Vol. 92, Issue 1, 279-287, January 2002
Lung albumin accumulation is spatially heterogeneous but not
correlated with regional pulmonary perfusion
Anthony J.
Gerbino1 and
Robb W.
Glenny1,2
Departments of 1 Medicine and 2 Physiology and
Biophysics, University of Washington, Seattle, Washington 98195
 |
ABSTRACT |
10.1152/japplphysiol.00353.2001.
The contribution of pulmonary
perfusion heterogeneity to the development of regional differences in
lung injury and edema is unknown. To test whether regional differences in pulmonary perfusion are associated with regional differences in microvascular function during lung injury, pigs were
mechanically ventilated in the prone position and infused with
endotoxin (Escherichia coli 055:B5, 0.15 µg · kg
1 · h1;
n = 8) or saline (n = 4) for 4 h.
Extravascular albumin accumulation and perfusion were measured in
multiple ~0.7-ml lung regions by injecting pigs with radiolabeled
albumin and radioactive microspheres, respectively. Extravascular
albumin accumulation was spatially heterogeneous but not correlated
with regional perfusion. Extravascular albumin accumulation was greater
in dorsal than ventral regions, and regions with similar albumin
accumulation were spatially clustered. This spatial organization was
less evident in endotoxemic than control pigs. We conclude that there
are regional differences in lung albumin accumulation that are
spatially organized but not mediated by regional differences in
pulmonary perfusion. We speculate that regional differences in
microvascular pressure or endothelial function may account for the
observed distribution of extravascular albumin accumulation.
blood flow; interstitial; lipopolysaccharide
| |
INTRODUCTION |
THE CLINICAL IMPORTANCE
OF ventilator-induced lung injury (1) has focused
attention on the spatial heterogeneity of acute lung injury. The acute
respiratory distress syndrome (ARDS) is characterized by a spatially
heterogeneous distribution of radiographic opacities (15,
25) and transvascular protein leak (30), but the
etiology of this heterogeneity is incompletely understood. Because lung
involvement in ARDS is spatially heterogeneous even when the initial
insult to the lung is diffuse, we reasoned that there must be a
comediator of injury that converts homogeneous initial insults into
heterogeneous disease, and this comediator should itself be
heterogeneously distributed.
Could pulmonary perfusion be one such mediator? Although regional
differences in pulmonary perfusion have well-recognized consequences
for ventilation-perfusion matching and gas exchange (17),
the contribution of perfusion heterogeneity to lung injury has not been
explored. Regional perfusion may increase local endothelial permeability via its relationship with endothelial shear stress (11), neutrophil sequestration (23, 37), or
microvascular pressures (36). In addition, regional
perfusion may contribute to transendothelial flux of water and protein
without altering permeability via effects on capillary surface area.
To test the hypothesis that regional differences in pulmonary perfusion
are associated with regional differences in microvascular function
during acute lung injury, we measured regional extravascular albumin
accumulation and regional perfusion in the lung during porcine
endotoxemia. Porcine endotoxemia has prominent pulmonary vascular
effects, increases neutrophilic inflammation and endothelial permeability (10), and frequently complicates ARDS
(29). Thus this animal model of sepsis-associated acute
lung injury is well suited to study the relationship between perfusion
heterogeneity and local microvascular function. Regional pulmonary
perfusion and regional extravascular albumin accumulation were measured with radioactive microspheres and radiolabeled albumin, respectively, and their spatial distributions were analyzed and compared.
| |
METHODS |
Animal preparation.
Twelve pathogen-free pigs weighing 19-25 kg were chemically
restrained with ketamine (20 mg/kg) and xylazine (2 mg/kg) and anesthetized with a thiopental infusion (10 mg · kg1 · h1) adjusted to
achieve a surgical plane of anesthesia and suppress spontaneous
ventilation (range: 10-17
mg · kg1 · h1). Pigs
breathed air and were mechanically ventilated with a piston-pump ventilator via tracheostomy without positive end-expiratory pressure. Respiratory rate was set so that arterial PCO2
was 30-35 Torr before endotoxin infusion and was not changed
thereafter, and tidal volume was 12 ml/kg. Lungs were hyperinflated to
twice tidal volume every 15 min to prevent atelectasis. Central venous,
arterial, and pulmonary arterial catheters were placed. Pigs were given 20 ml/kg normal saline intravenously and placed in the prone posture to
minimize the effects of the pleural pressure gradient. The University
of Washington Animal Care Committee approved all experiments.
Study protocol.
Escherichia coli O55:B5 endotoxin (Sigma Chemical, St.
Louis, MO) was continuously infused at 0.15 µg · kg1 · h1 through a
femoral venous catheter for 4 h. The infusion rate was halved in
one pig because severe pulmonary hypertension developed (mean pressure:
~70 mmHg) and doubled in another because neither hypoxemia nor
pulmonary hypertension had developed by 90 min. Core body temperature,
mean systemic and pulmonary arterial pressure, pulmonary arterial
occlusion pressure, thermodilution cardiac output (in triplicate;
Edwards COM 2, Baxter, Irvine, CA), and arterial blood gases were
measured before endotoxemia and every 60 min during endotoxin infusion.
We used two radiolabeled tracers injected at different times to measure
regional extravascular albumin accumulation between 3 and 4 h of
endotoxemia. Two different albumin tracers were used to allow
separation of total lung albumin into its intravascular and
extravascular components. Human serum albumin labeled with 131iodine (131I-HSA) was injected over 1 min via a femoral venous catheter after 3 h of endotoxemia.
Arterial blood was collected every minute for 5 min and then every 5 min thereafter. 125I-HSA was then injected over 1 min via a
femoral venous catheter after 3 h and 50 min of endotoxemia, and
arterial blood was collected every minute thereafter. As
125I-HSA circulated, ligatures were loosely place around
the aorta, pulmonary artery, and pulmonary veins that had been exposed
just before injection of 125I-HSA by median sternotomy.
When 125I-HSA had circulated for ~7 min, mixed venous and
arterial blood were simultaneously collected, a 60-mg thiopental bolus
was given to ensure deep anesthesia, and the vascular ligatures were tightened.
Regional perfusion was measured by injecting ~1.5 million
15-µm-diameter microspheres labeled with 141cerium
(141Ce; NEN, Boston, MA) via a femoral venous catheter
after 3 h of endotoxemia. We used radioactive rather than
fluorescent microspheres because tissue radioactivity was already being
measured to calculate extravascular albumin accumulation. Thus the use
of radioactive microspheres allowed us to measure regional perfusion
without introducing additional steps to measure tissue fluorescence.
Microspheres were vortexed, sonicated, suspended in normal saline, and
manually injected over 30 s.
We performed median sternotomy at the end of each experiment to ligate
the pulmonary artery and veins, thereby trapping blood in the pulmonary
circulation. We initially trapped blood in the pulmonary circulation
for another purpose (which we subsequently abandoned) but continued to
adhere to this protocol in all pigs because of concern that
interstitial albumin would shift between lung regions more easily
(either within interstitial lymphatics or via vascular drainage) if the
pulmonary vascular compartment were open to the atmosphere at the end
of the experiment. Because radioiodine circulated only briefly (<15
min in all pigs) after the switch to the supine position for median
sternotomy, this procedure should have had little effect on regional
albumin flux. Although median sternotomy may have altered regional
blood volume, thereby shifting intravascular radioiodine between lung
regions, regional shifts in intravascular radioiodine do not affect
calculated rates of regional extravascular albumin accumulation.
Tissue processing.
Lungs were removed from the chest with ligatures in place, inflated
with 25-cmH2O airway pressure, and air-dried for 48-72 h. Lungs were encased in foam and cut into ~1.2-cm-thick transverse slices, and cylindrical tissue samples with diameters of 0.86 cm
(~0.7 ml) were obtained from each slice with a cylindrical coring
device. Sampling locations were predetermined using a template containing multiple 0.86-cm-diameter holes arranged along rectilinear coordinates with centers separated by 1.2 cm. Tissue samples were rejected if they overlaid the lung edge or if >20% of the sample was
composed of airways. This algorithm yielded 258-373 and
220-407 tissue samples per pig in the control and endotoxin group,
respectively, and sampled ~30% of the lung parenchyma (each core
sampled 40% of the surrounding lung, and ~25% of cored tissue
samples were inadequate based on the above criteria). To account for
variability in slice thickness, we confirmed the height of each
cylindrical tissue sample to the nearest 0.5 mm. The lobe and
x-, y-, and z-coordinates of each
tissue sample were recorded, as were the coordinates of the right and
left hilum.
Measurement of radioactivity.
Radioactivity of plasma and tissue samples was determined in a gamma
counter (Minaxi model 5550, Packard, Downers Grove, IL) and corrected
for decay time and spillover by using the matrix inversion method.
Tissue samples were counted for 5 min, resulting in mean total counts
per sample of 125,000, 85,000, and 43,000 for 131I,
125I, and 141Ce, respectively. Plasma samples
were counted for 15 min, resulting in mean total counts per sample of
820,000 and 417,000 for 131I and 125I,
respectively. Because the 131I-to-125I-HSA
plasma ratio in blood drawn at the end of each experiment is critical
to the calculation of extravascular albumin accumulation (see Eq. 3), we measured radioactivity in approximately thirty 1-ml
arterial and mixed venous samples collected at the end of each
experiment to minimize error in this measurement.
Regional pulmonary perfusion.
Regional perfusion (ml/min) was estimated by multiplying thermodilution
cardiac output by the fraction of lung sampled by our algorithm (i.e.,
~0.3) and the 141Ce signal in a sample and dividing by
the 141Ce signal in all tissue samples. Coefficients of
variation (standard deviation/mean) were calculated by using
volume-normalized measurements of regional perfusion.
Preparation of radioiodine.
HSA (0.1 mg) was labeled with 125I or 131I
(22) just before the experiment. Disposable PD-10 columns
prepacked with Sephadex G-25M (Pharmacia, Piscataway, NJ) were used to
remove unbound radioiodine, and the percent unbound iodine was
quantified by electrophoresis with cellulose polyacetate paper. One and
one-half millicuries of 131I-HSA and 100 µCi of
125I-HSA were used in each experiment. 131I-HSA
and 125I-HSA contained 0.6 ± 0.3 (SD) and 0.4 ± 0.2% unbound radioiodine, respectively, The mean difference between
unbound 131I and unbound 125I was 0.2 ± 0.3 (SD) % in the endotoxin group and 0.3 ± 0.5% in the control
group. Modeling the effects of unbound radioiodine on albumin
accumulation rates suggests that, when unbound percentages of
131I and 125I differ by 0.2%, albumin
accumulation rates are overestimated by only 0.4%. This model assumes
mean plasma and tissue activities for 131I and
125I and rapid passage of unbound radioiodine into all body
tissues in proportion to weight.
Regional albumin accumulation rates.
Regional extravascular albumin accumulation was measured by adapting a
dual-tracer method validated by Graham and Evans (20) in
whole rat lungs. In this model, the interstitial and vascular compartments are separated by a permeable membrane with interstitial albumin accumulation proportional to plasma albumin concentration. If
131I-HSA and 125I-HSA are given after
T1 and T2 minutes of
endotoxemia, respectively, and the pig is killed after
Tend minutes of endotoxemia (where Tend is time at the end of the experiment), the
activities of 131I (A1,i) and
125I (A2,i) in tissue sample
i are related to their intravascular and extravascular
components by Eqs. 1 and 2
|
(1)
|
|
(2)
|
where Ai is activity in piece
i (µCi); Vi is plasma volume in
sample i (µl); Li is rate of
albumin accumulation in the extravascular space in sample i
(expressed as a plasma clearance, µl/min);
C1(Tend) and
C2(Tend) are plasma activities (µCi/µl) of 131I and 125I,
respectively, at the end of the experiment; and
C1(t) and
C2(t) describe plasma activities
(µCi/µl) of 131I and 125I, respectively,
over time and are integrated over the time that each radioisotope
circulated. Li is the mean rate of extravascular
albumin accumulation in the lung and is assumed to be identical for
131I-HSA and 125I-HSA.
C1(t) and
C2(t) were generated by curve fitting
131I and 125I plasma activity, respectively,
over time to a double exponential using least squares regression.
A1,i, A2,i, C1(Tend), and
C2(Tend) were measured
directly in a gamma counter.
Equations 1 and 2 were solved for
Li, giving
![L<SUB><IT>i</IT></SUB><IT>=</IT><FR><NU>{A<SUB>2,<IT>i</IT></SUB><IT>−</IT>[A<SUB>1,<IT>i</IT></SUB><IT>·C</IT><SUB>2</SUB>(<IT>T</IT><SUB>end</SUB>)<IT>/C</IT><SUB>1</SUB>(<IT>T</IT><SUB>end</SUB>)]}</NU><DE><FENCE><LIM><OP>∫</OP><LL><IT>T</IT><SUB>2</SUB></LL><UL><IT>T</IT><SUB>end</SUB></UL></LIM><IT> C</IT><SUB>2</SUB>(<IT>t</IT>)<IT>·</IT>d<IT>t−</IT>[<IT>C</IT><SUB>2</SUB>(<IT>T</IT><SUB>end</SUB>)<IT>/C</IT><SUB>1</SUB>(<IT>T</IT><SUB>end</SUB>)]<IT>·</IT><LIM><OP>∫</OP><LL><IT>T</IT><SUB>1</SUB></LL><UL><IT>T</IT><SUB>end</SUB></UL></LIM><IT> C</IT><SUB>1</SUB>(<IT>t</IT>)<IT>·</IT>d<IT>t</IT></FENCE></DE></FR>](/content/vol92/issue1/fulltext/279/img003.gif)
|
(3)
|
Li was normalized for differences in the
volume of each tissue sample, giving extravascular albumin accumulation rates in units of microliters of plasma per minute per milliliter lung tissue.
This simple model of protein movement has been compared
(20) with a more complete description in the lung
(5) that includes osmotic and hydrostatic water flux,
convective and diffusive protein flux, back diffusion, and lymph flow.
Results of this model are similar to those of the more complete model
over a range of permeability surface area products and microvascular
pressures. This method has two principal advantages over more commonly
used methods. First, we used a molecular tracer to mark the
intravascular space, eliminating error due to variability in lung
hematocrit that significantly affects regional lung measurements
(3, 14). Second, we do not assume that either
125I-HSA or 131I-HSA is confined to the
vascular space. Typically, one tracer is assumed to remain in the
vascular space, despite some degree of transvascular leak. Failure to
account for transvascular leak introduces error (31),
especially when endothelial permeability is increased or a small
molecular tracer is used.
Variability in Li due to counting statistic error.
To determine whether error due to counting statistics accounts for the
observed variability in Li, we introduced counting statistic error in each measurement of radioactivity used to
calculate Li by assuming typical radioactive signals from experimental data. We recalculated
Li using Eq. 3 and repeated this
process 500 times, yielding a hypothetical distribution for
Li consistent with propagation of counting statistic error. Variability due to counting statistic error was expressed as the coefficient of variation (standard deviation/mean) of
this distribution.
We also examined whether gamma-counter error (counting statistic error,
machine error, error due to changes in tissue position within the
scintillation vial, etc.) could account for observed variability in
Li by calculating Li from
radioactive signals measured five times on consecutive days in a subset
of tissue samples in two pigs. Variability was expressed as the
coefficient of variation for repeated measurements in each tissue sample.
Statistical analysis.
Data are reported as means ± SD, except where noted.
Physiological data from different time points were compared using
ANOVA. Post hoc comparisons were evaluated for statistical difference using the Bonferroni correction for multiple comparisons. Differences between control and endotoxin groups were evaluated with unpaired t-tests. Distributions of albumin accumulation rates within
each pig were characterized statistically by their mean, interquartile distance, and skewness (distributions could not be normalized by log
transformation). We used interquartile distance rather than standard
deviation as the measure of distribution breadth, because interquartile
distance is a better measure of breadth when distributions are skewed.
Perfusion distributions were characterized by the coefficient of
variation (standard deviation of perfusion/mean perfusion). Least
squares linear regression was used to describe the correlation between
regional albumin accumulation rates and regional pulmonary perfusion in
each pig. In addition, least squares linear regression was used to
determine the correlation between mean albumin accumulation rate and
the following measurements: mean pulmonary arterial pressure at 3 h, arterial PO2 at 3 h, the percent
difference in unbound 131I and unbound 125I,
and the number of tissue samples per pig.
Linear gradients.
We analyzed rates of regional extravascular albumin accumulation as a
function of cranial-caudal and ventral-dorsal position and as a
function of distance from the ipsilateral hilum (i.e., hilar-peripheral
position). On average, tissue samples were separated by a maximum of
12 ± 1 (SD) cm in the ventral-dorsal direction, 21 ± 2 cm
in the cranial-caudal direction, and 14 ± 2 cm in the hilar-peripheral direction. Lungs were corrected for tilt before these
relationships were calculated so that x-, y-, and
z-axes reflected the true anatomic right-left,
ventral-dorsal, and cranial-caudal directions, respectively. Slopes of
least squares regression lines were used to describe rates of regional
extravascular albumin accumulation as a function of ventral-dorsal,
cranial-caudal, or hilar-peripheral position. Hilar-peripheral distance
was defined as the Euclidean distance between the center of a piece and
the ipsilateral hilum. A given slope was considered statistically different from zero if it lay outside 95% confidence intervals generated with a permutation test, as previously described
(16).
Spatial correlation.
The tendency for neighboring lung regions to have similar albumin
accumulation rates was quantified by calculating the correlation between albumin accumulation rates in adjacent tissue samples (18). Briefly, pairs of tissue samples in the same lobe
that were 1.2 cm apart (i.e., samples that were nearest neighbors) were
identified, and the correlation between albumin accumulation rates from
paired tissue samples was calculated. Spatial correlation was
considered significantly different than zero if it fell outside 95%
confidence intervals for a random spatial distribution that was
calculated using a permutation test (16).
| |
RESULTS |
Physiological data.
Endotoxemia significantly increased pulmonary arterial pressure and
systemic arterial pressure and caused trends toward decreased arterial
PO2 and cardiac output compared with the
control group (Fig. 1). Pulmonary
arterial, systemic arterial, and pulmonary arterial occlusion pressures
and body temperature increased, and cardiac output, arterial pH, and
PO2 decreased, over time in endotoxemic pigs.
In contrast, only cardiac output changed significantly over time in the
control group and did not change significantly after the first hour.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 1.
Physiological data over time for control (open symbols)
and endotoxin (solid symbols) groups. A: mean systemic blood
pressure (BP) and cardiac output (CO). B: mean pulmonary
arterial (PA) pressure (Ppa) and pulmonary arterial occlusion pressure
(Ppaocc). C: arterial PO2,
PCO2, and pH. Ppa and systemic BP increased,
and there were trends toward decreased arterial
PO2 and CO in the endotoxin compared with
control group. Values are means ± SD. * Significant difference
between control and endotoxin groups, P < 0.05.
|
|
Pulmonary perfusion heterogeneity.
Pulmonary perfusion was spatially heterogeneous in endotoxemic and
control pigs (Fig. 2, Table
1). Regional perfusion varied two- to
threefold in the control lungs, consistent with previous reports in
uninjured lungs (18). However, endotoxemia increased the
fractions of lung that were poorly and well perfused, relative to the
control group (Fig. 2).
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 2.
Frequency histograms for regional pulmonary perfusion in
1 endotoxemic and 1 control pig after 3 h. CV, coefficient of
variation for the perfusion distribution. Expected variability is seen
in the control lung, but regions with very low or high levels of
perfusion increase during endotoxemia.
|
|
Spatial distribution of extravascular albumin accumulation.
Extravascular accumulation of radiolabeled albumin in the lung was
spatially heterogeneous in the endotoxemic and control pigs (Fig.
3, Table 1). However, there was no
correlation between regional extravascular albumin accumulation and
regional pulmonary perfusion in either group (Fig.
4, Table 1). Albumin accumulation rates
varied by an average of 5% because of counting statistic error and 7%
(mean standard deviation: 0.008 µl · min1 · ml lung1)
because of gamma-counter error and, therefore, cannot explain regional
differences in albumin accumulation observed in this study.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 3.
Frequency histograms for regional albumin accumulation
rates in 1 endotoxemic and 1 control pig. Note higher mean, broader
distribution, and rightward skew of distribution in the endotoxemic
lung.
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4.
Regional albumin accumulation rates as a function of
regional pulmonary perfusion in 1 endotoxemic pig. Each point
represents a single ~0.7-ml tissue sample (n = 220).
R2, coefficient of determination. There is no
relationship between regional albumin accumulation and pulmonary
perfusion. Lungs in the control group showed a similar relationship and
are, therefore, not shown.
|
|
Variability in regional extravascular albumin accumulation was not
randomly distributed in the lung. Adjacent lung regions had similar
rates of albumin accumulation (i.e., showed positive spatial
correlation; Fig. 5, Table
2) and rates were greater in dorsal
(nondependent) compared with ventral lung regions (Fig. 6, Table 2). There were no significant
differences between cranial and caudal regions or between hilar and
peripheral regions.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 5.
Spatial distribution of regional albumin accumulation in
1 control pig. Each cube (n = 257) represents a single
~0.7-ml tissue sample. Samples are color coded according to their
albumin accumulation rate and are shown in their anatomic position in
the lung. Note the slight dorsal predominance of albumin accumulation,
variability that is independent of ventral-dorsal position, and
tendency for nearest neighbors to have similar albumin accumulation
rates (spatial correlation = 0.73). Clustering of similar albumin
accumulation rates in neighboring lung regions was less evident in
endotoxemic lungs, and the dorsal predominance was less consistently
seen (not shown).
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 6.
Regional albumin accumulation as a function of vertical
position in the lung in 1 endotoxemic pig. Each point represents a
single ~0.7-ml tissue sample (n = 220). Note
regression line with slope 0.15 µl · min1 · ml
lung1 · cm1 that is significantly
different than zero. Albumin accumulation is threefold greater in
nondependent lung regions in this pig. Note the considerable
variability among regions with the same vertical position.
|
|
Spatial correlation and ventral-dorsal gradients were less impressive
in endotoxemic pigs. Endotoxemia significantly decreased spatial
correlation compared with the control group, and endotoxemic pigs with
the highest albumin accumulation rates showed the weakest spatial
correlation (Table 2). Endotoxemic pigs also exhibited a dorsal
predominance of albumin accumulation less consistently (5 of 8 pigs)
than did control pigs (4 of 4 pigs) (Table 2) and a trend toward less
steep ventral-dorsal gradients (P = 0.14 for difference
from zero).
The distribution of albumin accumulation rates was skewed to the right
and constrained at the origin (Fig. 3). A fraction of tissue samples
(3 ± 2%) had albumin accumulation rates less than zero, but
two-thirds of these were close enough to the origin (i.e., within 0.016 µl · min1 · ml1) to be
explained by counting statistic error. The remaining tissue samples
with albumin accumulation rates less than zero were not unique in terms
of their location in the lung, physical appearance, or 125I
and 131I signals. Two pigs accounted for the majority of
these samples, and most pigs had only zero or one such sample.
Variability in the response to endotoxin.
The physiological response to endotoxin varied widely. Some endotoxemic
pigs had little or no increase in extravascular albumin accumulation, whereas others had threefold increases relative to
controls (Table 1). On average, the mean albumin accumulation rate was
twofold greater in endotoxemic than control pigs, a difference that
approached statistical significance (Table 1).
Variability in the response to endotoxin was mirrored by variability in
the pulmonary vascular and gas-exchange response to endotoxin.
Pulmonary arterial pressure (r = 0.69) and arterial PO2 (r = 
0.73) at 3 h
were associated with the mean albumin accumulation rate, suggesting
that variability in albumin accumulation rates between pigs reflected
biological variability in the response to endotoxin. In contrast, mean
albumin accumulation rate was not correlated with the number of tissue
samples per pig (r = 0.21) or difference between
unbound 131I and 125I (r = 0.01), suggesting that differences in lung expansion and unbound
radioiodine do not explain variability in albumin accumulation rates
between pigs.
| |
DISCUSSION |
We have demonstrated regional differences in the accumulation of
extravascular albumin in endotoxemic and control pig lungs that are not
explained by regional differences in perfusion. Variability in regional
albumin accumulation is not randomly distributed in the lung but rather
is spatially correlated and increased in dorsal (nondependent) lung
regions. This spatial organization is present but decreased during endotoxemia.
Methodological error.
Experimental error is unlikely to account for the observed variability
in regional albumin accumulation between tissue samples. The error
estimated by simulating counting statistic error or by repeating
experimental measurements was small compared with observed variability
in albumin accumulation rates. Fractions of unbound 131I
and 125I were small and similar in magnitude, making it
unlikely that differences in unbound 131I and
125I, which rapidly enter the extravascular space, impacted
measurements of extravascular albumin accumulation.
The majority of tissue samples with albumin accumulation rates less
than zero were very close to zero and were, therefore, consistent with
error due to counting statistics. Negative albumin accumulation rates
that were not explained by counting statistic error may have been due
to a greater fraction of unbound 125I than 131I
or more rapid extravascular accumulation of 125I-HSA than
131I-HSA (violating the assumption that
Li in Eqs. 1 and 2 are identical).
Determinants of regional extravascular albumin accumulation.
Regional extravascular albumin accumulation is determined by the net
effects of albumin movement from the vasculature into the interstitium
and from the interstitium into adjacent interstitial and lymphatic
spaces. Transvascular albumin movement is initially determined by
diffusive and convective mechanisms, including endothelial permeability, microvascular surface area, transendothelial water flux
(itself depending on Starling forces), and the solvent-drag reflection
coefficient for albumin. Once vascular-interstitial equilibrium has
been reached, however, extravascular albumin content is determined by
the plasma albumin concentration and the interstitial volume accessible
to albumin.
To directly assess the contribution of transvascular albumin flux,
interstitial volume, and lymph albumin clearance to extravascular albumin accumulation, interstitial albumin concentration and lymph drainage would have to be measured in each tissue sample. Although measurement of total lung lymph flow has been made in some species (8, 9, 26, 27), measurements of regional lymph flow are
not possible. Although regional interstitial volume could have been
measured with a molecular tracer such as diethylenetriamine pentaacetic
acid, the interstitial volume accessible to albumin would still be unknown.
Inference from published data suggests that transvascular albumin flux
is more important than lymph drainage or interstitial volume in
determining extravascular albumin accumulation during the 1-h
circulation time for 131I-HSA in this study. Interstitial
volume determines extravascular albumin accumulation only as
plasma-interstitial equilibrium is approached. However, studies in dogs
(26, 28) and sheep (33) suggest that it takes
>1 h for the vascular and interstitial spaces to reach equilibrium
[half-time >2 h for normal and 0.7 h for abnormal lungs
(26, 28, 33)]. Because the interstitial concentration [and, therefore, lymph concentration (34)] of
131I-HSA is small shortly after injection of
131I-HSA, lymph clearance of 131I-HSA is also
unlikely to have had a major influence on extravascular albumin
accumulation. In addition, lymph drainage of radiolabeled albumin
toward the hila should have resulted in a hilar-peripheral gradient in
albumin accumulation, but such a gradient was not observed. For these
reasons, we believe that regional extravascular albumin accumulation
primarily reflects local transvascular albumin flux, with smaller
contributions from lymph albumin clearance and regional interstitial volume.
Relationship between pulmonary perfusion and regional albumin
accumulation.
Endotoxemia increased the number of lung regions that received either
high or low levels of perfusion compared with perfusion in uninjured
lungs (Fig. 2). The increase in poorly perfused regions is consistent
with endotoxin's effects on microvascular dysregulation observed in
other studies (13, 21, 24) and may, in part, explain organ
dysfunction even when total perfusion to an organ is preserved. These
changes in perfusion heterogeneity during endotoxemia potentially
magnify the effect of regional perfusion on local transvascular albumin flux.
Although regional differences in pulmonary perfusion may be associated
with local differences in endothelial shear stress (11),
vascular pressures (36), neutrophil sequestration
(23, 37), and perfused vascular surface area, our data
argue that regional differences in pulmonary perfusion are not
associated with local accumulation of extravascular albumin. These data
represent the first comparison of regional pulmonary perfusion and
extravascular protein accumulation but are consistent with previous
data showing no association between regional extravascular lung water
and regional pulmonary perfusion in normal and edematous dog lungs
(32). Because the spatial resolution of our methods is
limited, we cannot exclude an association between regional perfusion
and protein accumulation at the capillary level.
Possible causes of regional differences in lung albumin
accumulation.
The spatial distribution of extravascular albumin accumulation observed
in this study is in agreement with previous data (2) demonstrating spatial heterogeneity in lymph and, therefore,
interstitial albumin concentration (34). Albertine et al.
(2) measured regional lymph albumin concentration in sheep
lung and found increased concentrations in nondependent regions.
Because extravascular lung water does not vary with vertical position
in the lung (3, 14), increased albumin concentration
measured by Albertine et al. (2) is consistent with
increased albumin content observed in this study. Albertine et al. also
reported considerable variability in albumin concentration in lung
regions with the same vertical position, consistent with the
variability that we observed in isogravitational lung (Figs. 5 and 6).
Any explanation for regional differences in albumin accumulation must
account for the vertical gradient, isogravitational heterogeneity, and
positive spatial correlation in regional albumin accumulation. The
nondependent predominance of lymph and interstitial albumin has been
explained by the presence of a vertical, gravitationally mediated
vascular pressure gradient (35) in the absence of a vertical interstitial pressure gradient (6). The resulting vertical gradient in microvascular driving pressures increases transvascular flux of water out of proportion to protein flux in
dependent regions, thus increasing regional lymph flow and "washing
out" interstitial albumin (7). This mechanism decreases interstitial albumin in dependent regions if equilibrium has already been established (7) and slows albumin accumulation in
dependent regions if equilibrium has not yet been reached.
We speculate that regional differences in microvascular driving
pressure explain variability within isogravitational lung regions and
positive spatial correlation, as well as the vertical gradient in
albumin accumulation. Recent observations suggest that vascular
geometry plays an important role in determining the spatial
heterogeneity of pulmonary perfusion (4, 19), including
heterogeneity within isogravitational regions (19) and
positive spatial correlation (18). We hypothesize that
regional differences in vascular geometry (e.g., differences in
regional vascular diameters, branching angles, vessel lengths, etc.)
also cause heterogeneity in regional microvascular pressures, including heterogeneity within isogravitational regions. Because neighboring lung
regions share a similar heritage in the vascular tree, regional microvascular pressures resulting from local differences in vascular geometry are more alike in adjacent than distant lung regions and,
therefore, show positive spatial correlation. Thus the theoretical relationship between microvascular pressure and interstitial albumin proposed by Blake and Staub (7) and the hypothesis that
vascular geometry, as well as gravity, determines local microvascular
pressure together explain the isogravitational heterogeneity, positive spatial correlation, and vertical gradient that we have observed.
Regional differences in microvascular permeability are another
plausible explanation for variability in extravascular albumin accumulation. Yoneda (38) postulated the existence of
regional differences in endothelial permeability because extravascular lung water was uniformly distributed, despite the gravitational gradient in vascular driving pressures. Seeking structural correlates of endothelial permeability in uninjured lungs, he found the complexity of interendothelial tight junctions, which is associated with endothelial barrier function (12), to be greater in
dependent than nondependent lung. If microvascular pressures are
determined by both vascular geometry and gravity, and microvascular
endothelium develop structural differences in response to local
microvascular pressures, then spatial heterogeneity in endothelial
structure and function may also explain the isogravitational
heterogeneity, positive spatial correlation, and the vertical gradient
that we have observed.
If pulmonary perfusion and microvascular pressure are linked because
they both depend on vascular resistance, and microvascular pressure
determines extravascular albumin accumulation, it seems intuitive that
regional perfusion and albumin accumulation should be closely
associated. How then do we explain the observed lack of correlation
between regional perfusion and extravascular albumin accumulation?
Although regional microvascular pressure and perfusion both depend on
vascular resistance, they do so in different ways. Local microvascular
pressure depends on the relative resistance in upstream (arterial) and
downstream (venous) vessels, whereas local perfusion depends on the sum
of arterial, venous, and capillary resistances supplying one region
relative to this sum in other regions of the lung. For example, two
lung regions in which the sums of arterial, venous, and capillary
resistances are equal will have identical perfusion. However, if the
ratio of arterial-to-venous resistance in the first region exceeds that
in the second region, vascular pressure would be lower in the first
than in the second capillary bed. Thus the design of the pulmonary
circulation does not require a tight association between regional
microvascular pressure (and, therefore, extravascular albumin
accumulation) and regional perfusion.
Implications.
Regional differences in extravascular albumin accumulation may reflect
regional differences in the propensity of lung to develop edema and, in
turn, develop ventilation-perfusion mismatch, hypoxemia, and
ventilator-induced lung injury. Consequently, determining the etiology
of these differences may improve the basic understanding of acute lung
injury. Although perfusion heterogeneity does not explain regional
differences in lung albumin accumulation, other possible mediators
should be identified and pursued.
Variability in regional albumin accumulation in the lung provides
additional evidence that spatial heterogeneity is fundamental to
biological systems and should be considered when overall organ function is assessed. In particular, researchers should exercise caution when attempting to draw conclusions about whole organ function
from measurements in a single region.
In conclusion, this study demonstrates that there are spatially
organized regional differences in extravascular albumin accumulation in
the lung. Although pulmonary perfusion is heterogeneous, regional differences in perfusion do not explain regional variability in extravascular albumin accumulation.
| |
ACKNOWLEDGEMENTS |
The authors acknowledge the excellent technical assistance of Dowon
An, Shen-sheng Wang, and Dr. Susan Bernard. We are indebted to Dr.
Kenneth A. Krohn for providing expertise regarding preparation of
radiolabeled albumin and for use of his radiochemistry laboratory and
to Dr. Michael M. Graham for expert advice in methods development.
| |
FOOTNOTES |
Address for reprint requests and other correspondence: A. J. Gerbino, Division of Pulmonary/Critical Care Medicine, Virginia Mason Medical Center, 1100 9th Ave., PO Box 900, Seattle, WA 98111 (E-mail: anthony.gerbino{at}vmmc.org).
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. Section 1734 solely to indicate this fact.
Received 17 April 2001; accepted in final form 31 August 2001.
| |
REFERENCES |
1.
[Anonymous].
Ventilation with lower tidal volumes compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network.
N Engl J Med
342:
1301-1308,
2000[Abstract/Free Full Text].
2.
Albertine, KH,
Schultz EL,
Wiener-Kronish JP,
and
Staub NC.
Regional differences in pleural lymphatic albumin concentration in sheep.
Am J Physiol Heart Circ Physiol
252:
H64-H70,
1987[Abstract/Free Full Text].
3.
Baile, EM,
Pare PD,
Dahlby RW,
and
Hogg JC.
Regional distribution of extravascular water and hematocrit in the lung.
J Appl Physiol
46:
937-942,
1979[Free Full Text].
4.
Beck, KC,
and
Rehder K.
Differences in regional vascular conductances in isolated dog lungs.
J Appl Physiol
61:
530-538,
1986[Abstract/Free Full Text].
5.
Bert, JL,
and
Pinder KL.
Pulmonary microvascular exchange: an analog computer simulation.
Microvasc Res
27:
51-70,
1984[Web of Science][Medline].
6.
Bhattacharya, J,
Gropper MA,
and
Staub NC.
Interstitial fluid pressure gradient measured by micropuncture in excised dog lung.
J Appl Physiol
56:
271-277,
1984[Abstract/Free Full Text].
7.
Blake, LH,
and
Staub NC.
Pulmonary vascular transport in sheep. A mathematical model.
Microvasc Res
12:
197-220,
1976[Web of Science][Medline].
8.
Brigham, KL,
Bowers R,
and
Haynes J.
Increased sheep lung vascular permeability caused by Escherichia coli endotoxin.
Circ Res
45:
292-297,
1979[Abstract/Free Full Text].
9.
Brigham, KL,
Bowers RE,
and
McKeen CR.
Methylprednisolone prevention of increased lung vascular permeability following endotoxemia in sheep.
J Clin Invest
67:
1103-1110,
1981.
10.
Brigham, KL,
and
Meyrick B.
Endotoxin and lung injury.
Am Rev Respir Dis
133:
913-927,
1986[Web of Science][Medline].
11.
Broccard, AF,
Hotchkiss JR,
Kuwayama N,
Olson DA,
Jamal S,
Wangensteen DO,
and
Marini JJ.
Consequences of vascular flow on lung injury induced by mechanical ventilation.
Am J Respir Crit Care Med
157:
1935-1942,
1998[Abstract/Free Full Text].
12.
Claude, P,
and
Goodenough DA.
Fracture faces of zonulae occludens from "tight" and "leaky" epithelia.
J Cell Biol
58:
390-400,
1973[Abstract/Free Full Text].
13.
Drazenovic, R,
Samsel RW,
Wylam ME,
Doerschuk CM,
and
Schumacker PT.
Regulation of perfused capillary density in canine intestinal mucosa during endotoxemia.
J Appl Physiol
72:
259-265,
1992[Abstract/Free Full Text].
14.
Flick, MR,
Perel A,
Kageler W,
and
Staub NC.
Regional extravascular lung water in normal sheep.
J Appl Physiol
46:
932-936,
1979[Abstract/Free Full Text].
15.
Gattinoni, L,
Pelosi P,
Vitale G,
Pesenti A,
D'Andrea L,
and
Mascheroni D.
Body position changes redistribute lung computed-tomographic density in patients with acute respiratory failure.
Anesthesiology
74:
15-23,
1991[Web of Science][Medline].
16.
Gerbino, AJ,
Altemeier WA,
Schimmel C,
and
Glenny RW.
Endotoxemia increases relative perfusion to dorsal-caudal lung regions.
J Appl Physiol
90:
1508-1515,
2001[Abstract/Free Full Text].
17.
Gerbino, AJ,
McKinney S,
and
Glenny RW.
Correlation between ventilation and perfusion determines 
A/
heterogeneity in endotoxemia.
J Appl Physiol
88:
1933-1942,
2000[Abstract/Free Full Text].
18.
Glenny, RW.
Spatial correlation of regional pulmonary perfusion.
J Appl Physiol
72:
2378-2386,
1992[Abstract/Free Full Text].
19.
Glenny, RW,
Lamm WJE,
Albert RK,
and
Robertson HT.
Gravity is a minor determinant of pulmonary blood flow distribution.
J Appl Physiol
71:
620-629,
1991[Abstract/Free Full Text].
20.
Graham, MM,
and
Evans ML.
A simple, dual tracer method for the measurement of transvascular flux of albumin into the lung.
Microvasc Res
42:
266-279,
1991[Web of Science][Medline].
21.
Hinshaw, LB.
Sepsis/septic shock: participation of the microcirculation: an abbreviated review.
Crit Care Med
24:
1072-1078,
1996[Web of Science][Medline].
22.
Hunter, WM,
and
Greenwood FC.
Preparation of iodine-131 labeled human growth hormone of high specific activities.
Nature
194:
495-496,
1962[Medline].
23.
Kuebler, WM,
Kuhnle GE,
and
Goetz AE.
Leukocyte margination in alveolar capillaries: interrelationship with functional capillary geometry and microhemodynamics.
J Vasc Res
36:
282-288,
1999[Web of Science][Medline].
24.
Lam, C,
Tyml K,
Martin C,
and
Sibbald W.
Microvascular perfusion is impaired in a rat model of normotensive sepsis.
J Clin Invest
94:
2077-2083,
1994.
25.
Maunder, RJ,
Shuman WP,
McHugh JW,
Marglin SI,
and
Butler J.
Preservation of normal lung regions in the adult respiratory distress syndrome. Analysis by computed tomography.
JAMA
255:
2463-2465,
1986[Abstract/Free Full Text].
26.
Parker, JC,
Crain M,
Grimbert F,
Rutili G,
and
Taylor AE.
Total lung lymph flow and fluid compartmentation in edematous dog lungs.
J Appl Physiol
51:
1268-1277,
1981[Abstract/Free Full Text].
27.
Parker, JC,
Falgout HJ,
Parker RE,
Granger DN,
and
Taylor AE.
The effect of fluid volume loading on exclusion of interstitial albumin and lymph flow in the dog lung.
Circ Res
45:
440-450,
1979[Abstract/Free Full Text].
28.
Parker, JC,
Parker RE,
Granger DN,
and
Taylor AE.
Vascular permeability and transvascular fluid and protein transport in the dog lung.
Circ Res
48:
549-561,
1981[Abstract/Free Full Text].
29.
Parsons, PE,
Worthen GS,
Moore EE,
Tate RM,
and
Henson PM.
The association of circulating endotoxin with the development of the adult respiratory distress syndrome.
Am Rev Respir Dis
140:
294-301,
1989[Web of Science][Medline].
30.
Sandiford, P,
Province MA,
and
Schuster DP.
Distribution of regional density and vascular permeability in the adult respiratory distress syndrome.
Am J Respir Crit Care Med
151:
737-742,
1995[Abstract].
31.
Taylor, AE,
and
Parker JC.
Pulmonary interstitial spaces and lymphatics.
In: Handbook of Physiology. The Respiratory System. Circulation and Nonrespiratory Functions. Bethesda, MD: Am. Physiol. Soc, 1985, sect. 3, vol. I, chapt. 4, p. 167-230.
32.
Tsang, JY,
Baile EM,
and
Hogg JC.
Relationship between regional pulmonary edema and blood flow.
J Appl Physiol
60:
449-457,
1986[Abstract/Free Full Text].
33.
Vaughan, TR, Jr,
Erdmann AJ, 3rd,
Brigham KL,
Woolverton WC,
Weidner WJ,
and
Staub NC.
Equilibration of intravascular albumin with lung lymph in unanesthetized sheep.
Lymphology
12:
217-223,
1979[Web of Science][Medline].
34.
Vreim, CR,
Snashall PD,
Demling RH,
and
Staub NC.
Lung lymph and free interstitial fluid protein composition in sheep with edema.
Am J Physiol
230:
1650-1653,
1976.
35.
West, JB.
Distribution of blood flow in isolated lung: relation to vascular and alveolar pressures.
J Appl Physiol
19:
713-724,
1964[Abstract/Free Full Text].
36.
West, JB,
and
Mathieu-Costello O.
Stress failure of pulmonary capillaries: role in lung and heart disease.
Lancet
340:
762-767,
1992[Web of Science][Medline].
37.
Wiggs, BR,
English D,
Quinlan WM,
Doyle NA,
Hogg JC,
and
Doerschuk CM.
Contributions of capillary pathway size and neutrophil deformability to neutrophil transit through rabbit lungs.
J Appl Physiol
77:
463-470,
1994[Abstract/Free Full Text].
38.
Yoneda, K.
Regional differences in the intercellular junctions of the alveolar-capillary membrane in the human lung.
Am Rev Respir Dis
126:
893-897,
1982[Web of Science][Medline].
J APPL PHYSIOL 92(1):279-287
8750-7587/02 $5.00
Copyright © 2002 the American Physiological Society
TOP
ABSTRACT
INTRODUCTION
