Exposure of metals and PAH through local foods and risk of cancer in a historically contaminated glasswork area

Contents lists available atScienceDirect

Environment International

Exposure of metals and PAH through local foods and risk of cancer in a

historically contaminated glassworks area

Ingela Helmfrid

, Stefan Ljunggren

, Reza Nosratabadi

, Anna Augustsson

, Monika Filipsson

,

Mats Fredrikson

, Helen Karlsson

, Marika Berglund

a_{Occupational and Environmental Medicine Center, and Department of Clinical and Experimental Medicine, Linköping University, Linköping, Sweden} b_{Department of Biology and Environmental Sciences, Linnaeus University, Kalmar, Sweden}

c_{Department of Clinical and Experimental Medicine, Linköping University, Linköping, Sweden} d_{Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden}

A R T I C L E I N F O

Handling Editor: Hui Li Keywords:

Exposure Metals PAH

Consumption of local food Cancer

Contaminated land

A B S T R A C T

Background: Production of crystal glass and colored art glassware have been going on in the south-eastern part of Sweden since the 1700s, at over 100 glassworks and smaller glass blowing facilities, resulting in environ-mental contamination with mainly arsenic (As), cadmium (Cd), lead (Pb) and polycyclic hydrocarbons (PAH). High levels of metals have been found in soil, and moderately elevated levels in vegetables, mushrooms and berries collected around the glassworks sites compared with reference areas. Food in general, is the major exposure source to metals, such as Cd and Pb, and PAHs. Exposure to these toxic metals and PAH has been associated with a variety of adverse health effects in humans including cancer.

Objective: The aim of the present study was to evaluate the occurrence of cancer in a cohort from the con-taminated glasswork area in relation to long-term dietary intake of locally produced foods, while taking into account residential, occupational and life styles factors.

Methods: The study population was extracted from a population cohort of 34,266 individuals who, at some time between the years 1979–2004, lived within a 2 km radius of a glassworks or glass landfill. Register information on cancer incidence and questionnaire information on consumption of local foods (reflecting 30 years general eating habits), life-time residence in the area, life style factors and occupational exposure was collected. Furthermore, blood (n = 660) and urine (n = 400) samples were collected in a subsample of the population to explore associations between local food consumption frequencies, biomarker concentrations in blood (Cd, Pb, As) and urine (PAH metabolite 1-OHPy) as well as environmental and lifestyle factors. The concurrent exposure to persistent organic pollutants (POPs) from food was also considered. A case-control study was performed for evaluation of associations between intakes of local food and risk of cancer.

Results: Despite high environmental levels of Cd, Pb and As at glasswork sites and landfills, current metal ex-posure in the population living in the surrounding areas was similar or only moderately higher in our study population compared to the general population. Reported high consumption of certain local foods was associated with higher Cd and Pb, but not As, concentrations in blood, and 1-OHPy in urine. An increased risk of cancer was associated with smoking, family history of cancer, obesity, and residence in glasswork area before age 5 years. Also, a long-term high consumption of local foods (reflecting 30 years general eating habits), i.e. fish and meat (game, chicken, lamb), was associated with increased risk of various cancer forms.

Conclusions: The associations between consumption of local food and different types of cancer may reflect a higher contaminant exposure in the past, and thus, if consumption of local food contributes to the risk of ac-quiring cancer, that contribution is probably lower today than before. Furthermore, it cannot be ruled out that other contaminants in the food contribute to the increased cancer risks observed.

https://doi.org/10.1016/j.envint.2019.104985

Received 21 December 2018; Received in revised form 20 June 2019; Accepted 29 June 2019 ⁎_{Corresponding author.}

E-mail address:ingela.helmfrid@regionostergotland.se(I. Helmfrid).

Available online 15 July 2019

0160-4120/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

1. Introduction

Public health issues related to soil contamination are receiving in-creasing attention, where a complex interplay between different po-tential exposure pathways contributes to the total exposure. According to the European Environment Agency, there are about 2.5 million po-tentially contaminated sites within the EU member states (Van Liedekerke et al., 2014).

One example of industrial activity where the local soil environment has become severely contaminated is the historical production of glass in south-eastern Sweden. Artistic and crystal glass have been manu-factured here since the 1700s, at over 100 glassworks and smaller glass blowing facilities (Nordström, 1999), resulting in environmental con-tamination with mainly arsenic (As), cadmium (Cd) and lead (Pb), but also with polycyclic aromatic hydrocarbons (PAHs) which are emitted from the glasswork furnaces (Höglund et al., 2007;SEPA, 2009). Many of these contaminated sites have been classified to be of high risk to humans and the environment (SEPA, 2009).

Exposure to toxic metals and PAHs has been associated with a variety of adverse health effects in humans, including cancer (Akesson et al., 2014;Alicandro et al., 2016;Arisawa et al., 2001;Calderon et al., 2003;Dreij et al., 2005;Hellstrom et al., 2007;IARC, 2010;Julin et al., 2012; Lagerqvist et al., 2011; Steenland and Boffetta, 2000;Thomas et al., 2014). The International Agency for Research on Cancer has found evidence from a number of published papers that occupational exposure from glass manufacturing increases the risk of cancer (IARC, 1993). Also residence near glass manufacturing sites has been asso-ciated with increased risk of cancer (Nyqvist et al., 2017;Wingren and Axelson, 1992). In a former study from the Swedish glassworks region an increased risk of brain cancer mortality was found among people living in parishes with glassworks as compared to the Swedish popu-lation in general (Wingren and Axelson, 1992). In a recent register study within the same area, significantly elevated cancer incidences were observed for total malignant cancers, cancers in the digestive system, and for cancer in the prostate, lymphatic and hematopoietic system (Nyqvist et al., 2017).

In the general population, diet is usually the major exposure route for most metals, even though ingestion and inhalation of soil/dust can be relevant contributors around areas with elevated metal concentra-tions in the soil (Cao et al., 2016;EFSA, 2009a;EFSA, 2010;Glorennec et al., 2016;Parveen et al., 2018). In this context, it has been shown that there is a positive correlation between soil metal concentration and metal concentration in homegrown vegetables from glassworks villages, and that consumption of these vegetables may result in elevated ex-posure levels (Augustsson et al., 2018;Augustsson et al., 2015; Uddh-Soderberg et al., 2015). The major exposure routes for PAHs are food and smoking (EFSA, 2008). Food can be contaminated from environ-mental sources, and from certain home cooking practices, e.g. barbecue (EFSA, 2008).

The diffuse and widespread contamination in the area may sig-nificantly affect the contaminant exposure for local residents through the ingestion of local foodstuff. Most residents in the glassworks villages live in private houses and a survey that was carried out in the region in 2012 revealed that many residents cultivate their own vegetables (Filipsson et al., 2013). The Swedish glassworks region is also a forested area, where wild berries and mushrooms are picked and consumed. In addition, groundwater sample measurements at the glassworks site indicated elevated mean concentrations of As and Pb at most glass-works (Höglund et al., 2007;Nyqvist et al., 2017). It can thus be hy-pothesized that consumption of local food may contribute to elevated body burden of metals and PAHs and the increased cancer incidences observed in these historically contaminated areas.

Humans are also exposed to persistent organic pollutants (POPs), mainly from fatty food (Darnerud et al., 2006). Since these pollutants have been associated with a range of health effects, including cancer (Ali et al., 2016; Arrebola et al., 2016; Henriquez-Hernandez et al.,

2017;IARC, 2018), we explored the possible concurrent exposure to a number of these compounds.

The aim of the present study was to evaluate the occurrence of cancer in a population extracted from a cohort from the Swedish glassworks area in relation to long-term dietary intake of locally pro-duced foods, taking into account residential, occupational and life styles factors. Furthermore, we explored associations between reported local food consumption frequencies, biomarker concentrations and environmental as well as life styles factors.

2. Material and methods 2.1. Study area

In 2007, an inventory of soil metal concentrations in the glassworks study area showed high mean concentrations of As, Pb and Cd in sur-face soil at the glassworks sites (Höglund et al., 2007). At glass waste dumps, the surface soil metal concentrations were even higher (Höglund et al., 2007;Nyqvist et al., 2017). Glass waste materials have also been used as backfill material in for example gardens and common land around the glassworks, and elevated concentrations are thus also frequently found in private garden soils and common land (Augustsson et al., 2018; Augustsson et al., 2015; Uddh-Soderberg et al., 2015). Historically, extensive emissions of dust particles and volatile com-pounds, including metals and PAHs, associated with glass production have also occurred and contributed further to the general contamina-tion situacontamina-tion around the glassworks (Larsson et al., 1999).

2.2. Study population

The study population was extracted from a cohort with 34,266 in-dividuals who, at some time between the years 1979–2004 lived within a 2 km radius of a glassworks emission source (in total 12 glassworks and 2 glass landfills) (Supplemental material, Fig. 1) (Nyqvist et al., 2017). The cohort was matched with the National Cancer and Mortality Register (National Board of Health and Welfare) for the period 1979–2012. Details on emigration during the same period and data on who were still alive at the end of 2012 were obtained from Statistics Sweden.

All identified cancer cases above 18 years (n = 1200) in the cohort and a random sample of cancer-free control subjects (n = 7000, to compensate for an expected low response rate), living in the same area during the same time period as the cancer cases diagnosed, were se-lected. Current addresses and live status of the case and control in-dividuals were obtained from The Swedish Tax Agency. After checking, a total of 7939 individuals remained and who received an invitation/ information letter with unique login information and a link to an online questionnaire during 2014. Within two weeks a paper questionnaire was sent to the non-responders. A second reminder was sent out three weeks later.

The study was approved by the Regional Ethical Committee at Linköping University, Sweden.

2.3. Questionnaires

Information was collected on life-time residences (addresses and periods), life-time occupations (job title, assignment and time-period), smoking habits (active, former, sometimes or never smoker), number of child births, anthropometric data, and own, parental and siblings medical history, and information on general eating habits during the past 30 years of local as well as non-local food items (fre-quencies and type of fish, meat, vegetables, fruit, garden and forest berries, mushrooms) and drinking water source, e.g. private well or public water. Three consumptions groups were constructed (high, in-termediate and non/low,) based on the food frequency information. High consumption of local vegetables, fruit and berries was defined

as > 1 time/week, high consumption of local meat andfish was defined as > 1 time/month. Intermediate consumption was defined as 3 times/ year to 4 times/month of local vegetables or 3–11 times/year of local meat orfish. Non/low consumption of local food was defined as never or seldom consumed. Potential occupational exposure to the metals of interest was evaluated based on job title information, being a glass-worker or other metal glass-worker.

The response rate for the questionnaire study was 36% among cancer cases and 27% among controls. About 70% of the women and 74% of the men did not respond. The mean age of responders (57 yrs.; range 19–98 yrs) and non-responders (51 yrs.; range 19–100 yrs), re-spectively, was significantly different (p = 0.005). About 9% returned an empty questionnaire or notified via email or phone that they did not want to participate in the study. The major reasons for not participating were residing in the glasswork area few years, long time ago, sickness and old age.

2.4. Biological sample collection

The participants were invited, through the questionnaire, to donate blood andfirst void morning urine samples.

About 1100 participants agreed to donate samples. At the time of sampling, the participants who lived within approximately 50 km geographical distance from the study area were contacted by a nurse. During September 2014 to April 2016, biological samples were col-lected from 656 participants. Information regarding the biological sampling and informed consent was mailed to the participants prior the sampling occasion. The weight and height were controlled before the sampling. Additional data regarding physical activity, time point and what kind of last intake of food, as well as the recent intake offish, were collected via a questionnaire at the time of sample collection.

Blood was collected in two Sodium Heparin vacutainer tubes (4 mL, Sodium-Heparin, Vacuette®, Greiner). Samples were kept frozen in a portable freezer bag (−18 °C) for transport to the laboratory and stored in a freezer (−20 °C or −80 °C) prior to the analyses. First void-morning urine samples were collected in specially prepared urine home samples kits (urine cups, 12 cL, Papyrus Supplies) and two urine tubes (Falcon tubes, 50 mL, Sarstedt®) and handed over to the nurse at the time of blood sampling The urine cups and the urine tubes were tested free from metal contamination.

2.5. Biological sample analyzes

Analyses of biomarkers in blood and urine were performed at the Department of Occupational and Environmental Medicine, Linköping University.

The metal elements 208Pb (the sum of three isotopes at 206, 207 and 208), 111Cd and 75As were analyzed in whole blood according to Barany et al. (1997). All samples were analyzed with a standard ICP-MS benchtop instrument HP 4500 series 100 (Agilent Technologies Inc.) operated with Pt- tip and skimmer cones, Babington nebulizer and Scott-type spray chamber,fitted with a Cetac ASX500 auto sampler. The instrument settings and parameters: RF power set on 1100 W, Argon gasflow rate at 14 L/min for plasma and 1.1 L/min for carrier gas. The uptake speed was 0.3 rps, uptake time 40 s and stabilization time 30 s. The S/C temperature was 2 °C, 3 points per peak and 30 ms dwell time. The conditions of the instrument were optimized using the tuning solution before analyses. Internal standards (Sc, Rh, In and Bi were 1 ppm) and samples were introduced with a peristaltic pump. All tube and container batches were tested for contamination before sam-pling. All chemicals were of laboratory grade (Sigma-Aldrich St Louis, MO, USA).

Urinary levels of the PAH metabolite 1-hydroxypyren (1-OHPy) were detected using a method previously described byCarmella et al. (2004)) with slight modification. Briefly, urine samples were incubated overnight withβ -glucuronidase at 37 °C to deconjugate eventual me-tabolites. 1-OHPy was extracted using hexane and dried. Reconstituted samples in acetonitrile were separated by HPLC and 1-OHPy were de-tected byfluorescence with excitation/emission at 344/400 nm. Levels of 1-OHPy were quantified against a standard between 10 and 1000 pg/ mL. The levels of 1-OHPy were normalized against urinary creatinine. Urine creatinine was determined at the Clinical Chemistry laboratory, Linkoping University Hospital.

To explore potential health effects due to concurrent exposure of persistent organic pollutants (POPs), the concentrations of 16 poly-chlorinated biphenyls (PCBs), 5 organochlorine pesticides (OCPs), oc-tachlorinated dibenzo-p-dioxin (OCDD) and one polybrominated di-phenylether (BDE 47) were measured in plasma with a previously established method (Stubleski et al., 2018) in a 96-well high throughput system. The analyses were carried out at Man-Technology-Environment (MTM) Research Center, Örebro University, Örebro. POP levels were lipid adjusted against the total amount of plasma lipids calculated from Loading scatterplot (p1 vs. p2) Age Men Women Low education High education Metalwork -1,0 -0,8 -0,6 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 p1 -1,0 -0,8 -0,6 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 2 p Residence time B_Cd B_As B_Pb Non-smoker Former smoker Active smoker No metal/glasswork Glasswork

Non cons meat Intermed cons meat

High cons meat

Non fish cons late 3 days Fish cons late 3 days

Non cons fish

Intermed cons fish High cons fish

Non cons veg Intermed cons veg

High cons veg

Fig. 1. Loading plot displaying the PCA-analysis of re-lationship between the variables As, Pb and Cd concentra-tions in blood (μg/L) (n = 658) and the variables age, gender, smoking habit, (active, former, non-smoker), glasswork, metalwork, residence time in a glasswork area, fish consumption during the last three days and consump-tion of local vegetables, meat andfish. The x- and y-axes represent the PC1 (Principal Component 1) and PC2 scores, respectively. Thefirst two PC accounts for as much possible of the variation in the original data, and thefirst component axis (PC1) will stretch out in the direction where there is most variance of variable space and are clustered.

the measured plasma cholesterol and triglycerides using an established formula (Rylander et al., 2006). Only measures with a recovery 20% < x < 150% and above the limit of detection were used in the statistical analyses.

2.5.1. Analytical quality control

The limit of detection (LOD) of the different metals was calculated based on three standard deviations (3∗ SD) of the mean of the Chemical blank solution that was analyzed 10 times. LOD for the different metals (μg/L) in blood samples were: As (0.27), Cd (0.02), Pb (0.06). The lowest values for As should be interpreted with caution since the in-strument was not optimized for such low values. The LOD for the PAH metabolite 1-OHPy in urine samples was determined based on three standard deviations (3∗ SD) of the mean of the noise in the standards with the lowest concentrations. This resulted in a LOD of 0.33 pg/mL. LOD of POPs was calculated based on three standard deviations (3∗ SD) of the mean of the chemicals blank solution that was analyzed 24 times. Overall, levels of POPs were low, including PCBs (Supplemental materialTable 5). LOD for PCB 118 was 30.2 pg/mL and for PCB153 38.3 pg/mL.

Seronorm™ Certified Reference Materials, Seronorm Trace Elements Blood levels L-1 and L-2 (low and high levels, respectively) were ob-tained from Sero (Billingstad, Norway) and were used for analytical quality control. The Reference Materials were reconstituted according to the recommendation of the supplier by adding pure water and were analyzed between every tenth samples during the blood analytical runs. For external quality control the laboratory was certified through par-ticipating in an inter-laboratory quality program INSTAND e.V. (Düsseldorf, Germany) for blood every 6 month.

The analytical results of 1-OHPy (pg/mL) in urine were compared to reference sample with 1100 pg/mL spiked concentration (Supplemental material,Table 1b).

The results of the analytical quality control are shown in Supplemental material (Tables 1a,b,c). The analytical results were sa-tisfactory.

2.6. Statistics

Statistical analyses were performed with STATISTICA 64 version 13 (Statsoft©, Oklahoma USA). A case-control study was performed in the study population who responded to the questionnaire (questionnaire study), as well as in a subsample of the study population who also donated biological samples (biomonitoring study). Differences between groups within respective study were analyzed. Cross tabulation Pearson Chi-square test and One-way ANOVA were used to assess differences between categorical variables (gender, academic education, occupa-tion, residence in glasswork area before age 5 years, smoking status, consumption groups) and continuous variables (age, residence time), respectively.

In the case-control analyses, only subjects with a requirement of at least 5 years of residence time in a glasswork area (cases and controls) and a latency time requirement of at least 10 years before cancer di-agnosis (cases) or inclusion in the study (controls), were included in the statistical analyses.

Logistic regression analyses were performed with one analysis for all cancer cases vs. controls as well as for cases of specific cancer forms. Multivariable logistic regression was used to study associations between consumption frequencies of local food and adjusted for the covariates age, gender, and family (parents and siblings) history of cancer, edu-cation level, obesity and residence in a glasswork area the first five years of life (Tables 2aand2b). The criterion for covariates to be en-tered in the multivariate-adjusted model was based on the logistic re-gressions analyses and a p≤ 0.20. Only associations including at least 5 cases and 18 controls are reported. The odds ratio (OR) is used as a measure of relative risk. A p-value < 0.05 is considered to be sig-nificant.

The concentrations of the toxic metals Cd, Pb, and As in blood and PAH metabolite in urine (biomonitoring study) were evaluated in relation to consumption frequencies of local food and risk factors for cancer. The same procedure was used to explore the levels of PCB 118 (similar structure to dioxins) and PCB 153 (marker for total PCB) in plasma. Multivariate analysis NIPALS/PCA (Nonlinear Iterative Partial Least Squares/Principal Component Analyses) was used to identify a struc-ture of the dataset and assess co-variation of measured concentrations of biomarkers and reported consumption frequencies of local food and other selected covariates (work at glasswork, metal work, education, age, smoking habits, residence time in a glasswork area,fish intake the last three days). Factors with eigenvalues > 1 (Kaisers method) where included in the PCA analysis and a Scree plot (Cattels criteria) was used to graphically determine the optimal numbers of factors to retain.

Main effect ANOVA, followed by post hoc test Bonferroni, was used to evaluate differences in concentrations of metals (Cd, Pb, As) in blood and 1-OHPy in urine, respectively, in relation to reported food con-sumption frequencies/groups of locally produced foods. Variables to be included in the multiple ANOVA-models were selected based on the PCA-analyses. The statistical significance level was set at p < 0.05. 3. Results

3.1. Basic characteristics of the study population

In total, 2198 individuals (1005 men, 1193 women) answered the questionnaire. After applying the restriction criteria set for the case-control study (see Statistics section) a total of 1846 individuals (361 cases, 1485 controls) were included in the case- control analysis.

In total, 656 individuals (132 cases, 524 controls) participated in the biomonitoring study. When the same restriction criteria set for the case-control study were applied, 127 cases and 479 controls remained for the case control analysis. The other 50 participants had been living outside the 2 km radius from an emission source (glassworks or glass landfill) but within 50 km from glassworks at the time of blood sam-pling.

Basic characteristics of the participants are presented inTable 1. In the questionnaire study, the cancer cases were significantly older, had longer residing time in a glasswork area, and included a larger pro-portion of former smokers and of long-term high consumers of locally caughtfish. A larger proportion of the controls had been residing in a glasswork area thefirst five years in life and had academic education. In the biomonitoring study, the cancer cases were significantly older, had longer residing time in a glasswork area, and a larger proportion of the controls had academic education.

3.2. Cancer risk in relation to potential risk factors

A significantly increased risk (odds ratio) for total cancer in the investigated glassworks area was associated with ever smoking, cancer diagnosis in the family (parents or siblings), obesity, and residence time in the glasswork area before age 5 years (Table 2a).

When stratifying by gender, residing in the glasswork area during thefirst five years was associated with significantly increased risk in women only, and obesity was significantly associated with increased risk in men (Table 2b). Family history of cancer was associated with increased risk for total cancer in both men and women.

Intake of drinking water from a private (drilled or dug) well in a contaminated area is a potential risk factor for increased contaminant exposure. Overall, no increased cancer risk from having a private well could be detected. No statistical analysis could be performed for the variable having a private well thefirst five years in life, due to few cancer case (n = 4) and controls (n = 15).

3.3. Exploratory data analysis

Associations between consumption of local food groups, biomarker concentrations in blood and urine and risk factors for cancer were evaluated by PCA for exploratory data analysis (Figs. 1 and 2). High consumption of all three local food groups, as well as age, active and former smoking and (total) residence time, were associated with the metals (Fig. 1). Four principal factors were retained based on eigen-value > 1 and a scree plot; accounting for 42% of the total variance. Fig. 1 shows the first two principal components which can explain about 26% of the total variance.

High consumption of vegetables and active smoking were associated with the PAH metabolite in urine (Fig. 2). Four principal factors were retained based on eigenvalue > 1 and a scree plot; accounting for 54% of the total variance.Fig. 2shows thefirst two principal components which can explain about 34% of the total variance.

High consumption of meat andfish was also associated with PCB 118 and 153 in plasma (not shown), however levels were low. 3.3.1. Association between consumption of local food and risk of cancer

It was hypothesized that living in a contaminated area and con-suming locally produced foods will increase the risk of cancer. Bivariate Table 1

Basic characteristics of the study population. A restriction criterion of at leastfive years of residence time in the contaminated area, within 2 km from glass work (controls) and a latency time of at least 10 years betweenfirst residence year and time of diagnosis (cases) were applied. In the subsample taking part in the biomonitoring study, study participants who at the time of sampling lived within a geographical distance of 50 km were included. Significant differences (p < 0.05) in bold.

Questionnaire study population N = 1846 Biomonitoring study (a subpopulation of the questionnaire study population) N = 656

All Cases N = 361

Controls N = 1485

Cases N = 127 Controls N = 479 Living within 50 kma

N = 50 Gender Men, Number (%) 863 (47) 174 (48) 689 (46) 62 (49) 220 (46) 25 (50)

Women, Number (%)

983 (53) 187 (52) 796 (54) 65 (51) 260 (54) 25 (50) Age (Year) Mean (SD) 59 (15) 69 (11) 56 (15) 70 (9.6) 58 (13) 54 (18)

Median 60 70 56 70 59 57

Range 19–98 19–98 19–97 33–89 29–92 20–92 Residence time in the glasswork area

(years)

Mean (SD) 36 (20) 41 (22) 35 (19) 47 (18) 40 (18) 26 (25)

Median 35 42 33 47 40 17

Range 5–98 5–98 5–91 6–88 5–91 1–80

Residence in glasswork area before age of 5 yrs

Number (%) 929 (50) 157 (43) 772 (52) 56 (44) 266 (55) 23 (46) Academic education Number (%) 665 (36) 97 (27) 568 (38) 35 (28) 153 (32) 18 (36) Working at glasswork Number (%) 162 (9.9) 37 (11) 125 (9.5) 12 (11) 48 (12) 5 (10) Other metal work Number (%) 204 (12) 32 (9.8) 172 (13) 14 (12) 74 (17) 8 (16) Smoking status: number (%) Active smoker 80 (4.4) 14 (3.9) 66 (4.4) 3 (2.4) 20 (4.2) 4 (8.0) Former smoker 789 (43) 181 (51) 608 (41) 62 (49) 209 (44) 23 (46) Never smoker 969 (53) 163 (46) 806 (54) 61 (48) 251 (52) 26 (52) Consumers of local vegetarian

products

Number (%) 1015 (75) 183 (78) 832 (74) 77 (79) 333 (83) 38 (76) Consumers of local meat products Number (%) 541 (47) 78 (44) 463 (48) 31 (40) 177 (50) 19 (38) Consumers of locally caughtfish Number (%) 158 (14) 34 (20) 124 (13) 27 (35) 109 (31) 11 (22)

Cancer ICD7

Digestive system Number (%) 151–154 41 (10) 15 (12) Rectum Number (%) 154 17 (4.3) 7 (5.5) Female genital Number (%) 171–174 99 (25) 33 (55) Breast Number (%) 170 76 (19) 25 (38) Male genital Number (%) 177–178 79 (20) 36 (58) Prostate Number (%) 178 73 (18) 33 (53) Kidney and bladder Number (%) 180–181 20 (5.1) 4 (3.1) Brain Number (%) 193 24 (6.1) 9 (7.1) Lymphatic and hematopoitic tissues Number (%) 200–208 23 (5.8) 10 (7.9) Other sites Number (%) 93 (23) 20 (16)

a _{Participants taking part in the biomonitoring study who lived outside the 2 km radius but within 50 km of glassworks at the time of biological sampling.}

Table 2a

Logistic regression analyses with total cancer variable. Restriction criteria of at leastfive years of residence time in the contaminated area and a latency time of at least 10 years betweenfirst residence year and time of diagnosis (cases) or inclusion in the study (controls) were applied. Significant ORs and p-values in bold.

Exposure variable Both gender

Exposed cases Exposed controls ORa_{(95% CI) P-value}a

Ever smoking 195 674 1.14 (1.01–1.29) 0.04

Parents or siblings have had cancer 189 558 1.28 (1.13–1.46) 0.0002

Academic education 97 568 1.05 (0.91–1.20) 0.53

Obesity 62 176 1.23 (1.04–1.45) 0.01

Working at glasswork 37 125 0.92 (0.74–1.13) 0.42

Other metalwork 32 172 0.92 (0.74–1.13) 0.43

Residence in glasswork area before age of 5 yrs 157 772 1.15 (1.01–1.31) 0.03 Residence time in glasswork area (years) 361 1485 1.00 (0.99–1.01) 0.86

logistic regression analysis for total cancer (and specific cancer forms) was performed followed by multivariable logistic regression analysis, which included covariates (Table 2aand2b) with p-value≤0.20.

A significant association with total cancer risk was seen for high consumption of localfish (Table 3). Due to the few cases of specific cancer forms, it was not possible to stratify for gender. Cancer in the digestive system was significantly associated with high consumption of locally produced meat (chicken, lamb, local game meat). Cancer in brain and nervous system was significantly associated with consump-tion of lamb (all consumpconsump-tion versus ever/never consumpconsump-tion of local lamb since there were too few cases in the respective consumption groups;Table 3).

Male genital/prostate cancer was associated with consumption of localfish. Non-significant decreased risk for prostate cancer was asso-ciated with high consumption of forest berries. We found no increased risks for female cancer and consumption of any local food. Reported high consumption of non-local food from places outside the study area was not associated with increased risk for total cancer or specific cancer forms (data not shown in table).

3.3.2. Associations between biomarker concentrations and risk factors for cancer

Active smoking as well as metalwork and glasswork were associated with increased Cd and Pb levels in blood and PAH in urine (Table 4). No significant differences in concentrations of As in blood in relation to

smoking or work were detected. Age was associated with increased concentration of Cd, Pb and As, and residence time was associated with increased concentrations of Pb and As. No associations were found between 1-OHPy concentrations and age or residence time.

High consumers of certain local foods had moderately higher Cd (vegetarian food) and Pb (meat andfish) concentrations in blood, and higher PAH (vegetarian food) concentrations in urine (adjusted for age, glass/metalwork and smoking;Table 5. No such trend was observed for As in blood.

In general, metal concentrations in blood and PAH concentrations in urine were similar or only moderately elevated compared to other po-pulation groups in non-contaminated areas (Supplemental material, Table 4a, b).

Evaluation of tertils of As, Cd and Pb concentrations in blood, and PAH in urine, showed no significant trends in cancer risks.

PCB 118 and 153 concentrations were moderately higher in the high consumption group of meat andfish however the levels were low and did not deviate from population levels in non-contaminated areas (Supplemental material,Table 5). No associations with cancer risk were detected.

4. Discussion

In this extensive study, based on register data, questionnaire in-formation and measurements of biomarkers in blood and urine, we Table 2b

Logistic regression analyses with total cancer variable. Restriction criteria of at leastfive years of residence time in the contaminated area and a latency time of at least 10 years betweenfirst residence year and time of diagnosis (cases) or inclusion in the study (controls) were applied. Significant ORs and p-values in bold.

Exposure variable Men Women

Exposed cases/controls ORa_{(95% CI) p-value}a _{Exposed cases/controls OR}a_{(95% CI) p-value}a

Ever smoking 103/319 1.13 (0.94–1.36) 0.19 92/355 1.13 (0.95–1.34) 0.16 Parents or siblings have had cancer 90/258 1.28 (1.06–1.54) 0.01 99/300 1.29 (1.08–1.53) 0.005 Academic education 48/229 1.23 (0.99–1.52) 0.05 49/339 0.90 (0.75–1.10) 0.30 Obesity 31/67 1.49 (1.16–1.92) 0.001 31/109 1.06 (0.84–1.33) 0.56 Working at glasswork 28/81 1.00 (0.77–1.29) 0.99 9/44 0.76 (0.51–1.13) 0.18 Other metalwork 26/136 0.94 (0.73–1.21) 0.64 6/36 0.93 (0.59–1.47) 0.75 Residence in glasswork area before age of 5 yrs 67/366 0.98 (0.82–1.19) 0.88 90/406 1.32 (1.10–1.59) 0.002 Residence time in the glasswork area (years) 174/689 0.99 (0.98–1.00) 0.19 187/796 1.01 (0.99–1.02) 0.12

a _{Adjusted for age and stratified on gender.}

Loading scatterplot (p1 vs. p2) Age Women Non-smoker Non barbeque Barbeque -1,0 -0,8 -0,6 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 p1 -1,2 -1,0 -0,8 -0,6 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 1,2 2 p Residence time 1-OHPy Men Former smoker Active smoker Low education High education Non glass/metalwork Glasswork Other metalwork

Non cons veg Intermed cons veg High cons veg

Non cons meat

Intermed cons meat High cons meat

Non cons fish

Intermed cons fish

High cons fish

Fig. 2. Loading plot displaying the PCA-analysis of re-lationship between the variables 1-OHPy concentrations in urine (ng/g crea) (n = 400) and the variables age, gender, smoking habit (active, former, non-smoker), glasswork, metalwork, residence time in a glasswork area, consump-tion of barbeque and consumpconsump-tion of local vegetables, meat andfish.

have explored metal and PAH exposure and cancer occurrence in re-lation to consumption of local food and other variables in a historically contaminated glasswork area. As expected, well known exposure sources such as smoking, glass and metalwork, as well as age, and re-sidence time in the area contributed to the total body burden of Cd and Pb (EFSA, 2013;Jarup and Akesson, 2009). Smoking and metal work also contributed to PAH exposure. Furthermore, smoking, age, obesity and family history of cancer, all well known risk factors for cancer (NIH, 2018), as well as residence in a glasswork area during thefirst five years in life, were identified as risk factors for total cancer.

Despite high environmental metal concentrations, overall current metal concentrations in blood were similar or only moderately higher in our study population compared to other subpopulations of the same age-range (Supplemental table 4) (Bjermo et al., 2013b;Helmfrid et al., 2015;Skerfving et al., 2012;Wennberg et al., 2015;Wilde et al., 2016).

Also, the concentration of 1-OHPy in urine in our study population was similar to populations in non-contaminated areas (Alhamdow et al., 2018;Li et al., 2008;Tombolini et al., 2018). We found no significant associations between cancer risk and current levels of metals in blood or with 1-OHPy in urine, but the exposure and body burden may very well have been higher in the past, partly due to inhalation of dust and vo-latile compounds that were released to air prior to the installation of technical solutions forflue gas purification, as well as a larger degree of general environmental contamination. The cancer risk was also higher in the past and has decreased over decades (Nyqvist et al., 2017).

Nevertheless, reported high consumption of certain local foods during the past 30 years was associated with higher Cd and PAH (ve-getables) and Pb and PCBs (meat,fish) concentrations in blood and urine (but not As), compared to reported non/low or intermediate consumption. These results are in agreement with our previous study in Table 3

Case-control analyses for the association between consumption of local food (high consumption vs never or seldom) and total cancer, cancer in digestive system (ICD7: 151–154), cancer in brain, nervous system (ICD7: 193) and male genital cancer (ICD7: 177–178), adjusted for covariates, and with a requirement of at least 5 years of residence time in the contaminated area (cases and controls) and with a latency time requirement of at least 10 years before cancer diagnosis (cases) or inclusion in the study (controls). Based on the bivariate logistic regressions analyses, the criterion for covariates to be entered in the multivariate-adjusted model was p≤ 0.20. Odds ratios (OR) and 95% confidence intervals (CI) are presented for outcomes significantly different from unity.

Exposure/consumption Logistic regression Multivariate logistic regression

Total cancer Exposed cases/controls ORa _{95% CI p-value Exposed cases/controls OR}b _{95% CI p-value}

Salmon, salmon trout, whitefish 19/42 1.82 1.14–2.92 0.01 17/40 1.77 1.07–2.92 0.03 Digestive system cancer

Game meat, chicken, lamb 10/180 2.01 1.07–3.74 0.03 8/165 2.21 1.09–2.86 0.03 Rectum cancer

Game meat, chicken, lamb 6/180 2.92 1.16–7.36 0.02 5/165 3.47 1.25–9.66 0.02 Brain and nervous system

All consumption of local lamb 5/127 2.62 1.34–5.14 0.004 5/124 2.65 1.34–5.24 0.005

Men Exposed case/control ORc _{95% CI p-value Exposed cases/controls OR}d _{95% CI p-value}

Male genital cancer

Forest berries 8/63 0.66 0.36–1.23 0.19 8/60 0.69 0.36–1.29 0.25 Salmon, salmon trout, whitefish 9/19 2.52 1.48–4.31 < 0.001 8/18 2.40 1.35–4.25 0.003 Prostate cancer

Forest berries 7/63 0.52 0.27–1.02 0.06 7/60 0.53 0.26–1.07 0.07 Salmon, salmon trout, whitefish 8/19 1.42 2.02–4.62 0.002 7/18 2.35 1.23–4.45 0.009

Significant ORs and p-value in bold. a _{Adjusted for age and gender.}

b _{Adjusted for age, gender, smoking, obesity, family history of cancer and residence in glasswork area before age 5 yrs.} c _{Adjusted for age.}

d_{Adjusted for age, smoking, obesity, education level, and family history of cancer.}

Table 4

Mean concentrations (SD) of Cd, Pb in blood (μg/L) (n = 658 individuals) and 1-OHPy in urine (ng/g crea) (n = 400 individuals) in smokers, non-smokers, glass/ metal work and non-glass/metalwork in the biomonitoring study (a subpopulation of the questionnaire study) N = 656. Significant p-value in bold.

Numbera _{B-Cd p-value}b _{B-Pb p-value}b _Numberc _{1-OHPy p-value}d

Smoking habit

Non-smoker 292 0.31 (0.20) Ref 15 (9.9) Ref 198 62 (73) Ref Former smoker 244 0.45 (0.41) < 0.0001 18 (12) 0.03 171 71 (84) 0.89 Active smoker 25 1.4 (0.69) < 0.0001 27 (17) < 0.0001 16 246 (209) < 0.0001 Work

Non glass/metal work 496 0.40 (0.36) Ref 16 (10) Ref 293 66 (73) Ref Glasswork 65 0.57 (0.65) 0.0001 25 (16) < 0.0001 49 68 (123) 1 Glass blower 10 0.79 (0.99) 0.0002 38 (17) < 0.0001 5 79 (41) 0.70 Glass grinder 14 0.59 (0.66) 0.02 25 (18) 0.001 9 123 (117) 0.02 Other metalwork 94 0.34 (0.20) 0.05 18 (9.3) 0.04 43 129 (153) < 0.0001

Table 4. Significant p-value in bold.

a _{Number of participants, who donate blood samples for metal analysis.} b _{Adjusted for age, gender, and glass/metalwork.}

c _{Number of participants, who donate urine samples for PAH-analysis.} d_{Adjusted for age, gender and smoking.}

a historically contaminated area, in which reported consumption of certain local vegetables, forest berries and mushrooms were associated with elevated Cd concentrations in blood and urine (Helmfrid et al., 2015). In parallel with the current study, samples of vegetables (home-grown lettuce and potatoes), berries and mushrooms from the same area were analyzed (Augustsson et al., 2018). In spite of elevated concentration of Cd, Pb and As in soil in adjacent gardens to glass works, metal concentrations in food were only moderately elevated and did not exceed the threshold limit values for the food products, but concentrations were higher in all crops closer to a glasswork compared to references areas further away (Augustsson et al., 2018). Although the metal concentrations in crops were generally below the permissible levels, a probabilistic exposure assessment showed that inhabitants living close to a contaminated glasswork area and who consumed local vegetables, berries and mushrooms could have significantly elevated intake of Cd, Pb and As than other populations (Augustsson et al., 2018). Elevated concentrations of PAH have earlier been shown in vegetables grown in contaminated areas (Paris et al., 2018;Wang et al., 2018). As expected, PCB concentrations in blood increased with in-creasing intake of fatty foods such as meat andfish, especially fish from the Baltic Sea, however the concentrations were low, and there is no known local source of PCB in the area.

In the present study, a high consumption of all localfish species was significantly associated with higher Pb levels in blood. Previous studies in the glasswork area have reported high levels of As, Pb and Cd in river sediments (Höglund et al., 2007). Fish can accumulate metals from food, water and sediments, the metal accumulation depends on both biological and ecological factors such as feeding habits, habitats, spe-cies, age and size offishes (Mendil et al., 2010;Zhao et al., 2012;Zohra and Habib, 2016). We did not measure metal concentrations in fish samples in the current study, but higher concentrations of metals infish muscle tissues have been found in areas with polluted sediments com-pared to areas with unpolluted sediments (Mendil et al., 2010;Zohra and Habib, 2016).

High consumption of lamb, game meat and chicken was sig-nificantly associated with elevated Pb exposure in our study. Earlier studies have observed that free range chicken, goat, sheep, cattle and horse grazing infields and close to metal process industry, dumps and mines are exposed to higher levels of Cd and Pb (Bortey-Sam et al., 2015; Farmer and Farmer, 2000; Pareja-Carrera et al., 2014). If the lamb and chicken consumed by the study participants in the present study were free ranged were not known.

Inorganic As, which is more toxic than the organic forms (WHO, 2001) was added to remove gas bubbles during the smelting process in the glass manufacturing (Magnusson, 1974). In the present study

relatively low As levels were found in the study population (blood: mean 8.4μg/L; range 0.55–48 μg/L). Unfortunately, we were not able to specify As in urine in this study. Consumption of anyfish during the latest three days was significantly associated with higher levels of As, probably due to exposure to the less toxic organic arsenic species that are typically found in high levels infish and sea foods (EFSA, 2009b). Despite of elevated metal concentrations in groundwater and sur-roundings soil, the private well water from households close to glass-works sites, had concentrations below the European drinking water criteria for As, Sb, Cd and Ba, and few drinking water samples had elevated (> 10μg/L) Pb concentrations (Augustsson et al., 2016;EFSA, 2009b;Livsmedelsverk, 2001;WHO, 2011).

In the present study, reported high consumption of localfish and meat (local game meat, chicken and lamb), but not vegetables, was associated with increased risk for various cancer types (among those digestive and prostate cancer). In the register study by Nyqvist et al. (Nyqvist et al., 2017), which included our study area, a dose-response relationship between the levels of Cd, Pb, and As in soil (three levels associated with distance from glass works), and cancer incidence for digestive cancer and prostate cancer was reported. For all three metals, a significantly increased risk for these cancer forms were detected in the highest contamination category, except for As which was significantly associated with an increased risk for digestive cancer both in high and medium pollution category area. In our previous case-control study in a historically contaminated area with elevated environmental levels of Pb, Cd and PCB, long-term high consumption of localfish showed as-sociation with total cancer, and with lymphoma, female breast cancer, cervix and corpus uteri cancer (Helmfrid et al., 2012). The associations between consumption of local food and different types of cancer may reflect a higher exposure in the past, and thus, if consumption of local food contributes to the risk of acquiring cancer, that risk is most probably lower today than previously. Other types of health effects that may be associated with high metal exposure were not investigated, i.e. osteoporosis, kidney disease or cardiovascular disease, all which have been associated with metal exposure (Akesson et al., 2014; ATSDR, 2007;Barregard et al., 2016; Larsson and Wolk, 2016;Nordberg and Nordberg, 2016). Furthermore, other contaminants in food have not been considered.

In the present study, former and active glass blowers and glass grinders had the highest levels of Cd and Pb in blood among glass-workers, but not higher than current smokers. In glassworks producing crystal glass, high concentrations of Pb in indoor air and in slag deposits inside of blow pipes have earlier been measured (Andersson et al., 1990). In addition, earlier studies of mortality in the Swedish glass-works industry showed that especially glass blower had an increased Table 5

Mean concentration of Cd, Pb in blood (μg/L) (n = 658 individuals) and 1-OHPy in urine (ng/g crea) (n = 400 individuals) in different consumption groups (non/ low, intermediate and high) of locally produced food (n = 656). Significant p-values in bold.

Exposure Consumption group Numbera _{B-Cd p-value}b _{B-Pb p-value}b _Numberc _{1-OHPy p-value}d

Root crops, vegetables Non/low 195 0.37 (0.35) Ref 17 (12) Ref 107 65 (47) Ref Intermediate 125 0.38 (0.36) 1 15 (9.7) 0.19 78 59 (50) 1 High 158 0.44 (0.40) 0.04 18 (12) 1 88 99 (149) 0.01 Fruit, berries Non/low 122 0.37 (0.31) Ref 16 (9.8) Ref 76 73 (111) Ref

Intermediate 256 0.37 (0.38) 1 17 (11) 1 152 69 (68) 1 High 145 0.46 (0.40) 0.06 18 (12) 0.50 72 77 (110) 1 Lamb, chicken, game meat Non/low 288 0.36 (0.31) Ref 16 (11) Ref 159 66 (54) Ref

Intermediate 88 0.41 (0.38) 0.43 17 (11) 0.67 53 105 (174) 0.009 High 94 0.42 (0.46) 0.19 20 (11) 0.0009 53 68 (69) 1 Allfish species Non/low 286 0.38 (0.35) Ref 16 (10) Ref 160 72 (76) Ref

Intermediate 124 0.35 (0.27) 0.80 17 (9.9) 1 69 76 (125) 1 High 47 0.46 (0.43) 0.11 22 (16) 0.003 32 75 (87) 1

a _{Number of participants, who donate blood samples for metal analysis.} b _{Adjusted for age, glass/metalwork and smoking.}

c _{Number of participants, who donate urine samples for PAH-analysis.} d_{Adjusted for age, smoking and consumption of barbeque.}

risk of mortality in stomach cancer, colon cancer and lung cancer (Wingren and Axelson, 1987). However, the main aim of our study was not to study occupational exposure and cancer risk, and only a small number of glass workers were included in the study.

There are several limitations in this study. The most notable lim-itation is the low response rate (about 30%). However, we included a larger number of controls than is usual to compensate for an expected low response rate. It is well known that the response rate in ques-tionnaire studies tend to decrease, especially if the individuals do not feel concerned or motivated to participate (Wenemark et al., 2010). Many of the non-participants had lived for only a short period in the study area, or responded that they were too old and/or sick to parti-cipate. The response rate was higher among cancer cases than controls, which was expected, due to that they often feel more motivated to answer questionnaires on life styles and health effects. The possibility of recall bias and exposure misclassification, with cancer cases system-atically overestimating their consumption of local food, have to be considered. Few cases of specific cancer forms in each consumption group and in each tertil of contaminant concentrations increase the uncertainty of the results, but the high number of controls compensate for that to some extent.

We used a comprehensive questionnaire containing questions on self-reported life-time residence, occupation, current and past food consumption habits during the past 30 years. There is obviously a risk for misclassification of exposure since it is difficult to remember fquencies and types of food consumed back in time. Generally, the re-sponders knew if they had consumed local food or not and could rea-sonably well assess approximate frequencies over time. All subjects in the study population were more or less exposed to metals and PAHs from several possible sources e.g. residing in a contaminated area, via drinking water, occupation, life style and general food, making it dif-ficult to obtain contrast in exposure and therefore difficult to establish significant associations. In the present study, none of the responders reported non-consumption of non-local food (e.g. food from other areas). In order to accomplish a reasonably good power (80%), the study population in the biomonitoring study should preferably be larger and missing information in the questionnaire should be less frequent. There was a falling off in recording in several of the food frequency questions. About 18% of the participants in the biomonitoring study did not answer the questions about consumption of local vegetables and about 30% of them did not answer the questions about consumption of local animal products.

Despite the limitations and the weaknesses described, it was pos-sible to establish associations between exposure sources and con-taminant concentrations in blood or urine, as well as for covariates associated with increased cancer risk. It could be concluded that the variables in the questionnaire reasonably well described differences in metal exposure in the study population. The strength in our study is the availability of reliable data from the national cancer registers in Sweden and the large amount of information collected through questionnaires, biomonitoring and environmental measurements.

5. Conclusions

Despite (historically) high environmental levels of Cd, Pb and As at glasswork sites and landfills, current metal exposure in the population living in the surrounding areas was similar or only moderately higher in our study population compared to the general population. That was true also for PAH exposure. No significant associations with increased cancer risk were detected in the highest tertil of Cd, Pb and As in blood or PAH in urine.

Reported high consumption of certain local food was associated with higher Cd (vegetarian food) and Pb (fish, meat), but not As, con-centrations in blood, and higher PAH (vegetarian food), and it was concluded that the questionnaire information on consumption of local foods reasonably well described differences in food consumption in the

study population.

An increased risk of cancer was associated with smoking, family history of cancer, obesity, and residence in glasswork area before age 5 years. Also, a high consumption of local foods, i.e. fish and meat (game, chicken, lamb), was associated with increased risk of various cancer forms, whereas a high consumption of forest berries was asso-ciated with a non-significant decreased risk for prostate cancer. The associations between habitual consumption of local food and different types of cancer may reflect a higher exposure in the past, and thus, if consumption of local food contributes to the risk of acquiring cancer, that contribution is probably lower today than before. Furthermore, it cannot be ruled out that other contaminants in the food contribute to the increased cancer risks observed.

Declaration of Competing Interest

We confirm that all authors have declared no actual or potential competing interests regarding the submitted article.

We declare that all authors a) have read the manuscript, b) agree the work is ready for submission to Environment International, and c) ac-cept responsibility for the manuscript's contents.

Acknowledgement

We express our gratitude to the participating population. We thank the research nurse Louice Eskilsson for excellent assistance in sample collection and for the dispatch of questionnaires together with Fredrik Nyqvist. We also gratefully acknowledge Anna-Lena Hällsten and Inger Nordén Larsson for their assistance in the study and to Bengt Ståhlbom, for his support. Finally, we gratefully acknowledge our retired collea-gues Gun Wingren (who initiated studies in the glasswork area) and Lennart Hellström for their help with the study design. The study was financed by the Swedish Environmental Protection Agency; Medical Research Council of Southeast Sweden; The Kamprad Family Foundation, Sweden, and by Occupational and Environmental Medicine, Linköping University Hospital, Sweden.

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps:// doi.org/10.1016/j.envint.2019.104985.

References

Akesson, A., Barregard, L., Bergdahl, I.A., Nordberg, G.F., Nordberg, M., Skerfving, S., 2014. Non-renal effects and the risk assessment of environmental cadmium exposure.

Environ. Health Perspect. 122, 431–438.

Alhamdow, A., Lindh, C., Hagberg, J., Graff, P., Westberg, H., Krais, A.M., Albin, M., Gustavsson, P., Tinnerberg, H., Broberg, K., 2018. DNA methylation of the cancer-related genes F2RL3 and AHRR is associated with occupational exposure to

poly-cyclic aromatic hydrocarbons. Carcinogenesis 39, 869–878.

Ali, I., Julin, B., Glynn, A., Hogberg, J., Berglund, M., Johansson, J.E., Andersson, S.O., Andren, O., Giovannucci, E., Wolk, A., Stenius, U., Akesson, A., 2016. Exposure to polychlorinated biphenyls and prostate cancer: population-based prospective cohort

and experimental studies. Carcinogenesis 37, 1144–1151.

Alicandro, G., Rota, M., Boffetta, P., La Vecchia, C., 2016. Occupational exposure to polycyclic aromatic hydrocarbons and lymphatic and hematopoietic neoplasms: a

systematic review and meta-analysis of cohort studies. Arch. Toxicol. 90, 2643–2656.

Andersson, L., Wingren, G., Axelson, O., 1990. Some hygienic observations from the glass

industry. Int. Arch. Occup. Environ. Health 62, 249–252.

Arisawa, K., Nakano, A., Saito, H., Liu, X.J., Yokoo, M., Soda, M., Koba, T., Takahashi, T., Kinoshita, K., 2001. Mortality and cancer incidence among a population previously

exposed to environmental cadmium. Int. Arch. Occup. Environ. Health 74, 255–262.

Arrebola, J.P., Fernandez-Rodriguez, M., Artacho-Cordon, F., Garde, C., Perez-Carrascosa, F., Linares, I., Tovar, I., Gonzalez-Alzaga, B., Exposito, J., Torne, P., Fernandez, M.F., Olea, N., 2016. Associations of persistent organic pollutants in serum and adipose

tissue with breast cancer prognostic markers. Sci. Total Environ. 566-567, 41–49.

ATSDR, 2007. Agency for Toxicity Substances and Disease Registry. Toxicological Profile

for Lead.

Augustsson, A., Uddh Soderberg, T., Jarsjo, J., Astrom, M., Olofsson, B., Balfors, B., Destouni, G., 2016. The risk of overestimating the risk-metal leaching to groundwater near contaminated glass waste deposits and exposure via drinking water. Sci. Total

Environ. 566-567, 1420–1431.

Augustsson, A., Uddh-Soderberg, T., Filipsson, M., Helmfrid, I., Berglund, M., Karlsson, H., Hogmalm, J., Karlsson, A., Alriksson, S., 2018. Challenges in assessing the health risks of consuming vegetables in metal-contaminated environments. Environ. Int.

113, 269–280.

Augustsson, A.L., Uddh-Soderberg, T.E., Hogmalm, K.J., Filipsson, M.E., 2015. Metal uptake by homegrown vegetables - the relative importance in human health risk

assessments at contaminated sites. Environ. Res. 138, 181–190.

Barany, E., Bergdahl, I., Schϋtz, A., Skerfving, S., Oskarsson, A., 1997. Inductively cou-pled plasma mass spectrometry for DirectMulti-element analysis of diluted human

blood andSerum. J. Anal. At. Spectrom. 123, 1005–1009.

Barregard, L., Sallsten, G., Fagerberg, B., Borne, Y., Persson, M., Hedblad, B., Engstrom, G., 2016. Blood cadmium levels and incident cardiovascular events during follow-up in a population-based cohort of Swedish adults: the Malmo diet and Cancer study.

Environ. Health Perspect. 124, 594–600.

Bjermo, H., Sand, S., Nalsen, C., Lundh, T., Enghardt Barbieri, H., Pearson, M., Lindroos, A.K., Jonsson, B.A., Barregard, L., Darnerud, P.O., 2013b. Lead, mercury, and cad-mium in blood and their relation to diet among Swedish adults. Food Chem. Toxicol.

57, 161–169.

Bortey-Sam, N., Nakayama, S.M., Ikenaka, Y., Akoto, O., Baidoo, E., Yohannes, Y.B., Mizukawa, H., Ishizuka, M., 2015. Human health risks from metals and metalloid via consumption of food animals near gold mines in Tarkwa, Ghana: estimation of the daily intakes and target hazard quotients (THQs). Ecotoxicol. Environ. Saf. 111,

160–167.

Calderon, J., Ortiz-Perez, D., Yanez, L., Diaz-Barriga, F., 2003. Human exposure to metals. Pathways of exposure, biomarkers of effect, and host factors. Ecotoxicol. Environ. Saf.

56, 93–103.

Cao, S., Duan, X., Zhao, X., Chen, Y., Wang, B., Sun, C., Zheng, B., Wei, F., 2016. Health risks of children's cumulative and aggregative exposure to metals and metalloids in a

typical urban environment in China. Chemosphere 147, 404–411.

Carmella, S.G., Le, K.A., Hecht, S.S., 2004. Improved method for determination of

1-hydroxypyrene in human urine. Cancer Epidemiol. Biomark. Prev. 13, 1261–1264.

Darnerud, P.O., Atuma, S., Aune, M., Bjerselius, R., Glynn, A., Grawe, K.P., Becker, W., 2006. Dietary intake estimations of organohalogen contaminants (dioxins, PCB, PBDE and chlorinated pesticides, e.g. DDT) based on Swedish market basket data. Food

Chem. Toxicol. 44, 1597–1606.

Dreij, K., Seidel, A., Jernstrom, B., 2005. Differential removal of DNA adducts derived from anti-diol epoxides of dibenzo[a,l]pyrene and benzo[a]pyrene in human cells.

Chem. Res. Toxicol. 18, 655–664.

EFSA, 2008. Polycyclic Aromatic Hydrocarbons in Food.

EFSA, 2009a. EFSA CONTAM. Scientific opinion of the panel on contaminants in the food chain on a request from the European Commission on cadmium in food. EFSA J. 980,

1–138.

EFSA, 2009b. EFSA Scientific Opinion on Arsenic in Food. Report from EFSA Panel on

Contaminants in the Food Chain (CONTAM).

EFSA, 2010. EFSA CONTAM. Scientific opinion on lead in food. EFSA J. 8, 1–151.

EFSA, 2013. Scientific Opinion on Lead in Food. Report from EFSA Panel on

Contaminants in the Food Chain (CONTAM).

Farmer, A., Farmer, A., 2000. Concentration of cadmium, lead, zinc in livestock feed and organs around metal production Centre in eastern Kazakhstan. Sci. Total Environ.

257, 53–60.

Filipsson, M., Augustsson, A., Nyholm, S., 2013. Consumption of vegetables and risk perception at contaminated glasswork sites in Sweden. In: Poster Vid SEE, ISES and

ISIAQ Environmental Health Conference, 19–23 August 2013, Basel, Switzerland.

Glorennec, P., Lucas, J.P., Mercat, A.C., Roudot, A.C., Le Bot, B., 2016. Environmental and dietary exposure of young children to inorganic trace elements. Environ. Int. 97,

28–36.

Hellstrom, L., Persson, B., Brudin, L., Grawe, K.P., Oborn, I., Jarup, L., 2007. Cadmium exposure pathways in a population living near a battery plant. Sci. Total Environ.

373, 447–455.

Helmfrid, I., Berglund, M., Lofman, O., Wingren, G., 2012. Health effects and exposure to polychlorinated biphenyls (PCBs) and metals in a contaminated community. Environ.

Int. 44, 53–58.

Helmfrid, I., Salihovic, S., van Bavel, B., Wingren, G., Berglund, M., 2015. Exposure and body burden of polychlorinated biphenyls (PCB) and metals in a historically

con-taminated community. Environ. Int. 76, 41–48.

Henriquez-Hernandez, L.A., Luzardo, O.P., Zumbado, M., Serra-Majem, L., Valeron, P.F., Camacho, M., Alvarez-Perez, J., Salas-Salvado, J., Boada, L.D., 2017. Determinants of increasing serum POPs in a population at high risk for cardiovascular disease. Results

from the PREDIMED-CANARIAS study. Environ. Res. 156, 477–484.

Höglund, L., Fanger, G., Yesilova, H., 2007. Resultat och tolkningar från undersökningar

av 25 glasbruksobjekt och 6 glasbruksåar. Kemakta konsult AB (in Swedish).

IARC, 1993. (International Agency for Research on Cancer) Beryllium, Cadmium,

Mercury, and Exposures in the Glass Manufacturing Industry.

IARC, 2010. Some Non-heterocyclic Polycyclic Aromatic Hydrocarbons and some Related

Occupational Exposures. Lyon, France. Distributed by World Health Organization.

IARC, 2018. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans.

Polychlorinated Biphenyls and Polybrominated Biphenyls.

Jarup, L., Akesson, A., 2009. Current status of cadmium as an environmental health

problem. Toxicol. Appl. Pharmacol. 238, 201–208.

Julin, B., Wolk, A., Johansson, J.E., Andersson, S.O., Andren, O., Akesson, A., 2012. Dietary cadmium exposure and prostate cancer incidence: a population-based

pro-spective cohort study. Br. J. Cancer 107, 895–900.

Lagerqvist, A., Hakansson, D., Lundin, C., Prochazka, G., Dreij, K., Segerback, D., Jernstrom, B., Tornqvist, M., Frank, H., Seidel, A., Erixon, K., Jenssen, D., 2011. DNA

repair and replication influence the number of mutations per adduct of polycyclic

aromatic hydrocarbons in mammalian cells. DNA Repair (Amst) 10, 877–886.

Larsson, S., Bergbäck, B., Eklund, M., Lohm, U., 1999. Historical emissions from Swedish crystal-glass production and future scenarios of soil lead pollution. Environ. Rev. 7,

53–60.

Larsson, S.C., Wolk, A., 2016. Urinary cadmium and mortality from all causes, cancer and cardiovascular disease in the general population: systematic review and

meta-ana-lysis of cohort studies. Int. J. Epidemiol. 45, 782–791.

Li, Z., Sandau, C.D., Romanoff, L.C., Caudill, S.P., Sjodin, A., Needham, L.L., Patterson, D. G., Jr., 2008. Concentration and profile of 22 urinary polycyclic aromatic hydro-carbon metabolites in the US population. Environ. Res.. 107, 320–331.

Livsmedelsverk, S.S., 2001. Statens Livsmedelsverks föreskrifter om dricksvatten.

Magnusson, J., 1974. Glas - glasets egenskaper och tillverkning (in Swedish). Liber

läromedel.

Mendil, D., Unal, O.F., Tuzen, M., Soylak, M., 2010. Determination of trace metals in different fish species and sediments from the river Yesilirmak in Tokat, Turkey. Food

Chem. Toxicol. 48, 1383–1392.

NIH. National Cancer Institute. 2018.

Nordberg, M., Nordberg, G.F., 2016. Trace element research-historical and future aspects.

J. Trace Elem. Med. Biol. 38, 46–52.

Nordström, O., 1999. Förteckning över glasbruk och hyttor i Sverige 1150–2000

Glasteknisk Tidskrift. vol. 54. pp. 77–84 (in Swedish).

Nyqvist, F., Helmfrid, I., Augustsson, A., Wingren, G., 2017. Increased Cancer incidence in the local population around metal-contaminated glassworks sites. J. Occup. Environ.

Med. 59, e84–e90.

Pareja-Carrera, J., Mateo, R., Rodriguez-Estival, J., 2014. Lead (Pb) in sheep exposed to mining pollution: implications for animal and human health. Ecotoxicol. Environ.

Saf. 108, 210–216.

Paris, A., Ledauphin, J., Poinot, P., Gaillard, J.L., 2018. Polycyclic aromatic hydrocarbons in fruits and vegetables: origin, analysis, and occurrence. Environ. Pollut. 234,

96–106.

Parveen, R., Saini, R., Taneja, A., 2018. Chemical characterization and health risk as-sessment of soil and airborne particulates metals and metalloids in populated

semi-arid region, Agra, India. Environ. Geochem. Health 40, 2021–2035.

Rylander, L., Nilsson-Ehle, P., Hagmar, L., 2006. A simplified precise method for adjusting serum levels of persistent organohalogen pollutants to total serum lipids.

Chemosphere 62, 333–336.

SEPA, 2009. (Swedish Environmental Protection Agency) Riktvärden för förorenad mark -modellbeskrivning och vägledning. Swedish Environmental Protection Agency (in

Swedish).

Skerfving, S., Löfmark, L., Rentschler, G., Lundh, T., 2012. Kadmiumhalter i blod och urin

hos skånska kvinnor 2010– med jämförelser bakåt till 1999/2000. In: Rapport från

Arbets- och miljömedicin, Lund.

Steenland, K., Boffetta, P., 2000. Lead and cancer in humans: where are we now? Am. J.

Ind. Med. 38, 295–299.

Stubleski, J., Kukucka, P., Salihovic, S., Lind, P.M., Lind, L., Karrman, A., 2018. A method for analysis of marker persistent organic pollutants in low-volume plasma and serum

samples using 96-well plate solid phase extraction. J. Chromatogr. A 1546, 18–27.

Thomas, L.D., Elinder, C.G., Wolk, A., Akesson, A., 2014. Dietary cadmium exposure and chronic kidney disease: a population-based prospective cohort study of men and

women. Int. J. Hyg. Environ. Health 217, 720–725.

Tombolini, F., Pigini, D., Tranfo, G., Paci, E., Carosi, I., Marini, F., Bauleo, L., Ancona, C., Forastiere, F., 2018. Levels of urinary metabolites of four PAHs and cotinine de-termined in 1016 volunteers living in Central Italy. Environ. Sci. Pollut. Res. Int. 25,

28772–28779.

Uddh-Soderberg, T.E., Gunnarsson, S.J., Hogmalm, K.J., Lindegard, M., Augustsson, A.L.M., 2015. An assessment of health risks associated with arsenic exposure via consumption of homegrown vegetables near contaminated glassworks sites. Sci. Total

Environ. 536, 189–197.

Van Liedekerke, M., Prokop, G., Rabl-Berger, S., Kibblewhite, M., Louwagie, G. 2014. Progress in the Management of Contaminated Sites in Europe. European Commission Joint Research Centre, Institute for Environment and Sustainability, Ispra (VA), Italy,

Wang, L., Li, C., Jiao, B., Li, Q., Su, H., Wang, J., Jin, F., 2018. Halogenated and parent polycyclic aromatic hydrocarbons in vegetables: levels, dietary intakes, and health

risk assessments. Sci. Total Environ. 616-617, 288–295.

Wenemark, M., Hollman Frissman, G., Svensson, T., Kristensson, M., 2010. Respondent satisfaction and respondent burden among differently motivated participiants in

health-related survey. Sage Journals 22, 1–14.

Wennberg, M., Lundh, T., Nilsson Sommar, J., Bergdahl, I., 2015. Time trends of lead and cadmium in the adult population of northern Sweden 1990-2014. In: Yrkes- och

miljömedicin i Umeå.

WHO, 2001. Envrionmental Health Criteria 224. Arsenic and Arsenics Compounds.

WHO, 2011. Guidelines for Drinking-Water Quality, forth edition. WHO press, Geneva.

Wilde, K., Forsgard, N., Hovgard, A., Sällsten, G., 2016. Kadmiumexponering och markör

för njurpåverkan hos yngre och medelålders kvinnor i Västsverige 2015.

Wingren, G., Axelson, O., 1987. Mortality in the Swedish glassworks industry. Scand. J.

Work Environ. Health 13, 412–416.

Wingren, G., Axelson, O., 1992. Cluster of brain cancers spuriously suggesting

occupa-tional risk among glassworkers. Scand. J. Work Environ. Health 18, 85–89.

Zhao, S., Feng, C., Quan, W., Chen, X., Niu, J., Shen, Z., 2012. Role of living environments

in the accumulation characteristics of heavy metals infishes and crabs in the Yangtze

River estuary, China. Mar. Pollut. Bull. 64, 1163–1171.

Zohra, B.S., Habib, A., 2016. Assessment of heavy metal contamination levels and toxicity

in sediments andfishes from the Mediterranean Sea (southern coast of Sfax, Tunisia).