Complex environmental contaminant mixtures and their associations with thyroid hormones using supervised and unsupervised machine learning techniques

Chia sẻ: _ _ | Ngày: | Loại File: PDF | Số trang:9

Thêm vào BST

Báo xấu

14
lượt xem 1
download

Download Vui lòng tải xuống để xem tài liệu đầy đủ

Bayesian kernel machine regression (BKMR) analysis was used to evaluate the univariate contaminant exposure effect as well as the contaminant mixture effects on levels of thyroid hormones. Significant and positive associations were found between total T3 and PC-2 (high positive nickel and cadmium loadings), total T3 and PC-3 (negative association with negative loading for nickel and positive loading for cadmium) and TSH and PC-1 (high positive loadings for organic contaminants).

Chủ đề:

Bình luận(0) Đăng nhập để gửi bình luận!

Lưu

Nội dung Text: Complex environmental contaminant mixtures and their associations with thyroid hormones using supervised and unsupervised machine learning techniques

Environmental Advances 4 (2021) 100054 Contents lists available at ScienceDirect Environmental Advances journal homepage: www.elsevier.com/locate/envadv Complex environmental contaminant mixtures and their associations with thyroid hormones using supervised and unsupervised machine learning techniques Eric N. Liberda a,∗, Aleksandra M. Zuk b, David S. Di a, Robert J. Moriarity c, Ian D. Martin c, Leonard J.S. Tsuji c a School of Occupational and Public Health and Environmental Applied Science and Management, 350 Victoria St, Ryerson University, Toronto M5B2K3, Ontario, Canada b School of Nursing, Queen’s University, Kingston K7L 3N6, Ontario, Canada c Department of Physical and Environmental Sciences, University of Toronto, Toronto M1C 1A4, Ontario, Canada a r t i c l e i n f o a b s t r a c t Keywords: Evaluating complex mixtures and their associated health eﬀects poses a challenge in human populations. Herein, Thyroid we assess the association between 17 organic and metal contaminants in blood with thyroid hormones in a remote Machine learning Indigenous (First Nations) region from Quebec, Canada (n=526). Using principal component analysis (PCA) to Contaminants reduce the number of variables, we generated varimax rotated principal component (PC) loadings of contaminants Indigenous on these uncorrelated synthetic axes. Associations with levels of thyroid hormones (TSH, free T4, and total T3) Exposure BKMR were conducted using multivariable linear regression methods with the participant PC loadings and adjusting for Principal component analysis covariates. Additionally, Bayesian kernel machine regression (BKMR) analysis was used to evaluate the univariate contaminant exposure eﬀect as well as the contaminant mixture eﬀects on levels of thyroid hormones. Signiﬁcant and positive associations were found between total T3 and PC-2 (high positive nickel and cadmium loadings), total T3 and PC-3 (negative association with negative loading for nickel and positive loading for cadmium) and TSH and PC-1 (high positive loadings for organic contaminants). No signiﬁcant observations were observed for free T4. BKMR provided additional insight into the PCA results and suggested that nickel, and not cadmium, was responsible for driving the observed eﬀects with this eﬀect remaining when evaluating the entire mixture. BKMR analysis did not support the association of TSH with organic contaminants that were found in the PCA regression. Our ﬁndings reinforced other studies which showed that metals such as nickel may alter thyroid hormone levels and highlighted how complex environmental mixtures interact with each other. These observations represent an important step to determining how complex mixtures of contaminants can be assessed in human populations, especially those living a subsistence lifestyle who may have high body burdens of contaminants, and to help understand the resultant net eﬀect of exposures on endogenous thyroid hormones utilizing novel machine learning statistical methods. Introduction and carcinogenic eﬀects (Sutcliﬀe and Harvey, 2015). Metabolic condi- tions, such as hyperthyroidism, have been associated with environmen- There are a variety of xenobiotic compounds that may aﬀect the tal contaminants − many of these compounds are still present in humans endocrine system, and it is well established that environmental con- even though some (e.g., legacy POPs) have been removed from use for taminants such as persistent organic pollutants (POPs), toxic metals, decades (Li et al., 2017; Nie et al., 2017; Sarkar et al., 2015). Addition- and metalloids play a role in altering the endocrine milieu in humans ally, the eﬀects of environmental contaminants on thyroid hormone dis- (Alvarez-Pedrerol et al., 2008; Blanco-Muñoz et al., 2016; Duntas and ruption is particularly important in prenatal exposure scenarios due to Stathatos, 2015; Jacobson et al., 2017; Jain, 2016; Jain and Choi, 2016). the development of the fetal brain (Bergman et al., 2013; Damstra et al., In fact, endocrine disruption has been implicated in adverse cardiovas- 2002; Préau et al., 2015; Zoeller, 2007). Thus, it is no surprise that there cular (Klein and Ojamaa, 2001), nervous system (Mnif et al., 2011), have been several investigations assessing various environmental con- ∗ Corresponding author. E-mail addresses: eric.liberda@ryerson.ca (E.N. Liberda), amz4@queensu.ca (A.M. Zuk), s1di@ryerson.ca (D.S. Di), rob.moriarity@utoronto.ca (R.J. Moriarity), ianmartin@mac.com (I.D. Martin), leonard.tsuji@utoronto.ca (L.J.S. Tsuji). https://doi.org/10.1016/j.envadv.2021.100054 Received 8 December 2020; Received in revised form 21 February 2021; Accepted 10 April 2021 2666-7657/Crown Copyright © 2021 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/)
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 taminants in pregnant women and their infants (Baccarelli et al., 2008; 2014b; Tsuji et al., 2005). Furthermore, although many of these legacy Berg et al., 2017, 2015; Dallaire et al., 2009). Therefore, it is impor- contaminants have been banned, they have been shown to have prefer- tant to assess the association that environmental contaminants may have ential deposition to northern latitudes (Wania and Mackay, 1996), hence with endogenous thyroid hormones, which are tightly regulated by the they present a consistent source of human exposure for years to come. hypothalamic–pituitary–thyroid axis, due to the role these xenobiotics A suite of contaminants known to alter thyroid hormone levels (e.g., may play in a variety of diseases and their eﬀects on cognitive function. organochlorine pesticides) were selected for analysis a priori. Addition- While thyroid hormone concentrations and their associations with ally, we included contaminants such as nickel, which have a paucity of contaminants vary by the compound under examination, studies com- information related to its eﬀects on altering human thyroid hormone monly ﬁnd that POPs, such as polychlorinated biphenyls (PCBs), poly- concentrations, as we wished to investigate their role in complex mix- brominated diphenyl ethers (PBDEs), and organochlorine pesticides, are ture eﬀects. Our analyses includes an examination of the overall mix- positively associated with thyroid stimulating hormone (TSH) and neg- ture eﬀect with free T4, total T3, and TSH in participants from the atively with thyroxine (T4) and triiodothyronine (T3) (Meeker et al., Eeyou Istchee using principal component analysis (PCA), an unsupervised 2007; Schell et al., 2008; Xu et al., 2014). However there are also al- machine learning technique, and Bayesian kernel machine regression ternative ﬁndings to these observations and research has shown that (BKMR), a supervised machine learning technique. the direction of these associations may be dependent upon which con- taminant is under investigation (eg., Blanco-Muñoz et al., 2016; Freire Materials and methods et al., 2012, 2013). Findings for metals and metalloids vary greatly de- pending on which element is being examined, but some toxic metals Data sources such as cadmium have been observed to be positively associated with thyroid hormone levels such as total T3 (Chen et al., 2013), while oth- The Eeyou Istchee traditional territory of the eastern James Bay Cree ers, such as selenium (a metalloid), is negatively associated with total consists of nine communities with varying degrees of isolation ranging T3 (Jain and Choi, 2016). Several studies have examined the eﬀects of from all-season roads access to airplane or boat access (Fig. 1). The Eeyou metal exposure on thyroid hormones levels and have found associations Istchee cross-sectional multi-community environment-and-health study, varying by contaminant (Dundar et al., 2006; Luo and Hendryx, 2014; known as the Nituuchischaayihtitaau Aschii (“Learn about ourselves and Meeker et al., 2009; Yorita Christensen, 2013); however, the role by our earth”), collected both personal data and clinical measurements, which this occurs has yet to be determined. Additionally, compounds such as, socio-demographic status, lifestyle data, and blood samples for such as perﬂuorooctane sulfonate (PFOS), perﬂuorooctanoate (PFOA), laboratory analyses. Participants were randomly selected with replace- and other perﬂuoralkyl acids (PFAAs) have also been found to alter thy- ment from the Beneﬁciaries List to achieve an acceptable sample size roid hormones concentrations (Jain, 2013; Li et al., 2017), and may that represented various age-sex groups. Based on the varying popula- thus form complex interactions between groups of legacy POP and toxic tion sizes, this resulted in a minimum of 150 persons being invited from metals. For some contaminants, such as nickel, there remains little infor- each community and represented approximately 20-30 persons per age- mation about its potential eﬀects on human thyroid hormones, although sex group. In total, there were 13,714 eligible participants, 3588 in- a recent study found no association between urinary nickel and thyroid vited, and we obtained an all-community participation rate of 48% for hormones (Castiello et al., 2020). Clearly, this poses a problem when as- those invited to the study (range: 35-68% participation per community) sessing human populations because exposures to environmental contam- (Nieboer et al., 2013). Due to community requests for reporting non- inants do not occur in isolation. For example, a person may be exposed disaggregated data, we only reported aggregated statistics. In total, 1730 to a wide variety of xenobiotics from several diﬀerent pathways and/or participants from the nine Eeyou Istchee communities provided written sources over time rather than a single (or similar group of) contami- informed consent in Cree, English, or French languages. The study ran nants. Therefore, it is possible that the eﬀect of one contaminant may from 2005 to 2009 and included all nine First Nation communities from co-vary with others resulting in synergistic, antagonistic, or potentiative the Eeyou Istchee. Due to the remote location of these communities com- observations which are incorrectly attributed to the single exposure un- bined with the scale of the study approximately two communities were der investigation. The issue of exposures to complex mixtures is not new sampled each year. Ethics approval was granted by McGill University, and this has been a topic of research in the endocrine disruption ﬁeld Laval University, McMaster University, and the Cree Board of Health with researchers assessing various groups of environmental contaminant and Social Services of James Bay. compounds and their related thyroid hormone outcomes for some time (Ghisari et al., 2015; Grønnestad et al., 2018; Taxvig et al., 2013). Fur- Study Population thermore, the mechanisms of how metals alter thyroid hormones has not been elucidated, although it has been postulated that they may af- Of the initial 1730 recruited participants, persons younger than 19 fect adrenal gland function and morphology, metabolism, synthesis of years of age and older than 80 years of age were excluded from anal- hormones, and even cellular apoptosis (Boas et al., 2006; Brouwer et al., ysis due to concerns of potential abnormal thyroid hormone measures 1998, 1986; Buha et al., 2018; Cheek et al., 1999; Rana, 2014). (Bremner et al., 2012; Gesing et al., 2012). Following case completion As part of the Eeyou Istchee Environment and Health study in north- with all covariate and exposure parameters, and after removing persons ern Quebec, Canada, we investigated the associations between several with medical chart conﬁrmed thyroid diseases (hyper- and hypothy- groups of environmental contaminants including POPs and metals with roidism), the ﬁnal cohort of participants was 526, of which 307 were thyroid hormone concentrations using machine learning methods. The female and 219 were male. Thyroid diseases such as hyper- and hy- remote communities within Eeyou Istchee are comprised of First Na- pothyroidism were conﬁrmed via medical chart review by a research tions persons who live (in part) a subsistence lifestyle and have higher nurse. All sampling occurred after fasting and at the same time of day. body burdens of contaminants when compared to non-First Nations per- Only contaminants with ≥25% of the data being above the limit of quan- sons from southern Canada [see (Liberda et al., 2014b, 2014a, 2018; titation were selected for analysis. Nieboer et al., 2017)]. Since these contaminants are found ubiquitously in the environment, those who lead a subsistence lifestyle, a classic path- Thyroid Hormone Measurements way of exposure to many contaminants, may be at higher risk to adverse endocrine system eﬀects due to the bioacumulative nature of some of TSH, free T4, and total T3 were measured on the ADVIA Centaur An- these compounds. Long-range transport and point source releases from alyzer automated chemiluminescent immunoassay system (Bayer Health mining and/or hydroelectric development are other potential sources Care Systems, Terrytown, NY). TSH, free T4, and total T3 had functional (Canadian Mining Journal, 2017; Gouin et al., 2004; Liberda et al., sensitivities of 0.01 mIU/L, 1.3 pmol/L, and 0.15 nmol/L, respectively 2
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 Fig. 1. Eeyou Istchee Territory, Quebec, Canada. (Bonnier-Viger et al., 2007). The respective reference ranges were 0.25- assessed by a clinical ﬁeld research nurse. Lipids were determined as 4.5 mIU/L for TSH, 8-27 pmol/L for free T4, and 1.3-3.1 nmol/L for total per Rylander et al. (2012). Iodine was not included as a covariate as T3, which took into account disease diagnostics (Bonnier-Viger et al., we have previously reported these values to be within normal ranges 2007). Although the use of reference ranges as a diagnostic tool is a (Tam et al., 2015), and further conﬁrmed via a sensitivity analysis that debated practice (Chaker et al., 2017), we included them to provide its inclusion in the models did not signiﬁcantly alter the observed re- context. sults. Age (Bremner et al., 2012), sex (Curtis et al., 2019), smoking status (Jain, 2013), blood lipids (Yorita Christensen, 2013), and BMI Environmental Contaminants Analysis (Yorita Christensen, 2013) were added as covariates in both the PCA and BKMR machine learning analyses as these variables are known a Assessment of metals included total mercury, lead, nickel, and cad- priori risk factors. Age, blood lipids, and BMI were used as continuous mium in whole blood (percent below detection: 1.52%, 0.19%, 2.09%, variables while sex and smoking status were categorical variables. 0.38%). Samples were assessed by inductively coupled plasma mass spectrometry (ICP-MS) on a Perkin Elmer Sciex Elan 6000 instrument. Statistical Analysis The methods and QA/QC have been previously reported (Bonnier- Viger et al., 2007; Nieboer et al., 2017). Blood plasma was assessed Descriptive statistics have been reported as arithmetic and geo- for nine PCBs (118, 138, 153, 156, 163, 170, 180, 183, and 187; metric means (± standard deviations (SD)), frequencies, or percent- percent below detection: 11.22%, 1.71%, 0.38%, 20.34%, 16.73%, ages, where appropriate. Unsupervised (PCA) and supervised (BKMR) 10.27%, 1.33%, 23.95%, 7.79%), and four organochlorine pesticides machine learning techniques were used to evaluate the mixture ef- or metabolites (p,p’-DDE, mirex, oxy-chlordane, and trans-nonachlor; fects of the contaminants on thyroid hormone concentrations, as we percent below detection: 0.38%, 19.77%, 6.65%, and 11.79%, respec- have done in the past (Liberda et al., 2021), and are described be- tively). Brieﬂy, samples were assessed on an Agilent 6890 gas chro- low. Analyses were conducted, and images generated, using v3.5.3 of R matograph equipped with an Agilent G2397A ECD and an Agilent 5973 (R Core Team, 2008) with the RStudio frontend (Rstudio Team, 2019) network mass detector. Full methods and limits of detection for this (v.1.2.1335) and BKMR (Bobb, 2017), tidyverse (Wickham, 2017), fac- analysis have been previously reported by Liberda et al. (2014b). Both toextra (Kassambara and Mundt, 2017), GADMTools (Decorps, 2012), elemental and organic analyses were performed at the Institute Na- and mapdata (Brownrigg, 2018) packages. Values were considered sig- tional De Santé Publique du Québec (INSPQ) which also serves as one niﬁcant when p
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 Table 1 Descriptive data. Percentile Variable N Mean Geometric mean SD. 25th 75th Demographic Data Age 526 39.78 37.51 13.88 29.20 48.00 BMI 526 33.78 33.15 6.56 29.55 37.85 Blood Lipid (g/L) 526 5.25 5.13 1.23 4.49 5.76 Female (%) 307 58% Current Smoker (%) 256 49% Thyroid Hormones TSH (mIU/L) 526 2.15 1.81 1.38 1.29 2.62 Free T4 (pmol/L) 526 15.21 15.07 2.09 13.73 16.40 Total T3 (nmol/L) 526 2.22 2.18 0.49 1.95 2.41 Contaminants Metals Mercury (nmol/L) 526 34.09 14.01 57.97 6.10 39.00 Lead (μmol/L) 526 0.23 0.15 0.25 0.07 0.31 Nickel (nmol/L) 526 20.80 18.24 10.10 16.00 25.00 Cadmium (nmol/L) 526 13.42 7.82 13.35 3.10 20.00 PCBs PCB-118 (μg/L) 526 0.22 0.06 0.44 0.02 0.21 PCB-138 (μg/L) 526 0.58 0.18 1.04 0.05 0.67 PCB-153 (μg/L) 526 1.35 0.41 2.43 0.12 1.40 PCB-156 (μg/L) 526 0.16 0.05 0.30 0.01 0.16 PCB-163 (μg/L) 526 0.22 0.06 0.44 0.02 0.23 PCB-170 (μg/L) 526 0.31 0.09 0.57 0.03 0.33 PCB-180 (μg/L) 526 1.14 0.32 2.10 0.09 1.18 PCB-183 (μg/L) 526 0.11 0.04 0.19 0.01 0.13 PCB-187 (μg/L) 526 0.43 0.12 0.78 0.03 0.47 Organochlorines Mirex (μg/L) 526 0.26 0.06 0.53 0.01 0.24 Oxychlordane (μg/L) 526 0.07 0.03 0.12 0.01 0.08 Trans-nonachlor (μg/L) 526 0.14 0.05 0.24 0.02 0.15 p’p’-DDE (μg/L) 526 2.41 1.18 3.40 0.46 2.68 SD = Standard deviation learning technique, PCA, was performed on the contaminant concen- terior inclusion probabilities (PIPs), examine the univariate exposure- trations which were previously transformed as log10 to improve the outcome relationships with all other exposures ﬁxed to their 50th quan- normality of the data distribution (Gauch, 1982; Green, 1979). To aid tile, and to summarize the overall contaminant mixture eﬀect on total in interpretation, the PCA was varimax rotated. Any non-detects were T3, free T4, and TSH hormones. imputed as 12 the detection limit. The PCA scores generated from the contaminant concentrations were used to explain variation in the thy- roid hormone measures using multivariable linear regression methods. Results It is important to note that using original concentration data as expo- sure variables in linear regression models is not advisable as the corre- Descriptive results lations between related contaminants are often very high – especially in lipophilic contaminants (r=0.90 or larger) (Wainman et al., 2016). All summary statistics are presented in Table 1. More than half Ultimately, this may lead to misleading interpretation of the eﬀects of (58.4%) of the participants were female and the mean age of the par- individual predictor variables due to the multi-collinearity biases of the ticipants was 39.8 (±13.9) years old. Approximately half of the partici- estimation of the individual regression coeﬃcients; the use of PCA rec- pants were current smokers (48.7%) and the mean blood lipid concen- tiﬁes this issue (Altland et al., 1999; Wainman et al., 2016). tration was 5.3 g/L. TSH, free T4, and total T3 mean concentrations were 2.15 mIU/L, 15.21 pmol/L, and 2.18 nmol/L, respectively. All thyroid Bayesian kernel machine regression (BKMR) hormone concentration means were within their respective reference ranges. As expected, body burdens of contaminants varied greatly with The supervised machine learning technique BKMR was used to ex- the highest concentration being observed for p,p’-DDE. amine the association between single contaminant and total mixture ef- fects with thyroid hormone concentration outcomes. Of particular im- portance, BKMR allows for the assessment of non-linear relationships Contaminant PCA Loadings to be evaluated (Bobb et al., 2018, 2014). BKMR utilizes Bayesian and supervised machine learning techniques to iteratively regress response The principal component (PC) loadings that were generated from variables (thyroid hormones) on the environmental contaminant mix- the participant’s contaminant body burdens are presented in Fig. 2. PC- ture while also adjusting for covariates (Chiu et al., 2018). Addition- 1 (74% of variation explained) was highly and positively loaded for all ally, BKMR was designed for cross-sectional studies (Ashrap et al., 2019) PCBs and organochlorines, likely grouped by lipophilicity. The 2nd PC and allows for hierarchical variable selection to ﬁt the model. For this axis (6% of variation explained) was most highly and positively loaded study, the BKMR model was operated with 100,000 machine iterations for nickel and cadmium. PC-3 (11% of variation explained) was highly by a Markov Chain Monte Carlo method with hierarchical selection to and negatively loaded for nickel and highly and positively loaded for group metals, PCBs, and organochlorine contaminants a priori. We uti- cadmium. In total, the three PC axes explained 92% of the original vari- lized BKMR to estimate the group and conditional (within-group) pos- ation in the contaminant body burdens of the participants. 4
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 Fig. 2. PCA loadings of environmental contaminants. Table 2 important group in the model, with oxychlordane being the most im- Principal component analysis regression results. portant contaminant within-group. For total T3, metals has the high- Dependent variable: est group PIP with nickel being the sole contaminant being chosen for Thyroid Hormone model inclusion. For TSH, PCBs were most important group, with PCB- Free T4 Total T3 TSH 118 followed by PCB-163 being the most important congeners within- group. The standardized exposure eﬀect on total T3 is shown in Fig. 3. PC-1 𝛽= 0.005 0.003 0.030∗ Increases in nickel concentration result in an increase in total T3 with C.I.= (-0.001, 0.010) (-0.004, 0.010) (0.005, 0.055) p= 0.086 0.366 0.019 other contaminants ﬁxed at their 50th percentile (Fig. 3A). The overall PC-2 𝛽= -0.003 0.042∗ -0.020 mixture eﬀect results in an increase in total T3 as quantile concentra- C.I.= (-0.016, 0.010) (0.026, 0.058) (-0.081, 0.040) tions increase (Fig. 3B). No signiﬁcant results for free T4 or TSH were p= 0.634 0.00000 0.507 observed (ﬁgures not shown). PC-3 𝛽= -0.001 -0.030∗ -0.029 C.I.= (-0.017, 0.014) (-0.048, -0.011) (-0.100, 0.043) p= 0.870 0.003 0.433 Discussion Observations 526 526 526 Note: ∗ p
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 Table 3 Bayesian Kernel Machine Regression (BKMR) Group and Conditional Posterior Inclusion Criteria (PIPs). Free T4 Total T3 TSH Contaminant Group PIPs Conditional PIPs Group PIPs Conditional PIPs Group PIPs Conditional PIPs Group 1: Metals 0.47 1.0 0.34 Mercury (nmol/L) 0.125 0.0 0.205 Lead (μmol/L) 0.585 0.0 0.260 Nickel (nmol/L) 0.066 1.0 0.179 Cadmium (nmol/L) 0.223 0.0 0.356 Group 2: PCBs 0.627 0.402 0.672 PCB-118 (μg/L) 0.249 0.104 0.229 PCB-138 (μg/L) 0.118 0.107 0.071 PCB-153 (μg/L) 0.074 0.103 0.084 PCB-156 (μg/L) 0.148 0.094 0.126 PCB-163 (μg/L) 0.086 0.135 0.130 PCB-170 (μg/L) 0.093 0.114 0.103 PCB-180 (μg/L) 0.089 0.137 0.095 PCB-183 (μg/L) 0.064 0.121 0.082 PCB-187 (μg/L) 0.077 0.084 0.081 Group 2: Organochlorines 0.849 0.483 0.513 Mirex (μg/L) 0.105 0.205 0.246 Oxychlordane (μg/L) 0.482 0.292 0.247 Transnonachlor (μg/L) 0.257 0.395 0.304 p,p’-DDE (μg/L) 0.156 0.108 0.204 Fig. 3. BKMR results for T3. A. Univariate dose-response model between T3 and scaled ﬁxed contaminant concentrations with credible interval. B. Overall risk model output for T3 in response to varying quantile concentration of entire complex mixture. also shown that nickel may decrease T3 concentrations in rats, not in- airborne particulate matter from mining operations can impact commu- crease it, as we observed (Cheng et al., 1997). Unfortunately, there are nities from the James Bay region (Liberda et al., 2015). Lastly, nickel few human studies assessing the association between, or the eﬀect of, contributions to human body burden may be due to cigarette smoke, nickel and its consequences on thyroid hormone concentrations, which however, our previous analyses found that cadmium, and not nickel, may ultimately play an important role in thyroid disruption. Interest- was associated with this exposure (Charania et al., 2014). ingly, in a murine study, Stangl and Kirchgessner (1998) found that As noted, our PCA analyses also suggested that cadmium was pos- nickel depressed status signiﬁcantly lowered all circulating thyroid hor- itively associated with total T3, a ﬁnding that Chen et al. (2013) also mones (free and total T4, and total T3), and thus it is possible that excess found, except in urinary cadmium. However, our additional BKMR anal- nickel could cause an increase in these hormones – as we observed – but ysis did not support this association. The association between the PC- this has not been conﬁrmed experimentally. 2 axis was most likely being driven by nickel and could be conﬁrmed While both PCA and BKMR analyses highlight the signiﬁcant associa- by the signiﬁcant association with PC-3 clearly highlighting the diﬀer- tion of nickel with total T3, it is important to note that this contaminant ences between nickel and cadmium. BKMR had the power to diﬀeren- is ubiquitous due to its use in stainless steel products (Sterzl et al., 1999). tiate between these two contaminants while PCA required additional While the collection and analyses of blood occurred using non-stainless examination. While the roles of metals in thyroid hormones is not fully steel products, the needle of the syringe was stainless steel. Additionally, understood, the mechanism for the associations observed with nickel cooking using acidic fruits and vegetables may cause more nickel to be or cadmium may be due to alterations in thyroid and adrenal gland released from the cookware contributing further to human body burden function and morphology, metabolism, synthesis, uptake mechanisms, (Christensen and Möller, 1978). It is probable that the nickel mines in or plasma transport eﬀects, and reduced apoptotic activity (Boas et al., this region (Canadian Mining Journal, 2017) may be a responsible for 2006; Brouwer et al., 1998, 1986; Buha et al., 2018; Cheek et al., 1999; the observed body burdens herein and we have previously reported that Rana, 2014). In some cases, the observations may be due to changes in 6
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 the iodothyronine deiodinases, transportation, or synthesis (Chen et al., diﬃcult. However, subsequent BKMR analyses bolstered the PCA results 2013; Gupta et al., 1997; Stangl et al., 1999) in addition to the pre- and provided further insight into the combined eﬀects of the contami- viously noted mechanisms. Thus, it is possible that the observed asso- nant mixtures. ciations of nickel and total T3 may be driven by its role in disturbing the hypothalamic–pituitary–thyroid axis, but the exact mechanism re- Conclusions mains to be elucidated. As this is a cross-sectional study, we cannot imply causality and therefore we report associations which should be We have previously reported on the iodine status from the com- veriﬁed through experimental methods. munities of the Eeyou Istchee territory of eastern James Bay (Quebec, PCA regression showed that PC-1 was signiﬁcantly and positively Canada) and have found that iodine status is suﬃcient at the population associated with TSH. Given that PC-1 is highly loaded for PCBs and level (Tam et al., 2015). However, it is possible that contaminants may organochlorines, this ﬁnding is not a surprise as other studies have found still play a role in thyroid disorder morbidity via a variety of diﬀerent this same correlation. The mechanisms by which various organochlo- mechanisms including receptor interference, thyroid antibody genera- rines and PCBs are causing the alterations in thyroid hormone concen- tion, and competitive binding (Boas et al., 2012; Sarkar et al., 2015). trations may not yet be fully elucidated, however, many of these mech- We have shown that supervised (BKMR) and unsupervised (PCA) ma- anisms previously noted have also been postulated for organic contam- chine learning techniques both identiﬁed nickel as being signiﬁcantly inants (Brouwer et al., 1998), and therefore the complexity of chemical and positively associated with total T3. While PCA analyses identiﬁed mixtures further complicate the understanding of experimental eﬀects PCBs and organochlorines as being signiﬁcantly and positively associ- and associations observed with singular contaminant exposures. More ated with TSH, BKMR analysis did not support this signiﬁcant associ- recently, an in vivo investigation into the mechanisms of DDE found that ation, but did suggest that PCBs were more important in positively al- may be exerting its thyroid disrupting eﬀects by altering genes related tering TSH levels than organochlorines in participants from this study. the hypothalamic pituitary-thyroid axis, speciﬁcally those related to thy- Taken together, nickel, and possibly PCBs, may play a role in modify- roid hormone synthesis and development (Wu et al., 2019). Marsan and ing the thyroid hormone milieu, but competing eﬀects appear to temper Bayse (2020) postulated that PCBs may exert their eﬀects via halogen the manifested outcomes due to contaminant mixture interactions. Im- bonding to the iodothyronine deiodinases resulting in thyroid hormone portantly, we note that while we have identiﬁed contributing contam- modulation (competitive inhibition) and that this interaction appears to inants that may alter thyroid concentrations, none of our participants be driven by the location of the chlorine groups on the molecules. Re- had thyroid hormone concentrations outside of the reference range. Fu- lated to this, Bai et al. (2018) reported that some PCB congeners appear ture studies should examine the role of nickel and PCBs using murine to act as thyroid receptor agonists while others as antagonists and this models to elucidate their mechanisms of action. This article represents could explain the varying ﬁndings between studies and their eﬀects on a signiﬁcant step towards assessing complex mixtures of organic and thyroid hormones. BKMR analysis did not support the signiﬁcant ﬁnd- metal xenobiotics using novel machine learning techniques by evaluat- ing between PCBs and organochlorines with TSH but did suggest that ing their univariate and complete mixture eﬀects on thyroid hormone PCBs were more important than organochlorines in contributing to the levels as a whole, rather than by single or small groups of similar con- association observed via the PCA. This observation may have occurred taminants. This article also highlights the importance of exposures to due to the ﬂexible and highly iterative and non-linear approach BKMR nickel and PCBs, as these may be more relevant to persons living a sub- uses, albeit at a signiﬁcant computational cost, or, simply due to the sistence lifestyle, and hence, may be at greater risk than those who are interaction of antagonistic and agonistic PCB congeners. minimally exposed. Body burdens of organic and elemental contaminants were greater than their non-Indigenous Canadian counterparts who were over 20 Credit Authorship Contribution Statement years of age (Health Canada, 2019, 2010). Speciﬁcally, this study found organic contaminant concentrations to be higher than Dene/Metis Eric N. Liberda: Conceptualization, Methodology, Validation, For- persons, but generally lower than those reported from both the mal analysis, Data curation, Writing - original draft, Writing - review & east and west coast of James Bay, Canada (Health Canada, 2010; editing, Visualization. Aleksandra M. Zuk: Conceptualization, Method- Liberda et al., 2014b; Philibert et al., 2009; Tsuji et al., 2005; ology, Validation, Formal analysis, Data curation, Writing - review & Walker and Van Oostam, 2001). With respect to elemental con- editing. David S. Di: Data analysis, interpretation, and editing. Robert taminants, this cohort had higher concentrations of blood cadmium J Moriarity: Data analysis, interpretation and editing. Ian D. Martin: (34.09 nmol/L vs. 13.5 nmol/L), nickel (20.8 nmol/L vs. 16.1 nmol/L), Conceptualization and data analysis, Leonard J. Tsuji: Conceptualiza- mercury (34.09 nmol/L vs. 24.0 nmol/L) and lead (0.23 μmol/L vs. tion, Methodology, Writing - review & editing. 0.1456 μmol/L) compared to a previous publication from this region (Nieboer et al., 2017). However, none of these mean values exceeded health guidelines with the exception of Cadmium, which we have pre- Funding viously shown was caused by smoking (Charania et al., 2014). While the ﬁndings of this analysis highlight the power of machine This scientiﬁc communication is a report from the Nituuchischaay- learning to identify associations with various dependent variables, this ihtitaau Aschii: Multi-Community Environment-and Health Study in article is not without limitations. First, the study is cross-sectional and Eeyou Istchee supported by the Cree People of Northern Québec, the hence causality and temporality cannot be implied. Second, although Cree First Nations and the Cree Board of Health and Social Services we have adjusted for various lifestyle eﬀects (e.g., smoking status), it of James Bay through ﬁnancial contributions from Niskamoon Corpo- is possible that unmeasured covariates such as time spent on the land ration. Additional funding for analyses was provided by the Institute (traditional activities) could contribute to the observed eﬀects. Third, of Indigenous Peoples’ Health, Canadian Institutes of Health Research while we adjusted for current smoking status, the data collected did (Grant #156396). The authors thank the participants from the Eeyou not record pack-years of cigarettes and hence could contribute to resid- Istchee territory in Quebec, Canada. ual confounding. Fourth, this study measures free (unbound free hor- mone, biologically active) T4, total (bound and unbound forms) T3, and Declaration of Competing Interest TSH, however associations with total T4 or free T3, for example, may have been present, but would be less clinically relevant and/or reliable. The authors declare that they have no known competing ﬁnancial Lastly, as PCA is a linear combination of original contaminant variables interests or personal relationships that could have appeared to inﬂuence with projected orthogonal axes, interpretation of results is sometimes the work reported in this paper. 7
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 References cadmium sources among the nine First Nations of Eeyou Istchee, northern Quebec, Canada. Environ. Sci. Process. Impacts 16. doi:10.1039/c4em00064a. Altland, H.W., Freund, R.J., Wilson, W.J., 1999. Technometrics. Academic Press Cheek, A.O., Kow, K., Chen, J., McLachlan, J.A., 1999. Potential mechanisms of thyroid doi:10.2307/1271353. disruption in humans: interaction of organochlorine compounds with thyroid recep- Alvarez-Pedrerol, M., Ribas-Fitó, N., Torrent, M., Carrizo, D., Grimalt, J.O., Sun- tor, transthyretin, and thyroid-binding globulin. Environ. Health Perspect. 107, 273– yer, J., 2008. Eﬀects of PCBs, p,p’-DDT, p,p’-DDE, HCB and beta-HCH on 278. doi:10.1289/ehp.99107273. thyroid function in preschool children. Occup. Environ. Med. 65, 452–457. Chen, A., Kim, S.S., Chung, E., Dietrich, K.N., 2013. Thyroid hormones in rela- doi:10.1136/oem.2007.032763. tion to lead, mercury, and cadmium exposure in the National Health and Nu- Ashrap, P., Sánchez, B.N., Téllez-Rojo, M.M., Basu, N., Tamayo-Ortiz, M., Peterson, K.E., trition Examination Survey, 2007-2008. Environ. Health Perspect. 121, 181–186. Meeker, J.D., Watkins, D.J., 2019. In utero and peripubertal metals exposure in rela- doi:10.1289/ehp.1205239. tion to reproductive hormones and sexual maturation and progression among girls in Cheng, W., Zhou, Z., Yang, Y., Yin, C., Chai, G.W., 1997. Eﬀect of nickel sulfate on the Mexico City. Environ. Res. 177. doi:10.1016/j.envres.2019.108630. concentrations of T3, T4 and TSH in serum of rat. Wei Sheng Yan Jiu 26, 53–59. Baccarelli, A., Giacomini, S.M., Corbetta, C., Landi, M.T., Bonzini, M., Consonni, D., Chiu, Y.H., Bellavia, A., James-Todd, T., Correia, K.F., Valeri, L., Messerlian, C., Ford, J.B., Grillo, P., Patterson, D.G., Pesatori, A.C., Bertazzi, P.A., 2008. Neonatal thyroid func- Mínguez-Alarcón, L., Calafat, A.M., Hauser, R., Williams, P.L., 2018. Evaluating eﬀects tion in seveso 25 years after maternal exposure to dioxin. PLoS Med. 5, 1133–1142. of prenatal exposure to phthalate mixtures on birth weight: a comparison of three doi:10.1371/journal.pmed.0050161. statistical approaches. Environ. Int. doi:10.1016/j.envint.2018.02.005. Bai, X., Yan, L., Ji, C., Zhang, Q., Dong, X., Chen, A., Zhao, M., 2018. A combination Christensen, O.B., Möller, H., 1978. Release of nickel from cooking utensils. Contact Der- of ternary classiﬁcation models and reporter gene assays for the comprehensive thy- matitis doi:10.1111/j.1600-0536.1978.tb03847.x. roid hormone disruption proﬁles of 209 polychlorinated biphenyls. Chemosphere 210, Curtis, S.W., Terrell, M.L., Jacobson, M.H., Cobb, D.O., Jiang, V.S., Neblett, M.F., Gerkow- 312–319. doi:10.1016/j.chemosphere.2018.07.023. icz, S.A., Spencer, J.B., Marder, M.E., Barr, D.B., Conneely, K.N., Smith, A.K., Mar- Berg, V., Nøst, T.H., Hansen, S., Elverland, A., Veyhe, A.S., Jorde, R., Odland, J.Ø., Sandan- cus, M., 2019. Thyroid hormone levels associate with exposure to polychlorinated ger, T.M., 2015. Assessing the relationship between perﬂuoroalkyl substances, thyroid biphenyls and polybrominated biphenyls in adults exposed as children. Environ. Heal. hormones and binding proteins in pregnant women; a longitudinal mixed eﬀects ap- A Glob. Access Sci. Source 18 (75). doi:10.1186/s12940-019-0509-z. proach. Environ. Int. 77, 63–69. doi:10.1016/j.envint.2015.01.007. Dallaire, R., Muckle, G., Dewailly, E., Jacobson, S.W., Jacobson, J.L., Sandanger, T.M., Berg, V., Nøst, T.H., Pettersen, R.D., Hansen, S., Veyhe, A.S., Jorde, R., Odland, J.Ø., Sandau, C.D., Ayotte, P., 2009. Thyroid hormone levels of pregnant inuit women and Sandanger, T.M., 2017. Persistent organic pollutants and the association with ma- their infants exposed to environmental contaminants. Environ. Health Perspect. 117, ternal and infant thyroid homeostasis: a multipollutant assessment. Environ. Health 1014–1020. doi:10.1289/ehp.0800219. Perspect. 125, 127–133. doi:10.1289/EHP152. Damstra, T., Barlow, S., Bergman, A., Kavlock, R., Van Der Kraak, G., 2002. Global As- Bergman, Å., Heindel, J.J., Kasten, T., Kidd, K.A., Jobling, S., Neira, M., Zoeller, R.T., sessment of the State-of-the-Science of Endocrine Disruptors. WHOpublication no. Becher, G., Bjerregaard, P., Bornman, R., Brandt, I., Kortenkamp, A., Muir, D., WHO/PCS/EDC/02.2 180 doi:10.1089/15305620252933437. Drisse, M.-N.B., Ochieng, R., Skakkebaek, N.E., Byléhn, A.S., Iguchi, T., Top- Decorps, J.P., 2012. Package “GADMTools.” GADMTools. R Packag version 2.3.0. pari, J., Woodruﬀ, T.J., 2013. The impact of endocrine disruption: a consensus Dundar, B., Öktem, F., Arslan, M.K., Delibas, N., Baykal, B., Arslan, Ç., Gultepe, M., Il- statement on the state of the science. Environ. Health Perspect. 121, a104–a106. han, I.E., 2006. The eﬀect of long-term low-dose lead exposure on thyroid function in doi:10.1289/ehp.1205448. adolescents. Environ. Res. 101, 140–145. doi:10.1016/j.envres.2005.10.002. Blanco-Muñoz, J., Lacasaña, M., López-Flores, I., Rodríguez-Barranco, M., González- Duntas, L.H., Stathatos, N., 2015. Toxic chemicals and thyroid function: hard Alzaga, B., Bassol, S., Cebrian, M.E., López-Carrillo, L., Aguilar-Garduño, C., 2016. facts and lateral thinking. Rev. Endocr. Metab. Disord. 16, 311–318. Association between organochlorine pesticide exposure and thyroid hormones in ﬂori- doi:10.1007/s11154-016-9331-x. culture workers. Environ. Res. 150, 357–363. doi:10.1016/j.envres.2016.05.054. Freire, C., Koifman, R.J., Sarcinelli, P., Rosa, A.C., Clapauch, R., Koifman, S., 2012. Boas, M., Feldt-Rasmussen, U., Main, K.M., 2012. Thyroid eﬀects of endocrine disrupting Long term exposure to organochlorine pesticides and thyroid function in chil- chemicals. Mol. Cell. Endocrinol. doi:10.1016/j.mce.2011.09.005. dren from Cidade dos Meninos, Rio de Janeiro. Brazil. Environ. Res. 117, 68–74. Boas, M., Feldt-Rasmussen, U., Skakkebæk, N.E., Main, K.M., 2006. Environmental chem- doi:10.1016/j.envres.2012.06.009. icals and thyroid function. Eur. J. Endocrinol. doi:10.1530/eje.1.02128. Freire, C., Koifman, R.J., Sarcinelli, P.N., Simões Rosa, A.C., Clapauch, R., Koif- Bobb, J.F., 2017. Package “BKMR”. Bayesian Kernel Machine Regression R package ver- man, S., 2013. Long-term exposure to organochlorine pesticides and thyroid sta- sion 0.2.0. tus in adults in a heavily contaminated area in Brazil. Environ. Res. 127, 7–15. Bobb, J.F., Claus Henn, B., Valeri, L., Coull, B.A., 2018. Statistical software for an- doi:10.1016/j.envres.2013.09.001. alyzing the health eﬀects of multiple concurrent exposures via Bayesian kernel Gauch, H.G., 1982. Multivariate Analysis in Community Ecology. Cambridge University machine regression. Environ. Heal. A Glob. Access Sci. Source Press, Cambridge doi:10.2307/2259746. doi:10.1186/s12940-018-0413-y. Gesing, A., Lewiński, A., Karbownik-Lewińska, M., 2012. The thyroid gland and the pro- Bobb, J.F., Valeri, L., Claus Henn, B., Christiani, D.C., Wright, R.O., Mazumdar, M., cess of aging; What is new? Thyroid Res. doi:10.1186/1756-6614-5-16. Godleski, J.J., Coull, B.A., 2014. Bayesian kernel machine regression for estimat- Ghisari, M., Long, M., Tabbo, A., Bonefeld-Jørgensen, E.C., 2015. Eﬀects of currently ing the health eﬀects of multi-pollutant mixtures. Biostatistics doi:10.1093/biostatis- used pesticides and their mixtures on the function of thyroid hormone and aryl tics/kxu058. hydrocarbon receptor in cell culture. Toxicol. Appl. Pharmacol. 284, 292–303. Bonnier-Viger, Y., Dewailly, E., Eageland, G., Nieboer, E., Pereg, D., 2007. Nituuchischaay- doi:10.1016/j.taap.2015.02.004. ihtitaau ASCHII Multi-community Environment-and-Health Study in Iiyiyiu Aschii. Gouin, T., MacKay, D., Jones, K.C., Harner, T., Meijer, S.N., 2004. Evidence for the Mistissini. “grasshopper” eﬀect and fractionation during long-range atmospheric transport of or- Bremner, A.P., Feddema, P., Leedman, P.J., Brown, S.J., Beilby, J.P., Lim, E.M., Wil- ganic contaminants, in: Environ. Pollut. pp. 139–148. 10.1016/j.envpol.2003.08.025 son, S.G., O’Leary, P.C., Walsh, J.P., 2012. Age-related changes in thyroid function: Green, R.H., 1979. Sampling Design and Statistical Methods for Environmental Biologists. a longitudinal study of a community-based cohort. J. Clin. Endocrinol. Metab. 97, Wiley. 1554–1562. doi:10.1210/jc.2011-3020. Grønnestad, R., Villanger, G.D., Polder, A., Kovacs, K.M., Lydersen, C., Jenssen, B.M., Brouwer, A., Morse, D.C., Lans, M.C., Gerlienke Schuur, A., Murk, A.J., Klasson- Borgå, K., 2018. Eﬀects of a complex contaminant mixture on thyroid hormones Wehler, E., Bergman, Å., Visser, T.J., 1998. Interactions of persistent environ- in breeding hooded seal mothers and their pups. Environ. Pollut. 240, 10–16. mental organohalogens with the thyroid hormone system: mechanisms and possi- doi:10.1016/j.envpol.2018.04.052. ble consequences for animal and human health. Toxicol. Ind. Health 14, 59–84. Guo, J., Lv, N., Tang, J., Zhang, X., Peng, L., Du, X., Li, S., Luo, Q., Zhang, D., doi:10.1177/074823379801400107. Chen, G., 2018. Associations of blood metal exposure with thyroid hormones in Brouwer, A., van den Berg, K.J., Blaner, W.S., Goodman, D.S., 1986. Transthyretin Chinese pregnant women: a cross-sectional study. Environ. Int. 121, 1185–1192. (prealbumin) binding of PCBS, a model for the mechanism of interference doi:10.1016/j.envint.2018.10.038. with vitamin A and thyroid hormone metabolism. Chemosphere 15, 1699–1706. Gupta, P., Chaurasia, S.S., Maiti, P.K., Kar, A., 1997. Cadmium induced alterations in doi:10.1016/0045-6535(86)90457-1. extrathyroidal conversion of thyroxine to triiodothyronine by type-I iodothyronine Brownrigg, R., 2018. Package ‘mapdata.’ mapdata Extra Map Databases. R Packag version 5’-monodeiodinase in male mouse. Horm. Metab. Res. doi:10.1055/s-2007-979009. 2.3.0. Health Canada, 2019. Fifth Report on Human Biomonitoring of Environmental Chemicals Buha, A., Matovic, V., Antonijevic, B., Bulat, Z., Curcic, M., Renieri, E.A., Tsatsakis, A.M., in Canada - Canada.ca. Schweitzer, A., Wallace, D., 2018. Overview of cadmium thyroid disrupting eﬀects Health Canada, 2010. Report on Human Biomonitoring of Environmental Chemicals in and mechanisms. Int. J. Mol. Sci. 19. doi:10.3390/ijms19051501. Canada-Results of the Canadian Health Measures Survey Cycle 1 (2007–2009). Canadian Mining Journal, 2017. NICKEL-COPPER: Midland, Altius discover new Jacobson, M.H., Darrow, L.A., Barr, D.B., Howards, P.P., Lyles, R.H., Terrell, M.L., showing near James Bay [WWW Document]. Can. Min. J. URL https://www. Smith, A.K., Conneely, K.N., Marder, M.E., Marcus, M., 2017. Serum Polybrominated canadianminingjournal.com/news/nickel-copper-midland-altius-discover-new- Biphenyls (PBBs) and Polychlorinated Biphenyls (PCBs) and Thyroid Function among showing-near-james-bay/ (accessed 2.19.21). Michigan Adults Several Decades after the 1973–1974 PBB Contamination of Live- Castiello, F., Olmedo, P., Gil, F., Molina, M., Mundo, A., Romero, R.R., Ruíz, C., Gómez- stock Feed. Environ. Health Perspect. 125. doi:10.1289/EHP1302. Vida, J., Vela-Soria, F., Freire, C., 2020. Association of urinary metal concentrations Jain, R.B., 2016. Association between arsenic exposure and thyroid function: with blood pressure and serum hormones in Spanish male adolescents. Environ. Res. data from NHANES 2007-2010. Int. J. Environ. Health Res. 26, 101–129. 182. doi:10.1016/j.envres.2019.108958. doi:10.1080/09603123.2015.1061111. Chaker, L., Bianco, A.C., Jonklaas, J., Peeters, R.P., 2017. Hypothyroidism. Lancet Jain, R.B., 2013. Association between thyroid proﬁle and perﬂuoroalkyl acids: data from doi:10.1016/S0140-6736(17)30703-1. NHNAES 2007–2008. Environ. Res. 126, 51–59. doi:10.1016/j.envres.2013.08.006. Charania, N.A., Tsuji, L.J.S., Martin, I.D., Liberda, E.N., Coté, S., Ayotte, P., Dewailly, E., Jain, R.B., Choi, Y.S., 2016. Interacting eﬀects of selected trace and toxic Nieboer, E., 2014. An examination of traditional foods and cigarette smoking as metals on thyroid function. Int. J. Environ. Health Res. 26, 75–91. doi:10.1080/09603123.2015.1020416. 8
E.N. Liberda, A.M. Zuk, D.S. Di et al. Environmental Advances 4 (2021) 100054 Kassambara, A., Mundt, F., 2017. Package “factoextra”. Extract and Visualize the Results Préau, L., Fini, J.B., Morvan-Dubois, G., Demeneix, B., 2015. Thyroid hormone of Multivariate Data Analyses. R Packag version 1.0.6. signaling during early neurogenesis and its signiﬁcance as a vulnerable win- Klein, I., Ojamaa, K., 2001. Thyroid hormone and the cardiovascular system. N. Engl. J. dow for endocrine disruption. Biochim. Biophys. Acta - Gene Regul. Mech. Med. 344, 501–509. doi:10.1056/NEJM200102153440707. doi:10.1016/j.bbagrm.2014.06.015. Levesque, H.M., Dorval, J., Hontela, A., Van Der Kraak, G.J., Campbell, P.G.C., 2003. Hor- R Core Team, 2008. R Foundation for Statistical Computing. Vienna, Asutria. monal, morphological, and physiological responses of yellow perch (perca ﬂavescens) Rana, S.V.S., 2014. Perspectives in endocrine toxicity of heavy metals–a review. Biol. to chronic environmental metal exposures. J. Toxicol. Environ. Heal. - Part A. Trace Elem. Res. 160, 1–14. doi:10.1007/s12011-014-0023-7. doi:10.1080/15287390309353773. Rstudio Team, 2019. RStudio: Integrated Development for R. RStudio, Inc., Boston MA. Li, Y., Cheng, Y., Xie, Z., Zeng, F., 2017. Perﬂuorinated alkyl substances in serum of the RStudio. southern Chinese general population and potential impact on thyroid hormones. Sci. Rylander, L., Björkdahl, C.-M., Axmon, A., Giwercman, A., Jönsson, B.A.G., Lindh, C., Rep. 7, 43380. doi:10.1038/srep43380. Rignell-Hydbom, A., 2012. Very high correlations between fresh weight and Liberda, E.N., Tsuji, L.J.S., Martin, D.I.D., Ayotte, P., Dewailly, E., Nieboer, E., 2014a. The lipid-adjusted PCB-153 serum concentrations: irrespective of fasting status, age, complexity of hair/blood mercury concentration ratios and its implications. Environ. body mass index, gender, or exposure distributions. Chemosphere 88, 828–831. Res. 134, 286–294. doi:10.1016/j.envres.2014.08.007. doi:10.1016/J.CHEMOSPHERE.2012.03.089. Liberda, E.N., Tsuji, L.J.S., Martin, I.D., Cote, S., Ayotte, P., Dewailly, E., Nieboer, E., Sarkar, A., Knight, J.C., Babichuk, N.A., Mulay, S., 2015. Skewed distribution of hypothy- 2014b. Plasma concentrations of persistent organic pollutants in the Cree of northern roidism in the coastal communities of Newfoundland, Canada. Environ. Int. 83, 171– Quebec, Canada: Results from the multi-community environment-and-health study. 175. doi:10.1016/j.envint.2015.05.017. Sci. Total Environ. 470–471. doi:10.1016/j.scitotenv.2013.10.048. Schell, L.M., Gallo, M.V., Denham, M., Ravenscroft, J., DeCaprio, A.P., Carpenter, D.O., Liberda, E.N., Tsuji, L.J.S., Peltier, R.E., 2015. Mining in subarctic Canada: Airborne 2008. Relationship of thyroid hormone levels to levels of polychlorinated biphenyls, PM2.5metal concentrations in two remote First Nations communities. Lead, p,p′- DDE, and other toxicants in Akwesasne Mohawk Youth. Environ. Health Chemosphere 139. doi:10.1016/j.chemosphere.2015.07.058. Perspect. 116, 806–813. doi:10.1289/ehp.10490. Liberda, E.N., Zuk, A.M., Tsuji, L.J.S., 2021. Heart rate variation and human body burdens Stangl, G.I., Kirchgessner, M., 1998. Comparative eﬀects of nickel and iron depletion on of environmental mixtures in the Cree First Nation communities of Eeyou Istchee. circulating thyroid hormone concentrations in rats. J. Anim. Physiol. Anim. Nutr. 79, Canada. Environ. Int. 146. doi:10.1016/j.envint.2020.106220. 18–26. doi:10.1111/j.1439-0396.1998.tb00625.x. Liberda, E.N., Tsuji, L.J.S., Martin, I.D., Ayotte, P., Robinson, E., Dewailly, E., Nieboer, E., Stangl, G.I., Schwarz, F.J., Kirchgessner, M., 1999. Cobalt deﬁciency eﬀects on trace ele- 2018. Source identiﬁcation of human exposure to lead in nine Cree Nations ments, hormones and enzymes involved in energy metabolism of cattle. Int. J. Vitam. from Quebec, Canada (Eeyou Istchee territory). Environ. Res. 161, 409–417. Nutr. Res. 69, 120–126. doi:10.1024/0300-9831.69.2.120. doi:10.1016/j.envres.2017.11.023. Sterzl, I., Procházková, J., Hrdá, P., Bártová, J., Matucha, P., Vera, &, Stejskal, D.M., Liu, Y., Chen, H., Zhang, L., Zhang, T., Ren, X., 2020. The association between thyroid Stejskal, V., Dm Stejskal, V., 1999. Mercury and nickel allergy: risk factors in fatigue injury and apoptosis, and Alterations of Bax, Bcl-2, and Caspase-3 mRNA/Protein and autoimmunity. Neuroendocrinol. Lett. 20, 221–228. expression induced by nickel sulfate in Wistar rats. Biol. Trace Elem. Res. 195, 159– Sutcliﬀe, C., Harvey, P.W., 2015. Endocrine Disruption and Human Health. Endocrine 168. doi:10.1007/s12011-019-01825-0. Disruption and Human Health doi:10.1016/B978-0-12-801139-3.00011-9. Luo, J., Hendryx, M., 2014. Relationship between blood cadmium, lead, and serum Tam, B., Tsuji, L.J.S., Martin, I.D., Liberda, E.N., Ayotte, P., Coté, S., Dewailly, É., thyroid measures in US adults - The National Health and Nutrition Examina- Nieboer, E., 2015. Iodine status of Eeyou Istchee community members of northern tion Survey (NHANES) 2007-2010. Int. J. Environ. Health Res. 24, 125–136. Quebec, Canada, and potential sources. Environ. Sci. Process. Impacts 17, 844–853. doi:10.1080/09603123.2013.800962. doi:10.1039/c4em00674g. Marsan, E.S., Bayse, C.A., 2020. Halogen bonding interactions of polychlorinated Taxvig, C., Hadrup, N., Boberg, J., Axelstad, M., Bossi, R., Bonefeld-Jørgensen, E.C., biphenyls and the potential for thyroid disruption. Chem. - A Eur. J. 26, 5200–5207. Vinggaard, A.M., 2013. In vitro - in vivo correlations for endocrine activity of doi:10.1002/chem.201903904. a mixture of currently used pesticides. Toxicol. Appl. Pharmacol. 272, 757–766. Meeker, J.D., Altshul, L., Hauser, R., 2007. Serum PCBs, p,p’-DDE and HCB doi:10.1016/j.taap.2013.07.028. predict thyroid hormone levels in men. Environ. Res. 104, 296–304. Tsuji, L.J.S., Wainman, B.C., Martin, I.D., Weber, J.-P., Sutherland, C., Elliott, J.R., doi:10.1016/j.envres.2006.11.007. Nieboer, E., 2005. The Mid-Canada Radar Line and First Nations’ People of the James Meeker, J.D., Rossano, M.G., Protas, B., Diamond, M.P., Puscheck, E., Daly, D., Paneth, N., Bay region, Canada: an evaluation using log-linear contingency modelling to analyze Wirth, J.J., 2009. Multiple metals predict prolactin and thyrotropin (TSH) levels in organochlorine frequency data. J. Environ. Monit. 7 (888). doi:10.1039/b500524h. men. Environ. Res. 109, 869–873. doi:10.1016/j.envres.2009.06.004. Wainman, B.C., Kesner, J.S., Martin, I.D., Meadows, J.W., Krieg, E.F., Nieboer, E., Mnif, W., Hassine, A.I.H., Bouaziz, A., Bartegi, A., Thomas, O., Roig, B., 2011. Eﬀect of Tsuji, L.J., 2016. Menstrual cycle perturbation by organohalogens and el- endocrine disruptor pesticides: a review. Int. J. Environ. Res. Public Health 8, 2265– ements in the Cree of James Bay. Canada. Chemosphere 149, 190–201. 2303. doi:10.3390/ijerph8062265. doi:10.1016/j.chemosphere.2015.12.056. Nie, X., Chen, Yi, Chen, Yingchao, Chen, C., Han, B., Li, Q., Zhu, C., Xia, F., Zhai, H., Walker, J., Van Oostam, J., 2001. Human Contaminant Trends in Arctic Canada - North- Wang, N., Lu, Y., 2017. Lead and cadmium exposure, higher thyroid antibod- west Territories and Nunavut Environmental Contaminants Exposure Baseline: Final ies and thyroid dysfunction in Chinese women. Environ. Pollut. 230, 320–328. Technical Report. Department of Health and Social Services, Government of North- doi:10.1016/j.envpol.2017.06.052. west Territories. Nieboer, E., Dewailly, E., Johnson-Down, L., Sampasa-Kanyinga, H., Château-Degat, M., Wania, F., Mackay, D., 1996. Tracking the distribution of persistent organic pollutants. Egeland, G., Atikessé, L., Robinson, E., Torrie, J., 2013. Nituuchischaayihti- Environ. Sci. Technol. doi:10.1016/s0926-3373(97)80026-4. taau Aschii Multi-community environment-and-health study in Eeyou Istchee Wickham, H., 2017. Package “tidyverse”. Easily Install and Load the “Tidyverse.”. R 2005- 2009: Final Technical Report [WWW Document]. Public Heal. Rep. Ser. Packag version 1.2.1. 4 Heal. Popul. Chisasibi QC Cree Board Heal. Soc. Serv. James Ba. URL Wu, L., Ru, H., Ni, Z., Zhang, X., Xie, H., Yao, F., Zhang, H., Li, Y., Zhong, L., 2019. Com- http://www.creehealth.org/sites/default/ﬁles/E-and-H . Technical Report.pdf (ac- parative thyroid disruption by o,p’-DDT and p,p’-DDE in zebraﬁsh embryos/larvae. cessed 9.2.19). Aquat. Toxicol. 216. doi:10.1016/j.aquatox.2019.105280. Nieboer, E., Martin, I.D., Liberda, E.N., Dewailly, E., Robinson, E., Tsuji, L.J.S., 2017. Body Xu, X., Liu, J., Zeng, X., Lu, F., Chen, A., Huo, X., 2014. Elevated serum polybrominated burdens, sources and interrelations of selected toxic and essential elements among diphenyl ethers and alteration of thyroid hormones in children from Guiyu, China. the nine Cree First Nations of: Eeyou Istchee, James Bay region of northern Quebec, PLoS One 9. doi:10.1371/journal.pone.0113699. Canada. Environ. Sci. Process. Impacts 19, 727–741. doi:10.1039/c7em00052a. Yorita Christensen, K.L., 2013. Metals in blood and urine, and thyroid function among Philibert, A., Schwartz, H., Mergler, D., 2009. An exploratory study of diabetes in adults in the United States 2007-2008. Int. J. Hyg. Environ. Health 216, 624–632. a ﬁrst nation community with respect to serum concentrations of p,p’-DDE and doi:10.1016/j.ijheh.2012.08.005. PCBs and ﬁsh consumption. Int. J. Environ. Res. Public Health 6, 3179–3189. Zoeller, R.T., 2007. Environmental chemicals impacting the thyroid: targets and conse- doi:10.3390/ijerph6123179. quences. Thyroid 17, 811–817. doi:10.1089/thy.2007.0107. 9