Associations between Children's Exposure to PM2.5 and their Serum Inflammatory Responses in Taiwan

Received: July 31, 2022
Revised: October 19, 2022
Accepted: October 30, 2022

 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:

Tseng, W.J., Lu, J.H., Chao, H.R., Tsai, M.., Chang, Y.T., Wang, L.J., Chen, C.C., Mansor, W.N.W., Jalaludin, J., Wang, C.L., Tsai, Y.I. (2022). Associations between Children's Exposure to PM_2.5 and their Serum Inflammatory Responses in Taiwan. Aerosol Air Qual. Res. 22, 220288. https://doi.org/10.4209/aaqr.220288

HIGHLIGHTS

• PM_2.5 had a positive association with children’s inflammatory responses in serum.
• PM_2.5 affected activation of tumor necrosis factor (TNF-α) in school children’s serum.
• School children living in Traffic-related-air-pollution PM_2.5 had the high serum TNF-α level.

Keywords: PM2.5, Tumor necrosis factor (TNF-α), Inflammatory responses, Traffic-related air pollution (TRAP), Health risk

1 INTRODUCTION

Airborne particulate, particularly respirable or fine particle matter (PM) whose aerodynamic diameter is less than 2.5 µm (PM_2.5), has been recognized as the main contribution to the worst air pollution in decades. The global concern towards PM_2.5 pollution due to the increase in adverse health effects including heart attack, cardiovascular disease, pulmonary disease including chronic obstructive pulmonary disease (COPD), lung cancer, chronic kidney disease (CKD), diabetes mellitus (DM), and neurological disorders is increasing (Shah et al., 2013; Meier et al., 2014; Wang et al., 2015; Feng et al., 2016; Kao et al., 2019; Shou et al., 2019; Grande et al., 2021; Li et al., 2021). PM_2.5 is emitted from vehicular engines, industrial processing, open biomass burning, chemical industries, cooking fume, resuspended road dust, incomplete combustion from fossil fuels, and agricultural and construction processing (Fann et al., 2012; Aunan and Wang, 2014; Loftus et al., 2015; Chao et al., 2018; Mukherjee and Agrawal, 2018; Chen et al., 2019; Yu et al., 2021). PM_2.5 from the long-range atmospheric transport (LRAT) or transboundary transportation through Asian dust storms or monsoon mainly contributed to PM_2.5 pollution in the winter and spring seasons of Taiwan (Liu et al., 2017; Shen et al., 2020; Yu et al., 2021). According to the report “Particulate Matters (PM_2.5) Precursor Demonstration Guidance” by the United States Environmental Protection Agency (U.S. EPA) (Mathias and Wayland, 2019), nitrogen oxides, sulfur dioxide, volatile organic compounds, aldehydes or ketones, and ammonia, which are the precursors of PM_2.5 associated with the traffic-related air pollutants (TRAPs), which was easily developed or converted into PM_2.5. TRAP-PM_2.5-bound chemicals might be linked with several inorganic or organic pollutants, such as carbon black, polycyclic aromatic hydrocarbons (PAHs), heavy metals, carbonaceous compounds, and chlorinated dioxins/furans (Chao et al., 2016). The PM_2.5 is characterized by large surface area and small particle diameter, which enables it to induce a variety of toxic responses (Xing et al., 2017; Chung et al., 2019, 2020). Several environmental factors such as temperature, relative humidity, transboundary pollution, and season variation might be related to the PM_2.5 concentrations in the local region of Taiwan (Chen et al., 2019). Compared with PM₁₀ (PM diameter less than 10 µm), the small size and large surface area of PM_2.5 render potentially harmful chemicals to adhere to these particles (Hassanvand et al., 2015; Pun et al., 2017; Xing et al., 2017; Chen et al., 2019; Hao et al., 2022).

PM_2.5 inhaled directly through the airway will deposit and penetrate the lung alveoli into the blood circulation (Han and Zhuang, 2021; Xie et al., 2021). TRAP PM_2.5 poses potential threats to human health by further activating the inflammatory responses, leading to DNA damage or adducts, and causing epigenetics modifications, which is detrimental to lung function (Li et al., 2014; Chao et al., 2018; Hisamuddin et al., 2020; Suhaimi et al., 2021; Putri Anis Syahira et al., 2020). In our previous study (Chao et al., 2018), compared with policemen working office jobs, policemen employed as traffic conductors exhibited significantly higher levels of serum proinflammatory biomarkers, such as tumor necrosis factor (TNF-α), which was attributed to heavy exposure to TRAP PM_2.5. The value of TNF-α equivalence (TNFα-EQ) in the serum from the exposed group (the traffic conductors) was 0.0302 ± 0.0048 pg TNFα-EQ mL^–1 was significantly higher than that whose level was 0.00440 ± 0.000800 pg TNFα-EQ mL^–1 from the reference control (the office policemen) (Chao et al., 2018). Taxi drivers exhibited systematic inflammation and high concentrations of serum cytokines (e.g., TNF-α) after they were exposed to TRAP (Brucker et al., 2013). PM_2.5-bound hazardous air pollutants (HAPs) (e.g., PAHs induced the activation of systematic inflammatory responses, including TNF-α, C-reactive protein (CRP), interleukin-1b (IL-1b), intercellular adhesion molecule-1 (ICAM-1), and vascular adhesion molecule-1 (VCAM-1), in the blood (Huang et al., 2013; Liu et al., 2017; Chao et al., 2018; Wang et al., 2018; Rahmatinia et al., 2021). PM_2.5 also induced cytokine release and oxidative stress to be involved lung cancer and chronic airway inflammatory diseases like asthma and COPD (Li et al., 2018).

PM_2.5 may trigger neurodevelopmental disruption or delay and may affect neurological behavior in humans. Recent studies also reported a negative association of TRAP PM_2.5 with neurological development or function, such as function deterioration of the central nervous system, delay of neurobehavior, and autism spectrum disorder (ASD) in the early age of life (Becerra et al., 2013; Raz et al., 2015; Sram et al., 2017). Aside from TRAP PM_2.5, ambient PM_2.5 is also a neurotoxicant that triggers harmful effects on the brain, especially during the utero, infancy, and early childhood (Sassá et al., 2011). According to the current epidemiology, prenatal or postnatal children exposed to severe PM_2.5 has a high risk of developing ASD and disrupting the motor and cognitive development of infants (Lertxundi et al., 2015; Talbott et al., 2015; Xi and Wu, 2021). A review article exploring the relationship of ASD and attention deficit hyperactivity disorder (ADHD) with PM_2.5 exposure concluded that although the evidence was insufficient to establish a causal relationship between ADHD and PM_2.5, ADHD development was linked to several factors, such as genome (e.g., transgenerational effects), in-utero exposure, and home environment that were easily influenced by PM_2.5, whereas high levels of PM_2.5 were associated with acceleration of ASD development (Xi and Wu, 2021). In an animal model, decreased responses of brain-derived neurotrophic factor (BDNF) expression and cAMP response element binding protein (CREB) phosphorylation were observed in the immature rats with high-dose exposure to PM_2.5 (Liu et al., 2021). PM_2.5 might also disrupt emotional and cognitive development (Liu et al., 2021). The evidence suggests that children's health is susceptible to PM_2.5 exposure and PM_2.5 is a potentially toxic precursor that causes neurological and neurobehavioral impairment.

 
2 METHODS

 
2.1 The Studying Participants

This cross-sectional study was followed up an established cohort from the Kaohsiung Chang Gung Memorial Hospital (KCGMH) ADHD case-control (KCGMH-ADHD cohort) study. The KCGMH-ADHD cohort was recruited from the Department of Child Psychiatry of KCGMH in southern Taiwan. Specially, patients with ADHD were recruited from the outpatient Department of Child Psychiatry of KCGMH, and children without ADHD were recruited from the communities surrounding KCGMH to serve as controls (Wang et al., 2022a). Of the 234 participants aged 6–12 years from the KCGMH-ADHD cohort, 183 children, comprising 115 ADHD patients (case) and 68 children without ADHD (control), were selected to participate in the present study based on completed questionnaires, inflammatory biomarker data, and submission of a complete and correct residential address. The study protocol was approved by the Institutional Review Boards (IRB) of the Human Ethical Committee of the KCGMH. All ethical standards outlined in the Declarations of Helsinki were followed. After receiving a detailed explanation of the study and its potential consequences, all participants’ parents or guardians signed written informed consent forms before the enrolment. The subjects were selected according to the criteria described in our previous study (Wang et al., 2022a). The inclusion criteria for the patients with ADHD were as follows: (1) verified clinical diagnosis of ADHD according to Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) at the age of 6–12 years; (2) drug-naïve; and (3) no recorded history of comorbid neurodevelopmental or psychiatric disorder, such as ASD, intellectual disability, psychotic disorder, bipolar disorder, epilepsy, or brain injury. The children without ADHD were selected based on the following requirements: (1) age-matched to the study group; (2) without diagnosis of ADHD; and (3) no recorded history of major neurological, neurobehavioral developmental, or psychiatric disorder including ASD, intellectual disability, psychotic disorder, major depressive disorder, substance dependence, or severe head trauma.

 
2.2 Neurological Measurements

The neurological and neurobehavioral status of the participants from the KCGMH-ADHD cohort was determined by an experienced child psychologist by the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) (Baron, 2005; Wang et al., 2022a). The behavioral symptoms were examined using the standard questionnaires of the Swanson, Nolan, and Pelham Version IV Scale (SNAP-IV), specifically the SNAP-IV parent form (answered by the child’s parents) and the SNAP-IV teacher form (answered by the child’s homeroom teacher).

The parent and teacher SNAP-IV forms each comprised a 26-item questionnaire evaluating ADHD symptoms, severity, and needs (Zhang et al., 2005; Bussing et al., 2008). These 26 items included symptoms of inattention (n = 9), hyperactivity/impulsivity (n = 9), and oppositional defiant disorder (n = 8). Each item was scored on a Likert scale ranging from 0 to 3. The present study only selected the core ADHD symptoms (inattention and hyperactivity/impulsivity subscales) for further evaluation.

 
2.3 Biochemical Tests of Inflammatory Markers

Blood samples from the participants were collected between 8:00 and 9:00 AM after an overnight fast. Approximately 2–4 c.c. of serum was collected from each participant. Each blood sample was centrifuged at 3000 rpm for 5 min and then stored at −80°C. Pre-inflammatory biomarkers including TNF-α, monocyte chemotactic protein (MCP)-1, interleukin 1b (IL-1b), IL-2, IL-6, IL-10, IL-13, IL-17, interferon-γ (IFN-γ), and IFN-α2 in the serum were determined using a custom multiplex Laminex immunoassay (Luminex, Austin, TX). Recombinant cytokines pre-diluted in pooled blank plasma from HCs were used as standards for measurement. To detect antigen‐reactive IgG, 50 µL of diluted Microparticle Cocktail was mixed with 50 µL of diluted serum and standard, and then incubated with detection antibodies for 2 h at room temperature (22 ± 3°C) with shaking at 800 rpm. After washing, 50 µL well^–1 of diluted Biotin-Antibody Cocktail was added and incubated for 1.5 h at room temperature with shaking. The plates were washed again and 50 µL well^–1 of phycoerythrin-conjugated streptavidin was added and incubated for 30 min at room temperature with shaking. Finally, the concentration of cytokines in the bead array was measured based on the fluorescence intensity calculated using the Flowmetrix software. All the biomarker measurements exceeded the limits of detection (LODs). The relative standard deviations of intra-assay precision and inter-assay precision were within 10% for all measured markers.

 
2.4 PM_2.5 in the Ambient Air

The long-term data of PM_2.5 in the ambient air from 12 TEPA air monitoring sites (Ciaotou, Nanzi, Mino, Linyuan, Fuxing, Fengshan, Qianzhen, Qianjin, Zuoying, Renwu, Xiaogang, and Daliao) during 2010−2019 and one air monitoring station (Ai-Gou) during 2016-2019 were obtained from the Environmental Protection Bureau Kaohsiung Government (KEPB) (Fig. 1 and Table S1). The historical PM_2.5 levels were recorded from the websites of TEPA and KEPB. Finally, three types of PM_2.5 exposure scenarios were determined.

Fig. 1. PM_2.5 concentrations in 12 air monitoring sites from TEPA or KEPB.

 
2.5 Statistical Analysis

Further statistical analysis was conducted on biomarkers with measurements exceeding the LODs. Descriptive analysis was used to determine the means and standard deviations (SD) of demographics, ADHD symptoms, SNAP-IV forms, IFN-α, IFN-γ, IL-16, IL-2, IL-6, IL-10, IL-13, IL-17, MCP-1, and TNF-α stratified by three PM_2.5 exposure scenarios. Evaluation of normality was performed using the Shapiro-Wilks or Kolmogorov-Smirnov method. An unpaired independent two-tailed student’s t-test assessed the differences in measured parameters among the three PM_2.5 exposure scenario groups. A logistic regression was used to examine levels of inflammatory biomarkers (e.g., TNF-α) among the three PM_2.5 exposure scenarios. The multivariate analysis was used to predict inflammatory biomarker levels by the PM_2.5 exposure scenarios after the age, gender, or ADHD were adjusted (Model 1 was adjusted for age and gender, Model 2 was adjusted for age, gender, and ADHD.). Model 3 was used to determine the association between ADHD and inflammatory responses with adjustment of age, gender, and PM_2.5 exposure scenarios. All statistical analyses were performed using Statistical Product and Service Solutions, version 12.0 (Chicago, Ill., USA). A p-value of less than 0.05 represented statistical significance.

 
3 RESULTS AND DISCUSSION

 
3.1 The Historical Data of PM_2.5 in the Studying Region

Upon examining the historical records and long-term exposure data for PM_2.5 in the study area (Kaohsiung City), we observed a substantial reduction (almost 50%) in PM_2.5 from 2010 to 2019 (Table S2). PM_2.5 emissions improved in the study area over the past decade. The study area is the economic center of southern Taiwan and is location of several key industries including manufacturing, steel-making, oil refining, freight transport, and shipbuilding. According to the big data analysis of PM_2.5 detected by PM_2.5 sensors in Kaohsiung from our previous study (Chen et al., 2019), four main patterns of PM_2.5 were concentrated in the industrial zones or the harbors in the northern and southern Kaohsiung. The TRAP PM_2.5 pattern was independent of the four main types of PM_2.5 patterns (Chen et al., 2019; Yu et al., 2021). In the present study, the Sanmin and Fengshan districts located in the central Kaohsiung were selected as the typical TRAP PM_2.5 areas in the present study. According to the historical and annual records of PM_2.5 by TEPA and KEPB and the PM_2.5 patterns in the studying areas from our previously published paper (Chen et al., 2019), PM_2.5 levels in the study areas could be categorized into three types, specifically low (Ciaotou, Nanzi, Mino, and Linyuan), high (Fuxing, Qianzhen, Qianjin, Zuoying, Renwu, Xiaogang, and Daliao), and TRAP (Sanmin and Fengshan) PM_2.5. Further statistical analyses were stratified by these three types of PM_2.5 exposure scenarios in the present study.

 
3.2 PM_2.5 and Serum Cytokines in School-children

The schoolchildren in the present study were categorized into three groups, namely low-PM_2.5, high-PM_2.5, and TRAP-PM_2.5 exposure groups, on the basis of their residential addresses (Table 1). Descriptive analyses of neurological or neurobehavioral development and inflammatory biomarkers were stratified by the three types of PM_2.5 exposure scenarios (low-PM_2.5, high-PM_2.5, and TRAP-PM_2.5 areas). Most of the SNAP-IV scores from the parent and teacher questionnaires displayed the significant differences (p < 0.05) among the three PM_2.5 exposure groups. ADHD symptoms were highly correlated with SNAP-IV records in the PM_2.5 exposure groups. Because the participants in this study were from the KCGMH-ADHD cohort (Wang et al., 2022a), the ratio of children with ADHD symptoms (62.8%) in this study was higher than that in the general Taiwanese population (3%–7%). The sex ratios (male/female) of the cohort in the present study, the prevalence of Taiwanese ADHD cases, and the Taiwanese general population were 2.59, 3.0, and 1.08, respectively. Xi and Wu (2021) found that air pollutants, including PM_2.5, was significantly associated with the development of ASD, but the occurrence of ADHD appeared to be connected to genetics, daily diet and nutrition, perinatal conditions, and home environment. Dunn et al. (2019) indicated that ADHD etiology was related to increased serum cytokine levels and the genetic polymorphism of relevant inflammatory pathways, which exacerbated ADHD through inflammatory mechanisms after children were exposed to environmental toxicants. In our previous study, we observed that, compared with children in the control group, children with ADHD had significantly lower plasma TNF-α levels (Wang et al., 2022b). Although PM_2.5 exposure could not be definitively linked to ADHD based on the current evidence, several ADHD patients might be comorbid with inflammatory and autoimmune disorders; therefore, the possible contribution of PM_2.5 exposure to ADHD was also adjusted for in the present study. PM_2.5 penetrates the alveolar cells of the lung directly through the airway and is further transported to the circulatory system by alveolar cells (Han and Zhuang, 2021; Xie et al., 2021). In addition to causing neurological disorders, PM_2.5 can lead to premature death caused by inflammatory responses or local or systematic disorders such as lung cancer, cardiovascular diseases, COPD, respiratory disease, and pneumonia (Pun et al., 2017; Mukherjee and Agrawal, 2018; Chowdhury et al., 2019).

In the present study, the IL-6 levels in the low-PM_2.5 participants were margin-significantly higher than those in the TRAP-PM_2.5 participants (p = 0.088). Compared with the high-PM_2.5 participants, the low-PM_2.5 participants had a borderline-significantly lower IL-10 level (p = 0.098). MCP-1 concentrations in the higher exposure groups including the high-PM_2.5 (p = 0.089) and TRAP-PM_2.5 (p = 0.081) participants were higher than those in the low-PM_2.5 exposure group, with borderline significance. Finally, the participants in the TRAP-PM_2.5 exposure group had the significantly higher TNF-α levels than those in the low-PM_2.5 exposure group (p = 0.016), whereas the participants in the high-PM_2.5 group exhibited higher TNF-α concentrations than the participants in the low-PM_2.5 area did, with the margin significance (p = 0.053).

Several articles have addressed the associations of PM_2.5 with the levels of inflammatory cytokines such as TNF-α in schoolchildren. A Mexican study investigated the systematic inflammation in the schoolchildren living in urban (Mexico City: high PM_2.5 area) and rural (Polotitlán: low PM_2.5 area) areas. They reported that children living in Polotitlán exhibited significantly lower serum levels of TNF-α (26.3 pg mL⁻¹, p = 0.005), IL-1b (4.43 pg mL⁻¹, p = 0.02), and IL-10 (3.54 pg mL⁻¹, p < 0.001) than those living in Mexico City (TNF-α, IL-1b, and IL-10: 66.5, 19.2, and 5.97 pg mL⁻¹, respectively) (Calderón-Garcidueñas et al., 2008). The present study focused on the relationship between PM_2.5 exposure and serum cytokine levels in children; our results are consistent with those reported in literatures, particularly regarding the associations between the adults’ exposure to PM_2.5 and their serum cytokine concentrations (e.g., TNF-α). In our previous study (Chao et al., 2016), the police officers who worked as traffic conductors were found to have significantly higher levels of serum TNF-α-EQ than the officers who worked in the office by testing an NF-κB reporter gene assay when they were exposed to PM_2.5 (the police officers carried PM_2.5 personal pumps for estimation of personal PM_2.5 exposure). Brucker et al. (2013) found that the serum TNF-α levels in taxi drivers (n = 39) were significantly higher than those in individuals with nonoccupational PM_2.5 exposure (n = 21) possibly due to the taxi drivers with occupational exposure to TRAP PM_2.5. For diabetes patients, the increased serum levels of IL-6 and TNF-α were positively related to higher levels of PM_2.5 (Schneider et al., 2010). Shakya et al. (2019) investigated air pollutants, including PM_2.5, in Kathmandu Valley, Nepal, during the spring and summer of 2014 and found PM_2.5 emissions of 124 and 45.2 µg m⁻³ in spring and summer, respectively. They observed that the serum levels of inflammatory cytokines, such as TNF-α and IL-1β, 2, 4, 6, 8, 10, 12, and 13, in traffic officers and train conductors were significantly lower in spring (high PM_2.5 period) than in summer (low PM_2.5 period), as revealed by an independent t-test and Wilcoxon test (p < 0.05). These findings of Shakya’s report conflict with those of other studies and the present report (Brucker et al., 2013; Calderón-Garcidueñas et al., 2008; Chao et al., 2016; Schneider et al., 2010; Shakya et al., 2019).

In Fig. 2, the PCA test with varimax rotation was used to examine the associations between systematic inflammation, schoolchildren's age and gender, and three types of PM_2.5 exposure scenarios. Four principal components (PCs) could account for 73.9% of the total variances if the eigenvalue was set to > 1.00. Rotated PC1 (RPC1) accounted for 45.5% of variance in IL-1b, IL-2, IL-10, IFN-α2, IL-6, IFN-γ, TNF-α, IL-13 levels. RPC2, comprising MCP-1 and the type of PM_2.5 exposure, accounted for 10.5% of variance. RPC3 and RPC4 were only associated with the variables of age (9.10% of variance) and sex (8.86% of variance), respectively. PCA tests revealed that the PM_2.5 exposure type and serum MCP-1 were closely correlated. Few studies have reported whether serum MCP-1 in children is affected by PM_2.5 exposure. Several cellular studies using RAW264.7 cells have revealed that PM_2.5 triggers the phosphorylation of NF-κB to activate the gene and protein expression of inflammatory responses (e.g., MCP-1 or TNF-α) (Fu et al., 2020; He et al., 2017). A Chinese research team revealed that BALB/c mice with continuous exposure to PM_2.5 exhibited increased MCP-1 expression from the bronchoalveolar lavage fluid (BALF), indicating that the expression of inflammatory cytokines in the lung was substantially induced by PM_2.5 (Yang et al., 2020).

Fig. 2. The associations between inflammatory biomarkers, age, and PM_2.5 scenarios groups were examined by the test of the principal component analysis.

Table 2 presents the odds ratios (ORs) of inflammatory biochemical markers for high-PM_2.5 and TRAP-PM_2.5 exposure groups in comparison with the reference group of children with low-PM_2.5 exposure group. The TRAP-PM_2.5 exposure group exhibited significantly higher ORs of IL-2 (OR = 3.61, 95% confidence interval (CI): 1.19–11.0, p = 0.023) and lower ORs of IL-13 (OR = 0.991, 95% CI: 0.983–0.999, p = 0.022) than the reference group. The OR of IL-1b (OR = 0.483 with 95%CI between 0.205 and 1.14, p = 0.095) in TRAP-PM_2.5 exposure group had the margin significantly higher than that of the reference control.

A few epidemiological studies have examined the associations between children's exposure to PM_2.5 and cytokine levels in their serum or blood. Calderón-Garcidueñas et al. (2008) showed that childhood exposure to high levels of PM_2.5 had significantly higher levels of IL-1β and IL-10 than those exposed to low levels of PM_2.5. For Nepalese traffic officers, serum levels of IFN-γ, IL-1β, IL-2, IL-6, IL-8, IL-10, IL-12, IL-13, and TNF-α were significantly higher during summer (TRAP-low-PM_2.5 season) than during spring (TRAP-high-PM_2.5 season) (Shakya et al., 2019). In the present study, our findings indicated that the children living in the TRAP-PM_2.5 areas had lower ORs of serum IL-13 than the controls living in the low-PM_2.5 areas, which is consistent with the result of the aforementioned Nepalese study (Shakya et al., 2019).

The inflammatory cytokines in children's serum could be predicted by the independent variables using the general linear models (GLM) in the present study (Tables 3 and 4). In the univariate analysis, serum TNF-α in children could be significantly and weakly explained by the variable of PM_2.5 exposure scenarios without adjusting for confounders factors such as age, gender, and ADHD (adjusted R² = 0.027, p = 0.032). As presented in Table 3, only two inflammatory biomarkers (i.e., MCP-1 and TNF-α) were statistically or marginally significant predictors in Model 1 (the variable of PM_2.5 exposure scenarios and the confounders of age and gender) and Model 2 (the variables of PM_2.5 exposure scenarios and the confounders of age, gender, and ADHD). In Model 1, TNF-α could be partially explained (adjusted R² = 0.033, p = 0.039) by the independent variables of age (p = 0.147), gender (p = 0.291), and PM_2.5 exposure (p = 0.031) in the multivariate analysis. The children’s serum TNF-α levels were significantly linked to their PM_2.5 exposure scenarios in Model 1. In Model 2, the children’s serum TNF-α levels could also be significantly predicted (adjusted R² = 0.055, p = 0.010) by age (p = 0.290), gender (p = 0.130), PM_2.5 exposure scenarios (p = 0.070), and ADHD (p = 0.025). ADHD was significantly correlated with serum TNF-α levels, and the correlation between TNF-α levels and PM_2.5 exposure scenarios was marginally or borderline significant in Model 2 in multivariate analysis. Gender was slightly and significantly correlated with ADHD occurrence (R² = 0.052, p = 0.002), and boys had a significantly higher ADHD prevalence (OR: 2.80, 95% CI: 1.44–5.43, p = 0.002) than girls. On the basis of the results of the multivariate analysis, ADHD was a major variable that significantly affected the GLM of Model 2. In Table 4, the results of Model 3 presented that ADHD is a significant predictor of serum TNF-α in children (p = 0.035), with adjustment for the confounders of age (p = 0.253), gender (p = 0.138), and PM_2.5 exposure scenarios (p = 0.044; adjusted R² = 0.054, p = 0.008). In Model 3, ADHD and PM_2.5 exposure scenarios were significantly associated with serum TNF-α. The environmental epidemiological studies have yet to provide adequate evidence on the association of ADHD occurrence in children with the severity of exposure to PM_2.5 (Suades-González et al., 2015; Xi and Wu, 2021); rather, PM_2.5 exposure scenarios and ADHD occurrence were found to be independent. In the present study, ADHD prevalence and SNAP-IV scores were inversely and significantly correlated with heavily PM_2.5 exposure scenarios in Table 1, and ADHD prevalence was significantly correlated with gender among the children. On the basis of our results, we determined that serum TNF-α in our cohort might be slightly or weakly correlated with living in the TRAP-PM_2.5 areas, and the influence of ADHD on inflammatory responses must also be adjusted for.

The findings of the present study were limited because this is an environmental epidemiological study and the in-vitro and in-vivo tests are not included. Firstly, the statistics were only used to examine the cause-effect with the inflammation in children’s serum after they were longitudinally exposed to PM_2.5. This cause-effect relationship was still needed to be tested by the cellular and animal experiments. Secondly, the three PM_2.5 exposure scenarios were determined by the long-term PM_2.5 data from TEPA air monitor sites. The subjects were assigned into the different PM_2.5 exposure scenarios based on the children’s household address. Our subjects might have complicated PM_2.5 exposure scenarios. Finally, compared with the general population in Taiwan, the KCGMH-ADHD cohort might have some sampling bias to influence the study’s outcomes.

According to most in-vitro, in-vivo, and epidemiological studies (Brucker et al., 2013; Calderón-Garcidueñas et al., 2008; Chao et al., 2016; Chen et al., 2018; Fu et al., 2020; He et al., 2017; Schneider et al., 2010; Yang et al., 2020), the elevated levels of inflammatory cytokines are linked to the severe PM_2.5 exposure. On the basis of our results, certain cytokine biomarkers in the serum, such as IL-2, IL-13, and TNF-α, in the schoolchildren were associated with exposure to PM_2.5, particularly for those children in the TRAP areas. The in-vitro and in-vivo studies usually used the high doses of PM_2.5 to test their models. Compared with the doses used in the in-vitro and in-vivo models, our epidemiological report used the relatively low doses to examine the cause-effect relationship in the present study. Several environmental epidemiological studies have demonstrated that PM_2.5 induces asthma morbidity in both children and adults, especially among those living in TRAP PM_2.5 areas (Loftus et al., 2015; Klümper et al., 2015; Orellano et al., 2017; Rosenquist et al., 2020; Jalaludin et al., 2014). Several review articles have reported that PM_2.5 might be involved in triggering and aggravating asthma and COPD through PM_2.5-induced cytokine release and oxidative stress production (Orellano et al., 2017; Li et al., 2018; Liu et al., 2022). Because few epidemiological studies have focused on the relationship between serum cytokines and PM_2.5 exposure in children, the hypothesis of PM_2.5-induced activation of systematic inflammation that causes asthma symptoms in schoolchildren is difficult to examine or evaluate. Further studies are warranted to examine the relationships between inflammatory responses and PM_2.5 exposure in children.

 
4 CONCLUSIONS

Currently, the evidence from these epidemiological studies was inadequate to verify the correlations between serum inflammatory biomarkers (e.g., inflammatory cytokines) in schoolchildren and outdoor PM_2.5. In this study, we examined the serum levels of cytokines in schoolchildren exposed to PM_2.5. Several inflammatory biomarkers except for TNF-α in schoolchildren’s serum were not linked with children exposure to PM_2.5. Furthermore, our results demonstrated that the serum levels of TNF-α in children were significantly correlated with living in high-PM_2.5 exposure areas, particularly TRAP exposure areas. The in-vitro and in-vivo studies were shown that pre-inflammatory and inflammatory responses were activated by the cells or animals after they were exposed to severe PM_2.5, but few epidemiological studies examined the cause-effect relationship between serum inflammation and children exposure to heavy PM_2.5 pollution. This study might provide the observation to show that TRAP PM_2.5 induced the activation of serum TNF-α after children were longitudinally exposed to TRAP.

 
ACKNOWLEDGMENTS

Our study was supported by grants from the Ministry of Science and Technology (MOST 109-2221-E-020-009-MY3 and MOST 111-2314-B-182-018-MY3), Chang Gung Memorial Hospital (CMRPG8K1281 and CMRPG8K1282), and Pingtung Christian Hospital (PS110005). This manuscript was edited by Wallace Academic Editing. The authors wish to thank the joined research teams in the Environmental Health and Risk Assessment Laboratory, National Pingtung University of Science and Technology and Department of Child and Adolescent Psychiatry, and Department of Pediatrics in Kaohsiung Chang Gung Memorial Hospital for their assistance with this study.

 
DISCLAIMER

The authors declare no competing interest.