INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During the last 20 yr, there has been a remarkable evolution in the ventilatory treatment of acute respiratory distress syndrome (ARDS). In the 1970s, high tidal volume (VT) and pressures were the rule, and the only recognized side effect was hypocapnia (1). The most important damaging factor for the lung was considered the high inspiratory oxygen fraction (FI_O2), as emphasized by the ECMO trial (2). In the 1980s, negative effects of high pressure/volume ventilation were demonstrated in animal models (3) and the concept of "lung rest" (4) was progressively accepted. This led to an increasing use of lower VT in the 1990s, to allow a more gentle treatment of the diseased lung. The accepted side effect was the hypercapnia (5), completely reversing the 1970s concepts.

Also, positive end-expiratory pressure (PEEP) has been often revisited during the last 20 yr, from the concepts of best PEEP (6), super PEEP (7), and minimal PEEP (8), to the current concepts, for which PEEP should be adequate to keep open the lung (9) and to prevent intratidal collapse and decollapse (10).

As a result of this long-lasting process of physiology-based conceptual evolution, the Consensus Conference of the American College of Chest Physicians (ACCP) recommended the following guidelines to ventilate patients with ARDS (13): (1) adequate PEEP to support oxygenation, without deleterious effects and evaluated by empirical trial; (2) low VT to avoid a plateau pressure higher than 35 cm H₂O; (3) FI_O2 below 60%. The low VT approach alone has been tested in prospective randomized trials (14, 15) with at least, in part, discouraging results. The lung protective strategy (16) (i.e., PEEP set 2 cm H₂O higher than the inflection point of the pressure-volume curve of the respiratory system and low VT) showed a significant improvement in survival compared with the control subjects, who experienced, however, an unexpected high mortality of 70%.

Although the pendulum is now towards the "low VT, relatively high PEEP and hypercapnia," we have forgotten that more than 30 yr ago it was clearly shown that low VT during anesthesia in normal lungs leads to progressive lung atelectasis with consequent hypoxemia (17). The atelectasis occurring with time, in this case, was not likely due to compression (which is an immediate phenomenon), but rather to a progressive gas reabsorption caused by a regional gas uptake greater than supply. The lung collapse was, in fact, directly related to the degree of hypoventilation and could be prevented by large tidal volumes, even delivered intermittently. Several reports came to the same conclusion, including studies in patients with ARDS (18, 19).

Thus, in this study, we tested the following hypotheses: (1) that the current recommended ventilation, i.e., PEEP thought to be sufficient to keep the lung open at end-expiration and inspiratory plateau pressure equal or lower than 35 cm H₂O maintain unresolved atelectasis; (2) that sighs should provide sufficient opening pressure to further recruit the lung and should provide sufficient volume to prevent new reabsorption atelectasis, which otherwise may develop if too low VT is used, despite a PEEP selected as currently recommended (13); (3) that all the above phenomena should be quantitatively different in ARDS from pulmonary and extrapulmonary origin as they may have, for a given plateau pressure, different transpulmonary pressures because of the differences in chest wall elastance (20).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study was approved by the institutional review board of our hospital, and informed consent was obtained from the patients' next of kin.

Study Population

We studied 10 consecutive, unselected patients, who met the ARDS criteria of the American European Consensus Conference on ARDS (21). None of them had asthma, chronic lung diseases, or cardiogenic pulmonary edema. The main demographic and clinical characteristics and the time from onset, as well as outcome, are summarized in Table 1.

Study Design

All the patients were nasotracheally intubated with a cuffed endotracheal tube, ventilated with a Siemens Servo ventilator 300 C (Siemens, Erlanger, Germany) and had an arterial and pulmonary artery thermodilution catheter inserted. During the study, the patients were sedated with fentanyl and diazepam and paralyzed with pancuronium bromide. The entire study lasted 4 h, which were divided in three periods: baseline period (2 h), sigh period (1 h), and no sigh period (1 h). Measurements were taken at the end of the baseline period, at 30 min and 60 min of the sigh period, and at 30 min and 60 min of the no sigh period.

Baseline Period

Our aim during this period was to provide the kind of mechanical ventilation that is recommended in ARDS, i.e., PEEP level thought to be sufficient to keep the lung open, and VT such as to maintain the plateau pressure equal or below 35 cm H₂O, accepting possible hypercapnia as a side effect, as recommended by the ACCP Conference (13) and by Amato and colleagues (16). To select PEEP we performed, prior to the baseline period, two maneuvers: a static pressure-volume (P-V) curve of the respiratory system, and a PEEP trial.

Pressure-volume curve. The static P-V curve was determined without previous homogenization for lung volume hystory, using an automatic supersyringe, and inflating the lung stepwise up to 1.4 L, starting from atmospheric pressure (100 ml per step, 2-s interval between the steps) (22). From the P-V curve the following parameters were derived: (1) Starting compliance (Cstart). This was computed as the ratio between the first 100 ml inflation and the corresponding pressure. (2) Inflation compliance (Cinf). This was computed as the slope of the P-V curve during inflation in its most linear segment. (3) Lower inflection point (LIP). This was computed as the pressure corresponding to the intersection between Cstart and Cinf lines. From these parameters, two possible PEEP levels for clinical use may be identified and defined: (1) "Pflex-PEEP" as PEEP 2 cm H₂O higher than the inflection point; (2) "linear-PEEP" as PEEP equal to the minimal pressure at which the slope of the P-V curve becomes linear (22).

PEEP trial. The PEEP trial was performed with the VT in clinical use (6 to 8 ml/kg), and applying 5, 10, 15, 20 cm H₂O PEEP. Inspiratory oxygen fraction and respiratory rate were maintained constant. If at a given PEEP level, the plateau pressure exceeded 35 cm H₂O, the VT was reduced until the plateau pressure was equal to 35 cm H₂O. This set (applied PEEP and VT with plateau not exceeding 35 cm H₂O) was maintained for at least 20 min, after which blood gases were determined. We defined "trial-PEEP" the PEEP at which the greater Pa_O2 improvement was observed. The Pa_O2 was considered arbitrarily not improved, if its increase was equal to or lower than 10 mm Hg.

Selection of PEEP. In this study we chose to apply the "trial-PEEP" according to the ACCP Conference recommendations (13). This (see RESULTS) was the highest PEEP selected by the three different methods and was therefore most likely to prevent collapse at end-expiration.

Ventilatory setting. After selecting PEEP and VT, the ventilatory setting has been defined and reported in Table 2. No intrinsic PEEP was detected in this series of patients.

Sigh Period

To provide a sigh function with the Siemens Servo ventilator 300 C, a software program was developed by one of us (P.C.), which determined automatically the volume or the pressure of the sigh, its rate (1, 2, 3, . . . /min), and its sequence (consecutive sighs or separated). We chose to deliver a sequence of three consecutive sighs per minute with the volume such as to reach 45 cm H₂O plateau pressure in volume control mode. The number of no-sigh VT was automatically reduced to maintain a minute ventilation (E) similar to that at baseline. The ventilatory setting during the sigh period is summarized in Table 2. As shown, PEEP, Paw, E, and FI_O2 were identical during the baseline period and the sigh period.

No-Sigh Period

After the sigh period, the baseline ventilation was reinstituted for 1 h, with the same ventilatory setting implemented during the baseline ventilation (Table 2).

Gas-Exchange and Hemodynamics

Gas tensions and pH were analyzed immediately after sampling from arterial and mixed venous blood. The venous admixture (VA/) was computed by the shunt equation, assuming a respiratory quotient equal to 1. Mean arterial (), pulmonary artery (), pulmonary wedge (), and central venous (PCV) pressures were measured with pressure transducers zeroed at the midaxillary line. Cardiac output (CO) was measured in triplicate by the thermodilution method, and the cardiac index (CI) was computed normalizing the CO for the body surface area.

The oxygen consumption (O₂) was computed using the reverse Fick equation, and the dead space fraction (VD/VT) was estimated rearranging the Bohr equation and assuming a respiratory quotient = 1, according to the following formula (23):

in which P_B is the atmospheric pressure. This formula implies that oxygen consumption equals the metabolic CO₂ production, and the latter equals the CO₂ excreted. The assumption that CO₂ equals CO₂ excreted holds true if the system is at steady state; the equilibration, when changing the alveolar ventilation, takes 20 to 30 min (24), and our samples were taken after 1 h of sighing, with the system at new steady state.

The effective ventilation/perfusion ratio was defined as: VA/CO_eff = (VE  VD)/CO · (1  VA/), in which VA is the alveolar ventilation and CO_eff is the cardiac output effectively perfusing the ventilated regions (i.e., CO  the shunt flow). The VA/CO_eff reflects the average ventilation to perfusion ratio occurring in the ventilated/perfused pulmonary units.

End-Expiratory Lung Volume and Lung Recruitment

The end-expiratory lung volume (EELV) was measured using a simplified closed-circuit helium dilution method (20) during an end- expiratory pause, maintaining the PEEP in use and inflating 1 to 1.5 L into the respiratory system 10 to 15 times. We defined recruitment (+EELV) as the differences between baseline EELV and the EELV measured at the end of the sigh period, and we defined derecruitment (EELV) as the differences between the EELV measured at the end of the sigh period and the EELV measured 1 h after the sigh interruption. It is important to emphasize that the conditions of the EELV measurement were identical (at end-expiration with identical PEEP). Thus the differences in EELV represent the lung volume gained or lost because of the different operating conditions (sighing and no sighing).

Respiratory Mechanics

Details have been previously published and will only be summarized here (20). Airway pressure (Pao) was measured proximal to the endotracheal tube, and the esophageal pressure (Pes) was measured with an esophageal balloon inflated with 0.5 to 1 ml of air. Gas flow was recorded with a heated pneumotachograph connected to a differential pressure transducer. Volume was obtained by digital integration of the flow signal. Both flow and pressure signals were recorded on a personal computer and processed via an analog-to-digital converter (Colligo, Elekton, Italy) at a sample rate of 200 Hz and were stored on diskettes for subsequent computer analysis.

Static elastance of total respiratory system, lung, and chest wall. Static elastance of the total respiratory system (Est,rs) was computed as Est,rs = Pao/VT, where Pao is the difference between end- inspiratory and end-expiratory airway pressure. Static elastance of the chest wall (Est,w) was computed as Pes/VT, where Pes is the difference between end-inspiratory and end-expiratory esophageal pressure. Static lung elastance (Est,L) was calculated as Est,L = Est,rs  Est,w.

Resistance of the total respiratory system, lung, and chest wall. Maximum (Rmax,rs) and minimum (Rmin,rs) resistance of the respiratory system were computed from Pao as (Pmax  P₂)/ and (Pmax  P₁)/, where Pmax is the maximal pressure value after occlusion, P₁ is the pressure recorded after the immediate drop from Pmax, P₂ is the plateau pressure, and is the flow immediate preceding the occlusion. Rmin,rs represents the "ohmic" resistive component of the respiratory system, and Rmax,rs includes Rmin,rs plus the additional respiratory resistance (DR,rs). Because there was no appreciable drop in Pes immediately after the occlusion, Rmin,rs reflects essentially the airway resistance (Rmin,L) and minimum chest wall resistance (Rmin,w) can be considered negligible. As a consequence, maximum chest wall resistance (Rmax,w) is entirely due to the viscoelastic properties of the chest wall tissues (i.e., Rmax,w = DR,w). Additional resistance of the lung (DR,L) was obtained as DR,rs  DR,w, whereas the sum of Rmin,L + DR,L gives the maximum lung resistance (Rmax,L). Rmax,rs, Rmax,L, and Rmin,L include the endotracheal tube resistance.

Statistical Methods

All data are expressed as mean ± standard deviation. Comparisons between different periods were performed using the analysis of variance. Individual comparisons with baseline were performed using the paired t test; Bonferroni's correction was applied for multiple comparisons. To compare pulmonary and extrapulmonary ARDS, Wilcoxon's test for unpaired data was applied. The least-squares method was used to perform linear regression analysis (25); a p value lower than 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

PEEP Selection

The results obtained with the PEEP trial and the values of "linear-PEEP" and "Pflex-PEEP" derived from the P-V curve of each individual patient are reported in Table 3. The PEEP trial, performed as described in METHODS, allowed us to select a PEEP ranging between 10 and 15 cm H₂O. Interestingly, we did not find any correlation between the PEEP values derived from the P-V curve, i.e., "Pflex-PEEP" and "linear-PEEP" and the "trial-PEEP" (r = 0.007 and r = 0.25, respectively). As shown in Table 3, the Pa_O2 at 20 cm H₂O of PEEP was not significantly different from the Pa_O2 at 15 cm H₂O of PEEP. More interestingly, the PEEP selected with the "trial-PEEP" led to an average Pa_O2 of 95.5 ± 15.9 mm Hg, which was significantly (p < 0.05) higher than the Pa_O2 recorded at 20 cm H₂O of PEEP (86.8 ± 17.7 mm Hg).

Gas Exchange

The behavior of the most important gas exchange variables is shown in Figure 1. It is important to remember that PEEP, FI_O2, mean airway pressure, and minute ventilation were kept constant throughout the study (Table 2). Despite that, the introduction of three sighs at 45 cm H₂O plateau pressure led to a significant improvement of Pa_O2 (average increase at 60 min, 44.8 ± 16.0 mm Hg [50.1 ± 21.7%], range from 16 to 65 mm Hg), together with a significant reduction in VA/ (average decrease at 60 min, 10 ± 10% [26.4 ± 26.4%], range from 0.1 to 29%) and Pa_CO2 (average decrease at 60 min, 3.6 ± 4.4 mm Hg [6.6 ± 8.5%], range from 2.2 to 11.3 mm Hg). During the study, pHa did not change significantly, being 7.34 ± 0.05 at baseline, 7.34 ± 0.06 at 60 min after application of sigh, and 7.33 ± 0.06 at 60 min after interruption of sigh. However, as shown in Figure 1, all the gas exchange variables returned to the baseline values by 30 min after sigh interruption, and they remained stable until the end of the study. It is worth noting, however, that the increase in Pa_O2 and the decrease in Pa_CO2 and shunt during sighing were not significantly correlated with the opposite change during the no-sigh period (r = 0.56, r = 0.32, r = 0.38, respectively), i.e., the patients in whom gas exchange improved were not the same as those whose gas exchange deteriorated after sigh interruption.

View larger version (52K): [in this window] [in a new window]   Figure 1.   Data are presented as mean ± SD. PaO2 (open bars) and PaCO2 (stippled bars) at the end of baseline period, at 30 min, and at 60 min of the sigh period, and at 30 min and 60 min of the no-sigh period. In the upper panel is shown the time course of the venous admixture ( VA/). *p < 0.05, **p < 0.01, compared to the baseline period.

Respiratory Mechanics

The behavior of the most relevant mechanical variables is reported in Figure 2. As shown, during the baseline period, the Est,rs and Est,L were abnormally elevated (34.3 ± 16.9 and 27.0 ± 16.7 cm H₂O · L¹, respectively), whereas Est,w was only slightly increased compared with normal (7.3 ± 3.9 cm H₂O · L¹). The EELV at PEEP was 1.49 ± 0.58 L. 

View larger version (47K): [in this window] [in a new window]   Figure 2.   Data are presented as mean ± SD. Static elastance of the total respiratory system (Est,rs) (stippled bars), of the lung (Est,L) (open bars), and of the chest wall (Est,w) (hatched bars), at the end of the baseline period, at 30 min, and at 60 min of the sigh period, and at 30 min and 60 min of the no-sigh period. In the upper panel is shown the end-expiratory lung volume (EELV) measured at the end of baseline period and at the end of the sigh period (60 min) and at the end of the no-sigh period (60 min). *p < 0.05, **p < 0.01, compared with baseline period.

Sighing led to a consistent and significant reduction in Est,rs and Est,L (average decrease at 60 min, 6.0 ± 6.6 cm H₂O · L¹ [16.3 ± 9.1%], range from 1.92 to 24.23 cm H₂O · L¹ and 6.1 ± 7.1 cm H₂O · L¹ [23.4 ± 13.9%], from 2.1 to 25.8 cm H₂O · L¹, respectively), whereas Est,w did not change significantly. After interruption of sighing, in the individual patient, all the mechanical variables returned to baseline values. As well as for gas-exchange, the increase in EELV during sigh period was not significantly correlated (r = 0.46) with its decrease during the no-sigh period.

The behavior of the resistances of the total respiratory system, lung, and chest wall is summarized in Table 4. During the sigh period Rmax,rs significantly decreased compared with baseline, mainly because of a reduction in Rmax,L. The reduction in Rmax,L was attributable to a reduction in Rmin,L and DR,L. After interruption of sighing all the resistances returned to baseline values.

Hemodynamics

The behavior of the most important hemodynamic variables is reported in Table 5. As shown, during baseline period we observed elevated CI, PVR, and Ppa and decreased systemic vascular resistances. The introduction of three sighs per minute led to a small but significant decrease of Ppa and PVR (average decrease at 60 min, 2.5 ± 1.9 mm Hg [8.5 ± 6.9%], range from 6 to 0 mm Hg and 30.8 ± 27.8 dyne · s · cm⁵ · m² [16.4 ± 15.6 %], range from 84.1 to 2.9 dyne · s · cm⁵ · m², respectively), which returned to the baseline values after the sigh interruption.

Lung Recruitment and Derecruitment, Gas Exchange, Respiratory Mechanics, and Hemodynamics

Lung volume recruited during the sigh period averaged 0.423 ± 0.370 L (58.8 ± 16.6%), range from 0.043 to 1.111 L, and the lung volume derecruitment after sigh interruption averaged 0.305 ± 0.243 L (19.3 ± 22.7%), range from 0.900 to 0.029 L (Figure 3). Recruitment was associated with a significant increase in Pa_O2, whereas derecruitment was associated with a decrease in Pa_O2 (Figure 3). Similar relationships with recruitment and derecruitment were observed for Pa_CO2 (r = 0.72, p < 0.01), and VA/ (r = 0.70, p < 0.01). Also, alterations of elastances of both the total respiratory system (r = 0.76, p < 0.01) and of the lung (r = 0.76, p < 0.01) were correlated with recruitment and derecruitment. Interestingly the DR,rs, which reflects the viscoelastic properties of the respiratory system, decreased with recruitment and increased with derecruitment (r = 0.69, p < 0.01) as well as the Rmin,L (r = 0.66, p < 0.01). In conclusion, both gas exchange and mechanical changes induced by sighing appeared recruitment-dependent.

View larger version (10K): [in this window] [in a new window]   Figure 3.   Changes of PaO2 (PaO2) as a function of changes of end-expiratory lung volume (EELV). The positive changes of EELV indicate recruitment, the negative changes indicate derecruitment. The relationship is described by the following equation: PaO2 = (2.4 ± 6.2) + (80.0 ± 13.0) · EELV, (r = 0.82, p < 0.01).

As shown in Figure 4, the derecruitment is a function of Pa_CO2 and the effective VA/Q_eff, i.e., more severe the hypoventilation and lower the effective VA/Q_eff ratio greater the derecruitment. Pa_CO2 and effective VA/Q_eff were also correlated (r = 0.84, p < 0.01) and it is interesting to note (Figure 4) that when VA/Q_eff is close to 1, and the Pa_CO2 is close to 40 mm Hg, the derecruitment is close to zero.

View larger version (13K): [in this window] [in a new window]   Figure 4.   Derecruitment after sigh interruption (EELV) as a function of alveolar ventilation to effective cardiac output ratio (VA/ COeff) (upper panel ), and of PaCO2 (lower panel ). The relationships are described by the following equations: EELV (L) = (0.910 ± 0.163) + (0.873 ± 0.225) · VA/COeff, r = 0.81, p < 0.01; EELV (L) = (0.251 ± 0.127) + (0.011 ± 0.002) · PaCO2 (mm Hg), r = 0.86, p < 0.01.

ARDS from Pulmonary and Extrapulmonary Disease

As shown in Table 1, in this series of 10 consecutive patients with ARDS, five had pulmonary and five extrapulmonary ARDS. Despite the limited number of patients, the quantitative differences in most of the variables considered above appear to be relevant. In particular, as previously observed, the Est,w was significantly higher in ARDSexp than in ARDSp (10.6 ± 2 cm H₂O · L¹ and 4.0 ± 1.3 cm H₂O · L¹, respectively, p < 0.01). The potential for recruitment with sigh appeared different between ARDSp and ARDSexp, being 0.160 ± 0.132 L and 0.688 ± 0.343 L, respectively, p < 0.01. Consequently, during the sigh period, the Pa_O2 increase and the VA/ decrease were significantly different between the two populations (Figure 5).

View larger version (17K): [in this window] [in a new window]   Figure 5.   Data are presented as mean ± SD. Differences in recruitment (EELV), in PaO2 (PaO2), and in venous admixture ( VA/) induced by sigh at 60 min in ARDS from pulmonary (ARDSp) and extrapulmonary (ARDSexp) origin. *p < 0.05, **p < 0.01, compared with ARDSp.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main findings of this study were that: (1) setting the ventilatory treatment according to the current recommendations does not exploit the full potential for lung recruitment and oxygenation; (2) introducing 45 cm H₂O pressure-limited sighs leads to further recruitment and improvement in oxygenation; (3) if we consider our three sequential sighs similar to a recruitment maneuver, it is evident that opening the lung for a limited period is not enough to prevent, with time, the loss of gas volume and gas-exchange deterioration, possibly because of reformation of atelectasis at PEEP levels used in this study; (4) the above effects are quantitatively different in pulmonary and extrapulmonary ARDS.

PEEP Selection

Current recommendations suggest to select a PEEP level "adequate" for oxygenation, avoiding excessive stretching, hemodynamic impairment (13), and prevention of intratidal collapse and decollapse (12). However, the method to select this "ideal" PEEP, if it exists, is still being debated. As shown in Table 2, some of the methods used to select PEEP gave different results. The PEEP levels derived from the P-V curve (the "linear-PEEP" and "Pflex-PEEP"), although in line with several previous reports (22, 26, 27), were significantly lower than the "trial-PEEP" and unrelated with the latter. As our goal was to apply the PEEP that most likely would prevent the end-expiratory collapse, we used as criterion for PEEP selection the "best" oxygenation. In fact, in the presence of intratidal collapse and decollapse, the oxygenation should be lower compared with an "open" lung at end-expiration. The PEEP selected according to oxygenation averaged 14 ± 2 cm H₂O, and a further increase to 20 cm H₂O did not further improve the oxygenation but, instead, deteriorated it. This indicates that at our selected PEEP, the lung was likely open throughout the respiratory cycle. Furthermore, we previously found, at the CT scan, that at 15 cm H₂O of PEEP the intratidal collapse and decollapse is likely avoided in the majority of patients with ARDS (11). The PEEP values used in the present study were similar, although slightly lower, to the PEEP used in the study of Amato and colleagues (16) with the lung protective strategy (i.e., 16.2 cm H₂O). We then thought that, in our study conditions, the selected PEEP could be adequate to likely prevent the end-expiratory collapse (or, at least most of it) and provide a satisfactory oxygenation.

Baseline Period

Our goal was to select a ventilatory setting satisfying the current recommendations, which are based on the "gentle treatment" of the lung (13) and on prevention of intratidal collapse and decollapse. As previously discussed, we applied a PEEP that was thought "adequate" to guarantee the respiratory cycle in a relatively open lung, and we limited the plateau pressure to avoid excessive alveolar stretching. The resulting VT (7.3 ± 1.2 ml/kg) and Pa_CO2 (52.7 ± 19.4 mm Hg) were similar to those reported in several current trials testing low VT in ARDS (14, 15), whereas the PEEP we used was higher, and the overall mode of our baseline ventilation was very similar to that used in the "lung protective strategy" of the study of Amato and colleagues (16).

Sigh Period, the Recruitment

Introducing three sighs per minute at 45 cm H₂O plateau pressure led to a further lung recruitment, indicating that some pulmonary units had opening pressure greater than 35 cm H₂O. We can neither define from our data the pressure required for full lung opening nor can we define the spectrum of opening pressures existing in the ARDS lung. It is possible, however, that the nature of atelectasis and its opening pressure are somewhat related, although only inferences may be made with the available data. We believe that two kinds of atelectasis possibly coexist in ARDS lung: the compression atelectasis and the reabsorption atelectasis. The compression atelectasis develops immediately, as shown by CT scan, because of the increased lung weight, and its anatomic basis is likely the collapse of the small airways at end-expiration (11). As some gas is left behind the collapsed airways, the transmural pressure required for opening is relatively low (12 to 20 cm H₂O) (28), i.e., "loose" atelectasis.

The reabsorption atelectasis develops with time, when the regional gas uptake exceeds the delivery, depending on the FI_O2, the regional A/, and likely the end-expiratory volume of the pulmonary units. As no gas is left in the pulmonary units, the transmural pressure required for opening may be as high as 30 to 35 cm H₂O (28), which may correspond to a plateau pressure as great as 70 cm H₂O, i.e., "sticky" atelectasis.

From the observed data, several interpretations may be advanced. It is possible that at end-expiration, with the PEEP levels we used, we still had small airway collapse. In this case, sigh would lead to two effects: (1) recruitment of some "sticky" atelectasis; (2) greater inflation, with stabilization of the pulmonary units previously collapsing and decollapsing, and consequent improvement of their A/ ratios. Alternatively, it is possible that at 14 cm H₂O of PEEP, the small airway collapse was prevented, as suggested by the PEEP trial. In this case the effect of sigh was just to recruit part of the "sticky" reabsorption atelectasis. In both cases it is clear that 35 cm H₂O of plateau pressure was not sufficient to fully recruit the lung.

The issue of full opening has been mainly emphasized by Bohn and Lachmann (29) who suggested a recruitment maneuver with pressures largely exceeding 35 cm H₂O plateau for short period of time to resolve the "sticky" atelectasis problem. Lower pressures were thought to be enough thereafter to maintain the lung open if adequate PEEP was provided.

No Sigh Period, the Derecruitment

Unfortunately, interrupting the sigh and resuming the recommended baseline ventilation led to a progressive derecruitment. We have then to explain why part of the pulmonary units that remained open at end-expiration at 14 cm H₂O of PEEP during the sigh period did collapse at the same PEEP when the sigh was abolished. Moreover, we have to explain why the patients who most recruited and improved Pa_O2, shunt, and Pa_CO2 during the sigh period were not the same patients who most derecruited and deteriorated gas-exchange after sigh withdrawal. These findings suggest, in fact, that the possible mechanisms underlying recruitment and derecruitment are not the same.

The derecruitment may occur through two basic mechanisms, one implying mechanical forces and one implying gas reabsorption. In our conditions, it is possible that abolishing the sigh, the unstable pulmonary units with more severe surfactant alterations (maybe the ones opened by sighs) undergo collapse. However, if recruitment and derecruitment would be only the function of mechanical forces, we would expect that the patients who recruited more, were the same as those who derecruited more. This was not the case in our study.

According to our data, one possible reason for derecruitment was the formation of reabsorption atelectasis caused by low A/ ratios. In fact, we found that derecruitment was greater with lower A/ and greater Pa_CO2. In these patients with severe ARDS, high FI_O2 used and low tidal volume may favor reabsorption atelectasis, which may be prevented by cyclic high tidal volumes (sigh) (30, 31).

A similar relationship was found by Bendixen and colleagues (17) during general anesthesia and more recently, formation of reabsorption atelectasis was noted by CT scan after recruitment maneuvers in normal anesthetized subjects (32).

In conclusion, although we lack a direct evidence, it is possible that recruitment is mainly a function of the inspiratory plateau pressure, and the derecruitment is mainly a function of A/Q_eff ratio and lack of adequate PEEP levels.

ARDS from Pulmonary and Extrapulmonary Origin

Recruitment appears to be different in ARDSp and ARDSexp (20). In fact, in ARDSp the amount of atelectasis is scarce, as the predominant damage is the consolidation of alveolar units. For a given applied plateau pressure during the sigh (for example, 45 cm H₂O), the "average" change of transpulmonary pressure, i.e., the difference between the plateau pressure and PEEP minus the difference between esophageal pressure at end-inspiration and end-expiration, is relatively high (24.0 ± 5.7 cm H₂O) because of the normal Est,w (4.0 ± 1.2 cm H₂O · L¹). Thus, ARDSp is characterized by low potential for recruitment, which is fully exploited by the high transpulmonary pressure (33). On the contrary, in ARDSexp the amount of atelectasis is much greater. For a given applied plateau pressure during the sigh (for example, 45 cm H₂O), the change in transpulmonary pressure is rather low (16.8 ± 4.6 cm H₂O) because of the high Est,w (10.7 ± 2.0 cm H₂O · L¹). Thus, ARDSexp is characterized by a high potential for recruitment, which, to be exploited, needs applied plateau pressures higher then in ARDSp. We have also to consider that our estimate of transpulmonary pressure does not take into account the possible regional variation, i.e., higher in nondependent and lower in the dependent lung regions.

Indeed, it is possible that in our study we did not fully exploit the potential for recruitment in ARDSexp since we applied, in both groups, the same plateau pressure to the respiratory system for recruitment. More generally, the ARDS origin, pulmonary or extrapulmonary, and Est,w should be taken into account when recruitment maneuver and sigh are set.

Clinical Consequences

This study has two limitations, which prevent its immediate translation in clinical practice: (1) the small number of patients; (2) the short observation period. We believe that the first point is of minor importance. Our purpose, in fact, was to investigate physiologic mechanisms and not to study outcomes. What has been clearly shown in this study is that using the recommended guidelines to ventilate patients with ARDS (i.e., relatively high PEEP selected according to the currently accepted methods and limited VT) leads to a persistence of unresolved atelectasis before and after the sigh. The most relevant clinical question is whether the presence of unresolved atelectasis (i.e., keeping the lung not fully open), is detrimental to the lung over a long-term period. We are not aware of any study that has specifically addressed this issue. However, there is some general consensus that the presence of atelectasis may increase the likelihood of infections and induce lung damage. It is believed that at the interface between open and atelectatic lung damaging shear forces are generated, with consequent lung damage, although this concept has more theoretical (34) than experimental evidence.

Indeed, if one believes that unresolved atelectasis is dangerous, the lung should be opened and kept open. Although opening may be accomplished in several ways, including the sigh, it is important to emphasize, as shown in this study, that the usual recommended ventilation may not be adequate to maintain the lung open.

In this study, we did not test the different possible systems to maintain the recruitment. It is possible that after the sigh-induced recruitment, a PEEP increase would be effective to maintain open the recruited lung regions. The cost, however, would be an increase in mean airway pressure, and, if the plateau pressure is maintained equal to or below 35 cm H₂O, a tidal volume reduction and further increase of Pa_CO2, as suggested by the findings of our PEEP trial (see Table 2). It is also possible that if tidal volume is excessively reduced, reabsorption atelectasis may develop with time, despite higher PEEP. Another possibility would be to maintain recruitment with ventilatory techniques designed to improve alveolar ventilation, such as high frequency oscillation, which has been shown to be effective in maintaining recruitment in animal models (35). It is possible that high frequency oscillation can maintain recruitment for hours not because of good overall A/Q_eff ratios but just because it keeps the end-expiratory pressures at very high levels. Finally, a continuous sigh at pressure, volume, and rate to be defined, may be another alternative to maintain recruitment and prevent reabsorption atelectasis. However, it remains to be determined which would be the most convenient approach, as their relative iatrogeneity has not been established yet.