INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Liquid ventilation with perfluorocarbon compounds having high oxygen and carbon dioxide solubility has been studied in animals for over 30 yr and in humans for the past several years (1). During total liquid ventilation, tidal volumes of preoxygenated perfluorocarbon are drained into and out of the lungs. Total liquid ventilation requires specialized equipment that is not commonly available. More recently, the technique of partial liquid ventilation (PLV) has been developed. During PLV, the lungs are filled with a volume of perfluorocarbon up to functional residual capacity. Ventilation continues with gas, using standard ventilators and typical settings which maintain normocapnia (1). Perflubron (LiquiVent; Alliance Pharmaceutical Corp., San Diego, CA) is a sterile perfluorochemical intended for use with ventilator support of acute respiratory distress syndrome. In animal studies using models of acute lung injury, PLV with perflubron has improved oxygenation, increased lung compliance, decreased histologic injury, and decreased short-term mortality, compared to gas ventilation (2).

The improvement in oxygenation during PLV has been attributed in part to hydrostatic elevation of alveolar pressure and recruitment of atelectatic or exudate-filled alveoli (2, 6, 7). That is, the dense perfluorocarbon (approximately 2 g/ml for perflubron) exerts an alveolar pressure proportional to its hydrostatic depth. Hemodynamic changes described during PLV have been minimal. With doses of perflubron up to FRC, pulmonary capillary wedge pressure (Pcw) and thermodilution cardiac output are reported to change little, if at all (5, 7, 8). In contrast, "gas positive end-expiratory pressure (PEEP)," PEEP as typically applied during positive pressure ventilation, exerts a uniform alveolar pressure independent of depth. PEEP has well-known deleterious effects on cardiac output and elevates Pcw. The increase in Pcw is attributable to increased pressure on the surface of the heart; the transmural left atrial filling pressure generally falls on PEEP (9). Thus, gas PEEP dissociates Pcw from a pressure it attempts to estimate, left atrial filling pressure.

Although these artifacts in pulmonary artery catheter (PAC) measurements during gas PEEP are well recognized, the accuracy of these values during PLV has not been validated. There are theoretical reasons why measurements made with a PAC may be biased when the lungs are filled with a dense liquid such as perflubron. First, the use of the Pcw to estimate left atrial pressure (Pla) is predicated on the presence of a continuous column of blood between the wedged catheter and the left atrium (LA). PLV redistributes blood flow away from dependent lung regions (12, 13). By elevating dependent alveolar pressure in excess of left atrial or even pulmonary arterial pressure, PLV may interrupt the fluid continuity between the LA and a previously placed PAC. If this occurs, the Pcw would overestimate Pla. Second, like gas PEEP, PLV could be expected to raise juxtacardiac pleural pressure. Thus, even if Pcw were equal to Pla, both may be artifactually elevated relative to transmural Pla. Finally, the heat capacity of perflubron is much greater than that of respiratory gases. It is possible that loss of the thermal indicator by conduction in the fluid-filled lungs could lead to a systematic overestimation of cardiac output by thermodilution during PLV. Given the brief time available for heat from the lung to be conducted to the thermal bolus, however, this source of error is unlikely to be significant.

Because PACs are frequently used to guide management of patients with acute respiratory disease syndrome (ARDS), we undertook this study to confirm their validity or estimate their error during PLV. In anesthetized, closed-chested sheep, we measured pericardial pressure and compared Pcw to directly measured Pla and thermodilution CO to CO measured with a flow probe. These comparisons were made during gas ventilation and PLV at varied levels of gas PEEP. We hypothesized that, like gas PEEP, PLV would cause Pcw to overestimate transmural Pla. We also hypothesized that the accuracy of thermodilution cardiac output measurements would not be affected by PLV.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgical Preparation

Experiments were performed in six adult sheep weighing 31.9 ± 9.5 (SD) kg. The animals were anesthetized with ketamine (30 mg/kg intramuscularly) followed by pentobarbital (15 mg/kg intravenously). Anesthesia was maintained with continuous supplemental pentobarbital (20 mg/kg/h intravenously) and the animals were paralyzed with pancuronium (5 mg/h) during data collection. They were ventilated (model 607; Harvard Apparatus Co.) via a cervical tracheotomy with 50% humidified O₂ at a tidal volume of 12 cm³/kg, 45% inspiratory time, and a rate of 18. This was maintained throughout the experiment. Via femoral cutdowns, arterial and venous catheters were placed (PE-200 tubing). A 7-Fr. Swan-Ganz catheter was advanced from a femoral vein to the wedged position in the pulmonary artery under waveform guidance. A median sternotomy was performed. The pericardium was opened, and an appropriately sized ultrasound transit-time flow probe (Transonic Systems, Ithaca, NY) was secured around the main pulmonary artery. An air-filled 2-mm-diameter catheter terminating in a flat latex balloon was sutured to the pericardium directly over the left ventricle. A 3-mm-diameter catheter was positioned in the left atrium via its appendage, and the pericardium was laced loosely closed. Bilateral chest tubes were placed through intercostal incisions and positioned in dependent and nondependent regions to keep the pleural cavity free of air and blood. The chest was closed airtight in layers with the catheters sealed with purse-string sutures, and the chest tubes placed to suction. To maintain catheter patency, the sheep was heparinized (5,000 units).

Physiologic Measurements

Vascular and respiratory pressures were transduced with strain gauge transducers (Statham p23) and recorded on a physiologic recorder (model 7D; Grass Instrument Co.). Transducers underwent multipoint calibration against a common pressure source (mercury manometer) prior to each experiment. They were checked periodically for zero drift during the experiment, and low amplitude pressures were recorded at high spacial resolution. Arterial (Pa), pulmonary arterial (Ppa), left atrial, pericardial (Pperi), proximal airway (Paw) pressures, and flow probe cardiac output (_FCO) were recorded continuously. Pcw was recorded intermittently after inflating the PAC balloon with 1.5 ml of air and observing the change in the Ppa to a characteristic Pcw waveform. All pressures were measured at end expiration; mean pressures were recorded by electronic filtration. Thermodilution cardiac output (_TCO) was recorded as the mean of three, 5-ml injections of iced saline at random relative to the respiratory cycle, calculated with a clinical cardiac output computer (model 9520; Baxter Edwards).

The balloon used to measure pericardial pressure was fashioned from a 3.5 × 3.5-cm length of latex Penrose tubing. This was sealed at its open ends with flexible silicone adhesive that incorporated the catheter at one corner. The working volume of the balloon was determined prior to each experiment as follows (14): The Pperi catheter and a large, partially filled but flaccid balloon were attached to separate calibrated transducers. The Pperi balloon was placed on a flat surface, covered with the flaccid balloon, and the two were weighted down under a flat steel plate. Air (< 0.5 ml) was added or removed from the Pperi catheter until the pressures recorded in the two balloons were equal between the range 0 and 20 mm Hg. In addition, the Pperi catheter had to record a pressure of 0 mm Hg with no weight on it (i.e., the working volume was below the unstressed volume), and the pressure had to remain zero when the balloon was curved in orthogonal planes around a cylinder the approximate diameter of the ovine heart.

Experimental Protocol

After the completion of surgery, 15-30 min was allowed to ensure hemodynamic stabilization. Pleural suction was briefly discontinued during data collection. The protocol consisted of three sequential periods of ventilation: gas ventilation (GV), PLV with 10 ml/kg perflubron (PLV10), and PLV with 30 ml/kg perflubron (PLV30). During each period of ventilation, positive end-expiratory pressures of 0, 5, 10, and 15 mm Hg (PEEP₀-PEEP₁₅, equal to 0, 7, 14, and 21 cm H₂O) were applied in random order. At each level of PEEP, 5 min was allowed to reach a steady state and then hemodynamic variables were recorded. After GV was completed, the lungs were partially filled with 10 ml/kg intratracheal perflubron (PLV10) over 5 min during brief interruptions of ventilation. Fifteen minutes was allowed for stabilization. The data collection at levels of PEEP was repeated, and then the lungs were filled with perflubron to a total dose of 30 ml/kg (PLV30). This approximates FRC (15). A meniscus of perflubron was visible in the ventilator tubing above the level of the heart in all animals at 0 PEEP. After 15 min, the data collection at different PEEP levels was repeated. Arterial blood gases were obtained on PEEP₀ during GV, PLV10, and PLV30. Finally, to confirm catheter position in vessels to dependent lobes, the PAC was advanced with the balloon deflated until the waveform took on the appearance of the Pcw. The sheep were euthanized by exsanguination. One milliliter of blue ink was injected into the distal port of the PAC. The chest was opened to confirm position of all intrathoracic catheters, and the location of blue lung staining noted.

Statistical Analysis

All data are given as mean ± standard error. Effects of PEEP at each perflubron dose were tested by one-way analysis of variance (ANOVA). Interactive effects of perflubron dose and PEEP on hemodynamic variables were tested by two-way ANOVA for repeated measures. The effect of perflubron dose on Pa_O2 was examined by one-way ANOVA. When significant effects were seen, data points were compared by least significant difference post-hoc testing. The relationship between thermodilution and probe cardiac output was determined by linear regression at each dose of perflubron and level of PEEP. Statistical significance was assumed at   0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ink staining indicated the PAC had been located in a dependent segment of lung in the costaphrenic recess and below the level of the heart in all animals. Table 1 shows arterial blood gas values at PEEP₀. In these animals with normal lungs, PLV significantly decreased Pa_O2 as has been previously observed (16). Table 2 shows the hemodynamic values during GV, PLV10, and PLV30 at all levels of PEEP. Off PEEP, Pcw agreed closely with Pla during both GV and PLV10. When the dose of perflubron was increased to 30 ml/kg, Pcw slightly but significantly exceeded Pla (6.9 ± 1.4 versus 5.0 ± 1 mm Hg). As PEEP was applied, both Pcw and Pla rose significantly. However, the increases were smaller during PLV (Figure 1) than during GV. This interaction between PLV dose and PEEP was significant (p < 0.005) for Pcw but did not reach significance for Pla (p = 0.10). Moreover, the difference between Pcw and Pla tended to increase with PEEP during GV (0.5 ± 0.7 mm Hg on PEEP₀ to 1.25 ± 0.6 mm Hg on PEEP₁₅, p = 0.24), but tended to fall during PLV30 (1.88 ± 0.9 to 0.5 ± 1 mm Hg from PEEP₀ to PEEP₁₅, p = 0.08; p = 0.056 for the interaction between PLV dose and PEEP).

View larger version (17K): [in this window] [in a new window]   Figure 1.   Pulmonary capillary wedge pressure (Pcw, mean ± SE) during gas ventilation (GV) and both doses of perflubron (PLV10 and PLV30) with 0-15 mm Hg PEEP. During PLV30, Pcw was higher off PEEP, but then rose little as PEEP was increased. * Significantly (p < 0.05) different from GV and PLV10. § Significantly different from PLV10 and PLV30.

Off PEEP, filling the lungs with perflubron had minimal effects on Pperi (Table 2). Although PEEP elevated Pperi, this increase was somewhat attenuated during PLV. Consequently, on 10 and 15 mm Hg PEEP, Pcw agreed more closely with transmural Pla (Platm) during PLV30 than during gas ventilation (Figure 2). Thus, when PEEP was applied during PLV, because of the smaller change in pericardial pressure, Pcw tended to represent Pla and Platm more accurately than it did during conventional ventilation.

View larger version (17K): [in this window] [in a new window]   Figure 2.   The difference between pulmonary capillary wedge pressure and transmural left atrial pressure (Pcw  Platm, mean ± SE) during gas ventilation (GV) and both doses of perflubron (PLV10 and PLV30) with 0-15 mm Hg PEEP. Off PEEP, Pcw more closely approximates left atrial filling pressure during PLV30, and this difference becomes significant at PEEP10 and PEEP15. * Significantly (p < 0.05) different from GV and PLV10. § Significantly different from GV.

As expected, during all modes of ventilation, PEEP decreased cardiac output. It decreased output similarly during GV and PLV. _TCO and _PCO correlated with regression slopes near one and intercepts not different from zero independent of the presence of perflubron (Figure 3).

View larger version (19K): [in this window] [in a new window]   Figure 3.   Correlation of cardiac output during all levels of PEEP as measured with the flow probe (PCO) and by thermodilution (TCO). Data are separately regressed from gas ventilation (GV) and the higher dose of perflubron (PLV30). The regression equations are shown. Both have slopes near unity and intercepts near zero.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We found that pulmonary capillary wedge pressure and cardiac output measurements made with the PAC were as accurate during PLV as during conventional gas ventilation. Unlike gas PEEP, PLV caused minimal elevation of the pressure surrounding the heart. Filling the lungs to FRC with perflubron also did not appear to dissociate Pcw from Pla, suggesting that the PAC remained in zone III lung. Finally, the similar agreement between thermodilution and directly measured pulmonary artery flow during PLV or GV suggests that thermal indicator loss to the perflubron is insignificant.

Critique of Preparation

This experiment was performed in sheep, whose anatomy differs from human in a few relevant aspects. The pulmonary arteries in these sheep were smaller than in adult humans. Therefore, a standard flotation PAC will wedge in a more proximal vessel than it would in a human patient. By subtending a larger volume of lung, the wedged segment may be more likely to include lung regions where zone III conditions persist despite perflubron filling. The depth of the sheep thorax is greater than in humans. During PLV, this would cause a greater range in alveolar (and pleural) pressure from nondependent to dependent lung regions. Finally, any alterations in chest wall compliance following sternotomy and closure would alter the extent to which alveolar pressure is transmitted to the pleural space. Therefore, these findings should be applied only cautiously to humans.

Accurate measurement of pressure on the surface of the heart is challenging. The ideal pressure sensor would be volumeless and insensitive to shape changes. Furthermore, pressure on various regions of the cardiac surface vary, and change differently with changes in lung volume (17, 18). We used a thin, flat, balloon within the loosely approximated pericardium over the lateral left ventricular surface. It was calibrated and tested ex vivo using standard methods (14). Nevertheless, our data suggest that this measurement of pericardial pressure may have underestimated the generalized PEEP-induced changes in cardiac surface pressure. During gas ventilation, 15 mm Hg PEEP caused an increase in Pperi averaging 3.75 mm Hg. This is somewhat less than the transmission of airway pressure to the pleural surface in animals with normal lungs reported by some (19, 20), but similar to changes in pericardial pressure reported by others (21). Reflecting this possible underestimation, we recorded small increases in Platm associated with decreases in cardiac output as PEEP was applied during both GV and PLV. This is not likely attributable to increased left ventricular afterload or impaired myocardial contractility, because PEEP causes neither (10, 11, 20). Both Huberfeld and coworkers (19) and Takata and Robotham (21), using balloons similar to ours, found that CPAP or PEEP decreased transpericardial pressure in hypervolemic animals. In the present study, no animal was hypervolemic. However, we cannot exclude that pericardial tension decreased with PEEP, attenuating the rise in intrapericardial pressure. In addition, it is possible the numerous nearby cannulae, flow probe, and catheters may have sheltered the pericardial balloon from nearby changes in pressure.

Nevertheless, within individual animals the position of the pericardial balloon and instrumentation remained unchanged for the duration of the experiment. Therefore, although PEEP-induced absolute changes in pericardial pressure may have been underestimated, relative changes during GV and PLV should still be comparable. Directly measured vascular pressures were also free from this potential artifact.

Pcw as measured through a PAC is widely used as a clinical estimate of the left ventricular preload. Numerous assumptions are implicit in that estimate. For example, since Pcw is measured relative to atmospheric pressure, it is assumed the pressure surrounding the left atrium is near ambient at end expiration. With the application of PEEP, this assumption may be incorrect. Generally, left atrial transmural pressure falls with PEEP, but Pcw rises. This effect is well recognized and expected, and was observed in the present study. Several techniques have been suggested to correct for this artifact from elevated pleural pressure (22).

In the perflubron-filled lung at end expiration, alveolar pressure will be determined by the depth of overlying perflubron. At a depth of 20 cm at the base of the lung, for example, alveolar pressure would be 40 cm H₂O. The beneficial effects of PLV on gas exchange have been attributed in part to this "liquid PEEP." This is believed, like conventional PEEP, to prevent expiratory alveolar collapse, and moreover to fill the alveoli with an oxygen-containing compound instead of exudate (2). We hypothesized that filling the lungs with perflubron would elevate the pressure surrounding the heart and thereby elevate both Pla and Pcw.

Previous investigators have measured lateral pleural pressure canine lungs during PLV and found that filling the lungs with perflubron increased the vertical pleural pressure gradient. Goldner and coworkers (25) reported lateral pleural pressure measured with flat balloons during PLV with perflubron in normal dogs. At the mid-thorax, they found essentially no change in pleural pressure. Our results are consistent with theirs. We found little change in pleural pressure measured in the pericardium, near the hydrostatic midpoint of the thorax, while they found little change at the mid-lateral pleural space. Barbas and coworkers (26) reported similar findings in three dogs with oleic acid lung injury undergoing PLV and PEEP.

The explanation for this may be as follows: Assume there is no change in the volume of the lungs when they are filled with perflubron, that is, 30 cm³/kg of liquid displaces exactly 30 cm³/kg of gas. With constant lung volume, there must be no change in the volume of the chest wall. If chest wall compliance is unchanged by PLV, then the pleural pressure averaged throughout the pleural surface must also be unchanged. It is clear that local pleural pressure in the dependent regions must rise from the weight of the perflubron above them. For mean pleural pressure to remain constant, pleural pressure in nondependent regions must fall. Dependent lung will tend to expand, and nondependent lung tend to retract. Somewhere in-between, the pleural pressure will remain unchanged. This may be at the approximate midpoint, where we measured pericardial pressure.

A second assumption necessary for Pcw to approximate Pla is that a static, continuous column of blood exists between the tip of the catheter and the large pulmonary veins. This occurs in dependent, zone III of the lung, where Pla exceeds alveolar pressure at end expiration (27). However, when alveolar pressure is raised preferentially in dependent regions by the addition of perflubron, zone III regions might be converted to zone II or I, where alveolar pressure exceeds Pla or Ppa, respectively. Our study cannot directly answer this physiological question. However, when the PAC was wedged conventionally by balloon inflation on PEEP0, Pcw and Pla agreed fairly well in the gas-filled and perflubron-filled lung. This indicates that at least one fluid-filled channel remains patent distal to the PAC, somewhere within the segment of lung perfused by the artery in which the catheter is wedged. Taken together, the close agreement between Pcw and Pla and the trivial changes in Pperi attributable to PLV indicate that the PAC retains its utility as a diagnostic and monitoring tool in the perflubron-filled lung.

When PEEP is applied, our findings suggest that the PAC may actually reflect both Pla and transmural Pla more accurately during PLV than during gas ventilation. During PLV, Pperi, mean Ppa, and the difference between both Pcw and Pla and Pcw and transmural Pla changed less with PEEP than they did during gas ventilation. It has been shown in other animal studies that during gas ventilation, the hemodynamic effects of PEEP are attenuated when hyperinflation of the lower lobes is prevented (28). Recent studies of animals receiving PLV have shown that tidal ventilation is distributed to nondependent lung regions (29). It is therefore likely that the volume changes induced by PEEP, in the perflubron-filled lung, are also largely confined to nondependant lung regions. First, any gas remaining in the lung during expiration would float above the perflubron. Second, dependent lung regions would be nearly fully distended by the perflubron alone, and regional pressure-volume relations would favor volume changes in nondependent regions with PEEP. In that case, although filling the lungs with perflubron may alter local pleural pressure, pulmonary blood flow, and impedence to right ventricular ejection in the dependent regions, those regions are thereafter somewhat protected from further changes induced by gas PEEP.

Nevertheless, PEEP caused nearly identical decrements in cardiac output during PLV as during GV. Two possible explanations for this are, first, that PEEP may induce different reflex effects during GV compared to PLV. Consistent with this explanation, heart rate tended to rise with PEEP during GV but fell with PEEP during PLV (p = 0.056 for the interaction of PEEP and PLV). Second, effects of PEEP on cardiac output depend in part on PEEP-induced changes in right atrial pressure, the downstream pressure for venous return. The right atrium and right ventricle are relatively nondependent structures in a supine sheep, filling the anterolateral cardiac fossa. With the application of PEEP, lung volume changes in nondependent regions would be similar during PLV and GV. Therefore, PEEP-induced changes in pleural pressure over the right atrium and in right atrial pressure may be similar during PLV and GV, resulting in equal decrements in venous return. Since right atrial pressure was not recorded in this study, however, we have no data to support or refute this hypothesis.

As expected, thermodilution CO was not biased by PLV. Mean flow velocity in the pulmonary artery in man or dog is 18-19 cm/s (30). Thus, mean transit time between the injection port in the right atrium and the distal thermistor is only about 1.5 s. During much of that time, the blood is surrounded by myocardium or mediastinal tissue, rather than lung. It is nearly inconceivable that sufficient heat would be lost from the perflubron to the saline bolus to bias cardiac output measurements.

In conclusion, we find that hemodynamic measurements made with the PAC remain at least as accurate during PLV as during gas ventilation in normal sheep lungs. PLV up to FRC filling doses causes little elevation in juxtacardiac Ppl. Fluid continuity is maintained between the site of balloon catheter wedging and the pulmonary veins. "Liquid PEEP" from PLV, in contrast to gas PEEP, has minimal hemodynamic effects. Furthermore, it tends to moderate the hemodynamic effects of gas PEEP. We speculate that is because it displaces gas in the lung, does not change end-expiratory lung volume, and prevents further hyperinflation of dependent lung regions when gas PEEP is added. Further study will be necessary to determine if these findings apply with variations in intravascular volume, or in the presence of acute lung injury where the effects of PLV on lung compliance and volume may be different than in normal lungs.