Urothelial Carcinogenesis in the Urinary Bladder of Rats Treated with Naveglitazar,a γ-dominant PPAR α/γ Agonist: Lack of Evidence for Urolithiasis as an Inciting Event

Abstract

Naveglitazar, a γ-dominant peroxisome proliferator-activated receptor (PPAR) α/γ dual agonist, was tested for carcinogenicity in F344 rats in a 2-year study. Changes in urine composition and urothelial morphology were characterized in a companion 18-month investigative study. A significant increase in neoplasms of the bladder occurred only in females of the high-dose group (14/60) in the carcinogenicity study. Trends toward increased cell proliferation in the urothelium were noted in both sexes at all time points evaluated in the 18-month study. Group means for urothelial mitogenesis were increased statistically significantly only in high-dose females at 12 and 18 months. Urothelial hyperplasia occurred in high-dose females at 18 months. Morphologic changes in the urothelium at earlier time points were limited to hypertrophy and decreased immunolabeling of the superficial cells for cytokeratin 20 (a marker of terminal differentiation in urothelial cells) in both males and females. No treatment-related changes in urinary parameters, including urinary sediments, were associated with the occurrence of urothelial proliferation. Urinary pH was unaffected by treatment in both males and females, but expected diurnal changes were demonstrated. Collectively, these data indicate that naveglitazar was associated with hypertrophic and proliferative effects on the urothelium, but a link with changes in urinary parameters was not demonstrated.

Introduction

Peroxisome proliferator-activated receptors (PPARs) play a critical role as regulators of lipid metabolism and insulin sensitization. Synthetic ligands for PPARs have been developed for the treatment of diabetes and dyslipidemia (Berger and Moller, 2002). Carcinogenicity has been identified as one of the primary issues in preclinical testing of PPAR γ and α/γ dual agonists. Administration of these PPAR agonists to rodents has been associated with the development of hemangiosarcomas in mice, liposarcomas and/or fibrosarcomas in rats, and transitional cell tumors of the urinary bladder and/or renal pelvis in rats (El Hage, 2005a, 2005b).

Mechanistic explanations for the occurrence of neoplasms related to the pharmacologic class are problematic because, with few exceptions, in vitro and in vivo work indicates that PPAR γ agonists generally have antiproliferative effects (Grommes et al., 2004). Differentiation and reversal of malignant changes in colon cancer have been associated with PPAR γ (Sarraf et al., 1998; Sarraf et al., 1999). Conversely, enhancement of colon carcinogenesis caused by activation of PPAR γ has been seen in a mouse model with inherent abnormalities in cell-cycle control (Lefebvre et al., 1998; Saez et al., 1998).

The immunohistochemical expression of PPAR γ appears to be increased in human bladder tumor cells compared to normal bladder cells, and the degree of this expression may be associated with tumor grade, recurrence, and/or progression (Possati et al., 2002; Yoshimura, Matsuyama, Hase, et al., 2003; Yoshimura, Matsuyama, Segawa, et al., 2003; Nakashiro et al., 2001). PPAR γ ligands inhibit the proliferation of normal and neoplastic urothelial cells in vitro (Nakashiro et al., 2001; Yoshimura, Matsuyama, Hase, et al., 2003). With normal urothelial cells in culture, PPAR γ agonists cause terminal transitional differentiation, which is maximized when epithelial growth factor receptor (EGFR) is inhibited (Varley et al., 2003; Varley et al., 2004; Kawakami et al., 2002; Spencer et al., 2004). One potential contrasting effect of PPAR γ ligands on urothelium is an apparent increase in expression of vascular endothelial growth factor (VEGF) by neoplastic urothelial cells in vitro, compared to normal urothelial cells (Fauconnet et al., 2002).

PPAR γ is normally expressed in the urothelium of the bladder and renal pelvis (Guan et al., 1997), and the above-cited in vitro studies suggest that PPAR γ agonists should cause decreased cell proliferation and increased differentiation in the urothelium. Additionally, compounds in this pharmacologic class tend to lack genotoxicity and to be excreted into the urine to only a slight degree. Taken together, these factors suggest that carcinogenicity caused by PPAR γ agonists in the rat bladder is not the result of a direct effect of the agonists but is more likely caused by some indirect effect. In the rat, the most commonly identified nongenotoxic mode of carcinogenicity of the lower urinary tract is increased formation of urinary solids, which leads to chronic irritation, proliferation, and eventually, neoplasia (Clayson et al., 1995; Cohen et al., 2002; Cohen, 2005).

The presence of critical components of the indirect mode of urothelial carcinogenesis in the rat was demonstrated with the PPAR α/γ agonist muraglitazar (Dominick et al., 2006). Muraglitazar caused an increased incidence of urinary bladder tumors in male rats associated with changes in urine composition leading to formation of urinary solids, chronic damage, and resultant proliferative changes. Dietary acidification prevented the cytotoxic, proliferative, and tumorigenic responses. Ragaglitazar, another PPAR α/γ agonist that caused urinary bladder neoplasms in long-term rat studies, caused hypertrophy of the urothelial cells in short-term studies (Oleksiewicz et al., 2005). Apparent proliferation of the urothelium was limited to 1 male rat treated with ragaglitazar, but the number of animals tested was relatively small, and the duration was limited to 2 to 3 weeks. Hypertrophy of the urothelium was not reported with muraglitazar (Dominick et al., 2006).

Naveglitazar is a nonthiazolidinedione (non-TZD), γ-dominant PPAR α/γ dual agonist. In vitro, naveglitazar binds selectively to PPAR γ with high affinity (IC50 = 0.024 μM, Ki = 0.022 μM) and to PPAR α with lower affinity (IC50 = 1.71 μM, Ki = 1.66 μM; Reifel-Miller et al., 2003). Two-year rodent carcinogenicity studies were conducted to support the development of naveglitazar as a chronic-use therapeutic agent. Because of the known potential for neoplasms of the urinary bladder in rats treated with PPAR γ and α/γ dual agonists, a study to characterize changes in urine composition, the occurrence of urinary solids, and urothelial morphology and mitogenesis in rats was conducted at the same time as the 2-year carcinogenicity study in rats. The test system, doses, and dosing routes for the urinalysis study were the same as those used for the carcinogenicity study. Analyses of freshly voided urine samples and assessments of urothelial histology were done at periods throughout the study to characterize the temporal nature of any changes.

Materials and Methods

Test Animals and Treatment

F344/Ntac rats were obtained from Taconic Farms (Germantown, New York). Rats were housed individually in stainless steel, wire-bottom cages. Environmental controls for the animal room were set to maintain a temperature of 19 to 25ºC, a relative humidity of 30% to 70%, and a 12-hour light/ 12-hour dark cycle. The animals were offered certified rodent pellets (#5002C, PMI, Richmond, Indiana) and water ad libitum. Animals were randomly assigned to treatment groups based on weight. Rats (60/sex/group for the carcinogenicity study and 50/sex/group for the urinalysis/characterization study) were assigned to groups receiving 0, 0.3, 1.0, or 3.0 mg naveglitazar/ kg body weight/day (males) or 0, 0.1, 0.3, or 1.0 mg naveglitazar/ kg body weight/day (females) by oral gavage. These doses resulted in approximately equal systemic exposures in males and females of the respective low-, mid-, and high-dose groups (i.e., AUC_0-24hr in males given 3.0 mg/kg was similar to or slightly in excess of that in females given 1.0 mg/kg). All doses, including control, were administered at a dose volume of 1 ml/kg in a vehicle composed of 1.0% (w/v) carboxymethylcellulose sodium, 0.5% (w/v) sodium lauryl sulfate, 0.085% (w/v) Povidone, 0.05% (v/v) Dow Corning Antifoam 1510-US in reverse osmosis water. Dosing was conducted in the morning, generally within 4 hours of the beginning of the light cycle.

Carcinogenicity Study

The rodent carcinogenicity study had a duration of 104 weeks. All rats dying at unscheduled intervals or sacrificed at termination were necropsied, and a standard set of tissues was collected. All collected tissues, including urinary bladder and both kidneys from all rats, were examined histologically. The incidence of neoplasms was analyzed by a survival weighted statistical method with significance levels based on relative historical frequency of specific neoplasms (Lin, 1997; Lin and Rahman, 1998).

Urinalysis/Characterization Study

Urine Collection and Analysis

Freshly voided urine was collected from all surviving animals for acclimation to the collection process during weeks 2, 12, 25, 38, 51, and 64. Samples were discarded without analysis.

Urinalysis was conducted on freshly voided urine samples collected within 2 hours after the end of the dark cycle (a.m.) and within 1 to 3 hours before the beginning of the dark cycle (p.m.) during weeks 3, 13, 26, 39, 52, 65, and 78. Rats were not fasted before urine collection, and all a.m. urine samples were collected within 2 hours of the beginning of the light cycle so that they would be representative of urine composition of the dark cycle (Cohen et al., 2007). Urine samples collected a.m. were collected before dosing so that rats were not handled for at least 18 hours before urine collection. Rats were sampled until 10 animals provided a sample of at least 100 μl from each sex/ group; only those samples of at least 100 μl were analyzed. Urine pH was determined within 30 minutes of collection with a pH microelectrode. Urine samples of at least 100 but less than 200 μl were diluted 1:1 with distilled water after pH measurement and analyzed for urine chemistry. For urine samples of 200 μl or greater, 100 μl of urine was removed after pH measurement, diluted 1:1, and analyzed for urine chemistry; the remaining sample was centrifuged and the sediment examined microscopically. Urine chemistry analyses included determination of concentrations of creatinine, calcium, phosphorus, sodium, potassium, chloride, and magnesium. Ratios of calcium, phosphorus, sodium, potassium, chloride, and magnesium to creatinine were calculated.

Urine samples were collected a.m. and p.m. during week 40 specifically for light microscopic examination of urine sediments. Freshly voided urine was collected until 10 animals each provided a sample of at least 100 μl from each sex/group. Urine pH was determined within 30 minutes of collection for samples of at least 100 μl with a pH microelectrode. After determination of pH, urine samples were centrifuged and the sediment examined microscopically.

Urine samples were collected a.m. during week 74 for scanning electron microscopy evaluation of urine sediments. Freshly voided urine was collected from a sufficient number of animals to provide samples of at least 300 μl from 5 animals/sex/group. Within 1 hour of collection, urine was mixed to resuspend sediments, and each entire urine sample was drawn into a syringe and pushed through a 25-mm diameter 0.22 μm filter (Millipore Corporation, Billerica, Massachusetts) in a 25-mm filter holder (Structure Probe Inc., West Chester, Pennsylvania). A vacuum was applied to the underside of each filter holder to pull the remaining liquid through the filter. The filter holders were placed upright in a desiccator chamber with desiccant and allowed to air-dry for at least 1 week. Sections of filter from each control and high-dose male and female were excised with a razor, processed, and examined by scanning electron microscopy.

Freshly voided urine samples collected during the acclimation sampling in week 77 were analyzed for calcium concentrations in acidified and nonacidified subsamples. A subsample of each urine sample of sufficient volume (60 μl or greater) was acidified with 2 μl of concentrated hydrochloric acid before analysis. Urine samples were not centrifuged before subsampling or analysis.

Urinary Bladder Examination

Ten rats/sex/group were sacrificed during each of weeks 27, 53, and 79. Rats selected for sacrifice were given an intraperitoneal injection of 12 mg bromodeoxyuridine (BrdU, Sigma-Aldrich Corp., St. Louis, Missouri) approximately 24 hours before scheduled necropsy. The BrdU dosing solution was 20 mg/ ml in phosphate-buffered saline. At necropsy, rats were anesthetized with isoflurane, the abdominal cavity was opened, and the urinary bladder was collected before the animals were terminated by exsanguination. A clamp was placed on the urethra, and urine was withdrawn from the bladder. The urinary bladder was inflated with Bouin’s solution (week 27) or 10% neutral buffered formalin (weeks 53 and 79) until the serosal surface of the bladder was smooth. Formalin was used as a fixative for the later sampling times to provide adequate fixation for transmission electron microscopy (TEM). An additional clamp was placed on the neck of the bladder, and the ventral surface of the bladder was marked with a permanent tissue marker. The bladder and proximal urethra, with both clamps, were removed and placed in the same fixative as used for inflation. A section of duodenum from each animal was collected and placed in the same fixative as used for the bladder.

After 4 hours in fixative, each bladder was transected longitudinally through the ventral tissue mark. The right half of each bladder and respective duodenum were placed in 70% ethanol. The right half of the bladder was then sectioned into 3 longitudinal strips, processed, and embedded in paraffin with the respective section of duodenum. The left half of each bladder was processed for scanning and TEM. Bladder sections collected week 27 were transferred to 70% alcohol for processing, and those collected weeks 53 and 79 were transferred to modified Karnovsky’s fixative (2.5% glutaraldehyde/2% formaldehyde in 0.12M cacodylate buffer [with 0.5 g calcium chloride/liter]).

Four replicate sections of the paraffin block containing urinary bladder were made from each animal sacrificed during week 27. Two replicate sections of each block were placed on regular glass slides and stained with hematoxylin and eosin. Two additional replicate sections from each block were placed on positively charged glass slides for immunohistochemical labeling. Ten replicate sections of the paraffin block containing urinary bladder were made from each animal sacrificed during weeks 53 and 79. Two replicate sections of each block with urinary bladder were placed on regular glass slides and stained with hematoxylin and eosin. An additional 8 replicate sections from each block were placed on positively charged glass slides for immunohistochemical labeling.

One slide of urinary bladder and duodenum from each animal on study was immunolabeled for BrdU by standard procedures (Eldridge et al., 1990) and counterstained with hematoxylin. All slides from all animals from each scheduled sacrifice were stained together in a single staining run, and positive and negative controls were included in each staining run. Cell proliferation was quantified for each animal that had evidence of BrdU incorporation in the section of duodenum and if 3 complete sections of urinary bladder mucosa were present on the slide. The number of BrdU-labeled epithelial cells within the entire length of bladder epithelium from each of the 3 bladder sections present on the slide was counted for each rat. BrdU-labeled cells were not counted in focal proliferative lesions of epithelium that projected above and/or below the normal mucosal layer. The unit length labeling index (ULLI) was expressed as the total number of labeled cells in all 3 sections evaluated for each animal. The Student’s t-test (two-sided, unequal variance) was used to test for statistical significance in ULLI between control and treatment groups. A p value of less than or equal to .05 was judged to be statistically significant.

Immunolabeling for uroplakin and cytokeratins 5, 17, and 20 was conducted on sections of urinary bladder collected during weeks 53 and 79. Antibodies used were mouse monoclonal antiuroplakin (Research Diagnostics Inc., Flanders, New Jersey), polyclonal goat antihuman cytokeratin 5 (Santa Cruz Biotechnology Inc., Santa Cruz, California), monoclonal mouse antirat cytokeratin 17 (Sigma-Aldrich Corp., St. Louis, Missouri), and monoclonal mouse antihuman cytokeratin 20.8 (Signet Laboratories Inc., Dedham, Massachusetts). Antigen retrieval was conducted before labeling for cytokeratins 5 and 20, using Declere (Cell Marque Corp., Rocklin, California) in a pressure cooker. The targeted expression in the rat urothelium was superficial cells for uroplakin and cytokeratin 20, basal cells for cytokeratin 5, and basal/intermediate cells for cytokeratin 17 (Romih et al., 1998). The distribution and intensity of labeling were evaluated qualitatively in conjunction with evaluation of hematoxylin and eosin stained slides.

Bladder specimens for electron microscopy were rinsed with 2 changes of distilled water for 15 minutes each and transferred to fresh modified Karnovsky’s fixative. The urinary bladder from control and high-dose animals (3-mg/kg males, 1-mg/kg females) from the week 27, 53, and 79 sacrifices were processed for evaluation by scanning electron microscopy (SEM). The bladder from selected control and high-dose males and females from the week 53 and 79 sacrifices (6/sex/group and 5/sex/group, respectively) were evaluated by light microscopy of toluidine-blue–stained plastic-embedded sections and by TEM.

Results

Carcinogenicity Study

Survival of females to week 104 was 46%, 48%, 57%, and 39% for the control, low-dose, mid-dose, and high-dose groups, respectively. Survival of males was 10%, 27%, 19%, and 0%, respectively. Much of the mortality in males occurred late in the study; survival in males at week 80 was 82%, 80%, 74%, and 50%, respectively. The incidence of neoplasms of the urinary bladder was increased in naveglitazar-treated female rats and unchanged in male rats. The incidence of transitional cell carcinomas in the bladder was 0, 0, 2, and 8 (of 60 per group) in female rats treated with 0, 0.1, 0.3, and 1.0 mg/kg, respectively. The incidence of transitional cell papillomas in the bladder was 0, 0, 0, and 6, respectively, in the female groups. Approximately half of the urinary bladder neoplasms in females (0, 0, 1, and 6, respectively) were grossly visible masses at necropsy. The incidence of transitional cell papillomas and carcinomas in females in the 1.0 mg/kg group was statistically greater than that in controls. The incidence of hyperplasia of the urothelium of the bladder was 0, 3, 3, and 28 in the same groups, respectively. Transitional cell neoplasms were not increased in incidence in male rats, with incidences of 0, 1, 0, and 1 in groups treated with 0, 0.3, 1.0, and 3.0 mg/kg, respectively. Hyperplasia of the transitional epithelium of the urinary bladder occurred in 4, 4, 6, and 18 males in the same groups, respectively. Females had no increases in incidences of neoplasms of other tissues. Males in all treatment groups had an overall increase in multiple types of soft-tissue sarcomas. The incidences and descriptions of these neoplasms and of nonproliferative lesions in other tissues will be reported elsewhere.

Urine pH and Chemistry

Administration of naveglitazar had no substantive effects on urine pH (Figure 1A, 1B). A few statistically significant differences that occurred in urine pH were minimal and sporadic and had no consistent effects by sex, time, or direction. There were no apparent differences in urine pH between males and females. Normal diurnal variation for urine pH was exhibited by all groups; pH tended to be approximately 7 in the a.m. samples and slightly less than 6.5 in the p.m. samples.

Urine chemistry parameters affected by naveglitazar administration were concentrations of calcium and inorganic phosphorus. These effects were apparent after week 27. Urinary calcium concentrations were decreased in a poorly dose-responsive manner for naveglitazar-treated males and females in both a.m. and p.m. samples (Figure 2A, 2B). The differences in females were most prominent in the p.m. samples. Calcium concentrations in males tended to be lower than those in females in both control and naveglitazar-treated rats, and treatment-related decreases were more consistently significant in males. Urinary inorganic phosphorus concentrations were inconsistently lower for males and females of the high-dose groups (Figure 3A, 3B). The inorganic phosphorus concentrations showed a strong diurnal variation; levels in males and females were fairly similar. Urinary creatinine concentrations were not significantly changed by treatment, and urinary calcium-to-creatinine and phosphorus-to-creatinine concentrations were altered by compound similarly to the absolute calcium and phosphorus concentrations (data not shown).

Results obtained from urine samples collected during week 77 indicated that acidification of urine samples had no effect on measured calcium concentrations (Table 1). Absolute calcium concentrations in urine were similar to those measured at other times and showed similar treatment-related decreases.

Urine Sediment Examination

There were no apparent differences caused by treatment in urine sediments in samples examined by light microscopy during weeks 3, 13, 26, 39, 52, 65, and 78. Results for week 3 are shown in Table 2. The majority of crystals present were consistent morphologically with triple phosphates (MgNH₄PO₄; Osborne et al., 1990), varied from none to >20 per microscopic field (semiquantitative) in individual samples, and had no apparent relationship to treatment. Sediments morphologically consistent with amorphous phosphates (amorphous calcium phosphates) were variably present in fewer animals and had no apparent relationship to treatment. Sediments were examined microscopically from samples that were >200 μl. This resulted in examination of 4 to 10 samples/group for most collection times. The numbers examined at week 26 were lower than at other collection times (3 to 9 males/group, 2 to 7 females/group), prompting the more thorough collection and examination for sediments at week 40.

In samples collected during week 40, the majority of crystals present were morphologically consistent with triple phosphates. The presence and relative numbers (semiquantitative) of crystals varied from none to >20 per rat in each group. The relative numbers (based on semiquantitative scores) of triple phosphate crystals appeared to be increased slightly by dose in both a.m. and p.m. samples in males and in p.m. samples only in females (Table 3), compared to the concurrent controls. The numbers of triple phosphate crystals in treated rats in p.m. samples were similar to the respective control values in the a.m. samples. Sediments morphologically consistent with amorphous phosphates were present predominantly in females and appeared to be slightly decreased by treatment. Crystals consistent with calcium oxalate were observed in low numbers in only 1 sample, a p.m. sample from a male in the mid-dose group.

There were no qualitative morphologic differences in urine sediment composition between control and compound-treated rats by scanning electron microscopic examination at week 74 (Table 4). The urine from both groups contained slight to moderate amounts of crystals morphologically consistent with triple phosphates and lesser numbers of crystals consistent with calcium oxalate. There was moderate variation among individual rats in the amounts of triple phosphate crystals present. The males had slightly higher incidence and relative numbers of triple phosphate crystals than females, but there were no apparent differences between control and compound-treated rats within each gender. Amorphous material and unidentified crystals were present in both control and compound-treated rats, with no apparent relationship to treatment.

Histopathology

Microscopic findings in the urinary bladder for the urinalysis/ characterization study are listed in Tables 5, 6, and 7 for weeks 27, 53, and 79, respectively. The earliest and most consistent treatment-related change was hypertrophy of the urothelium, predominantly involving the superficial cells (Figure 4). These cells appeared more cuboidal, with increased amounts of eosinophilic cytoplasm. No particular location within the bladder was noted. The change affected animals in all compound-treated groups and generally increased in incidence and severity with dose.

Hyperplasia of the urothelium was present in only 1 high-dose–group male at the 27-week sampling time and 1 high-dose–group female at the 53-week sampling time. Both of these lesions were minimal simple hyperplasia (diagnostic criteria of Frith et al., 1995). Hyperplasia was more common at week 79, with hyperplasia evident in 2 of 10 males and 6 of 10 females in the high-dose groups. Hyperplasia in 1 female was graded minimal and was similar to the lesion at week 53. The remaining hyperplastic lesions at week 79 were graded as slight. Four of the 5 slight hyperplastic lesions in females were identified as focal, multifocal, and/or nodular. One of these 4 was identified as being near the ureter. The exact location within the bladder of the other 3 could not be determined from the orientation of the tissue sections, but it could be determined that 2 of these were not in the ventral portion of bladder. Neoplastic lesions present at 79 weeks were a papilloma in a high-dose female and a carcinoma in a high-dose male.

Immunolabeling of the urothelium for cytokeratin (CK) 20 was reduced in males of the mid- and high-dose groups at 53 weeks and in females of the mid- and high-dose groups at 53 and 79 weeks. Label for CK 20 in control animals generally appeared as a granular label in the apical portion of most superficial cells (Figure 4C). A few cells, often in small segments, were not labeled. In the treated rats with reduced label, more superficial cells, often in larger segments, were not labeled (Figure 4D). A few females in the mid- and high-dose groups at week 53 had no significant CK 20 immunolabel in the urothelium.

Immunolabeling for CK 17 in controls at 53 and 79 weeks was cytoplasmic, granular, and present on the basal and intermediate layers of the transitional epithelium (Figure 4E). Immunolabeling for CK 17 was not changed in naveglitazar-treated rats with hypertrophy of the urothelium (Figure 4F). Immunolabeling for CK 17 was reduced in 1 high-dose male and 4 high-dose females at week 79. The reduction was usually segmental and associated with hyperplastic areas. However, in some hyperplastic areas of urothelium, label for CK 17 was distributed normally on the basilar and intermediate cells.

There were no changes in immunolabeling for uroplakin or CK 5 in treated rats compared to controls. Immunolabeling for uroplakin in all rats was cytoplasmic, granular, and present on the apical transitional epithelium with occasional weak label in the intermediate layers. Immunolabeling for CK 5 was cytoplasmic, diffuse, and present throughout the transitional epithelium.

Cell Proliferation

The mean BrdU ULLI in urothelium of control rats tended to be higher in males than in females and to increase more in males with increasing age (comparisons not analyzed statistically, Table 8). There were apparent increasing numerical trends for ULLI in naveglitazar-treated males and females at all sampling times, but mean ULLI was increased based on statistical significance only in high-dose females at weeks 53 and 79. In these female groups, the ULLI was approximately 3-fold to 6-fold greater than concurrent controls, and 5 of 9 or 6 of 10 individual rats had an ULLI that was greater than that of the highest concurrent control value.

In contrast, no significant increases in mean ULLI were observed in naveglitazar treated males. However, the mean ULLI for high-dose males at 79 weeks was approximately 4-fold greater than that of concurrent controls, with 4 of 10 animals having a ULLI greater than that of the highest control value.

The distribution of labeled cells in most rats was apparently random across most of the 3 sections of bladder. In some rats with higher ULLI, the labeled cells were associated with 1 section of bladder, but the distribution still tended to be across most of the section (e.g., from anterior to posterior). BrdU labeled cells were noted (subjectively) as being common in focal areas of hyperplasia in the 7 high-dose rats with diagnoses of slight hyperplasia, but the BrdU-positive cells were not enumerated within proliferative lesions that appeared nodular and/or expansile. One of the high-dose males with a relatively high ULLI at 79 weeks was the rat with the focal proliferative lesion diagnosed as a carcinoma. The carcinoma had abundant BrdU-positive cells, as did the urothelium in sections of the bladder without the carcinoma, but few BrdU-positive cells were in the epithelium overlying the carcinoma.

Electron Microscopy

In scanning electron micrographs of the urothelial surface, control animals generally had cells that were flat with indistinct cell borders and limited surface ruffling (Figure 5A). The urothelial surface of compound-treated animals had bulging of individual cells into the lumen, more distinct cell boundaries, more apparent variability in cell size, and more ruffling of the surface (Figure 5B). These changes were generally diffuse in the sections examined; the specific location in the bladder of the sections was not identified. The changes in compound-treated animals were similar over time (6 to 18 months). Only 1 compound-treated animal, a female at 18 months, had a polypoid lesion consistent with a nodular proliferative lesion of the urothelium within the tissues examined by SEM. The exact location (e.g., dorsal vs. ventral) of this lesion within the bladder was not determined.

The evaluation of toluidine-blue–stained sections and TEMs indicated that the superficial urothelial cells of compound-treated animals were more cuboidal than those of controls. Superficial cells of both control and naveglitazar-treated rats had variable numbers of lysosomes, but compound-treated animals had relatively more cells with numerous lysosomes compared to controls (Figure 5C, 5D). Minor changes that were inconsistently present in the superficial urothelial cells of compound-treated animals included increased variability in the shape of fusiform vesicles and intermediate filaments that were less electron-dense and less well organized.

Discussion

Neoplasms of the urinary bladder and renal pelvis have been reported with numerous compounds with PPAR γ or PPAR α/γ dual agonist activity (El Hage, 2005a, 2005b; Cohen, 2005). The increased incidence of neoplasms in the urinary bladder associated with naveglitazar treatment occurred only in females, with an incidence of 23% in the high-dose group. In contrast, the incidence of urinary bladder tumors in male rats treated with muraglitazar was significantly increased in 3 of 4 compound-treated groups and was 58% in the high-dose group (Tannehill-Gregg et al., 2006). The incidence of urothelial tumors in naveglitazar-treated female rats in the mid-dose group (2/60) was not significantly different from controls. However, these tumors are normally rare in female F344 rats, with background incidences ≤0.2% for papillomas and ≤0.1% for carcinomas (Goodman et al., 1979; Haseman et al., 1990). The occurrence of the neoplasm in 2 rats, with an apparent dose-response incidence, in a known target tissue for the pharmacologic class indicates that the urothelial neoplasms in the mid-dose females may also have been compound related. The lack of development of urothelial neoplasms in male rats in this study may have been related to the low survival of male rats, especially those of the high-dose group. However, survival of the males was sufficient through week 80 to detect a strong carcinogenic effect in the bladder, had it been present.

In the rat, the most commonly identified indirect (non–compound specific) mechanism of carcinogenicity of the lower urinary tract is increased formation of urinary solids, which leads to chronic irritation, cell proliferation, and eventually, neoplasia (Clayson et al., 1995; Cohen et al., 2002; Cohen, 2005). Because of the previously described association of PPAR α/γ dual agonists with neoplasms of the urothelium, we investigated the potential for naveglitazar to cause changes in urinary composition, inflammation, and cell proliferation. In the current studies, the increase in cell proliferation in the urothelium identified in high-dose female rats at 12 and 18 months correlated with the overall sex and dose occurrence of urothelial neoplasms in the carcinogenicity study. However, we did not demonstrate any significant treatment-related differences in urine composition, especially crystalluria, which could be causally linked to the proliferative changes. Additionally, there were no treatment-related changes in urothelial morphology, such as erosions and inflammation as determined by light microscopy, which would suggest sediment-mediated injury to the urothelium. Since we did not examine the urothelium histologically during the first 6 months of treatment or complete a full evaluation of the urinary bladder mucosa by SEM at any time point, we can not completely exclude the possibility of crystalluria-like injury of the urothelium during the study. However, since the mechanism of action is based on chronic irritation, one would expect crystalluria and/or urothelial damage to be detected at some point in the study by the methods we used if the mechanism were causative.

In the naveglitazar-treated rats with increased proliferation in the urothelium, there was no apparent predilection for a specific site in the bladder for the increase in cell counts or hyperplastic lesions, suggesting a diffuse stimulus rather than a localized stimulus secondary to injury. While the blocking scheme used for sections of bladder did not necessarily maintain the dorsal-ventral orientation of the sections, it did ensure that both dorsal and ventral sections were examined. Since no specific lesions indicative of irritation were noted by light microscopy and the BrdU-labeled cells were generally randomly distributed, a causal relationship to increased crystalluria could not be made. In this study, we expressed the BrdU labeling index as the total number of labeled cells in all 3 sections of bladder examined per animal and did not normalize the labeling index to the total number of cells, as has been recommended for urinary bladder mucosa (Cohen et al., 2007). Since the sections were uniformly prepared across all animals examined, labeling index was normalized to a unit length, which was the total length of urothelium examined for each animal. We evaluated the entire length of urothelial mucosa because of the very low background rate of labeled cells in the urothelium in many animals. The ULLI method is not influenced by changes in the total cell population, such as hyperplasia (Monticello et al., 1990), and thus gives a better indication of total number of replicating cells in a tissue. However, results could be influenced by differences in tissue preparation in an organ such as the urinary bladder, which could complicate interpretation of apparent effects in individual animals. Despite the limitations of the methods used, the statistically significant results for increased cell proliferation and neoplasms of the urothelium of the bladder both occurred in females of the high-dose group.

The earliest morphologic change in the urothelium identified in this study was hypertrophy, primarily of the superficial urothelial cells. This change, which occurred in both male and female rats in our studies, is consistent with early changes in the urothelium described with ragaglitazar, another PPAR α/γ dual agonist, in male Sprague-Dawley rats (Oleksiewicz et al., 2005). The urothelial hypertrophy with ragaglitazar was characterized as increased cell size by flow cytometry and increased protein content by protein/DNA measurements, so it was not an artifact of histologic preparation (e.g., differences in degree of distention of the bladder). Hypertrophy of the urothelium has been associated with regenerative responses in the urothelium (Molon-Noblot et al., 1992) and with mechanisms not related to proliferative events (e.g., alterations in urine volume; Cohen, 2005). In the current study, there were no treatment-related lesions of urothelial damage and no changes in urinary concentrations of creatinine to suggest changes in urine volume. Studies of urothelium from anuric human patients suggest that a lack of urine-derived factors, including mechanical factors, does not appear to modulate differentiation in normal urothelium (Stahlschmidt et al., 2005), but this does not rule out the possibility that abnormal urine composition or mechanical factors could modulate urothelial differentiation in the rat. Regardless of the potential mechanisms causing the hypertrophy, it occurred in both male and female rats and so did not correlate with the increased incidence of urothelial neoplasms that was limited to females.

We did not investigate the potential for decreased apoptosis as a mechanism of carcinogenesis in the urothelium or as a potential cause of the hypertrophy of the superficial cells. The rate of cell turnover in the normal urothelium is very low (Martin, 1972), and a decreased rate of apoptosis would be extremely difficult to quantify. In a study of the effects of muraglitazar on the urothelium in rats, there were no apparent treatment-related changes in apoptotic rate in the urothelium (Dominick et al., 2006). The reported apoptotic rate in control and treated rats in the muraglitazar study was <0.1%, which would equate to <1 apoptotic cell per the minimum of 500 cells counted per animal. A decreased rate of apoptosis could not reliably or readily be measured against this very low background rate.

In our studies with naveglitazar, we looked at expression of cytokeratins and uroplakin as potential markers of cell differentiation. The only change in these markers associated with naveglitazar treatment was decreased CK 20 expression in superficial cells. This decrease in label occurred in both males and females but appeared to be more severe and occurred in more sampling times in females compared to males. CK 20 is normally expressed in terminally differentiated superficial urothelial cells and appears after uroplakin is expressed in these cells (Veranic̆ et al., 2004; Harnden et al., 1995; Harnden et al., 1996). This decreased labeling for CK 20 could suggest an alteration in terminal differentiation or increased cell turnover of the superficial cells, but a relationship to possible changes in relative age of the cells is not known. Alterations in expression of CK 20 in the urothelium have been related to some forms of urothelial dysplasia in humans, but the dysplasia was associated with increased labeling for CK 20 (Harnden et al., 1996). In contrast, expression of CK 20 is associated with low malignancy potential in some urothelial neoplasms (Harnden et al., 1995; Harnden et al., 1999). The decreased labeling for CK 20 in rats in the current study may have been related to the minor differences in organization of terminal filaments observed by electron microscopy but was, overall, a change with no apparent relationship to the occurrence of proliferative changes in the urothelium. Labeling with CK 17, a marker for basal and intermediate cells, was reduced in just a few animals, usually in association with hyperplastic areas. This suggested an alteration of differentiation of cells in these overtly proliferative areas but offered no suggestion for changes in differentiation that might precede the proliferative changes. The lack of changes in immunolabeling for uroplakin in treated rats compared to controls suggests that the structure of the apical limiting membrane was not altered and that the barrier function of the urothelium was intact.

Urothelial cells have potential for rapid cell turnover in response to growth factors that are normally present in the urine. However, this normally occurs only in response to breaches in the surface limiting membrane and resultant exposure of the basilar cells to components of the urine (Messing et al., 1987; Messing, 1992). It is possible but unlikely that the hypertrophy of the superficial urothelial cells associated with naveglitazar treatment was associated with a loss of integrity that allowed increased exposure of the basilar cells to growth factors in the urine. The superficial cells appeared to have a normal expression of uroplakin, the protein associated with the asymmetric unit membrane of the superficial urothelial cells. Additionally, proliferative changes in the urothelium were rarely observed at the earlier sampling periods when hypertrophy of the urothelium was already established. A possible explanation for increased cell proliferation that we did not investigate is an increased expression of growth factor receptors in the urothelium. Transgenic mice with increased expression of EGFR have increased cell proliferation in the urothelium (Cheng et al., 2002).

We did not demonstrate any substantive differences in urinary precipitates between control and naveglitazar-treated rats. Urinary sediments observed in the routine urine examination at quarterly intervals were present with some variability but showed no apparent treatment-related effects. There was an apparent minor increase in incidence of triple phosphates in urine of naveglitazar-treated females, examined in the afternoon at week 40. However, the relative incidences in triple phosphate crystals in the treated females were still within the incidences seen at other times in controls. This minor difference in relative crystal numbers may have been related to differences in crystal formation, since control rats appeared to have a higher incidence of amorphous phosphates at this time. The overall sediment load was similar between control and treated females. Examination of filtrates of entire urine samples by SEM at 74 weeks showed no apparent treatment-related differences in incidence or types of urinary solids. Additionally, analysis of acidified and nonacidified urine samples at 77 weeks suggested that decreased urinary calcium concentrations were not caused by precipitation of calcium salts in the urine. As a critical part of determining the mode of action of urothelial carcinogenesis, light microscopic examination of the urothelium showed no evidence of responses to irritation (e.g., erosion and inflammation) that would be expected with excessive amounts of urinary sediments.

A limitation of this urinary characterization study is that it was 18 months in duration, and alterations in urinary parameters may have occurred after this time. However, naveglitazar-treated female rats had increased incidences of hyperplastic lesions and increased cell proliferation in the urothelium by 18 months, indicating that the causative factors for the proliferative events were present before this time. Another potential limitation was the timing of collection of urine, because of the strong diurnal alterations in urine composition in rats. Although we did not collect urine during the actual ‘dark’ phase in our study, our a.m. collection time was within 2 hours of lights on. This is within the time during which urine composition is still reflective of the night phase in the rat (Cohen et al., 2007). Additionally, the expected diurnal changes in pH were demonstrated to support the adequacy of sampling.

Decreased urinary calcium concentrations have been linked to increased rates of cell proliferation in urothelium (Reese and Friedman, 1978). These changes, however, were demonstrated in cell and organ cultures, in which the basilar urothelial cells could be in contact with the culture medium. In contrast, urothelial cells in vivo are segregated from the urine by the asymmetric unit membrane of the superficial cells and the tight junctions between the superficial cells. While urinary calcium concentrations were decreased in naveglitazar-treated rats, they were still higher than the calcium concentrations (0.3 mM) associated with the increased rates of urothelial cell proliferation in vitro. Additionally, urinary calcium concentrations were lowered by treatment in both male and female rats and were generally lower overall in male rats. Thus, there was no obvious relationship between decreased urinary calcium concentrations, which occurred in both sexes, and the increased incidence of urinary bladder neoplasms, which occurred only in females.

We did not detect any substantive treatment-related changes in urinary pH in the current study. However, we did demonstrate the expected diurnal variation in pH to ensure that we were adequately testing for this parameter. The urinary pH in the current study was near the breakpoint (pH ≥ 6.5) that has been identified as supporting the precipitation of calcium and magnesium salts in the urine of rats (Cohen, 2005). The urinary pH in both males and females was approximately 7 in the morning but was slightly below 6.5 in the afternoon. Conditions conducive to formation of calcium and magnesium salts may have been present during the dark phase but may not have been sustained during the entire day. Additionally, there were no apparent sex-related differences in urinary pH that could support the sexual dimorphism in induction of urothelial neoplasms. An additional study in female rats treated with naveglitazar and dietary acidification might answer the question of whether reductions in urinary pH and urinary solids would have prevented urothelial carcinogenesis. Dietary acidification of male rats treated with muraglitazar inhibited the formation of urinary solids and prevented urothelial carcinogenesis (Dominick et al., 2006). However, since we did not demonstrate crystalluria with naveglitazar, a dietary acidification study was considered unwarranted.

We did not investigate all known potential mechanistic steps that have been related to indirect causes of urinary bladder carcinogenesis in rats. Alterations in urinary citrate and oxalate levels were demonstrated as being key events associated with increased urinary precipitates in the urine of rats treated with muraglitazar (Dominick et al., 2006). Citrate measurements in our study might have added to the body of information for changes associated with PPAR agonists as a class. However, without demonstration of crystalluria for naveglitazar, such data would not have proven or disproven causation. Alterations in multiple components of urine, singly or in relationship to each other, can lead to precipitation of salts in urine (Cohen, 2005), and changes in 1 component may be offset by changes in other components relative to potential for crystalluria.

In the current study, there was no evidence of a direct or indirect compound-related cytotoxic effect on the urothelium. Therefore, proliferative effects independent of cytotoxic effects should be considered. Examples have been presented in which experimental treatments in rats have led to increased cell proliferation in the absence of apparent urothelial damage (Cohen et al., 1994). Possible causes for such effects include alterations in growth factors in the urine, relative expression of growth-factor receptors on the urothelium, and alterations of cell-cycle control in urothelial cells. Increased amounts of growth factors in the urine are an unlikely cause because normal rat urine contains ample amounts of epithelial growth factor to induce increased cell proliferation. The normal superficial urothelial barrier prevents the urinary growth factors from reaching the basilar urothelial cells. Direct effects of PPAR agonists on the urothelium should be considered. All 3 PPAR subtypes (α, γ, and δ) are expressed in the urinary system, and expression of PPAR γ is greatest in the urothelium (Guan et al., 1997). While the activity of PPAR γ agonists on urothelium in vitro is generally antiproliferative or cytostatic (Grommes et al., 2004; Spencer et al., 2004; Kawakami et al., 2002), the possibility exists that PPAR agonists could cause other changes that could secondarily lead to urothelial proliferation. Simultaneous activation of both PPAR α and γ has been shown to induce expression of genes in the urothelium that could be related to cell-cycle control (Egerod et al., 2005). PPAR agonists may also act through ligand-independent pathways, as shown by activation of mitogen-activated protein-kinase signaling of EGFR transactivation (Gardner et al., 2003). Additionally, PPAR agonists may cause localized changes in expression of other growth factors, such as VEGF (Fauconnet et al., 2002).

In summary, the current study could not identify a specific mechanistic cause for the increased cell proliferation and neoplasms of the urothelium associated with naveglitazar treatment of female rats.