TOXICOLOGICAL SCIENCES 111(1), 4–18 (2009) doi:10.1093/toxsci/kfp131 Advance Access publication June 12, 2009

REVIEW Hemangiosarcoma in Rodents: Mode-of-Action Evaluation and Human Relevance

Samuel M. Cohen,* Richard D. Storer,† Kay A. Criswell,‡ Nancy G. Doerrer,§,1 Vicki L. Dellarco,{ David G. Pegg,k Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 Zbigniew W. Wojcinski,kj David E. Malarkey,kk Abigail C. Jacobs,# James E. Klaunig,** James A. Swenberg,†† and Jon C. Cook‡ *University of Nebraska Medical Center, Omaha, Nebraska 68198; †Merck Research Laboratories, West Point, Pennsylvania 19486; ‡Pfizer, Inc., Groton, Connecticut 06340; §ILSI Health and Environmental Sciences Institute, Washington, DC 20005; {United States Environmental Protection Agency, Washington, DC 20460; kMichigan Technology and Research Institute, Ann Arbor, Michigan 48104; kjFulcrum Pharma Developments, Inc., Ann Arbor, Michigan 48103; kkNational Institute of Environmental Health Sciences, National Toxicology Program, Research Triangle Park, North Carolina 27709; #United States Food and Drug Administration, Silver Spring, Maryland 20993; **Indiana University School of Medicine, Indianapolis, Indiana 46202; and ††University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599

Received April 7, 2009; accepted June 9, 2009

be a convergence of multiple initiating events (e.g., hemolysis, Although rarely occurring in humans, hemangiosarcomas (HS) decreased respiration, adipocyte growth) leading to either have become important in evaluating the potential human risk of dysregulated angiogenesis and/or erythropoiesis that results from several chemicals, including industrial, agricultural, and pharma- hypoxia and macrophage activation. These later events lead to the ceutical agents. Spontaneous HS arise frequently in mice, less release of angiogenic growth factors and cytokines that stimulate commonly in rats, and frequently in numerous breeds of dogs. endothelial cell proliferation, which, if sustained, provide the This review explores knowledge gaps and uncertainties related to milieu that can lead to HS formation. the mode of action (MOA) for the induction of HS in rodents, and Key Words: hemangiosarcoma; angiogenesis; endothelial cells; evaluates the potential relevance for human risk. For genotoxic endothelial precursor cells; mode of action; human relevance; chemicals (vinyl chloride and thorotrast), significant information PPAR agonists. is available concerning the MOA. In contrast, numerous chemicals produce HS in rodents by nongenotoxic, proliferative mecha- nisms. An overall framework is presented, including direct and indirect actions on endothelial cells, paracrine effects in local tissues, activation of bone marrow endothelial precursor cells, and Although rarely occurring in humans (Weiss and Goldblum, tissue hypoxia. Numerous obstacles are identified in investigations 2008b), hemangiosarcomas (HS) have taken on significant into the MOA for mouse HS and the relevance of the mouse importance in the evaluation of the potential for human risk of tumors to humans, including lack of identifiable precursor lesions, a variety of chemicals, including industrial, agricultural, and usually late occurrence of the tumors, and complexities of pharmaceutical agents. In contrast to humans, spontaneous HS endothelial biology. This review proposes a working MOA for are known to arise frequently in mice and less commonly in HS induced by nongenotoxic compounds that can guide future research in this area. Importantly, a common MOA appears to rats (Elwell et al., 2004; Haseman et al., 1998; Ruben et al., exist for the nongenotoxic induction of HS, where there appears to 1997). In addition, several breeds of dogs are known to have high spontaneous incidences of HS (Priester and McKay, This manuscript expresses the opinions of the authors and does not reflect the 1980). views of their institutions or agencies. For DNA reactive, genotoxic chemicals such as vinyl This review summarizes the presentations and discussions from an chloride and thorotrast, which are known to induce HS in international Workshop titled ‘‘Hemangiosarcoma in Rodents: Mode-of- Action Evaluation and Human Relevance,’’ which was held December 4–5, humans, significant information is known concerning the mode 2008, in Arlington, VA. This Workshop was part of the Society of Toxicology of action (MOA). In contrast, numerous commercial chemicals Contemporary Concepts in Toxicology series, was cosponsored by the ILSI that produce HS in rodents act predominantly by non-DNA Health and Environmental Sciences Institute (HESI), and was a follow-up to reactive, nongenotoxic, proliferative mechanisms. An overall a HESI project focused on MOA for peroxisome proliferator–activated receptor framework is presented which includes several proposed agonists, many of which are known to induce hemangiosarcomas in mice. 1 To whom correspondence should be addressed at ILSI Health and MOAs involving various direct and indirect actions on Environmental Sciences Institute, 1156 Fifteenth St., NW, Second Floor, endothelial cells. These include evidence for a possible role Washington, DC 20005. Fax: (202) 659-3306. E-mail: [email protected]. of adipose tissue as a source for endothelial growth factors for

The Author 2009. Published by Oxford University Press on behalf of the Society of Toxicology. All rights reserved. For permissions, please email: [email protected] RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 5 Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021

FIG. 1. Proposed HESI MOA framework for PPAR-c agonist–induced HS. agents such as peroxisome proliferator–activated receptor reactive (Nyska et al., 2004; Park et al., 2002). A preliminary (PPAR) and retinoic acid receptor (RAR) agonists, a possible examination of chronic rodent bioassays on pesticides show role for bone marrow–derived endothelial precursor cells in that increased incidences of HS occur relatively infrequently response to a net positive angiogenic stimulus, and a potential (approximately 3% of about 400 compounds evaluated in the contributing role for hypoxia as a possible area for convergence rodent bioassay) (http://www.epa.gov/pesticides/carlist/). It is across multiple MOAs (Figs. 1 and 2). Possible relationships unclear if this small number might be due to the common use between angiogenesis, a normal process in numerous bi- of 18-month bioassays in mice for pesticides in contrast to ological phenomena, and vascular tumor induction are 24 months commonly used for pharmaceuticals. HS generally discussed, with many similarities identified but with evidence are observed late in a bioassay, that is, after 18 months, of significant differences. whether spontaneously or chemically induced. Given the small The relevance of rodent hemangiomas and HS to human number of examples, it is difficult to make generalizations. cancer risk is not always known. Vinyl chloride and vinyl However, this type of response does not appear to affect bromide are known human carcinogens associated with HS, a particular chemical class of pesticides. Pesticides resulting in and also cause this type of tumor in rodents (Kielhorn et al., HS are generally not DNA reactive, although positive results 2000; Morinello et al., 2002). Thus, it is reasonable to assume in in vitro genotoxicity assays (e.g., clastogenicity) have been that when this tumor type occurs in rodents, it might be reported. Tumorigenic responses appear to be more common relevant to humans, unless there is evidence (i.e., a known in the mouse compared to the rat, and most frequently occur in MOA) to the contrary. There are likely to be several different liver, spleen, heart, or subcutaneous adipose tissue. This tumor pathways to HS following chemical exposure in rats and mice. response is generally weak and sometimes not dose-related. Ultimately, these various pathways lead either to direct MOA data are not typically available for pesticides. interaction with DNA and consequent DNA damage (DNA Increased incidences of HS, primarily in mice, have been reactivity) or to an increase in cell proliferation which increases detected with a broad range of pharmaceutical and chemical spontaneous mutations leading to an accumulation of cells that agents (Table 1), including some calcium channel blockers, become dysfunctional with respect to the maintenance of antipsychotics, PDE-5 inhibitors, DPP-4 inhibitors, antiarrhyth- mature differentiated endothelium and normal angiogenic mics, gonadotropin receptor antagonists, antisense compounds, processes (Cohen and Ellwein, 1991). nitric oxide releasers, hemolytic compounds, and vascular In contrast to the DNA reactivity and genotoxicity of vinyl endothelial growth factor (VEGF) inducers (Jacobs, 2008). chloride and vinyl bromide, most of the chemicals and Furthermore, HS in mice have been seen with many PPAR-c pharmaceuticals associated with HS in rodents are non-DNA and dual a/c agonists which have led to restrictive measures 6 COHEN ET AL. Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021

FIG. 2. Proposed model for HS induction with comparison of phenotypic responses seen with two compounds that induce HS in mice, 2-BE (Laifenfeld et al., 2009) and pregabalin. regarding clinical trials (EMEA, 2006a). Several companies species. Thus, findings with pharmaceuticals complement what have discontinued development of these drugs as a conse- has been seen for other chemicals. quence. The occurrence of HS in mice with PPAR agonists and Understanding both the chemical MOA pathways to HS some other pharmaceutical classes has challenged regulators. and any contributing species-specific susceptibility factors is During discussions on carcinogenicity by the International essential to determine the relevance of these findings to Conference on Harmonisation of Technical Requirements for humans. Analysis of MOA and human relevance is based on Registration of Pharmaceuticals for Human Use (ICH), a framework developed over the past decade through the European regulators defended the approach that positive mouse ILSI Risk Science Institute and the World Health Organization carcinogenicity data did not necessarily lead to regulatory International Programme on Chemical Safety (Boobis et al., action (Van Oosterhout et al., 1997). The significance of 2006, 2008; Meek et al., 2003; Seed et al., 2005; Sonich- PPAR-induced HS in mice is challenging because the human Mullin et al., 2001). This review summarizes the pathology and relevance of these mouse tumors remains unknown based on biology of HS, discusses interspecies comparisons, compares current knowledge gaps. and contrasts physiologic and pathologic angiogenesis, high- For pharmaceuticals, some aspects of the MOA may be lights examples of genotoxic and nongenotoxic compounds associated with exaggerated pharmacological responses or off- that induce HS, proposes an MOA framework for non- target effects and may be relevant across species. However, genotoxic compounds, and highlights the knowledge gaps. MOAs may also be species specific due to (1) species Importantly, despite the number of compound-specific initiat- differences in responsiveness to the chemical MOA, and (2) ing events (e.g., hemolysis, decreased respiration, adipocyte the interplay of this responsiveness with the genetic suscep- growth) (Table 1) that trigger nongenotoxic compounds to tibility factors governing the different spontaneous rates of HS induce HS, there appears to be a common convergence, development across species. Different classes of pharmaceut- namely, dysregulated angiogenesis and/or erythropoiesis that icals likely have different MOAs for formation of HS, but the results from hypoxia and macrophage activation. These later MOAs may, especially in mice, have a common point of events lead to the release of angiogenic growth factors and intersection and interaction with the growth regulatory path- cytokines that stimulate endothelial cell proliferation, which, if way(s) that underlie the genetic predisposition to HS in this sustained, provide the milieu that can lead to HS formation. RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 7

TABLE 1 Nongenotoxic Chemicals or Pharmaceuticals that Induce Hemangiosarcomas in Rodents and their Potential MOA

Chemical or Primary drug class pharmacology Potential MOA Reference(s)

2-Butoxyethanol Industrial solvent Hemolysis coupled with macrophage Corthals et al., 2006 activation from iron accumulation leading to increase in ROS and local tissue hypoxia Carbaryl Insecticide—acetylcholinesterase inhibitor Unknown—macrophage activation? Bigot-Lasserre et al., 2000 Dronedarone Antiarrhythmic Unknown EMEA, 2006b Elmiron Analgesic—relief of urinary bladder pain Hemolysis leading to local tissue Abdo et al., 2003 Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 hypoxia Entecavir Antivirals—nucleoside analogs Unknown Physician’s Desk Reference, 2009 Fenretinide RAR agonist Unknown—PPAR-c–like MOA Cromwell and Goldenthal, 2006; (4-hydroxyphenyl since RAR forms a heterodimer Goodman, 1991 retinamide) with PPAR-c receptor? p-Chloroaniline, Industrial solvents Hemolysis leading to iron overload Nyska et al., 2004 p-nitroaniline in macrophages and increased ROS Pregabalin a2d subunit of voltage gated Decreased respiration leading Pegg et al., 2006; Physician’s Desk calcium—analgesic, anxiolytic, to local tissue hypoxia Reference, 2009 and antiseizure Olanzapine 5-HT2/D2 antagonist—antipsychotic Unknown Physician’s Desk Reference, 2009 Troglitazone PPAR-c agonist—type II diabetes mellitus Dysregulated angiogenesis Herman et al., 2002 leading to local tissue hypoxia Tadalafil PDE-5 inhibitor—erectile dysfunction Unknown Physician’s Desk Reference, 2009 Vildagliptin DDP-4 inhibitor—type II diabetes mellitus Dysregulated angiogenesis— EMEA, 2007 inhibitor of angiogenesis

PATHOLOGY AND BIOLOGY a fivefold elevated risk of angiosarcoma compared to the general population, but the association with radiotherapy was HS is an aggressive, malignant tumor of endothelial cells not strong. Other etiologic factors in addition to radiation- which is believed to arise from endothelium of blood vessels, induced DNA damage have been suggested, such as combined and includes a variety of endothelial types, such as hepatic effects of radiation and chemotherapy, lymphedema associated sinusoidal, venous, arterial, and capillary endothelia, or a bone with radiation and/or surgical treatment, and female hormonal marrow-derived stem cell (Weiss and Goldblum, 2008b). factors (Abbot and Palmieri, 2008). Because mature endothelial cells continue to proliferate in While HS are rare in humans (Weiss and Goldblum, 2008b), adulthood, they are at risk for neoplastic transformation. HS other vascular neoplasms are much more common (Weiss and can arise in any organ, but are more commonly found in heart Goldblum, 2008a). Kaposi’s sarcoma (Barillari and Ensoli, (right atrium), liver, spleen, lung, skin, soft tissues, mammary 2002), a related spindle cell sarcoma arising from lymphatic gland, and/or bone. Local invasion and metastasis are common, endothelium that forms vascular channels filled with blood with metastasis frequently involving the lung and/or liver. cells, occurs at a 310-fold higher incidence than HS in the HIV- Metastatic lesions can be difficult to distinguish from multi- infected population. It is caused by human herpes virus 8 centrically arising HS. (HHV-8) infection, and is associated with elevated levels of Established causes in humans differ widely in pathogenetic cytokines (IL-1, IL-6, IL-8), HHV-8–encoded vIL-6, and other mechanisms. Genotoxic mechanisms appear to account for factors (leukemia inhibitory factor, oncostatin M, and cardio- liver HS induction associated with vinyl chloride exposures tropin) likely secreted by spindle cells and that act on (197 cases, from Kielhorn et al., 2000) and thorotrast (29 cases, gp130 receptors on endothelial cells (Amaral et al., 1993). reported by Kojiro et al., 1985), with median latencies of Other benign vascular neoplasms include pyogenic granulo- approximately 21 and 29 years, respectively, from onset of mas, cherry angiomas, infantile hemangiomas, and a variety exposure to death. Breast angiosarcoma (Abbot and Palmieri, of capillary and cavernous hemangiomas occurring most 2008) is associated with radiotherapy after breast-conserving commonly in the skin and liver, but seen virtually in any therapy for breast cancer with relatively short latency (mean tissue (Requena and Sangueza, 1997). Pyogenic granuloma is 72 months). However, in one study (Virtanen et al., 2007), a relatively common benign vascular lesion of the skin and breast and gynecological cancer patients were found to have mucosa which is neither infectious nor granulomatous and 8 COHEN ET AL. whose exact cause is unknown. The lesion usually occurs in the investigation of MOA for chemically induced HS has been children and young adults as a solitary glistening red papule or the lack of a clearly defined, early precursor lesion. Although nodule that is prone to bleeding and ulceration. Cherry hemangiomas and HS are frequently combined in rodent angiomas are red papules on the skin containing an abnormal bioassays for incidences of vascular lesions, hemangiomas are proliferation of blood vessels, and are the most common kind generally considered independent lesions that do not necessar- of angioma in adults. A small fraction have been linked to ily progress to HS in humans or animals (Weiss and Goldblum, chemical exposures to mustard gas (Emadi et al., 2008; Firooz 2008b). However, evidence exists that some lesions may be et al., 1999; Hefazi et al., 2006; Ma et al., 2006), 2- preneoplastic (i.e., angiomatous hyperplasia in the National butoxyethanol (Raymond et al., 1998), bromides (Cohen Toxicology Program’s [NTP] butadiene study) and that et al., 2001), and cyclosporine (DeFelipe and Redondo, occasionally HS are induced concurrently with or even arise 1998). Hemangiomas are the most common benign tumor of within benign vascular neoplasms (Hardisty et al., 2007). In infants, appearing usually within the first 2 weeks after birth, humans, there is no evidence that HS arise from hemangiomas Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 and up to five times more commonly in girls than boys (Weiss or any other benign precursor (Weiss and Goldblum, 2008b). and Goldblum, 2008a). While 50% involute by age 5, The relationship of HS to other proliferative vascular regression takes longer for others and surgical treatment may changes may depend on a limited number of disturbances in be required. There is no evidence that this type of hemangioma the genetic control of physiologically essential angiogenic progresses to HS. stimuli. Candidates for dysregulation include connective tissue There are also a number of hereditary syndromes in humans growth factor (CTGF) and CD34. CTGF is a heparin-binding in which abnormalities of the vasculature and neoplasia of growth factor produced by fibroblasts after TGFb activation vessels or other tissues are both significant features, and for that promotes angiogenesis in wound repair, tissue regenera- which an interrelationship between the changes leading to tion, and skin fibrosis (Igarashi et al., 1998). In mesenchymal tumor formation in both epithelia and the vasculature are tumors, CTGF mRNA was expressed in fibroblasts of common and apparently genetically determined. These include dermatofibromas, some dermatofibrosarcomas, angiolipoma, (1) the syndromes (von Hippel-Lindau, Chuvash polycythe- angiomyeloma, and pyogenic granuloma, but not angiosar- mia) associated with mutations in the VHL gene which controls coma, suggesting that benign fibroblasts and/or vascular the stability of the hypoxia inducible factors (HIFs) regulating endothelial cells have the capability to express CTGF mRNA VEGF-mediated angiogenic responses (Foellmi-Adams et al., when activated, but expression is lost in malignant tumors. 2004; Hickey et al., 2007), and (2) the hamartomatous Another candidate is the CD34 antigen (Lanza et al., 2001), polyposis syndromes, Bannayan-Riley-Ruvalcaba (BRRZ) a sialomucin glycoprotein involved in cell adhesion found to be syndrome and Cowden syndrome (Pezzolesi et al., 2008; expressed in dermatofibrosarcomas and angiosarcomas as well Rosner et al., 2008). VEGF-A, a specific endothelial cell as other vascular proliferative lesions. It is also expressed in mitogen, acts locally and systemically at tyrosine kinase multipotent stromal cells (MSCs) from adipose tissue but not in growth factor receptors (VEGFR-1 and 2) to stimulate bone marrow MSCs (Lazennec and Jorgensen, 2008). In endothelial cell mitogenesis and recruit endothelial progenitor summary, it is clear that a number of different potential cells (Ferrara, 2002). Other factors with a predominantly local pathways to angiosarcoma exist and that some assumptions that action in regulating angiogenic stimuli are platelet-derived have been made about pathogenesis are unfounded. growth factor, transforming growth factor b (TGFb), fibroblast- derived growth factor (FGF), and hepatocyte growth factor (HGF) (Ferrara and Kerbel, 2005). The importance of VEGF INTERSPECIES COMPARISONS signaling pathways in angiogenesis and HS is highlighted by the finding of mutations in the phosphatase and tensin homolog A review of the spectrum of spontaneous and chemically (PTEN) gene in canine and human HS; other common tumors induced vascular neoplasms in animals reveals important such as endometrium, prostate, glioma, and melanoma; and 80 species and strain/breed differences in incidence, tissue and 60% of patients with Cowden and BRRZ syndromes, distribution, tumor pathology, and susceptibility to tumorigen- respectively (Dickerson et al., 2005; Maehama et al., 2001). esis. Spontaneous HS are extremely rare (< 0.001%) in humans PTEN is a tumor and angiogenesis suppressor gene which (Weiss and Goldblum, 2008b); however, exposure-related inhibits angiogenesis due to its lipid phosphatase activity on incidences as high as 25% have occurred in humans substrates in the phosphatidylinositol-3 kinase pathway occupationally exposed to vinyl chloride (Kielhorn et al., downstream from VEGF receptor signaling. Although these 2000). In contrast, HS are relatively common in domestic and genetic disorders may shed light on vascular proliferation and experimental animals. The incidence in the dog is about 2%, angiogenesis, the vascular tumors associated with these are and it varies by breed, with the German Shepherd dog and benign; HS are not a consequence of these disorders. Golden Retrievers at greatest risk (Priester and McKay, 1980). HS in rodent bioassays, either spontaneous or chemically HS spontaneously occur in about 0.1–2% of rats and 2–5% of induced, appear late in the 2-year studies. A major obstacle to mice (Elwell et al., 2004; Haseman et al., 1998; Ruben et al., RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 9

1997). In mice, the spontaneous incidence is higher in males endothelial cells staining positive for factor VIII–related than females and higher in B6C3F1, CD-1, and BALB/c strains antigen and VEGFR2. The authors concluded that HS than CBA/J. development upon long-term exposure was likely promoted For chemicals studied for carcinogenicity by the NTP (http:// by induced changes in endothelial cell DNA adduct formation, ntp.niehs.nih.gov/index.cfm?objectid ¼ E8F394BD-1422- apoptosis, proliferation of endothelial cells having undamaged 0E5D-AD05E6097BA3D507), there have been 290 positive and/or damaged DNA, and mutation rates. Further study tests (‘‘clear’’ or ‘‘some’’ evidence) in 550 studies in mice and (Moyer et al., 2004) indicated that damage to both hepatocytes rats, with 9% (25/290 studies) associated with induction of and sinusoidal endothelial cells may lead to dysregulated vascular neoplasms, and a few chemicals causing HS VEGF synthesis by hepatocytes and activation of KDR/flk-1 incidences of > 80% (Nyska et al., 2004). In only two of the by endothelium leading to the induction of sustained 25 studies were vascular neoplasms the only neoplasm induced endothelial cell proliferation, culminating in the development (4-chloro-o-toluidine and 2-biphenyl hydrochloride). In 19 of of hepatic HS. Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 25 studies, the vascular neoplasms were observed in mice only PPAR-c agonists are an example of nongenotoxic com- (10 in both sexes), in three of 25 studies in rat only, and in pounds associated with HS induction in mice (EMEA, 2006a). three of 25 studies, vascular neoplasms were seen in both mice In 2006, a HESI PPAR Agonist Pathology Working Group and rats. The vascular system was a common site for convened to review diagnostic criteria for rodent proliferative chemically induced tumors, following liver, lung, mammary mesenchymal lesions induced by this class of pharmaceuticals gland, hematopoietic system, forestomach, and thyroid. The in rats and mice, and found that the spectrum of vascular overall incidences of hemangioma in control animals in NTP lesions in mice included angiectasis, angiomatous hyperplasia, studies were 1 and 2% for male and female B6C3F1 mice and angiolipoma, hemangioma, and HS (Hardisty et al., 2007). 0.3 and 0.25% in male and female F344 rats, with liver (mice), With the exception of angiomatous hyperplasia and angioli- skin (mice and rats), uterus (mice and rats), and ovary (mice) poma, the vascular neoplasms were similar in appearance to being the most common spontaneous sites (Elwell et al., 2004; those described in publications by the Society of Toxicologic Nyska et al., 2004). For HS, the overall incidences in controls Pathology (STP) (Elwell et al., 2004; Ruben et al., 1997). were 5% for male and female B6C3F1 mice and 0.5 and 0.4% Interestingly, a small number of cases was observed where HS in male and female F344 rats, with liver (mice), skin (mice and was seen arising within angiolipomas. However, it remains rats), spleen (mice and rats), bone marrow (mice), heart (mice unclear whether the angiolipomas were truly precursor lesions and rats), uterus (mice and rats), and ovary (mice) being the for the HS. In rats, PPAR-c agonists did not induce vascular most common spontaneous sites. For the B6C3F1 hybrid neoplasms but increased the incidence of lipoma, liposarcoma, mouse strain used in NTP studies, there has been considerable fibrosarcoma, and mixed mesenchymal tumors, predominantly variability in spontaneous incidence rates in different reports in the subcutis (Hardisty et al., 2007). for hemangioma (1–15%) and HS (0–5%). The overall incidences of chemically induced HS in positive NTP studies were 22% (range 8–100%) and 25% (range 4–75%), re- PHYSIOLOGIC AND PATHOLOGIC ANGIOGENESIS spectively, for mice and rats. The five compounds with the highest reported incidence rates in NTP studies for HS (> 75% Physiological angiogenesis is an essential part of many incidence) were riddelliine, 2-methyl-1-nitroanthroquinone, continuously running processes, including normal tissue de- cupferron, tetrafluoroethylene, and o-nitrotoluene. The affected velopment, wound healing, and endometrial cycling (Carmeliet, sites in positive studies were general/multiple sites, liver, 2005; Ferrara and Kerbel, 2005). Vascular neoplasms are spleen, skin/subcutis, abdomen/mesentery, heart, urinary assumed to reflect aberrations in growth control of cells that are bladder, and nasal cavity. Potential mechanisms that have essential participants in normal vasculogenesis and angiogen- been suggested include iron overload/hemosiderosis/hemoly- esis. Physiologic angiogenesis is an intermittent, localized and sis, hormonal perturbations, reduced antioxidant defense tightly regulated response reflecting mainly an adaptive re- mechanisms, genotoxic events, induced effects on rates of cell sponse to local hypoxia, and is dependent on the local proliferation and/or apoptosis, and dysregulation of cytokines accumulation of HIFs (Harris, 2002; Hickey et al., 2007). HIFs and growth factors. activate key angiogenic factors such as VEGF, whose For riddelliine, a naturally occurring, genotoxic, pyrrolizi- pleiotropic effects include increased capillary permeability, dine alkaloid which induced a high incidence of hepatic HS in stimulation of endothelial cell migration and proliferation, and both rats and mice, short-term mechanistic studies (Nyska the synthesis and release of the matrix metalloproteases (MMP- et al., 2002) showed reduced mitoses, increased hypertrophy 2) involved in matrix degradation. Downregulation of inhibitory and fatty degeneration in hepatocytes, with endothelial cells factors such as angiopoietin 1 acting at the Tie2 receptor to showing karyomegaly, cytomegaly, decreased apoptosis, more maintain a quiescent state in mature endothelium is also required S-phase nuclei, and p53 positivity. Hepatocytes of treated to allow angiogenesis to proceed. In contrast, pathologic animals expressed higher VEGF immunopositivity and altered angiogenesis as occurs in cancer, retinopathies, vascular 10 COHEN ET AL. neoplasms, and other disease states is a sustained localized heterogeneity of the vasculogenic phenotype as measured by response characterized by tortuous, irregular, and leaky vessels. the number of circulating endothelial cells (Shaked et al., 2005) Activation of the ‘‘angiogenic switch’’ is thought to involve has also been shown to parallel angiogenic responsiveness in a balance between the growth factor activation pathways and the corneal micropocket assay, suggesting this as a potentially a plethora of angiogenesis inhibitors such as thrombospondin-1, useful surrogate clinical marker of angiogenic responsiveness 16-kDa prolactin, interferon a/b, platelet factor-4, and angio- that may have utility in predicting the progression of human statin (Carmeliet, 2005; Ferrara, 2002; Ferrara and Kerbel, 2005; angiogenic diseases. Harris, 2002; Kerbel and Folkman, 2002; Nyberg et al., 2005). Canine HS is reported to account for ~7% of dog cancers Recent advances in our understanding of tumor angiogenesis (Priester and McKay, 1980). Of the estimated 65 million dogs have also highlighted the important roles of circulating, bone in the US, 1.5–2.5 million will likely be stricken. Because the marrow–derived progenitor cells, including both endothelial and tumor is poorly responsive to conventional therapy, most dogs monocyte/macrophage lineage cells to pathological angiogen- will die from it within 1–8 months depending on the Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 esis (Ding et al., 2008; Lamerato-Kozicki et al., 2006). These therapeutic modalities provided. A study of the molecular findings suggest that HS may arise not only from transformation pathogenesis of canine HS showed no evidence of a canine of tissue-resident endothelial cell populations, but also from gamma herpes virus or mutations in the ras or VHL genes. circulating progenitors or adult stem cells recruited from bone However, mutations in the C-terminal domain of PTEN were marrow or possibly also from extramedullary sites of hemato- common, and may represent late events in tumor progression poiesis such as the liver and spleen. (Dickerson et al., 2005). Gene expression profiling clusters HS The coordination of the recruitment, proliferation, localiza- separately from nonmalignant (hemangioma) samples with tion, and differentiation of the different cell types that regulate enriched genes in HS in inflammation (IL-8, IL-6, PTGS2), angiogenic processes is exquisitely complex, involving many angiogenesis and vessel response (VEGF-A, bFGF, ADM, genes and offering a multitude of opportunities for genetic PTGER4), adhesion (CD44, CDH2, FN1, NCAM1, VCAM1, diversity (polymorphisms) to introduce species, strain, and ITGB3, and TNC), signaling/cell cycle/metabolism/mitosis individual animal variation in angiogenic responses. The (HGF, GUCY1A3, CDKN1a, CCND1, SLC2A3, SSP1), genetic diversity in angiogenesis-regulating genes has been invasion (MMP-9, PLAU), and patterning (HOX-A10) linked to increased susceptibility to multiple angiogenesis- (Tamburini et al., unpublished data). dependent diseases in humans, including cancer (Rogers and Researchers in canine HS have also tested the hypothesis D’Amato, 2006; Rogers et al., 2003, 2004). The spectrum of that genetic background (defined as ‘‘breed’’) influences genetic differences may involve one or a few genetic loci, such phenotypes and behavior of sporadic tumors. Results of this as in Mendelian traits that are rare in the population, are work have shown that tumor-restricted expression of proin- minimally affected by environmental influences, and do not flammatory and angiogenic genes in HS of Golden Retrievers exhibit phenotypes in progeny that exceed the range of parental cluster separately from HS of non-Golden Retrievers, suggest- phenotypes. For example, in subsets of individuals with ing that heritable factors mold phenotypes in sporadic, infantile hemangiomas (localized and rapidly growing regions naturally occurring tumors. Gene sets enriched in Golden of disorganized angiogenesis), missense mutations in the genes Retrievers compared to other breeds included genes overex- encoding VEGFR2 (KDR) and TEM8 (ANTXR1) result in pressed in plasmablastic cells (immune response), genes suppressed NFAT-dependent VEGFR1 expression and consti- downregulated during differentiation of 3T3L1 fibroblasts into tutive VEGFR2 signaling (Jinnin et al., 2008). Other examples adipocytes with IDX (mesenchymal differentiation), genes include mutations in receptors in TGFb/BMP signaling upregulated in pulpal tissue from extracted cavities (inflam- pathway genes (endoglin, ALK1) in hereditary hemorrhagic mation), genes known to be induced by hypoxia (angiogene- telangiectasias and Tie-2 mutations in venous malformations. sis), and genes upregulated by HGF (endothelial growth Genetic differences in angiogenic responses also result from response and survival) (Tamburini et al., unpublished data). complex traits affected by many loci that are common in the Ontogeny analysis showed that 25/77-signaling pathways population and can be strongly affected by environmental intersect at IL-8, and 6/77 pathways intersect at VEGFR1 influences. For example, polymorphisms in the human VEGF (FLT-1) with recurrent enrichment shown for IL-1 response gene show significant associations with a variety of conditions genes and acid ceramidase in Golden Retrievers with HS as and diseases (Rogers et al., 2004). Ongoing mapping studies compared to non-Golden Retrievers with HS. Future research (Rogers et al., 2003, 2004) have identified multiple loci that directions include efforts to determine how VEGFR-1 and -2 control angiogenic responsiveness in several mouse models activity and other metabolic pathways influence pathogenesis such as the corneal micropocket and laser-induced choroidal of canine HS and if canine HS may arise from a myeloid neovascularization assays, raising the possibility that this progenitor or, as previously suggested (Lamerato-Kozicki approach may ultimately reveal genetic loci in mice that et al., 2006), from multipotential bone marrow–derived stem modulate susceptibility not only to angiogenic responsiveness cells whose progeny arrest differentiation at the hemangioblast but also to spontaneous and/or chemically induced HS. Genetic or angioblast stage. RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 11

Together, these data suggest that understanding the MOA could have provided such a stimulus. Because identical DNA and human relevance of chemically induced HS in rodents will adducts are produced endogenously that also occur after vinyl need to consider not only the dose response relationships and chloride exposure, risk assessments based on extrapolations to chemical MOA for agents associated with HS induction, but low-dose exposure are less clear. Furthermore, if steatohepatitis also the intrinsic sensitivity and susceptibility of different is required for the induction of HS by VC, it is likely to be rodent test species and strains and the genetic basis for these a threshold toxicological phenomenon. species differences. The recent investigations of the Louisville VC cohort also demonstrate that routine liver chemistries are not adequate as tumor markers for liver angiosarcomas, but preliminary data VINYL CHLORIDE: A DNA REACTIVE CARCINOGEN suggest that hyaluronic acid might be a reasonable early marker for future clinical development of HS (Cave et al., 2008a). Vinyl chloride (VC) is a known animal and human carcinogen Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 that induces HS of the liver as its hallmark cancer following inhalation exposure. In humans and animals, VC carcinogenesis NON-DNA REACTIVE NONPHARMACEUTICAL is associated with relatively high exposures ( 50 ppm). VC is CHEMICALS metabolized primarily by CYP2E1 to the epoxide, chloro- ethylene oxide, a highly unstable chemical that reacts with local Carbaryl induced an increased incidence of HS in both genders cellular DNA and proteins, and is thought to have minimal of mice but not in rats (US EPA, 2004). However, the increased systemic exposure. Four DNA adducts have been identified and incidence only occurred in mice administered carbaryl at a dietary have had molecular dosimetry studies conducted (Morinello dose of 8000 ppm which was in excess of the maximum tolerated et al., 2002). The major DNA adduct is 7-(2-oxoethyl) guanine dose based on decreased weight gain. Carbaryl was non- (OEG), which comprises ~98% of the DNA adducts. This genotoxic in a battery of in vitro assays except for clastogenic adduct is lost from DNA primarily by chemical depurination and activity in an in vitro Chinese hamster ovary assay with S9 is not considered to be promutagenic. Three etheno adducts are activation. However, it was negative in an in vivo chromosomal also formed and are thought to be promutagenic. All four of the aberration study, and also was negative when tested for DNA adducts of VC exhibit supralinear exposure response carcinogenicity in the heterozygous p53 knockout mouse model. curves due to saturation of metabolic activation that results in a 2-Butoxyethanol (BE) has been reported to induce an greater number of adducts being formed per ppm at lower increase in liver HS in male B6C3F1 mice following chronic exposures and plateauing of the exposure response at high inhalation, but not in female mice or in either gender in rats exposures. Furthermore, four adducts are formed endogenously (Klaunig and Kamendulis, 2005). The mechanisms involved in from lipid peroxidation products. Thus, there is always a BE-induced HS formation are not clear, but data suggest the background of identical endogenous adducts present. By expo- involvement of oxidative damage subsequent to red blood cell 13 sing rats to [ C2]-VC, the half-lives of all four of the VC- (RBC) hemolysis and iron deposition and activation of Kupffer induced adducts were determined. N2,3-ethenoguanine(eG) has cells in the liver, events that exhibit a threshold in both animals been shown to have a half-life of ~150 days, suggesting that it is and humans (Corthals et al., 2006; Klaunig and Kamendulis, very poorly repaired, if at all. In contrast, 1,N6-ethenodeox- 2005; Park et al., 2002). In isolated mouse and rat hepatocytes, yadenosine (edA) and 3N4-ethenodeoxycytidine are actively neither BE or its major metabolite, 2-butoxyacetic acid (BAA), repaired and have half-lives of ~1 day, while OEG has a half-life increased oxidative DNA damage (OH-8-dG), lipid peroxida- of ~4 days as a result of chemical depurination. Hepatic HS in tion (MDA formation), or decreased vitamin E concentrations, animals and humans contain mutations in the p53 tumor while both ferrous sulfate (iron) and hemolyzed RBCs suppressor gene and ras oncogenes that are compatible with the produced dose-related changes in biomarkers of oxidative mutations caused by the three etheno adducts. The MOA for VC stress (Klaunig and Kamendulis, 2005). Comparatively, mouse fits well with a mutagenic MOA. hepatocytes were more sensitive to oxidative stress by iron and However, recent investigations on a worker cohort in hemolyzed RBCs when compared to the rat. In the Syrian Louisville that was exposed to high concentrations of VC Hamster Embryo cell transformation assay, BE and BAA did demonstrated that approximately 80% showed significant not induce cellular transformation. In contrast, iron produced steatohepatitis that was not related to alcohol ingestion, dose-related increases in cell transformation, OH-8-dG, and diabetes, or obesity (Cave et al., 2008b). The 25 workers DNA damage, effects that were prevented by coexposure to who developed HS also had VC-associated steatohepatitis, antioxidants (vitamin E or EGCG). In cultured endothelial suggesting the possibility that it was a significant contributor to cells, BE, BAA, and the aldehyde intermediate, butoxyace- the pathogenesis of the HS, possibly contributing an increased taldehyde, did not induce DNA damage, using the single cell proliferative stimulus to liver endothelial cells in which DNA gel electrophoresis assay (COMET) (Corthals et al., 2006). mutation had occurred from the VC exposure. Release of However, iron, hemolyzed RBCs, and hydrogen peroxide a variety of cytokines, such as IL-1 and IL-6, from hepatocytes, increased DNA damage in endothelial cells. 12 COHEN ET AL.

Additional studies showed that activated macrophages regulate cell growth and angiogenesis, quite possibly through increased both endothelial cell DNA damage and DNA RAR-independent pathways. The issue of off-target receptor synthesis, suggesting a role for macrophage activation in interaction is another of several critical issues that needs to be increased cell proliferation in endothelial cells (Corthals et al., addressed in extrapolating the observations in mice to humans. 2006). Complementary in vivo studies have also been Endothelial cell apoptosis appears to be induced by performed. In a subchronic study in mice and rats, BE induced fenretinide via increased production of ceramide by mecha- hemolysis in both species at all time points (7, 14, 28, 90 days) nisms not yet elucidated (Johnson, 2008). Fenretinide, like and doses (225 and 450 mg/kg [rats]; 225, 450, and 900 mg/kg several other retinoids, inhibits the growth of a variety of [mice]). Evidence of hemolysis was also shown by an increase neoplastic cell lines, including breast and oral cancers in iron-stained Kupffer cells in both species, at the two highest (Johnson, 2008). PPAR-c transactivation is induced, as is doses of BE. Increases in OH-8-dG and MDA were biphasic increased nitric oxide production. No information is available and species selective: increases were seen after 7 and 90 days, as to whether these effects, particularly the association with Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 selectively in mouse liver (Corthals et al., 2006). BE also PPAR-c activation, are associated with the induction of HS. produced a biphasic and species selective induction of DNA Likewise, two metabolites of fenretinide which do not bind to synthesis in mouse liver; endothelial cell DNA synthesis RAR have been identified, N-(4-methoxyphenyl) retinamide increased early after BE exposure (7 and 14 days), while DNA and 4-oxo-N-(4-hydroxyphenyl) retinamide, but their biolog- synthesis in hepatocytes increased only at 90 days. ical effects are mostly unknown. To assess the role of Kupffer cells in BE-induced endothelial Several possible factors have also been demonstrated to cell DNA synthesis, mice were given BE in the presence or be involved in the HS induced in mice by pregabalin. Most of absence of Kupffer cell depletion (via clodronate-encapsulated these tumors occurred in the tissues where spontaneous HS are liposomes [CL]) for 7 days. BE increased the number of commonly observed in mice, namely liver, spleen, and bone Kupffer cells (F4/80þ cells), while CL decreased Kupffer marrow (Criswell, 2008; Pegg et al., 2006). HS were induced cell number by > 90% (Corthals et al., 2006; Klaunig and in B6C3F1 and CD1 mice, but no tumors of any tissue were Kamendulis, 2005; Park et al., 2002). BE produced the induced in rats. anticipated effects, including increased hemolysis, iron-stained The precipitating event in pregabalin-associated HS de- nonparenchymal cells, and increased endothelial cell DNA velopment in mice appears to be decreased respiratory synthesis. However, all values were similar to controls in function, including decreased respiration rate and decreased Kupffer cell-depleted mice, suggesting that this cell type minute volumes, along with alterations in acid-base balance, participates in BE-induced endothelial cell DNA synthesis. gradually shifting to a relative alkalotic state (Criswell, 2008). Collectively, evidence from in vitro and in vivo studies show Effects on respiration are also observed in rats administered that BE does not elicit direct effects on the liver, and provides pregabalin but, unlike mice, rats appear to adequately suggestive evidence that the MOA for BE-induced liver HS compensate with no resultant change in acid-base balance. involves indirect effects including hemolysis, Kupffer cell The shift toward a more alkalotic state in mice, a species which modulation, and oxidative stress (Klaunig and Kamendulis, normally has significantly more acidic blood than rats or 2005; Nyska et al., 2004). humans, presumably could lead to decreased oxygen release from hemoglobin and subsequent local tissue hypoxia at high doses of pregabalin, initiating a cascade of events, including NON-DNA REACTIVE PHARMACEUTICALS the overproduction of megakaryocytes and platelets, bone marrow congestion, macrophage infiltration and accumulation Several classes of pharmaceuticals have been associated with of iron in the macrophages with consequent release of various increased incidences of HS (Table 1). Three of these classes endothelial growth factors, including VEGF and PDGF and (retinoids, pregabalin, and PPAR agonists) have MOA in- increased endothelial cell proliferation. However, the rats formation that has been published in the peer-reviewed appear to adequately compensate whereas the mice do not. scientific literature or elsewhere in the public domain and will be highlighted in this review. Fenretinide is a prototypical retinoid which induces nearly PPAR AGONISTS a 100% incidence of HS at several sites in mice, but like the PPAR agonists, most are in adipose tissue, either subcutaneous Troglitazone is the prototypical low affinity (lM) PPAR-c or in the abdominal cavity (EMEA, 2006a; Herman et al., agonist developed to treat Type II diabetes. In carcinogenicity 2002; Johnson, 2008). Unlike many of the nongenotoxic studies, it produced an increased incidence of HS in mice but pharmaceuticals associated with HS in mice, the tumors not rats (Herman et al., 2002). These tumors were primarily induced by fenretinide occurred earlier (e.g., as early as 1 seen in skin (usually in subcutaneous adipose tissue but year). These compounds bind the RAR, and evidence based on included in skin histologic section) consistent with its in vitro investigations demonstrates that they negatively pharmacologic action. The proposed HESI MOA framework RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 13 for PPAR-c–induced HS is presented below, using principally linkage between adipogenesis and angiogenesis. Whether these data from troglitazone, where appropriate, to illustrate these temporal changes also occur at the same dose levels as the components. An outline of the framework is illustrated in carcinogenicity study remains to be demonstrated. Figure 1. Troglitazone was also evaluated for direct effects on Troglitazone binds the PPAR-c receptor, resulting in endothelial cell cultures from microvascular endothelial cells a conformational change that activates the transcriptional obtained from skin and subcutaneous adipose tissue from mice regulatory activity of the receptor and correlates with its (MFP MVEC) or humans (human microvascular endothelial antidiabetic actions (Berger et al., 1996). Troglitazone and cell [HMEC1]; Kakiuchi-Kiyota et al., 2009). The microvas- several other PPAR-c/PPAR dual agonists induce HS in mice culature of skin and adipose tissue was evaluated because of (EMEA, 2006a; Herman et al., 2002). Muraglitazar, pioglita- the predilection for PPAR agonist–induced HS to develop at zone, and rosiglitazone also bind and activate PPAR-c, but do these sites (Hardisty et al., 2007). In cytotoxicity experiments, not produce HS; therefore, PPAR-c binding appears to be the LD50 for troglitazone on day 3 was 17.4lM in HMEC1 Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 necessary, but not sufficient, to produce HS (Pioglitazone SBA; and 92.2lM for MFP MVEC cells. These concentrations are Rosiglitazone SBA; Tannehill-Gregg et al., 2007; Waites et al., above the therapeutic blood level in humans and above the 2007). The reasons for induction of HS by some PPAR-c sarcomagenic blood levels in mice. Therapeutic blood levels in agonists and not others are not known. Possible hypotheses humans at an orally administered daily dose of 400 mg include quantitative differences in receptor activation, differ- produces a C-max of 3.4lM. Furthermore, there was no effect ential recruitment of coactivators and corepressors, off-target of troglitazone on viability of HMEC1 cells grown in medium receptor activities, or non-PPAR receptor activities, any of with reduced growth factors, a standard method for assessing which could potentially impact cell cycle kinetics. Troglitazone direct mitogenic effects. In contrast, troglitazone increased and the other PPAR agonists are nongenotoxic in vitro and in viability of the MFP MVEC cells at low concentrations, vivo. The HS induced by troglitazone did not show evidence of comparable to the blood concentrations of mice treated with mutations in p53 or ras genes, in contrast to what is frequently sarcomagenic doses of troglitazone. The increased viability seen in human HS (Duddy et al., 1999a, 1999b) appears to be due to a combination of increased cell Troglitazone induces an expansion of brown fat and proliferation (assessed by 3H-thymidine incorporation) and promotion of brown adipocyte differentiation, increases adi- decreased apoptosis (assessed by flow cytometry using pocytes in Zucker rats, and promotes adipocyte differentiation Annexin V and propidium iodide staining). These in vitro of 3T3 fibroblasts in vitro (Okuno et al., 1998; Tai et al., 1996; studies suggest a fundamental difference between mouse and Tafuri, 1996). Troglitazone-induced brown fat morphology human endothelial cells in their biologic response to troglita- shifts from a multilocular to unilocular phenotype from zone. Similar studies with other PPAR agonists will ultimately upregulation of the mitochondrial uncoupling protein (UCP-1). be needed to fully compare and extrapolate results between Similar effects have been reported with pioglitazone (Foellmi- mice and humans. Adams et al., 1996; Sears et al., 1996). Hence, PPAR-c Adipocytes are a rich source of angiogenic growth factors agonists induce genes in adipocytes that drive differentiation (e.g., bFGF, VEGF); it is well recognized that adipogenesis is and their growth. tightly linked with angiogenesis (Fukumura et al., 2003; Troglitazone induces HS in male and female mice in the 2- Hutley et al., 2001; Kershaw and Flier, 2004; Rupnick et al., year bioassay at both 400 and 800 mg/kg/day (Herman et al., 2002; Sierra-Honigmann et al., 1998). Troglitazone-induced 2002). C-max at these doses is 79.9 and 86.9lM, respectively. HS occur in highest frequency at sites associated with fat: At the doses of 400 and 800 mg/kg/day in mice, there is an adipose tissue, bone marrow, and skin/subcutis (Duddy et al., increased labeling index of endothelial cells in white and 1999a; Herman et al., 2002). There is some evidence that brown fat and in the liver after 4 weeks of administration, but troglitazone and other PPAR-c agonists increase angiogenic not at 1 or 2 weeks (Kakiuchi-Kiyota et al., unpublished growth factors such as VEGF and possibly others. For instance, observations). in vitro cultures of 3T3-L1 cells (an adipocyte differentiation In a 2-week time-course experiment using BrdU adminis- model) demonstrated that pioglitazone and rosiglitazone pro- tered via osmotic pumps, troglitazone, administered at a dose duced smaller increases in VEGF levels and adiponectin than 50% greater than the highest dose used in the 2-year bioassay, troglitazone, consistent with the ability of these agents to increased adipocyte proliferation 3.9- and 1.5-fold at 1 and 2 induce HS (Johnson, 2008). The relative contribution of weeks, respectively (Breider et al., 1999). Interscapular brown angiogenic and/or inhibition of antiangiogenic growth factors fat weight was increased at 2 weeks (1.21-fold), but not at 1 (Nyberg et al., 2005) needs further study but is inhibited by the week (Breider et al., 1999). Troglitazone also produced difficulty of these measurements. The large number of a numerical increase in endothelial cell proliferation at 2 angiogenic and antiangiogenic growth factors precludes in- weeks (3.2-fold), but not at 1 week (Breider et al., 1999). These dividual measurement but rather requires multiplex technolo- data demonstrate that adipocyte proliferation precedes endo- gies and/or transcriptomic approaches to holistically assess the thelial cell proliferation, consistent with the proposed temporal net angiogenic environment. 14 COHEN ET AL.

Dysregulated angiogenesis has been hypothesized as the Assessment of the hypothesis that HS arises from basis for HS induction based on the following observations. recruitment of EPCs; PPAR-c agonists stimulate adipocyte growth resulting in the Objective tools to identify preneoplastic lesions, if they release of angiogenic growth factors that stimulate angiogenesis exist; to support the growth of the fat pad. In vitro data demonstrate More data on species specificity of HS; that PPAR-c agonists can inhibit endothelial growth (Bishop- In vitro comparative (mouse, rat, human) endothelial cell Bailey and Hla, 1999; Murata et al., 2001). Hence, angiogen- proliferation studies (the limitation is the difficulty in culturing esis sufficient to support the growth of the fat pad is attenuated sinusoidal-derived endothelial cells); and resulting in local tissue hypoxia. Alternatively, local tissue Pharmacogenomic assessment of mouse susceptibility to hypoxia may occur due to the inability of angiogenesis to keep HS. pace with growth of the fat pad. The local tissue hypoxia results in HIF-1a activation which can activate VEGF secretion and/or Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 macrophage activation. Macrophage activation leads to cyto- kine release (IL-1, IL-6), generation of reactive oxygen species, SPECIES SPECIFICITY and inflammation (Corthals et al., 2006; Stienstra et al., 2008). All of these changes can promote angiogenesis. Whether any of The species specificity of HS is not well understood, these relationships are also applicable to the induction of PPAR especially in the case of nongenotoxic compounds. There are agonist–induced HS is unknown. several lines of evidence that suggest that the mouse is more In addition to local stimulation of endothelial cell pro- susceptible to the induction of HS than humans: liferation by angiogenic growth factors, an alternative hypoth- esis is that the target organ (i.e., fat pad in the case of PPAR-c (1) Spontaneous incidence in mice is higher than in rats or agonists) provides an angiogenic growth stimulus that recruits humans, which suggests a genetic component as discussed in circulating endothelial progenitor cells (EPCs) from the bone the Interspecies Comparisons section. marrow, and these stems cells seed the target organs leading to (2) Spontaneous cell proliferation rates of endothelial cells the formation of HS. The EPCs migrate to the target organ in across mice, rats, and humans correlate with the spontaneous response to a gradient of angiogenic growth factors. Rosigli- incidence of HS (Ohnishi et al., 2007). tazone and pioglitazone have been shown to recruit EPCs to (3) Troglitazone has been shown to stimulate cell pro- blood vessels or adipocyte tissue (Crossno et al., 2006; Werner liferation in mouse microvascular endothelial cells but not et al., 2007); however, they do not induce HS. A potential HMEC (Kakiuchi-Kiyota et al., 2009). differentiating factor may be that troglitazone also inhibits (4) Human disease states suggest that humans are more mitochondrial function unlike pioglitazone and rosiglitazone. resistant than mice for development of HS. For instance, Sustained proliferation of endothelial cells can result in Chuvask polycythemia is a hypoxia-sensing disorder charac- transformation from errors during DNA replication that are not terized by a homozygous mutation of the VHL gene (Gordeuk repaired. In addition to this classical mechanism, there is et al., 2004). These individuals have elevated levels of HIF-1a growing evidence that hypoxia can drive and maintain genetic throughout their life, yet develop hemangiomas but not HS. In instability and a ‘‘mutator’’ phenotype (Bristow and Hill, contrast, mice with a similar mutation in the VHL gene develop 2008). Additional contributing factors for the transformation of HS (Hickey et al., 2007). endothelial cells include possible indirect DNA damage by (5) Mice have lower antioxidant levels (e.g., Vitamin E) macrophage activation and/or reduced antioxidant levels, as than rats and humans which may contribute to their enhanced was shown with BE (Corthals et al., 2006). susceptibility to developing HS (Corthals et al., 2006).

KNOWLEDGE GAPS

In addition to the issues described above, the HESI MOA SUMMARY framework analysis for PPAR-c–induced HS identified other In summary, HS induced by a variety of chemicals in rodent knowledge gaps, several of which were discussed at the bioassays, particularly in mice, have increasingly become December 2008 Workshop. These knowledge gaps included a significant issue with respect to regulatory decisions, the following: especially for pharmaceuticals. A major difficulty in dealing Data on endothelial cell proliferation using dual labeling with these rodent HS has been the lack of understanding of an methodology across PPAR-c agonists; MOA. Although significant factors in the MOA have been Absence of evidence for adipokine secretion, and more elucidated for DNA reactive carcinogens such as vinyl compelling data on the role of angiogenic/antiangiogenic chloride, there are relatively little data available regarding growth factors; MOAs for non-DNA reactive carcinogens inducing HS in RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 15 mice, nor are there definitive data to address the issue of their Sankyo; GlaxoSmithKline; Merck Research Laboratories; human relevance. However, developments in basic biology, Pfizer, Inc.; sanofi-aventis; Takeda Pharmaceutical Company; particularly related to research on angiogenesis, are providing the Society of Toxicology Regulatory and Safety Evaluation significant advances in knowledge, and are beginning to Specialty Section; and the Society of Toxicologic Pathology. identify possible MOAs and experimental approaches to addressing critical questions. The MOA frameworks proposed in Figures 1 and 2 are the ACKNOWLEDGMENTS first report of a unified model for induction of HS by nongenotoxic compounds. Figure 1 summarizes the proposed The authors gratefully acknowledge the individuals who MOA Framework for PPAR-c agonist–induced HS. A critical presented at the December 2008 Workshop and to those who component in this MOA is the concept of dysregulated presented recent research findings at the Workshop poster angiogenesis which can result in local tissue hypoxia and the session. A list of Workshop presenters and their abstracts can Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 subsequent HIFa and/or macrophage activation. HIFa activa- be found at http://www.toxicology.org/ai/meet/hr-COF_mee- tion would lead to increased VEGF levels while macrophage ting.asp. Sincere appreciation is extended to the many scientists activation would increase IL-6, both of which can stimulate who attended the Workshop and participated in the active endothelial cell proliferation. The concept of dysregulated exchange of ideas and information that are summarized in this angiogenesis has been proposed for troglitazone, fenretinide, paper. Finally, the authors thank Dr Kendall Wallace for his and vildagliptin. Figure 2 proposes a general model for HS leadership and guidance during the HESI peer review of this induction incorporating these concepts and compares the manuscript prior to journal submission. phenotypic responses seen with two other compounds that Contributions to the Workshop and this manuscript were induce HS in mice, 2-BE and pregabalin. Both agents can supported, in part, by the Intramural Research Program of the induce local tissue hypoxia: 2-BE via its induction of National Institutes of Health National Institute of Environmen- hemolysis, and pregabalin via its respiratory suppression. Both tal Health Sciences. compounds have been shown also to activate macrophages which may play a contributing role in the MOA. Hence, there do appear to be areas where divergent initiating events lead to REFERENCES a convergence around hypoxia and macrophage activation. Table 1 demonstrates that there are potentially multiple Abbot, R., and Palmieri, C. (2008). Angiosarcoma of the breast following surgery and radiotherapy for breast cancer. Nat. Clin. Pract. Oncol. 5, 727–736. initiating events that lead to HS, while Figure 2 suggests that Abdo, K. M., Johnson, J. D., and Nyska, A. (2003). Toxicity and there may be convergence of these MOAs via dysregulated carcinogenicity of Elmiron in F344/N rats and B6C3F1 mice following angiogenesis or erythropoiesis leading to hypoxia coupled with 2 years of gavage administration. Arch. Toxicol. 77, 702–711. macrophage activation. This general model for HS provides the Amaral, M. C., Miles, S., Kumar, G., and Nel, A. E. (1993). Oncostatin-M opportunity to test whether convergence does occur for other stimulates tyrosine protein phosphorylation in parallel with the activation of nongenotoxic compounds that induce HS. p42MAPK/ERK-2 in Kaposi’s cells. Evidence that this pathway is important Regarding human relevance, there appear to be significant in Kaposi cell growth. J. Clin. Invest. 92, 848–857. differences between the responses of endothelial cells to Barillari, G., and Ensoli, B. (2002). Angiogenic effects of extracellular human immunodeficiency virus type 1 Tat protein and its role in the pathogenesis of various chemicals in different tissue sites, as well as between AIDS-associated Kaposi’s sarcoma. Clin. Microbiol. Rev. 15, 310–326. species. It remains unclear as to whether these responses Berger, J., Bailey, P., Biswas, C., Cullinan, C. A., Doebber, T. W., represent sufficient qualitative differences between rodents and Hayes, N. S., Saperstein, R., Smith, R. G., and Leibowitz, M. D. (1996). humans to impact the overall risk assessment for these Thiazolidinediones produce a conformational change in peroxisomal chemicals. The only means to ultimately address the question proliferator-activated receptor-gamma: Binding and activation correlate with of interspecies comparisons is to develop a better understand- antidiabetic actions in db/db mice. Endocrinology 137, 4189–4195. ing of the MOAs leading to the development of these tumors, Bigot-Lasserre, D., Chuzel, F., Debruyne, E., Urtizberea, M., and Carmichael, N. (2000). Tumorigenic potential of carbaryl in the P53 and to identify further biologic similarities and differences knockout mouse model. The Toxicologist 54, 1038. between rodents and humans. Bishop-Bailey, D., and Hla, T. (1999). Endothelial cell apoptosis induced by the peroxisome proliferator-activated receptor (PPAR) ligand 15-dioxy-D12, 14-prostaglandin J2. J. Biol. Chem. 274, 17042–17048. FUNDING Boobis, A. R., Cohen, S. M., Dellarco, V., McGregor, D., Meek, M. E., Vickers, C., Willcocks, D., and Farland, W. (2006). IPCS framework for The Society of Toxicology and the ILSI Health and analyzing the relevance of a cancer mode of action for humans. Crit. Rev. Toxicol. 36, 781–792. Environmental Sciences Institute provided financial and Boobis, A. R., Doe, J. E., Heinrich-Hirsch, B., Meek, M. E., Munn, S., logistical support for the December 2008 workshop. The Ruchirawat, M., Schlatter, J., Seed, J., and Vickers, C. (2008). IPCS following companies provided the financial support: Aclairo framework for analyzing the relevance of a noncancer mode of action for Pharmaceutical Development Group; AstraZeneca; Daiichi- humans. Crit. Rev. Toxicol. 38, 87–96. 16 COHEN ET AL.

Breider, M. A., Gough, A. W., Haskins, J. R., Sobocinski, G., and de la trophic and vascular characteristics (case report, 16-year post-exposure). Iglesia, F. A. (1999). Troglitazone-induced heart and adipose tissue cell Ecotoxicol. Environ. Saf. 69, 574–576. proliferation in mice. Toxicol. Pathol. 27, 545–552. European Medicines Agency (EMEA). (2006a). CHMP SWP Reflection Paper Bristow, R. G., and Hill, R. P. (2008). Hypoxia, DNA repair, and genetic on PPARs (Peroxisome Proliferator Activated Receptors). EMEA/341972/ instability. Nat. Rev. 8, 180–192. 2006, September 2006, London. Carmeliet, P. (2005). Angiogenesis in life, disease and medicine. Nature 438, European Medicines Agency (EMEA). (2006b). Withdrawal Public Assessment 932–936. Report of the Marketing Authorisation Application for Multaq (Dronedar- Cave, M., Falkner, K., Patel, M., Reynolds, L., Barve, S., and McClain, C. one). EMEA/CHMP/370798, 18 October 2006, London. (2008a). Biomarkers for vinyl chloride-induced hepatic hemangiosarcoma. European Medicines Agency (EMEA). (2007). European Public Assessment Hemangiosarcoma in Rodents: Mode-of-Action Evaluation and Human Report (EPAR) Galvus. EMEA/H/C/771, London. Relevance, Society of Toxicology Contemporary Concepts in Toxicology Ferrara, N. (2002). VEGF and the quest for tumour angiogenesis factors. Nat. Workshop, December 4-5, 2008 (Abstract). Available at: http://www. Rev. Cancer 2, 795–803.

toxicology.org/ai/meet/SOT-CCT08_WkshpMaterials.pdf Ferrara, N., and Kerbel, R. S. (2005). Angiogenesis as a therapeutic target. Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 Cave, M., Falkner, K., Patel, M., Reynolds, L., Barve, S., and McClain, C. Nature 15, 967–974. (2008b). Vinyl chloride-induced hepatic hemangiosarcoma: The Louisville Firooz, A., Komeili, A., and Dowlati, Y. (1999). Eruptive melanocytic nevi and experience. Hemangiosarcoma in Rodents: Mode-of-Action Evaluation and cherry angiomas secondary to exposure to sulfur mustard gas. J. Am. Acad. Human Relevance, Society of Toxicology Contemporary Concepts in Dermatol. 40, 646–647. Toxicology Workshop, December 4-5, 2008 (Abstract). Available at: Foellmi-Adams, L. A., Wyse, B. M., Herron, D., Nedergaard, J., and http://www.toxicology.org/ai/meet/SOT-CCT08_WkshpMaterials.pdf Kletzien, R. F. (1996). Induction of uncoupling protein in brown adipose Cohen, A. D., Cagnano, E., and Vardy, D. A. (2001). Cherry angiomas tissue. Synergy between norepinephrine and pioglitazone, an insulin- associated with exposure to bromides. Dermatology 202, 52–53. sensitizing agent. Biochem. Pharmacol. 52, 693–701. Cohen, S. M., and Ellwein, L. B. (1991). Genetic errors, cell proliferation, and Foellmi-Adams, L. A., Wyse, B. M., Herron, D., Nedergaard, J., and carcinogenesis. Cancer Res. 51, 6493–6505. Kletzien, R. F. (2004). Congenital disorder of oxygen sensing: Association Corthals, S. M., Kamendulis, L. M., and Klaunig, J. E. (2006). Mechanisms of of the homozygous Chuvash polycythemia VHL mutation with thrombosis 2-butoxyethanol-induced HS. Toxicol. Sci. 92, 378–386. and vascular abnormalities but not tumors. Blood 103, 3924–3932. Criswell, K. A. (2008). Epigenetic mode of action associated with induction of Fukumura, D., Ushiyama, A., Duda, D. G., Xu, L., Tam, J., Krishna, V., hemangiosarcoma in mice treated with pregabalin. Hemangiosarcoma in Chatterjee, K., Garkavtsev, I., and Jain, R. K. (2003). Paracrine regulation of Rodents: Mode-of-Action Evaluation and Human Relevance, Society of angiogenesis and adipocyte differentiation during in vivo adipogenesis. Circ. Toxicology Contemporary Concepts in Toxicology Workshop. December 4- Res. 93, 88–97. 5, 2008 (Abstract). Available at: http://www.toxicology.org/ai/meet/SOT- Goodman, D. (1991). Expert Panel to Review the Safety and Toxicity of 4- CCT08_WkshpMaterials.pdf Hydroxyphenyl Retinamide. D. Goodman, Chairperson. Summary Con- Cromwell, J. A., and Goldenthal, E. I. (2006). Hemangiosarcomas induced in clusions. National Institutes of Health, Bethesda, MD. September 5, 1991. mice by N-(4-hydrophenyl)retinamide. The Toxicologist 90, 428 (Abstract Gordeuk, V. R., Sergueeva, A. I., Miasnikova, G. Y., Okhotin, D., #2088). Voloshin, Y., Choyke, P. L., Butman, J. A., Jedlickova, K., Prchal, J. T., Crossno, J. T., Jr., Majka, S. M., Grazia, T., Gill, R. G., and Klemm, D. J. and Polyakova, L. A. (2004). Congenital disorder of oxygen sensing: (2006). Rosiglitazone promotes development of a novel adipocyte Association of the homozygous Chuvash polycythemia VHL mutation with population from bone marrow-derived circulating progenitor cells. J. Clin. thrombosis and vascular abnormalities but not tumors. Blood 103, Invest. 116, 3220–3228. 3924–3932. DeFelipe, I., and Redondo, P. (1998). Eruptive angiomas after treatment with Grommes, C., Landreth, G. E., and Heneka, M. T. (2004). Antineoplastic cyclosporine in a patient with psoriasis. Arch. Dermatol. 134, 1487–1488. effects of peroxisome proliferator-activated receptor c agonists. Lancet Dickerson, E. B., Thomas, R., Fosmire, S. P., Lamerato-Kozicki, A. R., Oncol. 5, 419–429. Bianco, S. R., Wojcieszyn, J. W., Breen, M., Helfand, S. C., and Hardisty, J. F., Elwell, M. R., Ernst, H., Greaves, P., Kolenda-Roberts, H., Modiano, J. F. (2005). Mutations of phosphatase and tensin homolog Malarkey, D. E., Mann, P. C., and Tellier, P. A. (2007). Histopathology of deleted from chromosome 10 in canine HS. Vet. Pathol. 42, 618–632. hemangiosarcomas in mice and hamsters and liposarcomas/fibrosarcomas in Ding, T. T., Kumar, S., and Yu, D. C. (2008). The role of endothelial rats associated with PPAR agonists. Toxicol. Pathol. 35, 928–941. progenitor cells in tumour vasculogenesis. Pathology 75, 265–273. Harris, A. L. (2002). Hypoxia—A key regulatory factor in tumour growth. Nat. Duddy, S. K., Gorospe, S. M., Bleavins, M. R., and de la Iglesia, F. A. (1999b). Rev. Cancer 2, 38–47. Spontaneous and thiazolidinedione-induced B6C3F1 mouse hemangiosarco- Haseman, J. K., Hailey, J. R., and Morris, R. W. (1998). Spontaneous neoplasm mas exhibit low ras oncogene mutation frequencies. Toxicol. Appl. incidences in Fischer 344 rats and B6C3F1 mice in two-year carcinogenicity Pharmacol. 160, 133–140. studies: A National Toxicology Program update. Toxicol. Pathol. 26, Duddy, S. K., Parker, R. F., Bleavins, M. R., Gough, A. W., Rowse, P. E., 428–441. Gorospe, S., Dethloff, L. A., and de la Iglesia, F. A. (1999a). P53 is not Hefazi, M., Maleki, M., Mahmoudi, M., Tabatabaee, A., and Balali-Mood, M. inactivated in B6C3F1 mouse vascular tumors arising spontaneously or (2006). Delayed complications of sulfur mustard poisoning in the skin and associated with long-term administration of the thiazolidinedione troglita- the immune system of Iranian veterans 16-20 years after exposure. Int. J. zone. Toxicol. Appl. Pharmacol. 156, 106–122. Dermatol. 45, 1025–1031. Elwell,M.R.,Mahler,J.F.,andRuecker,F.A.(2004).Proliferative Herman, J. R., Dethloff, L. A., McGuire, E. J., Parker, R. F., Walsh, K. M., and non-proliferative lesions in the heart and vasculature in Gough, A. W., Masuda, H., and de la Iglesia, F. A. (2002). Rodent mice, MCV-1. In Guides for Toxicologic Pathology. STP/ARP/AFIP, carcinogenicity with the thiazolidinedione antidiabetic agent troglitazone. Washington, DC. Toxicol. Sci. 68, 226–236. Emadi, S. N., Hosseini-Khalili, A., Soroush, M. R., Davoodi, S. M., and Hickey, M. M., Lam, J. C., Bezman, N. A., Rathmell, W. K., and Simon, M. C. Aghamiri, S. S. (2008). Mustard gas scarring with specific pigmentary, (2007). Von Hippel-Lindau mutation in mice recapitulates Chuvash RODENT HEMANGIOSARCOMA: MOA/HUMAN RELEVANCE 17

polycythemia via hypoxia-inducible factor-2a signaling and splenic (2003). A framework for human relevance analysis of information on erythropoiesis. J. Clin. Invest. 117, 3879–3889. carcinogenic modes of action. Crit. Rev. Toxicol. 33, 591–653. Hutley, L. J., Herington, A. C., Shurety, W., Cheung, C., Vesey, D. A., Morinello, E., Koe, H., Ranasinghe, A., and Swenberg, J. (2002). Differential Cameron, D. P., and Prins, J. B. (2001). Human adipose tissue endothelial induction of N2,3-ethenoguanine in rat brain and liver after exposure to vinyl cells promote preadipocyte proliferation. Am. J. Endocrinol. Metab. 281, chloride. Cancer Res. 62, 5183–5188. E1037–E1044. Moyer, C., Allen, D., Basabe, A., Maronpot, R. R., and Nyska, A. (2004). Igarashi, A., Hayashi, N., Nashiro, K., and Takehara, K. (1998). Differential Analysis of vascular endothelial growth factor (VEGF) and a receptor subtype expression of connective tissue growth factor gene in cutaneous fibrohis- (KDR/flk-1) in the liver of rats exposed to riddelliine: A potential role in the tiocytic and vascular tumors. J. Cutaneous Pathol. 25, 143–148. development of hemangiosarcomas. Exp. Toxicol. Pathol. 55, 455–465. Jacobs, A. C. (2008). Hemangiosarcoma and pharmaceuticals: An FDA Murata, T., Hata, Y., Ishibashi, T., Kim, S., Hsueh, W. A., Law, R. E., and perspective. Hemangiosarcoma in Rodents: Mode-of-Action Evaluation and Hinton, D. R. (2001). Response of experimental retinal neovascularization to Human Relevance, Society of Toxicology Contemporary Concepts in thiazolidinediones. Arch. Ophthalmol. 119, 709–717.

Toxicology Workshop. December 4–5, 2008 (Abstract). Available at: Nyberg, P., Xie, L., and Kalluri, R. (2005). Endogenous inhibitors of Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 http://www.toxicology.org/ai/meet/SOT-CCT08_WkshpMaterials.pdf angiogenesis. Cancer Res. 65, 3967–3979. Jinnin, M., Medici, D., Park, L., Limaye, N., Liu, Y., Boscolo, E., Bischoff, J., Nyska, A., Haseman, J. K., Kohen, R., and Maronpto, R. R. (2004). Association of Vikkula, M., Boye, E., and Olsen, B. R. (2008). Suppressed NFAT- liver hemangiosarcoma and secondary iron overload in B6C3F1 mice. The dependent VEGFR1 expression and constitutive VEGFR2 signaling in National Toxicology Program experience. Toxicol. Pathol 32, 222–228. infantile hemangioma. Nat. Med. 14, 1236–1246. Nyska, A., Moomaw, C. R., Foley, J. F., Maronpot, R. R., Malarkey, D. E., Johnson, T. E. (2008). Retinoid-induced hemangiosarcoma in mice: potential Cummings, C. A., Peddada, S., Moyer, C. F., Allen, D. G., Travlos, G., and insight from in vivo and in vitro studies. Hemangiosarcoma in rodents: Chan, P. C. (2002). The hepatic endothelial carcinogen riddelliine induces Mode-of-Action Evaluation and Human Relevance, Society of Toxicology endothelial apoptosis, mitosis, S phase, and p53 and hepatocytic vascular Contemporary Concepts in Toxicology Workshop, December 4-5, 2008 endothelial growth factor expression after short-term exposure. Toxicol. (Abstract). Available at: http://www.toxicology.org/ai/meet/SOT-CCT08_ Appl. Pharmacol. 184, 153–164. WkshpMaterials.pdf Ohnishi, T., Arnold, L. L., Clark, N. M., Wisecarver, J. L., and Cohen, S. M. Kakiuchi-Kiyota, S., Vetro, J. A., Suzuki, S., Varney, M. L., Han, H., (2007). Comparison of endothelial cell proliferation in normal liver and Nascimento, M., Pennington, K. L., Arnold, L. L., Singh, R. K., and adipose tissue in B6C3F1 mice, F344 rats, and humans. Toxicol. Pathol. 35, Cohen, S. M. (2009). Effect of the PPARc agonist troglitazone on 904–909. endothelial cells in vivo and in vitro: Differences between human and Okuno, A., Tamemoto, H., Tobe, K., Ueki, K., Mori, Y., Iwamoto, K., mouse. Toxicol. Appl. Pharmacol. 237, 83–90. Umesono, K., Akanuma, Y., Fujiwara, T., Horikoshi, H., et al. (1998). Kerbel, R., and Folkman, J. (2002). Clinical translation of angiogenesis Troglitazone increases the number of small adipocytes without the change of inhibitors. Nat. Rev. Cancer 2, 727–739. white adipose tissue mass in obese Zucker rats. J. Clin. Invest. 101, 1354–1361. Kershaw, E. E., and Flier, J. S. (2004). Adipose tissue as an endocrine organ. Park, J., Kamendulis, L. M., and Klaunig, J. E. (2002). Effects of 2- J. Clin. Endocrinol. Metab. 89, 2548–2556. butoxyethanol on hepatic oxidative damage. Toxicol. Lett. 126, 19–29. Kielhorn, J., Melber, C., Wahnschaffe, U., Aitio, A., and Mangelsdorf, I. Pegg, D. G., Wojcinski, Z., Criswell, K., Herman, J., and Anderson, T. (2006). (2000). Vinyl chloride: Still a cause for concern. Environ. Health Perspect. Epigenetic mode of action associated with induction of hemangiosarcoma in 108, 579–588. mice. The Toxicologist 90, 334. Klaunig, J. E., and Kamendulis, L. M. (2005). Mode of action of Pezzolesi, M. G., Platzer, P., Waite, K. A., and Eng, C. (2008). Differential butoxyethanol-induced mouse liver HS and hepatocellular carcinomas. expression of PTEN-targeting microRNAs miR-19a and miR-21 in Cowden Toxicol. Lett. 156, 107–115. syndrome. Am. J. Hum. Genet. 82, 1141–1149. Kojiro, M., Nakashima, T., Ito, Y., Ikezaki, H., Mori, T., and Kido, C. (1985). Physician’s Desk Reference Electronic Library. (2009). Thomson Micromedex. Thorium dioxide-related angiosarcoma of the liver. Pathomorphologic study Available at: www.thomsonhc.com. Accessed February 2, 2009. of 29 autopsy cases. Arch. Pathol. Lab. Med. 109, 853–857. Priester, W. A., and McKay, F. W. (1980). The occurrence of tumors in Laifenfeld,D.,Gilchrist,A.,Drubin,D.,Jorge,M.,Eddy,S.F.,Elliston,K.O., domestic animals. Natl. Cancer Inst. Monogr. 54, 1–210. Obert, L. A., Sokowloski, S. A., Gosink, M. M., Cook, J. C., et al. (2009). The role Raymond, L. W., Williford, L. S., and Burke, W. A. (1998). Eruptive cherry of hypoxia in 2-butoxyethanol-induced hemangiosarcoma. Toxicol. Sci. (in press). angiomas and irritant symptoms after one acute exposure to the glycol ether Lamerato-Kozicki, A. R., Helm, K. M., Jubala, C. M., Cutter, G. C., and solvent 2-butoxyethanol. J. Occup. Environ. Med. 40, 1059–1064. Modiano, J. F. (2006). Canine hemangiosarcoma originates from hemato- Requena, L., and Sangueza, O. P. (1997). Cutaneous vascular proliferations. poietic precursors with potential for endothelial differentiation. Exp. Part II. Hyperplasias and benign neoplasms. J. Am. Acad. Dermatol. 37, Hematol. 34, 870–878. 887–922. Lanza, F., Healy, L., and Sutherland, D. R. (2001). Structural and functional features Rogers, M. S., and D’Amato, R. J. (2006). The effect of genetic diversity on of the CD34 antigen: An update. J. Biol. Regul. Homeost. Agents 15, 1–13. angiogenesis. Exp. Cell Res. 312, 561–574. Lazennec, G., and Jorgensen, C. (2008). Concise review: adult multipotent Rogers, M. S., Rohan, R. M., Birsner, A. E., and D’Amato, R. J. (2003). stromal cells and cancer: Risk or benefit? Stem Cells 26, 1387–1394. Genetic loci that control vascular endothelial growth factor-induced Ma, H. J., Zhao, G., Shi, F., and Wang, Y. X. (2006). Eruptive cherry angiomas angiogenesis. FASEB J. 17, 2112–2114. associated with vitiligo: Provoked by topical nitrogen mustard? J. Dermatol. Rogers, M. S., Rohan, R. M., Birsner, A. E., and D’Amato, R. J. (2004). 33, 877–879. Genetic loci that control the angiogenic response to basic fibroblast growth Maehama, T., Taylor, G., and Dixon, J. (2001). PTEN and myotubularin: factor. FASEB J. 18, 1050–1059. Novel phosphoinositide phosphatases. Annu. Rev. Biochem. 70, 247–259. Rosner, M., Hanneder, M., Siegel, N., Valli, A., Fuchs, C., and Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman- Hengstschlager, M. (2008). The mTOR pathway and its role in human McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. genetic diseases. Mutat. Res. 659, 284–292. 18 COHEN ET AL.

Ruben, Z., Arceo, R. J., Bishop, S. P., Elwell, M. R., Kerns, W. D., Morris, D. C., and Graves, R. A. (1996). Activation of the nuclear receptor Mesfin, G. M., Sandusky, G. E., and Van Vleet, J. F. (1997). Proliferative peroxisome proliferator-activated receptor gamma promotes brown adipo- lesions of the heart and vasculature in rats, CD-1. In Guides for Toxicologic cyte differentiation. J. Biol. Chem. 271, 29909–29914. Pathology. STP/ARP/AFIP, Washington, DC. Tannehill-Gregg, S. H., Sanderson, T. P., Minnema, D., Voelker, R., Rupnick, M. A., Panigrahy, D., Zhang, C. Y., Dallabrida, S. M., Lowell, B. B., Ulland, B., Cohen, S. M., Arnold, L. L., Schilling, B. E., Waites, C. R., Langer, R., and Folkman, M. J. (2002). Adipose tissue mass can be regulated and Dominick, M. A. (2007). Rodent carcinogenicity profile of the through the vasculature. Proc. Natl. Acad. Sci. U. S. A. 99, 10730–10735. antidiabetic dual PPAR c and a agonist muraglitazar. Toxicol. Sci. 98, Sears, I. B., MacGinnitie, M. A., Kovacs, L. G., and Graves, R. A. (1996). 258–270. Differentiation-dependent expression of the brown adipocyte uncoupling U.S. Environmental Protection Agency (US EPA). (2004). EPA Interim protein gene: Regulation by peroxisome proliferator-activated receptor gamma. Mol. Cell. Biol. 16, 3410–3419. Registration Eligibility Decision (IRED), Carbaryl. Washington, DC: US Environmental Protection Agency, Office of Prevention, Pesticide, and Toxic Seed, J., Carney, E. W., Corley, R. A., Crofton, K. M., DeSesso, J. M., Substances, Case 0080.

Foster, P. M., Kavlock, R., Kimmel, G., Klaunig, J., Meek, M. E., et al. Downloaded from https://academic.oup.com/toxsci/article/111/1/4/1671803 by guest on 28 September 2021 (2005). Using mode of action and life stage information to evaluate the Van Oosterhout, J. P., Van der Laan, J. W., De Waal, E. J., Olejniczak, K., human relevance of animal toxicity data. Crit. Rev. Toxicol. 35, 663–672. Hilgenfeld, M., Schmidt, V., and Bass, R. (1997). The utility of two rodent Shaked, Y., Bertolini, F., Man, S., Rogers, M. S., Cervi, D., Foutz, T., species in carcinogenic risk assessment of pharmaceuticals in Europe. Regul. Rawn, K., Voskas, D., Dumont, D. J., Ben-David, Y., et al. (2005). Genetic Toxicol. Pharmacol. 25, 6–17. heterogeneity of the vasculogenic phenotype parallels angiogenesis: Virtanen, A., Pukkala, E., and Auvinen, A. (2007). Angiosarcoma after Implications for cellular surrogate marker analysis of antiangiogenesis. radiotherapy: A cohort study of 332,163 Finnish cancer patients. Br. Cancer Cell 7, 101–111. J. Cancer 97, 115–117. Sierra-Honigmann, M. R., Nath, A. K., Murakami, C., Garcı´a-Carden˜a, G., Waites, C. R., Dominick, M. A., Sanderson, T. P., and Schilling, B. E. (2007). Papapetropoulos, A., Sessa, W. C., Madge, L. A., Schechner, J. S., Nonclinical safety evaluation of muraglitazar, a novel PPAR a/c agonist. Schwabb, M. B., Polverini, P. J., and Flores-Riveros, J. R. (1998). Biological Toxicol. Sci. 100, 248–258. action of leptin as an angiogenic factor. Science 281, 1683–1686. Weiss, S. W., and Goldblum, J. R. (2008a). Benign tumors and tumor-like Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner- lesions of blood vessels. In Enzinger and Weiss’s Soft Tissue Tumors, 5th ed. Crisp,P., Grant,D.,Hartley,M., Knaap, A., Kroese, D., et al. (2001).IPCS-Mode (S. W. Weis and J. R. Goldblum, Eds.), pp. 633–679. Mosley Elsevier, of action of chemical carcinogenesis. Regul. Toxicol. Pharmacol. 34, 146–152. China. Stienstra, R., Duval, C., Keshtkar, S., van der Laak, J., Kersten, S., and Mu¨ller, M. (2008). Peroxisome proliferator-activated receptor c activation Weiss, S. W., and Goldblum, J. R. (2008b). Malignant vascular tumors. In promotes infiltration of alternatively activated macrophages into adipose Enzinger and Weiss’s Soft Tissue Tumors, 5th ed. (S. W. Weis and tissue. J. Biol. Chem. 283, 22620–22627. J. R. Goldblum, Eds.), pp. 703–732. Mosley Elsevier, China. Tafuri, S. R. (1996). Troglitazone enhances differentiation, basal glucose Werner, C., Kamani, C. H., Gensch, C., Bo¨hm, M., and Laufs, U. (2007). The uptake, and Glu1 protein levels in 3T3-L1 adipocytes. Endocrinology 137, peroxisome proliferator-activated receptor-c agonist pioglitazone increases 4706–4712. number and function of endothelial progenitor cells in patients with Tai, T. A., Jennermann, C., Brown, K. K., Oliver, B. B., MacGinnitie, M. A., coronary artery disease and normal glucose tolerance. Diabetes 56, Wilkison, W. O., Brown, H. R., Lehmann, J. M., Kliewer, S. A., 2609–2615.