Main

There is an increasing awareness of the need for an “open” lung volume approach to mechanical ventilation. Minimal lung injury observed during high-frequency oscillatory ventilation can be reproduced using conventional ventilation strategies that optimize positive end-expiratory pressure (PEEP) (1). This approach has not gained widespread clinical support in neonates because of concerns about the potential of persistently raised PEEP levels (PEEP 10 cm H2O) to impede venous return, decrease cardiac preload, and pulmonary blood flow and impair cerebral blood flow (2,3).

An alternative approach is offered by BiLevel ventilation [also known as biphasic positive airway pressure (BIPAP or PeV+)]. BiLevel offers a flexible mode of ventilation similar to assist-control (A-C) except that it cycles between a low and a high level of positive pressure or PEEP (PEEPL and PEEPH). Transitions between levels of PEEP can be triggered by the patient (synchronized transitions) or initiated by the ventilator according to preset time (tL or tH) on each level of PEEP (time-based transitions). Spontaneous breathing is achievable at either PEEP level, and these breaths can be pressure supported. It is possible that cycling PEEP between low and high levels may offer the opportunity to recruit additional lung volume by exploiting the nonlinear dynamic nature of the pressure–volume relationship, while providing some “relief” from the effects of persistently raised background PEEP on cardiovascular status.

Superior gas exchange and cardiovascular function with BiLevel ventilation were demonstrated in adult animal models of acute lung injury comparing BiLevel with pressure support ventilation (4) and in clinical trials of BiLevel versus volume-controlled ventilation in adults with acute respiratory distress syndrome (ARDS) (5,6). A reduced need for sedation and decreased time on the ventilator in BiLevel compared with A-C was shown in a large cohort study in adults recovering from cardiac surgery (7). BiLevel ventilation is being used increasingly as an alternative ventilation strategy to conventional ventilation in ARDS (8), and its use was found to be safe in infants beyond the newborn period (9). Nonsynchronized nasal biphasic pressure ventilation was reported to improve gas exchange compared with nasal continuous positive airway pressure (CPAP) in stable preterm infants recovering from respiratory distress syndrome (RDS) (10). Endotracheal BiLevel ventilation has not been assessed in preterm neonates with acute RDS.

The objective of this study was to compare BiLevel ventilation with A-C ventilation in a preterm lamb model of neonatal lung disease. We investigated gas exchange and early indicators of lung injury. We compared BiLevel ventilation using PEEPL of 5 cm H2O and PEEPH of 20 cm H2O with A-C ventilation with PEEP 5 cm H2O. We hypothesized that the use of an intermittent high PEEP in BiLevel ventilation would reduce lung injury compared with A-C ventilation while maintaining adequate gas exchange. Some of the results of this study were reported previously in abstract form (Schulzke SM et al., Biphasic (Bilevel) versus assist-control ventilation in preterm lambs, 2007 ERS Annual Meeting, September 15–19, 2007, Stockholm, Sweden, Abstract P2338).

MATERIALS AND METHODS

Ethics.

The investigations were approved by the Animal Ethics Committee of the Department of Agriculture and Food, Western Australia.

Animals, delivery, and postnatal care.

We used an established ovine model of lung disease in moderately preterm lambs (11). Twenty-one lambs were delivered by caesarean section at 134 d gestation (term = 146 d) to date-mated Merino ewes. One lamb died at delivery. Remaining lambs were weighed, dried, intubated with a 4.5-mm internal diameter cuffed endotracheal tube, and quasirandomized to an unventilated control group (UVC, n = 6) that was euthanized immediately before delivery by 50 mg/kg i.v. pentobarbitone (Abbott Laboratories, Sydney, Australia) or to one of two ventilation groups using a Puritan Bennett 840 ventilator (Nellcor Puritan Bennett, Pleasanton, CA): 1) BiLevel ventilation using a PEEPL of 5 cm H2O, a PEEPH of 20 cm H2O, and pressure support of 25 cm H2O (BiLevel, n = 7); 2) A-C with PEEP of 5 cm H2O and peak inspiratory pressure (PIP) of 30 cm H2O (A-C5, n = 7). The baseline ventilator settings for the first 10 min after delivery are defined in Table 1. Both BiLevel and A-C5 groups were ventilated using a pressure-controlled, volume-targeted, flow-triggered (sensitivity 0.2 L/min) ventilation strategy, i.e., the initial PIP of 30 cm H2O (equivalent to PEEPL + pressure support in BiLevel group) in each group was subsequently adjusted to achieve tidal volumes (VT) of 7–8 mL/kg. VT estimation and pressure adjustment in lambs ventilated with BiLevel were based on readings obtained on PEEPL. An example of differences in the pressure wave form between the ventilation modes is shown in Figure 1. All ventilated lambs had a set (backup) respiratory rate (RR) of 20 breaths/min and a fixed fractional inspired oxygen concentration (FiO2) of 0.4. In BiLevel mode, the ventilator cycled between the two preset levels of PEEP (PEEPL and PEEPH) based on preset time on PEEPL (tL, 2.0 s) and PEEPH (tH, 1.0 s) while delivering pressure support for spontaneous breathing throughout the ventilatory cycle. In A-C mode, the ventilator allowed for time-limited (inspiratory time 0.5 s) spontaneous inspirations and all spontaneous respiratory efforts were supported with the preset PIP. Lambs were not sedated to facilitate spontaneous breathing effort. After intubation and commencing ventilation, the lambs were positioned prone, covered with transparent occlusive wrap (NeoWrap; Fisher & Paykel Healthcare, Auckland, NZ), and kept warm on an infant radiant warmer (CosyCot; Fisher & Paykel Healthcare).

Table 1 Initial ventilator settings
Full size table
Figure 1
figure 1

Characteristics of BiLevel (A) vs assist-control (B) ventilation. In BiLevel mode, the ventilator cycles between two preset levels of PEEP: PEEPL and PEEPH. Transitions between levels of PEEP can be flow triggered by the patient (synchronized transitions) or initiated by the ventilator according to preset time on each level of PEEP (time-based transitions). Flow-triggered spontaneous breathing with pressure support is achievable on either level of PEEP. In A-C mode, the ventilator applies breaths that are pressure controlled and time cycled with preset PEEP, PIP, and tI. Flow-triggered spontaneous inspirations are achievable on PEEP but not PIP level. All spontaneous breaths are assisted. PEEPL, low level of PEEP; PEEPH, high level of PEEP; tL, time on PEEPL; tH, time on PEEPH; tI, inspiratory time; tE, expiratory time.

Full size image

Lung processing.

Ventilated lambs were euthanized (50 mg/kg pentobarbitone) at 120 min, and the tracheal tube was clamped for 3 min to facilitate oxygen absorption and lung collapse for subsequent determination of the pressure–volume relationship (12). Lambs were weighed to obtain dry weight; lungs were weighed, pressure-volume relationship was determined from excised lungs (including UVC group), and samples from the right lower lobe were taken for molecular analyses.

Time course of physiologic variables.

Variables measured by the ventilator's built-in hot-film anemometer and manometer [VT and minute volume, total RR, PIP, and mean airway pressure (MAP)], arterial pH, partial pressure of arterial oxygen (Pao2), and carbon dioxide (Paco2) were documented at 10, 20, 30, 45, 60, 90, and 120 min postdelivery. Blood from an umbilical arterial catheter was used for blood gas and pH measurements (ABL 800, Radiometer, Copenhagen, Denmark). Oxygenation index (OI) was calculated as: OI = MAP × FiO2 × 100 /Pao2 (13).

Histologic inflammation score.

The right upper lobe of each lung was inflation fixed with 10% formalin at 30 cm H2O pressure. Lobes were cut into 5-μm serial sections, and three hematoxylin and eosin-stained sections were selected randomly in a blinded fashion by one observer (G.R.P.). The amount of inflammation in UVC, BiLevel, and A-C5 lambs was graded using a score validated previously in this ventilated preterm lamb model (14). Each lung section was scored as 0 (no inflammatory cells in tissue or airspaces), 1 (a few cells), 2 (a moderate cell infiltration), and 3 (large numbers of inflammatory cells in airspaces and tissue), assuming a linear relationship between amount of inflammation and inflammatory score (14). Average scores for airspaces and tissue were calculated for each animal (15).

Messenger RNA (mRNA) extraction and quantitative PCR.

Messenger RNA (mRNA) expression of early markers of neonatal lung injury (16), including early growth response one (EGR1) (17), cysteine-rich protein 61 (CYR61) (18,19), and connective tissue growth factor (CTGF) (20,21), was measured using quantitative PCR (qPCR). RNA was extracted from the lung tissue (Qiagen RNeasy Mini Kit; Qiagen, Valencia, CA) and reverse transcribed into complementary DNA (cDNA) (M-MLV Reverse Transcriptase, RNase H Minus, Point Mutant Kit; Promega, Madison, WI). qPCR was performed using a Realplex Real-Time Multiplexing System (Eppendorf, Hamburg, Germany). The qPCR reaction conditions and primers for these genes were published previously (16). The mRNA levels of each gene for each animal were normalized using the ΔΔCt (cycle threshold) method. The mean mRNA level for each gene in control animals was set to 1 and the mean mRNA levels for each ventilation modality were expressed relative to the mean value in control animals.

Data analysis and statistics.

Results are shown as mean (SEM) unless specified otherwise. Statistics were analyzed using Stata v10.0 (Stata Corporation, College Station, TX) and Prism v5.0 (Graphpad Inc, San Diego, CA). One-way ANOVA with (inflammation score) and without (pressure–volume relationships, biochemical markers) repeated measures and Bonferroni's multiple comparison test were used for statistical comparisons of continuous variables, with treatment modality as the independent factor. Categorical data were analyzed using Fisher's exact test. Two-tailed paired (time course of physiologic variables) and unpaired (wet/dry weight ratios) t tests were used to compare outcomes that were only available in BiLevel and A-C5 groups. Statistical significance was accepted as p < 0.05.

RESULTS

Table 2 summarizes baseline values for UVC, BiLevel, and A-C5 groups. There was a trend toward a higher proportion of male lambs in the UVC group (p = 0.054). Body weight and cord pH were not different between the groups.

Table 2 Baseline variables
Full size table

Time course of physiologic variables.

Table 3 highlights differences in ventilatory variables and blood gases between BiLevel and A-C5 groups. Further details on the time course of physiologic variables are given in Figure 2. BiLevel lambs had higher MAP than A-C5 lambs (mean difference 0.9 cm H2O, p = 0.014). Respiratory rate in the BiLevel lambs was higher (mean difference 9.6 min−1, p = 0.005) and minute ventilation lower than in A-C5 lambs (mean difference −0.05 L · kg−1 · min−1, p = 0.007). Pao2 in BiLevel lambs was lower than in A-C5 lambs during the first 60 min (mean difference −31.1 mm Hg, p = 0.018) but not different between groups after 60 min (mean difference −17.8 mm Hg, p = 0.20). There were no differences between groups in OI, VT, PIP, Paco2, or pH.

Table 3 Impact of ventilation modality on physiological variables
Full size table
Figure 2
figure 2

Time course of changes in physiologic variables and blood gases: A, mean airway pressure (continuous lines) and peak airway pressure (dashed lines), B, minute ventilation, C, tidal volume, D, Pao2, E, Paco2, and F, oxygenation index. Circles, BiLevel; squares, A-C5. *p < 0.05; **p < 0.01.

Full size image

Pressure–volume relationship, wet/dry weight ratio, and inflammation score.

BiLevel lambs had a higher static compliance than both A-C5 lambs (p = 0.006) and UVC lambs (p < 0.001) (Fig. 3). There was no difference in wet/dry weight ratio between BiLevel and A-C5 lambs (p = 0.77). BiLevel lambs had lower inflammatory scores in lung tissue than A-C5 (p = 0.013) but not UVC lambs (p = 0.35) (Fig. 4).

Figure 3
figure 3

Pressure–volume relationship. The deflation limb of the static pressure–volume relationship obtained postmortem in BiLevel, A-C5 and UVC lambs. There were differences among the BiLevel (○), A-C5 (▪), and UVC (▴) groups with BiLevel lambs having higher static compliance than both A-C5 and UVC lambs. *p < 0.05; **p < 0.01.

Full size image
Figure 4
figure 4

Average inflammatory scores in lung tissue. BiLevel lambs had lower inflammatory scores than A-C5 but not UVC lambs. *p < 0.05.

Full size image

mRNA expression of EGR1, CYR61, and CTGF.

Figure 5 illustrates differences in mRNA expression of EGR1, CYR61, and CTGF. There was no difference in mRNA levels of EGR1 (p = 0.42), CYR61 (p = 0.23), and CTGF (p = 0.18) between BiLevel and A-C5 groups. BiLevel lambs had higher mRNA levels of EGR1 than UVC lambs (p = 0.006), but there was no difference in mRNA levels of CYR61 (p = 0.01) and CTGF (p = 0.04) between BiLevel and UVC lambs (adjusted p value indicating significance of multiple comparisons: p = 0.008). A-C5 lambs had higher EGR1 (p = 0.004), CYR61 (p = 0.001), and CTGF (p = 0.007) mRNA levels compared with UVC lambs.

Figure 5
figure 5

mRNA expression of EGR1, CYR61, and CTGF. The values for normalized mRNA expression of A, EGR1, B, CYR61, and C, CTGF in UVC, BiLevel, and A-C5 lambs are shown. Fold increase in mRNA value is shown relative to UVC group. Mean (SEM) levels for each ventilation modality are expressed relative to the mean UVC value. The adjusted p value indicating significance of multiple comparisons was p = 0.008. *p < 0.008.

Full size image

DISCUSSION

Summary of main findings.

We found that BiLevel ventilation in preterm lambs is feasible and provides gas exchange comparable with A-C5 ventilation immediately postdelivery. BiLevel ventilation may reduce acute lung injury compared with A-C5 ventilation, as demonstrated by improved pressure–volume relationship, lower lung tissue inflammatory scores compared with A-C5 group lambs, and CYR61 and CTGF mRNA levels of A-C5 but not BiLevel lambs being higher than those of UVC lambs.

Gas exchange.

Ventilation modality did not significantly influence OI and Paco2, suggesting that arterial oxygenation and elimination of CO2 in BiLevel and A-C groups were similar. BiLevel lambs had slightly higher MAP than A-C5 lambs (mean difference 0.9 cm H2O). The increase in MAP in BiLevel lambs was lower than expected because of the rapid decrease in pressure support as ventilation was weaned to target the predefined Paco2 range and VT of 7–8 mL/kg. Consequently, the small increase in MAP was restricted to the first 30 min of the study period. The clinical importance of this small difference in MAP is questionable because in a clinical setting, FiO2 and/or PIP would be adjusted to maintain Pao2 within target range and the accuracy of built-in manometers is limited (22). It is likely, therefore, that any differences in outcomes between the BiLevel and A-C5 groups are primarily related to PEEP strategy, rather than MAP per se.

A more intriguing finding is the difference in the time course of Pao2 between BiLevel and A-C5 group animals. After commencement of mechanical ventilation and postnatal adaptation, Pao2 in BiLevel lambs showed little change over time, whereas Pao2 in A-C5 lambs increased rapidly after delivery followed by a decline after 30 min, i.e., Pao2 in BiLevel lambs was lower than in A-C5 lambs during the first 60 min of ventilation. Previous studies in an adult pig model (23,24) and human adults (25–27) indicated arterial oxygenation in BiLevel ventilation to be at least equivalent to A-C and pressure support ventilation. Given that the use of periods of high PEEP arguably increases lung recruitment (28), the lower initial Pao2 in BiLevel lambs compared with the A-C5 group may have resulted from reduced pulmonary blood flow caused by high PEEPH (29), transiently offsetting the benefits of PEEP for lung recruitment. Confirmation of this hypothesis requires measurement of lung volume and hemodynamics in future studies. The lack of measurement of hemodynamics in this study precludes any conclusions regarding the impact of ventilation modality on cardiac function.

Influence of ventilation modality on lung injury.

As expected, pressure–volume relationship in both BiLevel and A-C5 groups demonstrated higher static compliance compared with UVC lambs because of poor lung aeration in the latter. The improved pressure–volume relationship and decreased histologic lung inflammation score in BiLevel versus A-C5 ventilation, and the trend toward lower mRNA levels of CYR61 and CTGF in BiLevel lambs highlights the relevance of ventilation modality on lung injury. Three major suspected risk factors of ventilator-induced lung injury in surfactant-depleted newborn lungs are overdistension by large tidal volumes (30), repetitive opening and closing of alveolar units because of low PEEP ventilation strategy (31), and oxygen toxicity (32). A few large breaths during resuscitation initiated lung injury in preterm lambs (33), even when used in combination with surfactant therapy (34). Considering the large airway dead space of lambs (30), it is unlikely that this mechanism plays a major role in our study. Our target VT of 7–8 mL/kg delivers a low normal alveolar VT providing “gentle ventilation” of preterm lambs (who have a large anatomical dead space) and was applied to both BiLevel and A-C5 groups. The striking difference between the BiLevel and A-C5 lambs in our study is the PEEP strategy. BiLevel lambs were cycling between PEEP levels of 5 cm H2O and 20 cm H2O (mean level of PEEP in BiLevel: 10 cm H2O), whereas A-C5 group lambs were ventilated with a constant PEEP of 5 cm H2O. PEEP of 7 cm H2O versus 4 cm H2O has been associated with a larger alveolar surfactant pool in surfactant-treated preterm lambs ventilated for 7 h postdelivery (35). Maintaining lung volume with relatively high levels of PEEP (up to 15 cm H2O) and small pressure differences between PIP and PEEP (open-lung concept) compared with conventional ventilation had a beneficial effect on static compliance and histologic lung injury score in a term newborn piglet model (36,37). Thus, it is possible that the higher level of PEEP in BiLevel compared with A-C5 group lambs had a protective effect on the development of ventilator-associated lung injury.

Oxygen radical injury is a known contributor to neonatal lung injury (38). Although all ventilated lambs in this study were exposed to a constant FiO2 of 0.4, the Pao2 in BiLevel lambs was lower than A-C group lambs during the first hour of ventilation. Because the threshold Pao2 to induce lung injury in preterm lungs is unknown (39), it cannot be excluded that the differences in Pao2 confounded the finding of reduced lung inflammatory changes in BiLevel lambs. Finally, asynchrony contributes to pulmonary morbidity in ventilated neonates (40). Although we did not formally assess breathing pattern, visual observation of the lambs and ventilator alarms did not indicate substantial asynchrony in any of the ventilation modes.

Relevance for clinical practice and further research.

This is the first study to demonstrate the feasibility of endotracheal BiLevel ventilation in a preterm neonatal animal model. There were no adverse effects associated with BiLevel ventilation and gas exchange was comparable with A-C5 ventilation.

Comparable Paco2 in BiLevel versus A-C5 lambs in the presence of similar VT (at PEEPL), lower minute ventilation, and higher RR suggest that the higher respiratory rate in the BiLevel lambs was due to very small tidal volumes delivered during PEEPH.

The PEEPH of 20 cm H2O for BiLevel lambs was selected based on our previous experience with these preterm lambs during HFOV, in which MAPs greater than 20 cm H2O were required to initially open the lung (41). The optimum PEEPH and the role of tidal volume at PEEPH remain to be investigated. Regardless, achievement of comparable Paco2 despite lower overall minute ventilation in the BiLevel group suggests more efficient ventilation in this group. We speculate that the use of BiLevel ventilation may preserve airway patency and reduce physiologic dead space, increasing available alveolar surface area for gas exchange. The trend toward a benefit of BiLevel ventilation for reducing the severity of lung injury is promising. Because of the nature of the intervention, caregivers could not be blinded to the mode of ventilation, thus, observer bias cannot be excluded. The findings of this study need confirmation in larger trials with longer duration of ventilation, assessment of potential confounders such as oxygen toxicity, and blinding of outcome assessors. Further studies would need to include a control group of pressure support ventilation, thereby adjusting for potential benefits of pressure support—a feature that was absent in our A-C5 group. Finally, transferability of our findings to human neonates can only be addressed in clinical trials powered to consider the effect of BiLevel ventilation on respiratory outcomes in the setting of other factors contributing to nature and severity of lung injury such as prenatal corticosteroids and chorioamnionitis.

In conclusion, in a preterm lamb model of RDS, BiLevel ventilation provides gas exchange equivalent to A-C5 ventilation. The trend toward reduced markers of ventilator-associated lung injury using BiLevel in preterm lambs warrants further investigation of this mode as a treatment for neonatal RDS.