Clinical Science (1999) 96, 461ñ466 (Printed in Great Britain)
98/00327
Faecal peritonitis causes oedema and neuronal
injury in pig cerebral cortex
M. C. PAPADOPOULOS*†, F. J. LAMB*, R. F. MOSS‡, D. C. DAVIES§, D. TIGHE*
and E. D. BENNETT*
*Department of Anaesthetics and Intensive Care Medicine, St. George’s Hospital Medical School, Cranmer Terrace,
London SW17 0RE, U.K., ‡Department of Electron Microscopy, St. George’s Hospital Medical School, Cranmer Terrace,
London SW17 0RE, U.K., and §Department of Anatomy and Developmental Biology, St. George’s Hospital Medical School,
Cranmer Terrace, London SW17 0RE, U.K.
A
B
S
T
R
A
C
T
Encephalopathy is a common complication of sepsis. However, little is known about the
morphological changes that occur in the brain during sepsis. Faecal peritonitis was induced in
pigs that were killed 8 h later and frontal cortex samples were taken immediately after death.
The tissue was investigated using light and electron microscopy and compared with frontal
cortex samples taken from sham-operated controls. Septic pigs had 49.5 % more perimicrovessel
oedema than sham pigs. However, the tight junctions between cerebral microvessel endothelial
cells appeared morphologically intact in both septic and sham pigs. Sepsis also resulted in
neuronal injury, disruption of astrocytic end-feet and swollen, rounded erythrocytes. These
morphological changes may be sufficient to underlie the clinical features seen in septic
encephalopathy.
INTRODUCTION
Septic shock is the commonest cause of death in the
Intensive Care Unit [1]. Progressive sepsis causes multiorgan failure and the mortality of patients with three or
more failing organs approaches 100 % [2]. The adverse
effects of sepsis on the liver, lung, kidney and heart have
been studied extensively [3–9]. There is panendothelial
injury from persistent, uncontrolled inflammation which
causes the deposition of fibrin and the formation of
thrombi in microvessels [10]. As a result, these organs
become underperfused [8] and accumulate activated
leucocytes [3], especially degranulating neutrophils [5]
which release lysosomal enzymes and superoxide radicals.
Although 50 % of patients with progressive systemic
sepsis develop encephalopathy, which is a marker for
poor prognosis, little is known about the effects of sepsis
on the brain [11]. Septic encephalopathy (SE) is a diffuse
brain dysfunction. Its clinical features vary from mild
confusion to coma and its pathogenesis is unclear. SE is
associated with breakdown of the blood–brain barrier,
since patients have high protein levels in the cerebrospinal
fluid [12], and in septic rodents "#&I-albumin [8], "%Camino acids [13] and colloidal iron oxide [14] pass from
the circulation into the brain parenchyma. However, the
nature of the blood–brain barrier impairment in SE is
unknown. Increased pinocytosis by cerebral microvessel
endothelia and swelling of the astrocytes have been
reported following electron microscopy of brains from
rabbits with endotoxaemia [14], but these early experiments lacked haemodynamically matched controls.
Therefore, in an attempt to investigate the effect of SE on
the blood–brain barrier, the morphology of frontal cortex
Key words : astrocytes, cerebral microvessels, faecal peritonitis, neuronal injury, oedema, septic encephalopathy, ultrastructure.
Abbreviation : SE, septic encephalopathy.
Correspondence : Dr D. C. Davies.
† Present address : Department of Neurosurgery, Atkinson Morley’s Hospital, Copse Hill, London SW20 0NE, U.K.
# 1999 The Biochemical Society and the Medical Research Society
461
462
M. C. Papadopoulos and others
in an established porcine model of sepsis [5–7,9] was
investigated by light and electron microscopy.
METHODS
Surgical procedures
Anaesthesia was induced in 10 adolescent middle white
pigs (25–30 kg) by intramuscular injection of azaperone
(2.0 mg\kg, Janssen-Cilag Ltd, High Wycombe, Bucks,
U.K.) and intraperitoneal injection of metomidate
(10.0 mg\kg, Janssen-Cilag Ltd). A tracheotomy was
performed, each pig was intubated and anaesthesia was
maintained throughout surgery with a mixture of 50 %
O , 48 % NO and 2 % halothane (Zeneca Pharma#
#
ceuticals, Wilmslow, Cheshire, U.K.). Catheters were
inserted into the internal jugular vein and descending
aorta and a Swan Ganz catheter (Baxter Edwards Critical
Care, Irvine, CA, U.S.A.) advanced into the pulmonary
artery. Faecal peritonitis was induced in five pigs as
follows : after laparotomy, an opening was made in the
caecum and 35 ml of its contents aspirated and diluted to
50 ml with normal saline. The caecotomy was then closed
and the faecal solution spread around the peritoneum.
Five sham pigs underwent laparotomy without caecotomy and the induction of faecal peritonitis. After closure
of the laparotomy in the pigs with faecal peritonitis and
the sham controls, the lungs were ventilated (Harvard
Ventilator, Harvard Ltd, Edenbridge, Kent, U.K.) with
room air. The tidal volume (12–15 ml\kg) was adjusted
to keep the arterial partial pressure of CO at 35–
#
45 mmHg (4.7–6.0 kPa). Anaesthesia was monitored
continuously and, at signs of lightening, was maintained
by administration of propofol (6–12 mg\kg, Zeneca
Pharmaceuticals) and alfentanil (50 mg\kg, Janssen-Cilag
Ltd). Pulmonary artery occlusion pressure was maintained at 10–12 mmHg by intravenous infusion of colloid
solution (6 % hespan or pentafract, Dupont Pharmaceuticals Ltd, Letchworth, Herts, U.K.) to compensate
for peritonitis-induced fluid loss from the blood. This
study was authorized by the U.K. Secretary of State
under the provisions of the Animals (Scientific Procedures) Act 1986.
Tissue removal and processing
Eight hours after the induction of anaesthesia, the pigs
were killed by injection of a lethal dose of phenobarbitone (0.6 ml\kg, Animal Care Ltd, York, U.K.). A 3-cm
hole was bored through the frontal bone, the meninges
were incised and reflected and a fixative solution (3 %
glutaraldehyde in 0.1 mol\l cacodylate buffer, pH 7.4 at
room temperature) was poured on to the surface of the
brain. A block of frontal cortex (approximately 1 cm$)
extending from the pial surface to the underlying white
matter was removed, cut into 1-mm thick strips per# 1999 The Biochemical Society and the Medical Research Society
pendicular to the brain surface and immersed in fixative
for 1 h. The tissue was then rinsed in cacodylate buffer
and post-fixed in cacodylate buffered 2 % osmium
tetroxide for 2 h. It was then washed in cacodylate buffer,
immersed in 30 % ethanol containing 2 % uranyl acetate
for 30 min, dehydrated through graded alcohols and
processed into Spur’s resin (Agar Scientific Ltd, Stansted,
Essex, U.K.). The tissue blocks were then coded so that
subsequent analysis could be performed ‘ blind ’.
Light microscopy and image analysis
Semi-thin sections (2 µm) of the embedded tissue were
cut, picked up on glass slides, stained with a mixture of
0.5 % Toluidine Blue, 0.5 % Methylene Blue and 1 %
boric acid for 1 min, washed, dried, placed on cover slips
and then viewed in a light microscope. Twenty images
were captured from the tissue of each pig and imported
into an image analysis system (Macintosh Performa 630)
running NIH image software (public domain). Each
image contained a microvessel (cerebral vessel with a
diameter of 10 µm) in cross-section. The following
measurements were made in triplicate from each image :
A, area of oedema surrounding the microvessel ; B, total
microvessel area enclosed by the basement membrane ;
and C, area of microvessel lumen. Mean values were
obtained and then endothelial cell area (BkC) and the
percentage microvessel area occupied by oedema (A\B
i100 %) were calculated for each image. The experimental codes were broken and mean values calculated
for each pig.
Electron microscopy
Thin sections (80–100 nm) were cut from the resinembedded tissue blocks perpendicular to the surface of
the cortex, picked up on copper grids and stained with
uranyl acetate [15] for 20 min followed by Sato’s lead
stain [16] for 20 min. The sections were examined in a
Zeiss 900 electron microscope and photomicrographs
were taken of tissue from each pig.
Statistical analysis
Values are expressed as meanspS.E.M. The data from
septic and sham pigs were compared using Student’s ttest.
RESULTS
Light microscopy
Oedema was present around cortical microvessels but
not around larger vessels. There was significantly (t l
3.34, P l 0.01) more perimicrovessel oedema in the
frontal cortices of pigs with faecal peritonitis than in
sham controls (Figure 1). Faecal peritonitis had no effect
Cerebral oedema and neuronal injury in sepsis
swollen and a few cytoplasmic vacuoles were occasionally
present (Figure 2F). The morphology of synapses visible
in both septic and sham pigs appeared normal.
DISCUSSION
Figure 1 Effect of sepsis on the mean cross-sectional areas
of microvessel lumina (Lu), microvessel endothelial cells (En)
and perimicrovessel oedema (Oe) in the frontal cerebral
cortex of pigs 8 h after the induction of faecal peritonitis,
compared with sham-operated controls
n l 5 for both experimental groups. *P
0.05.
on the cross-sectional area of microvessel endothelial
cells or their lumina (Figure 1). The oedema present was
equivalent to 160.8p21.3 % (meanpS.E.M.) of the total
microvessel cross-sectional area in septic pigs and
84.20p10.5 % in sham controls (t l 3.32, P l 0.012). No
inflammatory cells or lumen occlusion was seen in either
group.
Electron microscopy
Perimicrovessel oedema was evident in both experimental
groups, but it was far more marked in septic pigs (Figures
2A and 2B) than in sham controls (Figure 2C). Microvascular endothelial cells appeared normal with morphologically intact mitochondria and intercellular tight
junctions (Figure 2D) in all specimens. However, extravasated erythrocytes were occasionally observed in the
cortical parenchyma of septic pigs. Within the lumina of
cerebral microvessels from sham pigs, erythrocytes appeared normal (Figure 2C), but in septic pigs they were
swollen and more spherical (Figure 2B). Astrocyte endfeet were occasionally slightly swollen in sham pigs, but
in septic pigs they were frequently grossly swollen and
their perimicrovessel membranes ruptured (Figure 2B).
Scattered, apparently degenerating neurons were found
in irregular clusters within the frontal cortex of septic
pigs (Figure 2A). These neurons were dark, shrunken,
contained pyknotic nuclei, multiple vacuoles and degenerating mitochondria (Figure 2E). The remaining
neurons in the frontal cortex of pigs with faecal peritonitis
appeared to have more ‘ dilute ’ cytoplasm than those in
sham pigs. In sham pigs, the neuronal perikarya appeared
normal, although some of their mitochondria were
Faecal peritonitis resulted in extensive perimicrovessel
oedema in porcine frontal cortex. Although sham pigs
exhibited some oedema, probably resulting from 8 h of
anaesthesia and immobilization, there was 49.5 % more
oedema in septic pigs than in sham counterparts. The
oedema in septic pigs was associated with gross swelling
of perivascular astrocyte end-feet and their rupture and
detachment from microvessel endothelial cell walls. Such
swelling is known to result from disruption of the
blood–brain barrier [17]. This perivascular brain oedema
is likely to affect the passage of oxygen, nutrients and
cellular waste between the blood and brain parenchyma
and thus contribute to the symptoms observed in SE. In
addition to their role in maintaining the blood–brain
barrier, astrocytes transport energy substrates from the
blood to neurons in proportion to the level of synaptic
activity [18]. Therefore, the injury to astrocyte end-feet
in SE may impair this metabolic coupling and adversely
affect synaptic activity.
Dark, shrunken neurons were present in the frontal
cortices of pigs after only 8 h of experimental peritonitis.
This is unlikely to be an artefactual change, since such
neurons were not evident in similarly processed tissue
from sham-operated pigs. The cytoplasm of remaining
neurons in the frontal cortices of septic pigs appeared
‘ dilute ’. These observations suggest that factors associated with the oedema or its causative agent are
cytotoxic. It is not known whether the neuronal damage
that occurs in the pig model also occurs in patients with
SE or if long-term neurological\neuropsychiatric problems emerge in patients who recover from sepsis. These
possibilities require investigation because the effects of
SE in humans are currently considered to be reversible
[19,20]. The dark neurons present after 8 h in the porcine
model of sepsis probably represent early degenerative
alterations that could be reversible [21–23]. However,
since the duration of sepsis is considerably longer in
critically ill patients than in the pig model, their neuronal
damage may be even more severe.
The damage to the cerebral cortex resulting from 8 h of
sepsis in the current study was in some ways less than
that reported for hepatic sinusoids in the pig model 5 h
after the induction of faecal peritonitis [6]. The hepatic
sinusoids were found to be occluded by degranulating
neutrophils, lymphocytes and activated Kupffer cells and
there was endothelial hypertrophy and obliteration of the
space of Disse [6]. Leucocyte exudation into the
parenchyma of the frontal cortices of septic pigs was not
observed in the current study, in agreement with the
# 1999 The Biochemical Society and the Medical Research Society
463
464
M. C. Papadopoulos and others
Figure 2 Electron micrographs showing the effect of sepsis on the frontal cerebral cortex 8 h after the induction of faecal
peritonitis
(A) Perimicrovessel oedema (po) and dark neurons (dn) in a pig with faecal peritonitis. i1260. (B) A cerebral microvessel in a pig with sepsis, showing perimicrovessel
oedema (po) comprising swollen and disrupted astrocytic end-feet. The microvessel endothelial cell (end) surrounded by a basement membrane (bm) appears normal,
but within the vessel lumen, the erythrocytes [red blood cells (rbc)] can be seen to be swollen. i3200. (C) A cerebral microvessel in a sham-operated pig. The endothelial
cell (end) basement membrane (bm) is lined by non-oedematous astrocyte foot processes (afp). i3200. (D) Intact trilaminar junctions (arrows) between cerebral
microvessel endothelial cell (end) processes, ending in a juxta-luminal zonula occludens (zo) in a septic pig. The microvessel lumen contains an erythrocyte [red blood
cell (rbc)] and perimicrovessel oedema (po) is present outside the endothelial cell basement membrane (bm). i38 000. (E) A dark neuron in a septic pig containing
a pyknotic nucleus (pn), numerous vacuoles and degenerating mitichondria. i3000. (F) A neuron in a sham-operated pig with a normal nucleus (n), slightly swollen
mitochondria and a few vacuoles. i3000.
# 1999 The Biochemical Society and the Medical Research Society
Cerebral oedema and neuronal injury in sepsis
previous observation that intracerebral injection of various chemotactic cytokines fails to cause leucocyte infiltration of murine brain parenchyma [24]. During
sepsis, cytokines such as tumour necrosis factor α and
interferon β are released from macrophages and lymphocytes and play a major role in the initiation, propagation,
regulation and suppression of immune and inflammatory
responses [25,26]. Normal central nervous parenchyma is
an immunologically privileged site and is uniquely
resistant to leucocyte accumulation [27]. Moreover,
although septic shock is associated with a reduction in the
perfusion of many organs [4], the brain may be protected
from hypoperfusion by the selective redistribution of
blood flow from other parts of the body [28]. However,
in the current study, the erythrocytes of septic pigs were
often enlarged and rounded. This finding is in agreement
with those of previous studies [29,30] which showed that,
in sepsis, activated leucocytes generate oxygen free
radicals that damage the erythrocyte plasma membranes.
Damaged erythrocytes may be unable to squeeze through
cerebral microvessels and thus impair cerebral perfusion
in sepsis.
Since extensive liver damage occurs in the pig model of
sepsis [6], the cerebral changes observed in the current
study may be secondary to hepatic failure, a wellrecognized cause of encephalopathy [31]. Perivascular
cerebral oedema and damage to astrocyte end-feet have
been reported both in patients who died from fulminant
hepatic failure [32] and in rabbits with galactosamineinduced hepatic failure [33]. Both hepatic failure [34] and
sepsis [25,26] are characterized by high levels of circulating cytokines. The results of experiments using
cultures of bovine [35] and human [36] brain endothelium
suggest that cytokines increase blood–brain barrier permeability. In our experiments, the intercellular tight
junctions appeared intact. In an early study using rabbits,
Clawson et al. [14] found that endotoxaemia caused brain
oedema without apparently damaging the tight junctions.
Addition of tumour necrosis factor α to bovine cerebral
endothelial cultures also increased permeability without
affecting the morphology of tight junctions [35].
Faecal peritonitis is common, and many patients with
this problem require ventilation and the administration
of intravenous colloids to maintain constant pulmonary
artery occlusion pressures. Pigs were used in the current
study since they are of comparable size to humans. Their
size allows monitoring of haemodynamic variables [37]
in a similar way to that of patients in an Intensive Care
Unit. Haemodynamic parameters are difficult to measure
in rodents, which are also relatively resistant to sepsis
from their own faeces [37]. The similarities between our
experimental model and the clinical situation suggest that
the perivascular cerebral oedema present in septic pigs
may also occur in septic patients in an Intensive Care
Unit. The systemic effects of sepsis may arise not only
from faecal peritonitis but also from other infections [38].
Since encephalopathy may develop in septic patients with
Gram-negative bacteraemia, Gram-positive bacteraemia,
fungaemia and no causative organism identified [11], our
findings may be relevant to all septic patients, not only
those with faecal peritonitis.
Although faecal peritonitis results in perimicrovessel
oedema, disruption of astrocyte processes, neuronal
degeneration and erythrocyte damage in the frontal
cortices of pigs, the mechanisms by which they occur
remain to be elucidated. Endothelial cell tight junctions
appear morphologically intact in sepsis, but their functional status is not clear. Since the molecular structure of
the tight junction is known [39], immunohistochemistry
using antibodies against occludins may show whether
these glue-like tight junction proteins are disrupted in
sepsis, allowing microvessel leakage and the development
of oedema. If the junctions remain functionally ‘ tight ’ in
sepsis, then some other route such as pinocytosis must be
involved in the formation of perimicrovessel oedema. It is
not known whether the factors causing the oedema
directly are responsible for the other pathological changes
present in the septic pig frontal cortex or whether they
develop in response to the oedema. Whatever the cause,
they are likely to contribute to the symptomatology of
SE, and it is important to determine whether the different
inotropic agents commonly used to treat patients with
septic shock modulate the effects of sepsis on the brain.
ACKNOWLEDGMENT
We would like to thank the staff of the Biological
Research Facility, St. George’s Hospital Medical School,
for their invaluable, skilled technical assistance.
REFERENCES
1 Parrillo, J. E., Parker, M. M., Natanson, C. et al. (1990)
Septic shock in humans. Advances in the understanding of
pathogenesis, cardiovascular dysfunction and therapy.
Ann. Int. Med. 113, 227–242
2 Knaus, W. A., Draper, E. A., Wagner, D. P. and
Zimmerman, J. E. (1985) Prognosis in acute organ-system
failure. Ann. Surg. 202, 685–693
3 Coalson, J. J., Hinshaw, L. B., Guenter, C. A., Berrell,
E. L. and Greenfield, L. J. (1975) Pathophysiologic
responses of the subhuman primate in experimental septic
shock. J. Lab. Invest. 32, 561–569
4 Lang, C. H., Bagby, G. J., Ferguson, J. L. and Spitzer, J. J.
(1984) Cardiac output and redistribution of organ blood
flow in hypermetabolic sepsis. Am. J. Physiol. 246,
R331–R337
5 Tighe, D., Moss, R., Boghossian, S., Heath, F. M.,
Chessum, B. and Bennett, E. D. (1989) Multi-organ
damage resulting from experimental faecal peritonitis. Clin.
Sci. 76, 269–276
6 Tighe, D., Moss, R., Hynd, J. et al. (1990) Pretreatment
with pentoxifylline improves the hemodynamic and
histologic changes and decreases neutrophil adhesiveness in
a pig fecal peritonitis model. Crit. Care Med. 18, 184–189
7 Tighe, D., Moss, R., Heywood, G., Al-Saady, N., Webb,
A. and Bennett, D. (1995) Goal-directed therapy with
# 1999 The Biochemical Society and the Medical Research Society
465
466
M. C. Papadopoulos and others
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
dopexamine, dobutamine, and volume expansion : effects
of systemic oxygen transport on hepatic ultrastructure in
porcine sepsis. Crit. Care Med. 23, 1997–2007
Deng, X., Wang, X. and Andersson, R. (1995) Endothelial
barrier resistance in multiple organs after septic and
nonseptic challenges in the rat. J. Appl. Physiol. 78,
2052–2061
Tighe, D., Moss, R. and Bennett, D. (1998) Porcine hepatic
response to sepsis and its amplification by an adrenergic
receptor α agonist and a β antagonist. Clin. Sci. 95,
#
467–478 "
Bradley, J. R., Wilks, D. and Rubenstein, D. (1994) The
vascular endothelium in septic shock. J. Infect. 28, 1–10
Sprung, C. L., Peduzzi, P. N., Shatney, C. H. et al. (1990)
Impact of encephalopathy on mortality in the sepsis
syndrome. Crit. Care Med. 18, 801–806
Young, C. B., Bolton, C. F., Archibald, Y. M., Austin,
T. W. and Wells, G. A. (1992) The electroencephalogram
in sepsis-associated encephalopathy. J. Clin. Neurophysiol.
9, 145–152
Jeppson, B., Freund, H. R., Gimmon, Z., James, J. H., von
Meyenfeldt, M. F. and Fischer, J. E. (1981) Blood–brain
barrier derangement in sepsis : cause of septic
encephalopathy ? Am. J. Surg. 141, 136–141
Clawson, C. C., Hartmann, J. F. and Vernier, R. L. (1966)
Electron microscopy of the effect of gram-negative
endotoxin on the blood–brain barrier. J. Comp. Neurol.
127, 183–198
Lombardi, L., Prenna, G., Okolicsanyi, L. and Gautier, A.
(1971) Electron staining with uranyl acetate. Possible role
of free amino groups. J. Histochem. Cytochem. 19, 161
Sato, T. (1968) A modified method for lead staining of thin
sections. J. Electr. Microsc. (Tokyo) 17, 158–159
Norenberg, M. D. (1994) Astrocyte responses to CNS
injury. J. Neuropathol. Exp. Neurol. 53, 213–220
Tsacopoulos, M. and Magistretti, P. J. (1996) Metabolic
coupling between glia and neurons. J. Neurosci. 16,
877–885
Bolton, C. F., Young, G. B. and Zochodne, D. W. (1993)
The neurological complications of sepsis. Ann. Neurol. 33,
94–100
Chen, R. and Young, G. B. (1996) Metabolic
encephalopathies. Ballie' re Clin. Neurol. 5, 577–598
So$ derfeldt, B., Kalimo, H., Olsson, Y. and Siesjo$ , B. K.
(1983) Bicuculline-induced epileptic brain injury. Transient
and persistent cell changes in rat cerebral cortex in the
early recovery period. Acta Neuropathol. (Berlin) 62,
87–95
Auer, R. N., Kalimo, H., Olsson, Y. and Siesjo$ , B. K.
(1985) The temporal evolution of hypoglycemic brain
damage. I. Light and electron microscopic findings in the
rat cerebral cortex. Acta Neuropathol. (Berlin) 67, 13–24
Auer, R. N., Kalimo, H., Olsson, Y. and Siesjo$ m, B. K.
(1985) The temporal evolution of hypoglycemic brain
damage. II. Light and electron microscopic findings in the
rat hippocampus. Acta Neuropathol. (Berlin) 67, 25–36
24 Andersson, P. B., Perry, V. H. and Gordon, S. (1992)
Intracerebral injection of proinflammatory cytokines or
leucocyte chemotaxins induces minimal myelomonocytic
cell recruitment to the parenchyma of the central nervous
system. J. Exp. Med. 176, 255–259
25 Goris, R. J. (1990) Mediators of multiple organ failure.
Intensive Care Med. 16, S192–S196
26 Tighe, D., Moss, R. and Bennett, E. D. (1996) Cell surface
adrenergic receptor stimulation modifies the endothelial
response to SIRS. Systemic Inflammatory Response
Syndrome. New Horiz. 4, 426–442
27 Perry, V. H., Andersson, P. B. and Gordon, S. (1993)
Macrophages and inflammation in the central nervous
system. Trends Neurosci. 16, 268–273
28 Laptook, A. R., Stonestreet, B. S. and Oh, W. (1983) Brain
blood flow and O delivery during hemorrhagic
hypotension in the# piglet. Pediatr. Res. 17, 77–80
29 Todd, J. C., Poulos, N. D., Davidson, L. W. and Mollitt,
D. L. (1993) Role of the leukocyte in endotoxin-induced
alterations of the red cell membrane. Am. Surg. 59, 9–12
30 Langenfeld, J. E., Machiedo, G. W., Lyons, M., Rush, Jr,
B. F., Dikdan, G. and Lysz, T. W. (1994) Correlation
between red blood cell deformability and changes in
hemodynamic function. Surgery 116, 859–867
31 Mousseau, D. D. and Butterworth, R. F. (1994) Current
theories on the pathogenesis of hepatic encephalopathy.
Proc. Soc. Exp. Biol. Med. 206, 329–344
32 Kato, M., Hughes, R. D., Keays, R. T. and Williams, R.
(1992) Electron microscopic study of brain capillaries in
cerebral edema from fulminant hepatic failure. Hepatology
15, 1060–1066
33 Traber, P. G., Dal Canto, M., Ganger, D. R. and Blei,
A. T. (1987) Electron microscopic evaluation of brain
edema in rabbits with galactosamine-induced fulminant
hepatic failure : ultrastructure and integrity of the
blood–brain barrier. Hepatology 7, 1272–1277
34 Duchini, A. (1996) The role of central nervous system
endothelial cell activation in the pathogenesis of hepatic
encephalopathy. Med. Hypotheses 46, 239–244
35 Anda, T., Yamashita, H., Tsutsumi, K., Fujita, H.,
Tokunaga, Y. and Shibata, S. (1997) Effect of tumor
necrosis factor-alpha on the permeability of bovine brain
microvessel endothelial cell monolayers. Neurol. Res. 19,
369–376
36 Huynh, H. K. and Dorovini-Zis, K. (1993) Effects of
interferon-gamma on primary cultures of human brain
microvessel endothelial cells. Am. J. Pathol. 142,
1265–1278
37 Wichterman, K. A., Baue, A. E. and Chaudry, I. H. (1980)
Sepsis and septic shock – a review of laboratory models
and a proposal. J. Surg. Res. 29, 189–201
38 Eiderman, L. A., Putterman, D., Putterman, C. and
Sprung, C. L. (1996) The spectrum of septic
encephalopathy. Definitions, etiologies, and mortalities. J.
Am. Med. Assoc. 275, 470–473
39 Denker, B. M. and Nigam, S. K. (1998) Molecular
structure and assembly of the tight junction. Am. J.
Physiol. 274, F1–F9
Received 21 September 1998/4 January 1999; accepted 7 January 1999
# 1999 The Biochemical Society and the Medical Research Society