Associations between Personal Exposure to Metals in Fine Particulate Matter and Autonomic Nervous System Dysfunction among Healthy Adults

Received: April 19, 2020
Revised: June 20, 2020
Accepted: July 2, 2020

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.

Cite this article:

Liao, Y.H., Chen, W.L., Wang, C.C. and Lai, C.H. (2020). Associations between Personal Exposure to Metals in Fine Particulate Matter and Autonomic Nervous System Dysfunction among Healthy Adults. Aerosol Air Qual. Res. 20: 1842–1849. https://doi.org/10.4209/aaqr.2020.04.0156

HIGHLIGHTS

• Exposure to metals in PM_2.5 was associated with ANS dysfunction.
• Significant changes in SDNN, r-MSSD levels were related to iron in PM_2.5.
• The SDNN levels were significantly positively related to gallium in PM_2.5.

Keywords: Particulate matter; Metals; Autonomic nervous system.

INTRODUCTION

Human beings need sunlight, air, and water. Currently, air pollution, such as fine particulate matter, has become a major problem in daily life in many countries. In large cities, the major source of airborne particulate matter is emitted during the combustion of fuels. Industry emissions and vehicle emissions are common sources of fine particulate matter (PM_2.5). In addition, this kind of airborne particulate matter can absorb metals and polycyclic aromatic hydrocarbons (PAHs) (Schwela, 2000). In Taiwan, the metallic elements Fe, Cd, Pb, Zn, Cr, Mn and Ni were the largest components in airborne particulate matter and were produced by traffic and industry emissions (Fang et al., 2003).

In previous studies, exposure to airborne particulate matter was associated with death or cardiovascular risks (Riediker et al., 2004; Madsen et al., 2012; Krishnan et al., 2013; Fiordelisi et al., 2017; Fang et al., 2019). Increasing evidence implicates airborne particulate matter as a potential cause of central nervous system (CNS) dysfunction (Wang et al., 2017; Kim et al., 2018). One hypothesis for how airborne particulate matter affects the CNS is that particulate matter travels through the olfactory bulb and blood-brain barrier to the brain, with toxic effects on the CNS. In addition, particulatematter may directly or indirectly damage physiological barriers, such as the blood-brain barrier (Block and Calderon-Garciduenas, 2009). The other hypothesis is that particulate matter causes autonomic nervous system (ANS) dysfunction through an effect on the pulmonary neural reflex. In this way, particulate matter would affect one’s heart rhythm (Brook et al., 2010).

ANS activity can be evaluated by heart rate variability (HRV). Measurement of HRV is a non-invasive method to assess ANS imbalances (Moritani et al., 2005). Recent studies have suggested that air pollution is associated with an HRV reduction in elderly adults, healthy adults and infants (Adar et al., 2007; Wu et al., 2010; Mirowsky et al., 2015; Cowell et al., 2019). A decrease in HRV is related to ANS dysfunction. Moreover, metal particulate exposure affects the ANS (Chen et al., 2006). Nickel, manganese, and calcium in airborne PM were related to a decrease in HRV (Wu et al., 2012; Chuang et al., 2013). Even transient exposure to PM may cause ANS imbalances in young adults (Jia et al., 2018). Furthermore, although an effect of metal particulates on HRV has been reported (Shutt et al., 2017), no published studies have characterized the impact of airborne particulate matter and its metal components on adverse effects such as ANS dysfunction in healthy adults. Thus, we conducted a longitudinal research study to examine the associations between personal exposure to metals in fine particulate matter and ANS dysfunction in healthy adults.

METHODS

 
Participants

This was a longitudinal-design study conducted in northern Taiwan from February and June 2014. Initially, 82 young adults aged 20 and older were eligible. Participants were from districts A and B of northern Taiwan. District A (low-level exposure) had lower ambient PM_2.5 levels than district B (high-level exposure) according to the data obtained from the monitoring sites of the environmental protection administration (EPA). Healthy non-smokers were recruited. During February and June 2014, personal PM_2.5 exposure and in-PM_2.5 metal data were collected every two months. HRV was also measured three times every two months. We collected individuals’ air samples within the personal breathing zone for 24 hours (from 8:00 AM on day 1 to 8:00 AM on day 2). A self-administered questionnaire was also completed by the study subjects. An HRV assessment was recorded on the following morning (day 2). The questionnaire collected information regarding the sampling area, participants’ characteristics, cigarettes smoked, and indoor time. HRV data were collected the same day. A flow chart of the study design is shown in Fig. 1. Some participants were lost to follow-up during the study period; thus, we recruited new study participants in April and June. In total, 57 study subjects completed all 3 measurements, 17 study subjects completed 2 measurements, and 5 study subjects completed one measurement. The study was approved by the Institutional Review Board of Tri-Service General Hospital in accordance with the revised Declaration of Helsinki. All participants provided written informed consent.

Fig. 1. Flowchart of the exposure assessment.
 

PM_2.5 Personal Exposure Assessment

To assess PM_2.5 personal exposure, personal air sampling was performed by a 2.5-µm impactor (PEM; SKC Inc., PA, USA) and a pump (Gilian Gilair, Sensidyne Inc., FL, USA). We used a quartz fibre filter (2500 QAT-UP, Purtram, Conn., USA) to collect personal air samples. After the air was passed through a static neutralizer, the filters were weighed. We used a microbalance (Mettler-Toledo, MT5, Greifensee, Switzerland) with 1-µg reading to analyse the weight. The laboratory had a relative humidity of 60% and a constant temperature. The detection limit for mass concentration was 2.11 µg m^–3.

 
Metal Composition and Concentration Measurements in PM_2.5

Participants were recruited from 2 districts of an urban area of northern Taiwan. In the urban area of Taiwan, the major source of metals in PM_2.5came from vehicle emissions. Three of the most common metals in PM_2.5 were iron (Fe), nickel (Ni), and manganese (Mn) (Fang et al., 2003). Furthermore, between the 2 districts, district B was located close to industrial areas and residential areas. Because electronics technologies have expanded in industry, we were interested in the potential personal exposure to titanium (Ti) and gallium (Ga). Fe, Ti, Ni, Mn, and Ga composition and concentration measurements. To assess the participants’ exposure to heavy metals in PM_2.5, we collected personal breathing-zone air samples from every study participant for 24 hours. After sampling, a pressure bomb digestion system consisting of a 25 mL polytetrafluoroethylene (PTFE) vessel and a thermostatically controlled heating block (supplied by Berghof, Engen, Germany) was used for sample digestion (Wang et al., 1997). The Teflon filters were placed in 25 mL PTFE containers and a mixture of 3 mL nitric acid (HNO₃), 5 mL perchloric acid (HClO₄), and 2 mL hydrofluoric acid (HF) was added. The container was sealed, transferred into a high-pressure bomb, and heated at 170°C for 5 hr to facilitate complete dissolution. After digestion, the sample solution was transferred into a PTFE beaker and heated gently at roughly 80°C to allow the HF to evaporate. The residue was then diluted to 10 mL with 2% HNO₃ and measured with an ICP-MS instrument (ICP-MS, Agilent 7500cx, Agilent Technologies Inc., USA). In addition to sample analysis, a recovery test of the National Institute of Standards and Technology (NIST) standard reference materials (SRM) 1648 on blank filters was also performed. Details of operating conditions for the ICP-MS instruments were presented in a previous paper (Wang et al., 1997).

Quality assurance/quality control (QA/QC) procedures associated with these methods involved calibration, a recovery test using standard reference materials, and protocols for calculating the method detection limits (MDL). The average recoveries of metals ranged from 83 to 108 %. The relative standard deviation (RSD) of the data was less than 10%, representing a high degree of reproducibility. The detection limits for Cd, Ni, As, Pb, Cu, and Zn were 0.02, 0.20, 0.05, 2.94, 5.43, and 18.40 ng g^–1, respectively.

 
Ambient PM_2.5, Noise and Meteorological Data

We obtained ambient PM_2.5 levels every hour and temperature, noise and relative humidity data every day from one monitoring site each in districts A and B. These data were collected from the EPA monitoring sites.

HRV

To measure each HRV index, we used an HRV analyser (SA-3000P, MEDICORE CO., Korea). HRV indices included the standard deviation of normal to normal (SDNN) intervals and the square root of the mean of the sum of the squares of differences (r-MSSD) between adjacent NN intervals. Trained interviewers performed the HRV index measurements. Each time, we measured a complete five-minute segment of the NN interval, including the SDNN and r-MSSD. Namely, we calculated these HRV index on 5-min segment of NN interval we obtained.

Statistical Analyses

Statistical analyses were performed with Stata 12 software (College Station, TX, USA). Continuous data were described using means ± standard deviations, and numerical data were described using numbers and percentages. We described the distributions of PM_2.5 exposure and metals in PM_2.5 in geometric means (95% CI). To examine the trend of each measurement, trend tests were performed. To examine the distribution of the data, we used the Shapiro-Wilk normality test. The concentration of PM_2.5, metals in PM_2.5 and levels of HRV were not normally distributed (all p < 0.01). Thus, the independent and dependent variables were logarithmically (log) transformed. We obtained previous 24-hour mean measurements of the ambient PM_2.5 levels, temperature, noise, and relative humidity data to match the corresponding the HRV assessment of the study subjects according to districts A or B. Our study was a longitudinal research study, and the PM_2.5 concentrations and HRV levels were repeatedly measured three times every two months. A generalized estimating equation (GEE) with an autoregression of order 1 (AR[1]) was set to assess the effects of the concentration of PM_2.5 and metals in PM_2.5 on the levels of HRV (Zeger and Liang, 1986). We used GEE analysis to estimate Standard Errors (SEs) while taking into consideration of repeated measurements of same subject effect. The independent variables were the concentration of PM_2.5 and metals in PM_2.5, and the dependent variables were the levels of HRV. We used PM_2.5 and HRV data as continuous variable. We did not set a cut-off point for HRV levels to distinguish healthy or unhealthy. We adjusted the confounding variables of age, gender, sampling area, sampling time, day of the week, temperature, humidity, and noise. Temperature and humidity were common confounders of PM_2.5, and noise was a common confounder of HRV. A p value of < 0.05 was considered statistically significant.

RESULTS

 
Characteristics of the Study Subjects

Table 1 shows a total of 82 adults were included in the analysis, of whom 33 were male (40%) and 61% were from district A. The average age of the participants was 26.4 ± 3.6 years in district A and 21.9 ± 1.5 years in district B (p < 0.001). The mean body mass index (BMI) of the participants was 22.9 ± 3.3 kg m^–2. Among the participants, 89% did not smoke. Participants in district A spent more time indoors than those in district B (p < 0.001).

Comparison of Personal PM_2.5 Exposure in All Participants and Metals in PM_2.5 with Regard to Measurement times

The levels of PM_2.5 and metals in PM_2.5 exposure at each measurement are shown in Table 2. Personal PM_2.5 levels were highest in the first measurement (trend test, p < 0.001). Compared to district A, district B had higher personal PM_2.5 levels. In the first measurement, district B had the highest levels of nickel and manganese in PM_2.5 (trend test, p < 0.008 and p < 0.001, respectively). In the third measurement, district A had the highest level of iron in PM_2.5 (trend test, p < 0.001). In the first measurement, district A had the highest level of gallium in PM_2.5 (trend test, p < 0.001).

Comparison of HRV in All Participants with Regard to Measurement Times

The HRV indices of the participants are shown in Table 3. In the third measurement, the Log₁₀ SDNN and Log₁₀ r-MSSD levels were higher in district B than in district A. In the first measurement, district A had the highest level of Log₁₀ SDNN (trend test, p < 0.007).

Relationships among Personal PM_2.5 Exposure, Metals in PM_2.5 and HRV

There were no significant associations between Personal PM_2.5 exposure levels and Log₁₀ SDNN and Log₁₀ r-MSSD levels (Table 4). However, after adjusting for confounding variables, significant changes in the Log₁₀ SDNN and Log₁₀ r-MSSD were related to unit changes in Log₁₀ iron in PM_2.5 (β = –0.033, 95% CI = –0.060 to –0.0056, p < 0.05) and (β = –0.041, 95% CI = –0.075 to –0.0076, p < 0.05), respectively. Furthermore, the Log₁₀ SDNN levels were significantly positively related to Log₁₀ gallium in PM_2.5 (β = 0.054, 95% CI = 0.0064 to 0.10, p < 0.05). Namely, for every 1% increase in the Log₁₀ iron in PM_2.5, our Log₁₀ SDNN and Log₁₀ r-MSSD decreases by about 0.03% and 0.04%, respectively. For every 1% increase in the Log₁₀ gallium in PM_2.5, our Log₁₀ SDNN increases by about 0.05%.

DISCUSSION

 We conducted a longitudinal study that investigated the effects of heavy metals in the airborne particulate matter on the ANS of young adults. Participants in district A were older and spent more time indoors than those in district B. The Log₁₀ SDNN level continually decreased in district A during the three measurements. After we adjusted for determinants of airborne particulate matter exposure such as age, gender, sampling area, sampling times, temperature, and relative humidity, significant relationships between iron and gallium in PM_2.5 and HRV were observed.

Result Validity

This is the first publication to establish the relationship between metal particulates and decreased HRV among young, healthy subjects. There are several strengths in this study. This was a longitudinal-design study; thus, we were able to observe changes over time. Assessments of personal PM_2.5 exposure and metals in PM_2.5 were conducted. Since comorbidity and smoking may influence the HRV of the study subjects, we recruited healthy, non-smoking college students from two districts of northern Taiwan. We adjusted for sampling temperature and humidity, which are common covariables of particulate matter, and adjusted for sampling noise, which is common covariable of HRV. 

Synthesis of Previous Knowledge

PM_2.5 was highest in the first measurement (winter) in the present study. Because of northeastern monsoons in winter, Chinese haze may drive PM_2.5 towards northern Taiwan. Chinese haze consists of air pollutants (Zhang and Samet, 2015; Li et al., 2016). District B had higher PM_2.5levels than district A in all 3 measurements. In the first and second measurements, district B had significantly higher PM_2.5 levels than district A (p < 0.001 and p < 0.001, respectively). In the third measurement, district B also had higher PM_2.5 levels than district A, although the difference was not statistically significant (p = 0.21). District B comprises industrial areas and residential areas. By contrast, district A comprises commercial areas and residential areas. Thus, district B had higher PM_2.5 levels than district A. We also collected ambient PM_2.5 concentrations from the EPA monitoring site of district A and B. The ambient PM_2.5 levels and personal PM_2.5 exposure levels were moderately correlated (Spearman's rho = 0.58, p < 0.001). Furthermore, significant associations were not observed between the ambient PM_2.5 levels and study subjects’ SDNN and r-MSSDlevels (p = 0.19 and p = 0.43, respectively).

Most metals in PM_2.5, such as titanium, nickel, and gallium, were higher in winter than in spring and summer. These may be due to Chinese haze, wind direction and topography. In our study, iron levels in PM_2.5 were higher in summer than in spring and winter, unlike other metals in PM_2.5. Among the metals in PM_2.5, iron had the highest levels during the three sampling periods. Fang et al. (2003) found that the metallic element iron was one of the most abundant atmospheric aerosol particles in Taiwan. Since iron is also one of the most abundant metals in crustal particles, it likely came from dust and unpaved roads. Thus, iron in PM_2.5 was seldom affected by monsoons and did not show increased levels in winter (Li et al., 2016; Chang et al., 2018). Heavy metals in ambient particulate matter come from different sources. For example, aluminium, iron, sodium, magnesium, potassium, calcium and gallium are crustal elements. On the other hand, nickel, zinc and lead are anthropogenic elements (Geiger and Cooper, 2010). In Taiwan, Chen and Lin (2015) also found that iron in PM_2.5 mainly came from soil dust and crust.

In the first measurement, the SDNN of the total study subjects’ (district A and B) median was 52.58 msec (25–75 percentile = 40.40–63.73 (msec)) and the r-MSSD of the total study subjects’ median was 35.75 msec (25–75 percentile = 22.80–47.50 (msec)). In the second measurement, the SDNN of the total study subjects’ median was 51.89 msec SDNN = standard deviation of all normal to normal intervals; (25–75 percentile = 36.36–67.58 (msec)) and the r-MSSD of the total study subjects’ median was 34.42 msec (25–75 percentile = 23.23–49.24 (msec)). In the third measurement, the SDNN of the total study subjects’ median was 43.69 msec (25–75 percentile = 32.97–57.91 (msec)) and the r-MSSD of the total study subjects’ median was 28.09 msec (25–75 percentile = 21.86–44.47 (msec)). In previous studies, SDNN levels below 50 msec represented unhealthy and r-MSSD levels below 42 msec represented higher CV risk. Previous studies that compared short-term to 24-hour measurements showed that the HRV index of short-term HRV measurements may be lower than that of 24-hour measurements (Lin et al., 2005; Shaffer and Ginsberg, 2017).

In the present study, a significant decrease in HRV was related to an increase in iron particulates. The SDNN levels of HRV were significantly positively related to gallium particulates. A change in HRV represents an ANS imbalance and could cause an increased risk of mortality (Electrophysiology, 1996). Higher SDNN levels represented a lower risk of mortality, whereas SDNN levels below 50 msec represented an unhealthy status. Also, lower r-MSSD levels represented a higher risk of death (Shaffer and Ginsberg, 2017). In a randomized crossover study near steel plants, Shutt et al. (2017) found that the SDNN intervals of participants who were exposed to possible air pollution and metal elements significantly decreased. Chen et al. (2015) found that exposure to metal particulates may cause adverse autonomic responses in high-cardiovascular risk workers. In a previous study, many elements were associated with changes in HRV. In a panel study, Wu et al. (2012) found that calcium in airborne particulate matter decreased the r-MSSD of HRV. Cavallari et al. (2008) found that manganese in PM_2.5 was associated with night-time r-MSSDs of HRV in boilermakers. In an epidemiologic study, lead and vanadium concentrations in PM_2.5 were associated with changes in HRV in boilermakers (Magari et al., 2002). A Taiwanese study found that metal elements such as nickel may regulate HRV by increasing oxidative stress and causing inflammatory responses (Chuang et al., 2013). Gallium is present in only trace amounts in the natural environment. Currently, potential exposure to gallium is receiving attention due to expanding electronics and energy technologies (White and Shine, 2016). There is no established evidence to explain its effects on the human body; thus, it may need further investigation.

There were limitations in this study. First, we used sampling noise for adjustment rather than personal noise dosimeter data. In addition, the measurements were repeated in February, April and June over a six-month study period. As such, lagged effects could not be explored. Third, in our study, we did not collect data on personal exposure to other air pollutants and PM_2.5 components. Thus, various factors may still confound the associations between heavy metals in PM_2.5 and ANS dysfunction. Fourth, we used 24-hour mean PM_2.5 levels, metals in PM_2.5 concentrations and 5-minute mean SDNN and r-MSSD levels to investigate the associations between heavy metals in PM_2.5 and ANS dysfunction. Although the SDNN represented the overall HRV index clinically, the associations between heavy metals in PM_2.5 and ANS dysfunction cannot be explored throughout due to a lack of information on the 24-hour mean SDNN and r-MSSD. Finally, we did not differentiate indoor airborne particulate matter from outdoor airborne particulate matter.

CONCLUSIONS

In summary, elevated levels of iron in PM_2.5 are associated with decreases in both the SDNN intervals and the r-MSSDs between adjacent NN intervals of HRV. In addition, the levels of gallium in PM_2.5 were positively associated with the SDNN of HRV. The health effects of metal particulates should not be underestimated.

FUNDING

This study was supported by grants (NSC102-EPA-F-003-001 and 103-2314-B-016-007) from the Environmental Protection Administration and National Science Council of Taiwan, Republic of China and the Ministry of Science and Technology.

 
CONFLICT OF INTEREST

The author(s) declare no competing interests.

ACKNOWLEDGEMENTS

The authors wish to thank Dr. Saou-Hsing Liou for the developing research project of this study.

Aerosol Air Qual. Res. 20 :1842 -1849 . https://doi.org/10.4209/aaqr.2020.04.0156