Identification of potential markers for internal exposure to ambient ozone in oral cavity of healthy adults.

Methods
We aimed to evaluate the feasibility and suitability of the markers in oral microbiome for ambient ozone exposure. Between March and May in 2018, 97 healthy adults were examined on 2 or 3 occasions for oral swab sampling. Hourly concentrations of ambient ozone 1–7 days preceding sampling were collected. Mixed-effect models were fitted to examine the associations between ambient ozone and the diversity and taxon abundances of oral microbiome. Receiver operating characteristic (ROC) curves estimated the accuracies of markers to delineate between samples exposed to different concentrations of ambient ozone. The associations between the makers and lung function were further examined by linear mixed effect models.

Results
The averages of daily mean concentrations of ambient ozone (O3-daily), maximum 8-h means (O3-8hmax) and 1-h maximums (O3-1hmax) were respectively 72 μg/m³, 123 μg/m³ and 144 μg/m³. O3-daily was positively associated with α-diversity of oral microbiome, but the exposure-response curves only yielded positive associations in the range of O3-daily from 60 μg/m³ to 75 μg/m³. Results of O3-8hmax and O3-1hmax were consistent with these of O3-daily. With an interquartile range increase in O3-daily at lag04, the abundance of Proteobacteria decreased by 3.1% (95% CI: −4.0%, −2.2%) and Firmicutes increased by 3.3% (95% CI: 2.3%, 4.3%), whilst the Proteobacteria:Firmicutes ratio (P/F) decreased by 0.9 (95% CI: −1.5, −0.4). The areas under ROC curves for Proteobacteria, Firmicutes and P/F were 0.8535, 0.7569 and 0.8929, respectively. Proteobacteria and P/F were associated with forced expiratory volume in the first second and fractional exhaled nitric oxide significantly.

Conclusion
Ambient ozone disturbs oral microbial homeostasis. Proteobacteria, Firmicutes and their ratio may be potential markers for short-term ambient ozone exposure, and indicators of airway inflammation or lung function decline.