Journal ofNeurology, Neurosurgery, and Psychiatry 1991;54:427-434 427

Glial swelling following human cerebral J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from contusion: an ultrastructural study

R Bullock, W L Maxwell, D I Graham, G M Teasdale, J H Adams

Abstract contusion, zones of profoundly reduced The ultrastructural features of cerebral regional blood flow (rCBF) have been found contusion seen three hours to 11 days adjacent to contusions, and this may be a after were studied in 18 factor in the ischaemic neuronal necrosis patients undergoing surgery. Massive which is a prominent finding in pathological astrocytic swelling ("cytotoxic" oedema) studies. Alterations to the microvascular bed was seen three hours to three days after adjacent to cerebral contusion may therefore injury, maximal in perivascular foot be an important determinant of such changes.8 processes, and compressing some of the The purpose of this study was to determine underlying capillaries. The tight junc- ultrastructurally the changes in neuronal, glial tions were not disrupted. Neuronal and microvascular morphology which occur damage was most marked three to 11 with time following human cerebral con- days after ijury. The patho- tusion, and to determine their role in the physiological mechanisms leading to pathogenic processes which are known to oedema formation and neuronal degen- damage the after severe head injury. eration are discussed. Patients and methods Cerebral cortical contusions are generally Eighteen patients with severe cerebral con- regarded as the pathological hallmark of head tusion were studied. Clinical details are shown injury and are clinically important for several in table lB. In each patient, contusions were reasons.`5 Extensive swelling of the cortical shown on CT scan, and in seven patients, ribbon and underlying white matter may mass effect, acute clinical deterioration and occur acutely after cerebral contusions and associated haematomas necessitated urgent may cause mass effect, brain shift and raised operation, at which the contusion was resec- , and these events may ted, and biopsy taken from the margins of the cause secondary ischaemic or resected tissue. In a further four patients, ICP death.6 When widespread, contusions cause monitoring was initially performed to ascer- extensive destruction of the cortical ribbon tain the need to remove the contusion, and particularly at the tips of the temporal and ICP rose necessitating surgery. In the remain- frontal lobes and the orbital surface of ing seven patients, delayed neurological the frontal lobes, and this may lead to deterioration occurred up to 11 days after http://jnnp.bmj.com/ post traumatic epilepsy, and to the neuro- injury, precipitating the operation. Specimens psychological and behavioural sequelae, of resected cerebral contusion from these typical of severe head patients were compared with specimens Institute of injury.7 Neurological Sciences, Although previous studies have focused processed in an identical manner, from four Glasgow upon the pathogenesis,45 histopathology,1 dis- control patients, with macroscopically and R Bullock tribution' 3 and quantification of cerebral con- radiologically normal brain tissue at the site of D IGraham tusion, many aspects remain unresolved.3 biopsy, who had either temporal lobectomy G M Teasdale Why do some cerebral contusions swell and for epilepsy or resection of deep temporal on September 30, 2021 by guest. Protected copyright. J H Adams generate raised intracranial pressure, while glioma. Clinical details of these patients are Correspondence to: others do not? What proportion of the shown in table 1A. Rongeur specimens Mr R Bullock, Department neuronal necrosis which is seen adjacent to around 2 cubic mm in size were taken from of Neurosurgery, Institute of Neurological Sciences, cerebral contusions is due to primary contact contused grey and white matter and Southemn GReer Hospitw, and how much of it is due to secon- immediately immersed in buffered glutaralde- Govan Road, Glasgow G51 damage, 4TF, UK dary factors, such as delayed reduction in hyde (350 m osm). Specimens were then Received 31 January 1990 blood flow to the contused area? stored at 4°C for two to seven days until and in final revised form In studies in which we have mapped processing for electronmicroscopy. The site 30 July 1990. Accepted 18 September 1990 cerebral blood flow following human cerebral from which the biopsy was taken was recorded by the surgeon and related as closely as possi- ble to the CT appearance on the pre-operative Table IA Controlpatients-clinical data scan. The study was approved by the Ethics Committee of the Institute of Neurological Age Case (years) Tissue biopsied Indicationfor surgery Sciences, Glasgow. 1 65 Cortex and white matter right Deep temporal glioma, seizures temporal tip 2 56 Left frontal cortex Partial left frontal lobectomy for Processingfor electronmicroscopy glioma After immersion fixation, specimens were cut 3 36 Posterior right temporal cortex and Right temporal lobectomy for glioma white matter into 200 p thick sections using a vibrotome, 4 22 Right temporal cortex Right temporal lobectomy for severe and ultrasonicated to dislodge non-adherent epilepsy blood components from the vessel lumina. 428 Bullock, Maxwell, Graham, Teasdale, Adams

Table lB Patients with cerebral contusion-clinical data interest selected for transmission electron- J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from microscopy. Thin sections were cut for the Age Interval between Case (years) injury and biopsy CT diagnosis and clinical details transmission EM and routinely stained with lead citrate and uranyl acetate and examined 67 3 hours Fall. Right temporal contusion, small acute subdural haematoma, hemispheric swelling, high ICP (died) in a JEOL 100 S transmission electronmicros- 2 33 6 hours Fall. Right frontal lobectomy for massive brain swelling cope. when acute subdural haematoma removed. (Alcoholic with clotting defect-died) 3 68 13 hours RTA. Right temporal contusion with deterioration of consciousness. Good recovery 4 18 16 hours RTA. Large traumatic, right frontal intracerebral Results haematoma and overlying contusion, deterioration of Biopsies were obtained at time points ranging consciousness. Good recovery 5 52 17 hours Found unconscious and drunk. Left temporal contusion from three hours to 11 days after injury. In all ("Burst lobe") and acute subdural haematoma. Disabled the specimens the quality of fixation as judged 6 18 23 hours RTA. Bifrontal contusions. Left frontal intracerebral haematoma, deteriorating consciousness. Disabled by preservation of intracellular organelles was 7 29 27 hours Assault. Focal right parietal contusion and depressed acceptable. fracture. Good recovery 8 17 28 hours RTA. Left hemispheric swelling and temporal contusions-ICP monitoring -+ high ICP. Disabled 9 59 31 hours RTA. Acute right extradural, subdural haematomas and right frontal contusion. Deteriorating consciousness. Control patients Disabled Normal glial, neuronal and vascular ultra- 10 55 33 hours Fell while drunk. Delayed deterioration. . Bifrontal contusions. Thin left acute subdural haematoma. Good structure was present in all control patients, recovery and the neuropil and white matter density was 11 68 37hours RTA. Right temporal contusion, small acute subdural. normal (fig 1). Scanning electronmicroscopy ICP monitoring -+ high ICP. Good recovery 12 42 42 hours Found unconscious. Delayed deterioration. Bitemporal demonstrated pit-vesicle activity on the contusions. This left acute subdural. Good recovery 13 46 44 hours Fall. Delayed confusion. Right temporal contusion. ICP endothelial surface of the lumen of all vessels monitoring -. high ICP. Good recovery in the control patients (table 2). 14 59 49 hours Alcoholic epileptic. ? Fall. Delayed confusion. Bifrontal contusions, mass effect. Disabled 15 40 72 hours Fall. Right parietal acute subdural. Delayed ICP rise new low density. Right temporal -* re-opening craniotomy and right temporal lobectomy. Disabled Transmission EM findings in patients with 16 64 77 hours Fall. Multiple contusions. Left hemisphere, small cerebral contusion extradural haematoma. Right parietal. Delayed deterioration. Partial left temporal lobectomy. Disabled A) GROSS DISRUPTION OF TISSUE 17 44 5 days Hydrocephalic with shunt. Fell out of bed -. bifrontal In five specimens taken from three patients the contusions. Delayed coma. ICP high. Right frontal contusion evacuated. Disabled architecture of the cortical neuropil was dis- 18 72 11 days Fall. Persistently confused. Right frontal subacute rupted by red blood cells and plasma. When- subdural, small temporal contusion. (Burst lobe at surgery). Disabled ever this was seen, the adjacent neuropil was grossly abnormal, demonstrating marked vacuolation and lucency of astrocyte cyto- plasm (fig 2). Specimens were then fixed in 2% osmium tetroxide in phosphate buffer, and the slices were critical-point dried, mounted on scan- B) GLIAL CELL SWELLING ning electronmicroscopy stubs with silver The most consistent finding in both grey and paint and sputter-coated before viewing in a white matter in patients with cerebral con- JEOL 300 TS scanning electronmicroscope. tusions was severe swelling affecting par- http://jnnp.bmj.com/ Blood vessels were selected for particular ticularly astrocytes. This was associated with study and micrographs were taken at a mag- "loosening" and apparent pseudo-vacuolation nification of 7000 and 1500 times to allow of the neuropil to produce a "Swiss cheese" demonstration of endothelial surface structure appearance, particularly striking on SEM. and measure vessel size. After scanning elec- This appearance developed rapidly (figs 3, 4a) tronmicroscopy specimens were carefully within three hours of injury but was still

removed from the SEM stubs and washed present in specimens taken three days after on September 30, 2021 by guest. Protected copyright. three times in 100% acetone to remove the silver paint. The material was then embedded in araldite resin and cut into semi-thin sec- tions stained with toluidine blue and areas of

Table 2 The time intervalsfor various EMfeatures after cerebral contusion

Time interval at which EM occurred after injury EMfeature 3 hours-3 days Astrocyte swelling 3 hours-3 days Astrocyte podocyte swelling 3 hours-3 days Astrocytc podocyte disruption V,. 3 hours-7 days Vessel lumen compression . 3 hours-5 days "Swiss cheese vacuolation" ofneuropil on TM 3, 16 hours Red cell diapedesis 6 hours-3 days Endothelial fenestration on SEM 16 hours-3 days Intravascular clots adherent to walls 16 hours-3 days Gross tissue disruption + haemorrhage ' ...: 16 hours-5 days White matter "loosening" 16 hours-l days Peri-vascular red cells 3-11 days Pyknosis of neurons 5-7 days Sheet reaction-glycogen granules Figure I Thin section of cortical neuropilfrom control -pseudopodial processes from astrocytes 5, 11 days Leucocytic phageocytosis patient KW. Preservation ofintracytoplasmic organelles 11 days Collagen deposition is satisfactory and vacuolation ofneuropil is absent. (Magnification: x 2975). Glial swellingfollowing human cerebral contusion: an ultrastructural study 4z29

Figure 2 Thin section J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from through cortical neuropil I 16 hours after injury. Note 1* gross swelling of astrocytic a cytoplasm (lower right) and vacuolation of neuropil caused by swelling astrocytic process. Note mitochondria apparently lyingfree within the cytoplasm ofswollen astrocytes (arrow). (Magnification: x 5250).

Figure 5 Thin section through white matter capillary Figure 3 Scanning EM showing gross astrocyticfoot process swelling (a), with ofa section through disruption of thefoot process limiting membranes and cortical neuropil to show narrowing and compression ofvessel lumen (L). vacuolation surrounding a (Magnification: x 3750). blood vessel (arrow) ("Swiss cheese" appearance). C) SWELLING OF PERIVASCULAR PODOCYTIC (Magnification: x 1000). PROCESSES Swelling of astrocytic perivascular podocytic processes were the most obvious manifesta- tion of glial swelling. This was sufficiently severe to suggest compression of the adjacent vessel lumen in some cases (fig 4a, b). Disrup- tion of foot processes was seen in some specimens taken at periods from three hours to three days after injury (fig 5). In specimens which were taken at later periods after injury, astrocytic swelling appeared to be less exten- sive and at five days astrocytic processes con- tained glycogen deposits and pseudopodial injury. In the worst affected astrocytes, cyto- folds typical of the "sheet reaction" of plasmic disruption was present with complete Torvick (fig 6a, b).9 Comparison between the rupture of the cell membrane (fig 5). Organe- changes seen in cortex (fig 4a, b) and those in lles such as mitochondria were, however, still white matter (figs 4, 5, 6) should, however, be recognisable. In the specimens taken five days made with caution. http://jnnp.bmj.com/ after injury, astrocytic swelling was less marked although still apparent (fig 6a, b) and the astrocytes now contained deposits of D) VASCULAR CHANGES (TRANSMISSION ELEC- glycogen, typical of the "sheet reaction" des- TRONMICROSCOPY) cribed by Torvick (fig 6).9 In the majority of vessels seen within con- on September 30, 2021 by guest. Protected copyright.

I lp*P . ,i..., --- -. , -,. .4iAI"

n -t-S

VA .41

.,A .. .

AW

Figure 4 (A) Thin section through a cortical capillary, showingflattening ofvessel luwnen (L) and massive swelling of perivascular astrocyticfoot processes (a). The lumen is narrowed to such an extent that passage of red cells would be impossible. (Magnification: x 10 500); (B) Scanning EM preparation ofa larger blood vessel in cortical neuropil showing collapse of the vessel lumen andfolding of the tunica intima with adherent clot (arrow). (Magnification: x 460). *a. 430 Bullock, Maxwell, Graham, Teasdale, Adams

Figure 6 (A) Thin J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from section ofwhite matter takenfive days after contusion-the swelling of astrocyticfoot processes (a) is less marked with glycogen accumulation. [%*wo ; % s- * , ; The sheet reaction is apparent. (Magnification: x 8200); ; (B) Thin section of cortical neuropil in a different patient, operated F^fs i&4>2K uponfive days after head injury. Swollen astrocyte processes and accumulation ofglycogen granules therein are seen. ( The so- called "sheet reaction"). (Magnification: x 8750).

tused tissue, endothelial cell morphology was grossly normal and in those vessels where adjacent astrocytic swelling was not noticed, vessel lumina were not disrupted or occluded. No disruption of inter-endothelial tight junc- tions was noted, and in the majority of vessels studied, the morphology of the perivascular basement membrane was also normal. An increased number of cytoplasmic folds and processes on the luminal aspects of endo- thelium were seen within vessels and thin transmission sections showed a very irregular luminal profile to the endothelium at two days (fig 7). Inter-endothelial junctions were not disrupted despite gross disruption of the adjacent neuropil. This irregular endothelial profile was also seen at five days when the astrocytic sheet reaction had become well Figure 7 High magnification thin section ofpart of a established (fig 8).9 vessel wall two days after injury to show a thickened Counts of endothelial pinocytotic vesicles in irregular profile of the endothelium. Microvilli are present. Note that the interendothelial tightjunctions are

the vessels from contusion specimens taken at http://jnnp.bmj.com/ intact (arrow). Astrocyticfoot processes on the abluminal different times after injury, showed a decline surface of the basal lamina are grossly swollen. with time and the highest numbers of vesicles (Magnification: x 20 000). were seen in the control and acute patients (table 3). Delayed changes A PYKNOSIS AND SHRINKAGE Early white matter changes on transmission EM Pyknosis and shrinkage of both neurons and Twelve specimens of white matter from con- glia was noted when biopsies were taken at on September 30, 2021 by guest. Protected copyright. tused tissue were studied. Preservation of three or more days after injury (fig 9). Pyknotic myelin was remarkably good. In some patients, neurons were frequently surrounded by swelling of axoplasm was noted, but this was microglia which appeared to be carrying out never severe. Cytoplasmic swelling ofglial cells phagocytosis of damaged neuropil and cells. In within white matter was present but was less specimens from the two patients who were marked than that seen in the grey matter. studied five and 11 days after injury, extensive Accumulation of interstitial fluid within white leucocyte and microglial phagocytosis was seen matter was marked (fig 5). and deposition ofcollagen fibres was prominent

Table 3 Microvillus counts (SEM) andpinocytotic vesicle activity (TEM) after contusion

Endothelial pinocytotic Luminal microvillus counts per 25 p2 Time interval vesicle count mean (SD) after per p2 injury mean (SD) Capillaries (<10 p) (>1Op) Controls - 6 105 (3-51) 9-66 (3 77) 12 916 (3 81) Contusion 3-6 hours 6 125 (2-99) 2-92 (1 59) 22 19 (5-6) 13-23 hours 0-75 (0 45) 5 0 (2 5) 24-48 hours 4-72 (3-01) 5-343 (2 44) 4 914 (1-35) 2-5 days 1 5 (0 91) 7 166 (2 23) 8 45 (3 45) 11 days 7 91 (2 33) 12 31 (3 63) Glial swellingfollowing human cerebral contusion: an ultrastructural study 431

Figure 8 High the neuropil in tissue adjacent to cerebral J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from magnification thin section takenfive days after contusions was particularly noticeable on scan- cerebral contusion. Note n.g EM in comparison with the control numerous endothelial patients microvilli and (fig 3). accumulation ofglycogen granules in astrocytic perivascularfoot processes B VASCULAR AND PERIVASCULAR CHANGES (a). (Magnification: Vessel morphology was remarkably normal in x 26 000). all specimens studied. Tears in vessel walls were not seen. Blood clots occluding the lumen ofvessels were seen only occasionally, although this may relate to the vibration and ultrasonica- tion process which was used to visualise the lumina. In many vessels, however, adherent '~~1 red cells and leucocytes were found suggesting an abnormality of endothelial adhesion charac- teristics. Red cells were seen moving through fenestrations in the endothelial membrane to leave the vessel lumen, by a process of .. I Figure 9 Thin section of 11 a pyknotic cortical cell, .Alk diapedesis (fig 1 la, b). probably a neuron Perivascular accumulations ofred blood cells containing lipofuscin were a frequent finding and in several C .14 granules, taken 72 hours A ..o specimens, these caused streak-like disruption after head injury. Note marked shrinkage and of tissue architecture. peri-cellular vacuolation. (Magnification: x 11 000). Pit-vesicle activity Pit-vesicle activity was present in both control and contusion specimens, but was less exten- sive in the patients with contusions. In some areas, the endothelial luminal membrane was disrupted by coalescent vesicles to form irregular holes in the vessel walls, and in two specimens, red cells were seen passing through .. .i these (fig 1 la, b). Pit-vesicle activity in the endothelium was always accompanied by parallel development of microvillus formation, Figure 10 Thin section but the proportion of pit-vesicle and from cortical neuropil in microvillus activity was variable from patient RS 11 days after head injury. Note specimen to specimen, and no clear relation- leucocytic ship was seen, either to time after injury, or to

phagocytosis http://jnnp.bmj.com/ (M) collagenfibre other microscopic changes. When the number deposition (outline arrow) andfibroblast (F). of luminal microvilli per 25 gim2 were counted (Magnification: x 9250)., using SEM, and correlated with the number of vacuoles within the endothelial cytoplasm seen on transmission electronmicroscopy, (table 3) no relationship was seen in either the contusion or control material.

The mean number of endothelial micropin- on September 30, 2021 by guest. Protected copyright. ocytotic vesicles in control patients was 6-105 jim2 and 6-125 gm2 in contusion patients at three hours after injury, 4-72 jim2 at 24-48 hours after injury, and 1-5 pm2 at five days after injury. Thus there was a slight (but not statis- tically significant) decline over time in the number of endothelial pinocytotic vesicles in damaged tissue. Marked changes in luminal microvillus (fig 10), indicating the early stages of scar tissue activity were seen in the contusion material, at formation. In the patient studied 11 days after various times after injury (table 3). In injury, perivascular foot process swelling was capillaries, microvilli numbers dropped to their absent and the numbers of perivascular lowest levels when astrocytic and foot process microglia were increased, suggesting that the swelling was most marked, at 13-24 hours and stage of maximal cell swelling had been passed. slowly rose to near control values by 11 days. By contrast, in larger vessels microvillus activity rapidly rose to almost twice control Findings on scanning electronmicroscopy in levels in the first six hours after injury, then patients with cerebral contusion dropped to nearly a third ofcontrol levels at 24- A NEUROPIL 48 hours, and were approaching control levels The vacuolated "Swiss cheese" appearance of by 11 days. 432 Bullock, Maxwell, Graham, Teasdale, Adams

Figure 11 (A) Scanning J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from EMpreparation of the luminal surface ofa cortical arteriole taken 17 hours after head injury. ii. . . ' ...... ': S,'l B l l F '' ...... !.::. Note the largefenestr:ae in tsSrsht t!M..s.X.'B','''.il5.E.s.iy ; | l l 4l >lE vSX.. v wtleKi;Me;-. w the endothelial surface ...... < , . t: - :oR} , i . g ,S - eSr ; .; .. through which erythrocytes .. .8 ...... S X m t' ...... are seen to pass (arrow). * * r v . S: uS ' ...... Y ] X l | S X ...... >m A (Magnification: x 1780), 1|15 ; ..m: ze:: " e:: /aS *::'' '. .i:::: _.'.;: 't:: S e S,:::: ,g,.,>,>- SS ' -; S '' A A (B) High magnification view of the endothelial . .. , ,, .. W .. ,,, .. ,,<,-:2 t-<',, ' '' :}:; s ...... ' i ' : }:}':.} s { ; .. ...S?;: : ::.=A- ;S_ ..x >''M.: ?': . sA: :: " ' ' '? : - ' '": surface 17 hours after head injury. Note the | S . :. : fi:. ... . : :::: . :. ;. .' . .; . . . .. }} .e .. ... S: endothelial microvilli and m . . ei.e.... .i e.. .. ?2' : ' : Irr.... . w .,;.,l::: £' %:' }: ..:' : : ?F ?' ^.;.X .... .::-Si:': erythrocyte passing il??^X<§ffi? !. SS . .. ^ . ., . e,£s ....'"''' ....^ ...... through thefenestration. r ...... > <, S.r } ...... e U * x .. . e ... .:. * =;, U (Magnification: 9700). .:.. | i s¢ .:. .: . : .:':::.'.::. , .... :.::...... w @tD ,r" BA # _#

Local cerebral which occurs after contusion in humans may be Haemorrhage blood flow a secondary, delayed process. This is in con- trast to the findings in the rat model of cerebral ncreassed local : ' #:X. ngg . ... ,-fo Tissue ischaemia cerebrcovascular contusion described by Hekmatpanah, in Contusion resisstance which neuronal changes were well established by three hours after injury, both on light and electronmicroscopy." Contact injy Membrane _.. Neuronal Delayed Capillary lumen damage death "vasogenic" Measurement of rCBF in patients after compression oedema cerebral contusion, has shown zones of profoundly reduced CBF (around 15-20 ml/ \ / W"~~~~~1 100 gm/min), at the site ofthe contusion, using K+, releIase, Epithelial excitatory amino response SPECT mapping and PET techniques.815 acid rele,ase These have been shown as early as within 24 hours of injury and up to three weeks after \ injury. Local microcirculatory vascular factors swelling may therefore be important in the pathogenesis Figure 12 Possible pathophysiological mechanisms after cerebral contusion. of neuronal degeneration adjacent to cerebral contusions. This study demonstrates four major microvascular changes, which have also Discussion been seen in a variety of animal models: 1) Cerebral contusion has been studied exten- marked perivascular astrocytic process swell- sively by light microscopy in humans and by ing-sometimes sufficient to cause apparent light and electronmicroscopy in animal models, compression of the vessel lumen (fig 4); 2) but the ultrastructural changes in humans have reduced luminal pit-vesicle and microvillus http://jnnp.bmj.com/ not been reported.'"5"02 The majority of his- activity, (table 3; figs 7 and 8); 3) decreased topathological studies were largely descriptive, endothelial cytoplasmic vacuole formation but Lindenberg and Freytag have also (table 3), and 4) transvascular red cell speculated on the mechanisms of neuronal diapedesis (fig 1 1). damage. It is postulated that the streaky form of Our study suggests that the major structural neuronal ischaemic necrosis which is seen feature associated with reduced local tissue

surrounding contusion differs from the conven- perfusion is microvascular collapse and com- on September 30, 2021 by guest. Protected copyright. tional changes of ischaemia,45 13 and postulated pression due to astrocytic perivascular that impact injury propagates waves of shear podocytic process swelling together with some stress through brain substance which damage compromise of cerebral circulation mediated neurons by tearing axons and dendritic proces- by the development of endothelial microvilli ses.5 However, neither in the present ultra- (figs 4, 11). This accords with our previous structural study nor in that of Hekmatpanah in ultrastructural findings, those ofDietrich et al, a rat model could shearing of dendritic proces- and the pattern ofprofound perifocal reduction ses or axons be demonstrated." Moreover, the in rCBF seen in a model of intracerebral neuropathological studies of Adams and Gra- haematoma.8"1617 Perivascular astrocytic ham carried out in 635 patients dying after podocyte swelling was also prominent in the severe head injury have not documented an ultrastructural studies of Allen-performed in association between either retraction balls or a primate model ofhigh velocity missile injury, microglial stars, the hallmarks of diffuse axonal and Hekmatpanah, in the rat contusion injury, and cerebral contusion.'4 model.'01' We have not, however, seen the In this study, neurons demonstrating the vessel tears and thrombosis described by Hek- features of the ischaemic cell process were not matpanah in the rat model, and we have not seen in the specimens taken at three and six demonstrated disruption of inter-endothelial hours, but were clearly apparent in the tight junctions (fig 7). specimens taken at one day; by the third day, Although the specimens studied five days pyknosis and nuclear lysis were obvious (fig 9). after injury showed less marked astrocyte cell This suggests that the neuronal degeneration body and foot process swelling than those Glial swellingfollowing human cerebral contusion: an ultrastructural study 433

studied within the first few days, these changes with other studies,22 and may provide a J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from should be interpreted with caution. Serial mechanism for the peri-vascular accumulation biopsies were not taken from the same patient oferythrocytes, which is a frequent histological and the severity of the contusion was variable feature after contusion but which has not been from patient to patient. previously explained.'45 Vessel tears may For these reasons the mechanisms by which therefore not be necessary for perivascular perivascular astrocytic foot process swelling "haemorrhage". may resolve after a contusion cannot be Although we and other authors have shown inferred from this study. Although specimens that astrocytic swelling is well established after taken at five days or more after the injury show cerebral contusion as early as three hours evidence of astrocytic process sprouting, and after injury, both in this study and in animal apparent reconstitution of the foot process models, the cause for this swelling remains limiting membranes, (fig 6a) this may represent uncertain.""20 22 the response to a less severe lesion. To improve When cerebral blood flow falls sufficiently our understanding of these fundamental repair low to cause failure of neuronal energy mechanisms, serial biopsy studies will be metabolism, and membrane depolarisation needed, in an animal model, with a controlled (around 14-15 ml/100 gm/min) due to global severity of injury. or focal ischaemia, massive astrocytic swelling Early perifocal oedema has been demon- occurs as a buffering mechanism; astrocytes strated in microgravimetric studies after absorb free K+ ions in an attempt to maintain human contusion.819 The prominence of extracellular fluid homeostasis.26 27 Global astrocytic swelling early after cerebral con- ischaemia is, however, unlikely to occur as a tusion, and the integrity of the inter-endoth- very early event after focal human contusion elial tight junctions seen in this study and in and, moreover, some animal studies have animal models,1011 2021 suggests that this is a shown blood pressure and CBF to be transien- cytotoxic process. The relative paucity of tly elevated after head injury.2325 protein-rich extracellular fluid (except where Faden et al and Becker et al have demon- gross tissue disruption was present), (fig 5) strated massive release of excitatory amino within the neuropil in this study is also acids after fluid percussion injury, in tissues evidence against a vasogenic mechanism. immediately beneath the impact point, but not This agrees with the results of studies of distant from it.2829 Excitatory neuronal mem- blood-brain barrier permeability which we brane depolarisation may result from this have performed at various time intervals after glutamate release, to cause subsequent K+ion focal cerebral contusion in 20 patients. SPECT flooding ofthe extracellular space, and this may scanning after intravenous 9'Tc pertech- cause astrocytic swelling, which itself may netate, showed that blood-brain barrier chan- jeopardise focal tissue perfusion adjacent to a ges were seldom detectable before three days cerebral contusion, and induce further after injury, and were maximal at about 10 ischaemic damage.26 A pathological "vicious days.8 cycle" such as that shown in fig 12, may thus be The micro-anatomical basis for the delayed postulated, as a cause for neuronal death in of increased vascular development per- cerebral contusion. http://jnnp.bmj.com/ meability after contusion is difficult to deter- Vasoconstrictor substances such as K+, mine from this study, or from the animal serotonin, or prostaglandins may also be studies of other authors. Povlishock, using a released into the contused tissue, inducing the fluid percussion model of head injury, demon- profound fall in regional blood flow, which we strated that Horseradish peroxidase (HRP) have demonstrated on SPECT (8). passed rapidly through intact endothelial cells Further studies are needed to clarify the after intravascular injection, and accumulated causes of astrocytic swelling, and to explore beneath the basement membrane, and within the possibilities of therapy for preventing or on September 30, 2021 by guest. Protected copyright. endothelial vesicles.2" 22 HRP was found chiefly attenuating this astrocytic response, which may within endothelial cytoplasmic vesicles and the be detrimental to neuronal survival. number of vesicles increased rapidly within 15 to 30 minutes after trauma.2' In contrast to We are grateful to the neurosurgeons at the Institute of Neurological Sciences, Glasgow who assisted with obtaining these experimental observations, we have cerebral contusion and control specimens, and to Anne Semple demonstrated that after a contusion micro- for preparing the manuscript. pinocytotic vesicles decline in number with 1 Adams JH, Gennarelli TA, Graham DI. Brain damage in time (table 3). Therefore, it is unlikely that non-missile head injury: observations in man and sub- micropinocytosis is an important mechanism in human primates. In: Thomas Smith W, Cavanagh JB, eds Recent advances in neuropathology. Edinburgh: Churchill delayed post-contusional oedema. Livingstone, 1982:165-90. The role of pit-vesicle activity and 2 Adams JH, Gennarelli TA, Graham DI, et al. Changes in endothelial morphology of brain vasculature after con- microvillus formation after head injury is not trolled angular acceleration of the subhuman primate understood. Its presence within hours ofinjury head. Proc Physiological Soc 1987:92. 3 Adams JH, Scott G, Parker L, et al. The Contusion Index: a after "pure" suggests that quantitative approach to cerebral contusions in head it may be a non-specific response to the invaria- injury. Neuropathol Appl Neurobiol 1980;6:319-24. 4 Lindenberg R, Freytag E. Morphology of cortical con- ble hypertensive surge which occurs within tusions. Arch Pathol 1957;63:23-42. seconds after severe head injury.2 23-25 5 Lindenberg R, Freytag E. The mechanism of cerebral contusion. Arch Pathol 1960;69:440-69. We have observed coalescence of vesicles to 6 Miller JD, Gudeman SK, Kishore PS, et al. Computed form ragged holes in the endothelial luminal tomography in brain oedema. Advances in neurology Vol 28. New York: Raven Press; 1980:413-22. membrane, through which intact erythrocytes 7 Levin HS, Grossman RG, Rose JE, et al. Longterm are seen to pass (fig 11). This partly accords neuropsychological outcome in closed head injury. J 434 Bullock, Maxwell, Graham, Teasdale, Adams

Neurosurg 1979;50:412-22. Neurosurg 1985;63:64-8. J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.54.5.427 on 1 May 1991. Downloaded from 8 Bullock R, Teasdale GM, Wyper D, et al. Tomographic 19 Galbraith S, Cardoso E, Patterson J, et al. The water content mapping of CBF, CBV and BBB changes after focal head of white matter after head injury in man. In: Go KG, injury using SPECT: mechanisms for late deterioration. Baethmann A. Recent progress in the study and therapy of In: J Hoff, ed. Proc VIIth Int Symposium Intracranial brain oedema. New York: Plenum Press, 1984:323-9. Pressure and Brain Injury. Berlin: Springer-Verlag 20 Olsson Y, Rinder L. Studies on vascular permeability 1989:637-9. changes in experimental brain . Acta 9 Torvick R, Skjorten F. Electron microscopic observations Neuropathologica (Berl) 1968;11:183-200. on nerve cell regeneration and degeneration, after axon 21 Povlishock JT, Becker DP, Sullivan HG, et al. Vascular lesions. I. Changes in the nerve cell cytoplasm. Acta permeability alterations to horse radish peroxidase in Neuropathol 1971;17:248-64. experimental brain injury. Brain Research 1978;153: 10 Allen IV, Kirk J, Maynard RL, et al. An ultrastructural 223-39. study of experimental high velocity penetrating head 22 Povlishock JT. The morphopathologic responses to injury. Acta Neuropathologica (Berl) 1983;59:277-82. experimental head injuries ofvarying severity. In: Becker 11 Hekmatpanah J, Hekmatpanah CR. Microvascular altera- DP, Povlishock JT, eds. CNS Trauma Status Report. tions following cerebral contusion in rats. J Neurosurg Bethesda, Maryland: NIH, 1986:443-52. 1985;62:888-97. 23 Gennarelli TA, Segawa H, Wald U, et al. Physiological 12 Jenkins A, Maxwell WL, Graham DI. Experimental response to angular accelaration of the head. In: Gross- intracerebral haematoma in the rat: sequential light mann RG, Gildenberg PL, eds. Head Injury: basic and microscopical changes. Neuropathol Appl Neurobiol clinical aspects. New York: Raven Press 1982:129-39. 1989;15:477-86. 24 Maxwell WL, Irvine A, Adams JH, et al. Response of the 13 Little JR, Kerr FWL, Sundt TM. Microcirculatory obs- cerebral microvasculature to brain inmury. J Pathology truction and focal cerebral ischaemia: an electron micros- 1988;155:327-35. copic investigation in monkeys. Stroke 1976;7(1):25-30. 25 Nilsson B, Nordstrom CH. Experimental head injury in the 14 Adams JH, Doyle D, Ford I, et al. Diffuse axonal injury in rat. Pt 3. Cerebral blood flow and oxygen consumption head injury. Definition, diagnosis and grading. Histopath- after concussive impact acceleration. J Neurosurg ology 1989;15:49-59. 1977;47:262-73. 15 Tenjin H, Mizukawa N, Yamaki T, et al. The investigation 26 Bourke RS, Nelson KM. Further studies in the K + of cerebral haemodynamics in patients with cerebral dependent swelling ofprimate in vivo: the contusion: Evaluation of Positron Emission Tomography enzymatic basis of the K + dependent transport of (PET). J CBF Metabol 1989;9(Suppl 1):S87. chloride. J Neurochem 1972;19:663-85. 16 Dietrich WD, Busto R, Ginsberg MD. Cerebral endothelial 27 Plum F, Cooper AJL, Craig RP, et al. Glial cells: the silent microvilli and fenestration following global fore brain partners of the working brain. The Thomas E Duffy ischaemia. J Neuropathol Exp Neurol 1984;43:72-83. Memorial Lecture. J CBF and Metabol 1985;5(Suppl 17 Mendelow AD, Bullock MRR, Teasdale GM, et al. 1):S1. Intracranial haemorrhage induced at arterial pressure in 28 Becker DP, Katayama Y, Tamura T, et al. Excitotoxic ionic the rat. Pt II. Short term changes in local cerebral blood fluxes and neuronal dysfunction following traumatic brain flow measured by autoradiography. Neurol Res injury. J CBF Metabol 1989;9(Suppl 1):302. 1984;6:189-93. 29 Faden AI, Demediuk P, Panter SS, et al. The role of 18 Bullock MRR, Smith R, Favier J, et al. Brain specific gravity excitatory amino acids and NMDA receptors in traumatic and CT density measurements after human head injury. J brain injury. Science 1989;244:798-800. http://jnnp.bmj.com/ on September 30, 2021 by guest. Protected copyright.