¹University of Southern California/Norris Comprehensive Cancer Center, Los Angeles, California

²Department of Surgery, University of California San Francisco, San Francisco, CA, USA

Thymidylate synthase (TS) catalyses the conversion of deoxy-uridylate to deoxy-thymidylate and is essential for DNA synthesis. The human TS gene promoter is polymorphic, having either double or triple tandem repeats of a 28-bp sequence. Here we determined the significance of this polymorphism in humans and its prediction for clinical outcome of patients with metastatic colorectal cancer treated with 5-fluorouracil. The TS mRNA level was analyzed using RT-PCR. Individuals homozygous for the triple repeat variant (L/L) had 3.6 times higher TS mRNA levels compared to those homozygous for the double repeat variant (S/S) in tumor tissue (P = 0.004). We tested 50 patients with disseminated colorectal cancer who received 5-FU treatment to determine whether this TS polymorphism will predict clinical outcome. We found individuals with S/S genotype had a response rate of 50% (4/8) when compared to 9% (2/22) in those with L/L and 15% (3/20) in those with S/L genotype (P = 0.041). Patients with L/L had less severe side effects to 5-FU (P = 0.008). The data suggest that genotyping for the TS polymorphism may have the potential to identify patients more likely to respond to 5-FU based chemotherapy. The Pharmacogenomics Journal (2001) 1, 65-70.

TS-polymorphism; TS-gene expression; colorectal cancer; 5-FU resistance; fluoropyrimidine chemotherapy

5-FU, 5-Fluorouracil; TS, Thymidylate synthase; TP, Thymidine phosphorylase; DPD, Dihydropyrimidine dehydrogenase

INTRODUCTION

Colorectal cancer is a leading cause of morbidity and mortality with about 300000 new cases and 200000 deaths in Europe and USA each year.¹ Since its introduction 4 decades ago, 5-fluorouracil (5-FU) has remained the mainstay of chemotherapeutic regimens for colorectal cancer, both in the metastatic and adjuvant settings.² The mechanism of action of 5-FU is through the inhibition of thymidylate synthase (TS). This enzyme catalyses the intracellular conversion of deoxyuridylate to deoxythymidylate which is the sole de novo source of thymidylate, an essential precursor for DNA synthesis.³ The active metabolite of 5-FU, 5-fluorodeoxyuridylate (5FdUMP), binds to TS and inhibits it by forming a stable ternary complex.⁴ The human thymidylate synthase gene (hTS) is polymorphic with either double or triple tandem repeats of a 28 base-pair sequence downstream of the cap-site in the 5' terminal regulatory region.⁵ In in vitro studies, the activity of a reporter gene linked to the 5' terminal fragment of the hTS gene with triple tandem repeats was 2.6 times higher than that with double tandem repeats.⁶ Thus this polymorphic region appears to be functional and may modulate TS gene expression. In addition, we have previously shown that the intratumoral TS mRNA level is a determinant of response to 5-FU chemotherapy and survival in patients with colon and gastric cancer.^7,8 Hence if this polymorphism modulates the TS gene expression, then knowledge of the TS genotype could predict response to fluoropyrimidine chemotherapy and survival. Based on these previous findings, we conducted a pilot retrospective study to investigate the significance of this polymorphism on TS mRNA levels and on clinical outcome of patients with colorectal cancer.

RESULTS

TS mRNA Level and TS Genotype

We found that among 52 patients with metastatic colorectal cancer (3C-92-2 and CALGB trial) analyzed for TS mRNA expression, 15 (29%) were homozygous for the triple repeat, 26 (50%) were heterozygous, and 11 (21%) were homozygous for the double repeat variant within the human TS promoter region. Table 1 summarizes the mean intratumoral TS mRNA levels in tumor tissue according to the three different genotypes. There was a significant association between TS mRNA expression and increasing number of the 28-bp repeats within the 5'-untranslated region of the human TS gene. Subjects with the S/S genotype had significantly lower TS mRNA levels (2.60 (1.39, 4.87)) than both S/L subjects (5.53 (3.68, 8.31); P = 0.050) and L/L subjects (9.42 (5.51, 16.12); P = 0.004) by pair-wise comparison.

The TS mRNA level in normal liver tissue of 26 patients (CALBG) was also examined. We found seven patients (27%) with the L/L genotype, 14 patients (54%) with the S/L genotype, and five patients (19%) with the S/S genotype. A similar association was observed between TS mRNA expression and genotype: L/L group 8.18 (4.90, 13.67), the S/L group 4.41 (3.07, 6.33), and the S/S group 3.19 ((1.74, 5.85), P = 0.051, Table 2).

The TS mRNA expression level in tumor tissue was 2.45-fold increased compared to the adjacent normal liver tissue. The ratio of TS expression in tumor tissue to TS expression in normal tissue was inverse associated with the number of 28-bp tandem repeats in the 5'-untranslated region of the TS gene (Table 3, P < 0.001).

TS Genotype and Clinical Outcome Under Fluoropyrimidine Chemotherapy

We evaluated 50 patients with disseminated colorectal cancer (SWOG 9420 and 3C-92-2) for response and survival treated with protracted infusion of 5-FU. Table 4 summarizes the demographic information of these patients.

Patients confirmed for the S/S genotype showed a higher response rate (50% (4/8)) to 5-FU compared to 15% (3/20) in the S/L group and 9% (2/22) in the L/L group (P = 0.041, Fisher's Exact test, Table 5). Forty-nine out of the above 50 patients were evaluable for toxicity. A significant inverse association was seen between the number of 28-bp tandem repeats in the 5'-untranslated region of the TS gene and the severity of toxicity (P = 0.008, Jonckheere-Terpstra exact test, Table 5). In 63% (5/8) of patients with S/S genotypes a toxicity grade 3 ('severe') was observed compared to 27% (6/22) in the L/L group and 32% (6/19) in the S/L group. A mild toxicity (grade 1) was found in 41% (9/22) of patients homozygous for the L allele compared to 5% (1/19) in the heterozygous group and 0% (0/8) in the S/S group. Toxicity-grade 0 ('none') as well as grade 4 ('life-threatening') was not seen in any of those 49 patients. Patients with the S/S genotype had a longer median survival of 16.2 months when compared to 8.3 months and 8.5 months in those with the S/L genotype and L/L genotype, however it did not reach statistical significance (P = 0.37, Log-rank test, data not shown). The median follow-up was 2.9 years.

DISCUSSION

5-fluorouracil-based therapy of metastatic colorectal cancer has an overall response rate of 26% in our patient population.⁷ However there are inter-individual differences in response, survival and toxicity in patients treated with 5-FU. Recently, the measurement of intratumoral TS mRNA expression has been established as a predictor of response and survival to 5-fluorouracil chemotherapy. It has been shown that low expression levels of TS mRNA were associated with a higher probability of response to 5-FU-based treatment and longer survival. Patients with a TS mRNA level higher than 4.1 did not respond to 5-FU-based chemotherapy.⁷ Similar results were seen in gastric cancer patients treated with a 5 FU-based regimen.⁸

Regulation of intratumoral TS expression is not very well understood. Recent studies have demonstrated that mutant p53 is associated with higher TS protein and gene expression.^9,10 Further it has been reported that a 28-bp tandem repeat polymorphism is associated with TS activity in vitro.⁶

This study is demonstrating for the first time a significant association between TS mRNA level in both metastasized colorectal tumors and normal tissue and a 28-bp tandem repeat polymorphism in the 5'-untranslated region of the TS gene. A higher intra-tumoral TS mRNA expression was observed with an increasing number of the 28-bp tandem repeats. Recent data showed that TS protein expression in gastrointestinal tumors is associated with this TS polymorphism in the 5'-untranslated region.¹¹ These data suggest that this TS polymorphism may be a significant predictor of TS gene and protein expression.

In our study we also found an association between the TS genotype and the TS mRNA expression in normal liver tissue suggesting the significance of this polymorphism on TS gene regulation as it has been concluded from in vitro studies.⁶ Comparing the TS mRNA expression in metastasized tumor tissue and in adjacent normal liver tissue, we found significant higher TS levels in the tumor. These data suggest that the TS mRNA expression in tumor tissue may be due to a transcriptional or post-transcriptional process. Recently a 6-bp deletion in the 3'-untranslated region of the TS gene was revealed and suggested that this new polymorphism may alter mRNA instability and translation.¹²

We were able to show that the presence of triple repeats (L) predicted not only intratumoral TS mRNA expression, but also response to 5-FU based chemotherapy. Patients with the S/S genotype had the most favorable response to 5-FU therapy, because the intratumoral TS mRNA expression is significantly lower in this group. The lower number of patients homozygous for the S/S genotype might be due to the fact that they comprise only about 10-20% of the population.^13,14 The differences in the frequency of these tandem repeats in Chinese and Caucasians may explain differences in clinical outcome to 5-FU chemotherapy in those ethnic groups. Taking these observations together, our data suggest that screening of patients for this TS polymorphism may have the potential to select patients of whom 50% will respond to 5-FU-based treatment. Recently two novel alleles in the enhancer region of the TS gene containing four and nine copies of the tandem repeat have been found in African populations. The frequency of those alleles in Caucasians appears to be very low, however it might be interesting to evaluate whether an association between these high copy numbers and the TS mRNA expression can be found.¹⁵

TS is not the only enzyme critical in the 5-FU metabolism. It has been shown that high levels of thymidine phosphorylase (TP) and dihydropyrimidine dehydrogenase (DPD) gene expression are associated with low sensitivity to 5-FU treatment in gastrointestinal tumors. Recently our group demonstrated that using all three genes, TS, TP and DPD, we were able to predict response to 5-FU in all patients.^16,17 Using TS mRNA expression alone, patients with TS expression levels <3.5 had a response rate of 52%. In this study patients with the S/S genotype showed a response rate of 50% indicating the clinical significance of this TS polymorphism. Another mechanism of 5-FU resistance was shown to be associated with instability of the TS polypeptide, caused by an amino acid substitution.¹⁸

Based on our data, that the TS genotype predicts not only TS mRNA expression in metastasized colon tumors but also in normal liver tissue, we tested whether genotyping of TS would not only predict for response but also for toxicity to 5-FU. Patients with the L/L genotype had significantly less toxicity, when compared to the S/L and S/S genotype under 5-FU-based chemotherapy. This may be due to TS mRNA expression levels in normal tissue. The increased TS mRNA expression in both normal and tumor tissue of patients with the L/L genotype protects the cells against damage by 5-FU treatment due to the low efficacy of TS inhibition. The resulting decreased cell death rate leads to resistance (tumor tissue) and low toxicity (normal cells). On the other hand the data suggest that the lower TS mRNA level in the normal tissue of patients with S/S or S/L genotype may enhance the cytotoxic effects of 5-fluorouracil leading to more severe side effects in these patients.

The survival benefit of the S/S group with a median survival of 16.2 months vs 8.3 and 8.5 months in the S/L and L/L group did not reach statistical significance. However, taking all data together these findings suggest that genotyping for this TS polymorphism may be helpful in selecting patients who most likely benefit from 5-FU chemotherapy. The simple method of PCR amplification of genomic DNA from peripheral blood prior to commencing chemotherapy provides a tool to allow individual chemotherapy based on genomic profiling. Patients unlikely to respond to 5 FU, those with the L/L genotype, may be treated with other agents such as irinotecan or oxaliplatin whose mechanisms of action are independent of TS inhibition. In addition genotyping for the TS polymorphism may become an important tool to identify patients, who require close monitoring or adjustment of the 5-FU dose (S/S group). These findings may also apply to the newer fluoropyrimidine agents like capecitabine and UFT, which are increasingly being used for the treatment of a variety of cancers. Larger prospective clinical studies are necessary to validate our data from this retrospective pilot study.

PATIENTS AND METHODS

Patients Selection

Patients included in this study had disseminated colorectal cancer and were enrolled in the following protocols: Southwest Oncology Group protocol 9420 (24 patients) opened for accrual in May 1995 and closed in May 1999; University of Southern California protocol 3C-92-2 (26 patients) opened for accrual in September 1992 and closed in June 1995; and Cancer and Leukemia Group B protocol 9481 (26 patients). All patients signed an informed consent to participate in the clinical trial and for evaluation of the TS polymorphism. Genotyping for the TS polymorphism was performed on fresh-frozen or paraffin-embedded tissues in all patients. All specimens were taken from metastasized liver tumors. The relation between genotype and intratumoral TS mRNA was determined for 52 patients enrolled in the 3C-92-2 and CALGB 9481 protocols as fresh frozen specimens were not available for TS mRNA quantitation from patients enrolled in the SWOG 9420 protocol. Quantitation for mRNA levels in normal tissues was performed on 26 CALGB (protocol 9481) patients. Analysis of the genotype and clinical outcome was done on 50 patients enrolled in the SWOG 9420 and 3C-92-2, as clinical data are not yet available for the CALGB patients. All patients were previously untreated, had stage IV disease and received 5-FU as definitive therapy (Table 6). All patients had bi-dimensionally measurable disease at the time of protocol entry. Responders to 5-FU therapy were classified as those patients whose tumor burden was decreased by 50% or more (partial response) or completely disappeared (complete response) for at least 6 weeks. Non-responders included those with stable disease or cancer progression. Toxicity was graded according to the NCI criteria version 2 considering grade 0 'none', grade 1 'mild', grade 2 'moderate', grade 3 'severe', and grade 4 'life-threatening'. Survival was computed as the number of months from the initiation of chemotherapy with 5-FU to death of any cause. Patients who were alive at the last follow-up evaluation were censored at that time. The double tandem repeat variant of the TS gene was designated as the S allele whereas the triple repeat was designated as the L allele.

TS Gene Polymorphism

DNA was extracted from frozen or paraffin-embedded tissues using the QiaAmp kit (Qiagen, Valencia, CA, USA). The promoter region of the hTS gene was amplified by polymerase chain reaction (PCR) using the following primers: Primer 1 (sense): 5' GTGGCTCCTGCGTTTCCCCC-3', and Primer 2 (antisense): 5'-GCTCCGAGCCGGCCACAGGCATGGCGCGG-3' as previously described.¹⁴ Briefly, 25 l reaction mixture containing 1.25 mM MgCl₂ was transferred to a thermal cycler (PTC-100 TM, MJ Research Laboratories, Watertown, MA, USA) and amplified for 35 cycles. Each cycle consisted of 1 min at 96°C, 30 s at 60°C and 1 min at 72°C with a final extension phase at 72°C for 5 min. The PCR product was analyzed by electrophoresis on a 4% agarose gel. Homozygotes for the triple repeat variant L/L had 250-bp product, homozygotes for the double repeat variant S/S had 220-bp product and heterozygotes S/L had 220 and 250-bp products (Figure 1).

TS mRNA Quantitation

Messenger RNA was isolated from fresh frozen tissue samples using QuickPrep micro mRNA isolation kit (Pharmacia Biotech), based on the method described by Chomczynski and Sacchi.¹⁹ TS mRNA was quantitated using a quantitative RT-PCR method as previously described. Briefly, TS mRNA was transcribed to cDNA using reverse transcriptase and random hexamers. The cDNA was PCR amplified using TS primers TS 61: T7-"GGGAGA" GAGTTGACCA-ACTGCAAAGAGTG (bases 469-492 of the TS coding sequence) and TS 60: GATGTGCGCA-ATCATGTACGTGAG (bases 697-720 of the TS coding sequence.²⁰ PCR conditions, T7 RNA polymerase transcription and the quantitation procedure are described in detail by Horikoshi et al.²¹

Statistical Analysis

TS mRNA levels were log-transformed prior to the analysis. An analysis of variance (ANOVA) was performed to test for differences in TS expression by TS genotype. Analysis was done for the normal liver tissue and metastatic tumor tissue separately. The overall P-values were based on the F-test from the ANOVA. The SNK (Student-Newman-Keuls) method²² was used for the pair-wise comparisons of the S/S, S/L, and L/L groups. For each genotype, the geometric mean (ie after transformation then using the exponential transformation to convert back to original scale) and the associated 95% confidence interval were used to summarize the TS expression. Fisher's exact test and Jonckheere-Terpstra exact test were used to evaluate the association of genotype with the baseline, and clinical outcome (response to the chemotherapy of 5-FU, and the toxicity). To evaluate the association between TS genotype and survival, Kaplan-Meier plots and logrank test were used.^23,24 Median survival was based on Kaplan-Meier estimate. The relative risks were based on the observed and expected number of deaths as calculated in the Log-rank test statistics.²⁵

DUALITY OF INTEREST

None declared.

Acknowledgements

This work is funded by NIH grants R01 CA 82655, R01 CA74166 and P30 CA14089. JS is supported by Dr Mildred Scheel Stiftung, Bonn, Germany.

Figure 1 Electrophoresis of PCR products on 4% agarose gel showing single 220-bp and 250-bp band in homozygotes for S/S and L/L respectively. Double bands are seen in heterozygotes (S/L).

Table 1 TS genotype and TS mRNA levels in tumor tissue

Table 2 TS genotype in normal tissue

Table 3 Difference of TS expression in normal and tumor tissues by TS genes^a

Table 4 Basic demographic information

Table 5 TS genotype and clinical outcome for patients treated with 5-FU

Table 6 Summary of patients included in study