INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Acute lung injury in respiratory distress syndrome (RDS) is characterized by progressive deterioration in oxygenation and lung inflammation (1). In experimental ovine RDS, treatment with exogenous surfactant improves oxygenation and decreases lung injury (2). However, extremely premature lambs do not have sustained improvement in oxygenation despite repeated doses of surfactant (5), suggesting that factors other than surfactant deficiency contribute to lung injury in this setting. Indeed, the structural and functional pulmonary immaturity that characterizes RDS renders the extremely premature newborn uniquely susceptible to lung injury from oxygen toxicity and volutrauma (6, 7). In particular, volutrauma has been implicated in acute lung injury and in the evolution of chronic lung disease in RDS by causing increased lung vascular permeability (8), neutrophil-mediated inflammation (11), and altered production of inflammatory cytokines (15).

Although exogenous surfactant therapy has dramatically improved survival in clinical RDS, chronic lung disease continues to cause significant morbidity (16, 17). Thus, there is increasing interest in the potential role of newer therapies that may further decrease lung injury in the susceptible premature newborn. Lung recruitment techniques that minimize the adverse effects of tidal volume distention with gas (volutrauma) may attenuate lung injury. Two techniques that cause lung recruitment while potentially minimizing the attendant risks of volutrauma during gas ventilation are high-frequency oscillatory ventilation (HFOV) and partial liquid ventilation (PLV). HFOV decreases lung injury, in part, by avoiding large phasic pressure changes in the airways and alveoli (18). Because opening pressures are high in the underinflated, surfactant- deficient lung, PLV may be useful in lung recruitment by eliminating the adverse effects of airway and alveolar distention with gas in the presence of an air-liquid interface, thus allowing lung recruitment at lower inspiratory pressures (19). Inhaled nitric oxide (iNO) also holds promise as an adjunctive treatment for RDS because of its effects on pulmonary vasodilation, ventilation-perfusion matching, and lung inflammation (1, 20). However, the relative effectiveness of these three therapies in improving gas exchange and decreasing inflammation in severe RDS is unknown.

We hypothesized that the two lung recruitment strategies (HFOV and PLV with perflubron) would have similar effects on gas exchange and reduction in lung inflammation, and that both strategies would augment the response to low-dose iNO in premature lambs with severe RDS. The extremely premature lamb delivered at 115 d gestation (0.78% term) provides a unique model of acute lung injury characterized by progressive deterioration in oxygenation over a 4-h period during conventional tidal volume mechanical ventilation. We studied the individual and combined effects of iNO, HFOV, and PLV in 31 premature lambs using seven mechanical ventilation protocols.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mixed-breed (Columbia-Rambouillet) pregnant ewes were used in this study. All procedures and protocols were reviewed and approved by the Animal Care and Use Committee at the University of Colorado Health Sciences Center. This study included seven ventilator protocols that examine the independent and combined effects of iNO, HFOV, and PLV on gas exchange and lung inflammation in this premature (115 d gestation, 0.78 term; term = 147 d) ovine model of severe RDS (1).

Study Design

Ewes were sedated with intravenous pentobarbital sodium (2-4 g total dose) and anesthetized with intrathecal tetracaine hydrochloride (1% solution, 3 mg). A uterine incision was made under sterile conditions. The fetal head was exteriorized and placed in a rubber glove containing warm saline to prevent fetal breathing prior to controlled ventilation. A right paramedian skin incision was made in the neck after local infiltration with lidocaine (1% solution, 2-3 ml). Polyvinyl catheters (20-gauge; Martech Medical Products, Lansdale, PA) were advanced into the ascending aorta through the carotid artery and into the superior vena cava through the jugular vein. The aortic catheter was connected to a MP 100A Biopac System (Santa Barbara, CA). Calibrations of pressure transducers were performed with a mercury column manometer. Pancuronium (0.1 mg/kg) was administered to the fetus and a tracheotomy was performed with placement of a 3.0-mm i.d. endotracheal tube (HiLo Jet tube; Mallinckrodt). All animals were treated with exogenous surfactant (Infasurf, provided by E. A. Egan, M.D.) at an estimated dose of 3 ml/kg (105 mg phospholipid/kg) before the first breath. Mechanical ventilation was initiated with a continuous flow, time-cycled, pressure-limited neonatal ventilator (Infant Star; Infrasonics, San Diego, CA) at the following settings: peak inspiratory pressure (PIP), 35 cm H₂O (3.4 kilopascals [kPa]); positive end-expiratory pressure (PEEP), 6 cm H₂O (0.6 kPa); rate, 30 breaths per minute; inspiratory time, 1.0 s; and FI_O2 = 1.00. After 10 min of mechanical ventilation, the umbilical cord was ligated and the animal was transferred to a radiant warmer. A solution of 5% dextrose in normal saline and pentobarbital (1 mg/kg/h) was continuously infused at 10 ml/h. Sodium bicarbonate (1 milliequivalent [mEq]/kg) was infused to correct metabolic acidemia if the arterial pH was less than 7.25 with Pa_CO2 in the target range for this study (see below).

Blood samples for pH, PO₂, PCO₂, oxygen saturation, and methemoglobin were withdrawn anaerobically into Natelson glass pipettes and analyzed at 39.5° C using a Radiometer OSM3 blood gas analyzer (Copenhagen, Denmark). The NO gas (Scott Specialty Gases, Plumbsteadville, PA) used in these experiments was certified at a concentration of 450 ppm NO (chemiluminescence method) with less than 1% contamination by other oxides of nitrogen. Concentrations of inhaled NO and NO₂ were measured with an electrochemical sensor calibrated against a reference NO tank.

Experimental Design

After delivery to the radiant warmer, animals were treated with one of seven different ventilator strategies. The approaches to mechanical ventilation described below evolved from experience ventilating the very premature lamb, and emphasize optimizing gas exchange and Pa_O2 while minimizing the risk of air leak and adverse hemodynamic effects (1, 20).

Four groups were treated with conventional ventilation (intermittent mandatory ventilation, Infant Star): (1) control conventional ventilation (CV, n = 5); (2) CV + iNO (5 ppm), n = 5; (3) CV + PLV, n = 5; and (4) CV + PLV + iNO (5 ppm), n = 4.

These four groups allow for measurements of the response to hemodynamic modulation with iNO (CV), lung recruitment alone (CV + PLV), and iNO with lung recruitment (CV + PLV + iNO).

The following protocol was followed during the 4-h period of conventional mechanical ventilation. Mechanical ventilator settings were modified during the course of studies based on the results of serial arterial blood gas samples. Changes in PIP were determined by measurements of Pa_CO2. If Pa_CO2 was less than 35 mm Hg (4.7 kPa), then PIP was reduced to 30 cm H₂O (2.9 kPa). If subsequent measurements of Pa_CO2 were less than 35 mm Hg (4.7 kPa), then PIP was reduced to 25 cm H₂O (2.5 kPa). The maximum PIP delivered was 35 cm H₂O (3.4 kPa). If Pa_CO2 was greater than 45 mm Hg (6 kPa), then the ventilator rate was increased to a maximum of 60 breaths/min. The inspiratory time was then decreased to maintain an inspiratory to expiratory ratio of 1.0 or less. PEEP was changed according to Pa_O2. If Pa_O2 was less than 100 mm Hg (13.3 kPa), PEEP was maintained at 6 cm H₂O (0.6 kPa). With Pa_O2 greater than 100 mm Hg (13.3 kPa), PEEP was decreased to 5 cm H₂O (0.5 kPa). If Pa_O2 was greater than 200 mm Hg (26.7 kPa), PEEP was decreased to 4 cm H₂O (0.4 kPa).

Three groups of animals were treated with HFOV (Sensormedics Inc., Yerba Buena, CA): (1) Control HFOV, n = 5; (2) HFOV + iNO (5 ppm), n = 4; and (3) HFOV + PLV, n = 3. This design allows for comparisons of the effects of lung recruitment with HFOV alone or in combination with PLV and the separate effects of iNO. HFOV + iNO in this design is then conceptually equivalent to CV + PLV + iNO.

The following protocol was followed during the 4-h period of HFOV. The initial HFOV settings were: mean airway pressure, 22 cm H₂O; amplitude, 40 cm H₂O; frequency, 10 Hz; and % inspiration, 33. In preliminary studies we found that the use of lower mean airway pressures at the onset of HFOV treatment did not effectively recruit lung volume, and Pa_O2 was not optimized. Changes in amplitude were determined by measurements of Pa_CO2. If Pa_CO2 was less than 35 mm Hg (4.7 kPa), then the amplitude was reduced by 5 cm H₂O decrements until the target Pa_CO2 was reached. If Pa_CO2 was greater than 45 mm Hg (6 kPa), then the amplitude was increased by 5 cm H₂O increments until the target Pa_CO2 was reached. Frequency and % inspiration were not changed during the study. Mean airway pressure was changed according to Pa_O2. If Pa_O2 was less than 100 mm Hg (13.3 kPa), mean airway pressure was increased by 2 cm H₂O (0.2 kPa). If Pa_O2 was less than 200 mm Hg (26.7 kPa), mean airway pressure was decreased by 2 cm H₂O (0.2 kPa).

Perflubron (LiquiVent; Alliance Pharmaceutical, San Diego, CA) was administered in the perfluorocarbon (PFC) groups through the pressure-monitoring lumen of the HiLo jet tube at an estimated dose of 10 ml/kg over 10 min. Preliminary studies demonstrated rapid changes in ventilation and oxygenation after administration of PFC. Therefore, in animals treated with PFC, ventilator settings were decreased empirically in the following manner. During CV, PIP was decreased to 25 cm H₂O at the completion of PFC administration. During HFOV, mean airway pressure was decreased to 15 cm H₂O and pressure amplitude was decreased to 25 cm H₂O. Additional PFC liquid was added in 1-3 ml doses at each 30-min time point by disconnecting the ventilator and administering PFC liquid until the fluid was visible in the endotracheal tube. After each administration of PFC liquid and at 30-min intervals, the animal was briefly rotated side-to-side.

After the study, animals were killed with T-61 euthanasia solution (American Hoechst, Summerville, NJ) and body weight was recorded. The right lung was snap-frozen in liquid nitrogen for subsequent determination of myeloperoxidase (MPO) activity. MPO activity was determined in whole lung tissue as previously described (21).

Statistical Analysis

Statistical comparisons of within-group continuous variables were performed using one-way repeated measures analysis of variance (ANOVA). Where significant differences were identified, post hoc analysis was performed using Student-Neuman-Keul's test. The analysis of differences among the seven groups over time was performed using two-way ANOVA for repeated measures (with time and treatment group identified as independent variables and the interaction between them analyzed). Comparisons of responses to each intervention (among treatment groups) at the hourly time points and MPO data were performed using one-way ANOVA and the Student-Neuman-Keul's test. The level of statistical significance was set at p < 0.05; results are reported as mean ± standard error.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There were no significant differences in body weight among the seven study groups (mean for all animals = 1.84 ± 0.08 kg). There were no differences in baseline (pre-ventilation) Pa_O2, Pa_CO2, arterial pH, heart rate, or mean systemic arterial blood pressure among groups.

In the control CV group, Pa_O2 progressively deteriorated during the 4-h study period (Table 1). By the 2-h time point and for the duration of the experiment, Pa_O2 was lower in the control CV group compared with all other groups (p < 0.05). At the end of the experiment (4-h time point), Pa_O2 in the CV + PLV + iNO group was higher than in all other CV groups (p < 0.05; Table 1, Figure 1). Similarly, Pa_O2 at the 4-h time point was higher in the HFOV + iNO group than in all other HFOV groups (p < 0.05; Table 1, Figure 1).

View larger version (13K): [in this window] [in a new window]   Figure 1.   PaO2 for all seven study groups at the end of the experiment (4-h time point). Both HFOV and CV + PLV augment the response to iNO. *p < 0.05 versus all groups; **p < 0.05 versus all CV groups; ***p < 0.05 versus all HFOV groups.

There were marked differences among groups in mean airway pressure required to optimize Pa_O2 (Table 2). Among the CV groups, mean airway pressure was lowest for CV + PLV + iNO at the 1-h time point (p < 0.05), and the reduction in pressure was predominantly related to the addition of PLV rather than iNO (Table 2). Among the HFOV groups, mean airway pressure was markedly lower for HFOV + PLV treatment at all time points (Table 2; Figure 2A).

View larger version (13K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Figure 2.   Mean airway pressure (A) and phasic pressure change (B) for all seven study groups at the end of the experiment (4-h time point). Combining PLV with CV or HFOV allowed marked reductions in mean airway pressure and phasic pressure change. *p < 0.05 versus all CV groups; **p < 0.05 versus all HFOV groups.

There were also marked differences among groups in the phasic pressure change (PIP during CV and pressure amplitude during HFOV) required to maintain Pa_CO2 in the target range. In the CV control group, PIP was higher than all other conventionally ventilated groups by 2 h and for the remainder of the experiment (Table 3; Figure 2B). In contrast, the PIP for the CV + PLV + iNO group was lower than all other groups for the first 2 h of the experiment. For the HFOV groups, the combination of HFOV + PLV allowed lower pressure amplitudes during the entire 4-h study period (Table 3; Figure 2B).

Oxygenation was also assessed by analyzing the relationship between mean airway pressure and Pa_O2, expressed as the oxygenation index (OI; FI_O2 · mean airway pressure · 100 · 1/Pa_O2). By the 4-h time point, the OI was reduced in all treatment groups compared with the control CV group (p < 0.05; Figure 3). Treatment with iNO during CV decreased OI, but combining CV with either PLV or PLV + iNO caused a further decrease in OI (p < 0.05; Figure 3). HFOV treatment decreased OI compared with control CV, and combining either PLV or iNO with HFOV caused a further decrease in OI (p < 0.05; Figure 3).

View larger version (10K): [in this window] [in a new window]   Figure 3.   Oxygenation index for all seven groups at the end of the experiment (4-h time point). *p < 0.05 versus all groups; **p < 0.05 versus CV + iNO.

Lung neutrophil accumulation was measured using the myeloperoxidase (MPO) assay. Compared with control CV, iNO treatment during CV markedly decreased lung MPO activity (0.53 ± 0.19 versus 0.23 ± 0.03 u/g lung, p < 0.05; Figure 4). However, lung recruitment with either HFOV or CV + PLV also decreased lung MPO activity (p < 0.05; Figure 4).

View larger version (11K): [in this window] [in a new window]   Figure 4.   Lung neutrophil accumulation measured by myeloperoxidase activity for all seven study groups. Lung recruitment techniques and iNO independently decreased lung inflammation. *p < 0.05 versus all groups.

There were no significant differences among groups at any time point for Pa_CO2 (Table 4), arterial pH (Table 5), or mean systemic arterial blood pressure (Table 6).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The premature lamb delivered at 115 d gestation develops severe RDS manifested by hypercarbia and hypoxemia despite mechanical ventilation with high inspired oxygen concentrations (24, 25). Treatment with exogenous surfactant at delivery improves gas exchange initially; however, conventional tidal volume mechanical ventilation causes lung injury and inflammation with progressive deterioration over 4 h (1). Because of its similarity to severe clinical RDS and its particular susceptibility to acute lung injury during conventional gas ventilation, we used the extremely premature lamb to study the independent and combined effects of lung recruitment strategies (HFOV, PLV with perflubron) and selective pulmonary vasodilation (iNO) on gas exchange and lung inflammation. We found lung recruitment with HFOV and/or PLV with perflubron caused similar improvements in gas exchange and lung inflammation in the premature lamb with severe RDS, and that both strategies augmented the oxygenation response to low-dose iNO.

During conventional ventilation, the addition of PLV (perflubron) caused marked improvement in oxygenation and allowed the use of a lower mean airway pressure compared with control subjects. During HFOV, the mean airway pressure required to achieve a similar improvement in oxygenation was considerably higher than that of CV + PLV with perflubron. The use of higher mean airway pressure, however, was not associated with adverse hemodynamic effects nor increased lung inflammation in this model. There are practical limitations to the comparisons of airway pressures between HFOV and CV. Since the proximal measurement of airway pressure during HFOV likely overestimates the distal transmission of pressure (due to attenuation across the endotracheal tube and airways), direct comparisons between the airway pressures used during HFOV and CV may not be easily interpretable (26). Moreover, lung volume cannot be measured during HFOV without some interruption of the ventilator pattern. However, the addition of PLV to HFOV allowed for the mean airway pressure to be reduced by 45% compared with HFOV alone, due to the salutary effects of perfluorochemical liquids on decreasing the air-liquid interface at the alveolar surface (27).

Previous reports of combined treatment with HFOV and PLV have shown conflicting results. Baden and colleagues (28) compared HFOV with HFOV + PFC in piglets injured by saline lung lavage. Animals were treated with equivalent mean airway pressures during the experiment (20 cm H₂O). Although Pa_O2 initially increased more rapidly in the HFOV + PFC group, oxygenation was similar by 120 min of treatment. In contrast, Sukumar and coworkers (29) reported marked and sustained increases in Pa_O2 during HFOV + PLV compared with HFOV alone using a fixed mean airway pressure of 15 cm H₂O in premature lambs (127 d) with RDS. In the study of Sukumar and coworkers, HFOV-treated animals had profound hypoxemia (Pa_O2 = 17 ± 1.8 mm Hg) and hemodynamic deterioration by the end of the 5-h study period. In another study, Smith and colleagues (30) found that oxygenation was higher during CV + PLV than during HFOV + PLV in piglets after lung injury with saline lavage. The striking differences between studies are most likely related to differences in the HFOV management scheme employed. Earlier studies demonstrated that the lung protective effects of HFOV accrue when the technique is applied with an adequate recruiting pressure, thus sustaining lung volume, while ventilation is achieved using small phasic pressure changes (31, 32). If an inadequate recruiting pressure is employed, little benefit from HFOV would be expected (33). For example, in the study of Sukumar and coworkers (30), a relatively low mean airway pressure in the surfactant-deficient and poorly compliant lung would not be expected to cause lung recruitment without combining the approach with additional lung recruitment techniques (PLV). Indeed, in preliminary experiments designed to determine optimal initial settings for HFOV in the premature lamb, we found that the use of a lower mean airway pressure (< 20 cm H₂O) resulted in inadequate lung recruitment and lower Pa_O2 values.

Similarly, alternative conventional ventilation approaches may affect the response to treatment with PFC as demonstrated by Cox and associates (34). One limitation of our study is that a single ventilator strategy and PFC dose were used during CV. Based on experience gained from previous studies using the 115-d premature lamb, we chose initial ventilator settings and made changes during the course of the experiment that were designed to optimize gas exchange and minimize air leak (1, 20). A recent report suggested that the acute improvement in Pa_O2 after PFC adminstration was greater in premature lambs with RDS ventilated at 60 breaths/min when compared with animals ventilated at 30 breaths/min (35). However, there was no difference in Pa_O2 between groups by 4 h of age.

Another variable that may affect PLV responses in the premature lamb is exogenous surfactant treatment. All animals in this study were pretreated with surfactant to simulate the usual clinical conditions. It is possible that less striking changes in oxygenation would have occurred during CV if animals were not pretreated with surfactant. As demonstrated by Tarczy-Hornoch and colleagues (36), the combination of surfactant with PLV is more effective at improving lung compliance than either modality alone in the premature lamb.

We also found that low-dose iNO and the use of lung recruitment strategies independently reduced lung neutrophil accumulation. The beneficial effects of these therapies on early neutrophil accumulation may have important clinical implications in RDS. The neutrophil plays a substantial role in the inflammatory cascade, which contributes to lung injury and the evolution of the most important sequel of RDS, chronic lung disease (12). Sequestration of neutrophils in the lung is an early step in a complex inflammatory response mediated through the elaboration of oxyradicals, proteases, phospholipases, and lipid compounds (37). Therapies that reduce neutrophil accumulation in the lung in RDS could potentially modify the early inflammatory process that amplifies acute lung injury and contributes to the development of chronic lung disease. However, potential mechanisms that account for the reduction in lung inflammation in this study are only speculative. Lung injury during CV alone may cause increased expression of neutrophil adhesion molecules on the pulmonary vascular endothelium, thus promoting neutrophil accumulation within the lung. It is likely that both HFOV and PLV minimize lung injury and diminish the adverse effects of mechanical ventilation on the pulmonary vasculature. The putative role of nitric oxide in modulating neutrophil-endothelial cell interactions is better characterized (38), and there is recent evidence that endogenous NO production modulates neutrophil-endothelial cell interactions in the microcirculation (39). However, whether iNO decreases lung neutrophil accumulation by affecting the adhesion characteristics of the pulmonary vascular endothelial cell, the neutrophil, or both is unknown.

Thus, lung recruitment with either HFOV or CV + PLV causes similar improvements in oxygenation and lung inflammation in the premature lamb with RDS, and both strategies augment the response to low-dose iNO. However, rapid adjustments in ventilator settings may be necessary when combining lung recruitment techniques in order to avoid adverse hemodynamic effects. Although the advent of newer therapies for the treatment of hypoxemic respiratory failure holds the promise of improving the clinical management of critically ill infants with RDS, considerably more study is required to understand both the interaction of these therapies and the most effective therapeutic strategies before the safe clinical application of multi-modal lung recruitment is practical.