Chemical constituents of ambient particulate air pollution and biomarkers of inflammation, coagulation and homocysteine in healthy adults: A prospective panel study

Background Ambient air pollution has been associated with activation of systemic inflammation and hypercoagulability and increased plasma homocysteine, but the chemical constituents behind the association are not well understood. We examined the relations of various chemical constituents of fine particles (PM2.5) and biomarkers of inflammation, coagulation and homocysteine in the context of traffic-related air pollution. Methods A panel of 40 healthy college students underwent biweekly blood collection for 12 times before and after their relocation from a suburban campus to an urban campus with changing air pollution contents in Beijing. Blood samples were measured for circulatory biomarkers of high-sensitivity C reactive protein (hs-CRP), tumor necrosis factor alpha (TNF-α), fibrinogen, plasminogen activator inhibitor type 1 (PAI-1), tissue-type plasminogen activator (t-PA), von Willebrand factor (vWF), soluble platelet selectin (sP-selectin), and total homocysteine (tHcy). Various air pollutants were measured in a central air-monitoring station in each campus and 32 PM2.5 chemical constituents were determined in the laboratory. We used three different mixed-effects models (single-constituent model, constituent-PM2.5 joint model and constituent residual model) controlling for potential confounders to estimate the effects of PM2.5 chemical constituents on circulatory biomarkers. Results We found consistent positive associations between the following biomarkers and PM2.5 chemical constituents across different models: TNF-α with secondary organic carbon, chloride, zinc, molybdenum and stannum; fibrinogen with magnesium, iron, titanium, cobalt and cadmium; PAI-1 with titanium, cobalt and manganese; t-PA with cadmium and selenium; vWF with aluminum. We also found consistent inverse associations of vWF with nitrate, chloride and sodium, and sP-selectin with manganese. Two positive associations of zinc with TNF-α and of cobalt with fibrinogen, and two inverse associations of nitrate with vWF, and of manganese with sP-selectin, were independent of the other constituents in two-constituent models using constituent residual data. We only found weak air pollution effects on hs-CRP and tHcy. Conclusions Our results provide clues for the potential roles that PM2.5 chemical constituents may play in the biological mechanisms through which air pollution may influence the cardiovascular system.


Supplementary Methods
Air Pollution Measurement and Analysis Estimation of Additional Carbonaceous Fractions References Table S1. Spearman's correlation matrix for the ambient air pollutants and PM 2.5 chemical constituents during the study Figure S1. Percent changes with 95% CIs in circulatory biomarkers associated with IQR increases in major air pollutants and PM 2.5 constituents at average concentrations during the preceding 1 to 6 days prior to the blood collection. Figure S2. Percent changes with 95% CIs in circulatory biomarkers associated with IQR increases in selected PM 2.5 constituents in single-constituent models and two-constituent models using constituent residual data.

Air Pollution Measurement and Analysis
The BIT Liangxiang campus is about 2 kilometers from the nearest freeway and was surrounded by active construction sites within 2 kilometers during the study. There are also a certain number of industrial facilities located within several kilometers from this campus. In contrast, the BIT main campus is located in the downtown area of Beijing City along the northwest inner side of the 3 rd ring road that circles the city [1], and with no substantive construction activities or industrial facilities nearby. The air monitoring site in the BIT Liangxiang campus was installed on the rooftop of a three-storey building (about 10 meters high) with open vision and the monitoring site in the BIT main campus was installed on the rooftop of a five-storey building (about 15 meters high) within 200 meters from the 3 rd ring road. We used the following instruments and materials in the air pollution measurements: a digital dust monitor for real-time PM 2.5 concentration measurement (LD-3K; Sibata Scientific Technology Inc., Tokyo, Japan); SKC sampling systems for PM 2.5 mass collection on quartz-fiber filters and polytetrafluoroethylene filters (SKC Inc., Eighty Four, PA, USA); a model T15n enhanced CO measurer for real-time CO concentration measurement (Langan Products Inc., San Francisco, CA, USA); Ogawa passive samplers for NO X and NO 2 collection on cellulose fiber filters (Ogawa Air Inc., Osaka, Japan); a HOBO Pro V2 logger for real-time temperature and relative humidity measurements (Onset Corp., Pocasset, MA, USA). Instruments were calibrated according to manufacturer's specifications before field work in each study period. Field filter blanks for PM 2.5 sampling were used in 20% of the sample days, and the sampling filters were all stored at -30°C before laboratory analysis.
Daily mass concentrations of PM 2.5 samples were determined by standard weighing procedures before and after the sample collection using a XS105 Dual Range Scale (Mettler Toledo, Columbus, OH, USA) and PM 2.5 filter samples were analyzed in the laboratory for the following chemical constituents using professional techniques [2-5]: OC and EC by thermo/optical transmission method; SO 4 2-, NO 3 -, Cland Fby ion chromatography; K, Na, Ca, Mg, Al, Fe and Zn by inductively coupled plasma atomic emission spectrophotometry; and Ba, Sr, Mn, Ti, Cu, Cr, Ni, Cd, V, Mo, Co, Pb, Sn, Sb, As and Se by inductively coupled plasma mass spectrometry. Concentration values below the detection limit were assigned ½ of the detection limit for each constituent. Quality control was performed by reanalyzing 15% of the samples, and results generally showed good agreement between analyses (mean difference of ±15% for different constituents). NO X and NO 2 contents collected on cellulose fiber filters were determined by a spectrophotometer at a wavelength of 545 nm in the laboratory following manufacturer's specification [6]. The NO contents were calculated as the differences between the NO X and NO 2 contents.

Estimation of Additional Carbonaceous Fractions
Three additional carbonaceous fractions, including primary OC (POC), secondary OC (SOC), and particulate organic matter (POM), represent different components of organic constituents within ambient particles. The OC can be either released directly into the atmosphere as POC or produced from gas-to-particle reactions (SOC) [7]. These constituents may have different potentials to affect the cardiovascular health as partly demonstrated in a few recent studies [8,9].
The contributions of SOC and POC to measured ambient OC were estimated from OC and EC concentrations using EC as a tracer of primary combustion generated OC (i.e., "EC tracer method") [10][11][12]. A previous study suggested using the minimum OC/EC ratios of the ambient aerosols as primary OC/EC ratios and calculating SOC assuming that the meteorological conditions are not favorable for the SOC formation in some cases [10]. These cases include the lack of direct solar radiation, low ozone concentration and unstable air mass [13]. This method has been successfully applied in several previous studies estimating the SOC concentrations in Chinese urban areas [7,11,14]. In our study, we thus estimated the SOC concentrations by the following expression: SOC = OC tot -EC×(OC/EC) min (1) where SOC is the secondary organic carbon, OC tot the total measured ambient organic carbon, and (OC/EC) min the minimum OC/EC ratio of the ambient aerosols. To account for heterogeneous relationships between OC and EC over different periods, we used the minimum OC/EC ratio of each time period to estimate the SOC concentrations for that time period. The POC concentrations were calculated as the differences between OC tot and SOC concentrations.
In addition, we also estimated the concentrations of particulate organic matter (POM). POM can be estimated by multiplying the measured OC by a factor to compensate for other atoms such as H, O and N in the organic molecule [11]. A previous study suggested a factor of 1.6 for the estimation of POM in urban aerosols [15], and this factor has also been successfully applied in several studies in Chinese urban areas [11,16]. Therefore the factor of 1.6 was adopted in the current study.    Figure S1. Percent changes with 95% CIs in circulatory biomarkers associated with IQR increases in major air pollutants and PM 2.5 constituents at average concentrations during the preceding 1 to 6 days prior to the blood collection.

References
(1) Major air pollutants series: (A) hs-CRP, and (B) tHcy; and (2) PM 2.5 constituents series: (C) hs-CRP, (D) TNF-α, (E) Fibrinogen, (F) PAI-1, (G) t-PA, (H) vWF, (I) sP-selectin, and (J) tHcy. Figure S1 (1) Figure S2. Percent changes with 95% CIs in circulatory biomarkers associated with IQR increases in selected PM 2.5 constituents in single-constituent models and two-constituent models using constituent residual data. Solid triangle: effect estimate of the constituent from single-constituent model; solid squares: effect estimates of the constituent from two-constituent models adjusting for the other constituents as shown in the X-axis. (A) TNF-α and SOC at 4-d moving average, (B) vWF and Clat 4-d moving average, and (C) vWF and Al at 5-d moving average.