Predictive markers of capecitabine sensitivity identified from the expression profile of pyrimidine nucleoside-metabolizing enzymes.

Molecular markers predicting sensitivity to anticancer drugs are important and useful not only for selecting potential responders but also for developing new combinations. In the present study, we analyzed the difference in the sensitivity of xenograft models to capecitabine (Xeloda®), 5'-deoxy-5-fluorouridine (5'-DFUR, doxifluridine, Furtulon®) and 5-FU by comparing the mRNA levels of 12 pyrimidine nucleoside-metabolizing enzymes. Amounts of mRNA in the tumor tissues of 80 xenograft models were determined by real-time RT-PCR and mutual correlations were examined. A clustering analysis revealed that the 12 enzymes were divided into two groups; one group consisted of 8 enzymes, including orotate phosphoribosyl transferase (OPRT), TMP kinase (TMPK) and UMP kinase (UMPK), and was related to the de novo synthesis pathway for nucleotides, with mRNA expression levels showing significant mutual correlation. In the other group, 4 enzymes, including thymidine phosphorylase (TP) and dihydropyrimidine dehydrogenase (DPD), were involved in the salvage/degradation pathway of the nucleotides, and the mRNA levels of this group were dispersed more widely than that of the de novo group. Antitumor activity was assessed in 24 xenograft models for each drug. The antitumor activity of capecitabine and 5'-DFUR correlated significantly with the mRNA levels of TP and with the TP/DPD ratio, whereas the activity of 5-FU correlated significantly with OPRT, TMPK, UMPK and CD. In a stepwise regression analysis, TP and DPD were found to be independent predictive factors of sensitivity to capecitabine and 5'-DFUR, and UMPK was predictive of sensitivity to 5-FU. These results indicate that the predictive factors for sensitivity to capecitabine and 5'-DFUR in xenograft models may be different from those for 5-FU, suggesting that these drugs may have different responders in clinical usage.