Short-term variability of airway caliber--a marker of asthma?

doi: 10.1152/japplphysiol.00420.2006103:296-304, 2007. First published 3 May 2007;J Appl Physiol

Gregory G. KingChantale Diba, Cheryl M. Salome, Helen K. Reddel, C. William Thorpe, Brett Toelle andasthma?

a marker of−−Short-term variability of airway caliber

You might find this additional info useful...

20 articles, 11 of which you can access for free at: This article citeshttp://jap.physiology.org/content/103/1/296.full#ref-list-1

8 other HighWire-hosted articles: This article has been cited by http://jap.physiology.org/content/103/1/296#cited-by

including high resolution figures, can be found at: Updated information and serviceshttp://jap.physiology.org/content/103/1/296.full

can be found at: Journal of Applied Physiology about Additional material and informationhttp://www.the-aps.org/publications/jappl

This information is current as of June 6, 2013.

http://www.the-aps.org/. © 2007 the American Physiological Society. ISSN: 8750-7587, ESSN: 1522-1601. Visit our website at year (monthly) by the American Physiological Society, 9650 Rockville Pike, Bethesda MD 20814-3991. Copyrightphysiology, especially those papers emphasizing adaptive and integrative mechanisms. It is published 12 times a

publishes original papers that deal with diverse area of research in appliedJournal of Applied Physiology

by guest on June 6, 2013http://jap.physiology.org/

Dow

nloaded from

http://jap.physiology.org/content/103/1/296#cited-by

http://jap.physiology.org/

Short-term variability of airway caliber—a marker of asthma?

Chantale Diba,1,2 Cheryl M. Salome,1,2 Helen K. Reddel,1,2

C. William Thorpe,1,4 Brett Toelle,1,2 and Gregory G. King1,2,3,4

1The Woolcock Institute of Medical Research, Sydney; 2The University of Sydney, Sydney;3Department of Respiratory Medicine, Royal North Shore Hospital, Sydney; and4Cooperative Research Centre for Asthma, Sydney, Australia

Submitted 10 April 2006; accepted in final form 17 April 2007

Diba C, Salome CM, Reddel HK, Thorpe CW, Toelle B, KingGG. Short-term variability of airway caliber—a marker of asthma?J Appl Physiol 103: 296–304, 2007. First published May 3, 2007;doi:10.1152/japplphysiol.00420.2006.—Variability in airway caliberis a characteristic feature of asthma. Previous studies reported that thevariability in respiratory system impedance (Zrs), measured by theforced oscillation technique during several minutes of tidal breathing,is increased in asthma and may be a marker of inherent instability ofthe airways. The aims of this study were to determine if short-termvariability in impedance correlates with peak expiratory flow (PEF)variability or airway hyperresponsiveness (AHR). The SD of log-transformed impedance (lnZrsSD) was measured as a marker ofshort-term variability and compared with the diurnal variability ofPEF over 2 wk in 28 asthmatic and 7 nonasthmatic subjects and withAHR to histamine in a cohort of 17 asthmatic and 82 nonasthmaticsubjects. In addition, lnZrsSD was measured in eight nonasthmaticsubjects before and after methacholine challenge in the upright andsupine positions. There were no significant differences in lnZrsSDbetween asthmatic and nonasthmatic subjects (P � 0.68). Further-more, in asthmatic subjects, lnZrsSD did not correlate with diurnalvariability of PEF (rs � �0.12 P � 0.54) or with AHR to histamine(r � 0.10, P � 0.71). Neither methacholine challenge nor supineposture caused any significant change in lnZrsSD. We conclude thatour findings do not support previous reports about the utility ofshort-term variability of impedance. Our findings suggest that, usingstandard methods for forced oscillometry, impedance variability doesnot provide clinically useful information about the severity of asthma.

asthma severity; impedance

ASTHMA IS CHARACTERIZED by airways that vary markedly incaliber both spontaneously and during pharmacological andenvironmental stimulation. Indeed, an assessment of the capac-ity of the airways for rapid change in caliber, either as aresponse to bronchodilator drugs or bronchial challenge tests oras the diurnal variation in peak expiratory flow (PEF), is animportant tool in the diagnosis and monitoring of asthma (3–5).Airflow variability, measured by recording twice daily PEFover a 2- to 4-wk period, is strongly correlated with othermarkers of asthma severity such as airway hyperresponsive-ness (AHR) and, to a lesser extent, with asthma symptoms(12). However, peak flow monitoring is time consuming andburdensome, and a simple, rapid test to measure the variabilityof airway caliber would be useful in the diagnosis and man-agement of asthma.

It has been suggested that the intrinsic variability of airwaycaliber can be measured during tidal breathing over a period ofminutes (11). Respiratory system impedance (Zrs) measured

using the forced oscillation technique (FOT) provides a virtu-ally continuous measure of airway caliber throughout thebreathing cycle. Que et al. (11) found that the variability in Zrsduring this continuous recording was greater in asthmatic thanin nonasthmatic subjects. They suggested that this variabilityof Zrs represents the intrinsic, spontaneous variability in air-way caliber and configuration and that it might predict longer-term variability in airway caliber and thus be a useful predictorof severe or life-threatening asthma (11).

In this study, we set out to extend the findings of Que et al.(11) to determine if the short-term variability in impedancecould provide clinically useful information about the severityof asthma. To do this, we first evaluated several technicalvariables that may have affected the measurement of imped-ance variability. We then compared impedance variability inasthmatic and nonasthmatic subjects and measured the rela-tionship between impedance variability, diurnal peak flowvariability, and AHR in adult subjects. In addition, we mea-sured the effects of posture and methacholine challenge onimpedance variability in nonasthmatic subjects.

METHODS

Subjects

Asthmatic and nonasthmatic subjects aged between 17 and 75 yrwere recruited from staff and students at the University of Sydney andfrom the Asthma Centre at Royal Prince Alfred Hospital, Sydney,Australia. Asthmatic subjects had a previous diagnosis of asthma froma respiratory physician and either had AHR or had experiencedasthma symptoms requiring treatment in the past 12 mo. Nonasth-matic subjects had no history of respiratory disease, symptoms, orasthma medication use.

Data were also obtained from asthmatic and nonasthmatic subjectswho participated in a follow-up study of a community-based cohort,known as the “Belmont cohort,” which has been described previously(8). This cohort was established in 1982 from a random selection of8- to 10-yr-old children attending primary schools in Belmont, NewSouth Wales. Respiratory symptom questionnaires and histaminechallenge tests (21) were performed at recruitment and then every 2 yruntil age 18–20, then at age 23–25. The current assessment wasundertaken when the participants were aged between 28 and 30 yr. Ofthe 322 subjects who participated in this follow-up, we had completedata for impedance variability and AHR from 243 participants. In thiscohort, subjects were defined as asthmatic if they had experiencedwheeze in the last 12 mo and had current AHR (17). Nonasthmaticsubjects were defined using data both from the present study and fromhistorical data obtained over the 20-yr follow-up of the cohort. Thesenonasthmatic subjects had no previous history of wheeze, respiratory

Address for reprint requests and other correspondence: C. Diba, WoolcockInstitute of Medical Research, P.O. Box M77, Missenden Road, Camperdown,New South Wales, Australia, 2005 (e-mail: [email protected]).

The costs of publication of this article were defrayed in part by the paymentof page charges. The article must therefore be hereby marked “advertisement”in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

J Appl Physiol 103: 296–304, 2007.First published May 3, 2007; doi:10.1152/japplphysiol.00420.2006.

8750-7587/07 $8.00 Copyright © 2007 the American Physiological Society http://www. jap.org296


Dow

nloaded from


disease, or use of asthma medications, and no history of having AHRover the preceding 20 yr.

Current smokers or subjects who had greater than 10 pack-yearssmoking history were excluded from all analyses. These studies wereapproved by the Human Ethics Committee of the University ofSydney, and written informed consent was obtained from all subjects.

Study Design

Effect of technical variables on Zrs variability. Respiratory systemimpedance was measured both in the present study and that of Queet al. (11) using noncommercial devices that were built in-house.Although all of our devices conformed to international guidelines forFOT devices (6, 19), the FOT devices used in our studies differedslightly from that described by Que et al. (11). Consequently, we firstevaluated several technical variables that had the potential to affectmeasurements of impedance variability. Specifically, we comparedimpedance variability measured by a device that included an inertancetube, as described by Que et al. (11), and by our device that had a flowsplitter plus a resistance mesh. In addition, we examined the effect onimpedance variability of using pneumotachographs with differentdead spaces (Fleisch, dead space 130 ml; and Hans Rudolph, deadspace 60 ml) and of using oscillatory signals of high (�2.5 cmH2O)and low (�0.8 cmH2O) amplitude. Since many subjects found itdifficult to breathe on a mouthpiece continuously for 15 min, weexamined the validity of reducing the duration of the measurement bycomparing impedance variability measured during the first minute ofrecording with that measured during the whole 15-min period. Thesestudies required multiple experiments involving, in total, 17 nonasth-matic and 15 asthmatic subjects. Not all subjects participated in allexperiments.

Clinical implications of short-term Zrs variability. The clinicalimplications of short-term variability of impedance were assessed inthree populations. In the Belmont cohort, we compared short-termvariability of impedance in 82 nonasthmatic and 17 asthmatic adults,and we examined the relationship between short-term variability ofimpedance and AHR to histamine. Second, in a group of 28 asthmaticsubjects and seven nonasthmatic subjects, we examined the relation-ship between short-term variability of impedance and diurnal vari-ability of PEF measured over 2 wk. Finally, in eight nonasthmaticsubjects, we examined the effect of airway smooth muscle activationand changes in posture on short-term variability of impedance. In thisstudy, Zrs was measured during methacholine challenge on twoseparate days, in upright and supine positions.

In all studies subjects withheld short-acting �2-agonist for 6 h andlong-acting �2-agonists for 24 h before testing. Asthma symptoms,medication use, and smoking history were obtained by self-completedquestionnaire. Lung function was measured by spirometry accordingto American Thoracic Society criteria (1). Impedance was measuredby the FOT as described below and was always measured beforespirometry.

FOT and Data Processing

Zrs was measured continuously using the FOT during tidal breath-ing, from which the mean and variability of the absolute values of Zrswere calculated. The FOT device has been described previously (15,16). Briefly, a 6-Hz oscillation, generated by a loudspeaker, wasapplied to the airway opening via a three-way flow splitter thatallowed the subject to breathe normally from room air. The threeopenings of the splitter were attached to the distal opening of thepneumotachograph, to atmosphere, and to the loudspeaker. Flow wasmeasured using either a Fleisch (50 mm diameter, Phipps and Bird) orHans Rudolph (model 4830, flow range 0–400 l/min, Hans Rudolph,VacuMed, Ventura, CA) pneumotachograph. Differential pressurewas measured using a solid-state pressure transducer with range �2.5cmH2O (Sursense DCAL4, Honeywell). Pressure at the airway open-ing was measured using a similar transducer that had a higher range

(�12.5 cmH2O). The pressure and flow signals were low-pass filteredat 25 Hz and sampled at 300 Hz with a 16-bit analog-to-digitalconverter. Signals were then filtered using a bandwidth of 3 Hzcentered around the oscillation frequency 6 Hz to reduce potentialleakage from nonharmonic frequency components in the recordedsignal. The filtered signals were then divided into segments exactlyequal in duration to the period of the oscillation signal (1/6 s) andoverlapping so that each segment started 0.05 s from the previous one(i.e., 20 segments/s). The impedance and its real (resistance) andimaginary (reactance) parts were derived from this sequence ofsegments by Fourier analysis. This entailed estimation of the cross-spectrum and power spectra from three adjacent segments of thepressure and flow signals. Zrs, calculated at intervals of 0.1 s (i.e., 10estimates/s), was then defined as the 6-Hz component of the ratiobetween the pressure power spectrum and the pressure-flow cross-spectrum estimates (16).

Since the period of the oscillation component is exactly equal to thelength of the Fourier transform, tapered windows are not required (ordesirable), but reducing nonharmonic components (in particular thelarge low-frequency breathing component) is essential to avoid leak-age of their power into the desired frequency component. Filters (typeII Chebychev) were applied in both forward and time-reversed direc-tions to minimize phase distortions of the signals.

Erroneous and extreme Zrs values, which could have occurred ifthe glottis closed or the seal around the mouthpiece was lost duringtesting, were identified and excluded using a procedure identical tothat described by Que et al. (11). Each measurement of impedancewas plotted against flow, and large outlying data points at zero flow,as well as all negative respiratory system resistance (Rrs) values, weremanually excluded from analysis. Resistance was reported as percent-age of the predicted values of Pasker et al. (7).

To examine the effect of the inertance tube on short-term variabil-ity, we attached a 3.8-m-long tube with a 35-mm diameter onto theexpiratory port of the FOT. Bias flow was applied at a constant rate of12 l/min.

To examine the effect of different oscillatory signal pressures, theflow head was occluded, and the peak pressures were adjusted to �2.5cmH2O for high-amplitude and �0.8 cmH2O for low-amplitudemeasurements.

Diurnal Variability

PEFs and forced expiratory volume in 1 s (FEV1) were measuredtwice a day at home for 2 wk using a hand-held turbine-styleelectronic spirometer (MicroMedical DiaryCard, Rochester, Kent,UK). Three spirometric maneuvers were recorded immediately onwaking and before sleep, and the best PEF and FEV1 of each set ofthree were selected. The diary displayed only PEF in liters per minutebut also stored date, time, flow volume curves, FEV1, and forced vitalcapacity (FVC) for later review.

Individual PEF and FEV1 data points were examined and, forvalues that were 1.5 SDs outside the individual patient mean, thestored flow-volume loops were reviewed. Data from artefactual ma-neuvers, and from maneuvers where FEV1 equaled FVC, were ex-cluded from analysis (13).

Variability in airway caliber was calculated as within-day variabil-ity (diurnal variability) for both PEF and FEV1. Diurnal variabilitywas calculated for each day as amplitude percent maximum, i.e., theabsolute difference between the morning and evening value, ex-pressed as a percentage of the highest value for the day. The overalldiurnal variability was the mean of the 2-wk period (14). Data setswith �50% of days with both morning and evening measurementswere excluded from the final analysis.

Airway Responsiveness

In the Belmont cohort, airway responsiveness was measured usinghistamine diphosphate, administered via a DeVilbiss no. 45 hand-held

297SHORT-TERM VARIABILITY OF AIRWAY CALIBER

J Appl Physiol • VOL 103 • JULY 2007 • www.jap.org


Dow

nloaded from


nebulizer (DeVilbiss HealthCare, Somerset, PA) in doses rangingfrom 0.03 to 3.9 �mol histamine. Subjects with baseline FEV1 below70% predicted underwent a bronchodilator reversibility test with 200�g salbutamol. AHR was defined as a provocative dose of �3.9 �molhistamine producing a 20% fall in FEV1, or bronchodilator response�15% increase in FEV1.

Dose-response slope was calculated as percent fall in FEV1 permicromole histamine (5, 8) to provide a continuous variable forairway responsiveness for use in correlation analyses.

In nonasthmatic subjects, the effect on impedance variability ofsmooth muscle activation and posture was measured during metha-choline challenge in the upright and supine position. In this study,methacholine was administered via a DeVilbiss no. 646 nebulizerusing a nebulization dosimeter (Rosenthal French, Baltimore, MD)attached to oxygen at 138 kPa in doubling doses, ranging from 0.15 to199 �mol.

Data Analysis

All data were analyzed using Analyse-It for Excel (Analyse-ItSoftware, Leeds, UK). Individual frequency distribution curves of Zrsand lnZrs were constructed using bin sizes of 0.2 cmH2O � l�1 �s, withfrequency normalized by expressing it as a percentage of the totalnumber of measurements. The Kolmogorov-Smirnov statistic wascalculated for individual subjects using Zrs and the natural logarithmof Zrs and is reported as the mean of the individual values. Compar-isons of the Kolmogorov-Smirnov values were made using the Wil-coxon signed ranks test. The SD of lnZrs (lnZrsSD) was used as theprimary measure of the variability of Zrs (9). Results are expressed asmeans � SD unless otherwise stated. Correlations were examinedusing either Pearson’s (r) or Spearman’s (rs) correlation coefficient.Comparisons were made using paired or unpaired t-tests with asignificance level of 0.05.

RESULTS

Frequency Distribution of Impedance

Figure 1 shows typical traces of Zrs measured over 15 minin an asthmatic and a nonasthmatic subject. The distribution ofboth Zrs and lnZrs, in asthmatic and nonasthmatic subjects,deviated significantly from a Gaussian distribution (Table 1).However, lnZrs more closely approximated a normal distributionthan did Zrs as indicated by lower Kolmogorov-Smirnov valuesfor lnZrs obtained using both the high-amplitude oscillations(difference in mean Kolmogorov-Smirnov statistic, P � 0.01) andthe low-amplitude oscillations (P � 0.001). Figure 2 showsindividual normalized frequency distributions of lnZrs for bothasthmatic and nonasthmatic subjects. There were no significantdifferences between asthmatic and nonasthmatic subjects in thesmall proportion of data points excluded because of error orartefact (P � 0.69, high amplitude; P � 0.17, low amplitude).

Effect of Inertance Tube vs. Mesh

The major technical difference between the FOT device usedby Que et al. (11) and our FOT device was the inclusion of aninertance tube rather than the flow splitter and resistance meshused in our device. We compared lnZrsSD collected over 15min in five asthmatic and six nonasthmatic subjects using bothmethods. There was no significant difference in lnZrsSD be-tween asthmatic and nonasthmatic subjects using either theinertance tube (0.27 � 0.08 vs. 0.33 � 0.04, P � 0.19) or theresistance mesh (0.33 � 0.06 vs. 0.30 � 0.04, P � 0.32). Inthese subjects, compared with nonasthmatic subjects, asthmaticsubjects had lower FEV1 (86.6 � 19.9 vs. 102.8 � 11.3%predicted, P � 0.12) and higher Rrs (115.7 � 54.0 vs. 75.1 �

Fig. 1. Typical traces of respiratory system imped-ance (Zrs) over 15 min in a nonasthmatic subject (top)and an asthmatic subject (bottom). Measurementswere made at high-amplitude oscillations. The meanlnZrs was 1.10 and 0.96 for the nonasthmatic andasthmatic subject, respectively, while the standarddeviation of log-transformed Zrs (lnZrsSD) was 0.34and 0.29, respectively.

298 SHORT-TERM VARIABILITY OF AIRWAY CALIBER



Dow

nloaded from


21.7% predicted, P � 0.15), but the differences were notsignificant.

Effect of Oscillatory Amplitude andPneumotachograph Device

In asthmatic subjects there were no significant differencesbetween low- and high-amplitude oscillations in lnZrsSD,Rrs, or Zrs (Table 2). However, in nonasthmatic subjects,high-amplitude oscillations decreased lnZrsSD but had noeffect on Rrs or Zrs. There were no significant differences inlnZrsSD between the Fleisch and Hans Rudolph devices ineither asthmatic (P � 0.74) or nonasthmatic (P � 0.79)subjects at high amplitude. There were no significant dif-ferences in lnZrsSD between asthmatic and nonasthmaticsubjects at either high (P � 0.33, Fleisch; P � 0.21, HansRudolph) or low (P � 0.54, Fleisch; P � 0.61, HansRudolph) amplitude. In this group of subjects, comparedwith nonasthmatic subjects, asthmatic subjects had lowerFEV1 (87.3 � 17.1 vs. 104.7 � 14.2% predicted, P � 0.04),and higher Rrs (84.8 � 31.2% vs. 121.38 � 43.7% pre-dicted, P � 0.06).

Duration of Impedance Measurement

To determine if the duration of the impedance measurementscould be reduced, we compared the distribution and variabilityof lnZrs calculated from the full 15-min recording with datacalculated from only the first minute of recording, usingcombined data from asthmatic and nonasthmatic subjects. ThelnZrs data were not normally distributed over 1 min; however,the distribution was closer to normal than the distribution ofmeasurements recorded over 15 min (Table 1). There were nosignificant differences in lnZrsSD between the 1- and 15-min recordings in either asthmatic or nonasthmatic subjectsTable 1).

The variability of lnZrs, measured as lnZrsSD, did not differsignificantly between asthmatic and nonasthmatic subjects(P � 0.68, high amplitude; P � 0.32, low amplitude). Fur-thermore, there was no significant difference in the variabilityof resistance, measured by lnRrsSD, between asthmatic andnonasthmatic subjects (P � 0.99, high amplitude; P � 0.26,low amplitude). In these asthmatic subjects, compared with

nonasthmatic subjects, FEV1 was significantly lower (85.0 �14.6 vs. 103.2 � 14.3% predicted, P � 0.004) and Rrs wassignificantly greater (124.6 � 39.0 vs. 89.8 � 25.5% predicted,P � 0.01).

The variability of impedance, before natural log transforma-tion, measured by ZrsSD, was strongly dependent on airwaycaliber (Fig. 3). In measurements over 15 min and those over1 min, there was a highly significant correlation between Zrsand ZrsSD in both asthmatic (r � 0.73, P � 0.005, over 15min; r � 0.91, P � 0.0001, over 1 min) and nonasthmatic (r �0.90, P � 0.0001, 15 min; r � 0.81, P � 0.0001, 1 min)subjects (Fig. 3, A and C). However, after log conversion, thecorrelation between lnZrs and lnZrsSD was significantlydiminished in both asthmatic (r � �0.06, P � 0.85, 15 min;r � 0.38, P � 0.02, 1 min) and nonasthmatic (r � 0.32, P �0.28, 15 min; r � 0.40, P � 0.001, 1 min) (Fig. 3, B and D)subjects.

Impedance Variability in Asthmatic andNonasthmatic Subjects

To determine if we could detect differences in impedancevariability between asthmatic and nonasthmatic subjects, im-pedance was measured over 1 min using high-amplitude oscil-lations and a Fleisch pneumotachograph in the Belmont cohort(Table 3). Asthmatic and nonasthmatic subjects differed sig-nificantly in FEV1, Zrs, and Rrs. However, there was nosignificant difference in lnZrsSD between asthmatic and non-asthmatic subjects (Table 3). Furthermore, there was no sig-nificant correlation between lnZrsSD and AHR, measured bydose-response slope, in either asthmatic (r � 0.10, P � 0.71)or nonasthmatic subjects (r � �0.12, P � 0.27) (Fig. 4).Two of the 17 asthmatic subjects could not have AHR tohistamine measured since baseline FEV1 was less than 70%predicted.

Association with Diurnal Variability in Airway Caliber

The association between short-term variability, measured bylnZrsSD over 1 min, and diurnal variability of PEF and FEV1,measured over 2 wk, was assessed in 28 asthmatic subjects and7 nonasthmatic subjects (Table 4). As expected, the diurnalvariability of airway caliber was greater in asthmatic than

Table 1. Comparison of 15-min and 1-min FOT measurements in asthmatic and nonasthmatic subjects

Nonasthmatic Subjects Asthmatic Subjects

15 min 1 min P Value 15 min 1 min P Value

High amplitude

K-S (Zrs) 9.44�4.77 3.03�1.30 0.02* 6.79�3.72 1.52�0.74 0.008*K-S (lnZrs) 6.38�4.40 1.89�0.86 0.02* 4.50�2.72 1.61�0.63 0.008*lnZrsSD 0.27�0.07 0.30�0.08 0.21 0.29�0.09 0.24�0.05 0.14lnRrsSD 0.30�0.08 0.32�0.10 0.40 0.30�0.10 0.25�0.05 0.15

Low amplitude

K-S (Zrs) 8.02�1.99 1.92�0.73 0.03* 11.10�2.88 2.05�0.45 0.06*K-S (lnZrs) 2.04�1.12 1.07�0.41 0.09* 3.97�2.58 0.84�0.44 0.06*lnZrsSD 0.30�0.04 0.31�0.07 0.70 0.33�0.06 0.30�0.05 0.09lnRrsSD 0.32�0.05 0.34�0.07 0.36 0.32�0.05 0.33�0.07 0.20

Values are means � SD. Measurements were made at high oscillation amplitude (n � 9 nonasthmatic and 8 asthmatic subjects) and at low oscillation amplitude(n � 6 nonasthmatic and 5 asthmatic subjects). FOT, forced oscillation technique; K-S, Kolmogorov-Smirnov statistic; Zrs, respiratory system impedance;lnZrsSD, SD of log-transformed Zrs; Rrs, respiratory system resistance; lnRrsSD, SD of log-transformed Rrs. *Wilcoxon signed-ranked test. P values arecomparisons between 15-min and 1-min measurements.




Dow

nloaded from


nonasthmatic subjects, measured either by PEF (8.1 � 4.8%vs. 3.5 � 1.2%, P � 0.02) or FEV1 (7.6 � 5.2% vs. 2.9 �1.4%, P � 0.06). However, there was no significant differencein lnZrsSD between asthmatic and nonasthmatic subjects, and,in asthmatic subjects, lnZrsSD did not correlate with thediurnal variability of either PEF (rs ��0.12, P � 0.54) (Fig. 5) orFEV1 (rs ��0.04, P � 0.86).

Effect of Smooth Muscle Activation and Posture

Mean values for lnZrs and lnZrsSD in eight nonasthmaticsubjects at baseline and after methacholine challenge in theupright and supine positions are shown in Table 5. Comparedwith baseline in the upright position, lnZrs was increased bothby supine position and by methacholine challenge. However,lnZrsSD was not changed from the baseline upright values byeither supine position or methacholine challenge. In this groupof nonasthmatic subjects, both FEV1 (100.3 � 11.4% pre-dicted) and Rrs (82.5 � 30.5% predicted) were within thenormal predicted range.

DISCUSSION

Variability in airway caliber is a fundamental feature ofasthma, and the development of a simple and rapid test tomeasure such variation would be an important advance forboth diagnosis and monitoring of the disease. In this study,we set out to extend the findings reported by Que et al. (11),who suggested that such a test might be achieved by mea-suring spontaneous variation in airway caliber over a periodof minutes, using FOT, rather than the days or weeksrequired to assess diurnal variability of peak flow. However,we were unable to confirm these original findings. We foundthat the short-term variability in log impedance measuredusing FOT did not differ between asthmatic and nonasth-matic subjects, did not correlate with diurnal variability ofpeak flow or FEV1, did not correlate with AHR, and, innonasthmatic subjects, was not affected by posture ormethacholine challenge. The reasons for the differencebetween our studies and that of Que et al. (11) remain

Fig. 2. Normalized frequency distribution of Zrs in asthmatic (dotted lines) and nonasthmatic (solid lines) subjects. A and C (left panels) are on a linear scale.B and D (right panels) are on a logarithmic scale. A and B (top panels) are at high oscillation amplitude. C and D (bottom panels) are at low oscillation amplitude.




Dow

nloaded from


unclear but may reflect the small number of subjects in-cluded in the study of Que et al. (11).

The principles of the FOT are well established (2), and ourmeasurement device followed the same general design assystems described previously by ourselves (15, 16) and rec-ommended guidelines for measurement of impedance (6, 19).However, there were some differences between our FOT de-

vices and the device used by Que et al. (11). Systematicevaluation of several technical factors showed that neither theuse of a resistance mesh, instead of an inertance tube, nor theuse of a Hans Rudolph, rather than a Fleisch, pneumotacho-graph altered impedance or impedance variability. The onlyfactor that was found to affect impedance variability was theamplitude of the pressure oscillations. We found that, in a

Fig. 3. Correlation between mean and SD of impedance measured over 15 min of tidal breathing in 13 asthmatic (Œ) and 13 nonasthmatic (E) subjects (A andB) and 1 min of tidal breathing in 38 asthmatic and 102 nonasthmatic subjects (C and D). A and C show raw Zrs values, and B and D show values obtained afternatural log transformation.

Table 2. Comparison of low and high oscillatory amplitude and pneumotachograph device in asthmaticand nonasthmatic subjects

Nonasthmatic Subjects (n � 9) Asthmatic Subjects (n � 8)

Low High P Low High P

Fleisch device

lnZrsSD 0.27�0.10 0.22�0.10 0.01 0.30�0.10 0.27�0.10 0.08Rrs, cmH2O � l�1 � s 2.67�0.91 2.88�1.31 0.37 3.61�1.39 3.62�1.11 0.97Zrs, cmH2O � l�1 � s 2.88�1.00 3.08�1.38 0.38 3.96�1.65 3.94�1.33 0.87

Hans Rudolph device

lnZrsSD 0.27�0.11 0.22�0.09 0.02 0.30�0.10 0.28�0.08 0.43Rrs, cmH2O � l�1 � s 2.45�0.94 2.51�1.02 0.43 3.38�1.33 3.50�1.46 0.38Zrs, cmH2O � l�1 � s 2.62�0.98 2.66�1.04 0.62 3.68�1.56 3.85�1.73 0.22

Values are means � SD. P values are comparisons between low- and high-amplitude oscillations.




Dow

nloaded from


small group of nonasthmatic subjects, the variability of imped-ance was reduced with a high-amplitude oscillation, consistentwith an improvement in signal-to-noise ratio. Therefore, ourclinical measurements of impedance variability were madeusing a high-amplitude oscillation.

To increase the practicality of the measurement of imped-ance variability for use in the outpatient clinic and in epide-miological field studies, we examined the effect of reducing theduration of the measurement from 15 min to 1 min. Our dataconfirm the observation by Que et al. (11) that over 15 min,the distribution of lnZrs is nearly, but not quite, log-normallydistributed, and abbreviating the test to 1 min brought thedistribution closer to a log-normal distribution. There were nosignificant differences in impedance variability between mea-surements made over 15 min and 1 min in either asthmatic ornonasthmatic subjects. This observation, and the finding thatimpedance variability measured over 15 min did not differsignificantly between asthmatic and nonasthmatic subjects,

suggests that the differences in findings between our study andthat of Que et al. (11) cannot be attributed to differences in theduration of the measurement. The data-processing methods,including that used to remove artefacts due to glottal closure,swallowing, and leaks, were similar in our study and that ofQue et al. (11). In both studies, plots of Zrs against flow wereused to identify and exclude large values of Zrs at zero flow.There were no significant differences between asthmatic andnon-asthmatic subjects in the percentage of data excluded as arte-facts in either study.

To investigate the clinical significance of impedance vari-ability, impedance was measured over 1 min using a high-amplitude oscillation at 6 Hz in asthmatic and nonasthmaticsubjects. Impedance variability was not significantly differentin asthmatic and nonasthmatic subjects, although there was a

Fig. 4. Airway hyperresponsiveness to histamine was not significantly corre-lated with lnZrsSD in asthmatic (r � 0.10, P � 0.71) (Œ) and nonasthmatic(r � �0.12, P � 0.27) (E) subjects. DRS, dose-response slope.

Fig. 5. Diurnal variability of peak expiratory flow (PEF) was not significantlycorrelated with lnZrsSD (rs � 0.09, P � 0.62) in asthmatic (Œ) and nonasth-matic (E) subjects.

Table 3. Subject characteristics for asthmatic andnonasthmatic subjects from the Belmont cohort

Normal Subjects(n � 82)

Asthmatic Subjects(n � 17) P Value

M:F 46:36 7:10FEV1, %predicted 106.1�8.9 91.2�15.4 �0.001Log DRS 0.58�0.10 1.58�0.40 �0.001Rrs, cmH2O � l�1 � s 3.46�1.16 4.94�2.40 �0.001Rrs, %predicted 124.6�41.9 172.7�83.9 0.001lnRrsSD 0.25�0.06 0.28�0.08 0.14Zrs, cmH2O � l�1 � s 3.51�1.17 5.08�2.51 �0.001lnZrs 1.18�0.30 1.48�0.46 0.001ZrsSD 0.92�0.50 1.45�1.00 0.002lnZrsSD 0.24�0.06 0.27�0.08 0.09

Values are means � SD. M, male; F, female; %FEV1, percent predictedforced expiratory volume in 1 s; DRS, dose-response slope to histamine; %Rrs,percent predicted Rrs.

Table 4. Subject characteristics for asthmatic andnonasthmatic subjects who completed 2 wk ofPEF monitoring

Normal Subjects(n � 7)

Asthmatic Subjects(n � 28) P Value

M:F 2:5 15:13Age, yr 36.4�11.6 49.8�14.6 0.03FEV1, % predicted 109.0�13.6 81.8�17.2 �0.001Log DRS 0.51�0.06 1.01�0.38 0.002Rrs, cmH2O � l�1 � s 2.81�0.81 3.91�1.36 0.05Rrs, %predicted 99.6�26.4 134.3�35.0 0.02Zrs, cmH2O � l�1 � s 3.14�0.99 4.57�1.82 0.05lnZrs 1.06�0.32 1.40�0.39 0.04ZrsSD 0.99�0.75 1.41�0.98 0.31lnZrsSD 0.27�0.10 0.28�0.09 0.71LABA 4ICS 20

Values are means � SD. PEF, peak expiratory flow; DRS, dose-responseslope to methacholine; LABA, long-acting �2-agonist; ICS, inhaled cortico-steroids.




Dow

nloaded from


nonsignificant trend (P � 0.09) for increased impedance vari-ability in asthmatic subjects in the Belmont cohort. This raisesthe possibility that impedance variability might provide clini-cally useful information in some populations. However, themagnitude of the difference in impedance variability betweenasthmatic and normal subjects in the Belmont cohort (0.24 innormal subjects and 0.27 in asthmatic subjects) was smallerthan the difference seen in the study by Que et al. (11) (0.24 innormal subjects and 0.34 in asthmatic subjects). More impor-tantly, there was a very large overlap in impedance variabilitybetween normal and asthmatic subjects, suggesting that themeasurement would have no diagnostic utility. Furthermore,we found that impedance variability was not related to diurnalvariability of airway caliber or to AHR over a wide range ofseverity. These findings suggest that although it may be pos-sible to detect differences in impedance variability betweenasthmatic and normal subjects in some samples, these differ-ences are inconsistent, and impedance variability is not relatedto other well-validated markers of asthma.

We and others have previously examined the source ofvariability in Rrs values, hence Zrs values, during normal tidalbreathing (10, 16). Much of the variability is directly related toboth fluctuations in flow and in volume, both of which havebeen long recognized (20). The changes in volume are likelydue to direct increases in caliber due to changes in transmuralpressure generated by the swings in pleural pressure, albeitsmall, during tidal breathing. Thus it is possible that alterationsin the rate and depth of breathing could affect impedancevariability.

The subjects in our studies were well characterized. TheBelmont cohort is a general population sample of young adults(9), and impedance measurements were made in 243 subjectsduring a follow-up study of this cohort when they were aged 28to 30 yr. We set very strict criteria to define both nonasthmaticand asthmatic subjects to maximize the chance of detecting anydifferences in impedance variability. Of these, we identified 82subjects that met our strict criteria for nonsmoking, nonasth-matic controls and 17 that met the criteria for nonsmokingasthmatic subjects. Similarly, the 28 asthmatic subjects re-cruited from the outpatient clinic were well characterized andhad clinically stable asthma that ranged in severity fromintermittent to severe asthma, on the basis of the GlobalInitiative for Asthma (GINA) guidelines (3). In the study ofQue et al. (11), 7 of the 10 subjects were said to have“worsening” asthma, implying clinical instability, although theclinical criterion for worsening asthma was not defined, and nodetails of medication use were given. In asthmatic subjects,resting lung function, measured by mean FEV1% predicted,and by Zrs was similar in our outpatient group and thosestudied by Que et al. (11). In the nonasthmatic controls in both

the Belmont cohort and the laboratory studies, mean resistancevalues were close to normal predicted values (7), but mean Zrswas a little higher than the equivalent measurement in controlsubjects in the study of Que et al. (11). No anthropometric data,or measurements of resistance relative to predicted values,were provided for the nonasthmatic subjects in the study byQue et al. (11), so it is not clear whether their subjects weresimilar to ours. Thus, although the inclusion criteria for non-asthmatic and asthmatic subjects were similar in our study andthat of Que et al. (11), it is difficult to compare the subjects inthe two studies, and it is possible that the differences in thefindings of the two studies could be attributed to differences insubject characteristics.

Que et al. (11) suggested that an increase in short-termvariability in asthmatic subjects might be due to airway smoothmuscle activation combined with unloading of the airways,possibly resulting from peribronchial inflammation. Theymimicked this condition in nonasthmatic subjects using metha-choline challenge to cause airway smooth activation and supineposture to cause unloading and reported that the combinationincreased variability into the asthmatic range. We reasoned thateven if the differences between the studies were due to differ-ence in subject selection, we should still be able to detect achange in Zrs variability by changing posture and smoothmuscle activation. However, we could not reproduce the find-ings of Que et al. (11) and found no changes in lnZrsSD in thesupine position or after methacholine challenge. This suggeststhat the difference between studies may be more fundamentalthan differences in subject selection.

The importance of using log-transformed, rather than raw,Zrs values to measure impedance variability was highlightedboth in our study and in that of Que et al. (11). Both studiesshow clearly that the distribution of Zrs values is closer tolog-normal than to a normal Gaussian distribution. Thus theSD of the raw Zrs values is not a true representation of thevariance of such a skewed distribution. Furthermore, we haveshown that there is a very close correlation between mean Zrsand its SD, suggesting that the variability, measured by ZrsSD,is highly dependent on airway caliber. This effect is probablydue, in part, to a geometric amplification of small variations inairway resistance in airways of small diameter. While logtransformation substantially diminished the association be-tween caliber and variability both in our study and in that ofQue et al. (11), other studies have not made this adjustment. Arecent study in children (18), in which the short-term variabil-ity of airway caliber differed between asthmatic and nonasth-matic children, did not use log-transformed data and did notmake any analysis of the association between variability andairway caliber. It is likely that the variability measured in thisway is simply a proxy for differences in airway caliber. It has

Table 5. Effect of posture and MCh on impedance measures in normal subjects

Upright Baseline Upright MCh P* Supine Baseline P* Supine MCh P*

Rrs, cmH2O � l�1 � s 2.31�0.93 4.27�1.42 �0.001 2.78�0.89 0.05 5.21�1.21 �0.001Zrs, cmH2O � l�1 � s 2.48�0.97 5.00�1.73 �0.001 3.12�1.00 0.01 6.80�1.21 �0.001Zrs SD, cmH2O � l�1 � s 0.48�0.26 0.90�0.39 0.004 0.70�0.26 0.01 1.26�0.45 0.002lnZrs 0.82�0.40 1.52�0.42 �0.001 1.07�0.33 0.006 1.85�0.35 �0.001lnZrsSD 0.18�0.05 0.18�0.03 0.81 0.21�0.05 0.10 0.20�0.04 0.63

Values are means � SD; n � 8 normal subjects. Impedance variables were measured using the Hans Rudolph device at high oscillation amplitude. MCh,methacholine. *Significance of difference from upright baseline.




Dow

nloaded from


not been established whether the variability in airway caliberprovides any additional, clinically relevant information otherthan that available from airway caliber alone.

A simple and rapid test to measure airflow variability wouldbe valuable for diagnosis and monitoring of asthma. Airflowvariability is determined by a combination of factors thatinclude not only the intrinsic airway dynamics and smallmomentary changes in airway caliber but also larger-scaleenvironmental factors, such as circadian rhythms and allergenexposure. The hypothesis that the intrinsic variability of theairways can predispose the airways to be more susceptible tothe larger-scale factors has great appeal. Although the studyreported by Que et al. (11) implied that a test of intrinsicvariability might be possible, in our studies we were unable toreproduce their findings. Using standard measurement anddata-processing methods, and systematically eliminating anyeffects of equipment or protocol between the studies, we foundno difference in the variability of impedance during tidalbreathing between asthmatic and nonasthmatic subjects. Fur-thermore, the variability of impedance did not relate to mea-sures of asthma severity such as AHR and PEF variability andwas not affected by posture or smooth muscle activation.Although we have not been able to identify the source of thedifferences between our findings and those of Que et al. (11),our findings in large populations of well-characterized asth-matic and nonasthmatic subjects do not support the view thatmeasurement of short-term variability of airway caliber canprovide clinically useful information about airway status.

ACKNOWLEDGMENTS

We thank Dr. Geoffrey Maksym from the School of Biomedical Engineer-ing at Dalhousie University Canada for careful and insightful comments on themanuscript. We also thank Kitty Ng, from the Woolcock Institute of MedicalResearch in Sydney, study coordinator for the Belmont cohort.

GRANTS

This study was supported by the National Health and Medical ResearchCouncil and by the Cooperative Research Centre for Asthma. H. K. Reddel isfunded by Asthma Foundation of New South Wales.

REFERENCES

1. American Thoracic Society. Standardization of spirometry. 1994 update.Am J Respir Crit Care Med 152: 1107–1136, 1995.

2. Dubois AB, Lanphier EH, Marshall R. Resistance to breathing innormal subjects during simulated dives. J Appl Physiol 9: 5–10, 1956.

3. Global Initiative for Asthma. Workshop Report: Global Strategy forAsthma Management and Prevention. NIH Publication No. 02–3659.[http://www.ginasthma.org].

4. National Heart Lung, and Blood Institute. National asthma educationprogram expert panel report. Guidelines for the diagnosis and managementof asthma. J Allergy Clin Immunol 88: 425–534, 1991.

5. O’Connor G, Sparrow D, Taylor D, Segal M, Weiss S. Analysis of doseresponse curves to methacholine. An approach suitable for populationstudies. Am Rev Respir Dis 136: 1412–1417, 1987.

6. Oostveen E, MacLeod D, Lorino H, Farre R, Hantos Z, Desager K,Marchal F. The forced oscillation technique in clinical practice: method-ology, recommendations and future developments. Eur Respir J 22:1026–1041, 2003.

7. Pasker HG, Mertens I, Clement J, Van de Woestijne KP. Normalvalues of total respiratory input resistance and reactance for adult men andwomen. Euro Respir Rev 4: 134–137, 1994.

8. Peat JK, Salome CM, Berry G, Woolcock AJ. Relation of dose-responseslope to respiratory symptoms and lung function in a population study ofadults living in Busselton, Western Australia. Am Rev Respir Dis 146:860–865, 1992.

9. Peat JK, Salome CM, Sedgwick CS, Kerrebijn J, Woolcock AJ. Aprospective study of bronchial hyperresponsiveness and respiratory symp-toms in a population of Australian schoolchildren. Clin Exp Allergy 19:299–306, 1989.

10. Peslin R, Ying Y, Gallina C, Duvivier C. Within-breath variations offorced oscillation resistance in healthy subjects. Eur Respir J 5: 86–92,1992.

11. Que CL, Kenyon CM, Olivenstein R, Macklem PT, Maksym GN.Homeokinesis and short-term variability of human airway caliber. J ApplPhysiol 91: 1131–1141, 2001.

12. Reddel H, Salome C, Peat J, Woolcock A. Which index of peakexpiratory flow is most useful in the management of stable asthma? Am JRespir Crit Care Med 151: 1320–1325, 1995.

13. Reddel HK, Ware SI, Salome CM, Jenkins CR, Woolcock AJ. Pitfallsin processing home electronic spirometric data in asthma. Eur Respir J 12:853–858, 1998.

14. Ryan G, Latimer KM, Dolovich J, Hargreave FE. Bronchial respon-siveness to histamine: relationship to diurnal variation of peak flow rate,improvement after bronchodilator, and airway calibre. Thorax 37: 423–429, 1982.

15. Salome CM, Thorpe CW, Diba C, Brown NJ, Berend N, King GG.Airway re-narrowing following deep inspiration in asthmatic and non-asthmatic subjects. Eur Respir J 22: 62–68, 2003.

16. Thorpe CW, Salome CM, Berend N, King GG. Modeling airwayresistance dynamics after tidal and deep inspirations. J Appl Physiol 97:1643–1653, 2004.

17. Toelle BG, Peat JK, Salome CM, Mellis CM, Woolcock AJ. Toward adefinition of asthma for epidemiology. Am Rev Respir Dis 146: 633–637,1992.

18. Trubel H, Banikol WK. Variability analysis of oscillatory airway resis-tance in children. Eur J Appl Physiol 94: 364–370, 2005.

19. Van de Woestijne KP, Desager KN, Duiverman EJ, Marchal F.Recommendations for measurement of respiratory input impedance bymeans of the forced oscillation method. Eur Respir Rev 4: 235–237, 1994.

20. Vincent NJ, Knudson R, Leith DE, Macklem PT, Mead J. Factorsinfluencing pulmonary resistance. J Appl Physiol 29: 236–243, 1970.

21. Yan K, Salome C, Woolcock AJ. Rapid method for the measurement ofbronchial responsiveness. Thorax 38: 760–765, 1983.




Dow

nloaded from


Documents

Short-term variability of airway caliber--a marker of asthma?