Experimental models of traumatic brain injury: do we really need to build a better mousetrap?

Approximately 4000 human beings experience a traumatic brain injury each day in the United States ranging in severity from mild to fatal. Improvements in initial management, surgical treatment, and neurointensive care have resulted in a better prognosis for traumatic brain injury patients but, to date, there is no available pharmaceutical treatment with proven efficacy, and prevention is the major protective strategy. Many patients are left with disabling changes in cognition, motor function, and personality. Over the past two decades, a number of experimental laboratories have attempted to develop novel and innovative ways to replicate, in animal models, the different aspects of this heterogenous clinical paradigm to better understand and treat patients after traumatic brain injury. Although several clinically-relevant but different experimental models have been developed to reproduce specific characteristics of human traumatic brain injury, its heterogeneity does not allow one single model to reproduce the entire spectrum of events that may occur. The use of these models has resulted in an increased understanding of the pathophysiology of traumatic brain injury, including changes in molecular and cellular pathways and neurobehavioral outcomes. This review provides an up-to-date and critical analysis of the existing models of traumatic brain injury with a view toward guiding and improving future research endeavors.

Novel aspects of the neuropathology of the vegetative state after blunt head injury.

A detailed neuropathological study was undertaken of the brains of patients who had been assessed clinically as vegetative after blunt head injury. There were 35 cases, (33 male; median age 38 years) with a survival of 6.5-19 months (median 9): 17 were injured in a road traffic accident, 9 after assault and 6 after a fall; 3 were recorded as having had a lucid interval. There was an intracranial hematoma in 9 and the median contusion index was 4; raised intracranial pressure was identified in 25, grades 2 and 3 diffuse traumatic axonal injury was present in 25, ischemic damage in 15 and hydrocephalus in 27. Thalamic and hippocampal damage was present in 28 and stereological studies revealed a differential loss of neurons in three principal nuclei of the thalamus and in different sectors of the hippocampus. Immunohistochemistry provided evidence of an inflammatory reaction and in situ DNA fragmentation, features that are strongly indicative of a continuing neuronal loss in subcortical gray matter. These findings provide evidence for the importance of diffuse brain damage to white matter as the structural basis of the vegetative state after blunt head injury with contributions from neuronal loss in the thalami and the hippocampus. Although amyloid plaques and tau inclusions were identified in some, their contribution did not seem important in the ultimate clinical outcome.

There is differential loss of pyramidal cells from the human hippocampus with survival after blunt head injury.

The experimental literature has shown that neurons within sub-fields of the hippocampus possess differential sensitivities to cell loss after different types of insult to the brain. In humans, after blunt head injury, differential neuronal responses between sub-fields of the hippocampus up to 72 hours after injury have been documented. But, in only a small part of the literature have data for alterations in real numbers of neurons been provided. In this study the hypothesis was tested that, after severe blunt head injury in humans, the total number of neurons within a defined volume of brain tissue differed between different sub-fields of the hippocampus and between groups of patients with differing post-traumatic survivals. Stereological methods were used to measure total cross-sectional area of sub-fields of the hippocampus taken at the level of the lateral geniculate nucleus and count numbers of neurons within each of the CA1, CA2, CA3, and CA4 sub-fields of the hippocampus in patients. The patients used in this study were categorized as follows: Group 1 (early) had survived for 1 week or less; Group 2 (late) survived 6 months or longer after fatal severe head injury; and Group 3 (controls) consisted of age-matched patients that had no history of head injury or disease prior to death. There was a significant loss in cross-sectional area in sub-fields CA3 and CA4 at 1 week or less after injury and in sub-field CA1 at 6 months and greater survival. There was no change in CA2. There was loss of neurons from within a predefined volume of brain tissue in sub-fields CA1, CA3, and CA4 one week or less after injury. But there was no loss in CA2. There was continued loss of neurons from sub-fields CA1 and CA4 between 1 week and 6 months and greater survival, but there was no loss of neurons in sub-fields CA2 and CA3 within the same period. These novel data show that after human severe head injury there is first an acute loss (1 week or less survival) of pyramidal neurons in all hippocampal sub-fields except CA2. Second, there is an ongoing loss of neurons in sub-field CA1 and, most notably, in sub-field CA4, in patients surviving for more than 6 months. However, in neither group of patients is there loss of neurons from sub-field CA2.

Recent advances in neurotrauma.

The frequency of and outcome from acute traumatic brain injury (TBI) in humans are detailed together with a classification of the principal focal and diffuse pathologies, and their mechanisms in extract laboratory models are outlined. Particular emphasis is given to diffuse axonal injury, which is a major determinant of outcome. Cellular and molecular cascades triggered by injury are described with reference to the induction of axolemmal and cytoskeletal abnormalities, necrotic and apoptotic cell death, the role of Ca2+, cytokines and free radicals, and damage to DNA. It is concluded that TBI in humans is heterogeneous, reflecting various pathologies in differing proportions in patients whose genetic background (APOE gene polymorphisms) contributes to the outcome at 6 months. Although considerable progress has been made in the understanding of TBI, much remains to be determined. However, a deeper understanding of the pathophysiological events may lead to the possibility of improving outcome from rational targeted therapy.

Freeze-fracture and cytochemical evidence for structural and functional alteration in the axolemma and myelin sheath of adult guinea pig optic nerve fibers after stretch injury.

Recent work in animal models of human diffuse axonal injury has generated the hypothesis that, rather than there being physical disruption of the axolemma at the time of injury, a pertubation of the membrane occurs, which leads, over time, to a dysfunction of the physiology of the axolemmal. This dysfunction is posited to lead to a disruption of ionic homeostasis within the injured axon, leading to secondary axotomy some hours after the initial insult. We decided to test the hypothesis that membrane pump/ion channel activity or function is compromised and this would be reflected in structural changes within the axolemma and myelin sheath. We used freeze fracture and cytochemical techniques to provide evidence for change in membrane structure and the activity of membrane pumps after nondisruptive axonal injury in the adult guinea pig optic nerve. Within 10 min of injury, structural changes occurred in the distribution and number of intramembranous particles (IMPs) in the internodal axolemma. By 4 h, there was novel labeling for Ca-ATPase membrane pump activity at the same site. There was loss of IMPs from the nodal axolemma extending over several hours after injury. There was loss of both membrane pump Ca-ATPase and p-nitro-phenylphosphatase (p-NPPase) activity of the node. There was loss of ecto-Ca-ATPase activity but increased labeling for p-NPPase activity at sites of dissociation of compacted myelin. Quantitative freeze-fracture demonstrated statistically significant changes in membrane structure. We provide support for the hypothesis that structural and functional changes occur in the axolemma and myelin sheath at nondisruptive axonal injury.

Axonal cytoskeletal responses to nondisruptive axonal injury and the short-term effects of posttraumatic hypothermia.

In human diffuse axonal injury (DAI), axons are exposed to transient tensile strain. Over the ensuing several hours, injured axons enter a "pathological cascade" of events that lead to secondary axotomy. Use of animal models of traumatic axonal injury (TAI) has allowed description of a number of pathological changes before axotomy occurs, including structural and functional changes in the axolemma, disorientation, and/or loss of microtubules, either compaction and/or dispersion of neurofilaments together with focal compaction at sites where continuity of the axolemma is lost. Recent literature suggests that use of hypothermia may improve behavioral outcomes or reduce the number/density of injured axons in which axonal transport is altered after TAI. But there is presently no ultrastructural, pathological explanation as to how hypothermia may act at the level of the axon to reduce posttraumatic loss of axoplasmic transport. In this study, we tested the hypothesis that posttraumatic hypothermia may ameliorate (a) alteration of axonal transport and (b) early pathological changes in the axonal cytoskeleton prior to secondary axotomy. We have undertaken a pilot study within 4 h of stretch injury to adult guinea pig optic nerve axons as a model of TAI and applied stereological techniques to assess differences in pathology in animals either maintained at 37.5 degrees C or cooled to 32-32.5 degrees C for 2 or 4 h after injury. We provide quantitative evidence that posttraumatic hypothermia significantly reduces the number of axons labelled for beta-APP, a marker for disruption of fast axonal transport, and reduces the loss of microtubules and compaction of neurofilaments, which occurs in normothermic animals over the first 4 h after injury.

Axonal cytoskeletal changes after nondisruptive axonal injury. II. Intermediate sized axons.

Earlier studies of axonal cytoskeletal responses to stretch injury in the guinea pig optic nerve, a model of nondisruptive axonal injury such as occurs in human diffuse axonal injury, have demonstrated different cytoskeletal responses between the smallest and largest axons. But these form only approximately 3% of the total number of axons in the optic nerve. It was then posited that the pathology described in the latter axons may not be representative of the pathology in the majority of axons after stretch injury. In order to test this hypothesis, we carried out a quantitative, morphological analysis of structural changes in the cytoskeleton of intermediate (axonal diameter of 0.5-2.0 mM) sized axons at 4 h after stretch injury. Neurofilaments in axons up to 1.00 microm in diameter increased in number and in axons up to 1.50 microm diameter were compacted. This did not occur in larger axons (diameter of 1.51-2.00 microm) in the present study. However, there was focal compaction of neurofilaments in some of the larger fibers at sites where the integrity of the axolemma was lost. The response by microtubules to stretch injury differed from that of neurofilaments in that there was an increased spacing between microtubules and a loss of their number in axons of >1.51 microm diameter. We provide quantitative, morphological evidence (a) that the neurofilamentous cytoskeleton of different sized axons responds in different ways to stretch and (b) that the response by microtubules differs from that of neurofilaments.

Reproducible peracute glutamate-induced focal lesions of the normal rat brain using microdialysis.

The neurotoxic effect of the excitatory amino acid neurotransmitter glutamate was first demonstrated 20 years ago, but the recent development of potent glutamate antagonist drugs with effects against ischaemic damage in vivo and their introduction in clinical studies has made 'excitotoxicity' a major focus of current interest. Despite this, the factors influencing glutamate neurotoxicity in vivo are poorly understood, and the role of glutamate as a neurotoxin in vivo is contested. By using a microdialysis probe to deliver glutamate to the normal rat cortex, we have devised a reproducible model of peracute 'excitotoxic' damage. We have demonstrated that concentrations of over 20 mM glutamate in the perfusate kill neurons in the intact brain in less than 90 min -20 to 200 times more than that required for toxicity in mixed cell culture. The histological and ultrastructural features of the glutamate lesion are very similar to those of acute ischaemia, although their development is much more rapid after glutamate. True extracellular glutamate concentrations estimated from microdialysis studies (about 4 mM) are not far from our results. The reproducible quantifiable nature of the glutamate lesions in this model make it well suited to study the factors affecting the excitotoxic process in vivo.

Loss of axonal microtubules and neurofilaments after stretch-injury to guinea pig optic nerve fibers.

Axonal swellings, characterized by focal accumulations of membranous organelles at presumed sites of interrupted axonal transport, occur in diffuse axonal injury (DAI) in human, blunt head injury and in animal models of nondisruptive axonal injury. Membranous organelles are transported by fast axonal transport in association with microtubules. Although loss of microtubules has been documented at levels of injury severe enough to result in permeabilization of the axolemma to tracers such as horseradish peroxidase, there has been no detailed analysis of responses by microtubules in less severe or milder forms of nondisruptive axonal injury. To test the hypothesis that in less severe forms of axonal injury there is a rapid response by axonal microtubules that might provide an explanation for loss of fast axonal transport, we have carried out a morphometric analysis of microtubules in CNS axons after stretch-injury. There is loss of microtubules at nodes of Ranvier with nodal blebs within 15 min of injury, and in internodal axonal swellings between 2 and 4 h. There is a return to control values at nodes of Ranvier by 4 h, and at the internode by 24 h. There is no loss of microtubules at paranodes, although there is a reduction in their density in the first 2 h after injury. The greatest loss of microtubules occurs at sites of axolemma infolding. Hypothetical mechanisms that might lead to this loss resulting in focal disruption of fast axonal transport and the formation of axonal swellings are discussed.

A mechanistic analysis of nondisruptive axonal injury: a review.

Axons are particularly at risk in human diffuse head injury. Use of immunocytochemical labeling techniques has recently demonstrated that axonal injury (AI) and the ensuing reactive axonal change is, probably, more widespread and occurs over a longer posttraumatic time in the injured brain than had previously been appreciated. But the characterization of morphologic or reactive changes occurring after nondisruptive AI has largely been defined from animal models. The comparability of AI in animal models to human diffuse AI (DAI) is discussed and the conclusion drawn that, although animal models allow the analysis of morphologic changes, the spatial distribution within the brain and the time course of reactive axonal change differs to some extent both between species and with the mode of brain injury. Thus, the majority of animal models do not reproduce exactly the extent and time course of AI that occurs in human DAI. Nonetheless, these studies provide good insight into reactive axonal change. In addition, there is developing in the literature considerable variance in the terminology applied to injured axons or nerve fibers. We explain our current understanding of a number of terms now present in the literature and suggest the adoption of a common terminology. Recent work has provided a consensus that reactive axonal change is linked to pertubation of the axolemma resulting in disruption of ionic homeostatic mechanisms within injured nerve fibers. But quantitative data for changes for different ion species is lacking and is required before a better definition of this homeostatic disruption may be provided. Recent studies of responses by the axonal cytoskeleton after nondisruptive AI have demonstrated loss of axonal microtubules over a period up to 24 h after injury. The biochemical mechanisms resulting in loss of microtubules are, hypothetically, mediated both by posttraumatic influx of calcium and activation of calmodulin. This loss results in focal accumulation of membranous organelles in parts of the length of damaged axons where the axonal diameter is greater than normal to form axonal swellings. We distinguish, on morphologic grounds, between axonal swellings and axonal bulbs. There is also a growing consensus regarding responses by neurofilaments after nondisruptive AI. Initially, and rapidly after injury, there is reduced spacing or compaction of neurofilaments. This compaction is stable over at least 6 h and results from the loss or collapse of neurofilament sidearms but retention of the filamentous form of the neurofilaments. We posit that sidearm loss may be mediated either through proteolysis of sidearms via activation of microM calpain or sidearm dephosphorylation via posttraumatic, altered interaction between protein phosphatases and kinase(s), or a combination of these two, after calcium influx, which occurs, at least in part, as a result of changes in the structure and functional state of the axolemma. Evidence for proteolysis of neurofilaments has been obtained recently in the optic nerve stretch injury model and is correlated with disruption of the axolemma. But the earliest posttraumatic interval at which this was obtained was 4 h. Clearly, therefore, no evidence has been obtained to support the hypothesis that there is rapid, posttraumatic proteolysis of the whole axonal cytoskeleton mediated by calpains. Rather, we hypothesize that such proteolysis occurs only when intra-axonal calcium levels allow activation of mM calpain and suggest that such proteolysis, resulting in the loss of the filamentous structure of neurofilaments occurs either when the amount of deformation of the axolemma is so great at the time of injury to result in primary axotomy or, more commonly, is a terminal degenerative change that results in secondary axotomy or disconnection some hours after injury.

Axonal cytoskeletal changes after non-disruptive axonal injury.

In animal models of human diffuse axonal injury, axonal swellings leading to secondary axotomy occur between 2 and 6 h after injury. But, analysis of cytoskeletal changes associated with secondary axotomy has not been undertaken. We have carried out a quantitative analysis of cytoskeletal changes in a model of diffuse axonal injury 4 h after stretch-injury to adult guinea-pig optic nerves. The major site of axonal damage was the middle portion of the nerve. There was a statistically significant increase in the proportion of small axons with a diameter of 0.5 micron and smaller in which there was compaction of neurofilaments. Axons with a diameter greater than 2.0 microns demonstrated an increased spacing between cytoskeletal elements throughout the length of the nerve. However, in the middle segment of the nerve these larger axons demonstrated two different types of response. Either, where periaxonal spaces occurred, there was a reduction in axonal calibre, compaction of neurofilaments but no change in their number, and a loss of microtubules. Or, where intramyelinic spaces occurred there was an increased spacing between neurofilaments and microtubules with a significant loss in the number of both. Longitudinal sections showed foci of compaction of neurofilaments interspersed between regions where axonal structure was apparently normal. Neurofilament compaction was correlated with disruption of the axolemma at these foci present some hours after injury. We suggest that the time course of these axonal cytoskeletal changes after stretch-injury to central axons is shorter than those changes documented to occur during Wallerian degeneration.

Is beta-APP a marker of axonal damage in short-surviving head injury?

beta-Amyloid precursor protein (beta-APP), a normal constituent of neurons which is conveyed by fast axonal transport, has been found to be a useful marker for axonal damage in cases of fatal head injury. Immunocytochemistry for beta-APP is a more sensitive technique for identifying axonal injury than conventional silver impregnation. This study was designed to determine how quickly evidence of axonal damage and bulb formation appears. Using this method a variety of brain areas were studied from 55 patients who died within 24 h of a head injury. Immunocytochemical evidence of axonal injury was first detected after 2 h survival, axonal bulbs were first identified after 3 h survival, and the amount of axonal damage and axonal bulb formation increased the longer the survival time.

A new experimental model of contusion in the rat. Histopathological analysis and temporal patterns of cerebral blood flow disturbances.

The authors have devised a simple reproducible rodent model of focal cortical injury that uses a mechanical suction force applied through intact dura. The time course and pattern of changes in neurons, glia, and microvasculature were investigated using this model. Early traumatic disruption of the blood-brain barrier and hemorrhage do not occur in this model; however, many of the features of human contusion seen with light and electron microscopy are closely reproduced. At the site of injury, early swelling and lucency of neural dendritic processes have been shown to precede an astrocyte response. In the absence of perivascular hemorrhage, delayed perivascular protein leakage and polymorphonuclear infiltration of the damaged cortex occurs, which is suggestive of an acute inflammatory response. Cerebral blood flow (CBF) has been measured using 14C-iodoantipyrine autoradiography at 30 minutes, 4 hours, and 24 hours after induction of negative-pressure injury in rats anesthetized with halothane and in time-matched sham-operated controls. A significant reduction in blood flow in the sensorimotor cortex at the site of the injury was present at 30 minutes, 4 hours, and 24 hours after induction of the lesion, compared to the contralateral cortex (superficial lamina, ipsilateral 50 +/- 7 ml/100 g/minute, contralateral 112 +/- 26 ml/100 g/minute). The CBF was significantly reduced at the ipsilateral entorhinal cortex at 30 minutes postinjury but no significant reduction was demonstrated at later time points. Although marked alterations in CBF occurred in this cortical injury model, the magnitude and duration of the reduction in CBF are not consistent with those necessary for production of ischemic cell damage. These data indicate that this model of cortical injury can be used to examine biomechanical aspects of contusion without domination by ischemic pathophysiology.

Histopathological changes at central nodes of Ranvier after stretch-injury.

While the brain readily deforms when exposed to rotational loads as experienced in violent movements of the head, axons are able only to sustain tensile loads. Two discrete classes of axonal injury have been posited: disruptive axonal injury, where axons are physically torn or fragmented at the time of the insult, and nondisruptive axonal injury, where there is a hypothesised "perturbation" of the axolemma which leads to a cascade of pathobiological changes which result in axotomy over a period between 2 and 24 h after the initial insult. In the latter, it is posited that the node of Ranvier is that part of the axon which is the initial locus of axonal damage/ histopathological change. This paper describes the ultrastructure of nodal blebs, axolemma limited protrusions of the nodal axoplasm into the perinodal space, in which the nodal dense undercoating has been lost and aggregates of membranous profiles occur within the axoplasm. In addition, this paper provides novel data for disruption of the axonal cytoskeleton in nodes where blebs occur within 15 min of stretch-injury. The cytoskeletal disruption is visualised in thin sections as an almost total loss of microtubules together with a reduced density of neurofilaments within the nodal axoplasm. The loss of microtubules is posited to result in a disruption of fast axonal transport which results in the focal accumulation of membranous organelles in adjacent paranodal regions of the axon to form so-called "axonal swellings." Cytochemical and freeze-fracture studies provide evidence for structural reorganisation of the nodal axolemma after stretch-injury, and it is posited that these changes provide a route for uncontrolled influx of calcium which leads to loss of axonal integrity which potentiates axotomy. It is suggested that increased understanding of regulatory mechanisms that control ion channel activity will greatly increase our understanding of responses of neurones to trauma.

Ascorbate attenuates the systemic kainate-induced neurotoxicity in the rat hippocampus.

The neuronal damage induced by systemic administration of kainic acid reproduces the cellular and regional pattern of damage produced by repeated seizures. The ability of kainic acid to induce lipid peroxidation, and the ability of free radical inhibitors to prevent ischaemically-induced cell death, has led us to examine the possible role of free radicals in kainate-induced injury. Ascorbic acid was able to reduce kainate-induced damage of the rat hippocampus, measured by means of the gliotic marker ligand [3H]PK11195. Ascorbate was significantly effective at doses of 30 mg kg-1 and above, with total protection against kainate at 50 mg kg-1. Histologically, ascorbate at 50 mg kg-1 was able to prevent kainate-induced neuronal loss in the hippocampal CA1 and CA3a cell layers. The antioxidant was also effective when administered simultaneously with, or 1 h before the kainate. Protection was also obtained by allopurinol, 175 mg kg-1 and by oxypurinol, 40 mg kg-1. Ascorbate did not modify synaptically evoked potentials or long-term potentiation in hippocampal slices, ruling out any blocking activity at glutamate receptors. It is concluded that the neuronal damage produced by systemically administered kainate involves the formation of free radicals.

Prevention by a purine analogue of kainate-induced neuropathology in rat hippocampus.

Systemic injection of kainic acid produces a characteristic regional and cellular pattern of neuronal loss in the central nervous system by mechanisms which may be relevant to an understanding of neurodegenerative disorders. It has previously been found, by measuring the binding of a glial marker ligand, that analogues of adenosine, such as R-N6-phenylisopropyladenosine (R-PIA), can prevent kainate-induced damage of the hippocampus at doses as low as 10 micrograms/kg, i.p. The use of gliotic markers, however, is open to misinterpretation, and the present work was designed to re-examine purine protection against kainate using histological methods. The results show that R-PIA, at a dose of 25 micrograms/kg i.p. in rats, can protect against the neuronal damage caused by kainate and that this protection could be completely prevented by the simultaneous administration of 1,3-dipropyl-8-cyclopentylxanthine, indicating the involvement of adenosine A1 receptors in the protection.

Cytochemical evidence for redistribution of membrane pump calcium-ATPase and ecto-Ca-ATPase activity, and calcium influx in myelinated nerve fibres of the optic nerve after stretch injury.

There has been controversy for some time as to whether a posttraumatic influx of calcium ions occurs in stretch/nondisruptively injured axons within the central nervous system in both human diffuse axonal injury and a variety of models of such injury. We have used the oxalate/pyroantimonate technique to provide cytochemical evidence in support of such an ionic influx after focal axonal injury to normoxic guinea pig optic nerve axons, a model for human diffuse axonal injury. We present evidence for morphological changes within 15 min of injury where aggregates of pyroantimonate precipitate occur in nodal blebs at nodes of Ranvier, in focal swellings within axonal mitochondria, and at localized sites of separation of myelin lamellae. In parallel with these studies, we have used cytochemical techniques for localization of membrane pump Ca(2+)-ATPase and ecto-Ca-ATPase activity. There is loss of labelling for membrane pump Ca(2+)-ATPase activity on the nodal axolemma, together with loss of ecto-Ca-ATPase from the external aspect of the myelin sheath at sites of focal separation of myelin lamellae. Disruption of myelin lamellae and loss of ecto-Ca-ATPase activity becomes widespread between 1 and 4 h after injury. This is correlated with both infolding and retraction of the axolemma from the internal aspect of the myelin sheath to form periaxonal spaces which are characterized by aggregates of pyroantimonate precipitate, and the development of myelin intrusions into invaginations of the axolemma such that the regular profile of the axon is lost. There is novel labelling of membrane pump Ca(2+)-ATPase on the cytoplasmic aspect of the internodal axolemma between 1 and 4 h after injury. There is loss of an organized axonal cytoskeleton in a proportion of nerve fibres by 4-6 h after injury. We suggest that these changes demonstrate a progressive pathology linked to calcium ion influx after stretch (non-disruptive) axonal injury to optic nerve myelinated fibres. We posit that calcium influx, linked to or correlated with changes in Ca(2+)-ATPase activities, results in dissolution of the axonal cytoskeleton and axotomy between 4 and 6 h after the initial insult to axons.

The nature, distribution and causes of traumatic brain injury.

The identification and interpretation of brain damage resulting from a non-missile head injury is often not easy with the result that the most obvious structural damage identified postmortem may not be the most important in trying to establish clinicopathological correlations. For example patients with a fracture of the skull, quite severe cerebral contusions or a large intracranial haematoma that is successfully treated can make an uneventful and complete recovery if no other types of brain damage are present. However, not infrequently more subtle forms of pathology are present and ones that can only be identified microscopically. A systematic and pragmatic approach through the autopsy is therefore required and one that recognises the need for tissue to be retained in ways that are appropriate for cellular and molecular studies.

Axonal injury: a universal consequence of fatal closed head injury?

beta-Amyloid precursor protein immunostaining has recently been shown to be a reliable method for detecting the damage to axons associated with fatal head injury. In an attempt to compare the efficacy of this technique with conventional histological detection of axonal damage, we have reanalysed sections from a large well-characterised series of head-injured and control patients. The results indicate that the frequency of axonal injury has been vastly underestimated using conventional silver techniques, and that axonal injury may in fact be an almost universal consequence of fatal head injury.