Urothelial Carcinogenesis in the Urinary Bladder of Male Rats Treated with Muraglitazar,a PPARα/γ Agonist: Evidence for Urolithiasis as the Inciting Event in the Mode of Action

Abstract

Muraglitazar, a PPARα/γ agonist, dose-dependently increased urinary bladder tumors in male Harlan Sprague–Dawley (HSD) rats administered 5, 30, or 50 mg/kg/day for up to 2 years. To determine the mode of tumor development, male HSD rats were treated daily for up to 21 months at doses of 0, 1, or 50 mg/kg while being fed either a normal or 1% NH₄Cl-acidified diet. Muraglitazar-associated, time-dependent changes in urine composition, urothelial mitogenesis and apoptosis, and urothelial morphology were assessed. In control and treated rats fed a normal diet, urine pH was generally ≥6.5, which facilitates formation of calcium- and magnesium-containing solids, particularly in the presence of other prolithogenic changes in rat urine. Urinary citrate, an inhibitor of lithogenesis, and soluble calcium concentrations were dose dependently decreased in association with increased calcium phosphate precipitate, crystals and/or microcalculi; magnesium ammonium phosphate crystals and aggregates; and calcium oxalate-containing thin, rod-like crystals. Morphologically, sustained urothelial cytotoxicity and proliferation with a ventral bladder predilection were noted in treated rats by month 1 and urinary carcinomas with a similar distribution occurred by month 9. Urothelial apoptotic rates were unaffected by muraglitazar treatment or diet. In muraglitazar-treated rats fed an acidified diet, urine pH was invariably < 6.5, which inhibited formation of calcium- and magnesium-containing solids. Moreover, dietary acidification prevented the urothelial cytotoxic, proliferative, and tumorigenic responses. Collectively, these data support an indirect pharmacologic mode of urinary bladder tumor development involving alterations in urine composition that predispose to urolithiasis and associated decreases in urine-soluble calcium concentrations.

Introduction

Peroxisome proliferators-activated receptors (PPARs) represent a subfamily of nuclear hormone receptors containing 3 isoforms: PPARα, PPARγ, and PPARδ(β) (Dreyer et al., 1992) and are ligand-dependent transcription factors that regulate target gene expression by binding to specific peroxisome proliferator response elements (PPREs) in promoter regions of regulated genes (Kersten et al., 2000; Berger and Moller, 2002; Nahle, 2004). Each PPAR isoform heterodimerizes with the retinoid X receptor (RXR) prior to binding with a specific PPRE. Binding of an agonist with the PPAR-RXR heterodimer induces a conformational change that results in the recruitment of transcriptional co-activators and an increase in target gene expression (Berger and Moller, 2002).

PPARs are an important therapeutic target given their critical role in the regulation of lipid metabolism and insulin sensitization (Berger and Moller, 2002; Yki-Jarvinen, 2004; Lehrke and Lazar, 2005). The fibrate PPARα agonists are used clinically to treat dyslipidemia whereas the thiazolidinedione PPARγ agonists lower blood glucose levels and improve peripheral insulin sensitization, thereby providing an effective therapeutic option for the treatment of type 2 diabetes. There has been considerable interest in the development of dual PPARα/γ agonists given that dyslipidemia often accompanies hyperglycemia in type 2 diabetic patients. To date, no dual PPARα/γ agonists have been approved for the treatment of diabetes, in part, because of concern over the carcinogenic potential of these agents and the lack of mechanistic data exploring mode of action for the tumors observed at low systemic exposures.

Based on the FDA’s review of 2-year carcinogenicity studies in mice and rats for 6 PPARγ and 6 dual PPARα/γ agonists, the most commonly occurring tumor types with these agents occur in tissues with high PPAR expression and include hemangiosarcoma in mice, subcutaneous lipoma and/or liposarcoma in rats, and transitional cell tumors of the urinary bladder and/or renal pelvis in rats (El Hage, 2005). In contrast, the tumor triad commonly observed in male rats treated with fibrate PPARα agonists (hepatocellular, pancreatic acinar cell, and testicular Leydig cell tumors) (Klaunig et al., 2003) has not been observed with the dual PPARα/γ agonists. The limited evidence for liver tumor induction with dual PPARα/γ agonists is intriguing, since the mechanism of hepatocarcinogenesis involves PPARα-dependent pathways and many of the dual PPAR agonists are more potent at the α receptor than fibrates.

There has been considerable scientific discussion concerning the potential for a direct PPAR-mediated mechanism of urinary bladder carcinogenesis in rats treated with dual PPARα/γ agonists given the number of structurally diverse agonists associated with this tumor type and because PPARα and γ are expressed at high levels in normal urothelium of rodents (Jain et al., 1998; Eshler et al., 2001). However, most studies suggest PPARγ stimulates differentiation, inhibits growth, and/or induces apoptosis in normal and neoplastic urothelium (Nakashiro, et al., 2001; Guan, 2002; Kawakami et al., 2002; Possati et al., 2002;Yoshimura et al., 2003a, 2003b; Varley et al., 2004; Kassouf et al., 2006), although Fauconnet et al. (2002) have suggested PPARγ activation can potentiate urinary bladder carcinogenesis. Moreover, rats treated chronically with PPARα agonists do not develop transitional cell hyperplasia or neoplasia of the urinary bladder.

Since PPARγ agonists could potentially reduce levels of the anti-lithogenic organic ion citrate, via increased insulin sensitization and stimulation of lipogenesis (Madsen et al., 1964; Ball, 1966; Cusin et al., 1990), an alternative hypothesis would be that urinary bladder tumors develop, in part, as a consequence of pharmacologically mediated changes in urine composition that predispose the rat to urolithiasis and consequent urothelial injury and regenerative hyperplasia (Cohen, 2005).

Urolithiasis is a well established mode of urinary bladder tumorigenesis in rodents, particularly male rats (Clayson et al., 1995; Cohen, 1995, 1999; Cohen et al., 2002). The magnitude of the tumorigenic response is dependent on the form and burden of urinary solids with urothelial tumors reported via this mode for melamine (Melnick et al., 1984), uracil (Fukushima et al., 1992), silicates (Emerick et al., 1963; Okamura et al., 1992), sodium saccharin (Cohen et al., 1995a), sodium ascorbate (Cohen et al., 1995a, 1998), certain sulfonamides (Jackson et al., 1979), and carbonic anhydrase inhibitors (Jackson et al., 1979). The abundant calculi formed via treatment of rats with 3% melamine induce a marked proliferative response that leads to a 90% incidence of transitional cell carcinoma after treatment for 36 weeks (Ogasawara et al., 1995).

In contrast, the modest increases in endogenous calcium phosphate precipitate induced by treatment with sodium saccharin or sodium ascorbate resulted in relatively low incidences of transitional cell tumors, even after administration over two generations (Cohen et al., 1998). Importantly, formation of endogenous calcium and magnesium salts in rat urine is facilitated at a urine pH > 6.5; high urinary concentrations of protein, calcium, and phosphate; and high urinary osmolality (Cohen, 1995; Cohen and Lawson, 1995). The predisposition of male rats to develop urinary calcium phosphate precipitate and crystalluria when treated with certain drugs or chemicals has been ascribed, in part, to the higher levels of urine protein and the earlier onset of spontaneous nephropathy in male rats compared to female rats (Hard, 1995).

In the oral carcinogenicity study in rats with muraglitazar, a dose-related increased incidence of urinary bladder tumors, particularly transitional cell carcinomas, was observed in males given 5, 30, and 50 mg/kg, whereas only a low incidence of transitional cell hyperplasia was seen in females at these dose levels. At the 2 highest doses in males, there were dose-related increased incidences of calcium phosphate crystals and/or calculi and aggregates of magnesium ammonium phosphate (MgNH₄PO₄) crystals detected in sediment of fresh-void urine samples collected during week 90, providing preliminary evidence for urolithiasis as a potential mode of urinary bladder tumor development.

Because the urothelial tumorigenic response in the urinary bladder of male rats occurred at relatively low exposures, the mode of tumor development was fully investigated. Urinalyses were performed on fresh-void urine specimens collected at 4 time points during the day at 3-month intervals, since we anticipated marked variations in urine composition as a reflection of the normal diurnal variation that occurs due to the eating and drinking pattern of rats (Fisher et al., 1989; Cohen, 1995). The outcome of our investigations is summarized here and provides clear evidence for urolithiasis as the primary mode of urinary bladder tumor development in male rats treated with muraglitazar. It also suggests other factors, such as reduced urinary soluble calcium concentrations, may be contributory.

Materials And Methods

Animal Selection and Husbandry

Random-bred, barrier-raised, male Hsd:Sprague–Dawley SD (HSD) rats were obtained from Harlan Laboratories, Frederick, MD, and acclimated for approximately 2 weeks prior to the initiation of studies of 1, 3, and 21 months’ duration. The rats were approximately 6 to 7 weeks of age at study initiation and were individually housed in suspended stainless-steel wire-bottom cages in environmentally controlled rooms maintained on a 12-hour light-dark cycle and at a targeted humidity range of 30 to 70% and a targeted temperature range of 64 to 79°F. Water and certified rodent diet were provided ad libitum. The studies were conducted in full compliance with U.S. Food and Drug Administration Good Laboratory Practice regulations and in accordance with the Guide for the Care and Use of Laboratory Animals.

Experimental Design

Animals were randomly assigned to 6 groups (3 groups fed a normal diet and 3 groups fed an acidified diet) in each study. In the normal diet arm of the studies, daily doses of 1 or 50 mg/kg of muraglitazar were administered by oral gavage for up to 21 months to groups of rats fed HSD/Teklad Certified Global Diet #8728C, a standard rodent diet. In the acidified diet arm of the studies, muraglitazar was administered by oral gavage at the same doses and for the same duration to groups of rats fed HSD/Teklad Diet #8728C supplemented with 1% ammonium chloride (NH₄C1). In both arms of each study, comparable diet-matched control groups received the vehicle, 96% polyethylene glycol-400 and 4% 1 M NaOH, at 5 mL/kg. The muraglitazar doses of 1 and 50 mg/kg/day were previously shown to be noncarcinogenic and carcinogenic, respectively, in the urinary bladder urothelium of male HSD rats (Simutis et al., 2006).

Dose- and time-dependent effects were determined for urothelial morphology at months 1, 3, 6, 9, 12, 15, 18, and 21; for urothelial proliferation at months 1, 3, 6, 9, and 12 (normal and acidified arms) and at month 18 (acidified arm only); and for urothelial apoptosis at months 3, 6, 9, and 12. Urothelial cell proliferation and apoptosis were not assessed at months 15, 18, and 21 in animals fed a normal diet or at months 15 and 21 in animals fed an acidified diet since a pronounced urothelial tumorigenic response was apparent by month 15 in 50-mg/kg animals fed a normal diet and there was no hyperplastic or neoplastic effects in 50-mg/kg animals fed an acidified diet.

Urinalyses

Select urine chemistry parameters were measured in fresh-void urine collected from a minimum of 12 animals/group at approximately 2.5 and 16 hours after dosing during months 1, 6 (2.5 hours postdose only), 9, 12, 15, 18, and 21. Whenever possible, the animals/group designated for a specific necropsy were those for which urine samples were collected at the same interval. As the sample volume permitted, urine pH, creatinine, protein, soluble calcium, total calcium (acidified), phosphorous, soluble magnesium, total magnesium (acidified), sodium, potassium, and chloride were determined in the order presented. The rats had food and water available ad libitum, even during urine collection periods.

Additional urine samples were collected overnight (approximate 18-hour period) from 12 animals/group at months 1, 3, 6, 9, 12, and 15 for the determination of urine volume and creatinine, citrate, and oxalate levels. Urine was collected into a chilled 50 mL conical plastic tube and containing 100 μL of 1% sodium azide to inhibit bacterial growth. The rats were nonfasted and acclimated to the metabolism cages for approximately 48 hours prior to urine collection.

Urine Sediment Analyses

Fresh-void urine samples were collected from a minimum of 12 animals/group at approximately 2.5 and 16 hours post-dose during months 1, 6 (2.5 hours postdose only), 9, 12, and 15 (normal and acidified diets) and month 18 (acidified diet). Approximately 10 μL of each sample was wet mounted on a glass slide, coverslipped, and examined by light microscopy for evidence of crystalluria. In addition, urine sediment was obtained from fresh-void urine collected from at least 12 animals/group prior to dosing (within 2 hours of the start of the light phase) and at 2.5, 8, and 16 hours postdose during months 1, 6, 9, 12, and 15 (normal and acidified diets) and month 18 (acidified diet) for characterization of the morphology and composition of major inorganic components by scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (XDS).

At each sampling interval and time point, urine collection was staggered across groups throughout the collection period to control for potential temporal effects on urine composition. Urinary pH was determined immediately after urine collection with a microelectrode, and a 50–100 μL aliquot of urine was centrifuged at 4000 to 5000 × g for 10 minutes. Following centrifugation, the majority of the supernatant was removed and the urine sediment was resuspended in the remaining supernatant and pipetted onto a 0.22 μm Millipore filter placed on a vacuum apparatus to facilitate dispersion and drying of the sediment on the filter surface. The majority of filters were coated with gold prior to evaluation by SEM and XDS.

Necropsy Procedure

Approximately 1 hour prior to necropsy, 12 nonfasted animals/group were injected intraperitoneally with 100 mg/kg bromodeoxyuridine (BrdU) at months 1, 3, 6, 9, and 12 (normal and acidified arms) and at month 18 (acidified arm). The urinary bladder and kidneys were collected after achieving a deep plane (Stage IV) of anesthesia via an intraperitoneal (IP) injection of pentobarbital sodium (initial dose of 120 mg/kg). In addition, a section of duodenum was collected as a positive control tissue for the BrdU procedure (when applicable). The animals designated for each necropsy were generally those from which the most urine samples were collected at the 4 sampling time points prior to that necropsy interval.

For animals with no macroscopic evidence of urinary bladder masses, a ligature was placed around the neck of the urinary bladder with the ligature knot present on the ventral neck and the urinary bladders were inflated with Bouin’s fixative. The ventral portion of the urinary bladder was marked with India ink to facilitate identification of dorsal versus ventral bladder urothelium during tissue trimming. The prostate gland was dissected from the urethra, and the urinary bladder was excised in toto at the neck and placed in Bouin’s fixative for 2 to 4 hours. For animals with grossly visible urinary bladder masses, the lumen of the urinary bladder was opened, the location of the mass was noted, and then the urinary bladder was further fixed in 10% neutral-buffered formalin and transected longitudinally.

The lumen of the urinary bladder of all animals was examined with a dissecting microscope for the presence and location of lesions and/or masses on the urothelial surface and for the presence of microcalculi. In animals with no macroscopic evidence of urinary bladder masses, one half of the urinary bladder was incised into 2–4 mm lengthwise strips (maintaining ventral to dorsal orientation of the urinary bladder tissue). The urinary bladder strips and representative sections of the kidneys were processed, stained with hematoxylin and eosin, and examined by light microscopy. For the 1-, 3-, 6-, and 9-month necropsies, the remaining one-half of the urinary bladder was rinsed and placed in 70% ethanol and routinely processed for SEM and XDS evaluation.

At the 3-, 6-, 9-, and 12-month necropsies, the last 5 animals/group were euthanatized via an IP injection of pentobarbital sodium followed by exsanguination, and urinary bladders were inflation-fixed with 10% neutral buffered formalin. After fixation, urinary bladders were transected longitudinally, maintaining the ventral to dorsal orientation with Indian ink as previously described and then processed and evaluated for apoptosis utilizing the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay (Trevigen, Inc., Gaithersburg, MD).

Microscopic Evaluation of Urothelial Morphology

The urothelium of the renal pelvis and urinary bladder was assessed by light microscopy for evidence of cytotoxicity, inflammation, hypertrophy, hyperplasia, and neoplasia from month 1 to 21. Additionally, the urothelial surface was examined for evidence of cytotoxicity and proliferation by SEM at months 1, 3, 6, and 9. Lesions observed were recorded with respect to dorsal versus ventral bladder urothelium.

Assessment of Urothelial Proliferation and Apoptosis

At months 1, 3, 6, 9, and 12 (normal and acidified arms) and at month 18 (acidified arm), representative sections of the urinary bladder and proximal duodenum (positive control) from 12 animals from each control and 50 mg/kg muraglitazar group were sectioned and evaluated immunohistochemically for nuclear BrdU labeling. The number of animals suitable for scoring of urothelial BrdU staining (based on the presence of satisfactory duodenal crypt staining) was 11–12 rats/group at months 1 and 3, 9–11 rats/group at month 6, 5–10 rats/group at month 9, 8–12 rats/group at month 12, and 12 rats/group at month 18. Since an increased proliferative response was detected in the animals in the 50 mg/kg group fed a normal diet, BrdU labeling indices also were determined for animals in the 1 mg/kg/day group fed a normal diet. All urothelial cells (BrdU-stained and unstained cells) in a total of 2 dorsal and 2 ventral strips of urinary bladder were counted, and the urothelial proliferative response was mapped with respect to dorsal or ventral orientation.

The BrdU-labeling index (%) was calculated by dividing the number of labeled cells by the total number of cells counted x 100. An average of approximately 5,000 urothelial cells was manually counted per urinary bladder. After 3, 6, 9, and 12 months of dosing, the rate of urothelial apoptosis in the urinary bladder of 5 animals/group was assessed by the TUNEL assay. Since there was no drug-related effect on apoptosis at 50 mg/kg/day in either diet group, animals in the 1 mg/kg/day groups were not evaluated. All TUNEL-stained and unstained urothelial cells were counted and recorded with respect to dorsal or ventral orientation of each strip (2 dorsal and 2 ventral strips). Labeling index (%) was calculated by dividing the number of labeled cells by the total number of cells counted x 100. An average of approximately 5,000 urothelial cells was evaluated manually per urinary bladder.

Toxicokinetics and Urinary Drug Concentrations

AUC exposures for muraglitazar were determined from venous blood samples collected from 4 animals/group at approximately 1, 4, 8, and 24 hours after a daily dose during months 3 and 15. Additionally, urine concentrations of muraglitazar were determined in an approximately 5 ml aliquot of urine collected over 24 hours from 6 nonfasted rats/group during months 3 and 15.

Statistical Analysis

Intergroup differences in urinalysis parameters and labeling indices for the BrdU and TUNEL assays were analyzed by the Dunnett’s Test. All analyses were based upon comparisons to appropriate age- and diet-matched control groups.

Results

Urine pH, Electrolyte, and Protein Determinations

Mean and/or individual urine pH values of 50 mg/kg rats fed a normal diet were generally maintained at ≥6.5, the critical threshold for precipitation of calcium and magnesium salts (Cohen, 1995), at all sampling time points and intervals (Figure 1A). Moreover, urine pH in 50-mg/kg rats fed a normal diet was statistically significantly increased at several time points during months 15, 18, and 21 when compared to concurrent control values (months 18 and 21 data not shown). Urine pH values were consistently <6.5 at the 8-hour sampling time point at each monthly interval in the control and 1 mg/kg rats fed a normal diet, whereas it was generally above this threshold at the other time points. Mean urine pH in rats fed an acidified diet was invariably below the critical threshold of 6.5 (Figure 1B). Moreover, the diurnal changes in urine pH observed in animals fed a normal diet (generally lowest at 8 hours postdose and highest at 16 hours post-dose) were also evident in rats fed an acidified diet.

There were no consistent treatment-related effects on mean urine total calcium/creatinine ratios at 2.5 and 16 hours post-dose in muraglitazar-treated rats fed a normal diet (Figure 2A). Although total calcium levels appeared to be mildly decreased at 2.5 hours postdose during months 18 and 21 in 50-mg/kg rats fed a normal diet, a similar effect was not seen at earlier time points. At 16 hours postdose, total urine calcium levels at 1 and 50 mg/kg were generally similar to controls, with a trend toward slightly reduced levels at the high dose. In comparison, consistent, dose-dependent decreases in mean urine soluble calcium/creatinine ratios (up to approximately 60%) were noted at 2.5 and 16 hours post-dose in muraglitazar-treated rats fed a normal diet (Figure 2B).

Moreover, the decreases in soluble calcium/creatinine ratios at 50 mg/kg generally exceeded the decreases, when present, in total calcium levels. On an individual animal basis, soluble calcium levels were decreased up to 80% or to 0.4 mM by month 6 in 50-mg/kg rats fed a normal diet. Additionally, mean urine phosphorous/creatinine ratios were decreased dose dependently (up to approximately 50%) at 2.5 and 16 hours postdose (Figure 2C) in muraglitazar-treated rats fed a normal diet at most monthly intervals. Collectively, these data were consistent with drug-associated decreases in urinary soluble calcium concentrations and increased calcium-and phosphate-containing urinary solids.

Urinary magnesium/creatinine ratios were highly variable in urine samples collected from muraglitazar-treated rats at 2.5 and 16 hours postdose with no treatment-related effect clearly apparent (data not shown). For muraglitazar-treated rats fed an acidified diet, mean urine total and soluble calcium/creatinine ratios were generally comparable suggesting that nearly all of the calcium present was in soluble form. Moreover, the total and soluble calcium/creatinine ratios were decreased significantly (approximately 15 to 71%) across intervals at both dose levels (Figures 3A and 3B) when compared to diet-matched controls, suggesting muraglitazar reduced urinary excretion of calcium. In contrast to findings in 50-mg/kg rats fed a normal diet, there were no clearly drug-related changes in urine phosphorous levels at 2.5 or 16 hours postdose (Figure 3C). Urinary sodium, potassium, and chloride levels were unaffected by muraglitazar treatment in rats fed either a normal or acidified diet (data not shown).

Dose- and time-dependent decreases in mean urine protein/creatinine ratios were apparent at 2.5 and 16 hours post-dose in muraglitazar-treated rats irrespective of diet (Figure 4, 16-hour data not shown). In 50-mg/kg rats fed a normal diet, urine protein to creatinine ratios were decreased approximately 38 to 72% from months 6 through 21 with the largest decreases occurring at month 15 and beyond. In 1 mg/kg rats fed a normal diet, decreases in urine protein were less consistently observed and were generally not statistically significant. Dose- and time-dependent, minimal-to-moderate decreases (approximately 17 to 74%) in urine protein/creatinine ratios also were noted in muraglitazar-treated rats fed an acidified diet. These reductions in urinary protein contrasted with the age-related increases in mean urine protein that occurred in the diet-matched control rats.

Urinary Citrate and Oxalate Determinations

Muraglitazar induced dose- and time-dependent decreases in urine citrate concentrations and/or 18-hour output in animals fed the normal and acidified diets (Figure 5). Urine citrate concentrations were decreased by 1 month at 50 mg/kg and, to a lesser extent, by 6 months at 1 mg/kg in animals fed a normal diet. Similar dose- and time-dependent patterns were apparent with respect to 18-hour urine citrate output. In 50-mg/kg rats fed a normal diet, urine citrate concentrations decreased up to approximately 50% on a group mean basis and by up to 70 to 80% on an individual animal basis at months 6 and 9, with more pronounced decreases (relative to concurrent control values) occurring with longer-term dosing.

In animals fed an acidified diet, mean urine citrate concentration and output were also decreased dose dependently in 1- and 50-mg/kg rats. When compared to the diet-matched concurrent control values, the differences were generally of similar magnitude to those observed in animals fed a normal diet; however, urine citrate concentrations and output were generally much lower in control and muraglitazar-treated rats fed an acidified diet compared to rats fed a normal diet. This was an expected finding since acidification of the urine is known to increase renal proximal tubular absorption and thereby decrease excretion of citrate (Sheridan et al., 1983; Brennan et al., 1988). Importantly, the acidic urine of these rats markedly inhibited formation of calcium and magnesium-containing precipitates and crystalline material in the presence of the prolithogenic change of reduced urinary citrate levels.

At 50 mg/kg, mean urine oxalate concentrations and/or output were variably increased (11 to 56%) at all intervals in rats fed the normal and acidified diets (Figure 6). The increases in urine oxalate in 50 mg/kg rats fed a normal diet combined with pronounced reductions in urine citrate and urine pH ≥6.5 likely accounted for observation of thin, rod-shaped, calcium oxalate-containing crystals at this dose level. However, due to the inhibitory effect of urine acidification on urinary calcium salt formation, the potential for calcium oxalate crystalluria would be reduced markedly in animals fed an acidified diet relative to animals fed a normal diet.

Urine Sediment Analysis

Calcium and magnesium-containing microcrystalline precipitates, crystals, aggregates, and/or microcalculi were increased dose dependently in the urinary sediment of muraglitazar-treated male rats fed a normal diet (Table 1). Specifically, amorphous calcium phosphate precipitate was increased at 1 and 50 mg/kg during months 1 (Figures 7A and 7B), 6, 12, and 15; and MgNH₄PO₄ crystals and/or aggregates were increased at all intervals evaluated (Figures 7C, 7D, and 7F). The calcium phosphate precipitate and MgNH₄PO₄ crystals were sometimes associated with or aggregated within a web-like organic matrix (Figure 7E). Additionally, in animals fed a normal diet, increased polymorphic calcium phosphate crystals were detected in urine of 1- and 50-mg/kg rats during months 6 and 9, and a dose-dependent low incidence of jagged calcium phosphate microcalculi were observed at 1 mg/kg during month 9 and at 50 mg/kg during months 1, 9, (Figure 7F) and 15. Finally, calcium oxalate-containing thin, rod-like crystals were noted at an incidence that was dose-dependent during months 6 and 15 in animals fed a normal diet.

Acidification of the urine markedly reduced the burden of urine solids in control and muraglitazar-treated rats when compared to their counterparts fed a normal diet (Table 1). This reduction was most evident for calcium phosphate-containing solids. Acidification of the urine did not completely mitigate muraglitazar-related increases in MgNH₄PO₄ crystals and aggregates, but clearly reduced the incidence and severity of each. Moreover, the overall burden of MgNH₄PO₄ crystals and aggregates in muraglitazartreated rats fed an acidified diet was considerably less than that observed in control rats fed a normal diet.

Disposition of Grossly Visible Urinary Calculi and Urinary Bladder Masses

Three 50 mg/kg rats fed a normal diet had grossly visible kidney and/or urinary bladder calculi composed of MgNH₄PO₄ or organic material. One animal had bilateral MgNH₄PO₄ calculi present in the renal pelvis and an organic calculi in the urinary bladder, another had bilateral renal pelvic calculi composed principally of organic material (a likely matrix for MgNH₄PO₄ calculi formation), and the last animal had a bladder distended with numerous spherical to elongate MgNH₄PO₄ calculi. There were no grossly visible MgNH₄PO₄ calculi in control and 1 mg/kg animals fed a normal diet or in rats given muraglitazar and fed an acidified diet.

Urinary bladder masses were first detected after 9 months of treatment. With the exception of those masses that had progressed to a point at which the site of origin could not be determined, most were located in the ventral bladder (Figure 8), with the ventral apex the most common site. In 50 mg/kg rats, there were only 2 tumors (out of 41 observed) with a dorsal bladder orientation and these were detected late in the study (months 18 and 21). Even though this region of the bladder would be less susceptible to injury than the ventral bladder as a consequence of increased urinary solids, the predominantly ventral bladder disposition of masses was highly supportive of a primary etiology of urolithiasis. There were no masses observed in the urinary bladders of rats given muraglitazar for up to 21 months and fed an acidified diet.

Urothelial Histopathologic and Ultrastructural Changes

Light Microscopy

Muraglitazar-related histopathologic changes were present in the urinary bladder only in rats fed a normal diet (Table 2). Dose- and time-dependent simple urothelial hyperplasia of primarily the ventral bladder mucosa was noted by month 9 at 1 mg/kg and month 3 at 50 mg/kg. At 50 mg/kg, the lesions progressed from simple urothelial hyperplasia to nodular hyperplasia, papilloma, and carcinoma of the urothelium with a ventral bladder predisposition by month 9 (Table 2, Figure 9). In addition to the urinary bladder changes, urothelial hyperplasia of the renal pelvis was observed in one 50-mg/kg animal with bilateral renal pelvic MgNH₄PO₄ calculi.

The single incidences of nodular hyperplasia, transitional cell papilloma, and transitional cell carcinoma observed by light microscopy in 1 mg/kg rats fed a normal diet were not clearly drug related since a transitional cell papilloma occurred spontaneously at month 21 in a control rat fed an acidified diet and a comparable incidence of transitional cell carcinoma (5/130) occurred in control male rats treated for up to 24 months in the oral carcinogenicity study in rats (Simutis et al., 2006).

There were no muraglitazar-related histopathologic changes in the urinary bladder or kidney of rats fed an acidified diet. In the urinary bladder of rats fed an acidified diet, simple urothelial hyperplasia was noted in one 50-mg/kg animal and both simple and papillary urothelial hyperplasia were seen in another 50-mg/kg animal at month 21. These changes were considered not to be drug related since a transitional cell papilloma was detected in an acidified diet-matched control animal at the same time point.

Scanning Electron Microscopy

Ultrastructural evaluation of the urinary bladder at 1, 3, 6, and 9 months by SEM demonstrated dose-related and time-dependent superficial urothelial necrosis and associated regenerative urothelial hyperplasia in the ventral urinary bladder of rats fed a normal diet. In 1-mg/kg animals, superficial urothelial necrosis was focal and localized to the ventral dome of the bladder, and the urothelial hyperplasia was generally minimal in severity. One exception was a 1-mg/kg rat that had locally extensive superficial urothelial necrosis in the ventral dome of the urinary bladder without accompanying regenerative hyperplasia after 6 months of treatment. With the exception of a single animal with nodular urothelial hyperplasia at 9 months, there was no clear evidence of progression of the urothelial changes at 1 mg/kg. At 50 mg/kg, the urothelial changes were more pronounced with increased duration of dosing and characterized by focal to locally extensive areas of superficial urothelial necrosis and associated minimal to severe diffuse urothelial hyperplasia at 3, 6, and 9 months (Figures 10B, 10D).

In addition, nodular urothelial hyperplasia was noted in two 50-mg/kg rats at 6 months (Figure 10E), 3 months prior to the detection of similar lesions by light microscopy. At both 1 and 50 mg/kg, calcium phosphate precipitate was observed occasionally on the urothelial surface of the ventral apex of the bladder proximal to areas of superficial urothelial necrosis and hyperplasia (Figure 10C). The urinary bladder urothelium of rats in the control and muraglitazar-treated groups fed an acidified diet were within normal limits at all intervals evaluated (Figures 10A vs 10F).

Urothelial BrdU Immunohistochemistry

Urothelial cell proliferation, as measured by nuclear BrdU incorporation, was significantly increased at all treatment intervals in male rats administered 50 mg/kg muraglitazar and fed a normal diet (Figure 11). In 50-mg/kg male rats fed a normal diet, overall urothelial proliferation rates were statistically significantly increased approximately 2.3-fold at month 1; 9- to 10-fold at months 3, 6, and 9; and 4.5-fold at month 12. The lower BrdU labeling index at month 12 relative to months 3 through 9 likely reflected the fact that urinary bladders from 4 of 12 rats at this interval were excluded from the BrdU analysis due to the presence of transitional cell carcinomas in the urinary bladder of these animals. From months 1 through 9, there was considerable interanimal variability in the proliferative response at 50 mg/kg, and approximately 50% of the 50-mg/kg animals at months 1 and 3 had urothelial BrdU labeling indices similar to concurrent control mean values. With the exception of 3 animals (1 at month 1, 1 at month 3, and 1 at month 6), the urothelial cell BrdU labeling indices in 50-mg/kg rats were generally higher in the ventral than the dorsal bladder with ventral labeling indices 3 to 17 times greater than dorsal labeling indices (Figure 12). In rats administered 1 mg/kg muraglitazar and fed a normal diet, a minimal statistically insignificant increase in the overall urothelial proliferation rate was noted at month 1 and to a lesser extent, at month 3 (Figure 11). These increases were due primarily to a predominately ventral proliferative effect in 6 of 12 animals at month 1 and 2 of 12 animals at month 3 (data not shown). At the tumorigenic dose of 50 mg/kg, the urothelial proliferative response was completely prevented by dietary and consequent urinary acidification (Figure 11). Last, overall mean urothelial cell proliferation rates in control rats fed a normal and acidified diets were similar (0.03 to 0.18%) at all intervals and comparable to reported values (0.04 to 0.5%) (Oyasu, 1995) in adult rats indicating that urinary acidification did not alter the background rate of urothelial proliferation.

Urothelial Apoptosis

There were no muraglitazar-related effects on the rate of urothelial apoptosis in rats fed a normal or an acidified diet (Figure 13). Mean apoptotic indices in muraglitazar-treated rats were comparable to the diet-matched controls at all study intervals. Additionally, background apoptotic rates were comparable irrespective of diet.

Toxicokinetics and Urinary Drug Concentrations

Systemic exposures (AUC) to muraglitazar were dose related, increased approximately in proportion to dose, were generally comparable between months 3 and 15, and were somewhat lower in rats fed an acidified diet compared to rats fed a normal diet. (Table 3). Importantly, the mean AUCs achieved at 1 and 50 mg/kg during months 3 and 15 in animals fed a normal or an acidified diet were comparable to or higher than those noted in male rats at the same dose levels during month 6 in the oral carcinogenicity study (Simutis et al., 2006). Additionally, urine concentrations of muraglitazar were dose related, increased in a less than dose-proportional manner, and were similar between rats fed normal and acidified diets. Therefore, the absence of cytotoxic and proliferative urothelial responses in the urinary bladder of 50-mg/kg male rats fed an acidified diet could not be attributed to lower exposures to muraglitazar.

Discussion

The urinary bladder tumorigenic response in male rats treated with muraglitazar was considered to be pharmacologically mediated given that structurally diverse PPARγ or PPARα/γ agonists have been shown to similarly increase transitional cell tumors in the urinary bladder and, in fewer cases, in the renal pelvis of rats (Cohen, 2005; El-Hage, 2005). Therefore, to more fully assess the relevance of these rodent tumor findings to potential human risk at therapeutic exposures, it was critical that an investigative program be conducted to elucidate the mode of tumor development.

In our investigative studies with muraglitazar, the potential for a direct or indirect pharmacologic effect on the urothelium was explored because evidence of direct drug-related urothelial cytotoxicity or direct urothelial PPAR modulation of cell proliferation or apoptosis could potentially have major implications to the assessment of carcinogenic risk in humans. In contrast, demonstrating that the urinary bladder tumorigenic response was secondary to muraglitazar-induced changes in urine composition would support that this rat-specific finding had little or no relevance with respect to urinary bladder cancer risk in humans given the profound differences in normal urine composition between rats and humans (Cohen, 1995).

The outcome of our investigative studies strongly supports that the urinary bladder tumorigenic response in rats treated with muraglitazar was a consequence of pharmacologically mediated changes in urine composition rather than direct drug-related cytotoxicity or direct PPAR-mediated stimulation of urothelial cell proliferation or inhibition of apoptosis. Moreover, data from these investigations indicate that the critical event in the mode of tumor development was a pharmacologically mediated increase in urine calcium- and magnesium-containing precipitates, crystals, aggregates, and microcalculi. The persistently increased urinary solids were accompanied by hypocalciuria (soluble fraction) and resulted in chronic dose-dependent cytotoxicity and irritation of predominantly the ventral bladder urothelium, effects which led to sustained increases in urothelial proliferation with progression to urothelial neoplasia of principally the ventral bladder by 9 months of treatment. Neither muraglitazar nor its urinary metabolites would be expected to have contributed to the increase in urinary solids since less than 1% of the administered dose is excreted in the urine of HSD rats.

Key to the development of a prolithogenic urinary environment in muraglitazar-treated rats fed a normal diet was the combination of urine pH at ≥6.5, the critical threshold for precipitation of calcium- and magnesium-containing salts (Cohen, 1995), and the presence of dose- and time-dependent decreases in urine citrate levels and increases in urine oxalate levels. These urinary changes are known to result in a highly prolithogenic environment (Bisaz et al., 1978; McLean et al, 1990; Hamm and Hering-Smith, 2002; Caudarella et al., 2003) at the supersaturated levels of calcium and magnesium typical to rat urine.

As a consequence, the 50-mg/kg group fed a normal diet was particularly predisposed to the formation/precipitation of calcium and magnesium salts because urine pH was generally at or above 6.5 in most rats throughout the light and dark cycle and urine citrate was most profoundly reduced on an individual animal basis at this dose level. Marked diurnal variations in urine solids and chemistries were observed, as expected (Fisher et al., 1989). Moreover, collection of fresh-void urine samples at multiple times during the light and dark phase was essential to provide a complete picture of urine compositional changes induced by muraglitazar.

Citrate, as the predominant organic ion in urine (Brown et al., 1989) has been purported to account for up to 50% of the total inhibitory activity against calcium phosphate precipitation in normal urine (Bisaz, 1978; Hamm and Hering-Smith, 2002). The antilithogenic effect of urinary citrate is largely a consequence of its ability to reduce urinary saturation of calcium and magnesium salts by forming soluble complexes with these divalent cations (Pak, 1987). Furthermore, it directly inhibits spontaneous nucleation, growth, and aggregation of crystals formed from calcium salts (Lieske and Coe, 1996) and the crystallization of magnesium salts (McLean et al., 1990).

The trivalent species of citrate predominates in normal urine with levels regulated at the proximal tubule via the brush border sodium dicarboxylate cotransporter (NaDC-1), the activity of which is determined by acid-base balance (Hamm, 1990). Additionally, urine pH regulates renal excretion of citrate due to increased proximal tubular uptake of the protonated (divalent) species in an acidic urine (Brennan et al., 1988). This latter regulatory mechanism explains the basis for greater reductions in urine citrate levels in muraglitazar-treated rats fed an acidified diet. Importantly, the marked reductions in urine citrate levels in this group did not result in a prolithogenic environment because the proportion of divalent and trivalent phosphate ions available for complexing with calcium and magnesium are decreased markedly in acidic urine (Brown and Purich, 1992).

In the presence of low urine citrate and high urine pH as noted in rats administered 50 mg/kg muraglitazar and fed a normal diet, chelation of calcium by citrate would be reduced which could potentially promote calcium’s interaction with urinary acidic proteins, such as α_{2 u} globulin, albumin, or osteopontin (Umekawa et al., 1995), thereby promoting protein self-aggregation or protein-calcium interactions favoring crystal formation. Urinary protein aggregates could then serve as sites for heteronucleation, growth, and/or aggregation of calcium- and magnesium-containing crystals. In support of this notion, normal physiologic concentrations of citrate have been shown to reduce the potential for self-aggregation of the polyanionic glycoprotein Tamm–Horsfall protein (THP) via chelation of calcium, whereas low concentrations of citrate lead to self-aggregation of THP, which can then serve as a nidus for crystal formation (Hess et al., 1993). Moreover, calcium phosphate-containing precipitate has been shown to contain up to 5% protein, including α_2u-globulin and albumin (Cohen et al., 2000), and NCI-Black Reiter rats, a strain that does not synthesize α_2u-globulin, have reduced calcium phosphate precipitate-induced urothelial cell proliferation following oral administration of sodium saccharin or sodium ascorbate (Garland et al., 1994; Uwagawa et al., 1994). In male rats given 50 mg/kg muraglitazar, the branching weblike organic matrix associated with the calcium phosphate precipitate and MgNH₄PO₄ crystals may have also contributed to crystal formation, growth, and aggregation.

The mechanism whereby muraglitazar decreases urine citrate is not fully understood. Diet, bone, and peripheral tissues (including fat and muscle) are sources of serum citrate (Gordan and Craigie, 1960) while kidney (Simpson, 1983) and liver (Vang et al., 1966; Kaneshige, 1975) serve as the primary sites of citrate metabolism. Since dose-dependent decreases in serum citrate levels of up to 42% occur in male rats after 1 to 3 months of treatment at 1 and 50 mg/kg of muraglitazar (unpublished data), the reductions in urine citrate levels in the present investigation were considered primarily due to its decreased urinary filtration; however, increased renal proximal tubular uptake and metabolism of citrate also could be contributing factors.

The basis for the reduced serum citrate levels in muraglitazar-treated rats has not been established but likely involves PPARγ-mediated increased peripheral insulin sensitization and associated stimulation of lipogenesis since fatty acid synthesis in brown and white fat is promoted by insulin (Denton and Brownsey, 1983) and is citrate consuming (Ball, 1966). Moreover, insulin has been shown to increase lipogenesis (Madsen et al., 1964; Cusin et al., 1990) and the glucose utilization index (Cusin et al., 1990) in fat of normal rats and to stimulate phosphorylation of adipocyte ATP citrate lyase, the cytosolic enzyme regulating lipogenesis (Fried et al., 1981) and responsible for the formation of acetyl-CoA from citrate (Denton and Brownsey, 1983). Additionally, in a pre-adipocyte fibroblast cell line, PPARγ1 and γ2 isoforms both induced increased aconitase activity, the TCA cycle enzyme that coverts citrate to isocitrate, without similar increases in citrate synthase (Mueller et al., 2002), suggesting PPAR-mediated effects on intermediary metabolism of differentiating adipocytes also may be contributing to the reductions in serum citrate in muraglitazar-treated rats.

Although the mechanism for the muraglitazar-related increases in urinary soluble oxalate levels and calcium oxalate-containing thin rodlike crystals has not been determined, there is experimental evidence suggesting that PPARα regulates hepatic metabolism of glyoxalate, the primary source of urinary oxalates. The endogenous metabolism of glyoxylate to oxalate is catalyzed by lactate dehydrogenase (LDH) and, to a lesser extent, by glycolate oxidase (GAO) (Yanagawa et al., 1990; Holmes and Assimas, 1998) and accounts for up to 50–60% of urinary oxalate levels (Ogawa et al., 2000). The liver-specific enzyme alanine glyoxalate aminotransferase (AGT) (Pirulli et al., 2003) converts approximately 65 to 80% of the glyoxalate pool to glycine (Ogawa et al., 2000) and thereby plays a major role in regulating the size of the hepatic glyoxalate pool. The PPARα agonist, WY-14643, has been shown to inhibit the expression of AGT in mice (Kersten et al., 2001; Genolet et al., 2005), which would theoretically result in an increase in the glyoxalate pool.

Additionally, hepatic synthesis and urinary excretion of oxalate were increased in Sprague–Dawley (SD) rats administered clofibrate via pharmacologic induction of hepatic LDH and GAO activities (Sharma and Schwille, 1997). Thus, a compound with rodent PPARα activity, such as muraglitazar, could potentially increase urinary oxalate levels and thereby further increase the burden of urinary solids by inducing increased metabolism and/or reduced amino acid (glycine) conversion of glyoxalate. Furthermore, there is the potential for a male rat predisposition to hyperoxaluria since hepatic glycolate oxidase is induced by testosterone and inhibited by estrogen (Tiselius et al., 1980; Yoshihara et al., 1999) and because hepatic AGT activity is higher in female than in male SD rats (Yoshihara et al., 1999). There is also the potential for modulation of urinary citrate levels by increased glyoxylate since intravenous dosing of male Wistar rats with glyoxylate resulted in up to 33% reductions in urine citrate (Ogawa et al., 2000).

There was clear evidence from the urine sediment and ultrastructural evaluations that contact of the ventral urinary bladder urothelium with urinary solids was increased in muraglitazar-treated rats fed a normal diet. The potential for chronic irritation of the rat urothelium due to increased urinary solids is exacerbated because rats are quadrupeds and this orientation favors settling of solids to the anteroventral regions of the urinary bladder due to gravity. If excessive crystals or calculi are present at the urothelial surface, the urothelium in the ventral aspects of the urinary bladder is readily irritated with bladder contraction during urination (DeSesso, 1995).

Since the internal urethral orifice is along the same plane as the anteroventral wall of the rat bladder, urinary precipitates, crystals, aggregates, and calculi can remain in the bladder and irritate the urothelium for prolonged periods without interfering with the outflow of urine (DeSesso, 1995). In addition, calcium phosphate precipitate has been shown to be directly cytotoxic to epithelial cells (Wigler et al., 1979; Brash et al., 1987), including urothelial cells of rats (Cohen et al., 1995b, 2000). Therefore, formation of excessive calcium phosphate precipitates in the urinary bladder, as seen in muraglitazar-treated male rats, likely also contributed to the superficial urothelial injury by a mechanism independent of urothelial irritation. In fact, prolonged contact of cytotoxic calcium phosphate precipitates with the ventral bladder urothelium during periods of inactivity or sleep (i.e., early light phase) would be potentially a consequence of reduced urine citrate levels in an alkaline urine, conditions observed at a tumorigenic dose of muraglitazar in male rats.

The persistence of low levels of normal MgNH₄PO₄ crystals and aggregates in the urine of some animals given a tumorigenic dose of muraglitazar and fed an acidified diet was likely a consequence of the extremely low urine citrate levels in this group. That is, the inhibitory effect of mild urinary acidification on urine crystal formation was somewhat compromised by the exaggerated decreases in urine citrate concentrations and output resulting from the increased proximal tubular reabsorption of citrate that occurs in an acidic urine (Brennan et al., 1988). Importantly, the levels of MgNH₄PO₄ crystals and aggregates that remained in the urine of muraglitazar-treated male rats fed an acidified diet were invariably less than that observed in control male rats fed a normal diet.

Through the first 6 months of muraglitazar treatment, dose-and time-dependent morphologic changes in the urinary bladder were observed most readily by SEM and were characterized by focal to locally extensive superficial urothelial necrosis and associated regenerative urothelial hyperplasia in primarily the ventral bladder with progression to nodular hyperplasia and urothelial neoplasia by month 9. The progression of the urothelial proliferative lesions in the ventral urinary bladder in 50-mg/kg rats fed a normal diet continued throughout the course of treatment based on the increasingly higher incidence of transitional cell carcinoma over time. There was no clear indication of progression of the proliferative changes in the 1-mg/kg rats fed a normal diet.

In addition, there was no morphologic evidence of urothelial cytotoxicity or proliferation in the urinary bladders of 50-mg/kg rats fed an acidified diet in the presence of systemic exposures to parent drug that were markedly higher than those observed at the lowest tumorigenic dose of 5 mg/kg and comparable to those at 50 mg/kg in the oral carcinogenicity study in rats.

Transitional cell papilloma and carcinoma of the urinary bladder are rare spontaneous tumors in most laboratory rat strains (Kunze, 1992). In our historical control tumor database for male HSD rats, background incidence rates (ranges) of transitional cell papilloma and carcinoma are approximately 0.3% (0–5%) and 0.3% (0–2%), respectively, and are higher than the rates reported for Charles River SD rats (Giknis and Clifford, 2004), F344 rats (NTP Database, 2005), and Wistar rats (Giknis and Clifford, 2003), other strains commonly used in rodent carcinogenicity studies. Moreover, based on results of previous carcinogenicity studies in HSD rats conducted or sponsored by our laboratories, the incidence of proliferative urinary bladder lesions (urothelial hyperplasias and tumors) are generally higher in control male HSD rats administered a vehicle containing PEG-400 compared to those administered aqueous vehicles. The basis for this vehicle effect is unclear, but may involve vehicle-associated perturbations of urinary citrate excretion and associated increased crystalluria since PEG-300 has been shown to reduce citrate levels in the urine of SD rats (Beckwith-Hall et al., 2002). Interestingly, we have recently determined that male Wistar rats have significantly higher serum and urinary levels of citrate and considerably less urinary calcium- and magnesium-containing solids than male HSD rats, a finding of potential relevance since urinary bladder tumors have not been reported to date in Wistar rats treated with PPAR agonists.

Consistent with the pronounced urothelial proliferative changes observed microscopically at 50 mg/kg of muraglitazar, there was a sustained increase in urothelial proliferation, as measured by BrdU immunohistochemistry, after 1 to 12 months of dosing in rats fed a normal diet. In contrast, at 1 mg/kg, the mean BrdU labeling index was only mildly increased at month 1 and to a lesser extent, at month 3. The basis for the transient nature of the BrdU labeling response in 1-mg/kg male rats fed a normal diet is unclear, but may be related to age-related differences in urine composition that more readily predispose young male Harlan rats to increased urinary solids following treatment with muraglitazar.

Importantly, BrdU labeling indices were generally considerably higher in the ventral than in the dorsal bladder urothelium, a pattern consistent with a primary etiology of increased urinary solids rather than direct pharmacologic stimulation of urothelium. The associated increased urothelial proliferation rate in the dorsal bladder was not unexpected, since the dorsal apex has a somewhat ventral disposition in the intact animal and all regions of the bladder are susceptible to mechanical injury from increased urinary solids during bladder contraction. In muraglitazar-treated rats fed an acidified diet, the absence of alterations in urothelial BrdU labeling indices correlated with the lack of proliferative changes in the histopathologic and SEM evaluations.

Although data from our investigative studies strongly support urolithiasis as the primary mode of urinary bladder tumor development, the pathogenesis of the lesion likely involves other factors that could impact the mitogenic response at sites of urothelial injury. Specifically, the dose-and time-dependent hypocalciuria (reduced soluble calcium) noted in male rats given 50 mg/kg may have contributed to the mitogenic response at sites of urothelial injury since low soluble calcium concentrations similar to those noted in individual rats given 50 mg/kg muraglitazar have been shown to markedly stimulate urothelial mitogenesis of rat urothelium in vitro (Reese and Friedman, 1978).

Importantly, there was no evidence for a direct mode of action for muraglitazar-induced tumor development based on the distribution of the urothelial cytotoxic and proliferative responses. Moreover, muraglitazar is nongenotoxic, and the tumorigenic response was male rat specific even though systemic and urinary exposures to drug and metabolites were similar in male and female rats across the tumorigenic dose range. Furthermore, there was compelling evidence that the urinary bladder neoplastic response was a consequence of pharmacologically mediated alterations in urine composition given that urolithiasis and associated urothelial cytotoxic, proliferative, and tumorigenic responses were prevented by urinary acidification.

Although urinary crystals, precipitates, and calculi predispose to urinary bladder tumorigenesis in rodents, particularly male rats, there is no epidemiologic data implicating persistent crystalluria (i.e., individuals with inborn errors of metabolism such as cystinuria, xanthinuria, and hyperoxaluria) (Burin et al., 1995) as a risk factor for bladder cancer in humans, and there is contradictory epidemiologic evidence with respect to the relevance of long-standing bladder calculi in humans to increased risk for bladder cancer (Howe et al., 1985; Claude et al., 1986; Kjaer et al., 1989; LaVecchia et al., 1991; Groah et al., 2002). This apparent disparity in susceptibility to irritation-induced bladder tumors is considered, in part, due to postural and anatomic differences in the orientation of the urinary bladder in biped humans compared to quadruped rodents. In effect, unlike the rat, the anatomic orientation of the urinary bladder in humans favors clearance of potentially irritating urinary solids. Importantly, in clinical trials with muraglitazar, there was no evidence of an increased incidence of nephrolithiasis, urolithiasis, or increases in urinary crystals in diabetic patients.

Since PPARα and γ are expressed in normal urothelium of rodents and humans (Guan et al., 1997; Jain et al., 1998; Escher et al., 2001), the potential for a direct pharmacologic basis or component to the urinary bladder tumorigenic response in rats administered PPARγ and PPARα/γ agonists was investigated. However, urothelial PPARα and γ mRNA expression has been reported to be similar in male and female rats (Escher et al., 2001), which suggests that the predilection for male rats to develop bladder tumors following chronic treatment with muraglitazar could not be attributed to gender differences in urothelial PPARα or γ expression.

Furthermore, the predisposition of the urothelial proliferative response to ventral aspects of the urinary bladder cannot be ascribed to regional differences in PPAR α or γ expression since these receptors are similarly expressed in dorsal and ventral bladder urothelium of male and female rats and mice (manuscript in preparation). In addition, the absence of muraglitazar-related urinary bladder tumors in mice cannot be explained based on species differences in activation of these receptors since the in vitro binding affinity and transactivation potentials of muraglitazar at PPARα or PPARγ are similar in rats and mice (unpublished data), and systemic exposures to parent drug were higher in mice than in rats in the oral carcinogenicity studies (Simutis et al., 2006). And finally, we have recently demonstrated that urinary acidification via 1% dietary ammonium chloride does not alter the immunohistochemical expression profiles of PPAR (α, γ, and δ) or epidermal growth factor (total and phosphorylated) receptors or the expression of select target genes for these receptors after subchronic dosing (manuscript in preparation).

Collectively, these data support that direct urothelial PPAR activation by muraglitazar did not likely contribute to the urothelial proliferative and tumorigenic responses in rats and that urinary acidification did not obviate the proliferative response via a mechanism involving altered PPAR or EGF receptor expression or signaling.

In contrast to the work conducted by our laboratory, there are recent studies conducted with ragaglitazar where the authors have suggested that direct urothelial PPAR activation may be a factor in the observed tumorigenic response with that agent (Egerod et al., 2005; Oleksiewicz et al., 2005). In those studies, a tumorigenic dose of ragaglitazar induced urothelial hypertrophy of the urinary bladder and renal pelvis and associated increases in urothelial expression of the transcription factor Egr-1 and increased phosphorylation of cJun and ribosomal S6 protein in rats treated for up to 3 weeks. Treatment of rats with rosiglitazone or fenofibrate alone resulted in no clear induction of Egr-1 and a transient, less robust increase in phosphorylated S6 protein, whereas the combined treatment with rosiglitazone and fenofibrate increased these proteins to levels approximately one-half (Egr-1) or similar (phosphorylated S6) to those observed with ragaglitazar. Levels of phosphorylated c-Jun were not measured for the comparators. Although these findings suggest a PPARα and/or γ-mediated basis for the urothelial hypertrophic response observed, they do not rule out an indirect pharmacologic mechanism involving PPAR-mediated changes in urine composition.

As Oleksiewicz et al. (2005) indicated, urothelial hypertrophy has also been seen in rats treated with diuretics that do not cause urinary bladder cancer and with carbonic anhydrase inhibitors (Molon-Noblot et al., 1992), which induce urothelial hyperplasia and tumors by a mechanism involving urolithiasis. Furthermore, urothelial hypertrophy is a common, nonspecific response to a variety of cytotoxic, mitogenic, and carcinogenic stimuli in rodents (Cohen, 1983, 1989, 1998). Moreover, Egr1 and c-Jun are immediate/early genes with the former increasing in response to not only mitogens and growth factors, but also to stress stimuli (Lim et al., 1998; Thiel and Cibelli, 2002). Therefore, the urothelial hypertrophic and cellular protein changes observed could potentially have been due to PPAR-mediated changes in urine composition rather than direct urothelial PPAR activation.

There is a growing body of literature that suggests direct urothelial PPARγ activation is not likely contributing to the induction of urothelial tumors in rats treated with PPARγ or PPARα/y agonists. That is, the PPARγ agonist troglitazone has been shown to facilitate differentiation rather than proliferation of normal human urothelium via a PPARγ-dependent pathway, whereas the PPARα specific ligand clofibrate had no effect (Varley et al., 2004). Additionally, the natural PPARγ ligand 15-deoxy-Delta (12,14) prostaglandin J(2) and the PPARγ agonists troglitazone and pioglitazone dose-dependently suppressed proliferation of nonneoplastic (Nakashiro et al., 2001; Kawakami et al., 2002) and neoplastic (Nakashiro et al., 2001; Yoshimura et al., 2003a) human urothelial cells in vitro.

Furthermore, a new class of PPARγ agonists, 1,1-bis (3′-indolyl)-1-(p-substituted phenyl) methanes (C-DIMs), has recently been shown to have potent anti-proliferative effects on bladder cancer cells in vitro and to inhibit the proliferation and growth of orthotopically and subcutaneously implanted tumors in nude mice (Kassouf et al., 2006). Although a positive correlation was observed between PPARγ expression and tumor grade and progression in one study of human bladder tumors, the authors demonstrated a PPARγ agonist-mediated potent anti-proliferative effect on the same bladder tumor cells in vitro (Yoshimura et al., 2003b). Moreover, no positive correlation between PPARγ expression and tumor grade, stage, onset, or number was observed by others in a separate study (Possati et al., 2002). Rather, the presence of PPARγ expression was higher in papillary than in solid (more advanced) tumors and was associated with a low incidence of tumor recurrence or progression.

In conclusion, the mode of urinary bladder tumor development in male rats administered muraglitazar involves pharmacologically mediated changes in urine composition that predispose to urolithiasis rather than direct drug-related urothelial cytotoxicity or proliferation. This conclusion is supported by the demonstration of prolithogenic changes in urine concentrations of citrate and oxalate in a generally alkaline urine; increased urinary calcium and magnesium-containing solids; the primarily ventral to anteroventral disposition of the urothelial cytotoxic, proliferative, and tumorigenic responses; the ability to prevent the urothelial changes by mild urinary acidification; and, the absence of urinary bladder tumors in female rats and male and female mice at higher systemic exposures than that observed in male rats in the oral carcinogenicity study. At sites of urinary bladder urothelial injury, the regenerative response may have been exacerbated by the marked reductions in urinary soluble calcium levels. Although there may be additional, as yet undetermined, changes in urine composition contributing to the proliferative and tumorigenic responses in male rats treated with muraglitazar, data from our investigations suggest that the primary and critical step in the mode of action is the development of urolithiasis.