Comparative Effects of Vaporized Perfluorohexane and Partial Liquid Ventilation in Oleic Acid– induced Lung Injury

All animal procedures were approved by the Institutional Animal Care Committee and the Government of the State of Saxony, Germany, and conformed to the Guide for the Care and Use of Laboratory Animals .15 

Thirty female pigs weighing 31.1 ± 3.5 kg (range, 25–37 kg) were premedicated with ketamine (2–3 mg/kg intramuscular) and brought to the experimental operation room, where an ear vein was punctured. After that, the trachea was intubated with a cuffed endotracheal tube (7.5 ID), and an esophageal catheter (Erich Jaeger GmbH, Höchberg, Germany) was placed and advanced into the mid chest until optimization of the position was performed (see Instrumentation and Measurement Devices section). Thereafter, anesthesia was deepened and maintained with midazolam (0.5- to 1-mg/kg bolus plus 1–3 mg · kg⁻¹·h⁻¹intravenous) and ketamine (3- to 4-mg/kg bolus plus 5–10 mg · kg⁻¹·h⁻¹intravenous). Paralysis was achieved with 2 mg pancuronium bromide administered intravenously every hour. Volume status was maintained by means of the administration of a crystalloid solution (E153; Serumwerk Bernburg AG, Bernburg, Germany) at a rate of 10–15 mg · kg⁻¹·h⁻¹, with a colloid solution (10% hydroxyethyl starch; Fresenius Kabi Deutschland GmbH, Bad Homburg, Germany) being given as necessary to keep the hemoglobin concentration in blood approximately constant (normovolemia).

Volume-controlled ventilation was initiated using an anesthesia machine with a semiclosed circle system (Cato®; Drägerwerk AG, Lübeck, Germany). Ventilator settings were as follows: fraction of inspired oxygen (Fio²) of 1.0 (fresh gas flow of 4 l/min), PEEP of 5 cm H²O, and inspiratory:expiratory ratio of 1:1.7, which corresponds approximately to the physiologic value. We also used tidal volumes (V^T) of 9 ml/kg, which corresponds to an intermediary value between protective (6 ml/kg) and conventional ventilation (12 ml/kg). The reason for choosing this value was that in our experience, carbon dioxide elimination is easier to achieve while keeping inspiratory plateau pressures lower than 35 cm H²O. Respiratory rates were adjusted to an arterial carbon dioxide tension (Paco²) of 35–45 mmHg. Except for PEEP, which was readjusted later (see Protocol for Measurements section), all settings of the mechanical ventilator were kept constant throughout the experiments.

The right anterior surface of the neck was infiltrated with 5 ml lidocaine, 1%, and a paramedian skin incision was performed. After carefully preparing the external carotid artery, an indwelling catheter was inserted, through which the sensor of a continuous blood gas analysis system (Paratrend 7+®; Diametrics, St. Paul, MN) was introduced. Mean arterial pressure was continuously monitored with a CMS Monitor (Agilent Technologies, Böblingen, Germany). Also, a pulmonary artery catheter (Opticath®; Abbott Laboratories, Abbott Park, IL) was advanced through an introducer set placed in the external jugular vein until a typical pulmonary artery curve could be visualized. Mean pulmonary artery pressures were then continuously measured with the CMS Monitor. Urine was collected by means of a bladder catheter, which was inserted during a mini-laparotomy.

A heated Fleisch pneumotachograph No. 2 (Fleisch, Lausanne, Switzerland) connected to a differential pressure transducer (PXL12X5DN; Sensortechnics, Troy, NY) was placed between the Y-piece of the ventilator tubing and the endotracheal tube. Airway pressure (P^aw) was monitored by a second pressure transducer (SCX01DNC; SenSym ICT, Milpitas, CA), whose tip was placed at the proximal ending of the pneumotachograph (near to animal). The tip of the capillary of a respiratory mass spectrometer (LR6000; Logan Research, Rochester, United Kingdom) was placed at the distal ending of the pneumotachograph, and the respiratory gas mixture was sampled at 60 ml/min. Finally, the esophageal catheter balloon was inflated with 1–2.5 ml room air and connected to a third pressure transducer (SCX01DNC) to measure the esophageal pressure (P^es). The position of the esophageal catheter was optimized to obtain a correlation coefficient of approximately 1 between ΔP^awand ΔP^esduring thoracoabdominal compression maneuvers with a closed respiratory system, i.e. , without a phase lag in the plot the two signals, according to a method described elsewhere.16 

Arterial and mixed venous blood samples were analyzed for oxygen and carbon dioxide tension, pH, base excess, and [HCO³⁻] using an ABL 505 (Radiometer, Copenhagen, Denmark), and for arterial oxygen saturation and hemoglobin concentration using an OSM 3 Hemoximeter (Radiometer) calibrated for swine blood. Gas tension measurements were performed at 37°C and corrected for pulmonary artery temperature at sampling. The oxygen content in arterial, end-capillary, and mixed venous blood, as well as intrapulmonary right-to-left shunt (Q̇^s/Q̇^t), were calculated using standard formulae, as described elsewhere.11 

Gas concentrations measured by the mass spectrometer, airway flow measured by the pneumotachograph V̇, P^aw, and P^eswere digitized at 150 Hz and acquired by the Data Acquisition System, which is a main platform written for the LabVIEW environment (National Instruments, Austin, TX). The Data Acquisition System supports the integration of other subroutines which were developed for calibration, calculation of parameters, and control of processes (see Calculations of Respiratory Mechanics section).

The following gases and vapor were measured at 32 Hz by the mass spectrometer: N², O², CO², Ar, and vaporized perfluorohexane. Calibration was performed with a two-point linear procedure internal to the device, which requires two gas mixtures with at least one common gas. The first mixture contained 78% N², 17% O², and 5% CO²and was produced and certified by Messer Griesheim GmbH (Ludwigshafen, Germany). The second mixture contained 71.29% N², 19.19% O², 0.86% Ar, and 8.57% perfluorohexane and was obtained by vigorously shaking a chamber of variable volume containing room air and a liquid phase of perfluorohexane at 0°C. The fraction of perfluorohexane in the second mixture was estimated according to equation 1, which was derived by the National Institute of Standards and Technology (NIST) based on the data from Crowder et al.  17:

where F^PFHis the fraction of perfluorohexane vapor and T^Kis the temperature in °K.

For the calibration of the pneumotachograph, animals were ventilated with an intensive care mechanical ventilator (Evita 2; Drägerwerk AG, Lübeck, Germany) whose respiratory settings were the same as those of the anesthesia machine. Before the change of the ventilators, V^Tof the Evita 2 was verified with a side stream spirometry tester (Datex, Helsinki, Finland). The flow signal was recorded by the Data Acquisition System during 2 min, and the inspiratory flow was increased by 50% in the last 30 s of acquisition without changing V^T. Increase of flow was necessary to obtain a wider flow range where the calibration procedure should be performed. Signals were processed off-line by a subroutine written for Matlab® (MathWorks, Natick, MA) to calculate the coefficients of a third-order polynomial that best fitted to the flow data according to a procedure described elsewhere.18Calibration was accepted if the error of inspiratory and expiratory V^Tin each respiratory cycle was less than 3%. Correction of flow for the variation of the viscosity in the presence of perfluorohexane was performed using empiric factors obtained in our laboratory. Such procedures were necessary because we noticed that the calibration of the Fleisch pneumotachograph outside of the respiratory circuit with a 3-l syringe, as described by other investigators for spontaneous breathing,18was not appropriate for use during mechanical ventilation. After calibration, animals were disconnected from the Evita 2 and reattached to the anesthesia machine. Change of ventilators was always preceded by clamping of the endotracheal tube during inspiration to avoid loss of lung volume.

For the calculation of time delay between flow and gas concentration, a special syringe containing two gas mixtures separated by the piston was used, as described in detail in the work by Brunner et al.  19Briefly, the wall of the syringe has a leak where the tip of the capillary of the mass spectrometer is attached. When the piston is moved, a step change in gas concentration is applied to the tip of the capillary. If the volumes on both sides of the piston are known, the exact moment the piston passes at the tip of the capillary can be determined, and the delay in measuring the gas concentration step in the mass spectrometer can be assessed. In this work, the step change in gas concentration was achieved with a first gas mixture corresponding to room air and a second gas mixture consisting of 78% N², 17–19% O², and 3–5% CO². Time delay (T^D) between flow and gas concentration measured in this way can be divided into a viscosity-dependent part and a viscosity-independent part. For the purpose of simplification, it was assumed in this work that T^Dconsisted only of a viscosity-dependent part, which was calculated according equation 2:

where j is the sample, T^Dois the time elapsed from the application of a carbon dioxide step change at the tip of the capillary of the mass spectrometer until 50% of the final carbon dioxide value, and η⁰is the viscosity of the gas mixture used. The variable η̄ represents the average viscosity of the time sample j traveling in the capillary, whose calculation has been shown in the detail in the work by Brunner et al.  19Viscosities of isolated gases and mixtures were calculated according to Turner et al.  20 

P^awand P^eswere calibrated using a water column as reference.

Elastance and resistance variables were determined on-line according to a simplified version of the equation of motion, as described in equation 3:

where P is the motion pressure, V is volume, t is time, P^restcorresponds to the pressure at rest volume, and E and R represent elastance and resistance, respectively. Partitioning of E and R into lung and chest wall components (E^L, E^CW, R^L, and R^CW, respectively) was achieved by applying the appropriate motion pressures, as described in equations 4 and 5:

where RS represents the respiratory system values, P⁰is the residual airway pressure at the end-expiratory lung volume (approximately PEEP), and P¹is the residual pressure in the pleural space at end-expiration. Curve fitting to a one-compartment model was performed with data acquired during periods of 2 min, and parameter extraction was obtained using a recursive minimal least squares method according to a method described elsewhere.21After sudden changes in lung mechanics, values usually stabilized within 2 min. The chest wall components of elastance and resistance (E^CWand R^CW, respectively) were calculated as the difference between the respective respiratory system and lung components.

Functional residual capacity measurements were obtained with the multiple breath nitrogen washout method. The Fio²was reduced to 0.4, while leaving other settings of the anesthesia machine unchanged, and enough time was allowed for the respiratory system to stabilize (up to 15 min). After stabilization, the endotracheal tube was clamped at end-expiration until the animal was connected to the Evita 2 mechanical ventilator. The settings of the Evita 2 were the same as those from the anesthesia machine, except for the Fio², which was fixed at 0.8. Flow and gas concentration data were acquired by the Data Acquisition System for off-line processing. The washout maneuver was interrupted when differences between inspiratory and expiratory N²concentrations (F^IN2and F^EN2respectively) were negligible, i.e. , F^IN2− F^EN2< 1%, or remained unchanged longer than 1 min. After that, the endotracheal tube was clamped, and the animal was connected back to the anesthesia machine, whose settings were the same as before the washout maneuver.

Calculation of FRC followed the mass balance principle, according to equation 6:

where t^Bis the begin of the washout maneuver and t^Eis the end of the calculation period. Time t^Ewas defined as the time when FRC calculation achieves a plateau.

Cardiac output was measured by means of the thermodilution method. Four measurements with 10 ml iced saline 0.9% solution equally spread over the respiratory cycle were obtained and averaged.

Acute lung injury was induced by means of the central venous infusion of OA. Initially, 0.05 ml/kg OA was diluted in arterial blood to obtain 10 ml total volume. The mixture was gently shaken, and aliquots of 0.25–1 ml were administered carefully through the proximal lumen of the pulmonary catheter over 1 min, while increasing the parallel administration of crystalloid to a rate of 800–900 ml/h to minimize the impact of boluses of OA and improve volume status during injury. The injection of the mixture was stopped if the mean arterial pressure decreased to under 50 mmHg, and administration of adrenalin boluses was permitted if spontaneous recovery was not observed within 30 s. This procedure was repeated during 1–2 h, until the arterial oxygen tension (Pao²) decreased to below 200 mmHg and did not show a trend toward improvement within 10 min. If necessary, a second dose of OA was prepared in the same way and administered within this period, the total OA amount being situated within 0.03–0.125 ml/kg. During the injury period, the Pao²was assessed continuously with the Paratrend 7+ (Diametrics) to allow better control of injury, but decisions regarding continuation or interruption of the procedure were taken by means of arterial blood sampling and analysis with the ABL 505.

Figure 1shows the time course of events. After instrumentation, a period of 30 min was allowed for animals to stabilize. Baseline measurements were then obtained and injury was performed in 30 animals, as previously described. After injury (time 0), animals were assigned to one of the five therapy groups after opening sealed envelopes (n = 6/group): (1) control—no therapy with perfluorocarbon or gas ventilation (GV), (2) therapy with 2.5% perfluorohexane, (3) therapy with 5% perfluorohexane, (4) therapy with 10% perfluorohexane, and (5) PLV with perfluorooctane (30 ml/kg) (see Partial Liquid Ventilation section). Two hours thereafter, measurements were repeated (time 2), and the lungs were recruited by means of a continuous pressure of 40 cm H²O during 30 s. To avoid derecruitment and stabilize the lungs thereafter, PEEP was set at 20 cm H²O while other respiratory parameters remained unchanged. After that, PEEP was reduced in steps of 5 cm H²O up to the start value, with periods of 3 min being allowed at each level (20, 15, 10, and 5 cm H²O). Measurements of total respiratory elastance (E^RS= E^L+ E^CW) and Pao²were taken at each step. After this procedure, recruitment was repeated in the same manner, and PEEP was set at the level corresponding to the lowest E^RSlevel. If E^RSvalues were comparable between two steps (difference < 5 cm H²O/l), the higher Pao²level was used to set PEEP. Measurements were repeated 4 and 6 h after injury (times 4 and 6, respectively). Except for PEEP, mechanical ventilator settings were the same as before lung injury. Finally, animals were killed by means of the intravenous injection of 2 g thiopental and 50 ml KCl, 1 m.

Perfluorohexane (C⁶F¹⁴; ABCR, Karlsruhe, Germany) with a purity of 95% was used in this work. This compound was chosen because of its relatively low molecular weight (338 amu) and unique physicochemical properties among the perfluorocarbons, particularly the relatively high vapor pressure (273 mmHg at 30°C), which permits its administration in vapor form.22Accordingly, administration of perfluorohexane vapor was accomplished by means of two standard bypass vaporizers, type 19 n (Drägerwerk AG, Lübeck, Germany), which were modified by a scaled opening of the dosage control cone. Vaporizers were connected in series in the inspiratory limb of the anesthesia machine and opened to permit mixing of perfluorohexane with the fresh gas flow. Inspiratory and expiratory concentrations of perfluorohexane were measured continuously by the mass spectrometer, ensuring a controlled application.

Perfluorooctane (C⁸F¹⁸; ABCR) with a purity of 98% was used during PLV in this work. This substance was chosen because it has a relatively low vapor pressure at 37°C (60.9 mmHg), which permits its use without important losses by means of evaporation, and other physicochemical properties that are comparable to those of perfluorooctyl bromide, namely specific weight (1.74 mg/ml), solubility for oxygen (52.1 ml/100 ml), viscosity (0.85 cs °C), and surface tension (13.9 dyn/cm),22which has been used for PLV in different studies. Before administration, oxygen was bubbled in 30 ml/kg perfluorooctane at a rate of 4 l/min during 2 min. After that, animals were disconnected from the anesthesia machine, and aliquots of 10%, 20%, 30%, and 40% of the total amount were dropped into the endotracheal tube with 1- to 2-min intervals between consecutive applications. Compensation for evaporative losses was performed by means of administration of perfluorooctane every 2 h through a special side port constructed for the endotracheal tube, to avoid disconnection from the ventilator. The compensation volume was extrapolated from the work by Reickert et al.  23According to these authors, the maximum compensation rate for PLV with perfluorooctylbromide in adults should be 9.4 ± 3 ml/h. Because the vapor pressure of perfluorooctane is approximately fivefold that of perfluorooctylbromide and the volume of perfluorooctane used in our animals was approximately one half of that used in humans,2350 ml perfluorooctane was refilled every 2 h in our study.

Postmortem, a median sternotomy was performed to excise the heart and lungs en bloc . We opted not to use continuous positive airway pressure during this procedure because we were aware of the possibility that lungs treated with perfluorohexane or PLV might not collapse as easily as nontreated lungs at atmospheric pressure. Because one of the histologic aspects to be investigated was development of microatelectasis, such an effect would be more easily detected if no continuous positive airway pressure was used. The heart was dissected from the lungs, and eight samples from defined sectors were obtained. Blocks of lung tissue of approximately 8 cm³were taken from the outermost apex and the central region of both the upper and lower lobe of the right lung. The lung tissue blocks were immediately immersed in a 4% neutral-buffered formaldehyde solution and fixed for several days. After embedding the blocks in paraffin, slices of 5 μm were cut and stained with hematoxylin and eosin, using routine histologic techniques. The sections of the samples that were used for staining were randomly selected by assisting technicians unfamiliar with the experiment. To quantify the extension and severity of histologic lung injury, hematoxylin and eosin–stained sections were observed by an expert in lung pathology (M. K.) who was blinded to the groups. The pathologic features of the lung in adult respiratory distress syndrome can be characterized under the term diffuse alveolar damage  (DAD), which was first proposed by Katzenstein and Askin.24DAD corresponds to a stereotypic pathologic sequence of events after severe acute lung injury regardless of the initial cause. The scoring system used in this work was adapted and modified to include a weighting system similar to that proposed by Broccard et al.  25This modification was performed to improve the quantification of the pathologic features according to their extension. We restricted the numerous features described in the literature to seven basic ones, which, in our opinion, are more representative for our model, namely alveolar edema, interstitial edema, hemorrhage, inflammatory infiltration, epithelial damage, microatelectasis, and overdistension. These features were evaluated in four separated nonoverlapping fields of view of hematoxylin and eosin–stained section by light microscopy (Leica DM RB; Wetzlar, Germany) using magnifications of ×25, ×100, ×250, and ×400. In our scoring system, values from 0 to 3 represented the severity of the feature, as follows: 0 = normal appearance, 1 = slight effect, 2 = intermediate effect, and 3 = severe effect. A further part of our system was used to describe the extent of involvement in each field of view, as follows: 0 = lack of involvement of the feature, 1 = up to 25%, 2 = 25–50%, 3 = 50–75%, 4 = 75% to almost total involvement, and 5 = total involvement of the field. For each feature evaluated, severity was multiplied by the extent, leading to values in the range of 0–15. Values of all fields per lung (n = 16) were averaged. The sum of features corresponded to the DAD score.

Values are given as mean ± SD. Paired Student t  tests were used to assess the effects of lung injury. Comparability among groups at baseline and time 0, total amounts of OA per group, PEEP levels after lung recruitment, and histologic findings were tested with one-way analysis of variance. Repeated-measures two-way analysis of variance was used to determine the effects of therapies on time course of variables (time, group, and time * group effects). When significance was achieved, it was followed by post hoc  analysis (Student-Newman-Keuls test). Statistical analysis was performed using SPSS, version 11.5 (SPSS, Chicago, IL), and significance was accepted at P < 0.05.

Values of all variables investigated were comparable at baseline and immediately after lung injury (time 0). Also, the amount of OA needed to induce lung injury did not differ significantly among groups (P = 0.64). After the recruitment maneuvers, PEEP had to be increased in every animal to optimize the lung elastance (table 1). In animals treated with PLV, minimal elastance was obtained at PEEP values higher than in animals treated with perfluorohexane (P < 0.05) but comparable to GV. Animals treated with 5% perfluorohexane needed the lowest PEEP levels, which were significantly different from GV and PLV groups (P < 0.05).

Administration of OA resulted in a significant decrease of Pao²/Fio²in all groups at time 0 as compared with baseline (P < 0.05; fig. 2). At time 2, Pao²/Fio²values decreased further (P < 0.05) but were comparable among groups. Two hours after lung recruitment and optimization of PEEP (time 4), Pao²/Fio²values were higher than at time 0 and time 2 (P < 0.05) but were not significantly different among groups. By end of the observation period, improvement in oxygenation was still significant with respect to time 2 (P < 0.05), but Pao²/Fio²values of PLV and perfluorohexane groups were not significantly different from GV. After lung injury, Paco²increased and arterial blood pH decreased significantly (P < 0.05; table 2). A further increase in Paco²could be observed at times 2, 4, and 6 as compared with time 0 (P < 0.05). However, Paco²values at time 4 and 6 were comparable to those at time 2. No statistical significant differences among therapy groups and GV could be detected at any time point.

Animals showed a significant increase in Q̇^s/Q̇^tand decrease in Sv̄o²at time 0 (P < 0.05) as a result of the administration of OA (table 2). A further deterioration in Q̇^s/Q̇^t, but not Sv̄o², was observed at time 2 (P < 0.05). After the recruitment maneuver and optimization of the ventilation according to the pulmonary mechanics, both Q̇^s/Q̇^tand Sv̄o²improved significantly as compared with time 2 (P < 0.05), but in a similar fashion in PLV and perfluorohexane groups when compared with GV.

Administration of OA led to a significant increase in E^L(P < 0.05; table 3), and a slight deterioration was observed 2 h thereafter (P < 0.05). After recruitment of the lungs and increase of PEEP, E^Limproved significantly (P < 0.05) and remained at lower levels up to time 6. In contrast, E^CWshowed no statistical significant changes throughout the experiments (table 3). Elastance parameters of PLV and perfluorohexane groups did not differ significantly from GV at any time point.

After recruitment maneuvers (time 4 and 6), R^Lshowed a significant improvement (P < 0.05), but values remained higher than immediately after establishment of lung injury (time 0; P < 0.05). Compared with GV, PLV led to a consistent increase of R^Lafter time 2 toward the end of the observation period (P < 0.05). Also, treatment with 10% perfluorohexane was associated with higher R^Lvalues than GV at time 6 (P < 0.05). R^CWdecreased slightly after administration of OA (P < 0.05), but values did not change significantly thereafter. In addition, PLV and perfluorohexane groups did not differ significantly from GV at any time point.

Functional residual capacity decreased significantly after lung injury (P < 0.05), and 2 h thereafter, a further deterioration could be observed (P < 0.05; table 3). After recruitment of the lungs, FRC values increased significantly as compared with times 0 and 2 (P < 0.05) and remained at higher levels toward the end of the observation period. Although different PEEP levels among groups were required to stabilize the lungs (table 1), FRC values of perfluorohexane groups did not differ significantly from GV.

Mean arterial pressure and mean pulmonary arterial pressures values increased significantly after induction of lung injury by OA and were accompanied by an impairment of cardiac output (P < 0.05; table 4). Mean arterial pressure values decreased slightly but significantly at time 2 (P < 0.05), whereas mean pulmonary arterial pressure and cardiac output did not change as compared with time 0. After PEEP was set to optimize the mechanics of the respiratory system, mean arterial pressure and mean pulmonary arterial pressure values decreased (P < 0.05), but no change could be detected in cardiac output. However, PLV and perfluorohexane groups did not differ significantly from GV at any time point.

Results of histologic analysis are summarized in table 5. Perfluorohexane at 5% and 10%, but not 2.5% perfluorohexane or PLV, significantly reduced DAD scores as compared with GV (P < 0.05). Animals treated with 5% perfluorohexane achieved the lowest score compared with all other concentrations tested, PLV, and GV. Accordingly, analysis revealed that most features were improved with 5% perfluorohexane as compared with GV, with the exception of microatelectasis and overdistension. Also, 10% perfluorohexane importantly reduced intraalveolar edema and epithelial damage as compared with control animals. PLV led to less intraalveolar edema, hemorrhage, and inflammatory infiltration than in controls (P < 0.05). However, overdistension was more evident with PLV than in other groups, and microatelectasis was more pronounced with the liquid perfluorocarbon than with 5% perfluorohexane and GV.

The most important finding of this study was that the use of 5% perfluorohexane vapor permitted stabilization of the lungs at lower PEEP levels during the open lung approach and was associated with improved histologic scores as compared with PLV and GV. Despite such beneficial effects, neither the vapor nor the liquid perfluorocarbon led to further improvements in gas exchange and pulmonary mechanics during conventional ventilation or the open lung approach. These issues will be discussed systematically.

Our finding that vaporized perfluorohexane and PLV did not improve gas exchange during conventional ventilation is in disagreement with previous reports in the literature.11,12,26–29A possible explanation for this discrepancy lies in the protocol for induction of lung injury. We administered OA in relatively small aliquots during a period of 1–2 h, which may lead to a more severe injury. We chose this regimen to avoid spontaneous recovery and because the slow onset of injury should mimic the clinical picture more closely. It could also be argued that the lack of effect of perfluorohexane vapor on gas exchange lies in the concentrations used. The first works showing improvement in oxygenation with this modality used concentrations as high as 18%,11,13whereas we worked with 10%. We decided to test a lower range because a more recent investigation on dose-dependent effects performed by one of the authors (J.-U. B.) suggested that concentrations far below 18%, namely 4.5% and 9%, are associated with better protection against the development of lung injury.14Although it cannot be completely excluded that higher concentrations might have been beneficial in the current study, the observation that none of the parameters investigated showed a linear relation with the concentration used makes it unlikely that 18% perfluorohexane would have been beneficial. In fact, Kemming et al.  30could not detect a positive effect of 18% perfluorohexane on gas exchange in lung injury induced by intravenous endotoxin.

Another possible explanation for the lack of positive effects lies in the mechanical ventilation protocol used in our work. After a short period of 2 h, lungs were recruited according to the open lung approach, which per se  is able to improve gas exchange dramatically and may have masked possible beneficial effects of perfluorocarbons. In fact, an increase in Pao²could be observed in some of the animals of the PLV group immediately after the beginning of that therapy with the Paratrend 7+® device (data not shown) but lasted only a few minutes. It is noteworthy that most previous experimental studies dealing with vaporized perfluorohexane and PLV have not used respiratory strategies in their control groups, which are common during ventilation of patients with acute lung injury. Fujino et al.  31compared the effects of PLV versus  GV with PEEP set 1 cm H²O above the lower inflection point in the static pressure–versus –volume curve and observed that oxygenation increased in a similar fashion in both groups. In our study, we also observed a dramatic increase in Pao²/Fio²after recruitment was performed and PEEP was set to minimize elastance, which is consistent with previous clinical and experimental studies.32,33Nevertheless, oxygenation was comparable between PLV and the control group in our study too. In contrast to Fujino et al. ,31however, we did not observe a decrease in Paco²values in the PLV group as compared with GV. This discrepancy is explained by differences in the ventilatory strategies used. Fujino et al.  31worked with pressure-controlled ventilation, and lower PEEP values were necessary to stabilize the lungs in their PLV group. This resulted in higher driving pressures and V^Tvalues in the PLV than in the GV group.31In our study, animals were ventilated in a volume-controlled mode and, therefore, V^Tvalues remained constant.

In contrast to other reports in the literature,29,34PLV combined with relatively low PEEP values was not able to improve E^Lin our animals. Accordingly, filling the lungs with 30 ml/kg perfluorooctane did not result in significant changes of E^L. A similar value was observed in the groups treated with vaporized perfluorohexane.

The improvement of E^Lvalues after recruitment of the lungs is consistent with previous clinical and experimental studies.32,35,36Such effect is probably mediated by accommodation of edema in fluid-filled alveoli of lung-dependent regions.36Unfortunately, however, the use of vaporized perfluorohexane or PLV did not add any benefit to elastance parameters.

The finding that minimal elastance in animals treated with PLV was not different from GV suggests that the dynamic pressure–versus –volume curve was not shifted to the left during PLV, as reported by Fujino et al.  31This discrepancy is most probably explained by different methods used to optimize the respiratory system mechanics. Fujino et al.  31analyzed the static pressure–versus –volume curves, whereas we decreased PEEP from 20 cm H²O to 5 cm H²O in steps of 5 cm H²O and measured the elastance at each PEEP level. The definition of lower inflection point is somewhat arbitrary, but it has been usually related to the point, or zone, in the pressure-versus -volume curve above which elastance is lowest. Accordingly, Hickling37suggested that a decremental PEEP trial, as used in our work, is useful to determine the PEEP needed to maintain the lungs open. It is worth noting that the group treated with 5% perfluorohexane required the lowest PEEP levels for achieving minimal elastance, suggesting that this concentration was able to reduce the pressure needed to stabilize the lungs.

In contrast to elastance, resistance was influenced by administration of the perfluorocarbons. In a rigid tube, resistance is proportional to the viscosity of the streaming medium,38which is much higher for perfluorocarbons in liquid and vapor form than for air.39,40Therefore, it is not surprising that 2 h after instillation of perfluorooctane into the lungs, R^Lbut not R^CWvalues were higher in PLV than in GV animals. The marked overall decrease in R^Lvalues which could be observed after lung recruitment most probably resulted from an increase in the cross-sectional area of the airways that occurred at higher lung volumes. However, such effect seems to have been outweighed by the higher viscosity of the streaming medium in animals treated with PLV and 10% perfluorohexane, as suggested by increased R^Lvalues in those groups.

The dynamic in the variation of FRC in our study is consistent with previous reports in the literature, which showed that the open lung concept is able to restore the lung volume after injury with OA.41,42Although vaporized perfluorohexane did not contribute to an increase FRC, PEEP values used in the 5% perfluorohexane group were lower than in control animals, suggesting that this therapy may lead to an increase in aerated lung volume.

Although PLV was associated with improved intraalveolar edema, hemorrhage, and inflammatory infiltration, these beneficial effects were outweighed by microatelectasis and overdistension, leading to relatively high DAD scores. However, the finding that both microatelectasis and overdistension occurred simultaneously in animals treated with PLV seems paradoxical. Because of the relatively high density of perfluorooctane, this substance spreads preferentially in dependent lung zones. Therefore, if the lung volume becomes higher than the instilled volume of perfluorocarbon, it is theoretically possible that PLV shifts from ventral to dorsal lung zones, following the gravity gradient. We could observe that the meniscus of perfluorooctane in the endotracheal tube disappeared after the recruitment of the lungs, despite compensation for evaporation loss. Therefore, the open lung concept probably resulted in uneven distribution of liquid and, consequently, of histologic findings in PLV animals. It could be argued that use of lower liquid perfluorocarbon dosages, e.g. , 20–22 ml/kg, as used by Fujino et al. ,31instead of 30 ml/kg, could lead to a better performance of PLV, minimizing the findings of overdistension. However, lower perfluorooctane volumes could also have led to development of more microatelectasis. Although the dosage of PLV was relatively high in our study, the meniscus in the tube disappeared after recruiting the lungs. This observation suggests that perhaps dosages higher than 30 ml/kg are necessary to completely fill the lungs with liquid perfluorocarbons during the open lung approach.

In contrast to PLV, vaporized perfluorohexane led to a reduction of most features investigated, resulting in a consistent reduction of the DAD score. Interestingly, a nonlinear, dose-dependent effect of vaporized perfluorohexane was observed, with an optimum at 5%. In fact, treatment with 5% perfluorohexane proved superior to 10% perfluorohexane and PLV with regard to the potential of reducing lung injury. Curiously, this optimum represents approximately the same concentration that was associated with best protection against development of lung injury after a challenge with Ca-ionophore A23187 in the isolated and perfused rabbit lung.14 

It was beyond the scope of this study to investigate how the perfluorocarbons used could have contributed to lung protection in our animals. Possible mechanisms that could explain the effects of PLV include reduction of the inflammatory response43and stabilization of alveoli at end-expiration, which may result in better distribution of mechanical stress across lung zones and minimization of further lung injury of mechanical ventilation.44Perfluorohexane also pursues antiinflammatory and anticoagulatory properties that could explain, at least in part, our results.45In addition, vaporized perfluorohexane seems to have the potential to attenuate ventilator-induced lung injury by unknown mechanisms, as demonstrated recently by our group.46 

The use of PLV has been practically abandoned after clinical trials failed to demonstrate any beneficial effect of administration of perfluorocarbons as liquid during acute lung injury. Our results demonstrated that although the therapy with another administration form, namely vaporization, was superior to PLV to improve lung structure, such effect was not accompanied by improved functional parameters, i.e. , gas exchange and respiratory system mechanics, except to a reduction of the pressures needed to stabilize the lungs in the OA model of lung injury. Because the observation period was limited to 6 h, we do not know whether the improved histologic findings would prove beneficial in the long term. This issue remains speculative and requires that further studies with vaporized perfluorohexane be performed.