S100B Protein May Detect Brain Death Development after Severe Traumatic Brain Injury

Abstract

Despite improvements in the process of organ donation and transplants, the number of organ donors is progressively declining in developed countries. Therefore, the early detection of patients at risk for brain death (BD) is a priority for transplant teams seeking more efficient identification of potential donors. In the extensive literature on S100B as a biomarker for traumatic brain injury (TBI), no evidence appears to exist on its prognostic capacity as a predictor of BD after severe TBI. The objective of this study is to assess the value of including acute S100B levels in standard clinical data as an early screening tool for BD after severe TBI. This prospective study included patients with severe TBI (Glasgow Coma Scale score [GCS] ≤8) admitted to our Neurocritical Care Unit over a 30 month period. We collected the following clinical variables: age, gender, GCS score, pupillary alterations at admission, hypotension and pre-hospital desaturation, CT scan results, isolated TBI or other related injuries, Injury Severity Score (ISS), serum S100B levels at admission and 24 h post-admission, and a final diagnosis regarding BD. Of the 140 patients studied, 11.4% developed BD and showed significantly higher S100B concentrations (p<0.001). Multivariate analysis showed that bilateral unresponsive mydriasis at admission and serum S100B at 24 h post-admission had odds ratios (ORs) of 21.35 (p=0.005) and 4.9 (p=0.010), respectively. The same analysis on patients with photomotor reflex in one pupil at admission left only the 24 h S100B sample in the model (OR=15.5; p=0.009). Receiver operating characteristics (ROC) curve analysis on this group showed the highest area under the curve (AUC) (0.86; p=0.001) for 24 h S100B determinations. The cut off was set at 0.372 μg/L (85.7% sensitivity, 79.3% specificity, positive predictive value [PPV]=18.7% and negative predictive value [NPV]=98.9%). This study shows that pupillary responsiveness at admission, as well as 24 h serum S100B levels, could serve as screening tools for the early detection of patients at risk for BD after severe TBI.

Introduction

Despite improvements in the process of organ donation and transplants, the number of organ donors is progressively decreasing in developed countries. This decline is attributed primarily to a drop in the incidence of severe traumatic brain injury (TBI), and the mortality rate among these cases, resulting in fewer patients with brain death (BD) outcome.^1,2 Moreover, transplant teams require strict, efficient follow-up of neurocritical patients to identify potential organ donors. Consequently, different strategies have been developed to find the best practices associated with organ donation and transplant management. Recent research has shown that the presence of an in-house transplant coordinator improves the donation process.³ However, the comprehensive follow-up provided by this charge would be unlikely in donor hospitals, which depend upon a joint coordination office, or in large trauma centers with high volumes of neurocritical admissions. Therefore, the early detection of TBI patients at risk for BD is a priority for transplant teams seeking more efficient identification of potential donors.

There are numerous tools used in severe TBI management that offer prognostic information, including clinical findings, CT scan, and multimodal neuromonitoring techniques.^4–6 However, all of these options are limited when it comes to establishing a prognosis for BD. Most studies explore the association between neuromonitoring data and the timing of BD, although a few attempt to predict this result.^7–9 In light of these limitations, studies that use serum biomarkers to predict BD development after TBI have specifically focused on the S100B protein.^10–12

S100B, a calcium-binding cytosolic protein, is released into the blood from astroglial cells as a consequence of brain injury and the disruption of the blood–brain barrier (BBB). It is involved in the regulation of calcium fluxes and stimulates astrocytosis proliferation.^13–15 The properties of S100B make it one of the most relevant biomarkers for central nervous system damage, although there is evidence of its release in the presence of melanoma, and after muscle and adipose tissue injury.^16–19 This protein is eliminated by renal excretion; however, when brain injury is present, S100B levels may remain elevated over time.^12,20 S100B has proven more reliable in predicting outcome than other biomarkers of brain injury after severe head trauma.^21–27 However, we found no evidence in the literature on the prognostic capacity of S100B as a predictor of BD during the first days of hospitalization after severe TBI.

The objective of this study is to assess the value of including acute S100B levels in standard clinical data as an early screening tool for patients at risk for BD after severe TBI.

Methods

This study prospectively included patients with severe TBI (Glasgow Coma Scale score [GCS] ≤8), admitted to the Neurocritical Care Unit at the Virgen del Rocío University Hospital in Seville, Spain, from June 2008 through June 2011. Written consent was obtained from patients' families. The protocol was approved by the Virgen del Rocío Hospital Ethics Committee.

Exclusion criteria included hospital admission>6 h post-trauma, pregnancy, being<15 years of age, alcohol or other drug abuse at the time of injury that would affect consciousness level, diagnosis of mild or moderate TBI, acute or chronic renal failure, history of neoplastic processes such as melanoma, psychiatric illness, neurodegenerative disease, or previous neurological sequelae. These data were collected from hospital clinical records or in-depth interviews with family members.

Clinical and demographic variables collected included age, gender, GCS score at admission (after excluding hemodynamic, pharmacological, or metabolic interference in GCS determination), pupillary response at admission (both normal, one wide and unresponsive, or both wide and unresponsive), hypotension (systolic blood pressure<90 mm Hg for>30 min, despite resuscitation with volume and/or amines), pre-hospital desaturation (oxygen saturation O₂<90%), Injury Severity Score (ISS), isolated TBI (when Abbreviated Injury Scale [AIS] scores in locations other than the head were ≤2) or other associated extracranial lesions, and BD or non-BD outcome. CT scan findings at admission were catalogued in accordance with the Traumatic Coma Data Bank (TCDB) classification system.²⁸ For analytical purposes, these results were grouped based on the presence of diffuse lesion (types I and IV), evacuated mass lesion, or non-evacuated mass lesion.

Patient management was in conformity with The Brain Trauma Foundation clinical practice guidelines, and protocols published by our institution.^4,29,30 These protocols included body temperature control, head elevation in bed, seizure prophylaxis, avoidance of jugular outflow obstruction, sedation, intubation, mechanical ventilation, and complete volume resuscitation to maintain a cerebral perfusion pressure (CPP) minimum of 60–70 mm Hg. Space-occupying lesions>25 cm³ were surgically removed. When intracranial pressure (ICP) exceeded 20 mm Hg, therapeutic interventions were initiated step by step: external ventricular drainage when possible, moderate hypocapnia (PaCO₂: 30–35 mm Hg), deeper sedation and muscle relaxation, and relative hyperosmolar therapy with mannitol or hypertonic saline infusions. In cases in which the interventions failed to control ICP, second tier therapies were considered: noradrenaline for arterial hypertension, PtiO2 monitoring for hyperventilation (PaCO₂<30 mm Hg), and high dose barbiturate therapy.

BD diagnosis was initiated based on clinical criteria and deteriorating CT scan results revealing catastrophic TBI lesions, and following legal provisions for BD determination in Spain and internationally recognized clinical requirements.³¹ We confirmed diagnoses as unresponsive coma with structural causes, with the specific exclusion of hypotension, hypothermia, hydroelectrolytic and metabolic disturbances, and the effects of muscle relaxants/sedatives or other toxins. Clinically confirmed signs and symptoms of BD included absence of motor response, bilateral wide and unresponsive pupils, absence of corneal reflex, absence of vestibulo-ocular and oculocephalic reflexes, absence of gag reflex, absence of coughing in response to tracheal suctioning, negative atropine test (<10% increment from the basal rhythm after administration of 0.04 mg/kg atropine IV), and absence of respiratory drive (PaCO₂: 60 mm Hg or 20 mm Hg above normal baseline values).³¹

After the clinical examination, we performed confirmatory tests to establish the absence of cerebral circulation or electrical brain activity.^31–33 TCDB was used to confirm sonographic patterns compatible with cerebral circulatory arrest: diastolic–systolic separation, reverberating flow, and bilateral insonation of isolated systolic spikes in anterior circulation arteries and the basilar artery.^34–36 Electroencephalogram (EEG) was performed in cases without an acoustic signal, or with non-hermetic skull (craniotomy, ventricular drainage, wide cranial fracture). EEG patterns were recorded for 30 min, with amplification at 2 μV/mm and band frequencies set between 0.3 and 30 Hz. Electrodes were positioned at least 10 cm apart, with scalp locations on frontal, temporal, occipital, and parietal regions. Electrocerebral silence, null recordings or other symptoms, such as a flat EEG, confirmed the absence of electrical brain activity.³⁷

A venous blood sample was taken at admission (during the first 6 h post-trauma), and 24 h later, on all surviving patients. The samples, sent to the laboratory by means of a pneumatic tube system, were centrifuged at 3000 rpm for 10 min at room temperature, frozen at −80°C, and stored for analysis. Serum S100B levels were measured using an electrochemiluminescence assay (ECLIA), a commercially available test produced by Elecsys 2010 Immunoassay Systems (Roche Diagnostics, Germany). According to the manufacturer, this test lasts 18 min and requires a minimum probe volume of 20 μL of serum. Detection begins at 0.005 μg/L. Concentrations<39 μg/L can be quantified without dilution. The results are given in micrograms per liter (μg/L). The biochemist and technician performing the analyses were blind to clinical and radiological findings.

Statistical analysis

A descriptive analysis was performed using qualitative variables, represented in the tables as absolute frequencies and percentages. After applying the Kolmogorov–Smirnov or Shapiro–Wilks tests (depending upon subgroup size), quantitative variables were expressed using mean and standard deviation (SD) or median and interquartile range (IR) (P₂₅–P₇₅), depending upon whether or not they followed a normal distribution. Spearman's correlation coefficient (Spearman's ρ) was used to measure the statistical dependence between serum S100B concentrations and BD outcome. We applied the nonparametric Mann–Whitney U test to compare mean quantitative differences between subgroups, if variables were not normally distributed. The Kruskal–Wallis test was applied for multiple group comparisons. If significant differences were obtained, 95% confidence intervals (CI) were calculated. Pearson's χ² test or Fisher's exact test was performed to assess the degree of dependence between categorical variables. Multivariate analyses were performed using logistic regression. We applied this method to describe the relationship between a dichotomous dependent variable (BD outcome) and a set of independent variables (hypothetical predictive factors). This method selects the best set of predictive variables for BD among those reaching a significance level<0.15 in the univariate analysis. Odds Ratio (OR) with 95% CI was calculated for all variables entered into the model. The predictive capacity of each model was calculated using receiver operator characteristics (ROC) curve analysis. ROC curve analysis on S100B values was used to differentiate between patients in the BD and those in the non-BD outcome group. This analysis, which provide area under the curve (AUC) measures, 95% CI, and plotted coordinates (sensibility and false positives), was used to establish a cutoff value for patient classification. All statistical analyses were conducted using software from the Statistical Package for the Social Sciences (SPSS) (Version 18.0, Chicago, IL).

Results

A total of 146 patients were selected prospectively for this study. Five patients were excluded because they were admitted>6 h post-trauma or had had prior treatment in other hospitals, and one family declined participation on the part of their injured relative. A total of 140 severe TBI patients, 116 male (83.6%) and 24 female (16.4%), met the inclusion criteria. Mean age was 35.6 (SD, 15.3; 95% CI, 33.0–38.2). All CT scan results showed intracranial lesion at admission (TCDB I, n=0). Sixteen patients were declared BD. The moment of BD diagnosis, expressed by mean value, occurred on day 2 (inter-observer reliability [IR]: 1–7); 50% of the patients were diagnosed with BD within the first 48 h. Table 1 provides a summary of patients' demographic and clinical data.

Table 1.

Characteristics, Demographics, and Clinical Data on 140 Patients

Gender, male n (%)	116 (83.6%)
Age, mean (SD)	35.6 (15.3) (CI 95% 33.0–38.2)
Cause of TBI, n (%)
Traffic accident	84 (60.0)
Fall from height	35 (25.0)
Aggression	6 (4.3)
Hit by a motor vehicle	8 (5.7)
Other	7 (5.0)
GCS, n (%)
8	35 (25.0)
7	44 (31.4)
6	16 (11.4)
5	13 (9.3)
4	10 (7.1)
3	22 (15.7)
Pupillary response,an (%)
Both normal	95 (67.8)
One wide and unresponsive	33 (23.6)
Both wide and unresponsive	12 (8.6)
Arterial hypotension/desaturation,b
n (%)	45 (32.1)
TCDB, n (%)
I	0 (0)
II	58 (41.4)
III	16 (11.4)
IV	5 (3.6)
Evacuated Lesion	53 (37.9)
Non-evacuated lesion	8 (5.7)
No extracranial injuries, n (%)	37 (26.4)
Brain death, n (%)	16 (11.4)

^aAt admission.

^bPre-hospital.

TBI, Traumatic Brain Injury; GCS, Glasgow Coma Scale; TCDB, Trauma Coma Data Bank.

For the univariate analysis, we selected clinical variables that could influence BD outcome (see Table 2). Patients with BD diagnosis showed a significantly higher incidence of pupillary photomotor alterations at admission (p=0.006) and mass lesion (evacuated and non-evacuated) on CT scan (p=0.011). Mean serum S100B levels were higher in samples taken at admission (0.683 μg/L), and 24 h later (0.474 μg/L), in patients with BD outcome (p<0.001). No significant differences were found in median S100B values after dichotomization based on the following variables: age, gender, GCS score at admission, pre-hospital desaturation or hypotension, pupillary response at admission, and extracranial lesions associated with TBI. As shown in Figure 1, mean S100B concentrations for the BD group were consistently higher than those for the non-BD group throughout the study period. S100B concentrations at admission correlated negatively with the moment of BD diagnosis (at admission, r=− 0.653 (p=0.006) and at 24 h, r=− 0.825 (p<0.001).

Table 2.

Patients' Clinical Characteristics Based on Outcome Subgroups: Brain Death/Non-Brain Death

Brain Death	No	Yes	p
Results n (%)	124 (88.60)	16 (11.40)
Mean age (SD)	35.32 (14.71)	37.88 (19.87)	0.883
95% CI	32.71–37.94	27.29–48.46
Gender n(%)			0.999
Male	102 (83.06)	14 (87.50)
Female	21 (16.93)	2 (12.50)
Median ISS (P25-P75)	30 (IR:25–38)	38 (IR:25–48)	0.209
Pupillary response, n (%)			0.006
Both normal	89 (72.70)	6 (37.50)
One wide and unresponsive	30 (23.40)	3 (18.80)
Both wide and unresponsive	5 (3.90)	7 (43.80)
Arterial hypotension/desaturation,an (%)	35 (29.03)	8 (50)	0.713
No extracranial injuries, n (%)	30 (24.19)	7 (43.75)	0.227
GCS, mean (SD)	6.38 (2.11)	5.25 (1.88)	0.506
TCDB, n (%)			0.011
I-IV	74 (58.3)	5 (31.2)
Evacuated	45 (37.5)	8 (50.0)
Non evacuated	5 (4.2)	3 (18.8)
S100B at admission	0.309	0.683	<0.001
Median (P25-P75)	(0.191–0.555)	(0.298–1.902)
S100B at 24h	0.213	0.474	<0.001
Median (P25-P75)	(0.117–0.325)	(0.244–1.790)

Significant results are highlighted in bold.

^aPre-hospital.

ISS, Injury Severity Score; GCS, Glasgow Coma Scale; TCDB, Trauma Coma Data Bank.

FIG. 1.

Box and whisker plot of serum S100B levels (μg/L) in patients at admission (A), and 24 h post-admission (B), based on brain death or non-brain death outcome (both p<0.001). Symbols: ○, atypical value; *, extreme values.

Binary multiple regression analysis showed that ORs for bilateral unresponsive mydriasis at admission were 30.72 (95% CI, 7.32–128.90; p=0.001). The ORs for evacuated and non-evacuated mass lesions were 3.11 (95% CI, 1.88–10.93; p=0.007) and 10.5 (95% CI, 1.82–60.45; p=0.008), respectively. We also found that for each 1 μg/L increase in S100B value at admission, the OR for developing BD was 1.99 (95% CI, 1.21–3.32; p=0.008). In the 24 h sample, the OR for BD outcome was 5.37 (95% CI, 1.85–15.59; p=0.002). The results of this test indicate that the nearer the GCS score is to 3, the higher the probability of BD outcome (OR=1.34; [95% CI, 1.08–1.77; p=0.044]) (see Table 3).

Table 3.

Logistic Regression Analysis of Clinical Variables that Could Influence Brain Death Outcome

Clinical variables	p	OR (95% CI)
Gender	0.654	0.70 (0.14–3.31)
Age	0.530	1.01 (0.97–1.04)
ISS	0.031	1.047 (1.01–1.09)
GCS	0.044	1.34 (1.08–1.77)
Hypotensiona	0.096	2.44 (0.85–7.01)
Desaturationa	0.111	2.35 (0.82–6.73)
No extracranial injuries	0.103	2.44 (0.84–7.10)
Both pupils wide and unresponsive	0.001	30.72 (7.32–128.90)
Mass type TCDB
Evacuated	0.007	3.11 (1.88–10.93)
Non-evacuated	0.008	10.5 (1.82–60.45)
S100B
At admission	0.008	1.99 (1.21–3.32)
At 24 h	0.002	5.37 (1.85–15.59)

Significant results are highlighted in bold.

^aPre-hospital.

OR, odds ratio; CI, Confidence Interval; ISS, Injury Severity Score; GCS: Glasgow Coma Scale; TCDB, Trauma Coma Data Bank.

Significant variables and variables considered clinically relevant (GCS score, ISS, pre-hospital desaturation or hypotension, CT scan findings, pupillary photomotor alterations at admission, and S100B levels) were included in the multivariate analysis model. Variables that remained in the model included bilateral unresponsive mydriasis at admission (OR=21.35) and 24 h S100B value (OR=4.90), with an AUC of 0.872. Multivariate analysis without S100B values obtained an AUC of 0.71 (95% CI, 0.574–0.863; p=0.008). Despite the higher AUC, no significant differences were found between the two models. Given the strong association between bilateral unresponsive mydriasis and BD, we generated a second model that included only patients with photomotor reflex in at least one pupil. In this analysis, the only remaining variable in the model was the 24 h S100B sample (OR=15.5; AUC=0.96) (see Table 4).

Table 4.

Multivariate Analysis

Model 1	Clinical Variable	Bilateral unresponsive mydriasis at admission	S100B at 24 h
	OR (95%CI)	21.35 (5.50–82.54)	4.90 (1.43–16.72)
	p	0.005	0.010
	AUC (95%CI)	0.872 (0.776–0.968)
	p	<0.001

Model 2	Clinical Variable	S100B at 24 h
	OR (95%CI)	15.5 (1.93–121.82)
	p	0.009
	AUC (95%CI)	0.966 (0.637–1.00)
	p	0.001

Model 1, all patients; Model 2, patients with photomotor reflex in at least one pupil at admission.

AUC, area under the curve; CI, confidence interval; OR, odds ratio.

The ROC analysis showed that S100B values at admission, and 24 h later, could correctly predict patient evolution toward BD. The AUC for the admission sample was 0.71 (95% CI, 0.56–0.85; p=0.009), whereas the AUC for the 24 h sample reached 0.78 (95% CI, 0.66–0.90, p<0.001) (see Fig. 2A). The multivariate analysis showed bilateral unresponsive mydriasis at admission as exhibiting the highest prognostic capacity. For this reason, we used the ROC analysis to focus on patients with responsive photomotor reflex in at least one pupil at admission. Figure 2B displays the discriminatory capacity of S100B and AUC increases in both S100B determinations: 0.80 at admission (95% CI, 0.66–0.94; p=0.007) and 0.86 24 h later (95% CI, 0.72–0.99). To maximize the relationship between sensitivity and specificity, we used the highest AUC plot (24 h) from the patient sample with photomotor reflex in at least one pupil at admission to assign a cutoff value for serum S100B. The optimum cutoff was 0.372 μg/L. At this level, the diagnostic properties of S100B as a biochemical predictor of BD after TBI would be 85.7% sensitivity, 79.3% specificity, 18.7% PPV and 98.9% NPV. Table 5 displays the distribution of this subpopulation using the S100B cutoff.

FIG. 2.

(A) Receiver operating characteristic (ROC) analysis comparing sensitivity to specificity of S100B in serum to determine brain death outcome. (B) ROC analysis comparing serum S100B sensitivity and specificity in determining brain death outcome, with a special focus on patients with photomotor reflex in at least one pupil at admission. Dotted line: S100B levels at admission. Continuous line: S100B level at 24 h post-admission.

Table 5.

Distribution of Patients with Photomotor Reflex in at Least One Pupil at Admission Using 24 h S100B Cutoff Point

Cutoff value (μg/L)	Non BD	BD	Total, n (%)
S100B≤0.372	95	1	96 (75)
S100B>0.372	26	6	32 (25)
Total	121	7	128 (100)

BD, brain death.

Discussion

This study, designed to assess the risk of BD development after severe TBI by means of serum S100B levels, illustrates that these levels are indeed higher in patients with BD outcome. Concretely, S100B determination was found to be more effective as a predictor in patients with photomotor reflex in at least one pupil at admission, given that its predictive capacity in patients with bilateral unresponsive mydriasis at admission does not surpass the clinical variable itself. These results could be useful to transplant teams for more efficient and objective identification of post-TBI patients at risk for BD.

No significant correlations were found between the probability of developing BD and clinical factors related to poor outcome (age, gender, arterial hypotension, and pre-hospital desaturation). Furthermore, serum concentrations at admission and at 24 h, in addition to ISS and CT-detected mass lesions (evacuated and non-evacuated), had a significant predictive role in BD outcome. Our results concur with those of previous studies reporting that a low GCS score at admission increases the probability of BD development, and that bilateral unresponsive mydriasis at admission is a sign of very poor prognosis (OR=21.35).^12,38 In our study, serum S100B levels did not correlate with pupillary response at admission, suggesting the protein's predictive capacity for BD in patients regardless of pupillary state. However, its prognostic capacity in the presence of bilateral unresponsive mydriasis at admission is not clinically relevant. These findings led us to focus on the utility of S100B in patients with apparently lower clinical severity at admission, those with at least some pupillary photomotor reflex. This patient group made up 91.4% of our sample. We concur with other authors in that S100B values obtained>12 h post-trauma are more precise than early serum values.^10,24 Once the post-TBI resuscitation phase is complete, S100B values should more precisely reflect tissue damage and disruption to the BBB, as well as the possible advance of secondary injury. This, in turn, could help explain the underlying prognostic properties found in the 24 h sample. Therefore, our analysis would indicate that pupillary responsiveness at admission, as well as 24 h S100B levels, could serve as screening tools for the early detection of patients at risk for BD after severe TBI.

A review of the literature identified only three studies that focused on the role of serum S100B in BD outcome.^10–12 Two were case studies with very limited sample sizes, n=3 and n=15, respectively.^10,11 They reported that S100B levels were higher in patients with BD outcome. The series with the highest number of patients to date, by Dimopoulou et al., included 47 patients with severe TBI and supported the previous findings, although the authors themselves admitted to the limitations brought on by a relatively small sample size. Nevertheless, they recommended including this biomarker in routine neuromonitoring procedures.¹² As far as we know, our study sample of 140 severe TBI patients is higher than any found in the literature. Our results were similar to those reported by Dimopoulou et al., showing that serum S100B levels were twice as high in the BD outcome group as in the non-BD group. Recent research by Böhmer et al. focuses on the analysis of cerebrospinal fluid (CSF) in 20 severe head trauma patients.³⁹ The authors found early elevations (up to 3 days) of S100B in patients with BD outcome. However, we consider CSF sampling to be more complex than serum sampling, requiring the insertion of an intraventricular catheter, which is not always necessary in post-TBI management.

In Figure 3, we provide a simple action protocol. Its implementation would facilitate the flow of information and transplant strategies among donor hospitals, transplant centers, and organ procurement organizations in several ways. First, it would improve transplant coordinators' follow-up efforts on neurocritical patients, by serving as an early risk indicator. We chose 0.372 μg/L as the 24 h cutoff, to optimize sensitivity without excessively lowering specificity. We understand that this cutoff generates false positives. Therefore, its predictive utility should focus on patients whose S100B levels are below this threshold, and thereby represent a lower risk of BD outcome. In our study, transplant coordinators would be able to exclude 75% of patients with at least one preserved photomotor response at admission from follow-up, given their low probability of BD development. The remaining 25% could be considered at risk, albeit assuming a high rate of false positives. Second, this follow-up of a smaller percentage of patients would cut down on time and optimize existing strategies if BD is diagnosed. Studies show that shorter delays between the pronouncement of BD and requests for consent are associated with higher consent rates.⁴⁰ Time savings may also benefit organ viability, given that significant hemodynamic and inflammatory changes take place after BD, which can be detrimental to organ function.⁴¹ Research using experimental BD models reported inflammatory changes and progressive organ dysfunction after BD.^42–44 At the clinical level, biopsies of livers from BD patients revealed that later organ harvest increased the expression of apoptosis-related genes and the presence of inflammatory phenomena.^45,46 In light of this evidence, shorter intervals between the establishment of BD and organ extraction would result in less exposure to the systemic inflammation triggered by BD, reducing organ loss, increasing organ quality, and significantly improving graft survival rate, while shortening the gap between organ supply and demand. Finally, we underscore that the data generated from our results would be strictly informative for physicians and transplant coordinators, without direct therapeutic implications. By no means are they intended to jeopardize the treatment or support of patients above the cutoff point, be used to make a premature prognosis, or be a means of coercion for family members to sign organ donation consent forms prior to BD diagnosis.

FIG. 3.

Algorithm for screening patient risk for brain death (BD), for follow-up by transplant coordinators.

There are some limitations to this study, mainly regarding the lack of testing>24 h post-admission. By extending this sample period, we could have obtained determinations that would show later increases in S100B levels and optimize data on this biomarker. Another limitation may be, as some studies have reported, that the extracranial origin of the S100B protein could lead to false increases of its serum levels.^16,17 However, other research has shown that the effects of this extracranial origin are few, and are quickly eliminated.⁴⁷ Our results concur with research that found no global differences in S100B levels between patients with TBI-associated extracranial lesions and patients with isolated TBI, regardless of severity level.^19,48,49 Moreover, patients with extracranial lesions in our series were distributed homogeneously between the BD group and the non-BD group. It would be interesting to study other biomarkers of brain injury in conjunction with S100B, particularly as most research on BD prediction using serum markers has focused exclusively on S100B.^10–12 However, this line of research will be included in future studies. At present, we would like to continue our research with larger samples sizes, sufficient for the specific analysis of patients without pupillary alterations at admission. Finally, the present data were collected exclusively from one medical center, and, therefore, would require external validation by means of a multicenter study.

The strengths of this study rest on its large sample size and its focus: the utility of acute serum S100B values in identifying patients at risk for BD. To achieve a homogeneous patient sample and avoid an initial GCS score influenced by drugs or external factors (sedation, anemization, substance abuse, hypothermia, hydroelectric or acid-base equilibrium disorders), we determined the GCS reference score after reverting the effects of these agents. BD diagnosis was established using clinical data and instrumental testing in all cases. It is also important to stress that patients underwent uniform TBI management as stipulated in the clinical guidelines of The Brain Trauma Foundation and protocols established by our institution.

Conclusion

In conclusion, the determination of S100B in blood at 24 h post-TBI provides an early and sensitive biomarker for the prediction of BD, particularly in patients with photomotor reflex in at least one pupil at admission. These determinations, in turn, could be used to define early risk patterns for BD. Its inclusion in neuromonitoring techniques would facilitate early detection of potential donors by transplant teams. Nevertheless, further studies with larger samples would be needed to validate our results and establish an optimal cutoff value for S100B levels to predict BD after severe TBI.

Acknowledgments

This research was made possible by funding from the Junta de Andalucia Health Department 2010 Grant Fund (cod. S0430), and through the generous donation of Protein S100B Electrochemiluminescence Assay Kits from Roche Diagnostics, Mannheim, Germany. We thank the nursing staff at the Virgen del Rocio University Hospital for their collaboration with blood sample extractions related to this study.

Author Disclosure Statement

No competing financial interests exist.