Associations of PM 2.5 with Chronic Obstructive Pulmonary Disease in Shipyard Workers: A Cohort Study

The association between lung deposition of particulate matter with an aerodynamic diameter less than 2.5 µ m (PM 2.5 ) from welding fumes and lung function is unclear. We conducted a cohort study with a follow-up of 4 years in 115 shipyard workers to investigate the incidence rate of chronic obstructive pulmonary disease (COPD), and to evaluate the effects of welding fume PM 2.5 deposition on fractional exhaled nitric oxide (FeNO), lung function, and blood pressure using generalized estimating equation models. Personal exposure to welding fumes was measured, and deposition fractions of inhaled welding fume PM 2.5 in lung regions were estimated using multiple-path particle dosimetry. We observed the overall incidence rate of COPD to be 2.51 cases per 100 person-years. The incidence rate was higher in welding workers than in office workers and higher in non-smokers compared to smokers. In the overall cohort subjects, we observed that an increase in the interquartile range of PM 2.5 was associated with a 1.618-ppb decrease in FeNO, a 0.115-L decrease in FVC, a 0.091-L decrease in FEV 1 , a 0.520% increase in the FEV 1 /FVC ratio, a 0.259-L s –1 decrease in PEF, a 0.096-L s –1 decrease in FEF 25%-75% , a 0.215-L s –1 decrease in FEF 25% , and a 0.114-L s –1 decrease in FEF 50% (all p < 0.05). We observed that a


INTRODUCTION
Chronic obstructive pulmonary disease (COPD) is a progressive respiratory condition characterized by poorly reversible airflow limitations associated with increased inflammatory responses to noxious particles or gases in the respiratory tract (Vestbo et al., 2013).In 2019, the World Health Organization estimated that COPD ranked third in global causes of mortality (WHO, 2020).Smoking is considered a major risk factor for COPD.Nevertheless, according to the American Thoracic Society (ATS), occupational-related COPD cases contribute to approximately 15% of total COPD (Melville et al., 2010).Occupational exposures to gases, vapors, dust, and welding fumes, notably in non-smokers, could result in a 1.32-fold increased risk of COPD mortality (95% confidence interval (CI): 1.18-1.47)(Torén and Järvholm, 2014).
Chronic exposure to metal fumes, which contain numerous metals such as zinc, manganese, copper, cadmium, nickel, and aluminum, may cause serious health problems, especially in welding workers (Lai et al., 2021).Welders are regarded as a population at risk for COPD due to high levels of metal fume particulate matter less than 2.5 µm in aerodynamic diameter (PM2.5) in their breathing zone.Notably, previous studies showed that exposure to welding fumes was associated with declines in lung function in welders (Christensen et al., 2008).It was reported that the prevalence of COPD in welding workers was approximately 15.2% (Koh et al., 2015;Minov et al., 2015).Moreover, the prevalence of COPD in welding workers was higher than in office workers (p < 0.05) (Minov et al., 2015).Despite the high burden of COPD in welding workers, the welding-related incidence rate of COPD is still unavailable.
Multiple-path particle dosimetry (MPPD) is a computational algorithm to estimate PM deposition in lung regions (Asgharian et al., 2001).Specifically, an MPPD study demonstrated that coarse particles are mostly deposited in the head and nasal region, whereas ultrafine particles are deposited in deeper parts of the airway (i.e., the tracheobronchial and alveolar regions) (Manojkumar et al., 2018).Another MPPD study found that PM deposition was higher during the haze period compared to the non-haze period (Behera et al., 2015).However, associations between PM deposition of welding fume PM2.5 in lung regions and respiratory health are poorly understood.
Shipyards are major sources of exposure to welding fumes as determined in previous studies (Jeong et al., 2016;Chuang et al., 2018;Su et al., 2019).Our previous findings demonstrated that exposure to metal fume PM2.5 increased systemic oxidative stress and inflammation in shipyard workers (Lai et al., 2016).In this study, a cohort study was conducted with a follow-up of 4 years in a shipyard.The objective of this study was to examine the incidence rate of COPD, and to evaluate the effects of welding fume PM2.5 on lung function and blood pressure in these workers.

Study Subjects
To examine the effects of welding fumes on workers in a shipyard located in northern Taiwan, a dynamic cohort study was conducted from 2014 to 2017.There were 115 subjects in total enrolled in the study.Exclusion criteria for subjects were being aged younger than 20 or older than 70 years, having comorbidities (diabetes mellitus, heart failure, or coronary artery disease), and having experienced exacerbations within 1 month prior to the study.Data on the smoking status (nonsmoker, ex-smoker, or current smoker) and the use of personal protective masks were collected.Tungsten inert gas (TIG) welding was the main welding method used at the shipyard (Lai et al., 2016).Welding rods used in the present study mainly contained manganese, silicon, nickel, chromium, and molybdenum (Lin and Pan, 2012).The Ethics Committee of the Joint Institutional Review Board of Tri-Service General Hospital (IRB no.1-102-05-013) approved the study protocol.All subjects obtained oral and written information and provided informed consent prior to inclusion.The methods were performed according to approved guidelines.

Personal PM2.5 Exposure
The individual exposure assessment for PM2.5 was conducted for each participant as previously demonstrated (Chuang et al., 2018).Briefly, individual exposure to welding fume PM2.5 was obtained from a whole working day.Consecutive 8-h personal welding fume PM2.5 sampling was performed (09:00-17:00, Monday) by a model 200 Personal Environmental Monitor (PEM) equipped with an AirChek ® XR5000 air sampling pump (SKC, Eighty-Four, PA, USA) at a flow rate of 2 L min -1 .Next, PM2.5 was collected onto Teflon filter substrates (37 mm, pore size: 2-µm, Pall, Ann Arbor, MI, USA) within the breathing zone.We measured the personal exposure to welding fume PM2.5 once per year (the same month) for 4 years.

MPPD Model
Deposition fractions of inhaled welding fume PM2.5 in the human lungs (i.e., total lung, head and nasal, tracheobronchial, and alveolar regions) were estimated using the MPPD computational model (MPPD vers.3.04 for Windows, Applied Research Associates, Albuquerque, NM, USA) as previously reported (Asgharian et al., 2001).Briefly, deposition fractions were estimated for 3300 ml of functional residual capacity.The upper respiratory tract volume was predicted to be 50 mL.We assumed a tidal volume of 625 mL, a respiratory rate of 12 breaths min -1 , and an inspiratory to expiratory ratio of 1:2 with no pause between inhalation and exhalation.PM2.5 deposition, which was assumed to enter the lungs via the nasal airway, was estimated to have various diameter sizes (0.01-1 µm).

Fractional Exhaled Nitric Oxide (FeNO) Measurement
A handheld FeNO device (NObreath, Bedfont Scientific, Kent, UK) was employed to quantify FeNO in accordance with ATS guidelines (ATS, 2003).The method of FeNO detection was previously described (Fan et al., 2018).FeNO was measured at the beginning of the day following the previous day of exposure (Tuesday morning; post-exposure).FeNO measurements in each subject were conducted in triplicate.Values were averaged as the response.

Statistical Analysis
Tests of normality were used to identify if the data had a normal distribution.To minimize the influence of outliers, extreme values outside the 1 and 99 percentiles were replaced by a winsorization method (Tsai et al., 2012).Incidence rates of COPD were determined by dividing the number of incident cases by the total person-years of subjects at risk and are expressed per 100 personyears.Generalized estimating equation (GEE) models were employed to investigate associations of PM2.5 with FeNO, lung function, and blood pressure in the overall subject group, healthy subject group (FEV1/FVC ratio ≥ 70%), and COPD group (FEV1/FVC ratio < 70%).These associations were expressed as regression coefficients (β) multiplied by the interquartile range (IQR) of welding fume PM2.5 levels.These models were adjusted for the test year, age, work type (office or welding), working duration, and smoking status.To examine associations (β coefficients) of PM2.5 deposition in lung regions with FeNO, lung function, and blood pressure in the overall subject group, GEE models were employed.These models were adjusted for test year, age, work type (office or welding), working duration, smoking status, and COPD.We also adjusted for protective mask use in these models (Supplementary Material Fig. S1 and Fig. S2).Two-sided p values of < 0.05 were considered statistically significant.Data analyses were conducted by SPSS vers.22 for Windows statistical software (SPSS, Chicago, IL, USA).
Since most of the metal fume particles were less than 2.5 µm, we opted to use welding fume PM2.5 as the exposure assessment.We found that the average individual exposure to welding fume PM2.5 was 999.38 µg m -3 , which was lower than the U.S. Occupational Safety and Health Administration permissible exposure limits of 5 mg m -3 , averaged over an 8-hour period, for respirable fraction particles not specifically regulated (NIOSH, 2011;CDC, 2018;NIOSH, 2018;Lai et al., 2021).Nevertheless, PM2.5 concentrations in this study were higher than in our previous studies which reported that welding fume PM2.5 levels ranged 716-755 µg m -3 (Lai et al., 2016;Lai et al., 2021).We observed associations between welding fumes and adverse health effects by cross-sectional studies (Koh et al., 2015;Li et al., 2015); however, the long-term impacts on respiratory health among the workers remain unclear.

Incidence Rate of COPD
Six subjects developed COPD during the 4-year follow-up period.As observed in Table 3, the overall incidence rate of COPD was 2.51 cases per 100 person-years.The incidence rate of COPD in current smokers and ex-smokers (1.41 cases per 100 person-years) was lower than in nonsmoking workers (4.12 cases per 100 person-years).Furthermore, the incidence rate of COPD in welding workers (2.96 cases per 100 person-years) was higher than that in office workers 1309.03Definitions of abbreviations: FeNO, fractional exhaled nitric oxide; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; FEF25%-75%, forced expiratory flow (FEF) at 25%-75% of FVC; FEF25%, FEF at 25% of FVC; FEF50%, FEF at 50% of FVC; FEF75%, FEF at 75% of FVC; PM2.5, particulate matter less than 2.5 µm in aerodynamic diameter; BP, blood pressure.(1.43 cases per 100 person-years).The incidence rate of COPD in workers who used protective masks was 2.51 cases per 100 person-years, whereas the incidence rate of COPD in workers who did not use protective masks was 2.50 cases per 100 person-years.Previous studies reported that the incidence rate of COPD in the general population ranged 2.6-9.2cases per 1000 person-years (García Rodríguez et al., 2009;van Durme et al., 2009).Specifically, a prospective cohort involving subjects aged ≥ 55 years based on a Rotterdam study reported that the overall incidence rate of COPD was 9.2 cases per 1000 person-years (95% CI: 8.5-10.0)(van Durme et al., 2009).Another cohort including subjects aged 40-89 years reported  et al., 2007).The results suggested that the COP D incidence rate in shipyard workers was higher than in the general population.Previous studies also demonstrated the association of welding fume exposure with lung function decline and increased risk of COPD (Christensen et al., 2008;Koh et al., 2015;Grahn et al., 2021).Next, we observed the COPD incidence rate was higher in welding workers than in office workers.In a study including non-smoking workers aged 37-59 years, the prevalence of COPD in welding workers was higher than in office workers (p < 0.05) (Minov et al., 2015).Furthermore, a case-control study found that welders had a 6.4-increased odds (95% CI: 1.6-25.5) of developing COPD compared to office workers (Mastrangelo et al., 2003).The present study demonstrated that the incidence rate of COPD was higher in non-smokers than in smokers.This finding suggests that exposure to welding fume PM2.5 could be associated with higher risk of COPD in non-smoking welders than in smoking welders.Previous studies also reported the associations of welding fumes with declining lung function and COPD in non-smoking welders (Koh et al., 2015;Minov et al., 2015).We found that incidence rates of COPD in workers with protective masks and those without protective masks were similar.It was reported that protective mask use was positively associated with FVC (β = 0.263; p < 0.05) (Hamzah et al., 2016).However, previous findings also suggested that FEV1, FVC, and PEF were significantly reduced in healthy subjects wearing N95 masks and surgical masks compared to those wearing no mask (Fikenzer et al., 2020).These observations imply that the efficiency of personal protective masks for protecting against COPD is insignificant.Based on the findings of this study, shipyard workers are a population-at-risk for the development of COPD, especially among welders and non-smokers.

Associations of PM2.5 with FeNO and Lung Function
Associations of PM2.5 with FeNO and lung function in the overall subject group, healthy group, and COPD group are demonstrated in Fig. 1.In the overall subject group, we observed that one IQR (1309.03-µgm -3 ) increase in welding fume PM2.5 was associated with a 1.618-ppb decrease in FeNO, a 0.115-L decrease in FVC, a 0.091-L decrease in FEV1, a 0.520% increase in the FEV1/FVC ratio, a 0.259-L s -1 decrease in PEF, a 0.096-L s -1 decrease in FEF25%-75%, a 0.215-L s -1 decrease in FEF25%, and a 0.114-L s -1 decrease in FEF50%.In the healthy subject group, we found that one IQR (1450.80-µgm -3 ) increase in welding fume PM2.5 was associated with a 1.925-ppb decrease in FeNO, a 0.121-L decrease in FVC, a 0.102-L decrease in FEV1, a 0.404% increase in the FEV1/FVC ratio, a 0.283-L s -1 decrease in PEF, a 0.110-L s -1 decrease in FEF25%-75%, a 0.232-L s -1 decrease in FEF25%, and a 0.124-L s -1 decrease in FEF50%.In the COPD subject group, we observed no significant associations of welding fume PM2.5 with FeNO or lung function.
FeNO is a sensitive noninvasive marker of eosinophilic airway inflammation and asthma (Verini et al., 2010;Riise et al., 2011;Schneider et al., 2013).It was reported that each IQR increase in personal exposure to PM2.5 resulted in a 12.8% increase (95% CI: 5.5%-20.7%) in FeNO (Fan et al., 2018).Fig. 1.Associations of particulate matter with an aerodynamic diameter less than 2.5 µm (PM2.5) with fractional exhaled nitric oxide (FeNO), lung function, and blood pressure in overall cohort subjects, healthy subjects, and chronic obstructive pulmonary disease (COPD) subjects.Data are presented as regression coefficients multiplied by an interquartile range (IQR) increase in PM2.5 with the 95% confidence interval (CI).Generalized estimating equations were adjusted for test year, age, work type (office or welding), working duration, and smoking status.Values in bold characters with red color were deemed statistically significant (* p < 0.05).
However, the effects of welding fumes on FeNO are still unclear.Previous studies reported that the association between welding fume PM2.5 exposure and FeNO was insignificant (Lehnert et al., 2011;Järvelä et al., 2013;Lai et al., 2021).Moreover, various associated factors could lead to a decrease in FeNO levels such as smoking, physical exercise, and sputum induction (Gabriele et al., 2005;Abba, 2009).Therefore, this association observed in shipyard workers needs further investigation in future studies.
Next, we found that an increase in the IQR of welding fume PM2.5 was associated with declines in FVC, FEV1, and PEF in the overall subject group and healthy subject group.This was consistent with previous studies showing declines in FVC, FEV1, and PEF in welders compared to a control group (Meo et al., 2003).It was also reported that FVC, FEV1, and FEF25%-75% decreased over the years of welding (Kilburn and Warshaw, 1989;Humerfelt et al., 1993).Together, the reduced FVC, FEV1, and PEF in our study could suggest impairment of pulmonary function in welders.Next, we found that an increase in the IQR of PM2.5 was associated with declines in FEF25%-75%, FEF25%, and FEF50% in the overall subject group and healthy subject group.Our results were also in line with previous findings which demonstrated that welders had lower levels of FEF25%-75% than office workers (p < 0.05) (Chuang et al., 2018).A reduction in FEF25%-75% with normal FEV1, FVC, and FEV1/FVC could be a sensitive marker of small and distal airway obstruction (Marseglia et al., 2007;Raji et al., 2018).Impaired FEF25%-75% was thus employed to detect early stages of obstructive airway disease (asthma and COPD) (Riley et al., 2015;Kwon et al., 2020).It was also reported that exposure to high levels of welding fumes resulted in 3.8-fold increased odds (95% CI: 1.03-16.02) of COPD compared to exposure to low levels of welding fumes (Koh et al., 2015).Furthermore, occupational exposure to dust and fume was associated with an increased mortality of COPD (Torén and Järvholm, 2014).It was reported that exposure to welding fumes could lead to increases in oxidative stress biomarkers (i.e., H2O2 and 8-hydroxy-8-deoxyguanosine (OHdG)) (Graczyk et al., 2015;Lai et al., 2016).These reactive oxygen species (ROS) were associated with reduced lung function (i.e., FEV1 and FVC) (Schünemann et al., 1997;Ochs-Balcom et al., 2005) and COPD (Montuschi et al., 2000;Kinnula et al., 2007).However, a positive association was observed between welding fume PM2.5 and the FEV1/FVC ratio in subjects.This could have been due to a decrease in the magnitude of FVC compared to FEV1 (Barreiro and Perillo, 2004;Johnson and Theurer, 2014).Previous studies also demonstrated an annual increase of the FEV1/FVC ratio after exposure to welding fumes (Mur et al., 1989;Szram et al., 2013).Meanwhile, another study reported no significant association between exposure to welding fumes and the FEV1/FVC ratio (Christensen et al., 2008).This association thus needs to be clarified in future works.Together, our findings imply that exposure to chronic welding fume PM2.5 could lead to reduced lung function, small airway obstruction, and COPD.

Associations of PM2.5 Deposition with FeNO and Lung Function
Particle deposition in lung regions is vital for understanding the mechanism of declining lung function.As depicted in Fig. 2, we observed that a 1-µg m -3 increase in welding fume PM2.5 deposition in lung regions (i.e., the total lung region, the head and nasal region, the tracheobronchial region, and the alveolar region) was associated with decreases in FeNO, FVC, FEV1, PEF, FEF25%-75%, FEF25%, and FEF50%, and was associated with an increase in the FEV1/FVC ratio (β coefficients).Importantly, we found that the absolute values of β coefficients decreased in the following order: alveolar > head and nasal > tracheobronchial > total lung region.Therefore, our results suggested that metal fume PM2.5 deposition in the alveolar region was associated with the higher risk of COPD compared to other lung regions.This was consistent with our previous study which reported significant associations of PM2.5 deposition in the alveolar region of the lung lobes with emphysema severity and COPD (Tung et al., 2021).Furthermore, a previous study demonstrated that cigarette smoke particles are mostly deposited in the alveolar region (Sahu et al., 2013).It was reported that deposition fractions of cigarette smoke in the head and nasal, tracheobronchial, and alveolar regions were 0.163, 0.152, and 0.298, respectively (Sahu et al., 2013).The alveolar region plays an important role in the toxicological effects of inhaled welding fumes due to its permeability (Fröhlich and Salar-Behzadi, 2014).Welding fumes can affect the alveolar epithelium through endocytosis, oxidative stress, and proinflammatory mechanisms (Suri and Grigg, 2016;Yan et al., 2016;Wang et al., 2020).Furthermore, mitochondrial ROS can lead to alveolar smooth muscle proliferation, contraction, and remodeling (Sutcliffe et al., 2012;Wiegman et al., 2015;Pan et al., 2019).Together, our findings showed that PM2.5 deposition in lung regions, especially in the alveolar region, was associated with a lung function decline and an increased risk of COPD.
The main strength of our study is the occupational cohort of shipyard workers, with a followup time of 4 years, in which we examined the decline in lung function over the study duration.A limitation of our study is its small sample size.Moreover, the metal components of the welding fumes were not identified in our study.We did not account for other gaseous air pollutants (i.e., nitrogen dioxide, carbon monoxide, and volatile organic compounds) in this study.Other risk factors for COPD, such as nutrition and genetics, should be studied in future work.Age-related changes in lung function with PM2.5 deposition in lungs were not measured in this study and could have impacted the statistical data.PM2.5 personal exposure may be partially originated from other emission sources.The office workers could be exposed to PM2.5 welding fumes due to the office was located nearby the welding site.The generalizability may be limited because the majority of workers were men.

CONCLUSIONS
This is the first study reporting the incidence rate of COPD in shipyard workers.We observed that the overall incidence rate of COPD in the shipyard cohort was approximately 2.51 cases per 100 person-years.Higher incidence rates of COPD were observed in non-smokers and welders, which was associated with decreases in FeNO and lung function decline after exposure to welding fume PM2.5.Pulmonary effects by PM2.5 exposure in occupational settings is an urgent issue for worker health protection.

Fig. 2 .
Fig. 2. Associations (β coefficient) of particulate matter with an aerodynamic diameter less than 2.5 µm (PM 2.5 ) deposition in total lung, head and nasal, tracheobronchial, and alveolar regions with forced exhaled nitric oxide (FeNO), lung function, and blood pressure in overall cohort subjects.Generalized estimating equations were adjusted for test year, age, work type (office or welding), working duration, smoking status, and chronic obstructive pulmonary disease (COPD).Values in bold characters with red color were deemed statistically significant (* p < 0.05).

Table 1 .
Characteristics of cohort subjects in the shipyard.

Table 2 .
Particulate matter with an aerodynamic diameter less than 2.5 µm (PM2.5)deposition in the total lung, head and nasal, tracheobronchial, and alveolar regions.

Table 3 .
Incidence rates for overall, smoking-specific, work type-specific, and protective maskspecific chronic obstructive pulmonary disease (COPD) per 100 person-years in the study cohort.