Portal de Periódicos da CAPES

Associations between Exposure to Organochlorine Chemicals and Endometriosis: A Systematic Review of Experimental Studies and Integration of Epidemiological Evidence

2021; National Institute of Environmental Health Sciences; Volume: 129; Issue: 7 Linguagem: Inglês

10.1289/ehp8421

1552-9924

Komodo Matta, Meriem Koual, Stéphane Ploteau, Xavier Coumoul, Karine Audouze, Bruno Le Bizec, Jean‐Philippe Antignac, Germán Cano-Sancho,

Reproductive System and Pregnancy

Vol. 129, No. 7 ReviewOpen AccessAssociations between Exposure to Organochlorine Chemicals and Endometriosis: A Systematic Review of Experimental Studies and Integration of Epidemiological Evidence Komodo Matta, Meriem Koual, Stéphane Ploteau, Xavier Coumoul, Karine Audouze, Bruno Le Bizec, Jean-Philippe Antignac, and German Cano-Sancho Komodo Matta Oniris, INRAE, UMR 1329 Laboratoire d'Étude des Résidus et Contaminants dans les Aliments (LABERCA), Nantes, France , Meriem Koual Université de Paris, T3S, Institut national de la santé et de la recherche médicale (Inserm) UMR S-1124, Paris, France Service de Chirurgie Cancérologique Gynécologique et du Sein, Assistance Publique-Hôpitaux de Paris, Hôpital Européen Georges-Pompidou, Paris, France , Stéphane Ploteau Service de gynécologie-obstétrique, Centre d'investigation clinique–Femme Enfant Adolescent, Hôpital Mère Enfant, Centre Hospitalier Universitaire Hôtel Dieu, Nantes, France , Xavier Coumoul Université de Paris, T3S, Institut national de la santé et de la recherche médicale (Inserm) UMR S-1124, Paris, France , Karine Audouze Université de Paris, T3S, Institut national de la santé et de la recherche médicale (Inserm) UMR S-1124, Paris, France , Bruno Le Bizec Oniris, INRAE, UMR 1329 Laboratoire d'Étude des Résidus et Contaminants dans les Aliments (LABERCA), Nantes, France , Jean-Philippe Antignac Oniris, INRAE, UMR 1329 Laboratoire d'Étude des Résidus et Contaminants dans les Aliments (LABERCA), Nantes, France , and German Cano-Sancho Address correspondence to German Cano-Sancho, Oniris, Institut National de Recherche pour l'Agriculture, l'Alimentation et l'Environnement, Laboratoire d'Étude des Résidus et Contaminants dans les Aliments, Route de Gachet–La Chantrerie 44307, Nantes, France. Telephone: +33 240687880. Email: E-mail Address: [email protected] Oniris, INRAE, UMR 1329 Laboratoire d'Étude des Résidus et Contaminants dans les Aliments (LABERCA), Nantes, France Published:26 July 2021CID: 076003https://doi.org/10.1289/EHP8421AboutSectionsPDF Supplemental Materials ToolsDownload CitationsTrack Citations ShareShare onFacebookTwitterLinked InRedditEmail AbstractBackground:Growing epidemiological evidence suggests that organochlorine chemicals (OCCs), including 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), may play a role in the pathogenesis of endometriosis.Objectives:We aimed to systematically review the experimental evidence (in vivo and in vitro) on the associations between exposure to OCCs and endometriosis-related end points.Methods:A systematic review protocol was developed following the National Toxicology Program /Office of Health Assessment and Translation (NTP/OHAT) framework and managed within a web-based interface. In vivo studies designed to evaluate the impact of OCCs on the onset or progression of endometriosis and proliferation of induced endometriotic lesions were eligible. Eligible in vitro studies included single-cell and co-culture models to evaluate the proliferation, migration, and/or invasion of endometrial cells. We applied the search strings to PubMed, Web of Science, and Scopus®. A final search was performed on 24 June 2020. Assessment of risk of bias and the level of evidence and integration of preevaluated epidemiological evidence was conducted using NTP/OHAT frameworkResults:Out of 812 total studies, 39 met the predetermined eligibility criteria (15 in vivo, 23 in vitro, and 1 both). Most studies (n=27) tested TCDD and other dioxin-like chemicals. In vivo evidence supported TCDD's promotion of endometriosis onset and lesion growth. In vitro evidence supported TCDD's promotion of cell migration and invasion, but there was insufficient evidence for cell proliferation. In vitro evidence further supported the roles of the aryl hydrocarbon receptor and matrix metalloproteinases in mediating steroidogenic disruption and inflammatory responses. Estrogen interactions were found across studies and end points.Conclusion:Based on the integration of a high level of animal evidence with a moderate level of epidemiological evidence, we concluded that TCDD was a known hazard for endometriosis in humans and the conclusion is supported by mechanistic in vitro evidence. Nonetheless, there is need for further research to fill in our gaps in understanding of the relationship between OCCs and their mixtures and endometriosis, beyond the prototypical TCDD. https://doi.org/10.1289/EHP8421IntroductionHumans are exposed daily to complex mixtures of chemical pollutants, some of which may contribute to a disruption of endocrine functions and contribute to reproductive diseases. Chemicals with bioaccumulative properties are of particular concern, because they persist worldwide in the environment and ultimately may accumulate in human fatty tissues, despite the fact that many are now heavily regulated or banned (Pumarega et al. 2016; UNEP 2017). These chemicals include a large family of pollutants known as organochlorine chemicals (OCCs), characterized by their carbon–chlorine bonds, which comprise polychlorinated dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs), polychlorinated biphenyls (PCBs), and organochlorine pesticides (OCPs) (Bernes 1998; Jones and De Voogt 1999). A number of OCCs such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) have been extensively reported to interact with the aryl hydrocarbon receptor (AhR) and/or estrogen receptors and disrupt the nervous, immune, and endocrine systems at various life stages (Gore et al. 2015; Lawrence and Vorderstrasse 2013; Barouki et al. 2012).The female reproductive tract has been shown to be especially vulnerable to the presence of hormone-disrupting chemicals, potentially leading to early puberty, infertility, adverse pregnancy outcomes, fibroids, or endometriosis (Bruner-Tran et al. 2017; Gore et al. 2015; Ho et al. 2017; Hutz et al. 2006). Endometriosis is a hormone-dependent gynecological disease characterized by the presence and growth of ectopic endometrial tissues (Zondervan et al. 2018). The precise prevalence of endometriosis in the general population is largely unknown due to difficulties in diagnosis, underreporting, and unknown prevalence among asymptomatic individuals, but estimates of prevalence range widely from 5% to 45% of individuals who menstruate (Buck Louis et al. 2011; Rawson 1991). Many gaps also remain in fully understanding the etiology of endometriosis, but it is likely multifactorial, involving genetic and environmental factors (Sourial et al. 2014; van der Linden 1996).In a previous systematic review and meta-analysis, we synthesized the associations between OCCs and endometriosis in human epidemiological studies and evaluated the quality of the body of evidence using the comprehensive National Toxicology Program/Office of Health Assessment and Translation (NTP/OHAT) framework (Cano-Sancho et al. 2019). The overall conclusion of this review supported the existence of positive associations between exposure to PCDDs, PCBs, and OCPs and endometriosis, which is consistent with subsequently published reviews (Freger and Foster 2020; Wen et al. 2019). However, the level of evidence was deemed moderate, with serious risk of bias: Major methodological limitations in epidemiological research of endometriosis included the lack of population-based study designs, the inherent constraints in classifying control populations, the heterogeneity of endometriosis subtypes, the use of different OCCs biomarkers, and narrow background exposure distributions.Associations observed from epidemiological studies, however, provide an incomplete picture and fall short of being able to provide a biological explanation for the link between OCC exposure and endometriosis. Thus, from our point of view, experimental studies are necessary to advance our understanding of the underlying molecular mechanisms and provide the support of biological plausibility to the trends observed in human epidemiological studies. The nonhuman primate model of endometriosis is considered the most reliable because it mimics the pathophysiological conditions in women (Story and Kennedy 2004; Grümmer 2006). However, ethical considerations and high economic costs have favored the development of rodent models (Bruner-Tran et al. 2018). The process of surgically inducing lesions by implanting the animal's autologous uterine horn was developed in rats (Vernon and Wilson 1985) and later adapted to mice (Cummings and Metcalf 1995a).In addition, ex vivo and in vitro models, with primary cells and co-cultures from human biopsies, represent a straightforward way to gain insight into the molecular signaling pathways in endometrial cells (Fan 2020). To our knowledge, no studies to date are attempting to systematically gather and appraise the experimental evidence on OCC exposure and endometriosis, though some narrative reviews on the topic have been conducted (Birnbaum and Cummings 2002; Bruner-Tran and Osteen 2010; Guo et al. 2009; Rier and Foster 2003). In this context, the present work aimed to perform a systematic review applying the NTP/OHAT framework to a) systematically appraise experimental (in vivo and in vitro) studies reporting evidence on the associations between OCC exposure and endometriosis and b) to draw a conclusion based on the level of confidence of this body of evidence.MethodsThe present systematic review was conducted following the guidelines established in the NTP/OHAT handbook (NTP/OHAT 2015a), using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) principles. The protocol was registered in International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD42018102618) on 17 July 17 2018, peer-reviewed, and published in January 2019 (Matta et al. 2019). Study selection, data extraction, data synthesis, and risk of bias (RoB) assessment were performed and managed using the Health Assessment Workspace Collaborative (HAWC; https://hawcproject.org/), an open-source, modular web-based content management system with a user interface (Shapiro et al. 2018). Numerical data from plots and graph images were extracted using WebPlotDigitizer (version 4.3; https://automeris.io/WebPlotDigitizer/). The RoB assessment was performed using the NTP/OHAT's RoB rating tool to assess individual study quality (NTP/OHAT 2015b), and followed the guidance of the NTP protocol for the perfluorooctanesulfonic acid/perfluorooctanoic acid (PFOS/PFOA) monograph for the integration of in vivo and in vitro data (NTP 2016). Results are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al. 2009).Eligibility CriteriaThe literature search strategy was initially developed by using the list of all persistent organic pollutants (POPs) determined by the Stockholm Convention (UNEP 2017) to calibrate and refine the protocol during problem formulation. Accordingly, we developed a Populations, Exposures, Comparators, Outcomes (PECO) statement (Table 1) establishing the inclusion/exclusion criteria for our systematic review. In addition, we excluded any studies that did not contain original data, such as reviews, commentaries, editorials, or conference abstracts, as well as studies not available in English. Further details of problem formulation can be found in the previously published protocol (Matta et al. 2019).Table 1 Summary of Population, Exposure, Control, Outcome (PECO) statement for in vivo and in vitro studies.Table 1, in three columns, lists Codes, Inclusion, and Exclusion.InclusionExclusionPIn vivo: Experimental animal models where endometriosis can either occur spontaneously or be inducedIn vitro: Human endometrial cells or tissuesIn vivo: Observational epidemiological studiesIn vitro: Cancer cells, nonuterine cellsEIn vivo and in vitro:• Organochlorine chemicals (OCCs)In vivo and in vitro:• Pharmaceuticals, non-OCCsCIn vivo and in vitro:• Has reference or control groupIn vivo and in vitro:• Lacks reference or control groupOIn vivo:• Onset or aggravation of endometriosis• Proliferation or growth of induced endometriotic lesions• Presence-of "endometriosis-like" phenotypes with human reference standardIn vitro: Proliferation, migration, and/or invasion of endometrial cellsIn vivo and in vitro:• Outcomes unrelated to endometriosisLiterature Search and Study SelectionThe search string was developed to identify all relevant published evidence in experimental studies (in vivo, ex vivo, or in vitro) with primary data on the associations between controlled exposures to OCCs and endometriosis and endometriosis-related effects by a) reviewing MeSH terms and literature tags used by previously identified human epidemiology studies on endometriosis for relevant and appropriate terms; b) adapting existing chemical lists (UNEP 2017; Wassenaar et al. 2017); and c) extracting several potential endometriosis-related mechanistic outcomes (Liu and Zhao 2016). The complete search string can be found in Table S1. It comprised the exposure and outcome elements of the PECO statement nested through the Boolean operators "AND/OR." A comprehensive list of 33 persistent organic pollutants identified in the Stockholm Convention with suspected endocrine-disrupting potential can be found in Table S2 (UNEP 2017). Because the scope of our review focused on OCCs, brominated congeners (i.e., decabromodiphenyl ether, hexabromobiphenyl, hexabromocyclododecane, hexabromodiphenyl ether, heptabromodiphenyl ether, tetrabromodiphenyl ether, and pentabromodiphenyl ether) and PFOS, its salts, and perfluorooctane sulfonyl fluoride were not included in the search terms. In addition to this list of chemicals, a set of OCC-related terminology was used to capture any other studies with relevant exposure. Synonyms of each chemical name were retrieved using the PubChem and ChemSpider database on 27 June 2018. An initial search was performed on 9 January 2019 in the databases PubMed/Medline ( https://www.ncbi.nlm.nih.gov/pubmed), Scopus ( https://www.scopus.com), and Web of Science (WoS; https://webofknowledge.com), with syntax customized for each database. A subsequent manual search of the bibliographic references of relevant articles and existing reviews was performed. A final search update was performed on 24 June 2020. No filter was applied to limit the date or language of publication during the search. The identified studies were exported to EndNote in RIS format to pool records for manual duplicate removal. Studies underwent a two-phase screening process by two independent researchers (K.M. and G.C.S.), based on the predefined inclusion and exclusion criteria of the PECO statement (Table 1). Phase I Screening was based on title and abstracts, and Phase II Screening was based on full text for the studies not excluded in Phase I. Any discrepancies were resolved by consensus or with the opinion of a third researcher (J.P.A.).Data ExtractionEligible studies underwent data extraction within the HAWC module according to a predefined standardized data extraction process (Table S3), explained below in detail, with illustrative examples from Foster et al. 1997 in the supplemental materials (Figures S1–S5). Results are stored and available for download in Excel format ( https://hawcproject.org/assessment/812/).First, studies were identified by citation (short and full), and type(s) of study data were indicated (i.e., "Animal bioassay," hereafter referred to as in vivo and "in vitro"). Studies which contained both in vivo and in vitro data were indicated as such. In this initial stage of study preparation, general study data (i.e., conflicts of interest, funding source, author correspondence details) were extracted (Figure S1).Studies with in vivo data comprised at least one animal "experiment" (Figure S2). Data extracted within each experiment included study name, study type (acute, short term, subchronic, chronic, mechanistic, reproductive, developmental, etc.), chemical identifiers (name, CAS, source), chemical purity, vehicle used, and details of animal husbandry and diet and compliance with any guidelines for methods (Figure S3). Each experiment consisted of at least one "animal group." Data extracted within each animal group included animal species, strain, and sex, source of animals (laboratory or breeding details), life stage exposed and life stage assessed, observation duration, siblings (if animal groups were related), and any additional description of the animal group (Figure S4). A dosing regimen was also extracted for each animal group, including route of exposure, exposure duration, number of dose groups, control information (positive and/or negative controls), and any other information about dosing methodology (e.g., dose units, dose groups, details of dosing regimen). Dose–response information for end points was also extracted if such data were provided. Each animal group could have multiple end points, characterized by relevant biological system (i.e., reproductive), organ/tissue (i.e., endometrium), effect/effect subtype, diagnostic method, observation time, data set type (e.g., continuous, dichotomous, percent difference), variance type (standard error or standard deviation if reported or relevant), response units, data location in the literature, and other notes on the end point methodology and results (e.g., response direction that would be considered adverse, points of departure, monotonicity, statistical tests, trend tests, etc.) (Figure S4). A dose–response visualization was automatically generated for end points with dose–response data (Figure S5).A similar data extraction process was performed for in vitro studies. First, cell type data was extracted, including species, strain, sex, and tissue of origin (typically female humans), as well as culture type (i.e., primary culture, immortalized cell line, transfected cell line, etc.) and source of cell cultures. Such data were extracted for each relevant cell type used in each study if more than one cell population were used. Chemical exposure data were then extracted, including chemical name, CAS number, source, purity (and purity confirmation details if available), and dilution storage notes. Again, such data were extracted for each unique chemical exposure for each study. Following this initial data extraction, an "experiment" for each cell type was created. If there were multiple cell types in an in vitro study, a separate experiment was created for each one. In each in vitro experiment, data extracted included dosing information (dosing regimen, duration, units), serum information (percent serum, serum type, and/or description), and control information (positive, negative, and/or vehicle controls). As with in vivo experiments, in vitro experiments with multiple outcomes were characterized by chemical exposure, assay type, outcome effect, data location in the literature, data set type (e.g., continuous, dichotomous, percent difference), variance type (standard error or standard deviation if reported or relevant), response units, observation time, points of departure, monotonicity, statistical tests, trend tests, and any other notes on the end point of the results. Dose–response data for end points were also extracted when such data were provided. Dose–response visualizations were similarly automatically generated for such end points. Corresponding authors were contacted by email for data unavailable in the published articles and for clarification of methods and risk of bias questions, and authors were provided 2 wk for response.Synthesis of ResultsEnd points were grouped into primary or secondary end points, as previously explained in the published protocol (Matta et al. 2019). In brief, four primary end points were measured (two in vivo and two in vitro). For in vivo studies, primary end points aimed to be the corollary to endometriosis in humans thus included a) the spontaneous onset of endometriosis and b) the growth or proliferation of induced endometriotic lesions. For in vitro studies, primary end points included c) cell migration/invasion and d) cell viability/proliferation. Secondary end points included gene expression or protein levels within the signaling pathways regulating the primary end points, markers of disrupted steroidogenic pathways, inflammatory biomarkers, such as cytokines [i.e., interleukins (IL)], or markers of extracellular matrix remodeling [i.e., matrix metalloproteinases (MPPs)]. End point results were summarized in data pivot figures displaying the significance and direction of the effects across exposure doses and studies. The heterogeneity of included studies precluded a quantitative meta-analysis.Risk of Bias AssessmentThe NTP/OHAT RoB Rating Tool was specifically adapted to the research question in the HAWC interface (NTP/OHAT 2015b) and tailored to in vitro studies as previously reported (NTP 2016). Briefly, the RoB tool consists of a set of questions tailored to each experimental stream of evidence, addressing six main bias domains listed below (i.e., selection bias, performance bias, attrition bias, detection bias, selective reporting bias, and other) (Table 2).Table 2 Risk of bias analysis: domains of bias and questions.Table 2 has two columns, namely, Domain of Bias and Risk of Bias Question.Domain of biasRisk of bias questionSelection bias1. Was administered dose or exposure level adequately randomized?Selection bias2. Was allocation to study groups adequately concealed?Performance bias3. Were experimental conditions identical across study groups?*Performance bias4. Were the research personnel blinded to the study group during the study?Attrition bias5. Were outcome data incomplete due to attrition or exclusion from analysis?Detection bias6. Can we be confident in the exposure characterization?*Detection bias7. Can we be confident in the outcome assessment?*Selective reporting bias8. Were all measured outcomes reported?Other9. Were there any other potential threats to internal validity?Note: Key elements considered for the tiered classification are marked by an asterisk (*).Questions received one of five possible ratings: "Definitely Low Risk of Bias," "Probably Low Risk of Bias," "Probably High Risk of Bias," "Definitely High Risk of Bias," or "Not Reported," based on prespecified criteria (Supplemental Materials, Section 3). The rating was determined by two independent assessors (K.M. and G.C.S.) and then finalized by discussion and consensus, with consultation by an additional member of the review team or technical advisors as needed. In the event that additional information was needed to make a rating determination, authors were contacted with questions specific to the RoB question and provided 2 wk to respond. Each RoB rating was justified based on the established criteria and the study text, and is stored and available in HAWC. Based on these ratings, individual studies were ranked on a three-tier scale of bias allowing the classification of specific bodies of evidence in "not serious," "serious," or "very serious" RoB to support decision-making for confidence rating (NTP/OHAT 2015a). More details of the three-tier scale can be found in the next section or in the previously published protocol.Rating the Level of EvidenceFollowing the NTP/OHAT framework, we analyzed different domains affecting the confidence level for each primary end point–related body of evidence. Each stream of data was considered a body of evidence, and an assessment was performed for each primary end point to determine a confidence rating reflecting the confidence with which the study findings accurately reflect a true association between exposure to OCCs and the primary end points. The process is summarized in Table S4 and the NTP/OHAT systematic review handbook, based on GRADE guidelines (NTP/OHAT 2015a).Briefly, each body of evidence was given an initial confidence rating, which was subsequently downgraded or upgraded according to factors that decrease or increase confidence in the results (NTP/OHAT 2015a). The initial confidence rating of "high" was determined by the presence of all four main features determined by the study design for both in vivo and in vitro evidence (Table S5). This high initial confidence rating was then either downgraded or upgraded, depending on the presence or absence of certain cross-studies flaws (i.e., risk of bias, inconsistency, indirectness, imprecision, publication bias) or strengths (i.e., consistency, large magnitude of effect, dose response), respectively.Factors decreasing confidence.RoB: As previously mentioned, the NTP/OHAT's RoB-tiered approach considers three key elements of higher relevance to establish the classification criteria for each individual study [marked in Table 2 by an asterisk (*)]. Studies were subsequently categorized into three possible tiers depending on their responses to these key elements: Tier 1: Study must be rated as "Definitely Low" or "Probably Low" RoB for key elements AND have most other applicable items answered "Definitely Low" or "Probably Low"; Tier 2: Study meets criteria for neither Tiers 1 nor 3; Tier 3: Study must be rated as "Definitely High" or "Probably High" RoB for key elements AND have most other applicable items answered "Definitely High" or "Probably High." "Not Reported" was counted as "Probably High." Downgrading for RoB reflects the entire body of studies; therefore, the decision to downgrade was applied conservatively and reserved for cases with substantial RoB across most of the studies composing the body of evidence.Unexplained inconsistency: Studies were considered for downgrading when there was inconsistency in results that were not explained by study design features, such as differences in cell model/animal species, timing or route of exposure, or health outcome assessment.Indirectness/applicability: The following points were used to assess the directness in the present study: Differences in population (applicability) and relevance of the animal model to outcome of concern. In vivo studies: Studies conducted in mammalian model systems were assumed relevant for humans (i.e., not downgraded) unless compelling evidence to the contrary was identified during the course of the evaluation (e.g., a biological system not present in humans).In vitro studies: Cell models were evaluated on the basis of the biological relevance in humans (human primary cell cultures or human immortalized cell lines).Differences in outcome measures or directness of the end points to the primary health outcome(s). For example, onset of endometriosis would be a direct end point, whereas development of "endometriosis-like" phenotypes is less direct.Dose levels and route of administration in in vivo studies: External dose comparisons were used to reach confidence rating conclusions, because internal dosimetry in animal models can vary based on route of administration, species, age, diet, and other cofactors. The most commonly used routes of administration (i.e., oral, dermal, inhalation, subcutaneous injections) were considered direct for the purposes of establishing confidence ratings.The applicability of specific health outcomes or biological processes in animal models is outlined in the PECO inclusion/exclusion criteria, with the most accepted relevant/interpretable outcomes considered "primary," and less direct measures, biomarkers of effect, or upstream measures of health outcome considered "secondary."Imprecision: Imprecision is typically assessed with confidence intervals for meta-analyses, but because a meta-analysis was not performed due to the heterogeneity of outcome measurements, the overall effects of the studies were considered for imprecision. Studies with high variability of effect estimates were at risk of imprecision bias.Publication bias: Studies were considered for downgrading for publication bias when the study was uniformly small, especially when sponsored by industries, nongovernment organizations, or authors with conflicts of interest. Abstracts or other types of gray literature that do not appear as full-length articles within a reasonable time frame (3–4 y) may be another indication of publication bias; thus such literature has been excluded.Factors increasing confidence.Magnitude of effect: A large magnitude of effect was considered to upgrade the confidence.Dose–response: Confidence was upgraded for dose–response if there was sufficient evidence of monotonic dose–response/gradient.Consistency: Consistency across animal studies, dissimilar populations, and study types were potential reasons to upgrade the confidence: Across animal studies: consistent results reported in multiple experimental animal models or species. Finding the same direction of change in the same outcome in more than two species would constitute sufficient evidence that a causal relationship has been established, by standards set by the International Agency for Research on Cancer. Though the health effect studied here is not cancer, the same principle can apply.Across dissimilar populations: consistent results reported across populations that differ in factors such as time, location, and/or exposure.Across study types: consistent results reported from studies with different design features (e.g., between chronic and multigenerational animal studies).Rating the Level of EvidenceSubsequently, the confidence rating was translated to the level of evidence considering the presence or absence of the health effect and the direction or nature of the effect (Table S6). Level of evidence was established for each of four primary outcomes separately.Evidence Integration and Hazard IdentificationBased on the NTP/OHAT Hazard Identification Scheme (Figure S6) reported by Rooney et al. (2014), the level-of-evidence conclusion for human data can be considered together with nonhuman animal data to reach