Benchmark dose estimation for coke oven emissions based on oxidative damage in Chinese exposed workers

Abstract

Coke oven emissions (COEs) can cause oxidative stress of the body, which in turn induces the occupational lung disease and also increases the risk of other diseases. COEs are the major occupational hazard factors for coke oven workers. The aim of the study is to explore the influences of COEs exposure on oxidative damage and estimate the benchmark dose (BMD) of COEs. A group of 542 workers exposed to COEs and 237 healthy controls from the same city were recruited in this study. The corresponding measuring kits were used to determine the plasma biomarkers of oxidative damage level. Generalized linear models and trend tests were used to analyze the relationship between COEs exposure and biomarkers. EPA Benchmark Dose Software was performed to calculate BMD and the lower confidence limit of the benchmark dose (BMDL) of COEs exposure. A significant association was observed between COEs exposure and oxidative damage with T-AOC as a biomarker. The BMD of COEs exposure were 2.83 mg/m³ and 1.39 mg/m³ for males and females, respectively, and the corresponding BMDL were 1.47 mg/m³ and 0.75 mg/m³, respectively. Our results suggested that the exposure level of COEs below the current national occupational exposure limits (OELs) would induce oxidative damage, and the OEL of COEs based on the T-AOC damage was suggested at 0.03 mg/m³ in this study.

1. Introduction

Coke oven emissions (COEs) are a kind of environmental pollutants produced in the coking process, primarily comprised of particulates and volatile organic compounds, in particular polycyclic aromatic hydrocarbons (PAHs)(Hu et al., 2006). Polycyclic aromatic hydrocarbons (PAHs) are widely existing in general or occupational environments, and produced from volcanic eruptions and forest fires, as well as anthropogenic activity human factors, such as automobile exhaust cooking, and industrial production (Han et al., 2019). Growing epidemiological evidence showed that exposure to COEs/PAHs could increase the risk of respiratory diseases (Alhamdow et al., 2017), cardiovascular diseases (Bortey-Sam et al., 2017), and even lung cancer (Hong et al., 2016), among others. Workers in a coke plant working environment are constantly exposed to COEs containing PAHs and therefore more likely to suffer from PAHs exposure.

The need for stringent occupational exposure limits (OELs) of COEs to protect workers has been recognized and highlighted. The National Institute for Occupational Safety and Health (NIOSH) had recommended a time weighted average (TWA) of 0.20 mg/m³ as the exposure limit (benzene-soluble fraction)(NIOSH, 1997). The Occupational Safety and Health Administration had commanded a permissible exposure limit of 0.15 mg/m³ (benzene-soluble fraction)(OSHA, 1998). Of note, the current OELs were set mainly based on the no observed adverse effect level (NOAEL) protocol in animal experiments (Ringblom et al., 2014), which may not be generalizable to humans. Studies reported that coke oven workers had a 30% higher risk of lung cancer if they exposed to levels above the OELs (Rota et al., 2014). The current OELs were based on the tumorigenesis or other observed damage in workers or animals, and the studied endpoints may be inappropriate for assessing other possible health effects to damage for workers, such as oxidative damage.

Oxidative stress is the early biological events and common pathogenic mechanism of PAHs-related diseases (Xie et al., 2019). Metabolites of PAHs could generate reactive oxygen species (ROS) through redox cycling. A large amount of active oxygen would break the balance of oxidation and antioxidant system and induce oxidative stress, which would further induce oxidative damage and lipid peroxidation (Mena et al., 2009), and consequently lead to the occurrence and development of inflammatory diseases, such as cardiopulmonary diseases (Kim et al., 2013). PAHs may in part lead to the adverse effects by inducing oxidative damage. Some effect biomarkers were used to reflect oxidative damage levels induced by COEs/PAHs exposure in vivo, such as malonaldehyde (MDA)(Yang et al., 2015), catalase (CAT)(Singh et al., 2008). Accordingly, in this study, some biomarker were selected to further explore the association of COEs exposure and oxidative damage and assess benchmark dose of COEs exposure.

The benchmark dose (BMD), proposed by Crump, was considered as more reasonable than NOAEL protocol. The BMD was defined as a dose that corresponded to a specified increase in response rates (1–10%)(Crump, 1984). The BMD methods derived dose-response curves by fitting mathematical models to the all collected data (Lin and Tai-Yi, 2007), and considered the uncertainty of the results by calculating the statistical lower confidence limit of the BMD (BMDL)(USEPA, 2012). The BMDL was suggested to replace the NOAEL estimate, and has been widely used to analyze dose-response of toxicity data and to evaluate exposure limits of environmental pollutants and chemical substances (Niu et al., 2018; Wang et al., 2013).

Oxidative damage was affected by COEs exposure. Some biomarkers were used to reflect the oxidative stress levels in vivo. However, the association between COEs exposure and some biomarkers related to oxidative damage was not further assessed, and used to estimate the BMD of COEs exposure. Therefore, this study aims to further explore the association between COEs external exposure dose and biomarkers of oxidative stress, and to estimate the BMD and BMDL of COEs exposure based on oxidative damage in COEs exposed workers.

2. Objects, materials, and methods

2.1. Study subjects

A total of 779 subjects was recruited in this study. Exposure-group subjects were comprised of coke oven workers from a coking plant in Henan province. The inclusion criteria included aged 18–60 years old, occupationally exposed to COEs for more than one year and written informed consent. The excluded criteria were subjects with major organ function failure, other major diseases, tumor, or pregnant. As a result, the exposure-group were comprised of 542 coke oven workers, and the work locations of workers mainly covered coke oven top, coke oven side, and coke oven bottom. The similar inclusion and exclusion criteria were set for the control group except the occupational exposure history. Control-group subjects included 237 staffs of a public institution in the same period and region without a history of exposure to occupational COEs and other toxicants.

2.2. Collections of basic information and biological samples

All study subjects were informed of the study purpose and signed informed consent. Detailed occupational health questionnaires were conducted face-to-face by uniformly trained investigators. In the questionnaires, according to the recommendations of the World Health Organization (WHO), smoking was defined in this survey as smoking of more than one cigarette per day and lasting for more than half a year; drinking alcohol was defined as drinking alcohol for more than twice a week within 6 months from the survey day. We collected demographic information, occupational exposure, and a total of 3 ml peripheral blood samples of heparin anticoagulation from each study subject. Plasma oxidative stress levels were detected in collected heparin anticoagulant blood samples within 72 h. The study was approved by Zhengzhou University Life Science Ethics Review Committee.

2.3. Calculation of cumulative exposure dose

For the exposure group, coke oven workers with different work types were exposed to different levels of COEs (Keimig et al., 1986). We monitored the COEs exposure for different working types in different working sites in the present study. Under the working environment, the total suspended particulate (TSP) medium-flow sampler (Laoshan, Shandong, China) was placed on the furnace top, the furnace side and the furnace bottom for air sampling. The sampling point arrangement and sampling requirements were in strict accordance with the “Sampling Specification for Monitoring of Hazardous Substances in the Air of the Workplace” (GBZ159–2004). COEs samples were collected by middle volume sampler in different workplaces with the flow rate at 100 L/min for 4 h. The COEs concentrations were determined according to the standard of “Stationary source emission-Determination of benzene soluble particulate matter-Soxhlet extraction gravimetric method”(HJ690–2014). The main steps included weighing after sampling, extracting samples, and then calculating the concentration. Based on onsite detecting results and occupational exposure history, we could get the time weighted average concentration (TWA, mg/m³) and work time (year) of each subject. The cumulative exposure doses (CED) of COEs in each subject was assessed using the following equation: CED (mg/m³·year) = ΣC (mg/m³) × T (year), where C denoted the mean value of COEs exposure concentrations of each day in a workplace, T was working years. For the controls, we estimated the CED-COEs in living working environment using the aforementioned equation, and CED-COEs was mainly based on the concentration of COEs in their living environment and their age. The limit of detection methods was 0.004 mg/m³ and the coefficients of Variation (CV) ranged from 0.46% to 13%. Detailed implementation steps of sample collection and COEs concentration determination also were described in our previous study (Duan et al., 2019; Wang et al., 2019b).

2.4. Determination of oxidative damage biomarkers

The four common biomarkers were used to evaluate the oxidative stress from free radical and antioxidant enzyme levels were determined in this study, including hydrogen peroxide (H₂O₂), malondialdehyde (MDA), catalase (CAT), and plasma total antioxidant capacity (T-AOC) (Wu et al., 2013). The process of collection, pretreatment, determination of biological sample, and the evaluation and analysis of results were carried out in strict accordance with the guidelines of Quality Assurance Standard for Biological Monitoring in Occupational Health (MOH, 2006). The determination kits of plasma oxidative damage biomarkers were purchased from the biological company (Nanjing Jiancheng Institute of Bioengineering, China). In process of determination Oxidative damage biomarkers, reagent storage and preparation, as well as pre-experimental operations are performed in accordance with the instructions of each kit. According to the results of the pre-experiment and the technical parameters of the kit to adjust and find the best experimental conditions, and then the determination experiment was carry out.

The detection instruments included 10 ml centrifuge tube (Nanjing Vazyme Biotechnology Co., Ltd), HH-42 constant temperature water bath box (Changzhou Guohua instrument Co., Ltd), V-5800 pc scanning visible spectrophotometer (Shanghai METASH instrument Co., Ltd), and Vortex mixer.

2.4.1. Total antioxidant capacity (T-AOC)

The detection reagents, instruments, and the concrete measurement methods of T-AOC we used were consistent with the previous study (Zhang et al., 2019), Below shows the calculation of T-AOC:

$$\begin{gathered} \text{Plasma T}-\text{AOC}(\text{U}/\text{ml})=\frac{\text{ Total volume of reaction fluid }}{\text{ Sampling volume }} \\ \times \frac{\text{ Measuring OD value − control OD value }}{0.01} \\ \times \text{ Dilution multiple before sample test }÷30 \end{gathered}$$

2.4.2. Hydrogen peroxide (H₂O₂)

The sample tubes were divided into measuring tubes, standard tube, and blank tube, and the commensurable serum sample, standard sample, and distilled water were added into those three tubes, respectively. We vortexed and mixed, and then added the termination agent. The absorbance was measured at a wavelength of 405 nm.

$$\begin{gathered} {\text{The Plasma H}}_2{\text{O}}_2(\text{mmol}/\text{L})=\frac{\text{ Measuring OD value−blank OD value }}{\text{ Standard OD value − blank OD value }} \\ \times \text{concentrations of Standard sample }(163\text{mmol}/\text{L}) \\ \times \text{ dilution multiple before sample test } \end{gathered}$$

2.4.3. Catalase (CAT)

According to the instructions of the kit, the sample tubes were classified into control tubes and measuring tubes. Before a water bath (37 °C), we added blood plasma samples into the measuring tubes. After water bath, blood plasma samples were added into the control tubes. When the water bath reaction reaches 1 min, we added the terminator and mixed them well. Then, we measured the absorbance at a wavelength of 405 nm. A unit of activity is defined as the amount of H₂O₂ per milliliter of plasma that decomposes 1 μ mol per second, and was calculated using the following formula:

$$\begin{gathered} \text{CAT activity in plasma}(\text{U}/\text{ml})=\frac{(\text{ Control OD value }-\text{ measuring OD value })\times 1.0\times 271}{60\times \text{ Sampleing quantity }} \\ \times \text{ dilution multiple before sample test } \end{gathered}$$

2.4.4. Malondialdehyde (MDA)

The sample tubes were divided into measuring tubes, control tubes, and blank tubes. We added the commensurable absolute ethyl alcohol, standard samples (tetraethoxypropane, 10 nmol/ml), and plasma samples into the three tubes, respectively. These tubes were vortexed, mixed, and then placed in a water bath at the temperature of 37 °C for 40 min, and then kept for 10 min. We took the supernatant and measured the absorbance at a wavelength of 523 nm. The calculation formula was as follows:

$$\begin{gathered} \text{MDA in plasma (nmol/ml) }=\frac{\text{ Measuring OD − control OD }}{\text{ Standard OD − blank OD }} \\ \times \text{ concentrations of standard samples } \\ \times \text{dilution multiple before sample test} \end{gathered}$$

2.5. Statistical analysis

All data were analyzed in SPSS 21.0 (SPSS Inc, Chicago, USA). Data with a normal distribution including T-AOC, H₂O₂, CAT, MDA, and age (year), were described as mean ± SD. The CED-COEs was not the normal distribution and described in median (P₂₅, P₇₅). A two-independent sample t-test was performed to analyze the continuous variables, and a chi-square test was used to analyze categorical variables. After adjusting for gender, age, smoking, drinking, and the BMI index, the analysis of covariance was performed to assess the effects of gender, age, smoking status, drinking status, and BMI on T-AOC, H₂O₂, CAT, and MDA. In order to investigate the association between CED-COEs exposure and oxidative damage, CED-COEs exposure was treated as a continuous variable and a categorical variable (based on quartiles of CED-COEs) in generalized linear models, respectively.

After assessing the association between the cumulative exposure concentration of COEs and oxidative damage, Benchmark Dose Software (BMDS) version 2.63 (U.S.EPA) was used to calculate BMD and BMDL with a benchmark response (BMR) to the risk assessment of exposure. The BMD was a dose at which exposed levels could induce a specified increase in adverse response than the controls. The BMDL was a lower confidence limit of the BMD to explain the uncertainty of the BMD. The BMR level was the specified increase in adverse response, commonly between 1% and 10%. In this study, the selected BMR were 10%. Based on the technical guidelines of EPA for BMDS, the optimal model was selected according to the lowest Akaike’s Information Criterion (AIC) and P-value (> 0.1) from the goodness-of-fit test.

3. Results

3.1. Demographic characteristics of subjects

Table 1 shows that the differences in the age grouping, gender, smoking, and drinking between COEs exposed workers and controls were significant (P < 0.05), but not significant in BMI index (P > 0.05).

Table 1.

The demographic characteristics of subjects.

Characters		COEs-exposed group	Controls	x2/t	P
Age group(%)	≤40	273 (50.4)	141 (59.5)	5.51b	0.019
	>40	269 (49.6)	96 (40.5)
Gender (%)	Male	388 (71.60)	139 (58.6)	12.6b	<0.001
	Female	154 (28.4)	98 (41.4)
Smoking (%)	No	320 (41.0)	196 (82.7)	41.3b	<0.001
	Yes	222 (59.0)	41 (17.3)
Drinking (%)	No	246 (45.4)	137 (42.2)	10.2b	0.001
	Yes	296 (54.6)	100 (57.8)
	≤18.5	12 (2.20)	5 (2.10)
BMI, (kg/m2) (%)	18.5–24	223 (41.1)	107 (45.1)
	24–28	229 (14.4)	101 (42.6)	2.93b	0.403
	≥28	78 (14.4)	24 (10.1)
T-AOC (U/mL)		11.9 ± 4.80	15.4 ± 7.08	6.92a	<0.001
MDA (nmol/mL)		4.65 ± 1.43	5.07 ± 1.60	3.39a	0.001
H2O2 (mmol/L)		58.1 ± 46.7	70.9 ± 68.9	2.62a	0.009
CAT (U/mL)		3.66 ± 2.15	3.79 ± 3.05	0.51a	0.613

^at test.

^bχ² test.

As shown in Table 1, the T-AOC, MDA, and H₂O₂ in the COEs-exposed group were lower than these in the control group, respectively (11.9 ± 4.80 vs. 15.4 ± 7.08, P < 0.05; 4.65 ± 1.43 vs. 5.07 ± 1.60, P = 0.001; 58.1 ± 46.7 vs. 70.9 ± 68.9, P = 0.009). The difference of CAT between the two groups was not statistically significant (P > 0.05).

3.2. Demographic characteristics on T-AOC

Among the oxidative damage biomarkers, the change of T-AOC in two groups was consistent with most studies and we further analyzed the association of CED-COEs and T-AOC and explore the benchmark dose. The T-AOC in exposure group (11.9 ± 4.80) was lower than that in the control group (15.4 ± 7.08) (P < 0.001). The COEs exposed group had significantly lower T-AOC than the control group (P < 0.001) in subgroups of gender, age, smoking, drinking, and BMI except the group with BMI ≤ 18.5 kg/m² (P = 0.169). The T-AOC of females was lower than that of males in both COEs-exposed and control groups (P = 0.002 and = 0.001, respectively). The T-AOC varied by BMI index (P = 0.013). The results are shown in Table 2.

Table 2.

Effects of demographic characteristics on T-AOC.

Variable		N	COEs-exopsure	n	Controls	F		Pa
			(M ± S)		(M ± S)
Gender	Male	388	12.6 ± 4.76	139	17.4 ± 7.11	81.6		<0.001
	Female	154	10.1 ± 4.40	98	12.6 ± 6.06	15.8		<0.001
	F		9.73		11.9
	Pb		0.002		0.001
Age (year)	≤40	273	11.9 ± 4.92	141	15.1 ± 7.24	35.0		<0.001
	>40	269	11.9 ± 4.69	96	15.8 ± 6.84	63.8		<0.001
	F		1.18		0.51
	Pb		0.278		0.474
Smoking	No	320	11.1 ± 4.63	196	14.8 ± 6.91	52.1		<0.001
	Yes	222	13.1 ± 4.83	41	18.4 ± 7.14	36.4		<0.001
	F		3.88		1.77
	Pb		0.050		0.185
Drinking	No	246	11.1 ± 4.52	137	14.2 ± 6.91	36.9		<0.001
	Yes	296	12.6 ± 4.92	100	17.1 ± 6.99	57.4		<0.001
	F		0.022		0.221
	Pb		0.881		0.639
	≤18.5	12	11.3 ± 3.71	5	16.03 ± 13.78		2.46	0.169
	18.5–24	223	10.9 ± 4.63	107	14.34 ± 6.65	45.6		<0.001
BMI (kg/m2)	24–28	229	12.8 ± 4.97	101	16.10 ± 7.50	26.2		<0.001
	≥28	78	12.3 ± 4.40	24	18.26 ± 5.59	18.7		<0.001
	J-Tb		3.61		0.25
	Pb		0.013		0.862

P^a The comparison of T-AOC between the COEs-exposure group and the control group, adjusted for gender, age, smoking, drinking and BMI index. P^bThe comparison of T-AOC in the exposure or in the control group, adjusted for gender, age, smoking, drinking, and BMI index. J-T^b: Jonckheere-Terpstra test.

3.3. Differences in T-AOC by COEs exposure

In this study, the CED-COEs was not the normal distribution and described in Median (P₂₅, P₇₅). The CED-COEs was from 0.052 mg/m³·year to 13.147 mg/m³year in exposure group, and the median (P₂₅, P₇₅) was 1.093 (0.344, 2.143). The CED-COEs of controls ranged from 0.046 mg/m³·year to 0.120 mg/m³·year, and the median (P₂₅, P₇₅) was 0.074 (0.064, 0.090). To explore the relationship between COEs exposure and T-AOC damage, the total population, the males, and the females were all divided into four groups based on the quartiles of cumulative dose of COEs (CED-COEs), with the lowest group as the reference. The results showed a decreasing trend in the T-AOC in the total population, males, and females (Ptrend < 0.001) (Table 3).

Table 3.

Differences in T-AOC by different COEs exposure levels.

CED-COEs (mg/m3·year)	N	T-AOC	β (95% CI)b	χ2	Pa
CED-COEs in total			Ref
0 ~	195	15.3 ± 7.23
0.094 ~	197	12.6 ± 5.53	−3.17 (−4.33, 2.02)	29.2	<0.001
0.425 ~	194	11.7 ± 4.72	−4.12 (−5.22, −3.02)	53.9	<0.001
≥ 1.695	193	12.2 ± 4.76	−4.70 (−5.89, −3.53)	61.3	<0.001
Ptrend			<0.001
CED-COEs in males
< 0.106	133	17.3 ± 7.18	Ref
0.106 ~	130	13.0 ± 4.92	−3.59 (−4.61, −2.58)	48.4	<0.001
0.754 ~	132	12.8 ± 4.90	−3.99 (−5.04, −2.95)	56.1	<0.001
≥ 2.132	132	12.5 ± 4.69	−4.61 (−5.83, −3.39)	54.6	<0.001
Ptrend			<0.001
CED-COEs in females
< 0.084	61	13.3 ± 6.90	Ref
0.084 ~	65	11.2 ± 4.36	−1.44 (−2.81, −0.07)	4.25	<0.001
0.223 ~	64	9.73 ± 4.59	−4.07 (−5.35, −2.79)	38.7	<0.001
≥ 0.875	62	10.4 ± 4.28	−4.51 (−5.69, −3.33)	56.1	0.039
Ptrend			<0.001

^aGeneralized linear models were adjusted for age, gender, smoking, drinking, and BMI index.

^bP < 0.001 compared with the group that CED-COEs less than 0.094 mg/m³ in the total, 0.016 mg/m³ in males, and 0.084 mg/m³ in females.

3.4. Prevalence of T-AOC at different levels of CED-COEs

To further examine the association between different CED-COEs levels and oxidative damage, as well as to facilitate the calculation of the benchmark dose. With the benchmark dose technical guidance of the EPA (U.S.EPA) and considering the cases of T-AOC damage, a cutoff point of the T-AOC was defined as the 10th percentile of the controls in this study. The T-AOC values below the threshold suggested T-AOC damage. Given the difference in the T-AOC distribution for the total, males, and females, the T-AOC of ≤7.77 U/ml in the total population, 2264 9.25 U/ml in males, and ≤6.73 U/ml in females suggested T-AOC damaged. In the total subjects, males, and females, the number of the T AOC damaged cases was then counted in each exposure level. Chi-square linear trend tests showed a dose-response relationship between different CED-COEs levels and the prevalence of oxidative stress levels damaged in total subjects (χ² = 4.32, Ptrend = 0.038). Similar trends were observed in the males (χ² = 6.38, Ptrend = 0.011) and females (χ² = 6.32, Ptrend = 0.012). The results suggested that the T-AOC decreased with increasing CED-COEs levels in the total, males, and females (Table 4).

Table 4.

Trend tests for T-AOC damage at different CED-COEs levels.

CED-COEs (mg/m3·year)	Total			CED-COEs	Males			CED-COEs	Females
	Median value	+/−	%	(mg/m3·year)	Median value	+/−	%	(mg/m3·year)	Median value	+/−	%
<0.094	0.07	19/176	24.7	<0.106	0.07	13/120	9.77	<0.084	0.07	4/57	7.81
0.094 ~	0.22	34/160	17.5	0.106 ~	0.30	24/106	18.5	0.084 ~	0.26	9/56	12.5
0.425 ~	1.01	38/159	17.6	0.754 ~	1.35	25/107	18.9	0.223 ~	1.41	15/49	22.9
≥1.695	2.70	33/160	17.1	≥2.132	3.22	29/103	21.9	≥0.875	2.74	13/48	22.2
χ2		4.32				6.38				6.32
Ptrend		0.038				0.011				0.012

Notes: “+” represented subjects with T-AOC damage.” – “represented subjects without T-AOC damage.

3.5. BMD analysis for the BMD and BMDL of COEs exposure

The BMD and BMDL were estimated for the total population, the males, and female population. Based on technical guidance of the benchmark dose, we tested all models in the BMD modeling guidelines to find an optimal model, including gamma, Weibull, Quantal-Linear, Logistic, Log-Logistic, Probit, Dichotomous-Hill, Log-Probit, Multistage, and Gamma-BgDose. Finally, the Log-logistic model was considered as the optimal model and was used for the male and female population in this study. The Log-Logistic dose response model formula was P [T-AOC damaged] = background+(1-background)/[1 + e ^{(−intercept-slope×Log}(dose))]. The goodness-of-fit test value was poor for the total population after trying all models in BMD specified in the BMD modeling guidelines, i.e., an optimal model was not found for the total population.

From the Log-Logistic model and at 10% of the BMR level, the BMD and the BMDL were 2.83 mg/m³·year and 1.47 mg/m³·year for the males, respectively. The BMD and the BMDL were 1.39 mg/m³·year and 0.75 mg/m³·year for the females, respectively (Table 5). The dose-response curves of the males and the females in the log-logistic model were shown in Fig. 1.

Table 5.

The estimated BMD and BMDL of COEs exposure on T-AOC damage in different subjects.

Subjects	n	Log-Logistic model			BMD10aa	BMDL10a	AIC	goodness-of-fit
		background	intercept	slope				χ2	P
Total	779	-	-	-	-	-	685	6.73	0.035
Males	527	1	−0.68	1	2.83	1.47	484	3.63	0.163
Females	252	1	−0.67	1	1.39	0.75	281	1.60	0.449

Abbreviations: COEs: Coke Oven Emissions; BMD: benchmark dose; BMDL: lower 95% confidence limit of BMD; AIC: Akaike Information Criterion.

^aUnit of measurement, mg/m³·year.

Fig. 1.

Use of Log-Logistic model to estimate the BMD and BMDL of cumulative COEs exposure for T-AOC damage incidence. (a) The BMD and BMDL of cumulative COEs for T-AOC damage incidence in males; (b) The BMD and BMDL of cumulative COEs for T-AOC damage incidence in females; In figures, the X-axis (dose, mg/m³·year) described cumulative COEs dose, and the Y-axis (fraction affected, %) described the rates of T-AOC damage. Abbreviations: COEs, coke oven emissions; T-AOC, total antioxidant capacity; BMD and BMDL, benchmark dose and its 95% lower bound confidence limit.

Given the different retirement ages in males and females, the working duration varied by sex. Dividing the total working durations of 40 years and 35 years for males and females, respectively, we could calculate a reference value for the threshold limit value-time weighted average (TLV-TWA). Taking the BMDL = 1.47 mg/m³·year for males and the BMDL = 0.75 mg/m³·year for females participants as starting points, the COEs exposure limits were 0.037 mg/m³ and 0.021 mg/m³ for male and female workers in a working system of 8 h, respectively.

4. Discussion

Coke oven emissions (COEs) are a major occupational hazard factor for coke oven workers, which contain a large number of carcinogenic polycyclic aromatic hydrocarbon (PAHs) substances. Most evidence showed that PAHs was chemical carcinogens. When PAHs enters the body through many ways, hydroxyl radical, hydrogen peroxide (H₂O₂), and other reactive oxygen species will be produced in the process of metabolism (Fu et al., 2012). The oxidant/antioxidant keeps the balance under normal physiological conditions, and the antioxidant in vivo could protect against oxidative stress induced by ROS(Eom et al., 2013). When a large amount of external PAHs substances stimulated the human body, with increasing exposure of external stimulations, the increasing reactive oxygen species (ROS) would break the equilibrium between ROS and antioxidant capacity (Hong et al., 2009). The increasing ROS levels would lead to the lower antioxidant ability because of the depletion of antioxidant. Oxidative stress would be induced, and thereby causing oxidative damage of DNA (Zhai et al., 2012), protein and lipid peroxidation (Fu et al., 2012) to induce damage, such as in-flammation, tumor, and aging (Sohal and Weindruch, 1996). Oxidative stress may be an important mechanism in PAHs-induced disease health hazards.

Many biomarkers were used to assess the oxidative stress levels (Marrocco et al., 2017). The activity or content of those biomarkers could be the important indicators reflecting oxidative damage from the different perspectives in vivo. The malondialdehyde (MDA) was a kind of reliable biomarker to reflect the damage level of lipid peroxide (Del Rio et al., 2005). As one of the primary ROS that human generated, hydrogen peroxide (H₂O₂) could reflect the levels of free radicals in body (Wu et al., 2013). CAT reflected the enzyme activity in antioxidant system and mainly decompose H₂O₂ free radical (Wu et al., 2013). The total antioxidant capacity (T-AOC) could reflect the total antioxidant level of the body defense system inside the body (Sharifian et al., 2005).

The four biomarkers were assessed in this study, the results suggested that the T-AOC in exposed group was lower than that in the control group, and the MDA and H₂O₂ was lower than that in the control group. In general, the MDA content would increase under COEs/PAHs exposure environment in vivo (Yang et al., 2007, 2015). However, some studies also showed that there was no significant association between MDA and OH-PAHs in PAHs exposure environments (Jeng et al., 2011; Lu et al., 2016). In this study, the results showed that MDA decreased in the exposed group compared with that in the control group. As the final products of the peroxidation of polyunsaturated fatty acids, MDA may be influenced by many other factors in vivo or vitro, and then leading to the decreased MDA results. The H₂O₂ would increase under the stimulation of chemicals. In this study, the results showed that the H₂O₂ in COEs-exposed group was lower than that in control group. When oxidative stress occurred, other metabolic enzyme in body also could be activated to decompose the H₂O₂ and other peroxides (Agarwal et al., 2018). It may be that glutathione peroxidase (tGPx) or other antioxidases were activated to decompose H₂O₂ in the COEs-stimulated environment, and thus the H₂O₂ was lower in the COEs-exposed group. In this study, although the CAT activity in exposed group was slightly lower than that in control group, the difference was not significant (P > 0.05).

Environmental chemicals exposure usually leads to a lower T-AOC in the exposed population (Hormozi et al., 2018; Shraideh et al., 2018). A lower level of T-AOC was more likely to suffer from oxidative stress and cause diseases (Eom et al., 2013). In this study, the T-AOC in COEs exposure group was lower than that in control group, suggesting that T-AOC could be a suitable biomarker reflecting oxidative damage levels. We further analyzed the effects of demographic characteristics on T-AOC. The results showed that the effects of age and drinking on T-AOC were not statistically significant (P > 0.05). The T-AOC in smoking subjects were slightly higher than that in non-smoking subjects in both exposed and control group (Table 2). Similarly, Eom et al. study also reported that the T-AOC in smokers were higher than that in never-smokers (Eom et al., 2013). One possible explanation was that the large amount of ROS would be produced in smokers when comparing with the no-smokers and the T-AOC as the defense mechanism biomarker of body would increase by smoking. Our pervious study showed that the T-AOC was decreased in COEs exposure workers when compared with the controls, and the T-AOC in females was lower than that in males (Zhang et al., 2019). We further confirmed the association between T-AOC and COEs exposure in this study, suggesting that the T-AOC biomarker could be used to reflect oxidative damage of COEs exposure.

Coke workers with long-term exposure to COEs were more likely to suffer from oxidative stress than others. In 2007, the Ministry of Health of the People’s Republic of China set up the OELs at 0.1 mg/m³ of PCTWA (benzene-soluble fraction)(MOH, 2007). The OELs by the NIOSH and OSHA were 0.20 mg/m³ (benzene-soluble fraction) and 0.15 mg/m³ (benzene-soluble fraction), respectively. A meta-analysis reported that coke oven workers had a 30% higher risk of lung cancer if they were exposed to the COEs levels above the OEL (Rota et al., 2014). Current standards may still not be effective to prevent workers from suffering from non-cancer risks like oxidative damage.

The BMD method, proposed by Crump, has been extensively used to access the risk and calculate the safely limited value of occupational hazardous substance exposure, for example blood lead (Wang et al., 2011a), vinyl chloride monomer (Wang et al., 2011b), and formaldehyde exposure (Wang et al., 2019a). According to the EPA, the BMD method was considered as a better method than NOAEL for the risk assessment research (Rota et al., 2014). More information can be derived from the BMD method than the NOVEL method, and the BMD had a higher quantitative sensitivity and accuracy (Izadi et al., 2012). Consequently, the BMD method was proposed to define an acceptable and reasonable level of human potentially toxic substances exposure due to the above advantages (Barnes et al., 1995).

Our previous study assessed the COEs exposure levels from telomere damage levels using the BMD method, and the exposure limits of COEs were recommended at 0.024 mg/m³ in males and 0.018 mg/m³ in females (benzene-soluble fraction) (Wang et al., 2019b). In this study, we assessed the COEs exposure limits from oxidative damage levels, and we found that the reference COEs exposure limit in males and females were 0.037 mg/m³ and 0.021 mg/m³, respectively. The COEs exposure limits evaluated by biomarkers of peripheral blood telomere length and plasma T-AOC damage were lower than the current standards of 0.10 mg/m³ (The Ministry of Health of the People’s Republic of China), 0.15 mg/m³ (Occupational Safety and Health Administration, OSHA), and 0.20 mg/m³ (National Institute of Occupational Safety and Health, NOISH). Compared to the current standard from the traditional endpoints, the COEs exposure limits estimated were lower in our study. However, as the non-disease-specific biomarkers, the oxidative damage usually appeared in the early stage of actual cancer or diseases, and T-AOC may be a more sensitive biomarker to reflect the early health damage. Therefore, the health effect value of this study is more sensitive. This is the first study of a benchmark dose study of COEs based on T-AOC damage. Based on our results, the occupational exposure limit of COEs was recommended at 0.03 mg/m³ (benzene-soluble fraction).

Under the current standards, there are still health hazards reported, which gave us a hint that coke oven workers may not be sufficiently protected. The new reference threshold limits obtained by some new molecular biomarkers in early stage of diseases may be more sensitive, and may have better protection for exposed workers. It should be stated that the BMD values would be influenced by the different endpoints studied (Nishijo et al., 2014), more specific and sensitive biomarkers are in need to assess occupational exposure levels of COEs exposure in future studies. In addition, given the complicated working environment of coke oven workers, the suggested OEL obtained from this study should be interpreted cautiously, and more studies from other COEs-exposed environment or coking plants are needed to test and verify the feasibility of the obtained OEL based on oxidative damage levels.

In conclusion, the study was the first of it kind that estimated the exposure limit levels of COEs from the oxidative damage levels. The results suggested that an exposure level of COEs below the current national occupational exposure limits (OELs) would still induce oxidative damage in coke oven workers, and the OEL of COEs based on the T-AOC damage was recommended at 0.03 mg/m³ in this study. The results could shed light on a more stringent occupational exposure level standard for COEs exposure.