INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Partial liquid ventilation (PLV) with perfluorocarbon (1) is a new technique that has been successfully used in experimental studies to improve oxygenation via diseased lungs (2, 3). Several Phase II clinical studies have shown that this technique can be used to treat neonatal respiratory distress syndrome (4) and adult respiratory distress syndrome (ARDS) (5). Perfluorocarbons (PFCs) are almost biologically inert liquids that are immiscible with water and have remarkably high affinities for O₂ and CO₂ and low surface tensions. PFCs can improve respiratory mechanics of diseased lungs by filling collapsed units, making them recruitable for gas exchange ("the liquid PEEP effect") and by forming a film with a surface tension lower than that formed by inactivated surfactant or plasma in units not completely filled by PFC (6). The interaction of PFC with the material in airspaces is complex. There seems to be little interaction between PFC and surfactant in normal lungs (7). However, ventilation of diseased lungs with PFC removes cells debris and mucus (4, 8) and lavage with PFC recovers alveolar edema fluid (9), indicating that PFC interacts mechanically with the elements in the alveolar-airway lumen despite its hydrophobic properties. PFC may thus interfere with the contact of alveolar edema liquid with the exchange surface and redistribute edema in zones with different shapes and sizes or transport properties, which may affect alveolar liquid clearance.

The permeability of the alveolar-airway barrier is minimally affected by PLV in normal lungs (10). PFC protects the lungs against overinflation injury when there is alveolar flooding in most cases, but may, in some instances, result in gas trapping and consequently fails to provide protection (11). Air leaks have been reported in diseased lungs during PLV in rabbits (12). This is a concern because integrity of the alveolar- airway epithelial barrier must be preserved for the efficient clearance of alveolar edema (13). Indeed, edema fluid is absorbed from airspaces by alveolar epithelial cells following active sodium transport (14). Thus, transepithelial sodium transport must be able to generate an osmotic pressure difference across the epithelial barrier which, necessarily, should not be too damaged (15).

This study was conducted to evaluate the effect of PFC on alveolar-airway transport properties: Ringer absorption and permeability to nontransported solutes. These transport properties were evaluated in anesthetized, ventilated rats in which alveolar flooding was produced by instilling liquid into the trachea (16).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Male Wistar rats (Charles River, Saint Aubin les Elbeuf, France) weighing 304 ± 2.0 g were anesthetized (thiopental, 50 mg/kg intraperitoneally) and a tracheotomy was performed. The animal was paralyzed with an intravenous injection of a mixture of succinylcholine and heparin, and ventilated (Harvard rodent ventilator; Ealing, Les Ulis, France) with 2 ml tidal volume (VT), 60 counts per minute, 50% fraction of inspired oxygen (FI_O2) at a positive end-expiratory pressure (PEEP) of 3 cm H₂O. After 10 min of ventilation, a volume (1 to 4.5 ml, determined by weighing) of Krebs-Ringer bicarbonate supplemented with 0.4% bovine serum albumin and 10 mmol/L glucose (280 milliosmole [mosmol]) was instilled into the trachea over 30 s at the carina level, followed by 2 ml air. The maximal volume of Ringer bicarbonate instilled was limited by the need to prevent leakage from the tracheal catheter. The instillate contained 0.05 µCi ¹²⁵I-labeled human serum albumin (free ¹²⁵I < 2%), 0.1 µCi ³H-mannitol, and 0.05 µCi ¹⁴C-sucrose. Some of the rats instilled with 2.5 ml or less Ringer bicarbonate were then given 1 (n = 21) or 2 ml (n = 12) perfluorooctyl bromide (perflubron: LiquiVent; Alliance Pharmaceutical Corp., San Diego, CA) via the tracheal catheter followed by 2 ml air. The control rats (n = 29) were not given perflubron. Perflubron was also given before Ringer instillation to better appreciate its barrier effect at the alveolar level. One (n = 5) or 2 (n = 5) ml perflubron were slowly instilled followed by 2 ml Ringer solution. In all these experiments, ventilation was continued for 30 min after Ringer instillation. Rats were killed with a lethal dose of pentobarbital, the thorax was opened, and a sample of blood was obtained by cardiac puncture. As much as possible of the alveolar instillate was recovered by aspiration with a syringe fitted to the tracheal catheter. The blood was immediately centrifuged and the osmolarity of the plasma determined with a freezing point osmometer (Advanced Instruments, Needham Heights, MA). The instillate was centrifuged at 100 × g for 10 min and ¹²⁵I activity per unit volume of supernatant determined by -counting (Wallac Wizard, Turku, Finland). The albumin in the instillate samples was then precipitated with 200 µl 0.6 M trichloracetic acid, the tubes were centrifuged (600 × g for 10 min), and the activity per unit volume of ³H and ¹⁴C in the supernatants was determined with a scintillation counter (Rack Beta, Wallac). Spillover between channels was automatically corrected using standards.

The rate of liquid volume changes in airspaces (J_w) was calculated from the changes in ¹²⁵I-albumin radioactivity per unit volume during a period t, assuming that albumin does not cross the alveolar-airway barrier: J_w = V₀ (C₀  C_t)/(C_t t), where V₀ is the volume of Krebs-Ringer instilled (because ¹²⁵I-albumin does not distribute in perflubron) and C₀ and C_t the ¹²⁵I-albumin activity per unit volume in this liquid at the beginning and end of the 30-min period. Absorption results in a decrease in the liquid volume in airspaces (negative J_w). J_w was corrected for the difference in osmolarity between the rat plasma (302 ± 1.03 mosmol, n = 72) and the Krebs-Ringer instillate: J_wc = J_w × 280/plasma osmolarity.

The permeability-surface area product (PA) for ³H-mannitol (in all but one control rat because of technical problems) and ¹⁴C-sucrose (in 17 control rats, 17 animals given 1 ml PFC and eight given 2 ml PFC) were measured. PA products were calculated from J_wc and the changes in these tracer activities per unit volume according to Reference 17: PA = J_wc[1 + ln (C_t/C₀)/ln ((V₀ + J_wc t)/V₀)], where C₀, C_t, and V₀ have the same meaning as for ¹²⁵I-albumin (because neither mannitol nor sucrose is soluble in perflubron). This calculation assumes that the change in V₀ following the dissipation of the osmolarity gradient is almost instantaneous compared with alveolar liquid absorption. Neither mannitol nor sucrose is transported by epithelial cells; they leak passively from airspaces. The ratio of sucrose to mannitol PA thus indicates the size selectivity of the epithelial barrier.

¹²⁵I-Albumin radioactivity per milliliter was determined in plasma at the end of the experiment. Albumin leak was calculated assuming a two-compartment model in which airspaces are separated from blood by a simple membrane. Assuming that the albumin leak (ml/h) times exchange time (t = 30 min) is small (less than 0.1) compared with the plasma volume (V_p), the following simplified formula was used: Albumin leak = (C_p V_p)/(C_o t), in which C_p and C_o are ¹²⁵I-albumin activities per ml in the plasma and instillate, respectively. Rat plasma volume was estimated from the formula (18): V_p (ml) = 0.0291 · body weight (g) + 2.54.

Statistical Methods

Alveolar liquid absorption rates as a function of the volume of liquid in the lungs were compared by covariance analysis and Bonferroni's post hoc test. Regressions were done with the least mean squares method. Segmental regression was performed according to Gallant and Fuller (19). A value of p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Alveolar Ringer absorption increased as the instilled volume of Ringer solution increased (Figure 1a). The regression lines for absorption versus instilled Ringer volume in control animals and rats given perflubron were parallel (i.e., the slopes of these lines were not different: F = 1.45, not significant [NS] by covariance analysis). However, elevations of these regression lines were larger in rats given perflubron, indicating a significant difference between the mean absorption rates with and without perflubron (F = 11.1, p < 0.001 by covariance analysis). Comparison of the adjusted means of alveolar liquid absorption obtained by covariance analysis is shown in Figure 1b. Perflubron significantly increased the absorption rate (34% for 1 ml and 23% for 2 ml perflubron).

View larger version (25K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 1.   (a) Absorption of Krebs-Ringer bicarbonate from rat airspaces as a function of the volume of buffer instilled. The rate of absorption increased linearly with the instilled Ringer volume in controls not given perflubron (PFC). The same trend was observed with perflubron. The regression lines obtained by covariance analysis show that perflubron shifted the Ringer absorption upwards (solid line: controls; dashed line: 1 ml perflubron; dot-dash-dot line: 2 ml perflubron). (b) Mean liquid absorption adjusted by covariance analysis for the volumes of Ringer instilled expressed as percentage of control (Ringer alone) value. Absorption was significantly faster in the presence of perflubron. *p < 0.05; ***p < 0.001 compared with control values.

We compared the effect of instilling extra Ringer solution or perflubron on absorption by plotting the absorption rate against the total volume of liquid in airspaces (Ringer + perflubron, Figure 2a). Although the increase in absorption rate resulting from adding 1 more ml Ringer or perflubron was similar (NS by covariance analysis), absorption was approximately 50% slower when 2 ml perflubron was instilled instead of 2 ml Ringer solution. Comparison of adjusted means obtained with Ringer + perflubron as the covariate is shown in Figure 2b.

View larger version (25K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 2.   (a) Absorption of the Krebs-Ringer buffer as a function of the total volume of liquid (Ringer + perflubron: PFC) instilled into rat airspaces. There was no difference in absorption rates for control rats not given perflubron and rats given 1 ml perflubron after the Ringer instillation. Absorption was shifted downwards (p < 0.001 by covariance analysis) in rats given 2 ml perflubron (solid line: regression line for controls; dashed line: 1 ml perflubron; dot-dash-dot line: 2 ml perflubron). (b) Mean liquid absorption adjusted by covariance analysis for the volumes of Ringer + PFC instilled expressed as percentage of control (Ringer alone) value. Absorption was significantly slower in the presence of 2 ml perflubron. ***p < 0.001 compared with control values.

The effect of the volume of liquid in airspaces on mannitol PA is shown in Figure 3. Mannitol PA in control rats increased exponentially with the volume of Ringer solution instilled (mannitol PA = 0.07× 10^{(0.56 · V)}, R² = 0.91, p < 0.001). Adding 1 ml perflubron produced the same effect as instilling 1 more ml Ringer, except in two experiments during which the mannitol PA was larger than expected and two others during which it was smaller. By contrast, instilling 2 ml perflubron did not produce an increase in permeability as large as 2 additional ml Ringer in most (11 of 12) cases. Mannitol PA in the presence of 2 ml perflubron was similar to that observed with the same volume of Ringer in the lungs in the absence of perflubron (Figure 3).

View larger version (22K): [in this window] [in a new window]   Figure 3.   Log transform of mannitol PA, which distinguishes more clearly between animals given perflubron (PFC) and control rats. Mannitol PA increased exponentially (dotted lines: 95% confidence interval) with the volume of Ringer solution instilled in control rats. Permeability was similar in almost all animals given 1 ml perflubron when the total volume (Ringer + perflubron) of liquid in the airspaces was taken into consideration. However, it was larger in two instances and smaller in two others. Mannitol PA was significantly (p < 0.001) smaller in rats given 2 ml perflubron than in control rats with approximately (4 to 4.5 ml) the same amount of liquid in the airspaces. The permeability in these animals instilled with perflubron was approximately the same as that in control rats with 1.5 to 2 ml less liquid in their airspaces.

The relationship between sucrose and mannitol PA in control animals was not linear and was better described by two linear segments (R² = 0.996, p < 0.001 by segmental regression) than by a single line (comparison of the two models: F = 4.36, p < 0.05). The first segment (mannitol PA less than 1.04 ml/h) had a slope of 0.61. The slope increased to 0.83 when mannitol PA was above this limit. Perflubron administration did not affect this relationship (Figure 4a).

View larger version (20K): [in this window] [in a new window]   View larger version (18K): [in this window] [in a new window]   Figure 4.   (a) Relationship between sucrose and mannitol PA in control rats that were not given perflubron (PFC; n = 17), and in rats instilled with 1 ml (n = 17) and 2 ml (n = 8) perflubron. There was a change in slope, indicating a loss of selectivity when mannitol PA was larger than 1.04 ml/h (see RESULTS for details). The dashed line is the prolongation of the lower linear segment (slope = 0.61), whereas the continuous line depicts the upper segment (slope = 0.83) determined by segmental regression in control rats. Perflubron instillation did not affect this relationship, whatever the amount instilled. (b) Albumin leak from airspaces as a function of mannitol PA. Albumin leak increased linearly with mannitol PA in animals instilled with 4 ml or more Ringer solution. Note that albumin leak was insignificant in animals given perflubron, some of which had a total volume of liquid in airspaces equivalent to that of the rats instilled with Ringer solution alone.

Almost no ¹²⁵I-albumin was detectable in plasma when mannitol PA was less than 5 ml/h, which was the case in almost all animals given perflubron (Figure 4b). There was a linear relationship between albumin leak and mannitol PAs larger than 5 ml/h (r = 0.99, n = 13, p < 0.001).

Alveolar liquid absorption was similar to that in control rats in animals given 1 or 2 ml perflubron before 2 ml Ringer solution. Mannitol permeability was significantly larger than in controls in these rats. It was also significantly higher in rats given 1 ml perflubron before Ringer than in those given perflubron after Ringer. No significant difference was observed when perflubron volume was 2 ml (Table 1). Albumin leak was low (0.011 ± 0.0029 and 0.007 ± 0.0007 ml/h for 1 and 2 ml perflubron, respectively) and was not different from that of rats given perflubron after Ringer solution.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The important finding of this study is that PLV with perflubron increases alveolar liquid absorption when administered after the aqueous liquid and does not significantly affect the absorption when administered before the aqueous liquid. Further, larger doses of perflubron may oppose an increase in epithelial permeability when there is alveolar flooding.

Several studies examined PLV with perfluorocarbons (1) as a means to ameliorate gas exchange in experimental models of ARDS (6). This new technique has been used in human clinical trials with variable success (4, 5, 8). In addition to ameliorating oxygenation, PLV greatly improves lung mechanics (2, 20, 21), reduces ventilator-induced lung injury (11), and lessens histologic lesions during experimental ARDS (22). This latter may be partly due to the anti-inflammatory effects of PFC (25). There are no published data on the effect of PFC on the transport properties of the alveolar-airway barrier in normal or in injured lungs.

The transport properties of the alveolar epithelium are important for the clearance of alveolar edema. Alveolar edema fluid is absorbed from airspaces by epithelial cells after active sodium transport (14). This has been shown in intact animals (28), including rats (16), and in isolated lungs (17, 29). We used an in vivo model similar to that described by Jayr and coworkers (16) and found absorption rates of the same magnitude.

Alveolar liquid absorption increased linearly with the volume of Ringer instilled in the absence of perflubron. This increase may be caused by recruitment of additional exchange surface area and/or redistribution of the liquid over epithelial surfaces with different exchange properties rather than lung distension by high airway pressures (30). The observed increase in mannitol PA with instilled Ringer volume is consistent with both explanations. It is logical that the recruitment of exchange surface area results in an increase in mannitol PA because there should be a parallel increase in the length of paracellular pathways as additional transport sites are recruited. The exponential increase in mannitol PA with the volume of Ringer solution instilled suggests that the recruited areas did not have the same tightness as those covered by smaller liquid volumes. It is also possible that the barrier properties of the alveolar-airway epithelium are affected by large instillate volumes. This will be discussed later.

The influence of PFC on alveolar fluid absorption has not been previously considered. Arterial oxygenation is decreased during PLV in normal lungs because of the increase in the gas diffusion barrier with PFC (1). Similarly, PFC lining the alveolar walls may preclude efficient alveolar liquid absorption by coming between the epithelial cells and alveolar edema fluid. There seems to be little interaction between PFC and surfactant (and probably the underlying epithelial lining fluid) in normal aerated lungs (7). However, this may not be true in diseased lungs. Indeed, lung lavage with PFC recovers alveolar edema fluid (9), reflecting a mechanical interaction between the hydrophobic PFC and the aqueous phase in the alveolar-airway lumen. We find that instilling 1 or 2 ml perflubron after Ringer significantly increases Ringer rate of absorption. This increase may result from the displacement of the Ringer liquid by perflubron, which then spreads over the additional alveolar surface. This recruitment may occur because of the low surface tension of the PFC, which may facilitate the opening of regions excluded by the presence of a menisci of Ringer in their conducting airways.

By contrast, when perflubron was instilled before Ringer, no increase in absorption was observed. Among possible explanations, the most likely is that Ringer solution preferentially filled empty airspaces because of their lower mechanical impedance. In that case Ringer would cover approximately the same surface area per unit volume as when it was given alone. It has been reported that whereas perfluorocarbon lavage retrieves aqueous liquid present in the lungs (9), the opposite is not true (7). This suggests that if perflubron may disperse the Ringer solution present in distal airspaces, Ringer could hardly penetrate zones already filled with perflubron. The large mannitol permeability with 1 ml perflubron may be explained either by the larger amount of Ringer solution remaining in proximal airways with larger permeability, or by the local trapping by airway closure and overinflation, as discussed previously, or by both factors. It is possible that the larger volume of 2 ml perflubron flowed back in airways and displaced the Ringer solution that remained in these airways. Whatever the explanation, we should stress that tracheal instillation of saline is only a rough model of alveolar edema. During alveolar flooding, the edema fluid enters airspaces at a very distal level, even perhaps at the alveolar level (31) which would probably ensure a better interaction between the two immiscible liquids. In addition, iterative perflubron administration compensating for evaporation would disperse edema liquid. Finally, perflubron did not form a barrier to aqueous liquid absorption, whatever its administration modality.

Instilling 1 ml perflubron after Ringer increased alveolar liquid absorption to the same extent as did one additional mililiter Ringer (Figure 2). This suggests that this low dose of perflubron either increases the exchange surface area available for resorption or redistributes the Ringer to zones with different exchange properties. By contrast, 2 ml perflubron resulted in a smaller increase in absorption than instillation of 2 ml Ringer solution. Several explanations are possible. First, the perflubron may segregate some of the Ringer, excluding this liquid from exchange areas. Second, perflubron may redistribute alveolar fluid to more proximal airways having a different geometry and, thus, different surface-to-volume ratios (32), in the same way it can remove edema fluid by lavage (9). However it occurs, the redistribution of alveolar Ringer solution by perflubron results in a significant increase in its absorption rate (Figure 1b).

The effect of PFCs on alveolar epithelial permeability is a legitimate concern. The integrity of the alveolar-airway epithelial barrier is essential for efficient clearance of alveolar edema (13). There is evidence that the clearance of ^99mTc-DTPA is minimally affected by PLV with PFC in normal lungs (10). Instilling approximately 2 ml Ringer solution into the lungs of rats mechanically ventilated for several hours does not affect the permeability of the alveolar epithelium to macromolecules (16). The effect of instilling Ringer solution on the permeability of the barrier to small solutes is unknown, as the numerous studies on this topic have all been performed in nonventilated isolated lungs (33). We found mannitol PAs in the range 0.2 to 0.8 ml/h in animals instilled with 2 ml Ringer solution or less, in reasonable agreement with the value of 0.6 ml/h found by Berg and coworkers (34) for isolated rat lungs, however filled with a larger (5 ml) volume of liquid but not ventilated. We find an exponential increase in mannitol PA with the volume of Ringer in the lungs (Figure 3). Sucrose and mannitol PAs are linearly correlated, with a slope that agrees with the published ratios for isolated rat lungs (0.53 and 0.59 [34, 35]) and which reflects the fact that the movement of these small solutes across the alveolar-airway barrier is size-limited. Mannitol PA sharply increases when the volume of Ringer instilled is over 2 ml, reaching 10 to 20 ml/h for 4-ml instillates (i.e., 20 to 50 times the values for smaller volumes, Figure 4b). This increase is accompanied by a loss of size selectivity of the barrier to small solutes (Figure 4a), because the slope of the relationship between sucrose and mannitol PA increased to 0.83, a value not different from the ratio of their diffusion coefficients in free water. This loss of selectivity almost always occurred when 1 ml perflubron was instilled, but less frequently when 2 ml perflubron were instilled. It seems unlikely that areas with larger PAs for small solutes were recruited. More likely, the presence of large amounts of liquid with a high surface tension in airways resulted in the exclusion of some zones from ventilation and local overinflation in others because of a "baby lung" effect (36). Overinflation increases epithelial permeability (37, 38), leading in some instances to the leakage of albumin from airspaces (37). Such a leakage occurred in rats instilled with 4 ml Ringer solution. The failure of instilled perflubron to redistribute Ringer might explain why the mannitol PA was larger than expected in two experiments with 1 ml perflubron and did not drop in one experiment with 2 ml (Figure 3). We have already reported that small doses of perflubron may be insufficient to completely prevent overinflation when there is alveolar flooding (11). The instillation of 2 ml perflubron (and 1 ml in two instances) may have reopened these excluded territories in most cases (Figure 3). It is worth noting that 2 ml perflubron did not produce albumin leakage, even when the total volume of liquid in airspaces was over 4 ml. This is consistent with recent studies suggesting that large doses of PFC more effectively improve lung function after surfactant administration in preterm lambs (39). These increases in small solute permeability had little if any influence on alveolar liquid absorption, probably because they are too limited to affect the coupling between Na transport and fluid absorption. Similar observations have been made in isolated rat lungs, in which polycationic compounds increase the permeability to small solutes to a comparable extent without affecting alveolar liquid absorption (40).

Our findings therefore suggest that instilling PFC per se has beneficial effects on alveolar-airway transport and barrier properties when instilled after flooding. It is also possible that large doses of PFC better preserve epithelial permeability when there is alveolar flooding. Taken together, these observations provide additional justification for using PLV during the ventilation of patients with pulmonary edema.