INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The lower inflection point (LIP) of the elastic pressure-volume (Pel/V) curve of the respiratory system recorded from the elastic equilibrium volume has long been used as orientation for setting the optimal positive end-expiratory pressure (PEEP) in patients with acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) (1). LIP was thought to indicate the opening pressure of previously collapsed or compressed alveoli (4). Recent studies, however, show alveolar recruitment to be a complex phenomenon that cannot be fully apprehended by LIP alone (5). Furthermore, the shape of the Pel/V curve and the value of LIP may vary according to the end-expiratory equilibrium volume from which the curve is traced (6). Chest wall properties may also affect LIP (9, 10).

In addition, because LIP is measured on the inflation limb of the Pel/V curve after a prolonged expiration, it may detect a need for recruiting alveoli only in this particular situation. Alveolar derecruitment (V_DER) induced by decreasing PEEP levels may be of greater clinical relevance. Recording the Pel/V curve and measuring V_DER at each PEEP decrement may yield the range of alveolar closing pressures. Should alveolar closure occur within a narrow pressure range, analogous to a threshold, then decreasing the PEEP level stepwise would not result in substantial derecruitment until the decrement is spanning the threshold, which would produce massive derecruitment. Setting the PEEP level above the threshold would keep the lung open. This was our working hypothesis. Our alternative hypothesis was that alveolar closure would occur over a broad pressure range, so that derecruitment would increase gradually over many PEEP decrements, and no optimal PEEP level would emerge.

We used a modified automated method for fast bedside recording and analysis of multiple Pel/V curves and V_DER at decremental PEEP levels. To study the relationship between LIP and the pressure range over which V_DER occurred, we compared V_DER to the information derived from the LIP on Pel/V curves recorded from zero end-expiratory pressure (ZEEP), and we looked for correlations linking V_DER to oxygenation parameters and to compliance.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

The study was done at Henri Mondor hospital in Créteil, France. Our institution's ethics committee approved the protocol. Informed consent was obtained from the patients' next of kin. During the study period, patients who fulfilled criteria for ALI and who required mechanical ventilation with a fraction of inspired oxygen (FI_O2) equal to or greater than 0.5 for longer than 12 h were potentially eligible. Criteria for ALI were those defined by the American-European International Consensus Conference on ARDS (11). Exclusion criteria were chronic obstructive pulmonary disease, presence of a leaking chest tube, contraindication to sedation, supposed chest wall impairment, or clinical or hemodynamic instability. Supposed chest wall impairments were thoracic deformities, known or patent restrictive disease, recent thoracic or abdominal surgery, acute abdominal syndrome, and multiple trauma. Sixteen patients met our inclusion and exclusion criteria. Among them, 14 met criteria for ARDS (Pa_O2/FI_O2 < 200 mm Hg). The lung injury score (LIS) was computed as described by Murray and coworkers (12). Patient characteristics are reported in Table 1.

All patients received mechanical ventilation in volume-controlled mode (ServoVentilator 900C; Siemens-Elema, Lund, Sweden). Tidal volume was 6 to 8 ml/kg, respiratory rate was 18/min, inspiratory-to-total time ratio was 0.33, and postinspiratory pause-to-total time ratio was 0.05. The baseline PEEP level was 15 cm H₂O in 10 patients and 20 cm H₂O in 8 patients. This level was chosen in each individual patient as the highest level not associated with hemodynamic instability (defined as an arterial systolic blood pressure drop of 20 mm Hg or more) or an excessively high plateau pressure (40 cm H₂O or more). Patients were supine. Sedation and neuromuscular blockade were achieved by continuous infusion. Endotracheal suctioning was performed 5 min before data collection and was not repeated during data collection unless needed. Absence of leaks in the circuit was checked during a 12-s postinspiratory pause. In five patients, esophageal pressure was measured using a balloon catheter (Marquat; Boissy Saint Léger, France) connected to the built-in ventilator pressure transducer via the auxiliary equipment port. The balloon was filled with 1 ml of air and positioned in the lower third of the esophagus. An occlusion test was performed to assess the validity of esophageal pressure measurement (13). All patients had an arterial line for blood gas sampling and blood pressure monitoring. Blood gases were measured using an ABL 520 analyzer (Radiometer, Copenhagen, Denmark).

Recording of Multiple Elastic Pel/V Curves

Multiple elastic Pel/V curves were recorded and analyzed using a modification of a method described in detail previously and in the online data supplement (2, 6, 14). The ventilator and carbon dioxide (CO₂) analyzer (ServoVentilator 900C, CO₂ Analyzer 930; Siemens-Elema) were connected to a personal computer. The ventilator system transducers produced signals representing pressure in the expiratory line, ventilator flow rate, and CO₂ at airway opening. These signals were filtered to avoid aliasing and were converted from analog to digital at 50 Hz. The flow signal was calibrated under BTPS conditions with a 1-L syringe. Pressure was calibrated using a water manometer and CO₂ using a gas mixture with a known composition.

Whereas the original method was designed for acquiring one Pel/V curve per recording session, the modification allows fully automated acquisition of multiple Pel/V curves at decreasing PEEP levels during a single continuous recording session lasting no more than 5 min. Each curve was obtained after the first expiration at a new, lower, PEEP level. Between curve recordings, patients were ventilated at the baseline PEEP level for 10 breaths, which was the highest level tolerated (Figure 1). Four or five Pel/V curves were thus obtained, starting from PEEP 15 cm H₂O or 20 cm H₂O and ending at ZEEP.

View larger version (20K): [in this window] [in a new window]   Figure 1.   Original recordings of pressure (P) and flow () over tim e showing the method used to record multiple inspiratory elastic pressure-volume curves at various PEEP levels. In this example, the baseline PEEP value was 15 cm H2O. Each low constant flow insufflation for Pel/V curve recording was preceded by a prolonged expiration. Apart from the prolonged expirations, the flow and pressure patterns were similar at all PEEP levels. PEEP levels were separated by 10 breaths to allow restoration of ventilation history. Integration of the inspiratory and expiratory flow measured just before and during the prolonged expiration preceding the low-flow insufflation allowed computation of EELV between the baseline PEEP (PEEP 15 cm H2O in this example) and each lower PEEP level.

Analysis of Pel/V Curves

Volume changes were calculated as the integral of the flow signal. To determine whether the recording procedure disturbed CO₂ exchange, CO₂ elimination was calculated as the integral of flow multiplied by the CO₂ concentration.

The analysis method is detailed in the online data supplement. Each Pel/V curve was analyzed using a sigmoid model (6, 14) in which each curve has three segments separated by the LIP and the upper inflection point (UIP). The segment before LIP and the segment after UIP are curvilinear and have low compliance values. The steeper segment between the LIP and the UIP has higher compliance (C_LIN) and is considered linear. LIP and the UIP are defined as the points at which the statistical analysis indicates that the Pel/V curve begins to deviate from a straight line. Accordingly, LIP corresponds to the point at which the second derivative of the equation used for Pel/V curve mathematical fitting reaches its maximum value. Similarly, UIP is at the minimum value of the second derivative of the equation.

Chest wall Pel/V curves were obtained in five patients by plotting lung volume against esophageal pressure. The above-described procedure was then used to analyze chest wall Pel/V curves.

Measurement of End-expiratory Lung Volume

In each patient, a previously described method (6), further explained in Figure 1, was used to align all Pel/V curves on a common volume axis. The method assumes that lung volume returned to a similar level during the 10 breaths following each Pel/V curve recording. Hence, end-expiratory lung volume (EELV) at baseline PEEP before each prolonged expiration was taken as the reference volume and used for subsequent Pel/V curve alignment. The change in EELV induced by each PEEP decrease was measured by integrating the difference between inspiratory flow and expiratory flow measured just before and during the prolonged expiration preceding the low-flow insufflation, respectively (Figure 1). Measuring the change in EELV induced by each PEEP decrease allowed alignment of each Pel/V curve on the first curve recorded at baseline PEEP (15 or 20 cm H₂O). Thus, all the Pel/V curves constituted a family of Pel/V curves aligned on a common volume axis. This volume, reached during a 6-s expiration at ZEEP, was taken as zero for the family of Pel/V curves (Figure 2).

View larger version (14K): [in this window] [in a new window]   Figure 2.   After plotting the Pel/V curves for the various PEEP levels on the same graph, the volume derecruited from a given PEEP to ZEEP was measured. Alveolar derecruitment (VDER) (gray double arrow) is the volume lost between the Pel/V curve from PEEP and the Pel/V curve from ZEEP, at an elastic pressure (Pel) of 20 cm H2O. LIP, lower inflection point on the Pel/V curve from ZEEP; EELVZEEP, end-expiratory lung volume at ZEEP.

To verify the assumption that lung volume returned to a similar level during the 10 breaths following each Pel/V curve recording, the change in EELV (EELV) between the baseline (highest) PEEP and ZEEP was measured again, in each patient, after 15 min of ventilation at baseline PEEP. Fifteen minutes after the multiple Pel/V curve recording, EELV was on average 69 ± 63 ml larger than EELV measured at the end of the multiple Pel/V curve recording (1059 versus 990 ml). This volume difference was 6% of the EELV measured 15 min after multiple Pel/V curve recording, suggesting that EELV was perhaps not fully restored during multiple Pel/V curve recording. Because this might lead to slight underestimation of derecruitment induced by the lowest PEEP steps, in each patient we corrected the last two EELV measurements of the sequence for the measured volume difference in that patient (data not shown). Given that any underestimation would be greatest for the last EELV measurement, we arbitrarily decided to correct the last EELV by two-thirds and the next to last EELV by one-third of the total volume difference.

Measurement of Intrinsic PEEP and Alveolar Derecruitment

Intrinsic PEEP (PEEP_i) was measured in all patients during a 2-s end-expiratory pause after a prolonged expiration to ZEEP.

Volume derecruitment (V_DER) caused by a PEEP decrement was defined for each PEEP level as the volume difference between the Pel/V curve recorded at that PEEP level and the Pel/V curve recorded at ZEEP. V_DER was calculated at a Pel of 20 cm H₂O (Figure 2). The difference in V_DER between two PEEP levels was V_DER.

Protocol

The patients were ventilated at the highest PEEP level used during the study (20 or 15 cm H₂O), and multiple Pel/V curves were recorded at decreasing PEEP levels (15, 10, and 5 cm H₂O and ZEEP) as described above. Arterial blood gases were measured 20 min after the recordings, at the highest PEEP. Then, PEEP was decreased stepwise to 15, 10, and 5 cm H₂O, and arterial blood gases were measured after 20 min of ventilation at each of these PEEP levels. To avoid unnecessary hypoxemia, blood gases were not measured at ZEEP. Oxygen saturation was monitored during the study by pulse oximetry (HP Model 56S; Hewlett Packard, Germany).

Statistics

Results are reported as means ± SD. To evaluate within-patient differences between alveolar derecruitment and oxygenation parameters at the various PEEP levels studied, one-way analysis of variance for repeated measurements and Fisher's protected least significance test were used. Pel/V curve C_LIN values were compared using the Wilcoxon test for paired samples. Regression analysis (Spearman rho) was used when required. p values smaller than 0.05 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Feasibility of the Technique

The time needed to record a complete sequence of 4 to 5 Pel/V curves ranged from 4.2 to 5 min. No complications occurred during recordings. Mean CO₂ elimination rate was 0.97 ± 0.08 of the value immediately before the recording (p > 0.05), suggesting that the recording procedure induced no disturbances.

Alveolar Derecruitment and LIP

In all 16 patients, derecruitment occurred at each PEEP decrement, as shown in Table 2. As illustrated in Figure 3, V_DER was almost evenly distributed across the PEEP decrements, suggesting that the alveolar closing pressures were spread over a wide range. In the seven patients who were ventilated with 20 cm H₂O PEEP, V_DER values between PEEP 20 and 15 cm H₂O (p < 0.05) and between PEEP 10 and 5 cm H₂O (p < 0.05) were significantly larger than the one between PEEP 5 cm H₂O and ZEEP (Figure 3). All patients showed substantial derecruitment even with the first PEEP reduction. On average, V_DER induced by the first PEEP decrement (from 20 to 15 cm H₂O in seven patients and from 15 to 10 cm H₂O in nine patients) was 34 ± 11% of the derecruitment between the individual highest PEEP and ZEEP. No correlation was found between maximal V_DER and P_LIP.

View larger version (10K): [in this window] [in a new window]   Figure 3.   Mean values of the differences in alveolar derecruitment (VDER) among the PEEP levels studied in the overall population (n = 16) (bold line with black squares), in the subset ventilated with PEEP 20 cm H2O (n = 7) (dotted line with open circles), and in the subset ventilated with PEEP 15 cm H2O (n = 9) (dotted line with open triangles). The only significant differences were between PEEP 5-0 and both PEEP 20-15 and 10-5 in the subset ventilated with PEEP 20 cm H2O. *p < 0.05, compared with PEEP 20-15; p < 0.05, compared with PEEP 10- 5; NS, not significant.

Oxygenation

Reducing PEEP levels for about 20 min decreased Pa_O2/FI_O2 (228 ± 81 mm Hg at PEEP 20 cm H₂O; 172 ± 81 mm Hg at PEEP 15 cm H₂O; 126 ± 66 mm Hg at PEEP 10 cm H₂O; and 90 ± 48 mm Hg at PEEP 5 cm H₂O, p < 0.01). The drop in Pa_O2/FI_O2 associated with reducing PEEP from 15 to 5 cm H₂O was significantly correlated with V_DER over the same pressure interval: Pa_O2/FI_{O2 (15-5)} = 28.9 + 0.49 · V_{DER (15-5)} (Pa_O2/FI_{O2 (15-5)} in mm Hg and V_{DER (15-5)} in ml; rho = 0.6, p = 0.02) (Figure 4). The interval between 15 and 5 cm H₂O was chosen for this calculation because data were available in all subjects. No correlation was found between Pa_O2/FI_{O2 (15-5)} and P_LIP.

View larger version (8K): [in this window] [in a new window]   Figure 4.   Correlation between changes in alveolar derecruitment (VDER(15-5)) and in the PaO2/FIO2 ratio (PaO2/FIO2(15-5)) associated with reducing PEEP from 15 to 5 cm H2O.

Linear Compliance

C_LIN increased significantly and gradually as PEEP decreased (Table 3). A significant correlation was found between C_LIN at ZEEP (C_LINZEEP) and V_DER at PEEP 15 cm H₂O (V_DER15): V_DER15 = 130.1 + 3.7 · C_LINZEEP (V_DER15 in ml and C_LINZEEP in ml/cm H₂O; rho = 0.9, p = 0.001) (Figure 5). C_LINZEEP was also significantly correlated with the Pa_O2/FI_O2 ratio measured at PEEP 15 cm H₂O (Pa_O2/FI_O2(15)):Pa_O2/FI_O2(15) = 94.5 + 1.66 · C_LINZEEP; rho = 0.6, p = 0.03). Both correlations suggested that C_LINZEEP reflected the potential for alveolar recruitment.

View larger version (10K): [in this window] [in a new window]   Figure 5.   Correlation between linear compliance of the Pel/V curve above the LIP at ZEEP (CLINZEEP) and alveolar derecruitment at PEEP 15 cm H2O (VDER15).

Chest Wall and Lung Pel/V Curves

Among the five patients whose chest wall Pel/V curves were recorded, three (2, 11 and 16) had a pulmonary and two (1 and 15) an extrapulmonary cause of ALI. Chest wall had little or no influence on the shape of the Pel/V curve of the respiratory system. Only in one patient (16) did the chest wall Pel/V curve exhibit an LIP, at a pressure of 1.3 cm H₂O. A UIP was found in patients 1 (2.4 cm H₂O) and 16 (1.7 cm H₂O). Mean chest wall C_LIN was 244 ml/cm H₂O (range 114-463 ml/cm H₂O), and mean lung C_LIN was 43 ml/cm H₂O (range 1466 ml/cm H₂O).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main results of this study can be summarized as follows: (1) alveolar derecruitment and closure were spread over a wide pressure range; (2) in patients with ALI, even a small PEEP decrease from a level as high as 15 or 20 cm H₂O induced derecruitment; (3) oxygenation was closely correlated with recruitment maintained by PEEP; and (4) higher compliance above LIP on the curve traced from ZEEP was associated with a greater potential for alveolar recruitment.

Technical Issues and Limitations

The computer-controlled procedure used to record multiple Pel/V curves at decremental PEEP levels is a further development of a recently introduced technology (2, 6, 14). That this procedure is feasible and safe was demonstrated by our study. Whereas the original technique acquires a single Pel/V curve per recording session, our modification provides a fully automated sequence of multiple Pel/V curves during a single, continuous, 4- to 5-min recording session. CO₂ elimination was not disturbed during recording, and no other adverse effects were detected. Bias-free analysis and alignment of Pel/V curves on end-expiratory volume of the respiratory system at ZEEP allowed us to estimate derecruitment at decremental PEEP levels. PEEP levels were not studied in random order; rather, we worked from the highest PEEP level to ZEEP. In an effort to standardize lung volume, patients were ventilated with 10 normal breaths at the highest PEEP level before each Pel/V curve recording (Figure 1). However, we found that EELV between the baseline (highest) PEEP level and ZEEP after 15 min of ventilation was 69 ± 63 ml larger on average than EELV measured at the end of the multiple Pel/V curve recording session (1059 versus 990 ml). This slight volume difference indicated that EELV at baseline PEEP was not completely restored after the 10-breath period. Consequently, in each patient, we corrected the last two EELV measurements of the sequence for the measured volume difference in that patient. This did not affect the main results of the study, particularly our finding that substantial derecruitment occurred with the first PEEP decrement.

We used a 6-s expiration before the insufflation needed for Pel/V curve recording because zero flow is usually achieved after this duration. However, if the volume exhaled during this 6-s expiration was large, zero flow was not consistently achieved, being only closely approached in some cases. This may have influenced LIP determination because of possible interference with the PEEP_i on the Pel/V curve. PEEP_i has been reported to produce a fallacious LIP (15). In our study, however, all LIP values were well above PEEP_i values, as shown in Table 2, suggesting that PEEP_i did not interfere with LIP determination.

Effect of PEEP Decrease on Lung Volume and Significance of V_DER

V_DER is the lung volume lost during a single deep expiration to ZEEP and not regained when the airway pressure reaches 20 cm H₂O during a subsequent slow inflation. Several mechanisms could explain the shift in the Pel/V curves. In ARDS, most of the alveolar derecruitment induced by PEEP removal may occur abruptly because of the short time constant of the respiratory system, whereas inflation of collapsed lungs is a slow and uneven process (16). Successive reinflation of collapsed lung units has been demonstrated in recent studies (5, 17, 18). Moreover, a recent study showed a good correlation between alveolar recruitment evaluated by computed tomography and by Pel/V curves, supporting the possibility that derecruitment/recruitment may be the main cause of the shift in the Pel/V curves (19).

Identification of Alveolar Closing Pressures

Our working hypothesis was that our method of alveolar derecruitment assessment might identify a threshold for alveolar closure, thus helping to determine the PEEP level required to keep the lungs open. We reasoned that, for instance, if alveolar closure started at 15 cm H₂O, V_DER values at 20 cm H₂O and 15 cm H₂O would be similar. This was not the case. Although we decreased PEEP levels from a high baseline value (20 or 15 cm H₂O), we were unable to identify an "optimal" PEEP level because alveolar closure occurred gradually over a broad range of pressures in our patients with ALI. Heterogeneity in alveolar properties and particularly a high pleural pressure gradient caused by the increased lung density in ALI may explain this finding. The range of closing pressures may be at least as wide as the range of opening pressures (5, 20).

Meaning of LIP

The LIP correlated neither with maximum V_DER nor with the change in Pa_O2/FI_O2 at the various PEEP levels. Major derecruitment occurred even at the first PEEP decrement, and further derecruitment was noted at all subsequent decrements, independently from the LIP value. Thus, both derecruitment and recruitment occurred gradually, rather than abruptly at a given threshold. LIP seems to have no relation with PEEP-induced recruitment (6). In patients with no identifiable LIP on Pel/V curves, Vieira and coworkers (21) found that recruitment was minimal with PEEP, whereas a risk of overdistension was present even with relatively low PEEP levels. Our findings also suggest that LIP is only a qualitative marker for a recruitable lung. The LIP reflects recruitment after a prolonged expiration, which is probably different from recruitment during tidal ventilation. Matamis and coworkers (1) and, more recently, Amato and coworkers (3) have suggested that PEEP should be set at a level above the LIP. However, a number of factors influence LIP, including chest wall elastic properties (9, 10, 22). The presence of an LIP seems to indicate a need for recruiting alveoli but may be of little help in determining the optimal end-expiratory pressure. Measurement of recruitment or derecruitment may be more relevant for determining the optimal PEEP in ALI/ARDS patients.

Effect of PEEP on Linear Compliance

In all our patients, decreasing PEEP induced an increase in C_LIN of the Pel/V curves recorded during subsequent insufflation. The gradual decrease in C_LIN with increasing PEEP could, in theory, reflect recruitment of some parts of the lung and distension or overdistension of other parts. However, our results are in accordance with those of Hickling (7) suggesting that a relatively low value of C_LIN, as observed at high PEEP, may indicate good lung recruitment. The observations by Jonson and coworkers support this suggestion (6). In line with previous reports (16), our data suggest that a high C_LIN at ZEEP (measured above the LIP) may indicate a highly recruitable lung.

Alveolar Recruitment/Derecruitment and Gas Exchange

ARDS is characterized by severe hypoxemia and intrapulmonary shunting related to the presence of perfused nonventilated alveoli. Recruiting these alveoli should, in theory, reduce intrapulmonary shunting, thereby improving oxygenation. Falke and coworkers were among the first investigators to show that the PEEP-induced increase in functional residual capacity was key to the improvement in oxygenation (23). When methods for measuring recruitment became available, Ranieri and coworkers were able to show a significant correlation between alveolar recruitment and oxygenation in patients with ARDS (24). Our study further describes the correlation between individual changes in alveolar derecruitment and in oxygenation parameters.

Influence of Chest Wall Mechanics

To minimize a possible influence of chest wall mechanical disturbances on our findings, we did not include patients with clinical conditions considered to be often associated with major chest wall abnormalities. In the five patients whose chest wall Pel/V curves were recorded, chest wall had little or no influence on the shape of the Pel/V curve of the respiratory system. A single patient, with ARDS related to a pulmonary cause, had an LIP on the chest wall Pel/V curve, at a pressure of 1.3 cm H₂O, whereas P_LIP on the Pel/V curve of the respiratory system was 17.1 cm H₂O.

Conclusion

Our modified technique for fast recording of multiple Pel/V curves at various PEEP levels proved feasible and safe. Lung derecruitment, evaluated based on multiple Pel/V curves, was well correlated with oxygenation in individual patients. Derecruitment was found at each step of PEEP reduction, suggesting that alveolar closure occurred over a wide range of pressures. Substantial derecruitment occurred even at the first PEEP decrement, although the starting level was high, that is, 15 or 20 cm H₂O. LIP did not indicate the pressure at which maximal alveolar derecruitment occurred. Therefore, although the presence of an LIP on the respiratory system Pel/V curve indicates a need for lung recruitment, it may be of limited value for determining the PEEP level required to prevent alveolar collapse. Finally, greater linear compliance above LIP on the Pel/V curve recorded from ZEEP predicts greater improvements in alveolar recruitment and oxygenation with increasing PEEP.