| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Departments of 1 Medicine,
2 Anesthesiology, and
3 Physiology and Biophysics, The distribution
of gaseous pulmonary emboli is presumed to be determined by their
buoyancy. We hypothesized that regional pulmonary blood flow may also
influence their distribution. Therefore, pulmonary blood flow was
measured in supine, anesthetized dogs with use of 15-µm fluorescent
microspheres at baseline and during N2 embolism. The animals were
killed, and the lungs were excised, air-dried, and diced into
~2-cm3 pieces with weights and
spatial coordinates recorded. Embolism was defined as a >10% flow
decrease relative to baseline. Vertically, the incidence of embolism
increased substantially by 6 ± 1% per additional centimeter in
height compared with baseline (P = 0.0003). Embolism also increased radially by 3 ± 1%/cm from the
hilum (P = 0.002). There was a weaker
but statistically significant increase in embolism to pieces with
greater baseline flow, 9 ± 2% for every 1.0 increase in relative
baseline flow (P = 0.008). We conclude that the distribution of gaseous emboli is influenced by buoyancy and
flow dynamics within the pulmonary vasculature.
fluorescent microspheres; air embolism; pulmonary gas exchange; pulmonary circulation; decompression illness
PULMONARY AIR EMBOLISM has been a subject of interest
to clinicians and researchers for many years because of the severe
hemodynamic and gas exchange abnormalities that result (29). Many
studies of pulmonary air embolism have been done in the context of
decompression illness, which is a prominent problem for divers (43) but
may affect aviators and astronauts as well (36, 47). Victims can be
devoid of pain or other symptoms, even with concomitant extrapulmonary embolism. Pulmonary air embolism is also known to occur during surgical
procedures and traumatic injuries and has been demonstrated to cause
elevated pulmonary arterial pressure, elevated systemic arterial
pressure, elevated pulmonary vascular resistance, reduced cardiac
output, systemic hypoxia, and systemic hypercarbia (12, 13, 35, 44). It
is the N2 in the air that is
responsible for these effects.
When air bubbles enter the venous blood, they are carried to the right
heart and undergo turbulent mixing and disruption in the right
ventricle. If the bubbles are not of sufficient size to obstruct the
right ventricular outflow tract, they will be ejected into the
pulmonary arteries, possibly after being decreased further in size as a
result of right ventricular mixing. The emboli travel through the
pulmonary vessels and are trapped in the small arterioles and
capillaries (1, 35, 37), with almost no further passage into the left
heart and systemic circulation (6, 7). Little is known about where the
bubbles distribute after leaving the right heart and the dynamics of
their resorption over time.
In prior studies of the gas exchange abnormalities associated with
pulmonary air embolism (27, 48), it was shown that there was no
increase in dead space ventilation with
N2 embolism, despite the
obstruction of the pulmonary vessels. Instead, there was an increase in
the lung regions with high ventilation-to-perfusion ratios
( Two other studies have looked at the actual distribution of air emboli
in the lung vasculature. In the first study, dogs were studied in the
prone and supine positions, and the emboli were tracked using a
radioactive xenon tracer method (10). In the second study, physical
models of the pulmonary vessels were employed to observe how the
bubbles might behave at bifurcations and when lodging in small
arterioles and capillaries (11). In both studies it was concluded that
the buoyancy of bubbles would be the primary determinant of their
distribution; i.e., the bubbles would always float to the top. In
addition, it was concluded that >20- to 30-µm bubbles would not
pass through to the systemic circulation. However, there is a problem
with the radioactive xenon method, because it is limited in its
resolution and requires the investigator to look only in the plane of
PBF distribution that will presumably be affected. We hoped to obtain
more information about the distribution of the emboli in other spatial
dimensions. In addition, we wanted to know whether the initial
magnitude of blood flow in a pulmonary vessel increased the incidence
of embolization to that vessel, irrespective of its spatial position in
the lung. We theorized that dynamic factors such as the release of
inflammatory mediators or hypoxic pulmonary vasoconstriction could have
additional effects on the distribution of bubbles lodging in the
pulmonary vessels.
The purpose of this study, therefore, was to study the
three-dimensional distribution of PBF on a finer scale of resolution before, during, and after venous embolism with
N2 in an intact animal model. We
hoped to shed more light on the interactions between buoyancy and flow
in determining bubble location.
After approval of the experimental protocol by the Animal Care
Committee of the University of Washington, six mixed-breed dogs of
either gender [27.0 ± 2.9 (SD) kg body wt] were studied.
Anesthesia and Instrumentation
ABSTRACT
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
A/
)
and also an increase in areas with low
A/. Presumably, the pulmonary blood flow (PBF) redistributed to lung regions where the pulmonary vessels were not blocked with bubbles, and
these regions then became relatively overperfused and underventilated. This increased
A/
heterogeneity is likely responsible for the hypoxia and hypercarbia
that result when air embolism occurs. The gas exchange abnormalities
were no longer present 30 min after the cessation of embolism,
indicating that the emboli had completely resorbed or were not present
in sufficient size or quantity to cause
A/ aberrations.
MATERIALS AND METHODS
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
1 · h1
to maintain hemodynamic stability and suppress responses to any noxious
stimuli. After anesthesia was induced, an oral endotracheal tube was
placed, and the lungs were ventilated with atmospheric air by means of
a conventional piston-pump ventilator (dual-phase respirator pump model
608, Harvard Apparatus) set for a tidal volume of 15 ml/kg. The animals
were positioned supine for intubation and instrumentation and remained
in the supine position for the duration of the experiment.
A large-bore femoral arterial catheter was inserted for pressure monitoring and blood sampling; it was positioned so that the tip was in the abdominal aorta just caudal to the level of the diaphragm. Bilateral femoral venous catheters were placed. One catheter was used for administration of the N2 emboli and was positioned in the inferior vena cava so that the tip was just caudal to the level of the diaphragm. A vascular introducer sheath was placed in the other femoral vein to allow a balloon-tipped thermodilution pulmonary arterial catheter (Baxter Healthcare, Irvine, CA) to be floated into the pulmonary artery to monitor pulmonary arterial pressure, pulmonary capillary wedge pressure, blood temperature, and cardiac output. The introducer and pulmonary arterial catheter were also used to allow infusion of anesthetic drugs and maintenance fluids. An additional venous catheter was placed in one of the external jugular veins for injection of fluorescent-labeled microspheres.
Arterial blood pressure, pulmonary arterial pressure, airway pressure, and end-tidal CO2 were measured continuously and recorded on a data-management system (Mark 12, DMS 1000, Western Graphtec) with Validyne amplifiers. Cardiac output, blood temperature, and arterial and mixed venous blood gas values were measured at conditions of steady state and under the N2 embolism study conditions (see below). The end-tidal CO2 was measured with a mass spectrometer (medical gas analyzer MGA-1100, Perkin-Elmer). The cardiac output and blood temperature were measured with a cardiac output computer (model Sat-2, Baxter Edwards). Arterial and venous pH, PO2, and PCO2 were measured with a blood-gas analyzer (model ABL 330 Acid Base Laboratory, Radiometer, Copenhagen, Denmark) and corrected for temperature. The ventilatory rate of the lungs was adjusted to maintain the arterial PCO2 at 38.8 ± 2.7 (SD) Torr.
Venous N2 Embolism and Study Conditions
N2 bubbles were infused into the femoral vein at a rate of 0.2 ml · kg1 · min1
for 30 min with use of a constant-flow syringe infusion pump. This
embolic load was sufficient to more than double the pulmonary arterial
pressure after 5 min of bubble infusion. The bubbles were formed by
forcing the N2 through a PE-20
catheter, resulting in ~1.2-mm-diameter bubbles, as estimated using a
stroboscopic light, as described previously (27). However, because
these bubbles were allowed to be mixed in the right atrium and
ventricle before entering the pulmonary artery, it is likely that the
size range of the bubbles was broader as a result of this
disruption. The bubbles that were initially administered could break up
into smaller bubbles or coalesce into larger ones as a result of this mixing.
Because Hlastala et al. (27) had studied the gas exchange of the lungs at 15 and 30 min after initiation of embolization, we chose to study these same time periods. Therefore, arterial and venous blood-gas samples and hemodynamic measurements were taken, and fluorescent microspheres were injected at the following times: baseline, after 15 min of N2 embolism (N215), and after 30 min of N2 embolism (N230). In addition, hemodynamic measurements, blood samples, and a fluorescent microsphere injection were performed during the recovery phase at 15 and 30 min after the cessation of the N2 embolism (R15 and R30, respectively).
Fluorescent Microsphere Technique
After hemodynamic measurements were taken under each study condition, samples were drawn for arterial and venous blood-gas measurements. After blood sampling, ~2-3 × 106 15-µm-diameter fluorescent polystyrene microspheres (FluoSpheres, Molecular Probes, Eugene, OR) were injected via the jugular venous catheter over 30 s. Before injection, the microspheres were sonicated for 2 min and then vortexed. Each injection was followed by 20 ml of normal saline flush. Five colors (red, orange, blue-green, yellow-green, and crimson) were used, with the order of the injection of the colors randomized in each experiment.At the completion of the study, each animal was given an additional bolus of 5-7 mg/kg thiopental sodium followed by 1,000 U/kg heparin (Elkins-Sinn, Cherry Hill, NJ) and 3 mg/kg papaverine hydrochloride (YorPharm, Buffalo Grove, IL); then they were exsanguinated. The chest was opened by midline sternotomy, and catheters were placed in the main pulmonary artery and the left atrium to isolate the pulmonary circulation. The lungs were flushed free of blood with 1 liter of 2% dextran (Sigma Chemical, St. Louis, MO), removed intact, and then air-dried at total lung capacity (TLC).
The dried lungs were then positioned in a miter box so as to reproduce the isogravitational planes in vivo and were encased in polyurethane foam insulating sealant to provide a rigid orthogonal reference system. The lungs were then diced into 1.9-cm3 pieces, and each piece was weighed and assigned x, y, and z coordinates. The pieces were then soaked in 2-ethoxyethyl acetate (Cellosolve acetate, Aldrich Chemical, Milwaukee, WI), and the fluorescence was read in a luminescence spectrophotometer (model LS-50B, Perkin-Elmer) fitted with a standard cuvette reader and a red-sensitive photomultiplier tube. Regional PBF for each lung piece was calculated from fluorescent microspheres data according to the methods of Glenny et al. (16). Briefly, lung pieces with abnormal appearance or with >25% airway content on visual inspection were excluded from analysis, and in the remaining pieces the weight-normalized relative PBF (WNRF) was calculated. The hilar-to-peripheral (h) distance was determined according to the methods of Glenny et al. (17).
Data Analysis and Statistical Methods
Descriptive statistics. Because the primary goal of our study was to try to gain more information about the interactions between buoyancy and flow on bubble location, we decided to focus our data analysis on the distribution of PBF only during the baseline and embolism (N215 and N230) conditions.
Descriptive statistics of hemodynamic parameters, spatial flow gradients, and other variables are usually presented as means ± SE calculated across the six animals. Differences in hemodynamic parameters between pairs of conditions are analyzed using the two-sided paired t-test (N215 vs. baseline, N230 vs. N215, and N230 vs. baseline). In all cases, P < 0.05 was considered statistically significant. There has been no adjustment of P values for multiple testing, because each P value addresses a comparison between states with potentially unique effects.Effect of embolism on perfusion.
The effect of N2 embolism on
perfusion is described by two variables. First, a piece of lung was
considered to have substantial embolism if there was a strong enough
negative effect on perfusion that a decrease of >10% in flow
(relative to baseline) was observed. The 10% value is highly likely to
be a nonrandom decrease and was determined as follows. Approximately
1,000 or more fluorescent microspheres were injected per piece,
yielding a Poisson standard error for the average piece of 
3%
[
100% * 1/(1,000)0.5]. Random time
variation over 15 min in anesthetized dogs has been shown previously to
be ~5% (18). Pooling these two sources of random variation yields a
standard error of ~6%
[(52 + 32)0.5].
Finally, on the basis of normal approximation and with the assumption
of the null hypothesis of no effect of embolism, only ~5% of the
pieces would have a random fluctuation decreasing the flow from
baseline by >10% (6% * z0.95, where
z0.95 = 1.64). Although there is a series of approximations here, we can be confident that the vast majority of pieces with >10% decrease in flow have lost flow because of embolism. A dichotomous variable indicating a
>10% loss of flow compared with baseline (vs. a smaller loss or an
increase) was therefore calculated for all pieces at
N215 and, as a separate variable,
for all pieces at N230.
baseline and
N230 baseline.
Influence of spatial dimensions and baseline flow on perfusion. In this study our interest focused on spatial dimensions and baseline flow as factors that would influence which pieces were affected by embolism. Spatial dimensions were defined as x (lateral left-to-right gradient), y (vertical back-to-front gradient, which is also the gravitational gradient in this study), z (horizontal caudal-to-cranial gradient), and h (the radial distance to the hilum of each lung) (17). The baseline (preembolism) flow was hypothesized as a mechanism transporting more N2 bubbles to pieces with higher blood flow rates.
WNRF was plotted against each of the spatial dimensions of x, y, z, and h, and slopes of the relationships between flow and spatial dimension were determined using least squares regression. Changes in the slopes among the three study conditions of baseline, N215, and N230 were calculated by subtraction and compared using the single-sample t-test. Significant changes in regional PBF occurred only in the y and h distributions (see RESULTS). The x and z dimensions were therefore excluded from the next series of analyses. The influence of the baseline flow on flow changes can also be described by least squares regression coefficients, analogous to the spatial dimensions y and h. Least squares regression was therefore performed for the influence of y, h, and baseline flow on the dichotomous variable indicating a >10% loss of flow. All the slope coefficients are readily interpretable. For example, if the dichotomous indicator of embolism-affected pieces is regressed on the y dimension, yielding a slope coefficient for y of 4.2, it indicates that the incidence of embolized lung pieces increases, on the average, by 4.2% for each vertical centimeter in the lung. Similarly, a coefficient of1.4 for baseline flow when the change in flow
(e.g., N215 baseline) is
regressed on baseline flow indicates that, on the average, relative
flow decreases by 1.4% of the whole lung's mean relative flow when
one piece has 1.0 more relative flow unit at baseline than another.
The influence of spatial dimensions and baseline flow on the location
of N2 emboli is summarized by the
mean of the regression coefficients across the six animals. The null
hypothesis of "no effect" was tested using the two-sided
single-sample t-test comparing each
mean regression coefficient with zero. Correlations among y, h, and baseline suggested potential
confounding among these variables; therefore, regression models were
calculated 1) for each of these
three variables as a single predictor,
2) for all three of the variables as
joint predictors, and 3) in the case of baseline, which was most affected by confounding, by using the
residuals of the dependent variable regressed on
y and
h (a step that removes linear effects
of y and
h on the dependent variable) and then
regressing the residuals on baseline blood flow as the independent
variable. The third procedure provides the most conservative estimate
of the impact of baseline blood flow.
We also partitioned the total variation in flow of the combined
embolism and the baseline states into two components:
1) a fixed component due to
biological structure (the position of a piece in the whole lung, or the
"piece effect") and 2) a
component due to dynamic changes, where such changes include the
embolism, time variation, and methodological noise (the random
variation in microsphere distribution and laboratory measurements). The contribution of each source of variation, "piece" or
"change," can be expressed as two percentages adding up to 100%.
Details of the calculation are described by Bernard et al. (4).
Finally, we calculated the physical center of the embolism-affected
pieces as well as of the whole lung. The center is simply the mean
x, y, and
z coordinates of the pieces
considered. We tested the hypothesis that the embolism pieces are
centered at the same coordinates as all the pieces by using a paired
t-test on each dimension
(x, y, and
z). The shift from the center of the
whole lung to the center of the embolism is a descriptive statistic
that can help in visualizing the relative position of the embolized regions.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
All animals had a normal hemodynamic profile at the beginning of the
study, as shown in Table 1. The decrease in
heart rate, increases in systemic and pulmonary arterial pressures, and
decreases in arterial oxygenation seen in this study are consistent
with the results of other studies reporting the effects of venous air embolism (2, 5, 9, 13, 27, 46). Cardiac output did not change
significantly during the study period.
|
The primary focus of this study was to examine the spatial distribution of 15-µm fluorescent microspheres before, during, and after the venous embolization of N2 bubbles. Because fluorescent microspheres are a reliable indicator of regional PBF (16), we assumed that regions with a substantial decrease in PBF after the infusion of N2 would be indicative of the presence of emboli.
Analysis of Slopes in Each Spatial Dimension
The WNRF for each lung piece analyzed was plotted against each of the four spatial dimensions studied (x, y, z, and h), and slopes were determined as described in MATERIALS AND METHODS. Figure 1 shows an example of the baseline WNRF values plotted against the dorsal-to-ventral (y) dimension, and Fig. 2 shows the same animal and spatial dimension at N215. Figure 3 shows the change in slope between these two conditions as a result of subtraction and illustrates that, in almost every lung piece in the ventral aspect of the animal, there was a substantial flow decrease, whereas in the dorsal lung regions there were areas of dramatic increases and decreases in flow. Results of least squares regression of these plots show no statistically significant changes in the slopes of the blood flow distribution in the x and z distributions, but the changes in the y and h slopes were statistically significant. For each dimension the slopes for N230-N215 are quite small and nonsignificant, indicating that the embolism process had reached a relatively steady state by 15 min. The results of the least squares regressions in each spatial dimension are shown in Table 2. All subsequent analyses of changes in PBF distribution were performed only in the y and h distributions.
|
|
|
|
Regression Analysis in the Vertical and Hilar-to-Peripheral Dimensions
Data were analyzed for all animals to determine the incidence of embolism at N215 and N230 in the y and h dimensions for each animal. Embolism affected a large fraction of the lung. Among the six animals, 48-71% of pieces lost
10% of their relative flow
(N215 compared with baseline) with
a mean of 62% of pieces. The results were plotted as illustrated in
Figs. 4 and 5,
which indicate strong trends in the incidence of embolism across the y and
h dimensions. In addition, the
fraction of pieces with a >10% decrease in WNRF at
N215 and
N230 was plotted against the initial WNRF at baseline conditions, as illustrated in Fig.
6. The purpose was to determine whether
there was any tendency for N2
emboli to go to lung areas with a particular baseline flow (i.e.,
whether high-flow pieces demonstrate a greater incidence of embolism
than low-flow pieces). No striking trend of embolism incidence with
baseline flow is evident from Fig. 6, but, as noted below, a small but
statistically significant trend does exist after controlling for
confounding from the effects of y and
h.
|
|
|
Univariate and multivariate regression analyses were performed for the
influence of y, h, and baseline on the
dichotomous indicator of embolism. All animals were studied at
N215 and
N230, and the mean results across
the six animals are summarized in Table 3.
On the basis of the results of the regression analysis, we can confirm
the previous findings of Chang et al. (10, 11) that in the supine
position the emboli clearly distribute in a manner consistent with
their buoyancy. The influence of y is
by far the strongest of the three factors, as indicated by
R2 or
R2 for the
N215 or
N230 embolism incidence:
R2 is
24-25% for y, 6-7% for
h, and 2-3% for baseline. Values
for R2 are
somewhat smaller but show the same dominance for
y, followed by
h, then baseline. The slope of
embolism incidence is also stronger for
y (mean slope = 6%) than for
h (3-4% across types of analyses and times). In stepwise regression analyses of embolism incidence, factors were selected into the regression model in the order
y, h, and baseline for every animal.
Table 3 also shows only small and nonsignificant changes in slopes
between N215 and
N230. The h distribution of the emboli
documented in Table 3 has not been previously shown. The reversal of
sign of the mean slope for baseline from univariate to multivariate
analysis is notable and arises from the positive correlation and trend
of baseline flow with the vertical and radial dimensions. Thus, in
analyzing the influence of baseline flow on embolism, it is essential
to control for y and
h.
|
Regression Analysis Performed Against Baseline WNRF
To analyze whether more emboli distribute to areas where there is more flow while controlling for the strong influences of the y and h dimensions, we performed a second set of regression analyses. As a very strong control for the linear influence of y and h, we first regressed the dichotomous embolism indicator on y and h and then regressed the residuals from this first step on baseline. (The methodology is analogous to partial correlation.) The coefficient of baseline in the second regression analysis is, then, a conservative estimate of the trend in embolism incidence with baseline flow. The regression analysis was carried out for the embolism indicator at N215 and, separately, for N230. The same regression analyses based on residuals were also carried out using the actual flow change for each piece at N215 (i.e., flow at N215 minus flow at baseline). The null hypothesis of no effect of baseline flow was tested using the two-sided single-sample t-test comparing each mean regression coefficient with zero. The results of the regression analyses performed for incidence of embolism and change in flow vs. baseline are summarized in Table 4. The trend in incidence of embolism with baseline flow is positive and statistically significant at N215 (P = 0.01), and the decrease in actual flow vs. baseline is also statistically significant (P = 0.005). The trend in embolism with baseline at N230 is also positive but not significant. R2 is very small for all these analyses.
|
Because a statistical phenomenon called "regression to the mean"
can induce a significant negative association when a change is compared
with the starting value, we carried out an analysis of actual flow
change from baseline to R30, a flow change with a relationship to
baseline flow that would be primarily due to regression to the mean.
Regression analyses were performed on the actual flow changes (i.e.,
R30 baseline) vs. baseline flow, as described above, and the
results again were compared with zero by using the two-sided
single-sample t-test. These results
are also displayed in Table 4. The "baseline effect" is observed much less at R30 (slope = 5.4%) than at
N215 (slope = 35.8%). Therefore, the tendency of emboli to distribute to lung regions with
higher flow is not the result of regression to the mean. On the basis
of these results, we can conclude that there is a weak but significant
tendency for the emboli to distribute to lung regions that have a
higher PBF.
Center of Embolism and Center of Lung Studies
We calculated the center of the embolism by calculating the mean x, y, and z coordinates for all the pieces with a positive embolism indicator. Next, we calculated the center of the lung by determining the mean x, y, and z coordinates for each animal. These two centers were then compared using a single-sample t-test (2-tailed). The embolism pieces (those with the positive embolism indicator) are centered at a small but statistically significant distance from the center of the lung. The shift is 1.5 ± 0.3 (SE) cm in the vertical direction (P = 0.004 compared with zero shift) and 0.4 ± 0.1 cm (P = 0.004) radially from the hilum. The total span of the vertical and radial dimensions of the lung is ~20 cm, making the vertical and radial embolism shifts ~8 and 2% of each dimension, respectively. Mean shifts in the other dimensions were minor: 0.1 ± 0.3 and 0.2 ± 0.3 cm for x and z, respectively.Variance Components Study
Embolism induced enormous changes in flow distribution, as indicated by the variance components analysis: 57 ± 1 (SE) % of the overall variation in flow (across pieces and conditions) was due to embolism, time, and methodological noise. Conversely, 43% of flow variation was due to biological structure or the piece effect in the pattern of flow distribution common to both states. For minor interventions such as exercise, the piece effect is generally 70-80% [e.g., 80% for exercising horses (4)]. For no intervention but the passage of time, the combination of time and methodological noise usually contributes10% (17) to the total piece effect.
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Methodological Issues
Microspheres have been used to measure regional organ perfusion for many years, and this method has generally been accepted as reliable and well validated (24). However, because microspheres are particulates, it is possible that they may not behave comparably to blood when one is attempting to measure flow distributions at the level of capillaries (51). This issue was addressed by Bassingthwaigthe et al. (3) in their validation of microspheres for measurement of regional myocardial blood flow values. Regional organ perfusion as measured using a molecular tracer of blood flow (1-iododesmethylimipramine) was highly correlated with flow measured by radiolabeled microspheres. These investigators concluded that microspheres accurately measure regional blood flow, even when the region in question is supplied by an artery with a diameter only a few times larger than the 15-µm diameter of the microsphere. Because the regions of lung that were studied in this investigation are supplied by arterioles that average 50-100 µm in diameter, the use of 15-µm microspheres to measure regional organ perfusion should result in an accurate assessment of the true regional PBF.This investigation used fluorescent microspheres, rather than radiolabeled microspheres, to measure regional organ perfusion. To reliably measure regional PBF, the 15-µm fluorescent microspheres used in this study must have a distribution in the microcirculation that is similar to the distribution of blood and must be highly correlated with the measurements obtained using radiolabeled microspheres. Because the 15-µm fluorescent microspheres have a density very close to that of blood (1.02 g/ml), their distribution in the circulation should be similar to that of blood; indeed, this validation in dogs has been published (16). In addition, the number of microspheres in each lung piece must be great enough to limit the effect of method error on the statistical analysis. With an injection of 2-3 × 106 microspheres per condition, each tissue piece would have >350 microspheres, even when PBF to that piece was decreased by 75%. Because the entire lung was analyzed and, therefore, contained some subpleural pieces that might be smaller than the majority of pieces, the samples were normalized by weight to correct for differences in parenchymal volume. In addition, samples containing >25% airway tissue or samples that were noticeably abnormal (i.e., contained clot or had compacted tissue) were excluded from the statistical analysis.
In this particular experiment a potential problem with ensuring
adequate numbers of microspheres per piece of lung tissue is that of
occluding too many pulmonary vessels by using the microsphere technique. This would not normally be of concern if emboli were not
present. However, it is likely that there are adequate numbers of
patent pulmonary capillaries, even in the presence of the
N2 emboli. Microspheres tend to
lodge primarily in capillaries or in the terminal arteriole just before
its branching into capillaries and can move slowly along the length of
a capillary, eventually lodging in a <15-µm-diameter vessel (up to
20% smaller) (23). A recent study (15) in dog lungs gave a terminal
arteriolar diameter of 28 µm, which is substantially larger than the
size of a microsphere and implies that microspheres would lodge in capillaries. In another study (31) the authors summarize many of the
known studies of vascular morphometry and make model predictions which
indicate that there should be ~1 × 109 vessels of 10-15 µm in
diameter. The total number of microspheres injected per dog was 1.1 × 107 (only one of the five
injections required 3 × 106
microspheres), which would occlude ~1% of these vessels. Because it
is possible that 15-µm microspheres could lodge in series in a given
vessel (40), then even <1% of the vessels might be occluded by
microspheres. N2 emboli were
presumed to be present in a lung piece when there was a >10% (of
baseline) decrease in relative PBF. A mean of ~60% of the lung
pieces studied had a positive embolism indicator. This does not
necessarily mean that 60% of the 15-µm vessels were occluded.
Rather, it is more likely that 10% of the vessels in the affected 60%
of the lung (or a total of 6% of the lung) have been embolized.
Therefore, one could conclude that, between microspheres and
N2 emboli, ~90% of the lung
capillaries remained patent during the study.
The lungs were dried at TLC, meaning that alveolar volumes will be uniformly distended throughout the lung while it is drying. This may not be true of the lung at TLC in the intact animal. The configuration of the lung after drying may be different from that of the lung in vivo for other reasons as well. First, lung drying is performed with a constant pressure of ~35 cmH2O while the lung is suspended from the trachea, so that lung volumes (and, consequently, the linear dimensions of the 3 axes) are increased compared with the in vivo lung. In addition, the influence of gravity in a lung suspended from the trachea will differ from that in a supine animal, possibly resulting in a slight distortion of the lung configuration compared with the in vivo condition. Another difference is the lack of physical constraint imposed by the chest wall and diaphragm in the intact animal and the compression of the lung regions that lie below the heart when the animal is supine. It is anticipated that these conditions might impose a different shape on the intact lung. These factors are difficult to quantitate or evaluate, much less to correct for. However, because the effect of N2 embolism is so profound and results in such an extreme series of changes in the distribution of PBF, we expect the conformational differences between the air-dried and the in vivo lung to be unimportant with respect to the results of this study. In addition, because the fluorescent microspheres should remain fixed in the vessels after lodging in vivo, the relative perfusion changes to the different lung regions should remain fixed as well.
Vertical Distribution
Under baseline conditions, we clearly observed a predominance of PBF to the dorsal lung regions in the supine dog. This has been observed previously, and this dorsal flow predominance persists in animals in the prone position (17, 33, 49, 50). This finding suggests that gravity does not play a dominant role in determining the heterogeneity of PBF in quadrupeds when perturbations other than anesthesia, intubation, and mechanical ventilation are absent (4, 25). With N2 embolism, the distribution of PBF in the y gradient showed a strong tendency for the bubbles to distribute according to their buoyancy, i.e., a gravitational dependence such that the bubbles "float to the top." This strong y gradient was most evident in the results of the center-of-embolism study performed on our data. Because the total span of the y distance was ~20 cm, the center of embolism shifted from the center of the lung vertically by ~8% of this distance. This confirms the original findings in the supine dog and the model predictions made by Chang et al. (10, 11). However, our study is still limited, because it does not give any indication of the size or shape of the region(s) most impacted by the embolism.The y distribution of the emboli at N230 is not significantly different from that at N215, as shown in Table 3. This is probably because the process of embolization has reached a steady state by 15 min. The way in which such a steady state could result can be hypothesized as follows: vessels at the top of the lung are blocked by the bubbles that arrive when embolism initially starts, and therefore the subsequent incoming bubbles must distribute to the lower, dependent lung regions. As the lodged bubbles resolve with time, emboli could once again distribute to the nondependent portions of the lung. With continuous bubble administration balanced by continuous bubble resorption, it is possible that the distribution of emboli in the lung could become uniform, if it is assumed that the dog could survive a certain level of embolization for a long enough period of time (45). There is some evidence that this was beginning to occur by N230, as evidenced in Table 1. When the values for N215 and N230 are compared with the baseline values, the P values for the mean systemic arterial pressure and mean pulmonary arterial pressure are not significantly different at N230, but they are at N215. Also, the P values are smaller for the arterial PO2 and PCO2, although they are still significantly different from the baseline. Because this study used a decrease of regional flow as an indicator of the presence of an embolism, we also cannot say whether another embolism could lodge more proximally in the same blood vessel as a result of reestablishment of flow or even in the continued presence of an occlusion to flow, as has been seen previously (40).
Hilar-to-Peripheral Distribution
Evidence for an h distribution of gaseous emboli has not been previously presented, nor has it been theoretically postulated. However, it is logical to assume that such a distribution would occur with gaseous emboli, because the emboli would travel along the pulmonary arteries until they reached a vessel that was too small to permit further passage. However, when a bubble reaches a bifurcation in the vessel, the next determinant of the path of the bubble could be flow or buoyancy, as illustrated in Fig. 7. In this study we found that both factors influence the behavior of the bubble, and therefore both should exert an influence on bubble distribution at such a bifurcation. The relative contribution of each factor at such a vascular bifurcation remains to be determined.
|
The h gradients for PBF have been demonstrated for dogs (17, 19, 20, 22), humans (21), sheep (50), and ponies (30). Walther et al. (50), using methods similar to those for this study, concluded that the normal h distribution of PBF was likely a function of vascular anatomy due to branching vascular patterns and regional variations in capillary density. If the h blood flow gradient is reflective of the inherent structure of the pulmonary vessels, then an h distribution of gaseous emboli as seen in this study may also reflect the influence of vascular structure on bubble distribution. However, because the variance-components study that we performed indicates that, on average, only 43-44% of the flow variation could be explained by the configuration of the biological structure, then one might conclude that vascular structure does not play the most important role in determining the distribution of gaseous emboli.
Distribution According to Baseline Flow
The preferential distribution of N2 emboli to lung regions of initially higher flow is a new finding that reflects an advantage of the methodology used and the superior level of resolution that it affords. Although the influence of baseline flow on the distribution of gaseous pulmonary emboli is small, it is nevertheless statistically significant and evidently real. As shown in Table 1, there are no statistically significant changes in cardiac output between baseline and the two embolism conditions. Because previous studies of N2 embolism have shown that it does not produce an increase in shunt (27), then the total amount of blood flowing through the pulmonary circulation per minute should be approximately the same in each of these three conditions. Therefore, redistribution of PBF from lung regions of initially higher blood flow to lung regions of initially lower blood flow could be accounted for partly by the effects of the embolism.In addition, the distribution of emboli according to flow is most evident when the baseline flow is compared with the flow at N215, rather than at N230. This is because at N230, there are already bubbles present and severe aberrations of PBF have already resulted. Therefore, at N230 the bubbles are now distributing according to the flow patterns observed at N215, and not at baseline. This is why the flow effect is seen less prominently at N230.
Although the flow effect is small, it is most likely real and not the result of regression to the mean. This is evidenced by the fact that the baseline slope from the regression analysis at R30 is much smaller than the baseline slope at N215. Regression to the mean would tend to affect both slopes.
Embolism Methodology
Bubble size was estimated using a method that is not highly accurate; however, the measured size at the time and site of infusion will likely be quite different from the size that is embolized into the pulmonary vasculature. This is due to the passage of the bubbles through the right side of the heart, which mixes the bubbles in the blood and would likely result in a change in size of the infused bubbles. Although this may not represent the fate of bubbles in the venous circulation during decompression, it is a realistic reproduction of accidental venous air embolism in clinical scenarios. However, not many studies address the changes in size of bubbles, and those that do have studied the effects of different gas types on the size of the bubbles (8, 37, 39, 41). A recent modeling study (42) of stabilized bubbles in the circulation indicates that bubbles reach maximal size in the pulmonary small vessel beds because blood pressure is low, there is no inherent unsaturation, and gases from the alveoli diffuse into the bubble. Therefore, even very small bubbles that might have traversed other capillary beds will ultimately become lodged in the pulmonary bed. This might explain why it has been observed that the lung is such a highly effective filter of bubbles (6, 46). For gaseous emboli, it is likely that surface-active molecules in blood could aggregate at bubble surfaces and provide a stabilizing action, but there is no experimental evidence for this.Another consideration when studying distributions of gaseous emboli in the lungs is the reaction by the lung vasculature caused by the lodging of such emboli. Gaseous emboli cause microvascular injury and the release of inflammatory mediators by the lung (14, 34), which may then have systemic circulatory effects (28). We saw no changes in cardiac output resulting from these physiological effects. However, there are profound alterations in systemic and pulmonary arterial pressures, which may exert a different influence on the distributions of emboli that enter the lungs later, after embolism has been occurring for a while. We did not attempt to correlate the levels of circulating inflammatory mediators with the changes in regional distribution of PBF, although this might be an important endeavor for the future. We also did not attempt to inhibit the release of such inflammatory mediators to attempt to study a "pure buoyancy" effect. In this investigation our goal was to study the animal in as much of an intact and unaltered state as possible, mimicking actual accidental N2 embolism.
Conclusion
The results of this study demonstrate that the use of higher resolution methods to determine the distribution of PBF provides new information about such distributions. In addition to the influence of buoyancy, the distribution of gaseous emboli is influenced by regional lung blood flow. We also observed an h distribution of gaseous emboli, which has not been described previously and may be reflective of inherent lung vascular structure. Therefore, we conclude that the distribution of gaseous emboli in the pulmonary blood vessels is influenced not only by their differential density from blood but also by the flow dynamics within the pulmonary vascular tree.| |
ACKNOWLEDGEMENTS |
|---|
We acknowledge the excellent technical assistance of Erin Shade, Dowon An, and Emily Anderson. We gratefully acknowledge the input of Robb W. Glenny regarding the many facets of the fluorescent microspheres technique.
| |
FOOTNOTES |
|---|
Support for this project was provided in part by National Heart, Lung, and Blood Institute Grants HL-12174 and HL-24163.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Address for reprint requests and other correspondence: J. E. Souders, Pulmonary and Critical Care Medicine, Box 356522, University of Washington, Seattle, WA 98195-6522 (E-mail: jsouders{at}u.washington.edu).
Received 28 December 1998; accepted in final form 13 July 1999.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
1.
Albertine, K. H.,
J. P. Wiener-Kronish,
K. Koike,
and
N. C. Staub.
Quantification of damage by air emboli to lung microvessels in anesthetized sheep.
J. Appl. Physiol.
57:
1360-1368,
1984
2.
Atkins, C. E.,
C. E. Lehner,
K. A. Beck,
R. R. Dubielzig,
E. V. Nordheim,
and
E. H. Lanphier.
Experimental respiratory decompression sickness in sheep.
J. Appl. Physiol.
65:
1163-1171,
1988
3.
Bassingthwaighte, J. B.,
M. A. Malone,
T. C. Moffett,
R. B. King,
S. E. Little,
J. M. Link,
and
K. A. Krohn.
Validity of microsphere deposition for regional myocardial flows.
Am. J. Physiol.
253 (Heart Circ. Physiol. 22):
H184-H193,
1987
4.
Bernard, S. L.,
R. W. Glenny,
H. H. Erickson,
M. R. Fedde,
N. L. Polissar,
R. J. Basaraba,
and
M. P. Hlastala.
Minimal redistribution of pulmonary blood flow with exercise in racehorses.
J. Appl. Physiol.
81:
1062-1070,
1996
5.
Butler, B. D.,
J. Conkin,
and
S. Luehr.
Pulmonary hemodynamics, extravascular lung water and residual gas bubbles following low dose venous gas embolism in dogs.
Aviat. Space Environ. Med.
60:
1178-1182,
1989[Medline].
6.
Butler, B. D.,
and
B. A. Hills.
The lung as a filter for microbubbles.
J. Appl. Physiol.
47:
537-543,
1979
7.
Butler, B. D.,
and
B. A. Hills.
Transpulmonary passage of venous air emboli.
J. Appl. Physiol.
59:
543-547,
1985
8.
Butler, B. D.,
B. C. Leiman,
and
J. Katz.
Arterial air embolism of venous origin in dogs: effect of nitrous oxide in combination with halothane and pentobarbitone.
Can. J. Anaesth.
34:
570-575,
1987[Medline].
9.
Butler, B. D.,
R. Robinson,
T. Little,
J. E. Chelly,
and
M. F. Doursout.
Cardiopulmonary changes with moderate decompression in rats.
Undersea Hyperb. Med.
23:
83-89,
1996[Medline].
10.
Chang, H. K.,
L. Delaunois,
R. Boileau,
and
R. R. Martin.
Redistribution of pulmonary blood flow during experimental air embolism.
J. Appl. Physiol.
51:
211-217,
1981
11.
Chang, H. K.,
M. E. Weber,
J. Thomson,
and
R. R. Martin.
Hydrodynamic features of pulmonary air embolism: a model study.
J. Appl. Physiol.
51:
1002-1008,
1981
12.
Cockett, A. T.,
S. M. Pauley,
D. N. Zehl,
A. A. Pilmanis,
and
W. S. Cockett.
Pathophysiology of bends and decompression sickness. An overview with emphasis on treatment.
Arch. Surg.
114:
296-301,
1979
13.
Deal, C. W.,
P. Barton,
F. Fielden,
and
I. Monk.
Hemodynamic effects of pulmonary air embolism.
J. Surg. Res.
11:
533-538,
1971[Medline].
14.
Flick, M. R.,
J. M. Hoeffel,
and
N. C. Staub.
Superoxide dismutase with heparin prevents increased lung vascular permeability during air emboli in sheep.
J. Appl. Physiol.
55:
1284-1291,
1983
15.
Gan, R. Z.,
and
R. T. Yen.
Vascular impedance analysis in dog lung with detailed morphometric and elasticity data.
J. Appl. Physiol.
77:
706-717,
1994
16.
Glenny, R. W.,
S. Bernard,
and
M. Brinkley.
Validation of fluorescent-labeled microspheres for measurement of regional organ perfusion.
J. Appl. Physiol.
74:
2585-2597,
1993
17.
Glenny, R. W.,
W. J. E. Lamm,
R. K. Albert,
and
H. T. Robertson.
Gravity is a minor determinant of pulmonary blood flow distribution.
J. Appl. Physiol.
71:
620-629,
1991
18.
Glenny, R. W.,
N. L. Polissar,
S. McKinney,
and
H. T. Robertson.
Temporal heterogeneity of regional pulmonary perfusion is spatially clustered.
J. Appl. Physiol.
79:
986-1001,
1995
19.
Hakim, T. S.,
G. W. Dean,
and
R. Lisbona.
Effect of body posture on spatial distribution of pulmonary blood flow.
J. Appl. Physiol.
64:
1160-1170,
1988
20.
Hakim, T. S.,
G. W. Dean,
and
R. Lisbona.
Quantification of spatial blood flow distribution in isolated canine lung.
Invest. Radiol.
23:
498-504,
1988[Medline].
21.
Hakim, T. S.,
R. Lisbona,
and
G. W. Dean.
Gravity-independent inequality in pulmonary blood flow in humans.
J. Appl. Physiol.
63:
1114-1121,
1987
22.
Hakim, T. S.,
R. Lisbona,
and
G. W. Dean.
Effect of cardiac output on gravity-dependent and nondependent inequality in pulmonary blood flow.
J. Appl. Physiol.
66:
1570-1578,
1989
23.
Hales, J. R. S.,
and
W. J. Cliff.
Direct observations of the behaviour of microspheres in microvasculature.
Bibl. Anat.
15:
87-91,
1977.
24.
Heyman, M. A.,
B. D. Payne,
J. I. Hoffman,
and
A. M. Rudolf.
Blood flow measurement with radionuclide-labeled particles.
Prog. Cardiovasc. Dis.
20:
55-79,
1977[Medline].
25.
Hlastala, M. P.,
S. L. Bernard,
H. H. Erickson,
M. R. Fedde,
E. M. Gaughan,
R. McMurphy,
M. J. Emery,
N. L. Polissar,
and
R. W. Glenny.
Pulmonary blood flow distribution in standing horses is not dominated by gravity.
J. Appl. Physiol.
81:
1051-1061,
1996
26.
Hlastala, M. P.,
M. A. Chornuk,
D. A. Self,
H. J. Kallas,
J. W. Burns,
S. Bernard,
N. L. Polissar,
and
R. W. Glenny.
Pulmonary blood flow redistribution by increased gravitational force.
J. Appl. Physiol.
84:
1278-1288,
1998
27.
Hlastala, M. P.,
H. T. Robertson,
and
B. K. Ross.
Gas exchange abnormalities produced by venous gas emboli.
Respir. Physiol.
36:
1-17,
1979[Medline].
28.
Huang, K. L.,
and
Y. C. Lin.
Activation of complement and neutrophils increases vascular permeability during air embolism.
Aviat. Space Environ. Med.
68:
300-305,
1997[Medline].
29.
James, P. B.
Dysbarism: the medical problems from high and low atmospheric pressure.
J. R. Coll. Physicians Lond.
27:
367-374,
1993[Medline].
30.
Jarvis, K. A.,
E. P. Steffey,
W. S. Tyler,
N. Willits,
and
M. Woliner.
Pulmonary blood flow distribution in anesthetized ponies.
J. Appl. Physiol.
72:
1173-1178,
1992
31.
Krenz, G. S.,
J. H. Linehan,
and
C. A. Dawson.
A fractal continuum model of the pulmonary arterial tree.
J. Appl. Physiol.
72:
2225-2237,
1992
32.
Loewenherz, J. W.
Pathophysiology and treatment of decompression sickness and gas embolism.
J. Fla. Med. Assoc.
79:
620-624,
1992.
33.
Mure, M.,
K. B. Domino,
H. T. Robertson,
M. P. Hlastala,
and
R. W. Glenny.
Pulmonary blood flow does not redistribute in dogs with reposition from supine to left lateral position.
Anesthesiology
89:
483-492,
1998[Medline].
34.
Ohkuda, K.,
K. Nakahara,
A. Binder,
and
N. C. Staub.
Venous air emboli in sheep: reversible increase in lung microvascular permeability.
J. Appl. Physiol.
51:
887-894,
1981
35.
Oyama, Y.,
and
M. P. Spencer.
Cardiopulmonary effects of intravenous gas embolism; with special reference to fate of intravascular gas bubbles.
Jpn. Circ. J.
35:
1541-1549,
1971[Medline].
36.
Pilmanis, A. A.
(Editor).
Proceedings of the 1990 Hypobaric Decompression Sickness Workshop. Brooks AFB, TX: US Air Force Armstrong Laboratory, 1992. (Publ. AL-SR-1992-0005)
37.
Presson, R. G., Jr.,
K. R. Kirk,
K. A. Haselby,
J. H. Linehan,
S. Zaleski,
and
W. W. Wagner, Jr.
Fate of air emboli in the pulmonary circulation.
J. Appl. Physiol.
67:
1898-1902,
1989
38.
Presson, R. G., Jr.,
K. R. Kirk,
K. A. Haselby,
and
W. W. Wagner, Jr.
Effect of ventilation with soluble and diffusible gases on the size of air emboli.
J. Appl. Physiol.
70:
1068-1074,
1991
39.
Sergysels, R.,
N. Jasper,
L. Delaunois,
H. K. Chang,
and
R. R. Martin.
Effect of ventilation with different gas mixtures on experimental lung air embolism.
Respir. Physiol.
34:
329-343,
1978[Medline].
40.
Shirai, M.,
K. Sada,
and
I. Ninomiya.
Diameter and flow velocity changes in small pulmonary vessels due to microembolization.
J. Appl. Physiol.
65:
288-296,
1988
41.
Steffey, E. P.,
B. H. Johnson,
and
E. I. D. Eger.
Nitrous oxide intensifies the pulmonary arterial pressure response to venous injection of carbon dioxide in the dog.
Anesthesiology
52:
52-55,
1980[Medline].
42.
Van-Liew, H. D.,
and
M. E. Burkard.
Bubbles in circulating blood: stabilization and simulations of cyclic changes of size and content.
J. Appl. Physiol.
79:
1379-1385,
1995
43.
Vann, R. D.,
and
E. D. Thalmann.
Decompression physiology and practice.
In: The Physiology and Medicine of Diving (4th ed.), edited by P. B. Bennett,
and D. H. Elliott. London: Saunders, 1993, p. 376-432.
44.
Verstappen, F. T.,
J. A. Bernards,
and
A. F. Kreuzer.
Effects of pulmonary gas embolism on circulation and respiration in the dog. II. Effects on respiration.
Pflügers Arch.
368:
599-604,
1977.
45.
Verstappen, F. T.,
J. A. Bernards,
and
F. Kreuzer.
Effects of pulmonary gas embolism on circulation and respiration in the dog. III. Excretion of venous gas bubbles by the lung.
Pflügers Arch.
370:
67-70,
1977[Medline].
46.
Vik, A.,
A. O. Brubakk,
T. R. Hennessy,
B. M. Jenssen,
M. Ekker,
and
S. A. Slordahl.
Venous air embolism in swine: transport of gas bubbles through the pulmonary circulation.
J. Appl. Physiol.
69:
237-244,
1990
47.
Waligora, J. M.,
D. J. Horrigan, Jr.,
and
J. Conklin.
The effect of extended O2 prebreathing on altitude decompression sickness and venous gas bubbles.
Aviat. Space Environ. Med.
58, Suppl.:
A110-A112,
1987[Medline].
48.
Walmrath, D.,
R. Konig,
C. Ernst,
H. Bruckner,
F. Grimminger,
and
W. Seeger.
Ventilation-perfusion relationships in isolated blood-free perfused rabbit lungs.
J. Appl. Physiol.
72:
374-382,
1992
49.
Walther, S. M.,
K. B. Domino,
R. W. Glenny,
and
M. P. Hlastala.
Pulmonary blood flow distribution in sheep: effects of anesthesia, mechanical ventilation, and change in posture.
Anesthesiology
87:
335-342,
1997[Medline].
50.
Walther, S. M.,
K. B. Domino,
R. W. Glenny,
N. L. Polissar,
and
M. P. Hlastala.
Pulmonary blood flow distribution has a hilar-to-peripheral gradient in awake, prone sheep.
J. Appl. Physiol.
82:
678-685,
1997
51.
Yen, R. T.,
and
Y. C. Fung.
Effect of velocity distribution on red cell distribution in capillary blood vessels.
Am. J. Physiol.
235 (Heart Circ. Physiol. 4):
H251-H257,
1978.
This article has been cited by other articles:
|
|
 |
|
  A. J. Calderon, J. B. Fowlkes, and J. L. Bull Bubble splitting in bifurcating tubes: a model study of cardiovascular gas emboli transport J Appl Physiol, August 1, 2005; 99(2): 479 - 487. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z Dujic, D Bakovic, I Marinovic-Terzic, and D Eterovic Acute effects of a single open sea air dive and post-dive posture on cardiac output and pulmonary gas exchange in recreational divers Br. J. Sports Med., May 1, 2005; 39(5): e24 - e24. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Haller, E. Faltin-Traub, D. Faltin, and C. Kern Oxygen embolism after hydrogen peroxide irrigation of a vulvar abscess Br. J. Anaesth., April 1, 2002; 88(4): 597 - 599. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
