Content uploaded by Astrid Hjelde
Author content
All content in this area was uploaded by Astrid Hjelde
Content may be subject to copyright.
ORIGINAL ARTICLE
Vibeke Nossum á Astrid Hjelde á Alf O . Brubakk
Small amounts of venous gas embolism cause delayed
impairment of endothelial function and increase
polymorphonuclear neutrophil in®ltration
Accepted: 14 September 2001 / Published online: 10 November 2001
Ó Springer-Verlag 2001
Abstract Gas bubbles from decompression and gas
embolization lead to endothelial dysfunction and me-
chanical injury in the pig, rabbit and lamb. In the study
presented here, 0.01 ml air/min/kg was infused through a
catheter into the jugular vein in 12 rabbits for 60 min.
The endothelial response was measured using tension
measurements in the blood vessel wall, and morpholog-
ical changes where quanti®ed using light microscopy and
image processing. Percent lung water content was cal-
culated and used to estimate the severity of pulmonary
oedema. The infusion led to a signi®cant decrease in the
acetylcholine-mediated endothelial-dependent vasodila-
tation in the pulmonary artery 6 h after the infusion (6-h
group, n=6). A decrease in substance-P-mediated en-
dothelial-dependent vasodilatation was also detected.
No changes where seen in a group of rabbits examined
1 h after infusion (1-h group, n=6). The impaired en-
dothelial-dependent vasodilatation caused by the bub-
bles is probably biochemical in origin, since no visible
changes were seen in the endothelial layer. A signi®cant
increase in polymorphonuclear neutrophils was observed
in the 6-h group compared to the 1-h group. This study
demonstrates that small numbers of bubbles, corre-
sponding to ``silent bubbles'', lead to an impairment of
the endothelial-dependent vasoactive response.
Keywords Pulmonary artery á Neutrophils á
Endothelial cells á Vascular bubbles á Lung oedema
Introduction
Intravascular gas bubbles occur in the venous system
during most decompressions (Brubakk et al. 1986;
Eckenho et al. 1990). Venous gas bubbles may also
develop following laparoscopy, by accidental injection
or in cardiopulmonary bypass surgery (Hoka et al. 1997;
Johnston et al. 1993; Webb et al. 1997), and they lead to
endothelial damage (Nossum et al. 1999; Philp 1974;
Warren et al. 1973). As the number of gas bubbles
increases, the likelihood of endothelium damage also
increases, and such damage seems to be related to the
amount of gas present (Nossum et al. 1999).
The amount of endothelial damage resulting from
gaseous microemboli may be important because endo-
thelial cells are the source of many vasoactive factors,
including nitric oxide (NO). NO regulates the diameter
of blood vessels and blood ¯ow and it is an important
mediator of pulmonary vascular tone (Busse et al. 1993;
Stewart and Baour 1990). After increased shear stress
or binding of vasodilators, such as endothelium-depen-
dent vasodilators, to surface receptors, vascular endo-
thelial cells synthesize NO via activation of the enzyme
endothelial NO synthase (eNOS; Mu
È
lsch et al. 1989;
Rubanyi et al. 1986). NO then diuses into the under-
lying vascular smooth muscle cells, activates soluble
guanylate cyclase and initiates a cascade resulting in
smooth muscle relaxation (Fiscus 1988). If the endo-
thelial lining becomes disrupted or damaged by gas
emboli, endothelium-dependent vasodilatation could be
depressed. This may cause a reduction in regional per-
fusion (Helps et al. 1990) and an exaggerated response to
vasoconstrictor agents (Busse et al. 1993; Ku 1987;
Stewart and Baour 1990). A reduced production of
NO, a decrease in the density or number of receptors for
acetylcholine and substance P, or an altered function of
these receptors, can lead to loss of endothelium-depen-
dent pulmonary vasodilatation. Alternatively, there may
be an increased degradation of NO or inhibition of
eNOS activity (Fineman et al. 1999).
To determine whether infusion of air through the
heart and into the pulmonary artery impairs the endo-
thelium-dependent regulation of pulmonary vascular
tone, we studied the response to endothelium-dependent
vasodilators using an in vitro method for isolated
Eur J Appl Physiol (2002) 86: 209±214
DOI 10.1007/s00421-001-0531-y
V. Nossum (&) á A. Hjelde á A.O. Brubakk
Department of Physiology and Biomedical Engineering,
Norwegian University of Science and Technology,
7489 Trondheim, Norway
E-mail: vibeke.nossum@medisin.ntnu.no
Tel.: +47-73550553
Fax: +47-73598613
vessels. We compared the changes in endothelial-de-
pendent vasodilatation to acetylcholine and substance P,
and endothelial-independent vasodilatation to sodium
nitroprusside after 60 min of air infusion in two groups
of animals, 1 and 6 h after infusion. The endothelial
layer was examined using light microscopy to evaluate
possible mechanical damage to the endothelial lining.
Methods
Twelve locally bred New Zealand Black rabbits of both genders,
weighing 2.8±3.4 kg, were used. No signs of illness were detected in
these animals during the study period. The experiments were per-
formed in accordance with the Principles of laboratory animal care
(NIH Publication NO 85±23 revised 1985). The experimental pro-
tocol was approved by the Norwegian Committee for Animal
Experiments.
Anaesthesia
The rabbits were tranquillized with an intramuscular injection of
midazolam (5 mg; Dormicum, Homann-La Roche, Basel,
Switzerland) and ¯uanisone (7 mg) + fentanyl (0.22 mg; Hypnorm,
Janssen-Cilag, Saunderton, Buckinhamshire, UK). Within 60 min
the rabbits received an additional intramuscular injection of half of
the above dose of both midazolam and fentanyl/¯uanisone. Thir-
ty minutes before the observation period, the rabbits were given an
intramuscular injection of Buprenor®n (0.02±0.05 mg/kg; Temgesic,
Reckitt and Colman). Body temperature was measured with a
rectal probe and kept at 39.0 (0.5)°C by means of a heating blanket.
The animals were allowed to breathe spontaneously throughout the
experiment.
Infusion
A catheter (0.36 mm inner diameter) was placed into the jugular
vein and moved centrally. Air was infused for 60 min by a syringe
in a special build pump. The amount of air infused was 0.01 ml/kg/
min, corresponding to 8±10 bubbles/min. Blood-gases (oxygen and
carbon dioxide tension, PO
2
and PCO
2
, respectively) and pH were
monitored before, during and 1 and 6 h after air infusion (1-h
group and 6-h group, respectively).
Bubble detection
Gas bubbles were detected in the heart using a 5-MHz transducer
connected to an ultrasonic scanner (750 Vingmed, Horten, Nor-
way). The number of gas bubbles was evaluated using a grading
system from 0 to 5 (Eftedal et al. 1994): grade 0 is no bubbles, 1
represents an occasional bubble, 2 represents at least one bubble
every fourth heart cycle, 3 is at last one bubble every heart cycle, 4
is continuous bubbling, and 5 is massive bubbling. This scoring
system is approximately exponential compared with the number of
bubbles in the right ventricle (Eftedal et al. 1994). The grades ob-
served were converted to bubbles/cm
2
using the conversion table
given by Eftedal et al. (1998).
Observation period
The rabbits were divided into two groups of six animals, the 1-h
group was observed for 1 h after air infusion, and the 6-h group
was observed for 6 h after air infusion. After the observation pe-
riod, the animals were given a lethal intravenous dose of potassium
chloride, under anaesthesia (midazolam 5 mg and ¯uanisone 7 mg
+ fentanyl 0.22 mg). The lungs were immediately harvested.
Wet-dry weight of the lungs
The dry weight of the lung tissue was determined from a less than
1-g section of the left lung. The tissue was weighed (wet weight),
incubated at 120°C for 7 days, and then weighed again (dry
weight). Percent lung water content [(wet weight ± dry weight)/wet
weight´100] was used to estimate the severity of pulmonary
oedema.
Lung histology
From all rabbits selected samples from the right and left upper and
lower lung were ®xed in a solution consisting of 70% ethanol, 4%
formaldehyde and 5% acetic acid. Four samples per animal were
taken. On the next day the specimens were transferred to 80%
ethanol, before dehydration and embedding in paran for histo-
pathology. Sections were cut at 5 lm and stained with haemat-
oxylin-eosin-safran. An investigator who was blinded to treatment
estimated the accumulation of polymorphonuclear neutrophils
(PMNs) present in the tissue. Four ®elds from each lung section
were examined at ´400 with the aid of a Nikon YS2-H light mi-
croscope. This microscope was equipped with an eyepiece con-
taining a 10´10 graticule grid (0.5´0.5 cm). The number of grid
points falling on the tissue determined the ®eld area. By counting
the total number of PMNs in that ®eld divided by the number of
lung tissue grid points, the number of PMNs per unit lung tissue
was calculated. Each rabbit was represented by a mean value of
eight data points from both the left and right lung. The results are
expressed as mean number of PMNs per unit lung tissue, and one
unit lung tissue=0.25 mm
2
.
Tension measurements of isolated vessels
A modi®ed tissue-bath technique (Edvinsson et al. 1974; Ho
È
gesta
È
tt
et al. 1983) was used, as described previously by (Nossum et al.
1999, 2000). The pulmonary artery was carefully dissected from the
right lung with the aid of a dissection microscope. The vessels were
cut into cylindrical segments with length ranging from 1.0±1.5 mm
and with a diameter between 1 and 2 mm. Each cylindrical segment
was mounted on two parallel L-shaped metal prongs and immersed
in temperature-controlled (37°C) tissue baths containing a sodium-
Krebs buer of the following composition: 119 mM NaCl, 10 mM
NaHCO
3
, 1.2 mM MgCl
2
, 4.6 mM KCl, 1 mM NaH
2
PO
4
, 1.5 mM
CaCl
2
, and 11 mM glucose. Air comprising 5% CO
2
in O
2
was
bubbled continuously through the sodium-Krebs buer to keep it
at pH 7.4. The contractile capacity of each vessel segment was
examined by exposure to a potassium-rich (60 mM) Krebs buer
solution. The vessels were pre-contracted with cumulative doses of
noradrenaline and the relaxation response was tested with cumu-
lative doses of acetylcholine (10
±9
±10
±4
M) and substance P (10
±12
±
10
±7
M). The response depended upon how much of the endothelial
layer was damaged by the bubbles. The maximum relaxation re-
sponse (I
max
) was de®ned as the maximal dilatory response re-
gardless of the concentration induced by an agonist, and is
expressed as a percentage of the pre-contraction induced by a pre-
contracting agent. The performance of the vascular smooth muscle
cells was evaluated with cumulative doses of sodium nitroprusside
(10
±9
±10
±5
M). In addition, dose-response curves for all agonists
were calculated.
Silver nitrate staining
The segments were cut open in strip form and mounted carefully
with needles on a Para®lm-covered cork plate with the vessel lu-
men-side (endothelial layer) up. The mounted segments were
stained with silver nitrate using a method described by (Abrol et al.
1984). The stained segments were transferred to an object glass and
mounted using an aqueous mounting medium.
210
Microscopy and photography
Each segment was examined by light microscopy (Nikon Micro-
photo-FXA ¯uorescence microscopy) and photographed (Nikon
FX-35DX ) at ´250 (Fuji®lm ISO 100). Since each photograph only
partly covered the segment, several photographs of each segment
were taken. In order to claim reproducible and reliable results, the
photographs were taken in the same manner for each segment
(three or ®ve pictures). The photographs were also taken in the
same pattern in order to be as representative as possible, usually
three pictures from each segment.
Quanti®cation of endothelial damage
Endothelial damage was evaluated using an image-processing
program (Adobe Photoshop 5.0). All photographs from each seg-
ment were scanned into the computer. The area containing dam-
aged endothelial layer was coloured using the paint bucket function
to increase the contrast and to simplify the quanti®cation. The level
of damage was calculated from the pixel dimensions of the marked
and stained area (re¯ecting the endothelial rupture) and compared
to the pixel dimension of the whole picture (expressed as a per-
centage). This procedure was followed for all of the photographs
from each segment. The mean value for each animal was calculated
from the mean value of every segment. The value is a result of at
least 12 photographs and represents the ®nal percentage of endo-
thelial damage for this vessel.
Drugs
()Noradrenaline[+]-hydrogen-tartrate, substance P, acetylcho-
line and sodium nitroprusside-dihydrate (all Sigma) were dissolved
in saline or small amounts of distilled water. All concentrations
given are the ®nal molar concentration in the tissue bath during the
experiments.
Statistics
Data were subjected to analysis using the Mann-Whitney U and
Wilcoxon signed-rank tests for unpaired and paired data, as ap-
propriate. The level of statistical signi®cance was set at P<0.05.
The results are expressed as mean (SD).
Results
Pulmonary artery bubbles
Eight to 10 bubbles/min were infused into the superior
caval vein and corresponded to grade 1±2 when detected
with a ultrasonic scanner in the pulmonary artery. There
was no signi®cant dierence in this parameter between
the 1-h and 6-h groups: 0.08 (0.03) bubbles/cm
2
(1-h
group) and 0.09 (0.02) bubbles/cm
2
(6-h group). All
rabbits survived the infusion and observation period.
The measurement of blood-gases showed no change in
PCO
2
for the animals during the air infusion compared
to that observed prior to and after infusion.
Relaxation response
The 6-h group showed a lower (P=0.04) I
max
response
[48.3 (22.4)%] to acetylcholine compared to the 1-h
group [74.5 (15.5)%; Table 1]. From the dose-response
curves, it appears that the dose-related relaxation re-
sponse was lower at 10
±7
M for the 6-h group compared
to the 1-h group (P=0.003; Fig. 1). Although the dif-
ference was not signi®cant, a lower response was seen for
all other concentrations of acetylcholine in the 6-h group
compared to the 1-h group.
There was a lower I
max
in the 6-h group compared to
the 1-h group for substance P (Table 1). However, the
dierence between the 6-h group [42.9 (16.8)%] and the
1-h group [60.4 (10.)7%] was not statistically signi®cant.
Dose-response curves showed lower relaxation responses
for the 6-h group compared to the 1-h group at two
Table 1 Comparison of values for animals observed 1 h after the
infusion of air into the jugular vein (1-h group) and for those ob-
served 6 h after the infusion (6-h group). Maximum %-relaxation
values (I
max
) for acetylcholine, substance P and sodium nitro-
prusside, wet weight, polymorphonuclear leucocyte (PMN) in®l-
tration and mechanical damage for the 1-h and 6-h groups. Values
are presented as the mean (SD) in each group
Experimental group I
max (%)
Wet weight (%) PMN/unit lung
tissue
Mechanical
damage (%)
Acetylcholine Substance P Sodium nitroprusside
1-h group (n=6) 74.5 (15.5) 60.4 (10.7) 90.7 (5.9) 80.55 (2.65) 0.0597 1.9 (1.2)
6-h group (n=6) 48.3 (22.4)* 42.9 (16.8) 83.9 (9.5) 79.87 (1.00) 0.1202** 1.6 (0.5)
*P=0.04
**P=0.0004
Fig. 1 Dose-response curves for the responses to acetylcholine
(ACh, solid lines), substance P (SP, dotted lines) and sodium
nitroprusside (SNP, dashed lines) in the animals observed 1 h after
the infusion of air into the jugular vein (the 1-h group, n=6) and in
those observed 6 h after the infusion (the 6-h group, n=6)
211
concentrations (P=0.003; Fig. 1). Lower responses were
seen for every concentration in the 6-h group compared
to the 1-h group.
Application of the endothelial-independent agonist,
sodium nitroprusside resulted in no signi®cant dier-
ences in I
max
between the two groups. The dose-response
curves did not show any dierences between the two
groups (Fig. 1).
Wet weight and PMN in®ltration
The pulmonary water content was 80.6 (2.7)% in the 1-h
group and 79.9 (1.0)% in the 6-h group (Table 1). The
dierence was not signi®cant.
The results of the histological examination (left and
right lung) from the 1-h and the 6-h group are given in
Fig. 2. No signi®cant dierence was found between the
left and right lung, for both the 1-h and the 6-h group.
However, an increase in PMNs was observed in the 6-h
group compared to the 1-h group (P=0.004; Fig. 2).
Mechanical damage
Evaluation of the endothelial layer by light microscopy
did not reveal any mechanical damage for the two
groups (Fig. 3). The percent damage for the 1-h group
was 1.9 (1.2)%, and did not dier signi®cantly from that
of the 6-h group [1.6 (0.5)% mechanical damage;
Table 1].
Discussion
The results of this study show that the endothelium-
dependent response to vasoactive substances in the
pulmonary artery would change 6 h after the infusion of
small amounts of air bubbles. Furthermore, there was an
increase in PMN in®ltration at that time. However, no
dierence in oedema formation (wet weight) was found
between the 1-h and 6-h groups. There were no signs of
mechanical damage in the pulmonary endothelial layer,
as assessed by light microscopy, indicating a biochemical
disruption to the endothelial layer.
The change in the vasoactive response occurred 6 h
after the infusion, while the response seems to have been
unaected after 1 h. The loss of endothelium-dependent
vasoactivity was reduced for the I
max
to acetylcholine,
while the response to substance P was reduced at some
concentrations (dose-response). The endothelium-inde-
pendent response to sodium nitroprusside seems to have
been unaected by air bubbles in both groups, and
con®rms that the change in vasoactive response is only
related to the endothelial function and not to function in
the vascular smooth muscle layer. Albertine et al. (1984)
and Berner et al. (1989) showed that with air emboli-
zation, the pulmonary vascular endothelium is the site of
injury.
The water content of the lungs was not dierence
between the groups, while there was a greater in®ltration
of leucocytes in the animals that survived for 6 h after
the infusion of air. Hjelde et al. (1999) found a con-
nection between the duration of observation and PMN
accumulation in animals exposed to many bubbles.
Thus, it seems that PMN in®ltration increases with time.
The number of PMNs observed after 1 h was similar to
the number of PMNs seen in control animals observed
for 2 h without exposure to bubbles (Hjelde et al. 1999).
From the light microscope analysis, it appears that in
the present study the infusion of air did not result in
damage to the endothelial layer. However, following
exposure to more bubbles (Nossum et al. 1999, 2000),
mechanical injury to the endothelium does occur, in
addition to a decreased response to endothelium-
dependent vasodilators. Yet any mechanical damage
would probably have an acute eect and would also be
observed in the 1-h group.
The dose-response curves obtained for both acetyl-
choline and substance P revealed a large dierence be-
tween the 6-h group and the 1-h group. Normally,
Fig. 2 Number of polymorphonuclear leucocytes (PMN) per unit
of lung tissue for the left (open bars) and right lungs (solid bars)
from animals in the 1-h (n=6) and 6-h (n=6) groups. Values are
mean SD
Fig. 3 Photomicrograph of an intact endothelial layer from an
animal in the 6-h group that exhibited a decreased relaxation
response (magni®cation ´ 250)
212
PMNs circulate within the vasculature as unstimulated
cells and do no damage to the vascular endothelium.
However, these cells can become activated and dra-
matically increase oxygen uptake, resulting in the pro-
duction of oxygen metabolites, lysosomal enzyme release
and subsequent endothelial damage (Fantone and Ward
1982; Roberts 1988). The surface of the bubbles acts as a
foreign substance and is capable of activating the alter-
native complement pathway in vitro (Hjelde et al. 1995;
Ward et al. 1986, 1987). During activation of the com-
plement pathway, three anaphylactic peptides are re-
leased into the ¯uid phase, with C5a being the most
important. Intravascular complement activation leads to
acute lung injury, and PMNs play a key role in this
development (Czermak et al. 1998; Till et al. 1982).
Complement-activated PMNs, when in close contact
with lung vascular endothelium, may release toxic oxy-
gen metabolites that can destroy the endothelium (Sacks
et al. 1978; Tofukuji et al. 1998). Gaseous microemboli
can cause direct vascular injury as a result of transient
capillary obstruction (Feinstein et al. 1984).
Complement-activated PMNs are associated with
the production and release of highly reactive oxygen
species such as superoxide anion (O
2
±
·), hydrogen
peroxide (H
2
O
2
) and hydroxyl radical (OH·; Fantone
and Ward 1982; Roberts 1988) which, when in close
contact with lung vascular endothelium, can destroy
the endothelium (Sacks et al. 1978; Varani et al. 1985).
Hjelde et al. (1999) demonstrated an increase in pul-
monary neutrophil accumulation over 2 h in decom-
pressed rabbits (Hjelde et al. 1999). The activation of
PMNs leads, through a cascade of events, to the for-
mation of ONOO
±
, which reduces NO. A decrease in
NO will subsequently increase the expression of the
surface adhesion molecules that are responsible for
adhesion between stimulated PMNs and the endothe-
lium, and activate more PMNs. This explains the de-
crease in endothelial response and the accumulation of
PMNs that occurs after 6 h.
The dierence in endothelium response was not sig-
ni®cant for all concentrations of acetylcholine and sub-
stance P. Two animals in the 6-h group demonstrated an
I
max
response above 60% for both substances. There are
individual dierences in the endothelial response to high
amounts of gas bubbles from decompression (Nossum
et al. 1999, 2000). Hjelde et al. (1995) demonstrated an
inter-individual dierence in complement activation
when sera from divers were incubated in the absence or
presence of air bubbles in vitro (Hjelde et al. 1995).
Bergh et al. (1993) investigated complement activation
by air bubbles in vitro and found that the responsiveness
of the complement system to air bubbles in both rabbits
and humans varies considerably.
Gas bubbles may enter the pulmonary circulation ei-
ther as a result of pressure reduction or following acci-
dents or medical procedures. If few bubbles are present,
no clinical symptoms will be evident, and such bubbles
have been termed ``silent'' bubbles. The present study,
together with that of Hjelde et al. (1999) demonstrates
that small numbers of gas bubbles would aect the en-
dothelium and lead to increased PMN in®ltration in the
lungs. Contrary to ®ndings by (Nossum et al. 1999,
2000), the small number of gas bubbles did not lead to
mechanical disruptions in the endothelial layer, as eval-
uated by light microscopy, suggesting that the changes in
endothelial function is biochemical in origin.
Acknowledgements This work was supported by the Norwegian
Petroleum Directorate, Norsk Hydro, Esso Norge and Statoil
under the ``Dive contingency contract'' (No 4600002328) with
Norwegian Underwater Intervention (NUI). The help of Anne-Lise
Ustad, Arn®nn Sira and Tove Svartkjùnnli is gratefully acknowl-
edged.
References
Abrol RP, Hughes VM, Krueger CA, Cook DA (1984) Detection
of endothelium in cerebral blood vessels. J Pharmacol Methods
12:213±219
Albertine K, Wiener-Kronish J, Koike K, Staub N (1984) Quan-
ti®cation of damage by air emboli to lung microvessels in
anesthetized sheep. J Appl Physiol 57:1360±1368
Bergh K, Hjelde A, Iversen O-J, Brubakk AO (1993) Variability
over time of complement activation induced by air bubbles in
human and rabbit sera. J Appl Physiol 74:1811±1815
Berner M, Teague W, Scheerer R, Bland R (1989) Furosemide
reduces lung ¯uid ®ltration in lambs with lung microvascular
injury from air emboli. J Appl Physiol 67:1991±1996
Brubakk AO, Peterson R, Grip A, Holand B (1986) Gas bubbles in
the circulation of divers after ascending excursions from 300 to
250 msw. J Appl Physiol 60:45±51
Busse R, Mu
È
lsch A, Fleming W, Hecker M (1993) Mechanism of
nitric oxide release from the vascular endothelium. Circulation
87:18±25
Czermak B, Lentsch A, Bless N, Schmal H, Friedl H, Ward P
(1998) Role of complement in vitro and in vivo lung in¯am-
matory reactions. J Leukoc Biol 64:40±48
Eckenho R, Olstad C, Carrod G (1990) Human dose-response
relationship for decompression and endogenous bubble for-
mation. J Appl Physiol 69:914±918
Edvinsson L, Nielsen K, Owman C (1974) In¯uence of initial
tension and changes in sensitivity during amine-induced con-
traction of pial arteries in vitro. Arch Int Pharmacodyn Ther
208:235±242
Eftedal O, Koteng S, Brubakk AO (1994) Bubble grading in ul-
trasonic images. Undersea Hyperb Med Suppl 20:79±79
Eftedal O, Brubakk AO, Nishi RY (1998) Ultrasonic evaluation of
decompression: the relationship between bubble grades and
bubble numbers. Undersea Hyperb Med Suppl 25:35±35. 1998
Fantone JC, Ward PA (1982) Role of oxygen-derived free radicals
and metabolites in leukocyte-dependent in¯ammatory reac-
tions. Am J Pathol 107:395±418
Feinstein S, Shah P, Bing R, Meerbaum S, Corday E, Chang B,
Santillan G, Fujibayashi Y (1984) Microbubble dynamics vi-
sualized in the intact capillary circulation. J Am Coll Cardiol
4:595±600
Fineman JR, Wong J, Mikhailov T, Vanderford PA, Jerome HE,
Soifer SJ (1999) Altered endothelial function in lambs with
pulmonary hypertension and acute lung injury. Pediatr Pul-
monol 27:147±156
Fiscus R (1988) Molecular mechanisms of endothelium-mediated
vasodilation. Semin Thromb Hemost 14:12±22
Helps S, Parsons D, Reilly P, Gorman D (1990) The eect of gas
emboli on rabbit cerebral blood ¯ow. Stroke 21:94±99
Hjelde A, Bergh K, Brubakk AO, Iversen O-J (1995) Complement
activation in divers after repeated air/heliox dives and its pos-
sible relevance to DCS. J Appl Physiol 78:1140±1144
213
Hjelde A, Brubakk AO, Bergh K, Videm V, Ustad AL (1999) Eect
of anti-C5a antibody on blood-lung and blood-brain barrier in
rabbits after decompression. Undersea Hyperb Med 26:249±256
Ho
È
gesta
È
tt E, Andersson K-E, Edvinsson L (1983) Mechanical
properties of rat cerebral arteries as studied by a sensitive device
for recording of mechanical activity in isolated small blood
vessels. Acta Physiol Scand 117:49±61
Hoka S, Okamoto H, Yamaura K, Takahashi S, Tominaga R,
Yasui H (1997) Removal of retained air during cardiac surgery
with transesophageal echocardiography and capnography.
J Clin Anesth 9:457±461
Johnston W, Stump D, DeWitt D, Vinten-Johansen J, O'Steen W,
James R, Prough D (1993) Signi®cance of gaseous microemboli
in the cerebral circulation during cardiopulmonary bypass in
dogs. Circulation 88:319±329
Ku D (1987) Unmasking of thrombin vasoconstriction in isolated
perfused dog hearts after intracoronary infusion of air embolus.
J Pharmacol Exp Ther 243:571±576
Mu
È
lsch A, Bassenge E, Busse R (1989) Nitric oxide synthesis in
endothelial cytosol: evidence for a calcium-dependent and a
calcium-independent mechanism. Naunyn Schmiedebergs Arch
Pharmacol 340:767±770
Nossum V, Koteng S, Brubakk AO (1999) Endothelial damage by
bubbles in the pulmonary artery of the pig. Undersea Hyperb
Med 26:1±8
Nossum V, Hjelde A, Bergh K, Brubakk AO (2000) Lack of eect
of anti-C5a monoclonal antibody on endothelial injury by gas
bubbles in the rabbit after decompression. Undersea Hyperb
Med 27:27±35
Philp R (1974) A review of blood changes associated with com-
pression-decompression: relationship to decompression sick-
ness. Undersea Biomed Res 1:117±150
Roberts NA (1988) Free radicals, immunoglobulins and comple-
ment as mediators of in¯ammation. Baillieres Clin Rheumatol
2:211±232
Rubanyi G, Romero J, Vanhoutte P (1986) Flow-induced release of
endothelium-derived relaxing factor. Am J Physiol 250:1145±
1149
Sacks T, Moldow C, Craddock P, Bowers T, Jacob H (1978)
Oxygen radicals mediate endothelial cell damage by comple-
ment-stimulated granulocytes. An in vitro model of immune
vascular damage. J Clin Invest 61:1161±1167
Stewart D, Baour R (1990) Functional state of the endothelium
determines the response to endothelin in the coronary circula-
tion. Cardiovasc Res 24:7±12
Till G, Johnson K, Kunkel R, Ward P (1982) Intravascular acti-
vation of complement and acute lung injury. Dependency on
neutrophils and toxic oxygen metabolites. J Clin Invest
69:1126±1135
Tofukuji M, Stahl G, Agah A, Metais C, Simons M, Sellke F
(1998) Anti-C5a monoclonal antibody reduces cardiopulmo-
nary bypass and cardioplegia-induced coronary endothelial
dysfunction. J Thorac Cardiovasc Surg 116:1060±1068
Varani J, Fligiel S, Till G, Kunkel R, Ryan U, Ward P (1985)
Pulmonary endothelial cell killing by human neutrophils. Pos-
sible involvement of hydroxyl radical. Lab Invest 53:656±663
Ward C, Koheil A, McCullough D (1986) Activation of comple-
ment at plasma-air or serum-air interface of rabbits. J Appl
Physiol 60:1651±1658
Ward C, McCullough D, Fraser W (1987) Relation between
complement activation and susceptibility to decompression
sickness. J Appl Physiol 63:1160±1166
Warren B, Philp R, Inwood M (1973) The ultrastructural mor-
phology of air embolism platelet adhesion to the interface and
endothelial damage. Br J Exp Pathol 54:163±172
Webb WR, Harrison LH Jr, Helmcke FR, Camino-Lopez A,
Munfakh NA, Heck HA Jr, Moulder PV (1997) Carbon dioxide
®eld ¯ooding minimizes residual intracardiac air after open
heart operations. Ann Thorac Surg 64:1489±1491
214
