Scholarly article on topic 'Comparison of in vivo genotoxic and carcinogenic potency to augment mode of action analysis: Case study with hexavalent chromium'

Comparison of in vivo genotoxic and carcinogenic potency to augment mode of action analysis: Case study with hexavalent chromium Academic research paper on "Biological sciences"

CC BY
0
0
Share paper
Keywords
{Carcinogenicity / Genotoxicity / "Hexavalent chromium" / "Mode of action" / "Risk assessment"}

Abstract of research paper on Biological sciences, author of scientific article — Chad M. Thompson, Anne Bichteler, Julia E. Rager, Mina Suh, Deborah M. Proctor, et al.

Abstract Recent analyses—highlighted by the International Workshops on Genotoxicity Testing Working Group on Quantitative Approaches to Genetic Toxicology Risk Assessment—have identified a correlation between (log) estimates of a carcinogen’s in vivo genotoxic potency and in vivo carcinogenic potency in typical laboratory animal models, even when the underlying data have not been matched for tissue, species, or strain. Such a correlation could have important implications for risk assessment, including informing the mode of action (MOA) of specific carcinogens. When in vivo genotoxic potency is weak relative to carcinogenic potency, MOAs other than genotoxicity (e.g., endocrine disruption or regenerative hyperplasia) may be operational. Herein, we review recent in vivo genotoxicity and carcinogenicity data for hexavalent chromium (Cr(VI)), following oral ingestion, in relevant tissues and species in the context of the aforementioned correlation. Potency estimates were generated using benchmark doses, or no-observable-adverse-effect-levels when data were not amenable to dose-response modeling. While the ratio between log values for carcinogenic and genotoxic potency was ≥1 for many compounds, the ratios for several Cr(VI) datasets (including in target tissue) were less than unity. In fact, the ratios for Cr(VI) clustered closely with ratios for chloroform and diethanolamine, two chemicals posited to have non-genotoxic MOAs. These findings suggest that genotoxicity may not play a major role in the cancers observed in rodents following exposure to high concentrations of Cr(VI) in drinking water—a finding consistent with recent MOA and adverse outcome pathway (AOP) analyses concerning Cr(VI). This semi-quantitative analysis, therefore, may be useful to augment traditional MOA and AOP analyses. More case examples will be needed to further explore the general applicability and validity of this approach for human health risk assessment.

Academic research paper on topic "Comparison of in vivo genotoxic and carcinogenic potency to augment mode of action analysis: Case study with hexavalent chromium"

Contents lists available at ScienceDirect

Mutation Research/Genetic Toxicology and Environmental Mutagenesis

journal homepage www.elsevier.com/locate/gentox Community address www.elsevier.com/locate/mutres

Short communication

Comparison of in vivo genotoxic and carcinogenic potency to augment mode of action analysis: Case study with hexavalent chromium

CrossMark

Chad M. Thompson3 *, Anne Bichtelerb, Julia E. Ragerb, Mina Suhc, Deborah M. Proctor' Laurie C. Hawsb, Mark A. Harrisa

a ToxStrateggies, Inc., Katy, TX, United States b ToxStrategies, Inc., Austin, TX, United States c ToxStrategies, Inc., Mission Viejo, CA, United States

ARTICLE INFO

Article history:

Received 7 December 2015

Received in revised form 28 January 2016

Accepted 29 January 2016

Available online 6 March 2016

Keywords: Carcinogenicity Genotoxicity Hexavalent chromium Mode of action Risk assessment

ABSTRACT

Recent analyses—highlighted by the International Workshops on Genotoxicity Testing Working Group on Quantitative Approaches to Genetic Toxicology Risk Assessment—have identified a correlation between (log) estimates of a carcinogen's in vivo genotoxic potency and in vivo carcinogenic potency in typical laboratory animal models, even when the underlying data have not been matched for tissue, species, or strain. Such a correlation could have important implications for risk assessment, including informing the mode of action (MOA) of specific carcinogens. When in vivo genotoxic potency is weak relative to carcinogenic potency, MOAs other than genotoxicity (e.g., endocrine disruption or regenerative hyper-plasia) may be operational. Herein, we review recent in vivo genotoxicity and carcinogenicity data for hexavalent chromium (Cr(VI)), following oral ingestion, in relevant tissues and species in the context of the aforementioned correlation. Potency estimates were generated using benchmark doses, or no-observable-adverse-effect-levels when data were not amenable to dose-response modeling. While the ratio between log values for carcinogenic and genotoxic potency was > 1 for many compounds, the ratios for several Cr(VI) datasets (including in target tissue) were less than unity. In fact, the ratios for Cr(VI) clustered closely with ratios for chloroform and diethanolamine, two chemicals posited to have non-genotoxic MOAs. These findings suggest that genotoxicity may not play a major role in the cancers observed in rodents following exposure to high concentrations of Cr(VI) in drinking water—a finding consistent with recent MOA and adverse outcome pathway (AOP) analyses concerning Cr(VI). This semiquantitative analysis, therefore, may be useful to augment traditional MOA and AOP analyses. More case examples will be needed to further explore the general applicability and validity of this approach for human health risk assessment.

© 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license

(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

Recent analyses have highlighted the correlation between in vivo genotoxic potency and in vivo carcinogenic potency of carcinogens [1-4]. These potencies are based on quantitative dose-response analysis, specifically benchmark dose (BMD) modeling, of in vivo genotoxicity (e.g. peripheral blood micronucleus (MN) assay) and carcinogenicity in the most sensitive tissue from available car-cinogenicity studies. The correlation between in vivo genotoxic and carcinogenic potency was high (~0.8) among the chemicals evaluated despite the fact that the endpoints were not constrained

* Corresponding author at: ToxStrategies, Inc., 23123 Cinco Ranch Blvd., Suite 220, Katy, TX 77494, United States.

E-mail address: cthompson@toxstrategies.com (C.M. Thompson).

for tissue, species, or strain concordance [2]. Members of the Working Group on Quantitative Approaches to Genetic Toxicology Risk Assessment (QWG) of the International Workshops on Genotoxicity Testing (IWGT) posited that this in vivo genotoxic-ity/carcinogenicity correlation would likely be found to be higher once a sufficient number of cases could be compiled for data matched with respect to target organ, species, and strain [2].

The correlation between in vivo genotoxicity and carcinogenic-ity has important implications for risk assessment. For example, examination of this correlation was motivated by the desire to use faster, less expensive genotoxicity assays to predict in vivo carcinogenic potency when carcinogenicity data are lacking [1,3]. Moreover, genotoxicity studies require far fewer animals than the standard cancer bioassay, and are thus consistent with the three Rs of animal testing (i.e. replacement, reduction, refinement).

http://dx.doi.org/10.1016/j.mrgentox.2016.01.008

1383-5718/© 2016 The Authors. Published by Elsevier B.V. This is an open access article underthe CC BY license (http://creativecommons.org/licenses/by/4X>/).

Aside from such predictive applications, the correlation between in vivo genotoxic and carcinogenic potency can also inform the mode of action (MOA) of carcinogens that are suspected of having (or assumed to have) a mutagenic MOA. That is, we can expect the correlation between genotoxic and carcinogenic potencies to be strong in chemicals with a mutagenic MOA. When geno-toxicity cannot be readily demonstrated in target tissues using standard assays (e.g., OECD guideline studies) and physiologically-or environmentally-relevant exposure routes, strength is lent to alternative non-genotoxic MOAs such as endocrine disruption or cytotoxicity and compensatory regenerative hyperplasia.

Assessment of the mechanism(s) by which a chemical causes cancer is traditionally conducted through a weight of evidence (WOE) approach typified by MOA and adverse outcome pathway (AOP) analyses [5-8]. Such analyses have been published fortumors caused by hexavalent chromium [Cr(Vl)] [9-13] in a 2-year cancer bioassay conducted by the National Toxicology Program (NTP) [14]. Some of these MOA analyses have concluded that the intestinal tumors arise through a mutagenic MOA, whereas others have concluded that the intestinal tumors arise as the result of chronic toxicity to intestinal villi leading to crypt regenerative hyperpla-sia and ultimately to spontaneous mutations within stem cells of the crypt compartment (Table 1). Herein, we examine the relationship between in vivo genotoxicity and carcinogenicity of Cr(Vl) in the broader context of the aforementioned correlation that exists between in vivo genotoxic and carcinogenic potency for many carcinogens. This analysis provides insight into the likelihood of a mutagenic or non-mutagenic MOA for Cr(Vl)-induced intestinal tumors in mice. The findings are then discussed in the context of the aforementioned MOA and AOP analyses for Cr(Vl) [9-12].

2. Methods

Quantitative aspects of this review/analysis involve BMD modeling to derive upper and lower confidence intervals for estimates of in vivo genotoxic (in blood MN assays) and carcinogenic (in the most sensitive rodent tissue) potency. Much of the data are BMDL values taken directly from Soeteman-Hernandez et al. [1]. Additional BMD analyses in the current article were conducted on blood MN data for Cr(Vl) published by NTP [15]. Specifically, multiple MN datasets in different mouse strains were analyzed together using PROAST version 61.3 [16], testing sex and study experiment as covariates. Following recent methodology identifying dose-response shape to be the parameter of greatest biological interest [17], nested models were tested by allowing fewer and fewer parameters to vary by the covariates. Thus, it could be determined on which parameters (control response, maximum response, potency, and steepness) the datasets differed from one other, and whether it was appropriate to estimate a composite BMDL for the Cr(Vl) in vivo blood MN assays. Additional points of departure (POD) for Cr(Vl) genotoxic potency estimates in target tissues were taken from published literature (see Section 3).

3. Results and discussion

Fig. 1 recapitulates data presented in Soeteman-Hernandez et al. [1] and MacGregor et al. [2]. The filled circles represent BMDL estimates for in vivo genotoxic and carcinogenic potency for 23 datasets, where the cancer was categorized (by Soeteman-Hernandez et al.) as carcinoma in a single tissue (designated category A herein). The open circles represent similar estimates for 21 datasets where the cancer was categorized as a combination of various tumor types in different tissues (designated category B herein). Soeteman-Hernandez et al. acknowledged that these categories were more pragmatic than informed by histopathology

Fig. 1. Correlation between in vivo genotoxic and carcinogenic potency. Filled circles represent datasets where tumors were categorized by Soeteman-Hernandez et al., as 'carcinoma in a single tissue'; open circles represent datasets where the tumors were categorized as a combination of tumor types in different tissues. Black circles (open and filled) represent the lowest BMDL10 for cancer in the most sensitive tissue (y-axis) and lowest BMDL5 for MN in peripheral blood (x-axis). The red circle represents data for Cr(Vl) (expressed as mg/kg SDD); these data were taken directly from Tables 3 and 4 in Soeteman-Hernandez et al. [1] (see text). Blue symbols are placed along the x-axis based on analyses in the present article. The blue circle represents BMDL5 for a meta-analysis of MN data in standard laboratory mouse strains as opposed to the Am3 strain represented by the red circle (see text and Supplemental Fig. S1). The blue diamond represents a NOAEL for mutant frequency in the Big Blue® rat oral mucosa. The downward and upward pointing blue triangles represent the NOAELs from 7-day and 90-day duodenal MN assays, respectively (see text). These three NOAEL values are based on the absence of a statistically significant response at the highest study dose, i.e. LOAEL or BMD values would fall higher on the x-axis. The dashed line represents unity of the ratio between BMDL10 and BMDL5.

expertise. The dashed line in Fig. 1 indicates where the ratio of the logBMDL10 for carcinogenicity and the logBMDL5 for genotoxicity would be unity. Clearly, the BMDL10 values for tumor risk tend to be equal to or greater than the BMDL5 values for micronuclei, which is to be expected if genotoxicity were a precursor to tumorigenicity. lndeed, Soeteman-Hernandez et al. [1] concluded that a BMD10 for carcinogenicity might be predicted to be <100-fold higher than the BMD5 for in vivo genotoxicity.

Soeteman-Hernandez et al. [1] assigned the tumors from oral exposure to Cr(Vl) to category B (red circle, Fig. 1). As will be discussed in greater detail below, Cr(Vl) exposure caused adenomas and carcinomas in the duodenum and jejunum of B6C3F1 mice. There is no clear division between the duodenum and jejunum, and it is therefore questionable to consider these as 'different tissues'. ln fact, neoplasms in various portions of the small intestine are typically combined for analysis [14,18,19]. Moreover, adenomas and carcinomas of the small intestine are considered to be on a continuum of cancer progression and are thus typically combined for analysis [18,19]. As such, it is questionable to consider the tumorigenic responses to Cr(Vl) in mice as 'various tumour types'. Herein, we therefore examine the in vivo genotoxic and carcinogenic potency of Cr(Vl) in comparison to compounds from both categories, because Cr(Vl) might very well fit into either category. This also has the advantage of increasing the number of datasets in our analysis.

With regard to the in vivo genotoxic potency of Cr(Vl), there are five 90-day GLP peripheral blood MN datasets for Cr(Vl) (administered as sodium dichromate dihydrate, SDD), conducted by the NTP [15]. Of the five assays, three were negative, one was deemed equivocal because it failed to reach statistical significance, and another was positive (upper left quadrant of Table 2). Although Soeteman-Hernandez et al. [1] cite recent work on the commonality of dose-response shape across multiple studies and compounds [17] and note one of the strengths of analyzing clusters of datasets is the inclusion of data not showing a significant dose-response trend, the full set of blood MN assay results were not included in their analysis [1]. lnstead, Soeteman-Hernandez et al. [1] modeled

Table 1

Published MOAs and AOPs Relevant to Cr(VI).

MOAs AOP

McCarroll et al. [13] Thompson et al. [11] Becker et al. [12]

Interaction with DNA (KE1) Intestinal absorption (KE1) Interaction with villi (MIE)

Mutagenesis (KE2) Villus cytotoxicity (KE2) Villus cytotoxicity (KE1)

Cell proliferation (KE3) Crypt hyperplasia (KE3) Crypt hyperplasia (KE2)

Gastrointestinal tumors (KE4) Crypt cell mutation (KE4) Spontaneous mutation in crypt stem cells (KE3)

Tumorigenesis Preneoplasia (KE4)

Intestinal tumors (AO)

KE, key event; MIE, molecular initiating event; AO, adverse outcome.

Table 2

Comparison of in vivo genotoxicity and carcinogenicity of Cr(VI). In vivo Genotoxicity In vivo Carcinogenicity

SYSTEMIC

NTP [15] five 90-day blood MN assays: NTP [14] 2-yr Cr(VI) bioassay:

- B6C3F1, <88 ppm, M, (-) - No'systemic'tumors

- B6C3F1, <350 ppm, M, (-)

- B6C3F1, <350 ppm, F, (-)

- BALB/c, <88 ppm, M, (-)

- Am3-C57BL/6, <88 ppm, M, (+)

Target tissue genotoxicity: NTP (2008) 2-yr Cr(VI) bioassay:

- aTgF344 oral cavity, 180 ppm, cII MF in buccal or palate sides of gingiva surrounding upper molar teeth, (-) - Oral cavity tumors in F344 rats

- bB6C3F1, <180 ppm, 90 days, kras codon 12 GAT MF, (-) - Small intestine tumors in B6C3F1

- bB6C3F1, <180 ppm, 90 days, crypt MN (transverse sections), (-) mice

- bB6C3F1, <180 ppm, 7 days, crypt MN (transverse sections), (-)

- cB6C3F1, <180 ppm, 7 days, crypt MN (Swiss rolls), (-)

- cB6C3F1, <180 ppm, 7 days, crypt ^-H2AX, (-)

- dB6C3F1,180 ppm, 90 days, crypt ^-H2AX, (-)

(-) negative result; (+) positive result. a Thompson et al. [28]. b O'Brien et al. [30]. c Thompson et al. [31]. d Thompson et al. [32].

only the positive data in Am3-C57BL/6 mice (the red circle in Fig. 1). The Am3-C57BL/6 strain harbors the am3 allele of bacteriophage $X174, which can be used to detect A:T base pair mutations [20]. According to NTP, mutational studies were not carried out in this strain due to technical difficulties [15]. To our knowledge, this strain no longer exists and was not used in any other MN test conducted by the NTP, nor could we find any MN studies with this strain in PubMed.1 With such a limited historical database on this strain, the positive results are difficult to interpret or relate to other findings. Notably, negative MN results have also been reported in blood and bone marrow studies of BDF1 and Swiss albino mice exposed to Cr(VI) [22].

Using PROAST BMD modeling software [16], we assessed the similarity in BMD models of Cr(VI) genotoxic potency in the five NTP datasets available: the four datasets using standard laboratory strains (B6C3F1 and BALB/c mice) and the fifth using Am3-C57BL/6 mice. Two of the four studies reported testing very high Cr(VI) concentrations (1000 mg/L SDD) where the MN responses decreased or increased insignificantly versus controls (2.0 vs. 2.7; and 1.9 vs. 1.7); therefore, the 1000 mg/L SDD groups were dropped, resulting in steeper dose-responses from which to estimate the BMDL5. Starting with the one study containing both male and female datasets, we ran nested exponential models testing whether sex was related to within-group variation of the 4 parameters estimated: a, response

1 Only one published study resulted from the following query (conducted on Jan 22, 2016): (Micronucleus Tests[Mesh]) AND ("Am3-C57BL/6" OR "Am3" OR Bacteriophage phi X 174|Mesh] OR iE protein, bacteriophage X174|Supplementary Concept]); this study was the NTP report on Cr(Vl) [21].

at the control dose; b, potency; c, maximum fold-change in MN; d, steepness. The best model (highest log-likelihood) allowed only parameter a to vary by sex. Thus sex was not required as a covariate in the subsequent analyses estimating potency, as the remaining studies only included male mice. The two additional NTP studies in B6C3F1 and BALB/c mice were added incrementally, testing the study as a covariate, and the potency parameter remained most efficiently estimated as constant across all studies. However, adding the fifth study using Am3-C57BL/6 mice yielded a highly significant difference on the shape parameter for the study covari-ate; that is, the study with Am3-C57BL/6 mice did not conform in shape to the other four studies. We retained the BMDL5 estimated in the grouped analysis using the standard mice for the reasons outlined above, and also because the non-standard Am3-C57BL/6 experiment differed so greatly from the four studies where the dose-response shape parameters conformed well (Supplemental Fig. S1). Thus, this BMDL5 was generated from 4 out of the 5 datasets available, which were also the datasets that were considered negative or equivocal by the original study authors. The composite logBMDL5 for this analysis is shown as the blue circle in Fig. 1.

Overall, the generally negative MN findings indicate that Cr(VI) has low/no in vivo systemic genotoxicpotency—as measured by the blood MN test. Based on the apparent relationship between in vivo genotoxic and carcinogenic potency in Fig. 1, the weak MN results in the NTP studies predict low carcinogenic potency of Cr(VI). Indeed, at the conclusion of a 2-year cancer bioassay [14], Cr(VI) was found not to induce any tumors in rats or mice beyond the portal of entry, i.e. no 'systemic' tumors (upper right quadrant of Table 2).

Although systemic tumors were not observed, Cr(Vl) exposure was associated with an increase in oral cavity tumors in F344 rats and small intestine tumors in B6C3F1 mice (lower right quadrant of Table 2; Supplemental Fig. S2). Oral tumors in rats were significantly elevated relative to concurrent controls at the highest test concentration (180ppm Cr(Vl) or ~516mg/L SDD; Supplemental Fig. S2), which is 5 orders of magnitude higher than the average 0.001 ppm Cr(Vl) concentrations found in U.S. drinking water. No non-neoplastic or pre-neoplastic histopathological lesions have been observed in the rat (or mouse) oral mucosa in either the NTP bioassays or the 90-day drinking water studies that were conducted subsequently [14,15,23,24]. ln the 2-year cancer bioassay, the first incidence of oral mucosa tumor in rats was reported at 506 days [14]. There were no effects on survival; however, water consumption and body weight were significantly reduced at the higher dose groups. The overall oral and general health of these animals may have been compromised [25]. Notably, higher concentrations employed in earlier 90-day studies resulted in reduced water consumption and decreased body weight in rats and mice, as well as glandular stomach ulcers in rats [15]; thus, 180ppm Cr(Vl) was the highest drinking water concentration used in the 2-year cancer bioassay [14]. Cr(Vl) increased the incidence of diffuse epithelial hyperplasia and tumors (adenomas and carcinomas) in the small intestine of male and female mice (Supplemental Fig. S2). Notably, Cr(Vl) was more potent in inducing non-neoplastic diffuse epithelial hyperplasia than tumors. Similar to rats, there was no effect on survival, and the first incidence of tumor occurred well after the first year of the study. These intestinal tumors are quite rare, as the fungicide captan is the only other chemical to clearly induce small intestine tumors in mice in an NTP (or formerly NCl) cancer bioassay [26,27].

As highlighted in MacGregor et al. [2], genotoxicity tests should ideally be conducted in tissues that receive the highest exposure and/or are the site or carcinogenic response. At the time the NTP [15] study was completed, there were no genotoxicity data for Cr(Vl) in the oral mucosa or small intestine (i.e. the organs that developed tumors). More recently, several studies have been conducted in target tissues—all of which have failed to detect positive genotoxicity results (lower left quadrant of Table 2). For the oral cavity, a GLP transgenic rodent (TGR) mutation assay in Big Blue® TgF344 rats, conducted in accordance with OECD Guideline 488 principles, demonstrated robust increases in mutant frequency (MF) in the gingival mucosa of rats exposed to 10 ppm 4-nitroquinoline-1-oxide (4NQO) but not 180ppm Cr(Vl) [28].

Several genotoxicity studies in the small intestine of B6C3F1 mice (the same strain used in NTP cancer bioassay) have also reported negative results. Because kras is commonly mutated early in intestinal carcinogenesis, and kras codon 12 GGT^ GAT mutations (krasG12D) have been shown to increase intestinal hyperplasia

[29], kras codon 12 GAT mutations were assessed in scraped duodenal mucosa from mice exposed to <180 ppm Cr(Vl) for 90 days

[30]. No significant increases (pairwise or trend) in kras codon 12 GAT mutant frequency were observed [30]. Digital images of transverse Feulgen-stained duodenal sections from the same study were examined for MN in multiple assays. Using published methodology, MN were scored in 10 fully intact crypts for each mouse exposed to < 180 ppm for 90 days. ln a novel approach, MN were scored in crypt and villus regions (regardless of intactness) across three serial sections from each mouse—thereby increasing coverage of analysis. ln a third assay, the same serial slide-based analysis was conducted on transverse sections from mice exposed to <180ppmCr(Vl) for only 7 days. All studies were negative with regard to induction of MN in crypts; however, MN and karyorrhectic nuclei (KN) were elevated slightly in the villus regions of mice exposed to >60 ppm Cr(Vl) for 90 days (vide infra) [30].

ln a subsequent GLP in vivo MN study, digital images of Feulgen-stained duodenal sections prepared by the 'Swiss roll' technique were examined for MN [31]. While the positive control cyclophosphamide significantly elevated crypt MN and KN, exposure to <180 ppm Cr(Vl) did not induce crypt MN or KN. Sections from these mice were also stained with anti-^-H2AX to assess potential DNA damage. No differences in crypt ^-H2AX immunostaining were evident in mice exposed to Cr(Vl), whereas 7-H2AX immunostaining co-localized with MN and KN nuclei in cyclophosphamide-treated mice. Unstained Swiss roll duodenal sections from mice exposed to 180 ppm Cr(Vl) were also examined by X-ray fluorescence (XRF) microscopy to determine the localization of Cr within the duodenal mucosa. Cr fluorescence was detected in the villi, but little or no fluorescence was detected in the crypt regions (Supplemental Fig. S3, [31]).

As further assessment of the lack of genotoxicity in multiple assays, we conducted XRF mapping in transverse duodenal sections from mice exposed to 180 ppm Cr(Vl) for 90 days in order to determine whether Cr directly reaches the crypt compartment following long-term, high-dose exposure scenarios [32]. Similar to the findings after 7 days of exposure, Cr was readily detected in villi, but scarcely detected in crypts (Supplemental Fig. S3)—indicating that little or no Cr(Vl) absorption occurred in the crypt compartment after prolonged exposure. ln fact, the Cr levels detected in the crypt regions were just slightly above background. Considering that ingestion of 180 ppm Cr(Vl) results in elevations of Cr (likely as inert trivalent chromium, Cr(lll)) in blood and other tissues [14,23,33], it is not surprising to detect some small level of Cr in the vascu-larized crypt region. lt should be noted that Cr(lll) did not induce any tumors in a 2-yr cancer bioassay [34]. As in the aforementioned 7-day MN study, no changes in crypt ^-H2AX immunostaining was evident in mice exposed to 180 ppm Cr(Vl) for 90 days [32]. ln contrast, 7-H2AX immunostaining was observed in the tips of the villi, which is consistent with the aforementioned detection of MN and KN in villi, as well as the localization of Cr fluorescence in villi via XRF microscopy (vide supra).

Taken together, recent studies indicate that Cr(Vl)-induced genotoxicity does not occur in the crypt compartment where intestinal stem cells reside. Although there is clear dosimetry of Cr and signs of genotoxicity in intestinal villi (MN, KN, ^-H2AX immunostaining), the crypt compartment remains devoid of Cr signal with no evidence for increases in aberrant nuclei or kras codon 12 GAT MF. Given that toxic effects were apparently confined to intestinal villi, transverse sections from mice exposed to <180 ppm Cr(Vl) for 90 days were examined for signs of aberrant proliferative foci within villi that might be indicative of preneoplastic lesions, but no such lesions were observed [32]. A subsequent study examining both Cr(Vl) and B[a]P reported no 'atypical hyperplasia' (includes villus foci) in mice exposed to 5.5 ppm Cr(Vl) for ~150 days, but did observe such lesions in mice treated with B[a]P [35]. Based on the study findings reported in the lower left quadrant of Table 2 and discussed above, we conclude that Cr(Vl) has no/low in vivo geno-toxic potency in the gastrointestinal tract, which was the only site where tumors were observed in the 2-year NTP cancer bioassay.

The experimental data discussed above suggest that Cr(Vl) does not fit the expected pattern of exhibiting genotoxicity in tissues that respond most sensitively with regard to carcinogenicity. ln fact, we have attempted to compare the tissue specific data for Cr(Vl) summarized above with the broader correlation that is depicted in Fig. 1. For this exercise, we have taken the highest Cr(Vl) doses in the tissue-specific genotoxicity assays (in mg/kg SDD) as quantitative estimates for in vivo genotoxic potency of Cr(Vl) (i.e. no-observable-adverse-effect levels, NOAELs) in the target tissues, even though the genotoxicity results suggest the potency may be nil (see Table 2). Notably, a recent meta analysis of several risk assessments found that NOAEL and BMDL10 values tend to approximate one another

[36], thus these NOAELs might be roughly analogous to the BMDL values in Fig. 1. The NOAEL value from the Big Blue® rat oral mucosa study [28] is represented by the blue diamond, whereas the 7-day [31] and 90-day [30] duodenal MN NOAELs are represented by the downward and upward pointing blue triangles, respectively (Fig. 1). The latter symbol also represents negative results for kras codon 12 GAT mutations and crypt ^-H2AX immunostaining [30,32].

Notably, the blue diamonds and triangles representing target tissue genotoxic potency fall below the unity line in Fig. 1, and therefore do not follow the expected correlation implied by the larger set of compounds. In fact, even the blue circle for Cr(VI) blood MN data (conservatively derived from multiple negative datasets) also falls below the line. It should be noted that a main principle of the in vivo MN assay, which can detect clastogenic and aneu-genic damage, is that it be conducted in a proliferative tissue such as bone marrow, colon, small intestine, or stimulated liver [37-39]. The duodenum is therefore ideal for studying genotoxicity, as it is a highly proliferative tissue. Moreover, as noted in MacGregor et al. [2], genotoxicity tests should ideally be conducted in tissues that receive the highest dose of active metabolite or that are a site of carcinogenic concern. Following 90-days of Cr(VI) exposure, the highest levels of Cr have been observed in the duodenum [23,33], and the duodenum was the most sensitive target organ with respect to carcinogenicity from oral ingestion of Cr(VI) [14,27]. Thus, the duodenal MN data are more applicable than data from other tissues (e.g., blood) for assessing the genotoxic potency of Cr(VI) in the gastrointestinal tract.

The in vivo genotoxicity data for Cr(VI) provide two important insights. First, the mostly negative findings with respect to the in vivo blood MN assay was indeed predictive of a low/no 'systemic' carcinogenicity of Cr(VI) in both rats and mice, which is consistent with the general relationship described by Fig. 1. Second, the lack of genotoxicity in the very target organs that developed tumors lead to three potential conclusions with respect to the in vivo genotoxic potency of Cr(VI): 1) the genotoxicity assays were not sensitive enough to detect genotoxic damage, 2) the 'right' genotoxicity assays for Cr(VI) have not been conducted, or 3) the MOA does not involve direct genotoxicity.

With regard to the first possible conclusion accounting for the lack of Cr(VI)-associated genotoxicity, the in vivo genotoxicity assays assessing Cr(VI) used carcinogenic and near toxic concentrations. The MN assays were clearly negative in multiple intestinal analyses, and consistent with the lack of ^-H2AX immunostaining or XRF-based Cr detection in the crypt [30-32]. The ACB-PCR and Big Blue® TGR assays for detecting point mutations are both sensitive assays [40,41], and measured responses at all doses did not differ significantly from controls. For example, the ACB-PCR assay can detect mutations in the range of 10-5-10-1 [30,40]. Positive controls for genotoxic endpoints in the duodenum and oral mucosa induced unequivocal responses—thus positive results for Cr(VI) were unlikely to have been missed at the high, carcinogenic concentrations investigated.

With regard to the second possible conclusion, i.e. potential for not having conducted the 'right' genotoxicity assays, it must be acknowledged that standard assays have been used for assessing the genotoxicity of Cr(VI). Although the mutation analysis in the duodenum was specific for kras codon 12 GAT mutations, it is a mutation commonly found in intestinal tumors [42,43], and reported to induce proliferation in the intestine of mice [29]. Mutation analysis of the tumor tissue samples from mice in the 2-year cancer bioassay could provide insight into additional gene targets for ACB-PCR mutation analysis; however, such data are unavailable. We have also conducted toxicogenomic analyses for signs of aberrant Wnt/P-catenin signaling that might suggest mutations in the tumor suppressor gene, APC (adenomatous polyposis coli), or other genes in this pathway. Pathway enrichment analy-

sis using DAVID [44] identified many KEGG pathways associated with toxicogenomic responses to Cr(VI) at various exposures, yet no enrichment for Wnt/P-catenin signaling was observed (Supplemental Table S1). Toxicogenomic profiling analyses also indicated that the gene expression changes in the duodenum were more consistent with changes elicited by non-mutagenic than muta-genic carcinogens [45]. A TGR assay in Big Blue® mice might add additional weight; however, the current WOE—including the XRF mapping in the intestine and negative results in Big Blue® rats—suggests the results would be negative. Finally, other geno-toxicity assays in the duodenum might be informative (e.g., DNA methylation; comet assay, DNA repair capacity); however, positive responses would not necessarily support direct DNA reactivity or linear low-dose extrapolation for risk assessment purposes.

The third possible conclusion that can be drawn from the analyses herein, i.e. the MOA does not involve direct genotoxicity, is directly supportable by recent studies. Villus cytotoxicity and crypt hyperplasia have been observed in multiple studies on Cr(VI) in drinking water by H&E staining of transverse duodenal sections. A 90-day study by NTP observed 'diffuse epithelial hyperplasia' in the duodenum of male and female B6C3F1 mice exposed to >20ppm Cr(VI) [15]. A subsequent study reported 'crypt hyperplasia' in mice exposed to >60 ppm Cr(VI) for 90 days and 180 ppm Cr(VI) after only 7 days [23]. Cell proliferation was later quantified in Feulgen-stained sections by expansion of the crypt zone area and the number of crypt enterocytes [30], as well as the length of the crypt compartment [32]. Consistent with the XRF mapping of Cr in the duodenal mucosa, all of these studies have noted adverse effects in the villi, characterized by atrophy/blunting, vacuolation, and single cell necrosis. In a study conducted four years after Thompson et al. [23] in a different laboratory, using different pathologists and an alternative histological preparation (Feulgen staining of Swiss roll sections), crypt hyperplasia was again observed after only 7 days of exposure, as evidenced by an increase in the number of enterocytes per crypt in mice exposed to >20 ppm Cr(VI), as well as qualitative increases in the length/depth of the crypt compartment [31]. In a recent re-evaluation of H&E stained slides from NTP [14,15] and Thompson et al. [23,24], it was concluded that, "villus atrophy/blunting, enterocyte vacuola-tion, single cell necrosis, and crypt epithelial hyperplasia portray a process in which chemically-induced villus enterocyte cytotox-icity resulted in regenerative crypt epithelial hyperplasia" [46]. These data support that there is clear evidence for a MOA involving chronic cytotoxicity to intestinal villi that leads to chronic lifetime increases in crypt hyperplasia, and that this increase in cell division is likely the major contributor to intestinal carcinogenesis.

As noted earlier, tumors of the small intestine of mice are rare in the NTP/NCI database. According to Stout et al. [27], captan is the only other chemical to have induced duodenal tumors in an NTP/NCI study [26]. Studies on this fungicide (and structurally similar folpet) indicate a MOA involving villus toxicity (evidenced by blunting), crypt hyperplasia (evidenced by expansion of crypt compartment), and ultimately tumorigenesis [47,48]. The latterprocess is hypothesized to be the result of additional lifetime stem cell divisions required to chronically regenerate damaged mucosa [49]. Indeed, the U.S. EPA has concluded that, "captan induces adenomas and adenocarcinomas in the duodenum of the mouse by a nongeno-toxic MOA involving cytotoxicity and regenerative cell hyperplasia that exhibits a clear dose threshold...." [50].

Finally, in the genotoxicity versus cancer potency correlation analysis, it is worth investigating the identity of the two black circles well below the unity line in Fig. 1. One represents data for liver tumors induced by chloroform, which has long been considered to cause liver tumors through a non-mutagenic mechanism involving cytotoxicity and regenerative hyperplasia [51,52]. It is therefore not surprising that the BMDL5 for genotoxicity would be greater than

the BMDLio for carcinogenicity due to the apparent lack of involvement of direct genotoxicity in chloroform-induced carcinogenicity. The other black circle represents data for liver tumors induced by coconut oil acid diethanolamine condensate (CODC). According to the 2-year cancer bioassay for CODC [53], the increase in tumors "were associated with the concentration of free diethanolamine present as a contaminant in the diethanolamine condensate." It has been suggested that diethanolamine induces tumors in rodents via a non-mutagenic MOA involving choline deficiency, cytotoxicity, and increased cell proliferation [54,55]. Interestingly, it was the location of these circles below the unity line that motivated our investigation into the underlying data for these data points. Stated differently, the non-conformity of chloroform and CODC flagged these compounds as differing from the others. Investigation into the MOA for all datasets in Fig. 1 is beyond the scope of the present study, but the co-localizing of Cr(Vl) well below the unity line with chloroform and CODC suggests that Cr(Vl) indeed behaves differently from the broader set of compounds. We interpret this outcome, along with the empirical lack of genotoxicity in the crypt compartment, robust proliferation in the crypt compartment, and corroborative Cr localization outside the crypt compartment, as support for a non-genotoxic MOA for intestinal tumors.

4. Conclusions

The analyses herein highlight some of the complexities of conducting large-scale meta-analyses. For example, Soeteman-Hernandez et al. [1] elected to include the lowest BMDL5 values from MN studies, but in the case of Cr(Vl), this resulted in analysis of data from a rarely used strain of mice in lieu of multiple negative GLP studies in standard laboratory strains. Moreover, the notion of conducting dose-response analysis on genotoxicity data that are clearly negative (of which there are several for Cr(Vl)) deserves more consideration and comment from experts in the field. Some of the correlation analyses might also be influenced by how tumors are categorized. This, like genotoxicity endpoints, needs further comment from experts in veterinary pathology. Nevertheless, the work of Soeteman-Hernandez et al. [1] is to be commended and may ultimately advance human health risk assessment. lndeed, their work has been highlighted by experts in genotoxicity testing [2]. Herein, we have shown that the type of analyses conducted in Soeteman-Hernandez et al. [1] can be used to broadly inform the MOA of carcinogens in terms of the likelihood of genotoxic versus non-genotoxic mechanisms.

ln the case of Cr(Vl), several traditional MOA and AOP analyses have been published for the gastrointestinal tract (Table 1). Recent data provide prima facie evidence for a MOA involving villus cyto-toxicity and regenerative crypt hyperplasia—a MOA that has been recognized for other chemicals inducing similar duodenal lesions as Cr(Vl). Among the MOA and AOP analyses summarized in Table 1, the in vivo genotoxic and carcinogenic potency comparisons for Cr(Vl) herein comport with a non-mutagenic MOA. These quantitative analyses can therefore help select among competing MOAs and augment WOE evaluation. Additional case examples will be needed to further explore the utility of this approach in support of more traditional MOA and AOP analyses. For Cr(Vl), both approaches provide support for a non-mutagenic MOA for the tumors observed in the 2-year rodent cancer bioassay.

Acknowledgement

This work was supported by the Cr(Vl) Panel of the American Chemistry Council.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.mrgentox.2016.

01.008.

References

[1] L.G. Soeteman-Hernandez, G.E. Johnson, W. Slob, Estimating the carcinogenic potency of chemicals from the in vivo micronucleus test, Mutagenesis (2015).

[2] J.T. MacGregor, R. Frotschl, P.A. White, K.S. Crump D.A. Eastmond, S. Fukushima, M. Guerard, M. Hayashi, L.G. Soeteman-Hernandez, G.E. Johnson, T. Kasamatsu, D.D. Levy, T. Morita, L. Muller, R. Schoeny, M.J. Schuler, V. Thybaud, IWGT report on quantitative approaches to genotoxicity risk assessment ll. Use of point-of-departure (PoD) metrics in defining acceptable exposure limits and assessing human risk, Mutat. Res. 783 (2015) 66-78.

[3] L.G. Hernandez, J. van Benthem, W. Slob, Estimating the Carcinogenic Potency of Chemicals from in Vivo Micronucleus Test, RIVM, 2012.

[4] L.G. Hernandez, W. Slob, H. van Steeg, J. van Benthem, Can carcinogenic potency be predicted from in vivo genotoxicity data? A meta-analysis of historical data, Environ. Mol. Mutagen. 52 (2011) 518-528.

[5] G.T. Ankley, R.S. Bennett, R.J. Erickson, D.J. Hoff, M.W. Hornung, R.D. Johnson, D.R. Mount, J.W. Nichols, C.L. Russom, P.K. Schmieder, J.A. Serrrano, J.E. Tietge, D.L. Villeneuve, Adverse outcome pathways: a conceptual framework to support ecotoxicology research and risk assessment, Environ. Toxicol. Chem. 29(2010) 730-741.

[6] M.E. Meek, A. Boobis, l. Cote, V. Dellarco, G. Fotakis, S. Munn, J. Seed, C. Vickers, New developments in the evolution and application of the WHO/IPCS framework on mode of action/species concordance analysis, J. Appl. Toxicol. 34(2014) 1-18.

[7] C. Sonich-Mullin, R. Fielder, J. Wiltse, K. Baetcke, J. Dempsey, P. Fenner-Crisp, D. Grant, M. Hartley, A. Knaap, D. Kroese, l. Mangelsdorf, E. Meek, J.M. Rice, M. Younes, IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis, Regul. Toxicol. Pharmacol. 34 (2001) 146-152.

[8] U.S. EPA, Guidelines for Carcinogen Risk Assessment, EPA/F, Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, D.C, 2005.

[9] J. HaneyJr., Consideration of non-linear non-threshold and threshold approaches for assessing the carcinogenicity of oral exposure to hexavalent chromium, Regul. Toxicol. Pharmacol. 73 (2015) 834-852.

[10] Health Canada, Chromium in Drinking Water: Document for Public Consultation (2015).

[11] C.M. Thompson, D.M. Proctor, M. Suh, L.C. Haws, C.R. Kirman, M.A. Harris, Assessment of the mode of action underlying development of rodent small intestinal tumors following oral exposure to hexavalent chromium and relevance to humans, Crit. Rev. Toxicol. 43 (2013) 244-274.

[12] R.A. Becker, G.T. Ankley, S.W. Edwards, S.W. Kennedy, I. Linkov, B. Meek, M. Sachana, H. Segner, B. Van Der Burg, D.L. Villeneuve, H. Watanabe, T.S. Barton-Maclaren, Increasing scientific confidence in adverse outcome pathways: application of tailored Bradford-Hill considerations for evaluating weight of evidence, Regul. Toxicol. Pharmacol. 72 (2015) 514-537.

[13] N. McCarroll, N. Keshava, J. Chen, G. Akerman, A. Kligerman, E. Rinde, An evaluation of the mode of action framework for mutagenic carcinogens case study ll: chromium (VI), Environ. Mol. Mutagen. 51 (2010) 89-111.

[14] NTP, National Toxicology Program technical report on the toxicology and carcinogenesis studies of sodium dichromate dihydrate (CAS No. 7789-12-0) in F344/N rats and B6C3F1 mice (drinking water studies), NTP TR 546 NlH Publication No. 08-5887 (2008).

[15] NTP, National Toxicology Program technical report on the toxicity studies of sodium dichromate dihydrate (CAS No. 7789-12-0) administered in drinking water to male and female F344/N rats and B6C3F1 mice and male BALB/c and am3-C57BL/6 mice, NTP Toxicity Report Series Number 72, NlH Publication No. 07-5964 (2007).

[16] W. Slob, R. Cotton, proast61.3: benchmark dose modelling. R package version 61.3, RlVM2013.

[17] W. Slob, R.W. Setzer, Shape and steepness of toxicological dose-response relationships of continuous endpoints, Crit. Rev. Toxicol. 44 (2014) 270-297.

[18] A.E. Brix, J.F. Hardisty, E.E. McConnell, Combining neoplasms for evaluation of rodent carcinogenesis studies, in: C.H. Hsu, T. Stedeford (Eds.), Cancer Risk Assessment, John Wiley & Sons, 2010, pp. 699-715.

[19] E.E. McConnell, H.A. Solleveld, J.A. Swenberg, G.A. Boorman, Guidelines for combining neoplasms for evaluation of rodent carcinogenesis studies, J. Natl. Cancer lnst. 76 (1986) 283-289.

[20] J.B. Chen, V.N. Dobrovolsky, R.H. Heflich, Development of a mouse cell line containing the PhiX174 am3 allele as a target for detecting mutation, Mutat. Res. 444(1999)347-353.

[21] J. Bucher, NTP toxicity studies of sodium dichromate dihydrate (CAS No. 7789-12-0) administered in drinking water to male and female F344/N rats and B6C3F1 mice and male BALB/c and am3-C57BL/6 mice, Toxic. Rep. Ser. (2007) 1-G4.

[22] S. De Flora, M. lltcheva, R.M. Balansky, Oral chromium(Vl) does not affect the frequency of micronuclei in hematopoietic cells of adult mice and of transplacentally exposed fetuses, Mutat. Res. 610 (2006) 38-47.

[23] C.M. Thompson, D.M. Proctor, L.C. Haws, C.D. Hebert, S.D. Grimes, H.G. Shertzer, A.K. Kopec, J.G. Hixon, T.R. Zacharewski, M.A. Harris, lnvestigation of

the mode of action underlying the tumorigenic response induced in B6C3F1 mice exposed orally to hexavalent chromium, Toxicol. Sci. 123 ( 2011) 58-70.

[24] C.M. Thompson, D.M. Proctor, M. Suh, L.C. Haws, C.D. Hebert, J.F. Mann, H.G. Shertzer, J.G. Hixon, M.A. Harris, Comparison of the effects of hexavalent chromium in the alimentary canal of F344 rats and B6C3F1 mice following exposure in drinking water: implications for carcinogenic modes of action, Toxicol. Sci. 125 (2012) 79-90.

[25] M. Suh, C.M. Thompson, C.R. Kirman, M.C. Carakostas, L.C. Haws, M.A. Harris, [38 D.M. Proctor, High concentrations of hexavalent chromium in drinking water

alter iron homeostasis in F344 rats and B6C3F1 mice, Food Chem. Toxicol. 65 [39 (2014)381-388.

[26] NCI, Bioassay ofCaptan for possible Carcinogenicity, CAS No. 133-06-2, National Cancer Institute CARCINOGENESIS Technical Report Series, 15

(1977). [40

[27] M.D. Stout, R.A. Herbert, G.E. Kissling, B.J. Collins, G.S. Travlos, K.L. Witt, R.L. Melnick, K.M. Abdo, D.E. Malarkey, M.J. Hooth, Hexavalent chromium is carcinogenic to F344/N rats and B6C3F1 mice after chronic oral exposure, [41 Environ. Health Perspect. 117 (2009) 716-722.

[28] C.M. Thompson, R.R. Young, M. Suh, H.R. Dinesdurage, R.H. Elbekai, M.A. [42 Harris, A.C. Rohr, D.M. Proctor, Assessment of the mutagenic potential of

Cr(VI) in the oral mucosa of Big Blue((R)) transgenic F344 rats, Environ. Mol. Mutagen. 56 (2015) 621-628.

[29] Y. Feng, G.T. Bommer, J. Zhao, M. Green, E. Sands, Y. Zhai, K. Brown, A. [43 Burberry, K.R. Cho, E.R. Fearon, Mutant KRAS promotes hyperplasia and alters differentiation in the colon epithelium but does not expand the presumptive [44 stem cell pool, Gastroenterology 141 (2011) 1003-1013, e1001-e1010.

[30] T.J. O'Brien, H. Ding, M. Suh, C.M. Thompson, B.L. Parsons, M.A. Harris, W.A. [45 Winkelman, J.C. Wolf, J.G. Hixon, A.M. Schwartz, M.B. Myers, L.C. Haws, D.M. Proctor, Assessment of K-Ras mutant frequency and micronucleus incidence

in the mouse duodenum following 90-days of exposure to Cr(VI) in drinking water, Mutat. Res. 754 (2013) 15-21. [46

[31] C.M. Thompson, J.C. Wolf, R.H. Elbekai, M.G. Paranjpe, J.M. Seiter, M.A. Chappell, R.V. Tappero, M. Suh, D.M. Proctor, A. Bichteler, L.C. Haws, M.A.

Harris, Duodenal crypt health following exposure to Cr(VI): micronucleus [47

scoring, gamma-H2AX immunostaining, and synchrotron X-ray fluorescence microscopy, Mutat. Res. 789-790 (2015) 61-66.

[32] C.M. Thompson, J. Seiter, M.A. Chappell, R.V. Tappero, D.M. Proctor, M. Suh, [48 J.C. Wolf, L.C. Haws, R. Vitale, L. Mittal, C.R. Kirman, S.M. Hays, M.A. Harris, Synchrotron-based imaging of chromium and gamma-H2AX immunostaining

in the duodenum following repeated exposure to Cr(VI) in drinking water, [49

Toxicol. Sci. 143 (2015) 16-25.

[33] C.R. Kirman, S.M. Hays, L.L. Aylward, M. Suh, M.A. Harris, C.M. Thompson, L.C. [50 Haws, D.M. Proctor, Physiologically based pharmacokinetic model for rats and

mice orally exposed to chromium, Chem. Biol. Interact. 200 (2012) 45-64.

[34] NTP, National Toxicology Program technical report on the toxicology and [51 carcinogenesis studies of chromium picolinate monohydrate (CAS NO. 27882-76-4) in F344/N rats and B6C3F1 mice (feed studies), NIH Publication

No. 08-5897 (2008). [52

[35] F.J. Sanchez-Martin, Y. Fan, V. Carreira, J.L. Ovesen, A. Vonhandorf, Y. Xia, A. Puga, Long-term coexposure to hexavalent chromium and B[a]P causes tissue-specific differential biological effects in liver and gastrointestinal tract [53 of mice, Toxicol. Sci. 146 (2015) 52-64.

[36] J.A. Wignall, A.J. Shapiro, F.A. Wright, T.J. Woodruff, W.A. Chiu, KZ. Guyton, I. Rusyn, Standardizing benchmark dose calculations to improve science-based decisions in human health assessments, Environ. Health Perspect. 122 (2014) [54 499-505. [55

[37] S. Hamada, W. Ohyama, R. Takashima, K. Shimada, K. Matsumoto, S. Kawakami, F. Uno, H. Sui, Y. Shimada, T. Imamura, S. Matsumura, H. Sanada, K. Inoue, S. Muto, I. Ogawa, A. Hayashi, T. Takayanagi, Y. Ogiwara, A. Maeda, E.

Okada, Y. Terashima, H. Takasawa, K. Narumi, Y. Wako, K. Kawasako, M. Sano, N. Ohashi, T. Morita, H. Kojima, M. Honma, M. Hayashi, Evaluation of the repeated-dose liver and gastrointestinal tract micronucleus assays with 22 chemicals using young adult rats: summary of the collaborative study by the Collaborative Study Group for the Micronucleus Test (CSGMT)/The Japanese Environmental Mutagen Society (JEMS)—Mammalian Mutagenicity Study Group (MMS), Mutat. Res. 780-781 (2015) 2-17.

T. Morita, J.T. MacGregor, M. Hayashi, Micronucleus assays in rodent tissues other than bone marrow, Mutagenesis 26 (2011) 223-230. Y. Uno, T. Morita, M. Luijten, C. Beevers, S. Hamada, S. Itoh, W. Ohyama, H. Takasawa, Micronucleus test in rodent tissues other than liver or erythrocytes: report of the IWGT working group, Mutat. Res. 783 (2015) 19-22.

M.B. Myers, P.B. McKinzie, Y. Wang, F. Meng, B.L. Parsons, ACB-PCR quantification of somatic oncomutation, Methods Mol. Biol. 1105 (2014) 345-363.

OECD, Detailed Review Paper on Transgenic Rodent Mutation Assays, Series on Testing and Assessment, OECD, Paris, France, 2009. F. Luo, D.G. Brooks, H. Ye, R. Hamoudi, G. Poulogiannis, C.E. Patek, D.J. Winton, M.J. Arends, Mutated K-ras(Asp12) promotes tumourigenesis in Apc(Min) mice more in the large than the small intestines, with synergistic effects between K-ras and Wnt pathways, Int. J. Exp. Pathol. 90 (2009) 558-574. A. Wicki, R. Herrmann, G. Christofori, Kras in metastatic colorectal cancer, Swiss Med. Wkly. 140 (2010) w13112.

D.W. Huang, B.T. Sherman, R.A. Lempicki, Systematic and integrative analysis of large gene lists using DAVID bioinformatics resource, Nat. Protoc. 4 (2009). C.M. Thompson, J. Gregory Hixon, D.M. Proctor, L.C. Haws, M. Suh, J.D. Urban, M.A. Harris, Assessment of genotoxic potential of Cr(VI) in the mouse duodenum: an in silico comparison with mutagenic and nonmutagenic carcinogens across tissues, Regul. Toxicol. Pharmacol. 64 (2012) 68-76.

J.M. Cullen, J.M. Ward, C.M. Thompson, Re-evaluation and classification of duodenal lesions in B6C3F1 Mice and F344 rats from four studies of hexavalent chromium in drinking water, Toxicol. Pathol. 44 (2016) 279-289. S.M. Cohen, E.B. Gordon, P. Singh, G.T. Arce, A. Nyska, Carcinogenic mode of action of folpet in mice and evaluation of its relevance to humans, Crit. Rev. Toxicol. 40 (2010) 531-545.

E. Gordon, Captan: transition from 'B2' to 'not likely'. how pesticide registrants affected the EPA cancer classification update, J. Appl. Toxicol. 27 (2007)519-526.

S.M. Cohen, L.B. Ellwein, Cell proliferation in carcinogenesis Science 249 (1990) 1007-1011.

U.S. EPA, Captan; Cancer Reclassification; Amendment of ReregistrationEligibility Decision; Notice of Availability Federal Register 69 (2004) 68357-68360.

R.J. Golden, S.E. Holm, D.E. Robinson, P.H. Julkunen, E.A. Reese, Chloroform mode ofaction: implications for cancer risk assessment, Regul. Toxicol. Pharmacol. 26 (1997) 142-155.

U.S. EPA, Toxicological review of chloroform in support of summary information on the integrated risk information system (IRIS), U.S Environmental Protection Agency, Washington, D.C, 2001. NTP, National Toxicology Program technical report on the toxicology and carcinogenesis studies ofcoconut oil acid diethanolamine condensate (CAS NO. 68603-42-9) in F344/N rats and B6C3F1 mice (dermal studies), NIH Publication No. 01-3969 (2001).

MAK, The MAK Collection for Occupational Health and Safety (2015). H.W. Leung, L.M. Kamendulis, W.T. Stott, Review of the carcinogenic activity of diethanolamine and evidence of choline deficiency as a plausible mode of action, Regul. Toxicol. Pharmacol. 43 (2005) 260-271.