Ozone Reaction Products Associated with Biomarkers of Cardiorespiratory Pathophysiology

To the Editor:

Ground-level ozone (O₃), a regulated air pollutant in many countries, contributes significantly to the global burden of disease (1). People are advised to stay indoors when outdoor O₃ concentrations are elevated. This conventional wisdom implies that a substantial fraction of O₃ disappears as if into a black hole during outdoor-to-indoor transport. However, O₃ loss results in the generation of reaction products (2, 3). Hence, building inhabitants inhale not only O₃ but also airborne products derived from reactions that consume O₃. Some of these products (e.g., formaldehyde, methacrolein, and organic peroxides) are known to be toxic or irritating, although short-term exposures of humans to O₃ oxidation products in a chamber study failed to compromise respiratory function or neurobehavior performance (4, 5).

Ozone reaction products are present as a complex mixture of both gaseous and particulate species, and neither single nor classes of chemical species can be readily measured in a real-world setting to define this mixture. Here we present a novel approach that uses O₃ loss as a proxy for O₃ reaction products, calculated by subtracting indoor concentration from outdoor concentration at a given time. As inferred by theory (2) and demonstrated through measurements (3), O₃ loss is proportional to the net concentration of gas-phase O₃ products. Given that O₃ may be too reactive to reach the distal lung, whereas O₃ reaction products can, we hypothesized that O₃ loss exposure is adversely associated with more biomarkers of pulmonary and cardiovascular pathophysiology than O₃. Some of the results of this study have been previously reported in the form of an abstract (6).

Methods

Biomarkers and relevant exposure data were obtained from two panel studies, namely the Adults study (89 healthy adults, 22–52 yr old) and the Children study (43 children with asthma, 5–13 yr old) (7, 8). Each participant was repeatedly measured four times for cardiorespiratory biomarkers. Measured outdoor and indoor concentrations of O₃ and particulate matter ⩽2.5 μm in aerodynamic diameter (PM_2.5), coupled with time–activity data, were used to calculate personal O₃ exposure, O₃ loss exposure, and PM_2.5 exposure averaged over 12 hours, 24 hours, and 2 weeks before biomarker measurements. We also calculated 24-hour average personal exposures 1–2 days before biomarker measurements. Statistical analyses were conducted separately for the two studies. We used linear mixed-effects regression models to examine the associations of a biomarker with personal O₃ or O₃ loss exposure. Copollutants were PM_2.5 and O₃ loss in the analysis of O₃ effects and PM_2.5 and O₃ in the analysis of O₃ loss effects. Fixed-effects covariates were specific to each study and are shown in the footnotes of Figures 1 and 2.

Figure 1.

The percent change in adverse effects (increase or decrease) associated with a 10-ppb increase in ozone (O₃) exposure (controlled for particulate matter ⩽2.5 μm in aerodynamic diameter (PM_2.5) and O₃ loss exposure) and a 10-ppb increase in O₃ loss exposure (controlled for PM_2.5 and O₃ exposure) in the Children study. Only biomarkers that show a statistically significant association (P value < 0.05) with 12-hour, 24-hour (at any of lag days), or 2-week average O₃ or O₃ loss exposure are shown. If multiple significant or nonsignificant associations were found, the associations with the smallest P value are shown in the figure. Significant and adverse associations between biomarkers and O₃ loss exposures: FeNO (fractional exhaled nitric oxide) (12-h, 24-h, lag Day 2, 2-wk), Z₅ (24-h), X₅ (24-h), Fres (24-h, lag Day 1), FEV₁/FVC: (12-h). A positive value (y-axis) indicates an adverse effect, and a negative value indicates a beneficial effect. Fixed-effect covariates: Sex, age, ambient temperature and humidity, baseline eosinophil number (/μl), upper airway tract infection-like symptoms (binary), inhaled corticosteroid usage (binary), asthma exacerbation (binary). A sensitivity analysis for FEV₁/FVC: and FeNO by adjusting for height and weight showed similar results. Participant baseline characteristics: Age (mean [SD] = 7.8 [2.3], range = 5–13, and unit = years), sex (17 males and 26 females), eosinophil count (mean [SD] = 379 [265], range = 80–1260, and unit = /μl). CI = confidence interval; Fres = resonant frequency; MDA = malondialdehyde; X₅ = reactance at 5 Hz; Z₅ = impedance at 5 Hz.

Figure 2.

The percent change in adverse effects (increase or decrease) associated with a 10-ppb increase in ozone (O₃) exposure (controlled for particulate matter ⩽2.5 μm in aerodynamic diameter (PM_2.5) and O₃ loss exposure) and a 10-ppb increase in O₃ loss exposure (controlled for PM_2.5 and O₃ exposure) in the Adults study. Only biomarkers that show a statistically significant association (P value < 0.05) with 12-hour, 24-hour (at any of lag days), or 2-week average O₃ or O₃ loss exposure are shown. If multiple significant or nonsignificant associations were found, the associations with the smallest P value are shown in the figure. Significant and adverse associations between biomarkers and O₃ loss exposures: EBC NN (nitrate and nitrite in exhaled breath condensate) (12-h, lag Day 2), FeNO (fractional exhaled nitric oxide) (lag Day 1, lag Day 2, 2-wk), 8-OHdG (8-hydroxy-2′-deoxyguanosine) (12-h, lag Day 2), 20-HETE (12-h, lag Day 1, lag Day 2, 2-wk), P-selectin (12-h, lag Day 1, lag D 2), DBP (24-h, lag Day 1, lag Day 2, 2-wk), SBP (12-h, 24-h, lag Day 1, lag Day 2, 2-wk). A positive value (y-axis) indicates an adverse effect, and a negative value indicates a beneficial effect. Fixed-effect covariates: sex, age, ambient temperature and humidity, time since last eating (h), secondhand smoke exposure (h), current smoking status (binary), urinary 6-sulfatoxymelatonin (ng/ml, as a surrogate of circulating melatonin, is adjusted when EBC NN, EBC MDA, or 8-OHdG is the health outcome). A sensitivity analysis for FeNO by adjusting for height and weight showed similar results. Participant baseline characteristics: age (mean [SD] = 31.5 [7.6], range = 22–52, and unit = years), sex (64 males and 25 females), current smoker (n = 15). 20-HETE = 20-hydroxyeicosatetraenoic acid; AI = augmentation index; CI = confidence interval; DBP = diastolic blood pressure; MDA = malondialdehyde; PWV = pulse wave velocity; SBP= systolic blood pressure; VWF = von Willebrand factor.

Results and Discussion

The associations of the biomarkers with O₃ and O₃ loss exposure are summarized in Figure 1 for children and Figure 2 for adults. We found that O₃ loss exposure was more likely to be adversely associated with biomarkers of the lung (lower airway) and cardiovascular pathophysiology. Specifically, in children with asthma, we only observed an adverse effect of O₃ exposure (but not O₃ loss exposure) on nasal malondialdehyde (oxidative stress in the nasal cavity), whereas O₃ loss exposure (but not O₃ exposure) was significantly and adversely associated with fractional exhaled nitric oxide (airway inflammation), Z₅ (airway impedance), X₅ (airway reactance), Fres (airway reactance), and FEV₁/FVC: (airflow obstruction risk). Similarly, we found significant and adverse effects of O₃ loss exposure (but not O₃ exposure) on nitrate and nitrite in exhaled breath condensate (pulmonary oxidative stress), fractional exhaled nitric oxide, 8-OHdG (systemic oxidative stress), 20-HETE (vasoconstriction), diastolic blood pressure, and systolic blood pressure in adults. On the contrary, O₃ exposure was more strongly and adversely associated with P-selectin and von Willebrand factor (endothelial dysfunction) than O₃ loss exposure. The underlying mechanisms of how O₃ exposure induces P-selectin and von Willebrand factor in the circulatory system are still unclear. Our findings can be explained with biological plausibility from a toxicokinetic standpoint. Ozone may be too reactive to reach the distal lung and exert a direct acute effect on the distal lung and the circulatory system (9). Some O₃ reaction products, including those in ultrafine particles, can reach the alveolar region and potentially enter the circulatory system (10). These O₃ reaction products tend to have larger oxygen-to-carbon ratios and larger water solubilities than their precursors, which translate to larger partitioning to lung fluid and longer lifetimes in the lung (11).

The findings from the present analysis have important implications for cardiopulmonary health, considering that the current O₃ epidemiologic literature is far less consistent than the PM_2.5 literature. Ozone is reactive, and a large fraction is lost when entering the indoor environment in which people spend most of their time. Ozone loss generates reaction products (2, 3). In addition, people may not be aware that for the same indoor O₃ concentration, O₃ reaction products can be different in concentration, depending on the co-occurring outdoor O₃ concentration. This may cause different degrees of confounding from O₃ reaction products in various O₃ effects studies. We provide the first real-world evidence that O₃ loss, which is proportional to the net concentration of inhalable O₃ reaction products, was associated with biomarker concentration changes suggestive of adverse cardiorespiratory health effects. This approach can be used to disentangle the health impacts of O₃ and its oxidation products. Our findings challenge the conventional wisdom that indoor O₃ concentrations are too low to be a concern, as people may be affected by both the residual O₃ and O₃ loss indoors. The current analysis is also timely, given that tropospheric O₃ concentrations are increasing globally because of climate change-facilitated O₃ formation.

Limitations

We observed a high correlation (coefficient: 0.8–0.9) between O₃ and O₃ loss in the Adults study but not in the Children study (coefficient: −0.2 to 0.4), reflecting that the Adults study was conducted in dorms and offices equipped with central heating, ventilation, and air conditioning systems, whereas the Children study was performed in diverse residences. This high correlation in the Adults study and unmeasured confounders may have distorted our associations, leading to seemingly beneficial effects of O₃ and O₃ loss exposures for some biomarkers. Our exposure assessment was on the basis of time–activity data and pollutant concentrations, mainly in residences and offices (Adults study). We recommend addressing these limitations in future studies with larger sample sizes, wider age ranges, and improved exposure assessments. The incomplete overlap between the health outcomes in the Children and Adults studies limits a direct comparison. On the other hand, our findings were strengthened by observing O₃ loss associations with more biomarkers of different yet interconnected pathophysiologic processes.