Introduction
Head trauma is a leading cause of death and disability in children and young adults in the United States. , Annually, approximately 3000 deaths, 50,000 hospitalizations, and 650,000 emergency department visits are due to head trauma. The etiology of pediatric head trauma depends, in large part, on the patient’s age. In infants, falls are the most common cause of head trauma, but child abuse is the most common cause of severe head trauma. In older children, head trauma of all severities is most often due to falls or recreational activities. In adolescents, motor vehicle crashes and assault are the most common etiologies of severe head trauma. In all age groups, head trauma rates for males are higher than for females.
Standard definitions for pediatric head trauma do not exist. It is difficult, therefore, to compare studies or perform meta-analyses since the definition of the condition being studied differs so significantly between studies. The most commonly used terms in the literature are head trauma, traumatic brain injury (TBI), head injury, closed head injury, and concussion. In some contexts, TBI refers only to children with head trauma who have a brain injury detected by computed tomography (CT). In other contexts, TBI denotes any child who has had trauma to the head with or without CT evidence of intracranial injury (ICI). In some studies, children with a skull fracture are considered to have a TBI, whereas in others, they are not. Head trauma is the most general term that encompasses all traumatic injuries to the head and is not dependent on CT evidence of ICI. In this chapter, the term head trauma is used to refer to patients with: (1) a history of traumatic injury to the head with or without radiologic evidence of injury, and/or (2) evidence of trauma to the head on either physical examination or by radiologic evaluation, but without any history of trauma. The lack of a requirement for a history of trauma is particularly important in children with head trauma resulting from abuse since caretakers of these children often do not report a history of trauma or are unaware of a traumatic event. The term traumatic brain injury will be used to refer to those patients with head injury in whom there is either a skull fracture or ICI on head CT. Abusive head trauma (AHT) will refer to children with TBI caused by child abuse.
The gold standard for clinical classification of the neurologic status of a patient with head trauma is the Glasgow Coma Scale (GCS) score ( Table 46-1 ), a 15-point scale that evaluates brain injury severity based on motor, eye, and verbal responses. The range of scores is 3 to 15. A GCS score of 8 or less indicates severe head trauma, 9-12 indicates moderate head trauma, and 13-15 indicates mild head trauma. Although it is the gold standard for assessment of injury severity, the GCS score has several limitations. Most importantly, it is relatively insensitive to subtle brain injury. As a result, patients with a GCS score of 15 might still have underlying brain injury that is too subtle to result in a decrease in the GCS but can have long-term sequelae. The accuracy of the GCS score is particularly poor in young children in whom the verbal score is difficult to assign. , In these children the GCS score often underestimates injury severity. Although the infant face scale (IFS) was developed to address this limitation, its use in infants and young children has not yet been validated.
| Assessed Response | Score |
|---|---|
| Eyes Open | |
| Spontaneously | 4 |
| To speech | 3 |
| To pain | 2 |
| No response | 1 |
| Best Verbal Response | |
| Oriented | 5 |
| Confused | 4 |
| Inappropriate words | 3 |
| Incomprehensible sounds | 2 |
| No response | 1 |
| Best Motor Response | |
| Obeys commands | 6 |
| Localizes pain | 5 |
| Withdraws | 4 |
| Abnormal flexion to pain | 3 |
| Abnormal extension to pain | 2 |
| No response | 1 |
Whereas the GCS score is the gold standard for classification of injury severity, the head CT is the gold standard for identification of ICI. Although CT is able to detect acute intracranial hemorrhage, it is much less sensitive to other abnormalities such as punctate hemorrhages, small contusions, or mild diffuse axonal injury. As a result, most children with head trauma have a normal head CT even though some of them might have ICI which cannot be visualized by CT. The significance of this problem, particularly as it relates to outcome prediction after mild head trauma, has become clearer as neuroimaging techniques have become more sensitive for detecting both anatomic and functional abnormalities. Magnetic resonance imaging (MRI), diffusion weighted imaging, magnetic resonance spectroscopy, diffusion tensor imaging, and proton emission technology are all more sensitive than head CT for identification of subtle ICI. These neuroimaging techniques are not currently considered the gold standard for several reasons, including lack of accessibility, cost, the time to complete the tests, and for some tests, the need for sedation. In addition, MRI, the most easily accessible of the alternative imaging techniques, has the important limitation of being less sensitive than head CT for identifying skull fractures and acute hemorrhage.
To address some of the limitations of the GCS and head CT in the areas of diagnosis and assessment of injury severity and outcome prediction, recent research has focused on the use of biomarkers of brain injury. The remainder of this chapter therefore focuses on the use of biomarkers to enhance the current gold standards for evaluating head trauma.
The Use of Biomarkers of Injury in the Field of Pediatrics
The Brain vs. Other Organs
Physicians routinely use biomarkers to diagnose disease, assess disease severity, assist in disease prognosis, and evaluate treatment efficacy in many organs other than the brain ( Table 46-2 ). Injury and/or death to a cell often results in an increased concentration of a given biomarker, caused by either the release of that biomarker from the injured cell (e.g., creatine phosphokinase in patients with myocardial infarction) or the lack of excretion of a normally excreted chemical that results in its accumulation (e.g., blood urea nitrogen in patients with renal failure). Despite the robust biochemical response of the brain to injury ( Figure 46-1 ) and a growing number of publications related to brain biomarkers ( Figure 46-2 ), development of a useful brain biomarker has proved more difficult than the development of biomarkers for other organs. Perhaps the most important reason for the difficulty is the presence of the blood–brain barrier (BBB), which limits the amount of and size of the markers that can cross into the serum.
| Organ | Marker |
|---|---|
| Heart | Troponin, creatine phosphokinase (CPK)-MB fraction |
| Liver | Aspartate aminotransferase (AST), alanine aminotransferase (ALT) alkaline phosphatase, γ-glutamyl transpeptidase (GGPT) |
| Pancreas | Lipase, amylase |
| Muscle | CPK-MM fraction |
| Kidney | Blood urea nitrogen (BUN), creatinine (Cr) |
There are two possible sources for brain biomarkers. The first and most obvious source is the brain. After injury, biomarkers are released from brain tissue and pass into the cerebrospinal fluid (CSF) and serum. The passage from brain to CSF and serum likely occurs through a transiently more permeable BBB, although there is evidence of other non-BBB–related mechanisms of biomarker transport. Limited animal and adult human data suggest an increase in BBB permeability after head trauma, although the increase is likely variable and related to the location, severity, and type of injury.
One of the difficulties in evaluating a brain injury marker that is released from brain tissue is the issue of where to measure its concentration. Since it is not possible to directly measure biomarkers in the brain itself, the next best source would be the CSF, which bathes the brain. CSF is only available, however, in patients with severe head trauma in whom an extraventricular drain is placed for clinical care. Obtaining CSF in other head trauma patients is not routine and could be dangerous because of the possibility of herniation with lumbar puncture if increased intracranial pressure (ICP) is not recognized. In contrast, serum is easily available and routinely collected as part of clinical care after head trauma. One problem with measuring brain markers into the serum is that many of these markers (e.g., interleukins) are not specific to the brain and are also released by other organs. As a result, when these markers cross the BBB into serum, their concentration can be diluted by the background serum concentrations. This has been one of the barriers to identification of brain-derived serum markers.
The second source of brain biomarkers is the serum. In addition to the markers released from the brain after head trauma, there are also multiple signaling molecules and proteins that are released directly into the bloodstream from peripheral organs in response to head trauma. There are two important advantages of markers that are released directly into the serum: they are often present in higher concentrations than markers that cross the BBB and they are often detected sooner after injury. Interest in these peripherally derived markers has been much more recent than the long-standing interest in brain-derived markers. Several recent studies evaluating these peripheral markers have been performed and will be discussed subsequently.
Prior to discussing the potential roles of biomarkers in the care of children with head trauma, it is first helpful to review the biological properties of the brain-derived biomarkers that have been the focus of the published literature.
Candidate Biomarkers of Brain Injury: A 30-Year Odyssey Continues ( Table 46-3 )
Myelin-basic protein (MBP), one of the two most abundant proteins in myelin, was one of the earliest brain biomarkers evaluated. , The few clinical studies of serum MBP suggest that serum MBP is increased only after severe head trauma and/or intracranial hemorrhage. As a result of the small patient population in which it could be applied as well as development of more universally applicable biomarkers, MBP is no longer being actively pursued as a general biomarker of brain injury. In one specific clinical situation, however—the identification of AHT—MBP may be a useful biomarker. Unlike other biomarker concentrations that begin to rise immediately after injury and quickly decrease, serum MBP concentrations do not begin to increase until 24 to 48 hours after injury and remain increased for up to 2 weeks. The late increase is likely related to the association of MBP with traumatic axonal injury; although axonal injury occurs at the time of head trauma, the Wallerian degeneration of the axon and release of MBP takes several days. Because of the late rise and prolonged presence in the serum, MBP might be useful in identifying nonacute intracranial hemorrhage in children in whom the concentrations of other biomarkers have already decreased.
| Biochemical Marker | Abbreviation | Location | Serum Half-life |
|---|---|---|---|
| Myelin-based protein | MBP | Myelin | 12 hr |
| Creatinine phosphokinase—brain-specific fraction | CPK-BB | Brain, lungs | <1 hr |
| Neuron-specific enolase | NSE | Neurons, platelets, red blood cells, neuroendocrine cells | 24 hr |
| S100B | Not applicable | Astrocytes, chondrocytes, adipocytes | <2 hr |
| Glial fibrillary acidic protein | GFAP | Glial cells | ∼1 week |
| Cleaved tau | c-tau | Axons of central nervous system neurons, proteolytically cleaved after release to form c-tau | Unknown |
| α-II spectrin/spectrin breakdown degradation products 120, 145, and 150 | SBDP 120, 145, and 150 | Cortical membrane cytoskeletal protein | Unknown |
| Hyperphosphorylated neurofilament heavy chain | pNFH | Axons | Unknown |
As interest in MBP waned for the reasons discussed above, focus turned to the brain-specific fraction of creatinine phosphokinase (CPK-BB). By the early 1990s, however, CPK was abandoned as a possible brain biomarker because of concern about its sensitivity and specificity. Almost all of the literature since that time has focused on two biomarkers: neuron-specific enolase (NSE) and S100B. NSE is a glycolytic enzyme localized primarily in neuronal cytoplasm, although it is also present in small quantities in platelets and red blood cells. It is a marker of neuronal death and is increased after head trauma of all severities.
S100B is the major low-affinity calcium-binding protein in astrocytes. It is released by astrocytes that die or are irreversibly injured. Since the serum half-life of S100B is less than 100 minutes, increases in serum S100B after head trauma are transient. S100B concentrations can also be increased with significant noncranial injuries (e.g., pelvic fracture). This could limit its usefulness in the setting of multi-organ trauma. An important limitation is the high normative concentrations of S100B in children less than 2 years of age. Because of high baseline concentrations, a single value of S100B cannot be interpreted in young children with possible head trauma. However, serial concentrations in the same patient can provide important information about the progression and/or severity of injury as discussed later in this chapter.
Although the literature has focused on NSE and S100B, several papers have evaluated other brain-derived biomarkers including glial fibrillary acidic protein (GFAP) , and cleaved tau protein (c-tau). , There has also been a recent interest in structural markers of axonal injury, specifically, α-II spectrin and its degradation products, SBDP 120, 145 and 150 and hyperphosphorylated neurofilament heavy chain (pNFH).
Although the number of different markers that have been evaluated is small, new techniques such as gel-based and gel-free proteomics, which can be performed both in the serum and CSF, allow for screening of the entire genome of patients with and without head trauma. These techniques, although technically difficult and markedly expensive, will likely identify additional brain-derived and peripherally derived biomarkers that are more sensitive and specific than those discussed previously. In a recently published feasibility study, Haqqani and colleagues used gel-free proteomics to identify 95 uniquely expressed proteins in six children with severe head trauma compared with healthy adult controls. Although it is not possible to determine which of these proteins are brain-derived and which are peripherally derived, the source of the markers and even the physiological explanation for their increase or decrease after head trauma is less important than their statistical association with head trauma. Although a significant amount of additional research is needed, these new techniques have the potential to revolutionize the field of biomarker development and validation.
The Potential Role of Serum Brain Biomarkers
For purposes of discussion, the potential roles of biomarkers have been divided into five categories: (1) diagnosis, (2) differentiation of head trauma and TBI, (3) assessment of head trauma severity/outcome prediction, (4) development of treatment interventions, and (5) evaluation of treatment efficacy. It is important to recognize, however, that this division is somewhat arbitrary and that the categories overlap considerably. Furthermore, studies that may focus on one specific role of biomarkers might have implications for future research and/or clinical care in a different category. The remainder of the chapter focuses on these roles in the context of pediatric head trauma with an emphasis on AHT. However, because of the dearth of pediatrics data, it will also be necessary to discuss some of the related adult literature.
Diagnosis of Head Trauma
In patients who can communicate and/or whose head trauma is witnessed, biomarkers are not needed to make a diagnosis. Infants and young children, however, cannot communicate, and since the etiology of their head trauma is often abuse, the history provided sometimes is not accurate. When used in the context of identification of head trauma, biomarkers would serve as a “point to the brain” test, much the same way that liver function tests can direct the treating physician to the liver as a possible source of the patient’s symptoms. When used in this way, serum biomarkers might not provide information about the etiology of brain injury; an infant with hydrocephalus and one with AHT could both present with vomiting and both might have increased biomarkers.
The possible use of biomarkers to identify head trauma in infants and young children who present without a history of trauma and with nonspecific symptoms is the most well-studied application of biomarkers in this context. The clinical dilemma of how to differentiate the vomiting infant with a routine viral illness from the vomiting infant with head trauma presents itself hundreds of thousands of times every year in emergency departments (EDs) and pediatricians’ offices throughout the United States. The problem in this situation is that in the correct context (e.g., when there is a history of or external evidence of trauma), the need for neuroimaging in these infants would be obvious. When there is a lack of history of trauma and/or a lack of findings on physical examination that suggest trauma, however, it is unlikely that a physician would even consider head trauma in the differential diagnosis of a vomiting or fussy infant. In the case of undisclosed head trauma, when the diagnosis of head trauma is not considered and is missed, an infant will be returned to a violent environment where he or she can be reinjured or killed. The magnitude of the problem of missed diagnosis of AHT has been estimated to be at least 30% to 40% based on several studies. In a landmark study by Jenny and colleagues, the authors determined based on a review of previous medical records that 31.2% (54/173) of children diagnosed with AHT in their hospital had been seen previously by a physician after AHT but had not been recognized as having AHT at that time. Remarkably, 27.8% of the AHT patients were reinjured in the time between misdiagnosis and eventual proper diagnosis, and at least four children died as a result of the delay in diagnosis. The high rate of reabuse after misdiagnosis demonstrates the critical importance of accurate and timely diagnosis.
A study by the author and colleagues evaluated the possible use of serum biomarkers to screen for head trauma in high-risk infants. Children were eligible for the study if they were less than 1 year of age, had a temperature less than 38.3°C, and presented to the ED without a history of trauma and with symptoms that placed them at increased risk of having HT: vomiting without diarrhea, an apparent life-threatening event, irritability, fussiness, lethargy, or seizures or seizure-like activity.
Ninety-eight subjects were enrolled. Blood or CSF was collected at enrollment, frozen and batched for later measurement of NSE, S100B and MBP. A head CT was not performed as part of the study protocol, but was at the discretion of the treating physician. Because of the lack of a gold standard head CT in all subjects, all subjects were tracked by chart review and phone until 1 year of age or for 6 months after enrollment, whichever came later, to evaluate for subsequent evidence of possible child abuse. The authors hypothesized that if there was a missed case of AHT, it was likely that the child would be reinjured during the tracking period and re-present for medical care.
Based on the clinical diagnosis given at the time of discharge and follow-up data (i.e., the presence or absence of subsequent abuse), subjects were classified as: (1) AHT (i.e., the diagnosis of AHT was recognized at the time of enrollment), (2) no brain injury (NBI) (i.e., no brain injury was identified at enrollment or follow-up), (3) indeterminate (IND) (i.e., no brain injury was identified at enrollment, but the child was diagnosed with possible abuse during the follow-up period), or (4) “not classified” (i.e.. the child was identified with a brain injury that was not the result of AHT). The “not classified” group highlights the fact that if used as a “point to the brain” test, biomarkers would not provide information about the etiology of the brain injury: an infant with hydrocephalus and one with AHT could both present with vomiting and both might have increased biomarkers. An abnormal biomarker screen would simply suggest to the treating physician that neuroimaging should be considered.
Study subjects were then classified as true positives, false positives, true negatives, and false negatives based on whether biomarker concentrations were in agreement with clinical diagnosis. For example, a true negative would be a patient who was clinically identified as NBI and whose biomarker concentrations were normal. Sensitivity and specificity was calculated based on data from the NBI and AHT subjects only.
Of 98 subjects, 76% were classified as NBI, 14% as AHT, 5% as IND, and 5% as “not classified.” As part of clinical care, 100% of the AHT and 28% of the NBI subjects had a head CT scan performed. Using previously derived cut-offs for abnormal biomarker concentrations, NSE was 76% sensitive and 66% specific and MBP was 36% sensitive and 100% specific for AHT. S100B was increased in 90% of NBI subjects and therefore not specific for head trauma in this population. Of the five subjects who were identified at follow-up as possible victims of abuse (IND group), four had had an increased serum NSE concentration at enrollment, suggesting that these may have been missed cases of AHT. The results of this study suggest that NSE and MBP, but not S100B, might be able to identify children at high risk of AHT who would benefit from evaluation with head CT.
A more recent study measured the concentrations of peripherally derived brain markers after AHT. Using multiplex bead technology to simultaneously screen 45 different serum biomarkers, the author and colleagues compared biomarker concentrations between infants with mild AHT and infants who presented with the same symptoms (e.g., vomiting) but who did not have brain injury. There were significant group differences in the concentrations of 9 of the 45 markers screened: vascular cellular adhesion molecule, interleukin-12, matrix metallopeptidase-9, intracellular adhesion molecule, eotaxin, hepatocyte growth factor, tumor necrosis factor receptor 2, interleukin-6, and fibrinogen. Several of these markers (i.e., matrix metallopeptidase-9, interleukin-6, and fibrinogen) have previously been identified as being increased in the serum after head trauma in adults. It is not possible to determine whether the increase in serum concentrations of the biomarkers was due to increased leakage from brain into blood as a result of BBB dysfunction or as the result of a systemic response to brain injury. Knowing the source of the biomarkers (i.e., brain vs. peripheral), however, is not critical when they are being used in the context of a screening tool, since the goal is merely to identify that brain injury has occurred. It is important to recognize that several of these markers can be increased in the presence of fever , and in a wide variety of pediatric illnesses such as enteroviral meningitis (IL-6), influenza (IL-6), and rotavirus (IL-6, IL-10). As a result, their specificity could be decreased if they are used in certain populations such as infants with febrile seizures or diarrhea.
These two studies suggest that there could be a wide variety of both brain and peripherally derived biomarkers that might be useful for AHT screening. Further research is essential in order to prospectively validate these markers. In addition, a study in which all subjects undergo the gold standard head CT is critical in order to truly assess the sensitivity and specificity of the biomarkers.
Differentiation of Head Trauma and TBI
In the clinical scenario in which biomarkers are used to differentiate head trauma and TBI, the treating physician already recognizes that head trauma might be the source of the patient’s symptoms. The clinical dilemma is whether the possibility that the head CT will identify an ICI outweighs the radiation risk to the patient. The assumption is that a head CT is only necessary if it identifies TBI. Almost all of the literature related to using biomarkers to identify TBI has focused on S100B.
The use of biomarkers to differentiate head trauma from TBI initially focused on addressing a very specific clinical problem: how to differentiate adults with acute alcohol intoxication from those with alcohol intoxication and TBI without needing to perform a head CT on all of them. In one of the first studies to address this specific clinical problem, Mussack and colleagues enrolled 139 subjects during the Munich Oktoberfest in 2000. All subjects underwent head CT and had serum NSE and S100B concentrations measured. S100B concentrations, but not NSE concentrations, were significantly higher in patients with TBI vs. those without TBI. It was possible to identify a cut-off value of S100B that provided 100% sensitivity and 50% specificity for identification of patients with TBI.
A recent multicenter study by the same group in Germany enrolled 1309 adults with mild head trauma. Mild head trauma in this study was defined as a history of head trauma, GCS score 13 to 15, and one or more clinical risk factors that included vomiting, intoxication, and severe headache. All subjects underwent head CT. The subset with ICI on head CT were identified by S100B with 99% sensitivity and 30% specificity using previously derived cut-offs for S100B. The authors concluded that use of S100B could result in a 30% decrease in the use of head CT without missing any cases of ICI. Other adult studies have shown similar results. In a study by Romner and colleagues, the negative predictive value of an undetectable S100B serum level was 0.99, meaning that there was 99% probability that patients with a normal S100B did not have an ICI.
There is only a single pediatric study that has specifically addressed the ability of serum biomarkers to differentiate head trauma from TBI. A pilot study published in 2000 by Fridriksson and colleagues reported on 50 children aged 0 to 18 years with head trauma. Forty-five percent of these children had ICI identified on head CT. In this population, an increased serum NSE concentration was 77% sensitive and 52% specific for ICI. Unfortunately, the authors did not report on the specificity at a sensitivity of 100%, and no follow-up study has been published.
Assessment of the Severity of Head Trauma/Outcome Prediction after Head Trauma
Outcome after head trauma is directly related to the severity of injury. After severe head trauma, outcome prediction is important for family counseling, for selection of patients who are most likely to benefit from rehabilitation services, and for decision-making, particularly as it relates to the extent of care to provide and whether to continue or withdraw care. After mild head trauma, outcome prediction is most important for identification of patients who are at highest risk of having sequelae from their injury and in whom early rehabilitation would be most helpful. Since only 10% to 15% of patients with mild head trauma have sequelae, it is neither possible nor appropriate to offer rehabilitation to all mild head trauma patients. The ability to predict outcome using biomarkers could have important implications for children with AHT. Access to rehabilitation services for children in foster care is notoriously poor, and the ability to identify those children who would benefit most from rehabilitation services would allow concerted effort to be directed toward those children. The ability to quantitatively predict outcome in cases of AHT could also have important legal implications in cases of AHT, since physicians are often asked in court to predict a given child’s eventual outcome.
The adult literature related to biomarkers and outcome prediction is extensive and the interested reader is referred to a review article on the subject. Very briefly, numerous markers have been assessed including NSE, , S100B, , MBP, GFAP, , and cleaved tau protein. Overall, the data demonstrate that higher serum biomarker concentrations are consistently correlated with worse outcome.
The pediatric literature is limited and is listed in its entirety in Table 46-4 . As seen in that table, only a single study specifically focused on children with AHT. This study evaluated the relationship between serum NSE, S100B, and MBP concentrations measured immediately after injury and neurocognitive outcome 6 months after injury in children less than 3 years of age with AHT compared with children less than 3 years of age with non-abusive head trauma. Eligible children with AHT (n = 15) were matched for gender, ethnicity, socioeconomic status, and injury severity with children with non-abusive head trauma. Outcome was assessed using the Glasgow Outcome Scale (GOS) score (1 = good outcome, 5 = death), the Vineland Adaptive Behavior Scale (VABS), and an age-appropriate IQ measure. Biomarker concentrations were measured every 12 hours for up to 5 days after injury. Initial, peak and “time to peak” (i.e., number of hours between the time of injury and the time of the peak biomarker concentration) were calculated for all three biomarkers. In cases of AHT, “time of injury” was defined as the time at which medical care was sought.
