Pulsatile pulmonary microvascular pressure measured with vascular occlusion techniques

The periodic variations of the pulmonary microvascular pressure during pulsatile perfusion were studied in isolated left lower lobes of canine lungs by the arterial occlusion (AO) and double occlusion (DO) techniques. Sixteen AO and eight DO maneuvers evenly distributed within the pump cycle were performed for each of four frequencies: 36, 54, 72, and 90 beats/min. Nearly identical microvascular pressure contours were reconstructed from the AO and DO maneuvers by relocating the measured occlusion pressures in time. These contours lagged behind the pulmonary arterial pressure waveform. Their amplitude decreased from 25 to 14% of the arterial pulse pressure as the pump frequency was increased from 36 to 90 beats/min. The modulus of the pressure transfer function at the site of arterial occlusion decreased as the frequency increased. The phase was negative for all frequencies and it approached -90 degrees for the higher frequencies. Vasoconstriction induced by serotonin resulted in an increase of the magnitude of the AO pressure contour that was nearly proportional to the increase of the pulmonary arterial pulse pressure. In contrast, elevation of the lobar venous pressure to 10 mmHg increased the amplitude of the AO pressure contour, whereas it slightly decreased the pulmonary arterial pulse pressure. These experiments demonstrate that the AO and DO pressures fluctuate markedly during pulsatile perfusion. Their oscillations would be indicative of the pulsatility in the pulmonary microvascular bed.     

Theoretical analysis of occlusion techniques for measuring pulmonary capillary pressure.

We have developed a model including three serial compliant compartments (arterial, capillary, and venous) separated by two resistances (arterial and venous) for interpreting in vivo single pulmonary arterial or venous occlusion pressure profiles and double occlusion. We formalized and solved the corresponding system of equations. We showed that in this model 1) pulmonary capillary pressure (Pc) profile after arterial or venous occlusion has an S shape, 2) the estimation of Pc by zero time extrapolation of the slow component of the arterial occlusion profile (Pcao) always overestimates Pc, 3) symmetrically such an estimation on the venous occlusion profile (Pcvo) always underestimates Pc, 4) double occlusion pressure (Pcdo) differs from Pc. We evaluated the impact of varying parameter values in the model with parameter sets drawn either from the literature or from arbitrary arterial and venous pressures, being respectively 20 and 5 mmHg. Resulting Pcao-Pc differences ranged from 0.4 to 5.4 mmHg and resulting Pcvo-Pc differences ranged from -0.3 to -5.0 mmHg. Pcdo-Pc was positive or negative, its absolute value in general being negligible (< 1.1 mmHg).     

Phasic capillary pressure determined by arterial occlusion in intact dog lung lobes

In six open-chest dogs, electrocardiogram- (ECG) controlled pulmonary arterial occlusion was performed during the control period and during the infusions of serotonin and histamine. A temporal series of instantaneous pulmonary capillary pressure and the longitudinal distributions of vascular resistance and compliance were evaluated in the intact left lower lung lobe. In the control period, we found a significant phasic variation of pulmonary capillary pressure (Pc) with the cardiac cycle. The ratio of arterial to venous resistances (Ra/Rv) was 6:4, and the ratio of arterial to capillary compliances (Ca/Cc) was 1:11. During the infusions of serotonin and histamine, Pc showed similar phasic variations, despite significant hemodynamic changes induced by these agents. Serotonin predominantly increased Ra, whereas histamine predominantly increased Rv. The ratio of Rv to the total resistance decreased significantly from 0.42 to 0.32 during the infusion of serotonin and increased significantly to 0.62 during the infusion of histamine. The data suggest that phasic Pc determined by ECG-controlled arterial occlusion reflects the pulsatility in the pulmonary microvascular bed under control conditions and after alterations of the pulmonary vascular resistance by serotonin and histamine.     

Venous occlusion measurement of pulmonary capillary pressure: effects of embolization

The effects of pulmonary arterial embolization on calculated pulmonary capillary pressure as determined by the venous occlusion technique are examined using a simple pressure-flow model for the lung. It is predicted that pulmonary, arterial embolization can induce significant underestimation of pulmonary capillary pressure in flowing vessels. This underestimation is related to the percent of vessels embolized and the caliber of pulmonary arteries that are embolized (i.e., the size of the emboli). Experimental verification of these theoretical findings is necessary before the conclusions can be extended to the interpretation of venous occlusion experiments in the lung.     

Effects of arachidonate on permeability and resistance distribution in canine lungs

In this study, 14 canine lung lobes were isolated and perfused with autologous blood at constant pressure (CP) or constant flow (CF). Pulmonary capillary pressure (Pc) was measured via venous occlusion or simultaneous arterial and venous occlusions. Arterial and venous pressures and blood flow were measured concurrently so that total pulmonary vascular resistance (RT) as well as pre- (Ra) and post- (Rv) capillary resistances could be calculated. In both CP and CF perfused lobes, 5-min arachidonic acid (AA) infusions (0.085 +/- 0.005 to 2.80 +/- 0.16 mg X min-1 X 100 g lung-1) increased RT, Rv, and Pc (P less than 0.05 at the highest dose), while Ra was not significantly altered and Ra/Rv fell (P less than 0.05 at the highest AA dose). In five CP-perfused lobes, the effect of AA infusion on the pulmonary capillary filtration coefficient (Kf,C) was also determined. Neither low-dose AA (0.167 +/- 0.033 mg X min-1 X 100 g-1) nor high-dose AA (1.35 +/- 0.39 mg X min-1 X 100 g-1) altered Kf,C from control values (0.19 +/- 0.02 ml X min-1 X cmH2O-1 X 100 g-1). The hemodynamic response to AA was attenuated by prior administration of indomethacin (n = 2). We conclude that AA infusion in blood-perfused canine lung lobes increased RT and Pc by increasing Rv and that microvascular permeability is unaltered by AA infusion.     

Comparison of isogravimetric and venous occlusion capillary pressures in isolated dog lungs

Venous occlusion capillary pressures (Pcv) were simultaneously compared with isogravimetric capillary pressures (PcI) in the same isolated perfused dog lung preparations. For 26 determinations, PcI averaged 1.23 +/- 0.22 (SE) mmHg higher than Pcv. However, the two measurements of capillary pressure were highly correlated (r = 0.99), and the following regression equation was obtained: Pcv = 1.12 PcI - 2.1. Pcv could be easily measured several times in the same preparation, either by total venous occlusion or regional venous occlusion using a Swan-Ganz balloon catheter. In addition, Pcv did not require an isogravimetric state for its determination. These data suggest that the major sites of filtration and vascular capacitance in the pulmonary circulation reside in the microvessels and that the more easily determined Pcv is an adequate measure of the average capillary filtration pressure in the lungs.     

Validation of double vascular occlusion method for Pc,i in lung and skeletal muscle

Capillary pressures in isogravimetric lung and skeletal muscle measured with the double vascular occlusion technique (Pdo) were compared to those measured using the traditional gravimetric technique (Pc,i). Pressures were measured using both techniques in isolated blood-perfused canine lungs (n = 18), blood-perfused rat hindquarters before (n = 8) and after (n = 6) maximal dilatation with papaverine and in rat hindquarters perfused with an artificial plasma (n = 6). In both organs, regardless of vascular tone, the double vascular occlusion isogravimetric pressure was the same as the gravimetric Pc,i, and the two measurements were highly correlated. Lung: Pdo = -0.22 + 1.06 Pc,i (r = 0.85, P less than 0.01); hindquarter: Pdo = -1.03 + 0.99 Pc,i (r = 0.91, P less than 0.01). In addition, Pdo was the same at every combination of isogravimetric arterial and venous pressures tested. The results indicate that the more rapidly applied double vascular occlusion pressure yields an accurate measure of isogravimetric capillary pressure in isolated organs over a wide range of isogravimetric pressures.     

Determination of pulmonary microvascular reflection coefficient in sheep by venous occlusion

We devised a technique that permitted elevation of pulmonary pressures in unanesthetized sheep by occluding their pulmonary veins. Using this technique, we raised pulmonary capillary pressure from a baseline of 13.2 +/- 2.2 to 35.3 +/- 5.1 mmHg. This increased lung lymph flow (from 8.8 +/- 2.7 to 53.1 +/- 13.9 ml/h). We estimated the pulmonary microvascular oncotic reflection coefficient and found it to be 0.82 +/- 0.05 (SD). The filtration coefficient was 0.019 +/- 0.005 ml.mmHg-1.min-1. During the period of increased pressure, the animals had stable arterial pressures and cardiac outputs. None of the animals developed blood coagulation problems. These data illustrate the usefulness of pulmonary venous occlusion to elevate pulmonary microvascular pressure to obtain plasma-to-lymph protein concentration ratios independent of flow, allowing for the calculation of the oncotic reflection coefficient.     

Pulmonary blood flow distribution measured by radionuclide-computed tomography

Distributions of pulmonary blood flow per unit lung volume were measured with subjects in the prone, supine, and sitting positions by means of radionuclide-computed tomography of intravenously administered 99mTc-labeled macroaggregates of human serum albumin. The blood flow was greater in the direction of gravity in all 31 subjects except one with severe mitral valve stenosis. With the subject in a sitting position, four different types of distribution were distinguished. One type had a three-zonal blood flow distribution as previously reported by West and co-workers (J. Appl. Physiol. 19: 713-724, 1964). Pulmonary arterial pressure and venous pressure estimated from this model showed reasonable agreement with pulmonary arterial pressure and capillary wedge pressure measured by Swan-Ganz catheter in 17 supine patients and in 2 sitting patients. The method makes possible noninvasive assessment of pulmonary vascular pressures.     

Occlusion pressures vs. micropipette pressures in the pulmonary circulation

Because of the discrepancies between the arterial and venous occlusion technique and the micropuncture technique in estimating pulmonary capillary pressure gradient, we compared measurements made with the two techniques in the same preparations (isolated left lower lobe of dog lung). In addition, we also obtained direct and reliable measurements of pressures in 0.9-mm arteries and veins using a retrograde catheterization technique, as well as a microvascular pressure made with the double-occlusion technique. The following conclusions were made from dog lobes perfused with autologous blood at normal flow rate of 500-600 ml/min and pressure gradient of 12 mmHg. 1) The double-occlusion technique measures pressure in the capillaries, 2) a small pressure gradient (0.5 mmHg) exists between 30- to 50-micron arteries and veins, 3) a large pressure gradient occurs in arteries and veins greater than 0.9 mm, 4) the arterial and venous occlusion techniques measure pressures in vessels that are less than 900 microns diam but greater than 50 microns, very likely close to 100 microns, 5) serotonin constricts arteries (larger and smaller than 0.9 mm) whereas histamine constricts veins (larger and smaller than 0.9 mm). Thus three different techniques (small retrograde catheter, arterial and venous occlusion, and micropuncture) show consistent results, confirming the presence of significant resistance in large arteries and veins with minimal resistance in the microcirculation.     

Localization of the sites of pulmonary vasomotion by use of arterial and venous occlusion

In this study, we present a new approach for using the pressure vs. time data obtained after various vascular occlusion maneuvers in pump-perfused lungs to gain insight into the longitudinal distribution of vascular resistance with respect to vascular compliance. Occlusion data were obtained from isolated dog lung lobes under normal control conditions, during hypoxia, and during histamine or serotonin infusion. The data used in the analysis include the slope of the arterial pressure curve and the zero time intercept of the extrapolated venous pressure curve after venous occlusion, the equilibrium pressure after simultaneous occlusion of both the arterial inflow and venous outflow, and the area bounded by equilibrium pressure and the arterial pressure curve after arterial occlusion. We analyzed these data by use of a compartmental model in which the vascular bed is represented by three parallel compliances separated by two series resistances, and each of the three compliances and the two resistances can be identified. To interpret the model parameters, we view the large arteries and veins as mainly compliance vessels and the small arteries and veins as mainly resistance vessels. The capillary bed is viewed as having a high compliance, and any capillary resistance is included in the two series resistances. With this view in mind, the results are consistent with the major response to serotonin infusion being constriction of large and small arteries (a decrease in arterial compliance and an increase in arterial resistance), the major response to histamine infusion being constriction of small and large veins (an increase in venous resistance and a decrease in venous compliance), and the major response to hypoxia being constriction of the small arteries (an increase in arterial resistance). The results suggest that this approach may have utility for evaluation of the sites of action of pulmonary vasomotor stimuli.     

Pitfalls of venous occlusion method for determination of capillary pressure in humans

We tested the method of estimating capillary pressure from venous pressure transients obtained after sudden venous clamping in a hydrodynamic model. The basic principles were confirmed in the model, but it was found that when occlusion was caused over a relatively wide distance or in a predistended vessel, capillary pressure was overrated. This problem was due to volume backflow from the occlusion site, since it could be eliminated by placing a one-way valve upstream from the occlusion site. Upstream from the valve, the venous pressure transient accurately followed capillary pressure. Downstream, however, the reading of capillary pressure was impaired by the backflow volume squeezed between valve and occlusion clamp, which caused an immediate large pressure elevation. We also tested the method recently advanced to estimate capillary pressure in humans from venous pressure curves obtained after rapid venous occlusion with an air-filled compression cuff. With the cuff around the upper arm, venous pressure was recorded at different levels along the forearm. The tracings obtained from the dorsum of the hand and halfway along the forearm did not show the initial rapid upstrokes that might indicate the capillary pressure. Tracings obtained slightly below or above the cubital fossa were similar to those seen downstream from the one-way valve in the model. Extrapolation to zero-time, using the distally recorded curves as a template, yielded values equal to venous pressure. We conclude that although the problem of backflow can be circumvented by pressure recording distal from venous valves, the method of venous occlusion by a circular upper-arm cuff may not be appropriate to estimate capillary pressure in humans.     

Pulmonary hemodynamics and lung water content during splanchnic ischemic shock

Pulmonary hemodynamics and lung water content were evaluated in open-chest dogs during splanchnic arterial occlusion (SAO) shock. Mean pulmonary arterial pressure [Ppa = 13.0 +/- 0.6 (SE) mmHg] and pulmonary venous pressure (4.1 +/- 0.2 mmHg) were measured by direct cannulation and the capillary pressure (Ppc = 9.0 +/- 0.6 mmHg) estimated by the double-occlusion technique. SAO shock did not produce a significant change in Ppa or Ppc despite a 90% decrease in cardiac output. An 18-fold increase in pulmonary vascular resistance occurred, and most of this increase (70%) was on the venous side of the circulation. No differences in lung water content between shocked and sham-operated dogs were observed. The effect of SAO shock was further evaluated in the isolated canine left lower lobe (LLL) perfused at constant flow and outflow pressure. The addition of venous blood from shock dogs to the LLL perfusion circuit caused a transient (10-15 min) increase in LLL arterial pressure (51%) that could be reversed rapidly with papaverine. In this preparation, shock blood produced either a predominantly arterioconstriction or a predominantly venoconstriction. These results indicate that both arterial and venous vasoactive agents are released during SAO shock. The consistently observed venoconstriction in the intact shocked lung suggests that other factors, in addition to circulating vasoactive agents, contribute to the pulmonary hemodynamic response of the open-chest shocked dog.     

Compounds that increase cAMP prevent ischemia-reperfusion pulmonary capillary injury.

This study evaluated the physiological effects of compounds that increase adenosine 3',5'-cyclic monophosphate (cAMP) on changes in pulmonary capillary permeability and vascular resistance induced by ischemia-reperfusion (I-R) in isolated blood-perfused rabbit lungs. cAMP was elevated by 1) beta-adrenergic stimulation with isoproterenol (ISO, 10(-5) M), 2) post-beta-receptor stimulation of adenylate cyclase with forskolin (FSK, 10(-5) M), 3) and dibutyryl cAMP (DBcAMP, 1 mM), a cAMP analogue. Vascular permeability was assessed by determining the capillary filtration coefficient (Kf,c), and capillary pressure was measured using the double occlusion technique. The total, arterial, and venous vascular resistances were calculated from measured pulmonary arterial, venous, and capillary pressures and blood flow. Reperfusion after 2 h of ischemia significantly (P less than 0.05) increased Kf,c (from 0.115 +/- 0.028 to 0.224 +/- 0.040 ml.min-1.cmH2O-1.100 g-1). These I-R-induced changes in capillary permeability were prevented when ISO, FSK, or DBcAMP was added to the perfusate at reperfusion (0.110 +/- 0.022 and 0.103 +/- 0.021, 0.123 +/- 0.029 and 0.164 +/- 0.024, and 0.153 +/- 0.030 and 0.170 +/- 0.027 ml.min-1.cmH2O-1.100 g-1, respectively). I-R significantly increased total, arterial, and venous vascular resistances. These increases in vascular resistance were also blocked by ISO, FSK, and DBcAMP. These data suggest that beta-adrenergic stimulation, post-beta-receptor activation of adenylate cyclase, and DBcAMP prevent the changes in pulmonary vascular permeability and vascular resistances caused by I-R in isolated rabbit lungs through a mechanism involving an increase in intracellular levels of cAMP.(ABSTRACT TRUNCATED AT 250 WORDS)     

Effect of leukotrienes B4, C4, and D4 on segmental pulmonary vascular pressures

Leukotrienes (LTs) C4 and D4 are vasoconstrictors and are thought to increase both systemic and pulmonary vascular permeability. However, we and others have observed that LTC4 and LTD4 cause pulmonary vasoconstriction but do not increase the fluid filtration coefficient of excised guinea pig lungs perfused with a cell-depleted perfusate. To determine what vascular segments were exposed to an LT-induced increase in intravascular hydrostatic pressure we measured pulmonary arterial (Ppa), pulmonary arterial occlusion (Po,a), venous (Po,v) and double occlusion (Pdo) pressures in isolated guinea pig lungs perfused with a cell-depleted buffered salt solution before and after injecting 4 micrograms of LTB4, LTC4, or LTD4 into the pulmonary artery. All three LTs increased airway pressures and also increased Ppa, Po,a, and Pdo. Histamine (15 micrograms) as well as serotonin (20 or 200 micrograms) had the same effect. In excised rabbit lungs, histamine and serotonin increased only Ppa, and Po,a. LTC4 had no vasoactivity. There are marked species variations with regard to the activity and site of action of histamine, serotonin, and LTC4 on the pulmonary circulation.     

Reference cardiopulmonary values in normal dogs

The purpose of this project was to collate canine cardiopulmonary measurements from published and unpublished studies in our laboratory in 97 instrumented, unsedated, normovolemic dogs. Body weight; arterial and mixed-venous pH and blood gases; mean arterial, pulmonary arterial, pulmonary artery occlusion, and central venous blood pressures; cardiac output; heart rate; hemoglobin; and core temperature were measured. Body surface area; bicarbonate concentration; base deficit; cardiac index; stroke volume index, systemic and pulmonary vascular resistance indices; left and right cardiac work indices; alveolar partial pressure of oxygen (pO2) ; alveolar-arterial pO2 gradient (A-apO2); arterial, mixed-venous, and pulmonary capillary oxygen content; oxygen delivery; oxygen consumption; oxygen extraction; venous admixture; arterial and mixed-venous blood CO2 contents; and CO2 production were calculated. In the 97 normal, resting dogs, mean arterial and mixed-venous pH were 7.38 and 7.36, respectively; partial pressure of carbon dioxide (pCO2), 40.2 and 44.1 mm Hg, respectively; base-deficit, -2.1 and -1.9 mEq/liter, respectively; pO2, 99.5 and 49.3 mm Hg, respectively; oxygen content, 17.8 and 14.2 ml/dl, respectively; A-a pO2 was 6.3 mm Hg; and venous admixture was 3.6%. The mean arterial blood pressure (ABPm), mean pulmonary arterial blood pressure (PAPm), pulmonary artery occlusion pressure (PAOP) were 103, 14, and 5.5 mm Hg, respectively; heart rate was 87 beats/min; cardiac index (CI) was 4.42 liters/min/m2; systemic and pulmonary vascular resistances were 1931 and 194 dynes.sec.cm-5, respectively; oxygen delivery, consumption and extraction were 790 and 164 ml/min/m2 and 20.5%, respectively. This study represents a collation of cardiopulmonary values obtained from a large number of dogs (97) from a single laboratory using the same measurement techniques.     

Tumor necrosis factor-alpha does not cause lung edema in rabbits.

Although tumor necrosis factor-alpha (TNF) is a key mediator in the pathophysiology of sepsis and septic shock, its role in lung microvascular injury is controversial. In isolated blood-perfused rabbit lungs, we studied the microvascular effects of human recombinant TNF by measuring the capillary filtration coefficient (Kf,c) as an index of microvascular leakiness and the arterial and venous resistances and occlusion pressures to define the microvascular pressure profile. At the end of the experiments, the lung wet-to-dry weight ratio (W/D) was determined as an index of edema. TNF increased the pulmonary venous resistance slightly but did not affect Kf,c or W/D. Furthermore, TNF at different doses failed to increase W/D less than or equal to 8 h after in vivo administration. Our data suggest that 1) the pulmonary microvascular response to TNF differs from the systemic response, which is characterized by arteriolar vasodilation, and 2) TNF is insufficient to cause lung edema, both in vivo and in vitro. Thus the development of lung microvascular injury may require the combined action of TNF and other mediators.     

Bronchial collateral vessel micropuncture pressure in postobstructive pulmonary vasculopathy.

Postobstructive pulmonary vasculopathy, produced by chronic ligation of one pulmonary artery, markedly increases bronchial blood flow. Previously, using arterial and venous occlusion, we determined that bronchial collaterals enter the pulmonary circuit at the distal end of the arterial segment. In this study, we tested the hypothesis that pressure in bronchial collaterals (Pbr) closely approximates that at the downstream end of the arterial segment (Pao). We pump perfused [111 +/- 10 (SE) ml/min] left lower lobes of seven open-chest live dogs 3-15 mo after ligation of the left main pulmonary artery. Bronchial blood flow was 122 +/- 16 ml/min. We measured pulmonary arterial and venous pressures and, by arterial and venous occlusion, respectively, Pao and the pressure at the upstream end of the venous segment (Pvo). Pbr was obtained by micropuncture of 34 pleural surface bronchial vessels 201 +/- 16 microns in diameter. We found that Pbr (14.4 +/- 1.0 mmHg) was similar to Pao (15.0 +/- 0.8 mmHg) but differed significantly (P < 0.01) from Pvo (11.3 +/- 0.5 mmHg). In addition, Pbr was independent of systemic arterial pressure and bronchial vessel diameter. Light and electron microscopy revealed that, in the lobes with the ligated pulmonary artery, the new bronchial collaterals entered the thickened pleura from the parenchyma via either bronchovascular bundles or interlobular septa and had sparsely muscularized walls. We conclude that, in postobstructive pulmonary vasculopathy, bronchial collateral pressure measured by micropuncture is very close to the pressure in precapillary pulmonary arteries and that most of the pressure drop in the bronchial collaterals occurs in vessels > 350 microns in diameter.     

Pulmonary capillaries are recruited during pulsatile flow.

Capillaries recruit when pulmonary arterial pressure rises. The duration of increased pressure imposed in such experiments is usually on the order of minutes, although recent work shows that the recruitment response can occur in <4 s. In the present study, we investigate whether the brief pressure rise during cardiac systole can also cause recruitment and whether the recruitment is maintained during diastole. To study these basic aspects of pulmonary capillary hemodynamics, isolated dog lungs were pump perfused alternately by steady flow and pulsatile flow with the mean arterial and left atrial pressures held constant. Several direct measurements of capillary recruitment were made with videomicroscopy. The total number and total length of perfused capillaries increased significantly during pulsatile flow by 94 and 105%, respectively. Of the newly recruited capillaries, 92% were perfused by red blood cells throughout the pulsatile cycle. These data provide the first direct account of how the pulmonary capillaries respond to pulsatile flow by showing that capillaries are recruited during the systolic pulse and that, once open, the capillaries remain open throughout the pulsatile cycle.     

Venous occlusion pressure and vascular permeability in the dog lung after air embolization

Pulmonary edema has frequently been associated with air embolization of the lung. In the present study the hemodynamic effects of air emboli (AE) were studied in the isolated mechanically ventilated canine right lower lung lobe (RLL), pump perfused at a constant blood flow. Air was infused via the pulmonary artery (n = 7) at 0.6 ml/min until pulmonary arterial pressure (Pa) rose 250%. While Pa rose from 12.4 +/- 0.6 to 44.6 +/- 2.0 (SE) cmH2O (P less than 0.05), venous occlusion pressure remained constant (7.0 +/- 0.5 to 6.8 +/- 0.6 cmH2O; P greater than 0.05). Lobar vascular resistance (RT) increased from 2.8 +/- 0.3 to 12.1 +/- 0.2 Torr.ml-1.min.10(-2) (P less than 0.05), whereas the venous occlusion technique used to determine the segmental distribution of vascular resistance indicated the increase in RT was confined to vessels upstream to the veins. Control lobes (n = 7) administered saline at a similar rate showed no significant hemodynamic changes. As an index of microvascular injury the pulmonary filtration coefficient (Kf) was obtained by sequential elevations of lobar vascular pressures. The Kf was 0.11 +/- 0.01 and 0.07 +/- 0.01 ml.min-1.Torr-1.100 g RLL-1 in AE and control lobes, respectively (P less than 0.05). Despite a higher Kf in AE lobes, total lobe weight gains did not differ and airway fluid was not seen in the AE group. Although air embolization caused an increase in upstream resistance and vascular permeability, venous occlusion pressure did not increase, and marked edema did not occur.