ORIGINAL ARTICLE

Int J Clin Oncol (2007) 12:111–119 © The Japan Society of Clinical Oncology 2007DOI 10.1007/s10147-006-0634-x

Kotaro Miyake · Satoru Imura · Tomoharu Yoshizumi Tetsuya Ikemoto · Yuji Morine · Mitsuo Shimada

Role of thymidine phosphorylase and orotate phosphoribosyltransferase mRNA expression and its ratio to dihydropyrimidine dehydrogenase in the prognosis and clinicopathological features of patients with pancreatic cancer

AbstractBackground. Thymidine phosphorylase (TP), orotate phos-phoribosyltransferase (OPRT), and dihydropyrimidine de-hydrogenase (DPD) are important enzymes related to the metabolism of 5-fl uorouracil and its derivatives. In this study, we analyzed the expression of these enzymes and evaluated the association between the expression of these enzymes and clinicopathological features and prognosis in patients with pancreatic cancer.Methods. TP, OPRT, and DPD mRNA expressions were detected using a real-time reverse transcriptional-poly-merase chain reaction method or by immunohistochemis-try, using surgical specimens obtained from 25 patients with pancreatic cancer.Results. TP mRNA expression was lower in cases with an alpha infi ltration growth pattern than in cases with other infi ltration growth patterns (P < 0.05). OPRT mRNA ex-pression was higher in poorly differentiated-type cases than in differentiated type cases (P < 0.05). TP-, OPRT-, and DPD-positive stainings were found in 15 of 24 cases (63%), 10 of 19 cases (53%), and 14 of 21 cases (67%), respectively. There were signifi cant correlations or trends between the mRNA and protein expressions of TP, OPRT, and DPD. Patients with a low TP/DPD ratio survived signifi cantly longer than those with a high ratio (P < 0.05). Multivariate analysis demonstrated a signifi cantly poorer outcome in pa-tients with a high TP/DPD ratio compared with in patients with a low ratio (P < 0.05).Conclusion. The TP/DPD ratio might be useful as a prog-nostic factor in patients with pancreatic cancer.

Key words Pancreatic cancer · Thymidine phosphorylase · Orotate phosphoribosyltransferase · Laser-capture microdissection

Introduction

One of the most widely used antitumor chemotherapeutic agents for the treatment of a variety of gastrointestinal cancers is 5-fl uorouracil (5-FU). 5-FU is catabolized to di-hydrofl uorouracil by the fi rst- and rate-limiting enzyme, di-hydropyrimidine dehydrogenase (DPD).1 By itself, 5-FU is inactive, and it requires intracellular conversion to 5-fl uoro-2′-deoxyuridine 5′-monophosphate (FdUMP) by thymidine phosphorylase (TP), orotate phosphoribosyltransferase (OPRT), uridine phosphorylase (UP), and thymidine ki-nase (TK). 5-FU exerts its cytotoxic activity through the formation of a ternary complex with thymidylate synthase (TS) and 5,10-methylene-tetrahydrofolate, resulting in inhibition of TS and blockade of the DNA synthetic process.2,3

TP is the enzyme that regulates intracellular pyrimidine metabolism via the degradation of thymidine to thymine and it is highly expressed in the normal liver and intestine, and in various human cancer cells. TP has been recognized as a prognostic factor because it is the same protein as platelet-derived endothelial cell growth factor (PD-ECGF).4–8 It is a key enzyme in the conversion of 5′-deoxy-5-fl uorouridine (5′-DFUR), a prodrug of 5-FU, and capecitabine, a prodrug of 5′-DFUR, to 5-FU.9,10 OPRT is the fi rst-limiting enzyme in this 5-FU conversion, leading to the formation of FdUMP in the presence of 5-phosphoribosyl-1-pyrophosphate as a co-factor.11,12 Previous studies have demonstrated that adenovi-rus-mediated transduction of the OPRT gene resulted in a marked sensitization to 5-FU cytotoxicity in the cells of co-lon, gastric, hepatic, and pancreatic cancers.13,14 Thus, OPRT is an important enzyme in 5-FU activation. In addition, OPRT, which converts orotic acid to orotidine 5′-phosphate, is the rate-limiting enzyme in the de novo process of DNA and RNA synthesis. DPD is the fi rst- and rate-limiting en-

K. Miyake · S. Imura · T. Yoshizumi · T. Ikemoto · Y. Morine · M. Shimada (*)Department of Digestive and Pediatric Surgery, Institute of Health Biosciences, The University of Tokushima Graduate School, 3-18-15 Kuramoto, Tokushima 770-8503, JapanTel. +81-88-633-9276, 9277; Fax +81-88-631-9698e-mail: mshimada@clin.med.tokushima-u.ac.jp

Received: July 3, 2006 / Accepted: October 31, 2006

112

zyme for the catabolism of 5-FU.1 High DPD mRNA levels have been shown in various human cancers and cell lines with low sensitivity to 5-FU.15–17

Recently, it was revealed that the effi cacy of 5′-DFUR was correlated with the ratio of TP to DPD (TP/DPD ratio) activity in human cancer xenograft models.18 A clinical study using 5′-DFUR showed that patients with a high TP/DPD ratio in gastric cancer tissues had longer disease-free survival than patients with a low TP/DPD ratio.19 Further-more, both OPRT mRNA expression and the OPRT/DPD ratio might be useful as predictive parameters for the effi -cacy of fl uoropyrimidine-based chemotherapy for meta-static colorectal cancer.20

Although TP, OPRT, and DPD are important enzymes involved in 5-FU cytotoxicity and DNA synthesis, to our knowledge, this is the fi rst report about TP, OPRT, and DPD mRNA expressions in pancreatic cancer. In the pres-ent study, we investigated the relationship between TP, OPRT, and DPD expressions and clinicopathological fea-tures and survival in patients with pancreatic cancer.

Patients and methods

Patient characteristics

A total of 39 pancreatic cancer patients who underwent surgical treatment between March 1994 and March 2004 at Tokushima University Hospital were enrolled in this study. Among these patients, 14 patients were excluded because the condition of the formalin-fi xed, paraffi n-embedded tumor specimen was not adequate. Thus, TP, OPRT, and DPD mRNA were detected, using a real-time reverse tran-scriptional-polymerase chain reaction (RT-PCR) method, from the tumor specimens of only 25 patients with pancre-atic cancer.

Of the 25 patients, 16 were men and 9 women, with a mean age of 62.9 years, ranging from 36 to 79 years. None had re-ceived prior chemotherapy or irradiation before the surgery. Three patients did not receive any chemotherapy after sur-gery, 16 patients received 5-FU or tegafur uracil, and 7 pa-tients received gemcitabine, as adjuvant chemotherapy.

Patient characteristics and several clinicopathological features (age, sex, tumor size, histological type, infi ltration growth pattern, lymphatic invasion, venous invasion, local advance, lymph node metastasis, distant metastasis, and stage) were examined according to the criteria of the Classi-fi cation of pancreatic carcinoma, second English edition.21 This study was authorized in advance by the institutional re-view board of the University of Tokushima Graduate School, and all patients provided written informed consent.

Microdissection in primary tumors

A representative formalin-fi xed, paraffi n-embedded tumor specimen, which contained a central section of the cancer, was selected from each of the lesions by a pathologist after an examination of the hematoxylin-and-eosin-stained slides.

Then 10-µm-thick sections were stained with nuclear fast red to enable visualization of histology, and laser-capture microdissection (PALM Microlaser Technologies, Munich, Germany) was performed to ensure the presence of malig-nant cells in the microcentrifuge tube.

RNA extraction and cDNA synthesis from paraffi n-embedded tissues

RNA for each sample was extracted according to a previ-ously described method, with minor modifi cations.16,17,19 Briefl y, 600 µl of xylene was added to each tube. After cen-trifugation for 7 min at 20 800 g, the supernatant was dis-carded, and the washing step was repeated three times. The deparaffi nized materials were rehydrated in xylene/ethanol/water. The rehydration medium was removed after centrif-ugation for 7 min at 20 800 g. After discarding the last super-natant, the pelleted sections were resolved in 70% ethanol. Then 400 µl of buffer (4 M guanidine isothiocyanate solu-tion including 0.5% sarcosine and 8 µl 1 M dithiothreitol [DTT]) were added to the dried tissue and homogenized mechanically. For RNA demodifi cation, homogenates were heated at 95°C for 30 min. RNA was extracted from homog-enates by the addition of 50 µl of 2 M sodium acetate (pH 4.0), 500 µl of water-saturated phenol, and 100 µl of chloro-form-isoamyl mixture (49 : 1). RNA was recovered from the water phase by isopropanol precipitation and transferred to a new tube and precipitated with 10 µl glycogen and 400 µl isopropanol for 30 min at −20°C. After centrifugation for 7 min at 20 800 g, the pellet was washed with 500 µl 75% ethanol. After drying, the pellet was dissolved in 50 µl 5 mM Tris HCl (pH 8.0). Reverse transcription was carried out at 39°C for 45 min, using 400 units of Moloney murine leuke-mia virus (MMLV) reverse transcriptase, 1 × fi rst-strand buffer, 0.04 µg/µl random hexamers, 10 mM DTT, and 1 mM deoxynucleoside triphosphate.

PCR quantifi cation of mRNA expression

Target cDNA sequences were amplifi ed by quantitative PCR, using a fl uorescence-based real-time detection meth-od (ABI Prism 7900 Sequence Detection System, TaqMan; Applied Biosystems, Foster City, CA, USA) as previously described22. PCR was carried out for each gene of interest and β-actin was used as an internal reference gene. The 25-µL PCR reaction mixture contained 600 nmol/l of each primer; 200 nmol/l each of dATP, dCTP, and dGTP; 400 µmol/l dUTP; 5.5 mmol/l MgCl2; and 1 × TaqMan buffer containing a reference dye. The primers and probe sequen-ces used were as follows: TP primers, CCTGCGGACGGAATCCT and GCTGTGATGAGTGGCAGGCT, probe 6FAM (carboxyfl uorescein)-5′-CAGCCAGAGATGTGACAGCCACCGT-3′TAMRA (N,N,N0,N0-tetramethyl-6-carboxyrhodamine);20 OPRT primers, TCCTGGGCAGATCTAGTAAATGC and TGCTCCTCAGCCATTCTAACC, probe 6FAM-5′-CTCCTTATTGCGGAAATGAGCTCCACC-3′TAMRA;20 DPD primers, AGGACGCAA GGAGGGTTTG and GTCCGCCGAGTCCTTACTGA,

113

probe 6FAM-5′-CAGTGCCTACAGTCTCGAGTCTGCCAGTG3′TAMRA;23 β-actin primers, TGAGCGCGGCTACAGCTT and TCCTTAATGTCACGCACGATTT, β-actin probe 6FAM-5′-ACCACCACGGCCGAGCGG- 3′TAMRA. The PCR conditions were 50°C for 10 s and 95°C for 10 min, followed by 42 cycles at 95°C for 15 s and 60°C for 1 min. Relative gene expressions of TP and OPRT were determined based on the threshold cycles of each gene in relation to the threshold cycle of the corresponding in-ternal standard β-actin.

The use of β-actin as a reference gene avoids the need for RNA concentration measurement. The β-actin real-time PCR analysis also estimated the amount of extracted mRNA. The rise of the β-actin signal after 37 cycles under the described conditions indicated an insuffi cient amount of mRNA present for the subsequent TP and OPRT quan-titation. When measuring gene expressions in paraffi n-embedded tissues, the median value of the threshold cycle of β-actin was 26 cycles, ranging from 23 to 28.

Immunohistochemical staining

Sections of 4-µm-thick paraffi n-embedded tumor specimen were dewaxed and rehydrated with xylene and ethanol. After immersion in 10 mM citrate buffer (pH 6.0), the slides underwent microwave pretreatment for 10 min for optimal antigen retrieval. The sections were then incubated over-night at 4°C with rabbit polyclonal DPD and OPRT anti-bodies (1 : 200 for DPD dilution, and 1 : 500 for OPRT dilution; Taiho Pharmaceutical, Tokyo, Japan) and with mouse monoclonal TP antibody (1 : 100 dilution; Nippon Roche Research Center, Tokyo, Japan). The secondary an-tibody was biotin-labeled and was applied for 30 min. A streptavidin-LSA amplifi cation method (DAKO K0679, Dako Cytomation, Carpinteria, CA, USA) was carried out for 30 min, followed by peroxidase/dia minobenzidine sub-strate/chromagen treatment. The slides were counterstained with hematoxylin.

Positive expressions of TP, OPRT, and DPD proteins were determined by counting the number of tumor cells with cytoplasmic staining. The cases with staining were classifi ed into two groups according to the percentage of positively stained tumor cells in all fi elds: low expression (negative or positive in <30% of tumor cells) and high ex-pression (positive in ≥30% of tumor cells).24,25

Statistical analysis

Statistical analysis was performed using StatView 5.0J soft-ware (SAS Institute, Cary, NC, USA). The Mann-Whitney U-test or the Kruskal-Wallis test was used to compare TP or OPRT – and its ratio to DPD mRNA expression – with the clinicopathological features. Survival analysis was done by the Kaplan-Meier method and survivals were compared by the log-rank test. Cox’s proportional hazard regression model was used for a multivariate analysis. The χ2 test was used for correlations between the mRNA and protein ex-pressions of TP, OPRT, and DPD. A P value of less than 0.05 was taken as signifi cant.

Results

TP, OPRT, and DPD mRNA expressions in pancreatic cancer

The correlations between clinicopathological features and the mRNA expressions of TP, OPRT, and DPD are shown in Table 1. The median values of TP, OPRT, and DPD mRNA expressions were 5.10 (range, 2.58–37.09), 0.83 (range, 0.46–1.77), and 0.98 (range, 0.62–4.13), respectively. TP mRNA expression was lower in cases with an alpha in-fi ltration growth pattern than in cases with other patterns (P < 0.05). OPRT mRNA expression was higher in poorly differentiated cases than in differentiated cases (P < 0.05). DPD mRNA expression in pancreatic cancer showed no statistically signifi cant differences in relation to any clinico-pathological features.

Ratio of TP or OPRT to DPD in pancreatic cancer

The correlations between clinicopathological features and both the TP/DPD and OPRT/DPD ratios are shown in Table 2. The median values of the TP/DPD and OPRT/DPD ratios were 5.15 (range, 1.95–22.43) and 0.84 (range, 0.24–2.33), respectively. The TP/DPD and OPRT/DPD ra-tios showed no statistically signifi cant correlation with any clinicopathological features.

Immunohistochemical study of TP, OPRT, and DPD expressions

TP protein expression was immunohistochemically seen in the cytoplasm and nuclei of cancer cells, whereas OPRT and DPD protein expressions were seen in the cytoplasm. As shown in Fig. 1, positive stainings for TP, OPRT, and DPD were found (in 15 of 24 cases [63%], 10 of 19 cases [53%], and 14 of 21 cases [67%], respectively).

Correlations between mRNA and protein expressions of TP, OPRT, and DPD

The median values of TP, OPRT, and DPD mRNA expres-sions were 5.10 (range, 2.58–37.09), 0.83 (range, 0.23–1.77) and 0.98 (range, 0.29–4.13), and these means were selected as cutoff values to separate high and low mRNA expression of TP, OPRT, and DPD. There were signifi cant correlations between the mRNA and protein expressions of TP, OPRT, and DPD (P = 0.015, P = 0.040, and P = 0.026, respectively; Table 3).

Survival

The median survival time in the patients with high TP and OPRT mRNA expressions was not different from that in the patients with low TP and OPRT mRNA expressions. The median values of the TP/DPD and OPRT/DPD ratios in patients with high and low TP and OPRT mRNA expres-

Table 1. Correlation between clinicopathological features and TP, OPRT, and DPD mRNA expressions in pancreatic cancer

Variables Number of TP mRNA expression OPRT mRNA expression DPD mRNA expression patients

Median P value Median P value Median P value (n = 25)

Age (years) 62 (36–79)Sex Male 16 4.56 (2.58–11.93) 0.456 1.00 (0.53–1.77) 0.353 0.94 (0.62–4.13) 0.168 Female 9 6.03 (3.79–37.09) 1.01 (0.46–1.25) 1.55 (0.86–1.94)Tumor size ts1,2 11 6.16 (2.58–9.43) 0.543 1.01 (0.67–1.64) 0.514 1.33 (0.86–4.13) 0.619 ts3,4 14 3.69 (3.06–37.09) 0.88 (0.46–1.77) 1.03 (0.62–1.94)Histological type Differentiated 22 5.51 (2.58–6.84) 0.760 0.83 (0.46–1.64) 0.046 1.10 (0.62–4.13) 0.547 Poorly differentiated 3 5.03 (3.59–37.09) 1.55 (1.33–1.77) 0.99 (0.76–1.34)Infi ltration pattern Alpha 5 3.53 (2.58–11.93) 0.015 1.33 (0.55–1.77) 0.371 0.76 (0.62–1.34) 0.132 Beta and gamma 20 6.16 (3.06–37.09) 0.92 (0.46–1.64) 1.10 (0.73–4.13)Lymphatic invasiona

ly0 8 4.23 (2.58–37.09) 0.602 1.13 (0.76–1.77) 0.233 0.98 (0.62–4.13) 0.349 ly1,2,3 14 6.08 (3.06–9.53) 0.83 (0.46–1.40) 1.10 (0.88–1.94)Venous invasiona

v0 8 4.99 (2.58–9.43) 0.664 1.21 (0.55–1.40) 0.126 1.33 (0.62–4.13) 0.349 v1,2,3 14 5.57 (3.06–37.09) 0.83 (0.46–1.77) 0.97 (0.73–1.94)Local advance T2,3 13 6.03 (2.58–37.09) 0.931 0.83 (0.55–1.77) 0.680 1.02 (0.62–4.13) 0.670 T4 12 4.87 (3.06–11.93) 1.13 (0.46–1.25) 1.13 (0.86–1.94)Lymph node metastasis N0 9 5.10 (2.58–11.93) 0.801 1.01 (0.55–1.77) 0.390 0.99 (0.62–4.13) 0.396 N1,2,3 16 4.99 (3.06–37.09) 0.82 (0.46–1.64) 1.10 (0.73–1.94)Distant metastasis M0 22 7.60 (2.58–37.09) 0.695 0.83 (0.46–1.77) 0.184 0.99 (0.62–4.13) 0.338 M1 3 6.84 (3.06–11.93) 1.29 (1.24–1.33) 1.10 (1.34–1.58)Stage I,II,III 10 4.54 (2.58–37.09) 0.412 0.82 (0.55–1.77) 0.514 0.92 (0.62–4.13) 0.227 IV 15 6.03 (3.06–11.93) 1.99 (0.46–1.40) 1.23 (0.86–1.94)a In 3 patients, lymphatic and venous invasion are unclear

Table 2. Correlation between clinicopathological features and TP/DPD and OPRT/DPD ratios

Variables Number of TP/DPD ratio OPRT/DPD ratio

patients (n = 25)

Median P value Median P value

Sex Male 16 5.34 (3.80–10.79) 0.127 0.91 (0.59–2.33) 0.063 Female 9 4.14 (1.95–22.43) 0.46 (0.24–1.46)Tumor size ts1,2 11 5.06 (2.28–6.68) 0.303 0.90 (0.25–1.68) 0.935 ts3,4 14 5.35 (1.95–22.43) 0.84 (0.24–2.33)Histological type Differentiated 22 5.08 (1.95–22.43) 0.934 0.79 (0.24–1.68) 0.111 Poorly differentiated 3 5.10 (4.95–5.13) 1.66 (0.99–2.33)Infi ltration pattern Alpha 5 5.10 (4.95–5.78) 0.688 0.99 (0.89–2.33) 0.178 Beta and gamma 20 5.08 (1.95–22.43) 0.79 (0.24–1.68)Lymphatic invasiona

ly0 8 5.10 (2.28–22.43) 0.640 0.99 (0.25–2.33) 0.193 ly1,2,3 14 4.82 (1.95–10.79) 0.73 (0.24–1.23)Venous invasiona

v0 8 5.09 (2.28–9.57) 0.454 1.07 (0.25–2.33) 0.448 v1,2,3 14 5.10 (1.95–22.43) 0.79 (0.24–1.46)Local advance T2,3 13 5.10 (2.28–22.43) 0.522 0.90 (0.25–2.33) 0.680 T4 12 5.06 (1.95–10.79) 0.73 (0.24–1.46)Lymph node metastasis N0 9 5.13 (2.28–7.57) 0.316 0.90 (0.25–2.33) 0.618 N1,2,3 16 4.82 (1.95–22.43) 0.78 (0.24–1.68)Distant metastasis M0 22 5.06 (1.95–22.43) 0.281 0.89 (0.24–2.33) 0.791 M1 3 6.34 (5.10–7.57) 0.89(0.79–0.99)Stage I,II,III 10 5.09 (2.28–22.43) 0.722 0.90 (0.25–2.33) 0.744 IV 15 5.08 (1.95–10.79) 0.79 (0.24–1.46)a In 3 patients, lymphatic and venous invasiona are unclear

115

sions were 5.09 (range, 1.95–22.43) and 0.89 (range, 0.24–2.44), respectively, and these means were selected as cutoff values to separate high and low mRNA expression of TP/DPD and OPRT/DPD ratios.

As shown in Fig. 2, the log-rank test revealed that a higher TP/DPD ratio was associated with a poor outcome (P = 0.02), whereas the OPRT/DPD ratio was not so associ-ated (P = 0.23; data not shown).

Factors related to patient prognosis were also evaluated using Cox proportional hazard univariate regression analy-sis. In addition to the established risk factors for pancreatic cancer (lymphatic invasion, local advance, and stage), the TP/DPD ratio was also related to survival. Factors consid-ered to be univariately signifi cant were additionally ana-lyzed in a multivariate analysis. On multivariate analysis, only the TP/DPD ratio was found to be an independent prognostic factor (relative risk, 8.786; 95% confi dence in-terval [CI], 1.593–48.470; P < 0.005; Table 4).

The correlation between TP, OPRT, and DPD and the chemotherapeutic outcome is shown in Table 5. There was no signifi cant correlation between the chemotherapeutic regimen and survival.

Discussion

The correlation between the expressions of TS, DPD, TP, and OPRT and clinicopathological features in gastrointesti-nal cancer has already been reported by a large number of authors, based on biochemical assays, immunohistochemis-try, enzyme-linked immunosorbent assay (ELISA), or PCR methods. In the present study, TP, OPRT, and DPD mRNA expressions in pancreatic cancer were measured by real-time RT-PCR combined with laser-capture microdissection. Our study showed that TP mRNA expression correlated

a b

c d

e f

Fig. 1a–f. Representative photo-micrographs of tissue sections immunostained for thymidine phosphorylase (TP), orotate phosphoribosyl transferase , (OPRT), and dihydropyrimidine dehydrogenase (DPD). a Pancre-atic cancer with TP-positive ex-pression. The signal was detected in the cytoplasm and nuclei of cancer cells. b Pancreatic cancer with TP-negative expression. c Pancreatic cancer with OPRT-positive expression. The signal was detected in the cytoplasm of cancer cells. d Pancreatic cancer with OPRT-negative expression. e Pancreatic cancer with DPD-positive expression. The signal was detected in the cytoplasm of cancer cells. f Pancreatic cancer with DPD-negative expression. a–f × 200

116

Table 3. Correlations between mRNA and protein expression of TP, OPRT, and DPD

(a)TP mRNA Total TP protein P valueexpression n = 24 expression

High Low

High 13 11 2 0.015Low 11 4 7

(b)OPRT mRNA Total OPRT protein P valueexpression n = 19 expression

High Low

High 11 8 3 0.040Low 8 2 6

(c)DPD mRNA Total DPD protein P valueexpression n = 21 expression

High Low

High 12 11 2 0.026Low 9 3 5

Table 4. Univariate and multivariate analyses of prognostic factors in pancreatic cancer

Independent factors Univariate (P) Multivariate (P) Relative risk 95% confi dence interval

TP/DPD ratio 0.0016 0.0126 8.786 1.593–48.470Lymphatic invasion (ly0 versus ly1/2/3) 0.0136 0.5637Local advance (T2 versus T3/4) 0.0149 0.8786Stage (stage I/II/III versus IV) 0.0058 0.3030

0

20

40

60

80

100

0 1 2 3 4 5

(%)

p<0.02

Low TP/DPD (n=14)High TP/DPD (n= 7)

Surv

ival

rate

(%)

Time after operation (years)

with the infi ltration growth pattern and that OPRT mRNA expression correlated with histological type. Furthermore, patients with pancreatic cancer who had a high TP/DPD ra-tio had a poor prognosis compared with patients who had a low TP/DPD ratio. On the other hand, TP, OPRT, and DPD mRNA expressions were correlated well with TP, OPRT, and DPD protein expressions, respectively. These fi ndings suggest that TP and OPRT play an important role in pancre-atic cancer and that the TP/DPD ratio has prognostic value in patients with pancreatic cancer.

TP is identical to platelet derived-endothelial cell growth factor (PD-ECGF), which has angiogenic activity. Angio-genesis facilitates rapid tumor growth, enabling the vascu-larized tumor to extend vertically into the deep tissue beneath the basement membrane. Several reports have demonstrated that microvessel density is correlated with the metastasis and recurrence of cancer. TP protein expres-sion was reported to be correlated with proliferating cell nuclear antigen (PCNA) and to be a prognostic marker in human cancer. TP protein expression also suppressed apop-tosis induced by hypoxia in cancer cells.26 Furthermore, many reports indicate that cancer patients with high TP protein expression levels have a poor prognosis. Therefore,

TP has received additional attention as a possible prognos-tic factor.27,28 Although pancreatic cancer is well known to be a hypovascular tumor, angiogenesis may play an impor-tant role in its development. In the present study, TP mRNA expression in pancreatic cancer showed no correlation with any clinicopathological factors, except for the infi ltration pattern. TP may enhance the invasion of tumor cells through the induction of matrix metallo proteinase (MMP)-1 MMP-7, and MMP-9, and TP may regulate infi ltration as well as angiogenesis, possibly through regulating cell adhesion molecules. Further study is required to investigate the cor-relation between TP mRNA and proangiogenic and cell adhesion molecules such as vascular endothelial growth fac-tor (VEGF), basic fi broblast growth factor (bFGF), and the MMP family in pancreatic cancer.

OPRT, an enzyme which also converts 5-FU to an active nucleotide, is anticipated to play a key role in the fi rst step of the induction of the 5-FU antitumor effect, leading to DNA synthesis inhibition and RNA dysfunction. Thus, OPRT, together with TS, DPD, or TP is thought to be largely associated with the induction of the 5-FU antitumor effect. In addition, OPRT activity may be correlated posi-tively with the activity levels of TS and TK, which are the key enzymes for DNA synthesis in the de novo and salvage pathways of. In bladder carcinoma, it was reported that OPRT activity was upregulated compared with the activity in normal bladder and that OPRT may be of prognostic value.30 Although little is known about OPRT expression in

Fig. 2. Comparison of overall survival according to high and low TP/DPD ratios in tumor tissues. The survival of patients with a low TP/DPD ratio was signifi cantly longer than that of patients with a high ratio (P = 0.02; log-rank test)

117

pancreatic cancer, there was a positive correlation between OPRT mRNA expression and histological type in patients with pancreatic cancer in the present study. OPRT may play a potential role in regulating the malignant potential of pancreatic cancer.

We found that there was no relationship between DPD mRNA expression and any clinicopathological factors in pancreatic cancer. DPD is present mainly in the liver and more than 80% of a given 5-FU dose is thought to be catabolized in the liver.1,31 As a result, the expression level of DPD infl uences selective cytotoxicity and is important in predicting chemosensitivity to 5-FU.32,33 Another report has suggested that a higher expression of DPD protein was associated with advanced stage and vessel invasion in colorectal cancer.34 Because the regulation of DPD expres-sion and its involvement with tumor progression remains unclear, we could not reach any – even minimal – conclu-sions about the role of this enzyme, except in the catabolism of 5-FU.

The TP/DPD ratio shows a signifi cant correlation with sensitivity to fl uoropyrimidine analogues, such as 5′-DFUR, and to capecitabine, as it has been found that chemosensi-tivity to 5′-DFUR and capecitabine was refl ected by differ-ences between the catabolic and metabolic enzymes.18,19 In addition, some reports have stated that a higher TP/DPD ratio in a tumor was associated with poorer clinical out-come,26,35 while others reported that patients with higher TP/DPD ratios in the tumor tended to have longer recur-rence-free periods with 5′-DFUR and capecitabine adju-vant chemotherapy after surgery. In particular, patients with high TP and low DPD values had long disease-free survivals with adjuvant chemotherapy after surgery.19

In the present study, TP and DPD alone were inadequate to predict a patient’s prognosis. Therefore, we investigated the relative preponderance of TP and DPD mRNA expres-sion with respect to survival. Considering the negative ef-fect of TP mRNA expression and the favorable prognostic effect of DPD mRNA,35 we calculated the TP/DPD ratio for the combined analysis of TP and DPD mRNA status. The relative risk of cancer death from tumors with a high TP/DPD ratio was greater than the risk of cancer death from tumors with only high TP mRNA expression.

Indeed, we found that a higher TP/DPD ratio in the tu-mor showed a poor prognosis and that the TP/DPD ratio showed no correlation with any clinicopathological features in the patients with pancreatic cancer in this study. Further-more, although the TP/DPD ratio was related to survival and was an independent prognostic factor in our univariate and multivariate analyses, the established risk factors (lym-phatic invasion, local advance, and stage) were not signifi -cant in the multivariate analysis. On this basis, the established risk factors (lymphatic invasion, local advance, and stage) would appear to be of limited value in predicting tumor progression in pancreatic cancer. At the moment, it is still unclear whether the TP/DPD ratio itself is useful as a prog-nostic parameter or as a predictive indicator for the effi cacy of fl uoropyrimidine-based chemotherapy for pancreatic cancer.

The OPRT/DPD ratio, on the other hand, has been re-ported to be a predictor of the antitumor effect of tegafur-uracil (UFT) and leucovorin (LV) regimens. UFT contains the 5-FU prodrug tegafur, which is converted by the P450 drug-metabolizing enzyme in the liver36. Recent reports have disclosed that, among the three pathways (i.e., TP,

Table 5. Correlation between TP, OPRT, and DPD and chemotherapeutic outcome

Patient Age Sex TP OPRT DPD TP/DPD OPRT/DPD Postoperative Survivalnumber (years) chemotherapy (days)

1 65 F Low High High Low High 5-FU 154 2 60 M Low – – – – TAM 110 3 78 F High – High High – Tegafur uracil+TAM 76 4 36 M High High High High Low 5-FU+CDDP 193 5 70 F High – High Low – 5FU+TAM 353 6 68 M Low Low Low High High TAM 307 7 53 M Low High Low Low High 5-FU 137 8 60 F Low Low High Low Low 5-FU+CDDP 820 9 55 M Low Low Low Low High 5-FU+ADM+MMC 46410 75 F High High High Low Low Tegafur uracil 10911 71 F Low High Low Low High – 18112 57 M High Low Low High Low 5-FU+CDDP 15213 65 M Low Low Low Low High Tegafur uracil 208414 58 M Low – – – – 5-FU+CDDP 33715 58 M Low High High Low High Tegafur uracil 88016 74 M Low – – – – – 73017 52 M High High Low High High GEM 15118 65 F High High High Low Low 5-FU 116319 53 M High High High Low Low Tegafur uracil, GEM 57520 69 M High – – – – Tegafur uracil, GEM 27021 59 M High High High Low High 5-FU+CDDP+GEM 27022 77 M High Low High High Low GEM 14323 62 F High Low High High Low GEM 8724 54 M Low High Low Low High GEM 57025 79 F High Low High Low Low – 283

TAM, Tamoxifen; CDDP, cisplatin; ADM, adriamycin; MMC, mitomycin C; GEM, gemcitabine

118

DPD, and OPRT), the OPRT pathway was the most impor-tant for activating 5-FU. Hence, the ratio of OPRT and DPD was also evaluated in this study. The OPRT/DPD ratio was signifi cantly higher in responding tumors than in nonresponding ones, and patients with tumors with a high OPRT/DPD ratio survived longer than those with tumors with a low OPRT/DPD ratio. Therefore, further studies are necessary to evaluate whether the OPRT/DPD ratio could be adopted as a possible predictor of the effectiveness of other fl uoropyrimidine analogues.

In conclusion, using the TP/DPD ratio as a prognostic parameter for pancreatic cancer may help select patients for more intensive surgical approaches. However, the conclu-sions have been drawn from a retrospective study of a small number of patients. Moreover, since the time that these patients were treated, various advances have occurred, not only in surgical procedures but also in adjuvant therapies for pancreatic cancer. Further study is needed to investigate the relationship between TP and OPRT mRNA expression (and the ratio of these expressions to DPD mRNA expres-sion) and features such as prognosis and the antitumor ef-fect of fl uoropyrimidine-based chemotherapy in pancreatic cancer.

Acknowledgments This study was supported by part of The Board for Cancer Research Project, a cooperative project of TAIHO Pharma-ceutical Co., Ltd and The University of Tokushima.

References

1. Heggie GD, Sommadossi JP, Cross DS, et al. (1987) Clinical phar-macokinetics of 5-fl uorouracil and its metabolites in plasma, urine, and bile. Cancer Res 47:2203–2206

2. Pinedo HM, Peters GJ (1988) 5-Fluorouracil: biochemistry and pharmacology. J Clin Oncol 6:1653–1664

3. Santi DV, McHenry CS, Sommer H (1974) Mechanisms of interac-tion of thymidylate synthase with 5-fl uorodeoxyuridine. Biochem-istry 13:471–480

4. Miyazono K, Okabe T, Urabe A, et al. (1987) Purifi cation and properties of an endothelial cell growth factor from human plate-lets. J Biol Chem 262:4098–4103

5. Ishikawa F, Miyazono K, Hellman U, et al. (1989) Identifi cation of angiogenic activity and the cloning and expression of platelet-derived endothelial cell growth factor. Nature 338:556–562

6. Sumizawa T, Furukawa T, Haraguchi M, et al. (1993) Thymidine phosphorylase activity associated with platelet-derived endothelial cell growth factor. J Biochem 114:9–14

7. Haraguchi M, KMiyadera K, Uemura K, et al. (1994) Angiogenic activity of enzymes. Nature 368:198

8. Miyadera K, Sumizawa T, Haraguchi M, et al. (1995) Role of thy-midine phosphorylase activity in the angiogenic effect of platelet-derived endothelial cell growth factor/thymidine phosphorylase. Cancer Res 55:1687–1690

9. Miwa M, Cook A, Ishitsuka H. (1986) Enzymatic cleavage of vari-ous fl uorinated pyrimidine nucleosides to 5-fl uorouracil and their antiproliferative activities in human and murine tumor cells. Chem Pharm Bull 34:4225–4232

10. Ishitsuka H. (2000) Capecitabine: preclinical pharmacology stud-ies. Invest New Drugs 18:343–354

11. Shirasaka T, Shimamoto Y, Fukushima M. (1993) Inhibition by oxonic acid of gastrointestinal toxicity of 5-fl uorouracil without loss of its antitumor activity in rats. Cancer Res 53:4004–4009

12. Inaba M, Mitsuhashi J, Sawada H, et al. (1996) Reduced activity of anabolizing enzymes in 5-fl uorouracil-resistant human stomach cancer cells. Jpn J Cancer Res 87:212–220

13. Kanai F, Kawakami T, Hamada H, et al. (1998) Adenovirus-mediated transduction of Escherichia coli uracil phosphoribosyl-transferase gene sensitizes cancer cells to low concentrations of 5-fl uorouracil. Cancer Res 58:1946–1951

14. Inaba M, Sawada H, Sadata A, et al. (1999) Circumvention of 5-fl uorouracil resistance in human stomach cancer cells by uracil phosphoribosyltransferase gene transduction. Jpn J Cancer Res 90:349–354

15. Salonga D, Danenberg KD, Johnson M, et al. (2000) Colorectal tumors responding to 5-fl uorouracil have low gene expression lev-els of dehydropyrimidine dehydrogenase, thymidylate synthase, and thymidine phosphorylase. Clin Cancer Res 6:1322–1327

16. Ishikawa Y, Kubota T, Otani Y, et al. (2000) Dihydropyrimidine dehydrogenase and messenger RNA levels in gastric cancer: pos-sible predictor for sensitivity to 5-fl uorouracil. Jpn J Cancer Res 91:105–112

17. Etienne MC, Cheradame S, Fischel JL, et al. (1995) Response to fl uorouracil therapy in cancer patients: the role of tumoral dihy-dropyrimidine dehydrogenase activity. J Clin Oncol 13:1663–1670

18. Ishikawa T, Sekiguchi F, Fukase Y, et al. (1998) Positive correla-tion between the effi cacy of capecitabine and doxifl uridine and the ratio of thymidine phosphorylase to dihydropyrimidine dehydro-genase activities in tumors in human cancer xenografts. Cancer Res 58:685–690

19. Terashima M, Fujiwara H, Takagane A, et al. (2002) Role of thy-midine phosphorylase and dihydropyrimidine dehydrogenase in tumour progression and sensitivity to doxifl uridine in gastric can-cer patients. Eur J Cancer 38:2375–2381

20. Ichikawa W, Uetake H, Shirota Y, et al. (2003) Both gene ex-pression for orotate phosphoribosyltransferase and its ratio to dihydropyrimidine dehydrogenase infl uence outcome following fl uoropyrimidine-based chemotherapy for metastatic colorectal cancer. Br J Cancer 89:1486–1492

21. Japan Pancreas Society (2003) Classifi cation of pancreatic carci-noma, Second English edition. Kanehara, Tokyo

22. Gibson UE, Heid CA, Williams PM. (1996) A novel method for real time quantitative RT-PCR. Genome Res 6:995–1001

23. Ichikawa W, Takahashi T, Suto K, et al. (2004) Thymidylate synthase and dihydropyrimidine dehydrogenase gene expression in relation to differentiation of gastric cancer. Int J Cancer 112:967–973

24. Kamoshida S, Matsuoka H, Shiogama K, et al. (2004) Immuno-histochemical analysis of thymidylate synthase, p16INK4a, cyclin-dependent kinase 4 and cyclin D1 in colorectal cancers receiving preoperative chemotherapy: signifi cance of p16INK4a mediated cel-lular arrest as an indicator of chemosensitivity to 5-fl uorouracil. Pathol Int 54:564–575

25. Kamoshida S, Matsuoka H, Shiogama K, et al. (2004) Im-munohistochemical evaluation of thymidylate synthase (TS) and p16INK4 in advanced colorectal cancer: implication of TS expression in 5-FU-based adjuvant chemotherapy. Jpn J Clin Oncol 34:594–601

26. Konno S, Takebayashi Y, Aiba M, et al. (2001) Clinicopathological and prognostic signifi cance of thymidine phosphorylase and pro-liferating cell nuclear antigen in gastric carcinoma. Cancer Lett 166:103–111

27. Nishimura G, Terada I, Kobayashi T, et al. (2002) Thymidine phosphorylase and dihydropyrimidine dehydrogenase levels in primary colorectal cancer show a relationship to clinical effects of 5′-deoxy-5-fl uorouridine as adjuvant chemotherapy. Oncol Rep 9:479–482

28. Mori K, Hasegawa M, Nishida M, et al. (2000) Expression levels of thymidine phosphorylase and dihydropyrimidine dehydroge-nase in various human tumor tissues. Int J Oncol 17:33–38

29. Fujioka S, Yoshida K, Yanagisawa S, et al. (2001)Angiogenesis in pancreatic carcinoma. Cancer 92:1788–1797

30. Mizutani Y, Wada H, Fukushima M, et al. (2004) Prognostic sig-nifi cance of orotate phosphoribosyltransferase activity in bladder carcinoma. Cancer 100:723–731

31. Diasio RB, Lu Z (1994) Dihydropyrimidine dehydrogenase activity and fl uorouracil chemotherapy. J Clin Oncol 12:2239–2242

32. Etienne MC, Cheradame S, Fischel JL, et al. (1995) Response to fl uorouracil therapy in cancer patients: the role of tumoral

119

dihydropyrimidine dehydrogenase activity. J Clin Oncol 13:1663–1670

33. Nita ME, Tominaga O, Nagawa H, et al. (1998) Dihydropyrimidine dehydrogenase but not thymidylate synthase expression is associ-ated with resistance to 5-fl uorouracil in colorectal cancer. Hepato-gastroenterology 45:2117–2122

34. Hiroyasu S, Shiraishi M, Samura H, et al. (2001) Clinical relevance of the concentrations of both pyrimidine nucleoside phosphorylase

(PyNPase) and dihydropyrimidine dehydrogenase (DPD) in colorectal cancer. Jpn J Clin Oncol 31:65–68

35. Fujiwaki R, Hata K, Nakayama K, et al. (2000) Gene expression for dihydropyrimidine dehydrogenase and thymidine phosphory-lase infl uences outcome in epithelial ovarian cancer. J Clin Oncol 18:3946–3951

36. Kohne CH, Peters GJ. (2000) UFT: mechanism of drug action. Oncology 14:13–18