INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Currently, there is no consensus as to what occurs at the alveolar level during changes in lung volume (VL). Initially physiologists supported the simple intuitive assumption that lung inflation and deflation result from the summation of isotropic expansion and contraction of the individual alveoli (1- 8). These early studies used morphometric assessment of fixed lung tissue. The various fixation techniques were limited in their ability to distinguish alveoli from alveolar ducts and by artifactual distortion during preservation. The longstanding notion that lung pressure-volume hysteresis is the result of the pressure-volume hysteresis generated by 300 million alveoli is still a common held perception. However, VL change is much more complex. Gil and coworkers have described four possible processes by which the lung may change volume during deflation, including sequential recruitment-derecruitment (R/D) of alveoli, isotropic "balloonlike" alveolar volume (VA) change, simultaneous changes in alveolar size and shape, and crumpling (anisotropic volume change) of the alveolar surface (9). The relative contributions of these proposed mechanisms toward changes in VL, at various levels of inflation and deflation, remain unknown. The precise elucidation of normal and pathologic alveolar mechanics will allow accurate ventilation strategies to be developed and confidently applied.

We hypothesize that change in VL is largely the result of alveolar R/D rather than isotropic (VA) change. This study evaluates the method by which alveoli change volume during inflation from a degassed state to 80% TLC. We also investigate the pressure-volume relationship of single alveoli during tidal ventilation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgical Preparation

Fifteen mongrel dogs (15 to 20 kg) were anesthetized with ketamine (30 mg/kg, intramuscularly) and xylazine (2 mg/kg, intramuscularly) and then pretreated with atropine (0.05 mg/kg, intramuscularly) 10 to 15 min before intubation. Animals were intubated with a divided endotracheal tube (Rosch 13) and ventilated with a Harvard ventilator (Harvard Instrument Co., Cambridge, MA). Ventilator settings were fraction of inspired oxygen (FI_O2) = 0.21, tidal volume (VT) = 12 ml/ kg, respiratory rate = 15, and positive end-expiratory pressure = 0. Polyethylene tubing with 2 mm interior diameter was inserted through a right carotid artery cutdown to monitor systemic arterial pressure. The arterial pressure transducer (Argon Model 049-992-000A; Athens, TX) was leveled with the right atrium and pressure recorded on a Hewlett-Packard Monitor/Terminal (78534C; Palo Alto, CA) with a Hemodynamic Module (78551D; Palo Alto, CA). Samples of arterial blood were obtained for blood gas analysis (Model ABL5; Radiometer Inc., Westlake, OH) to ensure adequate oxygenation and ventilation (Pa_CO2 = 35 to 45 mm Hg, pH = 7.35 to 7.45, and Pa_O2/FI_O2 ratio > 450). A 7.5-French triple-lumen catheter was placed into the adjacent internal jugular vein for infusion of intravenous fluids and anesthetic agents. Continuous anesthesia with sodium pentobarbital (6 mg/kg/h) was delivered via a Harvard infusion pump (Model 907). With the animals in the right lateral decubitus position, we performed a left thoracotomy through the fourth intercostal space to allow access for the in vivo microscope (Figure 1). The tracheal divider was then positioned under direct vision to isolate ventilation to each lung. Separation of ventilation was confirmed by inflating the left lung to an airway pressure (Paw) of 20 mm Hg, occluding the lumen to the left side, and observing no decrease in left-sided Paw for 30 s. Paw of the left lung was measured with a Hewlett-Packard 267BC transducer with a Hewlett-Packard 2754A recorder.

View larger version (33K): [in this window] [in a new window]   Figure 1.   Schematic of experimental preparation demonstrating intubation with a divided endotracheal tube. Inflation from a degassed state to 80% TLC was performed with a Collins syringe. A spirometer and pressure transducer were connected "in-line" to measure exhaled volume and Paw respectively. Ventilation of the animal (right lung only) was maintained with a Harvard ventilator.

Protocols A and B

Two microscopy protocols were necessary to test our hypothesis that alveolar R/D accounted for VL change. In Protocol A, no suction was used to stabilize the pleural surface because this could prevent accurate observation of alveolar recruitment. However, in Protocol B we wanted to generate the pressure/volume curve of a single alveolus during tidal ventilation. This required the use of suction to stabilize the pleural surface and to allow filming of a single alveolus throughout the ventilation cycle.

Protocol A: Determination of R/D versus isotropic volume change during change in VL. Measurement of TLC. After surgical preparation, the tube to the left lung was clamped to allow degassing by oxygen absorption (approximately 10 min), while ventilation to the right lung was maintained throughout the experiment with 100% oxygen. The VT to the right lung was decreased, and the respiratory rate increased to maintain a consistent minute ventilation and avoid iatrogenic injury to the right lung. TLC of the left lung was established by inflating the degassed lung with a 1-L Collins syringe to 25 mm Hg over 1 min and recording the volume. This pressure correlates with TLC in canine models. The lung was then allowed to passively deflate to atmospheric pressure into a Collins 1-L spirometer. Pilot investigation with our open-chest model revealed that passive deflation was to residual volume (RV) not FRC. In the present study, deflation to RV was ensured by measuring volume expired into the spirometer (Figure 1).

In vivo microscopy. The lung was positioned for in vivo microscopy by supporting it slightly off the heart using a netting secured to the thoracotomy margin. This served to eliminate cardiac motion artifact. The epiobjective, metallurgic microscope (Wild-Heerbrugg, Switzerland, M50) was placed into the thoracotomy incision. Subpleural alveoli were observed and photographed at ×150 with dark field illumination. Still photomicrographs were taken with a 35-mm Nikkormat camera and Ektachrome 400 ASA film. The microscope stage was constructed from a round, flat piece of aluminum with a 22-mm-diameter glass coverslip glued to the center. This entire apparatus was lowered until the coverslip lightly touched the pleural surface. No suction was required to maintain contact. However, free movement of the pleural surface under the coverslip during changes in VL made it impossible to keep the microscope focused on the same field. The 40 still photomicrographs were later examined at each VL. To assure randomness, the camera position was altered periodically. Each series of 40 photomicrographs contained an average of 393 alveoli (range 222 to 564). The subpleural structures measured have been previously identified as true alveoli by scanning electron microscopy (10).

Experimental design. With TLC defined and the in vivo microscope in position, the left lung was again degassed and a ventilation cycle was performed in the following fashion. With the 1-L syringe the lung was slowly inflated from the degassed state to 80%TLC. Photomicroscopy was performed during ventilation from opening volume (OV, when alveoli first recruit), 20, 40, and 80% TLC and then upon deflation to RV. Inflation was halted for 15 to 30 s to secure focus and obtain still images. Still images were examined to determine alveolar number per (NA) microscopic field. These images were then projected onto a grid to determine alveolar area by point count method. Alveolar area data were later analyzed to determine change in VA.

Measurement of VA. To quantify VA change, we first needed to obtain the area of the alveoli being observed. Because the still images of subpleural alveoli are not perfectly round nor could we be certain of the plane of our observation, a direct measure of the alveolar radius was not obtainable. Indeed it is well known that massed alveoli are not spherical (Figure 3), and ascribing a valid measurement of alveolar radius is problematic. As an alternative method, still images were projected onto a screen where alveolar area was determined by point count method, measuring the actual planar area of the image, as previously described (11, 12). In this manner we determined the area of 223 (range 218 to 256) single alveoli per animal at each of the various levels of inflation (OV, 20, 40, 80% TLC).

View larger version (27K): [in this window] [in a new window]   Figure 3.   Schematic representing subpleural alveoli as viewed through the in vivo microscope. At end expiration three alveoli are observed across the microscope field (A). Upon inflation the alveoli are observed to recruit, and although the new number of alveoli would equal five, again the observed number of alveoli would be three (B).

We were concerned that a two-dimensional (area) measurement may underestimate isotropic changes within the alveolus. To maximize any possible isotropic changes of the alveoli observed, yet remain within the limits of our methods, we chose to calculate the VA change. We first determined the change in alveolar area by the point count method, as described previously, and used this to calculate the "mean" linear change ("mean" change in one dimension). With the assumption that the mean linear change is unbiased for three-dimensional directions, we were able to calculate the VA change assuming a uniform mean linear change in any direction outside of our two- dimensional plane of measurement.

If A = original measured area, and a = fractional change in area, then taking the 3 /2 power gives the fractional change in volume as [(1 = a)^3/2]  1.

Measurement of NA. Our protocol originally determined NA by counting the alveoli per photomicrograph at each VL. The total number of alveoli observed in a series of 40 still photographs averaged 393 per animal (range 222 to 564). However, pilot studies demonstrated that NA could not be accurately assessed by simply counting the number of alveoli per microscopic field. We noted that during inflation there was an increase in surface area and an obvious gross recruitment of alveoli (Figures 3A and 3B), although the field of observation had not changed. Frame-by-frame analysis of continuous photomicroscopy determined that as the lung was inflated, collapsed alveoli were recruited, displacing previously inflated alveoli in a lateral plane. NA obtained by this method would be artifactually low. Accordingly, to obtain the total NA at each test VL, the number of alveoli counted per photograph had to be adjusted for changes in pleural surface area.

To correct for this error, two pairs of dots were placed on the pleural surface of the left lung with a Secure Line laboratory marker. Each pair was placed 20 mm apart in the center of the lobe and oriented perpendicular to each other (dorsal-ventral and apical-basal). The distance between each set of dots was measured with a dial caliper (Mitertoyo D-22; Aurora, IL) at each test VL (OV, 20, 40, or 80% TLC). The distance between each set of dots was used as a diameter to calculate a circular area (Area = r²). The area was calculated for both the dorsal-ventral and apical-basilar radii for each test VL. This was done to correct for any directional inconsistencies in pleural surface area as expansion or contraction occurred with change in VL. After 15 complete inflation-deflation cycles, all area measurements were averaged and used to determine the true NA with the equation: NA (true) = NA (measured) × surface area (at a given VL).

Protocol B: determination of individual alveolar pressure-volume relationship. Surgical Preparation. The experimental preparation for Protocol A was duplicated with the following modifications: dogs were intubated with a standard endotracheal tube (6-French), both lungs were ventilated throughout the experiment and degassing was not performed. Additionally a spirometer was placed "in-line" along both the inspiratory and expiratory limbs of the ventilation circuit.

In vivo microscopy. The methods for microscopy of individual alveoli are described in detail elsewhere (10). Briefly, the same in vivo microscope was positioned as previously outlined in the METHODS here. For assessment of individual alveoli, we used 5 to 10 cm of suction to stabilize the lung so a single alveolus could be observed during positive pressure ventilation. Continuous microscopy was performed with an Arriflex motion picture camera with Kodak Ektachrome 7242 Tungsten film (3200K, 125 ASA; Rochester, NY) at a speed of 25 frames/second. A single switch was used to initiate filming and record airway and hemodynamic pressures so that physiologic data could later be correlated to video data. Although paper speed and film speed were not identical, time was used to calibrate the two readings.

Measurement of VA. VA was determined from continuous photomicroscopy. The area of a single alveolus was determined by the point count method as previously described. VA determinations were then made from the initial alveolar area measurements as described in Protocol A.

Determination of pressure-volume relationship. We determined the lung pressure-volume relationship by recording Paw obtained during tidal ventilation. VL were determined during inflation and deflation by spirometry.

The alveolar pressure/volume relationship was determined from continuous photomicroscopy of single alveoli during inflation and deflation. Single alveoli were chosen for calculation of VA, and Paw for these alveoli were simultaneously recorded. These data were then used to construct the pressure/volume curves.

Statistical Analysis

Differences in either VA or NA at various VL were analyzed using a one-way analysis of variance (ANOVA). Statistical significance between percent of relative contribution from either VA or NA toward VL change was assessed with the Student's t test.

Vertebrate Animal Use

Animals were euthanized with an overdose of pentobarbital (90 mg/ kg intravenously). The experiments described in this study were performed in adherence with the National Institutes of Health guidelines for the use of experimental animals in research. The protocol was approved by the Committee for the Humane Use of Animals at the SUNY Health Science Center, Syracuse, New York.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Determination of R/D versus Change in Isotropic VA

Still photomicrographs of subpleural alveoli at RV (Figure 2A) and 80% TLC (Figure 2B), demonstrating minimal change in alveolar size with a substantial increase in NA during an increase in VL. Deflation from 80% TLC (Figure 2B) back to RV (Figure 2C) confirms minimal change in alveolar size and substantial change (decrease) in NA. Inflation from OV (when alveoli first open from a degassed state; approximately 10% TLC) to 80% TLC was associated with a significant increase in both VA and NA. However, statistically significant changes in VA occurred only between OV and 20% TLC, whereas VA changes at VL greater than 20% were insignificant (Figure 4). More importantly, we documented that inflation from OV to 80% TLC yielded a steady, significant increase in NA as VL increased (Figure 4)

View larger version (88K): [in this window] [in a new window]   Figure 2.   Photomicrographs of subpleural alveoli as the lung is inflated from RV (A) to 80% TLC (B) and back to RV (C ). Lung inflation creates minimal change in VA with a substantial increase in NA.

View larger version (18K): [in this window] [in a new window]   Figure 4.   Change in VA and NA during inflation from OV (approximately 10% TLC) to 80% TLC. Values expressed are mean ± SD of 20 ventilation cycles performed on five individual animals. *p < 0.05 versus baseline (OV); +p < 0.05 versus all other points during lung inflation.

Determination of Alveolar Pressure-Volume Relationship

The pressure-volume curve of the left lung, during tidal ventilation, assumed conventional volume change and hysteresis, whereas the mean pressure-volume curve of individual alveoli demonstrated minimal hysteresis (Figure 5). Although substantial VA change occurs, it cannot account for the entire change in VL. Furthermore, by comparing the pressure-volume relationship of individual alveoli during tidal ventilation, with and without degassing, we established that minimal volume change or hysteresis occurs in either group once alveoli are open (Figure 6).

View larger version (18K): [in this window] [in a new window]   Figure 5.   Pressure-volume relationship of the isolated left lung compared with the pressure-volume relationship of an individual alveolus. Lung data are mean of 15 dogs. Alveolar data are mean of 40 alveoli for each dog.

View larger version (16K): [in this window] [in a new window]   Figure 6.   Pressure-volume relationship of individual alveoli inflated from RV (without degassing) and from a degassed state. Data are means of 40 alveoli from 15 dogs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The principal finding of this study is that with inflation from OV to 80% of TLC, there is minimal change in VA, but a significant increase in NA (recruitment). In our model, the relative contribution to change in VL by changes in NA is much greater than the relative contribution provided by change in VA. Although a classic lung pressure-volume hysteresis pattern is generated during positive pressure ventilation, very little alveolar pressure-volume hysteresis is measured. This calls into question the notion that VL change and its respective pressure-volume curve represents an isotropic volume increase of 300 million alveoli.

Ventilator-induced lung injury (VILI) is an important clinical problem in a significant percentage of patients receiving positive pressure ventilation. Currently there is debate over which ventilation strategy most successfully limits VILI (13, 14). The success of a particular strategy has been judged by improvements in lung function and gas exchange (13, 14), release of proinflammatory cytokines (15), and morbidity and mortality (13, 14). However, data supporting a particular ventilation strategy are empiric, and do not offer proof of a pathophysiologic mechanism. To establish ventilation strategies which best limit VILI, the following must be known: (1) what occurs at the alveolar level during change in VL, (2) how do alveolar inflation patterns change during acute lung injury, and (3) what are the biochemical changes associated with abnormal alveolar inflation. Our study investigated the first of these questions.

Alveolar mechanics during positive pressure ventilation have been studied for a number of years, and yet the principal mechanism by which alveoli change volume remains unclear. Early studies suggested that VL change is complex, with alveolar R/D, isotropic VA change, change in alveolar size and shape, and anisotropic VA change all accounting for VL change (9, 16). The literature suggests that the two mechanisms that predominate are isotropic VA change and R/D of alveoli. Still, the relative contribution of these two mechanisms toward change in VL and the quantitative levels of inflation at which they contribute remains unclear.

Initial research efforts found that uniform, isotropic alveolar expansion and contraction were responsible for the majority of changes in VL (1). The investigations of Staub (6), D'Angelo (1), and Forrest (5) used rapid freezing of fresh lung tissue at various levels of inflation and deflation. These studies indicate that alveolar shape remains relatively unchanged with changes in VL, suggesting that alveoli change volume by isotropic expansion and contraction. Support was obtained from Dunnill, who fixed dog lungs with formalin vapor and compared the regression line of alveolar surface area/alveolar volume to the regression line of alveolar surface area/VL. They noted that the lines were parallel and straight with a slope of 0.67, such that alveolar surface area changed to the 2/3 power of VL, mathematically confirming an isotropic (uniform) alveolar expansion (3). However, Forrest (5) later found that total VA expanded directly with VL and that recalculation of Dunnill's (3) data could yield either VA changing directly with or to the 2/3 power of alveolar surface area. Thus, the precision of Dunnill's data is not sufficient to distinguish between R/D or isotropic VA change as the major mechanism of lung inflation. Nonetheless, other investigations at that time utilized direct visualization of subpleural alveoli and yielded results similar to the morphometric evidence which seemed to establish a balloonlike expansion of alveoli as the primary mechanism of lung inflation (2, 4).

In 1950, C. C. Macklin using histologic assessment demonstrated little change in alveolar size with VL change and felt that alveolar ducts and sacs enlarged to accommodate increases in VL (17). The notion of constant alveolar size was initially supported by Radford who, by directly visualizing subpleural alveoli, found alveolar diameter to minimally increase or decrease with lung inflation and only slightly decrease with deflation (18). Unlike Macklin, Radford observed subpleural alveoli without fixation and concluded that VL change is a result of R/D of alveoli. Investigations by our group support Radford's conclusions as we have demonstrated little change in VA with tidal ventilation (10, 19).

The inconsistencies of these early investigations may be explained by use of different experimental preparations and the relative insensitivity of morphometric techniques (5). Lung tissue was fixed for morphometric study by vascular perfusion (9, 11, 16), formalin vapor (3), rapid freezing (6, 7, 20), and by lavage of fixative (21). Fixation of tissue allows analysis at only one VL per animal, which is a difficult and costly problem with large animal studies. Fixation is also associated with tissue shrinkage and distortion (10, 22, 23). Another problem with morphometric assessment is distinguishing between alveolar ducts and alveolar sacs. Direct visualization avoids these criticisms, but is limited to sampling of peripheral alveoli only.

To avoid the criticisms of fixation and the limitation of viewing subpleural alveoli, Smaldone and coworkers developed a unique technique in which they filled exercised dog lung with a monodispersed aerosol and observed the aerosol deposition by gravity at zero airflow (fixed breath hold) (24). The fraction of deposited/delivered aerosol reflects the cross-sectional geometry of the air spaces. By this mechanism, particle deposition in the absence of flow becomes inversely proportional to the mean linear intercept. Repeat measurements of the mean linear intercept concluded that the lung inflates by a progressive recruitment of alveoli and deflates by alveolar derecruitment. To avoid the difficulty in distinguishing alveolar ducts from alveolar sacs, Lum and coworkers used chord length-frequency distribution analysis of freeze-dried lung sections. Their data also imply that lung inflation is the result of alveolar recruitment (25). Although the investigations by Smaldone and Lum support our results, both used techniques which performed a "breath hold" (Smaldone 6 s, Lum 10 min). The study by Lum and coworkers noted that lungs held at continuous inflation to 30 cm H₂O demonstrated an increase in VL, later confirmed as recruitment by histologic assessment. In open-chest protocols static end-inspiratory pressures will create a constant transpulmonary pressure which could favor recruitment. Although our data agrees with the findings of Lum and coworkers and Smaldone and coworkers, Protocol A also used a breath hold (15 to 30 s) while still images were taken. However, Protocol B employed continuous photomicroscopy during continuous ventilation, and our findings were unchanged. This suggests that recruitment as an artifact of our experimental preparation is unlikely.

In this study we employed in vivo photomicroscopy as a simple and direct method of measuring the changes in VA and NA as VL is varied, without the inherent artifacts of tissue fixation. An additional difficulty with measuring subpleural NA in vivo is the need to correlate changes in NA with changes in the surface area of the lung. As stated in the METHODS we have normalized changes in NA, during change in VL, by factoring in the change in lung surface area. Similar to morphometric techniques, in vivo microscopy of subpleural alveoli also makes it difficult to determine whether an alveolus or alveolar duct is being observed, yet electron microscopy has confirmed that our observation are limited to alveoli only (10). Finally, the argument has been made that degassing of the lung alters pressure-volume hysteresis (26). Because our hypothesis sought to investigate the difference between isotropic volume change and R/D, we needed a method to normalize both parameters at zero before initiating inflation. Additionally, examination of the compliance curves from RV to 80% TLC demonstrates nearly identical hysteresis to the compliance curve observed after degassing (once alveoli are open), directly refuting the argument against degassing (Figure 6).

Finally, the presence of nonhomogeneity in alveolar inflation is to be expected. These data evaluate the subpleural surface only, which leaves opportunity for extrapolation to the remaining lung. Although morphologically distinct, it is of theoretical concern that the pleura and subpleural alveolar wall act in concert. The pleura could then either enhance or prevent change in alveolar size. Consider first if the pleura enhances change in VA. In this study, change in pleural surface area was large (RV to 80% TLC). If subpleural alveoli are tethered to the pleural and the microscope stage is perpendicular to the pleural surface, any change in alveolar size would occur in the two dimensions being observed. Calculation of volume from this change in area would, if anything, overestimate true change in volume. Next consider if the pleural limits change in VA. If subpleural alveoli are tethered to the pleura, minimal change in alveolar area requires minimal change in pleural surface area. In this study there was dramatic change in pleural surface area (RV to 80% TLC) with minimal change in VA. Previous investigations with this model by our group (5, 19) have observed large volume change of individual subpleural alveoli whereas change in pleural surface area was considerably less (FRC to 40% TLC) than in the present study. Although seemingly disparate results, the former investigations evaluated alveoli after surfactant deactivation, markedly altering alveolar behavior.

In summary, we have photographed subpleural alveoli during tidal ventilation and determined that alveoli do change isotropically. However in our model, R/D plays the predominant role in accommodating changes in VL during positive pressure ventilation. We also determined that the alveolar pressure- volume curve during positive pressure ventilation was unlike that of the whole lung. While the lung demonstrated classic pressure-volume curve with a wide hysteresis, the alveolar pressure-volume curve showed no hysteresis and VA did not increase with elevated Paw. These data support the view that VL change is primarily caused by R/D and not by the combined pressure-volume changes of 300 million alveoli. Amato and coworkers have provided a strategy (open-lung approach) for mechanical ventilation of patients with acute respiratory distress syndrome (13). This strategy assumes that the lower inflection point represents the point at which a majority of alveoli recruit while the upper inflection point represents the point at which alveoli begin to overdistend. However, our data suggest that alveolar recruitment is an ongoing process throughout lung inflation, beyond the lower inflection point. Although our findings do not invalidate the survival advantage determined by Amato and coworkers (13), our data call into question their interpretation of the whole lung pressure- volume curve and its relation to the "open-lung" approach. The manner in which surface forces and tissue interdependence influence alveolar mechanics at various lung volumes requires further investigation.