| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
A,
, and A/
during PLV: effects of nitroprusside and inhaled
nitric oxide
Departments of 1 Medicine (Pulmonary and Critical Care Unit), 2 Pediatrics, and 3 Anesthesia and Critical Care, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts 02114
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Partial liquid ventilation (PLV) with high-specific-weight perfluorocarbon liquids has been shown to improve oxygenation in acute lung injury, possibly by redistributing perfusion from dependent, injured regions to nondependent, less injured regions of the lung. Our hypothesis was that during PLV in normal lungs, a shift in perfusion away from dependent lung zones might, in part, be due to vasoconstriction that could be reversed by infusing sodium nitroprusside (NTP). In addition, delivering inhaled NO during PLV should improve gas exchange by further redistributing blood flow to well-ventilated lung regions. To examine this, we used a single transverse-slice positron emission tomography camera to image regional ventilation and perfusion at the level of the heart apex in six supine mechanically ventilated sheep during five conditions: control, PLV, PLV + NTP, and PLV + NO at 10 and 80 ppm. We found that PLV shifted perfusion from dependent to middle regions, and the dependent region demonstrated marked hypoventilation. The vertical distribution of perfusion changed little when high-dose intravenous NTP was added during PLV, and inhaled NO tended to shift perfusion toward better ventilated middle regions. We conclude that PLV shifts perfusion to the middle regions of the lung because of the high specific weight of perflubron rather than vasoconstriction.
sodium nitroprusside; positron emission tomography; partial liquid ventilation; ventilation; perfusion; alveolar ventilation
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
PARTIAL LIQUID
VENTILATION (PLV) with perfluorocarbon fluids (6)
has been proposed as a treatment for the acute respiratory distress
syndrome. PLV has been shown to provide improved gas exchange (4,
10-13, 15-17, 24, 25, 27, 28, 33) in animal models of
lung injury. Although several studies have attempted to identify the
mechanisms responsible for the changes in global gas exchange caused by
PLV in normal and acute respiratory distress syndrome lungs, little is
known about the regional effects of PLV on the normal lung. Before
understanding how PLV alters the ventilation-perfusion
(
A/
) relationships in lung injury, it is
critical to understand what happens in the normal lung.
Fuhrman et al. (6) showed a slight increase in shunt
during PLV, with relative desaturation as the fraction of inspired oxygen (FIO2) dropped below 0.30. However,
with blood gases alone, it is impossible to identify the regional
alveolar ventilation (A) and perfusion
() relationships that caused this decrease in oxygenation.
Furthermore, in the setting of PLV, the shunt fraction calculated from
the Berggren equation (2) may be unreliable, because the
underlying assumption that the end-capillary blood is fully saturated
with 100% inspired O2 in liquid-filled areas is questionable.
Multiple inert gas elimination technique (MIGET) during PLV of healthy
piglets (19) showed an increase in
A/ heterogeneity during PLV, which was
independent of perflubron dose. Ventilation heterogeneity was found to
be the major factor in this increase, but it was not possible to
determine whether there was a predominance of high or low
A/ during PLV. Furthermore, an increase in the Berggren shunt fraction was found in all animals. The authors speculated that this was a result of a combination of true shunt and
diffusion limitation in perflubron-filled lung regions. Recently, fluorescent microspheres were used to examine the distribution of
regional during PLV in healthy lambs (5). This
study showed a redistribution of away from dependent regions
with PLV, particularly in the hilar lung regions. These results were also supported using a similar technique in healthy pigs
(21). None of these techniques was able to provide
information regarding regional A, and therefore the
contributions to the shifts in between hypoxic pulmonary
vasoconstriction (HPV) and the high specific weight of the
perfluorocarbons remain unknown.
Our laboratory has described a method to quantify the anatomic
distribution of A, , and
A/ by using positron emission tomography
(PET) (20, 32). In this paper, we applied this method to
characterize the regional changes A/ caused
by PLV in the normal lung. In preliminary studies (35,
39), we observed that PLV caused a substantial upward shift in
and a decrease in A of dependent
liquid-filled regions. We theorized that the upward shift in
could be caused by buoyant forces resulting from the perfluorocarbon's
high specific weight and by local HPV induced by the lower
A in dependent regions. In this paper, we sought to
confirm our preliminary findings and to explore the mechanisms
responsible for the observed changes in . We reasoned that if
HPV was the dominant mechanism causing the upward shift in ,
then a global reduction of vascular tone with high-dose intravenous
sodium nitroprusside (NTP) should reverse the shift in . In
contrast, if buoyant forces were the dominant mechanism causing the
upward shift in during PLV, the reduction of pulmonary vascular
tone should result in no change in the distribution of or in a
further upward shift as buoyancy forces became more dominant. Finally,
we sought to determine the effects on regional of an inhaled
pulmonary vasodilator (NO) and compare them with the intravenous NO
donor (NTP).
| |
METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Experimental Setup
The experimental setup has been described in detail elsewhere (32). Briefly, the setup included a single-ring PET camera (PCR-1), a mechanical ventilator, and an infusion system. PCR-1 is a high-sensitivity stationary camera that is able to trigger image collection in synchrony with a signal from a mechanical ventilator. 13NN-labeled gas, produced by a cyclotron, was forced into solution with previously degassed saline, resulting in 13NN-labeled saline with specific activity ranging from 0.1 to 0.2 mCi/ml. The infusion system included a peristaltic pump and a remotely controlled solenoid valve system that allowed flushing with 13NN-labeled saline of the tubing leading to a right jugular venous line terminating in the superior vena cava.Animal Preparation
The animal care committee of the Massachusetts General Hospital approved all protocols and procedures. Six sheep weighing 13.2 ± 0.9 (SD) kg (range 12-14 kg) were anesthetized with thiopental sodium (30 mg/kg) and maintained under deep anesthesia with a continuous infusion. A tracheotomy was performed for insertion of a 7.0 endotracheal tube. The ventilator (Harvard Apparatus, Millis, MA) was set at a breathing frequency of 10 breaths/min, the inspiratory time was set to 30% of the breathing period, positive end-expiratory pressure (PEEP) was set to 5 cmH2O, and FIO2 was 1.0. Tidal volume (VT) was set to maintain normocapnic arterial blood gases (mean VT = 21.1 ± 2.9 ml/kg, PCO2 = 42.5 ± 6.8 Torr). The right femoral artery and vein were cannulated for pressure monitoring, blood gas sampling, and administration of intravenous fluids and/or intravenous NTP. A 7.5-French pulmonary artery thermodilution catheter (Baxter Healthcare, Deerfield, IL) was inserted in the left femoral vein for measurement of cardiac output, pulmonary arterial pressure, central venous and wedge pressures, and mixed venous blood gases. A right jugular venous catheter was inserted to the superior vena cava for injection of the 13NN during PET imaging (see below). Pancuronium bromide was administered in 0.2 mg/kg intravenous doses as needed to prevent respiratory efforts after adequate sedation was achieved. Physiological data collection included heart rate, arterial and pulmonary blood pressures, cardiac output, wedge pressure, and arterial and venous blood gases. Oxygen saturation was calculated from the blood gases by using an oxyhemoglobin dissociation curve for sheep described by Sharan and Popel (29). Shunt fraction was calculated by use of the Berggren shunt equation.PLV
Room temperature, nonpreoxygenated perflubron (C8F17Br; LiquiVent, Alliance Pharmaceutical, San Diego, CA) was instilled via the endotracheal tube. Perflubron doses (30 ± 7 ml/kg) were slowly poured into the airway over ~5 min, with breaths administered intermittently throughout the instillation. Dosing was complete when 30 ml/kg had been administered or a meniscus in the endotracheal tube was observed to be above the level of the trachea. The animal was rocked gently from side to side to facilitate even mixing within the lungs. Animals were ventilated for 20 min before imaging to ensure steady-state conditions.PET Imaging
The animals were positioned in the camera for a transverse imaging slice that included the apex of the heart as determined by a short transmission scan. To correct for gamma ray energy attenuation caused by the supporting structures, body tissues, and perflubron during PLV, transmission scans were performed for each condition studied. An imaging run consisted of a transmission scan and a series of emission scans imaging the fate of the radioactive tracer of 13NN-labeled saline injected into the superior vena cava. Infusate volume ranged from 3 to 12 ml depending on the tracer's specific activity to produce PET images with equivalent number of counts/voxel.The emission scans were collected in the following manner. First, the ventilator was interrupted at end exhalation to allow the lungs to reach functional residual capacity (FRC). Simultaneously, intravenous infusion of the tracer was started, and a collection of three consecutive 10-s images was initiated. At the end of collection of the third image, ventilation was resumed, and four consecutive 30-s images were collected as the tracer was washed out from the lung. These images were collected, and the transmission scans were gated by using a signal from the ventilator synchronized with the start of inhalation as described in detail elsewhere (32). The gating scheme yielded a set of two images for each scan, corresponding to the first and second halves of a respiratory cycle. Because inspiratory time of the ventilator was set at 30% of the breathing period, the first image included inspiration and most of expiration, whereas the second image mostly included the lung at FRC.
Protocol
Imaging runs were performed for each of five experimental conditions: 1) control gas ventilation (GV), 2) PLV with ~30 ml/kg of perflubron, 3) PLV during infusion of 320 µg/min intravenous NTP, 4) PLV with 10 ppm inhaled NO, and 5) PLV with 80 ppm inhaled NO. All conditions used an FIO2 of 1.0 except those cases that included addition of NO, in which the addition of a small flow of 800 ppm NO reduced the FIO2 slightly (i.e., FIO2 0.90 at 80 ppm NO and FIO2 0.98 at 10 ppm NO). The NO gas was mixed at the fresh gas intake of the ventilator, and its concentration was confirmed with an NO analyzer. After the first two conditions, the order of the remaining three conditions was randomly selected for each animal. Between imaging runs with NTP and NO, the animals were allowed to return to PLV control conditions, as evidenced by the return of mean arterial and pulmonary arterial pressures to within 10 percent of baseline values. Before each imaging run and the corresponding collection of physiological data, the lungs were inflated and sustained for 20 s at a pressure of 30-40 cmH2O to minimize the occurrence of microatelectasis and the loss of compliance. The animal was then allowed to reach a steady state (~10 min) before the initiation of the corresponding run.PET Image Analysis
PET images were initially corrected for camera sensitivity and for tissue attenuation. Image reconstruction was then performed with a convolution back-projection algorithm by using a Hanning filter yielding an effective spatial resolution of 10 mm. Images collected during the apnea period at FRC were reconstructed by using the second of the gated transmission scans, and images collected during the breathing washout period were reconstructed by using the sum of the two gated transmission scans. Resulting images consisted of an interpolated matrix of 159 × 159 voxels of 0.2 × 0.2 cm corresponding to a transverse slice of 5-mm thickness. Reconstructed images of local counts per voxel were then processed as discussed below to yield functional images of, A,
A/, and lung density.
Masking.
For each animal, masks defining the lung field at FRC were created by
applying a semiautomatic threshold algorithm to a template made as the
sum of all emission scans collected during the breath-hold period. The
algorithm assigned a value of unity to voxels within the lung field and
a value of zero to voxels outside the lung field. Masks were manually
refined by comparing them with the second image of the gated
transmission scans. The lung masks were then divided into 1-cm-high
horizontal regions of interest (ROIs) starting from the most dependent
lung region. The most nondependent ROI often had a vertical height much
less than 1 cm and was not considered if it contained less than 80 voxels. This yielded 12 ROIs (Fig. 1)
that were used to assess the vertical variation of ,
A, A/, and gas content as
described below.
|
Perfusion.
Because of the very low partition coefficient for nitrogen (
= 0.018) between gas and water (or blood), on arrival to the capillary
bed the tracer 13NN diffuses almost completely into the
alveolar airspace at first pass and remains there during apnea until
ventilation is resumed. Thus, after the period of arrival, regional
tracer content is directly proportional to . In this protocol,
was calculated from the sum of second and third images
collected during the apnea period. Average perfusion per voxel for each
1-cm-high ROI was calculated and then normalized by the average
perfusion per voxel of the whole lung field.
Ventilation.
Images were corrected for radioactive tracer decay back to the time of
bolus infusion. Regional specific alveolar ventilation for each voxel
was defined as the inverse of the exponential time constant
(
wo) of tracer removal during the washout as
|
(1) |
/A were created
by adding voxel by voxel the tracer content during the four images
collected during the washout, which approximates a perfusion-weighted
perfusion-ventilation ratio.
Fractional gas content. The transmission scans were processed to calculate fractional gas content within the lung field. As described above, a gated transmission scan yielded two images: the first image corresponded to the average local density during inhalation and the first second of exhalation, and the second image corresponded to the average local density during the last 3 s of exhalation when the lungs were mostly at FRC. The sum of these two images is proportional to the average local density during the entire breathing cycle. By using this information, it was possible to compare the average regional gas content over the breathing cycle with that at FRC for both gas and PLV as described in the APPENDIX.
A/ analysis.
A global lung A was estimated from the
VT and frequency by assuming an anatomic dead space based
on body weight (31). Global A/
ratio for the total lung was calculated by dividing the estimated
global A by the measured thermodilution cardiac output. Two types of plots were generated from the PET images to
analyze the matching between A and :
voxel-by-voxel scatter plots of mean-normalized A
vs. mean-normalized , and log(A/) distribution histograms. Voxel values of A and
were obtained from the corresponding images described above.
Outliers were removed from the data sets by eliminating any voxel with
negative ventilation, with perfusion greater than four times the mean,
or with a log(A/) value that fell outside of
three standard deviations from the mean. This process left intact
96.8 ± 0.4% SE of the data that were mean renormalized
and plotted as mean-normalized A vs. mean-normalized
. From these plots, the Pearson's product-moment correlation
coefficient (
), the variance of ventilation
(
A/
(
A) vs. mean
normalized log() allowed calculation of the same variables for
the logged data. Variance in the A/
distributions were also assessed using an equation derived by Wilson
and Beck (40), where A/,
A, and are measured according to the
following equation
|
(2) |
A/ values
were obtained by dividing voxel by voxel the image of
A by that of . A/
values ranging from 0.01 to 100 were grouped into 50 bins of equal
width. For each voxel, the fraction of total imaged ,
A, or volume (V) was added to the corresponding bin
of A/ for that voxel. This resulted
in histograms of A/ grouped by perfusion,
ventilation, or lung volume (within the image slice), each having
a total area of 1. These plots differ from those by Wagner et al.
(36) in that the first and last bins are not shunt and
dead space, but A/ ratios of 0.01 and 100, respectively, because we did not directly measure shunt and dead space
with this method. From these histograms, the standard deviations of the
A/ grouped by perfusion [SDlog(A/),], ventilation
[SDlog(A/),A], and lung volume [SDlog(A/),V] were calculated.
Statistical Analysis
The Student's two-tailed t-test for dependent samples was used to determine significant changes in physiological data and PET data. Four comparisons were made: PLV vs. control, PLV + NTP vs. PLV, PLV + NO at 10 ppm vs. PLV, and PLV + NO at 80 ppm vs. PLV. Statistical significance was defined by P < 0.05. The Bonferroni adjustment was also applied to reduce the inflation of type I error caused by multiple comparisons. Significance after the Bonferroni adjustment was defined by P < 0.0125. For a sample size of 6, there is 80% power to detect a standardized effect of 2.005 using a paired t-test with a 0.0125 two-sided significance level. For the few comparisons with sample sizes of five and four, the standardized effect increases to 2.477 and 3.448, respectively. Data are expressed as means ± SD unless otherwise noted. Statistical analysis was performed using STATISTICA '98 edition (StatSoft, Tulsa, OK).| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Global Physiological Parameters
Mean arterial pressure (MAP) did not change significantly during PLV compared with control or with PLV + NO at 10 or 80 ppm compared with PLV. However, the infusion of NTP during PLV significantly decreased MAP compared with PLV (Table 1). Mean pulmonary arterial pressure (MPAP) increased during PLV compared with control. When NTP or NO was added to PLV, MPAP decreased significantly in all cases compared with PLV. Pulmonary capillary wedge pressure dropped significantly with PLV + NTP or PLV + NO at 80 ppm compared with PLV. Peak inspiratory pressure rose by a mean of 5 cmH2O during PLV, and this was unchanged by the addition of NTP or NO at 10 or 80 ppm. There was a significant decrease in arterial PO2 (mean 47%) to 222 Torr during PLV. PLV + NO at 10 ppm increased arterial PO2 significantly to 334 Torr. Mixed venous PCO2 was significantly lower during PLV + NTP or PLV + NO at 10 ppm compared with PLV alone. Calculated Berggren shunt fraction was significantly greater during PLV than control. Shunt fraction decreased significantly compared with PLV in both NO conditions. Estimated globalA/ was near 1 except for PLV + NTP and
PLV + NO 80 ppm, where the possible corresponding nonsignificant
increases in cardiac output caused whole lung
A/ to fall in both conditions. However, A/ did not reach statistical significance in
either condition compared with PLV alone. All other physiological
variables did not show significant differences when compared with
either control or PLV conditions.
|
Regional Distribution of , A,
and A/ During the Control Condition (GV)
demonstrated clear
vertical gradients in the individual sheep images (Fig.
2, row 1). Local distributions
of A/ are visually different in each sheep (Fig. 3, row 1), with some
sheep demonstrating very homogeneous A/ (sheep 1 and
6) and others (sheep 4 and 5) showing
preexisting heterogeneity of A/. Despite this
appearance, when plotted vs. vertical height, relative
A/ was close to unity in
all sheep for most ROIs except for some of the most nondependent ones, where the ratio often deviated substantially in the positive or negative directions (Fig. 6, row 1). To quantify the
magnitude of the changes in the vertical axis, we analyzed the ,
A, and gas content images in 1-cm-high horizontal
ROIs and plotted the mean values for all six sheep (Fig.
4). In control conditions (Fig.
4A), relative and A showed clear
vertical gradients, increasing monotonically from ROI 12 (nondependent,
ventral) to ROI 2 (dependent, dorsal) with a small drop in both
and A in ROI 1. The vertical gradients were 17%/cm
for and 14%/cm for A. Fractional gas
content decreased consistently from nondependent to dependent ROIs,
with values ranging from 0.5 to 0.6 in the nondependent zones and 0.0 to 0.3 in the dependent zones. There was very little difference in
fractional gas content between the sum of the two gated images and
second gated image of gas content during control GV (see Fig.
4A).
|
|
|
During GV, there was good correlation between local values of
A and , yielding a mean value of
Avs. for all
sheep of 0.863 ± 0.025 (See Table 3). Correlation plots for a
representative sheep (sheep 5) are shown graphically in Fig.
7. The variance of A/
(A and during GV
that decreased during PLV. The good correlation between
A and during GV resulted in relatively
narrow distributions of A/ grouped by
A
[SDlog(A/,A)= 0.179 ± 0.015], by
[SDlog(A/,A) = 0.168 ± 0.010], or by organ volume
[SDlog(A/,V) = 0.193 ± 0.018] (see Table 3). This is illustrated graphically for
sheep 5 in Fig. 8. Although with substantially different
values, equivalent parameters calculated for the log-transformed data followed a similar pattern to the nontransformed data (see Table 3).
The calculated standard deviation of the A/
histograms, grouped by perfusion, ventilation, or organ volume, showed
similar changes to those of
Regional Distribution of ,
A, and
A/ During
PLV
consistently shifted away
from the most dependent lung regions toward middle zones during PLV
(Fig. 2, row 2). This shift was of variable magnitude and was best demonstrated in sheep 1. Despite the reduction in
relative in the dependent regions, there was a clear localized
area of low A/ in the most dependent regions
in the images (Fig. 3, row 2). The magnitude of these
changes is demonstrated when plotted against vertical height (Fig. 4).
PLV shifted away from the bottom third of the lung toward the
middle ROIs (Fig. 4B). A similar but stronger shift occurred
in A, leading to a substantial reduction in relative
A in the bottom third of the lung. Fractional gas
content decreased substantially in PLV in all but the most nondependent
ROI, becoming negligible in the three most depended ROIs. In contrast
to GV, during PLV there was a measurable difference in the fractional
gas content between the sum of the two gated images [mean lung volume
(MLV)], yielding a higher fractional gas content in middle ROIs than
that collected during the last 2 s of exhalation (close to FRC).
To illustrate the individual regional changes in caused by PLV,
NTP, and NO, they are plotted in Fig. 5
for each animal. The shift in relative during PLV compared with
control is substantial (Fig. 5, row 1). This amounted to an
increase of 2 and 11% in the nondependent and middle regions,
respectively, and a decrease of 13% in the dependent region (Table
2). Relative A changed similarly but more considerably. There was an increase of 8 and 17% in
the nondependent and middle regions, respectively, and a decrease of
24% in the dependent region (Table 2). The changes in
A and from control to PLV were statistically
significant except for the increase in in the nondependent
zone. The vertical distribution of relative
A/ was also substantially affected by PLV
(Fig. 6). PLV caused a systematic
decrease in relative A/ of
dependent ROIs and a concomitant increase in nondependent ones.
Relative A/ crossed from less than unity to
greater than unity in four animals above 4 cm from the most
nondependent lung, and in the other two animals it crossed above 2 cm
(Fig. 6, row 2).
|
|
|
The dispersion of local decreased in relation to control
(A worsened to a mean value of
Avs. = 0.478 ± 0.134 such that
A and . The A/ distribution widened
substantially during PLV
[SDlog(A/),A= 0.255 ± 0.034, SDlog(A/), = 0.272 ± 0.046 and
SDlog(A/),V = 0.305 ± 0.056, Table 3 and Fig. 8].
|
|
|
Regional Distribution of ,
A, and
A/ During PLV
+ NTP
A, , and fractional gas content
were of lesser magnitude and more heterogeneous when either NTP or NO
was added to PLV, and therefore the averages for these conditions are
not shown. The individual changes for each sheep in relative
are shown in Fig. 5, row 2. Sheep 1,
2, and 6 demonstrated an increase in relative
in dependent regions and a concomitant drop in middle regions.
In contrast, sheep 3 and 4 decreased to
dependent and nondependent ROIs, shifting it to middle ROIs, whereas
sheep 5 had virtually no vertical redistribution in .
Because the changes in A and were small,
there was little change in relative A/ (Fig.
6, row 3). The distributions of A/
systematically narrowed with NTP
[SDlog(A/),A = 0.210 ± 0.034, SDlog(A/), = 0.228 ± 0.066, and
SDlog(A/),V = 0.256 ± 0.069, Table 3 and Fig. 7]. The significant improvement in
Avs., although
these changes individually did not reach significance.
Regional Distribution of ,
A, and
A/ During PLV + NO
away from the most dependent
zones toward middle ones of the lung (Fig. 5, rows 3 and 4). For NO at
10 ppm, this amounted to an increase in relative of ~5% in
the middle ROIs (Table 2). For NO at 80 ppm, the shift resulted in an
increase in relative of 4% in the middle ROIs (Table 2).
Inhalation of NO at 10 ppm during PLV caused a modest but consistent
improvement in vertical A/ matching compared
with PLV particularly seen by a shift of
A/ toward unity in the most dependent
ROIs (Fig. 6, row 4). The same was true for NO at 80 ppm
(Fig. 6, row 5). There was little change in
Avs. by addition
of NO at 10 ppm or 80 ppm (Table 3 and Fig. 7) to PLV. There was a
significant improvement in
A/ with
NO at 10 and 80 ppm (Table 3).
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
To understand the mechanisms responsible for improved oxygenation
during PLV in injured lungs, we felt that it was important to
understand the effects of PLV in normal lungs. The major experimental findings of this study in healthy mechanically ventilated sheep were as
follows: 1) during GV there were vertical gradients of local
and A, with increasing values in the
gravitational direction. There was a high degree of spatial correlation
between these variables and narrow A/
distributions. 2) During PLV, and
A shifted away from dependent regions toward middle
regions of the lung. In the most dependent regions, the drop in
A was much greater than that in , generating
areas of very low A/. The correlation between
local A and decreased, and there was a
concomitant widening of the A/ distributions.
3) Intravenous infusion of high-dose NTP during PLV caused a
decrease in A/ distributions.
4) Inhaled NO at 10 ppm during PLV decreased
Limitations of Methods
Before discussing these results, it is important to acknowledge the limitations of our methods. We used a single-slice PET camera to obtain a transverse image of the lungs at the level of the cardiac apex. Therefore, one must be careful in extrapolating the results obtained from this slice to all regions of the lung. However, because we used healthy animals in this study, potential regional nonuniformities along the rostral-caudal axis can be expected to be minor. Gravitational effects caused by the high specific weight of perflubron can be expected to be similar along the rostral-caudal axis. Other effects, such as vasodilation from NTP or inhaled NO, however, may not be as predictable.We attempted to minimize carryover effects of either NTP or inhaled NO
by waiting 20 min between conditions and until baseline physiological
measurements returned to within 10% of their baseline values. We
cannot be certain, however, that residual local vasodilatory effects
were not present despite normal global physiological variables. However, we randomized the order of the last three conditions to
minimize the potential systematic changes in A,
, and A/ that might result from this effect.
We cannot be certain that NTP completely relaxed pulmonary vasomotor
tone. We performed pilot studies with 15% hypoxia and demonstrated
complete return of MPAP back to baseline values at the dose of NTP used
in this study. We must acknowledge that this does not prove that
regional returned to normal. In fact, cardiac output usually
increases with NTP, and because of nonlinear pressure-flow relationships, it is likely that conditions are different with NTP + hypoxia than in control. We believe that these differences are likely
to be small compared with the dramatic changes in during PLV,
such that if HPV were largely responsible for the shift in perfusion,
high-dose NTP should measurably change the pattern of .
We imaged with PET the local concentrations of the tracer
13NN to measure the distributions of A,
, and A/. Details of this technique
can be found elsewhere (20, 32). The feasibility of these
measurements depends on a low partition coefficient ( = 0.018)
for nitrogen between blood and alveolar air spaces. As a result,
virtually all injected 13NN diffuses into the air spaces at
first pass and remains there during breath hold. Although the partition
coefficient for blood/perflubron ( = 0.043) is higher than that
for blood/air, it is still sufficiently low to provide an adequate
assessment of gas exchange.
The PET camera used has a limited spatial resolution of ~10 mm.
Therefore, heterogeneity in A, , and
A/ occurring below this length scale could
not be assessed. However, the mean value of tracer activity within this
region is highly accurate. Because we delivered the 13NN
intravenously, the initial concentration of the tracer before the
washout depends on regional : units receiving very low will have little initial activity. Therefore the measurement of regional A reflects exclusively gas transport of
perfused units and excludes the ventilation of dead space units.
GV
As in a previous report using PET of blood flow distribution on supine dog lungs (32), we found a significant vertical gradient in lung perfusion in these supine sheep (Fig. 4A). The magnitude of the vertical gradient (17%/cm) was similar to the gradient reported in our laboratory with the same technique for dogs (15%/cm) (32). This number is higher than the vertical gradients of 7.2%/cm (38) in one study and 7.8%/cm (37) in another using fluorescent microspheres in sheep breathing room air with no PEEP. The differences in these gradients compared with ours may be due to methodological differences between PET and the microspheres method or due to differences in FIO2 or PEEP. We do not believe that the difference in gradient was caused by methodology because a recent comparison of both methods in the same animal yielded equivalent results (34). A PEEP of 5 cmH2O in the former microsphere study by Walther et al. (37) caused the vertical gradient to increase to 10.4%/cm. Thus part of the difference between our results and the microsphere data is likely due to our use of PEEP to prevent dependent microatelectasis. The remaining difference may be due to our use of a higher FIO2 because an FIO2 of 1.0 should cause relaxation of vascular tone, thus potentiating gravitational gradients. A previous study comparing the distribution of between supine
and prone dogs has suggested that, in lungs with minimal pulmonary
arterial tone (ventilated with FIO2 = 1), the effects of gravity and lung structure are balanced out for the
prone position but become additive in the supine position, resulting in
substantial vertical gradients (32). Ventilation with room
air may have globally increased basal pulmonary arterial tone and thus
uniformly increasing the contribution of regional vascular resistance
to distribution decreasing the relative importance of the
vertical direction to heterogeneity measured supine. It is also likely
that the mechanical ventilation of supine animals without PEEP used in
those studies may have potentiated the formation of dependent
atelectasis causing hypoxic vasoconstriction and thus reducing the
magnitude of the vertical gradient. Vertical gradients of 6%/cm and
7%/cm have been reported for supine mechanically ventilated primates
(8) and dogs (1, 9), respectively, also using
fluorescent microspheres. The reasons for the discrepancy in vertical
gradients are most likely similar to those mentioned above. We do not
think that species differences are responsible because of the small
variation in vertical gradients between the species studied with each
method (sheep = dogs in PET, and dogs = primates in microspheres).
Visually, it is clear that the vertical dependence in blood flow was
much greater in magnitude than any isogravitational variability seen in
the single-slice PET images (Fig. 2, row 1). We cannot, however, comment on the rostral-caudal gradient in this study. Doctor
et al. (5), using radiolabeled microspheres in supine lambs (conventional GV), found substantial vertical gradients in
regional with flow favoring dependent lung zones in all slices
except for the diaphragmatic region, where no gradient was observed.
The lack of PEEP in that protocol may have been responsible for that
finding, as discussed above.
During GV, we also observed a vertical gradient in A
favoring the dependent regions (14%/cm) but of lower magnitude than that corresponding gradient for (Fig. 4). As a result, there was a small gradient in A/ with values
increasing from dependent to nondependent zones (Fig. 6, row
1). In addition, there were two animals (Fig. 6, row 1,
sheep 1 and 4) that had a low
A/ region in the most nondependent zone.
These areas, however, should have had little effect on overall gas
exchange because of the low relative reaching that zone. The
vertical gradient in A was somewhat greater than the
vertical gradient of 9%/cm previously reported for supine dogs
(32) and may have been caused by species differences in
airway structure or bronchial smooth muscle tone.
Heterogeneity of A and during GV in this
study was higher than previously estimated by others. Wilson and Beck
(40) estimated a value of 0.07 for
The correlation
(Avs.) between
A and in our study was high (0.86) and
similar to 0.81 from our previous study in dogs (32).
Wilson and Beck (40) did not estimate the correlation between A and , because simultaneous
measurements of these variables were not available at that time. Mure
et al. (22) found
Avs. to be 0.76 in the supine position when using the fluorescent microspheres technique in pigs. However, because microsphere data for both A and are expected to be equally affected by
heterogeneity in piece size in that study, the overestimation of
Avs. caused by
the induced pseudocorrelation between these variables was not calculated.
Wilson and Beck (40) proposed a theoretical model to
estimate Avs.. Using this
model, we found that the estimated value of
A/ values. This was to be
expected because Eq. 2 is only exact when
A and data are log-transformed or 
log(
) or 
)
vs. mean-normalized