OBJECTIVE: To determine if serum levels of S100B are higher in children with CHT and ICI as detected by cranial CT and if long bone fractures affect the level of S100B in children with CHT and skeletal injury.
METHODS: Children <18 years of age who presented to an urban pediatric emergency department or were transferred from a referral hospital within 6 hours after accidental closed head trauma and who underwent cranial computed tomography were enrolled prospectively. Mean serum S100B levels for children with or without intracranial injury (ICI) and long-bone fractures were evaluated through analysis of covariance.
RESULTS: One hundred fifty-two children, 24 with ICI and 128 without ICI, were enrolled prospectively. Twenty-five children had long-bone fractures. Children with ICI were significantly younger than those without ICI (6.9 vs 9.8 years; P = .01). The time of venipuncture after injury was significantly later in children with ICI (P = .03). Mean S100B levels were significantly greater for children with ICI (212.9 vs 84.4 ng/L; P = .001), children with long-bone fractures (P = .008), and nonwhite children (P = .03). After controlling for time of venipuncture, long-bone fractures, and race, mean S100B levels were still greater for children with ICI (409 vs 118 ng/L; P = .001). The ability of serum S100B measurements to detect ICI, determined as the area under the curve, was 0.67.
CONCLUSIONS: After controlling for time of venipuncture, long-bone fractures, and race, S100B levels were still higher in children with ICI than in those without ICI. However, the ability of serum S100B measurements to detect ICI was poor.
Head trauma is a major cause of morbidity and death among children.1–3 Each year, >400000 children <14 years of age are evaluated in emergency departments (EDs) across the United States after closed head trauma (CHT).1 The majority of children with worrisome symptoms (such as headache, vomiting, and loss of consciousness) after CHT do not have intracranial injury (ICI), as detected with cranial computed tomography (CT).4–7 Many children with symptoms who present to the ED after CHT undergo cranial CT, because there is no other diagnostic method to detect ICI. Less than 10% of such studies in children demonstrate ICI,4–7 however, and routine diagnostic use of CT in children has been called into question because of the risks of radiation.8–10
In the past 10 years, researchers examined the extent to which serum biomarkers such as S100B, glial fibrillary acidic protein, and neuron-specific enolase could determine the presence of ICI after CHT in adults.11–14 Results of those studies suggested that S100B might be more useful than other biomarkers.15
The S100 proteins consist of dimers, each with a subunit (either S100A or S100B).16 S100B is concentrated primarily in the central nervous system and, to a lesser degree, in chondrocytes and adipocytes.17, 18 Because of its large molecular weight, S100B is able to pass into the systemic circulation only when the blood-brain barrier is disrupted, such as after ICI. S100B has a half-life of 6 hours in serum and is readily excreted in urine.19 Serum levels of S100B were demonstrated to be elevated in adults with both medical and traumatic causes of brain injury.20–22 Individuals with Down syndrome, epilepsy, or Alzheimer disease express higher baseline levels of S100 than do healthy individuals.23, 24
Because S100B is produced by chondrocytes, long-bone fractures may contribute to serum levels of S100B.25, 26 In those studies, long-bone fractures, in the absence of brain injury, increased S100B levels significantly, which suggests that such injuries may affect the ability of S100B to act as a diagnostic marker of brain injury in patients with multiple injuries.
Several studies examined the relationship between serum S100B levels and ICI in children with CHT.15–27 Berger et al15 measured the serum concentrations of S100B in 45 hospitalized children with mild, moderate, or severe traumatic brain injury. Serum levels of S100B were increased (>0.39 pg/mL) in almost one half of the children with mild, moderate, or severe inflicted or noninflicted brain injury, compared with control subjects. The increased S100B levels were transient, lasting <12 hours after injury, except for children with severe injury. Similarly, Spinella et al27 examined the difference in serum S100B levels in 136 healthy children and 27 children with traumatic brain injury. Serum S100B levels of ≥2 μg/L demonstrated 86% sensitivity and 95% specificity to predict poor neurologic outcomes after 6 months. The relationships between S100B levels, the time of injury, and the time of venipuncture and the effects of extracranial injuries, such as long-bone fractures, on S100B levels were not evaluated in those studies.
At present, serum biomarkers such as S100B are not used to distinguish, in the ED setting, which children have ICI after CHT. Therefore, we wished to determine whether serum levels of S100B could predict the presence of ICI in children shortly after CHT. We also sought to determine whether extracranial injuries, such as long-bone fractures, had effects on S100B levels in children with CHT and multiple injuries.
All patients <18 years of age who presented to the pediatric ED (PED) of Yale-New Haven Children's Hospital within 6 hours after CHT and, in the judgment of the treating PED physicians, required cranial CT or who were referred from other hospitals for evaluation and treatment of CHT and/or ICI and had undergone cranial CT were eligible for enrollment. Written informed consent was obtained from the parent or guardian. A standardized data entry sheet, recording the medical record number, mechanism and date of injury, initial Glasgow Coma Scale (GCS) score, presence of posttraumatic complaints such as loss of consciousness, headache, and vomiting, presence of ICI or skull fracture as detected through CT, existence of extracranial injuries (eg, long-bone fractures or solid-organ abdominal injuries), and need for inpatient admission or operative intervention for ICI, was used to enroll each patient. For this study, ICI was defined as any contusion or collection of blood within the cranial vault. Patients with skull fractures alone were not considered to have ICI. An attending pediatric radiologist and an attending pediatric neurosurgeon provided the final interpretations of each cranial CT study. If there was a discrepancy between the interpretations of the cranial CT study, then the radiologist's final report regarding the study served as the final interpretation.
For measurement of S100B levels, venous blood samples were obtained from all enrolled patients in the ED within 6 hours after CHT. Blood was allowed to clot for 30 minutes and then was centrifuged (800–1000 rpm for 10 minutes). Serum samples were frozen at −20°C until S100B concentrations were measured by using an enzyme-linked immunoassay (CanAg S100; Fujirebio Diagnostics Majnabbeterminalen, Göteborg, Sweden). The assay requires only 50 μL of serum and can be performed with hemolyzed samples. The assay takes ∼90 minutes to perform and, at the time of this study, cost $9 per patient. Exclusion criteria included history of seizure within 7 days before CHT, penetrating head injury, and history of preexisting developmental delay, encephalopathy, cerebral palsy, or Down syndrome.
We used an independent-sample t test to compare the mean serum S100B levels between children with ICI and children without ICI. In addition, analysis of covariance was used to compare S100B levels between ICI groups with adjustment for age, race, and gender. S100B levels were logarithmically transformed to comply with distributional assumptions; therefore, geometric means were calculated. Geometric means describe the center of a positively skewed distribution more accurately than do arithmetic means and are calculated by exponentiating the arithmetic mean of the logarithmically transformed values. In addition, we constructed a receiver operating characteristic curve to evaluate the S100B level that offered the best combination of sensitivity and specificity to differentiate between subjects with and without ICI. Nonparametric tests (χ2 tests or Fisher's exact tests) were used to describe categorical data. Statistical analyses were performed by using SPSS 13 for Macintosh (SPSS Inc, Chicago, IL).
At the PED of Yale-New Haven Children's Hospital, each year ∼350 children <18 years of age require CT studies for evaluation of CHT. Of those children, 15% have ICI. With a 2-sided significance level of .05, group sizes of 23 (ICI) and 127 (no ICI) would provide 80% power to observe a threefold increase in the level of S100B in the ICI group, compared with the no-ICI group. Enrollment continued until the desired sample size was achieved. The human investigations committee of Yale University School of Medicine approved this study.
Between April 2005 and October 2006, 620 children with CHT who underwent cranial CT within 6 hours after injury were evaluated at the PED of Yale-New Haven Children's Hospital. A convenience sample of 152 children (25%) was enrolled prospectively, including 24 children (16%) with ICI and 128 (84%) without ICI. Most patients (98%) required venipuncture as part of their ED evaluations, and patients were enrolled primarily between the hours of 1 pm and 12 am.
Children in the ICI group were younger than those in the no-ICI group (6.9 vs 9.8 years; P = .01). There were no significant differences between the 2 groups with respect to gender or race (Table 1). Twenty-five children (16%) had long-bone fractures and 6 (4%) had intraabdominal injuries. Significantly more children in the ICI group were referred from other hospitals (48% vs 5%; P = .0001) and had venipuncture performed >120 minutes after injury (62% vs 34%; P = .03) (Table 1).
Injury mechanisms were predominantly falls (42%), being struck by a vehicle as a pedestrian (23%), motor vehicle crashes (13%), sports collisions (9%), and being struck in the head with an object (4%). There were no differences with respect to mechanism of injury between children with ICI and those without ICI (Table 2).
ICI types included subarachnoid hemorrhage (10 patients), subdural hemorrhage (9 patients), intraparenchymal hemorrhage (7 patients), and epidural hemorrhage (5 patients). Seven patients (29%) had >1 type of ICI. Fourteen patients (9.2%) had skull fractures. There were no discrepancies in interpretation of cranial CT studies between the attending pediatric neurosurgeon and the attending pediatric radiologist. Two children with epidural hemorrhage required operative intervention. All children with ICI were discharged from the hospital after hospitalization.
There were no significant differences between the 2 groups with respect to loss of consciousness, vomiting, or amnesia after CHT. There were no significant differences between groups with respect to the presence of long-bone fractures or intraabdominal injuries. Children in the ICI group were more likely to have palpable scalp hematomas (91% vs 51%; P = .001) and GCS scores of <12 (29% vs 2%; P = .001). Children in the no-ICI group were more likely to complain of headache after CHT (48% vs 24%; P = .05) (Table 3).
Because of a positively skewed distribution (Figs 1 and 2), S100B levels were logarithmically transformed and geometric means were calculated. S100B levels were greater in the ICI group (212.9 vs 84.4 ng/L; P = .0001), in children with long-bone fractures (220 vs 83.2 ng/L; P = .001), and in nonwhite children (127.3 vs 80.5 ng/L; P = .03). With controlling for time of venipuncture, long-bone fractures, and race, S100B levels were still greater in the ICI group (278 vs 80.2 ng/L; P = .0001) (Table 4).
The ability of S100B levels to detect ICI, as determined by the area under the curve (AUC), was 0.67 (95% confidence interval: 0.55–0.80) (Fig 3). A serum level of 50 ng/L yielded the best combination of sensitivity, specificity, and positive and negative predictive values; this cutoff value was chosen on the basis of maximization of the Youden index (sensitivity + specificity − 1).28 With this value, the sensitivity was 75%, specificity 56%, positive predictive value 20%, and negative predictive value 90%. When patients with long-bone fractures were excluded, the AUC value was 0.69, sensitivity 73%, specificity 52%, positive predictive value 23%, and negative predictive value 89%.
We wished to determine whether serum S100B levels could predict the presence of ICI after CHT among children in the ED. A serologic assay might help to determine which children with CHT would most likely benefit from cranial CT to exclude ICI. Although we found that mean serum levels of S100B were higher in children with ICI than those without ICI, the ability of S100B levels to detect ICI, determined as the AUC, was only 0.67. Therefore, S100B measurements may not be an accurate screening tool to detect ICI after CHT in children in the ED.
In this study, children with both ICI and long-bone fractures had significantly higher S100B levels than did those without. Our results differ from those of Berger et al,15 who did not find that serum levels of S100B were affected by the presence of long-bone fractures. Our results suggest that extracranial sources of S100B, such as long-bone fractures, affect the specificity of S100B measurements to detect ICI. Our results are similar to those of 2 studies in the adult population that evaluated the relevance of extracranial sources of S100B in the setting of ICI.25, 26 We also found that S100B levels were higher in nonwhite children, regardless of the presence of ICI. To our knowledge, this has not been described previously. These findings would make S100B measurements impractical for screening for ICI in the ED setting, but the results would need to be replicated in a larger study.
Of our sample of 152 patients with CHT, 24 children (16%) demonstrated ICI in CT studies. It might be argued that MRI would reveal a wider ranger of traumatic lesions than CT alone. Therefore, we might have mistakenly categorized children with brain injuries undetectable through CT but potentially visible on MRI scans in the group of children without ICI. If such patients with ICI detectable only through MRI had elevated S100B levels, then such an error would falsely elevate the mean S100B value of the group without ICI detected through CT alone, thus narrowing the statistical difference between the 2 groups. Akhtar et al29 noted such a phenomenon. In the future, perhaps more-widespread use in the ED of “quick brain” MRI, which currently is being used at a few pediatric tertiary care centers to detect ventricular shunt malfunction, would be a more-sensitive modality to detect ICI in children with CHT, without radiation exposure.
In our study, the children with ICI detected through CT were significantly more likely to have GCS scores of <12 in the ED. If S100B levels correlate with GCS scores, then it may be argued that patients with ICI who have high S100B levels would be likely to have worse neurologic prognoses than patients with lower S100B values. Herrmann et al30 demonstrated such a phenomenon. Therefore, S100B levels perhaps could better serve in conjunction with GCS scores as predictive markers of long-term neurologic outcomes after traumatic brain injury.
There are several strengths and weaknesses of this study. The strengths include a large sample size (152 children) and diverse injury mechanisms (including falls, motor vehicle accidents, and sports collisions). Our study improved on previous studies in that we evaluated a larger cohort of patients with CHT, examined the relationship of S100B levels and the presence of extracranial injuries, such as long-bone fractures, and examined the relationship of S100B levels and the time after injury when venipuncture was performed to obtain serum for analysis. Weaknesses include the short half-life of S100B, a characteristic that may make it not useful for screening for ICI in children with CHT. A significantly greater number of our subjects with ICI were from other hospitals, which increased the time between injury and venipuncture, and might have had lower levels of S100B, which might have reduced the ability of S100B measurements to detect ICI in this group of children. Finally, the fact that the study population was enrolled as a convenience sample and the majority of the enrolled patients required venipuncture as part of their ED evaluations limits the ability to generalize the results of this study to a population of patients who require less-extensive ED evaluations for their injuries.
With controlling for time of venipuncture, long-bone fractures, and race, S100B levels were higher in children with ICI than in those without ICI. However, the ability of S100B measurements to detect ICI, determined as the AUC, was only 0.67. This study demonstrates that, although serum S100B levels are higher in children with ICI, measurement of serum levels of S100B may not be a practical screening tool for ICI in children who present to the ED after CHT. The short half-life of the protein and its higher average levels in nonwhite children and in children with long-bone fractures suggest that S100B may not be a sensitive or specific biomarker of ICI in children with CHT, especially those with multiple injuries. Further study of S100B measurement is necessary to determine whether it can serve as a useful adjunct in evaluations of children for ICI after CHT, as a means of reducing the number of CT studies to detect ICI, or can serve as a predictive marker of neurologic outcomes after traumatic brain injury.
Support for this study was through Clinical and Translational Science Awards grant UL1 RR0249139 from the National Center for Research Resources, a component of the National Institutes of Health.
- Accepted May 29, 2009.
- Address correspondence to Kirsten Bechtel, MD, Section of Pediatric Emergency Medicine, Department of Pediatrics, Yale University School of Medicine, 840 Howard Ave, First Floor, New Haven, CT 06504. E-mail:
Financial Disclosure: The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject:
Previous studies evaluated the use of serum S100B levels to detect ICI and to determine prognoses for hospitalized children with traumatic brain injury, but none evaluated their use to detect ICI in children presenting to the ED after CHT.
What This Study Adds:
This study describes the use of serum S100B measurement as a screening tool to detect ICI in children who present to the ED after CHT.
- ↵American Academy of Pediatrics, Committee on Quality Improvement; American Academy of Family Physicians, Commission on Clinical Policies and Research. The management of minor closed head injury in children. Pediatrics.1999;104 (6):1407– 1415
- Homer CJ, Kleinman L. Technical report: minor head injury in children. Pediatrics.1999;104 (6). Available at: www.pediatrics.org/cgi/content/full/104/6/e78
- Schutzman SA, Barnes P, Duhaime AC, et al. Evaluation and management of children younger than two years old with apparently minor head trauma: proposed guidelines. Pediatrics.2001;107 (5):983– 993
- Benz M, Benz MW. Reduction of cancer risk associated with pediatric computed tomography by the development of new technologies. Pediatrics.2004;114 (1):205– 209
- Vos PE, Lamers KJ, Hendriks JC, et al. Glial and neuronal proteins in serum predict outcome after severe traumatic brain injury. Neurology.2004;62 (8):1303– 1310
- ↵Herrmann M, Curio N, Jost S, et al. Release of biochemical markers of damage in neuronal and glial brain tissue is associated with short and long term neuropsychological outcome after traumatic brain injury. J Neurol Neurosurg Psychiatry.2001;70 (1):95– 100
- Copyright © 2009 by the American Academy of Pediatrics