Induction of tunica vaginalis mesotheliomas in rats by xenobiotics

To better understand the relevance of tunica vaginalis mesotheliomas (TVM) to human cancer risk, we examined the nature of TVM responses in 21 published rat cancer bioassays against the backdrop of the biology and molecular biology of mesothelium, and of spontaneous and treatment-induced TVM. Although relatively rare in all species including humans, TVM are seen most frequently in F344 male rats, as opposed to other rat strains, and are causally associated with the high background incidence of Leydig-cell tumors of the testes of these rats. Hormone imbalance brought about by perturbations of the endocrine system is proposed as a key factor leading to both spontaneous and treatment-associated TVM. Of 21 F344 rat studies with a treatment-associated TVM response, 7 were judged to have a nonsignificant to marginal response, 11 had a robust TVM response, and 3 were noninformative due to early mortality from other induced tumors. Of the 11 chemicals with robust responses, 8 were directly mutagenic in Salmonella and 3 are known to be mutagenic after metabolism. Only 2 of the 7 with nonsignificant to marginal responses were Ames test positive. TVM induction is a male F344 rat-specific event, and chemicals/agents that induce only TVM in the male F344 rat from a typical two-sex rat and mouse chronic bioassay are likely irrelevant in human risk assessment.

Keywords: Human relevance; genotoxicity; hormonal impalance; cancer bioassay; leydig cell tumors; acrylamide; F344 rat

Spontaneously occurring mesotheliomas have been documented in a wide range of animals but are relatively rare. They have been observed in humans, lower vertebrates, domesticated and laboratory-reared mammals, avian species, and marsupials, and occur in the thoracic and abdominal cavities ([77]; [34]; [34]) with rare reports of atriocaval mesotheliomas in cardiac chambers ([73]; [143]; [26]). Spontaneous mesotheliomas, which occur primarily in the scrotal sac and peritoneal cavity, have been documented in various rat strains, with the highest frequency occurring in male Fischer 344 rats ([158]; [40]; [144]; [63]). These peritoneal mesotheliomas occur in rats 20 to 24 mo of age or older and arise in the mesothelium investing the testis, epididymis, and scrotal sac, and may extend or seed into the peritoneal cavity.

Mesotheliomas can be induced by a wide variety of agents, including various forms of asbestos, other natural and man-made fibers, metals, viruses, synthetic estrogens, and individual chemicals ([77]; [76]; [145]). Recently, multiwalled nanotubes injected intraperitoneally have been shown to induce peritoneal mesotheliomas in mice ([160]). Depending upon the route of exposure, induced mesotheliomas also can occur in the thoracic or peritoneal cavity.

While the diagnostic terms for mesotheliomas used in the studies reviewed in this document include testicular mesothelioma, epididymal mesothelioma, peritoneal mesothelioma, and malignant mesothelioma, all are considered to have arisen in the tunica vaginalis mesothelium. Morphologically, tunica vaginalis mesotheliomas are typically less invasive and have fewer stromal components than the more familiar asbestos-induced pleural mesotheliomas. Tunica vaginalis mesotheliomas in rats rarely metastasize, and are confined to the scrotal sac and abdominal cavity.

This review provides a brief overview of basic biology, key events, and examples of xenobiotics that have been associated with tunica vaginalis mesotheliomas in F344 rats based on the National Toxicology Program (NTP) database and a search of literature. Key events are important but not necessarily sufficient steps in the process of tumorigenesis. This review was undertaken to understand and evaluate the relevance of this unique F344 rat tumor to human health risk assessment. Because tumors initiated by direct DNA interaction (genotoxic mechanisms) are regulated in a different fashion from those that arise from non-DNA reactive modes of action, it is important to understand the etiology of these tumors and whether their induction in the rat is relevant to humans. We postulate that the high incidence of Leydig-cell tumors in the F344 rat is causally linked to development of tunica vaginalis mesotheliomas.

During early embryogenesis the coelom is a common cavity of mesodermal origin that will ultimately form the pleural, peritoneal, and cardiac cavities and mesothelial linings. This mesoderm forms two types of epithelial cells, namely, mesothelium, which is a squamous cell that forms from mesoderm and lines body cavities, and endothelium, which is a squamous cell that lines vascular and lymphatic channels ([14]). During development, the septum transversum, which will become the future diaphragm, separates the pleuropericardial membranes from the peritoneal membranes to form the separate pleural and peritoneal cavities ([69]; [9]). The peritoneal cavity and its contained abdominal organs are lined by a single layer of flattened mesothelial cells supported by delicate fibrous connective tissue. Peritoneal mesothelium extends into the scrotum and lines the surfaces of the testes, epididymis, and mesorchium, where it is referred to as the tunica vaginalis. Since mesothelial linings in the pleural, peritoneal and scrotal cavities all derive embryologically from the same coelomic mesoderm, it is reasonable to expect the biology and pathobiology of thoracic and peritoneal mesothelium to have common attributes. Furthermore, since mesothelium embryologically derives from mesoderm, it is not surprising that new mesothelium can arise from existing adjacent mesenchyme during wound healing in serous cavities ([101]).

Spontaneously occurring mesotheliomas are rare and, in general, are more commonly seen in males. A comprehensive listing of mesotheliomas in animals and humans can be found in the publications by Ilgren ([76]; [77]). The background incidences of spontaneous mesotheliomas in rats, range from 0.2 to 5%. With rare exceptions, rat mesotheliomas occur in aged males, originate in the tunica vaginalis, and may spread by extension or seeding into the peritoneal cavity. Spontaneous mesotheliomas have been seen in Wistar, Sprague-Dawley, and other rat strains ([144]; [40]), but most reports and descriptions in the literature are based on examples in Fischer 344 males ([59]; [155]; Shibuya et al., 1990; [162]; [63]; [69]; [114]).

Chemical-exposure-associated tunica vaginalis mesotheliomas in rats have been identified in several cancer bioassays conducted for safety assessment or hazard identification, as well as in specific research studies. With the exception of reduced latency and an increased tendency to extend into the peritoneal cavity, the pathological features of these treatment-associated tumors are indistinguishable from those in concurrent controls and spontaneous cases reported in the literature.

The histomorphology of tunica vaginalis and peritoneal mesotheliomas in rats is similar in spontaneous and chemically induced lesions, and is histologically indistinguishable from tunica vaginalis and peritoneal mesotheliomas in other species, including humans. Mesotheliomas vary from complex papillary to sessile nodular growths with a sarcomatous component. Smaller papillary lesions consist of a fibrovascular stroma lined by a single layer of flattened to cuboidal mesothelial cells, while larger papillary structures may have areas covered by multiple irregular layers of mesothelial cells forming a pavement or stratified pattern. Tumor cells are cuboidal to polygonal with round to oval nuclei and a prominent nucleolus, and may be arranged in solid sheets, nests, or in glandular and tubular structures. They can form cystic structures in which connective tissue cyst walls are lined by flattened to cuboidal mesothelial cells. The tumor cells may occasionally contain iron-positive material, are mucicarmine positive, and are typically positive for hyaluronic acid. Intracellular keratin and vimentin, and Wilm's tumor 1 (WT-1), can be detected by immunohistochemistry. Mesotheliomas in rats can be classified as epitheliomatous, sarcomatous, or mixed. This classification scheme is consistent with classification of mesotheliomas in humans. As in humans, rat peritoneal mesotheliomas arising in the tunica vaginalis may have features of malignancy, including pleomorphism, cytological atypia, and local invasiveness. Ultrastructurally, mesothelioma cells rest on a distinct basal lamina, and have microvilli, junctional complexes, abundant cytofilaments, pinocytotic vesicles, dilated rough endoplasmic reticulum (RER) cisternae, and a prominent Golgi apparatus ([36]).

As with most well-studied cancers, a spectrum of lesion progression ranging from hyperplasia to benign neoplasia and ultimately to malignant neoplasia is characteristic of tunica vaginalis and peritoneal mesotheliomas in rats. Mesothelial hyperplasia ranges from a focal or multifocal increased density of usually plump to cuboidal mesothelial cells arranged as a single layer lining a serosal surface, to a blunt but small papillary projection lacking a fibrovascular stalk but sometimes associated with a small amount of connective tissue. Benign mesothelioma typically forms as a papillary structure with single and stratified layering of mesothelial cells lining a fibrovascular stalk. Non-papillary growth patterns of stratified mesothelial cells on a fibrous tissue base may also be considered benign. The mesothelial cells in benign mesotheliomas are generally cuboidal to polygonal and uniform. It is easy to appreciate that benign mesothelial lesions represent a morphological continuum with hyperplasia, and differences of opinion relative to diagnoses between the two would not be surprising. Malignant mesotheliomas have a spectrum of easily recognized morphological features including cellular and nuclear atypia, a pleomorphic growth pattern, and invasion through the serosa, and typically involve multiple sites throughout the peritoneal cavity. Because malignant mesotheliomas can form glandular and tubular structures, they must be distinguished from metastatic adenocarcinoma. The lack of a primary adenocarcinoma elsewhere in the body, and the use of immunohistochemical staining, support a diagnosis of malignant mesothelioma. Organizations such as the National Toxicology Program (NTP) do not subclassify mesotheliomas, but rather consider all mesotheliomas to be potentially malignant. In contrast, literature reports often consider mesotheliomas confined to the scrotal tunics, and without localized invasion, to be benign, while those that spread to the peritoneal cavity and are pleomorphic with cellular atypia and invasive features are generally considered malignant.

Distinguishing mesotheliomas from adenocarcinomas is an important consideration in diagnosis of human cases, especially for lesions in the thoracic cavity. Consequently, a battery of stains, including immunohistochemical stains, has been used to assist in the diagnosis ([140]) (Table 1). While these stains can help in differential diagnosis, they do not distinguish benign from malignant mesotheliomas ([53]). Several of these staining methods work well with rodent tissues (e.g., cytokeratin, WT-1, and calretinin) although not all have been applied to rodent mesotheliomas as yet. Before the advent of immunohistochemistry staining batteries, staining for the acid mucopolysaccharide hyaluronic acid was commonly used to distinguish mesotheliomas from adenocarcinomas. Hyaluronic acid can be identified by Alcian blue (pH 2.5) staining with and without hyaluronidase. Mesotheliomas are generally periodic acid Schiff (PAS) negative after diastase treatment.

Table 1.  Some immunohistochemical (IHC) stains helpful in distinguishing human mesotheliomas from adenocarcinomasa.

16 Note. No single IHC stain is sufficiently specific and sensitive for mesothelioma. A battery of IHC stains is helpful to confirm a diagnosis. Positive staining of mesotheliomas is most uniformily present in epithelial subtypes with stromal components often negative. HBME-1 = mouse monoclonal antibody raised against human mesothelioma cells. a[13]; [62]; [58]; Abutaily et al., 2002.

Mesothelial cells are relatively easy to culture in vitro, where they can undergo spontaneous as well as treatment-induced transformation and gain malignant phenotypes ([95]). Consequently, much of the literature on the biology of normal and transformed mesothelium derives from in vitro studies. Similarly, cells derived from spontaneously occurring and induced mesotheliomas have been studied with in vitro test systems. Based on the similar embryological origin of the pleural and peritoneal mesothelium, it is reasonable to assume a similar biology in cell cultures derived from either of these tissue sites.

Normal and spontaneously transformed rat mesothelial cells studied in vitro express cytochrome P-450 1A1 (CYP1A1) and CYP1B1 mRNAs, which are decreased in transformed cells and in asbestos-induced mesothelioma cells from Wistar rats ([95]). P-Glycoprotein, the mdr1 gene product, was not detected in normal mesothelial cells. Furthermore, mRNA for the Ah (aryl hydrocarbon) receptor and ARNT (aryl hydrocarbon receptor nuclear translocator), proteins that regulate induction of CYP enzymes via signal transduction in the cell nucleus, did not differ among the various cultured cells. The relevance of these in vitro findings relates to the biological functions of the studied proteins. The CYP enzymes potentially oxidize xenobiotics in some cases to metabolites that can induce cellular toxicity and carcinogenicity unless eliminated from the organisms by conjugation with glutathione or other cell substances, and in other cases detoxify xenobiotics to polar less toxic substances. P-Glycoprotein is a transmembrane pump that functions to eliminate xenobiotics from cells. Its absence in mesothelial cells suggests that the cells are not able to eliminate potentially harmful xenobiotics by this specific mechanism.

Insulin-like growth factors (IGFs) are polypeptides that are associated with cell proliferation and differentiation. Cell lines from normal rat mesothelium and from spontaneous rat peritoneal mesotheliomas express RNA transcripts for IGF2, but cell lines from asbestos-induced rat mesotheliomas do not ([151]). Since all three cell types have receptors for IGF2, as well as for IGF1 and insulin, the expression of IGF2 in the normal rat mesothelium and in the spontaneous mesothelioma indicates the probability that IGF2 is functioning as an autocrine growth factor, and suggests that asbestos-induced mesotheliomas arise through a different transformation pathway than do spontaneous mesotheliomas.

The basic immunobiology associated with mesotheliomas is poorly understood. Using a mouse model of malignant mesothelioma, [19]) showed significant production of the cytokines transforming growth factor (TGF)-beta, interlukin-6 (IL-6), IL1, and tumor necrosis factor (TNF), by the mesothelioma cells. The authors suggested that the elaboration of these factors by the mesothelioma cells is contributory to sabotaging antitumor host defenses and can induce perturbations in immune surveillance.

Oncogenes appear to play a minor role in the pathogenesis of mesotheliomas. [137]) found no point mutations in H-, K-, or N-ras proto-oncogenes, or the p53 tumor suppressor gene, in three ferric nitrilotriacetate-induced peritoneal mesotheliomas in Wistar rats. In an analysis of 17 human and 22 rat asbestos-induced mesotheliomas, no mutations in exons 12, 13, or 61 of the K-ras proto-oncogene were identified by direct DNA sequence analysis ([136]). There is some evidence, however, that the early response gene pathway leading to chronic stimulation of cell proliferation is involved in asbestos-induced rat mesotheliomas. A dose-dependent induction of c-fos and c-jun mRNA in rat mesothelial cells by asbestos leads to persistent induction of activator protein 1 (AP-1) transcription factors that drive the cell proliferation process ([71]). Thus, this early response gene pathway involved in asbestos-induced rat mesotheliomas leads to chronic stimulation of cell proliferation. The fibrous geometry of the particulates appears to be critical in induction of c-fos and c-jun in rat pleural mesothelial cells, with crocidolite and chrysotile asbestos causing a more dramatic increase in these early response genes than nonfibrous particles ([78]). There is also some evidence that this induction of c-fos and c-jun in rat mesothelial cells by asbestos is not directly triggered by active oxygen species generation. The initial response of rat mesothelial cells to active oxygen species is an increase in antioxidant enzymes followed by induction of c-fos and c-jun, secondary to a redox-sensitive component in the signaling cascade influenced by intracellular thiol (glutathione) levels. ([79]). Although there have been a number of studies of the role of c-fos and c-jun in asbestos-induced mesotheliomas, there have been no similar studies of chemically induced tumors.

In contrast to oncogenes, tumor suppressor genes (TSG) appear to play a more important role in mesothelial tumorigenesis. Alterations in tumor suppressor genes are characteristic of human malignant mesotheliomas ([8]) and are also seen in murine mesothelioma animal models ([87]). In general, TSG are important regulators of cell cycle machinery. In human malignant mesotheliomas there is frequent inactivation of Nf2 (neurofibromatosis type 2) and loss of p16(INK4a)—a p16 gene that regulates cell cycle—secondary to deletion of the CDKN2A/ARF (cyclin-dependent kinase inhibitor 2A/ADP ribosylation factor) locus. There are also indications of alterations in p19(ARF) (a tumor suppressor that attenuates degradation of p53), AKT (a family of genes that encode protein kinases), and WT-1. Genetic alterations in p16 and Nf2, both of which are important regulators of the cell cycle, have been identified in human malignant pleural mesothelioma and in asbestos-exposed, Nf2-deficient mice ([81]). These studies show a similar profile of tumor suppressor gene (TSG) alteration in asbestos-induced mesotheliomas in mice and humans. Inactivation of Nf2 is typically associated with tumors of neuroectodermal origin. P16/CDKN2A, as a tumor suppressor gene, is an important inhibitory protein that maintains the necessary balance between cyclin activation of cell proliferation and inhibition of the uncontrolled cell division that is characteristic of cancer cells. It is also potentially important in cell motility and invasiveness ([87]).

In another study, alterations of p16, 85% of which were homozygous deletions, were present in all 40 human malignant mesothelioma cell lines examined, and homozygous deletions were present in 5 of 23 (22%) primary malignant mesotheliomas ([29]). Nf2 mutations were detected in 8 of 15 (53%) human malignant mesothelioma cell lines, nearly all of which were confirmed in matched primary tumor DNAs ([18]). Asbestos-exposed Nf2(+/–) knockout mice had significantly accelerated mesothelioma development compared with similarly exposed wild-type littermates ([4]). Biallelic inactivation secondary to loss of the wild-type allele occurred in all the knockout mice and in 50% of the wild-type mice. Alterations in p19/Arf and p15/Cdkn2b were frequent in asbestos-treated mice hemizygous for Nf2, with similar alterations in human mesothelioma cell cultures ([100]). These same authors also noted loss of heterozygosity for Nf2, as was noted by [4]).

No p53 mutations were detected in an analysis of 17 human and 22 rat asbestos-induced mesothelioma tissue samples ([136]), and neither spontaneous rat mesotheliomas nor erionite-induced mesotheliomas in rats were found to have p53 alterations ([94]). On the other hand, there was a low rate of p53 mutations in mesothelioma cells from asbestos-treated Nf2 hemizygous mice ([100]). While p53 does not appear to play a major role in malignant mesotheliomas, there is an accelerated development of asbestos-induced mesotheliomas in heterozygous p53 +/– mice ([168]). As the tumors develop in these mice there is loss of heterozygosity accompanied by genetic instability, decreased apoptosis, and accelerated tumor growth and invasiveness. The murine Nf2+/– model of environmental carcinogenesis is remarkably similar to human malignant mesothelioma and recapitulates many molecular features of the human tumor ([5]).

The WT-1 suppressor gene is expressed in normal and neoplastic mesothelial cells in rats and humans ([174]), and immunohistochemical staining for WT-1 is useful in distinguishing mesotheliomas from adenocarcinomas and other neoplasms.

From these findings regarding tumor suppressor genes, it is apparent that there is considerable genetic instability in both human cases and mouse models of mesothelioma, and that multiple TSG are involved in mesothelial tumorigenesis. It is likely that multiple molecular events, interacting either sequentially or in the aggregate, are involved in the development of mesotheliomas.

AKT is a protein kinase that is important in mammalian cell signaling. It plays an important role in tumorigenesis and therapeutic resistance and is frequently inactivated in human malignant mesotheliomas, as well as in Nf2(+/–) mice ([4], [5]).

A number of different growth factors associated with proliferation of normal and neoplastic mesothelial cells have been documented, and much of what has been learned about these factors was generated from in vitro cell culture studies.

Normal rat pleural mesothelial cells exposed in vitro to long carcinogenic mineral fibers upregulate epidermal growth factor receptor (EGFR), with increases in EGFR protein occurring 24 h prior to initiation of the protein kinase mitogenic signaling cascade leading to increased cellular proliferation ([46]). Furthermore, fibers with greater potential to cause mesothelioma induce a more marked upregulation of EGFR than less carcinogenic fibers. The EGFR response is linked to phagocytosis of the mineral fibers by the rat mesothelial cells.

The bioactivity of TGF-beta in two mesothelioma cell lines established from spontaneous rat mesotheliomas was 30 to 70 times higher than in normal rat mesothelium ([97]). Based upon application of exogenous TGF-beta to the mesothelioma cell lines and normal rat mesothelial cells, the authors suggested that rat mesothelioma cells produced TGF-beta through an autocrine mechanism that stimulates their growth.

Using asbestos-induced murine mesothelioma models, it was noted that TGF-beta production by mesothelioma cells may permit their escape from immune surveillance based on downregulation of lymphocyte surface markers ([19]). TGF-beta 1 and 2 isoforms are expressed by both human and murine malignant mesothelial cells, and inhibition of TGF-beta by antisense RNA reduces the anchorage-independent growth of malignant mesothelial cells in vitro and their tumorigenicity in vivo ([48]). Inhibition of TGF-beta also led to increased T-lymphocyte infiltration into tumors. Thus, it appears that TGF-beta has tumor-enhancing effects in mesothelial tumorigenesis.

Altered expression of platelet-derived growth factor (PDGF) is characteristic of human mesotheliomas. There is no expression of PDGF in asbestos-induced rat mesotheliomas, although the PDGF receptors are highly expressed ([172]). The species differences between human and rat mesothelioma cells suggest that expression of PDGF may be species specific, at least for asbestos-induced mesotheliomas.

The growth factors TGF-beta, epidermal growth factor (EGF), and PDGF all independently stimulate a round of cell proliferation in serum-deprived, quiescent, primary normal human mesothelial cells ([54]). When the growth medium is supplemented with chemically denatured serum, these same growth factors can sustain continuous replication of mesothelial cells. Based on the responses to PDGF and TGF-beta, the authors concluded that mesothelial cells have growth regulatory properties similar to connective tissue cells. Normal human mesothelial cells secrete more TGF-beta than mesothelioma cell lines. In contrast, mesothelioma cell lines secrete more PDGF than normal human mesothelioma cells ([57]).

TGF-alpha is expressed in asbestos-transformed rat mesothelial cells but not in spontaneously transformed mesothelial cells, while both cell types express functional EGF receptors ([173]). Although TGF-alpha inhibits the growth of spontaneously transformed mesothelial cells, it also functions in an autocrine growth control fashion to stimulate growth of asbestos-transformed mesothelial cells ([173]). The implication of this study is that differences in mesothelioma etiology may be responsible for differences in the molecular biology of these neoplasms.

Based upon vascular endothelial growth factor (VEGF) expression levels and VEGF blocking by neutralizing antibodies in four human malignant mesothelioma cell lines, as well as in biopsies of malignant mesothelioma, VEGF appears to be a key regulator of malignant mesothelioma cell growth ([159]). Since malignant mesothelioma cells also express the tyrosine kinase-related VEGF receptors Flt-1 (a tyrosine-protein kinase) and kinase insert domain receptor (KDR), VEGF is believed to function as an autocrine growth factor in human malignant mesothelioma.

Cell lines from normal rat mesothelium, as well as spontaneous and asbestos-induced mesothelioma cell lines, all express IGF1, IGF2, and insulin receptors. However, there is ubiquitous expression of IGF2 (important in cell proliferation and differentiation) by normal rat mesothelium and spontaneous mesothelioma cell lines but not by asbestos-induced mesothelioma cell lines ([151]). Hence, IGF2 appears to function as an autocrine or paracrine growth factor in normal and spontaneously altered rat mesothelial cells. The authors suggested that changes in growth factor expression may be a consequence of different pathways of cell transformation.

Immunostaining of human malignant mesothelioma tissue specimens shows elevated expression of phosphorylated/activated AKT kinases, which are protein kinases important in mammalian cell signaling ([5]). Hepatocyte growth factor (HGF)/met receptor signaling in human and murine malignant mesothelioma cell lines is associated with HGF-inducible AKT activity, and suggests that this pathway may be amenable to targeted pharmacological therapy ([5]).

In a study of the gene expression profile of rat peritoneal mesotheliomas induced by o-nitrotoluene or bromochloroacetic acid, [92]) utilized Ingenuity Analysis Pathway software to identify 169 cancer-related genes. They identified activated IGF-1, p38 MAPkinase (p38 mitogen-activated protein kinase), Wnt/beta-catenin (wingless-type gene that is the homolog of the mouse int-1 oncogene), and integrin signaling pathways in these tumors. The authors concluded that the mesotheliomas induced by these two agents were similar to human mesotheliomas with respect to their cellular and molecular features.

In summary, based on several in vitro studies, effects on cell signaling and cell proliferative responses in normal and transformed mesothelium are influenced by several growth factors and cytokines functioning in an autocrine fashion.

Much of our knowledge of the molecular biology of mesotheliomas has been derived from studies using primary and established cultures of normal and transformed mesothelium, as well as cell lines derived from human and rodent mesotheliomas (see Biology/Molecular Biology section of this review). New cell lines are being continually established and described (e.g., [141]; [169]; [38]; [107]; [86]).

In a recent review, [87]) briefly discussed animal models of mesothelioma, including genetically modified mouse models. Intraperitoneal and intrapleural injections of rodents with asbestos results in malignant mesotheliomas which are similar to human mesotheliomas with regard to latency, patterns of growth, and development of ascites ([44]; [171]; [1]; [153]; [38]). Lymphatic metastasis and invasion of abdominal adipose tissue and diaphragm muscle resemble cases of diffuse malignant mesothelioma in humans ([4]). Murine peritoneal mesotheliomas have histopathological growth patterns and phenotypic markers including cytokeratins, N-cadherin, and WT1 that are seen in human diffuse malignant mesotheliomas ([85]).

While only a minority of human malignant mesotheliomas carry p53 mutations ([87]), heterozygous p53-deficient mice have accelerated development of asbestos-induced peritoneal mesothelioma ([168]). Heterozygous Nf2-deficient mice also show accelerated development and increased invasiveness of peritoneal mesotheliomas following exposure to crocidolite asbestos ([49]; [4]). The relevance of this model relates to common occurrence of molecular alterations in Nf2 in human malignant mesothelioma. A subset of asbestos-exposed heterozygous Nf2-deficient mice develop mesotheliomas with loss of p53, possibly due to the colocalization of Nf2 and p53 on mouse chromosome 11 ([87]). The reported cooperativity between Nf2 and p53 would be expected to increase the invasive and metastatic potential of the induced mesotheliomas ([110]; [109]). In asbestos-induced murine mesotheliomas in heterozygous Nf2-dificient mice, there is constitutive activation of the Akt pathway ([5]), a pathway frequently upregulated in human mesotheliomas and a key pathway in cell growth and proliferation. It is also noteworthy that the majority of mesotheliomas induced in heterozygous Nf2-deficient mice exhibit co-deletion of p16(Ink4a) and p19(arf) ([87]), which are frequently observed alterations in human malignant mesotheliomas ([5]).

Following the identification of SV40 contamination of polio vaccine in the mid-1950s and the induction of mesotheliomas in hamsters with polio vaccine extract, concern about a role of SV40 in human malignancies was raised (Fischer, 2005; [30]). Implication of SV40 as a cofactor in asbestos-induced human mesotheliomas, the observation that SV40 can immortalize human mesothelium in vitro, and the identification of SV40 sequences in human mesotheliomas have led to considerable controversy regarding a role for SV50 in human mesotheliomagenesis ([55]; [165]; [93]; [164]; [43]). While most of the research progress on the relationship of SV40 and human malignancies has come from studies of mesothelioma, inherent difficulties with epidemiological studies and technical concerns regarding molecular detection due to low copy numbers of DNA sequences in tumors contribute to the ongoing controversy ([149]; [80]). Identification of SV40 in tumors might merely indicate its presence as a passenger virus rather than demonstrating a causal relationship (Fischer, 2005). Genetically engineered mice with SV40 T-antigen under control of regulatory elements of the cytokeratin 19 gene develop several epithelial neoplasms in addition to a moderate frequency of mesotheliomas, but due to fertility problems this model is not readily available ([66]).

Hormone imbalance brought about by perturbations of the endocrine system has been proposed as a key event ultimately leading to both spontaneous and treatment- associated tunica vaginalis mesotheliomas in rats ([166]; [162], [156]). The feasibility of a hormonally driven process was originally appreciated based on the observation that diethylstilbestrol induced mesotheliomas on the genital organs in both sexes of dogs ([142]). Decreased testosterone in aging rats leads to Leydig-cell hyperplasia and ultimately Leydig-cell tumors ([166]). This aging change is especially dramatic in the F344 rat, which has a high spontaneous incidence of Leydig-cell tumors (range 88 to 96%), in contrast to other rat stocks used in chronic studies ([20]; [104]; [161]; [158]). For example, based on Leydig-cell hyperplasia, it has been proposed that testicular aging changes seen at 12 mo in F344 rats ([88]) are equivalent to testicular aging changes in 2-yr-old Wistar rats. The occurrence of Leydig-cell tumors, in turn, is postulated to be causally linked to development of tunica vaginalis mesotheliomas in F344 rats ([166]).

In the sexually mature rat, both leuteinizing hormone (LH) and leuteinizing hormone releasing hormone (LHRH) stimulate Leydig cells to produce testosterone ([23]; [147]). The testicular LH receptors and the serum testosterone levels decrease in rats between ages 4 and 18 mo. In this age range, the testicular LH receptors and testosterone levels are correlated and balanced. As the testosterone levels decline with age or as a stress-related disturbance ([138]), there is a compensatory increase in circulating LH to increase the level of testosterone. The compensatory action results in an increase (hyperplasia) of Leydig cells to increase testosterone levels. Ultimately the compensation is inadequate to maintain youthful levels of testosterone and the testicular–LH interaction strikes a new balance at a lower level ([6]). The ratio of the two is the same as before, but the levels are lower. LH continues to stimulate the Leydig cells to divide in an attempt to reach youthful levels of testosterone, resulting in progression of the proliferating Leydig cells from hyperplasia to Leydig-cell tumors. The testosterone–LH ratio changes once Leydig-cell tumors are formed. Leydig-cell tumors produce less testosterone than normal Leydig cells. Thus, an age- associated hormonal imbalance persists in older rats bearing Leydig-cell tumors. In addition to decreased testosterone, there is an increase in Leydig-cell LH receptors, an increase in serum progesterone, decreased prolactin, and decreased LH. In other words, the balance between testicular LH receptor levels and serum testosterone that was present during the 4 to 18 mo age interval changes, and the levels of the different hormones become unbalanced in the presence of Leydig-cell tumors.

Perturbations in the hypothalamic–pituitary–testis axis lead to Leydig-cell proliferation, based on circulating levels of both LH and LHRH and the number of their cognate receptors on Leydig cells. While it may at first seem counterintuitive, increases as well as decreases in prolactin levels can affect the hypothalamic–pituitary–testis axis and lead to Leydig-cell hyperplasia and Leydig-cell tumors.

The decrease in testosterone that ultimately leads to Leydig-cell proliferation can also be brought about by an age-related increase in prolactin production in rats ([106]; [45]; [24]; [166]). The increased prolactin leads to decreased gonadotrophin-releasing hormone (GnRH; LHRH) as well as decreased LH secretion. Since rat Leydig cells have LHRH receptors that are responsive to LH and LHRH, the hormonal cascade initiated by increased prolactin leads to reduced testosterone production, as is reflected by the decreased serum testosterone levels seen in the aging rat ([106]). It is important to note that while rat Leydig cells have LHRH receptors, human Leydig cells do not ([147]).

Alternatively, decreased prolactin production may occur secondary to the action of dopamine agonists on the hypothalamus ([147]; [33]). The decreased prolactin leads to a decrease in LH receptors on the Leydig cells and thereby results in reduced testosterone production. This then causes a compensatory increase in circulating LH and a sustained increase in circulating LH results in Leydig-cell hyperplasia and Leydig-cell tumors ([147].

The proof that age-related hormonal perturbation leads to Leydig-cell tumors in the rat is supported by experiments in which Leydig-cell hyperplasia and Leydig-cell tumors are prevented by testosterone supplementation ([27]; Fort et al., 1995). Similarly, the hormonal effects leading to Leydig-cell tumorigenesis can be mimicked by different classes of chemicals that act through the hypothalamic–pituitary–gonadal axis to ultimately affect LH and testosterone, and lead to Leydig-cell hyperplasia and Leydig-cell tumors ([156]). In addition, GnRH receptor agonists cause development of Leydig-cell tumors by binding to LHRH receptors on Leydig cells ([147]; [42]). This latter mechanism is unique to the rat since human Leydig cells do not have LHRH receptors ([147]).

Leydig-cell tumors and their accompanying alterations in systemic hormonal levels have pleotrophic effects on the tissues of the genital system, including decreased spermatogenesis, seminiferous tubule atrophy, and atrophy of seminal vesicles ([88]; [15]). Intratesticular androgen levels are significantly higher than circulating levels ([50]). One hypothesis is that alterations in androgen levels that accompany Leydig-cell tumors are reflected as a transudate in the interstitial fluid within the testes as well as in the tunica vaginalis fluid compartment. There is support for this hypothesis; the mesothelium bathed by the tunica vaginalis fluid is exposed to a higher concentration of the altered hormonal levels, probably by diffusion, than would occur following exposure via the circulatory system ([90]; [56]). Exposure of tunica vaginalis mesothelium to altered levels of androgens may trigger mitogenesis via mesothelial-cell production of growth hormones that operate in an autocrine fashion, as occurs in other male reproductive system tissues ([112]; [98]). The growth hormones released from the stimulated tunica vaginalis mesothelium include TGF-beta, PDGF, IGF2, and EGF, all of which stimulate mitogenesis. Continued enhanced proliferation of the tunica vaginalis mesothelium will lead to hyperplasia, with a subsequent increased probability for development of genetic damage and subsequent mesotheliomas.

An alternative hypothesis for induction of tunica vaginalis mesotheliomas secondary to Leydig-cell tumors in rats relates to the physical pressure or mechanical stress placed on the mesothelial cells lining the scrotal tunics by the enlarged testes ([162]). Since it is known that transformed mesothelium expresses autocrine growth factors that stimulate mitogenesis ([57]; [170]), it is hypothesized that physical pressure from testes enlarged by Leydig-cell tumors could lead to transformation and/or growth factor secretion by tunica vaginalis mesothelium. This possible mode of action is further supported by the observation that visceral pleural mesothelial cells release significant levels of the growth factor PDGF in response to mechanical forces ([175]). As is the case with virtually all studies of carcinogenesis, alternative modes of action are not necessarily mutually exclusive, and more than one may act in concert to produce an adverse effect.

While hormone imbalance and mechanical force represent most likely key events for induction of both spontaneous as well as treatment-associated increases in tunica vaginalis mesotheliomas in rats, and especially in the F344 rat, alternative pathways for induction of tumor development from exposure to xenobiotics are certainly plausible. Assuming that a xenobiotic agent or its metabolite can reach the tunica vaginalis mesothelium, both direct genotoxic action or indirect DNA damage via reactive oxygen species could also explain an increase above the low spontaneous background incidence of this tumor. Similarly, enhanced cell proliferation, possibly secondary to irritation, inflammation, or mechanical stress, could contribute to an increase of this low incidence spontaneous tumor. An association between chronic inflammation and both human pleural and rat peritoneal mesothelioma induction has been reported ([72]; [65]).

Evidence for an oxidative stress mode of action is supported by intraperitoneal injection of xenobiotics such as ferric saccharate or ferric nitrilotriacetate ([139]; [137]), as well as by oral exposure to potassium bromate ([96]; [39]; [177]), which produce reactive oxygen species (ROS) that can potentially have direct action on tunica vaginalis mesothelium. ROS are also considered important mediators in asbestos-induced mesotheliomas ([12]; [153]; [1]). Alternatively, increases in replicative DNA synthesis in mesothelium that could lead to mesothelioma development either by directly affecting cell cycle machinery or secondary to gene alterations in cell cycle machinery have been shown in testicular mesothelium following subchronic exposure to acrylamide ([99]).

From a review of several agents associated with increases in tunica vaginalis mesotheliomas in F344 rats and occasionally in other rat stocks, one or more of the already-described key events may be operating in the genesis of tunica vaginalis mesotheliomas. Likely modes of action for mesothelioma induction will be addressed for the specific xenobiotics associated with increases in this tumor and are described in the following sections. Twenty-one substances that were associated with increased incidences of tunica vaginalis mesothelioma in chronic rat carcinogenicity studies were identified in the National Toxicology Program database (Table 2) and in an extensive review of published literature, and their effects are described next.

Table 2.  Xenobiotics associated with tunica vaginalis mesotheliomas in rat studies (arranged alphabetically).

Most rat cancer bioassays with some evidence of mesothelioma induction reported by NTP or in the literature were conducted using F344 rats. The NTP studies utilized F344 rats from a closed colony, and the other studies used F344 rats from different commercial sources. Consequently, the sensitivity of F344 rats to spontaneous and induced mesotheliomas extends to different colonies of these rats. The specific studies presented next are arranged in order, by route of administration.

Various forms of asbestos and a variety of other durable fibers and agents, including ceramic fibers, silicon carbide, stone wool, slag wool, glass wool, erionite, and cellulose, induce peritoneal cavity mesotheliomas in rats by intraperitoneal (ip) injection ([171]; [37]; [108]; [111]; [113]; [84]; [94]). These same agents have been shown to induce pleural cavity mesotheliomas in experimental animals injected by the intrapleural route. The various intraperitoneal injection studies have been carried out in different strains such as Osborne-Mendel, Wistar, and F344, sometimes in females rather than males, and typically have used a single intraperitoneal injection. Adhesions and chronic inflammation generally accompanied the induced mesotheliomas, which occurred several months after treatment. These studies are not summarized or discussed in detail in the following material.

Three nonfibrous chemical agents, when introduced into the peritoneal cavity of rats, led to development of tunica vaginalis mesotheliomas. Based on the anatomy of the rat, fluid injected into the abdominal cavity can easily get into the scrotal sac and lead to exposure of the tunica vaginalis mesothelium. Direct-acting carcinogens such as nitrosamines or agents that bring about oxidative stress, either as a primary effect or secondary to peritoneal inflammation, can also cause tunica vaginalis mesothelioma when injected into the peritoneal cavity of rats.

In a comparison of three different rat strains, [16]) reported on induction of testicular mesotheliomas following a single intraperitoneal injection of methyl(acetoxymethyl)nitrosamine (DMN-OAc), a short-lived, direct-acting carcinogen (Table 3). In addition to the mesotheliomas, atypical mesothelial hyperplasia was noted in rats that didn't develop the tumor. The authors offered the opinion that testicular mesothelium has properties that are distinct from mesothelium elsewhere, and that the ability of mesothelium to respond to chemical carcinogens is an almost exclusive property of testicular mesothelium. In another publication, the authors showed that the spectrum of tumors induced by DMN-OAc in rats is dependent upon the route of administration ([16]).

Table 3.  Frequency of proliferative mesothelial lesions and Leydig-cell tumors following a single intraperitoneal injection of methyl(acetoxymethyl)nitrosamine in three strains of male rats.

17 Note. LCT = Leydig-cell tumor.

The average age of death for the treated rats ranged from 14.8 to 16 mo, while the average for control rats ranged from 17.2 to 20.9 mo. Although DMN-OAc is a direct-acting carcinogen and does not require metabolic activation, it is clear that genetics can influence the susceptibility to mesothelioma formation. Furthermore, the highest incidence of mesothelioma (46%) occurred in the Buffalo rat, which did not have any Leydig-cell tumors, suggesting that hormonal effects were not driving the response in this particular study. It is noted that even gavage administration of nitrosamines causes peritoneal mesotheliomas ([102]), suggesting that mesothelium may be especially sensitive to nitrosamine carcinogenesis. Methyl(acetoxymethyl)nitrosamine is mutagenic in the Ames test (see Table 29).

Daily intraperitoneal injections of ferric saccharate, which is a colloidal iron, and ferric saccharate plus nitrilotriacetic acid (NTA) for 3 mo resulted in a high incidence of mesotheliomas in male Wistar rats (Table 4) ([139]). NTA stabilizes the iron, which allows it to more efficiently induce ROS, which then promote lipid peroxidation, enhancing the carcinogenic action of iron.

Table 4.  Frequency of mesotheliomas following intraperitoneal injection of ferric saccharate in male Wistar rats.

The mesotheliomas were confined to the tunica vaginalis in the ferric saccharate group. Six of the 14 mesotheliomas in the ferric saccharate–NTA group were disseminated throughout the abdominal cavity.

Intramuscular injection site neoplasms have been induced by iron dextran complex [150]), indicating that injected iron can cause cancer at the site of injection. In the [139]) study, the mesotheliomas appeared to arise in the tunica vaginalis, presumably because the injection iron became localized in the testicular sac following intraperitoneal injection. The authors suggest free radical production with localized enhancement of the carcinogenic action of iron by NTA as the likely mode of action for mesothelioma induction. NTA is not mutagenic in the Ames or mouse lymphoma mutation tests, and does not produce chromosome damage in mammalian cells in vitro; there are no reported mutagenicity studies of its combination with ferric saccharate.

In an NTP bioassay of cytembena, a cytostatic agent, F344 rats and B6C3F1 mice received intraperitoneal injections 3 times per week for 104 wk (NTP TR 207, 1980). Cytembena produced a strong mesothelioma response which was the only tumor induced in male rats in this study (Table 5). Female rats had an increase only in mammary fibroadenomas, and had two malignant abdominal mesotheliomas at the high dose. No induced tumors were seen in mice of either sex.

Table 5.  Frequency of peritoneal and tunica vaginalis mesotheliomas in male F344 rats given 3 times weekly injections of cytembena for 2 yr.

There was significant, drug-related chronic inflammation in the peritoneal cavity in both sexes of rats, and the inflammation occurred at a greater frequency and severity in the females. While mesotheliomas occurred in 2/50 high-dose females, a significantly more robust response was seen in the males. There was no dose response; a maximum response was seen at both doses, and the mesotheliomas were present throughout the abdominal cavity, inclusive of the testis and epididymis. The induction of mesotheliomas in this study is most probably a consequence of inflammation, in combination with the sex predilection for tumor induction in the tunica vaginalis of male F344 rats. The mice in this study received higher doses than the rats, did not have chronic peritoneal inflammation, and did not have mesotheliomas. This observation serves to reinforce the commonly accepted observation that mice in cancer bioassays do not develop mesotheliomas, even following multiple direct intraperitoneal injections for 2 yr, and that rats are more sensitive to mesothelial tumorigenesis. Cytembena is mutagenic in the Ames test and produces chromosome damage in cultured mammalian cells, but did not induce chromosome damage in mouse bone marrow cells following ip injection (see Table 29).

Three inhalation 2-yr cancer bioassays resulted in induction of tunica vaginalis and associated peritoneal mesotheliomas in male F344 rats.

Ethylene oxide, a highly reactive alkylating agent used in chemical synthesis, and to a lesser extent for sterilization and fumigation, was tested by inhalation exposure in F344 rats at 10, 33, and 100 ppm ([157]). At the end of the 2-yr study there was an increased incidence of tumors in both sexes with increases in brain tumors in both sexes, mononuclear cell leukemia and mammary gland adenomas and adenocarcinomas in females, and peritoneal mesotheliomas in males (Table 6). There was a high incidence of Leydig-cell tumors in all groups of male rats and a variety of endocrine neoplasms in both male and female rats. [157] used two equally sized but separate control groups. A different inhalation study at 50 and 100 ppm in male F344 rats also resulted in an increased incidence of peritoneal mesotheliomas (Table 6) ([103]). This latter study also documented an increase in mixed cell gliomas in the brain and mononuclear-cell leukemia in the ethylene oxide-exposed males.

Table 6.  Frequency of mesotheliomas in male F344 rats exposed to thylene oxide vapor for 2 yr.

18 *Combined controls (1/49 and 1/48).

The overall frequency of mesotheliomas in the [157] study was not statistically significant by a two-tailed Fischer's exact test. However, there was a statistically significant trend test and the cumulative percent of rats developing mesothelioma was significantly increased in the 1000-ppm group versus the controls, from month 21 to study termination. The late-developing mesotheliomas were probably influenced by the altered hormonal milieu associated with age-associated Leydig-cell tumors in F344 rats. In the [103] study there was a dose-related increase in mesotheliomas with a statistically significant increase in the 100-ppm-exposed rats.

In both studies, treatment-associated mesotheliomas arose in the tunica vaginalis and some spread into the abdominal cavity. They were morphologically similar to spontaneously occurring mesotheliomas. While the mechanism for induction of mesotheliomas by ethylene oxide remains unclear, the spectrum of other lesions in endocrine tissues and testes potentially implicates a hormonal factor in their development. Ethylene oxide is positive in the Ames test (see Table 29) and most in vitro and in vivo genetic toxicity tests.

1,2-Dibromoethane is a multisite, trans-species carcinogen following inhalation exposure, and produces nasal, pulmonary, and mammary tumors, as well as hemangiosarcomas (NTP TR 210, 1982). Inhalation of dibromoethane for 2 yr produced a strong mesothelioma response in male F344 rats (Table 7). There was an increase in mammary fibroadenomas in female rats. Primary lung tumors, hemangio sarcomas, fibrosarcomas, nasal carcinomas, and mammary adenocarcinomas were induced in B6C3F1-exposed mice (NTP TR 210, 1982).

Table 7.  Frequency of mesotheliomas in male F344 rats exposed to 1,2-dibromoethane by inhalation for 2 yr.

There was a high Leydig-cell tumor frequency in the control and exposed groups. 1,2-Dibromoethane caused testicular degeneration that might explain the reduced number of Leydig-cell tumors in the high exposure rats. In an older NTP gavage study in Osborne-Mendel rats, increased forestomach and liver tumors, as well as hemangiosarcomas, were reported, but no mesotheliomas were present (NTP TR 86, 1978b).

The mechanism by which 1,2-dibromoethane induced mesotheliomas is unknown. Glutathione conjugation of 1,2-dibromoethane leads to formation of an episulfonium ion that is DNA reactive, suggesting a genotoxic effect. The typically high incidence of Leydig-cell tumors in the low exposure group and the known testicular toxicity even at low doses (www.epa.gov/iris) suggest a profound perturbation of hormonal balance that might have contributed to the robust mesothelioma response. 1,2-Dibromoethane is mutagenic in the Ames and mouse lymphoma tests, and produces chromosome damage in mammalian cells in culture and in mouse bone marrow cells (see Table 29).

A low incidence of malignant mesotheliomas in the peritoneal cavity, especially in the scrotal sac, was reported at 160 ppm dichloroethane (DCE) in an inhalation study using F344 rats (Table 8) ([116]). The mesotheliomas at this highest concentration exceeded the historical control, but the incidence was not statistically significantly increased compared to the concurrent control.

Table 8.  Frequency of mesotheliomas in male F344 rats exposed by inhalation of DCE for 2 yr (Nagano et al., 2006).

Other tumor responses in the Nagano study included subcutaneous fibromas and mammary fibroadenomas in male and female rats, as well as mammary adenomas and adenocarcinomas in the female rats. In an older National Cancer Institute (NCI) gavage bioassay in Osborne-Mendel rats, mesotheliomas were not observed (NTP TR 55, 1978a), and there was no mention of testicular Leydig-cell tumors in the study report. DCE was carcinogenic in B6C3F1 mice, causing mammary and endometrial tumors in females and lung tumors in both sexes (NTP TR 55, 1978a). In an older inhalation study in F344 rats, exposure to 50 ppm DCE did not result in a tumor response ([28]). DCE is mutagenic in the Ames and in vitro cytogenetics tests, but did not induce micronuclei in bone marrow of dosed male or female mice (see Table 29).

A dose feed study of ethyl tellurac in F344 rats produced an equivocal tunica vaginalis mesothelioma response that showed a statistically significant trend, but was not significant by pairwise comparison (Table 9) (NTP TR 152, 1979a). This was the only tumor response seen in rats in this study, and the chemical was judged to exhibit equivocal evidence of carcinogenicity. There was no mention in the report of Leydig-cell tumors.

Table 9.  Frequency of mesotheliomas in F344 rats administered ethyl tellurac in the diet for 2 yr.

The judgment to consider the ethyl tellurac bioassay as not positive was based on a nonsignificant pairwise statistical comparison to the concurrent control, and the historical control incidence (12/416; 2.9%) for the testing laboratory. An increased frequency of Harderian-gland adenomas in treated male and female mice was considered equivocal evidence of carcinogenicity. Ethyl tellurac is not mutagenic in the Ames test, is mutagenic in the mouse lymphoma test, and produced an equivocal increase in chromosome aberrations in cultured mammalian cells (Table 29).

Two prechronic and one carcinogenicity study on o-nitrotoluene have been conducted by the NTP (NTP Tox 23, 1992; NTP Tox 44, 1996a; NTP TR 504, 2002). Mesothelial hyperplasia and mesotheliomas involving the tunica vaginalis surface of the epididymis were seen in rats receiving 5000 and 10,000 ppm o-nitrotoluene in their diet for 13 wk (Table 10). A follow-up 26-week prechronic study was conducted to compare the tumor responses of o- nitrotoluene and o-toluidine HCl given at equimolar concentrations in the diet, and to investigate the role of intestinal flora in metabolism of o-nitrotoluene (NTP Tox 44, 1996a). This 26-wk study included a 13-wk o-nitrotoluene exposure, followed by an additional 13 wk on control diet (i.e., stop study). Mesothelial hyperplasia and mesotheliomas were seen at the 13-wk interim sacrifice, in the stop-exposure group at study conclusion, and in the rats continuously exposed to o-nitrotoluene for 26 wk (Table 11). The 2-yr cancer bioassay of o-nitrotoluene included dietary doses of 625, 1250, and 2000 ppm, and incorporated a 3-mo stop study in which rats were fed diets containing 2000 or 5000 ppm o-nitrotoluene followed by undosed feed for the remainder of the 2 yr. All stop-study rats, and all but 3 of the rats given 1250 ppm, died before the end of the 2 yr. The incidences of mesotheliomas in this study are summarized in Table 12.

Table 10.  Frequency of epididymal mesothelial lesions in rats receiving o-nitrotoluene in the diet for 13 wk (NTP Tox 23, 1992).

Table 11.  Frequency of epididymal and testicular mesothelial lesions in rats in the 26-wk dietary o-nitrotoluene study (NTP Tox 44, 1996a).

19 * Rats treated with [antibiotic] to alter the intestinal flora. nd, Not done. GI, gastgrointestinal.

Table 12.  Frequency of mesotheliomas in male F344 rats in the 2-yr feed study of o-nitrotoluene (NTP TR-504, 2002).

20 * Rates in animals that were alive at 104 wk.

In the 2-yr study, the mesotheliomas were located in the tunica vaginalis of the testis or epididymis with some cases extending into the abdominal cavity. The majority of the mesotheliomas in treated rats were large and locally invasive. o-Nitrotoluene is not mutagenic in the standard Ames test. However, its nitro group can be reduced by anaerobic gut flora to ultimately yield a DNA reactive metabolite. The formation of o-benzyl glucuronide is a critical step in leading to formation of the DNA-reactive metabolite. Basically, intestinal microflora hydrolyze the glucuronide and reduce the nitro group to form o-aminobenzyl alcohol. Upon re-absorption of the o-aminobenzyl alcohol, it is sulfated and binds to DNA.

Because reduction of the nitro group of o-nitrotoluene by anaerobic gut flora yields o-toluidine, which is mutagenic in the Ames test, a 26-wk study comparing equimolar doses of o-nitrotoluene and o-toluidine was conducted. The incidence of mesothelioma was greater, and the latency less, for rats administered o-nitrotoluene (NTP Tox 44, 1996a). Similarly, the liver effects, including cholangiocarcinomas, were greater for o-nitrotoluene than for o-toluidine. The lower potency of o-toluidine compared to o-nitrotoluene with respect to liver lesions and mesothelioma induction suggests that the effects of o-nitrotoluene involve more than the simple intestinal reduction of the nitro group. o- Nitrotoluene produced testicular degeneration in the 26-wk toxicity study as well as in the 2-yr cancer study. This would lead to hormonal perturbations, which, in the 2-yr study, were the likely cause of the reduced Leydig-cell tumors in the high-dose males. An associated Leydig-cell tumor reduction associated with testicular toxicity has been noted for other chemicals ([21]). There was clear evidence of carcinogenicity in treated mice based on increased frequencies of hemangiosarcomas, large intestinal carcinomas, and hepatocellular neoplasms.

o-Nitrotoluene was not mutagenic in the Ames test and did not induce chromosome aberrations in cultured mammalian cells, or micronuclei in mouse bone-marrow cells when given in the feed to males and females, or when given ip to male mice or male and female rats.

o-Toluidine is a trans-species carcinogen that produced tumors in both sexes of F344 rats and B6C3F1 mice. Tumor types included a variety of splenic and other tissue mesenchymal tumors, urinary bladder transitional-cell neoplasms, subcutaneous fibromas, hepatocellular neoplasms, hemangiosarcomas, and mammary-gland fibroadenomas. o-Toluidine HCl induced a low incidence of epididymis mesotheliomas in F344 rats in a 26-wk o-nitrotoluene/o-toluidine comparative study (NTP Tox 44, 1996a) (Table 13). An older cancer bioassay had documented a high overall incidence of mesotheliomas involving multiple tissues in the abdominal cavity and the scrotal tunica vaginalis (NTP TR 153, 1979a) (Table 14). An increase in mammary fibroadenomas was present in female rats.

Table 13.  Frequency of mesotheliomas in F344 rats receiving dietary o-toluidine HCl for up to 26 wk (NTP Tox 44, 1996a).

21 * One rat had mesothelial hyperplasia.

Table 14.  Frequency of mesotheliomas in F344 rats receiving dietary o-toluidine for 2 yr (NTP TR 153, 1979b).

The mesotheliomas in the 2-yr study (Table 14) were morphologically similar to spontaneous and treatment-related mesotheliomas in other studies. A few of the more fibrous mesotheliomas contained foci of osseous metaplasia. In light of the known genotoxicity of o-toluidine (see Table 29), it is likely that the mode of action for mesothelioma induction involves DNA damage to the tunica vaginalis mesothelium in addition to the contribution of hormonal imbalance associated with aging male F344 rats bearing Leydig-cell tumors. Hemangiosarcomas and hepatocellular neoplasms were increased in o-toluidine-treated mice. o-Toluidine was mutagenic in the Ames and mouse lymphoma cell tests, produced chromosome aberrations in mammalian cells in culture, and produced contradictory results in two mouse bone marrow micronucleus tests.

2,2-Bis(bromomethyl)-1,3-propanediol is a widely used flame retardant. It is genotoxic in a number of test systems. A dosed feed 2-yr bioassay in F344 rats, which included a 3-mo exposure stop study, produced a multi-site tumor response, including an increased incidence of mammary fibroadenomas in male and female rats (NTP TR 452, 1996b). There was a strong peritoneal mesothelioma response in the male rats (Table 15). Other tumor responses in rats were seen in the skin, Zymbal's gland, oral cavity, esophagus, forestomach, small and large intestines, urinary bladder, lung, thyroid gland, hematopoietic system, and seminal vesicle. Neoplastic responses were also present in both sexes of B6C3F1 mice.

Table 15.  Frequency of mesotheliomas in male F344 rats administered 2,2-bis(bromomethyl)-1,3-propanediol in the diet.

• 22 * Rates in animals that were alive at 104 wk.
• 2,2-Bis(bromomethyl)-1,3-propanediol is one of 14 brominated chemicals studied by the NTP in 2-yr rodent carcinogenicity studies. Thirteen of those 14 brominated chemicals were found to be carcinogenic, but only 3 (1,2-dibromoethane, 2,2-bis(bromomethyl)-1,3-propaneldiol, and potassium bromate) produced TVM. There are two hypotheses for the carcinogenic activity of brominated chemicals: (1) oxidative damage to DNA and other cellular constituents resulting from the induction of ROS, and (2) formation of DNA adducts when the C–Br bond is broken, leaving a carbon-containing electrophilic group. In oral administration studies with potassium bromate (see later discussion), which also produces mesotheliomas in male F344 rats, there is a significant increase in 8- hydroxydeoxyguanosine, which is a biomarker of oxidative damage ([96]; [91]; [152]). 2,2-Bis(bromomethyl)-1,3-propanediol is mutagenic in the Ames test and produces chromosome aberrations in cultured mammalian cells, but yielded equivocal results in a mouse bone-marrow micronucleus test.

Mesotheliomas were induced in male F344 rats in the dosed feed study of nitrofurazone (NTP TR 337, 1988) (Table 16). The mesothelioma response, which was not dose related, was considered equivocal evidence of carcinogenicity by the peer review panel, and arose in the tunica vaginalis, with some mesotheliomas spreading to the peritoneal cavity and invading the underlying soft tissue. There was a treatment-related increase in preputial adenomas and carcinomas, and a significant increase in mammary fibroadenomas in the female rats. Previous studies suggested that nitrofurazone was a mammary gland carcinogen. There was an increase of ovarian cancer in mice. Taken together, the tumor responses indicate the nitrofurazone may act through hormonal effects.

Table 16.  Frequency of mesotheliomas in male F344 rats administered nitrofurazone in the diet for 2 yr.

23 *Rates in animals that were alive at 104 wk.

Poor survival of the high-dose group is the likely reason for the decrease in mesotheliomas at the 620-ppm dose as compared to the lower dose. There was a dose-related decrease in Leydig-cell tumors, also partly a reflection of poor survival in the high-dose group. The obligatory role for nitro reduction in nitrofurazone-induced mutagenicity may be related to the widespread tumorigenicity in rats and mice ([89]). Nitrofurazone was mutagenic in the Ames and mouse lymphoma mutation tests and produced chromosome aberrations in cultured mammalian cells, but did not induce micronuclei in bone marrow cells of mice (see Table 29).

Pentachlorophenol is a wood preservative, as well as an herbicide, fungicide, and germicide. In a dosed feed study with pentachlorophenol, an increase in peritoneal mesotheliomas was seen in the stop-study F344 rats but not in the continuously exposed rats (NTP TR 483, 1999) (Table 17). A marginal increase in nasal carcinomas (1/50 versus 5/50) was also present in the stop-study males. No other treatment-related neoplasms were present in the males, and no treatment-related neoplasms were present in the female rats. Increases in liver and adrenal tumors and hemangiosarcomas were seen in pentachlorophenol-treated mice (NTP TR 349, 1989).

Table 17.  Frequency of mesotheliomas in a 2-yr study with continuous exposure and a stop study in which male F344 rats received pentachlorophenol in the diet for 1 yr (NTP TR 483, 1999).

24 * Rates in animals that were alive at 104 wk.

The mesotheliomas arose in the tunica vaginalis and had the histomorphological characteristics of the spontaneous and chemically induced mesotheliomas seen in other studies. Extension into the peritoneal cavity was evident in five of the mesotheliomas in the stop-study group and the one mesothelioma in the control. Pentachlorophenol was nonmutagenic in the Ames test, and only weakly positive in an in vitro chromosome aberration test in cultured mammalian cells, and did not induce micronuclei in mouse or rat bone-marrow cells.

Although pentachlorophenol is not mutagenic in bacterial test systems, one of its major metabolites, tetrachloro-p-hydroquinone, is genotoxic, covalently binds to DNA, and can induce oxidative damage to DNA. Oxidative damage, as assessed by 8-hydroxydeoxyguanosine adducts, has been found in livers of mice exposed to pentachlorophenol, as well as elevated hemoglobin adducts in males and females (NTP TR 483, 1999). Thus, it is probable that the mesotheliomas seen in rats exposed to the high dose of pentachlorophenol in the NTP study are a consequence of the oxidative damage to mesothelium of the tunica vaginalis. Given that there was also a high incidence of Leydig-cell tumors in the treated rats, the altered hormonal milieu associated with the proliferating Leydig cells may also have contributed to the development of tunica vaginalis mesotheliomas.

Tartrazine is a food, drug, and cosmetic coloring agent. In a 2-yr dosed water study using F344 rats, mesotheliomas were present only at the lower dose (Table 18) of tartrazine ([105]). There was a persistent decreased body weight gain in the 2% group starting at experimental wk 40. Based on a lower than expected incidence in the control group (historical incidence was 4.1%), absence of a positive trend, and absence of hyperplastic or preneoplastic lesions in the peritoneal cavity, the authors concluded that the occurrence of peritoneal mesotheliomas was not related to treatment. It is mentioned in the publication that the mesotheliomas are similar to those seen spontaneously in the F344 male. The incidence of Leydig-cell tumors in this study was greater than 94% in the control and low-dose groups and was 100% in the high-dose group. There was an increased incidence of endometrial stromal polyps in the low-dose female rats that the authors concluded was not treatment related.

Table 18.  Frequency of mesotheliomas in male F344 rats in a 2-yr dosed water study of tartrazine (Maekawa et al., 1987).

Tartrazine was not mutagenic in the Ames test but produced chromosome aberrations in cultured mammalian cells (see Table 29).

In a chronic dosed water study, terminated at 21 mo due to early tumor-induced mortality, there was induction of tumors at multiple tissue sites including a marginal peritoneal mesothelioma response in male F344 rats (Table 19) and a robust mammary gland adenocarcinoma response in female rats (NTP TR 372, 1990a). 3,3'-Dimethoxybenzidine hydrochloride was considered to have clear evidence of carcinogenicity based on statistically significant increases in tumors at multiple sites. It is a member of the aromatic amine class of chemicals, which when metabolically activated induce a variety of tumor types. Activation of ras oncogenes was identified in some of the induced epithelial tumors.

Table 19.  Frequency of mesotheliomas in male F344 rats administered 3,3'-dimethoxybenzidine hydrochloride in drinking water for up to 21 mo (NTP TR 372, 1990a).

The incidences of mesothelioma were not statistically significant by pairwise comparison, although there was a significant positive trend. There was significant early mortality in all treated males and females with greater than 50% mortality by wk 86. At study termination (94 wk) only 8 low-dose males were alive among the treated rats. The authors of the technical report suggested that the mesothelioma incidences might have been higher had the rats lived longer. 3,3'-Dimethoxybenzidine is mutagenic in the Ames and mouse lymphoma mutation tests, but did not induce chromosome aberrations in cultured mammalian cells (see Table 29).

In a 14-mo dosed water study in F344 rats, 3,3'-dimethylbenzidine HCl produced a peritoneal mesothelioma response (Table 20) that showed a positive trend and was statistically significant at the highest dose. The authors of the technical report attributed the mesotheliomas to the test chemical and suggested that the incidence of mesothelioma might have been higher except for the reduced survival in the two highest dose groups. 3,3'-Dimethylbenzidine HCl was considered to have clear evidence of carcinogenicity based on robust responses at multiple other tissue sites (NTP TR 390, 1991).

Table 20.  Frequency of mesotheliomas in male F344 rats administered 3,3'-dimethylbenzidine HCl in drinking water for 14 mo (NTP TR 390, 1991).

• 25 *Rates in animals that were alive at 104 wk.

3,3'-Dimethylbenzidine is a congener of 3,3- dimethoxybenzidine. Activation of the H-ras oncogene was detected in several epithelial neoplasms. 3,3'-Dimethylbenzidine was mutagenic in the Ames and mouse lymphoma mutagenicity tests, and induced chromosome aberrations in cultured mammalian cells (Table 29).

Potassium bromate is a rodent carcinogen and is nephrotoxic and neurotoxic in humans. Because potassium bromate is a by-product of water disinfection by ozonation, there has been interest in testing it for adverse effects by dosing in drinking water. Four drinking-water cancer bioassay studies have been conducted in F344 rats, and peritoneal mesotheliomas were induced in each study. The incidences of mesothelial responses in these studies are summarized in Tables 21, 22, and 23.

Table 21.  Frequency of mesotheliomas in a drinking-water study of potassium bromate in F344 rats reported by Kurokawa et al.

Table 22.  Frequency of mesotheliomas in the drinking-water study of potassium bromate in F344 rats (DeAngelo et al., 1998).

Table 23.  Frequency of mesotheliomas in the drinking-water study of potassium bromate in F344 rats (Wolf et al., 1998).

In the 1983 study, the earliest mesothelioma was observed after 72 wk of treatment ([96]). The mesotheliomas were frequently seeded throughout the abdominal cavity and were associated with massive hemorrhagic ascites, which, according to the authors, led to severe anemia and early death.

Mesothelioma responses in the 1986 study were observed at doses of 30 ppm and higher with a statistically significant increase at 500 ppm, but the tumor incidences between 30 and 250 ppm were not dose related (Kurokawa et al., 1986). The occurrence of Leydig-cell tumors was 95 to 100% in all groups, including the controls.

The origin of the mesotheliomas in this study was the tunica vaginalis mesothelium with involvement of the vaginal tunic, including the mesotheliomas that were present throughout the abdominal cavity. The TVM tended to be bilateral with some exceptions. Based on the book chapter by [69]) and a pathology peer review of this study, the additional peritoneal sites of mesothelioma are considered neither additional primary tumors nor metastases. TVM in F344 rats typically spread by extension and seeding rather than via vascular or lymphatic routes of metastasis.

The design of this study with interim time points permitted the opportunity to examine the temporal sequences associated with development of treatment-induced mesotheliomas. While all mesotheliomas were considered malignant by the authors, a single case of mesothelial hyperplasia, and 1 rat with a small mesothelioma confined to the parietal vaginal tunic, were seen at 52 wk. Spreading to other peritoneal sites was not present until after 78 wk of treatment. Spreading was by extension or implantation (i.e., seeding) and most commonly involved spleen, gastrointestinal tract, mesentery, and pancreas. This study showed the origin of the mesotheliomas to be in the tunica vaginalis.

An extensive reexamination of the study materials from the [177]) study was reported by [34]). Using cross sections of the rat testes to map the TVM, it was concluded that the mesorchium was the major tissue target site for potassium bromate-induced mesotheliomas. The authors discuss several factors that may contribute to TVM development. However, as with other brominated chemicals, oxidative damage ([39]) and formation of oxidative DNA adducts ([96]; [91]; [152]) are the most likely mode of action for induction of the TVM response. Potassium bromate is mutagenic in the Ames test (see Table 29).

Two separate bioassays in which acrylamide was administered to F344 rats in drinking water have been reported ([82]; [51]). Mesotheliomas of the tunica vaginalis were documented in both studies (Table 24). Only some of the mesotheliomas were present in the abdominal cavity, while all were present in the vaginal tunics of the scrotal sac. Neither publication tabulates the incidence of testicular Leydig-cell tumors; however, the laboratory study report for the [82] study shows that 57 of the 60 males in each group, including the control group, had Leydig-cell adenomas. A retrospective examination of study slides from the [51]) study was conducted by [74]), who found that the degree of morphological progression of the tunica vaginalis mesotheliomas was correlated with the size of the Leydig-cell tumors. The malignant mesotheliomas, as classified by Iatropoulos, were seen only in rats that had 75% or greater of their testicular parenchyma replaced by Leydig-cell tumors. The mesothelial tumors that they classified as hyperplasias were present in rats in which the Leydig-cell tumors occupied 24% or less of the testicular parenchyma.

Table 24.  Frequency of mesotheliomas in F344 rats in two separate studies in which acrylamide was administered in drinking water.

26 *Pooled control groups (4/102 and 4/102).

Acrylamide is not mutagenic in the Ames test, but produces chromosome aberrations in cultured mammalian cells. It produces chromosome aberrations and micronuclei in mouse, but not rat bone-marrow cells, and chromosome damage in male germ cells of rats and mice.

The probable modes of action for induction of TVM associated with exposure to acrylamide have been extensively reviewed [156]). Administration of acrylamide to rats produces a dose-related reduction in prolactin and testosterone, thought to be centrally mediated via the dopaminergic system ([52]; [2]; [3], [167]). The enhanced dopamine signal, with its associated decreases in prolactin secretion, would trigger downregulation of Leydig-cell LH receptors ([148]), reduced testosterone, and a compensatory increase in LH, which in turn stimulates proliferation of Leydig cells ([33]). The altered hormonal milieu is then reflected as a transudate in tunical vaginalis fluid, and the exposed tunica vaginalis mesothelium proliferates via an autocrine response to growth factor production. A physical stimulus affecting tunica vaginalis mesothelium from testes enlarged by Leydig-cell tumors may also lead to elaboration of growth factors by the mesothelium and an autocrine-mediated cell proliferative response.

The gavage administration of methyleugenol (in 0.5% methylcellulose) to F344 rats resulted in induction of multiple tumor target sites (NTP TR 491, 2000) with a strong mesothelioma dose response (Table 25). Fifty of 60 males and females received 300 mg/kg methyleugenol for 52 wk and then were administered methylcellulose vehicle, alone, for the next 53 wk.

Table 25.  Frequency of mesotheliomas in male F344 rats gavaged with methyleugenol for 2 yr (NTP TR 491, 2000).

27 *Rates in animals that were alive at 104 wk.

There was a dose-related increase of Leydig-cell tumors in core study rats. Of the 5 rats treated with 300 mg/kg and the 5 controls examined at 12 mo, all had Leydig-cell tumors, 9 of which were bilateral. Of the 50 remaining stop study rats, 5 had TVM. Mammary-gland fibroadenomas were also increased in dosed male rats. Other induced neoplasms included benign and malignant liver tumors, benign and malignant gastric neuroendocrine tumors, benign kidney tumors, and benign and malignant connective tissue tumors of the skin. Liver and glandular stomach neoplasms were increased in treated mice. While methyleugenol is not mutagenic in the Ames test, and does not induce chromosome damage in cultured mammalian cells or mouse bone marrow, its metabolism is associated with adduct formation, and beta-catenin mutations have been reported in methyleugenol-induced mouse liver tumors ([41]).

In a 2-yr gavage study of benzaldehyde in F344 rats (NTP TR 378, 1990c) using corn oil as the vehicle, a marginal TVM response (Table 26) was not considered related to treatment, and the chemical was judged not to be a carcinogen in rats. This judgment was influenced by lack of a dose response and the laboratory's mesothelioma historic control incidence of 8% in male F344 rats. The incidences of Leydig-cell tumors in the control and low-dose groups were greater than 90%, while only 63% of the high-dose males had Leydig-cell tumors. There was some evidence of treatment-related neoplasia in mice based on forestomach squamous-cell papillomas. Benzaldehyde was not mutagenic in the Ames test, but did induce mutations in the mouse lymphoma test, and it did not induce chromosome aberrations in cultured mammalian cells (see Table 29).

Table 26.  Frequency of mesotheliomas in male F344 rats administered benzaldehyde by gavage in corn oil (NTP TR 378, 1990c).

Exposure to glycidol produces a marked carcinogenic response with tumors at multiple sites in both sexes of F344 rats and B6C3F1 mice (NTP TR 374, 1990b). Peritoneal mesotheliomas are among the tumor responses in male rats that showed a dramatic increase in a 2-yr gavage study in which glycidol was administered in a water vehicle (Table 27).

Table 27.  Frequency of mesotheliomas in male F344 rats in a 2-yr water gavage glycidol study (NTP TR 374, 1990b).

28 *Rates in animals that were alive at 104 wk.

All mesotheliomas were present in the tunica vaginalis, many with extension into the abdominal cavity. They were classified into benign and malignant neoplasms. Mesotheliomas confined to the vaginal tunics were considered benign and those that spread into the abdominal cavity and/or had cytological features of malignancy were considered malignant. The histomorphological features of the malignant mesotheliomas included pleomorphism, cytological atypia, local invasiveness, and implant metastasis throughout the abdominal cavity. Malignant mesotheliomas were considered rapidly lethal; the first death attributed to mesothelioma occurred in a high-dose male at study wk 49.

Despite early tumor-associated mortality in the treated males, the control and dosed male groups all had high incidences of Leydig-cell tumors. Mammary-gland neoplasms were dramatically increased in female rats. Epithelial tumors were increased at multiple sites in treated mice. Glycidol is a direct alkylating agent, forming promutagenic adducts in DNA, and is mutagenic in the Ames test and produces chromosome damage in cultured mammalian cells and mouse bone marrow. The relationship between adduct formation and tumorigenesis is in part attributed to the relative susceptibility of the exposed tissue. The robust mesothelioma response observed in the glycidol study is most probably a consequence of the combined effects of localized genotoxicity and the susceptibility of tunica vaginalis mesothelium to the hormonal imbalance in F344 rats associated with aging and the development of Leydig-cell tumors.

Topical application of 2,3-dibromo-1-propanol produced a marginal mesothelioma response in male rats (Table 28) but clear evidence of carcinogenicity at other sites (NTP TR 400, 1993). There was also clear evidence of carcinogenicity in mice based on increased incidences of epithelial neoplasms.

Table 28.  Frequency of mesotheliomas in male F344 rats treated topically with 2,3-dibromo-1-propanol for 51 wk (NTP TR 400, 1993).

The study was terminated after 51 wk, because of reduced survival in the high-dose groups resulting from chemically induced neoplasms. Early mortality began at wk 45. Major induced tumors involved the nasal cavity, skin, oral cavity, and gastrointestinal tract. The incidence of Leydig-cell tumors was low because of early study termination, with the highest incidence of 34% seen in the low-dose group. However, the incidence of Leydig-cell hyperplasia was up to 56% in the low-dose group, suggesting that the paracrine hormonal secretion by the proliferating Leydig cells also contributed to the early appearance of tunica vaginalis mesotheliomas in treated rats. 2,3-Dibromo-1-propanol is mutagenic in the Ames and mouse lymphoma mutation tests and produces chromosome aberrations in cultured mammalian cells, but did not induce micronuclei in mouse bone marrow cells (Table 29).

Analyses of carcinogenicity and genotoxicity databases ([10]; [61], [60]) have shown that some tumor types/locations are associated with genotoxic chemicals and some are associated with nongenotoxic chemicals, although the association appeared to be less strong in the [61], [60]) compilations, which examined the NTP and other data sources than in [10]). In this latter study that examined only chemicals tested by the NTP ([10]), there was an association of some tumor sites with mutagenicity. That is, some tumors/tumor sites were responsive primarily to chemicals that were mutagenic in the Salmonella test, some were responsive primarily to chemicals that were not mutagenic in Salmonella, and other sites appeared to be responsive to both mutagenic and nonmutagenic carcinogens.

Genotoxicity in this context is defined as positive results in the Salmonella mutagenicity (Ames) test. A positive response in an in vitro mammalian cell chromosome aberration test, by itself, is not considered to be definitive evidence of genetic toxicity because of the predilection of this test to produce positive results as a secondary response to cell toxicity, or to high osmolarity or changes in growth medium pH ([22]; [154]; [115]).

Ashby and Tennant identified two chemicals among the NTP database that induced tunica vaginalis mesotheliomas, glycidol (Ashby and Tennant, 1991a) and 1,2-dibromoethane (Ashby and Tennant, 1991b), both of which were mutagenic in Salmonella. The [61], [60]) compilations of cancer site and mutagenicity do not distinguish tunica vaginalis mesotheliomas from other testicular tumors, and do not list mesothelioma as a tumor type.

Chemicals reported here to induce tunica vaginalis mesotheliomas (see Table 29) were classified according to the potency of their tumor induction, e.g., robust or nonsignificant- to- marginal, and their genetic toxicity. The criterion for a robust tumor response was that the magnitude of the highest incidence, regardless of dose, was > 18%. This criterion was determined by examining the incidence data, the likely mode of action, and/or the final interpretation of the specific cancer bioassays. For 1,2-dichloroethane the 10% (5/50) TVM response was not statistically significant versus the concurrent control (0/50) ([116]). The 10% TVM response in the low-dose animals in the benzaldehyde study was judged to be a noncarcinogenic response by the NTP peer review board (NTP TR 378, 1990c) because, although it was greater than the concurrent control, it was equivalent to the 8% historical control incidence in the testing laboratory. The 12% TVM response induced by tartrazine did not exhibit a dose response, and was considered to be not treatment related by the author ([105]). The nitrofurazone TVM response of 14% was seen at the lower dose without evidence of a dose response (NTP TR 337, 1998). A 16% TVM response in the ethyl tellurac study, although dose related, was considered equivocal by the NTP peer review board (NTP TR 152, 1979a). The TVM found in the acrylamide studies are considered centrally mediated and secondary to Leydig-cell tumors ([156]). The 18% pentachlorophenol response was seen only in a stop study where the dose exceeded the maximum tolerated dose (NTP TR 483, 1999). The gap between the 18% incidence of TVM in the pentachlorophenol study and a 24% incidence of TVM for methyl eugenol prompted selection of 18% as a cutoff incidence for classifying the potency of the TVM response. Latency, as defined by the week to first observed TVM, of less than 60 wk was a feature of robust responses (Table 29).

Table 29.  Genetic toxicity of chemicals associated with tunica vaginalis mesothelioma (TVM) induction in rats, arranged in order of decreasing maximum tumor frequency.

29 Note. Ames test, Salmonella mutagenicity result; in vitro cyto, chromosome aberrations in Chinese hamster cells in culture; in vivo cyto, chromosome aberrations or micronuclei in bone-marrow cells of treated mice. ip, Intraperitoneal administration; +, positive response; w+, weakly positive; –, negative; E, equivocal response; +/–, conflicting results; blank, not tested; NA, data not available. *Nitrotoluene is mutagenic in Salmonella when activated by enteric organisms. ** Nitrilotriacetic acid, by itself, is negative in the Ames test and was not tested in the chromosome aberration test. Nitrilotriacetic acid + ferric saccharate has not been tested in the Ames test or in chromosome aberration tests. *** Studies were terminated early due to other tumor formation; cannot be classified as to TVM potency. aPercent of treated animals with TVM; maximum response recorded. bStudy week at which earliest TVM was identified. cLevels of evidence of carcinogenicity for TVM in NTP studies: CE or P = clear evidence, SE = some evidence, EE = equivocal evidence, NE = no evidence. Detailed definitions of each category are presented on the NTP web site: http://ntp.niehs.nih.gov/?objectid=07027D3D-BADE-07F1-7521FFEC341D1CA9. Some of the chemicals in this table were considered positive for carcinogenicity based on tumors at sites other than tunica vaginalis mesothelium (see Table 30). dCombined results from testing different salts of the same parent chemical.

Regardless of the conclusions of [61], [60]), who found that genotoxic and nongenotoxic chemicals produced similar tumor induction patterns, there was a clear distinction between the chemicals that produced robust, and those that produced weak, tunica vaginalis mesothelioma responses. Of the 10 chemicals producing robust responses that had genetic toxicity test results, 8 (80%) were mutagenic in Salmonella. One of the outliers, nitrotoluene, requires anaerobic activation as present in vivo, in contrast to the aerobic conditions present in the Ames test. Where in vitro cytogenetics results were available, they supported the Salmonella results. In contrast, only two of the seven chemicals (29%) that produced nonsignificant-to-marginal responses were mutagenic in Salmonella. There were an additional three chemicals in this group that were negative in the Salmonella test but positive in the chromosome aberration test, one of which, acrylamide, also produced chromosome damage in the in vivo bone marrow test.

Three NTP studies were terminated early, i.e., less than 2 yr, because of mortality from other tumors. 2,3-Dibromo-1-propanol, 3,3'-dimethoxybenzidine 2HCl, and 3,3'-dimethylbenzidine 2HCl, which are all mutagenic in Salmonella, had overall TVM frequencies of 8, 10, and 7%, respectively. Because TVM tend to be late-occurring neoplasms, especially in controls, early study termination because of other tumor responses would not allow for adequate exposure time to fully assess the magnitude of a potential mesothelioma response in these three studies.

The chemicals in Table 29, and their putative metabolites, present a wide range of structures and chemical characteristics. Some, e.g., methyl(acetoxymethyl)nitrosoamine, glycidol, and ethylene oxide, can form DNA adducts. Others, e.g., nitrilotriacetic acid, methyl eugenol, potassium bromate, and ethyl tellurac, do not appear to have any direct DNA reactivity, but may induce their damage through the induction of reactive oxygen species. And others, e.g., acrylamide, are known to be both DNA reactive (through its metabolite, glycidamide) and capable of inducing oxidative stress and hormonal changes.

The one conclusion that is obvious from this compilation is that the induction of tunica vaginalis mesotheliomas is not confined to genotoxic chemicals. A significant fraction of these tumors are induced by chemicals that are considered to be nongenotoxic, presumably acting through a mechanism or mechanisms that do not involve direct DNA interaction.

The abundant literature dealing with pleural mesotheliomas associated with human exposure to asbestos and other fibers has not be covered in detail in this review, other than to compare and contrast the fiber-induced tumors with the chemical-induced tumors, where appropriate. Nonpleural mesotheliomas, including tunica vaginalis mesotheliomas, have been reported in humans ([70]). Spontaneous tunica vaginalis mesotheliomas are rare in humans, with fewer than 100 cases reported in the literature in the last 36 yr ([67]; [176]; [25]; [83]; [146]; [68]; [7]). Of 11,629 malignant mesothelioma deaths reported in the United Kingdom from 1968 to 1991, 0.09% were primary TVM ([13]). The epithelial, sarcomatous, and mixed subtypes of TVM are the same as asbestos-associated pleural mesotheliomas with tubulopapillary, epitheloid, cystic, fibrotic, and pleomorphic growth patterns ([13]; [62]; [75]; [58]). In contrast to the rat, the tunica vaginalis in the adult human does not directly connect to the peritoneal cavity. Consequently, tunica vaginalis mesotheliomas in humans are typically confined to the scrotal vaginal tunics, are locally invasive in about 50% of the cases, and, when metastatic, typically spread via the lymphatics ([67]). In a review of 74 human cases, lymph-node metastases occurred in approximately 15% of the cases of tunica vaginalis mesothelioma ([146]). A correlation of asbestos exposure with some cases of human tunica vaginalis mesothelioma has been suggested ([67]; [146]), but the association of asbestos and gonadal mesothelioma is not well made and many cases have had no known exposure to asbestos ([13]).

Of the 21 xenobiotics associated with a mesothelioma response in rats that are addressed in this document, 7 are judged to have a nonsignificant-to-marginal response, 3 are relatively noninformative with respect to their potency due to early study termination because of tumors other than mesotheliomas, and 11 exhibited a robust mesothelioma response. Highlights of the findings in these studies are summarized in Table 30, and the categorizations of the responses are in Table 29. If one excludes the three robust chemicals that were identified via the intraperitoneal route of exposure, where the xenobiotic would have direct contact with mesothelium, the remaining 18 studies were done using F344 rats. For the 11 chemicals with a robust mesothelioma response, a genotoxic mode of action may be associated with that target tissue response. However, the presence of Leydig-cell tumors in the F344 rats, and the evidence linking Leydig-cell tumors to tunica vaginalis mesotheliomas, suggest a contributory effect of the Leydig-cell tumor burden.

Table 30.  Summary of chemicals associated with tunica vaginalis mesotheliomas in rats, arranged by route of administration.

30 aLevels of evidence of carcinogenicity for tumors other than TVM in NTP studies: CE or P = clear evidence, SE = some evidence, EE = equivocal evidence, NE = no evidence. Detailed definitions of each category are presented on the NTP web site: http://ntp.niehs.nih.gov/?objectid=07027D3D-BADE-07F1-7521FFEC341D1CA9. Note. TVM = highest % incidence tunica vaginalis mesothelioma. LCT = Leydig cell tumor response [+ = < 79%; ++ = ≥80%]. MR = male rat;. FR = female rat. MM = male mouse. FM = female mouse. NR = not reported. SD = Sprague-Dawley. Buf = Buffalo Wist = Wistar. NTA = Nitrilotriacetate acid. ip = Intraperitoneal. Topic = topical. Inh = inhalation. Diet = dietary. Water = drinking water. Gav = gavage.

The occurrence of xenobiotic treatment-associated tunica vaginalis mesotheliomas by other than the intraperitoneal route is a feature unique to the male F344 rat. This tumor response is an increased incidence of a well-documented, low spontaneous background rate of tunica vaginalis mesothelioma in these rats. A key event associated with the xenobiotic induction of TVM in the F344 rat is the age-associated and high incidence of testicular Leydig-cell tumors. The local hormonal milieu in the tissues adjacent to the Leydig-cell tumors is altered and the hormonal imbalance is reflected as a transudate in the tunica vaginalis fluid. This, in turn, leads to an autocrine growth factor response in the tunica vaginalis mesothelium as a primary mode of action, resulting in mesothelial hyperplasia and ultimately mesothelioma. Since it has been shown that mesothelial cells respond to pressure or shearing forces by elaborating autocrine growth factors, the markedly enlarged testes from the Leydig-cell tumor burden can also initiate a mitogenic stimulus. Thus, a specific primary mode of action for developing tunica vaginalis mesotheliomas in the F344 rat is dependent upon enhanced mitogenesis caused by autocrine growth factors in the stimulated tunica vaginalis mesothelium. Given the extremely low incidence of Leydig-cell tumors in humans, a F344 rat tunica vaginalis mesothelioma response attributed to this primary mode of action is not considered relevant to human cancer induction.

To further understand the factors associated with Leydig cell biology, an expert panel of scientists identified seven mechanisms that could lead to Leydig-cell hyperplasia and adenoma formation ([31]). Two hormonal modes of action, namely, GnRH agonism and dopamine agonism, were considered not relevant to humans. GnRH agonism is unique to the rat since human as well as monkey and mouse Leydig cells do not express the LHRH receptor ([147]). Dopamine agonism leads to decreased prolactin secretion by the pituitary, which, in turn, leads to downregulation of Leydig-cell LH receptors, decreased testosterone, and a compensatory increased circulating LH to raise testosterone levels ([33]; [148]). The increased LH leads to Leydig-cell proliferation and ultimately to Leydig-cell tumors ([33]; [147]). This dopaminergic mode of action is unlikely in humans because the number of LH receptors per Leydig cell is 14 times less than in the rat, and Leydig-cell tumors are extremely rare in humans ([147]; [50]). Five additional hormonal modes of action for Leydig-cell tumor induction that are potentially relevant to humans include androgen receptor antagonism, 5-alpha-reductase inhibition, inhibition of testosterone biosynthesis, aromatase inhibition, and estrogen agonism. Rodents have greater sensitivity than humans to these hormonal effects. The expert panel recommended a margin-of-exposure (MOE) approach be used when a rodent Leydig-cell tumor response is attributable to one of these five modes of action. If the compound under investigation was mutagenic, then a case-by-case judgment regarding human health risk was recommended.

There are species and strain differences that indicate a tunica vaginalis mesothelioma response by other than the peritoneal route of exposure is specific to the F344 rat. Examination of the literature indicates that a tunica vaginalis response to xenobiotic exposure is generally not seen in other strains and stocks of rats, even following sustained increased LH levels ([148]). The aging F344 rat has a more advanced development of testicular changes, including Leydig-cell tumors, than other rat stocks ([88]) and a greater background incidence of testicular mesotheliomas. In several hazard identification cancer bioassays conducted, in parallel, in F344 rats and B6C3F1 mice, a tunica vaginalis mesothelioma response was never seen in mice, nor was a mesothelioma response seen in female rats. Consequently, the male F344 rat specificity of tunica vaginalis mesothelial tumorigenesis is not likely to be relevant to other species or pose a human cancer risk.

Among the xenobiotics reviewed in this report, some are direct alkylating agents with clear genotoxicity and a robust tunica vaginalis mesothelioma response (Tables 30 and 31). Robust TVM responses have been observed in rats exposed to alkylating agents such as glycidol (NTP TR 374, 1990b) and nitrosamines ([16]; [102]; [64]). The relationship between adduct formation and tumorigenesis is, in part, attributed to the relative susceptibility of the exposed tissue. It has been suggested that tunica vaginalis mesothelium, as opposed to mesothelium elsewhere in the body, has unique properties making it more responsive to chemical carcinogens ([16]). The robust mesothelioma response observed in the glycidol study is most probably a consequence of the combined effects of localized genotoxicity and the susceptibility of tunica vaginalis mesothelium to the hormonal imbalance in F344 rats associated with aging and the development of Leydig-cell tumors.

Another example of a robust tunica vaginalis mesothelioma response occurred following exposure to o- nitrotoluene. For this chemical, the formation of o-benzyl glucuronide is a critical step in leading to formation of DNA-reactive intermediates. Intestinal microflora hydrolyze the glucuronide and reduce the nitro group to form o-aminobenzyl alcohol. Upon reabsorption of the o-aminobenzyl alcohol, it is sulfated and binds to DNA. Two brominated chemicals, 2,2-bis(bromomethyl)-1,3-propanediol and potassium bromate, produced a robust mesothelioma response (Table 29). Hypotheses for the carcinogenic activity of brominated chemicals include oxidative damage to DNA and formation of DNA adducts when the carbon–bromine bond is broken ([96]; [91]; [152]; De Angelo et al., 1998). It is noted, however, that even for genotoxic xenobiotics producing a robust tunica vaginalis mesothelioma responses in male F344 rats, there are no mesotheliomas in female rats or in mice, thereby underscoring the unique sensitivity of the tunica vaginalis mesothelium in male F344 rats.

Some tissue-specific responses are characteristic of epigenetic modes of action in the nonsignificant to marginal tunica vaginalis mesothelioma responses (Table 29). Using acrylamide as an example, the adrenal pheochromocytoma, tunica vaginalis mesothelioma, and thyroid follicular adenoma responses in the male F344 rat ([82]; [51]) are consistent with rodent-specific targeting of endocrine-sensitive tissues and have little relevance to human cancer risk ([32]. Exposure of F344 and Sprague-Dawley rats to acrylamide has been shown to increase replicative DNA synthesis in these tumor target tissues, but not in nontarget tissues ([99]). Furthermore, blocking cytochrome P-450 activity, and thus the formation of the DNA-reactive metabolite of acrylamide, glycidamide, did not abolish replicative DNA synthesis in the tunica vaginalis mesothelium. From these findings, it is apparent that the tunica vaginalis mesothelioma response occurred through a mode of action independent of oxidative metabolism of the chemical to a DNA-reactive metabolite ([99]). Acrylamide also has dopaminergic activity in the F344 rats, which leads to decreased circulating prolactin followed by enhancement of spontaneous, age-associated Leydig-cell tumorigenesis ([52]). As a result, the tunica vaginalis mesothelioma response in acrylamide-treated F344 rats is most likely caused by a hormonally mediated and autocrine growth factor-driven mesothelial mitogenesis mode of action. A similar autocrine growth factor-driven mode of action, although not necessarily amplified by dopamine agonism, is believed to be a primary cause of the observed tunica vaginalis mesothelioma responses seen for other chemicals with a nonsignificant to marginal response (Table 29). Thus, these xenobiotics with a nonsignificant to marginal tunica vaginalis mesothelioma response that is unique to the F344 rat do not pose a significant risk for human carcinogenesis (see Table 31).

Table 31.  Human relevance framework for tunica vaginalis mesothelioma (TVM) induction in F344 rats secondary to enhanced mesothelial mitogenesis.

The primary conclusions based upon this review of tunica vaginalis mesotheliomas in rat bioassays are as follows:

• Tunica vaginalis mesotheliomas are low-incidence spontaneous neoplasms in rats that can be increased by treatment.
• Tunica vaginalis mesotheliomas in rats originate in the mesothelial lining of the scrotal sac, testes, epididymides, and mesorchium and can spread to the abdominal cavity by extension or seeding since the scrotal sac mesothelium is continuous with the peritoneal cavity mesothelium.
• A majority of chemicals that are associated with a nonsignificant-to-marginal tunica vaginalis mesothelioma induction are nongenotoxic based on the Ames test, whereas chemicals producing a robust response tend to be Ames test mutagens.
• The mesothelioma responses to xenobiotic exposure by other than the peritoneal route are male F344 rat specific. They are never seen in female F344 rats or in either gender of mice in conventional cancer bioassays, and have not been reported in other rat strains used for carcinogenicity testing.
• Spontaneous, as well as several, xenobiotic-associated tunica vaginalis mesotheliomas are causally associated with Leydig-cell tumors that lead to an autocrine growth factor-induced mesothelial mitogenesis.

R. R. Maronpot, E. Zeiger, E. E. McConnell, H. Kolenda-Roberts, and H. Wall were funded for this study by the North American Polyelectrolyte Producers Association; MAF was funded by SNF SAS.

Declaration of interest: The authors alone are responsible for the content and writing of the paper.