The study provided a comprehensive analysis on the short-term effects of various PM2.5 chemical constituents (including carbonaceous fractions, ions, crustal metals, transition metals, and other metals/metalloid elements) on inflammation, coagulation and homocysteine simultaneously for the first time in the context of traffic-related air pollution. Our study is mainly strengthened by the repeated-measures study design among a panel of healthy college students before and after their natural relocation from a suburban area to an urban area with changing air pollution contents. This allowed the study subjects to serve as their own controls well. Furthermore, the air pollution levels showed significant variations over the study and therefore the relations of air pollutants with circulatory biomarkers could be well investigated over wide exposure ranges.
In the present analysis, we selected an informative set of circulatory biomarkers which can predict cardiovascular disease risk [23–28]. Relations of these biomarkers with cardiovascular diseases have been well established, and changes in these biomarkers also have been associated with short-term air pollution exposure. However, results in this field are not consistent. For example, a previous panel study under urban exposure setting found significant positive effects of particles on hs-CRP, fibrinogen and PAI-1 but not on t-PA in young adults ; a crossover study did not find any significant differences in inflammatory biomarkers (CRP, fibrinogen and TNF-α) after exposure to particle rich air in young healthy individuals ; a population-based survey found a positive association between vWF and short-term PM10 exposure  whereas a panel study in patients with chronic pulmonary disease found inverse associations between vWF and various air pollutants including particles and gases ; panel studies by Delfino et al. [14, 15] found significant positive associations between sP-selectin and particles and its carbonaceous contents whereas ours did not (Figure 1). In addition, a previous controlled exposure study found that exposure to concentrated ambient fine and ultrafine particles—low in combustion-derived particles—did not affect fibrinolytic function (denoted by changes in t-PA and PAI-1 levels) in either middle-aged healthy volunteers or patients with coronary heart disease . This was in contrast with previous exposures to diesel exhaust in healthy men and male patients with coronary heart disease [33, 34] and thus highlighted the importance of particle constituent in determining the cardiovascular effects of particles in human subjects. However, evidence for roles of different particle constituents in the air pollution-related cardiovascular effects is generally lacking because most of the previous studies in this area only examined a limited number of particle constituents [e.g., OC, elemental carbon (EC), sulfate and nitrate], and studies investigating the health effects of multiple air pollutants/PM constituents are difficult due to several challenges .
In general, we found that gaseous air pollutants (CO, NOX, NO2 and NO) and several major PM2.5 constituents (EC, OC, POC and POM) related to traffic all showed their highest levels in the Urban Period 1 after the relocation, and levels of these air pollutants were lower in the Urban Period 2 (accompanied by lower PM2.5 levels) but still higher (except EC) than in the Suburban Period (Table 1). These suggest a strong impact of traffic emissions on the urban air pollution contents after the study subjects’ relocation. Interestingly, we found similar changes in several circulatory biomarkers along with these traffic-related air pollutants. Three inflammatory biomarkers (hs-CRP, TNF-α, and fibrinogen) and tHcy all showed higher mean levels in the Urban Periods than in the Suburban Period, and three of them (TNF-α, fibrinogen and tHcy) showed their highest median levels in the Urban Period 1. In contrast, hemostatic biomarkers (PAI-1, t-PA, vWF and sP-selectin) generally showed decreasing trends along with the decreasing PM2.5 levels across the three study periods, suggesting a different responsive pattern in the context of air pollution changes. In the current study, we further examined the health effects of a range of air pollutants and PM2.5 chemical constituents, which may provide clues for the different changes in circulatory biomarkers and serve as novel evidence for the chemical constituents that behind the air pollution-related cardiovascular effects.
By using various analytic approaches, we found consistent air pollution effects on several circulatory biomarkers of inflammation and coagulation, and PM2.5 chemical constituents may be more important than total PM2.5 when assess the cardiovascular effects of particulate air pollution. For example, we did not find significant associations between total PM2.5 and fibrinogen, PAI-1, t-PA and sP-selectin (Figure 1). In contrast, several PM2.5 constituents, including Mg, Fe, Ti, Co, Cd, Mn and Se, were found to have robust associations with these biomarkers (Table 3). Several other constituents, including SOC, NO3
-, Cl-, Al, Na, Zn and Mo, were also found to have robust associations with other biomarkers. Among these constituents, Zn, Co, NO3
- and Mn were shown to have robust associations with biomarkers independent of the other constituents (Figure 2). Of note, three of these constituents are transition metals (Zn, Co and Mn) whereas the other constituent is a secondary species (NO3
-). In a previous study we have demonstrated significant effects of PM2.5 transition metals on heart rate variability in healthy adults . In another experimental study, the water-soluble constituents of particles were found to shorten the whole-blood coagulation time significantly, and several metals included in this fraction were found to be responsible for this effect . Among the constituents examined in that study, Fe and Zn were two transition metals found to have the greatest capacity to reduce the coagulation time. In fact, the bioavailability of transition metals and their redox properties are considered very important for the toxic effects and oxidative damage in cardiopulmonary system [18, 37]. They are thought to be able to stimulate the production of reactive oxygen species when delivered to the airways, and then induce airway injury and inflammation followed by a series of cardiopulmonary responses . In a previous animal study, the authors found that the lung dose of bioavailable transition metal (Fe, Cu, Ni, V and Zn), but not instilled PM mass, was the primary determinant of the acute inflammatory response in rats . In addition, soluble metals in PM may include sulfates and nitrates, which support electron transport to produce oxidants and thus possess potential health impact . Although SO4
2- and NO3
- also have been examined by several previous studies [1, 12, 13, 40], evidence is generally limited for the causal association between PM sulfate and nitrate compounds . Further study is needed to clarity the potential health impact of sulfates and nitrates and their interactions with other PM constituents.
Epidemiologic evidence for the cardiovascular effects of specific PM constituents also has been growing. For example, the cardiovascular effects of carbonaceous constituents (e.g., EC, OC, POC, SOC) were frequently examined in previous studies and predominantly reflected the health impact of traffic- and combustion-related PM [10, 12–15]. Our study also found a consistent positive association between SOC and TNF-α (Table 3). However, the PM metal contents received much less attention as compared to the PM carbonaceous contents in previous studies. The consistent effects of metals found in our study suggest important roles that metals (especially the transition metals) may play in the PM-related short-term health effects. On the other hand, the health effects of a specific constituent may be a reflection of effects of emissions from a source, or of a common set of pollutants from a source . For example, we found that the association between SOC and TNF-α were confounded by OC and POM (see Additional file 1: Figure S2), which may suggest a synthetic effect from PM carbonaceous fractions related to traffic. Ambient air pollution involves many emission sources, including power plants, domestic heating, industrial processes, vehicular exhaust, biomass burning, etc. . Among the PM2.5 constituents, metals Fe, Zn, Cd, Mn and Pb come from the traditional source of industrial emissions, and metals Al, Mg, Ba and Ti (and part of Fe and Mn) originate from mineral aerosols that would be likely from resuspended road dust and long-range transported dust ; SO4
- and Se are mainly generated from coal burning whereas the chemical transformation that formed the secondary aerosols can also produce a lot of SO4
2- and NO3
[43, 44]; Cl- and F- generally represent the waste incineration and part of coal burning in urban area [43, 45]; additionally, some metals Fe, Zn, V, Mn and Pb may also be contained in traffic-related emissions in addition to the carbonaceous fractions and gases [43, 46, 47]. Our findings thus may also have potential implications for the development of relevant pollution abatement strategies that maximize benefits to the public health.
Our study also has several additional strengths in addition to the natural relocation study design. We used three different statistical models to address the collinearity between exposure variables and examine the consistency of air pollution effects. Our study subjects only consisted of young healthy adults free of smoking and cardiovascular compromises, and therefore could avoid confounding from variations in personal characteristics (e.g., old age, smoking, disease status, medication use, obesity, etc.). This study was conducted in spring and autumn seasons with modest climate conditions so as to avoid significant climate changes which may potentially confound the air pollution effect . However, our study has its limitations. We used air pollution data from central air-monitoring stations as a surrogates for participants’ recent air pollution exposures and might have introduced a nondifferential exposure error. However, this kind of exposure error is likely to cause a bias toward the null hypothesis and underestimate the air pollution effects . The fact that our study subjects lived within a small range (about 300 meters) around the air-monitoring stations ensured a relatively homogeneous exposure environment for the study subjects. As a result, the air pollution data from the central air-monitoring stations could well represent the subjects’ real air pollution exposures under natural conditions.