The usefulness of S100B, NSE, GFAP, NF-H ... - Springer Link

Acta Neurochir (2012) 154:93–103 DOI 10.1007/s00701-011-1175-2

CLINICAL ARTICLE

The usefulness of S100B, NSE, GFAP, NF-H, secretagogin and Hsp70 as a predictive biomarker of outcome in children with traumatic brain injury Jiří Žurek & Michal Fedora

Received: 18 June 2011 / Accepted: 15 September 2011 / Published online: 7 October 2011 # Springer-Verlag 2011

Abstract Background Predicting the long-term outcome after traumatic brain injury (TBI) is an important component of treatment strategy. Despite dramatically improved emergency management of TBI and apparent clinical recovery, most patients with TBI still may have long-term central nervous system (CNS) impairment. Methods Sixty-three patients with TBI were enrolled into the prospective study. Venous blood samples were taken at admission and every 24 h for a maximum of 6 consecutive days. Serum concentrations of the biomarkers S100B, neuron-specific enolase (NSE), GFAP, NF-H, secretagogin and Hsp70 were quantified immuno-luminometrically or by enzyme-linked immunosorbent assay. The outcome was evaluated 6 months after TBI using the Glasgow Outcome Scale (GOS) in all patients. Results The S100B levels in patients with worse outcome (GOS 4 or death) were already significantly higher at D0 (p < 0.001; p = 0.002). NSE levels were significantly higher in patients who died or had worse outcomes (p < 0.001; p = 0.003). Patients who had worse outcomes (GOS) or died had higher GFAP values (p < 0.001; p < 0.001), but their dynamics were similar over the same period. NF-H grew significantly faster in patients who had a worse GOS or died (p < 0.001; p = 0.001). Conclusions Although further prospective study is warranted, these findings suggest that levels of biomarkers J. Žurek (*) : M. Fedora Department of Anesthesia and Intensive Care, University Children‘s Hospital, Černopolní 9, Brno 62500, Czech Republic e-mail: [email protected] M. Fedora e-mail: [email protected]

correlate with mortality and may be useful as predictors of outcome in children with TBI. Keywords Biomarker . Brain injury . Outcome . Children

Introduction Head injuries are the leading cause of death in children over 1 year old and are the third leading cause in children under 1 year of age . The injury rate is about two times higher in boys than in girls, and mortality is more than threefold [22]. In Czech Republic, in 2009 there were 525,803 cases of injury of children and adolescents under 20 years old treated in the surgical ambulances. About 38 thousands cases of injury of persons aged 0–19 years were treated in hospital. The results of these surveys show 16% incidence of injuries among school children [17]. Traumatic head injuries in infants, children and adolescents are caused by different mechanisms—from frequent falls and motorized collisions with nonmotorized vehicles to child abuse or gunshot wounds. Traffic accidents are the leading cause of severe traumatic brain injury (TBI) [37]. Several typical questions confront clinicians regarding the extent of injury and prognosis. Because there is such marked heterogeneity in TBI, predicting outcomes is difficult, as is deciding on the optimal treatment. For patients who survive severe TBI, some make a good recovery, while others are left with severe neurological disability. In children, uncertainties about predicting outcome are even greater than in adults and are multifactorial [11]. There are increased difficulties in assessing both the severity of initial injury and outcome. Objective tools to

94

better understand the degree of injury and to prognosticate are urgently needed. The application of biomarkers could lead to expeditious diagnosis in the case of sedated, unconscious or polytraumatized patients. Although there is an increasing demand for protein biomarkers in the management of TBI patients and the animal/clinical research has boomed in this field over the past 10 years, no single brain-specific biomarker has yet been unanimously established for traumatic brain injury in routine clinical practice. The primary aim of our study was to observe how dynamic biomarkers of brain damage correlated with mortality and outcome. Our second aim was the correlation of biomarkers with neurological deficits. Potential biomarkers of brain injury S100B protein S100B protein belongs to a multigenic family of low molecular weight (9–13 kD) calcium-binding S100 proteins [14]. S100B is most abundant in glial cells of the central nervous system, predominately in astrocytes [8]. Intracellularly, S100B is involved in signal transduction via the inhibition of protein phoshorylation and regulation of enzyme activity, and by affecting calcium homeostasis [31]. Moreover, S100B is functionally involved in the regulation of cell morphology by interaction with elements of the cytoplasmatic cytoskeleton. S100B is actively secreted from astroglia via an unknown mechanism and also exerts extracellular functions. In cultured rat astrocytes, S100B release can be stimulated by 5-HT1a agonists [40] and is observed within a few minutes after activation of A1 adenosine or mGlu3 metabotropic glutamate receptors. Release may last up to 10 h [6]. S100B has a biological half-life of 2 h. It is not influenced by hemolysis, and it remains stable even if samples are not centrifuged and frozen immediately [36]. Protein S100B can be detected in both cerebrospinal fluid (CSF) and in blood serum, and its concentration has been shown to increase in CSF and/or serum after a number of cerebral diseases, including traumatic brain injury [39], cerebral infarction [15], subarachnoid hemorrhage [41] and cerebral infections [35]. Further experimental studies in rats have shown that S100B is significantly increased early after bilateral femur fracture [27], and after local ischemia and reperfusion of the liver, gut and kidney [23]. Neuron-specific enolase (NSE) Enolases are glycolytic enzymes occurring as a series of dimeric isoenzymes made of three immunologically distinct subunits, the α, β and γ chains. The isoforms γ γ and α γ

Acta Neurochir (2012) 154:93–103

are restricted to neurons, peripheral neuroendocrine tissue, and tumors of the amine precursor uptake and degradation system [7]. Structural damage of neuronal cells causes leakage of NSE into the extracellular compartment and the bloodstream. Thus, it can be detected in the serum after neuronal death secondary to a traumatic injury or a cerebrovascular accident [2]. However, like all biomarkers, NSE has limitations. Similar to S100B, however, there have been reports of false-positive values in the setting of combined CNS injury plus shock [24]. Glial fibrillary acidic protein (GFAP) GFAP is a monomeric intermediate filament protein found in the astroglial cytoskeleton and is not found outside the CNS. Intermediate filaments form networks that provide support and strength to cells. Astroglial cells support and nourish cells in the brain and spinal cord. If brain or spinal cord cells are injured through trauma or disease, astroglial cells react by rapidly producing more glial fibrillary acidic protein. Although its function is not fully understood, glial fibrillary acidic protein is probably involved in controlling the shape, movement and function of astroglial cells. Some researchers have suggested that astroglial cells play an important role in the functioning of other cells, including specialized cells that surround nerves (oligodendrocytes) and are involved in the production and long-term maintenance of myelin [9]. It is released after CNS damage and thus may serve as a marker of TBI. Hyperphosphorylated neurofilament NF-H Neurofilaments are the most abundant protein components of neurons and should therefore be released from damaged and dying neurons in large amounts. Due to their abundance it should be relatively easy to detect them, and, because they are only found in neurons, detection of these proteins points unambiguously to neuronal damage. Neurofilaments consist predominantly of three subunits, namely NF-L (low), NF-M (medium) and NF-H (heavy or high). NF-H contains unusual tandemly repeated peptides centered on the sequence lysine-serine-proline (KSP). These KSP repeats are very numerous, up to 60 in some mammalian species. In axonal neurofilaments essentially all of the serine residues are phosphorylated [33]. Several studies have shown that pNF-H is more resistant to calpain and other proteases than either NF-L or NF-M [34]. Taken together, these findings suggest that pNF-H might be an unusually good candidate for a biomarker of axonal injury and degeneration.


Secretagogin Secretagogin (SCGN, SEGN, CALBL, setagin) is a Cabinding protein that consists of six-EF hand Ca-binding domains and is characterized by a 32-kDa molecular weight. Generally, Ca-binding proteins seem to be involved in several pathological conditions in the CNS, such as epilepsy or neurodegenerative disorders (for example, Alzheimer's disease). In the brain, secretagogin expression is restricted to distinct subtypes of neurons, with its highest expression in the molecular layer of the cerebellum, in the anterior part of the pituitary gland, in the thalamus, in the hypothalamus and in a subgroup of neocortical neurons. Secretagogin was also detectable in tissue homogenates of various other brain regions, but only to a very minor extent [30]. Moreover, recent study shows that secretagogin is detectable in human serum after ischemic brain damage [10]. Heat shock proteins (Hsp) The so-called heat shock or stress response is a highly conservative cellular response to injury. This cellular defense mechanism is characterized by the increased expression of heat shock or stress proteins that provide the cell with increased protection from insults that would otherwise be lethal, a phenomenon referred to as thermotolerance or preconditioning [20]. In nonstressed cells, Hsps are present in low concentrations, while in stressed cells they accumulate at high levels [42]. The Hsp70 family The 70-kDa heatshock-related proteins compose a family of molecular chaperones that regulate cellular processes during normal and stress conditions [32]. Hsp70 is one of the most abundant of these proteins, accounting for as much as 1–2% of total cellular protein [42]. In humans, they are inducible in physiological conditions such as apoptosis, tissue development, hormonal stimulation and immune responses, as well as in response to stressors [28]. In the brain, Hsp70 is induced by a variety of pathological stimuli, including ischemia, excitotoxic insults and inflammatory responses, major determinants of acute neural injury in TBI [12]. Data for pediatric patients with brain injuries are not available at this time.

Materials and methods The prospective observational study took place from May 2007 to October 2009. The study protocol and informed consent approach were approved by the ethics committee of

95

the University Hospital, Brno. Parents provided informed written consent for their children to participate in this trial. We prospectively enrolled 63 pediatric patients aged 0– 19 years old. The criteria required for inclusion were TBIs occurring a maximum of 12 h before admission. TBIs were verified by computerized tomography (CT). Patients whose parents (legal guardians) opposed inclusion in the study were excluded. We recorded the sex, age, type of accident (road traffic accident, bike accident, fall, assault), Glasgow Coma Scale (GCS) score and Pediatric Trauma Scale (PTS) score for each patient. The Glasgow Coma Scale score was based on evaluation of the emergency measures taken, but without consideration of pharmacological interventions [37]. The Pediatric Trauma Scale (PTS) was selected as the trauma scoring tool for use in evaluating the severity of injury in the pediatric patients [38]. Brain CT scans were rated by radiologists blinded to the study and using the Marshall grading scale [21]. CT was repeated after neurosurgical intervention and whenever required by the clinical course. After diagnostic assessment and/or surgery, all patients were transferred to the pediatric intensive care unit and received standard neurointensive care, including intubation and mechanical ventilation, hemodynamic and intracranial pressure (ICP) monitoring based on the TBI therapeutic protocol. Intracranial hypertension was treated progressively by use of a standard stepwise protocol that included sedation, paralysis, mild hyperventilation (target paCO2 32–35 mmHg), osmotherapy with mannitol and use of barbiturates. Cerebral perfusion pressure was maintained at 60 mmHg (50 mmHg in infants) by lowering intracranial pressure to 20 mmHg (15 mmHg in infants) and by maintaining mean arterial blood pressure at 80 mmHg (70 mmHg in infants). ICP was monitored in the 15–20° position through an intraparenchymal pressure monitor (Codman, Johnson & Johnson Co., Raynham, MA, USA) placed in one of the frontal lobes. ICP was noted hourly on the subject’s data sheet. Lastly, 6 months after the primary injury, the study protocol documented outcome according to the Glasgow Outcome Scale (GOS) scoring 1–5 (1, dead; 2, persistent vegetative state; 3, severely disabled; 4, moderately disabled; 5, good recovery) and the cause of death [18]. Clinical outcome (any neurological deficit) was measured 6 months after the primary injury, based on neurological examination [including cranial nerve 1–12 assessment, deep tendon reflexes including pathological reflexes, sensory examination (including visual fields, audition and nociceptive/tactile responsiveness), cerebellar assessment and motor function testing, and other areas assessed] and electroencephalography. The diagnostician of the 6-month outcome was completely blinded to the clinical information.

96

Venous blood for protein measurement was sampled at admission and at the same time every morning, for a maximum of 6 days. S100B and NSE levels were measured immuno-luminometrically with a commercially available kit (Elecsys® analyzer, Roche Diagnostics). GFAP, pNF-H, secretagogin and Hsp70 were measured by sandwich enzyme immunoassay for quantitative measurement (BioVendor, Laboratorní medicína a.s., Czech Republic). The laboratory technicians who carried out the assays were completely blinded to the clinical information. Statistical methods Baseline levels of analyzed proteins and demographic characteristics were summarized using descriptive statistics (N, mean, standard deviation, median, minimum, maximum). The dynamics (kinetics) of the protein levels during the 6-day evaluated period were analyzed using analysis of variance (ANOVA). A compound symmetry co-variance matrix was adopted in order to account for the withinsubject correlation (repeated measure ANOVA). Parameters of interest (mortality, GOS, hospital stay duration) were dichotomized and analyzed in ANOVA models where all main effects of the parameter, main effect of time and their interaction were included. Graphs representing the resulting mean models along with individual case profile curves were given to illustrate the parameter behavior over the 6-day evaluation period. Significance levels of the difference in dynamics (represented by a parameter and time interaction effect) and the difference on D0 (represented by the parameter main effect on D0) of the parameters are presented in the tables. The analysis was performed on logarithmically transformed data to achieve an approximately normal distribution of the evaluated data.

Results Table 1 presents basic descriptive characteristics of the patient and the input parameters. Protein dynamics during hospitalization and their correlation with mortality, outcome and length of hospitalization The difference in day 0 (D0) is the difference that can be detected on the first collection from the patient (shortly after the accident). The value of the biomarker is the first indicator. The average difference is used if the biomarker values differ within 6 days. That is, the intial occurence of TBI was unknown, or if a patient was transferred from another hospital (biomarker value in the D0 is not


available). The difference in the dynamics shows how the levels differ within 6 days (it is simply illustrated in the table—either uniformly increasing or decreasing, or variously changing). The S100B levels in patients with worse outcome (GOS 4 or death) were already significantly higher at D0 (p < 0.001; p = 0.002). Compared to those who survived, the S100B values decreased significantly more slowly than in those who died. As assessed by the GOS, the dynamics of both groups were similar. Bordering are also higher values of S100B at the beginning of hospitalization for patients who were hospitalized longer. The difference is very discreet (Table 2). NSE levels were significantly higher in patients who died or had worse outcomes. Significantly different dynamics of this protein were evident only in patients who died when there was still a noticeable increase during the first 2 days. In patients who survived, the value of the protein decreased from the beginning. At day 4 (D4) a few patients had mild elevation (see gray profiles); however, the overall average does not show a "second peak" effect. In relation, the length of hospitalization does not show any difference (Table 3). Patients who had worse outcomes (GOS) or died had higher values of GFAP, but their dynamics were similar over the same period. Patients hospitalized longer had significantly different GFAP dynamics, but as can be seen from Table 4, this is rather a random variation (in small numbers of evaluable data in the later stages). This conclusion is supported by the non-significant difference in mean values for 6 days. Higher values of GFAP in patients with severe injuries are seen at the beginning, at D0 (Table 4). NF-H levels grew significantly faster in patients who had a worse GOS or died. For a total period of 6 days, it also reflected a significantly greater increase in patients who were hospitalized for longer times, even though the total value of NF-H remained approximately the same. The difference in the NF-H values appeared only during the 6 days. At D0 NF-H values are not significantly different (Table 5). Significant difference in the dynamics of Hsp70 in exitus is due to random fluctuations in the rather small number of assessable data. The average difference is not significant and supports this conclusion (Table 6). Secretagogin values are not significantly related to any of the studied parameters (Table 7). Neurological functional deficits and their correlation with the protein levels at admission Six months after the primary injury, neurological assessment was performed based on neurological examination. Statistical significance was demonstrated for S100B proteins, NSE and GFAP levels at admission and neurological follow-up at 6-month intervals. Patients who had


97

Table 1 Patient characteristics at admission Survivors

Non-survivors

p

N

Median (min–max)

N

Median (min–max)

Age [months] Pediatric Trauma Score Length of hospitalization [days] S100B [ug/ml] NSE [ug/l]

54 54 54 54 54

104.0 (2.0–220.0) 7.5 (3.0–12.0) 6.0 (1.0–24.0) 0.36 (0.04–14.83) 33.51 (12.51–97.90)

9 9 9 9 9

105.0 (8.0–171.0) 5.0 (−2.0–7.0) 5.0 (2.0–14.0) 1.60 (0.11–31.24) 61.84 (25.55–271.30)

0.448 0.002 0.492 0.077 0.020

GFAP [ng/ml] Hsp 70 [ng/ml] NF-H [pg/ml] Secretagogin [pg/ml]

54 54 54 54 N (%) 33 (86.8%) 21 (84%) 25 (80.7%) 5 (100%) 5 (83.3%) 19 (90.5%)

0.12 (0.00–47.65) 0.08 (0.00–15.74) 12.00 (12.00–1,442.00) 71.54 (11.00–915.09)

9 9 9 9

7.47 (0.00–46.32) 0.00 (0.00–4.38) 46.43 (12.00–1,482.00) 120.42 (11.00–2,934.33) N (%) 5 (13.2%) 4 (16%) 6 (19.3%) 0 (0%) 1 (16.7%) 2 (9.5%)

0.002 0.658 0.317 0.463

Boys Girls Car accident Bicycle accident Attack Fall

∼1.000 0.593

NSE, neuron-specific enolase; GFAP, glial fibrillary acidic protein; Hsp, heat shock protein; NF-H, neurofilament H

too high input values in the above-mentioned proteins had neurological deficits appear within 6 months (Table 8). The most common findings were hemiparesis, specific behavior, spastic quadriparesis and convergent strabismus.

Discussion Although extensive research has not found an ideal brain biomarker over the last decade, several biomarkers have

Table 2 S100B values within 6 days

S100B values within 6 days Factor

p (difference in

p (average difference in 6 days)

p (difference in dynamics)

D0)

Exitus (death)

0.002