We studied 19 consecutive unselected patients who met the ARDS criteria of the American European Consensus Conference 21. Exclusion criteria were refusal of consent, age under 18 years, chronic respiratory insufficiency (chronic obstructive pulmonary disease, asthma, restrictive respiratory insufficiency), intracranial hypertension, bronchopleural fistula, and the persistence of unstable hemodynamics despite appropriate support therapy. Patients were orally intubated, sedated with remifentanil (0.2 to 0.4 μg/kg per minute) and midazolam (4 mg/hour), paralyzed with cis-atracurium (15 mg/hour), and ventilated with an Evita 2 Dura ventilator (Dräger, Lübeck, Germany). All patients were equipped with a radial or femoral arterial catheter (Arrow Inc., Erding, Germany). pH, arterial partial pressure of oxygen (PaO₂), and arterial partial pressure of carbon dioxide (PaCO₂) were measured using an IL BGE™ blood gas analyzer (Instrumentation Laboratory, Paris, France). The patients were on volume-controlled mechanical ventilation with a V_T of 6 mL/kg of dry body weight and the highest respiratory rate allowing the maintenance of a PaCO₂ of less than or equal to 46 mm Hg without intrinsic positive end-expiratory pressure (PEEP) 10. The fraction of inspired oxygen (FiO₂) was set at 1, Ti/Ttot (ratio of time of inspiration to total time of breath) at 33%, and the PEEP at 3 cm H₂O above the lower inflection point (LIP) of the pressure-volume (P-V) curve 22 or at 10 cm H₂O in the absence of LIP.

Before the beginning of the study, volemic status of the patients was checked according to pulmonary artery catheter (if the patient needed one before study inclusion) or echocardiography. If necessary, fluid administration or vasopressor adaptation was performed. During the protocol, no fluid administration or vasopressor modification was allowed (in the absence of a life-threatening episode).

Following a 5-minute period of mechanical ventilation in zero end-expiratory pressure (ZEEP), mechanical ventilation was reset with PEEP 3 cm H₂O above the LIP. Following a 15-minute period of mechanical ventilation in PEEP, cardiorespiratory parameters were recorded and alveolar recruitment was measured by the P-V curve method 17232425. After the collection of these data, patients were randomly assigned to benefit from one of the two RMs. Following the first RM, the patient was ventilated with the initial ventilator settings. Cardiorespiratory and RV measurements were performed 5 and 60 minutes after RM. Before the second RM, a 5-minute period of ZEEP ventilation was performed (return to baseline) followed by a 15-minute period of PEEP ventilation. During both ZEEP periods, if oxygen saturation as measured by pulse oximetry (SpO₂) decreased below 92%, PEEP ventilation with the PEEP set at the initial value was resumed. After measurements of cardiorespiratory parameters and RV, the second RM was performed (crossover). Five and 60 minutes after this second RM, cardiorespiratory and RV measurements were performed. The time course of the protocol is summarized in Figure 1.

CPAP was performed by imposition of a pressure of 40 cm H₂O for 40 seconds without V_T 2627 (Figure 2a). As previously described 17, our method of performing RM, eSigh, consisted of increasing PEEP 10 cm H₂O above the LIP for 15 minutes, the patient being on volume-controlled ventilation (Figure 2b). If necessary, V_T was decreased to maintain P_plat below the upper inflection point (UIP) or below 35 cm H₂O if UIP could not be identified on the ZEEP P-V curve. During the RM, the maximum peak airway pressure was limited to 50 cm H₂O. In case of severe arterial hypotension (systolic arterial pressure of less than 70 mm Hg) or severe hypoxemia (SpO₂ of less than 80%), the RM was immediately stopped. A positive response to RM was defined a priori as a 20% increase in RV 5 or 60 minutes after RM 28.

PEEP-induced changes in end-expiratory lung volume (EELV) were measured using a heated pneumotachograph (Hans Rudolph, Inc., Shawnee, KS, USA) positioned between the Y-piece and the connecting piece. The pneumotachograph was previously calibrated by a supersyringe filled with 1,000 mL of air. The precision of the calibration was 3%. The respiratory tubing connecting the endotracheal tube to the Y-piece of the ventilator circuit was occluded by a clamp at end-expiration while the ventilator was disconnected from the patient. The clamp was then released and the exhaled volume measured by the pneumotachograph was recorded on a Macintosh Performa 6400 computer (Apple Computer, Inc., Cupertino, CA, USA) using AcqKnowledge 3.7 software (BIOPAC Systems, Inc., Goleta, CA, USA).

P-V curves of the respiratory system were measured on an Evita 2 Dura ventilator (Dräger) using the low constant flow method as described by Lu and colleagues 22. During the maneuver, the peak airway pressure was limited to 50 cm H₂O. P-V curves were measured in ZEEP and PEEP conditions. For each patient, alveolar recruitment was measured using the P-V curve method as follows: the P-V curves in ZEEP and PEEP conditions were constructed. Changes in EELV were then added on each volume that served for constructing the P-V curve in PEEP. The two curves were then placed on the same pressure and volume axes. RV was defined as the difference in lung volume between PEEP and ZEEP at an airway pressure of 15 cm H₂O 29. When patients have a diffuse loss of aeration in computed tomography (CT) scan, RV was the EELV following PEEP release 23.

Lung scanning was performed in the supine position from the apex to the diaphragm by means of a spiral Tomoscan SR 7000 (Philips, Eindhoven, The Netherlands). All images were observed and photographed at a window width of 1,600 Hounsfield units (HU) and a window level of -600 HU. The exposures were taken at 120 kV and 85 mA without contrast material 30. By institutional protocol and as previously described, lung scanning was performed at ZEEP by briefly disconnecting the patient from the ventilator (10 to 20 seconds). Electrocardiogram, pulse oxymetry, and systemic arterial pressure were continuously assessed throughout the CT procedure. The lowest value of hemoglobin oxygen saturation allowed during the imaging exam was 85% 3132.

Qualitative assessment of lung morphology was performed by two independent radiologists (AB and J-MG) by applying the 'CT scan ARDS study group' criteria, which establish three patterns of loss of aeration distribution: focal or lobar, diffuse, and patchy 31. Loss of aeration was defined as a homogeneous increase of pulmonary parenchyma attenuation obscuring the margins of vessels and airway walls 31. Patients showing a lobar or segmental distribution of loss of aeration, with the possibility of recognizing the anatomical structures such as the major fissura or the interlobular septa, were classified as having a focal ARDS 31.

In each patient, heart rate, systemic arterial pressure, and airway pressure were continuously recorded on the BIOPAC system (BIOPAC Systems, Inc.). Fluid-filled transducers were positioned at the midaxillary line and connected to the arterial catheter. Arterial blood pressures were measured at end-expiration and averaged over five cardiac cycles. The compliance of the respiratory system was calculated by dividing the V_T by the P_plat minus intrinsic PEEP.

The statistical analysis was performed using Statview 5.0 software (SAS Institute Inc., Cary, NC, USA). All data are expressed as mean ± standard deviation (SD). Baseline clinical characteristics were compared between RMs using the Student t test for parametric data and the Mann-Whitney U test for non-parametric data. After the verification of the normal distribution of quantitative data using the Kolmogorov-Smirnov test, changes in cardiorespiratory parameters were analyzed by a two-way analysis of variance for repeated measures (at baseline and 5 minutes and 1 hour after RM) and one grouping factor (RM method: CPAP and eSigh) followed by a Student-Newman-Keuls post hoc comparison test. The statistical significance level was fixed at 0.05.

The design of the present study (crossover study with the patient being his own control) required the return to baseline ventilation between each RM (ZEEP for 5 minutes). Such a design raises several questions. Was 5 minutes of ZEEP ventilation long enough to return to control values? Was it safe enough for ARDS patients? Is a short period of ZEEP ventilation really representative of conditions encountered in clinical practice? RV and oxygenation were not different at the two baselines (Table 2 and Figure 4), suggesting that the short period of derecruitment resulting from ZEEP ventilation was long enough to return to comparable conditions before each RM. In each individual patient, the 5-minute period of ZEEP ventilation could be achieved without severe oxygen desaturation imposing the reinstitution of PEEP (as anticipated in the study protocol). In clinical practice, despite the efforts of the medical team to limit episodes of acute derecruitment, such conditions nevertheless occur in patients with ALI: accidental disconnection from the ventilator, open-circuit endotracheal suctioning 33, endobronchial fiberoptic procedure with or without bronchoalveolar lavage, blind mini-bronchoalveolar lavage for the diagnosis of ventilator-associated pneumonia 34, and ventilator malfunction requiring ventilator replacement and changes of tracheostomy tubes and ventilator circuits. We recommend that, following such events, RMs be performed 1033, and therefore the experimental design of the present study can be considered as of clinical relevance.

In this study, we compared two different RM methods. The first one is the widely used CPAP 40 cm H₂O for 40 seconds 2635. We compared this method with an eSigh performed in volume control ventilation. In previous studies 3637, a conventional form of sigh was found to be inadequate as a recruitment method in ARDS lungs. Inflating pressure during a conventional sigh, though perhaps sufficient in magnitude, is exerted on the lung only briefly. This brevity of pressure application, in light of current knowledge, would not re-aerate and/or splint lung units with a heightened collapsing tendency 38. This limitation of a conventional sigh was shown again in a study by Pelosi and colleagues 36, in which the effect of improved oxygenation and decreased lung elastance seen during a sigh period was soon lost after its discontinuation. The PEEP level set after sigh was probably insufficient in this study. Safety and efficacy of an eSigh were established in several studies 11171939. As previously reported by our group 17 and in the present study, this method increased alveolar recruitment and oxygenation in ARDS patients without respiratory or hemodynamic complications.

RM-induced changes in hemodynamic parameters were limited to a decrease in arterial pressure during RM in non-responders. But in this study, patients did not benefit from cardiac output monitoring (that is, pulmonary artery catheter or echocardiography). This could underestimate the hemodynamic impact of RM 40. CPAP interruption, due to a drop in arterial pressure below 70 mm Hg, was required in two patients, whereas eSigh was well tolerated, with a smaller decrease in blood pressure. This adverse event was previously described, but it underscores a major concern for routine use of this procedure. In 16 patients after open heart surgery, Celebi and colleagues 41 have already described this difference between CPAP and high PEEP recruitment methods.

The present study shows that only eSigh significantly increases RV. Changes in these parameters are more significant than raw data. It must be pointed out that, at baseline, PEEP level was optimized according to the P-V curve. So PEEP-induced alveolar recruitment and EELV were relatively high at baseline; RM-induced RV appears inferior to that obtained with a standardized low PEEP. RV was assessed by the P-V curve method 29. In a previous study, Lu and colleagues 23 compared this method with the reference method (CT scan) and showed that RV measured by P-V curve is highly correlated with RV measured by CT scan, but the P-V curve method underestimates recruitment in patients with diffuse loss of aeration. When the whole lung is poorly or not aerated, PEEP-induced alveolar recruitment is exactly PEEP-induced changes in EELV. A further study, based on CT measurement of lung recruitment, is required to definitively confirm these results.

As previously demonstrated for PEEP and RM 1742, a weak but statistically significant correlation was found between RM-induced alveolar recruitment and RM-induced improvement in arterial oxygenation (Figure 5). In fact, alveolar recruitment is an anatomical phenomenon depending exclusively on the penetration of gas into poorly or non-aerated lung regions, whereas arterial oxygenation is a complex physiologic parameter depending on multiple factors such as lung aeration, regional pulmonary flow, mixed venous oxygen saturation, and cardiac index 4.

Changes in RV and increases in oxygenation are higher with eSigh versus CPAP. Different hypotheses may be proposed to explain these facts. First, alveolar recruitment is a time-dependent phenomenon and procedure duration could influence the response to RM. One CPAP may not be sufficient, and perhaps two or three consecutive CPAPs should be used 43. Second, several studies based on CT scan, P-V curves, or gas exchange have demonstrated that recruitment is a continuous and progressive phenomenon that depends not only on PEEP, but also on peak inflation pressure 44. eSigh was performed for 15 minutes with 3 cm H₂O P_plat below CPAP, but 7 cm H₂O P_max above CPAP. A significantly higher P_max may explain, in part, why 5 patients were CPAP responders whereas 11 were eSigh responders. During mechanical ventilation, a reduction in V_T decreases lung recruitment 8. We can hypothesize that RM without V_T failed to achieve alveolar recruitment. The third point is the pressure level during RM. The use of CPAP as an RM has been described previously 26 using 40 cm H₂O for all patients. Effective pressure, during RM, is different if PEEP is set at 8 or 18 cm H₂O. We believe that it is important to have knowledge of the pulmonary mechanics of patients in order to adapt the pressure level for optimal lung recruitment.

In ARDS patients ventilated with a lung-protective strategy, the effects of RM are discussed. In 17 patients with high PEEP and low V_T, Villagrá and colleagues 39 concluded that RMs have no short-term benefit on oxygenation and that regional alveolar overdistension capable of redistributing blood flow toward non-aerated lung regions can occur during RM. In 22 patients, Grasso and colleagues 45 found an increase in oxygenation and RV with diminished elastance in responders (early ARDS) after RM in patients with lung-protective strategy. PaO₂/FiO₂ decreased from 480 mm Hg (2 minutes after RM) to 300 mm Hg 20 minutes later. The mean PEEP value was 9 ± 2 cm H₂O. In the present study, in which the mean PEEP value was 14 ± 2 cm H₂O, we found significant effects of RMs and these effects persisted after 1 hour. As previously reported 46, our data suggest that lung morphology predicts the response to RM, but not baseline ventilator strategy or ARDS history 25. Indeed, patients with a diffuse loss of aeration are responders to RM, whereas non-responders have a focal loss of aeration predominant in the inferior and posterior lung areas 4247. In these patients, performing RM could induce overinflation of the previously healthy lung 17. Moreover, a high level of PEEP is fundamental to ensure the prolonged effect of RM. The mean PEEP was 5 cm H₂O higher than that of the study performed by Grasso and colleagues 45. Furthermore, FiO₂ was set at 1 throughout this study to 'standardize' measurements. In 'real life', a reduction in FiO₂ will limit oxygen-induced loss of aeration.