Evaluation of the Double-Tracer Gas Single-Breath Washout Test in a Pediatric Field Study

Background The early life origins of chronic pulmonary diseases are thought to arise in peripheral small airways. Predictors of ventilation inhomogeneity, a proxy of peripheral airway function, are understudied in schoolchildren. Research Question Is the double-tracer gas single-breath washout (DTG-SBW) measurement feasible in a pediatric field study setting? What are the predictors of the DTG-SBW-derived ventilation inhomogeneity estimate in unselected schoolchildren? Study Design and Methods In this prospective cross-sectional field study, a mobile lung function testing unit visited participating schools in Switzerland. We applied DTG-SBW, fraction of exhaled nitric oxide (Feno), and spirometry measurements. The DTG-SBW is based on tidal inhalation of helium and sulfur-hexafluoride, and the phase III slope (SIIIHe-SF6) is derived. We assessed feasibility, repeatability, and associations of SIIIHe-SF6 with the potential predictors of anthropometrics, presence of wheeze (ie, parental report of one or more episode of wheeze in the prior year), Feno, FEV1, and FEV1/FVC. Results In 1,782 children, 5,223 DTG-SBW trials were obtained. The DTG-SBW was acceptable in 1,449 children (81.3%); the coefficient of variation was 39.8%. SIIIHe-SF6 was independently but weakly positively associated with age and BMI. In 276 children (21.2%), wheeze was reported. SIIIHe-SF6 was higher by 0.049 g.mol.L−1 in children with wheeze compared with those without and remained associated with wheeze after adjusting for age and BMI in a multivariable linear regression model. SIIIHe-SF6 was not associated with Feno, FEV1, and FEV1/FVC. Interpretation The DTG-SBW is feasible in a pediatric field study setting. On the population level, age, body composition, and wheeze are independent predictors of peripheral airway function in unselected schoolchildren. The variation of the DTG-SBW possibly constrains its current applicability on the individual level. Trial Registration ClinicalTrials.gov; No.: NCT03659838; URL: www.clinicaltrials.gov

The early life origins of respiratory diseases (eg, COPD) are thought to arise in small airways of lung periphery. 1ecause of practical constrains, predictors of peripheral airway function (ie, ventilation inhomogeneity) remain understudied in large pediatric populations.The doubletracer gas single-breath washout (DTG-SBW) test may overcome these constraints.The DTG-SBW is a simple lung function test based on tidal inhalation and exhalation of Helium (He) and sulfur-hexafluoride (SF 6 ). 2,3The derived slope of phase III (SIII He-SF6 ) measures ventilation inhomogeneity of He and SF 6 , which differ in diffusive gas mixing properties in small airway compartments. 2,3[4][5][6] DTG-SBW may be a simple and accessible tool to allow for early detection of lung function alterations (ie, ventilation inhomogeneity) associated with negative respiratory disease outcomes.However, in unselected pediatric populations, feasibility and repeatability of the DTG-SBW, and predictors of the SIII He-SF6 , are unknown.][9] Previous studies suggest that high BMI is associated with dysanaptic lung growth, a nonproportional growth of the airways and lung, because adipose tissue and proinflammatory mediators affect lung growth and development.Pediatric wheeze and airflow limitation increase the risk of COPD in adults. 10is study addressed the following two research questions: (1) Is the DTG-SBW measurement feasible in a pediatric field study setting?, and (2) What are the predictors of the DTG-SBW-derived ventilation inhomogeneity estimate in a sample of schoolchildren?To accomplish this, we applied the DTG-SBW test in a large pediatric field study to assess its feasibility and reliability, and explore associations between SIII He-SF6 and anthropometric variables, wheeze, and standard lung function indexes.
Previous estimates of feasibility and intratest variability of the nitrogen single-breath washout test in children and adults ranged from 74% to 89% and 13% to 24%, respectively. 11,124][15] We hypothesized that the feasibility and intratest variability of the DTG-SBW applied in unselected schoolchildren in a field study setting were > 75% and < 25%, respectively.
We further hypothesized that SIII He-SF6 is associated with age and body composition, 7 wheeze, 9,16 spirometry indexes, and fraction of exhaled nitric oxide (FENO).

Study Design and Methods
LuftiBus in the School (LUIS) is a prospective cross-sectional observational field study in unselected school-aged children (ClinicalTrials.govNo. NCT03659838). 17Inclusion criteria were 6 to 17 years of age, German language skills, and consent to participate.There were no predefined exclusion criteria.A mobile lung function testing unit (motorbus) visited 37 schools in the canton of Zurich, the most populated canton in Switzerland, between 2013 and 2016. 17 We found an acceptable success rate; substantial test variation; and identified age, body composition, and wheeze as independent but relatively weak predictors of ventilation inhomogeneity.Interpretation: The test variation currently constrains the use of the DTG-SBW in children.However, the current data suggest that schoolchildren with wheeze have alterations in ventilation inhomogeneity which can be attributed to peripheral airway dysfunction.the study was representative to the Swiss-SEP distribution from families with at least one child living in the household from the canton of Zurich. 17LUIS took place throughout different seasons (e- Fig 1).A consecutively recruited convenience sample of the whole population was studied because the hardware for DTG-SBW including tracer gas supply became available later during the study.Details about study design, sample size estimates, and data collection have been previously described. 17Children performed lung function tests in the following sequence: DTG-SBW, FENO measurement, and spirometry.The ethics committee of the canton of Zurich approved the study (KEK-ZH-Nr No. 2014-0491).Parents or caregivers signed the informed consent form.Children assented verbally and those aged $ 15 years also signed the informed consent form.
Anthropometrics were measured on the bus on-site, and parental questionnaires were used to collect information on exposures, respiratory symptoms, diagnoses, and prescribed medication. 17heeze was specified as parental report of continuous whistling sound during expiration during one or more episodes in the past 12 months. 17dal DTG-SBW was performed in triplet using the Exhalyzer D (EcoMedics AG) according to recommendations. 18An inert doubletracer gas mixture containing 5% SF 6 , 26.3% He, 21% oxygen, and balance nitrogen was inhaled during a single tidal breath and tidally exhaled to functional residual capacity.The setup, protocol, and quality control criteria were in accordance with the European Respiratory Society consensus on inert gas washout testing and were previously described. 3,17,18The DTG-SBW was analyzed automatically followed by quality control using a customized software platform (LungSim based on Matlab R2014a [The MathWorks Inc]). 17Quality control was performed by two trained lung function technicians and included central overread.The DTG-SBW trials were categorized according to the quality control categories of A, B, or failed.The quality control protocol used can be found in e-Table 1.Only children who achieved at least two acceptable DTG-SBW trials were included.
The primary outcome measure was the mean SIII He-SF6 of all technically acceptable DTG-SBW curves of each subject.SIII He-SF6 was computed from the volumetric expirogram by fitting a linear regression slope to the molar mass signal between 65% and 95% of the expired volume.In addition, SIII He-SF6 was normalized for expired volume by multiplication with the expired tidal volume as a secondary outcome. 17Findings are reported in e-Appendix 1.[4][5][6] FENO (parts per billion [ppb]) was measured according to recommendations using a single-breath online method and a fast response chemiluminescence analyzer (CLD 88; EcoMedics AG). 19urther details on test performance and quality control have been previously described. 17FENO is a proxy of eosinophilic airway inflammation; FENO values $ 20 ppb can be considered elevated. 20irometry was performed using a standard spirometer (Masterlab; Jaeger) according to recommendations. 21Indices were FEV 1 and FEV 1 /FVC.Values were expressed as z score according to Global Lung Function Initiative reference equations. 17,22Lower limit of normal of FEV 1 and FEV 1 /FVC were set at # À1.645 z score as recommended. 21,22alysis Discrete variables were expressed as count (percentage), and continuous variables were expressed as mean AE SD or median (interquartile range [IQR]), as appropriate.Missing data were not imputed. 17Between-group differences were assessed using unpaired t tests for parametric estimates and Wilcoxon-Mann-Whitney test for nonparametric estimates.DTG-SBW test feasibility was determined as the success rate calculated as the percentage of children with at least two acceptable trials of all children attempting the test.Intratest repeatability was calculated as coefficient of variation.The success rate of DTG-SBW was calculated as the number of successful DTG-SBW trials as a percentage of all DTG-SBW trials performed per subject.
Associations were assessed using scatterplots, Pearson correlations, and univariable linear regression models.Potential predictors of SIII He-SF6 included age, sex, height, weight, and BMI z score; wheeze; and FENO, FEV 1 , and FEV 1 /FVC.A multivariable linear regression model was used to explore these variables as independent predictors of SIII He-SF6 .Variables were analyzed as continuous variables with their original scale, wheeze as a binary variable (ie, yes or no), and FENO as quintiles ensuring balanced observations per category.Regression model diagnostics were used to confirm underlying assumptions.P < .05 was considered statistically significant.All analyses were performed using STATA (USA Version 16.0; StataCorp LP).Figures were made using GraphPad Prism version 8.0.1 (GraphPad Software).

Results
In total, 3,870 children were enrolled into the LUIS study (Fig 1).The children's median age was 12.1 years (IQR, 9.3-14.0years), and one-half of the population were female.The DTG-SBW test was applied in 1,782 children (46.0%), who were slightly younger (0.7 years), had slightly lower Swiss-SEP (1.3 points), reported hay fever somewhat more frequently (2.7%), and had slightly lower FENO (2.6 ppb) than children not invited to perform the DTG-SBW.There were no systematic differences in anthropometric and lung function estimates between these children (e-Table 2).Anthropometric characteristics and lung function estimates can be found in Table 1 and e-Table 2.

Feasibility and Repeatability
In total, 5,223 DTG-SBW trials were obtained, of which 4,090 trials (78.3%) were of acceptable quality.Therefore, 1,449 out of 1,782 children (81.3%) successfully achieved DTG-SBW tests (e-Tables 3-5).DTG-SBW success rate was higher than the hypothesized success rate (75%).Children with successful DTG-SBW tests were 1.1 years older, had a lower Swiss-SEP, and reported wheeze more often than the children with unsuccessful tests; all other anthropometric and questionnaire data were comparable (e-Table 4).
In children with a successful DTG-SBW test, trial quality was rated higher more often.Frequency of higher trial quality control categories was associated with the number of acceptable trials (e-Fig 2, e-Tables 6-8) until a maximum of four trials.The mean SIII He-SF6 AE SD was À0.30 AE 0.42 g.mol.L À1 .The repeatability of SIII He-SF6 with a median intratest coefficient of variation of 39.8% (IQR, 22.0%-70.9%)was poorer than the hypothesized repeatability (25%).For more details on DTG-SBW feasibility and repeatability, we refer to e-Figure 3 and e-Table 9.

Predictors of Ventilation Inhomogeneity
SIII He-SF6 was associated with all preselected anthropometric variables except for sex.In univariable regression models, SIII He-SF6 was positively associated with age, height, weight, and BMI z score (Fig 2 ,  Table 2).In a multivariable regression model, only age and BMI remained independent predictors of SIII He-SF6 , increasing SIII He-SF6 by 0.013 g.mol.L À1 per 1-year increase in age and by 0.060 g.mol.L À1 per 1 z score increase in BMI, respectively.
In total, 276 children reported wheeze, 1,025 children had no wheeze, and 148 children had missing information regarding wheeze and were excluded from this analysis (Fig 1).Children with wheeze were slightly older (0.7 years), heavier (BMI, 0.2 z score), and reported atopic diseases more frequently (e-Table 10).
FENO was slightly higher (4.3 ppb) and spirometry was lower (FEV 1 , 0.21 z score) in children with wheeze than in children without wheeze (e-Table 10).SIII He-SF6 was associated with wheeze in univariable regression models, and it remained weakly positively associated with wheeze after adjustment for age and BMI z score (Table 2).SIII He-SF6 was higher by 0.049 g.mol.L À1 in children with wheeze compared with those without, but it was not associated with FENO or with the spirometry indices FEV 1 and FEV 1 /FVC (e-Table 11, Table 2).A post hoc analysis in a subgroup of children with a BMI z score > 1.0 showed similar results compared with the primary analysis in the whole cohort (e-Table 12).

Discussion
In this large pediatric field study setting, we found that the DTG-SBW measurement was feasible in a mobile bus lung function laboratory.Repeatability was poorer than hypothesized.We identified predictors of ventilation inhomogeneity in unselected schoolchildren.SIII He-SF6 was weakly positively associated with age, BMI, and wheeze but not with FENO or spirometry indices.On the population level in sufficiently large samples such as in our study, SIII He-SF6 captures a subtle signal of chestjournal.orgalterations in ventilation inhomogeneity, suggesting small airways dysfunction in children with wheeze.However, on the individual level, SIII He-SF6 does not seem sensitive enough to screen for alterations in ventilation inhomogeneity in unselected children.

Interpretation
In this field study, we found an acceptable success rate in unselected schoolchildren.The current success rate was higher than hypothesized (75%) but lower than previously reported (92%) in selected children within Associations between SIII He-SF6 and potential predictors were assessed using univariable and multivariable linear regression models.Predictors were age, sex, height, weight, and BMI and wheeze, FENO, FEV 1 , and FEV 1 /FVC.Wheeze was included as a binary variable (ie, yes or no), and FENO was included as datadriven quintiles ensuring balanced observations per category; all other variables were included as continuous variables with their original scale.The quintile boundaries for FENO were as follows: 0.0 to 4.9, 5.0 to 8.8, 8.9 to 13.8, 13.9 to 23.4, and 23.5 to 197.0 parts per billion, respectively.A multivariable linear regression model was used to assess which anthropometric variables were independent predictors of SIII He-SF6 , and the independent predictors age and BMI were used to adjust the association of SIII He-SF6 with wheeze.All associations described the change in SIII He-SF6 in g.mol.L À1 induced by 1-unit increase in the predictor.FENO ¼ fraction of exhaled nitric oxide; SIII He-SF6 ¼ phase III slope.research laboratory settings. 2Because of the field study conditions with possibly a more distracting environment than standard laboratories and children naive to the use of sealed mouthpieces, success rates were somewhat lower.This is supported by the observed learning effect during testing in this study.Previously reported success rates of other tidal breathing protocols were similar to our findings. 23In our study, the reason for DTG-SBW test failure was mainly variable breathing pattern.Because of time constraints, details of test failure were not recorded on-site.In a previous study performed in a lung function laboratory, variable breathing pattern accounted for 94% of DTG-SBW test failures in schoolaged children. 2 In that study, reasons for DTG-SBW test rejection were (1) variable tidal flows and volumes, (2) small tidal volumes lacking phase III of the expirogram, and (3) technical errors. 2 The coefficient of variation quantifying intratest variability of SIII He-SF6 was higher than previously reported (19%) for DTG-SBW, 2 but comparable with the phase III slope indices Scond and Sacin from the established multiple-breath washout test, supporting the reliability of the current analysis. 6,24The estimated mean value of SIII He-SF6 was close to zero in our study; therefore, small changes may have increased the coefficient of variation exponentially.The variability seen can be because of factors related to the field study setting, but estimation of the proportion of variability that can be attributed to the setting is challenging.It is well established, however, that the intratest variability for inert gas analysis is high, commonly thought to be because of effects of breathing.Interestingly, variability of SIII He-SF6 was associated with age and the variability in tidal volume in our study, but not with other potential explanatory variables (eg, the SIII He-SF6 value itself).These data suggest that phase III slope indices are prone to considerable inherent physiologic variability and tidal breathing.Normalization for tidal volume alone may not substantially decrease variability or increase sensitivity of the test. 22,25,26Current protocols for phase III slope measurement seem to require refinement prior to clinical routine application.The high intratest variability may dampen test sensitivity to estimate subtle physiologic signals in individuals.Further research is needed to identify potentially modifiable sources of test variability and assess the potential of alternate protocols to reduce intratest variability of the DTG-SBW.
Although it is established that in cystic fibrosis the lung clearance index correlates with structural airway changes detected in chest CT scan, there is one negative study for SIII He-SF6 . 27Multiple-breath washout or lung imaging were not obtained in this field study.However, these estimates would have allowed more in-depth assessment of the diagnostic performance of SIII He-SF6 .Our study provides further evidence that body composition is a predictor of lung function development.Our data are in line with previous findings suggesting age-dependent or height-dependent effects on ventilation inhomogeneity estimates (eg, lung clearance index from multiple-breath washout). 7,28Our data further suggest that unfavorable body composition estimated by BMI may modify ventilation inhomogeneity.Reasons remain speculative  We have excluded one outlier (BMI ¼ À1.7 z score and SIII He-SF6 ¼ 2.9 g.mol.L À1 ) in panel A to ease visualization.SIII He-SF6 ¼ phase III slope.
chestjournal.orgbut may partly relate to airway dysanapsis observed in children with high BMI. 28Indeed, we have recently shown that the spirometry indexes obtained in this cohort did not fit well the reference values from the Global Lung Function Initiative. 26Underestimation of FEV 1 and FVC in the current cohort was partly explained by BMI; however, FEV 1 /FVC was not affected.
Wheezy symptoms are common and account for considerable burden in pediatric health care.5][6] Interestingly, our study suggests that these alterations in ventilation inhomogeneity were independent of airway inflammation or airflow limitation.However, overlap in SIII He-SF6 values of children with vs without wheeze was considerable.
Comparable with other studies, peripheral airway function estimated by current inert gas tests appears largely normal in children with wheeze. 29Therefore, the difference in SIII He-SF6 in children with wheeze was relatively small, and adjustment for age and BMI further increased the CIs.Comparable with SIII He-SF6 , FENO, FEV 1 , and FEV 1 /FVC values were overlapping between children with vs without wheeze, suggesting overall relatively low pretest probability (ie, low prevalence) of lung function abnormalities in the current cohort.

Strengths and Limitations
The large sample size is a strength of this prospective study because it allows conclusive analyses of potential predictors of lung function.Our study allowed for thorough assessment of potential predictors of the SIII He-SF6 estimate, including anthropometric and lung function measures.The large sample of unselected schoolchildren supports the generalizability of our findings.Participation of schools was decided by the heads of the schools, which may have introduced selection to some extent.However, the Swiss-SEP for families participating in the study was representative for the canton of Zurich. 11Because the DTG-SBW test was introduced later in this study, only a subgroup of the LUIS study was invited to perform DTG-SBW.During this study period, the frequency of measurements varied over time.SIII He-SF6 was not influenced by timing of measurements (ie, seasonal effects).
The current protocol determined the sequence of testing to avoid influences from forced breathing maneuvers during spirometry on SIII He-SF6 and FENO.Tidal inhalation of inert gas during the DTG-SBW unlikely influenced subsequent FENO or spirometry measurements.
We report wheeze in 19% of our study population, whereas this was 8% for the total LUIS population.In the latter study, wheeze was defined as whistling or panting sound originating from the chest within the last 12 months.In the current analysis, we expanded the definition of wheeze by adding whistling or panting sound originating from the chest in response to triggers (eg, exercise, respiratory tract infection, cold air, other).
The proportion of variation in SIII He-SF6 in this unselected population that can be explained by wheeze was low.We acknowledge that questionnaire-based classification of wheeze may have been subject to recall and misclassification bias.Parent-reported wheeze may have been less precise than physicianreported wheeze.The sound of wheezing that parents notice unaided by a stethoscope (ie, audible wheeze) originates from trachea and larger bronchi, rather than from the peripheral small airways.We assume that misclassification rather led to underestimation of the strength of association between wheeze and SIII He- SF6 .Premature birth may affect lung development and alter ventilation inhomogeneity in some children.We were unable to explore possible effects of prematurity on SIII He-SF6 .

Interpretation
Our results suggest that DTG-SBW is feasible in children between 6 and 17 years of age.Data from younger children are scant and warrant further study. 2,4,30Despite good feasibility, the high variability and presumably low sensitivity to capture slightly increased ventilation inhomogeneity constrain its use in unselected individuals.Currently, the DTG-SBW is applicable in research settings and sufficiently large populations, or in selected individuals with high pretest probability of lung function abnormalities.3][4][5][6] Distinct interpretation of dynamics in SIII He-SF6 warrants further research.Future longitudinal studies are warranted to establish the minimal clinically important differences derived from variability estimates and patient-reported outcomes.
To conclude, the DTG-SBW measurement is feasible in pediatric field studies.However, relatively high variability of SIII He-SF6 appears to limit the interpretation.This makes DTG-SBW currently unsuitable in small populations with low pretest probability of impaired lung function.In the current relatively large population of unselected schoolchildren, age, body composition, and wheeze were identified as predictors of ventilation inhomogeneity estimated by SIII He-SF6 .Schoolchildren with wheeze may have alterations in ventilation inhomogeneity which can be attributed to peripheral airway dysfunction. Funding/Support Most children were born in Switzerland (88%) and predominantly of White European ancestry (75.8%).The distribution of the Swiss socioeconomic position index (Swiss-SEP) for families participating in Take-home Points Study Question: In a large pediatric field study of unselected schoolchildren, what are the success rates and test variation of the double-tracer gas singlebreath washout (DTG-SBW) measurement and what are the predictors of ventilation inhomogeneity estimated by the DTG-SBW?Results:

Figure 2 -
Figure 2 -A, B, Scatterplot of the double-tracer gas single-breath washout-derived SIII He-SF6 vs BMI (A) and FEV 1 (B).BMI and FEV 1 are expressed as z score.The closed circles display SIII He-SF6 values of children without wheeze, and open circles display values of children with wheeze.We have excluded one outlier (BMI ¼ À1.7 z score and SIII He-SF6 ¼ 2.9 g.mol.L À1 ) in panel A to ease visualization.SIII He-SF6 ¼ phase III slope.

TABLE 1 ]
Characteristics of Study Participants Data are presented as mean AE SD, No. (%), or median (interquartile range).All questionnaire data were parent reported.Asthma medication included any inhaled corticosteroids or short-acting or long-acting beta-agonists or systemic treatment (eg, leukotriene receptor antagonists).DTG-SBW ¼ doubletracer gas single-breath washout; FENO ¼ fraction of exhaled nitric oxide; LUIS ¼ LuftiBus in the School; ppb ¼ parts per billion; SIII He-SF6 ¼ phase III slope; Swiss-SEP ¼ Swiss socioeconomic position index.

TABLE 2 ]
Nonadjusted and Adjusted Association Between SIII He-SF6 and Potential Predictors Study setup, development, and data collection were funded by Lunge Zürich, and the analysis was funded by grants from Lungenliga Bern and Foundation KinderInsel.A.-C. K. is recipient of a Swiss Excellence Grant from the Swiss government.