Received date: August 31, 2015; Accepted date: September 18, 2015; Published date: September 23, 2015
Citation: Ruana Y, Liangb R, Liana H, Zhaoa X, Liua X et al. (2015) Endothelial Function and Short-term Exposure to Particulate Matter: A Systematic Review and Meta-analysis. J Environ Anal Chem 2:159. doi:10.4172/2380-2391.1000159
Copyright: © 2015 Ruana Y, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at Journal of Environmental Analytical Chemistry
Association of particulate matter with endothelial function Current evidence has confirmed the definite correlation between air pollution and cardiovascular disease. One of the mechanisms was the adverse effect on endothelial function. However, there is heterogeneity between the air pollution and endothelial function in different studies. We performed a meta-analysis to determine the direction and strength of the association. PubMed, EMBASE, Cochrane library and Web of Science were searched for a combination of keywords related to major air pollutants and to the indexes of endothelial function, including reactive hyperemia, flow-mediated dilation and nitroglycerin-mediated dilatation. Thirteen of reviewed articles with sufficient details met inclusion criteria. Descriptive and quantitative information was extracted from the included studies. A pooled estimate in a random effects model was computed, and change in endothelial function (95% CIs) were calculated for each increment of 10 μg/m3 in PM2.5, and 1 μg/ m3 scaled for black carbon. A 0.27% decrease in flow-mediated dilation was marginally associated with per 1 μg/m3 increase in PM2.5 in overall risk estimate analysis, with the greatest effect occurred in PM2.5 exposure for 1-day lag. Subgroup analyses have shown that the effect was modified by the factors as follows: American, subjects aged less than 55 years old, the proportion of female less than 50%, and the panel design. A 0.07 mm decrease in brachial artery diameter and 2.46% decrease in nitroglycerin-mediated dilation for per 1 μg/m3 increment in BC were observed in the pooled analysis from two studies(P=0.005 and P=0.043, respectively). No significant association was found between black carbon and flow-mediated dilation. Short-term exposure to PM2.5 and black carbon was associated with endothelial function, which provided the evidence for the increased risk of cardiovascular disease due to air pollution
Endothelial dysfunction; Flow-mediated dilation; Cardiovascular disease; Particulate matter; Meta-analysis.
A growing body of evidence has confirmed the association between air pollutants and increased risk of cardiovascular morbidity and mortality , either short-term or long-term exposure. What’s more, exposure to air pollution has been emerging as an important risk factor for the development and progression of cardiovascular disease (CVD), which has been set forth in the guidelines and scientific statement [2,3]. Many credible pathological mechanisms of PM-mediated cardiovascular effects have been elucidated, including systemic autonomicnervous system imbalance, systemic proinflammatory responses, thrombosis and coagulation, oxidative stress, vascular or endothelial dysfunction . Atherosclerosis is the major cause of CVD. In the earliest time, 2 clinical trials with 798 participants conducted in the Los Angeles, arrived at the conclusion that 10 μg/m3 higher in PM2.5 was associated with non-significant 4.2% elevation in common carotid intima-media thickness (CIMT), a marker of subclinical atherosclerosis . Afterwards, related studies have been increasing with heterogeneity between different studies concerning the strength of the association. In 2015, Provost et al. performed a meta-analysis and drew a conclusion that an increase of 5 μg/m3 PM2.5 was associated with a 1.66% thicker CIMT in the combined cross-sectional studies and a 1.04 μm per year greater CIMT progression in the combined longitudinal studies .
Endothelial dysfunction, as a possible and important mechanism, is an initial step in atherosclerosis and caused principally by loss of endothelium-derived nitric oxide . The noninvasive tools to evaluate the endothelial function are brachial artery ultrasound and peripheral arterial tonometry (PAT). The former is a commonly used and widely accepted measure of peripheral macrovascular endothelial function, reported as baseline arterial diameter (BAD), nitroglycerin-mediated dilation (NMD), flow-mediated dilation (FMD). The latter is commonly used to assess microvascular endothelial function via changes in finger pulse wave amplitude in response to reactive hyperemia. Indeed, both macrovascular endothelial dysfunction, as measured by flow-mediated dilation, and microvascular endothelial dysfunction have been found to be independent predictors of future cardiovascular events in large cohort studies in healthy individuals over and above traditional risk factor assessment .
Several studies have reported an association between endothelial function and particulate air pollution [9-12]. However, there is heterogeneity in the strength and direction of the reported association. Therefore, we performed a meta-analysis to evaluate the short-term exposure to particulate air pollution and endothelial function, such as BAD, NMD, FMD, and reactive hyperemia index (RHI).
Search strategy and selection criteria
PubMed, EMBASE, Cochrane library and Web of Science were searched to identify relevant articles that were published through Aug 1, 2015, according to the search strategy as follows: (“air pollution” OR “air pollutant*” OR “particle” OR “particulate matter” OR aerosol OR “PM2.5” OR “PM10” OR “Black carbon” OR “carbon black” OR sulfate OR nitrate) AND (“reactive hyperemia” OR “endothelial function” OR “endothelial dysfunction” OR “Peripheral arterial tonometry” OR “flow-mediated dilation” OR “endothelium-dependent dilatation” OR “RHI” OR “reactive hyperemia index” OR “vascular function” OR “Nitroglycerin-mediated vasodilation” OR “Nitroglycerin-mediated dilatation” OR “flow-mediated dilatation” OR “flow-mediated vasodilation”) (Table 1), with restriction of human and not review. We also considered references found in the literature search.
|Search criteria for Medline*|
|#1||"reactive hyperemia" OR "endothelial function" OR "endothelial dysfunction" OR "Peripheral arterial tonometry" OR "flow-mediated dilation" OR "endothelium-dependent dilatation" OR "RHI" OR "reactive hyperemia index" OR "vascular function" OR "Nitroglycerin-mediated vasodilation" OR "Nitroglycerin-mediated dilatation" OR "flow-mediated dilatation" OR "flow-mediated vasodilation"|
|#2||"air pollution" OR "air pollutant*" OR "particle" OR "particulate matter" OR aerosol OR "PM2.5" OR "PM10" OR "Black carbon" OR "carbon black" OR sulfate OR nitrate|
|#3||(#1) AND #2 NOT review Filters: Humans|
Table 1: Detailed search strategy. *Similar search criteria executed for EMBASE using specific Mesh terms.
The cross-sectional and longitudinal cohort studies were included to evaluate the association between particulate air pollution (PM10, PM2.5, Black carbon, sulfate and nitrate) and endothelial function as assessed by BAD, NMD, FMD and RHI. Only English language was including in our search. We excluded non-human studies, studies reporting gaseous pollutants exposure, studies reporting estimates other than absolute and percent change in outcome per change in the level of particulate air pollutants, and studies with incomplete data. There were 2073 non-duplicate abstracts searched from the four databases, which were evaluated by all the authors independently using the search algorithm. Articles relevant to our meta-analysis (n=33) were reviewed by two separate authors to identify the final studies based on our inclusion and exclusion criteria. If any disagreement, a third author determined and resolved the final results.
There were 6 cross-sectional studies and 7 longitudinal studies on the association between the particulate air pollutants and endothelial function. If a group published two or more papers based on the same study participants, only articles with more details were included in our analysis. Information extracted included citation data, authors’ names, publication year, data source, country, sample size, age distribution, sex distribution, study design, baseline exposure level, outcome measure, effect estimate, and standard error of effect estimate. For studies that reported multiple effect estimates, we extracted the estimate from the main model or model that reflected the greatest degree of control for potential confounders. For each included studies, we extracted mean change in the indexes of endothelial function.
All estimates were standardized to per 10 μg/m3 increase in PM2.5, per 1 μg/m3 for black carbon (BC), and per 1000/cm3 for particle number concentration (PNC). In all included studies, linear mixed or general linear models were used to analyze the association between air pollution and the indicators of endothelial function. Therefore, we pooled absolute and percent change (with upper and lower 95% CI) for an increment per unit in air pollutants based on the assumption of the linear association between the both factors. Effect estimates were derived from the point estimate of each separate study weighted by the inverse of the variance (1/SE2). The combined estimate was computed using the random effect model. Delayed and cumulative effects of air pollutants on endothelial function were evaluated with multiple lag patterns, (e.g., lag 0 meant the present day exposure of BP measurement, lag 1 meant the day before and so on; 5-day moving average (MA) referring to the average of the same day and the previous four days), in many studies. If there were several estimates for multiple lag patterns reported in the same article, we chose the lag pattern with the largest estimate size, which was most frequently used in all the selected studies to assess overall risk estimates. In addition, we pooled the various estimates according to lag pattern separately and only pooled estimates where more than two estimates were available. Heterogeneity between studies was detected by the forest plot, and I-squared (I2) statistic. Consistent with prior thresholds we considered an I2 statistic ≥ 50% to represent substantial heterogeneity and ≥ 75% to represent considerable heterogeneity. Meta-regression was used to assess potential sources of heterogeneity, such as year of publication, country, study design, sample size, and baseline level of exposure. Subgroup analyses were conducted to explore the impact of possible confounding factors. All statistical tests were two-sided and p values <0.05 were considered to be statistically significant. Analyses were conducted with STATA Version 12(Stata Corp., College Station, Texas, USA).
There were 2073 articles screened by title and abstract, 2224 of which were excluded, and the remaining 33 articles were reviewed by full text according to the inclusion criteria of the meta-analysis (Figure 1). Thirteen studies were considered for data extraction, a total of 11226 subjects (51% female) from 6 studies in a cross-sectional design [11-16] and 7 in a longitudinal design [9,17-22]. The number of subjects for both groups was 7934 and 3293, respectively. The mean age for the subjects in the included studies except one study  was more than 45 years old. The mean level of PM2.5 exposure among studies ranged from 6.8 to 28 μg/m3, ranged from 0.36 to 4.7 μg/m3 for BC concentration. Most of the studies focused on the short-term effect of air pollutants on endothelial function. The long-term effect was depicted only in the two studies [9,15] (Tables 2 and 3).
|Authors||Year||Location||Study period||Study design||Data source||Population||Number of participants||Age||% female|
|Briet et al. ||2007||Paris||2000-2006||Cross-sectional||study to evaluate the arterial effects of partial genetic deficiency in tissue kallikrein activity on endothelial function||healthy, nonsmoking male subjects||40||22||0|
|Brook et al. ||2011||Michigan||2005-2007||longitudinal||Non reported||non-smoking subjects||65||45||77|
|Brook et al. ||2011||Detroit||2005-2007||longitudinal||Non reported||non-smoking subjects||51||45||75|
|Karottki et al. ||2014||Copenhagen||2011-2012||cross-sectional||Non reported||non-smoking volunteers||78||55||42|
|Krishnan et al. ||2012||Six cities in US||2000-2002||longitudinal||MESA cohort||Subjects free of clinical cardiovascular disease||3040||61||49|
|Lanzinger et al. ||2014||North Carolina||2004-2005||longitudinal||Non reported||Subjects with T2DM||22||61||36|
|Liu L et al. ||2009||Ottawa||2007||longitudinal||Non reported||nonsmoking seniors||28||78||61|
|Ljungman et al. ||2014||Massachusetts||2003-2008||Cross-sectional||the Framingham Heart Study.||the participants from the study||2369||56||51|
|O’Neill et al. ||2005||Boston||1998-2002||Cross-sectional||4 clinical trials conducted at the Joslin Diabetes Center and Beth Israel Deaconess Medical Center in Boston||participants either had DM (type I or type II) or were at risk for diabetes||270||55||41|
|Schneider et al. ||2008||North Carolina||2004-2005||longitudinal||Non reported||subjects with T2DM||22||61||36|
|Wilker et al. ||2014||Boston||2001||cross-sectional||Framingham Offspring Study and Third Generation Studies||participants from Framingham study||5112||49||53|
|Zanobetti et al. ||2014||Boston||2006-2009||longitudinal||Non reported||Subjects with T2DM||64||64||50|
|Zhao et al. ||2014||Beijing||2012||cross-sectional||Non reported||Subjects with MtS||65||61||57|
Table 2: Demographic characteristics of the 13 studies. DM: Diabetes mellitus; T2DM: Type 2 diabetes mellitus; Mts: Metabolic syndrome; MESA: Multi-ethnic of atherosclerosis.
|Authors||Year||Exposure||Concentration||Method of exposure measurement||Exposure way||Outcome|
|Briet et al. ||2007||PM2.5||28||Fixed station gives hourly level of air pollutants||Short-term||FMD|
|Karottki et al. ||2014||Indoor PNC||12400||Monitored with Philips NanoTracer1000||Short-term||RHI|
|Outdoor PNC||3900||Monitored at an urban background station|
|Indoor PM2.5||11.8||gravimetrically on Fluoropore Membrane PTFE filter|
|Outdoor PM2.5||14.4||Monitored at an urban background station|
|Ljungman et al. ||2014||PM2.5||9.6||Using a tapered-element oscillating microbalance||Short-term||RHI|
|BC||0.7||Using an Aethalometer|
|PNC||20560||With a condensation particle counter|
|SO42-||3.2||Measured by x-ray fluorescence analysis of the PM2.5 filter samples|
|O’Neill et al. ||2005||PM2.5||11.5||Measured at a site established by the Harvard School of Public Health, located near the site where patients were examined||Short-term||NMD, FMD|
|Wilker et al. ||2014||PM2.5||10.9||Daily PM2.5 concentration was predicted by aerosol optical density(AOD), ground PM2.5 measurements from 78 monitoring stations, land use regression and meteorological||Long-term||FMD|
|Zhao et al. ||2014||BC||4.7||Measured by personal Aethalometer||Short-term||RHI|
Table 3: Characteristics of the cross-sectional studies included in the meta-analysis. PNC: Particle number concentration, with the unit of per cm3; BC: Black carbon, with the unit of μg/m3; The unit of PM2.5, PM10 and SO42- is μg/m3; NMD: Nitroglycerin-mediated dilation; FMD: Flow-mediated dilation; RHI: Reactive hyperemia index.
Association between PM2.5 and endothelial function
We found nine studies reporting on the association between PM2.5 and FMD, which reflected the macrovascular function as described above. The meta-analysis based on the largest estimates chosed from different lag patterns pooled as overall risk estimates showed that a 0.27% decrease in FMD was associated with per 10 μg/m3 increase in PM2.5 exposure with marginal significance (P=0.05) (Figure 2).
In addition, subgroup analysis for different lag pattern between PM2.5 and FMD has been performed, which indicated that the greatest effect was found for per 10 μg/m3 increase in 1-day lag PM2.5 exposure with 0.092% decline in FMD (P=0.014) (Figure 3). However, no significant association was found for other lag pattern and 5-day average exposure (all P>0.05). There were moderate heterogeneity between studies (I2=49.2%, P=0.046). Therefore, we performed further subgroup analysis on this issue, according to whether ambient or personal exposure, study design, the percentage of female, whether age>55 years or not and non-American or American (Figure 4 and Table 4), which demonstrated that larger effect in pooled analysis was found for American with a 0.551% decrease in FMD for per 10 μg/m3 increase in PM2.5 exposure (P=0.002), -0.457% for subjects aged less than 55 years old (P=0.048), -0.837% for the group with the proportion of female less than 50% (P=0.032), -0.357% for the group in the panel design (P=0.022). Meanwhile, meta regression in univariate analysis was conducted to find out the sources of heterogeneity and showed that the listed factors, including publication year, location, data source, population, age, sex and study design were not contributing to the heterogeneity (all P>0.05). However, multivariate meta-regression analysis was not performed, because the poor rationality resulted from few studies.
|Authors||Year||Exposure||Concentration||Method of exposure measurement||Exposure way||Outcome|
|Brook et al. ||2011||Outdoor||15.4||Monitored at a nearby state of Michigan air quality monitoring site||Short-term||BAD, NMD,FMD|
|Brook et al. ||2011||Personal||22.5||Using a modified personal DataRam mephelometer||Short-term||BAD, NMD,FMD|
|Krishnan et al. ||2012||10.6-24.7||Estimated at each participant's residence using a spatio-temporal model||Long-term||FMD, BAD|
|1-74||Based on daily central-site monitoring in each of the 6 cities||Short-term||FMD, BAD|
|Lanzinger et al. ||2014||13.6||Obtained from a monitoring site||Short-term||BAD, FMD|
|Liu L et al. ||2009||personal PM2.5M||6.30||An active personal monitor,which was a modified (pDR) that consisted of, in series, an personal data random access memory air sampling pump||Short-term||BAD, FMD|
|Indoor BC||0.36||by Aethalometers|
|Outdoor BC||0.71||by Aethalometers|
|Indoor||6.80||Measured using a DustTrak at each of the three nursing homes|
|Outdoor||15.30||Measured using a DustTrak,located close to the homes|
|Schneider et al. ||2008||13.60||Obtained at central monitoring sites||Short-term||NMD, FMD|
|Zanobetti et al. ||2014||Ambient||8.37||measured hourly at a central monitoring site||Short-term||BAD, FMD, NMD|
|Indoor||7.11||Measured by fine particle samplers|
|Home/trip-integrated (5-day)||9.18||Collect fine particles on Teflon filters to determine and BC mass concentration|
|ambient BC||0.61||measured hourly at a central monitoring site|
|HOME/trip-integrated BC||0.77||Collect fine particles on Teflon filters to determine and BC mass concentration|
|PNC||13270||measured hourly at a central monitoring site|
Table 4: Characteristics of the longitudinal studies included in the meta-analysis. PNC: Particle number concentration, with the unit of per cm3; BC: Black carbon, with the unit of μg/m3; The unit of , PM10 and SO42- is μg/m3; BAD: Baseline arterial diameter; NMD: Nitroglycerin-mediated dilation; FMD: Flow-mediated dilation.
There were 5 studies evaluating the association between PM2.5 and BAD, as well as NMD, and 2 studies for PM2.5 and RHI. The pooled estimates in the meta-analysis demonstrated no significant association in the random-effect model (P=0.40, P=0.33, and P=0.48, respectively) (Figure 5).
Association between other pollutants and endothelial function
BC, an important pathogenic component of PM2.5, has proved to be associated with cardiovascular morbidity and mortality . A 0.07 mm decrease in BAD and 2.461% decrease in NMD for per 1 μg/m3 increment in BC were observed in the pooled analysis from two studies(P=0.005 and P=0.043, respectively) [14,21,22]. Meanwhile, 5.9% decline in RHI was associated with 1 μg/m3 increment in BC according to the pooled results of the studies conducted by Ljungman et al.  and Zhao et al. , respectively (P=0.028) (Figure 6). However, we didn’t find the association between BC and FMD (P=0.928).
Limited studies on the association between PM10, sulfate and endothelial function, and the indicators of endothelial function were different for these studies, which made the difficulty for the overall risk estimate in the meta-analysis [13,13,22].
The finding of the meta-analysis based on the 13 current human studies comprising 11226 study participants has indicated that FMD is marginally and negatively associated with short-term exposure to PM2.5, with the largest effect for 1-day lag PM2.5 exposure. We also observed the association between BC and BAD, NMD, RHI, but not FMD. The effect of BC was calculated from only two or three studies. The association of other pollutants, including PM10, sulfate and nitrate,with endothelial function was not explored due to the limited studies and different indicators of endothelial function.
FMD and NMD were calculated by the extent of reactive hyperemia after the occlusion of brachial artery, which represented the endothelial function and comprised the combined effects of endothelial-dependent processes and endothelial-independent process. The former influenced the production and quenching of vasodilatory nitrogen oxide (NO), while the latter influenced vascular smooth muscle responsiveness to NO [24,25]. NMD, a measure of the change in BAD before and after administration of nitroglycerin (an exogenous source of NO), reflects autonomic vascular smooth muscle responsiveness occurring independently of endothelial NO production. A decrease in FMD but not NMD suggests an effect on endothelial function specifically, whereas a decrease in both outcomes suggests that part or all of the change is due to endothelial-independent effects. Combined with our results, the association of PM2.5 with FMD represented the dysfunction of the endothelial-dependent process, depending on the vascular condition of the subjects. However, the effect of BC on NMD found in the meta-analysis has been demonstrated the dysfunction of vascular smooth muscle responsiveness to NO. The generalization of the results should take caution due to the limited studies on BC and NMD.
The results from our pooled analysis absolutely provided the certain evidence of mechanisms of PM-mediated adverse effect on cardiovascular morbidity and mortality- endothelial dysfunction. It is a systemic process, not only reflecting atherosclerotic risk , but also serving as a prognostic marker for future cardiovascular events . As described in the meta-analysis, BAD, FMD, and peripheral arterial tonometry (PAT) (expressed as RHI) have been shown to correlate well with invasive measures. What’s more, there were 3 meta-analyses reporting the prognostic value of non-invasive endothelial function and the risk of adverse outcomes, which demonstrated that the pooled RR of overall CVD risk, cardiovascular events and all-cause mortality or only cardiovascular events per 1% increase in brachial FMD was 0.92 (0.88-0.95), 0.90 (0.88-0.92) and 0.87 (0.83-0.91), and the pooled RR per 0.1 increase in RHI for cardiovascular events was 0.85 (0.78- 0.93) [8,28,29].
The potential implication of the effect of particulate air pollution on endothelial function has deduced based on our result. The results were also verified by controlled exposure studies in a comparative way [30,31] and current air filtration-based intervention studies. Conclusion was drawn based on a randomized, double-blind, crossover study design published in 2008 by Brauner et al. that indoor air filtration significantly improved microvascular function by 8.1% . However, there are several studies with the results of no association between filtration and endothelial function . Just as the studies included in the meta-analysis with heterogeneity in results, there were several reasons to explain the potential factors, such as population sensitivity, models, the relative low baseline pollution concentrations, methods if vascular function measurement, the exposure error or indoor sources, and the composition of the particles.
There was moderate heterogeneity between studies included the meta-analysis, and the meta-regression found no factors listed explaining the source of heterogeneity according to univariate metaregression analysis. The subgroup analysis has shown that there were several potential effect modified factors, such as study design, age, gender, and American.
Although, the mechanisms of the effect of air pollution on endothelial function have been explored gradually, including the destruction of endothelium, the imbalance of vasoconstriction and vasodilation factor, such as increased endothelin-1 (ET-1), decreased nitric oxide bioavailability, inflammation with thrombosis/coagulation, as well as PM-mediated reactive oxygen species (ROS) [21,28].
There are several strengths and limitations in our meta-analysis. This is the first study to pool effect based on the current studies which have inconsistent results and the study areas are focused on the three different countries (United State, Canada, China), comprised developed and developing countries. The potential limitations as follows deserve consideration. First, the number of included studies was relatively small, particularly for BAD, NMD and RHI, which limited our ability to derive strong conclusion from these analysis and to explore the source of heterogeneity. Second, the measurement error in assessment of exposure was different in different studies, which may result in the biased estimates.
In conclusion, short-term exposure to PM2.5 was marginally and negatively associated with FMD, which reflected the imbalance of endothelial-dependent dilation. BC was associated with NMD and BMD based on a very small number of studies. The interpretation of the results based on the meta-analysis need to be careful.
Make the best use of Scientific Research and information from our 700 + peer reviewed, Open Access Journals