INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Arterial oxygenation is often improved in the prone position compared with the supine position in patients with adult respiratory distress syndrome (ARDS) (1). This is also found in animals with normal lungs (10) and with oleic-acid- induced acute lung injury (6, 13, 14). The prone position improves oxygenation by decreasing ventilation/perfusion (A/) heterogeneity (6, 13), as ventilation to dorsal lung regions is improved (4, 6). Airway closure in dorsal lung regions may be reduced because the gravitational pleural pressure gradient is more uniform when prone (12, 15). However, the improvement in Pa _O2 in patients with ARDS in the prone position is variable, with some patients responding dramatically, whereas others are "nonresponders" (2, 16). The reasons for the variability in response are unclear and are probably multifactorial.

The presence of abdominal distension may influence the improvement in arterial oxygenation with the prone position. Abdominal distension is commonly seen in patients with ARDS after trauma and abdominal surgery because of ileus, fluid administration, and obesity. Marked decreases in pulmonary function and Pa_O2 may occur in patients with severe abdominal distension caused by morbid obesity (17). Abdominal distension reduces respiratory system compliance and may reduce FRC (18, 19). In addition, in the upright position, the gravitational gradient of pleural pressure is increased in pigs by abdominal distension caused by intravascular volume infusion (20) and inflation of an intra-abdominal balloon (19). As the pleural pressure in the dependent lung regions becomes positive, airway closure and atelectasis can occur at FRC in these regions. In the prone position, the pleural pressure in the dependent lung regions becomes more negative than in the supine position, so that much less of the lung is below closing volume (12). We therefore studied pulmonary gas exchange in the supine and prone positions, with and without abdominal distension, in anesthetized, mechanically ventilated pigs. We hypothesized that the prone position would result in a more marked improvement in gas exchange in the presence of abdominal distension than with a normal abdomen.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General Preparation

The study was approved by the University of Washington Animal Care Committee. Eight pigs weighing 15.4 ± 2.4 kg were sedated and anesthetized with xylazine (2 mg/kg) and ketamine (20 mg/kg) injected intramuscularly, followed by ketamine and diazepam intravenously, initially as bolus doses and later as a constant intravenous infusion.

After tracheotomy and endotracheal tube insertion, all pigs were ventilated (NEMI, Medway, MA) with FI_O2 = 0.40, a respiratory rate of 27 breaths/min, and tidal volumes of 13 to 15 ml/kg to keep Pa_CO2 between 37 and 42 mm Hg. A catheter was inserted via cutdown into the carotid artery to monitor systemic blood pressure and heart rate. A pulmonary arterial catheter was inserted through the jugular vein to measure central pressures, cardiac output, and to collect mixed venous blood (Edwards COM 2; Baxter Edwards, Irvine, CA). A venous catheter was inserted into the femoral vein to infuse six inert gases (sulfur hexafluoride, ethane, cyclopropane, halothane, diethylether, and acetone) at a rate of 3 ml/min (21). The inert gas infusion was started at least 1 h before the first measurement. A catheter was inserted into the femoral artery for collection of arterial blood gases (ABL 4; Radiometer, Copenhagen, Denmark). FRC was measured using a helium dilution technique in the following manner: (1) ventilation was stopped and the endotracheal tube was opened to atmospheric pressure, (2) a 2-L syringe containing 300 ml of gas composed of 10% He-90% O₂ was connected to the endotracheal tube, and (3) the lungs were ventilated from the syringe 14 times to mix the helium in the syringe with air in the lungs. The initial and final helium fractions of the mixed gas were measured by mass spectrometry (Perkin-Elmer 1100 Medical Gas Analyzer; Perkin-Elmer, Beaconsfield, Buckinghamshire, UK).

Gastric and Abdominal Balloons

A gastric pressure sensor (thin rubber balloon, 6 cm long, connected to a pressure transducer by 70 cm of PE-100 tubing) was inserted via the esophagus though the lumen of a 16 F Foley catheter and positioned distal to the Foley balloon (19). The Foley catheter was inserted 35 cm into the esophagus so that the tip was well into the stomach. The balloon was then inflated with 7 ml of water, and the catheter was gently pulled back until the Foley balloon prevented further movement. This maneuver ensured placing the gastric sensor inside the stomach below the lower esophageal sphincter. The gastric pressure sensor was then partially inflated with 5 ml of air, a volume previously determined to be on the flat portion of its pressure-volume curve (19). Gastric pressure was taken to represent abdominal pressure (19).

An abdominal balloon was made by connecting a 30-cm tube to a sturdy inflatable rubber balloon (Punch-O-Ball, Big Top) (19). A 2-cm incision was made in the linea alba below the umbilicus, and the abdominal balloon was inserted into the peritoneal cavity and positioned in the center of the abdomen. The peritoneal and skin layers were closed separately, allowing the connecting tube to exit through the abdominal wall with a good seal. Air was eliminated from the balloon.

Experimental Protocol

All pigs were studied in four different study phases in random order. The pigs were placed either supine or prone, with and without water in the abdominal balloon. The abdominal balloon was filled with a volume (110 ± 22 ml/kg) large enough to raise the gastric pressure more than 15 cm H₂O in the supine position. In the prone position, the pigs were resting on their abdomen without other support. The respiratory rate (RR) and tidal volume (VT) were adjusted to maintain normocapnia. After a period of at least 30 min of stabilization to achieve steady-state conditions, gastric pressure, end-inspiratory airway plateau pressure, mean systemic arterial pressure (), mean pulmonary arterial pressure (), pulmonary wedge pressure (Pw), cardiac output (CO), FRC, arterial and mixed venous blood gases, inert gas partial pressures in arterial and mixed venous blood, and mixed expired gas were measured. The mixed exhaled gas samples were kept at a temperature above 40° C until analyzed to avoid condensation and loss of high soluble gases. The concentration of inert gases was measured on a gas chromatograph (Varian 300; Varian Associates, Walnut Creek, CA) equipped with a flame ionization detector and an electron capture detector. The gas extraction method of Wagner and colleagues (24) was used to determine the concentration of inert gases in the blood samples.

Calculations

FRC was calculated as FRC = V_i(He_i/He_fin)  V_i where V_i is the initial volume of the helium mixture and He_i is the initial concentration of helium and He_fin is the final concentration of helium. Respiratory system compliance was calculated as tidal volume/inspiratory plateau pressure. The alveolar-arterial difference in oxygen tension was calculated in the formula [AaPO₂ = FI_O2 × (P_B  P_H2O)  Pa_CO2/R  Pa_O2] where P_B is the barometric pressure, P_H2O is the vapor pressure of water, Pa_CO2 is the partial pressure of CO₂, and R is the respiratory gas exchange ratio. Inspiratory and mixed expiratory fractions of O₂ and CO₂ were measured on the Perkin-Elmer mass spectrometer. R was calculated as

(1)

where FE_CO2 is the fraction of expiratory CO₂, FI_CO2 is the fraction of inspiratory CO₂, and FE_O2 is the mixed fraction of expiratory O₂, all measured by mass spectroscopy.

Gas exchange was assessed by changes in A/ distribution predicted by the 50-compartment model of Wagner and colleagues (21, 23). Inert gas shunt (S/T), dead space (VD/VT), mean A/ ratios of A and distributions, and log standard deviation of the (log SD_Q^.) and log standard deviation of A (log SD_V^.) distributions were calculated from the 50-compartment model. In addition, the arterial-alveolar difference ([a- A]D) area was calculated from the retention and the excretion of inert gases (22). The (a-A)D area was equal to the retention component (R[a-A]D) area plus the excretion component (E[a-A]D) area (25). These indices were used because they are model-independent measures of A/ heterogeneity. R(a- A)D area and log SD_Q^. are comparable because they are parameters that are more sensitive to changes in the distribution (with reference to A/ ratio) and because they tend to increase in the presence of low A/ (22, 26). E(a- A)D area and log SD_V^. are parameters that are more sensitive to changes in the A distribution (with reference to A/ ratio); (a- A)D area does not have a counterpart in the 50-compartment model. Increases in any of the above parameters are indicative of increases in A/ heterogeneity.

Statistics

The data were analyzed by two-factor (abdominal distension and position) within-subject analysis of variance. Significant post-hoc differences were further analyzed by t test for differences among several means; p values less than 0.05 were considered significant. Data are presented as mean ± SD.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

, , and Pa_CO2 remained constant throughout the study (Table 1). With the normal abdomen, Pa_O2 increased by 16 ± 21 mm Hg (p < 0.05) (Table 1) in the prone compared with the supine position. However, the other cardiopulmonary measurements, including AaPO₂ and gas exchange variables were not statistically different in the supine and prone positions, although there was a trend for a decrease in AaPO₂ (p = 0.08) in the prone position (Tables 1 and 2). Respiratory system compliance was low in both positions (16 ± 3 ml/cm H₂O) (Table 1).

Abdominal distension decreased CO (p < 0.01), Pw (p < 0.05), Pa_O2 (p < 0.05), P_O2 (p < 0.01), and pH (p < 0.001) in the supine position (Table 1). Respiratory system compliance decreased (p < 0.01) and VT (p < 0.05), inspiratory plateau pressure (p < 0.001), and FRC (p < 0.05) increased with abdominal distension in the supine position. Abdominal distension increased A/ heterogeneity, as indicated by increases in log SD _Q. (p < 0.01) and R(a-A)D area (p < 0.05) (Table 2). VD/VT (p < 0.01) and mean A/ of distribution (p < 0.05) were also increased (Table 2).

In the presence of abdominal distension, the prone position decreased intragastric pressure from 24 ± 8 to 18 ± 9 cm H₂O (p < 0.01) (Table 1). The plateau pressure decreased (p < 0.05), although respiratory system compliance was not altered in the prone position (Table 1). Pa_O2 increased by 26 ± 18 mm Hg (p < 0.01) and AaPO₂ was decreased by 25 ± 23 mm Hg (p < 0.05) in the prone compared with the supine position in the presence of abdominal distension (Table 1). The decrease in AaPO₂ was secondary to a reduction in A/ heterogeneity, indicated by decreases in log SD _Q. (p < 0.01), (a-A)D area (p < 0.05), and R(a-A)D area (p < 0.01) (Table 2). Other gas exchange and cardiopulmonary variables did not change, except that P_O2 was increased (p < 0.01) (Tables 1 and 2) with the prone position.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The principal finding of this study was that the prone position in pigs improves Pa_O2 and pulmonary gas exchange to a greater degree in the presence of, than in the absence of, abdominal distention.

Prone Position with Normal Abdomen

When the abdomen was normal, the prone position resulted in a small, but statistically significant, increase in Pa_O2 (16 ± 21 mm Hg) (Table 1). On the average AaPO₂ decreased by 20 ± 28 mm Hg (p = 0.08) in the prone position (Table 2). As hemodynamics, P_O2, and Pa_O2 did not change in the prone position, the small increase in Pa_O2 is probably secondary to small decreases in A/ heterogeneity that were not statistically significant (Table 2). Because of the small sample size and interanimal variability, our study may have lacked the power to detect small changes in A/ heterogeneity.

The amount of increase in Pa _O2 with turning from the supine to the prone position is consistent with prior studies in dogs (10, 11), pigs (12), and anesthetized humans (27) with normal lungs. Using the multiple inert gas elimination technique, Beck and colleagues (10) found in dogs that the distribution of blood flow versus A/ ratio (e.g., log SD _Q.) was narrower in the prone than in the supine position. Species differences in collateral ventilation and the intensity of the hypoxic pulmonary vasoconstriction (HPV) response may contribute to discrepancies in gas exchange results in the study of Beck and colleagues (10) and in our study.

Greater improvements in arterial oxygenation (e.g., 100 to 200 mm Hg) in the prone position have been observed in animals with acute lung injury induced by oleic acid (13, 14). Using the multiple inert gas elimination technique, Albert and colleagues (13) demonstrated marked decreases in intrapulmonary shunt in dogs with oleic acid lung injury with turning from the supine to the prone position. Although arterial oxygenation improves in mechanically ventilated patients with ARDS in the prone position (1, 5, 7, 8, 16), increases in Pa_O2 are variable, with some patients responding dramatically, whereas others exhibit little or no response (2, 4, 16). The etiology for the variability in response is unclear and is probably multifactorial. Computed tomography (CT) scans in "nonresponders" demonstrated a redistribution of CT densities to the newly dependent ventral areas, so that inflation of previously atelectatic dorsal lung regions was counterbalanced by atelectasis in previously inflated ventral lung regions. In contrast, both ventral and dorsal lung regions remained ventilated in the prone position in animals with acute lung injury (6).

Prone Position with Abdominal Distension

We hypothesized that the presence of abdominal distension may influence the improvement in gas exchange in the prone position. Abdominal distension commonly accompanies ARDS after trauma and surgery because of ileus, fluid administration, and obesity. We found that when the abdomen was distended, the prone position resulted in a greater improvement in Pa_O2 (26 ± 16 mm Hg) accompanied by significant decreases in AaPO₂ and A/ heterogeneity (Tables 1 and 2). Log SD _Q. was decreased by approximately 16% and R(a- A)D area was decreased by approximately 30% in the prone position (Table 2). These results suggest that in the presence of abdominal distension the distribution of perfusion as a function of A/ ratio is more uniform in the prone position.

Our study is the first to demonstrate position-related changes in gas exchange in the presence of abdominal distension. Pelosi and colleagues (17) found in anesthetized obese patients that Pa _O2 improved from 130 ± 31 mm Hg in the supine position to 181 ± 28 mm Hg in the prone position. However, pulmonary gas exchange, using multiple inert gas elimination technique, was not measured. Mutoh and colleagues (12) found that the prone position increased Pa_O2 and decreased AaPO₂ during intravascular volume overload in pigs. As the respiratory effects of infusing large volumes of fluid are primarily secondary to abdominal distension (20), similar changes would have been predicted with abdominal balloon inflation as in our study.

Mechanisms by which the Prone Position Improves Oxygenation

This study does not elucidate the mechanism for the greater improvement of gas exchange in the prone position in the presence of abdominal distension. This will be the subject of future investigation. Based upon a variety of studies with different species, postures, ventilation modalities, presence or absence of lung disease, and type of abdominal distension, we hypothesize that the prone position improves gas exchange by decreasing the pleural pressure and increasing regional ventilation in the dependent lung near the diaphragm, to a greater degree in the presence of abdominal distension.

The prone position improves oxygenation in ARDS primarily by decreasing A/ heterogeneity (6, 13) by increasing ventilation to dorsal lung regions (6). Airway closure in dorsal lung regions may be less as the gravitational pleural pressure gradient measured in animals is more uniform when prone (12, 15). Abdominal distension caused by inflation of an intra-abdominal balloon (19) or by intravascular volume infusion (20) steepens the gravitational gradient of pleural pressure in pigs in the upright position. The pleural pressure in dependent lung regions becomes positive and may approach 3 to 4 cm H ₂O (19, 20). As airway closure occurs at a transpulmonary pressure of 2.3 cm in dogs (28), significant airway closure and atelectasis might occur at FRC in these regions when the abdomen is distended. In the prone position, the pleural pressure in the dependent lung regions in pigs with oleic-acid-injured lungs becomes more negative and the gradient becomes flatter (12). The absolute change in the pleural pressure gradient between the supine and prone positions is greater in the presence of abdominal distension induced by volume overload than in the normal pig (12). Therefore, a greater improvement in gas exchange would be expected, and was obtained in our study, in the presence of abdominal distension. Thus, our results are consistent with the above postulated mechanism for the improvement in gas exchange in the prone position.

The more uniform pleural pressure gradient in the prone position supports the observation that the distribution of ventilation in the prone position is more uniform than in the supine position in both awake, spontaneously ventilating humans (29, 30) and anesthetized, paralyzed, mechanically ventilated humans (30). Improvements in ventilation to dorsal lung regions are also observed in the prone position in the presence of acute lung injury (6). The prone position may improve ventilation to dorsal lung regions in the area of the diaphragm in the presence of abdominal distension, even though FRC is not reduced, because of changes in position of the diaphragm. However, this is speculative, and further work is necessary to determine posture-related changes in the regional ventilation distribution in the presence of abdominal distension.

Other potential explanations for the improved gas exchange in the prone position include better secretion removal, increases in FRC, changes in regional diaphragm motion, and redistribution of blood flow (2, 6, 9, 16, 17, 27). Enhanced elimination of secretions would not be expected in the presence of normal lungs. The remaining factors are also unlikely to play a role in the greater prone position-induced improvement in gas exchange in the presence of abdominal distension.

FRC was not affected by the posture change in the normal abdomen and in the presence of abdominal distension (Table 1). This result is consistent with most animal studies that reported improvements in Pa_O2, without an increase in FRC, in the prone position (12, 13, 31). Small increases in FRC (38%) have been observed by others in prone dogs (10). However, the prone position may increase FRC to a greater degree in anesthetized, paralyzed humans (17, 27). FRC increased by 53% (from 1.9 ± 0.6 to 2.9 ± 0.7 L) in normal patients (27) and by 122 % (from 0.89 to 0.33 ± 1.98 ± 0.86 L) in obese patients (17) during general anesthesia. Therefore, improvements in FRC may also contribute to improved oxygenation in the prone position in human patients.

A surprising finding of our study was that FRC was not decreased by abdominal distension, as occurred in the upright pig using an identical preparation (19). Differences in the results may be attributable to the differences in postures (e.g., supine/prone versus upright) used in the studies. In our study, FRC was increased by abdominal distension in the supine position (Table 1). Although this result is counterintuitive, the influence of abdominal distension on FRC is dependent upon the relative ability of the abdominal wall to protrude outward versus the diaphragm to protrude cephalad (18, 19). Interactions between the rib cage, abdomen, and diaphragm protect against reductions in lung volume with abdominal distension (18). Abdominal distension may increase FRC if the increased diameter of the upper abdomen distends the rib cage (18, 32). A widened rib cage will result if the increased abdominal pressure is directed toward the area of diaphragm and rib cage apposition and if the diaphragm is pushed cephalad at its insertion on the lower ribs (32). A bucket-handle rotation of the ribs may occur, which increases the size of the rib cage (32). In contrast, end-expiratory lung volume would be decreased if the diaphragm position is moved cephalad more than the rib cage is widened.

Changes in regional diaphragm motion and/or differences in diaphragm activation and pig-ventilator interactions may contribute to improvements in gas exchange in the prone position. Using three-dimensional CT scanning in anesthetized, paralyzed, mechanically ventilated humans, Krayer and colleagues (33) found that the motion of the diaphragm during mechanical ventilation in the supine position was uniform, whereas in the prone position, most motion occurred in the nondependent (dorsal) lung regions. However, the position of the resting diaphragm did not differ between the supine and prone positions (31, 33). The position and motion of the diaphragm during abdominal distension is not known. Displacement of the dome of the diaphragm into the pleural cavity with gastric distension has been indirectly inferred by changes in lung mechanics and end-expiratory volume (34), but it has not been directly studied. Posture-related changes in regional diaphragm motion have also not been studied in the presence of abdominal distension in any species.

Intragastric pressure was lower in the prone position than in the supine position with identical inflation of the abdominal balloon. Gastric pressure may differ from intra-abdominal pressure by several cm H₂O (19). Gastric pressure may be lower in the prone position because of the change in the gravitational direction of the abdominal contents. As gastric pressure was referred to atmospheric pressure, differences in the vertical height of the pressure transducer relative to hydrostatic abdominal pressure gradient would not account for the lower pressure in the prone position. Decreased intra-abdominal pressure in the prone position may alter diaphragm position and motion compared with the supine position.

Finally, redistribution of pulmonary blood flow along a gravitational gradient away from atelectatic lung regions is unlikely to occur. Recent evidence has suggested that the distribution of pulmonary blood flow is primarily determined by anatomic factors, whereas gravity plays a smaller role (11). The perfusion distribution is more evenly distributed and there is no gravitational gradient of flow in the prone position in normal and oleic-acid-injured animals (11, 14). Redistribution of flow away from atelectatic, injured areas in the previously dependent, dorsal lung does not occur (14).

Clinical Implications

It is not appropriate to generalize our experimental findings in pigs with normal lungs to the obese human patient with injured lungs. Species differences in lung physiology, the presence/absence of lung injury, and type of abdominal distension may all contribute to differences in results. All forms of abdominal distension are not equivalent. The distension of a balloon in the abdomen may mimic abdominal distension caused by pregnancy or an abdominal tumor more than that caused by obesity, fluid, or gas within the gastrointestinal tract since the pressure distribution generated by a given increase in volume may be different. Chronic distension is also likely to be different from acute distension because of changes in the compliance of the abdominal wall. Therefore, further research is necessary in obese, mechanically ventilated patients with ARDS before to advocating the use of the prone position in these patients.

It is possible that gas exchange during abdominal distension would be improved in the prone position more in humans than in piglets. Respiratory system compliance in the normal pig is about one-half of the compliance in humans (Table 1) because of low chest wall compliance (19, 20). In addition, the compliances of the diaphragm and abdominal wall, as well as the interactions between the rib cage, abdomen, and diaphragm, may be different in the two species. Abdominal distension caused experimentally by inflation of an abdominal balloon may be quite different physiologically than abdominal distension caused by obesity or ileus. FRC is markedly reduced in the presence of morbid obesity (17), in contrast to a small increase (supine position) or no change (prone position) in FRC observed with abdominal distension in the present study. Marked increases in FRC in obese humans (17) may further enhance the improvement in gas exchange in the prone position.

Gas exchange was improved and gastric pressures were reduced in this study even though the pigs were resting on their abdomens without other support. In many human studies, the shoulder girdle and pelvis of the patients were carefully supported to allow free protrusion and motion of the abdomen (2, 17, 27). Although our results may suggest that such support is unnecessary, caution is advised because of species-related differences in diaphragm, abdominal, and chest wall compliances. When parallel, hard rubber rolls that did not allow free chest and abdominal movement were used to support anesthetized patients, the prone position caused a 30 to 35% decrease in respiratory system compliance and an increase in peak airway pressure (35).

Conclusions

The prone position increased Pa_O2 and decreased AaPO₂ to a greater degree in pigs with abdominal distension than when the abdomen was normal. The improvement was due to a decrease in A/ heterogeneity rather than to a change in hemodynamics.