Clinical Track. The Beijing AIRCHD Study

Transcription

1 Clinical Track Extreme Levels of Air Pollution Associated With Changes in Biomarkers of Atherosclerotic Plaque Vulnerability and Thrombogenicity in Healthy Adults The Beijing AIRCHD Study Hongbing Xu,* Tong Wang,* Shengcong Liu, Robert D. Brook, Baihuan Feng, Qian Zhao, Xiaoming Song, Tieci Yi, Jie Chen, Yi Zhang, Yang Wang, Lemin Zheng, Sanjay Rajagopalan, Jianping Li, Wei Huang Rationale: The pathophysiologic mechanisms of air pollution associated exacerbation of cardiovascular events remain incompletely understood. Objective: To assess whether ambient air pollution can be a trigger of the vulnerable plaque and heightened thrombogenicity through systemic inflammatory pathways. Methods and Results: In Beijing AIRCHD study (Air Pollution and Cardiovascular Dysfunctions in Healthy Adults Living in Beijing), 73 healthy adults (mean±sd, 23.3±5.4 years) were followed up in 2014 to We estimated associations between air pollutants and biomarkers relevant to atherosclerotic plaque vulnerability, thrombogenicity, and inflammation using linear mixed-effects models and elucidated the biological pathways involved using mediation analyses. Receiver operating characteristic analyses were conducted to assess the ability of each biomarker to predict ambient air pollution exposures. High average concentrations of particulate matter in diameter <2.5 μm (91.8±63.8 µg/m 3 ) were observed during the study period. Significant increases in circulating biomarkers of plaque vulnerability, namely MMPs (matrix metalloproteinases; MMP-1, 2, 3, 7, 8, and 9), of 8.6% (95% CI, ) to 141.4% (95% CI, ) were associated with interquartile range increases in moving averages of particulate matter in diameter <2.5 μm, number concentrations of particles in sizes of 5 to 560 nm and black carbon, during the last 1 to 7 days before each participant s clinic visit. Higher air pollutant levels were also significantly associated with decreases in TIMP (tissue inhibitors of MMPs; TIMP-1 and 2), heightened thrombogenicity (shortened prothrombin time and increases in scd40l [soluble CD40 ligand], scd62p [soluble P-selectin], and fibrinogen/fibrin degradation products), and elevations in systemic inflammation (IL-1β [interleukin-1β], CRP [C-reactive protein], MIP- 1α/β [macrophage inflammatory protein-1α/β], srage [soluble receptor for advanced glycation end products], and IGFBP [insulin-like growth factor binding protein]-1 and 3). Receiver operating characteristic curves showed that several biomarkers can serve as robust pollutant-specific predictors with high versus low black carbon exposure (area under the receiver operating characteristic curve of [95% CI, ] for MMP-8 and [95% CI, ] for srage). Mediation analysis further showed that systemic inflammation can mediate 46% of the changes in MMPs and thrombogenicity associated with interquartile range increases in air pollutants. Conclusions: Our results suggest that air pollution may prompt cardiovascular events by triggering vulnerable plaque along with heightened thrombogenicity possibly through systemic inflammatory pathways. (Circ Res. 2019;124:e30-e43. DOI: /CIRCRESAHA ) Key Words: adult air pollutants air pollution carbon thrombosis Original received August 21, 2018; revision received December 17, 2018; accepted January 17, In December 2018, the average time from submission to first decision for all original research papers submitted to Circulation Research was days. From the Department of Occupational and Environmental Health, Peking University School of Public Health, Beijing, China (H.X., T.W., B.F., Q.Z., X.S., J.C., Y.Z., Y.W., W.H.); Division of Cardiology, Peking University First Hospital, Beijing, China (S.L., T.Y., J.L.); Division of Cardiovascular Medicine, University of Michigan, Ann Arbor (R.D.B.); Institute of Cardiovascular Sciences (L.Z.) and Institute of Systems Biomedicine (L.Z.), Peking University School of Basic Medical Sciences, Beijing, China; Division of Cardiovascular Medicine, Case Western Reserve Medical School, Cleveland OH (S.R.); and Key Laboratory of Molecular Cardiovascular Sciences of Ministry of Education, Health Science Center, Peking University, Beijing, China (H.X., T.W., S.L., B.F., Q.Z., X.S., T.Y., J.C., Y.Z., Y.W., L.Z., J.L., W.H.). *H.X. and T.W. contributed equally to this article. The online-only Data Supplement is available with this article at Correspondence to Wei Huang, Department of Occupational and Environmental Health, Peking University School of Public Health, No. 38 Xueyuan Rd, Haidian District, Beijing , China, whuang@bjmu.edu.cn or Jianping Li, Division of Cardiology, Peking University First Hospital, No. 8 Xishiku St, Xicheng District, Beijing, China, lijianping03455@pkufh.com 2019 American Heart Association, Inc. Circulation Research is available at DOI: /CIRCRESAHA e30

2 Xu et al Air Pollution and Plaque Vulnerability e31 Novelty and Significance What Is Known? Exposure to air pollution is associated with exacerbation of cardiovascular events and poses major threats to global public health. Thrombotic complication evolving from disruption of atherosclerotic plaques has been proposed to be responsible for air pollution attributable to cardiovascular events; however, the pathophysiologic mechanisms remain poorly elucidated. What New Information Does This Article Contribute? Acute exposures to high levels of ambient air pollutants were significantly associated with alterations in biomarkers of plaque vulnerability, namely MMPs (matrix metalloproteinases) and its tissue inhibitors. Higher air pollution levels were also related to activations of the coagulation and fibrinolytic system and increases in levels of systemic inflammation. The air pollution associated plaque vulnerability and thrombogenicity can be partially mediated by inflammatory responses. In the Beijing AIRCHD study (Air Pollution and Cardiovascular Dysfunction in Healthy Adults in Beijing), we have shown here for the first time that air pollution can prompt cardiovascular events by triggering vulnerable plaque along with heightened thrombogenicity possibly through systemic inflammatory pathways. Given the considerable global burden of cardiovascular diseases, these findings extend our current understandings on underlying biological mechanisms of air pollution induced preclinical cardiovascular responses. Our results also highlight the relevance of continued efforts to reduce air pollution emissions and to validate clinical approaches to reducing exposures at a personal level (eg, air filtration, N95 respirators), particularly among at-risk patients residing in heavily polluted environments. Nonstandard Abbreviations and Acronyms AIRCHD ALT AST BC CO CRP ET-1 FDP HDL IGF-1 IGFBP IL-1β IQR LDL LME MA MIP-1α/β MMP NF-κB NO 2 O 3 PAR1 PM PM 2.5 PNC PUHSC RAGE ROC scd40l scd62p SO 2 srage TIMP Air Pollution and Cardiovascular Dysfunctions in Healthy Adults Living in Beijing alanine aminotransferase aspartate aminotransferase black carbon carbon monoxide C-reactive protein endothelin-1 fibrinogen/fibrin degradation product high-density lipoprotein insulin-like growth factor-1 insulin-like growth factor binding protein interleukin-1β interquartile range low-density lipoprotein linear mixed effects moving average macrophage inflammatory protein-1α/β matrix metalloproteinase nuclear factor-κb nitrogen dioxide ozone protease-activated receptor-1 particulate matter particulate matter in diameter <2.5 μm particle number concentration Peking University Health Science Center receptor for advanced glycation end product receiver operating characteristic soluble CD40 ligand soluble P-selectin sulfur dioxide soluble receptor for advanced glycation end products tissue inhibitor of metalloproteinase Ambient air pollution a heterogeneous mixture of particulate matter (PM) and gaseous pollutants is one of the leading global risk factors for cardiovascular disease. 1 A large body of studies has shown that recent exposure to several ambient air pollutants during the prior days is linked to an increase in cardiovascular events, including acute ischemic events, in both cerebrovascular (stroke) and coronary vasculature (acute myocardial infarction). 2 Among the hypothesized mechanisms, it has been suggested that air pollution could trigger ischemic events in multiple vascular territories by promoting a globally vulnerable patient, triggering vulnerable plaques (atherosclerotic plaque instability) and vulnerable blood (heightened thrombogenicity). 2,3 Indeed, thrombotic complication evolving from disruption of atherosclerotic plaques has also been recognized as the most common cause of myocardial infarction and ischemic stroke. 4 Although exposure to PM and traffic-related air pollution has been associated with atherosclerotic burden in human and animal studies, 5,6 the pathophysiologic mechanisms whereby air pollution could acutely trigger cardiovascular e- vents remain incompletely understood. Meet the First Author, see p 664 Emerging evidence suggests that atherosclerotic plaque instability through alterations in MMP (matrix metalloproteinase) activity/expression might play a role in ambient air pollution associated triggering of clinically overt cardiovascular events Experimental and epidemiological studies have shown that MMPs can be responsible for altered vascular collagen and elastin composition, enhanced atherosclerotic burden and plaque vulnerability, as well as increased risk of cardiovascular mortality MMPs are often expressed in macrophages, endothelial cells, and smooth muscle cells at low levels, whereas they can be swiftly upregulated by proinflammatory cytokines. 15 MMPs are also important regulators involved in inflammatory responses and play critical roles in triggering thrombosis after plaque rupture along with cytokines, growth factors, and soluble adhesion molecules. 16 Compelling evidence has shown that pulmonary oxidative

3 e32 Circulation Research March 1, 2019 stress and inflammatory mediators (eg, cytokines) provoked by PM inhalation can spillover into the systemic circulation. 2 Hence, it is plausible that air pollution might exacerbate plaque instability (eg, altered MMP expression/activity) and blood thrombogenicity, in part, through systemic inflammatory pathways. Further, limited and recent studies showed some increased circulating MMPs in healthy adults from 2-hour controlled exposure to generated diesel exhaust particles, and alterations in MMPs differed across studies. 17,18 Lund et al 18 found that the impact of gasoline engine emission exposure on MMPs could be mediated through ET-1 (endothelin-1)-et A pathway; however, treatment with ET A receptor antagonist did not normalize expression/activity of MMPs, suggesting alternative or additional mechanisms that are likely responsible. Nevertheless, diesel exhaust particles/gasoline engine emission generated from laboratory may differ substantially from aged ambient particulates in urban environments, with regard to the levels and physical and chemical characteristics. Thus far, no study has evaluated the impact of ambient air pollution mixture on alteration of MMPs in human or the potential mediating role of systemic inflammation in MMPs and heightened blood thrombogenicity attributable to air pollution exposure. Under a real-world exposure scenario, the wide spectrum of extreme concentrations of ambient air pollutants in Beijing, China, presented a unique opportunity to assess the impacts of poor air quality on cardiovascular outcomes from a global perspective. Previously, we observed that the drastic changes in air pollution levels before and during the Beijing Olympics were associated with acute changes in biomarkers of inflammation, thrombosis, and measures of cardiovascular physiology in healthy adults. 19 Here, we hypothesized that ambient air pollution can be a trigger of the vulnerable plaque (increases in MMPs) and vulnerable blood (enhanced thrombogenicity), which might be, in part, through systemic inflammatory pathways. In this prospective follow-up study of Beijing AIRCHD (Air Pollution and Cardiovascular Dysfunctions in Healthy Adults Living in Beijing), with repeated measurements in 2014 to 2016, we examined the impacts of high levels of air pollution on biomarkers relevant to atherosclerotic plaque vulnerability, thrombogenicity, and systemic inflammation. We further assessed whether systemic inflammation responses can mediate the alterations in circulating MMPs and thrombogenicity with air pollution exposure. Methods The data that support the findings of this study are available from the corresponding author on reasonable request. Study Participants and Design The Beijing AIRCHD study is a multicenter collaborative project between Peking University Health Science Center (PUHSC) and the University of Michigan. 20 This collaborative project provided for the unprecedented capability to study the health impacts of air pollution across nearly the entire global range of daily ambient concentrations (eg, ambient particulate matter in diameter <2.5 μm [PM 2.5 ] in Southeast Michigan [6 12 μg/m 3 ] versus in Beijing [ μg/m 3 ]) in a well-harmonized experimental manner and provided preclinical evidence and strategies for the prevention of cardiovascular diseases. In Beijing AIRCHD study, 73 self-reported nonsmoking healthy adults at low cardiovascular risk were recruited and scheduled to participate in clinical examinations at baseline and subsequent 4 repeated visits in 2014 to We included participants who were self-reported nonsmoking healthy adults living in nonsmoking households aged 18 to 50 years and without any preexisting cardiovascular diseases or risk factors (eg, hypertension, diabetes mellitus, or treated hyperlipidemia). We excluded those who were on cholesterol or blood pressure [BP]-lowering medication and antioxidant uses, with fasting blood glucose >7.0 mmol/l and BP dysfunction (hypertension [resting BP >140/90 mm Hg] and hypotension [resting BP <100/50 mm Hg]) or at pregnancy. A unique identification number from 1 to 73 was assigned for each participant at enrollment. The baseline examination was conducted during November 1, 2014, to November 5, 2014, followed by 4 follow-ups with 3 to 4 months apart until January 2016 (visit 1: November 24, 2014, to January 8, 2014; visit 2: April 22, 2015, to June 1, 2015; visit 3: September 20, 2015, to November 9, 2015; visit 4: December 27, 2015, to January 18, 2016). Among the participants, 56 of them completed all 4 visits, 3 completed 3 visits, and 7 participants completed 2 visits. Thus, a total of 327 person-visits were compiled and applied in all analyses. Clinical examinations, including vascular function assessment, blood withdrawal, and urine collection, were performed at the Clinic of PUHSC, and the dates of each clinical examination were illustrated in Figure 1. This study was approved by the Institutional Review Board of PUHSC (No ), and written informed consent was obtained from each participant. Clinical and Biomarker Measurements In Beijing AIRCHD study, participants were scheduled to arrive at the clinic between 8 am and 10 am with fasting for each clinical examination. All of the participants had no symptoms of respiratory infection for at least 7 days before each follow-up visit and were rescheduled if they reported having cold or other infections. At each visit, participants were first asked to complete a standard questionnaire collecting information on age, sex, body mass index, waist-tohip ratio, smoking status, medical history, and residential address. After resting for 5 minutes, averages of 3 readings of right upper arm BP and heart rate were obtained in a sitting position using an autonomic sphygmomanometer (HEM-7200; Omron, Japan). Then, resting supine microvascular endothelial-dependent vasodilatation (reactive hyperemia index) was measured on the right index finger using the Endo-PAT2000 system (Itamar Medical, Caesarea, Israel) following standard protocol. These measurements of vascular function have been reported previously. 20 Finally, venous blood and spot morning urine samples were collected using established methods. 19,21 Blood samples were collected using vacuum tubes containing EDTA (BD Vacutainer K2 EDTA 7.2 mg, REF ; Becton Dickinson, NJ) for cell counts and hemoglobin measurements, and tubes were coated with lithium heparin (BD Vacutainer PST II, REF ) for plasma biochemical analyses, including albumin, total bilirubin, direct bilirubin, AST (aspartate aminotransferase), ALT (alanine aminotransferase), glucose, total cholesterol, HDL (high-density lipoprotein) cholesterol, LDL (low-density lipoprotein) cholesterol, and triglycerides. Tubes with sodium citrate (BD Vacutainer 0.109M, REF ) were used to collect plasma samples for thrombogenicity testing (prothrombin time and fibrinogen/ fibrin degradation products [FDPs]). All analyses of cell counts, hemoglobin, biochemistry, and thrombogenicity were performed within 4 hours of blood sample collection using standard automated clinical methods at PUHSC First Hospital. Blood samples collected using BD Vacutainer SST tubes (REF ) for further analyses were centrifuged and stored at 80 C until analyzed. Serum profiles of biomarkers relevant to plaque vulnerability (MMP-1, 2, 3, 7, 8, and 9 and TIMP [tissue inhibitor of MMP]-1 and 2), platelet activation (scd40l [soluble CD40 ligand] and scd62p [soluble P-selectin]), and systemic inflammation (srage [soluble receptor of advanced glycation end products], IL-1β [interleukin-1β], CRP [C-reactive protein], MIP-1α/β [macrophage inflammatory protein-1α/β], IGF- 1 [insulin-like growth factor-1], and IGFBP [insulin-like growth factor binding protein]-1 and 3) were measured using a bead-based multiplex flow cytometry with Aimplex Human Custom Kits according to the manufacturer s instructions ( resources/). The IGF-1/IGFBP-3 ratio was calculated reflecting IGF- 1 bioavailability. 22

4 Xu et al Air Pollution and Plaque Vulnerability e33 Figure 1. The dates of clinical examinations and time-series pattern of ambient air pollutants and plotted during the study period. Circles with color represent blood-draw dates. A, Concentration of particulates (PM 2.5 and BC) during the study period. B, Concentration of gaseous pollutants (NO 2 and O 3 ) during the study period. Baseline: November 1, 2014, to November 5, 2014; visit 1: November 24, 2014, to January 8, 2015; visit 2: April 22, 2015, to June 1, 2015; visit 3: September 20, 2015, to November 9, 2015; visit 4: December 27, 2015, to January 18, BC indicates black carbon; NO 2, nitrogen dioxide; O 3, ozone; and PM 2.5, particulate matter in diameter <2.5 μm. Urine samples were collected using Corning polypropylene centrifuge tubes (Corning, Inc, NY) and stored at 20 C until analyzed. Urinary cotinine (a proxy of potential environmental tobacco smoke exposure) was measured using a commercially available ELISA kit (Pomona, CA), and concentrations of cotinine were adjusted by urinary creatinine and reported as ng/mg creatinine. Measurements of Air Pollution and Meteorologic Parameters A fixed-location real-time air pollution monitoring station has been set up on PUHSC campus since January The sampling site (N , E ) is on the roof of PUHSC School of Public Health building (about 25 meters above ground level), which is 600 meters in distance from the 4th Ring Road (a major traffic road surrounding urban Beijing area) and 100 meters from a local expressway. Hourly concentrations of ambient PM 2.5 were measured using a beta attenuation monitor (BAM-1020; Met One Instruments, Inc). Five-minute number concentrations of particles (particle number concentration [PNC]) in sizes of 5 to 560 nm, including a total of 32 size distribution segments, were measured by a Fast Mobility Size Spectrometer (FMPS Model 3091; TSI). Daily average PNC in sizesegregated fractions of <50 nm (PNC 5 50 ), 50 to 100 nm (PNC ), and 100 to 560 nm (PNC ) were calculated on the basis of potential health effects of particles in small sizes. 23,24 Potential sources of particles <560 nm were assessed using the Positive Matrix Factorization. 25 Details on the measurement principle of FMPS Model 3091 and Positive Matrix Factorization method are provided in Sections 1 and 2 in the Online Data Supplement. At the same location, minute-to-minute black carbon (BC) concentrations were measured continuously using a dual-wavelength aethalometer (AE-33; Magee Scientific). Minute-tominute concentrations of sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), carbon monoxide (CO), and ozone (O 3 ) were measured by EC9800 series ambient gas analyzers (EcoTech Pty, Ltd, Australia). Hourly ambient temperature and relative humidity were measured by a Met One unit (Met One Instruments, Inc). Minute-to-minute measurements of ambient wind speed and barometric pressure were also obtained from Tsinghua meteorologic monitoring station ( cn/), which is 3 kilometers from the PUHSC campus. Daily averages were calculated from 9 am to 9 am for all environmental parameters, except for O 3 reported as maximum 8-hour average. In this study, 93% of study participants lived within 1 km from our air monitoring station. To validate the reliability of the pollutant concentrations measured at the fixed monitoring station as an air pollution exposure proxy for individuals, we performed correlation analysis across selected pollutants (PM 2.5 and BC) measured at ambient and personal levels (Section 3 in the Online Data Supplement).

5 e34 Circulation Research March 1, 2019 Table. Descriptive Statistics of Study Participants (N=73) All Participants (N=73) Women (n=48) Men (n=25) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Characteristics Age, y 23.3 (5.4) 23.0 (4.0) 23.4 (5.7) 23.0 (4.0) 23.3 (4.9) 23.0 (4.0) BMI, kg/m (3.5) 21.5 (5.8) 21.4 (3.1) 20.6 (2.6) 24.5 (3.4) 25.4 (4.5) WHR 0.80 (0.07) 0.80 (0.10) 0.77 (0.06) 0.77 (0.06) 0.83 (0.06) 0.84 (0.08) Systolic BP, mm Hg 110 (10) 108 (13) 106 (8) 105 (10) 117 (8) 115 (11) Diastolic BP, mm Hg 63 (8) 60 (12) 62 (8) 60 (12) 65 (8) 65 (12) Heart rate, bpm 69 (9) 70 (12) 70 (9) 71 (12) 68 (10) 67 (18) Reactive hyperemia index 1.6 (0.5) 1.5 (0.4) 1.6 (0.5) 1.5 (0.4) 1.7 (0.5) 1.6 (0.5) Red blood cell, cells/l 4.6 (0.4) 4.6 (0.6) 4.4 (0.3) 4.4 (0.4) 5.1 (0.3) 5.1 (0.3) White blood cell, 10 9 cells/l 5.5 (1.4) 5.4 (1.8) 5.4 (1.4) 5.4 (2.0) 5.7 (1.2) 5.6 (1.1) Platelet, 10 9 cells/l 222 (38) 215 (46) 221 (40) 216 (48) 225 (37) 215 (45) Hemoglobin, g/l 143 (13) 139 (17) 136 (8) 136 (7) 158 (7) 160 (13) Albumin, g/l 46.4 (2.6) 46.4 (4.1) 45.6 (2.4) 45.6 (3.7) 47.8 (2.5) 47.6 (3.8) TBIL, μmol/l 13.7 (6.1) 12.6 (6.2) 13.8 (6.6) 11.9 (6.4) 13.7 (5.1) 13.2 (4.2) DBIL, μmol/l 1.8 (1.1) 1.6 (1.4) 1.7 (1.0) 1.6 (1.3) 2.0 (1.1) 1.9 (1.1) AST, IU/L 17 (4) 16 (4) 16 (3) 16 (3) 19 (6) 18 (6) ALT, IU/L 16 (12) 13 (7) 12 (5) 11 (6) 23 (16) 19 (15) Fasting blood glucose, mmol/l 4.9 (0.4) 4.9 (0.6) 4.8 (0.4) 4.7 (0.5) 5.2 (0.4) 5.2 (0.5) TC, mmol/l 4.07 (0.70) 4.07 (0.79) 4.18 (0.65) 4.16 (0.71) 3.87 (0.76) 3.85 (0.99) HDL-C, mmol/l 1.37 (0.29) 1.34 (0.38) 1.45 (0.27) 1.42 (0.39) 1.22 (0.26) 1.21 (0.26) LDL-C, mmol/l 2.26 (0.58) 2.27 (0.80) 2.28 (0.54) 2.26 (0.73) 2.23 (0.67) 2.27 (1.06) Triglycerides, mmol/l 0.71 (0.29) 0.72 (0.40) 0.70 (0.29) 0.70 (0.38) 0.74 (0.29) 0.78 (0.36) Urinary cotinine, ng/mg Cr 20.8 (102.7) 2.6 (3.4) 8.2 (58.2) 2.3 (3.1) 44.6 (152.5) 2.7 (3.3) Measured biomarkers Plaque vulnerability MMP-1, ng/ml 3.80 (4.40) 1.99 (2.73) 3.87 (4.40) 1.99 (2.66) 3.78 (4.41) 1.97 (3.36) MMP-2, ng/ml 0.43 (0.79) 0.28 (0.15) 0.32 (0.12) 0.27 (0.15) 0.63 (1.28) 0.29 (0.13) MMP-3, ng/ml 4.22 (3.08) 3.94 (5.08) 3.34 (2.23) 3.48 (3.72) 5.60 (3.71) 5.60 (6.50) MMP-7, ng/ml 10.9 (17.7) 3.33 (11.3) 10.8 (16.5) 3.9 (11.4) 11.0 (19.6) 2.8 (11.5) MMP-8, ng/ml 1.46 (0.12) 0.61 (1.30) 1.38 (2.03) 0.52 (1.00) 1.60 (2.66) 0.70 (1.27) MMP-9, ng/ml 3.11 (2.62) 2.49 (2.72) 2.80 (1.95) 2.36 (2.50) 3.62 (3.38) 3.31 (2.92) TIMP-1, ng/ml 24.5 (13.5) 25.2 (20.1) 24.3 (11.4) 25.5 (15.4) 24.9 (16.4) 24.1 (21.4) TIMP-2, ng/ml 0.50 (0.73) 0.33 (0.30) 0.46 (0.43) 0.33 (0.29) 0.57 (1.05) 0.34 (0.31) Thrombogenicity Platelet activation scd40l, pg/ml 23.3 (94.8) 12.2 (10.0) 13.8 (9.9) 11.6 (9.2) 38.5 (152.0) 13.9 (12.5) scd62p, ng/ml 27.2 (18.4) 22.2 (21.9) 25.4 (17.2) 21.0 (21.5) 31.5 (26.0) 25.0 (24.1) Coagulation PT, s 11.3 (0.7) 11.2 (1.0) 11.3 (0.7) 11.1 (0.9) 11.2 (0.8) 11.2 (1.3) FDPs, mg/l 0.95 (2.24) 0.40 (0.90) 1.14 (2.69) 0.4 (1.1) 0.58 (0.66) 0.40 (0.70) Systemic inflammation Cytokines (Continued )

6 Xu et al Air Pollution and Plaque Vulnerability e35 Table. Continued srage, pg/ml 942 (866) 663 (925) 926 (847) 663 (838) 968 (893) 670 (933) IL-1β, pg/ml 13.3 (94.0) 1.6 (1.5) 4.5 (23.7) 1.5 (1.6) 27.6 (148.3) 1.7 (1.3) CRP, mg/l 0.54 (1.83) 0.19 (0.26) 0.32 (0.55) 0.16 (0.18) 0.88 (2.85) 0.25 (0.44) MIP-1α, pg/ml 17.5 (41.5) 10.7 (7.6) 12.4 (9.1) 10.2 (6.3) 25.8 (65.5) 11.8 (9.8) MIP-1β, pg/ml 403 (306) 337 (273) 369 (238) 319 (261) 458 (387) 371 (286) Growth factors All Participants (N=73) Women (n=48) Men (n=25) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Mean (SD) Median (IQR) IGF-1, pg/ml 98 (85) 62 (91) 100 (86) 65 (91) 94 (84) 58 (90) IGFBP-1, pg/ml 3002 (7684) 1628 (2127) 2649 (4535) 1697 (2049) 3584 (11 073) 1390 (1973) IGFBP-3, pg/ml 761 (403) 689 (523) 757 (391) 684 (524) 768 (425) 700 (521) IGF-1/IGFBP-3 ratio 0.13 (0.10) 0.09 (0.14) 0.13 (0.10) 0.09 (0.14) 0.13 (0.10) 0.09 (0.15) Descriptive statistics of biomarkers were calculated on the basis of all clinical visits. ALT indicates alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; BP, blood pressure; Cr, creatinine; CRP, C-reactive protein; DBIL, direct bilirubin; FDP, fibrinogen/fibrin degradation product; HDL-C, high-density lipoprotein cholesterol; IGF-1, insulin-like growth factor-1; IGFBP, insulin-like growth factor binding protein; IL-1β, interleukin-1β; IQR, interquartile range; LDL-C, low-density lipoprotein cholesterol; MIP-1α/β, macrophage inflammatory protein-1α/β; MMP, matrix metalloproteinase; PT, prothrombin time; scd40l, soluble CD40 ligand; scd62p, soluble P-selectin; srage, soluble receptor of advanced glycation end products; TBIL, total bilirubin; TC, total cholesterol; TIMP, tissue inhibitor of metalloproteinase; and WHR, waist-to-hip ratio. Statistical Analyses Descriptive statistics were conducted for all biomarkers and ambient environmental measurements. Spearman correlations were also performed across measurements of air pollutants and meteorologic parameters and biomarkers. Variables with heavily right-skewed distributions were log-transformed for further analyses. Linear mixed-effects (LME) models, which can account for correlation within subjects in the longitudinal panel study design, were conducted to estimate associations between air pollutants and biomarkers. Considering unequally time spaced clinical measurements in this study, the continuous first-order autoregressive covariance structure (corcar1) that represents an AR(1) autoregressive covariance structure for continuous covariates was selected to control for correlations between repeated measurements within each participant with potential decays over time. 26 All LME models were adjusted for seasonal variations by including sine and cosine terms of calendar dates of blood withdrawal. The potential impacts of meteorologic factors were controlled by 24-hour temperature and relative humidity before clinic visits using natural splines with 3 df based on lower Akaike Information Criterion. Because certain covariates may have unequal influences on the measured biomarkers, backward stepwise regression models were performed to select the key covariates for each biomarker according to literature and biological plausibility, 19,27 including age, sex, body mass index, waist-to-hip ratio, day of the week of clinical visit, month of the blood withdrawal, and creatinineadjusted urinary cotinine. The selected covariates in final models are summarized in Online Table I. Single-pollutant LME models were first generated to examine the associations of biomarkers with moving average (MA) concentrations of air pollutants during the last 24 hours before each participant s clinic visit (MA day 1 [1 MA day]), 1 to 2 days (2 MA days), 1 to 3 days (3 MA days), and so on, 7 MA days. Exposure-response curves were also derived from predictive models over broad air pollution exposure ranges to assess the observed associations obtained in single-pollutant models. In exploratory analyses, we first determined the ability to predict selected air pollutant exposures (PNC and BC) using receiver operating characteristic (ROC) curves based on measured biomarkers associated with PNC and BC exposures. 28,29 ROC curves were fitted modeling the predicted values of biomarkers derived from LME models and the air pollutants with the 75th percentile as cutoff points (6360 particles/m 3 for PNC , 7.4 μg/m 3 for BC). The area under the ROC curve values were calculated to assess the separation characteristics of ROC curves ( 0.90 as excellent, as good, as fair, and <0.70 as poor). 30 Second, we conducted mediation analyses to evaluate the effects of mediators (systemic inflammation) on the associations between exposures (selected air pollutants) and outcomes (MMPs, TIMPs, or blood thrombogenicity). This approach decomposes the total effect of air pollutants on measured biomarkers into a direct effect of exposure and an indirect effect (or mediation effect) accounting for mediators. 31 Two LME models were fitted with random intercepts in the single-mediator model, with 1 modeling the exposure-mediator association and the other modeling the mediator-outcome association. Considering potential correlations among measured biomarkers, the multiple-mediator model was also conducted by including all hypothesized mediators with significance at P <0.05 in the single-mediator model. 32 The detailed methods of mediation analyses for a single mediator and multiple mediators were provided in Section 4 in the Online Data Supplement. Two-pollutant model analyses were further developed to assess whether single-pollutant effects would be confounded by correlated co-pollutants. We stratified the analyses by participant s characteristics (sex and overweight status [body mass index 25 kg/m 2 ]) and examined the differential effect estimates within the subgroups. Lastly, sensitivity analyses were performed by (1) using subject-normalization approach to assess the heterogeneity in measured biomarker levels across participants 33 ; (2) excluding baseline observations to e- valuate the robustness of main results; (3) excluding participants with urinary cotinine levels above the 75th percentile (4.53 ng/mg creatinine) to evaluate whether the associations could be partially affected by potential passive smoking exposure; (4) additional adjustment for other meteorologic factors (ambient wind speed and barometric pressure). All estimates are reported as percentage changes with 95% CIs associated with interquartile range (IQR) increases in pollutant concentrations. Statistical significance was determined at P <0.05, and a Bonferroni correction was used at a significance level of P < (0.05/20) correcting for the number of measured biomarkers to control for potential type I error rate. All analyses were performed using R, version (R Project for Statistical Computing), and SAS statistical software, version 9.2 (SAS Institute, Inc, Cary, NC). Results In the Beijing AIRCHD study, the characteristics of study participants and biomarkers are summarized in the Table. Of 73 participants, 48 (66%) were women, with mean (SD) age

7 e36 Circulation Research March 1, 2019 Figure 2. Percentage changes in MMPs (matrix metalloproteinases) associated with interquartile range increases in pollutant concentrations. A, Percentage changes in MMP-1 associated with air pollutants. B, Percentage changes in MMP-2 associated with air pollutants. C, Percentage changes in MMP-3 associated with air pollutants. D, Percentage changes in MMP-7 associated with air pollutants. Error bars indicate 95% CIs. Significant associations (P<0.05) are shown in red; Bonferroni corrections with significance (P<0.0025) are indicated by asterisks. Moving average (MA) concentrations of air pollutants during the last 24 h before each participant s clinic visit are presented as 1 MA day, 1 to 2 d as 2 MA days, 1 to 3 d as 3 MA days and 7 MA days. BC indicates black carbon; CO, carbon monoxide; NO 2, nitrogen dioxide; O 3, ozone; PM 2.5, particulate matter in diameter <2.5 μm; PNCx, particle number concentrations in given size ranges (nm); and SO 2, sulfur dioxide. of 23.3 (5.4) years. Figure 1 depicts substantial variabilities in concentrations of ambient particulates (PM 2.5 and BC) a- cross the study period. Average daily concentrations of PM 2.5, PNC in sizes <100 nm, and BC 7 days before clinical visits were 91.8 µg/m 3, particles/cm 3, and 5.9 µg/m 3, respectively (Online Table II). Four sources of PNC in 5 to 560 nm were identified by Positive Matrix Factorization from gasoline vehicle emissions, aged vehicle emissions, nucleation, and secondary aerosols (Online Figure I). High correlations between ambient and personal PM 2.5 and BC concentrations within individuals supported for using fixed-station pollutant concentrations as exposure proxy in this study (Online Figure IV). Ambient PM 2.5 and BC, NO 2, and CO were highly correlated with correlation coefficients >0.8, whereas O 3 was negatively correlated with other pollutants (Online Figure VA). Significant correlations between MMPs and inflammatory biomarkers suggested potential correlations among the study biomarkers (Online Figure VB). In single-pollutant analyses, alterations in circulating levels of collagenases (MMP-1 and 8) were positively associated with IQR increases in multiple air pollutants (Figures 2A and 3A). We observed significant increases in MMP-1 of 18.3% (95% CI, ) to 141.4% (95% CI, ) and MMP-8 of 15.4% (95% CI, ) to 126.0% (95% CI, ) associated with IQR increases in PM 2.5, PNC , and BC at various exposure periods, with the largest effects observed at prior 7 MA days indicating accumulative effects during prolonged exposure period. For gelatinases (MMP-2 and 9), significant increases in MMP-2 of 8.6% (95% CI, ) to 51.5% (95% CI, ) and MMP- 9 of 13.7% (95% CI, ) to 46.5% (95% CI, ) were observed in associations with IQR increases in exposure to PNC , BC, NO 2, CO, and SO 2 (Figures 2B and 3B). Significant elevations in stromelysin-1 (MMP-3) and matrilysin-1 (MMP-7) were also observed in associations with IQR increases in air pollution exposure (Figure 2C and 2D). For TIMPs, significant reductions in TIMP-1 of 8.1% (95% CI, 14.6 to 1.1) to 25.9% (95% CI, 35.6 to 14.7) and TIMP-2 of 7.7% (95% CI, 14.6 to 1.1) to 16.0% (95% CI, 22.3 to 9.8) were found in associations with IQR increases in exposure to PM 2.5, PNC , BC, NO 2, and CO at prior 7 MA days (Figure 3C and 3D). For biomarkers relevant to platelet activation, significant increases in scd62p of 15.3% (95% CI, ) to 28.5% (95% CI, ) were observed in associations with IQR increases in PM 2.5, BC, NO 2, and CO at prior 3 to 7 MA days (Figure 4A). The increases in scd40l were at reduced magnitude with significance of 14.0% (95% CI, ) to 24.3% (95% CI, ), in associations with IQR increases in exposure to PM 2.5, BC, NO 2, and CO (Figure 4B). For coagulation

8 Xu et al Air Pollution and Plaque Vulnerability e37 Figure 3. Percentage changes in MMPs (matrix metalloproteinases) and TIMPs (tissue inhibitors of metalloproteinases) associated with interquartile range increases in pollutant concentrations. A, Percentage changes in MMP-8 associated with air pollutants. B, Percentage changes in MMP-9 associated with air pollutants. C, Percentage changes in TIMP-1 associated with air pollutants. D, Percentage changes in TIMP-2 associated with air pollutants. Error bars indicate 95% CIs. Significant associations (P<0.05) are shown in red; Bonferroni corrections with significance (P<0.0025) are indicated by asterisks. Moving average (MA) concentrations of air pollutants during the last 24 h before each participant s clinic visit are presented as 1 MA day, 1 to 2 d as 2 MA days, 1 to 3 d as 3 MA days and 7 MA days. BC indicates black carbon; CO, carbon monoxide; NO 2, nitrogen dioxide; O 3, ozone; PM 2.5, particulate matter in diameter <2.5 μm; PNCx, particle number concentrations in given size ranges (nm); and SO 2, sulfur dioxide. function, significant reductions in prothrombin time of 1.1% (95% CI, 2.0 to 0.2) to 4.2% (95% CI, 5.9 to 2.5) were observed in associations with IQR increases in most pollutants (except for PM 2.5 and O 3 ), with the largest decreases observed at prior 5 to 7 MA days (Figure 4C). Significant increases in FDPs of 30.4% (95% CI, ) to 246.9% (95% CI, ) were also observed in associations with IQR increases in most pollutants (except for NO 2 and O 3 ; Figure 4D). For biomarkers relevant to systemic inflammation, as shown in Figure 5A, significant increases in srage were observed in associations with IQR increases in most pollutants at prior 7 MA days, ranging from 12.2% (95% CI, ) to 59.3% (95% CI, ). Greater magnitude of increases in CRP was observed in associations with IQR increases in exposure to BC and SO 2 at before 7 MA days (Figure 6A). In addition, a positive association between IL-1β and exposure to NO 2, CO, and SO 2 at current day was observed ranging from 13.9% (95% CI, ) to 36.7% (95% CI, ; Figure 5B). Similar association patents were also observed for MIP-1α and MIP-1β (Figure 5C and 5D). Significant increases in IGFBP-1 and 3 were also observed in associations with IQR increases in PNC , BC, NO 2, and CO at prior 7 MA days (Figure 6B and 6C). As hypothesized, significant reductions in IGF-1/IGFBP-3 ratio of 17.8% (95% CI, 30.0 to 3.4) to 33.2% (95% CI, 45.6 to 18.0) were observed in associations with IQR increases in PNC 5 50, BC, NO 2, CO, and SO 2 at prior 7 MA days (Figure 6D). All descriptions and summary statistics of single-pollutant analyses are presented in Online Tables III through VII. Most of the observed changes in biomarkers were in associations with IQR increases in particles from gasoline vehicle emissions, aged vehicle emissions, and secondary aerosols (Online Tables III through VII). In exploratory analyses, exposure-response curves generally supported a relationship whereby higher exposures to pollutants were associated with greater magnitudes of alterations in biomarkers of plaque vulnerability, thrombogenicity, and systemic inflammation (Online Figure VI); however, attenuation of effect estimates with wider CIs on most selected biomarkers was observed at extremely high exposure levels. As shown in Figure 7, ROC analyses further showed that several biomarkers displayed as pollutant-specific predictors to discriminate responses from broad ranges of selected pollutants at cutoff points of 75th percentile, which yielded area under the ROC curve values for MMP-8 (0.994 [95% CI, ] for PNC and [95% CI, ] for BC) and for srage (0.985 [95% CI, ] for PNC and [95% CI, ] for BC). In mediation analyses, single-mediator models showed that BC-associated changes in biomarkers relevant to

9 e38 Circulation Research March 1, 2019 Figure 4. Percentage changes in biomarkers of platelet activation and coagulation associated with interquartile range increases in pollutant concentrations. A, Percentage changes in scd62p associated with air pollutants. B, Percentage changes in scd40l associated with air pollutants. C, Percentage changes in PT associated with air pollutants. D, Percentage changes in FDPs associated with air pollutants. Error bars indicate 95% CIs. Significant associations (P<0.05) are shown in red; Bonferroni corrections with significance (P<0.0025) are indicated by asterisks. Moving average (MA) concentrations of air pollutants during the last 24 h before each participant s clinic visit are presented as 1 MA day, 1 to 2 d as 2 MA days, 1 to 3 d as 3 MA days and 7 MA days. BC indicates black carbon; CO, carbon monoxide; FDP, fibrinogen/fibrin degradation product; NO 2, nitrogen dioxide; O 3, ozone; PM 2.5, particulate matter in diameter <2.5 μm; PNCx, particle number concentrations in given size ranges (nm); PT, prothrombin time; scd40l, soluble CD40 ligand; scd62p, soluble P-selectin; and SO 2, sulfur dioxide. thrombogenicity (eg, FDPs and scd62p) can be mediated through MMPs and systemic inflammation 46% in proportion, and those changes in MMPs can be mediated through systemic inflammation 39% (Online Table VIII). Results from multiple-mediator models were consistent with those observed in single-mediator models (Online Table IX). In specific, we observed that alterations of MMPs, TIMP-2, and FDPs attributable to air pollution exposure were greater among men, whereas no consistent changes were observed in overweight status stratified analyses (Online Tables X through XVI). Our primary results remained robust in sensitivity examinations, including 2-pollutant models and subject-normalization analyses, as well as in analyses excluding observations at baseline, excluding participants with urinary cotinine above the 75th percentile, with additional adjustments for ambient wind speed and barometric pressure (Online Figures VII and VIII). Discussion The Beijing AIRCHD study has shown here that high levels of ambient air pollutants, including different size-fractioned particulates, BC, NO 2, CO, and SO 2 can promote significant alterations in circulating MMPs and TIMPs among healthy young adults living in Beijing a heavily polluted megacity. Higher air pollution levels were also related to heightened blood thrombogenicity (reductions in prothrombin time and increases in scd62p, scd40l, and FDPs), as well as elevations in inflammation (IL-1β, CRP, MIP-1α/β, srage, and IGFBP-1 and 3). ROC analysis suggested that several biomarkers (eg, MMP-8 and srage) can serve as specific and robust indicators predicting air pollution exposure. The air pollution associated changes in biomarkers of atherosclerotic plaque vulnerability and thrombogenicity have been shown to be at least partially mediated by systemic inflammatory pathways. In Beijing AIRCHD study, our collective findings provide additional mechanistic evidence supporting that air pollution may contribute to a heightening of both plaque vulnerability and blood thrombogenicity and thereby trigger acute ischemic cardiovascular events across a number of vascular territories under a real-world exposure scenario (as shown in Figure 8). These findings extend our current understandings on potential pathways of air pollution-induced preclinical cardiovascular responses, which are novel and of significant public health importance. A large amount of evidence suggest that circulating MMPs can largely influence the stability of vulnerable plague and have been considered as biomarkers associated with clinical

10 Xu et al Air Pollution and Plaque Vulnerability e39 Figure 5. Percentage changes in biomarkers of inflammation associated with interquartile range increases in pollutant concentrations. A, Percentage changes in srage associated with air pollutants. B, Percentage changes in IL-1β associated with air pollutants. C, Percentage changes in MIP-1α associated with air pollutants. D, Percentage changes in MIP-1β associated with air pollutants. Error bars indicate 95% CIs. Significant associations (P<0.05) are shown in red; Bonferroni corrections with significance (P<0.0025) are indicated by asterisks. Moving average (MA) concentrations of air pollutants during the last 24 h before each participant s clinic visit are presented as 1 MA day, 1 to 2 d as 2 MA days, 1 to 3 d as 3 MA days and 7 MA days. BC indicates black carbon; CO, carbon monoxide; IL-1β, interleukin-1β; MIP-1α/β, macrophage inflammatory protein-1α/β; NO 2, nitrogen dioxide; O 3, ozone; PM 2.5, particulate matter in diameter <2.5 μm; PNCx, particle number concentrations in given size ranges (nm); srage, soluble receptor of advanced glycation end products; and SO 2, sulfur dioxide. manifestations of atherosclerotic burden and cardiovascular events. 12,14,34 The matrix-degrading functions of MMPs (eg, MMP-1) can also play an important role in the pathogenesis of atherosclerosis and have agonist activity for PAR1 (protease-activated receptor-1). 16,35 In patients with coronary artery disease, each 1-ng/mL increase in MMP-1 was found in association with increased risks of total plaque burden in odds ratios of to 1.37-folds. 34 Studies in vitro showed that activation of MMP-1/PAR1 was capable of induction of a procoagulant state, and heightened products of thrombogenicity (eg, FDPs) might further increase the vulnerability of atherosclerotic plaques via NF-κB (nuclear factor-κb) pathway, which might potentially explain our findings of air pollution associated alterations in MMPs along with significant changes in biomarkers of thrombogenicity Thus far, the Beijing AIRCHD is the first study examining the impacts of ambient air pollution mixtures on alteration of MMPs, in which significant increases in MMP-1 of 18.3% to 141.4% associated with IQR increases in air pollution exposures can be translated into 0.70 to 5.42 ng/ml elevations in MMP-1. Although the magnitude of adverse biological changes (eg, MMP-1) was relatively modest and the impact on any single individual is likely to be small, however, even small adverse health responses can translate into an enormous threat to public health if millions of people are impacted, most likely by instigating acute events among susceptible people with preexisting lesions. In Beijing AIRCHD study, we systematically examined a suite of biomarkers that are potentially regulating or altering the circulating levels of MMPs attributable to ambient air pollution. Studies showed that inflammatory stimuli can significantly elicit MMP expression/activity, 15 and activation of RAGE (receptor for advanced glycation end products) signaling can play an important role in triggering inflammatory cascades and plaque instability. 38 Circulating srage is primarily generated from the cleavage of cellular membrane-bound RAGE in endothelial cell, and elevated srage levels may reflect the expression of cell surface RAGE. 39 A study in vivo reported that diesel exhaust particle mediated RAGE activation induced significant increases in inflammatory cytokines. 40 IGF-1 as a potent anti-inflammatory mediator is capable of reductions in vascular inflammation and MMP expression and thereby increased plaque stability Circulating IGF-1 is predominantly regulated by IGFBPs, especially IGFBP-3 (a potent inhibitor of IGF-1 bioavailability). 22 Another important factor involving MMP expression/activity is TIMP, 44 which showed increases in TIMPs in relation to gasoline

11 e40 Circulation Research March 1, 2019 Figure 6. Percentage changes in biomarkers of inflammation associated with interquartile range increases in pollutant concentrations. A, Percentage changes in CRP associated with air pollutants. B, Percentage changes in IGFBP-1 associated with air pollutants. C, Percentage changes in IGFBP-3 associated with air pollutants. D, Percentage changes in IGF-1/IGFBP3 ratio associated with air pollutants. Error bars indicate 95% CIs. Significant associations (P<0.05) are shown in red; Bonferroni corrections with significance (P<0.0025) are indicated by asterisks. Moving average (MA) concentrations of air pollutants during the last 24 h before each participant s clinic visit are presented as 1 MA day, 1 to 2 d as 2 MA days, 1 to 3 d as 3 MA days, and 7 MA days. BC indicates black carbon; CO, carbon monoxide; CRP, C-reactive protein; IGF-1, insulin-like growth factor-1; IGFBP, insulin-like growth factor binding protein; NO 2, nitrogen dioxide; O 3, ozone; PM 2.5, particulate matter in diameter <2.5 μm; PNCx, particle number concentrations in given size ranges (nm); and SO 2, sulfur dioxide. engine emission exposure. 7 However, we observed significant decreases in TIMP-1 and TIMP-2 associated with air pollution exposure in this healthy population; mediation analysis suggested that heighten inflammation could be involved and further mechanistic studies are needed. Collectively, our findings that the observed alterations on MMPs associated with ambient air pollution might be partially mediated by inflammation pathways provide important evidence supporting the study hypothesis. There is also existing evidence, albeit inconsistent, that air pollution may heighten blood thrombogenicity. 45 Several studies reported positive associations between platelet activation (scd40l and scd62p) and PM air pollution in both healthy and susceptible adults. 19,46,47 Recent studies showed that alterations in prothrombin time and FDPs, indicators of activation of coagulation function and fibrinolytic system, were also associated with PM air pollution, 48,49 which are largely consistent with our findings. In this study, we extended the previous evidence to reveal that ambient air pollution associated thrombogenicity could be potentially mediated through alterations in MMPs, given that MMPs can be important regulators involving in inflammatory responses and thrombotic reactions. 16 Studies have shown that MMPs might be primarily responsible for the shedding of scd40l from activated platelets. 50,51 Considering all these evidence, we speculate that circulating MMPs might play important roles in ambient air pollution associated thrombogenicity. It would be of great public health and policy interest, from both standpoints of emission control priorities and regulatory perspectives, to understand the pathophysiologic mechanisms posed by what specific mixtures of ambient air pollutants might induce adverse cardiovascular responses. In Beijing AIRCHD study, the observed changes in biomarkers of vulnerable plaque (increases in MMPs) and vulnerable blood (enhanced thrombogenicity) were found more responsive to exposure of particulates in smaller size fractions (eg, ultrafine particles) and, in part, mediated by systemic inflammatory pathways. Inhaled ultrafine particles have been found to be capable of translocation from the lung to the circulation and selective accumulation at sites of vascular inflammation (especially in atherosclerotic plaques) and could be detected in blood and urine following exposure after 3 months. 52 Another potential clinical implication of our findings is that several candidate-specific biomarkers (eg, MMP-8 and srage) identified by ROC curves may serve as clinical relevant indicators for predicting air pollution associated cardiovascular dysfunctions. Collectively, these evidence suggest that pharmacological and air pollution interventions targeting the personal level (eg, anti-inflammatory agents, air filtration) may represent novel therapeutic strategies to lessen or mitigate the

12 Xu et al Air Pollution and Plaque Vulnerability e41 Figure 7. Receiver operating characteristic (ROC) curves for measured biomarkers distinguishing high (>75th percentile) and low (<75th percentile) levels of air pollutants. A, ROC curves for PNC exposure. B, ROC curves for black carbon (BC) exposure. We used 7-d moving averages of particle number concentration in sizes of 50 to 100 nm (PNC ) and BC as representative exposure metrics with cutoff points at the 75th percentile (6360 particles/m 3 for PNC , 7.4 μg/m 3 for BC) to model ROC curves and presented results of the area under the ROC curve >0.70. FDP indicates fibrinogen/fibrin degradation products; IGFBP, insulin-like growth factor binding proteins; MMP, matrix metalloproteinase; and srage, soluble receptor of advanced glycation end products. risk for cardiovascular events in susceptible people, particularly among patients with known atherosclerosis. 53,54 The Beijing AIRCHD study has several strengths. Most previous epidemiological studies exploring pathophysiologic mechanisms regarding the adverse cardiovascular responses attributable to air pollution have focused on associations among a few biomarkers, and air pollution exposure metrics were based on criteria air pollutants and lower variations in levels. In Beijing AIRCHD study, we measured a suite of circulating biomarkers relevant to plaque vulnerability, thrombogenicity, and systemic inflammation, some of which (eg, MMPs, IGFBPs, and srage) are novel and have seldom been applied to air pollution studies. In addition, we observed robust and relatively modest changes in biomarker levels associated with air pollution exposure in a homogenous group of nonsmoking healthy and young adults, without potential confounding from medication use, preexisting disease, and age-related susceptibility. However, several study limitations should be noted when interpreting the results of Beijing AIRCHD study. One potential study limitation is potential for type 1 errors/false positives in the analysis that can be introduced from multiple testing. Although the Bonferroni correction was used Figure 8. Schematic illustration showing the proposed mechanisms of atherosclerotic plaque vulnerability and thrombogenicity after air pollution exposure. CRP indicates C-reactive protein; IGF-1, insulin-like growth factor-1; IGFBP, insulin-like growth factor binding protein; IL-1β, interleukin-1β; MIP-1α/β, macrophage inflammatory protein-1α/β; MMP, matrix metalloproteinase; scd40l, soluble CD40 ligand; scd62p, soluble P-selectin; srage, soluble receptor of advanced glycation end products; and TIMP, tissue inhibitor of metalloproteinase.