INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Measuring pulmonary mechanics in infants with a history of wheezing disorders is technically difficult and time-consuming. Apart from technical problems there are physiological differences in infant lungs in comparison to adults that lead to controversial reports regarding their response to bronchodilator treatment (1, 2). This is particularly true when airway mechanics are assessed during flow limitation by the rapid chest compression technique because flows achieved during a maximal expiratory effort are dependent on airway caliber as well as airway wall compliance, both of which are likely to be altered by bronchodilator inhalation. Alternatively, changes in airway caliber alone can be assessed from measurements of airway resistance using the whole-body plethysmograph technique (Raw-p) (1). However, whole-body plethysmography is not widely used in infants because it is labor-intensive, the resulting estimates of Raw-p can be influenced by subjectivity of the operator, and the analysis of resistance is difficult because of significant nonlinearities during tidal breathing caused by ventilation heterogeneities, or nonlinearities in the airways pressure-flow relationship. Nevertheless, there is a commercially available plethysmograph (Jaeger, Wurzburg, Germany) that uses an automated algorithm that allows for observer- independent estimates of Raw-p.

Alternatively, respiratory mechanics in adults have been estimated from measurements of respiratory impedance, either input or transfer impedance. Input impedance (Zin[], where  is angular frequency, or 2f ) is defined as the ratio of the Fourier transforms of the pressure and flow signals measured at the airway opening (Pao[] and ao[], respectively) when pressure oscillations are applied at the airway opening. Transfer impedance, Ztr(), is defined as the ratio of pressure at the body surface (Pbs[]) and ao() when pressure oscillations are applied to the body surface. Numerous studies have used Zin() or Ztr() measurements to investigate the mechanical properties of the respiratory system in healthy adults (5) as well as adults with respiratory disease (9). However, there have been few studies reporting Zin() in infants (13) and even fewer reporting Ztr() (20, 21). In a recent study (19), our group found that Zin() in infants is insensitive to changes following methacholine challenge even with a significant decrease in the maximal expiratory flow at functional residual capacity. We concluded that the methacholine-induced increase in peripheral airway resistance resulted in increased shunting of flow into the airway wall shunt impedance as well as the gas compression impedance in the face mask. Thus, in infants, Zin() does not appear to be a sensitive measure of bronchoconstriction, and most likely to bronchodilation. Ztr() measurements in infants before and after some intervention (bronchodilation or bronchoconstriction) have not been reported so the sensitivity of these measurements is not known.

The goal of the current study was to determine whether Ztr() measurements and the resulting estimates of airway and tissue resistance are sensitive enough to indicate changes after albuterol inhalation in 16 infants with wheezing disorders. Ztr() derived estimates of airway resistance (Raw-z) were compared with Raw-p in 10 of the infants.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects and Protocols

The study was performed in 16 infants (7 girls, 9 boys) ranging in age from 15 to 88 wk (46 ± 22 wk, mean ± SD) whose anthropometric data are given in Table 1. The subjects studied had a history of episodic or recurrent cough or wheeze and were referred from the outpatient clinic for routine lung function tests (infant whole-body plethysmography). Patients with upper respiratory tract infections within the last 3 wk were excluded from the study. The addition of Ztr() measurements to the protocol was approved by the institutional review board and informed consent was obtained from the patients' parents.

Experimental Protocol

All measurements were initiated 15 to 20 min after feeding. The subjects were sedated (100 mg/kg chloral hydrate) and the measurements of lung function were performed during behaviorally defined quiet sleep. All infants were pretreated with nasal vasoconstrictors to reduce nasal resistance. During all measurements heart rate and oxygen saturation were monitored by pulse oximetry (Biox III; Ohmeda, Boulder, CO). First, plethysmographic measurements of Raw-p and thoracic gas volume (Vtg), as described subsequently, were made at baseline conditions. The infant was then moved on a tray to the Ztr() chamber without changing the body or head position. Baseline Ztr() was then measured as described subsequently. Heart rate, O₂ saturation, sleep behavior, and respiratory rate were not affected by the applied pressure signal. After these baseline measurements, 600 µg of albuterol was administered from a metered dose inhaler through a modified Babyhaler spacer device (1). Postalbuterol Ztr() was measured and the subject was returned to the plethysmograph for postalbuterol measurements of Raw-p and Vtg.

Plethysmographic measurements. Thoracic gas volume and Raw-p were measured by an improved version of an infant whole-body plethysmograph (Jaeger) that has been described previously (2). The infant was placed in the supine position inside the plethysmograph. A soft latex face mask sealed around the nose and mask to ensure an airtight fit was carefully manipulated into place for the measurements. When the box was closed the infant breathed air from the box through a triple valve system until thermal equilibrium between the infant and the box was reached. A differential pressure transducer was used to detect changes in box pressure (Pb) relative to a compensating chamber of similar volume and time constant. The air in the rebreathing system was humidified (100% relative humidity) and heated (37° C) until BTPS rebreathing conditions were achieved. CO₂ was continuously measured in the circuit. The phase relationship between flow (ao), measured by an infant size pneumotachometer (Jaeger), and Pb was displayed and monitored until box volume changes and mouth volume changes were in phase (2). The effective resistance (Raw-p) was calculated from the ao-Pb relationship using an algorithm as previously described (22). Changes in mouth pressure (Pm) were obtained after closing a shutter to occlude the airway while the infant made two or three respiratory efforts, breathing in most cases at a frequency of 30 to 40 breath/min. Vtg was measured from the angle of the Pb/Pm plot and corrected for instrument dead space. At least three baseline measurements of Raw-p and Vtg were made.

Ztr() measurements. Ztr() was measured using a prototype system developed by Collins Medical, Inc. (Braintree, MA). A head-out chamber surrounding the infant provided an enclosure by which pressure oscillations were applied to the chest wall. A pseudo random noise signal containing frequencies between 2 and 256 Hz in 2-Hz increments was generated by a computer and outputted via a digital-to-analog converter (model CIO-DAS16; ComputerBoards, Inc., Mansfield, MA). This signal was amplified and used to drive a loudspeaker mounted in the wall of the chamber. The pressure oscillations in the enclosure and applied to the body surface (Pbs) were detected by a pressure transducer (model SCXL05; Sensym, Milpitas, CA). The resulting flow oscillations at the airway opening (ao) were detected by measuring the pressure drop across a low dead space pneumotachometer (Jaeger) with a Sensym model SCXL04 pressure transducer. The pneumotachometer was placed in a low dead space face mask placed over the nose and mouth of the infant.

The Pbs(t) and ao(t) signals were amplified and band-passed filtered (2 to 256 Hz) (Collins Medical, Inc.), and digitized at a sampling rate of 1,024 Hz with an analog-to-digital converter (model CIO-DAS16; ComputerBoards, Inc.). The digitized signals in the time domain, Pbs(t) and ao(t), were converted to the frequency domain by discrete Fourier transforms. Ztr() was then computed using the technique of Michaelson and coworkers (25), from,

(1)

where G_PbsPbs is the power spectra of Pbs() and G_aoPbs is the cross-power spectra of Pbs() and ao(). The coherence function was computed from

(2)

where G_aoao is the power spectra of ao(). Overlapping of sequential blocks of time-domain data was not done.

Systems Identification Technique

The Ztr() data were analyzed using a lumped six-element model (Figure 1). Because only five of the six parameters in the DuBois model can be independently extracted from Ztr() data, gas compression compliance (Cg) was constrained to its value calculated from the plethysmographically determined Vtg. The other five parameters in this model were estimated by fitting the data and minimizing the weighted performance index, P.I.:

(3)

View larger version (6K): [in this window] [in a new window]   Figure 1.   Six-element model proposed by DuBois and coworkers (5) to analyze Ztr() data. Consists of airway branch with resistance (Raw) and inertance (Iaw), tissue branch with resistance (Rti), inertance (It), and compliance (Ct), separated by a gas compression compliance (Cg). Measurements made are Pbs, pressure at the body surface, and ao, flow at the airway opening.

where n = the number of data points, _i = frequency, Ztr,d(_i ) = Ztr data at frequency _i, and Ztr,m(_i, ) = model predicted Ztr at frequency _i using model parameter vector .

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ztr(), Raw-z, and Tissue Resistance (Rti)

Typical Ztr() spectra (mean ± standard deviation) in baseline and postalbuterol conditions are shown in Figure 2. Data below about 4 to 6 Hz and above approximately 200 Hz were noisy as evidenced by the large standard deviation and low coherence. The six-element model was able to fit only those data where the real and imaginary parts are concave downward, i.e., for f < approximately 140 Hz. We believe that for f > 140 Hz the distributed parameter properties of the airways become important and the system can no longer be modeled with only lumped parameter elements as in the DuBois model (8, 9). Thus, data only where the real and imaginary parts of Ztr() were concave downward and where the coherence was above 0.90 were used in the systems identification routine to estimate Raw-z and Rti-z (Table 2). On average, baseline Raw-z (80.6 ± 31.5 cm H₂O/L/s) (mean ± SD) was significantly (p < 0.01, paired t test) lower after albuterol inhalation (60.6 ± 22.2 cm H₂O/L/s). Baseline and postalbuterol Rti-z were not significantly different (2.3 ± 1.7 versus 2.7 ± 2.4 cm H₂O/L/s, respectively). In general, the response to albuterol inhalation was heterogeneous in that in two infants Raw-z had large decreases (greater than 40%), five had moderate decreases (between 20% and 40%), four had small decreases (between 5% and 20%), and four had little or no response (less than ± 5%), while one had a small increase. However, the five infants whose baseline Raw-z was the highest had the largest response to albuterol (Infants SM, KS, AJ, BV, and MM2). Raw-z represented on average 97 ± 3% of the total resistance of the respiratory system in baseline conditions and 96 ± 3% postalbuterol.

View larger version (25K): [in this window] [in a new window]   Figure 2.   Representative example of Ztr() (Infant BM); R plotted in upper panel is the real part of Ztr(), X in the middle panel is the imaginary part of Ztr(), and coherence is plotted in the lower panel. Filled circles are baseline, open circles are postalbuterol. In Ztr() plots continuous line is model fit to baseline data, and dashed line is the model fit to the postalbuterol data.

Vtg and Raw-p

There were no significant differences between baseline and postalbuterol Vtg (293 ± 75 and 285 ± 86 ml, respectively) or Raw-p (19.0 ± 7.0 and 22.5 ± 14.7 cm H₂O/L/s, respectively). Baseline Raw-p (21.3 ± 8.9 cm H₂O/L/s) was significantly (p < 0.001) smaller than baseline Raw-z (80.6 ± 31.5 cm H₂O/L/s). Raw-z and Raw-p were measured under baseline as well as postalbuterol conditions in 10 of the 16 infants (Table 2 and Figure 3) because six infants woke up before their postalbuterol Raw-p and Vtg could be measured. Even though there was no significant difference between baseline and postalbuterol Raw-p, there was a significant difference between baseline and postalbuterol Raw-z (p < 0.02, paired t test). There was no correlation between Raw-z and Raw-p in baseline and/or postalbuterol conditions.

View larger version (21K): [in this window] [in a new window]   Figure 3.   Baseline and postalbuterol airway resistances, Raw-z and Raw-p estimated from Ztr() data and from plethysmography, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There is a well recognized need for methods to assess lung function in infants to diagnose, quantify, and monitor airway diseases as well as to determine the efficacy of therapeutic modalities (26). Methods of measuring airway mechanics in infants have been basically of three types: (1) rapid chest compression techniques where expiratory flow during conditions of flow limitation is measured (27), (2) whole-body plethysmography that provides estimates of Raw-p and Vtg (1), and (3) techniques based on forced oscillations applied either to the airway opening (Zin[]) (13) or to the surface of the thorax (Ztr[]) (20, 21).

Maximal expiratory flows measured by rapid chest compression techniques are directly related to airway caliber and inversely related to airway wall compliance. As a result, measurements can be confounding because this technique does not allow the differentiation between changes in airway caliber and changes in airway wall compliance. For example, on the one hand after inhalation of a bronchodilator that increases airway caliber by relaxing airway smooth muscles maximal expiratory flows would increase if the increase in airway caliber has a greater impact than the increase in airway wall compliance. On the other hand, maximal expiratory flows would decrease if the increase in airway wall compliance has a greater impact than the increase in airway caliber. Although Vtg is measured relatively easily in infants by whole-body plethysmography, Raw estimates are very time-consuming (10 to 20 min) and technically difficult. The advantages of techniques using forced oscillations are that measurements can be made very quickly, they are technically less demanding, and there is the potential of extracting separate estimates of Raw and Rti, and tissue compliance.

The motivation to measure Ztr() instead of Zin() in this study was that we have recently shown that Zin() for frequencies between 2 and 128 Hz in infants is not sensitive to significant levels of bronchoconstriction (19). We then had to decide what frequency range to use, low frequencies (0.5 < f < 20 Hz) as was measured by Sly and coworkers (17), or higher frequencies as measured by others (18). Sly and coworkers argued that measurements should be made at low frequencies because Rti and compliance have a greater influence on impedance at these frequencies. This would be a logical choice based on studies in animals (30) indicating that bronchoconstriction is associated primarily with increased parenchymal (i.e., tissue) resistance and elastance. However, there is recent evidence that Rti in adult human asthmatics is not a dominant contributor to total pulmonary resistance (35). These results suggest that if one wants to monitor changes in airway caliber it would be better to measure impedance over a frequency range where it is influenced more by Raw than by Rti (i.e., f > 2 Hz). Also, there is a significant disadvantage in low-frequency impedance measurements because breathing must be suspended while measurements are being made. Sly and coworkers (17) were able to achieve 8- to 10-s periods of apnea in infants by provoking a Hering-Breuer reflex but this necessitated a complex system of valves and pressure sources. However, unlike low-frequency impedance measurements high-frequency impedance measurements can be made while the subject is spontaneously breathing. We thus reasoned that the most appropriate method of assessing lung function in infants was to measure Ztr() at frequencies above 2 Hz.

The results reported here indicate that Ztr() can be measured in infants to frequencies that exceed those necessary to provide statistically reliable estimates of Raw and Rti, and that Raw-z is sensitive to -agonist induced bronchodilation. If measurements are not made over a sufficiently wide range of frequencies, the model parameters are unreliable because they can vary over an unacceptably wide range without significantly affecting the quality of the model fit (36). Our results also indicate that in infants the majority of total respiratory system resistance over this frequency range is due to airway resistance, whereas tissue resistance contributes very little (i.e., < 3%). Moreover, the majority of the decrease in resistance resulting from albuterol inhalation occurred in the airways, not in the tissues.

Direct comparison of our Raw-z and Rti values extracted from Ztr() measurements can only be made with those reported by Marchal and coworkers (20). A significant problem with that study, however, is that their measurements were not made to high enough frequencies to obtain statistically reliable estimates. Nevertheless, these investigators used a computer model to predict the effect of upper airway wall shunt, which might be important in nasal breathing infants. They predicted that Raw-z would overestimate true Raw and that the amount of the overestimation increases in proportion to the increase in peripheral Raw. However, their simulations also predicted that the real part of Ztr() would increase with increasing frequency for f < approximately 14 Hz. We found no frequency-dependent increase in the real part of Ztr() even in the most severely obstructed infants (Figure 4). In fact, we found just the opposite, that is, a rapid frequency-dependent decrease with frequency for f < approximately 20 Hz.

View larger version (33K): [in this window] [in a new window]   Figure 4.   Example of Ztr() in severely obstructed infant (MM2) in baseline conditions (closed circles) and postalbuterol (open circles). Note significant f-dependent decrease in R for f < 28 Hz.

Comparison of our results with those of other studies reporting forced oscillatory measurements is complicated owing to differences in the models used to extract airway and tissue properties, differences in the frequency range of the measurements, as well as differences in the subjects studied. Three of the infants studied by Sly and coworkers who were similar to our subjects (i.e., a history of wheezy episodes but asymptomatic at the time of the measurements and a mean age of 51 wk) had airway resistances (9.5 ± 1.5 cm H₂O/L/s) much smaller than our infants even after bronchodilation (60.6 ± 22.2 cm H₂O/L/s). There is evidence that in infants, particularly with peripheral airway obstruction there is significant shunting of flow into upper airways, the face mask, and possibly the central airways when the forced oscillations are applied at the airway opening as in Zin measurements (19). This shunting of flow could account for airway resistance to be underestimated in Sly and coworkers' study. Comparison of our estimates of Rti to those reported by Sly and coworkers is not possible because they used a four-element model where the Raw and Rti were separated not by the topology of the system as in our model but instead by their frequency-dependent behavior. Thus, in their model all of the frequency-dependent behavior of the respiratory system was associated with Rti, whereas all of the frequency-independent behavior was associated with Raw. Consequently, their Rti is frequency-dependent, decreasing from 11 cm H₂O/L/s at 0.5 Hz to approximately 0.8 cm H₂O/L/s at 10 Hz, and there is no single value for Rti to which we can compare our values.

A troublesome result of the current study is that the Raw-z estimates were significantly different from, and not correlated with the plethysmographically determined airway resistance, Raw-p. There are several possible reasons for these differences. First, as argued by Marchal and coworkers (20) and discussed earlier Raw-z could overestimate airway resistance particularly with bronchoconstriction if there is significant upper airway wall shunting. We will discuss this further and provide evidence that we think supports the notion that airway wall shunting may influence Ztr data only with severe bronchoconstriction. A second possibility is that the position of the head and neck was different during the two different measurements. Because neck position is known to influence airway resistance (37) we took care to ensure that the relative position of the head and neck was identical during both the Ztr() and plethysmographic measurements. Finally, it is possible that the trachea may have been compressed by the seal around the subject's neck during the Ztr() measurements. However, in each subject we made sure that the sealing material was not compressing his neck.

One might argue, based on the comparison of baseline and postalbuterol Raw, that the Raw-z values are more realistic than the Raw-p values. First, because the infants studied had a history of wheezing one would expect some of the infants to be constricted in baseline conditions and have a reduction in Raw after albuterol inhalation. Indeed, Raw-z decreased after albuterol inhalation in nine of the 15 infants. However, only two infants had a moderate reduction in Raw-p after albuterol, AJ and HJ (decreases of 28% and 30%, respectively), whereas in three infants (BS, GD, and BV) it was significantly elevated (95%, 34%, and 138%, respectively).

Validity of model used to extract Raw-z and Rti. The model used to extract parameters from the Ztr data assumes a single-compartment lung and rigid airway walls. Although airway walls may not be rigid especially in infants (38, 39), the relative importance of nonrigid airway walls depends on their impedance (Zw[]) relative to the impedance of whatever they are in parallel with. When the forcing is done at the airway opening, as in Zin() measurements, Zw() is in parallel with that portion of the respiratory system that is chestward of the effective location of Zw(). It is generally thought that in healthy adults the impedance of the downstream portion of the respiratory system is small compared with Zw(). As a consequence, in healthy adults nonrigid airway wall behavior has a minimal influence on Zin(). However, if the peripheral airways become significantly obstructed, their impedance may increase to the extent that airway shunting does influence Zin() (40). Conversely, when the pressure oscillations are applied to the chest wall as in the Ztr() measurements Zw() is in parallel with the impedance of those airways mouthward of the effective location of Zw() plus the impedance of the pneumotachometer. These impedances are generally considered to be small relative to Zw() in healthy adult subjects as well as patients with obstructive airways disease (12). There is insufficient published data to speculate whether this is true in healthy or bronchoconstricted infants but our data suggest that this is the case.

What about the second model assumption, i.e., that the respiratory system behaves as a single compartment? It is generally assumed that the impedances of the separate, parallel airway-alveolar compartments are relatively uniform in healthy adults in which case their respiratory system behaves as a single-compartment system (5, 7). However, since airway obstruction is known to be heterogeneous it is possible that at some level of airway obstruction heterogeneity increases to the point where the single alveolar compartment model is no longer valid. Evidence that the single-compartment lung model assumption is not valid, or that there is significant airway wall shunting, is seen as a significant negative frequency dependence in the real part of Ztr() and Zin() at low frequencies extending to frequencies above approximately 2 Hz (25). Desager and coworkers (13) reported a large frequency-dependent drop in the real part of Zin() in wheezy infants, which implies either airway shunting or parallel inhomogeneities. However, Marchal and coworkers (20) did not see frequency dependence in the real part of Ztr() in either healthy infants or in infants with a history of airway obstruction. In the current study, there was a rapid decrease in the real part of Ztr() at very low frequencies in only the two infants (AJ and MM2) whose Raw-z was the largest in baseline conditions, and this frequency dependence disappeared in their postalbuterol Ztr() (Figure 4). Thus, only in cases of very severe bronchoconstriction do the model assumptions (i.e., that the respiratory system behaves like a single compartment, and/or that airway walls behave as rigid structures) begin to fail.

Previous studies in adults (7, 10) have demonstrated that Raw and Rti can be reliably estimated from Ztr(). This may not be the case in infants since airway wall shunting may have a greater influence because their airway walls are more compliant, the measurements must be made through a face mask, and the measurement includes the impedance of the nasal and nasopharyngeal cavities. We think that airway wall shunting in Ztr() is not a problem because there was no evidence of it in the impedance spectra of all but the two subjects who were most obstructed (i.e., had the highest Raw-z) in baseline conditions. An additional concern in infant Raw measurements is that their end-expiratory lung volume is actively determined, which influences the interrelationship between Vtg and airway resistance. Changes in end-expiratory lung volume after albuterol inhalation could influence this interrelationship in a complex way that could be reflected differently in the two different techniques used to estimate Raw. However, based on these results, we conclude that in future studies more detailed analysis of how Raw-z changes within the respiratory cycle and as a function of lung volume should be made. For technical reasons, this was not possible in the current study, which mainly focused on the feasibility of extracting Raw and Rti from Ztr() measurements.

Neither the plethysmographic nor impedance method measures Raw directly; instead, in both methods estimates of Raw are estimated using systems identification techniques. Airway resistance has been successfully estimated from Ztr measurements in adults (7, 10) and has been found to decrease after inhalation of a bronchodilator in asthmatic subjects (12). Furthermore, airway resistances derived from plethysmographic and transfer impedance measurements have been shown to be correlated in adults (12). However, in the present study we found that Raw-z and Raw-p were not correlated. We conclude from these findings that the Jaeger infant plethysmograph measures something different from the airway resistance estimated from Ztr. Additional studies are necessary to determine which of these parameters, Raw-p or Raw-z, more closely reflects changes in airway caliber in infants. Our study does, however, demonstrate that Ztr measurements in infants are technically far simpler to make than is Raw from plethysmography.

In conclusion, this study demonstrated that Ztr() can be measured in infants over a frequency range (4 to 140 Hz) that enables reliable estimates of Raw using systems identification techniques based on the DuBois model. Also, for the first time this study reports Raw and Rti in infants under baseline conditions as well as following inhalation of a bronchodilator. To more fully explore the clinical utility of Raw-z measurements in infants more detailed studies with larger numbers of subjects, and other diseases are necessary. Because Ztr() can be measured within seconds and measurements can be made during normal breathing the technique has the potential of assessing the interrelationship between lung volume and Raw during tidal breathing, which is particularly interesting in infants. Because of the ease and rapidity with which measurements can be made, this technique would be very useful as a noninvasive clinical tool in situations where multiple measurements are required such as studies to assess changes in airway caliber in response to cold air or methacholine challenge, as well as drug therapy.