Crooks et al.: Axonal Injury in Closed Head Injury 109
Axonal Injury in Closed Head Injury by Assault: a quantitative study D A CROOKS, MD C L SCHOLTZ, PhD FRCPA G VOWLES, SSe FIMLS S GREENWALD, PhD Department of Morbid Anatomy, The Royal London Hospital, London E1 1BB
S EVANS, MSe FIS Department of Epidemiology and Medical Statistics, The London Hospital Medical College, London E1 1BB
ABSTRACT Due to the controversy in the literature regarding the time course of axonal balloon formation in human material, we wished to determine if it was possible to diagnose axonal injury before the development of axonal balloonings. The hypothesis was that the presence of axonal swellings or axonal enlargements associated with a glial reaction could be used as a diagnostic aid in human axonal injury before 12 hours. The brains of eight individuals that survived for less than 48 hours following head injury, and also had evidence of axonal injury using the criteria ofVanezis et al. (1987), were systematically studied by looking at axonal swellings, axonal balloonings, reactive astrocytes, maximum diameter of axonal enlargements and density of axonal enlargements. Controls were eight selected cases without neurological disease. The variables studied were assessed in 25 fields from ten different areas of the brain, using silver stains and immunoperoxidase for glial fibrillary acidic protein (GFAP). Logarithms of one plus the count of each variable were taken from the raw data and these were analysed using percentile distribution and the median, the t-test, Mann-Whitney U test and the Wilcoxon signed rank test. We conclude that quantitation of axonal damage allows the detection of mild degrees of axonal injury that could be overlooked on routine examination, and that the criteria of axonal enlargements, rather than axonal balloonings, are indications of axonal damage, cannot be endorsed with the evidence provided.
INTRODUCTION It has been suggested from the evidence of experimental trauma that in all forms of closed head injury the basic pathology may involve axonal disruption and that the morphological basis for concussion is associated with definite and describable anatomical alterations (Alves and Jane, 1985).
The diagnosis of diffuse axonal injury (DAI) is of considerable importance to both the general pathologist and the forensic pathologist facing an almost normal looking brain in a patient who has remained unconscious since the moment of injury (Adams et al., 1980; Adams and Graham, 1984). Axonal injury may be the only evidence of a closed head injury, in the absence of macroscopic focal lesions (Adams et al., 1985; Simpson and Berson, 1987; Simpson et al., 1985). Although DAI is a lesion that occurs at the moment of injury (Adams et al., 1977, 1982b), the question of whether the diagnosis could be made during the first 12 hours post-trauma was a matter of controversy for some time. There is evidence that axonal balloonings are the first microscopical evidence of axonal damage (Adams et al., 1977, 1982b), and it is known that they take some time to develop (Adams, 1975). For example, Rand and Courville (1934),
110 Med. Sci. Law (1992) Vol. 32, No.2
Figure 1. Axonal enlargements in the subcortical white matter of the frontal lobe. Palmgren x 40
Figure 2. Axonal swellings in the deep white matter of the parietal lobe. Palmgren x 25
Peerless and Rewcastle (1967) and Imajo and Roessman (1984) were of the opinion that axonal balloonings could be identified by light microscopy at 2--4 hours post-trauma, but there is a remarkable absence of cases in which axonal injury has been diagnosed within 12 hours in larger series (Adams et al., 1977, 1980, 1982b, 1984, 1985; Pilz, 1983). Thus, the diagnosis of DAI in formalin-fixed human material within 12 hours is very difficult, if not impossible (Pitz, 1983; Adams et aI., 1989). Experimental evidence suggests that the development of axonal balloonings in mild head injury begins around 12-24 hours post-trauma (Povlishock et aI., 1983). This delay is reduced in severe head injury (Erb and Povlishock, 1988) where axonal balloonings are seen at around 6 hours post injury. It follows that although axonal damage might have occurred at the moment of injury, DAI cannot be diagnosed because axonal balloonings cannot be seen using light microscopy before some 12 to 18 hours. In an attempt to provide a solution to this problem, Vanezis et al. (1987) examined the brains of a series of patients who had severe head injury and who died within 72 hours without a lucid interval. The study concluded that the features ofDAI at 2, 6 and 10 hours were those of enlarged axons associated with increased GFAP positive
material. The conclusions of Vanezis et al. (1987) were not in accordance with previously reported findings in DAI' Due to the controversy found in the literature regarding the time course of axonal balloon formation in human material, and the importance of their presence in the early diagnosis of DAI as a sine qua non finding in the early phase (Vanezis et aI. 1987 VB Adams et al. 1977, 1982b, 1984, 1985, 1989). It was decided to test the following hypotheses: i)
an increase in axonal diameter associated with an increased GFAP reaction in astrocytes;
ii) alternatively, an increase in the density of axonal enlargements associated with an increased GFAP reaction in astrocytes, are the earliest axonal changes found in axonal damage due to head injury;
iiilaxonal enlargements may increase in number with longer survival in head injury; iv) axonal swellings identified in human material stained with silver impregnations are the forerunners of axonal balloonings. In summary, the hypothesis is that the proposed axonal changes and associated glial reaction could be used as a diagnostic aid in human DAI during a 12-hour survival.
Crooks et al.: Axonal Injury in Closed Head Injury 111
.
•• \1 • •
L~ ,. ••
,
,• •
Figure 3. Retraction bulbs in the splenium of the corpus callosum. Palmgren x 40
Figure 4. Reactive astrocytes in the occipital subcortical white malter. Immunoperoxidase for GFAP x 25
MATERIALS
matter of the parietal lobes were sampled from level P4; the occipital lobes were sampled from P9 and the temporal lobes were sampled from P3. Sections of the mid-brain and the middle and upper thirds of the pons were also prepared. 9~ sections were cut, and subsequently stained with haematoxylin and eosin, Palmgren silver stain, and with immunoperoxidase for GFAP.
Head Injury cases The brains of eight individuals, who survived for less than 48 hours following head trauma during an alleged assault, and who had Diffuse Axonal Injury (DAD, were obtained from the Forensic Department of the London Hospital Medical College. They were immersed in 10% formalin and suspended by the basilary artery for 2 to 6 weeks. The diagnostic criterion for DAI used in these subjects was that which is outlined in Vanezis et al. (1987). The cerebral hemispheres were sliced in the coronal plane (the first cut placed at the mammillary bodies), the cerebellum was cut perpendicular to the folia and the brainstem was sectioned horizontally. Ten blocks in total were taken, to include both sides of the cerebrum, cerebellum and brainstem. The splenium of the corpus callosum was taken from the P4level 1; two slices of the body of the corpus callosum that incorporated independent areas of the sub-cortical and deep white matter of the frontal lobes were sampled from the P2 and P3 levels; the sub-cortical and deep white
Control cases The brains of eight selected patients, each without neurological disease, were autopsied at The Royal London Hospital. They were fixed, sliced, processed and stained as described for the individuals with head injury. METHODS
Silver-stained sections from head injury and control individuals were mixed, and a number assigned to each one at random. Measurements were made without knowing the group to which the section belonged, and the clinicopathological correlation was carried out after assessing all of the silver stained slides.
1 'P'stands for Coronal level 'posterior' to the mammillary bodies.
112 Mad. Sci. Law (1992) Vol. 32, No.2
Astrocytes with large cytoplasm and elongated processes that showed a distinctive nucleus (Figure 4). The MD of 25 enlarged axons was measured and recorded using a crossed-micrometers graticule calibrated with an Agar stage micrometer, in which each division was 2.35 urn using the 40x objective. In addition, AE, AS, AB and RA were estimated in 25 fields in each zone and their density was calculated by dividing their number by the nominal volume of tissue. There was no restriction in the number of swell ings counted per axon, and only axonal balloonings that were in fields free of ischaemia, necrosis or haemorrhage were included in the study. Considering each zone separately, the statistical differences between the head injury and control subjects were analysed with the MannWhitney U (MWU) test, and the comparison of
Definition of the variables studied
Maximum diameter of enlarged axons (MD): The diameter in micrometers (um) ofaxons that appeared approximately twice the width of the surrounding axons Axonal enlargements (AE): Number of enlarged axons estimated in 25 fields in each zone (Figure 1). Only enlarged axons that appeared twice the width of the surrounding fibres were counted. Axonal swellings (AS): Axonal lobulations visualized in silver stains (Figure 2). Axonal balloonings (AE): Argyrophilic structures with a rounded or elliptical shape depending on the plane of section located at the end of a disrupted axon (Figure 3).
Reactive astrocytes (RA).
Tahle I. Head injury: summary of the correlation between all variables studied with the survival time
Variable
p values
r
Maximum diameter
0.021
0.9611
Axonal enlargements
0.043
0.9190
Axonal swellings
0.590
0.1179
Axonal balloonings
0.610
0.1087
Reactive astrocytes
0.410
03157
Pearson correlation coefficient (r) and p values.
Table II. Head injury and control: correlations between the variables studied.
Control
Head Injury Variables
p
p
values
MD vs. AS
-0.018
09623
0.357
0.3447
AE vs. AS
-0.244
0.5185
-0113
0.7648
AS vs. AB
0.506
0.1807
0.625
0.0982
MD vs. RA
-0.286
0.4497
-0.155
0.6822
AE vs. RA 0.208 0.5815 -0.548 .::0~.c=1-=-47,-,4~_ _ Spearman correlation coefficient p (Rho) and p values. MD-Maximum diameter; AS-Axonal swellings; AE-Axonal enlargements; AB-Axonal balloonings; RA-Reactive astrocytes.
Crooks et al.: Axonal Injury in Closed Head Injury 113
zones within each group independently was done with the Wilcoxon signed rank (WSR) test. In every individual, the mean of each zone for each variable studied was averaged to obtain the mean of the means for all zones. Subsequently, differences were analysed between the head injury and control groups using the Mann-Whitney U test. For all zones, correlation was sought between the mean of the means of all variables of each individual, and survival time in the head injury group using the Pearson correlation coefficient (1'). RESULTS Median and percentile distribution
When using the median and percentile distribution for comparing the two groups, the MD showed no significant differences but controls showed consistently higher values; there was little difference between the two groups in AE although the cerebellum and frontal lobe showed the highest densities in the head injuries; in both groups the splenium showed the highest densities for AS but only the head injury group showed AB. The highest densities of RA were found in the midline of the head injury individuals but this pattern was not observed in the control group. In general, the head injury group showed higher densities in all zones for all variables, except the MD and AB. Comparison between the two groups
When using the unpaired t-Test and the MWU test for comparing head injury and control groups, no significant differences were found when the MD and AB were analysed. When comparing AE, only 3 zones showed differences that achieved statistical significance: the frontal deep (p = 0.04) and sub-cortical white matter (p= 0.01) and the occipital subcortical white matter (p = 0.01). When all zones were compared, these differences were also statistically significant (p = 0.016). The AS showed differences of significance only in the rostral brainstem (p = 0.03) and when all zones were compared (p = 0.03), whereas the RA achieved statistical significance in the splenium (p = 0.01), cerebel-
lum (p = 0.03) and occipital sub-cortical white matter (p = 0.02). Comparison between zones within the same group
When using the paired t-Test and the WSR test for comparing the zones within the same group, the head injuries showed no significant differences between the zones when the MD, AE and AB were analysed. However, differences of statistical significance were observed when comparing the AS of the temporal sub-cortical white matter against the splenium (p = 0.04), frontal (p = 0.04) and parietal subcortical white matter (p = 0.04). In addition, significant differences were also found on analysing the RA between the body of the corpus callosum and the frontal deep white matter (p = 0.03), between the splenium of the corpus callosum and the deep parietal white matter (p = .04), and between the occipital sub-cortical white matter and the cerebellum (p = .04). In the control group, no significant differences were found between the zones when comparing MD, AE, AS or AB. It is interesting that differences in RA were statistical significant between the deep and sub-cortical white matter of the frontal lobe (p = 0.04).
Correlation with survival time
There was no significant correlation between any of the variables studied and the survival time. A summary of the correlation coefficient (1') and p values is shown in Table I. In this study, the estimation of survival time is not as accurate as when dealing with experimental material in the laboratory. This leads to difficulties when cases of different survival times are compared, since it takes time for the changes under study to develop. Differing survival times mean that each case will present changes related to the particular post-injury interval after which death took place, and the picture may be obscured when the survival time has been too short for changes to have appeared, or where the time is inaccurately recorded.
114 Mad. Sci. Law (1992) Vol. 32, No.2
Correlation between the variables studied
There was no significant correlation between any of the variables studied in the head injury or control group. A summary of the Spearman correlation co-efficient (p) and p values is shown in Table II DISCUSSION Axonal enlargements and maximum diameter of enlarged axons
The possibility that the increase in axonal diameter is a post-mortem change is suggested by the fact that the control group had consistently higher values for the maximum diameter of enlarged axons than had the head injury group, and controls had longer intervals between death and autopsy. Although the difference in the interval between death and post-mortem was not statistically significant, a significant correlation was found between the axonal enlargements and the interval between death and post-mortem in the control group (p = 0.0027), suggesting that the axonal enlargements increase with longer delays between death and autopsy. In addition, axonal enlargements are not seen in biopsies, although they have been described in post-mortem material (Vanezis at al., 1987; Adams et al., 1989). While Vanezis et al. (1987) considered that axonal enlargements were indicative of early DAI when seen in association with an increase in GF AP reaction, this contention has not been endorsed by Adams et al. (1989), since, in their experience, such axonal enlargements are often present in control specimens. Therefore if patients have survived for 15 to 18 hours they will diagnose DAI only if axonal balloonings can be identified. An increase in the diameter ofaxons as an early change in diffuse axonal injury is not seen in experimental models, and this may be the result of the good tissue preservation of experimental material. This is further evidence that the nature of these enlarged axons is artefactual. Results indicate that measurements taken for the maximum diameter ofaxons in the corpus callosum, deep and sub-cortical white matter of all lobes, rostral brainstem and cerebellum, are within normal limits, because
measurements were within the same range of those of the control group. This means that the maximum axonal diameter cannot be taken as evidence of axonal damage because, with regard to this variable, there is no detectable difference between mechanically traumatized and normal brains. It is difficult to interpret the significant differences that were found between the head-injured and control groups for the density of axonal enlargements. The presence of enlarged axons in normal brains suggests that there is no reason to believe that they are pathological. It is probable that they represent the upper range of normal dimensions. If this is so, the differences in AE between the two groups are not likely to be due to the injury per se. The reasons for these differences are at present obscure. Density of axonal swellings The occurrence of highest densities of axonal swellings in the spleni um of corpus callosum, in both head injury and control groups, makes their suitability as a diagnostic feature of axonal injury questionable, if the corpus callosum is considered independently and no axonal balloonings are identified. Axonal swellings or axonal balloonings can arise in this area when the pericallosal arteries are compressed due to shift and distortion of the brain, which can also produce focal ischaemia or haemorrhage (Lindenberg et al., 1955; Adams et aI., 1989), al though axonal balloonings have been described away from focal lesions in the corpus callosum in true axonal injury (Adams and Graham, 1984). It is interesting that axonal swellings identified in human head injury material using silver techniques have been taken as presumptive evidence of DAI in cases of short survivals, whether axonal balloonings are present or not (Vanezis et al., 1987). This criterion cannot be supported by the findings in the present study. Significant differences between the head injury and control groups were found in the rostral brainstem, although the head injury cases showed axonal swellings in more than one zone. Controls showed axonal swellings in three cases, in a maximum of four different areas, but
Crooks et al.: Axonal Injury in Closed Head Injury 115
no axonal swellings in the rostral brainstem. Axonal swellings were present in the brainstem of three head injury cases; they were absent in three cases and two cases were excluded. From the three cases that showed swellings in the brainstem, the values of density indicated that the brainstem was the third, second and first most severely affected area as assessed by the density of swellings. When the brainstem was the structure most severely affected, there was no involvement of the sub-cortical white matter of the cerebral hemispheres, whereas in the other two cases, axonal balloonings were seen in the brainstem and the sub-cortical white matter in the cerebral hemispheres. It is suggested that the significant differences found in the brainstem between the head injury and control groups may be presumptive evidence which would support a mixture of patterns of structural vulnerability to axonal damage in head injury by assault, where the brainstem can be affected with variable degrees of severity, ranking from the most affected to the least affected, depending on the mechanical parameters. The differences found in the head injury group between the temporal sub-cortical white matter and the other zones arise as the result of there being no swellings in the temporal subcortical white matter in seven cases out of eight. This may be related to the observation that the temporal lobe may not swirl as much as the parietal and occipital lobes in head impacts in non-human primates, using the lucite calvarium as suggested by Pudenz and Shelden (1946). Most probably, relative immobility of the temporal lobe may produce proportionally low shearing strains that will not exceed the shear modulus ofaxons, which has been estimated as 1.4 x 104 Newtons/m2 (estimated by Thibault as quoted by Erb and Povlishock, 1988), hence producing no axonal damage. Density of axonal balloonings Axonal balloonings were present in three cases, all of which survived longer than 12 hours. This is an important finding, since axonal injury is considered not to be strongly associated with head injuries by assault (Adams et al., 1977, 1980, 1984, 1985) although its prominence in
interpersonal violence has been stressed by other investigators (Imajo and Roessman, 1984; Imajo et al., 1987). The absence of a strong correlation between assaults and axonal injury (Adams et al., 1982a, 1982b) has been explained on the grounds of biomechanics (Gennarelli and Thibault, 1982; Adams et al., 1982a), in which DAI is associated with lower rates of deceleration, in contrast to falls and assaults which are associated with short acceleration/deceleration impulses. Attention must be drawn to the fact that these correlations were made at the time when the investigators were concentrating on the most severe end of the spectrum of axonal injury, and although the views held with respect to the definition of DAI have changed, there is no evidence that a reappraisal of milder degrees of axonal injury in assaults has been undertaken. Mild degrees of axonal injury have recently been investigated (Graham et al., 1989). In this study, although the severity ofDAI was correlated with the degree of lucidity, the external causes of the injury were not. The paucity of axonal balloonings in this material is most likely to be the result of the short survival time in half of the cases, although it is uncertain whether there was axonal damage in those that survived less than 12 hours. It must be noted that axonal balloonings were occasionally found in areas away from the midline when none were seen in the parasagittal sub-cortical white matter of the frontal and parietal lobes. Whether or not this is related to the mechanisms of injury operating at the moment of injury remains a matter of interest. Correlation with survival time
Quantitative evidence provided by this study suggests that the maximum diameter ofaxons that appear enlarged with respect to the surrounding axons does not increase with longer survival times (r = 0.021, p = 0.96) as has been suggested formerly by Vanezis et al. (1987). The statistical analysis of the data collected during this investigation demonstrated that the density of axonal enlargements showed no increase or decrease with respect to the survival time (r =0.043, p =0.92), thereby rejecting the
116 Mad. Sci. Law (1992) Vol. 32, No.2
third hypothesis stated in the introduction. The original observations of Vanezis et al. (1987) have been rejected in the present study since the values for the maximum diameter of enlarged axons and the number of axonal enlargements were within the normal limits for all survival times. The density of axonal swellings showed a weak oorrelation with the survival time (r = 0.59, p =0.12). This suggests that more data are needed to establish whether or not there is a relationship between axonal swellings and survival time in human material. The Pearson correlation coefficient showed that the density of axonal balloonings correlated weakly with survival time (r = 0.610, p = 0.11). Again, more data and more cases are needed. Correlation between variables
The data presented here, after the analysis for statistical associations, showed that there was no correlation between the maximum axonal diameter and density of reactive astrocytes (p = 0.286, p = 0.45) suggesting that the first hypothesis stated in the introduction must be rejected. Similarly, no statistical correlation was found between the density of axonal enlargements and density of reactive astrocytes (p = 0.208, p = 0.58), which suggests that the second hypothesis must also be rejected. The Spearman correlation coefficient indicates that no correlation exists between maximum axonal diameter and density of axonal swellings (p = 0.018. p = 0.96), or between axonal enlargements and axonal swellings (p = 0.244, p = 0.52). This statistical evidence suggests that the first and second hypotheses must be rejected, because no association has been found between the axonal enlargements described by Vanezis et al. (1987) and the axonal swellings which are equivalent to those described in silver stains in experimental head injury in the cat (Povlishock et al., 1983; Erb and Povlishock, 1988), and in the non-human primate (Gennarelli et al., 1982; Jane et al., 1985a, 1985b). Similarly, the Spearman coefficient showed no correlation between density of axonal swellings and density of axonal balloonings (p = 0.506, p = 0.18). This lack of correlation is probably due to the paucity of axonal balloon-
ings in the material included in this study, and it suggests that the fourth hypothesis must be rejected, with the proviso that more cases need to be studied. CONCLUSIONS
Areas within the zone that appeared more severely damaged than others may be those in which the dissipation of the kinetic energy transmitted to the brain has generated sufficient force to elicit secondary axonotomy. Changes in axonal diameter, or in the density ofaxons that may appear enlarged with respect to the surrounding fibres, cannot be taken as evidence of axonal damage, rejecting previous contentions postulated by Vanezis et al. (1987). High densities of axonal swellings detected in the corpus callosum of both controls and head-injury cases preclude the utilization of the corpus callosum alone in the diagnosis of axonal injury in closed head injury, since pathology other than trauma can give rise to their formation. More cases need to be studied to assess the correlation between the survival time and axonal swellings, axonal balloonings and reactive astrocytes. No correlation was found between either the maximum axonal diameter or the density of axonal enlargements and the glial reaction, nor between the maximum axonal diameter or the density of axonal enlargements and axonal swellings, nor between axonal swellings and axonal balloonings. The hypothesis suggested by Vanezis et al. (1987) regarding the early diagnosis of DAI using axonal enlargements or swellings rather than axonal balloonings as an indication of axonal damage cannot be endorsed. REFERENCES Adams J. H. (1975) The neuropathy of head injuries. In: Vinken P. J. and Bruyn G. W. (OOs) Handbook of Clinical Neurology 23, Injuries of th.e brain tuul skull, Part I. New York, American Elsevier, 35-65. Adams J. H. and Graham D. 1. (1984). Diffuse brain damage in non-missile head injury. In: Anthony P. P. and MacSween R. N. M. (OOs) Recent Advances in Histopath.ology 12. London, Churchill Livingston, 241-57.
Crooks et al.: Axonal Injury in Closed Head Injury 117
Adams J. H., Mitchell D. E., Graham D.!. and Doyle D. (1977) Diffuse brain damage of immediate impact type. Its relationship to 'primary brainstem damage' in head injury. Brain 100, 489-502. Adams J. H., Graham D. I., Scott G., Parker L. S. and Doyle D. (1980) Brain damage in fatal non-missile head injury, J. Clin. Pathol. 33, 1132-45. Adams J. H., Gennarelli T. A, Graham D.!. (l982a) Brain damage in non-missile head injury: observations in man and sub-human primate. In: Smith W. T. and Cavanagh J. B. (eds) Recent advances in Neuropathology 2. London, Churchill Livingston, 165-90. Adams J. H., Graham D. 1., Murray L. S. and Scott G. (l982b) Diffuse axonal injury due to non-missile head injury in humans: an analysis of 45 cases. Ann. New-ol. 12,557-63. Adams J. H., Graham D. I., Doyle D., Lawrence A E. and McLellan D. R (1984) Diffuse axonal injury in head injuries caused by a fall. Lancet 2, 1420-2. Adams J. H., Doyle D., Graham D. 1., Lawrence A E. and McLellen D.R (1985) Microscopic diffuse axonal injury in cases of head injury. Med. Sci. Law 25,265-9. Adams J. H., Doyle D., Ford 1., Gennarelli T. A, Graham D. 1. and McLellan D. R (1989) Diffuse axonal injury in head injury: definition, diagnosis and grading. Histopathology 15, 49-59. Alves W. M., Jane J. A (1985) Mild brain injury: damage and outcome. In: Becker D. P. and Povlishock J.T. (eds) Central Neroous System Trauma Status Report. Bethesda, Md., National Institute of Neurological and Communicative Disorders and Stroke, 255-70. Erb D. E., Povlishock J. T. (1988) Axonal damage in severe traumatic brain injury: an experimental study in cat. Acta Neuropathol76, 347-58. Gennarelli T. A, Thibault L. E. (1982) Biomechanics of acute subdural haematoma. J. Trauma 22, 680-6. Gennarelli T. A, Thibault L E, Adams J. H., Graham D. 1., Thompson C. J. and Marcincin R P. (1982) Diffuse axonal injury and traumatic coma in the primate. Ann Neurol. 12,564-74. Graham D. 1., Lawrence A E., Adams J. H., Doyle D., McLellen D. and Gennarelli T. A (1989) Pathology of mild injury. In: Hoff J. T., Anderson T. E., Cole T. M. (eds) Mild to Moderate Head Injury. London, Blackwell Scientific Publications, 63-75.
Imajo T, Roessman U. (1984) Diffuse axonal injury. Am. J. Forens. Moo. Palhol. 5, 217-22. Imajo T., Challener R C., Roessmann U. (1987) Diffuse axonal injury by assault. Am. J. Forens. Med. Pathol. 8, 217-19. Jane J., Rimel R W., Pobereskyn L. H. and Dacey R G. (1982) Outcome and pathology of minor head injury. In: Grossman RG.H. (ed) Seminars in Neurological Surgery. New York, Raven Press, 229-37. Jane J. A, Rimel R W., Alves W. M., Dacey R G. Jr, Winn H. Rand Colohan A R (1985a) Minor and moderate head injury: model systems. In: Dacey R G., Winn H. R, Rimel R W, Jane J. A (eds) Seminars in Neurological Surgery. New York, Raven Press, 27-33. Jane J. A, Steward 0., Gennarelli T. A (1985b) Axonal degeneration induced by experimental non-invasive minor head injury. J. Neurosurg, 62, 96-100. Lindenberg R, Fisher R, Durleacher S. H., Lowett W V. Jr and Freytag E. (1955) Lesions ofthe corpus callosum following blunt mechanical trauma to the head. Am. J. Pathol. 31, 297-317. Peerless S. J. and Rewcastle N. B. (1967) Shear injuries of the brain. Can. Med. Ass. J. 96,577-82. Pilz P. (1983) Axonal injury in head injury. Acta Neurochirurg (Suppl) 32, 119-23. Povlishock J. T., Becker D. P., Cheng C. L. Y. and Vaughan G. W. (1983) Axonal changes in minor head injury. J. Neuropath. Exp. Neurol. 42, 22i}42. Pudenz R H. and Shelden C. H. (1946). The 1ucite calvarium - a method for direct observation of the brain II. Cranial trauma and brain movements. J. Neurosurg. 3, 487-505. Rand C. W, Courville C. B. (1934) Histological change in the brain in cases of fatal injury to the head. Arch. Neurol. Psychiatry 31,527-55. Simpson R H. and Berson S. D. (1987) The post-mortem diagnosis of diffuse cerebral injuries, with special reference to the importance of brain fixation. S. Afr. Med. J. 71,10-14. Simpson R H. W, Berson D. S., Shapiro H. A (1985) The diagnosis of diffuse axonal injury in routine autopsy practice. Foren. Sci. Int. 27, 229-35. Vanezis P., Chan K K, Scholtz C. L. (1987) White matter damage following acute head injury. Form. Sci. Int. 35, 1-10.
Axonal Injury in Closed Head Injury by Assault: a quantitative study D A CROOKS, MD C L SCHOLTZ, PhD FRCPA G VOWLES, SSe FIMLS S GREENWALD, PhD Department of Morbid Anatomy, The Royal London Hospital, London E1 1BB
S EVANS, MSe FIS Department of Epidemiology and Medical Statistics, The London Hospital Medical College, London E1 1BB
ABSTRACT Due to the controversy in the literature regarding the time course of axonal balloon formation in human material, we wished to determine if it was possible to diagnose axonal injury before the development of axonal balloonings. The hypothesis was that the presence of axonal swellings or axonal enlargements associated with a glial reaction could be used as a diagnostic aid in human axonal injury before 12 hours. The brains of eight individuals that survived for less than 48 hours following head injury, and also had evidence of axonal injury using the criteria ofVanezis et al. (1987), were systematically studied by looking at axonal swellings, axonal balloonings, reactive astrocytes, maximum diameter of axonal enlargements and density of axonal enlargements. Controls were eight selected cases without neurological disease. The variables studied were assessed in 25 fields from ten different areas of the brain, using silver stains and immunoperoxidase for glial fibrillary acidic protein (GFAP). Logarithms of one plus the count of each variable were taken from the raw data and these were analysed using percentile distribution and the median, the t-test, Mann-Whitney U test and the Wilcoxon signed rank test. We conclude that quantitation of axonal damage allows the detection of mild degrees of axonal injury that could be overlooked on routine examination, and that the criteria of axonal enlargements, rather than axonal balloonings, are indications of axonal damage, cannot be endorsed with the evidence provided.
INTRODUCTION It has been suggested from the evidence of experimental trauma that in all forms of closed head injury the basic pathology may involve axonal disruption and that the morphological basis for concussion is associated with definite and describable anatomical alterations (Alves and Jane, 1985).
The diagnosis of diffuse axonal injury (DAI) is of considerable importance to both the general pathologist and the forensic pathologist facing an almost normal looking brain in a patient who has remained unconscious since the moment of injury (Adams et al., 1980; Adams and Graham, 1984). Axonal injury may be the only evidence of a closed head injury, in the absence of macroscopic focal lesions (Adams et al., 1985; Simpson and Berson, 1987; Simpson et al., 1985). Although DAI is a lesion that occurs at the moment of injury (Adams et al., 1977, 1982b), the question of whether the diagnosis could be made during the first 12 hours post-trauma was a matter of controversy for some time. There is evidence that axonal balloonings are the first microscopical evidence of axonal damage (Adams et al., 1977, 1982b), and it is known that they take some time to develop (Adams, 1975). For example, Rand and Courville (1934),
110 Med. Sci. Law (1992) Vol. 32, No.2
Figure 1. Axonal enlargements in the subcortical white matter of the frontal lobe. Palmgren x 40
Figure 2. Axonal swellings in the deep white matter of the parietal lobe. Palmgren x 25
Peerless and Rewcastle (1967) and Imajo and Roessman (1984) were of the opinion that axonal balloonings could be identified by light microscopy at 2--4 hours post-trauma, but there is a remarkable absence of cases in which axonal injury has been diagnosed within 12 hours in larger series (Adams et al., 1977, 1980, 1982b, 1984, 1985; Pilz, 1983). Thus, the diagnosis of DAI in formalin-fixed human material within 12 hours is very difficult, if not impossible (Pitz, 1983; Adams et aI., 1989). Experimental evidence suggests that the development of axonal balloonings in mild head injury begins around 12-24 hours post-trauma (Povlishock et aI., 1983). This delay is reduced in severe head injury (Erb and Povlishock, 1988) where axonal balloonings are seen at around 6 hours post injury. It follows that although axonal damage might have occurred at the moment of injury, DAI cannot be diagnosed because axonal balloonings cannot be seen using light microscopy before some 12 to 18 hours. In an attempt to provide a solution to this problem, Vanezis et al. (1987) examined the brains of a series of patients who had severe head injury and who died within 72 hours without a lucid interval. The study concluded that the features ofDAI at 2, 6 and 10 hours were those of enlarged axons associated with increased GFAP positive
material. The conclusions of Vanezis et al. (1987) were not in accordance with previously reported findings in DAI' Due to the controversy found in the literature regarding the time course of axonal balloon formation in human material, and the importance of their presence in the early diagnosis of DAI as a sine qua non finding in the early phase (Vanezis et aI. 1987 VB Adams et al. 1977, 1982b, 1984, 1985, 1989). It was decided to test the following hypotheses: i)
an increase in axonal diameter associated with an increased GFAP reaction in astrocytes;
ii) alternatively, an increase in the density of axonal enlargements associated with an increased GFAP reaction in astrocytes, are the earliest axonal changes found in axonal damage due to head injury;
iiilaxonal enlargements may increase in number with longer survival in head injury; iv) axonal swellings identified in human material stained with silver impregnations are the forerunners of axonal balloonings. In summary, the hypothesis is that the proposed axonal changes and associated glial reaction could be used as a diagnostic aid in human DAI during a 12-hour survival.
Crooks et al.: Axonal Injury in Closed Head Injury 111
.
•• \1 • •
L~ ,. ••
,
,• •
Figure 3. Retraction bulbs in the splenium of the corpus callosum. Palmgren x 40
Figure 4. Reactive astrocytes in the occipital subcortical white malter. Immunoperoxidase for GFAP x 25
MATERIALS
matter of the parietal lobes were sampled from level P4; the occipital lobes were sampled from P9 and the temporal lobes were sampled from P3. Sections of the mid-brain and the middle and upper thirds of the pons were also prepared. 9~ sections were cut, and subsequently stained with haematoxylin and eosin, Palmgren silver stain, and with immunoperoxidase for GFAP.
Head Injury cases The brains of eight individuals, who survived for less than 48 hours following head trauma during an alleged assault, and who had Diffuse Axonal Injury (DAD, were obtained from the Forensic Department of the London Hospital Medical College. They were immersed in 10% formalin and suspended by the basilary artery for 2 to 6 weeks. The diagnostic criterion for DAI used in these subjects was that which is outlined in Vanezis et al. (1987). The cerebral hemispheres were sliced in the coronal plane (the first cut placed at the mammillary bodies), the cerebellum was cut perpendicular to the folia and the brainstem was sectioned horizontally. Ten blocks in total were taken, to include both sides of the cerebrum, cerebellum and brainstem. The splenium of the corpus callosum was taken from the P4level 1; two slices of the body of the corpus callosum that incorporated independent areas of the sub-cortical and deep white matter of the frontal lobes were sampled from the P2 and P3 levels; the sub-cortical and deep white
Control cases The brains of eight selected patients, each without neurological disease, were autopsied at The Royal London Hospital. They were fixed, sliced, processed and stained as described for the individuals with head injury. METHODS
Silver-stained sections from head injury and control individuals were mixed, and a number assigned to each one at random. Measurements were made without knowing the group to which the section belonged, and the clinicopathological correlation was carried out after assessing all of the silver stained slides.
1 'P'stands for Coronal level 'posterior' to the mammillary bodies.
112 Mad. Sci. Law (1992) Vol. 32, No.2
Astrocytes with large cytoplasm and elongated processes that showed a distinctive nucleus (Figure 4). The MD of 25 enlarged axons was measured and recorded using a crossed-micrometers graticule calibrated with an Agar stage micrometer, in which each division was 2.35 urn using the 40x objective. In addition, AE, AS, AB and RA were estimated in 25 fields in each zone and their density was calculated by dividing their number by the nominal volume of tissue. There was no restriction in the number of swell ings counted per axon, and only axonal balloonings that were in fields free of ischaemia, necrosis or haemorrhage were included in the study. Considering each zone separately, the statistical differences between the head injury and control subjects were analysed with the MannWhitney U (MWU) test, and the comparison of
Definition of the variables studied
Maximum diameter of enlarged axons (MD): The diameter in micrometers (um) ofaxons that appeared approximately twice the width of the surrounding axons Axonal enlargements (AE): Number of enlarged axons estimated in 25 fields in each zone (Figure 1). Only enlarged axons that appeared twice the width of the surrounding fibres were counted. Axonal swellings (AS): Axonal lobulations visualized in silver stains (Figure 2). Axonal balloonings (AE): Argyrophilic structures with a rounded or elliptical shape depending on the plane of section located at the end of a disrupted axon (Figure 3).
Reactive astrocytes (RA).
Tahle I. Head injury: summary of the correlation between all variables studied with the survival time
Variable
p values
r
Maximum diameter
0.021
0.9611
Axonal enlargements
0.043
0.9190
Axonal swellings
0.590
0.1179
Axonal balloonings
0.610
0.1087
Reactive astrocytes
0.410
03157
Pearson correlation coefficient (r) and p values.
Table II. Head injury and control: correlations between the variables studied.
Control
Head Injury Variables
p
p
values
MD vs. AS
-0.018
09623
0.357
0.3447
AE vs. AS
-0.244
0.5185
-0113
0.7648
AS vs. AB
0.506
0.1807
0.625
0.0982
MD vs. RA
-0.286
0.4497
-0.155
0.6822
AE vs. RA 0.208 0.5815 -0.548 .::0~.c=1-=-47,-,4~_ _ Spearman correlation coefficient p (Rho) and p values. MD-Maximum diameter; AS-Axonal swellings; AE-Axonal enlargements; AB-Axonal balloonings; RA-Reactive astrocytes.
Crooks et al.: Axonal Injury in Closed Head Injury 113
zones within each group independently was done with the Wilcoxon signed rank (WSR) test. In every individual, the mean of each zone for each variable studied was averaged to obtain the mean of the means for all zones. Subsequently, differences were analysed between the head injury and control groups using the Mann-Whitney U test. For all zones, correlation was sought between the mean of the means of all variables of each individual, and survival time in the head injury group using the Pearson correlation coefficient (1'). RESULTS Median and percentile distribution
When using the median and percentile distribution for comparing the two groups, the MD showed no significant differences but controls showed consistently higher values; there was little difference between the two groups in AE although the cerebellum and frontal lobe showed the highest densities in the head injuries; in both groups the splenium showed the highest densities for AS but only the head injury group showed AB. The highest densities of RA were found in the midline of the head injury individuals but this pattern was not observed in the control group. In general, the head injury group showed higher densities in all zones for all variables, except the MD and AB. Comparison between the two groups
When using the unpaired t-Test and the MWU test for comparing head injury and control groups, no significant differences were found when the MD and AB were analysed. When comparing AE, only 3 zones showed differences that achieved statistical significance: the frontal deep (p = 0.04) and sub-cortical white matter (p= 0.01) and the occipital subcortical white matter (p = 0.01). When all zones were compared, these differences were also statistically significant (p = 0.016). The AS showed differences of significance only in the rostral brainstem (p = 0.03) and when all zones were compared (p = 0.03), whereas the RA achieved statistical significance in the splenium (p = 0.01), cerebel-
lum (p = 0.03) and occipital sub-cortical white matter (p = 0.02). Comparison between zones within the same group
When using the paired t-Test and the WSR test for comparing the zones within the same group, the head injuries showed no significant differences between the zones when the MD, AE and AB were analysed. However, differences of statistical significance were observed when comparing the AS of the temporal sub-cortical white matter against the splenium (p = 0.04), frontal (p = 0.04) and parietal subcortical white matter (p = 0.04). In addition, significant differences were also found on analysing the RA between the body of the corpus callosum and the frontal deep white matter (p = 0.03), between the splenium of the corpus callosum and the deep parietal white matter (p = .04), and between the occipital sub-cortical white matter and the cerebellum (p = .04). In the control group, no significant differences were found between the zones when comparing MD, AE, AS or AB. It is interesting that differences in RA were statistical significant between the deep and sub-cortical white matter of the frontal lobe (p = 0.04).
Correlation with survival time
There was no significant correlation between any of the variables studied and the survival time. A summary of the correlation coefficient (1') and p values is shown in Table I. In this study, the estimation of survival time is not as accurate as when dealing with experimental material in the laboratory. This leads to difficulties when cases of different survival times are compared, since it takes time for the changes under study to develop. Differing survival times mean that each case will present changes related to the particular post-injury interval after which death took place, and the picture may be obscured when the survival time has been too short for changes to have appeared, or where the time is inaccurately recorded.
114 Mad. Sci. Law (1992) Vol. 32, No.2
Correlation between the variables studied
There was no significant correlation between any of the variables studied in the head injury or control group. A summary of the Spearman correlation co-efficient (p) and p values is shown in Table II DISCUSSION Axonal enlargements and maximum diameter of enlarged axons
The possibility that the increase in axonal diameter is a post-mortem change is suggested by the fact that the control group had consistently higher values for the maximum diameter of enlarged axons than had the head injury group, and controls had longer intervals between death and autopsy. Although the difference in the interval between death and post-mortem was not statistically significant, a significant correlation was found between the axonal enlargements and the interval between death and post-mortem in the control group (p = 0.0027), suggesting that the axonal enlargements increase with longer delays between death and autopsy. In addition, axonal enlargements are not seen in biopsies, although they have been described in post-mortem material (Vanezis at al., 1987; Adams et al., 1989). While Vanezis et al. (1987) considered that axonal enlargements were indicative of early DAI when seen in association with an increase in GF AP reaction, this contention has not been endorsed by Adams et al. (1989), since, in their experience, such axonal enlargements are often present in control specimens. Therefore if patients have survived for 15 to 18 hours they will diagnose DAI only if axonal balloonings can be identified. An increase in the diameter ofaxons as an early change in diffuse axonal injury is not seen in experimental models, and this may be the result of the good tissue preservation of experimental material. This is further evidence that the nature of these enlarged axons is artefactual. Results indicate that measurements taken for the maximum diameter ofaxons in the corpus callosum, deep and sub-cortical white matter of all lobes, rostral brainstem and cerebellum, are within normal limits, because
measurements were within the same range of those of the control group. This means that the maximum axonal diameter cannot be taken as evidence of axonal damage because, with regard to this variable, there is no detectable difference between mechanically traumatized and normal brains. It is difficult to interpret the significant differences that were found between the head-injured and control groups for the density of axonal enlargements. The presence of enlarged axons in normal brains suggests that there is no reason to believe that they are pathological. It is probable that they represent the upper range of normal dimensions. If this is so, the differences in AE between the two groups are not likely to be due to the injury per se. The reasons for these differences are at present obscure. Density of axonal swellings The occurrence of highest densities of axonal swellings in the spleni um of corpus callosum, in both head injury and control groups, makes their suitability as a diagnostic feature of axonal injury questionable, if the corpus callosum is considered independently and no axonal balloonings are identified. Axonal swellings or axonal balloonings can arise in this area when the pericallosal arteries are compressed due to shift and distortion of the brain, which can also produce focal ischaemia or haemorrhage (Lindenberg et al., 1955; Adams et aI., 1989), al though axonal balloonings have been described away from focal lesions in the corpus callosum in true axonal injury (Adams and Graham, 1984). It is interesting that axonal swellings identified in human head injury material using silver techniques have been taken as presumptive evidence of DAI in cases of short survivals, whether axonal balloonings are present or not (Vanezis et al., 1987). This criterion cannot be supported by the findings in the present study. Significant differences between the head injury and control groups were found in the rostral brainstem, although the head injury cases showed axonal swellings in more than one zone. Controls showed axonal swellings in three cases, in a maximum of four different areas, but
Crooks et al.: Axonal Injury in Closed Head Injury 115
no axonal swellings in the rostral brainstem. Axonal swellings were present in the brainstem of three head injury cases; they were absent in three cases and two cases were excluded. From the three cases that showed swellings in the brainstem, the values of density indicated that the brainstem was the third, second and first most severely affected area as assessed by the density of swellings. When the brainstem was the structure most severely affected, there was no involvement of the sub-cortical white matter of the cerebral hemispheres, whereas in the other two cases, axonal balloonings were seen in the brainstem and the sub-cortical white matter in the cerebral hemispheres. It is suggested that the significant differences found in the brainstem between the head injury and control groups may be presumptive evidence which would support a mixture of patterns of structural vulnerability to axonal damage in head injury by assault, where the brainstem can be affected with variable degrees of severity, ranking from the most affected to the least affected, depending on the mechanical parameters. The differences found in the head injury group between the temporal sub-cortical white matter and the other zones arise as the result of there being no swellings in the temporal subcortical white matter in seven cases out of eight. This may be related to the observation that the temporal lobe may not swirl as much as the parietal and occipital lobes in head impacts in non-human primates, using the lucite calvarium as suggested by Pudenz and Shelden (1946). Most probably, relative immobility of the temporal lobe may produce proportionally low shearing strains that will not exceed the shear modulus ofaxons, which has been estimated as 1.4 x 104 Newtons/m2 (estimated by Thibault as quoted by Erb and Povlishock, 1988), hence producing no axonal damage. Density of axonal balloonings Axonal balloonings were present in three cases, all of which survived longer than 12 hours. This is an important finding, since axonal injury is considered not to be strongly associated with head injuries by assault (Adams et al., 1977, 1980, 1984, 1985) although its prominence in
interpersonal violence has been stressed by other investigators (Imajo and Roessman, 1984; Imajo et al., 1987). The absence of a strong correlation between assaults and axonal injury (Adams et al., 1982a, 1982b) has been explained on the grounds of biomechanics (Gennarelli and Thibault, 1982; Adams et al., 1982a), in which DAI is associated with lower rates of deceleration, in contrast to falls and assaults which are associated with short acceleration/deceleration impulses. Attention must be drawn to the fact that these correlations were made at the time when the investigators were concentrating on the most severe end of the spectrum of axonal injury, and although the views held with respect to the definition of DAI have changed, there is no evidence that a reappraisal of milder degrees of axonal injury in assaults has been undertaken. Mild degrees of axonal injury have recently been investigated (Graham et al., 1989). In this study, although the severity ofDAI was correlated with the degree of lucidity, the external causes of the injury were not. The paucity of axonal balloonings in this material is most likely to be the result of the short survival time in half of the cases, although it is uncertain whether there was axonal damage in those that survived less than 12 hours. It must be noted that axonal balloonings were occasionally found in areas away from the midline when none were seen in the parasagittal sub-cortical white matter of the frontal and parietal lobes. Whether or not this is related to the mechanisms of injury operating at the moment of injury remains a matter of interest. Correlation with survival time
Quantitative evidence provided by this study suggests that the maximum diameter ofaxons that appear enlarged with respect to the surrounding axons does not increase with longer survival times (r = 0.021, p = 0.96) as has been suggested formerly by Vanezis et al. (1987). The statistical analysis of the data collected during this investigation demonstrated that the density of axonal enlargements showed no increase or decrease with respect to the survival time (r =0.043, p =0.92), thereby rejecting the
116 Mad. Sci. Law (1992) Vol. 32, No.2
third hypothesis stated in the introduction. The original observations of Vanezis et al. (1987) have been rejected in the present study since the values for the maximum diameter of enlarged axons and the number of axonal enlargements were within the normal limits for all survival times. The density of axonal swellings showed a weak oorrelation with the survival time (r = 0.59, p =0.12). This suggests that more data are needed to establish whether or not there is a relationship between axonal swellings and survival time in human material. The Pearson correlation coefficient showed that the density of axonal balloonings correlated weakly with survival time (r = 0.610, p = 0.11). Again, more data and more cases are needed. Correlation between variables
The data presented here, after the analysis for statistical associations, showed that there was no correlation between the maximum axonal diameter and density of reactive astrocytes (p = 0.286, p = 0.45) suggesting that the first hypothesis stated in the introduction must be rejected. Similarly, no statistical correlation was found between the density of axonal enlargements and density of reactive astrocytes (p = 0.208, p = 0.58), which suggests that the second hypothesis must also be rejected. The Spearman correlation coefficient indicates that no correlation exists between maximum axonal diameter and density of axonal swellings (p = 0.018. p = 0.96), or between axonal enlargements and axonal swellings (p = 0.244, p = 0.52). This statistical evidence suggests that the first and second hypotheses must be rejected, because no association has been found between the axonal enlargements described by Vanezis et al. (1987) and the axonal swellings which are equivalent to those described in silver stains in experimental head injury in the cat (Povlishock et al., 1983; Erb and Povlishock, 1988), and in the non-human primate (Gennarelli et al., 1982; Jane et al., 1985a, 1985b). Similarly, the Spearman coefficient showed no correlation between density of axonal swellings and density of axonal balloonings (p = 0.506, p = 0.18). This lack of correlation is probably due to the paucity of axonal balloon-
ings in the material included in this study, and it suggests that the fourth hypothesis must be rejected, with the proviso that more cases need to be studied. CONCLUSIONS
Areas within the zone that appeared more severely damaged than others may be those in which the dissipation of the kinetic energy transmitted to the brain has generated sufficient force to elicit secondary axonotomy. Changes in axonal diameter, or in the density ofaxons that may appear enlarged with respect to the surrounding fibres, cannot be taken as evidence of axonal damage, rejecting previous contentions postulated by Vanezis et al. (1987). High densities of axonal swellings detected in the corpus callosum of both controls and head-injury cases preclude the utilization of the corpus callosum alone in the diagnosis of axonal injury in closed head injury, since pathology other than trauma can give rise to their formation. More cases need to be studied to assess the correlation between the survival time and axonal swellings, axonal balloonings and reactive astrocytes. No correlation was found between either the maximum axonal diameter or the density of axonal enlargements and the glial reaction, nor between the maximum axonal diameter or the density of axonal enlargements and axonal swellings, nor between axonal swellings and axonal balloonings. The hypothesis suggested by Vanezis et al. (1987) regarding the early diagnosis of DAI using axonal enlargements or swellings rather than axonal balloonings as an indication of axonal damage cannot be endorsed. REFERENCES Adams J. H. (1975) The neuropathy of head injuries. In: Vinken P. J. and Bruyn G. W. (OOs) Handbook of Clinical Neurology 23, Injuries of th.e brain tuul skull, Part I. New York, American Elsevier, 35-65. Adams J. H. and Graham D. 1. (1984). Diffuse brain damage in non-missile head injury. In: Anthony P. P. and MacSween R. N. M. (OOs) Recent Advances in Histopath.ology 12. London, Churchill Livingston, 241-57.
Crooks et al.: Axonal Injury in Closed Head Injury 117
Adams J. H., Mitchell D. E., Graham D.!. and Doyle D. (1977) Diffuse brain damage of immediate impact type. Its relationship to 'primary brainstem damage' in head injury. Brain 100, 489-502. Adams J. H., Graham D. I., Scott G., Parker L. S. and Doyle D. (1980) Brain damage in fatal non-missile head injury, J. Clin. Pathol. 33, 1132-45. Adams J. H., Gennarelli T. A, Graham D.!. (l982a) Brain damage in non-missile head injury: observations in man and sub-human primate. In: Smith W. T. and Cavanagh J. B. (eds) Recent advances in Neuropathology 2. London, Churchill Livingston, 165-90. Adams J. H., Graham D. 1., Murray L. S. and Scott G. (l982b) Diffuse axonal injury due to non-missile head injury in humans: an analysis of 45 cases. Ann. New-ol. 12,557-63. Adams J. H., Graham D. I., Doyle D., Lawrence A E. and McLellan D. R (1984) Diffuse axonal injury in head injuries caused by a fall. Lancet 2, 1420-2. Adams J. H., Doyle D., Graham D. 1., Lawrence A E. and McLellen D.R (1985) Microscopic diffuse axonal injury in cases of head injury. Med. Sci. Law 25,265-9. Adams J. H., Doyle D., Ford 1., Gennarelli T. A, Graham D. 1. and McLellan D. R (1989) Diffuse axonal injury in head injury: definition, diagnosis and grading. Histopathology 15, 49-59. Alves W. M., Jane J. A (1985) Mild brain injury: damage and outcome. In: Becker D. P. and Povlishock J.T. (eds) Central Neroous System Trauma Status Report. Bethesda, Md., National Institute of Neurological and Communicative Disorders and Stroke, 255-70. Erb D. E., Povlishock J. T. (1988) Axonal damage in severe traumatic brain injury: an experimental study in cat. Acta Neuropathol76, 347-58. Gennarelli T. A, Thibault L. E. (1982) Biomechanics of acute subdural haematoma. J. Trauma 22, 680-6. Gennarelli T. A, Thibault L E, Adams J. H., Graham D. 1., Thompson C. J. and Marcincin R P. (1982) Diffuse axonal injury and traumatic coma in the primate. Ann Neurol. 12,564-74. Graham D. 1., Lawrence A E., Adams J. H., Doyle D., McLellen D. and Gennarelli T. A (1989) Pathology of mild injury. In: Hoff J. T., Anderson T. E., Cole T. M. (eds) Mild to Moderate Head Injury. London, Blackwell Scientific Publications, 63-75.
Imajo T, Roessman U. (1984) Diffuse axonal injury. Am. J. Forens. Moo. Palhol. 5, 217-22. Imajo T., Challener R C., Roessmann U. (1987) Diffuse axonal injury by assault. Am. J. Forens. Med. Pathol. 8, 217-19. Jane J., Rimel R W., Pobereskyn L. H. and Dacey R G. (1982) Outcome and pathology of minor head injury. In: Grossman RG.H. (ed) Seminars in Neurological Surgery. New York, Raven Press, 229-37. Jane J. A, Rimel R W., Alves W. M., Dacey R G. Jr, Winn H. Rand Colohan A R (1985a) Minor and moderate head injury: model systems. In: Dacey R G., Winn H. R, Rimel R W, Jane J. A (eds) Seminars in Neurological Surgery. New York, Raven Press, 27-33. Jane J. A, Steward 0., Gennarelli T. A (1985b) Axonal degeneration induced by experimental non-invasive minor head injury. J. Neurosurg, 62, 96-100. Lindenberg R, Fisher R, Durleacher S. H., Lowett W V. Jr and Freytag E. (1955) Lesions ofthe corpus callosum following blunt mechanical trauma to the head. Am. J. Pathol. 31, 297-317. Peerless S. J. and Rewcastle N. B. (1967) Shear injuries of the brain. Can. Med. Ass. J. 96,577-82. Pilz P. (1983) Axonal injury in head injury. Acta Neurochirurg (Suppl) 32, 119-23. Povlishock J. T., Becker D. P., Cheng C. L. Y. and Vaughan G. W. (1983) Axonal changes in minor head injury. J. Neuropath. Exp. Neurol. 42, 22i}42. Pudenz R H. and Shelden C. H. (1946). The 1ucite calvarium - a method for direct observation of the brain II. Cranial trauma and brain movements. J. Neurosurg. 3, 487-505. Rand C. W, Courville C. B. (1934) Histological change in the brain in cases of fatal injury to the head. Arch. Neurol. Psychiatry 31,527-55. Simpson R H. and Berson S. D. (1987) The post-mortem diagnosis of diffuse cerebral injuries, with special reference to the importance of brain fixation. S. Afr. Med. J. 71,10-14. Simpson R H. W, Berson D. S., Shapiro H. A (1985) The diagnosis of diffuse axonal injury in routine autopsy practice. Foren. Sci. Int. 27, 229-35. Vanezis P., Chan K K, Scholtz C. L. (1987) White matter damage following acute head injury. Form. Sci. Int. 35, 1-10.
