Journal of Applied Physiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


J Appl Physiol 92: 745-762, 2002; doi:10.1152/japplphysiol.00377.2001
8750-7587/02 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (13)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chang, H.
Right arrow Articles by Hlastala, M. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chang, H.
Right arrow Articles by Hlastala, M. P.
Vol. 92, Issue 2, 745-762, February 2002

Spatial distribution of ventilation and perfusion in anesthetized dogs in lateral postures

Hung Chang1,5, Stephen J. Lai-Fook4, Karen B. Domino3, Carmel Schimmel2, Jack Hildebrandt1,2, H. Thomas Robertson1,2, Robb W. Glenny1,2, and Michael P. Hlastala1,2

Departments of 1 Physiology and Biophysics, 2 Medicine, and 3 Anesthesiology, University of Washington, Seattle, Washington 98195; 4 Center for Biomedical Engineering, University of Kentucky, Lexington, Kentucky 40506; and 5 Division of Chest Surgery, Department of Surgery, Tri-Service General Hospital, National Defense Medical School, Taipei, Taiwan


    ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES

We aimed to assess the influence of lateral decubitus postures and positive end-expiratory pressure (PEEP) on the regional distribution of ventilation and perfusion. We measured regional ventilation (VA) and regional blood flow (Q) in six anesthetized, mechanically ventilated dogs in the left (LLD) and right lateral decubitus (RLD) postures with and without 10 cmH2O PEEP. Q was measured by use of intravenously injected 15-µm fluorescent microspheres, and VA was measured by aerosolized 1-µm fluorescent microspheres. Fluorescence was analyzed in lung pieces ~1.7 cm3 in volume. Multiple linear regression analysis was used to evaluate three-dimensional spatial gradients of Q, VA, the ratio VA/Q, and regional PO2 (PrO2) in both lungs. In the LLD posture, a gravity-dependent vertical gradient in Q was observed in both lungs in conjunction with a reduced blood flow and PrO2 to the dependent left lung. Change from the LLD to the RLD or 10 cmH2O PEEP increased local VA/Q and PrO2 in the left lung and minimized any role of hypoxia. The greatest reduction in individual lung volume occurred to the left lung in the LLD posture. We conclude that lung distortion caused by the weight of the heart and abdomen is greater in the LLD posture and influences both Q and VA, and ultimately gas exchange. In this respect, the smaller left lung was the most susceptible to impaired gas exchange in the LLD posture.

pulmonary gas exchange; spatial gradients; fluorescent microspheres; mediastinal shift; regional blood flow; positive end-expiratory pressure


    INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES

STUDIES (11, 33) HAVE SUGGESTED that regional blood flow (Q) and ventilation (VA) are more uniform in the prone compared with the supine posture. This was attributed to greater Q and VA to the dorsal lung regions, offsetting the effects of gravity. Also, the reduction in the dependent lung volume by the weight of the heart might contribute to differences in regional perfusion and VA (3). In the lateral decubitus posture, the compression of the dependent lung by the heart might be greater than that in either supine or prone posture. Furthermore, the dependent lung has a smaller lung volume in the left (LLD) than in the right lateral decubitus (RLD) posture (35), consistent with differing effects of the heart between the two postures.

Q increases from the nondependent to dependent lung in the lateral decubitus posture in the human (6, 25, 28) and the dog (20, 37). In addition, the intrapulmonary ventilation distribution is altered in the lateral posture after anesthesia and mechanical ventilation (38, 39). Impaired gas exchange after anesthesia and mechanical ventilation has been attributed to a mismatch between ventilation and perfusion. The relative matching of pulmonary blood flow and ventilation distribution in lateral decubitus postures remains uncertain.

Positive end-expiratory pressure (PEEP) is often applied to improve arterial oxygenation during anesthesia with mechanical ventilation (28, 38, 39). In the lateral posture, PEEP has been shown to reduce the difference in ventilation between the two lungs (39). In these studies (28, 38, 39), ventilation was measured in relatively large regions, such as a single lobe or lung. Accordingly, direct evidence using a high spatial resolution measure of the VA distribution in the lateral decubitus posture after induction of anesthesia with PEEP is lacking.

We hypothesize that the distributions of VA and Q are different between the LLD and RLD posture because of differences in lung distortion caused by the weight of the heart and abdominal contents. PEEP reduces lung distortion due to heart weight, may affect the smaller left dependent lung to a greater degree, and produces a larger decrease in pulmonary vascular resistance in the LLD posture. PEEP may increase both VA and Q to dependent lung regions, particularly to the dependent left lung in the LLD posture.

This study examines the effect of posture and PEEP on Q, ventilation, and gas exchange in the lateral decubitus posture in anesthetized, mechanically ventilated dogs, using both aerosolized and intravenously injected fluorescent microspheres.


    METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES

Animal Preparation and Physiological Measurements

This study was approved by the University of Washington Animal Care Committee. Six healthy mongrel dogs of either sex [22.8 ± 2.8 (SD, n = 6) kg] were anesthetized with pentobarbital sodium (~4-8 mg/kg iv) and maintained with a pentobarbital infusion sufficient to achieve a surgical plane of anesthesia and eliminate spontaneous ventilation (~10-17 mg · kg-1 · h-1). Dogs were mechanically ventilated with air via tracheostomy [tidal volume (VT) of 15 ml/kg]. The respiratory rate was adjusted to maintain arterial PCO2 (PaCO2) between 35 and 40 Torr. Minute ventilation was measured with a spirometer. Catheters were placed in one femoral artery and both femoral veins. A pulmonary artery catheter was introduced into the right external jugular vein and used for measuring cardiac output (QT; thermal dilution) and core temperature (Tc). Systemic arterial (Pa), pulmonary arterial (Ppa), pulmonary capillary wedge (Ppcw), and airway pressure (Paw) were recorded continuously on a data management system (Western Graphic Mach 12 DMS 1000). For determination of anatomic dead space (VD), exhaled end-tidal PCO2 was digitally sampled with an infrared CO2 detector (Perkin-Elmer, Plumsteadville, PA), and expiratory airflow (V) was measured by pneumotachograph. Arterial and mixed venous blood gases were measured with an automated blood-gas analyzer (ABL 300, Radiometer, Copenhagen, Denmark) and corrected for temperature. Body temperature was maintained by using heat lamps and pads.

Study Protocol

Animals were studied in the right and left lateral decubitus postures with 0 or 10 cmH2O PEEP, in random order. The lungs were fully inflated (30-40 cmH2O) 5 min before each experimental measurement to remove atelectasis. In each trial, we measured Pa, Ppa, Ppcw, QT, Tc, VT, and arterial and venous blood gas composition immediately before fluorescent microsphere administration. Q was measured with intravenously injected microspheres, and VA was measured with fluorescent aerosols as described below. Functional residual capacity (FRC) was measured by He dilution during each experimental condition.

Multiple Inert Gas Measurements

Pulmonary gas exchange was characterized and analyzed by MIGET, the multiple inert gas elimination technique (46, 47). Distributions of ventilation-perfusion ratio (VA/Q) were estimated by use of a 50-compartment model (47). Inert gas shunt (QS/QT) and dead space (VD/VT) were obtained from the model. Data from five of six animals are presented since one animal was rejected because of the presence of technical errors.

Fluorescent-labeled Microsphere Technique

VA was measured in both the LLD and RLD posture with and without PEEP by delivering aerosolized orange, orange-red, yellow, or yellow-green 1-µm-diameter fluorescent microspheres (FluoSpheres, Molecular Probes, Eugene, OR) into the ventilator circuit during a 5-min period (40). Simultaneously, Q was measured by injecting blue-green, green, crimson, or red 15-µm-diameter fluorescent microspheres via the femoral venous catheter, in five increments over the 5 min. To avoid clumping, microspheres were sonicated and vortexed before administration. Microsphere colors were randomly varied across experiments.

Terminally, the animals were deeply anesthetized with pentothal (150 mg/kg iv); then saline, heparin (20,000 units), and papaverine (2 mg/kg) were administered. The animals were exsanguinated via the arterial cannula. After a median sternotomy, the left atrium and pulmonary artery were cannulated, the aorta was tied off, and the lungs were perfused with 2% dextran solution to remove the blood. The lungs were removed and dried by inflation (~25 cmH2O) to total lung capacity (TLC). The pleura was pierced in several locations with a needle to facilitate drying. To maintain a normal anatomic configuration, the apical and most ventrocaudal rims of the left and right lungs were joined by tissue glue.

After 7 days, the dried lungs were coated with polyurethane foam (Kwik Foam, DAP, Dayton, OH) and placed in a plastic-lined square box with the caudal-cranial axis of the lung parallel to the wall of box. The box was filled with a rapidly setting foam (Polyol and isocyanate, International Sales, Seattle, WA). The solid block was sliced into 1.2-cm-thick slices. With use of a miter box, the lung slices were diced into cubes with 1.2-cm sides. Each lung piece was weighed and assigned x0, y0, and z0 coordinates measured from the left, dorsal, and caudal lung edges, representing the left-right, dorsal-ventral, and caudal-cranial axes, respectively. Samples <0.008 g were discarded. Fluorescent dye was extracted by soaking each piece in 1.5 ml 2-ethoxy ethyl acetate (Cellosolve, Aldrich Chemical, Milwaukee, WI) for 4 days. Dye concentrations were measured with an automated luminescence spectrophotometer (Perkin-Elmer, model LS-50B, Norwalk, CT) at the dye-specific excitation and emission wavelength. A matrix inversion method (42) was used to correct the fluorescent signal spillover from adjacent colors.

Data Processing

Adjusting cube dimensions from TLC back to in situ FRC conditions. To estimate spatial gradients in blood flow and ventilation per regional lung volume at the time of microsphere injection, the dimensions of each cube were adjusted from their TLC measurements to estimated in vivo values using previous measurements from anesthetized dogs (20, 37). In the LLD posture, the ratios of the maximum lung lengths at FRC to those at TLC measured radiographically were 0.68 (0.85 in RLD posture), 0.76, and 0.84 in the vertical (x), dorsal-ventral (y), and caudal-cranial (z) axis, respectively. Adjusted vertical heights at FRC averaged 11.1 ± 0.9 and 14.0 ± 1.2 cm in the LLD and RLD posture, respectively, without PEEP; and 16.4 ± 1.5 cm for both postures with PEEP. The adjusted dorsal-ventral and caudal-cranial lengths averaged 13.1 ± 1.1 and 23.9 ± 2.3 cm, respectively.

A second adjustment was made for the nonuniform deformation caused by the vertical gradient in transpulmonary pressure (Ptp). At the adjusted lung height xi measured from the bottom of the lung, (Ptp)i is given by
(Ptp)<IT>i=</IT>G<IT>xi</IT> (1)
where G is the vertical Ptp gradient (0.5 cmH2O/cm height) measured in the lateral posture in the dog (1) and i refers to the ith piece. (Ptp)min is 0 cmH2O at the bottom of the lung at xmin = 0 (1) and increases linearly to (Ptp)max at the top of the lung at xmax, where min and max refer to minimum and maximum, respectively. Typically, xmax at FRC was 13 cm and (Ptp)max was 6.5 cmH2O. We used the Ptp-lung volume (PV) curve of an isolated dog lung (27) to determine the changes in length corresponding to different values of Ptp along the height of the lung (xi). Lung volume (Vi) at each xi is given by
V<IT>i=</IT>V<SUB>min</SUB><IT>+</IT>(V<SUB>max</SUB><IT>−</IT>V<SUB>min</SUB>)(Ptp)<IT>i/</IT>(Ptp)<SUB>max</SUB> (2)
where (Delta Ptp)i is the Ptp change from the value at Vmin and (Delta Ptp)max is the maximum change in Ptp from Vmin to Vmax. We assumed that the PV curve is linear in this Ptp range. Vmin (20% TLC) is the lung volume at (Ptp)min at the bottom of the lung xmin. Vmax (65%TLC) is the lung volume at (Ptp)max. The cube at mid-lung height (xmid), equal to (xmax + xmin)/2, is assumed to remain undistorted (u) with a cube length equal to the value after the reduction from TLC to FRC (Delta xu). The deformed length (Delta xi) for the ith cube at xi is assumed to vary as V1/3, given by the PV curve, at each (Ptp)i
&Dgr;xi/&Dgr;xu=(V<IT>i/</IT>V<SUB>mid</SUB>)<SUP>1<IT>/</IT>3</SUP> (3)
Vmid equals (Vmax + Vmin)/2. The cube lengths were reduced below and expanded above the undeformed mid-lung height. The x positions of the deformed cubes were obtained by summing the deformed cube lengths starting from the bottom. The y and z dimensions of all cubes at each xi were adjusted by multiplying the values obtained after adjustment from TLC to FRC by Delta xi/Delta xu (Eq. 3).

We assumed no vertical Ptp gradient in the lateral position with PEEP (2). In addition, lung volume at 10 cmH2O PEEP was ~85% TLC; thus cube length (proportional to V1/3) was 0.95 times the cube length at TLC. Accordingly, no adjustment was made to the dried lung lengths at 10 cmH2O PEEP.

Volume normalization of blood flow. Fluorescent intensity of each color microsphere representing Q or VA to each piece (cube) was converted to units of blood flow (ml/min) by dividing the fluorescence intensity of each piece by the sum of the fluorescent intensities of all pieces and then multiplying by QT (ml/min). Q and VA of each piece were then converted to units of milliliters per minute per unit regional lung volume at FRC by dividing by its volume (Vi). Vi at xi was calculated from its dry weight Wi, mean lung density (rho ), and the deformed cube length (Eq. 3)
V<IT>i=</IT>4.7W<IT>i</IT>(<IT>&Dgr;xi/&Dgr;xu</IT>)<SUP>3</SUP><IT>/&rgr;</IT> (4)
The term (Delta xi/Delta xu)3 adjusts the mean density for changes due to the Ptp gradient. The product of Wi and 4.7, lung wet-to-dry-weight ratio (43), is the wet weight. Mean lung density at FRC was equal to total lung wet weight (total dry weight × 4.7) divided by total lung volume at FRC (air volume; FRC + volume of tissue mass). Tissue density was 1 g/ml. The measured FRC values are summarized in Table 1. Lung dry weight averaged 10.1 ± 0.7 and 13.6 ± 0.95 g for the left and right lung, respectively. Left lung weight was 24% smaller than right lung weight. Mean lung density averaged 0.14 ± 0.02 and 0.08 ± 0.01 g/ml at FRC and 10 cmH2O PEEP, respectively.

                              
View this table:
[in this window]
[in a new window]
 
Table 1.   Effect of position and PEEP on physiological variables

VD was obtained by using Fowler's method (13) from the plot of exhaled CO2 concentration vs. exhaled volume. VD was estimated by averaging results from three consecutive concentration-volume plots. Total ventilation was calculated by multiplying frequency by VT after subtracting VD.

Gas exchange parameters: VA/Q and PO2. The VA/Q was used to calculate end-capillary PO2. We used the method described by Altemeier et al. (4) to calculate arterial PO2, PCO2, alveolar-arterial O2 differences (A-aDO2), and regional PO2 (PrO2) from the measured VA and Q fluorescent intensities, Hb concentration, body temperature, and mixed venous blood gases. With the measured VA/Q for each piece, the mass balance equations for O2, CO2, and N2, end-capillary O2, and CO2 contents were solved to obtain regional alveolar PO2 and PCO2 for each piece. Regional alveolar gas tensions of each piece were ventilation weighted, and end-capillary gas contents were perfusion weighted and summed to yield mixed alveolar gas tension and mixed arterial gas content.

Statistical Analysis

Q and VA per unit regional lung volume were used for all analyses. Values were presented as means ± SD. A paired t-test was used to evaluate a difference between two groups. ANOVA was used to evaluate differences among more than two groups. A P value < 0.05 was considered significant.

Spatial variations using multiple linear regression analysis. A multiple linear regression model (StatView v. 5.0.1, SAS) was used to characterize the magnitude of Q (and other variables VA, VA/Q, and PrO2) as a linear function of the rectangular coordinates, x [vertical height in the left (or right) lateral posture, left-to-right (or right-to-left) direction], y (dorsal-ventral direction), and z (caudal-cranial direction)
<A><AC>Q</AC><AC>˙</AC></A><IT>=</IT>I<IT>+ax+by+cz+dxy+eyz+fzx+gxyz</IT> (5)
We subtracted the distances of the center of mass in the x, y, and z directions from the original coordinate system used for locating each lung cube to describe blood flow in relation to the center of mass (x = y = z = 0). The x, y, z coordinate distances of the center of mass of left, right, and whole lung, relatively to the original coordinate axes (x0, y0, z0) are summarized in the APPENDIX. The intercept (I) represents the mean blood flow at the center of mass. The center of blood flow was located near (<1 cm) the center of mass. The coefficients (a-g) and intercept in the linear equation (Eq. 6) describe the mean blood flow at an arbitrary position (x, y, z) within the lung. The coefficients a, b, and c define the blood flow gradients with respect to the x, y, and z coordinate axes, respectively, at the center of mass. The coefficients d, e, and f describe the variation of a gradient in one coordinate axis with respect to another axis, for example, the partial derivative of Q with respect to x results in the following equation describing the blood flow gradient in the x direction
∂<A><AC>Q</AC><AC>˙</AC></A><IT>/∂x=a+dy+fz+gyz</IT> (6)
For y = z = 0 and at any x, the vertical gradient is equal to the coefficient a. At y = 0, the vertical gradient is equal to
∂<A><AC>Q</AC><AC>˙</AC></A><IT>/∂x=a+fz</IT> (7)
That is, it varies linearly with z, and the coefficient f is the slope of the vertical gradient in the z direction. Thus both blood flow and blood flow gradient in any coordinate axis can be calculated for any arbitrary position within the lung. Inclusion of the xy, xz, and yz terms avoids analyzing separate lung regions to obtain regional spatial gradients.

The regression analysis provides a P value, a measure of the reliability, for each constant in best-fit linear equation. The coefficient of determination (R2) represented the degree of variability due to spatial variation.

VA/Q heterogeneity. We evaluated the heterogeneity of VA, Q and VA/Q distributions using the coefficient of variation (sigma ), the standard deviation of regional VA and Q values divided by mean values. Variance (sigma 2) of VA/Q was computed from the variances of VA and Q with Pearson's correlation coefficient (rho c) between VA and Q (50)
&sfgr;<SUP>2</SUP><A><AC>V</AC><AC>˙</AC></A><SC>a</SC><IT>/</IT><A><AC>Q</AC><AC>˙</AC></A><IT>=&sfgr;</IT><SUP>2</SUP><A><AC>V</AC><AC>˙</AC></A><SC>a</SC><IT>+&sfgr;</IT><SUP>2</SUP><A><AC>Q</AC><AC>˙</AC></A><IT>−</IT>2<IT>&rgr;&sfgr;</IT><A><AC>V</AC><AC>˙</AC></A><SC>a</SC><IT>·&sfgr;</IT><A><AC>Q</AC><AC>˙</AC></A> (8)
Heterogeneity in VA/Q measured by the log-normal standard deviations (lnSDVA and lnSDQ) of VA- and Q-VA/Q distribution curves was obtained from MIGET and the fluorescent microspheres (FMS) data.

To separate the heterogeneity in VA, Q, and VA/Q due to spatial variations from that due to other factors (residual variation), the variance (mean summed squares) of the mean summed square of the residuals, (measured - predicted values)/mean value (50), was obtained from the multiple linear regression analysis. This analysis was repeated using a fourth-order regression equation (31 terms) with terms up to third order in each coordinate and up to fourth order in each term (xpyqzr, p + q + r <=  4). Preliminary analysis indicated no further decrease in the residual variance (or increase in R2) with a higher order regression equation. The residual variance/total variance equals 1 - R2, where R2 is the adjusted coefficient of determination from the regression analysis.


    RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES

Physiological Data

Table 1 summarizes the physiological data. Body position and PEEP had no effect on QT, heart rate, temperature, respiration rate, pH, mixed venous PO2, and hemoglobin. PEEP decreased Psa and increased Ppa, Ppcw, and peak Paw in both LLD and RLD postures. Arterial PO2 (PaO2) was greater in the LLD than RLD posture, whereas A-aDO2 was less in the LLD than RLD posture. The addition of PEEP increased PaCO2 by ~2 Torr in both postures. PEEP increased FRC by 65 and 76% in the LLD and RLD postures, respectively.

QS/QT and VD/VT showed no change between the LLD (0.15 ± 0.3 and 43.5 ± 4.8%) and RLD (0.18 ± 0.24 and 43.6 ± 4.6%) postures. PEEP increased VD/VT in both the LLD (49.1 ± 8.5%) and RLD (48.9 ± 6.7%) postures and decreased QS/QT in the LLD (0.03 ± 0.05%) but not the RLD (0.15 ± 0.24%) posture. QS/QT was similar to that measured by the FMS data. VD/VT was greater than that measured by Fowler's method (Table 1).

Microsphere data. In total, 1,378-1,654 lung pieces per animal were processed for Q and VA. We discarded lung pieces (135 ± 49) with >25% pulmonary airways and with fluorescent intensity (11 ± 10) outside the range of ±4 SD of any of the mean values. Analysis of blood flow and ventilation was carried out on 90.4 ± 5% of the total lung pieces. For the analysis of VA/Q and PrO2, we accepted data with the range of mean ± 3 SD of ln(VA/Q). This eliminated the pieces associated with dead space (very large VA/Q) and with shunt (very low VA/Q). The number of pieces eliminated averaged 3 ± 4% of the total.

Spatial Gradients in Q, VA, VA/Q, and PrO2

Multiple regression analyses revealed systematic variations of blood flow in the three coordinates. Because the blood flow distribution described by the multiple linear regression equation (Eq. 5) for each animal showed that most (6 of 7) coefficients were significant, we used the equation to describe Q, VA, VA/Q, and PrO2 for all conditions. The use of the equation was further justified because R2 for the best-fit regression was lower when only terms of one or two coordinates were included. The coefficients of the six animals were pooled for each condition, and the significance was tested to determine its validity in describing the blood flow distribution. In many instances, coefficients that were significant in a single animal proved to be not significant among the six animals. Coefficients were considered meaningful only if the coefficients of the six animals were significant. The present study showed values of R2 of ~0.40, indicating that ~40% of the variability in blood flow was attributed to spatial variations.

Significant vertical, dorsal-ventral and caudal-cranial spatial gradients of Q, VA, VA/Q, and PrO2 are summarized in Table 2. The complete set of pooled coefficients are given in Tables A2 and A3 (APPENDIX). Figure 1 shows Q plotted vs. vertical height (x coordinate) up the lung in the LLD and RLD postures with and without PEEP for a representative animal. The lines represent Q vs. height (x) at the center of mass (y = z = 0), determined from the regression analysis. Figures 2, 3, and 4 are equivalent data for VA, VA/Q, and PrO2.

                              
View this table:
[in this window]
[in a new window]
 
Table 2.   Significant* spatial gradients and intercepts of regression equationdagger



View larger version (45K):
[in this window]
[in a new window]
 
Fig. 1.   Blood flow (Q) per unit regional lung volume (ml · min-1 · ml-1) vs. lung height for a representative dog in the left lateral decubitus (LLD) posture without positive end-expiratory pressure (PEEP; A), LLD with 10 cmH2O PEEP (B), right lateral decubitus (RLD) posture without PEEP (C), and RLD with 10 cmH2O PEEP (D). R, right lung (open circle ); L, left lung (diamond ). Lines represent best-fit values from multiple linear regression analysis. R2 indicated that ~40% of the variability in blood flow was spatially determined. WL, whole lung; RL, right lung; LL, left lung. Independent and dependent axes have been interchanged for presentation.



View larger version (50K):
[in this window]
[in a new window]
 
Fig. 2.   Ventilation (VA) per unit regional lung volume (ml · min-1 · ml-1) vs. lung height for representative animal in LLD without PEEP (A), LLD with 10 cmH2O PEEP (B), RLD without PEEP (C), and RLD with 10 cmH2O PEEP (D). Lines represent best-fit values from multiple linear regression analysis.



View larger version (49K):
[in this window]
[in a new window]
 
Fig. 3.   VA-to-Q ratio vs. lung height for representative animal in LLD without PEEP (A), LLD with 10 cmH2O cm PEEP (B), RLD without PEEP (C), and RLD with 10 cmH2O PEEP (D). Lines represent best-fit values from multiple linear regression analysis.



View larger version (48K):
[in this window]
[in a new window]
 
Fig. 4.   Regional PO2 (PrO2) vs. lung height for representative animal in LLD without PEEP (A), LLD with 10 cmH2O cm PEEP (B), RLD without PEEP (C), and RLD with 10 cmH2O PEEP (D). Lines represent best-fit values from multiple linear regression analysis. Note that the low PCO2 values in the dependent lung in the LLD posture (A) was eliminated with the addition of PEEP (B).

Effect of Posture Without PEEP

Regional distribution of Q. As indicated by the x coefficient (a) for the whole lung, there was a significant negative (gravity-dependent) vertical gradient (Table 2, -0.27 and -0.42 ml · min-1 · ml-1 · cm-1) in Q that was greater in the RLD posture (-0.42) than in the LLD posture (-0.27), implying that blood flow in the dependent lung was less in the LLD than in the RLD posture.

A smaller blood flow in the dependent lung in the LLD than RLD posture was verified by the fact that total blood flow (% total) was less in the dependent left lung (37%) than in the nondependent lung (63%) in the LLD posture but was greater in the dependent right lung (64%) than in the nondependent lung (36%) in the RLD posture (Table 3). Given the same QT in both postures (Table 1), the fraction of the QT adjusted for tissue mass to either lung did not change with body position.

                              
View this table:
[in this window]
[in a new window]
 
Table 3.   Percent cardiac output and ventilation to left and right lung from microsphere data

In both left and right lung, the vertical gradients (-0.60 and -0.37) observed in the LLD posture were eliminated with body inversion to the RLD posture. These gradients represented a 70-150% change in the mean blood flow over the height of the lung (~15 cm) or 5-10% cm-1.

In the left lung, there was a positive dorsal-ventral gradient (0.26 and 0.18) in the LLD and RLD posture, with the ventral regions having the greater blood flow. The dorsal-ventral gradient in the left lung was accompanied with a negative caudal-cranial gradient in the LLD posture (-0.18), with blood flow greatest in the caudal regions.

Regional distribution of VA. For the whole lung, the largest vertical gradient in VA was observed in the RLD posture (Table 2, -0.58) and occurred in conjunction with a positive dorsal-ventral gradient (0.31) that was eliminated with inversion to the LLD posture. In the LLD posture, the only substantial gradient occurred in the dorsal-ventral direction (0.27).

Similar to the behavior in QT, total ventilation was smaller in the dependent lung than in the nondependent lung in the LLD posture but larger in the dependent lung in the RLD posture (Table 3). For constant ventilation in both postures, body position had no effect on ventilation in either lung. In the left lung in the RLD posture, significant vertical (-0.20) and dorsal-ventral (0.27) gradients were observed.

Relationship between regional and total blood flow and ventilation for each lung. The vertical gradients in Q and VA (Table 2) measured in this study might appear at first sight to be at odds with the total blood flow and ventilation values measured for each lung (Table 3). For the whole lung, the vertical gradient in Q in the LLD posture would suggest a greater blood flow in the dependent lung than in the nondependent lung (Table 2). On the other hand, the total blood flow to the dependent left lung in the LLD posture was clearly lower than that to the nondependent lung. This apparent discrepancy is due to the fact that the total blood flow to the lung is the product of the mean regional Q and the total lung volume. This relationship allowed the estimate of FRC to each lung. The smallest FRC was predicted to occur in the left lung in the LLD posture.

In the LLD posture, mean Q (Table A2) averaged 5.8 and 4.6 ml · min-1 · ml-1 in the left and right lung pieces, respectively. QT (3.6 l/min) was distributed 1.3 l/min (37%) to the left lung and 2.3 l/min (67%) to the right lung (Table 3). Thus the estimated FRC was 220 ml in the left lung and 500 ml in the right lung, resulting in a total FRC of 720 ml. This value was close to the measured value (758 ml) equal to the air volume (645 ml, Table 1) and wet tissue volume (113 ml) based on lung wet weight (4.7 × 24 g dry weight). The predicted left lung FRC was 31% total FRC. The expected FRC of a uniformly inflated left lung based on tissue mass was 43% total FRC. This value was reduced by ~25% via the vertical Ptp gradient, resulting in a left lung FRC of 32%, near the predicted value of 31%. Accordingly, the reduced blood flow to the dependent left lung in the LLD posture was consistent with a reduced FRC caused by the vertical Ptp gradient. Evidently vertical gradients in Q require knowing regional lung volume to accurately predict relative blood flow to each lung. A similar argument applies to VA measurements.

Regional distribution of PrO2 and VA/Q. The largest gradient in PrO2 (1.3, Table 2) was observed in the vertical direction for the whole lung in the LLD posture and was positive, indicating that PrO2 was smaller in the dependent (left) lung than in the nondependent (right) lung (intercepts, Table A3). This vertical gradient was abolished with inversion to the RLD posture and was accompanied with positive dorsal-ventral and caudal-cranial gradients.

To determine whether PrO2 observed in the dependent lung in the LLD posture was low enough to trigger a hypoxic vasoconstriction response, the minimum PrO2 value was obtained from regression analysis for the six animals studied. Substituting into Eq. 5, the linear equation with mean intercept and coefficients for the whole lung was as follows (Table A3)
Pr<SUB>O<SUB>2</SUB></SUB><IT>=</IT>113<IT>+</IT>1.3<IT>x+</IT>1.15<IT>y+</IT>0.5<IT>z−</IT>0.14<IT>xy−</IT>0.16<IT>yz−</IT>0.06<IT>zx+</IT>0.03<IT>xyz</IT>
The minimum value of PrO2 (81 ± 21 Torr) occurred in the dependent (x = -4 cm), dorsal (y = -5 cm), and caudal (z = -8 cm) regions of the lung (Fig. 6). The maximum value of PrO2 (124 ± 5 Torr) was located in the nondependent (x = +4 cm), ventral (y = +5 cm), and cranial (z = +8 cm) region of the lung. In the RLD posture, minimum and maximum PrO2 values evaluated at similar (x, y, z) values used for the LLD posture were 96 ± 10 and 117 ± 8 Torr, respectively (Fig. 6). Note that PrO2 was reduced below 100 Torr only in the dependent lung in the LLD posture.

The low PrO2 originating from the dependent caudal-dorsal regions of the dependent left lung in the LLD posture without PEEP was increased by inversion to the RLD posture (Fig. 4). That body inversion from the LLD to the RLD posture increased PrO2 in the caudal regions was consistent with the reduction or elimination of significant positive vertical, dorsal-ventral, and caudal-cranial PrO2 gradients for the whole lung (Table 2).

In the left and right lung, small but significant gradients in PrO2 and VA/Q were observed in all three coordinates (Table 2). However, the detection of a significant VA/Q (PrO2) gradient was associated with a significant PrO2 (VA/Q) gradient only in the left dependent lung in the LLD posture. In the left (dependent) lung in the LLD posture, a dorsal-ventral VA/Q gradient (0.053) was associated with significant PrO2 gradients (2.8).

Comparison Between Predicted and Measured Gas Exchange

Predicted PaO2 and PaCO2 calculated from regional VA/Q data did not differ from measured PaO2 and PaCO2 in both the LLD and RLD posture (Table 1), and the measured A-aDO2 was well predicted from the microsphere data in the LLD posture. However, the predicted values of A-aDO2 were significantly (P < 0.05) less in the LLD than in the RLD posture (Table 1).

Mean VA/Q and PrO2 Changes of Nondependent and Dependent Lung

In the LLD posture (intercepts, Table 2), VA/Q was greater in the nondependent right lung (1.42 ± 0.45) than in the dependent lung (0.93 ± 0.37). In the right lung, PrO2 increased with body inversion from the RLD (113 ± 7 Torr) to the LLD (119 ± 4 Torr) posture. This was associated with an increase in VA/Q from 1.09 ± 0.39 in the RLD posture to 1.40 ± 0.60 in the LLD posture. This behavior was accompanied with a dorsal-ventral VA/Q gradient that was significant (0.035) only in the RLD posture.

Regional Variations in the Spatial Gradients

The significant coefficients d, e, and f, shown in Tables A2 and A3 indicated that the spatial gradients varied along an orthogonal axis. These are discussed in the APPENDIX.

Effect of PEEP

Regional distribution of Q. In general, PEEP either reduced or eliminated the spatial gradients in Q and VA that occurred without PEEP (Table A2). The decrease in the vertical gradient with PEEP was associated with a decrease (30-50%) in the mean blood flow (intercepts, Table A2). With PEEP, the dependent right lung in the RLD posture had the greater blood flow, similar to the behavior without PEEP (Table 3). By contrast, in the LLD posture PEEP eliminated the left-right lung difference in blood flow measured without PEEP.

PEEP increased the low PrO2 originating from the dependent caudal-dorsal regions of the dependent left lung in the LLD posture (intercepts, Table A3). The latter effect of PEEP was consistent with the PEEP-induced reduction of the caudal-cranial VA/Q gradient (from 0.053 to 0.03, Table A3) and 50% reduction of the caudal-cranial PrO2 gradient (from 0.85 to 0.40) in the left lung.

Similar to the data without PEEP, with PEEP PaO2 and PaCO2 calculated from VA/Q data did not differ from measured PaO2 and PaCO2 in both the LLD and RLD posture, and the measured A-aDO2 was well predicted from the microsphere data (Table 1).

PEEP reduced the relatively high VA/Q of the nondependent right lung to a value closer to 1 (Table A3, 1.22 ± 0.24). In the right lung with PEEP, VA/Q increased with body inversion from the RLD posture (1.13 ± 0.24) to the LLD posture (1.22 ± 0.24).

In the left lung in the LLD posture, PEEP increased PrO2 from 105 ± 12 to 112 ± 7 Torr (Table A3). This behavior was associated with a PEEP-induced vertical PrO2 gradient (0.83) in conjunction with reduced dorsal-ventral and caudal-cranial PrO2 gradients.

Regional Perfusion Correlation Between Postures With and Without PEEP

Figure 5 shows Q to each lung piece in the LLD posture plotted against blood flow of the same piece in the RLD posture of one representative animal without PEEP (Fig. 5A) and with PEEP (Fig. 5B). Except for the left lung without PEEP (R = 0.66), blood flow to each piece was poorly correlated (R = 0.044-0.22) between LLD and RLD posture. In other words, lung pieces with high (or low) blood flow in the LLD posture received low (or high) blood flow in the RLD posture. This behavior was consistent with the gravity-dependent vertical gradients observed for both right and left lungs and for the whole lung in both postures, opposite to that found between the supine and prone position (19).


View larger version (30K):
[in this window]
[in a new window]
 
Fig. 5.   Correlation between pulmonary blood flow in the LLD and RLD measured in 1 representative animal without PEEP (A) and with PEEP (B). Dotted line is line of identity. Note that lung pieces with high (low) blood flow in the LLD posture received low (high) blood flow in the RLD posture, consistent with a gravity-dependent vertical gradient measured by multiple linear regression analysis (Fig. 1B).

Q and VA heterogeneity. The mean coefficients of variation of Q and VA (30-60%) were similar to reported values (19, 21, 29, 33, 40). Table 4 summarizes the heterogeneity in VA and Q evaluated by three methods: total and residual variances of VA, Q, and VA/Q; widths (lnSDVA and lnSDQ) of the VA-VA/Q and Q-VA/Q distribution curves measured by FMS and MIGET. The variance of the data based on regional values was similar to that based on uncorrected data. The coefficients of correlation (rho c) between VA and Q (Pearson's method) were similar to those computed by use of sigma VA, sigma Q, and sigma 2VA/Q in Eq. 8 (33). The total variance data showed that in the LLD posture PEEP reduced the variance in VA and VA/Q but not in Q, consistent with a more uniformly inflated lung at the higher lung volume. This PEEP-induced change in variance was undetected by the other two methods. Neither body position nor PEEP affected the total variance in Q. Neither PEEP nor body position changed the correlation between VA and Q. There was a tendency for heterogeneity measured by lnSDVA and lnSDQ to be greater with MIGET than with FMS, but the difference was only significant for Q in the LLD posture without PEEP. The broader distribution measured with MIGET than with topographical data has been noted in previous studies (15, 45). The difference has been attributed to the coarse scale of the VA/Q distribution inherent in MIGET (4, 5).

                              
View this table:
[in this window]
[in a new window]
 
Table 4.   Effect of position and PEEP on heterogeneity of VA and Q distribution: FMS and MIGET

Heterogeneity in VA, Q, and VA/Q that was attributed to residual variation as measured by the residuals of the regression analysis was reduced from 65 ± 16% of the total variance in the linear regression analysis to 41 ± 6% of the total variance in the fourth-order regression analysis (Table 4). These values are in line with the coefficients of determination (R2) of ~40 and 60% that implicated ~60 and 40% of the variability to residual variation. Similar to the total variance, PEEP reduced the residual variance of VA in the RLD posture. Neither body position nor PEEP affected the residual variance in Q. There was a tendency for PEEP to increase the residual-to-total variance fraction in VA and VA/Q, but this was only significant in the LLD posture. The coefficient of correlation (rho c) between VA and Q calculated by using the residual variances (0.59-0.75, Table 4) in Eq. 8 was similar to the correlation coefficient between VA and Q (0.57-0.71).


    DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES

In this study we used a high-resolution technique to describe the distribution of Q, VA, VA/Q, and PrO2 in the lateral decubitus posture. The high resolution in conjunction with multiple linear regression analysis allowed the spatial description of these variables at any arbitrary position relative to the three rectangular coordinate axes.

The major findings of this study are as follows. First, the gravity-dependent vertical gradient in Q was greater in the RLD than LLD posture (Fig. 1, A and B; Table 2). This was attributed to a reduced blood flow in the dependent left lung in the LLD posture (Table 3). Second, PEEP reduced the vertical gradient in both postures and eliminated the difference between postures (Fig. 1). PEEP or body inversion from the LLD to the RLD posture abolished the positive vertical, dorsal-ventral, and caudal-cranial gradients in PrO2 observed in the LLD posture (Table A3). Third, a positive vertical gradient in PrO2 was observed in the LLD posture with the dependent dorsal-caudal regions of the dependent lung having values below that (100 Torr) needed to invoke a hypoxic vasoconstriction response (31). This behavior was consistent with the reduced blood flow to the dependent lung in the LLD posture.

Methodological Issues

Fluorescent microsphere technique. The microsphere technique as implemented in this study has been validated in previous studies (17, 40). Regional deposition of aerosolized and injected microspheres allowed simultaneous measurements of ventilation and perfusion distribution that predicted regional gas exchange with high spatial resolution.

Volume adjustment to FRC and vertical Ptp gradient. Injected and aerosolized fluorescent microspheres were delivered in vivo near FRC, whereas the fluorescent signals were measured in vitro in the dried lung inflated to TLC. Accordingly, we made several adjustments to the weight of each piece to extrapolate to piece volume in vivo and to determine Q and VA per unit regional lung volume at FRC.

First, the lung volume of each piece was adjusted from TLC to FRC by reducing the cube lengths in the three dimensions. This adjustment resulted in an anisotropically inflated lung (20) and a homogeneous (constant) deformation along each axis. Second, we imposed a distortion to the vertical dimension (x) of each lung piece to produce a vertical Ptp gradient as previously measured (1). This distortion in the x dimension at each height was based on the PV curve of an isolated lung (27) and was applied to all y and z dimensions at the same height. This preserved the homogeneous deformation in the y and z coordinates and the anisotropy in regional volume at FRC, in effect producing changes in regional volume identical to those given by the PV curve (Eq. 2). These adjustments for the vertical Ptp gradient increased the (maximum) dependent (D)-to-nondependent (N) ratio (D/N) for Q by a factor of ~3, N/D in regional lung volume, and D/N in regional lung density (16). Implicit in this adjustment for lung density is the scaling of tissue mass to capillary density. Accordingly, regional lung density changes caused by the vertical Ptp gradient were the dominant contributor to the vertical gradient in Q and VA.

Prior studies have reported the vertical gradients of perfusion relative to the number of alveoli or piece weight at TLC. This paper presents perfusion gradients relative to the regional lung volume at the time of microsphere injections. The adjustment for the vertical changes in regional lung density produced vertical gradients in regional Q that were greater than those estimated in previous studies using TLC-measured regional volume (18, 19, 21, 22, 33, 34). These normalization issues need further evaluation, particularly in the supine and upright body positions under both normal and increased acceleration loads with relatively large Ptp gradients (1, 2).

We made no adjustment for Ptp gradients in the other two axes (y and z), in the absence of reported data. Blood volume was ignored in the calculation of lung density and regional lung volume because the dry weight used in the calculation of mean lung density was blood free.

Distribution of Regional Perfusion

Effect of gravity. The effects of gravity on the vertical gradient in blood flow in the lung have been described in terms of the relation among the Ppa and pulmonary venous (Ppv) and alveolar (Palv) pressures (48, 49). This theory predicts a decreasing blood flow up the height of the lung (in our nomenclature, a negative vertical gradient). Most of the vertical gradients measured in the present study are explainable, at least qualitatively, with the gravitational model. The vertical gradients in blood flow measured in the LLD and RLD posture with and without PEEP were expected and confirmed previous findings (6, 20, 25, 28, 37).

Effect of lung volume and vascular resistance. The fact that the gravity-dependent vertical gradients in Q decreased with a change from RLD to LLD posture and with PEEP indicates that factors other than gravity contributed to the blood flow distribution. A major factor was the lung volume-induced vascular resistance that changed with body position and PEEP. Pulmonary vascular resistance depends on lung volume (24); its changes with lung volume are different for zone 2 (Ppa > Palv > Ppv) and zone 3 (Ppa > Ppv > Palv) conditions (9). By this theory (24), Q is proportional to Ppa - Palv in zone 2 and Ppa - Ppv in zone 3. The increased flow down the lung is due to Ppa increasing down the lung in zone 2 and to Ppv-induced capillary recruitment or vascular distention in zone 3.

In the latter study (9), pulmonary blood flow and vascular resistance were measured vs. lung volume (%TLC) at constant values of Ppa - Palv and Ppv - Palv under both zone 2 and zone 3 conditions in isolated rabbit lungs. In zone 3, for Ppa-Palv of 18 mmHg, blood flow increased linearly with a decrease in lung volume from TLC to residual volume. In zone 2, blood flow increased as lung volume decreased from TLC to 50% TLC but decreased from 60% TLC to residual volume, a behavior consistent with the U-shape curve describing the relationship between vascular resistance and lung volume (44). Whatever the lung volume-induced differences in blood flow, blood flow was greater in zone 3 than in zone 2.

In the present study, without PEEP, Ppv averaged 8.4 cmH2O in the LLD posture and 6.1 cmH2O in the RLD posture (Table 1, Ppcw) relative to mid-heart level. With a mean Paw of 5 cmH2O (Table 1), the dependent lung was in zone 3 in both postures whereas the nondependent lung was predominantly in zone 2, more so in the RLD than in the LLD posture. PEEP increased Ppv to 10 cmH2O and mean Paw to 15 cmH2O in both postures, placing the nondependent lung and half of the dependent lung in zone 2. Pulmonary vascular resistance, (Ppa - Ppcw)/blood flow, was greater in the dependent than nondependent lung in both postures and increased with PEEP (Table 1).

The gravity-dependent vertical gradients in blood flow measured in both postures for the whole lung without PEEP (Table 2, Fig. 1, A and C) were consistent with the shift from zone 2 to zone 3 conditions down the lung, but these gradients were accentuated by the threefold increase in lung density down the lung. The removal of the density gradient with PEEP reduced the gradients to 44 and 33% in the LLD and RLD posture, respectively (Table 2, Fig. 1, B and D). Thus the two- to threefold increase in the gradient with the removal of PEEP was attributed almost entirely to the lung density gradient. A similar behavior was observed for both left and right lungs in the LLD posture (Fig. 1, A and B).

Blood flow was lower in the dependent left lung than in the nondependent lung (Table 3), consistent with the results of a previous study (28). This behavior was opposite to that predicted by gravity, indicating that factors other than gravity contributed to blood flow distribution in the LLD posture. One factor has been associated with the vascular structure (17, 28). The reduced blood flow to the dependent left lung was not caused by a heart weight induced lower lung volume per se because a reduced lung volume is associated with a lower vascular resistance in zone 3 (9). It is possible that a nonuniform lung distortion due to heart and abdominal weight might conceivably cause an increased vascular resistance. Another mechanism such as hypoxic vasoconstriction remains an alternative explanation in vivo, particularly in view of the positive gradient in PrO2 measured (see below).

Nongravitational gradients in Q. An important finding of this study relates to other nongravitational gradients in Q. Specifically, in the isogravitational (y-z) plane (x = 0), a positive dorsal-ventral gradient in Q occurred in the left dependent lung in the LLD posture, with blood flow maximal in the ventral regions (Table 2, 0.26). The smaller blood flow in the dorsal regions was opposite to that observed in the supine dog (10) in which the dorsal lung regions had the larger blood flow. Thus the present measurements would not support an intrinsic greater vascular conductance postulated for the dorsal lung regions (10). Thus extrinsic factors such as hypoxic vasoconstriction and lung distortion caused by the weight of the heart and abdomen might be crucial.

The positive dorsal-ventral gradient in Q observed in the left lung in the LLD posture occurred in conjunction with a negative caudal-cranial gradient in Q with Q increasing in the caudal region, in the absence of any gradient in the right lung. Both these gradients were abolished with PEEP, indicating a lung volume-induced relative shift of blood flow from the ventral-caudal to the dorsal-cranial regions.

The decrease in blood flow in the caudal-cranial direction for the whole lung in the LLD posture is consistent with the results of Greenleaf et al. (20) in the mechanically ventilated anesthetized dog. This contrasts to the absence of a caudal-cranial gradient in spontaneously breathing humans in the lateral decubitus posture (6).

Distribution of Ventilation

Effect of posture without PEEP. In contrast to the absence of a vertical VA gradient in the LLD posture (Table 2), total ventilation was greater in the nondependent than dependent lung in the LLD posture (Table 3). This difference was similar to the behavior in blood flow and was most likely due to a smaller FRC in the dependent left lung. Body inversion from LLD to RLD posture produced a substantial negative gradient in VA (-0.58) that was consistent with the greater total ventilation measured in the right dependent lung than in the nondependent lung. This behavior in the anesthetized dog is consistent with results from the anesthetized human in the lateral decubitus posture (38, 39). These results with anesthesia differed substantially from those in awake humans, showing greater ventilation in the dependent than nondependent lung in both the LLD and RLD posture (7, 25, 32). The latter behavior was explained by the vertical Ptp gradient causing a lower lung volume and greater lung compliance in the dependent lung regions. The absence of a caudal-cranial gradient in VA in the lateral decubitus posture in the anesthetized dog was consistent with the results in awake humans measured by using radioactive gas inhalation and external scintillation counters (7, 25). This behavior was attributed to a uniform Ptp in the horizontal direction.

The differences in ventilation measured between the anesthetized and awake state might be related to the anesthesia-induced reduction in FRC observed in both lungs (38, 39). The following factors might be involved. First, the nondependent lung would move from the upper low-complia