Introduction

Cyclic opening and closing of atelectatic alveoli with tidal breathing is known to be a basic mechanism leading to ventilator-induced lung injury [1]. To prevent alveolar cycling and derecruitment in acute lung injury, high levels of PEEP have been found necessary to maintain normal functional residual capacity and to counterbalance the increased lung mass resulting from edema, inflammation, and infiltrations [2, 3]. Therefore the application of high levels of PEEP is recommended [4] and is considered standard care for the patient with acute respiratory distress syndrome. However, recent data reported by Crotti and coworkers [5] suggest that recruitment and derecruitment are not identical in the presence of lung hysteresis, and that "threshold closing pressure" is substantially lower than "threshold opening pressure," which may put into question the routine use of high levels of PEEP.

Pulmonary extravascular thermal volume has been used in numerous clinical and experimental studies to assess extravascular lung water (EVLW) [6, 7, 8, 9]. EVLW is an established marker for the severity of lung injury [10, 11], and management based on EVLW measurement may even improve outcome in some critically ill patients [6]. The effect of PEEP on EVLW has been the subject of numerous studies, which have yielded conflicting results: increase [12, 13], decrease [14, 15, 16, 17], or no change [18, 19, 20]. It has not been shown, however, whether the changes in EVLW are actually correlated with the changes in edema and nonaerated lung volume, making EVLW a useful bedside marker of the amount of diseased tissue and lung recruitment. Therefore the aim of the present study was to test whether (a) EVLW is a valid marker of the severity of lung injury assessed by the increase in tissue mass, and (b) whether variations in EVLW with PEEP are useful indicators of lung recruitment estimated as a decrease in nonaerated lung tissue.

Methods

Animal preparation and measurements

After approval by the District Governmental Commission on the Care and Use of Animals 15 anesthetized sheep of mixed breed weighing 41.5±6.5 kg were used in the study. The animals were premedicated with xylazine hydrochloride (Xylazin, Sanofi-Ceva, Germany) at 5 mg intravenously and anesthetized with ketamine (Ketanest, Parke-Davis) at 6 mg/kg intravenously and midazolam 0.15 mg/kg (Dormicum, Roche). Anesthesia and muscle relaxation were maintained by continuous infusion of ketamine (10 mg/kg per hour), midazolam (1 mg/kg per hour), and pancuronium bromide (0.12 mg/kg per hour) throughout the experiment. The sheep were intubated with a 9-mm inner-diameter cuffed endotracheal jet tube (Hi-Lo Jet, Mallinckrodt Medical, St. Louis, Mo.,USA) and were ventilated with a Siemens Servo Ventilator 300 (Siemens-Elema, Solna, Sweden) in the volume-controlled mode with a positive end-expiratory pressure (PEEP) of 5 cmH2O, an inspiratory to expiratory ratio of 1:1 and a FIO2 of 1.0. A tidal volume of 12 ml/kg and a respiratory rate of 12–14 breaths/min were applied to maintain a PCO2 value within the range of 35–40 mmHg. A continuous infusion of lactated Ringer's solution was given at a rate of 5 ml/kg per hour. Gastric emptying was achieved using a large bore orogastic tube, which was removed thereafter.

Central venous and pulmonary artery pressures were measured using a 7.5-F flow-directed thermodilution fiberoptic pulmonary artery catheter (Opticath, Abbott, North Chicago, Ill., USA) advanced via the left internal jugular vein. The right femoral artery was cannulated with a 6 French percutaneous sheath (Super Arrow-Flex, Arrow, Reading, Pa.,USA). Continuous electrocardiographic monitoring was performed. For hemodynamic monitoring a Sirecust 1281 monitor (Siemens Medical Electronics, Danvers, Mass., USA) and Novotrans II (Medex, Hilliard, Ohio, USA) pressure transducers referenced to atmospheric pressure at midthorax level were used.

EVLW was measured via the thermal-green double indicator dilution technique using the COLD Z-021 computer (Pulsion Medizintechnik, Munich, Germany) [21]. With the COLD Z-021 the dye and thermal dilution were measured using a 4-F fiberoptic catheter (Pulsiocath, Pulsion) advanced through the 6-F sheath in the femoral artery into the descending aorta. EVLW was determined by injecting 10 ml 5% glucose solution containing 1 mg/ml indocyanine green at a temperature between 0° and 10°C into the right atrium via the proximal lumen of the pulmonary artery catheter. EVLW was calculated as: EVLW=cardiac output× (MTTthermo−MTTdye) [where MTT=3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide].

The occlusion method was used oo measure the elastance of the respiratory system and to partition it into its pulmonary and chest wall components [22]. Distal tracheal pressure (Ptrach) was measured by connecting an air-filled pressure transducer (Novotrans II, Medex) to the additional lumen of the Hi-Lo jet tube ending at the tip of the tube. The pressure signals were sampled on-line at a rate of 50 Hz using a Signal Conditioning System (SCXI, National Instruments, Austin, Tex.,USA). Expiratory tidal volume was obtained from the Servo Ventilator. Esophageal pressure (Pes) was measured with a thin-walled latex balloon sealed over one end of a polyethylene catheter (International Medical, Zutphen, The Netherlands) inflated with 0.5 ml air. The catheter was positioned at the lower one-third of the esophagus, as confirmed by computed tomography. The validity of Pes was verified by the occlusion test [22].

Lung lavage procedure

With the sheep in supine position the endotracheal tube was disconnected from the ventilator, and warmed saline was instilled from a height of 50 cm until a meniscus was seen in the tube. The fluid was retrieved via gravity drainage after 45 s of apnea followed by endotracheal suctioning. Instilled and retrieved volumes were measured. Between the lavages the sheep were manually ventilated with an FIO2 of 1.0 using an AMBU self-inflating bag. The lavage process was repeated until adequate impairment of gas exchange (defined as PaO2 <100 mmHg 15 min following the last lavage) and respiratory mechanics were achieved. After lung lavage further lung injury was established by ventilating the sheep with zero end-expiratory pressure (ZEEP) for 60 min, thereby inducing ventilator-induced lung injury.

Lung imaging and image analysis

The Imatron C-150XP electron beam computed tomography scanner (Imatron, San Francisco, Calif., USA) was used for the study. Scans of the whole lungs were performed within a 14- to 17-s end-expiratory hold for baseline and every level of end-expiratory pressure after injury during an expiratory hold maneuver (collimation 3 mm, table feed 4–5 mm, reconstruction interval 2–2.5 mm, scan length 28–35 cm, 130 kV, 128mAs, exposure time 200 ms/image). The cross-sectional images of the lungs were analyzed using a custom made program based on the graphic programming language G (Lab-View 5.1, National Instruments). The images were divided into square pixels (size 0.46 mm2) and the Hounsfield attenuation unit (HU) of each pixel was computed. After multiplying the square pixels by the slice thickness the whole lung volume was divided into voxels (volume 1.84 mm3). For further analysis the total Hounsfield attenuation range from −1000 to +300 HU was subdivided into 100 compartments spanning 13 HU, and each voxel was assigned to the corresponding compartment according to its HU value. Absolute lung volumes for each compartment were computed by adding the corresponding voxels. Total lung volume was obtained by adding the lung volume of each compartment. To further differentiate the total lung volume according to its degree of aeration, the total HU range was subdivided into four subranges: lung parenchyma showing densities ranging from −1000 to −900 HU was classified as overinflated, −900 to −500 HU as normally aerated, −500 to −100 as poorly aerated, and −100 to +300 HU as nonaerated (atelectatic).

As the mean computed tomography (CT) number of a given lung volume is correlated with the proportion of gas and tissue in this lung volume, gas and tissue volumes can be calculated, as described by Vieira et al. [23]. In the first step the volume of gas and tissue for each compartment of 13 HU was computed as: gas volume=volume×CT/−1000, and tissue volume=volume×(1−CT/−1000), where CT is the mean CT number of the compartment analyzed. In the second step the volume of gas and volume of tissue for the whole lung were calculated by adding the values of the volumes of gas and the volumes of tissue obtained for each compartment of 13 HU.

Experimental protocol

The sheep were intubated and placed supine in the CT scanner. After instrumentation was completed baseline measurements were obtained and computed tomographic images of the lung were obtained as described above. Thereafter lung injury was induced by saline washout. The animals were ventilated in the volume-control mode with an inspiratory to expiratory time of 1:1, an FIO2 of 1.0, and frequencies between 12–20/min depending on the PCO2 value. Peak inspiratory pressure was limited to 45 cmH2O. Following 60 min of ventilation with zero PEEP to establish the lung injury another set of CT images were obtained, along with the complete set of measurements made at baseline. PEEP was increased in 60-min intervals to 7, 14, and 21 cmH2O to recruit the collapsed parts of the lung. Before each PEEP increment another set of CT images and measurements were obtained. After completion of the protocol the animal were killed with high-dose thiopental (30 mg/kg) followed by potassium chloride.

Statistical analysis

All values are reported as mean ±SD unless otherwise specified. One-way analysis of variance was used to test the significance of differences between baseline conditions before and various PEEP levels after induction of lung injury followed by multiple comparison with Bonferroni's correction. Statistical significance was accepted at p<0.05. Linear regression analysis was used to evaluate the relationship between PEEP-induced changes in EVLW and nonaerated lung volumes and nonaerated lung volumes and transpulmonary shunt. Statistical software package SAS 6.12 (SAS Institute, Cary, N.C., USA) was used for analysis.

Results

Effect of lung lavage-induced lung injury

Lung injury resulted in a substantial decrease in PaO2, an increase in transpulmonary shunt and an increase in static lung elastance (Table 1). All animals survived the experimental protocol. The volume of saline instilled was 6,874±1,443 ml. The volume of saline retrieved was 6,596±1,532 ml, resulting in a positive balance of 278±288 ml. EVLW significantly increased from 9±2 to 36±9 ml/kg (p<0.001). Total lung volume remained constant 52±7 vs. 50±8 ml/kg, while lung tissue volume and nonaerated lung volume increased (21±3 to 37±5 ml/kg and 4±2 to 30±5 ml/kg, p<0.001, Table 2)

Table 1. Respiratory parameters pre- and postinjury (Ppeak peak inspiratory pressure, Pplat inspiratory plateau pressure, EVLW extravascular lung water, Q S /Q T transpulmonary shunt, E stat rs static elastance of the respiratory system, E stat cw static elastance of the chest wall, E stat L static lung elastance)
Table 2. Hemodynamic parameters pre- and postinjury (MAP mean arterial pressure, MPAP mean pulmonary artery pressure, PCWP pulmonary capillary wedge pressure, CVP central venous pressure, ITBV intrathoracic blood volume, CO COLD cardiac output determined by COLD, CO PAC cardiac output determined by PAC)

Effects of PEEP on EVLW, lung volumes, gas exchange, and respiratory mechanics

The main respiratory and hemodynamic parameters are summarized in Tables 1 and 2. Stepwise increase in PEEP to 7, 14, and 21 cmH2O significantly reduced EVLW (30±7, 24±8, and 18±4 ml/kg, p<0.001 vs. ZEEP). While total lung volume increased (50±8 ml/kg at ZEEP to 77±12 ml/kg at PEEP 21, p<0.001; Table 3), lung tissue volume remained constant. Nonaerated lung volume decreased (30±6 ml/kg at ZEEP to 8±4 ml/kg at PEEP 21, p<0.001). PEEP induced alveolar recruitment by decreasing nonaerated lung volume as can be seen by the shift to the left on the density histogram curves (Fig. 1). At a PEEP level of 21 cmH2O the histogram curve was almost completely shifted back to its baseline shape. At the highest level of PEEP a small but significantly overdistended lung volume was detected (0.2±0.1 ml/kg at ZEEP to 0.9±0.7 ml/kg at PEEP 21, p<0.001; Table 3)

Table 3. Expiratory lung volumes measured by electron beam computed tomography
Fig. 1.
figure 1

Lung density histogram curves pre- and postinjury at different levels of PEEP. Standard deviations are deleted to improve readability

Increasing PEEP improved oxygenation and reduced transpulmonary shunt without compromising oxygen delivery (Table 1). The moderate hypercarbia after injury was well tolerated by all animals and did not change with increasing levels of PEEP. Estat rs and Estat L significantly increased with lung injury (p<0.0001) and slowly decreased with increasing PEEP. Estat cw did not change significantly either with injury or with changes in PEEP.

PEEP-induced changes in EVLW (ΔEVLW, calculated as EVLW at PEEP 7, 14, and 21 cmH2O minus EVLW at ZEEP) showed a significant correlation to changes in nonaerated lung volume (r=0.67, p<0.0001, Fig. 2A). An even more significant correlation, however, was found for PEEP-induced changes in transpulmonary shunt (Δ Qs/Qt) and changes in nonaerated lung volume (r=0.83, p<0.0001; Fig. 2B).

Fig. 2.
figure 2

Correlation between PEEP-induced changes in nonaerated lung volume (ml/kg) and extravascular lung water (ml/kg) and between PEEP-induced changes in nonaerated lung volume (ml/kg) and transpulmonary shunt (%)

Discussion

Saline lung lavage, an injury model first described by Lachmann and coworkers [24] and widely used thereafter, significantly increased EVLW. It should be noted, however, that part of the initial increase in EVLW may be attributed to saline remaining in the lung after the washout procedure. As we applied ventilation with zero PEEP for 1 h and low PEEP (7 cmH2O) for another, however, a significant part of EVLW accumulation may have been due to an inflammatory response caused by the injurious ventilatory pattern.

In this model of surfactant-washout lung injury the stepwise increase in PEEP effectively reduced EVLW, confirming the findings of former studies. Russel and coworkers [15] in an oleic acid edema model found that EVLW was reduced when 10 cmH2O of PEEP was applied. Colmenero-Ruiz et al. [16], using the same model in pigs, showed that early application of 10 cmH2O of PEEP reduced EVLW, and that lowering tidal volumes from 12 to 6 ml/kg reduced EVLW even further. In another study [17] the same group observed that PEEP reduced EVLW in a time-dependent manner. While our findings of decreasing EVLW with increasing levels of PEEP are in accord with these data, the CT data obtained simultaneously raise concerns about the validity of the thermal dye dilution measurements in the model studied. Saline washout induced lung injury was characterized morphologically by a marked excess in lung tissue (from 21±3 preinjury to 37±5 ml/kg postinjury, p<0.001). Assuming that pulmonary blood pooling was unlikely because of generalized constriction of the pulmonary circulation and constant amounts of intrathoracic blood volume (Table 1), it can be hypothesized that this increase in lung tissue volume resulted mainly from the presence of an excess amount of EVLW and inflammatory cells [25]. This volume, representing increased lung mass due to edema and infiltrations, is prone to recruitment/derecruitment phenomena, which renders the lung at risk for ventilator-induced lung injury [1]. Protective ventilatory approaches should therefore focus on that "critical volume," and changes with therapeutic interventions (e.g., PEEP) should be detected by bedside monitoring such as the COLD system. In our model, however, a striking disparity between PEEP-induced changes in EVLW and lung tissue volume was found. While EVLW steadily decreased with each increment in PEEP, the amount of excess lung tissue volume remained constant for all levels of PEEP studied. PEEP significantly increased total lung volume, decreased nonaerated lung volume, and increased gas volume. Similar results were also observed in ARDS patients studied by the CT Scan ARDS Study group [26]: According to Malo and coworkers [20], PEEP reduces Qs/Qt by inflating previously flooded and collapsed air spaces (i.e., alveolar recruitment) and by redistributing excess alveolar water into the compliant perivascular space, thus alleviating the obstacle to pulmonary O2 transfer. Looking at the volume data and the changes in the density histogram curves, we conclude that in addition PEEP improves gas exchange by reestablishing "normal" gas to tissue volume ratios. Since at least in this model PEEP does not change the excess tissue volume, the "normalization" of the gas to tissue volume ratio is achieved by a significant increase in gas volume and subsequently in total lung volume compared to baseline. Although these changes raise concern about potential overinflation, we did not observe substantial increases in overinflated lung volume. This finding, however, must be interpreted with caution due to the methodological restrictions of CT discussed in detail elsewhere [26]. While all the CT findings are well in accord with existing data, the surprising finding of this study was that, in spite of constant excess tissue volume, PEEP induced a steady decrease in EVLW measured via thermal-green dye double-indicator dilution. Theoretically, EVLW determined by thermal dye dilution, should follow the course of excess tissue volume, as this excess mass increases the pulmonary extravascular thermal volume. The most likely explanation of why this was not the case is based on methodological considerations: since the indicators are blood borne, the accumulation of extravascular water cannot be detected in poorly perfused lung regions (a) as a result of using PEEP, (b) because of significant intravascular obstruction from emboli or in situ thrombosis (as has been shown to occur in ARDS), or (c) because severe hypoxia from vasoconstriction from hypoxia is present [27, 28, 29]. With vasoconstriction the time required for full recovery of the thermal indicator from the lung may be so prolonged that recirculation artifacts are likely. Vascular obstruction (from PEEP or thrombosis) may cause errors because the thermal indicator cannot equilibrate within the extravascular space if it is not delivered sufficiently close to reach this space by diffusion [27, 29].

Bedside assessment of PEEP-induced alveolar recruitment is of utmost importance in optimizing ventilator settings. As exact quantification of recruited lung volume is confined to CT analysis, and the role of the pressure-volume curve as a tool to set positive end-expiratory pressure is still arguable [4, 5], the present study focused especially on the suitability of bedside parameters for estimating recruitment. In this respect EVLW proved to be a valuable tool for monitoring PEEP-induced decrease in nonaerated lung volume, but it was not found to be superior to transpulmonary shunt. Bearing in mind the lack of correlation between EVLW and increased lung tissue mass, however, we conclude that measurement of EVLW in the setting of acute lung injury should be interpreted very cautiously unless the validity and limitations of this method in the setting of acute lung injury can be defined more precisely.

The finding that changes in transpulmonary shunt showed an excellent correlation with changes in nonaerated lung volume is not surprising. As all animals were ventilated with an FIO2 of 1, impairments in arterial oxygenation at ZEEP and improvement in oxygenation parameters observed with application of PEEP were dependent solely on pulmonary shunt (nonaerated lung volume) and not on venous admixture (poorly aerated lung volume). That fact deserves some comments, however, as it is in contrast to the findings of Malbouisson et al. [30]. Their group, using our method of describing alveolar recruitment (termed Gattinoni's method, but confined to a single juxtadiaphragmatic section), did not find a correlation between PEEP-induced recruitment and changes in PaO2. They described a new method in which PEEP-induced alveolar recruitment was computed as the volume of gas penetrating in poorly and nonaerated lung regions following PEEP and found a significant correlation with changes in PaO2 (r=0.76). This method, however, while appealing and well supported by changes in density histogram curves such as those shown above, is quite cumbersome and is limited by the fact that separation of PEEP-induced alveolar recruitment occurring in poorly and nonaerated lung regions is not possible [30]. By applying the relatively simple "Gattinoni's method" to the entire lung, thereby taking into account the cephalocaudal gradient in the diseased lung, we found a similar degree of correlation (r=0.83), leaving this method a valid and easy tool to assess PEEP-induced alveolar recruitment.

In summary, increasing levels of PEEP effectively decrease EVLW measured via thermal dye dilution in saline washout induced lung injury. This decrease is closely correlated with the decrease in nonaerated lung volume, making EVLW a potentially valuable bedside parameter indicating alveolar recruitment, being as good as measurements of transpulmonary shunt. As the excess tissue volume did not change with increasing PEEP, however, EVLW may not be a valid marker of the overall severity of lung disease.