Are Health Effects of Ambient PM2.5 Proportional to its Mass? Relevance of Toxicity Measurements in Predicting PM2.5 Health Effects

SUDHEER SALANA, Zhuying Dai, Haoran Yu, P. S. Ganesh Subramanian, Vishal Verma, University of Illinois at Urbana-Champaign

     Abstract Number: 96
     Working Group: Health-Related Aerosols

Abstract
Most epidemiological models use PM mass as a metric to predict health effects. However, there are several issues with such models. First, PM mass does not capture spatiotemporal variations in PM chemical composition. Second, the relationship between PM mass and its toxicity may also suffer from spatiotemporal heterogenity and hence the results of epidemiological models may not be extrapolated from one region to another. Therefore, alternative metrics are needed that can comprehensively represent PM health effects. In this study, we demonstrated the relevance of using PM toxicity measurements over PM mass using a large number of PM_2.5 samples (n>350) collected from fourteen different sites across four different continents (Asia, Europe, North and South America) and evaluating five PM toxicity surrogates [3 acellular oxidative-potential (OP) endpoints – OP_DTT, OP_GSH, OP_OH-SLF; 2 cellular endpoints cell-viability using Crystal-Violet assay and cellular-ROS using Dichlorofluorescein-diacetate (DCFH-DA) in a human-lung-epithelial cell-line (A549)]. Our results show that the variation in mass concentrations [Coefficient-of-Variation (CoV)~20-40%] within a site is much smaller than the variation in extrinsic toxicity measurements (CoV~40-100%) and the concentrations of different PM chemical species (CoV~25-100%). The relationship between mass and extrinsic toxicity varied spatially. Samples collected in UK and Chile showed a strong correlation (r>0.7) between mass and almost all the extrinsic toxicity endpoints. However, none of the extrinsic toxicity endpoints had a strong correlation (r<0.5) with PM mass in the samples collected from India and Midwest USA. Even the sites displaying high correlation between toxicity and mass, showed variable slope (toxicity vs. mass) indicating differential levels of intrinsic toxicity among these sites. For example, the average PM_2.5 mass concentrations were highest in India, but the intrinsic PM toxicity was lower than for most of the sites. Similarly, PM_2.5 mass concentrations in Midwest USA and Atlanta were much lower (~4-20 times) than those in Chile and India, but the intrinsic PM toxicities was comparatively high (~2 times). These results emphasize the need for alternative metrics of PM, to better represent PM induced health effects. OP could be one of such metrics, but more worldwide studies (such as the current one) are needed to understand its spatiotemporal distribution and confirm its health relevance based on epidemiological studies.