INTRODUCTION

Previous studies have examined changes in heart rate variability (HRV*) and repolarization associated with increased particulate matter (PM) concentrations on the same and previous few days. However, few studies have examined whether these health responses to PM occur within a few hours or even less. Moreover, it is not clear whether exposure of subjects to ambient or-controlled PM concentrations both lead to similar health effects or whether any of the subjects' individual characteristics modify any of their responses to PM. The aims of the cur- rent study were to investigate whether exposure to PM was associated with rapid changes (< 60 minutes or con- current hour up to a delay of 6 hours) in markers of car- diac rhythni or changes in total antioxidant capacity (a marker of protection against oxidative stress) and whether any PM effects on cardiac rhythm markers were modified by total antioxidant capacity, age, obesity, smoking, hypertension, exertion, prior myocardial infarction (MI), or medication.

METHODS

We obtained data from a completed study in Augsburg, Germany (a panel study in N= 109 subjects, including a group with type 2 diabetes or impaired glucose tolerance [IGT; also known as prediabetes]) and a group of other- wise healthy subjects with a potential genetic susceptibil- ity to detoxifying and inflammatory pathways (Hampel et al. 2012b), as well as three completed studies in Rochester, New York (the REHAB panel study of N= 76 postinfarction patients in a cardiac rehabilitation pro- gram [Rich et al. 2012b]; the UPDIABETES study of con- trolled exposure to ultrafine particles [UFPs, particles with an aerodynamic diameter < 100 nm] of N = 19 patients with type 2 diabetes [Stewart et al. 2010; Vora et al. 2014j; and the UPCON controlled-exposure study of concentrated UFP exposure in N = 20 young, healthy, life- time nonsmokers). Data included 5-minute and 1-hour values for HRV and repolarization parameters from elec- trocardiogram (ECG) recordings and total antioxidant capacity measured in stored blood samples. Ambient con- centrations of UFPs, accumulation-mode particles (AMP, particles with an aerodynamic diameter of 100-500 nm), fine PM (PM2.5, particles with an aerodynamic diameter 2.5 pm), and black carbon (BC) were also available. We first conducted factor analyses in each study to find subgroups of correlated ECG outcomes and to reduce the number of outcomes examined in our statistical models. We then restricted the statistical analyses to the factors and representative.outcomes that were common to all four studies, including total HRV (measured as the standard deviation of normal-to-normal [NN] beat intervals [SDNNj), parasympathetic modulation (measured as the root mean square of the successive differences [RMSSD between adjacent NN beat intervals), and T-wave morphol- ogy (measured as T-wave complexity). Next, we used addi- tive mixed models to estimate the change in each outcome associated with increased pollutant concentrations in the . concurrent and previous 6 hours and with 5-minute inter- vals up to the previous 60 minutes, accounting for the correlation of repeated outcome measures for each subject and adjusting for time trend, hour of the day, temperature, relative humidity, day of the week, month, and visit number. Because multiple comparisons were an issue in our. analyses, we used a discovery-and-replication approach to draw conclusions across studies for each research question.

RESULTS

In the Augsburg study, interquartile range (IQR) increases in UFP concentrations lagged 2 to 5 hours were associated with 1%-3% decreases in SDNN (e.g., lagged 3 hours in the group with a genetic susceptibility: -2.26%; 95% confidence interval [CI], -3.98% to -0.53%). In the REHAB study, similarly, IQR increases in UFP concentra- tions in the previous 5 hours were associated with < 3% decreases in SDNN (e.g., lagged 1 hour: -2.69%; 95% CI, -5.13% to -0.26%). We also found decreases in SDNN associated with IQR increases in total particle count-(a surrogate for UFP) in the UPDIABETES study (lagged 1 hour: -13.22%; 95% CI, -24.11% to -2.33%) but not in the UPCON study. In the Augsburg study, IQR increases in PM2.5 concen- trations in the concurrent hour and lagged 1-5 hours, AMP concentrations lagged 1 and 3 hours, and BC con- centrations lagged 1-5 hours were associated with -1%-5% decreases in SDNN (e.g., PM2.5 lagged 2 hours in the group with diabetes or IGT: -4.59%; 95% CI, -7.44% to -1.75%). In the REHAB study, IQR increases in PM2.5 concentrations lagged 5 and 6 hours and AMP concentra- tions in the concurrent hour and lagged up to 5 hours were associated with 1%-2% decreases in SDNN (e.g., PM2.5 lagged 4 hours: -2.13%; 95% CI, -3.91% to -0.35%). In the Augsburg study, IQR increases in PM2.5 concen- trations in the concurrent hour and BC lagged 1 and 6 hours were associated with 3%-7% decreases in RMSSD (e.g., PM2.5 concurrent hour in the group with diabetes or IGT: -7.20%; 95% CI, -12.11% to -2.02%). In the REHAB study, similarly, increases in PM2.5 concen- trations lagged 4 to 6 hours-though not AMP or BC con- centrations at any lag hour-were associated with -2.5%-3.5% decreases in RMSSD (e.g., PM2.5 lagged 5 hours: -3.49%; 95% CI, -6.13% to -0.84%). We did not find consistent evidence of any pollutant effects on T-wave complexity in 1-hour recordings. For 5-minute record- ings, there was no consistent evidence of UFP effects on SDNN, RMSSD, or T-wave complexity at any 5-minute interval within 60 minutes. We further concluded that these replicated hourly effects of UFP and PM2.5 on short-term measures of SDNN and RMSSD generally did not differ between the groups in the studies (i.e., type 2 diabetes, pre-diabetes/IGT, post- infarction, and healthy subjects). Last, we found no con- sistent evidence of effects of any pollutant on total anti- oxidant capacity and no consistent evidence of modification of our PM2.5-outcome associations by any of the potential effect modifiers.

ONCLUSIONS

Increased UFP concentrations were associated with decreased SDNN in both of the panel studies and one of the two controlled-exposure studies. We also found that decreased SDNN was associated with both increased PM2.5 and AMP concentrations in the previous 6 hours in the panel studies and that decreased RMSSD was associ- ated with increased PM2.5 concentrations in the previous 6 hours in the panel studies. We therefore concluded that the research questions were replicated. Our findings suggest that both UFPs and PM2.5 are associated with autonomic dysfunction within hours of exposure, which may in part. explain the previously reported risk of acute cardiovascular events associated with increased PM in the previous few hours. Despite the heterogeneity of the study populations,and protocols, our findings provided consistent evidence for the induction of rapid pathophysiological responses by UFPs and PM2.5- The absence of consistent associations between UFPs, PM2.5, and these outcomes when examining shorter time intervals indicates that the 5- to 60-minute responses may be less pronounced than the responses occurring within hours. However, the findings from the 5-minute intervals may have been affected by the variety of proto- cols and conditions from study to study as well as by the potential effects of underlying diseases (e.g., healthy indi- viduals versus individuals with diabetes or a recent cor- onary artery. event), physical activity, circadian rhythms, stress, and/or medications.