumani H, Teunissen C, Süssmuth S, et al. Cerebrospinal fluid biomark-
ers of neurodegeneration in chronic neurological diseases. Expert Rev
Mol Diagn. 2008;8:479–94.
[29] Yu H, Li H, Liu X, Du X, Deng B. Levels of serum S100B are associ-
ated with cognitive dysfunction in patients with type 2 diabetes. Aging
(Albany, New York). 2020;12:4193–203.