Traffic-related air pollution and alveolar nitric oxide in southern California children
- Sandrah P. Eckel1⇑,
- Zilu Zhang1,
- Rima Habre1,
- Edward B. Rappaport1,
- William S. Linn1,
- Kiros Berhane1,
- Yue Zhang2,
- Theresa M. Bastain1 and
- Frank D. Gilliland1
- 1Dept of Preventive Medicine, University of Southern California, Los Angeles, CA, USA
- 2Dept of Internal Medicine, University of Utah, Salt Lake City, UT, USA
- Sandrah P. Eckel, Dept of Preventive Medicine, University of Southern California, 2001 N. Soto Street, MC-9234, Los Angeles, CA 90089, USA. E-mail: eckel{at}usc.edu
Abstract
Mechanisms for the adverse respiratory effects of traffic-related air pollution (TRAP) have yet to be established. We evaluated the acute effects of TRAP exposure on proximal and distal airway inflammation by relating indoor nitric oxide (NO), a marker of TRAP exposure in the indoor microenvironment, to airway and alveolar sources of exhaled nitric oxide (FeNO).
FeNO was collected online at four flow rates in 1635 schoolchildren (aged 12–15 years) in southern California (USA) breathing NO-free air. Indoor NO was sampled hourly and linearly interpolated to the time of the FeNO test. Estimated parameters quantifying airway wall diffusivity (DawNO) and flux (J′awNO) and alveolar concentration (CANO) sources of FeNO were related to exposure using linear regression to adjust for potential confounders.
We found that TRAP exposure indoors was associated with elevated alveolar NO. A 10 ppb higher indoor NO concentration at the time of the FeNO test was associated with 0.10 ppb higher average CANO (95% CI 0.04–0.16) (equivalent to a 7.1% increase from the mean), 4.0% higher J′awNO (95% CI −2.8–11.3) and 0.2% lower DawNO (95% CI −4.8–4.6).
These findings are consistent with an airway response to TRAP exposure that was most marked in the distal airways.
Abstract
Indoor exposure to traffic-related air pollution is associated with distal airway inflammation in schoolchildren http://ow.ly/V98mT
Introduction
Traffic-related air pollution (TRAP) is an important urban exposure with adverse respiratory effects, but the mechanisms for these effects have yet to be established [1]. TRAP is a complex mixture of primary particulate matter and gaseous pollutants (e.g. nitrogen oxides (NOx)) [1]. Typical TRAP exposure metrics include location-based measures, such as distance from roadways, or measurements of specific components, such as NOx. In this article, we focus on nitric oxide (NO), a primary pollutant present in indoor and outdoor air, primarily from anthropogenic combustion sources (e.g. gas stoves and traffic) [2]. Once emitted, NO is rapidly oxidised and converted to nitrogen dioxide through photochemical processes or ozone (O3) titration [2]. Outdoor NO concentrations display strong seasonal, diurnal and day-of-week patterns, with peaks coinciding with rush-hours, and high spatial variability, with concentrations decaying rapidly away from roadway sources [2]. Indoor NO depends on outdoor NO, air exchange rates and indoor emission/removal rates [3, 4]. In the absence of indoor sources, indoor NO can be considered a marker of TRAP exposure in the indoor microenvironment [2].
The mechanisms of acute respiratory responses to TRAP exposure may be elucidated by studying the exhaled nitric oxide fraction (FeNO), a noninvasively assessed marker of lower respiratory tract inflammation endogenously produced in the airway and alveolar epithelium [5–7]. FeNO measured at the conventional 50 mL·s−1 exhalation flow rate (FeNO50) is responsive to both ambient air pollution (e.g. particles with a 50% cut-off aerodynamic diameter of 2.5 μm (PM2.5), 10 μm (PM10) and O3) and TRAP exposures [8–12]. When FeNO is measured at multiple flow rates (extended FeNO analysis), it can be partitioned into airway and alveolar sources, reflecting proximal airway inflammation and what is assumed to be distal inflammation, although confirmation by direct measurement is not available [7, 13].
Extended FeNO analysis can be used to study noninvasively the differential response of the proximal and distal airways to air pollution exposures. To our knowledge, there has been only one such study. Modig et al. [14] related low and high flow FeNO in adults to short-term averages of central site ambient air pollution. However, low and high flow FeNO are imperfect proxies for airway and alveolar NO [15] and it remains unknown whether short-term TRAP exposures have differential associations with proximal and distal lower respiratory tract inflammation in children. We hypothesise that there will be differential proximal and distal airway inflammatory responses to TRAP exposure, possibly due to ultrafine particles (UFPs), which are an important component of the TRAP mixture and have a larger deposition profile in the distal airways. Exposures to UFPs from traffic sources are difficult to assess directly. Measurement of NO concentrations, which are strongly correlated to traffic-related UFPs, offers a practical alternative method for exposure assessment [16].
FeNO concentrations are in the parts-per-billion range, so specific precautions, as described in the American Thoracic Society (ATS)/European Respiratory Society (ERS) recommendations [17] are required to avoid contamination with room air NO. For online measurement of FeNO at a fixed flow rate, inhalation of air with a high NO concentration produces an early peak in FeNO from mixed expiratory air, but the subsequent steady-state plateau concentration is thought to be unaffected by ambient NO concentrations due to the fast uptake of inhaled NO by haemoglobin in the pulmonary capillary blood [17, 18]. Inhalation of NO-free air for two or more tidal breaths before the FeNO test precludes the early peak. Earlier work in the southern California Children's Health Study (CHS) showed that offline FeNO was associated with room air NO, but online FeNO50 in the same children was not [19].
The goal of this study was to investigate the association between short-term exposure to TRAP and localised lower respiratory tract inflammation. Using cross-sectional data from the CHS, a population-based cohort study of schoolchildren, we examined the association of schoolroom indoor NO with no known indoor sources of NO (as a marker for exposure to TRAP in the indoor microenvironment) with airway and alveolar sources of NO estimated from online extended FeNO analysis in children breathing NO-free air.
Methods
Study population
Participants were children from a CHS cohort originally recruited in 2002–2003 from kindergarten or first-grade classrooms in southern California [20]. We included data from the 1635 children in eight communities who participated in the first year of extended FeNO assessment (from March to June 2010, with each community visited in two or three separate rounds) [21] and who returned the corresponding annual written questionnaire. Written informed consent was obtained from a parent/guardian on behalf of each child participant. The study protocol was approved by the University of Southern California Health Sciences Campus institutional review board.
Extended FeNO collection
FeNO was assessed online using three chemiluminescence analysers (model CLD88-SP with DeNOx accessory to provide NO-scrubbed air; EcoMedics, Duernten, Switzerland/Ann Arbor, MI, USA) in a protocol described previously [21–23]. Briefly, children were requested to perform nine FeNO manoeuvres, in the following order: three at the conventional 50 mL·s−1 target flow rate and two at each of the following target flow rates: 30, 100 and 300 mL·s−1. Immediately prior to each manoeuvre, the child breathed through a DeNOx scrubber for two or more tidal breaths followed by inhalation to total lung capacity and exhalation at the target flow rate. As described in greater detail in the online supplementary material, DeNOx zero checks were performed twice daily and FeNO data processing was based on the ATS/ERS guidelines for FeNO at 50 mL·s−1 [17] and an airway turnover search window [24]. FeNO50 and FeNO300 were calculated as the average of reproducible manoeuvres at 50 mL·s−1 and 300 mL·s−1, respectively.
Assessing indoor TRAP exposure using indoor NO measurements
FeNO was tested in empty schoolrooms with no known sources of NO. Indoor NO was measured approximately hourly, using the FeNO analysers, in the FeNO testing room. For each analyser, indoor NO values were linearly interpolated to the time of each participant's first FeNO manoeuvre and the interpolated value was used as the indoor TRAP exposure metric. Unanticipated sources of indoor NO were noted in field logs. We compared diurnal variation in test room NO to hourly ambient NO from central site monitors located in each community (online supplementary figs E1 and E2).
NO parameter estimation
The two-compartment model of NO in the lower respiratory tract [7] describes FeNO as a function of flow rate and three “NO parameters”: alveolar NO concentration (CANO), maximum airway wall NO flux (J′awNO) and airway wall tissue diffusing capacity (DawNO). As described in the online supplementary material, we estimated these NO parameters for each participant using a multilevel modelling extension [25] of the participant-specific nonlinear least squares model proposed by Eckel et al. [26]. For sensitivity analysis, we estimated NO parameters using the Högman and Meriläinen algorithm (HMA) [27] modified for CHS data [21, 26], which omits internal data checks found to bias NO parameter estimation [26].
Data analysis
For clarity of exposition, we collectively refer to the estimated NO parameters, FeNO50 and FeNO300 as “extended FeNO summaries”. We group them into summaries thought to reflect, to a greater extent, NO from proximal (FeNO50, J′awNO and DawNO) or distal (FeNO300 and CANO) airway sources. We related indoor NO exposure to each of the extended FeNO summaries using multiple linear regression models to adjust for potential confounders (age, sex, race/ethnicity, rhinitis history, ever report of doctor-diagnosed asthma, use of asthma medications in the past 12 months, secondhand tobacco smoke exposure, highest attained parental education, time of FeNO test, FeNO analyser and CHS community). To better satisfy modelling assumptions, FeNO50, J′awNO, DawNO and FeNO300 were natural log transformed. CANO was analysed without transformation. Sensitivity analyses included additional adjustment for longer-term time trends (either a four degrees of freedom natural cubic spline of test date or indicators of community-specific rounds of FeNO assessment) and various data exclusions. We evaluated evidence for modification of indoor NO exposure associations by asthma by including a multiplicative interaction term. We also considered single lag models with indoor NO lagged 0, 30 or 60 min and used the Akaike information criterion (AIC) [28] to select the best fitting model. All hypothesis tests were two-sided, with a significance level of 0.05. Analyses were performed using R version 3.1.0 [29].
Results
Participants were aged between 12 and 15 years old; more than half identified as Hispanic white (58.2%); most had some history of rhinitis (38.1% not current and 31.5% current); and 321 (19.6%) reported a physician diagnosis of asthma, of whom 48 reported taking inhaled corticosteroid medication in the past 12 months (table 1). The geometric means of FeNO50, J′awNO, DawNO and FeNO300 were 15.4 ppb, 876 pL·s−1, 18.8 pL·s−1·ppb−1 and 4.2 ppb, respectively. Mean CANO was 1.4 ppb. Table 2 shows that log FeNO50 was highly correlated with log J′awNO (Pearson's R=0.97), as has been recognised previously [26]. Notably, log FeNO300 was more highly correlated with proximal airway NO (R=0.93 for log J′awNO and R=0.96 for log FeNO50) than with alveolar NO (R=0.69 for CANO).
Indoor NO concentrations at the time of the FeNO test, our marker for indoor exposure to TRAP, ranged from 0.0 to 58.5 ppb (table 3). Most children (67.0%) had FeNO tested after 10:00 h, when indoor NO tended to be lower following the peaks from morning rush-hour and school drop-offs. On many testing dates, the diurnal patterns of indoor NO and outdoor NO at the community's central site monitor were similar, reflecting high (i.e. ∼1) outdoor-to-indoor infiltration rates and common meteorology (online supplementary figure E2). On March 17 and 18, 2010 in Mira Loma, the two dates with highest indoor NO, the diurnal patterns of schoolroom indoor NO and central site NO were similar.
We found robust evidence for an association of TRAP exposure in the indoor microenvironment, as assessed by indoor NO, with FeNO from distal sources (CANO and FeNO300). Associations were less consistent with FeNO from proximal sources. After adjusting for potential confounders (tables 4 and 5 and online supplementary tables E1 and E2), a 10 ppb higher indoor NO concentration was significantly associated with 0.10 ppb (95% CI 0.04–0.16 ppb) higher average CANO (equivalent to a 7.1% increase from the mean of 1.4 ppb to 1.5 ppb) and 6.5% (95% CI 0.3–13.1%) higher FeNO300 and nonsignificantly associated with 2.8% (95% CI −3.3–9.3%) higher FeNO50, 4.0% (95% CI −2.8–11.3%) higher J′awNO and 0.2% (95% CI −4.8–4.6%) lower DawNO. The CANO results were robust to various sensitivity analyses, including adjustment for longer-term time trends, exclusion of the aforementioned potentially influential dates with high indoor NO (Mira Loma; March 17 and 18, 2010) and using HMA estimation of NO parameters (tables 4 and 5 and online supplementary tables E3–E5).
Children with asthma had larger estimated effects of indoor NO for all extended FeNO summaries except DawNO (figure 1), although none of the differences in effects comparing children with asthma to those without were statistically significant (interaction p-values all ≥0.22).
Associations of lagged indoor NO (up to 1 h prior to the FeNO test) with CANO were estimated in the subset of 1212 children tested ≥1 h after the first test of the day for each analyser. Since indoor NO was only measured during FeNO testing, the 423 children tested within the first hour were excluded. The best-fitting model, by minimum AIC, used a 60 min lag (table 6) and estimated that a 10 ppb higher indoor NO concentration 60 min prior was associated with 0.12 ppb (95% CI 0.05–0.19 ppb) higher average CANO. Considering lags up to 2 h (and hence restricting the analysis even further to the n=811 children tested ≥2 h after the first test) produced slightly attenuated effect estimates and found that a model with 30-min lag had the smallest AIC (online supplementary table E6).
Discussion
This study investigated the association between short-term TRAP exposures in the indoor microenvironment, assessed using indoor NO, and regional lower respiratory tract inflammation based on extended FeNO analysis, in a large population-based cohort of schoolchildren. We found strong evidence for a positive association of indoor exposures to TRAP with FeNO from distal sources (CANO and FeNO300) in schoolchildren. Lagging indoor NO by 30–60 min slightly improved model fit, providing biological plausibility for our hypothesis of an acute distal airway inflammatory response. There was some (nonsignificant) evidence to support an association of indoor exposures to TRAP with FeNO from proximal sources (FeNO50 and J′awNO, but not DawNO) in children with asthma, similar to previous studies of TRAP and FeNO50 [11]. We speculate that the observed pattern of associations may potentially be due to higher deposition of UFPs, a toxic component of TRAP, typically strongly correlated with NO, in distal versus proximal airways. We did not measure traffic-related or total UFPs in our study, so this should be investigated in future work.
To our knowledge, extended FeNO analysis has been used in no studies of TRAP and in only one study of the respiratory health effects of ambient air pollution in adults, by Modig et al. [14]. Modig et al. [14] related FeNO50 and FeNO270 measured in 5841 adults to short-term averages of measurements of O3, NOx, and PM10 from a central site location. In multipollutant models, the authors found significant associations of both 120-h average O3 and 24-h average NOx with distal airway NO and less evidence for associations with proximal airway NO [30]. Specifically, an interquartile range (IQR) increase in O3 was associated with a 5.1% (95% CI 1.7–8.5%) increase in FeNO270 and a 3.6% (95% CI −0.1–7.5%) increase in FeNO50, while an IQR increase in NOx was associated with a 1.4% (95% CI 0.1–2.8%) increase in FeNO270 and a 0.2% (95% CI −1.3–1.7%) increase in FeNO50 [14]. In our data, log FeNO300 was more highly correlated with airway wall NO than with alveolar NO, highlighting the problem of using high-flow FeNO as a proxy for alveolar NO [15]. We expect that Modig et al. [14] would have observed clearer, more differential associations had they conducted extended FeNO modelling to partition multiple flow FeNO into proximal (J′awNO) and distal (CANO) sources.
TRAP has been related to FeNO at a single exhalation flow rate in other studies, many having larger exposure contrasts than our study. From repeat tests on 18 subjects breathing NO-free air, offline FeNO was on average 73% higher (p<0.001) on days when outdoor NO was high (246 μg·m−3) versus low (3.6 μg·m−3) [31]. In 29 nonsmoking elderly subjects breathing NO-free air, an 11.9 ppb increase in 24 h average central site NO (maximum 70.7 ppb) was associated with a 0.83 ppb (95% CI 0.26–1.40 ppb) increase in offline FeNO [32]. A crossover study of 28 healthy adults walking for 2 h in low and high traffic pollution sites (at the high traffic site NOx, black carbon and UFP number were 5–10-fold higher and highly correlated (>0.9)) found that exposure to high TRAP was associated with a 0.89 ppb increase in FeNO, with the maximum mean increase in FeNO at 30 min post-exposure [33]. In 57 adult asthmatics followed for 6 months, daily mean ambient UFP number concentration and NO were strongly correlated (Spearman's R 0.73–0.77) and UFP number concentrations had the strongest negative associations with daily peak expiratory flow measurements [34]. Particularly relevant to our motivating hypothesis, a study of 103 schoolchildren found strong associations of personal daily alveolar deposited UFP surface area with forced expiratory flow at 25–75% of forced vital capacity (negative association in all children, R2=0.1) and FeNO50 (positive association, only in the 16 children with asthma, R2=0.9) [35].
Strengths of this study include the large population-based cohort of schoolchildren, the online measurement of FeNO at multiple flow rates using sensitive chemiluminescence analysers [36] by trained field technicians using a rigorous protocol requiring two or more tidal breaths through a DeNOx scrubber (zero-checked twice daily) prior to each manoeuvre [17], and the use of a highly temporally and spatially resolved exposure metric for TRAP (indoor NO). Statistical estimation of NO parameters allowed for better separation of proximal and distal sources of NO than average FeNO from low and high target flow rates [15]. The Eckel et al. [26] method for estimating NO parameters has been shown to have good statistical properties and the mixed-effects modelling extension of this approach [25] borrows strength across participants to produce NO parameter estimates for all participants. Participant-specific models typically fail to produce NO parameter estimates for a subset of participants. Exclusion of these participants from subsequent analyses could bias results. Finally, the observed CANO association was robust in a wide array of sensitivity analyses.
This study has several limitations. The two-compartment model of NO assumes inhalation of NO-free air and that blood is an infinite sink for NO [37], ignoring any possible change in the production or consumption rates of NO in the tissue or in the ability of NO to bind to haemoglobin due to prolonged inhalation of air with elevated NO concentrations. However, in our study children breathed NO-scrubbed air and indoor NO was relatively low (<58.5 ppb). Additionally, indoor NO measurements are an imperfect surrogate for exposure to TRAP. Indoor NO measurements reflect the indoor environment where children spend time and are more spatially and temporally resolved than many common TRAP exposure metrics (e.g. central site measurements or distance to nearest major road), but they are not measurements of personal exposure. In our study, we expect that outdoor-to-indoor infiltration rates were high (i.e. ∼1). A typical testing session was conducted in an otherwise empty classroom, with no indoor source of NO, that opened to the outdoors (as is common in southern California), with the door left propped open to welcome the next child to be tested. Finally, because indoor NO was assessed by FeNO analysers only during testing sessions, lagged indoor NO (e.g. a 1-h lag) was available only in a subset of children, resulting in reduced power to detect an association due to reduced sample size and reduced exposure variability (excluded children were tested earlier, when NO tended to be highest).
Our results should also be interpreted in the context of the considerable body of literature investigating whether room air NO contaminates FeNO samples. All studies that failed to find an association used online FeNO measurement [18, 38–41]. Studies that found an association used offline FeNO sampling methods in participants breathing room air (e.g. as in [42]) or early online methods for FeNO measurement (e.g. as in [43]). In the first year of online FeNO measurement in the CHS, 386 children provided both online FeNO (breathing through a DeNOx scrubber) and offline FeNO (breathing through the offline test kit NO scrubber, with deadspace air discarded) [19]. We found that online FeNO was not associated with test indoor NO (Pearson's R=0.09, p=0.08) while offline FeNO was (R=0.30, p<0.0001), primarily due to incomplete removal of NO by the offline scrubber and residual contamination [19].
In conclusion, TRAP exposure in the indoor microenvironment, assessed using indoor NO, was associated with exhaled NO from alveolar sources in schoolchildren. CANO concentration differences were small but robust, consistent with an acute distal airway inflammatory response to TRAP exposures. Future studies could refine exposure assessment (e.g. personal exposure measurements) and study a larger sample of children with asthma.
Acknowledgements
The authors gratefully acknowledge the contributions of Children's Health Study participants and the exhaled nitric oxide fraction field team: Steve Howland, Ned Realiza and Lisa Valencia (Dept of Preventive Medicine, University of Southern California, Los Angeles, CA, USA).
Footnotes
Editorial comment in: Eur Respir J 2016; 47: 1304–1306.
This article has supplementary material available from erj.ersjournals.com
Support statement: This work was supported by the National Heart, Lung and Blood Institute (grants 5R01HL61768 and 5R01HL76647); the Southern California Environmental Health Sciences Center (grant 5P30ES007048) funded by the National Institute of Environmental Health Sciences (NIEHS); the Children's Environmental Health Center (grants 5P01ES009581, R826708-01 and RD831861-01) funded by the NIEHS and the Environmental Protection Agency; the NIEHS (grants 5P01ES011627, 1R01ES023262-01 and 1K22ES022987); the James H. Zumberge Research and Innovation Fund; and the Hastings Foundation. Funding information for this article has been deposited with FundRef.
Conflict of interest: Disclosures can be found alongside the online version of this article at erj.ersjournals.com
- Received July 20, 2015.
- Accepted November 22, 2015.
- Copyright ©ERS 2016