|Home | About | Journals | Submit | Contact Us | Français|
Fine particulate matter air pollution plus ozone impairs vascular function and raises diastolic blood pressure. We aimed to determine the mechanism and air pollutant responsible. The effects of pollution on heart rate variability, blood pressure, biomarkers, and brachial flow-mediated dilatation were determined in 2 randomized, double-blind, cross-over studies. In Ann Arbor, 50 subjects were exposed to fine particles (150 μg/m3) + ozone (120 ppb) for 2 hours on 3 occasions with pretreatments of an endothelin antagonist (Bosentan 250 mg), anti-oxidant (Vitamin C 2 g), or placebo. In Toronto, 31 subjects were exposed to 4 different conditions (particles + ozone, particles, ozone, and filtered air). In Toronto, diastolic blood pressure significantly increased (2.9 and 3.6 mm Hg) only during particle-containing exposures in association with particulate matter concentration and reductions in heart rate variability. Flow-mediated dilatation significantly decreased (2.0 and 2.9%) only 24 hours after particle-containing exposures in association with particulate matter concentration and increases in blood tumor necrosis factor-alpha. In Ann Arbor, diastolic blood pressure significantly similarly increased during all exposures (2.5 - 4.0 mm Hg), a response not mitigated by pretreatments. Flow-mediated dilatation remained unaltered. Particulate matter, not ozone, was responsible for increasing diastolic blood pressure during air pollution inhalation most plausibly by instigating acute autonomic imbalance. Only particles from urban Toronto additionally impaired endothelial function likely via slower proinflammatory pathways. Our findings demonstrate credible mechanisms whereby fine particulate matter could trigger acute cardiovascular events and that aspects of exposure location may be an important determinant of the health consequences.
Fine particulate matter air pollution < 2.5 μm in diameter (PM2.5) is a major world-wide cause of cardiovascular (CV) mortality1,2. Although PM2.5 mass is related to the degree of CV risk, specific sources (e.g. traffic), chemical components, and gases (e.g. ozone) contribute to the health consequences3-7. As gas and particle pollution are typically present together, it is important to understand their individual and combined effects1,2.
PM2.5 promotes CV events via several mechanisms1,2; however, vascular dysfunction likely plays an integral role. An imbalance in vasomotor tone and/or a related pro-hypertensive response could trigger ischemic cardiac events and contribute to the observed risk for heart failure and strokes8. Indeed, we demonstrated that a 2-hour-long exposure to concentrated ambient PM2.5 (CAP) plus ozone raises diastolic blood pressure (BP) and triggers vasoconstriction in healthy adults9,10. Diesel exhaust, a mixture of particles and gases, can also instigate endothelial dysfunction and cardiac ischemia11,12. Whilst the degree of vasoconstriction and BP elevation were both associated with the concentration of organic carbon within PM2.5 in our original studies13, some effect of ozone could not be discounted. Ozone could have directly impaired arterial function or interacted with the particles enhancing their toxicity3,14. As both are targets for regulations, it is important to elucidate the responsible pollutant(s)15.
In addition, the mechanism underlying the CV responses remained a matter of speculation9,10. Three pathways appeared most plausible at the time of this study design: pollution-induced systemic oxidative stress/inflammation, elevated endothelin (ET) levels or activity, and altered autonomic nervous system (ANS) balance2,16. The inhalation of PM2.5 has been shown to cause a systemic inflammatory response (e.g. elevation in pro-inflammatory cytokines such as interleukin-6 and tumor necrosis factor alpha). Previous studies suggest that particle-induced oxidative stress, within the lungs and/or systemically, likely plays a key role in initiating this response1,2. Additional experiments show that air pollution exposure is associated with rapid changes in ANS balance favoring sympathetic nervous system activation and parasympathetic withdrawal. Several other lines of evidence also suggest that PM inhalation can rapidly trigger an increase in circulating levels and/or bioactivity of ET within the vasculature1,2. Each of these responses may be potentially responsible for causing systemic vascular dysfunction and increasing BP by favoring pro-vasoconstrictive cellular signaling pathways after acute air pollution exposure2. Therefore, we explicitly designed this experiment in order to test the viability of each of these putative pathways.
This study aimed to determine the responsible air pollutant(s) (ozone, PM2.5, or their combination) and the most credible mechanism underlying our previously findings9,10. To meet these goals, we designed a controlled air pollution exposure study purposely involving well-coordinated experimental limbs performed concomitantly at 2 different sites. This design contributed to the increased sample size compared to previous studies9,11 and facilitated the investigation of the 2 different study aims along with our ability to discern whether responses differ between locations where PM2.5 can be dissimilar in composition and sources.
The study was approved by the human research ethics committees of St. Michael's Hospital and the University of Toronto as well as the Institutional Review Board of the University of Michigan. Subjects at both locations were healthy 18 to 50 year old non-smokers, without any CV disease or risk factor and not taking medications. We excluded subjects with a screening fasting total cholesterol >240 mg/dL or glucose >126 mg/dL.
The overall study design is illustrated in the online data supplement (please see http://hyper.ahajournals.org; Figure S1). The 2 specific site locations were selected in order to provide exposures to fine particles that differ in source and also likely chemical composition. In Ann Arbor, Michigan, exposures and CV testing were performed in the AirCare1 mobile facility adopted for humans and stationed at the University of Michigan North Campus17. Subjects were exposed to CAP + ozone for 2 hours on 3 separate occasions 2-4 weeks apart. Two hours before each exposure (1 hour prior to pre-exposure testing), subjects were given 1 randomized, double-blind oral pretreatment of Bosentan 250 mg, Vitamin C 2000 mg, or placebo. Neither the subject nor the investigative personnel were aware of the pre-treatment type during the study. In Toronto, Ontario, exposures and CV testing were performed at the Gage Occupational and Environmental Health Unit in downtown Toronto9. Subjects were exposed in a randomized, blinded fashion to 4 conditions for 2 hours (CAP + ozone, CAP, ozone, or filtered air) without pretreatments at least 2 weeks apart. Subjects were not aware of the order and could not discern the exposure type. All study personnel who performed the CV outcome measurements were blinded to the exposure-types during the study. Only the investigator responsible for generating the exposure was aware of its composition during the study period.
Subjects arrived fasting (>8 hours) to the facilities between 8-9 AM. Pre-exposure testing was first performed (1 hour duration). Afterwards, subjects underwent the 2-hour-long exposure, followed immediately by repeat testing upon completion. Subjects returned fasting the following morning for repeat testing between 8-9 AM. In Ann Arbor, subjects wore a 24 hour ambulatory BP monitor (SpaceLabs 90207 ABP Monitor) the day prior and after all exposures.
Ambient PM2.5 was concentrated to a target level of 150 μg/m3. In Toronto, CAP exposures were produced with a 2-stage Harvard virtual impactor system9. During filtered air exposures, a HEPA filter was inserted downstream of the concentrator. In Ann Arbor, CAP exposures were produced with a 3-stage Harvard virtual impactor system17. Ozone (120 ppb) was produced by an arc generator and added to the CAP airflow. Ozone was monitored continuously using a photometric analyzer (Dasibi, model 1008RS). Both human chambers were modified air-tight body plethysmographys with exposure air flows (15-20 L/min) entering the chamber via ducting ending in a face mask facilitating nasal inhalation. PM2.5 levels were monitored during exposures by a tapered element oscillating microbalance (model 1400a, Rupprecht & Patashnick, Albany, NY).
A PM2.5 filter sample was collected immediately upstream of the chambers on a 47 mm Gelman Teflon filter with a 2 μm pore size at an air flow of 8 L/min. The sample was analyzed gravimetrically for total mass on conditioned filters using a climate controlled clean room9,17.
Technicians at both sites performed all protocols using identical methodologies after combined training9. Subjects rested supine for >10 minutes prior to all testing periods in a temperature-controlled room. In Ann Arbor, the order of testing was: average of 3 supine BP levels (Omron 780), brachial artery diameter (BAD), flow-mediated dilatation (FMD), arterial compliance, nitroglycerin-mediated dilatation (NMD), and blood draws for biomarkers. In Toronto, the order of testing was: average of 3 supine BP levels (Oscar-1 or 2; SunTech), supine 10-minute electrocardiogram (ECG) using a Holter monitor which subsequently recorded continuously throughout the study, BAD, FMD, NMD, and blood draws for biomarkers.
BAD, FMD, NMD were measured using a Terason 2000 ultrasound with a 7.5-10.0 mHz linear array transducer with ECG-gated image acquisition and storage of digital loops (Teratech, Inc). Upper arm cuff inflation above the site of ultrasound image acquisition for 4 minutes was used to generate reactive hyperemia and images were obtained continuously from 50 to 120 seconds after cuff deflation. Peak FMD within this period was utilized as the study outcome for “endothelial function”. Image analyses were performed using semi-automated software (Medical Imaging Applications, Inc). Arterial compliance was measured by radial artery tonometry using the CVProfilor as described before (Hypertension Diagnostics, Inc)18.
BP and heart rate were measured during the exposures while seated within the chamber. At both sites, subjects wore the automated BP monitors on their left upper arm (Ann Arbor: Omron 780; Toronto: Oscar-1 or 2). Readings were measured at the start, at 30 minute intervals, and immediately after exposure completion while in the chamber. At both sites, subjects showed the readings obtained by the automated devices to the investigators but were blinded to the results. The average of the 2nd and 3rd BP and heart rate readings was used in Toronto (7 subjects had only a single reading which was used). In Ann Arbor, we only performed a single BP reading at each time point during the exposure and therefore only this single result was used in the analyses.
Holter monitoring was performed in Toronto with a SEER MC ambulatory digital recorder (GE Medical Systems). The 2-channel ECG data was recorded on an 8 Mb flash card and downloaded onto a MARS 8000 workstation (GE Medical Systems) to determine time and frequency domain heart rate variability (HRV)19. HRV measures were carried out by comparing a 10-min epoch at the start of all exposures while seating within the chamber to one just prior to the end of the exposures.
Screening labs were analyzed for fasting lipoproteins and glucose. In Toronto, venous blood was collected pre, post, and 24 hours post-exposures for ET-1, cytokines, complete blood count (CBC), and high sensitivity C-reactive protein (CRP). Blood samples were centrifuged and the resulting plasma (ET-1, cytokines) and serum (CRP) stored at −70°C. Whole blood was used to make smears for CBC with an automated analyzer (Coulter, model LH755, Beckman Coulter Canada, Mississauga, ON). A panel of 10 cytokines (see online supplement) that included tumor necrosis factor-alpha (TNF-α)) were analyzed using LiquiChip cytokine kits (Qiagen, CA) and a Luminex analyzer (Luminex, Austin, TX). CRP and lipids were analyzed using a Beckman Synchron LX analyzer (Beckman Coulter Canada, Mississauga, ON). ET-1 was analyzed using a QuantiGlo ET-1 Chemiluminescent Immunoassay (R & D Systems, Minneapolis, MN) and Fluostar Optima microplate luminometer (BMG labtechnologies, Durham, NC).
The study was designed and powered to assess the differential effect of the 4 exposure types in Toronto, as well as the effect of the exposures across the 3 different pre-treatments in Ann Arbor, on the outcomes of intra-chamber diastolic BP slope and change in BAD and FMD. Other outcomes are secondary analyses. Data were analyzed using SAS version 9.1 (SAS Institute, NC). In order to examine the exposure-induced change in outcomes, the post minus pre-exposure difference (Δ) was calculated for each variable at each time point. Differences between the post-exposure results versus the corresponding pre-exposure values for each outcome were tested by paired t-tests. Differences in Δ responses in the outcomes at all time points were compared with the corresponding Δ responses across the 4 different exposure types (Toronto) and the 3 pre-treatment limbs (Ann Arbor) by linear mixed model analyses. For BP measured during exposure, linear regression was used to determine the slope (mm Hg/30 mins) of the line fitted over the five BP time point measures for each subject's exposure during the 2 hour period. The slope (β) or post exposure Δs were then used as dependent variables in mixed models. Each model included a random subject effect with the fixed effects being CAP and/or ozone, and the CAP and ozone interaction. The effects of HRV and PM2.5 on diastolic BP change during exposure were examined using a mixed model that included the fixed effects: HRV exposure end minus start HRV (Δ), continuous CAP (integrated gravimetric exposure PM2.5 mass), ozone as dichotomous, and the CAP and ozone interaction. The effects of PM2.5 on FMD change during exposures were examined using a mixed model including the fixed effects: continuous CAP mass, ozone as dichotomous, and the CAP and ozone interaction. The Pearson correlation was used to test for bivariate associations between variables. Statistical significance is reported as p < 0.05.
Subject characteristics are shown in Table 1. The PM2.5 mass and ozone concentrations were not different among the exposure limbs in Ann Arbor, but significantly differed per design in Toronto (Table 2).
FMD and NMD were not impaired immediately after exposures (Table 3). However, FMD significantly decreased 24 hours post-CAP exposure compared to baseline level (Table 3). Moreover, the change in FMD that occurred 24 hours following both exposures containing CAP (CAP and CAP + ozone) was significantly different than the comparative change occurring after exposures without CAP (ozone and filtered air) (Table 3). We defined this mixed model analysis as a specific effect of “CAP-containing exposures” on outcomes (for other endpoints as well). The degree of decrease in FMD 24 hours post-exposures was significantly associated only with the 2-hour integrated gravimetric PM2.5 mass concentration (pooled results for all 4 exposure conditions: β =−2.3% per 100 μg/m3, p=0.010, n=117 by mixed model analysis accounting for ozone). Higher concentrations in CAP were thus associated with greater reductions in FMD. The changes in BAD (Table 3), BP, heart rate, and HRV measures (please see http://hyper.ahajournals.org; Tables S1 and S2) did not differ across the 4 different exposure conditions when measured at any time point outside the chamber.
WBC count and neutrophils increased immediately after “CAP-containing exposures” (Table 3). There was no significant differential change across exposure types in any other blood biomarker, including all cytokines and CRP (please see http://hyper.ahajournals.org; Table S3) or ET-1 (Table 3), except for a marginal effect on interleukin-4 likely due to chance. Interestingly, the log-normalized immediate post-exposure change in TNFα was significantly inversely correlated (r=−0.26 p=0.023) with the 24-hour post-exposure reduction in FMD (all 4 exposures pooled: n=77 observations). This means that greater post-exposure elevations in TNFα were associated with larger reductions in FMD 24 hours later. There was no other significant association among the changes in all other blood biomarkers with BP or FMD.
Diastolic BP significantly linearly increased during “CAP-containing exposures” (Figure 1), resulting in a significant 2.9 and 3.6 mm Hg elevation after 2 hours for the CAP and CAP + ozone exposures, respectively. Neither filtered air nor ozone caused a significant increase in diastolic BP (slope statistically not different from 0). The diastolic BP slope for the 2 hour period during all exposures pooled was associated in a linear mixed model analysis with reductions in several HRV components (Table 4) and with the 2 hour integrated PM2.5 gravimetric PM2.5 mass concentration (β =1.6 mmHg per 100 μg/m3 of PM2.5, p=0.01), even when accounting for the HRV changes. This means that greater increases in diastolic BP during exposure were associated with larger reductions in most HRV metrics (e.g. SDNN) and also with larger elevations in PM mass level. Although systolic BP increased after CAP-containing exposures (2.3-3.5 mm Hg, p≤0.05) and not after filtered air (1.5 mm Hg, p>0.2), this increase was not quite significant in mixed model versus filtered air and ozone (p=0.23). Ozone exposure was not associated with changes in HRV or BP.
BAD, arterial compliance, FMD, and NMD were not changed after exposures (Table 5). There was no significant differential change in any outcome in response to exposures at any time point (immediately and 24 hours post-exposures) when compared across the 3 different pre-treatment limbs (mixed model analysis).
Throughout the exposures, diastolic BP linearly increased (between 2.5 - 4.0 mm Hg after 2 hours) to a statistically comparable degree in each pre-treatment limb (Figure 2). Although there was a slight blunting of the response, the slope of the diastolic BP increase was not statistically different in the Bosentan limb compared to the other 2 pre-treatments limbs. Heart rate (slope: beats/min per 30 minutes) significantly increased to a statistically similar degree (p>0.5 by linear mixed model analysis) during all exposure limbs (placebo: 1.05 ± 0.24, p=0.001; Bosentan = 0.78 ± 0.27, p=0.003; Vitamin C = 0.99 ± 0.27; p=0.001). Systolic BP did not significantly increase.
This study provides several novel insights into the acute CV effects of air pollution exposure. First, PM2.5, not ozone, was responsible for raising diastolic BP and only transiently during the actual period of inhalation. Second, the results implicate ANS imbalance as the most plausible mechanism underlying this pro-hypertensive response. Third, PM2.5, not ozone, can additionally impair endothelial function 24 hours post-exposure. However, certain aspects of exposure location are major determinant of this response as it occurred only in downtown Toronto. Fourth, the endothelial dysfunction was triggered via a slower biological pathway, the most likely being systemic inflammation.
This is the first controlled human exposure study to investigate the effect of ozone upon the vasculature, moreover to compare the responses to those induced by PM2.5. The results provide clear evidence that it is PM2.5, not ozone, which causes both the endothelial dysfunction (Toronto) and pro-hypertensive response (both locations) within minutes-to-hours of exposure9-11. Ozone did not cause any adverse CV effects by itself, nor did it augment those induced by PM2.5. Furthermore, higher levels of PM2.5 exposure were associated with larger elevations in diastolic BP and greater reductions in FMD, supporting the linear dose-response relationship between exposure and CV risk1,2.
Our findings support the concept that aspects of exposure location (e.g. PM2.5 composition or sources) critically influence the breadth of CV responses. Particle exposures caused a similar diastolic BP elevation at both sites. On the other hand, only the fine particles from downtown Toronto, which are heavily influenced by local traffic, were capable of additionally triggering endothelial dysfunction despite the fact that mass concentrations were nearly identical between sites. Congruent with our findings, particles in Northern Scotland (low in combustion-derived compounds) were shown to have a neutral effect20, while diesel exhaust (high in combustion particles and gases) was capable of triggering endothelial dysfunction11.
At present we can only speculate about the specific particle constituents or sources accountable for the differential responses between locations, although we can rule out a role played by gases or ozone. The exposure inlet in Toronto faces a downtown urban roadway experiencing 30,000 vehicles per day21. The site is surrounded by a dense road network for several kilometers in all directions and exposures were conducted in the morning when the impact of fresh rush hour emissions on particle mass and composition was expected to be important22. In contrast, Ann Arbor is a smaller urban area with much less traffic density in the vicinity of the north campus location. The region surrounding the exposure site quickly becomes rural in the prevailing wind directions and thus the fine particles utilized in the experiments were more likely derived from long range transport of background aged aerosols23. Future analyses of the filters collected during exposures will provide more-specific information regarding the particulate constituents and sources responsible. Nonetheless, the present findings provide compelling evidence that PM2.5 derived from different locations poses varying degrees of risk to the CV system1,2.
Our findings implicate acute ANS imbalance as the most plausible mechanism for the pro-hypertensive response. It initiated and terminated rapidly and was associated with an increase in heart rate. The magnitude of diastolic BP elevation during exposures was also most strongly associated with decreases in HRV markers of parasympathetic ANS withdrawal (e.g. SDNN) (Table 4)19,24. This hypothesis is supported by the fact that human airways are lined with receptors and nerve endings that after stimulation by inhaled PM2.5 may be capable of altering reflex ANS pathways leading to a blunting of CV parasympathetic tone and a relative favoring of sympathetic activity24. Indeed, a recent study with dogs has corroborated that diastolic BP acutely increases during 5-hour long CAP inhalation. This response was mitigated by α-adrenergic receptor antagonism, thus further supporting the central mechanistic importance of the ANS25.
Although we had alternatively hypothesized that PM2.5 could raise BP due to a systemic oxidative stress-mediated impairment in basal nitric oxide-dependent bioavailabilty (favoring vasoconstriction)9-11, the BP change was unrelated to inflammatory biomarkers (Toronto) and not mitigated by a pretreatment dosage of Vitamin C (Ann Arbor) previously shown capable of acutely obviating tobacco smoke-induced oxidative stress and endothelial dysfunction26. Moreover, the time courses of the endothelial dysfunction and BP elevation were discordant in Toronto. Diastolic BP also increased in Ann Arbor without any change in FMD or arterial compliance. Therefore, it is not likely that the later-onset endothelial dysfunction was mechanistically responsible for the earlier BP increase or that oxidative stress was centrally involved. It has also been speculated that increased ET-1 bioactivity could be responsible27. However, the non-significant changes in plasma ET-1 (Toronto) were unrelated to the BP elevation. Moreover, the relatively large pre-treatment dose of Bosentan (Ann Arbor) effectively lowered pre-exposure diastolic BP, suggesting that it achieved physiologically-relevant endothelin A and B receptor blockade that should have significantly attenuated any hemodynamic actions of CAP exposure if they had occurred via this pathway. Finally, ET-mediated vasoconstriction is typically of slower onset and of more prolonged duration than the corresponding BP changes observed28.
The mechanisms responsible for the endothelial dysfunction after the CAP-containing exposures in Toronto are less clear. However, our findings could most plausibly be interpreted that it was triggered by a systemic pro-inflammatory cascade. It had the expected slower onset of occurrence than responses induced by the ANS. Moreover, the decrease in FMD was associated with an increase in post-exposure TNFα levels. Epidemiological and animal studies support this hypothesis in that PM2.5 exposure can cause an increase in a variety of inflammatory cytokines/mediators within the circulation and vasculature29,30. Although CRP and other measured blood cytokines remained largely unchanged by PM2.5, WBC count and neutrophils (also blunt markers of inflammation) increased after the CAP-containing exposures that were associated with endothelial dysfunction. It is possible that unmeasured pro-inflammatory factors derived from circulating leukocytes (e.g. myeloperoxidase as demonstrated elsewhere)31,32, the increase in TNFα itself, or other facets of inflammation that we did not measure (or accurately quantify by the cytokine tests) were responsible for impairing FMD33. Nevertheless, our results corroborate that ambient PM2.5 exposure, at least in some urban locations, is capable of triggering vascular endothelial dysfunction within a day11,32.
It is important to highlight in the context of short-term and rather low dose exposures performed on healthy humans that the adverse, yet modest, CV responses observed in this study are actually at the very least on par with past reports1,2,9,11,32. Nevertheless, the observed small degree of diastolic BP elevation and endothelial dysfunction pose little risk to healthy people. However, both are plausible instigators of ischemic events in susceptible individuals by triggering pre-existing atherosclerotic plaque instability and/or by impairing myocardial perfusion8. The BP increase could also help explain the associations between PM2.5 with strokes and heart failure exacerbations1,2. Moreover, these responses could very conceivably occur in an exaggerated manner in patients with CV risk factors or disease who have pre-existing ANS imbalance or endothelial dysfunction that render counter-balancing mechanisms less effectual12. For example, subjects with hypertension have been shown to have a greater increase in BP in response to ambient PM2.5 exposures than normotensives34. Diabetics are at greater risk for air pollution-mediated endothelial dysfunction than healthy patients35. These findings agree with observations that PM2.5 poses much greater acute risks to vulnerable individuals with pre-existing heart disease1,2. Together with our present results, the evidence supports that even shorter-term PM2.5 averages (e.g. hourly levels) than are currently regulated may need to be minimized in order to optimally protect vulnerable individuals from an acute CV event15,36.
This study was the largest human controlled air pollution exposure protocol yet completed (274 total exposures)9,11. Our well-coordinated design and uniform methods between 2 sites provided us the unique ability to compare CV responses to air pollution from different locations.
It is unclear why we did not replicate previous findings of brachial artery vasoconstriction after CAP + ozone exposures9. We either failed to identify its occurrence or this vascular territory behaved in a discordant manner between experiments. It is also uncertain why exposures elicited a greater effect upon diastolic than systolic BP. Further studies are required, but perhaps the underlying hemodynamic changes responsible reflected an isolated vasoconstriction without changes in cardiac output or arterial compliance (as suggested by the results in Table 5).
We also limited blood biomarker interpretation to the Toronto results because this was where we could account for any filtered air (placebo) effect on these secondary outcomes. The Ann Arbor site did not have a filtered air control, therefore the blood biomarkers were not analyzed after it was apparent from the results in the Toronto study that controlling for filtered air responses was essential for proper data interpretation. It is unclear why we did not observe the expected decreases in HRV after CAP-containing exposures19,24, despite the fact that reductions in HRV during all exposures pooled together were associated with the diastolic BP elevation. Perhaps the sample size was inadequate for discerning dichotomous HRV differences as the study was not powered for this outcome. Future studies that investigate the effect of exposures upon baroreflex sensitivity and/or direct microneurography measurement of sympathetic activity are warranted to help further corroborate our findings. A lack of adequate sample size and the very healthy nature of our subjects may also explain why there were no observed increases in cytokines after CAP exposures even though the increase in TNFα correlated with a reduction in FMD in the pooled analysis. Although the PM2.5 concentration was higher than typically observed over 24 hours, levels exceeding 150 μg/m3 can occur for one to two hour periods in many North American locations and are commonly encountered over even longer durations throughout developing nations37.
It is a possible limitation that gaseous pollutants (e.g. CO, NOx, SOx) were not measured during all exposures. However, these gases along with any volatile vapor phase pollutants are at (or below) ambient levels in the concentrator system9,10,13. As such, per our study design the concentrations of these pollutants were expected to be at similar and ambient levels during all exposures at both sites. They were thus not likely to differ between exposures to any meaningful degree, nor are they therefore likely to explain any of the responses. Though it is not impossible that these other pollutants (e.g. gases, vapor phase compounds) could have played a small role, the sum of our findings (particularly the independent statistical associations of BP and FMD with PM2.5 mass) along the concentrator systems utilized makes it much more likely that the fine particles were principally responsible for the observed results.
There were small differences in triglycerides, LDL-C, HDL-C and the female/male ratio of subjects between sites (Table 1). However, the values were all well within the normal range and all subjects were healthy without overt CV disease, risk factors, and/or any parameter known to alter the risk of air pollution exposure1,2. At both sites, these parameters were not associated with the FMD responses to exposures. In addition, even overt hyperlipidemia has not been shown to be a risk effect modifier to air pollution exposure1,2. Therefore, we do not believe that these small differences account for the observed FMD response differences between sites.
Our findings provide plausible mechanisms whereby even brief contact with high levels of real-world ambient PM2.5 can promote acute CV events. They also illustrate that aspects of exposure location and hence particle source and/or composition can play a critical role in determining the breadth of the adverse reactions. We believe it important to examine whether shorter (e.g. hourly) time frames of ambient PM2.5 reductions beyond existing 24-hour-long hour standards can have a positive impact upon the health of vulnerable populations34.
PM2.5 air pollution is a leading cause of world-wide mortality. Exposure to ozone is also associated with an increased risk of premature death. We demonstrated that PM2.5 exposure, not ozone, causes a significant increase in diastolic blood pressure and impairs endothelial function, but only from particles derived from an urban environment. The results implicate changes in autonomic nervous system activity as the cause for an increase in blood pressure, whereas the impaired vascular function was likely due to systemic inflammatory responses. These are important new insights into the mechanisms whereby air pollution imparts biological harm. The findings confirm that even transient contact with relevant concentrations of PM2.5 can rapidly instigate physiological responses potentially capable of triggering acute cardiovascular events in susceptible individuals. In addition to bolstering the veracity of the epidemiological associations linking air pollution with excess cardiovascular-related mortality, the findings suggest that the characteristics of the particles are likely to be important determinants of the health consequences following exposure. Overall, these study results further strengthen the justification for environmental regulations on ambient PM2.5.
HRV analyses were generously carried out at the Beth Israel Deaconess Medical Center, Boston, MA, under the direction of Dr. Murray Mittleman with support from the National Institutes of Environmental Health: P01 ES009825.
Sources of funding: This project was supported by a grant from the United States Environmental Protection Agency: 2002-STAR-G1 (CR830837) and from a National Institutes of Health General Clinical Research Center Grant: M01-RR000042; and support from Health Canada and Environment Canada (Toronto Study).