Gastrointestinal cancers are major causes of cancer death throughout the world 12. Recent advances of multimodal treatments have been improving the prognosis. Chemotherapy with combinations of new drugs including fluoropyrimidine, irrinotecan, and oxaliplatin, has greatly contributed to the prolonged prognosis 345678.

5-Fluorouracil (5-FU) is a key drug in combination chemotherapy and an evaluation of the predictability of 5-FU sensitivity is important to exclude those patients who would experience adverse effects. Among the molecular markers of 5-FU activity, thymidylate synthase (TS), dihydropyrimidine dehydrogenase (DPD), orotate phosphoribosyltransferase (OPRT), and thymidine phosphorylase (TP) are reported to be highly predictive of 5-FU sensitivity 91011121314151617181920212223. 5-FU is catabolized to dihydrofluorouracil and inactivated by DPD. Thymidylate synthase is an essential DNA synthetic enzyme that is suppressed by 5-fluoro-deoxyuridine-monophosphate (FdUMP), an active metabolite of fluorouracil 9. FdUMP and TS form covalent ternary complexes with 5, 10-methylene-tetrahydrofolate that subsequently inhibit DNA synthases 910. Colorectal cancer with both low DPD and low TS mRNA expressions has been reported to show greater antitumor effects in 5-FU-based chemotherapy 1112. Fluorouracil is converted to active metabolites by phosphorylation through three different pathways, and TP and OPRT are the key enzymes in two of these pathways 13. Thymidine phosphorylase is an enzyme that activates 5'-deoxy-5-fluorouridine to 5-FU and then 5-FU to 5-fluoro-2'-deoxyuridine. Orotate phosphoribosyltransferase is an enzyme that converts 5-FU to 5-fluorouridine-5'-monophosphate (FUMP) and is considered to predominantly inhibit RNA synthesis. High expression of TP in a tumor is correlated with a high response rate to 5'-deoxy-5-fluorouridine 1415 and high expression of OPRT in a tumor is correlated with sensitivity to 5-FU 1316. For these reasons, many studies have reported that the activities of these enzymes have been associated with sensitivity to 5-FU-based chemotherapy in gastric cancer 171819 and colorectal cancer 111220212223.

These observations were based on gene expressions evaluated by using fresh frozen materials, which were composed of cancer cells, stromal cells in the cancer tissues, and even normal tissues. Because gastric and breast cancers contain large amounts of stromal cells in the cancer tissues, gene expression evaluated by ordinary methods reflects that of the cancer tissues, but not cancer cells. To evaluate gene expression of the cancer cells alone, it is essential to isolate the cancer cells from the stromal cells. To achieve this, we used a laser capture microdissection plus the real time reverse transcription-polymerase chain reaction (RT-PCR) method (LCM+RT-PCR) on formalin-fixed paraffin-embedded (FFPE) samples. In our previous report, in which TS, DPD, and TP gene expressions in breast cancers were evaluated by this method, we disclosed that the gene expressions in cancer cells were significantly different from those in stromal cells 24. Gastric cancer, which contains large amounts of stromal cells in the cancer tissues, may show different gene expression between cancer cells and stromal cells as in breast cancers. In this study, gene expression levels of TS, DPD, TP, and OPRT in cancer cells of gastric cancer tissues were separately quantified from those in stromal cells by using the LCM+RT-PCR method. We also investigated those genes in colorectal cancer, which contains small amounts of stromal cells.

In this study, all the gene expressions were successfully estimated separately in cancer cells and cancerous stroma of human gastric and colon cancers using the LCM+RT-PCR technique. All of them showed significantly different expression levels between cancer cells and cancerous stroma.

In both gastric and colon cancers, higher gene expression levels of TS and OPRT were observed in cancer cells than in cancerous stroma and a lower gene expression level of DPD was observed in cancer cells than in cancerous stroma. These gene expressions might be controlled by a common regulatory mechanism in gastric and colon cancers. In our previous study, these genes showed the same tendency in breast cancer 24. TS is the enzyme for DNA synthesis and cell proliferation and because cancer cells grow more rapidly than normal cells, TS gene expression is thought to be up-regulated in cancer cells. Several studies have revealed that the expression of OPRT is increased in several types of carcinoma, including gastric and colorectal carcinomas 282930. OPRT is a nucleotide metabolic enzyme that is essential for cell proliferation. Thus, OPRT gene expression is thought to be up-regulated in cancer cells, like TS gene expression 3132. Evaluations of comparisons of the expression levels of DPD in cancer cells and in normal tissues are controversial. Differing results might be affected by the use of tumor tissue samples with various amounts of stromal tissue. However, in the recent study, the amount of stromal cells was taken into consideration, and the results agreed with those of previous reports that DPD expression in cancer cells was lower than in normal tissues and in stromal cells 3334. McLead et al. reported that the down-regulation of DPD expression may create a favorable environment for tumor growth. Low expression levels of DPD and decreased catabolism of uracil in cancer cells suggest that pyrimidine nucleotide pools increase 35. On the other hand, regarding the gene expression level of TP, the opposite result was observed. Although the gene expression level of TP was higher in cancer cells than in cancerous stroma in gastric cancer, it was lower in cancer cells than in cancerous stroma in colon cancer. It has been reported that certain solid tumors, including gastric and breast cancers, expressed elevated levels of TP as compared with stromal tissues 36373839. In colon cancer, several immunohistochemical studies reported that cancer cells had TP expression 3940, while other studies reported that most cells expressing TP were stromal cells, especially macrophages and lymphocytes 4142. It was shown by analyzing the expression level of TP in cancer cells and cancerous stroma separately that cancer cells had TP expression, although stromal cells had higher TP expression than cancer cells. These current results regarding the expression of TS, DPD, OPRT, and TP agree with the previous data.

In previous studies, biochemical assays, immunohistochemistry, enzyme-linked immunosorbent assay (ELISA), and reverse transcription-polymerase chain reaction (RT-PCR) have been used to evaluate the protein expression related to enzymes catabolizing 5-FU. Biochemical assays are often impossible to perform with minimal clinical samples. Many institutions have stored tissue samples as paraffin-embedded specimens after surgical resection, so immunohistochemical assays are convenient and inexpensive when performed on paraffin-embedded specimens. However, this method does not yield quantified results. Amplification by PCR may be performed on fresh frozen samples from resected cancer specimens to study gene expressions. However, the results are dependent on how promptly the samples were collected and stored. This method is not practical for the purpose of retrospective studies. In fresh frozen samples of cancer tissues, contamination with cancerous stroma and even normal tissue cannot be avoided. It is known that gastric and breast cancers contain large amounts of stromal tissue. We thought that the contaminations might influence the results of PCR, so we adopted LCM+RT-PCR. Laser capture microdissection provides selective isolation of defined cell populations from heterogeneous tissue sections 43. Moreover, the availability of real-time RT-PCR technology combined with the extraction of RNA from paraffin-embedded specimens allows quantitative and accurate measurement of gene expressions 2544. This novel method made it possible to analyze only cancer cells, so in spite of amounts of cancerous stroma, we could analyze intratumoral gene expressions equally. Furthermore, this method has another advantage. Because cancer cells and cancerous stroma are extracted separately within the same paraffin section, we can perform the analysis from a small amount of resected specimen.

The time when various chemotherapies, including monoclonal antibody therapy has come, but the chemotherapy with 5-FU-based regimen has still played an important role as the treatment for many cancers. Accordingly it is thought to be useful to estimate gene expressions of enzymes, that are related to catabolism of 5-FU, in Ca and Str separately by LCM+RT-PCR. Though we need further investigate the relationship between those gene expressions and 5-FU sensitivity in clinical setting, we believe that with this technique it may be possible to predict the sensitivity of the agents before treatment using a small amount of a biopsy specimen and it leads to establish a tailor-made treatment for cancer patients.

In this study, by using LCM+RT-PCR, we could analyze gene expressions in cancer cells and stromal tissues separately in FFPE gastric and colon cancer specimens. This method may make it possible to accurately analyze gene expressions in cancer cells from a small amount of a biopsy specimen in spite of the amount of stromal tissue and intratumoral heterogeneity.

The authors declare that they have no competing interests.

HM prepared the formalin-fixed paraffin-embedded (FFPE) samples and drafted the manuscript. KD and PVD performed the LCM+RT-PCR. HU performed the statistical analysis. KS oversaw this study. All authors read and approved the final manuscript.