Purpose: Peroxisome proliferator-activated receptor γ (PPARγ) is a ligand-activated transcription factor that has been implicated in carcinogenesis and progression of various solid tumors, including pancreatic carcinoma. We aimed to clarify the expression patterns of PPARγ in pancreatic ductal carcinomas and to correlate these to clinicopathologic variables, including patient survival.

Experimental Design: Array-based expression profiling of 19 microdissected carcinomas and 14 normal ductal epithelia was conducted. Additionally, Western blots of pancreatic cancer cell lines and paraffinized tissue of 129 pancreatic carcinomas were immunostained for PPARγ. For statistical analysis, Fisher's exact test, χ2 test for trends, correlation analysis, Kaplan-Meier analysis, and Cox's regression were applied.

Results: Expression profiles showed a strong overexpression of PPARγ mRNA (change fold, 6.9; P = 0.04). Immunohistochemically, PPARγ expression was seen in 71.3% of pancreatic cancer cases. PPARγ expression correlated positively to higher pT stages and higher tumor grade. Survival analysis showed a significant prognostic value for PPARγ, which was found to be independent in the clinically important subgroup of node-negative tumors.

Conclusions: PPARγ is commonly up-regulated in pancreatic ductal adenocarcinoma and might be a prognostic marker in this disease. Both findings corroborate the importance of PPARγ in tumor progression of pancreatic cancer.

Pancreatic ductal adenocarcinoma (PDAC) confers one of the highest mortality rates in malignant human tumors. In 2005, the estimated incidence for this disease in the United States alone reached 32,180 cases, with an equally high rate of deaths related to this disease (1). The high disease-specific mortality is due to the fact that pancreatic carcinomas lack early symptoms. Consequently PDACs are often detected when the local tumor extent is high and metastatic spread to regional lymph nodes or to distant organ sites has already occurred. Therefore, only ∼20% of patients are eligible for radical surgery. To date, no satisfactory progress in the treatment of pancreatic adenocarcinomas has been made and gemcitabine is the only substance that results in a short gain of patient survival time (2, 3). Most chemotherapeutic regimens target the proliferative potential of tumor cells, which is initially effective in many instances because uncontrolled proliferation is a hallmark of these cells. To specifically target other properties of tumor cells, a variety of candidate genes are being screened for their importance to tumor biology and eligibility as target genes for therapy. One of these candidates is peroxisome proliferator-activated receptor γ (PPARγ). PPARγ is a member of the nuclear receptor family of ligand-activated transcription factors of which three subtypes (α, β, and γ) have been identified thus far, which show different tissue distributions and ligand specificities (4). PPARγ was initially recognized as a key regulator of adipogenic differentiation and glucose homeostasis. Recent data supported evidence for participation of PPARγ in the biological mechanisms underlying the carcinogenic evolution, by affecting proliferation and differentiation of cancer cells in vitro and in vivo (5, 6). PPARγ expression has been described in a variety of human malignancies, including, among others, breast cancer, prostate cancer, glioblastoma cell lines, non–small cell lung carcinoma, ovarian cancer, and pancreatic carcinoma (713). To our knowledge, a detailed description of PPARγ expression in a larger cohort of clinically characterized pancreatic cancer cases is still missing. Therefore, we aimed to investigate the expression of PPARγ in a collection of pancreatic carcinomas and to correlate the data with clinicopathologic variables, including patient survival.

Patients. The study was carried out with institutional review board consent using surgically resected tissue of 129 patients with PDAC. Of these, 95 cases had surgery in Berlin between the years 1991 and 2000, whereas 34 cases underwent surgery in Dresden. Additionally, frozen tissue from 33 patients undergoing pancreatic surgery was included for expression profiling. The study group for immunohistochemistry consisted of 70 male and 59 female patients whose age ranged from 34 to 80 years at the time of diagnosis (median, 64). Complete follow-up data were available for all cases. The overall survival time ranged from 2 to 492 weeks (median, 72). One hundred two patients died during follow-up after a median overall survival time of 55 weeks.

Tumor histology was determined according to the criteria of the WHO. All cases were ductal adenocarcinomas, which were predominantly localized in the head of the pancreas (n = 118; 91.5%), four tumors were in corpus/cauda region, and in seven cases, the papilla duodeni major was involved. The stage of tumors was assessed according to International Union Against Cancer. The clinicopathologic characteristics of the tumor cohort are described in Table 1.

Table 1.

Clinicopathologic variables and PPARγ expression of the tumor cohort

CharacteristicsAll cases, n (%)PPARγ low, n (%)PPARγ high, n (%)P
All cases  79 (61.2) 50 (38.8)  
Age at diagnosis (y)     
    ≤64 73 (56.6) 44 (34.1) 29 (22.5) 0.856 
    >64 56 (43.4) 27 (27.1) 16 (16.3)  
Tumor stage     
    pT1 5 (3.9) 3 (3.2) 2 (1.6) 0.021* 
    pT2 40 (31.0) 30 (23.3) 10 (7.8)  
    pT3 78 (60.5) 45 (34.9) 33 (25.6)  
    pT4 6 (4.7) 1 (0.8) 5 (3.9)  
Nodal status     
    pN0 46 (36.5) 31 (24.6) 15 (11.9) 0.349 
    pN1 80 (63.5) 47 (37.3) 33 (26.2)  
Grade     
    G1 10 (7.8) 7 (5.4) 3 (2.3) 0.118* 
    G2 66 (51.2) 44 (34.1) 22 (17.1)  
    G3 53 (41.1) 28 (21.7) 25 (19.4)  
Residual tumor status     
    R0 94 (82.5) 58 (50.9) 36 (31.6) 0.452 
    R1 20 (17.5) 10 (8.8) 10 (8.8)  
CharacteristicsAll cases, n (%)PPARγ low, n (%)PPARγ high, n (%)P
All cases  79 (61.2) 50 (38.8)  
Age at diagnosis (y)     
    ≤64 73 (56.6) 44 (34.1) 29 (22.5) 0.856 
    >64 56 (43.4) 27 (27.1) 16 (16.3)  
Tumor stage     
    pT1 5 (3.9) 3 (3.2) 2 (1.6) 0.021* 
    pT2 40 (31.0) 30 (23.3) 10 (7.8)  
    pT3 78 (60.5) 45 (34.9) 33 (25.6)  
    pT4 6 (4.7) 1 (0.8) 5 (3.9)  
Nodal status     
    pN0 46 (36.5) 31 (24.6) 15 (11.9) 0.349 
    pN1 80 (63.5) 47 (37.3) 33 (26.2)  
Grade     
    G1 10 (7.8) 7 (5.4) 3 (2.3) 0.118* 
    G2 66 (51.2) 44 (34.1) 22 (17.1)  
    G3 53 (41.1) 28 (21.7) 25 (19.4)  
Residual tumor status     
    R0 94 (82.5) 58 (50.9) 36 (31.6) 0.452 
    R1 20 (17.5) 10 (8.8) 10 (8.8)  
*

χ2 for trends.

Cell lines. Pancreatic cancer cell lines were obtained from American Type Culture Collection: (Wesel, Germany); AsPC-1 was originally isolated from ascites of a patient with a G2 PDAC. Mia Paca-2, PANC1, and PancTuI stemmed from weakly differentiated (G3) primary PDACs. Capan-1 was isolated from a lymph node metastasis of a PDAC patients, and Colo357 and Panc89 were from liver metastases.

Expression profiling. We analyzed the expression of microdissected PDACs from 19 patients and microdissected normal duct cells from 14 patients undergoing pancreatic surgery for nonmalignant diseases, as described before. Briefly, freshly frozen tissue samples of PDAC were obtained from surgical specimens of patients who were operated at the Department of Visceral, Thoracic, and Vascular Surgery, Carl Gustav Carus, Technical University of Dresden (Dresden, Germany), and the Department of General Surgery, University of Kiel (Kiel, Germany) between 1996 and 2003. Normal pancreatic tissue was obtained from 14 patients who underwent pancreatic resection for other pancreatic diseases. These tissues were histologically normal tissues without dysplastic changes in the ducts and were taken from the distal parts of the resected pancreas. PDAC cells and normal ductal cells were dissected manually using a sterile injection needle. The estimated cellularity was 10,000 to 11,000 cells per microdissected sample. The cellularity of the dissections was ∼95%. The RNA was extracted, amplified, and hybridized to the U133 GeneChip set as described previously (14). The obtained data were normalized and condensed using the dChip software,5

which also was used to generate the heatmap. Identification of differential expression was done using the SAM software.6 Genes were assigned as differential expressed if the fold change was larger than two and the q value was below 5% resulting in a set of differentially expressed genes for PDAC. Pathway analysis of genes interacting with PPARγ was done using PathwayAssist 3.0 (Stratagene, Amsterdam, The Netherlands). Such genes were used to identify PPARγ-related genes within the set of differentially expressed genes.

Western blot of PPARγ. Pancreatic cell lines were cultivated in RPMI 1640 with 10% FCS (Invitrogen, Karlsruhe, Germany), lysed using Novex LDS sample buffer, and subjected to gel electrophoresis using the Novex Nupage system (Invitrogen, Karlsruhe, Germany). After completion, the proteins were transferred to a nitrocellulose membrane (Amersham Pharmacia Biosciences, Freiburg, Germany). PPARγ was detected by using the PPARγ-specific antibody (E8, Santa Cruz Biotechnology, Santa Cruz, CA) diluted 1:100 and the enhanced chemiluminescence plus Western blot detection system (Amersham Pharmacia Biosciences). After detection, the blots were stripped and β-actin was detected by using a mouse monoclonal antibody (Abcam, Heidelberg, Germany) diluted 1:5,000. The secondary antibody was diluted 1:25,000 in both cases.

Quantitative reverse transcription-PCR of PPARγ in cell lines. For the determination of PPARγ expression in PDAC cell lines, total RNA was prepared and total RNA (5 μg) reverse transcribed was done using SuperScript II reverse transcriptase and random hexamer primers (Invitrogen, Heidelberg, Germany) according to the manufacturer's suggestion. The genes were amplified with the SYBR Green Universal PCR Master Mix according to the manufacturer's instructions, with the ABI Prism 5700 Sequence Detection System (Weitenstadt, Germany) using the following primers: PPARγ, 5′-atgacagcgacttggcaata-3′ (forward) and 5′-gaatgtcttcaatgggcttca-3′ (reverse); β-actin, 5′-aatgctatcacctcccctgtgt-3′ (forward) and 5′-aagccaccccacttctctctaa-3′ (reverse).

Immunohistochemistry. Formalin-fixed paraffin-embedded tissue was freshly cut into slices of 4 μm. The sections were mounted on Superfrost slides (Menzel Gläser, Braunschweig, Germany), dewaxed with xylene, and gradually hydrated. Antigen retrieval was achieved by pressure cooking in 0.01 mol/L citrate buffer for 5 minutes. The primary PPARγ antibody (E8) was diluted 1:75 using a background reducing dilution buffer (DAKO, Hamburg, Germany). No other blocking agents were used. The primary antibody was incubated at room temperature for 1 hour. As a negative control, two slides were processed without primary antibody. Detection took place by the conventional labeled streptavidin-biotin (DAKO) method with alkaline phosphatase as the reporting enzyme according to the manufacturer's instructions. Fast Red (Sigma-Aldrich, Munich, Germany) served as chromogen. Afterwards, the slides were briefly counterstained with hematoxylin and aqueously mounted.

Evaluation of the immunohistochemical stainings. The slides were evaluated by two clinical pathologists. Classification of nuclear immunoreactivity was based on staining intensity and the amount of positive tumor and graded as 0, no staining; 1, weak; 2, moderate; and 3, strong. The additional cytoplasmic staining seen in a few cases was not evaluated.

Statistical analysis. The data were compiled with the software package SPSS, version 12.0 (Chicago, IL). Fisher's exact and χ2 tests were used to assess the statistical significance of the association between expression of PPARγ and clinicopathologic variables. Correlation analysis comprised Pearson's analysis for the array data and Spearman's analysis for the immunohistochemistry data. Univariate survival analysis was done according to Kaplan-Meier, and differences in survival curves were assessed with the log-rank test. Ps < 0.05 were considered significant. All statistics were accredited by the head biostatistician of the Tumor Centre, Charité University Hospital (Berlin, Germany).

Expression profiling

To analyze the expression of PPARγ in PDAC, we evaluated the obtained gene expression profiles from microdissected samples of normal ductal epithelia and adenocarcinomas. PPARγ was overexpressed in ductal adenocarcinomas by a factor of (median) 6.92 (P = 0.04; Fig. 1). We were then interested if target genes of PPARγ or genes regulating PPARγ function were also differentially expressed. These genes were identified by the PathwayAssist program. Of the eight identified target genes, (CEACAM5, MCM7, AGR2, JUND, AGTR1, LOR, NR5A2, and RORC) three are also overexpressed in pancreatic cancer (CEACAM5, MCM7, and AGR2), whereas the others are underexpressed. Of seven genes known to regulate PPARγ function (STRA13, EGR1, CEBPD, BMPR1A, BMP2, EGF, and FGFR1), the activator BMPR1A and the inhibitor STRA13 were up-regulated in PDAC, whereas the others were down-regulated (Table 2). In a stratified correlation analysis of PPARγ with these 15 genes in normal and tumor tissue, we found quite divergent results: in normal tissue, PPARγ was significantly correlated with CEACAM5 [correlation coefficient (cc), 0.997; P < 0.001], MCM7 (cc, 0.740; P = 0.002), JUND (cc, −0.725; P = 0.002), STRA13 (cc, 0.784; P = 0.001), EGR1 (cc, −0.570; P = 0.026), CEBPD (cc, −0.556; P = 0.032), BMPR1A (cc, 1; P < 0.001), and NR0B2 (cc, −0.526; P = 0.044). In tumor tissue, PPARγ correlated only to CEACAM (cc, 0.478; P = 0.033), JUND (cc, 0.554; P = 0.011), and FGFR1 (cc, 0.449; P = 0.047). This discrepancy of different correlations in normal and malignant tissue suggests an impaired function of PPARγ in tumor tissue. The coactivator of PPARγ retinoid X receptor α was not differentially expressed in PDAC (Fig. 2A and B; Table 2).

Fig. 1.

Box plot of arbitrary fluorescent units of PPARγ expression in microdissected pancreatic tissue. T, expression in microdissected PDAC; N, expression in microdissected normal duct epithelia. To confirm the differential expression, t test was applied (P = 0.04).

Fig. 1.

Box plot of arbitrary fluorescent units of PPARγ expression in microdissected pancreatic tissue. T, expression in microdissected PDAC; N, expression in microdissected normal duct epithelia. To confirm the differential expression, t test was applied (P = 0.04).

Close modal
Table 2.

PPARγ-associated genes differentially expressed in PDAC identified using gene expression analysis of microdissected samples

Probe set IDSymbolDescriptionF/C (T/N)
208510_s_at PPARG Peroxisome proliferative-activated receptor, γ 6.09 
Regulating PPARγ activity    
    204832_s_at BMPR1A Bone morphogenetic protein receptor, type IA 5.45 
    209478_at STRA13 Stimulated by retinoic acid 13 homologue 3.00 
    201694_s_at EGR1 Early growth response 1 0.53 
    226705_at FGFR1 Fibroblast growth factor receptor 1 0.49 
    203973_s_at CEBPD CCAAT/enhancer binding protein, δ 0.47 
    206254_at EGF Epidermal growth factor (β-urogastrone) 0.21 
    205290_s_at BMP2 Bone morphogenetic protein 2 0.14 
Regulated by PPARγ    
    201884_at CEACAM5 Carcinoembryonic antigen-related cell adhesion molecule 5 4.03 
    208795_s_at MCM7 MCM7 minichromosome maintenance deficient 7 2.58 
    209173_at AGR2 Anterior gradient 2 homologue 2.26 
    203752_s_at JUND Jun D proto-oncogene 0.49 
    228806_at RORC Retinoic acid receptor-related orphan receptor C 0.47 
    207720_at LOR Loricrin 0.47 
    205357_s_at AGTR1 Angiotensin II receptor, type 1 0.41 
    208343_s_at NR5A2 Nuclear receptor subfamily 5, group A, member 2 0.36 
    208711_s_at CCND1 Cyclin D1 2.82 
Binding partner of PPARγ    
    206074_s_at HMGA1 High mobility group AT-hook 1 2.49 
    206410_at NR0B2 Nuclear receptor subfamily 0, group B, member 2 0.47 
    202426_s_at RXRA Retinoid X receptor, α 1.04 
Probe set IDSymbolDescriptionF/C (T/N)
208510_s_at PPARG Peroxisome proliferative-activated receptor, γ 6.09 
Regulating PPARγ activity    
    204832_s_at BMPR1A Bone morphogenetic protein receptor, type IA 5.45 
    209478_at STRA13 Stimulated by retinoic acid 13 homologue 3.00 
    201694_s_at EGR1 Early growth response 1 0.53 
    226705_at FGFR1 Fibroblast growth factor receptor 1 0.49 
    203973_s_at CEBPD CCAAT/enhancer binding protein, δ 0.47 
    206254_at EGF Epidermal growth factor (β-urogastrone) 0.21 
    205290_s_at BMP2 Bone morphogenetic protein 2 0.14 
Regulated by PPARγ    
    201884_at CEACAM5 Carcinoembryonic antigen-related cell adhesion molecule 5 4.03 
    208795_s_at MCM7 MCM7 minichromosome maintenance deficient 7 2.58 
    209173_at AGR2 Anterior gradient 2 homologue 2.26 
    203752_s_at JUND Jun D proto-oncogene 0.49 
    228806_at RORC Retinoic acid receptor-related orphan receptor C 0.47 
    207720_at LOR Loricrin 0.47 
    205357_s_at AGTR1 Angiotensin II receptor, type 1 0.41 
    208343_s_at NR5A2 Nuclear receptor subfamily 5, group A, member 2 0.36 
    208711_s_at CCND1 Cyclin D1 2.82 
Binding partner of PPARγ    
    206074_s_at HMGA1 High mobility group AT-hook 1 2.49 
    206410_at NR0B2 Nuclear receptor subfamily 0, group B, member 2 0.47 
    202426_s_at RXRA Retinoid X receptor, α 1.04 

NOTE: Genes were classified using the Pathway Assist software [F/C (T/N) fold change of tumor above normal tissue]. For completeness, the fold change of the PPARγ coactivator retinoid X receptor α is also shown.

Fig. 2.

Expression of PPARγ and related genes in PDAC. A, PPARγ pathway containing PPARγ target genes in the lower part and genes that regulate PPARγ function in the upper part. Left of PPARγ, the coactivators of PPARγ. CCND1 is depicted on the right of PPARγ because it appears that it regulates and is regulated by PPARγ (green, underexpressed; red, overexpressed; blue, not differentially expressed; +, positive regulation; −, negative regulation). B, heatmap of PPARγ-related genes (red, high expression; blue, low expression).

Fig. 2.

Expression of PPARγ and related genes in PDAC. A, PPARγ pathway containing PPARγ target genes in the lower part and genes that regulate PPARγ function in the upper part. Left of PPARγ, the coactivators of PPARγ. CCND1 is depicted on the right of PPARγ because it appears that it regulates and is regulated by PPARγ (green, underexpressed; red, overexpressed; blue, not differentially expressed; +, positive regulation; −, negative regulation). B, heatmap of PPARγ-related genes (red, high expression; blue, low expression).

Close modal

Western blot and quantitative reverse transcription-pCR

Western blot of PPARγ with the E8 antibody revealed a specific band between 50 and 60 kDa, as expected (Fig. 3A). Expression of PPARγ protein corresponded well with the expression levels of PPARγ mRNA as seen on the microarrays (Fig. 2) and CT values of the reverse transcription-PCR (Fig. 3B). This indicates that the differences in expression seen in the primary tissue is represented in pancreatic cancer cell lines and also shows that the RNA and protein level of PPARγ are correlated.

Fig. 3.

Detection of PPARγ by Western blot and quantitative reverse transcription-PCR in pancreatic cancer cell lines. A, top lane, PPARγ; bottom lane, β-actin, which served as a loading control. Expression of PPARγ protein corresponded well with the expression levels of PPARγ mRNA as seen on the microarrays (Fig. 2). B, expression of PPARγ RNA in PDAC cell lines normalized to Capan-1 expression in percent. For PPARγ, quantitative reverse transcription-PCR was done and the obtained CT values were first normalized to β-actin as housekeeping gene. The obtained ΔCT values were then normalized to the ΔCT value from the Capan-1 cell line, which was arbitrarily set to 100%.

Fig. 3.

Detection of PPARγ by Western blot and quantitative reverse transcription-PCR in pancreatic cancer cell lines. A, top lane, PPARγ; bottom lane, β-actin, which served as a loading control. Expression of PPARγ protein corresponded well with the expression levels of PPARγ mRNA as seen on the microarrays (Fig. 2). B, expression of PPARγ RNA in PDAC cell lines normalized to Capan-1 expression in percent. For PPARγ, quantitative reverse transcription-PCR was done and the obtained CT values were first normalized to β-actin as housekeeping gene. The obtained ΔCT values were then normalized to the ΔCT value from the Capan-1 cell line, which was arbitrarily set to 100%.

Close modal

PPARγ immunostaining in pancreatic tissue

To validate the observed overexpression of PPARγ in PDAC, we did immunohistochemistry using a mouse monoclonal antibody. Within normal pancreatic acinar epithelium, staining for PPARγ was not detected. Normal ductal epithelium was mostly PPARγ negative or showed a very weak nuclear or cytoplasmic staining (Fig. 4A). Intraductal proliferations with atypia diagnostic of pancreatic intraepithelial neoplasia (PanIN) occasionally exhibited a nuclear pronounced PPARγ expression (Fig. 4B). In invasive carcinomas, a heterogenous nuclear immunoreactivity was noted, which reached from weak to strong (Fig. 4C and D). In total, 71.3% of carcinomas were PPARγ positive, 28.7% of cases were PPARγ negative. The positive carcinomas were further classified into weak (32.6%), moderate (25.6%), and strong (13.2 %) expression of PPARγ. For statistical analysis, we grouped carcinomas, which had either no or a weak PPARγ expression together (low level group, 61.3% of cases) and opposed them to cases with a moderate or strong expression of PPARγ (high level group, 38.7% of cases).

Fig. 4.

PPARγ immunohistochemistry. A, normal pancreatic parenchyma. No significant PPARγ expression is seen in normal ductal epithelium (central), whereas acinar epithelia exhibit a very weak cytoplasmic staining. B, moderate nuclear PPARγ expression in atypical epithelia of a PanIN lesion. C and D, moderate/strong nuclear PPARγ expression in pancreatic adenocarcinoma.

Fig. 4.

PPARγ immunohistochemistry. A, normal pancreatic parenchyma. No significant PPARγ expression is seen in normal ductal epithelium (central), whereas acinar epithelia exhibit a very weak cytoplasmic staining. B, moderate nuclear PPARγ expression in atypical epithelia of a PanIN lesion. C and D, moderate/strong nuclear PPARγ expression in pancreatic adenocarcinoma.

Close modal

In cross-tables, we did not find any significant association of PPARγ staining intensity with patient age, nodal status, lymphangiosis or hemangiosis carcinomatosa, and histologic grade. However, there is a significant association of higher PPARγ expression levels with higher pT stages (Table 1). However, Spearman's rank correlations showed a significant correlation of PPARγ immunoreactivity with tumor grade and pT stage (Table 3).

Table 3.

Correlation of PPARγ protein expression in pancreatic cancer with conventional clinicopathologic variables

StatisticpT statuspN statusGradingResidual tumor (R)Age
PPARγ cc 0.249 0.132 0.225 0.089 −0.119 
 P 0.004 0.140 0.010 0.347 0.179 
 n 129 126 129 114 129 
StatisticpT statuspN statusGradingResidual tumor (R)Age
PPARγ cc 0.249 0.132 0.225 0.089 −0.119 
 P 0.004 0.140 0.010 0.347 0.179 
 n 129 126 129 114 129 

Abbreviation: n, number of cases.

PPARγ expression and patient survival

Univariate survival analysis. In univariate survival analyses, cumulative survival curves were calculated according to the Kaplan-Meier method. Differences in survival were assessed with the log-rank test. The conventional prognostic markers histologic tumor grade, nodal status, and patient age, but also PPARγ immunoreactivity reached significance for different overall survival times (Fig. 5A; Table 4). No significant differences in survival time became apparent for patient gender or pT stage.

Fig. 5.

Survival analysis. Kaplan-Meier analysis of overall survival times stratified according to PPARγ expression in patients with pancreatic adenocarcinomas (dotted line, low level; continuous line, high levels). A, all cases (n = 129). B, subgroup of pT3/pT4 cases (n = 85). C, subgroup of G3 cases (n = 53). D, subgroup of node-negative cases (n = 46).

Fig. 5.

Survival analysis. Kaplan-Meier analysis of overall survival times stratified according to PPARγ expression in patients with pancreatic adenocarcinomas (dotted line, low level; continuous line, high levels). A, all cases (n = 129). B, subgroup of pT3/pT4 cases (n = 85). C, subgroup of G3 cases (n = 53). D, subgroup of node-negative cases (n = 46).

Close modal
Table 4.

Kaplan-Meier survival analysis of PPARγ expression and selected tumor variables

CasesEventsMedian survival (wk)SEP, log-rank test
Age at diagnosis (y) ≤64 73 53 84 0.047 
 >64 43 40 64  
Tumor stage pT1/pT2 45 35 71 12 0.937 
 pT3/pT4 84 67 76  
Nodal status pN0 46 29 96 23 0.001 
 pN1 80 71 60  
Grade G1/G2 76 54 84 <0.001 
 G3 53 48 52  
PPARγ expression low 79 58 78 10 0.048 
 high 50 44 64 12  
CasesEventsMedian survival (wk)SEP, log-rank test
Age at diagnosis (y) ≤64 73 53 84 0.047 
 >64 43 40 64  
Tumor stage pT1/pT2 45 35 71 12 0.937 
 pT3/pT4 84 67 76  
Nodal status pN0 46 29 96 23 0.001 
 pN1 80 71 60  
Grade G1/G2 76 54 84 <0.001 
 G3 53 48 52  
PPARγ expression low 79 58 78 10 0.048 
 high 50 44 64 12  

Survival analysis in patient subgroups

To assess if PPARγ might bear a pronounced prognostic effect in patient subgroups, we conducted an extensive Kaplan-Meier analysis of PPARγ expression stratified for pT stage, tumor grading, and nodal status. In the subgroup of pT3/pT4 tumors, high levels of PPARγ expression were associated with shorter overall survival times (median survival time, 97 versus 56 weeks; P = 0.001; Fig. 5B), which was not found in the group of pT1/pT2 tumors (P = 0.243). In the subgroup of G1/G2 tumors, high levels of PPARγ expression were associated with shorter overall survival times (median survival time, 128 versus 77 weeks; P = 0.011; Fig. 5C), which was not found in the group of G3 tumors (P = 0.726). Stratifying for nodal status, in node-negative cases, survival time was significantly shorter in the group of tumors with high levels of PPARγ expression (median survival time, 148 versus 52 weeks; P = 0.012; Fig. 5D), which again was not seen in node positive cases (P = 0.964).

Multivariate survival analysis

In the Cox regression model, we included pT stage (pT1/pT2 versus pT3/pT4), tumor grade (G1/G2 versus G3), patient age, nodal status (pN0 versus pN1), and PPARγ expression (0/1+ versus 2/3+). In the analysis of all cases, overall survival time was significantly dependent on nodal status, patient age, and tumor grade (Table 5). Analogous to the stratified univariate analysis, the Cox regression analysis in the subgroup of node-negative patients (n = 46) revealed that PPARγ expression was a significant indicator for shortened patient survival (relative risk, 2.46; P = 0.039) as shown in Table 6.

Table 5.

Cox regression model, including conventional variables and PPARγ expression in all cases (n = 129)

Overall survival (102 events)
Relative risk95% CIP
pT stage 1.061 0.694-1.624 0.784 
Grading 1.983 1.307-3.008 0.001 
Nodal status 1.864 1.196-2.905 0.006 
Age 1.026 1.002-1.050 0.033 
PPARγ 1.271 0.831-1.944 0.269 
Overall survival (102 events)
Relative risk95% CIP
pT stage 1.061 0.694-1.624 0.784 
Grading 1.983 1.307-3.008 0.001 
Nodal status 1.864 1.196-2.905 0.006 
Age 1.026 1.002-1.050 0.033 
PPARγ 1.271 0.831-1.944 0.269 
Table 6.

Cox regression model, including conventional variables and PPARγ expression in the subgroup of node-negative cases (n = 46)

Overall survival (29 events)
Relative risk95% CIP
pT stage 0.806 0.374-1.735 0.581 
Grading 1.763 0.787-3.947 0.168 
Age 1.030 0.986-1.076 0.180 
PPARγ 2.463 1.048-5.788 0.039 
Overall survival (29 events)
Relative risk95% CIP
pT stage 0.806 0.374-1.735 0.581 
Grading 1.763 0.787-3.947 0.168 
Age 1.030 0.986-1.076 0.180 
PPARγ 2.463 1.048-5.788 0.039 

PPARγ belongs to the nuclear hormone receptor superfamily that includes the receptors for steroid, thyroid hormone, vitamin D, and retinoid acids (15). Activated by binding of specific ligands, PPARγ forms a complex with retinoid X receptors (16). This complex binds to so-called peroxisome proliferator response elements in target genes, which activates transcription. PPARγ is expressed at high levels in fat tissue but has also been described in other tissues (e.g., muscle, adrenal gland, and liver; refs. 1720). PPARγ functions in many processes related to cellular development proliferation and differentiation. PPARγ is expressed in the majority of malignant human tumors, ranging from hematologic malignancies (21, 22) to solid tumors [e.g., gastric and colon cancer (23, 24), prostate, breast and ovarian cancer (8, 11, 25), non–small cell lung cancer (10), urothelial carcinoma (26), renal cell carcinoma (27), thyroid cancer (28, 29), esophageal squamous cell carcinoma (30), osteosarcoma (31), liposarcoma (32), glioma (33), and neuroblastoma (34)]. Administration of synthetic PPARγ ligands (thiazolidindiones) has shown marked antineoplastic effects on various cell lines and tumors. PPARγ agonists are antiproliferative, at least in part by up-regulating cyclin-dependent kinase inhibitors and they induce apoptosis in some tumor cell lines (3537) or differentiation of cells in others (25, 38). This might be explained by the stimulation of overexpressed but inactive PPARγ in tumors and might also be the case in PDAC because we observed overexpression of only three typical PPARγ target genes in PDAC and also an underexpression of five PPARγ target genes. This is also mirrored in the loss of correlations of PPARγ with PPARγ-associated genes we found in tumor tissue. These data indicate that other signal transduction pathways might be involved in regulating PPARγ target gene expression. Interestingly, STRA13, a gene reported to negatively regulate PPARγ, is overexpressed in PDAC. This overexpression might result in an inactivation of PPARγ and might not be compensated by the overexpression of the PPARγ activator BMPR1A resulting in PPARγ inactivation.

The stock of literature on PPARγ in pancreatic cancer thus far focused on cell lines and small tumor cohorts. Eibl et al. (35) found PPARγ expressed in six pancreatic cancer cell lines (AsPC-1, BxPC-3, Capan-2, HPAF-II, MIA Paca-2, and PANC1) and could show a decreased cell viability by treatment with PPARγ agonists, which was time and dose dependent. Farrow et al. describe PPARγ expression in the cell lines AsPC-1, SUIT-2, BxPC-3, and MIA Paca-2 and conducted invasion assays under PPARγ agonist treatment. They found a reduced invasiveness and reduced levels of proinvasive factors as tissue plasminogen activator, matrix metalloproteinase 3, matrix metalloproteinase 7, and urokinase-type plasminogen activator (39, 40). Sasaki et al. (12) found PPARγ mRNA in five of seven pancreatic carcinoma cell lines and also in five of seven cancer samples, whereas all four adjacent normal tissues showed no expression. Functionally, they could show an inhibition of anchorage-independent growth of cell lines. Motomura et al. described PPARγ expression on RNA and protein level in 4 of 4 cell lines and in 9 of 10 pancreatic cancer samples. No expression was found in normal ductal epithelium. Their immunohistologic analysis clearly showed the nuclear pattern that is to be expected of a transcription factor and which we found as well (41). In addition, a dose-dependent inhibition of cell growth by troglitazone accompanied by an up-regulation of p27 was seen. Because this could be blocked by p27 antisense oligonucleotides, the authors conclude that the PPARγ effect on cell growth is p27 mediated. Itami et al. found PPARγ expressed in 5 of 5 cell lines and conducted immunohistochemistry on 47 primary pancreas cancer samples and 15 liver metastases. Seventy-five percent of primary tumors and 80% of metastases showed a high rate of PPARγ expression. Unfortunately, no further correlation of PPARγ with clinicopathologic variables of the primary tumors was investigated (42).

To our knowledge, our study is the first description of PPARγ expression in a larger cohort of clinically characterized primary pancreatic tumors by immunohistochemistry. We used a monoclonal PPARγ antibody, which is applicable to paraffinized tissue and enabled us to analyze clinically characterized archive material. This antibody is widely used in PPARγ research (23, 28, 36, 42) and yields clear immunohistologic stainings we found reproducible and easy to analyze. We saw PPARγ expression in 71.3% of pancreatic carcinoma tissues. High rates of PPARγ expression were noted in 38.7% of carcinomas with a clear predominance in larger (pT3) and less well-differentiated (G3) tumors, still no correlation to nodal status was apparent. On survival analysis, higher rates of PPARγ expression where associated with shorter overall survival times were seen, which even reached an independent prognostic value in the subgroup of node-negative cases. In the subgroups of patients with pT3 stage disease or less aggressive tumors (G1/G2), PPARγ could as well define significant risk groups. PPARγ might therefore be considered a new candidate marker of patient prognosis in patients with pancreatic cancer, which clearly warrants further study. It seems paradoxical that we found PPARγ overexpressed in more aggressive tumors, although PPARγ ligands are known to decrease tumor growth. As stated earlier, this could hint at a disrupted PPARγ downstream pathway, which would be endorsed by our array data that does not show a stringent correlation of PPARγ RNA with typical PPARγ target genes. It also parallels the findings of Zhang et al. (11) who reported higher PPARγ levels in more aggressive ovarian cancers. The data on PPARγ expression in primary tumors and patient prognosis is sparse and divergent. Jiang et al. (43) found lower mRNA levels in breast cancer patients with high-risk tumors. Papadaki et al. (44) investigated breast cancer tissues immunohistochemically and found low rates of PPARγ to be prognostic of shorter disease-free survival times. It is of interest to note, however, that although they used the same antibody we used in our study, they reported a predominantly cytoplasmic and even membranous immunostaining, which they subjected to statistical evaluation.

Given the wealth of experimental data on the growth-inhibitory effect of agonistic PPARγ ligands on pancreatic cancer cell lines and the high rate of PPARγ expression found in pancreatic cancer, we think that this strongly suggests to further investigate the potential of PPARγ ligands for treatment of pancreatic cancer. This idea is also supported in consideration of the convenient toxicity profile of these drugs and the very limited treatment options that are available for patients with this disastrous disease. Still, there are two caveats. First, previous studies with PPARγ ligands in liposarcoma as well as in prostate and breast cancer have shown diverse effects and were only partially successful (32, 45, 46). Also has troglitazone, the PPARγ ligand used in the breast cancer study, meanwhile been withdrawn by the FDA because of hepatic toxicity. Second, there is evidence that PPARγ ligands might even further induce tumor malignancy in some tissues (4749). In conclusion, further studies are needed to clarify the possible role of PPARγ expression in pancreatic cancer and to elucidate its diagnostic or prognostic role and to clarify the therapeutic potential of PPARγ ligands.

Grant support: Deutsche Krebshilfe (70-2937-SaI).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Britta Beyer for the excellent technical assistance.

1
Jemal A, Murray T, Ward E, et al. Cancer statistics, 2005.
CA Cancer J Clin
2005
;
55
:
10
–30.
2
Haller D. New Perspectives in the management of pancreatic cancer.
Semin Oncol
2003
;
30
:
3
–10.
3
Clayton AJ, Mansoor AW, Jones ET, et al. A phase II study of weekly cisplatin and gemcitabine in patients with advanced pancreatic cancer: is this a strategy still worth pursuing?
Pancreas
2006
;
32
:
51
–7.
4
Kersten S, Desvergne B, Wahli W. Roles of PPARs in health and disease.
Nature
2000
;
405
:
421
–4.
5
Gelman L, Fruchart JC, Auwerx J. An update on the mechanisms of action of the peroxisome proliferator-activated receptors (PPARs) and their roles in inflammation and cancer.
Cell Mol Life Sci
1999
;
55
:
932
–43.
6
Fajas L, Debril MB, Auwerx J. Peroxisome proliferator-activated receptor-γ: from adipogenesis to carcinogenesis.
J Mol Endocrinol
2001
;
27
:
1
–9.
7
Watkins G, Douglas-Jones A, Mansel RE, Jiang WG. The localisation and reduction of nuclear staining of PPARγ and PGC-1 in human breast cancer.
Oncol Rep
2004
;
12
:
483
–8.
8
Segawa Y, Yoshimura R, Hase T, et al. Expression of peroxisome proliferator-activated receptor (PPAR) in human prostate cancer.
Prostate
2002
;
51
:
108
–16.
9
Cimini A, Cristiano L, Bernardo A, Farioli-Vecchioli S, Stefanini S, Ceru MP. Presence and inducibility of peroxisomes in a human glioblastoma cell line.
Biochim Biophys Acta
2000
;
1474
:
397
–409.
10
Theocharis S, Kanelli H, Politi E, et al. Expression of peroxisome proliferator activated receptor-γ in non-small cell lung carcinoma: correlation with histological type and grade.
Lung Cancer
2002
;
36
:
249
–55.
11
Zhang GY, Ahmed N, Riley C, et al. Enhanced expression of peroxisome proliferator-activated receptor γ in epithelial ovarian carcinoma.
Br J Cancer
2005
;
92
:
113
–9.
12
Sasaki T, Fujimoto Y, Tsuchida A, Kawasaki Y, Kuwada Y, Chayama K. Activation of peroxisome proliferator-activated receptor γ inhibits the growth of human pancreatic cancer.
Pathobiology
2001
;
69
:
258
–65.
13
Wang T, Xu J, Yu X, Yang R, Han ZC. Peroxisome proliferator-activated receptor γ in malignant diseases.
Crit Rev Oncol Hematol
2005
;
58
:
1
–14.
14
Grutzmann R, Pilarsky C, Ammerpohl O, et al. Gene expression profiling of microdissected pancreatic ductal carcinomas using high-density DNA microarrays.
Neoplasia
2004
;
6
:
611
–22.
15
Kota BP, Huang TH, Roufogalis BD. An overview on biological mechanisms of PPARs.
Pharmacol Res
2005
;
51
:
85
–94.
16
Marcus SL, Miyata KS, Zhang B, Subramani S, Rachubinski RA, Capone JP. Diverse peroxisome proliferator-activated receptors bind to the peroxisome proliferator-responsive elements of the rat hydratase/dehydrogenase and fatty acyl-CoA oxidase genes but differentially induce expression.
Proc Natl Acad Sci U S A
1993
;
90
:
5723
–7.
17
Zhu YJ, Crawford SE, Stellmach V, et al. Coactivator PRIP, the peroxisome proliferator-activated receptor-interacting protein, is a modulator of placental, cardiac, hepatic, and embryonic development.
J Biol Chem
2003
;
278
:
1986
–90.
18
Lemberger T, Desvergne B, Wahli W. Peroxisome proliferator-activated receptors: a nuclear receptor signaling pathway in lipid physiology.
Annu Rev Cell Dev Biol
1996
;
12
:
335
–63.
19
Auboeuf D, Rieusset J, Fajas L, et al. Tissue distribution and quantification of the expression of mRNAs of peroxisome proliferator-activated receptors and liver X receptor-α in humans: no alteration in adipose tissue of obese and NIDDM patients.
Diabetes
1997
;
46
:
1319
–27.
20
Kliewer SA, Forman BM, Blumberg B, et al. Differential expression and activation of a family of murine peroxisome proliferator-activated receptors.
Proc Natl Acad Sci U S A
1994
;
91
:
7355
–9.
21
Asou H, Verbeek W, Williamson E, et al. Growth inhibition of myeloid leukemia cells by troglitazone, a ligand for peroxisome proliferator activated receptor γ, and retinoids.
Int J Oncol
1999
;
15
:
1027
–31.
22
Zang C, Liu H, Posch MG, et al. Peroxisome proliferator-activated receptor γ ligands induce growth inhibition and apoptosis of human B lymphocytic leukemia.
Leuk Res
2004
;
28
:
387
–97.
23
Sato H, Ishihara S, Kawashima K, et al. Expression of peroxisome proliferator-activated receptor (PPAR)γ in gastric cancer and inhibitory effects of PPARγ agonists.
Br J Cancer
2000
;
83
:
1394
–400.
24
Sarraf P, Mueller E, Smith WM, et al. Loss-of-function mutations in PPAR γ associated with human colon cancer.
Mol Cell
1999
;
3
:
799
–804.
25
Mueller E, Sarraf P, Tontonoz P, et al. Terminal differentiation of human breast cancer through PPAR γ.
Mol Cell
1998
;
1
:
465
–70.
26
Guan YF, Zhang YH, Breyer RM, Davis L, Breyer MD. Expression of peroxisome proliferator-activated receptor γ (PPARγ) in human transitional bladder cancer and its role in inducing cell death.
Neoplasia
1999
;
1
:
330
–9.
27
Inoue K, Kawahito Y, Tsubouchi Y, et al. Expression of peroxisome proliferator-activated receptor γ in renal cell carcinoma and growth inhibition by its agonists.
Biochem Biophys Res Commun
2001
;
287
:
727
–32.
28
Sahin M, Allard BL, Yates M, et al. PPARγ staining as a surrogate for PAX8/PPARγ fusion oncogene expression in follicular neoplasms: clinicopathological correlation and histopathological diagnostic value.
J Clin Endocrinol Metab
2005
;
90
:
463
–8.
29
Marques AR, Espadinha C, Catarino AL, et al. Expression of PAX8-PPAR γ 1 rearrangements in both follicular thyroid carcinomas and adenomas.
J Clin Endocrinol Metab
2002
;
87
:
3947
–52.
30
Takashima T, Fujiwara Y, Hamaguchi M, et al. Relationship between peroxisome proliferator-activated receptor-γ expression and differentiation of human esophageal squamous cell carcinoma.
Oncol Rep
2005
;
13
:
601
–6.
31
Haydon RC, Zhou L, Feng T, et al. Nuclear receptor agonists as potential differentiation therapy agents for human osteosarcoma.
Clin Cancer Res
2002
;
8
:
1288
–94.
32
Tontonoz P, Singer S, Forman BM, et al. Terminal differentiation of human liposarcoma cells induced by ligands for peroxisome proliferator-activated receptor γ and the retinoid X receptor.
Proc Natl Acad Sci U S A
1997
;
94
:
237
–41.
33
Grommes C, Landreth GE, Schlegel U, Heneka MT. The nonthiazolidinedione tyrosine-based peroxisome proliferator-activated receptor γ ligand GW7845 induces apoptosis and limits migration and invasion of rat and human glioma cells.
J Pharmacol Exp Ther
2005
;
313
:
806
–13.
34
Sato Y, Sasaki H, Kobayashi Y, et al. Expression of PPAR-γ is correlated with the clinical course of neuroblastoma.
J Pediatr Surg
2003
;
38
:
205
–10.
35
Eibl G, Wente MN, Reber HA, Hines OJ. Peroxisome proliferator-activated receptor γ induces pancreatic cancer cell apoptosis.
Biochem Biophys Res Commun
2001
;
287
:
522
–9.
36
Elstner E, Muller C, Koshizuka K, et al. Ligands for peroxisome proliferator-activated receptorγ and retinoic acid receptor inhibit growth and induce apoptosis of human breast cancer cells in vitro and in BNX mice.
Proc Natl Acad Sci U S A
1998
;
95
:
8806
–11.
37
Chattopadhyay N, Singh DP, Heese O, et al. Expression of peroxisome proliferator-activated receptors (PPARS) in human astrocytic cells: PPARγ agonists as inducers of apoptosis.
J Neurosci Res
2000
;
61
:
67
–74.
38
Elnemr A, Ohta T, Iwata K, et al. PPARγ ligand (thiazolidinedione) induces growth arrest and differentiation markers of human pancreatic cancer cells.
Int J Oncol
2000
;
17
:
1157
–64.
39
Farrow B, O'Connor KL, Hashimoto K, Iwamura T, Evers BM. Selective activation of PPARγ inhibits pancreatic cancer invasion and decreases expression of tissue plasminogen activator.
Surgery
2003
;
134
:
206
–12.
40
Sawai H, Liu J, Reber HA, Hines OJ, Eibl G. Activation of peroxisome proliferator-activated receptor-γ decreases pancreatic cancer cell invasion through modulation of the plasminogen activator system.
Mol Cancer Res
2006
;
4
:
159
–67.
41
Motomura W, Okumura T, Takahashi N, Obara T, Kohgo Y. Activation of peroxisome proliferator-activated receptor γ by troglitazone inhibits cell growth through the increase of p27KiP1 in human. Pancreatic carcinoma cells.
Cancer Res
2000
;
60
:
5558
–64.
42
Itami A, Watanabe G, Shimada Y, et al. Ligands for peroxisome proliferator-activated receptor γ inhibit growth of pancreatic cancers both in vitro and in vivo.
Int J Cancer
2001
;
94
:
370
–6.
43
Jiang WG, Douglas-Jones A, Mansel RE. Expression of peroxisome-proliferator activated receptor-γ (PPARγ) and the PPARγ co-activator, PGC-1, in human breast cancer correlates with clinical outcomes.
Int J Cancer
2003
;
106
:
752
–7.
44
Papadaki I, Mylona E, Giannopoulou I, Markaki S, Keramopoulos A, Nakopoulou L. PPARγ expression in breast cancer: clinical value and correlation with ERβ.
Histopathology
2005
;
46
:
37
–42.
45
Hisatake JI, Ikezoe T, Carey M, Holden S, Tomoyasu S, Koeffler HP. Down-regulation of prostate-specific antigen expression by ligands for peroxisome proliferator-activated receptor γ in human prostate cancer.
Cancer Res
2000
;
60
:
5494
–8.
46
Burstein HJ, Demetri GD, Mueller E, Sarraf P, Spiegelman BM, Winer EP. Use of the peroxisome proliferator-activated receptor (PPAR) γ ligand troglitazone as treatment for refractory breast cancer: a phase II study.
Breast Cancer Res Treat
2003
;
79
:
391
–7.
47
Saez E, Rosenfeld J, Livolsi A, et al. PPAR γ signaling exacerbates mammary gland tumor development.
Genes Dev
2004
;
18
:
528
–40.
48
Saez E, Tontonoz P, Nelson MC, et al. Activators of the nuclear receptor PPARγ enhance colon polyp formation.
Nat Med
1998
;
4
:
1058
–61.
49
Lefebvre AM, Chen I, Desreumaux P, et al. Activation of the peroxisome proliferator-activated receptor γ promotes the development of colon tumors in C57BL/6J-APCMin/+ mice.
Nat Med
1998
;
4
:
1053
–7.