- Email: [email protected]
Dietary exposure to endocrine disrupting chemicals in metropolitan population from China: A risk assessment based on probabilistic approach
Chemosphere 139 (2015) 2–8
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Dietary exposure to endocrine disrupting chemicals in metropolitan population from China: A risk assessment based on probabilistic approach Dongliang He a,e, Xiaolei Ye b, Yonghua Xiao c, Nana Zhao b, Jia Long a, Piwei Zhang a, Ying Fan a, Shibin Ding a, Xin Jin a, Chong Tian a, Shunqing Xu d, Chenjiang Ying a,⇑ a Department of Nutrition and Food Hygiene, Hubei Key Laboratory of Food Nutrition and Safety, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China b School of Environmental Science and Public Health, Wenzhou Medical University, Wenzhou 325000, China c Wuhan Center for Disease Control and Prevention, Wuhan 430015, China d Ministry of Education Key Laboratory of Environment and Health, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China e Department of Preventive medicine, School of Public Health, University of South China, Hengyang 421001, China
h i g h l i g h t s Seven out of eight EDCs were detected in most raw and fresh food in China. NP, EE2, and BPA coexisted in food. Human exposure to NP, EE2 and BPA was far below TDI.
REEQs exposure in population was considerably higher than E2.
The risk of co-exposure to multiple EDCs should not be ignored.
a r t i c l e
i n f o
Article history: Received 29 March 2014 Received in revised form 16 April 2015 Accepted 14 May 2015
Keywords: Endocrine disrupting chemicals (EDCs) Dietary exposure Probabilistic modeling Risk assessment P 17b-Estradiol equivalent ( EEQs)
a b s t r a c t The intake of contaminated foods is an important exposure pathway for endocrine disrupting chemicals (EDCs). However, data on the occurrence of EDCs in foodstuffs are sporadic and the resultant risk of co-exposure is rarely concerned. In this study, 450 food samples representing 7 food categories (mainly raw and fresh food), collected from three geographic cities in China, were analyzed for eight EDCs using high performance liquid chromatography tandem mass spectrometry (HPLC–MS/MS). Besides estrone (E1), other EDCs including diethylstilbestrol (DES), nonylphenol (NP), bisphenol A (BPA), octylphenol (OP), 17b-estradiol (E2), 17a-ethinylestradiol (EE2), and estriol (E3) were ubiquitous in food. Dose-dependent relationships were found between NP and EE2 (r = 0.196, p < 0.05), BPA (r = 0.391, p < 0.05). Moreover, there existed a correspondence between EDCs congener and food category. Based on the obtained database of EDCs concentration combined with local food consumption, dietary EDCs exposure was estimated using the Monte Carlo Risk Assessment (MCRA) system. The 50th and 95th percentile exposure of any EDCs isomer were far below the tolerable daily intake (TDI) value identically. P However, the sum of 17b-estradiol equivalents ( EEQs) exposure in population was considerably larger than the value of exposure to E2, which implied the underlying resultant risk of multiple EDCs in food should be concern. In conclusion, co-exposure via food consumption should be considered rather than individual EDCs during health risk evaluation. Ó 2015 Elsevier Ltd. All rights reserved.
⇑ Corresponding author. E-mail addresses: [email protected] (D. He), [email protected] (X. Ye), [email protected] (Y. Xiao), [email protected] (N. Zhao), [email protected] (J. Long), [email protected] (P. Zhang), [email protected] (Y. Fan), [email protected] (S. Ding), [email protected] (X. Jin), tianchong [email protected] (C. Tian), [email protected] (S. Xu), [email protected] (C. Ying). http://dx.doi.org/10.1016/j.chemosphere.2015.05.036 0045-6535/Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Endocrine disrupting chemicals (EDCs), a group of substances found in the diet and environment, have characteristic activity of estrogens or androgenic activity. To date, a number of EDCs
D. He et al. / Chemosphere 139 (2015) 2–8
including nonylphenol (NP), bisphenol A (BPA), octylphenol (OP), 17b-estradiol (E2), 17a-ethinylestradiol (EE2), estrone (E1), estriol (E3), and diethylstilbestrol (DES) have been applied to manufacture various daily products widely and identified persistent in the environment worldwide (Casals-Casas and Desvergne, 2011; Nohynek et al., 2013). Toxicological studies have revealed that, to a certain extent, EDCs could mimic the effect of endogenous estrogens or interfere with estrogen signaling pathways (Pelekanou and Leclercq, 2011) and ultimately alter growth, development and reproduction of wildlife as well as human beings (Frye et al., 2012). Besides air, water and even skin contact (Lu et al., 2013), several studies revealed that consumption of food is another important source of EDCs for the non-occupational population (Trasande et al., 2012). With the increasing concern on the potential risk of EDCs to human via food intake, the European Food Safety Authority (EFSA) has established a series of tolerable daily intake (TDI) value, i.e., 50 lg/kg bw for BPA and EE2 (EFSA, 2006). As a supplementary, Danish researchers proposed a TDI value 5 lg/kg bw for NP (Nielsen et al., 2000). A few pilot studies have reported the occurrence of EDCs in food items collected in some regions, and the dose of independent exposure to them was well below TDIs. For examples, 7.50 lg/day for adult was found for BPA in Germany (Guenther et al., 2002) and around 30.0 lg/day was observed in Taiwanese (Lu et al., 2007). It is, however, insufficient to eventually eliminate the worry of health risk from co-exposure to multiple EDCs for the term 17b-estradiol equivalency factor (EEF) raised by a few recent studies (e.g., Jin et al., 2013). Since then, it is of persistent interest to precisely detect the EDCs concentrations in food and assess the risk of co-exposure to EDCs via food in consideration of the 17b-estradiol equivalent factor (EEF) (Table S1). A precise assessment of risk is usually based on two critical factors: methods of accurate detection and applicative statistic theory. Several analytical methods such as High Performance Liquid Chromatography (HPLC), and Liquid chromatography–tandem mass spectrometry (LC–MS/MS), have been developed and widely used for EDCs determination even at a trace level (Locatelli et al., 2011; Fayad et al., 2013; Wang et al., 2013). Deterministic point estimation and probabilistic modeling were the two commonly used statistical approaches for estimating exposure to environmental chemicals (Jensen et al., 2008). The deterministic approach estimates the dietary intake by multiply the mean chemical residual level and mean food consumption together and was of more understandable but reckons without the variation and uncertainty in both level and consumption data between individual (FAO/WHO, 2001). The probabilistic exposure assessment, on the other hand, takes account of the default of the former by generating distributions of the level and food consumption data and serving as model parameters inputted. The outputs based on the simulation could be specified at any percentiles etc., which make it more suitable for assessing the exposure of whole population (IPCS, 2008). Professional software like the Risk Simulation and the Monte Carlo Risk Assessment system (MCRA) are commonly used for such analyses. In the present study, we conducted a cross-sectional investigation at three Chinese cities, i.e., Wuhan (WH), Guangzhou (GZ) and Jinan (JN). Concentrations of eight selected EDCs in raw and fresh foods purchased from local markets were tested simultaneously, and analyzed using a Monte Carlo simulation to describe exposure to EDCs via food sources, based on the 17b-estradiol equivalents (EEQs) theory. Compared with the respective TDI value of chemical, the cumulative probability distribution of exposure to EDCs for the general population was discussed. To the best of our knowledge, this is the first attempt to model the exposure to EDCs from food sources in Chinese population via probabilistic approach. The result will not only reveal the contamination status of EDCs in
3
foods commonly consumed in China, but also be used for assessing the health risk of EDCs in Chinese population.
2. Materials and methods 2.1. Chemicals and reagents All EDCs standard chemicals were purchased from Sigma (St. Louis, MO, USA). b-Glucuronidase/arylsulfatase and all organic solvents (HPLC grade) including methanol, acetonitrile, acetone, hexane, and dichloromethane were purchased from Merck (Darmstadt, Germany). BPA-d1 and 13C2-E2 were obtained from CDN Isotopes Inc. (Quebec, Canada) and Dr. Ehrenstorfer (Augsburg, Germany), respectively. The ultrapure water used in this study was treated by Millipore pure water system (Millipore Co., France). Stock solutions of the standards were prepared in methanol at a concentration of 10.0 mg/L and stored at 20 °C. On the day of analysis, the stock solutions of standards were further diluted into a concentration of 10.0 lg/L and for preparation of a set of pooled calibration curves.
2.2. Sample collections The study protocol was designed for taking a representative sample in the eastern mainland China (Fig. S1). According to the geographical distribution of Chinese population, the region in the east of Hu’s line constituted 42.9% of national territory but encompassed 94.4% of Chinese population (Tu and Peng, 1994). In this region, Wuhan city (WH), Guangzhou city (GZ) and Jinan city (JN) were embraced into the follow-up studies, which are all metropolises located at different basins of large rivers (e.g., Yangtze, Pearl, and Yellow Rivers). At each site, raw and fresh food samples of different species and origins were sampled by trained assistants and stored in pre-cleaned amber glass bottles wrapped in aluminum foil. During the summer (July–September) of 2012, 7 food categories and 450 food items (Table S2), mainly raw and fresh food, were chosen to represent available varieties which were commonly consumed by the local residents on the basis of regional and national surveys on food consumption in China (Jin, 2008). The majority of food samples were purchased from local supermarkets, and a few items were purchased from large retail stores. Samples were transported to the laboratory on ice and stored at 20 °C until analysis.
2.3. Sample pretreatment The extraction for the objective compounds and clean-up was abided by procedure which had been reported elsewhere (Shao et al., 2005) and with minor modifications. Food samples were classified into two groups: solid foods and liquid foods (e.g., cooking oils). Solid foods were homogenized, weighed (5.0 g fresh weight), spiked with internal standards (BPA-d1 and 13C2-E2), and adjusted the PH value (0.2 M acetate, PH = 5.2), then incubated with 100.0 lL b-glucuronidase/arylsulfatase (37 °C, 24 h) to improve the separation of binding EDCs from food matrix (Schmidt et al., 2013). After evaporation to near dryness by nitrogen with a rotary evaporator, the extract was re-dissolved in 2.0 mL of 10% dichloromethane/hexane and purified with an OasisÒ HLB cartridge (60 mg/3 cc, Waters, USA). The final elute was concentrated to 1.0 mL, flowed through a 0.45 mL filter and stored at 4 °C until instrumental analysis. The cooking oil samples (5.0 g) were extracted using the same procedure described above, without the incubate step.
4
D. He et al. / Chemosphere 139 (2015) 2–8
2.4. Instrumental analysis A Waters UPLC-TQD system (Waters Co., USA) was used for separation and detection of target analytes. The analytical column was Acquity BEH C18 (2.1 100 mm, 1.7 lm, Waters, USA). The mobile phase was constituted of methanol and Ammonium acetate, used for LC separation at 35 °C with gradient elution. The mobile flow rate was 200.0 lL/min, and 10.0 lL of the sample extract was injected. The MS/MS was operated in an electrospray negative ionization multiple reaction monitoring (MRM) mode. The capillary voltage was held at 2.5 kV. The cone voltage and other parameters were listed in Table S3. 2.5. Quality Assurance and Quality Control (QA/QC) To avoid the potential contamination from analytical containers or reagents, no plastic was used in the experiment thoroughly, and all glassware utilized underwent even previously baked at 450 °C for 4 h. Each analysis run (50–60 samples) included full procedure blanks (n = 3), matrix spiked samples (n = 3) and at least six different concentrations of target analyte standards. The concentrations of target analytes in full procedure blanks were below the corresponding limit of detections (LODs). The average recoveries of target analytes spiked into full procedure blanks ranged from 91% to 113%, and those spiked into food matrices for solid foods (meats) ranged from 82% to 101% and for liquid foods (cooking oils) ranged from 79% to 105% (Table S4). The relative standard deviations (RSDs) for the target analytes were not higher than 10.00%. The standard calibrations were linear (r2 > 0.99) and spanned five orders of magnitude, and the results were corrected for the recoveries of the internal standards (BPA-d1 for NP, BPA, and OP, 13C2-E2 for EE2, E1, E2, E3, and DES). The LODs for NP, BPA, OP, EE2, E1, E2, E3, and DES were 0.05, 0.03, 0.05, 0.02, 0.05, 0.04, 0.04 and 0.02 ng/g, respectively. 2.6. Food consumption data The food consumption data were obtained from the nationwide dietary survey conducted in October 2002, which used a 7-day prospective food record with pre-coded questionnaire that included answering categories for the most commonly eaten foods in Chinese diet. In the present study, data of regional dietary patterns from Guangzhou, Wuhan, and Jinan were respectively cited, which were presented in the form of gram per 60 kg body weight (Jin, 2008). 2.7. Data analysis and Monte Carlo simulation For general statistical analysis, values detected below the LOQ were set to half the LOQ. Spearman’s rank correlation and correspondence analyses were conducted using the SPSS software version 12.0 (SPSS Inc., Chicago, IL, USA). A value of p < 0.05 was P denoted as statistical significance. We defined EEQs value as the co-effect of multiple EDCs congeners. The intake dose for food exposure to various EDCs was assumed to be an indicator of chronic exposure condition. The estimated dietary intake (EDI, ng/kg bw/day) for a typical adult was calculated on the basis of Eq. (1) which had been detailed elsewhere (Trudel et al., 2011) and with minor modifications:
EDI ¼
n X ðC ki ak Þ bw
ð1Þ
i¼1
where Cki (ng/g) is the concentration of congener i in the sample matrix of food group k, ak (g/day) is the amount of daily consumption of food group k, and bw is the body weight. In the present
study, since the data of food consumption cited were presented in gram per 60 kg bodyweight status (Jin, 2008), we integrated the bodyweight into consumption data and fixed it on 60 kg in calculation. Eq. (2) was used to estimate the integrated effect of EDCs:
EDIEEQs ¼
n X ðC ki f i ak Þ bw
ð2Þ
i¼1
where fi is the EEF value of EDCs congener. Individuals differ in food consumption and EDCs concentration, and the different value of Cki and ak would apparently result in various EDI values. If the mean or 95th percentile concentration of EDCs and food consumption were available for exposure calculation merely, the result derived of equations is going against the fact and reflect special exposure scenarios (50th or 95th percentile) only. Considered of the variation and uncertainty aroused from the EDCs level and individual intake of different foodstuff, the probability distribution assumptions of EDCs concentration was used to forecast the distribution of the exposure based on MCRA system (conducted by Software Crystal Ball 11.0) which is more compatible for dealing with the intrinsic EDCs level and food consumption variation or uncertainty. 3. Result 3.1. EDCs levels in food samples Most of the selected EDCs were detected in cereals but E1 was under the LOD in all samples (Table 1). NP, BPA and OP were found in all of the cereals. NP was the EDCs with the highest concentration and ranged from 5.88 to 79.08 ng/g. The concentrations were between 1.05 and 9.20 ng/g for OP, 0.96 and 29.04 ng/g for BPA. DES and EE2 were detected in the majority of cereal items (found in 45 out of 50 samples) at the levels ranged from below LOD to 0.71 and 35.04 ng/g, respectively. E2 and E3 were occasionally found in these samples. In addition, the EDCs concentrations (ng/g) in different categories of food from three cities were presented in Table S5. NP, BPA and EE2 shared common high detection rate in most of the food categories from the different cities. Similar to the results observed in cereals, detected rate of NP, OP, and BPA ranked the top three in vegetables (found in 62, 57 and 55 out of 62 samples) and the order of concentration was NP > BPA > OP. The other EDCs were also detected in some of vegetable samples and ranged from below LOD to different maximum values. Among the 51 meats and proceeded meat product samples, E2 and BPA got the highest detected rate (found in 51 out of 51 samples) and followed by NP and OP. However, the median concentration of NP (6.56 ng/g) stood first and followed by E2 (6.36 ng/g). Furthermore, both the detected rate and concentration of E2 take the leading level among similar compounds in fish and egg. The sum of EEQ of multiple EDCs in each food group was calculated based on EEF values and showed (Tables 1 and S5). In any P food group, the EEQs value was far greater than a single congener. Vegetable, cereal, meat, fish, fruit and egg ranked first to P sixth in term of median EEQs value and ranged from 1.95 to 15.82 ng/g. In addition, the concentration of EDCs also showed a matrix dependent distribution. The concentration of one congener usually increased with the increase or decrease of other EDCs level in a food category. The non-parametric Spearman correlation analysis revealed weak but positive relationships between NP and EE2 (r = 0.196, p < 0.05), NP and BPA (r = 0.391, p < 0.05), which suggested co-occurrence and similarity in food sources (Table S6). Correspondence analysis between EDCs level and food category was carried out. Two dimensions were adapted to represent main
5
D. He et al. / Chemosphere 139 (2015) 2–8 Table 1 Concentrations (ng/g fresh weight) of EDCs congeners in different food categories. OP
NP
EE2
BPA
E3
E2
P EEQs
Cereals and cereal products (n = 50) Positive samples 45 Mean 0.259 Median 0.245 Minimum
50 3.493 2.820 1.050 9.200
50 16.429 8.755 5.880 79.080
45 9.142 7.595
50 7.708 5.720 0.960 29.040
35 0.674 0.360
28 0.521 0.355
– 17.230 14.335 0.210 63.390
Vegetables (n = 62) Positive samples Mean Median Minimum Maximum
36 1.762 0.120
57 2.273 1.320
62 25.968 17.760 4.920 90.960
43 9.938 6.645
55 8.073 6.540
41 6.218 0.480
43 1.441 0.240
– 23.214 15.825 0.020 206.820
Fruits (n = 75) Positive samples Mean Median Minimum Maximum
25 0.137
51 0.322 0.240
69 4.552 4.250
57 0.689 0.770
70 3.588 2.280
36 0.443
62 0.987 0.870
– 2.525 2.580 0.240 6.020
Meat and meat products (n = 51) Positive samples 36 Mean 2.487 Median 0.190 Minimum
43 2.376 1.870
49 8.680 6.560
20 0.462
51 2.348 1.880 0.400 7.680
34 7.329 0.480
51 6.638 6.360 2.360 15.040
– 13.106 7.670 2.750 162.950
Fish and seafood (n = 99) Positive samples Mean Median Minimum Maximum
70 4.240 0.680
87 1.616 1.320
84 7.198 4.520
34 0.408
81 1.505 1.080
86 2.309 1.680
99 5.221 4.830 1.960 17.200
– 13.364 7.400 2.250 518.700
Eggs (n = 54) Positive samples Mean Median Minimum Maximum
39 0.458 0.355
36 1.439 0.380
31 0.427 0.435
0
33 0.405 0.320
35 0.319 0.295
54 1.274 1.230 0.440 2.710
– 2.091 1.955 0.440 4.720
Cooking oils (n = 59) Positive samples Mean Median Minimum Maximum
3 0.016
8 0.033
52 6.396 5.710
12 0.150
9 0.130
0
0
– 0.293
DES
LOD, limit of detection.
factors, which explained most of original variables. In the correspondence analysis plot (Fig. 1), the location of NP and oils, fruits and vegetables were close and all fell in the first quadrant. This result suggested that specific congener NP had higher contributions in oils, fruits and vegetables food categories. Similarly, we can deduce E2, E3, OP and DES were the main congeners occurred in meat, fish and egg food group for they were all in the third quadrant. BPA and EE2 were the primary contaminants in cereals rather than other congeners.
In addition, S-curve was adopted to present the cumulative freP quency and probability for exposure to EDCs and EEQs in populations (Fig. 2). The analysis of the intersections of curves and horizontal scale in the diagram suggested the exposure to seven P detected EDCs congeners and EEQs level could easily be quantified corresponding to any percentile scenarios. In the overlaid chart P (Fig. 3), the non-normal distributions of E2 and EEQs were P demonstrated. The vertical bars in distributions of EEQs, indicating the 50th forecasted values, were on the far right of E2. This result meant that the exposure to multiple EDCs in population is considerably larger than that of E2.
3.2. Human exposure to EDCs via food intake The distribution patterns of daily EDCs exposure via MCRA demonstrated the variation between individuals (Table 2). Both in mean (50th) and high (90th) scenarios, the NP exposure identically ranked first in three geographic cities, with 0.31 and 0.65 lg/kg bw/day for Guangzhou, 0.14 and 0.25 lg/kg bw/day for Wuhan, and 0.18 and 0.85 lg/kg bw/day for Jinan, respectively. P As for EEQs exposure, Guangzhou population at mean scenario but Jinan population at high scenario rated at the top of the list, at the value of 0.27 lg/kg bw/day and 0.79 lg/kg bw/day, respectively.
4. Discussion Currently, the use of EDCs in China is less subject to government regulation. In the present study, we investigated the selected EDCs in food which have been reported sporadically (Niu et al., 2012). The results demonstrated the EDCs were usually co-occurrence in foodstuffs. Based on the datasets of EDCs level and food consumption, the daily dietary exposure and their primary food sources was evaluated by probability simulation.
6
D. He et al. / Chemosphere 139 (2015) 2–8
at low and individually ineffective concentrations. The combined effects of EDCs compounds should therefore be included as well when we assess their ecological impacts on the environment (Jin et al., 2012). To date, there is no unanimously accepted parameter which has been recommended in field of risk assessment on co-exposure to P EDCs. As a pilot, EEQs value calculated based on the 17b-estradiol equivalency factor (EEF) was brought into this study for analyzing the co-exposure risk. In various food categories, P EEQs values were 10-fold higher than E2 values (Table 1), suggesting that a reasonable risk assessment should consider the co-exposure of multiple congeners. Except NP, OP and BPA were relatively high detection rates and concentrations but less contriP bution to EEQs for their low EEF value. The positive relationships P between DES and EEQs value (r = 0.887, p < 0.05) implied that DES was identified as the main contributor in all of the three matrixes for its higher EEF value. 4.2. Simulation for human exposure to EDCs Fig. 1. Diagram of corresponding analysis of EDCs and food categories. Projection of correspondence analysis showed the relationship between EDCs isomer and food category. On the basis of the statistical theory of correspondence analysis, there were corresponding relations between EDCs and food categories when their locations were close and fell in the same quadrant.
4.1. EDCs levels in Chinese food The ubiquitous presence of NP, BPA and OP in most of Chinese foodstuffs was revealed by analyzing eight EDCs compounds foodstuffs sampled from three regions of different river basins in China. The results were generally in line with the previous reports in other regions (Guenther et al., 2002; Lu et al., 2013). However, the concentrations in foodstuffs detected in this study were 5-fold higher than that noted by Guenther et al. (2002), 0.1– 19.4 ng/g wet weight. The detection rates of BPA in different food categories analyzed in our study were generally higher than those reported in previous studies (13 out of 27 for meat and 3 out of 10 for egg) (Shao et al., 2007; Cunha et al., 2011). While this may partially due to the differences in collection and number of food samples analyzed, the use of BPA in materials intended to come in contact with foodstuffs is increasing in recent years. Correspondence analysis was rarely used for the similar studies in the literature. In this study (Fig. 1), two representative dimensions explained the most original variables which suggested that the primary EDCs contamination in various food categories were specific. The non-parametric Spearman correlation analysis for the relative abundance of EDCs revealed positive relationships between various congeners, suggesting that people always co-expose to multiple EDCs via food consumption simultaneously. The evidence obtained indicates the potential and resultant health hazards of mixtures of EDCs even when each EDCs congener occurs
By using Monte Carlo method, we simulated the detected EDCs exposure on different scenarios synchronously and found that the forecasted median (50thpercentile), even high (95th percentile) exposures for NP, EE2 and BPA (with TDI value available) were far below the corresponding TDI values. The NP intake dose was lower than the value reported in a Taiwanese population (i.e., 38.59 ± 29.32 ng/day) (Lu et al., 2007) and the intake dose of BPA was far below the value estimated by another study in a Chinese population (489 ng/kg bw/day) (Liao et al., 2013). Such inconsistency may be partially contributed by the difference between the deterministic and the probabilistic approaches (Jensen et al., 2008). As described in our result (Table 1), the median values were generally lower than the mean values which suggested a non-normal distribution of EDCs. In fact, the majority of EDCs distributions analyzed by Monte Carlo simulation were identified as log-normal distribution and the representativeness of mean values should be questioned. If an inappropriate mean value (not normal distribution) was inputted into the equation for calculating the daily intake dose, the output value would be bias inevitably. Our results indicated that the co-exposure to multiple EDCs was considerably larger than the exposure to E2 independently (Fig. 3). It should be considered as an alert of the default in assessment on health risk of exposure to EDCs isomer and regardless of its combined effect. 4.3. Limitation Firstly, although the consumption data used in this study are the most recent data (released in 2008), they were collected ten years ago. Over the last ten years, consumption patterns have changed very likely due to the presence of new food products on
Table 2 Distribution of exposure (lg/kg bw/day) to various EDC congeners for Guangzhou, Wuhan and Jinan population. TDI value
DES OP NP EE2 BPA E3 E2 EEQs
NA NA 5.0 50.0 50.0 NA NA NA
Guangzhou population
Jinan population
Wuhan population
Distribution
P50
P90
Distribution
P50
P90
Distribution
P50
P90
Log Log Log Log Log G Log M.E
0.01 0.05 0.31 0.12 0.14 0.07 0.04 0.27
0.03 0.12 0.65 0.20 0.23 0.12 0.13 0.43
Log Log Log Log Log Log Log Log
0.00 0.02 0.14 0.11 0.04 0.01 0.02 0.16
0.02 0.03 0.25 0.23 0.12 0.20 0.05 0.79
Log Log Log Log Log Log Log Log
0.01 0.04 0.18 0.05 0.08 0.01 0.02 0.14
0.07 0.08 0.85 0.32 0.17 0.03 0.04 0.58
NA, no available; Log, lognormal distribution; G, gamma distribution; M.E, max extreme distribution.
D. He et al. / Chemosphere 139 (2015) 2–8
Fig. 2. The cumulative probabilities of EDCs congeners and population, respectively.
7
P EEQs exposure in population. A, B and C represented exposure (lg/kg bw/day) for Guangzhou, Wuhan and Jinan
P P Fig. 3. The distributions of EEQs and E2 exposure in population. The overlaid distributions of A, B and C described EEQs and E2 exposure for Guangzhou, Wuhan and Jinan population, respectively. The vertical bars represented the medians of the forecasted values.
8
D. He et al. / Chemosphere 139 (2015) 2–8
the market. Moreover, the EDCs concentrations were not nationwide monitoring data including all food categories but measured in 450 foodstuffs bought on the market very recently and the representativeness can be further improved. The deficiency of full scale in sampling can give rise to under-estimation in exposure. P Secondly, EEQs was brought into consideration of co-effect of exposure to multiple EDCs simultaneously; however, no available TDI value was authoritatively released. The significant difference P between E2 and EEQs can only suggest that even far below regulation TDI value of any single isomer is not enough to identify the exposure as harmless. 5. Conclusions Our study found no evidence for health risk of exposure to NP, EE2 and BPA for Guangzhou, Wuhan and Jinan population. However, the risk of co-exposure to multiple EDCs could not be ignored. NP and BPA were found to be the main congeners with higher detection rates and concentrations in the food samples collected in three China populations. Different congeners co-occurred with each other and such characteristics existed in various food categories. Acknowledgements We thank Dr. Terrance Ye for valuable comments on our manuscript. This work was supported by the National Basic Research Program of China (973 Program) Project Number: 2012CB722401, the National Natural Science Foundation of China (Grant Nos. 81030051, 81172674, 81273060), the Specialized Research Fund for the Doctoral Program of Higher Education of China (20110142110022), and the Fund Project of Hunan Province Education Office (14C0996). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2015.05.036. References Casals-Casas, C., Desvergne, B., 2011. Endocrine disruptors: from endocrine to metabolic disruption. Annu. Rev. Physiol. 73, 135–162. Cunha, S.C., Almeida, C., Mendes, E., Fernandes, J.O., 2011. Simultaneous determination of bisphenol A and bisphenol B in beverages and powdered infant formula by dispersive liquid–liquid micro-extraction and heart-cutting multidimensional gas chromatography–mass spectrometry. Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 28, 513–526. EFSA, 2006. Opinion of the Scientific Panel on Food Additives, Flavourings, Processing Aids and Materials in Contact with Food on a Request from the Commission Related to 2,2-bis (4-hydroxyphenyl) propane (bisphenol A), 1–75. FAO/WHO, 2001. Pesticide Residues Infood-2000. Report of the Joint Meeting of the FAO Panel of Experts on Pesticide Residue in Food and the Environment and the WHO Core Assessment Group on Pesticide Residues, Rome (Italy). Plant Production and Protection Paper 163. Fayad, P.B., Prevost, M., Sauve, S., 2013. On-line solid-phase extraction coupled to liquid chromatography tandem mass spectrometry optimized for the analysis of steroid hormones in urban waste waters. Talanta 115, 349–360.
Frye, C.A., Bo, E., Calamandrei, G., Calza, L., Dessi-Fulgheri, F., Fernandez, M., 2012. Endocrine disrupters: a review of some sources, effects, and mechanisms of actions on behaviour and neuroendocrine systems. J. Neuroendocrinol. 24, 144– 159. Guenther, K., Heinke, V., Thiele, B., Kleist, E., Prast, H., Raecker, T., 2002. Endocrine disrupting nonylphenols are ubiquitous in food. Environ. Sci. Technol. 36, 1676– 1680. IPCS, 2008. Uncertainty and data quality in exposure assessment. Part 1: guidance document on characterizing and communicating uncertainty in exposure assessment. Part 2: hallmarks of data quality in chemical exposure assessment. World Health Organization, International Programme on Chemical Safety (IPCS Harmonization Project Document No. 6), WHO, Geneva. Jensen, B.H., Andersen, J.H., Petersen, A., Christensen, T., 2008. Dietary exposure assessment of Danish consumers to dithiocarbamate residues in food: a comparison of the deterministic and probabilistic approach. Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 25, 714–721. Jin, S.G., 2008. A description on the Chinese national nutrition and health survey in 2002 (report 10). People’s Medical Publishing House, Beijing (in Chinese). Jin, S., Yang, F., Liao, T., Hui, Y., Wen, S., Xu, Y., 2012. Enhanced effects by mixtures of three estrogenic compounds at environmentally relevant levels on development of Chinese rare minnow (Gobiocypris rarus). Environ. Toxicol. Pharmacol. 33, 277–283. Jin, S., Yang, F., Xu, Y., Dai, H., Liu, W., 2013. Risk assessment of xenoestrogens in a typical domestic sewage-holding lake in China. Chemosphere 93, 892–898. Liao, C., Chen, L., Kannan, K., 2013. Occurrence of parabens in foodstuffs from China and its implications for human dietary exposure. Environ. Int. 57–58, 68–74. Locatelli, M.A., Sodre, F.F., Jardim, W.F., 2011. Determination of antibiotics in Brazilian surface waters using liquid chromatography–electrospray tandem mass spectrometry. Arch. Environ. Contam. Toxicol. 60, 385–393. Lu, Y.Y., Chen, M.L., Sung, F.C., Wang, P.S., Mao, I.F., 2007. Daily intake of 4nonylphenol in Taiwanese. Environ. Int. 33, 903–910. Lu, S.Y., Chang, W.J., Sojinu, S.O., Ni, H.G., 2013. Bisphenol A in supermarket receipts and its exposure to human in Shenzhen, China. Chemosphere 92, 1190–1194. Nielsen, E., Östergaard, G., Thorup, I., Ladefoged, O., Jelnes, O., Jelnes, J.E., 2000. Toxicological Evaluation and Limit Values for Nonylphenol, Nonylphenol Ethoxylates, Tricresyl, Phosphates and Benzoic Acid. The Danish Environmental Protection Agency, Copenhagen, Denmark, pp. 1–43. Niu, Y., Zhang, J., Wu, Y., Shao, B., 2012. Analysis of bisphenol A and alkylphenols in cereals by automated on-line solid-phase extraction and liquid chromatography tandem mass spectrometry. J. Agric. Food Chem. 60, 6116–6122. Nohynek, G.J., Borgert, C.J., Dietrich, D., Rozman, K.K., 2013. Endocrine disruption: fact or urban legend? Toxicol. Lett. 223, 295–305. Pelekanou, V., Leclercq, G., 2011. Recent insights into the effect of natural and environmental estrogens on mammary development and carcinogenesis. Int. J. Dev. Biol. 55, 869–878. Schmidt, L., Müller, J., Göen, T., 2013. Simultaneous monitoring of seven phenolic metabolites of endocrine disrupting compounds (EDC) in human urine using gas chromatography with tandem mass spectrometry. Anal. Bioanal. Chem. 405, 2019–2029. Shao, B., Zhao, R., Meng, J., Xue, Y., Wu, G.H., Hu, J.Y., 2005. Simultaneous determination of residual hormonal chemicals in meat, kidney, liver tissues and milk by liquid chromatography–tandem mass spectrometry. Anal. Chim. Acta 548, 41–50. Shao, B., Han, H., Tu, X., Huang, L., 2007. Analysis of alkylphenol and bisphenol A in eggs and milk by matrix solid phase dispersion extraction and liquid chromatography with tandem mass spectrometry. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 850, 412–416. Trasande, L., Attina, T.M., Blustein, J., 2012. Association between urinary bisphenol A concentration and obesity prevalence in children and adolescents. JAMA 308, 1113–1121. Trudel, D., Tlustos, C., von Goetz, N., Scheringer, M., Reichert, P., Hungerbuhler, K., 2011. Exposure of the Irish population to PBDEs in food: consideration of parameter uncertainty and variability for risk assessment. Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 28, 1–13. Tu, P., Peng, X., 1994. China’s population: status and challenges. China Popul. Today 11, 13–15. Wang, H.X., Wang, B., Zhou, Y., Jiang, Q.W., 2013. Rapid and sensitive analysis of phthalate metabolites, bisphenol A, and endogenous steroid hormones in human urine by mixed-mode solid-phase extraction, dansylation, and ultraperformance liquid chromatography coupled with triple quadrupole mass spectrometry. Anal. Bioanal. Chem. 405, 4313–4319.
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Dietary exposure to endocrine disrupting chemicals in metropolitan population from China: A risk assessment based on probabilistic approach Dongliang He a,e, Xiaolei Ye b, Yonghua Xiao c, Nana Zhao b, Jia Long a, Piwei Zhang a, Ying Fan a, Shibin Ding a, Xin Jin a, Chong Tian a, Shunqing Xu d, Chenjiang Ying a,⇑ a Department of Nutrition and Food Hygiene, Hubei Key Laboratory of Food Nutrition and Safety, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China b School of Environmental Science and Public Health, Wenzhou Medical University, Wenzhou 325000, China c Wuhan Center for Disease Control and Prevention, Wuhan 430015, China d Ministry of Education Key Laboratory of Environment and Health, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China e Department of Preventive medicine, School of Public Health, University of South China, Hengyang 421001, China
h i g h l i g h t s Seven out of eight EDCs were detected in most raw and fresh food in China. NP, EE2, and BPA coexisted in food. Human exposure to NP, EE2 and BPA was far below TDI.
REEQs exposure in population was considerably higher than E2.
The risk of co-exposure to multiple EDCs should not be ignored.
a r t i c l e
i n f o
Article history: Received 29 March 2014 Received in revised form 16 April 2015 Accepted 14 May 2015
Keywords: Endocrine disrupting chemicals (EDCs) Dietary exposure Probabilistic modeling Risk assessment P 17b-Estradiol equivalent ( EEQs)
a b s t r a c t The intake of contaminated foods is an important exposure pathway for endocrine disrupting chemicals (EDCs). However, data on the occurrence of EDCs in foodstuffs are sporadic and the resultant risk of co-exposure is rarely concerned. In this study, 450 food samples representing 7 food categories (mainly raw and fresh food), collected from three geographic cities in China, were analyzed for eight EDCs using high performance liquid chromatography tandem mass spectrometry (HPLC–MS/MS). Besides estrone (E1), other EDCs including diethylstilbestrol (DES), nonylphenol (NP), bisphenol A (BPA), octylphenol (OP), 17b-estradiol (E2), 17a-ethinylestradiol (EE2), and estriol (E3) were ubiquitous in food. Dose-dependent relationships were found between NP and EE2 (r = 0.196, p < 0.05), BPA (r = 0.391, p < 0.05). Moreover, there existed a correspondence between EDCs congener and food category. Based on the obtained database of EDCs concentration combined with local food consumption, dietary EDCs exposure was estimated using the Monte Carlo Risk Assessment (MCRA) system. The 50th and 95th percentile exposure of any EDCs isomer were far below the tolerable daily intake (TDI) value identically. P However, the sum of 17b-estradiol equivalents ( EEQs) exposure in population was considerably larger than the value of exposure to E2, which implied the underlying resultant risk of multiple EDCs in food should be concern. In conclusion, co-exposure via food consumption should be considered rather than individual EDCs during health risk evaluation. Ó 2015 Elsevier Ltd. All rights reserved.
⇑ Corresponding author. E-mail addresses: [email protected] (D. He), [email protected] (X. Ye), [email protected] (Y. Xiao), [email protected] (N. Zhao), [email protected] (J. Long), [email protected] (P. Zhang), [email protected] (Y. Fan), [email protected] (S. Ding), [email protected] (X. Jin), tianchong [email protected] (C. Tian), [email protected] (S. Xu), [email protected] (C. Ying). http://dx.doi.org/10.1016/j.chemosphere.2015.05.036 0045-6535/Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Endocrine disrupting chemicals (EDCs), a group of substances found in the diet and environment, have characteristic activity of estrogens or androgenic activity. To date, a number of EDCs
D. He et al. / Chemosphere 139 (2015) 2–8
including nonylphenol (NP), bisphenol A (BPA), octylphenol (OP), 17b-estradiol (E2), 17a-ethinylestradiol (EE2), estrone (E1), estriol (E3), and diethylstilbestrol (DES) have been applied to manufacture various daily products widely and identified persistent in the environment worldwide (Casals-Casas and Desvergne, 2011; Nohynek et al., 2013). Toxicological studies have revealed that, to a certain extent, EDCs could mimic the effect of endogenous estrogens or interfere with estrogen signaling pathways (Pelekanou and Leclercq, 2011) and ultimately alter growth, development and reproduction of wildlife as well as human beings (Frye et al., 2012). Besides air, water and even skin contact (Lu et al., 2013), several studies revealed that consumption of food is another important source of EDCs for the non-occupational population (Trasande et al., 2012). With the increasing concern on the potential risk of EDCs to human via food intake, the European Food Safety Authority (EFSA) has established a series of tolerable daily intake (TDI) value, i.e., 50 lg/kg bw for BPA and EE2 (EFSA, 2006). As a supplementary, Danish researchers proposed a TDI value 5 lg/kg bw for NP (Nielsen et al., 2000). A few pilot studies have reported the occurrence of EDCs in food items collected in some regions, and the dose of independent exposure to them was well below TDIs. For examples, 7.50 lg/day for adult was found for BPA in Germany (Guenther et al., 2002) and around 30.0 lg/day was observed in Taiwanese (Lu et al., 2007). It is, however, insufficient to eventually eliminate the worry of health risk from co-exposure to multiple EDCs for the term 17b-estradiol equivalency factor (EEF) raised by a few recent studies (e.g., Jin et al., 2013). Since then, it is of persistent interest to precisely detect the EDCs concentrations in food and assess the risk of co-exposure to EDCs via food in consideration of the 17b-estradiol equivalent factor (EEF) (Table S1). A precise assessment of risk is usually based on two critical factors: methods of accurate detection and applicative statistic theory. Several analytical methods such as High Performance Liquid Chromatography (HPLC), and Liquid chromatography–tandem mass spectrometry (LC–MS/MS), have been developed and widely used for EDCs determination even at a trace level (Locatelli et al., 2011; Fayad et al., 2013; Wang et al., 2013). Deterministic point estimation and probabilistic modeling were the two commonly used statistical approaches for estimating exposure to environmental chemicals (Jensen et al., 2008). The deterministic approach estimates the dietary intake by multiply the mean chemical residual level and mean food consumption together and was of more understandable but reckons without the variation and uncertainty in both level and consumption data between individual (FAO/WHO, 2001). The probabilistic exposure assessment, on the other hand, takes account of the default of the former by generating distributions of the level and food consumption data and serving as model parameters inputted. The outputs based on the simulation could be specified at any percentiles etc., which make it more suitable for assessing the exposure of whole population (IPCS, 2008). Professional software like the Risk Simulation and the Monte Carlo Risk Assessment system (MCRA) are commonly used for such analyses. In the present study, we conducted a cross-sectional investigation at three Chinese cities, i.e., Wuhan (WH), Guangzhou (GZ) and Jinan (JN). Concentrations of eight selected EDCs in raw and fresh foods purchased from local markets were tested simultaneously, and analyzed using a Monte Carlo simulation to describe exposure to EDCs via food sources, based on the 17b-estradiol equivalents (EEQs) theory. Compared with the respective TDI value of chemical, the cumulative probability distribution of exposure to EDCs for the general population was discussed. To the best of our knowledge, this is the first attempt to model the exposure to EDCs from food sources in Chinese population via probabilistic approach. The result will not only reveal the contamination status of EDCs in
3
foods commonly consumed in China, but also be used for assessing the health risk of EDCs in Chinese population.
2. Materials and methods 2.1. Chemicals and reagents All EDCs standard chemicals were purchased from Sigma (St. Louis, MO, USA). b-Glucuronidase/arylsulfatase and all organic solvents (HPLC grade) including methanol, acetonitrile, acetone, hexane, and dichloromethane were purchased from Merck (Darmstadt, Germany). BPA-d1 and 13C2-E2 were obtained from CDN Isotopes Inc. (Quebec, Canada) and Dr. Ehrenstorfer (Augsburg, Germany), respectively. The ultrapure water used in this study was treated by Millipore pure water system (Millipore Co., France). Stock solutions of the standards were prepared in methanol at a concentration of 10.0 mg/L and stored at 20 °C. On the day of analysis, the stock solutions of standards were further diluted into a concentration of 10.0 lg/L and for preparation of a set of pooled calibration curves.
2.2. Sample collections The study protocol was designed for taking a representative sample in the eastern mainland China (Fig. S1). According to the geographical distribution of Chinese population, the region in the east of Hu’s line constituted 42.9% of national territory but encompassed 94.4% of Chinese population (Tu and Peng, 1994). In this region, Wuhan city (WH), Guangzhou city (GZ) and Jinan city (JN) were embraced into the follow-up studies, which are all metropolises located at different basins of large rivers (e.g., Yangtze, Pearl, and Yellow Rivers). At each site, raw and fresh food samples of different species and origins were sampled by trained assistants and stored in pre-cleaned amber glass bottles wrapped in aluminum foil. During the summer (July–September) of 2012, 7 food categories and 450 food items (Table S2), mainly raw and fresh food, were chosen to represent available varieties which were commonly consumed by the local residents on the basis of regional and national surveys on food consumption in China (Jin, 2008). The majority of food samples were purchased from local supermarkets, and a few items were purchased from large retail stores. Samples were transported to the laboratory on ice and stored at 20 °C until analysis.
2.3. Sample pretreatment The extraction for the objective compounds and clean-up was abided by procedure which had been reported elsewhere (Shao et al., 2005) and with minor modifications. Food samples were classified into two groups: solid foods and liquid foods (e.g., cooking oils). Solid foods were homogenized, weighed (5.0 g fresh weight), spiked with internal standards (BPA-d1 and 13C2-E2), and adjusted the PH value (0.2 M acetate, PH = 5.2), then incubated with 100.0 lL b-glucuronidase/arylsulfatase (37 °C, 24 h) to improve the separation of binding EDCs from food matrix (Schmidt et al., 2013). After evaporation to near dryness by nitrogen with a rotary evaporator, the extract was re-dissolved in 2.0 mL of 10% dichloromethane/hexane and purified with an OasisÒ HLB cartridge (60 mg/3 cc, Waters, USA). The final elute was concentrated to 1.0 mL, flowed through a 0.45 mL filter and stored at 4 °C until instrumental analysis. The cooking oil samples (5.0 g) were extracted using the same procedure described above, without the incubate step.
4
D. He et al. / Chemosphere 139 (2015) 2–8
2.4. Instrumental analysis A Waters UPLC-TQD system (Waters Co., USA) was used for separation and detection of target analytes. The analytical column was Acquity BEH C18 (2.1 100 mm, 1.7 lm, Waters, USA). The mobile phase was constituted of methanol and Ammonium acetate, used for LC separation at 35 °C with gradient elution. The mobile flow rate was 200.0 lL/min, and 10.0 lL of the sample extract was injected. The MS/MS was operated in an electrospray negative ionization multiple reaction monitoring (MRM) mode. The capillary voltage was held at 2.5 kV. The cone voltage and other parameters were listed in Table S3. 2.5. Quality Assurance and Quality Control (QA/QC) To avoid the potential contamination from analytical containers or reagents, no plastic was used in the experiment thoroughly, and all glassware utilized underwent even previously baked at 450 °C for 4 h. Each analysis run (50–60 samples) included full procedure blanks (n = 3), matrix spiked samples (n = 3) and at least six different concentrations of target analyte standards. The concentrations of target analytes in full procedure blanks were below the corresponding limit of detections (LODs). The average recoveries of target analytes spiked into full procedure blanks ranged from 91% to 113%, and those spiked into food matrices for solid foods (meats) ranged from 82% to 101% and for liquid foods (cooking oils) ranged from 79% to 105% (Table S4). The relative standard deviations (RSDs) for the target analytes were not higher than 10.00%. The standard calibrations were linear (r2 > 0.99) and spanned five orders of magnitude, and the results were corrected for the recoveries of the internal standards (BPA-d1 for NP, BPA, and OP, 13C2-E2 for EE2, E1, E2, E3, and DES). The LODs for NP, BPA, OP, EE2, E1, E2, E3, and DES were 0.05, 0.03, 0.05, 0.02, 0.05, 0.04, 0.04 and 0.02 ng/g, respectively. 2.6. Food consumption data The food consumption data were obtained from the nationwide dietary survey conducted in October 2002, which used a 7-day prospective food record with pre-coded questionnaire that included answering categories for the most commonly eaten foods in Chinese diet. In the present study, data of regional dietary patterns from Guangzhou, Wuhan, and Jinan were respectively cited, which were presented in the form of gram per 60 kg body weight (Jin, 2008). 2.7. Data analysis and Monte Carlo simulation For general statistical analysis, values detected below the LOQ were set to half the LOQ. Spearman’s rank correlation and correspondence analyses were conducted using the SPSS software version 12.0 (SPSS Inc., Chicago, IL, USA). A value of p < 0.05 was P denoted as statistical significance. We defined EEQs value as the co-effect of multiple EDCs congeners. The intake dose for food exposure to various EDCs was assumed to be an indicator of chronic exposure condition. The estimated dietary intake (EDI, ng/kg bw/day) for a typical adult was calculated on the basis of Eq. (1) which had been detailed elsewhere (Trudel et al., 2011) and with minor modifications:
EDI ¼
n X ðC ki ak Þ bw
ð1Þ
i¼1
where Cki (ng/g) is the concentration of congener i in the sample matrix of food group k, ak (g/day) is the amount of daily consumption of food group k, and bw is the body weight. In the present
study, since the data of food consumption cited were presented in gram per 60 kg bodyweight status (Jin, 2008), we integrated the bodyweight into consumption data and fixed it on 60 kg in calculation. Eq. (2) was used to estimate the integrated effect of EDCs:
EDIEEQs ¼
n X ðC ki f i ak Þ bw
ð2Þ
i¼1
where fi is the EEF value of EDCs congener. Individuals differ in food consumption and EDCs concentration, and the different value of Cki and ak would apparently result in various EDI values. If the mean or 95th percentile concentration of EDCs and food consumption were available for exposure calculation merely, the result derived of equations is going against the fact and reflect special exposure scenarios (50th or 95th percentile) only. Considered of the variation and uncertainty aroused from the EDCs level and individual intake of different foodstuff, the probability distribution assumptions of EDCs concentration was used to forecast the distribution of the exposure based on MCRA system (conducted by Software Crystal Ball 11.0) which is more compatible for dealing with the intrinsic EDCs level and food consumption variation or uncertainty. 3. Result 3.1. EDCs levels in food samples Most of the selected EDCs were detected in cereals but E1 was under the LOD in all samples (Table 1). NP, BPA and OP were found in all of the cereals. NP was the EDCs with the highest concentration and ranged from 5.88 to 79.08 ng/g. The concentrations were between 1.05 and 9.20 ng/g for OP, 0.96 and 29.04 ng/g for BPA. DES and EE2 were detected in the majority of cereal items (found in 45 out of 50 samples) at the levels ranged from below LOD to 0.71 and 35.04 ng/g, respectively. E2 and E3 were occasionally found in these samples. In addition, the EDCs concentrations (ng/g) in different categories of food from three cities were presented in Table S5. NP, BPA and EE2 shared common high detection rate in most of the food categories from the different cities. Similar to the results observed in cereals, detected rate of NP, OP, and BPA ranked the top three in vegetables (found in 62, 57 and 55 out of 62 samples) and the order of concentration was NP > BPA > OP. The other EDCs were also detected in some of vegetable samples and ranged from below LOD to different maximum values. Among the 51 meats and proceeded meat product samples, E2 and BPA got the highest detected rate (found in 51 out of 51 samples) and followed by NP and OP. However, the median concentration of NP (6.56 ng/g) stood first and followed by E2 (6.36 ng/g). Furthermore, both the detected rate and concentration of E2 take the leading level among similar compounds in fish and egg. The sum of EEQ of multiple EDCs in each food group was calculated based on EEF values and showed (Tables 1 and S5). In any P food group, the EEQs value was far greater than a single congener. Vegetable, cereal, meat, fish, fruit and egg ranked first to P sixth in term of median EEQs value and ranged from 1.95 to 15.82 ng/g. In addition, the concentration of EDCs also showed a matrix dependent distribution. The concentration of one congener usually increased with the increase or decrease of other EDCs level in a food category. The non-parametric Spearman correlation analysis revealed weak but positive relationships between NP and EE2 (r = 0.196, p < 0.05), NP and BPA (r = 0.391, p < 0.05), which suggested co-occurrence and similarity in food sources (Table S6). Correspondence analysis between EDCs level and food category was carried out. Two dimensions were adapted to represent main
5
D. He et al. / Chemosphere 139 (2015) 2–8 Table 1 Concentrations (ng/g fresh weight) of EDCs congeners in different food categories. OP
NP
EE2
BPA
E3
E2
P EEQs
Cereals and cereal products (n = 50) Positive samples 45 Mean 0.259 Median 0.245 Minimum
50 3.493 2.820 1.050 9.200
50 16.429 8.755 5.880 79.080
45 9.142 7.595
50 7.708 5.720 0.960 29.040
35 0.674 0.360
28 0.521 0.355
– 17.230 14.335 0.210 63.390
Vegetables (n = 62) Positive samples Mean Median Minimum Maximum
36 1.762 0.120
57 2.273 1.320
62 25.968 17.760 4.920 90.960
43 9.938 6.645
55 8.073 6.540
41 6.218 0.480
43 1.441 0.240
– 23.214 15.825 0.020 206.820
Fruits (n = 75) Positive samples Mean Median Minimum Maximum
25 0.137
51 0.322 0.240
69 4.552 4.250
57 0.689 0.770
70 3.588 2.280
36 0.443
62 0.987 0.870
– 2.525 2.580 0.240 6.020
Meat and meat products (n = 51) Positive samples 36 Mean 2.487 Median 0.190 Minimum
43 2.376 1.870
49 8.680 6.560
20 0.462
51 2.348 1.880 0.400 7.680
34 7.329 0.480
51 6.638 6.360 2.360 15.040
– 13.106 7.670 2.750 162.950
Fish and seafood (n = 99) Positive samples Mean Median Minimum Maximum
70 4.240 0.680
87 1.616 1.320
84 7.198 4.520
34 0.408
81 1.505 1.080
86 2.309 1.680
99 5.221 4.830 1.960 17.200
– 13.364 7.400 2.250 518.700
Eggs (n = 54) Positive samples Mean Median Minimum Maximum
39 0.458 0.355
36 1.439 0.380
31 0.427 0.435
0
33 0.405 0.320
35 0.319 0.295
54 1.274 1.230 0.440 2.710
– 2.091 1.955 0.440 4.720
Cooking oils (n = 59) Positive samples Mean Median Minimum Maximum
3 0.016
8 0.033
52 6.396 5.710
12 0.150
9 0.130
0
0
– 0.293
DES
LOD, limit of detection.
factors, which explained most of original variables. In the correspondence analysis plot (Fig. 1), the location of NP and oils, fruits and vegetables were close and all fell in the first quadrant. This result suggested that specific congener NP had higher contributions in oils, fruits and vegetables food categories. Similarly, we can deduce E2, E3, OP and DES were the main congeners occurred in meat, fish and egg food group for they were all in the third quadrant. BPA and EE2 were the primary contaminants in cereals rather than other congeners.
In addition, S-curve was adopted to present the cumulative freP quency and probability for exposure to EDCs and EEQs in populations (Fig. 2). The analysis of the intersections of curves and horizontal scale in the diagram suggested the exposure to seven P detected EDCs congeners and EEQs level could easily be quantified corresponding to any percentile scenarios. In the overlaid chart P (Fig. 3), the non-normal distributions of E2 and EEQs were P demonstrated. The vertical bars in distributions of EEQs, indicating the 50th forecasted values, were on the far right of E2. This result meant that the exposure to multiple EDCs in population is considerably larger than that of E2.
3.2. Human exposure to EDCs via food intake The distribution patterns of daily EDCs exposure via MCRA demonstrated the variation between individuals (Table 2). Both in mean (50th) and high (90th) scenarios, the NP exposure identically ranked first in three geographic cities, with 0.31 and 0.65 lg/kg bw/day for Guangzhou, 0.14 and 0.25 lg/kg bw/day for Wuhan, and 0.18 and 0.85 lg/kg bw/day for Jinan, respectively. P As for EEQs exposure, Guangzhou population at mean scenario but Jinan population at high scenario rated at the top of the list, at the value of 0.27 lg/kg bw/day and 0.79 lg/kg bw/day, respectively.
4. Discussion Currently, the use of EDCs in China is less subject to government regulation. In the present study, we investigated the selected EDCs in food which have been reported sporadically (Niu et al., 2012). The results demonstrated the EDCs were usually co-occurrence in foodstuffs. Based on the datasets of EDCs level and food consumption, the daily dietary exposure and their primary food sources was evaluated by probability simulation.
6
D. He et al. / Chemosphere 139 (2015) 2–8
at low and individually ineffective concentrations. The combined effects of EDCs compounds should therefore be included as well when we assess their ecological impacts on the environment (Jin et al., 2012). To date, there is no unanimously accepted parameter which has been recommended in field of risk assessment on co-exposure to P EDCs. As a pilot, EEQs value calculated based on the 17b-estradiol equivalency factor (EEF) was brought into this study for analyzing the co-exposure risk. In various food categories, P EEQs values were 10-fold higher than E2 values (Table 1), suggesting that a reasonable risk assessment should consider the co-exposure of multiple congeners. Except NP, OP and BPA were relatively high detection rates and concentrations but less contriP bution to EEQs for their low EEF value. The positive relationships P between DES and EEQs value (r = 0.887, p < 0.05) implied that DES was identified as the main contributor in all of the three matrixes for its higher EEF value. 4.2. Simulation for human exposure to EDCs Fig. 1. Diagram of corresponding analysis of EDCs and food categories. Projection of correspondence analysis showed the relationship between EDCs isomer and food category. On the basis of the statistical theory of correspondence analysis, there were corresponding relations between EDCs and food categories when their locations were close and fell in the same quadrant.
4.1. EDCs levels in Chinese food The ubiquitous presence of NP, BPA and OP in most of Chinese foodstuffs was revealed by analyzing eight EDCs compounds foodstuffs sampled from three regions of different river basins in China. The results were generally in line with the previous reports in other regions (Guenther et al., 2002; Lu et al., 2013). However, the concentrations in foodstuffs detected in this study were 5-fold higher than that noted by Guenther et al. (2002), 0.1– 19.4 ng/g wet weight. The detection rates of BPA in different food categories analyzed in our study were generally higher than those reported in previous studies (13 out of 27 for meat and 3 out of 10 for egg) (Shao et al., 2007; Cunha et al., 2011). While this may partially due to the differences in collection and number of food samples analyzed, the use of BPA in materials intended to come in contact with foodstuffs is increasing in recent years. Correspondence analysis was rarely used for the similar studies in the literature. In this study (Fig. 1), two representative dimensions explained the most original variables which suggested that the primary EDCs contamination in various food categories were specific. The non-parametric Spearman correlation analysis for the relative abundance of EDCs revealed positive relationships between various congeners, suggesting that people always co-expose to multiple EDCs via food consumption simultaneously. The evidence obtained indicates the potential and resultant health hazards of mixtures of EDCs even when each EDCs congener occurs
By using Monte Carlo method, we simulated the detected EDCs exposure on different scenarios synchronously and found that the forecasted median (50thpercentile), even high (95th percentile) exposures for NP, EE2 and BPA (with TDI value available) were far below the corresponding TDI values. The NP intake dose was lower than the value reported in a Taiwanese population (i.e., 38.59 ± 29.32 ng/day) (Lu et al., 2007) and the intake dose of BPA was far below the value estimated by another study in a Chinese population (489 ng/kg bw/day) (Liao et al., 2013). Such inconsistency may be partially contributed by the difference between the deterministic and the probabilistic approaches (Jensen et al., 2008). As described in our result (Table 1), the median values were generally lower than the mean values which suggested a non-normal distribution of EDCs. In fact, the majority of EDCs distributions analyzed by Monte Carlo simulation were identified as log-normal distribution and the representativeness of mean values should be questioned. If an inappropriate mean value (not normal distribution) was inputted into the equation for calculating the daily intake dose, the output value would be bias inevitably. Our results indicated that the co-exposure to multiple EDCs was considerably larger than the exposure to E2 independently (Fig. 3). It should be considered as an alert of the default in assessment on health risk of exposure to EDCs isomer and regardless of its combined effect. 4.3. Limitation Firstly, although the consumption data used in this study are the most recent data (released in 2008), they were collected ten years ago. Over the last ten years, consumption patterns have changed very likely due to the presence of new food products on
Table 2 Distribution of exposure (lg/kg bw/day) to various EDC congeners for Guangzhou, Wuhan and Jinan population. TDI value
DES OP NP EE2 BPA E3 E2 EEQs
NA NA 5.0 50.0 50.0 NA NA NA
Guangzhou population
Jinan population
Wuhan population
Distribution
P50
P90
Distribution
P50
P90
Distribution
P50
P90
Log Log Log Log Log G Log M.E
0.01 0.05 0.31 0.12 0.14 0.07 0.04 0.27
0.03 0.12 0.65 0.20 0.23 0.12 0.13 0.43
Log Log Log Log Log Log Log Log
0.00 0.02 0.14 0.11 0.04 0.01 0.02 0.16
0.02 0.03 0.25 0.23 0.12 0.20 0.05 0.79
Log Log Log Log Log Log Log Log
0.01 0.04 0.18 0.05 0.08 0.01 0.02 0.14
0.07 0.08 0.85 0.32 0.17 0.03 0.04 0.58
NA, no available; Log, lognormal distribution; G, gamma distribution; M.E, max extreme distribution.
D. He et al. / Chemosphere 139 (2015) 2–8
Fig. 2. The cumulative probabilities of EDCs congeners and population, respectively.
7
P EEQs exposure in population. A, B and C represented exposure (lg/kg bw/day) for Guangzhou, Wuhan and Jinan
P P Fig. 3. The distributions of EEQs and E2 exposure in population. The overlaid distributions of A, B and C described EEQs and E2 exposure for Guangzhou, Wuhan and Jinan population, respectively. The vertical bars represented the medians of the forecasted values.
8
D. He et al. / Chemosphere 139 (2015) 2–8
the market. Moreover, the EDCs concentrations were not nationwide monitoring data including all food categories but measured in 450 foodstuffs bought on the market very recently and the representativeness can be further improved. The deficiency of full scale in sampling can give rise to under-estimation in exposure. P Secondly, EEQs was brought into consideration of co-effect of exposure to multiple EDCs simultaneously; however, no available TDI value was authoritatively released. The significant difference P between E2 and EEQs can only suggest that even far below regulation TDI value of any single isomer is not enough to identify the exposure as harmless. 5. Conclusions Our study found no evidence for health risk of exposure to NP, EE2 and BPA for Guangzhou, Wuhan and Jinan population. However, the risk of co-exposure to multiple EDCs could not be ignored. NP and BPA were found to be the main congeners with higher detection rates and concentrations in the food samples collected in three China populations. Different congeners co-occurred with each other and such characteristics existed in various food categories. Acknowledgements We thank Dr. Terrance Ye for valuable comments on our manuscript. This work was supported by the National Basic Research Program of China (973 Program) Project Number: 2012CB722401, the National Natural Science Foundation of China (Grant Nos. 81030051, 81172674, 81273060), the Specialized Research Fund for the Doctoral Program of Higher Education of China (20110142110022), and the Fund Project of Hunan Province Education Office (14C0996). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2015.05.036. References Casals-Casas, C., Desvergne, B., 2011. Endocrine disruptors: from endocrine to metabolic disruption. Annu. Rev. Physiol. 73, 135–162. Cunha, S.C., Almeida, C., Mendes, E., Fernandes, J.O., 2011. Simultaneous determination of bisphenol A and bisphenol B in beverages and powdered infant formula by dispersive liquid–liquid micro-extraction and heart-cutting multidimensional gas chromatography–mass spectrometry. Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 28, 513–526. EFSA, 2006. Opinion of the Scientific Panel on Food Additives, Flavourings, Processing Aids and Materials in Contact with Food on a Request from the Commission Related to 2,2-bis (4-hydroxyphenyl) propane (bisphenol A), 1–75. FAO/WHO, 2001. Pesticide Residues Infood-2000. Report of the Joint Meeting of the FAO Panel of Experts on Pesticide Residue in Food and the Environment and the WHO Core Assessment Group on Pesticide Residues, Rome (Italy). Plant Production and Protection Paper 163. Fayad, P.B., Prevost, M., Sauve, S., 2013. On-line solid-phase extraction coupled to liquid chromatography tandem mass spectrometry optimized for the analysis of steroid hormones in urban waste waters. Talanta 115, 349–360.
Frye, C.A., Bo, E., Calamandrei, G., Calza, L., Dessi-Fulgheri, F., Fernandez, M., 2012. Endocrine disrupters: a review of some sources, effects, and mechanisms of actions on behaviour and neuroendocrine systems. J. Neuroendocrinol. 24, 144– 159. Guenther, K., Heinke, V., Thiele, B., Kleist, E., Prast, H., Raecker, T., 2002. Endocrine disrupting nonylphenols are ubiquitous in food. Environ. Sci. Technol. 36, 1676– 1680. IPCS, 2008. Uncertainty and data quality in exposure assessment. Part 1: guidance document on characterizing and communicating uncertainty in exposure assessment. Part 2: hallmarks of data quality in chemical exposure assessment. World Health Organization, International Programme on Chemical Safety (IPCS Harmonization Project Document No. 6), WHO, Geneva. Jensen, B.H., Andersen, J.H., Petersen, A., Christensen, T., 2008. Dietary exposure assessment of Danish consumers to dithiocarbamate residues in food: a comparison of the deterministic and probabilistic approach. Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 25, 714–721. Jin, S.G., 2008. A description on the Chinese national nutrition and health survey in 2002 (report 10). People’s Medical Publishing House, Beijing (in Chinese). Jin, S., Yang, F., Liao, T., Hui, Y., Wen, S., Xu, Y., 2012. Enhanced effects by mixtures of three estrogenic compounds at environmentally relevant levels on development of Chinese rare minnow (Gobiocypris rarus). Environ. Toxicol. Pharmacol. 33, 277–283. Jin, S., Yang, F., Xu, Y., Dai, H., Liu, W., 2013. Risk assessment of xenoestrogens in a typical domestic sewage-holding lake in China. Chemosphere 93, 892–898. Liao, C., Chen, L., Kannan, K., 2013. Occurrence of parabens in foodstuffs from China and its implications for human dietary exposure. Environ. Int. 57–58, 68–74. Locatelli, M.A., Sodre, F.F., Jardim, W.F., 2011. Determination of antibiotics in Brazilian surface waters using liquid chromatography–electrospray tandem mass spectrometry. Arch. Environ. Contam. Toxicol. 60, 385–393. Lu, Y.Y., Chen, M.L., Sung, F.C., Wang, P.S., Mao, I.F., 2007. Daily intake of 4nonylphenol in Taiwanese. Environ. Int. 33, 903–910. Lu, S.Y., Chang, W.J., Sojinu, S.O., Ni, H.G., 2013. Bisphenol A in supermarket receipts and its exposure to human in Shenzhen, China. Chemosphere 92, 1190–1194. Nielsen, E., Östergaard, G., Thorup, I., Ladefoged, O., Jelnes, O., Jelnes, J.E., 2000. Toxicological Evaluation and Limit Values for Nonylphenol, Nonylphenol Ethoxylates, Tricresyl, Phosphates and Benzoic Acid. The Danish Environmental Protection Agency, Copenhagen, Denmark, pp. 1–43. Niu, Y., Zhang, J., Wu, Y., Shao, B., 2012. Analysis of bisphenol A and alkylphenols in cereals by automated on-line solid-phase extraction and liquid chromatography tandem mass spectrometry. J. Agric. Food Chem. 60, 6116–6122. Nohynek, G.J., Borgert, C.J., Dietrich, D., Rozman, K.K., 2013. Endocrine disruption: fact or urban legend? Toxicol. Lett. 223, 295–305. Pelekanou, V., Leclercq, G., 2011. Recent insights into the effect of natural and environmental estrogens on mammary development and carcinogenesis. Int. J. Dev. Biol. 55, 869–878. Schmidt, L., Müller, J., Göen, T., 2013. Simultaneous monitoring of seven phenolic metabolites of endocrine disrupting compounds (EDC) in human urine using gas chromatography with tandem mass spectrometry. Anal. Bioanal. Chem. 405, 2019–2029. Shao, B., Zhao, R., Meng, J., Xue, Y., Wu, G.H., Hu, J.Y., 2005. Simultaneous determination of residual hormonal chemicals in meat, kidney, liver tissues and milk by liquid chromatography–tandem mass spectrometry. Anal. Chim. Acta 548, 41–50. Shao, B., Han, H., Tu, X., Huang, L., 2007. Analysis of alkylphenol and bisphenol A in eggs and milk by matrix solid phase dispersion extraction and liquid chromatography with tandem mass spectrometry. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 850, 412–416. Trasande, L., Attina, T.M., Blustein, J., 2012. Association between urinary bisphenol A concentration and obesity prevalence in children and adolescents. JAMA 308, 1113–1121. Trudel, D., Tlustos, C., von Goetz, N., Scheringer, M., Reichert, P., Hungerbuhler, K., 2011. Exposure of the Irish population to PBDEs in food: consideration of parameter uncertainty and variability for risk assessment. Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 28, 1–13. Tu, P., Peng, X., 1994. China’s population: status and challenges. China Popul. Today 11, 13–15. Wang, H.X., Wang, B., Zhou, Y., Jiang, Q.W., 2013. Rapid and sensitive analysis of phthalate metabolites, bisphenol A, and endogenous steroid hormones in human urine by mixed-mode solid-phase extraction, dansylation, and ultraperformance liquid chromatography coupled with triple quadrupole mass spectrometry. Anal. Bioanal. Chem. 405, 4313–4319.