Peroxisome Proliferator-Activated Receptor γ and Growth Inhibition by Its Ligands in Uterine Endometrial Carcinoma

Peroxisome proliferator-activated receptor (PPAR) is a member of the nuclear hormone receptor superfamily of transcription factors. PPARs function as transactivation factors following heterodimerization with retinoid X receptors (RXR) and bind to its specific response elements, termed peroxisome proliferator-responsive elements, of various target genes (1). PPARs have a subfamily of three different isoforms: PPARα, PPARβ/δ, and PPARγ. All isoforms play important roles in the regulation of several metabolic pathways, including lipid biosynthesis and glucose metabolism. There are distinctive differences in the distribution of these receptor subtypes in humans. PPARα has been detected mainly in the myocardium, kidney, and liver; PPARβ/δ is expressed in most tissues; and PPARγ has been found primary in adipocytes and immune cells (2, 3).

PPARγ plays important roles in the regulation of lipid homeostasis, adipogenesis, insulin resistance, and development of various organs (4–6). Naturally occurring PPARγ ligand, a 15-deoxy-Δ^12,14-prostaglandin J₂ (15d-PGJ₂), activates PPARγ at micromolar concentrations in human in vivo (7–9). Synthetic PPARγ ligands are known as thiazolidinediones, including troglitazone, rosiglitazone, and pioglitazone. These synthetic ligands have been used for the treatment of insulin resistance in type II diabetes mellitus. In addition, thiazolidinediones have been proposed in differentiation-mediated therapy of various human carcinomas associated with high levels of PPARγ (10).

Various in vitro studies showed that PPARγ ligands have a potent antiproliferative activity for a wide variety of neoplastic cells (11). PPARγ agonist was reported to inhibit the proliferation of carcinoma cells, and phase II clinical trials using PPARγ ligands have recently been done as a novel therapy for advanced patients with breast carcinoma, histologically confirmed prostate carcinoma, liposarcoma, and metastatic colon carcinoma (12). The outcomes of these trials are, however, still controversial. Mueller et al. reported that one advanced prostate carcinoma patient had a dramatic decrease in serum prostate-specific antigen as a result of troglitazone treatment (13), but Burstein et al. reported troglitazone had little apparent clinical value among patients with treatment-refractory metastatic breast carcinoma (14). In addition, Debrock et al. reported no significant changes in the histologic appearance of the liposarcomas following treatment with rosiglitazone (15).

The expression and effectiveness of PPARγ has been extensively studied in breast, prostate, and colon carcinoma, but little is known about PPARγ in uterine endometrial carcinoma. In addition, obesity, excess estrogen, type II diabetes, and hypertension are some of the important risk factors of endometrial carcinoma (16–19), but the effects of PPARγ agonists to endometrial carcinoma are largely unknown. Therefore, in this study, we first examined the expression of PPARγ in endometrial carcinoma, correlated the findings, including the status of progesterone receptor, estrogen receptor α (ERα), ERβ, Ki-67, and p21 and the body mass index (BMI) of the patients, and examined the growth inhibition of the carcinomas by PPARγ agonists in vitro to clarify the possible biological and clinical roles of PPARγ in human endometrial malignancy.

Patients and tissues. All patients were informed of the purpose of this study, and informed consent was obtained before participation. The research protocol was approved by the Ethics Committee of Tohoku University Graduate School of Medicine. Uterine endometrial carcinoma tissues (49 well differentiated, 32 moderately differentiated, and 22 poorly differentiated; 66 stage I, 12 stage II, 22 stage III, and 3 stage IV) were obtained from 103 patients who underwent surgery at the Department of Obstetrics and Gynecology, Tohoku University Hospital (Sendai, Japan) from 1993 to 2004. All the subjects were Japanese. All endometrial carcinoma specimens were obtained after hysterectomy. Median follow-up time of the patients examined in this study was 60 months (range, 2-148 months). Disease-free survival and overall survival were calculated from the time of initial surgery to recurrence and/or death or the date of last contact. Survival times of patients still alive or lost to follow-up were censored in December 2004. Information regarding age, BMI, stage, grade, primary surgery, postoperative therapy and recurrence, and complications were all retrieved by a review of the chart. BMI was calculated by dividing weight in kilograms by the height in meters squared. We defined obesity in these patients as a BMI ≥ 25 (20). A standard primary treatment for endometrial carcinoma at Tohoku University Hospital from 1993 to 2004 was total abdominal hysterectomy, salpingo-oopholectomy, pelvic and/or para-aortic lymphadenectomy, and peritoneal washing cytology. A total of 85 of 103 (83%) patients underwent complete surgery. Six of the 85 patients had lymph node metastasis. The remaining 18 (17%) patients underwent total abdominal hysterectomy and salpingo-oopholectomy without lymphadenectomy because of obesity and/or poor performance status. None of these patients had received preoperative chemotherapy and/or hormonal therapy or pelvic irradiation. None of the patients used oral contraceptives. The lesions were classified according to the Histological Typing of Female Genital Tract Tumors by WHO and staged according to the International Federation of Gynecology and Obstetrics system (21, 22). About 60% of 103 patients received pelvic radiation therapy (50 Gy) or three to six courses of chemotherapy consisting of the cisplatin-based combination regimen CAP (cisplatin 60-70 mg/m², doxorubicin 40 mg/m², and cyclophosphamide 500 mg/body weight) after operation. Patients who had early-stage and low-grade disease (stage IA, grade 1, stage IA, grade 2, and stage IB, grade 1) and patients who were associated with poor performance status did not receive any adjuvant therapy. All of the archival specimens were retrieved from the surgical pathology files at Tohoku University Hospital.

Normal endometrial tissues were obtained from unaffected endometrium of normally menstruating females who underwent hysterectomy for cervical carcinoma in situ. Biopsy specimens from 23 cases were in the proliferative phase, and 19 cases were in the secretory phase. Fourteen had simple hyperplasia, 9 had complex hyperplasia with atypia, and 8 had complex hyperplasia without atypia. All specimens were routinely processed (i.e., 10% formalin fixed for 24-48 hours), paraffin embedded, and thin sectioned (3 μm).

Antibodies. Monoclonal antibodies for PPARγ were purchased from Perseus Proteomics Co. Ltd. (Tokyo, Japan). This antibody recognizes both human PPARγ1 and PPARγ2. Monoclonal antibody for β-actin was purchased from Sigma-Aldrich Co. (St. Louis, MO), monoclonal antibody for progesterone receptor was purchased from Chemicon International, Inc. (Temecula, CA), monoclonal antibodies for ERα and Ki-67 were purchased from Dako Cytomation Co. Ltd. (Glostrup, Denmark), monoclonal antibody was purchased for ERβ from Gene Tex, Inc. (San Antonio, TX), and monoclonal antibody was purchased for p21 from Pharmingen (San Diego, CA).

Immunohistochemistry. A Histofine kit (Nichirei, Tokyo, Japan), which employs the streptavidin-biotin amplification method, was used in this study. Antigen retrieval for PPARγ immunostaining was done by heating the slides in an autoclave at 120°C for 5 minutes (p21; 15 minutes) in citric acid buffer [2 mmol/L citric acid, 9 mmol/L trisodium citrate dehydrate (pH 6.0)]. Dilution of antibodies used in this study was as follows: 1:300 for PPARγ, 1:40 for progesterone receptor, 1:50 for ERα, 1:1,500 for ERβ, 1:50 for Ki-67, and 1:250 for p21. The antigen-antibody complex was visualized with 3,3′-diaminobenzidine solution [1 mmol/L 3,3′-diaminobenzidine, 50 mmol/L Tris-HCl (pH 7.6), 0.006% H₂O₂] and counterstained with hematoxylin. As a positive control for immunohistochemistry of PPARγ, normal mammary glands and omentum tissues were employed in this study. Immunohistochemical staining was evaluated by two independent observers (K.O. and T.S.) without knowledge of clinical outcomes. PPARγ immunoreactivity was detected in the nucleus, and the immunoreactivity was quantitated as the labeling index (LI). Five hundred carcinoma cells in each field were selected and the LI was determined as the percentage of positive cells per 500 carcinoma cells. For each tissue section, at least three fields were photographed under light microscopy at ×200. Cases with a PPARγ LI of >10% were considered PPARγ-positive endometrial carcinoma in this study. Interobserver and intraobserver differences were <5% in our present study.

Cell lines and media. Uterine endometrial carcinoma cell lines, Ishikawa 3-H-12, Sawano, and RL95-2, were cultured in phenol red–free RPMI 1640 (Sigma-Aldrich) supplemented with 10% fetal bovine serum (JRH Biosciences, Lenexa, KS) and 1% penicillin/streptomycin. Ishikawa cells were originally established from well-differentiated human endometrial carcinoma (23). Sawano cells were naturally raised cisplatin-resistant cells (24). RL95-2 cells are ER-positive endometrial carcinoma cell line.

Real-time PCR. Total RNA from tissues, Ishikawa, Sawano, and RL95-2, were extracted using TRIzol reagent (Invitrogen Life Technologies, Inc., Gaithersburg, MD), and reverse transcription kit [SuperScript II Preamplification System (Gibco-BRL Invitrogen Co., Carlsbad, CA)] was used in the synthesis of cDNA. The LightCycler System (Roche Diagnostics GmbH, Mannheim, Germany) was used to semiquantify the mRNA expression levels using real-time PCR. The primer sequences used in this study were as follows: PPARγ 5′-ATGCTGGCCTCCTTGATGAATAA-3′ and 5′-AGGGCTTGTAGCAGGTTGTCTTG-3′ (266 bp), RXRα 5′-TTCGCTATGCTCTCAG-3′ and 5′-ATAAGGAAGGTGTCAATGGG-3′ (113 bp), RXRβ 5′-GAAGCTCAGGCAAACACTAC-3′ and 5′-TCTTTGTTGTCCC-3′ (111 bp), RXRγ 5′-GCAGTTCAGAGGACATCAAGCC-3′ and 5′-GCCTCACTCTCAGCTCGCTCTC-3′ (352 bp), ribosomal protein L 13a 5′-CCTGGAGGAGAAGAGGAAAGAGA-3′ and 5′-TTGAGGACCTCTGTGTATTTGTCAA-3′ (125 bp), and p21 5′-GGAAGACCATGTGGACCTGT-3′ and 5′-GGATTAGGGCTTCCTCTTGG-3′ (178 bp). Settings for the PCR thermal profile were initial denaturation at 95°C for 1 minute followed by 40 amplification cycles of 95°C for 1 second, annealing at 58°C (RXRα), 62°C (PPARγ), 60°C (RXRβ and p21), 68°C (ribosomal protein L 13a and RXRγ) for 15 seconds, and extension at 72°C for 15 seconds. To verify amplification of the correct sequences, PCR products were purified and subjected to direct sequencing. Negative control experiments lacked cDNA substrate to check for the possibility of exogenous contaminant DNA. Fat tissue cDNA was used as a positive control. The mRNA levels were summarized as the ratio of ribosomal protein L 13a and subsequently evaluated as a ratio (%) compared with that of controls. The PCR products were separated electrophoretically on a 2% agarose gel and stained with ethidium bromide.

Immunoblotting. Cells were grown to 70% confluence in 10-cm plates, and after removal of culture medium with PBS, nuclear protein was extracted by conventional method. The protein concentration was measured by a Model 680 microplate reader (Bio-Rad, Hercules, CA) using Bradford reagent (Bio-Rad). In all, 60 μg nuclear protein of each sample was mixed with an equal volume of 2× concentrated SDS-PAGE sample buffer, boiled, and then transferred to nitrocellulose membrane (Hybond polyvinylidene difluoride, Bio-Rad). The membranes were incubated in blocking solution (PBS containing 3% nonfat milk and 0.1% Tween 20) and then incubated in 1:500 dilution of PPARγ antibody in PBS (containing 0.1% Tween 20) overnight at 4°C. After incubation with horseradish peroxidase–labeled anti-mouse IgM (Santa Cruz Biotechnology, Santa Cruz, CA), the antigen-antibody complex was visualized with enhanced chemiluminescence system (Amersham, Freiburg, Germany). β-Actin (Sigma-Aldrich) was used as internal positive control, and the dilution was 1:1,000.

Cell proliferation assay and apoptosis analysis. Cell proliferation was assessed by the Cell Counting Kit-8 (Wako Pure Chemical Industries, Osaka, Japan). We also examined the status of apoptosis in Ishikawa, Sawano, and RL95-2 cells using an apoptosis screening kit (Wako), which employed a modified terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling method (25). Ishikawa, Sawano, and RL95-2 cells were plated at 5 × 10³ per well in 96-well plates and cultured with phenol red–free RPMI 1640 containing 10% dextran-coated, charcoal-stripped fetal bovine serum at 37°C for 2 days in a CO₂ incubator. 15d-PGJ₂ was purchased from Biomol Research Laboratories, Inc. (Butler Pike, PA), dissolved in ethanol, and added to the wells (day 0), and the assay was done on days 0, 1, 3, and 5. Control cells were treated with vehicle only. For cell proliferation assay, WST-1 solution was added to each well of the plates and incubated at 37°C for 2 hours. The absorbance at 450 nm was measured in a microplate reader SpectraMax 190 (Molecular Devices Corp., Osaka, Japan). The cell number and apoptosis index were calculated according to the following equation: (cell absorbance value after test materials treated / vehicle control cell absorbance value) and were subsequently evaluated as a ratio (%) compared with that of controls.

Statistical analysis. The statistical analysis was done using StatView 5.0 (SAS Institute, Inc.) software. An association between immunoreactivity for PPARγ and clinicopathologic factors, results of cell proliferation assay, and results of real-time PCR were evaluated using Bonferroni/Dunn test, Mann-Whitney U test, or Wilcoxon test. Overall and disease-free survival curves were generated according to the Kaplan-Meier method and the statistical significance was calculated using the log-rank test. Results were considered significant when the P < 0.05.

Complications of endometrial carcinoma patients. The median BMI for carcinoma patients was 24.6 (range, 17.3-39.6). BMI ≥ 25 was 45 (44.6%) of endometrial carcinoma patients examined in our present study (Table 1). The number of patients with hypertension was 56 (55.4%) and the number of patients with type II diabetes was 24 (23.8%). A significant association was detected between obesity and disease-free survival (P = 0.0353; Fig. 1), but there were no significant correlations between obesity and overall survival of the patients (P = 0.0844; data not shown).

Immunohistochemistry and correlation with clinicopathologic variables. Positive immunoreactivity for PPARγ was detected in the nuclei of carcinoma cells, normal endometrial gland cells, and hyperplastic cells (Fig. 2A and C-E). PPARγ immunoreactivity was detected in 67 of 103 (65%) of carcinoma cases, 11 of 23 (48%) of proliferative-phase endometrium, 14 of 19 (74%) of secretory-phase endometrium, and 27 of 32 (84%) of endometrial hyperplasia. Immunoreactivity of PPARγ was significantly lower in endometrial carcinoma than in secretory-phase endometrium (P < 0.0001) and endometrial hyperplasia (P = 0.0015). Lower immunoreactivity was also detected in proliferative endometrium than in secretory-phase endometrium (P = 0.0163) and endometrial hyperplasia (P = 0.0008; Fig. 3A). Results of immunohistochemistry for PPARγ and their correlation with clinicopathologic variables, such as age, stage, and grade of the patients, are summarized in Table 2. PPARγ LI was significantly correlated with p21 LI (P < 0.0001). BMI in endometrial carcinoma patients was significantly associated with PPARγ LI of their carcinoma cells (P < 0.0001). No significant associations were detected between PPARγ immunoreactivity and overall survival (P = 0.6367) or disease-free survival of the patients (P = 0.1168; data not shown).

Real-time PCR analysis. We examined the expression of PPARγ mRNA in 28 endometrial carcinoma patients and 14 normal controls. PPARγ mRNA was abundantly expressed in secretory endometrium. There were significant differences between endometrial carcinoma and proliferative-phase tissues (P = 0.0115) or secretory-phase tissues (P = 0.0217; Fig. 3B), but no significant differences were detected between proliferative-phase and secretory-phase endometrium. In carcinoma cell lines, PPARγ mRNA was also expressed in Ishikawa, Sawano, and RL95-2 cells (Fig. 4A). It is well known that PPARγ heterodimerizes with RXRs, and we therefore examined the expression of RXRs-RXRα, β, and γ-mRNAs in these cells. These mRNAs were also expressed in Ishikawa, Sawano, and RL95-2 cells (data not shown).

Immunoblotting. Immunoblotting analysis was done using Ishikawa, Sawano, RL95-2, and MCF-7 cells to detect the presence of PPARγ protein. MCF-7 cells were used as a positive control. Immunoreactive bands corresponding to PPARγ were detected in all carcinoma cell lines (Fig. 4B).

Cell proliferation assay and apoptosis analysis. We examined the effects of naturally occurring PPARγ ligand on the cell growth in vitro (Fig. 5A-C). 15d-PGJ₂ markedly suppressed cell growth in Ishikawa, Sawano, and RL95-2 cells in both dose- and time-dependent manners. In all three cell lines, growth suppression was detected after 3 days of the 15d-PGJ₂ addition.

We also examined the status of apoptosis after adding 15d-PGJ₂. Apoptosis indexes of Ishikawa, Sawano, RL95-2 cells were not significantly altered under 10 and 20 μmol/L 15d-PGJ₂ treatments for 5 days (Fig. 6A-C).

The expression of p21 mRNA after 15d-PGJ₂ treatment was summarized in Fig. 7A to C. The p21 mRNA expression increased in both dose- and time-dependent manners.

In our present study, weak PPARγ immunoreactivity was detected in endometrial carcinoma tissues. PPARγ mRNA expression in carcinoma tissues was also lower than that in normal tissues. Similar results have also been reported in esophageal (26) and lung (27) carcinomas. Normal human ureter also expressed PPARγ protein, but there was a significant loss of PPARγ expression in high-grade transitional cell carcinomas (28). Therefore, the results of our study are also consistent with these results, which support the hypothesis that the PPARγ gene is a tumor suppressor gene, and dysfunction of PPARγ contributes to tumorigenesis (11). However, several studies also showed that PPARγ expression was more marked in carcinoma tissues than in normal tissues in several types of human malignancies. For instance, Zhang et al. analyzed 56 specimens of normal ovary and neoplasm using immunohistochemistry (29). Immunoreactive PPARγ was not detected in normal ovaries. However, PPARγ immunoreactivity in ovarian tumor tissues was significantly higher than in normal ovaries and benign ovarian tumors (29). Ikezoe et al. examined the expression of PPARγ in 339 clinical samples and 71 various cancer cell lines, including colon cancer, breast cancer, prostate cancer, lung cancer, osteosarcoma, glioblastoma, and leukemia. All of the cell lines and clinical samples expressed PPARγ as detected by real-time PCR and/or Western blot, but their expression levels varied widely among samples (30). Therefore, the results above indicated that the expression of PPARγ is dependent on tissue specificity and/or the mutational events that are required for cancer development.

We showed that PPARγ LI was higher in endometrium in the secretory phase than the proliferative phase. We examined previously the expression of RXRs in normal endometrium, endometrial hyperplasia, and endometrial carcinoma (31). RXRγ immunoreactivity was detected in the nuclei of epithelial cells of the secretory-phase endometrium but not of the proliferative phase. Loughney et al. also reported that intracellular concentration of all-trans retinoic acid, a ligand of RXRs, was elevated during the secretory phase because of a marked reduction of cellular retinoic acid protein type II mRNA (32). PPARγ was shown to heterodimerize RXRs, which is also consistent with the results of our study. PPARγ may also have antiproliferative effects in secretory-phase endometrium.

In in vitro experiments, 15d-PGJ₂ markedly inhibited cell proliferation of the endometrial carcinoma cell lines at 10 and 20 μmol/L. 15d-PGJ₂ is both an endogenous PPARγ ligand and a direct inhibitor of several other signal transduction pathways (33). 15d-PGJ₂ has been considered an endogenous ligand; Forman et al. reported that prostaglandin D₂ is the major prostaglandin in most tissues and PGJ₂ derivatives may be produced at several of these sites (8). Nosjean et al. also showed that prostaglandins were cyclooxygenase products, and a final product of this pathway, PGJ₂, is nonenzymatically converted into 15d-PGJ₂ (7). Parikh et al. reported the following three findings. (a) 15d-PGJ₂ activated PPARγ-dependent signaling systems more potently than other fatty acids that have been studied. (b) A representative synthetic cyclopentenone-prostaglandin, Δ¹²-PGJ₂, was actively transported into the nucleus in a time- and temperature-dependent manner with Michaelis-Menten kinetics, suggestive of carrier-mediated active transport. Finally, formation of immunoreactive 15d-PGJ₂ has been detected during the propagation and resolution of inflammation in association with PPARγ activation (34). Nakashiro et al. reported that low-dose 15d-PGJ₂ (0.5 μmol/L) almost completely inhibited the growth of nonneoplastic human urothelial cell line and 10 μmol/L 15d-PGJ₂ suppressed the growth of neoplastic urothelial cells (28). Several investigators also showed that 15d-PGJ₂ inhibited the growth of pancreatic carcinoma cells (35), lung carcinoma cells (36, 37), and gastric carcinoma cells (38). These antiproliferative effects of 15d-PGJ₂ discussed above are all considered to be mediated by several pathways: PPARγ dependent and PPARγ independent. The lipopolysaccharide-induced transcription responses of activator protein-1, nuclear factor-κB, and STAT1 can be repressed by 15d-PGJ₂ only in the presence of PPARγ (39). Two identified candidates have been proposed to mediate PPARγ-independent actions of 15d-PGJ₂: the nuclear factor-κB system and the extracellular signal-regulated kinase signaling pathway. We showed that 15d-PGJ₂ suppressed cell proliferation of the endometrial carcinoma cell lines. If 15d-PGJ₂ decreased cell proliferation through PPARγ-dependent pathways, 15d-PGJ₂ may represent a new therapy for human PPARγ-positive endometrial carcinoma.

In this study, PPARγ immunoreactivity was significantly correlated with that of p21 in endometrial carcinoma tissues, and expression of p21 was induced by 15d-PGJ₂ at mRNA levels in Ishikawa, Sawano, and RL95-2 cells. Apoptosis indexes of cell lines were not altered under 10 and 20 μmol/L 15d-PGJ₂ treatment for 5 days. Several studies showed that PPARγ ligands induced cyclin-dependent kinase inhibitors, such as p21, in various types of carcinoma cells (25, 40, 41). The p21 protein inhibits cyclin-dependent kinases and mediates cell cycle arrest and cell differentiation. A potential conserved consensus peroxisome proliferator-responsive element was detected in the promoter region of p21 gene (25, 40). Suzuki et al. reported that the expression of p21 was significantly induced by 15d-PGJ₂ at both mRNA and/or protein levels in MCF-7 breast carcinoma cell line and apoptosis index of MCF-7 cells was not significantly altered under the 15d-PGJ₂ for 3 days (25). Jung et al. also showed that a large portion of human cervical carcinoma cell line C-4II cells showed growth arrest at G₁ phase with the induction of p21 following ciglitazone treatment (41). They also reported that PPARγ ligands suppressed cervical cancer cell proliferation by inhibiting cell growth, not by triggering apoptosis at least in this cell line examined (41). Shen et al. also reported that 15d-PGJ₂ induced the expression of cyclin-dependent kinase inhibitor p21 protein in human chondrosarcoma cells in a p53-independent manner, which seems to be involved in the mechanism of inhibition of cell proliferation (40). Results of our present study are consistent with all of these studies reported previously and suggest that PPARγ also regulates the expression of p21 in endometrial carcinoma tissues.

Approximately 50% of endometrial carcinoma patients had obesity and hypertension and 24% of the patients had type II diabetes mellitus in our present study. These results are also consistent with reported results of previous studies (16, 18, 42, 43). The NIH defines a normal BMI as 18.5 to 24.9. Overweight is defined as a BMI between 25.0 and 29.9. Class I obesity is a BMI between 30 and 34.9, class II obesity between 35.0 and 39.9, and class III obesity as >40 (44). However, among Japanese women, there are fewer class I obese people than in Western countries (19). Therefore, the Japan Society for the Study of Obesity originally defined class 1 obesity as BMI between 25 and 29.9, class 2 obesity as between 30 and 34.9, class 3 obesity as between 35 and 39.9, and class 4 obesity as ≥40 (19). Therefore, in this study, which examined Japanese patients with endometrial carcinoma, we defined obesity as a BMI ≥ 25. In our study of Japanese women with endometrial carcinoma, patients with endometrial carcinoma had high BMI, but paradoxically obese women with endometrial carcinoma had longer disease-free survival than non-obese women with carcinoma. Everett et al. reported that, in 396 endometrial carcinoma women, women with BMI > 40 had a lower recurrence rate compared with those with BMI < 30 (4.7% versus 13%), but the difference did not reach statistical significance (P = 0.065; ref. 42). Anderson et al. also reported that disease-free survival increased significantly (P = 0.014), and recurrence rates decreased as BMI increased (45). Therefore, an inverse correlation between biological behavior and obesity seems to be observed in both Japanese and Western patients with endometrial carcinoma. Further investigation is needed to clarify the mechanisms.

There was a significant negative correlation between PPARγ expression and BMI in women with endometrial carcinoma (P < 0.0001). To the best of our knowledge, this is the first study that analyzed the correlation between PPARγ expression and BMI in endometrial carcinoma patients. Kadowaki et al. reported that PPARγ was a thrifty gene mediating high-fat diet-induced obesity, adipose hypertrophy, and insulin resistance (46). They also reported that genetic or environmental factors causing obesity might interact with the PPARγ gene, leading to differences in insulin sensitivity between subjects with and without this substitution in overweight and obese subjects. Trujillo et al. reported that circulating adiponectin, a hormone produced exclusively by the adipocyte, was a biomarker of the metabolic syndrome (47). Adiponectin functions as an insulin sensitizer by decreasing hepatic glucose output and thereby contributing to the regulation of whole-body insulin homeostasis. Combs et al. reported induction of adipose tissue adiponectin expression and subsequent increases in circulating adiponectin levels represented a novel potential mechanism for PPARγ-mediated enhancement of whole-body insulin sensitivity (48). PPARγ agonists are known to raise circulating adiponectin levels (48). Maso et al. proposed that the combined effects of low plasma adiponectin and high BMI contributed to a >6-fold excess risk (49). These results may also explain an inverse correlation between BMI and PPARγ expression in endometrial carcinoma detected in our present study, but further investigations are awaited for clarification.

In our present study, we showed the expression of PPARγ in human endometrial carcinoma and the effects of PPARγ ligand in endometrial carcinoma cells. These findings suggest that a PPARγ ligand, 15d-PGJ₂, has antiproliferative activity against endometrial carcinoma.

We thank Dr. Masato Nishida (National Kasumigaura Hospital) for providing Ishikawa 3-H-12 cells and Yoko Sugihashi (Department of Obstetrics and Gynecology, Tohoku University Graduate School of Medicine) for skillful technical assistance.