INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The possibility that developmental differences in airway function might be risk factors for subsequent wheezing disorders in infancy has been systematically studied using, among other methods, expiratory flow induced by the rapid thoracic compression (RTC) technique (1). However, during flow limitation, the maximum expiratory flow is related both to airway cross-sectional area and to airway wall elastance (6), so that the RTC technique cannot be used to distinguish between changes in airway caliber and airway wall properties. Developmental differences in airway wall properties rather than airway caliber (2) could be at least partly responsible for the association between excessive flow limitation and subsequent wheezing disorders in infants.

The hypothesis that airway wall properties might explain some physiological phenomena in infants, was raised by investigators looking into paradoxical bronchoconstrictor responses in infants (7). In infants pretreated with albuterol (a ₂-receptor agonist) the maximum flow at FRC (V'max_FRC) increased at low concentrations of aerosolized methacholine but declined at higher concentrations. They speculated that at low concentrations, methacholine isometrically increased airway smooth muscle tone (and thereby airway wall elastance) and facilitated wave propagation in the airways. The implication of this hypothesis is that a technique to measure airway wall properties may enhance our understanding of airway stability in the developing lung and of often unpredictable responses to bronchoactive drugs.

We have developed new methods to measure high-frequency input impedance (Zin) in infants (8). High-frequency impedance and particularly the frequency of the first antiresonant frequency (f_ar,1) has been shown to be related to wave propagation phenomena in the airways of dogs (12), human adults (15, 16), and infants (10). f_ar,1 is therefore related to airway wall compliance. The aim of this study was to use the high-speed interrupter technique (HIT) to determine whether f_ar,1 was different in asymptomatic infants with a history of wheezing disorders in comparison with age-matched healthy controls. If true, not only airway narrowing but also alterations in airway wall compliance might be associated with wheezing disorders in infants. This has not previously been demonstrated in vivo in human infants.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Subjects

The study was performed in 23 infants and young children (aged 36- 81 wk) with a history of transient or persistent cough or wheeze, who had been referred from the outpatient clinic for lung function tests for clinical purposes (Table 1). Some of the data of six wheezy infants have been used in a previous study (11) of methacholine challenge. However, the measurement protocols of the two studies were matched and coordinated. Sixteen infants suffered from persistent, and 7 from transient, wheeze. These subgroups were defined on clinical grounds as follows: transient wheezers had wheezing episodes only after upper respiratory tract infections and had no symptoms between these episodes, whereas persistent wheezers had wheezing with or without virus-induced episodes. All infants with transient wheezing have had three or fewer wheezing episodes. Infants with other specific diseases and infants with upper respiratory tract infection within the previous 3 wk were not included in the study. Similarly, measurements of 19 healthy infants of similar age were performed. Twenty-five of the 42 infants and toddlers were sedated with a single oral dose of triclofos sodium up to 150 mg/kg. Other infants were studied during natural sleep. Lung function was measured during periods of regular quiet breathing in the supine position. The HIT measurements were approved by the Ethics Committees of the Royal Postgraduate Medical School, Hammersmith Hospital (London, UK) and of the Leicestershire Health Authority (Leicester, UK), where the measurements were performed. Written consent was obtained from parents.

Experimental Protocol

The principles and technical details of the HIT have been described previously (10, 11, 17). Briefly, high-frequency respiratory input impedance was measured with a propeller valve that rapidly occluded the airway opening several times within a period of 0.15 s at the beginning of inspiration without disturbing the infant. The resulting pressure and flow oscillations were measured by the wave tube technique (18). Using spectral analysis (19), respiratory input impedance was calculated from pressure and flow. Transcutaneous oxygen pressure (Ptc_O2) (TMC3; Radiometer, Copenhagen, Denmark) and transcutaneous oxygen saturation (Stc_O2) (Biox 3740; Omeda, Omaha, NE) were observed. The head position was standardized as described by Desager and co-workers (20), but we did not tape the mouth for safety reasons. A firm face mask that covered the nose and the mouth was applied with a putty ring (therapeutic putty; Carter's, Bridgend, UK). The dead space in the face mask was reduced by partially filling it with putty. For 19 of the subjects it was possible to measure reliably the residual face mask dead space by water displacement. The mean dead space volume (± SD) was 6.12 (± 1.7) ml. There was no difference in dead space volume between healthy and wheezy infants in this subgroup.

In all infants we performed 7-10 Zin measurements (1 set) during quiet regular tidal breathing. In 10 healthy infants we performed 2 sets of measurements on two different days within the same week to test repeatability. Impedance measurements with a coherence below 0.9 were not accepted (10). From the impedance spectrum, we extracted the frequency of the first antiresonance (f_ar,1), defined by a zero crossing in the imaginary part in the presence of a relative maximum in the real part of the impedance spectrum. We also determined the relative maximum [Zin_re(f_ar,1)] in the real part at f_ar,1. We did not describe other features of the Zin spectrum from 32 to 100 Hz, since we knew from previous work (9, 11), that Zin was not sensitive to changes in airway mechanics in this frequency range.

Data Analysis

In all infants we calculated the mean and 95% confidence interval (CI) of the sets of f_ar,1 and Zin_re(f_ar,1). For the repeatability test in 10 infants, we determined the mean and 95% confidence intervals (CIs) of the differences of the corresponding paired values of f_ar,1 and Zin_re(f_ar,1), respectively. We then compared the two measurement sets using a paired t test. The mean age, weight, height, f_ar,1, and Zin_re(f_ar,1) values for the infants in the wheezing group and the healthy control group were compared by t tests, and the group sex distribution was compared by ² tests. Since f_ar,1 is related to airway path length, we investigated in further detail whether f_ar,1 was confounded by the covariates age, height, and weight, by using multiple linear regression analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The healthy and the wheezy groups were not significantly different in age, weight, height, and sex. Since we were not allowed to sedate the healthy infants, the distribution of sedated to nonsedated infants was unbalanced in the two groups (Table 1).

We were able to detect antiresonant frequencies in all subjects. A representative example of the mean (SD) Zin spectrum for one subject is shown in Figure 1. Repeat values of f_ar,1 regressed to the group mean but there was no systematic difference between the two sets of repeated measurements in 10 healthy infants: first set, mean (95% CI) f_ar,1 = 220 (9.5) Hz; second set, f_ar,1 = 215 (27.6) Hz (paired t test: p = 0.68). However, the 95% CI of the differences of the two sets of corresponding observations of f_ar,1 was high (27.3 Hz), indicating marked day-to-day variability. Similarly, repeat values of Zin_re (f_ar,1) regressed to the group mean but there was no systematic difference between the two sets of repeated measurements in 10 healthy infants: first set, mean (95% CI) Zin_re(f_ar,1) = 89.6 (19.6) cm H₂O · L¹ · s; second set, Zin_re(f_ar,1) = 91.4 cm H₂O · L¹ · s (12.1) (paired t test: p = 0.90). However, the 95% CI of the differences of the two sets of corresponding observations of Zin_re(f_ar,1) was high (32.1 cm H₂O · L¹ · s), indicating marked day-to-day variability.

View larger version (19K): [in this window] [in a new window]   Figure 1.   Example of a high-frequency impedance spectrum from a single infant (mean and SD of 10 measurements). The antiresonant frequency (far,1) is defined by the relative maximum in the real part [Zinre(far,1)] in the presence of a zero crossing in the imaginary part.

The mean (95% CI) first antiresonance in wheezy infants occurred at a lower frequency (175.7 ± 14.1 Hz, p < 0.005) than in the healthy control group (212.1 ± 21.1 Hz) (Table 1, Figure 2). When divided into subgroups of wheezers, the values of f_ar,1 for 16 persistent wheezers (179.6 ± 18.0 Hz) and 7 transient wheezers (166.6 ± 28.1 Hz) were significantly different from the healthy infants (p < 0.05 and p < 0.02, respectively), but not from each other.

View larger version (32K): [in this window] [in a new window]   Figure 2.   Group mean and 95% confidence intervals (CIs) of the averaged Zin spectra for infants with wheezing disorders (circles) and healthy infants (triangles). The frequency (far,1) was significantly lower in wheezing infants than in healthy infants (p < 0.005). Similarly, the relative maximum in the real part of Zin at far,1 [Zinre(far,1)] was significantly lower in the wheezy group in comparison with the healthy group (p < 0.0005).

Since f_ar,1 is related to airway path length, we investigated possible confounding biometric factors. We found that f_ar,1 was not significantly related to age (0.45 × age + 215 Hz, p = 0.082) and length (1.14 × length + 273.2 Hz, p = 0.0624) but to weight (7.38 × weight + 262.7 Hz, p = 0.011). Assuming normal distribution, the linear regression coefficient (healthy versus wheezy) without any covariates was 36.4 (standard error [SE] 11.8, t = 3.1, p = 0.0036). Including height as a single covariate the regression coefficient decreased to 26.6 (SE 10.8, t = 2.4, p = 0.019). Including weight as a single covariate the regression coefficient decreased to 29.7 (SE 11.8, t = 2.5, p = 0.016). Including age (PNA) as a single covariate the regression coefficient decreased to 33.2 (SE 11.9, t = 2.8, p = 0.008). Including weight, height, and age as covariates the regression coefficient decreased to 24.7 (SE 11.3, t = 2.18, p = 0.036). Linear regression analysis indicated that weight and height were relevant confounding covariates.

Also, the mean (95% CI) relative maximum in the real part at the first antiresonant frequency [Zin_re(f_ar,1)] was significantly higher in the healthy group (96.4 ± 13.1 cm H₂O · L¹ · s) than in the wheezy group (67.0 ± 10.5 cm H₂O · L¹ · s) (p < 0.0005). Even if the wheezy infants were separated into subgroups, Zin_re(f_ar,1) was significantly higher in the healthy group in comparison with the persistently wheezing group (66.2 ± 12.0 cm H₂O · L¹ · s) (p < 0.001) and transiently wheezing group (68.8 ± 27.3 cm H₂O · L¹ · s) (p < 0.05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Developmental differences in airway function are associated with subsequent development of wheezing disorders in infancy (1). Lower values of V'maxFRC measured by the rapid thoracic compression technique in early infancy have been interpreted as due to smaller airway caliber (2). But during flow limitation, no analysis of maximal expiratory flow-volume curves can distinguish between the effects of airway caliber and airway wall compliance on flow limitation. It might be especially important to distinguish these effects in infants since, as has been demonstrated in the lamb by Panitch and coworkers (21), that airway wall compliance is much higher than in later life and may lead to greater airway collapsibility. To understand the developmental basis for wheezing disorders, it is thus important to devise a technique featuring parameters that are related to airway wall compliance. We have developed such a technique (10, 11), based on the measurement in infants of high, frequency respiratory impedance (Zin), using the high-speed interrupter technique (HIT).

High-frequency respiratory impedance is influenced by the properties of airway walls as shown in dogs (12), human adults (15, 16), and infants (8). In human adults and infants a particular feature of the high-frequency Zin, the antiresonance, is related not to tissue properties of the lung but to wave propagation phenomena in the airways (10, 15). This implies that the frequency at which the first antiresonance occurs (f_ar,1) is a function of the factors determining wave propagation phenomena and the boundary conditions of the airways. The latter means that f_ar,1 depends on whether the airways behave like an open or a closed system at the distal end. It is not entirely clear whether the infant airways behave like an open tube system. In previous work we found that the second antiresonant frequency was three times higher than f_ar,1 in infants (10). In a rigid tube model this would occur if the tube were open at both ends; however, in a branching compliant tubular system this might not necessarily be the case. Wave propagation velocity in a compliant tube is related to gas density, airway path length, and airway wall compliance and to a minor extent to airway diameter in small tubes. In large-diameter rigid tubes waves will propagate at the free-field speed of sound, independent of tube diameter. In rigid tubes with diameters < 0.4 cm, wave propagation velocities with room air are decreased by > 5% of the free-field speed of sound (22). In the terminal airways of human adults, where the diameter is 0.08 cm, the wave propagation velocity is 62% of the free-field speed of sound (15). Thus in peripheral airways, wave propagation velocities are significantly reduced. A reduction in the wave propagation velocities in these distal airways would cause them to resonate at a lower frequency. Thus, theoretically airway diameter has a certain influence on wave propagation and therefore f_ar,1 and differences in baseline lung function could imply differences in airway diameter between healthy infants and infants with wheezing disorders.

One might be tempted to explain a decrease in f_ar,1 in the wheezy group by baseline differences in airway diameter. However, previous physiological measurements (11) disprove this hypothesis. These measurements during methacholine challenge suggest that in infants the influence of airway wall compliance on f_ar,1 is much more important than the influence of changing diameter. We have shown that f_ar,1 increases after methacholine challenge (11). If the effect of airway diameter on f_ar,1 were important, f_ar,1 would decrease and not increase as seen during the methacholine challenge. Change in airway wall compliance can theoretically be explained by increasing airway smooth muscle tone during the stepwise challenge procedure.

We aimed to determine whether f_ar,1, and hence airway wall compliance, was different in asymptomatic infants with a known history of wheezing and in healthy children. We found that f_ar,1 was significantly lower in wheezy infants than in age-matched healthy controls. f_ar,1 was not different between persistent and transient wheezers. These data suggest that infants with wheezing disorders have differences in airway wall compliance, even when they are asymptomatic at the time of measurement. It is unlikely that differences in f_ar,1 could be explained by changes in airway path length since age, weight, and height were not significantly different between groups. This was supported by linear regression analysis showing that weight and height were confounding covariates; but f_ar,1 was still significantly different between the healthy and wheezy groups after accounting for these covariates.

From the current data it cannot be concluded whether the lower values of f_ar,1 correspond to lower or higher airway wall compliance, since the interaction between f_ar,1 and wave propagation is highly complex (23). We can conclude only that it is different. We hypothesize that these differences might be acquired through postviral inflammatory changes in wall mechanics or, alternatively, may be due to developmental differences in wall mechanics. If they are acquired we can conclude that they persist even in the absence of obvious wheezing symptoms.

Our findings contrast with those of Chalker and co-workers in adults (24), who found f_ar,1 to be higher in chronic obstructive pulmonary disease (COPD). They also hypothesized that airway wall compliance might be different in this group of patients. We can think of two possible explanations for the increase in f_ar,1 in COPD, in contrast to a decrease in infants with wheezing disorders. First, the underlying pathology of COPD might be different from that in wheezy infants. An alternative explanation concerns the complex relationship between antiresonant frequency, wave propagation velocity, and the resonance frequency of the airway wall (23). Depending on whether f_ar,1 is higher or lower than the wall resonant frequency, alterations in wall compliance could lead to an increase or decrease in f_ar,1 (23). In adults this relationship might differ from that in infants and might also explain why methacholine challenge resulted in an increase in f_ar,1 in individual subjects (9, 11) whereas as a group, wheezy infants have lower f_ar,1 than healthy controls at baseline.

Other features of the high-frequency Zin spectrum are more complex. From previous work (9, 11) we know that Zin from 32 to 100 Hz in infants is not sensitive to changes in airway mechanics and therefore we omitted this frequency range. As mentioned above, at higher frequencies > 100 Hz, Zin is influenced by airway resistance, less influenced by tissue properties, and more influenced by airway wall properties. Airway wall compliance significantly influences Zin at frequencies surrounding the f_ar,1. We found that in wheezy infants the relative maximum in the real part of Zin at f_ar,1 [Zin_re(f_ar,1)] was significantly lower in the wheezy group than in healthy controls. At the moment this must be considered as an empirical finding, since detailed analysis of Zin_re(f_ar,1) would require system identification techniques using highly sophisticated distributed parameter models (13, 14, 16) based on detailed anatomical models of the infant bronchial tree, which are currently not available.

Limits of the Method

In previous work we have reported that short-term variability of f_ar,1 (coefficient of variation of 10 measurements) was between 5 and 10% in sedated wheezy infants (10). The current study showed that f_ar,1 varies more widely than this from day to day in healthy infants. Variation was random resulting in similar means (95% CI) of the first and second set of measurements (220 ± 9.5 Hz versus 215 ± 27.6 Hz, respectively). However, f_ar,1 in the wheezy group was still significantly lower. These data indicate that the technique may not be especially useful for longitudinal measurements in individual subjects, for instance for clinical purposes, but in this particular cross-sectional study it allows us to conclude that f_ar,1 in infants with wheezing disorders is different from that of their healthy controls.

As we have previously reported (11) the flow through infant airways is a function of both airway diameter and airway wall compliance. Changes in airway wall compliance might even counteract the effect of bronchoconstriction on airflow under certain circumstances, for example during methacholine challenge (11). We hypothesize that f_ar,1 shows large day-to-day variability because airway wall compliance is part of a complex regulatory system maintaining balanced flow through the airways. Other parts of this regulatory system may be lung volume, elastic recoil, and airway diameter.

Since previous work has shown that the characteristics of the face mask can influence f_ar,1 (8), great care was taken to minimize its volume and standardize the application of the face mask. This could be especially crucial in small newborn infants with a short airway path length, since theoretically the face mask dead space will have a significant effect on f_ar,1 by increasing parallel impedance (8). The face mask volume was small (mean, 6.1 ml). Even more important was the standardization of the procedures, which reduced the variation in the face mask dead space between subjects (group SD = 1.9 ml) and which thus avoided systematic bias between healthy and wheezy infants.

While both groups were not significantly different in age, weight, height, and sex, more measurements were made during natural sleep than during sedated sleep in the healthy group. It is therefore theoretically possible that sedation might have had an effect on airway wall mechanics and f_ar,1. Sedated or not, we performed measurements only during behavioral quiet sleep, and this should have reduced some of the potential bias. Sleep stage as well as sedation can influence upper and lower airway mechanics. Furthermore, upper and lower airway mechanics are not independent of each other. While these complex relationships have been described in adults (e.g., reference 25), similar studies in infants are outstanding. To separate the influence of sleep stage, sedation, and the upper and lower airway interaction in high-frequency impedance data several basic studies had to be done, a description of which exceeds the scope of this article, but that will be important topics of future studies. Nevertheless, it is also important to note that f_ar,1 is independent of wide changes in upper airway resistance (10).

Summary and Hypothesis for Future Research

Although Zin measurements give only indirect information on airway wall compliance, they can be assessed non-invasively in vivo, which is a major advantage since airway mechanics is crucially influenced by such factors as the elastic recoil of the surrounding tissue which exerts its effect only on airways in situ. This is therefore the first evidence that airway wall compliance in vivo is different in infants suffering from wheezing disorders, even when they are asymptomatic. Moreover, it allows the interpretation of forced expiratory flow volume curves by the RTC technique, to be more fully interpreted in infants.

Our findings have important implications for the study of airway structure-function relationships in wheezing disorders in infants. In future, studies of both airway caliber airway wall structure and mechanics will be needed. The current study design does not allow to distinguish whether these changes in airway wall compliance are acquired (e.g. by inflammatory remodeling) or whether they pre-existed the onset of wheeze. Prospective studies, commencing shortly after birth, before inflammatory injury occurs, will help to distinguish between (fetal) developmental determinants of airway function and postnatal remodeling.