INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In 1958 Burton and Patel demonstrated that pulmonary vascular resistance (PVR) changed in response to lung inflation (1). Since then there have been numerous studies done to explain the causes of and the conditions under which PVR changes occur. Whittenberger and coworkers used an open-chest dog model to show a hysteresis in the PVR response when inflation was compared with deflation (2). When PVR was plotted against airway pressures in these dogs, it was found to be higher on deflation. Because lung volume varied widely at similar airway pressures in the presence of pressure-volume hysteresis, they concluded that lung volume was not the determinant of this hysteresis. Sun and colleagues found both total PVR and middle vessel resistance increased on deflation (3). They asserted that surface tension acted as a counterbalance to the distending force of alveolar pressure. Since surface tension was lower on deflation, the overall pressure applied to the alveolar capillaries was higher on deflation, and therefore the PVR was elevated. In contrast, the study by Thomas and coworkers concluded that PVR changes were volume-dependent but there was essentially no difference between inflation and deflation (4).

Recent work in the field of acute lung injury during mechanical ventilation has shifted the emphasis away from airway pressures and barotrauma to overdistension and volutrauma (5, 6). It is well established that high airway pressure can cause increases in lung fluid in animal models. However, when overdistension was avoided in rabbits and lambs, by binding the animal's chests, this fluid increase was either blunted or prevented completely (7, 8). Maintaining appropriate lung volume to avoid end-expiratory alveolar collapse and repeated reopening is also crucial to a ventilation strategy that protects the lung (9, 10). One approach is to keep lung volume greater than closing volume (CV) to achieve optimal compliance and diminish pressure-volume hysteresis. Closing volume is the volume below which small airways collapse and increasing atelectasis causes a decrement in compliance. Closing volume can be directly measured using tracer gas washout studies. Phase IV of these washout curves represents CV and is closely correlated with the subtle inflection point on the lower portion of the deflation limb of the pressure-volume compliance curve (11, 12).

Our interest stemmed from a desire to understand what changes in PVR would result from positive pressure ventilation at appropriate lung volumes. We looked specifically at inflation and deflation of the lung using fixed blood flow, physiologic inflation volumes, and simple positive pressure inflation. Given current emphasis on avoiding volutrauma, overdistension, and alveolar collapse, we plotted PVR changes against lung volume. In pilot studies we found that PVR was decreased on deflation and that PVR hysteresis paralleled pressure-volume hysteresis (13). Hysteresis is defined as the failure of one of two related phenomena to keep pace with the other. Aside from visually inspecting two related curves to see a hysteresis, or difference in the paths, one could statistically compare the individual points for a difference. We compared mean PAPs (the dependent variable) over a range of volumes (the independent variable) to show statistically significant hysteresis between entire inflations and deflations. We hypothesized that: (1) there was a PVR hysteresis during lung ventilation; (2) like pressure-volume hysteresis, it would be absent when lung volumes were kept above CV; (3) this hysteresis would return when inflation occurred after the lung was allowed to deflate below CV.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Isolated Lung Lobe Preparation

Mongrel dogs weighing 19.6 ± 0.45 kg were anesthetized with 30 mg/ kg of intravenous pentobarbital sodium (Abbott Laboratories, North Chicago, IL). They were then intubated and mechanically ventilated at a rate of 10 to 15 breaths/min and tidal volume 15 ml/kg with a Harvard ventilator (Model 607; Harvard Apparatus, Mills, MA). The rate and tidal volume were adjusted to achieve adequate chest rise and normal arterial blood gases. The left chest was opened through the fifth intercostal space; the upper and cardiac lobes were tied and removed. The dogs were then anticoagulated with 10,000 U of heparin sodium through a femoral venous line. After 10 min, they were exsanguinated from a femoral arterial cannula into a graduated cylinder that contained another 10,000 U of heparin sodium. After phlebotomy and before cardiac arrest, the left lower lobe (LLL) artery was tied and the LLL bronchus was clamped during full inflation. The lobes were then removed. The perfusion circuit was filled with 400 ml of the shed blood and diluted with 100 ml of normal saline to a hematocrit of 36.6 ± 1.9. The LLL bronchus, vein, and artery were then cannulated. Care was taken to avoid the introduction of air bubbles into the artery and deflation of the lobe. The LLL was placed in a nonrestrictive plastic bag, perforated to allow drainage of any condensation or fluid, and placed on a platform. The flexible airway and perfusion catheters were suspended and their positions fixed by clamping for the course of the experiment. The lobe was then perfused at constant flow, 600 ml/min, via the lobar artery using a roller head pump (Sarns 8000 roller pump; Sarns, 3M Health Care, Ann Arbor, MI). The perfusion circuit consisted of a Plexiglas venous reservoir in which the blood was continuously stirred and recirculated to the LLL after passage through a blood filter, a water-jacketed heat exchanger maintained at 38 ± 0.5° C, and a bubble trap. The blood was returned to the reservoir via a large-diameter venous catheter. A screw clamp was placed on the venous catheter to keep the lobar vein pressure constant at 5 cm H₂O. Pump blood flow was calibrated daily by timed collections from the pulmonary arterial catheter. During the experiment, pulmonary artery pressures (PAP), pulmonary vein pressures (PVP), and bronchial pressures were monitored continuously via pressure transducers. Transducers were zero referenced to the midpoint of the lobe hilus with electrically meaned pressures continuously recorded on a polygraph (Grass Model 7d; Grass Instruments, Quincy, MA).

Inflation and Deflation Cycles

Once the isolated lobes were perfused, the bronchial clamp was released and the lobes were allowed to completely deflate. A syringe, filled with room air, was then attached to the bronchial catheter and inflation was initiated from complete collapse to total lobe capacity (TLC), 240 ml, in 20-ml aliquots (14, 15). The lobes were given time for the PAPs and airway pressures to stabilize between each stepwise increase or decrease. Pulmonary vascular resistance is normally calculated as the difference between mean PAP and left atrial pressure divided by the cardiac output. In this model, because blood flow and PVP were held constant during the inflation/deflation cycles, PAP changes were a direct index of PVR changes. The 18 lobes underwent inflation and deflation using one of two separate techniques, described subsequently. Data from individual lobes were excluded from analysis if there was a demonstrable bronchial air leak or the PAP tracing was not complete for the entire inflation/deflation cycle.

Group 1 lobes (n = 7) underwent a simple full inflation followed by a full deflation. This technique allowed measurement of PVR hysteresis and CV. The CV was visually determined from the deflation limb lower inflection point of the pressure-volume curve for each of the Group 1 lobes. This subtle inflection point is where the slope, which represents compliance, becomes less steep. The mean CV for the Group 1 lobes was used to estimate the CV for the Group 2 lobes.

Group 2 lobes (n = 11) underwent two deflation/inflation cycles, after an initial full inflation. These cycles, both beginning at TLC, had deflations ending above CV and below CV respectively. Because the range for CV of Group 1 lobes was from 40 ml to 120 ml, Group 2 lobes were not allowed to deflate below 140 ml of volume added initially. This insured no lobe would go below CV. The first full inflation (I1) was from complete collapse to TLC. After this, the first deflation (D1), ended above CV. From there the lobes were reinflated (I2) back to TLC and then fully deflated (D2). This deflation was followed, finally, by a full reinflation (I3) from below CV. The three inflations performed on the Group 2 lobes were to determine if PVR hysteresis would disappear with inflation beginning above CV and recur with inflation beginning below CV.

Statistics

Data are expressed as means ± SEM. Statistical analysis comparing inflation and deflation cycles was done using an analysis of variance for repeated measures. Post hoc subgroup paired comparisons between similar inflation/deflation limbs were done using t tests with a Tukey's adjustment of the p values. Individual points were analyzed with paired t tests. Significance was accepted at p < 0.05. Descriptive composite graphs of the inflation/deflation cycles were constructed by obtaining simple means of the individual points. The graphs were used as illustration and were not used to generate the statistical data.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The CV for Group 1 lobes was 83 ± 10 ml of volume added. Figure 1 depicts the composite pressure-volume compliance curve for Group 1 lobes. There is hysteresis between the inflation and deflation limbs. A hysteresis can also be seen when PAP is plotted against added volume. When volumes 140 ml through 220 ml were analyzed for all 18 lobes, the inflation limb mean PAP was 31 ± 1.6 cm H₂O whereas the deflation limb mean PAP was 28.3 ± 1.3 cm H₂O. This 2.7 cm H₂O difference represents an overall 9% decrease in PVR on deflation (p < 0.0001). Figure 2, a composite inflation/deflation curve for the seven lobes of Group 1, illustrates this hysteresis which became significantly larger as deflation continued. For each volume decrease from 220 ml to 140 ml the difference in PAPs increased. At 140 ml of volume added, the deflation limb PAP was 3.3 ± 0.42 cm H₂O lower than the inflation limb PAP, representing an 11% decrease in PVR (p < 0.01).

View larger version (21K): [in this window] [in a new window]   Figure 1.   Represents the composite pressure-volume compliance of the Group 1 lobes (n = 7). The area between the inflation limb (open squares) and the deflation limb (solid squares) is the hysteresis area. The subtle inflection point on the lower portion of the deflation limb is closing volume (CV). The CV for Group 1 lobes was 83 ± 10 ml.

View larger version (19K): [in this window] [in a new window]   Figure 2.   Depicts pulmonary artery (PA) pressure plotted against volume for the Group 1 lobes (n = 7). Because blood flow and pulmonary vein pressure are held constant, PA pressure is an index of PVR. The hysteresis between the inflation limb (open squares) and the deflation limb (solid squares) became significantly larger as deflation continued. For each volume decrease from 220 ml to 140 ml the difference in PA pressures increased. At 140 ml of volume added, the deflation limb PAP was 3.3 ± 0.42 cm H2O lower than the inflation limb PAP, representing an 11% decrease in PVR (p < 0.01).

The PVR hysteresis seen between the initial inflation (I1) and deflation (D1) in Group 2 is absent when a second inflation (I2) and deflation (D2) are added (Figure 3). The mean PAPs, for volumes 140 ml through 220 ml, for the first deflation (D1, 29.6 ± 2.1 cm H₂O), the second inflation (I2, 29.3 ± 2.2 cm H₂O), and the second deflation (D2, 28.9 ± 2.2), all of which began from above CV, are not statistically different from each other. They all, however, remain a mean of 2.8 cm H₂O or 9% (n = 9, p < 0.0001) lower than the initial inflation mean PAP (I1, 32.1 ± 2.8 cm H₂O). When the final inflation (I3) is added after a full deflation (D2), the PVR hysteresis returns. When volumes 60 ml through 240 ml were analyzed, I3 had a mean PAP of 27.5 ± 2.2 cm H₂O and was 0.8 cm H₂O or 3% greater than D2, mean PAP 26.7 ± 2.0 cm H₂O (n = 7, p < 0.0038). This represented a 30% return of hysteresis. The final inflation (I3) stands out because the mean PAP for each of the previous limbs incrementally decreases (I1 > D1 > I2 > D2), yet I3 has a higher mean pressure when compared with D2.

View larger version (25K): [in this window] [in a new window]   Figure 3.   Represents the composite PVR curves for all inflations and deflations of Group 2 lobes (n = 7). The mean PAPs for the first deflation (solid line, solid squares), the second inflation (dotted line, open squares), and the second deflation (dotted line, solid squares), all of which began from above CV, are not statistically different from each other. They all are a mean of 2.8 cm H2O or 9% (n = 9, p < 0.0001) lower than the first inflation (solid line, open squares). The third inflation (dashed line, open squares) was 0.8 cm H2O or 3% greater than the second inflation (n = 7, p < 0.0038). This represented a 30% return of hysteresis.

When changes in PAPs were compared with changes in airway pressure rather than volume, the deflation limb was neither consistently higher nor lower than the inflation limb (Figure 4). PVR on deflation was initially greater than on inflation, then less as deflation continued. At the isopressure point, where the two curves crossed, the lobes had over 100 ml more of added volume on deflation than on inflation. Figure 4 represents a composite of only Group 1 lobes, but similar results are seen if the PAPs of I1 and D1 of Group 2 lobes are graphed against airway pressure.

View larger version (18K): [in this window] [in a new window]   Figure 4.   Pulmonary artery (PA) pressure is plotted against airway pressure for the Group 1 lobes (n = 7). The deflation limb (solid squares) crosses the inflation limb (open squares) at approximately 9 cm H2O airway pressure.

The composite pressure-volume compliance curves for the Group 2 lobes (Figure 5) demonstrate similar hysteresis patterns to the PVR curves in Figure 3. In Figure 5, I1 again is distinct from D1, I2, and D2 (n = 8, p < 0.0001), which are clustered together. I3 also shows a return of hysteresis versus D2 (p < 0.0001), albeit somewhat diminished. The difference between the three inflations stands out in both Figures 3 and 5. There is hysteresis between I1 and I3 and their corresponding deflations because they begin from below CV. The second inflation (I2) showed no such hysteresis for PAP versus volume or volume versus pressure because it began from above CV after the majority of alveoli had been recruited.

View larger version (26K): [in this window] [in a new window]   Figure 5.   The composite pressure-volume compliance curves for the Group 2 lobes demonstrate remarkable similarities to the PVR curves in Figure 3. The first inflation (solid line, open squares) again is distinct from the first deflation (solid line, solid squares), the second inflation (dotted line, open squares), and the second deflation (dotted line, solid squares) (n = 8, p < 0.0001). The third inflation (dashed line, open squares) also shows a return of hysteresis versus D2 (p < 0.0001).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This is the first report that shows significantly decreased PVR on deflation. Previous studies have shown supporting findings if evaluated from the perspective of PVR versus volume. Both Thomas and Sun showed a decrease in PVR on deflation when PVR was plotted against changes in lung volume rather than changes in airway pressure (3, 4). In fact, some earlier work also showed the same perplexing relationship of PVR to airway pressure that we found (Figure 4), yet concluded that PVR was higher on deflation (2, 4). The explanation for the different hysteresis patterns seen in Figures 2 and 4 may be that both PAP and airway pressure result from volume added which is the independent variable. Notice the isopressure point on Figure 4 where inflation and deflation cross. The lobes have over 100 ml more volume on deflation at this point. The PVR we measured may be derived from the interaction of lung volume, the direction of volume change (inflation versus deflation), the airway pressure, and surface forces. We do not believe that the hysteresis pattern seen in Figure 4 can be explained by different behaviors of subpopulations of the pulmonary vascular bed, like septal versus corner capillaries, or by zonal conditions. The preferential opening of corner vessels in Zone I conditions was unlikely because the PAP was always higher than airway pressure. Given the lack of an intrapleural pressure gradient and an inflated lobe height of less than 8 cm, the effect of lung zones in our model was also unlikely.

Given the apparent inconsistent relationship of PVR hysteresis when ventilation is viewed from the perspective of airway pressure, we agree with Hildebrandt, who concluded that the dependence of PVR on airway pressure was "illusory" (16). We feel that comparing relative changes in lung volume and PVR, with the style of Burton and Patel (1), is at least as valid as comparing PVR with changes in airway pressure. The comparison of the relative changes of PVR and lung volume is extremely useful, especially because there are strong clinical concerns regarding both end-expiratory alveolar collapse and lung overdistension (6, 17).

Our study is the first to demonstrate the disappearance and return of PVR hysteresis when inflation/deflation cycles begin above and below CV respectively. Whereas CV in healthy lungs is lower than functional residual capacity (FRC), in many clinical conditions involving acute lung injury CV can be greater than FRC. The fact that PVR may increase when lung volume is allowed to go below CV during ventilation is an important clinical concept. The findings of Canada and coworkers in their acute lung injury model (18) provide a bridge from our isolated lung findings to the clinical arena. They found that in healthy dogs a positive end-expiratory pressure (PEEP) of 5 cm H₂O allowed for optimal PVR, oxygen delivery (DO₂), and compliance, while a PEEP of 10 cm H₂O was required in oleic acid-injured dog lungs to achieve the same results. They concluded that a strategy, maintaining end-expiratory lung volume with PEEP, was the logical clinical way to minimize PVR. In addition, they proposed that minimized PVR should be included in Suter's (19) definition of "best PEEP," which already included optimized DO₂, static compliance, Pa_O2, and minimized intrapulmonary shunt. Our findings add further clarification to how PEEP helps to minimize PVR. By maintaining lung volume above CV and avoiding the return of hysteresis, PVR is minimized. This is especially pertinent in acutely ill infants and young children who are already predisposed to having lung CV greater than FRC (20). This includes patients having elevated PVR secondary to respiratory syncytial virus bronchiolitis, congenital heart disease, or adult respiratory distress syndrome (ARDS).

Although the return of PVR hysteresis is small when I3 and D2 are compared, we believe this does represent a real return of hysteresis. The mean PAPs of each inflation or deflation, from I1 through D2, are progressively lower. If the PAP changes we found were an artifact of our isolated preparation, one would expect the mean PAPs to continue to drop as the vessels relaxed after being reperfused. The mean PAP of I2 was less than D1. Yet when I3 began from below CV, its mean PAP was greater than D2. The diminished hysteresis upon reinflation from below closing volume (I3) may be explained, in part, by gas trapping. The 40 to 60 ml of retained air in our lobes prevented full collapse. This air retention is common in work done with excised lungs, and was reported by both Thomas and Glaister (4, 14). Unfortunately gas trapping may blunt full return of both the pressure-volume and the PVR hysteresis as seen in Figures 3 and 5.

We believe that because CV is a function of elastic recoil of the lung, isolated lungs should still accurately demonstrate this feature in the absence of chest wall compliance. In addition, by using an isolated lung preparation we controlled many of the variables such as pulmonary blood flow and pulmonary vein pressures, which change constantly in vivo. This enabled us to study the pressure-volume and PVR changes without many confounding factors. In contrast to some previous studies (2, 3), we have used only physiologic lung volumes (14) and blood flows (21), rather than arbitrarily using airway pressures which may be unnecessarily high except in injured lungs. Because our roller head pump was resistance-independent, blood flow remained constant during inflation and deflation. The blood was diluted with 100 ml of saline which resulted in a mean hematocrit of 36.6 ± 1.9. This dilution was done to provide sufficient blood to perfuse the system. Because the viscosity of the blood was decreased, its effect on PVR and hysteresis may also have been decreased (22).

To date, the relationship of lung inflation to PVR, pulmonary blood flow, and capillary recruitment has not been fully elucidated. It has been well established that the "U-shaped curve" of PVR in response to lung inflation has a PVR nadir at FRC (1). The two-compartment model of alveolar and extra-alveolar vessels attempts to explain this phenomenon (23). Even this model is an oversimplification and fails to accommodate factors such as perfusion conditions (24), rheology (22), or vessel length and diameter (25). Many investigators suggest that surface tension plays a key role in changes of PVR with inflation (2, 3, 26, 27). Additionally, the roles of hypoxic pulmonary vasoconstriction (28), calcium channels (29), and prostaglandins (30) need to be considered.

Although our work did not focus on the etiology of PVR hysteresis, its close relationship to closing volume during ventilation was remarkable. This relationship could be either causal or coincidental. As lungs deflate below closing volume, both alveoli and terminal bronchioles collapse. This collapse results in increased surface tension and would require higher airway pressures to reinflate. Muscedere and coworkers demonstrated the injurious result of this low lung volume ventilation on both respiratory bronchioles and alveolar ducts in rats (10). The increased surface tension and resultant higher airway pressures on reinflation could cause the return of PVR hysteresis as well. If lungs are kept above CV, then no dramatic increase in surface tension occurs after deflation. Both airway pressures and PVR would remain relatively low during subsequent inflation.

Another explanation is that the relationship of PVR hysteresis to airway CV is coincidental with a vascular phenomenon. At low lung volume, pulmonary capillary structure could be altered as suggested by Burton and Patel (1). Their findings of capillary "gnarliness" could represent a process of capillary recruitment and derecruitment associated with lung volume. In support of this concept is Mazzone's work with pulmonary microcirculation. He showed that the internal fold depth or "pleat" in the pulmonary capillaries seemed to diminish with lung inflation (31). In addition, Lebecque and coworkers demonstrated that alveolar capillaries that became unpleated during inflation remained unpleated for periods of time during deflation (32). Weibel and coworkers found that both pulmonary capillary and alveolar volumes were greater on deflation when rat lungs perfused at constant flow were compared at isopressure points (33). Capillary surface area and diffusing capacity of the lungs for carbon monoxide (DL_CO) were also higher when deflation was compared with inflation. Cassidy and coworkers (34) found that in trained human volunteers, the DL_CO was 22% higher at the same lung volume on deflation after an inspiration to TLC. They concluded that DL_CO hysteresis was secondary to unpleating of the alveolar capillary membrane. We can speculate that if the alveolar capillaries are unpleated and have a larger surface area on deflation, they allow not only an increase in DL_CO but also a decrease in PVR. This unpleating is transient however and when deflation below CV occurs both the pleating and PVR hysteresis return.

In conclusion, there is a PVR hysteresis in isolated, blood-perfused dog lungs when inflation begins from complete collapse. Pulmonary vascular resistance is decreased on deflation, and the lower PVR is maintained when subsequent inflation/deflation cycles begin from above closing volume. The PVR hysteresis returns when inflation begins after the lung is allowed to deflate below closing volume. These findings complement recent work done using "open lung" strategies which attempt to ventilate in the so-called "safe zone" between areas of atelectasis and overdistension on the pressure-volume compliance curve (35). In clinical settings of elevated PVR which require mechanical ventilation, a strategy of maintaining lung volume above closing volume with PEEP may help to minimize PVR. Thus, PEEP required to maintain lung volume greater than CV may be another parameter to be included in the definition of best PEEP.