The present study indicates that a decline in lung function was associated with increased exposure to sulfate, ammonium, and OC fractions of PM2.5, but not with PM2.5 mass, in the multiyear panels of children with asthma who resided in two urban areas in Nagasaki prefecture, Japan. We observed that concentrations of the significant chemical species were higher in days largely contributed by the path of Northeast China origin (for sulfate and ammonium) or both the same path and local sources (for OC) than by other clusters, as suggested by the back-trajectory and cluster analyses.
Existing evidence about the differential effects of PM2.5 fractions on children’s lung function has been inconclusive, although it appears more evidence suggests that EC and black carbon are associated with asthma exacerbations among children . We found no evidence for EC and nitrate in this study; however, our findings were consistent with those of some previous studies. For example, exposure to sulfate, ammonium, EC, and/or OC fractions of PM2.5 was adversely associated with pediatric emergency department visits for asthma and wheezing , and upper respiratory infections  among children in the United States. A panel study in Southern China also reported an association between OC and lung function decline in healthy school children . Our previous study in the same city (Isahaya only) for adults with asthma showed consistent findings that severe respiratory decline was associated with increased sulfate and OC fractions of PM2.5 .
Different environmental conditions, characterized by geographical terrain, seasonal features (e.g., monsoon and seasonal prevailing wind), or major sources of air pollution that could potentially influence the fractions of PM2.5, might magnify the effects of certain chemical species on health for each study area. The secondary pollutants (sulfate and ammonium) and the mixed basis (OC) in our study were found to be associated with lung function decrements among children with asthma. We speculate that the effects of transboundary air pollution in our study area played a role in the negative associations in the particular season (spring) because relatively large contributions to the path from Northeast China to Nagasaki city were estimated for those chemical species by back-trajectory and cluster analyses. A previous study conducted in Fukuoka and Fukue Island, both located approximately 100 km away from the city of Nagasaki, also reported consistent observations from 2010 to 2013 that sulfate and OC were the largest contributing components to PM2.5, and greatly dominated by the inflow of long-range transported aerosols . A previous study on children with asthma in Nagasaki prefecture (the same but larger panel as the present study without PM2.5 composition sampling) also reported PEF reduction with transboundary air pollution, represented by AD events detected by light detection and ranging and suspended particulate matter . Consistent evidence for the AD events associated with pediatric emergency department visits due to bronchial asthma was also reported for school children during the spring in Nagasaki prefecture .
Given the anticipation of the potential effects of long-range transboundary air pollution in our study areas, we initially hypothesized that the results from the two neighboring cities of this study would be identical enough to be combined. However, the estimated trajectories representing prevailing paths and clusters for the PM2.5 mass and chemical species were different between the cities (i.e., the largest contribution from the path of Northeast China origin in Nagasaki city, whereas almost even contributions from the two paths of Northeast China and local origins in Isahaya city). The chemical species of PM2.5, which were associated with declines in lung function, also differed by city (i.e., sulfate and ammonium in Nagasaki city and OC in Isahaya city). Several possible explanations may support this discrepancy. We hypothesize that the different elements of topography between the two cities might result in different patterns of estimated trajectories. Nagasaki city, a large traditional seaport, is adjacent and widely open to the East China Sea, where the westerly wind could constantly flow into the city with no major obstacles (Figure S1). In contrast, Isahaya city is located inside Nagasaki prefecture, where a range of mountains could obstruct the influence of the westerly wind and long-range transport air pollution. Other possibilities exist regarding the different health risks estimated between cities. Although we attempted to adjust for potential individual and time-varying confounders in the models, some other factors related to different susceptibility in children between cities might play a role in modifying the health risks that we could not fully consider in this study. In addition, the different sources of the exposure data might partly play a role in the discrepancy, although a comparable method of sampling and filters were used.
The biological plausibility of the short-term effects of PM2.5 on asthma exacerbation has been well described, coupled with the potential pathways through respiratory tract injury, airway inflammation, oxidative stress, airway hyper-responsiveness, allergic sensitization, or activation of sensory nerves in the respiratory tract [1, 21]. We believe that the same mechanisms could apply to the adverse effects of sulfate, ammonium, and OC fractions of PM2.5, on lung function in children with asthma. Furthermore, our findings of the exposure-response associations for these fractions provide evidence that even low levels of the components of PM2.5 could trigger airway inflammation. In fact, the annual PM2.5 levels in our study areas have been relatively low and observed around the Japanese National Ambient Air Quality Standard (NAAQS) (annual mean of 15 µg/m3 and daily mean of 35 µg/m3), with a small number of days exceeding the daily standard (i.e., less than 5% for a year during our study periods) .
Nevertheless, our results should be interpreted with caution from a clinical perspective because it is arguable whether the PEF reduction of 1–2 L/min per IQR increase in the PM2.5 fractions could provide clinical implications for children with asthma. In addition, we observed that the odds of asthma worsening episodes associated with PM2.5 were fairly sensitive to cut-offs to define an episode. In particular, the estimated odds based on the cut-off of 20% PEF reduction were counterintuitive and inconsistent with those for another cut-off of 15% PEF reduction. We believe that the unexpected results for the cut-off of 20% may be due to well-guided self-management of the asthma action plan by the children with asthma and their parents enrolled in our study. Given that the PEF reduction > 20% for more than two days has been recommended by the Global Initiative for Asthma as a clinically important criterion for initiating the use of controller medication , our findings for the less strict cut-off of 15% PEF reduction suggest that the fractions of PM2.5 could be implicated in the subclinical status of asthma worsening among children.
A major strength of this study was that we obtained a large number of daily PEF measurements over three years in the spring, while most previous panel studies investigating the associations with PM2.5 exposures have been designed to collect lung function measurements over multiple days or a few months [24, 29–32]. The large number of PEF measurements in our study could increase the statistical power to detect a smaller effect size for the exposure-response associations among children with asthma, who are one of the most vulnerable populations to air pollution exposure. In addition, given varying levels of PM2.5, the inclusion of the three years of data in our study could lead to a more generalizable conclusion for the associations in the study areas.
This study had several limitations. First, we collected daily self-measured PEFs, which were not measured under the supervision of medical staff. Although we trained the children with asthma and their parents and excluded the training periods (used to become accustomed to the measuring device) from our analyses, the measurement error might be greater than that under supervision. In addition, the self-measurement performance could naturally be improved over time through daily repetition. We included the day of the year in the models to adjust for the possible time trends of the self-measured PEFs. Second, we collected ambient PM2.5 exposures derived from fixed-site monitors, but not personal exposures. Some previous studies have reported weaker associations between children’s lung function and ambient air pollution exposure, compared to those for personal exposures presumed to be a superior method of exposure assessment [29, 33]. However, this possible misclassification in our study would be non-differential, leading to bias towards the null . Third, we fitted the models including a single chemical species with no adjustment for other species, although the associations for each chemical species were estimated after adjusting for PM2.5 total mass. Lastly, our study did not take into account seasonal aeroallergens such as pollen because of data unavailability, although pollen could potentially be associated with asthma worsening, independently or synergistically with fine particulate matter .