INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experimental studies have shown that mechanical ventilation at high lung volume causes permeability pulmonary edema and diffuse alveolar damage (1). Previous injury sensitizes the lungs to this ventilator-induced lung injury (VILI) (2), perhaps because mechanical nonuniformity makes them more vulnerable to local overdistension. It was initially thought that lung involvement was diffuse and homogeneous during acute respiratory distress syndrome (ARDS). Anatomical studies with CT scans revealed that the distribution of densities is in fact markedly inhomogeneous (5, 6). Comparison of lung imaging and pulmonary mechanics during ARDS led to the concept of "baby lung" (6). These lungs are characterized by the coexistence of diseased nonrecruitable zones, zones that may be recruited by mechanical ventilation, and relatively normal zones that receive the bulk of ventilation. These normal zones may become overinflated and hence injured if their tidal volume is too large. This concept has recently been supported by the demonstration of improved survival during ARDS by simply reducing tidal volume (7). In this study the same reduction of tidal volume was applied in all patients. However, it has repeatedly been shown that the pressure and the volume that are considered safe for some ARDS patients may cause lung overdistension in others (6, 8). Conversely, arbitrary settings may result in an unnecessary reduction in tidal volume. It has been suggested that information from the inspiratory pressure-volume (PV) curve of the respiratory system could be used to tailor ventilator settings. For instance, the presence of an opening pressure (lower inflection point, LIP) could be used to adjust the positive end-expiratory pressure (PEEP) (11- 13). In addition to improving oxygenation, PEEP reduces the severity of VILI (14) and may lessen the damage produced by the repeated opening and closing of lung units in surfactant-depleted lungs (15, 16). However, PEEP may favor overinflation if tidal volume (VT) is not reduced (1, 17). It has been proposed that the tidal volume be adjusted according to PV curve analysis by limiting end-inspiratory pressures to below the decrease in slope seen at high pressure/volume, called the upper inflection point (UIP) (9, 10, 18). The UIP often seen in patients with ARDS has been ascribed to overinflation (9, 10), or to the end of recruitment (19, 20) during lung expansion. However, whether or not ventilator settings that would result in pressure/volume excursions above the UIP are deleterious remains unsettled, and has never been assessed experimentally.

The impact of pulmonary edema and the resulting decrease in ventilatable lung volume on the inspiratory limb of the respiratory system PV curve have not yet been evaluated. A better understanding of its significance is required before the UIP can be used to set tidal volume in patients. This study was designed to examine several hypotheses. The first was that the reduction in ventilatable lung volume (the baby lung effect) not only decreases the compliance of the lung (6, 21) but also affects the position of the UIP. We tested this possibility by obstructing the distal airways of rats by instilling a viscous liquid. The second was that the development of edema alters the PV curve essentially because of distal airway obstruction. We therefore examined the changes in PV curves during high volume ventilation in normal lungs and lungs injured by -naphthylthiourea (ANTU). And the third was that individual characteristics of the PV curve reflect the susceptibility of the lungs to the deleterious effects of high volume ventilation.

Our observations suggest that the position of the UIP is a marker of ventilatable lung volume and is both influenced by and predictive of the development of edema during mechanical ventilation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiments were performed on male Wistar rats (Charles River, St-Aubin-lès-Elbeuf, France) weighing 280-320 g. Rats were anesthetized by intraperitoneal injection of 50 mg/kg thiopental (Sigma, Saint-Quentin Fallavier, France). Rodents metabolize thiopental slowly and remain deeply anesthetized for at least 4 h with this dose. They were tracheostomized and paralyzed with succinylcholine (Sigma) and ventilated with a Harvard volume respirator (Ealing, Courtaboeuf, France). All experiments were conducted according to regulations of the French Ministry of Agriculture.

Effect of Instilling a Viscous Liquid on the Shape of the PV Curve

Six rats were instilled at the carina level with successive 100-µl aliquots of a viscous solution (50% wt/wt Ficoll 70; Pharmacia, Uppsala, Sweden, in 0.1 M NaCl) to produce an increasing obstruction of small airways. At the beginning of the experiment and after each instillation, rats were ventilated for 5 min with 2.5 ml VT, PEEP 3 cm H₂O to allow the solution to move distally before quasistatic PV curves were registered as described (22).

Effect of High-volume Ventilation Lung Injury on the PV Curve

Rats (n = 12) were ventilated with 10 ml VT, 25 bpm, zero end-expiratory pressure (HV). Such a high tidal volume (i.e., approximately 33 ml/kg) resulted in end-tidal airway pressure higher than that of the UIP. This ventilation was continued for 30 to 120 min until the PV curve was obviously altered (i.e., the UIP was displaced to a lower lung volume). PV curves were registered at the beginning and at the end of the HV period and at different stages of edema development.

Effect of Lung Injury Prior to High-volume Ventilation on the Shape of the PV Curve

Lung permeability edema was produced with ANTU as previously described (2). Rats were then left to breath spontaneously for 2 h before being paralyzed and ventilated for 10 min with the same HV settings as above. PV curves were registered at the beginning and end of the HV period.

Pressure-Volume Curves

Lower (LIP) and upper (UIP) inflection points were determined by segmental linear regression analysis (Prism3; GraphPad Software, San Diego, CA) to fit the quasistatic PV curve (see Figure 1a). Junction points situate the LIP and the UIP(s). In ANTU-injected rats, the LIP and UIP were also determined by visual inspection by an independent observer (E.R.). Values for LIP and UIP were validated by comparison with static PV curves (see online data supplement).

View larger version (17K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Figure 1.   (a) Inflation PV curves during the successive instillation of 50- to 100-µl aliquots of a Ficoll solution into airways. Before instillation (on the left), the PV curve has a normal, classic, shape (shape a). The upper inflection point (plain arrow) is determined by fitting the curve with joined linear segments. The instillation of Ficoll progressively distorted the shape of the curve and the curve shifted to the right. PV shape a remains recognizable at low pressure, below that of a lower inflection point, easily identifiable in the lowest curve (broken arrow). A second upper inflection point is found above the lower inflection point, at high airway pressures (bold arrow). (b) Changes in the PV curve produced by ventilation with a large tidal volume to produce pulmonary edema. The normal (shape a) PV curve is progressively deformed to a curve with no recognizable UIP at low airway pressure, a well-defined opening pressure, and a new UIP at high airway pressure (shape b curve). Compare with (a).

Measurement of Extravascular Lung Water

Extravascular lung water was determined as previously described (23).

Statistical Analysis

All results are expressed as means ± SEM. Comparisons were made by Student's paired t test. Regressions were performed using the unweighed least-squares method. Statistical significance was accepted at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

PV Curve Analysis

There was a reasonable correlation between the positions of the LIP determined by eye and by segmental regression (r = 0.54, p < 0.001). The correlation was much better for the UIP (r = 0.85, p < 0.001). The values determined by segmental analysis correlated better with the edema parameters than did those obtained by visual inspection. Thus, they are the only values given. However, the same conclusions can be drawn from LIP or UIP obtained by visual inspection.

Effect of Instilling Viscous Liquid on the Shape of the PV Curve

The PV curve before instillation consisted in an almost linear increase in volume with pressure below the UIP. This UIP was seen at low (< 20 cm H₂O) airway pressure and at high lung volume (after approximately 7 ml). It will be called UIP1. No LIP was discernible (Figure 1a). Instilling viscous liquid led to a decrease in the slope of the segment below UIP1 (Crs1), a decrease in the volume of the UIP1 (Vuip1) with little change in the pressure at which it was found, an increase in Pei, and the appearance of a LIP at a pressure above that of the UIP1 (Figure 1a). The pressure at which UIP1 was found (about 15- 18 cm H₂O) did not vary appreciably with the amount of liquid instilled, but Vuip1 gradually decreased, the downward concavity below UIP1 flattened, and UIP1 became less and less discernible. The curve shape characterized by a discernible UIP1 at low airway pressure will be called shape a whereas that characterized by a linear segment below the LIP (i.e., no discernible UIP1) and an UIP (UIP2) at a pressure higher than that of the LIP will be called shape b. As the amount of instilled liquid increased, there was a progressive rightward shift of the curve on the pressure axis and an increase in Pei. The changes in Vuip1 and Crs1 were significantly correlated (r = 0.93, p < 0.001, Figure 2a). The slope of the regression of the percentage changes in Vuip1 compared with those in Crs1 was 1.05 ± 0.06 and the intercept was not different from zero (i.e., changes were proportional). No correlation was found between the slope of the linear segment between the LIP and UIP2 (Crs2) and the volume (Vuip2) at which UIP2 was observed.

View larger version (17K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   View larger version (18K): [in this window] [in a new window]   Figure 2.   (a) Relationship between respiratory system compliance (Crs) and the volume of the upper inflection point at low airway pressure (Vuip) during the instillation of Ficoll into the airways of eight rats. Changes in both are closely correlated (p < 0.001). The envelope of these data is visualized by dotted lines. (b) Changes in Crs and Vuip produced by ventilating normal rats with a high tidal volume. These are similar to those produced by Ficoll instillation in airways (the envelope of the data of (a) is shown as a dotted polygon). (c) Changes in Crs and Vuip produced by ventilating rats with ANTU-induced lung injury with a high tidal volume. These are superimposable to those observed in intact rats.

Effect of HV on the Shape of the PV Curve and its Relation with Qwl

A representative example of the changes produced by HV on the shape of the respiratory PV curve is shown in Figure 1b. They are similar to those seen with liquid instillation. The slope of the regression relating the percentage changes in Crs1 and Vuip1 was 1.12 ± 0.04, not different from that of the changes produced by liquid instillation, and the intercept was not different from 0 (p = 0.82).

The differences between the Crs1 (Figure 3a) and Vuip1 (Figure 3b) values for the PV curves registered before and after HV were significantly correlated with Qwl (both r = 0.95, p < 0.001). Pei also increased, albeit less significantly, with Qwl (r = 0.65, p < 0.05). Indeed, Pei was correlated with Crs1 (r = 0.78, p < 0.01) and Vuip1 (r = 0.76, p < 0.01).

View larger version (13K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   Figure 3.   Correlations between the amount of pulmonary edema (extravascular lung water: Qwl) and the percentage change in respiratory system compliance (a) or the volume of the upper inflection point at low airway pressure (b) produced by mechanical ventilation of normal rats with a large VT for various times.

Effect of HV on the Shape of the PV Curve in Injured Lungs

We have previously shown that ANTU has only a minimal effect on lung water by itself, but sensitizes the lung to HV- induced pulmonary edema (24). HV produced edema of variable severity in ANTU-injected rats. All PV curves had shape a before HV. HV caused varying decreases in both Crs1 and Vuip1, without any change in the shape of the PV curve in most (n = 45) rats. Curves of shape b were observed in nine rats after HV. The changes in PV curve shape were similar to those of rats instilled with the largest amounts of liquid or intact rats submitted to HV of longer duration (Figure 2c). Shape a curves with a UIP1 below the LIP were seen in rats with mild edema (Qwl = 4.78 ± 0.20 ml/kg bw), whereas shape b curves without discernible UIP1 were seen in rats with significantly more severe edema (Qwl = 7.25 ± 0.33 ml/kg bw, p < 0.001). Changes in Crs1 and Vuip1 were negatively correlated with the amount of edema in rats after HV (Figure 4a and 4b). No correlation was found between the changes in compliance (Crs1 before HV, Crs2 after HV) in the volume of the UIP (Vuip1 before HV, Vuip2 after HV) and Qwl in rats with shape b curves after HV (Figure 4a and 4b). Changes in Pei were positively correlated with Qwl in rats with both shape a (r = 0.73, p < 0.001) and b curves after HV (r = 0.95, p < 0.001) and when pooling all rats (r = 0.84, p < 0.001). Pei was highly correlated with Crs1 and Vuip1 only (all 0.90, p < 0.001).

View larger version (13K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   Figure 4.   Correlations between the amount of pulmonary edema (extravascular lung water: Qwl) and the percentage change in respiratory system compliance (a) or the volume of the upper inflection point at low airway pressure (b) produced by mechanically ventilating rats with ANTU-induced lung injury with a large VT for various times. Closed circles: rats with a shape a PV curve at low airway pressure (n = 45); open circles: rats with a shape b PV curve (n = 9). There was no correlation when all data were pooled. This was the consequence of the absence of a correlation between Crs and Qwl in rats with shape b PV curves, whereas both Crs and Vuip decreased with increasing Qwl when only shape a PV curves are taken into consideration.

The percentage change in Crs1 was linearly correlated with the percentage change in Vuip1 (r = 0.97, p < 0.001) when shape a was preserved after HV. The slope (1.12 ± 0.045) was identical to that found in intact rats submitted to HV and the intercept was not different from zero (p = 0.66). This indicates that shape a curves were similarly modified in ANTU-injected rats and in normal rats (see also Figure 2c).

Relationship Between the Initial PV Curve Parameters and HV-induced Edema

Because Crs1 and Vuip1 were altered proportionally with the severity of edema in initially intact lungs, we suspected that Crs1 and Vuip1 were also affected by the ANTU-induced edema before HV. We have shown that pulmonary edema and reduction of the amount of ventilatable lung increase susceptibility to HV injury (2, 24). The Crs1 and Vuip1 values before HV may thus reflect different degrees of ANTU-induced edema, different amounts of ventilatable lung at low airway pressure, and, thus, susceptibility to ventilator-induced lung injury. Both Crs1 and Vuip1 measured before HV correlated with the severity of edema produced by this ventilation (Figure 5a and 5b). Pei before HV was also correlated, although less significantly, with Qwl (r = 0.43, p < 0.001).

View larger version (14K): [in this window] [in a new window]   View larger version (12K): [in this window] [in a new window]   Figure 5.   (a) Correlation between the amount of edema (extravascular lung water, Qwl) produced by high volume ventilation and the respiratory system compliance (Crs) measured before high volume ventilation. (b) Correlation between Qwl and the volume of the upper inflection point on the PV curve (Vuip) measured before high volume ventilation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The shape of the respiratory system PV curve has acquired new significance because several studies suggested that it could be used to adjust ventilator settings for ARDS patients so as to minimize the risk of VILI. Several distinctive elements of the curve including the value of LIP pressure (11, 18), the respiratory system compliance (25), the end-inspiratory pressure (26) and the position of the UIP (9, 10) have been considered meaningful for patient care. However, the meaning of the PV curve shape is not completely understood.

This study explores the changes in the shape of the respiratory system PV curve produced by pulmonary edema, and particularly the significance of the position of the UIP. We examined the inflation limb of the respiratory system PV curve because it is the measurement performed at the patient bedside. We show that the changes in it depend on the amount of water that accumulates in the lungs. When the curve is not distorted by massive recruitment (shape a), the decrease in compliance (Crs1) is linearly correlated with the accumulation of water in the lungs, whatever their initial state, intact or injured with ANTU (Figures 3a and 4a). This decrease in compliance is correlated with changes in Pei, suggesting that it is the consequence of a decrease in the amount of aerated lung volume ("baby lung" effect), for example, by the filling of distal airways (27, 28). There may be a proportional decrease in compliance (Crs1) and in the volume of the UIP1 if lung mechanical properties at low airway pressures remain unchanged, whereas the volume of ventilatable lung decreases. The volume of the PV curve before HV of Figure 1b was multiplied by scaling factors less than unity to simulate a reduction in volume to illustrate this (Figure 6 shows this vertical compression on the volume axis). The curves obtained agreed nicely with those observed at low airway pressure during the development of edema. The changes in the PV curves during HV-induced pulmonary edema are similar to those observed when lung volume is decreased by airway obstruction (Figure 2). These observations support the contention that most of the changes in the PV curve produced by pulmonary edema are due to airway obstruction (28).

View larger version (24K): [in this window] [in a new window]   Figure 6.   Same data as in Figure 1b. The PV curve before overinflation-induced edema (typical shape a) was distorted by compression on the volume axis (dotted curves) multiplying volume at a given pressure by a factor of less than unity. Compression factor values are shown for each curve. Note that the compressed curve agrees well with the subsequent curves at low (below the upper inflection point) airway pressures.

PV curves may differ markedly depending on the amount of edema and site of airway obstruction. Lung conditions that may generate shape a and shape b curves are illustrated in Figure 7. Shape a curves may be produced by initial inflation of a relatively large amount of ventilatable lung (states 1 and 2 in Figure 7a) until these zones become overinflated (state 3). Recruitment of excluded areas at higher pressure produces an increase in the PV slope (state 4, see also Figure 1a). In contrast, when the amount of ventilatable lung at FRC is minimal, the initial part of the PV curve (states 1-3 of Figure 7b) has a very low slope. Massive recruitment occurs when the opening pressure is reached, resulting in an abrupt increase in the PV slope (state 4 of Figure 7b, see also Figure 1b). The pressure that leads to recruitment of these excluded areas is the same in lungs with shape a and b curves, about 25 cm H₂O (see also Figure 1a). It has been observed in isolated lungs (31) and suggested on theoretical grounds (19) that this recruitment may alter the shape of the inflation PV curve. Compliance above the LIP (Crs2) may become larger than the real one and another UIP (UIP2) may appear at a different lung volume. Our observations illustrate this possibility. They call attention to the fact that the significance of the UIP is equivocal in lungs that are inflated by recruitment above the opening pressure. High compliance value above the LIP (like the lowest curve of Figure 1b) may indicate recruitment. Recruitment has been reported in ARDS patients (20) whose lungs were inflated from ZEEP after a prolonged expiratory pause, but was not observed during tidal ventilation with ZEEP or PEEP (32). It may thus be important to lessen recruitment (perhaps with the use of PEEP) while registering PV curves and by standardization for volume history (33, 34).

View larger version (22K): [in this window] [in a new window]   Figure 7.   Diagram showing possible consecutive lung states during inflation depending on the amount of edema. Airway pressures are the same for each of the successive states (numbered 1-4) in a and b. For the sake of simplicity, territories excluded from ventilation are depicted by airway closure. This does not reflect the precise site or the mechanism of exclusion from ventilation (alveolar spaces filled with liquid, foam in airways, bronchial edema, etc.). (a) Shows how a limited volume of excluded (but potentially recruitable) lung results in shape a PV curve at low airway pressure. 1-2: Inflation of ventilatable areas; 3: the distensibility of these zones is decreased because of overinflation; 4: recruitment of excluded territories at airway pressures above the opening pressure. (b) Shows how a large volume of excluded but potentially recruitable lung results in shape b PV curve. 1-3: Progressive inflation of a small number of open units; 4: simultaneous opening of excluded volumes resulting in a steep increase in the PV curve slope when airway pressure exceeds the opening pressure (see text for details).

The initial stage of ARDS of extrapulmonary origin is a permeability type edema (35) that is probably better reproduced by ANTU than by other models (saline lavage, for example) that result in surfactant depletion and massive collapse, as in neonatal respiratory distress syndrome (RDS). The present study shows that the lower the volume of the UIP in shape a curves (UIP1), the greater the risk of HV-induced lung injury in these slightly edematous lungs (Figure 5), supporting the notion that UIP1 in shape a curves reflects an increase in lung tissue stress. The nonlinearity of lung elastic properties at high lung volume has been attributed to the amount and arrangement of elastin and collagen fibers (36) and to changes in surface tension that affect alveolar duct micromechanics (37). The increased stiffness at high lung volume is probably due to the increased number of units that reach maximum extension, determined by the length of the poorly distensible collagen fibers. This may explain the changes that occur in fibrosis or emphysema (36). By analogy, several authors consider the UIP to reflect tissue overstretching in patients with ARDS (9, 10), whereas Ranieri and coworkers (25) have shown that a faint UIP may be caused by altered parietal mechanics. This is probably not the case in rats because the chest wall does not contribute significantly to the respiratory system PV curve shape (38).

Overinflation may occur even with normal tidal volumes (1). There is presently no well-defined way of evaluating the risk of overinflation during ARDS, although the idea that the UIP is an important determinant of overstretching is worth consideration. The volume of the UIP is closely correlated with compliance and is influenced in the same way by pulmonary edema when the PV curve keeps shape a: The volume of the UIP loses its association with Qwl (Figures 3 and 4), when the PV curve shape changes to shape b. This supports the contention that the volume of the UIP is a marker of lung tissue stretch when the PV curve shape is not altered by recruitment during inflation (shape a curve). As already explained, Crs in shape a curves is more representative of the mechanical properties of the lung tissue than of the thoracoabdominal wall in rats. The influence of the parietal wall makes it difficult to obtain a predictive value of Crs in patients because of the effect of posture on rib cage and abdominal compliance (39). The knowledge of a patient's Crs value would not help to tailor the tidal volume. Thus, a relative criterion indicating the susceptibility of an individual to VILI is probably more useful than an absolute one. This observation may be clinically relevant considering the proposal of ventilating patients using the "open lung" approach (40) that consists in applying appropriate PEEP levels to recruit most of the ventilatable lung at end expiration (18, 41, 42). It is conceivable that under such conditions, the inflation PV curve would keep shape a and Vuip its meaning as a marker of overinflation (see, for example, Hickling [19], Figure 6, right panel ).

It is impossible to deduce directly from our study whether the PV curve may be used to tailor the PEEP and tidal volume (i.e., the level of overall distension) so as not to cause extra injury, because of the possible larger contribution of parietal mechanics in patients. Further studies are thus needed to obtain a better understanding of the clinical meaning of the UIP and its possible relationship with lung compliance. Nevertheless, the demonstration that the volume of the UIP is correlated with Crs in a group of animals of similar morphology and that its value greatly influences the severity of VILI may be an important indicator.