INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A number of studies have recently demonstrated the ability of liquid ventilation to improve gas exchange and pulmonary function in a number of animal species experiencing acute respiratory failure (1). Liquid ventilation can either be performed by total liquid ventilation (TLV), in which a device is utilized to ventilate with perfluorocarbon the previously perfluorocarbon-filled lung, or as partial liquid ventilation (PLV), in which a conventional gas mechanical ventilator is used to ventilate with gas the partially perfluorocarbon-filled lung (2, 5, 10). Partial liquid ventilation offers the advantage that a specialized device is not necessary to apply this technique in the clinical setting. The first human experiences appear to demonstrate that gas exchange is increased during PLV in neonatal, pediatric, and adult patients with respiratory insufficiency (13). However, the mechanisms by which gas exchange is improved during PLV remain unclear. It has been previously demonstrated that atelectatic, dependent lung regions are reinflated during TLV (19). It has also been shown with plain radiography (20, 21) that the relatively dense perfluorocarbons are distributed specifically to the dependent regions of the lungs in both animal models and patients with adult respiratory distress syndrome (ARDS). However, the relative distribution of gas and perfluorocarbon during PLV has not been evaluated. The present study was, therefore, performed to evaluate the relative distribution of perfluorocarbon and gas during PLV in an oleic-acid animal model of respiratory failure.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Induction of anesthesia was performed in 14 sheep weighing 64.9 ± 6.4 kg with 250 mg ketamine, followed by a continuous infusion of ketamine (4 mg/kg/h) and midazolam (0.15 mg/kg/h). After conventional intubation, a midline neck incision was performed and the trachea isolated and cannulated with a 10-mm inner-diameter endotracheal tube (Hi-lo; Mallinckrodt Laboratories, Athlone, Ireland). The tracheostomy was specifically performed to provide a stable airway during the sheep's transport between the laboratory and the CT scanning suite. Gas mechanical ventilation was initiated and a bolus injection of 4 mg pancuronium was administered, followed by 0.1 mg/kg/h continuous infusion of pancuronium. Animals were volume-control ventilated with a Servo 900C (Siemens Elema, Solna, Sweden). A respiratory rate of 12 breaths/min and a tidal volume of 12-15 mg/kg was applied to maintain a Pa_CO2 value within the range of 35-40 mm Hg. A 16-gauge catheter (Insyte-W; Becton Dickinson, Franklin Lakes, NJ) was placed into the right carotid artery and a 7.5 French balloon-tipped catheter (Opticath, Abbott Laboratories, North Chicago, IL) was advanced into the pulmonary artery under transduced pressure guidance via the right internal jugular vein. For hemodynamic monitoring, a Sirecust 1281 monitor (Siemens Medical Electronics, Danvers, MA) and Novotrans II (Medex Inc., Hilliard, OH) pressure sensors were used. All hemodynamic parameters and ventilator variables were continuously recorded using the graphical programming software Labview^TM (National Instruments, Austin, TX). An orogastric tube was placed to reduce abdominal distension. The bladder was percutaneously cannulated with a CH 10 catheter (Cystofix, Braun Melsungen, Germany).

Experimental Protocol

After instrumentation, the sheep were placed supine in the CT scanner. A baseline chest CT scan was performed and baseline physiologic data, including systemic and pulmonary pressures and ventilator pressures and settings, were recorded on-line every minute. Baseline cardiac output (CO) and arterial and venous blood-gas data were also obtained. Lung injury was then induced by intravenous administration of 0.15 ml/kg oleic acid (C₁₈H₃₄O₂; Mallinckrodt Specialty Chemical Co., Paris, KY) added to 10 ml of aspirated blood in a 20-ml syringe. The oleic acid was emulsified in the blood by vigorous shaking. This blood was then injected, with continued shaking, into the right atrium through the proximal port of the pulmonary artery catheter over a 20-min period. The FI_O2 was increased to 1.0 and a positive end-expiratory pressure (PEEP) or 5 cm H₂O was applied. Hetastarch 10% at 10 ml/kg was infused intravenously during administration of oleic acid. Subsequently, animals received 10 ml/kg lactated Ringer's solution per hour for maintenance fluid. Initial tidal volume and respiratory rate remained unchanged. Blood-gas analysis, CO measurements, and physiological shunt (Q_p/Q_t) calculations were performed every 15 min. Bicarbonate was administered at a dose of 1 mEq/kg when the calculated negative base excess exceeded 5.0 mmol/L. The presence of arterial hypoxemia (Pa_O2 < 50 mm Hg with FI_O2 = 1.0), increased alveolar-arterial oxygen gradient (AaDO₂ = 600 mm Hg), and a measured mixed venous oxygen saturation (Sv_O2) = 50% were used to indicate severe respiratory failure and induction of lung injury. All three criteria were met by all animals. Once these criteria were surpassed, an "injury" chest CT scan (representing ARDS) was performed. Animals were then randomized to management with GV (n = 7) or PLV (n = 7) with perflubron (LiquiVent; Alliance Pharmaceutical Corp., San Diego, CA) for 90 min. Physiologic data as outlined previously and chest CT scans were obtained at baseline, after induction of lung injury, and after instillation of cumulative doses of 10 ml/kg, 20 ml/kg, and 30 ml/kg in the PLV animals. Data were gathered at baseline, after induction of lung injury, and following completion of the 90-min experimental period in the GV group. After the 90-min study period, all animals were killed with 30 mg/kg of high-dose thiopentone and potassium chloride.

Partial Liquid Ventilation

In the PLV group, gas ventilation was continued with ventilator settings of 12 breaths/min, tidal volumes of 12-15 ml/kg, PEEP = 5 cm H₂O, and FI_O2 = 1.0. Ten ml/kg perflubron was administered at 30-min intervals by transiently disconnecting the ventilator and draining perflubron from a reservoir into the endotracheal tube over 20 to 30 s. Chest CT scans were obtained at baseline, following induction of lung injury, and 20-30 min after each 10 ml/kg dose.

Gas Ventilation

Gas-ventilated animals were continued on conventional mechanical ventilation with settings of 12 breaths/min, tidal volumes of 12-15 ml/ kg, PEEP = 5 cm H₂O, and FI_O2 = 1.0. Chest CT scans were obtained at baseline, after induction of lung injury, and at the end of the 90-min experimental period.

Computed Tomography Imaging

Animals were placed supine in a General Electric 9800 Quick high-light (General Electric, Paris, France) CT scanner throughout the study. Cross-sectional images were performed using the following scanning parameters: 120 kV, 170 mA, and a 2-s scan time. A reconstruction filter appropriate for bone was used for viewing. The scanner was calibrated so that the density of air was 1,000 Hounsfield units (HU) and that of water was 0 HU. The field of view accommodated the size of the lung. After an initial scout image, five cross-sectional scans were defined and located at the apex of the lung (A), at the hilum (C), halfway between the apex and hilum (B), 1 cm superior to the diaphragm (E), and halfway between the hilum and diaphragm (D). A slice thickness of 1.5 mm was used. Each CT sequence was performed during either an end-inspiratory or end-expiratory hold. The end-inspiratory and end-expiratory hold was achieved using the appropriate hold option of the Servo 900C ventilator. A total of 30 cross-sectional scans in GV animals (5 cross-sectional scans × 2 [inspiration and expiration] × 3 [baseline scan, ARDS scan, 90-min postinjury scan]) and 50 cross-sectional scans in PLV animals (5 cross-sectional scans × 2 [inspiration and expiration] × 5 [baseline scan, ARDS scan, perflubron 10 ml/kg scan, perflubron 20 ml/kg scan, perflubron 30 ml/ kg scan]) were stored on magneto optical disc for further analysis.

Analysis of the Computed Tomography Scans

As part of this study, we initially evaluated the CT number of perflubron, which was 2,292 ± 49. Computed tomography attenuation numbers from the cross-sectional images were obtained by placing a line adjacent to the ribs along the chest wall over both lungs (Figure 1). This computer-assisted line was duplicated and shifted either 6 pixels (~ 0.5 cm) or 12 pixels (~ 1.0 cm) toward the middle of the thorax. The area between the chest-wall line and the 0.5-cm medial line (small segment) and the chest-wall line and the 1-cm medial line (large segment) were defined as the regions of interest (ROIs). The ROIs were defined in this fashion in order to obtain CT attenuation numbers of lung parenchyma while avoiding all major airway and vascular structures. In order to evaluate the regional distribution of attenuation numbers in a dependent to nondependent orientation, the ROIs were divided into 10 portions of equal size. With this division, 10 small segments and 10 large segments were defined where segment 1 described the most nondependent and segment 10 described the most dependent area (Figure 1). The mean attenuation number and the corresponding standard deviation of the data were calculated for each segmental ROI. This analysis was performed using software developed at our institution specifically for this purpose.

View larger version (114K): [in this window] [in a new window]   Figure 1.   A diagram demonstrating the technique by which the cross-sectional lung image was divided into 10 ventrodorsal segments. Three lines parallel to the internal, lateral aspect of the thorax were drawn 0.5 cm apart, which allowed development of 10 small (6 pixel, 0.5 cm) and 10 large (12 pixel, 1.0 cm) dependent/ nondependent oriented segments.

Alveolar Instability Index and Gas-Liquid Index

The use of the variation in radiologic density during the respiratory cycle to quantify the regional degree of alveolar collapse was first proposed by Wegenius and colleagues (22). The principle is based on the tendency of the individual alveolus to exist either air-filled/expanded or fluid-filled/collapsed at end-expiration. To assess the degree of alveolar collapse at end-expiration, the ratio of the CT attenuation number during inspiration to the CT attenuation number during expiration is calculated. The calculated number is defined as the alveolar instability index (AIX). In an ideal, healthy lung this ratio should be above, but nearly equal to, 1; although the CT number is more negative during inspiration, both the inspiratory and expiratory CT numbers are negative. As a result, a positive number greater than 1 is observed when the ratio is formed. Lung injury induces an increase in the variation between inspiratory and expiratory attenuation numbers because affected alveoli tend to collapse during expiration, which increases the AIX value. The AIX, then, is especially useful for determining whether alveoli are collapsed (AIX > 1) or inflated (AIX = 1) at end-expiration. We calculated the AIX for the CT numbers for each of the previously described 10 ventrodorsal segments of the lung at baseline, after induction of lung injury for both groups, and 90 min after lung-injury induction for the GV animals.

The concept of the AIX may also be applied during PLV. However, the lung consists of three phases during PLV: tissue/fluid, gas, and perflubron. The perflubron is a radiopaque perfluorochemical with a density of 2,292 ± 49 HU; as a consequence, mean attenuation numbers that are usually less than zero shift to the high positive range. In contrast to the alveolar instability index, therefore, the ratio typically obtained during PLV demonstrates values less than 1 (the inspiratory HU is less positive than the expiratory HU). For the purpose of this study, we defined the ratio of the mean attenuation during inspiration and expiration with PLV as the gas-liquid index (GLX). The GLX was calculated for the PLV group following administration of 30 ml/kg of perflubron. The GLX is useful for determining the volume of gas entering the lung region at end-inspiration: a value close to 1 is predominantly perflubron-filled, while a value much less than 1 is relatively more air-filled at end-inspiration. The GLX may be used, therefore, to evaluate the relative distribution of perflubron and air during an expiratory hold. The advantage of the GLX is that the absolute CT values are normalized, or indexed, providing an improved ability to analyze data that would otherwise have wide variance.

Measurement and Calculations of Hemodynamic and Gas Exchange Variables

The arterial and venous pH, Pa_CO2, Pa_O2, and hemoglobin were measured by an ABL 520 (Radiometer Copenhagen, Copenhagen, Denmark) blood-gas analyzer that also allowed estimation of the Sv_O2 and arterial oxygen saturation Sa_O2. The AaDO₂, pulmonary capillary oxygen content (Cc_O2), mixed venous oxygen content (Cv_O2), arterial oxygen content (Ca_O2), and Q_p/Q_t were calculated using the following formulas:

where

and where PAO₂ = alveolar oxygen partial pressure (mm Hg), P_B = barometric pressure (mm Hg), and PH₂O = partial pressure of H₂O (47 mm Hg at 37° C) (19).

where Sc_O2 and Pc_O2 are the predicted pulmonary capillary saturation and O₂ partial pressure, respectively, based on the PAO₂.

Thermodilution CO measurements were determined in triplicate using a Sirecust 1281 CO monitor with the mean value of the threefold measurement demonstrated.

Data Analysis

Initial analysis failed to show any significant differences in mean CT attenuation numbers between the right and left lung, small and large ROI segments, and among the five cross-sectional levels. Therefore, the small ROI segment CT attenuation data from the right and left lungs and from the five cross-sectional levels were averaged at inspiration or expiration for the gas or liquid ventilation animals. The CT number during an expiratory hold (HU_exp) in segments 1-5 (nondependent segments) and segments 6-10 (dependent segments) were averaged and compared with the Wilcoxon signed rank test in the GV and PLV animals. This allowed definition of the relationship of HU_exp and lung dependency (1).

The alveolar instability index (AIX) was calculated at baseline and after lung injury induction in both groups and 90 min after lung injury induction in the GV animals. The gas-liquid index (GLX) was calculated for the cumulative dose of 30 ml/kg of perflubron. Due to the small number of animals per group, we assumed that the measured data were not normally distributed. Therefore, the Wilcoxon signed rank test was used to compare intragroup changes in physiologic variables and the Wilcoxon sum rank test (U test) was performed for intergroup comparisons. The Kruskal-Wallis test was used to compare more than two groups. The level of significance was fixed at p < 0.05. All data are reported as mean ± SD and median and range (24).

Approval for use and care of animals was obtained from the animal care committee before all studies.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Physiologic data for the GV and PLV animals are presented in Table 1. No differences in baseline mean Pa_O2, Sv_O2, CO, mean arterial pressure (MAP), and Q_p/Q_t between the two groups were observed. Significant increases in Pa_O2 were observed during PLV when compared to GV during the experimental period. As shown in Table 1, Pa_O2 at 90 min was 40 ± 8 for GV animals and 108 ± 60 mm Hg for the PLV group after receiving 30 ml/kg perflubron; p = 0.004. Similarly, during PLV significant and sustained reductions in physiologic shunt were noted (Q_p/Q_t = 76 ± 12% in the GV and 40 ± 12% in the PLV group at 90 min; p = 0.004). The CO remained stable in both groups. A significant increase in Sv_O2 was noted in the PLV group when compared with the GV group (Sv_O2 at 90 min: GV = 25 ± 20; PLV = 63 ± 12%; p = 0.01).

A representative cross-sectional image at a level 1 cm above the diaphragm in a lung-injured animal during PLV (30 ml/kg perflubron) with an inspiratory and expiratory hold is demonstrated in Figure 2. Mean CT numbers demonstrated no differences in CT attenuation between the GV and PLV animals at baseline (p > 0.79) or after lung injury (p > 0.89). Therefore, the CT numbers during an inspiratory and expiratory hold for each of the 10 ventrodorsal regions were combined for the GV and PLV animals at those times. These data are demonstrated in Figure 3. Alveolar inflation for the combined groups was decreased during an expiratory hold in the dependent (posterior, segment 6-10) regions of the lungs at baseline and following induction of lung injury when compared with the nondependent (anterior, segment 1-5) regions (baseline: HU_exp = 713 ± 28 for segments 1-5, HU_exp = 526 ± 56 for segments 6-10, p = 0.001; injury: HU_exp = 609 ± 57 for segments 1-5, HU_exp = 243 ± 75 for segments 6-10, p = 0.001).

View larger version (82K): [in this window] [in a new window]   Figure 2.   A representative cross-sectional image at a point 1 cm above the diaphragm with an inspiratory and expiratory hold in one animal during PLV. Left image: exam no. 2519, slice no. 46, inspiratory hold; right image: exam no. 2519, slice no. 51, expiratory hold.

View larger version (33K): [in this window] [in a new window]   Figure 3.   Mean Hounsfield attenuation number at baseline (white bars) and following induction of lung injury (black bars) in the 10 ventrodoral segments. The GV and PLV data have been combined. The top graph (E ) shows an expiratory hold, while the bottom graph (I ) reveals similar data during an inspiratory hold.

A CT attenuation number approaching zero was observed in the dependent lung regions 90 min after induction of lung injury in the GV animals (Figure 4). This indicated that minimal gas aeration was effected in these regions. In contrast, increasingly negative CT numbers were observed in the nondependent regions, indicating increased aeration. A significant difference in the HU_exp between segments 1-5 and 6-10 was noted (HU_exp = 562 ± 108 and 165 ± 104, respectively, p = 0.01). The comparison of the CT number in the nondependent and dependent regions were significantly different during expiration following the 10 and 20 ml/kg doses of perflubron (10 ml/kg: nondependent, HU_exp = 1 ± 207; dependent, HU_exp = 381 ± 59, p = 0.01; 20 ml/kg: nondependent, HU_exp = 359 ± 294, dependent, HU_exp = 647 ± 85, p = 0.03). After the 30 ml/ kg dose, the comparison of nondependent and dependent measures just failed to reach statistical significance during expiration (HU_exp = 579 ± 338 for segments 1-5, HU_exp = 790 ± 149 for segments 6-10, p = 0.07); however, the tendency toward predominant distribution to the dependent lung regions could still be observed. As may be seen in Figure 4, in general, the CT numbers became increasingly positive as the lung region was more dependent, indicating increased accumulation of perflubron at end-expiration.

View larger version (35K): [in this window] [in a new window]   Figure 4.   Mean Hounsfield attenuation number in the 10 ventrodorsal segments during GV (white bars) and PLV (30 ml/kg perflubron) (black bars). The top graph (E ) demonstrates an expiratory hold, while the bottom graph (I ) reveals similar data during an inspiratory hold.

There were no statistically significant differences between the AIX in the GV and PLV groups at baseline nor after induction of lung injury. Therefore the AIX values of both groups for baseline and lung injury have been combined (Figure 5). The AIX is always greater than 1, and a value close to 1 indicates the regional alveoli are well inflated at end-expiration. Alveolar collapse was more prominent at end-expiration in the dependent than the nondependent lung regions at baseline and after induction of lung injury (AIX: baseline, nondependent = 1.04 ± 0.02, dependent = 1.13 ± 0.06, p = 0.001; injury, nondependent = 1.12 ± 0.05, dependent 1.44 ± 0.01, p = 0.001).

View larger version (16K): [in this window] [in a new window]   Figure 5.   Alveolar instability index (AIX) mean values for the 10 ventrodorsal segments in the GV and PLV animals at baseline (white bars) and after induction of lung injury (black bars). Values will always be greater than 1. The closer the value is to 1, the better inflated the regional alveoli are at end-expiration.

At 90 min after induction of lung injury in the GV animals, the mean AIX was 1.18 (range, 1.1-1.3) in the nondependent regions and 2.02 (range, 1.55-2.6) in the dependent regions, p = 0.01. The GLX at a similar time, but following a dose of 30 ml/kg perflubron, was 0.56 (range, 0.53-0.61) in the nondependent regions and 0.78 (range, 0.70-0.91) in the dependent segments in the PLV animals, p = 0.01 (Figure 6). Note that the GLX is always less than 1 during PLV. The further the value is from 1, the greater the relative volume of gas compared with perflubron that is present in the regional alveoli at end-inspiration.

View larger version (20K): [in this window] [in a new window]   Figure 6.   Alveolar instability index (AIX, white bars) and gas-liquid index (GLX, black bars) for the 10 ventrodorsal segments during GV 90 min after lung injury and during PLV after administration of 30 ml/kg perflubron. The AIX values will always be greater than 1 and GLX values less than 1. The closer the AIX is to 1, the better inflated the regional alveoli are at end-expiration. The further the GLX is from a value of 1, the more gas relative to perflubron is in the alveoli at end-inspiration. Bars represent mean values; lines, median values.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The objective of this study was to evaluate the effect of liquid ventilation on gas exchange and hemodynamics during acute lung injury, with the specific goals of describing the distribution of perflubron and gas during PLV in the setting of lung injury. The principal findings observed in this study were (1) gas exchange is better during PLV than during GV; (2) aeration is decreased in the dependent regions of the lungs during GV in the setting of lung injury; (3) perflubron is distributed predominantly to the dependent regions of the lungs; and (4) although gas is distributed to both the dependent and nondependent regions of the lungs during PLV, more gas, relative to perflubron, is distributed to the nondependent than dependent regions of the lungs with each inspiration.

Numerous studies have suggested that gas exchange and pulmonary compliance are improved in preterm and fullterm neonatal, pediatric, and adult animal models of lung injury with PLV. Such lung-injury models have included those induced by intravenous oleic acid administration, gastric acid aspiration, meconium aspiration, congenital diaphragmatic hernia, and saline lavage (1, 4, 6, 19, 20). One potential mechanism by which PLV may improve gas exchange involves recruitment of otherwise atelectatic injured lung, which is predominantly located in the dependent lung regions. Gattinoni and coworkers (25, 26) have demonstrated that the distribution of CT-measured lung density in patients with acute respiratory failure is dependent upon PEEP and body position. Specifically, such studies reveal that the dorsal, or dependent, lung regions in the supine patient demonstrate increased density or CT attenuation numbers approaching those of soft tissue. In contrast, the nondependent or ventral regions demonstrate CT attenuation numbers that are consistent with normal, inflated lung. Application of PEEP appears to decrease the density and to increase the inflation of the dependent lung regions. It has been previously demonstrated that perflubron affects alveolar recruitment in these dependent lung regions during total liquid ventilation (19). In the current study we wished to examine the hypothesis that perflubron is distributed predominantly to dependent, otherwise nonventilated regions of the lungs during PLV, while gas is distributed toward the nondependent lung regions. To do so, we applied a method of analysis in which CT density data were divided into 10 segments in a ventrodorsal orientation. After inducing lung injury and before initiating PLV, we documented an increase in mean CT attenuation number that was greater in the dorsal, or dependent, lung regions in both the GV and PLV animal groups. Shunt fraction simultaneously increased, indicating the presence of marked V/Q mismatch. After dose administration, perflubron was distributed predominantly to the dependent regions of the lungs, so that the CT density in the dependent regions was greater than that observed in the nondependent regions. In addition, the GLX was closer to 1 in the dependent than in the nondependent lung regions. This indicates a greater relative volume of gas when compared with perfluorocarbon in the nondependent than the dependent regions at end-inspiration. A simultaneous decrease in physiologic shunt was observed in the PLV animals when compared with the GV animals, which would lead one to suggest that V/Q matching was enhanced by improvement in ventilation after a perflubron dose of 30 ml/kg. We cannot exclude, however, the possibility that this reduction in shunt fraction is due in part to a redistribution of pulmonary blood flow from the dependent to the nondependent regions of the lungs, resulting from the effect of the relatively dense perflubron upon regional pulmonary vascular resistance and blood flow in the dorsal lung zones (27, 28). This redistribution of pulmonary blood flow may play a role in the improvement in V/Q matching observed during PLV in this model of lung injury.

A number of studies have demonstrated the ability of thoracic CT density measurements to quantify the degree of lung atelectasis after induction of lung injury (19, 22, 29). This is the first study, however, to use the technique with three phases present in the lung: perflubron (HU = 2,300), soft tissue (HU = 0), and air (HU = 1,000). Since it is reasonable to assume that the soft-tissue component remains stable during each ventilator cycle, two components ventilate the alveoli and determine the GLX value: a low-density component (gas) and a high-density component (perflubron). It is important to realize that perflubron is distributed to the distal airways during inspiration, whereas during expiration it partially flows back into and fills the central airways. The GLX values would suggest that relatively more perflubron than gas was present in the dependent than in the nondependent regions of the lung at end-inspiration. However, it should be emphasized that these data only allow one to determine the relative amounts, but not the absolute volume, of gas and perflubron distributed to each region with each ventilatory cycle; as mentioned previously, the presence of the three phases within the lungs precludes determination of absolute changes in regional perflubron and gas with each ventilator cycle.

In conclusion, in this model of respiratory failure we demonstrated improvement in gas exchange during PLV when compared with GV. Segmental CT-scan analysis supports the hypothesis that perflubron is predominantly distributed toward the dependent lung regions. During PLV, gas is distributed to all lung regions with more gas than perflubron distributed to the nondependent lung regions with each inspiration. These data suggest that a component of the enhancement in gas exchange that is observed during PLV is likely related to recruitment of atelectatic, dependent lung segments by perflubron. Whether redistribution of pulmonary blood flow induced by perflubron administration also plays a role in the observed improvement in ventilation/perfusion matching will require assessment by other studies that are currently underway.