INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Whereas the measurement of lung function is increasingly playing a role in the overall clinical management and assessment of the ventilated neonate, the traditional and commonly employed techniques are inappropriate or of low specificity for the partitioning of global respiratory or pulmonary mechanics into airway and tissue components. Respiratory disease in the newborn, however, is dominated by parenchymal disease processes such as surfactant deficiency, pulmonary interstitial emphysema, and pulmonary hemorrhage. Recent studies also suggest that chronic lung disease in the extremely preterm infant has a significant parenchymal component due to delayed alveolar development (1). To adequately describe the lung mechanics of neonatal patients, therefore, it is important to consider new approaches that may permit the noninvasive partitioning of lung function into airway and tissue parameters.

One particularly promising approach is the modification of the forced oscillation technique (FOT), initially proposed by DuBois and coworkers in 1956, as a means of obtaining detailed information on respiratory mechanics (4). Using the FOT, an impedance spectrum is obtained over the frequency range of the pressure signal (forcing function) applied at the airway opening. When the spectral content of the forcing function includes sufficiently low frequency components (< 2-4 Hz), the roughly inverse dependencies of tissue resistance and reactance on frequency become clearly evident, in association with the largely frequency-independent airway properties of resistance and inertance (5). Qualitatively speaking, this contrast in frequency dependence makes it possible to estimate separately the airway and tissue mechanical properties from input impedance data obtained noninvasively (6).

The major limitation to the low-frequency FOT in spontaneously breathing subjects and animals is the requirement of apnea during the external oscillations, as the small-amplitude oscillatory signal would be corrupted by the signals of spontaneous breathing up to several Hertz of frequency. This difficulty has been recently overcome in infants, by employing the Hering-Breuer reflex and hyperventilation to induce a short apneic interval. Using this method, Sly and coworkers (7) and Peták and coworkers (8) have successfully measured low-frequency (0.5-20 Hz) respiratory impedance in infants aged from 1 mo to 2 yr. Their studies have suggested that similar measurements using a low-frequency impedance spectrum may also be feasible in a neonatal population.

Changes in lung function occurring with increasing age and lung volume are exaggerated in the premature newborn infant compared with the populations studied by Sly and coworkers (7). In addition to increases in lung volume, major structural change is occurring in the architecture of the lung as it passes through the cannalicular and saccular phases of lung growth prior to the stage of alveolarization (9). The postnatal respiratory course of a group of infants at any one gestation may also vary, and may be influenced by the administration of maturational agents during the antenatal period and by postnatal ventilatory strategies and pharmacological therapies. To be clinically useful, therefore, a test of lung function in the neonate needs not only to be able to detect changes that occur with increasing gestation, but also to be able to discern disease severity within a gestational age group. The aim of this preliminary investigation, therefore, was to determine whether the low-frequency FOT technique is technically feasible and can meet these specifications in a population of intubated newborn lambs of varying gestation and disease severity. Furthermore, we aimed to use this technique to assess the relative contribution of airway versus tissue mechanical factors to the overall impedance of the immature lung.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Forced oscillation studies of postnatal lung mechanics were performed on singleton lambs (n = 37) sourced from a separate research study (10) designed to investigate the effects of timing and frequency of antenatal maternal glucocorticoid administration on postnatal lung function and growth retardation. The study was undertaken at the Medina Agricultural Research Station in Western Australia following ethical approval from the animal ethics committees of the Western Australia Department of Agriculture.

Antenatal Treatment

Merino ewes were date mated to deliver at 125 d, 135 d, or 146 d of gestation (term = 150 d). Each ewe received 150 mg medroxyprogesterone (Pharmacia, Upjohn, NSW, Australia) 4 d prior to commencement of the allocated management protocol. Commencing 3 wk prior to planned cesarean delivery, ewes in the 125-d and 135-d gestational groups were randomized to receive weekly intramuscular injections of either saline, a single dose of 0.5 mg/kg betamethasone (Celestone Chronodose, Shering, Australia) followed by two saline injections, or three weekly injections of 0.5 mg/kg betamethasone. The last intramuscular treatment was administered 7 d prior to delivery. Delivery was delayed until term in a further group of ewes that received three weekly intramuscular injections of saline commencing at 104 d.

Delivery

Ewes were sedated with 15 mg/kg ketamine and 0.1 mg/kg xylazine intramuscularly prior to spinal anesthesia with 4 ml of 2% lidocaine. A midline abdominal and uterine incision permitted delivery of the fetal head. The fetus was sedated (10 mg/kg ketamine and 0.1 mg/kg xylazine intramuscularly) and local anesthetic infiltrated into the anterior neck (2% lidocaine subcutaneously). Following tracheotomy, either a 4.5-mm-id (125 d) or 5.0-mm-id (135 d and 146 d), 10-cm-long endotracheal tube (TT) was secured in place to position the tip of the TT approximately 4 cm above the carina. The TT was then suctioned and the fetus was delivered.

Postnatal Care

Lambs were immediately weighed and commenced on intermittent mandatory ventilation using time-cycled pressure-limited ventilators (model BP 200; Bournes, Anaheim, CA) using fraction of inspired oxygen (FI_O2) of 1.0, inspiratory time of 0.7 s, positive end-expiratory pressure (PEEP) of 3 cm H₂O, and a respiratory rate of either 40 bpm (125 d and 135 d) or 30 bpm (146 d). The initial peak inspiratory pressure (PIP) was set to 35 cm H₂O and adjusted to maintain PCO₂ at approximately 50 mm Hg where possible. To prevent barotrauma, PIP was not increased beyond 40 cm H₂O. The investigators attending the lambs were blinded to their treatment randomization.

An umbilical arterial line was positioned to place the tip in the descending aorta. Supplemental pentobarbital (15 mg/kg) was administered as a slow intraarterial push if required to suppress spontaneous respiration. Lambs were kept warm using a radiant heater on an open cot and their rectal temperature was maintained at 39° C. Tidal volume was monitored by using a pneumotachograph (Model No. 3500; Hans Rudolph, Kansas City, NE); PIP was sensed with a pressure transducer (PC Microswitch; Honeywell, Perth, Australia) and adjusted as required to maintain a tidal volume (VT) of less than 10 ml/kg. Respiratory mechanics were measured using both multiple linear regression (MLR) analysis of transrespiratory pressure and tracheal flow (11) recorded during tidal breathing at 40 min of age and the forced oscillation technique (FOT) (see below). Lambs were then killed with 30 mg/kg pentobarbitone, the TT was clamped, and the lungs were excised and weighed.

Forced Oscillation Measurements

Apparatus. The setup for forced oscillation measurements was modified from that used by Sly and coworkers (7), and is illustrated in Figure 1. The TT was connected to both the ventilator and a loudspeaker-in-box. A screen pneumotachograph with a differential pressure transducer (model 33NA002D; IC Sensors, Milpitas, CA) was employed to measure oscillatory flow (V') with an identical transducer utilized for measurement of transrespiratory (Pao) pressure. Esophageal pressure (Pes) was measured using a 5 Fr catheter-tip micromanometer (MTC 5F; Dräger Medical Electronics, The Netherlands). The catheter tip was positioned in the distal third of the esophagus to obtain a smooth respiratory pressure waveform with minimal cardiac artifact, and the position was confirmed with a positive pressure occlusion test (12). Frequency calibration between the flow transducer and each of the pressure transducers was incorporated in the preprocessing of data. The measured signals were low-pass filtered at 25 Hz then digitized at 128 Hz by the analog-digital board of an IBM PC-compatible computer (Dell 486/66) and stored for further analysis.

View larger version (20K): [in this window] [in a new window]   Figure 1.   The experimental setup. During mechanical ventilation, the collapsible tube segments A and B, respectively, are open and clamped. Shortly before the oscillatory measurement at end expiration, segment B and tap C are opened to equilibrate the pressures in the trachea, the loudspeaker-in-box, and reference chamber; segment A is then clamped and tap C closed during oscillations. V', oscillatory flow; Pao, oscillatory component of the tracheal pressure; Pes, esophageal pressure.

Measurement of impedance. During normal ventilation, the connection to the loudspeaker was clamped. To measure the total respiratory input impedance (Zrs*) and chest wall impedance (Zw*), both uncorrected for the apparatus dead space, the flow from the ventilator was interrupted at end expiration (PEEP = 3 cm H₂O) and an ~ 1-s interval was allowed for the pressures in the airways, loudspeaker chambers, and reference box to equilibrate. The reference box was then isolated from the loudspeaker for an interval of approximately 6 s. During this time, the computer-generated pseudorandom signal, comprising 16 frequency components between 0.5 and 20 Hz with a 2-s periodicity, was led to the loudspeaker to produce a low-amplitude (< 3 cm H₂O peak to peak) forcing function at the airway opening. At the end of each measurement, the lamb was returned to its normal ventilation for 30 s before repeating the above procedure. On average, four recordings of Pao, Pes, and V' were obtained in each lamb.

The dead space between the pneumotachograph and the TT (see Figure 1) was considered a lumped shunt impedance of the compressible gas (Zds), and it was determined separately by measuring pressure and V' with the TT clamped at the tip and using the same pseudorandom excitation signal employed for Zrs*.

Computations. Zrs* and Zds were computed from the Pao and V' signals by using the fast Fourier transform with 2- s time blocks and 95% overlapping (5). The corrected total respiratory input impedance (Zrs) was calculated by considering Zrs* and Zds as parallel impedances (4). Zw was calculated similarly, from Zw* and Zds. The impedance of the lung (ZL) was considered in series with Zw and determined as ZL = Zrs  Zw.

The ZL spectra obtained in the same animal and condition were averaged and the mean impedance values were evaluated by fitting to them a model comprising an airway compartment with a frequency-independent (Newtonian) resistance (Raw) and an inertance (Iaw) and a constant-phase tissue compartment characterized by coefficients of tissue resistance or damping (GL) and elastance (HL) (6). Tissue hysteresivity () was calculated as the ratio of GL and HL (13). All frequencies were included in the model fitting except those coinciding with and corrupted by the heart rate and its harmonics.

Statistical Analysis

Results are expressed as the mean value and the standard deviation (SD). Comparison between the control animals at each gestation was performed using one-way ANOVA. The effects of gestation and antenatal betamethasone treatment at both 125 d and 135 d gestation were analyzed using a two-way ANOVA (general linear model with interactions). The differences between the mean values were considered significant when p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Group Parameters

The number, sex, body and lung weights, end expiratory volume (EEV), and deflation lung volume of each group of animals are shown in Table 1.

Impedance Spectra: Reproducibility and Model Fitting

Figure 2 illustrates how the ZL spectra were influenced by gestation and antenatal maternal betamethasone exposure. The impedance data are presented in terms of the real part (resistance, RL) and imaginary part (reactance, XL). The high reproducibility of the measurements is indicated by the small scatter of data at all frequencies except 0.5 Hz and those coinciding with the heart frequency or its harmonics. Qualitatively, the shift of the higher frequency resistance data represents differences in Raw, and the quasihyperbolic decrease in RL at low frequencies corresponds to the frequency-dependent resistive properties of the lung parenchyma (tissue damping). Similar changes in XL at low frequency reflect the elastic behavior of the lung parenchyma, whereas at the high frequencies, XL is dominated by the inertive properties of the lung. The average model fitting error (F) was 6.5 ± 1.5% and the differences in F between gestations or between treatment groups at any one gestational age were small with no systematic fitting error.

View larger version (20K): [in this window] [in a new window]   Figure 2.   Changes in pulmonary resistance (RL) and reactance (XL) versus frequency, associated with differing gestation (left) and in 125 d lambs with differing antenatal maternal betamethasone exposure (right). Mean ± SD values obtained from two to seven impedance measurements in a representative lamb from each group; solid lines indicate the best-fitting model curve to each spectrum.

Gestational Age

The results of the model fitting to ZL in control lambs at each of 125 d, 135 d and 146 d gestation are illustrated in Figure 3. The most significant changes associated with gestation were noted in the tissue parameters GL (p < 0.001) and HL (p < 0.001). GL decreased by 78.9% between 125 d and 135 d gestation and a further 51.1% in the interval from 135 d to 146 d gestation. Similar changes were observed in HL with a 76.7% decrease in the first time period and a smaller (39.1%) change in the subsequent interval. There was a decreasing trend in  with increasing gestation; however, this did not reach significance (p = 0.38). There was a small increase in Raw from 135 d to 146 d gestation, however this was clinically insignificant when compared with the large decreases in GL. There were no significant changes in Iaw (p = 0.74) with increasing gestation (p = 0.11). In contrast, Rrs obtained with the MLR technique decreased with increasing gestation from 54.3 ± 5.4 cm H₂O/L s¹ (125 d) to 26.1 ± 5.2 cm H₂O/L s¹ (146 d) p < 0.001) (Figure 4).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Gestation and pulmonary mechanical parameters at different gestational ages. Raw, airway resistance; Iaw, airway inertance; GL, tissue damping coefficient; HL, tissue elastance coefficient; , tissue hysteresivity. *,#p Values significantly different (p < 0.05) from the 125-d and 135-d data, respectively.

View larger version (8K): [in this window] [in a new window]   Figure 4.   Total respiratory resistance (Rrs) obtained with multiple linear regression analysis of tidal ventilation. Left: Rrs at 125 d (white bars), 135 d (gray bars), and 146 d (black bars) (*p < 0.05 on one-way ANOVA). Right: the effect of antenatal maternal betamethasone exposure (one or three doses) on Rrs at two gestational ages (left group, 125 d, right group, 135 d). White bars, placebo (saline); gray bars, one dose; black bars, three doses: *,# Values significantly different (p < 0.05) from the control and the one-dose data, respectively.

Single versus Repetitive Antenatal Maternal Glucocorticoid

The results of parameter estimation and two-way ANOVA analysis of the effects of single versus repetitive antenatal maternal glucocorticoid prior to delivery at either 125 d or 135 d are shown in Figure 5. The most pronounced effects associated with betamethasone exposure were noted in the tissue parameters. There was a significant decrease in the mean values of GL (p = 0.003) and HL (p = 0.002) among the different levels of treatment after allowing for effects of differences in gestation. There was an interaction between treatment and gestation for both GL and HL, which was evident as a clear dose-response to betamethasone exposure in 125-d lambs; no additional benefit of three doses over one dose of betamethasone was observed in the more mature 135-d lambs. There was a significant effect of betamethasone on  after accounting for the effects of differences in gestation (p = 0.04), observed as a reduction in  associated with multiple dose treatment compared with either control lambs or single-dose groups. There was no significant effect of betamethasone treatment on either Raw (p = 0.17) or Iaw (p = 0.46) after allowing for the effects of differences in gestation. The absence of a significant effect of steroid treatment on Raw contrasts with the results of MLR analysis, according to which Rrs decreases with increasing steroid exposure from 40.2 ± 3.1 cm H₂O /L s¹ (placebo) to 24.7 ± 3.3 cm H₂O /L s¹ (three doses) (p = 0.008) (Figure 4).

View larger version (21K): [in this window] [in a new window]   Figure 5.   Pulmonary mechanical parameters at two gestational ages and different doses of antenatal maternal betamethasone. Raw, airway resistance; Iaw, airway inertance; GL, tissue damping coefficient; HL, tissue elastance coefficient; , tissue hysteresivity. *,# p Values significantly different (p < 0.05) from the control and the one-dose data, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although Dorkin and colleagues have previously reported Zrs data between 4 and 40 Hz in intubated neonates (14), this is the first study using the FOT in a newborn model that incorporates low-frequency (< 4 Hz) forcing function components. Furthermore, as our measurements did not include the impedance of the chest wall they provide a unique insight into the mechanical changes occurring in the lung in association with advancing gestation and pharmacologically induced lung maturation. The use of this approach enables the separation of the airway and tissue mechanical parameters, and has provided the first mechanical evidence of disparity in rates of the in utero maturation of the fetal airway and tissue compartments, and the overwhelming contribution of the tissues to the resistive properties of the respiratory system in a premature mammalian species. The FOT may be a particularly useful tool in the premature and newborn subject in whom respiratory disease has a significant parenchymal component.

Methodological Issues

Our main objective was to obtain meaningful, partitioned measurements of lung mechanics in newborn lambs with varying disease severity using the FOT. In taking advantage of a separate research protocol to achieve this goal, however, we had several methodological constraints. The time available for measurements was limited to a brief interval (approximately 5 min) at the completion of the parent study being conducted in parallel, resulting in a variable number of measurements recorded on each lamb. In some instances, there was insufficient time to collect any data, and consequently unequal numbers in each gestation and treatment group were studied.

Our measurements were performed at a PEEP of 3 cm₂O to maintain the protocol employed by the parent study. This may have affected our measurements, as it is conceivable that 3 cm H₂O was below the alveolar opening pressure of some of the more premature and nontreated lamb groups. A more optimal approach would have incorporated a direct measure of lung volume to ascertain to what extent increases in lung volume associated with increasing gestation or antenatal betamethasone exposure could have accounted for the differences in lung mechanics observed between groups. As this was not practical in the current study, less direct measurements such as body weight, lung weight, end-expiratory volume (EEV), and deflation volume measurements in excised lungs were examined (see Table 1). Although in the mature animal, a clear relationship exists between lung volume and both body and lung weight, the nature of this relationship may be confounded by the presence of variable degrees of surfactant deficiency, alveolarization, and degree of exposure to antenatal fetal lung maturation treatments. The complexity of correcting measurements for changes in lung volume is highlighted by the conflicting results shown in Table 1. A degree of fetal growth retardation was observed in this study similar to that reported previously by Jobe and colleagues (10). In the animals used for this study, however, this was accompanied by a reduction in lung weight only in the 125 d gestation lambs, whereas static volumes obtained during pressure-volume measurements in excised lungs increased with both advancing gestation and steroid exposure. In contrast, however, although the EEV increased with advancing gestation, this parameter was not significantly altered by antenatal steroid exposure within a gestational age group.

The differences between the static and dynamic measures of lung volume raise the question of what is the most appropriate correction factor in the absence of an in vivo measurement of lung volume, and which of the parameters should be corrected. Previously published studies on the embryologic development of the ovine lung have shown that the airways have essentially completed development and branching prior to commencement of alveolarization and that in the gestational age range employed in this study, growth (and hence also increase in volume) is almost entirely restricted to the lung parenchyma (15, 16). It is thus inappropriate to correct measurements of airway resistance for changes in parenchymal volume. For the purposes of demonstrating the partitioning of lung mechanics into airway and parenchymal compartments, measurements have been reported in absolute values.

Gestation

Changes in the mechanical properties of the fetal lamb lung from 125 d to 146 d were determined by changes in the tissue properties. The decrease in elastance that we observed with increasing gestation was not surprising, and agrees with several previous investigations using cruder techniques in lambs (17) and in rabbits (18). This study provides the first evidence, however, that these changes are mirrored by an associated reduction in tissue resistance. The rate of change in GL and HL decreased with advancing gestation, suggesting that the rate of functional maturation slows as the lamb approaches term.

Previous lung function studies in lambs have reported that resistance decreases with advancing gestational age (17), similarly to the values of Rrs obtained with the MLR in our study. Values of Rrs or RL obtained with conventional measurements of lung function combine the airway and tissue resistive properties and thus do not reveal the dominant contribution of the parenchymal resistance at frequencies comparable to the breathing rate. We observed no change in Raw between 125 d and 135 d, and even a small increase in the term animals, which contrasts with the very large changes in GL in the same gestational age groups. Our study provides definitive evidence, therefore, that the decrease in the resistive properties of the lung with increasing gestational age primarily reflects changes in the properties of the parenchyma rather than the airways. From a methodological point of view it also follows that any inference to airway resistance from values of Rrs (or RL) obtained at a single (breathing) frequency would be largely misleading in the newborn lamb. At the functional level, the results of this study are in accord with those of earlier morphological studies showing that the fetal development of the conducting airways precedes that of the lung parenchyma (15, 16). Tepper and coworkers (19) have also demonstrated this phenomenon indirectly, by using V'max/FRC ratios in the postnatal infant. Our study, however, provides the first evidence of the dominant contribution of the tissues to changes in lung mechanics occurring during the final weeks of gestation through the use of a highly sophisticated, specific estimation technique of lung mechanics.

Steroids

The administration of antenatal glucocorticoids to women at risk of delivery prior to 34 wk completed gestation has been shown to reduce the morbidity and mortality associated with hyaline membrane disease in the newborn period (20). The mechanisms of action may be complex, and may vary between species with different functional effects. In animals, these functional changes have been shown to include increased compliance (21) and excised lung volume (22) resulting from both increased surfactant production and function and thinning of the alveolar walls. A reduction in capillary endothelial permeability with a consequent reduction in the amount of proteinaceous exudate within the alveolar space has also been noted (18, 22, 25, 26). It is not surprising, therefore, that we observed the change in lung mechanics associated with antenatal betamethasone to be dominated by changes in the tissue properties with clinically insignificant and inconsistent changes in the mechanics of the airways. Although the reduction in elastance following in utero exposure of fetal lambs to maternal betamethasone has been reported previously (22, 26, 27), this study represents the first report of the associated effect on tissue resistance.

Although there is widespread acceptance of the benefits of antenatal maternal betamethasone, issues such as the optimal frequency and timing of administration remain somewhat controversial. It has been common practice in many units to administer second and subsequent doses at regular intervals when delivery does not ensue within a prescribed period (1-3 wk) and where risk of premature delivery remains high. This practice has recently been questioned, however, with evidence that it may cause fetal growth retardation (10) and delayed myelination (28). Although our results suggest that the mechanical benefit of repeated over single antenatal betamethasone exposure may be influenced by gestational age at dosing, with seemingly little additional benefit of multiple dose betamethasone in the 135-d lamb subgroup; these data need to be interpreted with caution given the particularly small numbers and exclusively male nature of this group.

Of further interest was the observed decrease in  following three exposures to betamethasone. A similar trend in  was noted with increasing gestation, although this did not reach significance. Hysteresivity () reflects the relationship between dissipative and elastic processes within the lung, and has been shown to be fairly constant and species independent in mature healthy animals (13). The presence of emphysema and lung fibrosis has been shown to increase  in vivo in adult mice (29) and with advancing maturation in ex vivo rat lung parenchymal strips (30). No information is available, however, on whether the effect of rapid changes in lung development may influence the values of  in an intact animal model. The findings of our study may indicate that subtle changes in the mechanical coupling of the elastic and resistive properties of the lung occur with both functional and structural maturation of the lung. A recent morphometric analysis also demonstrated an altered balance among collagen, elastin, and ground substance composition of the parenchyma following antenatal steroids that influence the coupling of dissipative and elastic properties of the tissues (31). This interpretation would be consistent with the suggestion made by Fredberg and Stamenovic that a change in  reflects either a change in the hysteretic behavior of at least one of the tissue components, or a change in the mechanical interactions between the parenchymal components (13). This could include altered interaction between actin and myosin, changes in the surface-acting forces, buckling of the surfactant film at different points during both inflation and deflation, and changes in the thickness and density of the connective tissue network.

In conclusion, this study demonstrated that by the model-based evaluation of low-frequency impedance data, changes induced by gestation and antenatal steroid in pulmonary mechanical function can be detected in a neonatal lamb model. We have shown clearly that the changes in the mechanical parameters of the lungs in the newborn lamb are predominantly of parenchymal origin. The capacity to separate the mechanical properties of the airways and the tissues by means of a noninvasive impedance measurement technique represents an important and promising advance in the measurement of neonatal lung function. The incorporation of recent findings on tissue viscoelasticity, that is, the significance and frequency dependence of tissue resistance, in the evaluation of impedance data makes this technique useful not only to follow individual patients' mechanical status, but also to provide comparative lung function data for infants with different breathing frequencies. Longitudinal studies may ascertain the application of this technique in monitoring lung growth, development, and structural integrity in the premature infant with chronic lung disease.