Increases in daily levels of particulate air pollution (PM) measured at fixed outdoor monitoring sites have been associated with an increased risk for cardiovascular hospitalisation and mortality in numerous epidemiological studies.1 2 The associations have been especially strong among people with pre-existing cardiovascular disorders,3 4 and have been stronger for fine particles (PM_2.5, particles <2.5 µm in aerodynamic diameter) than for larger size fractions.5 Vehicular traffic and stationary combustion sources are responsible for a major part of fine particle emissions and there is increasing epidemiological evidence that particles from these sources are especially harmful.6 7 Ultrafine particles (<0.1 µm) are fresh combustion particles and have been found to be damaging, especially in toxicological studies.8 9

Cardiovascular diseases can be considered in essence to be inflammatory diseases, and correspondingly the hypothesised main pathway from particle exposure to clinically observable acute cardiovascular effects involves both local and systemic inflammation.10 Particles originating from combustion sources contain transition metals, quinones and polycyclic aromatic hydrocarbons which have oxidative properties.9 11 Oxidative stress caused by inhaled ambient particles has been associated in controlled human exposure studies with the induction of pulmonary inflammation.12 This localised inflammation in turn may develop into systemic inflammation impairing coronary endothelial function.10 Daily increases in particulate air pollution have been associated with increased circulating levels of inflammatory and coagulatory markers in epidemiological studies.13 14 However, exposure to particles has also been observed to have an effect on heart rate variability, a measure of autonomic control of the heart.15 16 It is not clear whether exposure to particles affects heart rate variability via systemic inflammation or by direct activation of neural receptors in the lungs.

Most epidemiological studies have examined the health effects of average PM levels over a few days before hospitalisation or death. However, some studies have associated hourly changes in outdoor PM levels with changes in heart rate variability and repolarisation dynamics, suggesting involvement of an additional immediate physiological response.17 18 Consistent with this, 1 h exposure to traffic has been associated with the onset of myocardial infarction.19 However, the effects of stress, noise and traffic-related air pollution could not be separated in the study. Exposure to particles has been hypothesised to trigger myocardial infarction by altering the autonomic tone of the heart, which would enhance the instability of a vascular plaque or initiate cardiac arrhythmias.10

We studied people with coronary heart disease to determine whether there is electrocardiographic evidence of an immediate (within hours) effect of PM_2.5 and ultrafine particles on the occurrence of exercise induced ST segment depression, an indicator for myocardial ischaemia. Daily PM_2.5 exposures have been shown to correlate longitudinally reasonably well with central outdoor measurements,20 although less is known about correlations at the hourly level. Thus, personal measurements of PM_2.5 were conducted for 22 h before a clinic visit to eliminate error in exposure estimation. We have previously demonstrated, using the same patient data, that the 24 h levels of outdoor PM_2.5 and ultrafine particles 2 days before a clinic visit are associated with ST segment depression.21

METHODS

Forty one subjects with coronary artery disease were followed-up with biweekly clinic visits and using personal and outdoor measurements of particulate air pollution for 4 months in Helsinki, Finland, as part of the European ULTRA study. The main inclusion criteria for the study were: a self-report of a doctor-diagnosed coronary artery disease, being a non-smoker, and age >50 years. The study population has been described in detail in a previous publication.21 All methods used in the study were conducted according to standard operating procedures.22

The occurrence of ST depressions >0.1 mV during a 6 min submaximal exercise test was used as the end-point in the current study. Ambulatory electrocardiograms were recorded with Oxford MR-63 tape recorders (Oxford Instruments, Abington, UK), and the recordings were analysed with the Oxford Medilog Excel II system (V7.5, Oxford Instruments) in one core laboratory.22 The study complies with the Declaration of Helsinki; the study protocol was approved by the ethical committee of the National Public Health Institute and written consent was obtained from all study participants.

During the winter/spring of 1999 (7 January–30 April), exposure to fine particles was measured in 1 min intervals during the 24 h preceding a clinic visit. The personal sampling system, build into an aluminium case, contained (in series) a PM_2.5 cyclone, a data logging photometer (pDR-1200 X, MIE, Bedford, MA, USA), a filter holder with a Teflon filter for gravimetric analyses of particles on a daily level, and a pump (AFC400S, BGI, Waltham, MA, USA).23 After every measurement, the study participants filled out a questionnaire on time spent in different microenvironments, and on cleaning and cooking activities.

Central outdoor hourly PM_2.5 concentrations were measured using the beta-attenuation method (Eberline FH 62 I-R, ESM Andersen, Erlangen, Germany) and the number concentration of ultrafine particles with the Electric Aerosol Spectrometer.24 All participants lived within 5 km of the measurement site.

Measurements of hourly and daily outdoor and personal particle concentrations were used in the analyses. Hour 0 was defined as the hour of a clinic visit, and thus particle concentration at lag 1 represented the exposure (or outdoor concentration) during the hour preceding a clinic visit. Further, 4 h personal PM_2.5 concentration was calculated as the average of hours 1–4, 8 h average took into account hours 1–8, and so on. The maximum averaging time used for personal measurements was 22 h, which was the length of the shortest measurements. The longitudinal correlation between personal and outdoor PM_2.5 was calculated as the median of subject specific correlation coefficients. Spearman’s correlation coefficients were used because of the rightly-skewed, non-normal nature of personal PM_2.5 data.

Only data from those with at least two successful visits were included in the analysis. Data were analysed using logistic regression and the statistical software R. A basic confounder model was built first without including particulate matter.22 The model included a dummy for each subject and linear terms for long-term time trend, 4 h average temperature and relative humidity, and difference in heart rate between a 5 min rest period in the supine position and the exercise test. We used penalised thin plate regression splines and the mgcv-procedure (v 1.3–12) in the generalised additive models framework to evaluate possible non-linear effects of confounders.25 The estimated degree of freedom for the smooth terms was less than 1.5 for all covariates, and they were considered to be linear. Consequently, generalised linear models were used in the final analyses.

The effect of extreme PM concentrations was evaluated by calculating an outlier concentration corresponding to three times the standard deviation of each averaging time, excluding observations higher than the value, and then repeating the procedure once.

For the analyses we standardised the measured photometric PM_2.5 concentrations by multiplying hourly concentrations with the average photometric to gravimetric PM_2.5 ratio during the specific 24 h period.

RESULTS

ST segment depressions during the exercise test were successfully recorded during 179 clinic visits for which personal hourly PM_2.5 data were also available. Thirteen people had visits both with and without ST segment depressions (31 ST events). There were 223 visits with outdoor PM_2.5 data, and during these visits 17 people had variation in the outcome (45 ST events). Typically clinic visits started between 10 and 11 am (median time).

The mean age of the study panel was 68 years (table 1). More than half of the study participants had had myocardial infarction, and roughly as many used daily β-receptor antagonists.

Personal PM_2.5 concentrations were lower than outdoor concentrations, and the difference increased with increasing averaging time (table 2). Exclusion of extreme personal PM_2.5 concentrations considerably decreased the maximum concentrations. Using the exposure questionnaires, the extremes were all linked to indoor PM sources: four out of six peaks (affecting multiple lags) were caused by cooking and two by environmental tobacco smoke. As we were interested mainly in the effects of outdoor air pollution, the results of the effects of air pollution on ST segment depressions are given without the extremes. Outdoor ultrafine particle concentrations decreased with increasing averaging time. One extreme value was identified among 1 h ultrafine particle concentrations.

The longitudinal correlations between personal and outdoor PM_2.5 were moderate for daily (r = 0.80) and 1 h average concentrations prior to a clinic visit (r = 0.70) but lower for other averaging times (r = 0.50–0.60) (table 3). Personal daily averages correlated poorly with the 1 h personal exposure just before a clinic visit (r = 0.54) but better with longer average exposures (r = 0.70–0.84). All outdoor PM_2.5 averages inter-correlated at least moderately (r = 0.69–0.98).

Spearman’s correlation coefficients between daily (1 h) outdoor ultrafine particle concentrations and outdoor and personal PM_2.5 measured just before a clinic visit were 0.32 (0.1) and 0.08 (0.08), respectively (not presented).

The 1 h average personal PM_2.5 exposure prior to a clinic visit was significantly associated with the occurrence of ST segment depressions, and the odds ratio for 4 h average concentrations was also considerable although not statistically significant (table 4). The 4 h average outdoor PM_2.5 concentration was statistically significantly associated with ST segment depressions, whereas daily outdoor PM_2.5 concentration was not associated with the end-point (table 4).

At first, the regression models for ultrafine particles did not converge. The problem was solved by only including in the models people having variation in the outcome. Ultrafine particles were not associated with ST segment depressions, and exclusion of the extreme 1 h value did not change the results.

Looking at individual hourly lags of outdoor PM_2.5 (fig 1), the only statistically significant odds ratio at the 5% level was observed at lag 2, but odd ratios were consistently elevated until lag 10. Results for personal PM_2.5 (not presented) were basically similar, but confidence intervals were somewhat larger, partly because of fewer measurements.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Odds ratios between 1 h outdoor PM_2.5 concentrations and ST segment depressions >0.1 mV. Odds ratios were calculated for an increase of 10 µg/m³ in PM_2.5. Vertical lines represent 95% confidence intervals.

DISCUSSION

Among elderly people with coronary heart disease, we observed an increased risk of exercise-induced ST segment depression in association with hourly variation in personal PM_2.5 exposure and outdoor levels over the 24 h before the exercise test. The associations were statistically significant for 1 h PM_2.5 exposure and 4 h outdoor concentration prior to the test. Variation in outdoor ultrafine particle concentrations was not associated with ST segment depressions during the same 24 h period.

In the present study we observed an association between exercise-induced ST segment depressions and PM_2.5 exposure and outdoor concentrations during the few hours before ST segment recording. Gold et al26 have reported an association among elderly subjects between ST segment depressions and outdoor levels of black carbon, an indicator for particles originating from combustion, in the 12 h period before ST segment recording. Although there was little evidence of an effect of black carbon on ST segment depressions during a stress test, an effect was observed during the recovery period. However, in their study, there was no evidence on the effects of fine particles or on delayed effects of black carbon.

We have previously shown an association between 2 days’ lagged outdoor air pollution levels, including PM_2.5 and ultrafine particles, and the occurrence of ST segment depressions using the same patient data.21 Taken together, our results suggest that there is both an immediate (within hours) and delayed (within days) effect of PM_2.5 on myocardial ischaemia, as indicated by an increased risk of ST segment depressions. Our results are thus in line with a previous study by Peters et al,27 where outdoor PM_2.5 levels were suggested to have independent immediate and delayed effects on the onset of myocardial infarction.

Both the initiation and progression of atherosclerosis and the onset of myocardial infarction and other acute forms of cardiovascular disease have been found to include an important inflammatory component.28 The effect of particulate air pollution on ST segment depression at 2-day lag could be explained by increased pulmonary and resulting systemic inflammation increasing blood viscosity and/or impairing endothelial function leading to vasoconstriction, both of which are mechanisms which would decrease oxygen supply to the heart.29 However, it is unclear whether systemic inflammation could explain the rapid, transient increase in the occurrence of ST segment depressions after short-term PM_2.5 exposure. Instead, the effect may be linked to changes in autonomic nervous control of the heart induced by particulate air pollution. A sudden decrease in heart rate variability has been observed to precede ischaemic events.30

In controlled human exposure studies, even hourly exposure to PM has been associated with a decrease in heart rate variability among elderly subjects.31 Heart rate variability and systemic inflammation have been proposed to be inversely associated,32 but the observed rapid response in the current study is potentially explained by direct induction of airway-mediated autonomic reflexes by pulmonary oxidative stress. Reactive oxygen species have recently been proposed to mediate the effect of particle exposure on the high frequency component of heart rate variability.33 The following sympathetic stress may cause rapid vasoconstriction of coronary arteries or increases in cardiac demand. β-Blockers increase parasympathetic (vagal) activity and could thus alleviate the stress.34 Consistent with this, we have previously observed stronger effects of particulate air pollution on ST segment depressions among people not taking β-blockers.21

Hourly number concentrations of ultrafine particles, a marker for traffic exhaust, were not associated with ST segment depressions, although a delayed effect was observed in our earlier study.21 It is likely that greater exposure misclassification on an hourly than on a daily level at least partly explains the discrepancy. In future studies, personal measurements of ultrafine particles should be conducted. The observation of an effect of very short-term PM_2.5 exposure on myocardial ischaemia has implications concerning local traffic emissions, because high ambient PM concentrations are encountered especially during traffic peaks. Exposure to in-vehicle PM_2.5 has recently been reported to be more strongly associated with heart rate variability and inflammatory markers than ambient or even roadside PM_2.5,35 and just 1 h exposure to traffic has been associated with myocardial infarction.19 In a recent controlled exposure study among men with stable coronary heart disease,36 brief exposure to diesel exhaust was associated with increased ischaemic burden quantified by ST segment analyses, which also supports our finding of a prompt effect of PM.

A major strength of the current study was the use of personal measurements for PM_2.5. By linking actual exposures to ST segment depressions, we avoided exposure misclassification (inevitable for outdoor measurements), which is caused by varying rates of infiltration in the home and the spatial variability of outdoor PM due to the presence of major roads.37 38 However, personal PM_2.5 levels were not substantially more strongly associated with ST segment depressions than were outdoor measurements. This observation can be explained if we assume that outdoor PM_2.5 is mainly responsible for the effect, in which case indoor sources affecting personal PM_2.5 levels only blur the underlying association and cancel some of the benefits of using personal measurements. The importance of indoor sources on PM_2.5 exposure became evident when we looked at extreme exposures, which were all linked to indoor sources. However, it should be noted that all the study participants lived within 5 km of the PM_2.5 outdoor measurement site. In most studies, study populations reside in much larger areas, in which case the benefits of personal measurements are likely to be clearer.

The main limitation of the study was the small sample size (41 subjects), which resulted in few subjects with variation in ST segment depressions, even when the study panel consisted of people more likely to experience ischaemia than the general population. As a consequence, confidence intervals were wide in the analyses, and one should clearly not focus on the exact odds ratios but on general trends. Small sample size may also in part explain the apparent lack of effect of ultrafine particles. The photometers used for personal measurements have their own limitations, caused mainly by the sensitivity of results to changes in relative humidity and particle density.39 We have previously reported a good correlation between gravimetric (filter collection method) and photometric PM_2.5 concentrations on a daily level.23 In the present study, we utilised the gravimetric method, the de facto standard for particle mass measurements, in the standardisation of the photometric concentrations in order to decrease measurement error.

In conclusion, hourly increases in exposure to fine particles were associated with a rapid increase in the risk of ST segment depressions. Together with other recent studies, this observation suggests that ambient particles have both an immediate and a delayed effect on myocardial ischaemia. The mechanisms of the two types of effects may differ.