Abstract
Traumatic brain injuries (TBIs) are a large societal and individual burden. In the first year of life, the vast majority of these injuries are the result of inflicted abusive events by a trusted caregiver. Abusive head trauma (AHT) in infants, formerly known as shaken baby syndrome, is the leading cause of inflicted mortality and morbidity in this population. In this review we address clinical diagnosis, symptoms, prognosis, and neuropathology of AHT, emphasizing the burden of repetitive AHT. Next, we consider existing animal models of AHT, and we evaluate key features of an ideal model, highlighting important developmental milestones in children most vulnerable to AHT. We draw on insights from other injury models, such as repetitive, mild TBIs (RmTBIs), post-traumatic epilepsy (PTE), hypoxic-ischemic injuries, and maternal neglect, to speculate on key knowledge gaps and underline important new opportunities in pre-clinical AHT research. Finally, potential treatment options to facilitate healthy development in children following an AHT are considered. Together, this review aims to drive the field toward optimized, well-characterized animal models of AHT, which will allow for greater insight into the underlying neuropathological and neurobehavioral consequences of AHT.
Introduction
Traumatic brain injury (TBI) is the leading cause of death and disability in children worldwide. 1 –5 Unfortunately, a large proportion of head injuries in young children are non-accidental injuries inflicted by trusted caregivers, with approximately 32–38 cases occurring per 100,000 infants. 6 –8 Abusive head trauma (AHT) is the leading cause of child abuse mortality and morbidity, 9 –11 and there is an increased likelihood for death to occur from an AHT compared to an accidental TBI. 12,13 When treating head injuries in hospitals, an estimated two-thirds of children under 2 years of age, 11 and ∼20% of children under 6.5 years of age, are reported to sustain injuries that could be attributed to abuse. 14 These statistics are likely an underestimate due to current challenges with AHT diagnosis, and consequently AHT remains a considerable societal burden.
The first description of caregiver-inflicted AHTs was in 1972, when J. Caffey noted a relation between shaking, subdural hematomas, and retinal hemorrhages. 15 Colloquially, this type of inflicted AHT had been referred to as “shaken baby syndrome,” “whiplash shaken infant syndrome,” “shaken impact syndrome,” “non-accidental head injury,” and “abusive head trauma.” 16 For this review, we will henceforth use the term AHT, as recommended by the American Academy of Pediatrics. 17 AHTs are most common when a child is 3–8 months old, 18 –20 and they occur when the child's body is grasped tightly and vigorously shaken, concomitantly either with or without a blunt impact and skull or intracranial injury. 10,21 This results in rapid acceleration/deceleration forces that overwhelm the child's underdeveloped neck muscles. 22 Infants have a large head weight to body ratio and higher brain water content than adults, which renders them more susceptible to brain injury. 23 The unique properties of infant brains, paired with a lack of parental knowledge of consequences, may culminate in AHT, and in turn persistent neurotrauma, lifelong behavioral disturbances, and retinal damage. 10,11,24
In this review, we provide an overview of the risk factors, symptoms, diagnosis, and manifestations of infant AHT, considering both isolated AHT and the cumulative burden of repetitive injuries. Then, we cover the important features and developmental timelines in human children that need to be accounted for in animal models of AHT. We then discuss current animal models of AHT and draw upon evidence from related models of other injuries that can guide the optimization of a robust model of AHT. We highlight the importance of an AHT model to display long-term deficits that are clinically relevant and worsen with repetitive injuries, closely mimicking the human condition. Finally, we describe treatment options used in more thoroughly examined pathologies, and how these may be translated in the future to ameliorate deficits from AHTs.
Risk factors for AHT
AHTs do not occur uniformly across children. There are numerous factors that increase the vulnerability for a child to experience an AHT, including premature birth, low birth weight, admission to the neonatal intensive care unit (NICU), sex, lower socioeconomic status (SES), or colic. 9,19,25 –28 Children born premature, with low birth weight, or who are admitted to the NICU, are more commonly subjected to AHT. 28,29 These children may exhibit decreased responsiveness to parental affection, leading to increased parental frustration and decreased parent-infant bonding, overall increasing the chance of an AHT. 15,27,29 –31 Male children are more commonly diagnosed with AHT than females. 9,25,27,32,33 Female brains develop slightly faster than males, 34 which may offer some degree of protection from a shaking injury and could therefore result in decreased reporting and clinical manifestations. However, it is important to consider that males are more likely to be born premature with low birth weight, 34 further increasing their vulnerability to experience a diagnosed AHT.
It has also been reported that AHT is more common in cohorts with a lower SES, possibly due to lower parental education level. 9,28,35 However, there may be reporting biases in AHT prevalence, as AHT is more likely to be misdiagnosed if both parents live with the child, when compared to families where the parents are separated. Further, AHT is more likely to be missed if the child is Caucasian compared to racial minorities. 6 Colic, or inconsolable crying, produces considerable frustration and can then result in AHT. Indeed, AHT incidences are tightly correlated with the ages where children generally cry the most. 19 Other precipitating factors for AHT including maternal mental health, having a child with disabilities, and fetal alcohol syndrome are outside the scope of this review, but are described elsewhere. 27,36
Symptoms and prognosis of AHT
Abusive head injuries are typically more severe than accidental head injuries. AHT is associated with a longer average stay in the hospital and a shocking six-fold increase in mortality. 13,14,37 Children are more likely to require neurosurgical intervention and experience seizures when the head injury is due to abuse as compared to accidental trauma. 12,38 Further, children who initially present with seizures have a worse prognosis than those who were seizure-free. 39 Interestingly, prognosis after AHT is worse in children with family instability, lower SES, more severe initial symptoms, or, importantly, the presence of an existing or prior inflicted brain injury. 40 This suggests that there are many factors that modulate AHT severity, pathology, and behavioral sequalae.
Known survivors of AHT have a considerable morbidity associated with the injury. 25,41 Although up to 25% of children exhibit a “symptom-free interval,” whereby they present as normal at hospital discharge 42 or at a 3-month follow-up, 25 this period may be followed by symptom onset and manifestation. 43 Specifically, it has been reported that global outcomes, fine motor skills, behavioral disabilities, and expressive language all decline as children age after an AHT. 41,44,45 It can take up to 4 months to see a reduction in brain growth, 6–12 months for the presentation of central nervous system (CNS) lesions, and up to 6 years for behavioral and social deficits to become apparent. 43 This highlights the presence of unknown latent periods, between injury and symptom presentation, that may occur in many children subjected to AHT, often resulting in delayed administration of treatment. The protracted manifestation of symptomatology highlights the progressive nature of this condition, where the pathology is intricately linked to the interaction between AHT-induced injury and ongoing brain development.
Regarding visual deficits, 40–65% of children are reported to have abnormal vision after an AHT. 25,41,42,46 Visual deficits can include visual agnosia, heterotopia, cortical blindness, or visual acuity deficits. 25,41 Children may also display speech and language deficits (48–64%), motor deficits (45–60%), sleep disorders (17–24%), attention deficits (28–79%), and behavioral disorders (48–53%). 25,41,46 Mental functioning is also perturbed, with children who are victims of AHT having decreased intelligence quotients, developmental quotients, cognitive development, working memory, mental organization, and adaptive behavior. 41,47 Together, it is clear that a wide range of aberrant developmental outcomes are associated with an AHT, with most children who are victims of AHT experiencing lifelong pathological sequelae, and few children left unaffected. Thus, further research is needed to better understand the underlying neuropathology, and to identify treatment strategies appropriate to mitigate the development of neurodevelopmental disorders.
Mild and repeated AHT
Mild traumatic brain injuries (colloquially known as concussions) account for 70–90% of all TBIs. 1 Although mild TBIs have historically been considered benign, it is now recognized that a single mild TBI can have severe psychological, behavioral, and neuropathological consequences that are amplified with each subsequent head injury (see reviews by Bailes and colleagues 48 and Howlett and associates 49 ). Similarly, the effects of mild, and especially, repeat AHT may result in short- and long-term disabilities that are never attributed to the undiagnosed AHT.
Recently, there has been interest in milder, undiagnosed AHT and how it likely precipitates long-term social, behavioral, and neuropathological deficits. The clinical pathology and consequences of a mild AHT are increasingly recognized, even when a child initially presents as “normal,” and the likelihood the AHT will go undiagnosed is increased if the child is awake, alert, and breathing normally when they present to the hospital. 24 Further occluding diagnosis, there may be a delay in bringing the infant to the hospital, and by the time an examination is completed, the acute symptoms may have resolved. 24,50 However, as noted above, approximately one-quarter of children who are initially symptom free will later demonstrate developmental disabilities. 43 There is a clear need for the development of more sensitive diagnostic tools and a greater understanding of symptomatology following mild AHT to aid in earlier and more complete detection of injuries sustained in infancy.
Misdiagnosis of AHT increases the risk for further abuse and AHTs. After a hospital admission following an AHT, if the child is returned home with their caregiver there is a 31–55% chance that the child will be subjected to further abuse. 51,52 Caregivers have reported that shaking behavior is repeated because it results in cessation of crying and the baby going to sleep, 50 reinforcing the abusive behavior. Indeed, in cases of admitted AHT, the primary reason given is frustration, with the amount of shaking force directly correlated to the degree of frustration. 50,53 Supporting this, the age window with the highest rates of AHT co-occurs with the highest incidence of child crying. 19
Judicial reports indicate that children typically receive from 2 to 30 bouts of AHT, and that shaking incidences are commonly habitual, occurring daily for weeks or months. 50 More recently, 3.9% of caregivers admitted to having smothered or shaken their child at least once in the past month. 54 Further, if a mother admitted to shaking her child in the first month after birth, there was a five-fold increase in likelihood she would repeat the behavior before the child is 6 months old. 55 An estimated 4/5 deaths could have been prevented with an accurate initial diagnosis; 73% of children diagnosed with an AHT had a preceding emergency department consultation or prior shaking event. 6,56 Further, 48% of children diagnosed with an AHT have healing fractures and/or skin lesions. 57 Overall, this indicates that AHTs are often not an isolated, singular event in a child's life, but a repeated abusive event that could potentiate the pathological sequelae. Using appropriate animal models will improve our understanding of the neuropathology, objective biomarkers, and possible treatments targeting repetitive mild AHT.
Diagnosis and Pathology of AHT
Clinical diagnosis of AHT
Diagnosis of an AHT when a child presents to the hospital is currently suboptimal, with an estimated one-third of cases either missed or misdiagnosed. 6 Given this solemn statistic, there has been a considerable effort by medical professionals to develop clear diagnostic tools. Symptoms of an AHT are non-specific, and may include nausea, lethargy, and poor feeding, which could also indicate a number of other issues and occlude diagnosis. 58 Physical findings of trauma are only present in around one-third of cases. 59,60 Children may present with apnea or abnormal breathing, although caregivers may report that the child suddenly turned blue and limp, or they just found the child unconscious. 61 In cases of confirmed AHT, the presence of a stroke is indicative of worse outcomes following injury, and should be considered when grading the injury severity. 62 Regarding milder AHTs, caregivers may not bring the child to the hospital if there are no overt, immediate consequences, precluding diagnosis altogether.
Nonetheless, there is a recognized “triad” of hallmark pathological characteristics for AHT diagnosis, which include subdural hemorrhaging, retinal hemorrhaging, and encephalopathy. 36,63,64 Hymel and colleagues developed the PediBIRN-7 in 2019, a screening tool comprising seven variables that reportedly diagnose AHT compared to accidental head trauma with high sensitivity and specificity in pediatric intensive care unit settings. 65 Electronic health records have also been utilized and screened to activate a triggering alert system. 66 These 30 triggers have been validated in emergency department settings in children less than 2 years old and resulted in the development of the child abuse clinical decision support system (CA-CDSS). 66 –68 This is a highly specific and sensitive tool that leads to high clinical guideline compliance for suspected child abuse in children; however, it is not currently implemented universally. 68 Fluid and imaging biomarkers may also provide diagnostic value, 69 but there is still a need for reliable clinical markers of AHT. At present, there is no single test to diagnose an AHT upon presentation to the hospital, and there is an inability to discern if it is the first—or possibly the most severe—of several AHTs that a child has experienced.
Currently, it is recommended that ophthalmic imaging be used as a supportive component in AHT diagnosis, and many techniques can be performed by experienced eye care clinicians including pediatric ophthalmologists, 70 or in some cases pediatric optometrists. Optical coherence tomography (OCT) has been invaluable in detecting retinal pathology and providing diagnostic and prognostic value in children with AHT. 70 –72 OCT is a rapid, non-invasive imaging modality that allows examination of the retina and optic nerve thickness and can provide an in-depth view of retinal layers. Several case studies have detected retinal pathology that can be related to eye function following an AHT. 73,74 For example, vision loss was recorded in most, but not all, eyes with traumatic retinoschisis following an AHT. 73 Although OCT aids in the diagnosis of suspected AHT in children, there is still a need for further analysis of AHT retinal pathology detected by OCT in both human children and animal models of AHT.
Magnetic resonance imaging (MRI), computed tomography (CT), and diffusion tension imaging (DTI) have been invaluable in facilitating diagnosis of AHT. However, there are still no objective imaging criteria capable of discerning AHT across patients. 61 MRI and CT scans provide different diagnostic information, with MRIs being more sensitive to hypoxic-ischemic injuries, 75 but poor at identifying acute hemorrhages. 76 In contrast, CT scans can detect hemorrhages but are poor at identifying shear-strain injuries and ischemic edema. 77 CT scans also detect unilateral or bilateral hemispheric hypodensity arising from acute subdural hematomas, termed “big black brain.” 78 –80 Big black brain is commonly identified in children with an AHT and can be detected as early as 3 hours after injury or may evolve over several days following injury. 81 Big black brain indicates widespread cerebral damage and poor prognosis. 78 –80,82
When examining images generated through MRI for children less than 2 years of age, ischemia in multiple brain regions is strongly predictive of abuse. 83 It is worth noting that imaging modalities of mild TBI in adults are generally unremarkable, despite known insults (e.g., from sports-related impacts), suggesting that diagnosing AHT using MRI or CT scans may have similar sensitivity limitations. 23,84 DTI studies reported differences in fractional anisotropy and apparent diffusion in adolescents after a mild TBI that is related to the severity of post-concussion symptomatology. 85 DTI allows for early detection of ischemic injuries, days before conventional CT scans detect damage. 86,87 Mixed clinical outcomes have been associated with DTI metrics. For example, poor outcomes have been linked to reduced axial diffusivity and these brain changes were further exacerbated as the severity of the injury increased. 87,88 However, another study indicated that DTI metrics were affected in children with an AHT, especially in those with retinal hemorrhages, but these were not correlated to visual acuity. 89
Post-mortem neuropathology in human AHT victims
Post-mortem examinations of children following a diagnosed case of AHT can aid in better understanding of injury pathophysiology, to identify potential clinical biomarkers and therapeutics to reduce morbidity for future cases of suspected AHT. Autopsies may help discern the neuropathology and progression of injuries and identify markers of prior trauma in children who succumb to their injuries. For example, mortality in children after AHT was associated with a high incidence of post-injury seizures, and presentation in the emergency department with cardiopulmonary arrest. 60
Perhaps the most informative autopsy study to date was conducted over 2 decades ago, to examine microscopic and macroscopic injuries in 53 children with an AHT 61,90 (Fig. 1). The most striking finding was that 82% of children died from raised intracranial pressure due to uncontrolled brain swelling. 61 The most common macroscopic injuries were skull fractures, subdural hemorrhages, and retinal hemorrhages, 61 in line with what is seen in children who survive AHT, 91 whereas the most prevalent microscopic injury was global hypoxic damage. 90 In children less than 2 years of age there is a strong association between hypoxic ischemic injury and AHT injury and morbidity. 92 The authors made a distinction between diffuse and vascular axonal damage, but still observed axonal damage in a combined 45% of children, with an additional 21% of children having axonal injury in the cervical spinal cord. 61,90 When distinguishing between injuries in younger infants (ages 2–3 months) and children (age older than 1 year), younger infants had subdural hemorrhages and skull fractures without extracranial injuries. 61
Abusive head trauma (AHT) in clinical, post-mortem, and animal models. Effects of AHT and the presence of its key features in clinical settings, autopsies, and animal models. Figure generated using Biorender.com.
Further, according to reports from doctors or paramedics prior to the autopsy, it was more common for younger infants to present with apnea, which was hypothesized to be driven by axonal injury at the craniocervical junction. 61,90 The older cohort of children, however, typically presented with severe extracranial injuries, particularly to the abdomen; larger subdural hemorrhages; and white matter damage. 61 These findings demonstrate that differential microscopic and macroscopic pathologies and peripheral injuries in children based on age may be of diagnostic value and guide examinations in hospital settings. Further, these pathologies should also be specifically recapitulated when developing animal models of AHT to ensure high clinical relevance of pre-clinical findings.
Differentiating accidental from abusive head trauma
There are several aspects of AHT discussed in section one, such as delayed presentation of symptoms, mechanism of injury, and the highly repetitive nature of AHTs that often aid in the differentiation between a child with accidental head trauma and one exposed to AHT. In addition to those discussed above, there is strong evidence that the manifestation of epilepsy, and retinal and orbital pathology, provides valuable insight into whether an infant was exposed to AHTs versus accidental head injury. Alarmingly, seizures are a very common outcome following AHT. 25,41 Regardless of the presence or absence of an initial immediate seizure following the traumatic incident, 93 there is a distinct possibility that the child may go on to develop post-traumatic epilepsy (PTE). PTE is defined as seizure development at least 1 week following a TBI, with the epilepsy initially provoked by the TBI. 94
Epileptogenesis occurs in three phases: a primary insult; a secondary latent period in which there are various biochemical, structural, and inflammatory changes with possible short bouts of epileptiform activity; followed finally by chronic, unprovoked seizure presentation. 93 Compared to the general incidence of PTE in pediatric patients of 10%, 95 AHT can lead to epilepsy in 20–48% of children, 25,39,46 with the majority of these cases being intractable. 25 Risk of epilepsy following a pediatric TBI is elevated by the presence of a subdural hemorrhage, 95,96 which has particular relevance when considering the high likelihood of subdural hemorrhaging following an AHT. 61
It is possible that there is subclinical epileptiform activity in children exposed to AHT, especially those with subdural hemorrhaging, that develops into epilepsy later in life. It is also plausible that individuals with idiopathic epilepsy may have been exposed to AHT incidents in childhood that were mild, undiagnosed, and/or unreported. A laboratory study by Malik and colleagues found that the proportion of rats that experience electrographic seizures following mTBI increased as the number of insults increased. 97 Therefore, children presenting with seizures after an AHT may have had more than one AHT incident prior to hospitalization, and the cumulative effects of several insults precipitated in seizure activity. Seizures and epilepsy are not currently used as diagnostic criteria, but their utility—especially when considering the general incidence of epilepsy following a pediatric accidental TBI—injury severity, and injury frequency, should be further examined.
In cases of suspected AHT, detection of retinal hemorrhages may have predictive value in the extent of brain damage and overall recovery. 98 Comparison of orbital pathology in children with an accidental head injury compared to AHT revealed significantly greater orbital tissue injury, optic nerve sheath hemorrhages, and intradural hemorrhages in children with an AHT. 99 Similarly, post-mortem analysis of the eyes of children who were victims of AHT reported that optic nerve sheath hemorrhages and retinal hemorrhages were ubiquitous. 100 Retinoschisis, where the retina separates from the back of the eye, may occur via acceleration/deceleration forces causing the retina to detach. 72,101 Retinoschisis does not present in every case of AHT, but is it rare in non-abusive instances, making it a highly specific indicator. 101,102
Animal models of TBIs
Considerations for animal models of AHT
Although AHT has been recognized as an important clinical problem for several decades, there remains a relative paucity of pre-clinical modeling to investigate the intricacies of brain injuries within this realm. Simply modeling pediatric TBI is not sufficient for understanding the pathophysiology of AHT. AHTs are generally associated with concomitant signs of prior abuse and bilateral brain trauma, and are often associated with protracted symptom presentation. 103,104 Moreover, age is an important variable, given that a majority of AHTs in human children occur before 8 months of age. 18,19 Combining these concepts, AHT should primarily be administered to very young animals, whereas the scope of pediatric TBI encompasses a larger developmental window. This idea is discussed further in the section regarding other pediatric animal models of relevance to AHT.
Despite its high prevalence and association with childhood mortality, there have been few attempts to create an animal model that fully recapitulates the important features of AHT (Fig. 1; see Table 1 for a summary of the current literature). Although acute neuropathology has been investigated with these models, the behavioral deficits that often plague children throughout development have only undergone cursory investigations. For example, one model in 3- to 7-day-old piglets reported cognitive dysfunction related to white matter injury at 8 and 11 days following injury. 105 Studies in animal models that have examined long-term deficits into adulthood indicate increased anxiety and memory deficits 106,107 ; however, these have not been replicated nor fully explored. Importantly, none of these models have been able to fully examine differences in injury severity and mortality from a single, severe AHT and multiple, milder AHTs.
Prior Models of AHT by Year of Model Development in Ascending Order
ACTH, Adrenocorticotropic hormone; AHT, abusive head trauma; EEG, electroencephalogram; HPA, hypothalamic-pituitary-adrenal; NMDA, N-methyl-D-aspartate; NMDAR, N-methyl-D-aspartate receptor.
Currently, as recently reviewed in depth elsewhere, 69 there is no consensus regarding the best methodology to induce an experimental AHT. However, there are several important considerations to account for when creating an animal model for any disease or insult, particularly for an injury that occurs during development. The timing and nature of different neurological developmental milestones can all influence the consequences of AHT, including the timing of white matter growth and myelination, biomechanics of an AHT in relation to brain growth spurt and skull formation, brain–immune system, blood–brain barrier (BBB) development, and brain and eye vascular development.
For example, diffuse and traumatic axonal injury is less common in AHT, leading to studies hypothesizing that unmyelinated axons may be more resistant to axonal damage. 61,90,108,109 Myelination occurs at the fastest rate in human children between 2 and 3 years old 110 and from p20 to 23 in mice. 111 Although the proportion of unmyelinated axons may alter vulnerability to axonal pathology, white matter injuries still occur in children with an AHT. 61,90,108 Injuries are induced in AHT rodent models before peak myelination, at a developmental period comparable to when AHT is commonly sustained in children, and this may result in less axonal damage than would be expected in fully myelinated axons. 61,108,109
White matter injuries may be better recapitulated and studied in gyrencephalic animals such as pigs and lambs that have closer structural homology to humans than rodents (see review by Murray and colleagues 112 ). Indeed, p3–5 piglets exhibit white matter injury, swollen axons, and axonal damage in TBI models involving head rotation on a single plane, 113 –115 and diffuse axonal injury and white matter damage has been reported in p7–10 lambs following an AHT induced by shaking. 116,117 Similar to human physiology, lambs have higher brain-water content, and have large brain movement within the skull, 118,119 but the lamb cranium is in line with the neck, whereas in the human this axis is perpendicular (see review by Vester and colleagues 120 ). Equivocal white matter development in an animal model to replicate axonal and white matter injury is imperative in AHT model development.
In human development, the peak growth spurt occurs before birth, between gestational week 36 and 40 (Fig. 2). 118 In rodents, this growth spurt occurs post-natally, between p7 and 10. 121 This therefore creates a discrepancy, as pre-clinical models may be administering insults in different windows of development, thereby misattributing changes in brain growth and volume over time. Skull development, cranial bone ossification, and suture integrity may be important factors when considering coup and countrecoup injuries related to how the brain moves in the head in response to whiplash motions, as well as the skull's ability to absorb the forces associated with blunt force impacts. 122 Pediatric skulls are more compliant and have more diffuse brain tissue distortion than adult skulls. 123 In human children, the fontanel closes at around 18 months, 124 correlating to an age where an AHT is more prevalent. In rodents, a fontanel is not present, but the skull sutures fuse between p7 and 12. 125 In the human infant, the head encompasses up to one-third of the entire weight of the child, 23,126 whereas a rat pup's brain weight reaches half of the adult weight by 7 days old. 127,128 Due to the elastic nature of an infant skull, impulsive loading via shaking produces minimal deformation of the skull, but rather leads to shear strains between the dura and subdural vessels. 23,127
Human and rodent pre- and post-natal development. Developmental timelines in human gestational week (GW) and post-natal month, and rodent embryonic day (ED) and post-natal day (P) in relation to peak age for an abusive head trauma (AHT; indicated in red). Figure generated using Biorender.com.
To reduce variability many animal models constrain the head, reducing the biomechanical relevance of the model; however, models in lambs allow the head and neck to move freely, leading to axial, lateral, and rotational movement of the head around the neck. 116,117,129 Taken together, species, age, greater skull flexibility, brain pre-growth spurts, and AHT injury biomechanics should be considered when developing animal models of AHT, particularly with an understanding of how each would affect clinical translation.
As stated earlier, acute subdural hemorrhages are one of the most common findings in post-mortem brains of children with an AHT, 61,90 often culminating in big black brain. 80,130 Only one animal model to date, performed in anaesthetized piglets, has been able to replicate the big black brain phenomenon. 130 Due to the invariable prevalence of CNS hemorrhage, 21,61,90 it is important to consider the development of vasculature and angiogenesis in animal models and how this may influence the observed neuropathology. The brain receives metabolites through the circulatory system and requires greater levels of metabolites to support the high levels of synaptogenesis and growth occurring early in brain development. 131 In line with drastic brain expansion and growth in humans, cerebral metabolism increases between the third trimester and the first 4 years of life, after which it slowly declines. 132 Hemorrhaging disrupts blood flow, which therefore alters delivery of metabolites and functional development of the vascular system.
Further, vascular beds and capillaries are expanding over the first few years of life, and angiogenesis occurs through short-distance sprouting, which is then pruned or maintained to reach adult levels by the early teen years. 133 Angiogenesis peaks shortly before birth in humans, at approximately 35 weeks of gestation, 134 whereas in mice the peak is shortly after birth. 135 Angiogenesis is guided by hypoxia, with a period of hypoxia preceding vascular sprouting. 136 Therefore, hypoxic-ischemic injuries induced by AHT may restructure angiogenesis to target injured areas. Notably, all animal models to date have indicated some form of subdural hemorrhaging (see Table 1), suggesting that this is a universally recognized feature to model in AHT. However, the longer-term consequences of hemorrhage and how subsequent angiogenesis could be reorganizing the developing vasculature, possibly leading to hyper- or hypoperfusion and the manifestation of behavioral deficits, has not yet been considered.
As noted above, retinal hemorrhages are a common clinical feature used to aid in the diagnosis of AHT. AHT models in p8 mice, 137,138 p7–10 lambs, 116,117 and p3–5 piglets 113 have all been found to exhibit retinal and/or ocular hemorrhages and damage. Although they are part of the CNS, eyes have vasculature and angiogenesis that develops separately from the brain. In human children, eyes do not reach complete vascularization until the first 3 months of post-natal life. 139 In human children, the average age of initial AHT is between 2.2 and 7.85 months, 18,20 when vascularization is primarily complete. By contrast, rodent eyes develop post-natally and eyelids are not open until p15, whereas vascularization is not complete until a few days prior to this at approximately p13. 140 However, retinal hemorrhages may be detected in animal models even when injuries occur before vascularization is complete (see Table 1 for ages of AHT). When using pre-clinical models aimed at distinguishing diagnostic factors of AHT using retinal hemorrhages, or treating hemorrhages to aid in normal retinal function, these differences in developmental timelines may prove important and should be considered. It remains unclear how the developmental stage of the retinal vascular system may influence AHT pathology and modulate the development of visual deficits.
The BBB is a highly selective, semi-permeable barrier that only allows specific molecules into the CNS under homeostatic conditions. BBB development occurs in parallel to angiogenesis and exhibits many adult functions early in fetal development. 141,142 The neonatal BBB differs from that in adults in that it still allows for increased metabolite delivery to support the developing brain. 141,143 Although the BBB develops quickly in post-natal rodents, it is still more vulnerable to injury at p7 compared to p21. 144 As such, infant rodents have perturbed BBB function following TBI, 145 hypoxic-ischemic injury, 146 and AHT. 147
In human children, the brain immune system is developed at birth 148 ; however, it does not reach adult immune system functionality until 7–8 years of age. 149 Further, microglia are the primary immune cells of the brain, and in human fetuses, immature microglia start developing as early as gestational week 5.5, 150 and have adult morphology by gestational week 35. 151 In contrast, the mouse immune system develops post-natally, with an increase in responsiveness and antigen development through the first 1–2 weeks of life. 152,153 Microglia develop from the yolk sac in the early embryonic period in mice but are not mature until p28–30. 154,155 It has been hypothesized that there is a “window of susceptibility” in which the developing brain is more sensitive to inflammatory stimuli and more vulnerable to TBI than in adulthood, and as such cannot be treated in the same way. 156,157 The caveat that the brain's immune system and BBB develop at different times and rates should be considered when deciding on appropriate ages to induce an AHT in an animal model, to accurately reflect relevant human developmental windows.
Finally, many animal models of AHT require anesthesia during induction of the insult. Although this may be beneficial for minimizing the pain and distress associated with the procedure, and for creating a reproducible injury, it may inadvertently have extensive implications for data interpretation. Primarily, anesthesia lacks translational relevance as children are awake, conscious, and likely distressed when subjected to an AHT. 19,27 From a biomechanical perspective, anesthesia prevents muscle tone from resisting whiplash motions in the neck. As stated above, seizures are a common outcome following human AHT, 25,41 and general anesthesia is known to prevent electrographic seizure activity. 158 Any long-term effects associated with initial seizure activity would therefore be prevented.
Pre-clinical evidence reports either a therapeutic benefit of commonly used anesthetics to the injured brain 159 or exacerbation of neurodegeneration, 160 with additional mixed effects reported for seizure outcomes. 161,162 Additionally, anesthesia protects from the effects of global ischemia. 163 By incorporating anesthesia in pre-clinical modeling, past studies may have missed substantial confounding pathogenic phenomena that could alter injury manifestation, reducing the translatability to human children injured by AHT. However, ethical limitations exist when inducing an AHT in gyrencephalic animal models without the use of anesthesia. As such, the use of anesthetics must be carefully balanced against species of choice in the animal model.
An ideal animal model for AHT
In an ideal world, creating an ethical animal model would take into consideration all the above variables, to provide robust and reproducible results that directly mimic the human scenario, for rapid clinical translation of findings (Fig. 1). This hypothetical AHT model would occur without anesthesia, be in a gyrencephalic animal, and create a range of injuries including mortality in a subset of individuals, concurrent bone fractures, retinal hemorrhages, and spontaneous PTE, alongside persisting behavioral and emotional impairments (as elaborated on below). Use of a gyrencephalic species with a freely moving head around the neck most accurately mimics whiplash motion associated with AHT, generating diffuse axonal injury within white matter tracts, and creating blood pooling at the craniocervical junction. However, in rodent, lamb, and pig models, a difference in the orientation and neck anatomy exists when comparing them to bipedal humans. There are also logistical considerations (e.g., limitations regarding housing, costs, technical expertise, use of anesthesia, and ethics) that factor into the use of large animals, which limit their use, and a lissencephalic species may need to be utilized instead.
Animal models should ideally create a range of injury severities from an AHT (or repeated AHT). Just as with concussion, there could be a large cohort of children with subclinical AHTs that do not require hospitalization. 27,42 If an AHT occurs without immediate overt consequences requiring hospitalization and a diagnosis, children could go on to develop a range of issues that are never linked to the AHT. AHT predisposes children to lifelong disabilities, and modeling several degrees of severity and frequency of AHT to examine the effects of these insults into adulthood will give insight to the mechanisms behind various degrees of pathology. Having several instances of an AHT in a single animal, with a progression of deficits, would provide invaluable insight into the compounding effects of repetitive abuse. As a further consideration, given that mortality is estimated in one-third of children after AHT, 25 animal models should correspondingly have some degree of associated mortality, whether after a single severe injury or after several milder injuries. Interestingly, one study in rats indicated a more severe phenotype from repetitive injuries compared to a single injury, but only saw mortality following the first injury, not in subsequent injuries. 164
Additionally, given the high rates of co-occurrence between physical abuse and childhood neglect, 165 we advocate for study designs that examine the role of neglect in outcome manifestation following AHT. Ideally, future studies should also aim to investigate concomitant peripheral injuries such as rib fractures, which co-occur with AHT in about a quarter of patients. 25 Finally, animal models should produce a range of acute and chronic behavioral impairments, which worsen with time and age, as identified in human populations. 25,41,45 It may be improbable to address all these considerations in a single experiment or model, but developing, characterizing, and expanding upon currently relevant models will help generate a more robust understanding of AHT in human children.
Other pediatric animal models of relevance to AHT
Although there is a dearth of research on AHT in animal models, we can also consider other relevant injury models to extrapolate and better understand AHT pathophysiology, comorbidities, and possible treatment strategies. In this section, we turn to pediatric models of TBI, PTE, hypoxic-ischemic encephalopathy (HIE), and maternal neglect to provide substantial new insight in guiding future AHT research (Table 2).
Models That Can Be Extrapolated to AHT
AHT, abusive head trauma; BBB, blood–brain barrier; HIE, hypoxic-ischemic encephalopathy; HPA, hypothalamic-pituitary-adrenal; NICU, neonatal intensive care unit; PTE, post-traumatic epilepsy; TBI, traumatic brain injury.
First and foremost, AHT is considered a closed head TBI. Fundamental differences between an AHT and other models of pediatric TBI are the age and primary mechanism of injury. In this discussion, an AHT must occur in very young animals, include whiplash acceleration/deceleration injuries, and there must be no penetrating injury to the skull. AHT also frequently occurs with hypoxia, whereas a pediatric TBI may not. Models of hypoxia and the relevance to AHT are discussed in the following paragraphs.
In AHT, the brain moving within the skull results in neuropathology: social, cognitive, and motor deficits similar to adult and adolescent TBIs sustained through different mechanisms, such as sports-related concussion. TBIs sustained in early childhood result in cortical thinning and volume loss in white matter tracts that can be correlated to behavioral dysfunction. 166,167 It appears that the neurometabolic cascade following an AHT is similar to that following a mild TBI. 36 In fact, following repetitive, mild TBI (RmTBI) there is an elevated immune response 168 that is similar to findings from repeated bouts of AHT. 147
Although single mTBIs in juvenile rats were not linked to reduced brain weight, 169 RmTBIs have been shown to drastically reduce cortical thickness and increase ventricle volume, 170,171 indicating cumulative insults result in more severe neuropathology. Additionally, a blunt impact injury directly to the exposed skull in p11 rats only culminated in significant behavioral and pathological deficits when repetitive injuries were administered. 164 Considering the likelihood that infants are subjected to more than one AHT incident, it is important to consider that these prior insults can prime the brain to lead to greater pathology and deficits. Experimental models have demonstrated that concomitant extracranial injuries can modulate neuroinflammation after TBI, and result in increased mortality and worse outcomes. 172 This combined insult exacerbation likely also occurs in the context of AHT.
When considering the brain specifically, healthy brain development is associated with an overall reduction in gray matter volume and an increase in white matter throughout the entire brain. 173 Increased myelination, particularly in the prefrontal cortex, is associated with increased attention, motor skills, and memory with age. 173,174 However, following moderate and severe TBI in children, there is an exacerbated reduction of gray matter and/or a decrease in white matter, with atrophy corresponding to behavioral and emotional impairments. 175 Further, pre-clinical evidence has suggested that the extent of progressive neurodegeneration after a TBI in the developing brain is enhanced compared to the adult brain, with a greater degree of cortical volume being lost over time after injury in pediatric mice compared to adult injured mice. 176 Additionally, persistent atrophy of white matter tracts, despite brain adaptation and maturation, is indicated following pediatric TBI. 166 Such findings highlight the complex interplay between post-injury neurobiology and ongoing neurodevelopment and emphasize the necessity for examination across multiple time-points to appropriately characterize AHT outcomes. For example, in the acute period, post-AHT mice exhibited a reduction in cortical white matter. 137 However, this was not examined at chronic time-points, nor considered relative to behavioral and emotional impairments.
As noted above, epilepsy following an AHT should be considered a type of PTE, and having an AHT in childhood may facilitate the development of a seizure disorder later in life. 95,97 Supporting this, in pre-clinical studies, severe TBIs induced with controlled cortical impact in early development lowered seizure threshold in adulthood, which was mediated by inflammation. 177,178 Therefore, a single AHT could similarly prime the brain for seizures and PTE. Epileptiform activity increases the morbidity associated with AHTs. 38,39 Addressing and treating seizures is important as studies in TBI models reveal that behavioral outcomes and rehabilitation is worse when the TBI induces a seizure compared to a TBI alone. 97,179,180 PTE is especially relevant when considering that some children have worse developmental outcomes over time 25,41 and a lack of seizures may be indicative of better recovery, but this is yet to be examined in animal models. However, treating seizures and PTE may not be feasible as a proportion of PTE is intractable, 179 and medications to date treat the seizure, not the underlying pathology resulting in epilepsy.
Expanding on this, sudden unexpected death from epilepsy (SUDEP) is an exclusionary diagnosis used when a person with epilepsy dies with no other known cause. 181 Although not currently explored, it may be of relevance to consider adding prior AHT to the exclusionary criteria as these insults may lead to lasting brain damage, epilepsy, and SUDEP. Overall, early diagnosis of epileptiform activity and prevention of further injuries may aid in recovery following AHTs, yet neither clinical studies nor pre-clinical models have investigated and characterized the relationship between AHT and PTE.
For many children, AHT often results in ischemic brain injuries. 61,90 Therefore, research from the disciplines of pediatric stroke and HIE, induced most often in rats using the Rice-Vannucci model (combination of carotid artery occlusion and systemic hypoxia), 182 can provide invaluable insight into the pathophysiology, outcomes, and treatments for AHT. At the time when AHTs are most prevalent (when a child is younger than age 1 year), 18,42 the BBB is still developing, and may have increased permeability and decreased resistance to injury. 183,184 However, hemorrhages and changes in BBB permeability can still result in alterations of cerebral oxygen delivery. Both diffuse and vascular axonal injury are reported similarly in autopsies of children with HIE and AHT, 90,109 suggesting that hypoxic-ischemic injury may be manifesting axonal damage in AHT. Regarding pathology, hypoxic-ischemic injuries in mice lead to dysfunction in working memory and learning in early adolescence. 185 Expanding on this, hypoxic-ischemic injuries in mice are detectable via MRI and these can be directly related to behavioral deficits measured in adulthood. 186 In addition, for children who experience post-traumatic seizure activity, the severity of the hypoxic-ischemic injury is correlated to the severity of the seizure. 187 However, it is unclear whether the degree of hypoxic-ischemic damage corresponds to the presence or severity of seizures and the behavioral disabilities following AHT.
Further, HIE can also result in cerebral palsy, a term used to describe a complex life-long group of disorders associated with disabilities in motor function due to a brain abnormality. 144,188 –190 AHT has similarly been implicated in cerebral palsy, with one report of children at a cerebral palsy center describing the disorder being induced by abuse in 9% of cerebral palsied children; however, 9% of children in this study were reported to have had cerebral palsy prior to being abused. 191 Although the evidence of correlation is weak at present, a possible causal relationship between HIE or AHT and cerebral palsy cannot be ignored.
AHT commonly co-occurs with other forms of childhood maltreatment. 64 Hence, a child with a missed diagnosis of AHT has a 30% chance of being subjected to further abuse. 6,24,64 In fact, childhood maltreatment is the strongest predictor of adulthood psychiatric disorders. 192 –194 Maternal neglect has been implicated in many negative behavioral and emotional outcomes. 195 –198 Similarly, early life maternal deprivation in animal models leads to increased despair and depression-like behaviors, 197,198 not just in the offspring but also in future generations via epigenetic modifications. 197 Conversely, rats with higher maternal care have advanced adaptive reactions to stress and decreased anxiety. 199 –201 As such, maternal neglect or stress may compound AHT pathology and neurobehavioral impairments. The risk for developing behavioral consequences after AHT may be exacerbated when considering children admitted to the NICU are at a higher risk for AHT, 35 and this forces early life separation between the parents/caregivers and child. Considering this, it may be important to consider the pathologies associated with maternal neglect, and the long-term and intergenerational effects of AHT.
Interventions and Treatments for AHT
Current interventions for AHT in human children
To date there are minimal treatment options for children diagnosed with AHT. Although there have been public health measures designed to prevent AHT before it occurs, these have had limited success at reducing incidence of AHT. 202 –204 In fact, there has been an increase in both the prevalence and mortality from AHT during the COVID-19 pandemic. 205,206 Although this clearly indicates that there continues to be a dire need for education and prevention, there must be examination of treatment options for children who, despite preventative measures, are still subjected to one or several AHTs through early development.
In-hospital treatment options for an AHT are dependent upon the severity of the abuse. The goal is to support maintenance of the child's airway, breathing, and circulation. 64 Treatments for AHT depend on individual symptom presentation for each child and are aimed at reducing secondary brain injuries. 64 Although these treatments may be effective for reducing primary and secondary brain damage, the long-term behavioral and neuropathological consequences of AHT also need to be addressed and treated, rather than focusing on individual symptoms and overt brain damage. Further, treatments for mild and/or repetitive AHT, which may still result in undiscernible neuropathology and behavioral deficits, have not been addressed in pre-clinical or clinical settings.
Potential treatments in animal models of AHT
Importantly, few pre-clinical studies have examined preventative strategies to eliminate or attenuate the myriad neurological and behavioral consequences from AHT, and only one of these models examined the long-term sequelae that may be relevant in human children. Following a particularly severe and repetitive AHT injury in a rat model, Tanaka and colleagues reported increased anxiety-like behavior and stress reactivity associated with dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis. 107 Younger rats were more susceptible to behavioral deficits in adulthood than their older counterparts 107 ; however, treatments to rescue these deficits were not explored.
In other studies, acute treatment with antioxidants has proven efficacious for reducing hemorrhaging and free radicals, 128 whereas administration of a glutamate receptor antagonist was effective at preventing primary damage and white matter thinning, 137,207 although this approach was found to exacerbate secondary damage in animal models of AHT. 207 This is unsurprising, as glutamate receptor antagonists in adults after TBI have failed to afford neuroprotection, and clinical trials in this space have largely been terminated early, possibly due to acute post-injury neurotoxicity. 208 Interestingly, antibiotics such as minocycline to target aberrant microglial activation following early life brain injury have also failed to prevent brain damage, and indeed caused additional behavioral deficits in rats with an AHT. 209 Finally, folic acid has shown promise as a potential therapeutic. When given to piglets after a single AHT, folic acid resulted in a transient reduction in motor and learning deficits, but this was not investigated beyond 6 days following injury. 210 In summary, there are currently no treatments showing promising indications to move to clinical trials for the amelioration of deficits in children after a severe AHT, and none targeting mild, subclinical AHTs.
Potential treatments for AHT from other relevant models
Given the lifelong burden of infant AHT, there is a dire need for effective therapeutic strategies. The above-mentioned animal models of TBI, PTE, HIE, and maternal neglect can guide AHT treatment and pre-clinical trials. Here we highlight how these prior models can guide the development of new treatment options for children following AHT. For example, following a moderate TBI in p21 mice, axonal damage and white matter pathology leading to behavioral deficits was attenuated by promoting myelination and supporting neuron health. 211 In RmTBIs in adult mice, treatment with tert-Butylhydroquinone and pioglitazone to reduce oxidative stress and neuroinflammation has been reported to synergistically reduce gene dysregulation and improve behavioral outcomes. 212 Administration of the anti-inflammatory compound anatabine 3 months following RmTBIs also improves cognitive deficits. 213 Although similar therapeutic avenues have yet to be explored in children and animal models of AHT, these findings support the pursuit of similar therapeutic strategies for AHT. An important treatment consideration is the delay between the AHT and symptom onset. As stated above, brain and behavioral abnormalities can take months to years to manifest. 25,41 –45 This is of relevance as many treatments are unable to be administered after an unknown, and possibly extended, period of time. Therefore, strategies such as environmental enrichment and oxytocin may be optimal as they demonstrate efficacy when delivered beyond the immediate post-injury time-point. 214,215
In the context of severe TBI in adult animals, environmental enrichment comprising cognitive and sensorimotor stimuli has been shown in several independent studies to attenuate histopathological damage and improve cognitive and motor impairments. 214,216,217 Similarly, social and environmental enrichment improved memory and anxiety-like behaviors following a moderate-severe TBI in pediatric mice. 218 However, enrichment in the context of AHT may be particularly difficult to implement, as it may necessitate change of caregivers, which would require significant ethical and legal considerations. There is a single case report demonstrating improvements in speech and language following a change to the infant's caregiver; however, no broad conclusions can be made from this. 25
Yet, although environmental enrichment may not be plausible following AHT, pharmacological agents to mimic the benefits of a supportive environment may hold promise for AHT treatment. For example, oxytocin, sometimes referred to as the “hormone of attachment,” is correlated to parent-infant bonding in humans. 219 As evidence of a key role for this hormone in normal development following adverse childhood events, a study involving blocking of oxytocin receptors in rats exposed to maternal separation was found to increase anxiety and decrease resilience, and this effect was rescued when oxytocin receptors were activated. 220 Further, a pediatric TBI in rats caused a reduction in social recognition that was accompanied by a reduction in inhibitory post-synaptic potentials for up to 8 weeks following injury. 215 Administering oxytocin immediately before behavioral testing rescued both of these deficits and highlighted the therapeutic efficacy of oxytocin with a delayed treatment paradigm. 215
Considering the detrimental effects of seizures following TBIs, and more specifically AHTs, treating immediate seizures should also be addressed. However, the development of PTE needs to also be considered separately, as the presence of an early seizure does not result in an increased risk for developing PTE, 179 nor does treatment of early seizures prevent PTE. Acute seizures and chronic PTE are distinct entities with varying pathogenesis and thus, anti-seizure effects cannot be equated to anti-epileptogenesis. 221,222 Unfortunately, there are currently no available treatments for pediatric or adult patients with epilepsy following a TBI. 222 Further, although prophylactic medications may reduce early seizures in adults, similar effects are not seen in children. 223
One promising field of research regarding therapeutics for early life brain injuries is HIE. HIE at birth is treated with therapeutic hypothermia, which has been shown to prevent excitotoxicity, cell death, and overall mortality. 224 Regarding pediatric TBIs, therapeutic hypothermia has offered inconsistent results, and may improve outcome scores, but failed to decrease mortality, secondary brain damage, or hospital admission length. 225,226 However, therapeutic hypothermia has not been trialed in animal models as an effective treatment for AHTs and would need to be administered immediately post-injury to have beneficial effects. Additionally, oxytocin also offers neuroprotection in pre-clinical models of HIE, 227 and pediatric TBIs, 220 but has yet to be trialed following AHT. Additional post-injury interventions used in other pathologies to treat immediate and long-term deficits may be of value to trial in pre-clinical animal models of AHT.
Conclusions
Despite AHT being the leading cause of death and disability in children and infants, there is a scarcity of research examining diagnosis and treatment strategies, especially regarding repetitive, mild AHT. We argue that, based upon the guidance of more highly studied injuries such as TBI, PTE, HIE, and maternal neglect, appropriate animal models can be developed that adequately replicate the clinically relevant features of AHT. Animal models at a suitable age range to mimic the development of a human child may provide diagnostic insight after early injuries to prevent the repetition of abuse, and they may help to develop clinical interventions to prevent neuropathology and resulting behavioral and social deficits seen chronically in survivors of AHT. Animal models need to be utilized in long-term studies to monitor the progression of neurobehavioral disorders and prevent the development of further deficits. Using insights from other models, strategies to promote white matter integrity, reduce oxidative stress, increase parent-child bonding, or enrich the post-injury environment may be worthwhile pursing in a pre-clinical setting for potential efficacy in this context. We advocate for increased pre-clinical modeling, in particular of mild and repetitive AHT, to drive the field forward toward protecting and treating the most vulnerable proportion of society, children.
Transparency, Rigor, and Reproducibility Summary
No new data were generated in this review. No human or animal subjects were used. A complete list of reviewed articles can be found in the references section. The authors agree to be available for contact to discuss the review and references.
Footnotes
Authors' Contributions
The authors contributed as follows. Sydney Harris: writing original draft (lead) and review and editing; Holly Chinnery: writing, review, editing, and supervision; Bridgette Semple: writing, review, editing, and supervision; Richelle Mychasiuk: writing, review, editing, and supervision.
Funding Information
This work has been supported by funding to Richelle Mychasiuk by the National Health and Medical Research Council (NHMRC) of Australia.
Author Disclosure Statement
No competing financial interests exist.