Dihydropyrimidine dehydrogenase pharmacogenetics in Caucasian subjects

Abstract

Aims

Dihydropyrimidine dehydrogenase (DPD) catalyses the reduction of pyrimidines, including the anticancer agent 5-fluorouracil (5FU). Impaired 5FU degradation, through low DPD activity, has led to severe, life-threatening or fatal toxicity after administration of 5FU. Complete DPD deficiency is associated with the inherited metabolic disease thymine uraciluria. Several mutations in the gene encoding DPD have recently been identified, but the phenotype-genotype concordance of these alterations in the general population has not been reported.

Methods

Mononuclear cells were isolated from whole blood and DPD activity was determined after ex vivo incubation with ¹⁴C-5FU followed by h.p.l.c. analysis of 5FU metabolites. Analysis of mutations in the DPD gene at an exon splice site, codons 534, 543, and 732, and a deletion at base 1897 (ΔC1897) were performed in 30 subjects with the lowest and 30 subjects with the highest enzyme activity using PCR-RFLP.

Results

DPD activity was measured in 226 Caucasian subjects and was highly variable (range 19.1–401.4 pmol min⁻¹ mg⁻¹ protein). Mutations were frequently observed at codons 543 (allele frequency 28%), 732 (allele frequency 5.8%), and 534 (allele frequency 0.8%), but were not associated with low DPD activity. There were no splice site or ΔC1897 mutations found in this population.

Conclusions

The five mutations analysed in this study are insufficient for identification of patients at risk for 5FU toxicity or thymine uraciluria. Both the splice site mutation and ΔC1897 are relatively rare in the general Caucasian population. Therefore, identification of further molecular alterations is required to facilitate the use of DPD analysis in genetic diagnosis and cancer therapeutics.

Introduction

The pyrimidines, uracil and thymine, are essential components of DNA and RNA synthesis. They are substrates for a number of enzymatic pathways, including reduction to 5,6 dihydro products by dihydropyrimidine dehydrogenase (DPD; E.C. 1.3.1.2) [1]. DPD activity is found in most human tissues, with highest activity in liver and mononuclear cells, and is the rate-limiting enzyme for the reduction of uracil and thymine to β-alanine and β-aminoisobutyric acid, respectively [1]. Perturbations in DPD may have important implications for neurodevelopment, as this pathway is the only endogenous source of the putative neurotransmitter β-alanine [1, 2]. Mononuclear cell DPD activity is used for clinical phenotype analysis and a 7- to 14-fold range in activity has been observed in population studies [3–6].

DPD deficiency has been associated with a neurological syndrome, thymine uraciluria, characterized by convulsive disorders and mental retardation [2]. Microcephaly, motor retardation, autism and growth retardation have also been observed [2]. Very low or undetectable fibroblast DPD activity and high concentrations of urinary, plasma, and cerebrospinal fluid uracil and thymine are found in infants with this syndrome. Pedigree analysis has demonstrated autosomal recessive inheritance of DPD deficiency [2, 7].

DPD is also a key enzyme for the degradation of the pyrimidine analogue 5-fluorouracil (5FU), thereby regulating its anticancer effect. 5FU has single agent activity in the treatment of colorectal tumours and is a component of combination therapy for breast, head/neck, and other malignancies. Greater than 80% of a 5FU dose is degraded via DPD metabolism [8]. Impaired degradation of 5FU, through low DPD activity, has led to severe, life-threatening or fatal diarrhoea, neutropenia, and/or neurotoxicity after administration of conventional doses of 5FU [7–9].

Several mutations in the DPYD gene which encodes for the DPD enzyme have been recently identified in children with thymine uraciluria and adults with 5FU-related toxicity, all of whom had DPD activity greater than 3 standard deviations below the population mean [6]. These include a G to A transition in a splice donor site which results in skipping of exon 14 [7, 9–11], point mutations at codons 534 and 543 [4, 12], and 732 [12], and a deletion of the cytosine at base 1897 (ΔC1897) which causes premature termination of the protein at codon 637 [13], as detailed in Table 1. Although a limited frequency analysis of several of the mutations has been performed, the phenotype-genotype concordance has not been previously evaluated in a healthy population.

Table 1.

PCR-RFLP information for analysis of five mutations in the DPYD gene. The positions of mutations within the DPYD cDNA are numbered from the translation start site, with the exception of the splice site which involves an intronic mutation.

In this report the frequency of five point mutations in the DPYD gene are described in a Caucasian population. In addition, the relationship between mononuclear cell DPD activity and molecular genotype is examined. By using PCR-based assays, such as that described in this report, prospective information on patient metabolic capacity may be obtained for diagnosis in paediatric genetics and prior to administration of potentially toxic chemotherapy.

Methods

DPD activity and genotype were analysed in 226 subjects randomly recruited from the West of Scotland Blood Transfusion Service in Glasgow, United Kingdom. Samples were collected from donors who were reported to be in good health and not taking any medications. After informed consent was obtained, DPD catalytic activity was measured in mononuclear cells from 20 ml heparinized blood, using a previously described method [3]. In brief, mononuclear cells were purified from 20 ml whole blood by ficoll-hypaque density centrifugation and stored at −20° C for a minimum of 15 days to avoid previously described changes in DPD activity during the 10–14 days after sample preparation and storage [3]. Cytosol was then isolated and incubated with ¹⁴C-5FU (Amersham, Buckinghamshire, UK). The production of 5FU metabolites were quantified by h.p.l.c. analysis with radiodetection. The assay has a coefficient of variation of 8% (n=20). Genomic DNA was purified from 5 ml whole blood using the Nucleon II extraction method (Scotlab, Paisley, UK) and was available for all subjects.

Patient DPD genotype was performed in DNA from 30 subjects with the lowest activity (19.1–69.7 pmol min⁻¹ mg⁻¹ protein) and in 30 subjects with the highest activity (193.6–401.4 pmol min⁻¹ mg⁻¹ protein) using recently described PCR-RFLP assays [4, 7, 12, 13]. In brief, PCR amplification was carried out in a mixture of 10 mm TrisHCl (pH 8.3), 50 mm KCl, 1.5–2.25 mm MgCl₂, 0.2 mm of each dNTP, 2.5 units of Taq polymerase, 100 ng of genomic DNA template, and 100 ng of the primers listed in Table 1. The reaction was carried out at 94° C×1 min, 58° C×1 min, 72° C×2 min for 31 cycles. The PCR products were digested under the conditions described in Table 1 and the resulting fragments were separated by gel electrophoresis and visualized by ethidium bromide staining.

The influence of gender or smoking status on DPD activity was evaluated using the Mann–Whitney test. A Spearman rank correlation was used to define the relationship between DPD activity and either age or order of subject recruitment.

Results

Mononuclear cell DPD activity was detectable in all 226 subjects analysed. DPD activity was highly variable, ranging from 19.1 to 401.4 pmol min⁻¹ mg⁻¹ protein, with a median value of 119.8 pmol min⁻¹ mg⁻¹ protein (Figure 1). The coefficient of variation (CV) of DPD activity was 46.6%. Median DPD activity was not influenced by gender (male 116.6 vs female 122.1 pmol min⁻¹ mg⁻¹ protein; P=0.56, Figure 2a) or age (r_s=0.08). Smoking status was available from all subjects, but was not associated with DPD activity (smoker 133.8 vs nonsmoker 117.2 pmol min⁻¹ mg⁻¹ protein; P=0.16, Figure 2b). No pattern between DPD activity and order of subject accrual was detected (r_s=0.03).

Figure 1.

Distribution of mononuclear cell DPD activity in 226 randomly selected Caucasian subjects.

Figure 2.

Influence of gender (a) and cigarette smoking (b) on mononuclear cell DPD activity. The bar represents the median values for each population.

The frequency of mutations in the gene encoding DPD was assessed using PCR-RFLP techniques. Mutational analysis was conducted for five mutant alleles at codons 534, 543, 732, the splice site mutation, and ΔC1897. Mutations were detected at codon 534 in 1/60 subjects (DPD activity 19.1 pmol min⁻¹ mg⁻¹ protein), resulting in an allele frequency of 0.8% (Table 2). Codon 543 mutations were found in 28/60 subjects (DPD activity 25.9–288.8 pmol min⁻¹ mg⁻¹ protein; 14 in low activity group and 14 in high activity group; Table 2). Five individuals were homozygous for alterations at codon 543 (30.8–241.9 pmol min⁻¹ mg⁻¹ protein; three in low activity group and two in high activity group) and 23 were heterozygous at this loci (25.9–288.8 pmol min⁻¹ mg⁻¹ protein), resulting in an allele frequency of 28%. Seven mutant alleles were detected at codon 732 (one homozygote, five heterozygous; 27.4–200.2 pmol min⁻¹ mg⁻¹ protein), for an allele frequency of 5.8% (Table 2). Four of the six subjects with mutations at codon 732 were in the low activity group, including the homozygote (27.4 pmol min⁻¹ mg⁻¹ protein). Neither the splice site mutation nor ΔC1897 were detected in this study. Three individuals had more than one mutation in DPYD. Two were heterozygous at codons 543 and 732 and had DPD activity values in the lower 10% of the population (25.9 and 53.4 pmol min⁻¹ mg⁻¹ protein). An additional subject who was heterozygous at codon 543 and homozygous for the mutant allele at 732 also had low DPD activity (27.4 pmol min⁻¹ mg⁻¹ protein).

Table 2.

Frequency of mutant alleles in the 30 subjects with the lowest activity and 30 subjects with the highest activity. Percentage allelic frequency is indicated in parenthesis.

Discussion

The characterization of DPD activity in 226 British Caucasian subjects showed a high degree of variability in mononuclear cell enzyme activity. Enzyme activity ranged from 19.1 to 401.4 pmol min⁻¹ mg⁻¹ protein and had a median activity of 119.8 pmol min⁻¹ mg⁻¹ protein. This degree of variability (CV=46.6%) is similar to that seen in North American Caucasian subjects (CV=33.9%) [6] and cancer patients from France and Scotland (CV=36.3–37.8%) [4, 5]. Although the median enzyme activity was lower in this study than that found in previous populations, the unimodal shape of the population DPD activity curve (Figure 1) was similar to that described in the other populations [4–6]. It is not clear whether the low median enzyme activity represents a true difference between this population and the other studies as simultaneous assessment of an alternative population using the same cell preparation and assay methodology was not performed. A similar degree of variation in DPD activity has also been described in tissue from human liver [14], head/neck cancer [15], colorectal tumours [16], and hepatocellular carcinoma [17]. Although median enzyme activity was slightly higher in female subjects (122.1 vs 116.6 pmol min⁻¹ mg⁻¹ protein) this difference was not statistically significant (P=0.56). The general trend for DPD activity to be higher in female subjects has also been described in other studies of mononuclear cell and liver DPD activity but these differences are small and are unlikely to have any clinical impact on 5FU metabolism in patients with cancer [6, 14]. As has been reported in other studies there was no influence of subject age on enzyme activity [3, 5]. Cigarette smoking has been reported to induce a number of drug metabolising enzymes. Although DPD activity was slightly higher in the 40 smokers vs the 186 non smokers (133.8 vs 117.2 pmol min⁻¹ mg⁻¹ protein), this was not statistically significant (P=0.16).

This study provides the first analysis of five specific mutations in the DPYD gene in a population of Caucasian subjects. The mutations had all been identified from analysis of cDNA from patients with severe toxicity to 5FU or children with thymine uraciluria. The approach used in this study was to perform genotype analysis in 30 subjects with the lowest DPD activity and 30 subjects with the highest DPD activity. This provided an enriched environment in which to identify those mutations which are associated with low enzyme activity and those which are general polymorphisms for loci within this gene, but does not allow definitive comments on population frequencies of the various mutations.

The presence of a G to A transition at codon 534 was detected as a heterozygous mutation in 1 of 60 individuals (0.8%). This is similar to that found in 78 colorectal cancer patients (1.3%) [4] and volunteer subjects from Taiwan (<0.8%), Japan (1%), Finland (3.3%) and African Americans (0.5%) [12]. The one individual with a codon 534 mutation in the current study had the lowest DPD activity (19.1 pmol min⁻¹ mg⁻¹ protein), suggesting that this mutation is associated with an impaired ability to degrade uracil, thymine, and 5FU. However, two previously described colorectal cancer patients with codon 534 mutations had DPD activity greater than the average value, making the impact of this mutation on pyrimidine degradation less clear [4].

Mutations at codon 543 were found in both the heterozygous and homozygous state in 28 of 60 individuals with an allele frequency of 28%. This frequency is nearly identical to that found in patients with colorectal cancer (26%) [4]. This is also similar to that found in Taiwanese (21%), Japanese (35%), and African Americans (22.7%) [12]. However, this allele was less common in Finnish subjects (7%) [12]. In both this study and the colorectal cancer population [4] a mutation at codon 543 was associated with a wide range of mononuclear cell DPD activity. This suggests that this mutation is a common polymorphism and alone is not associated with impaired DPD activity.

Mutations at codon 732 were also a relatively common finding as they were present in 6 of 60 individuals (allele frequency 5.8%). This is similar to that found in Japanese (4.4%) and Finnish (6.7%) subjects, but is more frequent then that found in Taiwanese (1.4%) and African American (1.9%) subjects [12]. The subjects with heterozygous mutations at codon 732 had a range of DPD activity suggesting that this mutation alone is not responsible for encoding low DPD activity. However, the one individual with a homozygous mutation at codon 732 had low mononuclear cell activity.

No subjects with ΔC1897 were detected, suggesting that alterations at this site are a rare occurrence in Caucasian subjects. No exon 14 splice site mutations were found amongst the 60 subjects studied. Our previous analysis of 75 colorectal patients found one individual heterozygous for a splice site mutation (1.3%) [4]. Combining the two studies suggests that the splice site mutation has an allelic frequency of 1 in 270. This allele was not found among 131 Taiwanese, 50 Japanese, or 105 African Americans, but had a 1% frequency in the Finnish population [12]. The finding in the Finnish population is of particular interest with the recent description of two Finnish children with thymine-uraciluria [18].

Three individuals had more than one mutation in DPYD; two were heterozygous for 543 and 732 and one was heterozygous for 543 and homozygous for 732. The three individuals had DPD activity values in the lower 10% of the population (25.9–53.4 pmol min⁻¹ mg⁻¹ protein). The mutations at codon 534 and 543 were initially identified in a patient who also has the splice site mutation [4]. This suggests that accumulation of DPYD mutations or the presence of modifying mutations may contribute toward the low activity phenotype of this enzyme. Future cDNA expression studies may be able to clarify the relative contributions of specific mutations to DPD deficiency [19].

At least five mutations in DPYD have been identified and many more are likely to be reported in the near future. The exon 14 splice site mutation, which has been found in both patients with cancer who experienced severe 5FU toxicity and children with thymine uraciluria, is clearly a rare event in the general Caucasian population. Further studies of various ethnic groups may suggest that analysis of this allele would be of therapeutic or diagnostic benefit in specific populations. The ΔC1897 results in premature termination of translation prior to the uracil binding site and no DPD activity was detectable in a homozygous mutant individual [13]. However, this study suggests that ΔC1897 is a relatively rare allele in the Caucasian population. While mutations at codons 543 and 732 occur more frequently in the Caucasian population, the phenotype-genotype analysis suggest that subjects heterozygous for these mutations do not encode a protein with low DPD activity. Whether the presence of homozygous codon 732 mutations or a combination of mutations contributes to low activity is yet to be evaluated. What is clear is that the five mutations are insufficient to provide a high level of diagnostic or predictive ability to identify patients who would be at risk for toxicity to 5FU or neonates at risk for mental retardation, epilepsy, or other features of developmental delay due to thymine uraciluria. Therefore, identification of further molecular alterations is required to facilitate the use of DPD analysis in paediatric genetics and cancer therapeutics.

The authors would like to thank Dr Rhona Watkins and her staff at the Glasgow and West of Scotland Blood Transfusion Service for their efforts during this study. This work was supported by a University of Aberdeen Faculty of Medicine Award, the Aberdeen Royal Hospitals Endowment Trust, and Wellcome Trust Grant 046607.