Investigations of Clofibrate in Alternative Carcinogenicity Models

Genotoxic chemicals can either be metabolically activated to electrophilic derivates which interact with DNA and other macromolecules, or directly bind to DNA. In contrast, nongenotoxic carcinogens induce cell transformation indirectly. Peroxisome proliferators are one of the most widely studied of the nongenotoxic carcinogens and have diverse industrial and therapeutic uses (Gonzalez, Peters, and Cattley 1998). Clofibrate is a nongenotoxic, peroxisome proliferator-activated receptor (PPAR) α agonist (NDA 1993), a member of a diverse class of fibric acid derivatives that are used therapeutically to treat dyslipidemia, coronary heart disease, diabetes, leukemia, proliferative inflammatory diseases, and weight loss (Yki-Jarvinen 2004). Clofibrate is considered a weak α agonist, with an EC₅₀ in transient transfection assays of approximately 50 μM (Henke et al. 1998). In humans taking clofibrate at the recommended maximum therapeutic dose of 500 mg, systemic exposure measured by area under the concentration curve (AUC) is approximately 1100 μg h/ml (DeSante et al. 1979). The clofibrate:PPARα complex heterodimerizes with the retinoid X receptor (RXR) and serves as a transactivating factor on DNA (Yki-Jarvinen 2004). PPARα ligands cause peroxisomal proliferation and an increase in liver size in rodents (Klaunig et al. 2003). Chronic administration of PPARα ligands, such as fenofibrate and clofibrate, typically lead to hepatocarcinogenesis in rodents (Kluwe et al. 1982; NDA 1993; Reddy, Rao, and Moody 1976). However, humans and nonhuman primates are considered resistant to peroxisomal proliferation and the developmentof liver tumors after exposure to PPARα ligands (Klaunig et al. 2003; Hoivik et al. 2004). Based on its mechanism of action and extensive human experience, clofibrate was selected for inclusion in the International Life Sciences Institute (ILSI) program in alternative carcinogenicity models (Robinson and MacDonald 2001). The purpose of the ILSI program in alternative carcinogenicity model testing was to evaluate the predictivity of several models to assess their utility as a replacement for the mouse study in the typical battery of carcinogenicity assessments (2-year rat and mouse studies). Compounds from diverse modes of action were selected to rigorously test the models and included immunosuppressive, genotoxic, enzyme-inducing, and peroxisome proliferator agents. An important component of the ILSI efforts was the cross-site validation of the studies. Chemicals were evaluated at multiple sites under similar study designs that were approved by a core committee. Factors controlled included lot number and source of test material, source of animals, and dose selection. GlaxoSmithKline, a participant in this effort, tested clofibrate in multiple alternative carcinogenicity models and this document provides an overall summary of the findings. Moreover, environmental, vehicle, and nontransgenic controls, as well as positive tumorigenic chemicals, were also evaluated in the alternative models.

These alternative carcinogenicity models tested included the rasH2 transgenic mouse model, the Tg.AC transgenic mouse model, the p53^{+ / −} knockout mouse model, the XPA^{− / −}knockout mouse model, the XPA^{− / −}/p53^{+ / −} double-knockout mouse model, and the neonatal mouse model (Robinson and MacDonald 2001). One major incentive for the use of these models in carcinogenicity studies is the shorter time to tumor formation, typically 6 months for p53^{+ / −}, rasH2, and Tg.AC mice; 9 months for XPA^{− / −} mice; and 1 year for the neonatal mouse. This yields obvious advantages in reduced resources and provides earlier results.

The p53^{+ / −} knockout mouse was genetically engineered to have one allele of the p53 tumor suppressor gene deleted to mimic some heritable forms of cancer in humans. The p53 tumor suppressor gene is critical for cell cycle control and DNA repair and suppresses DNA damage by inhibiting progression through the cell cycle and initiating apoptosis in response to damage (Tennant, French, and Spalding 1995; Tennant, Spalding, and French 1996). This gene is often found to be mutated or absent in human and rodent tumors (Donehower et al. 1992). The p53^{+ / −} knockout mouse is the model of choice for alternative carcinogenicity study protocols due to its accelerated tumorgenesis and generally is accepted for use in evaluating the carcinogenic potential of genotoxic agents (French, Storer, and Donehower 2001; Robinson and MacDonald 2001). The potential carcinogenicity of clofibrate was investigated in the p53^{+ / −}model following oral administration for a minimum of 26 weeks (Torrey et al. 2005a). Initial clofibrate doses were 50, 250, and 400 mg/kg/day for males and 50, 200, and 500 mg/kg/day for females. Doses were lowered to 25, 75, and 100 mg/kg/day for males and 25, 75, and 125 mg/kg/day for females due to unexpected mortality during the first week of dosing. No neoplastic response was observed in p53^{+ / −} mice after 6 months of exposure to clofibrate, at doses up to 100 mg/kg/day for males and 125 mg/kg/day for females (Table 1) as expected for nongenotoxic carcinogens (Storer et al. 2001), including in the liver. Clofibrate induced non-neoplastic findings in the adrenal gland, pancreas, and prostate.

The Tg.AC transgenic mouse alternative carcinogenicity model was created by fusing the v-Ha-ras gene into the FVB/N mouse strain. Genetic lesions in the Ha-ras gene are common occurrences contributing to human cancers (Tamaoki 2001). The transgene is transcriptionally dormant until activated by specific chemicals (Eastin et al. 2001), UV light (Trempus et al. 1998), or trauma (Cannon et al. 1997). This Tg.AC transgenic mouse model generates epidermal or forestomach squamous-cell papillomas/carcinomas in response to carcinogens after topical application or oral administration, respectively (Eastin et al. 2001). Moreover, the hemizygous Tg.AC mouse exhibits a higher average tumor burden and shorter latency period to tumor multiplicity compared to the homozygous strain. These data suggest that a rapid onset and a higher incidence of malignant tumors can be expected after treatment with various carcinogens in Tg.AC mice due to increased oncogene or tumor suppressor gene expression. This model detects both genotoxic and nongenotoxic carcinogens (Tennant et al. 2001). The potential carcinogenicity of clofibrate was investigated after oral (Torrey et al. 2005b) or dermal (Torrey et al. 2005c) administration in this model for a minimum for 6 months. Clofibrate was well tolerated at doses up to 500 (males) and 650 (females) mg/kg/day and up to 36 mg/200 μl after oral and dermal administration, respectively. Clofibrate is not carcinogenic (Table 1) when administered to Tg.AC mice by oral gavage for 6 months at doses up to 500 (males) and 650 (females) mg/kg/day, which did produce liver hypertrophy. In contrast, clofibrate produced a dose-related increase in the incidence of mice with cutaneous papillomas; and dose-related decreases in mean time to first tumor, mean multiplicity of tumors per mouse, and mean weeks to maximal yield, as well as numerous nonneoplastic microscopic lesions in the liver, kidney, spleen, and skin after dermal application (Table 1). After oral administration of 500 mg/kg/day, epithelial hyperplasia in the urinary bladder and prostate gland, and interstitial-cell hyperplasia in the testes were noted. The difference between carcinogenic potential of clofibrate between the dermal and oral studies may be due to a slightly lower systemic exposure (approximately 1.3×) noted in the oral study or insensitivity of the oral model to carcinogenesis (Tennant et al. 2001). Non-neoplastic nonproliferative findings after oral administration included: hepatic hypertrophy and hematopoietic (myeloid hyperplasia, myelodysplasia, lymphoid depletion, and erythropoiesis) in both sexes; reproductive (cystic degeneration and dilatation, hypospermia, spermatocele, dilated inspissated protein) and urogenital (tubular-cell hypertrophy, degenerative/regenerative nephropathy, necrosis/fibrosis) changes in male mice; congestion in the lung in male mice; gallbladder dilatation in female mice; and adrenal (intracellular lipofuscinosis and atrophy) and heart (eosinophillic myofibers) findings in mice of both sexes. After dermal application, nonneoplastic findings associated with clofibrate included hepatocellular hypertrophy, renal proximal tubular epithelial hypertrophy, increased splenic hematopoiesis, and acanthosis at the skin application site.

The rasH2 mouse is a hemizygous transgenic alternative carcinogenicity model carrying the human c-Ha-ras gene (five or six copies) within the promoter region which encodes the p21 protein, a key member of the signal transduction pathway for cell proliferation, differentiation, and death (Tamaoki 2001; Yamamoto et al. 1997; Yamamoto, Urano, and Nomura 1998). The transgene is expressed in the tumors and in normal tissues in man, but not in normal mouse tissues, with an increased level of p21 protein in transgenic mice compared with nontransgenic mice. Point mutations in the Ha-ras gene are common occurrences contributing to human cancers (Tamaoki 2001). Data suggest that a rapid onset and higher incidence of malignant tumors can be expected after treatment with various genotoxic carcinogens in rasH2 mice due to increased p21 gene expression (Qin et al. 1999; Tamaoki 2001). The carcinogenic potential of clofibrate was investigated in this model following oral administration at 50, 100, or 200 (males) and 50, 150, or 250 mg/kg/day (females) for 6 months (Nesfield et al. 2005a). Hepatocellular neoplasms were observed in clofibrate-treated rasH2 male mice after 6 months of treatment (Table 1). Clofibrate treatment (250 mg/kg/day) of female rasH2 mice also caused a slight increase in the incidence of neoplasms in the harderian gland, lungs, skin, spleen, tail, thymus, and uterus (Table 1). These neoplastic findings may be related to an increased incidence of peroxisome proliferation and/or other PPARα-related effects such as inhibition of apoptosis, oxidative stress, alteration of cell signaling molecules or gap junction communications (Klaunig et al. 2003). Non-neoplastic changes were found in the liver of transgenic mice of both sexes and in the kidneys of male mice.

The neonatal mouse assay for tumorigenicity has been used since 1959 (Flammang et al. 1997). Neonatal mice have a decreased time-to-tumor formation due to their developmental stage at the time of dosing (Fujii 1991) and typically require fewer doses than adult mice to cause tumors. These differences in susceptibility between adult and neonatal mice may be related to the peaks in DNA synthesis seen in young mice (Flammang et al. 1997). The neonatal mouse model offers a high sensitivity for genotoxic chemicals inducing tumors in a short period (McClain et al. 2001). The carcinogenic potential of clofibrate was studied following oral administration to neonatal mice at doses of 100, 250, and 500 mg/kg on days 9 and 16 post birth and observed for approximately 1 year for the development of tumors (Nesfield et al. 2005b). Clofibrate, at doses of up to 500 mg/kg did not produce a carcinogenic effect (Table 1) as expected for nongenotoxic carcinogens. Clofibrate-related non-neoplastic lesions (retinal atrophy and myocardial degeneration/fibrosis) were seen only in males at 500 mg/kg.

Theses data indicate that oral administration of the nongenotoxic rodent carcinogen clofibrate is hepatocarcinogenic in rasH2 mice after 6 months of exposure (Nesfield et al. 2005a) and produced papillomas in Tg.AC mice after dermal application (Torrey et al. 2005c). However, clofibrate is noncarcinogenic in p53^{+ / −} mice after 6 months of exposure (Torrey et al. 2005a) or in neonatal mice up to 1-year after treatment on litter days 9 and 16 (Nesfield et al. 2005b) or in Tg.AC mice after 6 months of oral exposure (Torrey et al. 2005b). Recently published data suggest that structural differences between the human and rodent PPARα receptor and/or tissue/species-specific coactivators/corepressors may be responsible for the differential response noted amongst mouse strains/models (Cheung et al. 2004). In addition, studies have shown species-related susceptibilityof inbred mice to carcinogenesis (Harris and Lawley 1994). Collectively, these data support the concept that alternative carcinogenicity models detect compounds with nongenotoxic carcinogenic potential in a shorter timeframe than conventional mouse carcinogenicity bioassays. In addition, these data show that the at least one model, the Tg.AC dermal model, has increased sensitivity in detecting tumors (skin papillomas) in rodents caused by the nongenotoxic rodent-specific carcinogen, clofibrate, at systemic exposures slightly higher (3×) than the systemic exposure observed in humans taking clofibrate at the recommended maximum therapeutic dose of 500 mg. Interestingly, the response to clofibrate was not consistent across the models, and the reasons for this are possibly due, in part, to differences in exposure and strain-related differences in susceptibility to carcinogenesis. The reasons for the varied responses warrant further investigation.