Current ventilatory strategies for 'lung-protective' ventilation in acute respiratory distress syndrome (ARDS) include low tidal volumes to avoid alveolar overdistension, adequate end-expiratory lung volume by positive end-expiratory pressure to prevent end-expiratory alveolar collapse, and inspiratory pressure limitation to minimize further stress and strain to the lung fibrous skeleton 1. The excessive and nonphysiological strain to lung structures is caused by high transpulmonary pressures (TPP), which in turn depend on the respiratory system elastance 2. High-frequency oscillatory ventilation (HFOV) offers several advantages over conventional ventilation. Oscillations in HFOV are superimposed on a constant fresh gas flow and induce active inspiratory and expiratory gas movement, resulting in high constant mean airway pressures at low tidal volumes. Atelectatic lung regions are reopened by the continuous distending airway pressure (CDP), and the superimposed small oscillations provide alveolar gas exchange for CO₂ removal 3. Recruitment maneuvers are beneficial at initiation of HFOV to ensure sufficient gas exchange area in the diseased lung 4.

In most clinical studies, HFOV is initiated by an initial lung volume optimization procedure (LVOP) with a CDP level 5 mbar above the effective mean airway pressure previously used at conventional ventilation 5678. The CDP is thereafter increased in a stepwise manner (2–5 mbar every 15–30 minutes) up to the maximum increase of PaO₂ or up to a predetermined CDP. This maneuver is followed by a stepwise reduction of the CDP (2 mbar every 30 minutes up to 4 hours) to maintain alveolar patency. Recruitment in a stepwise fashion is effective and safe with regard to hemodynamic impairment, but is also time consuming in adjusting an effective CDP.

Preclinical and clinical trials have been presented recently that used a recruitment maneuver with a high CDP, followed by a stepwise decrease of the airway pressure. Sedeek and colleagues used a continuous positive airway pressure of 50 mbar for lung recruitment in lung-lavaged sheep. The CDP was then set according to the maximal compliance on the pressure-volume 9. The Treatment with Oscillation and an Open Lung Strategy (TOOLS) trial used a standardized HFOV protocol in patients, which showed that the combination of HFOV and a high initial recruitment maneuver (with interrupted HFOV) resulted in a rapid and sustained improvement in oxygenation. The mean airway pressure was then titrated in a decremental fashion according to the oxygenation response 10.

In the present study we investigated the immediate effect of a modified LVOP by means of ongoing HFOV on hemodynamics and oxygenation prior to its clinical application. To demonstrate the feasibility and safety of this lung optimization procedure, HFOV was initiated with the CDP set to 45 mbar (for 2.5 minutes) during ongoing oscillation in six pigs after saline lung lavage, simulating an early-phase ARDS model. The CDP was thereafter reduced in a stepwise fashion of five mbar every five minutes with simultaneous measurement of the TPP. The effects upon gas exchange, systemic hemodynamics, and the pulmonary shunt proportion were observed at each CDP and TPP level.

All six animals (25 ± 2 kg bodyweight, mean ± standard deviation) completed the entire study protocol. Hemodynamic and gas exchange variables before and after induction of lung damage are presented in Table 1. The MAP and PAOP before initiation of HFOV complied with the predefined requirements; that is, PAOP of 13 (11–13) mmHg and MAP of 83 (82–85) mmHg. Repetitive lung lavages decreased the PaO₂/FiO₂ ratio (PCV pre-lavage, 559 (535–658) Torr vs PCV post-lavage, 99 (56–128) Torr; P < 0.05), and increased the pulmonary shunt fraction (PCV pre-lavage, 9.97% (8.8–11%) vs PCV post-lavage, 51.8% (49–55%); P < 0.05). The oxygenation index increased from PCV pre-lavage (1.4 (1.2–2)) to PCV post-lavage (16.3 (14.6–21.3)) (P < 0.05). The extravascular lung water increased from PCV pre-lavage (290 (241–311) ml) to PCV post-lavage (420 (354–463) ml) (P < 0.05).

The present experimental study investigated the effects upon gas exchange, TPPs, and hemodynamics of a modified HFOV initiation protocol to optimize the lung volume in a porcine model of acute lung injury. The protocol consists of a fast recruitment maneuver followed by a stepwise decrease of CDP. The fast stepwise reduction of CDP allows the identification of the lowest TPP level required to maintain improvement of oxygenation in each animal. This approach therefore offers an effective reduction of pulmonary shunt fraction and improvement in oxygenation without relevant adverse hemodynamic effects within a very short time.

In most clinical and experimental studies, HFOV is initiated with a LVOP with an initial CDP level 5 mbar above the mean airway pressure previously used in conventional ventilation 56812. The CDP is then increased in a stepwise fashion (2–5 mbar steps) every 15–30 minutes until a maximum in PaO₂ or a predetermined CDP is reached. Our study presents a fast lung optimization procedure that could be technically applied easily in the clinical scenario. In contrast to recent studies in animals and in patients, the CDP was directly set to 45 mbar for 2.5 minutes. The lowest possible TPP that still assures the improved oxygenation is subsequently titrated.

Recruitment maneuvers are beneficial at the initiation of HFOV to ensure sufficient gas exchange area in the diseased lung 4. In the Treatment with Oscillation and an Open Lung Strategy trial, the combination of HFOV and a high initial mean airway pressure recruitment maneuver without ongoing HFOV resulted in a rapid and sustained improvement in oxygenation 10. Although sustained inflation pressures up to 55 mbar are necessary to overcome the opening pressure of collapsed alveoli 13, the criteria of a fully recruited lung, defined as pulmonary shunt proportion < 0.1 14, was achieved at a CDP of 45 mbar. A CDP of 32 ± 6 mbar in piglets was able to reduce the shunt fraction < 10% 15. During the present study a single pressure step-up for alveolar recruitment was performed, and therefore no estimate can be made of whether a lower CDP and TPP would have been adequate to reopen the lung, and whether the TPP required to open alveolar units exceeds the TPP required to maintain alveolar patency 1516. Assuming that the TPP stays above a critical closing pressure, a significant alveolar derecruitment cannot be expected 1718. In other experimental studies with the lavage-injury animal model, the mean airway pressures were set according to the pressure-volume curve for the setting of the CDP during HFOV 91619. In their study, Sedeek and colleagues repeated recruitment maneuvers with a continuous positive airway pressure of 50 mbar for 1 minute until the PaO₂ was stable with a CDP set according to the maximal compliance on the pressure-volume curve 9.

We intended to use an alternative method without the pressure-volume curve. The P_aO₂/F_iO₂ ratio was therefore used as a criterion to identify the lowest TPP preventing alveolar derecruitment during the stepwise decrease of CDP. A TPP below this threshold was associated with an increased shunt > 10 and with an increased PaCO₂. An increased PaCO₂ at unchanged HFOV settings indicates a decreased alveolar surface available for gas exchange.

Previous studies in the lavage animal model showed that CO decreased at high mean airway pressures. Interestingly, better oxygenation values in those studies were found at lower mean airway pressure, suggesting in our study that eventually a lower CDP eventually would have been sufficient 1619. The lung recruitment maneuver had a marked effect on hemodynamics. These effects can easily be corrected by volume or by vasoactive drug usage. There was no relevant decrease of the intrathoracic blood volume as an indicator of reduced venous preload 20. Also, the fluid regime before initiation of HFOV may have attenuated a reduction of venous return.

At a CDP of 45 mbar, the SV and CO were decreased and, simultaneously, the cardiac filling pressures (LVEDP, right atrial pressure, and PAOP) were elevated. This can be explained by a compression of the heart into the cardiac fossa due to the transmission of high TPPs 2122. Impairment of hemodynamics can therefore be explained by the mechanical restriction of the heart. Systemic afterload did not have a major impact on the impaired SV and CO as measured by the systemic vascular resistance. Right ventricular dysfunction may occur when high airway pressures are applied, and the consecutive right ventricular output and left ventricular filling are impaired leading to SV and CO decreases. As the pulmonary vascular resistance was unaffected by a CDP of 45 mbar and the right ventricle was able to generate a pressure gradient (mean pulmonary artery pressure–right atrial pressure), we excluded right ventricular failure. During the deceleration CDP trial, the cardiac filling pressures returned to similar values as measured before the lung recruitment maneuver. This observation assures us that the hemodynamic effects are related to a pressure transmission of the CDP and the TPP.

The safe window for plateau pressures during conventional ventilation is considered between 30 and 35 mbar, but even those values can lead to a harmful TPP. Depending on the elastance of the respiratory system, volutrauma may occur from cyclic tidal overdistension. No recommendations regarding safe levels for CDP or TPP exist during HFOV. The lowest TPP levels possible, however, should be applied. The measurement and monitoring of the TPP is therefore helpful and of increased interest, as the TPP is the effective distending force of the lung, and ventilator-induced lung injury (VILI) depends on the TPP 2324. By measuring the TPP the mechanical ventilator settings could be set more individually with respect to lung and chest wall mechanical characteristics, which enables the identification of lung recruitment potential in relation to a potential risk of VILI. Such an individual approach may reduce the risk for further lung injury in patients with ARDS undergoing mechanical ventilation 2526. In a clinical scenario, a CDP higher than the safe window for plateau pressures under conventional ventilation should be avoided even if suggested by the LVOP presented in this study. Although the theoretical advantage of HFOV is the avoidance of volutrauma caused by tidal overdistension (due to minimal tidal volumes) in the case of excessive TPP, a compromise between oxygenation and potential VILI should be made to avoid a VILI and to accept lower but adequate oxygenation.

The esophageal pressure measured by an esophageal balloon is used as a surrogate parameter of the pleural pressure for calculation of the TPP. The pleural pressure and therefore the TPP are dependent on the chest wall elastance in experimental patients and ARDS patients. For a given mean airway pressure (CDP) the pleural pressure increases if the chest wall elastance is elevated, and consequently the TPP decreases. Chest wall elastance, however, depends on the pathophysiology of ARDS (for example, high chest wall elastance in extrapulmonary ARDS). The elevated pleural pressure (due to increased chest wall elastance) leads to a lower TPP compared with pulmonary ARDS with normal elastance 1272829.

We found a pressure difference of about 10 mbar between the applied CDP and the resulting TPP. The difference in mean intrathoracic pressure and CDP in HFOV increases with the oscillatory frequency, decreasing the tracheal tube diameter and the relative duration of the inspiratory time 30. The inspiration time:expiration time ratio of 33% in our study results in an increased tracheal tube resistance and, consequently, in a decreased mean intrathoracic pressure and TTP.

A major limitation of this study is that it was performed in a lavage animal model of ARDS and not in patients. Large-animal models, however, have contributed greatly to the understanding of the basic physiology of HFOV and its clinical application during severe lung injury 31. ARDS in patients is rarely and solely a result of surfactant deficiency only. Early institution of HFOV, however, and institution of HFOV in patients with high potential for recruitment is theoretically beneficial to outcome in ARDS 323334. The lavage model is easily recruitable and is probably the most comparable experimental ARDS model for early-phase ARDS. Although it is primarily a model of surfactant depletion, it has been shown that mechanical ventilation after lavages leads to neutrophil infiltration, cytokine expression, and capillary-alveolar protein leak (as indicated by the increased extravascular lung water in this study) 353637. The present study only observed immediate changes of hemodynamics and lung oxygenation. High airway pressures and subsequent lung overdistension may lead to mediator release and to VILI. These adverse events were not accessible in the present study. Also, we cannot report whether the optimized gas exchange remained stable over a longer time period. This raises the question of how often such recruitment maneuvers should be applied in clinical practice.

This short-term experimental observation study demonstrates the feasibility of a fast and safe lung optimization procedure during uninterrupted HFOV. This lung optimization procedure was well tolerated and resulted in a dramatic improvement of oxygenation. No clinically relevant adverse effects on hemodynamics or barotrauma occurred during the time period of approximately 1 hour. The stepwise decrease of CDP allowed the determination of the lowest TPP level necessary to maintain adequate oxygenation.

• A novel lung optimization procedure in uninterrupted HFOV is presented.

• The LVOP consists of a single CDP recruitment maneuver during uninterrupted HFOV, followed by a stepwise decrease of the CDP.

• This LVOP allows the determination of the lowest TPP that still maintains adequate pulmonary gas exchange.

• This LVOP shows no adverse effects in hemodynamics and oxygenation during its intervention in a porcine lavage ARDS model.

ARDS = acute respiratory distress syndrome; CDP = continuous distending pressure; CDP20 = continuous distending pressure of 20 mbar; CDP45 = continuous distending pressure of 45 mbar; CO = cardiac output; FiO₂ = fraction of inspired oxygen; HFOV = high-frequency oscillatory ventilation; LVEDP = left ventricular end-diastolic pressure; LVOP = lung volume optimization procedure; MAP = mean arterial pressure; PaCO₂ = arterial partial pressure of carbon dioxide; PaO₂ = arterial partial pressure of oxygen; PAOP = pulmonary occlusion pressure; PCV = pressure-controlled ventilation; SV = stroke volume; TPP = transpulmonary pressure; VILI = ventilator-induced lung injury.

The authors declare that they have no competing interests.

JK designed the study protocol, collected data, and drafted the protocol. MD designed and participated in the study protocol, collected data, and performed statistical analysis. SY collected data. KM coordinated the study and revised the manuscript.