Profound Dihydropyrimidine Dehydrogenase Deficiency Resulting from a Novel Compound Heterozygote Genotype¹

The antimetabolite 5-FU³ has been in clinical use for ∼45 years and has evolved as an important agent in the treatment of a large spectrum of tumors, including gastrointestinal, breast, and head and neck (1). Over the past 20 years, the pyrimidine catabolic pathway, in particular the first enzymatic step involving DPD, has been recognized as critical in determining the ultimate metabolism and, in turn, the pharmacology of 5-FU (2). Although the antitumor action (and host toxicity) of 5-FU depends on anabolism of the drug to cytotoxic nucleotides, studies from our laboratory have shown that 80–90% of an administered dose of 5-FU is rapidly converted into biologically inactive metabolites through the catabolic pathway (3). Increased DPD activity (with corresponding increased catabolism of 5-FU) has been correlated with resistance to 5-FU (4, 5). Conversely, decreased DPD activity (with corresponding decreased catabolism of 5-FU) has been shown to increase 5-FU half-life, thus increasing the amount of drug available for the anabolic (cytotoxic) pathway (6). Taken together, these data suggest that the variability of DPD activity in the normal population may account for observed differences in the pharmacokinetics and oral bioavailability of 5-FU (7).

The importance of DPD in 5-FU toxicity has been illustrated by patients with DPD enzyme deficiency. After administration of standard doses of 5-FU, these patients develop profound toxicity including mucositis, granulocytopenia, neuropathy, and (in some cases) death (8, 9, 10, 11). In 1988, our laboratory reported one of the first profoundly DPD-deficient patients who developed grade IV toxicity after CMF chemotherapy (8). The familial pedigree of this patient revealed that the proband’s son and daughter were partially DPD deficient (8). Later population studies in breast cancer patients have demonstrated that ∼5% are DPD deficient, with enzyme activity below the 95th percentile of a control population (12).

Since being identified as a pharmacogenetic disorder, there has been a steady increase each year in the number of case reports of DPD deficiency with severe toxicity secondary to treatment with 5-FU (13, 14). These reports, combined with the recent introduction of new generation fluoropyrimidine-based chemotherapy agents (e.g., capecitabine) have resulted in continued research to understand the molecular basis of DPD deficiency with the purification and characterization of the human protein (15), cloning and characterization of the DPYD gene (16) and the promoter (17), and the identification of 19 variant alleles in DPD-deficient patients (18, 19, 20).

This report describes the first compound heterozygote genotype in a profoundly DPD-deficient patient with segregation of mutations among immediate family members. The mutations responsible for profound DPD deficiency in this patient and partial DPD deficiency in the proband’s two children are identified, and the pattern of inheritance is elucidated. This comprehensive familial approach which uses both phenotype and genotype analysis should be used in future studies to determine which mutations in the DPYD gene result in DPD deficiency.

In 1988, a previously healthy 40-year-old white female began treatment for infiltrating ductal carcinoma with a left-modified radical mastectomy and adjuvant CMF chemotherapy. The complete toxicity profile was described previously in detail (8). Briefly, 11 days after the first cycle of CMF chemotherapy, the patient was found to be neutropenic (white blood count of 1,400 with 1% neutrophils and 1% bands). Twenty six days after cycle 1, the second cycle was administered at reduced doses. Fourteen days later, the patient returned to the clinic with a white blood count of 1,300. The patient received the third cycle on schedule but at further reduced doses. Fourteen days later, the patient was hospitalized with fever, neutropenia, and a 2-day history of ataxia. The patient continued to deteriorate neurologically over the next 7 days, eventually becoming completely unresponsive. Over the succeeding 2 months, she regained all function and demonstrated no neurological deficits. Approximately 1 year later, the patient died from progression of disease. Although the cDNA and the structure of the human DPYD gene was not known in 1988, genomic DNA was prepared from the proband’s PBM cells for future analysis.

Phenotypic analysis revealed that the proband was profoundly DPD deficient (no detectable enzyme activity). Pharmacokinetic analysis of the proband demonstrated altered 5-FU metabolism characterized by prolonged half-life (150 versus 13 min for normal controls) and decreased clearance (70 versus 594 ml/min/m² for normal controls; Ref. 8). Phenotypic analysis of the proband’s children revealed that both her son and daughter were partially DPD deficient. Both children were otherwise healthy without any characteristic phenotype.

DPD enzyme activity was originally determined in the proband, her father, two children, and available extended family members from PBM cells as described (8). In the present study, DPD enzyme activity was reevaluated in the PBM cells of the available surviving family members (the proband’s husband, two children, and three grandchildren) and controls using a recently described semiautomated radioassay for DPD enzyme activity (21). Both the proband and her father (tested in 1988) have died since publication of the original study. The proband’s nephew (identified as partially DPD deficient in 1988) was unavailable for participation in this study. Plasma uracil concentrations were also measured (in surviving family members and controls where DPD enzyme activity was obtained) as described previously (8, 9).

The theoretical basis and validation of quantitating DPD mRNA using the ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) is described in detail elsewhere (22). Briefly, total RNA was isolated from PBM cells of the proband’s husband, two children, and three grandchildren, along with controls using RNAzol (Biotecx, Houston, TX), following the manufacturer’s instructions. Total RNA was diluted to a final concentration of 5 ng/μl in water containing 12.5 μg/ml total yeast RNA as a carrier. All samples were stored at −70°C until analysis. Final DPD forward, reverse, and probe concentrations were 100, 200, and 100 nm, respectively. Thermal cycling conditions were 30 min at 48°C, followed by 10 min at 95°C and 40 cycles of 15 s at 95°C and 1 min at 60°C. The absolute standard curve method was used to determine the copy number of DPD mRNA (22).

In the present study, genomic DNA from the proband’s husband, two children, and three grandchildren was prepared from PBM cells using the Easy-DNA kit (Invitrogen, San Diego, CA), following the manufacturer’s instructions. Archived genomic DNA obtained from the proband in 1988 was maintained at −70°C until analysis. All 23 exons along with the flanking intronic regions of the DPYD gene were PCR amplified under conditions similar to those described previously by our laboratory (9). Amplification of exon 1 (along with 427 bp of the promoter region of the DPYD gene) was carried out with 1% DMSO (17). The primers used to amplify each exon are listed in Table 1. After amplification, PCR products were resolved on 2% agarose gels and purified using a Qiaquick Gel Extraction kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Samples were sequenced on an ABI 310 automated DNA sequencer using the dideoxynucleotide chain termination method (Applied Biosystems, Foster City, CA). Sequence reactions were repeated three times in each direction (using the same primers that were used to amplify each specific PCR amplicon) and analyzed using MacVector 4.1 sequence analysis software (IBI, New Haven, CT).

Although the proband could not be retested for DPD enzyme activity, several separate determinations in the original study demonstrated no detectable enzyme activity in the PBM cells (8). In the present study, we examined the DPD enzyme activity in the proband’s husband, two children, and three grandchildren. These data are summarized in Fig. 1 and Table 2. The proband’s husband (II-6) demonstrated DPD enzyme activity within the normal range. However, both the daughter (III-6) and son (III-7) demonstrated partial DPD deficiency, as reported originally (8). The proband’s three grandchildren (IV-1, IV-2, and IV-3), along with seven unrelated controls, demonstrated normal DPD enzyme activity (Fig. 1 and Table 2). The proband’s father (I-1) and nephew (III-2) are illustrated as partially DPD deficient in Fig. 1. These data were obtained from the original examination of this family (8) because the proband’s father is now deceased and the nephew was unavailable for participation in this study.

Examination of the proband’s plasma in 1988 demonstrated elevated (1500 ng/ml) uracil levels (8). In the present study, the plasma uracil levels in the proband’s husband (II-6), two children (III-6 and III-7), and three grandchildren (IV-1, IV-2, and IV-3) were examined and found to be in the normal range (Table 2).

DPD mRNA levels were quantitated by real-time quantitative PCR in the PBM cells of the proband’s husband (II-6), two children (III-6 and III-7), and three grandchildren (IV-1, IV-2, and IV-3) and are summarized in Table 2. No archived RNA was available from the proband (II-5) or her father (I-1). DPD mRNA levels were also quantitated in the PBM cells of seven unrelated control individuals (Table 2) with normal DPD enzyme activity. DPD enzyme activity correlated (R² = 0.98) with mRNA levels in family members with normal enzyme activity but not in the partially DPD-deficient children (III-6 and II-7; Fig. 2).

Because only archived genomic DNA was available from the proband, DPYD gene-specific primers (located in bordering intronic regions and listed in Table 1) were used to amplify the 23 exons (along with intron/exon splice junctions) in the proband’s DPYD gene. Sequence analysis of purified PCR amplicons showed that the proband contained two heterozygote mutations: (a) a G to A mutation in the GT 5′ splicing recognition sequence of intron 14, which results in a 165-bp deletion (corresponding to exon 14) in the DPD mRNA (DPYD*2A); and (b) a T1679G mutation (designated DPYD*13), which results in a I560S substitution. These data are summarized in the family’s pedigree (Fig. 1) and in Table 2. There were no other mutations or polymorphisms detected in the proband’s DPYD gene.

The proband’s husband (II-6, and father to both children) demonstrated neither the DPYD*13 nor DPYD*2A mutation in his DPYD gene. However, complete sequence analysis of the father revealed that he was a heterozygote for both the T85C (C29R, DPYD*9A) and the A496G (M166V) mutations (as shown in Fig. 1 and Table 2).

Sequence analysis of the proband’s two children revealed a segregation of mutations, with the daughter (III-6) demonstrating a heterozygote DPYD*13 T1679G (I560S) genotype and the son (III-7) demonstrating a heterozygote DPYD*2A (intron 14 G1A) genotype. The proband’s daughter (III-6) also demonstrated a heterozygote A496G (M166V) mutation. All three of the proband’s grandchildren (IV-1, IV-2, and IV-3) demonstrated a heterozygote genotype for the A496G (M166V) mutation (as summarized in Fig. 1 and Table 2). The locations of the mutations (within the DPYD gene) characterized in this study are shown in Fig. 3.

In the present study, we describe the genotypic characterization of one the first known profoundly DPD-deficient cancer patients. Surviving family members were analyzed to elucidate the pattern of inheritance and to determine the role of each mutation in DPD deficiency. Several earlier studies of DPD-deficient patients have reported homozygote mutations resulting in complete DPD deficiency (9, 11, 19, 20, 23, 24); however, this is the first report of a compound heterozygote genotype with segregation of mutations among immediate family members. These analyses demonstrate a codominant pattern of inheritance for this pharmacogenetic disease. In addition, two previously reported DPD mutations [T85C (C29R, DPYD*9A) and A496G (M166V)] thought previously to result in DPD deficiency (19, 23, 24) are shown in this study to have no functional significance to DPD activity.

To determine the molecular basis for DPD deficiency in the proband, archived genomic DNA (isolated from the patient’s PBM cells in 1988) was used as a template, and all 23 exons (including the intron/exon junctions) were PCR amplified. Primers in the flanking introns were designed based on data from the characterization of the human DPYD gene (16), promoter (17), and recent studies examining variant DPYD alleles (25). Complete sequence analysis revealed two heterozygous mutations in the proband: (a) a G to A mutation in the GT 5′ splicing recognition sequence of intron 14 (DPYD*2A); and (b) a T1679G mutation (now designated DPYD*13), which results in a nonconserved I560S substitution. Complete sequence analysis demonstrated that these alleles represent the only sequence differences in this DPD-deficient patient. The DPYD*2A mutation remains the most characterized and frequently observed allele associated with DPD enzyme deficiency (9, 11, 25). However, the DPYD*13 missense mutation has been reported in only one other patient with reduced DPD enzyme activity (26).

The proband’s husband (and father to both partially DPD-deficient children) demonstrated neither the DPYD*2A nor the DPYD*13 mutation in his DPYD gene (as shown in Table 2). However, complete sequence analysis of the father revealed that he was heterozygote for both the DPYD*9A and the A496G (M166V) mutations. Both of these mutations have been reported previously to be associated with DPD deficiency (19, 23, 24). However, the father demonstrated normal DPD activity. Although the frequency of these alleles in the general population remains to be determined, their identification in an individual with normal enzyme activity suggests that both are nonfunctional polymorphisms. This conclusion is supported by a recently published study demonstrating normal DPD enzyme activity in individuals with the DPYD*9A polymorphism (26) and by the identification of the A496G (M166V) genotype in the proband’s three grandchildren with normal DPD enzyme activity (Fig. 1). These results (combined with the autosomal codominant pattern of inheritance demonstrated in the proband and her children, see below) confirm that neither of these mutations alters DPD enzyme activity.

Sequence analysis of the proband’s two children demonstrated segregation of both mutations (DPYD*2A and DPYD*13) identified in the proband. The son demonstrated a heterozygote DPYD*2A genotype whereas the daughter demonstrated a heterozygote DPYD*13 genotype. This unique segregation of mutations revealed that the proband was a compound heterozygote containing one variant allele on each DPYD gene. Furthermore, this genotype allowed the independent assessment of the effect of each mutation on DPD enzyme activity. As shown in Table 2, both children demonstrated partial DPD deficiency, thus demonstrating that both forms of the DPYD gene were expressed and that this pharmacogenetic syndrome follows a codominant pattern of inheritance, not autosomal recessive as originally thought (8).

The molecular basis for DPD deficiency is best understood in the DPYD*2A mutation. A previous study by our laboratory demonstrated that a homozygote DPYD*2A genotype results in a 165-bp deletion (corresponding to exon 14) in the DPD mRNA (9). Western blot analysis demonstrated that the aberrant DPD mRNA was translated into a nonfunctional polyubiquitinated DPD protein. Thus, enhanced proteolysis provides a hypothetical mechanism for loss of DPD catalytic activity, similar to what has been reported with human thiopurine S-methyltransferase deficiency (27). In contrast, the DPYD*13 mutation results in a single nonconserved I560S substitution. Although this substitution does not occur in any recognized functional domain, amino acids located in regions critical for enzyme structure and/or catalytic function would be expected to be conserved across species. A recent report examining the conservation of the DPD enzyme demonstrates that the I560 position is 100% conserved across all five mammalian species (human, mouse, rat, bovine, and pig) examined (28) and suggests that this position is important in maintaining DPD enzyme activity.

The phenotypic characterization of this family demonstrating partial DPD deficiency together with a known genotype prompted us to examine other types of analysis (plasma uracil and DPD mRNA levels) that may, in the future, be used to identify DPD deficiency before treatment with 5-FU. Analysis of the partially DPD-deficient individuals (the proband’s daughter and son) showed plasma uracil levels in the normal range (<50 ng/ml; Table 2). Elevated uracil levels were detected only in the proband, who demonstrated profound (no detectable enzyme activity) DPD deficiency. These analyses suggest that only profoundly, not partially, DPD-deficient individuals have elevated plasma uracil levels. This is particularly important because a recent study reports that most DPD-deficient patients in a population of patients who demonstrated unanticipated toxicity secondary to treatment with 5-FU are partially deficient (31% partial versus 12% profound deficiency; Ref. 10). These data illustrate the potential limitations of determining plasma uracil levels for the identification and characterization of partially DPD-deficient patients.

In addition to uracil levels, DPD mRNA levels were also evaluated (Fig. 2 and Table 2). DPD mRNA levels quantitated from the proband’s husband and three grandchildren (individuals in the family having normal DPD activity) correlated with enzyme activity (Fig. 2). These data agree with previous studies from our laboratory demonstrating a linear relationship between DPD mRNA and activity levels (22). The partially DPD-deficient family members (III-6 and III-7) also demonstrated DPD mRNA levels in the normal range, suggesting that neither mutation (DPYD*2A nor DPYD*13) affects DPD mRNA levels. This is not particularly surprising because neither mutation would directly interfere with RNA transcription, nor do they occur in regions known to affect RNA stability (29). Although several recent pharmacogenomic studies have used DPD mRNA as a surrogate marker for DPD enzyme activity to predict 5-FU efficacy (30, 31), the present studies suggest that quantitation of DPD mRNA cannot be used to identify DPD-deficient patients with this genotype. Taken collectively, these data indicate that phenotypic analysis of DPD enzyme activity remains the most reliable method for the identification and characterization of this pharmacogenetic syndrome.

In summary, comparative phenotypic and genotypic analysis of this family has allowed us to conclude that: (a) a new compound heterozygote genotype resulting in profound DPD deficiency has been identified; (b) DPD deficiency exhibits an autosomal codominant pattern of inheritance, not autosomal recessive as thought originally; (c) two mutations reported previously to be associated with DPD deficiency [T85C (C29R, DPYD*9A) and A496G (M166V)] have no functional significance on DPD enzyme activity. Further analysis suggests that neither uracil nor DPD mRNA levels can be used to predict partial DPD deficiency for the mutations examined in this study. This approach demonstrates the usefulness of familial genotypic and phenotypic analyses to determine the functional significance DPYD mutations.