The Effects of Residential Greenness and Air Pollution on Oxidative Stress Levels in Urban and Peri-urban Residents of Beijing

Background and Aims Exposure to air pollution has been associated with increased risks of cardiopulmonary diseases, cancer, and mortality. Simultaneously, greenspace has also documented to be protective of mortality. Oxidative stress may be an intermediate biomarker in these processes. There has been little investigation on the effects of residential greenness and air pollution on oxidative stress. There are two aims of this study: 1) To explore the association of personal and ambient air pollution exposure with urinary oxidative stress levels; 2) To investigate the association of quantified-contemporary greenness with urinary oxidative stress levels and the interaction between greenness and air pollution exposure in affecting oxidative stress levels, respectively.

Methods In an existing panel study named AIRLESS (Effects of AIR pollution on cardiopuLmonary disEaSe in urban and peri-urban reSidents in Beijing), 123 residents living in an urban district (Haidian district) and 128 residents in a peri-urban district (Pinggu) of Beijing participated in the study. All participants were non-smokers, ≥ 49 years of age, and included 110 men and 140 women. Personal and ambient exposures to air pollutants were assessed for each participant during winter 2016 and summer 2017, respectively. Each participant was instructed to carry a validated personal air monitor (PAM) to measure particulate matters (PM1, PM2.5, and PM10), nitrogen dioxide, carbon monoxide, and ozone concentrations at a high spatiotemporal resolution seven consecutive days in each sampling season. We calculated contemporaneous green space coverage level by the average daily satellite-derived Normalized Difference Vegetation Index (NDVI) in the zone with 500m*500m grids by Google Earth Engine (GEE) using the Moderate-resolution Imaging Spectroradiometer (MODIS) dataset from NASA (National Aeronautics and Space Administration). We used the coordinates of the ambient air pollution monitoring stations and monitoring dates to match the NDVI data. Multiple oxidative stress biomarkers were measured, including urinary free malondialdehyde (MDA), urinary total MDA, and urinary 8-hydroxydeoxyguanosine (8-OHdG). All biomarkers were normalized by urinary creatinine in statistical analyses. Due to the right skewness, all biomarkers data were ln-transformed in the aim-specific analyses. (1) The association of personal and ambient air pollution exposure with the percent change of urinary oxidative stress biomarkers was estimated using linear mixed-effects models and the distributed lag linear model was used to investigate daily air pollutants’ hysteresis effects on the percent change of urinary oxidative stress biomarkers. (2) The association between tertiary NDVI with urinary oxidative stress biomarkers was estimated using linear mixed-effects regression and subgroup analysis was used to test the robustness of the association between quantified NDVI with the mean percent change of oxidative stress biomarkers using the linear mixed-effects model in different groups. (3) To explore the interaction between greenness and air pollution in affecting oxidative stress by each of these exposure variables, stratified analyses were conducted to examine whether air pollution exposure modifies the effect of greenness and whether greenness modifies the effect of air pollution. Tertiles of NDVI, personal PM2.5 exposure, and personal ozone exposure were used in these analyses.

Results We found positive associations of CO and ozone personal exposure, respectively, with percent change of the three oxidative stress biomarkers. The association tended to be significant only in the ozone model with the percent change of 8-OHdG [8.69% 95%CI: (2.98,14.39), p-value=0.004]. However, in the models of ambient air pollution, some non-significantly negative associations were observed. Consistent positive associations of ambient lag 1- and 2- day CO exposure with the percent change in levels of each of the three oxidative stress biomarkers were weakly persisted. However, the positive associations remained significant between CO exposure and total MDA (p value=0.033) only in lag2-day. In the analyses of greenness as the exposure variable, we observed that individuals who lived in greener areas tended to have lower levels of oxidative stress. Participants in the highest NDVI tertile (0.36-0.83) had significantly lower free and total MDA levels, mean and (95%CI) by -20.21% (-37.84%, -1.30%) and -17.77% (-32.89%, -2.16%), respectively, compared to the lowest NDVI tertile (0.11-0.25) (p-value =0.028). In the urban area, we found significant negative associations of NDVI with free MDA (p-=0.003), total MDA (p =0.005), and 8-OHdG (p=0.022), but not in the peri-urban area. In the modification (interaction) analyses, we observed negative estimates of quantified NDVI associated with each of the three biomarkers in the low personal ozone exposure group (Ozone≤18.7 ppb). We also observed negative estimates of quantified NDVI associated with free and total MDA in the low PM2.5 exposure group (PM2.5≤32μg/m3). In addition, we observed significant effects of personal ozone exposure on 8-OHdG [17.77 95%CI: (8.04, 27.56), p value=0.010]; personal CO exposure on free MDA [12.48 95%CI: (3.15, 21.85), p value=0.012 and total MDA [9.06 95%CI: (1.28, 16.64), p value=0.021] only in participants falling in the lowest NDVI tertile; and the positive associations were no longer significant in participants with higher tertiary NDVI.

Conclusion The protective effects of greenness on oxidative stress, especially in urban residents, elucidates the importance of green space in the urban built environment. Additionally, the adverse effects of air pollution exposure on oxidative stress indicates the noteworthiness of personal protection against air pollution exposure in urban residents.