INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary interstitial emphysema (PIE) is a major complication of respiratory distress syndrome (RDS) in preterm infants, usually resulting from mechanical stress (1). Antenatal glucocorticoids reduce the incidence and severity of RDS (2), although it is uncertain whether they decrease the risk of PIE, as very few clinical trials have specifically included pulmonary air leak as an outcome variable. Postnatal surfactant therapy offers a protective effect against PIE (3, 4) and there may be some additive effect of the combined use of antenatal glucocorticoids and postnatal surfactant (5). Contrary to these clinical observations, the lungs of preterm rabbits exposed prenatally to glucocorticoids are more susceptible to rupture during pressure-volume maneuvers (6).

During a series of studies examining the maturational response to direct fetal injection of glucocorticoids in preterm sheep (7), we observed a high incidence of PIE in fetuses delivered 24 h after receiving a single injection of either 0.5 mg/kg betamethasone or 0.5 mg/kg betamethasone plus 15 mg/kg L-thyroxine (T4). The high incidence of PIE was specific to animals delivered 24 h after treatment, as animals delivered 48 h or 7 d after treatment were no more likely to develop PIE than saline controls. The aim of the present study was threefold. First, we retrospectively examined ventilatory parameters and blood gas measurements to determine whether these outcomes were predictive of PIE. Second, we examined lung mechanics data to determine whether there was any indication that animals that developed PIE were being overventilated. By modeling the respiratory system using a volume-dependent single-compartment model we were able to partition total elastance into volume independent (E1) and volume-dependent (E2V) components. E2V provides an indication of overdistension (10). Finally, we examined whether differences in lung structure might account for the increased susceptibility to PIE. We measured parenchymal collagen and elastin content, as alterations in these matrix proteins are known to impact on the mechanical behavior of the lungs (11- 13) and they have previously been shown to be altered by exposure to glucocorticoids (14, 15). Alveolar wall thickness was also measured to ascertain whether excessive wall thinning might account for increased risk of PIE.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

During a series of studies examining the maturational response to antenatal hormone treatment in preterm sheep (7), we observed a high incidence of PIE in fetuses delivered 24 h after receiving a single injection of either 0.5 mg/kg betamethasone or 0.5 mg/kg betamethasone plus 15 mg/kg L-thyroxine (T4). We retrospectively examined ventilatory records and lung mechanics data in these animals to test for any indication of overventilation. We also examined the lungs for morphologic changes that might predispose these animals to PIE. We compared animals delivered 24 h after hormone treatment with animals delivered 48 h or 7 d after hormone treatment, which did not exhibit a high incidence of PIE, to determine whether there were significant structural or functional differences between hormone-treated animals with different treatment to delivery intervals, and whether these differences might relate to ventilator-induced lung injury. A total of 438 animals were delivered during these studies, 112 of which met the criteria for inclusion in the present study: single fetal intramuscular injection of 0.5 mg/kg betamethasone, betamethasone plus thyroxine (T4), or saline; delivery at 128 d gestation; delivery 24 h, 48 h, or 7 d after treatment.

Fetal Treatment

All hormone administration was by ultrasound-guided fetal injection as previously described (7). Fetuses received a single injection of saline or betamethasone (0.5 mg/kg, Celestone Soluspan; Schering Pharmaceuticals, Madison, NJ) with or without L-thyroxine (15 mg/kg, T4; Sigma, St. Louis, MO) 24 h, 48 h, or 7 d before delivery at 128 d (term = 150 d). Treatment protocols are outlined in Table 1. Four fetuses were injected with saline 24 h before delivery but were not mechanically ventilated before lung fixation. These animals were included as a separate group (unventilated control) for morphologic assessments. All other animals were ventilated for 40 min to determine lung function.

For delivery each ewe was sedated with ketamine (1 g intramuscularly) followed by spinal anaesthesia (2% lidocaine, 3 ml). The fetal head was exposed through midline abdominal and uterine incisions and the fetus was sedated (10 mg/kg ketamine). After administering local anesthetic (2% lidocaine, subcutaneously) a tracheotomy was performed and a 4.0 mm endotracheal tube secured in place. Lung liquid was removed by suction through the endotracheal tube. Animals were delivered and the umbilical cord cut.

Postnatal Measurements

After delivery, lambs were weighed, dried vigorously, and covered with plastic wrap to minimize heat loss. Animals were placed on an infant ventilator (Bournes, BP200; Riverside, CA) set to deliver 100% oxygen at a rate of 40 breaths/min, inspiratory time 0.75 s, and positive end-expiratory pressure (PEEP) 3 cm H₂O. Peak inspiratory pressure (PIP) was initially set at 35 cm H₂O. Both tidal volume (VT) and arterial carbon dioxide partial pressure (Pa_CO2) were monitored closely and PIP was adjusted to maintain adequate ventilation. No other ventilator setting was altered during the study. To minimize the risk of ventilator-induced lung injury, PIP was not permitted to exceed 40 cm H₂O. An arterial catheter was advanced to the level of the descending aorta via an umbilical artery and lambs were anesthetized by slow arterial infusion of pentobarbital sodium (15 mg/kg). Dextrose (5%) in water was infused at a rate of 12 ml/h. Temperature was maintained at 39° C. Animals were ventilated for 40 min.

A pressure transducer (model 8507C-2; Endevco, San Juan Capistrano, CA) and pneumotachograph (model 35-597; Hans Rudolph, Kansas City, MO) were placed between the tracheostomy tube and the ventilator to measure tracheal pressure (Ptr) and flow (), respectively. Volume (V) was obtained by integrating flow. An estimate of VT was obtained at this point to ensure that animals were not being ventilated at excessively high volumes (> 12 ml/kg) and PIP was adjusted accordingly. This initial VT estimate was taken at approximately 6 to 7 min after delivery. Dynamic respiratory mechanics were measured at 10 min and thereafter at 10-min intervals. Estimates of dynamic resistance (RRS) and elastance (ERS) were obtained by multiple linear regression analysis of pressure, flow, and volume signals according to a volume-dependent, single-compartment model (VDSCM):

(1)

(2)

where PA_EE is end-expiratory alveolar pressure, E1 is volume-independent elastance and describes the slope of the linear portion of the pressure-volume relationship, and E2V is volume-dependent elastance which describes nonlinear behavior at either extreme of the pressure-volume relationship (10). The inclusion of a volume-dependent elastance term in the model yields improved fits to the data (16).

Arterial oxygen (Pa_O2) and carbon dioxide (Pa_CO2) partial pressure were measured at 10-min intervals. Target Pa_CO2 was 45 to 50 mm Hg, however animals were permitted to become hypercarbic when target Pa_CO2 was not able to be attained at maximal PIP (40 cm H₂O).

PIE

The presence or absence of PIE was examined macroscopically when the chest was opened. PIE was characterized by the presence of bullae (air cysts) beneath the surface of the visceral pleura. No attempt was made to grade the severity of PIE. Of the group of fetuses delivered 24 h after hormone administration, approximately 70% (13 of 19) had macroscopic evidence of PIE (Table 1). The incidence was equally divided between the betamethasone group (6/9) and the T4 + betamethasone group (7/10). By comparison only two animals in each of the 48 h hormone and saline groups had PIE, and there was no PIE in the 7-d hormone group. For all functional and morphometric comparisons, the 24-h hormone group has been separated into +PIE (n = 13) and PIE groups (n = 6). Those animals that developed PIE in the saline and 48-h hormone groups have been excluded from all analyses.

Morphometry

As lung maturation is known to vary between regions of the lung (17, 18), all morphometric assessments were performed on the same region of each lung, the right cranial lobe. The right cranial lobe was fixed overnight at a pressure of 30 cm H₂O via bronchial instillation of Karnovsky's fixative. All morphometric measurements were performed blind by the same operator (K.E.W.). Lung samples from all 24-h hormone-treated animals were fixed for morphometric examination. In each of the other groups, a subset of animals was randomly chosen for morphometric assessment, and lungs from other animals were not fixed. These animals were representative of the group as a whole for all outcome variables examined (data not shown). Fixed lung tissue samples were available for unventilated (n = 4), saline (n = 12), and 24 h (n = 19), 48 h (n = 4), and 7 d (n = 6) hormone-treated animals. Each lobe was cut into 5-mm serial slices and three slices were randomly chosen for morphometric examination (19). One 5-µm section per slice (i.e., three sections per lobe) was stained with hematoxylin and eosin. Each hematoxylin-eosin section was assessed for evidence of alveolar damage at low magnification (×4-×10 objective) under light microscope. A scale was devised to quantify the degree of alveolar distortion (Table 2).

Digitized grayscale images from 10 nonoverlapping parenchymal fields were captured from each 5-µm section using a Sony 3CCD Color video camera interfaced with a Leica DMLS microscope and a Macintosh 8100/80AV computer. Images were examined at a final magnification of ×1,200. The number of points that fell on airspace and on alveolar septal tissue and the number of air/tissue tissue/air intercepts were counted by superimposing a linear point counting grid (21 lines/42 points). Alveolar wall thickness (TD) was determined as volume per unit area of alveolar surface according to the formula TD = (AWF * L_r)/2I₀ where AWF is the volume fraction of alveolar wall tissue, L_r is the length of the test line within the reference volume (parenchymal fields), and I₀ is the number of intercepts with the air-tissue interface.

Parenchymal Collagen and Elastin Content

Three 5-µm sections per lobe were stained with Picro-sirius red and three with Miller's elastic stain which selectively stain collagen (20) and elastin (21), respectively. Digitized images were captured as described previously to estimate collagen and elastin content. Both collagen and elastin could be distinguished from surrounding tissue on the basis of pixel density. Parenchymal elastin content was estimated from images of 10 parenchymal fields on Miller's elastic stained sections, as previously described (22). Elastin fraction (EF) was determined as the volume of elastin per unit volume of alveolar septal tissue:

(3)

Collagen fraction (CF) was similarly calculated on Picro-sirius red stained sections (22).

Statistical Analyses

Group differences in alveolar distortion scores were examined by the Kruskal-Wallis test. All other clinical, mechanical, and morphometric indices were examined by one-way analysis of variance (ANOVA). Where the difference between treatment groups was significant, post hoc pairwise comparisons between the control group and individual treatment groups were performed using Dunnett's multiple comparison test. Statistical significance was accepted at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Clinical Outcomes

Changes in VT and PIP were tracked throughout the study period to ascertain whether these parameters provided any indication that animals that developed PIE were being overventilated. Initial estimates of VT, taken at approximately 6 min postdelivery, were significantly higher in the 48-h hormone- treated group than in saline controls (p < 0.05, Figure 1A). Average VT in this group was almost double that of the control group, with 13 of the 32 animals having VT in excess of 12 ml/kg. Initial VT was also significantly higher in the 24 h PIE group of animals, with three of the eight animals being ventilated at greater than 12 ml/kg (p < 0.05, Figure 1A). Only one of the 19 animals in the 24-h +PIE group was being ventilated at greater than 12 ml/kg at this time, average VT being less than 8 ml/kg which was not significantly different from control animals. Despite the high VT, 6 min PIP was significantly lower in the 48 h hormone group than in control animals (p < 0.05, Figure 1B).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Initial ventilatory measurements recorded at 6 min postdelivery. SAL = saline control; 24H +PIE = 24-h hormone-treated with macroscopic evidence of PIE; 24H PIE = 24-h hormone-treated without PIE; 48H = 48-h hormone-treated; 7D = 7-day hormone-treated. (A) VT. Significant increase in 24H PIE and 48H groups versus control. (B) Peak inspiratory pressure. Significant decrease in 48H group. (C ) E2V. Significant decrease in 24H +PIE and 24H PIE groups. *p < 0.05 versus SAL.

During the ensuing study period VT increased slightly in control, 24 h +PIE, and 7-d hormone groups, and decreased slightly in the 24 h PIE, and 48-h hormone groups (Figure 2A). Peak pressure increased slightly in the control and 24 h +PIE groups and decreased in the 24 h PIE and 48 h hormone groups (Figure 2B). By 40 min, peak pressure was the only clinical variable that distinguished hormone-treated animals with PIE from those without PIE (Table 3). All hormone-treated groups, with the exception of those animals at 24 h that developed PIE, were able to be ventilated at lower peak pressures than control animals (p < 0.05, Table 3). Forty-minute VT varied between groups, being significantly higher in the 48 h hormone-treated group than in control animals (p < 0.05, Table 3).

View larger version (16K): [in this window] [in a new window]   Figure 2.   Time course of ventilatory parameters. SAL = saline control; 24H +PIE = 24-h hormone-treated with macroscopic evidence of PIE; 24H PIE = 24-h hormone-treated without PIE; 24H = 48-h hormone-treated; 7D = 7-day hormone-treated. SAL (), 24H +PIE (), 24H PIE (), 48H (), 7D (). (A) VT. 48H group consistently highest throughout study. (B) PIP. 24H +PIE group intermediate between SAL and other hormone-treated animals. (C ) E2V. Progressive increase in 24H +PIE group (p < 0.05), no change in other groups.

Blood gas measurements did not distinguish those animals that developed PIE from those that did not. Although there was some variability, particularly in Pa_O2, gas exchange at earlier time points was similar to that at 40 min. Pa_CO2 was 16 to 20 mm Hg lower in all hormone-treated groups compared with control animals (p < 0.05, Table 3). Pa_O2 was approximately 30 to 50% higher in all hormone-treated groups, with the exception of the 24 h PIE group, in which values were similar to control animals (Table 3).

Dynamic Mechanics

Total elastance was partitioned into E1 and E2V components. E2V was tracked over time to see whether PIE animals behaved differently to others. The first available measurement was taken at approximately 6 min postdelivery. At this time, E2V was greatest in the control and 7-d hormone groups, intermediate in the 48-h hormone group, and lowest in the 24 h +PIE and PIE groups (Figure 1C). Both the 24 h +PIE and 24 h PIE groups had significantly lower E2V than control animals (p < 0.05 for both groups). E2V was negative, indicating ventilation at low volumes, in 11 of the 19 animals in the 24 h +PIE group. During the study period E2V increased significantly in the 24 h +PIE group (p < 0.001, one-way repeated-measures ANOVA) but was relatively constant in all other groups (Figure 2C). Despite the progressive increase with time in the 24 h +PIE group, by 40 min, E2V was still lower in this group than in all other groups.

In most animals, both resistance and elastance improved slightly during the 40-min study period (data not shown). At 40 min total elastance (ERS) was significantly lower in all hormone-treated groups (p < 0.005, Figure 3A). The relative contribution of volume-dependent and volume-independent components to the reduction in ERS varied with treatment to delivery interval. E2V decreased markedly with a short (24 h) treatment to delivery interval, then increased progressively, returning to control levels by 7 d (p < 0.001, Figure 3A). Conversely, E1 decreased progressively with increasing treatment to delivery interval, although this trend just failed to reach statistical significance (p = 0.06). In the 24 h +PIE group, the reduction in ERS was due solely to a reduction in E2V and in the 24 h PIE group there was a small reduction in E1 and a relatively large reduction in E2V. In contrast, 48 h after treatment the reduction in ERS was the result of proportionate decreases in both E1 and E2V, and after 7 d the reduction in ERS was solely attributed to a decrease in E1. Resistance was not significantly affected by hormone treatment (Figure 3B).

View larger version (19K): [in this window] [in a new window]   Figure 3.   (A) Total elastance (ERS). Group mean values of ERS with standard error bars are shown. ERS is partitioned into volume-independent (E1, black) and volume-dependent (E2V, white) components. SAL = saline control; 24H +PIE = 24-h hormone-treated with macroscopic evidence of PIE; 24H PIE = 24-h hormone-treated without PIE; 48H = 48-h hormone-treated; 7D = 7-d hormone-treated. E2V decreased markedly 24 h after hormone injection, thereafter increasing progressively with treatment to delivery interval (p < 0.001). E1 decreased progressively with increasing treatment to delivery interval, although this trend was of borderline significance only (p = 0.06). (B) Total resistance (RRS). Group mean values with standard error bars are shown. *p < 0.05 versus SAL.

Alveolar Distortion and Wall Thickness

Almost all ventilated animals, including controls, exhibited some evidence of alveolar distortion, whereas there was no evidence of lung injury in unventilated control animals (Figure 4). Distortion of the parenchymal architecture was typically characterized by grossly enlarged respiratory bronchioles and alveolar ducts with shallow alveoli, interspersed with normal and atelectatic regions (Figure 4A). The presence of numerous spherical air "cysts" suggested the rupture of airspaces (Figure 4B). There was no evidence of alveolar distortion in lungs from unventilated control animals (Figure 4C). In general, parenchymal architecture of ventilated control animals was similar to that of their unventilated counterparts (Figure 4D).

View larger version (93K): [in this window] [in a new window]   View larger version (86K): [in this window] [in a new window]   View larger version (112K): [in this window] [in a new window]   View larger version (111K): [in this window] [in a new window]   Figure 4.   Five µm hematoxylin-eosin-stained sections. (A) 24-h betamethasone-treated animal (#93-92). Distortion of parenchymal architecture was characterized by grossly enlarged alveolar ducts (AD) and respiratory bronchioles (RB) with shallow or missing alveoli. (B) 24-h betamethasone-treated animal (#93-92). Rupture of airspaces characterized by the presence of spherical "air cysts" (arrows). (C ) Unventilated control animal (#93-UV2). No evidence of dilated alveolar ducts or respiratory bronchioles. (D) Ventilated control animal (#93-84). Parenchymal architecture similar to that of unventilated controls in most regions of lung. * denotes terminal bronchiole (TB). All images at original magnification ×125.

The alveolar distortion score differed significantly between groups (p < 0.005, Figure 5A), being most severe in the 24 h +PIE group (p < 0.05 versus ventilated saline control). The 24 h PIE group also showed considerable distortion, although this was not statistically significant compared with ventilated control animals (p = 0.09). The 48-h and 7-d hormone groups were comparable to ventilated controls.

View larger version (24K): [in this window] [in a new window]   Figure 5.   (A) Alveolar distortion score. Median values with 25th to 75th centile (box) and 10th to 90th centile (error bars) are shown. Data points below 10th centile or above 90th centile are represented by filled circles. UV = unventilated saline control; SAL = ventilated saline control; 24H +PIE = 24-h hormone-treated with macroscopic evidence of PIE; 24H PIE = 24-h hormone-treated without PIE; 48H = 48-h hormone-treated; 7D = 7-d hormone-treated. All ventilated groups showed some evidence of alveolar distortion (p < 0.0005). Distortion in the 24H +PIE group was significantly greater than in ventilated control group (p < 0.05). (B) Alveolar wall thickness. Group mean values with standard error bars are shown. Significant reduction 48 h after hormone treatment. *p < 0.05 versus SAL.

Mean alveolar wall thickness was approximately 4 µm in control animals. There was no indication of excessive wall thinning in the 24 h +PIE group, the mean value being comparable to control. There was, however, significant thinning of alveolar walls 48 h after hormone treatment (p < 0.05, Figure 5B).

Parenchymal Collagen and Elastin Content

Parenchymal CF varied significantly between groups (p < 0.0005, Figure 6). Those animals that developed PIE at 24 h exhibited a reduction in CF of approximately 35% (p < 0.001 versus ventilated control) whereas those animals that did not develop PIE at 24 h were similar to saline controls. By 48 h after hormone exposure, CF was almost 30% higher than in controls (p < 0.01 versus ventilated control). Parenchymal elastin fraction (EF) also varied significantly between groups (p < 0.05, Figure 6). EF in the 24 h +PIE group was approximately double that in the ventilated control group (p < 0.005) whereas other treated groups were similar to control animals.

View larger version (17K): [in this window] [in a new window]   Figure 6.   Parenchymal collagen and elastin fractions. Group mean values with standard error bars are shown for collagen (black) and elastin (gray). SAL = ventilated saline control; 24H +PIE = 24-h hormone- treated with macroscopic evidence of PIE; 24H PIE = 24-h hormone- treated without PIE; 48H = 48-h hormone-treated. Significant reduction in collagen fraction in 24H +PIE group and significant increase in 48-h hormone group compared with saline control. Significant increase in elastin fraction in 24H +PIE group compared with control group. *p < 0.05 versus SAL.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our aim in undertaking this study was to determine whether the high incidence of PIE that we observed in mechanically ventilated preterm lambs delivered 24 h after antenatal hormone injection could be attributed to either systematic overventilation of these animals, or to structural or functional differences in their lungs. The aggressive ventilation strategy employed in this study almost certainly increased the extent of alveolar injury (23, 24). However, in the absence of exogenous surfactant administration, the high ventilatory pressures used were necessary to maintain adequate ventilation, as the lungs of fetal sheep at this gestational age are extremely immature and surfactant-deficient. We have used this ventilation strategy in almost 1,000 preterm lambs, during the course of numerous studies (7, 25). The overall incidence of PIE in these studies is approximately 8%, and falls to less than 6% when animals delivered 24 h after hormone treatment are excluded. It should be emphasized that in the present study only those fetuses delivered 24 h after hormone injection were susceptible to developing air leak, although the criteria for adjusting PIP were uniformly applied across all treatment groups. With Pa_CO2 averaging more than 50 mm Hg and VT below 10 ml/kg, the operators (who were blinded to the treatment received by each lamb) were obliged to maintain relatively high peak ventilatory pressures in these PIE animals in order to maintain adequate ventilation. Our data do not support the hypothesis that these animals were overventilated in comparison to other animals VT, peak pressure, and Pa_CO2 fell within the defined ranges. Furthermore, E2V, an index which has previously been shown to provide an indication of overdistension (10), was lower in the 24 h +PIE group than in any other group. The increase in susceptibility to PIE appears to be a transient effect, occurring frequently 24 h but not 48 h or 7 d after hormone injection. Delivery times of less than 24 h also do not appear to be associated with a high incidence of PIE (28).

It may be argued that the high incidence of PIE in the 24 h hormone-treated group was due to volutrauma, as VT were higher in this group than in control animals. However, if high VT was the major contributing factor to the development of PIE, we would expect to see the highest incidence of PIE in the 48 h hormone group, in which a large proportion of animals had initial VT well in excess of 12 ml/kg, and average VT throughout the study period exceeding 10 ml/kg. However, the incidence of PIE in this group of animals was very low. Our findings therefore suggest that the lungs of animals delivered 24 h after hormone treatment do not tolerate high VT as well as animals delivered 48 h or more after hormone treatment. Similarly, it could be argued that PIE resulted from barotrauma, animals with PIE were ventilated at higher PIP than other hormone-treated animals. If this argument held true, we would see an equally high incidence of PIE in control animals, which were ventilated at similarly high PIP throughout the study period. However, the incidence of PIE is low in control animals, therefore we must conclude that the lungs of animals delivered 24 h after hormone treatment do not tolerate high ventilatory pressures as well as their control counterparts.

In contrast to other hormone-treated animals, those fetuses that developed PIE had relatively high E1 and low values of E2V. There are several possibilities to account for the low E2V in 24 h +PIE animals. Conceptually, the coefficient E1 describes the slope of the linear portion of the static pressure- volume curve, whereas E2V describes nonlinearities at either end (29). By moving the position of the VT loop up or down the volume axis, the contribution of E2V to ERS can be altered. A positive E2V can be indicative of overventilation and is observed when a portion of the tidal pressure-volume loop extends onto the upper nonlinear portion of the pressure-volume curve. Conversely, a negative E2V is observed when a portion of the tidal pressure-volume loop spans the lower nonlinear portion of the pressure-volume curve and may indicate underventilation. A reduction in PEEP or a longer expiratory time could decrease E2V by lowering the position of the VT loop on the volume axis (29). However, as all animals were ventilated with a fixed PEEP of 3 cm H₂O and expiratory time of 0.75 s, changes in these parameters do not account for the low E2V in animals with PIE.

Alternatively, a low E2V might indicate the presence of an air leak. As alveoli approach maximal distending volume during inspiration, a substantial air leak into the interstitium would enable the volume of air delivered by the ventilator to continue to increase unhindered by the opposing pressure of near maximally distended alveoli. If this were the case, a drop in E2V might be observed during the course of the study. There was no drop in E2V during the study in animals with PIE. However, if air leak was present at the time of the initial measurement at 6 min, E2V would already be low; therefore the possibility of a significant fall in E2V during the first few minutes after delivery cannot be excluded. It should also be noted that the accuracy of E2V, and other estimates of dynamic mechanics, may be greatly reduced where air leak is present.

The final possibility is an alteration in intrinsic tissue mechanics at high lung volumes as a result of structural changes in the lung parenchyma. A reduction in E2V may reflect an increase in the number of functional lung units, or an increase in elasticity of existing lung units at high lung volumes, whereas a reduction in E1 probably reflects increased elasticity of existing lung units at low to middle lung volumes (i.e., within the range of normal tidal breathing) (29). In the present study, 24 h after hormone treatment the lungs appeared to be less stiff at high lung volumes, whereas by 48 h elastance at low and high volumes was equally affected, and by 7 d the effect on elastance at high volumes appeared to subside and the effect on low volumes remained. Our finding of increased susceptibility to rupture coinciding with altered pressure-volume behavior at high volumes is consistent with previous findings (12).

The susceptibility to lung injury 24 h after betamethasone treatment might reflect underlying structural changes that render the lung parenchyma more fragile. Perhaps the most plausible explanation would be excessive thinning of alveolar walls. Glucocorticoids have been shown to promote thinning of alveolar septa (30, 31). However, in the present study, susceptibility to alveolar rupture did not appear to coincide with wall thinning. Alveolar wall thinning was evident only in animals delivered 48 h after hormone treatment, but these animals were not prone to lung injury. Conversely, wall thickness in animals delivered 24 h after hormone treatment was comparable to control animals, yet these animals were highly susceptible to ventilator-induced injury.

Increased susceptibility to alveolar rupture might alternatively be explained by changes in structural proteins. The network of collagen and elastin fibers in the alveolar interstitial matrix forms the structural framework of the lung parenchyma. Elastin with its high level of distensibility is thought to impact primarily at low and middle lung volumes, whereas collagen with its high tensile strength and low compliance acts primarily at high lung volumes, serving to limit maximal distension (32). Disruption of collagen deposition has previously been observed in weanling rats exposed to semicarbazide, a lathyrogen capable of interfering with collagen cross-linking (12). Lungs from semicarbazide-treated animals exhibited greater compliance at high lung volumes and ruptured at approximately 11 cm H₂O less than those of control animals. Exposure to collagenase has also been shown to increase susceptibility to rupture during excised lung pressure-volume curves (11), and to markedly reduce the force required to break lung parenchymal strips (33). Tensile strength of human lung parenchymal strips exhibits a strong positive correlation with collagen content and is greatly reduced after collagenase treatment (34). Glucocorticoids have been shown to decrease procollagen messenger RNA (mRNA) synthesis (35) and to inhibit activity of lysyl oxidase, an enzyme participating in cross-link formation (36). The rate of lung collagen turnover in adult animals is approximately 10%/day (37) and may be considerably higher during lung development (38). The reduction in interstitial collagen in animals with PIE in the present study may have contributed in part to the increased susceptibility to lung injury.

Perhaps more remarkable than the decrease in lung collagen was the very rapid, but transient increase in elastin 24 h after hormone treatment. Other researchers have also reported upregulation of elastin in fetal lungs after exposure to corticosteroids (15, 39). Pierce and colleagues demonstrated a 7- to 10-fold increase in tropoelastin mRNA levels within 24 h of exposure to dexamethasone in rat fetal lung organ culture, although no attempt was made in this study to examine the relationship between mRNA and protein levels (39). There are no previous reports in the literature of an increase in elastin (protein), of the magnitude observed in the present study, within 24 h of exposure to glucocorticoids. However, Lui and coworkers reported a 4-fold increase in soluble elastin in cultured neonatal rat lung fibroblasts within 24 h of exposure to another potent stimulus, retinoic acid (40), suggesting that hormonal intervention can lead to very rapid upregulation of protein synthesis. These investigators also found upregulation of protein to be a transient phenomenon, reporting a decline in protein levels after 48 h, although mRNA levels remained high for at least 96 h.

Physiologically, those animals that developed PIE were intermediate between control animals and other hormone-treated animals: gas exchange was comparable to other hormone-treated animals, but only when animals were ventilated at high ventilatory pressures. Whereas total elastance was comparable in hormone-treated animals with and without PIE, volume-dependent elastance was low in animals with PIE, which may be a reflection of structural differences, or alternatively may simply indicate the presence of an air leak. Importantly however, there is no indication from E2V, or other ventilatory indices, to suggest that animals with PIE were being overventilated. Both collagen and elastin content were significantly different in animals with PIE, which supports the argument that PIE may be associated with structural abnormalities. As glucocorticoids are only one of a number of hormones and growth factors that participate in fetal lung maturation, exposure to glucocorticoids alone should not be expected to simulate normal maturation, but rather to promote only some of the many changes necessary for the formation of functionally mature airspaces. As such, an inherent risk of precocious administration of glucocorticoids may be the promotion of "structural dysmaturation," which manifests as increased risk of PIE.